mGluR5/ERK signaling regulated the phosphorylation and function of glycine receptor α1ins subunit in spinal dorsal horn of mice

Inhibitory glycinergic transmission in adult spinal cord is primarily mediated by glycine receptors (GlyRs) containing the α1 subunit. Here, we found that α1ins, a longer α1 variant with 8 amino acids inserted into the intracellular large loop (IL) between transmembrane (TM)3 and TM4 domains, was expressed in the dorsal horn of the spinal cord, distributed at inhibitory synapses, and engaged in negative control over nociceptive signal transduction. Activation of metabotropic glutamate receptor 5 (mGluR5) specifically suppressed α1ins-mediated glycinergic transmission and evoked pain sensitization. Extracellular signal-regulated kinase (ERK) was critical for mGluR5 to inhibit α1ins. By binding to a D-docking site created by the 8-amino–acid insert within the TM3–TM4 loop of α1ins, the active ERK catalyzed α1ins phosphorylation at Ser380, which favored α1ins ubiquitination at Lys379 and led to α1ins endocytosis. Disruption of ERK interaction with α1ins blocked Ser380 phosphorylation, potentiated glycinergic synaptic currents, and alleviated inflammatory and neuropathic pain. These data thus unraveled a novel, to our knowledge, mechanism for the activity-dependent regulation of glycinergic neurotransmission.


Introduction
Glycine receptors (GlyRs) are ligand-gated chloride channels that mediate fast inhibitory synaptic transmission in the spinal cord, retina, brain stem, and other brain regions [1,2]. Glycinergic inhibition plays an important role in the modification of motor and sensory functions [1][2][3]. Four α subunits (α1-α4) and one β subunit have, to date, been identified to form pentameric GlyRs [4]. These subunits display spatiotemporal distribution and assign distinct biological properties to GlyRs. The protein level of the α2 subunit in the spinal cord declines rapidly after birth [2,4]. In adulthood, the vast majority of glycinergic transmission is generated by those GlyRs containing the α1 subunit [4,5]. The GlyR α3 subunit also contributes to spinal glycinergic transmission, albeit to a lesser extent than α1 [5]. Alternative  two variants for each α subunit [2]. α1 ins is a longer variant of the α1 subunit with 8 amino acids inserted into the intracellular large loop (IL) between transmembrane (TM)3 and TM4 domains [6]. α1 ins is expressed in the adult spinal cord and brain stem and is estimated to account for more than 30% of total α1 subunit [6]. However, the functional significance of this splice variant remains to be elucidated. Glycinergic transmission in the spinal cord dorsal horn negatively controls the excitability and responsiveness of nociceptive neurons. The reduced glycinergic inhibition is widely considered as a key contributor to central sensitization of nociceptive behaviors [7]. The activitydependent regulation of GlyRs number on plasma membrane and postsynaptic sites represents an important way to modulate inhibitory synaptic strength and plasticity [8]. The intracellular trafficking process of GlyRs is precisely controlled by posttranslational modifications. Protein kinases have been shown to phosphorylate GlyRs [9][10][11], which either depresses or boosts glycinergic currents through mechanisms involving the altered receptor endocytosis, exocytosis, and lateral diffusion on plasma membrane [9][10][11][12]. In addition to phosphorylation, ubiquitination has also been implicated in the modification of endocytosis and surface expression of GlyRs [13].
Group I metabotropic glutamate receptors (mGluRs) are Gq/11-protein-coupled receptors that include two subtypes, mGluR1 and mGluR5. Activation of Group I mGluRs triggers multiple intracellular signaling cascades and causes long-lasting changes of excitatory and inhibitory synaptic transmission in several brain regions [14]. mGluR-dependent synaptic plasticity correlates with a number of neuropsychiatric disorders such as depression, anxiety, schizophrenia, and pathological pain [14]. Here, we found that brief stimulation of mGluR5 in spinal cord dorsal horn suppressed glycinergic transmission through extracellular signal-regulated kinase (ERK)-dependent phosphorylation and ubiquitination of the GlyR α1 ins variant. We provided evidence that the removal of α1 ins -mediated glycinergic inhibition contributed to inflammatory pain.
The regulated synaptic transmission might result from the decrease of presynaptic glycine release or hypofunction of postsynaptic receptors. Previous studies have implicated that prolonged activation (>10 min) of mGluR1/5 by DHPG can stimulate spinal excitatory interneurons to produce endocannabinoids, which act as retrograde messengers to stimulate Type-1 cannabinoid (CB1) receptors expressed at inhibitory nerve terminals [18]. CB1 receptor activation reduces presynaptic glycine release and inhibits GlyR-IPSCs [18]. Here, we found that pretreatment with CB1 receptor antagonist AM251 (5 μM) for 30 min did not block the reduction of GlyR-IPSCs caused by brief DHPG application (S1 Fig), suggesting that short DHPG treatment was not sufficient to produce endocannabinoids. Rather, our data showed that postsynaptic loading of U-0126 or PD98059 eliminated the inhibitory effect of mGluR5 (Fig 1F), implicating a postsynaptic origin. To verify this result, we elicited the whole-cell glycinergic currents by a brief puff (5 s) of exogenous glycine (1 mM) onto the recorded neurons ( Fig 1G). Extracellular application of DHPG for 3 min noticeably decreased the amplitudes of GlyR currents (Fig 1G), which were blocked by intracellularly loaded U-0126 ( Fig 1G). Comparison of paired-pulse ratios of GlyR-IPSCs revealed no significant changes before and after DHPG exposure (Fig 1H), suggesting that mGluR5/ERK signaling suppressed the function or number of GlyRs on postsynaptic membrane.

GlyR α1 ins subunit was the specific target for mGluR5 regulation
Glycinergic inhibition in the adult spinal cord is predominantly produced by GlyRs with α1 subunit [2,5]. Alternative splicing of GLRA1 transcript can generate a longer variant, α1 ins [6]. To test which of the functional GlyRs subunits was regulated by DHPG, we recorded glycineactivated whole-cell currents in human embryonic kidney (HEK)293T cells coexpressing individual GlyR subunit and mGluR5a. Puffer application of glycine (1 mM, 10 ms) evoked robust membrane currents in cells expressing either α1 or α1 ins (Fig 2A). Activation of mGluR5a by DHPG did not affect the peak currents mediated by α1 (Fig 2A). However, α1 ins currents were potently inhibited by DHPG (Fig 2A). As was seen in spinal slices, intracellular loading of MEK inhibitor U-0126 (5 μM; Fig 2B) or PD98059 (20 μM; S2 Fig) blocked DHPG from depressing α1 ins currents. In the presence of PKC inhibitor chelerythrine (10 μM; Fig 2B) or Ro-32-0432 (3 μM; S2 Fig), however, DHPG still caused α1 ins inhibition. The mGluR5/ERK signaling might take effect by inducing α1 ins endocytosis. When the cells were pretreated with dynamin inhibitor dynasore (10 μM), no reduction of α1 ins currents was observed in response to DHPG challenge ( Fig 2B). We also investigated the effect of DHPG on GlyR α3L subunit, a longer α3 variant involved in pain modification [19], and found no reduction of α3L currents after DHPG exposure (Fig 2A). These data suggested that mGluR5 specifically inhibited α1 ins function.

Role of α1 ins in spinal nociceptive processing
To date, little is known about the biological properties of α1 ins . To address this issue, we developed a rabbit antibody against α1 ins (anti-α1 ins ). In transfected HEK293T cells, anti-α1 ins mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function antibody recognized α1 ins but not α1 or α3L (S3A Fig). A western blot also illustrated a specific reaction of the antibody with α1 ins (S3B Fig). By using this custom-made antibody, our data revealed abundant expression of α1 ins in the superficial dorsal horn of the spinal cord ( Fig 3A). A fraction of α1 ins puncta (27.0 ± 1.2%, n = 21 slices) were colocalized with inhibitory synaptic marker gephyrin (Fig 3B), and 25.8 ± 1.5% of inhibitory synapses expressed α1 ins (Fig 3B). Preincubation with excess antigen abolished anti-α1 ins immunosignals (S3B and S3C Fig), confirming the specificity of the antibody.
Next, we designed a short hairpin RNA (shRNA) against mouse α1 ins (shRNA-α1 ins ) to reveal α1 ins function. In transfected HEK293T cells, shRNA-α1 ins specifically reduced the protein level of α1 ins without any influence on that of α1 (S4A Fig). We injected the adeno-associated virus (AAV) encoding green fluorescent protein (GFP) and shRNA-α1 ins in the dorsal horn of mice and recorded GlyR-mediated miniature IPSCs (mIPSCs) in acute slices after 28 days. GFP fluorescence was restricted to the injected side and spread rostrocaudally for about 0.  Fig 3C). The frequencies of GlyR-mIPSCs displayed no significant difference between shRNA-α1 ins -and NC-injected mice (Fig 3C), suggesting a functional involvement of postsynaptic α1 ins in basal glycinergic transmission. Behavioral tests demonstrated that the mice developed mechanical pain hypersensitivity at days 7-28 after shRNA-α1 ins expression ( Fig 3D). A similar sensitization was observed for heat ( Fig 3E) and cold stimuli ( Fig 3F) after α1 ins knockdown. shRNA-α1 ins did not cause motor impairment, as evidenced by the similar performance of NC-and shRNA-α1 ins -injected mice on an accelerating rotarod ( Fig 3G) [3]. We then examined the expression of c-fos, a marker of neuronal activation. Compared to NC, shRNA-α1 ins significantly increased the number of c-fos-positive soma at day 28 post-viral injection (Fig 3H), suggesting that the synaptic inhibition generated by α1 ins was constitutively active in the negative control over neuronal excitability.
To examine whether mGluR5 inhibited glycinergic transmission by specific down-regulation of α1 ins function ex vivo, we recorded GlyRs-IPSCs in neurons expressing shRNA-α1 ins . Knockdown of α1 ins blocked DHPG from reducing the amplitudes of GlyRs-IPSCs (Fig 3I), while NC had no effect on the synaptic depression ( Fig 3I). To confirm the specificity of

ERK interacted with α1 ins
For many kinases, the full access to substrates has been deemed a key step to achieve their specificity in biological regulation. Since mGluR5 inhibited glycinergic responses through ERK, we tested whether this kinase physically interacted with α1 ins . From lysates of transfected HEK293T cells, anti-ERK antibody precipitated α1 ins (Fig 4A) rather than α1 (Fig 4B), suggesting the specificity of ERK binding to α1 ins . Coimmunoprecipitation experiments from spinal cord dorsal horn illustrated that ERK antibody pulled down GlyRs α1 ins and β subunits ( Fig  4C), while α3 was undetectable in ERK precipitates ( Fig 4C). Glutathione S-Transferase (GST) fusion of IL of α1 ins (GST-α1 ins -IL) also precipitated ERK from lysates of spinal cord slices ( Fig 4D). By comparison, GST-fused IL of α1 (GST-α1-IL) did not interact with ERK ( Fig 4D). DHPG (10 μM) treatment of slices for 3 min enabled GST-α1 ins -IL to precipitate more ERK ( Fig 4D). This increased binding was not due to altered ERK expression because there was no change of total ERK protein level after DHPG application ( Fig 4D). Since DHPG increased ERK phosphorylation (Fig 4D), we examined whether α1 ins associated with the active ERK. The results showed that more phosphorylated ERK was pulled down by GST-α1 ins -IL from DHPG-treated slices relative to control slices ( Fig 4D).
To determine whether the phosphorylated ERK (pERK) directly bound to α1 ins , we purified His-tagged phosphorylated ERK2 (His-pERK2) for in vitro GST pull-down assays [20]. GST-α1 ins -IL exhibited a high affinity for His-pERK2 ( Fig 4E). Such an interaction was not observed when GST or GST-α1-IL was incubated with His-pERK2 ( Fig 4E). His-pERK2 also failed to interact with GST-α3L-IL and GST-β-IL (Fig 4E), which harbored the ILs of the α3L and β subunits, respectively. By using nonphosphorylated His-ERK2, we found that none of the GlyR subunits bound to the inactive ERK2 ( Fig 4F).
The α1 ins subunit differed from α1 only by the spliced insert of 8 amino acids in the IL. The specific α1 ins /ERK interaction implicated that the spliced insert might be essential for ERK binding. Many proteins have been shown to interact with ERK through a consensus Ddocking site (R/K-X 2-6 -F-X-F, in which F represents a hydrophobic amino acid) [21]. By analyzing the amino-acid sequence of α1 ins , we found that the insertion of 8 amino acids (SPMLNLFQ) created a putative D-docking site. We therefore tested whether a synthetic peptide (FRRKRRHHKSPMLNLFQE), which encompassed the putative D-docking site, competed with α1 ins for ERK binding. This α1 ins -derived peptide (pep-α1 ins ) was made cell permeable by addition of human immunodeficiency virus-type 1 TAT sequence (referred to as TAT-pep-α1 ins ). In the absence of TAT-pep-α1 ins , GST-α1 ins -IL effectively pulled down His-ERK2 from lysates of HEK293T cells coexpressing constitutively active MEK1(S218D/S222D) mutant ( Fig 4G). Incubation with TAT-pep-α1 ins dose-dependently reduced the content of His-ERK2 pulled down by GST-α1 ins -IL ( Fig 4G). A TAT-fused scrambled peptide (referred to as TAT-Scram) had no effect even at the high dose of 10 μM ( Fig 4G). These data suggested that the D-docking site of α1 ins mediated the binding to ERK. To consolidate this result, we performed coimmunoprecipitation ex vivo. Spinal DHPG (10 nmol, 10 min) application enhanced ERK interaction with α1 ins (Fig 4H). Pretreatment with TAT-pep-α1 ins (200 pmol) for 30 min disturbed α1 ins /ERK interaction, whereas TAT-Scram had no effect ( Fig 4H).

ERK phosphorylated α1 ins at Ser380
Because ERK associated with and inhibited α1 ins , we hypothesized that this kinase might act to phosphorylate α1 ins . In the cytoplasmic region of α1 ins , there are only two serine-proline motifs, 380 SP and 326 SP, which served as the potential phosphorylation sites by proline-directed ERK kinase. To investigate whether these two serine residues regulated α1 ins function, we constructed α1 ins (S380A) and α1 ins (S326A) mutants in which Ser380 and Ser326 were substituted with alanine, respectively. In HEK293T cells, these mutants responded to exogenously applied glycine with large membrane currents ( Fig 5A). DHPG stimulation of coexpressed mGluR5a reduced the peak amplitudes of α1 ins (S326A) (Fig 5A). In α1 ins (S380A)-expressing cells, mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function however, the inhibitory effect of DHPG was blocked (Fig 5A), suggesting the importance of Ser380 in the regulation of α1 ins currents.
To test whether ERK was the kinase that catalyzed Ser380 phosphorylation, we conducted the kinase assays in vitro. In the absence of active ERK, pS380-Ab did not detect any phosphorylation signal on GST-α1 ins -IL (Fig 5D). Incubation with purified His-pERK2 noticeably elevated the phosphorylation level of GST-α1 ins -IL (Fig 5D). This phosphorylation was eliminated by Ser380 mutation to alanine (Fig 5D). The addition of TAT-pep-α1 ins in the reaction buffer also inhibited GST-α1 ins -IL phosphorylation in a dose-dependent manner (Fig 5E), implying that Ser380 was the amino-acid residue that was directly phosphorylated by ERK. We then examined α1 ins phosphorylation in spinal dorsal horn. The protein band recognized by pS380-Ab ex vivo predominantly migrated at 55 kDa (Fig 5F). To confirm the specificity of pS380-Ab, we incubated the antibody with excess immunizing antigen before western blot or treating the immunoblots with alkaline phosphatase, finding that both of them erased pS380 signals (Fig 5F). Intrathecal application of DHPG enhanced pS380-Ab signals ( Fig 5F) and meanwhile evoked spontaneous nociceptive behaviors (Fig 5G). Pretreatment with TAT-pep-α1 ins not only inhibited Ser380 phosphorylation induced by DHPG (Fig 5F) but also alleviated the nociceptive behaviors ( Fig 5G). We found that knockdown of α1 ins by intraspinal injection of AAV encoding shRNA-α1 ins mimicked DHPG by eliciting spontaneous pain behaviors ( Fig 5H). It was noteworthy that intrathecal application of DHPG at day 21 post-viral injection enhanced the nociceptive behaviors to a lesser degree in shRNA-α1 ins mice than in NC-injected mice ( Fig 5H). The maximum nociceptive responses observed at 5 min post-DHPG were only 183.2 ± 35.2% of pre-DHPG values in shRNA-α1 ins mice compared to 915.5 ± 301.2% in NC mice (p = 0.002 relative to shRNA-α1 ins mice, Mann-Whitney U test, n = 6 mice/group), suggesting that the painful responses elicited by shRNA-α1 ins partially occluded DHPG action. These results confirmed that down-regulation of α1 ins function was one of the important ways for mGluR5 to sensitize the nociceptive behaviors.

Ser380 phosphorylation induced the ubiquitination and endocytosis of α1 ins
As mentioned above, the polypeptide detected by pS380-Ab migrated at 55 kDa in the spinal dorsal horn compared to 48 kDa of intact α1 ins (Fig 5F). Previous studies have indicated that GlyR α1 subunit can be ubiquitinated when expressed in Xenopus oocytes [13], which causes about a 7-kDa upward mobility shift. Our data showed that spinal α1 ins was also a ubiquitinated protein, with the apparent molecular size of 55 kDa ( Fig 6A). Ubiquitin (Ubi) modification of α1 ins was activity-dependent. Stimulation of mGluR5 significantly enhanced α1 ins ubiquitination level (Fig 6A). When ERK was inhibited by U-0126, DHPG-induced α1 ins ubiquitination was substantially repressed (Fig 6A). A similar inhibition was also observed when TAT-pep-α1 ins was used to disturb ERK/α1 ins interaction ( Fig 6A). These data raised the possibility that Ser380 phosphorylation might favor α1 ins ubiquitination. To test this, we expressed Myc-α1 ins , Myc-α1 ins (S380A), or Myc-α1 ins (S380D) in neurons along with GlyR β subunit. The nonphosphorylatable Myc-α1 ins (S380A) inhibited, whereas phospho-mimicking Myc-α1 ins (S380D) occluded, the stimulatory effect of DHPG on Myc-α1 ins ubiquitination (Fig 6B), suggesting that Ser380 phosphorylation facilitated the conjugation of Ubi to α1 ins .
There are 10 potential ubiquitination sites (lysine residues) within the IL of α1 ins . We used DHPG to treat neurons expressing Myc-α1 ins /β subunits and immunoprecipitated Myc protein for mass spectrometry. This method identified Lys379 as a ubiquitinated site on Myc-α1 ins (Fig 6C). Mutation of Lys379 to arginine attenuated the basal ubiquitination of Myc-α1 ins (Fig 6D) and blocked DHPG from enhancing α1 ins ubiquitination level (Fig 6D). Lys379 mutation did not eliminate the Ubi signal completely (Fig 6D), suggesting that Lys379 was the major but not the sole site for Ubi modification.
The ubiquitinated cargos on plasma membrane can be recognized by endocytic machinery, which is a critical step for the initiation of endocytic process. By performing coimmunoprecipitation in cultured neurons, we found a physical interaction of Myc-α1 ins with epidermal growth factor receptor substrate 15 (Eps15) (Fig 6E), one of the key endocytic components that recruit the ubiquitinated proteins [23]. Importantly, the molecular interaction of Myc-α1 ins with Eps15 was regulated by Ser380 phosphorylation. Compared to Myc-α1 ins , Myc-α1 ins (S380A) pulled down less Eps15 (Fig 6E). In contrast, the Eps15 contents precipitated by Myc-α1 ins (S380D) were higher than those by Myc-α1 ins (Fig 6E). When Lys379 was mutated to arginine, the interaction between Myc-α1 ins (S380D) and Eps15 was attenuated (Fig 6E). Immunocytochemical analysis showed that shRNA knockdown of Eps15 (S6A Fig) prevented the decrease of surface Myc-α1 ins expression induced by DHPG (S6B Fig). To directly examine whether Ser380 phosphorylation led to α1 ins endocytosis, we transfected Myc-α1 ins or Myc-α1 ins (S380A) along with GlyR β subunit in neurons. DHPG induced a marked increase of internalized Myc-α1 ins immunoreactivity when compared to media control (Fig 6F and 6H). There was no difference in the immunofluorescence intensities of internalized Myc- mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function α1 ins (S380A) between control and DHPG-treated cells (Fig 6G and 6H). These data suggested that mGluR5 caused α1 ins endocytosis through Ser380 phosphorylation.

Potentiation of glycinergic neurotransmission by TAT-pep-α1 ins alleviated pathological pain
Glycinergic disinhibition following peripheral injury plays an important role in central sensitization of nociceptive behaviors. To examine the role of α1 ins in inflammatory pain, we injected formalin into left hindpaws of mice. Our data showed that Ser380 phosphorylation was significantly enhanced in the dorsal horns ipsilateral to formalin injection relative to contralateral sides (Fig 7A). Coincident with Ser380 phosphorylation was a robust α1 ins ubiquitination on the injured sides (Fig 7B). The α1 ins phosphorylation ( Fig 7A) and ubiquitination (Fig 7B) were inhibited when mGluR5 inhibitor MPEP (50 nmol) or ERK inhibitor U-0126 (2 nmol) was intrathecally applied for 30 min before formalin injection. Disruption of ERK/α1 ins interaction by TAT-pep-α1 ins (200 pmol) also blunted α1 ins phosphorylation ( Fig 7A) and ubiquitination ( Fig 7B) in formalin mice. The same dose of TAT-Scram, however, had no effect ( Fig  7A and 7B). Formalin triggers biphasic spontaneous pain behaviors (Fig 7C). We found that TAT-pep-α1 ins dose-dependently ameliorated the second-phase painful behaviors (Fig 7C and  7E). Compared to saline control, TAT-Scram had no effect on the second-phase responses (Fig 7C and 7E). The first phase was similar among saline-, TAT-Scram-, and TAT-pep-α1 instreated mice (Fig 7C and 7D). These data suggested that mGluR5/ERK-signaling-dependent α1 ins phosphorylation and ubiquitination closely correlated with inflammatory pain. We also tested the effect of TAT-pep-α1 ins (200 pmol) on the neuropathic pain induced by spared nerve injury. The results illustrated that interference with ERK/α1 ins interaction alleviated the mechanical allodynia (  To test whether disruption of ERK/α1 ins interaction potentiated glycinergic transmission in inflamed mice, we recorded GlyR-IPSCs in spinal slices prepared at 90 min post-formalin injection. The results showed that bath application of TAT-pep-α1 ins , but not TAT-Scram, enhanced the amplitudes of GlyR-IPSCs (Fig 7F). If α1 ins was virally knocked down before formalin injection, the synaptic potentiation by TAT-pep-α1 ins was eliminated ( In intact mice, glycinergic transmission was also insensitive to TAT-pep-α1 ins (S8F Fig), possibly because of low ERK activity at resting conditions. These data suggested that specific reinstatement of α1 ins -mediated glycinergic transmission attenuated inflammatory pain.

Discussion
GlyR α1 subunit is abundantly expressed in the adult spinal cord and brain stem and is responsible for the majority of glycinergic neurotransmission [2,5]. The importance of α1 in neuronal function is underlined by genetic studies showing that mutation or reduced expression of α1 in the spinal cord and brain stem incurs severe neurological disorders such as hyperekplexia [24-26]. The major finding in the current study was that α1 ins , a longer α1 variant, was involved in spinal glycinergic transmission. The α1 ins expression was restricted to the superficial dorsal horn of spinal cord, and knockdown of α1 ins did not affect motor coordination but elicited pain hypersensitivity, suggesting that α1 ins was required for spinal nociceptive processing. Activation of mGluR5 in the spinal cord dorsal horn has been shown to enhance nociceptive neuronal excitability, potentiate glutamatergic inputs, and play an important role in pathological pain [27-31]. We provided evidence that mGluR5 activation also attenuated glycinergic currents through ERK-dependent α1 ins phosphorylation. This novel, to our knowledge, signaling pathway might act synergistically with the enhanced glutamatergic transmission and neuronal excitability to sensitize the nociceptive behaviors. In support of this notion, α1 ins knockdown partially occluded DHPG action in inducing painful responses, and, more importantly, specific recovery of α1 ins -mediated glycinergic inhibition attenuated pathological pain.
The IL between TM3-TM4 of GlyRs subunits is a unique domain that displays the highest degree of variability. These loops bind to intracellular scaffolds and signaling components that are essential for the activity-dependent modification of glycinergic efficacy. Great efforts have been made to distinguish the biological characteristics of two GlyR α3 subunits, α3K and the longer α3L that contains additional 15 amino acids in the TM3-TM4 loop [32]. These two α3 variants are distributed throughout the central nervous system. However, they differ significantly in terms of channel desensitization, membrane distribution, and synaptic localization [33-36]. The alternatively spliced insert in α3L has been proposed to stabilize the secondary structure of the TM3-TM4 loop and regulate the channel gating [34]. The current study demonstrated that one of the important functions of the α1 ins insert was to constitute a D-docking site, a unique structure that allowed mGluR5/ERK signaling to decrease α1 ins -mediated glycinergic inhibition and evoke nociceptive behavioral sensitization. The deficiency of the spliced insert and corresponding D-docking site explained why α1 subunit was refractory to mGluR5 regulation. As with mGluR5, the adenosine A1 receptor has also been shown to regulate glycinergic responses mediated by α1 ins but not by α1 [15]. These data implied that α1 and α1 ins might respond distinctly to some G-protein-coupled receptors despite their high similarities in electrophysiological and pharmacological properties [6]. mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function Phosphorylation plays an important role in the dynamic modulation of ligand-gated ion channels. Most of the known phosphorylation sites on GlyRs have been mapped to the TM3-TM4 loop. The cyclic adenosine monophosphate (cAMP)-dependent protein kinase (PKA) can suppress glycinergic transmission by phosphorylating α3L at Ser346, which is involved in the regulation of nociceptive behaviors and respiratory rhythm [19,37]. Ser346 phosphorylation also enables allosteric modulators to interact with α3L and reverse inflammation-induced glycinergic disinhibition in the spinal cord [38]. PKC phosphorylation of GlyR β subunit disturbs the binding of the postsynaptic scaffold protein gephyrin and decreases GlyR accumulation at inhibitory synapses [11]. mGluR5 has been well known to regulate a wide range of cellular responses through PKC pathway. We tested the role of PKC in mGluR5 modification of GlyRs, finding that glycinergic responses were insensitive, at least in part, to PKC signaling initiated specifically by mGluR5. Previous studies have demonstrated that mGluR5 can stimulate a transient activation of ERK through protein phosphatase 2A and Homer1b/c pathways [16,17]. Our data revealed that ERK was essential for mGluR5 to suppress glycinergic currents. Upon binding to the D-docking site of α1 ins , ERK catalyzed the phosphorylation of Ser380 and led to the endocytosis of α1 ins . The activity-dependent Ser380 phosphorylation of α1 ins thus constituted a novel, to our knowledge, mechanism to modulate the efficacy of glycinergic transmission. We found that Ser380 phosphorylation was closely associated with inflammatory pain. Interference with ERK/α1 ins interaction decreased Ser380 phosphorylation in inflamed mice, resumed glycinergic transmission, and attenuated pain sensitization. The spliced insert also introduced a serine-proline motif ( 326 SP) in the TM3-TM4 loop of α1 ins . However, this SP motif was found to be dispensable for mGluR5 inhibition of α1 ins currents.
Our data showed that α1 ins was a ubiquitinated protein, and Lys379 was identified as the major ubiquitination site. The ubiquitination decreased the band mobility of α1 ins by about 7 kDa, suggesting that only one Ubi molecule was conjugated to one α1 ins subunit. Monoubiquitination generally regulates protein-protein interaction and the endocytic process [39]. The Ubi moieties on α1 ins were recognized by endocytic adaptor protein Eps15, which initiated α1 ins endocytosis to remove glycinergic inhibition [39]. Importantly, activation of mGluR5/ ERK signaling or Ser380 mutation to aspartic acid enhanced the ubiquitination level of α1 ins , indicating that α1 ins ubiquitination was regulated by Ser380 phosphorylation. Possibly, Ser380 phosphorylation facilitated the interaction of α1 ins with unidentified ubiquitination machinery that catalyzed the Ubi transfer cascade. Alternatively, Ser380 phosphorylation caused the conformational change in the intracellular large loop of α1 ins so that lysine residues became susceptible for Ubi conjugation reaction. Recently, HECT, UBA, WWE domain containing 1 (HUWE1) is identified as the E3 Ubi ligase that contributes to the ubiquitination and endocytosis of spinal GlyR α1 subunit during inflammatory pain [40]. HUWE1 knockdown enhances glycinergic transmission and generates an effective analgesic action against pain hypersensitivity [40]. The E3 Ubi ligases that ubiquitinate α1 ins subunit require further investigation.
Taken together, the current study demonstrated that GlyR α1 ins subunit served as a specific target for mGluR5/ERK signaling to reduce glycinergic inhibition and evoke spinal sensitization. Given the gating control by glycinergic inhibition over nociceptive sensory input through the spinal cord dorsal horn to higher brain regions, these data shed new light on a potential for α1 ins to treat pathological pain.

Ethics statement
The animal experiments were conducted in accordance with the guidelines of the Animal Care and Use Committee of Lanzhou University. The male C57BL/6J mice (10-12 weeks old) were purchased from the Experimental Animal Center of Lanzhou University (approval number: SCXK(GAN)-2013-0002) and selected randomly in each experiment. These animals were housed two to three per cage with free access to food and water on a 12 h light/dark cycle. Every effort was made to minimize the number and suffering of animals.

Animal models and drug delivery
Formalin (1.5%, 20 μl) was injected subcutaneously into the plantar surfaces of hindpaws. For spared nerve injury [41], the mouse was anaesthetized with sodium pentobarbital (60-90 mg/ kg, i.p.), and the skin of the left lateral thigh was incised to expose the sciatic nerve. After careful separation of the three nerve branches, the tibial and common peroneal nerves were ligated with 5.0 silk and transected, followed by removing a 2-3 mm portion of the tibial and common peroneal nerves distal of the ligation. Every effort was made to keep the sural nerve intact during the operation. The muscle and skin were then closed in layers. Intrathecal injection (5 μl) was achieved by direct lumbar puncture as described previously [20]. Intraspinal viral injection was conducted in sodium pentobarbital (60-90 mg/kg, i.p.)-anesthetized mice [42]. Briefly, the animals were immobilized on a stereotaxic frame after a laminectomy. A glass pipette attached to a 5-μl microsyringe was used to inject the viral vectors (30 nl/min) at a depth of mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function 0.2-0.3 mm from the dorsal surface of lumbar segment and 0.5 mm apart from the midline. After the injection, the muscle and skin were closed.

Cell cultures and transfection
The HEK293T cells were plated onto poly-D-lysine (0.1 mg/ml)-coated coverslips, maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin, and grown at 37˚C. The spinal cord neurons were cultured as previously described [43]. In brief, the mouse pups (postnatal day 1) were decapitated, and the spinal cords were removed into ice-cold Hank's Balanced Salt Solution containing 10 mM HEPES. After careful removal of all meninges, the dorsal quadrants of spinal cords were dissected out, chopped into small strips, and digested by papain (2 mg/ml) for 20-30 min at 37˚C. DMEM with 10% heat-inactivated fetal bovine serum was added to terminate the digestion reaction. After trituration, the cells were harvested by centrifugation at 1,000 × g and resuspended in neurobasal medium containing 2% fetal bovine serum, 2% heat-inactivated horse serum, 2% B27, 1% penicillin/streptomycin, and 2 mM L-glutamine. The neurons were plated onto poly-D-lysine-coated coverslips with the cell density adjusted to be 1.5 × 10 6 . The cultured cells were transfected with Lipo 6000 Transfection Reagent (Beyotime Institute of Biotechnology, Jiangsu, China) according to the manufacturer's instructions.

GST pull-down assay
GST-fused proteins were expressed in Escherichia coli BL21 cells and affinity purified with glutathione agarose beads (Sigma-Aldrich) [20]. His6-tagged recombinant proteins were purified with nickel-nitrilotriacetic acid column (Roche, Indianapolis, IN, USA) from lysates of transfected HEK293T cells and eluted by 0.25 M imidazole, 300 mM NaCl, and 50 mM NaH 2 PO 4 (pH 8.0) [20]. Amicon Ultra Centrifugal Filters (Millipore, Burlington, MA, USA) were used to concentrate and desalt the eluted His proteins. The His-tagged phosphorylated ERK2 was purified from lysates of HEK293T cells coexpressing MEK1(S218D/S222D) and His-ERK2 [20]. The purity of His protein was assessed by western blot and Coomassie blue staining. For GST pull-down, the purified His proteins (0.5 μM) or lysates (200 μg) of spinal dorsal horn or HEK293T cells were incubated with glutathione-agarose-bead-bound GST proteins in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris�HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 1.0% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, and inhibitor cocktail of phosphatases and proteases). After gentle rotation for 4 h at 4˚C, the beads were collected by brief centrifugation at 800 × g, washed for six times with RIPA buffer, and boiled in SDS sample buffer. Different concentrations of peptides were preincubated with His proteins for 1 h before pulldown assays.

LC MS/MS analysis
GST proteins purified by glutathione agarose beads or the immunoprecipitated Myc-α1 ins from neurons were separated by SDS-PAGE. The corresponding protein band was then excised and cut into 1-mm pieces. After in-gel digestion with trypsin (10 ng/μl) at 37˚C overnight, the peptides were extracted with 50% acetonitrile/0.1% trifluoroacetic acid and dried. The tryptic peptides were dissolved in 0.1% formic acid (solvent A); loaded onto a 5-cm-long, 75-μm-inner-diameter trap column packed with 5-μm C18 stationary phase; and separated by 15-cm-long, 75-μm-inner-diameter analytical column packed with 2-μm C18 stationary phase. The gradient was comprised of 5%-35% solvent B (0.1% formic acid in 80% acetonitrile) for 60 min, 35%-80% solvent B for 20 min, and 100% solvent B for 10 min at a constant flow rate of 300 nl/min on an EASY-nLC 1200 UPLC system (Thermo Fisher Scientific, Waltham, MA, USA). The eluted peptides were subjected to Thermo Scientific Obritrap Fusion Lumos Tribrid mass spectrometer. The electrospray voltage was 2.5 kV. The mass spectrometer was operated in the data-dependent mode, with a survey scan over an m/z range of 300-1,800 at a resolution of 120,000 in the Orbitrap. Data were processed using the Proteome Discoverer 2.1 software package (Thermo Fisher Scientific). Tandem mass spectra were searched against the amino-acid sequence of GST-α1 ins -IL or Myc-α1 ins . Trypsin was specified as the cleavage enzyme, allowing up to 2 missing cleavages. Mass error was set to 10 ppm for mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function precursor ions and 0.02 Da for fragment ions. Serine/threonine phosphorylation and lysine ubiquitination were allowed as variable modifications.

Immunohistochemistry
The mice were anesthetized with sodium pentobarbital at day 28 after viral injection and perfused through the ascending aorta with 4% paraformaldehyde in phosphate-buffered saline (PBS; 0.01 M). The lumbar enlargements of spinal cords were dissected out, fixed in the same fixative for 4 h, and cryoprotected in 30% sucrose overnight. The transverse or sagittal sections (16 μm) were cut on a cryostat, blocked with 10% normal goat serum (NGS) and 0.1% Triton X-100 in PBS for 12 h at 4˚C, and incubated with mouse anti-GFP or rabbit anti-c-fos antibody (Proteintech) at 4˚C for 72 h. To assay the synaptic distribution of α1 ins , the transverse slices of 2-mm thickness were cut on a chopper at 4˚C and fixed with 4% paraformaldehyde for 30 min before cryoprotection [19,45]. The transverse slices (16 μm) were blocked and incubated with rabbit anti-α1 ins and mouse anti-gephyrin antibody (Synaptic System) at 4˚C for 12 h. After five washes with PBS, the slices were incubated with Alexa Fluor 488-and Cy3-conjugated secondary antibodies for 1 h before image capture with a confocal laser scanning microscope (FV1000; Olympus, Tokyo, Japan).

Immunocytochemistry
HEK293T cells were transfected with Myc-α1 ins , Myc-α1, or Myc-α3L. At 48 h after transfection, the cells were fixed with 4% paraformaldehyde and 4% sucrose in PBS for 15 min. After three washes with PBS, the cells were permeabilized in PBS containing 0.25% Triton X-100 for 15 min and blocked with 10% NGS in PBS overnight at 4˚C. GlyRs were labeled by anti-α1 ins antibody for 2 h at 4˚C. After five washes with PBS, the cells were stained for 1 h at room temperature with Cy3-conjugated secondary antibodies in 10% NGS-containing PBS.
The cultured spinal neurons at 10-12 days in vitro were co-transfected with shRNA-Eps15 or NC along with Myc-α1 ins and GlyR β subunit (β:α = 50:1) [46]. At 72 h after transfection, the neurons were treated with DHPG in culture media for 3 min at 37˚C. After washing DHPG out for three times with prewarmed culture media, the neurons were incubated in the media for 10 min at 37˚C, followed by washes with PBS containing 4% sucrose and fixation with 4% paraformaldehyde and 4% sucrose in PBS for 20 min. Surface receptors were labeled by mouse anti-Myc antibody for 2 h at room temperature. The neurons were washed with PBS containing 4% sucrose and blocked in PBS containing 0.25% Triton X-100 and 10% NGS for 30 min. Neurons were then stained with chicken anti-MAP2 antibody (Novus Biologicals, Littleton, CO, USA). After five washes with PBS, surface Myc proteins and MAP2 were visualized by incubation for 1 h at room temperature with Cy3-conjugated goat anti-mouse and Alexa Fluor 405-conjugated goat anti-chicken secondary antibody.
To assay the endocytosis, the cultured spinal neurons at 10-12 days in vitro were transfected with GlyR β subunit and Myc-α1 ins or Myc-α1 ins (S380A). At 48 h after transfection, the surface-bound receptors were labeled with mouse anti-Myc antibody for 2 h at 4˚C. The neurons were washed 3 times with prewarmed culture media and treated with DHPG for 3 min at 37˚C. After washing DHPG out with prewarmed culture media, the neurons were maintained in the incubator for 10 min. Thereafter, the neurons were washed twice with ice-cold PBS and fixed with PBS containing 4% paraformaldehyde and 4% sucrose for 20 min at room temperature. The remaining surface anti-Myc antibody was visualized by staining with Alexa Fluor 488-conjugated anti-mouse IgG at room temperature for 1 h. The cells were then permeabilized and blocked in PBS containing 0.25% Triton X-100 and 10% NGS for 30 min. The cells were stained with chicken anti-MAP2 antibody. After five washes with PBS, the internalized mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function anti-Myc antibody and MAP2 were visualized by staining with Cy3-conjugated anti-mouse and Alexa Fluor 405-conjugated anti-chicken IgG before image capture. The internalization signals were divided by the signals of surface plus internalized receptors.

Behavioral tests
The animals were acclimatized to the testing environment for at least 1 h before intraplantar formalin or intrathecal DHPG injection. Immediately after the injection, we returned the animals to the chamber and observed the spontaneous pain behaviors for 0.5-1 h [47,48]. For the Von Frey test, a set of calibrated Von Frey monofilaments (Stoelting, Wood Dale, IL, USA) were applied perpendicularly to the plantar surfaces of hindpaws. The pattern of positive and negative withdrawal responses was converted to 50% paw withdrawal thresholds by using the up-down method [20]. The paw withdrawal latencies (PWLs) were measured by delivering a beam of light onto the plantar surfaces of hindpaws (with the cutoff of 10 s). The time between the onset of heat application and paw withdrawal was recorded automatically as PWL values [20]. Cold stimulation was delivered by dabbing acetone onto the plantar surfaces of hindpaws. The first 10-second activities were excluded, and the time spent on flicking and licking the paws for 60 s afterwards was recorded [49]. Motor function was tested on a rotarod that was accelerated from 0 to 40 RPM within 60 s. After training for two days, the mice were tested on the rod for three times, and the maximum RPM that caused the mice to fall was averaged [3].

Electrophysiological recordings
The mice (6-8 weeks old) were anesthetized with sodium pentobarbital, and the lumbar segment of spinal cord was isolated into ice-cold sucrose solution (in mM: 50.0 sucrose, 95.0 NaCl, 1.8 KCl, 0.5 CaCl 2 , 7.0 MgSO 4 , 1.2 NaH 2 PO 4 , 26.0 NaHCO 3 , 15.0 D-glucose [pH 7.4], bubbled with 95% O 2 + 5% CO 2 ). A transverse slice (350-μm thickness) with an intact L4 or L5 dorsal root was cut on a vibratome stage and perfused (5 ml/min) with oxygenated ACSF (32˚C-33˚C) in the recording chamber for at least 1 h before recordings. The lamina II outer neurons were visually identified under an Olympus BX51WIF microscope fitted with a 40× water immersion objective under fluorescence and transmitted light illumination. We performed the recordings on lamina II outer neurons because these neurons receive the inputs from unmyelinated peptidergic C nociceptors and myelinated Aδ nociceptors as well as from spinal glycinergic inhibitory interneurons [3,[50][51][52]. The reduced glycinergic inhibition contributes to the development of pathological pain [3,50]. The glass electrodes had the resistance The neurons were voltage-clamped at 0 mV with an Axon 700B amplifier (Molecular Devices, San Jose, CA, USA). To evoke GlyR-IPSCs, focal stimulation (0.1 Hz) was delivered through a glass pipette that was positioned adjacent to the recorded neurons [15]. The glycinergic component was pharmacologically isolated by adding bicuculline (10 μM), D-APV (50 μM), and CNQX (20 μM) in the external solution. The paired-pulse ratios were recorded and measured by delivering two successive electric stimuli at a 30-ms interval. For mIPSCs, tetrodotoxin (1 μM) was also included in the bath solution. The GABAergic IPSCs were isolated by strychnine (2 μM), D-APV, and CNQX. To record the dorsal-root-evoked EPSCs, the glass pipettes were filled with (in mM) 115 cesium methanesulfonate, 20 CsCl, 10 HEPES, 2.5 MgCl 2 , 4.0 Na 2 ATP, 0.4 Na-GTP, 0.6 EGTA, and 10 sodium phosphocreatine (pH 7.25; 295-300 mOsm). NMDAR-EPSCs were recorded at +40 mV in the presence of bicuculline, strychnine, and CNQX. The AMPAR-EPSCs were recorded at −70 mV in the presence of bicuculline and strychnine. The monosynaptic EPSCs were identified on the basis of the constant latency and mGluR5/ERK signaling inhibited glycine receptor α1 ins subunit function absence of conduction failure in response to high-frequency electrical stimulation (20 Hz). To evoke the whole-cell glycinergic currents in spinal slices [15], glycine (1 mM, 5 s) was perfused onto the recorded neurons through an electrically controlled microperfusion system. HEK293T cells co-transfected with pcDNA3.1 and pEGFP-N1 vector (reporter plasmid) were perfused at room temperature with the external solution containing (mM) 145.0 NaCl, 5.0 KCl, 2.0 CaCl 2 , 1.0 MgCl 2 , 10.0 HEPES, and 11.0 D-glucose (pH 7.3) [19]. When filled with the internal solution (in mM: 140.0 CsCl, 1.0 CaCl 2 , 2.0 MgCl 2 , 10.0 HEPES, 8.0 EGTA, 3.0 Na-ATP, and 0.1 Na-GTP [pH 7.2]; 295-300 mOsm), the recording pipettes had the resistance of 3-5 MO. The cells were voltage-clamped at −80 mV. To elicit the whole-cell glycinergic currents, glycine (1 mM, 10 ms) was dissolved in the external solution and rapidly applied onto the cells at an interval of 30 s. The series and input resistances were monitored online throughout each experiment. The recordings were collected for analysis unless the resistances changed by more than 15%. The current signals were filtered at 2 kHz and sampled at 5 kHz.