Vascular motion in the dorsal root ganglion sensed by Piezo2 in sensory neurons triggers episodic pain

Abstract

Spontaneous pain, characterized by episodic shooting or stabbing sensations, is a major complaint among neuropathic pain patients, yet its mechanisms remain poorly understood. Recent research indicates a connection between this pain condition and "clustered firing," wherein adjacent sensory neurons fire simultaneously. This study presents evidence that the triggers of spontaneous pain and clustered firing are the dynamic movements of small blood vessels within the nerve-injured sensory ganglion, along with increased blood vessel density/angiogenesis and increased number of pericytes around blood vessels. Pharmacologically or mechanically evoked myogenic vascular responses increase both spontaneous pain and clustered firing in a mouse model of neuropathic pain. The mechanoreceptor Piezo2 in sensory neurons plays a critical role in detecting blood vessel movements. An anti-VEGF monoclonal antibody that inhibits angiogenesis effectively blocks spontaneous pain and clustered firing. These findings suggest targeting Piezo2, angiogenesis, or abnormal vascular dynamics as potential therapeutic strategies for neuropathic spontaneous pain.

Keywords: Piezo2; angiogenesis; blood vessel; calcium imaging; dorsal root ganglion; neuropathic pain; pericytes.

Figure 1:. Vascularization of the DRG and vascular responses preceding clustered firing from DRG neurons of SNI mice

(A) Projection of 3D volume imaging of a SNI L4 DRG labeled with antibody to pan endothelial cell marker CD31 (green) and Alexa 633 dye (far red; label for elastin in arterial structures). (B) Three large branches of blood vessels on the surface of the same DRG (top view). (C, D) Immunostaining of the blood vessels identified by CD31 in DRG sections from normal (C) and SNI (D) mice. (E) Summary data of C and D showing increased number of neurons in normal and SNI DRGs, with cell bodies surrounded more than half by CD31+ blood vessels. **, p<0.001, t-test, n=6 per group. (F) qPCR analysis indicated that CD31 mRNA expression increased in SNI mouse DRGs on POD 2 and POD 28 compared to normal, uninjured DRGs. *p<0.05, **p<0.01, ANOVA, n=4. (G) Labeling with GAP43 (red) revealed neuronal sprouting in the DRG after SNI, closely associated with blood vessels (CD31, green). (H) Diagram showing the mating strategy to express Ca indicator in sensory neurons and the in vivo recording setup used in (I). (I) Examples of single frames from video recordings showing vasomotion preceding clustered firing in SNI mice in vivo, with arrows indicating blood vessel movements. Blood vessel visibility was enhanced by i.v. dextran fluorescein (MW 70,000). See Movie S3 for dynamics.

Figure 2:. Vasoconstrictors enhance neuropathic spontaneous pain and abnormal clustered firing of the sensory neurons in SNI mice.

(A) Systemic injection of phenylephrine (PE) increased spontaneous pain behaviors. Baseline pain was measured in sham animals (black) or after SNI was established (red, blue). PE (0.6 mg/kg, i.p) or vehicle (saline) was injected 2–3 hours later, and spontaneous pain was re-evaluated. Individual values are plotted as before-after points; bars indicate means. Black: in uninjured mice, baseline pain scores were low and unaffected by PE (p =0.52, paired t-test, n = 8. Red: after SNI, baseline pain increased and i.p. PE further increased spontaneous pain. ***, p<0.001, paired t-test, n = 21. Blue: Vehicle injection failed to show significant effects on baseline pain in SNI mice. p = 0.84., paired t-test, n =11. Conclusions were the same when analyzed for males only or females only. (B) Timelines of drug administration, and image recording setup. (C) Representative images (extracted frames from recordings) of the clustered firing before and after local PE application. (D) Time courses of calcium transients in individual neurons from the clusters before and after PE application. (E, F) Effect of PE on clustered firing. The number of clusters (E) and total number of neurons in clusters (F) was measured during 45 minutes of in vivo imaging (baseline). Then PE (0.6 mg/kg, i.p. or 0.3 mg/ml, 30 μl intra-DRG, “local”) or vehicle (buffered saline) was injected, and clusters observed for the next 45 minutes. In normal mice, no clusters were observed before or after i.p. PE (n = 5; black) or intra-DRG PE (n= 5; grey). In mice with established SNI (21–35 days), the number of clusters and total number of neurons per cluster were significantly increased by i.p. PE (n = 7; red) or local PE (n = 7; orange). In mice with established SNI, clustered firing was not affected by vehicle (saline) injection using the same recording protocols (n = 8 i.p., blue, and n = 7, local injection, light blue). Increase in clustered firing could also be observed in mice with microsympathectomy (“mSYMPX”) performed just prior to recording (n = 9; green). Clustered firing could still be induced in SNI mice in which the neuroma was disconnected from the DRG just prior to recording (n= 12; purple). **, p<0.01; *, p<0.05; n.s., not significant, paired t-test). (G) Example of single frames extracted from video recordings showing vasomotion preceding clustered firing in SNI mice in vivo. Clusters occurred at branches of blood vessels indicated by “*” in SNI DRG after PE (i.p.) and showing dynamic changes/displacement of small blood vessel indicated by the white arrow. Time between vascular activity and appearance of clustered firing is 4–5 sec.

Figure 3.

Angiotensin II (A-II) also increases spontaneous pain and clustered firing in SNI mice-an effect reduced by mechanical receptor blockers. (A) Baseline spontaneous pain was measured in normal animals or after SNI was established. Angiotensin II (A-II; 0.2 mg/kg, i.p.) was injected 3–4 hours later and spontaneous pain measured again. n.s., not significant; ***, p<0.001, paired t test. Normal, n = 8; SNI, n = 12. Conclusions were the same when analyzed for males only or females only. (B) Timelines of drug administration and image recording. (C) Effect of A-II on clustered firing. The number of clusters (top) and total number of neurons in clusters (bottom) were measured during 45 minutes of in vivo imaging (baseline). Then angiotensin II (0.6 mg/kg, i.p) was injected and clusters observed for the next 45 minutes. *, p<0.05, **, p<0.01, paired t test. N = 8 mice for i.p. In a second experiment, D-GsMTx4 and gsmtx4, blockers of mechanoreceptors, were injected locally into the DRG prior to the A-II injection, which blocked A-II induced clustered firing (n.s., not significant; n = 6). (D) Time to cluster start following i.p. injections of PE and A-II.

Figure 4.. Role of mechanoreceptors in clustered firing and neuropathic spontaneous pain.

(A) Timelines of local drug administration. (B) Effect of local mechanoreceptor blockers on clustered firing. Purple: After SNI was established, the L4 DRG was imaged for 45 minutes (“baseline”). D-GsMTx4 and gsmtx4 were injected into the DRG (2 μL, both at 50 μM) and the DRG imaged for another 45 minutes. Then blockers were re-injected, and PE given i.p. *, p<0.05; n.s., not significant, paired t-test (n=9). Blue: similar experiment format, but both PE and the blockers were injected together into the DRG (n=6, note, control for this experiment is intra DRG saline, shown in Figure 2E, F). (C) In situ hybridization confirming that some Na_V1.8-positive neurons as putative nociceptors also express Piezo2 in L4 DRG from SNI mice on POD 28. Scale bar=50 μm. (D) Immunostaining demonstrating increased expression of Piezo2 in the SNI DRG on POD 28 compared to contralateral normal DRG (***, p<0.001, t-test, n = 5 per group). (E) Piezo2 knockdown with siRNA in the DRG blocks PE-induced increase in spontaneous pain. After SNI was established, mice were videotaped for 3 continuous days. On the third day, 1–2 hours after baseline taping (“base”), the mice received PE (i.p) and spontaneous pain was measured again (“+PE”). siRNA (against Piezo2, black, or control non-targeting, red) was injected two days later (i.t. 5 μl x2), and on the third day after siRNA injection, spontaneous pain before and after PE i.p. injection was measured again. The effect of PE on spontaneous pain was lost after Piezo2 knockdown. n.s., not significant; *, p<0.05,**, p<0.01, ***, p<0.001, significant difference between the indicated groups, two-way ANOVAs with Šídák’s posttest (n=8 per group). 2-way ANOVA of the difference scores (PE – baseline) showed a significant effect of siRNA type (p = 0.02) and of time (pre vs. post siRNA injection, p = 0.02). After Piezo2 siRNA injection the PE-induced change in pain behavior was not significantly different from zero, while all other difference scores were significantly different from zero (p<0.01 to 0.001.) (F) Effect of Piezo2 knockdown on baseline and PE-induced clustered firing. siRNA directed against Piezo2 or nontargeting control was injected after SNI was established, 2 – 3 days prior to recording. After recording baseline clustered firing, PE was injected i.p. N = 7 for control siRNA group, n=8 for siRNA group. $, p<0.05, $$, p<0.01, significant overall effect of siRNA, **, p<0.01, n.s., not significant effect of PE, two-way ANOVAs with Šídák’s posttest. (G) Effect of Cre-mediated Piezo2 knockdown on spontaneous pain behaviors. AAV-Cre or AAV-control virus was injected into the ipsilateral paw at age postnatal day 16. After the animals reached adulthood, SNI was performed. Spontaneous pain was measured 4–5 weeks later before (“SNI base”) and 1–2 hours after i.p. injection of PE. Piezo2 knockdown reduced spontaneous pain (#, p<0.05, ###, p<0.001 compared to AAV-Cre group) and blocked the PE-induced increase in spontaneous pain behaviors ($$$, p<0.001, significant overall effect of the AAV factor in 2-way RM ANOVA; ### or ***, p<0.001; # p<0.05, n.s., not significant, Šídák’s posttest). Cre group: n = 13; Control AAV group: n=14. (H) Effect of CRE-mediated Piezo2 knockdown on baseline and PE-induced clustered firing. Virus treatment was as in (G). Cre group: n=8; RFP control group (n=7). $$, p<0.01, significant overall effect of Cre factor, **, p<0.01, n.s., not significant effect of PE, 2-way ANOVAs with Šídák’s posttest.

Figure 5.. Clustered firing is evoked by mechanically induced myogenic vascular responses in the DRG from SNI mice in vivo and in vitro.

(A-E) In vivo experiments. The L4 DRG was recorded in mice 3–5 weeks after SNI. After a 45-minute of baseline recording, a blunt glass micropipette was used to apply mechanical stimuli (~1 minute each) on visible blood vessels (A). (B) Sample images showing clustered firing triggered by gentle mechanical poking of the blood vessels in SNI DRG but not in normal, uninjured DRG. (C) Number of observed clusters normalized to the recording time (45 minutes for baseline, or total duration of mechanical stimuli). **, p<0.01, Wilcoxon matched pairs signed rank test. (D) Examples of calcium transients in response to poking a small blood vessel. (E) Summary data showing the percentage of blood vessel (b.v.) pokes that evoke clustered firing in the indicated conditions. & = significant difference when all experiments compared to SNI b.v. poke via ANOVA. * is for indicated comparisons (mixed effect model for the 3 groups with repeated measures in same animal), # unpaired t test for the 2 indicated comparisons. The poking-induced percent responses data in various experimental groups include data from animals used for the clustered firing data in Figures 2–4 (n=10). (F-H) Experiments with an ex vivo DRG preparation (n=5). (F) Diagram of the recording setup. Whole DRGs with nerve and root attached were removed from mice after SNI and mounted in a recording chamber perfused with ACSF at 35–36°. (G) Example still frames extracted from Movie S11 showing clustered firing evoked by gentle poking of a nearby blood vessel. Dashed lines showing the path of small blood vessels visualized under the microscope. The site of poking is indicated by *. 3a, the inset of frame #3 showing red blood cells moving in the capillary over the activated neuron. (H) Example still frames from Movie S12 showing the poking pipette, blood vessels visualized via i.v. injection of dextran, and activated neurons evoked by gentle poking.

Figure 6.

Peripheral nerve injury increased the density of pericytes in the DRG. (A, B) Representative image of pericytes identified by PDGFRβ (red) associated with tomato lectin labeled blood vessels (green) in normal and SNI mice on POD28. The asterisks indicate cell bodies (yellow) of the pericytes, where PDGFRβ signals overlap with tomato lectin-positive blood vessels. (C) Example of sympathetic (purple) innervations of the blood vessels (green) with pericytes surrounded (red) in SNI DRGs on POD28. #, TH-positive VGLUT3 sensory neurons. (D) Summary data showing the number of pericytes was higher in the SNI DRG compared to normal uninjured DRG (****, p <0.0001, t-test, n=5 per group). (E) qPCR measurements of NG2 expression showed it was increased in the SNI DRGs on POD2 and POD28. **, p<0.01; ***, p<0.001; n=4, one-way ANOVA. (F) qPCR measurements of Pdgfb mRNA expression showed it was increased in the SNI DRGs on POD2 and POD28. *, p<0.05; ***, p<0.01, one-way ANOVA; n=4. (G) qPCR measurements of Pdgfrb mRNA expression showed no change in the SNI DRGs on POD2 or POD28; n=4.

Figure 7.

Pericyte activation alters the flow of red blood cells in vitro and triggers mechanically mediated clustered firing in vivo in SNI mice (A) Imaging setup used to measure red blood cells (RBC) movement in the ex vivo preparation. DRGs from normal or SNI pirt-GCaMP6s mice were dissected out and placed in a recording chamber continuously perfused with ACSF. Arrow indicates blood vessels containing RBCs. (B) Pericyte activator U46619 at 30 μM increased RBC movement in SNI but not in normal DRGs. *, p<0.05, n=9; paired t-test. (C-E) Pericyte activation triggers clustered firing in SNI mice on POD 28 in vivo. Baseline clustered firing was recorded for 45 min followed by topical application of U46619 (10 μM and 30 μM). Image recording continued for an additional 45 min. *p<0.05, **p<0.01, one-way ANOVA. D: Normal control DRG, n=7; E: SNI DRG, n=7. (F) Pericyte activator-induced clustered firing was blocked by co-application with mechanical channel blockers, D-GsMTx4 and gsmtx4. p>0.05, one-way ANOVA, n=6. (G) Topical application of saline did not evoke any clustered firing in 7 mice.

Figure 8.. Anti-VEGF monoclonal antibody bevacizumab decreases spontaneous pain and blocks spontaneous and PE-evoked clustered firing.

(A) Nerve injury increased VEGF protein expression along blood vessels measured by the length of VEGF-positive blood vessels normalized by cellular area. ****p<0.0001, unpaired t-test, n=6 per group. (B) Experimental strategy and timeline for C-E. (C) Bevacizumab administered during the first 3 weeks (10 mg/kg, i.p., twice a week) of nerve injury decreased CD31-postive blood vessel density. (D) Bevacizumab administered during the first 3 weeks of nerve injury decreased baseline (*, p<0.05, **, p<0.01; ##, p<0.01; $,p<0.05; two-way RM ANOVA with Šídák’s posttest; n=8 per group) and PE-evoked (####,p<0.0001, two-way RM ANOVA; n=8 per group) spontaneous pain scores. (E) Bevacizumab administered during the first 3 weeks of nerve injury blocked baseline (*, p<0.05, **, p<0.01, t-test, n=7 per group) and PE-evoked clustered firing (ns, not significant, t-test, n=6 per group).