Activation of cortical somatostatin interneurons prevents the development of neuropathic pain

Abstract

Neuropathic pain involves long-lasting modifications of pain pathways that result in abnormal cortical activity. How cortical circuits are altered and contribute to the intense sensation associated with allodynia is unclear. Here we report a persistent elevation of layer V pyramidal neuron activity in the somatosensory cortex of a mouse model of neuropathic pain. This enhanced pyramidal neuron activity was caused in part by increases of synaptic activity and NMDA-receptor-dependent calcium spikes in apical tuft dendrites. Furthermore, local inhibitory interneuron networks shifted their activity in favor of pyramidal neuron hyperactivity: somatostatin-expressing and parvalbumin-expressing inhibitory neurons reduced their activity, whereas vasoactive intestinal polypeptide-expressing interneurons increased their activity. Pharmacogenetic activation of somatostatin-expressing cells reduced pyramidal neuron hyperactivity and reversed mechanical allodynia. These findings reveal cortical circuit changes that arise during the development of neuropathic pain and identify the activation of specific cortical interneurons as therapeutic targets for chronic pain treatment.

Figure 1. Elevated L5 pyramidal somatic Ca²⁺ activity in the S1 after peripheral nerve injury

(a) Schematic of spared nerve injury (SNI) and two-photon Ca²⁺ imaging timeline in the primary somatosensory cortex. Imaging was performed at different cortical depths (dashed boxes) from pial surface in awake, head-restrained mice expressing GCaMP6s in L5 pyramidal (PYR) neurons. (b) Measures of hind limb paw withdrawal threshold before and after SNI (n = 17 mice) and Sham operations (n = 12 mice) (2 d: P = 0.025; 7 d: P < 0.001; 14 d: P < 0.001; 31 d: P < 0.001, two-way ANOVA followed by Tukey’s test). (c) Fluorescence traces of representative L5 PYR soma expressing GCaMP6s in SNI/Sham mice at 1 month post-surgery. (d) Average total integrated Ca²⁺ activity over 2.5 min in L5 PYR 1 month after surgery (SNI: 116.2 ± 8.5 ΔF, n = 141 cells from 5 mice; sham: 41.6 ± 2.9 ΔF, n = 59 cells from 5 mice; t₁₉₈ = 5.581, P < 0.001, unpaired t test; SNI ipsilateral: 20.2 ± 1.0 ΔF, n = 106 cells, t₂₄₅ = 9.694, P < 0.001, unpaired t test). (e) L5 PYR soma Ca²⁺ activity at 1 month correlates with the hind limb paw withdrawal threshold in SNI mice (Pearson r = -0.33, P < 0.001), but not in Sham mice (Pearson r = 0.046, P = 0.609). (f) L5 PYR soma total integrated Ca²⁺ activity before (SNI: 40.3 ± 3.6 ΔF, n = 50 cells from 2 mice; Sham: 17.1 ± 1.2 ΔF, n = 45 cells) and after local application of TTX to sciatic nerve (SNI: 41.9 ± 3.7 ΔF, t₄₉ = 1.928, P = 0.06, paired t test; Sham: 11.6 ± 0.9 ΔF, t₄₄ = 12.08, P < 0.001, paired t test) at 2 weeks post-Sham/SNI. Data are presented as means ± s.e.m. *P < 0.05, ***P < 0.001. (c) Representative traces from experiments carried out on at least 5 animals per group.

Figure 2. Enhanced dendritic spine and branch activities in the S1 of neuropathic pain mice

(a) Representative two-photon images of active, apical tuft dendrites of L5 PYR neurons expressing GCaMP6s in SNI/Sham mice at 1 month post-surgery. Arrowheads point to spine heads measured. (b) Fluorescence traces of dendritic spines and adjacent shaft show increased Ca²⁺ transients in spines and shaft of SNI as compared to Sham mice. Light magenta bars indicate Ca²⁺ elevations in both spine heads and shaft, indicating Ca²⁺ spike generation. (c) Distribution of peak Ca²⁺ amplitude of dendritic spines over 1 minute (SNI: 282 ± 21 peak ΔF/F₀, n = 54 spines from 3 mice; Sham: 183 ± 13 peak ΔF/F₀, n = 47 spines from 3 mice; t₂₃₁ = 4.116, P < 0.001). (d) Distribution of average total integrated Ca²⁺ activity of dendritic spines over 1 minute (SNI 125 ± 8.2 ΔF, Sham 66.5 ± 4.1 ΔF, t₉₉ = 6.078, P < 0.001). (e) Average spine co-activity index in SNI/Sham mice (SNI 0.8 ± 0.02, n = 43 spines; Sham 0.63 ± 0.04, n = 40 spines; t₈₁ = 3.368, P < 0.01). (f) In both SNI and Sham mice, L5 cell spine co-activity correlates with the generation of apical tuft dendritic Ca²⁺ spike (SNI: Pearson r = 0.54, P < 0.001; Sham: Pearson r = 0.33, P < 0.05). (g) Fast-scanning of individual tuft Ca²⁺ events in SNI (rise time, 0.7 ± 0.2 s, decay time, 4.8 ± 1.5 s, n = 11 dendrites) and Sham mice (rise time, 0.6 ± 0.1 s, decay time, 4.5 ± 0.5 s, n = 9 dendrites). Average trace shown as magenta for SNI and green for Sham. (h) Average number of dendritic Ca²⁺ spikes per minute recording on dendrites (SNI, 4.1 ± 0.5 transients/min; Sham, 1.7 ± 0.2 transients/min; t₃₂ = 4.509, P < 0.001). (i) Distribution of peak Ca²⁺ amplitudes of dendritic Ca²⁺ spikes (SNI, 534 ± 38.5 ΔF/F₀, n = 147 spikes from 5 mice; Sham, 192 ± 19.4 ΔF/F₀, n = 117 spikes from 5 mice, t₂₆₂ = 7.393, P < 0.001). (j) Average total integrated Ca²⁺ activity detected with GCaMP6s over 2.5 min in active tuft apical dendrites (SNI, 80.2 ± 6.6 ΔF, n = 147 dendrites; Sham, 37.0 ± 4.0 ΔF, n = 117 dendrites, t₄₀₈ = 6.613, P < 0.001). Local application of either MK801 or TTX to L1 in SNI/Sham mice at 1 month post-surgery (SNI+MK801: 4.0 ± 0.6 ΔF, n = 30 dendrites from 2 mice, t₄₀₈ = 7.22, P < 0.001; Sham+MK801: 8.4 ± 1.2 ΔF, n = 38 dendrites from 2 mice, t₄₀₈ = 2.918, P = 0.011; SNI+TTX: 6.6 ± 0.7 ΔF, n = 37 dendrites from 4 mice, t₄₀₈ = 7.593, P < 0.0001; Sham+TTX: 5.6 ± 0.5 ΔF, n = 45 from 2 mice, t₄₀₈ = 3.402, P = 0.0022). Data are presented as means ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test in (c-e, h, i), two-way ANOVA followed by Bonferroni's test in (j). (a-b) Representative images and traces from experiments carried out on at least 3 animals per group.

Figure 3. Elevated dendritic Ca²⁺ spike generation in apical tuft branches and trunks of L5 pyramidal neurons in the S1 of neuropathic pain mice

(a) Two-dimensional projection of multiple sibling branches from an individual L5 PYR neuron in SNI mice. Five regions-of-interests (ROIs) (magenta box) corresponding to different dendritic branches were analyzed. All ROIs generated multiple Ca²⁺ transients during recording. (b) Percentages of local or global Ca²⁺ spikes observed on sibling branches from individual PYR neurons (SNI, 9.8 ± 1.2 transients/min, n = 24 cells from 3 mice; Sham, 4.8 ± 1.0 transients/min, n = 16 cells from 3 mice). Ca²⁺ events were analyzed over 2.5 min. (c) Fluorescence traces of apical dendritic trunks imaged at ∼300 μm below the pial in SNI/Sham mice. (d) Average total integrated Ca²⁺ activity over 2.5 min in apical trunk (SNI, 77.2 ± 4.8 ΔF, n = 102 trunks from 5 mice; Sham, 32.9 ± 2.3 ΔF, n = 103 trunks from 5 mice; t₂₅₅ = 9.34, P < 0.001). Local application of MK801 in SNI (16.7 ± 0.8 ΔF, n = 26 trunks from 2 mice, t₂₅₅ = 8.111, P < 0.001, two-way ANOVA followed by Bonferroni's test) and Sham mice (8.3 ± 0.9 ΔF, n = 28 trunks from 2 mice, t₂₅₅ = 3.4, P = 0.0016). (e) TTX locally applied to L1 (blue shaded circle) significantly reduced Ca²⁺ activity in apical trunk and L5 soma (noted by grey dashed lines) in SNI mice (before TTX: 1; trunk after TTX: 0.51 ± 0.02, n = 50 trunks, t₄₉ = 23.77, P < 0.001, paired t test; soma after TTX: 0.53 ± 0.03, n = 54 cells, t₅₃ = 17.19, P < 0.001; trunks and cells not related to same cells). Data are presented as means ± s.e.m. **P < 0.01, ***P < 0.001. (a,c) Representative traces from experiments carried out on at least 3 animals per group.

Figure 4. Alterations in inhibitory neuronal Ca²⁺ activity in the S1 of neuropathic pain mice

(a–d) Cartoon (left of traces) depicting cortical interneuron subtype imaged with GCaMP6s (highlighted by color). Representative fluorescence traces (right) and average total integrated Ca²⁺ activity over 2.5 min recordings of L2/3 SOM-positive somata and axon fibers expressing GCaMP6s in SNI/Sham mice at 1 month (soma: SNI 36.0 ± 4.7 ΔF, n = 117 cells from 4 mice, Sham 75.2 ± 7.7 ΔF, n = 68 cells from 3 mice; t₁₈₃ = 6.825, P < 0.001; axon: SNI 19.1 ± 1.1 ΔF, n = 80 fibers from 4 mice, Sham 42.5 ± 3.1 ΔF, n = 61 fibers from 3 mice; t₁₃₉ = 9.008, P < 0.001). (e–h) Representative fluorescence traces and quantification of average total integrated Ca²⁺ activity of L2/3 PV-positive (e–f; SNI, 40.0 ± 3.0 ΔF, n = 81 cells from 3 mice; Sham, 55.5 ± 5.1 ΔF, n = 66 cells from 3 mice; t₁₄₅ = 2.707, P < 0.01) and L2/3 VIP-positive (g–h; SNI, 90.0 ± 5.0 ΔF, n = 133 cells from 4 mice; Sham, 47.9 ± 3.6 ΔF, n = 144 cells from 4 mice; t₂₇₅ = 6.899, P < 0.001) neurons at 1 month. Data are presented as means ± s.e.m. **P < 0.01, ***P < 0.001, unpaired t test. (a, c, e, g) Representative traces from experiments carried out on at least 3 animals per group.

Figure 5. Acute activation of SOM neurons in the S1 reduces L5 pyramidal neuron activity in neuropathic pain mice

(a) Representative fluorescence traces of SOM neurons expressing GCaMP6s/hM₃Dq-mcherry before and after CNO injection in SNI mice. (b) Percent change in average Ca²⁺ activity of SOM somata over 2.5 min following CNO injection in SNI (382 ± 78%, n = 93 cells from 4 mice) and Sham mice (361 ± 114%, n = 61 cells from 5 mice). SOM Ca²⁺ activity significantly increased from baseline activity upon CNO injection (SNI: t₉₂ = 4.912, P < 0.001; Sham: t₆₀ = 3.168, P < 0.01, paired t test). No significant difference was found between SNI and Sham mice after CNO injection (t₁₅₂ = 0.163, P = 0.87, unpaired t test). (c) Representative fluorescence traces of apical dendrites of L5 PYR neurons expressing GCaMP6 before and after CNO injection to activate SOM/hM₃Dq-positive neurons in SNI mice. (d) Dendritic Ca²⁺ activity of L5 PYR neurons following CNO injection in SNI (-14.5 ± 5.5%, n = 29 dendrites from 3 mice; t₂₈ = 2.522, P = 0.0176, paired t test) and Sham mice (-11.5 ± 5.3%, n = 43 dendrites from 4 mice; t₄₂ = 2.149, P = 0.0374). (e) Representative fluorescence traces of L5 PYR neuron somata expressing GCaMP6s before and after CNO injection to activate SOM/hM₃Dq-positive neurons in SNI mice. (f) L5 PYR somatic Ca²⁺ activity following CNO injection in SNI (-17.9 ± 4.0%, n = 94 cells; t₉₃ = 4.449, P < 0.001, paired t test) and Sham mice (-13.7 ± 5.9 %, n = 87 cells; t₈₆ = 2.310, P < 0.05). (g) Percent change in average total integrated Ca²⁺ activity of SOM somata following CNO injection in naïve mice infected with hM₄Di. SOM Ca²⁺ activity significantly decreased from baseline activity upon CNO injection (-25.4 ± 10.6%, n = 20 cells from 2 mice, t₁₉ = 2.381, P = 0.03). (h, i) Inactivating SOM cells with hM₄Di exacerbated Ca²⁺ activity of L5 PYR dendrites (h, % change in L5 dendritic Ca²⁺ activity after CNO: Sham: 32.9 ± 14.1 %, n = 60 dendrites, t₅₉ = 2.35, P = 0.02, paired t test; SNI: 21.6 ± 6.6 %, n = 40 dendrite, t₃₉ = 3.195, P = 0.003, paired t test) and somata (i, % change in L5 PYR somata Ca²⁺ activity after CNO: Sham: 31.7 ± 5.0%, n = 60 soma, t₅₉ = 6.289, P < 0.001, paired t test; SNI: 13.9 ± 6.5%, n = 52 soma, t₅₁ = 2.126, P < 0.05, paired t test) in naïve mice. (j) Cartoon depicting design for determining mechanical allodynia before and after CNO-induced SOM cell activation or inactivation. Thresholds measured before and 20 min after CNO injection in SNI/Sham mice. (k) Withdrawal threshold after SOM activation by CNO in SNI (before CNO: 1.7 ± 0.1 g, after 20-min CNO: 4.0 ± 0.2 g, n = 14 mice; t₁₃ = 12.97, P < 0.001, paired t test) and Sham mice (before CNO: 4.8 ± 0.4 g, after 20-min CNO: 5.2 ± 0.2 g, n = 14 mice; t₁₃ = 1.685, P = 0.12, paired t test). (l) Withdrawal threshold before and after CNO injection in naïve mice infected with hM₄Di. Inactivating SOM cells had no significant effects on the animals' withdraw threshold (contralateral paw: P = 0.12, paired t test; ipsilateral paw: P = 0.13, paired t test). (m) Percent change in average total integrated Ca²⁺ activity of PV somata following CNO injection in SNI (185 ± 29.7%, n = 81 cells from 3 mice) and Sham mice (131 ± 24.5%, n = 48 cells from 3 mice). PV Ca²⁺ activity significantly increased from baseline activity upon CNO injection (SNI: t₈₀ = 6.243, P < 0.001; Sham: t₄₇ = 5.339, P < 0.001, paired t test). No significant difference was found between SNI and Sham mice after CNO injection (t₁₂₇ = 1.262, P = 0.21, unpaired t test). (n) Withdrawal threshold after PV activation by CNO in Sham (before CNO: 5.4 ± 0.3 g, after 20-min CNO: 4.5 ± 0.4 g, n = 8 mice; t₇ = 2.305, P = 0.06, paired t test) and SNI mice (before CNO: 1.9 ± 0.3 g, after 20-min CNO: 2.1 ± 0.1 g, n = 8 mice; t₇ = 0.7693, P = 0.47, paired t test). Data are presented as means ± s.e.m. *P < 0.05. **P < 0.01, ***P < 0.001. (a, c, e) Representative traces from experiments carried out on at least 3 animals per group.

Figure 6. Daily activation of SOM neurons following perpherial nerve injury prevents the development of neuropathic pain

(a) Top panel, experimental design for daily activation of SOM neurons following SNI or sham surgeries (CNO, orange syringe; vF, von Frey). For control, SNI and sham mice received saline injections. Bottom panel, cartoon of two-photon imaging planes corresponding to SOM somata (b), pyramidal somata (c,d) and pyramidal tuft dendrites (e,f). (b) Percent change in average total integrated Ca²⁺ activity of SOM somata over 2.5 min following single CNO injection in SNI (n = 23 cells from 2 mice) and Sham mice (n = 26 cells from 2 mice). CNO injection significantly increased from baseline activity of SOM cells for at least 12 h (P < 0.001) but not 24 h (Sham, P = 0.19; SNI, P = 0.29). (c) Fluorescence traces of L5 PYR neuron somata expressing GCaMP6s after 1-week of daily CNO activation of SOM cells in SNI/Sham mice. (d) Average total integrated Ca²⁺ activity over 2.5 min recording of L5 PYR somata following 1-week daily SOM activation in SNI (14.4 ± 0.7 ΔF, n = 170 soma from 5 mice) and Sham mice (14.4 ± 1.2 ΔF , n = 54 soma from 5 mice). No significant difference between SNI and Sham mice treated with CNO (t₃₇₇ = 7.171, P = 0.99). (e) Fluorescence traces of L5 apical tuft dendrites expressing GCaMP6s after 1-week of daily SOM activation in SNI and Sham mice. (f) Average total integrated Ca²⁺ activity over 2.5 min recording of L5 dendrites following 1-week daily SOM activation in SNI (14.1 ± 0.7 ΔF, n = 159 dendrites from 5 mice) and Sham mice (12.9 ± 0.8 ΔF, n = 37 dendrites from 5 mice). No significant difference between SNI and Sham mice treated with CNO (t₂₇₉ = 0.7357, P = 0.9251). (g) Withdrawal threshold under various conditions. Daily SOM activation significantly increased withdrawal threshold in SNI mice as compared to SNI mice with IP saline (P < 0.05 at day 5 and P < 0.001 at day 7, 14, 21, and 28). Data are presented as means ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001, paired t test in (b), two-way ANOVA followed by Bonferroni's test in (d, f), and Tukey’s test in (g). (c, e) Representative traces from experiments carried out on at least 3 animals per group.