AAV.CMV.HI. eGFP-Cre.WPRE.SV40 was from Addgene (Plasmid #105545). Shank3 siRNA (Cat# AM16708, ID:194441), Trpv1 siRNA (# AM16708, ID: 223137), and Trpm2 siRNA (# 1320001, ID: mss219854) were purchased from Thermo scientific. Control siRNA was purchased from Santa cruz biotech (#sc-37007), and RVG peptide (# AS-62566) was from Anaspec (Fremont, CA). Menthol (# W266507), capsaicin (# M2028), and LPS (# L2630) were purchased from Sigma.

Shank3 knockout mice (exons 4 to 22 deletion, Δ4–22, C57BL/6 background RRID: MGI:5800311) (8), Shank3 flox/flox mice (RRID: MGI: 5800310) (26), and Shank2 flox/flox mice were generated at Yong-Hui Jiang’s laboratory at Duke University Medical Center. Both Shank2 and Shank3 floxed mice were backcrossed to C57BL/6J for more than 10 generations. To generate sensory neuron-specific deletion of Shank3 and Shank2, the flox mice were crossed with Nav1.8-cre mice, provided by Dr. Rohini Kuner of University of Heidelberg (27). Adult mice (males and females, 8–10 weeks) were used for behavioral and biochemical studies. CD1 mice (males and females, 8–10 weeks, Charles River Laboratories) were used for some siRNA knockdown experiments. All the mouse procedures were approved by the Institutional Animal Care & Use Committee of Duke University. Two to five mice were housed in each cage under a 12-hour alternating light-dark cycle with ad libitum access to water and food. Sample sizes were estimated based on our previous studies for similar types of behavioral and biochemical analyses (24, 28).

The siRNA preparation and injection were conducted as we previously described (29). For siRNA delivery to the vagus nerve, we mixed 100 nM siRNA with 1 µM RVG peptide that can facilitate siRNA uptake by neurons (30). For AAV9 virus delivery to vagus nerve, we mixed 1 x 10⁹ virus in 10 µl of PBS. The vagal nerve was exposed as previously described (31). Mice were anesthetized with isoflurane and a midline cervical muscle and salivary glands were exposed by blunt dissection until the trachea was visible. The vagus nerve was gently separated from the artery, and bilateral peri-neural injections were conducted to deliver siRNA or virus (5 μl) using a glass electrode with tip size of 1 µm. The knockdown effects of siRNA and AAV9 virus were validated by examining the expression of Shank3, Trpm2, and Trpv1 in vagal sensory neurons in nodose ganglion (NG) using in situ hybridization.

The 5% menthol and 100 µM capsaicin were painted into left ear auricular acupoint or skin (50 µl) to measure the body temperature change. Mice were lightly anesthetized with 2.5% isoflurane and then painted by fine brush.

The systemic inflammation was generated by intraperitoneal injection of 1 mg/kg LPS (from Escherichia coli O111:B4) prepared in PBS (100 mg/ml stock). Body temperature measured before LPS injection, then at 0.5, 1, 3, 6, 24, 48, 72, and 96 hours after the LPS injection. The environmental temperature was maintained at 23 ± 0.5°C. To obtain a rectal temperature, the mouse is hand-restrained (briefly anesthetized) and placed on a horizontal surface (cage lid) with the tail lifted, and a probe (covered with Vaseline) was gently inserted into the rectum to a fixed depth of 2 cm. The rectal temperature was measured by the TA-29 flexible implantable probe and TC-344C temperature measuring unit (Warner Instruments., Hamden, CT, USA).

For ECG recording, the silver wires were implanted in the left abdomen below the heart and the right upper shoulder. Two electrodes were fixed to the incised tissues in mice using a 26 G needle under isoflurane anesthesia (32). The silver wires were exposed outside the neck region and put on the small jacket for the prevention of biting. The silver wires were exposed outside the neck region for the prevention of biting. After exposure to the menthol into ear acupoint region (concha area in outer ear), the ECG activity was measured in awake mice. The shaking induced electrical noise was excluded during data analysis. The analog electrical signal was amplified by a microelectrode ac amplifier (AM systems, model 1800) and converted by Digidata 1440A (Molecular device). Signals were filtered at 2 kHz and digitized at 5 kHz. Data were stored and analyzed with a personal computer using pCLAMP 10 (Molecular Devices) software.

Mice were briefly anesthetized with 4% isoflurane and transcardially with PBS and 4% paraformaldehyde. Following perfusion, nodose ganglia (NG) were collected and postfixed in the same fixative overnight at 4°C. NG tissues were embedded in OCT medium (Tissue-Tek), sectioned (14 μm) in a cryostat, and thaw-mounted onto Superfrost Plus slides (VWR). In situ hybridization was performed using the RNAscope system (Advanced Cell Diagnostics) following the manufacturer’s instructions. The RNA scope probes for mouse Shank3 (# 417371), Trpm2 (# 316831-C3), and Trpv1 (# 313331-C3) were designed by Advanced Cell Diagnostics and the RNAscope multiplex fluorescent assays were performed according to the manufacturer’s instructions. Pre-hybridization and hybridization, and then washing were performed according to standard methods. All images were acquired with the same settings using a confocal microscope under 20x magnification. For quantification, four non-adjacent NG sections from each animal and four animals per group were included. The intensity score (0: no signal, 1: weak signal; 2: medium signal, and 3: strong signal) was used to semi-quantify the mRNA signals blindly. We also quantified the number of mRNA-labeled punctate dots (puncta) in each neuron using Image J.

Total RNAs from nodose ganglia were extracted using the Direct-zol™ RNA MiniPrep Kit (Zymo Research), and 0.5-1 μg of RNA was reverse-transcribed using the iScript cDNA Synthesis^® (Bio-Rad). Specific primers, including Gapdh control, were designed using IDT SciTools Real-Time PCR software. We performed gene-specific mRNA analyses using the CFX96 Real-Time PCR system (BioRad). Quantitative PCR amplification reactions contained the same amount of Reverse transcription (RT) product, including 10 μl of 2× iQSYBR-green mix (BioRad) and 100-300 nM of forward and reverse primers in a final volume of 15 μl. Primer efficiency was obtained from the standard curve and integrated for calculation of the relative gene expression, which was based on real-time PCR threshold values of different transcripts and groups.

Trpm2 forward: TGCCTCACCTGCTCTTTGCC

Trpm2 reverse: TCTGTGTGTTCCTGCACCTA

TRPV1 forward: ATGTTCGTCTACCTCGTGTTCTTG

TRPV1 reverse: AGGCAGTGAGTTATTCTTCCCATCC

TRPM8 forward GTTGGACCTTGCCAGTGATGAG

TRPM8 reverse CCATTCTCCAGAAAGAGGCGGA

SCN10A forward ATGGAGGTCAGCCAGGACTACA

SCN10A reverse CTGTGAGGTTGTCCGCACTGAA

Gapdh forward: AGGTCGGTGTGAACGGATTTG

Gapdh reverse: GGGGTCGTTGATGGCAACA

To detect serum cytokine levels, we purchased mouse ELISA kits from R&D system (Minneapolis, MN, Cat# DY406 for IL-6). We collected whole blood from a facial vein (30 μl~60 μl). The collected blood was allowed to clot for 30 minutes at room temperature. The clot was then removed in a refrigerated centrifuge at 2,000 x g for 10 min to collect the supernatant (serum). To perform each ELISA assay, we used 5 μl of serum. We measured the absorbance of OD450 using a microplate reader (Bio-Rad) and then quantified the data by MPM6 software (BioRad).

All the data in the figures were expressed as mean ± SEM. We analyzed the biochemical or survival data using Prism GraphPad 6.0. One-Way ANOVA with Bonferroni post hoc or Two-Way ANOVA with Tukey’s posthoc test or unpaired t-test were used to assess biochemical, temperature, and ECG data. We analyzed the survival ratio using a Kaplan–Meier method and log-rank (Mantel-Cox) test. If the overall log-rank test was statistically significant (p < 0.05), we further performed pairwise comparisons of two groups. p < 0.05 was considered statistically significant.

To investigate an overall role of Shank3 in LPS-induced hypothermia, inflammation, and mortality, we employed Shank3 △e4-22 mutant mice, in which exons 4 to 22 were deleted and SHANK3 protein is completely lost (8). In the current study, we referred to △e4-22 mutant mice as Shank3 global knockout mice (GKO). First, we compared the baseline body temperature in wild-type (WT) and heterozygous (Shank3^+/- ) and homozygous (Shank3^-/- ) GKO mice. The rectal temperature analysis revealed that Shank3^-/- mice exhibited no change in body core temperature as compared to WT and Shank3^+/- mice ( Figure 1A ), suggesting that SHANK3 is not involved in the regulation of basal body temperature. Then, we challenged these mice with intraperitoneal (i.p.) injection of LPS (1 mg/kg) to induce sepsis (28). Herein, we found a rapid and transient hyperthermia in WT mice, as indicated by an elevation in rectal temperature at 30 minutes after LPS injection as compared to baseline (p = 0.0158, Figure 1A ). Interestingly, this hyperthermia was completely compromised in both Shank3^+/- and Shank3^-/- mice (p = 0.0106 and p = 0.0004, respectively) when compared with WT mice ( Figure 1A ). We also revealed that LPS-induced hypothermia started at 3 hours (p = 0.0008), peaked at 6 hours (p < 0.0001), and continued for 24 hours (p=0.0488) after LPS intervention in WT mice. Furthermore, as compared to WT mice, this hypothermia was observed within 1 hour (p = 0.0411) and lasted for at least 48 hours (p < 0.0001) after LPS exposure in Shank3^-/- mice. More strikingly, sepsis-induced hypothermia was significantly aggravated in both Shank3^+/- and Shank3^-/- mice (F (2, 408) = 48.24, p < 0.0001, two-way ANOVA, Figure 1A ). There was also significant difference between heterozygous and homozygous mice (F (1, 245) = 11.35, p = 0.0011, two-way ANOVA, Figure 1A ). This result suggested that loss of one copy of Shank3 is sufficient to cause hyperthermia defects and hypothermia exacerbation, which is reminiscent of autism patients with haploinsufficiency of SHANK3.

Next, we investigated whether sepsis-induced systemic inflammation was altered in Shank3 mutant mice. Pro-inflammatory cytokine IL-6 is well recognized as a biomarker for sepsis and sepsis-associated death (28). We detected robust increases of serum IL-6 levels at 6 hours after LPS injection in Shank3^-/- mice as compared to WT mice (F (1, 24) = 0.1450, p = 0.008226, two-way ANOVA, Figure 1B ).

Also, we evaluated survival rate following LPS injection among three genotypes. Low-dose LPS at 1 mg/kg resulted in a very low mortality (10%, Figure 1C ) in WT mice. As compared to WT mice, we found a slight increase (23.33%) in mortality in LPS-treated Shank3^+/- mice (p = 0.1062, Figure 1C ) but a significant increase (37.37%) in mortality in LPS-treated Shank3^-/- mice (p = 0.0152, Figure 1C ). Collectively, these data indicate that SHANK3 confers protection against hypothermia, systemic inflammation and septic death following LPS injection.

Vagus nerve stimulation has been shown to protect against sepsis and systemic inflammation (22). Recent studies reported Shank3 expression in peripheral sensory neurons of DRG (24, 25). However, the expression and role of SHANK3 in the vagus system is unclear. Using a sensitive RNA scope assay, we found extensive Shank3 mRNA expression in majority of vagal sensory neurons in nodose ganglia (NG, Figure 2A ). To evaluate an underlying contribution of peripheral SHANK3 to hypothermia regulation, we generated Shank3 conditional knockout mice by crossing Shank3-floxed mice with Nav1.8-Cre mice, leading to specific deletion of Shank3 in Nav1.8-expressing peripheral sensory neurons including NG neurons (33). In this study, we referred to Shank3^f/f mice as littermate control wildtype (WT) mice, and referred to Nav1.8-cre; Shank3^f/+ and Nav1.8-cre; Shank3^f/f mice as Shank3 conditional knockout (CKO) heterozygous and homozygous mice. Notably, Shank3 expression was substantially reduced in CKO mice (Nav1.8-cre; Shank3^f/f mice) as compared to Shank3^f/f mice (p = 0.0017, Figures 2A, B ). This result indicates a selective expression of Shank3 in NG sensory neurons and also validated Shank3 CKO mice.

As in GKO mice, body temperature analysis revealed no difference of basal body temperature between WT and CKO mice ( Figure 2C ), suggesting that Shank3 in sensory neurons does not contribute to basal body temperature. We next compared hypothermia phenotypes after LPS injection in three genotypes of WT and CKO mice. Herein, we reported a rapid and transient hyperthermia in Shank3^f/f mice, as indicated by the elevation in rectal temperature at 30 minutes after LPS injection as compared to baseline (p = 0.0473, Figure 2C ). Interestingly, this hyperthermia was completely compromised in both heterozygous and homozygous CKO mice (p = 0.0025 and p < 0.0001, respectively) when compared with Shank3^f/f mice. Furthermore, homozygous CKO mice exhibited more rapid (within 1 hour, p=0.0038) and more lasting (>48 hours, p < 0.0001) hypothermia, as compared to Shank3^f/f mice ( Figure 2C ), suggesting that Shank3 deficiency in sensory neurons accelerates the induction of hypothermia and delays the resolution of hypothermia after sepsis. Intriguingly, sepsis-induced hypothermia was significantly aggravated in both heterozygous and homozygous CKO mice (F (2, 448) = 58.38, p < 0.0001, two-way ANOVA, Figure 2C ). There was also significant difference between heterozygous and homozygous CKO mice (F (1, 286) = 16.47, p < 0.0001, two-way ANOVA, Figure 2C ). More strikingly, Shank3 CKO mice exhibited more remarkable changes (reductions) in rectal temperature when compared with Shank3 GKO (p = 0.0136, Figure 2D ), suggesting a dominant role of peripheral SHANK3 in temperature regulation.

Next, we investigated whether LPS-induced systemic inflammation would also be affected in Shank3 CKO mice. Serum IL-6 levels did not differ at 3 hours and 6 hours after LPS stimulation in WT and CKO mice. At 24 hours, IL-6 levels returned to baseline in WT mice but remained at the peak in CKO (F (3, 40) = 29.21, p = 0.0002, two-way ANOVA, Figure 2E ). Survival rate analysis showed that LPS significantly increased mortality (42.90%) in homozygous CKO mice (p=0.0083, vs. WT, Figure 2F ). Collectively, these data demonstrate that SHANK3 in Nav1.8-expressing peripheral sensory neurons confers protection against hypothermia, systemic inflammation and septic death after sepsis.

To investigate a specific role of peripheral SHANK3 in regulating body temperature, we challenged WT and Shank3 CKO mice in different conditions of hypothermia and hyperthermia, including cold room habituation at 4°C ( Figure 3A ), 5% isoflurane anesthesia ( Figure 3B ), intrathecal injection of PGE₂ (200 ng, Figure S3 ), 5% menthol application to the back and auricular region ( Figures 3C, D ), and auricular application of 100 μM capsaicin ( Figure 3E ). Cold room testing showed deficits in homozygous CKO mice (F (1, 60) = 19.49, p < 0.0001, two-way ANOVA, vs. WT) and heterozygous CKO mice (F (1, 56) = 5.568, p = 0.0168 vs. WT) in maintaining core temperature ( Figure 3A ). Interestingly, CKO mice exhibited rapid hypothermia compared to WT mice (F (2, 92) = 10.54, p < 0.05 vs. WT, Figure 3A ). Brief anesthesia to 5% isoflurane produced a rapid drop in body temperature in 10 min, but no difference was found between three genotypes (F (2, 42) = 3.094, p = 0.0558, two-way ANOVA, Figure 3B ). Intrathecal PGE₂ induced hyperthermia and fever in WT mice, which was impaired in Shank3 CKO mice (F (2, 48) = 3.835, p = 0.0285, two-way ANOVA, Figure S1 ). Next, we observed a rapid and transient hyperthermia after 5% menthol smearing in body trunk in Shank3^f/f mice ( Figure 3C ). This hyperthermia at 30 min was partially attenuated in homozygous CKO mice (p = 0.0218, vs. WT), although we did not observe any significant difference among three genotypes (F (2, 80) = 1.776, p = 0.176, two-way ANOVA, Figure 3C ). Next, we smeared 5% menthol to auricular vagal region to mimic vagus nerve stimulation (aVNS, Figure 3D ). This cold/cool stimulation induced hyperthermia in WT mice but not in homozygous CKO mice (F (1, 56) = 12.85, p = 0.0007, two-way ANOVA, Figure 3D ), which was consistent with the results in Shank3 GKO mice ( Figure S2A ). Similarly, aVNS with 100 μM capsaicin smearing produced hyperthermia in Shank3^f/f mice, which was compromised in both heterozygous and homozygous CKO mice (F (2, 92) = 22.51, p < 0.0001, two-way ANOVA, Figure 3E ). The same result was also observed in Shank3 GKO mice ( Figure S2B ).

Next, we recorded electrocardiogram (ECG) to assess VNS-induced heart rate (HR) change following auricular menthol smearing ( Figure 3F ). As expected, menthol-induced VNS resulted in a significant decrease of HR in WT Shank3^f/f mice, which was abolished in homozygous CKO mice (F (1, 14) = 11.73, p = 0.0041, two-way ANOVA, Figures 3G, H ). Collectively, these detailed results suggest that peripheral Shank3 in vagal sensory nerve plays an important role in maintaining core temperature following perturbations.

SHANK3 is expressed by different types of peripheral sensory neurons in mouse DRG, NG, and trigeminal ganglia (34). To confirm a specific role of peripheral SHANK3 in vagal nerve for regulating the LPS-induced effects, we used Shank3 siRNA to knockdown endogenous SHANK3 expression in vagal sensory neurons in NG. A short peptide derived from rabies virus glycoprotein (RVG) has been utilized to facilitate siRNA uptake by neurons including sensory neurons and enhance knockdown efficiency of siRNA (29, 30). We performed bilateral peri-vagal nerve delivery of Shank3 siRNA (mixed with the RVG peptide, 1:10) in CD1 mice ( Figure 4A ). In situ hybridization (ISH) revealed a profound reduction in Shank3 mRNA in NG neurons following Shank3 siRNA (100 nM) treatment (F (2, 9) = 48.70, p = 0.0002, one-way ANOVA, Figure 4B ), suggesting the successful establishment of SHANK3 knockdown by siRNA. We next challenged these mice with LPS on 3 days following siRNA injection. Intriguingly, we found that LPS-induced transient hyperthermia in control siRNA treated mice was completely lost after Shank3 siRNA treatment (p < 0.0001, Figure 4C ). Moreover, LPS-induced hypothermia was significantly aggravated by vagal SHANK3 knockdown (F (1, 75) = 90.52, p < 0.0001, two-way ANOVA, vs. control siRNA, Figure 4C ). Peri-vagal Shank3 siRNA (50 nM, 100 nM and 200 nM) enhanced LPS-induced hypothermia in a dose-dependent manner (F (5, 297) = 53.45, p < 0.0001, two-way ANOVA, Figure S3A ). Furthermore, we detected a robust increase of serum IL-6 levels at 3 hours after LPS injection (p = 0.0027, Shank3 siRNA vs. control siRNA, Figure 4D ). However, we did not detect any difference of mortality between control siRNA and Shank3 siRNA treatment, as CD1 mice are more resistant to sepsis than C57BL/6 mice ( Figures S3B–D ).

Next, we employed GFP fluorescence-labeled AAV9-Cre virus to produce a more persistent knockdown of endogenous SHANK3 expression in vagal sensory neurons, by using bilateral peri-vagal nerve delivery of AAV9 Cre virus in Shank3^f/f mice ( Figure 4E ). Peri-vagal nerve injection of GFP-labeled AAV9 Cre virus resulted in uptake of AAV9 Cre virus in NG neurons as a result of axonal transport from the vagal nerve ( Figure S4A ). Simultaneously, we detected a profound reduction in Shank3 mRNA in NG neurons following the same treatment (p = 0.0002, Figures S4A–C ), manifesting the successful establishment of SHANK3 knockdown by Cre virus in Shank3^f/f mice. We next challenged these mice with LPS at 2 weeks following the virus injection. We found that the Cre virus treatment abolished the LPS-induced transient hyperthermia (p = 0.002 vs. control virus) and exacerbated the LPS-induced hypothermia (F (1, 89) = 36.12, p < 0.0001, two-way ANOVA, vs. control virus, Figure 4F ). Moreover, the Shank3 virus treated mice exhibited a significant increase of serum IL-6 levels (F (1, 24) = 10.18, p = 0.0039, Two-Way ANOVA, vs. control virus, Figure 4G ) following LPS challenge. Vagal SHANK3 knockdown by Cre virus treatment also resulted in a significant reduction in survival rate after LPS intervention in Shank3^f/f mice (p = 0.0274, vs. control virus, Figure 4H ). Collectively, these data indicate that SHANK3 in vagal sensory neurons confers protection against LPS-induced hypothermia, systemic inflammation and septic death.

TRP channels are thermosensors in peripheral nervous system and detect ambient temperature changes (35, 36). First, we employed quantitative RT-PCR to measure mRNA expressions of Trpm2, Trpv1, Trpv4, Trpm8 and Scn10A (encoding Nav1.8) in NG of WT and GKO mice. Interestingly, we detected a robust reduction in Trpm2 (p = 0.00754) but not Trpv1, Trpv4, Trpm8 and Scn10A mRNA levels in NG of Shank3 ^−/− mice ( Figure 5A ). ISH revealed a broad expression of Trpm2 mRNA in majority of NG neurons in WT mice, but this expression was significantly lower in Shank3 ^−/− mice (p < 0.05, vs. WT, Figure 5B ).

To further verify the potential role of peripheral TRPM2 in vagal nerve in sepsis-induced hypothermia, Trpm2 siRNA (100 nM) was employed to knockdown endogenous TRPM2 expression in vagal sensory neurons of CD1 mice through bilateral peri-vagal nerve delivery ( Figure 5C ). ISH detected a significant reduction in Trpm2 mRNA in NG neurons following the siRNA treatment (p = 0.01331, Figure 5D ). Following LPS challenge, Trpm2 siRNA treatment exacerbated hypothermia (F (1, 150) = 120.4, P < 0.0001, vs. control siRNA, Figure 5E ) and increased serum IL-6 levels (F (1, 39) = 39.31, p < 0.0001, vs. control siRNA, Figure 5F ), without affecting mortality in CD1 mice ( Figure 5G ). Together, these data indicate that 1) TRPM2 is a downstream event of SHANK3 signaling in NG neurons and 2) loss of TRPM2 in vagal sensory neurons can recapitulate some of the key features of SHANK3 knockdown, including hypothermia and systemic IL-6 increase.

Dysfunction of SHANK2 protein is also a critical step for the development of autistic-like behaviors (2, 37). We therefore investigated whether SHANK2 is involved in LPS and menthol induced thermal regulation using Shank2 CKO mice with Shank2 deletion in Nav1.8⁺ sensory neurons. We did not find significant differences in LPS-induced hypothermia (F (2, 124) = 0.1497, p = 0.8611, two-way ANOVA, Figure 6A ) and survival rate ( Figure S5A ) between littermate control mice (Shank2^f/f ) and CKO mice (Nav1.8-cre; Shank2^f/+ and Nav1.8-cre; Shank2^f/f ). In parallel, we did not observe significant differences in menthol-induced hyperthermia (F (2, 64) = 0.3973, p = 0.6738, two-way ANOVA, Figure 6B ), capsaicin-induced hyperthermia ( Figure S5B ), and anesthesia-induced hypothermia ( Figure S5C ). Thus, Shank2 in peripheral sensory neurons is not involved in the regulation of body core temperature.

Given a functional link between SHANK3 and TRPV1 in DRG primary sensory neurons and heat sensation (24), we evaluated whether peripheral TRPV1 is important for sepsis-induced hypothermia. Trpv1 siRNA was employed to knockdown endogenous TRPV1 expression in NG neurons via peri-vagal nerve delivery in CD1 mice. ISH detected a substantial reduction in Trpv1 mRNA in NG neurons following Trpv1 siRNA (100 nM) injection [F (2, 9) = 115.8, p < 0.0001, Figure 6C ]. Compared to the control siRNA, Trpv1 siRNA had no significant effects on LPS-induced hypothermia (F (1, 70) = 0.648, p = 0.4235, two-way ANOVA, Figure 6D ) and serum expressions of IL-6 (F (1, 18) = 0.3131, p = 0.5827, Figure 6E ). These findings suggest that TRPV1 in vagal sensory neurons is not required for sepsis-induced hypothermia, systemic inflammation, and mortality.