A quantitative study of neurons which express neurokinin-1 or somatostatin sst2a receptor in rat spinal dorsal horn

Pergamon

PII:

Neuroscience Vol. 85, No. 2, pp. 459–473, 1998 Copyright  1998 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306–4522/98 $19.00+0.00 S0306-4522(97)00669-6

A QUANTITATIVE STUDY OF NEURONS WHICH EXPRESS NEUROKININ-1 OR SOMATOSTATIN sst2a RECEPTOR IN RAT SPINAL DORSAL HORN A. J. TODD,*‡ R. C. SPIKE* and E. POLGA u R*† *Laboratory of Human Anatomy, Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ, U.K. †Department of Anatomy, Histology and Embryology, University Medical School of Debrecen, Nagyerdei krt. 98, Debrecen H-4012, Hungary Abstract––The neurokinin-1 and somatostatin sst2a receptors have both been identified on spinal cord neurons. In this study we have estimated the proportions of neurons in different parts of the spinal cord which express these receptors, by using a monoclonal antibody against a neuronal nuclear protein named NeuN and combining the optical disector method with confocal microscopy. The NeuN antibody was initially tested on over 3200 neurons identified with antisera against a variety of compounds, including neuropeptides, enzymes and receptors, and also on astrocytes and oligodendrocytes. All of the neurons, but none of the glial cells that were examined possessed NeuN-immunoreactivity, which suggests that NeuN is a reliable marker for all spinal cord neurons. We found that approximately 45% of neurons in lamina I, 23–29% of those in laminae IV–VI and 18% in lamina X possessed the neurokinin-1 receptor, while the receptor was present on a smaller proportion of neurons in laminae II and III (6% and 11%, respectively). Thirteen percent of lamina I neurons and 15% of those in lamina II expressed the sst2a receptor. To provide further information about the types of neuron which possess the sst2a receptor, we searched for possible co-existence with the neurokinin-1 receptor as well as with GABA and glycine. sst2a and neurokinin-1 receptors were not co-localized on neurons in laminae I and II. All of the sst2a-immunoreactive neurons examined were also GABA-immunoreactive, and 83.5% were glycineimmunoreactive, indicating that the receptor is located on inhibitory neurons in the superficial dorsal horn. These results demonstrate the proportions of neurons in each region of the spinal cord which can be directly activated by substance P or somatostatin acting through these receptors. Levels of receptors can change in pathological states, and this method could be used to determine whether or not these changes involve alterations in the number of neurons which express receptors. In addition, the method can be used to estimate the sizes of neurochemically-defined populations of spinal cord neurons.  1998 IBRO. Published by Elsevier Science Ltd. Key words: spinal cord, immunocytochemistry, confocal microscopy, disector method.

The neuropeptides substance P and somatostatin are present in two distinct populations of fine-diameter primary afferent axons which terminate in the superficial laminae of the spinal cord,3,13–15 and can be released following different types of peripheral noxious stimulation.10,21,24,32 Both peptides are also present in interneurons within the dorsal horn11,16,18 and in axons descending from the brainstem and terminating within the spinal cord.20,30 Substance P-containing axons form a dense plexus in lamina I and the dorsal part of lamina II (which is derived ‡To whom correspondence should be addressed. Abbreviations: AMPA, á-amino-3-hydroxy-5-methylisoxazole-4-propionic acid; ChAT, choline acetyltransferase; CNPase, 2 ,3 -cyclic nucleotide-3 -phosphodiesterase; Cy5, cyanine 5.18; FITC, fluorescein isothiocyanate; GFAP, glial fibrillary acidic protein; GluR, glutamate receptor; LRSC, lissamine rhodamine; MOR, µ-opioid receptor; NK1, neurokinin-1; NOS, nitric oxide synthase; NPY, neuropeptide Y.

mainly from primary afferents), but in addition significant numbers of substance P-containing axons are present in the deeper laminae of the dorsal horn, the area around the central canal and in the motor nuclei of the ventral horn. Axons which contain somatostatin are present throughout laminae I and II, and the majority of these are thought to be derived from the numerous somatostatin-containing neurons in lamina II,2,23,44 with the primary afferent component making up a minority. Receptors for both peptides are present in the spinal cord, and in each case the laminar distribution of the receptors generally matches that of the corresponding peptide. Spinal neurons with the neurokinin-1 (NK1) receptor (on which substance P acts) show extensive immunostaining on somatic and dendritic membrane, such that antibodies against the receptor appear to reveal the entire dendritic arborization.27,60 The highest density of NK1 receptorimmunostaining is observed in lamina I,4,6,26–28,36,60

459

460

A. J. Todd et al.

and neurons in this lamina which possess the receptor are likely to be targets for substance P-containing primary afferents.53 NK1 receptor-immunoreactive cells are also present in moderate numbers throughout the remainder of the dorsal horn, with the exception of lamina II where they are infrequent.4,6,26,27 The low density of NK1 receptor-immunoreactive neurons in lamina II is surprising, since this lamina contains many substance P-containing primary afferents. However, dorsally-directed dendrites of neurons in laminae III and IV which possess the receptor pass up through the superficial dorsal horn and receive a substantial synaptic input from substance P-containing primary afferents.35 Radioligand binding has revealed a moderate density of somatostatin receptor in the superficial dorsal horn,57 matching the large number of somatostatincontaining axons in this area. Somatostatin acts on G-protein-coupled receptors, and five different somatostatin receptors (sst1–5) have so far been cloned,42 one of which (sst2) exists either as a longer form (sst2a) or a shorter form (sst2b), produced by alternative splicing.59 At least some of the somatostatin binding sites in superficial dorsal horn appear to be the sst2 receptor, since this has been localized to the region both with a specific ligand (BIM 23027),17 and also with an antibody directed against the predicted C-terminus of the sst2a receptor protein45 (however in situ hybridization studies have failed to detect sst2 mRNA in the mouse or rat spinal trigeminal nucleus,5,46 suggesting that there may be a low turnover of the protein). sst2a-like immunoreactivity was seen as a dense plexus in laminae I and II of the spinal cord, and immunoreactive cell bodies were observed within this plexus.45 Only occasional immunoreactive neurons were found in deeper parts of the dorsal horn. Immunocytochemical studies have demonstrated that several populations of neurons in the spinal cord can be identified with antibodies directed against transmitters, neuropeptides and enzymes, and this approach has been particularly useful in distinguishing different types of neuron in the superficial dorsal horn. GABA and glycine, the major inhibitory transmitters are present in certain neurons in each region of the gray matter, and in many cases co-exist in individual cells.56 Neurotensin and somatostatin are present in some non-GABAergic neurons in the superficial laminae, while neuropeptide Y (NPY) and galanin are found in neurons in this region with GABA but without glycine.47,55 Neurons which contain nitric oxide synthase (NOS) are found throughout the gray matter of the spinal cord,58 and most of them contain either GABA, glycine or both inhibitory transmitters.50 Cholinergic neurons are present in the dorsal horn (laminae III–VI) and around the central canal. Many of these cells also contain NOS,50 and those in lamina III have been shown to be GABAergic.52 Certain receptors for neurotransmitters and peptides have been identified on specific

populations of dorsal horn neurons. The great majority of cells which possess the NK1 receptor (including all of those in lamina I)26 and most lamina II cells with the µ-opioid receptor (MOR-1),22 are not GABA-immunoreactive. We have recently demonstrated that subunits of the á-amino-3-hydroxy-5methyl-isoxazole-4-propionic acid (AMPA) receptor are also differentially distributed in laminae I–III: glutamate receptor 1 (GluR1) is mainly associated with GABAergic neurons and GluR2/3 with nonGABAergic cells.49 Although the laminar distribution of spinal neurons which possess the NK1 and sst2a receptors has been described, there is little information available concerning the proportions of neurons in each lamina of the dorsal horn which possess the receptors, apart from an estimate that 5% of lamina I neurons have the NK1 receptor.6 Quantitative information is valuable, not only because it reveals the proportion of neurons that can be directly affected by the peptides, but also because it provides a baseline for determining whether alterations in receptor density (for example the up-regulation of NK1 receptors that follows peripheral inflammation or nerve injury1) involve a change in the number of neurons which express a receptor, or simply alteration in the level of expression by neurons that normally possess it. Part of the difficulty in carrying out quantitative studies results from the small size of neurons in certain parts of the spinal cord (in particular laminae I–III and lamina X), which means that they cannot reliably be distinguished from glial cells with nuclear stains. In this study we have used a monoclonal antibody against a nuclear protein (NeuN) which is apparently present in the great majority of neurons throughout the CNS, but not in glial cells.33 In cells which contain NeuN, the protein is invariably found in the nucleus and may also be present at lower levels in the cytoplasm.33 In the first part of the study, we examined the suitability of the NeuN antibody as a marker for all spinal neurons, by carrying out double-labelling experiments with a variety of antibodies against neuronal and glial antigens. In the second part, we used the NeuN antibody to determine the proportions of neurons which possessed NK1 receptor throughout the dorsal horn and the area around the central canal, and of neurons with sst2a in laminae I and II. At present there is little information concerning the types of spinal neuron which possess the sst2a receptor, and so in the final part of the study we have attempted to characterize these cells by determining whether they also possess the NK1 receptor or the inhibitory transmitters GABA and glycine. EXPERIMENTAL PROCEDURES

Immunocytochemistry Lumbar spinal cord segments were obtained from 11 adult male Albino Swiss rats (250–370 g, Glasgow University), which had been deeply anaesthetized and

NK1 and sst2a receptors in spinal cord perfused with a fixative containing 4% formaldehyde in 0.1 M phosphate buffer. Blocks of spinal cord were postfixed for 2–18 h, rinsed in buffer and cut into transverse sections (60 µm-thick) with a Vibratome. Before they were immunostained, the sections were immersed in blocking serum consisting of 5% donkey serum in phosphate-buffered saline with 0.3% Triton X-100, and this solution was also used as the diluent for all antibodies. Incubations in primary antibodies were for approximately 16 h (except for sections reacted with antisera against GABA or glycine, which were incubated for three to four days), and those in fluorescent or biotinylated secondary antibodies were for 2 h. Fluorescent secondary antibodies (Jackson Immunoresearch) raised in donkey against IgG of the appropriate species and labelled with fluorescein isothyocyanate (FITC), lissamine rhodamine (LRSC) or cyanine 5.18 (Cy5) were used at 1:100. The distribution of cells in the spinal cord with NeuNimmunoreactivity was studied in sections from two rats, which were incubated in NeuN antibody (mouse monoclonal; Chemicon; diluted 1:2000 or 1:10,000), followed by biotinylated donkey anti-rabbit IgG (Jackson Immunoresearch; diluted 1:500) and avidin–peroxidase (Sigma; diluted 1:1000). Enzyme activity was revealed with diaminobenzidine in the presence of hydrogen peroxide. To test the validity of NeuN as a marker for neurons, double immunofluorescence labelling was performed with antibody to NeuN (diluted 1:1000–2000) and each of the following neuronal markers: rabbit antisera against neurotensin, somatostatin, NPY and galanin (all from Peninsula Laboratories; diluted 1:1000); goat antiserum against choline acetyltransferase (ChAT; Chemicon; diluted 1:100); sheep antiserum against neuronal NOS12 (a gift from Dr P. C. Emson; diluted 1:2000); rabbit antisera against GluR1 and against GluR2/3 subunits of the AMPA receptor (Chemicon; each diluted 1:200); rabbit antiserum against MOR-1 (Gramsch Laboratories; diluted 1:1000); rat antiserum against a paraformaldehyde conjugate of glycine39 and rabbit antiserum against a paraformaldehyde conjugate of GABA39 (both gifts from Dr D. V. Pow; diluted 1:1000 and 1:5000, respectively). Sections were subsequently incubated in a mixture of FITC- or LRSC-anti-rabbit, anti-goat or anti-rat IgG and Cy5-, LRSC- or FITC-anti-mouse IgG. This selection of transmitters, neuropeptides, enzymes and receptors was chosen because antibodies raised against them (in species other than mouse) have been shown to stain populations of neuronal cell bodies in the spinal cord.22,25,47,49,50,54–56 Double-immunofluorescence labelling was also used to determine whether astrocytes or oligodendrocytes were NeuN-immunoreactive. For astrocytes, sections were incubated in antibody to NeuN and rabbit antiserum against glial fibrillary acidic protein (GFAP; Sigma; diluted 1:2000), followed by FITC-anti-rabbit IgG and Cy5-anti-mouse IgG, and subsequently stained with propidium iodide (Sigma; 1% in phosphate-buffered saline) in the presence of RNAse (Sigma; 10 mg/ml) to reveal cell nuclei. Oligodendrocytes were immunostained with a mouse monoclonal antibody against 2 ,3 -cyclic nucleotide-3 phosphodiesterase (CNPase), a protein which is present in their cell bodies.51 Since in this case both primary antibodies were mouse IgGs, sections were initially incubated in CNPase antibody (Sigma; diluted 1:200) and then in FITC-antimouse IgG, followed by antibody to NeuN and then Cy5-anti-mouse IgG. They were stained with propidium iodide in the presence of RNAse as described above. For the quantitative analysis of neurons expressing NK1 or sst2a receptors, sections from L4 segments were incubated in antibody to NeuN together with either rabbit antiserum against NK1 receptor60 (diluted 1:10,000) or guinea-pig antiserum against sst2a receptor (Gramsch Laboratories; diluted 1:2000). They were subsequently reacted with Cy5anti-mouse IgG and FITC-anti-rabbit or anti-guinea-pig IgG and stained with propidium iodide.

461

To determine whether any spinal neurons possessed both NK1 and sst2a receptors, sections were incubated in a mixture of rabbit antiserum against NK1 receptor and guinea-pig antiserum against sst2a, followed by LRSC-antirabbit IgG and either FITC-anti-guinea-pig IgG or biotinylated anti-guinea-pig IgG (Jackson Immunoresearch; 1:500). Sections treated with the biotinylated secondary antibody were finally incubated in avidin–FITC (Vector Laboratories; diluted 1:10,000). In order to determine whether sst2a receptor was present on GABAergic or glycinergic neurons, sections were incubated in a mixture of guinea-pig antiserum against sst2a and either rabbit antiserum against GABA39 or rat antiserum against glycine,39 followed by FITC-anti-guinea-pig IgG and LRSC-anti-rat or anti-rabbit IgG. Antibodies The NK1 receptor antiserum was donated by Dr S. Vigna, and had been raised against a synthetic peptide corresponding to the 15 amino acid residues at the C-terminus of the rat NK1 receptor, which was coupled to bovine thyroglobulin.60 The guinea-pig antiserum against sst2a was raised against the C terminal 15 amino acids of the peptide sequence of the receptor, coupled to keyhole limpet haemocyanin. Immunostaining with this antiserum was blocked by incubation with the peptide antigen (manufacturer’s specification). The antisera against paraformaldehyde-conjugates of GABA and glycine react with the corresponding amino acid in formaldehyde-fixed tissue sections and this staining is blocked by pre-absorption with a formaldehyde-conjugate of the corresponding amino acid.39 On dot-blots these antisera show no detectable cross-reactivity with the other amino acid or with glutamate, aspartate or glutamine.39 Analysis All sections reacted with fluorescent antibodies were examined with a confocal laser scanning microscope (Bio-Rad MRC 1024) equipped with a krypton-argon laser. The relationship of NeuN to the various neuronal markers (GABA, glycine, neurotensin, somatostatin, NPY, galanin, ChAT, NOS, GluR1, GluR2/3 and MOR-1) was examined in sections from two rats for each antibody. For each of these antibodies except that against ChAT, over 200 immunoreactive neurons in laminae I–III (approximately 100 cells/antibody from each rat) were initially identified by scanning to reveal the corresponding type of immunofluorescence, and then examined to determine whether they were also immunoreactive with the NeuN antibody. In addition, glycine- and NOS-immunoreactive neurons in the deep dorsal horn (laminae IV–VI) and in the area around the central canal (lamina X), and ChAT-immunoreactive neurons in lamina X and in the motor nuclei of the ventral horn (lamina IX) were tested for NeuN-immunoreactivity in the same way. For each of these antibodies, over 200 neurons in each region were examined (approximately 100 cells/antibody/region from each rat). The relation of NeuN-immunostaining to that seen with GFAP and CNPase antibodies was also examined in sections from two rats for each of these antibodies. For each antibody, over 200 immunoreactive cells throughout the gray matter were identified (approximately 100 cells/ antibody/rat) and then examined to see whether they were NeuN-immunoreactive. For the quantitative analysis of neurons which possessed NK1 or sst2a receptors, the optical disector method8 was carried out with confocal microscopy. For both receptor antibodies, two sections each from three rats were selected (before examining receptor immunostaining), and the dorsal horn on one side of the section was analysed. Since sst2aimmunoreactive neurons are rarely seen below lamina II,45 only laminae I and II and the overlying white matter were examined for this receptor. NK1 receptor-immunoreactive

462

A. J. Todd et al.

neurons are present throughout the entire dorsal horn, and in this case the spinal gray and white matter as far ventrally as the central canal were studied. The sections were scanned sequentially with each of the three lines of the laser through a 40 oil-immersion lens, and z-series consisting of 16 optical sections separated by 1 µm intervals were obtained. Each series was started at the most superficial plane for which tissue was present throughout the field of view. Since the field with this lens is a square of approximately 200 µm, for each Vibratome section numerous slightly overlapping z-series were constructed: approximately six series were required for the superficial dorsal horn (sst2a receptor) and 30 series for the entire dorsal horn (NK1 receptor) from each section. Merged images of Cy5 (representing NeuNimmunoreactivity) and propidium iodide fluorescence were constructed for each z-series and viewed with a computer program which allowed drawings to be made of structures appearing within confocal z-series (Neurolucida for Confocal; MicroBrightField Inc.). For each series of 16 optical sections, the 5th was designated as the reference section and the 12th as the look-up section. (This separation was chosen to prevent the possibility that neuronal nuclei could lie entirely between the reference and look-up sections). Neuronal nuclei (double-labelled with propidium iodide and Cy5) which were present within the reference section, but which had disappeared by the look-up section were drawn onto an outline of the spinal gray matter, and subsequently examined in the images obtained with the 488 nm line of the laser to determine whether they showed NK1- or sst2a-immunoreactivity. Low-magnification images of each section were obtained through 4 and 10 objective lenses by scanning with the 488 nm line (to reveal FITC) and also with light transmitted through a dark-field condenser. These images were superimposed on the drawings, after appropriate correction for the magnification of the different objective lenses, in order to allow the complete outline of the gray matter to be traced. Since lamina II appears as a dark region with dark-field optics (due to its lack of myelin), its dorsal and ventral borders could also be identified, and these were added to the drawings of the sections. Other laminar boundaries were drawn from an atlas of rat CNS.38 To determine whether any neurons possessed both NK1 and sst2a receptors, sections which had been reacted with both antibodies from three rats were scanned sequentially through a 40 oil-immersion objective lens with 488 and 568 nm lines to reveal sst2a and NK1 receptorimmunostaining, respectively. Altogether, 272 neurons with sst2a receptor- and 147 neurons with NK1 receptor-immunoreactivity in superficial dorsal horn were examined. The relationship between sst2a- and GABA- or glycineimmunoreactivity was examined in sections from two rats. Sections were scanned with a 40 oil-immersion objective lens and the 488 nm line to identify sst2a-immunoreactive neurons, and then with the 568 nm line to determine whether these neurons were GABA- or glycineimmunoreactive. For each amino acid, 100 sst2aimmunoreactive neurons from each of two rats were examined in this way. RESULTS

Appearance of NeuN in rat spinal cord The distribution of NeuN-immunoreactivity in rat spinal cord seen with immunoperoxidase staining (Fig. 1) was the same as that previously reported in the mouse.33 Immunoreactive cells were present throughout all parts of the gray matter but very few were seen in the white matter, with the exception of the lateral spinal nucleus and a popula-

tion of cells lying dorsal to the medial part of lamina I. A few immunoreactive cells were scattered elsewhere in the lateral and dorsal funiculi (Fig. 1). Ependymal cells lining the central canal were not NeuN-immunoreactive. Although the staining was generally strongest in the nucleus of immunoreactive cells, there was frequently staining in the perikaryal cytoplasm and this sometimes extended into proximal dendrites. The density of immunostaining varied considerably among immunoreactive cells. NeuN compared to neuronal and glial markers Over 2000 cells in laminae I–III (labelled with antisera to GABA, glycine, neurotensin, somatostatin, NPY, galanin, NOS, GluR1, GluR2/3 or MOR-1), 400 cells in laminae IV–VI (labelled with antisera to glycine or NOS), 600 cells in lamina X (labelled with antisera to glycine, NOS or ChAT) and 200 cells in the motor nuclei (labelled with antiserum to ChAT) were examined with doubleimmunofluorescence and confocal microscopy, and all of these cells were immunoreactive with the NeuN antibody (Fig. 2). In addition, during the examination of sections reacted with NeuN antibody and antisera to NK1 or sst2a receptors, all cells immunoreactive with each of the receptor antisera were also found to be NeuN-immunoreactive. In sections which had been reacted with NeuN antibody and stained with propidium iodide, many nuclei which were not NeuN-immunoreactive were seen: these were present throughout the gray and white matter of the spinal cord and were among the smallest nuclei observed. None of the astrocytes (immunostained with GFAP antiserum) was NeuNimmunoreactive (Fig. 3a–c). In sections immunostained with CNPase and NeuN antibodies, CNPaseimmunostaining was seen with FITC (Fig. 3d), while Cy5 revealed both CNPase- and NeuNimmunoreactivity (Fig. 3f). CNPase was restricted to the cytoplasm of oligodendrocytes, and although this was also stained with Cy5, the nuclei of these cells never showed Cy5-staining and were therefore not NeuN-immunoreactive. Neurokinin-1 and sst2a receptor-immunostaining The distribution of NK1 receptor-immunostaining was the same as that described previously:4,6,26–28,36 the strongest immunoreactivity was observed in lamina I (Figs 4c, 5b), there was moderate immunostaining in the deep dorsal horn (particularly in its medial part) and the area around the central canal (lamina X), while lamina II had a low level except where it was crossed by immunoreactive dendrites. The pattern of sst2a-immunoreactivity was also similar to that reported previously,45 with the densest immunostaining in lamina II, lighter staining in lamina I and little staining elsewhere in the gray matter (Figs 4d, 5e).

NK1 and sst2a receptors in spinal cord

463

Fig. 1. The distribution of NeuN-immunoreactivity in the spinal cord revealed with the immunoperoxidase technique. NeuN-immunoreactive cells are present throughout the gray matter of the cord. In the white matter, several cells are visible in the lateral spinal nucleus (LSN) and in the lateral part of the dorsal columns, while a few can be seen elsewhere in the dorsal and lateral funiculi. Scale bar=200 µm.

Examination of these sections at low magnification with dark-field microscopy and also with immunofluorescence to reveal receptor- or NeuN immunostaining revealed a variation in the arrangement of the superficial laminae across the mediolateral extent of the dorsal horn (Fig. 4). In the central part of the superficial dorsal horn, just beneath the medial part of Lissauer’s tract, the distance between the dark band seen with dark-field microscopy (which was taken to represent lamina II) and the dorsal limit of the gray matter, was substantially increased (Fig. 4a). Immunostaining with NeuN antibody often revealed that the dorsal part of lamina II had a much higher density of neuronal cell bodies than its ventral part (as reported previously in cat43 and rat31), and the line separating these two subdivisions of lamina II also deviated further from the dorsal edge of the gray matter in the central part of the horn (Fig. 4b). The dense band of sst2a-immunostaining, which in medial and lateral parts of the dorsal horn occupied the full thickness of lamina II was also displaced away from the dorsal edge in the central region (Fig. 4d), while the band of dense NK1 receptor-immunostaining (Fig. 4c) and of NK1 receptor-immunoreactive cell bodies (Figs 5, 6) was much thicker in this part. At high magnification both NK1 and sst2a receptor immunoreactivities were observed on the plasma

membrane of immunoreactive cells, and in single optical sections could be seen surrounding the cell body and in several cases, the emerging dendrites (Fig. 5b,e). In some neurons, weak cytoplasmic staining was also observed. The results of the quantitative analysis of NK1 and sst2a receptor-immunoreactive neurons are shown in Tables 1 and 2, and examples of sections which were analysed with the disector method are illustrated in Figs 5 and 6. NK1 receptorimmunoreactivity was present on a large proportion (45%) of neurons in lamina I, but a much lower proportion of those in laminae II and III (6% and 11%, respectively). In the deep part of the dorsal horn (laminae IV–VI) between 23 and 29% of neurons were NK1 receptor-immunoreactive, while 18% of neurons in lamina X were immunoreactive. Thirteen percent of neurons in lamina I and 15% of those in lamina II were sst2a-immunoreactive.

sst2a receptor compared with neurokinin-1 receptor, GABA and glycine Although the distribution of NK1 and sst2a receptor-immunoreactive neurons overlapped in laminae I and II, the two types of immunoreactivity were invariably present on different neurons, and no

464

A. J. Todd et al.

Fig. 2. Confocal images showing the relation of NeuN to GluR2/3 subunits of the AMPA receptor in lamina III (a,b), NOS in lamina IV (c,d) and ChAT in lamina X (e,f). In each case, all of the immunoreactive neurons seen in the left hand images (arrowheads in a,c,e) are also NeuN-immunoreactive (b,d,f). All of the confocal images of fluorescence in this and subsequent figures were obtained by scanning a single optical section with one line of the laser. Scale bar=20 µm.

examples of double-stained cells were observed (Fig. 7). Immunostaining in the spinal cord with the antisera against formaldehyde conjugates of GABA and glycine showed a generally similar distribution to

that seen with conventional antibodies raised against glutaraldehyde conjugates of these amino acids.54,56 GABA-immunoreactivity has previously been shown to be strongest in laminae I–III of the dorsal horn,54 and the GABA-formaldehyde antiserum revealed

NK1 and sst2a receptors in spinal cord

465

Fig. 3. Confocal images showing NeuN-immunostaining in relation to glial markers. (a) Two GFAPimmunoreactive astrocytes in lamina II are visible (arrowheads). (b) The nuclei of these and several other cells are seen stained with propidium iodide (PI). (c) The two astrocytes are not immunoreactive with the NeuN antibody. (d) CNPase-immunostaining revealed with FITC shows a single oligodendrocyte in lamina II, and its nucleus can be seen with propidium iodide (e). Scanning to reveal Cy5 also shows the cytoplasm of the oligodendrocyte, because Cy5-anti-mouse IgG has also bound to the CNPase antibody. In addition, neurons labelled with NeuN can be seen. The nucleus of the oligodendrocyte is not stained with Cy5, and is therefore not NeuN-immunoreactive. Scale bar=20 µm.

many immunoreactive neurons in this region (Fig. 8c), however fewer immunoreactive neurons could be distinguished with this antibody in deeper laminae. As has been shown previously with antibodies against glutaraldehyde conjugates of glycine,50,56 the glycine-formaldehyde antiserum revealed a smaller number of immunoreactive neurons in laminae I–II (Fig. 8d), many in the deeper laminae of the dorsal horn and also some immunoreactive cells in the intermediate gray matter and the area around the central canal. Of the 200 sst2a-immunoreactive neurons in laminae I and II which were tested with GABA or glycine antisera, all were GABAimmunoreactive (Fig. 8a,c), while 167 (83.5%) were glycine-immunoreactive (Fig. 8b,d).

DISCUSSION

Validity of NeuN as a neuronal marker The monoclonal antibody used in this study was raised by Mullen et al.33 against cell nuclei extracted from mouse brain, and was shown to react with a nuclear protein apparently specific to neurons, which was named NeuN. Immunocytochemistry with this antibody indicated that it labelled most, if not all, neurons in many parts of the CNS and peripheral ganglia, however a few neuronal populations such as cerebellar Purkinje cells, mitral cells of the olfactory bulb and retinal photoreceptors were not labelled with the antibody.33 The sparse labelling of white matter indicated that glia were probably not detected, and the lack of staining of astrocytes was

demonstrated directly by double-labelling with antibody against NeuN and antiserum against GFAP.33 Several subsequent studies have used the NeuN antibody to distinguish neurons from non-neuronal cells in sections of CNS tissue.19,37,61 While it is clear from previous reports and from the results of the present study that immunostaining with the NeuN antibody is restricted to neuronal cells, a potential limitation to its use in quantitative studies of spinal neurons is that certain neuronal populations in other CNS regions are consistently not labelled with the antibody. If significant numbers of neurons in the spinal cord were not NeuNimmunoreactive, this would invalidate its use in studies of the kind attempted here. Although it is not possible to test every type of neuron within the spinal cord, we surveyed large numbers of neurons in different parts of the gray matter. The antisera which we used are likely to have detected several populations of inhibitory interneurons (containing GABA, glycine, NOS, galanin or NPY25,47,50,55), excitatory interneurons (revealed with antibodies to neurotensin, somatostatin, MOR-1 or GluR2/322,49,55) as well as motor neurons. Every one of over 3200 neurons examined in this part of the study was immunoreactive with the NeuN antibody, and although we cannot rule out the possibility that there are neurons in the spinal cord that lack NeuN, it seems very unlikely that they are present in significant numbers. The strength of NeuNimmunoreactivity varied considerably between different neurons, from weak staining restricted to the nucleus to strong staining which filled the cell

466

A. J. Todd et al.

resenting only NeuN-immunoreactivity, as well as those showing merged NeuN- and propidium iodide fluorescence. The quantitative method

Fig. 4. Low magnification images of a section reacted with antiserum against NK1 receptor (a–c) or sst2a receptor (d). (a), (b) and (c) show the same field scanned with light transmitted through a dark-field (DF) condenser, with the 647 nm line of the laser (to reveal NeuN) and with the 488 nm line (to reveal NK1 receptor), respectively. (d) shows another section scanned with the 488 nm line to reveal sst2a receptor. The dashed lines in (a) show the limits of the translucent zone, which represents lamina II, and these have been superimposed onto the image showing NeuN-immunoreactivity (b), which reveals that the neuronal density in the outer part of lamina II is much greater than that in the inner part. The boxes in (c) and (d) show the areas scanned at higher magnification in Fig. 5. Scale bar=100 µm.

body and extended into primary dendrites. For this reason, it was necessary to look carefully for weak NeuN-immunoreactivity over nuclei, and this was most easily done by examining confocal images rep-

In order to provide reliable counts of cells in tissue sections it is necessary to use an unbiased sampling method, to avoid the over-representation of large cells that would otherwise occur. The disector method provides a simple way of obtaining accurate cell counts, and is based upon the fact that while cells vary in size, each cell (or cell nucleus) has only one upper or lower surface.8 Structures are counted only if their upper or lower surface lies between two planes (physical or optical sections) which are spaced such that no structure could lie entirely between them. Confocal microscopy is particularly well-suited to the disector method, since a series of optical sections can readily be obtained, and cells can be followed through the series. In the present study we chose optical sections 5 and 12 µm below the surface of the Vibratome section, and counted cells for which the lower surface of the nucleus lay between these planes. This meant that all the cells which were included in the sample are likely to have been sectioned through the cell body by the Vibratome, and therefore there should have been no problem resulting from inadequate penetration of antibodies. The reference section was set at 5 µm below the surface of the Vibratome section because many cells were cut by the reference section near the lower surface of their nucleus and were therefore very small. By following the cells through earlier sections in the series it was possible to view them where their cell bodies were larger. The sections used for the quantitative analysis were stained with propidium iodide because although NeuN-immunostaining was sometimes restricted to the nucleus, it frequently extended into perikaryal cytoplasm, and in these cells NeuNimmunoreactivity alone could not be used to identify nuclear boundaries. Because the Krypton-Argon laser allows the use of three different fluorescent dyes, the combination of NeuN antibody and propidium iodide to identify neuronal nuclei still left a third channel which could be used to detect the receptor antiserum. Since the NeuN antibody is a mouse monoclonal, the method used here could be applied to a wide variety of compounds which are present in neurons, and for which antisera raised in species other than mouse are available. This quantitative approach, together with studies of co-localization, should improve our understanding of the complex neurochemical populations of neurons in the superficial dorsal horn.55 Dark-field illumination of the dorsal horn reveals a band corresponding to lamina II which is dark due to the lack of myelin, and dorsal to this is a lighter region (Fig. 4a). We found that while this lighter

Fig. 5. High magnification scans to show NK1 receptor- (a–c) and sst2a receptor- (d–f) immunostaining in laminae I and II. In each case, the left hand image (a,d) was formed by merging a scan of Cy5 (blue, representing NeuN) and propidium iodide (red). In these images, neuronal nuclei appear purple, while nuclei of non-neuronal cells are red. Pure blue staining represents cytoplasmic NeuN immunoreactivity in neurons which are not ‘‘sectioned’’ through the nucleus. The central image (b,e) shows the same optical section scanned to reveal FITC, which represents either NK1 receptor (b) or sst2a (d). Because the receptor-immunostaining is mainly associated with the cell membrane, immunoreactive neurons appear as hollow green structures, and these have been marked with asterisks in each image. The right hand images (c,f) are produced by merging the scans of Cy5, propidium iodide and FITC. Scale bar=20 µm.

468

A. J. Todd et al.

the central part of the dorsal horn, and a similar finding has been illustrated in mid-lumbar spinal cord of the rat by Nakaya et al. (Fig. 15d in Ref. 36). In addition, NK1 receptor-immunoreactive neurons which are infrequent in the medial and lateral parts of lamina II were present throughout this thickened region in the central part (Fig. 6). In the present study we therefore defined the lighter region seen at the dorsal surface of the dorsal horn with dark-field optics as lamina I, and accordingly this lamina was much thicker (up to 100 µm) in the central part of the horn than in the lateral and medial parts (Figs 4, 6). In the medial half of the dorsal horn, lamina I has an indistinct, reticulated dorsal border31 and neurons are found at various positions in the overlying white matter of the dorsal columns (Fig. 1). Since there is no clear dorsal limit to lamina I, and these cells showed a similar pattern of immunostaining to that of lamina I cells with both NK1 and sst2a receptor antisera, they were included with those of lamina I in the quantitative analysis (Tables 1, 2). Neurokinin-1 and sst2a receptors

Fig. 6. Drawings of one of the sections used for the quantitative analysis of sst2a receptor-immunoreactivity, and one of those used for analysis of NK1 receptorimmunoreactivity. The symbols represent NeuNimmunoreactive cells whose nuclei (stained with propidium iodide) were visible on the reference section but which had disappeared on the look-up section. Filled symbols show cells which were sst2a or NK1 receptor-immunoreactive and open symbols show cells which were not receptorimmunoreactive.

region is extremely thin (between 10 and 20 µm) laterally and medially, it is much thicker (up to 100 µm) in the central part of the dorsal horn, as has been reported previously.34 The difference in thickness could be due either to a broadening in the central region of lamina I or else of the outer part of lamina II. In the cat dorsal horn, a discrete thickening of lamina I has been identified by Snyder48 and termed the ‘‘dorsal cap’’, and a similar thickening of lamina I was previously illustrated in the monkey.40 In their cytoarchitectonic study of rat spinal cord, Molander et al.31 observed a triangular thickening of lamina I just below the medial part of Lissauer’s tract (Figs 1–3 in Ref. 31), and suggested that this might be homologous to the ‘‘dorsal cap’’ identified by Snyder.48 The more medial location of this thickening in the rat, compared with that seen in cat and monkey, is probably due to the more medial position of Lissauer’s tract in the rat. In sections reacted with the NK1 receptor antiserum, the plexus of strongly immunoreactive dendrites (which is virtually restricted to lamina I in the medial and lateral parts of the superficial dorsal horn) was much thicker in

One of the main findings of the present study was that nearly half of the neurons in lamina I and approximately a quarter of those in laminae IV–VI possess the NK1 receptor, and can therefore presumably be influenced directly by substance P, released from primary afferents, local interneurons or pathways descending from the brainstem. Our result for lamina I neurons was much higher than that previously reported by Brown et al.,6 who estimated from sections reacted by an immunoperoxidase method and counterstained with Cresyl Violet, that approximately 5% of lamina I neurons had the receptor. The discrepancy between the two findings is probably due to the improved spatial resolution of the confocal microscope (Fig. 5), which allowed detection of the receptor on the plasma membrane of weakly immunoreactive neurons, together with the use in the present study of the NeuN antibody to exclude non-neuronal cells. We have previously shown that within lamina I, the NK1 receptor is restricted to neurons which are not GABA-immunoreactive,26 and which are therefore likely to be excitatory. Since GABAimmunoreactive neurons make up 28% of the population in lamina I,56 NK1 receptors must be present on the majority of neurons which are not GABAergic (approximately 63%). Some of the NK1 receptor-immunoreactive neurons in lamina I project to the brain,9,29 and it has been estimated that 77% of lamina I spinothalamic tract neurons29 and 70% of those which belong to the spinoparabrachial tract9 express NK1 receptors. This suggests that neurons belonging to these two tracts more often possess the receptor than other non-GABAergic neurons in lamina I.

NK1 and sst2a receptors in spinal cord

469

Table 1. Proportions of neurons which possessed NK1 receptor-immunoreactivity Total of neurons sampled

NK1 receptorimmunoreactive neurons

% of neurons that were NK1 receptorimmunoreactive

165 375 317 204 251 138 82 7 10

75 24 35 57 72 32 15 4 6

45.4 6.4 11 27.9 28.7 23.2 18.3 57.1 60

Lamina I* Lamina II Lamina III Lamina IV Lamina V Lamina VI Lamina X Lateral spinal nucleus Lateral white matter**

Results pooled from six sections (two from each of three rats) showing the numbers and proportions of neurons which were immunoreactive with NK1 receptor antiserum. *The counts for lamina I include cells in the overlying white matter. **This group consists of cells in the lateral funiculus which were not in the lateral spinal nucleus. Table 2. Proportions of neurons with sst2a receptor-immunoreactivity

Lamina I* Lamina II

Total of neurons sampled

sst2a receptorimmunoreactive neurons

% of neurons that were sst2a receptorimmunoreactive

165 328

22 48

13.3 14.6

Results pooled from six sections (two from each of three rats) showing the numbers and proportions of neurons which were immunoreactive with sst2a receptor antiserum. *The counts for lamina I include cells in the overlying white matter.

Fig. 7. Confocal images of part of laminae I and II from a section reacted with antisera to NK1 and sst2a receptors. The same optical section is shown (a) after scanning with the 568 nm laser line (to reveal LRSC/NK1 receptor), and (b) with the 488 nm line (to reveal FITC/sst2a receptor). Three NK1 receptor-immunoreactive cells (arrowheads) and several sst2a receptor-immunoreactive cells (six of which are marked with asterisks) are visible. All of these cells are only immunoreactive with one of the receptor antisera. Scale bar=20 µm.

Little was previously known about the types of neuron in superficial dorsal horn which express the sst2a receptor. Schindler et al.45 reported that sst2aimmunoreactive neurons were located at the border between laminae I and II, however we observed

immunoreactive cells throughout both laminae. We have been unable to achieve satisfactory immunostaining with this antiserum on tissue fixed with 1% glutaraldehyde, and therefore have not been able to combine detection of sst2a receptor with

470

A. J. Todd et al.

Fig. 8. Confocal images of lamina II to show the relationship between sst2a receptor and GABA or glycine. (a,c) The same optical section is shown scanned with the 488 nm line (to reveal FITC/sst2a receptor) and the 568 nm line (LRSC/GABA), respectively. Seven sst2a receptor-immunoreactive neurons (arrowheads) are present in this field, and all are also GABA-immunoreactive. Some cells which are not GABAimmunoreactive are indicated (asterisks), and these appear much darker than the surrounding neuropil. (b,d) A section scanned with the 488 nm line (FITC/sst2a receptor) and the 568 nm line (LRSC/glycine). Two sst2a-immunoreactive neurons are present, and both are also glycine-immunoreactive. Two of the cells which are not glycine-immunoreactive are indicated (asterisks). Scale bar=20 µm.

postembedding immunocytochemistry for GABA and glycine, as we have done for other receptors.22,26,49 As an alternative approach we used antisera raised against formaldehyde conjugates of these amino acids, and found that all sst2a receptor-immunoreactive neurons were GABAimmunoreactive, and most (83.5%) were glycineimmunoreactive. All glycine-immunoreactive neurons in laminae I and II are also GABAimmunoreactive,56 and the results of the present study therefore suggest that the sst2a receptor is restricted to inhibitory neurons in the superficial dorsal horn, and that it is particularly associated with those which use both GABA and glycine.56 Since we have previously shown that the NK1 receptor is almost exclusively present on neurons which are not GABA- or glycine-immunoreactive in laminae

I–III,26 this would explain why sst2a and NK1 receptors were not co-localized on neurons in this region. There is relatively little information concerning the functions of somatostatin within the dorsal horn, although the present results suggest that the peptide can influence a significant proportion of inhibitory neurons in the superficial laminae. Randic and Miletic41 found that somatostatin had an inhibitory action when applied to cells recorded in laminae I and II of the cat, however these effects might have been indirect, or mediated through another receptor, since they also found that cells in lamina V were inhibited by somatostatin, and these presumably did not express sst2a receptors. It has been reported that somatostatin applied intrathecally has an analgesic action,7 however its clinical usefulness is limited by apparent neurotoxicity.62 In order to have a better

NK1 and sst2a receptors in spinal cord

understanding of the role of somatostatin within the superficial dorsal horn, we will need to know more about the actions of somatostatin on the sst2a receptor, the circuitry involving inhibitory neurons that express the receptor, and whether or not other somatostatin receptors are present in the region. CONCLUSION

It is already known that somatostatin and substance P are contained in different populations of primary afferents14 and that their release into the superficial dorsal horn can be evoked by different

471

patterns of noxious stimulation: noxious mechanical, chemical and thermal cause increased substance P release,10,21,24 whereas only noxious thermal stimuli are adequate in the case of somatostatin.24,32 The results of the present study provide further evidence for a functional separation between the two peptides, by showing that they are likely to act on different neurons with the superficial laminae. Acknowledgements—We are grateful to Drs S. Vigna, P. C. Emson and D. V. Pow for generous gifts of antisera, to Drs D. J. Maxwell and S. A. S. Shehab for helpful discussion and to Mrs C. Watt for technical assistance. The work was supported by the Wellcome Trust.

REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

21. 22.

Abbadie C., Brown J. L., Mantyh P. W. and Basbaum A. I. (1996) Spinal cord substance P receptor immunoreactivity increases in both inflammatory and nerve injury models of persistent pain. Neuroscience 70, 201–209. Alvarez F. J. and Priestley J. V. (1990) Anatomy of somatostatin-immunoreactive fibres and cell bodies in the rat trigeminal subnucleus caudalis. Neuroscience 38, 343–357. Barber R. P., Vaughn J. E., Slemmon J. R., Salvaterra P. M., Roberts E. and Leeman S. E. (1979) The origin, distribution and synaptic relationships of substance P axons in rat spinal cord. J. comp. Neurol. 184, 331–352. Bleazard L., Hill R. G. and Morris R. (1994) The correlation between the distribution of NK1 receptor and the actions of tachykinin agonists in the dorsal horn of the rat indicates that substance P does not have a functional role on substantia gelatinosa (lamina II) neurons. J. Neurosci. 14, 7655–7664. Breder C. D., Yamada Y., Yasuda K., Seino S., Saper C. B. and Bell G. I. (1992) Differential expression of somatostatin receptor subtypes in brain. J. Neurosci. 12, 3920–3934. Brown J. L., Liu H., Maggio J. E., Vigna S. R., Mantyh P. W. and Basbaum A. I. (1995) Morphological characterization of substance P receptor-immunoreactive neurons in rat spinal cord and trigeminal nucleus caudalis. J. comp. Neurol. 356, 327–344. Chrubasik J., Meynadier J., Blond S., Scherpereel P., Ackerman E., Weinstock M., Bonath K., Cramer H. and Wunsch E. (1984) Somatostatin, a potent analgesic. Lancet ii, 1208–1209. Coggeshall R. E. (1992) A consideration of neuronal counting methods. Trends Neurosci. 15, 9–13. Ding Y.-Q., Takada M., Shigemoto R. and Mizuno R. (1995) Spinoparabrachial tract neurons showing substance P receptor-like immunoreactivity in the lumbar spinal cord of the rat. Brain Res. 674, 336–340. Duggan A. W., Hendry I. A., Morton C. R., Hutchison W. D. and Zhao Z. Q. (1988) Cutaneous stimuli releasing immunoreactive substance P in the dorsal horn of the cat. Brain Res. 451, 261–273. Gibson S. J., Polak J., Bloom S. R. and Wall P. D. (1981) The distribution of nine peptides in rat spinal cord with special emphasis on the substantia gelatinosa and on the area around the central canal (lamina X). J. comp. Neurol. 201, 65–80. Herbison A. E., Simonian S. X., Norris P. J. and Emson P. C. (1996) Relationship of neuronal nitric oxide immunoreactivity to GnRH neurons in the ovariectomized and intact female rat. J. Neuroendocrinol 8, 73–82. Ho¨kfelt T., Elde R., Johansson O., Luft R. and Arimura A. (1975) Immunohistochemical evidence for the presence of somatostatin, a powerful inhibitory peptide, in some primary sensory neurons. Neurosci. Lett. 1, 231–235. Ho¨kfelt T., Elde R., Johansson O., Luft R., Nilsson G. and Arimura A. (1976) Immunohistochemical evidence for separate populations of somatostatin-containing and substance P-containing primary afferent neurons in the rat. Neuroscience 1, 131–136. Ho¨kfelt T., Kellerth J. O., Nilsson G. and Pernow B. (1975) Substance P: localization in the central nervous system and in some primary sensory neurons. Science 190, 889–890. Ho¨kfelt T., Ljungdahl A., Terenius L., Elde R. and Nilsson G. (1977) Immunohistochemical analysis of peptide pathways possibly related to pain and analgesia: enkephalin and substance P. Proc. natn. Acad. Sci. U.S.A. 74, 3081–3085. Holloway S., Kidd E. J., Feniuk W. and Humphrey P. P. A. (1996) A quantitative autoradiographical study on the distribution of somatostatin sst2 receptors in the rat central nervous system. Neuropharmacology 35, 1121–1129. Hunt S. P., Kelly J. S., Emson P. C., Kimmel J. R., Miller R. J. and Wu J. Y. (1981) An immunohistochemical study of neuronal populations containing neuropeptides or ã-aminobutyrate within the superficial layers of the rat dorsal horn. Neuroscience 6, 1883–1898. Jansen A. S. P. and Loewy A. D. (1997) Neurons lying in the white matter of the upper cervical spinal cord project to the intermediolateral cell column. Neuroscience 77, 889–898. Johansson O., Ho¨kfelt T., Pernow B., Jeffcoate S. L., White N., Steinbusch H. W. M., Verhofstad A. A. J., Emson P. C. and Spindel E. (1981) Immunohistochemical support for three putative transmitters in one neuron: coexistence of 5-hydroxytryptamine, substance P- and thyrotropin releasing hormone-like immunoreactivity in medullary neurons projecting to the spinal cord. Neuroscience 6, 1857–1881. Kantner R. M., Kirby M. L. and Goldstein B. D. (1985) Increase in substance P in the dorsal horn during a chemogenic nociceptive stimulus. Brain Res. 338, 196–199. Kemp T., Spike R. C., Watt C. and Todd A. J. (1996) The µ-opioid receptor (MOR-1) is mainly restricted to neurons that do not contain GABA or glycine in the superficial dorsal horn of the rat spinal cord. Neuroscience 75, 1231–1238.

472 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57.

A. J. Todd et al. Kiyama H. and Emson P. C. (1990) Distribution of somatostatin mRNA in the rat nervous system as visualized by a novel non-radioactive in situ hybridization histochemistry procedure. Neuroscience 38, 223–244. Kuraishi Y., Hirota N., Sato Y., Hinoh Y., Satoh M. and Takagi H. (1985) Evidence that substance P and somatostatin transmit separate information related to pain in the spinal dorsal horn. Brain Res. 325, 294–298. Laing I., Todd A. J., Heizmann C. W. and Schmidt H. H. H. W. (1994) Subpopulations of GABAergic neurons in laminae I–III of rat spinal dorsal horn defined by coexistence with classical transmitters, peptides, nitric oxide synthase or parvalbumin. Neuroscience 61, 123–132. Littlewood N. K., Todd A. J., Spike R. C., Watt C. and Shehab S. A. S. (1995) The types of neuron in spinal dorsal horn which possess neurokinin-1 receptors. Neuroscience 66, 597–608. Liu H., Brown J. L., Jasmin L., Maggio J. E., Vigna S. R., Mantyh P. W. and Basbaum A. I. (1994) Synaptic relationship between substance P and the substance P receptor: light and electron microscopic characterization of the mismatch between neuropeptides and their receptors. Proc. natn. Acad. Sci. U.S.A. 91, 1009–1013. Mantyh P. W., DeMaster E., Malhotra A., Ghilardi J. R., Rogers S. D., Mantyh C. R., Liu H., Basbaum A. I., Vigna S. R., Maggio J. E. and Simone D. A. (1995) Receptor endocytosis and dendrite reshaping in spinal neurons after somatosensory stimulation. Science 268, 1629–1632. Marshall G. E., Shehab S. A. S., Spike R. C. and Todd A. J. (1996) Neurokinin-1 receptors on lumbar spinothalamic neurons in the rat. Neuroscience 72, 255–263. Millhorn D. E., Seroogy K., Ho¨kfelt T., Schmued L., Terenius L., Buchan A. and Brown J. C. (1987) Neurons in the ventral medulla oblongata that contain both somatostatin and enkephalin immunoreactivities project to nucleus tractus solitarii and spinal cord. Brain Res. 424, 99–108. Molander C., Xu Q. and Grant G. (1984) The cytoarchitectonic organization of the spinal cord in the rat. I. The lower thoracic and lumbosacral cord. J. comp. Neurol. 230, 133–141. Morton C. R., Hutchison W. D., Hendry I. A. and Duggan A. W. (1989) Somatostatin: evidence for a role in thermal nociception. Brain Res. 488, 89–96. Mullen R. J., Buck C. R. and Smith A. M. (1992) NeuN, a neuronal specific nuclear protein in vertebrates. Development 116, 201–211. Nagy J. I. and Hunt S. P. (1983) The termination of primary afferents within the rat dorsal horn: evidence for rearrangement following neonatal capsaicin treatment. J. comp. Neurol. 218, 145–158. Naim M., Spike R. C., Watt C., Shehab S. A. S. and Todd A. J. (1997) Cells in laminae III and IV of the rat spinal cord which possess the neurokinin-1 receptor and have dorsally-directed dendrites receive a major synaptic input from tachykinin-containing primary afferents. J. Neurosci. 17, 5536–5548. Nakaya Y., Kaneko T., Shigemoto R., Nakanishi S. and Mizuno N. (1994) Immunohistochemical localization of substance P receptor in the central nervous system of the adult rat. J. comp. Neurol. 347, 249–274. O’Rourke N. A., Sullivan D. P., Kaznowski C. E., Jacobs A. A. and McConnell S. K. (1995) Tangential migration of neurons in the developing cerebral cortex. Development 121, 2165–2176. Paxinos G. and Watson C. (1986) The Rat Brain in Stereotaxic Coordinates. 2nd edn. Academic, London. Pow D. V., Wright L. L. and Vaney D. I. (1995) The immunocytochemical detection of amino-acid neurotransmitters in paraformaldehyde-fixed tissues. J. Neurosci. Meth. 56, 115–123. Ralston H. J. (1979) The fine structure of laminae I, II and III of the macaque spinal cord. J. comp. Neurol. 184, 619–642. Randic M. and Miletic V. (1978) Depressant actions of methionine-enkephalin and somatostatin in cat dorsal horn neurons activated by noxious stimuli. Brain Res. 152, 196–202. Raynor K. and Reisine T. (1992) Somatostatin receptors. Crit. Rev. Neurobiol. 16, 273–288. Rexed B. (1952) The cytoarchitectonic organization of the spinal cord in the cat. J. comp. Neurol. 96, 415–496. Ribeiro-da-Silva A. and Cuello A. (1990) Ultrastructural evidence for the occurrence of two distinct somatostatincontaining systems in the substantia gelatinosa of rat spinal cord. J. chem. Neuroanat. 3, 141–153. Schindler M., Sellers L. A., Humphrey P. P. A. and Emson P. C. (1997) Immunohistochemical localization of the somatostatin sst2(A) receptor in the rat brain and spinal cord. Neuroscience 76, 225–240. Sen˜aris R. M., Humphrey P. P. A. and Emson P. C. (1994) Distribution of somatostatin receptors 1, 2 and 3 mRNA in rat brain and pituitary. Eur. J. Neurosci. 6, 1883–1896. Simmons D. R., Spike R. C. and Todd A. J. (1995) Galanin is contained in GABAergic neurons in the rat spinal dorsal horn. Neurosci. Lett. 187, 119–122. Snyder R. L. (1982) Light and electron microscopic autoradiographic study of the dorsal root projections to the cat dorsal horn. Neuroscience 7, 1417–1437. Spike R. C., Kerr R., Maxwell D. J. and Todd A. J. (1998) GluR1 and GluR2/3 subunits of the AMPA-type glutamate receptor are associated with particular types of neuron in laminae I–III of the spinal dorsal horn of the rat. Eur. J. Neurosci. 10, 324–333. Spike R. C., Todd A. J. and Johnston H. M. (1993) The coexistence of NADPH diaphorase with GABA, glycine and acetylcholine in rat spinal cord. J. comp. Neurol. 335, 320–333. Sprinkle T. J. (1989) 2 ,3 -cyclic nucleotide 3 -phosphodiesterase, an oligodendrocyte-Schwann cell and myelinassociated enzyme of the nervous system. Crit. Rev. Neurobiol. 4, 235–301. Todd A. J. (1991) Immunohistochemical evidence that acetylcholine and glycine exist in different populations of GABAergic neurones in lamina III of rat spinal dorsal horn. Neuroscience 44, 741–746. Todd A. J. (1997) A method for combining confocal and electron microscopic examination of sections processed for double- or triple-labelling immunocytochemistry. J. Neurosci. Meth. 73, 149–157. Todd A. J. and McKenzie J. (1989) GABA-immunoreactive neurones in the dorsal horn of the rat spinal cord. Neuroscience 31, 799–806. Todd A. J. and Spike R. C. (1993) The localization of classical transmitters and neuropeptides within neurons in laminae I–III of the mammalian spinal dorsal horn. Prog. Neurobiol. 41, 609–646. Todd A. J. and Sullivan A. C. (1990) A light microscope study of the coexistence of GABA-like and glycine-like immunoreactivities in the spinal cord of the rat. J. comp. Neurol. 296, 496–505. Uhl G. R., Tran V., Snyder S. H. and Martin J. B. (1985) Somatostatin receptors: distribution in rat central nervous system and human frontal cortex. J. comp. Neurol. 240, 288–304.

NK1 and sst2a receptors in spinal cord 58.

473

Valtschanoff J. G., Weinberg R. J. and Rustioni A. (1992) NADPH-diaphorase in the spinal cord of rats. J. comp. Neurol. 321, 209–222. 59. Vanetti M., Kouba M., Vogt G. and Ho¨llt V. (1992) Cloning and expression of a novel mouse somatostatin receptor (SSTR2B). Fedn Eur. biochem. Socs Lett. 311, 290–294. 60. Vigna S. R., Bowden J. J., McDonald D. M., Fisher J., Okamoto A., McVey D. C., Payan D. G. and Bunnett N. W. (1994) Characterization of antibodies to the rat substance P (NK-1) receptor and to a chimeric substance P receptor expressed in mammalian cells. J. Neurosci. 14, 834–845. 61. Wolf H. K., Buslei R., Schmidtkastner R., Schmidtkastner P. K., Pietsch T., Wiestler O. D. and Blumcke I. (1996) NeuN—a useful neuronal marker for diagnostic histopathology. J. Histochem. Cytochem. 44, 1167–1171. 62. Yaksh T. L. (1994) Spinal somatostatin for patients with cancer. Risk-benefit assessment of an analgesic. Anesthesiology 81, 531–533. (Accepted 9 December 1997)