Paternal highly-motivated cocaine-seeking experience enhanced cocaine self-administration in offspring
We first established the cocaine SA procedure in outbred SD rats to generate F0 generations. Rats were randomly assigned to three groups, saline self-administration group (SSA), cocaine self-administration group (CSA), and cocaine yoked-injection group (CY). SSA and CSA groups were subjected to voluntary lever-pressing for saline or cocaine for 30 days under a fixed ratio program (5-day FR1, followed by 25-day FR5), while CY was kept in identical but lever-omitted chambers, programmed to passively receive the same dose of the cocaine while the paired CSA animal received each cocaine injection (Fig. 2A). A progressive ratio test (PR) succeeded in the FR sessions and was used to evaluate motivation for cocaine seeking in CSA (Fig. 2B). To discern rats with high seeking motivation for cocaine, we scored the 47 CSA rats using the PR lever pressed, and a clear drop in score appears on the top eighth of the 47 individuals. In this way, we designated the top 7 rats (14.9%) as “highly motivated” rats (Fig. 2C), which approaches the rate (< 20%) of drug users become addicted 1,20. Seven “highly motivated” rats, together with their yoked animals, and seven SSA controls, were randomly chosen and mated with naïve female rats to generate F1 (Fig. 2D). SSA-F1, CY-F1, and CSA-F1 rats (7 litters respectively) were obtained, and 3–4 adult male rats from each litter were subjected to the cocaine self-administration tests (Fig. 2D). Compared with SSA-F1 and CY-F1, the CSA-F1 rats exhibited higher lever presses for cocaine injections during the FR program and higher break points in the PR program (Fig. 2E, FR5: PCSA−F1 vs CY−F1 < 0.001, PCSA−F1 vs SSA−F1 =0.008; PR: PCSA−F1 vs CY−F1 = 0.006, PCSA−F1 vs SSA−F1 =0.011). At the same time, we counted rats of different litter origins separately to ensure that the enhanced cocaine-seeking behavior presented by CSA-F1 stemmed from inter-group differences induced by the paternal cocaine acquisition paradigm rather than individual variance (Figure S1A, B). We also conducted dimension reduction analysis with behavioral characteristics (active, inactive lever in drug, no-drug session in FR and PR) during the self-administration. Interestingly, F1 rat's behavioral performance diverged in the dimension-UMAP1 according to parental drug intake, but diverged in the dimension-UMAP2 according to parental drug-seeking experience (Fig. 2F, Figure S1C, D). Moreover, correlation analysis showed that offspring's cocaine-seeking motivation was not correlated paternal cocaine intake (Fig. 2G), which restated our previous findings that paternal highly-motivated drug-seeking experience, but not drug exposure per se, leads to a higher level of drug-seeking in the F1 generation.
Paternal highly-motivated cocaine seeking induced differentiated transcriptional responses in cortical and mesolimbic regions of F1 generation
To comprehensively analyze the potential mechanism underlying the increased drug-seeking in CSA-F1, RNA-sequencing was performed in CSA-F1, together with CY-F1 and SSA-F1 as control, in seven brain regions including the orbitofrontal cortex (OFC), medial prefrontal cortex (mPFC), nucleus accumbens (NAc), dorsal striatum (dStr), dorsal hippocampus (dHip), amygdala (BLA), and ventral tegmental area (VTA). We included drug-naïve (Naïve) or cocaine self-administration (Coc. SA) conditions, in hope of modeling differences in the innate and cocaine abuse-induced neuronal plasticity changes over time (Fig. 3A).
As indicated by the behavioral assessment of cocaine SA tests in F1, the vulnerability to drug addiction in descendants depends on the paternal cocaine acquisition paradigms. Therefore, we focus on the genes regulated by paternal passive cocaine exposure (CY-F1) and highly-motivated cocaine-seeking (CSA-F1) in a drug-free state (naïve) and after cocaine self-administration (Coc. SA) (Fig. 3B, line 3) and the genes affected by cocaine administration in all F1 groups (Figure S3A-C). First, we used DEG counts to grossly evaluate the extent of changes in all groupwise comparisons. Four brain regions, including dHip, OFC, dStr, and VTA, exhibit significant between-group variance (DEG count > 500) under naïve or Coc.SA-experienced states (Fig. 3B). Among these regions, OFC exhibited cocaine self-administration-magnified between-group variance (Fig. 3B), and 76.3% of DEGs showed a greater magnitude of change in CY-F1 (Fig. 3C) in response to cocaine self-administration (Figure S3B-D). While in dHip, the significant between-group variance under drug-free state diminished after Coc. SA (Fig. 3B, 3D). Interestingly, the transcriptional profiles of OFC and dHip in naïve and Coc. SA states were not strong correlated (Fig. 3H, I, Figure S4A, B, E, F, I). Besides, a significantly larger number of DEGs were observed in dStr and VTA in response to Coc.SA experience (Fig. 3B). However, the changes in dStr and VTA between CSA-F1 and CY-F1 were positively correlated between naïve and Coc.SA states (Fig. 3J, K, Figure S4I). In addition, the percentage of classified changes in VTA was relatively stable under both naïve and Coc. SA states (Fig. 3F). Therefore, in summary, differences across groups in cortical structures, OFC and dHip, were not maintained after cocaine self-administration (Fig. 3H, I, R OFC = 0.19; R dHip = 0.05) whereas subcortical mesolimbic structures, dStr and VTA, exhibited stable features (Fig. 3J, K, R OFC = 0.47; R dHip = 0.52).
Pattern analysis disentangles the transcriptomic effects of paternal cocaine exposure, motivation, and passive infusion on offspring
In the context of our study, we aimed to investigate the factors that distinguish offspring of paternal highly-motivated cocaine-seeking (CSA-F1) from those of passive cocaine infusion (CY-F1). A notable distinction between these groups is that while both CSA-F0 and CY-F0 were exposed to cocaine, only CSA-F0 had voluntary access to cocaine (Fig. 4A). Although both groups received the same dose of cocaine, it is important to consider that cocaine, being a psychoactive drug 21,22, may induce different psychoactive effects in the two groups due to the absence of operant behavior in CY rats 23. The act of drug-taking involves the immediate ingestion or use of the drug to experience its effects, and both CSA and CY rats undergo this process. However, the crucial difference lies in the motivation behind drug-seeking. CSA rats are driven by a strong craving to obtain the drug, while CY rats are passive and unable to actively seek or acquire drugs. Additionally, the passive acquisition of drugs by CY rats may generate potential stress, that may obstruct drug-seeking behavior in offspring. Therefore, three key factors contribute to the paternal influence on offspring: 1) the direct effects of the drug itself, 2) the drug craving induced by voluntary drug-seeking, and 3) the potential stress induced by passive acquisition (Fig. 4A).
To disentangle the effects of the three factors, we unsupervised classified genes into 12 expression patterns (Figure S5). Following a screening for biological relevance, we assigned genes exhibiting Pattern A expression profiles as representative of "drug exposure" factors (cocaine exposure-induced changes, i.e., consistent significant changes in CSA-F1 vs. SSA-F1 and CY-F1 vs. SSA-F1). Pattern B genes were attributed to the “highly motivated” category (Significant in CSA-F1 vs. SSA-F1 & CY-F1, but not changed in CY-F1 vs. SSA-F1). Similarly, Pattern C genes were associated with the "passive infusion" impact (Significant in CY-F1 vs. SSA-F1 & CSA-F1, but not changed in CSA-F1 vs. SSA-F1) (Fig. 4A, S5). The DEG counts and GO enrichment analysis within each pattern reveal that distinct paternal factors lead to specific transcriptional changes in the F1 generation (Figure S5, S6). Notably, genes related to posttranscriptional gene silencing and protein transport were uniquely influenced by Pattern A, the paternal "pharmacological" factor, in OFC, with no significant enrichment observed in Pattern B and C clusters. Conversely, dStr, dHip, and VTA were primarily affected by Pattern B, relating to the "highly motivated" factor (Figure S5, S6). Overall, the paternal psychoactive effects differ from the pharmacological effects, which exert wide-ranging influence on multiple brain regions, particularly mesolimbic system. These transcriptomic changes are likely to have significant implications for nervous system development and neuronal function (Figure S6).
In order to capture the full breadth of information in the transcriptome data and avoid overlooking key details through simple gene classification, we employed gene-weighted gene co-expression network analysis (WGCNA) 24 to summarize gene expression modules within each brain region. The impact of paternal factors on each gene module was assessed based on the enrichment degree of pattern genes (Fig. 4B-E). In order to thoroughly investigate the impact of paternal psychoactive influences on the neurological function of offspring, we specifically directed our attention to the genes annotated to the "Synapse" ontology term within the enriched modules associated with pattern B and pattern C (Fig. 4F-I). Consequently, the genes within the selected modules exhibited significant enrichment in signaling pathways associated with cell junctions, glutamate receptors, and signal release (Fig. 4J-M). Notably, these changes were observed to be predominantly up-regulated in dStr (Fig. 4H, L), while exhibiting a down-regulated pattern in the dorsal hippocampus dHip (Fig. 4G, K) and VTA (Fig. 4I, M), and no significant changes in OFC (Fig. 4F, J). Specifically, the synaptic differences in the dHip were exclusively observed in the drug-naïve state (Fig. 4G, Figure S7E). Conversely, the differences in synapses within the dorsal striatum and OFC exhibited more pronounced disparities between groups after cocaine self-administration (Fig. 4F, H, Figure S7D, F). In contrast, the group differences in the VTA remained consistent across conditions (Fig. 4I, Figure S7G). The synaptic differences observed in VTA were found to be independent of cocaine exposure in the F1 generation, indicating that the paternal psychoactive effects on the VTA may exhibit stability and have long-lasting impacts. Considering the close neural connections between the striatum and VTA, it is plausible that the transcriptomic differences observed in the dorsal striatum could stem from the stabilized functional changes in VTA (Fig. 4M).
Paternal highly-motivated cocaine-seeking reduced basal dopamine level and increased neural activation induced by cocaine in F1 rats
Dopamine-releasing neurons located in the ventral tegmental area (VTA) play crucial roles in reward-related and goal-directed behaviors 25–27. Given the central role of the VTA in these processes, we hypothesized that transcriptomic changes in the VTA may have wide-ranging functional implications. To explore this hypothesis, we conducted preliminary investigations focusing on two aspects: transmitter level and co-activation. (Fig. 5A-F, Figure S8A, B). HPLC-based monoamine neurotransmitter quantification of naïve F1-generation rats (n = 5 per group) revealed reduced dopamine and dopamine metabolites content in the OFC, NAc, and dStr of CSA-F1 (Fig. 5A, B, Figure S8A, B). mPFC DOPA level exhibited a significant positive correlation with OFC and VTA in CSA-F1 (Fig. 5C).
To further assess cocaine-induced neuronal activation, 10 mg/kg of cocaine was intraperitoneally injected into each naïve F1 rats and then sampled after 1hr. C-Fos+ cell density within each region was calculated (Fig. 5D, E). Differential c-Fos+ cell density was observed in NAc, dStr, and VTA (Wilcoxon rank-sum test, NAc, PCSA vs CY = 0.026; dStr, PCSA vs CY = 0.038; VTA, PCSA vs CY = 0.026). The results showed that highly-motivated paternal cocaine-seeking caused enhanced activation in the offspring upon cocaine exposure (Fig. 5E). Paternal cocaine exposure has been found to potentially induce a heightened co-activation pattern in the reward circuitry of offspring (Figure 5F, left). Specifically, significant differences were observed in the co-activation between the OFC and dStr in CSA-F1 when exposed to cocaine ༈Figure 5F, right༉, suggesting a potential influence of paternal high drug-seeking motivation. However, no significant difference in co-activation was observed between CSA-F1 and CY-F1 groups in this analysis. (Fig. 5F). These findings highlight that paternal cocaine-seeking or -taking impacts the offspring not only at the transcriptomic level but also induces neural activity changes across multiple brain regions. However, the precise relationship between these neural activity changes and the corresponding transcriptome alterations requires further exploration.
Modular transcription factor regulation underlies the coordinated transcriptional changes across brain regions of F1 generation
In pursuit of understanding the origins of transcriptomic changes in various reward-associated brain regions, we embarked on investigating whether paternal “highly motivated” factors elicit synergistic effects across these regions. To establish the presence of coordinated transcriptional signatures, we ranked genes in each region based on their P-values (CSA-F1 vs SSA-F1 and CY-F1 vs SSA-F1) and then utilized rank-rank hypergeometric overlap (RRHO2) 28,29 to assess the similarity of gene ranking between different regions (Figure S9). Our findings indicated that in the naïve state of CSA-F1, there was a more coordinated up- and downregulation of transcriptional signatures between the OFC, mPFC and dStr, BLA, dHip, VTA, as revealed by RRHO analysis (maximum Fisher’s exact test (FET) P < 1.0 × 10− 50) (Figure S9 up-left). To gain deeper insights into the molecular mechanisms underlying the transgenerational inherent vulnerability to cocaine reinforcement across multiple brain regions, we implemented multi-brain-region co-expression network analyses in SSA-F1, CSA-F1, and CY-F1 rats. This approach allowed us to integrate data from all seven brain regions studied, utilizing an established pattern gene set (union of A, B, C pattern genes in Figure S5). We next used module differential connectivity (MDC) analysis (detail provided in Supplementary Materials) to quantify changes in network connectivity between groups (Fig. 6A, Figure S10A, B). We screened the modules that were potentially differentially regulated in CSA-F1 based on MDC (Fig. 6A). The enrichment of the GO term reiterates the hypothesis that neuroplastic changes of reward circuitry affected by paternal highly-motivated cocaine-seeking induce vulnerability to cocaine reinforcement in descendants. (Fig. 6B).
To find the key nodes in the transcriptional regulation process, we performed transcription factor prediction analysis using ChEA3 30. Based on the predicted transcription factors candidate genes could be clustered into three (Fig. 6C). The Cluster I is primarily regulated by NEUROD6, THRA, and THRB, the Cluster II by MYT1L, ZNF25, and PEG3, and Cluster III by SIX3, POU3F4, and SP9. Besides, Cluster II and III share some transcriptional factors, such as DACH2 and CSRNP3 (Fig. 6C). However, genes in Clusters I and III are differently expressed in VTA and dStr, while Cluster II does not show a specific distribution pattern (Figure S10D). Additionally, the main signaling pathways involved in each cluster and downstream of transcription factor were identified (Fig. 6D). Results showed that transcription factors have a preference for certain clusters of genes. For example, genes in Cluster I cluster are involved in neuron migration and dendrite development regulated by NEUROD6, genes in Cluster II are involved in exocytosis and response to cocaine regulated by MYT1L, and genes in Cluster III are involved in neurotransmitter signaling and regulation of membrane potential regulated by POU3F4 (Fig. 6D). Next, the top 15 genes with the most molecular interactions in the co-expression network were chosen as hub genes to generate a co-expression network (Fig. 6E). The final network revealed that genes in Cluster II and III act as the hub nodes. Notably, MYT11L was identified as a key transcriptional regulator and hub of the gene network, leading to the hypothesis that it plays a crucial role in the transmission of vulnerability to cocaine across generations (Fig. 6E).
In conclusion, Transcriptional hubs that underly shared transcriptional changes across multiple reward-associated brain regions were identified in CSA-F1 rats, and these hubs may lead to the specific transcriptomic profile of reward circuitry in CSA-F1 rats. These findings plan a general direction for the subsequent exploration for the transgenerational effects of paternal highly-motivated drug-seeking.