TLR signaling adaptor protein MyD88 in primary sensory neurons contributes to persistent inflammatory and neuropathic pain and neuroinflammation

Increasing evidence suggests that neuro-immune and neuro-glial interactions are critically involved in chronic pain sensitization. It is well studied how immune/glial mediators sensitize pain, but how sensory neurons control neuroinflammation remains unclear. We employed Myd88 conditional knockout (CKO) mice, in which Myd88 was deleted in sodium channel subunit Nav1.8-expressing primary sensory neurons, to examine the unique role of neuronal MyD88 in regulating acute and chronic pain, and possible underlying mechanisms. We found that baseline pain and the formalin induced acute inflammatory pain were intact in CKO mice. However, the late phase inflammatory pain following complete Freund’s adjuvant injection and the late phase neuropathic pain following chronic constriction injury (CCI), were reduced in CKO mice. CCI induced up-regulation of MyD88 and chemokine C-C motif ligand 2 expression in DRG neurons and macrophage infiltration into DRGs, and microglia activation in spinal dorsal horns in wild-type mice, but all these changes were compromised in CKO mice. Finally, the pain hypersensitivity induced by intraplantar IL-1β was reduced in CKO mice. Our findings suggest that MyD88 in primary sensory neurons plays an active role in regulating IL-1β signaling and neuroinflammation in the peripheral and the central nervous systems, and contributes to the maintenance of persistent pain.

Scientific RepoRts | 6:28188 | DOI: 10.1038/srep28188 as pain or itch 10,[12][13][14][15][16] . MyD88 is also expressed in nociceptive DRG neurons and has been implicated in chemotherapy-induced neuropathic pain in rats 8 . In addition to nociceptive neurons, Na v 1.8 is also expressed in low-threshold A-fibers (e.g., Aβ fiber) DRG neurons 17,18 . However, research on sensory neuron-specific role of TLR signaling is hampered by the lack of tissue (e.g., sensory neuron) specific knockout mice. To this end, we generated MyD88 conditional knockout (CKO) mice by deleting MyD88 selectively in sodium channel Na v 1.8-expressing nociceptive neurons 19 . We found that paclitaxel induced innate and adaptive immunity in DRGs was impaired in CKO mice 19 . In this study, we further investigated the unique role of neuronal MyD88 in acute vs. chronic pain. Here we show that selective deletion of Myd88 in Na v 1.8-expressing primary sensory neurons in CKO mice leads to reductions in complete Freund's adjuvant (CFA) induced inflammatory and chronic constriction injury (CCI) induced neuropathic pain in the maintenance phase, without affecting basal pain and acute inflammatory pain. Further, we show that MyD88 is required for CCI induced up-regulation of the chemokine C-C motif ligand 2 (CCL2, also named as monocyte chemoattractant protein-1 (MCP-1)) in DRG neurons. Finally, macrophage infiltration in DRGs and microglia activation in spinal dorsal horns after CCI are reduced in CKO mice.

Results
Primary deletion of Myd88 adaptor protein in small-sized DRG neurons in CKO mice. We first examined the expression of MyD88 in DRG neurons of littermate mice (WT) and MyD88 CKO mice, in which Myd88 protein was deleted in Na v 1.8-expressing primary sensory neurons, as described in our previous study 19 . Double immunostaining showed that MyD88 was expressed in majority DRG neurons, including both smallsized C-fiber nociceptive neurons (NF-200 − ) and some large-sized A-fiber neurons (NF-200 + , Fig. 1A) in WT mice. While the MyD88 expression was unaltered in large diameter (NF-200 + ) DRG neurons, its expression was dramatically reduced in small-sized (NF-200 − ) neurons in CKO mice (Fig. 1A,C). The quantitative analysis from immunohistochemistry showed that the percentage of MyD88 + neurons was reduced from ~52% (2368/4550) in WT DRGs to ~24% (1373/5559) in CKO DRGs (two-tailed unpaired T-test, t = 13.44, P < 0.0001) (Fig. 1B). Furthermore, size frequency analysis revealed that MyD88 immunoreactivity was mainly localized in small-sized DRG neurons (Fig. 1C). Size frequency analysis also showed loss of MyD88 is small DRG neurons (two-way ANOVA, F (1,249) = 74.48, P < 0.0001, Fig. 1C). By contrast, the size distribution of total DRG neurons did not Six DRG sections were counted per mouse. * P < 0.05, * * * P < 0.001, two-way ANOVA compared with WT; n = 4~5 mice/group. Scientific RepoRts | 6:28188 | DOI: 10.1038/srep28188 change in CKO mice, suggesting that there is not developmental defect of DRG neurons in CKO mice (Fig. 1D). Although our data showed MyD88 loss in small-sized DRG neurons in naïve CKO mice, MyD88 loss may also occur in some large Aβ fiber neurons under inflammatory pain condition. Indeed, functional upregulation of Na v 1.8 was found in Aβ fiber neurons after inflammation 17 . Baseline pain and acute inflammatory pain are normal in CKO mice. We subsequently examined whether depletion of MyD88 in nociceptive neurons would affect normal pain perception (baseline pain sensitivity) and acute inflammatory pain. CKO mice displayed normal motor function in Rota-rod test ( Fig. 2A). CKO mice also have intact baseline pain, by exhibiting normal mechanical and thermal nociception in the Randall-Selitto test and tail-flick test (Fig. 2B,C), respectively. We also tested formalin induced acute inflammatory pain in WT and CKO mice. Both the first phase (0-10 min) and the second phase (10-60 min) nocifensive responses were unaltered in CKO mice (Fig. 2D,E). Two-way ANOVA of Phase II response revealed no significant effect of phenotype (F (1,108) = 0.27, P = 0.616, Fig. 2E). These results suggest that MyD88 in primary sensory neurons is not necessary for generating acute pain.
Persistent inflammatory pain and neuropathic pain are reduced in CKO mice. Next, we examined persistent inflammatory pain induced by intraplantar injection of CFA and persistent neuropathic pain induced by CCI in WT and CKO mice. Notably, CFA induced mechanical allodynia was only impaired in the maintenance phase (3 d and 7 d) but not in the induction phase (2 h and 1 d) in CKO mice (Fig. 3A,B). Two-way  Fig. 3E). Immunohistochemistry assay also confirmed the increased expression of MyD88 in DRG neurons 10 d after CCI (Fig. 4A), assessed by both percentage of positive neurons (50.1% (2465/4863): 60.1% (2480/4125)), two-tailed unpaired t-test, t = 3.16, P = 0.004, Fig. 4B) and intensity of immunostaining (two-tailed unpaired t-test, t = 2.26, P = 0.034, Fig. 4C). Size frequency analysis revealed that increased MyD88 immunoreactivity was mainly localized in small-sized DRG neurons 10 d after CCI (two-way ANOVA, F (1,275) = 8.22, P = 0.0045, Fig. 4D). Together, these results suggest that selective deletion of Myd88 in DRG neurons contributes to the maintenance of chronic inflammatory and neuropathic pain.
To test the mechanisms by which MyD88 regulates IL-1β signaling within DRG neurons, we cultured DRG neurons and exposed them to IL-1β (10 ng/ml) for 3 h. As shown in Fig. 5D, MyD88 immunoreactivity was increased in DRG neurons after IL-1β treatment, compared with vehicle control. Two-tailed unpaired T-test of MyD88 in primary sensory neurons contributes to neuronal CCL2 expression and macrophage activation/infiltration in DRGs. Chemokines such as CCL2 have been implicated in neuroinflammation and chronic pain sensitization 21 . CCL2 was shown to increase Na v 1.8 activity and excitability of DRG neurons 21,22 . Notably, CCL2 expression was found to be dependent on MyD88 in murine mammary carcinomas cells 23 . We postulated that primary sensory neurons recruit macrophages to DRGs through chemokine release. Nerve injury not only induced the activation and infiltration of IBA1-expressing macrophages in DRGs but also caused the up-regulation of CCL2 in DRG neurons (Fig. 6A,B,E,F). Of interest CCI induced IBA1 expression in DRGs was compromised in CKO mice (one-way ANOVA, F (2,12) = 18.75, P = 0.0002, Fig. 6B). Further characterization of macrophages showed that IBA1 is largely co-localized with macrophage markers CD68 or F4/80 in DRGs of CCI mice (Fig. 6C,D). Finally, CCI induced CCL2 expression in DRG neurons, which was compromised in CKO mice (One-way ANOVA, F (2,6) = 53.17, P = 0.0002, Fig. 6E,F). These results suggest that MyD88 in primary sensory neurons contributes to neuronal CCL2 expression and macrophage activation/infiltration in DRGs of chronic neuropathic pain mice.

MyD88 in primary sensory afferents contributes to microglia activation in spinal dorsal horn.
Since primary sensory neurons project to the spinal dorsal horn, we test whether MyD88 in primary sensory afferents regulates the microglia activation in spinal dorsal horn. As expected, we found more IBA1-positive microglia in the spinal dorsal horns of WT mice post CCI procedure compared with sham mice. However, there are less activated microglia in CKO mice compared with WT mice (Fig. 7A,B). One-way ANOVA of statistical analysis revealed a significant effect of phenotype (F (2,8) = 45.54, P = 0.0213, Fig. 7B). This result suggests that MyD88 contributes to microglia activation in spinal cords of mice post CCI procedure.

Discussion
Increasing evidence suggests that neurons expecially primary sensory neurons express TLRs, including TLR3, TLR4, TLR5, TLR7 10,12-16 . TLR3, TLR4, and TLR7 are mainly expressed by small diameter C-fiber neurons that are nociceptive or/and pruriceptive 10,12,14 , whereas TLR5 is primarily expressed in large diameter Aβ neurons that are essential for eliciting mechanical allodynia in neuropathic pain 13 . MyD88 is primarily expressed in small diameter DRG neurons 8 , although some large-sized neurons also express MyD88 (Fig. 1A,C). Intrathecal injection of MyD88 inhibitor inhibited chemotherapy induced neuropathic pain 8 . Consistently, spinal nerve ligation induced neuropathic pain was reduced in Myd88 KO mice 24 . Although these studies support an important role of MyD88 in neuropathic pain, they are not able to prove a specific role of nociceptor MyD88 in acute and chronic pain. Also, global Myd88 deficiency results in both cognitive and motor impairments in mice 25 . Therefore, we generated Myd88 CKO mice in which Myd88 was deleted in Na v 1.8 + sensory neurons (Fig. 1A) 19 . Our data showed that motor function, basal pain perception, and formalin induced acute inflammatory pain were all intact in CKO mice. Consistently, MyD88 has limited involvement in TLR7 mediated TRPA1 activation and acute pain following miRNA let-7b stimulation 26 .
Another interesting finding of this study is that late phase but not early phase inflammatory and neuropathic pain was reduced in CKO mice. This result is consistent with that of global deletion of Myd88 24 . Furthermore, TLR7 ligand let-7b induced persistent mechanical allodynia but not acute pain depends on MyD88 26 . It is conceivable that MyD88 in DRG neurons controls late phase inflammatory and neuropathic pain via gene regulation, which may explain the delayed effects in inflammatory and neuropathic pain in MyD88 CKO mice (Fig. 3A-E). In particular, MyD88 is required for nerve injury induced upregulation of CCL2, in DRG neurons (Fig. 6E,F), consistent with another report in which the release of CCL2 was dependent on MyD88 pathway in murine mammary carcinomas cells 23 . Previous studies showed that CCL2, upregulated in DRG neurons after nerve injury 27 , could increase Na v 1.8 activity and excitability of DRG neurons 22 , leading to an enhanced neuropathic pain state. In addition, CCL2 could enhance neuropathic pain via neuron-macrophage interaction in DRGs. We postulate that secretion of CCL2 from DRG neurons could cause activation of resident macrophages and infiltration of circulating macrophages to DRG tissue. Interestingly, CCI induced IBA1 expression, a marker for macrophage, which co-expresses with CD68 or F4/80 (Fig. 6C,D), was also abrogated in CKO mice (Fig. 6A,B). In another study, global deletion of MyD88 reduced IBA1-positive macrophage infiltration by 50% in dorsal horn of spinal cord following L5 spinal nerve ligation 21,24 . Thus, MyD88 mediated CCL2 expression in primary sensory neurons also regulates neuroinflammation in DRGs (Fig. 6A,B) and in spinal cord (Fig. 7). However, the contribution of sensory neuron-MyD88 to overall inflammation could be limited, since we only found moderate reduction in CFA induced edema in CKO mice (Fig. 3C). Given the marked deficits in innate and adaptive immunity in the peripheral nervous system 19 , sensory neuron-MyD88 may maintain chronic pain, especially neuropathic pain, by controlling neuroinflammation in DRGs and spinal cord. We also found that pain hypersensitivity induced by intraplantar IL-1β was reduced in CKO mice ( Fig. 5A-C), suggesting that neuronal MyD88, presumably expressed by nerve terminals in hindpaw skin, contributed to IL-1β produced pain. Indeed, IL-1 receptor is also expressed in DRG neurons, and IL-1β was shown to rapidly and directly activate nociceptors to generate action potentials 28 . Since intraplantar IL-1β induced mechanical allodynia and heat hyperalgesia was reduced in the early phase (1 and 3 h) in CKO mice, MyD88 in nerve terminals may regulate IL-1β induced acute pain via posttranslational modification that occurs within minutes or tens of minutes. Interestingly, recent research has shown different immune cells mediate pain hypersensitivity in male and female mice, and male dependent TLR4 signaling in the spinal cord 29 . Although male mice were used in the present study, female animals are worth testing in the future.
In summary, our findings have demonstrated that MyD88 expressed by Na v 1.8 + primary sensory neurons contributes to persistent inflammatory pain and neuropathic pain in the maintenance phase, by regulating CCL2 expression in DRG neurons and neuroinflammation in the peripheral and central nervous systems (e.g., macrophage activation and infiltration in DRGs, and microglia activation in spinal cord). MyD88 in sensory neurons is also required for intraplantar IL-1β induced acute pain via possible posttranslational regulation. Together, our data further support an important role of the TLRs/MyD88 pathway in sensory neuron that detects and responds to both exogenous pathogens (PAMP) and endogenous danger signals (DAMP) in chronic pain.

Materials and Methods
Animals. MyD88 flox mice were purchased from Jackson Labs. Na v 1.8 cre transgenic mice were kindly provided by Rohini Kuner (University of Heidelberg) 30 . Breeding colonies were maintained by mating MyD88 f/f with Na v 1.8 cre -MyD88 f/f mice as previous reported 19 . The homozygous conditional knockout mice (MyD88 f/f with Na v 1.8 cre ) referred to as CKO mice, whereas the MyD88 f/f littermates were used as wild type control mice (WT). All animal experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All the experimental procedures with animals were approved by the Institutional Animal Care & Use Committee (IACUC) of Duke University. Adult male mice (> 8~10 weeks) were used and kept under a 12-hour light/dark cycle. Animals were habituated to the testing environment daily for at least two days before behavior experiments. WT and CKO mice were not randomized. All the behavioral tests and tissue quantification were done by individuals who were blinded to the treatment or genotypes of the mice. Young male mice (4~6 weeks) were used for DRG neuron culture. Reagents. We purchased CFA, formalin, IL-1β , paraformaldehyde, DNAse I and Opti MEM culture medium from Sigma-Aldrich Company (St. Louis., MO). Collagenase, trypsin inhibitor and dispase-II were provided by Roche Diagnostics (Mannheim, Germany). B27 supplement was purchased from Invitrogen Company (Carlsbad, CA).

Behavioral tests. Motor function test.
A Rota-rod system (IITC Life Science Inc., Woodland Hills, CA) was used to assess the animal motor function. Mice were tested for three trails separated by 10 min intervals. During the tests, the speed of rotation was accelerated from 2 to 20 r.p.m. within 3 min. The falling latency was recorded and averaged.
Randall-Selitto test. We used Randall-Selitto Analgesy-meter (Ugo Basile, Comerio-Varese, Italy) to examine mechanical sensitivity by applying ascending pressure to the tail of a mouse and determined the mechanical pain threshold when animal showed a clear sign of discomfort or escape, with a cut-off threshold of 250 g to avoid tissue damage.
Tail immersion test. Tail immersion test was used to assess heat pain sensitivity by keeping the tail of a mouse in hot water at 48, 50, or 52 °C and recorded the tail flick latency, with a cut-off time of 10 seconds. Acute inflammatory pain model. Mice were given intraplantar injection of 10 μ l of diluted formalin (5% in saline) to induce acute inflammatory pain. The spontaneous pain behavior was video-recorded and the duration of licking and flinching was measured in a blinded manner in 5 min bins for 60 min after the formalin injection. Persistent inflammatory pain model. Mice were given intraplantar injection of 20 μ l CFA to induce persistent inflammatory pain.
Paw edema measurement. To assess the CFA induced inflammation and edema, paw volume was determined by water displacement plethysmometer (Ugo Basile).
Neuropathic pain model. To induce neuropathic pain, we performed CCI model 31 . Under isoflurane anesthesia, the sciatic nerve of mouse was exposed and then three ligatures (6-0 prolene) were ligated around the nerve proximal to the trifurcation. The ligatures were loosely tied and a distance between each ligature was one millimeter. The sham group received the same surgery but without nerve ligation.
Intraplantar injection of IL-1β induced pain. Mice were given intraplantar injection of 10 μ l IL-1β (10 ng) for behavioral test. Mechanical allodynia and thermal hyperalgesia were tested by using von Frey test and Hargreaves test, respectively.
Von Frey test. Mice were put in a box on elevated metal mesh floor and stimulated hind-paw with a series of von Frey hairs with logarithmically incrementing stiffness (0.08-2.00 grams, Stoelting, Wood Dale, IL). The hairs were presented perpendicular to the plantar surface, and determined the 50% paw withdrawal threshold (PWT) using Dixon's up-down method. For testing mechanical allodynia, we also checked paw withdrawal frequency in response to a subthreshold von Frey hair stimulation (0.16 g, 10 times).
Hargreaves test. For testing heat sensitivity, we put mice in plastic boxes and measured the hindpaw withdrawal latency using Hargreaves radiate heat apparatus (IITCLife Science). The radiant heat intensity was adjusted so that basal paw withdrawal latency was between 9 and 15 s, with a cut-off of 20 s to prevent paw tissue damage.
Immunohistochemistry. Spinal cord and DRG tissue section procedures were described in our previous report 19 . After blocking, the sections were incubated overnight at 4 °C with anti-MCP-1 (rabbit, 1:500; R&D Systems, Minneapolis, MN; mouse, 1:400; Thermo Fisher Scientific, Rockford, IL) or anti-IBA1 (rabbit, 1:2000; Wako, Osaka, Japan) primary antibodies, anti-NF-200 (mouse, 1:5000; Millipore, Billerica, MA), anti-CD68 (rat, 1:500; AbD Serotec, Raleigh, NC), and anti-F4/80 (rat, 1:500; eBioscience, San Diego, CA), followed by mixed FITC-and Cy3-conjugated secondary antibodies (1:400; Jackson ImmunoResearch, West Grove, PA). To improve MyD88 immune staining, antigen retrieval procedure was used. In brief, slides were first processed in citrate buffer (10 mM Aitric Acid, 0.05% Tween-20, pH6.0) at 95~100 °C for 30 min, and then the slides were incubated with anti-MyD88 antibody (rabbit, 1:1000; Abcam, Cambridge, MA) at 4 °C for overnight. For double immunofluorescence, sections were washed with PBS and then incubated with a mixture of polyclonal and monoclonal primary antibodies, followed by a mixture of Cy3-and FITC-conjugated secondary antibodies. For quantification of immunostaining, 3-6 DRG sections were selected from each animal and 4~6 animals were analyzed blindly in each group. To determine the percentage of labeled neurons in DRGs, the number of positive neurons (3 times of background staining) was divided by the total number of neurons. For IBA-1 staining, the number of positive cells was counted and the density of labeled cells (per square mm) was determined. Image-Pro Plus 5.0 software (Media Cybernetics, Silver Spring, MD) was used to analyze the images.
Primary culture of DRG neurons and immunocytochemistry. Young mouse (4~6 weeks) DRGs were removed aseptically and incubated with collagenase (1.25 mg/ml)/dispase-II (2.4 units/ml) at 37 °C for 90 min, then digested with 0.25% trypsin at 37 °C for 8 min, followed by 0.25% trypsin inhibitor. Cells were mechanically dissociated with a flame polished Pasteur pipette in the presence of 0.05% DNAse I. DRG cells were placed on glass cover slips and grown in a neurobasal defined medium (with 2% B27 supplement) with 5 μ M AraC and 5% carbon dioxide at 37 °C. DRG neurons were grown for 24 h before use. For immunocytochemistry, cultured DRG neurons were exposed to IL-1β (10 ng/ml in Opti-MEM) for 3 h, and then fixed with 4% paraformaldehyde for 20 min and followed by immunofluorescence process with anti-Myd88 antibody (rabbit, 1:200; Cell Signal, Beverly, MA). After immunostaining, 4′ , 6′ -diamidino-2-phenylindole (DAPI; 0.1 mg/ml) was added at room temperature for 5 min to stain all the nuclei of cells. The stained sections were examined with a Nikon fluorescence microscope, and images were captured with a CCD Spot camera. Five random regions from each cultured slice, total 4 slices per group were examined. The intensity of fluorescence was analyzed using ImageJ software (NIH; Bethesda, MD).

Statistical analyses.
All data were expressed as mean ± S.E.M. Data were analyzed using Student's t-test (two groups), one-way or two-way ANOVA followed by post-hoc Bonferroni test with GraphPad Prism 5.0 software (La Jolla, CA) and were considered to be statistically significant at P < 0.05.