- Email: [email protected]
Glycine immunoreactivity localized in the cochlear nucleus and superior olivary complex
Neurosc&nce Vol. 22, No. 3, pp. 897-912, 1987 Printed in Great Britain
0306-4522/87 $3.00 + 0.00 F’crgamon Journals Ltd
GLYCINE IMMUNOREACTIVITY LOCALIZED IN THE COCHLEAR NUCLEUS AND SUPERIOR OLIVARY COMPLEX R. J. WENTHOLD,D. HUE, R. A. ALTSCHIJLER and K. A. RBEKS Laboratory of Neuro-otohuyngology, National Institute of Neurological and Communicative Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892, U.S.A. Akatrae-Polyclonal antibodies were made in rabbits against glycine conjugated to bovine serum albumin with glutaraldehyde and were used for lmmunocytochemical studies in the cochlear nucleus and superior olivary nucleus of the guinea-pig. Antibodies selective for glycine were prepared by affinity chromatography. By dot-blot analysis this preparation showed a strong recognition of glycine conjugates and relatively little recognition of conjugates of most other amino acids tested. However, there was a slgnitkant reaction with conjugates of ala&e and beta-alanine, and this cross-reaction could not be removed by tinity chromatography without eliminating the preparation’s recognition of glycine. The affmity-puriiied preparation showed only a weak recognition of conjugates of gamma-aminobutyrate (GABA) which was detectable at high concentrations of primary antibody. Immunocytochemical studies showed several intensely staining cell bodies in the cochlear nucleus and superior olivary complex. Most immunoreactive cell bodies in the cochlear nucleus were in the dorsal cochlear nucleus, being present in both the superflcial and deep layers. Scattered immunoreactive cells were ptesent in the ventral cochlear nucleus. Intense staining of cell bodies was seen in the medial nucleus of the trapezoid body, and these cells appear to correspond to the principal cells of that nucleus. Punctate labelling, suggestive of immunoteactive presynaptic terminals, was also apparent, particularly in the ventral cochlear nucleus and lateral superior olive. In the ventral cochlear nucleus, immunoreactive puncta were found around unlabeled cell bodies, at times nearly covering the perimeter of the cell. A population of glycine-immunoreactive cell bodies in the super8clal dorsal cochlear nucleus also labeled with anti-GABA antibodies as determined through double-labeling studies. However, glycine-positive cells in the deep dorsal cochlear nucleus were not labeled with anti-GABA antibodies, and some populations of GABA-positive cells in the supetlicial layers were not labeled with anti-glycine antibodies. In the hippocampus intense staining of cell bodies and puncta was seen with anti-GABA antibodies while essentially no staining was seen with anti-glycine antibodies. These results suggest that anti-glycine antibodies can be useful for immunocytochemical identification of glycinergic neurons. From this study several populations of putative glycinergic neurons am identified in the auditory nuclei of the brain stem using these antibodies. Some populations of GABA-containing neurons also contain high levels of glycine or a related molecule.
There is strong
cord.” High-affinity uptake of glycine has been demonstrated in the cochlear nucleus,m and endogenous glycine is released from cochlear nucleus slices in a calcium-dependent manner.% Glycine inhibits neurpharmacological experiments, indicate that gly- onal firing in the cochlear nucleus with iontophoretic cinergic synapses are most abundant in the spinal application, and strychnine reduces glycine-induced cord and brain stem. Results of several studies responses. 8f161(1 In the superior olivary complex strongly support a role for glycine as a neuro- (SOC), iontophoretic studies indicate that the neurtransmitter in auditory nuclei of the brain stem. The onal pathway originating in the medial nucleus of distribution of glycine shows a sevenfold concen- the trapezoid body (MNTB) and terminating in the tration range within the cochlear nucleus of the cat, ipsilateral lateral superior olive (LSO) may be glyand the highest levels are similar to those in spinal cinergic20 These results are supported by high levels of [3H]strychnine bindir#’ as well as immunocytochemical analysis showing heavy staining for Address for correspondence:Dr R. J. Wenthold, La&atory of Neuro-otolaryngology, Bldg 36, Room 5DO8, glycine postsynaptic receptors in the LS0.37 NIH, Bethesda, MD 20892, U.S.A. Until recently glycinergic synapses could not be Abbreviations: AVCN, anteroventral cochlear nucleus; readily identified using immunocytochemical techBSA, bovine serum albumin; DCN, dorsal cochlear niques. However, with the purification and develnucleus; FITC, fluorescein isothlocyanate; GABA. opment of monoclonal antibodies against the glycine gamma aminobutyrate; LSG, lateral superior olive; postsynaptic receptor,‘U6*27such synapses can now be MNTB, medlal nucleus of the trapexoid body; PVCN, posteroventral cochlear nucleus; SGC, superior olivary identified.32*”Our recent work using these antibodies complex; VCN, ventral cochlear nucleus. shows a widespread and dense distribution of glycine evidence that glycine is a major inhibitory neurotransmitter in the mammalian central nervous system.3*‘o*4’*4L45*46 Studies on the distribution and uptake of glycine, as well as results of
897
898
R. J.
WENTHOLD el al
receptors throughout the cochlear nucleus and SOC.*.” With the exception of the MNTB-LSO pathway, there are no indications concerning the identities of neurons giving rise to the presynaptic elements of these putative glycinergic synapses. Based on studies of other neurotransmitters, neurons releasing glycine could be studied by measuring selective biosynthetic enzymes for glycine or by measuring glycine itself. No enzyme selective for the neurotransmitter pool of glycine has yet been identified. However, the presence of high levels of free glycine may be a potential indicator of neurotransmitter glycine, since the distribution of glycine rather closely follows the reported neurotransmitter role of glycine This suggests that the in brain and spinal cord. 3~10.‘1 direct localization of glycine may be useful in identifying glycinergic neurons. The feasibility of immunocytochemical localization of neurotransmitters has been demonstrated for gamma-aminobutyrate (GABA) by Storm-Mathisen et al.” and confirmed Glycine immunoreactivity has by several others. 133’*y) recently been demonstrated in the central nervous system.639*35In the present study we have sought to produce selective antibodies against glycine conjugated to bovine serum albumin (BSA) and to use them to determine the distribution of glycine in the brain stem auditory
nuclei.
EXPERlMENTAL PUOCEDUPES
Glycinc was conjugated to BSA with glutarakkhy& using the method of Storm-Math&n er al.” Rabbits were injected with 2mg of glycincBSA conjugate in compkte Freund’sadjuvant. Booster injections of I mg of conjugate in incompkte Freund’sadjuvantwem given at 4 and 6 weeks and 6 and 12 months. Animals wem bkd 1 week aider the last three booster injections. while all bleeds gave antibodies that appeared to recognize conjugated glycine. antibodies obtained from the second bleed gave the most intense qxciiic staining with lowest background and was used in inununoeytochemicol studies preeentd here.Aritkenua was
at&t&y-purifkdby pasaagcthrough columns of amino acid conjugates. The tirst column contained glycine conjugated to ovalbumin with glutaraldehyde attached to cyanogen bmmide-activatedSepharose.Bound antibodies were duted with 0.1 M acetic acid, pH 2.0. and immediately neutral&d with potassium phosphate. This fraction was then paned through a column of GABA conjugated to BSA with
glutuakkhyde attached to cyanogcn bromide-activated !kpkoae. The unbound fraction was collected, BSA w-as added to a final concentration of 1?A, and stored at - 30°C. For do&k-labeling studies, antibodies against GABA c&ugatai to BSA with ghrtar&ehyde were made in guitxm-pigs following methods similar to those for producing m&t antil~~Iies in rabbits as previously deeaibai.q Booster mjections were given 2 and 4 we& aI?er the initial injection of antigen. Antibodies were. a!Bnity-purhhd by paaaing &rough a cohunn of gIycine conjugated to BBA with gItuaraldehyde attached to cyanogan Luankkaetivatal fiepharose.The unntained fraction was used for immunocytocbemkal analysis of GABA. Antibodies wcn character&d u&g immunoblotn of ovalimmin-wnjugata! amino acids. Four microllrpmr of
cowere applied to nitrocelluloae paper and incubatiata and waabes were done in Tris-buffered saline. Afker incubatin in primary antibody and washes, the nitroc&dose paper was incubated in goat anti-rabbit IgG
conjugated to peroxidase (Bio-Rad Laboratories) at a dilution of l/lSOO. For guinea-nix antibodies the avidin-biotin procedure was used.Peroxida& activity was visualized with 4-chloro-I-naphthol. For immunocytoehemical studies female NIH strain guinea-pigs (ISO-250g) were perfused with 5t&lOOml of phosphate-buffered saline. pH 7.4, followed by fixative containing 4% parafonnaldehyde and 0.25% glutaraldehyde in 0.1 M sodium cacodylate at 4°C. Tissue was postfixed for 1h at 4°C in the same fixative and then transferred to cold phosphate-buffered saline. Vibratome sections were cut at 50 pm. For immunocytochemical localization sections were preincubated in phosphate-buffered saline containing 10% normal goat serum for 1h. Sections were then incubated in primary antibody overnight at 4°C at dilutions ranging from l/60 to l/600. Subsequent steps followed the avidin-biotin method of Hsu et of.” (Vector Labs) and peroxidase activity was visualized using diaminobenzidine. For adsorption controls 200 ~1 of antibody (at l/20 dilution) was mixed with 100~1 of amino acid conjugate (4 mg protein/ml), diluted to 2 ml with phosphatebuffemd saline, and incubated for I h at room temperature. Immunocytochemical studies were then performed as described above. For double-labeling studies to localize glycine and GABA, tissue was incubated with both primary antibodies. Rhodamine-labeled goat anti-rabbit IgG and fluoreacein isothiocyanate (FlTQlabeled goat anti-guinea-pig IgG were used to localize both primary antibodies on the same tissue section. Controls were performed as outlined by Wessendorf and Elde.” It was found that the secondary antibodies cross-reacted to some extent with the inap propriate primary antibody; particularly, goat anti-rabbit IgG recognized guinea-pig antibodies. This cross-reactivity was eliminated by passing the goat anti-rabbit antibodies through a column of immobilized guinea-pig IgG and the goat anti-guinea-pig antibodies through a column of immobilized rabbit IgG. Double-labeled fhtorescent sections were examined with a Z&s Universal microscope with epifluorescence illumination using a 50 W mercury light source. For rhodamine, the Zeiss 487715 filter combination was used. For FITC, the Zeiss 487716 filter combination was used together with a 515-565 nm bandpass filter to control red emissions. RESULTS
Properties of anti-glyciite adbodies Immunoblot analysis of crude anti-glycine serum shows reaction with ovalbumin conjugates of several amino acids in addition to glycine (Fig, 1). Several approaches were tried to effectively &ity purify glycine-specific antibodies for use in immuaocytochemistty. Using a technique analogous to that successfully used for purifying anti-G-A antibodie~,~ the serum was passed tbrougb a column of immobilized GABA conjugated to BSA. T&I was effective in removing cross-rerrcting antibodies as determined with immunoblotting. However, wbik certain cell populations were intensely tinad with this affinity-pur#kd prepamtion, imm~nocytochemical studies showed a weak general thbding of most neuronal all bodies; St&ring of nuclei was greatest. Preincutuation of the antibody witb Washed membranes from guinea-pig brain decrea& this staining. It is po&blc that this baokgroand staining was due to antipresent in the rabbits before immunization with the glycine conjupte; however,
Glycine immuaoreactivity in brain stern
899
6 GABA GLY ASP GLU ALA B-ALA TAU GLN Fig. 1. Immuaoblot analysis of rabbit anti-glyciae and guinea-pig anti-GABA antibodies. Gvslbuaria conjugates of amino acids were applied to aitrocellulose and reacted with aatibodies. (1) Whole serum of rabbit anti-glyciae at a dilution of l/250. (2) Rabbit anti-glyciae which bound to glyciae-ovalbtia column and eluted with acid, l/250 dilution. (3), (4) Atliaity-purified anti-glyciae eluted from the glyciae-ovalbumia column sad unbound to the GABA-BSA column; this preparation was used for immuaocytochemistry. Dilutions of primary antibody were l/250 (3) and l/l00 (4). (5) Whole senaa of guinea-pig anti-GABA at a dilution of l/250. (6) AfEaity-purified guinea-pig aati-GABA at a dilution of l/250. Bound rabbit antibody is detected using atliaity-purified goat anti-rabbit immuaoglobulias conjugated to horseradish peroxidase, and bound guinea-pig antibody is detected using the avidia-biotia procedure. Peroxidase was visuslixed with 4-&loro-1-aaphthol.
the same pattern of staining was seen in all bleeds of both animals used to produce antibodies. These animals were immunized at different times. This background staining could be removed by passing the serum through a column of immobilized glycine conjugated to ovalbumin. The antibodies recognizing conjugated glycine were bound to the column and were eluted with weak acid. This fraction contained antibodies that recognize conjugates of glycine as well as those of other amino acids (Fig. 1). Passage of this fraction through a column of immobilized GABA conjugated to BSA removed antibodies recognizing conjugates of several other amino acids while retaining those which recognize glycine as determined by immunoblotting. However, antibodies recognizing alanine and beta-alanine were not removed by this treatment and could not be removed using columns of conjugates of these amino acids without eliminating selective staining for glycine (Fig. 1). The purified antibodies also showed a very minor recognition of GABA, which can be seen only using high concentrations of antibody (Fig. 1). This crossreactivity was not removed by additional passage through the GABA-BSA column.
Antibody binding to both immunoblots and tissue sections was eliminated by preincubation with glycine-BSA conjugates, while neither free glycine at concentrations up to 1 mM nor conjugates of GABA, ghmnnate and aspartate affected the binding (Fig. 2C, D). Several concentrations of glutaraldehyde were used for fixation, and 0.25% was used for all findings reported here. Lower concentrations gave substantially reduced staining, while higher concentrations did not increase specific staining and often increased background staining. Specific staining was not seen in the absence of glutaraldehyde. Antibodies made in guinea-pig against GABA conjugated to BSA were similar to those previously made in rabbits as determined by immunoblotting. Significant reaction was seen with beta-ala&e and no reaction with glycine was detected (Fig. 1). Cochlear nucleus Glycine-immunoreactive cell bodies were most abundant in the dorsal cochlear nucleus (DCN). Based on size at least two classes of labeled cell bodies were seen as shown in Figs 2 and 3. The population with the larger cell bodies, with average diameters at
Fig. 2.
Glycine immunoreactivity in brain stem
901
Fig. 3. Distribution of glycine-immunoreactive cell bodies in the JXN of the guinea-pig. Primary antibody was used at a dilution of l/200. Labeled cells are drawn from a single representative section through the DCN. Gpen circles indicate larger immunoreactive cells with average diameters of 16pm (n = 54), and closed circles show smaller cells with average diameters of 10 pm (n = 58). Asterisks show large immunoreactive cells occasionally found in deeper layers of the DCN. Diameters were obtained from immunoreactive cells with a visible nucleus.
16 pm, was contined mostly to the superficial layers. These cells were usually spherical and often large labeled processes extended from their soma. These cells are similar in shape, size and distribution to a population of DCN cells labeled with anti-GABA and anti-glutamate decarboxylase antibodies.“” In the rat DCN, glutamate decarboxylase-immunoreactive cells have been identified as cartwheel cells, stellate cells and Golgi cells.” Therefore, the present results suggest that one or more classes of these cells may contain high levels of glycine. The other class of glycine-immunoreactive cells is made up of smaller cells with average diameters of 10 pm. The cell bodies are spherical or irregular (Fig. 2B). They were most often found in the deeper layers of the DCN, but some smaller cells were also found in the superficial layers. Finally, a few labeled cells which may constitute an additional class were present in the deep DCN. These cells were larger than those present in the superficial DCN and were often seen with labeled processes. Immunoreactive labeling of all cells was eliminated with prior incubation of the antibody with glycin+BSA conjugate, but labeling was not a&c&l by preincubation with GABA-BSA conjugates (Fig.
2C, D). Fusiform cells and granule cells in the DCN were not stained with anti-glycine antibodies. Glycine-immunoreactive cell bodies were also numerous in the ventral cochlear nucleus (VCN Fig. 4A), unlike results obtained with anti-GABA antibodies.” Two types of glycine-immunoreactive cells were seen. Large cells usually containing labeled processes were most abundant near the eighth nerve root (Fig. 4B), and less frequently found in the posteroventral cochlear nucleus (PVCN) and the rostra1 anteroventral cochlear nucleus (A-VCN, Fig. 4C). These cells were among the largest in the VCN with diameters sometimes exceeding 25 pm. The second class of glycine-immunoreactive cells was made up of small spherical cells which were most abundant in the granule cell cap of the AVCN (Fig. 4D). These appear to be slightly larger than granule cells in this region, which can be distinguished by their very weak labeling with the antibody, but similar in sixe to the smaller immunoreactive cells in the DCN. Unlabeled cell bodies throughout the VCN were often discerned by the heavy covering with immunoreactive puncta which outlined most of the cell body of neurons in the AVCN (Fig. 5A). In the DCN
Fig. 2. Glycine4mmunoreactive labeling in the DCN. Immunoreactive cell bodies are present in the superlkial (A) and deeper (B) layers of the DCN. This labeling is not eliminated by prior incubation of the antibody with GABA-BSA conjugate (C), but is completely absent after incubation with glycine-BSA conjugate (D). Primary antibody used at dilutions of l/600 (A) and l/200 [(B)-(D)]. Bars = 20pm [(A), (B)l, 55pm Kc), @)I.
902
R. J. Wmmcm et al.
Fig. 5. Immunoreactive pun&
around unlabeled cell bodies in the rostrai AVCN {A) and f&form ceif layer of the DCN (IQ. Them may be immuno~~tive making synapses on c&i bodies. Primary antibody nsed at dilutions of I/Zoo. Bar = 6 em.
preaynaptic terminals
Fig. 4. Glycbimmuno~tive laheliug in the VCN. (A) !kattcred large immunorcactive c&s near the anditory nerve root. f3) Higher tnodific&ion of large and small immunomactive {C) Ilarge ~~~r~ve cell in eaudal AVCN. Labeled pun&a are present ~ou~out the section. (D) I~~o~c~ve cells in the gnu&e cell cap of the AVCX IntCn*iy iabeled small ceils @ITOWS) am shown with weakly labeled gramde celIs (aI_rowheads). Primary antibody used at diMions of I/200 [(A)-(C)] and l/60 (ID). Bars = 35 pm (A), 20 wn t(B), (C)l, I3 pm P).
dls.
Fig. 6 904
905
Glycine immunoreactivity in brain stem
such patterns were not as frequently seen, but some puncta were seen adjacent to cell bodies in the fusiform cell layer (Fig. 5B). While analysis at the ultrastructural level is required for identification, the labeled puncta appear to be immunoreactive presynaptic Wminals. The pattern obtained with the anti-glycine antibodies is similar to that obtained with anti-GABA antibodies. Electron microscopic studies have shown these puncta to be GABAimmunoreactive presynaptic terminals.@ Furthermore, studies in the spinal cord using an anti-glycine antibody similar to the one used here have confirmed through electron microscopy the immunoreactive labeling of presynaptic tennim~ls.~~ The widespread punctate labeling around cells in the VCN suggests that most major groups of large neurons in this region receive terminals containing glycine. A dense glycinergic input to cells in the AVCN and PVCN is consistent with the heavy labeling of these cells with anti-glycine receptor antib0dies.l Superior olivary complex Intense immunoreactive labeling was seen in the superior olivary complex (Fig. 6A), with the heaviest labeling in cells in the MNTB (Fig. 6B). The neurons of the MNTB have been described in the cat.‘lJ2 Three types of neurons are present in this nucleus with the principal neurons being the most numerous. These cells have a spherical or oval perikaryon and an eccentric nucleus. Based on distribution, size and shape, the principal neurons of the MNTB appear to be glycine-immunoreactive. Cell labeling is also present in the ventral nucleus of the trapezoid body, and scattered cell labeling is seen in the LSO. In addition to cell latiling, the LSO contains heavy punctate labeling around unlabeled cell bodies, resembling immunoreactive presynaptic terminals (Fig. 6C).
Relationship to GABA-immunoreactive labeling The population of glycine-immunoreactive cells in the superficial DCN appears similar to a population of cells labeled with anti-GABA and anti-glutamate decarboxylase antibodies.” To determine if cells are immunoreactive for both GABA and glycine, doublelabeling studies were performed using anti-glycine antibodies made in rabbits and anti-GABA antibodies made in guinea-pig. Fluorescent second antibodies were affinity-purified to eliminate cross-
reactivity. These studies indicate that some cells in the superficial DCN contain both immunoreactivities (Figs 7 and 8). Some cells in this area were immunoreactive only for GABA, many of ,Wse appear smaller than the co-labeled cells and may, represent a class of smaller GABA-containing cells found in the DCN (Fig. 7). Cells heavily labeled for glycine and not GABA were not routinely seen, although occasionally cells weakly immunoreactive for glycine and negative for GABA were observed (Fig. 8). Cells in the deep DCN were immunoreactive only for glycine. In addition to immunoreactive labeling of cell bodies in the superficial DCN, punctate labeling and fiber labeling are also present. While many of these structures appear immunoreactive for both GABA and glycine, some puncta were labeled with only one antibody (Fig. 8). This labeling was further investigated in the VCN. There, most punctate labeling showed only one immunoreactivity, but doublelabeled puncta were seen throughout the VCN. This double labeling was most commonly found in the PVCN (Fig. 9). Therefore, as is the case with cell bodies, pun&a in the cochlear nucleus can bc immunoreactive for GABA, glycine or both. The co-labeling of some neurons in the DCN with antibodies against GABA and glycine suggested that this may be a general property of GABAergic neurons. To address this question, GABA- and glycineimmunoreactive labeling was studied in other areas of the central nervous system. The difference in labeling patterns obtained with the two antibodies was most apparent in the hippocampus where intense neuron labeling was seen with anti-glycine antibodies (Fig. IO). The only glycine-immunoreactive labeling seen there was a very light labeling of scattered small neurons near the surface of the hippocampus (Fig. 10, arrows).
DISCUSSION These experiments show that antibodies made against conjugated glycine can be used for the immunocytochemical localization of glycine in the central nervous system. In determining whether or not anti-glycine antibodies are useful for studying the distribution and properties of glycinergic neurons, two questions must be addressed. The first is whether glycine itself is a valid marker for glycinergic neurons. Glycine, unlike GABA which has been shown to be
Fig. 6. Glycinaimmunorcactive labeling in the superior olivary complex. (A) Low magnification illustration showing intense labeling of cell bodies in the medial nucleus of the trapezoid body (M) and labeling of cells, puncta and fibers in the lateral superior olive (L). Bar = 150 pm. (LI)Higher magni6cation of labeled principal cells in the MNTB. Bar = 18 pm. (C) Higher magnification of labeling in the LSO showing scattered labeled cells (arrows). Puncta, which may represent immunoreactive preqnaptic terminals, are present throughout the LSO and are often seen outlining unlabeled cell bodies (asterisk). Bar = 18 pm. Primary antibody used at dilutions of l/200.
Fig. 7. Co-Iocaltition of gIycine and GABA ~rn~~or~~~t~~ in c&l bodies in the sqerki~ IXX. {A) GA3A~imm~no~a~i~ c&s labeled with ~inea-~i~ anti-GABA and FIX-goat anti-~i~ea-pig IgG. (B) Glycine-immunoreactivt: cells labeled with rabbit anti-glycine and rhodaminegoat anti-rabbit IgG. A cell body (arrowhead) labeled with antW3ABA antibodies is not labeled with anti-glycine antihodie’s. Primary antibodies were used at ditutions of l/60. Bar = IS pm.
a r&bie immu~oCytochemica1marker for GABAergiC~~~~~13,2~.25.~1,~ is involved in many functions in the centraE nervous system in addition to he&g a ~ur~~r~s~~~er. High Ievets of glydne could re@ect any one of these functions. Such may be true for glutamate and aspartate, since the immunocyto~hemi~af d~s~~b~~onsof these amino acids do IKP~ ~&r&e with neuron& ~~~a~~ons b&eved to be asp&&erg&zor ~u~~~Cr~c.** However, previous studies have shown that the distribution af glycine is more rest&M, and in areas where it is not thought to be a major ne~o~~tter, g&&e levels are much tower than in areas where it is believed to be a ~urot~~it~r,~,‘“*4* For exampIe, the: concentration of g&C&ein the Hindus is more than t&&ii lower than in the spina$ cord? *r immune~@oC&mk~f results support these Endings showing little glyGinGnmun0~Ctive staining in the hippocampus, In the brain stem large gradients of g&tine ~~~~o~a~tivjt~ are seen. Within the cochtear
nucleus and SOC only certain celi ~~p~l~~~~n~ are intensely immunore~tive whib others are: f25smi: dally unfab&ed, Thjs may suggest that the
leve1 in neurons neC&sary to support the nonneurotransmitter functions is rather uniform indifferent populations of neurons and that it is muck lower than levels found in neurons where g&&e is a 1 n~uro~~~t~er. The second question concerns whether or not. antibodies made against &Eycinea= &mtive em&&h to suggest that the substance lo&i%~.I is &&ne. From the ~rn~~~ob~~t data it appt%rs that antiglycine antibodies do not appreciably recogr&C GXXI&gates of several other amino acids including !J@ putative ~urotr~~~~~~~~ g&tam&e and abbe.. Ekwwer, thy do show a & astir of ajtz- -, @es of GABA, alanine:and heta-&nine. The crossreactivity with GABA could not be removed by additional passage through a column of GABA CWjag&e, and the ~r~~-~~~v~t~ with afanine and
GIycine
imnunoreactivity in brain
stem
907
Fiis 8. Co-Idtion of@yeine and GABA ~~0~~~ incm11 bodka and pacta inthe XX (A) GABA-hnnuan-ve IX& kbekd with guinea-pig %nti-GABA and FITC-goat anti ilyinea-pig IgG. (B) Glyeine immunorwctitity d&e&d with rabbit anti-glyeine and rho&n&e-goat
anti-rabbit I&k Closed arrowhead chaws B cell body lightly lab&d with arhglycine antibadjca and unfabebd with anti_GABA mtibodks. Arrows show puma positive for both antibodies and open arm&m& show pmcta immunoreective fix only one antibody. Primary aatibodies were med at d&tiom of Wi43.Bar = 15pm.
beta-riianine could noi be removed without destraying the ~rnun~t~~ty for g&tine. In the sense that they cannot distinguish between giycine, alanine and beta-alanine, the anti-glycine antibodies are simiXar to the giycine postsynaptic receptor. Ala-
nineand
beta-alanine are strong antago&s and compete with g&he for binding to the receptor.i**45 Therefore, in immunocytochemical studies some immunoreactitity may be duo to cram-reactivity of the antibodies with alanine and beta-&nine. Both of
908
R. J. WENTHOLD et al
Fig. 9. Co-localization of glycine and GABA immunoreactivities in the PVCN. (A) GABA immunoreactive labehng using guinea-pig anti-GABA and FITC-goat anti-guinea-pig IgG. (B) Glycineimmunoreactive labeling using rabbit anti-glycine and rhodaminegoat anti-rabbit IgG. Arrows show examples of puncta labeled with both antibodies, and arrowheads show puncta positive for only one antibody. Primary antibodies were used at diiutions of l/60. Bar = 9 pm.
Fig. 10. Comparison of inmunoreactive labeling in the hippocampus with anti-GABA (A) and anti-glycine (B) antibodies. Cell bodies and pun&a are inter&y labeled with anti-GABA antibodies while only vety weak staining of sotue oell bodies near the surface of the hippocampus @ITOW)is present with anti-glycine antibodies. Primary antibodies were rabbit anti-GABA at a dilution of 1/2WO and rabbit anti-giycine at a dilution ol l/200. P indicates pyramidal cell layer. Bar = 55 pm.
P
c
910
R. J.
WENTHOLD er al
these amino acids are rather abundant in the central nervous system; for example, in both whole brain and the cochlear nucleus, the concentration of alanine is about one half that of GABA.19.% The weak reaction of anti-glycine antibodies with GABA appears to be negligible for immunocytochemical studies since many neurons which label with anti-GABA antibodies are not labeled with anti-glycine antibodies; this possible relationship with GABA is discussed further below. Finally, the antibodies against glycine may cross-react with unknown large or small molecules in certain neurons. This, of course, must always be a concern with any immunochemical study. An unexpected finding of the immunocytochemical studies was that the anti-glycine antibodies label a population of neurons in the cochlear nucleus that also label with anti-GABA antibodies. Although double labeling was done only with anti-GABA antibodies, this same population of neurons has been previously shown to be immunoreactive for glutamate decarboxylase. ‘W However, anti-glycine antibodies do not label only GABAergic neurons and do not label all populations of GABA-containing neurons. Cells immunoreactive for glycine in the deep DCN and the large immunoreactive cells in the VCN do not label for GABA. The flattened cells and small round cells in the superficial DCN which are GABApositive do not appear labeled with anti-glycine antibodies. Therefore, it appears that only some cells that are GABAergic based on GABA and glutamate decarboxylase labeling in the cochlear nucleus are labeled with anti-glycine antibodies. This point is clearly illustrated in the hippocampus where cell bodies and puncta are intensely stained with antiGABA antibodies and unlabeled with anti-glycine antibodies. In the cerebellum Golgi cells are labeled with anti-glycine antibodies, while stellate and basket cells are unlabeled (Ref. 6 and unpublished observation). The cells of all three classes are labeled with anti-GABA antibodies and are thought to be GABAergic.*’ High-affinity uptake studies on cerebellar slices suggests there are two populations of Golgi cells based on their uptake preferences for glycine and GABA.” Whether glycine and GABA immunoreactivities are in the same or different populations of Golgi cells awaits double-labeling studies. There are several possible explanations for the apparent colabeling of some neurons with both antibodies. The anti-glycine antibodies may recognize GABA or the anti-GABA antibodies may recognize glycine. This is unlikely since very little cross-reactivity is evident from the immunoblots. Furthermore, only some populations of cells labeled by one antibody are also labeled with the second antibody. Therefore, a crossreactivity between GABA and glycine could be a reasonable explanation only if GABA or glycine is recognized differently in different populations of GABAergic or glycinergic neurons. For example, the type of macromolecule to which the amino acid is linked during fixation may influence its immuno-
reactivity. Alternatively, the amount of GABA may vary considerably between populations of GABAergic neurons, and only those with very high concentrations of GABA will be recognized by the anti-glycine antibodies. A second possibility is that one or both antibodies recognize a third molecule. Two candidates are alanine and beta-alanine with which both anti-glycine and anti-GABA cross-react to some extent. However, while the anti-GABA antibodies used in the present study were not purified to remove antibodies recognixing alanine and beta-alanine, earlier studies on the cochlear nucleus, using anti-GABA antibodies with these contaminating antibodies removed, labeled the same population of cells in the superficial DCN.” Furthermore, this same population of neurons wntains glutamate decarboxylase immunoreactivity.23~“’ Therefore, since the GABA immunocytochemical findings can be supported by the presence of glutamate decarboxylase, it suggests that the glycine antibody recognizes a molecule in some GABAergic neurons. A third possibility is that both GABA and glycine are present in these neurons at relatively high levels. Again, since glutamate decarboxylase is also present in these neurons, it would imply that a subpopulation of GABAergic neurons has high levels of glycine. This does not necessarily imply that both are acting as neurotransmitters, although both amino acids seem to he enriched in the same presynaptic terminals as well as the same cell bodies. The present results show that several populations of neurons in the cochlear nucleus and SGC are enriched in glycine immunoreactivity. The intense labeling of cells in the MNTB and of fibres and terminals in the LSO is consistent with the neuronal pathway originating with the principal cells. of the MNTB and terminating in the LSG being glycinergic, as was suggested by pharmacological studies on the chinchilla.M Experiments directed at localizing the glycine postsynaptic receptor using [‘I-IJstrychnine and antibodies against the postsynaptic receptor also show intense labeling in the L!3G.37” In the cochlear nucleus most glycine-immunoreactive cells are in the DCN. While the immunoreactive cells cannot be assigned to any morphological group based on the present studies, the fact that the neurons labeled in the superficial DCN are also labeled with antiGABA’O and anti-glutamate decarboxylase” antibodies indicates that these are either cartwheel, stellate or Golgi cells. The glycine-immunoreactive ceils in the deep DCN could correspond to any of the small cells that have been noted in this region.‘~” The fact that most glycine-immunoreactive cell bodies are found in the DCN suggests that glycine-containing presynaptic terminals in the VCN may originate with these cells. The distribution, size and shape of the large immunoreactive cells in the VCN indicates that they may be the giant cells which have been reported to project to the contralateral cochlear nucleus in the cat.‘*’ The
Glycine ~~o~~ty
properties of tbese large cells in the VCN would be consistent with glycine being their neurotransmitter. This pathway is reported to be inhibitory, and this inhibition has been found to be antagonized by ~~.‘~~’ However, anterograde transport studies showed that the input of tbis pathway to the cochlear nucleus was relatively sparse,’ suggesting that the abundance of glycinergic synapses in the cochlear nucleus, based on glycine-immunoreactive staining and glycine postsynaptic receptor distribution, is not primarily due to &is pathway.
in brain stem
911
does not appreciably recognize conjugates of other amino acid neurotransmitters. Such antibodies may then be useful in identifying glycinergic neurons. Based on staining with this antibody, we have identiged several ~p~tio~ of neurons in the co&ear nucleus and SGC that may use glycine as a neurotransmitter. However, questions remain concerning the labeling of some populations of GABAcontaining neurons with anti-glycine antibodies. This finding suggests that some GABAergic neurons also contain high concentrations of glycine or a similar substance.
CONCLUSIONS Our antibody preparation against glycine conjugated to BSA recognims conjugates of glycine but
thank Drs J. Fex, J. C. Adams and A. N. Van Den Pol for criticaBy reading this manuscript.
Acknowle&ements-We
1. Adams J. C. and Warr W. B. (1976) origins of axons in the cat’s acoustic stria determined by injeotion of horseradish peroxidase into severed tracts. J. cotup. Newof. 170, 107-122. 2. Altscluder R. A., Bets H., Pamkkal M. H., Reeks K. A. and Wenthold R. J. (1986) IdentiScatioa of glycinergic synapses in the co&ear nucleus through immunocytochemical Mcahxation of the postsynaptic ramptor. B&t Res. %9,31&O. 3. A~rison hf. H.. Da& E. C.. Shank R. P. and McBride W. J. (1975) Neurochemi& evidence for aivcine as a tnwmitter aid a model for its ~~p~~ ~~~~~ti~. In ~e~~i~ C~~~~ ~~~o~~~ (eds Beri S., Clarke D. D. and S&eider D.), pp. 37-63. Raven Press, New York. 4. bergerS. J., CarterJ. G. and Lowry0. H. (1977)The distributionof glycine, GABA, glutamateand sspartatein rabbit spinal cord, cerebellum and hippocampus. J. Neurochem. 2S, 149-158. 5. Brawer J. R., Merest D. K. and Kane E. C. (1974) The neuronal amhitecture of the cochlear nucleus of the cat. J. camp. Neural. l!%, 251-299. 6. Camp&on G., Buijs R. M. and Get&d M. (1986) Glycine neurons in the brain and spinal cord. Antibody production and ~~~~ loathration. Bruin Rez 376,400-405. 7. Cant N. B. and Gaston K. C. (1982) Pathways connecting the right and left cochkar nuclei. J. camp. Net&. 212, 313-326. 8. Caspary D. M., Havey D. C. and Faingold C. L. (1979) Effects of microiontophoretically applied glycine and GABA on net&ml response~pattems in the &chJear nuclei. &utir Ees. 17& 179-185. 9. Casoarv D. M.. Rvbak L. P. and Fainaold C. L. (1985) The effects of inhibitorv and excitatorv amino-acid ne&otr&smitte& on the response proper& of brain&m auditory neurons. In Awdiroiy B&tern&try &d. Dres&er D.), pp. 198-224. Charles C. Thomas, SpringBeld, Illinois. in the ~ oentral nervous system. I&u. 10. Curtis I), R. and Johnston G. A. R. (1974) Amino acid ~~~ Ply&d 69,98-188. 11. Godfrey D. A., Carter J. A., Bergcr S. J., Lowry D. H. and Matschinsky F. M. (1977) Quantitative histochemical mapping of candidate transmitter ammo acids in the cat co&ear nuckus. J. Hivtochem. Cytochem. U, 417431. 12. Graham D., PfeifRr F., Simmler R. and Betr H. (1985) Purifkation and characterization of the Blycine receptor of pig spinal cord. Biochemirtry 2& 990-997. 13. Hodgaon A. J., Per&c B., Erdei A., Chubb I. W. and Somo8yi P. (1985) Antisara to y-aminobutyric acid. f. Production and &ara&&ation using a new modes system. J. Histech. Cyrochem. 33,229-239. 14. Hsu S-M., Raine L. and Fanger H. (1981) Use of a~bio~~x~ complex (ABC) in i~~o~~~ techniques: A comparison between ABC and unlabeled antibody (PAP) mums. J. Histochem. Cytockm. 29, 577-580. 15. Lorente de No R. (1933) Anatomy of the eighth nerve. III. General plan of structure of the primary cochlear nuclei. taryngoscojre, St. Louis 43, 327-350. 16. Martin M. R. (1985) The pharmacology of amino acid receptors and synaptic transmission in the cochlear nucleus. In Atditary Biochemistry (cd. llkscb D.), pp. 184-197. Charks C. Thomas, Springfield, Bhnois. 17. Martin M. R., Dickson J. W. and Fex J. (1982) Biie, strychnine and depressant ammo acid msponses in the anteroventral co&ear nucleus of the cat. J. Nenropharma. 21,2OI-207. 18. Martin M. R. and Fenix L. P. (1983) Comparison of the effects of bkucu&ne and strychnine on brain stem auditory evoked potentials in the cat. Br. J. Pharmac. T&75-77. 19. McGccc P. L. and McGeer E. G. (1981) Amino acid neurotransmitters. In Bade Neurochembtry (cds Siegel G. J., Albers R. W., Agranoff B. W. and Katzman R.), pp. 233-253. Little, Brown & Co., Boston. 20. Moore M. J. and Caspary D. M. (1983) Strychnine blocks binaural inhibition in lateral superior olivary neurons. J. Neurosci. 3,237~242, 21. Moreat D. K. (1968) The collateral system in the media nucleus of the trapexoid body of the cat, its neuronal amhiteeture and m&ion to the oiivo-co&lear bundle. Bruin Res. $288-311. 22. More& D. K. (1973) Auditory neurons of the brain stem. Adu. Oto-rhino&r. 20, 337-356. 23. Mugnaini E. (1985) GABA neurons in the superg&! layers of the rat dorsal cochkar nucleus: Light and electron microsoopic immunochemistry. J. amp. Neural. 235, 61-81. 24. Gttemen 0. P. and Storm-Math&n J. (1984) Glutamate- and GABA-containing neurons in the mouse and rat brain, as demonstrated with a new immunocyto&emical technique. J. camp. Neurol. 229,374-392.
R. J.
912
WENTHOLD
eI al.
25. Ottersen 0. P. and Storm-Math&n
J. (1984) Neurons containing or accumulating transmitter ammo acids. In Vol. 3, Classical Transmitters and Transmitter Receptors in the CNS (eds BjBrklund A., Hiikfelt T. and Kuhar M. J.), pp. 141-246. Elsevier, Amsterdam. 26. Pfeiffer F., Graham D. and Betx H. (I 982) Purification by affinity chromatography of the glycine receptor of rat spinal cord. J. biol. Chem. 257, 9389-9393. 27. Pfeiffer F., Simler R., Grenningloh G. and Betx H. (1984) Monoclonal antibodies and peptide mapping reveal structural similarities between the subunits of the glycine receptor of rat spinal cord. froc. MIX Acad. Sci. U.S.A. 81, 72247227. 28. Pirsig W., Pfalx R. and Sandanaga M. (1968) Postsynaptic auditory crossed efferent inhibition in the ventral cochlear nucleus and its blocking by strychnine nitrate (guinea pig). Kumumo~o med. J. 21, 75-82. 29. Pourcho R. G. and Goebel D. J. (1985) Immunocytochemical demonstration of glycine in retina. Brain Res. 348, Handbook
of Chemical
Neuroanatomy,
339-342. 30. Schwartz 1. R. (1981) The differential distribution of label following uptake of ‘H-labeled amino acids in the dorsal cochlear nucleus of the cat. Expl Neural. 73, 601-617.
31. Sequela P., Geffard M., Buijs R. M. and Le Moal M. L. (1984) Antibodies against gamma-aminobutyric aicd: specificity studies and immunocytochemical results. Proc. noun. Acod. Sci. U.S.A. 81, 388-392. 32. St. John P., Owen D. G., Barker J. L., Pfeiffer F. and Betx H. (1984) Monoclonal antibodies to purified receptor bind to glycine receptors on mouse spinal neurons in culture. Neurosci. Abstr. 10, 7. 33. Storm-Math&n J., Leknes A. K., Bore A. J., Vaaland J. L., Edminson P., Huang F. M. S. and Ottersen 0. P. (1983) First visualization of glutamate and GABA in neurones by immunocytochemistry. Norure St, 517-520. 34. Triller A., Cluzeaud F., Pfeiffer F. V. and Kom H. (1985) Distribution of gJycine receptors at central synapses: an immunoelectron microscopy study. 1. Cell Biol. 101, 683-688. 35. Van Den Pol A. N. and Gores T. (1986) Glycine immunoreactive neurons and presynaptic boutons in the spinal cord. Neurosci. Abstr. 12, 771. 36. Wenthold R. J. (1979) Release of endogenous glutamic acid, aspartic acid and GABA from cochlear nucleus slices. Brain Res. 162, 338-363. 37. Wenthold R. J., Betx H., Reeks K. A., ParakkaJ M. H. and AJtschuJer R. A. (1985) Lo&x&ion of glycitxqic synapses
38.
39. 40. 41. 42. 43. 44. 45. 46. 47. 48.
in the cochlear nucleus and superior olivary complex with monoclonal antibodies ape&c for the erycine receptor. Neurosci. Abstr. 11, 1048. Wenthold R. J. and Gulley R. L. (1977) Aspartic acid and glutamic acid levels in the cochkar nucleus after auditory nerve lesion. Brain Res. 138, 111-123. Wenthold R. J. and Martin M. R. (1984) Neurotranamitters of the auditory nerve and central auditory system. In Hearing Science (cd. Berlin I.), pp. 341-369. College-Hill Press, San Diego. Wenthold R. J., Zempel J. M., Parakkal M. H., Reeks K. A. and Altachuler R. A. (1986) Immunocyto~hamical localization of GABA in the cochlear nucleus of the guinea pig. Brok Res. 3sc, 7-18. Werman R. and Aprison M. H. (1966) Glycine: The search for a spinal cord inhibitory tranarnitter. In Structure ad Function of Inhibitory Neuronal Mechanisms (cds von Euler C., Skoglund S. and Swerbag U.), pp. 473-486. Pargamon Press, New York. Werman R., Davidoff R. A. and Aprison M. H. (1967) Inhibition of motoneurones by iontophoresis of glycine. Norure, Land. 214, 681683. Wessendorf M. W. and Elde R. P. (1985) Characterization of an immunotluorescence technique for the demonstration of coexisting neurotransmitters within nerve Bbres and terminals. J. Histochem. Cytoekm. 3& 98c994. Wilkin G. P., Csillag A., Balax~~R., King&uy A. E., Wilson J. E. and Johnson A. L. (1981) Localization of high a!Bnity 3H-glycine transport sites in the cereb&r cortex. Brain Res. 216, I l-33. Youna A. B. (1984) Glvcine recrmtors in the nervous svsttan. In Brain Recepror hfetho&loaies: Part B Amino Acti, Pep&, Psychwciive brugs (A Marangos P. J., Campbell I. C. and Cohen R. M.), pp 37-57. Academic Rass, Orlando. Young A. B. and McDonald R. L. (1983) Glycine as a spinal cord neurotranamitter. In Hun&ok of the Spind Cord, Vol. I (ed Davidoff R. A.), pp. l-44. Marcel Dekker, New York. Young E. D. and Brownell W. E. (1976) Responses to tones and noise of single cells in dorsal cochkar nucleus of unanesthetixed cats. J. Neurophysiol. 39, 282-300. Zarbin M. A., Warms@ J. K. and Kuhar M. J. (1981) Glycine receptor: Light microscopic autoradiographic localization with tritiated strychnine. J. Neurosci. 1, 532-547. (Accepted
I2 January 1987)
0306-4522/87 $3.00 + 0.00 F’crgamon Journals Ltd
GLYCINE IMMUNOREACTIVITY LOCALIZED IN THE COCHLEAR NUCLEUS AND SUPERIOR OLIVARY COMPLEX R. J. WENTHOLD,D. HUE, R. A. ALTSCHIJLER and K. A. RBEKS Laboratory of Neuro-otohuyngology, National Institute of Neurological and Communicative Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892, U.S.A. Akatrae-Polyclonal antibodies were made in rabbits against glycine conjugated to bovine serum albumin with glutaraldehyde and were used for lmmunocytochemical studies in the cochlear nucleus and superior olivary nucleus of the guinea-pig. Antibodies selective for glycine were prepared by affinity chromatography. By dot-blot analysis this preparation showed a strong recognition of glycine conjugates and relatively little recognition of conjugates of most other amino acids tested. However, there was a slgnitkant reaction with conjugates of ala&e and beta-alanine, and this cross-reaction could not be removed by tinity chromatography without eliminating the preparation’s recognition of glycine. The affmity-puriiied preparation showed only a weak recognition of conjugates of gamma-aminobutyrate (GABA) which was detectable at high concentrations of primary antibody. Immunocytochemical studies showed several intensely staining cell bodies in the cochlear nucleus and superior olivary complex. Most immunoreactive cell bodies in the cochlear nucleus were in the dorsal cochlear nucleus, being present in both the superflcial and deep layers. Scattered immunoreactive cells were ptesent in the ventral cochlear nucleus. Intense staining of cell bodies was seen in the medial nucleus of the trapezoid body, and these cells appear to correspond to the principal cells of that nucleus. Punctate labelling, suggestive of immunoteactive presynaptic terminals, was also apparent, particularly in the ventral cochlear nucleus and lateral superior olive. In the ventral cochlear nucleus, immunoreactive puncta were found around unlabeled cell bodies, at times nearly covering the perimeter of the cell. A population of glycine-immunoreactive cell bodies in the super8clal dorsal cochlear nucleus also labeled with anti-GABA antibodies as determined through double-labeling studies. However, glycine-positive cells in the deep dorsal cochlear nucleus were not labeled with anti-GABA antibodies, and some populations of GABA-positive cells in the supetlicial layers were not labeled with anti-glycine antibodies. In the hippocampus intense staining of cell bodies and puncta was seen with anti-GABA antibodies while essentially no staining was seen with anti-glycine antibodies. These results suggest that anti-glycine antibodies can be useful for immunocytochemical identification of glycinergic neurons. From this study several populations of putative glycinergic neurons am identified in the auditory nuclei of the brain stem using these antibodies. Some populations of GABA-containing neurons also contain high levels of glycine or a related molecule.
There is strong
cord.” High-affinity uptake of glycine has been demonstrated in the cochlear nucleus,m and endogenous glycine is released from cochlear nucleus slices in a calcium-dependent manner.% Glycine inhibits neurpharmacological experiments, indicate that gly- onal firing in the cochlear nucleus with iontophoretic cinergic synapses are most abundant in the spinal application, and strychnine reduces glycine-induced cord and brain stem. Results of several studies responses. 8f161(1 In the superior olivary complex strongly support a role for glycine as a neuro- (SOC), iontophoretic studies indicate that the neurtransmitter in auditory nuclei of the brain stem. The onal pathway originating in the medial nucleus of distribution of glycine shows a sevenfold concen- the trapezoid body (MNTB) and terminating in the tration range within the cochlear nucleus of the cat, ipsilateral lateral superior olive (LSO) may be glyand the highest levels are similar to those in spinal cinergic20 These results are supported by high levels of [3H]strychnine bindir#’ as well as immunocytochemical analysis showing heavy staining for Address for correspondence:Dr R. J. Wenthold, La&atory of Neuro-otolaryngology, Bldg 36, Room 5DO8, glycine postsynaptic receptors in the LS0.37 NIH, Bethesda, MD 20892, U.S.A. Until recently glycinergic synapses could not be Abbreviations: AVCN, anteroventral cochlear nucleus; readily identified using immunocytochemical techBSA, bovine serum albumin; DCN, dorsal cochlear niques. However, with the purification and develnucleus; FITC, fluorescein isothlocyanate; GABA. opment of monoclonal antibodies against the glycine gamma aminobutyrate; LSG, lateral superior olive; postsynaptic receptor,‘U6*27such synapses can now be MNTB, medlal nucleus of the trapexoid body; PVCN, posteroventral cochlear nucleus; SGC, superior olivary identified.32*”Our recent work using these antibodies complex; VCN, ventral cochlear nucleus. shows a widespread and dense distribution of glycine evidence that glycine is a major inhibitory neurotransmitter in the mammalian central nervous system.3*‘o*4’*4L45*46 Studies on the distribution and uptake of glycine, as well as results of
897
898
R. J.
WENTHOLD el al
receptors throughout the cochlear nucleus and SOC.*.” With the exception of the MNTB-LSO pathway, there are no indications concerning the identities of neurons giving rise to the presynaptic elements of these putative glycinergic synapses. Based on studies of other neurotransmitters, neurons releasing glycine could be studied by measuring selective biosynthetic enzymes for glycine or by measuring glycine itself. No enzyme selective for the neurotransmitter pool of glycine has yet been identified. However, the presence of high levels of free glycine may be a potential indicator of neurotransmitter glycine, since the distribution of glycine rather closely follows the reported neurotransmitter role of glycine This suggests that the in brain and spinal cord. 3~10.‘1 direct localization of glycine may be useful in identifying glycinergic neurons. The feasibility of immunocytochemical localization of neurotransmitters has been demonstrated for gamma-aminobutyrate (GABA) by Storm-Mathisen et al.” and confirmed Glycine immunoreactivity has by several others. 133’*y) recently been demonstrated in the central nervous system.639*35In the present study we have sought to produce selective antibodies against glycine conjugated to bovine serum albumin (BSA) and to use them to determine the distribution of glycine in the brain stem auditory
nuclei.
EXPERlMENTAL PUOCEDUPES
Glycinc was conjugated to BSA with glutarakkhy& using the method of Storm-Math&n er al.” Rabbits were injected with 2mg of glycincBSA conjugate in compkte Freund’sadjuvant. Booster injections of I mg of conjugate in incompkte Freund’sadjuvantwem given at 4 and 6 weeks and 6 and 12 months. Animals wem bkd 1 week aider the last three booster injections. while all bleeds gave antibodies that appeared to recognize conjugated glycine. antibodies obtained from the second bleed gave the most intense qxciiic staining with lowest background and was used in inununoeytochemicol studies preeentd here.Aritkenua was
at&t&y-purifkdby pasaagcthrough columns of amino acid conjugates. The tirst column contained glycine conjugated to ovalbumin with glutaraldehyde attached to cyanogen bmmide-activatedSepharose.Bound antibodies were duted with 0.1 M acetic acid, pH 2.0. and immediately neutral&d with potassium phosphate. This fraction was then paned through a column of GABA conjugated to BSA with
glutuakkhyde attached to cyanogcn bromide-activated !kpkoae. The unbound fraction was collected, BSA w-as added to a final concentration of 1?A, and stored at - 30°C. For do&k-labeling studies, antibodies against GABA c&ugatai to BSA with ghrtar&ehyde were made in guitxm-pigs following methods similar to those for producing m&t antil~~Iies in rabbits as previously deeaibai.q Booster mjections were given 2 and 4 we& aI?er the initial injection of antigen. Antibodies were. a!Bnity-purhhd by paaaing &rough a cohunn of gIycine conjugated to BBA with gItuaraldehyde attached to cyanogan Luankkaetivatal fiepharose.The unntained fraction was used for immunocytocbemkal analysis of GABA. Antibodies wcn character&d u&g immunoblotn of ovalimmin-wnjugata! amino acids. Four microllrpmr of
cowere applied to nitrocelluloae paper and incubatiata and waabes were done in Tris-buffered saline. Afker incubatin in primary antibody and washes, the nitroc&dose paper was incubated in goat anti-rabbit IgG
conjugated to peroxidase (Bio-Rad Laboratories) at a dilution of l/lSOO. For guinea-nix antibodies the avidin-biotin procedure was used.Peroxida& activity was visualized with 4-chloro-I-naphthol. For immunocytoehemical studies female NIH strain guinea-pigs (ISO-250g) were perfused with 5t&lOOml of phosphate-buffered saline. pH 7.4, followed by fixative containing 4% parafonnaldehyde and 0.25% glutaraldehyde in 0.1 M sodium cacodylate at 4°C. Tissue was postfixed for 1h at 4°C in the same fixative and then transferred to cold phosphate-buffered saline. Vibratome sections were cut at 50 pm. For immunocytochemical localization sections were preincubated in phosphate-buffered saline containing 10% normal goat serum for 1h. Sections were then incubated in primary antibody overnight at 4°C at dilutions ranging from l/60 to l/600. Subsequent steps followed the avidin-biotin method of Hsu et of.” (Vector Labs) and peroxidase activity was visualized using diaminobenzidine. For adsorption controls 200 ~1 of antibody (at l/20 dilution) was mixed with 100~1 of amino acid conjugate (4 mg protein/ml), diluted to 2 ml with phosphatebuffemd saline, and incubated for I h at room temperature. Immunocytochemical studies were then performed as described above. For double-labeling studies to localize glycine and GABA, tissue was incubated with both primary antibodies. Rhodamine-labeled goat anti-rabbit IgG and fluoreacein isothiocyanate (FlTQlabeled goat anti-guinea-pig IgG were used to localize both primary antibodies on the same tissue section. Controls were performed as outlined by Wessendorf and Elde.” It was found that the secondary antibodies cross-reacted to some extent with the inap propriate primary antibody; particularly, goat anti-rabbit IgG recognized guinea-pig antibodies. This cross-reactivity was eliminated by passing the goat anti-rabbit antibodies through a column of immobilized guinea-pig IgG and the goat anti-guinea-pig antibodies through a column of immobilized rabbit IgG. Double-labeled fhtorescent sections were examined with a Z&s Universal microscope with epifluorescence illumination using a 50 W mercury light source. For rhodamine, the Zeiss 487715 filter combination was used. For FITC, the Zeiss 487716 filter combination was used together with a 515-565 nm bandpass filter to control red emissions. RESULTS
Properties of anti-glyciite adbodies Immunoblot analysis of crude anti-glycine serum shows reaction with ovalbumin conjugates of several amino acids in addition to glycine (Fig, 1). Several approaches were tried to effectively &ity purify glycine-specific antibodies for use in immuaocytochemistty. Using a technique analogous to that successfully used for purifying anti-G-A antibodie~,~ the serum was passed tbrougb a column of immobilized GABA conjugated to BSA. T&I was effective in removing cross-rerrcting antibodies as determined with immunoblotting. However, wbik certain cell populations were intensely tinad with this affinity-pur#kd prepamtion, imm~nocytochemical studies showed a weak general thbding of most neuronal all bodies; St&ring of nuclei was greatest. Preincutuation of the antibody witb Washed membranes from guinea-pig brain decrea& this staining. It is po&blc that this baokgroand staining was due to antipresent in the rabbits before immunization with the glycine conjupte; however,
Glycine immuaoreactivity in brain stern
899
6 GABA GLY ASP GLU ALA B-ALA TAU GLN Fig. 1. Immuaoblot analysis of rabbit anti-glyciae and guinea-pig anti-GABA antibodies. Gvslbuaria conjugates of amino acids were applied to aitrocellulose and reacted with aatibodies. (1) Whole serum of rabbit anti-glyciae at a dilution of l/250. (2) Rabbit anti-glyciae which bound to glyciae-ovalbtia column and eluted with acid, l/250 dilution. (3), (4) Atliaity-purified anti-glyciae eluted from the glyciae-ovalbumia column sad unbound to the GABA-BSA column; this preparation was used for immuaocytochemistry. Dilutions of primary antibody were l/250 (3) and l/l00 (4). (5) Whole senaa of guinea-pig anti-GABA at a dilution of l/250. (6) AfEaity-purified guinea-pig aati-GABA at a dilution of l/250. Bound rabbit antibody is detected using atliaity-purified goat anti-rabbit immuaoglobulias conjugated to horseradish peroxidase, and bound guinea-pig antibody is detected using the avidia-biotia procedure. Peroxidase was visuslixed with 4-&loro-1-aaphthol.
the same pattern of staining was seen in all bleeds of both animals used to produce antibodies. These animals were immunized at different times. This background staining could be removed by passing the serum through a column of immobilized glycine conjugated to ovalbumin. The antibodies recognizing conjugated glycine were bound to the column and were eluted with weak acid. This fraction contained antibodies that recognize conjugates of glycine as well as those of other amino acids (Fig. 1). Passage of this fraction through a column of immobilized GABA conjugated to BSA removed antibodies recognizing conjugates of several other amino acids while retaining those which recognize glycine as determined by immunoblotting. However, antibodies recognizing alanine and beta-alanine were not removed by this treatment and could not be removed using columns of conjugates of these amino acids without eliminating selective staining for glycine (Fig. 1). The purified antibodies also showed a very minor recognition of GABA, which can be seen only using high concentrations of antibody (Fig. 1). This crossreactivity was not removed by additional passage through the GABA-BSA column.
Antibody binding to both immunoblots and tissue sections was eliminated by preincubation with glycine-BSA conjugates, while neither free glycine at concentrations up to 1 mM nor conjugates of GABA, ghmnnate and aspartate affected the binding (Fig. 2C, D). Several concentrations of glutaraldehyde were used for fixation, and 0.25% was used for all findings reported here. Lower concentrations gave substantially reduced staining, while higher concentrations did not increase specific staining and often increased background staining. Specific staining was not seen in the absence of glutaraldehyde. Antibodies made in guinea-pig against GABA conjugated to BSA were similar to those previously made in rabbits as determined by immunoblotting. Significant reaction was seen with beta-ala&e and no reaction with glycine was detected (Fig. 1). Cochlear nucleus Glycine-immunoreactive cell bodies were most abundant in the dorsal cochlear nucleus (DCN). Based on size at least two classes of labeled cell bodies were seen as shown in Figs 2 and 3. The population with the larger cell bodies, with average diameters at
Fig. 2.
Glycine immunoreactivity in brain stem
901
Fig. 3. Distribution of glycine-immunoreactive cell bodies in the JXN of the guinea-pig. Primary antibody was used at a dilution of l/200. Labeled cells are drawn from a single representative section through the DCN. Gpen circles indicate larger immunoreactive cells with average diameters of 16pm (n = 54), and closed circles show smaller cells with average diameters of 10 pm (n = 58). Asterisks show large immunoreactive cells occasionally found in deeper layers of the DCN. Diameters were obtained from immunoreactive cells with a visible nucleus.
16 pm, was contined mostly to the superficial layers. These cells were usually spherical and often large labeled processes extended from their soma. These cells are similar in shape, size and distribution to a population of DCN cells labeled with anti-GABA and anti-glutamate decarboxylase antibodies.“” In the rat DCN, glutamate decarboxylase-immunoreactive cells have been identified as cartwheel cells, stellate cells and Golgi cells.” Therefore, the present results suggest that one or more classes of these cells may contain high levels of glycine. The other class of glycine-immunoreactive cells is made up of smaller cells with average diameters of 10 pm. The cell bodies are spherical or irregular (Fig. 2B). They were most often found in the deeper layers of the DCN, but some smaller cells were also found in the superficial layers. Finally, a few labeled cells which may constitute an additional class were present in the deep DCN. These cells were larger than those present in the superficial DCN and were often seen with labeled processes. Immunoreactive labeling of all cells was eliminated with prior incubation of the antibody with glycin+BSA conjugate, but labeling was not a&c&l by preincubation with GABA-BSA conjugates (Fig.
2C, D). Fusiform cells and granule cells in the DCN were not stained with anti-glycine antibodies. Glycine-immunoreactive cell bodies were also numerous in the ventral cochlear nucleus (VCN Fig. 4A), unlike results obtained with anti-GABA antibodies.” Two types of glycine-immunoreactive cells were seen. Large cells usually containing labeled processes were most abundant near the eighth nerve root (Fig. 4B), and less frequently found in the posteroventral cochlear nucleus (PVCN) and the rostra1 anteroventral cochlear nucleus (A-VCN, Fig. 4C). These cells were among the largest in the VCN with diameters sometimes exceeding 25 pm. The second class of glycine-immunoreactive cells was made up of small spherical cells which were most abundant in the granule cell cap of the AVCN (Fig. 4D). These appear to be slightly larger than granule cells in this region, which can be distinguished by their very weak labeling with the antibody, but similar in sixe to the smaller immunoreactive cells in the DCN. Unlabeled cell bodies throughout the VCN were often discerned by the heavy covering with immunoreactive puncta which outlined most of the cell body of neurons in the AVCN (Fig. 5A). In the DCN
Fig. 2. Glycine4mmunoreactive labeling in the DCN. Immunoreactive cell bodies are present in the superlkial (A) and deeper (B) layers of the DCN. This labeling is not eliminated by prior incubation of the antibody with GABA-BSA conjugate (C), but is completely absent after incubation with glycine-BSA conjugate (D). Primary antibody used at dilutions of l/600 (A) and l/200 [(B)-(D)]. Bars = 20pm [(A), (B)l, 55pm Kc), @)I.
902
R. J. Wmmcm et al.
Fig. 5. Immunoreactive pun&
around unlabeled cell bodies in the rostrai AVCN {A) and f&form ceif layer of the DCN (IQ. Them may be immuno~~tive making synapses on c&i bodies. Primary antibody nsed at dilutions of I/Zoo. Bar = 6 em.
preaynaptic terminals
Fig. 4. Glycbimmuno~tive laheliug in the VCN. (A) !kattcred large immunorcactive c&s near the anditory nerve root. f3) Higher tnodific&ion of large and small immunomactive {C) Ilarge ~~~r~ve cell in eaudal AVCN. Labeled pun&a are present ~ou~out the section. (D) I~~o~c~ve cells in the gnu&e cell cap of the AVCX IntCn*iy iabeled small ceils @ITOWS) am shown with weakly labeled gramde celIs (aI_rowheads). Primary antibody used at diMions of I/200 [(A)-(C)] and l/60 (ID). Bars = 35 pm (A), 20 wn t(B), (C)l, I3 pm P).
dls.
Fig. 6 904
905
Glycine immunoreactivity in brain stem
such patterns were not as frequently seen, but some puncta were seen adjacent to cell bodies in the fusiform cell layer (Fig. 5B). While analysis at the ultrastructural level is required for identification, the labeled puncta appear to be immunoreactive presynaptic Wminals. The pattern obtained with the anti-glycine antibodies is similar to that obtained with anti-GABA antibodies. Electron microscopic studies have shown these puncta to be GABAimmunoreactive presynaptic terminals.@ Furthermore, studies in the spinal cord using an anti-glycine antibody similar to the one used here have confirmed through electron microscopy the immunoreactive labeling of presynaptic tennim~ls.~~ The widespread punctate labeling around cells in the VCN suggests that most major groups of large neurons in this region receive terminals containing glycine. A dense glycinergic input to cells in the AVCN and PVCN is consistent with the heavy labeling of these cells with anti-glycine receptor antib0dies.l Superior olivary complex Intense immunoreactive labeling was seen in the superior olivary complex (Fig. 6A), with the heaviest labeling in cells in the MNTB (Fig. 6B). The neurons of the MNTB have been described in the cat.‘lJ2 Three types of neurons are present in this nucleus with the principal neurons being the most numerous. These cells have a spherical or oval perikaryon and an eccentric nucleus. Based on distribution, size and shape, the principal neurons of the MNTB appear to be glycine-immunoreactive. Cell labeling is also present in the ventral nucleus of the trapezoid body, and scattered cell labeling is seen in the LSO. In addition to cell latiling, the LSO contains heavy punctate labeling around unlabeled cell bodies, resembling immunoreactive presynaptic terminals (Fig. 6C).
Relationship to GABA-immunoreactive labeling The population of glycine-immunoreactive cells in the superficial DCN appears similar to a population of cells labeled with anti-GABA and anti-glutamate decarboxylase antibodies.” To determine if cells are immunoreactive for both GABA and glycine, doublelabeling studies were performed using anti-glycine antibodies made in rabbits and anti-GABA antibodies made in guinea-pig. Fluorescent second antibodies were affinity-purified to eliminate cross-
reactivity. These studies indicate that some cells in the superficial DCN contain both immunoreactivities (Figs 7 and 8). Some cells in this area were immunoreactive only for GABA, many of ,Wse appear smaller than the co-labeled cells and may, represent a class of smaller GABA-containing cells found in the DCN (Fig. 7). Cells heavily labeled for glycine and not GABA were not routinely seen, although occasionally cells weakly immunoreactive for glycine and negative for GABA were observed (Fig. 8). Cells in the deep DCN were immunoreactive only for glycine. In addition to immunoreactive labeling of cell bodies in the superficial DCN, punctate labeling and fiber labeling are also present. While many of these structures appear immunoreactive for both GABA and glycine, some puncta were labeled with only one antibody (Fig. 8). This labeling was further investigated in the VCN. There, most punctate labeling showed only one immunoreactivity, but doublelabeled puncta were seen throughout the VCN. This double labeling was most commonly found in the PVCN (Fig. 9). Therefore, as is the case with cell bodies, pun&a in the cochlear nucleus can bc immunoreactive for GABA, glycine or both. The co-labeling of some neurons in the DCN with antibodies against GABA and glycine suggested that this may be a general property of GABAergic neurons. To address this question, GABA- and glycineimmunoreactive labeling was studied in other areas of the central nervous system. The difference in labeling patterns obtained with the two antibodies was most apparent in the hippocampus where intense neuron labeling was seen with anti-glycine antibodies (Fig. IO). The only glycine-immunoreactive labeling seen there was a very light labeling of scattered small neurons near the surface of the hippocampus (Fig. 10, arrows).
DISCUSSION These experiments show that antibodies made against conjugated glycine can be used for the immunocytochemical localization of glycine in the central nervous system. In determining whether or not anti-glycine antibodies are useful for studying the distribution and properties of glycinergic neurons, two questions must be addressed. The first is whether glycine itself is a valid marker for glycinergic neurons. Glycine, unlike GABA which has been shown to be
Fig. 6. Glycinaimmunorcactive labeling in the superior olivary complex. (A) Low magnification illustration showing intense labeling of cell bodies in the medial nucleus of the trapezoid body (M) and labeling of cells, puncta and fibers in the lateral superior olive (L). Bar = 150 pm. (LI)Higher magni6cation of labeled principal cells in the MNTB. Bar = 18 pm. (C) Higher magnification of labeling in the LSO showing scattered labeled cells (arrows). Puncta, which may represent immunoreactive preqnaptic terminals, are present throughout the LSO and are often seen outlining unlabeled cell bodies (asterisk). Bar = 18 pm. Primary antibody used at dilutions of l/200.
Fig. 7. Co-Iocaltition of gIycine and GABA ~rn~~or~~~t~~ in c&l bodies in the sqerki~ IXX. {A) GA3A~imm~no~a~i~ c&s labeled with ~inea-~i~ anti-GABA and FIX-goat anti-~i~ea-pig IgG. (B) Glycine-immunoreactivt: cells labeled with rabbit anti-glycine and rhodaminegoat anti-rabbit IgG. A cell body (arrowhead) labeled with antW3ABA antibodies is not labeled with anti-glycine antihodie’s. Primary antibodies were used at ditutions of l/60. Bar = IS pm.
a r&bie immu~oCytochemica1marker for GABAergiC~~~~~13,2~.25.~1,~ is involved in many functions in the centraE nervous system in addition to he&g a ~ur~~r~s~~~er. High Ievets of glydne could re@ect any one of these functions. Such may be true for glutamate and aspartate, since the immunocyto~hemi~af d~s~~b~~onsof these amino acids do IKP~ ~&r&e with neuron& ~~~a~~ons b&eved to be asp&&erg&zor ~u~~~Cr~c.** However, previous studies have shown that the distribution af glycine is more rest&M, and in areas where it is not thought to be a major ne~o~~tter, g&&e levels are much tower than in areas where it is believed to be a ~urot~~it~r,~,‘“*4* For exampIe, the: concentration of g&C&ein the Hindus is more than t&&ii lower than in the spina$ cord? *r immune~@oC&mk~f results support these Endings showing little glyGinGnmun0~Ctive staining in the hippocampus, In the brain stem large gradients of g&tine ~~~~o~a~tivjt~ are seen. Within the cochtear
nucleus and SOC only certain celi ~~p~l~~~~n~ are intensely immunore~tive whib others are: f25smi: dally unfab&ed, Thjs may suggest that the
leve1 in neurons neC&sary to support the nonneurotransmitter functions is rather uniform indifferent populations of neurons and that it is muck lower than levels found in neurons where g&&e is a 1 n~uro~~~t~er. The second question concerns whether or not. antibodies made against &Eycinea= &mtive em&&h to suggest that the substance lo&i%~.I is &&ne. From the ~rn~~~ob~~t data it appt%rs that antiglycine antibodies do not appreciably recogr&C GXXI&gates of several other amino acids including !J@ putative ~urotr~~~~~~~~ g&tam&e and abbe.. Ekwwer, thy do show a & astir of ajtz- -, @es of GABA, alanine:and heta-&nine. The crossreactivity with GABA could not be removed by additional passage through a column of GABA CWjag&e, and the ~r~~-~~~v~t~ with afanine and
GIycine
imnunoreactivity in brain
stem
907
Fiis 8. Co-Idtion of@yeine and GABA ~~0~~~ incm11 bodka and pacta inthe XX (A) GABA-hnnuan-ve IX& kbekd with guinea-pig %nti-GABA and FITC-goat anti ilyinea-pig IgG. (B) Glyeine immunorwctitity d&e&d with rabbit anti-glyeine and rho&n&e-goat
anti-rabbit I&k Closed arrowhead chaws B cell body lightly lab&d with arhglycine antibadjca and unfabebd with anti_GABA mtibodks. Arrows show puma positive for both antibodies and open arm&m& show pmcta immunoreective fix only one antibody. Primary aatibodies were med at d&tiom of Wi43.Bar = 15pm.
beta-riianine could noi be removed without destraying the ~rnun~t~~ty for g&tine. In the sense that they cannot distinguish between giycine, alanine and beta-alanine, the anti-glycine antibodies are simiXar to the giycine postsynaptic receptor. Ala-
nineand
beta-alanine are strong antago&s and compete with g&he for binding to the receptor.i**45 Therefore, in immunocytochemical studies some immunoreactitity may be duo to cram-reactivity of the antibodies with alanine and beta-&nine. Both of
908
R. J. WENTHOLD et al
Fig. 9. Co-localization of glycine and GABA immunoreactivities in the PVCN. (A) GABA immunoreactive labehng using guinea-pig anti-GABA and FITC-goat anti-guinea-pig IgG. (B) Glycineimmunoreactive labeling using rabbit anti-glycine and rhodaminegoat anti-rabbit IgG. Arrows show examples of puncta labeled with both antibodies, and arrowheads show puncta positive for only one antibody. Primary antibodies were used at diiutions of l/60. Bar = 9 pm.
Fig. 10. Comparison of inmunoreactive labeling in the hippocampus with anti-GABA (A) and anti-glycine (B) antibodies. Cell bodies and pun&a are inter&y labeled with anti-GABA antibodies while only vety weak staining of sotue oell bodies near the surface of the hippocampus @ITOW)is present with anti-glycine antibodies. Primary antibodies were rabbit anti-GABA at a dilution of 1/2WO and rabbit anti-giycine at a dilution ol l/200. P indicates pyramidal cell layer. Bar = 55 pm.
P
c
910
R. J.
WENTHOLD er al
these amino acids are rather abundant in the central nervous system; for example, in both whole brain and the cochlear nucleus, the concentration of alanine is about one half that of GABA.19.% The weak reaction of anti-glycine antibodies with GABA appears to be negligible for immunocytochemical studies since many neurons which label with anti-GABA antibodies are not labeled with anti-glycine antibodies; this possible relationship with GABA is discussed further below. Finally, the antibodies against glycine may cross-react with unknown large or small molecules in certain neurons. This, of course, must always be a concern with any immunochemical study. An unexpected finding of the immunocytochemical studies was that the anti-glycine antibodies label a population of neurons in the cochlear nucleus that also label with anti-GABA antibodies. Although double labeling was done only with anti-GABA antibodies, this same population of neurons has been previously shown to be immunoreactive for glutamate decarboxylase. ‘W However, anti-glycine antibodies do not label only GABAergic neurons and do not label all populations of GABA-containing neurons. Cells immunoreactive for glycine in the deep DCN and the large immunoreactive cells in the VCN do not label for GABA. The flattened cells and small round cells in the superficial DCN which are GABApositive do not appear labeled with anti-glycine antibodies. Therefore, it appears that only some cells that are GABAergic based on GABA and glutamate decarboxylase labeling in the cochlear nucleus are labeled with anti-glycine antibodies. This point is clearly illustrated in the hippocampus where cell bodies and puncta are intensely stained with antiGABA antibodies and unlabeled with anti-glycine antibodies. In the cerebellum Golgi cells are labeled with anti-glycine antibodies, while stellate and basket cells are unlabeled (Ref. 6 and unpublished observation). The cells of all three classes are labeled with anti-GABA antibodies and are thought to be GABAergic.*’ High-affinity uptake studies on cerebellar slices suggests there are two populations of Golgi cells based on their uptake preferences for glycine and GABA.” Whether glycine and GABA immunoreactivities are in the same or different populations of Golgi cells awaits double-labeling studies. There are several possible explanations for the apparent colabeling of some neurons with both antibodies. The anti-glycine antibodies may recognize GABA or the anti-GABA antibodies may recognize glycine. This is unlikely since very little cross-reactivity is evident from the immunoblots. Furthermore, only some populations of cells labeled by one antibody are also labeled with the second antibody. Therefore, a crossreactivity between GABA and glycine could be a reasonable explanation only if GABA or glycine is recognized differently in different populations of GABAergic or glycinergic neurons. For example, the type of macromolecule to which the amino acid is linked during fixation may influence its immuno-
reactivity. Alternatively, the amount of GABA may vary considerably between populations of GABAergic neurons, and only those with very high concentrations of GABA will be recognized by the anti-glycine antibodies. A second possibility is that one or both antibodies recognize a third molecule. Two candidates are alanine and beta-alanine with which both anti-glycine and anti-GABA cross-react to some extent. However, while the anti-GABA antibodies used in the present study were not purified to remove antibodies recognixing alanine and beta-alanine, earlier studies on the cochlear nucleus, using anti-GABA antibodies with these contaminating antibodies removed, labeled the same population of cells in the superficial DCN.” Furthermore, this same population of neurons wntains glutamate decarboxylase immunoreactivity.23~“’ Therefore, since the GABA immunocytochemical findings can be supported by the presence of glutamate decarboxylase, it suggests that the glycine antibody recognizes a molecule in some GABAergic neurons. A third possibility is that both GABA and glycine are present in these neurons at relatively high levels. Again, since glutamate decarboxylase is also present in these neurons, it would imply that a subpopulation of GABAergic neurons has high levels of glycine. This does not necessarily imply that both are acting as neurotransmitters, although both amino acids seem to he enriched in the same presynaptic terminals as well as the same cell bodies. The present results show that several populations of neurons in the cochlear nucleus and SGC are enriched in glycine immunoreactivity. The intense labeling of cells in the MNTB and of fibres and terminals in the LSO is consistent with the neuronal pathway originating with the principal cells. of the MNTB and terminating in the LSG being glycinergic, as was suggested by pharmacological studies on the chinchilla.M Experiments directed at localizing the glycine postsynaptic receptor using [‘I-IJstrychnine and antibodies against the postsynaptic receptor also show intense labeling in the L!3G.37” In the cochlear nucleus most glycine-immunoreactive cells are in the DCN. While the immunoreactive cells cannot be assigned to any morphological group based on the present studies, the fact that the neurons labeled in the superficial DCN are also labeled with antiGABA’O and anti-glutamate decarboxylase” antibodies indicates that these are either cartwheel, stellate or Golgi cells. The glycine-immunoreactive ceils in the deep DCN could correspond to any of the small cells that have been noted in this region.‘~” The fact that most glycine-immunoreactive cell bodies are found in the DCN suggests that glycine-containing presynaptic terminals in the VCN may originate with these cells. The distribution, size and shape of the large immunoreactive cells in the VCN indicates that they may be the giant cells which have been reported to project to the contralateral cochlear nucleus in the cat.‘*’ The
Glycine ~~o~~ty
properties of tbese large cells in the VCN would be consistent with glycine being their neurotransmitter. This pathway is reported to be inhibitory, and this inhibition has been found to be antagonized by ~~.‘~~’ However, anterograde transport studies showed that the input of tbis pathway to the cochlear nucleus was relatively sparse,’ suggesting that the abundance of glycinergic synapses in the cochlear nucleus, based on glycine-immunoreactive staining and glycine postsynaptic receptor distribution, is not primarily due to &is pathway.
in brain stem
911
does not appreciably recognize conjugates of other amino acid neurotransmitters. Such antibodies may then be useful in identifying glycinergic neurons. Based on staining with this antibody, we have identiged several ~p~tio~ of neurons in the co&ear nucleus and SGC that may use glycine as a neurotransmitter. However, questions remain concerning the labeling of some populations of GABAcontaining neurons with anti-glycine antibodies. This finding suggests that some GABAergic neurons also contain high concentrations of glycine or a similar substance.
CONCLUSIONS Our antibody preparation against glycine conjugated to BSA recognims conjugates of glycine but
thank Drs J. Fex, J. C. Adams and A. N. Van Den Pol for criticaBy reading this manuscript.
Acknowle&ements-We
1. Adams J. C. and Warr W. B. (1976) origins of axons in the cat’s acoustic stria determined by injeotion of horseradish peroxidase into severed tracts. J. cotup. Newof. 170, 107-122. 2. Altscluder R. A., Bets H., Pamkkal M. H., Reeks K. A. and Wenthold R. J. (1986) IdentiScatioa of glycinergic synapses in the co&ear nucleus through immunocytochemical Mcahxation of the postsynaptic ramptor. B&t Res. %9,31&O. 3. A~rison hf. H.. Da& E. C.. Shank R. P. and McBride W. J. (1975) Neurochemi& evidence for aivcine as a tnwmitter aid a model for its ~~p~~ ~~~~~ti~. In ~e~~i~ C~~~~ ~~~o~~~ (eds Beri S., Clarke D. D. and S&eider D.), pp. 37-63. Raven Press, New York. 4. bergerS. J., CarterJ. G. and Lowry0. H. (1977)The distributionof glycine, GABA, glutamateand sspartatein rabbit spinal cord, cerebellum and hippocampus. J. Neurochem. 2S, 149-158. 5. Brawer J. R., Merest D. K. and Kane E. C. (1974) The neuronal amhitecture of the cochlear nucleus of the cat. J. camp. Neural. l!%, 251-299. 6. Camp&on G., Buijs R. M. and Get&d M. (1986) Glycine neurons in the brain and spinal cord. Antibody production and ~~~~ loathration. Bruin Rez 376,400-405. 7. Cant N. B. and Gaston K. C. (1982) Pathways connecting the right and left cochkar nuclei. J. camp. Net&. 212, 313-326. 8. Caspary D. M., Havey D. C. and Faingold C. L. (1979) Effects of microiontophoretically applied glycine and GABA on net&ml response~pattems in the &chJear nuclei. &utir Ees. 17& 179-185. 9. Casoarv D. M.. Rvbak L. P. and Fainaold C. L. (1985) The effects of inhibitorv and excitatorv amino-acid ne&otr&smitte& on the response proper& of brain&m auditory neurons. In Awdiroiy B&tern&try &d. Dres&er D.), pp. 198-224. Charles C. Thomas, SpringBeld, Illinois. in the ~ oentral nervous system. I&u. 10. Curtis I), R. and Johnston G. A. R. (1974) Amino acid ~~~ Ply&d 69,98-188. 11. Godfrey D. A., Carter J. A., Bergcr S. J., Lowry D. H. and Matschinsky F. M. (1977) Quantitative histochemical mapping of candidate transmitter ammo acids in the cat co&ear nuckus. J. Hivtochem. Cytochem. U, 417431. 12. Graham D., PfeifRr F., Simmler R. and Betr H. (1985) Purifkation and characterization of the Blycine receptor of pig spinal cord. Biochemirtry 2& 990-997. 13. Hodgaon A. J., Per&c B., Erdei A., Chubb I. W. and Somo8yi P. (1985) Antisara to y-aminobutyric acid. f. Production and &ara&&ation using a new modes system. J. Histech. Cyrochem. 33,229-239. 14. Hsu S-M., Raine L. and Fanger H. (1981) Use of a~bio~~x~ complex (ABC) in i~~o~~~ techniques: A comparison between ABC and unlabeled antibody (PAP) mums. J. Histochem. Cytockm. 29, 577-580. 15. Lorente de No R. (1933) Anatomy of the eighth nerve. III. General plan of structure of the primary cochlear nuclei. taryngoscojre, St. Louis 43, 327-350. 16. Martin M. R. (1985) The pharmacology of amino acid receptors and synaptic transmission in the cochlear nucleus. In Atditary Biochemistry (cd. llkscb D.), pp. 184-197. Charks C. Thomas, Springfield, Bhnois. 17. Martin M. R., Dickson J. W. and Fex J. (1982) Biie, strychnine and depressant ammo acid msponses in the anteroventral co&ear nucleus of the cat. J. Nenropharma. 21,2OI-207. 18. Martin M. R. and Fenix L. P. (1983) Comparison of the effects of bkucu&ne and strychnine on brain stem auditory evoked potentials in the cat. Br. J. Pharmac. T&75-77. 19. McGccc P. L. and McGeer E. G. (1981) Amino acid neurotransmitters. In Bade Neurochembtry (cds Siegel G. J., Albers R. W., Agranoff B. W. and Katzman R.), pp. 233-253. Little, Brown & Co., Boston. 20. Moore M. J. and Caspary D. M. (1983) Strychnine blocks binaural inhibition in lateral superior olivary neurons. J. Neurosci. 3,237~242, 21. Moreat D. K. (1968) The collateral system in the media nucleus of the trapexoid body of the cat, its neuronal amhiteeture and m&ion to the oiivo-co&lear bundle. Bruin Res. $288-311. 22. More& D. K. (1973) Auditory neurons of the brain stem. Adu. Oto-rhino&r. 20, 337-356. 23. Mugnaini E. (1985) GABA neurons in the superg&! layers of the rat dorsal cochkar nucleus: Light and electron microsoopic immunochemistry. J. amp. Neural. 235, 61-81. 24. Gttemen 0. P. and Storm-Math&n J. (1984) Glutamate- and GABA-containing neurons in the mouse and rat brain, as demonstrated with a new immunocyto&emical technique. J. camp. Neurol. 229,374-392.
R. J.
912
WENTHOLD
eI al.
25. Ottersen 0. P. and Storm-Math&n
J. (1984) Neurons containing or accumulating transmitter ammo acids. In Vol. 3, Classical Transmitters and Transmitter Receptors in the CNS (eds BjBrklund A., Hiikfelt T. and Kuhar M. J.), pp. 141-246. Elsevier, Amsterdam. 26. Pfeiffer F., Graham D. and Betx H. (I 982) Purification by affinity chromatography of the glycine receptor of rat spinal cord. J. biol. Chem. 257, 9389-9393. 27. Pfeiffer F., Simler R., Grenningloh G. and Betx H. (1984) Monoclonal antibodies and peptide mapping reveal structural similarities between the subunits of the glycine receptor of rat spinal cord. froc. MIX Acad. Sci. U.S.A. 81, 72247227. 28. Pirsig W., Pfalx R. and Sandanaga M. (1968) Postsynaptic auditory crossed efferent inhibition in the ventral cochlear nucleus and its blocking by strychnine nitrate (guinea pig). Kumumo~o med. J. 21, 75-82. 29. Pourcho R. G. and Goebel D. J. (1985) Immunocytochemical demonstration of glycine in retina. Brain Res. 348, Handbook
of Chemical
Neuroanatomy,
339-342. 30. Schwartz 1. R. (1981) The differential distribution of label following uptake of ‘H-labeled amino acids in the dorsal cochlear nucleus of the cat. Expl Neural. 73, 601-617.
31. Sequela P., Geffard M., Buijs R. M. and Le Moal M. L. (1984) Antibodies against gamma-aminobutyric aicd: specificity studies and immunocytochemical results. Proc. noun. Acod. Sci. U.S.A. 81, 388-392. 32. St. John P., Owen D. G., Barker J. L., Pfeiffer F. and Betx H. (1984) Monoclonal antibodies to purified receptor bind to glycine receptors on mouse spinal neurons in culture. Neurosci. Abstr. 10, 7. 33. Storm-Math&n J., Leknes A. K., Bore A. J., Vaaland J. L., Edminson P., Huang F. M. S. and Ottersen 0. P. (1983) First visualization of glutamate and GABA in neurones by immunocytochemistry. Norure St, 517-520. 34. Triller A., Cluzeaud F., Pfeiffer F. V. and Kom H. (1985) Distribution of gJycine receptors at central synapses: an immunoelectron microscopy study. 1. Cell Biol. 101, 683-688. 35. Van Den Pol A. N. and Gores T. (1986) Glycine immunoreactive neurons and presynaptic boutons in the spinal cord. Neurosci. Abstr. 12, 771. 36. Wenthold R. J. (1979) Release of endogenous glutamic acid, aspartic acid and GABA from cochlear nucleus slices. Brain Res. 162, 338-363. 37. Wenthold R. J., Betx H., Reeks K. A., ParakkaJ M. H. and AJtschuJer R. A. (1985) Lo&x&ion of glycitxqic synapses
38.
39. 40. 41. 42. 43. 44. 45. 46. 47. 48.
in the cochlear nucleus and superior olivary complex with monoclonal antibodies ape&c for the erycine receptor. Neurosci. Abstr. 11, 1048. Wenthold R. J. and Gulley R. L. (1977) Aspartic acid and glutamic acid levels in the cochkar nucleus after auditory nerve lesion. Brain Res. 138, 111-123. Wenthold R. J. and Martin M. R. (1984) Neurotranamitters of the auditory nerve and central auditory system. In Hearing Science (cd. Berlin I.), pp. 341-369. College-Hill Press, San Diego. Wenthold R. J., Zempel J. M., Parakkal M. H., Reeks K. A. and Altachuler R. A. (1986) Immunocyto~hamical localization of GABA in the cochlear nucleus of the guinea pig. Brok Res. 3sc, 7-18. Werman R. and Aprison M. H. (1966) Glycine: The search for a spinal cord inhibitory tranarnitter. In Structure ad Function of Inhibitory Neuronal Mechanisms (cds von Euler C., Skoglund S. and Swerbag U.), pp. 473-486. Pargamon Press, New York. Werman R., Davidoff R. A. and Aprison M. H. (1967) Inhibition of motoneurones by iontophoresis of glycine. Norure, Land. 214, 681683. Wessendorf M. W. and Elde R. P. (1985) Characterization of an immunotluorescence technique for the demonstration of coexisting neurotransmitters within nerve Bbres and terminals. J. Histochem. Cytoekm. 3& 98c994. Wilkin G. P., Csillag A., Balax~~R., King&uy A. E., Wilson J. E. and Johnson A. L. (1981) Localization of high a!Bnity 3H-glycine transport sites in the cereb&r cortex. Brain Res. 216, I l-33. Youna A. B. (1984) Glvcine recrmtors in the nervous svsttan. In Brain Recepror hfetho&loaies: Part B Amino Acti, Pep&, Psychwciive brugs (A Marangos P. J., Campbell I. C. and Cohen R. M.), pp 37-57. Academic Rass, Orlando. Young A. B. and McDonald R. L. (1983) Glycine as a spinal cord neurotranamitter. In Hun&ok of the Spind Cord, Vol. I (ed Davidoff R. A.), pp. l-44. Marcel Dekker, New York. Young E. D. and Brownell W. E. (1976) Responses to tones and noise of single cells in dorsal cochkar nucleus of unanesthetixed cats. J. Neurophysiol. 39, 282-300. Zarbin M. A., Warms@ J. K. and Kuhar M. J. (1981) Glycine receptor: Light microscopic autoradiographic localization with tritiated strychnine. J. Neurosci. 1, 532-547. (Accepted
I2 January 1987)