Neuropathic pain promotes adaptive changes in gene expression in brain networks involved in stress and depression

Abstract

Neuropathic pain is a complex chronic condition characterized by various sensory, cognitive, and affective symptoms. A large percentage of patients with neuropathic pain are also afflicted with depression and anxiety disorders, a pattern that is also seen in animal models. Furthermore, clinical and preclinical studies indicate that chronic pain corresponds with adaptations in several brain networks involved in mood, motivation, and reward. Chronic stress is also a major risk factor for depression. We investigated whether chronic pain and stress affect similar molecular mechanisms and whether chronic pain can affect gene expression patterns that are involved in depression. Using two mouse models of neuropathic pain and depression [spared nerve injury (SNI) and chronic unpredictable stress (CUS)], we performed next-generation RNA sequencing and pathway analysis to monitor changes in gene expression in the nucleus accumbens (NAc), the medial prefrontal cortex (mPFC), and the periaqueductal gray (PAG). In addition to finding unique transcriptome profiles across these regions, we identified a substantial number of signaling pathway-associated genes with similar changes in expression in both SNI and CUS mice. Many of these genes have been implicated in depression, anxiety, and chronic pain in patients. Our study provides a resource of the changes in gene expression induced by long-term neuropathic pain in three distinct brain regions and reveals molecular connections between pain and chronic stress.

Fig 1. The SNI model of neuropathic pain induces mechanical allodynia and enhances anxiety- and depression-like behaviors. (A)

Top: experimental timeline; * denotes days where mechanical nociceptive thresholds were measured. Bottom: Spared nerve injury model (SNI). (B) Mechanical allodynia 50% mechanical thresholds were determined using Von Frey filaments before surgery (baseline) and every 2 weeks thereafter. (n = 16 mice per group, F_(1,120) = 155, p < 0.001; two-way ANOVA and Holm-Sidak adjustment for multiple comparisons). (C to F) Anxiety-like behaviors assessed in an elevated plus maze (C; n = 9 mice per group, t₍₁₆₎ = 3.03, p = 0.008; unpaired t-test), in an open field test (D; n = 9 mice per group, t₍₁₆₎ = 2.24, p = 0.04; unpaired t-test)_, for sucrose consumption (E; n = 9 mice per group, t₍₁₆₎ = 4.62, p < 0.001; unpaired t-test), and in a forced swim test (F; n = 9, t₍₁₆₎ = −3.15, p = 0.004; unpaired t-test). Data in (B to F) are mean ± SEM, * p < 0.05.

Fig 2. The NAc and PAG show robust co-upregulation patterns of genes in correspondence with long term neuropathic pain. (A)

RRHO maps comparing the overlap of SNI-induced differentially expressed genes. Threshold free comparisons of up- and down- regulated genes are shown between NAc and PAG (left), NAc and mPFC (middle), and mPFC and PAG (right). (B) Venn diagram showing number of co-upregulated genes between NAc and PAG (left); enriched gene-ontology terms of co-upregulated genes (right). (C) Protein-protein interaction maps of enriched GO term negative-regulation of biosynthesis. (D) RT qPCR validation of SNI-induced HDAC5 upregulation in the NAc (n=6 mice per group, t₍₁₁₎ = −2.24, p = 0.047; unpaired t-test) and PAG (n = 16 mice per group, t₍₃₀₎= −2.32, p = 0.027; unpaired t-test). Data are mean ± SEM are reported, * p < 0.05.

Fig 3. Genetic elimination of HDAC5 attenuates emotional but not nociceptive effects of long-term neuropathic pain, and enhances antidepressant efficacy. (A)

50% mechanical thresholds were compared between HDAC5 WT and HDAC5 KO mice before sham or SNI surgery (baseline) and up to 60 days thereafter. Measurements were made using Von Frey filaments. No significant difference was observed between groups at any time-point (F_(1,324) =1.11, p = 0.355), and there was no interaction between genotype, typed of surgery, and time (F_{(8, 324)} = 1.36, p = 0.214); n = 10 mice per group. SNI treated HDAC5 WT and HDAC5KO mice were monitored (B) in the sucrose preference test (n = 10 mice per group, t₍₁₈₎ = −2.2, p = 0.041; unpaired t-test). (C) in the EPM test (n = 10 mice per group, t₍₁₈₎ =−2.3, p = 0.033; unpaired t-test). (D) in the open field test (n = 10 mice per group, t₍₁₈₎ = −2.3, p = 0.035; unpaired t-test). (E) and in the Von Frey assay following chronic duloxetine (5 mg/kg every 12 hours; n = 7 mice per group, F_(1,48) = 9.0; P = 0.01; two-way ANOVA and Holm-Sidak adjustment for multiple comparisons). Data are mean ± SEM, *p < 0.05

Fig 4. Long-term neuropathic pain corresponds with robust and distinct transcriptome changes in the mPFC, NAc, and PAG. (A)

RNA sequencing experimental timeline and design comparing mPFC, NAc, and PAG from mice exposed to SNI or sham. 6 biological replicates were used per condition for each brain region, composed of 2 pooled samples/biological replicate. (B) Hierarchical cluster heat maps of the mPFC, NAc, and PAG showing relative expression of genes across sham and SNI samples. (C) Numbers of significantly upregulated or downregulated differentially expressed genes in SNI vs. Sham mice in mPFC, NAc, and PAG brain regions. Genes reported showed at least log2(1.50) fold change compared to sham. (D) Left: VennPlex assessment of gene expression overlaps in the mPFC, NAc, and PAG in SNI vs. sham. Right: Common differentially expressed genes across brain regions, data is expansion of Venn diagram on left. Down arrow = down-regulation; Up arrow = up-regulation.

Fig 5. SNI-induced gene expression patterns indicated common signal transduction pathways are affected in the mPFC, NAc and PAG and show robust overlap with genes previously implicated in anxio-depressive states and pain. (A)

Comparison of IPA derived canonical signaling pathways affected by gene expression patterns in mPFC, NAc, and PAG samples from SNI vs. sham. (B) Overlap of differentially expressed genes from present RNAseq study (mPFC, NAc, and PAG combined) to genes previously implicated in depression, anxiety, and pain, mostly from post-mortem human brain samples or mutant KO mouse behavioral studies.

Fig 6. qPCR validation of genes of interest and behavioral effects of chronic calpain inhibition (A)

Expression of the indicated genes in each brain region. Top, mPFC: Capn11 (t₍₁₆₎ = −2.83, p = 0.012), Capn2 (t₍₁₀₎ = −0.02, p = 0.669), Capn1 (t₍₁₀₎ = 1.44, p = 0.168), Btg2 (t₍₁₀₎ = 2.35, p = 0.041), Cyr61 (t₍₁₀₎ = 2.48, p = 0.033), Tph2 (t₍₁₀₎ = 2.61, p = 0.037), Dusp1 (t ₍₁₄₎ = 4.27, p < 0.001), F2rl1 (t₍₁₀₎ = −1.28, p = 0.23. Middle, NAc: Capn11(t₍₁₀₎ = −3.06, p = 0.012), Capn2 (t₍₁₀₎ = 0.61, p = 0.558), Capn1 (t₍₁₀₎ = −0.15, p = 0.883), Btg2 (t₍₁₀₎ = 4.14, p = 0.002), Ccl3 (t₍₁₀₎ = −2.26, p = 0.048), Cyp2e1 (t₍₁₂₎ = 2.54, p = 0.026), Fos (t ₍₁₀₎ = 4.57, p = 0.001), Oacyl (t₍₁₀₎ = 3.27, p = 008). Bottom, PAG: Capn11 (t₍₁₂₎ = −4.15, p < 0.001), Capn2( t₍₁₀₎ = −0.91, p = 0.383), Capn1 (t₍₁₀₎ = −0.36, p = 0.727), Mt1 (t₍₁₀₎ = −6.14, p < 0.001), Slc17a8 (t₍₁₀₎ = 3.12, p = 0.011), Sgk1 (t₍₁₀₎ = −3.03, p = 0.013), Vgf (t ₍₁₄₎ = − 4.15, p < 0.001), Pde6g (t₍₁₀₎ = −0.92, p= 0.380). Data are mean fold change ± SEM from n = 6–9 mice per group . values determined by unpaired t-tests. (B) Timeline of MDL 28170 or saline treatments and behavior assessment. (C) Mechanical thresholds after two weeks of treatment with MDL28170 (n = 5 mice per group; F_(3,39) = 7.12, P = 0.001; two-way ANOVA and Holm-Sidak adjustment for multiple comparisons). (D and E) Percent open arm exploratory behavior in the EPM test (D) and immobility time in the forced swim test (E) in SNI mice treated with saline or MDL 28170 for two weeks, no significant effect found by unpaired t-tests. * p < 0.05.

Fig 7. SNI and chronic stress cause similar gene expression patterns in the mPFC, NAc, and PAG. (A)

Chronic unpredictable stress (CUS) experimental timeline; CUS was administered for 4 consecutive weeks, after which behavior was assessed or animals were sacrificed and brains extracted for further analysis. (B) In the novelty suppressed feeding test, CUS increased latencies to eat (n = 11, t₍₂₀₎ = −2.7, p = 0.015; unpaired t-tests). (C) CUS decreased sucrose consumption rates (n = 11, t₍₂₀₎ = −5.3, p < 0.001; unpaired t-tests). (D) Genes regulated by CUS and SNI in the mPFC: Btg2 (n = 7, t₍₁₂₎ = 2.8, p = 0.017; unpaired t-tests), Cyr61 (n = 7 mice per group, t₍₁₂₎ = 7.3, p < 0.001; unpaired t-tests), Dusp1 (n = 6, t₍₁₀₎ = 4.4 p < 0.001; unpaired t-tests). (E) Genes regulated by CUS and SNI in the NAc: Btg2 (n = 7, t₍₁₂₎ = 2.8, p = 0.042; unpaired t-tests), Capn11: (n = 9, t₍₁₆₎ = −2.5, p = 0.024; unpaired t-tests), Ccl3 (n = 7, t₍₁₂₎ = −3.65, p = 0.003; unpaired t-tests), Fos (n = 7, t₍₁₂₎ = 3.3 p = 0.006; unpaired t-tests). (F) Genes regulated by CUS and SNI in the mouse PAG: Sgk1 (n = 7 mice per group, t₍₁₂₎ = −2.4, p = 0.034; unpaired t-tests), Slc17a8 (n = 7, t₍₁₂₎ = 3.6, p = 0.004; unpaired t-tests), Capn11 (n = 7, t₍₁₂₎ = − 4.5 p < 0.001; unpaired t-tests). Data are mean fold change ± SEM, *p < 0.05.