Signal processing in brainstem auditory neurons which receive giant endings (calyces of Held) in the medial nucleus of the trapezoid body of the cat

321

Hearing Research, 49 (1990) 321-334 Elsevier

HEARES

01427

Signal processing in brainstem auditory neurons which receive giant endings (calyces of Held) in the medial nucleus of the trapezoid body of the cat John J. Guinan

Jr. and Robert Y.-S. Li

Depariment of Electrical Engineering and Computer Science, and Research Loboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts, U.S.A. and Eaton-Peabody Laboratov of Auditory Physiology and Department of Otoiaryngologv, Massachusetts Eye and Ear Infirmary, Boston, Massachusetts, U.S.A. (Received 17 July 1989; accepted 17 October 1989)

In the medial nucleus of the trapezoid body (MNTB), each principal neuron receives one large axonal ending (a calyx of Held) and many small endings. In this same region, microelectrode recordings show unusual ‘unit’ waveforms which have two components separated by about 0.5 ms. We show that the first component (C,) of such a waveform corresponds to a spike from the calyx of Held and that the second component (C,) corresponds to a spike from the MNTB principal neuron. There are two kinds of evidence for these correspondences. First, electrical stimulation of calyciferous axons in the contralateral trapezoid body evokes C, spikes with latencies of 0.1-0.3 ms. These latencies are too short for there to be an intervening synapse and are consistent with C, being a presynaptic spike. Second, shocks in the lateral superior olive (which receives projections from MNTB principal-neurons) evoke ‘A spikes in the MNTB which can be shown by their waveshapes and mutual refractoriness with C, spikes to result from antidromic activation of the neurons producing C, spikes. Spontaneous and sound-evoked responses in dozens of cats anesthetized by barbiturates or Ketamine always had a C, spike following each C, spike. This implies that there is normally one-to-one spike transmission from the calyx of Held input to the MNTB principal neuron output. The small endings on MNTB principal neurons are also capable of evoking spikes. Electric shocks (and in one case, sound), evoked long latency (l-3 ms) ‘LC,’ spikes, which (by mutual refractoriness and waveshape) are from the same neural elements as C, and ‘A’ spikes. Since LC, spikes are not preceded by C, spikes, LC, spikes must be mediated by small axonal endings on MNTB principal neurons. We found some evidence of inhibition, possibly recurrent inhibition, in MNTB principal neurons. In a few neurons, sound or shocks inhibited ‘A’ spikes or LC, spikes. In some cases, after each C, spike, LC, spikes were blocked or reduced in amplitude for several milliseconds. Our data firmly establish that there is fast, secure spike transmission from calyces of Held to MNTB principal neurons and suggest that under some circumstances there is additional signal processing in MNTB principal neurons. Auditory System: Synapses; Neural Signal Processing

Introduction In the medial nucleus of the trapezoid body (MNTB), there are synaptic endings called calyces of Held which are among the largest in the mammalian central nervous system (Fig. 1, left). A single calyx envelopes a large fraction of the soma

Correspondence to: John J. Guinan Jr, Eaton Peabody Laboratory of Auditory Physiology, Massachusetts Eye and Ear Infirmary, 243 Charles Street, Boston MA 02114, U.S.A. 0378-5955/90/$03.50

of a single MNTB principal neuron, forming synapses which have ultrastructural features typical of chemically mediated synapses (Ramon y Cajal, 1909; Lenn and Reese, 1966; Morest, 1968, 1973). Microelectrode recordings in the MNTB reveal many ‘units’ with unusual waveforms consisting of two distinct components, C, and C,, separated by about 0.5 ms (Fig. 1, lower right). Such ‘compound waveforms’ have been recorded in the anteroventral cochlear nucleus (AVCN) (Pfeiffer, 1966; Bourk, 1976) in the MNTB (Guinan et al., 1972a,b), and in the ventral portion of the ventral nucleus of the lateral lemniscus (Adams, 1978). In

0 1990 Elsevier Science Publishers B.V. (Biomedical

Division)

322

CALYX OF HELD

3

1

:

?$!! MNTB / PRINCIPAL NEURON _# -7

‘y

COMPOT

Tl (Cil)

x

EA

COMPONENT 2 (C2)

Fig. 1. (Left) An artist’s rendition of a calyx of Held ending on an MNTB principal neuron. Each principal neuron also receives many small endings such as the two shown on the upper right dendrite. The dashed cell outline at right is a nearby MNTB principal neuron which receives endings from a collateral of the calyx. (Adapted from Morest et al., 1972). (Lower right) A typical compound waveform showing C, and Cz components. Unit 129-18.

all of these regions, there are neurons which receive very large presynaptic endings. The correlation between the presence of large endings and the recording of compound waveforms suggests that the large endings play an important role in producing compound waveforms. For compound waveforms recorded in the AVCN, Pfeiffer (1966) proposed that C, is due to the firing of a presynaptic element and C, is due to the firing of a postsynaptic element. As pointed out by Pfeiffer, C, by itself resembles an extracellular recording of a postsynaptic spike. Pfeiffer also found that C, could be further decomposed into two components, C,, and C,,, and that in 20-258 of cochlear nucleus neurons with compound waveforms, C,, occasionally fails to develop. He suggested that C,, might be the discharge of the initial segment or axon hillock of the neuron, and that C,, might be the discharge of the soma-dendritic process, both of which fit the interpretation that C, is postsynaptic. Because C, precedes C, by a time appropriate for a synaptic delay, Pfeiffer assumed that C, represented a presynaptic spike.

We have tested Pfeiffer’s prespike + postspike hypothesis for the ‘compound waveforms’ of the MNTB. With this hypothesis, C: is due to a calyx of Held, or some nearby part of the presynaptic axon, and C, is due to the MNTB principal neuron contacted by that calyx of Held. The prespike + postspike hypothesis is interesting in that it implies that there is one-for-one spike transmission by these neurons, and that both presynaptic and postsynaptic events can be monitored by a single extracellular electrode. If the prespike + postspike hypothesis is true, such recordings might be useful for determining the circumstances under which there are deviations from one-for-one transmission and for determining the effects of drugs or other manipulations on transmission at this synapse. In addition to a single calyx, each MNTB principal neuron receives many small endings both on its soma and on its dendrites (Mores& 1968). Synaptic contacts made by these smaller endings are presumed to be both excitatory and inhibitory (i.e., some have round vesicles and asymmetric pre- versus post-synaptic densities, and others have flattened or pleomorphic vesicles and symmetric densities) (Lenn and Reese, 1966; Nakajima, 1971; Morest, 1973; Jean-Baptiste and Morest, 1975). A second goal of the work presented here was to determine the effects of the smaller synapses on signal processing in MNTB principal neurons. * Methods The surgery, acoustic systems, and recording methods were the same as in Guinan et al. (1972a). Briefly, adult cats were anesthetized with Dial in urethane (100 mg/ml diallyl barbiturate, 400 mg/ml monethyurea and 400 mg/ml urethane) or with sodium pentobarbital, and the bulla and middle-ear cavities were opened on both sides. The brainstem was approached ventrally through the basi-occipital bone. Cats were placed in a soundproof chamber, and sounds were delivered by closed acoustic assemblies similar to those described by Kiang et al. (1965). We recorded unit

* Preliminary reports of this work have been presented previously (Li and Guinan, 1971; Guinan, 1979).

323

activity with indium-filled glass pipette electrodes with platinum tips (3-10 pm diameters) or, in a few cases, with pipette electrodes filled with normal saline (tips 3 to 10 pm). Electric shocks (monophasic pulses I 0.1 ms duration) were delivered through a variety of bipolar electrodes. Usually, we used concentric electrodes with outer diameters of 400-700 pm and the last few hundred pm uninsulated. These had inner conductors which tapered to a point and extended 500-1000 pm past the outer conductor with the last half uninsulated. We also used electrodes consisting of two parallel wires with uninsulated tips 200-500 pm in length separated by 300-1000 pm. Contamination of all-or-none ‘spike’ responses by stimulus artifacts and evoked responses was minimized by the ‘subtract-average’ technique. With this technique, an average ‘artifact + evoked

A

c\ II +

-

1 ms

100 pv

[

response’ was subtracted from the records containing spikes (e.g., Fig. 2). This technique requires that the spikes be clearly discernible in each trace. Near-threshold shocks were used, so that there would be some responses in which there were no spikes. This procedure worked well (i.e., it produced traces with relatively flat baselines) when the evoked response was relatively constant from trial to trial. The technique has been applied to the data in Figs. 4, 8, and 9. Results ‘Units’* with ‘compound waveforms’ were recorded in the MNTB in the region which contains calyces of Held. We define ‘compound waveforms’ as all-or-none waveforms with two distinct comby 0.3-0.6 ms. ponents, C, and C,, separated Usually, C, is positive and C, is negative, but either component can be diphasic (positive-negative). A typical penetration through the MNTB by an electrode with a platinum-ball tip yielded several ‘units’ with compound waveforms. Similar waveforms, but with poorer signal-to-noise ratios, were obtained with electrolyte-filled pipettes. We tested the prespike + postspike hypothesis for compound waveforms in two steps. If C, represents a spike from a calyx of Held, or some nearby part of the presynaptic axon, then C, should be evoked with a short latency by shocks which excite the presynaptic axon. Accordingly, the first test was to determine whether C, spikes with appropriate properties were evoked by shocks to regions containing calyciferous axons. Similarly, if C, represents a postsynaptic spike in an MNTB principal neuron, it should be possible to activate the MNTB principal neuron antidromitally and to record C, spikes alone. The second test was to determine whether appropriate C, spikes were evoked by shocks to regions containing axons of MNTB principal neurons. In addition to these tests, various anatomical data are

SHOCK TO LATERAL LEMNISCUS Fig. 2. An example of the reduction of baseline stimulus artifact and evoked response by the ‘subtract-average’ technique. (A) 10 superimposed responses evoked by shocks to the ipsilateral lateral lemniscus. (B) An average of 8 responses which did not contain spikes. (C) The waveforms in (A) minus the waveform in (B). Unit 129-18.

Since these waveforms are due to two neurons which produce all-or-none responses, each waveform might be considered to be a response of two units. Nonetheless, we will refer to the entities which produce these waveforms as ‘units’ because in most circumstances they behave as single units.

324

considered to establish that the presynaptic and postsynaptic elements which generate the C, and C, spikes are calyces of Held and MNTB principal neurons, respectively. C, is a presynaptic spike from a calyx of Held We tested the hypothesis that C, is a spike from a calyx or nearby calyciferous axon by determining whether C, could be directly excited by shocks to the presynaptic axon. It is well established (e.g., Mores& 1968; Warr, 1982; Tolbert et al., 1982; Glendenning et al., 1985) that MNTB calyces of Held arise from large axons (ranging up to 10 pm) which cross the midline in the trapezoid body and originate from globular bushy cells in the contralateral cochlear nucleus (see Fig. 3). We delivered electric shocks to the contralateral trapezoid body (CTB) in the region traversed by the calyciferous axons (Fig. 3). With sharp thresholds, these CTB shocks evoked all-or-none waveforms with C, and C, components similar to those evoked by sound. The latency of C, (measured from the beginning of the shock to the foot of C,) was typically 0.1 to 0.3 ms (Fig. 4, left). This latency is too short to include a synaptic delay. C, followed the shocks one-for-one up to very high shock rates (500/s or more). The sharp

LATERAL LEMNISCUS

threshold, short latency, and high following rates demonstrate that C, represents a spike in an axon which was electrically stimulated without a synapse intervening between the stimulation and recording points. Since C, occurs with normal sound stimulation it must be an orthodromic response, as opposed to an antidromic response. That C, is a direct axonal response which is orthodromic, supports the interpretation that C, is a spike in a presynaptic element in the MNTB. We also found that compound waveforms were evoked by electrical shocks to the ipsilateral lateral lemniscus. These responses had sharp thresholds, relatively fixed latencies (see Fig. 2) and followed high shock rates. Again, such response properties are consistent with C, representing a spike in an axon which was electrically stimulated without an intervening synapse. We did not find evoked compound waveforms from shocks at other stimulating electrode locations (the lateral superior olivary nucleus (LSO) and the dorsomedial periolivary nucleus (DMPO)). The locations at which we were able electrically to evoke C,s without intervening synapses are consistent with the presynaptic elements being calyces of Held. The large calyciferous axons course through the contralateral trapezoid body;

I COCHLEAR NUCLEUS

\

DENDRITE OF MNTB PRINCIPAL NFIIRON

HYPOTHETICAL INTERNEURON

ELECTRODE

ELECTRODE RECORDING ELECTRODE

CONTRALATERAL TRAPEZOID ELECTRODE

BODY

Fig. 3. A schematic showing electrode positions in relation to input and output pathways of an MNTB principal neuron. calyciferous axons and MNTB principal neurons have additional collaterals which are not shown here.

Some

325

“A”SPIKE

[

1 ms

100 uv

1

t

t

CTB SHOCK

LSO SHOCK

Fig. 4. Examples of compound waveforms (C, +C,) and ‘A spikes from three different units. In each row, the spikes at left and right were recorded at the same location. Each panel shows many superimposed traces. In a few traces, the shocks failed to evoke a spike response. (Left) ‘Compound Waveforms’ consisting of C, +C, components evoked by shocks to the contralateral trapezoid body (CTB). The middle C, spikes and the bottom C, spikes are diphasic. (Right) ‘A’ spikes evoked by shocks in or near the lateral superior olivary nucleus (LSO) The middle ‘A’ spikes are diphasic. Top to bottom: Units 128-14,128-10,25-12.

some provide collaterals to the DMPO and some provide collaterals which ascend in the lateral lemniscus (e.g., Morest, 1968; Smith et al., 1987; Friauf and Ostwald, 1988). Presumably, C,s evoked by lateral lemniscus shocks are due to action potentials first traveling antidromically in the lateral-lemniscus collateral and then orthodromically in the branch to the calyx. The 0.8 ms latency typically found for C,s evoked by lateral lemniscus shocks are consistent with a 10 mm

distance and an average axon size of about 3 pm. That no compound waveforms were found with DMPO stimulation may be because we used DMPO stimulation only a few times and/or because the collaterals to the DMPO are so small that an antidromic spike in the collateral may not invade the main calyciferous axon. In summary, the observations on shock-evoked C,s are consistent with the prespike + postspike hypothesis and with the C,s being produced by calyces of Held. C, is a postsynaptic spike from a neuron which receives a calyx of Held We tested the hypothesis that C, represents a postsynaptic spike from a neuron which receives a calyx of Held, i.e., a MNTB principal neuron, by determining whether C, could be evoked by shocks which were expected to activate MNTB principal neurons antidromically. With orthodromic excitation, C, was always followed by C,, but we would not expect the reverse to be true for antidromic activation. Antidromic activation should evoke a spike resembling C, alone. To activate MNTB principal neurons antidromically, we delivered electric shocks in or near the LSO (see Fig. 3) or the DMPO. Both of these receive axonal projections from MNTB principal neurons (Morest, 1968; Spangler et al., 1985; Glendenning et al., 1985; Zook and DiCaprio, 1988). LSO or DMPO shocks evoked ‘A’ spikes in the MNTB (Fig. 4, right). These ‘A’ spikes had short latencies (usually 0.1 to 0.3 ms), sharp thresholds and followed high (several hundred per set) shock rates. Thus, ‘A’ spikes represent spikes in axons which were electrically stimulated without a synapse intervening between the stimulation and recording points. It seems reasonable that an ‘A’ spike might represent antidromic activation of the same neural element which produces a C, spike. ‘A’ spikes were always found at a location where there was a compound-waveform unit with C, spikes similar in shape to the ‘A’ spikes (Fig. 4). However, ‘A spikes sometimes had higher amplitudes than C, spikes. To test whether or not ‘A’ spikes and ‘C2’ spikes were from the same neural elements, we determined whether or not these spikes were mut-

326

ually refractory (when one spike occurs, the other cannot occur at the same time or for a short time). This was done by comparing responses from paired stimuli which separately evoked ‘A’ spikes or ‘Cz’ spikes. Usually, the responses were evoked by shocks to orthodromic and antidromic stimulating electrodes but sometimes orthodromic spikes were provided by spontaneous activity * _ Two examples which show that ‘A’ spikes never occurred immediately after C, spikes are shown in Fig. 5. Usually the interval over which C, spikes blocked ‘A’ spikes was about 1 ms. Presumably, the blocking of ‘A’ spikes by C, spikes is due to refractoriness of the axon hillock and/or soma membrane which produces the spikes and also to collisions of spikes traveling in principal-neuron axons. Tests similar to those in Fig. 5 which showed clear evidence for refractoriness (or spike collisions) were obtained on over 25 units. The reverse, the blocking of C, spikes by previously occurring ‘A spikes, was more difficult to demonstrate, because, with the required timing, the C, spikes, the ‘A spikes, and the shock artifacts tended to coincide and obscure each other. ‘A’ spikes only blocked C, spikes for about 0.5-0.7 ms following the ‘A’ spike. An example which shows that C, spikes never occurred immediately after ‘A’ spikes is shown in Fig. 6. On over 6 units, spike signal/ noise ratios were adequate so that we obtained unequivocal tests showing the blocking of C, spikes by ‘A’ spikes. Together, the two kinds of tests leave little doubt that C, spikes and ‘A’ spikes both represent activity from the same neural elements. The above data support the conclusion that the C, and C, components of compound waveforms are due to presynaptic and postsynaptic spikes respectively. Both C, and ‘A’ spikes can be evoked directly (i.e., without intervening synapses) from shocks at distant electrodes, and ‘A’ spikes are from the same neural element as C, spikes. C, follows C, by a delay which is appropriate for a synaptic delay; however, no other spike precedes or follows an ‘A’ spike. All of these are consistent with the interpretation that C, is a spike in an * Sound-evoked spikes were difficult to use because, at a fixed sound level, the spike latencies and sound-evoked gross potentials were very variable.

Cl c2

I

“A” SPIKES

“A” SPIKES

I

Jr I-

Y-Z-

4

4

4

CTB SHOCK

LSO SHOCK

LSO SHOCK

Fig. 5. Two demonstrations that ‘A’ spikes are refractory after C, spikes. (Left) CTB shocks followed 1.2 ms later by shocks delivered near the LSO. (Left top) 22 responses superimposed. (Left bottom) Four individual responses. The CTB shocks evoked C, +C, spikes in 290 out of 400 trials examined. The LSO shocks evoked ‘A’ spikes in 108 out of 110 trials in which there was no C, +C, response but did not evoke an ‘A’ spike in any of the 290 trials in which there were C, +C, spikes. Unit 128-14. (Right) Shocks near the LSO which normally evoked an ‘A’ spike every time but did not when immediately preceded by the C, component of spontaneously occurring C, + C, spikes. (Right top) 20 superimposed responses. (Right bottom) Four individual responses. Abbreviations as in Fig. 4. Unit 128-9.

element which is presynaptic to the element which produces C, . The data are also consistent with the hypothesis that the postsynaptic elements which produce C, and ‘A’ spikes are MNTB principal neurons. We were most successful in evoking ‘A’ spikes with electrodes in the LSO, a nucleus which receives a

321

Cl k

“A”SPIKES rc2

C2V ’

<

1 ms

C2b

B

CTB SHOCK Fig. 7. Examples of failure of all (A) or part (B) of the second component of compound waveforms. (A) Two superimposed traces obtained near the beginning and after several hundred ms of the response to 667 shocks/s. Unit 17-8. (B) Six superimposed traces obtained from the response to 500 shocks/s. In this ‘unit’, the second component consists of two subcomponents, C,, and C,,. There are four compound-waveform responses, one of which shows a failure of C,,. The C, components may be large and positive because the electrode is very close to the soma membrane providing spikes with the same time course as the membrane current (see Freygang and Frank, 1959). Unit 126-6.

P-

CTB SHOCK

LSO SHOCK

Fig. 6. Demonstration that C, spikes are refractory after ‘A spikes. LSO shocks paired with CTB shocks. (Top) 20 superimposed responses. (Bottom) Four individual responses. The LSO shocks evoked ‘A’ spikes in approximately half of the trials. The CTB shocks were well above threshold and evoked C, + C, spikes every time in the absence of LSO shocks (C, is difficult to see because it is brief, has a very short latency and is partly obscured by the onset of ‘A’ spikes; C, +C+ spikes from this unit evoked at a lower shock level are shown in Fig. 5, left). In the responses to over 400 shock pairs, an ‘A’ spike or a C, spike was seen on every trial, but in no case was there both an ‘A’ spike and a C, spike. Abbreviations as in Fig. 4. Unit 128-14.

massive innervation from MNTB principal neurons (Rasmussen, 1967; Spangler et al., 1985; Glendenning et al., 1985; Zook and DiCaprio, 1988). We also evoked ‘A’ spikes with an electrode in the DMPO, which receives axons from MNTB principal neurons and also is traversed by some principal-neuron axons on their way to the LSO (Mores& 1968; Spangler et al., 1985; Glendenning et al., 1985; Zook and DiCaprio, 1988). Finally, the principal neuron is the only cell type in the MNTB which receives a large presynaptic ending (Morest, 1968). The C, component consists of separable C,, and C,, components With CTB shocks at very high rates (2 500/s), C, sometimes failed totally (Fig. 7A), or partially (Fig. 7B), in an all-or-none fashion. The partial

328

failures reveal that, at least under some conditions, C, consists of two separate components, C,, and C,, (Fig. 7, bottom). It is not clear, however, whether our C,, and C,, components correspond to the two subcomponents of C, found by Pfeiffer (1966). With excitation by high-level tones, Pfeiffer (1966) saw failures of a component similar to C,, on 20-25% of units with compound waveforms in the AVCN. We have made comparable observations from our recordings of compound waveforms in AVCN. However, for compound wave-

forms in the MNTB, we have seldom, if ever, seen failures of C,, with tones at any level. Excitatory effects of non-calycine inputs on MNTB principal neurons In some MNTB neurons, we found spike responses which appear to be due to activation of principal neurons by the small, non-calycine, inputs. These ‘LC,’ spikes resembled C, and ‘A’ spike but had considerably longer latencies and were not immediately preceded (or followed) by C, spikes (Fig. 8: A,D). LC, spikes were evoked

4 CTB SHOCK

&TB SHOCK

CTB SHOCK

CTB SHOCK

SPONTANEOUS

Fig. 8. Examples of long latency (LC,) and compound waveform (C, +C,) spikes. Each row shows data from a single electrode location. Abbreviations as in Fig. 4. (Top) Data from a run in which bursts of CTB shocks (3 ms between shocks, 16 shocks per burst, 10 bursts per s) were presented at the threshold for evoking C, + C, spikes. (A) four traces with only LC, spikes (each of these was the first spike evoked by a shock burst). (B) four traces with both C, + C, and LC, spikes (none of these was the first spike evoked by a shock burst). (C) a C, + C, response which was the first spike response evoked by a shock burst. Note that C, and LC, spikes which followed within a few ms of other C, or LC, spikes (panel B) were consistently smaller than C, and LC, spikes which did not immediately follow other C, or LC, spikes (panels A, C). Unit 128-10. (Bottom) (D) LC, spikes evoked by CTB shocks (an early part of the trace was very variable and has been blanked out). As the CTB shock level was increased, LC, spikes first appeared with - 3.5 ms latency, jumped to - 2 ms latency, went back to the longer latency, and disappeared at the highest shock level. Shown here is an intermediate level of CTB shocks which evoked LC, spikes in both time ranges. (E) Spontaneously occurring C, +C, spikes. Unit 131-16.

329

dromic, synaptically evoked spikes in MNTB principal neurons, but since each LC, was not preceded by a C, spike, they are not evoked by the CalyX.

In two cases LC, spikes appeared to be evoked or influenced by sound. In one case, LC,-like spikes were evoked by tone bursts (1.15 kHz, 25 ms duration, 6/s) *. In another case, LC,-like spikes were evoked by shocks paired with a variety of sounds, but not by these stimuli presented alone. We call these ‘LC,-like’ spikes because their waveshapes were almost identical to the C, spikes of a compound-waveform unit recorded at the same location, but we did not prove (or test) that the LC,-like and C, spikes were from the same unit by a demonstration of mutual refractoriness * *.

LATERAL

LEMNISCUS

SHOCK

CTB SHOCK Fig. 9. Demonstration that LC, spikes are refractory after C, spikes. CTB shocks which evoked LC, spikes, paired with shocks to the lateral lemniscus which evoked C, +C, spikes. (Top) 20 superimposed responses. (Bottom) 3 individual responses. C, +C, spikes were evoked in 195 out of 672 trials. LC, spikes were evoked in 74 out of the 477 trials in which there was no C, +C, spike but never in trials in which there was a C, +C, spike. If C, +C, spikes and LC, spikes were independent, following the 195 C, +C, spikes there would be about 30 LC, spikes. Unit 129-18.

by short bursts of CTB shocks but not usually by the first shock of a burst or by a single CTB shock. From the similarity of LC, and C, waveforms, and the refractoriness of LC, spikes following C, spikes (Fig. 9), we conclude that LC, spikes are due to the same neural elements as C, spikes. LC, spikes resembled C, and ‘A’ spikes, but they had long (several ms), variable latencies, indistinct thresholds and they never followed every shock burst, even for shock levels several times threshold. These properties suggest that one or more synapses are interposed between the axons stimulated and the elements producing LC, spikes. With this interpretation, LC, spikes are ortho-

Inhibitory effects of non-calycine inputs on MNTB principal neurons We searched extensively but did not find inhibition which blocked C, following C,. Compound waveforms were recorded in dozens of cats anesthetized with barbiturates (Dialurethane or sodium pentobarbital) and in two anesthetized by Ketamine. We looked at spontaneous activity and responses to sounds including monaural or binaural clicks, noise bursts, tone bursts and two-tone complexes. We did not find any units which normally had compound waveforms and which sometimes fired with C, alone (excluding units in which C, became permanently absent following injury by the electrode (see Pfeiffer, 1966)). Except during the presentation of shocks at high rates, under the conditions of our experiments each input spike from the calyx of Held (C,) was always followed

* This unit was unique among compound waveform units in that sound did not evoke compound waveform spikes. In addition, CTB shocks evoked compound waveform spikes with an unusually low threshold, and, at threshold, there was almost no gross evoked potential at the recording electrode (i.e., few other axons were excited). These indicate that the CTB stimulating electrode was very close to the axon stimulated. We hypothesize that the CTB electrode damaged the calyciferous axon and prevented the normal activation of compound waveforms by sound. * * These were excluded from consideration in Guinan (1979) because they were not proven LC, responses.

330

Discussion -40 -

Q iii -60 3 -I

s

-80

I

o---

n

Excitatorv Inhibitory

1

-100

! 8 “,“‘I 0.2 SOUND

1.0 FREQUENCY

I 10 (kHz)

Fig. 10. Tuning curves for excitation of compound waveforms (circles and solid lines) and for inhibition of ‘A’ spikes (boxes and dashed lines) on unit 25-3.

by one output spike (C,) from the principal neuron. In one unit, we found sound-evoked and shock-evoked inhibition of ‘A’ spikes but not of C, spikes. For this unit, either CTB shocks or contralateral sound bursts blocked ‘A’ spikes, which were normally evoked by shocks to the LSO (raising the level of the LSO shocks did not remove the block). This block was apparent 5-15 ms after a tone burst. The tuning curve for blocking ‘A’ spikes after each tone burst was very similar to the tuning curve for evoked compoundwaveform responses during the tone bursts (Fig. 10). In several cases we found what appeared to be inhibition of LC, spikes. In a few cases, shocks to one CTB electrode inhibited LC,-like spikes evoked by a second CTB electrode. In one of these units, each time a C, + C, spike was present, the LC,-like spike was absent. In another case, CTB shocks evoked both C, + C, spikes and LC, spikes, but the LC, spikes which followed C, + C, spikes were smaller than LC, spikes which did not follow C, + C, spikes (Fig. 8: A,B).

Validity of the prespike + postspike hypothesis Our evidence strongly supports Pfeiffer’s (1966) prespike + postspike hypothesis applied to neurons in the MNTB. The sharp thresholds, short latencies and high following rates of C, and ‘A’ spikes, and the mutual refractoriness of C, and ‘A’ spikes, provide a persuasive demonstration that the C, and C, components of compound waveforms are due to extracellularly recorded presynaptic and postsynaptic spikes of a neuron in the MNTB. Furthermore, a large body of data (reviewed below) are consistent with the hypothesis that prespikes and postspikes are due to calyces of Held and MNTB principal neurons, respectively. Individual presynaptic spikes have been detected with an extracellular electrode in several places outside of the mammalian central nervous system (see Eccles, 1964). Of particular interest are extracellular recordings at the squid giant synapse, where waveforms very similar to our compound waveforms have been recorded. At this synapse, intracellular recordings from both presynaptic and postsynaptic cells have confirmed the identification of the initial, positive component as a spike in the presynaptic axon and the following, negative component as a spike in the postsynaptic axon (Takeuchi and Takeuchi, 1962). Although we interpret C, and ‘A’ spikes as being produced by the same neural elements, for some neurons, the C, spikes were smaller than the ‘A’ spikes. One difference between C, and ‘A’ spikes is that C, spikes are evoked by EPSPs but ‘A’ spikes are not. The increased conductance during an EPSP might shunt some of the spike current, thereby decreasing the spike amplitude. The association of compound waveforms with calyces of Held on MNTB principal neurons Three main pieces of evidence lead us to associate each MNTB compound waveform with a spike in a calyx of Held followed by a spike in the contacted MNTB principal neuron: (1) the strong evidence for the validity of the prespike + postspike hypothesis for compound waveforms, (2) the correlation between the locations at which compound waveforms are recorded and the loca-

331

tions of principal neurons receiving calyces of Held (Guinan et al., 1972b), and (3) the fact that the calyx of Held is the only large presynaptic element in the MNTB and therefore the only good candidate for being able to produce a presynaptic spike which could be recorded extracellularly. The following other evidence is also consistent with this interpretation. (1) Compound waveform units in the MNTB respond to contralateral sound and calyces of Held come from axons which originate in the contralateral cochlear nucleus. (2) The stimulus-electrode locations at which C, and ‘A’ spikes were evoked are consistent with the spikes being due to calyces of Held and principal neurons, respectively. (3) The tonotopic map of complex waveform units fits well with the projection pattern of MNTB principal neurons to the LSO, the tonotopic map within the LSO, and the fact that the inhibitory characteristic frequencies (CFs) of LSO neurons are almost identical to their excitatory CFs (Boudreau and Tsuchitani, 1968; Guinan et al., 1972b; Spangler et al., 1985). Since C, is a recording from a calyx of Held and calyces of Held originate from globular bushy cells, recordings of C, are from globular bushy cells. Identified bushy cells in the posterior division of AVCN (the region which contains globular bushy cells) had post-stimulus-time histograms which were primarylike, primarylike with notch (pri-notch), onset type L, chopper, or phase-locked (Rhode et al., 1983; Rouiller and Ryugo, 1984; Smith and Rhode, 1987). With the exception of the one chopper cell (Rouiller and Ryugo, 1984) these results are in good agreement with the firing patterns of compound waveform units in the MNTB (from Guinan et al., 1972a: 13 primarylike, 26 class B [in present terminology: pri-notch, and any onset type L which has a notch], and 9 phase-locked).

Effects of the giant endings on MNTB principal neurons The prespike + postspike hypothesis and our observation that C, always followed C, (except with high-rate shocks) imply that there is normally one-for-one spike transmission from the calyx of Held to the MNTB principal neuron, at least for our experimental conditions. This one-for-one transmission is consistent with the large number

of synaptic contacts between each calyx of Held and MNTB principal neuron, and the synaptic morphology which is characteristic of excitatory synapses (Lenn and Reese, 1966; Mores& 1973; Jean-Baptiste and Mores& 1975). The anatomical similarity of calyces of Held on MNTB principal neurons with endbulbs of Held on spherical cells in AVCN and with the large endings on neurons in the ventral part of the VNLL suggest that in all three regions the compound waveforms are generated by similar mechanisms (Pfeiffer, 1966; Adams, 1978). Several differences between our results from the MNTB and Pfeiffer’s (1966) from the AVCN may be due to the giant endings in the MNTB being larger than the giant endings in the AVCN. One difference is that with sound stimulation, Pfeiffer found selective failure of a component of AVCN neurons, similar to our C,, in 20-25s but we never found such failures in MNTB neurons. We presume that the giant endings in the MNTB evoke larger EPSPs than the giant endings in the AVCN, thereby making spike transmission more secure. Another difference is that in the AVCN a sequence of spike shapes thought to represent cell injury by the electrode was found only for the C, component (Pfeiffer, 1966), whereas in the MNTB both components showed this evidence of cell injury (although less often for the C, component). We presume that this difference is due to the larger size of the giant endings in the MNTB as compared to the AVCN, but other factors may also be important such as the degree to which the endings are invaded by spikes, the size of the electrodes, and the electrode position relative to the endings. Excitatory effects of small endings on MNTB principal neurons The LC, spikes demonstrate that MNTB principal neurons can be excited by endings other than calyces of Held. Although we found some cases in which LC,-like spikes were evoked by sound, or sound + shocks, we have too few data to say under what conditions LC, spikes are evoked in a normal animal. It is interesting to find, in a neuron in which there is one-for-one transmission from a giant ending, that other endings can also excite spikes.

332

There are several candidates for the presynaptic axons which mediate LC, spikes. Since LC, spikes are evoked by shocks to the contralateral trapezoid body, axons which cross in the trapezoid body are likely to be involved. Morest (1968) described two main classes of axons which cross in the trapezoid body and provide small endings on MNTB principal neurons: calyciferous axons and axons from medium-sized fibers on their way to the MSO. It is likely that endings from both of these fiber classes are excitatory since both provide excitatory synapses elsewhere (at the calyx or at branches to the MS0 and LSO, respectively). In a few cases, the threshold for evoking LC, spikes with CTB shocks was comparable to the threshold for evoking a gross response at the MNTB recording electrode. In these very low threshold cases, the large fibers may be the only ones stimulated (the largest fibers have the lowest thresholds); this suggests that synapses from collaterals of calyciferous axons are able to evoke LC, spikes. It is also possible that CTB shocks evoke LC, spikes through pathways with one or more interposed neurons. The identity of the interposed neurons is unknown, and all neurons in the MNTB may be considered as possible candidates. A significant body of physiologic data are available for two MNTB neural classes: ‘compound waveform’ units and ‘off’ units (Guinan et al., 1972a,b). Compound waveform units (i.e., MNTB principal neurons) are probably not the interposed neurons (especially for a single-interneuron pathway) because synapses from axons of MNTB principal neurons are inhibitory in the LSO and seem likely to be inhibitory elsewhere (Glendenning et al., 1985; Spangler et al., 1985; Tsuchitani, 1988a). A few ‘off’ units were encountered in the present study and were found to respond to CTB shocks with a response just after the termination of a train of shocks. Although some LC, spikes were similar in their timing to spikes from these ‘off units, most LC, spikes occurred during the train of CTB shocks. We think that ‘off’ units are probably not the interposed neurons. Inhibitory effects of small endings on MNTB principal neurons We found relatively little direct evidence for inhibition in MNTB principal neurons. In one

clear-cut case, sound or shock stimuli which evoked C, + C, responses also evoked inhibition of ‘A’ spikes at a longer latency (see Fig. 10). One interpretation of these data is that the firing of the principal neuron (or alternately a collateral of the calyciferous axon through one or more intemeurons) evoked recurrent inhibition which prevented antidromic activation of the principal neuron. Recurrent inhibition in MNTB principal neurons might explain several other findings. (1) In some cases, both C, + C, spikes and LC, spikes were evoked by the same stimuli but never on the same response trace. Perhaps, each initial C, + C, response evoked recurrent inhibition which blocked the later LC, spike. (2) For the case shown in Fig. 8 top, a conductance increase in the MNTB principal neuron evoked by recurrent inhibition might explain why C, and LC, responses which occurred shortly after a previous C, or LC, response were smaller than C, or LC, responses which did not immediately follow a C, or LC, response. (3) Perhaps it is not by chance that the only neuron on which we found sound-evoked LC,-like spikes was also the only neuron on which there were no sound-evoked C, + C, responses. Perhaps, under most conditions, sound-evoked C, + C, spikes produce sufficient recurrent inhibition to block a normally present sound-evoked excitatory effect, so that sound does not normally evoke LC, spikes. (4) With recurrent CTB shock bursts (the optimal stimuli for evoking LC, spikes), LC, spikes were never evoked on every trial at any shock level, and there was a tendency for periods with no LC, responses to alternate with periods with LC, spikes on most trials. If recurrent inhibition were present, then each LC, spike would evoke recurrent inhibition which might block subsequent LC, responses to the CTB-evoked excitation. The alternating periods with and without LC, spikes might be due to the inhibition requiring some time (or several LC, spikes) to build up. Although we were able to demonstrate inhibition in MNTB principal neurons which could block LC, spikes and ‘A’ spikes, it is not clear whether the inhibition can be strong enough to block transmission from the calyx. We never saw block of C, following Cr. However, our stimuli may normally evoke a combination of excitation and inhibition from the small endings. Perhaps a

333

stimulus which evoked pure inhibition would be powerful enough to block transmission from the calyx. One possibility is that the effects of the inhibitory synapses on MNTB principal neurons were reduced by the anesthetic. We did two experiments with Ketamine as an anesthetic. In these animals, C, spikes appeared to follow C, spikes consistently. However, these experiments are not conclusive, because few units were held long enough for extensive testing. The physiologic role of calyces of Held and MNTB principal neurons It seems likely that the role of calyces of Held and MNTB principal neurons is to provide fast, secure transmission of spikes and short-latency inhibition in the lower auditory system. It is clear from the data we have presented, that the calyx synapse is excitatory. There is also strong evidence that the effect of MNTB principal neurons is inhibition, at least in the LSO (Glendenning et al., 1985; Spangler et al., 1985; Tsuchitani, 1988a; also see Moore and Caspary, 1983). Thus, the principal neuron, with its one-to-one inhibitory output, acts as a sign inverter of the excitatory inputs from cochlear nucleus globular cells. Even though it contains a synapse, the crossed, inhibitory pathway through large trapezoid-body fibers and MNTB principal neurons is very fast. This is illustrated by a comparison with the crossed, excitatory pathway through medium-sized trapezoid body fibers. Bourk (1976) found that latencies for antidromic stimulation from the contralateral side were about 0.5 ms for units which probably correspond to globular cells and large trapezoid body fibers. To this we add 0.4 ms for transmission at the calyx-principal-neuron synapse and 0.1 ms for conduction in the postsynaptic axon (from our antidromic stimulation experiments this is usually enough time for the spike to arrive at the LSO), to give a total latency of 1.0 ms. In contrast, Bourk (1976) found latencies of about 1.0 ms for antidromic stimulation from the contralateral side for units which probably correspond to co&ear nucleus spherical cells and medium-sized trapezoid body fibers. Thus, conduction times to the opposite side are comparable in these two systems despite the interneuron in the

inhibitory pathway. For the main inputs to the LSO, the crossed, inhibitory pathway may be comparable in latency to the uncrossed, excitatory pathway in that the uncrossed input has less distance to travel from the CN but originates from spherical cells which have longer latencies in response to sound than globular cells (see Bourk, 1976). In addition to the LSO, MNTB principal neurons project to several other brainstem nuclei including the dorsomedial periolivary nucleus, the ventromedial periolivary nucleus, the intermediate nucleus of the VNLL, and perhaps to medial olivocochlear efferent neurons (Morest, 1968; Glendenning et al., 1985; Spangler et al, 1985; Zook and DiCaprio, 1988). Many of these nuclei are also innervated by the calyciferous axons (Morest, 1968; Smith et al., 1987; Friauf and Ostwald, 1988). Presuming that synapses from calyciferous axons are excitatory and synapses from principal neurons are inhibitory, neurons which receive inputs from both might receive a balanced combination of excitation and inhibition. This balance will be changed if transmission from the calyx to the MNTB principal neuron is inhibited, or if extra spikes, such as LC, spikes, are added. Summary Compound waveforms recorded in the MNTB, and elsewhere, appear to be due to presynaptic spikes from large terminal endings, followed securely by spikes evoked in the postsynaptic cell. MNTB principal neurons provide fast one-to-one transmission from calyciferous axons and produce short-latency inhibition in the LSO and elsewhere. Non-calycine synapses also excite MNTB principal neurons, and the balance of excitation and inhibition in principal neurons may affect many other neurons. Acknowledgements We thank the many members of the EatonPeabody Laboratory who provided help throughout this work. We also thank Drs. D.K. Morest and W.B. Sewell for comments on the manuscript. Portions of this work were submitted by R. Y-S.

334

Li as a thesis to the Department of Electrical Engineering, Massachusetts Institute of Technology. This work was supported by NIH grants R01 NS01344, PO1 GM14940, PO1 NS13126, and ROl NS20269. References Adams, J.C., (1978) Morphology and physiology in the ventral nucleus of the lateral lemniscus. Sot. Neurosci. Abstr. 4, 1. Boudreau, J.C. and Tsuchitani, C. (1968) Binaural interaction in the cat superior olive S segment. J. Neurophysiol. 31, 442-454. Bourk, T.R. (1976) Electrical responses of neural units in the anterovental cochlear nucleus of the cat. PhD Dissertation. Dept. of Elec. Eng. and Comp. Sci. MIT, pp l-385. Eccles, J.C. (1964) The Physiology of Synapses. Springer, Berlin. Freygang, W.H. and Frank, K. (1959) Extracellular potentials from single spinal motoneurons. J. Gen. Physiol., 42, 749760. Friauf, E, and Ostwald, J. (1988) Divergent projections of physiologically characterized rat ventral cochlear nucleus neurons as shown by intra-axonal injection of horseradish peroxidase. Exp. Brain Res. 73, 263-284. Glendenning, K.K., Hutson, K.A., Nudo, R.J. and Masterton, R.B. (1985) Acoustia Chiasm II: Anatomical basis of binaurality in lateral superior olive of cat. J. Comp. Neurol. 232, 261-285. Guinan, J.J. Jr. (1979) Physiology of MNTB neurons receiving large endings. Sot. Neurosci. Abstr. 5, 21. Guinan, J.J., Jr., Guinan, S.S. and Norris, B.E. (1972a) Single auditory units in the superior olivary complex I: Responses to sounds and classifications based on physiological properties. J. Neuroscience 4, 101-120. Guinan, J.J., Jr., Norris, B.E. and Guinan, S.S. (1972b) Single auditory units in the superior ohvary complex II: Tonotopic organization and locations of unit categories. J. Neurosci. 4, 147-166. Jean-Baptiste, M. and Morest, D.K. (1975) Transneuronal changes of synaptic endings and nuclear chromatin in the trapezoid body following cochlear ablations in cats. J. Comp. Neural. 162, 111-134. Kiang, N.Y.S., Watanabe, T., Thomas, E.C. and Clark, L.F. (1965) Discharge patterns of single fibers in the cat’s auditory nerve. MIT Press, Cambridge. Lenn, N.J. and Reese, T.S. (1966) The fine structure of nerve endings in the nucleus of the trapezoid body and the ventral cochlear nucleus. Am. J. Anat. 118, 375-390. Li, R.Y-S. and Guinan, J.J. Jr. (1971) Antidromic and orthodromic stimulation of neurons receiving calyces of Held. Massachusetts Institute of Technology, RLE Quarterly Progress Report 100, 227-234. Moore, M.J. and Caspary, D.M. (1983) Strychnine blocks binaural inhibition in lateral superior olivary neurons. J. Neurosci. 3, 237-242. Morest, D.K. (1968) The collateral system of the medial nucleus

of the trapezoid body of the cat, its neuronal architecture and relation to the olive-co&ear bundle. Brain Research 9, 288-311. Mores& D.K. (1973) Auditory neurons of the brain stem. Adv. Oto-Rhino-Laryng. 20, 3377356. Morest, D.K., Kiang, N.Y.S., Kane, E.C., Guinan, J.J.Jr. and Godfrey, D.A. (1973) Stimulus coding at caudal levels of the cat’s auditory nervous system: II, Patterns of synaptic organization. In: Basic Mechanisms in Hearing, Academic press, N.Y. pp 479-509. Nakajima, Y. (1971) Fine structure of the medial nucleus of the trapezoid body of the bat with special reference to two types of synaptic endings. J. Cell. Biol. 50, 121-134. Pfeiffer, R.R. (1966) Anteroventral cochlear nucleus: waveforms of extracellularly recorded spike potentials. Science 154, 667-668. Rambn y Cajal, S. (1909) Histologic du Systeme Nerveux de I’Homme et des vertebres, Chapter 28, Vol. 1. Instituto Rambn y Cajal, Madrid, Spain. Rasmussen, G.L. (1967) The efferent connections of the cochlear nucleus. In: Sensorineural Hearing Processes and Disorders. Little Brown Co., Boston, pp 61-75. Rhode, W.S., Oertel, D. and Smith, P.H. (1983) Physiological response properties of cells labeled intracellularly with horseradish peroxidase in cat ventral cochlear nucleus. J. Comp. Neural. 213, 448-463. Rouiller, E.M. and Ryugo, D.K. (1984) Intracellular marking of physiologically characterized cells in the ventral cochlear nucleus of the cat. J. Comp. Neurol. 225, 167-186. Smith, P.H., Camey, L.H. and Yin, T.C.T. (1987) Projections of globular bushy cells in the cat. Sot. Neurosci. Abstr. 13, 547. Smith, P.H. and Rhode, W.S. (1987) Characterization of HRPlabeled globular bushy cells in the cat anteroventral cochlear nucleus. J. Comp. Neural. 266, 360-375. Spangler, K.M.. Warr, W.B. and Henkel, C.K. (1985) The projections of principal cells of the medial nucleus of the trapezoid body in the cat. J. Comp. Neural. 238, 249-262. Takeuchi, A. and Takeuchi, N. (1962) Electrical changes in pre- and postsynaptic axons of the giant synapse of loligo. J. Gen. Physiol. 45, 1181-1193. Tolbert, L.P., Morest, D.K. and Yurgelun-Todd, D.A. (1982) The neuronal architecture of the anteroventral co&ear nucleus of the cat in the region of the cochlear nerve root: Horseradish peroxidase labelling of identified cell types. Neuroscience 7, 3013-3052. Tsuchitani, C. (1988) The inhibition of cat lateral superior olive unit excitatory responses to binaural tone bursts. I: The transient chopper response. J. Neurophysiol. 59, 164183. Warr, W.B. (1982) Parallel ascending pathways from the co&ear nucleus: Neuroanatomical evidence of functional specialization. In: W.D. Neff (Pd.), Contributions to sensory physiology. Vol. 7. Academic Press, NY, pp l-38. Zook. J.M. and DiCaprio, R.A. (1988) Intracellular labeling of afferents to the lateral superior olive in the bat, Eptesicus fuscus. Hear. Res. 34, 141-148.