{"paper_id":"352f5119-e3f2-4776-872f-90fe8cd8f0e4","body_text":"Title: Anxiety and risk-taking behavior maps onto opioid and alcohol polysubstance consumption \npatterns in male and female mice \nRunning title: Polysubstance use, sex, and behavior \n \nMakenzie Patarinoa,b,c, Ziheng Christina Wangc, Katrina Wonga,c, Suhjung Janet Lee, BSc,  \nEmma Skillena,c, Richa Naga,c, Britahny Baskin, MAa,b,c, Abigail G. Schindler, PhDa,b,c,d,e* \n  \naDepartment of Psychiatry and Behavioral Sciences, University of Washington, Seattle, WA, \nUSA 98195. \nbGraduate Program in Neuroscience, University of Washington, Seattle, WA, USA 98195. \ncVA Northwest Geriatric Research Education and Clinical Center, VA Puget Sound Health Care \nSystem, Seattle, WA 98108, USA. \ndVA Northwest Mental Illness Research, Education, and Clinical Center, VA Puget Sound Health \nCare System, Seattle, WA 98108, USA.  \neDepartment of Medicine, University of Washington, Seattle, WA 98195, USA. \n  \n*Corresponding Author \nAbigail G. Schindler, PhD \nVA Puget Sound Health Care System, S182 \n1660 South Columbian Way, Seattle, WA 98108 \nPhone: 206-716-5644 Fax: 206-764-5437    Email: aschind@u.washington.edu \n  \nKeywords: Polysubstance, Substance Use Disorder, Alcohol, Ethanol, Opioids, Fentanyl, \nAnxiety, Novelty, Locomotion, Sex differences, Individual differences  \n \nThis article was submitted as a preprint to Bioarchive () \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nAbstract \nPolysubstance use is prevalent in the population but remains understudied in preclinical models. \nAlcohol and opioid polysubstance use is associated with negative outcomes, worse treatment \nprognosis, and higher overdose risk; but underlying mechanisms are still being uncovered. \nExamining factors that motivate use of one substance over another in different contexts in \npreclinical models will better our understanding of polysubstance use and improve translational \nvalue. Here we assessed baseline anxiety-like and locomotive behavior and then measured \nvoluntary consumption of multiple doses of alcohol and fentanyl in group housed male and \nfemale mice using our novel Socially Integrated Polysubstance (SIP) system. Fifty-six male \n(n=32) and female (n=24) adult mice were housed in groups of 4 for one week with continuous \naccess to food, water, two doses of ethanol (5% and 10%) and two doses of fentanyl (5 ug/ml \nand 20 ug/ml). Our analyses revealed sex differences across multiple domains – female mice \nconsumed more liquid in the dark cycle, had higher activity, a higher preference for both ethanol \nand fentanyl over water, and their fentanyl preference increased over the seven days. We then \nused machine-learning techniques to reveal underlying relationships between baseline \nbehavioral phenotypes and subsequent polysubstance consumption patterns, where anxiety- \nand risk-taking-like behavioral phenotypes mapped onto discrete patterns of polysubstance use, \npreference, and escalation. By simulating more translationally relevant substance use and \nimproving our understanding of the motivations for different patterns of consumption, this study \ncontributes to the developing preclinical literature on polysubstance use with the goal of \nfacilitating better treatment outcomes and novel therapeutic strategies. \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nIntroduction \nPolysubstance use, or the longitudinal, sequential, or simultaneous use of multiple substances, \nis a persistent and growing concern globally. Clinical populations that engage in polysubstance \nuse experience detrimental outcomes including worsened substance use disorder (SUD) \nseverity, mental and physical health status, treatment response, and mortality, as well as \nincreased risk for overdose, suicide, and infectious and sexually transmitted disease1,2. In one \nlongitudinal study, persistent polysubstance use was associated with the poorest biological \naging and midlife health and financial/social preparedness\n3. Nearly all individuals with a SUD \nadditionally consume other substances and the majority have at least one other diagnosed \nSUD\n4. \n \nThe increasing prevalence of alcohol and opioid co-use is a pressing concern – from 2002 to \n2012, there was a 15-fold increase in the number of individuals with AUD and comorbid OUD\n5.  \nAlcohol and opioid co-use accelerates the progression of problematic use and is more harmful \nthan either substance used alone\n6 (the number of deaths resulting from opioid overdose also \ninvolving alcohol increased 5.5 times between 1999 and 20177). Thus, there is an urgent need \nto address alcohol and opioid polysubstance use to limit harms and improve outcomes. \n \nThe choice to use one or more substances may depend on life history, current environment, and \npersonality type. Experiencing stressful life events is predictive of polysubstance use8–13. \nAdditionally, maladaptive coping (including aggressive, reactive, or substance-driven coping) is \nthought to play a role in mediating polysubstance use\n14–16. Behavioral or psychological \nphenotypes of an individual may also influence which substance or substances to use, and if \nthat choice remains constant in all cases or is circumstance dependent. For example, both \npreclinical and clinical studies have shown that higher levels of anxiety and novelty-seeking are \ncorrelated with increased alcohol consumption, albeit with different patterns of use17–19. There \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nmay be a relationship between anxiety-like or reward behaviors and increased opioid \nconsumption, but findings remain mixed20,21. However, these types of investigations have been \nlimited to a single substance, so conclusions about pre-existing behaviors or personality traits \nand polysubstance use remain limited. Sex differences in substance use are also thought to be \na critical factor, but little focus has been placed on understanding polysubstance use in relation \nto biological sex.  \n \nWhile work to further characterize polysubstance use patterns in clinical populations is ongoing, \npreclinical models present a viable line of research to investigate underlying motivations and \nmechanisms. Preclinical polysubstance use research typically involves alcohol, nicotine, or \ncocaine, with limited studies on cannabinoids, hallucinogens, and opioids. Within preclinical \nopioid research, heroin is commonly administered over prescription opioids or fentanyl. Even \nthough alcohol and opioid co-use is quite common, animal studies involving the combination of \nthese two substances are lacking. Furthermore, many current studies lack additional features of \nrealistic human substance use, such as a group-housed social environment during use and \nvoluntary, continuous access to multiple substances and concentrations. \n \nTo address these gaps, the current study investigated voluntary intake of alcohol, fentanyl, and \nwater in a group-housed environment in adult male and female mice. To do this, we utilized the \nSocially Integrated Polysubstance (SIP) system, which allows rodents to remain group-housed \nwhile self-administering substances with continuous monitoring and intake measurement\n22. \nPrevious research using SIP cages in our lab revealed differences in activity and flavor \npreference between male and female rodents, offering insights into how sex may influence \nsubstance preference and behavior patterns. \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nMaterials and Methods \n \nAnimals \nAll experiments utilized female and male (as determined by genital appearance at weaning) \nC57BL/6 mice from Jackson Labs aged 9-11 weeks of age at time of arrival to VA Puget Sound. \nMice were housed by sex in cages of four on a 12:12 light:dark cycle (lights on at 06:00), and \nwere given ad libitum food and water. All animal experiments were carried out in accordance \nwith AAALAC guidelines and were approved by the VA Puget Sound IACUC. Mice were \nacclimated to the VA for one week following arrival and subsequently handled for an additional \nweek prior to experimental manipulation. To increase rigor and reproducibility, the study \nincluded at least two cohorts of mice each run at separate times.  \n \nBaseline behavioral testing \nOne week prior to housing in the SIP cages, animals were tested in the open field and then at \nleast 24 hours later in the elevated zero maze to assess locomotion and anxiety-like behavior. \nOn each day of testing, animals were allowed at least 30 minutes to acclimate to the testing \nroom.  \nOpen field box (OFB): Mice were allowed 5 minutes to explore a large circular open space (1 \nmeter diameter) and their movements were recorded from above and analyzed with Anymaze \n(Wood Dale, IL). Decreased time spent in the middle of the OFB is indicative of an anxiety-like \nphenotype.  \nElevated zero maze (EZM): Mice were allowed 5 minutes to explore an elevated zero maze \n(Maze Engineers, Skokie, IL) and their movements were recorded from above and analyzed \nwith Anymaze (Wood Dale, IL). Decreased time spent exploring the open arms is thought to \nreflect anxiety-like behavior. \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nRFID transponder implantation \nEach RFID transponder (Euro I.D., Koln, Germany) is coated in a biocompatible glass material \nand is 2.12 mm x 12 mm diameter. At least 72 hours prior to SIP cage housing, each \ntransponder is sterilized and injected subcutaneously behind the shoulder blades of an \nanesthetized mouse (5% isoflurane) using the provided syringe applicator.  \n \nSocially Integrated Polysubstance (SIP) system  \nAs previously described, the SIP system enables group housed mice to self-administer multiple \ndifferent substances in a home-cage setting while still maintaining individual intake levels on a \nsecond-to-second time scale (Wong et al., 2023). The current study employed a setup with six \ndrinking stations in a rectangular home cage design (3 drinking stations on each long wall). Mice \nwere housed for seven days with continuous access to water (2 drinking stations, one on each \nwall), two different doses of ethanol (5% and 10%) and two different doses of fentanyl (5 ug/ml \nand 20 ug/ml). Cages were checked daily, and food was available ad libidum. Custom Python \nscripts were used to integrate the RFID and VDM data streams via common timestamps and \nare available at https://github.com/grace3999/SIP_Polysubstance\n. \n \nUnsupervised machine learning (cluster analysis) of baseline behavioral testing \nGiven the high degree of collinearity across the 12 behavioral parameters collected from the \nOFB and EZM, we first performed a dimensionality reduction step using Principal Component \nAnalysis (PCA). We then used the first three principal components (explaining over 75% of \nmodel variance) in a K-means cluster-based approach. Cluster stability was assessed as \npreviously described\n23, using the scores for homogeneity, adjusted Rand, and adjusted mutual \ninformation criterion and a bootstrap approach with repeated random assignment of initial \ncluster centroids. K=3 clusters was chosen based on the above evaluation metrics.  \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nStatistical analysis \nData are expressed as mean ± SEM. Differences between groups were determined using a two-\ntailed Student’s t-test, one-way analysis of variance (ANOVA), or two-way (repeated measures \nwhen appropriate) ANOVA followed by post hoc testing using Bonferroni’s Multiple Comparison. \nReported p values denote two-tailed probabilities of p ≤  0.05 and non-significance (n.s.) \nindicates p > 0.05. Statistical analysis and visualization were conducted using Graph Pad Prism \n9.0 (GraphPad Software, Inc., La Jolla, CA) and with custom Python scripts \n(https://github.com/grace3999/SIP_Polysubstance).  \n \n \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nResults \nFifty-six male (n=32) and female (n=24) adult mice were housed in groups of 4 for one week in \nthe Socially Integrated Polysubstance (SIP) system with continuous access to water, two doses \nof ethanol (5% and 10%) and two doses of fentanyl (5 ug/ml and 20 ug/ml). Visits to the drinking \nchambers were collected at 100Hz and drinking data was collected at 1Hz. Across the 56 mice, \nthe data set included over 650,000 RFID data points and over 45,000 drinking data points.  \n \nVisit summary data by sex: Female mice spent more time than male mice in the drinking \nchambers in total (Student’s unpaired t-test, t[54]=5.12, p<0.0001) (Figure 1a) and when \nanalyzed across days (two-way RM ANOVA: interaction effect F[6,324]=1.47, p>0.05, main \neffect Sex F[1,54]=26.3, p<0.0001, main effect Day F[6,324]=5.7, p>0.0001; Bonferroni Multiple \nComparison Test (BMCT) post hoc) (Figure 1b), and both sexes decreased time spent in the \nchambers as days in the SIP system progressed. Female mice spent more time in the drinking \nchambers specifically during the dark cycle (with both sexes showing decreased time spent \nduring the light vs. dark cycle) (two-way RM ANOVA: interaction effect F[1,54]=24.9, p<0.001, \nmain effect Sex F[1,54]=26.3, p<0.0001, main effect Cycle F[1,54]=206.3, p>0.0001; BMCT \npost hoc) (Figure 1c). Likewise, female mice spent more time in the drinking chambers during all \nhours of the dark cycle and some hours of the light cycle (two-way RM ANOVA: interaction \neffect F[23,1242]=10.9, p<0.001, main effect Sex F[1,54]=26.3, p<0.0001, main effect Zeitgeber \nF[23,1242]=121.2, p>0.0001; BMCT post hoc) (Figure 1d). Heat maps depicting average time \nspent in the drinking chambers for male and female mice across days and zeitgeber time are \nshown in Figure 1e.  \n \nDrinking summary data by sex: Female and male mice did not differ in the total amount of liquid \nconsumed (Student’s unpaired t-test, t[54]=1.1, p>0.05) (Figure 1f). When analyzed across \ndays, there was a significant interaction effect but no main effect of sex (two-way RM ANOVA: \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\ninteraction effect F[6,324]=6.4, p<0.001, main effect Sex F[1,54]=1.2, p>0.05, main effect Day \nF[6,324]=3.8, p>0.0001; BMCT post hoc) (Figure 1g), with both sexes increasing amount \nconsumed as days in the SIP system progressed. When examined by light/dark cycle, there \nwas a significant interaction effect but no main effect of sex (with both sexes showing a \ndecrease in amount consumed during the light vs. dark cycle) (two-way RM ANOVA: interaction \neffect F[1,54]=5.6, p<0.02, main effect Sex F[1,54]=1.2, p>0.05, main effect Cycle \nF[1,54]=356.7, p>0.0001; BMCT post hoc) (Figure 1h). Likewise, when examined by zeitgeber \ntime, there was a significant interaction effect with female mice consuming more liquid during all \nhours of the dark cycle and some hours of the light cycle (two-way RM ANOVA: interaction \neffect F[23,1242]=3.1, p<0.001, main effect Sex F[1,54]=1.2, p>0.05, main effect Zeitgeber \nF[23,1242]=3.0, p>0.0001; BMCT post hoc) (Figure 1i). Heat maps depicting the average \namount consumed in drinking chambers for male and female mice across days and zeitgeber \ntime are shown in Figure 1j.  \n \nVisit individual substance data by sex: When summarized across all days in the SIP system, \nthere were significant main effects of sex and substance type but no significant interaction effect \n(two-way RM ANOVA: interaction effect F[4,216]=0.3, p>0.05, main effect Sex F[1,54]=26.3, \np<0.0001, main effect Substance F[4,216]=25.0, p>0.0001; BMCT post hoc) (Figure 2a). \nPotential differences in time spent in the drinking chambers for males vs. females across \nsubstances and light/dark cycle was examined using a three-way ANOVA (three-way RM \nANOVA: 3 way interaction effect F[4,216]=0.5, p>0.05) (Figure 2b; see Table 1 for statistical \nresults). Finally, we examined potential differences in male vs. female mice across days for \neach substance separately (Figure 2c; see Table 2 for statistical results). Heat maps depicting \nthe average time spent in each substance drinking chamber for male and female mice across \ndays and zeitgeber time and the total time spent in each substance drinking chamber across \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nindividual mice are shown in Figure 3a and 3b, respectively. Finally, Figure 3c shows a raster \nplot of chamber visits for four example mice (2 male and 2 female). \n \nDrinking individual substance data by sex: When summarized across all days, there was a \nsignificant interaction effect (two-way RM ANOVA: interaction effect F[4,216]=4.6, p<0.001, \nmain effect Sex F[1,54]=1.2, p>0.05, main effect Substance F[4,216]=23.7, p>0.0001; BMCT \npost hoc) (Figure 2d). Potential differences in amount consumed for males vs. females across \nsubstances and light/dark cycle was examined using a three-way ANOVA (three-way RM \nANOVA: 3 way interaction effect F[4,216]=4.4, p<0.01) (Figure 2e; see Table 3 for statistical \nresults). Finally, we examined potential differences in male vs. female mice across days for \neach substance separately (Figure 2f; see Table 4 for statistical results). Heat maps depicting \nthe average amount of each substance consumed for males and females across days and \nzeitgeber time are shown in Figure 3d. Heat maps depicting the amount of each type of \nsubstance consumed across individual mice are shown in Figure 3e. Finally, Figure 3f shows a \nraster plot of chamber visits for four example mice (2 male and 2 female). \n \nWhen examining total intake, preference for both alcohol (Student’s unpaired t-test, t[54]=2.37, \np<0.05) (Figure 2g) and fentanyl (Student’s unpaired t-test, t[54]=3.1, p<0.01) (Figure 2h) was \nhigher in females. When examined across days, there were significant main effects of Sex and \nDay but no significant interaction effect for ethanol preference (two-way RM ANOVA: interaction \neffect F[6,324]=1.6, p>0.05, main effect Sex F[1,54]=8.2, p<0.01, main effect Day F[6,324]=1.8, \np>0.05; BMCT post hoc) (Figure 2i). Conversely, when examining fentanyl preference across \ndays, there was a significant interaction effect and significant main effects of Sex and Day (two-\nway RM ANOVA: interaction effect F[6,324]=2.9, p<0.01, main effect Sex F[1,54]=9.3, p<0.01, \nmain effect Day F[6,324]=6.0, p<0.001; BMCT post hoc) (Figure 2j). Finally, we examined dose \npreference in males vs. females for alcohol (doses available were 5% and 10%) and fentanyl \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\n(doses available were 5 ug and 10 ug/ml) (Figure 2k-n). When examining total intake, dose \npreference for alcohol was not significantly different between males and females (Student’s \nunpaired t-test, t[54]=1.54, p>0.05) (Figure 2k) nor was dose preference for fentanyl (Student’s \nunpaired t-test, t[54]=0.67, p>0.05) (Figure 2l). When examined across days, there was only a \nsignificant main effect of Day but not Sex and no significant interaction effect for ethanol dose \npreference (two-way RM ANOVA: interaction effect F[6,168]=1.0, p>0.05, main effect Sex \nF[1,54]=1.7, p>0.05, main effect Day F[6,324]=6.5, p<0.001; BMCT post hoc) (Figure 2m). \nLikewise, when examining fentanyl dose preference across days, there was only a significant \nmain effect of Day but not Sex and no significant interaction effect for ethanol dose preference \n(two-way RM ANOVA: interaction effect F[6,324]=1.0, p>0.05, main effect Sex F[1,54]=1.7, \np>0.05, main effect Day F[6,324]=2.7, p<0.05; BMCT post hoc) (Figure 2n).  \n \nBaseline behavioral clustering: In addition to finding differences between male and female mice, \nwe also hypothesized that baseline behavioral phenotypes might map on to subsequent \npolysubstance use profiles. One week prior to the start of housing in the SIP cages, mice were \ntested in the OFB and EZM. While male and female mice did not differ significantly in locomotor \nor anxiety-like metrics in the OFB (Figure 4a-f) or in the EZM (Figure 4g-l) (see Table 5 for \nstatistical results), there was large amount of variability across animals, leading us to \nhypothesize that we could identify phenotypic sub-groups by using an unsupervised cluster-\nbased approach. Given the high degree of collinearity across the 12 behavioral parameters \ncollected from the OFB and EZM, we first performed a dimensionality reduction step using \nPrincipal Component Analysis (PCA) (Figure 4m-o). We then used the first three principal \ncomponents (explaining over 75% of model variance) in a K-means cluster-based approach. \nAnalysis of cluster stability supported a three-cluster solution (Table 6; Figure 4p-r). Using k=3, \nthere is a non-significant trend for a different distribution of cluster assignment across male and \nfemale mice (Chi\n2 = 4.5, p=0.1) (Figure 4s). To determine whether OFB and EZM behavior \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\ndiffered across clusters, we assessed the 12 behavioral parameters when grouped by cluster \nassignment. Behavior across clusters differed significantly on all 12 parameters examined \nexcept for EZM open arm time (see Table 7 for statistical results) (Figure 4t-ae).  \n \nVisit summary data by behavioral cluster: There was no significant difference across clusters for \ntotal time spent in the drinking chambers (one-way ANOVA: F[2,53]=1.22, p>0.05) (Figure 5a). \nWhen analyzed across days, there was a significant effect of Day but not Cluster (two-way RM \nANOVA: interaction effect F[12,318]=1.62, p>0.05, main effect Cluster F[2,53]=1.2, p>0.05, \nmain effect Day F[6,318]=5.7, p<0.0001; BMCT post hoc) (Figure 5b). Likewise, when examined \nby light/dark cycle, there was a significant effect of Cycle but not Cluster (with all clusters \nshowing decreased time spent during the light vs. dark cycle)  (two-way RM ANOVA: interaction \neffect F[2,53]=1.3, p>0.05, main effect Cluster F[2,53]=1.2, p>0.05, main effect Cycle \nF[1,53]=151.3, p<0.0001; BMCT post hoc) (Figure 5c). Finally, when examined by zeitgeber \ntime, there was a significant effect of Time but not Cluster (two-way RM ANOVA: interaction \neffect F[46,1219]=1.2, p>0.05, main effect Cluster F[2,53]=1.2, p>0.05, main effect Time \nF[23,1219]=103.6, p<0.0001; BMCT post hoc) (Figure 5c). Heat maps depicting the average \ntime spent in the drinking chambers for males and females across days and zeitgeber time are \nshown in Figure 5e.  \n \nDrinking summary data by behavioral cluster: There was no significant difference across \nclusters for total liquid consumed (one-way ANOVA: F[2,53]=0.84, p>0.05) (Figure 5f). When \nanalyzed across days, there was a significant effect of Day but not Cluster (two-way RM \nANOVA: interaction effect F[12,318]=0.43, p>0.05, main effect Cluster F[2,53]=0.84, p>0.05, \nmain effect Day F[6,318]=8.7, p<0.0001; BMCT post hoc) (Figure 5g). Likewise, when examined \nby light/dark cycle, there was a significant effect of Cycle but not Cluster (with all clusters \nshowing decreased time spent during the light vs. dark cycle)  (two-way RM ANOVA: interaction \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\neffect F[2,53]=0.01, p>0.05, main effect Cluster F[2,53]=0.8, p>0.05, main effect Cycle \nF[1,53]=318.4, p<0.0001; BMCT post hoc) (Figure 5h). When examined by zeitgeber time, there \nwas a significant effect of Time but not Cluster (two-way RM ANOVA: interaction effect \nF[46,1219]=0.8, p>0.05, main effect Cluster F[2,53]=0.9, p>0.05, main effect Time \nF[23,1219]=97.2, p<0.0001; BMCT post hoc) (Figure 5i). Heat maps depicting average time \nspent in the drinking chambers for male and female mice across days and zeitgeber time are \nshown in Figure 5j.  \n \nVisit data by individual substance and behavioral cluster: When summarized across all days in \nthe SIP system, there was a significant interaction between Cluster and Substance and main \neffect of Substance type but no main effect of Cluster  (two-way RM ANOVA: interaction effect \nF[8,212]=2.63, p<0.01, main effect Cluster F[2,53]=1.2, p>0.05, main effect Substance \nF[4,212]=26.8, p>0.0001; Benjamini/Hochberg FDR correction) (Figure 6a). Potential \ndifferences in time spent in the drinking chambers for each cluster across substances and \nlight/dark cycle was examined using a separate two-way ANOVA for each cluster. For clusters 0 \nand 1 there was a significant main effect of Cycle but no main effect of substance or interaction \neffect. Conversely, for cluster 2 there were significant main effects of Cycle, Substance and a \nsignificant interaction (Figure 6b; see Table 8 for statistical results). Finally, we examined \npotential differences across days and behavioral clusters for each substance separately (Figure \n6c; see Table 9 for statistical results). Heat maps depicting average time spent in each type of \nsubstance drinking chambers for each cluster across days and zeitgeber time are shown in \nFigure 7a.   \n \nDrinking data by individual substance and behavioral cluster: When summarized across all days \nin the SIP system, there was a significant interaction between Cluster and Substance and main \neffect of Substance type but no main effect of Cluster (two-way RM ANOVA: interaction effect \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nF[8,212]=2.55, p<0.05, main effect Cluster F[2,53]=0.85, p>0.05, main effect Substance \nF[4,212]=23.5, p>0.0001; Benjamini/Hochberg FDR correction) (Figure 6d). Potential \ndifferences in amount consumed for each cluster across substances and light/dark cycle was \nexamined using a separate two-way ANOVA for each cluster (Figure 6e). For clusters 0 and 2 \nthere was a significant interaction effect and significant main effects of Cycle and Substance. \nConversely, for cluster 1 there was only a significant main effect of Cycle but no main effect of \nsubstance or interaction effect (Figure 6e; see Table 10 for statistical results). Next, we \nexamined potential differences across days and behavioral cluster assignment for each \nsubstance separately (Figure 6f; see Table 11 for statistical results). Heat maps depicting the \naverage amount of each substance consumed for each cluster across days and zeitgeber time \nare shown in Figure 7b.   \n \nWhen examining total intake, neither preference for alcohol nor fentanyl was significantly \ndifferent across clusters (one-way ANOVA: F[2,53]=2.6, p>0.05, Figure 6g; one-way ANOVA: \nF[2,53]=2.2, p>0.05, Figure 6h). When examined across days, there was a significant main \neffect of Day but not Cluster or interaction effect for both ethanol and fentanyl preference \n(Figure 6i-j; see Table 12 for statistical results). Finally, we examined potential differences \nacross behavioral clusters in the dose preference for alcohol and fentanyl (Figure 6k-n). When \nexamining total intake, dose preference for alcohol was not significantly different across clusters \n(one-way ANOVA: F[2,53]=1.9, p>0.05) (Figure 6k). Conversely, dose preference for fentanyl \nwas significantly different across clusters (one-way ANOVA: F[2,53]=4.4, p<0.05) (Figure 6l). \nWhen examined across days, there was only a significant main effect of Day for ethanol dose \npreference, while there were significant main effects of Day and Cluster for fentanyl dose \npreference (Figure 6m-n; see Table 13 for statistical results).  \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nDiscussion \nThis study aimed to better understand how individual differences influence alcohol and opioid \npolysubstance use in male and female mice. We identified multiple parameters related to \ndrinking activity that differed according to sex. We also uncovered three discrete clusters of \nmice based on behavioral phenotypes that had unique drinking patterns. Together our results \ndemonstrate the utility of studying polysubstance use in group housed mice and support the \noverarching notion that baseline behavioral phenotypes map onto substance use and \npreference patterns.  \n \nThe first outcome that we measured was activity level, determined by number of visits to and \ntime spent in the drinking chambers (registered by RFID sensor). While number of visits and \ntime spent in the drinking chambers is an imperfect measure of activity, it gives an initial \nbaseline to build from. Both male and female mice decreased time spent in the chambers \nacross the seven days in the SIP system, but female mice spent more time in the drinking \nchambers each day. This agrees with previous rodent studies that found increased locomotion \nin female rodents compared to males after chronic alcohol, fentanyl, or morphine \nadministration\n24–27. It is unclear why differences in locomotion exist between male and female \nrodents following alcohol and/or opioid consumption. One possible explanation could be \ndifferences in metabolism and how these substances physiologically affect males and females, \nor potentially differences in the rewarding or aversive neural properties of a substance. \nImportantly, we did not track estrous cycle in the female mice. While changes in estrous cycle \ncould potentially influence the reinforcing effects of fentanyl, previous studies have shown that \nestrous cycle likely does not impact locomotor behavior\n28,29.  \n \nWhen looking at intake across the five available substances (water, 5% and 10% alcohol, 5% \nand 10% fentanyl), there were sex differences in substance intake pattern and preference. On \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\naverage, male mice consumed the most water, followed closely by 5% fentanyl, small amounts \nof 5% and 10% ethanol, and the lowest volume of 20% fentanyl. The highest total intake for \nfemale mice was 5% fentanyl, then water, closely followed by 20% fentanyl, 5% ethanol, and \nthe smallest volume of 10% ethanol.  Males had a slight preference for alcohol over water and a \nmoderate preference for fentanyl over water, while females had a moderate preference for \nalcohol and a strong preference for fentanyl over water. In females, the preference for alcohol \nover water decreased over time, but fentanyl preference escalated over time. Fentanyl \npreference remained generally consistent for the male mice. There were no statistical \ndifferences in dose preference between male and female mice.  \n \nOur results generally corroborate trends seen previously. Female mice tend to consume higher \namounts of ethanol24,30  and fentanyl31–33 relative to their body weight compared to male mice. \nAnother study found that female rats drank larger volumes of a 5% dose of ethanol compared to \nmale mice, as well as compared to other higher doses of ethanol, showing the importance of \nincluding multiple doses of substances24. There is also evidence in both human and rodent \nstudies that females will escalate from initial and moderate substance consumption to \ndisordered use or addiction more quickly than males\n32,34, which mirrors what we saw with the \nfemale mice escalating fentanyl preference during the seven days.  \n \nOne striking result from this study is the high variability in consumption, not only between mice \nbut also across days within individual mice. The constant access and voluntary consumption \nmodel of the SIP system provides an abundance of data regarding the timing and dose \npreference patterns for each individual mouse. When looking to the clinical literature to uncover \nmotivators underlying choice in substance use, it appears choice is often driven by stress-\nrelated experience, social environment, or personality traits such as impulsivity and maladaptive \ncoping strategies\n9,15,35–37. To test this concept using our SIP system, we decided to assess \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nlocomotion and anxiety-like behaviors one week prior to housing in the SIP cages to investigate \nany correlations between behavior and substance use patterns.  \n \nOur initial examination revealed no significant sex differences on any of the twelve parameters \nin the open field (OFB) and elevated zero maze (EZM) tests. While behavioral tests have some \ndegree of variability, typically female rodents show lower anxiety-like behaviors, with no sex \ndifferences in novelty-seeking behavior (although this can depend on estrous phase)38,39. \nBecause there was a considerable amount of variability across mice in our study, we \nhypothesized that the range of behavioral profiles might map on polysubstance use patterns. \nAfter dimensionality reduction and an unsupervised clustering analysis based on the 12 \nbehavioral parameters, three distinct groups of mice were revealed. The composition of male \nand female mice in cluster zero had 7 males and 4 females, cluster one had 5 females and 1 \nmale, and cluster two had 24 males and 15 females; this distribution was trending but non-\nsignificant when tested statistically. The clusters did statistically differ in 11 of 12 behavioral \nparameters (all except EZM open arm time) which suggests we identified three distinct \nbehaviorally phenotypic subgroups. Cluster 0 was defined by higher anxiety-like behaviors, \nincluding less distance traveled in the center of the OFB and in the open arms of the EZM and \nlongest latency to enter the center area/open arms. Cluster 1 had the longest time spent in the \ncenter of the OFB and open arms of the EZM, and shortest latency to enter the center \narea/open arms, suggesting lower anxiety-like behavior. \n \nFinally, we projected the three clusters onto the substance consumption data. Although this \nwould not prove a causal relationship between behavioral phenotypes and polysubstance use \npatterns, it certainly provides beneficial insight and highlights predictive ability. There were \nmeaningful differences in consumption patterns between the three clusters, with cluster 0 \ndrinking a high amount of 5% fentanyl and a moderate amount of water; cluster 1 consuming a \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nhigh amount of 20% fentanyl and a moderate amount of 5% fentanyl; and cluster 2 consuming a \nhigh amount of water, moderate 5% fentanyl and small amount of 5% ethanol. There were no \nsignificant differences between clusters for ethanol or fentanyl preference over water, or for \nalcohol dose preference, but fentanyl dose preference was higher for cluster 1 compared to \nclusters 0 and 2 and increased over the course of the seven days of substance access.  \n \nTaken all together, it appears that cluster 1 consists of majority female mice, shows lower \nanxiety-like behavior, and preferentially consumes a higher dose of fentanyl. Previous studies \nhave found mixed results relating anxiety-like and novelty-seeking behaviors with higher opioid \nconsumption21. In our study, Cluster 1 showed more exploratory and less anxious behavior and \nthe highest consumption of fentanyl. Surprisingly, the cluster with the highest anxiety-like \nbehavior (cluster 0) did not have the highest preference for ethanol, as has been shown before \nin the literature\n17–19. This could be because the mice had access to fentanyl in addition to the \nethanol, the 24-hour access of the alcohol, the concentration of ethanol, or because there were \nno stressors prior to substance availability.  \n \nTo our knowledge, there are only two other studies that consist of simultaneous or sequential \n(respectively) voluntary administration of an opioid and alcohol\n5,40. Both studies used oxycodone \nin limited access operant chamber models, and specifically captured the effect of forced \nwithdrawal from oxycodone on alcohol consumption. In line with our research, Wilkinson et al., \nalso found that male and female rats with access to oxycodone consumed less alcohol than rats \nthat only had access to sucrose. Neither study conducted behavioral testing before alcohol or \nopioid administration. While there are some meaningful differences that prevent direct \ncomparison between these studies and our experiments here, a main takeaway is the persistent \nexistence of sex differences in polysubstance use and behavioral profiles across a variety of \nhousing conditions and access paradigms.  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\n \nOur results provide an initial characterization of some of the fundamental parameters \nsurrounding polysubstance use in a preclinical model, and the interpretation is constrained by \nthe scope of the experiment. We began with seven days in the SIP cages, and while we \nobserved escalation of consumption and dose preference, a longer experimental timeline will be \ncritical to understand the transition from casual substance use to development of an SUD-like \nphenotype. We provided continuous access to both alcohol and fentanyl, and an intermittent \naccess paradigm may reveal different patterns of use. We relied on drinking chamber visits to \ndetermine activity, which could not accurately reflect total locomotion. The addition of a stressor, \nor period of extinction/deprivation of a substance would also help improve our understanding of \ndrug seeking and motivations for consumption. Age of first exposure is known to have \nsignificant implications for future substance consumption and behavioral and biological \noutcomes; so inclusion of animal models across the lifespan is important as well\n4,24,41,42. Future \nstudies should investigate the mechanisms underlying drug metabolism and pharmacology and \nhow it affects other related behaviors, including sex differences. Physiological measures and \nbiomarkers could play an important role in predicting future substance consumption patterns, \nconsequences of substance use, and treatment outcomes. \n \nThe SIP system provides an enriched social environment and voluntary consumption of multiple \nsubstances, and the possibilities for future studies using the SIP system are nearly unlimited. It \noffers the opportunity to continue interrogating the role of sex differences in substance use. It is \npertinent to acknowledge that our preclinical models do have limitations in uncovering the multi-\nfaceted and societal-driven motivations to consume substances that are cited in clinical studies, \nbut some indicators such as anxiety-like behaviors and stress responses are preserved across \nspecies. These basic behaviors may help us to reveal critical factors that influence substance \nuse. Overall, we hope this study underscores the need for more preclinical research on \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\npolysubstance use to better understand the patterns of consumption, treatment outcomes, and \nnovel therapeutic strategies. \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nReferences \n1. Crummy, E. A., O’Neal, T. J., Baskin, B. M. & Ferguson, S. M. One Is Not Enough: \nUnderstanding and Modeling Polysubstance Use. Front Neurosci 14, 569 (2020). \n2. Midanik, L. T., Tam, T. W. & Weisner, C. Concurrent and simultaneous drug and alcohol \nuse: Results of the 2000 National Alcohol Survey. Drug and Alcohol Dependence 90, 72–80 \n(2007). \n3. Meier, M. H. et al. Preparedness for healthy ageing and polysubstance use in long-term \ncannabis users: a population-representative longitudinal study. Lancet Healthy Longev 3, \ne703–e714 (2022). \n4. Compton, W. M., Valentino, R. J. & DuPont, R. L. Polysubstance use in the U.S. opioid \ncrisis. Mol Psychiatry 26, 41–50 (2021). \n5. Wilkinson, C. S. et al. Voluntary alcohol intake alters the motivation to seek intravenous \noxycodone and neuronal activation during the reinstatement of oxycodone and sucrose \nseeking. Sci Rep 13, 19174 (2023). \n6. Preston, K. L., Jobes, M. L., Phillips, K. A. & Epstein, D. H. Real-time assessment of alcohol \ndrinking and drug use in opioid-dependent polydrug users. Behav Pharmacol 27, 579–584 \n(2016). \n7. Tori, M. E., Larochelle, M. R. & Naimi, T. S. Alcohol or Benzodiazepine Co-involvement With \nOpioid Overdose Deaths in the United States, 1999-2017. JAMA Network Open 3, e202361 \n(2020). \n8. Cascalheira, C. J. et al. High-risk polysubstance use among LGBTQ+ people who use drugs \nin the United States: An application of syndemic theory. International Journal of Drug Policy \n118, 104103 (2023). \n9. DiGuiseppi, G. T., Davis, J. P., Christie, N. C. & Rice, E. Polysubstance use among youth \nexperiencing homelessness: The role of trauma, mental health, and social network \ncomposition. Drug and Alcohol Dependence 216, 108228 (2020). \n10. John, W. S. et al. Prevalence, patterns, and correlates of multiple substance use disorders \namong adult primary care patients. Drug and Alcohol Dependence 187, 79–87 (2018). \n11. McCabe, S. E., West, B. T., Jutkiewicz, E. M. & Boyd, C. J. Multiple DSM-5 substance use \ndisorders: A national study of US adults. Human Psychopharmacology: Clinical and \nExperimental 32, e2625 (2017). \n12. Tomczyk, S., Isensee, B. & Hanewinkel, R. Latent classes of polysubstance use among \nadolescents—a systematic review. Drug and Alcohol Dependence 160, 12–29 (2016). \n13. Tomczyk, S., Pedersen, A., Hanewinkel, R., Isensee, B. & Morgenstern, M. Polysubstance \nuse patterns and trajectories in vocational students — A latent transition analysis. Addictive \nBehaviors 58, 136–141 (2016). \n14. Compas, B. E. et al. Coping, emotion regulation, and psychopathology in childhood and \nadolescence: A meta-analysis and narrative review. Psychological Bulletin 143, 939–991 \n(\n2017). \n15. Elam, K. K., Mun, C. J., Connell, A. & Ha, T. Coping strategies as mediating mechanisms \nbetween adolescent polysubstance use classes and adult alcohol and substance use \ndisorders. Addictive Behaviors 139, 107586 (2023). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\n16. Metzger, I. W. et al. An examination of the impact of maladaptive coping on the association \nbetween stressor type and alcohol use in college. Journal of American College Health 65, \n534–541 (2017). \n17. Bahi, A. Individual differences in elevated plus-maze exploration predicted higher ethanol \nconsumption and preference in outbred mice. Pharmacology Biochemistry and Behavior \n105, 83–88 (2013). \n18. Hayton, S. J., Mahoney, M. K. & Olmstead, M. C. Behavioral Traits Predicting Alcohol \nDrinking in Outbred Rats: An Investigation of Anxiety, Novelty Seeking, and Cognitive \nFlexibility. Alcoholism: Clinical and Experimental Research 36, 594–603 (2012). \n19. Pelloux, Y., Costentin, J. & Duterte-Boucher, D. Differential involvement of anxiety and \nnovelty preference levels on oral ethanol consumption in rats. Psychopharmacology (Berl) \n232, 2711–2721 (2015). \n20. O’Brien, C., Vemireddy, R., Mohammed, U. & Barker, D. J. Stress reveals a specific \nbehavioral phenotype for opioid abuse susceptibility. J Exp Anal Behav 117, 518–531 \n(2022). \n21. Pelloux, Y., Costentin, J. & Duterte-Boucher, D. Novelty preference predicts place \npreference conditioning to morphine and its oral consumption in rats. Pharmacology \nBiochemistry and Behavior 84, 43–50 (2006). \n22. Wong, K. et al. Socially Integrated Polysubstance (SIP) system: An open-source solution for \ncontinuous monitoring of polysubstance fluid intake in group housed mice. Addiction \nNeuroscience 7, 100101 (2023). \n23. Baskin, B. et al. Repetitive Blast Exposure Increases Appetitive Motivation and Behavioral \nInflexibility in Male Mice. Front Behav Neurosci 15, 792648 (2021). \n24. Foo, J. C., Skorodumov, I., Spanagel, R. & Meinhardt, M. W. Sex- and age-specific effects \non the development of addiction and compulsive-like drinking in rats. Biol Sex Differ 14, 44 \n(2023). \n25. Gaulden, A. D. et al. Effects of Fentanyl on Acute Locomotor Activity, Behavioral \nSensitization, and Contextual Reward in Female and Male Rats. Drug Alcohol Depend 229, \n109101 (2021). \n26. Karami, M. & Zarrindast, M. R. Morphine sex-dependently induced place conditioning in \nadult Wistar rats. Eur J Pharmacol 582, 78–87 (2008). \n27. Zhan, B., MA, H.-Y., WANG, J.-L. & LIU, C.-B. Sex differences in morphine-induced \nbehavioral sensitization and social behaviors in ICR mice. Dongwuxue Yanjiu 36, 103–108 \n(2015). \n28. Babb, J. A., Constantino, N. J., Kaplan, G. B. & Chartoff, E. H. Estrous cycle dependent \nexpression of oxycodone conditioned reward in rats. Sci Rep 13, 13946 (2023). \n29. Hinds, N. M., Wojtas, I. D., Gallagher, C. A., Corbett, C. M. & Manvich, D. F. Effects of sex \nand estrous cycle on intravenous oxycodone self-administration and the reinstatement of \noxycodone-seeking behavior in rats. Front Behav Neurosci 17, 1143373 (2023). \n30. Flores-Bonilla, A. & Richardson, H. N. Sex Differences in the Neurobiology of Alcohol Use \nDi\nsorder. Alcohol Res 40, 04 (2020). \n31. Hammerslag, L. R. et al. Effects of the glucocorticoid receptor antagonist PT150 on stress-\ninduced fentanyl seeking in male and female rats. Psychopharmacology 238, 2439–2447 \n(2021). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\n32. Little, K. M. & Kosten, T. A. Focus on fentanyl in females: Sex and gender differences in the \nphysiological and behavioral effects of fentanyl. Frontiers in Neuroendocrinology 71, 101096 \n(2023). \n33. Malone, S. G., Keller, P. S., Hammerslag, L. R. & Bardo, M. T. Escalation and reinstatement \nof fentanyl self-administration in male and female rats. Psychopharmacology 238, 2261–\n2273 (2021). \n34. Becker, J. B. Sex differences in addiction. Dialogues Clin Neurosci 18, 395–402 (2016). \n35. Biolcati, R. & Passini, S. Development of the Substance Use Motives Measure (SUMM): A \ncomprehensive eight-factor model for alcohol/drugs consumption. Addictive Behaviors \nReports 10, 100199 (2019). \n36. Bonfiglio, N. S., Renati, R., Agus, M. & Penna, M. P. Development of the motivation to use \nsubstance questionnaire. Drug and Alcohol Dependence 234, 109414 (2022). \n37. Mahu, I. T., Barrett, S. P., Conrod, P. J., Bartel, S. J. & Stewart, S. H. Different drugs come \nwith different motives: Examining motives for substance use among people who engage in \npolysubstance use undergoing methadone maintenance therapy (MMT). Drug and Alcohol \nDependence 229, 109133 (2021). \n38. Fritz, A., Amrein, I. & Wolfer, D. P. Similar reliability and equivalent performance of female \nand male mice in the open field and water‐ maze place navigation task. Am J Med Genet C \nSemin Med Genet 175, 380–391 (2017). \n39. Lovick, T. A. & Zangrossi, H. Effect of Estrous Cycle on Behavior of Females in Rodent \nTests of Anxiety. Front. Psychiatry 12, (2021). \n40. Amico, K. N., Arnold, M. E., Dourron, M. S., Solomon, M. G. & Schank, J. R. The effect of \nconcurrent access to alcohol and oxycodone on self-administration and reinstatement in \nrats. Psychopharmacology 239, 3277–3286 (2022). \n41. Montemayor, B. N., Noland, M. & Barry, A. E. College students mandated to substance use \ncourses: Age-of-onset as a predictor of contemporary polysubstance use. Journal of \nAmerican College Health 0, 1–8 (2022). \n42. Pitkänen, T., Lyyra, A.-L. & Pulkkinen, L. Age of onset of drinking and the use of alcohol in \nadulthood: a follow-up study from age 8–42 for females and males. Addiction 100, 652–661 \n(2005). \n \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nTables: \n \nTable 1: time spent in chamber, male vs. female across substances and \nlight/dark cycle \nChamber 3-way ANOVA F(df) P Effect \nSex x Cycle F[1,54] = 21.9 <.0001 2 way interaction \nSex x Substance F[4,216] = 0.3 >0.05 2 way interaction \nCycle x Substance F[4,216] = 18.16 <.0001 2 way interaction \nSex F[1,54] = 26.3 <.0001 Main \nCycle F[1,54] = 221.5 <.0001 Main \nSubstance F[4,216] = 24.6 <.0001 Main \n \n \nTable 2: time spent in chamber \n    Chamber Interaction F(df) Interaction P Sex F(df) Sex P Day F(df) Day P\nwater F[6,324] = 1.9 0.08 F[1,54] = 4.9 0.03 F[6,324] = 3.6 0.002\netoh 5% F[6,324] = 2.6 0.02 F[1,54] = 13.8 <.0001 F[6,324] = 2.2 0.04 \netoh 10% F[6,324] = 1.2 0.3 F[1,54] = 20.7 <.0001 F[6,324] = 0.9 0.4  \nfent 5ug F[6,324] = 2.1 0.06 F[1,54] = 4.8 0.03 F[6,324] = 6.2 <.0001\nfent 20ug F[6,324] = 3.5 0.002 F[1,54] = 37.8 <.0001 F[6,324] = 6.2 <.0001\n \n \nTable 3: amount consumed, male vs. female across substances and \nlight/dark cycle \nChamber 3-way ANOVA F(df) P Effect \nSex x Cycle F[1,54] = 5.6 <0.05 2 way interaction \nSex x Substance F[4,216] = 4.6 <.01 2 way interaction \nCycle x Substance F[4,216] = 23.35 <.0001 2 way interaction \nSex F[1,54] = 1.2 >0.05 Main \nCycle F[1,54] = 362.2 <.0001 Main \nSubstance F[4,216] = 22.5 <.0001 Main \n \nTable 4: amount consumed \n    Chamber Interaction F(df) Interaction P Sex F(df) Sex P Day F(df) Day P\nwater F[6,324] = 1.1 0.34 F[1,54] = 5.9 0.02 F[6,324] = 1.5 0.16  \netoh 5% F[6,324] = 3.1 0.006 F[1,54] = 2.8 0.1 F[6,324] = 1.6 0.1  \netoh 10% F[6,324] = 2.0 0.07 F[1,54] = 0.11 0.74 F[6,324] = 1.2 0.35  \nfent 5ug F[6,324] = 3.4 0.003 F[1,54] = 2.5 0.12 F[6,324] = 12.7 <.0001\nfent 20ug F[6,324] = 2.0 0.07 F[1,54] = 11.1 0.002 F[6,324] = 1.0 0.45  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nTable 5: baseline behavioral clustering \n \nBehavior Test \nStudent's unpaired t-\ntest P \nOFB distance t[54] = 1.33 >0.05 \nOFB speed t[54] = 1.21 >0.05 \nOFB CT t[54] = 1.29 >0.05 \nOFB CE t[54] = 1.14 >0.05 \nOFB CL t[54] = 0.5 >0.05 \nOFB CD t[54] = 0.9 >0.05 \nEZM distance t[54] = 0.43 >0.05 \nEZM speed t[54] = 0.34 >0.05 \nEZM OAT t[54] = 0.5 >0.05 \nEZM  OAE t[54] = 1.52 >0.05 \nEZM OAL t[54] = 1.1 >0.05 \nEZM OAD t[54] = 1.1 >0.05 \n \n \nTable 7: three cluster behavioral parameters \nBehavior Test one-way ANOVA P  \nOFB distance F[2,53] = 44.1 <.0001 \nOFB speed F[2,53] =  42.7 <.0001 \nOFB CT F[2,53] =  11.8 <.0001 \nOFB CE F[2,53] = 26.8 <.0001 \nOFB CL F[2,53] = 7.0 <.01 \nOFB CD F[2,53] =  32.9 <.0001 \nEZM distance F[2,53] = 16.6 <.0001 \nEZM speed F[2,53] = 15.7 <.0001 \nEZM OAT F[2,53] = 1.6 >0.05 \nEZM  OAE F[2,53] = 28.3 <.0001 \nEZM OAL F[2,53] = 8.2 <.001 \nEZM OAD F[2,53] = 6.7 <.01 \nTable 6: cluster stability \n   \nk homogeneity completeness V measure \nadj. rand \ninfo. adj. mutual info\n3 0.79 0.67 0.72 0.71 0.71 \n4 0.76 0.74 0.75 0.72 0.73 \n5 0.77 0.77 0.77 0.67 0.74 \n6 0.74 0.77 0.76 0.63 0.72 \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nTable 8: time spent in chambers for each cluster across substances and light/dark cycle  \nChamber Interaction F(df) Interaction P Cycle F(df) Cycle P Substance F(df) Substance P\nCluster 0 F[4,40] = 1.7 >0.05 F[1,10] = 16.0 <0.01 F[4,40] = 2.4 >0.05 \nCluster 1 F[4,20] = 2.2 >0.05 F[1,5] = 14.0 <0.05 F[4,20] = 2.7 >0.05 \nCluster 2 F[4,152] = 41.7 <0.001 F[1,38] = 192.3 <.001 F[4,152] = 47.7 <.001 \n \nTable 9: time spent in chamber \n    Chamber Interaction F(df) Interaction P Cluster F(df) Cluster P Day F(df) Day P  \nwater F[12,318] = 1.4 0.15 F[2,53] = 0.67 0.52 F[6,318] = 3.6 0.002 \netoh 5% F[12,318] = 1.3 0.23 F[2,53] = 0.2 0.82 F[6,318] = 2.15 0.048 \netoh 10% F[12,318] = 1.8 0.048 F[2,53] = 1.6 0.22 F[6,318] = 0.9 0.47 \nfent 5ug F[12,318] = 0.9 0.5 F[2,53] = 2.6 0.08 F[6,318] = 6.1 <.0001 \nfent 20ug F[12,318] = 2.0 0.021 F[2,53] = 11.7 <.0001 F[6,318] = 6.15 <.0001 \n \nTable 10: amount consumed for each cluster across substances and light/dark cycle  \nChamber Interaction F(df) Interaction P Cycle F(df) Cycle P  Substance F(df) \nSubstance\nP \nCluster 0 F[4,152] = 19.7 <0.001 F[1,38] = 240.9 <0.001 F[4,152] = 18.1 <.001 \nCluster 1 F[4,20] = 1.8 >0.05 F[1,5] =28.0 <0.01 F[4,20] = 2.6 >0.05 \nCluster 2 F[4,152] = 17.7 <0.001 F[1,38] = 240.9 <.001 F[4,152] = 18.2 <.001 \n \nTable 11: amount consumed \n    Chamber Interaction F(df) Interaction P Cluster F(df) Cluster P Day F(df) Day P  \nwater F[12,318] = 0.2 0.9 F[2,53] = 2.5 0.09 F[6,318] = 1.5 0.2 \netoh 5% F[12,318] = 1.1 0.35 F[2,53] = 0.1 0.87 F[6,318] = 1.6 0.14 \netoh 10% F[12,318] = 0.7 0.74 F[2,53] = 0.12 0.89 F[6,318] = 1.0 0.37 \nfent 5ug F[12,318] = 0.9 0.56 F[2,53] = 1.6 0.21 F[6,318] = 12.1 <.0001 \nfent 20ug F[12,318] = 4.1 <.0001 F[2,53] = 19.3 <.0001 F[6,318] = 1.0 0.4 \n \nTable 12: ethanol/water and fentanyl/water preference across days \n  Substance Interaction F(df) Interaction P Cluster F(df) Cluster P Day F(df) Day P  \nAlcohol F[12,318] = 0.8 >0.05 F[2,53] = 2.3 >0.05 F[6,318] = 3.2 <.01 \nFentanyl F[12,318] = 0.7 >0.05 F[2,53] = 1.9 >0.05 F[6,318] = 5.7 <.001 \n \nTable 13: ethanol and fentanyl dose preference across days \n  Substance Interaction F(df) Interaction P Cluster F(df) Cluster P Day F(df) Day P  \nAlcohol F[12,318] = 0.6 >0.05 F[2,53] = 0.7 >0.05 F[6,318] = 6.2 <.001 \nFentanyl F[12,318] = 1.7 >0.05 F[2,53] = 5.6 <.01 F[6,318] = 2.7 <0.05 \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nABBREVIATIONS \n \nAUD: Alcohol Use Disorder \nBMCT: Bonferroni Multiple Comparison Test \nCD: Center Distance \nCE: Center Entries \nCL: Center Latency \nCT: Center Time \nEtOH: Ethanol \nEZM: Elevated Zero Maze \nFent: Fentanyl \nOAD: Open Arm Distance \nOAE: Open Arm Entries \nOAL: Open Arm Latency \nOAT: Open Arm Time \nOFB: Open Field Box \nOUD: Opioid Use Disorder \nPC: Principal Component \nSIP: Socially Integrated Polysubstance \nSUD: Substance Use Disorder  \n  \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nDISCLAIMER \nThe views expressed in this scientific presentation are those of the author(s) and do not reflect \nthe official policy or position of the U.S. government or Department of Veteran Affairs. \n  \nDECLARATIONS \nEthics approval and consent to participate \nAll animal experiments were conducted in accordance with Association for Assessment and \nAccreditation of Laboratory Animal Care guidelines and were approved by the VA Puget Sound \nInstitutional Animal Care and Use Committee. \n  \nConsent for publication \nNot applicable. \n  \nAvailability of data and materials \nThe data in this study are available from the corresponding author upon reasonable request. \n  \nCompeting interests \nThe authors declare that the research was conducted in the absence of any commercial or \nfinancial relationships that could be construed as a potential conflict of interest. \n \nFunding \nThis work was supported by grants from NIAAA Training Grant 5T32AA007455 (MP), NIDA \nTraining Grant 2T32DA007278-26 (BMB), UW NAPE Summer Undergraduate Research \nProgram NIDA DA048736 (KW), UW NAPE Pilot Program NIDA DA048736 (AGS), and \nDepartment of Veteran Affairs (VA) Basic Laboratory Research and Development (BLR&D) \nCareer Development Award 1IK2BX003258 (AGS). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\n \nAuthors' contributions \nThe work presented here was carried out in collaboration among all authors. MP, KW, BB, and \nAGS contributed to conception and design of the study. MP, ZCW, KW, SJL, ES, RN, BB, and \nAGS collected and analyzed data. MP and AGS wrote the first draft of the manuscript. All \nauthors contributed to manuscript revision, read, and approved the final manuscript. \n  \nAcknowledgements \nWe would like to thank Scott Ng Evans, Traci J. Weber, Cindy Pekow, DVM, Kari Koszdin, \nDVM, and Lena Strait-Bodey for technical assistance and veterinary care. \n  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nFigure Legends \n \nFigure 1: Activity and consumption in combined drinking chambers by sex \nA-E: Time spent in the drinking chambers in male and female mice in total (a), across days (b), \nacross light/dark cycle (c), and across Zeitgeber time (d); heatmaps shown in e. F-J: Amount of \nliquid consumed by male and female mice (normalized to body weight) in total (f), across days \n(g), across light/dark cycle (h), and across Zeitgeber time (i); heatmaps shown in j. Student's t-\ntest (a,f); Two-way RM ANOVA post hoc BMCT (b-d. g-h). **p \n≤  0.01, ****p ≤  0.0001. Values \nrepresent mean ± SEM. \n \nFigure 2: Activity and consumption in individual substance/dose drinking chambers by \nsex \nA-C: Time spent in each drinking chamber in male and female mice in total (a), across light/dark \ncycle (b), across days (c). D-F: Amount of liquid consumed by male and female mice \n(normalized to body weight) for each individual substance/dose combination in total (d), across \ndays (e), across light/dark cycle (f). G-J: Alcohol and fentanyl preference over water in total (g,h) \nand across days (i,j). K-N: Alcohol and fentanyl dose preference in total (k,l) and across days \n(m,n). Two-way RM ANOVA post hoc BMCT (a,c,d,f,i,j,m,n). Three-way RM ANOVA post hoc \nBMCT (b,e); Student's t-test (g,h,k,l). **p ≤  0.01, ****p ≤  0.0001. Values represent mean ± SEM. \n \nFigure 3: Heatmap and example raster plots for individual substances and mice  \nA. Heatmap of time spent in each drinking chamber in male and female mice across days and \nZeitgeber time. B. Heatmap of total time spent in each drinking chamber for individual mice. C. \nRaster plots of drinking chamber visits for 2 example male (left) and female (right) mice across a \nsingle day. D. Heatmap of amount consumed for each substance in male and female mice \nacross days and Zeitgeber time. E. Heatmap of amount consumed for each substance for \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nindividual mice. F. Raster plots of individual drinking events for 2 example male (left) and female \n(right) mice across a single day (same mice used as in c).  \n \nFigure 4: Baseline behavioral testing and cluster analysis \nA-F: Behavioral parameters measured in the OFB in male and female mice. G-L: Behavioral \nparameters measured in the EZM in male and female mice. M-O. PCA dimensionality reduction \nof 12 behavioral parameters measured in OF and EZM: explained variance by PC (m). Heatmap \nof PC loadings by behavioral parameter (n). PCA biplot (o). P-Q: Unsupervised k-means \nclustering metrics using first three behavioral PCs. S: Behavioral cluster assignment by sex. T-\nY: Behavioral parameters measured in the OFB by behavioral cluster. Z-AE: Behavioral \nparameters measured in the EZM behavioral cluster. Student's t-test (a-l); Chi\n2 (s); One-way \nANOVA post hoc BMCT (t-ae). **p ≤  0.01, ****p ≤  0.0001. Values represent mean ± SEM. \n \nFigure 5: Activity and consumption in combined drinking chambers by cluster \nA-E: Time spent in the drinking chambers for each cluster in total (a), across days (b), across \nlight/dark cycle (c), and across Zeitgeber time (d); heatmaps shown in e. F-J: Amount of liquid \nconsumed by mice in each cluster (normalized to body weight) in total (f), across days (g), \nacross light/dark cycle (h), and across Zeitgeber time (i); heatmaps shown in j. Student's t-test \n(a,f); Two-way RM ANOVA post hoc BMCT (b-d. g-h). **p \n≤  0.01, ****p ≤  0.0001. Values \nrepresent mean ± SEM. \n \nFigure 6: Activity and consumption in individual substance/dose drinking chambers by \ncluster \nA-C: Time spent in each drinking chamber for each cluster in total (a), across light/dark cycle \n(b), across days (c). D-F: Amount of liquid consumed by mice in each cluster (normalized to \nbody weight) for each individual substance/dose combination in total (d), across days (e), \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nacross light/dark cycle (f). G-J: Alcohol and fentanyl preference over water in total (g,h) and \nacross days (i,j). K-N: Alcohol and fentanyl dose preference in total (k,l) and across days (m,n). \nTwo-way RM ANOVA post hoc BMCT (a,c,d,f,i,j,m,n). Three-way RM ANOVA post hoc BMCT \n(b,e); Student's t-test (g,h,k,l). **p ≤  0.01, ****p ≤  0.0001. Values represent mean ± SEM. \n \nFigure 7: Heatmap and example raster plots for individual substance and mice \nA. Heatmap of time spent in each drinking chamber for each cluster across days and Zeitgeber \ntime. D. Heatmap of amount consumed for each substance by mice in each cluster. \n \n \n  \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nMale Female Male Female\nA. B.\nC.\n D.\nE.\nF. G.\nH. I.\nJ.\nFigure 1\n****\n****\n****\n****\nns\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nA. B.\nC.\nMale Female\nMale FemaleD.\nG.\nL.\nF.\nE.\nH. I. J.\nK. L. M. N.\nFigure 2\n****\n****\n****\n****\n****\n****\n**** ****\n****\n****\nns\nns\n****\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nEtOH05 EtOH10 Fent05 Fent20Water\nEtOH05 EtOH10 Fent05 Fent20Water\nA. B.\nE.D.\nC.\nF.\nFigure 3\nMaleFemaleMaleFemale\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nA. B. F.E.C. D.\nG. H. L.K.I. J.\nM.\nO.\nR.\nQ.P.\nS.N.\nT. U. X. Y.\nZ. AA. AD. AE.AC.AB.\nFigure 4\nns ns ns ns ns ns\nns ns\nns ns\nns ns\nns\nns\nns\n**** **** **** **** **** ****\n**** **** **** **** **** ****\n**\n**\nns ns\n****\n* *\n*\nV.\n**** ****\n****\nW.\n**** ****\n******** ******\n**** **** ****ns\n***\n***\n***\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nCluster 0 Cluster 1 Cluster 2 Cluster 0 Cluster 1 Cluster 2\nA. B.\nD.\nF. G.\nC. H. I.\nE. J.\nFigure 5\nns\nns\nns\n****\nns\nns\nns\n****\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nCluster 0 Cluster 1 Cluster 2\nCluster 0 Cluster 1 Cluster 2\nFigure 6\nA. B.\nC.\nD.\nG.\nF.\nE.\nH. I. J.\nK. L. M. N.\n****\n****\n** *\n**\nnsns\nns\nns\nnsns * *\nns ns\nns ns\n****\n**** ****\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint \n\nWaterEtOH05EtOH10Fent05Fent20\nCluster 0 Cluster 1 Cluster 2 Cluster 0 Cluster 1 Cluster 2\nFigure 7\nA. B.\n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted August 23, 2024. ; https://doi.org/10.1101/2024.08.22.609245doi: bioRxiv preprint","source_license":"Public-Domain","license_restricted":false}