- Email: [email protected]
An efferent inhibition of auditory afferents mediated by the goldfish mauthner cell
Neuroscience Vol. 24, No. 3, pp. 829636, Printed in Great Britain
0306-4522/88
1988
$3.00 + 0.00
Pergamon Press plc Q 1988 lBR0
AN EFFERENT INHIBITION OF AUDITORY AFFERENTS MEDIATED BY THE GOLDFISH MAUTHNER CELL J.-W. LIN* and D. S. FABER Department of Physiology, State University of New York at Buffalo, Buffalo, NY 14214, U.S.A. A~act-Intra~ilular recordings from goldfish auditory afferents revealed a hy~~la~~tion triggered by a single impulse in the Mauthner cell. The firing of either one of the two Mauthner cells alone was sufficient to evoke this potential change. The all or none hyperpolarization, which could only be recorded in some auditory fibers, presumably was an inhibitory postsynaptic potential. The inhibitory postsynaptic potential typically had a latency of 6 ms, an amplitude of 1 mV and a half-decay time of 6.8 ms; it could block or delay impulses evoked by direct current injection and could attenuate the amplitude of excitatory postsynaptic potentials evoked by sound pulses. However, this inhibitory postsynaptic potential did not reduce the amplitudes of electrotonic coupling potentials produced by antidromic impulses in the Mauthner ceil. We propose that the inhibitory postsynaptic potential is generated at the dendrites of the auditory fibers, i.e. in the ear, rather than at the central terminals of the aBerent, where the antidromi~ coupling potentials originate. The possibility that the inhibitory postsynaptic potential actually represented disfacilitation due to an efferent inhibition of the hair cells, which tonically depolarize the saccular fibers, was ruled out because depolarization of these fibers increased the inhibitory postsynaptic potential amplitude. Possible morphological substrates for the efferent inhibition and the behavioral significance of this inhibition are discussed
The Mauthner (M-) cells in the goldfish medulla are a pair of giant interneurons that mediate a startle reflex. When one of the M-cells fires an action potential it causes a massive contraction of the trunk musculature contralateral to its soma and produces a rapid tail flip. 3~5334 The tail flip can be expected to activate the sensory organs of the vestibular system, and stimulate the hair cells in the ear, due to the body movement which compresses or expands the air bladders coupled to the ear.8,33.It will also stimulate the lateral line organs which are sensitive to the turbulence around the fish body.22a These sensory stimuli may represent “noise” to the CNS in the context of information processing. One possible mechanism for preventing this sensory input, which may be unnecessary, from entering the CNS is to inhibit the sensory organs during the movement. This type of inhibition occurs during the escape reflex mediated by the lateral giant axons of the crayfish where the axons trigger an efferent inhibition of sensory afferents and reiay interneurons.’ The latency and duration of this inhibition coincides with those of the animal’s movements. In the goldfish M-cell system, an efferent inhibition triggered by the M-cell has been demonstrated for the lateral line system, and the time course of this inhibition correlates well with that of the locomotion associated with the startle reflex.28.29This report is concerned with a short latency efferent inhibition of the auditory afferents *Present address: Department of Physiology and Biophysics, New York Medical Center, 550 First Avenue, New York, NY 10016, U.S.A. Abbreuiarions: CRN, cranial relay neuron; EOR, extracellular orthodromic response;. EPSP, excitatory postsynaptic potential; IPSP, inhibitory postsynaptic potentiai; M-cell, Mauthner cell.
(saccular nerve Mauthner cell.
fibers) triggered
by the goldfish
EXPERIMENTAL PROCEDURES Goldfish (Cmmius auratus), 10-t2cm long, were perfused with tap water containing the anesthetic tricaine, 70 mg/l, and immobilized with d-tubocurarine (l-3 pg/g body weight) throughout the experiments. The surgical and recording techniques were generally similar to those described previously. 9~2’However, the head was rotated laterally so that the entry of the saccular fibers into the brain could be visualized directly. Micr~l~tr~es used for intracellular recordings from the saccular fibers were filled with 1.25M KC1 (30-40 Ma). and the M-cell electrodes contained 2.5 M KC1 (6-10 MR). Experimental data were recorded on tape for later analysis. A Nicolet 1074 computer was used for on-line signal averaging. The recordings from the saccular fibers were obtained outside the brain, at the point where they entered the medulla. Sometimes, a second electrode was used for simultaneous intracellular recordings from the M-cell (Fig. 1A). The saceulac nerve could be visualized directly, and the identity of individual fibers was further confirmed by their physiological responses to sound. An example is shown in Fig. 1B, where a short duration 40~0Hz sound pulse was able to evoke excitatory postsynaptic potentials (EPSPs) and action potentials in the saccular fiber and EPSPs in the M-cell. The EPSPs recorded in the afferent were presumably mediated by transmitter released by the hair cells in the ear,ls whereas the EPSPs in the M-cell (upper trace in B) were due to the impulses in some of the saccular fibe.rs.gJ3 The visual and physiologi~l identifi~tions were also supported by the mo~hologi~l observation that individual saccular fibers stained with horseradish peroxidase could be traced to the sensory epithelium of the ear.23 RESULTS When the M-cell was activated antidromically,
its impulse was often followed by an all-or-none hyperpolarization in an ipsilateral saccular fiber (Fig. 1C). This response was not observed in ah recorded 829
830
J.-W. LIN and D. S. FABER
Aff.fiber
-9-
2mcrec
I
> E
Fig. 1. Identilication of saccular fibers and demonstration of the elferent IPSP. (A) General experimental arrangement. The saccular fiber recordings were obtained outside the brain, where the rierve entered the medulla and could be visualixed directly. Somttimes a second electrode was used to reccrd from the M-cell in the lateral dendrite or soma. (B) Simultaneous recordings from a saccular fiber (lower trace: AK fiber) and the M-cell lateral dendrite (upper trace: Lat. dent.) of sound evoked depolarizations (400 Hz sound pulse for 30ms). (C) Antidromic activation of the M-cell, indicated by the truncated impulses (upper traces), also evoked a later hyperpohrixation in the saccular fiber (lower trace). When the antidromic stimulation failed to fire the M-cell, there was no hyperpolarixation of the fiber. ‘I’he depolarixation following the M-cell antidromic impulse is a collateral inhibitory IPSP, inverted by chloride loading. Note that the amplitudes of the collateral IP!IP and the hyperpolarixation fluctuated in parallel, the largest responses having been evoked by the same stimulus. B and C are from separate experiments.
saccular fibers, and it was seen in both the afferents which terminate on the lateral dendrite as large myelinated club endings and are electrotonically coupled to that process,*2~23and those which are not presynaptic to the M-cell. For example, in the record of Fig. lC, which is from an experiment different from that shown in B, the afferent fiber was not electrically coupled to the M-cell, as there was no coupling potential associated with the M-cell antidromic impulse. 23The all or none hyperpolarixation was the same in both fiber types; it had a mean amplitude of 1.O f 0.85 mV (mean + standard deviation) (n = 37), a mean latency of 5.99 + 0.73 ms (n = 36), and an average half-decay time of 6.81 f 2.08 ms (n = 32). It should be noted that Mcell activation did not produce any significant extracellular held potentials in the saccular nerve, which otherwise would have distorted the postsynaptic potential waveform. The example. of Fig. 1C also illustrates that the amplitude of the hyperpolarization fluctuated in parallel with that of the depolarizations which fotlowed the antidromic spike in the M-cell. The latter was an inhibitory postsynaptic potential (IPSP) mediated by a collateral pathway and inverted by chloride loading. I4 We propose that the hyperpolarization of the saccular fibers is an IPSP mediated by an efferent inhibitory pathway (efferent IPSP) that is closely coupled to the collateral inhibitory network of the M-cell.
The efferent IPSP could also be activated by an impulse in the contralateral M-cell. In the exam@ shown in Fig. 2, the spinal stimulus activated only the M-cell contralateral to the recorded saccular fiber, as the collateral IPSP in the ipsilateral M-cell (A,) was not preceded by an antidromic imp&e. Under this condition, an efferent IPSP was still observed in the saccular fiber (A,). In the same experiment, the second M-cell could be activated by incmaaing the stimulus strength; the amplitudes and time course of efferent IPSPs activated by the firing of one (5) and both M-cells (A.+) were essentially the same. Therefore, as with the b&cell colIatera1 inhibitory network,‘,‘” the firing of either one of the two &cells can activate a full-sized efferent IPSP bilaternlly. The small hyperpolarizations seen in Fig. 2A2, when the spinal stimulus was below threshold for the M-axon, were not observed consistently and were not studied further. Finally, the role of the M-41 in the activation of the efferent IPSP was directly demonstrated by intracellular current injections in the axan of the ipsilateral M-cell (Fig. 2&). The e&rent IPSP thus activated had an amplitude and time course similar to that obtained with spinal stimulation (Fig. 2B,). Given the similar properties of the IPSPs evoked by spinal and intracellular stimulations, the first approach was used to activate the M-cell(s) in most of the experiments. The inhibitory effect of the efferent IPSP could
831
Efferent inhibition of auditory fibers via M-cell outputs
n=16
Fig. 2. M-cell firing can evoke efferent IPSPs bilaterally. (A,) and (A,) Simultaneous recordings from a saccular fiber and the ipsilateral M-cell, respectively. The spinal stimulus (arrow in A,) was adjusted for selective activation of the contralateral M-cell, at threshold, as indicated by the characteristic all or none collateral IPSP in the ipsilateral M-cell (A,). The superimposed traces show that the efferent IPSP was only evoked when the contralateral M-cell was fired (A,). Note that when the stimulus was below threshold for both M-axons, there was still a small hyperpolarization in the afferent. (A,) and (AJ Averaged traces (n = 16) of the efferent IPSPs activated by the contralateral M-cell alone (A,) or by the firing of both M-cells (AJ. The two IPSPs are essentially identical. Activation of the ipsilateral M-cell required a stimulation intensity greater than that used to fire only the contralateral one and was signalled by the presence of an antidromic coupling potential in the afferent (dot in A,). Traces in (A2)(A4) were obtained from the same saccular fiber. (B,) and (B2) In a different fiber, the efferent IPSP evoked by spinal stimulation (B,) had a time course similar to that of the IPSP due to direct intracellular activation of the ipsilateral M-cell (Br). The antidromic coupling potential mediated by the ipsilateral M-cell is indicated by the dot in (B,); it was masked in (B,), by “cross talk” associated with the intracellular stimulus. The traces in (B,) and (B2) are the averages of 32 sweeps each.
be demonstrated as a block or delay of impulses activated by direct current injections, as shown in Fig. 3A, and A,, respectively. In that experiment, the M-cell’s action potential was recorded extracellularly in the vicinity of the axon cap’, and the inhibitory effect was coupled to the firing of the ipsilateral cell. A more physiologically relevant and more striking inhibitory effect was observed by pairing the efferent IPSP with EPSPs evoked by sound. When a 400 Hz sound pulse which evoked a subthreshold EPSP in a saccular fiber (upper trace, Fig. 3B,) was paired with a spinal stimulation capable of evoking an IPSP in the same fiber (middle trace, B,), the EPSP was markedly attenuated (lower trace, B,). This inhibition was not due to algebraic summation of the two responses. Rather, it was associated with a strong shunting effect, since the amplitude of the net EPSP or difference potential (upper trace, B,), obtained by subtracting the IPSP from the paired response, was much smaller than that of the control EPSP itself. The magnitude of this conductance change closely followed the time course of the IPSP. When the latency of the sound pulse was increased so that the first peak of the EPSP coincided with the later phase of the IPSP, the inhibition was less pronounced (middle and lower traces, B2). Therefore, the efferent IPSP appeared to be generated at the dendrite of the HSC24;3-D
saccular fibers, i.e. in the ear, where its conductance change can be expected to effectively attenuate the amplitude of EPSPs produced by hair cells. To further pinpoint the location of the conductance change associated with the efferent IPSP, we examined the alternative explanation that there was a presynaptic inhibition mediated by axoaxonal synapses at the central endings of the saccular fibers.6 This possibility can be ruled out, at least for the endings terminating on the M-cell. If the IPSP were generated at the central axonal terminals, it should have resulted in a conductance increase there, and would in turn have attenuated the antidromic coupling potential recorded in the saccular fiber. However, when the coupling coefficient, defined as the ratio of the amplitude of the coupling potential recorded in a saccular fiber to that of the M-cell antidromic impulse recorded in its lateral dendrite, was measured during the peak of the efferent IPSP, this parameter was only 12.4 f 10.6% (n = 8) lower than in the control condition. In contrast, the same IPSP produced a 61.8 + 16% (n = 8) attenuation in the amplitude of the sound-evoked EPSPs. An example of this result is shown in Fig. 4. In A, the lower trace was obtained from the M-cell and antidromic impulses were activated by paired spinal stimulations, with the latency of the second impulse coinciding with the
J.-W.
832
LIN and D. S. FABER
A2
Smsec
BI
:I!.
‘.
.
.
82
1.
‘.:’ .
I
.
,:. 1
.I ,i $
Net
n=16
EPSP
! ii
Fig. 3. In~bito~ effect of the efferent IPSP. (A,) and (A,) Direct activation of a saccular fiber (lower traces) was i&Wed (A,) or delayed (AZ) when preceded by antidromic activation of the M-cell. Upper traces are extracellular recordings from theI+ce& which?n both cases was activated (dots) in one of the two superimposed records. ‘Ihe e&rent IPSP recorded in this fiber (not shown) had an amplitude of 0.38 mV and a latency of 7.3 ms, (II,) and (Br) Interactions, in another experiment, between the e&rent mSP and auditory evoked HPWs. (B,) A WHa sound pulse with a duration of 25 ms evoked a. compound EPSP in a saazular fiber (upper trace), while activation of the lWcclI(e) produced an IPSP (middls trace). When the two stimuli were paired So that tbe initial peak of the EPSP coincided with that of the IIW, the excitation was greatly attenuatad (lower trace). (82) Net EPSP, de&ted as the digerence between the paired response and the IPSP alone, for the timing conditions of (8,) (upper trace) and for cases where the sound pulses were delayed by 5 ms (middle) r&d 10 ms (lower). Arrows indicate the time that spinal ~rn~~n was applied. As the net EPgP is less than the control one, the ~~tow e&t is more than that pm&ted by linear sumrndion of the two potentials. Note that the level of inl@ition was decreased when the BPSPs occurred dtig the falling phase of the IPSP. All traces in (B,) and (BJ are averages of 16 sweeps.
peak of the efferent IPSP. The smaller amplitude of the second impulse was due to tbe shunting effect produced by tbe collateral IPSP in the M-cell soma. Both antidromic spikes produced coupling potentials (dots on tbe upper trace of A) in tbe saecular fiber, and the coupling coe@cients mediated by the fbst and second impulses were 0.067 and 0.064 respectively. In the same fiber, the reduction of the sound-evoked EPSP was as much as 76%. The reduction was calculated by comparing tbe amplitude of the first major peak of the control EPSP (upper trace, B) with the remaining depolarization measured from the net EPSP at the same latency as tbe major peak (lower trace, B). This example clearly illustrates that the
inbibitory conductance increase at the central terminal is quite small, if present at all. Thus, the results of Figs 3 and 4 together suggest tbe ~~~~n~ change associated with tbe IPgP is most l&&y generated in the ear rather than at the central terminals of the afferents. This conclusion is consistent with electron microscopic studies of the synaptic endings on the M-cell, which failed to reveal any axoaxonal synapses on the saccular nerve terminals.2s36 In addition, it is clear that tbe e&rent IPSP cannot he due to electrotonic transmission of the M-cell’s collateral inhibitory response, as tbese two potentials have opposing polarities when tbe M-cell is Cl--loaded (Fig. 4A).
Efferent inhibition of auditory fibers via M-cell outputs
a
B
1 ‘! .. .:
-: ;I
4 .*
.: :
*;
i
:I s
M-cell
Smsec
*
<
:t
Net
EPSP
-Ihi
1 Omsec Fig. 4. Localization of the site of IPSP generation. (A) Tbe efferent IPSP does not shunt antidromic coupling potentials. Paired spinal stimulations were used to activate the M-cell (lower trace, M-cell) and the second antidromic impulse was timed to occur at the peak of the IPSP (upper trace, saccular fiber). The amplitude-of the second imp&e was attenuated bv the collateral IPSP. which had been inverted bv chloride loading. Both impulses produced electrotonic coupling potentials in the saccular fiber (dots in the upper trace). The coupling coefficient for the two spikes were the same, indicating that the input resistance of the saccular fiber at its central terminal remained the same during the IPSP. (B) In the same fiber, the initial peak of the sound-evoked EPSP had a control amplitude of 0.96 mV (upper trace) whereas when it was paired with the efferent IPSP, the net EPSP measured at the same latency was only 0.23 mV (lower
trace). The EPSP reduction can only be attributed to a conductance increase in the dendrite of the saccular fiber.
An alternative explanation for the efferent IPSP is based on the assumption that the hair cells in the ear depolarize the afferents tonically; thus, the hyperpolarization could be mediated by the known efferent inhibition of the hair cells themselves.” If this were the case, a depolarization of the saccular fibers, produced by current injection, should have reduced the driving force of the tonic excitatory inputs and, therefore, decreased the amplitude of this tonic
833
EPSP. An inhibition of the hair cells would then further decrease, or eliminate, the diminished tonic excitatory input and would produce a smaller efferent IPSP. On the other hand, if the IPSP was mediated by a conductance change in the afferents such depolarizations should increase the amplitude of the IPSP, by increasing its driving force. The observed voltage sensitivity of efferent IPSPs evoked during subthreshold depolarizations and limited hyperpolarizations supports the latter possibility. In the example of Fig. 5, a 1.6 nA depola~~ng current pulse injected into the saccular fiber produced approximately an 8 mV depolarization and increased the IPSP amplitude from 1 mV to 1.58 mV. A hyperpolarizing current of the same intensity reduced the IPSP amplitude to 0.75mV (not shown). Similar observations were made in more than 60% of the fibers tested. We were, however, not able to invert the IPSP (i) with large hy~~la~~tions, due to the instability of the electrode resistance during the injection of large currents, or (ii) by chloride loading, presumably due to a large spatial separation between the electrode and the site of conductance change. In fact, it is not possible to clearly demonstrate that the IPSP is due to a chloride dependent conductance change, rather than to a potassium dependent one. DISCUSSION
The evidence presented in this report indicates that the activation of the M-cell can evoke an IPSP in auditory afferents and this IPSP is probably generated in the ear. This is the first intracellular demonstration of an efferent inhibitory potential on such sensory afferents in the vertebrate central nervous system, although mo~holo~cal studies have already reported the presence of presumptive inhibitory synapses on auditory and vestibular afferents.27~30~3’ Our results can be contrasted with the finding that vestibular afferents in the toadfish are instead excited by efferent stimulation, an effect which has been
> E v-
Fig. 5. Effect of shifting membrane potential on the amplitude of the efferent IPSP. Superimposed traces show the IPSP recorded in a saccular fiber at resting membrane potential, - 70 mV, and the larger response obtained during a 20 ms, 1.6 nA depolarizing current pulse. ‘Ihe electrotonic coupling potentials evoked by the M-cell antidromic impulse are indicated by the solid circles. Both traces are the averages of 16 sweeps.
834
J.-W. LAN and D. S. FABEK
linked to an arousal preceding movements associated with feeding, fighting, etcX Comparisons between the efferent IPSP and the inhibition of the hair cells” suggest that they are mediated by separate pathways. First, the inhibition of the hair cells was cholinergic and blocked by d-tubocurarine at a dose of l-2 mg/kg body weight.” In our experiments, the same dosage was routinely used to immobili~ the animal and it did not interfere with the recording of IPSPs. Second, the latency of the efTerent IPSP was shorter than that reported for the inhibition of the hair cells. Specifically, the 6 ms latency of the IPSP, measured from the artifact produced by spinal stimulation to the onset of the IPSP, includes the following components: (1) the time required for impulse conduction from the spinal cord to the medulla, at least 0.5 ms’, and (2) at least one synaptic deIay, from the M-cell to an interneuron, about 0.4 ms.” Therefore, the latency of the efferent IPSP would be no more than 5 ms if the efferent fibers were activated directly by a stimulating electrode located on the saccular nerve. In contrast, the inhibition of the hair cells had a latency of 6-7 ms when it was produced by a direct stimulation of the nerve. Third, the peak of the efferent IPSP, which coincided with the maximal inhibition of the sound-evoked EPSPs (Fig. 3B1), had a latency of about 779ms whereas the peak inhibition of the hair cells occurred 12 ms after the stimulation of the saccular nerve. Fourth, the hair cell inhibition lasts about 40ms while the duration of the efferent IPSP ranged from 9.28 to 19 ms, with one exception of 34 ms. Finally, it has been reported that inhibition of hair cells never produced any hy~rpolari~tion in the saccular fibers. This observation is consistent with our finding that the saccular fiber recordings normally showed a noise-free baseline, and did not seem to be depolarized by any tonic excitatory input, i.e. there was a low frequency of spontaneous EPSPs. Therefore, the different pharmacological properties and time courses of the two inhibitions enable us to distinguish between them. In confirmation. ultrastructural studies of the goldfish ear demonstrated that there were two types of synapses on the dendrites of the saccular fibers. one being the excitatory junctions established by the hair cells while the other type, which was observed less frequently, was suggested to be inhibitory and might mediate the inhibition reported here.27 The rare occurrence of inhibitory synapses on the auditory afferents is consistent with our experience that not every fiber impaled exhibited an efferent IPSP. The time course of the efferent inhibition can account for the presynaptic component of the third type of inhibition defined by Furukawa et af.” In that report, they found that the firing of the M-cell evoked a feedback inhibition in the lateral dendrite (dendritic inhibition), which can be differentiated from the collateral in~bition located primarily at the soma on the basis of latency, duration, sensitivity to procaine
and inhibition of the EPSPs activated by eighth nerve stimulations, Coupled to this dendritic inhibition was a reduction in the extracellular field potential recorded in the M-cell axon cap (extracellular orthodromic response; EOR) following saccular nerve stimulation. In fact, it is now known that the EOR originated from a population of inhibitory interneurons that were activated by eighth nerve via electrical synapses.35 The EOR ampIitude, which presumably reflects the number of the inhibitory interneurons activated, then provides an estimate of the number of eighth nerve fibers that were activated. The efferent IPSP and the feedback inhibition of the EOR have the same latencies, peak times and durations. Therefore, we propose that the decreased excitability of the saccular fibers is partly due to the action of the efferent pathway. However, this still does not totally account for the proposed dendritic inhibition, which may last longer than the efferent IPSP.‘~‘3 Finally, it should be noted that the dendritic inhibition and the inhibition of the EOR could be completely blocked by strychnine,” which suggests a glycinergic mechanism, although it is generally accepted that acetylcholine is the transmitter for efferent inhibitionI The firing of one M-cell typically activates a collateral inhibitory pathway which prevents further impulse activity in both M-ceils during the execution of the startle reflex.’ Our findings add a new component to this feedback network: the firing of the M-cell produces bilateral inhibition of the auditory fibers. The components of the proposed pathway postsynaptic to the M-axon are shown in the scheme of Fig. 6. First, since the efferent inhjbition is bilateral, it is likely that the circuit involves interposed relay neurons similar to or the same as those which are in the collateral inhibitory network” and excite cranial motoneurons involved in head-level components of the startle response.i8 In some experiments, when the recording quality of the saccular fiber was ideal, we noted that the amplitudes of the efferent and collateral IPSPs fluctuated in parallel (see Fig. 1C for an example). This observation suggests that both inhibitory systems may share a common pathway, and the so-called cranial relay neurons (CRNs) are the most logical candidates for this role. Specifically, it is likely that CRNs monosynaptically activate the efferent inhibitory interneurons (Fig. 6) whose relatively slow impulse conduction velocity would account for the 56ms latency.” Then, the ~varia~ons of the IPSP amplitudes would be due to the occasional failure of some CRNs (however, see Ref. 36). The interneurons mediating the efferent inhibition evoked by the M-cell have not been identified. In our previous morphological study of the saccular inputs to the M-cel12’, where horseradish peroxidase was applied to the cut end of the saccular nerve, we found a cluster of stained cell bodies located at the level of the IV ventricle and near to the M-axon (unpublished observations). Similar to some
Efferent inhibition of auditory fibers via M-cell outputs
CiiN
Fig. 6. Schematic representation of the M-cell’s feedback inhibitory network. The dotted line delineates the midline of the fish brain. M-cell send axons cross the midline and monosynaptically activate cranial relay neurons (CRNs), one of which is shown. The CRNs in turn are excitatory to a group of inhibitory interneurons (Int) which produce the collateral IPSP in the M-cell, and to cranial motoneurons, such as those in the trigeminal nucleus Q’). Efferent inhibitory intemeurons are presumably also activated by the CRNs. As the synapses interposed between these two groups of cells have not been defined, they are represented by the rectangular box. The responses are bilateral since each CRN is postsynaptic to both M-cells. The efferent intemeurons inhibit VIIIth nerve fibers which terminate on the M-cell as well as those which do not. The latter group synapse on the secondary neurons of the acousticovestibular nuclei (broken circle). Although it is likely that the efferent inhibition of the hair cells (see text) is also coupled to the M-cell feedback inhibitory pathway, this connection is not included, due to the lack of definitive evidence for it.
of the efferents of the lateral
line organs,36 these neurons sent their axons rostrally and turned laterally at the level of the M-cell before exiting from the medulla. It remains to be determined whether these efferent axons make svnaotic contact with hair cells or with saccular fibers. Finally, it also is not clear
835
what types of saccular fibers, in addition to those which give rise to large myelinated club endings, are inhibited by the M-cell network, because we found that not all the saccular fibers were inhibited. The characterization of physiological and morphological properties of the inhibited fibers may provide further insight into the specific sensory information that the M-cell network controls. The physiological significance of this efferent inhibition has been stated in general terms in the introduction. To further evaluate the specific functions of the efferent inhibition, it is necessary to compare its time course with that of the tail tlip evoked by the M-cell. The latency between the firing of the goldfish M-cell and the beginning of a tail flip is about 10-12 ms,29 and the duration of the body movement defined as Kinematic stage 1 by Webb32 was about 20 ms.4 Apparently, the efferent inhibition, with an initial latency of 6 ms and a peak latency of 8 ms, starts and reaches its peak before the beginning of the body movement. Furthermore, in a restrained fish the inhibition is terminated at about 26 ms after the firing of the M-cell, or before the first stage of the tail Ilip is completed (30-32 ms). Therefore, the main function of the efferent inhibition appears to be the termination of auditory evoked inputs before the fish moves. Presumably, the auditory inputs which initiate the M-cell firing become “irrelevant” once the escape “command” is issued by the M-cell, and the temporary silence of the saccular fibers may prepare the animal for the swimming behavior which follows the tail fli~.~ It should also be pointed out that afferent suppression cannot be accomplished as effectively by inhibition of the hair cells, if the afferents are capable of firing repetitively after they are activated.’ On the other hand, the latency and duration of the efferent inhibition of the hair cells described by Furukawa” suggest this latter process may be more appropriately timed to block the auditory noise generated by the tail Ilip itself. In addition, the inhibition of the hair cells may have an added advantage of preserving the “gain” of the hair cellafferent synapses, because this synapse habituates after repetitive stimulation.i6 Acknowledgements-We thank Dr Joseph Fetch0 for comments on the manuscript, Julie Lakatos for graphics and Jan Jordan for typing. This work was supported by NIH Grant No. NSl5335.
REFERENCES
1. Bryan J. S. and Krasne F. B. (1977) Presynaptic inhibition: the mechanism of protection from habituation of the crayfish lateral giant fibre escape response. J. Physiol., Land. 271, 369-390. 2. Dijkgraaf S. (1963) The functioning and significance of the lateral line organs. Biol. Rev. 38, 51-105. 3. E&on R. C. and Bombardieri R. A. (1978) Behavioral functions of the Mauthner neuron. In Neurobiology of the Maurhner Cell (eds Faber D. S. and Kom H.). 221-224. Raven Press. New York. 4. Eaton R. C., Bombardieri R. A. and Meyer D.‘I. (1977) The Mauthner-initiated startle response in teleost fish. J. exp. Biol. 66, 65-8 1. 5. Eaton R. C. and Hackett J. J. (1984) The role of the Mauthner cell in fast-starts involving escape in teleost fishes. In Neural Mechanisms o~Srarrfe Eehuuior (ed. Eaton R. C.), pp. 213-266. Plenum Press, New York. 6. Eccles J. C., Eccles R. M. and Magni F. (1961) Central inhibitory action attributable to presynaptic depolarization produced by muscle afferent volleys. J. Physiol., Land. 159, 147-166.
J.-W. LIN and D. S. FABER
X36
7. Faber D. S. and Korn H. (1978) Eiectrophysiology of the Mauthner cell: Basic properties, synaptic mechanisms. and associated network. In ~eu~o~~u~~g~qfzhe ~~~z~~e~ Cetl (eds Faber D. S. and Kern H.). pp. 47-132. Raven Press. New York. 8. Fay R. R., Hillery C. M. and Bolan K. (1982) Representation of sound pressure and particle motion information in the midbrain of the goldfish. Camp. Bioclrem. Ph~siol. 71A. 181-192. 9. Furshpan E. J. and F&kawa T. (i962) Intracellular and extracellular responses of the several regions of the Mauthner cell of the goldfish. J. Neurophysiol. 25, 732-771. 10. Furukawa T. (1966) Synaptic interaction at the Mauthner cell of goldfish. Prog. Brain Res. 21A, 44-70. Il. Furukawa T. (1981) Effects of efferent stimulation on the saccule of goldfish. J. Physiol., Land. 315, 203-215. 12. Furukawa T., Fukami Y. and Asada Y. (1963) A third type of inhibition in the Mauthner cell of goldfish. J. Neurophysioi. 26, 759-774. 13. Furukawa T., Fukami Y. and Asada Y. (1964) Effects of strychnine and procaine on collateral inhibition of the Mauthner cell of goldfish. Jup. J. Pbysjo~. 14, 386-399. 14. Furukawa T. and Furshpan E. J. (1963) Two inhibitory mechanisms in the Mauthner neurons of goldfish. J. Neurophysiol. 26, 140-175. 15. Furukawa T. and Ishii Y. (1967) Effects of stat+ binding of sensory hairs on sound reception in the goldfish. sup. J. Physiol. 17, 572-588. 16. Furukawa T. and Matsura S. (1978) Adaptive rundown of excitatory post-synaptic potentials at synapses between hair cells and eighth nerve fibres in the goldfish. J. Physiol., Land. 276, 193-209.
17. Guth P. S., Norris C. H. and Bobbin R. P. (1976) The pharmacology of transmission in the peripheral auditory system. Pharmac. Ret~. 28, 95-125. 18. Hackett J. T. and Faber D. S. (1983) Mauthner
axon networks mediating supraspinai components of the startle response in the goldfish. N~ro~~~~ce &, 3 17-33 1. 19. Hackett J. T. and Faber D. S. (1983) Relay neurons mediate collateral inhibition of the goId&h Mautlmer cell. Brain Res. 264, 302-306. 20. Highstein S. M. and Baker R. (1985) Action of the efferent vestibular system on primary agerents in the toadfish, Opsanus tau. J. Neurophysiol. 54, 370-384. 21. Kom H. and Faber D. S. (1975) An electrically mediated inhibition in gold&h medulla. J. Newophysioi. 38.452-471. 22. Lin J.-W. (1986) Physiology and Morphology of Identified Mixed Excitatory Synapses on the Goldfiab Mauthner Cell.
Ph.D. thesis, SUNY at Bufialo. Buffalo, NY. 23. Lin J.-W. and Faber D. S. (In press) Synaptic transmission mediated by single club endings on the goldfish Mauthner cell. I. Characteristics of electrotonic and chemical postsynaptic potedais. J. Neurosci. In press. 24. Lin J.-W., Faber D. S. and Wood M. R. (1983) Organized projection of the goidftsfi saccnlar nerve onto the Mauthner cell lateral dendrite. Brain Res. 274, 3 19-324. 25. Nakajima Y. (1974) Fine structure of the synaptic endings on the Mauthner cell of the goldfish. J. camp. Newof. ISa, 35742. 26. Nakajima Y. and Kohno K. (1978) Fine structure of the Mauthner cell: Synaptic topography and comparative study. In Neurobiology of the Muutbner Cell (eds Fabr D. S. and Kom H.), pp. 133-166. Raven Press, New York. 27. Nakajima Y. and Wang D. W. (1974) Morphology of afferent and efferent synapses in the hearing organ of the golclftsh. J. camp. Neural. 156, 403-416. 28. Russell I. J. (1974) Central and peripheral inhibition of lateral line input during the startle response in goldfish. Bruin Res. So, 517-522. 29. Russell I. J. (1976) Central in~bitjon of lateral line input in the medulla of the goldfish by neurones which control active body movements. J. camp. Physiof. 111, 335-358. 30. Spoendlin H. (1973) The innervation of the cochlear receptor. In Basic Mechanisms in Hear&g (ed. Msller A. R.), pp. 185-234. Academic Press, New York. 31. Takasaka T. and Smith C. A. (1971) The structure and innervation of the .&eon’s basilar _ Dapilla. J. ultrastruct. Rex _ _ .., 35, 2[M5. 32. Webb P. W. (1976) The effect of size on the fast-start performance of rainbow trout Salmo gairdneri, and a consideration of piscivorous predator-prey interactions. J. exp. Biol. 65, 157-177. 33. Weber E. H. (1820) De sure et auditi hominis et animals, Pars I. De aure anbnaltun aquutikm, Lipsiae. 34. Yasargil G. M. and Diamond J. (1968) Startle-response in teleost fish: an elementary circuit for neural discrimination. I
Nature, Land. 226, 241-243. 35. Zottoli S. J. and Faber D. S. (1980) An identifiable class of s~toacoustic
in~~eurons with bilateral projections in the goldfish medulla. Neuroscience 5, 1287-I 302. 36. Zottoli S. J. and van Home C. (1983) Posterior lateral line tierent and efferent pathways within the central nervous system of the goldfish with special reference to the Mauthner cell. J. camp. Neurol. 219, 10&l 11. (Accepted 20 August 1987)
0306-4522/88
1988
$3.00 + 0.00
Pergamon Press plc Q 1988 lBR0
AN EFFERENT INHIBITION OF AUDITORY AFFERENTS MEDIATED BY THE GOLDFISH MAUTHNER CELL J.-W. LIN* and D. S. FABER Department of Physiology, State University of New York at Buffalo, Buffalo, NY 14214, U.S.A. A~act-Intra~ilular recordings from goldfish auditory afferents revealed a hy~~la~~tion triggered by a single impulse in the Mauthner cell. The firing of either one of the two Mauthner cells alone was sufficient to evoke this potential change. The all or none hyperpolarization, which could only be recorded in some auditory fibers, presumably was an inhibitory postsynaptic potential. The inhibitory postsynaptic potential typically had a latency of 6 ms, an amplitude of 1 mV and a half-decay time of 6.8 ms; it could block or delay impulses evoked by direct current injection and could attenuate the amplitude of excitatory postsynaptic potentials evoked by sound pulses. However, this inhibitory postsynaptic potential did not reduce the amplitudes of electrotonic coupling potentials produced by antidromic impulses in the Mauthner ceil. We propose that the inhibitory postsynaptic potential is generated at the dendrites of the auditory fibers, i.e. in the ear, rather than at the central terminals of the aBerent, where the antidromi~ coupling potentials originate. The possibility that the inhibitory postsynaptic potential actually represented disfacilitation due to an efferent inhibition of the hair cells, which tonically depolarize the saccular fibers, was ruled out because depolarization of these fibers increased the inhibitory postsynaptic potential amplitude. Possible morphological substrates for the efferent inhibition and the behavioral significance of this inhibition are discussed
The Mauthner (M-) cells in the goldfish medulla are a pair of giant interneurons that mediate a startle reflex. When one of the M-cells fires an action potential it causes a massive contraction of the trunk musculature contralateral to its soma and produces a rapid tail flip. 3~5334 The tail flip can be expected to activate the sensory organs of the vestibular system, and stimulate the hair cells in the ear, due to the body movement which compresses or expands the air bladders coupled to the ear.8,33.It will also stimulate the lateral line organs which are sensitive to the turbulence around the fish body.22a These sensory stimuli may represent “noise” to the CNS in the context of information processing. One possible mechanism for preventing this sensory input, which may be unnecessary, from entering the CNS is to inhibit the sensory organs during the movement. This type of inhibition occurs during the escape reflex mediated by the lateral giant axons of the crayfish where the axons trigger an efferent inhibition of sensory afferents and reiay interneurons.’ The latency and duration of this inhibition coincides with those of the animal’s movements. In the goldfish M-cell system, an efferent inhibition triggered by the M-cell has been demonstrated for the lateral line system, and the time course of this inhibition correlates well with that of the locomotion associated with the startle reflex.28.29This report is concerned with a short latency efferent inhibition of the auditory afferents *Present address: Department of Physiology and Biophysics, New York Medical Center, 550 First Avenue, New York, NY 10016, U.S.A. Abbreuiarions: CRN, cranial relay neuron; EOR, extracellular orthodromic response;. EPSP, excitatory postsynaptic potential; IPSP, inhibitory postsynaptic potentiai; M-cell, Mauthner cell.
(saccular nerve Mauthner cell.
fibers) triggered
by the goldfish
EXPERIMENTAL PROCEDURES Goldfish (Cmmius auratus), 10-t2cm long, were perfused with tap water containing the anesthetic tricaine, 70 mg/l, and immobilized with d-tubocurarine (l-3 pg/g body weight) throughout the experiments. The surgical and recording techniques were generally similar to those described previously. 9~2’However, the head was rotated laterally so that the entry of the saccular fibers into the brain could be visualized directly. Micr~l~tr~es used for intracellular recordings from the saccular fibers were filled with 1.25M KC1 (30-40 Ma). and the M-cell electrodes contained 2.5 M KC1 (6-10 MR). Experimental data were recorded on tape for later analysis. A Nicolet 1074 computer was used for on-line signal averaging. The recordings from the saccular fibers were obtained outside the brain, at the point where they entered the medulla. Sometimes, a second electrode was used for simultaneous intracellular recordings from the M-cell (Fig. 1A). The saceulac nerve could be visualized directly, and the identity of individual fibers was further confirmed by their physiological responses to sound. An example is shown in Fig. 1B, where a short duration 40~0Hz sound pulse was able to evoke excitatory postsynaptic potentials (EPSPs) and action potentials in the saccular fiber and EPSPs in the M-cell. The EPSPs recorded in the afferent were presumably mediated by transmitter released by the hair cells in the ear,ls whereas the EPSPs in the M-cell (upper trace in B) were due to the impulses in some of the saccular fibe.rs.gJ3 The visual and physiologi~l identifi~tions were also supported by the mo~hologi~l observation that individual saccular fibers stained with horseradish peroxidase could be traced to the sensory epithelium of the ear.23 RESULTS When the M-cell was activated antidromically,
its impulse was often followed by an all-or-none hyperpolarization in an ipsilateral saccular fiber (Fig. 1C). This response was not observed in ah recorded 829
830
J.-W. LIN and D. S. FABER
Aff.fiber
-9-
2mcrec
I
> E
Fig. 1. Identilication of saccular fibers and demonstration of the elferent IPSP. (A) General experimental arrangement. The saccular fiber recordings were obtained outside the brain, where the rierve entered the medulla and could be visualixed directly. Somttimes a second electrode was used to reccrd from the M-cell in the lateral dendrite or soma. (B) Simultaneous recordings from a saccular fiber (lower trace: AK fiber) and the M-cell lateral dendrite (upper trace: Lat. dent.) of sound evoked depolarizations (400 Hz sound pulse for 30ms). (C) Antidromic activation of the M-cell, indicated by the truncated impulses (upper traces), also evoked a later hyperpohrixation in the saccular fiber (lower trace). When the antidromic stimulation failed to fire the M-cell, there was no hyperpolarixation of the fiber. ‘I’he depolarixation following the M-cell antidromic impulse is a collateral inhibitory IPSP, inverted by chloride loading. Note that the amplitudes of the collateral IP!IP and the hyperpolarixation fluctuated in parallel, the largest responses having been evoked by the same stimulus. B and C are from separate experiments.
saccular fibers, and it was seen in both the afferents which terminate on the lateral dendrite as large myelinated club endings and are electrotonically coupled to that process,*2~23and those which are not presynaptic to the M-cell. For example, in the record of Fig. lC, which is from an experiment different from that shown in B, the afferent fiber was not electrically coupled to the M-cell, as there was no coupling potential associated with the M-cell antidromic impulse. 23The all or none hyperpolarixation was the same in both fiber types; it had a mean amplitude of 1.O f 0.85 mV (mean + standard deviation) (n = 37), a mean latency of 5.99 + 0.73 ms (n = 36), and an average half-decay time of 6.81 f 2.08 ms (n = 32). It should be noted that Mcell activation did not produce any significant extracellular held potentials in the saccular nerve, which otherwise would have distorted the postsynaptic potential waveform. The example. of Fig. 1C also illustrates that the amplitude of the hyperpolarization fluctuated in parallel with that of the depolarizations which fotlowed the antidromic spike in the M-cell. The latter was an inhibitory postsynaptic potential (IPSP) mediated by a collateral pathway and inverted by chloride loading. I4 We propose that the hyperpolarization of the saccular fibers is an IPSP mediated by an efferent inhibitory pathway (efferent IPSP) that is closely coupled to the collateral inhibitory network of the M-cell.
The efferent IPSP could also be activated by an impulse in the contralateral M-cell. In the exam@ shown in Fig. 2, the spinal stimulus activated only the M-cell contralateral to the recorded saccular fiber, as the collateral IPSP in the ipsilateral M-cell (A,) was not preceded by an antidromic imp&e. Under this condition, an efferent IPSP was still observed in the saccular fiber (A,). In the same experiment, the second M-cell could be activated by incmaaing the stimulus strength; the amplitudes and time course of efferent IPSPs activated by the firing of one (5) and both M-cells (A.+) were essentially the same. Therefore, as with the b&cell colIatera1 inhibitory network,‘,‘” the firing of either one of the two &cells can activate a full-sized efferent IPSP bilaternlly. The small hyperpolarizations seen in Fig. 2A2, when the spinal stimulus was below threshold for the M-axon, were not observed consistently and were not studied further. Finally, the role of the M-41 in the activation of the efferent IPSP was directly demonstrated by intracellular current injections in the axan of the ipsilateral M-cell (Fig. 2&). The e&rent IPSP thus activated had an amplitude and time course similar to that obtained with spinal stimulation (Fig. 2B,). Given the similar properties of the IPSPs evoked by spinal and intracellular stimulations, the first approach was used to activate the M-cell(s) in most of the experiments. The inhibitory effect of the efferent IPSP could
831
Efferent inhibition of auditory fibers via M-cell outputs
n=16
Fig. 2. M-cell firing can evoke efferent IPSPs bilaterally. (A,) and (A,) Simultaneous recordings from a saccular fiber and the ipsilateral M-cell, respectively. The spinal stimulus (arrow in A,) was adjusted for selective activation of the contralateral M-cell, at threshold, as indicated by the characteristic all or none collateral IPSP in the ipsilateral M-cell (A,). The superimposed traces show that the efferent IPSP was only evoked when the contralateral M-cell was fired (A,). Note that when the stimulus was below threshold for both M-axons, there was still a small hyperpolarization in the afferent. (A,) and (AJ Averaged traces (n = 16) of the efferent IPSPs activated by the contralateral M-cell alone (A,) or by the firing of both M-cells (AJ. The two IPSPs are essentially identical. Activation of the ipsilateral M-cell required a stimulation intensity greater than that used to fire only the contralateral one and was signalled by the presence of an antidromic coupling potential in the afferent (dot in A,). Traces in (A2)(A4) were obtained from the same saccular fiber. (B,) and (B2) In a different fiber, the efferent IPSP evoked by spinal stimulation (B,) had a time course similar to that of the IPSP due to direct intracellular activation of the ipsilateral M-cell (Br). The antidromic coupling potential mediated by the ipsilateral M-cell is indicated by the dot in (B,); it was masked in (B,), by “cross talk” associated with the intracellular stimulus. The traces in (B,) and (B2) are the averages of 32 sweeps each.
be demonstrated as a block or delay of impulses activated by direct current injections, as shown in Fig. 3A, and A,, respectively. In that experiment, the M-cell’s action potential was recorded extracellularly in the vicinity of the axon cap’, and the inhibitory effect was coupled to the firing of the ipsilateral cell. A more physiologically relevant and more striking inhibitory effect was observed by pairing the efferent IPSP with EPSPs evoked by sound. When a 400 Hz sound pulse which evoked a subthreshold EPSP in a saccular fiber (upper trace, Fig. 3B,) was paired with a spinal stimulation capable of evoking an IPSP in the same fiber (middle trace, B,), the EPSP was markedly attenuated (lower trace, B,). This inhibition was not due to algebraic summation of the two responses. Rather, it was associated with a strong shunting effect, since the amplitude of the net EPSP or difference potential (upper trace, B,), obtained by subtracting the IPSP from the paired response, was much smaller than that of the control EPSP itself. The magnitude of this conductance change closely followed the time course of the IPSP. When the latency of the sound pulse was increased so that the first peak of the EPSP coincided with the later phase of the IPSP, the inhibition was less pronounced (middle and lower traces, B2). Therefore, the efferent IPSP appeared to be generated at the dendrite of the HSC24;3-D
saccular fibers, i.e. in the ear, where its conductance change can be expected to effectively attenuate the amplitude of EPSPs produced by hair cells. To further pinpoint the location of the conductance change associated with the efferent IPSP, we examined the alternative explanation that there was a presynaptic inhibition mediated by axoaxonal synapses at the central endings of the saccular fibers.6 This possibility can be ruled out, at least for the endings terminating on the M-cell. If the IPSP were generated at the central axonal terminals, it should have resulted in a conductance increase there, and would in turn have attenuated the antidromic coupling potential recorded in the saccular fiber. However, when the coupling coefficient, defined as the ratio of the amplitude of the coupling potential recorded in a saccular fiber to that of the M-cell antidromic impulse recorded in its lateral dendrite, was measured during the peak of the efferent IPSP, this parameter was only 12.4 f 10.6% (n = 8) lower than in the control condition. In contrast, the same IPSP produced a 61.8 + 16% (n = 8) attenuation in the amplitude of the sound-evoked EPSPs. An example of this result is shown in Fig. 4. In A, the lower trace was obtained from the M-cell and antidromic impulses were activated by paired spinal stimulations, with the latency of the second impulse coinciding with the
J.-W.
832
LIN and D. S. FABER
A2
Smsec
BI
:I!.
‘.
.
.
82
1.
‘.:’ .
I
.
,:. 1
.I ,i $
Net
n=16
EPSP
! ii
Fig. 3. In~bito~ effect of the efferent IPSP. (A,) and (A,) Direct activation of a saccular fiber (lower traces) was i&Wed (A,) or delayed (AZ) when preceded by antidromic activation of the M-cell. Upper traces are extracellular recordings from theI+ce& which?n both cases was activated (dots) in one of the two superimposed records. ‘Ihe e&rent IPSP recorded in this fiber (not shown) had an amplitude of 0.38 mV and a latency of 7.3 ms, (II,) and (Br) Interactions, in another experiment, between the e&rent mSP and auditory evoked HPWs. (B,) A WHa sound pulse with a duration of 25 ms evoked a. compound EPSP in a saazular fiber (upper trace), while activation of the lWcclI(e) produced an IPSP (middls trace). When the two stimuli were paired So that tbe initial peak of the EPSP coincided with that of the IIW, the excitation was greatly attenuatad (lower trace). (82) Net EPSP, de&ted as the digerence between the paired response and the IPSP alone, for the timing conditions of (8,) (upper trace) and for cases where the sound pulses were delayed by 5 ms (middle) r&d 10 ms (lower). Arrows indicate the time that spinal ~rn~~n was applied. As the net EPgP is less than the control one, the ~~tow e&t is more than that pm&ted by linear sumrndion of the two potentials. Note that the level of inl@ition was decreased when the BPSPs occurred dtig the falling phase of the IPSP. All traces in (B,) and (BJ are averages of 16 sweeps.
peak of the efferent IPSP. The smaller amplitude of the second impulse was due to tbe shunting effect produced by tbe collateral IPSP in the M-cell soma. Both antidromic spikes produced coupling potentials (dots on tbe upper trace of A) in tbe saecular fiber, and the coupling coe@cients mediated by the fbst and second impulses were 0.067 and 0.064 respectively. In the same fiber, the reduction of the sound-evoked EPSP was as much as 76%. The reduction was calculated by comparing tbe amplitude of the first major peak of the control EPSP (upper trace, B) with the remaining depolarization measured from the net EPSP at the same latency as tbe major peak (lower trace, B). This example clearly illustrates that the
inbibitory conductance increase at the central terminal is quite small, if present at all. Thus, the results of Figs 3 and 4 together suggest tbe ~~~~n~ change associated with tbe IPgP is most l&&y generated in the ear rather than at the central terminals of the afferents. This conclusion is consistent with electron microscopic studies of the synaptic endings on the M-cell, which failed to reveal any axoaxonal synapses on the saccular nerve terminals.2s36 In addition, it is clear that tbe e&rent IPSP cannot he due to electrotonic transmission of the M-cell’s collateral inhibitory response, as tbese two potentials have opposing polarities when tbe M-cell is Cl--loaded (Fig. 4A).
Efferent inhibition of auditory fibers via M-cell outputs
a
B
1 ‘! .. .:
-: ;I
4 .*
.: :
*;
i
:I s
M-cell
Smsec
*
<
:t
Net
EPSP
-Ihi
1 Omsec Fig. 4. Localization of the site of IPSP generation. (A) Tbe efferent IPSP does not shunt antidromic coupling potentials. Paired spinal stimulations were used to activate the M-cell (lower trace, M-cell) and the second antidromic impulse was timed to occur at the peak of the IPSP (upper trace, saccular fiber). The amplitude-of the second imp&e was attenuated bv the collateral IPSP. which had been inverted bv chloride loading. Both impulses produced electrotonic coupling potentials in the saccular fiber (dots in the upper trace). The coupling coefficient for the two spikes were the same, indicating that the input resistance of the saccular fiber at its central terminal remained the same during the IPSP. (B) In the same fiber, the initial peak of the sound-evoked EPSP had a control amplitude of 0.96 mV (upper trace) whereas when it was paired with the efferent IPSP, the net EPSP measured at the same latency was only 0.23 mV (lower
trace). The EPSP reduction can only be attributed to a conductance increase in the dendrite of the saccular fiber.
An alternative explanation for the efferent IPSP is based on the assumption that the hair cells in the ear depolarize the afferents tonically; thus, the hyperpolarization could be mediated by the known efferent inhibition of the hair cells themselves.” If this were the case, a depolarization of the saccular fibers, produced by current injection, should have reduced the driving force of the tonic excitatory inputs and, therefore, decreased the amplitude of this tonic
833
EPSP. An inhibition of the hair cells would then further decrease, or eliminate, the diminished tonic excitatory input and would produce a smaller efferent IPSP. On the other hand, if the IPSP was mediated by a conductance change in the afferents such depolarizations should increase the amplitude of the IPSP, by increasing its driving force. The observed voltage sensitivity of efferent IPSPs evoked during subthreshold depolarizations and limited hyperpolarizations supports the latter possibility. In the example of Fig. 5, a 1.6 nA depola~~ng current pulse injected into the saccular fiber produced approximately an 8 mV depolarization and increased the IPSP amplitude from 1 mV to 1.58 mV. A hyperpolarizing current of the same intensity reduced the IPSP amplitude to 0.75mV (not shown). Similar observations were made in more than 60% of the fibers tested. We were, however, not able to invert the IPSP (i) with large hy~~la~~tions, due to the instability of the electrode resistance during the injection of large currents, or (ii) by chloride loading, presumably due to a large spatial separation between the electrode and the site of conductance change. In fact, it is not possible to clearly demonstrate that the IPSP is due to a chloride dependent conductance change, rather than to a potassium dependent one. DISCUSSION
The evidence presented in this report indicates that the activation of the M-cell can evoke an IPSP in auditory afferents and this IPSP is probably generated in the ear. This is the first intracellular demonstration of an efferent inhibitory potential on such sensory afferents in the vertebrate central nervous system, although mo~holo~cal studies have already reported the presence of presumptive inhibitory synapses on auditory and vestibular afferents.27~30~3’ Our results can be contrasted with the finding that vestibular afferents in the toadfish are instead excited by efferent stimulation, an effect which has been
> E v-
Fig. 5. Effect of shifting membrane potential on the amplitude of the efferent IPSP. Superimposed traces show the IPSP recorded in a saccular fiber at resting membrane potential, - 70 mV, and the larger response obtained during a 20 ms, 1.6 nA depolarizing current pulse. ‘Ihe electrotonic coupling potentials evoked by the M-cell antidromic impulse are indicated by the solid circles. Both traces are the averages of 16 sweeps.
834
J.-W. LAN and D. S. FABEK
linked to an arousal preceding movements associated with feeding, fighting, etcX Comparisons between the efferent IPSP and the inhibition of the hair cells” suggest that they are mediated by separate pathways. First, the inhibition of the hair cells was cholinergic and blocked by d-tubocurarine at a dose of l-2 mg/kg body weight.” In our experiments, the same dosage was routinely used to immobili~ the animal and it did not interfere with the recording of IPSPs. Second, the latency of the efTerent IPSP was shorter than that reported for the inhibition of the hair cells. Specifically, the 6 ms latency of the IPSP, measured from the artifact produced by spinal stimulation to the onset of the IPSP, includes the following components: (1) the time required for impulse conduction from the spinal cord to the medulla, at least 0.5 ms’, and (2) at least one synaptic deIay, from the M-cell to an interneuron, about 0.4 ms.” Therefore, the latency of the efferent IPSP would be no more than 5 ms if the efferent fibers were activated directly by a stimulating electrode located on the saccular nerve. In contrast, the inhibition of the hair cells had a latency of 6-7 ms when it was produced by a direct stimulation of the nerve. Third, the peak of the efferent IPSP, which coincided with the maximal inhibition of the sound-evoked EPSPs (Fig. 3B1), had a latency of about 779ms whereas the peak inhibition of the hair cells occurred 12 ms after the stimulation of the saccular nerve. Fourth, the hair cell inhibition lasts about 40ms while the duration of the efferent IPSP ranged from 9.28 to 19 ms, with one exception of 34 ms. Finally, it has been reported that inhibition of hair cells never produced any hy~rpolari~tion in the saccular fibers. This observation is consistent with our finding that the saccular fiber recordings normally showed a noise-free baseline, and did not seem to be depolarized by any tonic excitatory input, i.e. there was a low frequency of spontaneous EPSPs. Therefore, the different pharmacological properties and time courses of the two inhibitions enable us to distinguish between them. In confirmation. ultrastructural studies of the goldfish ear demonstrated that there were two types of synapses on the dendrites of the saccular fibers. one being the excitatory junctions established by the hair cells while the other type, which was observed less frequently, was suggested to be inhibitory and might mediate the inhibition reported here.27 The rare occurrence of inhibitory synapses on the auditory afferents is consistent with our experience that not every fiber impaled exhibited an efferent IPSP. The time course of the efferent inhibition can account for the presynaptic component of the third type of inhibition defined by Furukawa et af.” In that report, they found that the firing of the M-cell evoked a feedback inhibition in the lateral dendrite (dendritic inhibition), which can be differentiated from the collateral in~bition located primarily at the soma on the basis of latency, duration, sensitivity to procaine
and inhibition of the EPSPs activated by eighth nerve stimulations, Coupled to this dendritic inhibition was a reduction in the extracellular field potential recorded in the M-cell axon cap (extracellular orthodromic response; EOR) following saccular nerve stimulation. In fact, it is now known that the EOR originated from a population of inhibitory interneurons that were activated by eighth nerve via electrical synapses.35 The EOR ampIitude, which presumably reflects the number of the inhibitory interneurons activated, then provides an estimate of the number of eighth nerve fibers that were activated. The efferent IPSP and the feedback inhibition of the EOR have the same latencies, peak times and durations. Therefore, we propose that the decreased excitability of the saccular fibers is partly due to the action of the efferent pathway. However, this still does not totally account for the proposed dendritic inhibition, which may last longer than the efferent IPSP.‘~‘3 Finally, it should be noted that the dendritic inhibition and the inhibition of the EOR could be completely blocked by strychnine,” which suggests a glycinergic mechanism, although it is generally accepted that acetylcholine is the transmitter for efferent inhibitionI The firing of one M-cell typically activates a collateral inhibitory pathway which prevents further impulse activity in both M-ceils during the execution of the startle reflex.’ Our findings add a new component to this feedback network: the firing of the M-cell produces bilateral inhibition of the auditory fibers. The components of the proposed pathway postsynaptic to the M-axon are shown in the scheme of Fig. 6. First, since the efferent inhjbition is bilateral, it is likely that the circuit involves interposed relay neurons similar to or the same as those which are in the collateral inhibitory network” and excite cranial motoneurons involved in head-level components of the startle response.i8 In some experiments, when the recording quality of the saccular fiber was ideal, we noted that the amplitudes of the efferent and collateral IPSPs fluctuated in parallel (see Fig. 1C for an example). This observation suggests that both inhibitory systems may share a common pathway, and the so-called cranial relay neurons (CRNs) are the most logical candidates for this role. Specifically, it is likely that CRNs monosynaptically activate the efferent inhibitory interneurons (Fig. 6) whose relatively slow impulse conduction velocity would account for the 56ms latency.” Then, the ~varia~ons of the IPSP amplitudes would be due to the occasional failure of some CRNs (however, see Ref. 36). The interneurons mediating the efferent inhibition evoked by the M-cell have not been identified. In our previous morphological study of the saccular inputs to the M-cel12’, where horseradish peroxidase was applied to the cut end of the saccular nerve, we found a cluster of stained cell bodies located at the level of the IV ventricle and near to the M-axon (unpublished observations). Similar to some
Efferent inhibition of auditory fibers via M-cell outputs
CiiN
Fig. 6. Schematic representation of the M-cell’s feedback inhibitory network. The dotted line delineates the midline of the fish brain. M-cell send axons cross the midline and monosynaptically activate cranial relay neurons (CRNs), one of which is shown. The CRNs in turn are excitatory to a group of inhibitory interneurons (Int) which produce the collateral IPSP in the M-cell, and to cranial motoneurons, such as those in the trigeminal nucleus Q’). Efferent inhibitory intemeurons are presumably also activated by the CRNs. As the synapses interposed between these two groups of cells have not been defined, they are represented by the rectangular box. The responses are bilateral since each CRN is postsynaptic to both M-cells. The efferent intemeurons inhibit VIIIth nerve fibers which terminate on the M-cell as well as those which do not. The latter group synapse on the secondary neurons of the acousticovestibular nuclei (broken circle). Although it is likely that the efferent inhibition of the hair cells (see text) is also coupled to the M-cell feedback inhibitory pathway, this connection is not included, due to the lack of definitive evidence for it.
of the efferents of the lateral
line organs,36 these neurons sent their axons rostrally and turned laterally at the level of the M-cell before exiting from the medulla. It remains to be determined whether these efferent axons make svnaotic contact with hair cells or with saccular fibers. Finally, it also is not clear
835
what types of saccular fibers, in addition to those which give rise to large myelinated club endings, are inhibited by the M-cell network, because we found that not all the saccular fibers were inhibited. The characterization of physiological and morphological properties of the inhibited fibers may provide further insight into the specific sensory information that the M-cell network controls. The physiological significance of this efferent inhibition has been stated in general terms in the introduction. To further evaluate the specific functions of the efferent inhibition, it is necessary to compare its time course with that of the tail tlip evoked by the M-cell. The latency between the firing of the goldfish M-cell and the beginning of a tail flip is about 10-12 ms,29 and the duration of the body movement defined as Kinematic stage 1 by Webb32 was about 20 ms.4 Apparently, the efferent inhibition, with an initial latency of 6 ms and a peak latency of 8 ms, starts and reaches its peak before the beginning of the body movement. Furthermore, in a restrained fish the inhibition is terminated at about 26 ms after the firing of the M-cell, or before the first stage of the tail Ilip is completed (30-32 ms). Therefore, the main function of the efferent inhibition appears to be the termination of auditory evoked inputs before the fish moves. Presumably, the auditory inputs which initiate the M-cell firing become “irrelevant” once the escape “command” is issued by the M-cell, and the temporary silence of the saccular fibers may prepare the animal for the swimming behavior which follows the tail fli~.~ It should also be pointed out that afferent suppression cannot be accomplished as effectively by inhibition of the hair cells, if the afferents are capable of firing repetitively after they are activated.’ On the other hand, the latency and duration of the efferent inhibition of the hair cells described by Furukawa” suggest this latter process may be more appropriately timed to block the auditory noise generated by the tail Ilip itself. In addition, the inhibition of the hair cells may have an added advantage of preserving the “gain” of the hair cellafferent synapses, because this synapse habituates after repetitive stimulation.i6 Acknowledgements-We thank Dr Joseph Fetch0 for comments on the manuscript, Julie Lakatos for graphics and Jan Jordan for typing. This work was supported by NIH Grant No. NSl5335.
REFERENCES
1. Bryan J. S. and Krasne F. B. (1977) Presynaptic inhibition: the mechanism of protection from habituation of the crayfish lateral giant fibre escape response. J. Physiol., Land. 271, 369-390. 2. Dijkgraaf S. (1963) The functioning and significance of the lateral line organs. Biol. Rev. 38, 51-105. 3. E&on R. C. and Bombardieri R. A. (1978) Behavioral functions of the Mauthner neuron. In Neurobiology of the Maurhner Cell (eds Faber D. S. and Kom H.). 221-224. Raven Press. New York. 4. Eaton R. C., Bombardieri R. A. and Meyer D.‘I. (1977) The Mauthner-initiated startle response in teleost fish. J. exp. Biol. 66, 65-8 1. 5. Eaton R. C. and Hackett J. J. (1984) The role of the Mauthner cell in fast-starts involving escape in teleost fishes. In Neural Mechanisms o~Srarrfe Eehuuior (ed. Eaton R. C.), pp. 213-266. Plenum Press, New York. 6. Eccles J. C., Eccles R. M. and Magni F. (1961) Central inhibitory action attributable to presynaptic depolarization produced by muscle afferent volleys. J. Physiol., Land. 159, 147-166.
J.-W. LIN and D. S. FABER
X36
7. Faber D. S. and Korn H. (1978) Eiectrophysiology of the Mauthner cell: Basic properties, synaptic mechanisms. and associated network. In ~eu~o~~u~~g~qfzhe ~~~z~~e~ Cetl (eds Faber D. S. and Kern H.). pp. 47-132. Raven Press. New York. 8. Fay R. R., Hillery C. M. and Bolan K. (1982) Representation of sound pressure and particle motion information in the midbrain of the goldfish. Camp. Bioclrem. Ph~siol. 71A. 181-192. 9. Furshpan E. J. and F&kawa T. (i962) Intracellular and extracellular responses of the several regions of the Mauthner cell of the goldfish. J. Neurophysiol. 25, 732-771. 10. Furukawa T. (1966) Synaptic interaction at the Mauthner cell of goldfish. Prog. Brain Res. 21A, 44-70. Il. Furukawa T. (1981) Effects of efferent stimulation on the saccule of goldfish. J. Physiol., Land. 315, 203-215. 12. Furukawa T., Fukami Y. and Asada Y. (1963) A third type of inhibition in the Mauthner cell of goldfish. J. Neurophysioi. 26, 759-774. 13. Furukawa T., Fukami Y. and Asada Y. (1964) Effects of strychnine and procaine on collateral inhibition of the Mauthner cell of goldfish. Jup. J. Pbysjo~. 14, 386-399. 14. Furukawa T. and Furshpan E. J. (1963) Two inhibitory mechanisms in the Mauthner neurons of goldfish. J. Neurophysiol. 26, 140-175. 15. Furukawa T. and Ishii Y. (1967) Effects of stat+ binding of sensory hairs on sound reception in the goldfish. sup. J. Physiol. 17, 572-588. 16. Furukawa T. and Matsura S. (1978) Adaptive rundown of excitatory post-synaptic potentials at synapses between hair cells and eighth nerve fibres in the goldfish. J. Physiol., Land. 276, 193-209.
17. Guth P. S., Norris C. H. and Bobbin R. P. (1976) The pharmacology of transmission in the peripheral auditory system. Pharmac. Ret~. 28, 95-125. 18. Hackett J. T. and Faber D. S. (1983) Mauthner
axon networks mediating supraspinai components of the startle response in the goldfish. N~ro~~~~ce &, 3 17-33 1. 19. Hackett J. T. and Faber D. S. (1983) Relay neurons mediate collateral inhibition of the goId&h Mautlmer cell. Brain Res. 264, 302-306. 20. Highstein S. M. and Baker R. (1985) Action of the efferent vestibular system on primary agerents in the toadfish, Opsanus tau. J. Neurophysiol. 54, 370-384. 21. Kom H. and Faber D. S. (1975) An electrically mediated inhibition in gold&h medulla. J. Newophysioi. 38.452-471. 22. Lin J.-W. (1986) Physiology and Morphology of Identified Mixed Excitatory Synapses on the Goldfiab Mauthner Cell.
Ph.D. thesis, SUNY at Bufialo. Buffalo, NY. 23. Lin J.-W. and Faber D. S. (In press) Synaptic transmission mediated by single club endings on the goldfish Mauthner cell. I. Characteristics of electrotonic and chemical postsynaptic potedais. J. Neurosci. In press. 24. Lin J.-W., Faber D. S. and Wood M. R. (1983) Organized projection of the goidftsfi saccnlar nerve onto the Mauthner cell lateral dendrite. Brain Res. 274, 3 19-324. 25. Nakajima Y. (1974) Fine structure of the synaptic endings on the Mauthner cell of the goldfish. J. camp. Newof. ISa, 35742. 26. Nakajima Y. and Kohno K. (1978) Fine structure of the Mauthner cell: Synaptic topography and comparative study. In Neurobiology of the Muutbner Cell (eds Fabr D. S. and Kom H.), pp. 133-166. Raven Press, New York. 27. Nakajima Y. and Wang D. W. (1974) Morphology of afferent and efferent synapses in the hearing organ of the golclftsh. J. camp. Neural. 156, 403-416. 28. Russell I. J. (1974) Central and peripheral inhibition of lateral line input during the startle response in goldfish. Bruin Res. So, 517-522. 29. Russell I. J. (1976) Central in~bitjon of lateral line input in the medulla of the goldfish by neurones which control active body movements. J. camp. Physiof. 111, 335-358. 30. Spoendlin H. (1973) The innervation of the cochlear receptor. In Basic Mechanisms in Hear&g (ed. Msller A. R.), pp. 185-234. Academic Press, New York. 31. Takasaka T. and Smith C. A. (1971) The structure and innervation of the .&eon’s basilar _ Dapilla. J. ultrastruct. Rex _ _ .., 35, 2[M5. 32. Webb P. W. (1976) The effect of size on the fast-start performance of rainbow trout Salmo gairdneri, and a consideration of piscivorous predator-prey interactions. J. exp. Biol. 65, 157-177. 33. Weber E. H. (1820) De sure et auditi hominis et animals, Pars I. De aure anbnaltun aquutikm, Lipsiae. 34. Yasargil G. M. and Diamond J. (1968) Startle-response in teleost fish: an elementary circuit for neural discrimination. I
Nature, Land. 226, 241-243. 35. Zottoli S. J. and Faber D. S. (1980) An identifiable class of s~toacoustic
in~~eurons with bilateral projections in the goldfish medulla. Neuroscience 5, 1287-I 302. 36. Zottoli S. J. and van Home C. (1983) Posterior lateral line tierent and efferent pathways within the central nervous system of the goldfish with special reference to the Mauthner cell. J. camp. Neurol. 219, 10&l 11. (Accepted 20 August 1987)