Abstract
α2δ-1 (encoded by the Cacna2d1 gene) is a newly discovered NMDA receptor-interacting protein and is the therapeutic target of gabapentinoids (e.g., gabapentin and pregabalin) frequently used for treating patients with neuropathic pain. Nerve injury causes sustained α2δ-1 upregulation in the dorsal root ganglion (DRG), which promotes NMDA receptor synaptic trafficking and activation in the spinal dorsal horn, a hallmark of chronic neuropathic pain. However, little is known about how nerve injury initiates and maintains the high expression level of α2δ-1 to sustain chronic pain. Here, we show that nerve injury caused histone hyperacetylation and diminished enrichment of histone deacetylase-2 (HDAC2), but not HDAC3, at the Cacna2d1 promoter in the DRG. Strikingly, Hdac2 knockdown or conditional knockout in DRG neurons in male and female mice consistently induced long-lasting mechanical pain hypersensitivity, which was readily reversed by blocking NMDA receptors, inhibiting α2δ-1 with gabapentin or disrupting the α2δ-1–NMDA receptor interaction at the spinal cord level. Hdac2 deletion in DRG neurons increased histone acetylation levels at the Cacna2d1 promoter, upregulated α2δ-1 in the DRG, and potentiated α2δ-1–dependent NMDA receptor activity at primary afferent central terminals in the spinal dorsal horn. Correspondingly, Hdac2 knockdown-induced pain hypersensitivity was blunted in Cacna2d1 knockout mice. Thus, our findings reveal that HDAC2 functions as a pivotal transcriptional repressor of neuropathic pain via constitutively suppressing α2δ-1 expression and ensuing presynaptic NMDA receptor activity in the spinal cord. HDAC2 enrichment levels at the Cacna2d1 promoter in DRG neurons constitute a unique epigenetic mechanism that governs acute-to-chronic pain transition.
SIGNIFICANCE STATEMENT Excess α2δ-1 proteins produced after nerve injury directly interact with glutamate NMDA receptors to potentiate synaptic NMDA receptor activity in the spinal cord, a prominent mechanism of nerve pain. Because α2δ-1 upregulation after nerve injury is long lasting, gabapentinoids relieve pain symptoms only temporarily. Our study demonstrates for the first time the unexpected role of intrinsic HDAC2 activity at the α2δ-1 gene promoter in limiting α2δ-1 gene transcription, NMDA receptor-dependent synaptic plasticity, and chronic pain development after nerve injury. These findings challenge the prevailing view about the role of general HDAC activity in promoting chronic pain. Restoring the repressive HDAC2 function and/or reducing histone acetylation at the α2δ-1 gene promoter in primary sensory neurons could lead to long-lasting relief of nerve pain.
- chromatin
- dorsal root ganglion
- epigenetics
- histone modification
- synaptic plasticity
- transcriptomics
Introduction
Neuropathic pain is a chronic debilitating condition that afflicts millions of people worldwide. Current treatments are unsatisfactory, and our knowledge of the underlying mechanisms governing neuropathic pain development remains limited. Both sustained changes in gene expression in primary sensory neurons and synaptic plasticity in the spinal dorsal horn are integral to the development of chronic neuropathic pain (Woolf and Salter, 2000; Zhou et al., 2011; Laumet et al., 2015). Among many upregulated genes in the dorsal root ganglion (DRG) caused by nerve injury, α2δ-1 (encoded by the Cacna2d1 gene) is the best known and clinically validated target in the treatment of neuropathic pain (Gee et al., 1996; Fuller-Bicer et al., 2009). α2δ-1 mediates the therapeutic effect of gabapentinoids, including gabapentin and pregabalin, which are first-line drugs for treating neuropathic pain. Peripheral nerve injury induces an early and sustained increase in α2δ-1 expression in DRG neurons (Luo et al., 2001; Newton et al., 2001; Zhang et al., 2021). Another hallmark of neuropathic pain is increased glutamatergic input augmented by NMDAR receptor hyperactivity in the spinal dorsal horn (Zhang et al., 2009; Zhou et al., 2012; Chen et al., 2014b; Li et al., 2016; Xie et al., 2017). Previous studies revealed that in neuropathic pain, excess α2δ-1 proteins physically interact with NMDARs to potentiate NMDAR synaptic trafficking and activity in the spinal dorsal horn independently of voltage-gated Ca2+ channels (Chen et al., 2018, 2019; Zhang et al., 2021). However, little is known about how nerve injury initiates and maintains the high expression level of α2δ-1 and associated NMDAR activity to sustain chronic pain.
Chromatin structure and the modification status of histone tails critically control gene transcription (Strahl and Allis, 2000; Jiang et al., 2008), and altered expression levels of pronociceptive and antinociceptive genes in the DRG are causally related to the development of chronic pain after nerve injury. For example, G9a, a histone methyltransferase, facilitates chronic pain development after nerve injury by silencing antinociceptive genes, including genes associated with 40 voltage-gated K+ channels and µ-opioid and cannabinoid receptors (Laumet et al., 2015; Zhang et al., 2016; Luo et al., 2020). Also, nerve injury induces sustained DNA hypomethylation of many genes in the DRG (Garriga et al., 2018). However, altering histone methylation or DNA methylation does not normalize α2δ-1 expression levels in the injured DRG (Laumet et al., 2015; Garriga et al., 2018). Acetylation is another common modification of histones and serves as a crucial epigenetic mechanism for gene transcription. The acetylation status of lysine residues in histone tails is dynamically regulated by the opposing activities of various histone acetyltransferases and histone deacetylases (HDACs). HDAC proteins are grouped into four classes based on function and DNA sequence similarity, and Class I HDAC subtypes (HDAC1, HDAC2, HDAC3, and HDAC8) are located primarily in the nucleus. The functions of Class I HDACs in the nervous system are highly heterogeneous. In the brain, HDAC1 is predominantly expressed in glia, whereas HDAC2 is abundant in mature neurons and is involved in memory formation (Broide et al., 2007; Guan et al., 2009; Hanson et al., 2013). At present, the roles of individual HDAC subtypes in regulating the transcription of Cacna2d1 and the development of neuropathic pain remain unclear.
Here, we present our findings that nerve injury diminishes HDAC2 occupancy at the promoter region of Cacna2d1 in the DRG, which induces histone hyperacetylation and promotes the transcription of Cacna2d1. Strikingly, knockdown or conditional knockout (KO) of Hdac2 in DRG neurons causes a neuropathic pain-like phenotype, which is maintained by α2δ-1–dependent NMDAR activation in the spinal dorsal horn. Thus, our study uncovers HDAC2 as a crucial epigenetic repressor of chronic pain development via constitutive suppression of Cacna2d1 transcription in primary sensory neurons and associated synaptic NMDAR activity in the spinal cord. This information advances our understanding of the epigenetic basis of chronic neuropathic pain.
Materials and Methods
Animal models
All procedures and experimental protocols were approved by the Institutional Animal Care and Use Committee at The University of Texas MD Anderson Cancer Center and conformed to guidelines from the National Institutes of Health on the ethical use of animals. All animals were housed (two to three per cage for rats, and four to five per cage for mice) on a standard 12 h light/dark cycle, maintained in pathogen-free conditions, and received food and water ad libitum. Male Sprague Dawley rats (8–10 weeks old, Envigo) were used for spinal nerve ligation (SNL) to induce neuropathic pain as described previously (Kim and Chung, 1992). Briefly, rats were anesthetized with 2–3% isoflurane, and the left L5 and L6 spinal nerves were exposed and ligated with 6–0 silk sutures separately under a surgical microscope. Sham surgery (the same surgical procedure without nerve ligation) was used as the control.
Cacna2d1 KO mice were originally obtained from the Medical Research Council (stock #6900) and were generated as described previously (Fuller-Bicer et al., 2009). Hdac2 conditional knockout (Hdac2-cKO) mice were generated by crossing female mice carrying LoxP recombination sites flanking exons 5 and 6 of the Hdac2 gene (Hdac2fl/fl; Guan et al., 2009; Wilting et al., 2010) and male mice with Advillin promotor-driven Cre-recombinase expression (AvlCre+/−; X. Zhou et al., 2010; da Silva et al., 2011). AvlCre+/−::Hdac2fl/- mice obtained from the first generation were crossed again with female Hdac2fl/fl mice to obtain AvlCre/+::Hdac2fl/fl (Hdac2-cKO mice).
To ablate the Hdac2 gene from DRG neurons in adult mice, we crossed Hdac2fl/fl mice with tamoxifen-dependent inducible Avl-CreERT2 (stock #032027, The Jackson Laboratory; Lau et al., 2011). Tamoxifen (catalog #T5648, Sigma-Aldrich) was dissolved in corn oil (catalog #C8267, Sigma-Aldrich) and injected intraperitoneally (75 mg/kg/day for 5 consecutive days) to induce Hdac2 deletion in DRG neurons in Avl-CreERT2::Hdac2fl/fl mice (Hdac2-icKO). The Advillin Cre-induced target gene KO occurs in 87% of DRG neurons in Avl-CreERT2 mice (Woo et al., 2014) and in 84% of DRG neurons in the AvlCre line (Zappia et al., 2017). Both male and female adult mice (8–14 weeks of age) were used for electrophysiological and behavioral studies. Data were pooled from males and females because no evident sex difference was detected during the course of our study. After weaning (3 weeks after birth), mice were ear tagged and genotyped using ear tissues with PCR. Littermates without Cre were used as controls (Hdac2fl/fl). Genotyping primers are listed in Table 1.
List of primers used in this study
Intrathecal catheter placement and siRNA treatment
For intrathecal catheter implants, rats were anesthetized with 2–3% isoflurane. A small incision was made at the back of the neck, and a PE-10 catheter (8 cm) was inserted via a small opening made on the atlanto-occipital membrane of the cisterna magna so that the catheter tip reached the lumbar enlargement (Chen and Pan, 2001). The animals were allowed to recover for 5–7 d before intrathecal siRNA injections. Rat Hdac2 siRNA (MISSION® siRNA Product, siRNA ID #SASI-RN02-00300570), rat Hdac3 siRNA (MISSION® siRNA Product, siRNA ID #SASI-RN01-00031908), mouse Hdac2 siRNA (MISSION® siRNA Product, siRNA ID #SASI_Mm01_00100699), and control siRNA (catalog #SIC001, MISSION siRNA Universal Negative Control #1) were obtained from Sigma-Aldrich. The sequence for rat Hdac2 and rat Hdac3 siRNA was GAUAUCGGGAAUUAUUAUU[dT][dT] and GAGUUCUGCUCCCGCUAUA[dT][dT], respectively.
The sequence for mouse Hdac2 siRNA was CUGCUAAAUUAUGGUUUAU[dT][dT]. i-Fect (catalog #NI35150, Neuromics) was used to dissolve and deliver the siRNA via intrathecal injection, as we described previously (Laumet et al., 2015; Zhang et al., 2018). Intrathecal injection in mice was performed using a lumbar puncture technique, as previously described (Huang et al., 2020; Zhang et al., 2021). A mixture of 2 µg of siRNA in 10 µl of i-Fect was intrathecally injected daily for 5 consecutive days.
Nociceptive behavioral tests
For measurement of tactile sensitivity, mice or rats were habituated on a wire-grid panel for 30 min before testing. A series of von Frey filaments (Stoelting) was applied to the plantar surface of the hindpaw. A quick withdrawal or flinching of the paw was considered a positive response. In the absence of a response, we applied the filament of the next greater force. We used the up-down method to calculate the tactile threshold that produced a 50% likelihood of a withdrawal response as previously described (Chaplan et al., 1994).
For measurement of mechanical nociception in rats, we used an analgesiometer (Ugo Basile) to perform the paw pressure test. A foot pedal was pressed to activate a motor that applied a linearly increasing force to the dorsal surface of the hindpaws. The pedal was immediately released when the animal displayed pain by either withdrawing the paw or vocalizing. Each trial was repeated two or three times at 2 min intervals, and the mean value was used as the force needed to produce a withdrawal response (Chen et al., 2014b, 2018). To quantify the pressure withdrawal threshold in mice, we used a Pincher Analgesia Meter (model 2450, IITC Life Science) to apply pressure with increasing force on the midplantar glabrous surface of hindpaws as previously described (Zhang et al., 2018; Jin et al., 2022). Brisk withdrawal or vocalization was considered a positive response and recorded as the withdrawal threshold. The cutoff was set to 400 g to minimize potential tissue injury.
For measurement of heat sensitivity, mice or rats were tested on a Plantar Analgesia Meter (model 400, IITC Life Science). The plantar surface was maintained at 30°C, and animals were habituated for 30 min before each test. A mobile radiant heat stimulus was applied to the plantar surface of the hindpaw until the animal displayed a withdrawal response or paw licking, as described previously (Chen et al., 2014a, b). The cutoff was set to be 30 s to minimize potential tissue injury.
Immunoblotting
Total proteins were extracted from the lumbar DRG and dorsal spinal cord tissues (at L5/L6 levels in rats and L3–L5 levels in mice) using an extraction buffer (50 mm Tris-HCl, pH 7.4, 1% NP-40, 1% sodium deoxycholate, 150 mm NaCl, 1 mm EDTA, 1 mm Na3VO4, and 1 mm NaF in the presence of a proteinase inhibitor cocktail; Sigma-Aldrich). In the same experiments, synaptosomes were prepared from dorsal spinal cords at L3–L5 levels in mice, as described previously (Chen et al., 2018; Chen et al., 2019). In brief, tissues were gently homogenized with 15 slow strokes using glass homogenizer in 2 ml of ice-cold Syn-PER reagent (Thermo Fisher Scientific) containing a protease inhibitor cocktail (Sigma-Aldrich). The homogenate was centrifuged at 1200 × g for 10 min at 4°C to remove the nuclei and large debris. The supernatant was centrifuged at 15,000 × g for 20 min at 4°C to obtain the synaptosome pellet. The pellet was then dissolved in RIPA buffer (Thermo Fisher Scientific) containing a protease inhibitor cocktail and centrifuged at 15,000 × g for 10 min at 4°C to obtain the synaptosomal protein in the supernatant.
The protein was quantified using a DC protein assay kit (Bio-Rad). A quantity of 30 µg of protein from each sample was loaded and separated on 4–12% Bis-Tris SDS-PAGE gel (Invitrogen). The primary antibodies used for immunoblotting are listed in Table 2. The protein bands were detected with an enhanced chemiluminescence kit (Thermo Fisher Scientific), and the protein band density was visualized and quantified using an Odyssey Fc Imager (LI-COR Biosciences).
List of antibodies used in this study
Immunohistochemistry
Mice were deeply anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and perfused with PBS and then 4% paraformaldehyde in 0.1 m PBS. The DRG was dissected, postfixed for 2 h with 4% paraformaldehyde followed sequentially by 10, 20, and 30% sucrose solutions in PBS, and frozen in Tissue-Tek optimal cutting temperature compound. Slices were cut to 20 µm thick, collected onto SuperFrost Plus glass slides, and air dried for 1 h at 25°C. After washing with PBS, the sections were incubated in a blocking solution containing 1% bovine serum albumin in PBS for 20 min, incubated with the primary antibodies overnight at 4°C, and then incubated with the corresponding secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 568 (1:200; Invitrogen) for 1 h at 25°C. The primary antibodies used for immunocytochemistry are listed in Table 2. Images were acquired with a confocal laser-scanning microscope (Zeiss).
RNA sequencing
An RNA-sequencing library was constructed using RNAs from the mouse L3–L5 DRGs and sequenced using the Illumina Platform with paired-end reads (150 bp, mean 20 million reads) per sample (Novogene). Reads were mapped to the mouse genome (mm10) using Hisat2 (Kim et al., 2015). The mapped files were converted and sorted into BAM files using samtools (Li et al., 2009). Then the reads of each gene were calculated using featureCounts (Liao et al., 2014). Differentially expressed genes were analyzed using DESeq2 (Love et al., 2014). Gene Ontology (GO) Enrichment Analysis was performed using PANTHER online software (Mi et al., 2019). Pathway enrichment of differentially expressed genes was analyzed using Ingenuity Pathway Analysis (Krämer et al., 2014). Raw sequencing datasets have been deposited in the Gene Expression Omnibus (accession no. GSE145125).
Quantitative PCR
Total RNAs were extracted from the lumbar DRG and dorsal spinal cord tissues using TRIsure (catalog #BIO-38 032, BioLine). Samples were treated with RNase-free DNase (catalog #79254, QIAGEN), and 1 µg of RNA was used for reverse-transcription with a RevertAid RT Reverse Transcription Kit (catalog #K1691, Thermo Fisher Scientific). A total of 2 μl of five-times diluted cDNA was added to a 20 μl reaction volume with SYBR Green Real-Time PCR Mix (catalog #A25780, Thermo Fisher Scientific). Real-time PCR was run on a QuantStudio 7 Flex Real-Time PCR System (Applied Biosystems). The thermal cycling conditions were as follows: 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 45 s (Zhang et al., 2018; Ghosh et al., 2022). The primers used are listed in Table 1.
Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) assays were performed as previously described (Zhang et al., 2016; Zhang et al., 2018; Ghosh et al., 2022). Briefly, lumbar DRG tissues (L3–L5 in mice and L5/L6 in rats) were isolated and cross-linked with 2% formaldehyde for 10 min at 25°C. The crossed tissues were lysed with lysis buffer and sonicated to fragments of 200–1000 bp using a water bath sonicator (Qsonica) at 4°C (40 cycles, 30 s on and 30 s off). Chromatin was pulled down using the magnetic beads Dynabeads Protein G (catalog #10003D, Thermo Fisher Scientific) conjugated with indicated antibodies listed in Table 2. After de-cross-linking, DNA was recovered using a QIAquick PCR Purification Kit (catalog #28104, Qiagen). Data were analyzed and corrected by input (10% of the amount used for ChIP) or H3. Primers used for the ChIP assay are listed in Table 1.
Electrophysiological recordings in spinal cord slices
The lumbar spinal cord was obtained through laminectomy in mice anesthetized with 3% isoflurane. Spinal cords were placed in ice-cold artificial CSF containing the following (in mm): 234 sucrose, 3.6 KCl, 1.2 MgCl2, 2.5 CaCl2, 1.2 NaH2PO4, 12 glucose, and 25 NaHCO3, presaturated with 95% O2 and 5% CO2. Transverse slices were cut to 400 µm thick in ice-cold artificial CSF using a vibratome. The slices were preincubated in Krebs solution containing the following (in mm): 117 NaCl, 3.6 KCl, 1.2 MgCl2, 2.5 CaCl2, 1.2 NaH2PO4, 11 glucose, and 25 NaHCO3, oxygenated with 95% O2 and 5% CO2 at 34°C for at least 1 h. The slices were transferred into a glass-bottom recording chamber (Warner Instruments) and perfused continuously with Krebs solution at 3 ml/min at 34°C.
Neurons from the lamina II outer zone were selected for whole-cell patch-clamp recordings because they are involved in processing nociceptive input from primary afferent nerves (Li et al., 2002; Santos et al., 2007; Wang et al., 2018). A glass pipette (5–10 MΩ) was filled with an internal solution containing the following (in mm): 135 potassium gluconate, 5 tetraethylammonium chloride, 2 MgCl2, 0.5 CaCl2, 5 HEPES, 5 EGTA, 5 Mg-ATG, 0.5 Na-GTP, 1 guanosine 5'-O-(2-thiodiphosphate) trilithium salt, and 10 lidocaine N-ethyl bromide, pH adjusted to 7.2–7.4, with 1 m KOH and osmotic pressure to 290–300 mOsmol. EPSCs were elicited by electrical stimulation (0.2 ms, 0.6 mA, and 0.1 Hz) of the dorsal root and recorded at a holding potential of –60 mV. Monosynaptic EPSCs were identified on the basis of the constant latency and absence of conduction failure of evoked EPSCs in response to a 20 Hz electrical stimulation (Li et al., 2002; H.Y. Zhou et al., 2010; Chen et al., 2018). For paired-pulse stimulation, two EPSCs were electrically evoked by a pair of stimuli given at 50 ms intervals. The paired-pulse synaptic response, a measure of presynaptic plasticity, was expressed as the ratio of the amplitude of the second EPSC to the amplitude of the first EPSC (H.Y. Zhou et al., 2010; Xie et al., 2016).
To measure quantal glutamate release from presynaptic terminals, miniature EPSCs (mEPSCs) were recorded from lamina II outer neurons in the presence of 0.5 μm tetrodotoxin at a holding potential of –60 mV (Li et al., 2002; Chen et al., 2014a). The input resistance was continuously monitored, and the recording was terminated if the input resistance changed by >15%. Signals were recorded using an amplifier (MultiClamp 700B, Molecular Devices), filtered at 1–2 kHz, digitized at 10 kHz, and stored for off-line analysis. All drugs were freshly prepared in artificial CSF before the experiments and delivered via syringe pumps to reach their final concentrations. Tetrodotoxin (catalog #HB1035) and 2-amino-5-phosphonopentanoic acid (AP5; catalog #HB0252) were purchased from Hello Bio.
Study design and statistical analysis
Data are presented as means ± SEM. Data collection was randomized, and the investigators performing behavioral tests and electrophysiological recordings were blinded to the genotype and treatment. In electrophysiological experiments, only one neuron was recorded from each spinal cord slice, and at least five animals were used for each group. The amplitude of the evoked EPSCs was quantified by averaging 10 consecutive EPSCs with Clampfit 10.0 software (Molecular Devices). The amplitude and frequency of mEPSCs were analyzed using the MiniAnalysis peak detection program (Synaptosoft). The cumulative probability of the amplitude and the interevent interval of the mEPSCs were compared using the Kolmogorov–Smirnov test, which estimates the probability that two distributions are similar (Li et al., 2010; Chen et al., 2014b). RNA-sequencing data were generated using DRG tissues from wild-type (WT) and Hdac2-cKO mice in triplicates, and differentially expressed genes were identified by setting the threshold at p < 0.05. The Student's t test was used to compare two groups. We used one-way and two-way ANOVAs to compare more than two groups. Statistical analysis was performed using GraphPad Prism (version 8) software; p < 0.05 was considered to be statistically significant.
Results
Nerve injury increases histone acetylation and diminishes HDAC2 enrichment at the Cacna2d1 promoter in the DRG
As a first step toward understanding how α2δ-1 expression is controlled by epigenetic mechanisms, we determined whether histone acetylation levels at the Cacna2d1 promoter in the DRG are altered by nerve injury. We performed L5 and L6 SNL in rats, a well-characterized animal model of chronic neuropathic pain (Kim and Chung, 1992). ChIP and quantitative PCR (ChIP-PCR) assays using DRGs removed from rats 3 weeks after surgery showed that SNL significantly increased the acetylation levels of histone H3 and H4 at the promoter region of Cacna2d1 [p = 0.0012, t(9) = 3.825 for H3 acetylation (H3ac); p = 0.0083, t(9) = 3.366 for H4 acetylation (H4ac); Fig. 1A] but not the housekeeping gene Gapdh. Furthermore, SNL significantly increased the levels of H3K9 acetylation (H3K9ac; p = 0.0009, t(5) = 6.807) and H4K5 acetylation (H4K5ac; p = 0.0013, t(8) = 3.353), two histone marks associated with active gene transcription (Guan et al., 2009; Laumet et al., 2015), at the Cacna2d1 promoter in the DRG (Fig. 1B).
Nerve injury diminishes HDAC2 occupancy and causes histone hyperacetylation at the Cacna2d1 promoter in the DRG. A, Total H3ac and H4ac levels at the promoters of Cacna2d1 (–32 to 60 bp) and Gapdh (–120 to –23 bp) in the DRG from rats subjected to sham surgery or SNL; **p < 0.01, two-tailed t test, n = 10 rats per group. B, H3K9ac and H4K5ac levels at the promoters of Cacna2d1 and Gapdh in the DRG from rats subjected to sham surgery or SNL; **p < 0.01, ***p < 0.001, two-tailed t test, n = 6 rats per group for H3K9ac, n = 9 rats for H4K5ac. C, D, The occupancy of HDAC2 (C) and HDAC3 (D) at the promoters of Cacna2d1 and Gapdh in the DRG from rats subjected to sham surgery or SNL; ***p < 0.001, two-tailed t test, n = 6 rats per group for HDAC2, n = 7 rats per group for HDAC3. Data are expressed as means ± SEM.
Among the four subtypes of Class I HDACs (HDAC1, HDAC2, HDAC3, and HDAC8), HDAC2 is the most abundant subtype expressed in DRG neurons and is upregulated in the injured DRG (Laumet et al., 2015). HDAC3 is also expressed in DRG neurons and plays a role in axonal regeneration (Hervera et al., 2019). In contrast, HDAC1 is mainly present in glial cells, and HDAC8 is minimally expressed in the DRG (Laumet et al., 2015). Thus, we next used the ChIP-PCR assay to determine whether nerve injury affects the occupancy of HDAC2 and HDAC3 at the Cacna2d1 promoter. Remarkably, our promoter occupancy assay showed that SNL caused a large reduction in the enrichment of HDAC2 at the Cacna2d1 promoter (p < 0.002, t(5) = 9.814; Fig. 1C). However, the enrichment of HDAC3 at the Cacna2d1 promoter in the DRG did not differ significantly between SNL and sham control groups (Fig. 1D). These data raised the possibility that nerve injury increases Cacna2d1 transcription in the DRG by diminishing HDAC2 binding and consequent histone hyperacetylation at the Cacna2d1 promoter.
HDAC2, but not HDAC3, constitutively inhibits mechanical pain hypersensitivity
Because HDAC2 enrichment at the Cacna2d1 promoter in the DRG is diminished by nerve injury, we next used loss-of-function analysis to determine the role of constitutive HDAC2 in regulating pain hypersensitivity. We first used siRNA-induced HDAC2 knockdown to define the role of HDAC2 at the spinal cord level in the control of nociception. Intrathecal injection of Hdac2-specific siRNA in rats for 5 d caused ∼50% reduction in HDAC2 expression levels in the DRG and dorsal spinal cord (Fig. 2A–C). Strikingly, treatment with Hdac2-specific siRNA, but not the control siRNA, gradually and profoundly reduced the hindpaw withdrawal thresholds in response to tactile and noxious pressure stimuli (p < 0.0001, F(5,66) = 11.42 for tactile threshold; p < 0.0001, F(5,66) = 17.62 for pressure threshold; Fig. 2E). Treatment with Hdac2-specific siRNA did not significantly alter the heat withdrawal latency (Fig. 2E). In contrast, treatment with Hdac3-specific siRNA for 5 d had no significant effects on the tactile, pressure, or heat withdrawal threshold in rats (Fig. 2D,E).
HDAC2 knockdown at the spinal cord level causes pain hypersensitivity. A, B, Representative blots and quantification show the protein levels of HDAC2 in the DRG (A) and spinal cord (B) of rats treated with control siRNA or Hdac2-specific siRNA; **p < 0.01, ***p < 0.001, two-tailed t test, n = 6 rats per group. C, D, Quantitative PCR shows the mRNA levels of Hdac2 (C) and Hdac3 (D) in the DRG of rats treated with control siRNA, Hdac2-specific siRNA, or Hdac3-specific siRNA; **p < 0.01, ***p < 0.001, two-tailed t test, n = 6 rats per group. E, Time course of changes in withdrawal thresholds tested with von Frey filaments, pressure, and heat stimuli in rats treated with Hdac2-specific siRNA, Hdac3-specific siRNA, or control siRNA. Two-way ANOVA showed a significant interaction between the siRNA type and treatment time (F(5,66) = 13.47, p < 0.001 for tactile threshold; F(5,66) = 15.75, p < 0.001 for pressure threshold); ***p < 0.001, compared with baseline (day 0), two-way ANOVA with Tukey's post hoc test (n = 6 rats in control siRNA group, n = 7 rats in Hdac2-siRNA and Hdac3-siRNA groups). Data are expressed as means ± SEM.
To specifically determine the role of HDAC2 in primary sensory neurons in nociceptive control, we generated Hdac2 conditional knockout (Hdac2-cKO, AvlCre/+::Hdac2fl/fl) mice by crossing Hdac2fl/fl mice (Guan et al., 2009) with AvlCre/+ mice (X. Zhou et al., 2010) so that Hdac2 in DRG neurons was selectively ablated (Fig. 3A–E). Consistent with rats treated with Hdac2-specific siRNA, the tactile and pressure withdrawal thresholds were much lower in Hdac2-cKO mice than in age-matched littermate WT controls (p < 0.0001, t(19) = 6.881 for tactile threshold; p < 0.0001, t(18) = 5.852 for pressure threshold; Fig. 3F). However, the heat-elicited withdrawal latency was similar between Hdac2-cKO and WT mice (Fig. 3F).
HDAC2 ablation in primary sensory neurons induces pain hypersensitivity. A, Representative immunocytochemical labeling shows the presence of HDAC2 (green)- and NeuN (red)-labeled neurons in the DRG section from Hdac2-cKO and WT control mice. B, Representative blots and quantification show the HDAC2 protein level in the DRG of Hdac2-cKO and WT mice; ***p < 0.001, two-tailed t test, n = 6 mice per group. C, The mRNA level of Hdac2 in the DRG of Hdac2-cKO and WT mice; ***p < 0.001, two-tailed t test, n = 5 mice per group. D, Representative blots and quantification show the HDAC2 protein level in the spinal cord of Hdac2-cKO and WT mice, n = 6 mice per group. E, The mRNA level of Hdac2 in the spinal cord of Hdac2-cKO and WT mice, n = 4 mice per group. F, Withdrawal thresholds tested with von Frey filaments, pressure, and heat stimuli in 10-week-old Hdac2-cKO mice and WT mice; ***p < 0.001, two-tailed t test (n = 10 WT mice, n = 11 Hdac2-cKO mice for von Frey filament test; n = 10 mice per group for pressure test; n = 14 mice per group for heat test). G, Representative blots and quantification show the HDAC2 protein level in the DRG and spinal cord of Avl-CreERT2/+::Hdac2fl/fl (Hdac2-icKO) and WT mice treated with tamoxifen; ***p < 0.001, two-tailed t test, n = 6 mice per group. H, Time course of changes in withdrawal thresholds tested with von Frey filaments, pressure, and heat stimuli in tamoxifen-induced Hdac2-icKO and WT mice. Two-way ANOVA showed a significant interaction between the genotype and treatment time (F(3,40) = 44.7, p < 0.001 for tactile threshold; F(3,40) = 11.04, p < 0.001 for pressure threshold); ***p < 0.001, compared with baseline, two-way ANOVA with Tukey's post hoc test; n = 5 WT mice, n = 7 Hdac2-icKO mice. Data are expressed as means ± SEM.
Because deletion of Hdac2 in premature mice may affect DRG neuronal development, we generated mice with tamoxifen-dependent inducible deletion of Hdac2 in DRG neurons by crossing Hdac2fl/fl mice with Advillin-CreERT2 mice (Lau et al., 2011). We used tamoxifen treatment to induce Hdac2 conditional KO in DRG neurons in adult Avl-CreERT2::Hdac2fl/fl mice (Hdac2-icKO; Fig. 3G). Before tamoxifen treatment, the tactile, pressure, and heat withdrawal thresholds were similar between Avl-CreERT2::Hdac2fl/fl and WT mice. Tamoxifen treatment had no effect on the tactile, pressure, or heat withdrawal thresholds in WT control mice (Fig. 3H). In contrast, Hdac2-icKO mice showed profoundly reduced tactile and pressure withdrawal thresholds, and these reductions lasted at least 3 weeks (p < 0.0001, F(3,40) = 84.92 for tactile threshold; p < 0.0001, F(3,40) = 11.16 for pressure threshold; Fig. 3H). The heat withdrawal latency did not differ significantly between Hdac2-icKO and WT mice (Fig. 3H). Together, these findings reveal a crucial role of constitutive HDAC2 in primary sensory neurons in suppressing mechanical pain hypersensitivity.
HDAC2 in primary sensory neurons tonically inhibits Cacna2d1 transcription and histone acetylation at the Cacna2d1 promoter
In subsequent experiments, we used loss-of-function approaches to determine the role of constitutively expressed HDAC2 in the DRG in regulating α2δ-1 expression. Increased α2δ-1 expression at the spinal cord level increases both presynaptic and postsynaptic NMDAR activity in the spinal dorsal horn (Chen et al., 2018). We first used quantitative PCR and immunoblotting to determine whether siRNA-induced Hdac2 knockdown at the spinal cord level in rats alters α2δ-1 expression. Indeed, the mRNA (p = 0.0007, t(10) = 4.838 for DRG; p = 0.0054, t(10) = 3.537 for spinal cord) and protein (p < 0.0001, t(10) = 6.463 for DRG; p = 0.0026, t(10) = 3.994 for spinal cord) levels of α2δ-1 in DRG and dorsal spinal cord tissues were significantly greater in rats intrathecally treated with Hdac2-specific siRNA than in rats treated with control siRNA (Fig. 4A–D).
HDAC2 deficiency increases α2δ-1 expression in the DRG and spinal cord. A, B, Representative blots and quantification (A) and quantitative PCR (B) show the protein and mRNA levels of α2δ-1 in the DRG from Hdac2-specific siRNA- or control siRNA-treated rats; ***p < 0.001, two-tailed t test, n = 6 rats per group. C, D, Representative blots and quantification (C) and quantitative PCR (D) show the protein and mRNA levels of α2δ-1 in the spinal cord from Hdac2-specific siRNA-treated or control siRNA-treated rats; **p < 0.01, two-tailed t test, n = 6 rats per group. E, F, Representative blots and quantification (E, n = 10 mice) and quantitative PCR (F, n = 6 mice) show the protein and mRNA levels of α2δ-1 in the DRG from Hdac2-cKO or WT mice; ***p < 0.001, two-tailed t test. G, H, Representative blots and quantification (G) and quantitative PCR (H) show the protein and mRNA levels of α2δ-1 in the spinal cord from Hdac2-cKO or WT mice; n = 6 mice per group. I, Representative blots and quantification show the α2δ-1 protein level in dorsal spinal cord synaptosomes from Hdac2-cKO or WT mice; **p < 0.01, two-tailed t test, n = 7 mice per group. Data are expressed as means ± SEM.
We also measured α2δ-1 expression in the DRG obtained from WT and Hdac2-cKO mice. The mRNA and protein levels of α2δ-1 in the DRG were much higher in Hdac2-cKO mice than in WT mice (p = 0.0007, t(10) = 4.822 for mRNA levels; p < 0.0001, t(18) = 6.127 for protein levels; Fig. 4E,F). Although the total protein and mRNA levels of α2δ-1 in the spinal cord did not differ significantly between WT and Hdac2-cKO mice, the α2δ-1 protein level in dorsal spinal cord synaptosomes was significantly greater in Hdac2-cKO than in WT mice (p = 0.003, t(12) = 3.707; Fig. 4G–I). These results suggest that HDAC2 constitutively represses α2δ-1 expression in DRG neurons and their central terminals in the spinal dorsal horn.
We then used ChIP-PCR to determine the occupancy of HDAC2 in the Cacna2d1 genomic sequence in the mouse DRG. HDAC2 was highly enriched at the promoter region (71–164 bp) of Cacna2d1, near its transcriptional start site (p < 0.0001, F(2,15) = 42.05; Fig. 5A). However, HDAC2 was not enriched at the 3′-untranscribed region (440,221 to 440,390 bp) of Cacna2d1 (Fig. 5A). Also, there was no enrichment of HDAC2 at the promoter region of Gapdh in the DRG (Fig. 5A).
HDAC2 in primary sensory neurons restrains histone acetylation at the Cacna2d1 promoter in the DRG. A, The occupancy of HDAC2 at the promoter region (71 to 164 bp) of Cacna2d1 in the mouse DRG. The 3′ untranslated region (UTR; 440,221 to 440,390 bp) of Cacna2d1 and promoter of Gapdh (–53 to 85 bp) were used as controls. IgG was used as a negative control for HDAC2 immunoprecipitation (HDAC2-IP); ***p < 0.001, one-way ANOVA with Tukey's post hoc test, n = 6 mice per group. B, Total H3ac and H4ac levels at the Cacna2d1 and Gapdh promoters in the DRG from Hdac2-cKO and WT mice; **p < 0.01, ***p < 0.001, two-tailed t test; n = 8 mice per group. C, The H3K9ac, H4K5ac, and H3K14ac levels at the Cacna2d1 and Gapdh promoters in the DRG from Hdac2-cKO and WT mice; **p < 0.01, two-tailed t test; n = 6 mice per group for H3K9ac and H3K14ac, n = 10 mice per group for H4K5ac. Data are expressed as means ± SEM.
We next determined whether constitutive HDAC2 controls histone acetylation at the Cacna2d1 promoter in the DRG. ChIP-PCR assays showed that the total H3 and H4 acetylation levels at the promoter of Cacna2d1 (p = 0.0003, t(14) = 4.759 for H3ac; p = 0.0031, t(14) = 3.563 for H4ac), but not of Gapdh, in the DRG were much greater in Hdac2-cKO mice than in WT control mice (Fig. 5B). Furthermore, we determined which specific histone acetylation marks at the promoter of Cacna2d1 are the substrates of HDAC2 in the DRG. We selected three histone acetylation marks, H3K9ac, H3K14ac, and H4K5ac, because they are generally associated with active gene transcription in neural tissues (Guan et al., 2009; Laumet et al., 2015; Yamakawa et al., 2017). The levels of H3K9ac (p = 0.0024, t(10) = 4.032) and H4K5ac (p = 0.0099, t(18) = 2.885) at the promoter of Cacna2d1, but not of Gapdh, in the DRG were significantly higher in Hdac2-cKO than in WT mice (Fig. 5C). However, the H3K14ac level at the Cacna2d1 promoter in the DRG did not differ significantly between Hdac2-cKO and WT mice (Fig. 5C). Collectively, these data indicate that constitutive HDAC2 in DRG neurons is a key epigenetic repressor that inhibits Cacna2d1 transcription by reducing histone acetylation at the promoter of Cacna2d1.
Constitutive HDAC2 in DRG neurons restrains activity of presynaptic NMDARs in the spinal dorsal horn by limiting α2δ-1 expression
α2δ-1 is essential for increased activity of presynaptic NMDARs at primary afferent central terminals in neuropathic pain conditions (Chen et al., 2018; Chen et al., 2019; Zhang et al., 2021). Based on this knowledge, we next sought to determine whether HDAC2 expressed in DRG neurons constitutively suppresses presynaptic NMDAR activity in the spinal dorsal horn. We performed whole-cell voltage-clamp recordings of mEPSCs of lamina II neurons in spinal cord slices from Hdac2-cKO and WT mice. The frequency, but not the amplitude, of mEPSCs of lamina II neurons was significantly greater in Hdac2-cKO mice than in WT mice (p = 0.0058, n = 12 neurons in WT group, n = 13 neurons in Hdac2-cKO group; Fig. 6A,B). This increased frequency of mEPSCs in Hdac2-cKO mice was abolished within 5 min after bath application of 50 μm AP5, a specific NMDAR antagonist. Similarly, pretreatment of spinal cord slices from Hdac2-cKO mice with 100 μm gabapentin, an α2δ-1 inhibitory ligand (Gee et al., 1996; Fuller-Bicer et al., 2009), for 30 min normalized the increased frequency of mEPSCs in lamina II neurons (n = 13 neurons; Fig. 6A,B). This increased presynaptic NMDAR activity in the spinal cord by HDAC2 deficiency is similar to that in neuropathic pain conditions (Xie et al., 2016; Chen et al., 2018; Chen et al., 2019). Treatment with gabapentin has no effect on spinal NMDAR activity in WT control mice (Chen et al., 2018; Chen et al., 2019; Deng et al., 2019).
HDAC2 in DRG neurons constitutively restrains α2δ-1–dependent presynaptic NMDAR activity in the spinal dorsal horn. A, B, Representative current traces and cumulative plots (A) and mean changes (B) of mEPSCs of spinal lamina II neurons before (Baseline), with (AP5), and after (Washout) bath application of 50 μm AP5. Slice recordings were performed using spinal cords from five WT mice (n = 12 neurons), six Hdac2-cKO mice (n = 13 neurons), and five Hdac2-cKO mice pretreated with 100 μm gabapentin (n = 13 neurons). Two-way ANOVA (B) showed there was a significant main effect for AP5 treatment (F(2,69) = 26.34, p < 0.001) and a significant interaction between the genotype and AP5 treatment (F(2,69) = 3.497, p = 0.0358). C, D, Original recording traces and cumulative plots (C) and mean changes (D) show the effect of pretreatment with Tat-fused α2δ-1CT peptide or control peptide on the frequency and amplitude of mEPSCs of spinal lamina II neurons before (Baseline), with (AP5), and after (Washout) bath application of 50 μm AP5. Spinal cords slices were obtained from Hdac2-cKO mice and pretreated with 1 μm α2δ-1CT peptide (n = 12 neurons from 5 mice) or 1 μm control peptide (n = 13 neurons from 6 mice).Two-way ANOVA (D) showed there was a significant main effect for AP5 treatment (F(2,69) = 3.441, p < 0.0376) and a significant interaction between the control peptide and α2δ-1CT peptide treatment (F(2,69) = 3.239, p = 0.0452); *p < 0.05 versus respective baseline; #p < 0.05, ##p < 0.01 versus the baseline in the WT or control peptide group, two-way ANOVA followed by Tukey's post hoc test. Data are expressed as means ± SEM.
α2δ-1 physically interacts with NMDARs via its C terminus, an intrinsically disordered protein region, and a Tat-fused α2δ-1 C terminus peptide (α2δ-1CT peptide) effectively disrupts the α2δ-1–NMDAR interaction (Chen et al., 2018). Pretreatment of spinal cord slices from Hdac2-cKO mice with 1 μm α2δ-1CT peptide, but not 1 μm Tat-fused control peptide, for 30 min normalized the increased frequency of mEPSCs in Hdac2-cKO mice (n = 13 neurons in control peptide group, n = 12 neurons in α2δ-1CT peptide group; Fig. 6C,D). These results suggest a critical role of α2δ-1–bound NMDARs in HDAC2 deficiency-induced presynaptic NMDAR hyperactivity.
Next, we recorded EPSCs of spinal lamina II neurons monosynaptically evoked by dorsal root stimulation to determine the role of HDAC2 in regulating the activity of NMDARs expressed at primary afferent central terminals. The amplitude of evoked EPSCs of lamina II neurons was significantly larger in Hdac2-cKO mice than in WT mice (p = 0.0080; n = 11 neurons per group; Fig. 7A,B). Bath application of AP5 rapidly reduced the amplitude of evoked EPSCs of lamina II neurons in Hdac2-cKO mice, but not in WT mice (Fig. 7A,B). Pretreatment of spinal cord slices with gabapentin (n = 12 neurons) or α2δ-1CT peptide (n = 11 neurons), but not the control peptide (n = 11 neurons), also normalized the increased amplitude of evoked monosynaptic EPSCs in Hdac2-cKO mice (Fig. 7C,D).
HDAC2 in DRG neurons tonically inhibits α2δ-1–dependent NMDAR activity at primary afferent central terminals. A, B, Representative current traces (A) and mean changes (B) of EPSCs in spinal lamina II neurons monosynaptically evoked from the dorsal root and EPSCs evoked by a pair of pulses before (Baseline), with (AP5), and after (Washout) bath application of 50 μm AP5. Slice recordings were performed using spinal cords from six WT mice (n = 11 neurons), six Hdac2-cKO mice (n = 11 neurons), and five Hdac2-cKO mice pretreated with 100 μm gabapentin (n = 12 neurons). Two-way ANOVA (B) showed there was a significant main effect for AP5 treatment on the EPSC amplitude (F(2,60) = 4.199, p = 0.0196) and a significant interaction between the genotype and AP5 treatment (F(2,60) = 3.283, p = 0.0444). C, D, Original recording traces (C) and mean changes (D) show the effect of pretreatment with Tat-fused α2δ-1CT peptide or control (Cont) peptide on the amplitude of evoked EPSCs and paired-pulse ratio of evoked EPSCs in spinal lamina II neurons before (Baseline), with (AP5), and after (Washout) bath application of 50 μm AP5. Spinal cord slices were obtained from Hdac2-cKO mice and pretreated with 1 μm α2δ-1CT peptide or 1 μm control peptide (n = 11 neurons from 5 mice per group). Two-way ANOVA (D) showed there was a significant main effect for AP5 treatment on the EPSC amplitude (F(2,60) = 6.395, p = 0.003) and a significant interaction between the genotype and AP5 treatment (F(2,60) = 3.405, p = 0.0397); *p < 0.05, **p < 0.01 versus respective baseline, #p < 0.05, ##p < 0.01 versus the baseline value in the WT or control peptide group; two-way ANOVA followed by Tukey's post hoc test. Data are expressed as means ± SEM.
In addition, the paired-pulse ratio of monosynaptically evoked EPSCs in lamina II neurons was significantly smaller in Hdac2-cKO mice than in WT mice (p = 0.0088; Fig. 7A,B). Bath application of AP5 inhibited the first evoked EPSCs more than the second evoked EPSCs, resulting in an increase in the paired-pulse ratio in Hdac2-cKO mice (Fig. 7A,B). Treatment of spinal cord slices with gabapentin or α2δ-1CT peptide, but not the control peptide, also normalized the decreased paired-pulse ratio of evoked EPSCs in Hdac2-cKO mice (Fig. 7C,D). Collectively, these findings are consistent with the notion that HDAC2 in DRG neurons normally suppresses the activity of presynaptic NMDARs at primary afferent central terminals by limiting the availability of α2δ-1 proteins.
HDAC2 in DRG neurons constitutively inhibits pain hypersensitivity by suppressing α2δ-1–NMDAR activity
The important roles of α2δ-1 and NMDARs at the spinal cord level in neuropathic pain are well documented (Zhou et al., 2012; Chen et al., 2018; Chen et al., 2019). We sought to determine whether HDAC2 in DRG neurons controls pain hypersensitivity via α2δ-1–dependent NMDAR activity. In Hdac2-cKO mice, we first tested the effect of intrathecal injection of 5 µg AP5, a specific NMDAR antagonist, on tactile and pressure withdrawal thresholds tested with von Frey filaments and a pressure stimulus. AP5 completely reversed the pain hypersensitivity in Hdac2-cKO mice (n = 6 mice per group; Fig. 8A). Similarly, intraperitoneal injection of 60 mg/kg or 100 mg/kg gabapentin rapidly increased the tactile and pressure withdrawal thresholds that had been reduced in Hdac2-cKO mice (n = 6 mice for WT; n = 8 mice for Hdac2-cKO; Fig. 8B). Furthermore, intrathecal administration of 1 µg α2δ-1CT peptide, but not 1 µg Tat-fused control peptide, completely reversed the pain hypersensitivity in Hdac2-cKO mice (n = 6 mice per group; Fig. 8C). However, treatment with gabapentin or α2δ-1CT peptide had no effect on the tactile and pressure withdrawal thresholds in WT mice.
HDAC2 deficiency-induced pain hypersensitivity is maintained by α2δ-1–bound NMDARs at the spinal cord level. A–C, Time course of the effect of AP5 [5 μg, intrathecal (i.t.); A], gabapentin (60 mg/kg or 100 mg/kg, i.p.; B), and α2δ-1CT peptide (1 μg, i.t.; C) on withdrawal thresholds tested with von Frey filaments or a pressure stimulus in Hdac2-cKO and WT mice. Two-way ANOVA showed a significant interaction between the genotype and treatment time (A; F(4,50) = 7.765, p < 0.001 for tactile threshold; F(4,50) = 16.58, p < 0.001 for pressure threshold), (B; F(8,95) = 5.076, p < 0.001 for tactile threshold; F(8,75) = 2.879, p < 0.001 for pressure threshold), (C; F(8,75) = 5.932, p < 0.001 for tactile threshold; F(8,75) = 5.346, p < 0.001 for pressure threshold); *p < 0.05, **p < 0.01, ***p < 0.001, compared with baseline (BL), two-way ANOVA with Tukey's post hoc test, n = 8 mice per group for Hdac2-cKO + gabapentin; n = 6 mice per group for other groups. D–F, Time course of the effect of AP5 (5 μg, i.t.; D), gabapentin (60 mg/kg, i.p.; E), and α2δ-1CT peptide (1 μg, i.t.; F) on withdrawal thresholds tested with von Frey filaments or a pressure stimulus in Hdac2-specific siRNA- or control siRNA-treated rats. Two-way ANOVA showed a significant interaction between the siRNA treatment and drug treatment time (D; F(4,55) = 6.373, p = 0.0003 for tactile threshold; F(4,55) = 4.599, p = 0.0028 for pressure threshold), (E; F(4,55) = 4.339, p = 0.004 for tactile threshold; F(4,55) = 4.865, p = 0.002 for pressure threshold), (F; F(8,85) = 3.825, p = 0.0007 for tactile threshold; F(8,85) = 5.172, p < 0.001 for pressure threshold); *p < 0.05, **p < 0.01, ***p < 0.001, compared with BL, two-way ANOVA with Tukey's post hoc test; n = 6 rats in the control siRNA group; n = 7 rats in the Hdac2-siRNA group. Data are expressed as means ± SEM.
In rats intrathecally treated with Hdac2-specific siRNA, similar treatment with AP5, gabapentin, or α2δ-1CT peptide normalized the already reduced tactile and pressure withdrawal thresholds (n = 6 rats per group for control siRNA; n = 7 rats per group for Hdac2-specific siRNA; Fig. 8D–F). In addition, we assessed and confirmed the same reversal effects of AP5, gabapentin, and α2δ-1CT peptide on the pain hypersensitivity present in tamoxifen-induced Hdac2-icKO mice (n = 5 WT mice; n = 7 Hdac2-icKO mice; Fig. 9A–F). Together, these data indicate that HDAC2 constitutively represses pain hypersensitivity by limiting α2δ-1–dependent NMDAR activity at the spinal cord level.
α2δ-1–bound NMDARs at the spinal cord level maintain pain hypersensitivity in Hdac2-icKO mice. A, B, Time course of the effect of gabapentin (100 mg/kg, i.p.) on withdrawal thresholds tested with von Frey filaments (A) and a pressure stimulus (B) in Hdac2-icKO or WT mice. C, D, Time course of the effect of AP5 (5 μg, i.t.) on withdrawal thresholds tested with von Frey filaments (C) and a pressure stimulus (D) in Hdac2-icKO or WT mice. E, F, Time course of the effect of α2δ-1CT peptide (1 μg, i.t.) on withdrawal thresholds tested with von Frey filaments (E) and a pressure stimulus (F) in Hdac2-icKO or WT mice. Two-way ANOVA showed a significant interaction between the genotype and drug treatment time (F(8,80) = 56.94, p < 0.0001 in A; F(8,80) = 9.475, p < 0.0001 in B; F(8,80) = 34.94, p < 0.0001 in C; F(8,80) = 18.33, p < 0.0001 in D; F(8,80) = 38.85, p < 0.0001 in E; F(8,80) = 13.5, p < 0.0001 in F); *p < 0.05, **p < 0.01, ***p < 0.001, compared with baseline (time 0), two-way ANOVA with Tukey's post hoc test; n = 7 Hdac2-icKO mice; n = 5 WT mice. Data are expressed as means ± SEM.
Constitutive HDAC2 in DRG neurons restrains pain hypersensitivity mainly via limiting α2δ-1 expression
To investigate other gene targets constitutively regulated by HDAC2 in DRG neurons, we performed RNA-sequencing analysis using L5 and L6 DRGs from Hdac2-cKO and WT mice. A total of 210 genes showed a significant increase, whereas 455 genes showed a significant decrease, in Hdac2-cKO mice compared with WT mice (Fig. 10A,B; Extended Data Figs. 10-1, 10-2). These differentially regulated genes were then used for GO analysis. The 210 genes upregulated in Hdac2-cKO are enriched in several biological processes, including ion transport, regulation of secretion, T cell activity, intracellular signal transduction, and inflammatory responses (Fig. 10C; Extended Data Fig. 10-2). The 455 downregulated genes are enriched in other biological processes, including cellular morphogenesis, cell differentiation, cytoskeletal organization, myelination, and G-protein-coupled receptor signaling pathway (Fig. 10D; Extended Data Fig. 10-3). Ingenuity Pathway Analysis indicated that genes with differential expression caused by ablating Hdac2 in DRG neurons are mainly involved in calcium signaling, actin cytoskeleton signaling, tight junction signaling, and Cdc42 signaling (Fig. 10E; Extended Data Fig. 10-3).
Transcriptome analysis of DRG tissues from WT and Hdac2-cKO mice. A, Heat map data show that ablating Hdac2 in DRG neurons significantly downregulated 455 and upregulated 210 genes (p < 0.05) in the DRG. RNA-sequencing data were obtained from three Hdac2-cKO (cKO1, cKO2, and cKO3) and three WT (WT1, WT2, and WT3) mice. B, Volcano plot shows differentially expressed genes in the DRG induced by ablating Hdac2 in DRG neurons. C, D, GO analysis shows the enrichment pathways of upregulated genes (C) and downregulated genes (D). E, Ingenuity Pathway Analysis (IPA) shows the top eight pathways in which differentially expressed genes are enriched (Extendend Data Figs. 10-1, 10-2, 10-3).
Figure 10-1
All genes in RNA-sequencing data from Hdac2-cKO (Excel file). Download Figure 10-1, XLSX file.
Figure 10-2
Upregulated and downregulated genes in DRG in Hdac2-cKO (Excel file). Download Figure 10-2, XLSX file.
Figure 10-3
GO and IPA pathway analyses of differentially regulated genes in Hdac2-cKO (Excel file). Download Figure 10-3, XLSX file.
Notably, RNA-sequencing data showed that the mRNA level of Cacna2d1 in the DRG was significantly increased in Hdac2-cKO mice compared with WT mice (Extended Data Figs. 10-1, 10-2). We also manually curated a list of differentially regulated genes potentially relevant to nociceptive transduction and transmission. Except for significant changes in a few putative nociceptive genes (e.g., Il1b and P2ry1), the mRNA levels of pronociceptive genes in the DRG implicated in neuropathic pain, including Csf1, Grm5, Scn9a, Scn10a, Tnf, and Tlr4, did not differ significantly between WT mice and Hdac2-cKO mice (Extended Data Figs. 10-1, 10-2). In addition, the mRNA levels of many antinociceptive gene targets, such as potassium channels (e.g., Kcna4, Kcnd2, and Kcnq2), Oprm1, Oprd1, and Cnr1, in the DRG were similar between Hdac2-cKO mice and WT mice (Extended Data Figs. 10-1, 10-2).
Finally, we used Cacna2d1 KO mice to determine to what extent HDAC2 deficiency causes pain hypersensitivity via α2δ-1. We injected intrathecally Hdac2-specific siRNA via lumbar puncture in Cacna2d1 KO mice and WT mice and measured their withdrawal tactile and pressure withdrawal thresholds. Treatment with Hdac2-specific siRNA for 5 d in WT mice led to a gradual and profound reduction in the withdrawal thresholds (p < 0.0001, F(5,126) = 15.50 for tactile threshold; p < 0.0001, F(5,126) = 6.42 for pressure threshold; Fig. 11A–C). By contrast, the reduction in the tactile withdrawal threshold induced by treatment with Hdac2-specific siRNA was largely blunted in Cacna2d1 KO mice (p = 0.0001, F(2,261) = 30.19; Fig. 11A). Furthermore, treatment with Hdac2-specific siRNA failed to reduce the pressure withdrawal threshold in Cacna2d1 KO mice (Fig. 11B). These findings suggest that constitutive HDAC2 represses pain hypersensitivity primarily by inhibiting α2δ-1 expression at the spinal cord level.
HDAC2 deficiency-induced pain hypersensitivity is blunted in Cacna2d1 KO mice. A, B, Time course of changes in the withdrawal thresholds tested with von Frey filaments (A) and a pressure stimulus (B) in WT mice treated intrathecally with control siRNA or Hdac2-specific siRNA and in Cacna2d1 KO mice treated intrathecally with Hdac2-specific siRNA. Two-way ANOVA showed a significant interaction between the genotype and siRNA treatment time (F(10,126) = 3.698, p = 0.0002 in A; F(10,126) = 3.107, p = 0.0014 in B); *p < 0.05, **p < 0.01, ***p < 0.001 versus respective baseline (day 0); #p < 0.05 versus WT mice treated with Hdac2-specific siRNA; two-way ANOVA with Tukey's post hoc test, n = 8 mice per group. C, The Hdac2 mRNA level in the DRG and spinal cord of Cacna2d1 KO and WT mice treated with control siRNA or Hdac2-specific siRNA; ***p < 0.001, one-way ANOVA with Dunnett's post hoc test, n = 6 mice per group. Data are expressed as means ± SEM. D, Schematic illustrates the epigenetic control of α2δ-1 expression and synaptic NMDAR activity by HDAC2 in primary sensory neurons and their central terminals.
Discussion
Our study provides substantial new evidence that HDAC2 in primary sensory neurons, via dynamic regulation of histone acetylation status at the Cacna2d1 promoter, plays a central role in synaptic plasticity associated with neuropathic pain (Fig. 11D). Excess α2δ-1 proteins produced by injured DRG neurons mainly amplify nociceptive input by recruiting presynaptic NMDARs at primary afferent central terminals (Chen et al., 2018; Chen et al., 2019; Zhang et al., 2021; Zhou et al., 2021). Thus, sustained upregulation of α2δ-1 in primary sensory neurons plays a key role in the development and maintenance of neuropathic pain. In this study, our promoter occupancy analysis revealed that nerve injury diminished HDAC2 enrichment at the Cacna2d1 promoter in the DRG. Importantly, the decrease in HDAC2 binding was accompanied by increased histone H3 and H4 acetylation, particularly H3K9ac and H4K5ac levels, at the Cacna2d1 promoter. Our findings support the idea that diminished HDAC2 occupancy at the Cacna2d1 promoter alters the chromatin architecture to promote Cacna2d1 transcription in DRG neurons after nerve injury. Our discovery of the crucial role of HDAC2 in regulating α2δ-1 expression was unexpected, because nerve injury increases the HDAC2 level in bulk DRG tissues (Laumet et al., 2015). However, global changes in HDAC2 levels in whole DRG tissues cannot predict the abundance of HDAC2 locally at the gene promoter. This critical difference reiterates the importance of determining the binding site occupancy at individual gene promoters to appropriately identify the function of HDAC subtypes in regulating specific gene transcription and chronic pain. Our findings pinpoint that HDAC2 dissociation at the Cacna2d1 promoter in the DRG constitutes a unique epigenetic mechanism that controls the development of chronic neuropathic pain. It remains unclear how nerve injury diminishes HDAC2 occupancy at the Cacna2d1 promoter in the DRG. HDAC2 may interact with transcription factors and corepressors, such as SP3, FOXO3a, and coREST (Guan et al., 2009; Formisano et al., 2015; Peng et al., 2015; Yamakawa et al., 2017). Further research is required to define how various transcription factors and corepressors control HDAC2 binding at the Cacna2d1 promoter in DRG neurons. Until the exact mechanism underlying nerve-injury-induced diminishment of HDAC2 binding at the Cacna2d1 promoter is known, it is difficult to restore HDAC2 enrichment at the Cacna2d1 promoter in the injured DRG.
Our study reveals that HDAC2 in DRG neurons constitutively controls mechanical pain hypersensitivity. We found that Hdac2 knockdown at the spinal cord level or Hdac2 conditional KO in DRG neurons caused a neuropathic pain-like phenotype. Importantly, this phenotype occurred concomitantly with an increase in histone acetylation at the Cacna2d1 promoter and in α2δ-1 expression. Overexpression of full-length α2δ-1, but not α2δ-1 with C terminus mutations, at the spinal cord level causes a neuropathic pain-like phenotype (Chen et al., 2018; Li et al., 2021). Because HDAC2 is highly enriched at the Cacna2d1 promoter in the normal DRG, it can constitutively repress Cacna2d1 transcription via inhibiting histone acetylation at its promoter. Intriguingly, HDAC2 deficiency increased mechanical hypersensitivity, but not cutaneous heat sensitivity. It is unclear why HDAC2 knockdown or cKO increased mechanical, but not heat, sensitivity. Consistent with this phenotype, most patients with neuropathic pain experience mechanical, but not heat, hypersensitivity (Baron and Saguer, 1993; Campbell and Meyer, 2006; Vollert et al., 2017). It is possible that the discrepancy in pain phenotypes caused by nerve injury and HDAC2 knockdown results from α2δ-1 upregulation in different types of DRG neurons. Traumatic nerve injury likely damages both mechano- and heat-sensitive primary afferents to increase α2δ-1 expression in all DRG neurons (Newton et al., 2001; Zhang et al., 2021), whereas α2δ-1 expression controlled by constitutive HDAC2 may be predominantly expressed in mechanosensitive DRG neurons. This could explain the effect of gabapentinoids on both mechano- and heat-hypersensitivity associated with neuropathic pain (Chen et al., 2001; Chen et al., 2019). Remarkably, the pain hypersensitivity caused by HDAC2 deficiency was fully reversed by blocking NMDARs with AP5, inhibiting α2δ-1 with gabapentin, and disrupting the α2δ-1–NMDAR interaction with α2δ-1CT peptide, interventions that are also highly effective for reducing neuropathic pain caused by traumatic nerve injury, small-fiber neuropathy, and chemotherapy (Chen et al., 2018; Chen et al., 2019; Zhang et al., 2021). Our study thus implicates an important protective role of HDAC2 in primary sensory neurons in the transition from acute to chronic neuropathic pain by constitutive repression of Cacna2d1 transcription.
Another salient finding of our study is that HDAC2 in DRG neurons constitutively regulates synaptic NMDAR activity in the spinal dorsal horn via controlling the expression level of α2δ-1. We showed that the activity of presynaptic NMDARs expressed at primary afferent central terminals is augmented in Hdac2-cKO mice, supporting the critical function of HDAC2 in tonically suppressing nociceptive glutamatergic input to dorsal horn neurons. HDAC2 knockdown also potentiates excitatory synaptic transmission in hippocampal neurons (Hanson et al., 2013), suggesting a wide-ranging role of HDAC2 in regulating synaptic plasticity in the nervous system. NMDARs are expressed at primary afferent central terminals in the spinal cord (Liu et al., 1994); however, these presynaptic NMDARs remain functionally inactive under normal conditions and become active in painful conditions (Chen et al., 2018; Chen et al., 2019; Deng et al., 2019; Huang et al., 2020). We have shown that synaptic expression of NMDARs at primary afferent central terminals requires α2δ-1 proteins (Chen et al., 2018; Zhang et al., 2021; Zhou et al., 2021; Huang et al., 2022). Increased expression of α2δ-1 interacts with NMDARs via its C terminus to augment their synaptic trafficking, causing tonic activation of synaptic NMDARs at the spinal cord level in neuropathic pain (Chen et al., 2018). Our findings suggest that NMDARs that are present at primary afferent central terminals are latent because of limited availability of α2δ-1 and its interaction with NMDARs. This notion is supported by our findings that inhibiting α2δ-1 with gabapentin or disrupting the α2δ-1–NMDAR complex reversed the potentiated NMDAR activity in the spinal dorsal horn caused by ablating HDAC2 in DRG neurons. Thus, HDAC2 in DRG neurons normally restrains nociceptive input to the spinal cord by inhibiting α2δ-1 expression and ensuing NMDAR activity at primary afferent central terminals.
RNA-sequencing data from Hdac2-cKO mice confirmed that Cacna2d1 is the key nociceptive gene regulated by HDAC2 in the DRG. We cannot, however, exclude the possibility that other gene targets are involved in HDAC2 deficiency-induced pain hypersensitivity. By inspecting the list of differentially regulated genes from RNA-sequencing data, we noted that some gene targets implicated in promoting pain, such as interleukin-1β (Il1b) and purinergic receptor P2Y1 (P2ry1), are also upregulated in the DRG of Hdac2-cKO mice. It is possible that the transcription of Il1b and P2ry1 is also controlled directly by HDAC2. Interleukin-1β and purinergic receptor P2Y1 may promote nociception by increasing the excitability of DRG neurons (Nakamura and Strittmatter, 1996; Stemkowski and Smith, 2012). However, their roles in the development of chronic pain are uncertain because nerve injury reduces the expression level of P2ry1 in the DRG (Laumet et al., 2015) and because intrathecal injection of interleukin-1β attenuates inflammatory pain (Souter et al., 2000). Importantly, we showed in this study that HDAC2 knockdown with siRNA had little effect in inducing pain hypersensitivity in Cacna2d1 KO mice. This finding further supports our interpretation that HDAC2 in DRG neurons constitutively restrains pain hypersensitivity predominantly via regulating α2δ-1 expression.
HDAC subtypes have diverse functions in the nervous system (Broide et al., 2007; Guan et al., 2009). Previous studies, largely using nonspecific HDAC inhibitors, provided only conflicting results about the functions of individual HDAC subtypes in pain regulation. In this regard, some HDAC inhibitors, including suberoylanilide hydroxamic acid, LG325, and MS-275, slightly reduce nerve injury-induced neuropathic pain (Denk et al., 2013; Laumet et al., 2015; Sanna et al., 2017), whereas other HDAC inhibitors, such as JNJ-26481585, cause pain hypersensitivity in normal animals (Capasso et al., 2015). Interestingly, systemic treatment with JNJ-26481585 increases α2δ-1 expression in the spinal cord, and gabapentin effectively reverses mechanical pain hypersensitivity caused by JNJ-26481585 (Capasso et al., 2015). Thus, by extension, our findings suggest that JNJ-26481585 may cause pain hypersensitivity mainly via inhibiting HDAC2 activity in the DRG and spinal cord.
In summary, we discovered that HDAC2 in primary sensory neurons is a pivotal epigenetic regulator of synaptic plasticity and nociception. The physiological role of HDAC2 is to function as a key transcriptional repressor to restrain pain hypersensitivity by constitutively inhibiting α2δ-1 expression in primary sensory neurons and ensuing presynaptic NMDAR activity in the spinal dorsal horn. Correspondingly, the loss of HDAC2 binding at the Cacna2d1 promoter could facilitate the transition from acute to chronic pain after nerve injury by potentiating α2δ-1 expression and α2δ-1–dependent synaptic NMDAR activity. This information not only advances our mechanistic understanding of epigenetic control of synaptic plasticity in neuropathic pain but also can guide the treatment of adverse effects of HDAC inhibitors that are clinically used for treating several types of cancers. Pain is likely associated with HDAC inhibitors that impair HDAC2 activity, and α2δ-1–bound NMDARs can be targeted for treating this painful condition. Furthermore, because of long-lasting α2δ-1 upregulation in neuropathic pain conditions, gabapentinoids relieve pain symptoms only temporarily. Our study suggests that restoring the normal function of HDAC2 and/or reducing histone acetylation at the Cacna2d1 promoter in primary sensory neurons may lead to long-lasting relief of neuropathic pain.
Footnotes
- Received April 13, 2022.
- Revision received October 3, 2022.
- Accepted October 10, 2022.
This work was supported by National Institutes of Health–National Institute of Neurological Disorders and Stroke Grant NS101880 and the N.G. and Helen T. Hawkins Endowment. We thank Sarah Bronson at MD Anderson Cancer Center for proofreading the manuscript.
The authors are employees of the University of Texas System, which currently holds a patent for targeting α2δ-1–bound glutamate receptors for treating diseases and disorders.
- Correspondence should be addressed to Hui-Lin Pan at huilinpan{at}mdanderson.org
- Copyright © 2022 the authors