Projection of the Marginal Shell of theAnteroventral Cochlear Nucleus toOlivocochlear Neurons in the Cat

Y. YE, D.G. MACHADO, AND D.O. KIM*

Department of Neuroscience, Division of Otolaryngology, Department of Surgery,University of Connecticut Health Center, Farmington, Connecticut 06030-1110

ABSTRACTThe marginal shell of the anteroventral cochlear nucleus is anatomically and physiolog-

ically different from its central core. Previous studies suggest that neurons in the marginalshell are well suited to encode the intensity of acoustic stimuli. To investigate the projectionsof the marginal shell, a focal injection (,100 nl) of a mixture of biotinylated dextran amine(BDA) and 3H-leucine was made into the marginal shell of the cat combined with injection ofcholera toxin subunit-B (CTB) into the cochleas. Following a 7-day survival, the cats wereperfused. Axons and swellings labeled with BDA and olivocochlear neurons labeled with CTBwere immunocytochemically stained black and brown, respectively. 3H-leucine labels werevisualized by autoradiography. Labeled neural structures were examined via light micros-copy. We found that swellings labeled with BDA, sometimes doubly labeled with BDA and3H-leucine, were in close apposition with dendrites and/or somata of olivocochlear neuronsidentified with CTB labeling. Double labeling with BDA and 3H-leucine signifies that thelabel was anterogradely transported. The results support the conclusion that the anteroven-tral cochlear nucleus projects to medial olivocochlear neurons bilaterally and to lateralolivocochlear neurons ipsilaterally. Furthermore, the results are consistent with the inter-pretation that the marginal shell provides a source of the above-mentioned projections.Together with information in the literature, the present anatomical results support a hy-pothesis that the marginal shell provides information about stimulus intensity as a part of areflex (or feedback gain control) system comprising the cochlea, cochlear neurons, cochlearnucleus, medial olivocochlear neurons, and cochlear outer hair cells. J. Comp. Neurol. 420:127–138, 2000. © 2000 Wiley-Liss, Inc.

Indexing terms: cochlear mechanics; feedback; reflex; cholera toxin subunit B; biotinylated

dextran amine; tritiated leucine

The marginal shell of the ventral cochlear nucleus(VCN) consists of the granule cell layer, the small cell cap,and the “cell poor rind,” which encapsulate the VCN cen-tral core of large cells. The granule cell layer covers thesurface of the VCN, and the small cell cap (Osen, 1969) issubjacent to the granule cell layer. The cell-poor rind isusually located along the medial and ventral borders ofthe VCN (Liberman, 1993). The small cell cap in thehuman (Moore and Osen, 1979; Paxinos and Huang, 1995)is considerably expanded and constitutes a large portion ofthe cochlear nucleus (CN).

The marginal shell of the VCN is morphologically dif-ferent from its central core (Cant, 1993). Liberman (1991,1993) found that nearly all cat type I auditory nerve fibersinnervating the small cell cap have low spontaneous rates.In contrast, the VCN core receives a mixture of auditory

nerve fibers having high and low spontaneous rates. Col-laterals of type II auditory nerve fibers, which originatefrom the outer hair cells, often terminate in the VCNmarginal shell in rodents (Brown et al., 1988a; Brown andLedwith, 1990). Besides ascending inputs, the VCN mar-ginal shell also receives descending inputs from the supe-rior olivary complex (Shore and Moore, 1998), includingcollaterals of medial olivocochlear neurons (Brown et al.,

Grant sponsor: NIDCD, NIH; Grant number: DC00360.*Correspondence to D.O. Kim, Department of Neuroscience, University

of Connecticut Health Center, Farmington, CT 06030-1110.E-mail: kim@neuron.uchc.edu

Received 10 August 1999; Revised 30 December 1999; Accepted 30 De-cember 1999

THE JOURNAL OF COMPARATIVE NEUROLOGY 420:127–138 (2000)

© 2000 WILEY-LISS, INC.

1988b; Benson and Brown, 1990), and from the auditorymidbrain (Shore and Moore, 1998) and cortex (Feliciano etal., 1995; Weedman and Ryugo, 1996). There is little evi-dence for termination of medial olivocochlear collaterals inthe VCN central core. The VCN marginal shell also re-ceives vestibular (Burian and Gstoettner, 1988; Kevetterand Perachio, 1989; Zhao et al., 1995) and somatosensory(Itoh et al., 1987; Weinberg and Rustioni, 1987; Wrightand Ryugo, 1996) inputs.

Physiological characteristics of the marginal shell of theanteroventral cochlear nucleus (AVCN) are likewise dif-ferent from those of the AVCN core. A large majority ofmarginal-shell neurons exhibit low spontaneous rates andwide dynamic ranges (as wide as 89 decibels), whereas alarge majority of central-core units exhibit narrow dy-namic ranges (Ghoshal and Kim, 1996a; 1997). Some neu-rons of the marginal shell are not driven by sounds,whereas essentially all neurons of the central core are(Ghoshal and Kim, 1996b). Because auditory nerve fiberswith low spontaneous rates have wider dynamic rangesthan their high spontaneous-rate counterparts (Sachs andAbbas, 1974; Evans and Palmer, 1980), the afferent inputnearly exclusively from low spontaneous-rate nerve fibersto the marginal shell (Liberman, 1991) is thus expected toenhance encoding of the acoustic stimulus intensity in thisregion (Kim et al., 1995). Moreover, the input of medialolivocochlear neurons’ collaterals to the marginal shell isexpected to counteract the suppressive effects of the me-dial olivocochlear neurons on the output of the cochlearamplifier (Davis, 1983; Neely and Kim, 1983), therebyhelping marginal-shell neurons to represent accuratelythe absolute stimulus intensity (Brown et al., 1988b; Kimet al., 1995; 1998). In brief, the unique afferent and effer-ent inputs to the VCN marginal shell should make thisregion ideally suited to encode intensity of the acousticstimulus.

Although the descending portion of the cochlear feed-back gain-control circuit, from medial olivocochlear neu-rons to the cochlea, has been well documented (see, e.g.,Warr, 1992), the ascending portion of the circuit, from theCN to medial olivocochlear neurons, is poorly understood.The goal of this study is to evaluate the hypothesis thatthe AVCN marginal shell projects to medial olivocochlearneurons, thereby conveying stimulus intensity information.

MATERIALS AND METHODS

General procedures

This study used four laboratory cats, which were pig-mented, female, of body weight 2.0–3.5 kg, and with cleanears and no history of diseases or hereditary abnormali-ties. The cats were purchased from a commercial breederof research cats. The procedures used in this study fol-lowed the Public Health Service Policy on Humane Careand Use of Laboratory Animals and were approved by theAnimal Care Committee, University of Connecticut HeathCenter. Each cat was initially anesthetized with an intra-peritoneal injection of sodium pentobarbital, 35 mg/kg. Anintravenous (IV) catheter and an intratracheal tube wereinstalled. Deep anesthesia was maintained by administer-ing sodium pentobarbital through the IV catheter or byapplying halothane, 0.1–0.6%, mixed with nitrous oxideand oxygen, through the intratracheal tube. Animal bodytemperature was maintained near 38°C.

A craniotomy was performed on the left posterior fossa,and a portion of the cerebellum overlying the left CN wasaspirated. Using surface landmarks of the AVCN, a mi-cropipette was guided toward the AVCN using a micro-drive. Insofar as the target injection site was the marginalshell of the AVCN, a shallow structure compared to theAVCN central core, it was important to ascertain accu-rately when the electrode tip first touched the surface ofthe AVCN. For this purpose, we applied a swept-tonestimulus and monitored the voltage signal picked up bythe electrode (the same micropipette containing tracerswith an inner diameter of about 25 mm) as it graduallyadvanced toward the AVCN surface. The appearance ofdetectable evoked neural activities of multiunits signifiedthe initial contact of the electrode with the AVCN surface.We advanced the electrode a short distance (100–200 mm)beyond that point. Recording multiunit responses alsoallowed us to determine the characteristic frequency of theinjection site in each of the cats, except for one as a resultof equipment malfunction. A small amount (,0.1 ml) of amixture of biotinylated dextran amine (BDA) (MolecularProbes Inc., Eugene, OR), 10%, 10,000 MW, and 3H-leucine (L-leucine- 4,5,-3H; ICN Biomedicals, Inc., Irvine,CA), 50 mCi/ml, was injected into the marginal shell of theAVCN using a pressure injector (Picospritzer; GeneralValve Corp., East Hanover, NJ), 20–100 psi, 40–200 msecduration, several pulses. Injection of a mixture of BDAand 3H-leucine was used previously by Warr and Beck(1996).

The bulla was opened, and cholera toxin subunit-B(CTB; List Biological Laboratories, Inc., Campbell, CA),8–10 ml, 1%, was injected into the cochlear scala tympanithrough a micropipette (inner diameter about 30 mm),which impaled the round window membrane. Injection ofCTB was made into both cochleas of each cat using thepressure injector described above, 25 psi, 50 msec pulseduration, over a period of about 20 minutes for each co-chlea. Upon withdrawal of the micropipette, the roundwindow membrane sealed by itself without showing anydetectable fluid leakage. The experimental approach ofthis study is summarized in Figure 1.

Immunohistological processing

After a survival period of 7 days, each cat was anesthe-tized as described above and perfused transcardially, ini-tially with saline and subsequently with a solution con-taining 4% paraformaldehyde, 0.1% glutaraldehyde, and0.1 M sodium phosphate-buffered saline (PBS), pH 7.4, atroom temperature. The brain tissue was kept in 30% su-crose for 1 day. Frozen sections were cut at 50 mm in thetransverse plane. The procedure for processing BDA labelwas similar to that of Veenman et al. (1992). The brainsections were incubated in 0.3 % H2O2 for 15–20 minutesand rinsed in 0.1 M PBS plus 0.1% Triton X-100. Thesections were incubated in ABC solution (Vector Labora-tories, Inc., Burlingame, CA) overnight at 4°C, and rinsedin 0.1 M PBS. The sections were treated with diaminoben-zidine intensified with cobalt chloride and nickel ammo-nium sulfate (Adams, 1981). Thus, BDA-labeled struc-tures were visualized with grayish black reactionproducts.

The sections were then incubated with 5% normal horseserum (Jackson Immunoresearch Laboratories, Inc., WestGrove, PA) for 2 hours at room temperature and in aprimary antibody (goat anti-CTB, List Biological Labora-

128 Y. YE ET AL.

tories, Inc.), diluted to 1:20,000, plus 2.5% normal horseserum, 4°C, under agitation for 3–4 days. The sectionswere then rinsed and incubated with a secondary antibody(biotinylated anti-goat IgG, made in horse; Vector Labo-ratories, Inc.) diluted to 1:2,000 in 0.1 M PBS, plus 1%normal horse serum, 24–48 hours, 4°C, under agitation.The sections were incubated in ABC overnight at 4°C andtreated with diaminobenzidine without metal intensifica-tion. Thus, CTB-labeled structures were visualized withreddish brown reaction products.

Alternate sections (one half of the total set) were pro-cessed by autoradiography (Oliver, 1984) with an expo-sure time of 8 weeks at 4°C. A subset of the sections(usually three-fourths of the total set) was counterstainedwith cresyl violet, and the remainder was cleared.

Microscopic examination

A light microscope (Leitz Diaplan, Leica, Inc., Allendale,NJ) was used to identify axons and axonal swellings la-beled with BDA (grayish black) and/or 3H-leucine (silvergrains) and somata and dendrites of olivocochlear (OC)neurons labeled with CTB (reddish brown). Close apposi-tions between BDA-labeled axonal swellings and CTB-labeled dendrites/somata were examined under a 1003oil-immersion objective (Leitz PL Fluotar, NA 5 1.32). Thecriterion for acceptance as close apposition between anaxonal swelling and a dendrite/soma was that the twostructures should be in the same focal plane with novisible gap between the two under a 1003 objective. Im-ages were digitized through a color video camera con-nected to a computer. Camera lucida drawings were madefor CTB-labeled OC neurons with closely apposed BDA-labeled axons and swellings. An injection site was definedas the area identified by the presence of dense and uni-form deposits of BDA reaction product throughout theneuropil at the time of examination. A computer-interfaced microscope system (Neurolucida, Microbright-

field, Colchester, VT) was used to trace injection sites. Thesame system was also used to measure sizes of axonalswellings and thicknesses of axons (for the axonal seg-ments between swellings) contacting OC neurons under a1003 objective.

Regions containing OC neurons

By using Warr and Guinan’s (1979) criteria, OC neu-rons were divided into medial and lateral olivocochlear(MOC and LOC) neurons. We designated the locations ofOC neurons into various regions. In defining these regionsof OC neurons, we used the nomenclature of the superiorolivary complex nuclei described by Guinan et al. (1972).We found that OC neurons were located not only insidevarious cell-dense periolivary nuclei but also in cell-poorareas, which are not included in any periolivary nucleus.This difference between the nature of a periolivary nu-cleus and that of an OC region led us to introduce a newset of terminology of OC regions, shown in Figure 2 anddescribed below.

Four regions of MOC neurons were defined as follows: 1)R-MOC, a rostral SOC region, rostral to the medial supe-rior olive (MSO); it includes a rostral part of the medialnucleus of the trapezoid body and a part of the ventralnucleus of the lateral lemniscus; some of the MOC neu-rons in this region are located as lateral as LOC neurons(Adams, 1983; Warr, 1992); 2) V-MOC, a region ventral tothe MSO; it includes the ventral nucleus of the trapezoidbody and a surrounding area; the rostral end of the MSOrepresents a landmark that separates V-MOC from R-MOC; 3) VM-MOC, a region ventromedial to the MSO; itincludes most of the medial nucleus of the trapezoid bodyand the cell-poor area between the MSO and the medialnucleus of the trapezoid body; 4) DM-MOC, a region dor-somedial to the MSO; it includes the dorsomedial perioli-vary nucleus and a surrounding area.

Fig. 1. Schematic diagram of the experimental approach used inthis study. AVCN, anteroventral cochlear nucleus; BDA, biotinylateddextran amine; cereb., cerebellum; CTB, cholera toxin subunit B;3H-leucine, tritiated leucine; LOC, lateral olivocochlear neuron; LSO,

lateral superior olive; MOC, medial olivocochlear neuron; MSO, me-dial superior olive; RW, round window; 4th Vent, fourth ventricle;5ST, spinal trigeminal tract; 6, abducens nerve root; 7, facial nerveroot.

129PROJECTION OF AVCN SHELL TO OLIVOCOCHLEAR CELLS

Analogously, five regions of LOC neurons were definedas follows: 1) M-LOC, a marginal region of the lateralsuperior olive (LSO); it includes the immediate marginssurrounding the LSO and the dorsal hilus of the LSO;Warr et al. (2000) introduced the designation of a sub-population of the cat LOC neurons to be those in theM-LOC region; 2) R-LOC, a region rostral to the LSO; itincludes the anterolateral periolivary nucleus and a ros-tral part of the lateral nucleus of the trapezoid body andsurrounding areas; R-LOC lies between the rostral ends ofthe LSO and MSO; 3) VL-LOC, a region ventrolateral tothe LSO; it includes most of the lateral nucleus of thetrapezoid body and a surrounding area; 4) D-LOC, a re-gion dorsal to the LSO; it includes the dorsolateral peri-olivary nucleus and a surrounding area; the rostral end ofthe LSO represents a landmark that separates R-LOCfrom VL-LOC and D-LOC; 5) C-LOC, a region caudal tothe LSO; it includes the posterior periolivary nucleus anda surrounding area; the caudal end of the LSO representsa landmark that separates C-LOC from D-LOC andVL-LOC.

RESULTS

The BDA injection sites are shown in Figure 3. Thecenters of the injection sites were located in the dorsolat-eral part of the marginal shell of the AVCN in all of thefour animals. Nomenclature and parcellation of the CNare adapted from Brawer et al. (1974) and Liberman(1991, 1993).

Figure 4 shows color images of examples of CTB-labeled(reddish brown) dendrites and somata of OC neurons inclose apposition with axonal swellings labeled with BDA(grayish black). The axons and axonal swellings of theAVCN marginal shell neurons were well labeled by BDA.Somata and long stretches of dendrites of OC neuronswere well labeled with CTB in the present study, a resultconsistent with results of Vetter and Mugnaini (1992).BDA-labeled axons tended to be thin (mean diameter 0.5mm). Many of the axonal swellings were small (see, e.g.,Fig. 4F,J,M), whereas some were larger (see, e.g., Fig.4C,K). Some of these axonal swellings and axons weredoubly labeled with BDA and 3H-leucine (silver grains;

Fig. 2. Schematic diagram of the cat superior olivary complex(SOC) and periolivary region containing olivocochlear neurons.A–D: Four transverse sections of the SOC. C, caudal; D, dorsal;DM, dorsomedial; L, lateral; LOC, lateral olivocochlear; LSO, lat-

eral superior olive; M, marginal; MOC, medial olivocochlear; MSO,medial superior olive; R, rostral; V, ventral; VL, ventrolateral; VM,ventromedial; 7N, facial nucleus.

130 Y. YE ET AL.

see, e.g., Fig. 4A–D,G). Because autoradiographic silvergrains are present on the surface of a tissue slide, whereasneural structures (such as swellings) are generally withina tissue section, visualizing a structure and the corre-sponding silver grains required taking images at multiplefocal planes. Double labeling of axons/swellings with BDAand 3H-leucine, as shown in Figure 4, provides evidencethat the tracers were anterogradely transported, because3H-leucine is specifically transported anterogradely(Cowan et al., 1972), whereas BDA is transported bothantero- and retrogradely.

These images demonstrate that the colors (brown andblack) of the two reaction products (CTB and BDA) areclearly distinguishable. The OC cells shown in Figure

4A,E are LOC cells in the ipsilateral SOC. The rest of theOC cells are MOC cells on the contralateral side (Fig.4B–D,G–I) and on the ipsilateral side (Fig. 4F,J–M). Thelocation of each OC cell, in terms of the OC region definedin Figure 2, is indicated in the legend to Figure 4. Theseexamples provide good evidence that axons of AVCN neu-rons at the injection site make close contacts with thedendrites and somata of MOC and LOC cells.

Camera lucida drawings of the CTB-labeled MOC neu-rons and BDA-labeled axons/swellings (represented inFig. 4) in the contralateral and ipsilateral sides are shownin the left and right panels of Figure 5, respectively.Whereas the photographic images of Figure 4 capture onlythe structures that are in a particular plane of focus, the

Fig. 3. Camera lucida drawings of BDA injection sites in four cats.Each column represents one animal. The middle row represents thecenter of the injection site. The top and bottom rows represent therostral and caudal ends of the injection site, respectively. All sectionswere cut in the transverse plane at a thickness of 50 mm. The numberat the lower right corner of each panel represents the section number.The characteristic frequency (CF) of each injection site was 13.8 kHz

for Q87, unknown for Q88, 4.0 kHz for Q89, and 6.0 kHz for Q93. A,anterior part of PVCN; AA, anterior part of the anterior division ofAVCN; AN, auditory nerve; AP, posterior part of the anterior divisionof AVCN; CG, cerebellar granular cell layer; D, dorsal; DAS, dorsalacoustic stria; DCN, dorsal cochlear nucleus; G, granule cell layer; L,lateral; PD, posterodorsal part of the AVCN; PV, posteroventral partof the AVCN; R, cell-poor rind; S, small cell cap.

131PROJECTION OF AVCN SHELL TO OLIVOCOCHLEAR CELLS

Fig. 4. Examples of images of axons (arrows) and axonal swellings(arrowheads) that were labeled with BDA (black) and in close appo-sition with dendrites or somata (open arrows) of OC neurons retro-gradely labeled with CTB (reddish brown). Silver grains, which weretaken at different focal planes, represent 3H-leucine labels. All imageswere taken with a 1003 oil-objective lens, and dorsal is upward unlessotherwise indicated. A1,A2: R-LOC, ipsi. Thin lines in A1 and A2 (andother panels) indicate that the image is a montage constructed with

parts taken at multiple focal planes. B1,B2: VM-MOC, contra., ro-tated 45° clockwise. C1,C2: VM-MOC, contra. D1–D3: VM-MOC,contra. E1–E3: R-LOC, ipsi. F1,F2: V-MOC, ipsi. G1,G2: VM-MOC,contra. H: R-MOC, contra., rotated 90° clockwise. I: V-MOC, contra. J:V-MOC, ipsi., rotated 55° counterclockwise. K: V-MOC, ipsi., rotated90° clockwise. L: VM-MOC, ipsi., rotated 30° clockwise. M: MOCneuron’s dendrite, R-MOC, rotated 35° clockwise.

camera lucida drawings represent structures in multiplefocal planes superimposed and thus demonstrate moreparts of each cell. BDA-labeled swellings were found incontact with proximal dendrites (see, e.g., Fig. 5E,H) dis-tal dendrites (see, e.g., Fig. 5A,K) or somata (see, e.g., Fig.5B,G) of MOC cells. A series of BDA-labeled axonal swell-ings connected by an axon, resembling stringed beads,was often observed traveling together and maintainingclose contacts with an MOC neuron’s dendrite (see, e.g.,Fig. 5C,D,K).

Camera lucida drawings of CTB-labeled LOC neuronsand BDA-labeled axons/swellings on the ipsilateral sideare shown in Figure 6. Two of the LOC neurons, shown inFigure 6A,C, are represented in Figure 4A,E, respectively.The LOC neurons tended to be smaller than the MOCneurons, and a group of LOC neurons were sometimesclustered together (see, e.g., Fig. 6D). The contacts ofBDA-labeled axonal swellings on the LOC neurons tendedto be on the somata (Fig. 6, all five neurons), althoughcontacts of axonal swellings on both dendrites and somatawere also observed (Fig. 6C). In the case of the neuronshown in Figure 6C (also Fig. 4E), an axon collateral(arrow, Fig. 6C) was seen coming off a parent axon andcontacting the soma.

The axonal swellings contacting OC neurons have var-ious sizes as noted above. Their distributions are shown inFigure 7A,B. The results in this figure represent swellingsassociated with a random sample of 11 MOC cells (thesame as those shown in Figs. 4, 5) and 11 LOC cells

Fig. 5. A–K: Camera lucida drawings of MOC neurons in closeapposition with BDA-labeled axonal swellings. The left and rightgroups represent MOC neurons in the contralateral and ipsilateral

SOC, respectively. These MOC neurons are also shown in color pic-tures in Figure 4. The locations of the neurons are given in the legendof Figure 4.

Fig. 6. Camera lucida drawings of LOC neurons, in the ipsilat-eral SOC, in close apposition with BDA-labeled axonal swellings.Two of these LOC neurons are shown as color pictures in Figure 4.The locations of the neurons are as follows. A: R-LOC. B: M-LOC.C: R-LOC. D: M-LOC. E: M-LOC.

133PROJECTION OF AVCN SHELL TO OLIVOCOCHLEAR CELLS

(including the cells shown in Fig. 6). The swellings incontact with MOC neurons (mean 5.2 mm2) were signifi-cantly (P , 0.01, analysis of variance) larger than those incontact with LOC neurons (mean 3.3 mm2). The differencein the size of the swellings suggests 1) that there may bea difference in functional characteristics of the two groupsof putative synapses and/or 2) that there may be twodifferent cell types in the AVCN marginal shell projectingto MOC or LOC neurons separately.

Distributions of BDA-labeled axonal thicknesses con-tacting MOC and LOC neurons are shown in Figure 7C,D.It is not known whether these axons are myelinated. Thethickness of a BDA-labeled portion of an axon is expectedto correspond to the axoplasm excluding any myelinsheath that may be present. The axons associated withMOC and LOC neurons had similar thicknesses withmeans of 0.54 and 0.52 mm, respectively (P . 0.05, anal-ysis of variance). The fact that these axons are thin sug-gests that they may be lightly myelinated or unmyeli-nated and that the axons transmit signals slowly, at leastover their termination zones where the thicknesses weremeasured.

Figure 8 shows distributions of OC neurons, which arein close apposition with BDA-labeled axonal swellings, in

Fig. 7. Distributions of areas (A,B) and thicknesses (C,D) of BDA-labeled axonal swellings contacting MOC and LOC neurons.

Fig. 8. Distribution of MOC and LOC neurons, which are in closeapposition with BDA-labeled axonal swellings, in various OC regions.C, caudal; D, dorsal; DM, dorsomedial; M, marginal; R, rostral; V,ventral; VL, ventrolateral; VM, ventromedial. This figure represents116 OC neurons from three cats.

134 Y. YE ET AL.

various OC regions. The largest component (52%) wascontalateral MOC neurons. The next large component(29%) was ipsilateral MOC neurons. A relatively smallproportion (19%) corresponded to ipsilateral LOC neu-rons. The numbers of these OC neurons in various OCregions are listed in Table 1. The findings of the presentstudy are summarized schematically in Figure 9A.

DISCUSSION

Effective and virtual injection sites

An effective injection site is defined to be “the volume oftissue which has sustained the uptake and subsequenttransport of a tracer”, whereas a virtual injection site isthe volume of tissue “which is identified by the presence ofdense and uniform deposits of reaction product through-out the neuropil at the time of microscopic examination”(Mesulam, 1982, p 63–64). In results obtained with horse-radish peroxidase as a tracer, a virtual injection site wasobserved to expand between 2 and 8 hours and contractbeyond 18 hours following injection (Hedreen andMcGrath, 1977; Vanegas et al., 1978; Mesulam, 1982).Consequently, an effective injection site can be substan-tially different from a virtual injection site.

Although we are not aware of a corresponding study ofBDA injection sites, it is possible that analogous phenom-ena may exist for BDA injection sites. The injection sitesdescribed above correspond to virtual injection sites. It ispossible that the sizes of the effective injection sites maybe underestimated by the virtual injection sites in thepresent study considering that a long (7-day) survivalperiod was used. The reason for our choice of the 7-daysurvival period was based on a relatively slow axonaltransport of BDA, 2.2 mm/day (Oliver and Beckius, 1993).Thus, it is possible that the effective injection sites of thepresent results might have encroached into the centralcore of the AVCN and that the axonal contacts with OCneurons described in this paper may partly, or even en-tirely, represent axons originating from the AVCN centralcore.

In a study conducted in our laboratory, it was observedthat axons labeled with BDA injected into the marginalshell and those with BDA injected into the central core ofthe AVCN were markedly different. Axons associated withinjection into the core were thicker (more than half were

thicker than 1.5 mm) and projected to both the principalnuclei (i.e., lateral and medial superior olives) and peri-olivary areas. In contrast, those associated with injectioninto the shell were thinner (92% were thinner than 1.5mm) and projected nearly exclusively to periolivary areaswith little projection to the principal nuclei (Ye and Kim,1998). The thicknesses and spatial distributions (withinthe superior olivary complex) of BDA-labeled axons of thepresent study, where the center of the virtual injectionsite was within the marginal shell of the AVCN in all fouranimals, closely correspond to those of the latter group ofaxons described above. These observations support theinterpretation that the present results represent, to alarge extent, projections from the AVCN marginal shell toOC neurons. Results obtained with injection of BDA and3H-leucine into the AVCN central core combined withinjection of CTB into the cochlea are presently unavail-able. A future study of such a nature should be valuable inaddressing the questions of whether both the marginalshell and central core of the AVCN project to OC neuronsand what differences may exist between the projectionsfrom the shell and core of the AVCN to OC neurons.

Main finding

This study provides information about the ascendingportion of the MOC and LOC reflex (or feedback) circuits.Together with the physiology of AVCN marginal-shellneurons (Ghoshal and Kim, 1996a, 1997) and afferentinputs to the marginal shell (Liberman, 1991, Liberman,1993), the present anatomical results support a hypothe-sis that the AVCN marginal shell provides informationabout stimulus intensity to MOC neurons as a part of afeedback gain-control system consisting of the cochlea,cochlear neurons, cochlear nucleus, MOC neurons, andcochlear outer hair cells. A projection from the AVCNmarginal shell to MOC neurons fulfils a requirement ofthe above hypothesis. The projection of the AVCN mar-ginal shell to LOC neurons has come as an unexpectedfinding. This was not anticipated, because little experi-mental or theoretical insight is available regarding thefunction of the LOC reflex circuit.

Ascending projections to OC neurons

A few previous studies have described ascending projec-tions of the CN to OC neurons. Robertson and Winter(1988) observed that the VCN projected to MOC neuronsin the guinea pig. Because of a technical limitation of theirmethod, they could not determine whether the VCN alsoprojected to LOC neurons. Thompson and Thompson(1991) found, also in the guinea pig, that the posteroven-tral CN projected to MOC and LOC neurons. Kim et al.(1995) reported preliminary results obtained with injec-tion of BDA into the AVCN marginal shell of the catcombined with horseradish peroxidase injection into thefloor of the fourth ventricle; BDA-labeled axonal swellingswere observed in close apposition with OC neurons. Thepresent paper is an extension of that of Kim et al. (1995).The present and earlier studies agree that the VCNprojects to MOC and LOC neurons. The Robertson andWinter and the Thompson and Thompson studies usedlarger injections into the VCN, and so information aboutthe marginal shell was not provided separately from thatabout the central core. In contrast, the present study usedsmaller injection sites centered in the AVCN marginalshell, providing further information about the source of

TABLE 1. Number of Olivocochlear (OC) Neurons in Close Apposition WithAxonal Swellings Labeled With Biotinylated Dextran Amine (BDA) Injected

Into AVCN Marginal Shell1

Contra./ipsi. Region

Cat no.

Q872 Q88 Q89 Q93

Contra. R-MOC 10 1 3V-MOC 23 4VM-MOC 7 19 2DM-MOC 1 1

Ipsi. R-MOC 2V-MOC 8 15 3VM-MOC 4 2DM-MOC

Ipsi. M-LOC 4 1R-LOC 1 1VL-LOC 3 3D-LOC 3 2C-LOC 4

1See Figure 2 legend and Materials and Methods for definitions of the locations.2Data for cat Q87 are not complete because of partial loss of the tissue.

135PROJECTION OF AVCN SHELL TO OLIVOCOCHLEAR CELLS

the observed projection. The present study also extendsinformation to another subdivision of the CN, i.e., fromposteroventral CN to anteroventral CN, and another spe-cies, i.e., from guinea pig to cat.

The spatial distributions of OC neurons contacted byBDA-labeled axonal swellings were different among theanimals of the present study, particularly for MOC neu-rons on the contralateral side (Table 1). For example, theMOC neurons on the contralateral side tended to be con-centrated in V-MOC and VM-MOC for cats Q88 and Q89,respectively. The injection site of Q89 was more posteriorto that of Q88. The characteristic frequency of Q89 was 4kHz and that of Q88 was unknown. There may be a spatialrelationship between the location of AVCN marginal-shellneurons and the OC neurons to which the marginal-shell

neurons project. Further studies are needed to addressthis question.

Cell types of the AVCN marginal shell

The marginal shell of the AVCN has been described tocontain multiple cell types: granule cells, Golgi cells,unipolar brush cells, chestnut cells, mitt cells, elongatecells, and stellate (or multipolar) cells of small, medium,and large sizes (Cant, 1992, 1993; Weedman et al., 1996;Morest, 1997; Doucet and Ryugo, 1997; Ferragamo et al.,1998). It is not clear whether each of the above namesuniquely corresponds to a distinct cell type or whethersome of the multiple names correspond to a common celltype. Based on the results of the present study, it is notknown which of these cell types in the marginal shell

Fig. 9. A: Schematic diagram of the findings of the present study,illustrating ascending projections from the AVCN marginal shell toOC neurons. B: Descending projections from OC neurons to the leftcochlea, adapted from Warren and Liberman (1989), with permission

from Elsevier Science. AVCN, anteroventral cochlear nucleus; cereb.,cerebellum; LOC, lateral olivocochlear neuron; LSO, lateral superiorolive; MOC, medial olivocochlear neuron; MSO, medial superior olive;4th Vent., fourth ventricle.

136 Y. YE ET AL.

project to OC neurons. Future studies with injections of aretrograde tracer into a region containing OC neurons(e.g., VM- and V-MOC, VL- and D-LOC) should help toaddress this question.

Comparison of ascending and descendingprojections involving OC neurons

The components of the ascending projections of theAVCN marginal shell to OC neurons (Fig. 9A) are com-pared to the components of the descending projections ofOC neurons to the cochlea in Figure 9B (adapted fromWarren and Liberman, 1989). The descending projectionsto one cochlea include four components representing thetwo types of OC neurons on both sides of the brainstem. Incontrast, the ascending projections from one AVCN mar-ginal shell include only three components representingthose to MOC neurons on both sides and to LOC neuronson the ipsilateral side; the results of Thompson andThompson (1991) are consistent with this. Consequently,LOC neurons on the right side do not receive ascendingprojections from the left AVCN marginal shell but they doproject to the left cochlea. The neural circuit implies thatLOC neurons, in part, receive ascending signals from onecochlea but send descending signals to both cochleas. It isinteresting to note that the contralateral to ipsilateralratio for the ascending projections of the AVCN marginalshell to MOC neurons (1.8, representing 52%/29%; Fig.9A) is comparable to the corresponding ratio of descendingprojections from MOC neurons to the cochlea (2.4, repre-senting 26%/11%; Fig. 9B). The present results were ob-tained with injections of CTB into both cochleas in eachanimal in order to maximize the number of labeled OCneurons. Consequently, the results do not provide infor-mation about to which of the cochlea the labeled OC neu-rons project.

Labeling with3H-leucine

The injection sites of 3H-leucine (not shown) were largerthan those of BDA. Thus, axons labeled with 3H-leucinemay have originated from an area larger than the BDAinjection sites. Consistently with this, there were manyaxons that were labeled only with 3H-leucine and not withBDA. Because 3H-leucine was used in combination withBDA, however, axons and swellings doubly labeled by bothtracers were interpreted to have originated from BDAinjection sites.

As was described for Figure 4 above, some axons weredoubly labeled with BDA and 3H-leucine. Other axons,however, were labeled with BDA only. One potential rea-son for the absence of 3H-leucine label is that the axonsmay have been located too far away from the surface of thetissue section, such that the radioactive emission from 3Hwas too weak to produce silver grains in the emulsion onthe tissue surface. Another potential reason why someaxons were labeled with BDA only is that the BDA label insome cases may be retrogradely transported. Consistentlywith this, we did observe that some somata in the SOCwere labeled with BDA. However, the numbers of suchBDA-labeled somata was quite small. We believe that thefirst reason indicated above (i.e., axons being away fromthe surface) accounts for most of the axons being labeledwith BDA only in the present material.

Future studies

Conducting a study similar to the present one withinjection of a retrograde tracer into one cochlea (someipsilaterally and others contralaterally) would be desir-able to obtain further information about the reflex circuitsinvolving the two cochleas. Electron microscopic examina-tion of synaptic ultrastructures of the contacts betweenAVCN marginal shell and OC neurons, demonstrated onlyby light microscopy in this study, would also be desirablein future studies. Such electron microscopic studies plusneurotransmitter-related immunocytochemical studiesshould help to elucidate whether these contacts consist ofexcitatory or inhibitory synapses and which neurotrans-mitters are involved in this projection.

ACKNOWLEDGMENTS

We thank Drs. D.L. Oliver, E.M. Josephson, and D.K.Morest for help in anatomical methods; Dr. W.B. Warr forcomments on an early version of the manuscript; and Drs.A.M. Thompson and N.B. Cant for discussion.

LITERATURE CITED

Adams JC. 1981. Heavy metal intensification of DAB-based HRP reactionproduct. J Histochem Cytochem 29:775.

Adams JC. 1983. Cytology of periolivary cells and the organization of theirprojections in the cat. J Comp Neurol 215:275–289.

Benson TE, Brown MC. 1990. Synapses formed by olivocochlear axonbranches in the mouse cochlear nucleus. J Comp Neurol 295:52–70.

Brawer JR, Morest DK, Kane EC. 1974. The neuronal architecture of thecochlear nucleus of the cat. J Comp Neurol 155:251–300.

Brown MC, Ledwith JV. 1990. Projections of thin (type-II) and thick(type-I) auditory-nerve fibers into the cochlear nucleus of the mouse.Hearing Res 49:105–118.

Brown MC, Berglund AM, Kiang NYS, Ryugo DK. 1988a. Central trajec-tories of type II spiral ganglion neurons. J Comp Neurol 278:581–590.

Brown MC, Liberman MC, Benson TE, Ryugo DK. 1988b. Brainstembranches from olivocochlear axons in cats and rodents. J Comp Neurol278:591–603.

Burian M, Gstoettner W. 1988. Projection of primary vestibular afferentfibres to the cochlear nucleus of the guinea pig. Neurosci Lett 84:13–17.

Cant NB. 1992. The cochlear nucleus: neuronal types and their synapticorganization. In: Webster DB, Popper AN, Fay RR, editors. The mam-malian auditory pathway: neuroanatomy. New York: Springer-Verlag.p 66–116.

Cant NB. 1993. The synaptic organization of the ventral cochlear nucleusof the cat: the peripheral cap of small cells. In: Merchan MA, Juiz JM,Godfrey DA, Mugnaini E, editors. The mammalian cochlear nuclei:organization and function. New York: Plenum. p 91–105.

Cowan WM, Gottlieb DI, Hendrickson AE, Price JL, Woolsey TA. 1972. Theautoradiographic demonstration of axonal connections in the centralnervous system. Brain Res 37:21–51.

Davis H. 1983. An active process in cochlear mechanics. Hearing Res9:79–90.

Doucet JR, Ryugo DK. 1997. Projections from the ventral cochlear nucleusto the dorsal cochlear nucleus in rats. J Comp Neurol 385:245–264.

Evans EF, Palmer AR. 1980. Relationship between the dynamic range ofcochlear nerve fibres and their spontaneous activity. Exp Brain Res40:115–118.

Feliciano M, Saldana E, Mugnaini E. 1995. Direct projections from the ratprimary auditory neocortex to nucleus sagulum, paralemniscal regions,superior olivary complex and cochlear nuclei. Auditory Neurosci 1:287–308.

Ferragamo MJ, Golding NL, Gardner SM, Oertel D. 1998. Golgi cells in thesuperficial granule cell domain overlying the ventral cochlear nucleus:morphology and electrophysiology in slices. J Comp Neurol 400:519–528.

Ghoshal S, Kim DO. 1996a. Marginal shell of the anteroventral cochlear

137PROJECTION OF AVCN SHELL TO OLIVOCOCHLEAR CELLS

nucleus: intensity coding in single units of the unanesthetized decere-brate cat. Neurosci Lett 205:71–74.

Ghoshal S, Kim DO. 1996b. Marginal shell of the anteroventral cochlearnucleus: acoustically weakly-driven and not-driven units in the un-anesthetized decerebrate cat. Acta Otolaryngol 116:280–283.

Ghoshal S, Kim DO. 1997. Marginal shell of the anteroventral cochlearnucleus: single-unit response properties in the unanesthetized decere-brate cat. J Neurophysiol 77:2083–2097.

Guinan JJ Jr, Norris BE, Guinan SS. 1972. Single auditory units in thesuperior olivary complex II: locations of unit categories and tonotopicorganization. Int J Neurosci 4:147–166.

Hedreen JC, McGrath S. 1977. Observations on labeling of neuronal cellbodies, axons and terminals after injection of horseradish peroxidaseinto rat brain. J Comp Neurol 176:225–246.

Itoh K, Kamiya H, Mitani A, Yasui Y, Takada M, Mizuno N. 1987. Directprojections from the dorsal column nuclei and spinal trigeminal nucleito the cochlear nuclei in the cat. Brain Res 400:145–150.

Kevetter GA, Perachio AA. 1989. Projections from the sacculus to thecochlear nuclei in the Mongolian gerbil. Brain Behav Evol 34:193–200.

Kim DO, Parham K, Zhao HB, Ghoshal S. 1995. The olivocochlear feedbackgain control subsystem: ascending input from the small cell cap of thecochlear nucleus? In: Flock A, Ottoson D, Ulfendahl M, editors. Activehearing. New York: Elsevier. p 31–51.

Kim DO, Ghoshal S, Ye Y. 1998. Integration of ascending and descendingsignals representing stimulus intensity in the marginal shell of theanteroventral cochlear nucleus. In: Palmer AR, Reese A, SummerfieldAQ, Meddis R, editors. Psychophysical and physiological advances inhearing. London: Whurr Publishers Ltd. p 195–203.

Liberman MC. 1991. Central projections of auditory-nerve fibers of differ-ing spontaneous rate. I. Anteroventral cochlear nucleus. J Comp Neu-rol 313:240–258.

Liberman MC. 1993. Central projections of auditory nerve fibers of differ-ing spontaneous rate. II. posteroventral and dorsal cochlear nucleus.J Comp Neurol 327:17–36.

Mesulam MM. 1982. Principles of horseradish peroxidase neurohistochem-istry and their applications for tracing neural pathways. In: MesulamMM., editor. Tracing neural connections with horseradish peroxidase.New York: John Wiley and Sons. p 1–151.

Moore JK, Osen KK. 1979. The cochlear nuclei in man. Am J Anat 154:393–418.

Morest DK. 1997. Structural basis for signal processing: challenge of thesynaptic nests. In: Sika J, editors. Acoustical signal processing in thecentral auditory system. New York: Plenum. p 19–32.

Neely ST, Kim DO. 1983. An active cochlear model showing sharp tuningand high sensitivity. Hearing Res 9:123–130.

Oliver DL. 1984. Dorsal cochlear nucleus projections to the inferior collicu-lus in the cat: a light and electron microscopic study. J Comp Neurol224:155–172.

Oliver DL, Beckius GE. 1993. Ascending projections from the cochlearnucleus to the inferior colliculus and their interactions with projectionsfrom the superior olivary comples. In: Merchan MA, Juiz JM, GodfreyDA, Mugnaini E, editors. The mammalian cochlear nuclei: organizationand function. New York: Plenum Press. p 335–347.

Osen KK. 1969. Cytoarchitecture of the cochlear nuclei of the cat. J CompNeurol 136:453–484.

Paxinos G, Huang XF. 1995. Atlas of the human brainstem. New York:Academic Press.

Robertson D, Winter I.M. 1988. Cochlear nucleus inputs to olivocochlearneurones revealed by combined anterograde and retrograde labellingin the guinea pig. Brain Res 462:47–55.

Sachs MB, Abbas PJ. 1974. Rate versus level functions for auditory-nervefibers in cats: tone-burst stimuli. J Acoust Soc Am 56:1835–1847.

Shore SE, Moore JK. 1998. Sources of input to the cochlear granule cellregion in the guinea pig. Hearing Res 116:33–42.

Thompson AM, Thompson GC. 1991. Posteroventral cochlear nucleus pro-jections to olivocochlear neurons. J Comp Neurol 303:267–285.

Vanegas H, Hollander H, Distel H. 1978. Early stages of uptake andtransport of horseradish peroxidase by cortical structures, and its usefor the study of local neurons and their processes. J Comp Neurol177:193–212.

Veenman CL, Reiner A, Honig MG. 1992. Biotinylated dextran amine as ananterograde tracer for single- and double-labeling studies. J NeurosciMethods 41:239–254.

Vetter DE, Mugnaini E. 1992. Distribution and dendritic features of threegroups of rat olivo-cochlear neurons: a study with two retrograde chol-era toxin tracers. Anat Embryol 185:1–16.

Warr WB. 1992. Organization of olivocochlear efferent systems in mam-mals. In: Webster DB, Popper AN, Fay RR, editors. The mammalianauditory pathway: neuroanatomy. New York: Springer-Verlag. p 410–448.

Warr WB, Beck JE. 1996. Multiple projections from the ventral nucleus ofthe trapezoid body in the rat. Hearing Res 93:83–101.

Warr WB, Guinan JJ Jr. 1979. Efferent innervation of the organ of Corti:two separate systems. Brain Res 173:152–155.

Warr WB, Ye Y, Boche JE, Kim DO. 2000. Olivocochlear neurons in the cat:possible homologues of rodent shell and intrinsic neurons revealed byretrograde transport of cholera toxin-B. Assoc Res Otolaryngol MeetingAbstr Vol 23.

Warren EH, Liberman MC. 1989. Effects of contralateral sound onauditory-nerve responses. I. Contributions of cochlear efferents. Hear-ing Res 37:89–104.

Weedman DL, Ryugo DK. 1996. Projections from auditory cortex to thecochlear nucleus in rats: synapses on granule cell dendrites. J CompNeurol 371:311–324.

Weedman DL, Pongstaporn T, Ryugo DK. 1996. Ultrastructural study ofthe granule cell domain of the cochlear nucleus in rats: mossy fiberendings and their targets. J Comp Neurol 369:345–360.

Weinberg RJ, Rustioni A. 1987. A cuneocochlear pathway in the rat.Neuroscience 20:209–219.

Wright DD, Ryugo DK. 1996. Mossy fiber projections from the cuneatenucleus to the cochlear nucleus. J Comp Neurol 365:159–172.

Ye Y, Kim DO. 1998. Projections from the cat AVCN marginal shell andcentral core to the superior olivary complex: axonal morphology. AssocRes Otolaryngol Meeting Abstr Vol 21. p 90.

Zhao HB, Parham K, Goshal S, Kim DO. 1995. Small neurons in thevestibular nerve root project to the marginal shell of the anteroventralcochlear nucleus in the cat. Brain Res 700:295–298.

138 Y. YE ET AL.