348 ( – , Electrophorus M. V. L. BENNETT Molopterurus Sachs’ Main 9 Astroscopus ( 0 ) Gnothonemus Gymnorchus
which are described in the following chapter. Electrosensory systems can function in communication between fish. Some strongly electric fish also have weakly electric organs.
Strongly electric fish must have been known to primitive man, and Torpedos are said to have been used by the Romans in a primitive, and probably subconvulsant, form of shock therapy ( Kellaway, 1946). Recognition
that the discharge was electric came soon after the development of the Leyden jar allowed widespread study of electricity by scientists10. ELECTRIC ORGANS 349 and experience of shocks. The Torpedo and electric eel were studied by many early physiologists, but the stargazer escaped the attention of the scientific community until just before Dahlgren and Silvester described its organ in 1906.
Recognition of weakly electric fish was relatively delayed because of the imperceptibility of their discharges, and the electric activity of many was unknown until the 1950s (Lissmann, 1951, 1958; Coates et al., 1954;
Grundfest, 1957). Howcver, all the groups had been recognized from morphological evidence by the early 1900s although their organs were sometimes classified as “pseudoelectric.” Historical references are given in the earlier reviews by Grundfest (1957) and Keynes (1957), and the historical essays of Kellaway (1946) and Mauro (1969) are charming.
Relatively brief reviews of recent work are given in Bennett (1970) and Grundfest ( 1967).
Electric fish have been studied in part because of their remarkable physiological abilities and the ease of detection of their discharges. Also, it has been hoped that they might reveal a great deal about normal function. The argument is that evolution may have exaggerated some aspect of organ or cell that makes a general phenomenon more understandable or easier to study. The giant axon of the squid is an outstanding example of a cell that has evolved in such a way-toward increased size-that makes feasible many kinds of experiments that are much more difficult in other tissues ( Hodgkin, 1964).
The generating cells of electric organs are modified from muscle fibers except in the starchiness family of the larger group, the gymnotids. In the sternarchids the myogenic part of the organ has been lost and the organ is modified from nerve fibers; that is, it is neurogenic. The evidence of origin will be discussed in Section II,G,l,f. In a number of species the generating cells are flattened and for this reason have been termed electroplaque( s ) , electroplate( s ) , or clectroplax( es). However, in many of the more recently described organs, the generating cells have quite complex shapes. Although electroplaque still seems a natural term for flattened cells, the author has introduccd the more general term electrocyte to refer to any generating cell whatever its shape (Bennett, 1970).
Electric organs generally are rather gelatinous, and a large fraction of their volunic is extracellular space. They contain a considerable amount of connective and other accessory tissues as well as blood vessels and motor nerves that control the discharge. As will be seen below the connective tissue can be important in channeling the flow of current.
Electrocytes work on the same general principles as ordinary nerve and muscle cells: potentials are generated across membranes. In all known cases the potentials result from selective permeability and passive movement of ions down their concentration gradients. But it is likelyElectric catfish Stargazers Gymno tidae 1 species, Gymnotus curapo Sternopygidae 4 or 5 genera, a number of species Rhamphichthyidae 2 monospecific genera Sternarchidae About 9 genera, a number of species Malapteruridae 1 species, Malapterurus electricus Uranoscopidae 1 electric genus, Astroscopus; several species
Weak Freshwater, South
America
Weak Freshwater, South
America
Weak Freshwater, South
America
Weak Freshwater, South
Smerica
Strong, more than
300 V d
Strong? about 5 V Marine, Western in air from small Atlantic animals8 a Arranged phylogenetically; see Greenwood et al. (1966) for more detailed classifications. From Bennett (1970).
From Bennett et al. (1961). c From Albe-Fessard (1950a).
From Remmler (1930).
* From Bennett and Grundfest (1961~).
w B352 M. V. L. BENNETT that some electrocytes will be found to have electrogenic pumping [movement of ions that involves net current flow and is linked to reactions such as adenosine triphosphate ( ATP) hydrolysis (see Hodgkin, 1964; Albers, 1967)l. When the membranes on opposite faces of the generating cells are at the same potential that at rest is the resting potential, no current flows (Fig. 2A). When the membranes are at different potentials, current flows in a circuit that involves the two membranes, the cell cytoplasm, and the external medium (Fig. 2B). Figure 2C represents current flow around an electrocyte when the two faces are at different, but uniform, potentials ( dotted lines). These currents are associated with potential changes in the external medium, and surfaces at equal potentials ( equipotential surfaces) are also diagrammed in Fig. 2C (solid lines). The current flow and field that an electric fish sets up around itself is essentially like that of Fig. 2C but larger since many cells in series and parallel are active at the same time. If an object such as a hand is present in the external medium, a potential difference is present across it and current flows through it. The current from a strongly electric fish is of sufficient magnitude to excite nerves, muscles, or receptors in the hand. An object distorts the electric field if it is of different conductivity than the medium, and the distortions, if large enough, can be detected by a fish’s electrosensory system (see Chapter 11, this volume).
The relatively large size of external potentials generated by electric organs (Table I ) as compared to other excitable tissues is not a result of larger membrane potentials although they may be slightly larger in a few instances. Rather the large outputs are a result of ( a ) arrangement of a cell’s membranes in such a way as to maximize current outside the cell; ( b ) synchronous activity of many cells arranged in series and parallel; ( c ) to some degree, lower membrane resistances; and ( d ) accessory structures tending to channel current flow. The different kinds of
adaptation will be discussed in connection with the various types of organ.
The types and patterns of electric organ discharge can be divided into several categories. The strongly electric organs all produce essentially monophasic pulses. They are all active intermittently, as indeed they must be, for the power outputs are so large that the fish can maintain them for only short periods. The organ is normally silent and is
generally, if not exclusively, discharged in response to appropriate external stimuli. These stimuli may be tactile, chemical, electric, or perhaps visual. Responses are single pulses or trains of pulses usually of fairly
constant size ( Fig. 3A,A’).
Generally, weakly electric organs of freshwater fish continually emit pulses of rather constant size. The pulses may be monophasic, diphasic,10. ELECTRIC ORGANS 353 + + + + + + + + A (=) – – – + + + + + + + + Resting, no current flow + + + + + + + + + + I + Upper face active D 7 External potential Indifferent electrode Face 2 r r———- ‘ Monopolarly recorded j to active face ‘ C 1 potential external L—–7—i I active face I t Fig. 2. Equivalents of electrocytes during rest and activity. ( A ) Diagram of a cell at rest; equal potentials are opposed. ( B ) When the upper face generates an overshooting action potential, two potentials act in the same direction and current flows as indicated by arrows. ( C ) Potentials (solid lines) and lines of current flow (dotted lines) around a thin electrocyte indicated by the heavy
horizontal line. If the polarity of the cell corresponds to that in ( B ), the potential is negative above the thin horizontal line, which is the zero potential, and positive below it. For reasons of mathematical simplicity the diagram is for very long ribbon shaped cell of zero thickness. The membranes are assumed to act as current sources corresponding to a shell of magnetic dipoles. The isopotentials are separated by equal increments and meet at the two edges of the cell. ( D ) Electrical equivalent of a resting cell. The resistance of the external current path can be
represented by re,, rez, and res. ( E ) Equivalent of an active cell. The small arrows indicate the direction of current flow.354 M. V. L. BENNETT
Fig. 3. Patterns of electric organ discharge. ( A ) An electric catfish 7 cm long.
Potentials recorded between head and tail in a small volume of water with head negativity upward. Mechanical stimulation evokes a train of five pulses ‘ which attain a maximum frequency of 190/sec. (A‘) Single pulses can also be evoked (faster sweep speed). (B-D) Weakly electric gymnotids immersed in water, discharges recorded between head and tail, head positivity upward. (A) A variable frequency gymnotid, Gymnotus; pulses are emitted at a basal frequency of approximately 35/sec. Tapping the side of the fish at the time indicated by the downward step in the lower trace causes an acceleration up to about 65/sec. The acceleration persists beyond the end of the sweep. The small changes in amplitude result from movement of the fish with respect to the recording electrodes. ( B )
Faster sweep showing the pulse shape. ( C ) Sternopygus, a constant, low frequency gymnotid. The pulse frequency is about 55/sec. The horizontal line indicates the zero potential level. ( D ) Sternarchus, a constant, high frequency gymnotid. The frequency is about 800/sec. The horizontal line indicates the zero potential level.
Calibrations in volts and milliseconds. From Bennett ( 1968a). triphasic, or even more complex. The patterns of emission fall into two categories. In one, the responses are brief pulses separated by long intervals.
These species generally accelerate their discharges when presented with almost any kind of stimulus (Fig. 3B,B’). Acceleration results in an increased rate of testing the environment but may also represent a signal to another fish (Bullock, 1970; Black-Cleworth, 1970; Moller, 1970 ) .
In the second category the duration of the pulses is as long or longer than the intervals between them (Fig. 3C,D). Generally pulses are emitted at a very constant frequency that can be very high. Recently, it has been found that in most species weak electric stimulation at a frequency close to that of the discharge causes small shifts in frequency
( Watanabe and Takeda, 1963; Bullock, 1970). These changes apparently represent an attempt to avoid “jamming” of the electrosensory system by the applied signal. Other small changes of frequency may function in communication ( Bullock, 1970; Black-Cleworth, 1970).
These two discharge patterns of weakly electric organs have been termed variable and constant frequency (although the terms now need a modifier such as relatively). They have also been called buzzers and 10. ELECTRIC ORGANS 355 hummers from the sound one perceives when the electric pulses are fed into a loudspeaker system. Variable frequency species include the mormyrids and many gymnotids. Constant frequency species include the
remainder of the gymnotids and Gymnurchus. Gymnarchus, mormyrids, and several gymnotids are able to cease discharging for brief periods.
This response appears to represent hiding by keeping quiet electrically and may also function in communication. Discharge patterns of weakly electric organs in marine fish are intermittent, but are poorly known under natural conditions.
11. ELECTRIC ORGANS AND ELECTROCYTES
A. Methods
An introduction to methods of establishing the properties of electrocytes may be helpful at this point. Microelectrodes are placed intracellularly and at various sites externally. Potentials at the tips can then be recorded by suitable amplifiers and associated equipment. A recording is termed monopolar if it represents the potential difference between the electrode tip and a distant grounded electrode in the vessel containing the preparation. The distant electrode is often called the indifferent electrode because it is sufficiently far from the active cell that moving it around does not detectably affect the recorded potential. One may use a differential amplifier to subtract the potential recorded through one microelectrode from that recorded through another; the result is differential recording. Monopolar recordings can also be subtracted after the experiment to obtain the same result.
In discussing properties of single electrocytes it is useful to have an equivalent circuit. As a first approximation one can consider an electrocyte to be a flattened cell with uniform potentials across each face that differ from each other during activity. Because of the uniformity of potential, each face can be given a simple equivalent circuit like those
used for other membranes. (Of course a “naked” cell could not have a discontinuous change in potential at the edge of the cell for this would require an infinite current density. However, an electrocyte is often situated in an insulating connective tissue sheath that adheres closely to its edges, and the potentials over the faces are indeed quite uniform. In any case the nonuniformities are not important for most considerations of membrane properties.) Separate branches of the equivalent circuits for each membrane can be assigned to different ion species, each branch with a particular internal (equilibrium or Nernst) potential and a con356 M. V. L. BENNETT ductance that may or may not vary during activity. Alternatively, each membrane can be considered to consist of a single branch with a single
internal potential and conductance, both of which can change. The membrane capacity, which may or may not be significant in electrocytes, is in parallel with the other branch or branches of the membrane equivalent. Generally, but not always, the resistance of the cytoplasm is negligible, and in the equivalent circuit of the entire cell the two membranes are directly connected together but oriented in the opposite direction. The circuit of the electrocyte can then be represented as a three terminal network, two terminals just external to each face and one inside the cell (Fig. 2D). The external terminals can be considered to be connected by a resistive path through the tissue and medium surrounding the cell. Another resistance leading to the indifferent electrode is required if monopolar recording is employed. This resistance is connected into the resistance of the external path thereby dividing it into two components. The placement of the connection depends on the symmetry with respect to the indifferent electrode of the material making up the external resistance. There are thus three resistances in the external path which are labeled T,,, re2,a nd resi n Fig. 2D.
Consider what is recorded when one membrane of a cell generates a spike and the other membrane does not change its properties (Fig. 2E ) .
Appropriate differential recordings show the potentials across each face and across the external medium. The recording across the active face gives the internal potential of the membrane less whatever internal voltage drop there is due to current flow through the opposed, inactive face and external path. Differential recording across the inactive face gives the passive voltage drop across this face that results from current flow. The potential across the external path is the voltage drop across the external resistances and also the difference between the potentials across
the opposed membranes. Monopolar recording outside the two faces gives potentials of opposite sign; for a conventional depolarizing response the recording is negative going outside the active face and positive going
outside the inactive face. The potentials are smaller than when differential recording is used because they represent voltage drops across part of the resistance around the external circuit ( r e l and rr2 in Fig. 2D). A monopolar recording by an intracellular electrode cannot distinguish between the two faces. The spike is smaller than that recorded across the active face by the voltage drop across T , ~in Fig. 2D. The spike is larger than the passive drop across the inactive face by the voltage drop across rez.
Obviously the same kind of inferences can be made from monopolar and differential recording since monopolar records can be subtracted to10. ELECTRIC ORGANS 357 give the equivalent of differential recording. However, differential recordings often are more easily interpreted and electronic subtraction is much easier than the same process done graphically. An additional advantage is that interfering potentials resulting from activity of distant
cells tend to be about the same size at different monopolar electrodes and are thus subtracted out by differential recording.
It is useful and often essential to pass stimulating currents through the faces. When current is passed from an intracellular electrode to an indifferent electrode, it flows outward through both faces in amounts depending on their resistances, capacities, and any responsiveness they exhibit as well as on the external resistances. Each face is depolarized and a response of one face may obscure a response of the other. For this reason it may be desirable to pass current between an external electrode outside one face and the indifferent electrode. (Use of two external electrodes, one outside each face, would be preferable in some ways but has not been carried out because of added technical complexity.) In this case part of the applied current flows inward through one face and outward through the other. Since this current polarizes the faces in opposite directions, it is generally simpler to evaluate the responsiveness of the two faces.
Electrocytes have been studied both in situ and in isolated tissue
bathed in suitable physiological saline. For in situ work normal organ
discharge can in many species be stopped by spinal section. Curare
which blocks vertebrate neuromuscular transmission also blocks nerveelectrocyte
transmission. It can be used to immobilize a fish as well as
to block neurally controlled activity of myogenic electrocytes, but of
course it prevents study of nerve-electrocyte transmission.
In a number of instances it is possible to apply currents to columns of
cells or even to part of the intact animal. By means of a bridge circuit
the potential resulting from voltage drops across fixed resistances in
series with the active tissue can be subtracted from the records, and
information similar to ( and supplemental of) that obtained from single
cells can be obtained ( Albe-Fessard, 1950b; Bennett, 1961; Bennett et al.,
1961). The success of experiments of this kind is dependent on the series
arrangement of the active membranes.
B. Membrane Properties
The surface membranes of electrocytes exhibit a number of different
passive and active properties. In considering these properties, it is convenient
to think of the cell surface as containing various kinds of sites358 M. V. L. BENNETT
where different and more less-specific ions can enter or leave the cell.
These sites are intermixed in varying degrees and proportions to account
for the different kinds of membrane activity. The membrane separating
the sites has a bimolecular lipid core that presumably is of very high
resistance and inactive in ion movement. Most of the membrane capacity
can be assigned to this part of the surface. For most cells the specific
capacitance is about 1 pF/ cm2 while membrane resistance-reflecting
the number and nature of sites for ionic movement-can vary over a
range of six or seven orders of magnitude. This concept of localized
sites of ion movement has received strong support from recent experiments
using artificial bimolecular lipid membranes. These membranes
are similar to ordinary membranes in dimensions and capacitance but
have an extraordinarily high resistivity. A number of compounds lower
the resistance into the physiological range by providing sites for movement
of ions or by acting as carrier molecules (e.g., Cass et d., 1970).
In impulse responses of electrocytes there appear to be three types of
sites that change their properties as a function of membrane potential.
When the membrane is moderately depolarized, the conductance at
sodium sites increases (sodium activation) allowing influx of sodium and
further depolarization. This process is responsible for the active rising
phase of the spike. If depolarization is maintained, the sodium conductance
decreases again (sodium inactivation) and this change is a
factor in return of the membrane potential to its resting level. At the
resting potential the inactivated sodium sites gradually recover their
ability to increase in conductance when the membrane is depolarized.
(Any remaining activated sites rapidly return to their low, resting conductance
at the resting potential and can be rapidly activated again
without further delay.) One kind of potassium site in electrocytes increases
in conductance when the membrane is depolarized (potassium
activation or delayed rectification), but the change is delayed compared
to sodium activation. The increased potassium conductance tends to
restore the cell to the resting potential where the conductance gradually
returns to normal. If the membrane is kept depolarized, the increase in
potassium conductance can also reverse (potassium inactivation), a
process which is generally much slower than either the reversal at the
resting potential or sodium inactivation. So far these changes are like
those of ordinary nerve (Hodgkin, 1964; Hille, 1970). Another kind of
potassium site decreases in conductance when an outward current is
passed through it. This change is very rapid in onset and reverses very
rapidly when the outward current is reduced. It resembles and probably
has the same mechanism as anomalous or outward rectification in muscle10. ELECTRIC ORGANS 359
(Adrian et al., 1970; cf. Bennett, 1970). It can also be considered a kind
of inactivation ( Bennett and Grundfest, 1966).
The resting membrane potential is apparently largely determined by
potassium selective sites and the concentration gradient of potassium,
although the cells may also be significantly permeable to chloride. In
some electrocytes large hyperpolarizations cause the conductance of the
resting membrane to decrease which can lead to what have been termed
hyperpolarizing responses ( Bennett and Grundfest, 1966). Since the
conductance decreases by a large fraction of the resting value, these
responses must involve a change in the potassium selective sites of the
resting membrane. There is no known physiological significance of these
responses.
In myogenic electrocytes another kind of a response is mediated by
sites sensitive to the neurotransmitter acetylcholine released by the presynaptic
nerve fibers. The transmitter increases the permeability at these
sites, presumably to sodium and potassium ions, and the resulting current
flow generates a postsynaptic potential (PSP). The change in conductance
is virtually independent of the membrane potential, a property
which has been termed electrical inexcitability ( Grundfest, 1957; Bennett,
1964). An important consequence of there being an electrically inexcitable
conductance increase is useful in establishing the chemical
mediation of a postsynaptic response (Bennett, 1966). If the membrane
is made sufficiently inside positive prior to the release of transmitter, the
PSP current that would have depolarized the cell reverses and flows to
make the inside of the cell less positive; the response is inverted. The
potential at which the current reverses (loosely speaking the equilibrium
potential by analogy with Nernst potentials) is generally slightly negative
to the zero potential; this is the primary reason for believing that
permeability is increased to both sodium and potassium (see N.
Takeuchi, 1963).
Sites of the kinds described are combined in a number of ways in
membranes comprising the faces of different electrocytes. These combinations
are as follows:
(1) Spike generating membrane. This membrane contains sodium
sites, and may or may not have potassium activation. Probably anomalous
rectification is present in spike generating membranes of all myogenic
organs. PSP generating sites may or may not be present.
( 2) Membrane exhibiting delayed rectification without spike activity.
This membrane appears to be spike generating membrane that has lost
the sodium mechanism. PSP generating sites may or may not be present.
( 3) Postsynaptic potential generating membrane. Postsynaptic poten360 M. V. L. BENNETT
ca re
1 1 2 3
1 2 3
1 2 3
i i i
1 2 3 r3’
Fig. 4.. Activity of different kinds of electrocytes and their equivalent circuits.
The upper part of each circuit, labeled only in A, represents the innervated or
stalk face; the lower part represents the uninnervated or nonstalk face. The resistance
of the extracellular current path is represented by re. The potentials that are recorded
differentially across the two faces (V, and V,) and across the entire cell (V,)
are drawn to the right of each circuit (intracellular positivity and positivity outside
the uninnervated face shown upward; note that Vi – V, = Ve). Placement of
electrodes for recording these potentials is indicated in C. The successive changes
in the membrane properties are shown by the numbered branches of the equivalent
circuits, and their times of occurrence are indicated on the potentials. A lower
membrane resistance is indicated by fewer zigzags in the symbol. Return to resting
condition is omitted. ( A ) Electrocytes of strongly electric marine fish and discshaped
electrocytes of rajids. The innervated face generates only a PSP. ( B ) Cupshaped
electrocytes of rajids. The innervated face generates a PSP and the
uninnervated face exhibits delayed rectification. ( C ) Electrocytes of the electric
eel and a few other gymnotids. The innervated face generates an overshooting
spike. ( D ) Electrocytes of mormyrids and some gymnotids. Both faces generate
a spike, and V, is diphasic. In some the spike across the uninnervated face is
longer lasting and the second phase of V, predominates. These potentials are10. ELECTRIC ORGANS 361
tial generating sites may be the only responsive elements in a membrane
as well as occurring intermixed in membrane of the preceding kinds.
(4) Low resistance inexcitable membrane. This membrane has a
low resistance owing to a large potassium conductance, and it generates
a resting potential; its properties do not change when the potential across
it is changed. In some electrocytes there may be a significant C1
conductance.
( 5 ) High capacity inexcitable membrane. This membrane has few
sites for ionic movement, and its resistance is high. As a result, during
activity virtually all of the current through it is capacitative. Its capacity
per unit area of plasma membrane is probably similar to that of other
membranes, that is, 1 pF/cm2. Where membrane of this kind has been
studied by electron microscopy, the surface is highly convoluted on an
ultrastructural level, and therefore the capacity per unit area referred to
macroscopic area is larger than the values for ordinary cells. [A similar
situation obtains in muscle because of the transverse tubular system (cf.
Gage and Eisenberg, 1969) .] In electrocytes of the gymnotid Sternopygus
one membrane has an effective capacity so large that it probably
results from shift in ionic distributions on either side of the membrane
(see Section 11, C, 1, d).
In different electrocytes the two faces are made up of different combinations
of the foregoing kinds of membrane. The equivalent circuits at
rest and during activity are diagrammed in Fig. 4; they will be discussed
in detail in sections concerning the different kinds of fish in which they
occur.
There is good indirect evidence that sites mediating the different
components of electrically excitable responses are independent of each
other on a microscopic scale and that they are also separate from PSP
generating sites (which often occur in the same general region of the
cell). To be sure the spatial resolution of electrophysiological techniques
may not allow measurement of the separation of different kinds of membrane
responsiveness. Nonetheless, physicochemical considerations and
separability of different kinds of electric activity under various electrical
and pharmacological treatments strongly support the idea of separate
shown by dashes. ( E ) Electrocytes of the electric catfish. The stalk face is of
higher threshold and generates a smaller spike (indicated by a smaller battery
symbol) and the external potential is entirely negative on the nonstalk side distant
from the stalk. ( F ) Electrocytes of Gymnarchzrs. The uninnervated face acts as a
series capacity and the external response has no net current flow. The summation
of a second response (2’,3’) on the first is shown by dashes. Electrocytes of
sternarchids and Eigenmanniu may operate similarly. From Bennett ( 1970).362 M. V. L. BENNE’IT
channels mediating different response components (see, especially, Hille,
1970; but also Bennett, 1961; Grundfest, 1966). Another argument for
separability of different components is the separate occurrence in
different cells and in macroscopically different regions of a single cell.
Electrocytes provide much of the comparative evidence of this kind.
A frequent characteristic of electrocytes is a kind of impedance
matching between the faces (Bennett, 1961). The lower the membrane
resistance the greater the electrical output of the cell, but also the greater
the leakage at rest, and the greater the exchange between Na’ and K+
across spike generating membrane during activity. It would thus be
inefficient to have one face of much lower resistance than the other (or
of much lower resistance than the series resistances of the external
medium and cytoplasm). One finds that a low resistance inexcitable face
generally is of lower resistance than the opposed, excitable face at rest
(Fig. 4A,C). The excitable face reduces its resistance during activity so
that under these conditions the resistances of the two faces are more
closely matched. In electrocytes where both faces become active and
decrease their resistances, the resting resistances are more or less equal.
Often this kind of impedance matching has morphological correlates,
which will be discussed in respect to individual cases and in Section 11, F.
C. Marine Electric Fish
There are three groups of marine fish possessing electric organs, the
torpedinids (electric rays), the rajids (skates or rays), and members of
the teleost genus Astroscopus (one group of stargazers). Because of
functional similarities between the electric organs it is convenient to consider
these three groups together, although other characteristics make
them the “lowest” and the ‘%ighest” fish possessing electric organs. The
electrocytes of Astroscopus, the torpedinids, and some of the rajids are
the simplest known. There is only the one response component in the
innervated face and no response in the uninnervated face.
1. Astroscopus
The electric stargazers are a small group of several species that occur
along the western coast of the North and South Atlantic. There may also
be a representative on the Pacific side of the Panamanian isthmus
(Dahlgren and Silvester, 1906). These fishes are unusual in being somewhat
flattened dorsoventrally. Their eyes are located on the dorsal
surface of the head and look virtually straight upward, hence the name
stargazer. Their habit is to burrow into sand leaving only their eyesFig. 5. Innervation and structure of electrocytes of the stargazer. The upper
picture shows the dorsal surface of a single cell teased from formalin fixed material
and stained with methylene blue. A nerve bundle enters from the lower left and
forms profuse, sometimes anastomosing branches. Finer branches are not resolvable.
Capillaries ( c ) are also seen. The lower micrograph is a vertical section through
the organ, dorsal surface uppermost, following osmic acid fixation. The innervated,
dorsal surfaces are smooth; the uninnervated, ventral surfaces have long papillae.
Two nerve bundles are seen on the left. The edge of a cell between two others
(lower right) illustrates the characteristic irregular layering of the cells. Just to
the right of center is a vertical fissure (arrow) which may represent edge to edge
apposition of two cells. From Bennett and Grundfest (196lb).
363364 M. V. L. BENNElT
protruding when they are practically invisible except for the two small
black spots of their pupils. From this position they gulp down unwary
minnows passing overhead. The electric organs lie just behind the eyes
( Fig. 1 ) , and they are in point of fact modified from extraocular muscles
(Dahlgren, 1914, 1927). Other members of the same family, Uranoscopidae,
are very similar in appearance but lack electric organs.
The electrocytes are large flattened cells that lie in the horizontal
plane and are densely innervated on their dorsal surface by branches of
the very large oculomotor nerves (Fig. 5 ) . The ventral surface has many
short processes or papilli that markedly increase its surface. This surface
is further increased by many small invaginating tubules or canaliculi
(Mathewson et al., 1961; Wachtel, 1964). About 150-200 layers of cells
are arranged in series, one above the other (Dahlgren and Silvester,
1906). Each layer of cells in parallel consists of about four large cells
(approximately 5 mm in diameter in a 20-cm fish) surrounded by about
10 smaller ones. A single large cell can also spiral around to overlap
itself to some extent.
The organ discharges are pulses making the dorsal surface negative
(and presumably also making the inside of the mouth positive, although
this has not been directly verified). Responses to handling the fish range
from single pulses up to trains of several tens of pulses at frequencies of
about 5&100/sec. Trains of pulses are also emitted when the fish is capturing
small minnows (Pickens and McFarland, 1964). The pulses are
about 5 msec in duration. The amplitude is somewhat variable, particularly
at the beginning of a train but is up to about 5 V recorded from a
fish 20-30 inches in length with its dorsal surface in air (Fig. 6 ) . The
voltage across the organ is reduced somewhat if the discharge is evoked
while the fish is immersed in seawater. The pulses are generated synchronously
by the two organs.
The effectiveness of the discharge in aiding the capture of prey has
not been demonstrated, but the unstimulated fish emits pulses rarely, if
ever, at other times. Large fish can reach 40 cm in length, and the discharge
is sufficiently strong that it is easily detected when the fish are
handled (Dahlgren and Silvester, 1906). Even a small fish can cause
mild discomfort if one’s hands are quite wet and have a number of
minor cuts (personal observation).
Only the innervated (dorsal) face of the electrocytes is active during
organ discharges (Bennett and Grundfest, 1961b). The responses are
PSPs and are monophasic depolarizations of about the Same shape and
duration as the organ pulses. The innervated membrane does not respond
at all to depolarization but behaves linearly; it is electrically inexcitable.
The response that is evoked by stimuli applied directly to the organ is10. ELECTRIC ORGANS 365
Fig. 6. Patterns of electric organ discharge of Astroscopus. ( A,B) Simultaneous
records from the two organs in each of two different fish. A probe electrode was
placed on the skin over each organ with the dorsal surface of the fish in air. The
reference electrode was on the ventral surface and grounded; dorsal negativity up.
( C ) Same fish as in B, but recording with the animal covered by seawater. (D-F)
Simultaneous recordings as in A but from a fish which had had one of its electric
organs denervated. The electrode on the denervated organ (lower traces) registered
only a small pickup of the activity of the other organ.
mediated by the presynaptic nerve fibers. The uninnervated face is of
very low resistance and generates only a resting potential. (Presumably
it is inexcitable, but this point has not been established critically because
the membrane is of such low resistance that it may not have been
depolarized sufficiently to excite it. However, one doubts that this
membrane would have electrically excitable sites when they would be
nonfunctional and when the innervated face lacks them.) The equivalent
circuit during responses is shown in Fig. 4A. At rest the two faces
generate equal potentials and no current flows. During activity the
resting potential virtually disappears across the innervated face and
the conductance of this face to (presumably) sodium and potassium
greatly increases. Current flows inward across the innervated face and
outward across the uninnervated face. [It is to be expected that the
innervated face actually generates a potential of -10 to -20 mV (see
N. Takeuchi, 1963). Note that the potential in the circuit is provided
by what we have termed the inactive face.]
Monopolar recordings on which this description is based are shown
in Fig. 7. (The straight horizontal line is a reference trace.) A brief
stimulus is applied to the dorsal surface of the organ by means of a
pair of small wire electrodes 1 msec after the start of the sweep. The
recording trace goes off screen for about 0.2 msec during the stimulus366 M. V. L. BENNE’JT
_-
F — A
C
D
F
H
D B
-A E
H G
-‘+I;
I msec
Fig. 7. Activity of a single electrocyte of Astroscopus. Monopolar recording
ill zjizjo. Stimuli are applied by a pair of fine wire electrodes close to the site of
microelectrode penetration. The diagram on the right indicates positions of the
recording electrode in successive records inferred from appearance and disappearance
of resting potential and changes in response amplitude and sign. From
Bennett and Grundfest ( 1961b).
and then returns more slowly toward its initial potential before the
response begins with a latency of about 1 msec. As the electrode penetrates
the cells a regular sequence of potential changes is observed that
serves to identify electrode position as well as to characterize the
responses. Immediately dorsal to the most superficial cell the response
is a monophasic negativity of about 15 mV amplitude ( A ) . The sign
of the potential indicates that the underlying membrane is passing
inward current. As the electrode is advanced the steady potential shifts
about 90 mV negative which represents the resting potential across
the innervated membrane ( B ) . Simultaneously the response becomes
a positive-going or depolarizing response of about 60 mV indicating
that the electrode has crossed an active membrane. If it were differentially
recorded across the innervated face (record B minus record A )
the response would be about 75 mV in amplitude.
As the electrode is further advanced, the steady potential shifts back
to its initial value signaling passage of the electrode through the cell
into the underlying extracellular space ( C ) . The resting potentials
developed by the two faces are equal, and thus no current flows through
the cell at rest. The response recorded outside the innervated face is
virtually identical to that recorded in the cell. This indicates that the
resistance of the uninnervated face is very low compared to the resistance
in the external path. The response recorded differentially across
the entire cell would of course be almost exactly like that recorded
across the innervated face. When the electrode is further advanced,
the steady potential again shifts negative indicating penetration of the
second cell ( D ) . Simultaneously the response amplitude decreases by
about half indicating that the innervated membrane of this cell cornprises
a significant fraction of the resistance of the external path. Further
advances show shifts in the steady potential as the electrode leaves10. ELECTRIC ORGANS 367
and enters cells (E-H). The amplitude decreases as the electrode
crosses innervated faces but not as it crosses uninnervated faces because
of their relatively high and low resistances, respectively.
The neural mediation of these responses is indicated by the latency
which remains about 1 msec even if very suprathreshold stimuli are
used. If spike generating membrane were present, a much shorter
latency could be obtained (an example is given in Fig. 18). Neural
mediation is confirmed by pharmacological data and the effects of
denervation (see below). Responses like those of Fig. 7 result from
single nerve fibers for their amplitude varies in an all-or-none manner
as stimulus strength is varied.
Other experiments involving passage of current during responses
show that the resistance of the innervated faces is decreased during
neurally mediated activity but is not affected by depolarization alone.
Thus during activity the resistance of the innervated face moves toward
that of the uninnervated face in the kind of impedance matching discussed
in the preceding section. The resting resistances of the two faces
correlate with the degree of surface elaboration seen in the fine structural
studies; the lower resistance membrane is much more elaborated
( Mathewson et al., 1961). (The negligible voltage drop across the
innervated faces need not obtain in responses of the entire organ when
all the cells are active in series and much larger currents flow. In organ
discharge there would also be a smaller potential across the innervated
face because of its internal resistance. )
The PSPs of the electrocytes are cholinergic as they presumably are
in the muscle fibers of origin. They are greatly reduced by the blocking
agent curare and prolonged by the anticholinesterase eserine. The cells
are depolarized by acetylcholine, presumably the actual transmitter,
and the related compound carbamylcholine. Denervated cells retain
their ability to respond to these drugs, but the response to electrical
stimulation disappears confirming its neural mediation.
The microelectrode experiments demonstrate that unitary responses
of the electrocytes are essentially of the same duration, shape, and sign
as the organ discharges. If 150 cells in series give rise to a 5 V pulse
about 30 mV per cell is required, a value exceeded somewhat in the
microelectrode experiments. The discrepancy is explicable as a result
of greater current flow during synchronous activity and failure of some
of the cells to fire. Furthermore, cell counts, organ discharges, and PSP
amplitudes may have varied in the different animals used for the
measurements. Although most of the presynaptic fibers are active during
a response (see Section 111), some do fail to fire on occasion as indicated
by the variable response amplitude, and a few inactive cells in series370 M. V. L. BENNETT
Fig. 9. Innervation of torpedinid electrocytes. From the main organ of Narcitie.
(Upper) A single cell teased out from forinalin fixed material and stained with
methylene blue. Four nerve fibers run across the surface from different points on10. ELECTRIC ORGANS 371
would markedly increase the internal resistance of the organ. In spite
of the quantitative discrepancy the principles of series summation and
synchronous activation are well illustrated by this electric organ.
2. TORPEDINIDS
The electric rays are a cosmopolitan group of marine fishes. The
electric organs of the genus Torpedo are diagrammed in Fig. 1. They
are large flattened organs on each side of the head that extend completely
through the “disc” from dorsal to ventral surfaces. The Torpedos
are rather slow moving, and the use of the organ discharge in capture
of much faster swimming prey has been studied in the European
species, Torpedo murmorata (Belbenoit, 1970). When a fish comes near
a Torpedo resting on the bottom, it swims forward and upward and
then emits pulses. Small fishes can be stunned. The Torpedo drops
back to the bottom over any immobilized prey. If it has been successful,
it consumes the prey while continuing to emit pulses at a low frequency.
The delay between movement and discharge indicates that the organ
does not function in prey detection. The detection of prey appears to
be mechanical or perhaps electrical. The effectiveness of the organs in
predation is confirmed for the American species, T. nobilianu, by the
presence of large and rapidly swimming fish in the stomach contents
(Bennett et al., 1961).
Each electric organ in T. nobiliuna is made up of some 500-1000
closely packed and roughly circular columns of electrocytes that run
from top to bottom of the organ (Fig. 8). The electrocytes are thin
(10-30 p ) discs and have the same diameter as the columns; about
1000 are stacked one above another as in a role of coins. The columns
can be quite large in diameter, 5-7 mm in a large Torpedo (1 meter
across ) .
The cells are profusely innervated on their ventral surfaces. Nerves
run in the interstices between columns and 5-7 nerve fibers enter the
space between cells at roughly equal intervals around the periphery;
each fiber innervates a sector of the cell (Fig. 9). The nerves arise
from the electromotor lobe in the medulla (perhaps including parts of
the seventh, ninth and tenth cranial nerves, Fig. 8). Electron microthe
periphery. The fibers branch profusely, but each innervates a separate segment
of the surface. The inset shows schematically the columnar arrangement of the
electrocytes. (Lower) A region from the lower right of the isolated cell is shown
at higher magnification. Nuclei of the electrocytes are stained as well as many fine
nerve branches. Modified from Bennett and Grundfest (1961a) and Grundfest
(1957).372 M. V. L. BENNETT
scopic examination reveals that the innervation is very dense (Fig. 10).
The organ discharges of a large Torpedo are monophasic pulses, positive
on the dorsal surface and about 5 msec in duration and SO V in amplitude
recorded in air (Bennett et al., 1961). The internal resistance of
the organs is low, and the power output at the peak of the pulses can
exceed 1 kW.
Because the electrocytes of Torpedo are very thin, it is difficult to
use two or more electrodes for differential recording across the two
faces and the entire cell. However, a single electrode can be advanced
into a column of cells, and successive changes in resting potential and
response can be used to determine electrode position as described in
Fig. 10. Fine structure of torpedinid electrocytes ( Torpedo murmorutu). The
central portion of the figure shows a perpendicular section through three electrocytes
(EP), dorsal surface upwards. The ventral surface is densely covered with
nerve endings ( n ) . On the left is shown a region of the uninnervated surface at
higher magnification. Several external openings of the many canaliculi are seen ( A ? ) .
The rectangles (Bt) indicate branch points of the canaliculi. Basement membrane
material extends deeply into the canalicular network. On the right is shown a
higher magnification of the innervated surface. Nerve profiles contain vesicles ( v )
as well as small granules ( g ) , possibly glycogen, and lie embedded in the electrocyte.
Occasion folds ( j ) extend quite deeply into the postsynaptic cell. From
Sheridan et al. (1966).10. ELECTRIC ORGANS 373
respect to Astroscopus (Fig. 7). The responses of the electrocytes of
Torpedo are essentially the same as in Astroscopus (Bennett et d., 1961).
The active face is the innervated face, which however is on the ventral
side. Again, this face does not respond to depolarization; it is electrically
inexcitable. The response that is evoked by external stimulation is a
monophasic depolarization, that is, a PSP mediated by activity of the
presynaptic fibers. Its origin as a neurally mediated response is attested
to by the irreducible delay when stimulating with closely applied electrodes
and by the block of the response when curare is applied to the
innervated face. In a column of tissue the response can be inverted
by polarizing currents [the first tissue in which PSPs were shown to
invert ( Albe-Fessard, 1951; Bennett et al., 196l)l. The uninnervated
face is of very low resistance and generates only a resting potential.
The area of this face is greatly increased by numerous invaginating
canaliculi ( Fig. 10).
The maximum response amplitude observed in a single cell is about
90 mV, which is somewhat larger than the recorded resting potentials.
However, the cells are very thin and the canaliculi extend across at
least half the cell thickness; a supposedly intracellular electrode might
always be partly in the extracellular space outside the uninnervated
face. Probably the full resting potentials are not recorded and like other
PSPs those in Torpedo do not overshoot the resting potential. The duration
of the PSPs is about 5 msec. Their latency with locally applied
stimuli is 2-3 msec, an appreciable fraction of which may be conduction
time in fine branches of the innervating fibers.
The PSPs in Torpedo electrocytes are cholinergic in that they are
blocked by curare and dihydro-P-erythroidine and prolonged by the
cholinesterase inhibitors eserine and physostigmine. Also, the cells are
depolarized by actylcholine and carbamylcholine.
The synthesis of organ discharge from the responses of the single
electrocytes simply requires synchronous activation as in Astroscopus.
The duration of the PSPs is about the same as that of the organ discharge,
and the amplitude of the PSPs is sufficient that the number of
cells in series could produce the discharge amplitude.
Narcine is a smaller relative of Torpedo. I t has a bilateral electric
organ, the “main” organ, that resembles that of Torpedo both morphologically
and physiologically ( Bennett and Grundfest, 1961a). Nurcine
has an additional (bilateral) electric organ, the accessory organ, that
lies at the posterior margin of the main organ (Mathewson et al., 1958).
This organ differs in one important respect from the main organ. The
responses to low frequency stimuli are very small, but when the nerve
is stimulated at up to 100/sec, the responses are considerably augmented374 M. V. L. BENNETT
and prolonged. However they are still somewhat smaller than in the
main organ of Torpedo. Although the normal discharge of the accessory
organ is unknown, these findings suggest that the discharge involves
repetitive and fused responses of the single cells similar to what is
observed in the rajids (see the next section), It seems likely that the
accessory organ is used either in an active electrosensory system or in
communication, although there is no direct evidence for this suggestion.
The torpedinids like most other elasmobranchs have ampullae of Lorenzini,
which are electroreceptors (see Chapter 11, this volume). A
number of other small torpedinids are known, some of which live in
the deep seas and are blind (Lissmann, 1958). Possibly their main
electric organs are used in electrolocation. It is not known whether
these species have accessory organs.
3. RAJIDS
The skates or ordinary rays are a large cosmopolitan group of marine
fish comprising six or more genera and many species. They are weakly
electric, but unlike freshwater species they emit pulses only infrequently.
The discharges are sufficiently inconspicuous that a major taxonomic
work on the group “Fishes of the Western North Atlantic” (Bigelow and
Schroeder, 1953) makes no mention of the fact that these fish are electric.
Other rays (suborder Myliobatoidea ) apparently lack electric organs.
The electric organs of rajids are located in the tail in the center of
the most lateral bundle of longitudinally running muscle fibers (Fig. 1).
The organs are spindle-shaped and run most of the length of the tail.
They are much greater in length than in diameter. The electrocytes are
oriented anteroposteriorly and are innervated on their anterior faces.
Each cell lies in a small connective tissue compartment. Two types of
electrocytes have been described, the cup-shaped and the disc-shaped
(Fig. 11). However, these terms are not particularly descriptive of the
morphological differences. Cup-shaped cells lie at the anterior margin
of their connective tissue chamber. Often they are convex posteriorly,
which accounts for their name. Both faces are relatively smooth at the
light microscopic level of resolution. Electron microscopy reveals a
relatively small number of tubules invaginating into the innervated
face and a somewhat greater number in the uninnervated face (Mathewson
et al., 1961). Disc-shaped cells lie nearer the posterior of their
chambers. The posterior, uninnervated faces have a large number of
protuberances tens of microns in diameter and length (Fig. 11). Probably
there are more invaginating tubules in these faces than in cupshaped
cells. Both classes of cells contain striated filamentous material10. ELECTRIC ORGANS 375
Fig. 11. Anatomy of rajid electrocytes. Longitudinal sections of electric organ
from R. erinacea, which has cup-type cells (A,C) and from R. eglanturia, which
has disc-type cells (B,D). Rostra1 surface up, hematoxylin and eosin stain, A,B,
low power; C,D, high power. Cup-type cells are usually smooth on both surfaces
( A ) although there may be a few processes on the caudal face in some cells (A,
lower right). The cells lie against the caudal walls of the connective tissue chambers
( c ) that contain them. The caudal face of disc-type cells usually has many processes
although in some regions it may be relatively smooth (B, lower right). The
cells lie anteriorly in their connective tissue chambers ( c ) . Both types of cell contain
striated material in a central area (C,D). There is often a short process or stalk ( s )
that is a remnant of the muscle fiber from which the cell develops (Ewart, 1892).
From Bennett ( 1961).
(Fig. 11, Wachtel, 1964) that reveals their myogenic origin.
The response properties of the two kinds of cell are somewhat
different. The disc-shaped cells are physiologically similar to those of
other marine electric fish, but the cup-shaped cells are more complex
(Bennett, 1961). Some intergrading of the two extreme physiological
types apparently does occur, but too few species have been studied
to be sure of the extent of the correlation between form and physiological
functioning. There is evidence that the differences between cup- and
disc-shaped electrocytes are associated with other morphological characteristics
and that the rajids can be divided into two groups (Ishiyama,
1958).
The organ discharges are monophasic and head negative. The fish376 M. V. L. BENNETT
can only sometimes be provoked into discharging by mechanical stimulation.
Under these conditions the organ discharges are usually irregular
and variable in size and duration (Fig. 12). The maximum pulse amplitude
is of the order of a volt recorded in air, but tens of millivolts if
the fish is in seawater (D). Sometimes more regular pulses are emitted
(A), and preliminary results with long-term recording from animals
in holding tanks suggest that “spontaneous” discharges or those evoked
by light touch under these conditions are more constant in amplitude
(A, B. Steinbach, unpublished data). In any case the discharges that
have been observed are long lasting compared to the responses of individual
electrocytes, and the discharges most probably involve fused
repetitive activity of many cells.
Unlike the electrocytes of strongly electric marine fish, cup-shaped
electrocytes respond to depolarization. However, the response involves
only delayed rectification. There is no sodium activation or other
electrically excitable component leading to a regenerative response;
the response to graded depolarizing pulses is graded. An example is
shown in Fig. 13. For small depolarizing (outward) currents the cell
behaves linearly with the same resistance as for hyperpolarizing current.
The resulting potentials rise slowly. and more or less exponentially to
a steady state value; the slowness of rise results from charging of the
membrane capacity. For larger depolarizing currents there is an early
peak of depolarization which then decreases as the rectification “turns
on.” The peak depolarization is always less than the steady state poten-
Fig. 12. Discharge of rajid electric organs. (A-D) Raja erinacea, which has
cup-type electrocytes. ( A‘,B’ ) Raja eglantaria, which has disc-type electrocytes.
Discharges are evoked by vigorous prodding and recorded differentially between
tip and base of tail, caudal positivity up. Upper trace, higher gain. In both forms,
discharge is asynchronous and variable in amplitude and duration. All records are
in air except D, for which the animal is immersed in seawater while regularly
responding as in B. The discharge is greatly reduced in amplitude. From Bennett
(1961).10. ELECTRIC ORGANS
Peak
.’.x . . +J
Rqa
20 mV
377
Fig. 13. Intracellularly applied polarization of cup-shaped electrocytes from
R. eriinacea. ( A ) Superimposed records of responses (lower traces) to depolarizing
currents (upper traces). For larger currents, the voltage reaches an initial peak,
then falls to a much lower steady value. (B) Records as in A in response to
hyperpolarizing currents. The voltage change increases approximately exponentially
toward a steady level. Graph: voltage-current relationship for the same cell as A
and B, but using additional data. The relation is linear for hyperpolarizing current.
For larger depolarizing currents, the initial peaks fall somewhat below the potentials
they would have reached if the cell had the same resistance as for hyperpolarizing
current. The potentials at the end of the pulses are much lower and continue to
decrease as further current is applied. Modified from Bennett (1961).
tial would be that corresponded to the resistance for small depolarizing
and hyperpolarizing currents. If a current pulse is terminated on the
rising phase of the initial depolarization, the potential immediately
begins to return toward the base line; there is no tendency of the
potential to continue in the depolarizing direction (Fig. 14). These
findings lead to the conclusion that there is no regenerative component
in the response. The increase in conductance is indicated not only by
the reduction in potential during the current but also by the more rapid
drop in potential following cessation of the current after the conductance
increase has been produced ( Fig. 14). The conductance increase caused
by a brief stimulus lasts a few tenths of a second. It is associated with
a small depolarization from the resting potential (Fig. 14D). The nature
of the permeability change underlying the conductance increase is unclear.
There is suggestive but incomplete evidence that it may be an
increase in C1 permeability ( Bennett, 1961; Grundfest, 1967).373 M. V. L. BENNETT
Fig. 14. Responses of cup-shaped electrocytes to depolarizing pulses of
different strengths and durations ( R. erinacea ). Lower trace: intracellular voltage.
Upper trace: current applied through second intracellular electrode. ( A-C ) Superimposed
records of constant current pulses of different durations, magnitude
increasing from A to C. ( D ) Superimposed records of a large and a small pulse
of the same length; the traces cross following the end of the pulses. Whatever the
current strength or duration, the voltage starts to decrease immediately on cessation
of the current. Following the peaks, the voltages decrease more rapidly than
if the pulses are stopped before them. The time course of the decrease deviates
from exponential because the cells are not isopotential. In D a small depolarization
is seen to be maintained following the larger stimulus. From Bennett (1961).
When the nerve supply to an electrocyte is stimulated, a PSP is produced
that is graded in several all-or-none steps indicating that the
cell is innervated by several fibers (Fig. 15C). The PSPs are cholinergic
(Brock and Eccles, 1958). The larger PSPs activate the delayed rectification,
and these PSPs decay more rapidly because of the shortened
time constant of the cell.
In most cells the delayed rectification is found primarily in the uninnervated
face; it thereby acts to increase external current flow in the
medium around the electrocytes. At rest the resistance of innervated
and uninnervated faces is relatively high. Depolarization by PSPs
generated in the innervated face increases the conductance of the uninnervated
face and thus allows more current to flow out through this
face and around the cell. (Delayed rectification in the innervated face
is maladaptive; it increases microscopic current loops in this face and
by loading the PSP generating membrane tends to reduce current flow
around the cell.)
The effect of the delayed rectification on external potentials can be
illustrated with reference to Figs. 4B and 15. The external response
to a small PSP that does not turn on the delayed rectification is a singlepeaked
potential which is positive outside the uninnervated face (Fig.
15C ) . However, the peak of the external potential is earlier than that of10. ELECTRIC ORGANS 379
Fig. 15. Effect of delayed rectification on PSPs of cup-shaped electrocytes,
R. erinacea. Recording and stimulating as in the diagram except that nerve stimulating
electrodes are omitted. Upper trace: intracellular stimulating current. Second
trace: potential across innervated face ( V , ) . Third trace: potential across uninnervated
face ( V2). Fourth trace: potential across cell, posterior positivity up
(Vx). A large PSP produces a posterior positive external potential that has two
peaks ( A ) . The second peak is blocked when the conductance increase is prevented
by hyperpolarization ( B ) that also prolongs and increases the height of the response.
Small PSPs that do not activate the delayed rectification also produce external
potentials with a single peak (C, superimposed traces of large and small responses).
When the delayed rectification is activated by a depolarizing pulse and a PSP is
evoked, the external potential is increased and the potentials across the faces are
reduced ( D, superimposed responses with and without applied depolarizing pulse).
The time constant of the uninnervated membrane is reduced so that all the potentials
have the same time course and the peak of the external potential occurs later than
the initial peak of responses when the delayed rectification has not been activated.
From Bennett ( 1961).
the potential across the uninnervated face, a result that indicates that
there is a capacitative component of the current. During the rising phase
of the PSP the capacity of the uninnervated face is being charged and
more capacitative current is flowing. At the peak of the potential no
capacitative current is flowing and the current is entirely resistive. When
a large PSP activates the delayed rectification, the initial part of the
external record is of the same shape as that during small PSPs. However,
after this period the delayed rectification turns on, outward current
through the uninnervated face increases, and the external potential
rises to a second peak (Fig. 15A,C). If the cell is hyperpolarized so
that even a large PSP fails to activate the delayed rectification, the
external response again has only a single peak that is earlier than the
peaks of the transmembrane potentials (Fig. 15B). Note that in B the380 M. V. L. BENNETT
transmembrane potentials are slowly falling as they are for small PSPs.
Moreover, hyperpolarization increases the amplitudes of the potentials
because the PSP conductance change is unaffected while the driving
force, the difference between the membrane potential and the PSP
reversal potential, is augmented (see Bennett, 1961). If a large PSP
is evoked after a depolarizing pulse that activates the delayed rectification,
the external potential is enlarged, the voltage drop across the
external response has only a single peak that occurs at nearly the same
time as the peak of the transmembrane potentials (Fig. 150). In addition,
the external potential is enlarged, the voltage drop across the
uninnervated face is reduced, and the PSP across the innervated face
is also reduced because of the greater electrical load placed on it. A
PSP that turns on the conductance increase causes the same changes
in subsequent responses as does directly applied depolarization.
The utilization of the delayed rectification to reduce the resistance
of the innervated face is an interesting adaptation. It allows the fish to
maintain a high resting resistance but to achieve a large external current
during activity. A possible disadvantage to this mechanism is that the
increase in conductance is delayed, a property that would not appear
to be very significant in an organ discharge that involves fused and
presumably repetitive responses. The earliest PSPs would cause the
conductance increase which would remain activated during the later
PSPs. The double peak of the initial external responses of single cells
is not seen in organ discharges although there may be an inflection on
the rising phases (Fig. 12A). The difference is undoubtedly a question
of synchronization because if electric stimuli are used to evoke synchronous
PSPs in lengths of organ the external responses have the same
shape as the responses of single cells.
D. Freshwater Electric Fish
1. GYMNOTIDS
The gymnotids are a diverse group of fish living in fresh waters of
tropical South America. They are often divided into six families (Table
I ) , but the relationship between the families is obvious from the
similarities in body shape and in many other characteristics as well
as from the possession of electric organs (Fig. 1). This group includes
the electric eel, probably the best known electric fish, and a moderate
number of other species that have only weakly electric organs. Because
of their characteristic elongate shape, the weakly electric gymnotids10. ELECTRIC ORGANS 381
are commonly called knife fish. The greater part of the length of the
fish contains only muscle, spinal column, and electric organ. The viscera
are located in the anterior end of the body, and the anus and genital
papilla are just behind the chin. The anal fin begins slightly more
posteriorly and runs to near the caudal end of the fish. The electric
organ runs along most or all of the length of the body and is innervated
by spinal nerves. A caudal fin is absent in all but the sternarchids, in
which it is greatly reduced. There is little muscle caudal to the anal
fin, and this region, sometimes called the caudal filament, contains
mostly electric organ.
Gymnotids can regenerate new posterior regions if part is removed
(Ellis, 1913). Often in fish taken from the wild the posterior is regenerated
indicating that there were encounters with predators in
which only the front end of the fish escaped. Ordinary swimming movements
of the gymnotids are quite different from those of almost all
other fish. The body is held straight, and waves along the anal fin
provide the propulsive force. Most species appear to swim about equally
well forward and backward and often investigate objects by swimming
backward toward them (see Gymnarchus, Section 11, C, 3 ) . Most of
the body musculature is not involved in movement of the anal fin and
is probably used only in emergency movements when the body is rapidly
flexed in the more usual fishlike manner.
The species of gymnotids are poorly described. Keys are available
for a few areas of South America, but it is doubtful that these are
complete. Tropical fish dealers, the usual source of supply of gymnotids,
often bring in fish from other areas that are clearly new species. On
the Rio Negro expedition of the R. V. Alpha Helix, perhaps one-third
of the gymnotid species collected were undescribed taxomically. While
there are difficulties in working on undescribed species, classification
into genera is generally simple, and generic characteristics are apparently
the most important ones from a physiological point of view. Nevertheless,
it may be desirable to save specimens of a particular species studied
for subsequent identification when the systematics of gymnotids is
clarified.
Table I1 and Fig. 16 present a key to gymnotids that is modified
from that of Ellis (1913). A second system has been added, one that
is nondestructive and applicable to living specimens. Accurate classification
sometimes requires dissection of the specimen, but a reasonable
identification can usually be achieved without it, particularly if an
oscilloscope is available to observe electric organ discharges.
a. The Electric Eel. The electric eel was the first electric fish for which
the cellular mechanisms of the discharge were elucidated (Keynes and382
Table I1
Key to Families and Genera of Gymnotids
M. V. L. BENNETT
The gymnotids are recognizable from their elongate body and anal fin; the pectoral
fills are small; the caudal is present only in the sternarchids in which it is very small;
other fins are absent. There are two African species, Notopterus ofcr and Xenomystus
nigri, that are also commonly known as knife fish. They are similar in appearance to
Some weakly electric gymnotids, and confusion might arise if place of origin were unknown.
African knife fish are easily distinguished by the short “tentacle” from the anterior
naris and the anal fin that extends a short distance around the tip of the tail to form
a false caudal fin (as is also true of the electric eel). Notoptcrurus has a small dorsal fin.
I n the following key the species name is given, if only a single species has been described.a
a. Head flattened dorsoventrally, lower jaw projects beyond upper so that the large
mouth opens somewhat upwards
b. Strongly electric; electric organ occupies much of the body caudal to the
abdominal cavity; body not scaled; anal fin extends around the end of the tail
to form a false caudal. Electrophoridae. Electrophorus elcctricus
bb. Weakly electric; scaled; slender cylindrical tail extends beyond anal fin. Gymnotidae.
Gymnotus carapo
%a. Head round in cross section or compressed laterally, lower jaw projecting little if
c. Caudal fin absent; tail beyond anal fin slender and usually cylindrical; no dorsal
any, mouth small or large
filament
d. Snout short, Sternopygidae6
e. Teeth in both jaws, color uniform or with longitudinal stripes
f. Color uniformly dark except sometimes a lighter stripe along the posterior
lateral line; body compressed laterally; orbital margin free, i.e., there
is a distinct and deep cleft between the eye ball and adjacent skin;
posterior air bladder long and conical; organ discharge more or less
sinusoidal a t a frequency of 50-150/sec. Sternopygus
ff. Color light; fairly transparent to pale yellow, often with several darker
longitudinal stripes; body compressed laterally; eye covered by a thin
membrane; posterior air bladder small, nearly spherical; organ discharge
more or less sinusoidal a t a frequency of 250-600/sec. Eigcnmannia
fff. Very long caudal filament, dorsal profile of head concave. Rabdolichops
longicaudatusc
ee. Teeth absent, color dark but patterned; brown to black mottled or with
slightly diagonal banding; organ discharge consists of brief pulses separated
by much longer intervals
g. Body compressed laterally, depth increases from posterior of head to
shortly after beginning of anal fin, then decreases (Fig. 16); profile
rounded; accessory organs in head region (Fig. 32) ; lies on side when resting
on a flat surface. Steatogenys elegansd
gg. Body less compressed, depth near greatest a t posterior of head (Fig. 16);
profile shows a protruding mouth; no rostra1 accessory organs; rarely
lies on side. Hypopomus
dd. Snout long and tubular. Rhamphichthyidae
g. Body entirely scaled. Rhamphichthys rostratusC
gg. Sides not scaled in anterior region. Gymnorhamphichthys hypostomus
cc. Caudal fin present but quite small; dorsal filament (see Fig. 42) closely adherent
to back but may be separated in fixed specimens; organ discharge frequency
700/sec or more. Sternarchidaee10. ELECTRIC ORGANS 383
h. Snout long, tubular, and down curving. Sternarchorhynchus oryrhynchusf
hh. Snout long, tubular, horizontal and straight
i. Mouth large, opens a t least one-third the distance back to the level of the
ii. Mouth small, opens less than one-sixth the distance back to the level of the
eye; upper profile of head markedly convex. Sternarchorhamphus
eye, upper profile quite straight. Orthosternarchus tamandua“
hhh. Snout not long unless mouth is very large
j. Teeth present in both jaws; some located externally on what appear to be
swollen lips. Oedemognathus exodon
k. Dorsal region of body virtually entirely scaled from head posteriorly to
1. Mouth large, its angle extending a t least as far posteriorly as the
11. Mouth small, its angle extending no farther posteriorly than thepostekk.
Scales absent from much of the body above the lateral line anterior to
the dorsal filament (shaded in Fig. 16); scales near lateral line much
larger than more dorsally. Porotergus
jj. Teeth present in both jaws but inside mouth
origin of dorsal filament
anterior margin of the eye. Sternarchus
rior naris. Sternarchella
jjj. Teeth absent from upper jaw
m. Lower jaw toothless and with a distinct midline groove into which the
beaklike upper jaw fits; middorsal region of body scaled anteriorly.
Adontosternarchus
mm. Lower jaw sometimes toothed and fitting into a midline groove in the
upper jaw; middorsal region of body naked anteriorly. Sternarchogiton
a Modified from Ellis (1913), Eigenmann and Allen (1942), and Gery and Vu-Tbn-
Tu& (1964) who should be consulted for more extensive descriptions.
Sternopygus, Eigcnmannia, Hypopornus, and Steatogenys are normally grouped together
as the family Sternopygidae. Properties of the electrocytes and their innervation
indicates that Hypopomus and Steatogenys are more closely affiliated to the Rhamphichthyidae.
Greenwood et al. (1966) observe “Peculiar specializations in [the gymnotids]
are many, and a complete study may alter the family arrangement accepted here.”
The author has not seen members of these genera nor are there any physiological
data from them.
Only one species is described in the taxonomic literature, but there is a second species
described here that has somewhat different rostra1 electric organ.
There is a proposed revision of Sternarchidae and Sternarchus to Apteronotidae and
Aptcronotus on grounds of priority (1800 vs. 1801). As Sternarchus has been used for a
long time and provides the basis of many other generic names in the family, the author
stands with Ellis (1913) who uses this name and derivatives.
f Although this genus has been considered to be monospecific, two apparent species
of each were found on the Rio Negro expedition of the Alpha Helix.
Martins-Ferreira, 1953; Altamirano et al., 1953). The eel emits two
classes of pulses, small pulses about 10 V in amplitude and large pulses
some 500 V or more in amplitude. All the pulses are monophasic, head
positive, and about 2 msec in duration. The voltage increases with the
length of the fish, which can exceed 1.5 meters, but even a baby 7-10
cm long can emit close to 1OOV. When the animal is resting, the small10. ELECTRIC ORGANS 385
pulses are emitted at quite low frequencies down to a few per minute.
When actively swimming, the pulse rate is usually increased to about
30 or more per second. Large pulses are emitted only in high frequency
bursts at a frequency of several hundred per second. The first pulse
in such a burst is small, but the amplitude increases to the maximum
within 2 or 3 pulses ( Albe-Fessard, 1950a). The weak pulses are implicated
in the electrosensory system (see Chapter 11, this volume); the
strong pulses have offensive or defensive value. The emission of moderate
to high frequency pulses by one eel attracts other eels, and they also
increase their rates of discharge (Bullock, 1970).
The electric tissue occupies much of the cross section of the body
in the posterior three-fourths of the fish, and it is divided into three
(bilateral) organs. There is axial musculature dorsal to the organs and
a smaller amount of muscle ventrally that controls the anal fin (Fig. 17).
In the dorsal and posterior region is the organ of Sachs, and in the
ventral and anterior region is the main organ (Fig. 1). Beneath these
organs is Hunter’s organ. The electrocytes are more widely separated
and somewhat larger in Sachs’ organ than in the main organ, but otherwise
the two organs are similar (Luft, 1957; Couceiro and Akerman,
1948). The main organ generates the greater part of the high voltage
discharge used offensively or defensively. The organ of Sachs generates
the greater part of the low voltage pulses involved in the electrosensory
system but also contributes in a minor way to the high voltage discharge.
Hunter’s organ apparently functions like main organ anteriorly and
Sachs’ organ posteriorly ( Albe-Fessard and Chagas, 1954).
The single electrocytes are flattened in the anterior posterior axis
(Fig. 17). They are ribbon-shaped and tend to run from the medial
septum to the lateral margin of the fish. In an adult there are dorsoventrally
about 25 in the main organ and 10 in Hunter’s organ. There
are some 6000 in the anteroposterior axis ( Albe-Fessard, 1950a). Each
cell is contained in a connective tissue chamber, and successive layers
of cells in the anterior posterior direction are fairly accurately aligned,
one behind the other. The alignment is somewhat less good in the
dorsoventral direction.
The electrocytes of the main organ occupy perhaps one-fifth to onehalf
the chamber volume; the fraction is smaller in Sachs’ organ (Luft,
1957; Couceiro and Akerman, 1948). The cells are innervated on their
posterior faces by spinal nerves. The posterior faces have a moderate
number of short papilli or stalks protruding from them that increase
the surface area; there are also some tubules in this face (Fig. 17). The
innervation is primarily on the stalks and is much less dense than in
electrocytes of torpedinids. The anterior faces have a large number ofA
Rostra1 Caudal
Spinal cord
I Swim. bladder
Fig. 17. Structure of eel electric organ, ( A ) Diagram of gross morphology of the organ showing series arrangement
of electrocytes (“plaques”) on the left and parallel arrangement on the right. From Altamirano d al. (1953). ( B ) Light
micrograph of a section through a single electrocyte ( e ) . The caudal, innervated surface (on the left) has some small
processes from it. The innervation appears quite sparse and few nerve fibers ( n ) are visible. Relatively stout processes
come off the anterior, uninnervated face and there is a pronounced layer of increased density associated with the mem- 5
brane. (C) Low and high (inset) magnification electron micrographs of the uninnervated surface showing the extensive 5
canalicular network responsible for the density associated with this face seen in B. The canaliculi open to the exterior r
(arrows) and branch profusely. Basement membrane material extends into the canaliculi, ( D ) High magnification m
electron micrograph of the innervated face showing a vesicle filled nerve terminal (n). The surface proliferation provided
by canaliculi (openings indicated by arrows) is much smaller at this face. Magnification for inset in C same as
that for D. Micrographs provided through the courtesy (and skill) or Drs. F. E. Bloom and R. Barnett. H
r-1388 M. V. L. BENNETT
papilli and an extensive network of small canaliculi that greatly increase
the surface area compared to the area that a simple planar membrane
would have (Fig. 17, see also Mathewson et al., 1961).
The properties of eel electrocytes are only slightly more complex
than those of electrocytes of the stargazer and Torpedo. The innervated
face responds to depolarization by generating a spike that overshoots
the resting potential by about 50 mV. The uninnervated face is of very
low resistance and does not become excited. The responses of a single
cell are shown in Fig. 18. Current is applied by external electrodes in
order to depolarize the innervated face, and differential recording between
two microelectrodes is used. In Fig. 18A-C one electrode is
advanced into the cell and then into the underlying extracellular space.
When the “exploring” electrode is in the cell the internally negative
resting potential of -90 mV is recorded across the innervated face
(Fig. 18B). An adequate stimulus evokes a spike that overshoots the
zero potential by about 50 mV. The response arises from the depolariza-
Fig. 18. Responses of electrocytes of the electric eel. Right: recording differentially
as one electrode is advanced through a cell; positivity of this electrode
shown upward. Current passed through the cell by external electrodes in order
to depolarize the innervated face. (A) Both electrodes external to the innervated
face; no response is seen (there is a brief diphasic stimulus artifact). ( B ) One
electrode is advanced into the cell. The inside negative resting potential of about
90 mV and an overshooting action potential about 140 mV in amplitude are
recorded. ( C ) When the exploring electrode is advanced to outside the uninnervated
face, the resting potential disappears, but the spike is essentially
unchanged. From Keynes and Martins-Ferreira ( 1953). (E-F) Differential recording
across the innervated face, the upper trace shows the zero potential. Stimuli
hyperpolarizing the innervated face can evoke PSPs ( E ) that arise after a latency
of 1-2 msec and that if sufficiently large initiate a spike ( F ) . From Altqmirano
et al. ( 1955). Calibrations are the same for A-F.10. ELECTRIC ORGANS 389
tion produced by the stimulus; the delay can be made much shorter
if a stronger stimulus is given. When the exploring electrode is advanced
out through the uninnervated face of the cell, the resting potential
disappears demonstrating that the resting potentials across the two faces
are equal (Fig, 1SC). The response amplitude however does not
change. As discussed in respect to electrocytes of Astroscopus this
finding indicates that the resistance of the uninnervated face is very
low. Direct measurements using current pulses confirm its low resistance,
which is much lower than that of the innervated face. The
degree of surface elaboration seen cytologically correlates with the
different resistances.
Stimulation of the nerve supply evokes PSPs that can depolarize the
innervated membrane to the point where it generates a spike. Current
applied as in Fig. 18A-C but in the opposite direction hyperpolarizes
the innervated face and does not excite it. Such stimuli can excite nerve
fibers in the tissue that then produce PSPs that arise after a delay of
about 2 msec (Fig. 180). If enough nerve fibers are stimulated the
PSP initiates a spike (Fig. 18E).
The ionic basis of the action potential has been well studied in electrocytes
of the eel. The cells are depolarized by high potassium solutions
approximately as predicted by the Nernst relation, which indicates
that at rest the potassium permeability of the cell predominates and
the potential is largely determined by the intra- and extracellular concentrations
of potassium (Higman et al., 1964). The inward current
responsible for the rising phase of the spike is sodium dependent
( Keynes and Martins-Ferreira, 1953). Furthermore, it is eliminated by
the pharmacological agent tetrodotoxin which is a specific blocking
agent for the increase in sodium permeability produced by depolarization
(Nakamura et al., 1965).
Unlike the squid axon and many other tissues, the eel electrocytes
lack K activation or delayed rectification, but they do have anomalous
rectification in the innervated face (Nakamura et al., 1965). Actually,
it makes sense for the cells to have anomalous rectification and not to
have K+ activation. For maximum effectiveness as an electric organ, the
circuit for all the current carried inward by Na’ should be completed by
current in the external environment, not by local currents in the innervated
face. The time constant of the cells is sufficiently short that
the membranc rapidly returns to the resting potential without the
restoring effect of delayed rectification (as is also very nearly true of
myelinated nerve fibers, see Frankenhaeuser and Huxley, 1964). The
anomalous rectification decreases the conductance enough to result in
a severalfold decrease in eddy currents in the innervated face,Fig. 19. Electric organ of Gymnotus. (A) Cross section at two magnifications
of region 2 cm from the tip of the tail of a fish about 10 cm long; higher power
view on right of the region indicated. The connective tissue tubes enclosing the
four columns of electrocytes are numbered from dorsal to ventral; the section
grazes electrocytes of column 1 and passes through nearly the maximum diameter
of cells of column 3. The spaces dorsal to the tubes are fixation artifacts. At this
level most of the body cross section is occupied by muscle. The vertebral column
and spinal cord ( c ) are seen. ( B ) Photographs of a dissected unstained preparation.
The cells and their main innervation are outlined on the right-hand copy, The
caudal innervation of the most dorsal cells is also visible. Electrocytes of all four
tubes are numbered. Modified from Bennett and Grundfest (1959).10. ELECTRIC ORGANS 391
The excitability properties of the innervated face lead to an unusual
sequence of conductance changes during the response (Morlock et al.,
1969). At the start of the spike, depolarization causes sodium activation
and increased conductance and further depolarization; depolarization
also decreases the conductance of anomalously rectifying channels, but
the net result is increased conductance, Sodium inactivation ensues,
and the total conductance decreases to below the resting value while
the membrane potential falls from the spike peak; at a sufficiently low
potential anomalous rectification reverses and conductance rises to the
resting level. This diphasic sequence of conductance changes, increase
followed by decrease, contrasts to the monophasic increase seen in
nerve (Cole and Curtis, 1939; Tasaki and Freygang, 1955).
One might ask why the innervated membrane should not just have
a high resting resistance instead of anomalous rectification. One possible
reason is to provide a conductance in inactive cells for flow of current
generated by active cells in series with them. As will be discussed further
in Section 111, A, the cells are not all active at the same time except in the
largest discharges. The voltage of these responses is accounted for by
synchronous activity of some 6000 cells in series each producing over
100 mV.
b. Gymnotus. This species is weakly electric; its organ discharge is a
fraction of a volt recorded in water and not much more than a volt
recorded in air. Its body shape is similar to that of the eel, but the electric
organ is much smaller (Fig. 1). When undisturbed and resting it normally
emits pulses at about 50/sec. The pulses are approximately triphasic,
initially head negative, and about 1 msec in duration (Fig. 3B,
B’) . Mechanical stimuli, light, electric fields, and resistance changes can
cause moderate transient accelerations of the discharge, and when the
animal is feeding, the discharge frequency can briefly exceed 200/sec
( Fig. 3, Lissmann, 1958; Bennett and Grundfest, 1959; Black-Cleworth,
1970). When swimming around its tank the fish maintains a fairly steady
frequency somewhat above the resting level. Gymnotus is also capable
of ceasing its discharge completely for brief periods, a response that may
sometimes represent hiding or “listening.” Both accelerations and cessations
can be involved in communication between other members of the
same species and a relatively potent stimulus for inducing cessation of
firing is weak electric pulses at a frequency close to that of the organ
discharge.
The electric organ runs longitudinally from just behind the chin to the
tip of the caudal filament. The organ lies dorsal to the anal fin but
extends beyond it rostrally to the cleithrum as well as caudally (Fig. 1).
The organ on one side consists of about four longitudinal columns of392 M. V. L. BENNETT
drum-shaped cells, the flat faces of which are oriented anteroposteriorly
(Fig. 19). The cells are about a millimeter in diameter and 300p thick
in a fish 20 cm long. The number of columns is reduced anteriorly, and
there may be several additional columns caudally. In each column there
is one electrocyte per segment and there are about 90 segments in the
animal. Each column of cells is enclosed in a connective tissue tube that
is divided into chambers by loose septa between the cells. The cells of
all but the most dorsal column are innervated by a number of fibers on
their posterior faces. Aside from innervation the two faces are very
similar. They are quite smooth with relatively few inpocketings and
canaliculi (Schwartz et al., 1971). The cells of the most dorsal column
have their main innervation on the anterior face, but they have a few
fibers ending on their posterior faces as well (Szabo, 1961d). As will be
shown below, these cells behave physiologically like the more ventral
cells but are oriented in the opposite direction. No obvious function of
the posterior innervation was observed in the early physiological study of
these cells (Bennett and Grundfest, 1959), but since the presence of the
posterior innervation was not rwognized at that time the question could
well be reinvestigated.
The single electrocytes of Gymnotus generate external potentials
that are diphasic (Fig. 20) in contrast to the monophasic discharges of
the eel. This form of potential is produced because both faces generate
spikes. The lower threshold face is the posterior, innervated face in all
but the dorsal column of cells in which it is the anterior face. When the
nerve supply is activated or when current is applied by an intracellular
electrode, the lower threshold, innervated face fires first. Current flows
inward through this face and outward through the opposite, uninnervated
face which becomes excited, but with some delay with respect to
firing of the innervated face. By this time the spike of the lower threshold
(innervated) face is decreasing, and current flows in the reverse direction
along the axis of the cell. This pattern of activity is indicated by the
external recordings shown in Fig. 20. The monopolarly recorded external
potentials are of opposite sign outside the two faces. External to the
innervated face, the potential is negative during PSPs, and when a spike
arises, goes rapidly more negative. During the later part of the monopolarly
recorded intracellular spike, the potential external to the innervated
face reverses to go positive, indicating that the uninnervated face
has a larger potential across it than the innervated face. The potential
outside the uninnervated face has the same shape and about the same
amplitude as that outside the innervated face but is opposite in sign.
External to the edges of the cell the potentials are very small. This
feature and the opposite polarity of potentials external to the two faces10. ELECTRIC ORGANS 393
I msec
Fig. 20. Responses of electrocytes of dorsal and ventral columns (Gymnotus).
( A ) Records from an electrocyte of the most dorsal column. ( B ) Records from a
cell of the third column ventrally. Monopolar recordings outside rostral face (upper
traces ) , caudal face ( middle traces ) , and intracellularly ( lower traces ) . ( AI,Bl )
Weak stimulation of the nerves evokes a depolarizing PSP which is associated
with negativity external to the rostral face in A,, and external to the caudal face
in B,, reflecting the different innervation of the two classes of cell. (A,,B2) Stronger
stimuli evoke similar intracellular spikes in the two cells, but the diphasic external
potential is initially positive outside the rostral face in A, and outside the caudal
face in Bz. (AI,BS) Current passed through the intracellular electrode evokes an
external response that is initially negative outside the innervated face in each case
indicating that this is the lower threshold membrane. From Bennett and Grundfest
( 1959).
demonstrate the longitudinal orientation of the cell. ( I t should be noted
that there will be some local circuit or eddy currents within each face
that will not contribute significantly to the externally recorded potentials;
for this reason the synapse in the innervated face could cause it to fire
first even if it were not of lower threshold.) That the innervated face is
indeed of lower threshold is indicated by the effect of intracellularly
applied current, which depolarizes both faces equally if the external
resistances are equal (see Fig. 2). This procedure always excites the
innervated face first, even when the external resistance is somewhat
greater on the innervated side.
As will be shown later with respect to Gymnarchus (Section 11, C, 3 ) ,
a diphasic external potential can arise if one face is inactive and behaves
as a series capacitance. The excitability of both faces of Gymnotus elec394 M. V. L. BENNETT
trocytes can be clearly demonstrated by stimulating with externally
applied currents that run along the axis of the cell; these currents hyperpolarize
one face and depolarize the other. An experiment of this kind is
illustrated in Fig. 21. When a stimulating cathode is placed external to
the uninnervated face, the applied current tends to depolarize this face
and hyperpolarize the innervated face ( Fig. 21B,). The uninnervated
face fires first as indicated by initial positivity of the response recorded
external to the innervated face. The spike recorded across the innervated
face arises from a hyperpolarized level of membrane potential confirming
that this face is not initiating the spike. The innervated face is excited by
the activity of the uninnervated face, and negativity external to this face
follows the positive phase. When the stimulating electrode external to the
uninnervated face is an anode, the innervated face is depolarized and
fires first; there is initial negativity outside this face (Fig. 21A1). Since
the uninnervated face is hyperpolarized by the applied current, its
firing is delayed compared to that evoked by neural or intracellular
stimulation and the spike recorded across the innervated face has two
quite distinct components. (Thc two spikc components are only barely
recognizable in the monopolar recordings of Fig. 20. )
Usually if stimulation only moderately above threshold is applied as
in Fig. 21A,,B,, the initially active face can excite the opposite face that
is being hyperpolarized by the stimulus. However, strong stimuli can
cause sufficient hyperpolarization of this face that the spike of the
initially firing face cxcitvs it only partially if at all. Correspondingly,
there is failure of the second spikc component recorded intracellularly
and disappearance or reduction of the second phase of the extcrnally
recorded responses (Fig. 21A2,B2).
Although in Gymnotus the single cells generate diphasic extcrnal
potentials, the overall organ discharge is triphasic. This configuration
results from the opposite orientation and earlier firing of the most dorsal
electrocytes (Fig. 22). These cells fire about ?d msec earlier than the
more ventral cells, all of which fire synchronously. The activity of thc
anterior faces of the dorsal cells results in the initial head negativity. The
activity of their uninnervated faces is simultaneous with the activity of
thc innervated faccs of the ventral cclls and summates with it to cause
the second, head negative phase. Activity of the uninnervated faces of the
ventral cclls causes the final, head negative phase. Corresponding to the
number of cells active, the initial phasc is the smallest, the second phase
is the largest, and the final phase is somewhat smaller than the second
phase.
In Gymnotus the electric organ has at its rostra1 region a small
number of modified cells that fire earlier than the main organ and10. ELECTRIC ORGANS 395
JL;
A,– +;
Fig. 21. Effects of axial stimulation on electrocytes of Gymnotus. Stimulation
by rectangular current pulses and recording as indicated in inset except that the
indifferent electrode is much farther away. Upper trace: differential recording
across innervated face. Lower trace : monopolar recording external to innervated
face. (A1,A2) Anodal stimulation external to the uninnervated face. (A1) The
stimulus is moderately above threshold and initiates a two component spike. The
external record is initially negative indicating that the innervated face generates
a spike first. (Az) Stronger stimulation largely blocks the second spike component
and reduces the positive phase in the external record; evidently the stimulus
hyperpolarizes the uninnervated face sufficiently that it is only partially excited
by the spike of the innervated face. (The residual positivity may represent local
response in or capacitative currents through the uninnervated face, see Section 11,
D, 3 . ) (B1,BZ) Cathodal stimulation external to the uninnervated face. ( B , ) A
two component spike is initiated but the external record is initially positive indicating
that the uninnervated face fires first. (Bz) Stronger stimulation causes failure of the
second spike component and a large reduction in the negative phase of the external
record indicating failure of excitation of the innervated face. The small intracellular
positivity and aysociated external negativity that develops after a latency of about
1 msec ia a PSP resulting from stimulation of the nerve supply. Modified from
Bennett and Grundfest ( 1959).
apparently generate monophasic external potentials as do eel electrocytes.
The detailed operation of this part of the organ has not been investigated.
It resembles in its operation the rostra1 accessory organs of
several other gymnotids ( see below).
c. Hypopomus. At least three species of Hypopomus have been
studied electrophysiologically, but the correspondence to the taxonomically
named species is somewhat uncertain. Pulses are emitted at a basal
frequency of 5-10/ sec, and again there are large accelerations during
swimming or when the fish is stimulated. One species can maintain its
discharge rate quite constant at two or more levels, the higher ones
generally associated with greater activity (Bullock, 1970). Cessation of
discharge has also been observed (Bullock, 1970; Black-Cleworth, 1970).
The pulses are about 2 msec in duration and of the order of 1 V in
amplitude; they are monophasic head positive in one of the species
studied and diphasic initially head positive in another.396 M. V. L. BENNElT
Fig. 22. Constitution of the organ discharge in terms of the responses of the
different columns of electrocytes ( Gymnotus). Left, intracellularly recorded responses
of individual cells in different columns at the three levels of the fish
indicated in the center diagram. The potentials are somewhat distorted by pickup
from other active cells. In each case the lowest trace is the monopolarly recorded
organ discharge at the tip of the tail which is used as a time reference; the vertical
line indicates onset of organ discharge. ( A ) The electrocytes of the most dorsal
column (upper traces) fire about 0.2 msec earlier than the cells of the three
ventral columns (middle three traces). (B) At this level the electrocytes of the
most dorsal column also fire earlier than those in the next two columns ventrally,
but the discharge slightly precedes the firing of the dorsal column in A. ( C ) Only
one column continues to this level. The electrocytes are caudally innervated and
probably represent those in the third column. The response arises approximately
at the same time as does activity in the caudally innervated electrocytes in A.
Right, organ activity recorded at the tail as a summation of the potentials produced
by the different columns. Broken vertical line indicates the start of discharge of
the three columns of caudally innervated cells. The rostrally innervated cells contribute
a smaller, earlier diphasic potential which is of opposite sign. From Bennett
and Grundfest (1959).
As in Gymnotus the electric organ runs from just behind the chin
region to the tip of the caudal filament. The organ consists of 3-5
longitudinally running columns on each side in which there is one
electrocyte per segment ( Bennett, 1961). The cells are cylindrical, about
0.3 mm in diameter, and 0.2 mm thick in a 15-cm fish. All the electrocytes
are innervated by a small number of nerve fibers on a short process or
stalk from their posterior faces. In respect to innervation the electrocytes
closely resemble those of the main organ of Steatogenys (Fig. 29).
In the species of Hypopomus (probably H . urtedi) with the diphasic
organ discharge, the potential is initially head positive when it is10. ELECTRIC ORGANS 397
recorded at the head of the fish or some distance rostrally to it. At the
tail and caudal to it, the potential is inverted (Fig. 2 3 ) . Along the side
of the fish the discharge is triphasic, probably because of asynchronous
discharge of anterior and posterior cells.
The single cells resemble those of Gymnotus in that both faces
generate spikes and the resulting external potentials are diphasic; however,
all the cells are oriented in the same direction. The response of an
electrocyte to intracellularly applied depolarizing current is shown in
Fig. 24 recorded both monopolarly and differentially. The innervated
(posterior) face is lower threshold because it fires first under these con- +-
I rnsec
Fig. 23. Organ discharge of Hypopornus. A 16.5 cm fish is placed in a
shallow plastic tray, 45 cm long by 24 cm wide by 5 cm deep, and held by a
gauze tube lengthwise against the long side, midway between the surface and
bottom. Three electrodes are used. One is fixed at the head, and one is moved
at 1.5 cm intervals along the axis of the fish; these two electrodes record
differentially against an electrode at the midpoint of the opposite side of the tray.
The records are therefore somewhat larger but similar in form to what would
be obtained by monopolar recording in a very large volume of water. The brief
discharges occur at about 5/sec when the fish is unstimulated. The discharge at
the rostra1 fixed electrode triggers the oscilloscope and is used to align the records
in the figure obtained from the exploring electrode (positivity of this electrode up).
At the tip of the tail ( C ) the potential is diphasic and initially tail negative. Moving
the electrode farther caudally, it becomes smaller, but is similar in form (records
above C). Moving rostrally along the body, the potential becomes triphasic and
then diphasic, initially head positive, at the tip of the snout ( R ) . The potential
remains diphasic but decreases as the electrode is farther advanced (records below
R ) . From Bennett (1961).398 M. V. L. BENNETT
Fig. 24. Response of an electrocyte of Hypopomus to intracellularly applied
depolarizing current. Superimposed threshold pulses that do and do not excite the
cell. ( A ) Monopolar recording; ( B ) differential recording as in diagrams. Upper
trace: stimulating current. The transmembrane potentials in B (V, and V,) start
from the same level. The external record in B (V,) is shown with positivity
external to the uninnervated face upward. Both faces are excited, the spike of
the innervated face is lower threshold or faster rising and initially there is negativity
external to this face. The spike of the caudal face is longer lasting and
larger, and there is a longer lasting negativity external to this face. The characteristics
of the spikes are more clearly seen from the differential recordings ( B ) . The
external records in A are mirror images, showing the longitudinal orientation of
the cell. From Bennett (1961).
ditions as indicated by initial negativity outside this face and initial
positivity outside the uninnervated face. Differential recording across the
two faces shows clearly that, although the spikes in each face are of
about the same peak amplitude, the spike generated by the innervated
face is considerably briefer than that generated by the uninnervated
face. Thus, the head-negative phase of the external potential is larger
than the head-positive phase.
External stimulation of the electrocytes also demonstrates that both
faces generate spikes (Fig. 25). When current is passed in order to depolarize
the uninnervated face and hyperpolarize the innervated face,
the external record is initially head negative (Fig. 25A). The spike in
the uninnervated face soon excites the briefer spike of the innervated
face and the external record goes transiently more head positive, The
longer lasting spike of the uninnervated face causes the external record
to go head negative again. When current is passed that depolarizes the
innervated face and hyperpolarizes the uninnervated face, the brief spike
of the innervated face is evoked (Fig. 25B). However, this activity does
not excite the long-lasting spike of the uninnervated face, which is
apparently too hyperpolarized by the applied current. [A small head
negativity occurs in the external record. This phase may represent a sub10. ELECTRIC ORGANS 399
– 7
I rnsec –
$3
f l
Fig. 25. Stimulation of an electrocyte by externally applied currents. Hypopomus,
same cell as Fig. 24. Stimulation and recording as in diagram. Upper trace: stimulating
current. Middle traces (V, and V2): potentials across the two faces starting
from the same level. Lower trace: potential between the external electrodes, rostral
positivity up. When anodal current is passed ( A ) , the caudal (stalk) face is
hyperpolarized and the rostral ( uninnervated) face is depolarized. The rostral
face is excited first as indicated by initial rostral negativity externally. This activity
excites the brief caudal spike which produces a rostral-positive phase in the external
record which then becomes rostral-negative again. Oppositely directed currents
( B ) excite the caudal face first. The caudal spike fails to fire the rostral face,
which is hyperpolarized by the stimulus. Only a small rostral negativity occurs in
the external record. From Bennett (1961 ),
threshold response of the uninnervated face or discharge of the capacity
of this face (see Gymnurchus, Section 11, D, 3 ) .]
When the nerve supply is stimulated, excitation occurs first in the
stalk and then propagates into the main part of the cell. The potentials
in stalk and body can be quite different (Bennett, 1961) as is illustrated
with respect to the species of Hypopomus with a monophasic discharge
(Fig. 28).
The external potential generated by the electric organ of the monophasically
discharging Hypopomus remains relatively constant in shape
as an exploring electrode is moved along the fish and becomes very small
just caudal to the center of the fish (Fig. 26). Evidently the fish behaves
more like a dipole than the diphasically discharging Hypopomus, perhaps
because of better synchronization between cells but also because the
discharge of the individual cells is monophasic and the total output is
therefore less sensitive to slight failures of synchronization.
The single cells have properties similar to those of the electric eel
( Fig. 27). However, the uninnervated ( anterior) faces have a resistance
about half that of the innervated faces (Fig. 28D) and a spike in the
innervated face causes an appreciable voltage drop across the uninnervated
face. Still the external potentials differentially recorded across the400 M. V. L. BENNETT
Fig. 26. Potentials generated by the monophasically discharging Hypopomus.
Procedure as for Fig. 23, but recorded with the fish in the center of a somewhat
smaller container. The distance calibration gives recording sites with respect to
the fish referred to the level at the start of the records.
a
I msec
Fig. 27. Response of an electrocyte of monophasically discharging Hypopomus
to intracellularly applied depolarization. Upper trace: stimulating current. ( A,C )
Monopolar recording. Second trace: intracellular; third trace: external to uninnervated
face; fourth trace: external to innervated face. (B,D) Differential
recording. Second trace: across innervated face; third trace: across uninnervated
face; fourth trace: across entire cell. Superimposed traces of threshold stimulation
with and without a response. The innervated face generates a spike, but there is
only a small potential across the uninnervated face.10. ELECTRIC ORGANS 401
A =-. B-n-
2 msec
Fig. 28. Neural excitation and stimulation by axial currents of an electrocyte
of monophasically discharging Hypopomus. First trace: stimulating current applied
by an electrode external to the uninnervated face; next two traces starting from
the same level: differential recording across the cell’s faces; fourth trace: differential
recording across the membrane of the stalk (one electrode external to the innervated
face was used). ( A ) Three superimposed traces showing two subthreshold
PSPs and one that initiated a spike. The spike in the stalk is larger than that
across the main part of the innervated face of the cell. ( B ) A stronger and longer
lasting stimulus excited the stalk and innervated face directly as indicated by the
short latency. A PSP is still evoked and appears inverted on the peak of the spike
in the stalk. ( C ) A brief cathodal stimulus causes a neurally mediated response
to be initiated in the stalk (as indicated by latency and abrupt rise from the
base line). ( D ) A strong cathodal stimulus prevents the stalk impulse from invading
the body of the cell. A large amplitude but abbreviated spike remains in the stalk.
The innervated face is depolarized by about 40 mV but fails to be excited.
cells are about 50 mV in amplitude. Corresponding to the different resistances
the uninnervated face is more proliferated by tubules and canaliculi
than is the innervated face (Schwartz et al., 1971). The form of the
recorded spike potentials depends on the form of the stimulus. A brief
pulse near threshold produces a potential that rises more slowly than it
falls, whereas the opposite is true of long-lasting stimuli. The response
to brief pulses more closely resembles the externally recorded organ
discharge and presumably corresponds more closely to activation by way
of the synapses on the stalks.
During neural activation the impulse arises in the stalk and propagates
into the cell body. An external stimulating electrode can be used
both to evoke the neurally mediated response and to excite the cell
directly. A response evoked by a brief anodal pulse applied by an electrode
external to the uninnervated face is shown in Fig. 28A. Two small
PSPs subthreshold for initiating a spike are seen in the stalk, but there is402 M. V. L. BENNETT
little spread to the main part of the cell. Stronger stimulation causes a
larger PSP, and a spike is initiated in the stalk that is earlier, larger, and
longer lasting than the spikes recorded across the faces of the cell. With a
stronger and longer lasting stimulus the spike is initiated directly in the
stalk (or posterior face) as indicated by the short latency of the response
(Fig. 28C). At the time that the PSP occurs in Fig. 28A there is a depression
in the peak of the spike in the stalk. This response is the PSP that is
inverted because the potential during the spike exceeds the reversal
potential of the PSP (see del Castillo and Katz, 1956). If a cathodal
stimulus is applied external to the uninnervated face a large depolarization
of this face can be produced without exciting it, demonstrating its
inexcitability (Fig. 28D). Stimulation of this polarity hyperpolarizes the
innervated face, and if a prior brief stimulus initiates a neurally evoked
response in the stalk (Fig. ZSC), this activity can fail to invade the main
part of the innervated face of the cell. In this event only a small potential
is observed across the innervated face even though the spike recorded in
the stalk remains large.
The role of stalks in electrocyte function remains obscure. It will be
considered again in Section 11, F.
One unidentified species of Hypopomus has been studied in which
external recording suggests that electrocytes in about the anterior twothirds
of the body generate monophasic potentials and the posterior cells
generate diphasic potentials. This possibility requires exploration using
microelectrode techniques. No species of Hypopomus has been observed
to have a rostra1 accessory organ like those of Gymnotus, Steatogenys,
and G ymnorhamphicthys.
d. Steatogenys. Steatogenys is quite similar in appearance to
Hypopomus and is often confused with it by fish suppliers. The main
organ of Steatogenys appears morphologically identical to that of the
diphasically discharging Hypopomus (Fig. 29), and the form of discharge
recorded near the tail or distant from the fish is similar (Fig. 30).
The frequency of resting discharge is more like that of Gymnotus, and
various stimuli cause moderate accelerations [although there is some
dispute as to this point (see Bullock, 1970)l.
As in Hypopomus innervation occurs on a short stalk from the
posterior face (Fig. 29). Both faces of the cells generate spikes and
morphologically the two faces are similar although there is a somewhat
greater proliferation of the uninnervated face ( Schwartz et al., 1971).
Recordings across the faces and stalk of an electrocyte of the main organ
are shown in Fig. 31. A brief stimulus is applied to the nerve supply and
evokes a threshold synaptic potential. This PSP is much larger in the
stalk than across either face of the cell. The impulse arises in the stalk404 M. V. L. BENNETT r,; 0.5 msec
Fig. 30. Electric organ discharge of Steutogenys ekguns. Recorded monopolarly
along the side of the fish in a large dish of water. Time relations are taken from
a simultaneous recording from a stationary electrode in the submental region.
Records on the left are from the intact fish. The potentials are essentially diphasic
in the tail region, but there is an early negative phase near the head (see Fig. 35).
The early negative phase is lacking from the records on the right which were taken
after removal of the rostral accessory organs. A minimum-sized potential is recorded
4 cm rostral to the tip of the tail and the diphasic potentials are of opposite sign
on either side of this point.
v- I msec
Fig. 31. Response of an electrocyte of the main organ of Steutogays eleguns.
Recording and stimulating as indicated in the diagram. A brief stimulus is given to
the nerve supply just after the beginning of the sweep and evokes a PSP just
threshold for initiating a spike in the stalk (Vs,tw o superimposed sweeps). The
spike of the innervated (stalk) face (Vt ) rises more rapidly and is briefer than that
of the uninnervated face (V,) and the external potential (V,) is diphasic, initially
positive outside the uninnervated face. The PSP is considerably larger in the stalk
than across either face of the cell and the spike in the stalk is longer lasting.10. ELECTRIC ORGANS 405
and rapidly propagates into the main part of the cell. The external
record is initially positive on the side of the uninnervated face, which
indicates that the spike in the innervated (stalk) face rises more rapidly,
although this is not clear in the figure. The impulse in the uninnervated
face then becomes larger and a phase of head negativity appears in the
external record. This latter phase is considerably larger than the initial,
head-positive phase.
Steatogenys has an accessory electric organ in the chin region
(Lowrey, 1913) which was called the submental filament by taxonomists
( Fig. 32). In Steatogenys elegans ( a relatively large species reaching
15-20 cm long) the submental organ consists of a single column of cells
innervated on their anterior faces. The posterior part of the organ can be
seen through the overlying epidermis; the anterior part lies more deeply
in a fold of dermis and is not visible in the intact animal. There is an
,PO 0
Fig. 32. Rostra1 accessory organs of Steatogenys elegans. ( a ) Appearance in
intact fish; both postopercular organ (POO) and submental organ (SMO) are
visible. ( b ) Overlying tissue is removed to show the full extent of the organs and
the nerve ( N ) coming from spinal nerves to innervate them. The most rostra1
electrocytes of the main organ are also shown. Spinal cord, C; main organ, MO;
longitudinal nerve plexus, P; and rib, R.406 M. V. L. BENNETT
additional electric organ behind each operculum. The postopercular
organ is for part of its length a double column of cells; all the cells are
innervated on their posterior faces. These accessory organs are innervated
by spinal nerves that run rostrally from the most anterior level of
the main organ (Fig. 32). Electrocytes of the accessory organs have
several stalks instead of the single one found in cells of the main organ
( Fig. 33). They are difficult to penetrate with microelectrodes because
of the connective tissue surrounding the cells. However the isolated
organs can be studied, and it can be concluded that they generate
monophasic action potentials as do the cells of the eel and the monophasically
discharging Hypopornus, and they have a similar marked
proliferation of the uninnervated face ( Schwartz et al., 1971).
When a column is stimulated by an axial current depolarizing the
innervated faces, a monophasic spikelike potential is produced with a
very short latency (about 0.1 msec in Fig. 34A). This response is positive
with respect to the uninnervated face, and because of its short latency it
can be identified as a directly excited spike. This potential is followed at
a latency of about 1 msec by a smaller potential of similar sign that
evidently represents PSPs and perhaps spike activity in the stalks. If a
brief axial stimulus is given that hyperpolarizes the innervated face and
depolarizes the uninnervated face, there is no short latency response
indicating that the uninnervated face is inexcitable (Fig. 34C). This
mode of stimulation does excite the nerve fibers; the delayed response is
still present, but it becomes a full-sized spike since the innervated faces
are not refractory as in Fig. 34A. The neural origin of the delayed
response is confirmed by the effects of curare, a drug which blocks
transmission at cholinergic synapses like those of nerve-electrocyte junctions.
Following curare treatment the delayed response is absent for both
polarities of stimulation, although the directly excited response persists
unchanged (Fig. 34B,D).
The accessory organs fire about 1 msec before the main organ (Fig.
35). The submental electrocytes are innervated on their anterior faces,
and the potential external to the anterior end of the organ is largely
negative. The postopercular electrocytes are innervated on their posterior
faces, and negativity is recorded external to the posterior end of this
organ. At the same time the potential in the gill opening and in the
mouth is positive. These electrocytes act to make the interior of the head
positive. If the accessory organs are removed, the discharge in the head
region becomes the simple diphasic potential of main organ (Fig. 30).
There is another much smaller species of Steatogenys (5-7 cm)
which has submental electrocytes that are all innervated on a single stalk
from their posterior faces. This species lacks a separate group of postopercular
electrocytes. These cells also have a monophasic discharge and10. ELECTRIC ORGANS 407
Fig. 33. Anatomy of submental electric organ of Steatogenys elegans. Romanes
silver stain of paraffin embedded material. A longitudinal section, rostral to the
right. Two electrocytes ( e ) are shown, each with several stalks ( s ) on the rostral
face. Nerve fibers ( n ) come from the longitudinally running nerve trunk in the upper
part of the figure to end on the stalks. The sheath (sh) around the organ is seen
at the bottom.
function to make the interior of the head positive. The largest external
potential is recorded at the posterior of the organ; there is very little
potential at the anterior end of the organ under the chin. The direction
of current flow with respect to the anterior posterior axis through this
organ is opposite to that in the submental organs of Steatogenys elegans,408 M. V. L. BENNETT
Fig. 34. Response of the submental organ of Steatogenys elegans. Recording
in a bridge circuit (Bennett, 1961). Rostra1 positivity up. ( A ) A brief stimulus
depolarizing the innervated (caudal) faces evokes a brief spikelike potential at
short latency followed by a smaller potential. (B) Following curarization ( about
0.1 mM) the brief spikelike potential is unaffected but the longer latency component
is abolished. ( C ) Before curarization a brief stimulus hyperpolarizing the
innervated faces evokes a response with a latency of about 1 msec. ( D ) Following
curarization the same stimulus evokes no response.
but the transepidermal potentials are affected similarly because opposite
ends of the organ are connected to the exterior and interior of the animal.
The function of these accessory organs is presumably to increase sensitivity
of the electrosensory system in the head region, but there are no
experimental data.
e. Gymnorhamphichthys. Gymnorhamphichthys usually rests buried
in sand during the day and emerges to swim around and feed at night.
When in the sand it emits pulses at a quite constant rate of about 10-15/
sec. When actively swimming it increases its discharge rate to a new and
also quite constant frequency which it maintains for prolonged periods.
Disturbances also cause increase in discharge rate. It has been shown to
exhibit a circadian rhythm of motor activity and (perhaps secondarily)
of discharge frequency under conditions of constant darkness ( Lissmann
and Schwassmann, 1965).
The main organ of Gymnorhamphicthys is cytologically and, as inferred
by recording from the intact fish, electrophysiologically like that
of Steatogenys. Gymnorhamphichthys also has submental electrocytes as
does Steatogenys. They are located beneath the dermis and thus are not
obvious, which accounts for their not being seen previously. Each cell
has multiple stalks from its posterior faces. External recording indicates
that the submental cells fire before the main organ and generate mono10. ELECTRIC ORGANS 409
Fig. 35. Responses recorded in head region of Steatogenys elegans. Methods
as for Fig. 30. The dotted lines indicate the sites of recording of the various traces.
The three traces with arrows are recorded 2.5 cm rostral to the snout, 4.5 cm
caudal to the snout and near the tip of the tail. There are large early negativities
external to the rostral end of the submental organ and external to the caudal end
of the postopercular organ. There are corresponding positivities just inside the
mouth and gill opening.
phasic action potentials. The related genus Rhumphichthys has not been
studied with respect to its electric organs.
f. Sternopygus and Eigenmunnia. Sternopygus and Eigenmannia are
closely related and the organ discharges are so similar that they may be
considered together. The discharges consist of head-positive pulses
superimposed on a head-negative base line such that there is little dc
component in the organ discharge; that is, averaged over one discharge
cycle, there is little or no net current flow (Fig. 3C). The frequency in
Sternopygus is about 5&100/sec, and in Eigenmannia it is about 250-
6OOIsec.
The frequency is quite constant, and ordinary forms of stimulation do
not affect it. Weak electric stimuli of nearly the same frequency as the
animal‘s own discharge evoke the “jamming avoidance” response, that is,
a small shift in frequency away from that of the interfering stimulus
(Watanabe and Takeda, 1963; Bullock, 1970; see Chapter 11, this
volume). Sternopygus can also stop discharging completely under these
conditions.
The electric organ is located in the same position as in Hypopomus410 M. V. L. BENNETT
Fig. 36. Anatomy of the electric organ of Eigenmannia. Organ isolated from
osniic acid fixed material. Dorsal up, rostra1 to the right viewed from the lateral
surface. ( A ) Low power view. The five columns of electrocytes are numbered
dorsoventrally. Two segmental nerves ( n ) run ventroposteriorly and give off
branches to the electrocytes. A complete cell of column 4 is seen most clearly. The
small spots on the cells are nuclei. The posterior, innervated ends of the cells are
more darkly stained. There is little extracellular space between cells, and the cells
usually overlap somewhat in the longitudinal axis. ( B ) The anterior end of a cell
of column 3 lies lateral to the next cell anteriorly, and is similarly overlapped by
the cell caudal to it. ( C ) Details of innervation of a cell from column 4. ( D )
Innervation of the anterior cell from column 5 in A. Magnifications the same in
A and B and in C and D.10. ELECTRIC ORGANS 411
and there is no specialized region rostrally. The electrocytes are tubular
to spindle-shaped with the long axis parallel to the anteroposterior axis
of the fish. The cells are 1-2 mm long and 0.2 mm in diameter in a
Sternopygus 15 cm long and somewhat larger in an Eigenmunnia of
similar size (Fig. 36). They are innervated on their posterior faces. There
are about 6 columns on each side in Eigenmunnia and about 10 in
Sternopygus. The cells are accurately aligned in columns in Eigenmannia,
but they are separated by only small spaces; the amount of extracellular
space between cells is much smaller than the cells themselves. The alignment
is less regular in Sternopygus, but the amount of extracellular space
is similarly small.
It may not be obvious from records such as those of Fig. 3C that the
discharge consists of head-positive pulses superimposed on a headnegative
base line, and the distinction between pulse and interpulse
interval may be even less obvious in the faster firing Eigenmunniu. The
distinction is readily established experimentally (Fig. 37). An appropriately
timed stimulus to the spinal cord can cause the complete disappearance
of a pulse ( Fig. 37H) or cause a pulse to be generated earlier (Fig.
37C). The mechanism of block appears to be that the evoked volley in
descending fibers finds the spinal neurons refractory from their immediately
preceding response (initiated by the center in the medulla controlling
the discharge, see Section 111, A), and the evoked volley also
collides with the next command volley from the medulla preventing
Fig, 37. Effects of spinal stimulation on organ discharge
of Sternopygus. Potential recorded differentially
along caudal portion of tail, rostra1 positivity up. ( A )
Normal discharge. ( E I ) A brief stimulus to the spinal
.fuU-i
i jlJUl-0 o cord is given at successively later times in the discharge
10 msec cycle.412 M. V. L. BENNETT
it from reaching the spinal neurons. The stimulus however has no
noticeable effect on the following pulse indicating that the organ is
controlled rostral to the spinal cord and that the antidromic volley does
not invade the controlling nucleus. [There is actually a very slight advance
of subsequent pulses ( see Bennett et d., 1967c) .]
In Sterrwpygus the head-negative potential between responses can
also be demonstrated by section of the spinal cord which blocks further
pulse activity ( Fig. 38). The head-negative potential remains immediately
after spinal section but decays away over the next minute or so.
Repetitive stimulation of the organ for some seconds causes at least a
partial restoration of the head-negative potential which decays away
again on cessation of stimulation [although somewhat more rapidly than
following spinal section (Bennett, 196l)I.
The origin of these potentials has been elucidated by microelectrode
studies of the electrocytes. The pulse component of the organ discharge is
generated by the posterior end of the cell, and the head-negative component
is generated by the anterior end. The cells are long compared to
Fig. 38. Organ discharge of Sternopygus, effect of spinal section and repetitive
stimulation. Potential recorded differentially along caudal portion of tail, rostral
positivity up, zero level indicated by horizontal line and base line traces. Two
preparations. ( A ) Two complete normal discharge cycles are shown on the left.
The potential is head positive and head negative for approximately equal times.
Spinal section abolishes the pulses, but a head-negative potential about equal to
that during regular organ discharge remains (lower line at time 0). This potential
decays over approximately 2 min (right). (B) Stimulation of organ inactivated by
spinal section. Separate stimulating and recording electrodes, head positivity up.
Brief pulses are passed through a condenser to give diphasic stimuli with no dc
component. The pulses are oriented so that the initial brief phase excites the innervated
faces. Their response appears superimposed on the slow opposite phase ( left ).
Repetitive stimulation at 50/sec for 15 sec develops a head-negative potential that
on cessation of stimulation decays over 15-20 sec (right). From Bennett (1961).10. ELECTRIC ORGANS 413
Fig. 39. Spikes and PSPs in an electrocyte of Sternopygus during organ discharge.
Two intracellular recording electrodes are used, one 700 p rostral to the other. Both
record differentially against a closely applied external electrode to minimize pickup
from other cells. The second trace gives the potential at the caudal electrode with
the upper trace as its zero level. The fourth trace gives the potential at the rostral
electrode with the third trace showing its zero level. When the cell is firing
regularly, as in A, the spikes and the potential between spikes are greater at the
caudal end. Repeated hyperpolarizing pulses are passed through a third intracellular
electrode in order to block the spikes and, after some seconds, the records become
as in B. The resting potential is equal at the two ends of the cell and only PSPs are
produced, which are larger at the caudal end. From Bennett (1961).
cells of weakly electric gymnotids discussed up to this point, and the
potential can be quite different at the two ends of the cell. The spike and
PSPs are larger at the posterior end of the cell (Fig. 39), and the threshold
current is lower when applied through an electrode at this end of the
cell. Longitudinal stimulation as in Figs. 21 and 25 has not been carried
out. However, in the caudal filament there is little tissue other than the
organ, and it can be stimulated by external electrodes. This procedure
shows the posterior faces to be capable of generating a spike and the
anterior faces to be inexcitable just as demonstrated for the submental
organs of Steatogenys (Fig, 34).
In a cell that has not been stimulated for some time, the resting
potential is the same at the two ends of the cell. Repetitive stimulation
causes a depolarization of the cell to develop, but the depolarization is
greater at the anterior end of the cell (Fig. 40). On cessation of stimulation
the potential of the two ends of the cells slowly equalizes again.
Similar changes are seen in neural activation (Fig. 39). From these
records it is clear that, during spikes, current flows from caudal to rostral
in the interior of the cell, thus generating the head-positive phase of the
organ discharge. When the cell has been active at the normal frequency
for a sufficient period, current flows caudally in the cell in the interval
between spikes thus generating the head-negative component of the overall
discharge.
Insufficient experimental data are available to warrant a deailed discussion,
but a possible explanation of the depolarization of the anterior
face can be given. During the spike sodium current flows inward through
the innervated face while outward current at the anterior end is carried414 M. V. L. BENNETT
……………………… ……. Peak of spike
-5 …………………… I …..
Fig. 40. Development and decay of slow depolarization produced by repetitive
direct stimulation of a single cell of Sternopygus. Organ inactivated by spinal
section. Three electrodes are used as in diagram. Brief pulses adequate to excite
the cell are applied at l/sec and 50/sec. Sample records are shown above graph
with the connecting lines indicating when they are taken. Upper traces: potential
at the caudal end of the cell and base line. Lower traces: potential at the rostral
end of the cell and stimulating current serving also as a base line. Graph: data
of complete experiment, resting potentials at the beginning taken as 0. Open
circles: caudal. Crosses: rostral. Resting potentials determined at the end of the
experiment are 64 mV at the rostral electrode and 67 mV at the caudal electrode.
Repetitive stimulation at 50/sec causes a steady level of depolarization to develop,
which is greater at the rostral end of the cell. On return to stimulation at l/sec
the depolarization subsides in about 20 sec. From Bennett ( 1961).
by potassium ions. Potassium accumulates in the restricted extracellular
space external to the anterior face and tends to depolarize it. This
mechanisni of ion accumulation provides an explanation of why there is
so little extracellular space in this organ in contrast to those of other
electric fish previously discussed. The question remains as to what ions
carry the outward current through the innervated face during the headnegative
phase of the organ discharge. It cannot be potassium if the
concentration of this ion is the same outside the two faces as it would be
except for cells at either end of the organ.
Spike generation in the electrocytes has an unusual feature: The
resistance during the peak of the spike exceeds that at rest by a factor of
1.5 to 2. This property may be seen from the application of brief pulses
between the spikes and at their peaks (Bennett, 1961) and is confirmed
by ac impedance measurements from the cells of the caudal filament
(Fig. 41). The sequence of conductance and permeability changes is
probably as follows. During the rising phase of the spikes the conduct10. ELECTRIC ORGANS 415
– 20 rnsec
Fig. 41. Impedance changes during organ discharge of Sternopygus. The tip of
the tail was placed in a bridge circuit. Upper trace: bridge signal after moderate
amplification. Lower trace: bridge signal after filtering out low frequencies and much
greater amplification. Amplitude and frequency of ac input 0.5V and 5 kc. The
bridge could be approximately balanced for both spike peaks and between spikes
( A ) , but large imbalances occurred on the rising and falling phases. When the
bridge was balanced for the falling phase ( B ), the imbalance at the spike peaks was
somewhat greater than between spikes, and balance was approached during the rising
phase. From Bennett ( 1961).
ance rises as a result of sodium activation. The conductance then decreases
again at the spike peak because the conductance of anomalously
rectifying membrane is a large fraction of the resting conductance and
decreases more than enough to compensate for the increase in sodium
conductance. Sodium inactivation ensues and the potential begins to fall.
However, on the falling phase the conductance rises to above the
maximum observed on the rising phase; this change is ascribable to
potassium activation or delayed rectification. Except for the conductance
increase on the falling phase this pattern of changes is like that in the
electric eel for which the data quite adequately demonstrate the proposed
mechanism (see Nakamura et d., 1965; Morlock et al., 1969). The
similar conductances during spikes and at rest may be functional in
maintaining an ac organ discharge. Exposure to media of different conductivities
will load both phases of organ discharge similarly. Therefore
less dc component will be associated with the new amplitude of organ
discharge than there would be if the conductance during spikes were
much lower than that between spikes.
Conductance changes associated with organ discharge are the same
in Eigenmunnia and in Sternopygus. Insufficient physiological data are
available to distinguish whether the absence of a dc component in the
discharge results from a mechanism like that in Sternopygus or whether
the uninnervated face acts as a series capacity as in Gymnarchus and
probably also the sternarchids (see Sections 11, D, 3 and 11, D, 1, g). The
two different mechanisms might be characterized as a polarization416 M. V. L. BENNETT
capacity in contrast to a true membrane dielectric capacity. Ultrastructural
data suggest that Eigenmannia may employ a dielectric
capacity. The uninnervated face of its electrocytes has many fine branching
tubules or canaliculi that greatly increase the surface; the situation
thus resembles that in Gymnurchus (Schwartz et al., 1971). In Sternopygus,
the innervated and uninnervated faces are similar and rather smooth,
their areas being little increased compared to surface seen at the light
microscope level of resolution. ( A question here is why there is little
extracellular space in the electric organ of Eigenmanniu.) The significance
of an ac discharge will be discussed in Section 11, F.
g. Sternarchids. This is the largest of the gymnotid families in terms
of numbers of genera and species (Tables I and 11). The group is characterized
by the presence of a small caudal fin. There is also the so-called
dorsal filament that arises about the middle of the back and runs
posteriorly for perhaps a third of the body length (Fig. 42). The dorsal
filament is in vivo closely adherent to the back of the fish, and back and
filament are contoured so that the filament is practically invisible; it may
separate and become obvious in preserved specimens.
The sternarchids discharge their electric organs at the highest fre-
— =n=mmrnmm—m1—
Enlarged nodes
Fig. 42. Anatomy of the electric organ of Sternarchus. The upper diagram shows
the location of the organ in the fish. The middle diagram shows the organ and nerves
running to it and the course of a single fiber from its origin in the spinal cord. The
lower diagram represents a single fiber and its nodal structure.10. ELECTRIC ORGANS 417
quency of any known electric fish. Depending on the species, individual,
and temperature the frequency ranges from about 700 to 1700 impulses
per second ( Steinbach, 1970). The discharge shape varies somewhat with
the species, and will be treated separately below. The frequency is
generally highly stable and is not affected by ordinary stimuli. Weak
electric stimulation at closely neighboring frequencies does evoke small
shifts in frequency, the “jamming avoidance” response ( Larimer and
MacDonald, 1968; Bullock, 1970; see Chapter 11, this volume). Several
species have been observed to accelerate their discharge briefly, emitting
a “chirp” (Bullock, 1970). These changes may function in intraspecific
communication and in species recognition.
The electrocytes of the sternarchids turn out to be spinal neurons
rather than cells of myogenic origin (de Oliveira Castro, 1955; Bennett,
1966, 1970). The best-studied species is Sternarchus albifrons because
this is the one most commonly obtained by tropical fish dealers who call
it the black ghost knife fish. The fish, electric organ, and electrocytes are
shown diagrammatically in Fig. 42. The organ runs longitudinally just
ventral to the spinal column over most of the length of the body. The
axons of the spinal neurons descend from the cord and enter the electric
organ (Fig. 43). They then run anteriorly for several segments, turn
around and run posteriorly to end blindly at about the same level that
they entered the organ. This course is established by dissection of single
fibers from their point of entry into the organ until their termination.
Evidently the organ has lost its myogenic electrocytes and enlarged the
axons that formerly innervated them. This origin is confirmed by the
existence of a nerve plexus in Gymnotus, Hypopomus, Steatogenys, and
Gymnorhumphichthys in the same location as the sternarchid electric
organ. In Gymnotus axons in this plexus have been traced to the (myogenic)
electrocytes. The plexus is absent in Sternopygus, Eigenmannia,
and Electrophorus, a fact which has implications for the phylogenetic
relations within the gymnotid group.
There are characteristic changes in the nerve fibers as they run along
in the electric organ. In both its anteriorly and posteriorly running parts
it becomes greatly dilated and can exceed 100p in diameter. Before it
enters the organ and where it turns around to run posteriorly, it is of the
usual diameter for a large myelinated fiber, 10-2Op. It tapers gradually
before it terminates. The nodal structure also changes characteristically
along the fiber (Fig. 44). The nodes become somewhat enlarged just
after the fiber enters the organ. As the fiber dilates, the nodes become
very narrow (although area is difficult to estimate because of the increased
fiber diameter). As the fiber tapers near its most anterior part,
there are several very long nodes. The nodes become of ordinary size
again as the fiber turns around. This sequence repeats itself in the10. ELECTRIC ORGANS 419
caudally running section of the fiber. There are first moderately enlarged
nodes, then narrowed ones, then very enlarged ones.
The organ discharge of Sternarchus consists of diphasic pulses that
are initially head positive (Figs. 3D and 45). As would be expected from
the neural origin of the electric organ, the discharge is unaffected by a
dose of curare that is sufficient to paralyze the animal completely.
The contribution of the single cells to the organ discharge is disclosed
by microelectrode experiments. Just anterior to where the fiber enters
the organ, a large impulse is recorded in the fiber during the headpositive
phase of organ discharge (Fig. 46A). This impulse is smaller
more anteriorly and the peak is delayed; thus, like the cells of Sternopygus,
these cells are not isopotential. The response in this part of the fiber
accounts for the head-positive phase of the discharge because current
runs anteriorly along the interior of the cell at this time. In the
posteriorly running section of the fiber the impulse is larger anteriorly
(Fig. 46B). This activity occurs during the head-negative phase of the
organ discharge and current flows in the cell in such a way as to give
rise to the head-negative phase. The anatomy and recordings indicate
that impulses coming from the spinal cord propagate along the electrocyte
in the organ and that the activity of the anteriorly running part
excites the posteriorly running part. Because the impulse is smaller at
the distal end of each section of the fiber, it is likely that these regions
are inexcitable like the uninnervated faces of eel electrocytes. The
current paths and intracellular potentials during one discharge cycle are
diagrammed in Fig. 47.
There is evidence that the inactive regions of the cells act as a series
capacity rather than as a series resistance. The effects when one face of an
electrocyte acts as a capacity are discussed in detail in respect to
Gymnarchus, for which the experimental evidence is more complete
(Section 11, D, 3 ) . The most important effect is that the organ discharge
Fig. 43. Electric organ of Sternarchus. Formalin fixed material, dissected and
stained with methylene blue. Dorsal to the top, anterior to the right. ( A ) The
electric organ (0) following removal of its connective tissue sheath. Single longitudinally
running electrocytes are visible. Nerves to the organ ( n ) and to more ventral
tissues ( s ) are also seen. (B) Following dissection away of all electrocytes from
nerves caudal to those shown and of most fibers entering the organ from the more
caudal nerve in the figure. Three descending fibers of this nerve run approximately
parallel in their origin position (arrows); two others are deflected anteriorly; the
central part of a sixth is broken off shortly after entry into the organ (arrow). The
more anterior nerve is undissected; its fibers diverge in the organ. Most of them lie
medial to the fibers from the more caudal nerve. (C) Higher magnification of the
left part of B. The five numbered fibers can be traced from the top left to the lower
right. The smaller fibers of the ventral segmental nerve ( s ) are visible. Fiber 4
passes medial to this nerve. Calibrations on the lower right.420 M. V. L. BENNETT
Fig. 44. Structure of Stemarchus electrocytes. (A-D) Photographs of a cell isolated
by dissection following osmic acid fixation. ( A ) The fiber near its entry into the
organ; 4 nodes (arrows) are visible; the 3 to the left appear somewhat enlarged.
( B ) A dilated part of the fiber in its longitudinally running course; 2 nodes of
peculiar structure are seen (arrows). ( C ) The thin region of the fiber where it turns10. ELECTRIC ORGANS 421
– 0.5, 2.5 msec
Fig. 45. Electric organ discharge of Stemarchus. Recorded differentially between
head and tail at two sweep speeds. Upper records; prior to curarization. Lower
records: after a dose of curare (10 mg/kg) that completely paralyzed the fish, organ
discharge was unaffected.
has no dc component. At a constant frequency of activity current flows
as follows. During a spike at one end of the cell the charge on the
capacity of the other end is made more positive from the negative resting
level; after the spike the capacity recharges to its original level, and no
net current flows over the discharge cycle. The evidence for this property
in sternarchids is largely comparative (M. V. L. Bennett and A. B.
Steinbach, unpublished data). The organ discharge of all sternarchids
has little or no dc component. In Sternarchus this could result, as in
Gymnotus, from successive firing of opposite faces or ends of the cells.
However, in Sternurchorhamphus the organ discharge consists of monophasic
head-negative pulses superimposed on fairly level head-positive
base line with no dc component (Fig. 48A,B). The electrocytes lack
a rostrally running portion; the fibers turn and run caudally on entering
the organ. The monophasic appearance of the discharge is consistent
with activity of only the rostra1 regions of the fibers. The head positivity
between impulses could be generated by a polarization capacity as in
Stemopygus. It seems more likely that the mechanism is like that of
around to run posteriorly again; 5 nodes are seen; the 2 to the right appear somewhat
enlarged. ( D ) The caudal termination of the fiber, at least three very large nodes
(double arrows) and 3 small internodes ( i ) are seen. The nature of the adhering
structures near the fiber termination is unclear. (E-F) Toluidine blue stained sections
of osmic acid fixed Epon embedded material. ( E ) Arrows indicate a large node in
the dilated fiber on the right and a small node in the dilated fiber on the left.
(F) Arrows indicate a node in a narrow region of a fiber that is turning around a t
the anterior limit of its course. Note that the other fibers in this section are cut transversely.
Magnifications the same in A-l3 and in E-F.422 M. V. L. BENNETT
– I msec
Fig. 46. Intracellular recordings from fibers in the electric organ of Sternarchus.
Differential recording with respect to a closely applied external electrode to minimize
contributions from other cells. Upper trace: monopolar recording of organ discharge.
Middle trace: intracellular recording from anterior electrode. Lower trace: intracellular
recording from an electrode in the same fiber several millimeters posteriorly.
( A ) Recording from an anteriorly running fiber segment as indicated by the larger
size and earlier peak of the potential at the posterior electrode. The impulse occurs
during the initial phase of organ discharge (negative going when monopolarly
recorded at this level of the organ). ( B ) Recording from a caudally running fiber
segment as indicated by the larger size and earlier peak of the potential at the
anterior electrode. The impulse occurs during the second phase of organ discharge.
Gymnurchus, and the head-negative phase results from charging of
the dielectric capacity of the caudal portion of the fiber. That the
potential between pulses is quite level only requires that the time
constant of the system be long compared to the interval between pulses.
Fig. 47. Diagram of potentials and current flow at successive regions of the
electrocytes during activity. The upper potential tracing represents an organ discharge.
The directions of current flow during the spikes in the different regions are
indicated by the arrows on the right.10. ELECTRIC ORGANS 423
In some Adontosternarchus the main organ discharge has a small
head-negative phase followed by a large head-positive phase separated
by a more or less level base line. However, the base line is head positive
so that again there is no net current flow during a discharge cycle (Fig.
48C,D). The electrocytes have both rostrally and caudally running
portions, but the rostrally running section is shorter than the caudally
running section. Thus, the electrocytes present an intermediate condition
between those of Sternarchus and Sternarchorhamphus, The small headpositive
phase of the pulse is ascribable to the small anteriorly running
portion of the fiber; the large head-negative phase is ascribable to the
large posterioriy running portion. The most likely way for the headnegative
level between pulses to arise is for each portion of the fiber
to have a series capacity.
These data do not require Sternarchus electrocytes to have a series
capacity. However, if the fish is made anoxic the second phase of the organ
discharge appears to fail. In spite of this change, the organ discharge still
has no dc component. It seems probable then that both anteriorly and
posteriorly running regions have a series capacity. In some individual
Sternarchus the initial, head-positive phase is larger than the second,
head-negative phase, although there is still no dc component. Correspondingly
in some individuals, presumably the same ones, the rostrally
running portion of the electrocytes is larger than the caudally running
6 . 2 . 5 mSec
Fig. 48. Organ discharge of Sternurchorhamphus and Adontosternarchus.
(A, B ) Recorded monopolarly at the tail of Sternurchorhamphus, positivity down
(equivalent to head positivity up). ( C, D) Recorded differentially between head
and tail of Adontosternarchus, head positivity up. Horizontal lines indicate zero
potential levels. Faster sweep speed in A and C.424 M. V. L. BENNETT
portion. Unfortunately, it is not possible to test these mechanisms by
activating the organ over a wide range of frequencies. The cells continue
to fire spontaneously after spinal section, and synchronous low frequency
activity cannot be obtained.
One sternarchid, Adontosternarchus, has an accessory organ in the
chin region (M. V. L. Bennett and A. B. Steinbach, unpublished data).
This organ generates a potential of 20-25 mV amplitude outside the
chin (Fig. 49). It is absent from the five or more other genera of sternarchids
examined on the Rio Negro expedition of the R. V. Alpha Helix.
It is made up of fibers that are probably modified from electrosensory
nerve fibers for the fibers end in the skin in what appear to be modified
electroreceptors. Consistent with its sensory origin the impulse frequency
is set in the chin organ itself, and impulses proceed centripetally in the
afferent fibers which can be sectioned without affecting organ discharge.
Because of its peripheral origin the chin organ can fire at a somewhat
different frequency from the main organ (Fig. So).
Since electroreceptors themselves can, under certain conditions, generate
maintained oscillations (see Chapter 11, this volume), it is necessary
to distinguish the neurogenic potentials of the chin organ from
potentials generated by electroreceptors that of course the fish also
possesses. The most important difference is morphological. The fibers
of the chin organs have very dilated myelin sheaths and peculiar nodal
structures that closely resemble the characteristics of electrocytes in the
main organ. Electroreceptor afferent fibers are ordinary myelinated fibers.
Futhermore, if an electrode is advanced into the organ from the
surface, the polarity of discharge does not invert until the electrode
is deep into the tissue of the chin. In contrast, the oscillations at electroreceptors
invert when the epidermis is crossed. Finally, the chin organ
oscillations are little affected by external loading; the frequency is
nearly constant whether the chin is in air or immersed in physiological
saline. In contrast, electroreceptors do not oscillate when they are
electrically loaded to only a small degree; generally, they will oscillate
only when the skin is allowed to dry in air.
Fig. 49. Main and accessory organ discharges of Adontostemarchus. Recorded
monopolarly with respect to a distant electrode in a large volume of water. There
is a relatively large potential in the chin region ( a ) and at the tip of the tail ( h ) .
The potentials are smaller elsewhere over the head (b-e) and near the middle of
the body (f, g ) .10. ELECTRIC ORGANS 425
A B
– 0.5 msec
Fig. 50. Lack of synchronization of main and accessory organs of Adontosternarchw.
Recording monopolarly from chin organ (upper trace) and tip of tail (lower
trace). ( A ) When the sweep is triggered by the accessory organ discharge, repetitive
superimposed sweeps show that the phase of main organ discharge changes. ( B )
When the sweep is triggered by the main organ discharge, the phase of the accessory
organ discharge changes in different sweeps.
While measurements to date are inadequate, it is probable that discharge
of the chin organ has little dc component. The function of the
chin organ is obscure, but the potentials produced are large enough to
activate receptors in the head region. Presumably the chin organ plays
the same role as the rostra1 accessory organs of other gymnotids.
2. THEE LECTRICCA TFISH
The electric catfish, Malupterurus electricus, is the only silurid known
to be electric. It is also distinguishable from other catfish by the absence
of rays in its dorsal fin. The electric organ lies in the skin surrounding
the body over most of the length of the fish (Fig. 1). It is innervated
by two giant neurons, one on either side of the spinal cord in the first
spinal segment (Fig. 72). Each neuron sends out a single axon that
innervates the several million electrocytes on that side of the body. The
electrocytes are shaped rather like lily pads. The main part of the cell
is disc-shaped, about 1 mm in diameter and 2040 p thick in a fish
15 cm long (Fig. 51). From a convoluted region in the center of the
caudal face (called the rosette), a stalk protrudes that is about as long
as the radius of the disc-shaped part. The cell is innervated on the tip
of this stalk by a branch of the axon from the giant cell.
The electric organ of Malapterurus was long thought to be of glandular
origin, in part because of its location and because the side opposite to
the innervation of the electrocytes became negative. It thus violated
the rule formulated by Pacini concerning innervation and polarity (who426 M. V. L. BENNETI
Fig. 51. Anatomy of the electric organ of the electric catfish. ( A ) The dissected
ventral surface of the fish showing the organ (0) and the nerve (ne), artery ( a e ) ,
and vein (ve) running to it (from Rosenberg, 1928). ( B ) The body of a single
electrocyte dissected out following forrnalin fixation, and stained with methylene
blue. ( C ) Higher magnification of the rosette region of another cell, Romanes’
silver stain.10. ELECTRIC ORGANS 427
is better known for the corpuscles that bear his name) and generated
potentials similar in sign to those known from some glandular tissue.
It turns out that the discharge polarity is explicable in terms of ordinary
mechanisms of excitability (Keynes et al., 1961), and recent physiological
and embryological data are consistent with a muscle origin for
the organ (Johnels, 1956).
The organ discharges are primarily head-negative pulses 1-2 msec
in duration (Fig. 3A,A’). The head-negative pulse is preceded by a
very much smaller head-positive phase that was not observed in earlier
studies (Fig. 52). This early phase results from activity of the stalks
as will be discussed below. Thus, the organ really does not violate
Pacini’s rule. The pulses are emitted infrequently by an undisturbed
fish. However, when the animal is feeding or mechanically stimulated
a few pulses or long trains of pulses can be emitted. Prey detection
appears to precede pulse emission and the organ presumably functions
in prey capture (Bauer, 1968). Inputs from taste receptors, which are
i
1.5 rnsec
Fig. 52. Electric organ discharge of the electric catfish. Recorded between head
and tail of a fish about 15 cm in length in a small container of aquarium water. High
gain (upper traces) and low gain (lower traces), head negativity upwards, faster
sweep in A. Two superimposed sweeps, one with and one without discharges to show
base line. The primary head-negative discharge is preceded by a small head
positivity. Single pulses (A) or trains (B) are emitted in response to touch.428 M. V. L. BENNETT
found over much of the body surface, are powerful excitants. However,
the effectiveness of the discharge in stunning small fish is minimal. The
amplitude recorded in air is about 150 V from an animal 30 cm in
length and even a small fish 5 cm long can generate 30 V (Keynes et al.,
1961). The amplitudes are considerably reduced in water but are still
uncomfortably strong to someone trying to pick up the fish. The fish
reaches a length of at least 50 cm, and 350 V discharges have been
reported ( Keynes, 1957).
It is interesting that the cells produce an almost entirely head-negative
potential, although they are innervated on their posterior faces; in fact,
the mechanism is not so different from that in the diphasicly discharging
Hypopornus (Fig. 24). For the experiments of Figs. 53 and 54 a
propagated impulse is set up by stimulation through a large monopolar
electrode pressed against the main part of the cell distant from the
recording site. A monopolar recording electrode detects a negative
spike external to the nonstalk face (Fig. 53A) that is propagated in
from the stimulation site. If this electrode is advanced into the cell, it
records an inside negative resting potential and an overshooting spike
(Fig. 53B). Further advance of the electrode through the cell causes
a loss of resting potential and recording of a positive external potential
associated with the spike. Data of these kinds show the electrocyte
activity to be of conventional polarity, but the shape of the external
potentials is unlike that in other fish discussed up to this point.
1 msec
Fig. 53. Responses of a single electrocyte of the electric catfish. Monopolar
recordings during advance of a microelectrode through the single cell. The horizontal
trace shows the dc level. Stimulation by an external electrode distant from the
recording site and recording as indicated in the inset diagram. A long-lasting pulse
is used to minimize interference by stimulus artifacts. ( A ) Recorded outside the
rostral face, the response is largely negative but is preceded by a small positive
phase. ( B ) When the electrode is advanced into the cell, it records an internally
negative resting potential and an overshooting spike with a shoulder on the falling
phase. ( C ) When the electrode is further advanced to lie outside the rostral face,
the resting potential disappears and a positive going response is recorded. Both
external recordings have inflections on their rising phases at the time of the brief
peak of the intracellular recording. From Keynes et al. ( 1961).10. ELECTRIC ORGANS 429
The origin of the peculiar shape of the external potential becomes
clearer in differential recordings (Fig. 54) from which it appears that
the stalk face generates a small brief spike (followed by a much smaller
hump) and the nonstalk face generates a larger and longer lasting spike.
The sequence of potential changes during propagation of a spike thus
is as follows (see Fig. 4E): outward current depolarizes each face, and
the potential is initially positive outside each face as seen in the monopolar
recordings of Fig. 53A,C. But negligible potential is produced
across the cell because the positivities are about equal. The nonstalk
face, being of lower threshold, begins to pass inward current that flows
out through the stalk face depolarizing it further; the external potential
becomes positive outside the stalk face and negative outside the nonstalk
face. When the stalk face is excited, its activity opposes the spike of
the noiistalk face, and, as a result, there is a reduction in the external
potential or at least in its rate of rise. The spike in the stalk face then
terminates while the spike in the nonstalk face continues. The external
potential becomes much larger and then declines as the spike terminates.
This sequence of firing also explains the shape of the spikes in the two
faces. The spike in the nonstalk face often has two peaks; the larger
potential generated initially is ascribable to reduced loading when the
two faces fire together. Evidently this first maximum need not be
accompanied by maximal sodium activation because the potential can
rise to a second maximum after the brief spike of the stalk face. The
A B
Fig. 54. Responses of a single electrocyte of the electric catfish. Monopolar
and differential recordings from three separate electrodes as diagrammed, stimulation
by a distant external electrode. ( A ) The monopolar recordings are similar to those
in Fig. 53. ( B ) The differential recordings show the external response to be monophasic
head negative with an inflection on the rising phase (Vi), the response of
the rostra1 face to be a relatively long-lasting spike with two peaks (V2), the
response of the caudal face to be a brief spike with a small longer lasting and
probably passive component ( V3). From Keynes et aZ. ( 1961).430 M. V. L. BENNETT
residual potential across the stalk face following its brief spike is
ascribable to voltage drop across this face produced by activity of the
nonstalk face.
What is the significance of this sequence of activity? The resistance of the two faces at rest is about equal. The brief spike in the stalk face is apparently a rapid way to turn on delayed rectification in this membrane,
and by increasing its conductance to increase external current flow. Electron microscopic observations show that both faces have a moderate number of branching tubules that increase their surface areas (Mathewson et al., 1961). The proliferation is slightly greater in the nonstalk face. The similarity in areas is consistent with the suggestion
that both surfaces increase their conductances during a response.
Actually, the stalk face may not be capable of generating an all-or-none
response. The response of this face to depolarization (applied by an
external electrode) apparently is graded, and a propagated spike cannot
be set up by external stimuli that depolarize this face and hyperpolarize
the nonstalk face. It is even possible that this membrane only exhibits
delayed rectification and totally lacks an inward current mechanism.
The initial maximum seen in some external records would then be
ascribable to capacitative current as in cup-type electrocytes of rajids
(Fig. 15). The steep rise and fall of the potential across the nonstalk
face makes this explanation unlikely, but it does not contradict available
data.
From the location of the synapses and the observation that spikes
can propagate along the body of the cell, it appears that a PSP sets up
a spike in the stalk that propagates to invade the rest of the cell. As
would be expected, neurally evoked responses in the body of the cells
are similar in shape to directly evoked ones. Transmission at the synapses
is cholinergic in that curare blocks neural excitation of the organ and
a high concentration of cholinesterase is found histochemically at the
synapses ( Couteaux and Szabo, 1959).
Activity of the stalk can be recorded by electrodes placed (under
visual control) close to the rosette, i.e., the site where the stalk joins
the disc-shaped part of the cell. External stimulation in this region can
excite the cell even if the polarity is such as to depolarize the stalk face
(Fig. 55). External to the rosette opposite the stalk (upper trace) an
initially positive potential is observed apparently resulting from activity
of the stalk, because it is much smaller a short distance away (lower
trace). If the stimulus strength is increased, repetitive firing is produced
(Fig. 53B), which is never observed with stimulation of either polarity
distant from the rosette. If the stimulus strength is increased further,
the impulses fail to invade the body of the cell, but small biphasic10. ELECTRIC ORGANS 431
external potentials remain that are initially positive opposite to the
stalk (Fig. 55C,D). Evidently the hyperpolarization of the nonstalk face
is so large that the stalk impulse cannot excite it. The predominance of
the negative phases in the external records indicates that some excitation
of membrane on the nonstalk side still occurs, although the activity
does not propagate to the remainder of the cell. However, if the stalk
impulse occurs at the end of the stimulus when hyperpolarization of
the nonstalk face is terminated, the impulse in the stalk becomes able
to invade the body of the cell as indicated by a large external response
(Fig. 55D). A stimulus over the rosette that depolarizes the nonstalk
face can also excite the cell, but the threshold is higher than for the
opposite polarity of stimulation (Fig. 55E). When excited in this way
the component of the external potential associated with firing of the
stalk face is larger than when recorded distant from the rosette (Figs.
53 and 54).
One may question why the inflection that is seen on the rising phase
of responses externally recorded close to single cells is generally not
seen in the organ discharge. Since propagation time over the cell is an
Fig. 55. Responses recorded near the stalk of an electrocyte of the electric
catfish. External recording and stimulation as in the diagram, upper trace nearer
the center of the rosette. Anodal stimuli outside the rostra1 face except in E. ( A )
A single largely negative spike is evoked that is preceded by a small but distinct
positivity at the more central electrode. ( B ) A stronger stimulus evokes two
responses. ( C ) A still stronger stimulus evokes two responses of the stalk but
invasion of the body of the cell is delayed and blocked for the first and second
responses, respectively. ( D ) A stronger stimulus evokes three responses of the stalk,
but the first two fail to invade the body of the cell. The third occurs at the end
of the stimulus and does invade the body of the cell. ( E ) Cathodal stimulation
can excite the cell at this site, but the threshold is higher than for anodal
stimulation.432 M. V. L. BENNETT
appreciable fraction of the duration of the spike of the stalk face, the
activity of the stalk face is not synchronous in the different regions of
the cell. It thus fails to produce a distinct component in the overall
organ discharge which at any instant is an average of the contributions
of all regions of all the cells.
3. Gymnurchm
The monospecific genus Gymnarchus is found in tropical Africa. It
is closely related to the Mormyridae, but the mode of swimming is
remarkably similar to that of gymnotids. The animal moves with a
straight body by undulations of the dorsal fin. As far as locomotion is
concerned, it is like an upside-down gymnotid (Fig. 1). Movement
appears equally easy forward and backward, and often the animal
seems to investigate strange objects with the tip of its tail. The electric
organ pulses are emitted at a frequency of about 25O/sec (Fig. 56). The
frequency is not altered by ordinary kinds of stimuli, and “jamming
avoidance” has not been observed ( Bullock, 1970). Novel stimuli that
perhaps startle the fish may evoke a sudden cessation and weak electric
pulses may also be effective (Lissmann, 1958; Szabo and Suckling, 1964;
Harder and Uhlemann, 1967).
The electric organ of Gymnurchus consists of four columns of electrocytes
on each side of the body, one above the other. Each column runs
to the tip of the caudal filament but their anterior extent varies (Fig. 1).
The electrocytes are flattened cylinders innervated on their posterior
faces by spinal nerves. Both faces are moderately convoluted. However,
on a microscopic scale the uninnervated face has a large number of
small canaliculi and processes that greatly increase its surface, while
– 1.2 rnsec
Fig. 56. Electric organ discharge of Gymnarchus. Recorded from an animal
about 30 cm in length in a small container of aquarium water. Zero potential level
indicated by the horizontal line. (A, B) Normal organ discharge at fast and slow
sweep speeds.10. ELECTRIC ORGANS 433
the innervated face is relatively smooth and has only a few canaliculi
(Schwartz et al., 1971).
The electric organ discharge consists of what appear to be headpositive
pulses superimposed on an almost level base line (Fig. 56).
However, the potential goes head-negative between pulses and as in
Sternop ygw, Eigenmannia, and sternarchids there is virtually no net
current flow during one complete organ cycle. (There is a small headpositive
bump between the large pulses; this component is apparently
a result of nerve activity.) The absence of net current flow is a result
of the properties of the uninnervated face of the electrocytes. The
innervated face generates an ordinary spike when depolarized either
synaptically or by applied currents. The uninnervated face has a large
capacity and is of high resistance and inexcitable. Current flow through
it is virtually entirely capacitative. During a spike in the innervated
face the charge on the capacity of the uninnervated face is made more
positive; between spikes the charge on the capacity tends to return to
its initial value. The resulting external potential is diphasic and initially
positive outside the uninnervated face.
The sequence of potential changes and the cell equivalent circuit
are shown in Fig. 3E. The external medium behaves like an ohmic
resistance (at the frequencies of the electric organ discharge). The
external voltage is then proportional to the amplitude of transmembrane
current, and the integral over time of the external voltage gives a
measure of total current flow. Since the charge on the capacity of the
uninnervated face before and after a spike becomes the same during
a steady frequency of firing, no net current can flow through the
capacity and the time integrals of the positive and negative phases of
the external potentials must be equal. This they are observed to be
to within the 1 or 2% accuracy of the measurement.
The response of the single cells to depolarization applied by an
intracellular electrode is shown in Fig. 57 recorded both monopolarly
(Fig. 57A,B) and differentially (Fig. 57C,D). The monopolarly recorded
potentials external to the two faces are diphasic and opposite
in sign (Fig. 57A,B) establishing the longitudinal orientation of the
cells. The potential outside the innervated face is initially negative
indicating that this face becomes active. The differentially recorded
potentials show a simple spike potential across the innervated face. The
potential across the uninnervated face is also a monophasic depolarization,
but its peak is much lower and somewhat later. During the falling
phase, however, the potential across this face exceeds that across the
innervated face, which accounts for the head-negative phase in the434 M. V. L. BENNETT
f i n t e r i 7 6 ,4— B-L-
2 ( ! !
r L i
Posterior u8-67– -r–
Q
Anterior
Jr
V” –
Poster lor
2 rnsec 4 rnsec
Fig. 57. Responses of a single electrolyte of Gymnarchus. Recording and stimulation
as indicated in the diagrams. Faster sweep in A and C. ( A , B ) Outside the two
faces monopolar recordings are diphasic and of opposite polarity. (C, D) Differentially
recorded the response across the innervated face is a large monophasic
spike. The potential across the uninnervated face V, has a much smaller peak amplitude
but decays away more slowly. The resulting external potential is diphasic
initially head positive.
external records. It is not obvious from these data that the uninnervated
face is purely passive and acts as a series capacity. These properties
are established by passing longitudinal currents with an external stimulating
electrode (Fig. 58). The results may be understood with respect
to the equivalent circuit shown in Fig. 59.
If a long-lasting pulse of either sign is applied, there are transient
potential changes across the innervated face at the onset and termination
of the pulse, but the potential across this face is back very close to the
resting value after 20 msec. After the capacity of the uninnervated face
is charged to its new value, no current flows through it and there is
no voltage drop across the resistance of the innervated face (Fig. 58A).
For pulses of this magnitude records for either direction of current are
symmetrical. When depolarized by over 50 mV the uninnervated face
is affected by a small pulse of current exactly as when it is at the resting
potential (Fig. 58B). Both these results indicate the passivity of the
uninnervated membrane.
Although no steady state voltage drop is produced across the innervated
face by maintained current pulses applied externally, there
are transients associated with charging and discharging the uninnervated
face at onset and termination of the currents. If an anodal stimulus is
applied outside the uninnervated face, the innervated face is depolarized10. ELECTRIC ORGANS 435
Anterior
-3 4%
F”
Posterior
Fig. 58. Responses of an electrocyte of Gymnarchtm to externally applied currents.
Stimulation external to the uninnervated face and differential recording as
indicated. ( A ) Oppositely directed current pulses of approximately equal amplitude
produce approximately symmetrical potentials across the two faces. ( B ) When a
brief pulse is superimposed on a long-lasting pulse that depolarizes the uninnervated
face by about 50 mV, the brief pulse produces about the same change in potential
as when it is given alone. Four superimposed sweeps with one, both, and no pulses.
( C ) The onset of an anodal stimulus depolarizes the innervated face; superimposed
sweeps of threshold stimulation. ( D ) The termination of a cathodal stimulus depolarizes
the innervated face; the threshold amplitude for stimulation by a longlasting
stimulus is the same as in C .
by the initial surge and hyperpolarized following termination of the
stimulus (Fig. 58C). If the stimulus is cathodal, the innervated face
is hyperpolarized initially and depolarized after the termination of the
stimulus (Fig. 58B). Provided the pulses are long lasting compared to
the time constant of the system, the voltage across the innervated face
must be identical at onset and termination of stimuli of equal amplitude
but opposite polarity. This is experimentally true for Gymna~chus electrocytes;
in Fig. 58 the threshold for firing at the onset of a cathodal
1-9
Anterior,
uninnervoted face. External medium
R,, = R, = 40 kR (400.Qcm2)
Re2 C, = 005pF(5pF/cm21
Re, f Re, = 70 kR
Posterior,
Fig. 59. Equivalent circuit of an electrocyte of Gymnarchzls. Intracellularly
applied current ( i ) and extracellularly applied current (I) are indicated. Rel, Re?, and
R,, represent the resistance of the external medium. Negligible current flows through
the resistance of the uninnervated face and the capacity of the innervated face and
these elements are shown by dashed lines. The tentative values given for membrane
parameters are referred to membrane area not corrected for surface convolutions.436 M. V. L. BENNETT
stimulus ( C ) is the same as the threshold for firing at the termination
of an anodal stimulus (D). Furthermore, if brief pulses are used to
evoke responses of the innervated face during maintained current pulses,
the amplitudes of the responses are the same as without maintained
current, although they occur when the uninnervated membrane is either
markedly depolarized or hyperpolarized.
One may ask if the membranes of the electrocytes have unique
properties that give rise to the unusual function of the uninnervated
face, From the equivalent circuit (Fig. 59) it is clear that intracellularly
applied current and differential recording across the innervated face
allow evaluation of the resistance of this face ri. The resistances rel or
rez can be evaluated by differentially recording across the cell while
passing current through an electrode external to one or the other face.
Where monopolar recording shows the external potentials outside the
two faces to be equal, then rel = rez. The time constant of decay of
potentials following either intra- or extracellular stimulation is given
by ( rel + ren + r i ) C. Thus from the measured resistances and time
constant one can obtain the capacity.
Although the data available are restricted, preliminary values are
given in Fig. 59. These assume a uniform planar membrane on each
surface. The resistance of the innervated face is quite reasonable. The
capacity of the uninnervated face is if anything too small on the assumption
that its capacity is the usual 1 pF/cm2 and takes into account the
great increase in membrane area resulting from the numerous tubules
and processes of this face. The measurements indicate that the resistance
of the uninnervated face is at least 50 times that of the innervated face.
The great increase in area observed on a fine structural level means
that the uninnervated membrane has a resistance of perhaps 250-500
times that of the innervated face. This is a high value, perhaps 200,000
a ern2, but it is not unprecedentedly great (Bennett and Trinkaus, 1970).
If the capacity of the innervated face were the usual 1 pF/cm2 its time
constant would be 0.4 msec. This value is much shorter than the time
constant of the uninnervated face and is in reasonable agreement with
the experimental observations. The significance of an organ discharge
without net current flow is discussed in Section 11, F.
4. MORMYRIDS
The mormyrids are found in freshwaters of tropical Africa. They are
a large group of some 11 genera and large numbers of species. All that
have been studied are weakly electric. A key to the genera is presented
in Table I11 and Fig. 60. Identification as to species is generally un10. ELECTRIC ORGANS 437
Table I11
Key to the Genera of Mormyrids
The mormyrids are identifiable as small scaled, ray finned fish without adipose fin
(rayless fin caudal to the dorsal), without spines in front of the fins, and without barbels.
The caudal fin is small and generally on a peduncle (Fig. 60). The mouth is small and
there are teeth in both jaws. The eye is covered by a thin epithe1ium.a
a. Anal very short compared to dorsal (ca. one-third the length, Fig. 60A). Mormyrus
(many species)
aa. Anal very long (5 times the length of dorsal, Fig. 60B). Hyperopisus
aaa. Anal about same length as dorsal (no more than 30% difference)
b. Ventral closer to anal than pectoral; body elongate (Fig. 60C). Zsichthys (one
species, Z . henryi)
bb. Ventral closer to pectoral than anal or equidistant between them
c. Teeth in several rows, villiform; terminal mouth with a chin appendage (Fig.
6OD,E). Genyomyrus (one species, G. donnyi)
cc. Teeth in a single row in both jaws, 10-36 in each row; no chin appendage
(Fig. 60G,H)
d. Nares farther from the eye than each other; mouth terminal (at the front
dd. Nares close together and close to the eye; mouth always inferior (Fig. 601).
of the head, Fig. 60F) or inferior. Mormyrops (many species)
Petrocephalus (many species)
ccc. Teeth in a single row in both jaws, 3-10 in each
e. One naris close to the corner of the mouth (Fig. 60J). Stomatorhinus (many
species)
ee. Both nares distant from the corner of the mouth
f . Dentition of upper and lower jaw similar
g. Mouth inferior or subinferior without a chin appendage or elongated
mouth (Fig. 60K). Marcusenius (many species)
gg. Mouth terminal (Fig. 60L) or on elongated jaws, sometimes greatly
so (Fig. 60N); sometimes with short or long chin appendage (Fig.
60M,N). Gnathonemus (many species)
ff. A large pair of incisors in the lower jaw, without a chin appendage
h. Teeth of upper jaw fine conical, anal rather shorter than dorsal (Fig.
hh. Teeth of upper jaw obtuse or bicuspid, anal and dorsal of same length.
60P). Myomyrus (two species)
Paramyomyrus (one species, P . aequipinnis)
Modified from Poll (1957).
reliable with the possible exception of those areas for which keys are
available ( e.g., Greenwood, 1956; Pelligrin, 1923). Identification of
species in our earlier work (Bennett and Grundfest, 1961c) was kindly
provided by Dr. M. Poll of the Muske Royal du Congo Belge, Brussels,
Belgium.
The organ discharges are brief pulses that are emitted somewhat
irregularly at a few per second when the animal is resting, but the discharge
can accelerate to 40 or more per second during swimming (Lissmann,
1958). Accelerations are evoked by most modalities of stimulation10. ELECTRIC ORGANS 439
N
Fig. 60. Characteristics of representative mormyrid genera. ( A ) Mormyrus
caballus. ( B ) Hyperopisus bebe. ( C ) Isichthys henryi. (D,E) Genyomyrus donnyi;
side view and chin process. (F,G) Momnyrops deliciosus, side view and dentition.
( H,I) Petrocephalus sauvagei, side view and dentition. ( J ) Stomatorhinus corneti.
( K ) Marcusenius plagiostoma. ( L ) Gnathonernus leopoldianw. ( M ) Gnathonernus
petersii. ( N ) Gnathonemus numenius. ( 0,P) Myomyrus mucrodon, side view and
dentition. From Poll (1957).10. ELECTRIC ORGANS 441
including resistance changes and weak electric pulses (Szabo and
Fessard, 1965; Moller, 1971 ) . Accelerations can be classically conditioned
as well as operantly conditioned in an avoidance paradigm (Mandriota
et al., 1965, 1968). Brief interruptions of discharge can also occur in
response to novel stimuli and in interaction with other electric fish.
The electric organs lie just anterior to the caudal fin (Fig. 1). They
are made up of four columns of cells and each column contains 100-200
cells in series. The cells are accurately aligned one behind the other.
In many species the body narrows close to the tail to form a “caudal
peduncle” and almost the entire cross section of the fish in this region
is made up of electric organ. Each column of electrocytes occupies one
quadrant and the only other structures are spinal column, skin, and
tendons to the caudal fin.
The cells are innervated by spinal neurons that lie in three segments
in the central region of the electric organ that extends over 8 or 10
segments. The nerve fibers end on stalks as in Mahpterurus and some
gymnotids, but the stalks are much more complex (Szabo, 1958, 1961a;
Bennett and Grundfest, 1961~)A. large number of stalks arise from
the posterior faces (Fig. 61). These stalks fuse repeatedly (binarily )
to form a greatly reduced number of stalks before they are innervated.
In electrocytes of Movmyrus there may be more than 10 separate stalk
systems, each with its own site of innervation. In Mormyrops there are
one or two. In Gnuthonemus, Petrocephalus, and Marcusenius there is
only one.
There is a further complication in some species of Gnathonemus.
After varying numbers of fusions the stalks turn anteriorly and pass
Fig. 61. Anatomy of mormyrid electrocytes. ( A-C) Single isolated cells stained
with methylene blue, ( A ) Caudal surface of a cell from Mormyrus ruma The
darkly stained, thick, branching nerve trunk innervates ( i ) at least eleven separate
systems of stalks ( s ) . ( B ) Rostral surface of an electrocyte of Gnathonemus compressirostris
with penetrating stalks. The regions surrounding the penetrations ( p )
are darkly stained, elsewhere the stalks ( s ) and body of the cell are only lightly
stained. A few stalks are seen running to the site where they are all fused and
innervated ( i ) . Under high magnification details of the stalk system could be
seen, and the stalks on half the rostral surface of this cell are drawn in Fig. 67C.
( C ) Nonpenetrating stalk system torn off an electrocyte of G. compressirostris
and heavily stained. The innervation ( i ) may be seen surrounding the stalk ( s )
at the central region. ( D ) Parasagittal section through the electric organ of G.
tamndua stained with hematoxylin and eosin. Rostral surface up. Bodies of two
electrocytes, about 30 p thick and heavily stained, run horizontally across the figure.
Fine stalks ( s ) , about 10 p in diameter, arise from the caudal surface, fuse, and
pass through penetrations ( p ) in the bodies of the cells to the rostral surfaces
where they leave the plane of section. The regions of innervation ( i ) were included
in the section and lie rostral to the cell bodies. Darkly stained nuclei are
seen in the stalk and in the bodies of the cells. From Bennett and Grundfest (1961~).442 M. V. L. BENNETT
through holes in the body of the cells (Fig, 61). They undergo their
final stages of fusion on the anterior side of the cells. The degree of
fusion before penetration depends on the species and correlates with
the size of the initial head-negative phase of organ discharge as will
be discussed below (Fig. 67). A similar penetrating stalk system occurs
in Mormyrops (Grosse and Szabo, 1960) and in one of two specimens
of a species of Hyperopisus that I have examined. As far as is known
Marcusenius and Petrocephalus have only nonpenetrating stalk systems.
The form of organ discharge varies with the species. In Gnuthonemus,
Marcusenius, Petrocephalus, and Hyperopisus it is essentially diphasic,
initially head-positive although there is a small initial head-negative
phase if there is a penetrating stalk system (Figs. 62, 67). The discharges
can be quite brief, 0.5 msec or less in duration. The discharge of
Mormyrops is also essentially diphasic, but the initial large phase may
be head-positive or head-negative. The head-negative discharges occur
in specimens in which the innervation is on the posterior side, but the
stalks are penetrating (Grosse and Szabo, 1960; Fig. 62B). In Mormyrus
-.. ..: . . .v:
I ,
I .
” –
0 5 rnsec
Fig. 62. Electric organ discharges of representative mormyrids. Recorded in
a small volume of aquarium water between head and tail, head positivity up. ( A )
From Mormyrus. ( B ) From Mormyrops. The electrocytes of this species have
penetrating stalks and are innervated on the caudal side. There are few penetrations
and an initial head positive phase can be seen only with higher gain recording.
( C ) From Petrocephalus. ( D ) From Hyperopisus. The electrocytes from the
specimen of this species with penetrating stalks were innervated anteriorly and
had large numbers of penetrations; correspondingly the initial head-negative phase
is quite large. Superimposed sweeps with ( 2 in D) and without organ discharges
to show baseline.10. ELECTRIC ORGANS 443
the discharge is also diphasic, but the second, head-negative phase is
much larger than the initial phase (Fig. 24) and the entire response
can be longer lasting (Bennett and Grundfest, 1961~S; zabo, 1961).
The single cells operate in a manner similar to those of electric
fish already described (Bennett and Grundfest, 1961~)B. oth faces of
the main part of the cells generate spikes. The external potentials are
diphasic to nearly monophasic depending on the relations between the
two faces.
The contributions of the two faces of an electrocyte from a fish
producing a diphasic output are shown in Fig. 63. A microelectrode
recording monopolarly is advanced through the cell from the rostral
to caudal side (Fig. 63A-C). These records are then subtracted to give444 M. V. L. BENNETT
siveness of the two faces, the surfaces appear similar electron microscopically;
both surfaces are moderately increased by tubules, that of
the anterior face slightly more so (Schwartz et al., 1971).
In Mormyrus r u m the organ discharge is predominantly head
negative. Both faces of the cells appear to generate spikes, although
it is difficult to be sure what fraction of responsiveness on the caudal
side is to be assigned to the stalks (Fig. 64). The spike of the anterior
face is longer lasting and the external potential is predominantly head
negative. The initial slow head-positive phase of the organ discharge
appears to be a result of activity in the stalks and can be observed with
external stimulation of the single cells (Fig. 64A’,B’).
The location of the innervation on the stalks indicates that impulses
arise in this region and propagate to involve the body of the cells. Action
potentials can be recorded in the stalks, and the cell to which they go
can be identified by intracellular stimulation in stalks or body of the
cell. Impulses in stalks are longer lasting than those in the main part
of the cells (Fig. 65). The PSPs that initiate the spike activity can be
Fig. 64. Responses of electrocytes of Mormyrus. A,B and A’,B are experiments
with different cells. ( A ) Three recording electrodes are close to the site of intracellular
stimulation. Simultaneous traces show ( from above down ) stimulating
current and differential recordings across the cell (rostral negativity up) and
across the rostral and caudal membranes. The spike of the caudal face is briefer
than that of the rostral face. The external response is largely rostral negative and
lacks the initial rostral positivity of the organ discharge. ( B ) The same recording
electrodes, but the response is evoked by intracellular stimulation about 1.5 mm
from the recording site. The responses are somewhat shorter but are otherwise
similar to those in A. The small potentials following the spikes result from excitation
of neighboring electrocytes by activity of the penetrated cell. ( A’,B’) Responses
evoked by brief externally applied stimuli. Lower traces: recording across the cell,
rostral positivity up. Upper traces: recording across the rostral face in A’ and the
caudal face in B . The transmembrane potentials are like those in A and B, but
the external response much more closely resembles the organ discharge. Modified
from Bennett and Grundfest (1961~).10. ELECTRIC ORGANS 445
Fig. 65. Recording from an electrocyte stalk of Gnathonernus tamandua. The
stalks are penetrating and the organ discharge is triphasic in this species. Three
electrodes are used as in diagrams; the closely neighboring pair of electrodes are
about 1 mm from the other one. V,, upper trace; V,, lower trace. (A,B) Before
and after penetration of a stalk on the rostral side of the cell, stimulation in a
stalk on the caudal side. Note the polarity of the triphasic external response in A.
About the same potential appears to be superimposed on the broader intracellular
spike in B. ( C ) Stimulation in the rostral stalk, recording in the body of the cell
and in the stalk on the caudal side. The broad spike in the stalk has superimposed
on it the response generated by the body of the cell. The early part of the brief
response is obscure, but it clearly has a late positive phase. This polarity contrasts
to the terminal negativity of the superimposed brief response on the opposite side
of the cell in B. From Bennett and Grundfest (1961~).
recorded with an appropriately placed electrode and are diagrammed
in Fig. 73. The difference in spike shape, the recording of PSPs, and
the fact that an electrode can enter and leave a stalk before penetrating
the body of its cell allow the identification of the records as from stalks.
The discharges of Gnathonernus are very brief, and all regions of
each face of the cells must fire quite synchronously. Direct measurement
shows that a neurally evoked impulse reaches all regions of the posterior
face of the cell highly synchronously. In contrast, stimulation in the
body of the cell does not excite the cell synchronously. Because of the
stalks an impulse initiated at one part of the body of the cell appears
to be conducted along it with a very nonuniform velocity. Close to the
site of stimulation (within 1 mm in the experiment of Fig. 66, right)
the impulse propagates slowly, at a velocity of about 2 meters/sec.
However, the impulse arrives at distant parts of the body of the cell
nearly simultaneously. Evidently, the evoked impulse propagates antidromically
(backward) up the stalks near the site of stimulation and
then proceeds orthodromically to excite much of the cell in nearly the
normal time relations. In cells of Morrnyrus the conduction velocity446 M. V. L. BENNETT
mm
0
0.5
10
20
A
B
C
D
0.5 mm ventral
1.0 mm ventral
2 5 mm ventral
1g% ‘d.orOsm alm
5 msec 0 5 msec
Fig. 66. Conduction along electrocytes of mormyrids. Three electrodes are
used: one for intracellular stimulation, one for intracellular recording close to the
stimulation site, and one exploring electrode that records at various distances
along the edge of the cell. The records on the right are from Mormyrus rume. The
response close to the site of stimulation is shown on the zero trace. Intracellular
responses recorded by the other microelectrode as it is moved along the cell are
below, The traces are displaced downward an amount proportional to the distances
between the two recording electrodes. The spike recorded at the same time with
the fixed electrode is used to position the traces in the horizontal (time) axis.
Since the latency of the responses at the stimulation site vanes, the stimulus
artifacts do not align. The conduction velocity given by the slope of the broken
line is 0.45 meter/sec. On the left are records from Gnathonemus tamandua.
Procedure as for Mormyrus except that the exploring electrode records external
to the rostra1 face. The external responses in this species are triphasic because of
the large number of stalk penetrations (Fig. 67). The conduction velocity appears
to be about 2 meters/sec up to a distance of 1 mm but is much faster between
1 mm and 2.5 mm. Modified from Bennett and Grundfest (1961~).
along the cells is more or less uniform (Fig. 66, left). Not only is the
stalk system divided into separate parts but also the stalks are much
finer and the conduction velocity in them is likely to be lower.
There appears to be an evolutionary progression in going from
the multiple innervation sites to the single site. Mormyrus is apparently
the least complex stage after diffuse innervation (which is found in
the related form Gymnarchus). In Mormyrops the number of sites is
greatly reduced, and in Gnathonemus and most other genera it reaches
its final limit of one. The organ discharge shows a parallel progression
toward brevity, a fact that suggests more precise synchronization can
be obtained when there are fewer sites of innervation. We will return
to the question of synchronization in Section 111, B.
The stalks play an additional roll in the form of the organ discharge.
It is observed that fish possessing electrocytes with penetrating stalks
all produce triphasic organ discharges; there is an initial phase of head
the equivalent of differential recording ( Fig. 63B’,C’). Both rostral and
caudal faces generate spikes. (In this particular experiment the separation
between the two spikes is more marked than it often is, and
external potentials that more closely resemble the organ discharges are
usually obtained, Figs. 65 and 66.) Corresponding to the similar respon-
Fig. 63. Response of an electrocyte of Gnuthonemus. Three electrodes are
used, one for intracellular stimulation (1.6 msec pulse), another for intracellular
recording near the stimulation site (A-C, upper trace), the third an exploring
electrode about 2 mm distant (lower trace). ( A ) The exploring electrode is
external to the rostral face and records a diphasic initially positive potential. ( B )
It is in the cell and shows the resting potential and a double-peaked spike. ( C ) It
is advanced to lie outside the caudal face and records a diphasic response opposite
to that in A. Time calibrations begin from the peak of the intracellular spike in
the upper traces. (A’) Superimposed tracings of records made with the exploring
electrode aligned with respect to the peak of the intracellularly recorded spike.
(B’,C) Potentials across the caudal and rostral faces and diphasic potential across
the entire cell (rostral positivity up) obtained by graphical subtraction of the
monopolar records. From Bennett and Grundfest ( 1961~).10. ELECTRIC ORGANS 447
negativity that precedes the predominantly diphasic response and that
is absent in fish without penetrating stalks (Fig. 67). It is explicable
as a result of longitudinal currents flowing along the organ due to
impulses in the stalks passing through the bodies of the cells. The
sequence of changes in direction of current flow is diagrammed in
Fig. 68. In agreement with this explanation the size of the initial phase
is greater the greater the number of penetrations (Fig. 67). No data
are available on the functional significance of the initial head negativity.
It would appear to be detectable in a fish where it is quite large (such
as Gnathonernus tamandua), but in species where it is very small, it
would seem to be of no significance at all. Behavioral studies may help
to resolve these questions.
The embryological formation of the stalk system is an intriguing
problem, particularly where the stalks penetrate the body of the cell.
The cells are multinucleate and presumably arise by fusion of a number
of cells (Szabo, 1961~)I.n any case the morphogenetic movements required
and the control of cell fusion would appear to be highly involved.
In spite of this apparent complexity, closely related species and perhaps
even different individuals of the same species can have penetrating or
0.2 msec
Fig. 67. Correlation of amplitude of initial head negativity with number of
penetrations by stalks. Camera lucida drawings of one-half of a representative
electrocyte from each of four species of Gnathonmus are shown with tracings of
the discharge of their organs. Black area indicates zone of innervation. The potentials
drawn at the same time scale but with amplitudes normalized to equal height
of the head-positive phase. ( A ) Gnathonemus compressirostris, individuals which
produce no initial head negativity do not have penetrating stalks. ( B ) The largest
initial head-negative phase is in the discharge of G. tamundua, the electrocytes
of which have many penetrations (175 in the drawing). ( C ) A specimen of G .
compressirostris which has a small initial head-negative phase has electrocytes
with a medium number of penetrations (25 in the drawing). (D) The organ
discharge of a specimen of G. moorii shows initial head negativity only in high
gain recordings. Its electrocytes have very few penetrations (5 in the drawing).
From Bennett and Grundfest ( 1961~).448 M. V. L. BENNETT
–
–
–
B …………T… J *; .;\: + ;;-.. …….. ………….. …………… ” …….
+
+
+
Fig. 68. Current flows generating triphasic pulses in electrocytes with penetrating
stalks. Diagrams show a region near a single penetration during different stages
of activity. Active membrane is indicated by dotted outlines. Arrows show direction
of current flow. Resulting potential is shown in lower right. (A) Head negativity
is produced when the stalk activity, initiated near the site of innervation, is passing
through the penetration. ( B ) Head positivity results when the impulse in the stalk
excites the caudal face. ( C ) Head negativity is again produced when the rostra1
face becomes active. From Bennett and Grundfest ( 1 9 6 1 ~ ) .
nonpenetrating stalk systems (Grosse and Szabo, 19600; Bennett and
Grundfest, 1961~).
E. Some Quantitative Considerations
The amplitudes of responses of individual electrocytes in several
of the strongly electric fish tend to be somewhat larger than responses
in other excitable cells. The PSPs in electrocytes of marine strongly
electric fish are up to 90 mV in amplitude, and the peak is near zero
resting potential in Astroscopus. This amplitude is larger than in any
other known cell, but one reason is that most other cells are electrically
excitable and their responses obscure the PSP. The PSP at the neuromuscular
junction of frog twitch muscle fiber would also be very large;
the reversal potential of the PSP is similar to that in the electric fish,
and the conductance change is a substantial fraction of the resting
conductance (Takeuchi and Takeuchi, 1959). In the electric eel
the spike amplitude is about 150 mV, and the sodium equilibrium
potential appears to be about +70 mV (Altamirano, 1955; Nakamura
et uZ., 1965). This compares with values of up to about 130 mV for
other spike generating cells. Voltage amplitudes of membrane responses
in electrocytes are otherwise unexceptional. Of course, the externally
recorded responses of individual cells of even weakly electric organs10. ELECTRIC ORGANS 449
are generally much larger than those of ordinary nerve and muscle cells.
The values of membrane resistance in electrocytes are far below
the values for muscle [about 3000 fi cm2 for frog twitch muscle (Eisenberg
and Gage, 1969)l. Correspondingly, the current outputs are much
greater. In eel electrocytes the resting resistance of the innervated face
is about 25 0 cm2 according to Nakamura et al. (1965); a value of
19 fi cm2 is reported by Cohen et al. (1969) using a different and
perhaps more accurate technique. The resistance of the uninnervated
face is about 0.2 fi cm2 (Keynes and Martins-Ferreira, 1953). During
activity the resistance of the innervated face decreases to about 1 0 cm2
(calculated from Nakamura et al., 1965), and the peak inward current
is about 60 mA/cm2. These values approach the resistance of the extracellular
fluid and cytoplasm in series with the innervated face. Because
of leakage and eddy currents, decrease of the membrane resistances
to much below the series resistance would make the cells less efficient.
In Torpedo the resting resistance across an entire cell is 5-30 fi cm2
( Albe-Fessard, 1950b; Bennett et al., 1961). The peak current during
activity is about 75 mA/cm2 and if the driving force for each cell is
about 90 mV, the series resistance during activity is about 1.2 fi cm2
per cell.
Although the resting and active resistances of the electrocytes are
far below those of muscle membranes, the values are comparable to
those of the node of Ranvier [resting resistance, 30 0 cm2; active
resistance, 2 0 cm2; peak inward current 10 mA/cm2 (Dodge and
Frankenhaeuser, 1959; Frankenhaeuser and Huxley, 1964) 1. Furthermore,
the figures for electrocytes are calculated from macroscopic areas
disregarding the considerable surface convolutions, and a more realistic
comparison would be referred to actual areas of plasma membrane. On
the reasonable assumption that the plasma membranes of the electrocytes
have a capacity of 1 pF/cm2, one may estimate the true membrane
area from the capacities measured per macroscopic area, which in the
eel is about 15 pF/cm2 (Cohen et al., 1969) and in Torpedo is about
5 pF/cm2 ( Albe-Fessard, 1950b). Thus, the actual membrane parameters
probably differ by factors of 5-15 from the reported ones (factors
which appear consistent with the increase in surface seen morphologically,
references below), Weakly electric organs tend to have somewhat
higher membrane resistances, several tens to several hundred Q
cm2 referred to macroscopic surface. Insufficient data are available for
an accurate comparison among weakly electric organs in terms of area
of plasma membrane, but the membranes-with the exception of membranes
acting as a series capacity-all have a shorter time constant than
muscle and presumably are of lower resistivity.
The response characteristic in which electrocytes are most out450 M. V. L. BENNETT
standing is frequency of firing. The electric organs of sternarchids discharge
constantly at frequencies from 700 to 1700/ sec, Eigenmannia
operates at 2sO-600/ sec, and Gymnarchus at about 250/sec. Sternop ygus
discharges at 50-100/sec, but it is like Eigenmannia on a 50% duty cycle
in which the interval between pulses is about equal to the pulse duration.
The higher values are unequaled for maintained frequency by
muscle or other nerve. In fish other than the sternarchids the nerveelectrocyte
synapse is chemically transmitting ( and cholinergic ) so that
chemically mediated transmission can also operate at high frequencies.
However, in gymnotids and Gymnarchus the synapses between the
controlling neurons, which fire at the full organ frequency, may well
all be electrically transmitting ( see Section I11 ).
The high frequency of maintained activity suggests that there
could be a considerable power output per unit weight of organ. As yet
no satisfactory data are available, but many of the cells contain large
numbers of mitochondria suggesting a high metabolic rate (Schwartz
et al., 1971). The peak pulse power of the strongly electric organs is
very large, but the pulses can be emitted at a high rate only for short
times. The energy output presumably represents ions running down
preexisting concentration gradients that are restored relatively slowly.
Electrocytes of strongly electric organs do not have high densities of
mitochondria ( Mathewson et al., 1961; Wachtel, 1964; Sheridan, 1965;
Bloom and Barnett, 1966), and it would be predicted that the power
output per gram of cell that could be maintained would be lower than
for high frequency cells.
In marine electrocytes the site of production of electric energy is
somewhat paradoxically the innervated face which is inexcitable in the
strongly electric fish. It is across this face that the potential is found
during organ discharge. In the eel the contribution of the two faces
is about equal, as it is in many cells of which both faces generate spikes.
For electrocytes in which one face acts as a capacitance the energy is
produced entirely across the innervated face.
F. Adaptation and Convergent EvoIution in EIectric Organs
The evolutionary origin of electric organs will be discussed in the
following chapter, because it appears intimately linked with the development
of electrosensory systems. Ignoring the intermediate stages
of evolution, one notes a number of convergences in electric organ
systems that extend beyond simply the production of electricity (Table
I V ) . Some of these convergences have obvious adaptive value. For
example, it was recognized early that electric organs are relatively10. ELECTRIC ORGANS 451
Table IV
Convergences in Evolution of Electric Organsa
No spikes, marine
Organ flattened
Spikes, fresh water
Organ elongate
Strongly electric organ
Weakly electric organ
Accessory weak organ
Intermittently active
Continuously active
“Constant” frequency
Variable frequency
Can cease firing
Monophasic, both faces respond
Monophasic, one face series R
Diphasic, one face series C
Diphasic, both faces spike
Triphasic
Innervation on stalks
Teleos ts
Gymno tids
-1-
x x
x x
—
—
-X
a Characteristics are shown along the left side; different groups are shown along the
top. A particular group is indicated as having a particular characteristic if a t least one
member of the group has it; a question mark indicates that only suggestive evidence is
available. Thicker vertical lines separate groups in which electric organs are virtually
certain to be separately evolved. A convergence is indicated when a particular characteristic
appears in more than one of the separately evolved groups. Other convergences,
such as development of chin accessory organs, may occur within groups.
I n accessory organs.452 M. V. L. BENNETT
flattened in the two kinds of strongly electric marine fish and elongated
in the two kinds of strongly electric freshwater fish. This difference is
ascribable to the different resistances of the environment. The low
resistance of seawater requires a high current, low voltage output; the
high resistance of freshwater requires a low current, high voltage output.
Another striking difference between marine and freshwater electric
fish is the absence of spike generating membrane in electrocytes of the
three marine groups and its presence in those of the three freshwater
groups. While it is natural to ascribe this difference to differences in
loading by the environments, it is difficult to do so convincingly
(Bennett, 1961, 1970) since each organ consists of many elements in
series-parallel array and the most energetically efficient kind of membrane
would appear to be the best in both environments. Yet there is
almost sure to be some advantage to PSP membrane for the marine
environment because in each case the muscle giving rise to the electric
organ does possess spike generating membrane (Bennett et al., 1961;
Grundfest and Bennett, 1901, and unpublished data).
An interesting convergence is in the development of stalks. These
structures are most remarkable in mormyrids but are also prominent
in catfish electrocytes. Stalks are present but seem of no great importance
in gymnotids. Stalks do allow a smaller PSP current to excite the
electrocytes since the input resistance of the stalks is higher, but the
advantages of this are far from obvious if one considers marine electrocytes
where the entire output consists of PSPs. The stalks act in synchronization
of Ging of mormyrid electrocytes, but the same output
could be produced by diffusely innervated cells, at least in the species
with longer lasting discharges and the more primitive stalk systems.
Another generalization concerns the presence of a dc component
in the organ discharge. All strongly electric fish emit dc pulses; perhaps
this leads to more effective shocking of prey or predator. A few weakly
electric fish that fire at low frequencies also emit pulses that are dc or
have a large dc component, but in most weakly electric fish including
all high frequency species, the organ discharge has little or no dc component.
The absence of a dc component in the organ discharge allows
the fish to have a dual electrosensory system in which one set of receptors
is sensitive to low frequency signals arising in the environment
and another set detects distortions in the high frequency field produced
by the electric organ (see Chapter 11, this volume).
In Gymnurchus, the sternarchids, and perhaps Eigenmannia the
absence of a dc component is achieved by modification of one electrocyte
face to pass only capacitative current. In Sternopygus the uninnervated
face apparently acts as a polarization capacity. In others10. ELECTRIC ORGANS 453
such as Gymnotus and most mormyrids, the opposed faces of the electrocytes
are active sequentially to achieve the same effect. Other convergences
in terms of pulse shape and patterns of organ discharge can
be found. Mormyrids are similar to the variable frequency gymnotids,
and Gymnarchus is similar to the constant frequency gymnotids. The
more striking differences are sure to be associated with differences in
the operation of the sensory part of the system, and some of the subtler
ones may be associated with species recognition.
The fine structure of electrocytes shows a few correlations with
their functional properties. Where one face is low resistance and inexcitable
and the other face generates PSPs or spikes, the surface of
the low resistance face is greatly increased by projections or invaginations
[ torpedinids, electric eel, monophasically discharging Hypopomus,
accessory organs of Steatogenys (Mathewson et al., 1961; Sheridan,
1965; Bloom and Barnett, 1966; Schwartz et al., 1971)l. The area of
the active face may also be increased but to a smaller extent. The relative
areas correlate with the changes in resistances during activity
and are such that these resistances tend to become more nearly equal.
The probable significance of this feature in terms of increased efficiency
was noted in Section 11, B. In Gymnarchus one face acts as a series
capacity and this face is markedly increased in area; the morphological
relations are similar in Eigenmannia (Schwartz et al., 1971). In
Sternopygus both uninnervated and innervated faces are quite smooth,
which supports the hypothesis that the very large apparent capacity
is a result of shifts of ionic concentrations rather than a dielectric
capacity. Furthermore, the resistances during and between spikes are
about equal.
A number of electrocytes in which both faces generate spikes have
been studied at the fine structural level [Gymnotus, Steatogenys main
organ, Gnathonemus, and Malapterurus (Mathewson et al., 1961;
Schwartz et al., 1971)l. In each case there is some proliferation of both
faces, but the degree is always less than in the inexcitable faces of
monophasically responding cells. The proliferation in diphasically responding
cells is somewhat greater in the uninnervated or nonstalk
face; that in all but Malapterurus is the higher threshold, later firing
face. The morphological difference may be an adaptation to the fact
that the earlier firing face must operate through the series resistance
of the later firing face at rest and that the latter should therefore have
a low resting impedance. In Gymnotus at least the time constant of
the later firing face is greater than that of the earlier firing face, and
a significant fraction of the early outward current through the later
firing face may be capacitative. The activity of the later firing face454 M. V. L. BENNETT
occurs during the phase of increased conductance owing to delayed
rectification (potassium activation) in the earlier firing face, and the
area of this face need not be so great. One would not be surprised to
find that delayed rectification was more pronounced in the earlier firing
face, but this possibility has not been investigated. Delayed rectification
is more marked in Gymnotus cells than in those of Steatogenys, which
may account for the smaller degree of proliferation of both surfaces
in Gymnotus as compared to Steatogenys.
Although some nerve-electrocyte junctions are continuously active
at very high rates, no morphologically distinctive features have been
found (Schwartz et al., 1971). All investigated do not appear significantly
different from neuromuscular junctions.
It is a bit disappointing that current electron microscopic techniques
have not revealed any further correlations with function. One cannot
yet distinguish between a high resistance inexcitable membrane and a
low resistance excitable membrane. This failure probably reflects the
small fraction of membrane surface actually involved in ionic movements
( cf. Hille, 1970). Perhaps freeze-cleaving techniques, which
allow one to examine relatively large areas of membrane en face, will
provide sufficient resolution. After all, it is to be expected that ionic
channels, although themselves small, have associated with them protein
molecules well within the range of resolution of the electron microscope.
An important kind of adaptation in electric fish is the arrangement
of connective tissue. In many electric organs, connective tissue appears
to provide an insulating sheath that channels current flow along the
axis of the organ. An example from Gymnotus is given in Fig. 69. In
Torpedo the resistance of the skin on the dorsal and ventral surfaces
of the organ is lower than that over the rest of the body (Bennett et al.,
1961). This difference tends to maximize current in the external medium
and minimize current flowing through the animal’s body.
An old but frequently recurring question is why strongly electric
fish do not shock themselves. Their resistance to their own discharges
as well as those of other fish is no doubt contributed to by the heavy
fat and connective tissue layers that surround much of the nervous
system. For example, the electromotor axons of Malapterurus are rather
small, but they are surrounded by a sheath so that their apparent
diameter is about 1 mm (in a fish about 15 cm long). This sheath
becomes attenuated as the fiber branches, but it nevertheless extends
beyond the synaptic terminals and part way down the electrocyte stalks
(Mathewson et al., 1961). One would not be surprised if the hearts
of strongly electric fish were less easily driven into fibrillation by shock
or if the electroconvulsive threshold were lower for their neural tissue,
but no data have been obtained.
Another answer to the question of why electric fish do not shock10. ELECTRIC ORGANS 455
Fig. 69. Influence of connective tissue on external fields of electrocytes of
Gymnotus. The diagram shows portions of the connective tissue tubes of the three
ventral columns of electrocytes, two of which are indicated by heavy lines in each
tube (see Fig. 19). Septa between chambers are shown as dotted lines. Rostra1
is toward the top. The lower cell in tube I11 was stimulated intracellularly and a
third electrode explored the external field. Upper and lower records of each pair
represent the response before and after penetrating the connective tissue tubes.
The diphasic external response is initially positive on the rostral side. The external
potentials are considerably larger inside tube I11 than elsewhere. The potentials
decrement relatively little along the tube. A subthreshold PSP is evoked in the
rostral cell of tube 11. From Bennett and Grundfest (1959).
themselves is that they do. Often an eel will twitch when it discharges
its main organ. Self-stimulation becomes more prominent when the fish
is out of water because the voltage developed across the animal’s
own tissue is increased. The electric catfish will normally not move
when it discharges, but if a cut is made through the organ to the body
wall each discharge will cause a twitch. Evidently opening of the
inner lining of the organ allows more current to flow through the body.
When mormyrids are placed in air, even they may show signs of selfstimulation.
Organ discharges cause twitches of tail muscles and small
after-discharges of the organ, probably because of electrical stimulation
of the spinal cord (Bennett and Grundfest, 1961~).
G. Embryonic Origin and Development of Electric Organs
There is now little doubt that except in sternarchids all electric
organs evolved as modifications of muscle. The muscle groups of origin
can be located at any point along the body from eye muscles in Astro456 M. V. L. BENNETT
scopus to branchial muscles in torpedinids, to pectoral muscles in
Mahpterurus, to axial and tail muscles in the remainder. In electrocytes
of Astroscopus, rajids, and mormyrids striations are easily seen under
the light microscope. The orientation however is quite unrelated to
the cell axis. Electron microscopy shows well-organized Z lines and thin
filaments in both Astroscopus and rajids. Gymnurchus electrocytes have
similar but less regular structures (Schwartz et al., 1971). In the
mormyrid Gnethonemus there are both thick and thin filaments and
Z lines. The filaments are not very well aligned, however, and the cross
bridges of normal muscle are certainly reduced in number if not absent
altogether. Sarcoplasmic reticulum, associated in normal muscle with
control of tension, is lacking in most regions or very sparse. It is not
surprising then that the cells do not move, although it would be
interesting to know the biochemical “deficiency,” if any, in the contractile
machinery.
Electrocytes of the other three evolutionary lines of electric fish,
torpedinids, gymnotids, and Mahpterurus, all have fine but disorganized
filaments without Z lines (Wachtel, 1964; Schwartz et al., 1971). The
origin of these tissues from muscle is less obvious in the adult, although
the development of the organ from myoblast-like tissue has been
demonstrated histologically (Johnels, 1956; Keynes, 1961; see also
Fritsch, 18%). The physiological and pharmacological properties
strongly support derivation from muscle. Some histological aspects of
electric organ development have also been described for Gymnurchus,
Astroscopus, and Mormyrus (Dahlgren, 1914, 1927; Szabo, 1961~).
An interesting problem is the ontogeny of excitability in electrocytes
of marine fish. Rajid electrocytes develop from rather normal appearing
muscle fibers (Ewart, 1892). If they are normal, the spike generating
membrane present in tail muscle of the adult (Grundfest and Bennett,
1961) would have to be lost. Embryological material is readily obtainable
for this group unlike that for most other electric fish. Increase in the
number of electrocytes during growth may also provide an opportunity
for experimental analysis. Gymnotids regenerate their tails including
electric organs (Ellis, 1913) and electric eels apparently add layers of
electrocytes as they increase in length (Keynes, 1961).
The existence of a neurogenic accessory organ in the chin region of
Adontosternarchus (Section 11, D, 1, g ) has developmental as well as
evolutionary implications. It suggests a kind of preadaptation but one not
involving a purposeful change toward a structure that has survival
advantage only at some later time. A probable early stage in the chin
organ development was maintained oscillatory activity of a group of
electroreceptors (see Chapter 11, this volume) that served as a weakly
electric organ for other receptors. The afferent fibers would then have10. ELECTRIC ORGANS 457
become modified to resemble the fibers of the main electric organ. The
independent evolution of rostra1 accessory organs in other gymnotids
attests to the value of these organs. Is it to be supposed that the sensory
fibers of the primitive accessory organ gradually enlarged over many
generations to repeat the evolutionary sequence followed in the main
organ? It seems much more likely that having evolved the developmental
mechanism to make large generating nerve fibers in its main
organ, the fish merely evolved the ability to apply the same mechanism
to the chin fibers. In molecular terms, DNA coded information required
to make the main organ fibers was also present in the DNA of the
sensory neurons; the fish then acquired a way of expressing this information
at the different site. In brief, the argument is that there is evolution
of a mechanism of turning-on in a new part of the organism a previously
evolved very complex developmental sequence. This process appears
much more probable than evolution of a very similar complex sequence
all over again.
Evolution of a structure in one part of the body and then the sudden
appearance of a very similar structure in a quite different part of the
body may not be uncommon in phylogenetic history. Another example
from electric fish is Hunter’s organ of the eel, which appears to be
evolved from a different muscle group than the main and Sachs’ organs,
but which has very similar electrocytes (see Section 11, D, 1, a ) . Other
examples are discussed in Bennett ( 1970). In accessory electric organs of
other gymnotids and Narcine the innervation suggests that the organs
developed by migration of electrocytes from the main organ rather than
by development from a different and local muscle group (see Sections
11, C, 2 and 11, D, 1, d).
This concept of preadaptation also raises the possibility that electric
organs may have arisen in elasmobranchs only once. The argument for
separate evolution is based on the muscles of origin being different in
torpedinids and rajids. The organs could have originated in one muscle
type in a common ancestor, then “jumped to a second muscle type, and
finally have been lost at the first site. The rajids seem the more generalized
of the two, and a hypothetical electric common ancestor would
probably have been rajid-like. The same argument appears inapplicable
to separate evolution of electric organs in different teleost groups because
of their much wider evolutionary separation.
H. Electrocytes as Experimental Material
Many electrocytes are very large cells that are of low resistance and
easily studied by microelectrode techniques. Their greatest advantage458 M. V. L. BENNETT
over muscle is that they do not move when stimulated. However, the
large size and low resistance is often a disadvantage because the membrane
potential may not remain uniform over the cell when it is stinwlated
by a single intracellular microelectrode; that is, the cells are not
“space-clamped ( cf. Bennett, 1970). Thus, qualitative results can be
obtained using microelectrodes, but quantitative characterization of
membranes in terms of Hodgkin-Huxley parameters as has been carried
out for other tissues may not be possible (Hodgkin and Huxley, 1952;
Frankenhaeuser and Huxley, 1964).
Because electrocytes have a uniform orientation, it is sometimes
feasible to use external electrodes on columns of cells or even on the
entire electric organ. This kind of preparation allows satisfactory impedance
measurements which can be referred to the single cell (e.g., Fig.
41; see also Albe-Fessard, 1950b). Crude but useful current clamp
measurements may also be possible where the properties of the
individual cells are sufficiently alike that the current density through
them is essentially uniform (e.g., Bennett et al., 1961). The uniform
orientation also has allowed measurements of thermal and optical
changes associated with activity; most important, these changes can be
studied under different degrees of electrical loading of the tissue (Cohen
et al., 1969; other references in Bennett, 1970).
Single eel electrocytes can be isolated. The single cell can then be
placed between two baths and transcellular current restricted to a limited
area by pressing the innervated face of the cell against a plastic sheet
with a mall hole through it. Current application by external electrodes is
more or less uniform because the uninnervated face is inexcitable and of
low resistance. This preparation has been used in voltage clamp and
impedance measurements, although it is not clear that space clamping is
possible even under these conditions (Nakamura et al., 1965; Morlock et
al., 1969; cf. Bennett, 1970). As shown in respect to the squid axon, space
clamping may fail if access resistance through the surrounding solution
is too large compared to membrane resistance (K. S. Cole, 1968). It is
difficult or impossible to clamp the squid axons showing the largest inward
currents, and eel electrocytes pass considerably larger currents.
Another problem is whether the small stalks on the innervated faces protrude
far enough to be nonisopotential with neighboring membrane (as
in several weakly electric gymnotids, Figs. 28 and 31). In the eel
preparation the effects of membrane outside the edges of the window
also must be evaluated. These factors require careful analysis as did
voltage clamp of the squid giant axon, and it will be necessary to use
exploring microelectrodes to verify space clamping. In spite of the
possible or real shortcomings, gymnotid electrocytes allow voltage clamp10. ELECTRIC ORGANS 459
ing superior in respect to temporal resolution to what has been possible
with skeletal muscle fibers (Adrian et al., 1970). One important measurement
that has not been obtainable from muscle fibers is the high speed
of onset and reversal of anomalous rectification (Bennett and Grundfest,
1966; Nakamura et al., 1965; Morlock et al., 1969). The primary factor
is that the time constant of the excitable membrane is much lower in
electrocytes and there is less interference from capacitative currents.
Isolated eel electrocytes can be of value in flux measurements
( Higman et al., 1964), although the spatial uniformity of concentration
as well as potential becomes important and the conductances are SO large
that intracellular concentrations can change rapidly ( Karlin, 1967). Still
the large amounts of synaptic as well as spike generating membrane and
the possibility of approximately controlling transmembrane potential
during drug application and flux measurements are features not readily
available in muscle.
Electrocytes do not appear as good as muscle for electrophysiological
study of many aspects of synaptic transmission. The presynaptic fibers
are no larger than those in muscle and usually do not end in a localized
but accessible region. The low input resistance of the cells makes them
less useful for study of miniature PSPs and actions of restricted synaptic
areas. The low input resistance and wide distribution of synaptic membrane
also impede studies using iontophoretic application of drugs, which
remains the best method for study of kinetics of drug action. Nonetheless,
through experimental simplicity, electrocytes have been and should
continue to be useful in studies involving relatively gross application of
drugs during current application and recording of PSPs (e.g., Karlin,
1969).
The kinds of problems where electrocytes are likely to be particularly
useful are in the area of biochemistry. The organs are a rich source of
acetylcholinesterase ( Leuzinger and Baker, 1967), and much of the work
characterizing the enzyme has been done on material of this origin.
Moreover, the eel electric organ is probably the richest known source of
the sodium-potassium transport ATPase and has been used in studying
this enzyme ( Albers, 1967; Post et al., 1969). Eel electric organ does not
have a particularly high concentration of mitochondria compared to a
number of muscles, which suggests that its mean metabolic rate is not
particularly high. Nevertheless, the only work it does is generation of
electricity. Thus, it is reasonable that its sodium-potassium ATPase levels
should be very high. From the much greater concentration of mitochondria
in electrocytes of repetitively active weakly electric fish
(Schwartz et al., 1971) one would expect there to be considerably higher
levels of the transport ATPase. Of course, the weight of organ available460 M. V. L. BENNETT
is measured in grams rather than kilograms. When macromolecules of
excitable membranes are to be isolated, electric organs are probably a
good place to start. The ability to choose by choice of organ the type of
excitability present in terms of PSP or spike generating membrane may
be helpful in the future isolation procedures.
The electric organ of Torpedo is now being used for the isolation of
vesicles containing acetylcholine (Israel et al., 1968). Electric organs
seem to be much better tissues than guinea pig brain in which to look
for vesicles containing acetylcholine when only a very small fraction of
interneuronal synapses can involve this transmitter. Another probably
useful tissue for vesicle isolation is the electromotor lobe of the Torpedo
brain. Available evidence indicates that synapses on the electromotor
neurons are chemically transmitting, but the transmitter is unlikely to be
acetylcholine (see Section 111, B ) . The electromotor lobes provide a
tissue sample of up to a gram of what appear to be neurons of a single
type with a single class of synaptic ending on them. While the weight of
tissue is small compared to the kilograms of electric organ, the size is
very large compared to other neuronal groups of comparable
homogeneity.
111. NEURAL CONTROL OF ELECTRIC ORGANS
For most electric fish it is of considerable importance that the firing
of individual electrocytes be synchronous. Synchrony leads to an output
that is larger in terms of both voltage and power because inactive cells
act as a shunt or series resistance and thus reduce the amount of current
that active cells produce outside the fish. When the discharge is diphasic
or triphasic, synchronization is particularly important because slightly
out of phase addition leads to cancellation. As has been seen in the
section on electrocyte activity, organ discharge generally involves one
highly synchronized response of each cell. Only weakly electric organs of
marine fish appear to have discharges that involve fused and repetitive
activity of the electrocytes.
There is another fundamental consideration in control of organ discharge;
namely, that two spike generating membranes arranged in series
tend to desynchronize each other’s activity. Inward current generated by
one excitable membrane tends to hyperpolarize the corresponding membrane
of the next cell in series with it and thus prevent its firing (see
Fig. 2). To achieve synchronous activity each cell in series must be
separately innervated and its discharge controlled centrally. Of course,
cell membranes in parallel may tend to excite one another as an action10. ELECTRIC ORGANS 461
potential propagates along a single membrane. Still separate cells are
always separately innervated. One exception to the rule of central control
of organ discharge is known, the chin organ of Adontosternarchus. Since
this organ consists of a single layer of parallel elements, the spontaneous
activity of the cells tends to be synchronized and the discharge frequency
is set in the organ itself without central control (see Section 11, D, 1, g).
As might be expected, this organ often operates at a somewhat different
frequency from the main organ.
Control of electric organ discharge can be divided into two problems,
how the fish “decides” to discharge its organ, and, having reached the
decision, how it activates the different generating elements synchronously.
The “decision” to discharge the organ is reached, depending on
the kind of fish, by a small group of neurons in the higher spinal cord,
medulla, or perhaps midbrain ( Bennett et al., 1967a,b,c; Bennett, 1968a).
These cells make up the “command nucleus,” and when they fire synchronously,
the “command signal” to discharge the organ is initiated.
This activity is then transmitted to the electrocytes either directly or
through one or more neural relays.
Neurons of the command nucleus are probably spontaneously active
in the continually discharging forms; that is, they are pacemaker neurons
in a manner analogous to pacemaker cells in the heart. In species that
discharge only intermittently, the command neurons receive excitatory
inputs and perhaps inhibitory inputs as well. When these inputs reach
threshold, the neurons “decide” that the organ will be discharged. A very
significant feature of the pacemaker or command cells is that they are
coupled to each other by means of “electrotonic synapses.” The coupling
is the basis of the highly synchronous firing that is observed in command
neurons of the electric organ systems. The electrotonic synapses provide
resistive pathways between cells that for this reason behave as if they are
part of the same core conductor; potentials spread between the cells in the
same way as they spread electrotonically along an axon. Two important
properties of electrotonic synapses are relevant to their functioning in
electric organ control. Current can flow in either direction across the
synapses, and current begins to flow without delay when pre- and postsynaptic
potentials differ. These properties differentiate electrotonic
synapses from chemically transmitting synapses in which transmission is
basically in one direction and postsynaptic current is delayed with respect
to presynaptic impulses (by about ?h msec at room temperature). Electrotonically
mediated PSPs are delayed with respect to presynaptic potentials
because of the capacity of the postsynaptic cells, but generally the
delay is very short compared to the delay at chemically transmitting
synapses (Bennett, 1966).
The electrotonic synapses mediate rapid-acting positive feedback be462 M. V. L. BENNETT
tween cells; a relatively more depolarized cell tends to depolarize and
excite its less depolarized neighbors and is itself made less depolarized
and inhibited by them. Thus, the neurons tend to fire synchronously. The
synapses are both excitatory and inhibitory, and it seems reasonable
to call them synchronizing synapses (Bennett, 1968b). One may ask
whether mutually excitatory, chemically transmitting synapses could also
mediate synchronization, As will be seen below synchronization in most
of the electric organ systems is so precise that the negligible delay of
electrically mediated transmission is required; the delay associated with
chemically mediated transmission would be too great.
The experimental demonstration and several properties of electrotonic
transmission are illustrated by Fig. 70. In this experiment neighboring
cells in the oculomotor nucleus of the stargazer are penetrated by
microelectrodes; the antidromic spikes evoked by stimulation of the
oculomotor nerve are shown in Fig. 70A. Depolarizing or hyperpolarizing
current applied in either cell spreads to the other cell (Fig. 70C-F).
If one or both electrodes are placed in a just extracellular position the
recorded voltages are very greatly reduced. They do not disappear completely
because the currents develop some voltage across the volume
resistance of the neural tissue. It can be concluded that the voltages
recorded intracellularly involve a special, junctional relation between
A – ‘ 6 C-E-
\&2 -&-.-s:
— I–.-B= D q FL-l?
~ z e c
a
% 5msec _ I T
10 msec In
“-r-.w
Fig. 70. Electrotonic coupling of oculomotor neurons innervating the electric
organ of the stargazer Astroscopus. Two neighboring cells are penetrated by independently
mounted electrodes. Their antidromic spikes are shown in A. The
upper trace shows the antidromic volley recorded at the point of exit of the
oculomotor nerve from the cranial cavity. When current is passed through the
electrode in the cell of the upper trace both depolarization (C) and hyperpolarization
( D ) spread from cell to cell (superimposed sweeps with and without a pulse;
current strength shown on the lower trace). When current is passed through the
other electrode, depolarization ( E ) and hyperpolarization ( F ) also spread from
cell to cell. When depolarization of the first cell is adequate to evoked spikes, there
are corresponding small deflections in the second cell in addition to the maintained
depolarization (B, increasing stimulus strength from top to bottom). The voltage
gain is the same in B-F; the sweep speed is the same in C-F. From Bennett
( 1968b ) ,10. ELECTRIC ORGANS 463
cells and are not merely a result of proximity. Other less direct methods
of demonstrating electrotonic transmission may be useful; for example,
a PSP of very short latency can be assumed to be electrically mediated
(cf. Bennett, 1966).
The morphological basis of electrotonic coupling has been investigated
in a number of systems (Pappas and Bennett, 1966; Bennett et al.,
1967a,b,c; Kriebel et al., 1969; Pappas et al., 1971; these papers include
other references ) . In every case close membrane appositions occur (Figs.
71 and 72) which are rare or absent in neighboring regions and which do
not occur at synapses for which there is evidence that transmission is
chemically mediated. These close appositions are believed to be the site
of current passage between cells. They are probably what have recently
been termed gap junctions (Revel and Karnovsky, 1967; Brightman and
Reese, 1969). This name arises from the appearance in perpendicular
sections of a 20-30 A gap between membranes following suitable fixation
procedures. The gap is penetrable by marker substances applied in the
extracellular space. The gap is not uniform but is made up of a more or
less hexagonal lattice of channels. There is evidence that in the spaces
outlined by the lattice, channels separated from extracellular space cross
the junctional complex and interconnect the cell cytoplasms. The intercytoplasmic
channels are perhaps 10A in diameter and provide sites for
movement of ions and other small molecules between cytoplasms of the
coupled cells (Payton et al., 1969; Pappas et al., 1971). This class of
junctions is distinct from tight junctions, which are appositions where
extracellular space appears completely occluded in perpendicular sections,
and where there is no hexagonal structure in tangential sections
( Brightman and Reese, 1969). (Both types were formerly called tight
junctions. The current nomenclature in this area is confusing and will
probably be revised when there is more general argreement as to morphological
and physiological properties of the junctions and criteria for
their identification. ) These latter junctions generally occur in epithelia
where they form complete rings around cells (zmulae occludentes;
singular, zonula occludens) that prevent transepithelial leakage through
intercellular clefts ( Farquhar and Palade, 1963; Brightman and Reese,
1969). There is no evidence that zonulae occludentes form low resistance
channels between cell cytoplasms because no cells are known to be both
electronically coupled and joined exclusively by these junctions.
Electrotonic coupling between cells of the same kind can be mediated
by electrotonic synapses between somata or dendrites (Fig. 71). Alternatively,
cells can be coupled by way of presynaptic fibers that form
electrotonic synapses on the cells; current then spreads from one postsynaptic
cell to another through the presynaptic fibers (Fig. 72). (Since10. ELECTRIC ORGANS 465
activity can normally be conducted in either direction across some of
these synapses, pre- and postsynaptic denote the usual morphological
relations rather than the direction of impulse propagation. ) Coupling can
also be mediated by both presynaptic fibers and dendrodendritic
synapses in the same nucleus ( Pappas and Bennett, 1966).
A. Pathways and Patterns of Neural Activity
The circuitry of electric organ command systems in teleosts that have
been studied physiologically is diagrammed in Fig. 73. The physiological
analysis has generally involved microelectrode recording from the entire
animal paralyzed or anesthetized.
1. THE ELECTRICCA TFISH
The simplest control system is in the electric catfish (Bennett et al.,
1967b). Two neurons lie in the first spinal segment, one on either side
(Fig. 72A). Each neuron innervates all the electrocytes on its side and
each impulse fires the electrocytes once. The two neurons are closely
coupled electrotonically, and hyperpolarization produced in one cell
spreads to the other (Fig. 74C ) . The coupling is so close that an impulse
initiated in one cell propagates into the other (Fig. 74B). The pathway
of coupling is by way of presynaptic fibers (Fig. 72). Excitatory inputs
gradedly depolarize the cells (Fig. 74A,D), and when one cell is excited
the other must fire also. This explains why no stimulus can be found that
excites one cell without exciting the other.
That the two cells comprise the command nucleus is indicated by the
gradual rise in potential when organ discharge is evoked by cutaneous
stimulation (Fig. 74D). When PSPs in one cell exceed threshold, both
cells are depolarized since the presynaptic fibers end on both; once initiated
the impulse rapidly propagates between the cells. The conduction
Fig. 71. Morphological basis of electrotonic coupling: dendrodendritic junctions
between electromotor neurons of the mormyrid Gnathonemus. ( A ) In a silver
stained preparation ( Romanes’ method) of the medullary relay nucleus, a thick
bridge appears to connect the two cell bodies without there being any intervening
membrane. ( B ) With the electron microscope such cells are seen not to have
cytoplasmic continuity. The cell bodies ( s ) are separated by membranes the ends
of which are indicated by the arrows. Blood vessels (bv) and myelinated nerve
fibers are seen. Axon terminals ( a ) make contact with the cells; one on the lower
right shows a portion of the myelin sheath. ( C ) At a similar region of apposition
between spinal electromotor neurons, higher magnification shows large regions where
the membranes appear fused. In this very thin section the central dark region
appears as a series of dots. Osmic acid fixation. From Bennett et al. (1967a).10. ELECTRIC ORGANS 467
requires less than 0.2 msec, which is more rapid than could be achieved
by chemically mediated transmission ( Fig. 74A). [The delay is greater
when one cell is directly stimulated, for the unstimulated cell is much
less depolarized when the impulse arises ( Fig. 74B) .]
2. WEAKLYE LECTRIGCY MNOTIDS
In weakly electric gymnotids the command system has more than
one neural level (Bennett et al., 1967~)T. he organ frequency is set by
what has been termed the pacemaker nucleus, which is in the medulla.
(From a functional point of view it would be difficult for synchronous
activity to arise within a group of neurons that are as widely separated
as the spinal neurons innervating the electrocytes. ) The pacemaker
nucleus lies in the midline and contains some 30 to 200 cells depending
on the species. It activates a relay nucleus also in the midline of the
medulla, and this nucleus in turn activates the spinal neurons. There are
50 or so medullary relay neurons and some hundreds to thousands of
spinal neurons. A single spike occurs at each level before each organ
pulse (Fig. 75). In the pacemaker neurons there is a gradually rising
depolarization between spikes (Fig. 75A,C). This depolarization is
similar to pacemaker potentials in other tissues, and the cells appear to
be spontaneously active. In the relay cells the potential between spikes
is quite flat and the discharges arise abruptly from a level base line (Fig.
75B,D). These cells are clearly activated from a higher level which is in
fact the pacemaker nucleus.
Fig. 72. Morphological basis of electrotonic coupling: axosomatic and axodendritic
electrotonic synapses on giant electromotor neurons of the electric catfish.
( A ) Toluidine blue stained thick section that passes through the nuclei of the
two giant cells. The central canal (arrow) is just ventral to the cells. A number
of small dendrites come off the somata, but there is no apparent direct connection
between the cells. ( B ) Electron micrograph of two axosomatic synapses (a,a’) on
the cell. At the upper one ( a ) the myelin sheath is seen to terminate in the plane
of section. There is a region of close apposition of axon and soma membranes
at this synapse (between arrows), but probably not a t the other one. Terminal
a’ contains many vesicles, but there are relatively few in a. The former may be
one of the chemically transmitting inhibitory synapses that occur on these cells.
Near these endings there is a relatively large amount of extracellular space ( e )
filled with granular material. (C) Higher magnification of an axodendritic synapse.
The axonal side is to the left and a presynaptic vesicle ( v ) is seen in it. A central
dark line is formed at the close membrane apposition extending between the
arrows. ( D ) An axodendritic synapse like that in C, but the close membrane
apposition is cut more tangentially and striations appear with a periodicity of
about 100A. ( E ) Diagram of current flow where cells are coupled by way of
presynaptic fibers. The cell on the left is more depolarized. All sections of osmic
acid fixed material. Modified from Bennett et al. (1967b).468 M. V. L. BENNETT
Malaplerurus Gymnot ids Astroscopus
Midbrain
–
Medulla
–
Spinal cord
Periphery -i i –
Mormyrids
Pacemaker:
?@% *– dsuobomr!=i nan t—-
– –
Medullary
relay –
Spinal
relay
:l&ro p la qx
1
Organ ‘2Gec
X Dendrodendritic A~~~~~~~~)ei~ec~tri cal A Axosornatic- chemical
Fig. 73. Neural circuitry controlling electric organs of teleosts. The modes
of transmission are diagrammed as shown in the key below. Where axosomatic
synapses are indicated, axodendritic synapses are also found. The mode of transmission
to the command nucleus is known only in the electric catfish, although it
is indicated as chemically mediated in the others. Where there is a question mark,
the cells have not been definitely localized, but several of their properties can be
inferred. The command nucleus of the stargazer Astroscopus is now known to be
in the medulla, not the midbrain. In Malapterurus, the gymnotids, and Astroscopus,
a single command volley at each level precedes each organ discharge. In mormyrids,
the activity is more complex and is diagrammed for each level. The dotted lines
indicate the thresholds of the cells in the two pacemaker nuclei. The dominant
pacemaker at any given time is the one receiving more excitatory inputs and firing
before the other. From Bennett (1968a).
That the discharge frequency is set in the pacemaker nucleus is
established by results like those in Fig. 76. If a hyperpolarizing pulse is
applied in one cell, the next spike in that cell and each subsequent one
is delayed (Fig. 76B,D). The descending volleys in the spinal cord are10. ELECTRIC ORGANS 469
2 msec 10 msec
Fig. 74. Properties of the giant electromotor neurons of the electric cadish.
( A ) Upper and lower traces, recording from right and left cells, respectively. Brief
stimuli of gradually increasing strength are applied to the nearby medulla (several
superimposed sweeps, the stimulus artifact occurs near the beginning of the sweep).
Depolarizations of successively increasing amplitude are evoked until in one sweep
both cells generate spikes. ( B ) Two electrodes in the right cell, one for passing
current (shown on the upper trace) and one for recording; one recording electrode
in the left cell. The traces from the recording electrodes are the lower ones starting
from the same base line. When an impulse is evoked in the right cell by a depolarizing
pulse, the left cell also generates a spike after a short delay. ( C ) When
a hyperpolarizing current is passed in the right cell, the left cell also becomes
hyperpolarized, but more slowly and to a lesser degree (display as in B ) . (D)
When organ discharge is evoked by irritating the skin, a depolarization gradually
rises up to the threshold of the giant cell and initiates a burst of three spikes (lower
traces, base line indicated by superimposed sweeps ). Each spike produces a response
in the organ (upper trace, recorded at high gain and greatly reduced in amplitude
because curare is used to prevent movement). Modified from Bennett et al. ( 1967b).
not changed in duration or desynchronized; they are reset in phase by the
same amount as the spikes in the pacemaker neuron and follow the spikes
at the normal interval. Thus, hyperpolarization in one cell affects the
entire pacemaker nucleus and resets its phase of firing. Moreover, since
hyperpolarization spreads between cells, the interaction between cells
can be inferred to be electrotonic. (This has been directly demonstrated
also.) The coupling between pacemaker cells obviously leads to synchronization
of their firing. Each cell is spontaneously active and would
get out of step with the others if there were not coupling-or positive
feedback-between them. The accuracy of synchronization is so great
as to require the speed of electrotonic coupling; synchronization by
mutually excitatory chemically transmitting synapses would involve too
great a delay.
The coupling of gymnotid pacemaker neurons is not as close as in the
catfish electromotor neurons and does not allow impulses to propagate470 M. V. L. BENNETT
2 ‘ U L
10 msec
Fig. 75. Responses of pacemaker and
2 msec
relay cells in a weakly electric gymnotid
Gymnotus. Upper traces: activity in the spinal cord and peripheral nerves leading
to the electric organ (recorded by needle electrodes at high gain in a curarized
animal). Lower traces: intracellular recordings in pacemaker ( A,C) and relay
(B,D) neurons. Faster sweep in C and D where the dotted lines indicate the
times of firing of the cells in relation to the descending activity. From Bennett
et al. (1967~). – c 4 L 4. 4
A l 1 – D 4
I 1 I!.
B l I
Fig. 76. Effect of polarization in a single pacemaker cell of a weakly electric
gymnotid Gymnotus. Recording as in Fig. 75 except that current applied through
the recording electrode is indicated on the lower trace. Two superimposed sweeps
in each record, one with and one without applied current. The sweeps are triggered
by the spike of the pacemaker cell. Faster sweep in A and B. (A,C) A depolarizing
pulse that evokes a spike advances the next and subsequent spikes but does not
desynchronize or itself cause any descending activity. (B,D) A hyperpolarizing
pulse retards the next and subsequent spikes but does not desynchronize the
descending activity. From Bennett et al. (1967~).10. ELECTRIC ORGANS 471
from one cell to another. As shown in Fig. 76A,B an impulse in one
pacemaker cell does not cause a descending volley in the spinal cord;
it only advances the phase of firing. Thus, there is some spread of depolarization
to the other cells but not a propagated impulse (unless the
cells are very near to firing). The coupling is sufficiently close to synchronize
the cells, and, when all the cells are firing in the vicinity of a
cell penetrated by a microelectrode, it is extremely difficult to block
that cell’s activity by hyperpolarizing current. While a spike in any one
cell affects its neighbors but little, many cells firing together can produce
a large and very suprathreshold depolarization in an inactive cell.
As it happens, the medullary relay neurons are also electrotonically
coupled in most gymnotids that have been studied. Coupling is by way
of the pacemaker fibers afferent to them. At this level the coupling does
not serve to keep the cells from firing out of phase, for each cell is
always excited once per organ discharge by the large descending volley
from the pacemaker nucleus. Presumably the coupling serves to synchronize
relay cell firing, either because asynchrony was present in the
initial pacemaker volley or because it has arisen in transmission from the
pacemaker nucleus. The spinal relay neurons have been adequately
studied only in the electric eel (Bennett et al., 1964; Bennett, 1968a).
These cells, too, are electrotonically coupled and by way of the descending
fibers ending on them (Pappas and Bennett, 1 W ) . Preliminary
experiments indicate that the spinal relay neurons are similarly organized
in Gymnotus. Again, coupling would tend to increase synchronizatio?.
Sherrington (1906) called the motoneuron the final common path for
muscle fiber activity. In the gymnotids the final common path for
electric organ discharge extends three neurons back into the nervous
system. Each level involves a number of neurons and a certain amount
of signal shaping may go on in the relays, but these are only small
extensions of the Sherrington concept.
3. Asmoscomrs
In the stargazer the command signal is initiated in what is probably
a midline nucleus in the medulla and is relayed in the very large oculomotor
nucleus (Bennett, 1968a, and unpublished data). Neurons are
electrotonically coupled in both command and relay nuclei (Fig. 70).
In the relay both dendrodendritic synapses and the presynaptic fibers
mediate coupling. The organ discharges are somewhat variable in size
and apparently do not always involve every oculomotor neuron (Fig. 6).
The experimental evidence indicates that the command nucleus is not472 M. V. L. BENNETT
well synchronized. The oculomotor neurons do not simply relay the
signal; they also “decide” whether or not a sufficiently large command
volley is present to call for a discharge, and this decision is not always
unanimous in spite of the coupling between cells.
Often firing of the command neurons is in pairs of spikes, and PSPs
in the oculomotor neurons show two components. Apparently, a few
oculomotor neurons fire twice in normal activity and give rise to the
delayed component on the falling phase of some discharges (Fig. 6).
4. RAJrnS
In the skate the electric organ is probably controlled by a midline
nucleus in the medulla since stimuli applied in this region evoke organ
discharge (Szabo, 1955, 1961b). There is no evidence as to whether this
nucleus is a relay or pacemaker, or whether the neurons are electrotonically
coupled. This is a particularly interesting example because the
discharge is repetitive and probably asynchronous. While it seems likely
that positive feedback between neurons is involved in the decision to
discharge the organ, the speed of electrical transmission is apparently
not required. If feedback is still mediated electrically, it would suggest
that electrical coupling may be involved in other slow systems as well.
5. MORMYRIDS
In the mormyrids control of organ discharge is more complex than in
the fish discussed up to this point (Bennett et al., 1967a; Bennett, 1968a).
There is some evidence that the pacemaker nucleus is bilateral and that
and organ discharge can be initiated by a command signal arising on
either side (Fig. 73). This activity is relayed through a midline nucleus
in the medulla to the spinal neurons and then to the electric organ.
Although each pacemaker nucleus is able to initiate a command volley
independently, either because it is spontaneously active or because it
receives tonic excitatory inputs, pacemaker activity on the two sides is
coordinated. An impulse arising in what may be considered the dominant
nucleus at that time propagates from the medullary relay “antidromically”
into the other pacemaker nucleus, the subordinate one. (Transmission
at the pacemaker synapses is apparently electrotonic and allows
“antidromic” propagation. ) Thus both nuclei are excited, and the pacemaker
potential of both nuclei is returned to the level of hyperpolarization
that immediately follows spike activity; pacemaking in both nuclei
is reset. The cells of the two nuclei then begin to depolarize again, but
those in the dominant nucleus reach the firing level first, again resetting
the activity of both nuclei.10. ELECTRIC ORGANS 473
In the mormyrids there is not simply a single impulse at each level
of the electric organ control system (Fig. 73). PSPs from what is presumably
a single spike in either pacemaker nucleus initiate a two-spike
discharge in the medullary relay (the second spike propagates back to
the dominant pacemaker and both spikes propagate to the subordinate
pacemaker). Firing in “doublets” appears to be a property intrinsic to
the cell membrane rather than a result of a long-lasting PSP because the
cells give a two-spike discharge in response to a brief intracellularly
applied current pulse. The two medullary relay volleys descend the
spinal cord and cause two PSPs in spinal neurons that innervate the
electrocytes. These PSPs are chemically mediated. The first PSP initiates
a three-spike discharge. Firing in triplets appears intrinsic to the cell
membrane of the spinal cells as does the doublet firing of the medullary
cells. However, the third spike is somewhat labile and the second PSP
from the medullary relay guarantees its occurrence. The three electromotor
neuron spikes are propagated out to the synapses on the stalks
of the electrocytes, but only a single postsynaptic spike is produced. The
first PSP is very small, the second is greatly facilitated (increased in
amplitude) but still subthreshold, and the third is facilitated sufficiently
to reach threshold. The PSPs are not recorded external to the fish because
the synapses are on stalks far from the body of the cells and there
is little or no longitudinal current associated with them. (A similar
firing pattern could be detected externally if the cells were diffusely
innervated over one face.) The significance of this peculiar sequence
of signal transformations is unknown. It also occurs in Mormyrus and
Mormyrops in some species of which the discharge is longer lasting
and the requirement for synchronization is not so great.
Both medullary and spinal neurons are closely coupled electrotonically,
and an impulse in one cell propagates to all the other cells of that
nucleus. (An antidromic impulse in an axon generally fails to invade
the cell body, at least in part because of the low input resistance resulting
from the close coupling of the cells.) The neurons are connected
by thick dendrodendritic bridges in both nuclei (Fig. 71), and coupling
by way of presynaptic fibers probably occurs in the medullary relay
as well.
Interpretation of the peculiar multiple firing in the mormyrid control
system is not likely to be aided by comparison to Gymnurchus. Light
microscopy suggests that there are four interconnected nuclei in the
medulla, but they have not been studied physiologically ( Szabo, 1961b ) .
All that is known is that a single volley descends the spinal cord, and
a single PSP in the electrocytes initiates each organ discharge, An interesting
aspect of the mormyrid control system is that the command474 M. V. L. BENNETT
system “informs” the sensory system that an organ discharge is coming
(Bennett and Steinbach, 1969; see Chapter 11, this volume). The pathways
mediating this activity and the effects on afferent volleys are just
being explored.
6. TORPEDINIDS
In Torpedo there is a command nucleus on each side that is located
deep in the medulla (Szabo, 1954; Albe-Fessard and Buser, 19%).
These nuclei activate the very large electromotor nuclei that lie on the
dorsal surface of the medulla (Fig. 8). The electric organ discharges
are synchronous on the two sides, and it is probable that either command
nucleus can activate all the relay cells on both sides. NO data are
available about interaction between the two command nuclei, but some
coordinating mechanism is likely to be found. Preliminary data indicate
that the electromotor cells, which comprise a relay, are not coupled
to each other and that transmission from the command nucleus is
chemically mediated (cf. Saito, 1966). Single volleys from the command
nucleus excite single volleys in the relay cells that produce the single
PSPs in the electrocytes comprising the individual organ discharges.
7. THEE LECTRIECE L
The mechanism of control in the eel shows an interesting variation.
All the electrocytes fire together during the large discharges, but only
Sachs’ organ and posterior part of Hunter’s organ are active during
the weak discharges. As in weakly electric gymnotids, the organs are
controlled by a single midline relay nucleus in the medulla (Bennett
et al., 1964; Bennett, 1968a). The pacemaker nucleus has not yet been
found in the eel. Each command volley from the medullary relay excites
all the spinal neurons that innervate the electrocytes, and the command
volley passes out ventral roots to anterior as well as posterior parts of
the organ. At low frequencies PSPs in the main organ and the anterior
part of Hunter’s organ are small and do not excite the cells, but they
are large enough to excite cells in Sachs’ organ and in the posterior
part of Hunter’s organ (Albe-Fessard and Chagas, 1954). When the
interval between command volleys becomes as small as several milliseconds,
the PSPs in the main organ and anterior part of Hunter’s
organ greatly increase in amplitude and become adequate to excite the
cells of these organs. Thus the command to excite the main organ consists
of high frequency activity in the same neurons that control Sachs’
organ. Only a single bulbospinal relay system is required to control
both organs. This increase in simplicity is obtained at the cost of several10. ELECTRIC ORGANS 475
milliseconds’ increase in the minimum latency at which the main organ
can be activated. Probably the control system has evolved its dual
functioning from an earlier stage in which there was only a single
weakly electric organ. An intriguing problem is how pacemaker activity
for the two organs is controlled.
B. Synchronization of Electrocyte Activity
Given that the command nucleus has reached a decision to fire the
electric organ, that is, has itself fired, the individual electrocytes must
be excited synchronously. This is a significant problem because the
different parts of the electric organ may be quite far apart. Nerve
tracts running directly to the different parts of the organ could not
fire the cells synchronously unless they were much more rapidly conducting
than any known nerves.
Two basic mechanisms are known to contribute to synchronization;
both involve utilization of conduction time in nerve fibers or electrocyte
stalks to equalize overall latency of the command signal in reaching
the electrocytes. In one mechanism the nerve fibers run more or less
directly to the more distant part of the organ, but take a more devious
path to the nearer parts, thus tending to equalize path length and
thereby conduction time (Fig. 77A). In the second mechanism shorter
paths to nearer parts of the electric organ are of lower conduction
velocity, again tending to equalize conduction time (Fig. 77B; fibers
conducting more slowly are indicated as being of smaller diameter).
Lower conduction velocity may also be found in only a portion of the
path to the electrocytes; slowly conducting collaterals may branch off
from the main rapidly conducting path to innervate the nearer parts
of the organ (Fig. 77C).
Equalization of path length is an obvious feature of a number of
electric organs where the different nerves enter the electric organ and
run for some distance before giving off branches that return to end
near the point of entry. Compensatory differences in conduction velocity
apparently occur in the stalk system of mormyrid electrocytes where
differences in conduction distance are easily visualized and where
synchronization between parts of the body of the cell is very precise
(Fig. 61). As would be expected, the shorter paths involve stalks that
are smaller in diameter. Probably most systems use a combination of
the two mechanisms.
In the electric eel differences in conduction time down the spinal
cord are compensated for by both increased delay at the spinal relays476 M. V. L. BENNETT
B-
….
Fig. 77. Mechanisms of compensatory delay. Neural pathways are diagrammed
leading to terminations in the periphery at different distances from a rostral
command center. A command volley arises at the large arrow and the dotted lines
represent times of arrival of impulses at equal time intervals afterward. ( A )
Equalization of path length. The paths to the nearer cells are made more devious
so that all paths are of nearly equal length. ( B ) Compensatory differences in
conduction velocity. The paths leading to the periphery are direct, but conduction
is slower in the shorter, thinner paths. ( C ) Localized compensatory delays. Thin
terminal branches in which conduction velocity is reduced are longer in the paths
leading to the nearer parts of the periphery. From Bennett (1968a).
and increased delay from activity in ventral roots to spike initiation in
the electrocytes ( Albe-Fessard and Martins-Ferreira, 1953). TWO lines
of evidence indicate that the delay at the spinal relay arises in
conduction time in collateral branches of descending axons. First,
transmission from descending fibers to electromotor neurons is electrotonic
and the PSPs are sufficiently rapidly rising that they can be delayed
very little at the synapses themselves (Bennett, 1966). Second,
action potentials at intermediate delays can be recorded in collaterals
of the descending fibers. Preliminary morphological observations indicate
that these collaterals are thinner at the anterior region of the spinal
relay nucleus (R. M. Meszler and M. V. L. Bennett, unpublished data).
The efferent axons entering the ventral roots are also thinner in the
anterior regions of the spinal cord, and reduced conduction velocity
probably contributes to the greater peripheral delay at the anterior of
the organ.
The early firing of rostral accessory organs in a number of gymnotids
and of the dorsal column of electrocytes in Gymnotus presumably involves
similar mechanisms to those in the eel. There is no indication
of an earlier firing pacemaker or medullary relay, or of an earlier firing
group of spinal neurons. Apparently the relatively delayed firing of the
majority of electrocytes is achieved peripherally. It may be that early10. ELECTRIC ORGANS 477
firing of the accessory organs really represents delayed firing of the main
organ.
At chemically transmitting synapses such as those on electrocytes,
compensatory delays could in principle arise in “synaptic delay” [which
at the neuromuscular junction at least results from time required for
release of transmitter ( Katz and Miledi, 1965a,b) 1. Compensatory delays
may occur at a number of chemically transmitting synapses in electromotor
systems. One instance is in the spinal electromotor nucleus of
the mormyrid Gnathonernus in which PSPs are synchronous at the two
ends of the nucleus, although the descending volley arrives at the
anterior end about 0.3 msec earlier than at the posterior end (M. V. L.
Bennett and E. Aljure, unpublished data).
A further instance is provided by the electric organ of the electric
catfish. A single large axon branches to innervate all the electrocytes
on one side of the organ. There is some equalization of path length
for the axon enters the organ somewhat posteriorly and runs both
anteriorly and posteriorly from the site of entry. Stimulation of the
axon at its point of entry leads to a discharge that is about the same
duration as the responses of single cells (Fig. 78C). Stimulation of a
small branch to a piece of organ from the head end of the fish produces
a response that has a longer latency than stimulation of a similar
preparation from the tail of the fish (Fig. 78B,B’). This difference in
latency is just sufficient to compensate for the difference in conduction
2 msec 2 msec
Fig. 78. Compensatory delays in the catfish electric organ. (A,A’) Responses
of a piece of electric organ dissected from near the tail of a fish about 20 em long.
The tissue has 10 em of motor nerve attached to it and is stimulated at each end
of this length of nerve. ( B ) Response of a piece of organ from same fish dissected
near head, stimulated via the nerve close to point of entry into tissue. (B’)
Response of another piece of organ, similarly stimulated, but taken from a position
10 cni nearer tail. ( C ) Response of the electric organ on one side from another
large fish, stimulated orthodromically at the central end of the nerve. ( C ) The
response of the same organ, stimulated “antidroniically” at the caudal end of the
nerve. From Keynes et al. ( 1961).478 M. V. L. BENNETT
time down the main trunk of the axon (A,A’). The significance of the
synaptic delay can be well illustrated by recording the output of one
entire side of the organ when stimulating the nerve at different sites.
If the nerve is stimulated at its caudal end rather than its site of entry
into the organ, the differences in delay add to the differences in conduction
time and a markedly asynchronous discharge results ( Fig. 78C’).
The site of these compensatory delays is unknown; it could be in
presynaptic nerve branches and terminals or at the synapses themselves.
In the electric catfish it could also be in the stalks of electrocytes. However,
no systematic differences in length of stalks have been observed
in teased preparations, and the variability of stalk length is quite great
in all parts of the organ. In both mormyrids and the electric catfish,
morphological investigations may reveal that terminals are longer and
thinner where delays are greater. In this case one would be less inclined
to consider synaptic delay as a factor in the compensatory mechanism.
C. Organization of Electromotor Systems
As discussed here the control of electric organs is reasonably well
described from one to several neural levels back into the nervous
system. Provided one can accept the proposition that the pacemaker
cells are autoactive, the analysis is reasonably complete for the very
regular basal discharge frequency found in many species. Certainly one
would like to know the ionic basis of the pacemaker activity, but in
respect to impulse traffic, the origins and pathways are well defined.
The pathways of afferents to the command nuclei in variable frequency
and intermittently discharging forms are by and large unknown, although
most if not all sensory modalities can be excitatory and responses
can be of quite short latency (e.g., Fig. 3B). The slight modifications
of frequency in some high and constant frequency species also require
investigation. The relative fixity of jamming avoidance responses suggests
fairly direct electrosensory inputs, while the brief modulations that
perhaps have a signaling function probably involve much more complex
pathways (Bullock, 1970). The involvement of higher centers is also
indicated by the fact that changes in organ discharge rate can be
operantly and respondently conditioned ( Mandriota et al., 1965, 1968)
as well as serve for signaling ( Black-Cleworth, 1970; Moller, 1971).
An interesting problem is raised by the ability of Gymnorhamphichthys
and some Hypopomus to maintain organ discharge frequency constant
at low, high, and even intermediate levels. This ability suggests that
there are neurons that maintain a tonic, asynchronous but quite constant10. ELECTRIC ORGANS 479
excitatory input to the pacemaker nucleus. This input seems too constant
to be due to casual stimulation of receptors resulting from movement,
and an activating system that is itself quite stable would seem to be
required to explain the data.
Pathways mediating the complete inhibition of discharge in Gymnotus,
Hypopomus, and Sternopygus are also not worked out. The strong
excitatory drive to the medullary relay in Gymnotus and probably also
Hypopomus suggests that this nucleus is not involved. Probably the
inhibition operates on the pacemaker nucleus. If so, it might be considered
surprising that in Gymnotus the discharge always starts and
stops abruptly with only a small degree of slowing below the basal
rate before and after complete cessation. However, similar observations
have been made during microelectrode recording from the pacemaker
nucleus in which activity at this level did indeed cease when discharge
ceased and restarted when renewed discharge was evoked by spinal
stimulation. These sudden changes in frequency are consistent with
calculations from the Hodgkin-Huxley equations, which predict that
maintained firing of the squid axon in response to a steady current
can occur only at rather high frequencies ( Stein, 1967).
Unlike Gymnotus, Sternopygus does not start up again by emitting
full-sized organ pulses; there is a gradual recovery of pulse amplitude
and synchronization. While this observation might represent a difference
in the pacemaker nucleus, it might also result from properties of lower
level synapses. If the spinal cord is sectioned in Gymnotus a single brief
stimulus to the cord will still cause a full-sized organ discharge. However,
in Sternopygus moderately prolonged stimulation at about the
normal organ frequency is required to restore the discharges to their
full amplitude. Granted that transmission at the spinal relays is electrotonic,
the most likely site of facilitation in Sternopygus is at the
nerve-electrocyte synapses. The absence of cessation of discharge in
the higher frequency species Eigenmannia and sternarchids is perhaps
a result of the fact that discharge does not stop following spinal section
but continues asynchronously.
The investigations to date have been largely restricted to elucidation
of what may be considcred the final common path for organ discharge.
A number of modifications to the original concept of final common
path appear necessary to describe control of electric organs. A group
of neurons at a particular level can fire as a single neuron, and the
common path may involve several levels of neurons. The highest level
determines the frequency of discharge; the lower levels act only as
relays and perhaps also do some signal shaping (or as in the mormyrids,
transformation of impulse number). The command path may bifurcate480 M. V. L. BENNETT
(and then cease to be common). The command to a particular organ
may be expressed in terms of frequency.
Several of these modifications are already required for ordinary
motor systems. For example, some muscle fibers are innervated by
more than one nerve fiber, which is analogous to paired command
nuclei. If tension of a muscle fiber rather than any contraction at all
is considered, the command may be coded in terms of frequency of
firing. Furthermore, it is obvious that we do not understand all about
the control of an electric organ or a muscle fiber when we know the
final common path to it. We also need to know all the fibers afferent
to neurons in that path. We have taken the decision to fire the organ as
meaning the firing of the highest level nucleus of the final common
path. The decision could also be defined as activity in any subset of
fibers afferent to the highest nucleus that can cause the nucleus to fire.
Similarly, the relevant output of a whole muscle, its tension, may be
achieved (or coded for) by many different combinations of fibers active
at different frequencies.
For freshwater weakly electric fish and for the electric catfish the
neural circuitry guarantees that every electrocyte receives neural excitation
in every discharge. The operation of the system is essentially
all-or-none and an organ discharge is present or absent. It should be
recognized that amplitude can vary minorly as a result of refractoriness
(and perhaps facilitation), but these changes must be secondary to
changes in frequency. The situation is similar in the eel except that
the amplitude normally varies over a wide range. In all these fish and
to a good approximation in Torpedo and the stargazers as well, the
behavior of the animal with respect to its discharge can be characterized
by a single series of time intervals, a simplicity which gives the system
some attractiveness for quantitative study.
Two additional characteristics are frequently found in electric organ
control systems. Where a group of neurons fire synchronously there
is likely to be positive feedback between the cells. In command nuclei
in which a highly synchronous volley arises, positive feedback must
be present and must be electrically mediated because chemically
mediated transmission is too slow. In relay nuclei there is no absolute
requirement for feedback, but cells are often coupled electrotonically,
presumably to increase synchronization. Another characteristic often
found in electric organ control systems is a tendency toward reduced
numbers of cells at higher levels. Apparently the decision to discharge
the electric organ is always reached in a group of cells that
are few in number compared to the final generating cells. The most
striking example is found in the electric catfish, in which two neurons10. ELECTRIC ORGANS 481
control the several million electrocytes. Similar but less extreme pyramids
of numbers are found in other electric fish. This aspect of neural organization
may be denoted the committee principle. This term is chosen
in analogy with decision making by committees which tends to be more
rapid the smaller the committee.
One might question whether there could be a command ‘‘nucleus’’
containing a single cell, a committee of one. An example is provided
by the Mauthner cells of lower vertebrates which can be considered
single cell command systems for the axial musculature on either side
of the body (Furukawa and Furshpan, 1963; Diamond and Yasargil,
1969).
One could propose a number of reasons for the existence of control
systems containing more than one cell, such as protection against loss
of neurons and production of enzymes for maintenance of synaptic
transmission. At lower levels progressive increase in a number of neurons
may be equivalent to progressive increase in size. The two giant neurons
of the electric catfish are able to support a very small synaptic area
on a large number of electrocytes, but it seems likely that a large
number of neurons are required to provide the vast synaptic area in the
electric organs of Torpedo. Phylogenetic or ontogenetic factors may
also be responsible for the presence of the relays. Evolution may yet
progress to single-celled command “nuclei” although in most instances
the multicelled nuclei do about as well as required by the electric
organ.
It is interesting to compare the function of relay nuclei to transmission
of impulses along an axon. At a node of Ranvier an impulse
that has been attenuated in electrotonic propagation from the preceding
node triggers the generation of a new impulse that is of full amplitude.
The node acts as a pulse restoring element, and decrementless conduction
is thereby made possible. A relay nucleus potentially does more
than this. The volley of impulses in the fibers efferent from the relay
may be more synchronous than that in the afferent fibers, or all the
efferents may become excited when only a fraction of the afferents are
active (as appears to occur occasionally in Astroscopus) . These functions
do not require coupling of the cells; all that is necessary is that a
number of afferent fibers converge on each relay cell.
Although midline nuclei have developed in a number of control
systems, bilateral command nuclei may be fairly common because of
the basically bilateral organization of the nervous system. If either of
the two command nuclei can excite an entire effector system, synchronization
between the two is not important. However, it would appear
functional for each command to reset the phase of firing in both com482 M. V. L. BENhTETT
mand nuclei. Probably this resetting occurs in the mormyrids by actual
invasion of the impulses, but mutual inhibition could also be a mechanism
in other systems. For example, crossed inhibition like that between
Mauthner cells (Furukawa and Furshpan, 1963; A. A. Auerbach, unpublished
data) would be equally effectivc in an electric organ system.
In the hatchetfish each Mauthner fiber activates the muscles depressing
both pectoral fins, and these cells thus constitute a bilateral command
system for the depressor muscles ( Auerbach and Bennett, 1969a,b).
Contraction of these muscles causes the animals to dart upward in an
escape reflex. Mutual excitation like that proposed for the mormyrid
pacemaker nuclei would not work for the Mauthner cells, because each
innervates the axial musculature on one side and near simultaneous
excitation could lead to a counter-productive attempt to flip the tail to
both sides simultaneously. Actually there is evidence that in the goldfish
each Mauthner fiber excites motoneurons on one side and inhibits them
on the other side but at a shorter latency (Diamond and Yasargil, 1969).
Simultaneous activity of both Mauthner fibers excites no axial motoneurons.
If one Mauthner fiber fires, it causes muscle contraction on
one side and prevents the other Mauthner fiber from exciting the contralateral
niotoneurons for about 100 msec.
Recognition of the requirement for speed of transmission in synchronized
systems has been useful in predicting sites where electrically
mediated transmission has subsequently been found including neurons
controlling sonic muscles ( Bennett, 1966; Pappas and Bcnnett, 1966)
and oculoinotor neurons (Kriebel et al., 1969). In mediating activity
that is not very precisely synchronized, the speed of electrotonic
synapses might not be required. Examples are the bursting behavior
of respiratory neurons and various invertebrate cardiac ganglia. Electrotonic
coupling does occur in a number of moderately synchronized
systems in both vertebrates (cf. Bennett, 1968b) and invertebrates
(Hagiwara et al., 1959, Willows and Hoyle, 1969). The extent to which
synchronization of slow systems is mediated by positive feedback and
the degree to which the positive feedback is mediated electrically remain
to be worked out. Electrotonical transmission appears to provide a
simpler and perhaps more efficient means of positive feedback between
cells (Bennett, 1968b).
One may question whether firing of an electric organ resembles
other decision processes. The final common path concept of a decision
requires extension even in some very simple systems. On the other
hand, it is reasonable to think that the features observed in simple
effector systems havc some more general relevance. Negative feedback
has been emphasized as a common property in neural control systems10. ELECTRIC ORGANS 483
mediating homeostasis. It is also common in sensory and motor pathways
where it may increase spatial resolution. Yet many phenomena involve
recruitment of large numbers of cells in ways that appear to require
mutual reinforcement of neuronal activities, that is, positive feedback.
IV. CONCLUSIONS AND PROSPECTS
The study of electric organs and electrocytes has illuminated many
aspects of membrane physiology. Their primary usefulness is a result
of evolution which has exaggerated different membrane functions in
different electrocytes. For example, the concept of independent sites
that pass different ions or groups of ions becomes more reasonable
given that specific kinds of permeability can occur in isolation in different
regions of a single cell or in different cells. Examples are the isolated
occurrence of membrane generating postsynaptic potentials (in Astroscopus
and torpedinids ) , of membrane exhibiting delayed rectification
(in rajids), and of the electrically excitable sodium system without
delayed rectification (in the electric eel), This macroscopic separation
suggests microscopic separation, which in many single membranes can
only be inferred from functional arguments or pharmacological data.
These specializations of different kinds of membrane cannot really
be said to have been responsible for major advances in electrophysiological
knowledge. Yet as new morphological, biochemical, and biophysical
techniques become available, electrocytes may provide the best
tissues for their evaluation.
Although one cannot determine electrophysiologically whether or
not sodium and potassium channels of ordinary spike-generating membrane
are separate, one can begin to think about isolating them biochemically.
Electric organs would appear to provide a good starting
material just as they are proving useful in the characterization of
cholinesterase and sodium-potassium ATPase. By selection of cell type,
one can select different kinds of membrane with relatively great degrees
of purity compared to most other tissues.
The neural systems controlling electric organs have provided a large
number of examples of electrically mediated transmission, which meets
the functional requirement for rapid communication between cells. This
mode of transmission also proves to be able to mediate many functions
often considered as restricted to chemically mediated transmission. The
correlation between morphologically close apposition and electrotonic
coupling was considerably strengthened by the work on electromotor484 M. V. L. BENNETT
systems. This correlation helps to validate morphological identification
of electrical transmission in other systems where electrophysiological
analysis is not so simple.
It is not known whether there is any relevance to higher systems
of the organizational principles deduced from electric organ systems.
The next level of analysis of the electric organ systems may be no
easier than study of less specialized systems that are of more general
interest. Some knowledge is being obtained of afferent pathways from
electroreceptors in weakly electric fish (see Chapter 11, this volume)
which have important inputs to the electric organ control system. Both
operant and respondent conditioning of the control system can be
obtained and conditioned response latency can be very short. It is not
unreasonable that the complete neural pathway of the conditioned
response could be obtained in these cases. The central connections are
minimally explored; one knows what goes in and one can go from the
electric organ several synapses antidromically. The rewards for filling
in the gap could be great, and prospects for at least some progress
are bright.
ACKNOWLEDGMENTS
Some of the hitherto unpublished work was carried out in the laboratory of
Neurophysiology, College of Physicians and Surgeons, Columbia University, with the
support of Dr. H. Grundfest. Work on the Rio Negro expedition of the R. V. Alpha-
Helix was in collaboration with A. B. Steinbach and allowed considerable advance
in understanding of the sternarchids. The author is greatly indebted to Dr. T. H.
Bullock for his assistance in that remote place. Supported in parts by grants from the
National Institutes of Health (5PO1 NB 07512 and HD-04248) and the National
Science Foundation ( GB-6880).
REFERENCES
Adrian, R. H., Chandler, W. K., and Hodgkin, A. L. (1970). Slow changes in potassium
permeability in skeletal muscle. 1. Physiol. (London) 208, 645-668.
Albe-Fessard, D. (1950a). Les caractkres de la dhcharge des poissons 6lectriques.
Arch. Sci. Physiol. 4, 299-334.
Albe-Fessard, D. ( 1950b). Proprikks hlectriques passives du tissue klectrogkne des
poissons Alectriques. Arch. Sci. Physiol. 4, 413434.
Albe-Fessard, D. ( 1951 ). Modification de l’activith des organes Alectriques par
des courants de l’origine exterieure. Arch. Sci. Physiol. 5, 43-73.
Albe-Fessard, D., and Buser, P. ( 1954). Analyse microphysiologique de la transmission
rkflexe au niveau du lobe klectrique de la torpille (Torpedo nmmorata).
1. Physiol. (Paris) 46, 923-946.10. ELECTRIC ORGANS 485
Albe-Fessard, D., and Chagas, C. (1954). Etude de la sommation a la jonction nerf-
Blectroplaque chez le gymnote (Electrophorus electricus). J. Physiol. (Paris)
Albe-Fessard, D., and Martins-Ferreira, H. (1953). R81e de la commande nerveuse
dans la synchronisation du functionnement des Blkments de I’organe Blectrique
du gymnote, Electrophorus electricus L. J. Physiol. (Paris) 45, 533-546.
Albers, R. W. (1967). Biochemical aspects of active transport. Ann. Rev. Biochem.
36, 727-756.
Altamirano, M. (1955). Electrical properties of the innervated membrane of the
electroplax of electric eel. J. Cell. Comp. Physiol. 46, 249-278.
Altamirano, M., Coates, C. W., Grundfest, H., and Nachmansohn, D. (1953). Mechanisms
of bioelectric activity in electric tissue. I. The response to indirect and
direct stimulation of electroplax of Electrophorus electricus. J. Gen. Physiol. 37,
91-110.
Altamirano, M., Coates, C. W., and Grundfest, H. (1955). Mechanisms of direct
and neural excitability in electroplaques of electric eel. J. Gen. Physiol. 38, 319-
360.
Auerbach, A. A., and Bennett, M. V. L. ( 1969a). Chemically mediated transmission
at a giant fiber synapse in the central nervous system of a vertebrate. J. Gen.
Physiol. 53, 183-210.
Auerbach, A. A., and Bennett, M. V. L. (196913). A rectifying synapse in the central
nervous system of a vertebrate. J. Gen. Physiol. 53, 211-237.
Bauer, R. ( 1968). Untersuchungen zur Entladungstatigkeit und zum Beutefangverhalten
des Zitterwelses Malapterurus electricus Gmelin 1789 ( Siluroidea,
Malapteruridae, Lacep, 1803). 2. Vergleich. Physiol. 59, 371402.
Belbenoit, P. ( 1970). Comportement alimentaire et dBcharge associBe chez Torpedo
marmorata ( Selacchii, Pisces). 2. Vergleich. Physiol. 67, 205-216.
Bennett, M. V. L. (1961). Modes of operation of electric organs. A m . N. Y. Acad.
Sci. 94, 458-509.
Bennett, M. V. L. (1964). Nervous function at the cellular level. Anti. Rev. Physiol.
26, 289-340.
Bennett, M. V. L. (1966). Physiology of electrotonic junctions. Ann. N. Y. Acad. Sci.
137, 509-539.
Bennett, M. V. L. (1968a). Neural control of electric organs. In “The Central
Nervous System and Fish Behavior” (D. Ingle, ed.), pp. 147-169. Univ. of
Chicago Press, Chicago, Illinois.
Bennett, M. V. L. ( 1968b). Similarities between chemically and electrically mediated
transmission. In “Physiological and Biophysical Aspects of Nervous Integration”
(F. D. Carlson, ed.), pp. 73-128. Prentice-Hall, Englewood Cliffs, New Jersey.
Bennett, M. V. L. (1970). Comparative physiology: Electric organs. Ann. Rev.
Physiol. 32, 471528,
Bennett, M. V. L., and Grundfest, H. (1959). Electrophysiology of electric organ in
Gymnotus carapo. J. Gen. Physiol. 42, 1067-1104.
Bennett, M. V. L., and Grundfest, H. (1961a). The electrophysiology of electric
organs of marine electric fishes. 11. The electroplaques of main and accessory
organs of Narcine brasiliensis. J. Gen. Physiol. 44, 805-818.
Bennett, M. V. L., and Grundfest, H. (1961b). The electrophysiology of electric
organs of marine electric fishes. 111. The electroplaques of the stargazer, Astroscopus
y-graecum. J. Gen. Physiol. 44, 819-843.
46, 823-840.486 M. V. L. BENNETT
Bennett, M. V. L., and Grundfest, H. ( 1 9 6 1 ~ )S. tudies on morphology and electrophysiology
of electric organs. 111. Electrophysiology of electric organs in
Mormyrids. In “Bioelectrogenesis” ( C. Chagas and A. Paes de Carvalho, eds. ),
pp. 113-135. Elsevier, Amsterdam, 1961.
Bennett, M. V. L., and Grundfest, H. (1966). Analysis of depolarizing and hyperpolarizing
inactivation responses in Gymnotid electroplaques. J. Gen. Physiol.
Bennett, M. V. L., and Steinbach, A. B. (1969). Influence of electric organ control
system on electrosensory afferent pathways in mormyrids. In “Neurobiology of
Cerebellar Evolution and Development” ( R . LlinAs, ed.), pp. 207-214. Am.
Med. Assoc., Chicago, Illinois.
Bennett, M. V. L., and Trinkaus, J. P. (1970). Electrical coupling between embryonic
cells by way of extracellular space and specialized junctions. J. Cell Biol.
44, 592-610.
Bennett, M. V. L., Wiirzel, M., and Grundfest, H. (1961). The electrophysiology of
electric organs of marine electric fishes. I. Properties of electroplaques of
Torpedo nobiliana. J. Gen. Physiol. 44, 757-804.
Bennett, M. V. L., Gimbnez, M., Nakajima, Y., and Pappas, G. D. (1964). Spinal
and medullary nuclei controlling electric organ in the eel, Electrophorus. Biol.
Bull. 127, 362.
Bennett, M. V. L., Pappas, G. D., Aljure, E., and Nakajima, Y. (1967a). Physiology
and ultrastructure of electrotonic junctions. 11. Spinal and medullary electromotor
nuclei in Mormyrid fish. J. Neurophysiol. 30, 180-208.
Bennett, M. V. L., Nakijima, Y., and Pappas, G. D. (196713). Physiology and ultrastructure
of electronic-junctions. 111. Giant electromotor neurons of Malapterurus
electricus. 3. Neurophysiol. 30, 209-235.
Bennett, M. V. L., Pappas, G. D., Gimknez, M., and Nakajima, Y. ( 1 9 6 7 ~ )P.h ysiology
and ultrastructure of electrotonic junctions. IV. Medullary electromotor nuclei
in gymnotid fish. J. Neurophysiol. 30, 236-300.
Bigelow, H. B., and Schroeder, W. C. (1953). “Fishes of the Western North
Atlantic, Part Two. Sawfishes, Guitarfishes, Skates and Rays.” Sears Found.
Marine Res., Yale University, New Haven, Connecticut.
Black-Cleworth, P. ( 1970). The role of electrical discharges in the non-reproductive
social behavior of Gymnotus carapo ( Gymnotidae, Pisces ) . Animal Behavior
Monographs 3, 3-77.
Bloom, F. E., and Barnett, R. (1966). Fine structural localization of acetylcholineesterase
in electroplaque of the electric eel. J. Cell Biol. 29, 475-495.
Brightman, M. W., and Reese, T. S. (1969). Junctions between intimately apposed
cell membranes in the vertebrate brain. J. Cell Biol. 40, 648-677.
Brock, L. G., and Eccles, R. M. (1958). The membrane potentials during rest and
activity of the ray electroplate. J. Physiol. ( L o n d o n ) 142, 251-274.
Bullock, T. H. (1970). Species differences in effect of electroreceptor input on
electric organ pacemakers and other aspects of behavior in electric fish. Brain,
Behav. Evol. 2, 85-118.
Cnss, A., Finkelstein, A., and Krespi, V. (1970). The ion permeability induced
in thin lipid membranes by the polyene antibiotics nystatin and amphotericin
B. J. Gen. Physiol. 56, 100-124.
Coates, C. W., Altamirano, M., and Grundfest, H. (1954). Activity in electrogenic
organs of knifefishes. Science 120, 845-846.
50, 141-169.10. ELECTRIC ORGANS 487
Cohen, L. B., Hille, B., and Keynes, R. D. (1969). Light scattering and birefringence
changes during activity in the electric organ of Electrophorus electricus. J.
Physiol. (London) 203, 489-509.
Cole, K. S. (1968). “Membranes, Ions and Impulses.” Univ. of California Press,
Berkeley, California.
Cole, K. S., and Curtis, H. J. (1939). Electric impedance of the squid giant axon
during activity. J. Gen. Physiol. 22, 649-670.
Couceiro, A., and Akerman, M. (1948). Sur quelque aspects du tissu Clectrique de
1’Electrophorus electricus ( Linnaeus ). Anais Acad. B r a d Cien. 20, 383-395.
Couteaux, R., and Szabo, T. ( 1959). Sikge de la lonction nerf-Clectroplaque dans les
organes Clectriques a klectroplaque pCdiculCes. Compt. Rend. 248, 457460.
Dahlgren, U. (1914). Origin of the electric tissues of Gymnarchus niloticus. Carnegie
Znst. Wash. Publ. 183, 161-194.
Dahlgren, U. ( 1927). The life history of the fish Astroscopus (the “Stargazer”). Sci.
Monthly 24, 348-365.
Dahlgren, U., and Silvester, C. F. (1906). The electric organ of the stargazer,
Astroscopus ( Brevoort ). Anat. Anz. 15, 387403.
del Castillo, J., and Katz, B. (1956). Biophysical aspects of neuromuscular transmission.
Progr. Biophys. Biophys. Chem. 6, 121-170.
De Oliveira Castro, G. (1955). Differentiated nervous fibers that constitute the electric
organ of Sternachus albifrons Linn. Anais Acad. Brasil. Cien. 27, 557-560.
Diamond, J,, and Yasargil, G. M. (1969). Synaptic function in the fish spinal cord:
Dendritic integration. Progr. Brain Res. 31, 201-209.
Dodge, F. A., and Frankenhaeuser, B. (1959). Sodium currents in the myelinated
nerve fibre of Xenopus laevis investigated with the voltage clamp technique.
J. Physiol. (London) 148, 188-200.
Eigenmann, C. H., and Allen, W. R. (1942). “Fishes of Western South America.”
Univ. of Kentucky, Lexington, Kentucky.
Eisenberg, F. A., and Gage, P. W. (1969). Ionic conductances of the surface and
tubular membranes of frog sartorius fibers. J. Gen. Physiol. 53, 279-297.
Ellis, M. M. (1913). The gymnotid eels of tropical America. Mem. Carnegie Museum
6, 109-204.
Ewart, J. C. (1892). The electric organ of the skate: Observations on the structure,
relations, progressive development, and growth of the electric organ of the skate.
Phil. Trans. Roy. SOC. London B183, 389420.
Farquhar, M. G., and Palade, G. E. (1963). Junctional complexes in various
epithelia. J. Cell Biol. 17, 375-412.
Frankenhaeuser, B., and Huxley, A. F. (1964). The action potential in the myelinated
nerve fibre of Xenopus laevis as computed on the basis of voltage clamp
data. J. Physiol. (London) 171, 302-315.
Fritsch, G. (1890). “Die elektrischen Fische. 11. Die Torpedineen.” Von Veit,
Leipzig.
Furukawa, T., and Furshpan, E. J. (1963). Two inhibitory mechanisms in the
Mauthner neurons of goldfish. J. Neurophysiol. 26, 140-176.
Gage, P. W., and Eisenberg, R. S. Capacitance of the surface and transverse tubular
membrane of frog sartorius muscle fibers. J. Gen. Physiol. 53, 265-278.
GCry, J., and Vu-Tin-Tu&. ( 1964). Gyrnnorhamphichthys hypostomus petiti, ssp.
nov. un curieux Poisson gymnotoide ar6necole. Vie Milieu, Suppl. 17, 485-498.
Greenwood, P. H. (1956). The fishes of Uganda-11. The Uganda Journal 20, No. 2,
129-143.