Colocalization of glycine and GABA in synapses on spinomedullary neurons

BRAIN RESEARCH Brain Research 690 (1995) 127-132

ELSEVIER

Short communication

Colocalization of glycine and GABA in synapses on spinomeduUary neurons D.J. M a x w e l l

*, A.J. T o d d , R. K e r r

Laboratory of Human Anatomy, Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK Accepted 16 May 1995

Abstract

Spinomedullary neurons of the postsynaptic dorsal column pathway in adult cats were retrogradely labelled with horseradish peroxidase: Postembedding immunogold reactions were performed with antisera which recognise GABA or glycine to determine if synaptic boutons in contact with these neurons contain both transmitters. Analysis of series of ultrathin sections revealed that synaptic profiles with strong immunogold reactions for GABA usually also displayed strong immunogold reactions for glycine. Pre-embedding immunocytochemistry was performed on sections containing labelled cells with a monoclonal antibody which recognises the glycine receptor-associated protein, gephyrin. Many synapses onto postsynaptic dorsal column neurons were associated with gephyrin-like immunoreactivity and these typically contained irregularly shaped vesicles. Immunogold reactions showed that synaptic profiles apposed to gephyrin-immunoreactive junctions contained GABA and glycine. The evidence suggests that glycine is a neurotransmitter at synapses on spinomedullary neurons and that it is colocalized with GABA. Keywords: Gephyrin; Spinal cord; Retrograde tracing; Immunogold reaction; Postsynaptic dorsal column pathway

Spinomedullary neurons of the postsynaptic dorsal column (PSDC) pathway are located mainly in laminae I I I - V of the dorsal horn of the spinal cord and have axons which project through the dorsal columns and terminate in the dorsal column nuclei [14-16]. PSDC neurons are excited by mono- and di-synaptic inputs from cutaneous and muscle receptors and are inhibited through di- and poly-synaptic pathways from low and high threshold cutaneous afferents [5]. The organization of receptive fields is complex and several types of inhibitory process appear to modify the sensitivity of excitatory fields [8]. A previous study showed that many of the profiles which were presynaptic to PSDC neurons display immunoreactivity for GABA, while almost all of the profiles which were not immunoreactive for GABA were immunoreactive for glutamate [6] and it was therefore suggested that most of the inhibitory boutons associated with these neurons contain GABA. However it is probable that GABA is not the only neurotransmitter involved in these inhibitory processes. It is now well established that GABA is colocalized with glycine in neurons in various regions of the central ner-

* Corresponding author. Fax: (44) (141) 330-4299. 0006-8993/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0 0 0 6 - 8 9 9 3 ( 9 5 ) 0 0 6 1 3 - 3

vous system (e.g. see [10]) including the cell bodies of many dorsal horn neurons [18] and a recent study has shown that these transmitters are also colocalized in synaptic boutons of the dorsal horn [17]. Game and Lodge [4] have shown that two phases of inhibition occur in cells of laminae IV and V which are evoked by electrical stimulation of cutaneous nerves. The early phase of inhibition is most sensitive to strychnine whereas the later phase is more sensitive to bicuculline. This evidence implies that glycine and GABA are both involved in the inhibition of dorsal horn neurons. The purpose of the present study was to investigate the possibility that some boutons which make synaptic connections with PSDC neurons use both GABA and glycine as inhibitory transmitters. Two series of experiments were carried out. Initially post-embedding immunogold reactions were performed on ultrathin sections containing retrogradely labelled PSDC neurons and boutons were examined through series of sections to establish if they displayed immunoreactivity for GABA and glycine. In a second series of experiments pre-embedding immunocytochemistry was performed with a monoclonal antiserum which recognises the glycine receptor associated protein, gephyrin [11,19] in order to determine if gephyrin is present at synapses on labelled PSDC neurons and if there

128

D.J. Maxwell et al. / B r a i n Research 690 (1995) 127-132

~: P S D C

PSDC

B

~ ' ~~ : .

~=

""

..

Fig. 1. A and B are consecutive sections through part of a PSDC cell (the area within the box in C) which were reacted by using the postembedding immunogold technique to reveal the presence of GABA (A) or glycine (B). A profile (*), which contains irregularly shaped vesicles and makes a symmetrical synapse, shows strong immunogold reactions for both amino acids. A second bouton (~r), is negative for both amino acids. The arrows in C indicate granules of reaction product within the PSDC neuron. Scale bars: A and B = 0.5 /xm; C = 2 /xm.

D.J. Maxwell et aL / Brain Research 690 (1995) 127-132

was any relationship between these synapses and GABAand glycine-immunoreactive profiles. Four adult cats (2.5 kg) were used in the study. They were deeply anaesthetized with sodium pentobarbitone (40 mg/kg; i.p.) and surgery was performed under sterile conditions. Postsynaptic dorsal column neurons were retrogradely labelled by implanting agar-horseradish peroxidase (HRP) pellets in small, unilateral lesions of the left dorsal columns at C 3. An additional lesion was made in the left dorsolateral funiculus at C 4 to ensure that the sample of PSDC neurons was not contaminated with neurons which project through this region of the white matter. Details of the surgery and preparation of pellets have been published previously [2,3,6]. Animals were allowed to recover from the anaesthetic and, following a 72-h survival period, were re-anaesthetized (sodium pentobarbitone; 40 mg/kg, i.p.) and transcardially perfused for 15 s with warm saline (37°C) containing sodium nitrite (0.1%) and heparin (100 /x/ml). This was followed by 1 1 of warm (37°C) fixative which contained 1% formaldehyde and 1% (2 animals) or 2.5% glutaraldehyde (2 animals) in 0.1 M phosphate buffer (pH 7.6) and two litres of cold (4°C) fixative. When perfusion was completed, spinal segments L 6 - S 1 and C 3C 5 were removed and placed in the same fixative for 8 h at 4°C. Transverse sections (50 /xm thick) were cut through the lumbosacral cord with a Vibratome and serial transverse sections (90/xm thick) were cut through the implant site with a freezing microtome. Both groups of sections were reacted with hydrogen peroxide in the presence of 3',3-diaminobenzidine (DAB) to visualize the presence of HRP, however the reaction product in the Vibratome sections to be used for gephyrin immunoreactions (from animals perfused with 1% glutaraldehyde in the fixative) was intensified with cobalt chloride which produces a black/dark blue reaction product [1]. Implant sites at C3 were examined; lesions were confined to the left dorsal column and there was minimum spread of reaction product into the grey matter of the cord. Vibratome sections were wet mounted and those containing labelled neurons were selected. Sections from animals perfused with 2.5% glutaraldehyde fixative were osmicated, dehydrated and flat embedded for combined light and electron microscopy (see [6]), while sections from animals fixed with 1% glutaraldehyde were reacted for gephyrin immunoreactivity. The procedure was identical to that described previously [7,17]. Briefly, sections were treated for 30 min in 1% sodium borohydride, rinsed several times in buffer, and incubated overnight with a monoclonal antibody directed against gephyrin (mAb 7a; Boehringer Mannheim) at a dilution of 1:4000. The properties of this antibody have been described in detail elsewhere (for a full discussion see [7]). The sections were then processed by using a Vector ABC Elite kit and peroxidase was visualized with DAB. On this occasion the reaction product was not intensified in order to produce a brown colour which would contrast with the black reaction

129

product associated with retrogradely-transported HRP in labelled PSDC cells. These sections were osmicated and prepared for combined light and electron microscopy. Labelled cells were examined with a light microscope and sections containing selected neurons were then attached to blocks of cured resin. Series of ultrathin sections were cut and collected on single slot nickel grids coated with Formvar. Usually two or three sections were collected on each grid. Alternate grids from these series were processed to reveal either GABA or glycine immunoreactivity by using the postembedding immunogold method [12]. The procedure was similar to that used by Todd et al. [17], except that phosphate-buffered saline and Tris-buffered saline solutions contained 0.3 M sodium chloride and the primary antisera to glycine and GABA were diluted to 1:1000-1:4000 in phosphate-buffered saline containing 1% bovine serum albumen. Triton X-100 (0.1%) was included in all solutions. Anti-rabbit IgG adsorbed to 10 nm gold spheres (British Biocell International; diluted 1:20) was used to detect the presence of primary antisera. The two antisera raised against glutaraldehyde-conjugated GABA and glycine were donated by Dr. D.V. Pow. The production and properties of these antisera have been described in detail elsewhere [13]. They do not cross-react with inappropriate amino acid conjugates in dot-blots and preincubation of primary antisera with appropriate amino acids conjugated to bovine serum albumen greatly reduces immunogold reactions in spinal tissue (see Fig. 5 of [17]). Profiles were examined through series of sections to assess the quality of immunogold reactions. The same profile was always viewed in at least two serial sections on the same grid (i.e. reacted either with anti-GABA or anti-glycine antiserum) to determine if the pattern of immunoreactivity was consistent for each of the two antisera. Cell bodies of PSDC neurons labelled with granules of reaction product could be visualized in laminae III-V of the lumbosacral dorsal horn in sections when viewed with a light microscope. They had a similar appearance and location to cells described in previous studies of PSDC neurons [2]. The cells could easily be identified also when viewed with the electron microscope as a consequence of the presence of numerous electron-dense packages of reaction product within them (Fig. 1C, 2A). Six PSDC cells (three from each of the animals perfused with the higher concentration of glutaraldehyde) were examined through series of 24-36 ultrathin sections and boutons which were presynaptic to the cells were examined on sections reacted with GABA and glycine antisera. Strong immunogold reactions for these amino acids were present over synaptic vesicles and mitochondria of some profiles and the density of labelling associated with them was at least four times greater than the density associated with PSDC neurons. Most of the profiles with strong immunogold reactions for GABA also displayed strong reactions for glycine and vice versa (Fig. 1), however some profiles were observed which possessed only

130

D.J. Maxwell et al. / B r a i n Research 690 (1995) 127-132

i

D.J. Maxwell et al. / Brain Research 690 (19951 127-132

G A B A - o r glycine-immunoreactivity. Twenty profiles which showed a strong immunoreaction for GABA were selected and followed through the series; all of these profiles consistently displayed a strong reaction for glycine also. Profiles which were both GABA- and glycine-immunoreactive were typically about 0.75 /~m in diameter, contained irregularly shaped vesicles and formed symmetrical synaptic junctions with PSDC neurons (Fig. 1) and therefore they are similar to the population of GABA-immunoreactive boutons described by Maxwell et al. [6]. In sections processed with antibody to gephyrin, light microscopy revealed that the cell bodies of PSDC neurons were associated with numerous immunopositive puncta which were located on the membranes of somata and proximal dendrites. This was confirmed by ultrastructural analysis (Fig. 2). Reaction product was associated with postsynaptic membranes of some of the synapses on PSDC neurons (Fig. 2A and B). Boutons associated with gephyrin-immunoreactive synapses had a similar morphological appearance to the population of profiles which contain GABA [6] (Fig. 2B). Immunogold reactions were performed on 3 PSDC neurons (one from one of the animals and two from the other) and profiles were examined through series of 24-45 ultrathin sections. Most of the boutons associated with gephyrin-immunoreactive synapses displayed strong immunogold reactions for both GABA and glycine (Fig. 2C and D) but some boutons were weakly labelled for glycine or GABA or, very occasionally, for both amino acids. It is probable that these latter observations represent 'false-negatives' as immunogold reactions on sections processed for gephyrin were generally weaker than those in tissue specifically prepared for immunogold reactions (see Figs. 1 and 2). Although no attempt was made to quantify numbers of immunoreactive profiles, it is clear that many GABA-immunoreactive boutons on PSDC neurons are also immunoreactive for glycine. Glycine is known to have metabolic functions, but the presence of gephyrin-like immunoreactivity at synapses where the presynaptic bouton showed both GABA- and glycine-immunoreactivity strongly suggests that glycine functions as a transmitter at these synapses. It is likely that most of the GABA/glycine-containing boutons arise from spinal interneurons since these two amino acids are known to co-exist in many dorsal horn neurons [18]. Inhibition of PSDC neurons evoked by light tactile or noxious stimulation of the skin remains effective

131

following a cold block of the thoracic cord at T13 [9] and electrical stimulation of low-threshold axons in cutaneous nerves elicits inhibitory postsynaptic potentials in PSDC neurons which have di-synaptic latencies [5]. Furthermore Game and Lodge [4] have shown that electrical stimulation of low-threshold cutaneous axons produces inhibition of dorsal horn cells which is biphasic and involves GABA and glycine. It is likely therefore that many of the interneurons which contain colocalized GABA and glycine are monosynaptically excited by low-threshold cutaneous afferents and that these in turn modulate transmission through the PSDC pathway by inhibiting the activity of PSDC neurons.

Acknowledgements Financial support from the Wellcome Trust and the MRC is gratefully acknowledged. Some preliminary parts of this study were performed at the Department of Preclinical Veterinary Sciences, University of Edinburgh, U.K. We thank Dr David Pow for the generous gift of antisera and H. Anderson and M. Hughes for excellent technical assistance.

References [1] Adams, J.C., Technical consideration on the use of horseradish peroxidase as a neuronal marker, Neuroscience, 2 (19771 141-145. [2] Enevoldson, T.P. and Gordon, G., Postsynaptic dorsal column neurons in the cat: a study with retrograde transport of horseradish peroxidase, Exp. Brain Res., 75 (1989) 611-620. [3] Enevoldson, T.P., Gordon, G. and Sanders, D.J., The use of retrograde transport of horseradish peroxidase for studying the dendritic trees and axonal courses of particular groups of tract cells in the spinal cord, Exp. Brain Res., 54 (1984) 529-537. [4] Game, C.J.A. and Lodge, D., The pharmacology of the inhibition of dorsal horn neurones by impulses in myelinated cutaneous afferents in the cat, Exp. Brain Res., 23 (1975) 75-84. [5] Jankowska, E., Rastad, J. and Zarzecki, P., Segmental and supraspinal input to cells of origin of non-primary fibres in the feline dorsal columns, J. Physiol., 290 (1979) 185-2(10. [6] Maxwell, D.J., Ottersen, O.P. and Storm-Mathisen, J., Synaptic organization of excitatory and inhibitory boutons associated with spinal neurons which project through the dorsal columns of the cat, Brain Res., 676 (1995) 103-112. [7] Mitchell, K., Spike, R.C. and Todd, A.J., An immunocytochemical study of glycine receptor and GABA in laminae 1-III of rat spinal dorsal horn, J. Neurosci., 13 (19931 2371-2381.

Fig. 2. A shows a retrogradely labelled PSDC neuron. The inset is the area contained within the box at a higher magnification. The arrows indicate gephyrin-like immunoreactivity which is present at synaptic junctions. (scale bars: main picture = 10 kLm; inset = 0.5 /xm) B: this illustrates two boutons which are presynaptic to a PSDC neuron. The small arrows indicate the presence of gephyrin-like immunoreactivity at a synaptic junction which is formed by a bouton containing irregularly shaped vesicles. The large arrow indicates a gephyrin-negative synapse which is formed by a bouton containing round vesicles and forming an asymmetrical synaptic junction (scale = 0.5 /xm). C and D illustrate two sections from a series through a bouton which forms a gephyrin-immunoreactive synapse with a PSDC neuron. The bouton displays strong immunogold reactions for glycine (C) and GABA (D). Scale bars = 0.5 /zm.

132

D.J. Maxwell et al. /Brain Research 690 (1995) 127-132

[8] Noble, R. and Riddell, J.S., Cutaneous excitatory and inhibitory input to neurones of the postsynaptic dorsal column system in the cat, J. Physiol., 396 (1988) 497-513. [9] Noble, R. and RiddeU, J.S., Descending influences on the cutaneous receptive fields of postsynaptic dorsal column neurones in the cat, J. Physiol., 408 (1989) 167-183. [10] Ottersen, O.P., Storm-Mathisen, J. and Somogyi, P. Colocalization of glycine-like and GABA-like immunoreactivities in Golgi cell terminals in the rat cerebellum: a postembedding light and electron microscopic study, Brain Res., 450 (1988) 342-353. [11] Pfeiffer, G., Simler, R., Greenigloh, G. and Betz, H., Monoclonal antibodies and peptide mapping reveal ultrastructural similarities between the subunits of the glycine receptor of the rat spinal cord, Proc. Natl. Acad. Sci. USA, 81 (1984) 9389-9393. [12] Phend, K.D., Weinberg, R.J. and Rnstioni, A., Techniques to optimize post-embedding single and double staining for amino acid neurotransmitters, J. Histochem. Cytochem., 40 (1992) 1011-1020. [13] Pow, D.V. and Crook, D.K., Extremely high titre polyclonal antisera against small neurotransmitter molecules: rapid production, charac-

[14] [15] [16]

[17]

[18]

[19]

terisation and use in light- and electron-microscopic immunocytochemistry, .L Neurosci. Methods, 48 (1993) 51-63. Rustioni, A., Non-primary afferents to the nucleus gracilis from the lumbar cord of the cat, Brain Res., 51 (1973) 81-95. Rustioni, A., Non-primary afferents to the cuneate nucleus in the brachial dorsal funiculus of the cat, Brain Res., 75 (1974) 247-259. Rustioni, A. and Kaufman, A.B., Identification of cells of origin of non-primary afferents to the dorsal column nuclei of the cat, Brain Res., 27 (1977) 1-14. Todd, A.J., Spike, R.C., Chong, D. and Neilson, M., The relationship between glycine and gephyrin in synapses of the rat spinal cord, Eur. J. Neurosci., 7 (1995) 1-11. Todd, A.J. and Sullivan, A.C., Light microscopic study of GABAlike and glycine-like immunoreactivities in the spinal cord of the rat, J. Comp. Neurol., 296 (1990) 496-505. Triller, A., Cluzeaud, F., Pfeiffer, F., Betz, H. and Korn, H., Distribution of glycine receptors at central synapses; an immunoelectron microscope study, ./. Cell Biol., 101 (1985) 683-688.