THE JOURNAL OF COMPARATIVE NEUROLOGY 371521-632 (1996)

Morphological Subclasses of Lateral

Olivocochlear Terminals?
Ultrastructural Analysis of Inner Spiral
Bundle in Cat and Guinea Pig
MITSUAKI SATAKE AND M. CHARLES LIBERMAN
Department of Otology and Laryngology,Harvard Medical School, Boston, Massachusetts 02115
(M.S.,M.C.L.);Eaton-Peabody Laboratory ofAuditory Physiology,Massachusetts Eye and Ear
Infirmary, Boston, Massachusetts 02114 (M.S., M.C.L.);Department of Health Sciences and
Technology,Massachusetts Institute of Technology, Boston, Massachusetts 02 115 (M.C.L.);
Department of Otolaryngology,Tohoku University School of Medicine, Sendai 980, Japan (M.S.)

ABSTRACT
The lateral olivocochlear efferent pathway terminates in vesicle-filled swellings in the inner
spiral bundles under inner hair cells (IHCs) and has been suggested to include at least two
chemically distinct subclasses (see, e.g., Vetter et al. 119911 Synapse 72-43]. In the present
study, the ultrastructure and peripheral targets of vesicle-filled swellings in the IHC area of the
cat and guinea pig cochleas were quantitatively analyzed to determine 1)whether morphologi-
cal subclasses could be defined based on swelling size or on the density, size, or shape of clear
and dense-cored vesicles and 2) whether swellings with different postsynaptic targets differed
morphologically. In both cat and guinea pig, all swellings contained large, round, clear vesicles
and a variable number of dense-core vesicles. Although evidence of clear-cut subclasses was not
compelling, the smallest swellings tended to be rich in dense-core and poor in clear vesicles and
rarely formed synaptic contacts. Most of the larger swellings, which tended to contain few
dense-core vesicles and a rich complement of clear round vesicles, formed synapses with radial
afferent fibers. However, there were no morphological differences between swellings contacting
afferents originating on the modiolar vs. pillar sides of the IHC (the source of afferents with low
and high spontaneous discharge rates, respectively). We conclude that 1)if distinct y-amino
butyric acid (GABA)ergicand cholinergic subclasses of lateral olivocochlear (LOC) fibers exist,
then the vesicle morphology of their terminals does not differ as it does in the central nervous
system and that 2) if peptide neurotransmitters, such as calcitonin gene-related peptide and
enkephalins, are packaged in dense-core vesicles, then the LOC terminals synapsing with IHC
afferent fibers are not particularly rich in these peptides. o 1996 Wiley-Liss, Inc.

Indexing terms: cochlea, efferent, synaptic vesicles, morphometry

The lateral olivocochlear (LOC) system is made up of
approximately 1,500 neurons on each side of the brainstem
that originate in the region of the lateral superior olive and
that project via unmyelinated axons predominately to the
ipsilateral cochlea (Warr, 1975; Warr and Guinan, 1979;
Guinan et al., 1983). Within the cochlea (Fig. 11, each fiber
enters the longitudinally oriented inner spiral bundle (ISB)
and tunnel spiral bundle (TSB) under the inner hair cells
(IHCs). While traveling in these bundles, each LOC axon
produces numerous en passant swellings and short branches
ending in terminal swellings (Smith, 1961; Liberman,
1980b; Brown, 1987a). It is estimated that, on average, each
LOC fiber gives rise to roughly 200 such swellings along its
course (Brown, 1987a). The major peripheral targets of
these LOC terminals are the unmyelinated segments of the
afferent radial fibers (RFs) in the region immediately below
their synaptic contacts with IHCs (Smith, 1961; Liberman,
1980b); however, synaptic contacts between apparent LOC
terminals and IHCs as well as other efferent-like profiles in
the IHC area have been reported (Liberman, 1980b).
The functional significance of this profuse innervation of
the IHC area is unknown at present (see, e.g., Liberman,
1990); however, there is anatomical evidence suggesting
Accepted March 22,1996
Address reprint requests to M. Charles Liberman, Eaton-Peabody Labora-
tory, Massachusetts Eye and Ear Infirmary, 243 Charles Street, Boston, MA
02114.

O 1996 WILEY-LISS, INC.

622
Fig. 1. Schematic illustration of the position of the inner spiral
bundle (ISB) and tunnel spiral bundle (TSB) with respect to the inner
hair cell (IHC). Three afferent fibers are illustrated; a single IHC is
actually contacted by 10-30 different radial fibers (RFs). Two of the
that the innervation of RFs is related somehow to the
functional differences among RFs according to spontaneous
discharge rate (SR). Ultrastructural studies suggest that,
although virtually all RFs receive synapses from LOC
terminals, those RFs with high thresholds and low SRs
(which are found exclusively on the modiolar side of the
IHC; Fig. 1) are much more heavily innervated than the
high-SR, low-threshold RFs found on the pillar side of the
IHCs (Liberman et al., 1990).
A large amount of immunocytochemical literature on the
olivocochlear (OC) system suggests that LOC neurons can
contain a number of neuroactive substances, including acetyl-
choline (ACh), y-amino butyric acid (GABA), calcitonin
gene-related peptide (CGRP), enkephalins, and dynorphins
as well as dopamine (for review, see Eybalin, 1993). Al-
though there are few true co-localization studies, a number
of pieces of evidence suggest that, at least in rodents, there
may be two fundamentally different subpopulations within
the LOC system (Schwartz and Ryan, 1986; Ryan et al.,
1987): one cholinergic subsystem co-containing ACh, CGRP,
and enkephalins (Altschuler et al., 1983; Vetter et al., 1991;
Saffeidine and Eybalin, 1992) and a separate GABAergic
system (Fex and Altschuler, 1984; Eybalin and Pujol, 1987;
Eybalin et al., 1988; Eybalin and Altschuler, 1990; Vetter et
al., 1991). If this is true, then it should be possible to
identify the LOC terminals of these two major subgroups in
routine electron micrographs through the ISB area. CGRP
and enkephalins are reported to be contained in dense-core
vesicles (Sliwinska-Kowalska et al., 1989), whereas ACh
and GABA are thought to be packaged in clear vesicles. The
distinction between dense-core and clear vesicles is easily
M. SATAKE AND M.C. LIBERMAN
illustrated fibers are shown contacting the modiolar side of the IHC
(light stippling), and one is shown contacting the pillar side of the IHC.
Fibers contacting the two sides of the IHC are known to differ in
spontaneous discharge rates and thresholds (Liberman, 1982a).
made in routine micrographs. Furthermore, in other ner-
vous tissues studied, ACh and GABA are typically stored in
vesicles that appear different: round vs. pleomorphic, respec-
tively (see, e.g., Helfert et al., 1992). Vesicle shape is also
quantified easily in routine electron micrographs.
The purpose of the present study was to systematically
apply ultrastructural morphometry to all vesicle-contain-
ing swellings in the ISB of selected cochlear regions from
cat and guinea pig in order to search for morphological
subtypes, defined mainly on the basis of vesicle content.
The peripheral targets of all these vesicle-containing swell-
ings were also determined (if any were visible within the
serial sections available), allowing us to address whether
there were any systematic morphological differences among
swellings according to whether they synapsed with RFs or
IHCs or whether they synapsed with low-SR or high-SR
RFs.
MATERIALS AND METHODS
Tissue preparation
Animals used in this study included two cats and three
guinea pigs (see Table 1).The cats were anesthetized with
intraperitoneal injections of Dial in urethane, and the
guinea pigs were anesthetized with intraperitoneal injec-
tions of Nembutal followed by intramuscular injections of
Innovar-Vet. Both cats and guinea pigs were used in
physiological experiments prior to perfusion and prepara-
tion of cochleas for electron microscopy. The physiological
ULTRASTRUCTURE OF' LATERAL OLIVOCOCHLEAR TERMINALS 623
TABLE 1. Cochlear Frequency Regions and Respective Numbers of section was obtained (magnification roughly x 3,000).These
Sections Analyzed
were used to 1) identify and number all the vesicle-filled
Frequency correlate No. of serial swellings to be analyzed, 2 ) to search for synaptic contacts
Animal ID (kHd sections between vesicle-filled swellings and their targets, and 3 ) to
Cat 1 0.25 102 identify the postsynaptic targets to whatever extent pos-
Cat I 2.00 54 sible. Second, a series of higher power views ( x 20,000) were
Cat 1 16.00 66
Cat 2 0.25 8 taken of each swelling in each section to be analyzed. These
Cat 2 16.00 8 micrographs were photographically enlarged to ~ 4 3 , 0 0 0to
Guinea pig 1 0.25 7
Guinea pig 1 0.50 12 measure the swelling size and were also photographically
Guinea pig 2 3.50 5 enlarged to ~ 9 0 , 0 0 0to measure size and shape of clear
Guinea pig 3 0.50 15
vesicles. The magnification of the electron microscope was
checked at each session by photographing a calibrated grid
in the microscope and printing the grid along with the
experiments included measurement of compound action micrographs obtained at that session.
potentials (CAP) from the round window and/or single- Measurements of swelling size and vesicle size were made
fiber recordings from the auditory nerve (cat 1only). Based by manually tracing the outline of the swellings or vesicles
on these physiological measures of cochlear function, all (in the appropriate micrographs), digitizing the outline
ears used in this report were normal at the time of the with a scanner (at 300 dots per inch), and analyzing the
perfusion; however, these tests do not assay the state of the digitized outlines with NIH Image (version 1.47). At the
OC system. Morphological data from one of the cats (cat 1) magnification used for the tracings, the scanner resolution
formed the basis for the analysis in a previous study of the produced a pixel size equal to roughly 2% of the average
afferent innervation of the organ of Corti (Liberman et al., diameter of a vesicle. In addition to the area and perimeter
1990). All experimental protocols for the treatment and use of any digitized outline, this software package computes the
of these animals were approved by the IACUC at the roundness, which is defined as the ratio of the minor to the
Massachusetts Eye and Ear Infirmary. Animals were killed major axis of the outlined form. Reproducibility checks
by exsanguination while they were deeply anesthetized. involving retracing and redigitizing the same vesicle popula-
Fixation of the cochleas was effected by intralabyrinthine tions typically yielded test-retest errors of less than 5%.
perfusion of a solution containing 2.5% glutaraldehyde in
0.065 M phosphate buffer with 0.005% CaC12 (0.45 mM),
pH 7.2, while the animal was still deeply anesthetized. RESULTS
Following perfusion for roughly 5 minutes, the cochlea was Databases and inclusion criteria
dissected out and postfixed in the same solution overnight
at 4°C. After postfixation, the cochleas were rinsed (phos- Figure 1 is a schematic illustration of the ISB, which
phate buffer) and osmicated (1%Os04 in phosphate buffer) consists of a number of separate fascicles of longitudinally
for 1-2 hours, dehydrated in ethanol, and embedded in oriented fibers running among the supporting cells in the
epoxy resins via propylene oxide. For one of the guinea pigs IHC area. The ISB contains the peripheral terminals of the
(guinea pig 31, an en bloc staining step (with uranyl acetate LOC efferent fibers, which synapse primarily with the
saturated in distilled water) was inserted between the dendrites of afferent RFs. It may also contain spiraling
osmication and the dehydration steps. portions of medial OC neurons destined for the outer hair
After oven curing the epoxy, the cochleas were processed cells (OHCs) as well as brief spiral courses of some RFs
as plastic-embedded surface preparations in which the (Liberman, 1980a; Brown, 1987a,b).Outer spiral fibers, the
entire organ of Corti, from apex to base, was dissected into a afferent innervation of OHCs, aImost never travel in the
series of pieces that were reembedded in epoxy (for ease of ISB (Brown, 1987b; Simmons and Liberman, 1988).
handling), thinned with sanding discs to roughly 200 pm in The purpose of the present study was to characterize the
thickness, and then glued to microscope slides. Each 200 pm ultrastructure of all vesicle-containing swellings in the IHC
wafer contained a piece of the organ of Corti roughly 1 mm area, looking for possible morphological subclasses of the
in length, with the basilar membrane parallel to the slide. LOC system. Typical electron micrographs of portions of
The length of the organ of Corti can be measured with a the ISB in cat and guinea pig are illustrated in Figure 2.
drawing tube on a light microscope, and the frequency Many of the larger neuronal profiles contain vesicles. Some
correlate of any cochlear piece can be estimated by using contain mainly clear vesicles (see, e.g., Fig. 2, vesicles 1and
cochlear frequency maps, which are available for the cat a), some contain a mixture of clear and dense-core vesicles
and the guinea pig (Liberman, 1982b; Liberman and Gao, (see, e.g., Fig. 2, vesicles 3, 4,and 51, whereas still others
1995). contain an extremely high percentage of dense-core vesicles
Selected regions of the organ of Corti were removed from (see, e.g., Fig. 2, vesicles 6 and 7). Some of the large, clear
the slide, mounted on epoxy blanks, and serially sectioned vesicle-containing profiles (see, e.g., Fig. 2, vesicle 1)can be
in the radial plane on an ultramicrotome. Serial-section seen to synapse (Fig. 2, arrow) with other nonvesiculated
ribbons were collected on Formvar-coated slot grids, stained profiles (putative RFs) and, thus, are prime candidates for
with uranyl acetate and lead citrate, and examined in an OC terminals. However, still other profiles contain only a
electron microscope. The numbers of serial sections col- modest number of vesicles, perhaps only three to five per
lected in each of the cochlear regions analyzed from each of cross section (see, e.g., Fig. 2 , vesicle 7); it is not immedi-
the five animals are summarized in Table 1. ately obvious how to classify such vesicle-poor profiles. The
smallest profiles (see, e.g., Fig. 2, vesicle 8) contain only
Electron microscopy and morphometry neurotubules and neurofilaments and presumably corre-
Analysis of ultrastructure involved three steps. First, a spond to the nonsynaptic interswelling portion of the
series of low-power views of the inner spiral bundles in each spiraling OC neurons.
 Fig. 2. Electron micrographs of typical radial sections through the (3, 4, and 5), a swelling with a particularly high ratio of dense-core to
inner spiral bundle in cat (top) and guinea pig (bottom). Features clear vesicles (6), a profile with very few clear vesicles (71, and a
highlighted include swellings containing only clear vesicles (1and 21, neuronal profile containing only neurotubules and neurofilaments (8).
swellings containing a few dense-core as well as numerous clear vesicles Scale bars = 1 pm.
 18
ULTRASTRUCTURE OF LATERAL OLIVOCOCHLEAR TERMINALS
300
A: Cat 0-1 DCV/sq.pm
1-4 DCV/sq.pm
4-1 6 DCV/sq.pm
=-
.-
Y
0
16-64 DCVlsq.pm
v) 0 0
c 200 0 200
0°F
0 0
0 0
I I I 1
0 1 2 3
Swelling Area (sq. pm)
Fig. 3. Analysis of swelling size (x-axis),clear-vesicledensity (y-axis),
and dense-core vesicle density (symbol type) for cat (A) and guinea pig
(B). Swelling area is defined as the maximum cross-sectional area
among the serial sections. Clear and dense-core vesicle densities
represent the average density seen across all sections through each
If the release of neuroactive substances occurs primarily
at localized swellings along or at the ends of neuronal
processes where vesicles congregate (en passant or terminal
swellings, respectively), then the functionally important
profiles to evaluate would be those containing vesicles and
representing obvious swellings of the axoplasm. Light
microscopic studies of labeled LOC neurons suggest that
their swellings in the ISB area are, at most, about 2-3 pm
in diameter. Thus, to assess from serial ultrathin sections
whether a profile represents a localized swelling requires, at
most, 4-5 pm (or 50-60 serial sections).
The analysis in the present study encompassed nine
cochlear regions from five different animals (two cats and
three guinea pigs), as summarized in Table 1. The numbers
of serial ultrathin sections analyzed were greater than 50 in
each of three cochlear regions from one cat (cat 1)and were
between five and 15 in all other cochlear regions from the
other cat and from all three guinea pigs. Thus, in cat 1,for
which extensive serial sections were available, the criterion
for inclusion in the database was that the profile appeared
to travel longitudinally in the ISB and constituted a local-
ized swelling in the neuron, demonstrating an increase of at
least 50% in diameter. Virtually all profiles included by
these criteria contained at least a few “vesicles.” Further-
more, virtually all vesicles visible within the ISB were
contained within such profiles (for the present study,
vesicles are defined as membranous profiles with areas
from 800 to 300 nmz, i.e., diameters from circa 32 to
62 nm). For those regions with fewer than 20 sections
analyzed, the database comprised all vesicle-containing
profiles, without explicit attempts at reconstruction-based
demonstration that each profile represented a localized
swe11ing.
625
300
B: Guinea Pig
0
0
o n
100 0
0
I I I I
0 1 2 3
Swelling Area (sq. pm)
swelling Le., the total number of vesicles counted in all relevant
sections divided by the summed cross-sectional areas of the swelling
across all relevant sections). Data for cat are from the 2.0 and 16.0 kHz
regions of cat 1; data from guinea pig are from all cochlear regions of
guinea pigs 1and 2.
In the Discussion, evidence will be presented to suggest
that all the vesicle-containing swellings analyzed arise from
the thin unmyelinated fibers of the LOC system. Although
such an origin has not been directly demonstrated, they will
be referred to as “LOC swellings” as a convenient short-
hand.
Swelling size and the density of clear
and dense-core vesicles
Micrographs, such as those in Figure 2, suggest differ-
ences in the relative proportion of clear and dense-core
vesicles as well as in the overall density of vesicles among
the swellings in the ISB. Of course, random micrographs
cannot reveal whether such variability reflects systematic
differences among swellings or simply the tendency for
clear and dense-core vesicles to cluster in complicated ways
within a homogeneous population of swellings.
Systematic analysis of swelling size and the density of
clear and dense-core vesicles for all ISB swellings was
undertaken to look for evidence of discrete subpopulations.
Results for cat 1, for which extensive section series were
available, are illustrated in Figure 3A. Although there is no
compelling evidence for discrete subpopulations, there are
clear interswelling differences in size and density of clear
and dense-core vesicles. Furthermore, there is a clear
tendency for the most dense-core vesicles to be seen among
the smallest swellings (maximum cross-sectional areas less
than 0.5 p,m2). A similar tendency is seen among the
profiles from the guinea pig cochleas (Fig. 3B): Those
profiles with the most dense-core vesicles (16-64 per pm2)
are always relatively small in diameter (maximum area less
than about 0.6 pm2) and relatively low in clear-vesicle
626
70 i
60
50-
-I
a,
0
LL
c
40-
0
L In each cochlear region evaluated, all neuronal profiles in
2 30-
E available sections were subjected to morphometric analysis.
* 20-
J
-m-T
0 0.2 0.4 0:6 ’ i
Fiber Diameter (pm)
Fig. 4. Distribution of diameters for all of the ISB neurons analyzed
from cat 1.Fiber diameter represents the average value for all sections
available from the extraswelling portion of each fiber.
content. Note also that dense-core vesicles are generally
more common in the guinea pig than in the cat, as can also
be seen from an examination of Figure 2. Although the
swelling sizes varied considerably, the interswelling por-
tions were more uniform in diameter: In the cat data shown
in Figure 3A, the interswelling diameters ranged from 0.1
to 0.7 pm (Fig. 4),with roughly 90% less than 0.4 pm in
diameter.
The analysis of the three cochlear regions from cat 1also
included a complete reconstruction of the RFs, as reported
in a previous publication (Liberman et al., 1990). For the
sample of swellings illustrated in Figure 3A,this analysis
allows us to examine possible correlations between swelling
size or the content of clear or dense-core vesicles on the one
hand and the nature of the peripheral targets (if any) on the
other hand. In other words, for each of the reconstructed
swellings, we could assess whether any synaptic contact
was made, and, if contact was made, what was the nature of
the postsynaptic element.
Figure 5A illustrates the proportion of these swellings
making synaptic contacts varies with swelling size: A very
high proportion (90%) of large swellings (i.e., those with
areas greater than 0.7 pm2) are seen to make synaptic
contact with RFs; a very small proportion (5%)of smaller
swellings are seen to make any synapses. Furthermore, it
appears that none of the numerous small swellings with
relatively few clear vesicles contact RFs. Comparison of
Figure 3A with Figure 5A also reveals that few of the small
swellings with high dense-core vesicle content synapse with
RFs: For example, of the numerous small (<0.5 pm2)
swellings with dense-core vesicle density > l / p m 2(Fig. 3A),
only one synapsed with an RF. The analysis in Figure 5A
also includes a comparison in the morphology of LOC
swellings contacting pillar-side vs. modiolar-side RFs (see
Fig. 5 , key): There are no compelling differences with
respect either to average size or to content of clear or
dense-core vesicles. Only two LOC swellings were seen to
synapse with the IHC (see Liberman et al., 1990), both were
on the modiolar side, and neither contained any dense-core
vesicles. Numerous vesicle-containing profiles were in inti-
M. SATAKE AND M.C. LIBERMAN
mate contact with the IHC without forming clear synaptic
specializations: Many of these formed synapses on RFs very
near the RFiIHC synapse. Figure 5B shows that these
IHC-contacting profiles included both large and small
swellings; however, most contained very few dense-core
vesicles.
Size and shape of clear vesicles
the IHC area containing at least 20 clear vesicles across all
The mean area and roundness of clear vesicles contained in
all swellings evaluated is plotted in Figure 6.
The mean size of clear vesicles ranged from about 1,250
to 2,500 nm2. This corresponded to average diameters (if
perfectly spherical) of roughly 40-60 nm. Vesicle diameters
in this range are typically classified as “large” (see, e.g.,
Cant, 19931, although relatively few studies classifying
terminals based on vesicle size provide actual measure-
ments. There appears to be a relationship between mean
size and the characteristic frequency (CF) region analyzed:
In Figure 6A-C, there is a tendency for the largest vesicles
to be seen in the most apical regions. However, there also
seem to be systematic differences between individual ani-
mals of the same species: e.g., the mean vesicle sizes at the
0.25 kHz region are significantly larger in cat 1than in cat
2. Thus, the apparent size difference between the 3.5 kHz
region and the 0.25 kHz region in Figure 6C could also be
due to the fact that the two regions were sampled from
different animals. Whatever their significance, these size
differences do not represent magnification drift in any step
of the documentation process: Their accuracy was verified
by simultaneously reanalyzing selected examples from dif-
ferent cochlear regions or animals.
There is little variability in the average shape of the clear
vesicles in any cochlear region analyzed from cat or guinea
pig. Mean roundness values in the vicinity of 0.85-0.90, as
seen here, would be termed “round” in existing classifica-
tion schemes of synaptic vesicles (see, e.g., Helfert et al.,
1992). Although the vesicle shape appears to be quite
homogeneous in all cochleas analyzed, the data shown in
Figure 6A-C were from cochleas prepared without en bloc
staining in uranyl acetate, a procedure typically used in
studies of vesicle shape in central nervous system tissues
(see, e.g., Helfert et al., 1992). Thus, the measurement of
vesicle size and shape was repeated in one cochlear region
from one guinea pig for which the cochlea was treated with
uranyl acetate prior to embedding. Figure 6D shows that
the results from this cochlea are not significantly different
from those obtained without the en bloc uranyl step: All
swellings in this cochlea would also be characterized as
containing large, round vesicles.
The data in Figure 6 do not suggest any obvious subtypes
of vesicle-containing swellings distinguishable on the basis
of clear vesicle morphology. However, these data represent
only mean values of the vesicle size and roundness on a
swelling-by-swelling basis. The data in Figure 7 compare
the distributions of size and roundness in individual swell-
ings, giving a visual comparison of the variances of the two
measures. When normalized to the means in this fashion,
these distributions of vesicle size and roundness also sug-
gest that the swellings are quite homogeneous with respect
to the sizes and shapes of the clear vesicles they contain.
Finally, in the one cat for which the section series was
extensive enough to allow complete reconstruction of much
ULTRASTRUCTURE OF LATERAL OLIVOCOCHLEAR TERMINALS
A I
No RF Synapses
0-1 DCV/sa.un
x
I
I
4-16 DCV/sq.Fm
0
0
1 I I I
3 1 2 3
Swelling Area (sq.pm)
Fig. 5. Swelling size, clear vesicle density, and dense-core vesicle
density for ISB swellings coded according to their peripheral target.
Data are the same as those from Figure 3A, except with different
symbol coding (see key). In A, the symbol code identifies those swellings
synapsing with RFs (circles and triangles) from those not synapsing
(crosses) and further identifies swellings synapsing with RFs destined
for the pillar side of the IHC (triangles) vs. the modiolar-side (circles).
of the surrounding afferent innervation, the mean size and
roundness of the clear vesicles was compared between
swellings contacting modiolar-side RFs, pillar-side-RFs,
and those not seen to contact any RFs at all. As shown in
Figure 8, there does not appear to be any appreciable
difference in the size or roundness of clear vesicles as a
function of the swelling target defined in this way.
DISCUSSION
Central origin(s) of the ISB swellings analyzed
Existing ultrastructural studies of the IHC area strongly
suggest that all fibers spiraling in the ISBs are part of the
OC system. First, afferent fibers to IHCs and OHCs take a
fundamentally radial course through this region, either
climbing directly to the IHC or passing directly between
inner pillar cells (well below the ISB) to cross on the floor of
the tunnel to the OHCs (Spoendlin and Gacek, 1963;
Liberman and Brown, 1986; Brown, 1987a,b; Simmons and
Liberman, 1988). Second, severing the OCB in the brain-
stem leads to complete disappearance of all the ISBs in cat
(Liberman, 1990). Third, autonomic innervation of the
cochlea is confined to the modiolus and osseous spiral
lamina (Spoendlin, 1981; Hozawa et al., 1989).
Thus, the swellings analyzed in the present study must
arise from the OC system. Light microscopic evidence from
reconstruction of single HRP-filled OC fibers suggests that
all swellings in the ISB region arise from thin collaterals of
spiraling OC fibers that do not project to the OHC region
(Liberman and Brown, 1986; Brown, 1987a). Such fibers
are studded with small en passant swellings in the ISB and
TSB, the sizes of which are comparable to those seen here.
627
300
No IHC Contacts
0-1 DCV/sq.prnl
x
o I A
200
x o
100
0
-
I I I I
1 1 2 3
Swelling Area (sq.pm)
In B, the symbol code identifies those swellings intimately contacting
(but not synapsing with) IHCs (open symbols) from those not contact-
ing (crosses) and further identifies the side of the IHC contacted (circles
for modiolar, triangle for pillar). Two swellings synapsing on RFs
contacting the basal pole of the IHC have been classified arbitrarily as
pillar side: Both had dense-core vesicle density < l/pm2.
On the other hand, OC fibers projecting to OHCs do not
produce en passant or terminal swellings in the IHC area
(Liberman and Brown, 1986; Brown, 1987a) and are large
in caliber: only 8% of OC fibers projecting to OHCs have
diameters less than 3.5 pm (Brown, 1987a), whereas 88%of
the interswelling diameters measured in this study were
less than 3.5 pm (Fig. 4).
A variety of studies, including immunocytochemical and
intracellular or extracellular labeling, suggest that all OC
swellings in the IHC area arise from the lateral (L)OC
system of small cells in and/or around the lateral superior
olivary nucleus (evidence reviewed by Liberman et al.,
1990). There may also be LOC projections to the OHC area
(see below), but there is no evidence that the medial (M)OC
system projects to the IHC area. Thus, the vesicle-filled
swellings analyzed here must correspond to peripheral
projections of LOC neurons.
If a subset of LOC neurons gives off swellings only in the
OHC region, then they would have been missed in the
present study. Two pieces of evidence suggest such a
projection may indeed exist. First, in the brainstem, only
LOC cell bodies are immunopositive for CGRP and/or
enkephalin (Altschuler et al., 1984a; Abou-Madi et al.,
1987; Lu et al., 1987; Vetter et al., 1991; Saffeidine and
Eybalin, 1992),yet fibers immunopositive for CGRP and/or
enkephalins have been reported in the OHC area, especially
in apical cochlear regions (Altschuler et al., 1984b, 1985;
Fex and Altschuler, 1984; Eybalin et al., 1988). Second, a
population of thin beaded fibers, similar morphologically to
spiraling ISB fibers, crosses the tunnel to OHC areas in the
apical half of the cochlea (Liberman et al., 1990).
628 M. SATAKE AND M.C. LIBERMAN
1
A: Cat 1 B: Cat 2
0.9 8
0
lr:
0.8

0 0.25kHz 0.7 0 0.25kHz

t& 2.0kHz 0 16.0kHz

0.6 0.6 I I I

1000 1500 2000 2500 1000 1500 2000 2500

1
C: Guinea Pig D. Guinea Pig (Uranyl)
0.9

A
0.8

1 0 0.25kHz
0.5 kHz
&% 3.5kHz
A
1 0.7

1 A 0.5kHz/
I I I 0.6 I I I
1000 1500 2000 2500 1000 1500 2000 2500
Mean Vesicle Area (sq. nm.) Mean Vesicle Area (sq. nm.)
Fig. 6. Scatter plots of mean roundness vs. mean area for clear vesicles are included.
... - -
The maximum . . number of.vesicles
. traced
. -per-”
vesicles from a population of ISB swellings in the cat (A,B) and guinea swelling was 200: lfcursory examination suggested that more than ZOO
pig (C,D) cochleas. Only the case in D was treated en bloc with uranyl vesicles were present across all of the available sections in the series,
acetate. Each symbol in A-D represents mean data for all the vesicles then the sections evaluated from that swelling were spaced at regular
measured in one swelling. Within each scatter plot, swellings are pooled intervals within the series. In this and in all subsequent figures, vesicle
across two or three cochlear regions, with different symbol types (see roundness is defined as the ratio of minor-to-major axis, as assessed by
key). All swellings for which there were at least 20 measurable clear NIH Image software.

The present study did not specifically evaluate swellings
in the TSB area; however, light microscopicreconstructions
of OC neurons show free and profuse intercommunication
between these two bundles: Single OC neurons send collat-
erals to both ISB and TSB, and no single neuron sends
branches only to the TSB (Brown, 1987a). In summary,
existing evidence suggests that the present analysis should
provide a complete sample of the LOC projection to the IHC
area.
Sites of neurotransmission:
Swellings, vesicles, and synapses
According to the present study, virtually all swellings in
the ISB contain clear vesicles, and vesicle aggregates occur
only within localized swellings. Furthermore, a high propor-
tion of the larger swellings synapsed with RFs (Fig. 5A),
whereas few small swellings formed synapses with any
structures (Fig. 5A). Thus, present morphological evidence
implies that neurotransmission occurs at most of the larger
swellings (cross-sectional area >0.7 pm2) in the ISB and
may not occur at most smaller swellings. The vesicles
within these smaller swellings may be in transit to more
distally located synaptic zones. Alternatively, they could be
released in the absence of clear synaptic contacts and affect
the function of the organ of Corti in a more diffuse fashion.
The small and large swellings seen in this study may arise
from the “spiral” and “terminal” portions, respectively,
seen in light microscopic studies of LOC fibers (Brown,
ULTRASTRUCTURE OF LATERAL OLIVOCOCHLEAR TERMINALS
a, 401~
a,
2 10
a,
n
0 0
0.5 1 1.5
0:5 i 1.5
Vesicle Area
(Normalized)
Fig. 7. A-F: Normalized distributions of area (A,C,E) and round-
1
ness (B,D,F) for all clear vesicles measured in three cochlear regions
(top, middle, and bottom) of one cat cochlea. Each plot contains
numerous superimposed traces; each trace represents data from one
swelling. Values of area and roundness for each vesicle (x-axes) were
1987a). Existing light microscopic data suggest that the
projection of a single LOC neuron in the ISB consists of 1)
an initial spiral portion in which there are small en passant
swellings and no branching and 2) a terminal portion in
which several collaterals are given off, each of which shows
numerous large en passant swellings (intermingled with
smaller ones) as well as terminal branchlets with large
terminal swellings. Although swelling sizes were not explic-
itly measured in Brown's light microscopic study, his
calibrated drawings suggest that the smaller en passant
swellings from the spiral portion correspond to areas less
than 0.5 Fm2, i.e., those with a low probability of forming
synapses (Fig. 5A; present study), whereas the larger ones
from the terminal portion correspond to areas greater than
0.5 Fm2 (with a high probability of synapsing with RFs).
The average number of swellings seen in the terminal
regions of LOC fibers is 120 (Brown, 1987a). If all these
make synaptic contact with RFs, then the estimated total of
1,500 LOC neurons projecting to one cochlea (Warr, 1975)
would create a total of 180,000 synaptic swellings distrib-
uted among the roughly 45,000 RFs, for an average ratio of
629
4 0 1 ~,
30
20
10
2 0.5 1 1.5 2
2
I
10
00.5
i 1.5
Vesicle Roundness
(Normalized)
2
normalized to the mean area or roundness for that swelling; the y-axes
show the relative proportion of vesicles of each normalized area or
roundness expressed as a percentage of the sample for that swelling.
Only swellings with at least 50 measurable clear vesicles are included in
these plots.
exactly four LOC swellingsiRF. Existing ultrastructural
data from serial section reconstructions suggest that each
high-SR fiber in the auditory nerve receives, on average,
from five to ten LOC synapses (Liberman et al., 1990).
Because a single LOC swelling often gives rise to multiple
synapses (as defined in these ultrastructural studies), the
two sets of numbers are probably in good agreement.
Relevance to putative subpopulations
within the LOC system
A significant body of immunohistochemical evidence
suggests that the LOC system comprises at least two major
subdivisions. Although it is far from established, one
appealingly simple hypothesis is that one division is cholin-
ergic and the other is GABAergic (Vetter et al., 19911, with
the two subtypes present in roughly equal numbers. Immu-
nocytochemical studies of the auditory brainstem have
suggested that, at least in rodents, the cholinergic neurons
may co-contain CGRP (Vetter et al., 1991) and enkephalins
(Altschuler et al., 1983; Abou-Madi et al., 1987; SafTeidine
630 M. SATAKE AND M.C. LIBERMAN
1
A: 2.0 kHz B: 16.0 kHz
0.9 r,
x x

0.8

U. /

v Synapse on Modiolar RFs

0 Synapse on Pillar RFs
0.6 0.6 I I I
1000 1500 2000 2500 1000 1500 2000 2500
Mean Vesicle Area Mean Vesicle Area
Fig. 8. Scatter plots comparing mean roundness vs. mean area of Data are all from one cat. A shows swellings from the 2.0 kHz region,
clear vesicles for lateral olivocochlear (LOC) swellings synapsing with and B shows swellings from the 16.0 kHz region. All other conventions
RFs on modiolar side (triangles) vs. pillar side (circles). Swellings of data display are as described for Figure 5.
(crosses) did not synapse with RFs within the serial sections analyzed.

and Eybalin, 1992), whereas the GABAergic population the base (Fex and Altschuler, 1984; Eybalin et al., 1988), we
may not contain CGRP, enkephalins, or any of the neuroac- might expect the pleomorphic vesicle-containing swellings
tive constituents thought to exist in the LOC system as a to be more common in the apical turns, i.e., cochlear
whole, e.g., dynorphins (Altschuler et al., 1985) and dopa- frequency regions tuned below roughly 2.5 kHz.
mine (Eybalin et al., 1993). The results of the present study are clearly not consistent
The co-localization patterns suggested from brainstem with these expectations. First, there is no evidence for
studies, to some extent, are corroborated by immunohisto- swellings containing pleomorphic vesicles, even in apical
chemical studies of the cochlea. Although no explicit colocal- turns of the guinea pig, where the evidence for GABAergic
ization studies have been executed for the periphery, a innervation is the strongest (Fig. 6). It is unlikely that there
number of investigators have reported that, for a variety of is any fundamental inadequacy in our methodology, given
immunological markers (cholinergic: Eybalin and Pujol, the fact that identical processing and analysis methods
1987; Vetter et al., 1991; opioid: Altschuler et al., 1984b; yielded numerous clear examples of pleomorphic and flat
GABAergic: Fex and Altschuler, 1984; Eybalin and Altsch- vesicles (with average roundness values from 0.50 to 0.85)
uler, 1990; Vetter et al., 1991), some of the fibers and/or when they were applied in our laboratory to electron
swellings in the ISB are reactive, whereas others are not. micrographs of ultrathin sections of neuropil in the co-
Swellings positive for enkephalin (Altschuler et al., 198413; chlear nucleus (Berglund et al., 1993). However, it is
Eybalin et al., 1985b), GABA (Eybalin et al., 1988), CGRP possible that peripheral nervous tissues behave differently
(Takeda et al., 1986, 1987; Sliwinska-Kowalska et al., from those in the central nervous system. Another possibil-
1989), and dopamine (Eybalin et al., 1993) have all been ity is that, in cochlear efferents, GABA is packaged in round
shown to make synaptic contact with RFs. vesicles. Indeed, recent immunoelectron microscopic stud-
If this hypothesis of cholinergic and GABAergic LOC ies have suggested that, in the inferior colliculus of cat,
subtypes is correct, then we might expect one population of whereas 97% of GABA-immunopositive terminals con-
ISB swellings to display round, clear vesicles (containing tained pleomorphic vesicles; a subpopulation of 3% con-
ACh) and numerous dense-core vesicles (containing CGRP tained round vesicles (Oliver and Beckius, 1992). Further-
and enkephalins) and a second population to contain only more, in cerebellar cortex, immunopositivity for GABA may
pleomorphic clear vesicles (containing GABA). Such expec- be found in both small-ovoid and large-spheroidal or pleo-
tations are based on 1)immunoelectron microscopic studies morphic vesicles (Hamori et al., 1990). However, it is also
of the cochlea suggesting that both CGRP and enkephalins possible that the GABA activity reported in the ISB is not
in the ISB are concentrated in dense-core vesicles (Altsch- associated with synaptic transmission. No clear resolution
uler et al., 198413; Sliwinska-Kowalska et al., 1989) and 2) of this question is possible at present.
studies of the central nervous system suggesting that ACh Although we found no compelling evidence for division of
and GABA are packaged in round and pleomorphic clear the ISB swellings into morphological subtypes, in both cat
vesicles, respectively (see, e.g., Van der Want and Nunes and guinea pig, the population of swellings was not homoge-
Cardozo, 1988; Helfert et al., 1992; Oliver and Beckius, neous: The highest dense-core vesicle contents were seen in
1992; Mize et al., 1994; Benson et al., 1996). Given the small swellings with few clear vesicles (Fig. 3), whereas
further observation that GABAergic innervation in the ISB large swellings tended to have few dense-core vesicles.
region is stronger in the apical half of the cochlea than in Furthermore, the population of small, dense-core vesicle-
 ULTRASTRUCTURE OF LATERAL OLIVOCOCHLEAR TERMINALS
rich swellings rarely synapsed with RFs (nor did they come
into intimate contact with IHCs), whereas almost all large
swellings synapsed with RFs (Fig. 5). It is possible that
these small and large swellings represent different LOC
subtypes. Given the fact that dense-core vesicles are likely
to be where neuroactive substances, such as CGRP and the
opioid peptides, are concentrated, it is interesting to note
that the swellings richest in dense-core vesicles rarely
formed synapses with RFs or other cochlear structures.
Although swellings that are immunopositive for neuroac-
tive peptides have been reported to synapse with RFs
(Altschuler et al., 1984b; Eybalin et al., 1985a,b, 1993;
Sliwinska-Kowalska et al., 19891, the present study sug-
gests that the majority of peptide-rich swellings do not form
clear synapses in the IHC area.
ACKNOWLEDGMENTS
This research was supported by an R 0 1 grant from the
NIDCD (DC-00188). The assistance of L.W. Dodds and D.F.
O’Grady in the preparation of ultrastructural materials is
gratefully acknowledged.
LITERATURE CITED
Abou-Madi, L., P. Pontarotti, G. Tramu, A. Cupo, and M. Eybalin (1987)
Coexistence of putative neuroactive substances in lateral olivocochlear
neurons of rat and guinea pig. Hearing Res. 30.135-146.
Altschuler, R.A., M.H. Parakkal, and J. Fex (1983) Localization of enkephalin-
like immunoreactivity in acetylcholinesterase-positivecells in the guinea-
pig lateral superior olivary complex that project to the cochlea. Neurosci-
ence 9t621-630.
Altschuler, R.A., J. Fex, M.H. Parakkal, and F. Eckenstein (1984a) Colocal-
ization of enkephalin-like and choline acetyltransferase-like immunore-
activities in olivocochlear neurons of the guinea pig. J. Histochem.
Cytochem. 32839-843.
Altschuler, R.A., M.H. Parakkal, J.A. Rubio, D.W. Hoffman, and J. Fex
(1984b) Enkephalin-like immunoreactivity in the guinea pig organ of
Corti: Ultrastructural and lesion studies. Hearing Res. 16t17-31.
Altschuler, R.A., D.W. Hoffman, K.A. Reeks, and J. Fex (1985) Localization
of dynorphin B-like and alpha-neoendorphin-like immunoreactivities in
the guinea pig organ of Corti. Hearing Res. 17.249-258.
Benson, T.E., A.M. Berglund, and M.C. Brown (1996) Synaptic input to
cochlear nucleus dendrites that receive medial olivocochlear synapses. J.
Comp. Neurol. 365:2741.
Berglund, A.M., K. Jacob, and M.C. Liberman (1993) Morphometry of
synaptic vesicles in anteroventral cochlear nucleus: Some correlations
with terminal origins. Association for Research in Otolaryngology.
Abstracts of theXVIth Midwinter Meeting, p. 118.
Brown, M.C. (1987a) Morphology of labeled efferent fibers in the guinea pig
cochlea. J. Comp. Neurol. 26Ot605-618.
Brown, M.C. (1987b) Morphology of labeled afferent fibers in the guinea pig
cochlea. J. Comp. Neurol. 260:591-604.
Cant, N.B. (1993) The synaptic organization of the ventral cochlear nucleus
of the cat. the peripheral cap of small cells. In M.A. Merchan, J.M. Juiz,
D A. Godfrey, and E. Mugnaini (eds): The Mammalian Cochlear Nuclei:
Organization and Function, New York: Plenum Press, pp. 91-106.
Eybalin, M. (1993) Neurotransmitters and neuromodulators of the mamma-
lian cochlea. Physiol. Rev. 73:309-373.
Eybalin, M., and R.A. Altschuler (1990) Immunoelectron microscopic local-
ization of neurotransmitters in the cochlea. J. Electron Microsc. Tech.
15t209-224.
Eybalin, M., and R. Pujol(1987) Chohne acetyltransferase ( C U T ) immuno-
electron microscopy distinguishes a t least three types of efferent syn-
apses in the organ of Corti. Exp. Brain Res. 65.261-270.
Eybalin, M., L. Abou-Madi, J. Rossier, and R. Pujol (1985a) Electron
microscopic localization of N-terminal proenkephalin (synenkephalin)
631
immunostaining in the guinea pig organ of Corti. Brain Res. 358:354-
359.
Eybalin, M., A. Cupo, and R. Pujol (1985b) Met-enkephalin-Argb-Gly’-Leu*
in the organ of Corti: High performance liquid chromatography and
immunoelectron microscopy. Brain Res. 331r389-395.
Eybalin, M., C. Parnaud, M. Geffard, and R. Pujol (1988) Immunoelectron
microscopy identifies several types of GABA-containing efferent syn-
apses in the guinea-pig organ of Corti. Neuroscience 24r29-38.
Eybalin, M., G. Charachon, and N. Renard (1993) Dopaminergic lateral
efferent innervation of the guinea-pig cochlea: Immunoelectron micros-
copy of catecholamine-synthesizingenzymes and effect of 6-hydroxydopa-
mine. Neuroscience 54r133-142.
Fex, J., and R.A. Altschuler (1984) Glutamic acid decarboxylase immunore-
activity of olivocochlear neurons in the organ of Corti of guinea pig and
rat. Hearing Res. I5t123-131.
Guinan, J.J. Jr., W.B. Warr, and B.E. Norris (1983) Differential olivoco-
chlear projections from lateral versus medial zones of the superior
olivary complex. J. Comp. Neurol. 221:358-370.
Hamori, J., J. Takacs, and P. Petrusz (1990) Immunogold electron micro-
scopic demonstration of glutamate and GABA in normal and deaffer-
ented cerebellar cortex: Correlation between transmitter content and
synaptic vesicle size. J. Histochem. Cytochem. 38:1767-1777.
Helfert, R.H., J.M. Juiz, S.C. Bledsoe, Jr.,J.M. Bonneau, R.J. Wenthold, and
R.A. Altschuler (1992) Patterns of glutamate, glycine and GABA immu-
nolabeling in four synaptic terminal classes in the lateral superior olive
of the guinea pig. J. Comp. Neurol. 323t303-325.
Hozawa, K., R.S. Kimura, and T. Takasaka (1989) Sympathetic nervous
system in guinea pig cochlea. Ear Res. Jpn. 20: 111-120.
Liberman, M.C. (1980a) Morphological differences among radial afferent
fibers in the cat cochlea: An electron-microscopic study of serial sections.
Hearing Res. 3t45-63
Liberman, M.C. (1980b) Efferent synapses in the inner hair cell area of the
cat cochlea: An electron microscopic study of serial sections. Hearing
Res. 3:189-204.
Liberman, M.C. (1982a) Single-neuron labeling in the cat auditory nerve.
Science 216t1239-124 1.
Liberman, M.C. (1982b) The cochlear frequency map for the cat: Labeling
auditory-nerve fibers of known characteristic frequency. J. Acoust. Soc.
Am. 72.1441-1449.
Liberman, M.C. (1990) Effects of chronic cochlear de-efferentation on
auditory-nerve response. Hearing Res. 49:209-224.
Liberman, M.C., and M.C Brown (1986) Anatomy and physiology of single
olivocochlear neurons in the cat. Hearing Res. 24.17-36.
Liberman, M.C., and W.Y. Gao (1995) Chronic cochlear de-efferentation and
susceptibility to permanent acoustic injury. Hearing Res. 90:158-168.
Liberman, M.C., L.W. Dodds, and S. Pierce (1990) Afferent and efferent
innervation of the cat cochlea: Quantitative analysis using light and
electron microscopy. J. Comp. Neurol. 301:443-460.
Lu, S.M., L. Schweitzer, N.B. Cant, and D. Dawbarn (1987) Immunoreactiv-
ity to calcitonin gene-related peptide in the superior olivary complex and
cochlea ofcat and rat. Hearing Res. 31:137-146.
Mize, R.R., R.H. Whitworth, B. Nunes-Cardozo, and J. Van der Want (1994)
Ultrastructural organization of GABA in the rabbit superior colliculus
revealed by quantitative postembedding immunocytochemistry. J. Comp.
Neurol. 341t273-287.
Oliver, D.L., and G.E. Beckius (1992) Fine structure of GABA-labeled axonal
endings in the inferior colliculus of the cat: Immunocytochemistry on
deplasticized ultrathin sections. Neuroscience 46:455-463.
Ryan, A.F., I.R. Schwartz, R.H. Helfert, E. Keithley, and 2.-X. Wang (1987)
Selective retrograde labeling of lateral olivocochlear neurons in the
brainstem based on preferential uptake of 3H-D-aspartic acid in the
cochlea. J. Comp. Neurol. 255,606-616.
Saffeidine, S., and M. Eybalin (1992) Triple immunofluorescence evidence
for coexistence of acetylcholine, enkephalins and calcitonin gene related
peptide within efferent (olivocochlear) neurons of rats and guinea pigs.
Eur. J. Neurosci. 4t981-992.
Schwartz, I.R., and A.F. Ryan (1986) Amino acid labeling patterns in the
efferent innervation of the cochlea: An electron microscopic autoradio-
graphic study. J. Comp. Neurol. 246t500-512
Simmons, D.D., and M.C. Liberman (1988) Afferent innervation of outer
hair cells in adult cats, I: Light-microscopic analysis of fiber labeled with
horseradish peroxidase. J. Comp. Neurol. 27Ot132-144.
632
Sliwinska-Kowalska, M., M. Parakkal, M.E. Schneider, and J. Fex (1989)
CGRP-like immunoreactivity in the guinea pig organ of Corti: A light
and electron microscopy study. Hearing Res. 4283-96.
Smith, C.A. (1961) Innervation pattern of the cochlea. Ann. Otol. Rhinol.
Laryngol. 84.504-527.
Spoendlin, H.H. (1981) The autonomic innervation of the inner ear. Adv.
Oto-Rhino-Laryngol. 27: 1-13.
Spoendlin, H.H., and R.R. Gacek (1963) Electron microscopic study of the
efferent and afferent innervation of the organ of Corti of the cat. Ann.
Otol. Rhinol. Laryngol. 72.660-686.
Takeda, N., M. Kitajiri, S. Girgis, C.J. Hillyard, I. MacIntyre, P.C. Emson, S.
Shiosaka, M. Tohyama, and T. Matsunaga (1986) The presence of a
calcitonin gene-related peptide in the olivocochlear bundle in rat. Exp.
Brain Res. 61.575-578.
Takeda, N., K. Doi, N. Mori, H. Yamazaki, M. Tohyama, and T. Matsunaga
M. SATAKE AND M.C. LIBERMAN
(1987) Localization and fine structure of calcitonin gene-related peptide
(CGRPI-like immunoreactive nerve fibres in the organ of Corti of guinea
pig by immunohistochemistry. Acta Otolaryngol. (Stockholm) 103:567-
571.
Van der Want, J.J.L., and J.J. Nunes Cardozo (1988) GABA Immuno-
electron microscopic study of the nucleus of the optic tract in the rabbit.
J. Comp. Neurol. 271.229-242,
Vetter, D.E., J.C. Adams, and E. Mugnaini (1991) Chemically distinct rat
olivocochlear neurons. Synapse 7-21-43.
Warr, W.B. (1975) Olivocochlear and vestibular efferent neurons of the feline
brain stem: Their location, morphology and number determined by
retrograde axonal transport and acetylcholinesterase histochemistry. J.
Comp. Neurol. 161.159-182.
Warr, W.B., and J.J. Guinan, J r. (1979) Efferent innervation of the organ of
Corti: two separate systems. Brain Res. 173:152-155.