Brief exposure to oral antibiotics has age-dependent effects on morphine reward and gene expression in the medial prefrontal cortex of adolescent and adult mice

Abstract

Adolescence is a critical period for the initiation of problematic drug use, which significantly increases the risk of developing substance use disorders later in life. This heightened vulnerability is partly attributed to the immaturity of the prefrontal cortex, a brain region both essential for decision-making and implicated in drug reward. During adolescence, peripheral systems, such as the gut microbiome, also undergo substantial changes. Emerging evidence suggests that disruptions to the gut microbiome can influence gene expression and drug reward behaviors in rodent models. In this study, we investigated the effects of oral antibiotics on morphine reward and prefrontal cortical gene expression in adolescent and adult mice. Using oral antibiotics to transiently disrupt the microbiome, we found that short-term antibiotic exposure reduced morphine place preference specifically in adolescent mice. In a separate cohort, we observed that antibiotic treatment altered the transcriptomic response to morphine in the medial prefrontal cortex across all age groups. Notably, the transcriptomic changes induced by antibiotics and morphine were age-specific, with distinct gene expression patterns observed in adolescents compared to adults. These findings establish a foundation for future research into the role of the gut microbiome in opioid reward and highlight potential gene pathways underlying age-dependent differences in opioid sensitivity.

1. Introduction

Opioid use disorder (OUD) is a major crisis in the United States and around the world. This condition is associated with increased prevalence of chronic health conditions^1,2 and risk of death from overdose^3,4. Understanding the biological and environmental factors that contribute to the development of OUD can help identify policies and treatments to help curb opioid use and treat OUD. One predictive factor in diagnosis of a substance use disorder (SUD) is drug use at a young age; age of drug use initiation inversely correlates with risk of developing an SUD later in life^5,6. For many people, drug use starts during adolescence⁷- a time of life when individuals undergo many physical and social changes. Understanding how adolescents respond to drugs of abuse is necessary to reduce early drug use. Additionally, given the relationship between drug use at young ages and SUD diagnosis later in life, this knowledge could be leveraged to help decrease the total number of SUD diagnoses and overdose deaths.

One contributing factor in the propensity of adolescents to experiment with drugs of abuse is an underdeveloped prefrontal cortex (PFC)⁸. Proper PFC functioning is necessary for healthy working memory and decision-making^9,10, but this region has also been implicated in drug addiction. Dysfunction in this region has been observed in patients with SUD¹¹, and it has been proposed that individuals with pre-existing abnormalities in PFC function are more prone to problematic drug use¹¹. While the rodent frontal cortex is not as complex as humans’, the rodent PFC also contributes to decision-making and drug reward^12–14 and is rapidly developing during adolescence^15–17.

While the PFC of adolescents is undergoing many changes, the physiological effects of adolescence are not restricted to the brain. Puberty occurs during adolescence, as well as changes in body composition and rapid growth during the teenage years. Partially due to changing energy needs and diets, the adolescent gut microbiome undergoes drastic shifts in humans^18,19 and is undergoing changes between the juvenile period and adulthood in rodents^20,21. Increasing evidence suggests that the gut microbiome has far reaching impacts on health and disease, including disorders of the central nervous system. In fact, causal studies in animals have shown that changes to the microbiome have drastic impacts on outcomes in models of neurodegenerative²² and psychiatric disorders^23–28. While most studies are conducted in adulthood, alterations to the microbiome in adolescence also show effects on anxiety-like behaviors²³ that last into adulthood²⁹ and produce alterations to social behavior after traumatic brain injury³⁰. Additionally, one week of antibiotics (Abx) treatment during early postnatal life, during pre-weaning, or immediately post-weaning produced differing effects on behavior and gene expression in the brain during adulthood³¹, demonstrating that the behavioral and molecular effects of microbiome manipulation differ depending on age of microbiome disruption.

In addition to measures of anxiety and depression, prior work has shown that the microbiome can influence drug reward and reinforcement^32,33. Our lab and others have found that manipulation of the microbiome alters cellular and behavioral responses to cocaine^34–37 and opioids^38–41. We have shown that knockdown of the microbiome with Abx decreases morphine conditioned place preference (CPP)³⁹ and increases self-administration of low-dose fentanyl and increases breakpoint during progressive ratio³⁸ in adults.

Additionally, there are several studies investigating how Abx alters the molecular and cellular response to drugs of abuse^36,38–40. In studies employing sequencing or proteomic analyses, more differentially expressed genes (DEG) or proteins were found in animals with a reduced microbiome and receiving cocaine³⁶ or opioids^38,39. Additionally, pathway analysis on sequencing data from the nucleus accumbens (NAc) identified more upregulated genes related to histone and chromatin modification in animals with microbiome knockdown and drug exposure compared to drug exposure alone. These studies focused on the NAc, a brain region strongly implicated in drug reward and reinforcement. However, other studies suggest that the mPFC is also sensitive to changes to the gut microbiome^30,42–45. Notably, microbiome disruption with Abx or germ-free housing altered expression of genes involved in calcium signaling and genes that were associated with glutamatergic synapses within the mPFC⁴⁴. This occurred concomitantly with attenuated fear extinction and reduced dendritic spine formation during fear extinction in this region⁴⁴. Reconstitution of select bacteria during early life reversed fear extinction deficits, implicating the microbiome in this effect⁴⁴. Additional studies show upregulation of myelin-related genes within mPFC in mice raised in germ-free housing, an effect that was reversed by a return to conventional housing during a sensitive period post-weaning⁴².

Given the importance of the mPFC in drug reward^12–14, and the fact that the mPFC is changing more dramatically than the NAc during adolescence¹⁵, the current study measures morphine reward after Abx and explores the molecular consequences of morphine in the mPFC with and without oral Abx. Unlike our prior work that employed 14 days Abx treatment before behavior, a shorter exposure period of Abx is utilized to restrict treatment effects to within the early adolescent period (starts during the juvenile period PND25 and ends PND35) and compares results to those found in young adults (starts PND55 and ends PND65). Data from this study explores how brief Abx exposure alters morphine reward and the mPFC transcriptome in adolescents and adults.

2. Materials and Methods

2.1. Animals

Male and female adolescent (PND 25 at start of drink treatment) and adult (PND 55 at start of drink treatment) C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME, USA) were used for all experiments. Mice were housed 4 or 5/cage with age- and sex-matched animals in a humidity and temperature-controlled vivarium on a 12h light:dark cycle. Mice were left undisturbed in the colony for four to seven days before the start of drink solutions. At age PND 25 or PND 55, cages were randomly assigned to control H₂O or antibiotics (Abx) treatment. All animal procedures were approved by the IACUC at Mount Sinai School of Medicine and all animal procedures conformed to ARRIVE guidelines. Separate groups of mice were used for conditioned place preference and sequencing experiments.

2.2. Drugs

Morphine sulfate pentahydrate was provided by NIDA drug supply program and was dissolved in isotonic saline. Morphine and saline were injected subcutaneously at a volume of 10ml/kg.

2.3. Drink solutions

Mice were placed on control H₂O or a cocktail of antibiotics on PND 25 or PND 55. Antibiotics consisted of 0.5 mg/ml vancomycin (Chem-Impex International), 2 mg/ml neomycin (Fisher Scientific), 0.5 mg/ml bacitracin (Research Products International), and 1.2 microg/ml pimaricin (Infodine Chemical) dissolved in drinking water^36,38,39. Mice remained on drink solutions for five days before the start of behavior or injections and all mice remained on their drink solutions throughout the entire study.

2.4. Conditioned place preference

Adolescent and adult male and female mice (n = 11–14, 5–8/sex) on H₂O or Abx underwent conditioned place preference (CPP) for morphine using Med Associates boxes and software as previously described^35,39. The testing apparatus consisted of three distinct chambers- two larger chambers to either side of a small middle chamber. The end chambers contained unique wall patterns and floor textures to distinguish them from each other (one chamber had gray walls and a small grid floor and the other chamber had black and white striped walls and a larger grid floor). The entire CPP task took place over 5 consecutive days: one day of pre-test, three conditioning days, and one test day. For pre-test day, mice were placed into the small middle chamber and were allowed to explore all three chambers for 30 minutes. Morphine chamber was assigned such that there was no baseline preference within any group. On conditioning days, mice received injections of saline (s.c.) and were confined to one end chamber for 60 mins in the morning (saline chamber). No less than 3 hours later, mice were injected with 5.6 or 10 mg/kg morphine (s.c.) and were confined to the opposite end chamber for 60 mins (morphine-paired chamber). On test day, mice were again allowed to explore all three chambers of the CPP apparatus for 30 mins. Mouse locomotion during saline and morphine conditioning sessions and mouse location during pre-test and test sessions were recorded by laser beam breaks. CPP score was calculated as: [(time in morphine-paired chamber on test day) – (time spent in saline chamber on test day)].

2.5. 16S sequencing

Separate groups of adolescent and adult male and female mice on H₂O or Abx were given once daily injections of saline or 15 mg/kg morphine (s.c.) for 5 days (n = 5 for all groups). Twenty-four hours after their last injection, mice were rapidly decapitated. Cecal content was collected and flash frozen on dry ice. DNA was isolated from cecal contents using DNeasy PowerSoil kit (Qiagen) via kit instructions with the additional bead beating step. DNA concentration was quantified using a NanoDrop1000. PCR amplification was achieved using primers (341F/805R) targeting the variable V3 and V4 region of the 16S rRNA bacterial genome. DNA fragments were sequenced on an Illumina NovaSeq (2× 250 bp). Using Divisive Amplicon Denoising Algorithm 2⁴⁶, data were filtered, dereplicated, and paired ends were merged. Chao1 and Simpson Indices were used to assess alpha diversity. Principle coordinates analysis plots were generated using the unweighted Unifrac distance as an assessment of beta diversity using QIIME2 software⁴⁷. Observed taxonomic units (OTU) were identified by comparing their genetic sequences to reference bacterial genomes using SILVA (Release 132)⁴⁸ with confidence set to 0.7. Male and female data were sequenced and analyzed separately.

2.6. RNA sequencing

2.6.1. Tissue collection and RNA extraction

The same mice used for 16S sequencing were used for RNA sequencing (n = 5 for all groups). Twenty-four hours after their last injection, all mice were euthanized as described above. Whole brain was removed and sectioned using a 1mm partitioned brain mold. The dorsal mPFC, containing prelimbic and anterior cingulate cortices, was identified in the rostral most 4mm of brain dorsal and medial to the corpus callosum. A 12g blunt needle was used to remove a 2mm thick section of mPFC. Tissue punches were flash frozen on dry ice and stored at −80° C. RNA was extracted using RNeasy Mini kit (Qiagen) via kit instructions.

2.6.2. Sample preparation and sequencing

Libraries were prepared using the Fast RNA-seq Library Prep Kit V2 (abclonal Technology). Fragments were read to 150 bp (paired-end) on an Illumina HiSeq2500 to an average depth of 25M reads. Reads were removed if they contained adaptor contamination, when nucleotides were uncertain for 10% or more of the read, or the base quality was less than 5 for 50% of the read. Whole samples were excluded from analysis if the difference between their adjusted number of read counts and the average number of read counts were >80% greater or less than the range of the remaining samples’ read counts. Two adult male H₂O-Sal and two adolescent Abx-Sal samples were excluded from further analysis due to low number of reads and two male samples (one adolescent H₂O-Sal and one adolescent Abx-Mor) and two female samples (two adolescent H₂O-Mor) were excluded for having too many reads. Male and female samples were sequenced and analyzed separately.

2.6.3. RNA-sequencing bioinformatics

Reads were aligned to the Mus musculus 10 reference genome using the aligner STAR v2.7.10a⁴⁹. Aligned reads were normalized using DESeq2⁵⁰ and differential expression analysis was conducted with ExpressAnalyst⁵¹ software. Pairwise comparisons of all genes were conducted between adolescent and adult controls (H₂O-Sal) to determine age-specific gene expression signatures. Additional comparisons were conducted between each treatment group (Abx-Sal, H₂O-Mor, and Abx-Mor) and their age- and sex-matched control (H₂O-Sal) to determine the effects of each treatment within each age and sex. Genes were considered differentially expressed if they met a statistical significance of p < 0.05 after FDR correction. Significantly up- and downregulated gene lists for each pairwise comparison were separately uploaded to G:profiler⁵² and significant pathways were identified using an FDR corrected p < 0.05.

To generate heatmaps of the adolescent signature genes, the log 2 fold change of raw counts from genes significantly differentially regulated between adolescent and adult controls (H₂O-Sal) were rank ordered from largest downregulated to largest upregulated using the website Morpheus (Broad Institute, https://software.broadinstitute.org/morpheus). Next, the log 2 fold change in expression of the adolescent-specific genes were calculated between the adolescent treatment groups (Abx-Sal, H₂O-Mor, and Abx-Mor) and adult controls (H₂O-Sal). Gene expression information from treatment groups were aligned to the adolescent control group to visualize overall alterations in the developmental gene signature across treatments.

For pathway term heatmaps, pathway term titles were searched for select words of interest and grouped by category for each comparison. For the “development” category, pathway terms containing the words “development” or “genesis” were counted. The “vascular” category counted pathways with the prefix “vascul-“, or “angio-“, or the word “vessel”, the “histone modification / transcription” category counted “histone”, “chromatin”, “DNA”, or “transcription”, the “immune” category counted “immune”, “cytokine”, or “microglia”, and the “extracellular matrix” counted “extracellular matrix” or “collagen”. If a pathway term contained more than one keyword, it was only counted once for that category. The number of unique terms in each category were counted and normalized to total term number for each comparison.

Finally, weighted gene co-expression network analysis (WGCNA) was performed as done previously^36,53 using the WGCNA R package⁵⁴. WGCNA was performed with individual female sample weights first determined with the ‘signed hybrid’ network (i.e. negatively correlated genes are assumed not connected). Signed adjacency matrices were generated using a soft-power parameter via a scale-free topological fit to at least a power of 0.8 whilst maximizing mean connectivity. From the determined scale of independence and mean connectivity calculated, we used a soft thresholding power of 7. Subsequently, a topological overlap matrix (TOM) was generated through the adjacency matrix to perform hierarchical clustering. To identify unique modules, a one-step network was constructed via blockwise modules with unsigned topological overlap matrices. Distinct modules were identified and labeled using the Dynamic Tree Cut method. Notably, only networks with a minimal network size of > 40 genes were used for subsequent analysis.

To determine relationships between modules and treatments, Pearson correlation coefficients between module eigengenes or treatment status were calculated. Importantly, heatmaps and pathway analyses for WGCNA figures included all genes assigned to a specific module (i.e. eigengenes with scores less than 0.99 were included). Modules were chosen for further analysis if they were significant in at least one treatment group, with priority given to modules showing significant positive correlations in Abx-Mor groups. Of the 11 modules meeting this criterion in adolescent Abx-Mor females, we chose one module that also had a significantly negative correlation in adolescent H₂O-Mor group (grey) and another module that had the largest number of associated pathways (green). Additionally, of the four modules that had a significant positive correlation in adult Abx-Mor only, we chose to examine the pathway with the largest number of associated pathways (yellow). For these three modules, the top 25 driver genes or “nodes” within these modules were identified using cytoHubba’s Maximal Clique Centrality (MCC) feature within Cytoscape⁵⁵, which measures the relatedness or connectedness of all the nodes in the module. However, all nodes from these modules were included in pathway analysis via G:profiler⁵².

2.7. Statistics

Conditioned place preference score for adolescents and adults were analyzed separately using two-way ANOVAs with drink and morphine dose as between-subjects factors. A priori planned comparisons were conducted between H₂O and Abx mice at each morphine dose using Sidak’s multiple comparisons. In mice receiving 10 mg/kg morphine conditioning, locomotion after saline and morphine injections were analyzed separately in both adolescents and adults using mixed effects analysis with drink as a between-subjects factor and conditioning day as a within-subject factor. Sidak’s multiple comparison test was used to compare H₂O to Abx at each conditioning day. This study was not powered to detect sex differences in behavior, so males and females were analyzed together.

The alpha diversity metrics of Chao1 and Simpson were analyzed using two-way ANOVAs. A priori pairwise comparisons were conducted between the control samples (H₂O-Sal) and each treatment group (Abx-Sal, H₂O-Mor, and Abx-Mor) within each age using Sidak’s multiple comparisons. PERMANOVA tests with 999 permutations were used to analyze beta diversity using the R statistical package vegan (http://github.com/vegandevs/vegan).

3. Results

3.1. Brief Abx reduces morphine place preference in adolescents but not adults.

To determine the effects of brief Abx exposure on morphine reward in adolescents and adults, mice underwent morphine CPP after five days of Abx or H₂O. Pre-adolescent (PND 25) and adult (PND 55) male and female mice were given Abx in their drinking H₂O or remained on control H₂O for five days before CPP (Fig. 1A). Interestingly, there was a significant effect of Abx reducing adolescent (F_(1,45) = 6.277, p = 0.0159), but not adult (F_(1,49) = 0.3898, p = 0.5353) morphine place preference (Fig. 1B–C). Planned comparisons indicated a significant difference between adolescent mice treated with H₂O and Abx at the 10 mg/kg dose (p = 0.0247). There were no significant pairwise comparisons in adult mice, only a main effect of morphine dose (F_(1,49) = 8.898, p = 0.0044) that was not present in adolescents (F_(1,45) = 0.6184, p = 0.4358).

Fig. 1.

Adolescents display reduced morphine place preference after brief Abx. (A) Experimental timeline. (B) Morphine conditioned place preference in adolescent mice. There was a significant effect of Abx (p = 0.0159), and a significant pairwise comparison between H₂O and Abx at the 10 mg/kg morphine dose (p = 0.0247), but no other significant effects. (C) Morphine conditioned place preference in adults. There was a significant effect of morphine dose (p = 0.0044) but no other significant effects in adults. *p < 0.05, males (●), females (▼). (D) Locomotor counts during morphine (●) and saline (■) conditioning sessions in adolescent 10 mg/kg treated mice. (E) Locomotor counts during morphine (●) and saline (■) conditioning sessions in adult 10 mg/kg treated mice. *p < 0.05, **p < 0.01, ***p < 0.001 compared to Abx mice after morphine. ^p < 0.05 compared to Abx mice after Saline.

During conditioning, locomotion in response to saline and morphine injections were recorded in mice undergoing 10 mg/kg morphine conditioning to assess basal activity levels and locomotor sensitization to morphine. In adolescents, Abx treated mice had greater locomotor counts after saline (Fig. 1D, significant effect of drink, F_(1,22) = 7.304, p = 0.013) but Abx did not alter locomotion after saline in adult mice (Fig. 1E, F_(1,25) = 2.910, p = 0.1004). However, adolescent Abx mice demonstrated a dampened morphine sensitization (significant effect of drink, F_(1,22) = 15.71, p = 0.0007 and significant interaction F_(2,44) = 17.37, p < 0.0001). Planned comparisons demonstrate that Abx reduced sensitization on day 2 (p = 0.0066) and day 3 (p = 0.0001). In contrast, Abx did not affect adult locomotor response to morphine over days (effect of drink: F_(1,24) = 3.939, p = 0.0587, but significant interaction F_(2,47) = 9.379, p = 0.0004) but planned comparisons between H₂O and Abx treated adults showed a reduced sensitization on day 3 (p = 0.0028).

3.2. Brief Abx alters the microbiome of adolescents and adults.

Given age-specific differences observed in morphine reward, we next determined if Abx exposure altered the microbiome of adolescent and adult mice receiving morphine and if this differed by age. In a separate group of animals that did not undergo behavioral testing, mice received five days of saline or morphine injections (15 mg/kg, s.c.) after 5 days of Abx or H₂O as described (Fig. 2A). Abx consistently decreased the richness of the microbiome as measured by Chao1 index (Fig. 2B & D; significant effect of treatment males: F_(3,31) = 49.02, p < 0.0001 and females: F_(3,30) = 98.28, p < 0.0001; all pairwise comparisons between H₂O-Sal and Abx treated groups p < 0.01). However, morphine alone affected richness in females (Fig. 2D; adolescents: p = 0.0053 and adults p = 0.0004). In contrast to richness, brief Abx decreased microbiome evenness in some adult Abx treated groups, but no adolescent Abx groups (Fig. 2C & E; Simpson index significant effect of treatment males: F_(3,31) = 12.07, p < 0.0001 and females: F_(3,30) = 227.0, p < 0.0001. Significant pairwise comparisons in adult male Abx-Mor p < 0.0001, adult female Abx-Sal p < 0.0001, and adult female Abx-Mor p = 0.0017).

Fig. 2.

Brief Abx alters the microbiome of adolescents and adults. (A) Experimental timeline. (B & D) Chao1 index of species richness in male (B) and female (D) mice. (C & E) Simpson index of species evenness in male (C) and female (E) mice. (F & G) Principal coordinate analysis plots using unweighted Unifrac distance of microbiome samples from male (F) and female (G) mice. * p < 0.05, ** p < 0.01, p < 0.001, **** p < 0.0001.

Statistical analysis of beta diversity in males indicated significant effects of age (F_(1,31) = 7.852, p < 0.001), drink (F_(1,31) = 14.523, p < 0.001), and drug treatment (F_(1,31) = 1.883, p = 0.026), as well as significant interactions between age and drink (F_(1,31) = 3.366, p = 0.002) and age and drug treatment (F_(1,31) = 1.608, p = 0.010) (Fig. 2F). Similarly, beta diversity analysis of female samples identified significant effects of age (F_(1,31) = 4.363, p < 0.001), drink (F_(1,31) = 6.763, p < 0.001), and drug treatment (F_(1,31) = 1.951, p = 0.017) as well as a significant interaction between age and drink (F_(1,31) = 7.377, p < 0.001) (Fig. 2G). Together, this suggests adolescent and adult microbiomes differed at baseline and that both Abx and morphine altered the microbiome in males and females.

3.3. Gene expression in mPFC differs between adolescent and adult mice.

Since brief Abx reduced morphine CPP only in adolescent mice but changed the microbiome of both adolescents and adults, we next wanted to investigate differences in gene expression within the mPFC. Any age-dependent gene expression changes in mPFC could help explain behavioral differences observed in morphine CPP, given the importance of this region in opioid CPP¹⁴ and its rapid development during adolescence.

We first investigated baseline age differences in gene expression by comparing adolescent H₂O-Sal to adult H₂O-Sal in males and females. There were more differentially regulated genes between adolescent and adult males (Fig. 3B, 437 downregulated, 555 upregulated) than females (Fig. 3F, 23 downregulated, 40 upregulated). Pathway analysis identified several pathways related to synapse pruning in males, such as “neuron projection development” and “synaptic signaling” which were predicted to be downregulated in adolescents (Fig. 3C) and “regulation of neuron apoptotic process” which was upregulated (Fig. 3D). In contrast, several upregulated terms in adolescent females included those related to myelination, such as “ensheathment of neurons”, “gliogenesis”, “oligodendrocyte differentiation”, and “myelination” (Fig. 3H). As expected, age differences in mPFC gene expression seem to reflect known developmental changes that occur during adolescence. To investigate whether morphine or Abx influence the adolescent-specific gene signatures, heatmaps depicting fold-change expression in DEGs from the adolescent gene signature were generated and fold-change of adolescent H₂O-Mor and adolescent Abx-Mor compared to adult controls of corresponding genes were aligned beneath (Fig. 3E & I). In both males and females, the developmental signature seems to weaken over treatments, visualized as fewer genes appearing as dark red or dark blue in Abx-Sal, H₂O-Mor, and Abx-Mor groups (Fig. 3E & I). However, most genes that are up- or downregulated in adolescent H₂O-Sal are regulated in the same direction or do not differ from adult H₂O-Sal at all, suggesting that Abx and morphine do not produce opposite effects on gene regulation in these adolescent signature genes.

Fig. 3.

Age differences in gene expression at baseline are only moderately altered by Abx and morphine. (A) Experimental timeline. (B & F) Volcano plots showing the number of differentially expressed genes (DEG) in male (B) and female (F) adolescent H₂O-Sal mice when compared to sex-matched adult H₂O-Sal mice after FDR correction. Red dots represent significantly upregulated genes and blue dots represent downregulated genes. (C & G) The top ten predicted pathways from male (C) and female (G) adolescent H₂O-Sal downregulated DEGs using an FDR corrected -log(p). (D & H) The top ten predicted pathways from male (D) and female (H) adolescent H₂O-Sal upregulated DEGs using an FDR corrected −log(p). (E & I) Heatmap depicting log(2)fold change gene expression of DEGs identified between male (E) and female (I) adolescent H₂O-Sal and adult sex-matched H₂O-Sal in each adolescent treatment group. Top row depicts expression change between adolescent H₂O-Sal and adult H₂O-Sal ordered from most downregulated on the left (blue) to the most upregulated on the right (red). The bottom rows reflect log(2)fold change gene expression between adolescent Abx-Sal (second row) compared to adult H₂O-Sal, adolescent H₂O-Mor compared to adult H₂O-Sal (third row), and adolescent Abx-Mor compared to adult H₂O-Sal (bottom) aligned with the corresponding DEG in the top row between adolescent and adult controls (H₂O-Sal).

3.4. Abx and morphine alter mPFC gene expression differently in adolescents and adults.

As we have demonstrated previously in the NAc, Abx during drug exposure drastically enhances the number of differentially regulated genes when compared to drug administration alone^36,39. In males, Abx-Mor adolescents had 1,170 and adults had 723 DEGs compared to 103 and 1 DEGs in H₂O-Mor adolescent and adult mice, respectively (Fig. 4A). Unlike our prior work in the NAc, Abx-Sal in adults altered expression of 102 genes- a larger effect than produced by H₂O-Mor. The majority of DEGs in Abx-Mor adolescents were unique to this treatment group, but this was also true for adults (Fig. 4B). Pathway analysis of unique adolescent Abx-Mor DEGs identified upregulated pathways relating to transcription and DNA binding, while downregulated pathways varied but included “response to wounding” and “basement membrane”, as well as pathways related to synapses such as “synaptic receptor adaptor activity” and “structural constituent of postsynaptic density” (Fig. 4C).

Fig. 4.

Treatment effects on gene expression differ by age in male mPFC. (A) Total number of differentially expressed genes in each treatment group when compared to their age-matched control (H₂O-Sal). (B) Upset plot depicting the total number of unique differentially expressed genes in each treatment group or set of treatment groups. Groups are indicated in the bottom portion of the figure by singular black dots or a series of black dots connected by a vertical line. (C) Significant pathways from unique DEGs from adolescent Abx-Mor (compared to adolescent H₂O-Sal). Pathways from downregulated genes are in blue (left) and pathways from upregulated genes are in red (right). (D) Heatmaps depicting the ratio of pathways containing selected keywords to total number of pathways in each comparison. Downregulated terms are in blue (left) and upregulated terms are in red (right).

Next, given the plethora of pathways identified across all pairwise comparisons, we performed an a priori keyword search of pathway names. Given the hypothesis that Abx would alter the mPFC transcriptome of adolescents differently than adults, we first searched for pathway terms related to development. Additionally, based on our prior work^36,39 we searched for terms related to histone modification and transcription. Finally, based on emerging literature, we looked for terms related to vasculature⁵⁶, immune functioning^57,58, and the extracellular matrix^30,59. There were several terms relating to development that were both up- and downregulated in male Abx-Mor mice only, although some development terms were downregulated in adult Abx-Sal (Fig. 4D). Like what we have observed in the NAc after Abx-Mor³⁹, adult Abx-Mor males had a higher ratio of upregulated histone modification and transcription terms than adult H₂O-Mor. Interestingly, this effect was not present in adolescents, where there were a higher ratio of these terms in H₂O-Mor mice (Fig. 4D). Vascular and immune terms were predicted downregulated in adolescent Abx-Mor and extracellular matrix was predicted to be downregulated in adult Abx-Sal and Abx-Mor.

As observed in males, female Abx-Mor had the largest number of DEGs- 1,220 DEGs in Abx-Mor adolescents and 264 in Abx-Mor adults compared to 27 and 110 in H₂O-Mor adolescents and adults (Fig. 5A). Similar to males, the adolescent Abx-Mor group had the greatest number of unique DEGs. Pathway analysis of the unique adolescent Abx-Mor DEGs produced predicted downregulated pathways including those related to cell adhesion, such as “regulation of cell-substrate adhesion” and “cell-cell signaling”. However, upregulated pathways were similar to males and included pathways related to transcription such as “DNA binding” and “transcription coactivator activity” (Fig. 5C).

Fig. 5.

Treatment effects on gene expression differ by age in female mPFC. (A) Total number of differentially expressed genes in each treatment group when compared to their age-matched control (H₂O-Sal). (B) Upset plot depicting the total number of unique differentially expressed genes in each treatment group or set of treatment groups. Groups are indicated in the bottom portion of the figure by singular black dots or a series of black dots connected by a vertical line. (C) Significant pathways from unique DEGs from adolescent Abx-Mor (compared to adolescent H₂O-Sal). Pathways from downregulated genes are in blue (left) and pathways from upregulated genes are in red (right). (D) Heatmaps depicting the ratio of pathways containing selected keywords to total number of pathways in each comparison. Downregulated terms are in blue (left) and upregulated terms are in red (right).

In females, development terms were up- and downregulated in most groups except adolescent H₂O-Mor and all Abx-Sal groups (Fig. 5D). Additionally, all morphine treated groups had several histone modification terms that were upregulated, as well as some that were downregulated in adults (Fig. 5D). Vascular terms were found to be predicted downregulated in adolescent Abx-Mor and adult H₂O-Mor; extracellular matrix was predicted to be downregulated adolescent Abx-Mor only.

Finally, we examined global patterns of gene expression change using a threshold free analysis. Weighted gene coexpression network analysis (WGCNA) uses a data-driven approach to identify groups of genes that change their expression in the same direction across all samples or within select groups. Notably, WGCNA was not performed on male subjects as no detected network met our inclusion criteria previously outlined (i.e. networks with > 40 genes). In females, of 44 total modules, 36 produced significant correlations in at least one treatment group (Fig. 6A). Of these, we prioritized modules with differential directional regulation between adolescent H₂O-Mor and adolescent Abx-Mor (grey module) and the two modules with the most associated pathways that were significantly regulated in adolescent (green) or adult (yellow) Abx-Mor.

Fig. 6.

WGCNA identified unique expression modules in female adolescents and adults. (A) Correlation heatmap of selected modules. The module color is indicated to the left of the correlation plot. Positive correlations are shown in red and negative correlations are in blue. Asterisks indicate significant correlations (* p < 0.05, ** p < 0.01, **** p < 0.0001). Only modules that were significant in at least one treatment group are shown. Adolescent and adult control groups also not shown for clarity. The modules surrounded by black boxes were selected for further analysis (grey, yellow, and green). (B) The top 25 nodes in the grey module (top). The top 8 pathways from all nodes in the grey module (bottom). (C) The top 25 nodes in the green module (top). The top 9 pathways from all nodes in the green module (bottom). (D) The top 25 nodes in the yellow module (top). The top 9 pathways from all nodes in the yellow module (bottom). Node size adjusted separately for each module using MCC score (legend on bottom right).

The grey module, significantly negatively correlated in adolescent H₂O-Mor and positively correlated in adolescent Abx-Mor, contains 3 main clusters of genes. One cluster contains histone variants- reflected in the pathways “RNA polymerase II transcriptional regulatory region sequence-specific DNA binding” and “DNA-binding transcription factor activity, RNA polymerase II-specific”. Another contains genes involved in interferon gamma signaling and innate immune response (Rsad2, Ifit3, Gbp4, Gbp6, and Zpb1). Terms associated with this cluster are “alpha-beta T cell activation”, “defense response to virus”, and “regulation of cytokine production”. The third main cluster contains several different G-protein coupled receptors (Galr2 and Galr3) and G-protein coupled receptor modulators (Gper1 and Npw). The green module, that was significantly positively correlated in adolescent Abx-Mor, contained two clusters of genes- notably one encoding calcium channel subunits (Cacna1a, Cacna1d, Cacna1f, Cacna1g, Cacna1h, Cacna2d2, Cacna2d4) and another encoding collagens and proteins involved in collagen synthesis (P3h1, P3h3, and P4ha2). Overall, pathways indicated that nodes in this module are involved in calcium signaling (“voltage-gated calcium channel activity”) and extracellular matrix construction (“extracellular matrix structure constituent conferring tensile strength”) as well as MAP kinase signaling and DNA repair. Finally, the yellow module (significant in adult Abx-Mor only), contained some histone modifiers such as Ep300, Setdb2, and Kmt2c, and several other DNA binding proteins (Smad4, Arid1b, Arid2, and Nr3c1) indicated by the pathway “transcription regulator activity”. This module also contained varied other significant pathways such as “action potential” and “voltage-gated channel activity”.

4. Discussion

This collection of studies identified an age-specific effect of brief Abx on morphine reward and gene expression after morphine in the mPFC of mice. In contrast to our past study that found reduced place preference in adult male mice³⁹, here we found that adolescents, but not adults, had reduced morphine place preference after Abx. In the current study, we employed a shorter exposure period of Abx (5 days instead of 14 days), and this experimental difference could explain the discrepancy with our earlier work. Our work here suggests that adolescents are more sensitive to the behavioral effects of Abx since adults showed reduced CPP after 14 days pretreatment but adolescents demonstrated that same effect after only 5 days. We also observed a substantially reduced locomotor sensitization in Abx-treated adolescents, in line with the CPP results. This work adds support to previous studies demonstrating that adolescent microbiome disturbance, even if brief, can have marked effects on behavior^23,30,31. We also observed a small but significant baseline hyperlocomotion in Abx treated adolescents. This could reflect another age-specific effect of Abx, such as increased exploration of a novel space or enhanced anxiety-like behavior. Increased anxiety is unlikely given that numerous studies have shown that adult germ-free mice or mice on Abx demonstrate less anxiety-like behavior compared to controls⁶⁰. However, more experiments need to be conducted to determine the cause and relevance of hyperlocomotion to morphine reward in the current paradigm. Additionally, since our behavioral studies were not powered to analyze sex differences, it is possible that the reduction in CPP after adolescent Abx could be sex dependent. All the female adolescent Abx mice had negative place preference scores at 10 mg/kg. However, the current work needs to be repeated with appropriate sample sizes to make definitive conclusions.

Since adolescents are undergoing many physiological changes, identifying the mechanism underlying the age effect of microbiome disturbance on morphine reward is challenging. Given that both the mPFC and the microbiome are undergoing significant changes during adolescence^15–20, we investigated how brief Abx, together with morphine, changes the microbiome and gene expression within the mPFC. Sequencing of the microbiome confirmed underlying age differences in the microbiome and found that brief Abx changed the microbiome in all groups of mice tested. However, the overall effects of Abx on measures of alpha diversity were mild compared to that observed in prior studies^36,38,39. As previously seen, Abx decreased species richness or abundance (measured as Chao1 index) in all groups. Interestingly, species abundance was also decreased in H₂O-Mor females at both ages, an effect we have not observed in adult male animals after opioids^38,39. However, species evenness was only reduced after Abx in adult mice, potentially suggesting that adults are more sensitive to the microbiome-altering effects of Abx. This is contrary to behavior, where adolescents are more sensitive to the reward-reducing effects of Abx. It is important to note that Abx could produce off-target effects not directly investigated in this study. In our prior work we have shown that 2 weeks of Abx exposure does not affect circulating corticosterone levels in adult male mice³⁵. However, other Abx protocols have been shown to increase gut inflammation⁶¹, gut permeability⁶², and release of cytokines⁶³, which could influence the outcomes of this study. Taken together, the effect of brief Abx on morphine CPP is unlikely to be explained solely by age differences in the extent of microbiome disruption produced by Abx- other physiological effects of Abx need to be considered in future experiments.

To further examine the biological locus of our behavioral results, this study measured gene expression changes in the mPFC of adolescents and adults after morphine with and without Abx. First, we identified an adolescent-specific gene expression signature by comparing gene expression between adolescent and adult H₂O-Sal controls. Unsurprisingly, genes involved in synapse pruning and myelination were differentially regulated between adolescents and adults. However, there was a sex effect- males showed downregulation of synapse related terms and females showed upregulation of myelination-related terms. While both increases in myelination and decreases in synapse density occur during adolescence, these sex dependent DEG changes suggest they might occur differently in males and females or might occur at slightly different ages in males and females.

As in our previous dataset³⁹, Abx-Mor mice expressed more DEGs than age- and sex-matched H₂O-Mor groups. Overall, adolescent Abx-Mor expressed the most total DEGs and the most unique DEGs compared to all other groups. Despite differences in total numbers of DEGs, pathway analysis indicated that all Abx-Mor mice had upregulation of pathways affecting gene transcription and histone modification, an effect we have seen previously in the NAc³⁹. Given that so many of the Abx-Mor DEGs between adolescents and adults do not overlap, it is possible that the transcription factors and chromatin modifiers within these pathways are not the same in every Abx-Mor group. Unlike in adult males that only had altered histone modification pathways in Abx-Mor groups, females and adolescent male H₂O-Mor mice also had upregulated histone modification pathways as well.

We also identified several development pathways that were predicted to be up and downregulated in many groups- not just adolescents. In fact, every Abx-Mor group had up- and downregulated terms related to development. This is not true of H₂O-Mor groups where only adults had differences in development terms and adolescents had none. Surprisingly, we found no overall pattern of vascular, extracellular matrix, or immune-related terms across comparisons. These terms appeared infrequently, although when they were present, they were always downregulated and almost exclusively in Abx-treated groups. Despite their sparse appearance in our data, the mechanisms described in these rare terms could be important in understanding how adolescents and adults differ in their response to Abx and morphine.

Finally, threshold free analysis using WGCNA identified several noteworthy modules in female mice. Analysis of the grey module, which was positively correlated in adolescent Abx-Mor females and negatively correlated in adolescent H₂O-Mor, identified histone variants and genes involved in the innate immune response. Currently, any relationship between these disparate groups of genes has not been explored, especially in the context of cortical functioning or responses to opioids. We also analyzed the green and yellow modules, which were upregulated in adolescent and adult Abx-Mor mice, respectively. The green module highlighted calcium signaling and extracellular matrix and the yellow module highlighted some histone modifying genes.

Other noteworthy limitations of the current study are that different groups of mice were used for behavior and sequencing analysis and the doses and injection regimen of morphine differed between mice undergoing CPP and mice used for sequencing. While drawing definitive conclusions in the current study are not possible, the sequencing data might still inform future studies seeking to understand the effect of Abx on adolescent morphine reward. These sequencing results highlight new mechanistic avenues to explore to further understand the adolescent behavioral response to morphine observed in the current study. Adolescent Abx-Mor mice had a greater number of DEGs compared to adult Abx-Mor, regardless of sex. Pathway analysis of unique upregulated genes in adolescent Abx-Mor showed DNA binding or transcription-related terms, suggesting that differences in histone modification and gene transcription could underlie the adolescent-specific effect on morphine reward shown here. In our prior work, we identified similar upregulated pathways in the NAc after morphine CPP in adult male mice³⁹, and in adult male rats after abstinence from cocaine self-administration³⁶. This suggests that alterations to gene transcription occur in response to Abx and drugs of abuse in both NAc and mPFC at both ages, but the exact transcriptional signature likely differs between adolescents and adults, at least in the mPFC.

Given the limitations of the current study and possible off-target effects of Abx, future work will examine the importance of the microbiome in the effect of Abx observed here. We have previously shown that replenishment of short-chain fatty acids (SCFA) can reverse the reduction in adult morphine CPP³⁹. SCFA are metabolites produced by the microbiome⁶⁴ that have a variety of cellular effects, including on histone modification⁶⁵. Future work can replenish the microbiome through fecal microbiota transplant or can replenish SCFAs to support a role of the microbiome or its metabolites in adolescent-specific morphine reward.

While much work is needed to determine the causal relationship between transcriptional changes in cortex to behavior, this study provides many promising avenues to investigate including age-specific histone modifications. Understanding how adolescents respond to drugs of abuse and understanding how environmental stimuli influence these responses is necessary to develop policies to reduce adolescent drug use.