INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Qualitatively different pain sensations can be elicited from the skin by different stimulus modalities, such as mechanical, thermal, or chemical stimuli. Regardless of the modality (i.e., including electrical), a sudden noxious stimulus causes two discriminable pain sensations: a short-latency "first" or sharp pain, and a long-latency "second" or burning pain. The former depends on conduction in A primary afferent fibers, while the latter depends on C-fibers (Campbell and Meyer 1996; Lewis and Pochin 1937; Price 1988). Yet, distinct central substrates for these different sensations have not been differentiated.

The available evidence supports the concept that the lamina I spinothalamic tract (STT) projection comprises discrete sensory channels that could provide the basis for distinct sensations, that is, virtual "labeled lines" (Craig et al. 2001). Lamina I of the spinal and trigeminal dorsal horn is an integral component of the central representation of pain, temperature, and itch sensations (see reviews by Craig 2000a; Perl 1984). It receives monosynaptic input from A- and C-fibers, it contains morphologically distinct, modality-selective neurons, it is the main output from the superficial dorsal horn, and it projects to the forebrain in the lateral STT, which is critical for these sensations. Our recent identification of histamine-selective lamina I STT cells provides a clear demonstration of the selectivity inherent in this projection system (Andrew and Craig 2001a): such cells receive input from a specific subset of very slowly conducting C-fibers that are selectively responsive to histamine; they are distinct with respect to ongoing activity, central conduction velocities, and thalamic projections; their temporal response profile parallels itch sensation in humans; and their axons in the lateral STT seem to be critical for the sensation of itch. Thermoreceptive-specific (COOL and WARM) lamina I STT cells are similarly unique in their functional and anatomical characteristics and can be directly associated with human thermal sensation (Andrew and Craig 2001b; Craig and Dostrovsky 2001; Craig et al. 2000, 2001; Dostrovsky and Craig 1996; Han et al. 1998).

Two major types of nociceptive cutaneous lamina I STT neurons have been identified whose functional roles have not yet been clearly distinguished (Craig 2000a). Nociceptive-specific (NS) cells, dominated by A input, and polymodal nociceptive (HPC, for heat, pinch, and cold) cells, dominated by C-fiber input, differ significantly in their afferent input, their sensitivity to cold, their ongoing discharge, their central conduction velocities, and their somatal shapes, yet they display similar stimulus-response profiles to simple noxious mechanical and noxious heat stimuli, similar projections to thalamus, and similar responses to morphine. Nonetheless, our quantitative analyses indicate that their median thresholds to noxious heat and pinch differ, suggesting that the response profiles of these cells might be physiologically differentiable with more refined testing (Craig et al. 2001).

To functionally dissect the possible roles of NS and HPC lamina I STT cells, the experiments reported in this and the preceding paper (Andrew and Craig 2002) used novel stimulus paradigms that distinguish different aspects of human pain sensation. The preceding paper reported the correspondence between the responses of NS cells and mechanically induced first pain, and this paper reports the correspondence between the responses of HPC cells and thermally induced second pain. Here, we used a method introduced in a recent psychophysical study by Vierck et al. (1997), which we refer to as the repeated brief contact heat paradigm. In this paradigm, a hot stimulus is used that is perceived only as warm if contacted very briefly. Repeated brief contacts reliably elicit a strongly augmenting second (burning) pain sensation, but only a weak first (sharp) pain sensation. Most strikingly, Vierck et al. (1997) showed that this sensation displays a rapid "reset" phenomenon, in which the temporal summation begins again from baseline following very short intertrial intervals. The present experiments demonstrate that the responses of HPC lamina I STT can explain the second pain sensation described by Vierck et al. (1997). Furthermore, this paradigm distinguishes HPC from NS cells, whose responses parallel the resulting first pain sensation, consistent with the correspondence described in the preceding paper. A preliminary account of this work has been given (Craig and Andrew 1999).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The data reported in this article were obtained from 18 adult cats of either sex, anesthetized with pentobarbital sodium. The procedures for anesthesia, surgical preparation, microelectrode recording, unit isolation, antidromic identification, and unit classification based on natural and quantitative stimulation were the same as described in the preceding paper and in prior descriptions of lamina I STT cells from this laboratory (Andrew and Craig 2002; Craig et al. 2001). The procedures were approved by the local Institutional Animal Care and Use Committee, and they conform to the guidelines of the American Physiological Society and the National Institutes of Health.

Repeated brief contact heat stimulation

For the present experiments, thermal stimuli were applied with a thermoelectric stimulator (20 × 20 mm; Marlow Industries, Dallas, TX) attached to a custom-built, microprocessor-controlled pneumatic piston, which enabled the thermode to be applied to the skin repeatedly in a reproducible manner. The thermode had a thermocouple glued to its surface for feedback control of its temperature by a custom-built power amplifier. The temperature used for analysis was the mean of repeated measurements at the interface between the thermode and the skin with a separate, calibrated thermocouple (0.005-in. type T, Omega CL23A, Stamford, CT). The thermode was first placed evenly on the ventral glabrous hindpaw with moderate pressure, and then the piston was retracted to the hold position, 1 in. above the skin. The microprocessor controlled the position of the piston, the dwell time, and the interstimulus interval; the advance speed and compliance were maintained constant by an air pressure regulator. An analog output signal provided a record of the command voltage governing the piston's position.

For each unit, a series of trials consisting of repeated contacts of the preheated thermode with the hindpaw were performed using different temperatures and temporal parameters. Measured temperatures of 34.0, 45.5, 48.5, 53.0, and 57.7°C (skin-thermode interface 20-s plateau temperatures) (see Craig et al. 2001), which for the purpose of this article are abbreviated to 34, 46, 49, 53, or 58°C, were used for different trials in ascending sequence. For each temperature, the thermode was applied 15 times each in trials at interstimulus intervals (ISIs) of 5, 3, and 2 s, usually but not always in descending order, with ~3 min between each ISI trial and ~5 min between each temperature step. The contact dwell time on the ventral hindpaw was ~0.7 s (calibrated by means of thermocouple measurements on the hindpaw). Stimulus sequences at a given temperature were in some cases repeated two or more times, using a different ISI sequence. To test the reset phenomenon, the trial for the highest temperature at 3 or 2 s ISI was followed by another trial at the same ISI, using a manually controlled intertrial interval of 4-12 s (even multiples of 2-6 times the ISI). The reset phenomenon was also tested by repeating this pairing in separate trials.

Skin temperatures were measured during trials in three cats by placing a 0.005-in. calibrated thermocouple on the skin surface (in 2 cases) or ~1 mm below the skin surface (1 case) of the glabrous central pad. The peak temperatures during each stimulus contact and the final baseline temperatures between stimulus contacts were measured from the DC recording.

The data were stored and analyzed on a PC using a Power1401 and the program Spike2 (Cambridge Electronic Design, Cambridge, UK) or on a UNIX computer (Masscomp 5400) using custom software. The recorded unit spikes were summed in bins that corresponded to the intervals between the onset of successive thermode contacts. The discharge frequencies were calculated for each stimulus contact and plotted with respect to stimulus number for each trial at each different temperature and each different ISI. Mean frequencies were averaged relative to stimulus number across units. Comparisons between population response curves were made statistically using one-way repeated measures ANOVA and post hoc t-tests with the programs SigmaStat/SigmaPlot (SPSS, Chicago, IL) and CSS Statistica (Statsoft, Tulsa, OK), with P < 0.05 as the significance criterion. Trend analyses of population response curves were performed (as directed in CSS Statistica) using a fixed effects, repeated measures ANOVA with no independent variable and with all stimulus contacts as dependent variables.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 20 lamina I STT units were recorded in these experiments; about half were classified as NS (n = 9) and about half as HPC (n = 11), based on their responses to natural cutaneous stimulation with standard innocuous and noxious mechanical and thermal stimuli (Craig et al. 2001). All had receptive fields that were restricted to or included a major portion of the ventral glabrous hindpaw (e.g., digits 3 and 4 and the lateral third of the central pad), and many were responsive to all glabrous hindpaw skin. One of the NS cells was responsive only to noxious heat stimuli, whereas the remainder were sensitive both to heat and to pinch. All units were antidromically activated from the ventral aspect of the ventrobasal complex in lateral thalamus, and nearly all were antidromically activated from medial thalamus (the submedial nucleus). The mean conduction latencies of these NS and HPC cells were significantly different at 118 ± 41.7 (SD) ms and 77 ± 27.0 ms (P < 0.02), respectively, indicating conduction velocities of 2.9 and 4.4 m/s for this sample. The NS cells had less ongoing discharge (µ = 0.3 ± 0.3 SD impulses/s, range 0.0-0.6) than the HPC cells (µ = 1.6 ± 1.2 impulses/s, range 0.0-4.8, P < 0.01). Thus the general characteristics of these units were consistent with prior work (Craig et al. 2001).

Systematic data were obtained with stimulus trials at the different stimulus temperatures and ISIs from 16 lamina I STT cells, comprising 8 NS and 8 HPC units. The incomplete observations made in the other four units were consistent with the following results.

Lamina I STT population response

At sufficiently high stimulus temperatures and sufficiently short ISIs, the population of lamina I STT neurons displayed responses to the repeated brief contact heat stimuli that augmented with successive stimuli. This temporal summation showed a progressive increase with increasing stimulus temperature and with decreasing ISI, and it was maximal for the highest temperature at the shortest ISI. Figure 1 shows the family of response curves for a single HPC lamina I STT cell (it32u2), organized according to ISI at the four different stimulus temperatures. These graphs display a clear dependence on temperature and a clear inverse relationship with ISI. Specifically, the unit showed a weak, delayed response at the 2-s ISI with the 49°C stimulus, larger responses at 54°C, and the greatest response at the 2-s ISI with the 58°C stimulus. The greatest augmentation occurred between the 3rd and 10th contacts. Like this unit, none of the lamina I STT units responded to repeated brief contact heat stimuli unless the stimulus temperature was well above the thresholds of nociceptive lamina I STT cells as measured with 20-s heat pulses, which overall range from 42.7 to 53.0°C (HPC median ~45.5°C, NS median ~43.0°C) (Craig et al. 2001).

View larger version (11K): [in this window] [in a new window]   Fig. 1. The family of response curves for unit it32u2, showing the evoked discharge frequency following each stimulus contact and prior to the next, for each ISI at each stimulus temperature (°C).

Figure 2 (top half) presents original records showing the action potentials discharged by this HPC neuron in response to each of the 1st 12 stimuli with the 58°C stimulus in the trials at the 3-s ISI (left) and the 2-s ISI (right). Each successive record is aligned with the stimulus contact period, which is indicated in the top trace. These records show that the first one or two stimuli in each trial evoked little or no response, even with this high temperature. Successive stimulus contacts elicited a markedly increasing response until a plateau was reached. The responses to successive contacts did not overlap, because the response to each contact was complete before the onset of the next contact, even at the end of the trials. (Thus bin times equal to each ISI were used for averaging.) In addition, these records show clearly that the spike discharge to each brief contact stimulus usually lagged the stimulus onset by >1 s (that is, occurred after the stimulus offset), consistent with activation of this HPC lamina I STT cell by C-nociceptors. This delay was observed in every cell that showed such a summating response. For comparison, the bottom half of Fig. 2 illustrates the original records from NS lamina I STT cell it23u6, which showed a weak response (see following text) with a latency following stimulus onset of <100 ms, consistent with activation by A-nociceptors. Note that the temporal pattern of the HPC cell responses is consistent also with the possibility of active inhibition during the period of the mechanical contact.

View larger version (25K): [in this window] [in a new window]   Fig. 2. Top half: the neural spike records for polymodal nociceptive (HPC) unit it32u2 for each of the 1st 12 stimulus contacts of the 3-s interstimulus interval (ISI) series (left) and the 2-s ISI series (right) at 58°C. The records are aligned with the stimulus contact time shown in the top trace and are shown in serial order from top to bottom. Whereas no response was elicited by the initial contacts, the responses quickly augmented and then reached a plateau. The responses occurred at a latency of ~1-1.5 s following the stimulus contact, which lasted 0.7 s. Decrementing spike size at increased discharge rates is commonly observed in lamina I spinothalamic tract (STT) neurons [see Fig. 1A in the preceding paper (Andrew and Craig 2002; or Craig and Kniffki 1985)]. Bottom half: the neural spike record for nociceptive specific (NS) unit it32u6, shown with the same conventions as above. In contrast to the HPC cell, only a weak response was obtained that showed little augmentation, and the responses occurred at a latency of <100 ms following the stimulus contact. The same unit is shown in Fig. 6.

The family of curves displayed in Fig. 3 summarize the mean responses of all 16 nociceptive lamina I STT units to these stimuli, organized according to ISI at the 4 different temperatures. The same family of curves is displayed in Fig. 4, organized instead according to temperature at the three different ISIs. For comparison, at the top of each figure the corresponding graph is shown that describes the psychophysical sensation of second pain experienced by human subjects in response to the equivalent stimuli, as reported by Vierck et al. (1997). For this comparison, it is important to note that the peak responses of laminae I and V STT cells elicited by stimulation of glabrous skin in the primate occur at ~53°C (Craig, unpublished results; Surmeier et al. 1986; see also Campbell and Meyer 1996), whereas for cat lumbosacral lamina I STT neurons they occur at ~58°C (Craig and Serrano 1994; Craig et al. 2001); this difference is probably due to differences in the thermal conductance properties of cat and primate glabrous skin.

View larger version (14K): [in this window] [in a new window]   Fig. 3. The family of mean response curves for the entire population of 16 lamina I STT units, showing the evoked discharge frequency following each stimulus contact, for each ISI at each stimulus temperature (°C). For comparison, at the top is the graph showing the dependence on ISI of the human sensation of second pain, reproduced from Vierck et al. (1997), with permission.

View larger version (17K): [in this window] [in a new window]   Fig. 4. The family of mean response curves for the entire population of 16 lamina I STT units, showing the evoked discharge frequency following each stimulus contact, for each stimulus temperature (°C) at each ISI. For comparison, at the top is the graph showing the dependence on stimulus temperature of the human sensation of second pain, reproduced from Vierck et al. (1997), with permission.

The mean data illustrated in Fig. 3 show that at the lowest temperature used (46°C), no response was elicited from lamina I STT cells, and that at the next highest temperature (49°C), a response was obtained only after 8-10 stimulus contacts at the shortest ISI (2 s). However, at the higher temperatures (53 and 58°C), there was a progressive increase that was clearly dependent on the ISI. At the highest temperature (58°C), the progressive increase in summation observed at 5-, 3-, and 2-s ISIs matches the progressive increase with decreasing ISI shown in the graph from Vierck et al. (1997); in both sets of curves, there is only a slow augmentation at the longest ISIs, whereas at the shortest ISI, there is a rapid increase in response, especially between the 3rd and 10th stimuli, which achieves approximately an 8-fold increase and then plateaus or declines slightly. Statistical analyses of the responses of the population of 16 lamina I STT cells using one-way repeated-measures ANOVAs (across stimulus contacts at different ISIs) confirm these qualitative descriptions: in brief, the three response curves at 58°C are all significantly different (P < 0.001), the two top curves at 54°C are different from each other (P < 0.001) and from the lowest curve (P < 0.01), the top curve at 49°C is different from the bottom two (P < 0.001), and all of the indicated curves have a positive correlation with stimulus number (trend analysis, P < 0.003), whereas the lowest curves do not (P > 0.07 or more).

In Fig. 4, the same family of curves viewed according to temperature shows that there was considerable summation only for ISIs 3 s, and that at the shortest ISI, there was a graded, direct dependence on temperature. Again, the correspondence with the psychophysical data reported by Vierck et al. (1997) and shown in the top panel is clear. Similarly, statistical analyses support the visual impressions of these graphs: the four curves at 2-s ISI are all different (P < 0.001), and the top three show a positive trend with stimulus number (P < 0.001); the two top curves at 3-s ISI are different (P < 0.001) and show a positive trend (P < 0.001), and the top curve (58°C) at 5-s ISI is different (P < 0.001) from the others and has a positive trend (P < 0.003).

In trials on four units studied early in this experimental series, we applied the repeated brief contact heat stimuli several times, and we varied the order of the ISI and temperature presentation sequences; the patterns of the responses were reproducible with no evidence of significant order effects, using 3- to 5-min intertrial intervals (ITIs). However, there were differences in response patterns related to the classification of the 16 nociceptive lamina I STT units, as described further in the text.

Skin temperature measurements

We measured the effects of these stimuli on skin temperature with a thermocouple placed directly on the hindpaw in two animals or with an intradermal thermocouple inserted ~1 mm below the surface of the central pad in one animal. Similar results were obtained in all three cases. Figure 5 presents the families of curves that show the means of the peak temperatures achieved during each brief contact stimulus at the shortest ISI or the highest temperature. These graphs reveal that the peak skin temperatures increased with successive stimuli during each trial, and that they increased slightly more for increasing stimulus temperatures and decreasing ISIs, as would be expected. However, these curves all showed similar trajectories, and there were only slight differences between them, in contrast to the nonlinear augmentation observed in the neural responses. This contrast indicates that the temporal summation observed in lamina I STT cells cannot be explained by the increases in the temperature of the skin. Furthermore, the highest peak skin temperatures achieved did not reach the mean noxious heat threshold of nociceptive lamina I STT units as measured with fast-rising 20-s heat pulses (Craig et al. 2001), although lower thresholds probably would be measured with slow-rising heat stimuli (Yeomans and Proudfit 1996). These temperature measurements are consistent with the comparable measurements made in primate skin using repeated brief contact heat stimuli by Vierck et al. (1997) (see Fig. 8).

View larger version (21K): [in this window] [in a new window]   Fig. 5. The peak skin temperature values measured in 3 cases during each stimulus contact for each temperature at the shortest ISI (top panel) and for each ISI at the highest temperature (bottom panel). The temperatures were measured with a thermocouple placed on the glabrous skin (n = 2) or ~1 mm below the surface (n = 1), which produced similar numbers. These curves do not parallel the summation observed in the lamina I STT cells.

Differentiation of HPC and NS cells

The response curves of all individual units were examined, and a clear difference between HPC and NS cells was immediately apparent. Of the eight HPC lamina I STT units, all showed progressively increasing response summation at 54 and 58°C. Five of these eight HPC units also showed summation at 49°C (e.g., Fig. 1). In stark contrast, most (5) of the eight NS lamina I STT units showed no summation at all, and in the three that did, only weak (less than 3-fold) summation was observed, only at the highest temperature (58°C), and only after six or seven stimuli.

The middle panels in Fig. 6 document the original response records for an individual HPC cell (it32u2) and an individual NS cell (it23u6) at all three ISI trials at the highest stimulus temperature (58°C). These records clearly show the enormous enhancement of the HPC cell's response at the shorter ISIs (for comparison, note that this is the same unit illustrated in Figs. 1 and 2). The NS cell illustrated in this panel is the cell that showed the greatest enhancement among all NS cells at the 2-s ISI; all other NS cells had even weaker responses. For completeness, the traces in the bottom panels of Fig. 6 show responses to the standard quantitative thermal stimuli used for characterization (HPC left, NS right); as described quantitatively by Craig et al. (2001), HPC cells respond to noxious cold below a median threshold of ~24°C, and both HPC and NS cells respond in a graded fashion to 20-s noxious heat pulses, with NS cells having lower median thresholds.

View larger version (32K): [in this window] [in a new window]   Fig. 6. Comparison of the mean response curves of 8 HPC lamina I STT cells (left) and 8 NS lamina I STT cells (right) following each stimulus contact for each ISI at the highest stumulus temperature (58°C). The top curves for HPC and NS cells are significantly different (ANOVA) beginning with the 4th stimulus contact (post hoc t-tests). The error bars indicate SDs. In the middle are shown the original records for HPC cell it32u2 (left) and NS cells it23u6, with the histogrammed response (1-s bins), the spike record, and the stimulus record. Afterdischarges (left) following these trials always waned within 1-2 min, that is, before the next trial. At the bottom are shown the representative thermal stimulus-response functions that characterize an HPC cell (left; sensitive to noxious cold and noxious heat, binned at 1 s) and an NS cell (right; sensitive to noxious heat at a lower threshold).

The top panels in Fig. 6 illustrate the mean response curves of all HPC cells (left) and all NS cells (right) at the highest temperature used (58°C) for the three different ISIs; the error bars indicate the SDs of the population responses. The population of HPC units showed strong temporal summation, with a clearly enhanced discharge that increased progressively with stimulus number and with decreasing ISI. The summation was especially evident between the 3rd and 10th contact. In contrast, the mean response of the NS units showed only a weak and much delayed enhancement, even at the shortest ISI at this high stimulus temperature.

Statistical comparison of the maximal curve in each graph (i.e., 58°C at the 2-s ISI) revealed that the population responses of the HPC cells and the NS cells are significantly different (repeated measures ANOVA, P < 0.0001); the individual response points within these curves are different (from P < 0.004 to P < 10⁵, post hoc t-tests) beginning with the fourth stimulus. In Fig. 7, these two curves are compared directly with a graph of psychophysical reports from Vierck et al. (1997), which shows a clear correspondence between the maximal curve for HPC units and their maximal curve for second pain sensation elicited by the repeated brief contact heat paradigm, and also a clear correspondence between the maximal curve for NS units and their maximal curve for first pain sensation. The only apparent difference between the mean HPC lamina I STT cell activity and the mean human second pain report is the decrement in the HPC response after stimulus 12. 

View larger version (21K): [in this window] [in a new window]   Fig. 7. Comparison of the mean response curves of HPC and NS lamina I STT cells (bottom) to the maximal repeated brief contact heat stimulus trial with the mean report of human subjects to the maximal trial in the study of Vierck et al. (1997), reproduced with permission.

Reset phenomenon

We systematically examined the responses of three HPC lamina I STT units to paired stimulus trials, to determine whether they displayed the same reset phenomenon observed by Vierck et al. (1997). In this phenomenon, human subjects reported that the temporal summation produced by repeated brief contact heat stimulation is reset following very short ITIs. That is, Vierck et al. (1997) showed that a series of heat stimuli of sufficiently high temperature and short ISI produce a nonlinear temporal enhancement of the second pain report, but that a brief delay eliminates this enhancement, so that the pain due to an immediately subsequent trial (after the ITI) begins again at baseline and then generates a large enhancement again.

Figure 8 compares the responses of the same HPC cell illustrated earlier (it32u2, bottom left) to closely repeated trials at a 3-s ISI with the corresponding chart (top left) from the study by Vierck et al. (1997). The data from Vierck et al. show that the summating human pain report is reset to baseline at short ITIs of 5 and 6 s. Like the human psychophysical reports, the HPC lamina I STT cell displayed a clear summation that reset to baseline following short ITIs of 6, 9, and 12 s (i.e., 1, 2, or 3 skipped thermode contacts). In other words, following the ITI, the response of the cell to the brief contact heat stimulus returned to a level near baseline and then showed the same pattern of summation as it had to the preceding stimulus sequence. This was reproducible on subsequent trials. Thus the cell showed a reset phenomenon that directly corresponds with the reset of the human second pain report shown in the top left panel observed by Vierck et al. (1997). This reset phenomenon was observed in all three HPC cells that we tested, at both 3-s and 2-s ISIs.

View larger version (27K): [in this window] [in a new window]   Fig. 8. The "reset" phenomenon. At the top are shown graphs from Vierck et al. (1997) (reproduced with permission) showing summation in the human 2nd pain report that begins again near baseline following brief intertrial intervals (ITIs; left) and the skin temperature (right), which does not. At the bottom are shown comparable measurements for unit it32u2; the summating response to repeated brief contact heat stimuli began again near baseline following brief ITIs (left), but the skin temperature did not (right).

For comparison, the graph in the bottom right panel of Fig. 8 shows measurements of peak and baseline skin temperatures (immediately following stimulus contacts) from a thermocouple within the dermis of the glabrous pad of the cat's hindpaw during the reset trials for the unit illustrated on the left. The comparable measurements in human skin from Vierck et al. (1997) are presented in the top right panel. The temporal profile of the temperature of the skin, both the cat hindpaw and the human glabrous hand, is entirely different from that of the HPC cell's response and the human pain report. In both cases, the peak (or the baseline) temperature achieved a plateau much more rapidly than the augmenting response. Furthermore, after the brief ITI, the skin temperature in both cat and human remained elevated from the preceding stimulus trial, which directly contrasts with the reset phenomenon observed in the corresponding profiles of the HPC cell response and the human psychophysical reports.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present results demonstrate that HPC lamina I STT cells display responses to repeated brief contact heat stimuli that parallel the corresponding psychophysical data in humans for second pain sensation (Vierck et al. 1997). In contrast, NS cells do not; instead, they display responses that parallel the corresponding data for human first pain sensation in the same paradigm. Together with the converse parallels with mechanically induced first pain described in the preceding article (Andrew and Craig 2002), our observations provide strong evidence distinguishing NS and HPC lamina I STT cells, and they reveal a double dissociation indicating that NS and HPC lamina I STT cells can be directly associated with qualitatively different aspects of pain sensation, that is, with first (sharp) and second (burning) pain, respectively. The following text compares the findings from the present study with prior physiological results within the context of the neural basis for second pain sensation and then addresses the significance of these combined observations for our understanding of the fundamental role of the lamina I STT projection.

Physiological basis of second pain sensation

It is generally agreed that a sudden noxious event, such as a brief heat pulse, an acute impact or an electrical shock, elicits two qualitatively distinct pain sensations, a short-latency sharp or pricking sensation and a longer-latency dull or burning sensation, which are denoted as "first" and "second" pain (Campbell and Meyer 1996; Lewis and Pochin 1937; Price 1988). Findings based on latency measurements, nerve blocks, and microneurographic recording and stimulation support the association of these distinct sensations with A- and C-fiber activity, respectively (Mackenzie et al. 1975; Ochoa and Torebjörk 1989). They can be distinguished experimentally by different stimulus paradigms. Maintained mechanical stimulation with a small probe produces a sensation of first, sharp pain that is pricking or stinging (Adriansen et al. 1984; Andrew and Greenspan 1999). Repeated noxious heat pulses generated with a radiant heat lamp (Hardy et al. 1952) or a thermoelectric element (Price et al. 1977) at ISIs 3 s cause an augmenting sensation of second, burning pain that is unpleasant and distressing. Price and colleagues confirmed J. D. Hardy's original conjecture that the temporal summation of second pain occurs centrally, by showing that the summation occurs even with alternating stimulation on different sites and is reduced by dextromethorphan, a central analgesic (Price and Dubner 1977; Price et al. 1977).

Our understanding of the temporal summation of second pain was advanced significantly by Vierck et al. (1997), who developed the method we refer to as the repeated brief contact heat paradigm. It had previously been thought that repeated mechanical contact would inhibit pain sensation (a tenet of the "gate control theory" of pain, or an "epicritic" effect on "protopathic" sensation), yet they showed that a hot stimulus that was perceived only as warm if contacted briefly would, if contacted repeatedly, build up a profound burning pain sensation and yet elicit only a delayed, weak sensation of sharp pain. They acquired a family of curves that demonstrated the dependence of the temporal summation of second pain on temperature and ISI and its dissociation from skin temperature. Most strikingly, they discovered an important reset phenomenon, whereby this temporal summation is reinitiated from baseline following very short intertrial delays (the omission of even a single brief contact) in normal subjects. Increasing activation of C-nociceptors and the summation of second pain with repeated stimulation are probably important aspects of neuropathic and central pain (Price 1988; Vierck et al. 1997, 2000); indeed, dysfunction of this reset phenomenon could conceivably underlie the so-called "wind-up" pain often described in chronic pain patients.

The present data show that HPC lamina I STT cells in cats display activity that parallels in all respects the augmenting second pain sensation elicited by the repeated brief contact heat paradigm. The correspondence is clear with regard to temperature, ISI, degree of augmentation, and temporal profile. The responses of HPC cells show a similar overall pattern of temporal summation, with a similar degree of summation at similar ISI and temperature parameters. They show the same dependence on stimulus temperature and inverse dependence on ISI. They show the same pattern of progressive augmentation between the 3rd and the 10th stimuli. The present recordings were obtained in barbiturate-anesthetized cats (see DISCUSSION in Craig et al. 2001), which might explain the slight decrement late in the mean HPC response, yet the cross-species comparison certainly indicates that HPC lamina I STT cells can provide the basis for the augmenting human second pain sensation resulting from repeated brief contact heat stimulation. Most strikingly, there is an unmistakable and convincing correspondence in that they display the same reset phenomenon at short ITIs.

In contrast, NS cells do not show such augmenting activity, and their discharge parallels the pattern of weak first pain described by Vierck et al. (1997) in response to the repeated brief contact heat stimulus. This corroborates their association with first pain by our findings using maintained mechanical stimulation described in the preceding paper (Andrew and Craig 2002). Together, these observations are consistent with the dominance of NS lamina I STT cells by A-nociceptor input and the dominance of HPC cells by monosynaptic polymodal C-nociceptor input (Craig and Kniffki 1985; Craig et al. 2001).

The correspondence between the present findings and those of Vierck et al. (1997) supports the conclusion that activity in cutaneous polymodal nociceptive HPC lamina I STT neurons produces a burning pain sensation, and this is consistent with other evidence. Prior physiological and psychophysical findings indicate a correspondence between HPC activity and the burning sensation evoked by the thermal grill illusion of pain or by noxious cold (Craig and Bushnell 1994), which is similar psychophysically to the burning sensation caused by noxious heat (Boring 1942; Morin and Bushnell 1998). The direct association of an unpleasant burning pain sensation with the activity of polymodal C-nociceptors that is conveyed by HPC lamina I STT cells is also strongly supported by the observation that heat, cold, and (most tellingly) pinch all evoke a burning or dull pain sensation when only C-fiber conduction remains during a peripheral nerve block (MacKenzie et al. 1975; Ochoa and Torebjörk 1989).

This conclusion is necessarily qualified by the recognition that the forebrain integration of HPC activity with other types of ascending activity (including NS cells) remains to be determined. Under normal circumstances, HPC cells are not the only type of ascending cell activated by noxious cutaneous stimuli. Further, our prior work with the thermal grill illusion demonstrated that HPC lamina I STT activity is integrated at the thalamocortical level with the activity of COOL lamina I STT cells (Craig and Bushnell 1994; Craig et al. 1996). Yet, the present findings in cat offer for the first time a clear and parsimonious explanation for the observations of Tommerdahl et al. (1996), who showed in monkeys that repeated brief contact heat stimulation causes augmenting activation of neurons in cortical area 3a and reduced activity in areas 3b and 1. The present findings can explain their results, because in primates nociceptive lamina I STT activity is relayed topographically by a specific thalamic nucleus (VMpo) to the dorsal margin of posterior insula and to area 3a (see Craig 2000a) (note that these physiological observations by Tommerdahl et al. in area 3a of monkey cortex have been confirmed in this laboratory: Craig, unpublished results). We are not aware of any other similar observations in the thalamus or cortex of monkeys or humans.

The present data contrast with earlier work by Price and colleagues (Price et al. 1978), who proposed that a different population of STT cells might be the basis for second pain. Using a sequence of four 1.5-s triangular heat pulses produced by a thermoelectric element at a 3-s ISI, they reported that NS and wide-dynamic-range (WDR) STT cells in both laminae I and V of monkeys displayed a decrementing "early" response and an incrementing "late" response. They associated such responses with first and second pain, respectively, tacitly presuming that the same neurons could generate two qualitatively different sensations in rapid succession. Because they found that only WDR STT cells also showed long afterdischarges to a slow brush stimulus, they concluded that WDR neurons must be important for both the temporal summation and the persistence that characterize pain sensation. In a subsequent study in spinalized rats, Price and colleagues reported that only deep WDR cells, and not superficial NS cells, encoded pain intensity over long durations (45 min) of repetitive (5 s on-off) noxious heat stimulation (Coghill et al. 1993). Their interpretation was thought to be consistent with the "gate control theory" of pain and with the widely held view that WDR lamina V STT cells that project to the ventroposterior nucleus (VP) and thence to S1 cortex have an essential role in human pain sensation (Price 1988; Willis and Westlund 1997).

It must be noted immediately that there are significant methodological differences between the present study and the work of Price et al. (1978). They applied repeated heat pulses with a stationary thermode, whereas we used repeated brief contact heat stimuli. Also, at the time of their work, neither lamina I STT cells that project to the main lamina I thalamic projection target (VMpo) in primates nor HPC lamina I cells had been recognized yet. These differences may be pertinent to our contrasting observations; indeed, the characteristics of HPC cells can explain other discrepancies in the prior literature (see Craig 2000b,c). Nevertheless, it is particularly noteworthy that the actual data they illustrated on the so-called "late" response of STT cells to four consecutive heat pulses (their Fig. 6B) showed an increase with the second pulse but then a flat plateau across the second, third, and fourth stimuli. This quite clearly differs from the progressively augmenting second pain sensation that they reported psychophysically with the same stimulus and that Vierck et al. (1997) reported with the repeated brief contact heat paradigm. Yet, they interpreted these responses as parallel to the sensation, apparently because of the stark contrast with the progressive decrement in C-fiber responses they had recorded with the same stimulus. This discrepancy, which contrasts with the clear and comprehensive correspondence observed in the present experiments between HPC activity and the human sensation of second pain, strongly recommends that the repeated brief contact heat paradigm of Vierck et al. (1997) be used to reexamine and compare the responses of lamina I and lamina V STT cells in the monkey.

For such a comparison, the reset phenomenon revealed by the work of Vierck et al. (1997) and displayed by HPC lamina I STT cells provides a particularly key test, because this striking phenomenon clearly differs from the long-term, persistent "windup" characteristically induced by repetitive C-fiber activation in WDR lamina V cells. Such windup has been thought by many investigators to be related to central sensitization, secondary hyperalgesia and neuropathic pain (see Li et al. 1999; Price 1988; Willis 1997; Woolf 1996; Woolf and King 1987), but its temporal persistence is inconsistent with the rapidity of the reset phenomenon. This contrast validates the need for a reconsideration of the role of the modality-ambiguous WDR lamina V STT cells in sensation (see Craig 2000b,c). Indeed, preliminary data already obtained in this laboratory with the repeated brief contact heat paradigm suggest that WDR lamina V STT neurons in monkeys respond with an immediate increment and then a plateau, in agreement with the actual data of Price et al. (1978) described above; furthermore, they maintain the same plateau across a brief intertrial delay, that is, they do not display the reset phenomenon, in stark contrast to the responses of HPC lamina I neurons and the human second pain sensation elicited by repeated brief contact heat. Thus the findings in the present study indicate that the activity of HPC lamina I STT cells corresponds directly, and perhaps uniquely, with the C-fiber-induced sensation of second pain.

Fundamental role of lamina I neurons

Analyses of the functional anatomy of lamina I neurons have led to the concept that they constitute an interoceptive (or, homeostatic) afferent pathway comprising several distinct sensory channels that carry modality-selective activity representing different aspects of the physiological condition of all tissues of the body to sensory and homeostatic integration sites and, in primates and humans, to a direct cortical representation of the condition of the body (Craig 2000a). At present, we recognize HPC, NS, thermoreceptive-specific (COOL, WARM), deep-responsive (muscle, joint), histamine-(itch)-selective, and mustard oil-(chemo-noci)-selective lamina I STT cells, and there are likely several other classes that remain to be documented more clearly, such as visceroreceptive, metaboreceptive, and C-fiber mechanoreceptive (Andrew and Craig 2001a,b; Craig et al. 2001; Wilson et al. 2002). These robust and biologically distinct classes of ascending sensory neurons differ in multiple respects, e.g., their afferent and descending inputs, their projections, and their functional, morphological, and chemical properties. The present studies and the prior evidence support, and in our view compel, the conclusion that the activity in these discrete classes of neurons corresponds directly to psychophysically distinct sensations (cool, warm, itch, first pain, second pain, muscular aches or cramps, and so on), that is, that these classes constitute qualitatively distinct sensory channels, or "labeled lines."

This is clearly the only reasonable conclusion for thermal or itch sensations, because the lamina I spinothalamocortical projection to dorsal posterior insular cortex in primates and humans provides the only known ascending pathway with functional and anatomical characteristics that specifically parallel these sensations (Andrew and Craig 2001a,b; Craig et al. 2000; Drzezga et al. 2001). However, the idea that pain is a specific sensation represented by specific neural elements has, for at least 100 yr, been fervently contentious [see reviews: Fields 1987; Melzack and Wall 1983; Perl 1984, 1996; Rey 1995; e.g., adherents of specificity were accused of being a "cult" by Wall (1995)].

The chief empirical objection to the idea of specificity in the field of pain research is that, under pathological conditions, pain can be elicited by touch (i.e., mechanical allodynia); therefore many have concluded that pain must result from integration with the tactile senses. Thus the view has been widely held for the past 25 years that "multireceptive" neurons within the somatosensory system are responsible for pain, and that WDR lamina V STT cells are the convergent pain-and-touch cells proposed by the "gate control theory" (Melzack and Wall 1983; Willis and Westlund 1997).

The contrasting view is that pain and touch are subserved by different neural systems. The essential conceptual difference, in our view, that is compelled by the available evidence on the lamina I STT projection system is the recognition that pain is a specific aspect of interoception, instead of exteroception (Craig 1996, 2000a; Craig et al. 2000). Pain is a sensation related to the condition of the body, rather than to the properties of objects that contact the body (exteroception) or to the movements of the body (proprioception), and it arises from any tissue of the body, of which skin is but one example. In subprimate mammals, pain is a motivational (distress) signal indicating that homeostatic (autonomic) control systems cannot rectify the physiological condition of a portion of the body, i.e., that homeostasis is imbalanced, and it drives necessary compensatory behaviors that affect life-critical processes extending from cardiorespiratory activity to food and water intake to social interactions, i.e., the goal-directed homeostatic behaviors guided by the so-called limbic system. The spinal and brain stem projections of lamina I directly indicate this association with homeostasis. Encephalization in primates, and most particularly in humans, produced a highly resolved interoceptive representation in the cortex, a sensory image of the physiological condition of the body, that is provided by the direct lamina I spinothalamocortical projection (by way of VMpo) and the parallel projection from the nucleus of the solitary tract (by way of VMb) (Beckstead et al. 1980). In thalamus and in cortex, this pathway is topographically organized orthogonal to the tactile representation in the exteroceptive somatosensory system, to which it is connected at the representation of the tongue and mouth. The interoceptive representation comprises several discrete modalities that engender sensations, including pain, temperature, itch, tickle (sensual touch), and deep and visceral sensations (muscle ache, air hunger, cardiovascular activation, vasomotor flush, and so on). This concept emerges directly from the functional anatomical findings, but its fundamental philosophical features were recognized by authors such as W. James (1890) and Sir Charles Sherrington (1900); for example, the latter considered pain not as an aspect of touch but as part of the sense of the "material me" (and later coined the term "interoception").

According to this concept, the confound of touch-evoked neuropathic pain (allodynia) would be explained by abnormal low-threshold activation of otherwise specifically nociceptive neurons within the lamina I spinothalamocortical system. In fact, such sensitization of lamina I STT neurons has been described (Craig and Kniffki 1985; Woolf et al. 1994), and recent anatomical results indicate that the neural basis for allodynia in a behavioral model of neuropathic pain may be direct activation of nociceptive lamina I projection neurons by low-threshold mechanical stimulation (Bester et al. 2000). Although it is not yet clear to what extent this may occur by disinhibition, sensitization, or anatomical plasticity (Blomqvist and Craig 2000; Cervero and Laird 1996; Wasner et al. 1999), these data strongly indicate that allodynia can result from low-threshold activation of the specific, interoceptive pain system, rather than from dysfunctional integration within the tactile, lemniscal system.

Thus the findings described in this and the preceding paper, indicating that first and second pain can be regarded as qualitatively distinct sensations subserved by distinct neural elements both peripherally and centrally, are consistent with this fundamental concept of the role of the lamina I STT projection system. Conceptual recognition of these sensory channels as unique components of an interoceptive (homeostatic) afferent system could simplify the interpretation of behavioral and clinical data on the characteristics of acute and chronic pain and on the integration of pain with homeostatic processing. Within this conceptual context, the present findings suggest that these distinct neurobiological channels probably engage somewhat different sets of forebrain neurons, initiate different behavioral and homeostatic drives, and are subject to different descending controls. The identification of HPC cells as a substrate for the augmenting sensation of second pain, with an unpleasant, distressing character, could have particular clinical significance for chronic pain, which is often characterized by temporally summating ("windup") pain; as noted above, the mechanisms underlying the reset phenomenon could include inhibitory processes that might be dysfunctional in central or neuropathic pain (see Pagni 1998), and thus an understanding of the pharmacology of this phenomenon in HPC cells could be critical for the development of novel therapies.

Finally, many issues remain to be explored. These two nociceptive sensory channels must be integrated hierarchically in the brain stem and in the forebrain with each other (when co-active) and with other ongoing interoceptive (homeostatic) modalities, such as thermal sensation, itch, and metaboreceptive activity representing salt and glucose levels or immune system function (Bickford 1938; Craig and Bushnell 1994; Craig et al. 2001; Critchley et al. 2001; Damasio et al. 2000; Saper 2000; Watkins and Maier 2000). These levels of integration all remain to be elucidated. As noted earlier, the functional relationship of these ascending lamina I STT projections in the forebrain with the concomitant activity of other types of ascending neurons needs to be clarified. The HPC and NS categories clearly contain subclasses, such as HPC cells dominated by input from one submodality or tissue layer and NS cells sensitive only to pinch or only to heat, which could have differentiable functional roles (Andrew and Craig 2002; Craig and Kniffki 1985; Craig et al. 2001). In contrast to innocuous thermal sensation, nociceptive lamina I neurons coactivate insular cortex, area 3a, and area 24c, at least, and the integration of multiple coactivated forebrain sites is not understood. Last, further comparison of the activity of these neurons with other psychophysical aspects of human pain sensation (e.g., Bushnell et al. 1984; Robinson et al. 1983) and the interactions with other modalities will help determine how closely our perceived sensations reflect activity in these distinct sensory channels.

Conclusions

The present findings demonstrate that HPC lamina I STT neurons have unique characteristics that parallel the augmenting second pain sensation that humans perceive in response to the repeated brief contact heat paradigm of Vierck et al. (1997). In contrast, NS lamina I STT cells show a response profile that matches the first pain sensation reported with this paradigm. These associations are consistent with the predominant A-nociceptor input to NS cells and the predominant polymodal C-nociceptor input to HPC cells. Together with the converse responses to maintained mechanical stimulation reported in the preceding article (Andrew and Craig 2002), these findings provide a double dissociation indicating that NS and HPC lamina I STT neurons are robust and distinct classes that provide substrates for qualitatively distinct aspects of pain sensation. The present results add further support for the concept that lamina I STT neurons convey modality-selective activity in discrete sensory channels (virtual "labeled lines") that are integrated in the forebrain to produce highly resolved interoceptive sensations, such as pain, temperature, and itch, related to the physiological condition of the body. These results in cat strongly recommend deeper analyses of STT cells in primates and investigation of the integration of these unique sensory channels at higher levels with other aspects of homeostatic activity and with other pathways.