SHANK3 deficiency impairs heat hyperalgesia and TRPV1 signaling in primary sensory neurons

SUMMARY

Abnormal pain sensitivity is commonly associated with autism spectrum disorders (ASDs) and affects the life quality of ASD individual. SHANK3 deficiency was implicated in ASD and pain dysregulation. Here we report functional expression of SHANK3 in mouse dorsal root ganglion (DRG) sensory neurons and spinal cord presynaptic terminals. Homozygous and heterozygous Shank3 complete knockout (Δe4–22) results in impaired heat hyperalgesia in inflammatory and neuropathic pain. Specific deletion of Shank3 in Nav1.8-expressing sensory neurons also impairs heat hyperalgesia in homozygous and heterozygous mice. SHANK3 interacts with transient receptor potential subtype V1 (TRPV1) via Proline-rich region and regulates TRPV1 surface expression. Furthermore, capsaicin-induced spontaneous pain, inward currents in DRG neurons, and synaptic currents in spinal cord neurons are all reduced after Shank3 haploinsufficiency. Finally, partial knockdown of SHANK3 expression in human DRG neurons abrogates TRPV1 function. Our findings reveal a peripheral mechanism of SHANK3, which may underlie pain deficits in SHANK3-related ASDs.

eTOC blurb

In this issue of Neuron, Han et al., describe how SHANK3, expressed by peripheral primary sensory neurons, regulates TRPV1 function and heat hyperalgesia after inflammation and nerve injury, and therefore offer a mechanistic insight into pain dysregulation in autism.

INTRODUCTION

Individuals with autism spectrum disorders (ASDs) show distinct neurobehavioral deficits, characterized by impairments in social interactions, deficits in communication as well as repetitive behaviors (Jiang and Ehlers, 2013; Kaiser and Feng, 2015; Lord et al., 2000). ASDs are also characterized by a wide sensory abnormalities related to touch, smell, taste, and pain (Allely, 2013; Klintwall et al., 2011). Environmental factors, including pesticides and fungicides have been linked to autism and neurodegeneration (Lord et al., 2000; Pearson et al., 2016). Pain is poorly assessed in ASD patients. They appear to have higher tolerance to painful injuries such as bumps and cuts. As many as 70% of ASD patients show self-injurious behaviors, including head banging, skin picking, and self-biting (Furniss and Biswas, 2012). In sharp contrast, ASD children also display exaggerated behavioral responses to pain (Mandell et al., 2005; Nader et al., 2004; Rattaz et al., 2013). It was argued that ASD children have deficits in facial and behavioral expression of pain, which may lead caregivers to interpret this as pain insensitivity (Allely, 2013). This discrepancy results from our poor understanding of mechanisms underlying pain dysregulation in ASD. Interestingly, a recent study demonstrated that dysfunction of peripheral mechanosensory neurons contributes to tactile and behavioral deficits in mouse models of ASDs (Orefice et al., 2016).

Numerous genes have been implicated in ASDs including those that regulate synaptic transmission (Jiang and Ehlers, 2013). SHANK/ProSAP family proteins act as synaptic scaffolding proteins at postsynaptic density (PSD) of excitatory glutamatergic synapses (Sheng and Kim, 2000). Human genetic evidence indicates that SHANK family genes are strong causative genes for idiopathic ASDs. SHANK3 is one of the best implicated genes in ASDs, and SHANK3 mutations were found in ~2% of ASDs (Leblond et al., 2014; Moessner et al., 2007; Yi et al., 2016). Shank3 mutant mice exhibit dysregulation of glutamatergic synapses in their sizes, shapes, and functions, which is reminiscent of the phenotypes observed in mouse models of fragile X syndrome, Rett syndrome, and Angelman syndrome (Jiang and Ehlers, 2013; Wang et al., 2014a; Wang et al., 2011; Yoo et al., 2014). SHANK3 maps to the critical chromosome region of 22q13.3. Strikingly, clinical and genetic evaluation of 201 patients with Phelan–McDermid syndrome (Phelan et al., 2001), also known as 22q13 deletion syndrome which includes the deletion of entire SHANK3 gene, demonstrates decreased pain sensitivity in 77% of patients (Sarasua et al., 2014).

Different lines of Shank3 mutant mice, with deletions or mutations in different exons (exons 4–7, 4–9, 11, 13–16, and 21), have been reported. However, these lines of Shank3 mutant mice show variable molecular, synaptic, and behavioral phenotypes, in part due to different manipulations of Shank3 isoforms (Jiang and Ehlers, 2013). To develop a more relevant ASD model, we generated Shank3 complete deficient mice by deleting exons 4 to 22. This deletion causes the absence of the entire SHANK3 protein coding sequence (Wang et al., 2016). Shank3^Δe4–22 mouse has high construct validity because the deletion of Shank3 in mice closely mimics the SHANK3 defects found in most patients with ASDs. Notably, Shank3^Δe4–22 mice demonstrate robust behavioral phenotypes that recapitulate the major features in SHANK3-related ASD and PMS and highlight postsynaptic modulation of this gene in brain regions such as striatum (Wang et al., 2016).

In this study, we investigated how SHANK3 regulates pain sensitivity in Shank3^Δe4–22 mice. Specifically, we examined the role of SHANK3 in dorsal root ganglion (DRG) primary sensory neurons of the peripheral nervous system. We also examined SHANK3 expression in presynaptic sites in spinal cord central terminals of DRG neurons. We have revealed a peripheral and presynaptic role of SHANK3 in pain transduction and transmission via regulating transient receptor potential ion channel subtype V1 (TRPV1). Our data have also demonstrated that happoinsufficiency of SHANK3 causes profound defects in TRPV1 function in mouse and human DRG neurons.

RESULTS

Heat hyperalgesia but not baseline pain is impaired in Shank3 (Δe4–22) mutant mice

To investigate a specific role of Shank3 in pain regulation, we tested baseline pain and inflammatory and neuropathic pain in Shank3 complete knockout mice, in which exons 4 to 22 were deleted (Figure 1A). In this study, we referred Δe4–22^−/− mice as Shank3^−/− mice. We first compared the baseline thermal and mechanical sensitivity in homozygous (Shank3^−/−), heterozygous (Shank3^+/−), and wild-type (WT) mice. Hot plate test indicated that Shank3^−/− mice exhibited no change in withdrawal latency at different temperatures (50, 53, and 56 °C), compared to Shank3^+/− and WT mice (Figure 1B). Von Frey test also showed no difference among three genotypes in paw withdrawal threshold to punctate mechanical stimuli (Figure 1C). Similarly, the cold sensitivity in cold plate and temperature preference test was unaltered in Shank3^+/− and Shank3^−/− mice (Figures 1D and S1A). Furthermore, Hargreaves test showed intact response to radiant heat in Shank3^−/− mice (Figure 1E). Together, these data suggest that baseline pain sensitivity, i.e. normal pain perception and temperature sensation do not change in Shank3 global knockout mice.

Figure 1. Shank3 global mutant mice exhibit reductions in heat hyperalgesia after inflammation and nerve injury and also in capsaicin-induced pain.

(A) Schematic illustration of Shank3 complete knockout mice, in which exons 4 to 22 are deleted. Δe4–22^−/− mice = Shank3^−/− mice.

(B–D) Baseline pain sensitivity in homozygous (Shank3^−/−), heterozygous (Shank3^+/−), and wild-type (WT, Shank3^+/+) mice. n.s., no significance, one-way ANOVA. n = 5–13 mice/group.

(B) Hot plate test for heat sensitivity at 50, 53, and 56 °C.

(C) Von Frey test for punctuate mechanical sensitivity.

(D) Temperature preference test at 50/15 °C and 21/5 °C.

(E and F) Hargreaves test for basal heat sensitivity before CFA injection (E) and CCI (F) and heat hyperalgesia after the CFA injection and CCI in three genotypes. Note that heat hyperalgesia is reduced in both Shank3^−/− and Shank3^+/− mice. ^#p < 0.05, one-way ANOVA; *p < 0.05, two-way ANOVA; n = 5–7 mice / group.

(G) Time course of intraplantar capsaicin (10 µg) induced spontaneous pain in three genotypes. ^#p < 0.05, one-way ANOVA; *p < 0.05, two-way ANOVA; n = 5–9 mice / group. n.s., not significant.

(H) Capsaicin-induced neurogenic inflammation, evaluated by plasma extravasation in ipsilateral paw in Evan Blue Dye (EBD) test 30 min after intraplantar capsaicin. Contralateral paw was included as control. ^#p < 0.05, one-way ANOVA; n = 5–6 mice / group. Abs, absorbance.

All the data are mean ± S.E.M.

Next, we investigated whether chronic pain is altered in Shank3 mutant mice. Intraplantar injection of complete Freund’s adjuvant (CFA) induced persistent inflammatory pain for > 1 week, as characterized by heat hyperalgesia (Figure 1E). This heat hyperalgesia was significantly impaired in both Shank3^+/− mice and Shank3^−/− mice (F_{(2, 60)} = 34.4790, p<0.0001, Two-way ANOVA, Figure 1E). There was also significant difference between homozygous and heterozygous mice (F_{(1, 40)} =7.7231, p =0.0083, Two-way ANOVA, Figure 1E). This result indicates that loss of one copy of Shank3 is sufficient to cause pain defects, which is reminiscent of autism patients with haploinsufficiency mutations of SHANK3. However, CFA-induced edema was normal after Shank3 deficiency (Figure S1B), suggesting that inflammatory pain defect was not a result of impaired inflammation. Chronic constriction injury (CCI) also produced long-lasting heat hyperalgesia (> 4 weeks, Figure 1F). This neuropathic pain symptom was reduced in both Shank3^−/− and Shank3^+/− mice (Figure 1F). Inflammatory and neuropathic pain is also characterized by mechanical allodynia, a reduction in paw withdrawal threshold in von Frey test. Shank3^+/− and Shank3^−/− mice displayed a mild but significant reduction in mechanical allodynia after CFA (F_{(2, 72)}=6.9008, p = 0.0018, Two-way ANOVA) and CCI (F_{(2, 90)}=7.5570, p=0.0009, Two-way ANOVA) (Figures S1C and S1D). In contrast, acetone test showed that nerve injury-induced cold allodynia was unaltered in Shank3 mutant mice (Figure S1E).

Given a substantial reduction in heat hyperalgesia in Shank3 mutant mice, we postulated that heat transduction in sensory neurons could be impaired in these mice. The capsaicin receptor TRPV1 is expressed in C-fiber nociceptive neurons and upregulated in DRG neurons after inflammation and nerve injury (Ji et al., 2002; Obata et al., 2004). TRPV1 regulates heat transduction and is especially critical for the development of heat hyperalgesia (Caterina et al., 2000; Davis et al., 2000; Eskander et al., 2015). We tested capsaicin-induced pain in WT and Shank3 mutant mice. Intraplantar injection of capsaicin (10 µg) induced rapid (1–5 min) spontaneous pain (licking and flinching behavior) in WT mice, but this spontaneous pain was compromised in Shank3^−/− and Shank3^+/− mice (Figure 1G). Intraplantar capsaicin also resulted in neurogenic inflammation (paw edema), showing increased plasma extravasation in hind paw tissue in Evan Blue test (Figure 1H). The capsaicin-evoked plasma extravasation was reduced in Shank3^−/− and Shank3^+/− mice (Figure 1H). Interestingly, heterozygous and homozygous mice displayed similar defects in capsaicin responses (Figure 1G and 1H). Thus, Shank3 mutant mice have defects in both inflammation- and nerve injury-induced heat hyperalgesia due to compromised TRPV1 signaling.

To determine if there is sensory neuron loss in mutant mice, we quantified the number of Nissl-stained neurons from every fifth DRG sections of WT and Shank3^−/− mice, but did not see significant loss of DRG neurons in Shank3^−/− mice (1756 ± 191 in WT DRGs vs. 1821.7±181 in Shank3^−/− DRGs, p>0.05, Figure S2A). The percentages of TRPV1⁺, CGRP⁺, IB4⁺, and NF200⁺ neurons in Shank3^−/− mice were unaltered (Figure S2A). Skin innervation of PGP9.5⁺ nerve fibers and spinal cord innervations of CGRP⁺, IB4⁺, and NF200⁺ axonal terminals were also normal in KO mice (Figure S2, B and C). Neither did we see neuronal loss in the spinal cord (Figure S2D). Together, we conclude that the deficit of heat hyperalgesia in Shank3 mutant mice is not due to cell loss and innervation deficits of DRG neurons.

SHANK3 is expressed by primary sensory neurons and their spinal cord central terminals

We employed several different approaches to examine SHANK3 expression in DRG primary sensory neurons. RT-PCR analysis revealed the expression of all the Shank3 isoforms including Shank3a, Shank3b, Shank3c, Shank3d, and Shank3e in DRG, as well as spinal cord, cerebellum, and cortex tissues (Figure 2A). Western blotting showed SHANK3 protein expression in DRG and spinal cord of WT mice not Shank3^−/− mice, with lower expression levels compared with that of cerebellum and cortex (Figure 2B). We also examined the expression of different Shank3 isoforms in DRGs after CFA-induced inflammation and nerve injury (CCI) but did not find any upregulations of these isoforms, despite a moderate downregulation in Shank3 isoforms (Figure S1F). It is likely that pain-induced neuronal activity may cause the down-regulation, because neuronal activity has been shown to downregulate Shank3 isoforms in cortical neurons (Wang et al., 2014b).

Figure 2. Characterization of SHANK3 expression in mouse DRG and spinal cord.

(A) RT-PCR showing Shank3 isoforms in DRG, spinal cord (SC), cerebellum (CB), and cortex (CT).

(B) Western blot showing SHANK3 expression in DRG, SC, CB, and CT. We loaded 50 or 5 µg proteins for DRG/SC and CB/CT samples, respectively.

(C) IHC showing SHANK3 expression in DRG neurons of WT but not KO mice. Scale bar, 50 µm.

(D) Double staining of SHANK3/Substance P (SP), SHANK3/IB4, and SHANK3/NF200 in DRG sections. Scale bars, 50 µm.

(E) Size distribution of SHANK3⁺ neurons and all neurons in DRG sections. A total of 1896 neurons (including 1245 SAHNK3⁺ neurons) from 5 mice were analyzed.

(F) Double staining of SHANK3/IB4 in the superficial dorsal horn of control and resiniferatoxin (RTX) treated mouse. RTX (1 mg/kg body weight) was given daily for 3 days and mice were sacrificed 24 h after the last injection. Box 1 and box 2 are enlarged in the upright panels (box 1’ and box 2’) and further enlarged in the bottom panels (box 1” and box 2”). Scale bars, 100 µm in top left panels and 20 µm in other panels.

(G) Density of SHANK3-IR punctate in laminae I-IIo and IIi of the superficial dorsal horn of control and RTX treated mice. ^#p < 0.05, Student’s t-test, n = 3 mice/group. n.s., not significant.

All the data are mean ± S.E.M.

Immunohistochemistry revealed that SHANK3 immunoreactivity (IR) is widely expressed in mouse DRG neurons, but SHANK3-IR in DRGs was lost in Shank3^−/− mice (Figure 2C). Double staining showed that SHANK3-IR was highly expressed in small-sized peptidergic (Substance P⁺) and non-peptidergic (IB4⁺) neurons, as well as in large-size myelinated neurons (NF200⁺) in mouse DRG (Figure 2D). Cell size distribution analysis showed broad expression of SHANK3 in DRG neurons of different sizes, with predominant expression in small-sized neurons (<600 µm², Figure 2E). We also observed SHANK3-IR in nerve fibers in the hindpaw glabrous skin, due to axonal transport of SHANK3 from DRG cell bodies to peripheral terminals (Figure S3A). In situ hybridization also showed wide expression of Shank3 mRNA in DRG neurons of WT mice but not Shank3^−/− mice (Figure S3B). As in mouse DRG neurons, SHANK3 was widely expressed in rat DRG neurons too (Figures S3C and S3D).

Dense SHANK3-IR was also found in the spinal cord dorsal horn, especially in the superficial layers (laminae I-III) of WT mice but not in KO mice (Figures 2F and S3E). As in DRG neurons, SHANK3-IR is present in both peptidergic (SP⁺) and non-peptidergic (IB4⁺) axonal terminals in the superficial dorsal horn (Figures 2F, S3E, S3F). Interestingly, ablation of TRPV1-expressing C-fibers with systemic treatment of the neurotoxin resiniferatoxin (RTX) completely blocked the heat pain (Liu et al., 2010) and significantly reduced SHANK3-IR in the laminae I and IIo (Figures 2F and 2G). This result strongly suggests that SHANK3 in the superficial dorsal horn is originated, at least partly, from primary afferents. As expected, SHANK3-IR was also observed in many neurons in the deep dorsal horn (Figure S3E). Therefore, SHANK3 is expressed at both presynaptic sites (primary afferent terminals) and postsynaptic sites (neuronal somata) in the spinal cord dorsal horn.

TRPV1 currents in DRG neurons are substantially reduced in Shank3 mutant mice

To determine the molecular and neural mechanism by which SHANK3 regulates pain, we employed patch clamp recordings to assess TRPV1 function in dissociated small-diameter (<25 µm) DRG neurons of WT, Shank3^+/− and Shank3^−/− mice. Remarkably, inward currents induced by 100 nM capsaicin were substantially reduced in heterozygous mice and almost abolished in homozygous mice (Figures 3A and 3B). The percentage of DRG neurons responding to 100 nM capsaicin decreased from 55% (12/22) in WT mice to 30% (18/33 and 10/31) in Shank3^+/− and Shank3^−/− mice, respectively. Inward currents induced by high dose of capsaicin (1 µM) were also drastically reduced in Shank3^+/− and Shank3^−/− mice (Figure 3B). In parallel, capsaicin-induced intracellular Ca²⁺ increase was also reduced in Shank3-deficient neurons (Figures S4A and S4B). Interestingly, the TRPV1 defects in homozygous and heterozygous mice were comparable, and no statistic difference was found between these two genotypes (Figures 3A and 3B), indicating that haploinsufficiency of Shank3 is sufficient to disrupt TRPV1 function.

Figure 3. Capsaicin responses in DRG and spinal cord neurons are reduced in Shank3^+/− and Shank3^−/− mice.

(A) Traces of 100 nM capsaicin-induced inward currents in small-sized DRG neurons from WT, Shank3^+/−, and Shank3^−/− mice.

(B) Amplitude of capsaicin (100 nM and 1 µM) currents in DRG neurons from three genotypes. ^#p < 0.05 vs. WT control; n.s., not significant, one-way ANOVA, n = 10–20 neurons/group.

(C) Traces of transient sodium currents and Nav1.7-mediated sodium currents in small-sized DRG neurons of WT and KO mice. Nav1.7-mediated sodium currents were isolated by 10 nM Pro-toxin II. Right, quantification of the currents. n.s., not significant. n = 10–14 neurons/group.

(D) Traces of action potentials in small-sized DRG neurons of WT and KO mice. Right, quantification of firing frequency of action potentials. n = 12–13 neurons/group.

(E) Deficiency of TRPV1 currents in DRG neurons of Shank3^−/− mice is rescued by re-expression of WT Shank3 (Shank3-Res) but not mutant Shank3 (Shank3-INS) in Shank3-deficient neurons. Right, quantification of TRPV1 currents. ^#p < 0.05 vs. Shank3^−/− control; n.s., not significant, oneway ANOVA, n = 6–9 neurons/group.

(F) Traces of mEPSCs in lamina IIo neurons of spinal cord slices of WT and Shank3^−/− mice and the effects of capsaicin (100 nM and 1 µM).

(G) Frequency of mEPSCs in lamina IIo neurons of spinal cord slices of WT and Shank3^−/− mice. ^#p < 0.05, n.s., not significant, two-way ANOVA, n = 6 neurons/group.

(H) Spontaneous pain induced by intrathecal capsaicin (500 ng, i.t.) is reduced in Shank3^+/− and Shank3^−/− mice. ^#p < 0.05, one-way ANOVA. n = 5–10 mice/group.

All the data are mean ± S.E.M.

Human genetics strongly suggests that SCN9A, the human gene encoding sodium channel subunit Nav1.7, critically regulates pain sensitivity (Cox et al., 2006; Raouf et al., 2010). Interestingly, transient sodium currents, as well as Nav1.7-mediated sodium currents (isolated by Protoxin-II) (Sokolov et al., 2008) were normal in Shank3-deficient DRG neurons compared with WT neurons (Figure 3C). Neither did we find changes in neuronal excitability in Shank3^−/− mice: Shank3-deficient neurons and WT neurons fired action potentials at the same rate (Figure 3D). Thus, sodium channels such as Nav1.7 may not contribute to pain defects in Shank3^−/− mice.

To rescue TRPV1 deficiency in Shank3-deficient neurons, we overexpressed Shank3 these deficient mouse DRG neurons through electroporation of Shank3a construct. Notably, overexpressing Shank3-gfp in Shank3-deficient DRG neurons increased TRPV1 currents to the levels of WT neurons (Figures 3E). Thus, mouse Shank3 gene is critical for the function of TRPV1 in primary sensory neurons.

Human SHANK3 mutation with a guanine insertion in exon 21 (InsG3680) is strongly associated with ASD (Durand et al., 2007). Over-expressing this SHANK3 mutant in Shank3-deficient DRG neurons failed to completely rescue TRPV1 currents (Figure 3E), despite higher expression of this mutant than WT Shank3 in DRG cultures after the transfection (Figure S4C). This result suggests an active role of SHANK3 mutation in regulating the TRPV1 function.

Capsaicin-induced spinal cord synaptic plasticity and pain following spinal application is compromised in Shank3 mutant mice

TRPV1 in central terminals of nociceptors regulates glutamate release from presynaptic terminals (Park et al., 2011). Next, we tested basal and capsaicin-evoked excitatory synaptic transmission in spinal cord slices of WT and Shank3^−/− mice, by recording miniature excitatory postsynaptic currents (mEPSCs) in lamina IIo neurons. These interneurons are predominantly excitatory, receive input from TRPV1-expressing primary afferents, and send output to lamina I projection neurons (Park et al., 2011; Todd, 2010). We found normal mEPSCs in Shank3-deficient lamina IIo neurons: both the frequency and amplitude of mEPSC were comparable in WT and Shank3^−/− mice (Figures 3F, 3G, S4D, S4E). Capsaicin produced dose-dependent increases in mEPSC frequency but not mEPSC amplitude in WT mice; but these mEPSC frequency increases were substantially reduced after Shank3 deficiency (Figures 3F and 3G). Thus, SHANK3 is also required for TRPV1-mediated synaptic plasticity (mEPSC frequency increase) via presynaptic modulation.

Given an important role of TRPM8 in cold sensation (Bautista et al., 2007), we tested the function of TRPM8 in DRG and spinal cord neurons in WT and Shank3^−/− mice. Menthol induced comparable increases in intracellular Ca²⁺ in DRG neurons and mEPSC in spinal lamina IIo neurons of WT and Shank3^−/− mice (Figures S4, F–H), suggesting an intact TRPM8 signaling after Shank3 deficiency. TRPC4 is expressed in DRG neurons and regulated in neuropathic pain (Staaf et al., 2009). Gadolinium chloride (GdCl₃), an agonist of TRPC4, induced comparable inward currents in DRG neurons and mEPSC frequency increases in spinal cord neurons in WT and Shank3 KO mice (Figures S4, I–K), suggesting that TRPC4 function does not require SHANK3.

Spinal capsaicin injection via intrathecal (i.t.) route induces robust spontaneous pain via central TRPV1 at primary afferent terminals (Park et al., 2011). This spontaneous pain was reduced in both Shank3^+/− and Shank3^−/− mice (Figures 3H). Capsaicin also elicited phosphorylation of extracellular signal-regulated kinase (pERK) in superficial dorsal horn neurons, an essential step for central sensitization in persistent pain (Ji et al., 2003). The capsaicin-induced pERK was abolished in Shank3^−/− mice (Figure S5). Thus, Shank3 is also important for central TRPV1 signaling in the spinal cord.

Interaction of SHANK3 with TRPV1 in native DRG neurons and heterologous ND7/23 cells

Our behavioral and electrophysiological data support a functional interaction of SHANK3 and TRPV1. We further examined biochemical interactions between these two important proteins. Double staining revealed co-localization of TRPV1 with SHANK3 in mouse DRG neurons (Figure 4A). TRPV1 is also co-localized with SHANK3 in spinal axonal terminals in the superficial dorsal horn (Figure 4B). Co-IP experiment showed possible interaction of SHANK3 and TRPV1 in mouse DRG (Figure 4C). Surface biotinylation analysis revealed a reduction of TRPV1 surface expression in Shank3-deficient DRG neurons (Figure 4D). We further tested SHANK3/TRPV1 interaction in heterologous ND7/23 cells, which are hybrid cells of neuroblastoma and DRG neurons (Dunn et al., 1991). Co-IP experiment validated the SHANK3/TRPV1 interaction in ND7/23 cells-expressing SHANK3 and TRPV1 (Figures 4E and 4F). Surface biotinylation analysis also showed increased surface expression of TRPV1 in ND7/23 cells expressing Shank3 (Figure 4G). These results suggest that SHANK3 control the function TRPV1 by modulating its trafficking and surface expression.

Figure 4. Characterization of SHANK3 and TRPV1 interaction in mouse DRG neurons and ND7/23 cells.

(A) Double staining showing co-localization of SHANK3 and TRPV1 in DRG neurons: red channel and green channel are from single focal plane of a DRG section, and the merged image is the orthogonal view (XY, XZ, YZ) of z stack. Scale bar, 50 µm.

(B) Double labeling of SHANK3 and TRPV1 in axonal terminals in superficial spinal dorsal horn. Small box in the top panel is enlarged in 3 separate boxes with single and merged images. Scale bars, 50 µm (top) and 5 µm (bottom).

(C) Co-IP showing SHANK3/TRPV1 interaction in DRGs. DRG lysate was immunoprecipitated with TRPV1 antibody, and immunoblotted with SHANK3 or TRPV1 antibody as indicated. This experiment was repeated three times.

(D) Surface biotinylation showing decreased surface expression of TRPV1 in cultured DRG neurons of Shank3^−/− mice. Right, quantification of surface and lysate TRPV1 expression in WT and Shank3 KO mice. ^#p < 0.05, Student’s t test, n = 4 cultures/group.

(E,F) Co-IP analysis in ND7/23 cells. Lysate of ND7/23 cells overexpressing SHANK3-GFP and TRPV1 was immunoprecipitated with GFP antibody (E) or TRPV1 antibody (F) and then immunoblotted with anti-SHANK3, or anti-TRPV1 antibody. Each experiment was repeated three times.

(G) Surface biotinylation showing increased surface expression of TRPV1 after Shank3 overexpression in ND7/23 cells. Right, quantification of surface and lysate expression of TRPV1 in ND7/23 cells. ^#p < 0.05, Student’s t test, n = 3 cultures/group. TfR, transferrin receptor.

All the data are mean ± S.E.M.

SHANK3a consists of multiple domains for protein-protein interactions: ankyrin repeats (ANK), SH3 domain, PSD-95/Dlg/ZO-1 (PDZ) domain, Proline-rich region, and sterile alpha motif (SAM) domain (Figure 5A) (Jiang and Ehlers, 2013; Sheng and Kim, 2000). To define how SHANK3 interacts with TRPV1, we constructed 4 expression vectors: SHANK3a, PDZ^554–686, Proline-rich^891–1299, and Homer-SAM^1302–1730 and over-expressed them with TRPV1 constructor in ND7/23 cells. We found the region^891–1299 of Proline-rich domain but not PDZ^554–686 and Homer-SAM^1302–1730 domains was sufficient for the SHANK3/TRPV1 interaction (Figure 5B). Competing experiment further showed that Proline-rich^891–1299 domain and SHANK3a competed for the interaction with TRPV1 in ND7/23 cells (Figure 5C). Transfection of WT mouse DRG neurons with Proline-Rich^891–1299 region of Shank3a drastically reduced TRPV1 currents (Figure 5, D and E), suggesting a role of this SHANK3 region in regulating TRPV1 function in native DRG neurons. It is noteworthy that Human SHANK3 mutation InsG3680 resulted in a truncated SHANK3 ending at the site 1227 of the Proline-rich^891–1299 region (Durand et al., 2007), and this mutation also affected TRPV1 function in DRG neurons (Figure 3E).

Figure 5. Prolin-rich domain is essential for the SHANK3 and TRPV1 interaction in heterologous cells and DRG neurons.

(A) Diagram of full length SHANK3a and SHANK3a mutant proteins.

(B,C) Co-IP showing the interaction between TRPV1 and the Proline-rich (Pro) domain of SHANK3 in ND7/23 cells. Each experiment was repeated three times.

(B) Interaction of TRPV1 with SHANK3 mutants. SHANK3 or its mutants were transiently co-expressed together with TRPV1, and cell lysate was immunoprecipitated with GFP antibody and then immunoblotted with GFP or TRPV1 antibody as indicated.

(C) Proline-rich region competes the interaction between SHANK3 and TRPV1. SHANK3 mutants were co-transfected with full length SHANK3 and TRPV1, and cell lysate was immunoprecipitated with TRPV1 antibody, and then immunoblotted with GFP, TRPV1, or SHANK3 antibody.

(D,E) Competition experiment showing that transfection of WT mouse DRG neurons with Proline-Rich^891–1299 region of SHANK3a profoundly inhibits TRPV1 currents.

(D) Traces of TRPV1 currents in DRG neurons.

(E) Amplitudes of inward currents. ^#p < 0.01 vs. WT control, unpaired t-test, n = 8–10 neurons/group.

All the data are mean ± S.E.M.

Inflammatory pain, heat pain, and TRPV1 signaling are impaired after sensory neuron-specific Shank3 deletion

To confirm a specific role of peripheral SHANK3 in pain regulation, we generated Shank3 conditional KO (CKO) mice by crossing Shank3-floxed mice with Na_V1.8-Cre mice (Figure 6A), leading to specific deletion of Shank3 in nociceptive neurons as well as some low-threshold A-fiber neurons (Agarwal et al., 2004; Shields et al., 2012). We then compared baseline pain and inflammatory pain in three genotypes: littermate control mice (Shank3^f/f), heterozygous mice (Na_V1.8-cre;Shank3^f/+), and homozygous mice (Na_V1.8-cre;Shank3^f/f) (Figure 6B). We found a profound reduction (80%) of SHANK3-IR in the superficial dorsal horn (laminae I-III) in Na_V1.8-cre;Shank3^f/f mice (Figure 6, C and D), further supporting that SHANK3-IR in the superficial dorsal horn is largely originated from primary sensory neurons. Conditional heterozygous mice also showed a 60% reduction in SHANK3-IR in laminae I-III (Figure 6D). Remarkably, baseline heat sensitivity in hot plate and Hargreaves tests was impaired in homozygous mice showing increased withdrawal latency (Figure 6, E–G). However, temperature preference test showed no difference among three genotypes in mild test temperatures (21/21 °C and 15/25°C, Figure S6A). We also failed to see changes in body temperature (rectal temperature) in Shank3 conditional mutant mice and Trpv1 null mice (Figure S6B). Deficit in baseline heat sensitivity did not occur in heterozygous mutant mice (Figure 6, E and F). CFA-induced heat hyperalgesia was reduced in both homozygous and heterozygous CKO mice (Figure 6G). Von Frey test showed comparable baseline mechanical sensitivity in three genotypes, but CFA-induced mechanical allodynia was slightly decreased in homozygous and heterozygous mice (Figure S6C). Capsaicin-induced spontaneous pain was substantially reduced in homozygous and heterozygous CKO mice following intraplantar or intrathecal injection (Figure 6, H and I). It is noteworthy that SHANK3-IR in deep dorsal horn neurons (laminae IV-V) was unaltered (Figure S6D). Collectively, these data from Shank3 CKO mice strongly suggest that deletion of Shank3 in sensory neurons can recapitulate all pain defects in Shank3 global KO mice. Moreover, Shank3 homozygous but not heterozygous CKO mice showed heat deficit, a reminiscent of Trpv1 KO mice (Caterina et al., 2000).

Figure 6. Shank3 in sensory neurons is required for heat transduction, CFA-induced heat hyperalgesia, and capsaicin-evoked pain.

(A) Schematic illustration of Shank3 conditional knockout mice with Shank3 deletion in Nav1.8 -expressing sensory neurons.

(B) Genotyping of Shank3 conditional knockout mice for Shank3^f/f and Nav1.8-Cre genotypes.

(C) Double staining of SHANK3 and IB4 on spinal cord dorsal horn from Shank3^f/f and Nav1.8-Cre;Shank3^f/f mice. Scale bar, 100 µm.

(D) Quantification of SHANK3-IR in laminae I-III and IV-V of spinal cord from Shank3^f/f, Nav1.8-Cre;Shank3^f/+, and Nav1.8-Cre;Shank3^f/f mice. ^#p < 0.05, one-way ANOVA, n =5 mice/group.

(E) Hot plate test for heat sensitivity in three genotypes, ^#p < 0.05, one-way ANOVA; *p<0.05, two-way ANOVA, n = 5–10 mice/group. n.s., not significant.

(F) Temperature preference test at 50/15 °C and 21/5 °C in three genotypes, ^#p < 0.05, one-way ANOVA, n = 6 mice/group.

(G) CFA-induced heat hyperalgesia in three genotypes. ^#p < 0.05, one-way ANOVA; *p<0.05, two-way ANOVA; n = 5–10 mice/group. n.s., not significant.

(H, I) Spontaneous pain induced by intraplantar capsaicin (10 µg, i.pl., H) or intrathecal capsaicin (500 ng, i.t., I) in three genotypes. ^#p < 0.05, one-way ANOVA; *p<0.05, two-way ANOVA , n = 5–10 mice/group. n.s., not significant.

All the data are expressed as mean ± S.E.M.

SHANK3 expression in human DRG neurons regulates TRPV1 signaling

To evaluate the translational potential of this study, we further tested if SHANK3 would regulate TRPV1 function in human DRG neurons obtained from non-diseased donors. Immunohistochemistry and size frequency analysis showed very broad expression of SHANK3 in all types of human DRG neurons (Figure 7, A and B). We also observed high-level of co-localization of SHANK3 and TRPV1 in human DRG neurons (Figure 7C). Co-localization of SHANK3/TRPV1 was further observed in axons in human spinal nerve (Figure 7C), suggesting active axonal transport of both proteins. To define a critical role of SHANK3 in regulating TRPV1 function, we developed an effective method to knockdown SHANK3 expression in dissociated human DRG neurons using small-interfering RNA (siRNA). Figure 7D shows that siGLO-labeled siRNA could be up-taken by all human DRG neurons in primary cultures. Treatment of human DRG cultures with selective SHANK3 siRNA (100 nM) lead to a partial reduction in SHANK3-IR (Figure 7E), which can mimic haploinsufficiency of SHANK3 in human. Patch clamp recordings in siGLO-labeled small-diameter human DRG neurons (<50 µm) showed that capsaicin-induced inward currents were substantially suppressed (85 and 79% inhibition at 0.1 and 1 mM capsaicin, respectively) by SHANK3 siRNA, despite a partial (50%) knockdown of SHANK3 expression (Figure 7, E–G). By contrast, the SHANK3 siRNA treatment did not change TRPV1 mRNA levels in human DRG cultures (Figure S7), arguing against the possibility of regulating TRPV1 expression by SHANK3. Competing experiment by over-expression of the Proline-rich^891–1299 region further suggested an active role of this SHANK3 region in TRPV1 function regulation in human primary sensory neurons (Figure 7, H and I). Therefore, SHANK3 haploinsufficiency also causes a profound impairment of TRPV1 function in human DRG neurons.

Figure 7. SHANK3 regulates TRPV1 function in human DRG neurons.

(A) SHANK3-IR in human DRG neurons. Upright, enlarged SHANK3 immunostaining in the box of left panel. Low right, Nissl staining in the same box. Scale bars, 500 µm (left) and 200 µm (right).

(B) Size distribution of SHANK3-IR neurons and all neurons in human DRG sections. A total of 2199 neurons from 3 human DRGs were analyzed.

(C) Double labeling of SHANK3 with TRPV1 in human DRG and spinal nerve. Scale bars, 200 µm.

(D) Immunocytochemistry showing knockdown of SHANK3 expression by SHANK3 siRNA treatment (100 nM, 48 h) in cultured human DRG neurons. Scale bar, 200 µm.

(E) Intensity of SHANK3-IR neurons after SHANK3 siRNA and control (con) siRNA treatment. ^#p < 0.05, Student’s t-test, n = 25–31 neurons/group.

(F and G) Inhibition of TRPV1 currents in human DRG neurons following SHANK3 siRNA treatment (100 nM, 48 h).

(F) Typical traces of capsaicin-induced inward currents in cultured human DRG neuron.

(G) Quantification of capsaicin-induced inward currents in human DRG neurons. ^#p < 0.05 vs. control siRNA, unpaired t-test, n = 10–11 neurons/group.

(H and I) Proline-rich^891–1299 domain of SHANK3a is required for TRPV1 function in human DRG neurons.

(H) Typical traces of capsaicin-induced inward current in cultured human DRG neuron.

(I) Quantification of capsaicin-induced inward currents in human DRG neurons. ^#p < 0.05 vs. control, unpaired t-test, n = 6–7 neurons/group.

All the data are expressed as mean ± S.E.M.

DISCUSSION

Using Shank3 complete knockout mice with deletion of exons 4 to 22 (Δe4–22 Shank3^−/−), we have made several interesting findings in this study. First, SHANK3 is broadly expressed in cell bodies and central terminals of DRG primary sensory neurons. Second, same as in human SHANK3-related disorders, Shank3 haploinsufficiency in mice causes marked deficits in inflammatory pain and neuropathic pain with predominant reductions in heat hyperalgesia. Third, Shank3 CKO mice with Shank3 deficiency in Nav1.8-expressing sensory neurons can recapitulate all the pain defects in global knockout mice and display additional deficits in baseline heat sensitivity. Fourth, SHANK3 regulates TRPV1 function in both primary sensory neurons and presynaptic terminals and is also essential for capsaicin-evoked pain at both peripheral and central sites. Fifth, SHANK3 interacts with TRPV1 via the Proline-rich region of SHANK3 and regulates the trafficking and surface expression of TRPV1. Finally, SHANK3 is expressed in human DRG neurons and SHANK3 knockdown dramatically disrupts TRPV1 currents in human sensory neurons.

Peripheral modulation of pain and heat sensation by SHANK3 via interaction with TRPV1

To the best of our knowledge, this is the first study to demonstrate a peripheral role of SHANK3. We found broad SHANK3 expression in primary sensory neurons of both mouse and human DRGs. All the major Shank3 isoforms, including Shank3a, Shank3b, Shank3c, Shank3d, and Shank3e and full length SHANK3 protein expressed in mouse DRGs. Importantly, specific deletion of Shank3 in Nav1.8-expressing sensory neurons recapitulated all the pain deficits in Shank3 global KO mice, including severe defect in heat hyperalgesia and mild defect in mechanical allodynia. Thus, peripheral SHANK3 plays an essential role in driving heightened pain sensitivity after inflammation and nerve injury. It is noteworthy that Shank3 deletion does not cause neuronal loss in DRG and spinal cord. Neither does this deletion affect peripheral and central innervations of DRG neurons. Thus, pain defects in Shank3 mutant mice are not caused by developmental defects in sensory neurons and their projections.

TRPV1 is one of the most investigated transduction molecules in pain research (Caterina and Julius, 2001; Patapoutian et al., 2009). Little is known about scaffold proteins for TRPV1, although A-kinase anchoring protein (AKAP) regulates PKA phosphorylation of TRPV1 and hyperalgesia (Jeske et al., 2008). Our findings demonstrated that SHANK3 is a scaffold protein for TRPV1 and critically regulates TRPV1 function in mouse and human primary sensory neurons. First, SHANK3 is co-localized with TRPV1 in sensory neurons and nerve axons of mouse and human DRGs. Second, capsaicin-induced inward currents are dramatically reduced in mouse DRG neurons after Shank3 haploinsufficiency. Strikingly, siRNA-mediated partial knockdown of SHANK3 resulted in a substantial reduction in TRPV1 currents in human DRG neurons. Third, Shank3 re-expression in Shank3-deficint DRG neurons rescued TRPV1 currents. Fourth, capsaicin-induced spontaneous pain and inflammation- and nerve injury-induced heat hyperalgesia are markedly reduced in Shank3 global and conditional mutant mice. Fifth, capsaicin-induced neurogenic inflammation but not CFA-induced edema is abrogated in Shank3 mutant mice. Co-IP experiment further demonstrated possible physical interaction of SHANK3 and TRPV1 in both native DRG neurons and heterologous ND7/23 cells. As a scaffold protein, SHANK3 appears to control TRPV1 function in sensory neurons via regulating the surface expression of TRPV1. SHANK3 is known to interact with NMDAR and mGluR5 via PDZ and Homer binding motif, respectively (Sheng and Kim, 2000; Tu et al., 1999). Of great interest SHANK3 interacts with TRPV1 via the Proline-rich region of SHANK3. This region is essential for the SHANK3 binding to TRPV1 and also important for regulating TRPV1 function in mouse and human DRG neurons. Consistently, an ASD-associated human mutation (InsG3680), which generates a truncated protein ending of SHANK3 in this region (Durand et al., 2007), also disrupted the TRPV1 function in DRG neurons. Future studies are warranted to determine the specific sequences on both proteins that are responsible for the SHANK3/TRPV1 interaction.

Our results also show that neuronal excitability did not change after Shank3 deficiency in DRG neurons, suggesting that SHANK3 specifically regulates the transduction but not the conduction of pain in primary sensory neurons. In addition to capsaicin-induced pain and injury-induced heat hyperalgesia, basal heat sensitivity in all the tests was impaired in homozygous CKO mice but not global knockout mice. This discrepancy suggests that peripheral and central SHANK3 may regulate heat sensitivity in a different even opposite way. A deficit in peripheral heat transduction could be masked by enhanced central processing of heat signaling in global KO mice. It is conceivable that loss of TRPV1 function in Shank3-deficient sensory neurons is responsible for this altered heat sensitivity in CKO mice. SHANK3 may also regulate additional thermal sensors or signaling molecules in the PNS and CNS. Notably, TRPV1 primarily regulates heat hyperalgesia after tissue injury but only has a partial role in heat-induced nociceptive responses (Caterina et al., 2000; Davis et al., 2000). Apart from TRPV1, several thermal sensors have been implicated in extreme heat (TRPV2) and painful heat such as TRPM3 and calcium-activated chloride channel anoctamin 1 (ANO1) (Caterina et al., 1999; Cho et al., 2012; Vriens et al., 2011). By contrast, cold sensitivity and TRPM8 function do not change in Shank3 KO and CKO mice.

Presynaptic modulation of pain by SHANK3

Almost all SHANK3-related studies focus on postsynaptic mechanisms in the brain. Strikingly, we saw prominent SHANK3 expression in central terminals in the superficial dorsal horn. These SHANK3-expressing axonal terminals are presynaptic components of the first-order nociceptive synapses that co-express TRPV1 (Todd, 2010). The presynaptic expression of SHANK3 is also supported by the evidence that SHANK3-IR is dramatically reduced in the superficial dorsal horn after C-fiber ablation and Shank3 deletion in primary sensory neurons. As a functional measurement of presynaptic modulation, TRPV1-medicated mEPSC frequency increase in spinal lamina IIo neurons is substantially reduced in Shank3 mutant mice. Spinal TRPV1 was also implicated in the generation of mechanical allodynia (Eskander et al., 2015; Kim et al., 2012). Consistently, mechanical allodynia after CFA and CCI was partially reduced in Shank3 mutant mice. However, defect in heat hyperalgesia is the primary phenotype in these mice. Heat hyperalgesia can be regulated by both peripheral and central TRPV1. Inflammation not only induces heat hyperalgesia but also increases TRPV1 expression in DRG neurons and axonal transport of TRPV1 to skin nerve terminals (Ji et al., 2002). SHANK3 is also present in axonal terminals and presynaptic components of hippocampal neurons (Halbedl et al., 2016), but the function of this presynaptic SHANK3 was not investigated in this study. Remarkably, deletion of the autism gene Mecp2 in DRG mechanosensory neurons results in loss of prespresynaptic inhibition in the spinal cord, tactile hypersensitivity, and behavioral phenotypes of ASD (Orefice et al., 2016). It remains to be investigated if spinal inhibitory synaptic transmission is also altered after SHANK3 deficiency.

Concluding remarks

Abnormal sensory integration is commonly associated with ASD patients and significantly affect the life quality of the individuals. Pain in ASD patients is difficult to assess in part due to their deficits in communication. It is generally believed that ASDs are a group of developmental brain disorders due to dysregulation of excitatory and inhibitory synaptic transmission in the cortex (Jiang and Ehlers, 2013; Tabuchi et al., 2007). However, peripheral and presynaptic mechanisms of ASDs have begun to be revealed (Orefice et al., 2016). It is conceivable that different autism genes in sensory neurons may regulate different sensory modalities, such as pain and touch by Shank3 and Mecp2, respectively. Remarkably, SHANK3 deletion in PMS results in high pain threshold in 77% patients (P=0.0256), and high pain tolerance is becoming more common with increased age (89%, P=0.00037). PMS patients also exhibit significant self-injury behaviors such as biting (46%, P=0.0006) and hair pulling (41%, P=0.0163) but insignificant touch hypersensitivity (46%, P=0.699) (Sarasua et al., 2014).

Using the newly generated Shank3 complete knockout mice with deletion of exons 4 to 22 (Wang et al., 2016), we have demonstrated that like human SHANK3 deficiency, haploinsufficiency of Shank3 in mice leads to substantial impairment in TRPV1 function and heat hyperalgesia, and these deficits can be reproduced by Shank3 deletion in mouse and human sensory neurons. Our data strongly suggest a peripheral mechanism of ASDs by which SHANK3 expression in primary sensory DRG neurons controls the surface expression and function of TRPV1 to regulate heat pain transduction and heat hyperalgesia associated inflammatory and neuropathic pain. Our results also revealed a presynaptic mechanism of SHANK3 by which SHANK3 expression in superficial spinal presynaptic terminals controls TRPV1-mediated neurotransmitter release in spinal cord pain circuit under injury conditions. Thus, our findings have offered a mechanistic insight into pain deficits in ASDs. On the other hand, targeting peripheral SHANK3 may also help alleviate chronic pain in non-ASD patients.

Given an important role of SHANK3 in postsynaptic modulation of glutamatergic synaptic transmission and well-demonstrated role of spinal cord postsynaptic plasticity in driving central sensitization and chronic pain (Ji et al., 2003), SHANK3 could also contribute to enhanced pain states after inflammation and nerve injury via modulating postsynaptic plasticity in spinal cord dorsal horn neurons. Therefore, SHANK3 modulates inflammatory and neuropathic pain via peripheral, presynaptic, and postsynaptic mechanisms. However, defects in heat transduction and heat hyperalgesia in Shank3 mutant mice could be primarily mediated by peripheral SHANK3 via its interaction with TRPV1.

EXPERIMENTAL PROCEDURES

Reagents

We purchased capsaicin and complete Freund’s adjuvant (CFA), menthol, and gadolinium chloride from Sigma-Aldrich, siRNA from Origene, resiniferatoxin from LC Laboratories, and ProTx-II from Peptide institute, Inc (Osaka, Japan).

Animals

Shank3 complete knockout mice (exons 4 to 22 deletion, Δe4–22 ^−/−, C57BL/6 background) and Shank3^flox/fllox mice (RRID: MGI:5800311) were generated at Duke University Medical Center (Wang et al., 2016). Nav1.8-cre; Shank3^flox/flox mice were generated by mating Nav1.8-cre mice with Shank3^flox/flox mice (RRID: MGI: 5800310). Adult male C57BL/6 mice (8–12 weeks) were used for behavioral and biochemical studies. Young C57BL/6 mice (5–8 weeks) were used for electrophysiological studies in spinal cord slices. All the animal procedures were approved by the Institutional Animal Care & Use Committee (IACUC) of Duke University. See Supplemental Experimental Procedures for details.

Animal models of pain and intrathecal injection

To produce inflammatory pain, diluted CFA (20 µl) was injected into the plantar surface of a hindpaw. To produce chronic constriction injury (CCI) in mice, 3 loose silk ligatures were tied around the sciatic nerve. Capsaicin was injected via intraplantar (3 or 10 µg) or intrathecal route (0.5 µg) to induce spontaneous pain. For intrathecal injection, spinal cord puncture was made with a 30G needle. See Supplemental Experimental Procedures for details.

Behavioral testing

Capsaicin-induced spontaneous pain was assessed by counting the time (seconds) spent on flicking, biting, and flinching for the first 5 min after the injection. Mechanical sensitivity was tested with von Frey hairs and thermal sensitivity was tested using Hargreaves apparatus. Two temperature choice test was used to assess temperature preference. Cold allodynia was assessed by acetone test. See Supplemental Experimental Procedures for details.

Whole-cell patch clamp recordings in dissociated mouse DRG neurons

DRGs were aseptically removed from mice and digested with collagenase and dispase-II for 120 min. Cells were placed on glass cover slips coated with poly-D-lysine and grown in a neurobasal defined medium for 24 h before experiments. Whole-cell voltage- and current-clamp recordings were performed at 28°C to measure capsaicin-induced currents, transient sodium currents and action potentials, respectively. See Supplemental Experimental Procedures for details.

Whole-cell patch clamp recordings in dissociated human DRG neurons

Non-diseased human DRGs were obtained from National Disease Research Interchange (NDRI). Postmortem L3–L5 DRGs were dissected from six donors and delivered in ice-cold culture medium. Human DRG cultures were prepared as previously reported (Xu et al., 2015). DRGs were digested for 120 min with collagenase Type II and dispase II. DRG neurons were plated on poly-D-lysine–coated glass coverslips and grown in Neurobasal medium. Whole-cell patch-clamp recordings in small (<50 µm) human DRG neurons were conducted at room temperature. See Supplemental Experimental Procedures for details.

Spinal cord slice preparation and patch clamp recordings

The L3-L5 lumbar spinal cord segment was removed from mice under urethane anesthesia (1.5 −2.0 g/kg, i.p.) and kept in pre-oxygenated ice-cold artificial cerebrospinal fluid (aCSF) solution. Transverse slices (300–400 mm) were cut on a vibrating microslicer. The whole cell patch-clamp recordings were made from lamina IIo neurons in voltage clamp mode to record spontaneous EPSCs (sEPSCs) and miniature EPSCs (mEPSCs). See Supplemental Experimental Procedures for details.

Immunohistochemistry in rodent and human tissues

Animals were perfused through the ascending aorta with 4% paraformaldehyde. Spinal cord, skin DRG, and nerve sections were cut in a cryostat. The sections were blocked with 2% goat or horse serum and then incubated overnight at 4°C with primary antibodies followed by Cy3- or FITC-conjugated secondary antibodies. Human DRGs were fixed in 4% paraformaldehyde for 24 h and processed for immunofluorescence. See Supplemental Experimental Procedures for details.

Cell culture and transfection

ND7/23 cells were cultured in high glucose DMEM supplemented with 10% fetal bovine serum. Transfection was performed with LipofectamineTM 2000 Reagent. For electroporation, freshly disassociated mouse or human DRG neurons were transfected with 4D–Nucleofector™ System (Lonza). For siRNA transfection, 100 pmol siRNA was mixed with 5 pmol siGLO control and 1.5 ml TransIT®-2020 Transfection Reagent (Mirus). See Supplemental Experimental Procedures for details.

Immunoprecipitation

Transfected ND7/23 cells or cultured primary DRG neurons were lysed in ice-cold immunoprecipitation buffer. The lysate was immunoprecipitated with 0.5 mg of a rabbit anti-GFP antibody or goat anti-TRPV1 antibody and then incubated with protein G-Sepharose beads (Roche). See Supplemental Experimental Procedures for details.

Surface protein biotinylation

Plasma membrane protein expression was detected after protein biotinylation. Solubilized lysates were pulled down on Streptavidin Agarose Resin overnight on rotor at 4°C. The pellet was then incubated in SDS-PAGE loading buffer and sample buffer for Western blot analysis. See Supplemental Experimental Procedures for details.

Western blot

The samples were separated on an SDS-PAGE gel, transferred, and probed with primary antibodies against SHANK3 (rabbit, 1:5000, provided Paul Worley from Johns Hopkins Medical School). See Supplemental Experimental Procedures for details.

RT-PCR

Tissues were rapidly isolated in RNase free conditions. Total RNAs were extracted using RNeasy Plus Mini kit (Qiagen). RNAs (0.5–1 µg) were reverse-transcribed using the SuperScript III reverse transcriptase (Invitrogen). See Supplemental Experimental Procedures for details.

Statistical analyses

All data were expressed as mean ± s.e.m. Biochemical, electrophysiological, and behavioral data were analyzed using Student’s t-test (two groups) and one-way or two-way ANOVA followed by post-hoc Bonferroni test. The criterion for statistical significance was p <0.05. See Supplemental Experimental Procedures for details.

Highlights.

• SHANK3 is expressed in primary sensory neurons and spinal cord central terminals

• Shank3 haploinsufficiency in sensory neurons leads to impaired heat hyperalgesia

• SHANK3 in mouse and human DRG neurons regulates TRPV1 function

• SHANK3 interacts with TRPV1 and modulates presynaptic pain transmission