Molecular rhythm alterations in prefrontal cortex and nucleus accumbens associated with opioid use disorder

Abstract

Severe and persistent disruptions to sleep and circadian rhythms are common in people with opioid use disorder (OUD). Preclinical evidence suggests altered molecular rhythms in the brain modulate opioid reward and relapse. However, whether molecular rhythms are disrupted in the brains of people with OUD remained an open question, critical to understanding the role of circadian rhythms in opioid addiction. Using subjects' times of death as a marker of time of day, we investigated transcriptional rhythms in the brains of subjects with OUD compared to unaffected comparison subjects. We discovered rhythmic transcripts in both the dorsolateral prefrontal cortex (DLPFC) and nucleus accumbens (NAc), key brain areas involved in OUD, that were largely distinct between OUD and unaffected subjects. Fewer rhythmic transcripts were identified in DLPFC of subjects with OUD compared to unaffected subjects, whereas in the NAc, nearly double the number of rhythmic transcripts was identified in subjects with OUD. In NAc of subjects with OUD, rhythmic transcripts peaked either in the evening or near sunrise, and were associated with an opioid, dopamine, and GABAergic neurotransmission. Associations with altered neurotransmission in NAc were further supported by co-expression network analysis which identified OUD-specific modules enriched for transcripts involved in dopamine, GABA, and glutamatergic synaptic functions. Additionally, rhythmic transcripts in DLPFC and NAc of subjects with OUD were enriched for genomic loci associated with sleep-related GWAS traits, including sleep duration and insomnia. Collectively, our findings connect transcriptional rhythm changes in opioidergic, dopaminergic, GABAergic signaling in the human brain to sleep-related traits in opioid addiction.

Fig. 1. Rhythmic transcripts are largely distinct in unaffected comparison subjects and subjects with opioid use disorder.

A In the dorsolateral prefrontal cortex (DLPFC), there were 730 rhythmic transcripts detected in unaffected comparison (UC) subjects and 339 in subjects with opioid use disorder (OUD). Notably, only 19 transcripts were rhythmic in both UC subjects and subjects with OUD. Fisher’s exact test indicated lack of overlap in rhythmic transcripts in the DLPFC between UC and OUD subjects (p > 0.35). B Rank–rank hypergeometric overlap was used as a threshold-free approach to confirm the lack of overlap in rhythmicity patterns in the DLPFC of UC subjects and subjects with OUD. C Heatmap of the top 200 circadian transcripts identified in the DLPFC of UC subjects (left), with transcripts peaking across the day. Expression levels are Z-transformed for each transcript, and the transcripts are ordered by their circadian phase value (peak hour). Each column represents a subject and the subjects are ordered by time of death. The top 200 rhythmic transcripts identified in UC subjects are then plotted for subjects with OUD (right), indicating disrupted rhythmicity of normally rhythmic transcripts in subjects with OUD. D The top 200 rhythmic transcripts identified in OUD subjects in the DLPFC (left) are then plotted in UC subjects (right). E Canonical circadian transcripts (NR1D2, ARNTL, CIART) were rhythmic in the DLPFC of UC subjects, but were not rhythmic in OUD subjects. F In the nucleus accumbens (NAc), there were 349 rhythmic transcripts detected in UC subjects and 738 in subjects with OUD. Notably, only 14 transcripts were rhythmic in both UC subjects and subjects with OUD. Fisher’s exact test indicated lack of overlap in rhythmic transcripts in the NAc between UC and OUD subjects (p > 0.65). G Rank–rank hypergeometric overlap was used as a threshold-free approach to confirm the lack of overlap in rhythmicity patterns in the NAc of UC subjects and subjects with OUD. H Heatmap for the top 200 circadian transcripts identified in the NAc of UC subjects (left). The top 200 rhythmic transcripts identified in UC subjects are then plotted for subjects with OUD (right). I The top 200 rhythmic transcripts identified in OUD subjects in the NAc (left) are then plotted in UC subjects (right). J Canonical circadian transcripts (NR1D2, ARNTL, CIART) were rhythmic in the DLPFC of UC subjects, but were not rhythmic in OUD subjects.

Fig. 2. Distinct peak times for rhythmic transcripts in subjects with opioid use disorder (OUD) compared to unaffected comparison (UC) subjects.

A In the dorsolateral prefrontal cortex of UC subjects, transcripts generally peaked at either ZT4 or ZT16, ~12 h apart. Rhythmic transcripts in the DLPFC in subjects with OUD did not peak at consistent times. B In the nucleus accumbens (NAc), rhythmic transcripts in UC subjects generally peaked at ZT10. Rhythmic transcripts in the NAc of subjects with OUD peaked at either ZT11 or ZT23, ~12 h apart. C Transcripts peaking at ZT4 in the DLPFC of UC subjects were enriched for pathways related to rhythms (e.g., circadian rhythm-related genes, sleep) and neurotransmission (e.g., negative regulation of NMDA receptor-mediated neuronal transmission), while transcripts peaking at ZT12 were enriched for immune-related pathways (e.g., adhesion of symbiont to host, negative regulation of innate immune response). Rhythmic transcripts in the DLPFC of OUD subjects were enriched for regulation of neurotrophin TRK receptor signaling pathway and positive regulation of receptor internalization. D Rhythmic transcripts in the NAc of OUD subjects were enriched for apoptotic cleavage of cellular proteins. In the NAc of OUD subjects, rhythmic transcripts peaking at ZT11 were enriched for morphine addiction, glial cell-derived neurotrophic factor receptor signaling, ECM glycoproteins, and synaptic transmission, GABAergic, while transcripts peaking at ZT23 were enriched for opioid signaling, voltage-gated potassium channels, and synapse-related pathways (e.g., regulation of postsynapse organization, chemical synaptic transmission).

Fig. 3. Scatterplots indicating rhythmicity for transcripts that were significantly more or less rhythmic in the dorsolateral prefrontal cortex (DLPFC) and nucleus accumbens (NAc) in subjects with opioid use disorder (OUD) compared to unaffected comparison (UC) subjects.

Each dot indicates a subject with x-axis indicating the time of death (TOD) on ZT scale (−6 to 18 h) and y-axis indicating transcript expression level. The red line is fitted sinusoidal curve. A Scatterplots for the DLPFC indicating rhythmicity of APBA2, FAT3, and AC083798.2 in UC subjects (left), which are significantly less rhythmic in subjects with OUD (right). B Scatterplots for the DLPFC indicating lack of rhythmicity of TMEM119, ADPRH, and HELLS in unaffected comparison subjects (left), and these transcripts are significantly more rhythmic in subjects with OUD (right). C The top pathways represented by transcripts that are less rhythmic in the DLPFC in OUD are related to netrin signaling and eicosanoid signaling, and the top IPA-predicted upstream regulators are PER1 and PER2. D The top pathways represented by transcripts that are more rhythmic in the DLPFC in OUD are related to inositol and dopamine, and the top IPA-predicted upstream regulators are TBX21 and ZC3H12C. E Scatterplots for the NAc indicating rhythmicity of POLR2F, DNM1P46, and HNRNPA1P7 in UC subjects (left), which are significantly less rhythmic in subjects with OUD (right). F Scatterplots for the NAc indicating lack of rhythmicity of ATXN3, STX2, and CARNMT1 in UC subjects (left), and these transcripts are significantly more rhythmic in subjects with OUD (right). G The top pathways represented by transcripts that are less rhythmic in the NAc in OUD are related to RhoA signaling and autophagy, and the top IPA-predicted upstream regulators are monobutyl phthalate and the antipsychotic penfluridol. H The top pathway represented by transcripts that are more rhythmic in the NAc in OUD is Dopamine-DARPP32 Feedback, and the top IPA-predicted upstream regulators is Dopamine.

Fig. 4. OUD associated gene networks in the NAc.

A Weighted gene co-expression network analysis (WGCNA) was used to generate co-expression modules, with the network structure generated on each brain region separately. The identified modules that survived module preservation analysis were arbitrarily assigned names and module differential connectivity (MDC) analysis compared the identified modules in OUD and unaffected comparison subjects. A MDC analysis indicated a gain of connectivity in the NAc for the OUD-1, OUD-2, and OUD-3 modules. Node size indicates the degree of connectivity for that transcript. Blue nodes indicate rhythmic transcripts and yellow halos indicate transcripts that were significantly more rhythmic in OUD. Edges indicate significant co-expression between two particular transcripts. B Pathway enrichment analysis within the NAc OUD-1 module, the NAc OUD-2 module, and the NAc OUD-3 module. Warmer colors indicate increasing −log₁₀ p-value.

Fig. 5. Rhythmic transcripts in the dorsolateral prefrontal cortex (DLPFC) and nucleus accumbens (NAc) enrich for genetic associations with opioid and sleep-related traits.

Genome-wide associated studies (GWAS) have identified loci associated with various sleep-related traits and opioid dependence. We investigated whether rhythmic transcripts, as well as transcripts that were significantly more or less rhythmic in subjects with opioid use disorder (OUD), were enriched for genetic associations with sleep-related traits and opioid dependence. A In the DLPFC of unaffected comparison (UC) subjects, there was significant enrichment of rhythmic transcripts for genes associated with insomnia and long sleep duration. In NAc of OUD subjects, there was enrichment of rhythmic transcripts and transcripts that were more rhythmic in OUD in total sleep duration. There were no significant associations in the NAc of UC subjects. B Insomnia and morningness were associated with transcripts that were significantly less rhythmic in DLPFC of OUD subjects. Transcripts in the DLPFC that were more rhythmic in OUD subjects were enriched for genes associated with short sleep duration, while transcripts that were more rhythmic in the NAc of OUD subjects were enriched for total sleep duration. No significant enrichments were identified for opioid dependence.