Male and female C57Bl/6NTac mice (referred to as B6 throughout; Taconic, Hudson, New York) were bred at the University of Pennsylvania. Upon weaning, mice were group-housed (3–5/cage) with ad libitum access to food and water under a 12-h light/dark cycle (lights on at 0700). All procedures were approved by the University of Pennsylvania Animal Care and Use Committee and were in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Morphine sulfate and remifentanil were obtained from NIDA Drug Supply (Research Triangle Park, NC) and dissolved in 0.9% saline.

Osmotic minipumps (model 1004; Alzet, Cupertino CA) providing either 10 mg/kg/day morphine or saline were implanted subcutaneously in 8–13-week-old female mice which were then allowed to recover for 3 days. Implanted females were co-housed with drug naïve B6 male counterparts for 2 weeks for mating. Cages were examined for pups 3x daily. Litters from implanted females were transferred to a surrogate dam within 4–8 h after birth (mice typically give birth between 1:00 am and 4:00 am and first checks were at 8:00 am). Surrogate dams were actively nurturing naïve pups of the same age (+2–4 days). Drug or saline-exposed pups were co-housed with 2 naïve pups for 8–16 h to facilitate surrogate nurturing, after which naïve pups were removed from cage. Third trimester equivalent exposure began on the evening of postnatal day 1 (PND1) and continued until the morning of PND14. Morphine was delivered at a dose of 10 mg/kg in a volume of 10 mL/kg and administered subcutaneously twice daily at 9:00 AM and 5:00 PM. The last injection of morphine was given on the morning of PND14. Developmental milestones were always measured just prior to the 5:00 PM injection. Controls received saline in an equal injection volume. The treatment of the dam determined the pup injection treatment (morphine or saline) during PND 1–14. Four different cohorts of pregnant dams were used to generate the data for this study. Cohort size, and developmental milestones and withdrawal statistics for each cohort are shown in Supplementary Figure 1. From PND 1–21, pup weight was examined daily and developmental milestones including eye opening, surface righting, forelimb grasping, and extinguishing pivoting behavior were assessed as described by Robinson et al. (2020).

After collection, brains were frozen in isopentane and stored at −80 degrees. Morphine levels were quantitated in fetal and pup mouse brain by LC-MSMS detection of the m/z 286.15 to m/z 165.03, which refers to morphine molecular ion to fragment ion transition, in reconstituted protein precipitation extracts from calibrator, quality control and experimental sample 1:1 brain:PBS homogenates. Briefly, excised brain tissue was homogenized and mixed with 0.4x volume equivalent (eq.) of aqueous ISTD solution (714 ng/mL morphine-D3 deuterated standard; Cerilliant Corp., Round Rock, Tx). Protein was precipitated with 14x volume eq. of 50:50 methanol:acetonitrile containing 0.1% formic acid. Samples were prepared for LC-MSMS analysis via centrifuge, vacuum drying, and 1x volume eq. reconstitution in water. LC-MSMS analysis was conducted on a Sciex Triple Quad™ 7,500 equipped with a Sciex ExionLC™ AD liquid chromatography system and controlled using Sciex OS software (Ver.2.0). Seven microliters of reconstituted extract were injected onto an Inertsil^® ODS-3 (3 × 100 mm; 4 μ) analytical column. The column was maintained at 40°C and was eluted with a gradient of 5% methanol in water containing 0.1% formic acid (Mobile Phase-A) and methanol containing 0.1% formic acid (Mobile Phase-B) at a flow of 0.2 mL/min. The gradient progressed from 0 to 15% MP-B over 2 min then increased to 35% MP-B over 0.5 min. MP-B was maintained at 35% for 3.5 min before being decreased to 0% over 0.5 min. The column was equilibrated at 0% MP-B for 7 min, resulting in a 13.5 min total run time. The eluent was volatilized within an OptiFlow^® Pro ion source (fixed position electrode at 2,200 V, temperature 400°C, N2 heater gas at 50 psi, nebulization N2 gas at 30 psi and N2 curtain gas at 40 psi) prior to MRM monitoring for molecular to fragment ion transitions of morphine (m/z 286.2à165.0; EP = 11/CE = 55/CXP = 24; Avg. Rt = 4.29 min). Calibration curves were generated within the OS software by 1/x^2 weighted, linear regression of nominal analyte to ISTD concentration ratios to their respective analyte/ISTD area ratios (y = 0.00127x + 2.89e-4, R2 = 0.998). Morphine concentration data was analyzed with GraphPad Prism 9 (GraphPad Software) and is displayed as mean. Morphine concentration data was analyzed by one-way ANOVA with post hoc testing conducted using Bonferroni’s multiple-comparisons correction (Table 1).

Ultrasonic vocalizations, somatic withdrawal, and hot plate latency measurements were taken at 8, 24, and 48 h post the final injection of saline or morphine, with different pups being tested at each timepoint. Individual pups were separated from the dam and littermates and placed in a petri dish with fresh bedding. Ultrasonic vocalizations were recorded for 5 min as described in Robinson et al. (2020). DeepSqueak software was used to perform spectrographic analysis of USVs (Coffey et al., 2019). A novel detection network was made using USVs from a previous cohort of over 50 pups, based on the DeepSqueak Mouse Call Network, to detect USVs from each recording. This network did not detect call type. Next, the pup was transferred to a clean 300 ml glass beaker in a different room. Pup activity was filmed for 10 min and later scored for total numbers of wall-climbs (forepaws moving back and forth against wall of beaker) and jumps (all four paws of the ground). The frequency of each behavior was summed to create the withdrawal score. Finally, nociceptive sensitivity was measured on a 55°C hot plate. Latency to show any of the following behaviors within 30 s was measured: licking hind paws, wall climbing, sudden shaking of hind paws, and jumps. The first behavior observed was considered the latency score. Data shown in Figure 1 are from several different timepoints to determine the optimal time to evaluate withdrawal behaviors. Based on this timecourse, all subsequent withdrawal data (Supplementary Figure 1) are from 24 h after the last morphine injection.

For all adult behaviors, mice were group housed (N = 3–5 per cage) and were tested between 12 and 25 weeks of age. Mice were acclimated to testing rooms for at least 30 min prior to the test for all assays except self-administration. For self-administration studies, mice were housed in the same room as Med Associate operant chambers immediately following surgery. Cohort 2 was used for EPM, TST and Learned Helplessness and tested in that order with 2 days between each test. Cohort 3 was split, with half the mice used for CPP and the other half used for self-administration.

Brains were dissected from PND 1 pups 6 h after discovery. Brains were dissected from post-natal exposure only (PND 14) or 3-trimester exposure (3-tri) 6 h after the last morphine or saline injection. This time point was selected to equate morphine levels among all groups at time of death. Total RNA was extracted from the right brain hemisphere using the miRNeasy mini kit (Qiagen, Netherlands) and RNA integrity assessed in an Agilent Tapestation 4200 using the RNA ScreenTape (Agilent, Wilmington, DE). All samples had RIN values > 8. mRNA libraries were generated using the NEBNext Ultra™ II RNA Library Prep Kit for Illumina (New England BioLabs, Ipswitch, MA). The number of animals per group was similar (N = 5–7 animals), and the quality controls, library construction and sequence parameters were also identical across all groups.

Libraries were sequenced on a NovaSeq 6000 at a depth of 30 million total reads/sample using paired-end sequencing of 150 base pairs (PE150), to a depth of 30 million total reads/sample. Reads were processed using the NGS-Data-Charmer pipeline.¹ In brief, the pipeline removed adaptors and low-quality bases from the reads using Trim Galore.² Reads were then mapped to the mouse reference genome (Mus Musculus, GRCm38/mm10) using HISAT2 (version 2.2.1), and duplicated fragments were removed using Picard MarkDuplicates.³ Read counts for the GENCODE gene annotation (version M25) (Frankish et al., 2023) were generated using featureCounts (v2.0.1).

Differential expression analysis between two conditions (e.g., Morphine and Saline) was performed in R (version 4.1.1) with DESeq2 (v1.32.0) package. In brief, DESeq2 provides statistical platform for determining differential gene expression using a model based on the negative binomial distribution. Genes were assigned as differentially expressed if the (adjusted) (nominal) p-value < 0.05. Gene ontology (GO) analyses were performed via ClusterProfiler (v3.8.1). The Canonical Pathways and Upstream regulators were generated through the use of Ingenuity Pathway Analysis (IPA; QIAGEN Inc.⁴). Enrichment for Autism Spectrum Disorder (ASD) gene sets was assessed using Fisher’s exact tests, and P-values were adjusted using the Benjamini-Hochberg method.

For gene expression, snap-frozen samples were homogenized in QIAzol Lysis Reagent (Qiagen). Total RNA was extracted from one brain hemisphere with TriZol reagent (Invitrogen) and purified with the miRNeasy mini kit according to the manufacturer’s instructions (Qiagen), nano drop purity was A260/280 = 1.8–2.2 and A260/230 ≥ 1.8. 1000 ng of each RNA were reverse-transcribed using the High Capacity RNA-to-cDNA Kit (Applied Biosystems, Foster City, CA, USA). Real-time qPCR was performed in a Step-One Plus system (Applied Biosystems) using TaqMan assays following manufacturer protocol. The following Taqman probes were used in this study: GAPDH (Mm99999915_g1), Gad2 (Mm00484623_m1), Mag (Mm00487538_m1), Myrf (Mm01194959_m1). The individual value for each gene, performed in duplicate, was normalized to GAPDH levels. Relative quantification was performed using the ΔΔCt method (Livak and Schmittgen, 2001) and was expressed as fold change relative to control by calculating 2-ΔΔCt.

Behavioral data were analyzed using mixed effects models. Non-normal count data (EPM entries into open arms and failures to escape in the learned helplessness paradigm) were analyzed using generalized linear mixed effects models with a Poisson probability distribution and a log link. All other data were analyzed using linear mixed effects models. Models for developmental milestones, time in open arms of the EPM, time immobile in the TST, PR breakpoint, and nest shredding included fixed effects of drug treatment (morphine or saline). Models for body weight, withdrawal behaviors, learned helplessness, CPP, IVSA over trials, activity during morphine conditioning sessions, social affiliation, and sleep data included fixed effects of treatment and timepoint (day, hour, trial, or session) and their interaction. All models included litter as a random effect and repeated measures designs also included subject as a random effect. Mixed effects modeling was performed within R (R Core Team, 2021) using package “lme4” (Bates et al., 2015), p values were obtained using “afex” (Singmann et al., 2021) and “lmerTest” (Haubo, 2021), and post hoc tests were performed using package “phia” (De Rosario-Martinez, 2015) with the Holm-Bonferroni correction for multiple comparisons. A p < 0.05 was considered statistically significant. Psuedo-R² values were calculated from the mixed effects models (Nakagawa and Schielzeth, 2013). Effect sizes are reported as the marginal R²m value, which takes into account the model fixed effects only.

A separate cohort of mice was used to quantify morphine concentrations in pups throughout gestation and in the early post-natal period. Mice underwent the 3-trimester exposure paradigm as described in Section “Materials and methods” and pup brain tissue was collected at the following timepoints: embryonic day (ED) 10, ED16, ED20, post-natal day (PND) 1, and PND6. For all embryonic timepoints, pups and brain tissue were collected via cesarean section at 9 AM of that day. For ED10 6 embryos were collected, and two brains pooled to produce a total of 3 samples for analysis. For ED16, 12 embryos were collected from two dams and 2 brains were pooled to produce a total of 6 samples. By ED 20, brains were collected from 3 dams and were of sufficient weight to use as individual samples. For post-natal timepoints, brains were collected across two different litters, 6 h after their final morphine injection. Control brains were collected from PND14 mice who did not have any morphine exposure. No significant differences were found between experimental groups, and all experimental groups had significantly higher morphine concentrations compared to controls (p < 0.05). The higher morphine levels observed in ED10 may reflect a decreasing amount of morphine being delivered as the dam gains weight over pregnancy.

Following 3-trimester opioid or saline exposure, (Figure 1A), male but not female morphine-treated pups had significantly lower body weights (Figures 1B, C) which persisted through PND 21 but normalized once mice reached adulthood at 12 weeks (data not shown). In subsequent cohorts both sexes showed significant weight loss (Supplementary Figure 1). Additionally, morphine-treated pups of both sexes took longer to reach several developmental milestones, including surface righting, grasping a rod with forelimbs, and extinguishing of pivoting behavior (Figures 1D, E).

After the last morphine injection on the morning of PND14, there were no sex differences in withdrawal behaviors (data not shown), and therefore mixed-sex groups were used to assess signs of spontaneous withdrawal at 8 h, 24 h, and 48 h. Morphine-treated mice had increased ultrasonic vocalizations (USVs) when separated from the litter, beginning at 8 h after the last injection (Figure 1F), but the full withdrawal syndrome did not emerge until 24 h after the last injection. Morphine-treated mice showed increased somatic signs of withdrawal (Figure 1G) and increased thermal sensitivity, measured by decreased latency to remove a paw or jump in a hotplate assay (Figure 1H). As all withdrawal signs were significantly different at the 24 h time point, subsequent cohorts of mice used for adult behavior studies were tested for withdrawal at this time and showed similar developmental delays and withdrawal behavior (Supplementary Figure 1; characteristics of all cohorts used in this study).

The robust withdrawal behavior seen after opioid exposure in our paradigm mimics key features of human neonatal opioid withdrawal syndrome (NOWS). For behavioral studies, mice who were exposed to 3-trimesters of morphine and experienced subsequent withdrawal will be referred to as “NOWS mice,” and their saline-exposed counterparts as “saline mice.” For transcriptome studies, mice treated only in utero will be described as “PND1” and those treated only post-natally from PND 1–14 as “PND14” and mice treated for all 3-trimseters but not assayed for withdrawal behaviors as “3-Tri.”

To examine forebrain transcriptomic changes that might elucidate mechanisms of withdrawal, delayed development, and any long-term behavior changes, we generated transcriptomic signatures following our “3-trimester” exposure model (3-Tri). In addition, we also examined transcriptomes from animals that received opioids only during the gestational period (PND1) or only during the last trimester from PND 1–14 (PND 14). We sought to determine whether transcriptomic signatures vary based on the window of exposure, perhaps contributing to the discrepancies in the literature regarding acute and long-term outcomes. We have previously reported on the withdrawal and adult behavior profile of PND14 mice (Robinson et al., 2020).

As evidenced in representative Volcano plots (Figure 2A), the majority of the genes exhibited a low amplitude of change in expression level, with most between log2FC (+)0.50 to (-(-)0.50 in all exposure paradigms. Following all three opioid exposure protocols there were a greater number of differentially expressed genes (DEGs) in RNA derived from males than from females (Figures 2B–D), with no DEGs meeting criteria for adjPvalue < 0.05 in the PND14 females (Figure 2D). Therefore, in subsequent analyses nominal P values were sometimes used, and the cutoff is presented for each analysis. Cutoffs for log2FC were not utilized due to the small amplitude of transcript abundance changes. We also noted a greater number of DEGs following exposure restricted to the in utero period (PND1) relative to post-natal only (PND14) or 3-trimester (3-Tri). These data imply that gene expression partially normalized post-natally even in the presence of continued morphine exposure. Although these data do not allow for a definitive mechanism for this phenomenon, it is possible that the presence of early life stressors associated with opioid exposure protocols contributes to the alteration in transcriptional programs. Not only did gene expression largely normalize with continued exposure after birth, but the specific DEGs markedly changed (Figure 2B and Supplementary DEG lists). Using the pAdj cutoff of < 0.05, comparison between males in all three exposure protocols showed only 3 DEGs common to all. These 3 genes, Cobl, Parm1, and Ust, play critical roles in neuronal development, including regulation of the axonal growth cone, cortical actin cytoskeleton, axon dendrite development and embryonic axis specification. There were no overlapping DEGs amongst the female groups. As there were no differentially expressed genes in the female PND14 group using the pAdj cutoff of < 0.05, we compared DEGs between males and females in either PND1 or 3-Tri exposure. There were a number of shared DEGs in both PND1 and in the 3-trimester cohort, the comparison with the highest level of overlap between groups when measured by percentage of total DEGs (Supplementary Figure 2).

To explore the biologic implications of the altered gene expression, we performed Gene Ontology (GO) analysis (Ashburner et al., 2000). First, using the cutoff pAdj < 0.5 as was used for the Venn diagrams, we looked at individual sets of up and downregulated genes, restricted to males in which this analysis was possible in all three exposure paradigms (Figure 3). Notably, there was some overlap between groups for the top 3 up or downregulated, enriched pathways, but in opposite directions, e.g., “response to axon injury” was downregulated in PND1 and upregulated in PND14. To incorporate all three exposure paradigms and both sexes (6 groups) in the GO analysis, we relaxed the stringency and included all DEGs meeting the criteria of a nominal p value < 0.01. GO Biological Processes confirmed what was evident from the analysis of individual DEGs, in that most of the enriched pathways in the females do not reach the significance cutoff and overlap between the groups for the top hits was limited as anticipated (Figure 4, females left and males right and Supplementary Figures 4, 5). In the males, in both the PND1 and 3-trimester cohorts, there was an enrichment for genes involved in processing of proteins, including translation, quality control, localization and the role of the endoplasmic reticulum, albeit not in identical GO Biological Processes. There was essentially no overlap with the PND14 male cohort, in which enriched pathways included myelination and neuron projection morphogenesis and guidance. Other recurring themes and terms in the Cellular Component and Molecular Function enriched GO categories included the extracellular matrix, ribosome, mitochondria, and pre- and post-synaptic membrane (Supplementary Figures 4, 5).

Enrichment of GABA-related processes was restricted to the male PND1 cohort in both the GO Molecular and Cellular categories, i.e., synaptic, while myelin-related processes were enriched in the PND14 male cohort. To further compare and contrast the biologic significance of the DEGs in the three exposure windows, we used Ingenuity Pathway Analysis (IPA), again with relaxed stringency, to identify Canonical Pathways and Upstream Regulators (selected pathways and regulators are shown in Figures 5A, B, and the top pathways and regulators are shown in Supplementary Figures 5A, B). Importantly, the IPA and GO analyses showed significant overlap of pathways. For example, the top Canonical Pathways (Figure 5A and Supplementary Figure 6) identified Opioid Signaling Pathway as significantly enriched in 5/6 groups. RAS GTPase binding and activity, which is required for the effect of opioids on development via Nf1 (Xie et al., 2016), was prominent in the GO Biological pathways enriched in the PND14 males and in overlapping DEGs between the male groups. Furthermore, Axonal Guidance Signaling was enriched throughout all cohorts in the IPA analysis, as was Protein Kinase A Signaling, critical for opioid and dopamine signal transduction (Svenningsson et al., 2005). The PND1 group remains notable for a high level of enrichment for mitochondrial functions, the activity of which is increasingly linked to neurodevelopment (Mahony and O’Ryan, 2021). Finally, we observed that GABA is a significant Upstream Regulator in the PND1 female and PND14 male cohorts, along with the PND1 males.

To complete the comparisons between groups, we used the Rank Rank Hypergeometric Algorithm (RRHO) which compares gene lists without use of a threshold (Plaisier et al., 2010). There was no concordance noted between any of the female groups (not shown), but there was a moderate degree of discordance between PND1 and 3-trimester, and a high degree of concordance between PND14 and 3-trimester (Supplementary Figure 7), consistent with the IPA Canonical Pathway analysis.

There was a preponderance of enriched GO pathways relevant to GABA in males and/or the synapse in both sexes, which have been implicated in neurodevelopmental abnormalities, particularly autism spectrum disorder (ASD) (Di et al., 2020). qPCR confirmed that Gad2 mRNA was increased by 25% in PND1 males but not females (Figure 6A). We therefore analyzed the intersection between our gene lists and those that catalog evidence of association with ASD. Using ASD human cortical transcriptomic gene sets comprising differentially expressed genes and ASD-associated co-expression modules, we detected significant enrichments overall in male DEG lists across all exposure windows via a Fisher’s Exact Test (Figure 6B). We observed enrichment of male and female DEGs with co-expression module M1 (Gupta et al., 2014) in PND 1 and PND 14. This module is enriched for neuronal markers and synaptic genes and, importantly, is downregulated in post-mortem brains from autistic patients. We also found an enrichment of male DEGs with module M6 from the same study, and ASD genetic risk factors from the SFARI database (Abrahams et al., 2013), but only in PND 1 males; this dataset was also enriched for neuronal markers but upregulated in post-mortem brains from autistic patients. We also observed an enrichment of DEGs from males in the 3-Tri exposure paradigm with module M16 from a different study (Voineagu et al., 2011), which contains genes involved in immune and inflammatory responses and is upregulated in autistic brains. Lastly, we detected a significant enrichment of DEGs in both males and females with targets of FMRP (Darnell et al., 2011), the protein that when mutated causes Fragile-X syndrome. These points to a potential disruption of FMRP function caused by morphine exposure. Only two of the ASD databases showed no enrichment; ASD Module 5 (Gupta et al., 2014) and Satterstrom (Satterstrom et al., 2020).

In the PND14 cohort, one of the most highly enriched GO Biological Process pathways in males was Myelination, and in the IPA Canonical Pathway analysis, Axonal Guidance Signaling was very prominent in multiple cohorts. Analysis of IPA Upstream Regulators highlighted Tcf7l2 (Figure 7A) and Sox2, both known to be involved in myelination and/or oligodendroglial differentiation (Fu et al., 2012; Zhang S. et al., 2018). The top DEGs in both sexes following all three opioid exposures contained multiple, but different, myelin-associated genes, most prominently demonstrated by the IPA map of DEGs regulated by Tcf7l2 in PND14 males, almost all of which were downregulated, indicated in green. Mbp, encoding myelin basic protein, narrowly missed the cutoff, with a log2FC of (-)0.40 and a an adjPvalue = 0.06.

We assayed several of the DEGs using qPCR, with the knowledge that qPCR validation of DEGs derived from bulk RNA preparation and with such low log2FC values is notoriously difficult. Myelin regulatory factor (Myrf), a transcription factor required for late stages of oligodendroglial differentiation (Emery et al., 2009), trended toward down-regulation in morphine treated PND14 males. Surprisingly, myelin associated glycoprotein (Mag), a constituent of compact myelin which inhibits neurite outgrowth (Quarles, 2007), was unchanged in morphine treated PND14 males, but significantly downregulated in PND14 females, despite the fact that it was unchanged in the RNA sequencing analysis in females. Consistent with the RNA sequencing data, Myrf and Mag mRNA levels were unchanged in both the male and female 3-trimester cohorts (Figure 7B).

We proceeded to assay behaviors commonly examined in animal models of in utero/perinatal opioid exposure, with the knowledge that analysis of the RNA seq data yielded data consistent with an impact on developmental pathways, including neuronal circuitry, that might alter adult behavior. Prior work had reported that gestational exposure to opioids increases anxiety- and depressive-like behavior in adolescence and adulthood (Klausz et al., 2011; Ahmadalipour et al., 2015), while others have reported the opposite or no effect (Walters et al., 2005; McPherson et al., 2007; Tan et al., 2015). Therefore, we investigated behaviors related to these phenotypes in our NOWS mice using the elevated zero maze, forced swim test, and development of learned helplessness.

In the elevated zero maze, there was no difference in the time spent in the open arms between adult saline and NOWS mice of either sex (Figure 8A) or in the number of entries into the open arms (Figure 8B). In the tail suspension test, there was no difference in the time spent immobile between saline and NOWS mice of either sex (Figure 8C). In a learned helplessness paradigm, there was no difference between saline and NOWS mice of either sex in the number of failures to escape (Figure 8D) or the average escape latency across trials in the test period (Figures 8E, F).

Saline and NOWS adult mice were tested for morphine reward using the conditioned place preference (CPP) procedure. Using a dose of morphine previously shown to elicit CPP (Walters et al., 2005), we found both saline and NOWS mice exhibit a robust preference for the morphine paired side of the chamber regardless of sex (Figure 8G, H). Additionally, there was no difference in the development of locomotor sensitization during CPP conditioning sessions in either male (Figure 8I) or female mice (Figure 8J). To determine if contingent administration of a clinically relevant opioid in adulthood is impacted by 3-trimester opioid exposure, saline and NOWS mice underwent 10 days of remifentanil self-administration (0.1 mg/kg/infusions) in 2-h daily sessions. There was no difference between sex or treatment in the number of nose pokes, or the amount of drug infused, during the training, FR1, or FR2 portions of the self-administration test (Figure 8K). Additionally, there was no difference in the progressive ratio breakpoint between NOWS and saline mice (Figure 8L).

While there is no identified association between in utero opioid exposure and Autism Spectrum Disorders (ASD) in humans, we had identified an overlap with ASD datasets in our transcriptome analysis detailed above (Figure 4). Therefore, we tested whether 3-trimester opioid exposure and subsequent withdrawal causes specific behaviors associated with neurodevelopmental disorders like ASD (Hol et al., 1996; Niesink et al., 1996). In a 3-chamber social affiliation assay, we found that male NOWS mice displayed a significant preference to interact with the non-social cylinder compared to the social cylinder (Figure 9A). In contrast, female NOWS mice showed a significant preference for the cylinder when the stimulus mouse was present (Figure 9B).

To explore other phenotypes associated with ASD, we examined sleep behaviors, which to our knowledge have not previously been assayed in mouse models of in utero/post-natal opioid exposure. Sleep was defined as 40 consecutive seconds or more of inactivity as measured by the COMPASS system, which has been shown to correlate with EEG sleep measures in mice (Brown et al., 2016). Male NOWS mice exhibited significantly different sleep-wake patterns than saline mice over a 36-h period that included two dark periods and one light period (Figure 9C). There was no effect in NOWS males in the length of sleep bouts (Figure 9D). However, in female mice, there was no difference in overall sleep patterns between NOWS and saline mice (Figure 9E), but female NOWS mice had significantly longer sleep bouts during the light period compared to saline mice (Figure 9F).