Immunocytochemical Localization of G lutamate Decarboxylase in Rat Spinal Cord BARBARA J. M c L A U G H L I N , ~ ROBERT BARBER,2 KIHACHI SAITO,* EUGENE ROBERTS,Z3 AND J. Y. W U 2 Division of Neurosciences, City of Hope National Medical Center, Duarte, California 91010 and 1 Department of A n a t o m y , University of Tennessee Center f o r t h e Health Sciences, Memphis, Tennessee 38163
2
ABSTRACT The GABA synthesizing enzyme, glutamate decarboxylase (GAD), has been localized by light and electron microscopy in the rat lumbosacral spinal cord using a peroxidase-labeling antibody technique. The light microscopic localization shows heavy, punctate reaction product for GAD in the dorsal horn laminae 1-111. Moderately heavy reaction product is also seen in the deeper dorsal horn laminae IV-VI, the medial aspect of the intermediate gray (lamina VII) and the region around the central canal (lamina X). A moderately light concentration of GAD reaction product is observed in the ventral horn, and punctate deposits of reaction product also are seen on motoneuron cell bodies. The punctate distribution of reaction product for GAD in both ventral and dorsal horns, as visualized by light microscopy, corresponds to GADcontaining synaptic terminals seen by electron microscopy in comparable regions of the spinal gray. Many more GAD-positive terminals are observed in dorsal horn laminae 1-111 than in deeper laminae IV-VI. GAD-containing terminals in the dorsal horn are presynaptic to dendrites and cell bodies. GADcontaining terminals presynaptic to other axon terminals are observed also, and they are more numerous in laminae I1 and 111. In the ventral horn motor nuclei, GAD-psitive knobs are presynaptic to large and small dendrites and motoneuron cell bodies. In addition, small GADcontaining terminals also are presynaptic to larger axonal terminals which are in turn presynaptic to motoneuron somata. The observation of GAD-containing terminals presynaptic to dendrites and cell bodies in both dorsal and ventral horns is compatible with the evidence suggesting that GABA terminals may mediate postsynaptic inhibition of spinal interneurons and motoneurons. The additional finding of GAD-positive terminals presynaptic to other axonal terminals in the dorsal horn and motor nuclei is consistent with the growing evidence that GABA also may be the transmitter mediating presynaptic inhibition via axo-axonal synapses i n the spinal cord.
There is growing evidence that 7-aminobutyric acid (GABA) is a major neurotransmitter in the vertebrate central nervous system. A good deal of this evidence has evolved from biochemical, physiological and morphological studies on the cerebellum, which have received further support from the recent studies locating the GABA synthesizing enzyme, glutamate decarboxylase (GAD, E.C., 4. 1. 1. 15), by an immunoperoxidase method (Saito et al., '74; McLaughlin et al., '74). In those studies, GAD was localized in inhibitory and presumably GABA synaptic terminals in the cerebellar cortex and deep cerebellar nuclei. In the present study, we extend this J. COMP. NEUR.,164: 3 0 5 3 2 2
immunocytochemical technique to another area of the CNS, the spinal cord, where GABA is thought to mediate two types of spinal inhibition. Biochemical studies have shown that both GABA and its synthesizing enzyme, GAD, are present in the dorsal and ventral horns of the spinal gray, but are more concentrated in the superficial laminae of the dorsal horn (Albers and Brady, '59; Graham and Aprison, '69; Miyata and Otsuka, '72). Further studies have sug3 Please send reprint requests to: Dr. Eugene Roberts, Division of Neurosciences, City of Hope National Medical Center, 1500 E. Duarte Road, Duarte, California 91010.
305
306
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
gested that GABA is associated with interneurons located in the dorsal horn mediating presynaptic inhibition of primary afferent terminals (Eccles et al., '63; Schmidt, '64; Barker and Nicoll, '73; Miyata and Otsuka, '72; Otsuka and Konishi, '75). Other studies investigating the pharmacological action of GABA in the ventral horn have suggested that GABA is associated with interneurons mediating postsynaptic inhibition of motoneurons (Curtis et al., '68a; Curtis, '69). The present study examines the light and electron microscopic distribution of GAD-containing synaptic terminals in the rat lumbosacral spinal cord, with particular attention given to the localization of GADpositive terminals in dorsal horn laminae I-VI and in the motor nuclei of the ventral horn. MATERIALS AND METHODS
GAD was purified from the mouse brain synaptosomal fraction. Production of antibody was achieved by four bi-weekly injections of 150 fig of GAD into rabbit infkascapular muscle as described previously (for details, see Wu et al., '73; Saito et al., '74). Horseradish peroxidase (Type VI, Sigma), goat anti-rabbit IgG (Miles Laboratories), 3,3'-diaminobenzidine . 4 HCl (Sigma), acetone-dried mouse liver and mouse brain and spinal cord powders were all used in the immunocytochemical procedures. Goat anti-rabbit IgG was labeled with peroxidase as described previously (Saito et al., '74). Glutaraldehyde ( 8 % ) was purchased from Polysciences. All other chemicals were reagent grade.
Imm unochemis try procedure To remove the non-specific staining involved in the immunocytochemical procedures, 0.2 ml of anti-GAD serum or serum from non-immune rabbits was pretreated with a mixture of 75 mg liver and 25 mg spinal cord acetone powders. Peroxidaselabeled goat anti-rabbit IgG (0.2 ml containing 9 mg of protein) was treated with a mixture of 75 mg brain and 25 mg spinal cord powders. All steps were performed with a constant gentle agitation at room temperature. Tissue slices were incubated with 4-fold diluted rat serum for ten minutes prior to the experiment. The spinal cord slices were then incubated for 30 minutes with a 6-
fold diluted anti-GAD or control serum which contained goat IgG (1 mg/ml) and 0.1% Triton X-100 (Hartman, '73). The slices were washed with phosphate buffered saline (PBS) for three hours and then incubated for 30 minutes with peroxidaselabeled goat anti-rabbit IgG (1.2 mg/ml) containing 0.1% Triton X-100. After washing with PBS for another three hours, the slices were incubated with 3,3'-diaminobenzidine . 4 HCl (30 mu50 ml PBS) containing H202(0.006%) for ten minutes. The slices to be used for light microscopy were washed in PBS overnight at 4°C; the slices for electron microscopy were washed one hour in PBS prior to osmication.
Light microscopy Young adult Sprague-Dawley rats were anesthetized with 35% chloral hydrate (1 mVkg body weight) and fixed by vascular perfusion with 600 ml of 4% paraformaldehyde, in 0.12 M Millonig's ('61) buffer and 0.002% CaC12,pH 7.2-7.4 at 4°C. The animals were placed in plastic bags in the refrigerator for two hours, after which transverse segments of lumbosacral spinal cord were dissected out and placed in the above fixative for 48 hours at 4°C. The segments of spinal cord were then sliced transversely into 2 mm lengths and placed in separate, numbered containers in rostra1 to caudal order. Prior to cryotomy, the tissue was infiltrated with 30% sucrose for 24 hours. The tissue was then placed on a chuck, surrounded by 30% sucrose, and then quick-frozen in isopentane which was cooled by liquid nitrogen. The chuck and tissue were then transferred to the cryostat and allowed to equilibrate to - 20°C before cutting 40 f i thick transverse sections of the entire spinal cord face. The sections were picked up from the blade with a warm, moist slide and collected in 37°C PBS, and were washed overnight in a large volume of PBS at 4°C. The unmounted 40 p tissue sections were rinsed again in PBS and then processed for the immunocytochemical procedure. Following immunochemical treatment, the tissue was rinsed in distilled water, postfixed in 0.01% aqueous OSmium tetroxide for 30 seconds, rinsed in distilled water and mounted on a slide in glycerin. For more permanent storage, sections were dehydrated, cleared in xylene, and mounted in Permount.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
Electron microscopy Adult Sprague-Dawley rats were anesthetized with an intraperitoneal injection of 35% chloral hydrate (1 mYkg body weight) and then perfused through the heart with a fixative containing 4% paraformaldehyde and 0.1% glutaraldehyde in 0.12 M Millonig's ('61) phosphate buffer, pH 7.2, with 0.002% CaClz added. After the perfusion, the lumbosacral spinal cord was removed and stored overnight at 4 C in 4 % phosphate-buffered paraformaldehyde without glutaraldehyde. The spinal cord was sectioned transversely into 75 p slices with a Sorvall TC-2 tissue sectioner and collected in PBS in preparation for the immunocytochemical procedure described above. After processing the tissue for immunochemistry, the control and experimental transverse slices were hand cut into quarters so that each quarter contained a ventral or a dorsal horn. The quarter slices were then postfixed one hour in 2% osmium tetroxide in 0.12 M Millonig's buffer, pH 7.2, followed by rinses in 30% acetone and distilled water and block staining in 2% aqueous uranyl acetate for 20 minutes. The slices were dehydrated in graded alcohols, rinsed in propylene oxide and flat embedded in EponAraldite. The dorsal horn slices were embedded in either of two ways: (1) with the transverse slide face down in Beem capsule caps into which the cut apical end of the capsule is inserted; or (2) in silicone rubber molds with the medial sagittal edge of the dorsal horn forming the block face. The ventral horns were flat embedded only in the Beem capsules. Ultrathin sections of dorsal horns embedded in capsule caps were cut with a diamond knife in the transverse plane of the cord so that several laminae were in the same section; dorsal horns embedded in rubber molds were cut parasagittally so that all laminae were present in the same section; ventral horns were cut transversely so that the entire thin section contained only motor nuclei. Ultrathin sections were stained with lead citrate (Venable and Coggeshall, '65) and examined in either a Hitachi 11B or Philips EM 301 electron microscope. Cytoarchitectural lamination in the rat spinal cord was determined according to the procedure and criteria of Rexed ('52, '54) using paraffm embedded sections of O
30 7
the lumbosacral region stained with Luxol Fast Blue and cresyl violet or cresyl violet alone. In preparation for electron microscopy, photomicrographs of these stained sections were enlarged photographically and approximate measurements were obtained for each dorsal horn lamina, which were then used to identify the various laminae in thin sections. Due to the limited penetration of the reagents used to demonstrate reaction product for GAD in tissue slices embedded for electron microscopy, no attempts were made to quantitate the number of labeled terminals observed in each lamina. In addition, this penetration variability probably accounted for differences in the electron density of reaction product around synaptic vesicles and in terminals of any given field and, as a result, necessitated a qualitative rather than a quantitative analysis of the label. RESULTS
Lzg ht microscopy The dorsal gray matter of the rat spinal cord contains a higher concentration of GAD reaction product than the ventral spinal gray. Reaction product is heavily concentrated in the dorsal horn laminae 1-111 and two intense bands of GAD product are seen in this region. One dense band of punctate reaction product extends from the medial aspect of the dorsal horn to the most lateral aspect (fig. 1) and conforms to Rexeds lamina I11 medially and lamina I1 laterally. A thinner dense band follows the layer of marginal cells, conforming to lamina I of Rexed. The remaining dorsal horn (laminae IV-VI), the medial aspect of the intermediate gray (lamina VII), and the area around the central canal (lamina X) appear to have moderately heavy concentrations of punctate GAD-positive product (figs. 6, 7). The more lateral regions of the intermediate spinal gray (fig. 9) and the ventral horn motor nuclei (lamina IX) (fig. 8 ) exhibit a moderately light concentration of GAD-positive product. The remaining area of the ventral spinal gray has only light concentrations of GAD-positive product (fig. 10). Most of the neurons in the spinal cord have some punctate reaction product for GAD associated closely with their somata and large dendrites (fig. 9). The neuropil
308
McLAUGHLJN. BARBER, SAITO, ROBERTS, WU
throughout the spinal cord exhibits G A D positive product in different concentrations that vary roughly by region as illustrated in the diagram in figure 3. The region around the central canal exhibits a moderately heavy concentration of GAD reaction product. At higher magnifications, a few of the cells near the central canal appear to project an apical process into the central canal and resemble the cell bodies of spinal CSF-contacting neurons (Vigh and Vigh-Teichmann, '73). These cells have several GAD-positive deposits on their somata (fig. 7), which may correspond to GADcontaining synaptic terminals on the somata of these specialized neurons. In contrast, the motor nuclei are a moderately light concentration area, but the motoneurons themselves have punctate GAD deposits on their cell bodies (fig. 8). The punctate distribution of GAD product in both ventral and dorsal horns as visualized in the light microscope corresponds to the GAD-containing synaptic terminals seen with the electron microscope in comparable regions of the spinal gray (see Electron microscopy). No reaction product for GAD is seen in the control spinal cord slices treated with non-immune serum (fig. 2) and prepared for light or electron microscopy. Electron microscopy In the experimental tissue, electronopaque reaction product for peroxidaselabeled GAD is localized in synaptic terminals. The reaction product is usually aggregated around synaptic vesicle and mitochondrial membranes and along the presynaptic membrane specialization of each labeled terminal. Some of the GADpositive terminals in the rat spinal cord contain round synaptic vesicles (fig. 17) and make asymmetrical synaptic contacts (fig. 18), but the majority of GADpositive knobs contain pleomorphic or flattened vesicles and make symmetrical synaptic contacts (figs. 11, 12). Many more labeled synaptic terminals are seen in the superficial dorsal horn laminae 1-111 than in the deeper laminae IV-VZ, which corresponds with the light microscopic observations described above. Many of the labeled terminals in laminae 1-111 are presynaptic to dendrites but GADpositive terminals that are presynaptic to
cell bodies are also observed (figs. 11-13). In the transversely sectioned dorsal horns, GADpositive knobs which are presynaptic to other axonal terminals are observed occasionally in laminae I1 and 111. When the dorsal horns are sectioned along the sagittal plane, many more axo-axonal synaptic complexes are observed in these two laminae, in which GAD-containing terminals are presynaptic to other axonal terminals which are, in turn, presynaptic to other dendrites (figs. 14-16). The GAD-containing presynaptic terminals in these axoaxonal complexes always contain flattened or pleomorphic vesicles and appear to make symmetrical synaptic contacts with the postsynaptic terminals. The postsynaptic terminals of these complexes, on the other hand, usually contain rounded vesicles and make asymmetrical synaptic contacts with other profiles. GAD labeled synaptic terminals are not Fig. 1 Light photomicrograph of the dorsal horn from a rat lumbosacral spinal cord slice treated with immune serum. Arrows point to intense band of punctate GAD positive product in the substantia gelatinosa. Arrowheads point to GAD product in marginal layer. X 70. Fig. 2 Control dorsal horn from rat lumbosacral spinal cord treated with non-immune serum. Dashed line outlines the dorsal horn, X 70. Dense bodies are red blood cells. Fig. 3 The diagram shows different spinal cord regions numbered 4 through 10 which represent different concentrations of reaction product for GAD. Corresponding light photomicrographs from these regions are numbered accordingly. Long arrows represent GAD positive product in neuropil; Short arrows represent GAD positive product associated with neuronal somata. n., neuron; d., dendrite; cc, central canal. Figures 4-10 magnified X 1,000. Fig. 4 Neuropil with intense concentration of GAD product (arrows) &om lateral substantia gelatinosa. Fig. 5 Neuropil with intense concentration of GAD positive product (arrows) &om medial substantia gelatinosa. Fig. 6 Neuropil with moderately heavyconcentration of GAD positive product (arrows). Fig. 7 Region around central canal (cc). Punctate reaction product for GAD (arrows) is associated with CSF contacting neuron (n). Fig. 8 Motor neuron with punctate reaction product (arrows) studding the soma1 surface. Fig. 9 Neuron in region of moderately light concentration of reaction product. Both the soma (n) and the dendrite (d) of this neuron are contacted by punctate reaction product for GAD (arrows). Fig. 10 A n area from the spinal cord region where the concentration of GAD positive product (arrows) is the lowest.
IMMUNOCYTOCHEMICAL LOCALIZATION O F G A D I N SPINAL CORD
309
3 10
McLAUGHLIN, BARBER, SAITO, ROBERTS, W U
Fig. 11 Electron micrograph from the substantia gelatinosa (SG) (lamina 11) of spinal cord slice treated with immune serum. Numerous electron-opaque (GAD-positive synaptic terminals (arrows) are seen in this region which correspond to the heavy punctate reaction product seen in figures 4 and 5. Other adjacent terminals i n the field are not GADpositive. X
13,100.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD I N SPINAL CORD
311
Fig. 1 2 Higher magnification of GAD-positive terminals (arrows) in substantia gelatmosa (SG). X 21,800.
as numerous in the deeper dorsal horn laminae IV-VI. Both axosomatic and axodendritic GAD-positive presynaptic knobs are observed (fig. 19). Occasional G A D positive presynaptic axo-axonal contacts are observed, such as in figure 20, in which two GADcontaining terminals in lamina IV are presynaptic to a larger axonal terminal, which is not GADpositive and which is presynaptic to another axonal terminal and a dendrite.
In the ventral horn motor nuclei, GADcontaining synaptic terminals are less numerous than in the dorsal horn. Presynaptic GAD-positive terminals are observed on large and small dendritic profiles (fig. 21) as well as on motoneuron cell bodies. Some of the labeled knobs appear to be presynaptic to larger axonal terminals which synapse on motoneuronal somata (fig. 22). These terminals which are postsynaptic to the smaller GADpositive knobs
312
M c L A U G H L I N , BARBER, SAITO, ROBERTS, WU
Fig. 13 Numerous GADcontaining synaptic terminals (arrows) on neuronal cell bodies (N) and dendrites (D) in dorsal horn lamina 111. x 16,300.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
313
Fig. 14 Axo-axonal synapse in substantia gelatinosa (lamina 11) in which a GAD-positive terminal (T,) is presynaptic (small arrow) to another terminal (T2) which is not GAD-positive but which is itselfpresynaptic (large arrow) to a dendrite (D). x 52,400.
Fig. 15 Axo-axonal synapse in lamina 111 in which a GAD containing terminal (TI) is presynaptic (small arrow) to a n unlabeled terminal (T2) which is presynaptic (large arrows) to several dendrites (d). X 66,900.
314
McLAUGHLIN, BARBER, SAITO, ROBERTS, W U
Fig. 16 Axo-axond synapses located a t the interface between lamina I1 and 111 in which which two GADpositive terminals (TI) (T2) a r e presynaptic (small arrows) to a terminal (Tn) is not GAD-positive a n d is presynaptic (large arrows) to two dendrites (d). x 79,400.
Fig. 1 7 T w o GAD-positive terminals (tl) (h) i n lamina I1 which make symmetrical synaptic contacts with a dendrite (D). One GAD terminal (t,) contains predominantly rounded vesicles. Another rounded vesicle terminal (ta) makes a n asymmetrical contact with the same dendrite a n d is not GAD-positive. x 35,800. Fig. 18 GAD containing terminal t, in lamina I, which contains rounded synaptic vesicles a n d makes asymmetrical contacts (arrows) with a dendrite (d). An adjacent rounded vesicle terminal ( t 2 ) is not GAD-positive. x 43,700.
IMMUNOCYTOCHEMICAL LOCALIZATION O F GAD IN SPINAL CORD
315
Fig. 19 Fewer GAD-positive terminals (arrows) are seen in dorsal horn lamina IV than i n lamina-111. One terminal i n the field is presynaptic to a neuron soma (N) and another is presynaptic to a dendrite (D). A large terminal (t) i n the vicinity is not GAD-positive. X 12,500.
containing flattened vesicles, are characteristically large in size, contain rounded vesicles and make multiple synaptic contacts with the motoneuron, in which numerous dense bodies are associated with the subsynaptic membrane (fig. 22). Such terminals are thought to be of dorsal root origin (Conradi, '69b; McLaughlin, '72). DISCUSSION
Although GAD and GABA levels are comparatively lower in the spinal cord than
in other regions of the central nervous system (Albers and Brady, '59; Okada et al., '71; Fahn, '75) the relatively high concentration of GABA and its synthesizing enzyme in the dorsal horn, nevertheless, has led to a consideration of this amino acid as a possible neurotransmitter in the spinal cord. Pharmacological studies (Curtis et al., '68a,b; Curtis, '69) have demonstrated that iontophoretically applied GABA has a hyperpolarizing effect on both interneurons and motoneurons in the dorsal and
3 16
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
Fig. 20 Axo-axonal synapses i n lamina IV i n which two GAD-positive terminals ( t l ) (f) appear to be presynaptic (small arrows) to a large terminal (t3) which is not GAD-positive and which is uresvnautic (large arrows) to another axonal terminal (t4) and dendrite (d).
ventral spinal gray and suggest that GABA may function as a postsynaptic inhibitory transmitter in these regions via axodendritic and axosomatic synapses. Similar studies (Eccles et al., '63; Schmidt, '64; Barker and Nicoll, '72, '73; Davidoff, '72; Levy, '74) have demonstrated that GABA has a depolarizing effect on dorsal root fibers and suggest that GABA may act also to mediate presynaptic inhibition of primary afferent terminals via axo-axonal synapses, in addition to mediating postsynaptic inhibition on spinal interneurons and motoneurons. The present immunocytochemical localization of GAD in the rat lumbosacral spinal cord reveals a high concentration of synaptic terminals containing GAD in the outermost dorsal horn laminae 1-111 and fewer GADcontaining terminals in the deeper laminae IV-VI. This finding correlates well with the biochemical evidence that both GAD and GABA are more concentrated in the dorsolateral region of the dorsal horn (Albers and Brady, '59; Miyata and Otsuka, '72; Otsuka and Ko-
nishi, '75), as well as with recent autoradiographic work (Ljungdahl and Hokfelt, '73) showing intense uptake of [3H]-GABA by boutons in laminae 1-111. The majority of these GADpositive terminals in the dorsal horn laminae are presynaptic to dendrites and neuronal cell bodies and may belong to neurons mediating postsynaptic inhibition in the dorsal horn (Curtis et al., '68b). It has been suggested that the GABA interneurons mediating postsynaptic inhibition in the spinal cord are located in the dorsal regions of the cord (Curtis et al., '71), but other workers (Davidoff et al., '67) have suggested that GABA is not concentrated in spinal interneurons and instead may be the inhibitory transmitter of supraspinal structures which terminate in the spinal gray. While an investigation into the origin of GAD-containing or GABA terminals in the spinal cord is of utmost importance, the immunocytochemical method used in this study is not sufficiently sensitive to detect GADcontaining cell bodies and axonal pathways, which would allow such GABA neurons to be properly identi-
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
317
Fig. 21 Large dendrite (D) in the ventral horn motor nucleus surrounded by synaptic terminals, two of which are GADpositive (arrows). x 12,500.
fied. Future studies combining surgical ablation of certain spinal cord pathways with the GAD immunocytochemical technique, however, may help to determine whether they are indeed of supraspinal or interneuronal origin. For example, spinal cord transections could be done to determine if the GAD-containing terminals (below the transected cord) are from neurons in supraspinal regions. If there is a considerable decrease in the number of observed GAD-positive terminals below the transected cord, then it could be concluded that many of the GAD-containing and presumably GABA terminals are of supraspinal origin. Similar experiments, such as combined dorsal root (Barber et al., '74) and cord transections above and below the spinal cord level to be examined,
could be done to determine if the GADpositive terminals belong primarily to spinal interneurons. The drop in glycine concentration and not GABA, following hypoxic destruction of ventral horn spinal interneurons has led to the suggestion that glycine is released by interneurons located primarily in the ventral horn and is the major transmitter mediating segmental spinal inhibition (Davidoff et al., '67). Further studies using presumed antagonists to glycine and GABA (Curtis et al., '71) suggest that prolonged inhibition of motoneurons appears to be mediated by GABA inhibitory terminals on motoneuronal dendrites or on excitatory terminals which are presynaptic to motoneurons, whereas glycine inhibitory terminals are located predominantly on
3 18
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
Fig. 22 A small GAD-containing terminal (t,) presynaptic (small arrow) to a larger terminal (tz) which is not GAD-positive and which is presynaptic (large arrow) to a motoneuron (MN). X 55,500.
motoneuron cell bodies. The immunocytochemical localization of GAD-containing terminals on dendrites in the motor nuclei and on axonal terminals presynaptic to motoneurons is consistent with this latter proposal, while the relatively large number of GAD-containing terminals observed synapsing with motoneuron somata suggests that GABA neurons, as well as glycine neurons may play an important role in the inhibitory events taking place on motoneuron somata. The immunocytochemical localization of GAD in terminals which are presynaptic to other axonal terminals in both the dorsal horn laminae and ventral horn motor nuclei is consistent with the growing evidence that GABA is the transmitter mediating presynaptic inhibition of primary afferent terminals (Eccles et al., '63; Schmidt, '64; Curtis et al., '71; Davidoff, '72; Davidoff et al., '73; Barker and Nicoll, '72, '73; Miyata and Otsuka, '72; Levy, '74; Otsuka and Konishi, '75). The largest number of these GAD-containing presynaptic components of axo-axonal synapses are observed in dorsal horn laminae I1 and 111. This observation is of interest with regard to the studies of Miyata and Otsuka ('72) and Otsuka and Konishi ('75), in which GABA levels in the dorsolateral region of the dorsal horn are markedly de-
creased along with the size of the dorsal root potentials after hypoxic destruction of spinal interneurons in that area and which suggest that the GABA interneurons acting in presynaptic inhibition are located in the superficial dorsal horn laminae. Rethelyi and Szentagothai ('69) have suggested that other neurons located in deeper dorsal horn laminae may also be presynaptic to primary afferent terminals and function in presynaptic inhibitory pathways. In order to satisfy this structural prerequisite for presynaptic inhibition of an axonal terminal synapsing on a dorsal root terminal, as well as to strengthen the mounting evidence that GABA is the presynaptic transmitter, we currently are studying the immunocytochemical distribution of GAD in the spinal cord after dorsal rhizotomies (c.f. Barber et al., '74), to determine whether some of the GADcontaining terminals are in fact presynaptic to primary afferent terminals. Heimer and Wall ('68) have shown heavy dorsal root degeneration in the same dorsal horn laminae 11 and I11 of the rat spinal cord where GAD-containing terminals are the most numerous. Future studies combining dorsal rhizotomies and GADimmunocytochemistry (Barber et al., '74) will determine if GADpositive terminals are indeed presynaptic to some of these
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
degenerating dorsal root terminals and whether or not there is a regional difference in the axo-axonal synaptic arrangements as Ralston ('68) has shown in the cat dorsal horn, in which degenerating dorsal root terminals in laminae I11 are presynaptic to other axonal terminals but in deeper laminae IV-VI are postsynaptic to axonal terminals. Perhaps the strongest morphological support for a GABA presynaptic inhibitory pathway on primary afferent terminals comes from the present observations in the ventral horn motor nuclei, of GAD reaction product in small terminals which are presynaptic to larger terminals synapsing on motoneurons. The smaller GAD-containing terminals correspond to Conradi's ('69a) P-type (presynaptic) terminals in the cat spinal cord and the larger post-synaptic terminals correspond to Conradi's M-type (monosynaptic) terminals on motoneurons, which are of dorsal root origin (Conradi, '69b; McLaughlin, '72). It has been postulated h m observations of vesicle morphology in terminals of known synaptic action that synaptic terminals containing rounded vesicles are excitatory and terminals containing flattened vesicles are inhibitory (Uchizono, '65, '66, '67). This postulate, however, implies that vesicle shape may also reflect a difference in chemical constituents between excitatory and inhibitory terminals, namely in the synaptic transmitter and its synthesizing enzymes, but does not consider that some terminals may contain the same synaptic transmitter and synthesizing enzyme, and be either excitatory or inhibitory, such as GABA terminals in the spinal cord. The majority of GADcontaining and presumably GABA synaptic terminals in the spinal cord contain flattened or pleomorphic vesicles and make symmetrical synaptic contacts, in which both pre- and post-synaptic membranes are equally electron-dense (Ftype terminals). Some of these GADcontaining F knobs, which are presynaptic to dendrites and somata, are thought to hyperpolarize (inhibit) (Curtis et al., '68a,b; Curtis, '69). Other GAD-positive F knobs which are presynaptic to axonal terminals, are thought to depolarize (excite) (Eccles et al., '63; Schmidt, '64; Curtis et al., '71; Davidoff, '72;Davidoff et al., '73; Barker and Nicoll, '72, '73; Levy, '74). The re-
319
maining sample of observed GAD-containing terminals which are presynaptic to dendrites and presumed to have inhibitory functions, contain rounded vesicles and make asymmetrical contacts in which the postsynaptic membrane is more electrondense (R-type terminals). While it is possible that some of these discrepancies between vesicle shape and proposed synaptic action may be due to variations in the fixation procedure (Lund and Westrum, '66; Larramendi et al., '67; Bodian, '70) we assume that they are minimal on the basis of a more recent study (Valdivia, '71) indicating that when the osmolarity of the phosphate buffer is kept in the general range used in the present study, variations in vesicle morphology are minimal. In conclusion, our observations of GAD-containing terminals belonging either to R- or Ftypes which do not correlate necessarily with the proposed synaptic action, suggest that, at least for the spinal cord, vesicle shape and morphology of synaptic contacts may not indicate synaptic action and may not in fact be related to the type of synaptic transmitter or its synthesizing enzymes. ACKNOWLEDGMENTS
We would like to thank Dr. John G. Wood for his constant help and encouragement throughout this work and we also thank Dr. James E. Vaughn for helpful discussions and a critical review of the manuscript. The technical assistance of Sally Santos in the immunocytochemical procedures is also gratefully acknowledged. Supported by Grants NS01615 from the National Institute of Neurological Diseases and Stroke, National Institutes of Health, and MH-22458 from the National Institutes of Mental Health and USPHS Grant RR-05423 awarded to J. G. Wood and B. J. Mc Laughlin. LITERATURE CITED Albers, R. W., and R. 0. Brady 1969 The distribution of glutamic decarboxylase in the nervous system of the Rhesus monkey. J. Biol. Chem., 234: 926-928. Barber, R. P., B. J. McLaughlin, K. Saito and E. Roberts 1974 Light microscopic localization of glutamate decarboxylase in boutons of rat spinal cord before and after dorsal rhizotomy. Proc. SOC.for Neuroscience, 4th Annual Meeting, p. 127. Barker, J . L., and R. A. Nicoll 1972 Gammaaminobutyric acid: role in primary afferent depolarization. Science, 376: 1093-1045.
320
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
1973 The pharmacology and ionic dependency of amino acid responses in the frog spinal cord. J. Physiol., 228: 259-277. Bodian, D. 1970 An electron microscopic characterization of classes of synaptic vesicles by means of controlled aldehyde fixation. J. Cell Biol., 44: 115-124. Conradi, S. 1969a Ultrastructure and distribution of neuronal and glial elements on the motoneuron surface in the lumbosacral spinal cord of the adult cat. Acta Physiol. Scand., Suppl., 332: 548. 1969b Ultrastructure of dorsal root boutons o n lumbosacral motoneurons of the adult cat, as revealed by dorsal root section. Acta Physiol. Scand., Suppl., 332: 85115. Curtis, D. R. 1969 The pharmacology of spinal postsynaptic inhibition. In: Prog. i n Brain Res. Vol. 31,K. Akert and P. G. Waser, eds. Elsevier, Amsterdam, pp. 171-189. Curtis, D. R., A. W. Duggan, D. Felix and G. A. R. Johnston 1971 Bicuculline, a n antagonist of GABA and synaptic inhibition i n the spinal cord of the cat. Brain Res., 32;69-96. Curtis, D. R., L. Hosli and G. A. R. Johnston 1968b A pharmacological study of the depression of spinal neurons by glycine and related amino acids. Exp. Brain Res., 6: 1-18. Curtis, D. R., L. Hosli, G. A. R. Johnston and I. H. Johnston 1968a The hyperpolarization of spinal motoneurones by glycine and related amino acids. Exp. Brain Res., 5: 235-258. Davidoff, R. A. 1972 Gamma-aminobutyric acid antagonism and presynaptic inhibition i n the frog spinal cord. Science, 175: 331-333. Davidoff, R. A,, L. T. Graham, Jr., R. P. Shank, R. Werman and M. H. Aprison 1967 Changes i n amino acid concentrations associated with loss of spinal interneurons. J. Neurochem., 14: 1 0 2 5 1031. Eccles, J. C., R. Schmidt and W. D. Willis 1963 Pharmacological studies o n presynaptic inhibition. J. Physiol., 168: 50M30. Fahn, S . 1975 Regional distribution studies of GABA and other putative neurotransmitters and their enzymes. In: GABA i n Nervous System Function. E. Roberts, T. N. Chase and D. Tower, eds. Raven Press, New York, pp. 16S186. Graham, Jr., L. T., and M. H. Aprison 1969 Distribution of some enzymes associated with the metabolism of glutamate, aspartate, y-aminobutyrate and glutamine in cat spinal cord. J. Neurochem., 16: 559-566. Hartman, B. K. 1973 Immunofluorescence of dopamineg-hydroxylase. Application of improved methodology to the localization of the peripheral and central noradrenergic nervous system. J. Histochem. Cytochem., 2 1 : 312-332. Heimer, L., and P. D. Wall 1968 The dorsal root distribution to the substantia gelatinosa of the rat with a note on the distribution in the cat. Exper. Brain Res., 6: 89-99. Larramendi, L. M. H., L. Fickenscher and N. Lemkey-Johnston 1967 Synaptic vesicles of inhibitory and excitatory terminals in the cerebellum. Science, 1 5 6 : 967-969. Levy, R. A. 1974 GABA: a direct depolarizing action at the mammalian primary afferent terminal. Brain Res., 76: 155160.
-
Ljundahl, A., and T. Hokfelt 1973 Autoradiographic uptake patterns of [3H] GABA and [3H] glycine in central nervous tissues with special reference to the cat spinal cord. Brain Res., 62:
587-595. Lund, R. D., and L. E. Westrum 1966 Synaptic vesicle differences after primary formalin fixation. J. Physiol., 185: 7-9 P. McLaughlin, B. J. 1972 Dorsal root projections to the motor nuclei in the cat spinal cord. J. Comp. Neur., 144: 461474. McLaughlin, B. J., J. G. Wood, K. Saito, R. Barber, J. E. Vaughn, E. Roberts and J.-Y. Wu 1974 The fine structural localization of glutamate decarboxylase i n synaptic terminals of rodent cerebellum. Brain Res., 76: 377-391. Millonig, G. 1961 Advantage of a phosphate buffer for OsOl solutions i n fixation. J. Appl. Phys., 32: 1637 (abstract). Miyata, Y., and M. Otsuka 1972 Distribution of y-aminobutyric acid in cat spinal cord and the alteration produced by local ischaemia. J. Neurochem., 1 9 : 1833-1834. Okada, Y., C. Nitsch-Hassler, J. S. Kim, I. J. Bak and R. Hassler 1971 Role of y-aminobutyric acid (GABA) in the extrapyramidal motor system. I. Regional distribution of GABA in rabbit, rat, guinea pig and baboon CNS. Exp. Brain Res., 13: 514518. Otsuka, M., and S. Konishi 1975 GABA i n the spinal cord. In: GABA i n Nervous System Function. E. Roberts, T. N. Chase and D. Tower, eds. Raven Press, New York, pp. 197-202. Ralston, H. J., I11 1968 Dorsal root projections to dorsal horn neurons i n the cat spinal cord. J. Comp. Neur., 132: 303-330. Rethelyi, M., and J. Szentitgothai 1969 The large synaptic complexes of the substantia gelatinosa. Exp. Brain Res., 7: 258-274. Rexed, B. 1952 The cytoarchitectonic organization of the spinal cord i n the cat. J. Comp. Neur., 96: 415-496. 1954 A cytoarchitectonic atlas of the spinal cord in the cat. J. Comp. Neur., 100:
297480. Saito, K., R. Barber, J.-Y. Wu, T. Matsuda, E. Roberts and J. E. Vaughn 1974 Immunohistochemical localization of glutamic acid decarboxylase in rat cerebellum. Proc. Nat. Acad. Sci., (Washington), 71: 269-273. Schmidt, R. F. 1964 The pharmacology of presynaptic inhibition. In: Prog. in Brain Res. Vol. 12.J. C. Eccles and J. P. SchadC, eds. Elsevier, Amsterdam, pp. 119-134. Uchizono, K. 1965 Characteristics of excitatory and inhibitory synapses in the central nervous system of the cat. Nature, 207: 642643. 1966 Excitatory and inhibitory synapses in the cat spinal cord. Jap. J. Physiol., 16: 570-
575. 1967 Inhibitory synapses on the stretch receptor neurone of the crayfish. Nature, 214: 833-834. Valdivia, 0. 1971 Methods of fixation and the morphology of synaptic vesicles. J. Comp. Neur.,
142: 257-274. Venable, J. H., and R. Coggeshall 1965 A simplified lead citrate stain for use in electron microscopy. J. Cell Biol., 25: 407408.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD Vigh, B., and I. Vigh-Teichmann 1973 Cornparative ultrastructure of the cerebrospinal fluidcontacting neurons. Int. Rev. Cytol., 35: 189.251.
321
Wu, J.-Y., T. Matsuda and E. Roberts 1973 Purification and characterization of glutamate decarboxylase from mouse brain. J. Biol. Chem., 248: 30293034,
2
ABSTRACT The GABA synthesizing enzyme, glutamate decarboxylase (GAD), has been localized by light and electron microscopy in the rat lumbosacral spinal cord using a peroxidase-labeling antibody technique. The light microscopic localization shows heavy, punctate reaction product for GAD in the dorsal horn laminae 1-111. Moderately heavy reaction product is also seen in the deeper dorsal horn laminae IV-VI, the medial aspect of the intermediate gray (lamina VII) and the region around the central canal (lamina X). A moderately light concentration of GAD reaction product is observed in the ventral horn, and punctate deposits of reaction product also are seen on motoneuron cell bodies. The punctate distribution of reaction product for GAD in both ventral and dorsal horns, as visualized by light microscopy, corresponds to GADcontaining synaptic terminals seen by electron microscopy in comparable regions of the spinal gray. Many more GAD-positive terminals are observed in dorsal horn laminae 1-111 than in deeper laminae IV-VI. GAD-containing terminals in the dorsal horn are presynaptic to dendrites and cell bodies. GADcontaining terminals presynaptic to other axon terminals are observed also, and they are more numerous in laminae I1 and 111. In the ventral horn motor nuclei, GAD-psitive knobs are presynaptic to large and small dendrites and motoneuron cell bodies. In addition, small GADcontaining terminals also are presynaptic to larger axonal terminals which are in turn presynaptic to motoneuron somata. The observation of GAD-containing terminals presynaptic to dendrites and cell bodies in both dorsal and ventral horns is compatible with the evidence suggesting that GABA terminals may mediate postsynaptic inhibition of spinal interneurons and motoneurons. The additional finding of GAD-positive terminals presynaptic to other axonal terminals in the dorsal horn and motor nuclei is consistent with the growing evidence that GABA also may be the transmitter mediating presynaptic inhibition via axo-axonal synapses i n the spinal cord.
There is growing evidence that 7-aminobutyric acid (GABA) is a major neurotransmitter in the vertebrate central nervous system. A good deal of this evidence has evolved from biochemical, physiological and morphological studies on the cerebellum, which have received further support from the recent studies locating the GABA synthesizing enzyme, glutamate decarboxylase (GAD, E.C., 4. 1. 1. 15), by an immunoperoxidase method (Saito et al., '74; McLaughlin et al., '74). In those studies, GAD was localized in inhibitory and presumably GABA synaptic terminals in the cerebellar cortex and deep cerebellar nuclei. In the present study, we extend this J. COMP. NEUR.,164: 3 0 5 3 2 2
immunocytochemical technique to another area of the CNS, the spinal cord, where GABA is thought to mediate two types of spinal inhibition. Biochemical studies have shown that both GABA and its synthesizing enzyme, GAD, are present in the dorsal and ventral horns of the spinal gray, but are more concentrated in the superficial laminae of the dorsal horn (Albers and Brady, '59; Graham and Aprison, '69; Miyata and Otsuka, '72). Further studies have sug3 Please send reprint requests to: Dr. Eugene Roberts, Division of Neurosciences, City of Hope National Medical Center, 1500 E. Duarte Road, Duarte, California 91010.
305
306
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
gested that GABA is associated with interneurons located in the dorsal horn mediating presynaptic inhibition of primary afferent terminals (Eccles et al., '63; Schmidt, '64; Barker and Nicoll, '73; Miyata and Otsuka, '72; Otsuka and Konishi, '75). Other studies investigating the pharmacological action of GABA in the ventral horn have suggested that GABA is associated with interneurons mediating postsynaptic inhibition of motoneurons (Curtis et al., '68a; Curtis, '69). The present study examines the light and electron microscopic distribution of GAD-containing synaptic terminals in the rat lumbosacral spinal cord, with particular attention given to the localization of GADpositive terminals in dorsal horn laminae I-VI and in the motor nuclei of the ventral horn. MATERIALS AND METHODS
GAD was purified from the mouse brain synaptosomal fraction. Production of antibody was achieved by four bi-weekly injections of 150 fig of GAD into rabbit infkascapular muscle as described previously (for details, see Wu et al., '73; Saito et al., '74). Horseradish peroxidase (Type VI, Sigma), goat anti-rabbit IgG (Miles Laboratories), 3,3'-diaminobenzidine . 4 HCl (Sigma), acetone-dried mouse liver and mouse brain and spinal cord powders were all used in the immunocytochemical procedures. Goat anti-rabbit IgG was labeled with peroxidase as described previously (Saito et al., '74). Glutaraldehyde ( 8 % ) was purchased from Polysciences. All other chemicals were reagent grade.
Imm unochemis try procedure To remove the non-specific staining involved in the immunocytochemical procedures, 0.2 ml of anti-GAD serum or serum from non-immune rabbits was pretreated with a mixture of 75 mg liver and 25 mg spinal cord acetone powders. Peroxidaselabeled goat anti-rabbit IgG (0.2 ml containing 9 mg of protein) was treated with a mixture of 75 mg brain and 25 mg spinal cord powders. All steps were performed with a constant gentle agitation at room temperature. Tissue slices were incubated with 4-fold diluted rat serum for ten minutes prior to the experiment. The spinal cord slices were then incubated for 30 minutes with a 6-
fold diluted anti-GAD or control serum which contained goat IgG (1 mg/ml) and 0.1% Triton X-100 (Hartman, '73). The slices were washed with phosphate buffered saline (PBS) for three hours and then incubated for 30 minutes with peroxidaselabeled goat anti-rabbit IgG (1.2 mg/ml) containing 0.1% Triton X-100. After washing with PBS for another three hours, the slices were incubated with 3,3'-diaminobenzidine . 4 HCl (30 mu50 ml PBS) containing H202(0.006%) for ten minutes. The slices to be used for light microscopy were washed in PBS overnight at 4°C; the slices for electron microscopy were washed one hour in PBS prior to osmication.
Light microscopy Young adult Sprague-Dawley rats were anesthetized with 35% chloral hydrate (1 mVkg body weight) and fixed by vascular perfusion with 600 ml of 4% paraformaldehyde, in 0.12 M Millonig's ('61) buffer and 0.002% CaC12,pH 7.2-7.4 at 4°C. The animals were placed in plastic bags in the refrigerator for two hours, after which transverse segments of lumbosacral spinal cord were dissected out and placed in the above fixative for 48 hours at 4°C. The segments of spinal cord were then sliced transversely into 2 mm lengths and placed in separate, numbered containers in rostra1 to caudal order. Prior to cryotomy, the tissue was infiltrated with 30% sucrose for 24 hours. The tissue was then placed on a chuck, surrounded by 30% sucrose, and then quick-frozen in isopentane which was cooled by liquid nitrogen. The chuck and tissue were then transferred to the cryostat and allowed to equilibrate to - 20°C before cutting 40 f i thick transverse sections of the entire spinal cord face. The sections were picked up from the blade with a warm, moist slide and collected in 37°C PBS, and were washed overnight in a large volume of PBS at 4°C. The unmounted 40 p tissue sections were rinsed again in PBS and then processed for the immunocytochemical procedure. Following immunochemical treatment, the tissue was rinsed in distilled water, postfixed in 0.01% aqueous OSmium tetroxide for 30 seconds, rinsed in distilled water and mounted on a slide in glycerin. For more permanent storage, sections were dehydrated, cleared in xylene, and mounted in Permount.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
Electron microscopy Adult Sprague-Dawley rats were anesthetized with an intraperitoneal injection of 35% chloral hydrate (1 mYkg body weight) and then perfused through the heart with a fixative containing 4% paraformaldehyde and 0.1% glutaraldehyde in 0.12 M Millonig's ('61) phosphate buffer, pH 7.2, with 0.002% CaClz added. After the perfusion, the lumbosacral spinal cord was removed and stored overnight at 4 C in 4 % phosphate-buffered paraformaldehyde without glutaraldehyde. The spinal cord was sectioned transversely into 75 p slices with a Sorvall TC-2 tissue sectioner and collected in PBS in preparation for the immunocytochemical procedure described above. After processing the tissue for immunochemistry, the control and experimental transverse slices were hand cut into quarters so that each quarter contained a ventral or a dorsal horn. The quarter slices were then postfixed one hour in 2% osmium tetroxide in 0.12 M Millonig's buffer, pH 7.2, followed by rinses in 30% acetone and distilled water and block staining in 2% aqueous uranyl acetate for 20 minutes. The slices were dehydrated in graded alcohols, rinsed in propylene oxide and flat embedded in EponAraldite. The dorsal horn slices were embedded in either of two ways: (1) with the transverse slide face down in Beem capsule caps into which the cut apical end of the capsule is inserted; or (2) in silicone rubber molds with the medial sagittal edge of the dorsal horn forming the block face. The ventral horns were flat embedded only in the Beem capsules. Ultrathin sections of dorsal horns embedded in capsule caps were cut with a diamond knife in the transverse plane of the cord so that several laminae were in the same section; dorsal horns embedded in rubber molds were cut parasagittally so that all laminae were present in the same section; ventral horns were cut transversely so that the entire thin section contained only motor nuclei. Ultrathin sections were stained with lead citrate (Venable and Coggeshall, '65) and examined in either a Hitachi 11B or Philips EM 301 electron microscope. Cytoarchitectural lamination in the rat spinal cord was determined according to the procedure and criteria of Rexed ('52, '54) using paraffm embedded sections of O
30 7
the lumbosacral region stained with Luxol Fast Blue and cresyl violet or cresyl violet alone. In preparation for electron microscopy, photomicrographs of these stained sections were enlarged photographically and approximate measurements were obtained for each dorsal horn lamina, which were then used to identify the various laminae in thin sections. Due to the limited penetration of the reagents used to demonstrate reaction product for GAD in tissue slices embedded for electron microscopy, no attempts were made to quantitate the number of labeled terminals observed in each lamina. In addition, this penetration variability probably accounted for differences in the electron density of reaction product around synaptic vesicles and in terminals of any given field and, as a result, necessitated a qualitative rather than a quantitative analysis of the label. RESULTS
Lzg ht microscopy The dorsal gray matter of the rat spinal cord contains a higher concentration of GAD reaction product than the ventral spinal gray. Reaction product is heavily concentrated in the dorsal horn laminae 1-111 and two intense bands of GAD product are seen in this region. One dense band of punctate reaction product extends from the medial aspect of the dorsal horn to the most lateral aspect (fig. 1) and conforms to Rexeds lamina I11 medially and lamina I1 laterally. A thinner dense band follows the layer of marginal cells, conforming to lamina I of Rexed. The remaining dorsal horn (laminae IV-VI), the medial aspect of the intermediate gray (lamina VII), and the area around the central canal (lamina X) appear to have moderately heavy concentrations of punctate GAD-positive product (figs. 6, 7). The more lateral regions of the intermediate spinal gray (fig. 9) and the ventral horn motor nuclei (lamina IX) (fig. 8 ) exhibit a moderately light concentration of GAD-positive product. The remaining area of the ventral spinal gray has only light concentrations of GAD-positive product (fig. 10). Most of the neurons in the spinal cord have some punctate reaction product for GAD associated closely with their somata and large dendrites (fig. 9). The neuropil
308
McLAUGHLJN. BARBER, SAITO, ROBERTS, WU
throughout the spinal cord exhibits G A D positive product in different concentrations that vary roughly by region as illustrated in the diagram in figure 3. The region around the central canal exhibits a moderately heavy concentration of GAD reaction product. At higher magnifications, a few of the cells near the central canal appear to project an apical process into the central canal and resemble the cell bodies of spinal CSF-contacting neurons (Vigh and Vigh-Teichmann, '73). These cells have several GAD-positive deposits on their somata (fig. 7), which may correspond to GADcontaining synaptic terminals on the somata of these specialized neurons. In contrast, the motor nuclei are a moderately light concentration area, but the motoneurons themselves have punctate GAD deposits on their cell bodies (fig. 8). The punctate distribution of GAD product in both ventral and dorsal horns as visualized in the light microscope corresponds to the GAD-containing synaptic terminals seen with the electron microscope in comparable regions of the spinal gray (see Electron microscopy). No reaction product for GAD is seen in the control spinal cord slices treated with non-immune serum (fig. 2) and prepared for light or electron microscopy. Electron microscopy In the experimental tissue, electronopaque reaction product for peroxidaselabeled GAD is localized in synaptic terminals. The reaction product is usually aggregated around synaptic vesicle and mitochondrial membranes and along the presynaptic membrane specialization of each labeled terminal. Some of the GADpositive terminals in the rat spinal cord contain round synaptic vesicles (fig. 17) and make asymmetrical synaptic contacts (fig. 18), but the majority of GADpositive knobs contain pleomorphic or flattened vesicles and make symmetrical synaptic contacts (figs. 11, 12). Many more labeled synaptic terminals are seen in the superficial dorsal horn laminae 1-111 than in the deeper laminae IV-VZ, which corresponds with the light microscopic observations described above. Many of the labeled terminals in laminae 1-111 are presynaptic to dendrites but GADpositive terminals that are presynaptic to
cell bodies are also observed (figs. 11-13). In the transversely sectioned dorsal horns, GADpositive knobs which are presynaptic to other axonal terminals are observed occasionally in laminae I1 and 111. When the dorsal horns are sectioned along the sagittal plane, many more axo-axonal synaptic complexes are observed in these two laminae, in which GAD-containing terminals are presynaptic to other axonal terminals which are, in turn, presynaptic to other dendrites (figs. 14-16). The GAD-containing presynaptic terminals in these axoaxonal complexes always contain flattened or pleomorphic vesicles and appear to make symmetrical synaptic contacts with the postsynaptic terminals. The postsynaptic terminals of these complexes, on the other hand, usually contain rounded vesicles and make asymmetrical synaptic contacts with other profiles. GAD labeled synaptic terminals are not Fig. 1 Light photomicrograph of the dorsal horn from a rat lumbosacral spinal cord slice treated with immune serum. Arrows point to intense band of punctate GAD positive product in the substantia gelatinosa. Arrowheads point to GAD product in marginal layer. X 70. Fig. 2 Control dorsal horn from rat lumbosacral spinal cord treated with non-immune serum. Dashed line outlines the dorsal horn, X 70. Dense bodies are red blood cells. Fig. 3 The diagram shows different spinal cord regions numbered 4 through 10 which represent different concentrations of reaction product for GAD. Corresponding light photomicrographs from these regions are numbered accordingly. Long arrows represent GAD positive product in neuropil; Short arrows represent GAD positive product associated with neuronal somata. n., neuron; d., dendrite; cc, central canal. Figures 4-10 magnified X 1,000. Fig. 4 Neuropil with intense concentration of GAD product (arrows) &om lateral substantia gelatinosa. Fig. 5 Neuropil with intense concentration of GAD positive product (arrows) &om medial substantia gelatinosa. Fig. 6 Neuropil with moderately heavyconcentration of GAD positive product (arrows). Fig. 7 Region around central canal (cc). Punctate reaction product for GAD (arrows) is associated with CSF contacting neuron (n). Fig. 8 Motor neuron with punctate reaction product (arrows) studding the soma1 surface. Fig. 9 Neuron in region of moderately light concentration of reaction product. Both the soma (n) and the dendrite (d) of this neuron are contacted by punctate reaction product for GAD (arrows). Fig. 10 A n area from the spinal cord region where the concentration of GAD positive product (arrows) is the lowest.
IMMUNOCYTOCHEMICAL LOCALIZATION O F G A D I N SPINAL CORD
309
3 10
McLAUGHLIN, BARBER, SAITO, ROBERTS, W U
Fig. 11 Electron micrograph from the substantia gelatinosa (SG) (lamina 11) of spinal cord slice treated with immune serum. Numerous electron-opaque (GAD-positive synaptic terminals (arrows) are seen in this region which correspond to the heavy punctate reaction product seen in figures 4 and 5. Other adjacent terminals i n the field are not GADpositive. X
13,100.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD I N SPINAL CORD
311
Fig. 1 2 Higher magnification of GAD-positive terminals (arrows) in substantia gelatmosa (SG). X 21,800.
as numerous in the deeper dorsal horn laminae IV-VI. Both axosomatic and axodendritic GAD-positive presynaptic knobs are observed (fig. 19). Occasional G A D positive presynaptic axo-axonal contacts are observed, such as in figure 20, in which two GADcontaining terminals in lamina IV are presynaptic to a larger axonal terminal, which is not GADpositive and which is presynaptic to another axonal terminal and a dendrite.
In the ventral horn motor nuclei, GADcontaining synaptic terminals are less numerous than in the dorsal horn. Presynaptic GAD-positive terminals are observed on large and small dendritic profiles (fig. 21) as well as on motoneuron cell bodies. Some of the labeled knobs appear to be presynaptic to larger axonal terminals which synapse on motoneuronal somata (fig. 22). These terminals which are postsynaptic to the smaller GADpositive knobs
312
M c L A U G H L I N , BARBER, SAITO, ROBERTS, WU
Fig. 13 Numerous GADcontaining synaptic terminals (arrows) on neuronal cell bodies (N) and dendrites (D) in dorsal horn lamina 111. x 16,300.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
313
Fig. 14 Axo-axonal synapse in substantia gelatinosa (lamina 11) in which a GAD-positive terminal (T,) is presynaptic (small arrow) to another terminal (T2) which is not GAD-positive but which is itselfpresynaptic (large arrow) to a dendrite (D). x 52,400.
Fig. 15 Axo-axonal synapse in lamina 111 in which a GAD containing terminal (TI) is presynaptic (small arrow) to a n unlabeled terminal (T2) which is presynaptic (large arrows) to several dendrites (d). X 66,900.
314
McLAUGHLIN, BARBER, SAITO, ROBERTS, W U
Fig. 16 Axo-axond synapses located a t the interface between lamina I1 and 111 in which which two GADpositive terminals (TI) (T2) a r e presynaptic (small arrows) to a terminal (Tn) is not GAD-positive a n d is presynaptic (large arrows) to two dendrites (d). x 79,400.
Fig. 1 7 T w o GAD-positive terminals (tl) (h) i n lamina I1 which make symmetrical synaptic contacts with a dendrite (D). One GAD terminal (t,) contains predominantly rounded vesicles. Another rounded vesicle terminal (ta) makes a n asymmetrical contact with the same dendrite a n d is not GAD-positive. x 35,800. Fig. 18 GAD containing terminal t, in lamina I, which contains rounded synaptic vesicles a n d makes asymmetrical contacts (arrows) with a dendrite (d). An adjacent rounded vesicle terminal ( t 2 ) is not GAD-positive. x 43,700.
IMMUNOCYTOCHEMICAL LOCALIZATION O F GAD IN SPINAL CORD
315
Fig. 19 Fewer GAD-positive terminals (arrows) are seen in dorsal horn lamina IV than i n lamina-111. One terminal i n the field is presynaptic to a neuron soma (N) and another is presynaptic to a dendrite (D). A large terminal (t) i n the vicinity is not GAD-positive. X 12,500.
containing flattened vesicles, are characteristically large in size, contain rounded vesicles and make multiple synaptic contacts with the motoneuron, in which numerous dense bodies are associated with the subsynaptic membrane (fig. 22). Such terminals are thought to be of dorsal root origin (Conradi, '69b; McLaughlin, '72). DISCUSSION
Although GAD and GABA levels are comparatively lower in the spinal cord than
in other regions of the central nervous system (Albers and Brady, '59; Okada et al., '71; Fahn, '75) the relatively high concentration of GABA and its synthesizing enzyme in the dorsal horn, nevertheless, has led to a consideration of this amino acid as a possible neurotransmitter in the spinal cord. Pharmacological studies (Curtis et al., '68a,b; Curtis, '69) have demonstrated that iontophoretically applied GABA has a hyperpolarizing effect on both interneurons and motoneurons in the dorsal and
3 16
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
Fig. 20 Axo-axonal synapses i n lamina IV i n which two GAD-positive terminals ( t l ) (f) appear to be presynaptic (small arrows) to a large terminal (t3) which is not GAD-positive and which is uresvnautic (large arrows) to another axonal terminal (t4) and dendrite (d).
ventral spinal gray and suggest that GABA may function as a postsynaptic inhibitory transmitter in these regions via axodendritic and axosomatic synapses. Similar studies (Eccles et al., '63; Schmidt, '64; Barker and Nicoll, '72, '73; Davidoff, '72; Levy, '74) have demonstrated that GABA has a depolarizing effect on dorsal root fibers and suggest that GABA may act also to mediate presynaptic inhibition of primary afferent terminals via axo-axonal synapses, in addition to mediating postsynaptic inhibition on spinal interneurons and motoneurons. The present immunocytochemical localization of GAD in the rat lumbosacral spinal cord reveals a high concentration of synaptic terminals containing GAD in the outermost dorsal horn laminae 1-111 and fewer GADcontaining terminals in the deeper laminae IV-VI. This finding correlates well with the biochemical evidence that both GAD and GABA are more concentrated in the dorsolateral region of the dorsal horn (Albers and Brady, '59; Miyata and Otsuka, '72; Otsuka and Ko-
nishi, '75), as well as with recent autoradiographic work (Ljungdahl and Hokfelt, '73) showing intense uptake of [3H]-GABA by boutons in laminae 1-111. The majority of these GADpositive terminals in the dorsal horn laminae are presynaptic to dendrites and neuronal cell bodies and may belong to neurons mediating postsynaptic inhibition in the dorsal horn (Curtis et al., '68b). It has been suggested that the GABA interneurons mediating postsynaptic inhibition in the spinal cord are located in the dorsal regions of the cord (Curtis et al., '71), but other workers (Davidoff et al., '67) have suggested that GABA is not concentrated in spinal interneurons and instead may be the inhibitory transmitter of supraspinal structures which terminate in the spinal gray. While an investigation into the origin of GAD-containing or GABA terminals in the spinal cord is of utmost importance, the immunocytochemical method used in this study is not sufficiently sensitive to detect GADcontaining cell bodies and axonal pathways, which would allow such GABA neurons to be properly identi-
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
317
Fig. 21 Large dendrite (D) in the ventral horn motor nucleus surrounded by synaptic terminals, two of which are GADpositive (arrows). x 12,500.
fied. Future studies combining surgical ablation of certain spinal cord pathways with the GAD immunocytochemical technique, however, may help to determine whether they are indeed of supraspinal or interneuronal origin. For example, spinal cord transections could be done to determine if the GAD-containing terminals (below the transected cord) are from neurons in supraspinal regions. If there is a considerable decrease in the number of observed GAD-positive terminals below the transected cord, then it could be concluded that many of the GAD-containing and presumably GABA terminals are of supraspinal origin. Similar experiments, such as combined dorsal root (Barber et al., '74) and cord transections above and below the spinal cord level to be examined,
could be done to determine if the GADpositive terminals belong primarily to spinal interneurons. The drop in glycine concentration and not GABA, following hypoxic destruction of ventral horn spinal interneurons has led to the suggestion that glycine is released by interneurons located primarily in the ventral horn and is the major transmitter mediating segmental spinal inhibition (Davidoff et al., '67). Further studies using presumed antagonists to glycine and GABA (Curtis et al., '71) suggest that prolonged inhibition of motoneurons appears to be mediated by GABA inhibitory terminals on motoneuronal dendrites or on excitatory terminals which are presynaptic to motoneurons, whereas glycine inhibitory terminals are located predominantly on
3 18
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
Fig. 22 A small GAD-containing terminal (t,) presynaptic (small arrow) to a larger terminal (tz) which is not GAD-positive and which is presynaptic (large arrow) to a motoneuron (MN). X 55,500.
motoneuron cell bodies. The immunocytochemical localization of GAD-containing terminals on dendrites in the motor nuclei and on axonal terminals presynaptic to motoneurons is consistent with this latter proposal, while the relatively large number of GAD-containing terminals observed synapsing with motoneuron somata suggests that GABA neurons, as well as glycine neurons may play an important role in the inhibitory events taking place on motoneuron somata. The immunocytochemical localization of GAD in terminals which are presynaptic to other axonal terminals in both the dorsal horn laminae and ventral horn motor nuclei is consistent with the growing evidence that GABA is the transmitter mediating presynaptic inhibition of primary afferent terminals (Eccles et al., '63; Schmidt, '64; Curtis et al., '71; Davidoff, '72; Davidoff et al., '73; Barker and Nicoll, '72, '73; Miyata and Otsuka, '72; Levy, '74; Otsuka and Konishi, '75). The largest number of these GAD-containing presynaptic components of axo-axonal synapses are observed in dorsal horn laminae I1 and 111. This observation is of interest with regard to the studies of Miyata and Otsuka ('72) and Otsuka and Konishi ('75), in which GABA levels in the dorsolateral region of the dorsal horn are markedly de-
creased along with the size of the dorsal root potentials after hypoxic destruction of spinal interneurons in that area and which suggest that the GABA interneurons acting in presynaptic inhibition are located in the superficial dorsal horn laminae. Rethelyi and Szentagothai ('69) have suggested that other neurons located in deeper dorsal horn laminae may also be presynaptic to primary afferent terminals and function in presynaptic inhibitory pathways. In order to satisfy this structural prerequisite for presynaptic inhibition of an axonal terminal synapsing on a dorsal root terminal, as well as to strengthen the mounting evidence that GABA is the presynaptic transmitter, we currently are studying the immunocytochemical distribution of GAD in the spinal cord after dorsal rhizotomies (c.f. Barber et al., '74), to determine whether some of the GADcontaining terminals are in fact presynaptic to primary afferent terminals. Heimer and Wall ('68) have shown heavy dorsal root degeneration in the same dorsal horn laminae 11 and I11 of the rat spinal cord where GAD-containing terminals are the most numerous. Future studies combining dorsal rhizotomies and GADimmunocytochemistry (Barber et al., '74) will determine if GADpositive terminals are indeed presynaptic to some of these
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD
degenerating dorsal root terminals and whether or not there is a regional difference in the axo-axonal synaptic arrangements as Ralston ('68) has shown in the cat dorsal horn, in which degenerating dorsal root terminals in laminae I11 are presynaptic to other axonal terminals but in deeper laminae IV-VI are postsynaptic to axonal terminals. Perhaps the strongest morphological support for a GABA presynaptic inhibitory pathway on primary afferent terminals comes from the present observations in the ventral horn motor nuclei, of GAD reaction product in small terminals which are presynaptic to larger terminals synapsing on motoneurons. The smaller GAD-containing terminals correspond to Conradi's ('69a) P-type (presynaptic) terminals in the cat spinal cord and the larger post-synaptic terminals correspond to Conradi's M-type (monosynaptic) terminals on motoneurons, which are of dorsal root origin (Conradi, '69b; McLaughlin, '72). It has been postulated h m observations of vesicle morphology in terminals of known synaptic action that synaptic terminals containing rounded vesicles are excitatory and terminals containing flattened vesicles are inhibitory (Uchizono, '65, '66, '67). This postulate, however, implies that vesicle shape may also reflect a difference in chemical constituents between excitatory and inhibitory terminals, namely in the synaptic transmitter and its synthesizing enzymes, but does not consider that some terminals may contain the same synaptic transmitter and synthesizing enzyme, and be either excitatory or inhibitory, such as GABA terminals in the spinal cord. The majority of GADcontaining and presumably GABA synaptic terminals in the spinal cord contain flattened or pleomorphic vesicles and make symmetrical synaptic contacts, in which both pre- and post-synaptic membranes are equally electron-dense (Ftype terminals). Some of these GADcontaining F knobs, which are presynaptic to dendrites and somata, are thought to hyperpolarize (inhibit) (Curtis et al., '68a,b; Curtis, '69). Other GAD-positive F knobs which are presynaptic to axonal terminals, are thought to depolarize (excite) (Eccles et al., '63; Schmidt, '64; Curtis et al., '71; Davidoff, '72;Davidoff et al., '73; Barker and Nicoll, '72, '73; Levy, '74). The re-
319
maining sample of observed GAD-containing terminals which are presynaptic to dendrites and presumed to have inhibitory functions, contain rounded vesicles and make asymmetrical contacts in which the postsynaptic membrane is more electrondense (R-type terminals). While it is possible that some of these discrepancies between vesicle shape and proposed synaptic action may be due to variations in the fixation procedure (Lund and Westrum, '66; Larramendi et al., '67; Bodian, '70) we assume that they are minimal on the basis of a more recent study (Valdivia, '71) indicating that when the osmolarity of the phosphate buffer is kept in the general range used in the present study, variations in vesicle morphology are minimal. In conclusion, our observations of GAD-containing terminals belonging either to R- or Ftypes which do not correlate necessarily with the proposed synaptic action, suggest that, at least for the spinal cord, vesicle shape and morphology of synaptic contacts may not indicate synaptic action and may not in fact be related to the type of synaptic transmitter or its synthesizing enzymes. ACKNOWLEDGMENTS
We would like to thank Dr. John G. Wood for his constant help and encouragement throughout this work and we also thank Dr. James E. Vaughn for helpful discussions and a critical review of the manuscript. The technical assistance of Sally Santos in the immunocytochemical procedures is also gratefully acknowledged. Supported by Grants NS01615 from the National Institute of Neurological Diseases and Stroke, National Institutes of Health, and MH-22458 from the National Institutes of Mental Health and USPHS Grant RR-05423 awarded to J. G. Wood and B. J. Mc Laughlin. LITERATURE CITED Albers, R. W., and R. 0. Brady 1969 The distribution of glutamic decarboxylase in the nervous system of the Rhesus monkey. J. Biol. Chem., 234: 926-928. Barber, R. P., B. J. McLaughlin, K. Saito and E. Roberts 1974 Light microscopic localization of glutamate decarboxylase in boutons of rat spinal cord before and after dorsal rhizotomy. Proc. SOC.for Neuroscience, 4th Annual Meeting, p. 127. Barker, J . L., and R. A. Nicoll 1972 Gammaaminobutyric acid: role in primary afferent depolarization. Science, 376: 1093-1045.
320
McLAUGHLIN, BARBER, SAITO, ROBERTS, WU
1973 The pharmacology and ionic dependency of amino acid responses in the frog spinal cord. J. Physiol., 228: 259-277. Bodian, D. 1970 An electron microscopic characterization of classes of synaptic vesicles by means of controlled aldehyde fixation. J. Cell Biol., 44: 115-124. Conradi, S. 1969a Ultrastructure and distribution of neuronal and glial elements on the motoneuron surface in the lumbosacral spinal cord of the adult cat. Acta Physiol. Scand., Suppl., 332: 548. 1969b Ultrastructure of dorsal root boutons o n lumbosacral motoneurons of the adult cat, as revealed by dorsal root section. Acta Physiol. Scand., Suppl., 332: 85115. Curtis, D. R. 1969 The pharmacology of spinal postsynaptic inhibition. In: Prog. i n Brain Res. Vol. 31,K. Akert and P. G. Waser, eds. Elsevier, Amsterdam, pp. 171-189. Curtis, D. R., A. W. Duggan, D. Felix and G. A. R. Johnston 1971 Bicuculline, a n antagonist of GABA and synaptic inhibition i n the spinal cord of the cat. Brain Res., 32;69-96. Curtis, D. R., L. Hosli and G. A. R. Johnston 1968b A pharmacological study of the depression of spinal neurons by glycine and related amino acids. Exp. Brain Res., 6: 1-18. Curtis, D. R., L. Hosli, G. A. R. Johnston and I. H. Johnston 1968a The hyperpolarization of spinal motoneurones by glycine and related amino acids. Exp. Brain Res., 5: 235-258. Davidoff, R. A. 1972 Gamma-aminobutyric acid antagonism and presynaptic inhibition i n the frog spinal cord. Science, 175: 331-333. Davidoff, R. A,, L. T. Graham, Jr., R. P. Shank, R. Werman and M. H. Aprison 1967 Changes i n amino acid concentrations associated with loss of spinal interneurons. J. Neurochem., 14: 1 0 2 5 1031. Eccles, J. C., R. Schmidt and W. D. Willis 1963 Pharmacological studies o n presynaptic inhibition. J. Physiol., 168: 50M30. Fahn, S . 1975 Regional distribution studies of GABA and other putative neurotransmitters and their enzymes. In: GABA i n Nervous System Function. E. Roberts, T. N. Chase and D. Tower, eds. Raven Press, New York, pp. 16S186. Graham, Jr., L. T., and M. H. Aprison 1969 Distribution of some enzymes associated with the metabolism of glutamate, aspartate, y-aminobutyrate and glutamine in cat spinal cord. J. Neurochem., 16: 559-566. Hartman, B. K. 1973 Immunofluorescence of dopamineg-hydroxylase. Application of improved methodology to the localization of the peripheral and central noradrenergic nervous system. J. Histochem. Cytochem., 2 1 : 312-332. Heimer, L., and P. D. Wall 1968 The dorsal root distribution to the substantia gelatinosa of the rat with a note on the distribution in the cat. Exper. Brain Res., 6: 89-99. Larramendi, L. M. H., L. Fickenscher and N. Lemkey-Johnston 1967 Synaptic vesicles of inhibitory and excitatory terminals in the cerebellum. Science, 1 5 6 : 967-969. Levy, R. A. 1974 GABA: a direct depolarizing action at the mammalian primary afferent terminal. Brain Res., 76: 155160.
-
Ljundahl, A., and T. Hokfelt 1973 Autoradiographic uptake patterns of [3H] GABA and [3H] glycine in central nervous tissues with special reference to the cat spinal cord. Brain Res., 62:
587-595. Lund, R. D., and L. E. Westrum 1966 Synaptic vesicle differences after primary formalin fixation. J. Physiol., 185: 7-9 P. McLaughlin, B. J. 1972 Dorsal root projections to the motor nuclei in the cat spinal cord. J. Comp. Neur., 144: 461474. McLaughlin, B. J., J. G. Wood, K. Saito, R. Barber, J. E. Vaughn, E. Roberts and J.-Y. Wu 1974 The fine structural localization of glutamate decarboxylase i n synaptic terminals of rodent cerebellum. Brain Res., 76: 377-391. Millonig, G. 1961 Advantage of a phosphate buffer for OsOl solutions i n fixation. J. Appl. Phys., 32: 1637 (abstract). Miyata, Y., and M. Otsuka 1972 Distribution of y-aminobutyric acid in cat spinal cord and the alteration produced by local ischaemia. J. Neurochem., 1 9 : 1833-1834. Okada, Y., C. Nitsch-Hassler, J. S. Kim, I. J. Bak and R. Hassler 1971 Role of y-aminobutyric acid (GABA) in the extrapyramidal motor system. I. Regional distribution of GABA in rabbit, rat, guinea pig and baboon CNS. Exp. Brain Res., 13: 514518. Otsuka, M., and S. Konishi 1975 GABA i n the spinal cord. In: GABA i n Nervous System Function. E. Roberts, T. N. Chase and D. Tower, eds. Raven Press, New York, pp. 197-202. Ralston, H. J., I11 1968 Dorsal root projections to dorsal horn neurons i n the cat spinal cord. J. Comp. Neur., 132: 303-330. Rethelyi, M., and J. Szentitgothai 1969 The large synaptic complexes of the substantia gelatinosa. Exp. Brain Res., 7: 258-274. Rexed, B. 1952 The cytoarchitectonic organization of the spinal cord i n the cat. J. Comp. Neur., 96: 415-496. 1954 A cytoarchitectonic atlas of the spinal cord in the cat. J. Comp. Neur., 100:
297480. Saito, K., R. Barber, J.-Y. Wu, T. Matsuda, E. Roberts and J. E. Vaughn 1974 Immunohistochemical localization of glutamic acid decarboxylase in rat cerebellum. Proc. Nat. Acad. Sci., (Washington), 71: 269-273. Schmidt, R. F. 1964 The pharmacology of presynaptic inhibition. In: Prog. in Brain Res. Vol. 12.J. C. Eccles and J. P. SchadC, eds. Elsevier, Amsterdam, pp. 119-134. Uchizono, K. 1965 Characteristics of excitatory and inhibitory synapses in the central nervous system of the cat. Nature, 207: 642643. 1966 Excitatory and inhibitory synapses in the cat spinal cord. Jap. J. Physiol., 16: 570-
575. 1967 Inhibitory synapses on the stretch receptor neurone of the crayfish. Nature, 214: 833-834. Valdivia, 0. 1971 Methods of fixation and the morphology of synaptic vesicles. J. Comp. Neur.,
142: 257-274. Venable, J. H., and R. Coggeshall 1965 A simplified lead citrate stain for use in electron microscopy. J. Cell Biol., 25: 407408.
IMMUNOCYTOCHEMICAL LOCALIZATION OF GAD IN SPINAL CORD Vigh, B., and I. Vigh-Teichmann 1973 Cornparative ultrastructure of the cerebrospinal fluidcontacting neurons. Int. Rev. Cytol., 35: 189.251.
321
Wu, J.-Y., T. Matsuda and E. Roberts 1973 Purification and characterization of glutamate decarboxylase from mouse brain. J. Biol. Chem., 248: 30293034,
