Positive Regulatory Domain I–binding Factor 1 Mediates Peripheral Nerve Injury–induced Nociception in Mice by Repressing Kv4.3 Channel Expression

C57BL/6 male and female mice weighing 25 to 30 g were obtained from Jie Si Jie Experimental Animal Co., Ltd. (China) and maintained in accordance with the established standards of animal care and procedures of the Institutes of Brain Science and State Key Laboratory of Medical Neurobiology (approval No. 31320103906; Fudan University, Shanghai, China). In addition, all experimental protocols were approved by the Medical Experimental Animal Administrative Committee of Fudan University (Shanghai, China) and were conducted in accordance with the ethical guidelines of the National Institutes of Health (Bethesda, Maryland) and the International Association for the Study of Pain (Washington, D.C.) for care and use of laboratory animals under pain research.¹²  Animals were numbered, randomly assigned to different experimental groups using a random number table, and tested in sequential order. All efforts were made to minimize the number of animals used as well as their suffering. After experimentation, the mice were euthanized via carbon dioxide inhalation. All behavioral tests, electrophysiologic recordings, calcium imaging, and biochemical experiments were performed by experimenters who were blinded to the treatments. All experiments were conducted between 9:00 am and 6:00 pm.

The chronic constriction injury–induced nociception model, a common model for nerve injury, was carried out according to previously published methods.¹³  Briefly, the mice were anesthetized with 40 mg/kg sodium pentobarbital. The left common sciatic nerve was exposed and then loosely ligated with 6-0 silk threads at the sciatic nerve bifurcation at intervals of approximately 1 mm. The sham group also underwent sciatic nerve exposure but without ligation. The complete Freund’s adjuvant–induced nociception model, a model for inflammation-associated acute pain induction, was established in mice as described previously.¹⁴  In brief, 15 µl of complete Freund’s adjuvant (1 mg/ml; Sigma-Aldrich, USA) was subcutaneously injected into the left hind paw, and 15 µl of saline was injected as a control. The capsaicin-induced acute pain mouse model was established as described previously.¹⁵  In brief, 10 µl of vehicle (10% ethanol and 10% Tween 80 in saline) containing 1 μg capsaicin (Sigma-Aldrich) or 10 µl of vehicle alone was injected into the plantar surface of the left hind paw.

Mechanical allodynia was assessed using Von Frey hair (Stoelting, USA) testing to measure the paw withdrawal threshold. The mice were placed in a Plexiglas cage with a wire mesh floor and were allowed to habituate for 30 min before testing. This habituation was continued for 2 to 3 days, and during this time, the basal pain hypersensitivity threshold was recorded. Measurements were performed according to the previously published method.¹⁶  Briefly, a series of calibrated Von Frey hairs (0.04, 0.07, 0.16, 0.4, 0.6, 1.0, and 1.4 g) were used to vertically stimulate the central part of the plantar surface of the hind paw. The response to each hair was measured five times at 5-min intervals. The paw withdrawal threshold was defined as the lowest hair force in grams that produced at least three withdrawal responses in five tests.

Thermal hyperalgesia was measured using an IITC Plantar Analgesia (IITC Life Science Inc., USA) to measure the paw withdrawal latency.¹⁷  The mice were placed into Plexiglas chambers with a temperature-controlled glass platform. The heat source was focused on the central part of the plantar surface of the hind paw. The stimulus was shut off when the hind paw moved or after 20 s to prevent tissue damage. The time from the onset of radiant heat to the endpoint was the paw withdrawal latency. The radiant heat intensity was adjusted to obtain a basal paw withdrawal latency of 10 to 12 s. Thermal stimuli were delivered three times to each hind paw at 5-min intervals.

Measurements of cold hyperalgesia were based on a previously published method.¹⁸  Powdered dry ice was compacted into the barrel of a 12-mm-diameter plastic syringe and used to deliver a cold stimulus to the plantar surface. Briefly, mice were enclosed in individual Plexiglas chambers on an elevated glass platform that was 2.5 mm thick, and cold stimulation was delivered to the animal’s plantar hind paw from below. Care was taken to ensure a smooth surface of the dry ice, and light pressure was applied to make contact between the dry ice and the glass. The withdrawal latency was measured with a stopwatch. Withdrawal was defined as any action to move the paw away from the cold glass. The maximum time allowed for withdrawal was 20 s to avoid potential tissue damage, and cold stimuli were delivered three times to each hind paw at 5-min intervals.

Mice were placed on a moving rotarod (Ugo Basile, Italy) programmed to accelerate from 5 to 40 revolutions per minute, over 300 s. After 3 days of training, the time the mice managed to remain on the rotarod was averaged over three consecutive trials.

The CatWalk XT system (Noldus Information Technology, The Netherlands) was used for quantitative assessment of gait parameters and footfalls in rodents. CatWalk is a verified system in the research of several pain models, such as spinal cord injury, osteoarthritis, and neuropathic pain, and has proven to be a reliable method for measuring nonreflexive nociceptive-associated behavior.^19,20  The test involves mice passing through a dark tunnel with a glass plate from one side to the other. Their footprints are illuminated by fluorescent light from the glass plate and captured by a high-speed camera positioned underneath the plate. The captured images are processed by CatWalk XT software. Data were considered for mice that passed through the camera’s photographing area between 0.5 and 5 s and within a maximum run variation of 60% or less. The print area (the contacting area between the hind paw and glass) and the single stance (the duration of the contralateral or ipsilateral hind paw touching the glass plate in the step cycle) were measured to assess nociceptive response.

Dorsal root ganglion microinjection was performed as described previously.²¹  Briefly, the mice were anesthetized with sodium pentobarbital, a midline incision was made in the lower lumbar back region, and the lumbar articular process was exposed and removed. The exposed dorsal root ganglion was injected with adeno-associated virus 5 containing PRDM1 short hairpin RNA and a green fluorescent protein expression marker (adeno-associated virus 5–PRDM1–short hairpin RNA–green fluorescent protein; 3.4E + 12 virus genomes [v.g.]/ml); adeno-associated virus 5–scrambled–short hairpin RNA–green fluorescent protein (3.7E + 12 v.g./ml); adeno-associated virus 5–human synapsin–PRDM1-3×Flag (3.8E + 12 v.g./ml); or adeno-associated virus 5–human synapsin–green fluorescent protein (3.4E + 12 v.g./ml) viral solution (0.5 µl) through a glass micropipette connected to a micro syringe with heat glue and filled with mineral oil. The pipette remained 10 min after injection. The short hairpin RNA targeting the sequence of mouse PRDM1 (Gene Bank Accession NM_007548.4) was designed as follows: 5’-GCATTAGTTCCAATGGCTT-3’. A scrambled sequence was also designed as a negative control (Scram): 5’-CGCTGAGTACTTCGAAATGTC-3’. The adeno-associated virus vectors were designed and generated by Shanghai Taitool Bioscience (China). Intraspinal injection was performed as described previously.²²  The mice underwent hemilaminectomy at the thoracic 13−lumbar 1 vertebral segments. After exposure of the spinal cord, each mouse received adeno-associated virus 5–human synapsin–GCaMP6s (6.5E + 12 v.g./ml) viral solution (0.5 µl). The tip of the glass micropipette reached the superficial laminae of the spinal dorsal horn.

Dorsal root ganglions were lysed with TRIzol reagent (Invitrogen, United Kingdom), and total RNA was extracted according to the manufacturer’s instructions. The mRNA level was then measured using reverse transcription quantitative polymerase chain reaction performed with SYBR Green real-time polymerase chain reaction Mater Mix (Toyobo, Japan). Gene expression was normalized to the expression of the standard housekeeping gene glyceraldehyde-3-phosphate dehydrogenase using the ΔΔCT method. Polymerase chain reaction was performed under the following conditions: 95°C for 3 min followed by 45 cycles at 95°C for 10 s, 65°C for 30 s, 72°C for 10 s, and 72°C for 10 min. Primer sequences are listed in Supplemental Digital Content 1 (https://links.lww.com/ALN/C522).

Mice were deeply anesthetized with sodium pentobarbital and perfused with 4% paraformaldehyde. The lumbar 4 and lumbar 5 dorsal root ganglions were harvested. Dorsal root ganglion sections were cut on a cryostat (Leica 1900, Leica, Germany) and mounted onto slides for immunofluorescence. After being blocked with 5% donkey serum and 0.5% Triton X-100 for 2 h at room temperature, the sections were incubated overnight at 4°C with primary antibodies, including rabbit anti-PRDM1 (1:200, Cell Signaling, USA), mouse anti-NeuN (1:200, EMD Millipore, Germany), mouse anti–glutamine synthetase (1:200, EMD Millipore), mouse anti–calcitonin gene-related peptide (1:200, Abcam, USA), biotinylated isolectin B4 (1:400, Sigma-Aldrich), mouse anti–neurofilament-200 (1:400, Sigma-Aldrich), mouse anti-Kv4.3 (1:100, Sigma-Aldrich), and chicken anti–green fluorescent protein (1:3,000, Aves Labs). Sections were then incubated in Alexa Fluor 488– or Alexa Fluor 594–conjugated secondary antibodies (1:200, Thermal Scientific, USA) containing Hoechst for 2 h at room temperature. Images of samples were collected using confocal microscopy (FluoView FV1000; Olympus, Japan). The numbers of PRDM1-positive neurons were calculated manually or using National Institutes of Health ImageJ Software.

After mice were deeply anesthetized with sodium pentobarbital, the unilateral lumbar 4 and lumbar 5 dorsal root ganglions were removed and immediately frozen in liquid nitrogen. The dorsal root ganglions were ground in cold (4°C) radioimmunoprecipitation assay lysis buffer (Beyotime Institute of Biotechnology, China) containing a cocktail of protease inhibitors. After incubation on ice for 15 min, homogenates were centrifuged at 12,000g for 20 min at 4°C. Protein concentrations were measured using a BCA Protein Assay Kit (Beyotime Institute of Biotechnology). Protein samples were then denatured at 95°C and separated by 10% sodium dodecyl sulphate-polyacrylamice gel electrophoresis. After the samples were transferred to polyvinylidene fluoride membranes and blocked with 5% nonfat milk, the membranes were incubated overnight at 4°C with primary antibodies, including rabbit anti-PRDM1 (1:200, Cell Signaling), rabbit anti-Kv4.3 (1:1,000, Alomone Lab, Israel), mouse anti-Flag (1:1,000, GNI, Japan), and rabbit anti–glyceraldehyde-3-phosphate dehydrogenase (1:1,000, Cell Signaling). The membranes were then incubated with horseradish peroxidase–conjugated anti-mouse or anti-rabbit secondary antibodies for 2 h at room temperature. Signals were finally detected using enhanced chemiluminescence (Pierce), and the bands were visualized with the FluorChem E system (ProteinSimple, USA).

Total RNA was extracted using the miRNeasy Mini Kit (Qiagen, GmBH, Germany) according to the manufacturer’s instructions, and the RNA integrity number was assessed by an Agilent Bioanalyzer 2100 (Agilent Technologies, USA). Qualified total RNA was further purified using the RNAClean XP Kit (Beckman Coulter, Inc., USA) and RNase-Free DNase Set (Qiagen, GmBH). RNA sequencing libraries were prepared according to the Illumina TruSeq protocol (Illumina, USA). Levels of gene expression were determined using edgeR software (Illumina),²³  with discovery rate adjustment, and were reported in fragments per kilobase of transcript per million mapped reads. A gene was considered to be expressed if it had a fragments per kilobase per million greater than 1. For a gene to be called differentially expressed, it had to meet two criteria: a greater than 2.0-fold expression level change and P value < 0.05.

Fresh dorsal root ganglions were rapidly removed from anesthetized mice and stabilized for 3 min using stabilization buffer. Then, the dorsal root ganglions were incubated in 1% formaldehyde for 20 min at room temperature, and the reaction was stopped by adding glycine. After being washed three times with phosphate-buffered saline, the dorsal root ganglions were incubated in lysis buffer for 30 min on ice. Finally, the dorsal root ganglions were sonicated (30 s on: 30 s off) in chromatin immunoprecipitation dilution buffer using a water bath sonicator at 4°C for 10 min. Chromatin was pulled down using the following antibodies: nonspecific mouse immunoglobulin G (Bangs Laboratories, USA) and PRDM1 (Cell Signaling). Ten percent of the sample for immunoprecipitation was used as input. After chromatin precipitation, quantitative polymerase chain reaction was performed using the primers shown in Supplemental Digital Content 2 (https://links.lww.com/ALN/C523).

Clearing of the dorsal root ganglion and spinal cord with clear, unobstructed brain imaging cocktails was performed as described previously.²⁴  After fixation with 4% paraformaldehyde overnight, dorsal root ganglion and spinal cord samples were washed three times with phosphate-buffered saline for 1 h each with shaking at room temperature. The samples were then cleared with 1/2-Reagent-1 (a 1:1 mixture of Reagent-1 [urea, quadrol, Triton X-100, and distilled H₂O] and phosphate-buffered saline), with shaking at 60 revolutions per minute at 37°C for 3 h. Reagent-1 was applied for further clearing under the same conditions and replaced every 2 days until distinct optical transparency was achieved. The samples were then washed with phosphate-buffered saline with gentle shaking three times at room temperature for 1 h each. The samples were incubated with 4’,6-diamidino-2-phenylindole with shaking at 60 revolutions per minute at 37°C for 24 h followed by three washes with phosphate-buffered saline with shaking at room temperature for 1 h each. The 1/2-Reagent-2 (a 1:1 mixture of Reagent-1 [urea, sucrose, triethanolamine, and distilled H₂O] and phosphate-buffered saline) was then applied to clear the sample a final time with shaking at 60 revolutions per minute at 37°C for 24 h, and Reagent-2 was applied for further clearing and replaced every day until distinct optical transparency was achieved. All clearing procedures were performed in darkness to avoid fluorescence bleaching.

Three-dimensional images of transparent dorsal root ganglion and spinal cord samples were acquired using a light-sheet microscope (Lightsheet Z.1, Carl Zeiss, Germany). All images were processed with Imaris 9 (Bitplane, Switzerland) or Arivis (Vision4D, Germany), and surface visualization was used for three-dimensional rendering.

Four weeks after viral injection into the dorsal root ganglions, freshly dissociated dorsal root ganglion cultures were prepared. Cells were placed on glass coverslips coated with poly-D-lysine and grown in neurobasal-defined medium at 37°C with 5% CO₂ for 24 h before experiments. Whole cell current clamp recordings were performed at room temperature to measure action potentials and the resting membrane potential with an Axonpatch 200B amplifier and a Digidata 1440A digitizer (Axon Instruments, USA). The recording chamber (300 µl) was continuously superfused at a flow rate of 1 to 2 ml/min. Series resistance was compensated (>80%), and leak subtraction was performed. Only green fluorescent protein–positive cells were selected for study. The pipette solution contained the following (in mM): 126 K-gluconate, 10 NaCl, 1 MgCl₂, 10 EGTA, 2 Na-ATP, and 0.1 Mg-GTP, adjusted to pH 7.3 with KOH. The external solution contained the following (in mM): 140 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 10 N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid, and 10 glucose, adjusted to pH 7.4 with NaOH. In current clamp experiments, the action potentials were evoked by current injection steps. The resting membrane potential was measured without a current injection.

For in vivo Ca²⁺ imaging, GCaMP6s-expressing mice were anesthetized by sodium pentobarbital supplemented with 1 to 2% isoflurane in air if needed. Body temperatures were maintained at 37°C with a heating blanket. The skin was incised at the thoracic 12−lumbar 2 level. The paravertebral muscle overlying thoracic 12−lumbar 2 was removed. The laminectomy was performed at thoracic 13−lumbar 1, and a custom-made spinal chamber was attached. The laminectomized region was then covered with agarose (type III-A, Sigma-Aldrich; 1% in Ringer solution), and a glass coverslip was placed to seal the laminectomized area. Two-photon Ca²⁺ imaging was performed using an Olympus FVMPE-RS with a 25 × 1.05 WMP water-immersion lens, and the fluorescence signals were measured at a frame scanning rate of 30 Hz (512 × 512 pixel images). The InSight X3 laser (Spectra Physics, USA) was tuned at 920 nm for two-photon excitation of GCaMP6s. Movies were imported into ImageJ (http://rsbweb.nih.gov/ij/), and fluorescence traces were analyzed from somatic regions of interest and expressed as relative percentage changes (Δfluorescence/fluorescence) after background subtraction. Thermal pain stimulation was performed using a custom-built heating plate applied to the plantar surface of the hind paw. The temperature of the heating plate was increased from a baseline temperature of 25°C to 45°C or decreased from 45°C to 25°C. Temperature changes occurred as a ramp of 0.7°C/s to the target temperature of 45°C (maintained for 10 s) before returning to baseline.

Data are presented as mean ± SD. No formal a priori statistical power calculations were conducted, and the sample sizes were based on our previous knowledge and experience with this design. There were no missing data. All data from different groups were verified for normality and homogeneity of variance using Kolmogorov–Smirnov and Brown–Forsythe tests before analysis. For pain hypersensitivity behavioral experiments, differences between groups for threshold, latency, print area, and single stance results were determined using two-way repeated-measures ANOVA followed by the post hoc Bonferroni multiple comparison test. For reverse transcription quantitative polymerase chain reaction, Western blot, and chromatin immunoprecipitation–quantitative polymerase chain reaction experiments, group differences were determined using paired or independent unpaired two-tailed Student’s t tests and one-way ANOVA followed by the Tukey post hoc test. Electrophysiologic data were tested using two-way ANOVA (for multiple time points) followed by post hoc Bonferroni or unpaired two-tailed Student’s t tests. For in vivo Ca²⁺ imaging experiments, group differences were determined using unpaired two-tailed Student’s t tests. Differences were considered statistically significant if P < 0.05. No outliers were observed, and no data were excluded from statistical analyses. All statistical analyses were performed using GraphPad Prism 7.0 software (GraphPad, USA).

Peripheral nerve injury causes many molecular changes in the dorsal root ganglion, and these changes are critical for the development of neuropathic pain. To determine whether PRDM1 is modulated in response to peripheral nerve injury, we examined the expression of PRDM1 in the mouse dorsal root ganglion after chronic constriction injury.¹³  Compared with the contralateral side, the ipsilateral (injured) lumbar 4 and lumbar 5 dorsal root ganglions expressed statistically higher levels of Prdm1 mRNA on days 3 (fold change: 2.0 ± 0.8; two-tailed paired t test, t₍₅₎ = 4.01; n = 6; P = 0.032) and 7 (fold change: 1.6 ± 0.6; two-tailed paired t test, t₍₅₎ = 4.30; n = 6; P = 0.022) after chronic constriction injury (fig. 1A). As a control, sham surgery did not affect the basal mRNA expression level of Prdm1 in ipsilateral lumbar 4 and lumbar 5 dorsal root ganglions (fig. 1B). Furthermore, PRDM1 protein was expressed at higher levels on days 3 (fold change: 3.1 ± 0.6; two-tailed paired t test, t₍₅₎ = 15.28; n = 6; P < 0.0001) and 7 (fold change: 2.7 ± 0.6; two-tailed paired t test, t₍₅₎ = 12.28; n = 6; P < 0.0001) but not on day 14 after chronic constriction injury (fig. 1C). As a control, sham surgery did not alter the basal expression of the protein level of PRDM1 in ipsilateral lumbar 4 and lumbar 5 dorsal root ganglion (fig. 1D). The number of PRDM1-positive neurons was also greater in the injured dorsal root ganglion compared with the dorsal root ganglion of the sham group on day 7 after chronic constriction injury (chronic constriction injury day 7: 81.5 ± 7.0% vs. sham day 7: 17.8 ± 3.1%; two-tailed unpaired t test, t₍₁₀₎ = 20.42; n = 6; P < 0.0001; fig. 1, E and F).

We further examined whether other pain hypersensitivity models altered PRDM1 expression in the dorsal root ganglion. A similar increase in PRDM1 expression was found after plantar injection of complete Freund’s adjuvant. The levels of Prdm1 mRNA (Supplemental Digital Content 3, https://links.lww.com/ALN/C524, fig. S1A) and protein (Supplemental Digital Content 3, https://links.lww.com/ALN/C524, fig. S1, B and C) on days 3 and 7 after complete Freund’s adjuvant injection on the injured side were higher than those on the contralateral side. However, saline injection did not change the basal expression of Prdm1 mRNA (Supplemental Digital Content 3, https://links.lww.com/ALN/C524, fig. S1D) or protein (Supplemental Digital Content 3, https://links.lww.com/ALN/C524, fig. S1, E and F) in ipsilateral lumbar 4 and lumbar 5 dorsal root ganglions. We further determined whether acute pain also changes PRDM1 expression in the dorsal root ganglion by injecting capsaicin into the plantar side of the hind paw. Unexpectedly, the amount of PRDM1 protein was not altered in the ipsilateral lumbar 4 or lumbar 5 dorsal root ganglion (Supplemental Digital Content 3, https://links.lww.com/ALN/C524, fig. S1, G and H), suggesting that the response is limited to chronic pain.

To further characterize PRDM1 expression after nociceptive pain, we analyzed its distribution pattern in the mouse dorsal root ganglion. By co-labeling PRDM1 with NeuN (a neuronal marker) or glutamine synthetase (a marker for satellite glial cells) and Hoechst (a marker for cellular nuclei), we determined that PRDM1 was localized in the nuclei of dorsal root ganglion neurons (fig. 2A) and could not be detected in satellite glial cells (fig. 2B). We further examined the distribution of PRDM1 in various subpopulations of dorsal root ganglion neurons. The increased PRDM1 was mainly distributed among nociceptive neurons expressing three different marker proteins: calcitonin gene-related peptide (fig. 2C), isolectin B4 (fig. 2D), and neurofilament-200 (fig. 2E). We also performed a control experiment to verify PRDM1 antibody specificity via injection of short hairpin RNA in the mouse dorsal root ganglion (Supplemental Digital Content 4, https://links.lww.com/ALN/C525, fig. S2). Taken together, our results show that PRDM1 expression in dorsal root ganglion nociceptive neurons increases in the early stage of peripheral nerve injury.

To determine the role of PRDM1 in normal pain perception and peripheral nerve injury–induced nociception in the dorsal root ganglion, we injected adeno-associated virus 5–expressing Prdm1 short hairpin RNA or control scrambled short hairpin RNA (Scram) into ipsilateral lumbar 4 and lumbar 5 mouse dorsal root ganglions 4 weeks before chronic constriction injury and assessed nociceptive response (fig. 3A). Four weeks after the injection, we observed the expression of the virus in the dorsal root ganglion by immunofluorescence staining (Supplemental Digital Content 5, https://links.lww.com/ALN/C526, fig. S3A). We used tissue clearing technology to make the dorsal root ganglion transparent and performed three-dimensional image reconstruction to observe the expression of the virus in the dorsal root ganglion (Supplemental Digital Content 6, https://links.lww.com/ALN/C527). As expected, Prdm1 short hairpin RNA reduced basal expression and blocked the chronic constriction injury–induced increase in PRDM1 protein expression in ipsilateral lumbar 4 and lumbar 5 dorsal root ganglion on day 7 after chronic constriction injury (fold change: naive + short hairpin RNA: 0.3 ± 0.2 vs. naive + Scram: 1.0 ± 0.4; chronic constriction injury + short hairpin RNA: 0.9 ± 0.2 vs. chronic constriction injury + Scram: 2.2 ± 0.7; one-way ANOVA; n = 6; P < 0.001; fig. 3, B and C) and mRNA (fold change: naive + short hairpin RNA: 0.3 ± 0.1 vs. naive + Scram: 1.0 ± 0.5; chronic constriction injury + short hairpin RNA: 1.2 ± 0.7 vs. chronic constriction injury + Scram: 1.9 ± 0.4; one-way ANOVA; n = 6; P = 0.040; fig. 3D). In behavioral assays, PRDM1 knockdown increased the paw withdrawal threshold to mechanical stimulation (short hairpin RNA: 3.5 ± 1.4 g vs. Scram: 1.2 ± 0.2 g; two-way ANOVA; n = 8; P < 0.0001; fig. 3E) and paw withdrawal latency to thermal (short hairpin RNA: 13.0 ± 0.9 s vs. Scram: 10.6 ± 1.6 s; two-way ANOVA; n = 8; P = 0.021; fig. 3F) or cold (short hairpin RNA: 6.9 ± 0.9 s vs. Scram: 5.2 ± 0.8 s; two-way ANOVA; n = 8; P = 0.001; fig. 3G) stimulation on the ipsilateral side in naive mice. PRDM1 knockdown also attenuated the hyperalgesia induced by chronic constriction injury. Consistently, injection of Prdm1 short hairpin RNA also reduced the mechanical, thermal, and cold hyperalgesia caused by complete Freund’s adjuvant injection (Supplemental Digital Content 5, https://links.lww.com/ALN/C526, fig. S3, B through D). However, dorsal root ganglion injection or knockdown of PRMD1 did not alter mouse motor activity (Supplemental Digital Content 5, https://links.lww.com/ALN/C526, fig. S3E).

In addition, we employed CatWalk gait analysis to evaluate nonreflexive nociceptive behavior after knockdown of PRMD1 in dorsal root ganglion. We measured alterations in two parameters for the ipsilateral hind paw of chronic constriction injury mice: the print area and the single stance (% ipsilateral/contralateral). Our results showed that PRDM1 knockdown reversed the chronic constriction injury–induced decrease in the print area (short hairpin RNA: 83.1 ± 14.3% vs. Scram: 21.8 ± 19.0%; two-way ANOVA; n = 8; P < 0.001; fig. 3H) and single stance (short hairpin RNA: 82.3 ± 17.4% vs. Scram: 29.8 ± 17.9%; two-way ANOVA; n = 8; P < 0.0001; fig. 3I). Representative images of the print area in the Scram and short hairpin RNA mouse groups are shown (fig. 3, J and K). We also repeated these experiments to verify the effect of PRDM1 knockdown in the dorsal root ganglion on the pain threshold and neuropathic pain hypersensitivity in female mice with similar results (Supplemental Digital Content 7, https://links.lww.com/ALN/C528).

To assess potential therapeutic effects of PRMD1 knockdown in treating established nociceptive response, we injected short hairpin RNA 7 days after the chronic constriction injury surgery. Our results show that chronic constriction injury–induced mechanical allodynia (fig. 4A), thermal hyperalgesia (fig. 4B), and cold hyperalgesia (fig. 4C) were relieved 2 weeks after dorsal root ganglion injection of PRDM1 short hairpin RNA. Furthermore, dorsal root ganglion injection of PRDM1 short hairpin RNA also reversed chronic constriction injury–induced decreased in the percentage of the print area (chronic constriction injury + short hairpin RNA: 68.0 ± 13.3% vs. chronic constriction injury + Scram: 22.8 ± 12.8%; two-way ANOVA; n = 8; P < 0.0001; fig. 4D) and single stance (chronic constriction injury + short hairpin RNA: 75.3 ± 15.6% vs. chronic constriction injury + Scram: 45.1 ± 15.1%; two-way ANOVA; n = 8; P = 0.046; fig. 4E). Representative images show the print area in the sham + Scram, chronic constriction injury + Scram, and chronic constriction injury + short hairpin RNA male mouse groups (fig. 4, F and G). Similar results were obtained in female mice (Supplemental Digital Content 8, https://links.lww.com/ALN/C529). Taken together, our results suggest that knockdown of PRDM1 can relieve hyperalgesia induced by peripheral nerve injury.

To determine whether PRDM1-dependent effects on nociceptive response in injured dorsal root ganglion neurons may be accounted for by an impact on neuronal excitability, we performed whole cell current clamp recordings on day 7 after chronic constriction injury, with assessment of the action potentials and resting membrane potential in ipsilateral lumbar 4 and lumbar 5 dorsal root ganglion neurons from mice injected with Prdm1 short hairpin RNA or Scram. Compared with neurons in the Scram group, those in the Prdm1 short hairpin RNA group showed a lower action potential firing rate in response to superimposed positive current injections ranging from 90 to 160 pA (fig. 5, A and B), but no difference was observed in action potential amplitude (fig. 5C). Furthermore, injection of Prdm1 short hairpin RNA into the dorsal root ganglion hyperpolarized the resting membrane potential (chronic constriction injury + short hairpin RNA: −61.7 ± 11.4 mV vs. chronic constriction injury + Scram: −50.0 ± 10.5 mV; two-tailed unpaired t test, t₍₅₃₎ = 3.96; n = 25 in the chronic constriction injury + short hairpin RNA group and n = 30 in the chronic constriction injury + Scram group; P < 0.001; fig. 5D) and increased the rheobase (chronic constriction injury + short hairpin RNA: 36.8 ± 25.1 pA vs. chronic constriction injury + Scram: 23.7 ± 15.0 pA; two-tailed unpaired t test, t₍₅₃₎ = 2.40; n = 25 in the chronic constriction injury + short hairpin RNA group and n = 30 in the chronic constriction injury + Scram group; P = 0.020; fig. 5E) and action potential threshold (chronic constriction injury + short hairpin RNA: −28.9 ± 11.1 mV vs. chronic constriction injury + Scram: −16.5 ± 8.4 mV; two-tailed unpaired t test, t₍₅₃₎ = 4.68; n = 25 in the chronic constriction injury + short hairpin RNA group and n = 30 in the chronic constriction injury + Scram group; P < 0.0001; fig. 5F). Collectively, these results suggest that endogenous PRDM1 regulates the excitability in injured dorsal root ganglion neurons, whereas abnormal firing of injured dorsal root ganglion neurons is normalized by PRDM1 knockdown.

To further confirm that the increase in PRDM1 in the dorsal root ganglion at the early stage of nerve injury can indeed induce nociception, we injected adeno-associated virus 5 that expresses full-length Prdm1 (adeno-associated virus 5–Prdm1) into the unilateral lumbar 4 and lumbar 5 dorsal root ganglions of naive mice. Adeno-associated virus 5–green fluorescent protein was used as a control. Four weeks after injection, the expression levels of Prdm1 mRNA (fold change: PRDM1 overexpression: 5.2 ± 1.6 vs. green fluorescent protein: 1.0 ± 0.3; two-tailed unpaired t test, t₍₁₀₎ = 6.23; n = 6; P < 0.0001; fig. 6A) and PRDM1 protein (fold change: PRDM1 overexpression: 2.5 ± 0.7 vs. green fluorescent protein: 1.0 ± 0.4; two-tailed unpaired t test, t₍₆₎ = 4.68; n = 6; P = 0.001; fig. 6, B and C) in the dorsal root ganglion of adeno-associated virus 5–Prdm1–injected mice were higher than those in the dorsal root ganglion of adeno-associated virus 5–green fluorescent protein–injected mice. The overexpression of Flag-tagged PRDM1 was verified by Western blotting with an anti-Flag antibody (fig. 6D). Behavioral results showed that the paw withdrawal threshold to mechanical stimulation (PRDM1 overexpression: 0.6 ± 0.3 g vs. green fluorescent protein: 1.2 ± 0.2 g; two-way ANOVA; n = 13; P < 0.0001; fig. 6E) and paw withdrawal latency to thermal (PRDM1 overexpression: 5.2 ± 1.3 s vs. green fluorescent protein: 9.8 ± 1.7 s; two-way ANOVA; n = 13; P < 0.0001; fig. 6F) and cold (PRDM1 overexpression: 3.4 ± 0.5 s vs. green fluorescent protein: 5.3 ± 0.6 s; two-way ANOVA; n = 13; P < 0.0001; fig. 6G) stimulation of mice injected with adeno-associated virus 5–Prdm1 were lower than those of mice injected with adeno-associated virus 5–green fluorescent protein. Similar results were also observed in female mice (Supplemental Digital Content 9, https://links.lww.com/ALN/C530). These findings indicate that overexpression of PRDM1 in the dorsal root ganglion of naive mice leads to mechanical allodynia and thermal and cold hyperalgesia, which are similar to nociception symptoms.

To identify genes that may be regulated by PRDM1 in the injured dorsal root ganglion after chronic constriction injury, we performed RNA sequencing analysis on injured dorsal root ganglions from short hairpin RNA– and Scram-injected mice. The scatter plot shows the differentially expressed genes (fig. 7A), and the differentially expressed gene heat map depicts their hierarchical clustering (fig. 7B). Among the differentially expressed genes, Cacna1g (Ca_V3.1),²⁵  Cxcl5,²⁶  Cxcl12,²⁷ Cxcr2,²⁸ Ednra,^29,30 Htr2b (5-HT_2B),³¹ Kcnj2 (Kir2.1),³² Mmp9,³³  and Prkcg (PKCγ)^34,35  are classically involved in peripheral sensitization of neuropathic pain (table 1). The identification of differentially expressed genes by RNA sequencing was confirmed by reverse transcription quantitative polymerase chain reaction, for which we selected seven differentially expressed genes for validation (fig. 7, C through I). Among these differentially expressed genes, only Kcnd3 was upregulated after PRDM1 knockdown. This gene encodes Kv4.3, a member of the A-type voltage-gated K⁺ channel family that is selectively expressed in nociceptive dorsal root ganglion neurons and is inversely associated with the development of chronic pain.^36,37  Suppression of Kcnd3 expression by PRDM1 in injured dorsal root ganglions might contribute to the function of PRDM1 in promoting neuron excitability and nociceptive response. Because PRDM1 acts as a transcriptional repressor, the downregulation of the other genes may be associated with decreased peripheral sensitization after PRDM1 knockdown rather than direct suppression by PRDM1.

To verify a role of Kv4.3 in mediating PRDM1-dependent peripheral nerve injury–induced nociception in the dorsal root ganglion, we evaluated Kv4.3 protein expression after PRDM1 knockdown. Injection of Prdm1 short hairpin RNA but not Scram increased the basal expression level of Kv4.3 protein in lumbar 4 and lumbar 5 dorsal root ganglions of sham-operated mice (fold change: sham + short hairpin RNA: 1.5 ± 0.5 vs. sham + Scram: 1.0 ± 0.2; one-way ANOVA; n = 6; P = 0.026) and statistically reversed the decrease in Kv4.3 protein expression on day 7 after chronic constriction injury in injured dorsal root ganglions (fold change: chronic constriction injury + short hairpin RNA: 1.2 ± 0.3 vs. chronic constriction injury + Scram: 0.4 ± 0.1; one-way ANOVA; n = 6; P = 0.003; fig. 8, A and B). Furthermore, overexpression of PRDM1 through injection of Prdm1-overexpressing virus into the dorsal root ganglion of naive mice reduced the levels of Kcnd3 mRNA (fold change: PRDM1 overexpression: 0.4 ± 0.2 vs. green fluorescent protein: 1.0 ± 0.3; two-tailed unpaired t test, t₍₁₀₎ = 4.39; n = 6; P = 0.001; fig. 8C) and Kv4.3 protein (fold change: PRDM1 overexpression: 0.4 ± 0.1 vs. green fluorescent protein: 1.0 ± 0.3; two-tailed unpaired t test, t₍₆₎ = 2.98; n = 6; P < 0.001; fig. 8, D and E). These data strongly suggest that PRDM1 is involved in downregulation of the Kv4.3 channel after peripheral nerve injury.

To test whether the Kcnd3 gene is a direct target of PRDM1, we performed site-directed chromatin immunoprecipitation–quantitative polymerase chain reaction assays with an antibody specific for PRDM1 at five different regions of the Kcnd3 gene locus (fig. 8F). The PRDM1-binding activity of two regions (R3 and R4) of the Kcnd3 gene in the injured dorsal root ganglion was higher in the chronic constriction injury group than in the sham group on day 7 after chronic constriction injury (fig. 8G). Thus, these results are consistent with the possibility that PRDM1 transcriptionally represses the Kcnd3 gene in the injured dorsal root ganglion via enhanced binding to the Kcnd3 gene promoter region.

To directly visualize spinal dorsal horn neuron responses after dorsal root ganglion PRDM1 knockdown, we performed in vivo two-photon imaging of thermal stimuli–evoked calcium signals in spinal dorsal horn neurons in mice. GCaMP6s virus was injected into the spinal cord at the level of the lumbar 4–lumbar 5 dorsal horn (fig. 9A), and the transfection efficiency was verified by GCaMP6s labeling using immunohistochemistry with a green fluorescent protein antibody 4 weeks after injection (fig. 9B). We used tissue clearing technology to make the spinal cord transparent and then performed three-dimensional image reconstruction processing to clearly observe the expression of the GCaMP6s virus in the spinal dorsal horn (Supplemental Digital Content 10, https://links.lww.com/ALN/C531). Stable images were achieved by using a custom clamp to hold the spinous process (fig. 9C), and consistent labeling of green fluorescent protein was detected in the lumbar 4–lumbar 5 spinal dorsal horn.

Because temperature-sensing dorsal root ganglion neurons are mainly projected to the superficial laminae of spinal dorsal horn neurons, we focused on the changes in calcium signals in neurons less than 200 μm below the surface of the spinal cord. We examined the response to thermal pain hypersensitivity stimulation (target temperature: 45°C; maintained for 10 s) and observed robust, reliable calcium transients in subsets of spinal dorsal horn neurons in control Scram-treated mice (fig. 9, D and F; Supplemental Digital Content 11, Movie S3, https://links.lww.com/ALN/C532). However, we observed a decrease in the number of neurons specifically activated by thermal stimulation in Prdm1 short hairpin RNA–treated mice (fig. 9, E and F; Supplemental Digital Content 12, Movie S4, https://links.lww.com/ALN/C533). Furthermore, the peak response amplitudes of the short hairpin RNA group were lower than those of the Scram group (Δfluorescence/fluorescence: naive + short hairpin RNA: 84.6 ± 19.8% vs. naive + Scram: 210 ± 58.3%; two-tailed unpaired t test, t₍₁₅₄₎ = 17.73; n = 76 cells in naive + short hairpin RNA group and n = 80 cells in naive + Scram group; P < 0.0001; fig. 9G). Together, these results confirm the important role of PRDM1 in the dorsal root ganglion in the perception of peripheral nociceptive stimuli.

Our study reveals the alteration and functional significance of PRDM1 in a peripheral nerve injury–induced nociception animal model. The principle findings are fourfold: (1) peripheral nerve injury leads to an increase in PRDM1 expression in injured dorsal root ganglion neurons; (2) this increase is associated with the downregulation of Kv4.3 channel expression in injured dorsal root ganglion neurons caused by peripheral nerve injury; (3) blocking the increase in PRDM1 restores Kv4.3 channel expression, reduces the hyperexcitability of injured dorsal root ganglion neurons, and attenuates peripheral nerve injury–induced nociception; (4) PRDM1 overexpression in naive mouse dorsal root ganglion neurons diminishes Kv4.3 channel expression and induces hyperalgesia. Because Kv4.3 channel downregulation contributes to nociception genesis,^3,38  PRDM1, as a transcriptional suppressor, likely participates in peripheral nerve injury–induced nociception genesis by repressing Kv4.3 channel expression in the injured dorsal root ganglion.

PRDM1 is a member of the PRDI-BF1 and RIZ homology domain–containing protein family, which contains 16 to 17 family members in mammals. PRDM1 was first identified as a repressor of interferon-β gene expression⁶  but was later discovered to be required for the differentiation of mature B lymphocytes into plasma cells.³⁹  Recently, PRDM1 has also been shown to participate in certain biologic processes in the nervous system. For example, PRDM1 is expressed in both mesodermal cells and neural plates involved in nervous system development. Knockout of PRDM1 in zebrafish leads to a loss of neural crest cells and Rohon–Beard sensory neurons, which ultimately leads to abnormal peripheral nervous system development.¹⁰  The sox10 gene is a primary effector of PRDM1 in the neural crest, and islet1 lies downstream of PRDM1 in the development of Rohon–Beard sensory neurons. PRDM1 is widely expressed in most tissues, including the nervous system, with expression reported in the dorsal root ganglion in mice.¹¹  However, the distribution pattern of PRDM1 in the mouse dorsal root ganglion was not clear before this study. We detected PRDM1 in the nuclei of dorsal root ganglion neurons but not in satellite cells. Furthermore, we revealed that PRDM1 is expressed in calcitonin gene–related peptide–, isolectin B4–, and neurofilament-200–labeled neurons, suggesting that PRDM1 is mainly distributed in medium- and small-sized nociceptive dorsal root ganglion neurons. We demonstrated activation of the PRDM1 gene in the dorsal root ganglion at the transcriptional level, with increased expression of PRDM1 mRNA and protein in the dorsal root ganglion after peripheral nerve injury. PRDM1 upregulation occurred at the onset of behaviorally expressed pain hypersensitivity and persisted for 7 days after peripheral nerve injury. These findings suggest that PRDM1 may function in peripheral nociceptive transmission in the early period after peripheral nerve injury. Consistent with this possibility, injection of Prdm1 short hairpin RNA into mouse dorsal root ganglions specifically blocked the peripheral nerve injury–induced increase in PRDM1 and attenuated nociception. Notably, our findings are also supported by a recent clinical report that reveals that PRDM12 (a member of the PRDI-BF1 and RIZ homology domain–containing family) is essential for human perception of pain, as pathogenic mutations cause a congenital loss of pain perception in humans.⁴⁰  This clinical finding further indicates that the PRDI-BF1 and RIZ homology domain–containing family may play a role in pain perception and provides clinical support for our study.

Peripheral nerve injury can cause the upregulation or downregulation of many dorsal root ganglion genes, including those involved in inflammatory responses,⁴¹  G protein–coupled receptors,³¹  neurotrophins,⁴²  and ion channels.²⁵  Strikingly, our RNA sequencing data showed that knockdown of PRDM1 restored the expression of the Kv4.3 channel gene in injured dorsal root ganglions. The Kv4.3 channel belongs to a family of voltage-gated K⁺ channels that can be activated transiently and inactivated rapidly.⁴³  Many voltage-gated K⁺ channels contribute to the regulation of dorsal root ganglion neuron excitability and pain hypersensitivity.³⁸  Peripheral nerve injury results in the downregulation of voltage-gated K⁺ channel gene expression, reduction in outward voltage-gated K⁺ currents, increased abnormal excitability, and ectopic spontaneous discharge of primary sensory neurons in the dorsal root ganglion, all of which may increase pain hypersensitivity.^4,44  We observed a decrease in Kv4.3 channel expression within 7 days after peripheral nerve injury, which is consistent with a previous report.⁴  A previous study has reported that in the absence of pain stimulation, Prdm1 expression is low and Kv4.3 expression is much higher in sensory neurons of the dorsal root ganglion.⁴⁵  Therefore, there is little overlap in the expression of these two proteins. As shown in the current work, PRDM1 expression is dependent on the pain level. Although reverse coupling of PRDM1 expression with Kv4.3 can lead to the detection of either Prdm1 or Kv4.3 transcripts under certain conditions, these proteins are not usually coexpressed.⁴⁵  However, we cannot rule out that the repression of Kv4.3 in response to nociceptive pain may involve the coordinate function of additional corepressors. Although numerous studies have reported changes in voltage-gated K⁺ gene expression or current changes after peripheral nerve injury, whether these transcriptional changes serve to initiate and maintain peripheral nerve injury–induced voltage-gated K⁺ channel gene decrease in chronic pain remained unresolved before this study. Our results demonstrate that knockdown of PRDM1 restores Kv4.3 channel expression in the injured dorsal root ganglion, whereas mimicking the chronic constriction injury–induced increase in dorsal root ganglion PRDM1 via genetic overexpression decreases the expression of the Kv4.3 channel. These findings suggest that PRDM1 may negatively regulate Kv4.3 channel expression in the dorsal root ganglion after chronic constriction injury. PRDM1, as a transcriptional repressor, modulates the expression of many genes by directly binding DNA or acting as a scaffold to recruit multiple corepressor proteins, such as G9a⁶  and histone deacetylase.⁵  However, peripheral nerve injury–induced gene expression changes likely involve multiple transcriptional regulators, and we therefore performed chromatin immunoprecipitation–quantitative polymerase chain reaction to determine whether PRDM1 regulation of Kv4.3 channel gene expression in peripheral nerve injury–induced nociception may involve direct interaction of PRDM1 at the Kcnd3 gene promoter. Consistent with this possibility, chronic constriction injury increased the occupancy of PRDM1 at the Kcnd3 promoter in injured dorsal root ganglion. Thus, our data provide evidence for the essential role of PRDM1-mediated inhibition of Kv4.3 channel gene expression in dorsal root ganglion neurons in peripheral nerve injury–induced nociception through direct transcriptional repression. Although PRDM1 is generally known for mediating transcriptional repression, our RNA sequencing data revealed that PRDM1 knockdown also normalizes the expression of many upregulated genes in the injured dorsal root ganglion, such as Cacna1g, Cxcl12, and Ednra. The downregulation of these genes in PRDM1 knockdown dorsal root ganglions could potentially be explained by a decrease in peripheral sensitization upon restoration of Kv4.3 channel expression.

Hyperexcitability of dorsal root ganglion neurons is a common phenomenon in various peripheral nerve injury–induced nociception animal models and is often associated with a reduction in K⁺ currents. Among the affected K⁺ channels, voltage-gated K⁺ channels, which are protein complexes composed of ion-conducting integral protein α subunits and auxiliary cytoplasmic β subunits, play an important role in setting the resting membrane potential and controlling action potential repolarization and frequency.⁴  We demonstrated that knockdown of PRDM1 normalizes abnormal firing of injured dorsal root ganglion neurons, indicating that PRDM1 knockdown reduces spontaneous depolarization of the membrane in dorsal root ganglion neurons upon restoration of Kv4.3 channel expression. Peripheral nociceptive information is transmitted to the dorsal horn of the spinal cord through dorsal root ganglion transformation, resulting in increased responsiveness of dorsal horn neurons.⁴¹  Therefore, we assessed spinal dorsal horn neuron responses by recording calcium responses to heat stimuli in Scram-injected and PRDM1 short hairpin RNA–injected mice. Our data reveal that PRDM1 knockdown in the dorsal root ganglion also reduces the response of spinal dorsal horn neurons to peripheral nociceptive stimuli, potentially because of a decrease in the excitatory input of primary nociceptive neurons in the dorsal root ganglion.

In summary, our study reveals that the transcriptional repressor PRDM1 triggers Kv4.3 channel downregulation in the injured dorsal root ganglion after peripheral nerve injury. We demonstrated that blocking the peripheral nerve injury–induced increase in PRDM1 expression in the dorsal root ganglion restores the expression of the Kv4.3 channel, reduces the hyperexcitability of injured dorsal root ganglion neurons, and attenuates nociception (fig. 10). As a limitation, our results were obtained in a mouse model rather than in patients with chronic pain; although experimentation in the mouse has the advantage of enabling detailed mechanistic insight, the applicability to human pain responses will require future verification. Furthermore, we have not elucidated mechanisms for PRDM1-dependent upregulation of several pain-associated genes that were identified in our analysis. The regulation of these genes may contribute to PRDM1 function in response to nociceptive pain hypersensitivity or other neural processes. Nevertheless, the new knowledge gained from this study advances our understanding of transcriptional mechanisms of peripheral nerve injury–induced nociception development and indicates that PRDM1 may represent a novel target for preventing and treating neuropathic pain.

The authors thank Professor Yuqiu Zhang, Ph.D. (Institutes of Brain Science, Fudan University, Shanghai, China), for her guidance with electrophysiologic and Ca²⁺ imaging experiments.

This work was supported by the National Natural Science Foundation of China (grant Nos. 81500947 to C.W.; 31871076 and 31320103906 to T.B.; Beijing, China), Shanghai Municipal Science and Technology Major Project (No. 2018SHZDZX01; Shanghai, China), and ZJLab (China).

The authors declare no competing interests.