Prospects for finding the mechanisms of sex differences in addiction with human and model organism genetic analysis

Abstract Despite substantial evidence for sex differences in addiction epidemiology, addiction‐relevant behaviors and associated neurobiological phenomena, the mechanisms and implications of these differences remain unknown. Genetic analysis in model organism is a potentially powerful and effective means of discovering the mechanisms that underlie sex differences in addiction. Human genetic studies are beginning to show precise risk variants that influence the mechanisms of addiction but typically lack sufficient power or neurobiological mechanistic access, particularly for the discovery of the mechanisms that underlie sex differences. Our thesis in this review is that genetic variation in model organisms are a promising approach that can complement these investigations to show the biological mechanisms that underlie sex differences in addiction.


| INTRODUCTION
Sex and gender differences in many aspects of drug use, drug effect and substance use disorders (SUDs) are well-documented, but the biological mechanisms underlying these differences and their implications for risk, prevention and treatment are poorly understood.
Although human genetic studies remain the standard for establishing genetic etiology for a complex neuropsychiatric condition like addiction, studies attempting to investigate sex-specific differences in processes of addiction largely lack the phenotypic breadth, power and neural tissue access needed to discover the underlying molecular mechanisms. In contrast, genetic studies in model organisms benefit from lower sample size requirements, and a wealth of in vivo technologies for research into mechanisms.
Model organism studies have showed or corroborated observations of sex and gender differences in addiction. Therefore, there is potential for genetic mapping of these phenomena to identify mechanisms of sex differences in addiction.
In this review, we highlight the epidemiological evidence for sex differences in SUDs, the influence of the environment on such sex differences and the promise and challenge of genetic analysis for discovering the biological mechanisms that underlie sex differences in addiction through the use of human and model organism discovery genetics. We highlight the major gender differences in phenomena related to SUDs, and the progress made toward studying these phenomena in human genetic and conventional model organisms studies. The literature on these phenomena is quite uneven with respect to the classes of drugs investigated across species and genetic backgrounds but show some general phenomena across classes of drugs. In other cases, the magnitude and direction of effects vary across strains and species. We have chosen examples from the literature that reflect this diversity.
Overall, few human genetic studies have succeeded in identifying sex-specific loci for addiction because of inadequate statistical power.
Although human genetic studies have been attempted for each major class of drugs, these studies are few in number and sex differences have been studied in even fewer. We have summarized these studies to call attention to this dearth of results. We argue that model organism studies show many sex differences in addiction-related phenomena, and that with recent improvements to genetic analysis methods in model organisms, these are now amenable to genetic investigation. Model organisms have benefits of (a) increased minor allele frequency, (b) controlled genetic background variation, (c) controlled environments, (d) fully ascertained drug exposure histories and (e) access to neurobiological intermediate phenotypes that serve to increase the observed effect size of genetic loci, revealing previously undiscovered causal mechanisms underlying sex differences that can be further interrogated.

| SEX AND GENDER DIFFERENCES IN ADDICTION EPIDEMIOLOGY
There are many documented differences in drug use and addiction between men and women. The Treatment Episode Data Set provided by the Substance Abuse and Mental Health Services Administration indicates sex differences in prevalence of substance use, the particular substances involved, the age at which drug use is initiated and drug use patterns. 1 These differences are brought about by an interplay of biological traits, and social and environmental influences. Socioeconomic status, life experiences and cultural factors including societal norms and gender expectations shape drinking and drug use. 2 According to the 2017 National Survey on Drug Use and Health, 53.6% of males and 45.7% of females 12 years or older use an illicit drug in their lifetime. 3 Among US adults 18 years of age and older, 71.8% of males and 59.6% of females consumed alcohol within a year. 4 Although men still consume more alcohol and illicit drugs than women, in the prevalence of substance use has become more similar across men and women over time. This has been attributed to societal factors including economic development and modernization, alcohol culture and gender equality, such that countries with greater equality show a narrowing of the gender gap in prevalence of use 5,6 and substance problem severity. 7

| GENDER DIFFERENCES IN ADDICTION TRAJECTORY
Males and females exhibit differences in the substances they use and in patterns of drug use. For example, men are more likely to smoke marijuana, while women are more likely to use alcohol and prescription drugs, including benzodiazepines and sedatives. [8][9][10] Women initiate drug use at lower doses than men, 11,12 but their drug use escalates more rapidly to addiction. Women report using substances to relieve stress or negative emotions, 13 and these negative emotions may be attributed to increased rates of sexual abuse, trauma and other stressors that are related to the cultural status of women. 14 By contrast, men more frequently report peer influence and reinforcing properties as reasons for drug use. 15,16 Environmental risk factors contribute to addiction vulnerability differently in each sex. For instance, women who experience spousal abuse often report being coerced to use drugs/alcohol 17 and suffer from increased risk of substance use problems. 18,19 Maternal and neonatal exposure to drugs, raise additional concerns for women and their exposed offspring. 20,21 Sexual minorities comprising the LGBTQ individuals experience higher rates of SUDs. 22,23 This is also attributed to societal environmental issues, history of traumatic life experience and other factors. [24][25][26] The extent to which biological factors play an intrinsic role in substance use in these populations beyond the effects of chromosomal sex on environmental influences is unknown.
Subjective effects of drugs in women are affected by the stage of the menstrual cycle. 27 Women have greater subjective responses to cocaine in the follicular phase of the menstrual cycle 28,29 when levels of estrogen are rising and progesterone levels are low. During the luteal phase, when progesterone levels are highest (estrogen levels are also elevated at this time), women report a reduction in the positive subjective effects of cocaine. 28,29 The physiological and positive subjective effects of cocaine are attenuated by exogenous progesterone in women, but not men. 30 Functional magnetic resonance imaging studies indicate changes in the reward-related neural system across the menstrual cycle and heightened reward responsivity shortly before ovulation. 31,32 Such changes in the functioning of reward-related circuits may underlie the premenstrual increases in drug cravings. 32 Taken together, these data highlight the importance of sex hormones in the modulation of drug effects in women. 33 Treatment outcomes also differ between the genders. Men and women face unique challenges in cessation of drug use. Women tend to enter treatment sooner after becoming substance dependent 34,35 and present with more co-occurring mood and anxiety disorders that complicate treatment. 36 Women are more susceptible to craving and face a greater risk of relapse following abstinence. [37][38][39] In summary, gender differences in human addiction are likely the result of sex-specific biological mechanisms that interact with sociocultural influences and life stressors that affect individuals of different genders differently. The complex biopsychosocial interactions underlying addiction make discovery of the biological basis of sex differences in humans challenging, yet identifying these biological mechanisms is critical for the development of more precisely tailored preventative and therapeutic interventions. 40

| Twin and adoption studies
It has long been appreciated that a family history of SUD is one of the strongest risk factors for the development of drug addiction. 42 Heritability estimates of SUDs obtained through family, adoption and twin studies [43][44][45] indicate a strong contribution of genetic factors in addiction.
In surveys of adult twins, heritability of addictive disorders has been estimated in a range of 0.39 for hallucinogens to 0.72 for cocaine. [46][47][48] A small number of these studies have assessed the heritability of substance use by gender of the subjects [49][50][51][52] and interactions of gender and genetic factors were ignored because researchers typically collapsed data across drug class and gender because of small sample sizes. Heritability estimates for cocaine use disorder (CUD) have been reported to be lower in women compared with men. 52 However, no mechanism has emerged to explain the heritability difference in CUD for males and females. 53 Recent meta analytic studies of alcohol use disorder (AUD) indicate that gender differences are more likely attributable to social influences on alcohol exposure and use than sexspecific genetic influences. 54,55 Genome wide linkage studies, genome-wide association studies (GWAS), whole genome sequencing and exome sequencing have begun to identify specific genes and genetic variants that explain the heritability of SUDs, although few such studies have reported gender specific genetic effects (Table 1).

| Linkage studies
Early genome-wide linkage studies 56,57 have reported chromosomal locations harboring risk loci for cocaine, 58 opioid, 59,60 nicotine 61,62 and alcohol dependence. 63 As noted above, few of these historical studies have analyzed the effect of sex on genetic linkage to substance dependence. One example is a study investigating the differential risk of opioid dependence in males and females, 64 which identified a significant sex-specific locus associated with opioid dependence and several other suggestive loci. The positional information showed by these legacy studies can now be combined with convergent data from more recent genomic strategies to implicate genes that may play a role in sex differences in substance dependence.

| Genome-wide association studies
For highly complex diseases like addiction that are influenced by vast numbers of genetic variants, population-based genome-wide association approaches are better suited to identify risk loci with relatively small effects compared with family-based genetic studies, 65,66 but the sample size requirements are substantial. Early studies suffered from low genetic marker density, 67 but increased power and precision in subsequent studies has allowed identification of some causal genes. 68,69 For example, the nuclear transcription factor PKNOX2 has been identified as a sex-specific candidate gene for composite substance dependence in women of European origin. 70 In the absence of heroic sample sizes, the power of GWAS studies is typically insufficient to detect statistically significant associations 53 especially in the context of sex differences. The limitations of small sample sizes have been overcome by combining GWAS data from multiple studies to allow statistical meta-analyses. 71,72 Significant genome-wide findings have now emerged for alcohol for example, 73 opioids for example, 74 cannabis for example 75 and nicotine dependence for example 76 (Table 1).
A recent GWAS 77 of alcohol consumption level and AUD using the large multiancestry sample (N = 274 424) from the Million Veteran Program 78 reported 18 genome-wide significant loci. The sexstratified sampling methodology allowed detection of female-specific signals despite the predominantly male sample, but power requirements precluded comparisons of Polygenic Risk Score by sex. In another recent GWAS analysis, a sex-specific variant at ADGRV1 was also identified with effects on opioid dependence risk in African American males. 79

| SEX DIFFERENCES AND GENETIC MECHANISMS OF RISK FACTORS IN ADDICTION AND RELAPSE
Some GWAS studies have incorporated the stress-related risk factors that contribute to the development of drug addiction and in addiction T A B L E 1 Number of genome-wide significant (P ≤ 5 × 10-8) associations for various substance dependence categories including alcohol, nicotine, opioids, cocaine, methamphetamine and cannabinoids, documented on GWAS catalog (https://www.ebi.ac.uk/gwas/) as of May  relapse susceptibility in GWAS. As noted above, the psychosocial and cultural factors that influence addiction epidemiology are largely attributable to stress. Stress increases vulnerability to drug addiction, 80,81 and this phenomenon differs between males and females. 82 Females, irrespective of their drug dependency status, report significantly higher anxiety, stress and negative mood during distress 37 Given that sex differences in gene expression and splicing patterns are widespread in the adult human brain, 89 studies of sex differences in gene expression in drug-exposed brains could provide insight into the mechanisms of addiction, but in humans, variability of drug exposure history and the challenges inherent in postmortem brain limit the utility of this approach.
Integrating genome-wide genetic findings with tissue-specific gene expression genetics could reveal additional biological mechanisms underlying substance dependence. Using this approach, Huggett and Stallings have identified a SNP associated with cocaine dependence and detected three genes (two loci) underlying this predisposition that displayed robust enrichment in numerous brain regions, including the hippocampus. 90  Model organisms, particularly rodents, have many conserved neurobiological and physiological features of humans and display specific facets of addiction-related behavior and neurobiological endophenotypes. As such, they can be used to identify causal mechanisms in brain-behavior relationships, including neurobiological and behavioral consequences of chronic substance exposure. 96 Animal research can show specific neurobiological mechanisms (eg, molecular, cellular or pharmacological) that mediate specific aspects of addiction. A distinct advantage of animal models is that the effects of an identified mechanism can be directly tested through specific neural manipulations on processes that mediate addictive behaviors. 97   Trait term "response to addictive substance trait," 39% exhibited strain by sex differences. As this number grows, and as genetic mapping studies in these populations are performed, discovery of mechanisms of addiction-related sex differences will be increasingly possible.
Reservations about including females have been based on the assumption that female hormonal cycles introduce substantial "noise" and complicate experimental studies relative to male-only studies, 103 but increased variability in females is not consistently observed. 102 In a genetic analysis of hundreds of measures from widely used behavioral assays in BXD recombinant inbred mice, within strain variability is similar for males and females for most assays. 104 In recent years, the number of studies that have included female subjects has increased. However, studies that have explicitly investi-

| SEX AS A MODULATOR OF GENETIC EFFECTS UNDERLYING RODENT ADDICTION-RELATED BEHAVIORS
There are several reviews addressing sex differences in SUD and addiction. 109 HR-LR phenotypes predict rapidity of acquiring cocaine selfadministration HR females self-administer more cocaine than HR males and both LR groups.
Preference for and reaction to novelty Exposure to PNS selectively facilitated the rate of acquisition and overall drug intake of males on an escalating-doses (Continues) Sex differences in behavioral effects of addictive drugs are widely observed, but the extent and magnitude of these differences vary across species, strains and even vendors ( Table 2). For example, female rodents exhibit heightened sensitivity to psychomotor stimulant [116][117][118] and reinforcing properties of cocaine. [119][120][121][122][123][124][125] Various sex differences are also observed for the other drug classes under multiple behavioral paradigms. [126][127][128][129][130] Although the magnitude and direction of the differences may vary in different species and strains, this is a "feature," not a "bug." It is precisely this variation that we are harnessing in the use of genetic variation to discover the biological mechanisms of sex differences. The

| GENETIC EFFECTS CAN INFLUENCE THE DETECTION OF SEX DIFFERENCES
Despite the general consensus that females are more sensitive to many drug effects, some studies report either no difference 138 or decreased sensitivity compared with males. 139,140 These disparate outcomes are caused by methodological variability and differences in age, species and strain of animals used ( Ovariectomy interacted with genetic background to alter anxiety-like behavior.
investigations of sex differences in addiction-related behaviors using rodents have relied on a single strain, with results that do not generalize across species or strains. The case in point is illustrated by a study of thermal nociception and morphine anti-nociception that discovered sex differences in some strains but not others. 131 Similar sex by genotype interactions are observed in locomotor stimulating effects of cocaine between male and female mice from genetically divergent strains 141 and are likely to be important for other drug classes as well. 130,142 The influences of sex and genotype has also been used to investigate other addiction-related traits. In a study with DO mice that harbor heterogenous genetic backgrounds, a sexspecific correlation of exploratory traits to drug-self administration was observed. 143 In a study with the inbred founder population of DO mice, sign tracking, a trait characterized by the tendency to pursue cues that predict the reward, 144,145 was observed to be influenced by both sex and strain of the mice, and across strains males had a greater sign tracking range than females. 143 Given the influence of genetic background on sex differences in drug response, it is possible to use genetic variation to identify the mechanisms of these differences.

| GENETICS AND GENOMIC TOOLS CAN SHOW THE NATURE OF SEX DIFFERENCES IN ADDICTION-RELATED BEHAVIOR
Rodent studies allow the evaluation of the nature of sex differences and to what extent they are attributable to chromosomal or hormonal influences. Sex differences are influenced by multiple separable and/or interacting sex-biasing factors. 146 Sex chromosome complement, and associated dosing of X and Y chromosome genes, is one such mechanism. These influences can be studied using the four-core genotypes mouse model. 147 In addition, gonadal phenotype and associated gonadal secretions, including sex steroids, can elicit both organizational effects (ie, slowly emerging and long-lasting effects of hormones that are initiated by, but not actively maintained by, steroid levels) or activational effects (ie, steroid effects that are induced and maintained by current hormone levels). It is possible to tease these latter effects out experimentally using specialized paradigms and approaches including measurements of steroid levels, gonadectomy and hormone replacement strategies. 148 Notably, all these mechanisms combine and interact with one another to sustain sex differences in phenotypes of interest, and once a sex difference is detected, additional studies using hormonal or chromosomal studies can provide further insight into how the genetic influence is modified by sex.
Quantitative trait locus (QTL) mapping is one of the primary genetic strategies used to show mechanisms underlying sex differences. Modern behavioral QTL studies of sex-specific loci are readily performed in recombinant inbred mouse strains C57BL/6JxDBA/2J (BXD), experimental crosses of closely related strains (C57BL/6J (B6) and C57L/J (C57)), 149 collaborative cross (CC) and DO mouse populations. 104,[150][151][152][153] Even though mice are unable to recapitulate the entire psychobiological diagnostic construct of addiction observed in humans, a set of alcohol and drugrelated phenotypes can be identified in mouse and humans to compare QTL data between the species. 154 Syntenic mapping of these traits allows us to determine to what extent similar genes influence a range of drug-related behaviors between the two species. Identification of sex-specific loci is facilitated by the reduced environmental and genetic variability in rodent genetic studies. For example, syntenic sex-specific QTL have been discovered that regulate alcohol consumption 155,156 and that mediate effects of opioids 157 using rodent models.
Minor allele frequencies are typically higher in mouse populations than human populations, rendering possible the detection of small effect alleles. Selective breeding for behavioral phenotypes that correlate with drug-seeking behaviors enriches and genetically fixes risk alleles. 158,159 In a study with high responder and low responder rats selectively bred based on exploratory locomotion in a novel environment, seven genome-wide significant loci accounted for approximately one-third of total variance and two-thirds of genetic variance selected for this trait. 160 Selective breeding has been applied to the investigation of several addictionrelated traits 114,161,162 that exhibit sex differences, including alcohol consumption 161 and cocaine self-administration. 162 Detection of small effect alleles, coupled to precisely defined and controlled phenotyping, allows the identification of previously unknown biological mechanisms of addiction.
Gene expression is also influenced by sex by genotype interactions, and expression QTL that represent genomic loci responsible for differential transcriptional regulation have been identified.
Through correlational analysis, we have been able to move beyond sex-specific QTL and identify sex-specific gene expression networks. 163 Such mechanisms of sex by genotype regulation of the molecular phenotypic variation have been observed in mental health issues in human 164 and are expected to emerge in addiction relevant regions of the brain. 165 High-diversity mouse populations, such as the CC and DO populations (for review, 166 ) with known and reproducible genetic variation provide a valuable platform for studying the mechanisms that drive sex dimorphic addiction-related traits.

| GENETIC ANALYSES OF SEX AS A MODULATOR IN GENE BY ENVIRONMENT INTERACTIONS
Model organisms allow the study of mechanisms of the interplay among, environmental and genetic interactions that contribute to sex differences in addiction vulnerabilities. Modulation of the interactions among stress-and drug-related traits by sex has been investigated in rodents for alcohol use 167 and other drugs. [168][169][170][171] Sex-dependent outcomes of gestational (prenatal) stress augment the rewarding and neurochemical-stimulating effects of the drug in rodents. 172,173 A recent study has identified sex-specific QTL that modulate responsiveness to cocaine following prenatal stress in offspring of BXD recombinant inbred mice. 174 Both sex 175,176 and strain 177,178 influence the response to stressors in the unpredictable chronic mild stress paradigm, and many sex differences in mouse behaviors are attributable to interactions with environmental variables. 179,180 Therefore, genetic studies can show mechanisms of sex differences in the stress response, and their role in addiction-related behaviors.

| CONCLUSIONS
In conclusion, genetic variation in humans and model organisms can

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.