Brain, Vol. 123, No. 3, 601-619,
March 2000
© 2000 Oxford University Press
Cytoarchitectonic and immunohistochemical characterization of a specific pain and temperature relay, the posterior portion of the ventral medial nucleus, in the human thalamus
1 Division of Cell Biology, Department of Biomedicine and Surgery, Faculty of Health Sciences, University of Linköping, Sweden and 2 Division of Neurosurgery, Barrow Neurological Institute, Phoenix, USA
Correspondence to:
A. Blomqvist, Division of Cell Biology, Department of Biomedicine and Surgery, Faculty of Health Sciences, University of Linköping, S-581 85 Linköping, Sweden E-mail: andbl{at}mcb.liu.se
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Previous studies in the macaque monkey have identified a thalamic nucleus, the posterior portion of the ventral medial nucleus (VMpo), as a dedicated lamina I spinothalamocortical relay for pain and temperature sensation. The dense plexus of calbindin-immunoreactive fibres that characterizes VMpo in primates enables its homologue to be identified in the human thalamus by immunohistochemical labelling for calbindin. We have now analysed in detail the cytoarchitectonic characteristics of VMpo and its relationship with immunoreactivity for calbindin, substance P and calcitonin gene-related peptide (CGRP) in the human thalamus. The area in the posterolateral thalamus in which dense calbindin-immunoreactive fibre terminations are present coincides nearly completely with a distinct region that contains small to medium-sized cells with round or oval shapes that are aggregated in clusters separated by cell sparse areas. This region, which we identify as VMpo, is located posteromedial to the ventral posterior lateral (VPL) and ventral posterior medial (VPM) nuclei, ventral to the anterior pulvinar and centre médian nuclei, lateral to the limitans and parafascicular nuclei and dorsal to the medial geniculate nucleus. Calbindin-immunoreactive fibres enter VMpo from the spinal lemniscus and form large patches of dense terminal-like staining over clusters of VMpo neurons. A few of these clusters also display terminal-like substance P labelling. Small bursts of CGRP staining are intercalated between the calbindin-labelled clusters, but there is little or no overlap between these two markers. CGRP immunoreactivity is also present over small, non-clustered neurons in the calbindin-negative area that separates VMpo from the VPL and VPM nuclei, which we denote as the posterior nucleus (Po). These observations provide a concise description of VMpo in the human thalamus. Further, they suggest that the lamina I spinothalamic tract fibres (represented by calbindin and probably also substance P immunoreactivity) and vagal-solitary-parabrachial afferents (represented by CGRP immunoreactivity) form closely related, but separate, termination fields that can be considered to represent different aspects of enteroceptive information regarding the physiological status of the tissues and organs of the body. The location of VMpo and the adjacent Po fits with clinical descriptions of the thalamic area from which pain, temperature and visceral sensations can be evoked by microstimulation, and where nociceptive and thermoreceptive neurons have been recorded in humans. It also corresponds to the area in which infarcts cause analgesia and thermoanaesthesia and can lead to the paradoxical development of central pain.
spinothalamic tract; posterior complex; calbindin; calcitonin gene-related peptide; substance P
CeM = central medial nucleus; CGRP = calcitonin gene-related peptide; CL = central lateral nucleus; CM = centre médian nucleus; eml = external medullary lamina; H = habenular nucleus; Hl = lateral habenular nucleus; Hm = medial habenular nucleus; LD = lateral dorsal nucleus; LG = lateral geniculate nucleus; Li = limitans nucleus; LP = lateral posterior nucleus; mc = magnocellular portion of medial geniculate nucleus; MD = mediodorsal nucleus; MDvc = ventrocaudal region of the mediodorsal nucleus; MG = medial geniculate nucleus; Pc = paracentral nucleus; pc = posterior commissure; Pf = parafascicular nucleus; Pg = perigeniculate nucleus; Pla = anterior pulvinar nucleus; Pll = lateral pulvinar nucleus; Plm = medial pulvinar nucleus; Po = posterior nucleus; R = reticular thalamic nucleus; rf = retroflex bundle; Sg = suprageniculate nucleus; SN = substantia nigra; Sub = subthalamic nucleus; V.c. = nucleus ventrocaudalis; VLa = anterior ventral lateral nucleus; VLp = posterior ventral lateral nucleus; VM = ventral medial nucleus; VMb = basal portion of ventral medial nucleus; VMpo = posterior portion of ventral medial nucleus; VP = ventroposterior nuclei; VPI = ventral posterior inferior nucleus; VPL = ventral posterior lateral nucleus; VPLa = anterior ventral posterior lateral nucleus; VPLm = medial ventral posterior lateral nucleus; VPLp = posterior ventral posterior lateral nucleus; VPM = ventral posterior medial nucleus; VPMpc = parvocellular VPM; ZI = zona incerta
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
In contrast to most other sensory modalities, the neuroanatomical substrates for pain and temperature sensation in the forebrain have only recently begun to be elucidated. Major advances in this field have come through functional anatomical studies in non-human primates, which have identified substrates that underlie findings from functional imaging and microelectrode studies in humans. In particular, studies in the monkey have traced the central connections of spinal cord neurons that respond specifically to natural thermal and noxious stimuli and have shown that their thalamic and cortical projections correspond with brain structures activated in humans experiencing pain and temperature sensations. These studies have demonstrated the presence in the posterior thalamus of a relay nucleus specific for pain and temperature sensation (Craig et al., 1994
). This nucleus, the posterior portion of the ventral medial nucleus (VMpo), receives via the spinal and trigeminal lemniscus a dense, topographically organized input from nociceptive- and thermoreceptive-specific neurons in the most superficial portion (lamina I) of the medullary and spinal dorsal horn (Craig et al., 1994, 1999; Blomqvist et al., 1996; Zhang and Craig, 1997a). Together with two other lamina I spinothalamic termination sites, the ventrocaudal region of the mediodorsal nucleus (MDvc) and the ventral posterior inferior nucleus (VPI) (Craig et al., 1994), this pathway provides input to the four main cerebral cortical regions (Friedman and Murray, 1986; Craig, 1995; Craig and Zhang, 1996) that, in human imaging studies, show activation with noxious and thermal stimuli (Jones et al., 1991; Talbot et al., 1991; Casey et al., 1994; Craig et al., 1996).
The correspondence between these functional anatomical studies in experimental primates and the findings from human imaging studies is consistent with accumulating evidence that strongly supports the concept that the lamina I spinothalamic projection system is critical for pain sensation. Lamina I spinothalamic tract neurons are selectively responsive to stimuli that cause pain in humans, such as noxious heat and noxious cold and noxious pinch, as well as the thermal grill illusion of pain (Christenson and Perl, 1970; Craig and Kniffki, 1985; Craig and Bushnell, 1994; Dostrovsky and Craig, 1996a; Han et al., 1998). Their responses are attenuated by morphine in analgesic doses (Craig and Serrano, 1994), in accordance with clinical observations. Their axons ascend contralaterally in the middle of the lateral funiculus, i.e. in the lateral spinothalamic tract, which is critical for pain and temperature sensation (May, 1906; Kuru, 1949; Nathan and Smith, 1969; Craig, 1991; Ralston and Ralston, 1992). Recent studies in rats have shown that selective, partial destruction of lamina I nociceptive cells (by internalization of a conjugate of substance P and the ribosome-inactivating protein saporin) produces marked attenuation of behavioural responses to tissue-damaging stimuli while sparing the responses to mild noxious stimuli (Mantyh et al., 1997). Thus, the lamina I projection system transmits information that is relevant to clinical pain, and the termination sites of this pathway in the human brain need to be studied.
The present report deals with the characterization of the lamina I termination field in the human homologue of the macaque VMpo nucleus. In the monkey, VMpo is characterized by dense fibre- and terminal-like calbindin immunoreactivity (Craig et al., 1994). Double-labelling experiments have shown co-localization of calbindin immunoreactivity with anterogradely transported tracer substance in lamina I terminals in VMpo (Craig et al., 1994), suggesting that calbindin is a useful marker for lamina I fibres. Consistent with this, many lamina I spinothalamic tract cell bodies are calbindin-immunoreactive (Zhang and Craig, 1997), and a calbindin-positive fibre bundle is present in the middle of the lateral spinal funiculus in primates and humans (Craig et al., 1998), where lamina I fibres ascend (Craig, 1991).
Prior immunohistochemical staining of the human thalamus demonstrated dense calbindin labelling in a position homologous to VMpo in the macaque thalamus (Craig et al., 1994). This location coincides with the clinically described region where microstimulation can evoke discrete pain and temperature sensations and where neurons that respond to pain and temperature stimuli have been recorded (Lenz et al., 1993a, b; Davis et al., 1996, 1999). In the present study we extend our previous observations on the human VMpo by describing in detail its cytoarchitectonic characteristics, in order to provide an anatomical basis for the clinical work on pain processing in this part of the thalamus. To further elucidate the functional organization of the posterolateral thalamus, we also analyse how the calbindin-labelled area relates to the distribution of immunoreactivity for substance P and calcitonin gene-related peptide (CGRP), two neuromodulators associated with pain and visceral sensory processing, respectively. A preliminary account has been given in abstract form (Blomqvist et al., 1998).
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The present findings are based on the analysis of 10 human thalami, obtained from normal autopsy material with appropriate permissions from the authorities at the University Hospital of Linköping and in accordance with Swedish law. The subjects were of both sexes, and aged between 50 and 86 years. They had all died from non-neurological causes.
The brains were removed between 1 and 3 days after death. The diencephalon was dissected free and divided along the midline. Either the left or the right side was taken. It was blocked so that the dorsal and ventral surfaces were cut parallel to a plane through the anterior and posterior commissures (the horizontal plane); the anterior surface was cut perpendicular to the horizontal plane (the frontal plane) or in a plane perpendicular to the long axis of the cerebrum (the transverse plane); and the lateral surface was cut in the sagittal plane. The blocks were immersed for 2 weeks in a fixative containing 4% paraformaldehyde in 0.1 M phosphate buffer, and then for 1 week in fixative to which 30% sucrose had been added. They were rinsed in a phosphate buffer-sucrose solution for 1 day before sectioning.
The thalami were cut on a freezing microtome at 50 µm. Two thalami were cut in the horizontal plane, four were cut in the frontal plane, one was cut in the transverse plane and three were cut in the sagittal plane. Sections were collected in three or four sets: one set was used for thionin staining and one for calbindin immunohistochemistry. Set 3 and set 4 (when available) were stained for substance P and/or CGRP immunoreactivity, or left in reserve.
Before immunohistochemistry, sections were treated to reduce endogenous peroxidase activity. They were immersed in graded concentrations of methanol (40%, 60%, 80%) in PBS (phosphate-buffered saline) and in 100% methanol to which had been added 1% H2O2; each step lasted 20 min. Following rehydration in decreasing concentrations of methanol, sections were rinsed in PBS and placed in primary antibody solution at 4°C for 3 days during constant agitation. Monoclonal mouse anti-calbindin-D antiserum was obtained from Sigma (St Louis Mo., USA, clone CL-300), and was diluted 1 : 2000; polyclonal rabbit anti-substance P antiserum was from Incstar (Stillwater, Minn., USA), and was diluted 1 : 6000; polyclonal rabbit anti-CGRP antiserum was a gift from Dr Lars Terenius (Arvidsson et al., 1989), and was diluted 1 : 40 000. The vehicle consisted of PBS with 0.2% Triton X-100, 0.2% bovine serum albumin and 3.3% normal serum. Controls included preadsorption of the primary antibody with the appropriate antigen, when available, or omission of the primary antibody. Bound primary antibody was detected with the peroxidase–antiperoxidase (Sternberger, 1979) or ABC (avidin–biotin–peroxidase complex) (Hsu et al., 1981) methods. Secondary antibodies and PAP (peroxidase–horseradish peroxidase) complexes were obtained from Dakopatts (Älvsjö, Sweden), and biotinylated antibodies and avidin–HRP (horseradish peroxidase) were obtained from Vector (Burlingame, Calif., USA), and each was used according to the manufacturer's instructions. After processing with 3,3'-diaminobenzidine tetrahydrochloride (0.075%) and H2O2 (0.01%) for 3–5 min, sections were rinsed, mounted on gelatinized slides, defatted in a chloroform/ethanol mixture, dehydrated, cleared and coverslipped with DPX (BDH Laboratory Supplies, Poole, UK).
Sections were analysed in a Wild M3C dissecting microscope equipped with a camera lucida. Nuclear groups were delineated and named as suggested by Hirai and Jones (1989b). The localization of immunolabelling in relation to cytoarchitectonically identified structures was determined by viewing sections superimposed on each other, using capillaries for alignment. Low-power micrographs were taken with a Nikon Multiphot macrophotography stand and 4 x 5-inch black-and-white Kodak 4125 Professional Copy Film. Medium- and high-power digital images were made with a Leaf Microlumina scanner (3380x2253 pixels) and x1, x2 and x20 planapo objectives, and Adobe Photoshop was used to enhance contrast, colour-code and apply labels.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
We identified VMpo in the human thalamus first on the basis of a dense plexus of immunoreactive fibres and terminals labelled with a monoclonal antibody for 28K calbindin (Sigma), because such labelling characterized VMpo in the macaque monkey. This enabled us to recognize the characteristics of the human VMpo as a distinct cytoarchitectonic region that is distinguishable from its neighbours in the posterolateral thalamus. In the following, we present first an overview of the neighbourhood relationships of VMpo in the human thalamus as seen in low-power views of thionin- and calbindin-stained sections in the frontal, sagittal and horizontal planes (Figs 1–4
). Then we document the details of these cytoarchitectonic relationships with mid- and high-power photomicrographs (Figs 5 and 6). Finally, we describe the relationship of the calbindin labelling in VMpo to the labelling pattern for the putative neurotransmitters substance P and CGRP (Fig. 7). The illustrations are taken from five of the 10 cases examined; although a few of the cases display some autolytic changes, they provided supporting immunohistochemical observations.
|
|
|
|
|
|
|
Overview
The photomicrographs and the drawings in Figs 1 and 2
, respectively, illustrate the location of the human VMpo in frontal sections. In the frontal plane, VMpo is an oval structure that extends ~3–3.5 mm mediolaterally and 2 mm dorsoventrally. It is posteromedial to the ventral posterior lateral (VPL) and ventral posterior medial (VPM) nuclei, ventral to the anterior pulvinar (Pla) and centre médian (CM) nuclei, lateral to the limitans (Li) and parafascicular (Pf) nuclei and dorsal to the medial geniculate nucleus (MG). A narrow region separates VMpo from VPM/VPL and the basal portion of the ventral medial nucleus (VMb), which we designate as the posterior nucleus (Po; more detail below). At its caudal pole, VMpo is intermixed with cell groups that belong to the suprageniculate nucleus (Sg; for more detail, see below).
The series of drawings in Fig. 3 show that in the sagittal plane the VMpo is a rounded structure measuring ~2 x 2 cm, located along the ventroposterior aspect of Pla and CM, with Sg and the medial pulvinar nucleus (Plm) situated caudally and caudodorsally, respectively. Caudoventrally the VMpo adjoins MG. The VMb is anterior to the VMpo. Note, however, that the VMpo, at its anterolateral aspect, is adjacent to VPM/VPL. Ventral to VMpo is the external medullary lamina (see also Fig. 6A and B).
The series of drawings of horizontal sections in Fig. 4 further demonstrate the relationship between VMpo and VPL, VPM, VMb, and the ventral posterior inferior (VPI) nuclei. In this plane, VMpo is egg-shaped, and it extends ~3.5 mm mediolaterally and 2 mm anteroposteriorly. Anterior to VMpo are, from medial to lateral, the CM, VMb/VPM, and VPI/VPL nuclei. Medial to VMpo are Sg and Li, and posterior are Sg, Plm and MG.
Calbindin labelling
The VMpo is characterized by the dense fibre- and terminal-like immunoreactivity for 28K calbindin (Figs 1 and 5) that is obtained with the Sigma monoclonal antibody. This contrasts with the calbindin negativity in VPM/VPL, in the lateral and medial geniculate nuclei and in CM that is seen when this antibody is used. [This pattern differs from that obtained (Rausell and Jones, 1991b; Morel et al., 1997) with the monoclonal anti-calbindin antibody of P. C. Emson (see below).] Although VMpo is the most conspicuous calbindin-immunoreactive structure seen in the human thalamus when stained with this antibody (Fig. 1), other regions also express calbindin immunoreactivity. For example, portions of the central lateral nucleus are also densely labelled (Fig. 1), and small patches of calbindin immunoreactivity are present in VPI ventral to VPM/VPL, and also along the border between VPM and Pla (Fig. 1D). In sagittal and horizontal sections, particularly, the strong terminal-like calbindin staining in VMpo (Fig. 5D and F) contrasts with the weaker labelling found in parts of VMb. In Po, the narrow region that partially surrounds VMpo, there is no or sparse calbindin immunoreactivity.
Comparison of adjacent thionin-stained and immunostained sections reveals that the patches of dense calbindin staining within VMpo are situated over large clusters of neurons (Figs 1 and 5A and B). The calbindin labelling in VMpo originates from calbindin-positive fibres that enter VMpo from the spinal lemniscus in the midbrain. The photomicrograph in Fig. 6B (from a case other than that shown in Fig. 1) clearly demonstrates the labelled ascending fibre bundle that is the source of the calbindin labelling in VMpo. In this particular section, the calbindin labelling in VMpo has a sac-like appearance, with an opening towards the midbrain through which labelled fibres enter. In the accompanying thionin-stained section (Fig. 6A), which is separated from the calbindin-stained section by 250 µm, it can be seen that the opening of VMpo towards the midbrain is located in a gap formed between Sg cells dorsally and the MG ventrally. This photomicrograph also shows a thick, calbindin-negative, mediolaterally oriented fibre bundle ventral to VMpo, which is the medial extension of the external medullary lamina that contains medial and spinal lemniscal fibres directed towards the VPM/VPL complex. In the thionin-stained section adjacent to the calbindin-stained section in Fig. 6B (not shown due to cutting damage), this fibre bundle can be followed between VMpo and MG towards the bottom of the VPL nucleus.
Although dense calbindin fibre labelling is characteristic of VMpo, the area of dense calbindin labelling does not completely outline VMpo, as determined on the basis of its cytoarchitectonic characteristics (see below). First, the cell-sparse areas between the VMpo cell clusters display no or faint calbindin immunoreactivity. Secondly, calbindin-negative VMpo cell clusters also exist that do not coincide with calbindin fibre staining. This can be seen in Fig. 5E and F, where only the medial part of VMpo displays dense labelling (although this lateral portion is labelled in more dorsal sections). The frontal sections in Fig. 5A and B also show several VMpo cell clusters that are only weakly labelled or are unlabelled. These observations are consonant with the finding that many, but not all, lamina I spinothalamic tract (STT) cells are calbindin-positive (Zhang and Craig, 1997b), and they suggest the presence of functional subdivisions within VMpo, consistent with the immunohistochemical organization described below.
Nonetheless, the details of the calbindin labelling and their correspondence with the distinguishable cytoarchitectonic characteristics enable recognition of VMpo. For example, within the ambiguous region of the most posterior thalamus, the terminal-like calbindin labelling in VMpo differentiates it from the adjacent nuclei. The adjacent Li and Sg also contain calbindin labelling; however, in these nuclei, and in most other calbindin-positive areas except VMpo, it is primarily the cell bodies, and few fibres, that are immunolabelled. This can be seen in the horizontal sections shown in Fig. 5E and F, and it is demonstrated clearly in the pair of frontal sections shown in Fig. 6C and D (from a case other than those shown in Figs 1 and 6A and B, and cut at a slightly different angle). In these frontal sections, two clusters of VMpo cells that coincide with dense terminal-like calbindin labelling are surrounded by groups of larger, more distinctly thionin-stained Li and Sg neurons, cells which are immunoreactive for calbindin, as illustrated in the high-power photomicrographs in Fig. 6E and F. The MG nucleus at the ventrolateral aspect of VMpo is calbindin-negative. Accordingly, the only profile in the area of the posterior region of the thalamus that displays dense terminal-like labelling for calbindin is VMpo; thus, such labelling is a characteristic and distinguishable feature of this nucleus.
Cytoarchitectonic characteristics
In the frontal plane, VMpo displays small to medium-sized, round or oval cells (20–30 x 20 µm) that are nearly all of average staining intensity and that are oriented mediolaterally. They are conspicuously aggregated in three to five large clusters, which are separated by cell sparse areas (Figs 5A and G and 7A). The lobulated appearance of VMpo and its uniformly sized cells differentiate it from the larger, darkly stained, heterogeneous and clumped cells of VPM and VPL (Fig. 5H), and make it easily distinguishable from the small, lightly stained but densely packed cells of the Pla and CM and from the more darkly stained but densely packed cells in Pf. Similarly, the clustered organization, smaller size and average staining density of the VMpo cells separate them from the densely packed, darkly stained cells of Sg and the larger, angular, and intensely stained cells of Li (Figs 5A and 6A, C and E).
In sagittal sections, close examination reveals that VMpo in this plane appears as a rounded, pale region (Fig. 5C) that contains small to medium-sized, mostly round neurons (20 x 20 µm) that are loosely collected in groups. It is easily distinguishable from the VPM and from the VMb with its lightly stained but rather densely packed cells. The pale staining of VMpo in this plane of section is attributable to the fact that its cells are cut perpendicular to their long axis. This makes it stand out as a circumscribed area when examined at low power. [It can be seen as a pale, round structure just posterior to VMb in the series of low-power micrographs of sagittal sections shown in Jones' recent description of the human thalamus (see Fig. 9.4F on p. 449 in Jones, 1997).]
In horizontal sections, VMpo displays mediolaterally oriented neurons (25 x 15 µm) in a lobulated, rather coherent, egg-shaped structure (Fig. 5E). A feature which is best appreciated in the horizontal sections (Fig. 4) is the zone of separation between VMpo and the adjacent nuclei CM, VMb, VPM, VPI and VPL, which we denote as the posterior nucleus (Po). This zone is continuous with a larger area of mostly small, fusiform, pale-staining neurons (Fig. 5E) located lateral to VMpo and anterior to the MG. It is seen as a zone separating VMpo from VPM/VPL also in the frontal sections (Figs 2, 5A and 7A) and as a gap between VMpo and VMb in the sagittal sections (Fig. 3). Thus, the Po is a curved structure that surrounds the anterior, lateral and ventral aspects of the VMpo. Its anterior portion is bordered laterally by the VPL and its posterior portion by the caudal end of the thalamic reticular nucleus. Ventrolaterally, it merges with VPI, which contains cells similar to those in Po, but also scattered larger, more darkly stained neurons. Dorsally, Po is limited by Plm and Pla and ventrally by the external medullary lamina. It is distinguished also by the immunohistochemical labelling described below.
Immunohistochemical organization
To further characterize the VMpo region, we stained for two peptides, substance P and CGRP, that are found in the posterior thalamic region of other species and that may represent input from two different sources, the spinal dorsal horn (Battaglia and Rustioni, 1992) and the parabrachial nucleus (Yasui et al., 1989), respectively. Substance P immunoreactivity suggestive of terminal labelling is present in small patches within VMpo. Substance P-labelled fibres enter VMpo from the spinal lemniscus along the same route as the calbindin-positive fibres. A few patches of terminal-like substance P staining are located in Po and in VPI, whereas there is no substance P labelling in VMb. The substance P staining in VMpo generally overlaps with the calbindin staining, although substance P-labelled clusters that are not associated with calbindin staining are also present. However, the substance P staining is much sparser than the calbindin staining and occupies only a small area of VMpo, as demonstrated in the series of adjacent frontal sections shown in Fig. 7A–D.
The relationship in the posterior thalamus between immunoreactivity for CGRP and the calbindin and substance P staining in VMpo is also shown in Fig. 7. The CGRP labelling is found in rather densely labelled patches in Po, separating VMpo from VPM/VPL. This can be seen clearly in both frontal and in horizontal sections. A few patches of CGRP staining are also present in VPI, but only sparse CGRP staining is present in VMb. CGRP immunoreactivity is also present within VMpo; however, it never overlaps the areas of dense calbindin labelling. Rather, it is intercalated between these areas, separating the VMpo cell clusters. The details of these relationships are shown in the two colour composites shown in Fig. 7. In Fig. 7E, the calbindin and CGRP staining in adjacent horizontal sections have been superimposed, and in Fig. 7F the calbindin, CGRP and substance P staining in adjacent frontal sections are displayed together.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
By comparing immunohistochemical staining by a monoclonal antibody for calbindin that serves as a marker for lamina I spinothalamic projections with cytoarchitectonic criteria, we have identified a nucleus in the human thalamus, the VMpo, that can be regarded as a specific relay for pain and temperature sensations. We have described and documented the cytoarchitectonic characteristics and neighbourhood relationships of VMpo, and we have shown that part of the VMpo nucleus contains substance P-immunoreactive fibres and terminals, whereas CGRP immunoreactivity forms a closely related but distinct terminal field that is intercalated between VMpo and the VPM/VPL and VMb nuclei.
Comparison with prior anatomical studies on the human thalamus
The denotation of the pain and temperature relay as VMpo reflects its close relationship to the VMb, the vagal–solitary–parabrachial relay, and their parallel cortical projections to the insular cortex (see below). The term VMpo has not been used before, but it adheres to the traditional system of nomenclature, as summarized for the human thalamus by Hirai and Jones (1989b). The region of VMpo was included in their suprageniculate/posterior complex (Sg/Po); similarly, in the recent human thalamic atlas by Morel and colleagues (Morel et al., 1997), the VMpo also seems to be included in the so-called posterior complex. In the atlas by Hassler, which is used in many clinical studies (Hassler, 1982), the VMpo seems to correspond to the limitans portae and the adjacent part of the V.c. [nucleus ventrocaudalis] portae (see also discussion by Mehler, 1966), denotations that literally describe the location of the opening against the midbrain for ascending spinal and lemniscal fibres, where VMpo is situated (Fig. 6A and B). However, the regions designated limitans portae and V.c. portae by Hassler also include other structures (cf. Hirai and Jones, 1989b), and in both the macaque and the human brain VMpo can be distinguished cytoarchitectonically and immunohistochemically as a separate entity within the posterior region (Craig et al., 1994; present study).
Although the VMpo has not been distinguished by others, several previous tract-tracing studies in monkeys have recognized the dense spinothalamic input to the Sg/Po region (Mehler et al., 1960; Boivie, 1979; Berkley, 1980; Burton and Craig, 1983; Ralston and Ralston, 1992; Rausell et al., 1992). Further, it has been demonstrated that the Sg/Po region is the main target for the spinothalamic tract fibres that ascend in the middle of the lateral funiculus in the monkey (Ralston and Ralston, 1992), i.e. the spinothalamic fibres that originate from neurons in lamina I of the superficial dorsal horn (Craig, 1991; Craig et al., 1998). Of great interest and salience are the degeneration studies on the spinothalamic tract in the human thalamus by W. R. Mehler, who in one particular article (Mehler, 1966) provided a focused description of a dense termination in a distinguishable portion of the posterior region, which he referred to as the `caudal V.c.'. His sagittal drawing in that article shows a dense spinothalamic termination field located in an area that matches the calbindin-stained VMpo nucleus as described in the present study. The observations by Mehler are consonant with the idea that the calbindin staining in the VMpo represents a lamina I spinothalamic input, viz. calbindin-immunoreactive fibres could be followed from the spinal lemniscus into the VMpo (Fig. 6B); many lamina I cells, both in the human and the macaque spinal cord, are calbindin-immunoreactive (Zhang and Craig, 1997b; Craig et al., 1998); calbindin-immunoreactive lamina I spinothalamic tract (STT) cells have been demonstrated in the macaque (Zhang and Craig, 1997b; Craig et al., 1998); and a calbindin-immunoreactive fibre bundle is present in the middle part of the lateral funiculus of the spinal cord in both humans and macaques, at a position corresponding to that of ascending anterogradely labelled lamina I STT axons in the lateral spinothalamic tract (Craig et al., 1998). [Note that others erroneously suggested that lamina I spinothalamic tract fibres travel in the dorsolateral funiculus and would not be involved in the analgesia and thermanaesthesia produced by cordotomies (Apkarian et al., 1985; Apkarian and Hodge, 1989, based on observations of retrograde labelling following lesions of the dorsolateral or ventrolateral spinal quadrants).] Their analysis did not discriminate the middle of the lateral funiculus, which is the location of the ascending lamina I fibres as shown directly with anterograde tracing from concise injections in the superficial dorsal horn (Craig, 1991) and with spinal lesions in behaving animals (Norrsell, 1979). In primates, which have a large corticospinal tract, lamina I STT fibres are located just ventral to the dentate ligament, which is the location of the calbindin-immunoreactive fibre bundle in the monkey and human spinal cord (Craig et al., 1998) and the location of efficacious cordotomy lesions that cause hypalgesia and thermoanaesthesia in humans (e.g. Nathan and Smith, 1979; for review, see Craig and Dostrovsky, 1999).] Taken together with the direct observation of double-labelled calbindin-immunoreactive and PHA-L-positive lamina I terminals in VMpo of the macaque monkey and their co-extensive distribution (Craig et al., 1994), these observations indicate that VMpo is the main terminal site of lamina I spinothalamic tract fibres also in the human brain.
Comparison with physiological studies in human thalamus
The location of VMpo in relation to the ventral posterior nucleus (VPM and VPL) and its stereotaxic coordinates [A, -0.5 to 2.0; L, 12.0 to 16.0; H, 1.0 to +1.0 (Craig et al., 1994)] fit with clinical descriptions of the thalamic area from which pain and temperature sensations can be evoked and in which nociceptive- and thermoreceptive-specific neurons have been recorded in humans. This is consistent with the characteristics of VMpo in the macaque (Craig et al., 1994). Thus, in patients suffering from movement disorders, microstimulation of the thalamic region ventroposterior to the main tactile relay nucleus [nucleus ventrocaudalis (V.c.; Hassler and Riechert, 1959; Hassler, 1982)] has been found to elicit pain and/or temperature sensations (Lenz et al., 1993a, b; Davis et al., 1996, 1999), and the same area, referred to as the posteroinferior region of V.c. in one laboratory (Lenz et al., 1993b), has been shown to contain neurons that appear to be nociceptive-specific (Lenz et al., 1993a) or thermoreceptive-specific (Dostrovsky et al., 1996; Davis et al., 1999). In contrast, electrical stimulation of the V.c. proper seldom elicits pain or thermal sensations in non-pain patients (Lenz et al., 1993a, b; Davis et al., 1996), and although nociceptive responses have been recorded in V.c. proper, it does not appear to contain nociceptive-specific cells (Lenz et al., 1993a). In the above-mentioned studies the electrodes were advanced from anterodorsal to posteroventral; pain and temperature sensations were evoked when the trajectories passed beyond the face region of V.c. but more rarely when they passed beyond the hand or leg region, suggesting that the stimulation sites where pain and thermal sensations are evoked are located behind the medial aspect of the cutaneous core of V.c. This is consistent with the location of VMpo posterior and ventral to the VPM nucleus, as demonstrated by the present study.
In a recent physiological study in humans, Davis and colleagues identified specific cold-sensitive neurons in the human thalamus, and they showed that microstimulation at these sites in awake humans induces localized cold sensations that grade with stimulus intensity (Davis et al., 1999). They reported that these neurons are located ventroposteriorly and medially to the V.c. nucleus, and they concluded that this location coincides with that of the human VMpo described in our initial report (Craig et al., 1994). This is consistent with the functional anatomical identification of thermoreceptive (cold)-specific neurons in the macaque VMpo (Craig et al., 1994), and also with the observations that lamina I spinothalamic neurons that project to VMpo (Dostrovsky and Craig, 1996a) constitute the only identified cold-sensitive spinal thermosensory-specific projection, and that innocuous cold produces functional (PET imaging) activation in the insular cortex, where VMpo projects (Craig, 1995), but not in the primary somatosensory cortex (Craig et al., 1996).
Comparison with lesion studies in humans
The location of the VMpo is also consistent with the general region in the human thalamus (i.e. its ventral posterolateral part) in which infarcts or neurosurgical lesions can produce analgesia and thermanaesthesia (Head and Holmes, 1911; Davison and Schick, 1935; Hassler and Riechert, 1959; Bettag and Yoshida, 1960; Mark et al., 1960, 1963; Leijon et al., 1989; Bogousslavsky et al., 1988; Bowsher et al., 1998), although no lesion has been described that was restricted to the area of VMpo. Further, neurosurgical lesions that were targeted at the highly visible CM nucleus because it was once thought to be a target of spinothalamic fibres (Bowsher, 1957) were reported in some cases to relieve intractable pain successfully without loss of innocuous tactile sensibility (Mark et al., 1963; Sano, 1977), and it is possible that such lesions could have involved the adjacent VMpo nucleus.
Studies in the macaque monkey have shown that the VMpo projects to a distinct region in the dorsal margin of the mid/posterior insula, directly posterior to the gustatory cortex that receives input from VMb (Craig, 1995). Nociceptive neurons have been recorded in this portion of the insular cortex in the macaque (Dostrovsky and Craig, 1996b). The correspondence of this pathway with pain and temperature sensation in humans is further supported by human imaging studies, which show a strong activation of the contralateral insula with noxious or thermal stimulation (Jones et al., 1991; Casey et al., 1994; Coghill et al., 1994; Craig et al., 1996). Similarly, clinical lesions in the region of insular cortex reportedly result in reduced pain and temperature perception or asymbolia for pain (Biemond, 1956; Berthier et al., 1988; Masson et al., 1991; Greenspan and Winfield, 1992; Schmahmann and Leifer, 1992).
Unfortunately, thalamic lesions that produce analgesia and thermoanaesthesia can also produce central pain, either immediately or after some delay (Bogousslavsky et al., 1988; Leijon et al., 1989; Bowsher et al., 1998; for further references see Pagni, 1998). As emphasized by Bowsher and colleagues (Bowsher et al., 1998), ~61% of their large sample of central pain patients had lesions in the ventroposterior thalamic region. Many such cases are ascribed to infarct or haemorrhage within the territory supplied by the thalamogeniculate branches of the posterior cerebral artery (Head and Homes, 1911; Pagni, 1998). The possible mechanisms behind the development of central pain following posterior thalamic lesions have recently been discussed in detail (Craig, 1998). In short, it has been proposed that at least some types of central pain are due to a release phenomenon caused by interruption of the thermosensory pathway and the subsequent loss of thermosensory integration and of cold inhibition of the central structures that mediate the perception of burning pain (Craig, 1998). The latter structures include the target of lamina I spinothalamic tract fibres in MDvc that provides input to the anterior cingulate cortex and that is spared by lateral thalamic lesions that typically evoke central pain (Bogousslavsky et al., 1988; Boivie and Leijon, 1991; Schmahmann and Leifer, 1992; Pagni, 1998). Imaging studies in humans have shown that the anterior cingulate cortex is activated by noxious stimuli and is a critical site for the affective-motivational aspects of thermal pain and other adverse sensations (Hsieh et al., 1994; Craig et al., 1996; Rainville et al., 1997).
Comparison with prior calbindin staining in VPM/VPL
In a prior study in macaque monkeys, Rausell and Jones (1991a, b, 1992) described small calbindin-positive neurons within VPM and VPL that they considered to constitute a `matrix' in these and other dorsal thalamic nuclei (Jones, 1998), and they reported that these neurons receive ascending spino- and trigeminothalamic projections. In order to alleviate any possible confusion, we emphasize that the calbindin staining pattern they described is not the same as that observed in the present material, due to the use of different antibodies and different tissue treatment regimens. The antibody they used (obtained privately from P. C. Emson) apparently preferentially stained calbindin-like immunoreactivity in cell bodies, whereas the commercially available antibody we used (from Sigma) preferentially stained fibres and terminals. Clearly, these different monoclonal antibodies recognize epitopes that are different, or that are differently exposed, in these cellular compartments. Accordingly, calbindin-positive cells rather than fibres were emphasized in their studies, and they focused on such cells in the VPM and VPL nuclei. They did not analyse the Sg/Po region, they did not recognize the VMpo nucleus, and they only briefly mentioned terminal-like calbindin staining. Nonetheless, we did observe lightly stained patches of calbindin-positive cells at some of the same locations reported by Rausell and Jones (1991b, 1992; see also Morel et al., 1997). For example, as seen in Fig. 1D, there are patches of calbindin immunoreactivity particularly along the medial border of VPM and also interspersed in the VPM/VPL complex. However, these areas contain calbindin-labelled cells and few fibres. These areas may represent the regions described by Rausell and Jones (1991b, 1992) in the macaque, and if so, they probably receive spinothalamic input from lamina V cells and contain non-selective nociceptive neurons (cf. Kenshalo et al., 1980; Iwata et al., 1992; Ralston and Ralston, 1992; Bushnell et al., 1993; Lenz et al., 1993a; Apkarian and Shi, 1994). Thus, although these sites may also play some role in pain transmission, it is certain to be different from that subserved by the VMpo (Perl, 1984; Craig, 1998).
Significance of immunoreactivity for substance P and CGRP
The present study demonstrated patches of substance P-like immunoreactivity that partially overlap with the dense terminal-like calbindin staining in VMpo. Hirai and Jones also noted similar patches of tachykinin-immunoreactive fibres within their Sg/Po region in human thalamus (see Fig. 2 in Hirai and Jones, 1989a; note that this area is different from their so-called `putative nucleus submedius'). The substance P-labelled fibres we observed appeared to enter VMpo by the same route as the calbindin-labelled fibres, suggesting that they may also be spinothalamic fibres. In cats and rats, some spinothalamic terminations in the posterior thalamus are substance P-immunoreactive (Battaglia et al., 1988), and some substance P-positive superficial dorsal horn neurons project to the thalamus and other sites (Noguchi et al., 1992; Blomqvist and Mackerlova, 1995). However, the substance P labelling we observed was much sparser than the calbindin labelling, and it can account for only a small portion of the lamina I spinothalamic tract input, suggesting that the majority of the spinothalamic tract neurons that project to the VMpo expresses other neurotransmitters in addition to glutamate (Blomqvist et al., 1996). Because lamina I spinothalamic tract cells are both functionally and morphologically distinct (Han et al., 1998), it is possible that substance P is expressed by a functional subset of these neurons. Similarly, in the monkey spinal cord substance P (NK-1) receptors are preferentially found on nociceptive lamina I neurons, whereas thermoreceptive neurons lack such receptors (Yu et al., 1999), and in the owl monkey spinal trigeminal nucleus, thermoreceptive lamina I trigeminothalamic tract neurons are devoid of a substance P innervation, in contrast to other lamina I trigeminothalamic tract cells (Craig et al., 1999).
Our cytoarchitectonic and immunohistochemical observations reveal that the VMpo is not immediately adjacent to the VPL, VPM and VMb nuclei, but is separated from them by a region of small, dispersed and faintly labelled neurons around which patches of CGRP-immunoreactive fibres occur. This dense CGRP staining, presently localized to the Po nucleus, contrasted with the sparse CGRP staining in the VMb. CGRP labelled terminal fibres are observed through most of the VMb (or VPMpc) nucleus in the rat (e.g. Kruger et al., 1988b, c; Yasui et al., 1989), the only other species in which CGRP staining so far has been investigated in the thalamus. These terminations originate preferentially from the external medial parabrachial subnucleus (Kruger et al., 1988a, b; Yasui et al., 1989) and convey visceral afferent information (Cechetto and Saper, 1987; Yasui et al., 1989) and perhaps aversive properties of gustatory stimuli (Yamamoto et al., 1994; Halsell and Travers, 1997). In contrast, CGRP-negative neurons in the central medial parabrachial subnucleus seem to be the obligatory relay for gustatory activity related to taste quality originating in the nucleus of the solitary tract in the rat (Norgren and Leonard, 1971; Yasui et al., 1989; Halsell and Travers, 1997), and these parabrachial subnuclei have distinct termination patterns within VMb in the rat (Yasui et al., 1989; Karimnamazi and Travers, 1998). Similarly, different regions in the macaque thalamus relay taste and visceral information to the cerebral cortex (Pritchard et al., 1989). Consequently, our observations suggest an even clearer separation of taste and visceral information in the human posterior thalamus, consistent with the non-peptidergic nature of the gustatory pathway (Kruger and Mantyh, 1989), with the direct projection of gustatory activity to VMb from the nucleus of the solitary tract in primates (Beckstead et al., 1980) and with the long-suspected encephalization of visceral (enteroceptive) sensation in humans (Bishop, 1959). Accordingly, the CGRP-positive area between VMb and VMpo we have observed could constitute the vagal afferent visceral representation in the human thalamus. This suggestion, of course, remains tentative until comparable anatomical data in monkeys are available.
Functional organization of the VMpo region
The VMpo nucleus can be identified in the human, the macaque monkey (Craig et al., 1994) and the New World owl monkey (Blomqvist et al., 1996; Craig et al., 1999). [In cats and rats, a primordial homologous region may exist as a parvocellular area adjacent to VMb (Craig, 1996).] Tracing evidence and electrophysiological evidence in both the macaque and the owl monkey indicate an anteroposterior topography within VMpo. Trigeminal inputs are represented anteromedially, with the forelimbs and hindlimbs represented successively more posterolaterally (Craig et al., 1994, 1999; Blomqvist et al., 1996). If a similar topography exists in the human VMpo nucleus, then the head, face and mouth are represented posterior and ventromedial to the VPM nucleus, and the arms and legs are represented posterior and ventromedial to the VPL nucleus (Fig. 2). This would be consistent with the findings in clinical studies, in which pain and temperature sensations evoked by microstimulation have often been referred intraorally when the electrode trajectories passed through the face and arm region of V.c. (Lenz et al., 1993b), and on the trunk and leg (including deep and visceral structures) when electrode trajectories passed more laterally (Lenz et al., 1994; Davis et al., 1995). A similar topography could exist in the CGRP zone, the putative general visceral relay that we denote as the Po nucleus; its larger ventrolateral part could represent vagal and pelvic nerve afferent information from thoracic, abdominal and pelvic organs, and its smaller dorsomedial region, intercalated between the VMpo and the VPM, could represent general visceral information from non-gustatory afferents in the facial and glossopharyngeal nerves. This suggestion also fits with a recent clinical study, in which taste sensations were evoked by thalamic microstimulation adjacent to an area from which general visceral sensations from the throat and noxious sensations from the mouth were elicited (Lenz et al., 1997).
From a global viewpoint, the VMpo, the Po and the VMb may be considered together as the thalamic representations of activity that signals the physiological condition of the somatic and visceral tissues of the entire body. This is in line with the idea that the lamina I neurons are part of an afferent homeostatic (or enteroceptive) system that signals the physiological status and integrity of the body, which includes the specific sensations of pain and temperature (Craig, 1996a, b, 1998). The topographic relationships between the lamina I representation, the general visceral representation and the gustatory (special visceral) representation in the posterior thalamus are similar (though rotated) to the topographic relationships between these representations in the parabrachial nucleus (Fulwiler and Saper, 1984; Hermanson and Blomqvist, 1996), the main sensory relay for homeostatic afferent information to the hypothalamus. This organizational scheme also appears to reflect the general developmental pattern of the alar plate (Opdam et al., 1976). Finally, it is important to note emerging evidence in humans and rats that suggests that sensory afferent information may also be incorporated within this pathway (Veening and Coolen, 1998; Vallbo et al., 1999).
Conclusions
In summary, clinical observations and experimental studies in humans and non-human primates have provided evidence that the thalamic VMpo nucleus is a critical structure for pain and temperature sensation. In the present study we have described its cytoarchitectonic relationships and some of its chemoarchitectonic characteristics. These observations should provide a solid basis for future studies on the processing of pain and temperature in the human thalamus, under both normal and pathological conditions.
|
Acknowledgments |
|---|
We thank Dr Jörgen Boivie for critical reading of the manuscript and Ludmila Mackerlova and Maribeth Tatum for help with photography. This study was supported by grants from the Swedish Medical Research Council (project 7079), NIH grant NS25616 and by the James R. Atkinson Memorial Pain Research Fund administered by the Barrow Neurological Foundation.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Apkarian AV, Hodge CJ. A dorsolateral spinothalamic tract in macaque monkey. Pain 1989; 37: 323–33.[ISI][Medline]
Apkarian AV, Shi T. Squirrel monkey lateral thalamus. I. Somatic nociresponsive neurons and their relation to spinothalamic terminals. J Neurosci 1994; 14: 6779–95.[Abstract]
Apkarian AV, Stevens RT, Hodge CJ. Funicular location of ascending axons of lamina I cells in the cat spinal cord. Brain Res 1985; 334: 160–4.[ISI][Medline]
Arvidsson U, Cullheim S, Ulfhake B, Hökfelt T, Terenius L. Altered levels of calcitonin gene-related peptide (CGRP)-like immunoreactivity of cat lumbar motoneurons after chronic spinal cord transection. Brain Res 1989; 489: 387–91.[ISI][Medline]
Battaglia G, Rustioni A. Substance P innervation of the rat and cat thalamus. II. Cells of origin in the spinal cord. J Comp Neurol 1992; 315: 473–86.[ISI][Medline]
Battaglia G, Spreafico R, Rustioni A. Substance P-immunoreactive fibres in the thalamus from ascending somatosensory pathways. In: Bentivoglio M, Spreafico R, editors. Cellular thalamic mechanisms. Amsterdam: Excerpta Medica; 1988. p. 365–74.
Beckstead RM, Morse JR, Norgren R. The nucleus of the solitary tract in the monkey: projections to the thalamus and brain stem nuclei. J Comp Neurol 1980; 190: 259–82.[ISI][Medline]