A distinct neuronal ensemble of prelimbic cortex mediates spontaneous pain in rats with peripheral inflammation

Abstract

The absence of a comprehensive understanding of the neural basis of spontaneous pain limits the development of therapeutic strategies targeting this primary complaint of patients with chronic pain. Here we report a distinct neuronal ensemble within the prelimbic cortex which processes signals related to spontaneous pain in rats with chronic inflammatory pain. This neuronal ensemble specifically encodes spontaneous pain-related behaviors, independently of other locomotive and evoked behaviors. Activation of this neuronal ensemble elicits marked spontaneous pain-like behaviors and enhances nociceptive responses, whereas prolonged silencing of its activities alleviates spontaneous pain and promotes overall recovery from inflammatory pain. Notably, afferents from the primary somatosensory cortex and infralimbic cortex bidirectionally modulate the activities of the spontaneous pain-responsive prelimbic cortex neuronal ensemble and pain behaviors. These findings reveal the cortical basis of spontaneous pain at the neuronal level, highlighting a distinct neuronal ensemble within the prelimbic cortex and its associated pain-regulatory brain networks.

Fig. 1. Pivotal role of the prelimbic cortex (PL) in spontaneous pain behaviors.

a Spontaneous behaviors within 30-min recording sessions 0–14 days after CFA injection into left hindpaw plantar. n = 10 rats for 0, 1, 3, 14-day CFA groups, n = 8 rats for 7-day CFA group, n = 7 rats for 0, 1, 7, 14-day Sham group, n = 6 rats for 3-day Sham group. Two-way ANOVA with Bonferroni post hoc test. b Left: Schematic illustrating the relative positions of recording electrodes in pain-associated brain areas in rats with CFA injection. PL prelimbic cortex, IL infralimbic cortex, AcbC nucleus accumbens core, AcbSh nucleus accumbens shell, S1 primary somatosensory cortex, MD medial thalamus, dCA1 dorsal hippocampal CA1, BLA basolateral amygdala, CeA central amygdala, vCA1 ventral hippocampal CA1; refs, reference and ground wires in the cerebellum. Right: An example of simultaneous recordings of LFPs (band-pass filtered 0.5–100 Hz, upper) and behavioral signals (bottom) during a recording session. The diagram was created with BioRender.com. c Granger causality comparisons between the onsets and intermissions of spontaneous paw lifting behaviors between pairs of two recorded regions. The T-values (left) and P-values (right) were measured by the point-to-point two-sided paired t-tests across either two recorded regions. d Increased granger causality values of information flows from PL to other regions during periods with and without spontaneous paw lifting. Paired t-test. n = 22 recording sessions from 10 rats. **p < 0.01, ***p < 0.001, Two-way ANOVA with Bonferroni post hoc test. e. Power spectrum density (PSD) comparisons between the onsets and intermissions of spontaneous paw lifting among ten brain regions. The T-values (left) and P-values (right) measured by the two-sided paired t-tests across the frequency. f Increased gamma band power (32–75 Hz) of the PL during spontaneous paw lifting compared to the intermissions. Red and blue lines showed the mean value of PSD during the presence and absence of spontaneous paw lifting, respectively. Shaded areas define ± SEM. Two-sided paired t-tests, black bar showed significant FDR corrected p values. Data for A, D were presented as mean values ± SEM. Source data for A and D were provided as a Source Data file.

Fig. 2. Prelimbic neurons responding to spontaneous pain behaviors.

a Enhanced c-Fos expression of PL neurons in rats with CFA injection (day 3). Left: Representations of immunostaining of the neurons of PL (upper) and IL (bottom) in CFA (right) and Sham groups (left). Right: Statistical results. Three slices from each rat, n = 4 rats for IL-CFA group, n = 5 rats for PL and IL-Sham groups. Scale bars: 250 µm. *p < 0.05, **p < 0.01, two-sided unpaired t-test. b Electrophysiological recording sites and an example of representative high-pass filtered traces. c Bi-modal distribution of spike durations was used to identify cells as putative interneurons (gray) and pyramidal neurons (yellow). d Representative neurons showing excitatory (upper), inhibitory (middle) and neutral (bottom) responses to CFA-induced spontaneous paw lifting. Left: The spike trains were binned at a 10-s window, with episodes of spontaneous paw lifting shaded by red. The averaged firing rate was marked by blue lines. Right: The averaged firing rate during the spontaneous paw lifting were marked by red vertical lines. e The proportion of excitatory responsive neurons in the PL but not in the IL paralleled the level of spontaneous paw lifting. *p < 0.05, Chi-square test. f Schematic illustrating electrophysiological recordings of PL using silicon probes during the ibuprofen administration. g Administration of ibuprofen alleviated spontaneous paw lifting in rats with inflammatory pain. n = 6 rats, ***p < 0.001, two-sided paired t-test. h Normalized firing rate changes of putative pyramidal neurons in PL after ibuprofen administration. n = 51 units from 3 rats. The dashed lines indicated the edges of firing rates with significant increment (red) or reduction (blue). i Proportions of changed firing rates (outer pie charts) in PL neurons responding to spontaneous paw lifting behaviors (inner ring chart) after ibuprofen administration. *p < 0.05, Chi-square test for trend. j Decreased spontaneous firing rates of excitatory responsive to paw lifting neurons after ibuprofen administration. n = 11 units, *p < 0.05, two-sided Wilcoxon signed-rank test. Data for A, E, G, I and J were provided as a Source Data file. The diagrams B and F were created with BioRender.com.

Fig. 3. A prelimbic neuronal ensemble specifically encodes spontaneous pain behaviors.

a Experimental timeline and daily recording protocol for in vivo electrophysiology before and after CFA injection. b Stimuli-evoked responses of PL pyramidal neurons in different evoked modalities. Heatmap rows represented the Z score-transformed average PSTH for individual neurons, and columns represented time bins relative to stimulus onset. The right column of the heatmap indicated the subpopulations of neurons showing excitatory (red), inhibitory (blue) and neutral responses to spontaneous paw lifting behaviors, respectively. c Pattern comparisons of PL neuronal populations which showed significant responses between spontaneous paw lifting and locomotion (upper)/self-grooming (bottom) behaviors. Left: Venn diagrams of amounts of significantly responding neurons to the behaviors, and the overlay showing the synclastic responses. (i.e., positive or negative responses in both given modalities). Right: Distributions of hypergeometric probabilities for each comparisons showing whether two neuronal response patterns were similar or significantly different. A significant difference was observed between spontaneous paw lifting and other behaviors. d Pattern comparisons of the neuronal populations showing significant responses between spontaneous paw lifting and different evoked stimuli. Left: Matrix of amounts of responding neurons to the modality combinations among evoked stimuli and spontaneous paw lifting. Right: Hypergeometric probabilities indicated the significance of the neurons responding to the modality combinations. Stars and triangles represented significantly similar and different responsive patterns to the modality combination, respectively. Significantly different patterns were observed between spontaneous paw lifting and multiple evoked modalities (pin, 15-g hair, 2-g hair, laser and sound), but not between spontaneous paw lifting and any other evoked modality. e The population activities of all PL neurons significantly distinguished a majority of evoked behaviors except brush. Left: The confusion matrix of the prediction accuracies. Right: p values of the prediction performance. f The performance in predicting evoked behaviors using spontaneous pain-irrelevant PL neurons.

Fig. 4. Labeling the spontaneous pain-related neuronal ensemble in the PL via RAM system.

a The RAM labeling system. Left: Virus strategy. Right: Timeline of virus injection, Dox switching, activated neuronal labeling and electrophysiological recording. The diagram was created with BioRender.com. b, c Characterization of labeled PL neurons in spontaneous pain induced by CFA using markers of glutamatergic (EAAC1⁺) and GABAergic (GAD⁺) neurons. Typical examples (b) and a quantitative summary (c) are shown. Scale bar, 250 µm. Three slices from each rat, n = 3 rats. d The raster plot (upper) and PSTH (bottom) of a representative opto-tagged spike unit activated by 1-Hz light pulse. e The proportion of the neurons responding to spontaneous paw lifting (inner pie charts) and 1-Hz light pulse (outer rings). n = 91 units from 3 rats. f Averaged PSTH of the tagged (green, activated), relevant (blue, inhibited) and irrelevant (orange, no significant change) neuronal subpopulations to a single light pulse. Shaded areas were defined by SEM. n = 133 units from 7 rats. g PSTHs of recorded neurons (left) in the 1-Hz protocol. Neurons were classified by a descending order of firing rate during the 1-s light-on period. Pie graphs (right) showing the proportion of neurons. h PSTHs of recorded neurons in the 3-min, 20-Hz protocol. Neurons were ordered in line with (g). i Average firing rate of the tagged and relevant neurons after optogenetic activation. n = 21 tagged, 56 irrelevant, 46 relevant units, respectively. *p < 0.05, ***p < 0.001, Friedman test. j Evoked nociceptive responses of recorded PL neurons to noxious laser on CFA-injected and uninjected hindpaws with and without optogenetic activation of spontaneous pain ensemble. Bin, 100 ms. k Averaged Z score changes. n = 21 tagged, 56 irrelevant, 46 relevant units respectively. **p < 0.01, two-way repeated measures ANOVA. Evoked spectrogram (l) and gamma power (m) of PL by noxious laser on CFA-injected and uninjected hindpaws with and without optogenetic activation. Bin, 50 ms. n Nociception-evoked gamma power after activating the spontaneous pain neuronal ensemble. n = 7 rats. *p < 0.05, two-way repeated measures ANOVA. Data for I, K and N were presented as mean ± SEM and provided as a Source Data file.

Fig. 5. Optogenetic activation of the spontaneous pain-tagged ensemble of PL aggravates pain behaviors in rats with inflammation.

a Diagram showing the timeline of behavioral tests. b Representative images of ChR2-EGFP expression and fiber site in the PL. PL spontaneous pain labeled neurons were re-activated (c-Fos⁺) following light stimulation compared to sham group and control virus group. Scale bar, 200 μm. 3 slices from each rat, n = 3 rats. ###p < 0.001 for RAM-CFA vs RAM-Sham, ***p < 0.001 for RAM-CFA vs vector-CFA, two-sided unpaired t-test. c Anatomical maps displaying the area of ChR2-EGFP expression across the anterior-posterior PL in CFA rats. n = 6 rats. d Optogenetic activation of PL spontaneous pain neuronal ensemble induced remarkable spontaneous paw lifting behaviors on day 3 after CFA injection. Bin = 10 s, n = 9 rats in RAM-Sham group, n = 10 rats in RAM-CFA and vector-CFA groups. *p < 0.05, two-sided paired t-test. e Systemic administration of ibuprofen (upper) but not peripheral administration of lidocaine (bottom) abolished light-induced aggravating spontaneous paw lifting behaviors in rats with inflammation. Bin = 10 s, n = 6 rats per group, *p < 0.05, n.s.: no significant difference, two-sided paired t-test. f Experimental design for real-time place avoidance (RTPA). g Activation of PL spontaneous pain neuronal ensembles resulted in increased place avoidance on day 5 after CFA injection. The n value was the same as that of (d). Left, time spent in paired and unpaired zone, **p < 0.01, two-sided paired t-test. Middle, RTPA scores, **p < 0.01, two-sided one sample t-test. Right, representative locomotor tracks. h Activation of PL spontaneous pain ensemble increased defecation in the light-paired zone compared with the unpaired zone. n = 10 rats, *p < 0.05, two-sided Mann–Whitney test. Optogenetic activation of the PL spontaneous pain-tagged neuronal ensemble exacerbated thermal hyperalgesia (i) and mechanical allodynia (j) in inflammatory pain. Dash line, baseline pain threshold. The n value was the same as that of (d), **p < 0.01, ***p < 0.001, two-sided paired t-test. Data for C, G and H were presented as mean ± SEM. Data for B-E, G-J were provided as a Source Data file.

Fig. 6. Chemogenetic silencing of the spontaneous pain-tagged neuronal ensemble of PL promotes overall recovery from chronic inflammatory pain.

a Diagram showing the timeline of behavioral tests. b Representative images of hM4Di-EGFP expression in the PL. PL spontaneous pain neurons were inhibited following administration of CNO. Scale bar, 200 μm. Three slices from each rat, n = 3 rats. ## p < 0.01 for RAM-CFA vs. RAM-Sham, ***p < 0.01 and **p < 0.01 for RAM-CFA vs. vector-CFA, two-sided unpaired t-test. c Temporary inhibition of PL spontaneous pain neurons alleviated spontaneous paw lifting behaviors on CFA 3 day. n = 9 rats. *p < 0.05, two-sided paired t-test. d Continuous CNO delivery (in CFA 3–7 d) relieved spontaneous paw lifting behaviors on CFA 7 day. n = 9 rats. ###p < 0.001 for RAM-CFA vs. RAM-Sham, *p < 0.05 for RAM-CFA vs. vector-CFA, two-sided unpaired t-test. e Experimental design for conditioned place preference (CPP). f Inhibition of PL spontaneous pain neurons induced preference to the paired chamber in CFA. n = 7 rats in RAM-CFA group, n = 8 rats in RAM-Sham and vector-CFA groups. Left, time spent in paired and unpaired chamber, *p < 0.05, two-sided paired t-test. Middle, CPP scores, *p < 0.05, two-sided one sample t-test. Right, representative locomotor tracks. Temporary inhibition of PL spontaneous pain neurons alleviated thermal hyperalgesia (g) and mechanical allodynia (h) on CFA 3 d. Dash line, baseline pain threshold. n = 9 rats. **p < 0.01, two-sided paired t-test. Continuous CNO delivery (in CFA 3–7 d) increased thermal (i) and mechanical pain thresholds (j) on CFA 7 d, and accelerated the recovery from thermal hyperalgesia (k) and mechanical allodynia (l) in inflammatory pain. n = 9 rats. ##p < 0.01 for RAM-CFA vs. RAM-Sham, *p < 0.05, ***p < 0.001 for RAM-CFA vs. vector-CFA, two-sided unpaired t-test for (i, j). ##p < 0.01, ###p < 0.001 for RAM-CFA-injected vs. RAM-Sham-CFA-injected; **p < 0.01, ***p < 0.001 for RAM-CFA-injected vs. vector-CFA-injected, two-way repeated measures ANOVA with Bonferroni’s post hoc test for (k, l). Data for D, F, I, K, J and L were presented as mean ± SEM. Data for C, D F-L were provided as a Source Data file.

Fig. 7. S1→PL and IL→PL projections differentially modulate PL spontaneous pain-responsive neuronal activities and pain behaviors.

a Anterograde viral labeling strategy for S1→PL pathway (left) and representative image of PL neurons receiving projection from the S1 (right). b The viral injection strategy and optrode site for electrophysiology recording in vivo. Activation of S1→PL projection terminals increased the firing rates of spontaneous pain-excitatory responsive neurons (c, n = 28 units), but not inhibitory responsive neurons (d, n = 12 units) in the PL. *p < 0.05, Friedman test. e Activation of IL→PL projections exacerbated spontaneous paw lifting behaviors. Left: Distribution of time spent on paw lifting behaviors. Blue shadow represented the light on condition. Right: Quantification of spontaneous paw lifting behaviors time. Bin = 10 s, n = 5 rats, *p < 0.05, two-sided paired t-test. f Activation of S1→PL projections aggravated thermal hyperalgesia (left) and mechanical allodynia (right). n = 5 rats, *p < 0.05, ***p < 0.001, one-way repeated measures ANOVA. g Anterograde viral labeling strategy for IL→PL pathway (left) and representative image of PL neurons receiving projection from the IL (right). h The viral injection strategy and optrode site for electrophysiology recording in vivo. Activation of IL→PL projections decreased the firing rates of spontaneous pain-excitatory responsive neurons (i, n = 14 units), but not inhibitory responsive neurons (j, n = 10 units) in the PL. *p < 0.05, Friedman test. k Activation of IL→PL projection terminals relieved spontaneous paw lifting behaviors. Left: Distribution of time spent on paw lifting behaviors. Blue shadow represented the light on condition. Right: Quantification of spontaneous paw lifting behaviors time. Bin = 10 s, n = 5 rats, *p < 0.05, two-sided paired t-test. l Activation of IL→PL projections alleviated thermal hyperalgesia (left) and mechanical allodynia (right). n = 5 rats, *p < 0.05, ***p < 0.001, one-way repeated measures ANOVA. For the boxplots of C, D, I and J, the top and bottom of the boxes, the center lines and the whiskers represented the 75%/25%, the median and min/max values of the data, respectively. Data for C–F and I–L were provided as a Source Data file. The diagrams A, B, G and H were revised from Brain maps 4.0.

Fig. 8. Summary of a distinct neuronal ensemble regarding the spontaneous pain induced by peripheral inflammation in the prelimbic cortex.

Main findings of this study elucidated a neuronal ensemble that encodes spontaneous pain behaviors in the PL, distinguished from other behaviors. The activity of these neurons inhibited other neurons, thus amplifying the nociception. The S1 and the IL modulated spontaneous pain neuronal activity and pain behaviors bidirectionally. The picture was revised from Brain maps 4.0, with the original drawing.