The use of opioid drugs, which includes prescription pain relievers, heroin, and synthetic opioids such as fentanyl, ranges along a continuum from clinical therapeutics to abuse/addiction to fatal overdose. In 2017, 1.7 million Americans suffered from use disorders related to prescription pain relievers and > 650,000 suffered from heroin use disorder (1). In 2019, nearly 50,000 deaths in the USA were attributed to opioid overdose (2, 3). The total economic burden of opioid use disorder and fatal opioid overdose in 2017 was estimated to be $1.02 trillion (4). Therefore, opioid use disorder and overdose constitutes a significant national public health crisis with attendant social and economic impacts. At present, FDA-approved treatments of opioid dependence are limited to buprenorphine, methadone, and naltrexone. However, the steady increase in heroin-related morbidity and mortality over the past two decades in the USA alone and the chronic nature of opioid use disorder points to repeated cycles of cessation of drug use and relapse (5), and the need for more effective and longer lasting treatments.

Opioid pharmacological actions are mediated primarily by a series of opioid receptors expressed by central and peripheral neurons, as well as by neuroendocrine, immune and ectodermal cells (6). The three main opioid receptors in the CNS- mu, delta and kappa- mediate analgesia and euphoria and are sites of action for buprenorphine (partial mu agonist), methadone (mu agonist), and naltrexone (nonselective opioid antagonist). For the most part, these CNS receptors have been the focus of study for understanding the addictive properties of the opioid drugs. However, it is also well known that long-term opioid use disorder is associated with alterations in gastrointestinal (GI) function to include constipation, reduced motility, and increased bacterial translocation by compromising gut barrier function (7–9). These signs of disrupted GI function suggest the possibility that the gut microbiome has been altered (10).

The bulk of the microbiome resides in the GI tract and is composed of bacteria, viruses, archaea, and fungi. It has been estimated that the human GI system contains > 1000 bacterial species and approximately 4 X 10¹³ microorganisms [roughly the same number of human cells (11)], and the gut microbiome expresses ~100 times as many genes as the host human genome (12, 13). Normal functioning of the gut microbiome is essential to the maintenance of human health. A disruption in gut microbiome composition (i.e., dysbiosis) has been linked to numerous diseases, including cancer, diabetes, obesity, immune dysfunction, and inflammatory bowel disease (14, 15). Emerging research is also showing that gut microbiome dysbiosis can play a role in numerous neurologic [e.g., Parkinson disease, Alzheimer disease (16)] and psychiatric conditions [e.g., autism (17), depression, and anxiety (18)], and in eating disorders (19).

Research on gut microbiome involvement in substance use disorders lags well behind most other health disorders, but an increasing number of publications are documenting drug-induced alterations in it and modulation of opioid effects by the gut microbiome. For instance, morphine causes significant alterations in the gut microbiome of animals (9, 20–23) and humans (24–28). The morphine-induced alteration in the gut microbiome causes impairment in reward and sensory responses (29, 30). The gut microbiome has also been shown to mediate morphine analgesic tolerance (31, 32) and withdrawal (33, 34). Exposure to morphine during pregnancy (35) or in newborn rats (36) induces a dysbiosis that persists from the neonatal period into adulthood. Other opioids, including oxycodone (37–39), fentanyl (40, 41), and heroin (22), have also been shown to cause dysbiosis in the gut microbiome. The interactions of opioid drugs with the gut microbiome have been also described in numerous review articles (10, 42–49).

The studies cited above showing gut microbiome alterations in opioid dependence models used noncontingent (e.g., slow-release morphine pellets or minipumps, injection) drug administration. However, it is not known if long-access opioid self-administration has effects on the gut microbiome. Therefore, to simulate contingent opioid use in humans more closely, outbred rats were trained to self-administer heroin using operant schedules of reinforcement. The effect of heroin self-administration on gut microbiome status was determined by 16S rRNA gene sequencing of fecal samples collected before onset of heroin self-administration, at the completion of long-term self-administration, and after post-extinction reinstatement. The results document that long-access heroin self-administration causes a significant disruption of the gut microbiome.

The goal of the present study was to determine if long access heroin self-administration disrupts the gut microbiome in outbred heterogeneous rats. The NMcwiWFsm : HS rats are outbred rodent populations that are known for their high genotypic and phenotypic variability. This strain closely resembles the variation found in human populations (62), and increases the translational relevance of this study. The escalation of heroin intake across the self-administration training sessions, the decrease in active lever responses at extinction, and the recovery of these responses at cue-induced reinstatement, indicate the presence of addiction-like behaviors, rather than non-disordered opioid use. However, these responses were similar in males and females, indicating that both sexes are responsive to the rewarding and reinforcing effects of heroin self-administration. This was consistent with a lack of sex differences in opioid intake at fixed and progressive ratios reported by Ren and Lotfipour in a fentanyl self-administration model (63). While sex effects were not observed in heroin intake, our results show that the gut microbiome was significantly altered after the self-administration and extinction/reinstatement phases of contingent heroin intake by comparison to saline-yoked controls. Heroin did not alter α-diversity of the gut microbiome at any of the experimental phases but it did significantly alter β-diversity in the self-administration and extinction/reinstatement phases. Sex differences were also observed for β-diversity during the self-administration and extinction/reinstatement phases. These differences were not detected prior to heroin use, which is consistent with previous reports that the gut microbiome does not vary by sex in control rats (64, 65). These changes in β-diversity confirm that the structure of the gut microbial community was changed significantly by heroin self-administration. Furthermore, these differences in the gut microbiome structure across the experimental phases varied by sex ( Figures 4B, C ). These results contrast with the study by Ren and Lotfipour, where sex differences were found after opioid self-administration in α-diversity but not in β-diversity. Given that baseline sex variances in α-diversity were reported in Sprague Dawley rats by these authors, it is possible that these differences could be explained by the divergence in drug and rat strains.

To expand the examination of the consequences of drug-induced alterations in β-diversity, the effects of heroin self-administration on the taxonomic makeup of the gut microbiome were examined. Changes at the ASV ( Figure 5 ) and phylum levels ( Figure 7 ) of analysis for the self-administration phase were in close agreement and showed that Bacteroidetes (48%) and Firmicutes (43%) were the taxonomic levels at which the majority (>90% together) of alterations were observed. This pattern was expected in light of the fact that the rat microbiome is dominated by these two phyla (66). The relative abundances of the genus Bacteroides as well as the Lachnospiraceae and Muribaculaceae families were increased to the greatest extent during the self-administration phase. Members of the Bacteroides genus have numerous commensal effects, yet many are pathogenic. B. fragilis is very commonly found in infections and after disruption of the intestinal wall or other perforations (67). This anaerobic species often participates in secondary phases of infection of the gut following the acute phase of infection by E. coli (67). These pathogenic roles of the Bacteroides genus are consistent with the opioid-induced potentiation of infection virulence and increases in systemic bacterial dissemination (68), exacerbation of Gram-positive sepsis (69), as well as the impairment of intestinal epithelial repair in HIV-infected humanized mice (70). Members of the Lachnospiriceae family, such as those of the Bacteroides genus, have beneficial effects including production of short chain fatty acids and maturation of the immune system. However, this family has also been implicated in several disease states, most of which are characterized as inflammatory conditions (e.g., metabolic disease, irritable bowel disease (71)). This is consistent with the known ability of morphine to cause gut barrier disruption and systemic inflammation (9). The alterations in Muribaculaceae in the heroin self-administration phase is paradoxical. A total of 7 different Muribaculaceae ASVs were significantly altered in this phase and 5 of these were increased in relative abundance while 2 were decreased. Muribaculaceae abundance is negatively correlated with inflammation (72) and its abundance has been shown to decline in morphine treated humanized HIV rats and their controls (73). This present result is difficult to link to heroin specifically as increases in the Muribaculaceae would be expected to counteract opioid induced gut inflammation. On the other hand, it is possible that these increases represent a protective response that was of insufficient magnitude to alter heroin self-administration. The differences in the taxonomic makeup of the gut microbiome across the heroin phases by sex were also examined at the ASV ( Figure 6 ) and phylum levels ( Figure 8 ). While most of the bacterial taxa differentially abundant in heroin vs saline-yoked controls at the ASV level were also identified by sex, some ASVs were exclusive to sex. This is the case for Prevotella 9 (ASV18) and Dubosiella (ASV91) for the self-administration phase, as well as Lactobacillus (ASV15, ASV297) and Gastranaerophiliales (ASV77) for the extinction phase ( Figure 6 ). Likewise, comparisons in phyla structure in males and females across phases identified taxa that were specifically changed by sex, but not by heroin treatment. Members of Spirochaetes, Actinobacteria, Verrucomicrobia and Tenericutes were differentially abundant in males compared to females ( Figure 8 ). These results contribute to the sex-dependent differences in β-diversity identified in each experimental phase.

At the end of the extinction/reinstatement phase, fewer taxa were altered in relative abundance (N=13) than in the self-administration phase (N=23), possibly reflecting a rebalancing of the gut microbiome after cessation of heroin intake. The Muribaculaceae family showed the greatest number of changes in the extinction/reinstatement phase, with 4 showing increased abundance and one showing a decrease. The taxa showing the greatest increases in extinction were Ruminiclostridium 6 and Coriobacteriaceae. Ruminoclostridium 6 has been associated with high fat diet (74) and diabetes-induced inflammation (75), and Coriobacteriaceae abundance is increased significantly in irritable bowel disease (76). This finding is consistent with opioid-induced disturbances in gut inflammation (9, 31) and suggests that this effect is quite persistent. Significant decreases in Mucispirillum were observed in the extinction/reinstatement phase as well. Muscispirillum schaedleri has been shown to protect the gut from colitis (77) suggesting that the large reduction in its abundance in the extinction/reinstatement phase could contribute further to gut inflammation. Similarly, the decrease in Ruminiclostridium 5 would be predicted to worsen gut health because this genus is a short chain fatty acid producer (78).

PICRUSt2 analysis was undertaken to predict metabolic alterations that would be expected based on bacterial marker gene sequences. While the low variance of microbial functional profiles can affect the accuracy of PICRUSt2 base metrics, and degrade its performance outside of human datasets (79), there are advantages in using this tool. PICRUSt2 is becoming increasingly popular in microbial ecology studies as it contains an updated and larger database of gene families and reference genomes than other similar tools (58, 80). This predictive tool allows for initial exploration of hypotheses prior to shotgun sequencing or metabolomics analyses. In this study, PICRUSt2 was used in an exploratory manner and confirmatory studies can follow. The results from this analysis were in consensus with the literature on changes in opioid-induced metabolomics measured from metabolites in serum and urine [see review by Dinis-Oliveira (81, 82)]. Pathways predicted to be reduced by the PICRUSt2 analysis have been documented in studies on opioid-induced metabolomics to include vitamin biosynthesis (83, 84), carbohydrate biosynthesis (85, 86), propionic acid production (87), nucleotide biosynthesis (81, 88), guanine degradation (89) and fatty acid and lipid biosynthesis (84, 85). Those pathways predicted to increase also aligned with existing literature on opioid induced metabolomics alterations. These include amino acid and amine biosynthesis (85, 86, 89–91), nucleotide and nucleoside degradation, and increased depletion of energy sources (82, 89). Indicators of oxidative stress related to opioid-induced increases in lipid peroxidation and membrane disruption (89) were also consistent with the PICRUSt2 predictions. Together, pathways predicted by PICRUSt2 to be increased or decreased by heroin and other opioids suggest an overall reduction in the tricarboxylic acid cycle, increased energy expenditure and altered amino acid metabolism, and agree with alterations in opioid metabolomic seen in vivo (81, 89). These results add a new dimension to existing metabolomics research on opioids by suggesting at least some of these changes are mediated at the level of the gut microbiome.

The magnitude of the gut microbiome changes resulting from long-access heroin self-administration seen presently appeared to be smaller than observed in other studies of opioid-induced dysbiosis, particularly as it relates to the relative abundance of taxa. The total dose of heroin self-administered in the present experiment amounted to approximately 10 mg/kg over 15 days, or an average daily dose of 0.75 mg/kg/day. Despite the lower doses achieved with heroin self-administration, this method allowed for the study of volitional use of heroin. Other studies used higher doses (20-80 mg/kg) and administered opioids via slow-release pellets or minipumps, or by systemic injection (20–23). These results agree with the observation by Taylor and colleagues (29) showing that differing opioid regimens influence the gut microbiome in a specific manner. The results of our study also agree with previous studies showing significant alterations in β-diversity (20, 22, 31) without changes in α-diversity (21, 22) after opioid administration.

This study has a number of strengths. First, it is the first to study to assess the effects of heroin self-administration on the gut microbiome. Second, the use of a long-access, contingent heroin self-administration model with heterogeneous stock rats as subjects to better reflect behavioral and genetic variance associated with human opioid abuse. Third, it recapitulates the sex dependent effects of opioids on various pharmacological and behavioral outcomes. Fourth, it establishes similarity of the effects of heroin with those of morphine and other opioids on the gut microbiome and potentially identifies biomarkers for heroin self-administration using MaAsLin2 analysis. Weaknesses of this study include sequencing at the 16S rRNA level which does not always allow identification of altered bacteria at the species level (as opposed to whole genome sequencing), lack of analyses of gut inflammation and function, and lack of a correlation between gut microbiome changes and behavioral outcomes such as heroin intake. Additional studies are underway to address these weaknesses.

The raw data supporting the conclusions of this article was deposited in the SRA database, BioProject # PRJNA1066745.

The animal study was approved by Institutional Care and Use Committee at Medical University of South Carolina. The study was conducted in accordance with the local legislation and institutional requirements.

JG: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. AW: Writing – review & editing. BZ: Data curation, Writing – review & editing. DJK: Writing – review & editing. DF: Writing – review & editing. NC: Writing – review & editing, Methodology. RC: Methodology, Writing – review & editing. LS: Writing – review & editing. JM: Writing – review & editing, Data curation. GH: Writing – review & editing, Investigation. BK: Writing – review & editing, Formal analysis, Investigation, Methodology, Resources. PK: Formal analysis, Investigation, Methodology, Writing – review & editing, Resources. DMK: Formal analysis, Investigation, Methodology, Writing – review & editing, Conceptualization, Funding acquisition, Resources, Writing – original draft. MA-P: Conceptualization, Formal analysis, Methodology, Writing – original draft, Writing – review & editing, Data curation, Investigation, Supervision.