Molecular signaling underlying bulleyaconitine A (BAA)-induced microglial expression of prodynorphin

Bulleyaconitine (BAA) has been shown to possess antinociceptive activities by stimulation of dynorphin A release from spinal microglia. This study investigated its underlying signal transduction mechanisms. The data showed that (1) BAA treatment induced phosphorylation of CREB (rather than NF-κB) and prodynorphin expression in cultured primary microglia, and antiallodynia in neuropathy, which were totally inhibited by the CREB inhibitor KG-501; (2) BAA upregulated phosphorylation of p38 (but not ERK or JNK), and the p38 inhibitor SB203580 (but not ERK or JNK inhibitor) and p38β gene silencer siRNA/p38β (but not siRNA/p38α) completely blocked BAA-induced p38 phosphorylation and/or prodynorphin expression, and antiallodynia; (3) BAA stimulated cAMP production and PKA phosphorylation, and the adenylate cyclase inhibitor DDA and PKA inhibitor H-89 entirely antagonized BAA-induced prodynorphin expression and antiallodynia; (4) The Gs-protein inhibitor NF449 completely inhibited BAA-increased cAMP level, prodynorphin expression and antiallodynia, whereas the antagonists of noradrenergic, corticotrophin-releasing factor, A1 adenosine, formyl peptide, D1/D2 dopamine, and glucagon like-peptide-1 receptors failed to block BAA-induced antiallodynia. The data indicate that BAA-induced microglial expression of prodynorphin is mediated by activation of the cAMP-PKA-p38β-CREB signaling pathway, suggesting that its possible target is a Gs-protein-coupled receptor – “aconitine receptor”, although the chemical identity is not illustrated.


Results
CREB activation mediates BAA-induced prodynorphin expression in primary microglia. We first assessed the stimulatory effects of BAA on phosphorylation of the transcription factors CREB and NF-κB in primary cultures of microglia. Incubation of 1 μM of BAA for 1 hour induced a robust increase in CREB phosphorylation by 62%, measured by Western blot, compared to the control group (P < 0.05 using unpaired and two-tailed Student's t-test; Fig. 1A). In contrast, the same BAA concentration did not significantly affect NF-κB p65 phosphorylation (Fig. 1B).
Since BAA-increased dynorphin A level measured by the enzyme-linked immunosorbent assay was in parallel with the expression of the cellular prodynorphin gene encoding dynorphin A in spinal microglia from neonatal and adult rats 9 , prodynorphin expression was only measured in this study. Indeed, BAA incubation (1 μM) of cultured primary microglial cells for 2 hour significantly increased expression of prodynorphin mRNA by 2.3 fold, measured by qRT-PCR. The stimulatory effect of BAA was completely inhibited by KG-501 (25 μM) pretreatment, although it did not significantly alter the basal prodynorphin expression (P < 0.05 using two-way ANOVA followed by the post-hoc Student-Newman-Keuls test; Fig. 1D).
Furthermore, KG-501 was also given to neuropathic rats induced by tight ligation of spinal nerves. Intrathecal injection of normal saline (10 μL) or KG-501 (10 μg) 25 did not significantly affect baseline paw withdrawal thresholds in response to mechanical stimulation in both contralateral and ipsilateral paws, whereas intrathecal BAA (300 ng) markedly and time-dependently increased mechanical thresholds in the ipsilateral paws (but not in the contralateral paws) of saline-pretreated rats. BAA mechanical antiallodynia was totally suppressed by 30 minute-pretreatment of KG-501 (P < 0.05 using two-way ANOVA followed by a post-hoc Student-Newman-Keuls test; Fig. 1E).
p38 activation mediates BAA-induced prodynorphin expression in primary microglia. MAPKs are a family of signaling molecules to transduce extracellular stimuli into intracellular responses in a wide variety of circumstances. The MAPK family includes p38, ERK and JNK, and is activated through different upstream transduction chain reactions 26 . In order to assess whether BAA affected MAPK activation, we measured phosphorylation of p38, ERK and JNK using Western blot. Treatment with 1 μM BAA in cultured primary microglia for 1 hour significantly induced p38 phosphorylation by 2.5 fold (P < 0.05 using the unpaired and two-tailed Student's t-test; Fig. 2A). However, the same concentration of BAA failed to activate ERK (Fig. 2B) or JNK (Fig. 2C). , and blockade effects of the specific CREB activation inhibitor KG-501 on BAA-induced CREB phosphorylation (C) and prodynorphin expression (D) in primary cultures of microglia, and spinal mechanical antiallodynia (E) in neuropathic rats. Primary microglial cells were collected from the spinal cord of 1-day-old neonatal rats. For the CREB and NF-κB phosphorylation, 1 μM BAA was added into primary microglia culture for 1 hour and the levels of phosphorylated CREB and NF-κB p65 were determined using Western blot. The representative blots are shown in the upper panels of each panel. The dynorphin A precursor prodynorphin mRNA expression was measured using qRT-PCR and normalized to GAPDH mRNA. For mechanical antiallodynia, neuropathic rats were induced by tight ligation of L5/L6 spinal nerves and mechanical thresholds were measured in both contralateral and ipsilateral paws. Data are expressed as means ± SEM (n = 3 in each treatment with two independent repeats in cultured primary microglial cells, and n = 6 in each group of neuropathic rats). a,b P < 0.05 using the unpaired and two-tailed Student's t-test or two-way repeated measures ANOVA followed by the post-hoc student-Newman-Keuls test, compared with the control and BAA groups, respectively. For phosphorylation, 1 μM BAA was added into microglia culture for 1 hour and levels of phosphorylated p38, ERK, JNK and CREB were determined by using Western blot. The representative blots are shown in the upper panels of each panel. For gene expression, the MAPK inhibitors were incubated 30 minutes before BAA treatment and primary microglial cells were collected 2 hours later. Expression of the dynorphin A precursor prodynorphin gene was determined using qRT-PCR and normalized to GAPDH mRNA. For mechanical antiallodynia, neuropathic rats were induced by tight ligation of L5/L6 spinal nerves, and mechanical thresholds were measured in both contralateral and ipsilateral paws. (F) Blockade effect of the p38 inhibitor SB203580 on BAA-induced CREB phosphorylation in cultured primary microglia. Data are expressed as means ± SEM (n = 3 in each treatment with two independent repeats in cultured primary microglial cells, and n = 6 in each group of neuropathic rats). a,b P < 0.05 using the unpaired and two-tailed Student's t-test or two-way repeated measures ANOVA followed by the post-hoc student-Newman-Keuls test, compared with the control and BAA groups, respectively. To further determine whether p38 activation was causally associated with BAA-stimulated prodynorphin expression, the selective p38 inhibitor SB203580, ERK inhibitor U0126, and JNK inhibitor SP600125 were employed. SB203580 (50 μM) 27 , U0126 (50 μM) 28,29 and SP600125 (50 μM) 30 treatment did not significantly alter basal levels of prodynorphin expression, whereas 30 minutes SB203580 pretreatment completely blocked 1 μM BAA-increased prodynorphin expression (P < 0.05 using two-way ANOVA followed by a post-hoc Student-Newman-Keuls test). In contrast, neither SP600125 nor U0126 significantly suppressed BAA-increased prodynorphin expression (Fig. 2D).
In neuropathic rats, mechanical thresholds were measured in both contralateral and ipsilateral paws. Intrathecal injection of 10 μL of normal saline or 10 μg each of SB203580, U0126 and SP600125 31-33 did not significantly affect baseline mechanical withdrawal response in either contralateral or ipsilateral paws. In saline-pretreated rats, intrathecal BAA (300 ng) in the ipsilateral paws produced a time-dependent mechanical antiallodynia, which was completely prevented by the 30 minutes SB203580 pretreatment (P < 0.05 using two-way ANOVA followed by a post-hoc Student-Newman-Keuls test), but not with U0126 and SP600125 (Fig. 2E).
We further assessed the p38-dependency of BAA-induced CREB phosphorylation. Treatment with SB203580 (50 μM) did not significantly affect the basal levels of CREB phosphorylation, but nearly completely inhibited BAA (1 μM)-upregulated CREB phosphorylation (P < 0.05 using two-way ANOVA followed by the post-hoc Student-Newman-Keuls test; Fig. 2F).
There are four isoforms of p38 MAPK, i.e., α, β, γ, and δ, among which α and β are the major isoforms in the mature nervous system and the targets of most p38 inhibitor drug development 26,34 . To identify which p38 isoforms mediate BAA-induced prodynorphin expression, p38α and p38β expression in cultured primary microglia was specifically silenced using the RNA interference technology. Compared to the nonspecific oligonucleotide control, transfection with siRNA/p38α (0.4 μM) for 5 hours significantly reduced p38α mRNA expression by 62%, measured by qRT-PCR (P < 0.05 using one-way ANOVA followed by a post-hoc Student-Newman-Keuls test), whereas it did not significantly affect p38β mRNA expression (Fig. 3A). Western blot also showed the identical data, i.e., compared to the nonspecific oligonucleotide control, siRNA/p38α significantly reduced levels of p38α protein by 66% (P < 0.05 using one-way ANOVA followed by a post-hoc Student-Newman-Keuls test), but not p38β protein levels (Fig. 3B). Furthermore, transfection with 0.4 μM siRNA/p38β significantly reduced p38β mRNA by 64% and p38β protein by 55%, compared to the nonspecific oligonucleotide control (P < 0.05 using one-way ANOVA followed by a post-hoc Student-Newman-Keuls test), but did not affect expression of p38α mRNA or protein (Fig. 3C,D).
Further studies were undertaken to identify which isoform of p38 mediated BAA-induced spinal mechanical antiallodynia in neuropathy. Five groups of neuropathic rats received multiple daily intrathecal injections of the vehicle (Dotap, 40 μg/day), nonspecific siRNA oligonucleotide (5 μg/day), siRNA/p38α (5 μg/day), or siRNA/ p38β (5 μg/day) for 7 days. The mechanical thresholds in both contralateral and ipsilateral paw were measured each morning before injection. On the eighth day, the rats received a single intrathecal injection of saline (10 μL for the first vehicle group) or BAA (300 ng for the remaining four groups). Mechanical thresholds were measured once 1 hour after BAA injection. As shown in Fig. 4A, consecutive daily intrathecal injections of siRNA/p38α and siRNA/p38β over 7 days did not significantly alter mechanical thresholds in either contralateral or ipsilateral paws, compared with the vehicle or nonspecific siRNA oligonucleotides control groups. On the 8th day, single intrathecal injection of BAA (300 ng) produced marked mechanical antiallodynia in the ipsilateral paws. Knockdown of p38β gene using siRNA/p38β completely prevented BAA mechanical antiallodynia (P < 0.05 using two-way ANOVA followed by the post-hoc Student-Newman-Keuls test), whereas multiple daily injections of siRNA/p38α was not able to block BAA-induced mechanical antiallodynia (Fig. 4B).
After completion of the behavior test, all rats were immediately sacrificed and spinal lumbar segments were obtained for later gene and protein assessment. Compared to the nonspecific siRNA oligonucleotides control, multiple daily intrathecal injections of siRNA/p38α significantly reduced p38α mRNA level by 51% (P < 0.05 using one-way ANOVA followed by post-hoc Student-Newman-Keuls test), but not p38β mRNA (Fig. 4C). Western blot showed similar data, i.e., siRNA/p38α significantly reduced p38α protein expression by 61% (P < 0.05 using one-way ANOVA followed by post-hoc Student-Newman-Keuls test), but not p38β protein (Fig. 4D). In contrast, multiple daily intrathecal injections of siRNA/p38β significantly reduced p38β mRNA level by 62% and protein by 50% (P < 0.05 using one-way ANOVA followed by post-hoc Student-Newman-Keuls test), but not p38α mRNA or protein (Fig. 4E,F).

The cAMP-PKA signaling mediates BAA-induced prodynorphin expression in primary microglia.
Scientific RepoRts | 7:45056 | DOI: 10.1038/srep45056 Figure 3. Differential effects of siRNA/p38α and siRNA/p38β on expression of p38α (A,B) and p38β mRNA and protein (C,D), bulleyaconitine A-induced p38 phosphorylation (E) and prodynorphin gene expression (F) in primary cultures of microglia. Primary microglial cells were collected from the spinal cord of 1-day-old neonatal rats. For the gene knockdown, siRNA/p38α and siRNA/p38β were transfected into primary microglia for 5 hours and microglia were cultured for additional 24 hours after refreshed the culture medium. Expressions of the mRNA of p38α, p38β and prodynorphin and protein of p38α and p38β referred to gapdh were determined using qRT-PCR and Western blot, respectively. For the phosphorylation assay, primary microglial cells were treated with 1 μM BAA for 1 hour and then subjected to Western blot analysis of phosphorylated p38 with an anti-phospho-p38 antibody. The representative blots are shown in the upper panels of each panel. Data are expressed as means ± SEM (n = 3 per group in each treatment with three independent repeats). a,b P < 0.05 using one-way or two-way repeated measures ANOVA followed by the post-hoc student-Newman-Keuls test, compared with the nonspecific oligonucleotide control and BAA groups, respectively. To assess whether BAA-induced prodynorphin expression and mechanical antiallodynia were through the cAMP/PKA signaling, the specific adenylate cyclase inhibitor DDA and PKA activation inhibitor H-89 were employed in cultured primary microglia and neuropathic rats. Treatment with 1 μM BAA for 2 hours significantly , and expression of spinal p38α (C and D) and p38β (E and F) mRNA and protein in neuropathic rats. Neuropathic rats, induced by tight ligation of L5/L6 spinal nerves, and then received multiple-daily intrathecal injection of the vehicle (DOTAP, 30 μg/day), nonspecific oligos (5 μg/day), siRNA/p38α (5 μg/day) or siRNA/p38β (5 μg/day) for 7 days. The paw withdrawal thresholds were measured once a day using electronic von frey filaments. On the eighth day, a single bolus of saline (10 μL) or BAA (300 ng) was intrathecally injected and mechanical nociceptive behaviors were quantified 1 hour after injection. Spinal lumbar enlargements were immediately obtained after completion of behavioral assessments. Expressions of the mRNA of p38α, p38β and prodynorphin and protein of p38α and p38β referred to gapdh were determined using qRT-PCR and Western blot, respectively. The representative blots are shown in the upper panels of each panel. Data are expressed as means ± SEM (n = 6 in each group). a,b P < 0.05 using one-way or two-way repeated measures ANOVA followed by the post-hoc student-Newman-Keuls test, compared with the control and BAA groups, respectively.  . Primary microglial cells were collected from the spinal cord of 1-day-old neonatal rats. For the cAMP measurement, these microglia were incubated with 1 μM BAA for 30 minutes and the intracellular cAMP accumulation was assayed using a commercial ELISA kit. For phosphorylation of PKA and p38, 1 μM BAA was incubated with cultured microglia for 1 hour, and subjected to Western blot analysis. The representative blots are shown in the upper panels of each panel. For gene expression, the MAPK activation inhibitors were incubated 30 minutes before BAA treatment, and primary microglial cells were collected 2 hours later. The level of dynorphin A precursor prodynorphin mRNA was determined using qRT-PCR and normalized to GAPDH mRNA. For mechanical antiallodynia, neuropathic rats were induced by tight ligation of L5/L6 spinal nerves, and mechanical thresholds were measured in both contralateral and ipsilateral paws. Blockade effects of DDA (E) and H-89 (F) on BAA-induced p38 phosphorylation were also assessed. Data are presented as means ± SEM (n = 3 in each treatment with two independent repeats in cultured primary microglial cells, and n = 6 in each group of neuropathic rats). a,b P < 0.05 using the unpaired and twotailed Student's t-test or two-way repeated measures ANOVA followed by the post-hoc student-Newman-Keuls test, compared with the control and BAA groups, respectively.
Pretreatment of neuropathic rats with DDA (1 μg) and H-89 (10 μg) 37 did not significantly affect baseline paw withdrawal thresholds in response to mechanical stimulation in both contralateral or ipsilateral paws, but completely prevented BAA-induced mechanical antiallodynia (P < 0.05 using two-way ANOVA followed by a post-hoc Student-Newman-Keuls test; Fig. 5D).
To further determine whether BAA-induced p38 phosphorylation was cAMP/PKA-dependent, DDA and H-89 were applied in cultured primary microglia. We found that pretreatment (1 hour before) with DDA (100 μM) and H-89 (10 μM) did not significantly affect the basal p38 phosphorylation level, but completely blocked BAA-upregulated p38 phosphorylation (P < 0.05 using two-way ANOVA followed by the post-hoc Student-Newman-Keuls test; Fig. 5E,F).
A GsPCR possibly mediates BAA-induced prodynorphin expression in primary microglia. To explore Gs-protein involvement in BAA-mediated cAMP production and subsequent prodynorphin expression, we employed a Gs inhibitor NF449 in cultured microglia. NF449 was identified as a selective inhibitor of the Gs signaling, with limited effects on the prototypical Gi/Go and Gq signaling 38 . BAA treatment (1 μM) for 30 minute significantly increased intracellular cAMP level, whereas NF449 treatment (10 μM) 39 in cultured primary microglial cells did not significantly affect the basal intracellular cAMP level. However, 30 minutes NF449 pretreatment completely inhibited BAA-increased cAMP accumulation (P < 0.05 using two-way ANOVA followed by a post-hoc Student-Newman-Keuls test; Fig. 6A).
Furthermore, intrathecal BAA injection (300 ng) produced a marked mechanical antiallodynia in the ipsilateral paws in a time-dependent manner. 30 minutes pretreatment with intrathecal NF449 (10 μg) 40,41 did not significantly change baseline paw withdrawal thresholds in either contralateral or ipsilateral paws, but completely prevented spinal BAA mechanical antiallodynia in the ipsilateral paws (P < 0.05 using two-way ANOVA followed by the post-hoc Student-Newman-Keuls test; Fig. 6C).

Discussion
In the current study, we have dissected and proposed signal transduction pathways for aconitines (represented by BAA) to stimulate prodynorphin expression in microglia. By binding to an unclarified GsPCR -"aconitine receptor" which is possibly specifically localized on microglial cell membrane, BAA, other aconitines (such as aconitine, bullatine A and lappaconitine) and the Aconitum extracts are able to activate Gs-protein and in turn trigger adenylate cyclase and produce a large amount of cAMP, which further stimulates PKA and subsequent p38β phosphorylation. The p-p38β then translocates to the nucleus and activates the transcription factor CREB, leading to an increase in prodynorphin transcription. The proposed signal transduction pathway is presented in Fig. 8. The illustration of the entire signal transduction pathway of aconitines further supports the notion that their antinociception is through spinal microglial prodynorphin expression and dynorphin A release, but not through the interactions with voltage-gated sodium channels which is not involved in the Gs-PKA-p38-CREB pathway. Furthermore, our results also highlight the selective role of p38β in pain transduction and provide an opportunity to identify the unclarified GsPCR, i.e., "aconitine receptor".
CREB, a member of the basic leucine zipper family of the transcription factors, regulates a large and diverse group of gene transcription and function in cellular responses to various physiological stimuli, including neurotransmitters, growth factors, and cell stress 20 . Specifically, the transcriptional activity induced by phospho-CREB initiates binding to CRE present in the target gene promoter 52 . Previous studies showed that CREB regulated the transcription of prodynorphin gene in neurons 22,23 , while our current study demonstrated that BAA induced a robust CREB phosphorylation and prodynorphin expression in primarily cultured microglia, which was completely inhibited by the selective CREB inhibitor KG-501, indicating that BAA-stimulated microglial prodynorphin expression was mediated by CREB activation. In contrast, BAA did not induce phosphorylation of NF-κB, Scientific RepoRts | 7:45056 | DOI: 10.1038/srep45056 which is also a pleiotropic factor to promote transcription of a wide variety of genes. In microglia, activated NF-κB translocates into the nucleus and regulates the synthesis and release of proinflammatory cytokines to play a pivotal role in neuroinflammation 53 . However, bullatine A and lappaconitine did not significantly inhibit microglial expression of proinflammatory cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-6, and IL-1β 11,12 . Both results rule out the possible involvement of NF-κB in mediation of BAA signaling transduction in microglia.
CREB is initially identified as a PKA substrate and phosphorylation of the serine-133 site makes it function as a stimulus-dependent transcriptional activator 21 . Another alternative pathway of the p38/CREB is also observed later, i.e., CREB phosphorylation can also be directly induced by p38 MAPK at the same serine 133 site 22 . Thus, these classic and alternative pathways do exist in different cells and respond to different gene signaling 54,55 . In our study, we found that selective p38 activation inhibitor SB203580 was able to entirely abolish BAA-activated microglial CREB phosphorylation and prodynorphin expression, suggesting that p38 MAPK activation plays an obligatory intermediate role in PKA-induced CREB's regulation of prodynorphin expression in microglia.
We demonstrated that BAA specifically activated p38 without a significant effect on ERK or JNK activation. The selective p38 inhibitor but not the ERK or JNK inhibitor could entirely eliminate BAA-increased prodynorphin expression in cultured primary microglia. Our current data were supported by recent studies showing that aconitine treatment of myocardial cells induced p38 phosphorylation and cell apoptosis, both of which were significantly attenuated by SB203580 treatment 56 , and that the microglial inhibitor minocycline completely inhibited aconitine-, BAA-and bullatine A-increased dynorphin A gene and protein expression and antinociception in neuropathy [9][10][11]42 . Minocycline has been shown to inhibit spinal p38 activation 57,58 , particularly in a neuropathic pain condition 59 , although it was also shown to inhibit ERK 60 and JNK activity 61 . Taken altogether, our data . Primary microglial cells were collected from the spinal cord of 1-day-old neonatal rats. For the cAMP measurement, microglial cells were incubated with 1 μM BAA for 30 minutes and subjected to ELISA analysis of the intracellular cAMP level using a commercial ELISA kit. For gene expression, the cultured cells were collected 2 hours after incubation with test reagents and subjected to qRT-PCR analysis of the dynorphin A precursor prodynorphin mRNA referred to GAPDH mRNA. For mechanical antiallodynia, neuropathic rats were induced by tight ligation of L5/L6 spinal nerves, and mechanical thresholds were measured in both contralateral and ipsilateral paws. Data are presented as means ± SEM (n = 3 in each treatment with two independent repeats in cultured primary microglial cells, n = 6 in each group of neuropathic rats). a,b P < 0.05 using two-way repeated measures ANOVA followed by the posthoc student-Newman-Keuls test, compared with the control and BAA groups, respectively. revealed that BAA specifically activated p38 and subsequent CREB phosphorylation leading to prodynorphin expression. Furthermore, our recent study showed that GLP-1 receptor agonism-induced β-endorphin expression was solely mediated by a p38-dependent CREB activation 62 . These data are of significance in pain transmission and transduction, as activation of p38 is an important intracellular signaling event which exclusively occurred in microglia and in turn regulated pain hypersensitivity, whereas activation of ERK and JNK activation usually occurs in dorsal horn neurons or astrocytes (but not in microglia) after spinal nerve injury 63 . Furthermore, p38 is a downstream target of several signal molecules 27 and activation of cAMP/PKA to phosphorylation of MAPKs has been shown to be an important signal transduction process. We demonstrated that BAA-induced phosphorylation of p38 and prodynorphin expression were completely suppressed by DDA and H-89, highlighting that BAA-induced p38 phosphorylation is cAMP/PKA-dependent in microglia.
However, our finding seems contradictory as it is known that p38 phosphorylation was also involved in expression of pro-inflammatory cytokines, IL-1β, IL-6, and TNF-α 64,65 . It is true that p38 has α, β, γ, and δ isoforms and that α and β are the major isoforms in the mature nervous system 26,34,66 . The p38 α isoform was responsible for expression of pro-inflammatory cytokines [67][68][69][70] . While p38β appeared to be the most abundant isoform in spinal microglia 34 , it was not involved in inflammation and tissue damage 69,70 . We recently evidenced that GLP-1 receptor agonism-induced β-endorphin expression and mechanical antiallodynia in neuropathy were fully mediated by p38β activation, whereas lipopolysaccharides-induced expression of proinflammatory cytokines was partially mediated by p38α activation 62 . Our data further confirmed that p38β activation mediates BAA-induced prodynorphin expression in cultured primary microglial cells.
Many investigators have demonstrated that nerve injury would induce p38 phosphorylation in spinal cord microglia, which contributed to development and maintenance of neuropathic pain 26 . For instance, p38 activation in the neuropathic pain state exclusively occurred in microglia, but not in neurons or astrocytes in the dorsal horn 71 . However, there are contradictory findings regarding p38 as a target molecule for treatment of chronic pain in experimental animals, especially in established neuropathic pain. It was reported that selective p38 activation inhibitor SB203580 cannot reverse neuropathic pain when given to animals after neuropathy was established 72 . Our recent study reported that knockdown of spinal p38α or p38β expression was not antiallodynic 62 . Our current results demonstrated that multiple-daily injections of siRNA/p38α or siRNA/p38β did not alter mechanical thresholds in the ipsilateral paws in neuropathic rats. All these results together suggest that spinal p38 or its α or β isoform does not regulate basal pain states.
Our study revealed that BAA treatment dose-dependently enhanced cAMP accumulation and PKA phosphorylation in cultured primary microglial cells. BAA-induced prodynorphin expression in microglia were completely blocked by DDA and H-89, and particularly the specific Gs-protein inhibitor NF449, suggesting that BAA-increased prodynorphin expression was mediated by the Gs/cAMP/PKA signaling pathway. Thus, we speculate that a GsPCR expressed on microglial cell membrane could be the target molecule for BAA to stimulate prodynorphin expression and dynorphin A release. We tentatively name the unidentified GsPCR as the "aconitine receptor".

Figure 8. Illustration of the proposed signal transduction pathway of bulleyaconitine A (BAA), bullatine A, lappaconitine, and the Aconitum extracts-induced prodynorphin expression in microglia.
Following activation of the "aconitine receptor", the Gs-protein, cAMP/PKA, p38β (but not ERK, JNK or p38α) MAPK and CREB (but not NF-κB) signals are successively activated, which mediates prodynorphin expression.
GPCRs are the largest family of receptors to respond for a variety of stimuli, ranging from neurotransmitters and hormones to light and mechanical pressure, and represent targets for approximately 40% of all approved drugs in the clinic. It would be extremely valuable to illustrate the chemical identity of the "aconitine receptor" on microglia. In an effort to identify the "aconitine receptor", we assessed the blockade ability of several specific GPCR antagonists in neuropathy on BAA mechanical antiallodynia, which has been demonstrated to be induced by spinal microglial dynorphin A expression (Li et al., 2016a;Li et al., 2016b;Huang et al. 11 ; Huang et al. 42 ). The GPCR receptors assessed in our study included the β-adrenoceptor (and β 1 -adrenoceptor), α-adrenoceptor, corticotrophin-releasing factor receptor, adenosine A1 receptor, formyl peptide receptor, D1/D2 dopamine receptor, and GLP-1 receptor. However, all the antagonists failed to alter BAA mechanical antiallodynia in neuropathy. It is worth noting that the D1/D2 dopamine receptor antagonist levo-tetrahydropalmatine did not alter BAA-induced mechanical antiallodynia, suggesting that BAA-induced dynorphin A expression is not involved in activation of D1/D2 dopamine receptors in microglia, whereas activation of D1 dopamine receptor was reported to stimulate dynorphin A expression in a cAMP/PKA/CREB signaling-dependent manner in nucleus accumbens and striatonigral neurons (Carlezon et al. 23 ; Muschamp and Carlezon, 2015). Further studies are warranted to utilize techniques, such as GsPCR binding or functional screening assays of large scale and chemical proteomics, to identify the particular GsPCR involved in BAA-induced spinal microglial dynorphin A expression and mechanical antiallodynia in animals.
Primary microglia were seeded into 6 or 24-well plates and grown overnight. The siRNA-DOTAP complex was then added into the plates with 300 μL/500 μL basic Dulbecco's modified Eagle's medium (DMEM) to make the final siRNA concentration of 5 μg/mL. The plates were then incubated in a humidified incubator with 5% CO 2 at 37 °C for 5 hours. Cells were further cultured for 24 hours after refreshing with the cultural medium. For the in vivo experiments, the siRNA-DOTAP complex was formulated in the same DOTAP:RNA mass ratio of 1:8 and intrathecally injected into rats for successive 7 days. Experimental animals. Male adult (200 ± 20 g body weight) and 1-day-old Wistar neonatal rats were purchased from the Shanghai Experimental Animal Institute for Biological Sciences (Shanghai, China). The adult rats were maintained four per cage, received food and water ad libitum, and housed in the Shanghai Jiao Tong University Experimental Animal Center (Shanghai, China) with thick sawdust bedding at standard room temperature (22 ± 2 °C), under conditions of a 12/12-hr reversed light-dark cycle (7:00 am-7:00 pm). The adult rats were accustomed to the laboratory environment for 3-5 days before our experiments. Experimental groups (n = 6 in each group except for the target screening study as indicated in the text) were randomly assigned, and the investigator was blinded for the behavior tests. This study was approved by the Animal Care and Welfare Committee of Shanghai Jiao Tong University and carried out in accordance with the animal care guidelines of the US National Institutes of Health.
Primary culture of microglial cells. Microglial cells were isolated from the spinal dorsal horn of 1-day-old neonatal rats. The isolated spinal cords were minced and then incubated with 0.05% trypsin solution. Dissociated cells were suspended in DMEM supplemented with 10% (vol/vol) fetal calf serum and penicillin (100 U/mL) and streptomycin (100 μg/mL) and plated in 75-cm 2 tissue culture flasks (1 × 10 7 cells/flask) that were pre-coated with poly-L-lysine and maintained in a 5% carbon monoxide incubator at 37 °C. After culture for 8 days, microglial cells were prepared as floating cell suspensions by shaking the flasks at 260 rpm for 2 hours. The aliquots were transferred to plates and unattached cells were removed by washing with serum-free DMEM. Harvested microglial cells showed a purity >95%, as determined by the OX42 immunoreactivity (Gong et al., 2014).

RNA isolation and quantitative reverse transcription-polymerase chain reaction (qRT-PCR).
The spinal dorsal lumbar enlargements were collected and homogenized using an electronic microhomogenizer at 4,000 rpm for 10 seconds in TRIzol (Invitrogen, Carlsbad, CA, USA) on ice, while total RNA from cultured microglial cells was also isolated directly using TRIzol. The RNA samples of 1 μg each were reversely transcribed into cDNA using a ReverTra Ace qPCR RT-Kit (Toyobo Co., Osaka, Japan). qPCR was then carried out in a Mastercycler ep realplex (Eppendorf, Germany) using Realmaster Mix (SYBR Green I) (Toyobo Co.) according to a previous study 73 . The fold change was calculated using the 2 −ΔΔCt method after normalization to GAPDH. The primers were GAPDH (5′-CCA AGG TCA TCC ATG ACA AC-3′ and 5′-TCC ACA GTC TTC  TGA GTG GC-3′) 51 and PDYN (5′-ACT GCC TGT CCT TGT GTT CC-3′ and 5′-CCA AAG CAA CCT CAT  TCT CC-3′) 74 .
Protein extraction and Western blot. Cultured microglia were seeded into 6-well plates at a density of 5 × 10 6 cells per well and grown overnight and then harvested and lysed in the radioimmunoprecipitation assay (RIPA) lysis buffer with addition of the phosphates inhibitor cocktail and the protease inhibitor cocktail (Biotool, Houston, USA). Briefly, cells were scraped into the lysis buffer and homogenized by passing through a 21G needle and cell lysates were centrifuged at 13,000 g at 4 °C for 15 minutes. The supernatants were then collected and the protein concentration was quantified using a bicinchoninic acid (BCA) assay according to the manufacturer's protocol (Beyotime Institute of Biotechnology, Jiangsu, China). The homogenates were diluted 1:1 (v/v) with 5× sodium dodecyl sulfate (SDS) sample buffer (Bio-Rad, Hercules, CA, USA) and boiled for 10 minutes at 95 °C in the SDS sample buffer. The protein samples were then separated in 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels and transferred onto a polyvinylidene fluoride (PVDF) membrane using the electrophoretic method. The membrane was then blocked in 5% skim milk powder in Tris-based saline-Tween 20 (TBST) at room temperature for one hour, and then incubated at 4 °C overnight with rabbit monoclonal primary antibodies raised against phospho-p38, phospho-ERK, phospho-JNK, p38α, p38β, phospho-CREB (Cell Signaling Technology, Danvers, MA, USA) at a dilution of 1:1,000, anti-phospho-PKA catalytic subunit (Santa Cruz Biotechnology) at a dilution of 1:200, and a mouse monoclonal antibody against GAPDH (Protein Tech Group, Chicago, USA) at a dilution of 1:4,000, respectively. The membranes were then washed four times in TBST (10 minutes each) and further incubated 1 hour with a secondary peroxidase conjugated goat anti-rabbit and goat anti-mouse antibody, and washed four times with TBST. Protein bands were visualized using the IRDye 800-conjugated affinity purified goat anti-rabbit IgG and IRDye 700-conjugated affinity purified goat anti-mouse IgG (Cell Signaling Technology, Danvers, MA, USA) in the Odyssey Infrared Imaging system (Li-Cor Biosciences, Lincoln, NE, USA). The band intensity was quantitated using a computer-assisted image analysis program (ImageJ Software, National Institutes of Health, Bethesda, MD, USA) and compared to the house-keeping protein, i.e., the band intensity ratios of phospho-p38/GAPDH, phospho-ERK/GAPDH, phospho-JNK/GAPDH, phospho-PKA/GAPDH, p38α/GAPDH, p38β/GAPDH, and phospho-CREB/GAPDH were obtained to quantify the respective target protein expression 51,75 . Intracellular cAMP assay. Cultured microglial cells (5 × 10 6 ) were seeded into 6-well plates and treated with the phosphodiesterase inhibitor IBMX (Sigma-Aldrich) at 0.5 mM in serum-free medium for 30 minutes before treatment with different concentrations (10 −10 , 10 −9 , 10 −8 , 10 −7 , 10 −6 , and 10 −5 M) of bulleyaconitine A. 30 minutes after incubation, microglial cells were lysed using the R&D Cell Lysis Buffer and cAMP release was determined using a commercial cAMP enzyme immunoassay kit (R&D Systems, Minneapolis, USA) according to the manufacturer's protocols. The concentration of cAMP was calculated using a standard curve using the Assay Zap (Biosoft, Cambridge, UK) and normalized. Data represented two independent experiments with triplicate 76 .

Intrathecal catheterization and injection in rats.
A 18-cm polyethylene catheter (PE-10: 0.28 mm inner diameter and 0.61 mm outer diameter) (Clay Adams, Parsippany, NJ, USA) with a volume of 13 μL was inserted into the rat lumbar level of the spinal cord under inhaled isoflurane anesthesia (4% for induction and 1% for maintenance) run by an anesthesiameter (Ugo Basile Gas Anesthesia System, Comerio, Italy). Two days after recovery from anesthesia, the installation of the catheter in the spinal cord was verified by administering 4% lidocaine (10 μL followed by 15 μL of saline for flushing). Rats that did not induce motor impairment following insertion of the intrathecal catheter and developed immediate bilateral paralysis of the hindlimbs following intrathecal lidocaine were selected for the study. For the intrathecal delivery, 10 μL of the drug solution was administered in a 50-μL microinjector (Shanghai Anting Micro-Injector Factory, Shanghai, China) followed by 15 μL saline flush 77 . Rat neuropathic pain model and behavioral assessment of mechanical allodynia. The unilateral ligation of two spinal nerves was performed under inhaled isoflurane anesthesia run by the anesthesiameter. The left L5 and L6 spinal nerves were isolated and tightly ligated with 6-0 silk thread. After ligation, the wound was sutured and the rats were allowed to recover. As intrathecal injection was needed in this study, the intrathecal catheterization was performed in rats at the same time just before spinal nerve ligation. Of the spinal nerve-ligated animals, only those with marked unilateral allodynia to mechanical stimulation (hindlimb withdrawal thresholds in the operated side <8 g) and no major impairment were included in the study. Drug testing started 2-4 weeks after spinal nerve ligation surgery 78,79 ).
To evaluate mechanical allodynia, the animals were acclimatized for at least 30 minutes to the test environment, namely a plexiglass box on a metal grid (0.5 × 0.5 cm). The hindpaw withdrawal threshold was measured by using a 2450 CE Electronic Von Frey hair (IITC Life Science Inc, Woodland Hill, CA, USA). The electronic hand-held transducer with a No. 15 monofilament (with forces ranging from 0.1 to 90 g) was applied perpendicularly to the medial surface of the hindpaws with the increasing force until the rat suddenly withdrew or licked the hindpaw. The lowest force to produced a withdrawal response was considered to be the threshold. Three repeated measurements were made at intervals of approximately 5 minutes and were averaged for each hindpaw at each time-point 78,79 . Statistical analysis. For the concentration-response curve analysis, the parameters, i.e., the minimum, maximum (E max ), median effective concentration (EC 50 ) and Hill coefficient (n) were calculated from individual concentration-response curves. To determine the parameters of the concentration-response curves, values of response (Y) were fitted by non-linear least-squares curves to the relation Y = a + bx, where x = (C) n / (EC 50 n + (C) n ), to give the value of EC 50 and b (E max ) yielding a minimum residual sum of squares of deviations from the theoretical curve 80 .
Data were presented as means ± standard error of the mean (SEM), and there were no missing data in this study. Statistical significance was evaluated using unpaired and two-tailed Student's t-test or a one-way or two-way repeated measured analysis of variance (ANOVA) with the 5.01 Prism program (GraphPad Software Inc., San Diego, CA, USA). This was followed by a post-hoc Student-Newman-Keuls test when a statistically significant drug (dose) effect was observed (for the one-way ANOVA, the factor was drug (dose); for the two-way ANOVA, the factors were drug (dose), time, and their interaction). Probability values were two-tailed, and the statistical significant criterion P value was set to be 0.05.