Two vesicular glutamate transporters, VGLUT1 and VGLUT2, have recently been identified, and it has been reported that they are expressed by largely nonoverlapping populations of glutamatergic neurons in the brain. We have used immunocytochemistry with antibodies against both transporters, together with markers for various populations of spinal neurons, in an attempt to identify glutamatergic interneurons in the dorsal horn of the mid-lumbar spinal cord of the rat. The great majority (94-100%) of nonprimary axonal boutons that contained somatostatin, substance P or neurotensin, as well as 85% of those that contained enkephalin, were VGLUT2-immunoreactive, which suggests that most dorsal horn neurons that synthesize these peptides are glutamatergic. In support of this, we found that most somatostatin- and enkephalin-containing boutons (including somatostatin-immunoreactive boutons that lacked calcitonin gene-related peptide and were therefore probably derived from local interneurons) formed synapses at which AMPA receptors were present. We also investigated VGLUT expression in central terminals of primary afferents. Myelinated afferents were identified with cholera toxin B subunit; most of those in lamina I were VGLUT2-immunoreactive, whereas all those in deeper laminae were VGLUT1-immunoreactive, and some (in laminae III-VI) appeared to contain both transporters. However, peptidergic primary afferents that contained substance P or somatostatin (most of which are unmyelinated), as well as nonpeptidergic C fibres (identified with Bandeiraea simplicifolia isolectin B4) showed low levels of VGLUT2-immunoreactivity, or were not immunoreactive with either VGLUT antibody. As all primary afferents are thought to be glutamatergic, this raises the possibility that unmyelinated afferents, most of which are nociceptors, express a different vesicular glutamate transporter.

To determine whether GABA and glycine can act as cotransmitters at synapses in the rat spinal cord, we have compared the ultrastructural distribution of GABAA-receptor beta 3 subunit with that of the glycine receptor-associated protein gephyrin and combined this with postembedding detection of GABA and glycine. We also used a dual-immunofluorescence method to confirm that gephyrin was associated with the glycine-receptor alpha 1 subunit throughout the cord. GABAA beta 3-subunit immunoreactivity was restricted primarily to synapses, and at a majority of these synapses the presynaptic axon was GABA-immunoreactive. Many synapses showed both GABAA beta 3 and gephyrin immunoreactivity, and at most of these synapses GABA and glycine were enriched in the presynaptic axon. These results strongly support the idea that cotransmission by GABA and glycine occurs in the spinal cord.

Peripheral nerve injury leads to structural and functional changes in the spinal dorsal horn, and these are thought to be involved in the development of neuropathic pain. In the chronic constriction injury (CCI) model, abnormal 'dark' neurons and apoptotic nuclei have been observed in laminae I-III of the dorsal horn in the territory innervated by the injured sciatic nerve. These findings have been taken as evidence that there is significant neuronal death in this model, and it has been suggested that loss of inhibition resulting from death of GABAergic inhibitory interneurons contributes to the neuropathic pain. However, loss of neurons from the dorsal horn has not been directly demonstrated in neuropathic models, even though this issue is of considerable importance for our understanding of the mechanisms that underlie neuropathic pain. In this study, we have looked for evidence of neuronal death by using a stereological method (the optical disector) with NeuN-immunostaining, and examining spinal cords of naïve rats, and of rats that had undergone CCI or sham operations. All of the CCI animals showed clear signs of thermal hyperalgesia. However, the numbers of neurons in laminae I-III of the ipsilateral dorsal horn in these animals did not differ significantly from those on the contralateral side, nor from those of sham-operated or naïve animals. These results do not, therefore, support the suggestion that there is significant neuronal death in the dorsal horn in this model.

In order to provide information about the function of glycine in the superficial three laminae of the rat dorsal horn and the possible coexistence of glycine and GABA at synapses in this region, we have carried out a combined study involving preembedding immunocytochemistry with a monoclonal antibody to the glycine receptor and postembedding immunocytochemistry with antiserum to fixed GABA. Glycine receptor-like immunoreactivity was present at axodendritic and axosomatic synapses in all three laminae, and at dendrodendritic synapses in lamina II. Although axons that formed axoaxonic synapses were often presynaptic at immunoreactive axodendritic synapses (and thus probably contained glycine), the axoaxonic synapses themselves did not show glycine receptor-like immunoreactivity. Many of the profiles that were presynaptic at glycine receptor-immunoreactive synapses showed GABA-like immunoreactivity. These results suggest that glycine acts as a postsynaptic inhibitory transmitter at various types of synapses in laminae I-III, and that it may coexist with GABA at many synapses in this region. However, it appears that while glycine and GABA may both be released at axoaxonic synapses, either glycine does not act as a transmitter at these synapses, or else it acts at an atypical receptor that was not recognized by the antibody used in this study.

In order to determine whether somatostatin coexists with GABA or glycine in neurones in rat spinal dorsal horn, a combined pre- and post-embedding immunocytochemical study was carried out. One hundred six somatostatin-immunoreactive neurones located in lamina II and the dorsal half of lamina III were tested with antiserum or monoclonal antibody to GABA and none of these cells showed GABA-like immunoreactivity. However, 8 out of 13 somatostatin-immunoreactive neurones located deeper in the dorsal horn (ventral lamina III and lamina IV) showed glycine-like immunoreactivity, and 6 of these were also GABA-immunoreactive. We have previously shown that neurotensin-immunoreactive neurones in laminae II and III are also not immunoreactive when tested with GABA antiserum (Todd et al.: Neuroscience 47:685-691, 1992), and a double-labelling fluorescence method was therefore used to compare the distribution of somatostatin and neurotensin within the superficial dorsal horn. The two types of peptide-immunoreactivity were never found in the same profile. These results suggest that somatostatin and neurotensin are present in different populations of non-GABAergic neurones in rat superficial dorsal horn, but that some somatostatin-containing neurones in the deeper part of the dorsal horn contain glycine, with or without GABA.