Research ArticlePain

MicroRNA-30c-5p modulates neuropathic pain in rodents

See allHide authors and affiliations

Science Translational Medicine  08 Aug 2018:
Vol. 10, Issue 453, eaao6299
DOI: 10.1126/scitranslmed.aao6299

Targeting a microRNA for neuropathic pain

Neuropathic pain is a debilitating condition resulting from nerve damage. Often, patients do not achieve pain relief with currently available therapies. A better understanding of the mechanisms mediating neuropathic pain could help to identify more effective therapies. Now, Tramullas et al. show that neuropathic pain in rodents was associated with increased expression of the microRNA miR-30c-5p in the brain, cerebrospinal fluid, and plasma. In addition, miR-30c-5p was up-regulated in plasma and cerebrospinal fluid from patients with peripheral ischemia–induced pain. Inhibiting this microRNA in the rodent model produced analgesic effects. The results suggest that targeting miR-30c-5p might be an effective strategy for treating neuropathic pain.

Abstract

Neuropathic pain is a debilitating chronic syndrome that is often refractory to currently available analgesics. Aberrant expression of several microRNAs (miRNAs) in nociception-related neural structures is associated with neuropathic pain in rodent models. We have exploited the antiallodynic phenotype of mice lacking the bone morphogenetic protein and activin membrane-bound inhibitor (BAMBI), a transforming growth factor–β (TGF-β) pseudoreceptor. We used these mice to identify new miRNAs that might be useful for diagnosing, treating, or predicting neuropathic pain. We show that, after sciatic nerve injury in rats, miR-30c-5p was up-regulated in the spinal cord, dorsal root ganglia, cerebrospinal fluid (CSF) and plasma and that the expression of miR-30c-5p positively correlated with the severity of allodynia. The administration of a miR-30c-5p inhibitor into the cisterna magna of the brain delayed neuropathic pain development and reversed fully established allodynia in rodents. The mechanism was mediated by TGF-β and involved the endogenous opioid system. In patients with neuropathic pain associated with leg ischemia, the expression of miR-30c-5p was increased in plasma and CSF compared to control patients without pain. Logistic regression analysis in our cohort of patients showed that the expression of miR-30c-5p in plasma and CSF, in combination with other clinical variables, might be useful to help to predict neuropathic pain occurrence in patients with chronic peripheral ischemia.

INTRODUCTION

Chronic pain is a common debilitating disease that poses a severe clinical and economic burden for patients and the health system (1). Often, patients with neuropathic pain that arises after injuries of the somatosensory system do not respond to currently available treatments (2).

Our incomplete understanding of the mechanisms that sustain the development and chronification of neuropathic pain has hampered the design of new pharmacological approaches directed at the pathogenesis of the disease (3). Recent studies showed a link between deregulation of microRNA (miRNA) activity within nociception-related networks and the long-lasting changes in gene expression associated with aberrant processing of nociceptive inputs (46). miRNAs are small, noncoding RNAs involved in the posttranscriptional regulation of gene expression; they bind to complementary sequences of target mRNAs to induce either translational repression or mRNA degradation (7). Many miRNAs have been described so far, and although in silico computational analysis identified a high number of potential targets, only few have been unequivocally validated by experimental approaches (8). miRNAs have been shown to play crucial roles in the fine tuning of physiological functions and in the pathogenesis of many diseases. miRNA-targeted therapeutics showed some promising preclinical results (9), and one miRNA therapeutic entered clinical trial (10). Studies in rodent models revealed that neuropathic pain development is associated with modulation of the expression of several miRNAs in nociception-related neural structures (1120).

miRNAs released from cells can access bodily fluids, where they can be detected with high sensitivity and specificity (21). The value of circulating miRNAs for diagnostic, prognostic, and therapeutic stratification purposes is a matter of growing interest in clinical medicine (21, 22). Objective and reliable biological indicators of chronic pain would benefit our understanding of pain pathophysiology and analgesia-related mechanisms, improve diagnostic accuracy, and facilitate the development of novel analgesics (23).

Transforming growth factor–β (TGF-β) constitutes a family of pleiotropic, contextually acting cytokines (24). Emergent evidence supports a protective role for TGF-β signaling against the pathological neural plasticity underlying neuropathic pain in animal models (2529). We have demonstrated that mice deficient in bone morphogenetic protein and activin membrane-bound inhibitor (BAMBI), a TGF-β decoy receptor, present TGF-β gain of function in the nervous system, which protects mice against the development of allodynia after a peripheral nerve injury (NI) (25, 28, 29). Here, we discovered a candidate miRNA, miR-30c-5p, with potential value as an etiologic agent and therapeutic target for treating neuropathic pain. We also provided evidence that the interaction between miR-30c-5p and its target TGF-β modulated the endogenous opioid system. In addition, our data in patients with critical leg ischemia suggest that measuring miR-30c-5p concentration in plasma and cerebrospinal fluid (CSF) might help to determine the likelihood to develop neuropathic pain.

RESULTS

miRNA expression profiles in the spinal cord of wild-type and BAMBI−/− mice depend on their respective neuropathic pain–related phenotype

Our previous work demonstrates that deficiency of BAMBI amplifies TGF-β–mediated signals and protects the mice against neuropathic pain development after sciatic NI (29). Taking advantage of the antiallodynic phenotype of BAMBI−/− mice, we made it our first objective to identify potential miRNA candidates that could be involved in neuropathic pain. We analyzed by next-generation sequencing the differential expression profiles of miRNAs in the spinal dorsal horn (SDH) of BAMBI−/− and wild-type (WT) mice 14 days after NI. At that time point, WT mice (NI-WT) showed mechanical allodynia, determined with von Frey monofilaments (30), whereas BAMBI−/− mice (NI-BAMBI−/−) still showed normal responses to mechanical stimuli (Fig. 1A). Compared with NI-WT, the SDH of NI-BAMBI−/− mice showed four significantly down-regulated miRNAs (miR-30c-5p, P < 0.05; miR-181a-5p, P < 0.01; miR-100-5p, P < 0.05; miR-148a-3p, P < 0.05) and one overexpressed miRNA (miR-16-5p, P < 0.01; Table 1). The expression changes of this set of miRNAs were validated by qPCR in another series of sham and NI mice (Fig. 1B). miR-30c-5p was the only miRNA whose expression significantly increased in allodynic NI-WT but did not in nonallodynic NI-BAMBI−/− (repeated-measures ANOVA: “genotype × nerve injury,” P < 0.001; Fig. 1B).

Fig. 1 The antiallodynic phenotype of BAMBI−/− mice is associated with differential miRNA expression profiles in the SDH.

(A) Nocifensive responses to mechanical stimuli in WT and BAMBI−/− NI and sham mice (n = 8 to 10 per group). **P < 0.01, ***P < 0.001 versus NI-BAMBI−/− [two-way repeated-measures analysis of variance (ANOVA) with Bonferroni correction]. (B) miRNA relative expression (RE) validated by quantitative polymerase chain reaction (qPCR; n = 7 per group). **P < 0.01 versus sham; #P < 0.05, ##P < 0.01, ###P < 0.001 versus WT (two-way ANOVA with Bonferroni correction).

Table 1 miRNAs differentially expressed in the SDH from NI-WT and NI-BAMBI−/− mice.

Next-generation sequencing. n = 3 per group. P with Benjamini-Hochberg correction.

View this table:

Sciatic NI in rats induces neuropathic pain and miR-30c-5p up-regulation in nociception-related areas, CSF, and plasma

NI and sham rats were tested for mechanical sensitivity. As expected, 10 and 21 days after NI, rats showed mechanical allodynia (Fig. 2A). Plasma and CSF samples from the same rats were collected from the retro-orbital sinus and the cisterna magna, respectively, at baseline and on days 10 and 21 after NI, when mechanical allodynia was fully developed. At the indicated time points, the rats were sacrificed, and the expression of miR-30c-5p was determined in the lumbar SDH and dorsal root ganglion (DRG) by qPCR and in situ hybridization. Compared with sham rats, NI rats exhibited a significant up-regulation of miR-30c-5p in both the SDH (ANOVA, P < 0.01) and the DRGs (ANOVA, P < 0.05) on days 10 and 21 after NI. Moreover, the expressions of miR-30c-5p in CSF (repeated-measures ANOVA, P < 0.001) and plasma (repeated-measures ANOVA, P < 0.01) were significantly increased from the baseline values (Fig. 2B).

Fig. 2 miR-30c-5p is overexpressed in nociception-related areas, CSF, and plasma of neuropathic rats.

(A) Nocifensive responses to mechanical stimuli and threshold force (g) required to elicit responses 50% of the time in NI (n = 9) and sham rats (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 versus sham (one-way repeated-measures ANOVA with Bonferroni correction). (B) Top: miR-30c-5p relative expression in SDH and DRG in NI and Sham rats. *P < 0.05, **P < 0.01 versus sham (one-way ANOVA with Bonferroni correction). Bottom: miR-30c-5p relative expression in CSF and plasma collected from the same rat before (basal) and after NI (10 and 21 days). *P < 0.05, **P < 0.01 versus basal (one-way repeated-measures ANOVA with Bonferroni correction). (C) Peroxidase in situ hybridization of miR-30c-5p in spinal cord sections in sham and NI rats and in sham rats stained with scrambled miRNA probe (Scr; negative control). (D) Fluorescence in situ hybridization of miR-30c-5p (green) in SDH sections from sham and NI rats. (E) Combined miR-30c-5p (green) in situ hybridization and immunofluorescence for GFAP (top, red) or Iba1 (bottom, red) in SDH sections. (F) Fluorescence in situ hybridization of miR-30c-5p (green) in DRG cell dissociates in sham and NI rats.

Peroxidase and fluorescence in situ hybridizations (Fig. 2, C and D) were performed to show the topographic and cellular distribution of miR-30c-5p in the SDH from sham and NI rats. Positive signals were detected in the SDH from NI rats. In situ hybridization, combined with immunofluorescence staining of glial markers [glial fibrillary acidic protein (GFAP) for astrocytes and Iba1 for microglia], showed that miR-30c-5p in the SDH does not colocalize with glial markers (Fig. 2E), suggesting that miR-30c-5p expression was mostly neuronal. In contrast, fluorescence in situ hybridization in squash preparations (31) of isolated DRG cells showed that miR-30c-5p is expressed in both neurons and satellite glia (Fig. 2F).

In addition, our analysis revealed that the mechanical sensitivity thresholds after NI negatively correlated with miR-30c-5p expression in the SDH, DRG, CSF, and plasma (fig. S1, A to D). We found a significant (P < 0.01) and positive correlation between miR-30c-5p expression in CSF and plasma (fig. S1E), which suggests that miR-30c-5p released by neural cells into the extracellular fluid would access the CSF and, subsequently, the bloodstream. To support a flux of miRNAs between nervous tissue, CSF, and bloodstream, we injected the exogenous miRNA Caenorhabditis elegans–miR-39 (cel-miR-39) into the cisterna magna or in the tail vein on day 10 after NI. The expression of cel-miR-39 in CSF, plasma, SDH, and DRG was determined by qPCR (fig. S2). Cel-miR-39 administered into the cisterna magna was highly expressed in CSF at all examined time points after injection. The maximal expression in plasma was detected 6 hours after the injection and was detected in the SDH and DRG 12 hours after the injection (fig. S2, A to C, left). When administered into the tail vein, cel-miR-39 was detected in plasma 3 hours after the injection and remained detectable at 12 hours; the maximal expression in CSF, SDH, and DRG was detected at 12 hours (fig. S2, A to C, right panels).

The recruitment of miR-30c-5p to the miRNA-induced silencing complexes is enhanced in the spinal cord after NI in rats

We analyzed whether spinal overexpression of miR-30c-5p after NI is associated with a parallel increase in miR-30c-5p bound to Argonaute2 (Ago2) protein in the miRNA-induced silencing complexes (miRISCs), which would reflect the functional impact of miR-30c-5p overexpression on the translation of its targeted mRNAs (7). To this end, Ago2 was immunoprecipitated with a specific antibody (Ab) in SDH lysates from NI and sham rats (32). As shown in fig. S3, no differences between sham and NI rats were observed in Ago2 mRNA (fig. S3A) and protein (fig. S3B) expressions in total lysates (input) nor in Ago2 immunoprecipitates (fig. S3C). However, the presence of miR-30c-5p in Ago2 immunoprecipitates was significantly higher (P < 0.05) in NI than in sham rats (fig. S3D). These results indicate that the interaction of miR-30c-5p with Ago2 was enhanced in the SDH of neuropathic rats.

Pharmacological targeting of miR-30c-5p modulates pain sensitivity after sciatic NI in rats

To determine the functional role of miR-30c-5p on pain development, we assessed in vivo the consequences of miR-30c-5p pharmacological modulation on pain sensitivity. NI rats received three injections of miR-30c-5p inhibitor (100 or 200 ng/10 μl) or mismatch inhibitor into the cisterna magna at the time of NI or sham interventions and on days 4 and 7 after surgery. The responses to mechanical stimuli were monitored for a 2-month follow-up period. Spontaneous pain (33), cold allodynia (34), and thermal hyperalgesia (35) were assessed on day 10 after NI. The lower dose of miR-30c-5p inhibitor significantly delayed the onset of mechanical allodynia from the 7th to the 21st day after NI (two-way repeated-measures ANOVA, P < 0.01; Fig. 3A) and prevented miR-30c-5p up-regulation in the SDH, CSF, and plasma (Fig. 3B). The rats were also protected against spontaneous pain, cold allodynia, and thermal hyperalgesia (Fig. 3C). The higher dose of miR-30c-5p inhibitor induced a long-lasting preemptive effect for the entire 2-month follow-up period (Fig. 3A). The administration of miR-30c-5p inhibitor 11, 7, and 4 days before NI also delayed the onset of allodynia, but the protective effect was shorter compared to the treatment starting the day of NI (fig. S4, A and B). miR-30c-5p inhibitor injected to sham rats did not modify their nocifensive mechanical and thermal responses (fig. S5, A and B).

Fig. 3 miR-30c-5p inhibitor prevents and reverses neuropathic pain in rats subjected to NI.

(A) Top: Intracisternal administration protocol of miR-30c-5p inhibitor [100 ng/10 μl (n = 9) or 200 ng/10 μl (n = 5)] or mismatch inhibitor (n = 6). Bottom: Time course of the threshold force (g) required to elicit responses 50% of the time for the groups indicated in legend. *P < 0.05, **P < 0.01, ***P < 0.001 versus NI + mismatch inhibitor (two-way repeated-measures ANOVA with Bonferroni correction). (B) miR-30c-5p relative expression in the SDH, CSF, and plasma in sham and NI rats ± miR-30c-5p inhibitor. *P < 0.05, **P < 0.01 versus sham; #P < 0.05, ###P < 0.001 versus NI + mismatch inhibitor (n = 6 per group; one-way ANOVA with Bonferroni correction). (C) Thermal hyperalgesia, cold allodynia, and spontaneous pain for the groups indicated in the legend. *P < 0.05, versus sham; #P < 0.05, ##P < 0.01 versus NI + mismatch inhibitor (n = 4 to 6 per group; Kruskal-Wallis with Dunn correction). (D) Top: Intracisternal administration protocol of miR-30c-5p inhibitor or mismatch inhibitor, starting (arrows) on day 7 (early) or on day 22 (late) after NI. Bottom: Time course of the threshold force (g).*P < 0.05, **P < 0.01, ***P < 0.001 versus NI + early mismatch inhibitor; ##P < 0.01, ###P < 0.001 versus NI + late mismatch inhibitor (n = 8 to 10 per group; two-way repeated-measures ANOVA with Bonferroni correction). (E) Fluorescence signals showing CY3-miRNA inhibitor within the spinal cord 12 hours after the injection into the cisterna magna. Bottom left: Lack of signal after vehicle.

We subsequently evaluated the effectiveness of miR-30c-5p inhibitor at suppressing already established neuropathic pain (Fig. 3D). An early cycle of miR-30c-5p inhibitor administered during the initial stages of allodynia (200 ng/10 μl; days 7, 9, 11, and 13 after NI) produced a rapid, progressive, and long-lasting antinociceptive effect; the nociceptive threshold recovered baseline values after the third injection of the inhibitor and lasted for the 2-month follow-up period (Fig. 3D). Moreover, administration of a late cycle of miR-30c-5p inhibitor for 2 weeks (days 22, 24, 26, and 28 after NI) to rats suffering from severe allodynia induced pain relief of rapid onset (48 hours after the first administration), and the analgesic effect lasted until the end of the experiment (5 weeks after the last injection; Fig. 3D).

Uptake of the miR-30c-5p inhibitor by spinal cells was evidenced by injecting a fluorescent-labeled miRNA inhibitor (200 ng/10 μl) into the cisterna magna (Fig. 3E). miR-30c-5p inhibitor injected in the tail vein, at higher doses (three injections of 400 and 2000 ng each), did not prevent the development of neuropathic pain nor did it reverse the established allodynia (fig. S6, A to C).

To confirm the role of miR-30c-5p in modulating neuropathic pain, we injected a synthetic miR-30c-5p mimic (100 ng/10 μl; days 0, 1, and 3 after NI) or a scrambled mimic (control) in the cisterna magna of NI rats (fig. S7A). The responses to mechanical stimuli were monitored for a 7-day follow-up period. miR-30c-5p mimic accelerated the onset of allodynia from days 7 to 5 after NI (fig. S7, B and C). Neither miR-30c-5p mimic nor the scrambled mimic modified the nocifensive responses of sham rats (fig. S7, B and C). The treatment with miRNA mimic increased the expression of miR-30c-5p in the SDH (fig. S7D).

TGF-β is involved in the effects of miR-30c-5p on neuropathic pain in rodents

The antiallodynic phenotype of BAMBI−/− mice depends on an increased TGF-β signaling (28, 29), and these mice did not overexpress miR-30c-5p in the SDH after NI. Therefore, we assessed whether the protective effect of TGF-β on neuropathic pain involves a modulatory effect on miR-30c-5p expression. WT mice received a 2-week subcutaneous infusion of recombinant TGF-β1 (rTGF-β1) or vehicle, starting at the time of NI. Fourteen days after NI, control WT mice that received vehicle injection displayed allodynia. Conversely, WT mice treated with rTGF-β1 did not show mechanical allodynia (Fig. 4A). Moreover, control WT mice treated with vehicle showed miR-30c-5p overexpression in the SDH 14 days after NI, whereas treatment with rTGF-β1 prevented miR-30c-5p up-regulation (Fig. 4B). The SDH expression values of miR-30c-5p negatively correlated with the mRNA expression of TGF-β1 (Fig. 4C). Similar to the in vivo results, the administration of rTGF-β1 (10 ng/ml) to the medium in cultured cells (SH-SY5Y neuroblasts, fibroblasts, and vascular smooth muscle cells) resulted in miR-30c-5p down-regulation (Fig. 4D).

Fig. 4 TGF-β1 is involved in the effects of miR-30c-5p on neuropathic pain in vivo and in vitro.

(A) Nocifensive responses of sham and NI mice treated with a 14-day subcutaneous infusion of rTGF-β1 or vehicle (n = 5 per group) to mechanical stimuli. *P < 0.05, **P < 0.01, ***P < 0.001 versus NI + rTGF-β1 (two-way repeated-measures ANOVA with Bonferroni correction). (B) miR-30c-5p expression in the SDH from sham and NI mice ± rTGF-β1. *P < 0.05 versus sham; ##P < 0.01 versus NI (ANOVA with Bonferroni correction). (C) Correlation between TGF-β1 mRNA and miR-30c-5p in the SDH from NI rats [Pearson’s coefficient (r); P < 0.001]. (D) miR-30c-5p expression changes induced by rTGF-β1 (10 ng/ml) in cultured cells. *P < 0.05, **P < 0.01 versus respective control (Student’s t test). (E and F) Relative expression of TGF-β1 in the SDH of sham and NI rats treated with miR-30c-5p mimic (E) or miR-30c-5p inhibitor (F). #P < 0.05, ###P < 0.001 versus NI (ANOVA with Bonferroni correction). (G) TGF-β1 fold change induced by miR-30c-5p mimic and miR-30c-5p inhibitor in SH-SY5Y cells. *P < 0.05, **P < 0.01 versus scrambled mimic (Scr; purple bars) or mismatch inhibitor (Mism; red bars) (one-way ANOVA with Bonferroni correction). (H) Relative Luciferase activity (RLU) after cotransfection of SH-SY5Y cells with pMIR-luc containing the 3′UTR of TGF-β1 or the empty vector and miR-30c-5p mimic (10 nM). **P < 0.01 (Student’s t test). (I) Intracisternal administration protocol of miR-30c-5p inhibitor or mismatch inhibitor. Pretreated animals received an intracisternal injection of a neutralizing Ab anti–TGF-β or an irrelevant immunoglobulin [sham + mismatch inhibitor + IgG1 (n = 3), NI + miR-30c-5p inhibitor + TGF-β Ab (n = 4), NI + miR-30c-5p inhibitor + IgG1 (n = 3), NI + mismatch + IgG1 (n = 5)]. (J) Time course of the threshold force (g) required to elicit responses 50% of the time for the groups indicated in the legend; arrow shows the time of injection of anti–TGF-β (TGF-β Ab) or irrelevant immunoglobulin (IgG1). **P < 0.01, ***P < 0.001 versus NI + miR-30c-5p inhibitor + IgG1 (two-way repeated-measures ANOVA with Bonferroni correction). (K to M) Double immunofluorescence staining of TGF-β (green) and GFAP (red) (K and L) or Iba1 (red) (M) in preparations of SDH cell dissociates from NI rats. (N) Immunofluorescence staining of TGF-β (green) in DRG cell dissociates from NI rats. (O) GFAP (red) and nonspecific TGF-β immunostaining (green) after neutralizing the Ab with a specific blocking peptide.

To confirm the direct relationship between miR-30c-5p and TGF-β1 in the context of neuropathic pain, we showed that NI rats treated with miR-30c-5p mimic exhibited a small but significant (ANOVA, P < 0.05) down-regulation of TGF-β1 mRNA in the SDH, as compared with the scrambled mimic (Fig. 4E). Conversely, the miR-30c-5p inhibitor induced TGF-β1 overexpression compared with the mismatch inhibitor (ANOVA, P < 0.001; Fig. 4F).

In addition, we transfected SH-SY5Y neuroblasts with miR-30c-5p mimic and miR-30c-5p inhibitor to respectively overexpress and knockdown miR-30c-5p (fig. S8). miR-30c-5p mimic significantly (ANOVA, P < 0.01) reduced the expression of TGF-β1, whereas miR-30c-5p inhibitor had the opposite effect (ANOVA, P < 0.01), increasing TGF-β1 expression (Fig. 4G).

The structure-based Probability of Interaction by Target Accessibility (PITA) algorithm (http://genie.weizmann.ac.il/pubs/mir07/mir07_exe.html) (36) predicts the presence of binding sites for miR-30c-5p in the 3′ untranslated region (3′UTR) sequence of TGF-β1. PITA’s notation (7.1.1) indicated that the seed region is a 7-mer, with one mismatch and one guanine-uracile wobble.

To assess whether TGF-β1 is posttranscriptionally regulated by miR-30c-5p, we cotransfected SH-SY5Y neuroblasts with a pMIR-REPORT luciferase vector containing the 3′UTR of TGF-β1 together with miR-30c-5p mimic (10 nM) or a scrambled mimic. Transfection of miR-30c-5p mimic, but not the scrambled mimic, resulted in a significant (P < 0.01) reduction of luciferase activity (Fig. 4H), suggesting that TGF-β1 mRNA was targeted by miR-30c-5p in SH-SY5Y cells.

In a series of NI rats, we assessed the involvement of TGF-β in the antiallodynic effect of miR-30c-5p inhibitor (200 ng; days 0, 4, and 7 after NI; Fig. 4, I and J). On day 10 after NI, when miR-30c-5p inhibitor–treated rats were fully protected against neuropathic pain, the animals received a single intracisternal injection of a neutralizing monoclonal Ab anti–TGF-β (TGF-β Ab; 50 μg/10 μl) or an irrelevant isotype-matched immunoglobulin [immunoglobulin G1 (IgG1)]. TGF-β Ab antagonized the antiallodynic effect of miR-30c-5p inhibitor within 24 hours after injection, but IgG1 did not. TGF-β Ab did not modify the mechanical sensitivity of sham rats (Fig. 4J).

To determine the regional distribution of TGF-β immunoreactivity in the nervous system, we performed double immunofluorescence of TGF-β and glial markers [GFAP, astrocytes; ionized calcium-binding adapter molecule 1 (Iba1), microglia] in preparations of dissociated neurons and glia from the SDH (30). Our results evidenced that TGF-β was localized both in neurons and glial cells (Fig. 4, K to M). In DRG, TGF-β was detected in neurons and satellite glia (Fig. 4N). Pretreatment of the preparations with a specific blocking peptide prevented TGF-β immunostaining (Fig. 4O).

The antiallodynic effect of miR-30c-5p inhibitor involves the endogenous opioid system

We have previously demonstrated that the TGF-β–dependent antiallodynic effect in mice depends on an increased activity of the endogenous opioid system (28, 29). Therefore, we postulated that the antiallodynic effect of miR-30c-5p was mediated by endogenous opioids. To test this hypothesis, NI rats treated with miR-30c-5p inhibitor or mismatch inhibitor subsequently received the opioid antagonist naloxone (1 mg/kg, subcutaneously) on day 30 after NI (Fig. 5, A and B). Rats treated with miR-30c-5p inhibitor, which showed normal nocifensive responses at this time point, exhibited mechanical allodynia immediately after the injection of naloxone, and mechanosensitivity recovered normal values within 24 hours (Fig. 5C). Naloxone did not modify mechanosensitivity in sham and NI rats treated with the mismatch inhibitor.

Fig. 5 The antiallodynic effect of miR-30c-5p involves the endogenous opioid system.

(A) Intracisternal administration protocol of miR-30c-5p inhibitor (n = 5) or mismatch inhibitor (n = 5). NI rats received naloxone on day 30. (B) Nocifensive responses to mechanical stimuli for the groups indicated in figure. *P < 0.05, **P < 0.01, ***P < 0.001 versus NI + miR-30c-5p inhibitor (two-way repeated-measures ANOVA with Bonferroni correction). (C) Threshold force (g) required to elicit responses 50% of the time in the animal groups indicated in figure **P < 0.001 versus baseline (two-way repeated-measures ANOVA with Bonferroni correction).(D) Rats treated with a cycle of miR-30c-5p mimic (n = 4) or scrambled mimic (n = 3) received an injection of naloxone or saline before restraint stress exposure. (E) Paw withdrawal latencies to thermal stimuli in the animal groups indicated in figure. ***P < 0.001 versus respective baseline (white bars); ##P < 0.01 versus scrambled after stress (two-way repeated-measures ANOVA with Bonferroni correction).

To further support the relationship between miR-30c-5p and the endogenous opioid system, we evaluated the effects of miR-30c-5p mimic in the acute restraint model of stress-induced analgesia, which is dependent on the activity of the endogenous opioid system (37). Rats were treated with intracisternal miR-30c-5p mimic (100 ng/10 μl; days 0, 1, and 4) or scrambled mimic and, 24 hours after, they were exposed to restraint stress for 30 min. Thermal nociception was assessed before and immediately after stress (Fig. 5D). The rats treated with scrambled mimic exhibited significantly higher paw withdraw latencies after stress than at baseline (two-way repeated-measures ANOVA, P < 0.001). Naloxone prevented the stress-induced analgesia. Consistent with a modulatory role of miR-30c-5p on the endogenous opioid activity, the animals treated with miR-30c-5p mimic did not exhibit stress-induced analgesia (Fig. 5E).

Patients suffering from ischemic neuropathic pain present elevated miR-30c-5p expression values in plasma and CSF

The potential translation of the experimental data into humans was assessed in a small cohort of patients suffering from severe leg ischemia associated with pain with neuropathic characteristics [n = 25; Douleur Neuropathique 4 (DN4) score ≥ 4] that lasted more than 4 months. Patients with no evidence of neuropathic pain, either with leg ischemia or without ischemia, served as controls (n = 35). The characteristics of the patients are depicted in Table 2.

Table 2 Clinical and demographic characteristics of the patients.
View this table:

Circulating miR-30c-5p was determined by relative expression and number of copies per microliter (38). Patients affected by neuropathic pain presented significantly higher plasma (P < 0.001) and CSF (P < 0.01) expression of miR-30c-5p than patients without pain. This higher expression was unrelated to the ischemic condition, diabetes, or gender (Fig. 6). miR-30c-5p could be detected in T lymphocytes (CD3+), B lymphocytes (CD19+), monocytes (CD14+), and granulocytes (CD15+), with monocytes and T lymphocytes showing the highest expression (fig. S9).

Fig. 6 Patients suffering from neuropathic pain exhibited increased expression values of miR-30c-5p in plasma and CSF.

(A) miR-30c-5p values in CSF (left) and plasma (right) in patients with neuropathic pain (n = 35) and in pain-free controls (n = 25). **P < 0.01, ***P < 0.001 versus controls (Student’s t test). (B to D) Influence of ischemia (B), diabetes (C), and gender (D) on CSF (left) and plasma (right) miR-30c-5p (one- or two-way ANOVA with Bonferroni correction).

Circulating miR-30c-5p values in either plasma or CSF predict neuropathic pain in patients with critical limb ischemia

The capability of circulating miR-30c-5p in plasma and/or in CSF, either alone or in conjunction with clinical parameters (age, sex, diabetes, hypertension, and obesity), to estimate the likelihood of suffering from neuropathic pain was assessed by logistic regression analysis. In our small cohort, when analyzed individually, miR-30c-5p expression in plasma and CSF, diabetes mellitus, and age behaved as significant independent positive predictors of neuropathic pain occurrence (Table 3, models #1 to #4). The model obtained if diabetes and age were combined with miR-30c-5p in plasma [Table 3 (model #9) and Fig. 7A] presented the highest area under the receiver operator characteristic (ROC) curve (0.93), with a sensitivity of 80% and a specificity of 94%. When the same analysis was performed using miR-30c-5p in CSF [Table 3 (model #10) and Fig. 7B], the area under the ROC curve was 0.90, with a sensitivity of 83% and a specificity of 90%. The relative goodness of fit of the candidate models was further assessed by the Akaike information criteria (AIC; Table 3) (39). AIC reflects the amount of information lost when a given model is used to represent reality; therefore, the preferred model is the one that has the lowest AIC value. The best AIC was shown by model #9 (age, diabetes, and miR-30c-5p in plasma). The second best model (Table 3, model #7) has 0.19 times as high a probability as the best model of minimizing the information loss.

Table 3 Diagnostic in silico models.

The models include the following as independent variables: circulating miR-30c-5p in plasma and CSF, diabetes mellitus, and age analyzed individually or combined. AUC, area under the ROC curve; CI, confidence intervals; OR, odds ratio.

View this table:
Fig. 7 Multiple logistic regression models predictive of neuropathic pain occurrence in patients.

(A) Diagnostic in silico models: The models include the following as independent variables: circulating miR-30c-5p in plasma and CSF, diabetes mellitus, and age analyzed individually or combined. (B) ROC curves of the models including different variables (as shown in figure).

DISCUSSION

Here, we have exploited the antiallodynic phenotype of mice lacking the TGF-β decoy receptor BAMBI (25, 28, 29) to identify potentially relevant miRNAs involved in neuropathic pain establishment. Next-generation sequencing and qPCR showed that miR-30c-5p was up-regulated in the SDH of neuropathic WT mice but remained unchanged in allodynia-resistant BAMBI−/− mice. In rats, neuropathic pain was associated with miR-30c-5p overexpression in the SDH and DRGs. The functional significance of miR-30c-5p up-regulation was supported by the direct correlation between the intensity of allodynia and the expression of miR-30c-5p in nociception-related areas. Moreover, the enrichment of miR-30c-5p bound to Ago2 in the miRISC obtained from SDHs of neuropathic rats suggests that the posttranscriptional modulation of mRNAs targeted by miR-30c-5p is increased under the neuropathic pain condition. Our results point to a relevant contribution of miR-30c-5p to allodynia development and make it a promising target for treating neuropathic pain states.

Here, intracisternal administration of miR-30c-5p inhibitor prevented the development of neuropathic pain in NI rats in a dose-dependent manner. The lower dose delayed the onset of mechanical allodynia for about 2 weeks, and the higher dose prevented allodynia development throughout the 2-month follow-up period (duration of the experiment).

miR-30c-5p inhibitor also exerted long-lasting pain relief when administered either in the early stages of allodynia or after a long period of maximally intense neuropathic pain. Time course experiments showed that both early- and late-treated rats recovered normal mechanical sensitivity within the first week of treatment and remained free of allodynia for the complete follow-up period (2 months). Such a long-lasting effect raises the possibility that those animals, if followed-up longer, would never have regained the neuropathic condition. From a clinical point of view, this finding is particularly relevant because, in real-life patients, the neurological origin of pain is often missed initially, and the administration of specific treatments is significantly delayed (40).

The present study and others (2529) support a protective role for TGF-β against neuropathic pain development. Although we cannot exclude relevant roles for other miR-30c-5p targets, our results suggest that TGF-β1 is a target of miR-30c-5p under neuropathic pain conditions; this hypothesis is based on the following results: (i) TGF-β1 exerted antiallodynic effects in rats. (ii) TGF-β1 mRNA was down-regulated by miR-30c-5p mimic and up-regulated by miR-30c-5p inhibitor. In addition, the mRNA expression of TGF-β1 in the SDH of NI mice correlated inversely with the expression of miR-30c-5p. (iii) miR-30c-5p has a seed-matching sequence in the 3′UTR region of TGF-β1, and TGF-βs have been predicted in silico to be targeted by miR-30c-5p (8). Using a luciferase reporter construct containing the 3′UTR sequence of TGF-β1 (41), we confirmed that miR-30c-5p mimic inhibited the expression of the reporter gene in SH-SY5Y neuroblasts.

Our results also support bidirectional modulatory networks between TGF-β1 and miR-30c-5p, as previously reported in models of hepatic fibrosis (42). TGF-β gain of function produced by either genetic (BAMBI deficiency) or pharmacological (administration of rTGF-β1) tools prevented miR-30c-5p up-regulation and neuropathic pain development. In addition, the administration of rTGF-β1 to SH-SY5Y neuroblasts induced miR-30c-5p down-regulation. The reversion of the antiallodynic effect of miR-30c-5p inhibitor by a neutralizing Ab to TGF-β strongly suggests that the antagonistic relationship between miR-30c-5p and TGF-β signaling is a relevant pathophysiological mechanisms underlying neuropathic pain.

Disruption of the homeostatic balance in the descending pain regulatory systems, toward descending facilitation, contributes to pain chronification after injury (43). In particular, several studies showed association between chronic pain states and dysfunction of the descending inhibitory opioid system (44). Our previous findings (28, 29) and present results show that the endogenous opioid system is a downstream effector of the TGF-β–induced antiallodynic effect. BAMBI−/− mice, which present TGF-β gain of function, and mice treated with rTGF-β did not develop allodynia and do not show miR-30c-5p up-regulation after NI. Moreover, the antiallodynic effect of miR-30c-5p inhibitor was reversed by the opioid antagonist naloxone. We propose that, under neuropathic pain conditions, TGF-β1 down-regulation subsequent to miR-30c-5p overexpression in nociception-related areas could result in less efficient pain modulation by the descending opioid system. According to this hypothesis, miR-30c-5p inhibitor would prevent the expression of signs of enhanced pain by amplifying TGF-β1 signaling in pain modulatory circuits. Therefore, the effect of miR-30c-5p modulation is displayed only when the activity of the descending pain control plays a critical role in pain modulation, as occurs in NI animals (43). In addition, our results showed that acute stress-induced analgesia, which is mediated by endogenous opioids (37), was prevented by miR-30c-5p mimic.

In the clinic, determining accurately whether chronic pain states are of neuropathic origin is a challenge (40). The diagnosis is based on anamnesis and clinical evaluation of highly subjective symptoms (allodynia, hyperalgesia, and dysesthesia) that are suggestive, but not pathognomonic, of neural damage. There are currently no objective indicators that can be used to identify individuals who suffer or are at risk of developing neuropathic pain (23).

Despite the great deal of interest in the study of circulating miRNAs as minimally invasive biomarkers in human diseases (21), little attention has been paid to chronic pain (45, 46), and very few studies have dealt with neuropathic pain (16, 4749). Herein, circulating miR-30c-5p in CSF and plasma increased after the establishment of the neuropathic state in rats. Moreover, the nocifensive responses to mechanical stimuli correlated directly with the expression of miR-30c-5p in both CSF and plasma. Our results thus suggest that analysis of circulating miR-30c-5p in pain states might provide useful information on pain assessment and management.

Patients suffering from neuropathic pain in the context of severe chronic leg ischemia exhibited elevated expression of miR-30c-5p in plasma and CSF. Ischemia can be discarded as a confounding factor because the circulating values of miR-30c-5p in pain-free patients were similar regardless of their ischemic condition. These findings suggest that miR-30c-5p expression in human biofluids is differentially regulated according to the painful condition of the neuropathy. Moreover, univariate logistic regression analyses showed that circulating miR-30c-5p, either in plasma or in CSF, correlated with neuropathic pain severity. However, further analysis in multiple large independent cohorts of patients suffering from pain of different etiology is necessary to determine the diagnostic value of miR-30c-5p. Although the CSF miRNA profile should better reflect that of the brain tissue and yield insights into disease pathophysiology, in terms of clinical applicability, blood is preferable over CSF because it is accessible by performing a safer and less invasive procedure.

Consistently, with a previous report showing that age, diabetes, and peripheral arterial disease are the main contributors to an increased risk of suffering neuropathic pain in the general population (50), inclusion of age and diabetes together with circulating miR-30c-5p, either in plasma or in CSF, in a multiple logistic regression framework, substantially improved the diagnostic power in our small cohort of ischemic patients. On the basis of the area under the ROC curve and the AIC, the model including miR-30c-5p in plasma, age, and diabetes as independent variables was the best of the candidate set for neuropathic pain risk assessment. Although the prediction model was internally validated by the bootstrapping method, further studies are warranted to validate our findings in independent cohort of patients and to assess the usefulness of circulating miR-30c-5p to predict individuals at higher risk for developing postamputation phantom limb pain, which is a complication extremely challenging to treat (51).

An important mode of intercellular communication in the nervous system is mediated by miRNAs released by cells into the extracellular space, incorporated in extracellular vesicles, or bound to high-density lipoproteins or Argonaute (52). The CSF contains specific signatures of miRNA expression for various central nervous system (CNS) pathologies (53). In addition, neuron-derived miRNAs have been detected in the bloodstream of patients with neurological and psychiatric disorders and brain tumors (53). Here, we found a positive correlation between circulating miR-30c-5p in plasma and CSF in NI rats. This finding suggests that miR-30c-5p released by neural cells into the extracellular fluid would access the CSF and, subsequently, the systemic bloodstream. In support of this hypothesis, our results showed that administration of miR-30c-5p inhibitor into the cisterna magna resulted in a reduction of miR-30c-5p expression in the nervous system, as well as in CSF and plasma. In addition, the exogenous-miRNA cel-miR-39, which has no homologous sequences in rats (53), was detected in the SDH, DRG, CSF, and plasma after its injection into either the cisterna magna or the tail vein in NI rats. These results suggest a bidirectional traffic of miRNAs, in a stable form, between nervous system, CSF, and bloodstream. Thus, in rats, miR-30c-5p released by neural cells could reach the CSF and, subsequently, the systemic circulation through the physiological CSF reabsorption into venous sinuses. However, we cannot discard a contribution from peripheral sources to circulating miR-30c-5p.

In contrast with rats, patients showed no relationship between circulating miR-30c-5p in plasma and CSF. Moreover, miR-30c-5p in plasma was more effective than its expression in the CSF at predicting neuropathic pain in our cohort of ischemic patients. These findings raise the possibility that, under the ischemic condition, peripheral sources of miR-30c-5p contributing to neuropathic pain might exist. The access of miR-30c-5p from the bloodstream to the CNS would be facilitated by the disruption of the blood-brain barrier that occurs during the neuropathic pain condition (54). The ubiquitous expression of miR-30c-5p in almost all tissues, including white blood cells, platelets, and vascular endothelial cells (5557), makes it difficult to ascertain the actual source of peripheral miR-30c-5p in our patient’s cohort.

There are a number of limitations in this study. Although sciatic injury in rodents is a widely used model of peripheral neuropathic pain, there are many differences with the clinical scenario (etiology and time course of the painful neuropathy, comorbidities, age and sex of the affected subjects, pharmacological treatments, etc.) that prompt to caution when tempted to extrapolate directly experimental results into clinical practice. For example, diabetes mellitus, which affects 44% of our cohort, underlies an increased susceptibility to neuropathic pain (1). In addition, the chronic inflammation state, yielded by impaired vascular function, age, and diabetes, can dysregulate the balance of miRNA expression (58).Therefore, the impact of a number of confounders in our observations is unavoidable.

MATERIALS AND METHODS

Study design

Here, we assessed the potential value of miR-30c-5p as etiologic agent and therapeutic target in neuropathic pain. We performed studies in patients with critical leg ischemia suffering from neuropathic pain, in rodent models of allodynia induced by peripheral NI, and in cultured cells. The following experimental studies were designed: (i) BAMBI-deficient mice subjected to sciatic nerve crush injury served to identify, by next-generation sequencing and qPCR, the relationship between miR-30c-5p up-regulation in the SDH and the neuropathic pain state. (ii) The cellular localization of miR-30c-5p in DRG and SDH was determined in rats subjected to the spared NI model of neuropathic pain (NI rats) by in situ hybridization combined with immunofluorescence. In addition, we assessed the relationship between pain severity and miR-30c-5p expression, determined by qPCR, in neural tissues and bodily fluids. (iii) The functional consequences of miR-30c-5p modulation on allodynia establishment and recovery were assessed in NI rats treated with a miR-30c-5p mimic and a miR-30c-5p inhibitor. (iv) The inhibitory role on TGF-β1 by miR-30c-5p under neuropathic pain conditions was assessed using pharmacological tools (rTGF-β and TGF-β Ab) both in vivo (NI rodents) and in vitro (luciferase reporter experiments in SH-SY5Y neuroblasts). (v) The administration of the opioid antagonist naloxone and the restrain stress-induced analgesia model were used to assess the regulatory role of miR-30c-5p on the endogenous opioid system. (vi) The dysregulation of miR-30c-5p in CSF and plasma and the capability of circulating miR-30c-5p to estimate the likelihood of suffering neuropathic pain were assessed by qPCR and logistic regression analysis in a cohort of patients with lower limb critical ischemia, suffering from neuropathic pain, and control patients without pain.

In each experimental series, the surgical intervention and the administration of treatments were performed by an operator that was not involved in the subsequent behavioral evaluation of the animals. Each animal was coded with a microchip, and the researcher who carried out the rest of the experiment was not aware of their group allocation. The investigators that performed surgery/treatments and behavioral assessing were rotated. The investigator that analyzed the nocifensive behavioral data did know which animals were grouped together but not what treatment these groups received.

Studies in patients

The study followed the Declaration of Helsinki guidelines for investigation on human subjects. The Cantabria Institutional Ethics and Clinical Research Committee approved the study, and all patients gave written informed consent.

We prospectively studied 25 patients with chronic peripheral ischemia who suffered from persistent neuropathic pain symptoms of more than 4 months duration and underwent surgery for lower extremity revascularization or amputation under regional anesthesia in the Cardiovascular Surgery Unit at the University Hospital Valdecilla in Santander, Spain. The control group consisted of patients who lacked painful pathologies and underwent elective surgery under regional anesthesia (n = 35). Demographic and clinical characteristics of the patients and their preoperative pharmacological treatments are shown in Table 2.

The neuropathic pain condition was assessed using the well-validated questionnaire, DN4 (59). To minimize interobserver variability, all patients were uniformly examined and interviewed by the same physician. The questionnaire has components of how the pain feels to the patient and assesses the occurrence of hypoesthesia and allodynia.

During the spinal anesthesia procedure, a CSF sample (100 μl) was collected. A peripheral venous blood sample (10 ml) was collected from an antecubital vein 24 hours preoperatively. To minimize platelet degranulation and hemolysis, we drew the blood without a tourniquet, using a syringe with a wide-gauge needle, and the blood was then gently transferred to a collection tube containing EDTA. Within 30 min of collection, the plasma was separated by centrifugation at 1500g for 15 min, and platelet-poor plasma was obtained by recentrifuging the plasma for another 10 min at 1500g, all at room temperature. CSF and plasma aliquots were immediately frozen on dry ice and stored at −80°C until assayed.

Studies in rodents

The experiments were performed in 8-week-old male WT (C57BL/6) and BAMBI knockout (BAMBI−/−) mice (in a C57BL/6 genetic background) and in 8-week-old (250 to 300 g) male Sprague-Dawley rats. The study was approved by the University of Cantabria Institutional Laboratory Animal Care and Use Committee (reference IP0415) and conducted in accordance with the guidelines from directive 2010/63/EU of the European Parliament and the International Association for the Study of Pain.

Two different models of neuropathic pain were used. In mice, the sciatic nerve crush injury model was used as previously described (28, 29, 60). Briefly, mice were anesthetized with isoflurane (1.5 to 2%), and the left common sciatic nerve was exposed, isolated from surrounding connective tissue and crushed for 7 s using smooth forceps. Sham-operated mice underwent the same surgical procedure, but the nerve was exposed and left intact. In rats, the spared NI model (61) was used. The left common sciatic nerve was exposed under isoflurane anesthesia, and the tibial and common peroneal branches of the nerve were sectioned. Sham-operated rats underwent the same procedure, but the sciatic and its branches were left intact. The development of neuropathic pain was assessed.

The following treatments were administered: miR-30c-5p mimic and miR-30c-5p inhibitor (mirVana, Thermo Fisher Scientific) were diluted in a mixture of Lipofectamine and artificial CSF or saline and injected into the cisterna magna (100 to 200 ng/10 μl) or in the tail vein (400 ng/100 μl). CY3-labeled miRNA inhibitor (Thermo Fisher Scientific) diluted in a mixture of Lipofectamine and artificial CSF was injected into the cisterna magna (200 ng/10 μl). Cel-miR-39 (Qiagen) diluted in a mixture of Lipofectamine and artificial CSF or saline was injected into the cisterna magna (300 ng/10 μl) or in the tail vein (300 ng/100 μl). The opioid antagonist naloxone (Sigma-Aldrich) was diluted in saline and injected (100 μl; 1 mg/kg, subcutaneously). Recombinant TGF-β1 (R&D Systems) diluted in 4 nM hydrochloric acid (HCl) plus albumin (2 mg/ml) in phosphate-buffered saline was administered using osmotic minipumps (6.2 ng/hour, 14 days; Alzet 1002). A monoclonal neutralizing TGF-β Ab [1D11.16.8 clone (62)] diluted in artificial CSF (50 μg/10 μl) was administered into the cisterna magna.

Statistical analysis

GraphPad Prism 5.01, Predictive Analytics SoftWare (PASW) 22 [Statistical Package for the Social Sciences (SPSS) Inc.], and Stata 14/SE (StataCorp) packages were used. Data from mice and rats were expressed as means ± SEM. Differences between two independent groups were assessed using the two-tailed Student’s t test. Differences between multiple groups were analyzed by one- or two-way repeated-measures ANOVA followed by Bonferroni post hoc test. Correlations between mRNA, miRNA expression values, and the mechanical threshold were performed using Pearson’s correlation analysis. Next-generation sequencing analysis was performed using Mann-Whitney test with Benjamini-Hochberg’s correction.

Patient’s data sets were assessed with the D’Agostino and Pearson omnibus normality test. Continuous variables were expressed as the mean ± SD if Gaussian and as median (25th and 75th interquartile range) if non-Gaussian. To assess statistical dependence between quantitative variables, we used the Spearman’s rank correlation coefficient (rho).

Predictors of neuropathic pain in patients were identified via logistic regression analysis: All models combining age, diabetes, and miR-30c-5p plasma/miR-30c-5p CSF were estimated. The Hosmer-Lemeshow test was used to evaluate goodness of fit of the model. A post hoc assessment of the regression model was performed with the bootstrapping method, with 1000 iterations. The ROC curve was calculated to assess the capability of the model to discriminate patients with neuropathic pain symptoms from pain-free patients. To assess the best model among candidates, we applied the AIC, which can be interpreted as the amount of information loss by each model (39). We also estimated their relative likelihood as Embedded Image, where AICi is the AIC for model i, and AICmin is the minimum of the AIC obtained. The relative likelihood reflects the probability of each model to minimize the information loss versus the best model. Raw data are shown in table S1.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/453/eaao6299/DC1

Methods

Fig. S1. Correlation of miR-30c-5p expression with allodynia severity after sciatic NI in rats.

Fig. S2. qPCR amplification curves of cel-miR-39.

Fig. S3. Recruitment of miR-30c-5p by Ago2 to miRISCs in the spinal cord after NI in rats.

Fig. S4. Preinjury administration of miR-30c-5p inhibitor delays neuropathic pain development after sciatic NI in rats.

Fig. S5. Administration of miR-30c-5p inhibitor into the cisterna magna neither influenced the baseline response to mechanical nor thermal stimuli in sham-operated rats.

Fig. S6. Administration of miR-30c-5p inhibitor into the tail vein do not prevent the development of neuropathic pain neither reverse the established allodynia after sciatic NI.

Fig. S7. miR-30c-5p mimic accelerates allodynia development after sciatic NI in rats.

Fig. S8. Modulation of miR-30c-5p expression in SH-SY5Y neuroblastoma cells.

Fig. S9. Relative miR-30c-5p expression in peripheral white cells from control patients.

Table S1. Raw data.

REFERENCES AND NOTES

Acknowledgments: We thank A. Cayón, N. García, R. Moreta, M. Navarro, M. T. Berciano, and J. F. Cryan. Funding: This work was supported by Ministerio de Economía y Competitividad of Spain (SAF2013-47434-R and SAF2016-77732-R), Fondo Europeo de Desarrollo Regional, Sociedad Española del Dolor, and Fundación Tatiana Pérez de Guzmán el Bueno (R.F.). Author contributions: M.T. and M.A.H. designed the study, interpreted data, and wrote the manuscript. R.d.l.F. was responsible for the study of patients. R.F., S.V., R.G., R.d.l.F., and M.C. performed the experiments in animals and contributed to data analysis and interpretation. J.L. provided statistical support and contributed to data analysis. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All the data are included in the manuscript or in the Supplementary Materials.
View Abstract

Navigate This Article