Ensemble encoding of nociceptive stimulus intensity in the rat medial and lateral pain systems
© Zhang et al; licensee BioMed Central Ltd. 2011
Received: 11 May 2011
Accepted: 24 August 2011
Published: 24 August 2011
Skip to main content
© Zhang et al; licensee BioMed Central Ltd. 2011
Received: 11 May 2011
Accepted: 24 August 2011
Published: 24 August 2011
The ability to encode noxious stimulus intensity is essential for the neural processing of pain perception. It is well accepted that the intensity information is transmitted within both sensory and affective pathways. However, it remains unclear what the encoding patterns are in the thalamocortical brain regions, and whether the dual pain systems share similar responsibility in intensity coding.
Multichannel single-unit recordings were used to investigate the activity of individual neurons and neuronal ensembles in the rat brain following the application of noxious laser stimuli of increasing intensity to the hindpaw. Four brain regions were monitored, including two within the lateral sensory pain pathway, namely, the ventral posterior lateral thalamic nuclei and the primary somatosensory cortex, and two in the medial pathway, namely, the medial dorsal thalamic nuclei and the anterior cingulate cortex. Neuron number, firing rate, and ensemble spike count codings were examined in this study. Our results showed that the noxious laser stimulation evoked double-peak responses in all recorded brain regions. Significant correlations were found between the laser intensity and the number of responsive neurons, the firing rates, as well as the mass spike counts (MSCs). MSC coding was generally more efficient than the other two methods. Moreover, the coding capacities of neurons in the two pathways were comparable.
This study demonstrated the collective contribution of medial and lateral pathway neurons to the noxious intensity coding. Additionally, we provide evidence that ensemble spike count may be the most reliable method for coding pain intensity in the brain.
The ability to evaluate the intensity of painful stimuli is a major property of the nociceptive system. Multiple studies have shown that discrimination of nociceptive stimulus intensity is the function of the lateral pain pathway, which participates in the sensory discrimination of pain [1–3]. Kenshalo et al. concluded that nociceptive neurons in the primary somatosensory cortex (SI) are involved in the encoding process by which monkeys perceive the intensity of noxious thermal stimulation . Apkarian and Shi found that the proportion of responsive neurons in the lateral thalamus of squirrel monkey was increased with the noxious stimulus intensity .
In recent years, more and more studies accepted that the medial pain pathway could also have the ability to encoding the noxious stimulus intensity . Coghill et al. employed a fMRI study of human subjects and suggested that a highly distributed, bilateral supraspinal mechanism engaged in the processing of pain intensity, including the brain areas in both the lateral and medial pain pathways, such as somatosensory cortex, anterior cingulate cortex (ACC) and the dorsomedial and centromedian thalamus . Buchel et al. also provided evidence using human fMRI that the ACC have a role in coding pain intensity . In an electrophysiological study of rat primary sensorimotor cortex (SmI) and ACC, Kuo and Yen have investigated the single-unit responses and found a noxious intensity-related activation in the recorded areas .
Although many studies have proved the ability of noxious intensity coding in the supraspinal pain circuits, it is still unclear about the neural encoding patterns in both pain pathways, and whether the coding ability of the medial pain system is comparable to that of the lateral pain system. In this study, a multiple single-unit recording technique for parallel recording of thalamo-cortical activity in the rat medial and lateral pain pathways was employed. Nociceptive responses were elicited in the hindpaw by laser pulses with progressively increasing intensity. Spike counts as well as several other commonly used coding schemes were computed to reveal the intensity coding strategy in the parallel pain pathways.
Number of neurons responding to stimuli of different intensities
Laser energy intensity (mJ)
Ratios of WDR neurons in each brain region of interest
In the present study, we investigated how the intensity of pain stimuli is encoded by neural ensembles within central sensory and affective pain pathways. We used a multichannel single-unit recording technique to measure neuronal activity within the SI, VPL, ACC, and MD in response to laser-evoked pain in the rat paw. The main findings of the present work were that the noxious laser stimulation evoked uniform double-peak responses in all of the recorded brain regions. The coding capability of the medial (affective) pathway neurons for noxious stimulus intensity can be comparable to that of the lateral (sensory) pathway, demonstrated by various analysis methods including the neuron number coding, firing rate coding, as well as mass spike count (MSC) coding. Furthermore, our data indicate that the MSCs best represent the coding strategy of neurons for the noxious intensity differentiation in the parallel pain pathways.
The present study showed that the laser stimulus evoked various types of single-unit responses in the four recorded regions, including single-peak, double-peak, brief, as well as long-lasting neuronal activity. However, computing the mean firing rate revealed uniform double-peak responses in all the brain regions with short latency (< 100 ms) and long latency (300-500 ms). The results were consistent with previous studies by Kuo and Yen in which they observed short- and long- latency laser-evoked cortical responses in the rat primary sensorimotor cortex (SmI) and ACC . The laser-evoked early and late peaks have been generally thought to coincide with the perception of the first (fast or pricking) and second (slow or burning) pain, respectively, since the laser beam could selectively and synchronously activate cutaneous thermonociceptors which conduct in the Aδ and C fiber ranges .
The SI has been reported to be involved in the C-fiber-mediated nociceptive processing in different animal models. Altered sensory processing related to hyperalgesia has been reflected in the C-fiber laser-evoked potentials in the rat SI cortex . By recording the nociceptive C-fiber evoked potentials from the surface of SI, Kalliomäki et al. demonstrated that following laser stimuli there are at least two different C fiber projections to SI, one somatotopically organized, and one more diffused organized input. Moreover, the spinal NMDA-receptor mechanisms may amplify the acute transmission of nociceptive C fiber input to SI in a frequency-dependent way [12, 13].
The experimental data is rare about the activation of medial pain system by the A-fibers. However, in a very recent study, by means of retrograde transneuronal transport of rabies virus, the electrophysiological evoked responses in ACC were recorded with such short latencies (less than 20 ms) . They also demonstrated that ascending somatosensory pathways to the ACC most likely originated from the myelinated Aδ and Aβ fibers but not from unmyelinated C fibers.
It has long been believed that the somatosensory cortex and sensory thalamic nuclei are responsible for the sensory discrimination of pain, including discriminant of the intensity, quality, duration, and location of painful stimuli [15–17]. In an early study investigating the thalamic neuronal response to noxious stimuli in macaque monkeys, Kenshalo et al. found that the discharge frequency of VPL neurons increased progressively in response to presentation of graded noxious heat stimuli . Exploring the effect of laser intensity on evoked potentials in several brain regions in surgical patients, including the primary somatosensory cortex, Ohara et al. showed that the peak amplitudes of both early (N2) and late (P2) pain-related components correlated with laser energy intensity . In the present study, we found that the mean firing rates, the number of responsive neurons, and the total spike counts in all of the four recorded regions constituting the pain network increased with laser stimulus intensity. Significant correlations were observed between these parameters and laser energy, suggesting the involvement of parallel pain pathways in the processing of noxious intensity. Thus, our results extend the previous findings to the single-unit level and underscore the importance of the sensory thalamus and cortex in coding noxious stimulus intensity.
On the other hand, in contrast to the convergence of data reported for the lateral pain pathway, the role of medial pathway neurons in encoding noxious intensity is still controversial. Several brain imaging studies from about a decade ago did not support a role for the ACC in coding pain intensity. Neither regional blood flow nor blood oxygenation level-dependent signal in the ACC correlated with noxious stimulus intensity [20–22]. Those imaging studies in human subjects conflicted with findings from animal studies using electrophysiological recording in monkeys and rabbits that provided evidence for the involvement of the ACC and the medial thalamus in noxious intensity coding [23–26]. More recently, an fMRI study using human subjects demonstrated pain intensity-related activation of the ACC and the dorsomedial and centromedian thalamus . These findings challenge the traditional view that the sensory-discriminative processing of pain is confined to the sensory pathway. In the current study, both neuronal and ensemble activation responses support a role of the medial pain pathway in pain intensity coding. The single-unit analysis of medial pathway neurons in the present study showed that firing rates, number of responsive neurons, and MSCs correlated with noxious stimulus intensity. Moreover, our ensemble data suggested that the medial and lateral pain systems may both be important for intensity discrimination. These results expand our understanding of pain perception by highlighting a role of the medial pain pathway.
Three kinds of coding methods were investigated in this study: neuron number coding, firing rate coding, and MSC coding. The former two are classic methods that have been widely used in sensory-related encoding studies [27–29]. The proportions of WDR neurons in the total sample of neurons in the SI and VPL were higher than those in the ACC and MD, suggesting that the sensory pathway may better encode nociceptive stimulus intensity than the affective pathway. This result seems to agree with the traditional view. For instance, when Kuo and Yen compared the neuronal responses to noxious laser-heat stimuli in the ACC and SmI, they found that ACC neurons were less responsive than SmI neurons .
On the other hand, we did not detect differences in coding efficiency between the medial and lateral pathways when MSC coding was used. This result is consistent with our discriminant analysis results. The correlation coefficients using spike count coding were obviously greater than those obtained when either frequency or number of coding neurons were used, indicating that the MSC coding strategy represented stimulus intensity better than the other two did. Because spike count analysis considers neuronal number and frequency coding simultaneously, there is no loss of information. Recent studies have shown that spike count is very useful, and more efficient and available, for neuronal encoding [30–33]. Luna et al. tested five possible candidate codes, including inter-spike intervals, mean spiking rate, mean burst rate, and absolute number of spikes or bursts, to determine the optimal neural code for discriminating two consecutive vibrotactile stimuli. They found that only the spike count code was consistent with the psychophysical measures . Another recent study successfully used the spike count code to find the best time scale for predicting monkeys' choices in a speed change detection and discrimination task . One might argue that the spike count code is, in essence, the same as the firing rate code. However, a normalized firing rate, such as that used in this study, does not take baseline neuronal activity into consideration, and thus may interfere with the assessment of coding capability. On the other hand, this result also indicates that the neuronal ensemble may represent the painful stimuli with a random/alternative manner, in which some neurons of the ensemble respond in one trial while other neurons in another trial. While keeping the MSC constant which bring about a steady representation of the stimulus intensity, this coding method has the advantage of giving neurons a chance to relax and restore the energy consumed. This will be extremely useful when dealing with repeated stimulation, as in the case of our current experimental protocol. Therefore, our results suggest that a strategy based on the MSC may be the best neural code for the central nervous system to use to discriminate different intensities of sensory stimuli.
In the present study, we analyzed the central coding strategies of nociceptive stimulus intensity within the parallel pain pathways at both single neuron and neural ensemble levels, and compared three types of coding schemes, i.e., neuron number coding, firing rate coding and mass spike count coding. We found significant correlations between laser intensity and the number of responsive neurons, the firing rates, as well as the mass spike counts (MSCs). The coding capacities of medial and lateral pain pathway neurons were comparable. In addition, the degree of correlation using MSC coding was higher than that with either frequency coding or number coding, suggesting that the MSC coding may be a more reliable method than neuron number coding or firing rate coding for elucidating nociceptive information transmission and processing in the brain.
Eight awake adult male Sprague-Dawley rats, weighing 250-300 g, were used. Animals were housed individually, with ad libitum access to food and water, under a reverse 12:12 light/dark cycle (the light phase commenced at 8:00 am). All experiments were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (U.S.A.). The research protocol was approved by the Institutional Animal Care and Use Committee of the Chinese Academy of Sciences. The protocol was designed to minimize pain and distress and the number of animals used in the study.
Four microelectrode arrays were implanted in the brain by making four small craniotomies. Animals were anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and then mounted on a Kopf stereotaxic apparatus (David Kopf Instruments, USA). Supplementary doses of anesthetic (one third of the original dose) were given when necessary. Coordinates for the craniotomies were determined according to a rat brain atlas . ACC was 3.2 mm rostral to Bregma (3.2 A), 0.8 mm lateral (L) to midline and 2.8 mm ventral (V) to the surface of the skull. SI was 1.0 mm caudal to Bregma (-1.0 A), 2.0 L and 2.0 V. MD thalamus was -2.3 A, 0.8 L and 5.5 V. VPL thalamus was -3.0 A, 3.0 L and 6.0 V. Four bundles of eight stainless steel Teflon-insulated microwires (50-μm diameter, Biographics, Winston-Salem, NC) were slowly lowered into the unilateral target areas. The microwires were connected to pins on the headset which was secured to the cranium with dental cement using skull screws as anchors. Ground wires were securely attached to the skull screws. Animals received penicillin (60,000 U, i.p.) before surgery and were housed individually in cages for 7 days while they were recovering from the surgical procedure.
Prior to the experiment, each animal was handled daily and habituated to a transparent plastic recording chamber (40 × 40 × 30 cm3) for 7 d. On each of the recording days, rats were allowed a 30-min period to become familiar with the chamber.
Stimuli (wavelength, 10.6 μm; beam diameter, 2.5 mm; pulse duration, 20 ms) were generated from a CO2 laser stimulator (Institute of Optical Precise Engineering and Physics, Chinese Academy of Sciences, China) in single-pulse mode [8, 9]. The CO2 laser beam projection was guided by a helium laser producing a red light spot. Laser pulses were applied to the plantar surface of the hind paw contralateral to the brain recording sites. Neuronal activity and observations of nocifensive behaviors (paw withdrawal or licking) were recorded. Each recording session consisted of about 40 trials with the laser pulses delivered at the same energy level. To minimize tissue damage, sensitization, and habituation, the stimulation site was varied and the interstimulus interval was ≥ 40 s. A total of 6 sessions were given. In each session the output energy of the laser was increased to a higher level (120, 130, 140, 150, 160 and 180 mJ). The laser pulse at the lowest intensity was perceived as a mildly painful sensation when applied to the experimenter's skin.
Neuronal activities were recorded extracellularly via stainless steel microwires and the signals were passed from the headset assemblies to a preamplifier. The signals were band-pass filtered between 0.5 and 5 kHz (6 dB cutoffs) before being sent to a multichannel spike-sorting device. Spike activities were monitored by a computer with a 20-μs time resolution and identified by setting proper boxes for amplitude and duration with a PC-based software Magnet (Biographics, Inc). The identity of a clearly sorted single neuron was verified by graphical capture of action potential waveforms which were inspected continuously to ensure that the same neuron was recorded during each experimental recording session. Waveform time stamps were stored in a database for off-line analysis with the commercially available PC-based programs Stranger (Biographics Inc., USA), Neuroexplorer (Plexon Inc., Dallas, TX, USA), Matlab (MathWorks Inc. USA), and Prism (GraphPad Software, Inc., USA). Behavioral responses were also recorded during each session using Magnet software and a video camera (Vigour, VG-2012, China).
Data were expressed as the percentage of paw withdrawal or licking, which were calculated according to the following formula: (number of stimulations producing paw withdrawal or licking/total number of stimulations) × 100%. The data were then analyzed by one-way analysis of variance (ANOVA), followed by a linear trend analysis across stimulus intensities. The accepted level of statistical significance was p < 0.05.
Bin counts for each trial were calculated using the analysis program Neuroexplorer. A 10-ms bin size was used for the computation of peri-stimulation time histograms (PSTHs) with a time range of -1.0 to +1.0 s relative to the stimulus onset. The firing rates for each unit and each time bin were then transferred into Z-scores: Z = (X - M)/S, where X is the actual firing rate obtained from PSTH, and M and S are the mean and standard deviation of the basal neuronal firing rate. An increase or decrease in firing rate that was more than two standard deviations from the baseline (-1 - 0 s) for at least three consecutive bins was considered to be an excitatory or inhibitory response. A cluster technique (K-means, SPSS) was used to sort neuronal responses based on the similarities in patterns of excitation or inhibition around stimulation events. Z-scores were arranged in clusters to visualize the response pattern of neuronal populations.
where N is the number of single units in a given brain area, and C i is the spike counts of the i th single unit in that area during a given time epoch around the stimulation.
The means and standard errors of these contributions from all rats were computed over time. Then the mean values of R all , R m and R l from 0 to 1 s post-stimulation were compared using one-way ANOVA.
After completion of the stimulation and recording sessions, rats received an overdose of sodium pentobarbital. Recording sites were marked by electrophoretically depositing iron (20 μA DC current, 20 s duration, and anode at the electrode) at the tips of selected wires. Animals were perfused with 4% paraformaldehyde and 5% potassium ferricyanide solution. The brains were post-fixed in a solution of 20% sucrose followed by a solution of 30% sucrose and then stored at 4°C until time of sectioning. Forty-micron-thick coronal sections were cut through the ACC, SI, and the MD and VP thalamus. The slides were stained with hematoxylin and eosin (H-E). Recording sites were determined under a light microscope. The iron deposits were easily identified as blue dots. Data from sites found to be outside the target areas were removed from the overall analysis.
FL is currently the Professor and Director of the Key Laboratory of Mental Health in the Institute of Psychology, Chinese Academy of Sciences; a committee member of the Chinese Association for the Study of Pain; first started the multichannel single-unit study of pain in behaving animals in China. JYW is the Associate Professor of the IPCAS, the first author of the earliest papers on ensemble neuronal pain responses in the thalamocortical pathways in behaving animals. YZ is a PhD student and NW a postdoc of FL. DJW is the Director of the Neuroscience Research Institute of North Carolina, one of the founders of the multiple channel single-unit recording technique in awake animals. JYC is the Associate Professor of the NRINC, who helped FL to establish the electrophysiological lab in Beijing.
Anterior cingulate cortex
Medial dorsal thalamus
primary somatosensory cortex
ventral posterior lateral thalamic nucleus
principal component analysis
peri-stimulation time histograms
primary sensorimotor cortex
wide dynamic range
Mass spike count
functional magnetic resonance imaging
This work was funded by a NNSF grant (30700223) and a CAS grant (KSCX2-EW-Q-18) to JYW, NNSF grants (30570577, 30770688, 30970959, and 61033011), CAS grants (KSCX2-YW-R-254 and KSCX2-EW-J-8), and a NIH grant (TW008038) to FL.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.