Physiological and morphological characterization of efferent neurones in the guinea pig cochlea

Hewing

Rrsearch.

63

20 (1985) 63-11

Eisevier

HRR 00653

Physiological and morphological characterization guinea pig cochlea

of efferent neurones in the

Donald Robertson and Mark Gummer Department

of Physiology,

University

(Received

ofWestern

Australia,

29 April 1985; accepted

Nedlands,

9 August

W.A. 6009, Australra

1985)

Efferent neurones within the intraganglioni~ spiral bundle of the guinea pig cochlea were characterized in terms of their response properties, and their pattern of ter~nation witbin the receptor organ revealed by intracellular labelling with horseradish peroxidase. Ail neurones subsequently identified as efferent neutones had clear features of their response properties which distinguished them from primary auditory afferents. They had long latency, low maximum discharge rate and low levels of spontaneous activity under Nembutal/Innovar anaesthesia. The pattern of discharge was extremely regular, revealed by symmetrical interspike interval histograms. 49.4% responded best to ipsilateral, 43.3% to contralateral sound and a third group of 7.3% responded equally well to either ipsiiateral or contralateral sound. In cochleae in good physiological condition, these efferents were as sensitive and as sharply tuned as primary afferents with the same characteristic frequencies (CFs). Ail efferents fully traced in histological processing terminated on the outer hair cells. Several efferents showed extensive branching beneath the inner hair ceils which might represent en passant synapses with other neuronal elements. There was clear evidence of tonotopic organization of the efferent projection. The site of termination on the outer hair cells in most instances was very close to the region of the cochlea predicted from the fibres’ CF and the known place-frequency map for primary afferent neurones in the guinea pig. cochlear

efferent,

response

properties,

horseradish

peroxidase

Introduction The efferent olivocochlear pathway in mammals has its cell bodies of origin in the vicinity of the superior olivary complex and projects to both ipsiand contralateral cochleae (Warr, 1975,19X; Warr and Guinan, 1979; White and Warr, 1983). Here the efferent neurones form synaptic terminals on both receptor hair cells and on the dendrites of the primary afferent neurones. Recent work in the cat (Warr, 1978: Warr and Guinan, 1979) has classified the efferents into two major systems. A lateral system of small unmyelinated axons projects mainly ipsilaterally from the region of the lateral superior olive (LSO) to terminate exclusively in the region of the inner hair cells, probably on afferent dendrites. A medial-ventral system of large myelinated axons arises from cells in the medial and ventral trapezoid nuclei and the medial periolivary complex and projects predomin~tly 0378-5955/85/$03.30

Q 1985 Elsevier Science Publishers

injection

contralaterally to form the large synaptic endings on outer hair cells. There have been several studies of the physiological responses of the olivocochlear neurones to sound (Cody and Johnstone, 1982; Fex, 1962, 1965; Liberman and Brown, 1985; Robertson, 1984; Rupert et al., 1968). They are characterized mainly by their long latency, and low and regular discharge rate. In some of these studies, the ipsilateral cochlea had to be destroyed in the course of the surgical approach. Others were undertaken at a time when knowledge of the normal and abnormal response properties of the primary cochlear afferents was incomplete. More recent studies (Cody and Johnstone, 1982; Robertson, 1984) in which care was taken to monitor the sensitivity of primary afferents in the same animal, report that neurones with efferent-like properties are as sharply tuned and as sensitive as primary auditory afferents. This is in contrast to earlier work in which efferents

B.V. (Biomedical

Division)

64

were reported to be relatively insensitive and broadly tuned. In only two of these previous studies, those of Robertson in the guinea pig (1984) and Liberman and Brown in the cat (1985), were the neurones which were classed as efferent on the basis of response properties, positively identified by intracellular labelling with neuronal tracers. In the experiments reported here, recordings were obtained from efferent neurones in the intraganghonic spiral bundle of the guinea pig cochlea. The responses of these neurones to both ipsilateral and contralateral sound were characterized and they were filled by intracellular injection with horseradish peroxidase (HRP) for subsequent identification of the site of termination within the inner ear. Brief reports of the techniques used and some preliminary results of this study have appeared elsewhere (Robertson, 1984, 1985). Methods Experiments were performed on young pigmented guinea pigs (180-350 g). They were anaesthetized with 25 mg/kg of Nembutal and 4 mg/kg of Innovar-Vet, paralysed by intramuscular injection of Alloferrin and artificially respirated. Supplementary doses of Innovar were given every hour and Nembutal every 2 h. In some experiments, animals were anaesthetized with 1.5 g/kg of LJrethane and allowed to breathe unassisted. In all experiments, pure tone and white noise search stimuli were delivered to both ears through closed calibrated hollow ear bars. The sound source for the left ear was a l/2-inch reverse-driven condenser microphone (Briiel & Kjaer type 4134) and for the right ear was a Beyer DT 49 loudspeaker connected to the ear bar by a 4.5 cm length of polyethylene tubing. Recordings were obtained of single neurone activity in the basal turn of the left cochlea using glass pipette microelectrodes filled either with 150 mM KC1 or 10% HRP (Sigma Type VI) in 50 mM KC1 and 20 mM Tris-HCl pH 7.4. The micropipettes (tip diameter -G0.15 pm) were fabricated on a Brown-Flaming micropipette puller (Brown and Flaming, 1977). HRP-filled electrodes

were bevelled from starting resistances of 150 -200 Mfz to final resistances of 80-95 MS2. The surgical approach shown schematically in Fig. 1 was similar to that described previously for recordings of primary afferent neurone activity in the spiral ganglion (Robertson and Manley, 1974). In the present experiments, however, the final hole in the bony osseous spiral lamina was made on the peripheral edge of Rosenthal’s canal, where the main intraganglionic spiral bundle containing the olivocochlear efferents courses along the cochlear spiral. The success of the experiment was dictated primarily by the accurate placement and uncluttered nature of this hole which was made with a hand-held steel pick. Throughout the course of these surgical procedures, careful attention was paid to the physiological condition of the cochlea by monitoring the threshold of the gross N, response to tone bursts ranging from 2 to 30 kHz (Cody et al., 1980; Johnstone et al., 1979). With care the surgery could be completed without significant change in N, thresholds across this frequency range. Data recorded from single neurones included mean spontaneous firing rate collected over a 10 s period, latency in response to tone bursts of 0.5 ms rise time, 250 ms duration and at a rate of l/s, frequency-threshold response curves (tuning curves) for both ipsilateral and contralateral sound, and input-output curves. For afferent neurones, tuning curves are determined with an automatic computer-controlled regime. For efferent neurones tuning curves were measured using audiovisual criteria. This was necessary because of the long HW-filled

intraganplimic bmdke

spiral

microelectrode

i organ of cwti

Fig. 1. Schematic illustration of the surgical approach to efferent fibres in the intraganglionic spiral bundle of the guinea pig cochlea.

latency of the acoustically-evoked response of many efferent neurones, and the relatively brief period for which recordings could be obtained in many instances. Interspike interval histograms of both spontaneous and acoustically-driven activity were also collected from many afferent and efferent neurones. The acoustic cross-talk between left and right ears was quantified using a form of bioassay already described by Caird et al. (1980). It was assumed that the response of primary auditory afferents in the left cochlea to sound in the right ear was due solely to leakage of sound across the head. A comparison of thresholds to left and right ear stimulation at the most sensitive or characteristic frequency (CF) of a number of primary afferents could thus be used to quantify this interaural cross-talk. The surgical approach used to obtained afferent neurones with CFs below 12 kHz was that described by Alder and Johnstone (1978). After physiological characterization of an efferent neurone was complete, when HRP-filled electrodes were used, and a stable negative resting potential of > 15 mV had indicated a clean impalement, HRP was injected into the neurone by passage of 3-10 nA depolarizing current pulses of 50 ms duration at a rate of 10/s for up to 5 min. In early experiments 500 ms pulses at a rate of l/s were used. Post injection survival times ranged from 1.5 to 4 h. Subsequent histological processing which allowed visualization of the peripheral processes of the injected cell has been described in detail previously (Robertson, 1984). Results Cochlear openings The use of the N, threshold audiogram enabled the functional condition of both left and right cochleae to be assessed at all times. This technique has been shown to be a reliable indicator of the threshold at the CF for fibres from the 2-30 kHz place in the guinea pig cochlea (Cody et al., 1980; Johnstone et al., 1979). Fig. 2A shows results from one animal in which surgical opening of the left cochlea and osseous spiral lamina overlying the intraganglionic spiral bundle caused no significant deterioration in N, sensitivity. Several hours after opening most animals showed a significant de-

60,

A

49 I

20;

t I

\

24

Fig. 2. N, audiograms before and after completion of left cochlea surgery for recording from efferents. (A) Successful opening of left cochlea with no significant loss of sensitivity after opening. (B) Animal which showed loss of N, sensitivity caused by surgical opening of left cochlea. (C) Right (unoperated) cochlea before opening of left cochlea shown in B, and several hours after opening. n . before cochlear opening; 0. after cochlear opening.

66

terioration in the N, sensitivity to high frequency tones in the opened cochlea with no alteration in the Ni response of the unopened cochlea (Fig. 2C). Fig. 2B shows a case of sudden deterioration of the N, on the operated side caused by opening the cochlea. Such changes in N, sensitivity could be readily monitored and their reflection in single fibre responses, both afferent and efferent, were easily interpretable. In about 50% of animals, single-fibre data from both afferents and efferents were obtained without significant deterioration having occurred in the N, sensitivity of the operated cochlea. The white noise search stimulus presented to each ear was maintained at an intensity more than 20 dB above the N, threshold for that ear and at least 15 dB below the N, cross-talk threshold of the opposite ear, to compensate for any deterioration in sensitivity over the course of the experiment. Basic properties of effeerent new-ones In the vicinity of the intraganglionic spiral bundle, judicious o~entation of the microelectrode enabled recordings to be obtained from single neurones with response properties quite distinct from primary auditory afferents. These neurones responded to adequate acoustic stimulation with extremely regular discharge pattern in contrast to the stochastic discharge typical of primary afferents. The regularity of discharge is evident from Fig. 3A and B which show chart records of the firing of one efferent neurone and in Fig. 3C-G showing typical examples of the interspike interval histograms derived from efferent neurones. These are compared in Fig. 3H to the skewed interval histogram typical of primary afferents driven at similar discharge rates. The responses to ipsilateral and contralateral sound were characterized for a total of 247 efferent neurones. In terms of their response to acoustic stimulation these neurones with extremely regular discharge pattern fell into three major classes. One group (49.4%), which we have called ipsilaterally activated, was most sensitive to ipsilateral sound of a particular frequency and if they responded at all to contralateral sound of the same fxequency, it was at a stimulus intensity which exceeded the interaural cross-talk threshold. A second major group (43.3%), designated as con-

tralaterally activated, was most sensitive to contralateral sound of a particular frequency and if they responded to ipsilateral sound it was again only at levels exceeding the cross-talk threshold. A third minor group (7.3%), called ipsi-contra activated, responded with roughly equal sensitivity to sound in either ear. The most sensitive frequency was very similar for each ear and the response to sound in either of the two ears could not be explained by interaural cross-talk alone. In the subset of animals which showed significant deterioration of N, sensiti~ty in the operated cochlea (50%), the ratio of ipsilaterally activated (51.7%) to contralaterally activated (40.4%) efferent neurones was similar. This indicates that alterations in ipsilateral sensitivity caused by the surgical procedure do not significantly alter the distribution of the major efferent groups determined by the methods described above. Fig. 4 shows a scatter plot of ipsilateral versus contralateral thresholds for 178 units encountered with the characteristic regular discharge pattern described above. The separation into three groups can be clearly seen. The dotted lines in this figure show the limits of interaural cross-talk obtained from primary afferents across the range of most sensitive frequencies encountered. All units which are truly monaural should fall within the boundaries defined by these lines. The fact that some units did not respond to sound in the least sensitive ear even above cross-talk threshold implies the existence of binaural suppression in some cases. A further distinguis~ng feature of these neurones was their long latency in response to sound at their most sensitive frequency. This was measured visually from the oscilloscope screen or from computer-generated peri-stimulus histograms at a sound intensity sufficient to cause the neurone to fire with its maximum discharge rate. Fig. 5 shows the minimum latency for all cells in which it was measured as a function of the cells’ most sensitive frequency. Also plotted on the same graph are the data of Evans (1972) showing the latency of response of primary auditory afferents in the guinea pig cochlea to click stimuli. For all efferent neurones the latency of response is clearly considerably longer than that for primary afferents. Fig. 6 shows

the distribution

of latencies

in the three

67

A

6kHz-~PSILATERAL

*I;.-

GE9WO2 IPSI EFF. CF - Ski+ 40 d8 SPL

CF - 9.ZkHz

0

20

40

60

.o

80

INIERSPlKE INTERVAL (A

25: ,

20

40

60

0

80

1 202 ,

20

40

60

60

INTERSPIKE INTERVAL (ms)

INTERSPIKE INTERVAL (.s)

1 G847/03 AFFERENT

25X

~~~~~

20%

4

0%

0

to

40

60

INTERSPIKE INTERVAL

80

0 ._

20

40

60

INTERSPIKE !NlERVAL (ms)

80

0

20

43

60

INTERSPIKE IHIERVAL imr)

Fig. 3. (A,B) Chart records of the response of a single ipsilaterally activated efferent neurone to the presentation of 250 ms tone pips at CF (6 kHz). Numbers beside traces denote dB SPL. (A) Response to ipsilateral sound. (B) Response to contralateral sound. (C-H) Interspike interval histograms of the response of 6 neurones to a continuous CF tone. Ordinate: percentage of the total number of spikes (always greater than 1000) collected from any one interval bin. (C) Efferent shown in Fig. 3A.B. (D-G) Interval histograms from other typical efferent neurones. (H) Typical skewed distribution exhibited by an afferent neurone.

6R 130 120 110 100

a-J90 z80 0q70

”

10 L 10

/ / I I III I I I II 11 20 30 40 50 60 70 80' 90 100 110 120 IPSI CF THRESHOLD,dB SPL

I

Fig. 4. Ipsilateral and contralateral CF thresholds for 178 aiqle efferents. 0, units showing clear monaural sensitivity. 0, units which responded to sound in both ears below interaural crosstalk threshold. Dotted lines show the limits of ipsilateral and contralateral thresholds calculated for purely monaural units on the basis of interaural cross-talk. Arrows indicate units for which one ear threshold was beyond the maximum sound pressure available at the fibre’s CF.

C

!

I

I_.::, 0

70 r

60-

0

0 0

5

IO

15

20

25

30

:i

35

40

45

50

MIN. LATENCY

55

50

(ms)

Fig. 6. Distribution of minimum latencies to CF tones for the three response classes: A, ipsilaterally activated; B, contralaterally activated; C, ipsi-contra activated.

c 0

response classes for all units in which it was measured. The mean latency for 70 of the contralaterally activated group of neurones was 23.1 f 13.1 ms and for the ipsilaterally activated group 24.8 + 12.7 ms (n = 90). The small ipsi-contra group had a mean minimum latency of 18.9 -t 7.5 ms (n = 8).

o-

-------------_\

0.2

I

- --

1

_____

I

10

___

I

1

20 30

CF , kHz

Fig. 5. Distribution of minimum latency of efferents to CF tones, as a function of fibre CF. Dotted line shows maximum latencies of afferent neurones to clicks from Evans (1972).

Tuning cwves and sensitiviry In agreement with two recent reports from this laboratory (Cody and Johnstone, 1982; Robertson. 1984), the efferent neurones recorded from in the present study showed sensitivity and frequency selectivity similar to that of primary auditory neurones. When the cochlea from which recordings

69

80

i

FREQUENCY,

i20

kHz

1

i 100

i

=: 80 60

/ 40 c 20+ I

01

I

.5

1

1

5

10

FREQUENCY,

20 30 kHz

120

fi

vi

m u

IflO I

were being obtained was in good condition as estimated by N, thresholds and single afferent neurone tuning curves, then the efferent neurones showed sensitivity at CF and sharpness of tuning curves commensurate with those abundantly reported for primary afferents in the guinea pig cochlea. This applied to all three major response classes. Typical examples of tuning curves are shown in Fig. 7. The small group of ipsi-contra efferents showed very similar CF when tested in either ear, though there were often differences in tuning curve shape (Fig. 7C). In the present experiments the efferent neurones encountered at a particular location in the intraganglioni~ spiral bundle showed a wide range of characteristic frequencies, in contrast to the tight range of CFs encountered for the primary afferents in the spiral ganglion (Robertson et al., 1980; Robertson and Manley, 1974). For this reason, direct comparison of tuning curves of afferents and efferents was not always possible in the same animal. One example however is shown in Fig. 8. It is clear that there is little difference in sensitivity and sharpness of tuning between primary afferents and efferent neurones of the same CF, provided there has been no deterioration in the condition of the cochlea during the total recording period. In animals in which the ipsilateral cochlea suffered deterioration, either as a direct result of the surgery, or as a function of time after opening the cochlea, there was a systematic alteration in the sensitivity and tuning properties of the ipsilaterally-activated group which was consistent with the known alterations in response properties of the primary afferents (Evans, 1972. 1984; Robertson and Manley, 1974) which must ultimately provide the input to these efferent neurones. The sensitivity of the contralaterally activated group remained unaltered in the face of changes in ipsilateral N, threshold. Spontaneous

01 t .5

I

1

s FREQUENCY,

10

activity

Under Nembutal

and Innovar

anaesthesia,

most

20 30 kHz

Fig 7. Representative tuning curves to preferred stimulation for (A) ipsilaterally activated, (3) contralaterally activated, and (C) ipsi-contra activated efferents (two units shown with open

symbols response to ipsilateral and closed symbols response to contralateral stimulation). All units shown are from cochleas in which no significant deterioration of N, threshold had occurred for frequencies in the region of the fibres’ CF.

found is shown in Fig. 9. When spontaneous activity was present it was generally regular in nature and exhiblted a symmetrical interspike interval histogram similar to those shown above for acoustically driven activity. in a second series of experiments, guinea pigs were anaesthetized with Urethane instead of Nembutal and Innovar. There was a significant difference found in the spontaneous activity of neurones with otherwise identical response properties to those described above. In (Jrethaneanaesthetized animals, 50.3% showed non-zero rate of spontaneous activity, compared with 11.1% in the Nembutal-I~ovar group. The mean rate of spontaneous firing for all neurones with non-zero rates was 8.1 ~fr7.1 and 4.1 f 4.3 spikes/s in the Urethane and Nembutal-Innovar groups, respectively. There was thus a clear tendency towards high spontaneous firing rates in the Urethaneanaesthetized group. The properties of primary afferent neurones did not differ notably between the two groups of animals. Other features of re-

80

4

.5

1

5

10

FREQUENCY,

20

30

20

30

kHz

01 f

.s

4

1

5 FRELIUENCY.

10

NO. UN1

kHz

Fig. 8. Direct comparison of tuning curves of (A) primary afferent, and (B) efferent neurones in the same cochlea. Open circles in B denote contralaterally activated, and closed circles ipsiiaterally activated efferents. (C) N, audiograms before and after cochlear opening.

efferent neurones encountered showed low spontaneous firing rates. In many cases the spontaneous firing rate was zero and the neurone’s presence was onIy apparent when search stimuli were presented. The ~st~bution of spontaneous firing rates

SPONT. RATE CAP's per sec.) Fig. 9. Distribution of mean spontaneous firing rates in Nembutal/Innovar anaestbetized guinea pigs.

71

GEl28/08 CF = 1OkHz PST at 42 d0 SPL 250 ms tone P.D.F. Usec) 60

150 Time (ms>

200

250

GB36/03

300

C

CF = 3kHz

FIRING 3o RATE, AP's per sec.

FIRING 3o RATE, AP's per sec. 20

20

-t CO 7kb CBf02106

J 0

" 4 ,,i 20

40

d8

SPL6

Fig. 10. (A) Typical peri-stimulus-time histogram shown above histogram. (B) Typical input-output tones. (C) Input-output curves of an ipsi-contra

80

1

40

d% SPL6 of efferent neurone responding to CF tone. Time of presentation of tone burst is curves of two contra- and one ipsilaterally activated efferents responding to CF activated efferent.

72

other two groups, with a low maximum discharge rate. Interestingly, the slope of the input-output curves was not always the same for ipsilateral and contralateral sound presentation, with the firing rate generally showing a lower rate of increase as a function of contralateral sound pressure. In several efferents there was also clear evidence of binaural interaction. Fig. II shows examples of binaural facilitation and binaural inhibition of two ipsilaterally activated efferent neurones. These effects of simultaneous presentation of ipsi- and ~ntr~ateral acoustic stimuli were found at levels below the cross-talk threshold.

sponse properties of efferent neurones in Urethane-anaesthetized animals will be presented in another report. Input-output curves

The firing rate of efferents as a function of sound-pressure level was measured by taking the mean rate over 10 presentations of a 400 ms tone burst at the fibres’ characteristic frequency. Examples of such input-output curves for one ipsilaterally and two contr~aterally activated efferents are shown in Fig. 10B together with a representative pea-stimuius-time histogram (Fig. 10A). The periodic fluctuations in the histogram reflect the regularity of interspike intervals shown by efferent neurones. Typically all efferents studied showed a low maximum discharge rate, the mean for 22 ipsilaterally activated efferents being 40.3 f 13 and 37.6 f 10.7 spikes/s for 7 contralaterally activated efferents. Some input-output data was also collected for the rare ipsi-contra activated group. An example is shown in Fig. IOC. They appeared to behave in a manner quatitativel~ similar to the

_-

A

Morphology of HRP-injected efferents

A total of 19 injected fibres with properties such as those described above were retrieved in subsequent histological processing. Five of these could not be traced to their site of termination within the organ of Corti but were presumed to be efferent neurones because of their long (several millimetres) spiral course within the intraganglionic spiral bundle.

GB38/06

GB127/02

CF = 6kHz

CF = 17kHz

40. ,q ,‘.’

FIRING

RATE, AP's per sec.

ap

308 IPSf

RATE, AP's per sec.

1

ALONE

20-

--.+CO 37d9 SPL

20-

-r-*CO 47dB SPL +. +CO 57dB SPL

04 0

IO-

a TO

40

dB SPL6

80

1

of i8 0

20

40

dB SP:

Fig. 11. Examples of (A) binaural facilitation and (B) suppression for two ipsilaterally activated efferents.

80

I0

All the fibres which were traced to the receptor organ showed terminal arborizations forming synapses on the outer hair cells. This was the case for three physiologically characterized contralaterally activated efferents and 11 characterized ipsilaterally activated efferents. No successful fills were obtained for the sparse ipsi-contra group of efferents. There was a wide range of morphological types when the details of the terminal arborization pattern are considered. In both major response groups some efferents showed a simple arborization pattern, with only one or two branches radially crossing the tunnel of Corti to terminate on the outer hair ceils in a very restricted region (Fig. 12A,D,E). Other efferents showed extensive branching be-

_.__----

OHC __-----

-

Ii_1C--------

_---

- -__

--

- - --

_________

~

a

i’

A

E

_----

-----------.----_

/...

_,

Fig. 12. Camera lucida drawings of the termination pattern of (A-C) three ipsilaterally activated efferents (with CF: A = 13.5 kHz, B = 16 kHz, C = 11 kHz). IHC, inner hair cells; OHC, outer hair cells. Calibration bar -100 pm. (D-F) Three contralaterally activated efferents (with CF: D = 8 kHz, E = 11 kHz, F = 10 kHz). Calibration bar = 100 pm. Base of cochlea is on left in al1 eases.

__-----.

neath the inner hair cells giving rise to numerous tunnel crossing fibres of varying diameter which finally synapsed with the outer hair cells (Fig. 12B,C,F). Even in these neurones however, examples of which were found in both major response types, the length of organ of Corti over which terminations were found rarely extended over more than 150 pm (range = 55-220 pm). Only one instance was found (Fig. 12C) in which a long process extended apically beneath the inner hair cells after sending out one tunnel crossing branch. This apical spiral fibre sent branches across the tunnel at intermittent sites along its extent, before the density of HRP reaction product became too faint for it to be traced further. All the tunnel crossing branches of this ipsilaterally activated efferent formed synapses with the outer hair cells. Many of the fine processes within the organ of Corti showed an intermittent beaded appearance reminiscent of en passant varicosities. This was particularly true of the fine fibre branches beneath the inner hair cells. No efferents were retrieved which only showed ramifications beneath the inner hair cells and did not synapse with outer hair cells. Neither were any successful retrievals achieved of the rare class of ipsi-contra activated efferents. For all efferents‘retrieved except that shown in Fig. 12C with a long apical-going process, there was an obvious relationship between the fibre’s CF, either to ipsi- or contralateral sound and the region of the sensory epithelium on which the outer hair cell terminations were found. Fig. 13 shows the relationship between the measured fibre CF and the CF calculated from the known afferent place-frequency map (Liberman, 1982; Robertson et al., 1980; Robertson and Manley, 1974) and the location of the centre of the terminal field. There is a clear tendency for efferents of a given CF to terminate at locations where primary afferent neurones emanating from the inner hair cells have the same CF. However, it is obvious that the efferent terminations cover a wider segment of the basilar membrane than that related to a single afferent neurone. This tonotopicity in the location of outer hair cell terminations appeared to hold for both ipsi- and contralaterally activated efferents.

TRUE CF. kHz

Fig. 13. Relationship between fibre CF (horizontal axis) and location of centre of terminal field expressed as mm from basal end of basilar membrane for 12 efferents. Estimated fibre CF calculated from afferent place-frequency map of Robertson et al. (1980). 0, ipsilaterally activated efferents; 0. contralaterally activated efferents.

Discussion

Several important points emerge from the present data. Firstly, the experiments provide direct confirmation that the fibres with response properties described are indeed efferent fibres. The pattern of termination and general morphology of HRP-injected single neurones in this study agrees in qualitative terms with that described in numerous anatomical studies of the olivocochlear efferents (Brown, 1985; Lorente de No, 1937; Robertson, 1984; Smith, 1975; Smith and Haglan. 1973; Spoendlin, 1972; Spoendlin and Gacek, 1963) and there seems little likelihood that they could be outer hair cell afferents. The pattern of innervation is strikingly different from that of the inner hair cell afferent population (Liberman, 1982; Lorente de No, 1937; Robertson, 1984; Spoendlin. 1972). On the basis of anterograde transport of radiolabelled amino acids, Warr and coworkers concluded that in the cat virtually all the lateral olivocochlear neurones form synapses only underneath the inner hair cells. The fact that in this study all efferents of both ipsi- and contralaterally

15

activated groups, whose peripheral terminations could be traced, formed synapses on the outer hair cells, would lead us to suggest that they are equivalent to the medial system described in the cat by Warr and coworkers (Warr, 1975, 1978; Warr and Guinan, 1979). A recent preliminary report by Liberman and Brown (1985) provides direct confirmation of these results in the cat. These workers have traced both the peripheral terminations and cell bodies of origin of four neurones with similar response properties to the ones we have described. All four neurones, terminating on outer hair cells, were traced to large cell bodies in the brainstem in the region of origin of the medial olivocochlear system. Guinan and coworkers (1983) have reported a pattern of autoradiographic labelling over the outer hair cells which suggests that many of the efferents of the medial system form patchy, punctate terminations over long lengths of the receptor epithelium. Only one of the fibres retrieved in this study (Fig. 12C) and one described in an earlier report by one of us (Robertson, 1984), did show a termination pattern consistent with their findings. The efferent neurone shown in Fig. 12C with its long apically directed process is also strikingly similar to one shown in earlier Golgi studies (Smith, 1975). However, most of the fibres we retrieved did not fit this pattern, with the spread of their terminations being relatively restricted. It may be that some branches of these efferent neurones were not filled with HRP or stained too lightly for detection. The short injection and post-injection survival times could both reduce the yield of complete fills. This may be especially important if the neurones branch and the branch points are central to the injection site, for instance in the modiolus. Consequently, the fibre reconstructions showing the more extensive ter~nation patterns should perhaps be considered as a conservative estimate of the spread across the organ of Corti of a single efferent neurone. Our failure to find any efferents which terminated only beneath the inner hair cells suggests several possibilities. Either we did not record from any efferents in this group, or alternatively we did not manage to fill any of them successfully with HRP. If the axons of the lateral system are indeed thinner and unmyelinated in the guinea pig

as they are in the cat, then they may well be more difficult to inject and trace successfully. A further possibility is that the distinction between medial and lateral system inne~ation patterns is not as clearcut in the guinea pig as it is in the cat. A recent report by Brown (1985), however, provides direct evidence that the guinea pig does indeed possess a population of efferents terminating exclusively under the inner hair cells. Neither the present experiments, nor those recently reported by Liberman and Brown (1985) have succeeded in filling the small group of ipsi-contra activated neurones. It is tempting to speculate that these may belong to the lateral system. In our HRP-labelled material, some of the efferents ter~nating on outer hair cells have ample opportunity to make, and exhibit morphological features suggestive of, en passant synapses beneath the inner hair cells before crossing the tunnel of Corti to the outer hair cells. The HRP-injetted material should be examined at the electronmicroscopic level to see if en passant synapses are indeed made by the cells we have characterized. Another important area concerns the detailed nature of the response properties of these identified efferent neurones. Our observations on regularity of discharge, long latency and low maximum discharge rates are in good qualitative agreement with previous descriptions of olivocochlear efferents recorded with other surgical approaches (Fex, 1962, 196.5; Rupert et al., 1968). However, our clear finding that the sensitivity and tuning of efferents measured by our techniques are comparable to primary auditory afferents is in conflict with some earlier reports in which olivocochlear efferents are described as being broadly tuned and insensitive compared to primary afferents (Klinke and Galley, 1974). Our findings in this regard are, however, in agreement with two more recent reports from this laboratory (Robertson, 1984, 1985). and findings by Liberman and Brown in the cat (Liberman, 1982; Liberman and Brown, 1985). We believe that the answer lies in careful monitoring of the condition of the cochlea during the experiment. The N, audiogram enables far better detection of loss of sensitivity in particular regions of the cochlea than does the simple click N, or the cochlear microphonic so often favoured by earlier researchers. The extreme lability of the cochlear

76

transduction process and the broadening and desensitization of primary afferent tuning curves which occurs as a consequence of many insults are now well established (Evans, 1972, 1984; Manley and Robertson, 1976; Robertson et al., 1980; Robertson and Manley, 1974). In the present study we did see instances of broadly tuned and insensitive efferents but these were always in situations where the N, sensitivity in the frequency range of the fibres’ CF had deteriorated. This suggests that broadly tuned and insensitive efferents are a consequence of abnormal insensitive and broadly tuned incoming afferent drive to these neurones. The observation that the location of efferent terminations is tonotopically related to fibre CF in a manner very similar to the afferent placefrequency map, confirms an earlier report by one of us (Robertson, 1984), in which only ipsilateral sound stimulation was employed. It now appears that this holds true both for the ipsi- and contralaterally activated groups of efferents. Fibres with extensive spiral trajectories to their branches under the inner hair cells (Fig. 12C) cannot be strictly said to conform to this scheme. However, overall the results suggest that many efferent neurones project to the outer hair cells at a location along the cochlear spiral very close to the inner hair cells from which these same neurones probably receive their sharply tuned afferent input. In view of current notions that the outer hair cells may play a role in modulating the overall input to the inner hair cells, these findings have interesting implications for central nervous system control of the functioning of the peripheral auditory receptor.

Lastly, it should be borne in mind that these results were all obtained in anaesthetized animals. Our observation that the spontaneous firing rate of efferents, normally very low under Nembutal and Innovar anaesthesia, is significantly elevated under Urethane anaesthesia, implies that the level of central nervous system depression should be taken into account when thinking of the possible role and properties of this efferent pathway. We are currently performing experiments to see whether other properties of the efferents are also modified by the use of different anaesthetic regimes. In addition, the large range of latencies found in this study is consistent with the notion

that many efferents may be responding to sound via a polysynaptic pathway involving brainstem centres higher than the level of the superior olivary nuclei. Acknowkdgements This work was supported by grants from the National Health and Medical Research Council, the Australian Research Grants Scheme and the University of Western Australia. M.G. was a University Research Students~F recipient. References Alder, V.A. and Johnstone, B.M. (1978): A new approach_ to the guinea pig auditory nerve. J. Acoust. Sot. Am. 64, 684-687. Brown, M.C. (1985): Peripheral projections of labelled efferent nerve fibres in the guinea pig cochlea: An anatomical study. In: Abstr. VIIith Midwinter AR0 Meeting, 9-10. Brown, K.T. and Flaming, D.G. (19773: New microelectrode techniques for intrac&tlar work in small cells. Neuroscience 2, 813-827. Caird, D., Gottl, K.-H. and Klinke, R. (1380): Interaural attenuation in the cat, measured with single fibre data. Hearing Res. 3, 257-263. Cody, A.R. and Johnstone, B.M. (1982): Acoustically evoked activity of single efferent neurons in the guinea pig cochlea. J. Acoust. Sot. Am. 72,280-282. Cody, A.R., Robertson, D., Bredberg, G. and Johnstone, B.M. (1980): Electrophysiological and morphological changes in the guinea pig cochlea following mechanicat trauma to the organ of Corti. Acta Otolaryngol. 89, 440-452. Evans, E.F. (1972): The frequency response and other properties of single fibres in the guinea pig cochlear nerve. J. Physiol. (London) 226, 263-287. Evans, E.F. (1984): Effects of hypoxia on the tuning of single co&Lear nerve fibres. J. Physiol. (London) 238, 65-67. Fex, J. (1962): Auditory activity in centrifugal and centripetal cochlear fibres in the cat: A study of a feedback system. Acta Physiol. Stand. 55, Suppl. 189. Fex, J. (1965): Auditory activity in uncrossed centrifugal cochlear fibres in cat: A study of a feedback system JI. Acta Physiol. Stand. 64, 43-57. Guinan, J.J., Warr, W.B. and Norris, B.E. (1983): Differential olivococblear projections from lateral versus medial zones of the superior olivary complex. 3. Comp. Neural. 221, 358-370. Johnstone, J.R., Alder, V.A., Johnstone, B.M., Robertson, D. and Yates, G.K. (1979): Co&ear action potential and single unit thresholds. J. Acoust. Sot. Am. 65, 254-257. K&&e, R. and Galley, N. (1974): Efferent innervation of the vestibular and auditory receptors. Physiol. Rev. 54,316-357.

77

Liherman, M.C. (1982): The cochlear frequency map for the cat: Labelling auditory nerve fibres of known characteristic frequency. J. Acoust. Sot. Am. 72, 1441-1449. Liberman. M.C. and Brown. M.C. (1985): Intracellular labelling of olivocochlear efferents near the anastomosis of Oort in cats. In: Ahstr. VIIIth Midwinter AR0 Meeting, 13-14. Lorente de No (1937): The neural mechanism of hearing. Laryngcwope 47, 373-377. Manley. G.A. and Robertson. D. (1976): Analysis of spontaneous activity of auditory neurones in the spiral ganglion of the guinea pig cochlea. J. Physiol. (London) 258. 323-336. Robertson. D. (1984): Horseradish peroxidase injection of physioI(~gicalIy characterized afferent and eFferent neurones III the guinea pig spiral ganglion. Hearing Res. 15. 113-121. Robertson. D. (1985): Centrifugal control mechanisms in mammalian hearing. Proc. Austrai. Physiol. Pharmacoi. Sot. (in press).

Robertson, D. and Manley, G.A. (1974): Manipulation of frequency analysis in the cochlear ganglion of the guinea pug. J. Comp. Physiol. 91, 3633375. Robertson, D., Cody, A.R.. Bredberg, G. and Johnstone, B.M. (I 980): Response properties of spiral ganglion neurones in cochleas damaged by direct mechanical trauma. J. Acoust. Sot. Am. 67, 1295-1303. Rupert. A.L.. Moushegian. G. and Whitcomb, M.A. (1968): Olivocochlear neuronal responses in medulla of cat. Expt. Neural. 20, 575-584.

Smith. C.A. (1975): Inne~ation of the cochlea in the guinea pig by use of the golgi stain. Ann. Otol. Rhinol. Laryngol. 84, 443-459. Smith, CA. and Haglan, B.J. (1973): Goigi stains on the guinea pig organ of Corti. Acta Otolaryngol. 75, 203-210. Spoendlin, H. (1972): Innervation density of the cochlea. Acta Otolaryngol. 73. 235-248. Spoendlin, II. and Gacek, R.R. (1963): Electronmicroscopic study of the efferent and afferent innervation of the organ of Corti in the cat. Ann. Otol. Rhinol. Laryngol. 72. l-27. Warr, W.B. (197.5): Olivocochlear and vestihular efferent neurons of feline brainstem: their location, morphology and number determined by retrograde axonal transport and acet~icholinesterase hist~hemist~. J. Camp. Neurol. 161. 159-182. Wax. W.B. (1978): The olivocochlear bundle: its origins and terminations in the cat. In: Evoked Eiectrical Activity in the Auditory Nervous System, pp. 43-62. Editors: R.F. Naunton and C. Fernandez. Academic Press. New York. Warr. W.B. and Guinan, J.J. (1979): Efferent innervation of the organ of Corn: two separate systems. Brain Res. 173. 152-155. White. J.S. and Warr, W.B. (1983): The dual origins of the olivocochlear bundle in the albino rat. J. Comp. Neurol. 219. 203-214.