Correlations between neuronal morphology and electrophysiological features in the rodent superficial dorsal horn

doi:10.1113/jphysiol.2001.012890

Correlations between neuronal morphology and electrophysiological features in the rodent superficial dorsal horn

T. J. Grudt and E. R. Perl

Department of Cell and Molecular Physiology, University of North Carolina-CH, Chapel Hill, NC 27599-7545, USA

ABSTRACT

Top
Abstract
Introduction
Methods
Results
Discussion
References

Relationships between the morphology of individual neurones of the spinal superficial dorsal horn (SDH), laminae I and II, and their electrophysiological properties were studied in spinal cord slices prepared from anaesthetized, free-ranging hamsters. Tight-seal, whole-cell recordings were made with pipette microelectrodes filled with biocytin to establish electrophysiological characteristics and to label the studied neurones. Neurones were categorized according to location and size of the somata, the dendritic and axonal pattern of arborization, spontaneous synaptic potentials, evoked postsynaptic currents, pattern of discharge to depolarizing pulses and current-voltage relationships. Data were obtained for 170 neurones; 13 of these had somata in lamina I and 157 in lamina II. Stimulation of the segmental dorsal root evoked a prompt excitatory response in almost every neurone sampled (161/166) with nearly 3/4 displaying putative monosynaptic EPSCs. The majority of neurones (133/170) fitted one of several distinctive morphological categories. To a considerable extent, neurones with a common morphological configuration and neurite disposition shared electrophysiological characteristics. Five of the 13 lamina I neurones were relatively large with extensive dendritic arborization in the horizontal dimension and a prominent axon directed ventrally and contralaterally. These presumptive ventrolateral projection neurones differed structurally and electrophysiologically from the other lamina I neurones, which had ipsilateral, locally arborizing axons and/or branches entering the dorsal lateral funiculus. One hundred and twenty lamina II neurones fitted one of five morphological categories: islet, central, medial-lateral, radial or vertical. Central cells were further divided into three groups on functional features. We conclude that the spinal SDH comprises many types of neurones whose morphological characteristics are associated with specific functional features implying diversity in functional organization of the SDH and in its role as a major synaptic termination for thin primary afferent fibres.

(Received 21 June 2001; accepted after revision 17 December 2001)
Corresponding author E. R. Perl: Department of Cell and Molecular Physiology, University of North Carolina-CH, Chapel Hill, NC 27599-7545, USA. Email: erp{at}med.unc.edu

INTRODUCTION

Top
Abstract
Introduction
Methods
Results
Discussion
References

The superficial dorsal horn (SDH) of the mammalian spinal cord, laminae I and II, is a major termination region for thin primary afferent fibres (Ranson, 1914; Sindou et al. 1974; Snyder, 1977; Light & Perl, 1979). Initially, the proposal, and eventually the establishment, that such fine afferent fibres are important in transmitting information related to pain and temperature linked the SDH to these sensations (reviewed in Perl, 1984). Direct evidence for the SDH function has come from the demonstration that it contains neurones selectively responsive to noxious and/or thermal stimuli (Christensen & Perl, 1970; Kumazawa & Perl, 1978; Light et al. 1979; Craig, 1998). In particular, lamina I (the marginal zone) has been shown to contribute substantially to the spinoreticular and spinothalamic pathways of the contralateral, lateral spinal cord (see reviews: Perl, 1984; Light, 1992; Craig, 1998).

In spite of its importance, remarkably little is known about the relationship of inherent functional features of the region's component neurones to their architecture and interconnections. The principal available information has come from studies addressing morphological characteristics using either Golgi staining or horseradish peroxidase labelling in conjunction with electrophysiological recording. Interpretations or deductions have differed considerably, suggesting either variability in SDH neuronal geometry or the presence of numerous neuronal categories. The shape of the somata and the principal dendritic origins has led to categorization of lamina I neurones as fusiform, pyramidal or multipolar (Gobel, 1978a; Lima & Coimbra, 1986; Beal et al. 1989; Zhang et al. 1996). There is less agreement about lamina II. Ramon y Cajal (1909) described three types of neurones in the substantia gelatinosa (lamina II) of the young cat: vertically oriented 'limitroph' cells with dendrites directed sagitally, 'central' cells located in the middle of lamina II with bushy dendrites directed rostro-caudally, and 'transverse' cells with processes directed medio-laterally. Gobel (1975, 1978b), also utilizing Golgi-stained tissue, defined four main types of neurones in the cat trigeminal lamina II: 'islet' neurones with abundant sagitally oriented dendrites, 'stalked' neurones with dendritic processes oriented vertically and sagitally, 'arboreal' (stellate-like) and 'border' cells. In addition, Gobel's reports (1975, 1978b) mention two other varieties, 'spinal' and 'inverted stalk' cells. In monkey lamina II, only stalked and islet cells were reported by Price et al. (1979), while Beal & Cooper's (1978) analysis of the primate describe heterogeneity to the extent that classification was impossible, a conclusion also reached for other species (Light et al. 1979; Réthelyi et al. 1989). In the rat, Todd & Lewis (1986) found that neurones resembling Gobel's stalked and islet cells made up the majority of lamina II cells with the remainder composed of several groups with some characteristics of Gobel's minor categories; however, a number of neurones were said to have structural features of more than one group. Beal and colleagues have described much morphological variability in rat lamina II neurones (Beal, 1983; Bicknell & Beal, 1984; Beal et al. 1989) differentiating between neurones displaying long projecting axons and those with more local or regional outputs. In lamina II of the human spinal cord, Schoenen (1982) recognized four morphological types of neurones that only partially fitted categories identified in other species.

The main purpose for the present study was to compare intrinsic electrophysiological properties of individual SDH neurones to their geometry. Previous investigations have indicated electrophysiological heterogeneity of SDH neurones (Thomson et al. 1989; Yoshimura & Jessell, 1989a,b; Lopez-Garcia & King, 1994). We describe here that features such as action potential firing pattern, responses to primary afferent volleys, current-voltage relationships and the frequency of spontaneous miniature events often correlate with morphological characteristics. These correlations define distinctive categories of SDH neurones. The neuronal subsets recognized are distinguished by laminar location, a particular geometry and neurite orientation as well as by a constellation of electrophysiological attributes. Some of these observations have been reported at scientific meetings (Grudt & Perl, 1998, 1999).

METHODS

Top
Abstract
Introduction
Methods
Results
Discussion
References

Deeply anaesthetized (1.6 g kg^-1 urethane), young (3-5 weeks postpartum), free-ranging hamsters were killed by exsanguination and cardiac arrest during removal of the spinal cord. All procedures conformed with the Institutional Animal Care and Use Committee of the University of North Carolina-CH. Sagittal or transverse lumbar spinal slices (900-1100 µm) with attached segmental dorsal roots were prepared using a vibrating microtome and maintained at room temperature in artificial cerebral spinal fluid (ACSF) equilibrated with 95 % O₂ and 5 % CO₂. ACSF consisted of (mM): NaCl 125, NaHCO₃ 26, NaH₂PO₄ 1.25, KCl 2.5, CaCl₂ 2, MgCl₂ 1 and D-glucose 26.

Tight-seal, whole-cell recordings, selectable by amplifier setting as current clamp or voltage clamp, were obtained with patch-type pipette electrodes visually guided into the SDH. The pipette internal solution consisted of (mM): potassium gluconate 130, NaCl 5, CaCl₂ 1, MgCl₂ 1, EGTA 11, Hepes 10, Na₂ATP 4. The recording electrodes were back-filled with internal solution containing 0.5 % biocytin to label neurones for morphological analysis.

Command voltages or currents were generated through the recording electrode by an Axopatch 200B amplifier (Axon Instruments, Inc., Union City, CA, USA) using a PC digital computer and the pCLAMP 6 computer program (Axon Instruments). Current-voltage curves were constructed by holding neurones at -60 mV in voltage clamp and stepping for 500-800 ms every 5 s, initially to -50 mV and then in 10 mV increments to -120 mV. The firing pattern of each neurone was determined in current clamp by passing 1 s long depolarizing pulses through the recording electrode from a holding potential of -60 mV; these were graded from subthreshold to considerably supraliminal. When differences in firing pattern in near-threshold and supraliminal depolarizations were noted, it is stated in the presentation and was considered in the classification.

A suction electrode was used to stimulate the segmental dorsal root (DR) with graded 0.2 ms pulses. Conduction velocities of primary afferent fibres evoking monosynaptic EPSCs were estimated from the latency of the evoked response and the length of the DR (~10 mm); an allowance of 1 ms was used for synaptic delay. In experiments measuring the compound action potential of the DR, the distal end of the root was stimulated using the same suction electrode employed in the whole-cell experiments. The resulting compound action potential, recorded from the central end of the DR with a second suction electrode, was used to determine the conduction velocities of the DR fibre groups (A $alpha$ $beta$ : > 5 ms^-1, A $delta$ : 4-0.8 ms^-1, C: < 0.8 ms^-1) . These values obtained from compound potential recordings in the present experiments are very similar to those published by Lawson et al. (1997).

Following completion of the electrophysiological recording, the slices were fixed in 4 % paraformaldehyde and 4 % sucrose in 0.1 mM phosphate buffer, stored overnight in 30 % sucrose in phosphate buffer, then frozen and sectioned transversely or parasagittally at 60 µm. Biocytin-containing neurones were visualized by an immune/horseradish peroxidase reaction using the avidin-biotin complex technique (ABC kit, Vector Laboratories). Images of the labelled neurones were studied using a Nikon compound optical microscope. Micrographs of these images were captured through the microscope with a digital camera (Optronics DEI-470) and a digital computer using the Image-Pro Plus program. Images were manipulated only by adjustment of contrast or for the preparation of illustrations to adjust for uniform magnification.

Location of the somata and neurites of the labelled neurones was determined using bright field and dark field microscopy. The SDH or laminae I, II outer (II_o) and II inner (II_i) appeared as a distinctive dark bands in dark field illumination due to the relative absence of myelinated fibres. Dimensions of dendritic arbors in the rostro-caudal plane were measured in spinal cord slices sectioned parasagittally or calculated from the number of transverse, 60 µm sections containing dendrites. The dendritic arbor extent in the medio-lateral plane was determined in slices sectioned transversely, while the dorsal-ventral dimension was established from measurements on both sagittally and transversely sectioned slices. Dimensions are given without taking account of shrinkage in the histological preparation.

Bicuculline, strychnine, 4-AP (4-aminopyridine) and TTX (tetrodotoxin) were obtained from Sigma (St Louis, MO, USA).

RESULTS

Top
Abstract
Introduction
Methods
Results
Discussion
References

Electrophysiological and morphological data were obtained on 170 superficial dorsal horn (SDH) neurones whose stained somata and processes were recovered histologically. The biocytin-labelled material showed all cell bodies were within the confines of the relatively narrow band forming the SDH; however, their shape and the distribution of their neurites varied greatly.

Morphological features

No single morphological feature distinguished one neurone from another. To characterize neurones structurally, the following were given weight: the laminar location and size of the soma; the relative size, number and branching pattern of dendrites; the laminar distribution and principal orientation of the dendritic tree; the size, branching pattern and distribution of the axon.

Figure 1 plots the distribution of the relative locations of recovered neuronal somata. Thirteen of these neurones had somata located just inside the white matter overlying the dorsal horn and external to the small neurones of lamina II. These 13 were judged to be in lamina I. The other 157 neurones had somata scattered over lamina II_o and II_i. Neurones differed not only in location, but also in the size of the somata and the size and extent of their dendritic and axonal processes (Figs 2-4).

View larger version
[in this window]
[in a new window]

Figure 1. Locations of the recorded SDH (laminae I, II_o and II_i) neurones
Each neurone studied was plotted as a circle on a schematic drawing of the lower lumbar dorsal horn. The position of each neurone was established using both darkfield microscopy, in which the myelin-poor SDH appears dark, and bright field cytoarchitectonic markers of the laminae.

The measurements mentioned subsequently and provided in Table 1 were made on histologically fixed material after processing for visualization of the biocytin label, a processing that is known to result in some tissue shrinkage. We estimated shrinkage in thickness of sections produced by histological processing by comparing dimensions of freshly cut slices before and after processing using a confocal microscope. The thickness of a histological section decreased by about 30 % in the development of the biocytin label. This change was possibly due to the removal of H₂O during mounting and sealing of a coverslip over the processed section as well as pressure by the coverslip. Therefore, it seems probable that shrinkage varied according to the orientation of the histological slice relative to the dimension under consideration. The majority of our observations were made on transverse spinal cord slices, which were then sectioned sagitally for histology. Thus, the dimensions given should be considered approximate and significant only for comparison of essentially identically treated material.

The somata in the sample varied from 19 to 30 µm in the largest dimension. On average, a subset of lamina I cells were the largest and certain neurones in the centre of lamina II_i, the smallest; distinctions that could be expected from past histological studies (Waldeyer, 1889; Ramon y Cajal, 1909; Pearson, 1952; Réthelyi & Szentágothai, 1973; Beal et al. 1989). In evaluating the size of the neuronal somata and arborizations of their neurites, one needs to be mindful that the observations are on tissue from small and relatively young rodents.

The size, the nature and orientation of neurite arborization differed between cells in what proved to be a systematic fashion. Several morphologically distinct categories were evident, principally on the grounds of the arrangement and distribution of the dendrites and the pattern and region of axonal arborization.

The branching structure of dendrites differed (Figs 2-4). In certain groups of neurones, dendrites were relatively thick while in others they were thin. Some subsets of neurones had many dendritic branches and a dense arborization while others had few branches and a much sparser dendritic tree. Almost all neurones in the sample exhibited some dendritic spines; however, the density of spines varied substantially from one subset to another. The orientation of the dendritic tree proved to be an important distinguishing feature. In confirmation of earlier morphological literature (e.g. Scheibel & Scheibel, 1968), most lamina II neurones had dendritic arbors that were primarily orientated in the rostro-caudal dimension and relatively flattened in the medio-lateral and dorso-ventral directions. Exceptions to the more common dendritic disposition served to set apart certain neuronal categories.

View larger version
[in this window]
[in a new window]

Figure 2. Micrographs of biocytin-stained lamina I neurones
Two examples are shown of each category as captured from 60 µm parasagittal histological sections. Dorsal is shown up. A, lamina I projection cells. Note the thick ventrally directed axon in A1, which in adjacent sections extended ventral to the central canal and turned toward the contralateral side of the spinal cord. B, lamina I non-projection cells. Axons of non-projection lamina I neurones, identified as uniform thin profiles originating from the soma (or on occasion a proximal dendrite) arborized locally.

View larger version
[in this window]
[in a new window]

Figure 3. Micrographs of biocytin-stained lamina II neurones exemplifying classified categories
Two examples of each category are shown. Images were captured from 60 µm parasagittal histological sections except D2, which was taken from a 60 µm transverse section. Dorsal is shown up. See text for additional details. A, islet cells. The dendritic and axonal arbors of this class extended further rostro-caudally than any other type of lamina II neurone. Although the soma was occasionally located in II₀, most of the dendritic and axonal arborizations were in lamina II_i. B, radial cells. The soma of these neurones was usually located near the border between lamina II_o and II_i. C, central cells. This class had a general configuration similar to islet cells, but with a much smaller dendritic expansion in the rostro-caudal direction. D, medial-lateral cells. A group of three neurones notable for being the sole lamina II neurones with dendrites that extended substantially in the medio-lateral dimension. Dendrites of the cell shown in D1 were present in five adjacent 60 µm parasagittal sections. The image illustrated in D2 is from a 60 µm transverse section. E, vertical cells. A partially heterogeneous group which had in common a dendritic arbor expanding more prominently dorso-ventrally than other lamina II neurones.

The projection of the axon also served to distinguish neurones morphologically. The axon was identified by its generally thinner and relatively constant size from the soma outward and the absence of spines. The axon was strikingly thick in certain neurones of lamina I. The distribution, trajectory and degree of branching of the axon also varied from one category to another.

View larger version
[in this window]
[in a new window]

Figure 4. Micrographs of unclassified lamina II cells
About 25 % of lamina II cells in the sample did not fit the criteria for any of the categorized groups. A, a lamina II neurone that had a smaller and less dense dendritic tree than islet cells and longer, less compact dendritic arborization than central cells (Fig. 3C). B, a lamina II neurone with dendrites spreading in several directions with a less dense arbor than the radial category (Fig. 3B). C, a lamina II neurone with vertically directed dendrites with a much denser arborization than the vertical category (Fig. 3E). In addition to morphological distinctions from the most similar of the categorized class, unclassified neurones exhibited functional properties that differed from those of classified groups.

View larger version
[in this window]
[in a new window]

Figure 5. Whole-cell, voltage clamp recording of spontaneous synaptic currents from a SDH neurone
Membrane potential was held at -50 mV. Spontaneous inward (downward) and outward (upward) postsynaptic currents are prominent.

Physiological features

Spontaneously appearing currents, inward and/or outward, were readily visible in the relatively low noise, tight-seal, whole-cell recordings. An example appears in Fig. 5. In essentially identical preparations and conditions (-50 or -60 mV holding potential), spontaneous inward currents declined in size or disappeared in the presence of the AMPA/KA glutamate receptor antagonist, CNQX (6-cyano-7-nitroquinoxaline-2,3-dione, 10 µM) or at holding potentials near 0 mV (Bao et al. 1998; J. Bao & E. R. Perl, unpublished observations). Such ongoing inward currents in the present work were considered to be spontaneous excitatory postsynaptic currents (sEPSCs). They ranged in amplitude from 5 pA to over 60 pA, averaging 15-20 pA (Figs 5-10). In the presence of TTX (1 µM), in most instances larger background inward currents (over 40 pA) disappeared while those under 30 pA were relatively unchanged in frequency (not shown). At least a few spontaneous outward currents appeared in recordings from almost all cells; some exhibited substantial numbers (2-10 s^-1) as illustrated in Fig. 5. In contrast to the spontaneous inward currents, under our conditions spontaneous outward currents decreased in amplitude at holding potentials of -70 mV and were suppressed completely by the glycine receptor antagonist strychnine (1 µM), the GABA_A receptor antagonist bicucullin (10 µM), or the two together (Bao et al. 1996; J. Bao & E. R. Perl, unpublished observations). For these reasons, we considered the ongoing outward currents to be spontaneous inhibitory postsynaptic currents (sIPSCs). Certain categories of cells, distinguishable on other grounds, tended to have more frequent spontaneous EPSCs or IPSCs than others (Table 2).

Neuronal input resistance (R_in) ranged from about 150 M $Omega$ to over 1 G $Omega$ (Table 2). Systematic differences in R_in appeared between certain groups of neurones. Criteria for a neurone to be considered in 'good' condition was an input resistance of over 100 M $Omega$ and a low current (< 100 pA) necessary to maintain voltage clamp at the targets of -50 or -60 mV.

Resting membrane potential (RMP) measured when initially going into the whole-cell mode varied from -40 to over -60 mV (Table 2). In at least one category of neurones, the RMP was substantially less than in the remainder of the sample.

The initiation and repetition of action potentials generated by a 1 s depolarizing pulse distinguished certain categories. Some classes of neurones responded tonically with repeated action potentials at relatively regular rates in the vicinity of 10 per s (e.g. Figs 6B2a, 7A1a, 7A2a and 8B2a). Other cells gave only a transient response consisting of a few action potentials during the step depolarization (Figs 6A2a, 8A1a, 8A2a, 9A1a and 9A2a). Still others consistently delayed the initiation of the first action potential and thereafter responded with a number of action potentials either in a steady, regular fashion or a bursting pattern (Figs 7B1a, 7B2a, 8B1a and 10Ba).

Certain voltage-dependent currents were differentially displayed across the sample. One was a slowly ebbing inward current that reached a plateau between 500 and 800 ms after hyperpolarizing steps to -70 mV or greater (e.g. Figs 6A1c, 7A2c, 8B1c and 8B2c). This inward current was blocked by Cs⁺ (2 mM) in 4/4 neurones and was unaffected by 2 mM Ba²⁺ (2/2) making it comparable with the hyperpolarization-activated inward current called I_h (Mayer & Westbrook, 1983; Yoshimura & Jessell, 1989a,b; Lüthi & McCormick, 1998). A smaller proportion of neurones exhibited a sizeable outward current at the end of hyperpolarizing steps (Figs 8A1c, A2c and B1c), which appeared similar to the current known as I_A (Conner & Stevens, 1971; Rudy, 1988; Yoshimura & Jessell, 1989b)_. By themselves, neither I_h nor I_A proved sufficient to specify a neuronal category; however, in combination with other features including firing pattern, they were part of the constellation of characteristics distinguishing certain groups.

In all but a few neurones, stimulation of the segmental dorsal root (DR) evoked inward currents. A low coefficient of variation of the latency of successive responses to graded DR stimuli suggested many of these were monosynaptically mediated responses (Yoshimura & Jessell 1989a; Li & Perl, 1994). In similar slice preparations, CNQX (10 µM) suppressed such DR-evoked responses suggesting mediation by AMPA/KA glutamate receptors (Bao et al. 1998; J. Bao & E. R. Perl, unpublished observations). Certain groups of SDH neurones were characterized by exceptionally large or relatively small DR-evoked responses. Since large differences in amplitude of DR-evoked responses consistently related to other features, we considered that they reflected the pattern of connectivity to the recorded cells. In the illustrations of DR-evoked activity in Figs 6-10, the maximal response is shown.

Often, the DR-evoked currents had complex shapes with both inward and outward components. Some of the evoked inward currents exhibited successive notches (e.g. Figs 7A2b, 9B1b and 9B2b) that were related to the intensity of the DR stimulation. A number of these later components also had small variations in latency on successive trials, suggesting probable generation through monosynaptic connections from more slowly conducting primary afferent fibres. The calculated conduction velocity for the initial deflection of the DR-evoked responses varied for different neuronal categories, ranging from an average of 0.3 m s^-1 to over 0.8 m s^-1 (see Table 2). Relating delays due to conduction to the timing of deflections of the compound action potential recorded from dorsal roots (not shown), indicated that C component primary afferent fibres often made monosynaptic connections to certain SDH neurones. Other neuronal groups appeared to receive monosynaptic inputs from DR A $delta$ fibres.

We found no evidence of monosynaptic linkage from the rapidly conducting A $alpha$ $beta$ fibres. The shortest latency DR-evoked EPSCs, fitting monosynaptic criteria in our sample, yielded dorsal root conduction velocities < 4 ms^-1, even with the allowance of 1 ms for synaptic delay, values that are in the A $delta$ fibre range. This contrasts with a report that describes monosynaptic dorsal root A $alpha$ $beta$ input to a substantial number of lamina II neurones in young postpartum rat (Park et al. 1999). These differences could reflect species differences or distinctions related to relative development in the two species. Alternatively, it is possible that relatively weak excitatory inputs from A $alpha$ $beta$ fibres were masked in our recordings by DR-evoked inhibitory currents, although such masking was not evident during graded DR stimulation.

Classification

Functional features often overlapped or were shared by neurones with quite different forms and locations, yet it proved possible to make a coherent classification of most of the sample by using a combination of morphological and functional criteria. In part, the categorization was influenced by the laminar description of the spinal grey matter based on cytoarchitectural characteristics that differentiated lamina I (marginal zone) from lamina II (substantia gelatinosa) and divided the latter into II outer (II_o) and II inner (II_i) parts (Rexed, 1952; see also Brown & Réthelyi, 1981). We also were swayed by the notable differences in dendritic arrangement, distribution and density. In general, the categories derived from our material confirm and extend reports by previous investigators (Ramon y Cajal, 1909; Pearson, 1952; Scheibel & Scheibel, 1968; Matsushita, 1969; Gobel, 1975, 1978a,b; Sugiura, 1975; Mannen & Sugiura, 1976; Beal & Cooper, 1978; Beal, 1979, 1983; Price et al. 1979; Schoenen, 1982; Abdel-Maguid & Bowsher, 1984, 1985; Bicknell & Beal, 1984; Bowsher & Abdel-Maguid, 1984; Lima & Coimbra, 1986; Beal et al. 1989). Structurally, seven distinct neuronal types were recognized. One morphological type was further subdivided into three groups on the basis of functional features. One hundred and thirty three of the 170 neurones were thus classified into one of the nine categories. The remaining 37 neurones differed from the seven morphological categories and were unclassified. We speculate that had our studied population been larger, additional types may have become recognizable that could accommodate some or all of the 37 unclassified cells.

Lamina I neurones

The 13 lamina I neurones varied in shape. Two distinct categories were recognized: putative ventralateral tract projection (VL-projection) neurones and neurones presumed not to project in that pathway (non-projection). Additional categories may exist; however, the sample may have been too limited for their definition. VL-projection neurones were characterized by a relatively thick axon that ran ventrally and medially toward the contralateral spinal cord. Cells in the non-projecting group had axons that were directed ipsilaterally and that frequently exhibited extensive branching within the SDH itself.

Dendrites of lamina I neurones were relatively sparse and typically spread both medio-laterally and rostro-caudally capping part of the dorsal horn grey matter. The dorso-ventral dendritic extension was largely confined to laminae I and II_o. The extent of the dendritic arborization, therefore, was mostly in a horizontal plane. All lamina I neurones had some dendritic spines, but they were not a dominant feature (see Fig. 2 and Table 1).

The thick axons of lamina I VL-projection neurones (Fig. 2A1) in the most medial sections available, coursed to positions ventral to the central canal. None of the VL-projection neurones gave off axon collaterals that were distributed in the vicinity of the cell's dendritic arbor; however, two of the five had an axon collateral that entered the dorsal lateral funiculus (DLF) to lie lateral and ventral to the most lateral extension of the dorsal grey matter. It has been suggested that in rodents the DLF is part of Lissauer's tract (Ranson, 1914).

The axonal arborization of non-projection lamina I neurones varied (Fig. 2B1 and B2). Six of the eight cells had a limited arborization in lamina I itself while the other two had extensive branching in lamina II. One neurone, classified as non-projection, had an axon collateral running longitudinally in the DLF and therefore may have represented a different type of projection neurone from the VL-projection category.

Table 2 summarizes electrophysiological features of lamina I neurones. Input resistances of the VL-projection neurones were considerably lower than those of non-projection cells and substantially lower than any other categories in the sample. All of the projection neurones (5/5) and most (5/8) of the non-projection neurones exhibited I_h (Fig. 6A1c, A2c, B1c); however, the amplitude of I_h of the VL-projection neurones (160 ± 32.1 pA) was notably greater than for non-projection cells (55.0 ± 12.4 pA).

View larger version
[in this window]
[in a new window]

Figure 6. Electrophysiological properties of lamina I neurones
Two examples are shown for each neuronal category. The responses are from the cells whose morphology is illustrated in Figure 2. A1 and 2, projection cells. B1 and 2, non-projection cells. From above down: a, current clamp recordings of action potential firing patterns in response to a depolarizing step from -60 mV; b, voltage clamp recordings of the synaptic response elicited by stimulating a segmental dorsal root at a holding potential of -50 mV; c, current-voltage relationship obtained by holding the cell at -60 mV in voltage clamp and stepping in 10 mV increments initially from -60 to -50 mV and then stepwise to -120 mV.

In current clamp, projection cells (4/5) discharged action potentials transiently when depolarized by a near-threshold step (Fig. 6A1a and A2a). Larger step depolarizations evoked sustained discharge at a near constant interspike interval. One projection cell differed by firing bursts of high frequency action potentials. It seemed likely that the transient discharge at threshold depolarization may have resulted from inactivation of I_h during the step; the depolarizing steps were started at -60 mV, a level at which I_h was active. This was confirmed in one neurone, which when subject to a near-threshold depolarization from -50 mV gave a sustained discharge similar to that appearing with larger steps. All non-projection lamina I neurones produced sustained discharge to depolarization steps (Fig. 6B1a and B2a).

Electrical stimulation of the segmental dorsal root evoked complex inward currents in each VL-projection neurone (e.g. Fig. 6A1b and A2b). All VL-projection cells were judged to have one or more monosynaptic inputs from primary afferent fibres. In contrast, only 4/7 non-projection lamina I neurones exhibited putative monosynaptically evoked responses to DR stimulation. (One non-projection neurone was recorded in a spinal cord slice without intact segmental dorsal roots and DR-evoked activity could not be examined).

The DR-evoked responses in the VL-projection neurones had significantly shorter latencies than in non-projection neurones, implying that the former were initiated by more rapidly conducting fibres. The calculated average conduction velocity for DR fibres evoking the initial EPSCs in VL-projection neurones was 0.82 m s^-1; for non-projection neurones it was 0.35 m s^-1 (Table 2). Therefore, non-projection cells had latencies for excitation consistent with a principal monosynaptic input from the C-fibre component of the dorsal root.

Lamina II

The labelled neurones with somata in lamina II exhibited diverse dendritic arrangements and orientations. The location of the cell body in lamina II is known to be related to characteristics of the dendritic arbor (Ramon y Cajal, 1909; Sugirua, 1975; Beal & Cooper, 1978; Light et al. 1979). Our findings largely confirmed these past observations. Neurones with somata located in lamina II_o often had a relatively extensive ventral dendritic spread with processes extending to lamina II_i or into superficial portions of lamina III. In contrast, neurones with somata in lamina II_i had dendritic trees elongated in the rostro-caudal direction, and limited both in dorso-ventral and the medio-lateral dimensions.

Five lamina II morphological categories were recognized. Certain functional characteristics accompanied the morphological distinctions and in instances the latter added a unique defining feature. Because the pattern of neurite distribution has strong implications for neuronal connectivity, we gave priority to structural aspects in nomenclature. The following description begins with neuronal categories exhibiting particularly coherent features rather than the size of the group or the location of somata relative to the lamina II landmarks.

Islet neurones

Nineteen lamina II neurones had in common a dense lamina II_i dendritic tree, markedly elongated in the rostro-caudal direction, with a much more limited spread in the dorso-ventral and medio-lateral dimensions. Examples are shown in Fig. 3A1 and A2 (also see Table 1). The rostro-caudal dendritic extension was greater than for any other lamina II cells and in some instances approached that of lamina I VL-projection neurones. Cell bodies of nine of the 19 were located in lamina II_i, four were in lamina II_o and six at the border of II_i and II_o. Because the configuration of their dendrites and the location of their cell bodies were similar to neurones labelled 'islet' cells by Gobel (1975) for a category common in the substantia gelatinosa of the spinal trigeminal nucleus of the cat, we used that name.

Islet cell axons arborized extensively in the vicinity of the dendritic tree although axon branches extended considerably rostrally and caudally of the dendritic distribution. In some cases the axonal tree exceeded 1000 µm in extent. The axons had a moderate to high density of swellings and often had one or more collaterals entering lamina III and travelling ventrally for a short distance.

Islet cells had an average resting transmembrane potential (RMP) nearly 10 mV less than the mean for all lamina II neurones (Table 2). All islet cells responded to step depolarization with a sustained repetition of action potentials (Fig. 7A1a and A2a) and showed an I_h-like current (Fig. 7A1c and A2c), which averaged 60.8 pA. They also exhibited a moderate degree of inward rectification. Most islet cells (17/19) displayed spontaneous EPSCs (e.g. Fig. 7A1c) and the majority (13/19) also showed spontaneous IPSCs.

View larger version
[in this window]
[in a new window]

Figure 7. Electophysiological properties of lamina II islet (A1 and 2) and radial (B1 and 2) neurones
Recordings were obtained from neurones for which morphology is shown in Fig. 3A1 and 2 (islet cells) and Fig. 3B1 and 2 (radial cells). See the legend for Fig. 6 and Methods for description of the experimental procedures.

Dorsal root stimulation evoked monosynaptic EPSCs in 17 of 19 islet neurones (Fig. 7A1b and A2b) and only polysynaptic EPSCs in the other two islet cells. The estimated conduction velocity of the afferent fibres evoking monosynaptic EPSCs in the islet group was under 1 m s^-1 in 18/19 neurones and averaged less than 0.35 m s^-1, suggesting that input to the islet cells stemmed from C-afferent fibres.

The mean amplitude of the maximal monosynaptic EPSCs recorded from islet cells was larger (427 ± 122 pA) than the mean of those recorded from other lamina II neurones (127 ± 11 pA) and their composite maximal evoked EPSCs were larger than maximal EPSCs evoked in lamina I neurones. Particularly large evoked EPSCs were recorded from 11 of the 19 islet neurones that had monosynaptic latencies suggesting initiation by a remarkably congruent set of afferent fibres with slow conduction velocities (0.29 ± 0.02 m s^-1). The amplitude of these large EPSCs graded positively with DR stimulus intensity while response latency changed little.

Radial neurones

Seventeen cells with cell bodies located near the border between lamina II_o and II_i had dendrites that radiated in all directions. Examples of this morphology appear in Fig. 3B1 and B2. The electrophysiological characteristics for the group, while relatively consistent, were not distinctive. Because of their unique dendritic arrangement, we labelled this group 'radial' neurones. Radial cells appeared similar to those described by Bicknell & Beal (1984) as 'star shaped' or by Schoenen (1982) as 'stellate' neurones. The majority of the dendrites of radial neurones were located in lamina II_o and II_I; about one-half of the group had dendrites extending as far in the dorso-lateral direction as lamina I. The dendritic arbor was regularly very restricted in the medio-lateral plane and was usually greater in the rostro-caudal than in the dorso-ventral direction (Table 1). In some examples, visualized in parasagittal spinal cord sections, the dendritic arbor appeared to be nearly circular. Spines were present on radial cell dendrites at low to moderate density.

Radial cell axons showed few divisions in the vicinity of the dendritic tree and typically branched at the rostral or caudal levels of the dendritic arbor. The axon branches coursed longitudinally in lamina II_o (16/17 cells) and in all cases passed into II_i and the superficial portion of lamina III. All but two of the well-stained radial neurones had an axon collateral passing to the DLF that ran rostrally, caudally or in both directions, to the edge of the slice. The branch destined for the DLF usually (12/14 cells) travelled ventrally from the soma into lamina III and then laterally; these branches had fewer varicosities than those in the SDH. In one instance, the DLF axonal branch passed dorsal of the soma into the white matter.

Action potentials of radial neurones were delayed from the onset of the depolarizing step. During this delay the membrane potential slowly decreased until firing commenced. After the delay, the discharge was typically irregular or consisted of high frequency bursts of short duration action potentials (Fig. 7B1a and B2a). Voltage clamp curves routinely showed inward rectification. A small I_h-like current (24.5 ± 3.0 pA) was noted in 4/17 cells.

Dorsal root volleys evoked monosynaptic EPSCs in 14 of 17 radial cells. The calculated conduction velocity of the primary afferent fibres mediating monosynaptic EPSCs ranged widely (0.17-2.12 m s^-1). Five of the presumed monosynaptic EPSCs were initiated by activity in fibres conducting over 1 m s^-1, presumably thinly myelinated fibres. The majority of radial neurones also had polysynaptically DR-evoked EPSCs and IPSCs. Although variable, the mean frequency of spontaneous EPSCs (3.6 ± 0.44 s^-1) was relatively high (see Fig. 7B2c and Table 2).

Central cells

Ramon y Cajal (1909) described neurones commonly situated in the mid-zone of the substantia gelatinosa (lamina II) that he called 'central cells' presumably because of their location. We have used the same designation for a group of lamina II cells that matched his account.

A total of 46 of the 170 neurones were categorized as central cells. The majority of these cell bodies were located in lamina II_I and a lesser proportion were in II_o. Their dendrites generally were located in II_i so most synaptic contacts would be likely to occur in that lamina.

Central neurones had a moderately dense dendritic arbor, extended in the rostro-caudal direction, but much shorter than islet neurones (see Table 1, Fig. 3). There was limited dendritic distribution in the medio-lateral and dorso-ventral planes. Some central cells had an asymmetric dendritic arbor, one that was larger either rostral or caudal of the soma.

The axonal arborization of central cells was moderately extensive in the vicinity of the dendritic tree. Unlike islet cells, axonal arbors were not restricted to lamina II_i; branches travelled the rostral-caudal direction throughout the SDH from lamina I to II_i as well as into the superficial part of lamina III. The axonal distribution usually spread beyond the limits of the dendritic branches.

View larger version
[in this window]
[in a new window]

Figure 8. Electrophysiological properties of lamina II transient central I_A (A1 and 2) and medial-lateral (B1 and 2) neurones
The data for the medial-lateral neurones are from neurones for which morphology is displayed in Fig. 3D1 and 2. See the legend for Fig. 6 and Methods for description of the experimental procedures. Note the difference in the horizontal axis (time) in A1c, A2c and B1c from that for B2c. The holding potential in A1c, A2c and B1c was -50 mV; for B2c it was -60 mV. Although both some transient central and the medial-lateral categories exhibit a transient outward I_A-like current, they differ substantially in other physiological as well as in morphological characteristics (see Fig. 3).

View larger version
[in this window]
[in a new window]

Figure 9. Electrophysiological properties of lamina II transient central (non- I_A) (A1 and 2) and tonic central (B1 and 2) neurones
The data were obtained from the transient central (non-I_A) neurones for which morphology is shown in Fig. 3C1 and 2. See the legend for Fig. 6 and Methods for details of the experimental procedures.

While the 46 central cells shared a common dendritic configuration, they were functionally diverse. Two major groups emerged from the pattern of action potential discharge evoked by a step depolarization. Thirty-five of 46 central cells fired a few action potentials and then were silent when depolarized; these are labelled 'transient central cells' (see Figs 8A1a, 8A2a, 9A1a and 9A2a). Eleven of the 46 central cells differed by firing action potentials to the depolarizing step in a sustained or tonic fashion (Fig. 9B1a and B2a) and are designated 'tonic central cells'.

Transient central cells were further divided by the presence or absence of the potassium current, I_A, (Figs 8A1c, 8A2c, 9A1c and 9A2c) into 'transient central I_A' and 'transient central non-I_A' categories. The K⁺ channel antagonist 4-AP at 100 µM (2 cells) reduced the A-type outward current by about 30 % and and at 1 mM (3 cells) by 62 %. The time constant of the A current in the central cells was 565 ± 50.6 ms. The discharge pattern of central neurones expressing I_A could be influenced by this current; hyperpolarization after an action potential could set the mechanism for activation by a subsequent depolarization. I_A would counter depolarization, lengthening the interval to generation of the next impulse.

Most transient central cells (21/35) displayed a monosynaptically evoked EPSCs to DR stimulation. Cells with I_A showed a DR monosynaptic EPSC more commonly than those lacking I_A. Table 2 shows that for transient cells, the average conduction velocity of primary afferents evoking monosynaptic EPSCs was relatively slow (< 0.35 m s^-1), suggesting a direct C-fibre projection.

All 11 tonic central cells had DR-evoked monosynaptic EPSCs (Fig. 9B1b and B2b). The calculated afferent conduction velocity for these monosynaptic responses was more rapid (average 0.48 m s^-1) than for those of the transient central group. In five tonic central neurones, the monosynaptic EPSCs were initiated by afferent fibres conducting over 1 m s^-1. These observations suggest that at least part of the tonic central group receives excitatory input from DR A $delta$ fibres, a feature that distinguishes them from other central and from islet cells. Nine of the 11 tonic central cells had their somata and most of their dendritic arbor in lamina II_o. The latter and the more rapid conduction velocities of their DR-afferent input suggest that the tonic central group are excited by a different primary afferent population from other central lamina II neurones.

The majority of transient central cells (22/35) had prominent evoked outward currents or IPSCs (Figs 8A1b, 9A1b, 9A2b), but polysynaptic EPSCs were rare. A relatively large DR-evoked IPSC was present in a similar proportion in those with and those without I_A. In about one-half of the transient central cells with DR-evoked IPSCs (e.g. Fig. 9A1b) the outward currents were slowly rising and long-lasting. Evoked IPSCs of this form were blocked by the GABA_A receptor antagonist bicuculline (10 µM, n = 6) but not by the glycine receptor antagonist strychnine (1 µM, n = 4). A monosynaptic EPSC was often seen just before or during the early part of the long-lasting IPSCs.

Medial-lateral neurones

Three neurones differed by having a much larger dendritic span in both the the medio-lateral and the dorso-ventral planes than islet, radial or central cells (Fig. 3D1 and D2). The medial-lateral extension of their sparsely branched dendrites matched those of the horizontally orientated neurones of lamina I (Table 1), prompting the name medial-lateral neurones. In the dorso-ventral direction their dendrites distributed from lamina I to the superficial part of lamina III.

Axons of medial-lateral neurones had a sparse to moderate arborization distributed from lamina I to the superficial portion of lamina III, extending both rostrally and caudally beyond the limits of the dendritic branches.

Electrophysiological features also set these three neurones apart. Two of the three were the only neurones in the sample other than the 11 transient central cells that exhibited an I_A-like transient outward current following a hyperpolarizing step (Fig. 8B1c and B2c). This I_A-like current measured at a holding potential of -50 mV following a step to -120 mV decayed in 150 ± 51.5 ms, less than one-third of the duration of I_A by transient, central I_A neurones. Unfortunately, the limited appearance of the medial-lateral cells in the sample precluded examination of this outward current's pharmacological profile.

Medial-lateral cells fired action potentials on depolarization in a sustained, repetitive pattern (Fig. 8B1a and B2a). The tonic discharge as well as the absence of DR-evoked IPSC set them apart functionally from transient central I_A neurones. The average sEPSC frequency was relatively high in these cells, again differing from the transient central I_A cells (Table 2).

Vertical neurones

Thirty-five cells were grouped together because of a distinctive vertical orientation of their sparse wide dendrites. This group had a lesser transverse (medio-lateral) dendritic spread than medial-lateral cells (Fig. 3E1 and E2). As for most lamina II neurones, their dendritic arbors were elongated in the rostro-caudal plane. This category contained neurones morphologically similar to the 'limitroph' neurones described by Ramon y Cajal (1909) and of Gobel's (1975, 1978b) 'stalked' cells. The stalked feature was not present in all of this category leading us to designate the category 'vertical' neurones.

While the somata of the majority of vertical cells (21/35) were located in lamina II_o, 12 were positioned either in the outer part of lamina II_i or in the border between lamina II_o and II_i. In a number of instances the entire dendritic arbor was located ventral to the soma, but in other examples dendritic branches were positioned dorsally as well. These cells had dendrites distributed from lamina I (18/35 cells) to lamina III (29/35 cells). All had a significant portion of the dendritic spread in lamina II_i and most had part of the arbor as well in lamina II_o (33/35 cells). Seven cells sent dendrites deep to lamina III raising the possibility that they might represent a separate subset. The density of dendritic spines varied considerably, ranging from rare to frequent.

Axons of vertical neurones had variable arborization patterns. Most axons had branches extending rostral and caudal of the dendritic arbor. Although 15/35 cells had one collateral that arborized in lamina I, only 2/35 cells appeared to have an extensive axonal projection there. Many of the vertical neurones (24/35) gave off axon collaterals in lamina II_o and nearly the same number (23/35 cells) in lamina II_i. Eight vertical neurones had axonal branches extending deep into the dorsal horn and also into the DLF. The axon of one of these passed ventral to the cell somata into lamina III before returning laterally to enter the DLF; the axons of the other seven cells with branches destined for the DLF passed laterally along the dorsal surface of the slice. Axonal varicosities appeared in variable numbers in this category but as with other types of neurones, varicosities in branches running longitudinally in the DLF were infrequent.

None of the vertical neurones discharged action potentials transiently during a depolarizing step. Seventeen vertical neurones discharged throughout the step with a relatively constant interspike interval (Fig. 10Aa). The other 18 neurones began discharging after a delay following the start of depolarization (Fig. 10Ba); seven of these firing in a sustained and regular manner, five in a sustained but irregular pattern, and six in high frequency bursts. All vertical cells exhibited inward rectification and 12 of the 35 showed a small I_h-like current (mean amplitude 33.6 ± 4.5 pA).

F10 View larger version
[in this window]
[in a new window]

Figure 10. Electrophysiological properties of lamina II vertical neurones
The displayed data are from the neurones for which morphology is shown in Fig. 3E1 and E2. See the legend for Fig. 6 and Methods for details of the experimental procedures.

Vertical neurones had DR-evoked monosynaptic EPSCs with short latencies, suggesting a relatively rapid conduction velocity for the primary afferent input (Table 2). Monosynaptic, dorsal root-evoked EPSCs were recorded in 26/34 cells and for 10 of these the calculated conduction velocity of the primary afferent excitation was > 1 m s^-1. Therefore, vertical cells are excited by at least some primary afferent fibres conducting in the A $delta$ range and more rapidly than those synapsing on islet or central neurones. Collectively, vertical cells showed a high frequency of spontaneous EPSCs (Table 2). Spontaneous EPSCs were present at > 1 s^-1 for 34/35 vertical neurones; sEPSC frequencies as high as 31 s^-1 were noted.

The vertical neurones typically showed a prolonged DR-evoked polysynaptic EPSC whose duration increased with increasing stimulation intensity. A polysynaptic EPSC was manifest 120 ms after the stimulus in 26/34 cells, a time later than any observed monosynaptic EPSCs. In 14/26 neurones, the evoked polysynaptic EPSCs were particularly large (50-100 pA in peak amplitude). Only 6/26 vertical neurones exhibited polysynaptic IPSCs after DR stimulation and those were small in amplitude.

Unclassified lamina II neurones

About 25 % (37/157) of the lamina II cells did not fit any of the categories described above. Ten of these had somata in lamina II_i with dorsally extending dendrites; otherwise this group was morphologically and physiologically diverse.

Fifteen of the unclassified neurones fired action potentials transiently when depolarized. While this feature mimicks the transient discharge of central cells, the morphology and other features of the unclassified group differed substantially from those of the transient central categories. A morphologically diverse group of 10 neurones of the unclassified category exhibited large I_h (199 ± 28 pA), but their other features did not match those of any classified category showing I_h.

DISCUSSION

Top
Abstract
Introduction
Methods
Results
Discussion
References

This study documents a systematic diversity among the neurones that constitute the spinal SDH. This finding is consistent with past work that describes the SDH as composed of neurones with differing morphologies and a varying responsiveness to peripheral stimulation (e.g. Light et al. 1979; Beal et al. 1989; Han et al. 1998). Our analysis demonstrates that specific groups of neurones exhibit relationships between microscopic structural features and cellular functional attributes. At this juncture it is possible only to speculate about the place in the function of the neurones identified in the present work in the intact animal.

Two important features of the outermost layers of the spinal dorsal horn emerge from the present study. Firstly, at the very least, the SDH includes nine categories of neurones distinguishable by a combination of morphological and functional criteria. In these categories, morphological features such as location, cellular geometry and the distribution of neurites are paralleled by a constellation of characteristics related to connectivity and excitability. Secondly, the synaptic input to neurones of the region from primary afferent fibres is potent. Amazingly, in a preparation of the spinal cord one spinal segment or less in thickness, nearly all of the neurones sampled (161/166) exhibited responses evoked by stimulation of a segmental dorsal root, with the bulk (73 %, 122/166) displaying putative monosynaptic EPSCs.

The strong excitatory projection of segmental dorsal root fibres to SDH neurones is not surprising. Both classical and modern morphological studies have described a dense termination of primary afferent fibres in the region (Ramon y Cajal, 1909; Ranson, 1914; Earle 1952; Pearson, 1952; Szentagothai, 1964; Réthelyi & Szentágothai, 1973; Light & Perl, 1979; Bullitt et al. 1988). The robust linkage of peripheral afferent fibres to SDH neurones does emphasize the region's importance for processing and transmitting somatosensory messages.

Lamina I neurones

Our sample of neurones with somata in lamina I, the marginal zone, were divisible into two categories, those with and those without an axon passing ventromedially in the direction of the contralateral, ventrolateral column. These two groups also differed substantially in size, the VL-projection group was notably larger. The two lamina I categories also differed in three functional features: (1) the VL-projection group had a distinctly lower R_in than the non-projection category; (2) the conduction velocity of DR fibres initiating the shortest latency, monosynaptic response was significantly higher (mean of 0.82 m s^-1) for the VL-projection category than for other lamina I neurones (mean of 0.35 m s^-1), suggesting that some VL-projection cells had direct A $delta$ dorsal root input while the non-projection neurones primarily received monosynaptic C-fibre connections; (3) typically the frequency of spontaneous EPSCs was considerably higher in the non-projection class than in the VL-projection group.

Previous investigations have described between one and four types of neurones in lamina I, differentiated mainly morphologically (Ramon y Cajal, 1909; Gobel, 1978a; Schoenen, 1982; Beal et al. 1989; Han et al. 1998). Distinctions between accounts in these earlier reports may be attributable to differing species, varying criteria and ages of the animals. It appears certain that lamina I contains more than one type of neurone, a point supported by our observations on the hamster. It seems reasonable to propose that neurones, with a relatively thick axon coursing in the direction of the opposite ventralateral white matter, contribute to crossed spinoreticular or spinothalamic pathways and, therefore, convey activity important for certain somatasensory functions (Light, 1992). Our five lamina I projection cells were not separable on morphological grounds; only one of the five differed from the others by firing bursts of action potentials at high frequency upon depolarization rather than producing a sustained, regular discharge. A larger sample is needed to evaluate the possibility that different lamina I cells contributing to the contralateral, lateral ascending tracts in the rodent have functional relationships to morphological configuration mimicking the arrangements proposed by Han et al. (1998) for the cat and other species.

Lamina II neurones

Our sample established five morphological categories of lamina II neurones: islet, central, medial-lateral, radial, and vertical. Central cells were further divisible into three subtypes on the basis of functional differences, suggesting the hamster lamina II neuronal constituents to number at least seven categories. Certain lamina II groupings resemble previously identified varieties. Our islet neurones appear morphologically equivalent to those given the same name in the cat, monkey, rat and human being (Gobel, 1975, 1978b; Bennett et al. 1979; 1980; Price et al. 1979; Schoenen, 1982; Beal, 1983; Bicknell & Beal, 1984; Steedman et al. 1985; Todd & Lewis, 1986; Beal et al. 1989). They represent a category that Gobel (1978b) argued, on structural grounds, to be inhibitory interneurones. Neurones with features of islet cell morphology have been reported to contain GABA and glycine (Todd & McKenzie, 1989; Todd & Sullivan, 1990), a finding consistent with an inhibitory function. It is interesting that islet neurones had lower resting membrane potentials than any other SDH neurones. While this latter feature was distinctive in our sample, it might prove a difficult criterion to use for identifying a neurone during an electrophysiological recording.

The most common morphological type in our lamina II sample, central neurones, had a pronounced rostro-caudal orientation similar to Beal's (1989) 'short islet' cells. However, the group we called 'islet' differed electrophysiologically from the smaller central cells by a distinctively lower resting membrane potential, by the large amplitude of DR-evoked EPSCs, and by the strikingly greater rostro-caudal extent of their dendritic trees. Central cells that fire transiently when depolarized often had large IPSCs evoked by primary afferent input. Both the inherent transient discharge of such cells and such IPSCs could be expected to limit their duration of discharge. From the arrangement of dendrites and axonal distributions, it seems likely that the three varieties of central neurones play distinctive parts in integration of activity between adjacent parts of lamina II, although the nature of those roles remains to be established.

Radial neurones appear comparable to (a) 'star-shaped' neurones described by Bicknell & Beal (1984), (b) a group of cells described by Todd & Lewis (1986) that represented a significant (15 %) part of their lamina II population and (c) Schoenen's (1982) 'stellate' category that comprised a major part of his sample of human lamina II neurones. Thus, as the islet and central categories, radial neurones appear to be a fixture in the organization of lamina II, but one likely to serve different functions from islet and central cells.

Morphologically, our vertical class resembles Ramon y Cajal's limotroph neurones (1909), the stalked and arboreal cells described by Gobel (1975, 1978b) and neurones also called vertical by Beal (1983). This group varied considerably in action potential firing pattern and in details of their morphology; however, we were unable to establish features that justified further subdivision. In addition to the dorso-ventral orientation of dendritic trees, vertical neurones generally had predominantly excitatory EPSCs mediated by relatively rapidly conducting DR afferent fibres, and a relatively high frequency of sEPSCs. As a group, vertical neurones tended to exhibit DR-evoked, long-lasting polysynaptic EPSCs, to show few IPSCs and to fire action potentials tonically until the end of a depolarizing step. The latter properties can be expected to produce prolonged activity following activation by primary afferent fibres, suitable for excitatory interneuronal function. Gobel (1978b) proposed 'stalked' cells to be excitatory interneurones projecting from lamina II to lamina I because he observed axonal projections from them to lamina I. Such stalked neurones were included in our vertical category; however, we failed to find a significant number of vertical-type neurones with an axonal arborization directed largely to lamina I.

The medial-lateral neurones defy the general rostral-caudal orientation of lamina II and in this sense mimic Ramon y Cajal's transverse neurones of the chicken dorsal horn (1909). It is conceivable that such transversely arranged cells link together groups of neurones of lamina II whose input is organized according to body region.

General comments

Some lamina I (3/13) and lamina II (69/157) neurones had an axon collateral travelling rostrally and/or caudally in Lissauer's tract. Axons of lamina II cells in Lissauer's tract have long been recognized to extend rostrally or caudally one or more segments (Ramon y Cajal, 1909; Ransom, 1914; Earle, 1952; Pearson, 1952; Szentágothai, 1964). Such propriospinal connections are probably involved in modulating responses of neurones in adjacent segments and may be related to integrating afferent inputs from neighbouring areas of the peripheral body.

Heterogeneity in electrophysiological properties among the various SDH neurones is no surprise. Previous work on SDH neurones has shown that action potential firing patterns vary (Thomson et al. 1989; Lopez-Garcia & King, 1994) and I_h and I_A currents are expressed by part of the population (Yoshimura & Jessell, 1989b). To a considerable extent, we found these features and differences related to other physiological or structural characteristics; however, some electrophysiological features are unlikely, in and of themselves, to distinguish neurones of the SDH. For instance, islet cells, one class of central cells and some vertical cells all discharge tonically when depolarized. Yet, these neurones are quite diverse in morphology and other electrophysiological properties. In another case, many SDH neurones, belonging to otherwise quite different categories, possess I_h.

Nonetheless, functional features can be determined while a neurone is living. Is it possible to predict the category of a particular neurone at the time of recording? A reasonable answer to that question is, perhaps. If, in a tight-seal, whole-cell recording from a location near the outermost portion of the hamster SDH, a particular neurone had an input resistance of under 200 M $Omega$ , a stable resting potential of -40 mV or greater, a robust DR-evoked EPSC at a latency suggesting an input from fibres conducting more rapidly than 0.7 m s^-1, there would be a good possibility that the recording was from a VL-projection type neurone. A recording from a similar locus on a neurone with an input resistance of over 500 M $Omega$ and an sEPSC frequency over 4 Hz would, in turn, suggest a non-projection lamina I cell. Lamina II predictions could be more difficult. Recording from a neurone that gave a transient response to threshold depolarization and that exhibited I_A would make it likely that the recording was from a transient I_A central cell. On the other hand, a recording from a neurone that gave a tonic response to near-threshold depolarization and showed an I_A type of outward current would suggest the medial-lateral class of neurone. Finally, a recording from a lamina II location with a DR-evoked EPSC suggestive of input from dorsal root fibres conducting close to 1 m s^-1 and exhibiting sEPSCs at a frequency in excess of 5 Hz could be a vertical neurone.

To reiterate, our observations emphasize that SDH is a highly complex region in which the preponderence of neurones receive strong input directly and indirectly from thin primary afferent fibres. The richly intertwined arrangements of dendrites and axons of many neural components of the region implies substantial processing and modulation of the incoming activity. The number of classes of neurones differentiated by structural and functional attributes suggest strongly that the region subserves a diverse range of functions that include nociception and pain, temperature sense, itch and possibly other somatosensory functions (Kumazawa & Perl, 1978; Light et al. 1979; Craig, 1998; Han et al. 1998; Andrew & Craig, 2001). Better insight into the functional interaction between the neural elements clearly is needed to understand the part played by this region in sensory activity related to the body.

REFERENCES

Top
Abstract
Introduction
Methods
Results
Discussion
References

ABDEL-MAGUID, T. E. & BOWSHER, D. (1984). Classification of neurons by dendritic branching pattern. A categorization based on Golgi impregnation of spinal and cranial somatic afferent and efferent cells in the adult human. Journal of Anatomy 138, 689-702 [Medline]

ABDEL-MAGUID, T. E. & BOWSHER, D. (1985). The gray matter of the dorsal horn of the adult human spinal cord, including comparisons with general somatic and visceral afferent cranial nerve nuclei. Journal of Anatomy 142, 33-58

ANDREW, D. & CRAIG, A. D. (2001). Spinothalamic lamina I neurons selectively sensitive to histamine: a central neural pathway for itch. Nature Neuroscience 4, 72-77 [Medline]

BAO, J., LI, J. & PERL, E. R. (1996). Primary afferent inhibition of spinal laminae I and II neurons. Society for Neuroscience Abstracts 22, 878

BAO, J., LI, J. & PERL, E. R. (1998). Differences in Ca²⁺ channels governing generation of miniature and evoked excitatory synaptic currents in spinal laminae I and II. Journal of Neuroscience 18, 8740-8750 [Abstract/Full Text]

BEAL, J. A. (1979). The ventral dendritic arbor of marginal (lamina I) neurons in the adult primate spinal cord. Neuroscience Letters 14, 201-206 [Medline]

BEAL, J. A. (1983). Identification of presumptive long axon neurons in the substantia gelatinosa of the rat lumbosacral spinal cord: A Golgi study. Neuroscience Letters 41, 9-14 [Medline]

BEAL, J. A. & COOPER, M. H. (1978). The neurons in the gelatinosal complex (laminae II-III) of the monkey (Macaca mulatta): A Golgi study. Journal of Comparative Neurology 179, 89-122 [Medline]

BEAL, J. A., NANDIA, K. N. & KNIGHT, D. S. (1989). Characterization of long ascending tract projection neurons and non-tract neurons in the superficial dorsal horn (SDH). In Processing of Sensory Information in the Superficial Dorsal Horn of the Spinal Cord, ed. CERVERO, F., BENNETT, G. J. & HEADLEY, P. M., pp. 181-197. Plenum, New York

BENNETT, G. J., ABDELMOUMENE, M., HAYASHI, H. & DUBNER, R. (1980). Physiology and morphology of substantia gelatinosa neurons intracellularly stained with horseradish peroxidase. Journal of Comparative Neurology 194, 809-827 [Medline]

BENNETT, G. J., HAYASHI, H., ABDELMOUMENE, M. & DUBNER, R. (1979). Physiological properties of stalked cells of the substantia gelatinosa intracellularly stained with horseradish peroxidase. Brain Research 164, 285-289 [Medline]

BICKNELL, H. R. & BEAL, J. A. (1984). Axonal and dendritic development of substantia gelatinosa neurons in the lumbosacral spinal cord of the rat. Journal of Comparative Neurology 226, 508-522 [Medline]

BOWSHER, D. & ABDEL-MAGUID, T. E. (1984). Superficial dorsal horn of the adult human spinal cord. Neurosurgery 15, 893-899 [Medline]

BROWN, A. G. & RÉTHELYI, M. (1981). Spinal Cord Sensation-Sensory Processing in the Dorsal Horn. Scottish Academic Press, Edinburgh

BULLITT, E., STOFER, W. D., VIERCK, C. J. & PERL, E. R. (1988). Reorganization of primary afferent terminals in the spinal dorsal horn of the primate caudal to antereolateral chordotomy. Journal of Comparative Neurology 270, 549-558 [Medline]

CHRISTENSEN, B. N. & PERL, E. R. (1970). Spinal neurons specifically excited by noxious or thermal stimuli: marginal zone of the dorsal horn. Journal of Neurophysiology 33, 293-307 [Medline]

CONNER, J. A. & STEVENS, C. F. (1971). Voltage clamp studies of a transient outward membrane current in gastropod neural somata. Journal of Physiology 213, 21-30 [Medline]

CRAIG, A. D. (1998). A new version of the thalamic disinhibition hypothesis of central pain. Pain Forum 7, 1-14

EARLE, K. M. (1952). The tract of Lissauer and its possible relation to the pain pathway. Journal of Comparative Neurology 96, 93-111

GOBEL, S. (1975). Golgi studies of the substantia gelatinosa neurons in the spinal trigeminal nucleus. Journal of Comparative Neurology 162, 397-416 [Medline]

GOBEL, S. (1978a). Golgi studies of the neurons in layer I of the dorsal horn of the medulla (trigeminal nucleus caudalis). Journal of Comparative Neurology 180, 375-393 [Medline]

GOBEL, S. (1978b). Golgi studies of the neurons in layer II of the dorsal horn of the medulla (trigeminal nucleus caudalis). Journal of Comparative Neurology 180, 395-414 [Medline]

GRUDT, T. J. & PERL, E. R. (1998). Neurons of the superficial dorsal horn of the spinal cord from electrophysiologically and morphologically heterogeneous groups. Society for Neuroscience Abstracts 24, 2079

GRUDT, T. J. & PERL, E. R. (1999). Correlations of morphological and physiological features for neurons in the superficial spinal dorsal horn. Society for Neuroscience Abstracts 25, 148

HAN, Z.-S., ZHANG, E.-T. & CRAIG, A. D. (1998). Nociceptive and thermoreceptive lamina I neurons are anatomically distinct. Nature Neuroscience 1, 218-225 [Medline]

KUMAZAWA, T. & PERL, E. R. (1978). Excitation of marginal and substantia gelatinosa neurons in the primate spinal cord: indications of their place in dorsal horn functional organization. Journal of Comparative Neurology 177, 417-434 [Medline]

LAWSON, S. N., CREPPS, B. A. & PERL, E. R. (1997). Relationship of substance P to afferent characteristics of dorsal root ganglion neurons in guinea pig. Journal of Physiology 505, 177-191 [Abstract]

LI, J. & PERL, E. R. (1994). Adenosine inhibition of synaptic transmission in the substantia gelatinosa. Journal of Neurophysiology. 72, 1611-1621 [Medline]

LIGHT, A. R. (1992). In The Initial Processing of Pain and Its Descending Control: Spinal and Trigeminal Systems, ed. GILDENBERG, P. H. Karger, Basel, Switzerland

LIGHT, A. R. & PERL, E. R. (1979). Re-examination of the dorsal root projection to the spinal dorsal horn including observations on the differential termination of coarse and fine fibers. Journal of Comparative Neurology 186, 117-132 [Medline]

LIGHT, A. R., TREVINO, D. L. & PERL, E. R. (1979). Morphological features of functionally defined neurons in the marginal zone and substantia gelatinosa of the spinal dorsal horn. Journal of Comparative Neurology 186, 151-172 [Medline]

LIMA, D. & COIMBRA, A. (1986). A Golgi study of the neuronal population of the marginal zone (lamina I) of the rat spinal cord. Journal of Comparative Neurology 244, 53-71 [Medline]

LOPEZ-GARCIA, J. A. & KING, A. E. (1994). Membrane properties of physiologically classified rat dorsal horn neurons in vitro: Correlation with cutaneous sensory afferent input. European Journal of Neuroscience 6, 998-1007 [Medline]

LÜTHI, A. & MCCORMICK, D. (1998). H-Current: Properties of a neuronal and network pacemaker. Neuron 21, 9-12 [Medline]

MANNEN, H. & SUGIURA, Y. (1976). Reconstruction of neurons of dorsal horn proper using Golgi-stained serial sections. Journal of Comparative Neurology 168, 303-312 [Medline]

MATSUSHITA, M. (1969). Some aspects of the interneuronal connections in cat's spinal gray matter. Journal of Comparative Neurology 136, 57-80 [Medline]

MAYER, M. L. & WESTBROOK, G. L. (1983). A voltage clamp analysis of inward (anomalous) rectification in mouse spinal sensory ganglion neurones. Journal of Physiology 340, 19-45 [Abstract]

PARK, J.-S., NAKATSUKA, T., NAGATA, K., HIGASHI, H. & YOSHIMURA, M. (1999). Reorganization of the primary afferent termination in the rat spinal dorsal horn during post-natal development. Developmental Brain Research 113, 29-36 [Medline]

PEARSON, A. A. (1952). Role of gelatinous substance of spinal cord in conduction of pain. Archives of Neurology and Psychiatry 68, 515-529

PERL, E. R. (1984). Pain and nociception. In Handbook of Physiology, section 1, The Nervous System, vol. III, Sensory Processes, ed. DARIAN-SMITH, I., pp. 915-975. American Physiological Society, Bethesda, USA

PRICE, D. D., HAYASHI, H., DUBNER, R. & RUDA, M. A. (1979). Functional relationships between neurons of marginal and substantia gelatinosa layers of primate dorsal horn. Journal of Neurophysiology 42, 1590-1608 [Medline]

RAMON Y CAJAL, S. (1909). Histologie du Système Nerveux de l'Homme et des Vertébrés, vol. 1. Maloine, Paris

RANSON, S. W. (1914). The tract of Lissauer and the substantia gelatinosa Rolandi. American Journal of Anatomy 16, 97-126

RÉTHELYI, M. & SZENTÁGOTHAI, J. (1973). Distribution and connections of afferent fibres in the spinal cord. In Handbook of Sensory Physiology. Somatosensory System, ed. IGGO, A., pp. 207-252. Springer-Verlag, Berlin

RÉTHELYI, M., LIGHT, A. R. & PERL, E. R. (1989). Synaptic ultrastructure of functionally and morphologically characterized neurons of the superficial dorsal horn of cat. Journal of Neuroscience 9, 1846-1863 [Abstract]

REXED, B. (1952). The cytoarchitectonic organization of the spinal cord in the cat. Journal of Comparative Neurology 96, 415-466

RUDY, B. (1988). Diversity and ubiquity of potassium channels. Neuroscience 25, 729-749 [Medline]

SCHEIBEL, M. E. & SCHEIBEL, A. B. (1968). Terminal axonal patterns in the cat spinal cord. Brain Research 9, 32-58 [Medline]

SCHOENEN, J. (1982). The dendritic organization of the human spinal cord: The dorsal horn. Neuroscience 7, 2057-2087 [Medline]

SINDOU, M., QUOEX, C. & BALEYDIER, C. (1974). Fiber organization at the posterior spinal cord-rootlet junction in man. Journal of Comparative Neurology 153, 15-26 [Medline]

SNYDER, R. (1977). The organization of the dorsal root entry zone in cats and monkeys. Journal of Comparative Neurology 174, 47-69 [Medline]

STEEDMAN, W. M., MOLONY, V. & IGGO, A. (1985). Nociceptive neurones in the superficial dorsal horn of cat lumbar spinal cord and their primary afferent inputs. Experimental Brain Research 58, 171-182 [Medline]

SUGIURA, Y. (1975). Three dimensional analysis of neurons in the substantia gelatinosa Rolandi. Proceedings of Japan Academy 51, 336-341

SZENTÁGOTHAI, J. (1964). Neuronal and synaptic arrangement of substantia gelatinosa Rolandi. Journal of Comparative Neurology 122, 219-239

THOMSON, A. M., WEST, D. C. & HEADLEY, P.

ABDEL-MAGUID, T. E. & BOWSHER, D. (1984). Classification of neurons by dendritic branching pattern. A categorization based on Golgi impregnation of spinal and cranial somatic afferent and efferent cells in the adult human. Journal of Anatomy 138, 689-702	[Medline]
ABDEL-MAGUID, T. E. & BOWSHER, D. (1985). The gray matter of the dorsal horn of the adult human spinal cord, including comparisons with general somatic and visceral afferent cranial nerve nuclei. Journal of Anatomy 142, 33-58
ANDREW, D. & CRAIG, A. D. (2001). Spinothalamic lamina I neurons selectively sensitive to histamine: a central neural pathway for itch. Nature Neuroscience 4, 72-77	[Medline]
BAO, J., LI, J. & PERL, E. R. (1996). Primary afferent inhibition of spinal laminae I and II neurons. Society for Neuroscience Abstracts 22, 878
BAO, J., LI, J. & PERL, E. R. (1998). Differences in Ca²⁺ channels governing generation of miniature and evoked excitatory synaptic currents in spinal laminae I and II. Journal of Neuroscience 18, 8740-8750	[Abstract/Full Text]
BEAL, J. A. (1979). The ventral dendritic arbor of marginal (lamina I) neurons in the adult primate spinal cord. Neuroscience Letters 14, 201-206	[Medline]
BEAL, J. A. (1983). Identification of presumptive long axon neurons in the substantia gelatinosa of the rat lumbosacral spinal cord: A Golgi study. Neuroscience Letters 41, 9-14	[Medline]
BEAL, J. A. & COOPER, M. H. (1978). The neurons in the gelatinosal complex (laminae II-III) of the monkey (Macaca mulatta): A Golgi study. Journal of Comparative Neurology 179, 89-122	[Medline]
BEAL, J. A., NANDIA, K. N. & KNIGHT, D. S. (1989). Characterization of long ascending tract projection neurons and non-tract neurons in the superficial dorsal horn (SDH). In Processing of Sensory Information in the Superficial Dorsal Horn of the Spinal Cord, ed. CERVERO, F., BENNETT, G. J. & HEADLEY, P. M., pp. 181-197. Plenum, New York
BENNETT, G. J., ABDELMOUMENE, M., HAYASHI, H. & DUBNER, R. (1980). Physiology and morphology of substantia gelatinosa neurons intracellularly stained with horseradish peroxidase. Journal of Comparative Neurology 194, 809-827	[Medline]
BENNETT, G. J., HAYASHI, H., ABDELMOUMENE, M. & DUBNER, R. (1979). Physiological properties of stalked cells of the substantia gelatinosa intracellularly stained with horseradish peroxidase. Brain Research 164, 285-289	[Medline]
BICKNELL, H. R. & BEAL, J. A. (1984). Axonal and dendritic development of substantia gelatinosa neurons in the lumbosacral spinal cord of the rat. Journal of Comparative Neurology 226, 508-522	[Medline]
BOWSHER, D. & ABDEL-MAGUID, T. E. (1984). Superficial dorsal horn of the adult human spinal cord. Neurosurgery 15, 893-899	[Medline]
BROWN, A. G. & RÉTHELYI, M. (1981). Spinal Cord Sensation-Sensory Processing in the Dorsal Horn. Scottish Academic Press, Edinburgh
BULLITT, E., STOFER, W. D., VIERCK, C. J. & PERL, E. R. (1988). Reorganization of primary afferent terminals in the spinal dorsal horn of the primate caudal to antereolateral chordotomy. Journal of Comparative Neurology 270, 549-558	[Medline]
CHRISTENSEN, B. N. & PERL, E. R. (1970). Spinal neurons specifically excited by noxious or thermal stimuli: marginal zone of the dorsal horn. Journal of Neurophysiology 33, 293-307	[Medline]
CONNER, J. A. & STEVENS, C. F. (1971). Voltage clamp studies of a transient outward membrane current in gastropod neural somata. Journal of Physiology 213, 21-30	[Medline]
CRAIG, A. D. (1998). A new version of the thalamic disinhibition hypothesis of central pain. Pain Forum 7, 1-14
EARLE, K. M. (1952). The tract of Lissauer and its possible relation to the pain pathway. Journal of Comparative Neurology 96, 93-111
GOBEL, S. (1975). Golgi studies of the substantia gelatinosa neurons in the spinal trigeminal nucleus. Journal of Comparative Neurology 162, 397-416	[Medline]
GOBEL, S. (1978a). Golgi studies of the neurons in layer I of the dorsal horn of the medulla (trigeminal nucleus caudalis). Journal of Comparative Neurology 180, 375-393	[Medline]
GOBEL, S. (1978b). Golgi studies of the neurons in layer II of the dorsal horn of the medulla (trigeminal nucleus caudalis). Journal of Comparative Neurology 180, 395-414	[Medline]
GRUDT, T. J. & PERL, E. R. (1998). Neurons of the superficial dorsal horn of the spinal cord from electrophysiologically and morphologically heterogeneous groups. Society for Neuroscience Abstracts 24, 2079
GRUDT, T. J. & PERL, E. R. (1999). Correlations of morphological and physiological features for neurons in the superficial spinal dorsal horn. Society for Neuroscience Abstracts 25, 148
HAN, Z.-S., ZHANG, E.-T. & CRAIG, A. D. (1998). Nociceptive and thermoreceptive lamina I neurons are anatomically distinct. Nature Neuroscience 1, 218-225	[Medline]
KUMAZAWA, T. & PERL, E. R. (1978). Excitation of marginal and substantia gelatinosa neurons in the primate spinal cord: indications of their place in dorsal horn functional organization. Journal of Comparative Neurology 177, 417-434	[Medline]
LAWSON, S. N., CREPPS, B. A. & PERL, E. R. (1997). Relationship of substance P to afferent characteristics of dorsal root ganglion neurons in guinea pig. Journal of Physiology 505, 177-191	[Abstract]
LI, J. & PERL, E. R. (1994). Adenosine inhibition of synaptic transmission in the substantia gelatinosa. Journal of Neurophysiology. 72, 1611-1621	[Medline]
LIGHT, A. R. (1992). In The Initial Processing of Pain and Its Descending Control: Spinal and Trigeminal Systems, ed. GILDENBERG, P. H. Karger, Basel, Switzerland
LIGHT, A. R. & PERL, E. R. (1979). Re-examination of the dorsal root projection to the spinal dorsal horn including observations on the differential termination of coarse and fine fibers. Journal of Comparative Neurology 186, 117-132	[Medline]
LIGHT, A. R., TREVINO, D. L. & PERL, E. R. (1979). Morphological features of functionally defined neurons in the marginal zone and substantia gelatinosa of the spinal dorsal horn. Journal of Comparative Neurology 186, 151-172	[Medline]
LIMA, D. & COIMBRA, A. (1986). A Golgi study of the neuronal population of the marginal zone (lamina I) of the rat spinal cord. Journal of Comparative Neurology 244, 53-71	[Medline]
LOPEZ-GARCIA, J. A. & KING, A. E. (1994). Membrane properties of physiologically classified rat dorsal horn neurons in vitro: Correlation with cutaneous sensory afferent input. European Journal of Neuroscience 6, 998-1007	[Medline]
LÜTHI, A. & MCCORMICK, D. (1998). H-Current: Properties of a neuronal and network pacemaker. Neuron 21, 9-12	[Medline]
MANNEN, H. & SUGIURA, Y. (1976). Reconstruction of neurons of dorsal horn proper using Golgi-stained serial sections. Journal of Comparative Neurology 168, 303-312	[Medline]
MATSUSHITA, M. (1969). Some aspects of the interneuronal connections in cat's spinal gray matter. Journal of Comparative Neurology 136, 57-80	[Medline]
MAYER, M. L. & WESTBROOK, G. L. (1983). A voltage clamp analysis of inward (anomalous) rectification in mouse spinal sensory ganglion neurones. Journal of Physiology 340, 19-45	[Abstract]
PARK, J.-S., NAKATSUKA, T., NAGATA, K., HIGASHI, H. & YOSHIMURA, M. (1999). Reorganization of the primary afferent termination in the rat spinal dorsal horn during post-natal development. Developmental Brain Research 113, 29-36	[Medline]
PEARSON, A. A. (1952). Role of gelatinous substance of spinal cord in conduction of pain. Archives of Neurology and Psychiatry 68, 515-529
PERL, E. R. (1984). Pain and nociception. In Handbook of Physiology, section 1, The Nervous System, vol. III, Sensory Processes, ed. DARIAN-SMITH, I., pp. 915-975. American Physiological Society, Bethesda, USA
PRICE, D. D., HAYASHI, H., DUBNER, R. & RUDA, M. A. (1979). Functional relationships between neurons of marginal and substantia gelatinosa layers of primate dorsal horn. Journal of Neurophysiology 42, 1590-1608	[Medline]
RAMON Y CAJAL, S. (1909). Histologie du Système Nerveux de l'Homme et des Vertébrés, vol. 1. Maloine, Paris
RANSON, S. W. (1914). The tract of Lissauer and the substantia gelatinosa Rolandi. American Journal of Anatomy 16, 97-126
RÉTHELYI, M. & SZENTÁGOTHAI, J. (1973). Distribution and connections of afferent fibres in the spinal cord. In Handbook of Sensory Physiology. Somatosensory System, ed. IGGO, A., pp. 207-252. Springer-Verlag, Berlin
RÉTHELYI, M., LIGHT, A. R. & PERL, E. R. (1989). Synaptic ultrastructure of functionally and morphologically characterized neurons of the superficial dorsal horn of cat. Journal of Neuroscience 9, 1846-1863	[Abstract]
REXED, B. (1952). The cytoarchitectonic organization of the spinal cord in the cat. Journal of Comparative Neurology 96, 415-466
RUDY, B. (1988). Diversity and ubiquity of potassium channels. Neuroscience 25, 729-749	[Medline]
SCHEIBEL, M. E. & SCHEIBEL, A. B. (1968). Terminal axonal patterns in the cat spinal cord. Brain Research 9, 32-58	[Medline]
SCHOENEN, J. (1982). The dendritic organization of the human spinal cord: The dorsal horn. Neuroscience 7, 2057-2087	[Medline]
SINDOU, M., QUOEX, C. & BALEYDIER, C. (1974). Fiber organization at the posterior spinal cord-rootlet junction in man. Journal of Comparative Neurology 153, 15-26	[Medline]
SNYDER, R. (1977). The organization of the dorsal root entry zone in cats and monkeys. Journal of Comparative Neurology 174, 47-69	[Medline]
STEEDMAN, W. M., MOLONY, V. & IGGO, A. (1985). Nociceptive neurones in the superficial dorsal horn of cat lumbar spinal cord and their primary afferent inputs. Experimental Brain Research 58, 171-182	[Medline]
SUGIURA, Y. (1975). Three dimensional analysis of neurons in the substantia gelatinosa Rolandi. Proceedings of Japan Academy 51, 336-341
SZENTÁGOTHAI, J. (1964). Neuronal and synaptic arrangement of substantia gelatinosa Rolandi. Journal of Comparative Neurology 122, 219-239
THOMSON, A. M., WEST, D. C. & HEADLEY, P.