Colony-stimulating factor 2 (CSF2) as a gut microbiome dependent immune factor that alters molecular and behavioral responses to cocaine in male mice

Abstract

Cocaine use disorder is a condition that leads to tremendous morbidity and mortality for which there are currently no FDA-approved pharmacotherapies. Previous research has demonstrated an important role for the resident population of bacteria of the large intestine, collectively dubbed the gut microbiome, in modulating brain and behavior in models of cocaine and other substance use disorders. Importantly, previous work has repeatedly shown that depletion of the gut microbiome leads to increased cocaine taking and seeking behaviors in multiple models. While the precise mechanism of these gut-brain signaling pathways in models of cocaine use is not fully clear, and intriguing possibility is through gut microbiome influences on innate immune system function. In this manuscript we identify the cytokine colony stimulating factor 2 (CSF2) as an immune factor that is increased by cocaine in a gut microbiome dependent manner. Peripherally injected CSF2 crosses the blood-brain barrier into the nucleus accumbens, a brain region central to behavioral responses to cocaine. Treatment with peripheral CSF2 reduces acute and sensitized locomotor responses to cocaine as well as reducing cocaine place preference at high doses. On a molecular level, we find that peripheral injections of CSF2 alter the transcriptional response to both acute and repeated cocaine in the nucleus accumbens. Finally, treatment of microbiome depleted mice with CSF2 reverses the behavioral effects of microbiome depletion on the conditioned place preference assay. Taken together, this work identifies an innate immune factor that represents a novel gut-brain signaling cascade in models of cocaine use and lays the foundations for further translational work targeting this pathway.

Fig. 1.

Multiplex analysis to identify cocaine × microbiome responsive cytokines. (A) Experimental schematic. (B) Heatmap of all cytokines with significant difference in any groups compared to H₂O-Sal controls. (C) G-CSF (aka CSF3) demonstrated a main effect of cocaine (P=0.008) but no effect of Abx or interactions. (D) IP-10 (aka CXCL10) exhibited a main effect of Abx, but no effects of cocaine or interactions. (E) While IL-1α showed effects in preliminary pairwise analysis, there were no main effects or interactions on two-way ANOVA. (F) CSF2 exhibited an expression pattern where it was increased by cocaine only in microbiome intact animals. There were main effects of cocaine (P=0.0008), Abx (P=0.01), and a cocaine × Abx interaction (P=0.0015). All data shown as mean ± S.E.M.; * P<0.05; **P<0.01; *** P<0.001 – Holm-Sidak post-hoc test. N=11–15/group.

Fig. 2.

CSF2 effects on locomotor response to cocaine. (A) Across a single session pretreatment with 10mcg/kg CSF2 reduces locomotor activation in response to cocaine. (B) CSF2 pretreatment does not affect locomotor response to acute saline injection. (C) Two-way ANOVA shows that CSF2 treatment reduces cocaine locomotor response, but main effect of cocaine persists. (D) Injection with lipopolysaccharide (LPS), which can induce sickness behavior, reduces acute locomotor activity, but CSF2 treatment does not. (E) PBS control mice show sensitization of locomotor response to cocaine after five days of treatment, but repeated CSF2 injections blunt locomotor sensitization. (F) Repeated injections of CSF2 does not affect locomotor activity over time. Data are presented as mean ± S.E.M.; **P<0.01; *** P<0.001 on Holm-Sidak post hoc test.

Fig. 3.

Effects of CSF2 on cocaine conditioned place preference. (A) Experimental paradigm for CPP. Mice were injected with PBS or CSF2 (10mcg/kg, i. p.) one hour prior to the start of any behavioral testing on each day. (B) Examination of CPP for low and high dose cocaine showed no significant main effect of dose. However there was a main effect of CSF2 treatment and a significant dose × CSF2 interaction. On post-hoc analysis, CSF2 treated mice show significantly reduced preference for 7.5 mg/kg cocaine. Data are presented as mean ± S.E.M.; ** P<0.01 Holm-Sidak post hoc test.

Fig. 4.

Effects of acute cocaine and CSF2 on NAc gene activation. One hour after an acute saline or cocaine injection levels of gene expression in the NAc were measured. (A) Levels of Cfos, as a marker of neuronal activation, were increased by cocaine with further increase by cocaine + CSF2. (B) Levels of the gene for the CSF2 ligand binding receptor, Csf2ra, were increased by acute cocaine and CSF2 treatments. Data are presented as mean ± S.E.M.; * P<0.05, ** P<0.01 Holm-Sidak post hoc test.

Fig. 5.

RNA sequencing analysis of the nucleus accumbens following CSF2 and cocaine treatment. (A/B) Volcano plot and key enriched gene ontology terms for mice treated with CSF2 only. (C/D) Volcano plot of differentially expressed genes and key gene ontology terms for mice treated with cocaine alone. (E/F) Volcano plot highlighting differentially expressed genes and bar plot of key enriched gene ontology terms for mice receiving injections of both CSF2 and cocaine. (G-I) Transcription factor score bar graphs for top transcription factors predicted to be upstream of upregulated differentially expressed genes using Enrichr analysis. Differentially expressed genes identified as FDR corrected P value < 0.1. N.S. Not significant; ** P<0.01; **** P<0.0001.

Fig. 6.

Patterns of overlap of differentially expressed genes between treatment groups. (A) Venn diagram plot of all differentially expressed genes in each group relative to controls. Circle size reflects number of genes in each category. (B) Key enriched gene ontology pathways from genes uniquely downregulated by CSF2-Cocaine treatment. (C) Venn diagram showing overlap of upregulated genes from each group. (D) Representative gene ontology enrichment from genes uniquely upregulated by CSF2-Cocaine.

Fig. 7.

Antibiotics, but not CSF2 or cocaine, alters makeup of the gut microbiome. Measures of alpha diversity after the indicated treatments showed marked effects of antibiotic treatment on both observed OTUs (A) and Shannon diversity index (B). There were no significant effects of or interactions with cocaine or CSF2 treatment on either measure. (C) Donut plots showing fractional composition of bacterial phyla for the indicated treatment groups. Notably, these plots indicate percent of present bacteria in each sample. The total density of bacteria is much lower in antibiotic treated animals.

Fig. 8.

CSF2 treatment reverses effects of microbiome depletion on cocaine CPP. (A) Experimental schematic for antibiotic treatment and cocaine place preference. (B) Conditioned place preference results for all groups at cocaine dose of 3.75 mg/kg. (C) CPP results for 7.5 mg/kg cocaine. * P<0.05; ** P<0.01; *** P<0.001; ****P<0.0001 – Holm-Sidak post hoc test.