Relationship of membrane properties, spike burst responses, laminar location, and functional class of dorsal horn neurons recorded in vitro

This work shows that grouping spinal neurons by response to afferent input provides the strongest correlation to both passive and active membrane properties.

Abstract

The input-output and discharge properties of neurons are shaped by both passive and active electrophysiological membrane properties. Whole cell patch-clamp recordings in lamina I–III neurons in an isolated preparation of the whole spinal cord of juvenile rats with attached dorsal roots and dorsal root ganglia were used to further define which of these properties provides the most impactful classification strategy. A total of 95 neurons were recorded in segment L5 and were classified based on the responses to L4 dorsal root stimulation. The results showed that high-threshold and silent neurons had higher membrane resistance and more negative resting membrane potential than low-threshold or wide-dynamic-range neurons. Rheobase in low-threshold and wide-dynamic-range neurons was significantly lower than that of high-threshold or silent neurons. Four types of firing patterns were identified in response to depolarizing current injections. Low-threshold cells most frequently showed a phasic firing pattern characterized by a short initial burst of action potentials, single spiking or irregular firing bursts at the onset of a depolarizing pulse. High-threshold and wide-dynamic-range neurons were characterized by tonic firing with trains of spikes occurring at regular intervals throughout the current pulse. The majority of silent neurons displayed a delayed onset of firing in response to current injection. These results indicate that the passive membrane properties of spinal neurons are tuned to optimize the responses to particular subsets of afferent stimuli.

NEW & NOTEWORTHY

This work shows that grouping spinal neurons by response to afferent input provides the strongest correlation to both passive and active membrane properties.

the spinal dorsal horn is the first site for the processing and integration of sensory information from the skin, deep tissues, and viscera and, as such, greatly influences somatosensation and nociception. Yet, the fundamental organization of the dorsal horn remains incompletely defined. A major stumbling block has been the struggle to adequately define the various types of neurons that are present and their specific functions. Several classification schemes have been used to categorize spinal neurons. Anatomic location within the dorsal to ventrally organized layers and morphological features have been used as one approach (Grudt and Perl 2002; Ruscheweyh and Sandkuhler 2002). Alternatively, the functional responses of cells to afferent inputs or to depolarizing current injection have also been used (Grudt and Perl 2002; Lopez-Garcia and King 1994; Prescott and DeKoninck 2002). More recently, studies have focused on the properties of spinal neurons in mice expressing fluorescent markers at specific genetic loci related to neurotransmitter content using slice preparations (Hughes et al. 2012, 2013; Punnakkal et al. 2014; Zhang et al. 2010, 2013). Although some basic organizing principles have emerged from this work, there remains a need to synthesize the findings from these various approaches.

Studies of spinal neurons in intact anesthetized animals have traditionally used a categorization system based on the responses to natural cutaneous stimuli. Low-threshold (LT) neurons were those that responded only to innocuous stimuli, high-threshold (HT) neurons [also often termed nociceptive-specific (NS) neurons] were those responded only to noxious stimuli, and wide-dynamic-range (WDR) neurons were those that showed graded responses to both innocuous and noxious stimuli. Specific pain signaling functions have been suggested for subsets of spinal neurons based on this classification system. Subsets of lamina I NS spinothalamic neurons signal in response to acute thermal (Christensen and Perl 1970; Kumazawa and Perl 1978) and cooling stimuli in a fashion that parallels psychophysical reports to these stimuli (Han et al. 1998). Differences in activity between these subsets of lamina I NS cells correlate with the perception of burning from noxious cold (Craig and Bushnell 1994) and provide a reasonable physiological correlate to the burning observed in central pain syndrome (Craig 1998). Meanwhile, lamina III–V WDR spinothalamic neurons appear to encode for the discrimination of acute noxious from non-noxious mechanical stimuli (Dubner et al. 1989; Willis 1993), the increased pain to mechanical stimuli in zones of secondary hyperalgesia (Dougherty and Willis 1992; Simone et al. 1991) and acute and chronic anginal pain (Blair et al. 1984).

Data from studies using in vivo whole cell techniques have shown that NS and silent neurons have passive membrane properties that would require higher amounts of synaptic current to drive the cells toward spike threshold than would WDR or LT neurons (Weng and Dougherty 2002). Thus, NS and silent neurons would be more likely to passively filter inputs from many primary afferents, whereas WDR or LT neurons would be more likely follow the synaptic inputs impinging upon them. Additionally, WDR and LT neurons have passive membrane properties that would support a more rapid activation of subliminal excitatory receptive fields. The general implication of these data is that the response properties of spinal cells are tuned by their passive biophysical membrane properties.

Yet, as noted above, several differing classification schemes have affected the extent to which the many whole cell studies can be better integrated (Grudt and Perl 2002; Lopez-Garcia and King 1994; Prescott and DeKoninck 2002; Ruscheweyh and Sandkuhler 2002). The strongest correlations between physiological properties of cells have been based on the responses to intracellular current, laminar location, and based on anatomic or neurochemical classification of cells. The most commonly used classification scheme has been based on the action potential (AP) responses of neurons to the injection of depolarizing current have included cells showing single spikes, phasic self-limiting bursts, tonic trains, and delayed onset spikes (Grudt and Perl 2002; Lopez-Garcia and King 1994; Prescott and DeKoninck 2002; Ruscheweyh and Sandkuhler 2002; Schneider 2003; Szûcs et al. 2003; Thomson et al. 1989). The goal in the present study was to use an in vitro preparation of the whole spinal cord with dorsal roots attached wherein each of these schemes could be assessed and compared for that which yielded the most robust correlations with active and passive membrane properties.

MATERIALS AND METHODS

Experiments were performed on 79 Sprague-Dawley rats at 15–21 days of age of both sexes. Pups were housed with their dam in standard polypropylene cages until the day of the experiments. All experiments were conducted with approval of the Institutional Animal Care and Use Committee at the M.D. Anderson Cancer Center and were in complete compliance with National Institutes of Health guidelines for the care and use of laboratory animals.

Preparation of the isolated spinal cord with attached dorsal roots.

Neurons were recorded in vitro from an isolated whole spinal cord using blind patch with attached dorsal roots and dorsal root ganglia (DRG) (Bagust et al. 1985; Chen et al. 2004). Rats were decapitated while under deep isoflurane anesthesia, and a block of tissue containing the vertebral column was rapidly excised and put into ice-cold (2–3°C) artificial cerebrospinal fluid (ACSF). The spinal cord was rapidly dissected together with the attached dorsal and ventral roots and DRG. The ventral roots were dissected free at the junction to the DRG, and the capsule of the DRG was cut with fine scissors. The intact spinal cord was transferred to a recording chamber, secured in place with the dorsal surface up, and continuously perfused with oxygenated (95% O₂ and 5% CO₂) ACSF at room temperature (22°C) at 30 ml/min. The composition of the ACSF was (in mM) 120 NaCl, 3 KCl, 25 NaHCO₃, 2.5 CaCl₂, 0.5 MgCl₂, and 12 glucose (pH 7.4). A period of at least 120 min was allowed for the preparation to stabilize before any recordings were collected.

Whole cell recording methods.

Glass micropipettes (no. 9-000-2312, outer diameter: 0.060 in. and inner diameter: 0.0445 in., Drummond Scientific, Broomall, PA) were fashioned (model PP-830 microelectrode puller, Narishige) with a tip size of 0.75–1.0 μm [estimated by bubble pressure (Mittman et al. 1987)] and filled with a solution approximating the internal environment of neurons (130 mM K-gluconate, 5 mM NaCl, 1 mM CaCl₂, 1 mM MgCl₂, 11 mM EGTA, 10 mM HEPES, 1 mM GTP, 0.01 mM GMP, and 2 mM Mg-ATP, adjusted to pH 7.35, 290 mosmol, with 1% neurobiotin). The electrical resistance of filled micropipettes ranged from 6 to 8 MΩ. Electrodes were mounted in a carrier (E45W-MxxPH, Warner Instruments), and 15–50 millibars of internal positive pressure were applied. The electrode was advanced into the dorsal horn with a hydraulic microdrive (model 640, David Koph) at an angle of ∼45° from the horizontal at the dorsolateral aspect of the spinal cord just ventral to the dorsal root entry zone. This approach allowed the most direct access to the spinal gray matter that was closest to the surface at this point. Negative square current pulses (0.1 nA, 50 ms) were applied through the electrode at 0.5 Hz using an AMPI Master 8. The electrode potential was amplified (Adams List SEC 05L/H) and monitored on a digital oscilloscope (Nicolet no. 4094) as well as captured and analyzed on a personal computer (Spike 2, version 2.6, Cambridge Electronics Devices). As a single cell was approached, positive pressure was released, and a gigaOhm seal allowed to form. Negative pressure was applied to the electrode as necessary to encourage seal formation. Once the seal was stable, a negative pressure transient was used to rupture the cell membrane and establish whole cell recording while in bridge mode. Only neurons with a resting membrane potential of at least −40 mV, stable baseline recordings, and evoked spikes that overshot 0 mV were used for further experiments and analysis. The input resistance was monitored, and the recording was abandoned if this changed by >15%. Similarly, the recording was abandoned with decay of membrane potential by >5 mV or loss of spike overshoot.

Measurement of passive and active membrane properties.

Resting membrane potential was measured 1 min after the establishment of the whole cell configuration to allow stabilization of the recordings, and 2 min of spontaneous cellular membrane activity was then recorded. Membrane responses to 500-ms hyperpolarizing 20-pA current steps delivered at 3-s intervals were first recorded followed by responses to 500-ms depolarizing 10-pA current steps delivered at 3-s intervals. Finally, responses to electrically evoked primary afferent stimuli were recorded. Glass pipette electrodes filled with ACSF were used to evoke responses in neurons by stimulation of the L5 root and to monitor the amplitude of these stimuli in the adjoining L4 root. Electrodes were placed over the proximal part of the roots 3–4 mm from their insertion site, and suction was applied to secure the roots onto the electrode tips. The distal parts of the roots were placed on silver wire reference electrodes mounted on the side of the recording chamber. The exposed sections of the roots and DRGs were covered with petroleum jelly and gel foam soaked in ACSF, respectively. Dorsal root reflexes were evoked from root L4 by delivery of constant-current electrical pulses to root L5 that were 500 ms in duration. The threshold stimulus strength was defined as the current needed to evoke a dorsal root potential in L4 that was two times background noise. Neuronal responses to electrical stimuli delivered at 2, 10, and 100 times threshold (T) were then recorded from the intact spinal cords. LT neurons were defined as those that responded maximally to 2T stimulation, HT neurons were those that responded only to 10T or 100T stimulation, and WDR neurons were those that responded in a graded fashion across all stimulus intensities. Neurons failing to show AP responses were classified based on the presence of graded subthreshold potentials evoked by root stimuli. Those showing graded excitatory postsynaptic potentials (EPSPs) were classified as EPSP neurons, whereas those that showed graded inhibitory postsynaptic potentials (IPSPs) were classified as IPSP neurons.

Neurobiotin intracellular labeling.

After completion of the electrophysiological recording, L4–L5 spinal segments were immersed in 4% paraformaldehyde for 48 h and transferred to 30% sucrose phosphate buffer for cryoprotection. The spinal cord was frozen and sectioned transversely at 50 μm. Neurobiotin-containing neurons were recovered by an immune/horseradish peroxidase reaction using the avidin-biotin complex technique (ABC kit). To detect neurobiotin, the tissue was incubated with Texas red avidin (1:500, Vector Laboratories) or 3,3′-diaminobenzidine (Vector Laboratories). Images of the labeled neurons were viewed using a Nikon D-FL compound fluorescent microscope, and images were captured using a Cool-Snap CF camera system. Neurons were categorized by soma size and shape and by the number, size, and orientation of primary dendrites (Grudt and Perl 2002).

Data analysis and interpretation.

Membrane properties including resistance, time constant, capacitance, and current-voltage functions were derived from the responses to negative intracellular current injections. Spike threshold, duration, amplitude, latency, afterhyperpolarization (AHP) height, and AHP duration were measured for spontaneous APs and for each set of evoked APs. The APs chosen for analysis were the first evoked at rheobase for positive intracellular current-evoked spikes, the first spike at threshold dorsal root intensity, and the first spike occurring in the spontaneous record with a preceding silent period of 15 s or greater. Statistical differences in all neurophysiological parameters between groups were determined using ANOVA with Duncan's post hoc comparisons. Differences in the distribution of firing patterns or proportions were determined using χ²-analyses with Bonferoni adjustments as needed. P values of <0.05 were considered significant.

RESULTS

Classification of neurons.

A total of 95 neurons were recorded from the L5 segment in 79 isolated spinal cord preparations using blind whole cell patch clamp. Neurons were categorized by functional responses to dorsal root afferent stimuli, by anatomic location, by patterns of AP responses to intracellular current injection, and by cellular morphology where recovery was successful. Data are initially presented using the categorization based on responses to dorsal root afferent stimuli as the main organizing principal as the most clear differences among groups emerged with this basis. Differences that emerged using alternate grouping variables are discussed where appropriate.

Representative responses of the four classes of neurons as defined by their evoked responses to 500-ms L5 dorsal root stimuli are shown in Fig. 1A. Eighteen neurons were classified as LT cells. LT neurons showed maximal AP responses at stimuli 2T, with threshold defined as the minimum current to evoke a dorsal root reflex that was two times background in the adjoining L4 dorsal root. Root stimulation typically produced brief excitatory potentials in LT cells that were commonly followed by inhibitory potentials (Fig. 1A, first column, top left). Whereas excitatory potentials did not grade with stimulus intensity, inhibitory potentials often did. The representative cell shown in Fig. 1 demonstrated a more prolonged inhibitory potential at 10T and 100T than at 2T. AP responses were typically phasic in LT neurons and composed of one to three spikes within decelerating bursts.

Fig. 1.

Representative whole cell patch-clamp recordings (bridge mode) of the five functional classes of dorsal horn neurons identified in laminae I–III taken from the isolated whole spinal cord in vitro are shown in A. Low-threshold (LT), wide-dynamic-range (WDR), nociceptive-specific (NS), and silent neurons were defined by the pattern of response to electrical stimulation of the attached L5 spinal dorsal root. Stimulating intensity is indicated the far left, and levels of resting membrane potential are indicated below each recording trace. 2T, 10T, and 100T represent stimulating intensity in times of threshold to activate a dorsal root potential in the adjoining L4 dorsal root. The diagram in B shows the arrangement of electrodes. HT, high threshold; EPSP, excitatory postsynaptic potential; IPSP, inhibitory postsynaptic potential.

A total of 20 neurons were categorized as WDR cells. These neurons showed AP responses that graded with increasing stimulus intensity (Fig. 1A, second column) that rode upon prolonged excitatory potentials. Spikes were contained within trains that were commonly of constant frequency but, on occasion, in other patterns (see below). Inhibitory membrane events in response to dorsal root stimulation were infrequently observed in WDR cells.

Twenty-nine neurons were classified as HT neurons (also often called NS cells). These neurons showed no AP responses and at best only very small membrane depolarization at stimulus intensities below 10T. In contrast, HT neurons did commonly show brief membrane hyperpolarization at 2T intensity stimuli (Fig. 1A, third column). Membrane hyperpolarization was also commonly present in HT cells at 10T stimuli; however, this was followed by membrane depolarization at this intensity that was sufficiently strong in many cells to result in AP responses. Membrane hyperpolarization was uncommon with 100T stimuli in HT neurons. Spike responses with high-intensity stimuli in HT cells were similar to those of WDR cells in that sustained trains of APs were commonly observed.

Twenty-eight neurons showed no AP responses to any dorsal root stimuli. These cells nevertheless showed membrane events induced by dorsal root stimulation that were either composed of subthreshold EPSPs or IPSPs. Eleven cells showed an EPSP pattern, as illustrated by the representative cell in Fig. 1A, column four, wherein low-intensity stimuli produced small membrane depolarization that graded with stimulus intensity. The remaining 17 neurons showed graded membrane hyperpolarization like that shown by the representative cell in Fig. 1A, column five. The distribution and mean depth of the neurons grouped by functional class are shown in Table 1. All neuron types were sampled in lamina I–III, although HT and EPSP neurons tended to be found shallower than the other neurons. However, this was not statistically significant.

Table 1.

Cellular properties by functional class

	LT	WDR	HT	EPSP	IPSP
n	18	20	29	11	17
Depth, μm	126.3 ± 21.8	137.8 ± 19.6	111.5 ± 14.1	129.8 ± 25.1	158.9 ± 23.3
RMP, mV	−48.4 ± 1.8	−50.0 ± 2.0	−57.2 ± 3.2*#	−59.6 ± 3.9**#	−57.2 ± 3.2*#
R, MΩ	355.8 ± 37.2	376.9 ± 33.9*	567.2 ± 37.2*#	489.6 ± 67.3*	507.5 ± 50.8*#
τ, ms	35.2 ± 6.3	44.9 ± 5.4	47.9 ± 4.2*	34.1 ± 6.6#	41.0 ± 4.2
C, pF	93.4 ± 15.2	104.9 ± 12.2	81.2 ± 9.5**#	91.8 ± 20.3	78.9 ± 20.0*#
Rheobase, nA	46.2 ± 20.0	35.0 ± 10.7	34.8 ± 3.9	45.9 ± 10.0	23.6 ± 4.3*#
AP threshold, mV	−32.5 ± 2.5	−31.7 ± 2.9	−32.0 ± 2.4	−29.3 ± 5.7	−31.1 ± 2.5
AP duration, ms	5.0 ± 0.68	4.3 ± 0.50	6.6 ± 0. 16	5.1 ± 0.82	7.2 ± 0.16

Values are means ± SE; n, number of cells.

LT, low threshold; WDR, wide dynamic range; HT, high threshold; EPSP, excitatory postsynaptic potential; IPSP, inhibitory postsynaptic potential; RMP, resting membrane potential; R, resistance; τ, time constant; C, capacitance; AP, action potential.

^*Different vs. the LT group;

^#different vs. the WDR group. One symbol indicates P < 0.05; two symbols indicate P < 0.01.

Passive membrane properties.

Numerous differences were found between the passive membrane properties of cells of differing functional classes. Table 1 shows the mean values (and SEs) for the passive membrane properties measured on the establishment of whole cell configuration (resting membrane potential) and those calculated (membrane resistance, capacitance, and time constant) after the intracellular passage of hyperpolarizing current steps. HT, EPSP, and IPSP neurons showed significantly more negative resting membrane potential [F(4,90) = 4.07, P = 0.001; post hoc P values shown in Table 1] but higher membrane resistance than LT or WDR cells [F(4,90) = 5.92, P = 0.0007; posthoc P values shown in Table 1]. The time constant and membrane capacitance showed a more complicated pattern of differences and similarities among functional classes (see Table 1).

Clustering of neurons by anatomic location but independent of functional responses to dorsal root stimulation or intracellular current injection revealed differences in passive membrane properties unique to each lamina (Table 2). The resting membrane potential of lamina III neurons was significantly more negative than cells located in lamina I or II [F(2,93) = 3.37, P = 0.04; post hoc P values shown in Table 2]. Similarly, the membrane resistance of cells in lamina III was significantly less than that of cells in lamina I or II [F(2,93) = 2.42, P = 0.03; post hoc P values shown in Table 2]. Finally, the time constant of cells in lamina III was shorter than cells from lamina II [F(2,93) = 2.17, P = 0.04; post hoc values in Table 2]. There were no differences between cells of different lamina in membrane capacitance.

Table 2.

Cellular properties by depth in the spinal cord

	Lamina I	Lamina II	Lamina III
n	37	47	11
Depth, μm	56.9 ± 12	153.9 ± 55.4	315.1 ± 15.4
RMP, mV	−52.5 ± 10	−48.5 ± 28.8	−60.3 ± 13.9**#
R, MΩ	491.7 ± 196.9	468.1 ± 203.5	338.7 ± 126.7*#
τ, ms	42.4 ± 23.8	41.9 ± 24.8	23.9 ± 8.9#
C, pF	81.5 ± 15.7	77.8 ± 10.8	67.0 ± 41.1
Rheobase, nA	0.036 ± 0.007	0.035 ± 0.007	0.055 ± 0.016*#
AP threshold, mV	−31.4 ± 11.6	−31.2 ± 13.1	−33.7 ± 10.7
AP duration, ms	6.19 ± 1.3	5.7 ± 0.38	4.1 ± 1.9

Values are means ± SE; n, number of cells.

^*P < 0.05 vs. lamina I;

^**P < 0.01 vs. lamina I;

^#P < 0.05 vs. lamina II.

Analysis of the passive membrane properties by both functional class and anatomic location revealed that the functional class of neurons was the more reliable sorting criteria. Neurons of a given functional class showed no differences in resting membrane potential or membrane resistance when sorted by anatomic location. Moreover, there were no differences in time constants among LT, WDR, EPSP, and IPSP neurons of different spinal laminae or in the capacitance of these cells from different spinal lamina. The only differences that emerged in this comparison were that the time constant of lamina III HT cells and capacitance of lamina III LT cells were significantly smaller than cells of matching functional class in more superficial layers. Similarly, the membrane resistance of HT, EPSP, and IPSP cells was significantly greater than that of LT and WDR cells in both laminas I and II and showed the pooled trend for cells within lamina III. These differences achieved statistical significance only in comparison of the capacitance between lamina I HT, EPSP, and IPSP cells with that of lamina I LT cells.

Membrane excitability.

Figure 2 shows comparisons of the responses of neurons evoked by root stimulation based on functional class. The bar graph in Fig. 2A, left, shows comparisons of the mean evoked spikes for LT, WDR, and HT neurons across the intensities of root stimulation that were tested. As noted above, LT cells showed a maximal response at very low stimulation intensity and a flat stimulus-intensity profile. WDR and HT cells showed graded responses to stimulus intensity, but with WDR cells encoding responses across the stimulus range, whereas HT cells only responded at intensities at 10T or greater. The pie charts in Fig. 2B show the proportions of spike patterns to root stimulation in each of the functional groups. Cells were classified as showing a single spike, a phasic spike pattern, or a tonic spike pattern based on the response at 100T root stimulation strength. LT cells showed phasic responses to root stimulation composed of either single spikes or short bursts of spikes. This distribution was significantly different than that observed for WDR cells, where the majority of cells showed tonic spike trains to root stimulation and the remainder showed graded phasic spike trains. HT neurons showed a mixture of spike-burst responses to root stimulation that included single spike and phasic and tonic spike trains. The proportions of these patterns were significantly different among all three classes of neurons.

Fig. 2.

Membrane excitability properties varied among the functional groups of spinal neurons. A: the left bar graphs show the comparisons of the mean evoked spikes for LT, WDR, and HT neurons across the intensities of root stimulation. LT cells showed a flat stimulus-intensity profile. WDR cells encoded a graded response across the stimulus range, whereas HT cells only responded at intensities at 10T or greater stimulus intensity. The bar graphs in A, right, show comparisons of the mean maximum amplitude of the graded EPSP and IPSP responses of neurons evoked that did not show action potential (AP) responses by graded root stimulation (top right). B: the pie charts show the distribution of spike patterns evoked at 100T root stimulation in each of the functional groups. The majority of LT cells showed single spike responses, and the rest showed phasic responses to root stimulation. The majority firing pattern from WDR cells showed tonic spike trains to root stimulation. HT neurons predominantly showed a phasic or single spike response with a less frequent tonic pattern to root stimulation. C: in the top bar graphs, a significant percentage of all cells that were studied showed IPSPs to root stimulation. IPSP components in the responses to root stimulation were much less frequent in WDR and EPSP cells. The bottom bar graphs in C show that spontaneous activity was most prevalent in LT, WDR, and HT cells and less so in silent neurons. One symbol indicates P < 0.05; two symbols indicates P < 0.01; three symbols indicates P < 0.001. *Within-group comparisons for stimulus strength versus the baseline response. ^ΔDifferences in total spike responses between functional groups. The arrows in the pie charts indicate the comparisons made.

The bar graphs in Fig. 2A, right, show the mean maximum amplitude of the evoked EPSPs and IPSPs in the two groups of silent neurons that did not show AP responses to root stimulation. Both groups of neurons showed graded responses with increasing stimulus strength. As shown in the bar graphs in Fig. 2C, a significant percentage of all cells that were studied showed IPSPs to root stimulation. By definition, all IPSP cells showed this response, but, in addition, nearly half of LT cells and a third of HT cells showed IPSPs to root stimulation. IPSP components in the responses to root stimulation were much less frequent in WDR cells and absent in all but one EPSP cell. Finally, the bar graphs in Fig. 2C show that spontaneous activity was quite common in all cells. This was most prevalent in the cells with AP responses to root stimulation and less so in EPSP and IPSP neurons.

The analog traces in Fig. 3, top, show representative samples of AP responses to three graded steps of depolarizing intracellular current injection. The bottom trace is at rheobase for each cell, and the next two traces are in 20-pA steps from the first trace. All cells were categorized based on the response to two times rheobase in that often times the rheobase pattern was found to transition from one pattern to another with stronger stimuli as noted below.

Fig. 3.

The top traces show representative AP responses to the three graded steps of depolarizing intracellular current injection. The bottom trace is at rheobase for each cell, and the next two traces are in 20-pA steps from that. All cells were categorized based on the response at two times rheobase. Single spike cells often showed a delay with injection of LT current, and the numbers of spikes in tonic cells always graded with increasing current. The pie charts show the proportions of spike patterns to current injection in each functional group. The majority of firing patterns in WDR and HT cells were tonic spike trains with the balance split between phasic and single spike patterns. ***P < 0.001 vs. the incidence of a tonic spike train vs. the LT group.

Single spike cells (n = 26, Fig. 3, top left) showed just that: a single spike to current injection. Single spike cells often showed a delay with injection of LT current, and this was shortened by stronger current injection. Single spike responses at rheobase or twice rheobase were often observed to transition to phasic or tonic responses with higher current injection. This was observed in 4 of 18 LT cells, 7 of 20 WDR cells, 10 of 29 HT cells, 3 of 11 EPSP cells, and 2 of 17 IPSP cells.

Phasic cells (Fig. 3, center column, n = 13) showed brief self-limiting trains of 2 to 5 spikes with current injection. The spike burst elicited in phasic cells occasionally persisted throughout the current injection at higher strengths, resulting in cells transitioning to a tonic pattern. This was observed in 3 of 20 WDR, 7 of 29 HT, 2 of 11 EPSP, and 1 of 17 IPSP cells. No LT cells categorized as phasic showed transition to tonic and no cells of any functional category showed a change in pattern from phasic cells to a single spike.

Tonic cells (n = 56; Fig. 3, top right) showed sustained APs throughout the duration of current injection. The numbers of spikes in tonic cells always graded with increasing current except in two cells. One EPSP neuron and one LT neuron showed a transition from a tonic pattern at rheobase to a phasic pattern at two times rheobase. No cells showed a change in response pattern from tonic to a single spike. This was the most common pattern of response to root stimulation observed and occurred in 7 of 18 LT cells, 15 of 20 WDR cells, 23 of 29 HT cells, 5 of 11 EPSP cells, and 6 of 17 IPSP cells.

The vast majority of WDR and HT cells showed tonic patterns of response. The majority of LT neurons showed either single spike or phasic response patterns, although a little more than one-third also showed tonic responses. These proportions were significantly different between LT versus WDR and HT cells. IPSP neurons showed a pattern of responses similar to that of LT neurons, whereas EPSP neurons split more or less evenly between the three response patterns.

Grouping of cells by response pattern to intracellular current injection revealed few significant differences (Table 3). Single spike cells had a significantly lower membrane resistance than phasic or tonic cells [F(2,93) = 4.66, P = 0.01]. Single spike cells also showed a higher mean rheobase to current injection than either phasic or tonic cells [F(2,93) = 6.73, P = 0.002]. Finally, single spike cells also had a significantly shorter time constant than tonic cells [F(2,93) = 3.81, P = 0.03].

Table 3.

Cellular properties by response pattern to intracellular current injection

	Single Spike	Phasic	Tonic†
n	21	18	56
Depth, μm	147.4 ± 21.5	107.2 ± 17.5	130.4 ± 10.8
RMP, mV	−50.3. ±2.24	−52.8 ± 1.9	−50.7 ± 6.3
R, MΩ	357.4 ± 38.8	504.4 ± 51.9*	497.1 ± 25.4*
τ, ms	28.9 ± 4.8	40.8 ± 4.4	45.1 ± 3.4*
C, pF	78.9 ± 31.9	62.7 ± 10.7	83.2 ± 8.3
Rheobase, nA	0.068 ± 0.02	0.023 ± 0.004*	0.031 ± 0.03*
AP threshold, mV	−27.7 ± 1.9	−31.7 ± 2.5	−32.9 ± 1.9
AP duration, ms	5.5 ± 0.76	8.2 ± 0.25	5.0 ± 0.4

Values are means ± SE; n, number of cells.

^†Includes eight delay firing pattern neurons.

^*P < 0.05 vs. single spike.

The scatter and line plots in Fig. 4 show several examples of the four types of spike trains that were shown by cells with tonic or phasic responses to intracellular current injection. The x-axis designates the number of spike intervals that comprised each spike train, and the y-axis shows the rate between each of these intervals. Figure 4A shows an accelerating spike train, Fig. 4B shows a decelerating spike train, Fig. 4C shows a plateau or constant train, and Fig. 4D shows several examples of cells with irregular spike trains. The pie charts show the proportions of phasic and tonic LT, WDR, HT, EPSP, and IPSP cells with each of these spike train structures. The ratios of these patterns showed several differences among the functional classes. LT cells showed significantly less decelerating spike trains than shown by HT or EPSP cells and, in addition, showed more plateau spike trains than EPSP cells. WDR cells showed more frequent plateau responses than EPSP cells and significantly fewer decelerating spike trains than either EPSP or IPSP cells. Finally, HT cells showed no accelerating spike trains, although this was only significantly different compared with EPSP and IPSP cells. Current-voltage plots were also defined for each of the functional classes, but no significant differences were observed between cells of different functional class. Finally, the mean spike threshold of each AP in each spike train up to the first five was measured. All neurons showed scaling of the spike threshold from an initial mean near −35 mV up to a mean of −30 mV or greater, but no significant differences among types were found (data not shown).

Fig. 4.

Scatter and line plots showing several examples of the four types of spike trains that were shown by cells with tonic or phasic responses to intracellular current injection. The x-axis designates the number of spike intervals that comprised each spike train, and the y-axis shows the rate between each of these intervals. A: accelerating spike train. B: decelerating spike train. C: plateau or constant train rate. D: several examples of cells with an irregular spike trains. The pie charts show the proportions of phasic and tonic LT, WDR, HT, EPSP, and IPSP cells with each of these spike train structures. The ratios of these patterns showed several differences among the functional classes. LT cells showed significantly fewer decelerating spike trains than shown by HT or EPSP cells and, in addition, showed more plateau spike trains than EPSP cells. WDR cells showed more frequent plateau responses than EPSP cells and significantly fewer decelerating spike trains than either EPSP or IPSP cells. HT cells showed no accelerating spike trains, although this was only significantly different compared with EPSP and IPSP cells. Two symbols indicate P < 0.01; three symbols indicate P < 0.001.

The analog traces in Fig. 5, top, show examples of the patterns of rebound spikes that were observed after the release of hyperpolarizing steps of intracellular current injection. Single spikes, phasic spike train bursts, and tonic trains were observed. Notably, in many of the cells that showed phasic or tonic rebound spike responses, pronounced sag voltage was also observed (arrows). The bar graphs in Fig. 5, middle, show the percentages of cells in each functional class where rebound APs and sag voltages were observed. EPSP and IPSP neurons showed significantly fewer rebound spikes than the cells that showed AP responses to root stimulation. In contrast, HT and EPSP neurons showed significantly more frequent sag voltages to hyperpolarizing steps than the other groups of cells. The pie charts in Fig. 5, bottom, show the proportions of neurons in each functional group showing the three types of rebound AP spike patterns. Single spikes were the most commonly observed pattern across all neuron types. IPSP cells showed the most frequent tonic responses, but this was not statistically different from the other cells given the small numbers of neurons.

Fig. 5.

The top analog traces show representative examples of the patterns of rebound spikes that were observed after the release of hyperpolarizing steps of intracellular current injection. Single spike (left), phasic (middle), and tonic spike trains (right) were observed. The arrow indicates sag currents that were typically seen in cells with phasic or tonic rebound spike responses. The bar graphs show the percentages of cells in each functional class where rebound APs (right) and sag currents (left) were observed. Silent (EPSP and IPSP) neurons showed significantly fewer rebound spikes than other cell types to root stimulation. HT and EPSP neurons showed significantly more frequent sag currents to hyperpolarizing steps than the other groups of cells (*P < 0.05; **P < 0.01). The bottom pie charts show the proportions of functional neurons showing the three types of rebound AP spike patterns. There were no significant differences in these proportions.

Figure 6 shows the relationships between the spike patterns evoked by all three types of stimuli that were used to elicit APs for each of the functional classes of cells. The pattern used in this analysis for the root stimulation was that for the maximal response in each class (2T for LT cells and 100T for WDR and HT cells), whereas the two times rheobase threshold was used for the responses to intracellular depolarizing and hyperpolarizing steps. The responses of LT cells were shown to be clustered to the low center portion of the graph in Fig. 6A, indicating that these cells tended to show single or phasic responses to all three types of stimuli. In contrast, the responses of the WDR and HT cells tended to cluster to the right lower portions of the graphs shown in Fig. 6, B and C, respectively. This indicates that WDR and HT cells tended to show similar responses to root stimulation and intracellular current injection. WDR cells, most especially, showed tonic or phasic responses to these stimuli, whereas HT cells most especially showed single or phasic spike responses. Finally, EPSP and IPSP cells were clustered to the left front bottom portion of the graphs shown in Fig. 6, D and E, indicating that these cells, by definition, showed no responses to root stimulation but, in addition, rarely showed APs on rebound from hyperpolarizing steps. The responses to depolarizing intracellular current injection were variable for these cells.

Fig. 6.

Relationships between spike patterns evoked by all three types of stimuli that were used to elicit APs for each of the functional classes of cells. Data acquired at 2T were used for LT cells and data acquired at 100T were used for WDR and HT cells to dorsal root stimuli; data acquired at two times rheobase were used for the intracellular depolarizing current stimuli. Responses of LT cells were found to be clustered to the low center portion of the graph with many single or phasic responses to all three types of stimuli. Responses of WDR and HT cells tended to cluster to the right lower portions of the graphs, respectively, indicating a prevalence of tonic or phasic response to both stimuli. Silent (EPSP and IPSP) cells were clustered to the left front lower portion of the graphs, indicating that these cells, by definition, showed no responses to root stimulation and few rebound spikes to hyperpolarizing steps but with variable responses to depolarizing intracellular current injection.

The analog traces in Fig. 7 show representative APs evoked by each type of stimulus as well as for spontaneous APs for each of the functional classes, whereas the bar graphs in Fig. 8 show the group mean data derived from these spikes. The surprising result that comes from this analysis is the variability in the AP data based on the type of evoking condition. For example, as shown in Fig. 8A, the top left graph shows that the spike threshold assessed for the first spike evoked by depolarizing current injection was significantly more positive than the threshold for spikes evoked under any other condition for all functional cell groups. Interestingly, comparison of the mean AP thresholds between functional groups showed that with the exception of the threshold for rebound spikes in the EPSP group, the thresholds were overall very similar within each type of evoking condition. Figure 8B shows that the mean current as calculated from the initial baseline membrane potential to the AP threshold was greatest for the rebound spikes. An obvious caveat here is that the “baseline” membrane potential was that induced by an injection of hyperpolarizing current. The mean discharge rates, as shown in Fig. 8C, were highly variable, in that most cells of all classes showed either phasic or tonic spike pattern responses to current injection and in response to root stimulation, whereas rebound and spontaneous spikes were most commonly single AP events. AP latency was also most prolonged for the rebound spikes in each functional group except for EPSP neurons. This group of neurons tended to show very gradual depolarizing ramps preceding spontaneous APs. AP duration and AHP duration were mostly constant both between and within each functional group, although with some notable exceptions. As a group, rebound APs tended to be more prolonged than APs evoked under other conditions. The afferent evoked APs showed very sharp spikes with short latency and prolonged AHP. Finally, note that IPSP neurons showed very prolonged AHP after spontaneous discharges.

Fig. 7.

Representative spontaneous and stimulus-evoked APs are shown for each of the functional classes of neurons. Comparison of the spikes at the left indicated that the spike threshold assessed for the first spike evoked by depolarizing current injection was significantly more positive than the threshold for spikes evoked under any other condition for all functional cell groups.

Fig. 8.

The bar graphs summarize the AP parameters occurring in spontaneous spike trains and evoked by dorsal root stimulation or current inject. A: means (and SEs) for AP threshold (first spike used for all evoked responses). B: means (and SEs) for rheobase calculated from the initial baseline membrane potential to the AP threshold. This was greatest in all classes of cells for the rebound spikes. C: mean (and SE) discharge rates occurring in spontaneous trains and evoked. These were found to be highly variable for all cell types. D: means (and SEs) for AP latency. The rebound spikes in each functional group except EPSP cells showed the most prolonged latency. This group of EPSP neurons tended to show very gradual depolarizing ramps preceding spontaneous APs. E and F: AP duration (E) and afterhyperpolarization (AHP; F) duration were mostly constant both between and within each functional group, although with some notable exceptions. IPSP neurons showed very prolonged AHP after spontaneous discharges. Afferent-evoked APs displayed a sharp spike with a short latency and AHP duration.

Anatomic classifications.

Morphology was studied in 26 neurons, including 5 LT, 5 WDR, 8HT, 3 EPSP, and 5 IPSP cells and classified based on the descriptions provided by Grudt and Perl (2002). Central cells had a rounded cell body and moderately extensive arbor. Eight of these neurons were recovered, four neurons with their soma located in lamina IIi and four neurons in lamina IIo. Three of these neurons showed HT responses to root stimulation, two neurons showed a WDR response, two neurons showed an LT response, and the last neurons showed an IPSP response. Similarly, the cells showed a mixture of response patterns to intracellular current injection. HT cells showed tonic and phasic patterns, WDR cells showed a tonic pattern, LT cells showed a tonic and a phasic pattern, and the IPSP cell showed a single spike response.

Five islet cells with rounded or fusiform somata with tight arbors in the medial-lateral direction but extensively running in the rostrocaudal direction were observed. The soma of three cells were located in lamina I, one cells was in lamina IIo, and the last cell was in lamina IIi. Three of these neurons showed HT responses to root stimulation, one neuron had an EPSP response, and the last neuron had an IPSP response. EPSP and IPSP cells both showed phasic response patterns to intracellular current injection, whereas two of the HT neurons showed tonic responses, and the last neuron showed a single spike pattern to intracellular current injection.

Medial-lateral neurons had somata that were typically rounded with a medially and laterally directed sparse arbor. Five of these neurons were recovered, two neurons with soma in lamina IIo and the other neuron in lamina IIi. Two of these neurons showed an EPSP response to root stimulation, two neurons showed a LT response, and the last neuron showed a WDR response profile. EPSP and WDR neurons both showed tonic responses to intracellular current injection, whereas LT neurons showed phasic responses.

Finally, eight vertical cells, typically with a pyramidal or elongated somata and dendrites running ventrally from the soma, were observed. Of these, four cells were localized to lamina I, three cells were in lamina IIo, and the last cell was in lamina IIi. responses to root stimulation resulted in two cells being classified as WDR neurons, two cells being classified as HT neurons, three cells being classified as IPSP neurons, and one cells being classified as a LT neuron. All HT and WDR neurons and one IPSP neuron showed tonic responses to intracellular current injection. The LT cell and one IPSP cell showed single spike responses, and the last IPSP cell showed a phasic response.

Analysis of cell morphological properties based on classification by the functional response to root stimulation yielded a number of differences. The mean size of the soma of the HT neurons was the largest for all cells but only showed significant differences in rostrocaudal extent compared with LT neurons and in dorsoventral and mediolateral extent compared with WDR cells (Table 4). Similarly, the dendritic arbor sizes in HT cells were significantly larger in rostrocaudal extent than that of either LT or WDR neurons (Table 4). Finally, as shown in Table 5, there were few significant differences observed in membrane properties among cells grouped by morphological classification. The membrane resistance of medial-lateral cells was significantly less than that of central and vertical cells, and vertical cells showed a significantly lower threshold than that of islet cells.

Table 4.

Morphological properties by functional class

	LT	WDR	HT	EPSP	IPSP
n	5	5	8	3	5
Soma dimensions, μm
R-C	16.1 ± 1.2*	16.7 ± 1.6	19.5 ± 0.8	16.9 ± 1.3	17.8 ± 0.9
D-V	18.4 ± 2.2	16.0 ± 1.5*	19.7 ± 1.3	15.5 ± 1.3	17.2 ± 1.4
M-L	15.6 ± 2.6	15.6 ± 1.6*	18.9 ± 0.9	15.0 ± 1.6	15.8 ± 1.5
Dendritic dimensions, μm
R-C	74.3 ± 13.2*#	78.9 ± 15.5*#	256.0 ± 34.9	215.1 ± 44.4	228.6 ± 21.9
D-V	152.7 ± 30.6	118.9 ± 17.7	159.0 ± 24.1	163.8 ± 18.5	112.0 ± 30.5
M-L	122.9 ± 33.3	120.6 ± 21.1	92.0 ± 15.6	118.1 ± 21.9	80.2 ± 10.9

Values are means ± SE; n, number of cells.

R-C, rostrocaudal extent; D-V, dorsoventral extent; M-L, mediolateral extent.

^*P < 0.05 vs. HT neurons;

^#P < 0.05 vs. IPSP neurons.

Table 5.

Cell properties based on cell morphology

	Central	Islet	Medial-Lateral	Vertical
n	8	5	5	8
Depth, μm	120.3 ± 40.5	89.2 ± 19.0	134.3 ± 21.9	122.8 ± 31.1
RMP, mV	−53.0 ± 3.6	−47.3 ± 4.0	−62.7 ± 10.6	−50.6 ± 2.8
R, MΩ	552.8 ± 112.3*	464.0 ± 74.3	289.8 ± 73.4	505.9 ± 73.5*
τ, ms	40.2 ± 7.3	38.1 ± 16.0	22.1 ± 4.5	56.1 ± 7.2
C, pF	73.2 ± 17.3	71.3 ± 32.0	129.0 ± 64.1	131.0 ± 71.2
Rheobase, nA	0.15 ± 0.006	0.32 ± 0.017	0.017 ± 0.008	0.02 ± 0.006
AP threshold, mV	−27.3 ± 6.3	−18.6 ± 5.75*	−44.6 ± 15.6	−33.3 ± 4.18
AP duration, ms	7.8 ± 2.4	5.9 ± 1.7	8.6 ± 1.7	6.3 ± 1.3

Values are means ± SE; n, number of cells.

^*P < 0.05 vs. medial-lateral neurons.

DISCUSSION

The present study characterized the relationship of membrane properties, spike burst responses, laminar location, and functional class of dorsal horn neurons recorded in vitro from the intact spinal cord. The important finding in these data is that the functional properties of spinal cells are correlated with their passive biophysical membrane properties, response to intracellular current injection, and, to some extent, the anatomic class of neurons.

HT, EPSP, and IPSP neurons were found to have higher membrane resistance and more negative resting membrane potentials than LT or WDR neurons. In addition, rheobase in LT and WDR neurons was significantly lower than that of HT, EPSP, or IPSP neurons. These data indicate that HT, EPSP, and IPSP neurons have passive membrane properties that would require higher amounts of synaptic current to drive the cells toward spike threshold than would WDR or LT neurons. The analysis also indicated that WDR and HT cells tended to show similar responses to root stimulation and intracellular current injection, most especially tonic or phasic response patterns to these stimuli. From the analysis of rebound spikes and spontaneous activity, EPSP and IPSP cells displayed lower activity than other neurons that had AP responses to root stimulation. Hence, HT, EPSP, and IPSP neurons would be more likely to passively filter inputs from primary afferents, whereas WDR or LT neurons would be more likely to follow synaptic inputs impinging upon them. Lopez-Garcia and King (1994) used a spinal cord-hindlimb preparation and also found cell types and results similar to those we reported here. WDR neurons were mainly observed to exhibit a tonic pattern of discharge to dorsal root stimulation, and NS neurons mostly showed a phasic burst. LT cells displayed either tonic or phasic responses. Many IPSP neurons showed rebound excitation, whereas EPSP cells displayed tonic or single spikes. NS cells tended to have the highest membrane resistance, but they only had a small sample of cells, and this did not achieve statistical significance from the other groups. In vivo recordings of spinal neurons have shown results both similar and differing from those reported here. Our own in vivo results are entirely consistent with the present findings in that HT neurons as defined by responses to cutaneous stimuli has lower resting membrane potential and lower membrane resistance than LT or WDR neurons (Weng and Dougherty 2002). This again is consistent with these neurons have passive membrane properties conducive to filtering low-intensity stimuli impinging upon them. Similarly, Reali et al. (2011) concluded that the intrinsic membrane properties of spinal neurons recorded in vivo were related to spike-burst properties. On the other hand, Graham et al. (2004) could not find a relationship between cutaneous and spike-burst responses.

The patterns of response to intracellular current injection showed an interesting association with functional class. Although one-third of LT neurons showed a tonic response, the majority showed either a phasic or single spike response. In contrast, by far the majority of HT and WDR neurons showed tonic responses. The patterns of response to intracellular current injection in EPSP and IPSP cells were almost identical to LT neurons. A notable difference in the data reported here from other similar studies is that we found only eight neurons with a delayed firing pattern and we found no neurons with a gap (irregular) pattern (Punnakkal et al. 2014; Ruscheweyh and Sandkuhler 2002). Given this relative infrequency, we collapsed delayed firing cells into the tonic category as the spike train once initiated assumed this pattern, albeit with a delay that averaged 0.16 s to the first spike. The delay decreased with higher current injection such that the occurrence of the first spike became shortened to an average of 0.06 s. The occurrence of the delay pattern was seen most frequently in HT (n = 3) and WDR (n = 2) neurons but also found in one of each of the other functional classes. A delayed firing pattern has been suggested as reflecting activation of a transient voltage-dependent outward current, presumably an A-current (Punnakkal et al. 2014). A-currents have suggested as being prevalent in neurons expressing the vesicular glutamate transporter in mice but very infrequent in either GABA- or glycine-containing neurons (Punnakkal et al. 2014). Studies in spinoparabrachial and rostral spinoperiaqueductal gray lamina I neurons have also suggested that gap firing and A-currents are common in these cells, whereas, in contrast, caudal spinoperiaqueductal gray neurons showed burst firing and LT Ca²⁺ currents (Grudt and Perl 2002; Ruscheweyh and Sandkuhler 2002). Delayed responses have been shown as especially prevalent in excitatory dorsal horn interneurons positive for PKC-γ or NK1, and these cells also express A-currents specifically mediated by Kv4.2 and modulated by metabotropic glutamate receptor 5 (Hu and Gereau 4th 2011). The paucity of delayed pattern neurons observed here could be explained by an inadvertent oversampling of inhibitory neurons, although this seems unlikely given that excitatory neurons comprise an estimated 65% of the total population (Todd 2010). Rather, the more likely explanation is that perhaps our criteria for categorization were too strict and/or that the majority of the phasic neurons observed here were excitatory. In addition, the five classes of AP response types to current injection in spinoparabrachial and spinoperiaqueductal gray cells appeared to be organized based on depth in the dorsal horn (Ruscheweyh and Sandkuhler 2002). A tonic firing pattern was most frequent in lamina I and III but not found in lamina II. A delayed firing pattern was observed in lamina I and II but not lamina III. An initial burst or single spike firing pattern was found only in lamina II. Perhaps given that a signifcant portion of our smaple was from lamina III, this introduced a seeming imbalance in the frequency of the delay pattern of responses.

The laminar location of neurons showed little correaltion with the functional properties of neurons in this study. HT cells tended to be found in the more superficial layers of the spinal cord, but this was not signifcantly different from the other functional classes of cells. In addition, lamina III neurons, as a group, showed a shorter time constant and higher rheobase than cells in the more superficial layers. It is well known that projection neurons are located in lamina I and III-V (Todd 2010; Willis Jr. and Coggeshall 2004). The physiological properties of projection neurons have revealed a number of distinguishing characteristics (Grudt and Perl 2002; Ruscheweyh and Sandkuhler 2002). Projection cells in the superficial dorsal horn all showed a distinctly low resistance, a short latency to evoked response, and increased incidence of monosynaptic input. Some of these cells were further suggested to have direct Aδ dorsal root input. Meanwhile, nonprojection lamina I neurons primarily received monosynaptic C-fiber inputs, and the frequency of spontaneous EPSCs was considerably higher in nonprojection cells than in projection cells. Tonic response neurons showed higher rheobase, and all showed hyperpolarization-actived current (I_h) in voltage-clamp studies. Nonprojection cells with tonic spike trains also often showed I_h (Grudt and Perl 2002; Ruscheweyh and Sandkuhler 2002).

Prescott and DeKoninck (2002) identified four classes of AP response types in lamina I spinal neurons. Tonic neurons were typically found to be fusiform, and 11 of 15 neurons showed rebound spiking. Phasic neurons were typically pyramidal, and half of these showed rebound APs. Delayed cells were typically multipolar, and none of these showed rebound spiking. Finally, single spike neurons were typically multipolar, and only 1 of 14 neurons showed rebound discharges (Prescott and DeKoninck 2002).

Schneider (2003) identified three classes of lamina III–V neurons in the hamster. Twenty-two of 30 tonic firing cells were presumed projection neurons; 42 of 45 cells were phasic, and 12 of 17 cells had delayed firing pattern and were presumed interneurons. In deep dorsal horn neurons (lamina V–VII), Szûcs et al. (2003) used 100-pA current injection to demonstrate tonic, phasic, or single spikes in all cells. They also separated the tonic cells to fast versus slow tonic firing patterns. Some of their results were similar with ours. Phasic cells also showed pronounced spike accommodation and deceleration. In slow firing neurons, there was a significantly lower membrane resistance than phasic or tonic neurons. Single firing neurons showed a smaller maximum AP amplitude and significantly lower AHP. Unlike our results in the present study, they found no differences in resting membrane potential between deep dorsal horn neurons of different classes based on intracellular current injection.

Among nonprojection neurons, stronger correlations to laminar location have been revealed for neurons with particular types of surface immunohistochemical markers or neurotransmitter content. Parvalbumin-containing cells with properties like inhibitory neurons (predominantly tonic spike burst responses, islet, and central-like morphology) were found concentrated in lamina Iii and lamina III (Hughes et al. 2012). These did not have any distinguishing passive membrane properties but had higher rheobase, and almost all demonstrated I_h and expressed the hyperpolarization-activated cyclic nucleotide-gated (HCN) subtype 4 channel (Hughes et al. 2012). Calretinin-expressing neurons were concentrated in lamina I and II and included two classes (Smith et al. 2015). The typical group was ∼85% of all those expressing calretinin and were characterized by a delayed firing pattern, large A-current, and a central, radial, or vertical morphology. The atypical group, in contrast, was characterized by a tonic or phasic response pattern, I_h, and islet cell morphology. Both responded to noxious peripheral stimulation, although this was stronger in the typical cells. Passive membrane porperties were not remarkable among these cells (Smith et al. 2015).

A unique aspect of the present study was the opportunity to compare patterns of resposnes of single cells to dorsal root input and direct current injection. A study similar to this compared the responses of cells to step current injection to that of current injected in a pattern mimicking that observed in vivo from cells while a noxious cutenous input was applied (Graham et al. 2007). The overall discharge frequency and rehobase were similar under both stimulus conditions, but the latency to discharge and duration showed variation. In addition, phasic and tonic neurons under step current injection showed similar patterns of response to pinch current, whereas single spike neurons remained resistant to pinch current responses. Here, LT and WDR cells tended to show a consistent pattern of response to intracellular current injection and to afferent stimualtion. In contrast, HT neurons tended to show a tonic pattern of response to intracellular current injection but often showed single spike or phasic responses to root stimulation. We also found that the threshold to activate neurons was more positive with intracellular current injection than to afferent stimulation. Interestingly, the threshold to afferent stimulation approximated that observed for spontaneous APs. Other properties of APs were constant over the various conditions.

Several studies have sought to demonstrate a correlation between functional and morphological class in spinal neurons. The anatomic classification scheme was based largely an especially influential study by Grudt and Perl (2002). They defined five morphological classes of lamina II neurons. Islet cells displayed tonic spike burst responses and I_h in voltage clamp. Central cells with I_h or without I_h typically showed transient spike burst responses. Medial-lateral cells typically showed delayed tonic spike responses and I_h in voltage clamp. Radial cells displayed delayed spike responses, and about one-quarter showed I_h. Finally, vertical cells were split between those with tonic and delayed firing patterns, and one-third of them showed I_h. Hantman et al. (2004) reported that glutamate decarboxylase (GAD)67-negative enhanced green fluorescent protein (GFP)-positive cells typically demonstrated tonic evoked spike discharges to intracellular current injection, I_h, and a central cell morphology. Non-GFP-positive cells typically demonstrated phasic spike burst discharges and infrequent I_h. Maxwell et al. (2007) found that all GAD67-positive cells showed an islet cell morphology and a tonic spike-burst response pattern. In addition, most (4 of 6) vertical cells were vesicular glutamate transporter2 positive but had variable spike burst properties. Meanwhile, Gassner et al. (2013) showed that a large proportion (47%) of spinal GAD67-negative enhanced GFP-expressing neurons had an islet cell morphology, whereas, in contrast, a large proportion of non-enhanced GFP-positive neurons showed an inverted stalked cell morphology. Daniele and MacDermott (2009) observed that GAD67-positive cells in lamina IIo showed both vertical cell and islet cell morphological shapes. They found 65% of GAD67-positive neurons showed a HT response to dorsal root stimulation with the balance split between WDR (13.5%), LT (7.5%), and silent (15%) response patterns. Finally, Schoffnegger et al. (2006) found that GAD67-positive neurons displayed a mix of responses to intracellular current injection, including phasic (40%), gap (30%), tonic (20%), and single spike or delayed firing patterns. Neither these patterns of response to current injection nor passive membrane properties were affected by peripheral nerve injury (Schoffnegger et al. 2006). Our results showed little correlation between morphological type and functional properties, perhaps not surprising given the relatively small numbers of neurons that were studied. Yet, an important finding nevertheless emerged from these results, as cells of a particular morphological types demonstrated a variety of functional responses based on afferent input. This indicates that whole sets of each of the morphological types are devoted to the processing of all intensities of peripheral inputs.

In summary, there are still many controversies in attempting to correlate functional categories related with encoding properties of discharge in spinal dorsal horn neurons. The results presented here suggest that functional class of the neurons is reliable sorting criteria to predict both passive and spike-burst evoked responses in spinal neurons. Additional studies using both isolated and in vivo approaches combined with more sophisticated labeling strategies should help better define these processes.

GRANTS

This work was supported by National Institutes of Health Grants NS-046606 and CA-200263 and the H.E.B. Professorship in Cancer Research.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

P.M.D. conception and design of research; P.M.D. and J.C. performed experiments; P.M.D. and J.C. analyzed data; P.M.D. and J.C. interpreted results of experiments; P.M.D. and J.C. prepared figures; P.M.D. and J.C. drafted manuscript; P.M.D. and J.C. edited and revised manuscript; P.M.D. and J.C. approved final version of manuscript.