Biological intersection of sex, age, and environment in the corticotropin releasing factor (CRF) system and alcohol

Abstract

The neuropeptide corticotropin-releasing factor (CRF) is critical in neural circuit function and behavior, particularly in the context of stress, anxiety, and addiction. Despite a wealth of preclinical evidence for the efficacy of CRF receptor 1 antagonists in reducing behavioral pathology associated with alcohol exposure, several clinical trials have had disappointing outcomes, possibly due to an underappreciation of the role of biological variables. Although he National Institutes of Health (NIH) now mandate the inclusion of sex as a biological variable in all clinical and preclinical research, the current state of knowledge in this area is based almost entirely on evidence from male subjects. Additionally, the influence of biological variables other than sex has received even less attention in the context of neuropeptide signaling. Age (particularly adolescent development) and housing conditions have been shown to affect CRF signaling and voluntary alcohol intake, and the interaction between these biological variables is particularly relevant to the role of the CRF system in the vulnerability or resilience to the development of alcohol use disorder (AUD). Going forward, it will be important to include careful consideration of biological variables in experimental design, reporting, and interpretation. As new research uncovers conditions in which sex, age, and environment play major roles in physiological and/or pathological processes, our understanding of the complex interaction between relevant biological variables and critical signaling pathways like the CRF system in the cellular and behavioral consequences of alcohol exposure will continue to expand ultimately improving the ability of preclinical research to translate to the clinic.

1. Introduction

Corticotropin-releasing factor (CRF) is a 41 amino acid neuropeptide first characterized in 1981 (Vale et al. 1981). In the nearly forty years since its discovery, CRF has risen to a prominent role in the function of neural circuitry and behavior, particularly in the context of stress and anxiety (Ohmura and Yoshioka, 2009). CRF is a critical mediator of the endocrine stress response and hypothalamic-pituitary-adrenal (HPA) axis activation following stress exposure. CRF release from the paraventricular nucleus of the hypothalamus (PVN) stimulates the release of adrenocorticotropic hormone (ACTH) from the anterior pituitary, which in turn prompts the release of glucocorticoids (including cortisol/corticosterone) from the adrenal glands (Rivier et al., 1982). CRF exerts its effects on physiology and behavior via actions at two receptors, CRF1 and CRF2 (Chen et al., 1993; Liaw et al., 1996), which are both expressed throughout the limbic system and cortex. In particular, CRF1 and CRF2 are expressed in the amygdala, PVN, cortex, and bed nucleus of the stria terminalis (BNST), regions that have been implicated in stress responsivity and anxiety disorders (Koob, 2013). These features have also made CRF an important player in the development of alcohol use disorder (AUD).

There is a strong, established relationship between alcohol and anxiety. Rates of AUD are higher in patients who have comorbid anxiety disorders (Grant et al., 2015). Patients with anxiety disorders appear to progress from problematic alcohol use to AUD more rapidly (Kushner et al., 2011), and patients with both AUD and an anxiety disorder have higher relapse rates than those with AUD alone (Kushner et al., 2005). Work with both human participants and rodent models suggests that the transition from alcohol use to alcohol dependence and AUD is characterized by increases in anxiety-like behavior, and that alcohol use in dependent subjects is motivated in part by a desire to alleviate this anxiety or negative affect (Koob, 2013). Additionally, in both human participants and preclinical animal models stress exposure facilitates the reinstatement of drug seeking and consumption in previously abstinent subjects (Mantsch et al., 2016). These findings point to an important role for brain stress circuitry in mediating the maladaptive effects of alcohol exposure. Alcohol activates the HPA axis in both human AUD patients and rodent models of alcohol consumption (Wand and Dobs, 1991; Waltman et al., 1993; Inder et al., 1995; Gianoulakis et al., 2003). Additionally, both acute (Rivier and Lee, 1996) and chronic (Rivier et al., 1990) alcohol exposure activate CRF neurons in the paraventricular nucleus (PVN) of adult rats in a CRF-dependent manner (Rivier and Lee, 1996; Rivier, 1999). A wealth of preclinical evidence has implicated the CRF/CRF1 signaling system in the development of alcohol dependence, and CRF1 antagonists have repeatedly been found to reduce alcohol self-administration in rodents (see Agoglia et al., 2018 for review.) However, recent clinical trials with CRF1 antagonists for the treatment of AUD in humans have had disappointing outcomes (Kwako et al., 2015; Schwandt et al., 2016). In evaluating the failure of such a large animal literature to translate to the human condition, the role of biological variables such as sex have been suggested as a potential explanation (Spierling and Zorrilla, 2017).

As of 2015, the National Institutes of Health (NIH) have required the consideration of sex as a biological variable in preclinical research for both cell lines and animal models (NIH, 2015). Reactions to this directive have been mixed (Richardson et al., 2015; Shansky and Woolley, 2016), and a recent survey of NIH study section members has shown that some scientists remain skeptical as to the feasibility and value of including biological variables such as sex in preclinical research (Woitowich and Woodruff, 2019). The disparity in research between male and female subjects is substantial, and the impact of this male subject bias negatively impacts the health of human women and preclinical science (Miller, 2001; Fisher and Ronald, 2010; Beery and Zucker, 2011; Yoon et al., 2014). Although the vast majority of the preclinical investigation of CRF1 antagonists as potential treatments for AUD was conducted in male rodents, the clinical trials evaluating these drugs in human participants have been conducted in both males and females (Kwako et al., 2015) or exclusively in females (Schwandt et al., 2016). Perhaps the difficulty in translating the extensive preclinical research on CRF to clinically-approved AUD treatments comes, in part, from this reduced focus on biological variability. Some have argued that sex is being given select prominence over other relevant biological variables such as age or environment (Woitowich and Woodruff, 2019). Although some of this advocacy for other biological variables may not have been made in earnest, age and environment are important determinants of neurobiology and behavior. This review will highlight important developments in the role of sex, age (especially adolescence), and housing conditions in the activity of the CRF system, and how these disparate influences may have particular relevance for the role of CRF in the development of alcohol dependence and AUD going forward.

2. Sex differences

2.1. Sex differences in alcohol use disorder and alcohol-related behaviors

Alcohol use and abuse affect both sexes, but historically AUD has been more prevalent in men than women (Grant et al., 2015). Recent data indicate that this trend may be reversing, as drinking and AUD in women are increasing at greater rates compared to men (Grant et al., 2017). Between 2002 and 2013 alcohol use and AUD prevalence increased generally within the United States, but women exhibited a 57.9% increase in problem drinking and an 83.7% increase in AUD as compared with 15.5% and 34.7%, respectively, in men. Over a similar time period, female binge drinking has also been on the rise in the US whereas male binge drinking has remained more or less stable (Grucza et al., 2018). These changes are troubling in light of research that indicates more severe consequences of alcohol drinking in women versus men. Women are more sensitive to several pharmacological effects and consequences of acute alcohol consumption, including sedation and intoxication, impaired driving ability, and propensity for injury (Miller et al., 2009; Arnedt et al., 2011; Zeisser et al., 2013). Women also appear to be at higher risk for negative consequences of chronic alcohol consumption, including negative affect, liver diseases, cancers, heart disease, brain damage, death, and more rapid transition from alcohol use to AUD (Bekman et al., 2013; Agabio et al., 2016, 2017).

Several lines of evidence suggest that women may be particularly vulnerable to stress-related drinking. When asked about their motivation for beginning to use alcohol, adolescent boys are more likely to report drinking in order to experience the enhancing effects of alcohol, whereas adolescent girls are more likely to report drinking in order to cope with stress (Kuntsche and Muller, 2012). Patients with anxiety disorders such as post-traumatic stress disorder (PTSD) are more likely to develop AUD than the general population, but this relationship is stronger in women than in men (Conway et al., 2006). Alcohol use is more motivated by coping in women than men (Lehavot et al., 2014), and PTSD more commonly precedes substance abuse in women than men (Sonne et al., 2003). Additionally, in male patients, substance use disorders are often a precursor to subsequent mood disorders, whereas in female patients mood disorders usually manifest prior to substance use disorders (Zilberman et al., 2003). In women but not men, a history of stressful events during the previous year was positively correlated with risk for AUD (Verplaetse et al., 2018). These differences in stress-motivated drinking have been hypothesized by some researchers to explain the faster pace at which women transition from alcohol use to alcohol abuse (Becker et al., 2012).

Preclinical alcohol research has also demonstrated sex differences in alcohol intake, neurobiological effects of alcohol exposure, and the development of alcohol dependence (All studies in section 2 were conducted in adult rodents unless otherwise specified.). Whereas in primates (human and non-human) males typically consume more alcohol than females, female rodents typically consume more alcohol than males when adjustments are made for body weight (see Becker and Koob, 2016 for review.) Female rats have also been shown to be more sensitive to the rewarding effects of alcohol (Torres et al., 2014) and less sensitive to the deleterious effects of alcohol withdrawal (Devaud and Chadda, 2001; Varlinskaya and Spear, 2004; Veatch et al., 2007), a pattern which may underlie their increased propensity to self-administer. Sex differences in cue or stress-induced reinstatement of alcohol seeking/responding are mixed, with some studies reporting greater sensitivity in females (Roger-Sanchez et al., 2012; Bertholomey et al., 2016), whereas others report greater sensitivity in males (Randall et al., 2017). Evidence from mice also suggests that females are more sensitive to stress-induced increases in voluntary alcohol consumption (Cozzoli et al., 2014), which recapitulates similar evidence from human women discussed above. Efforts to identify a biological basis for these sex differences have been inconclusive; evidence for hormonal regulation of alcohol consumption has been particularly variable (Erol et al., 2019). It seems likely that a diversity of neurotransmitter and hormonal signaling systems may play a role in sex differences in alcohol-related behaviors, with the CRF system representing one such potential regulator.

2.2. Sex differences in CRF

An extensive literature has characterized sex differences in the function of the HPA axis both under normative conditions and in response to acute and chronic stressors (see Panagiotakopoulos and Neigh, 2014 for review). Here we focus on some features of the CRF system that appear to be modulated by sex in humans and preclinical animal models. Anxiety disorders are more common in women than men, especially disorders such as PTSD that are characterized by hyperactive central CRF levels in humans (Kasckow et al., 2001). CRF administration to healthy adult participants stimulates a greater HPA axis response in adult women versus men (Gallucci et al., 1993; Born et al., 1995). Several important insights about sex differences in CRF signaling have come from rodent models. In rats, the expression of Crf mRNA is higher in the female PVN and may be less sensitive to stress than in the male PVN (Duncko et al., 2001; Viau et al., 2005; Iwasaki-Sekino et al., 2009; Babb et al., 2013; Janitzky et al., 2014; Lenglos et al., 2015). Crf mRNA is also greater in the female versus male amygdala (Janitzky et al., 2014), as is CRF1 binding (Weathington et al., 2014). The female rat BNST also has more CRF + neurons than the males (Funabashi et al., 2004). Crhr1 mRNA is greater in the female rat dorsal raphe (Lukkes et al., 2016) and within specific cell types in the rat hippocampus (Hiroshige et al., 1973; Williams et al., 2011). Crf gene expression is subject to regulation by estrogen (Vamvakopoulos and Chrousos, 1993; Lalmansingh and Uht, 2008), suggesting the possibility of hormonal mechanisms that regulate sex differences in Crf transcription and expression. The effects of stress on Crf gene expression are also dependent upon sex, but these differences appear to be sensitive to the type and timing of stressors. Prenatal stress increases Crf mRNA in the female but not male PVN of adult rats (Zohar and Weinstock, 2011), but acute restraint stress in adult rats increases PVN Crf mRNA to a greater extent in males versus females (Sterrenburg et al., 2012). Others have reported greater increases in the number of PVN neurons that contain Crf mRNA following acute restraint stress in females versus males (Babb et al., 2013). In the amygdala, females are insensitive to early life stress-induced changes in central amygdala (CeA) CRF (Prusator and Greenwood-Van Meerveld, 2017) and CRF1 gene expression (Brunton et al., 2011). However, acute foot shock in adult rats induced greater increases in CRF expression in females versus males (Iwasaki-Sekino et al., 2009). Chronic mild variable stress (CMVS) also induces differential epigenetic regulation of the CRF gene in male and female rats, which was associated with increased activation of the CeA in males but not females (Sterrenburg et al., 2011). These diverse findings together suggest two conclusions: first, that Crf gene expression is often higher in female rodents within brain regions that are relevant to anxiety and addiction; second, that females and males appear to differ in the immediate versus lasting effects of stress on Crf gene expression.

There are also sexual dimorphisms in CRF regulation of stress-related behaviors. Central administration of CRF provokes anxiety-like behavior in mice (Sherman and Kalin, 1987; Gargiulo and Donoso, 1996), and females appear to be more sensitive to this effect than males (Wiersielis et al., 2016). A line of mice bred to overexpress CRF demonstrated increased anxiety-like behavior versus wild type for female but not male subjects (Million et al., 2007), and forebrain-specific CRF overexpression results in increased anxiety-like behavior and impaired avoidance behavior that was more pronounced in females (Toth et al., 2014, 2016). Also in forebrain, oxytocin neurons dampen the anxiogenic effects of CRF in males but not females via the release of CRF binding protein, a finding that was related to higher levels of CRF in females (Li et al., 2016). Within the locus coeruleus (LC), a norepinephrinergic nucleus that is critical in the stress response, neurons from female mice are more sensitive to low dose CRF and exhibit blunted adaptation to stress as compared with neurons from male mice (Curtis et al., 2006). Within the dorsal raphe, CRF1 antagonists blunt anxiety responses to stressors in male but not female mice (Howerton et al., 2014). A wealth of data from Bangasser and colleagues has pointed to significant sex biases in CRF1 signaling (Bangasser et al., 2010, 2013). CRF1 appears to be more biased towards G_s signaling in females and towards β-arrestin in males, and females lack the robust CRF1 internalization response following stress exposure that is evident in males. These differences may be related to increased stress sensitivity and anxiety disorders in females (Valentino et al., 2013). Together, these results indicate a generally increased sensitivity of female rodents to the effects of CRF on anxiety-like behavior as well as a blunted ability of the CRF system to adapt to stress exposure. These sex differences may be related to disparities in mood disorders between men and women.

2.3. Sex differences in the role of CRF in alcohol use and abuse

As is the case for much of neuroscience research, females are understudied compared to males in addiction biology and information regarding the role of sex in CRF-alcohol interactions is lacking. However, the results of several studies strongly suggest that this is an important area for future research. In preclinical rodent models, sex differences in the effects of alcohol on the CRF system are evident. Alcohol-stimulated increases in corticosterone have been shown to be more dramatic in female versus male rats (Ogilvie and Rivier, 1997; Pronko et al., 2010), suggesting that alcohol may have greater potential to activate the CRF system in female rodents. Chronic exposure to an alcohol liquid diet produced divergent effects on CRF1 receptor localization in the LC of male and female rats, with alcohol driving increased cytoplasmic CRF1 in males and increased plasma membrane CRF1 (and increased activation of LC neurons) in females (Retson et al., 2015a). These effects were not altered by estrous phase in female rats. In rats with a history of chronic alcohol drinking, an acute injection of the physiological stressor lipopolysaccharide (LPS) induced a blunted corticosterone response as compared with control rats, and this effect was more pronounced in females than males (Silva and Madeira, 2012). In the same study, females with a history of alcohol drinking also failed to demonstrate an increase in Crf mRNA in the PVN following LPS injection that was exhibited by control females, a neuroadaptation that was absent in males. Female rats in these experiments were ovarectomized prior to LPS injection, with half the group receiving estradiol injections prior to LPS and the other receiving a vehicle injection. No differences between the two female groups emerged, suggesting that ovarian hormone status did not influence the ability of LPS to provoke a CRF mRNA response in females. Female but not male mice deficient in the HPA regulator β-endorphin consumed less alcohol as compared with WT, and the elevated Crf mRNA within the extended amygdala of β-endorphin deficient females was normalized by alcohol drinking (Nentwig et al., 2019). Alcohol drinking also activated CeA CRF neurons and enhanced the sensitivity of these neurons to stress in male rats, effects that were absent in neurons from female rats (Retson et al., 2015b). Together, these findings indicate that the CRF systems are reactive to alcohol in both male and female rodents, but there may be sex differences in the sensitivity of CRF to alcohol across different brain regions. PVN CRF may exhibit more adaptations to alcohol in females whereas CeA CRF may be more sensitive to alcohol-induced plasticity in males. Further investigation of these and other brain regions involved in the pathogenesis of AUD is needed to clarify these possible sex differences and determine their functional consequences. Brain regions exhibiting sexual dimorphism in the CRF systems are depicted in Fig. 1, with emphasis on the three regions (PVN, CeA and LC) that also show divergent responses of the CRF systems to alcohol between males and females.

Fig. 1. Sexually dimorphic components of the CRF systems.

A) Sagittal section of rodent brain depicting selected nuclei that have been shown to display sex differences in CRF expression or function. Shading indicates regions that have also been shown to display sex-dependent effects of alcohol on the CRF systems. Inserts illustrate the specific sex differences observed in each brain region. B) In female but not male rats, chronic alcohol drinking blunts the ability of LPS to increase Crf mRNA in the PVN. C) In the male but not female CeA, chronic alcohol consumption increased activation of CRF neurons. D) In the male LC, chronic alcohol consumption increases cytoplasmic expression of CRF1R, whereas in the female LC chronic alcohol consumption increases membrane-bound CRF1R. HIPP = hippocampus, PVN = paraventricular nucleus, BNST = bed nucleus of the stria terminalis, CeA = central amygdala, DR = dorsal raphe, LC = locus coeruleus, LPS = lipopolysaccharide, CRF = corticotropin releasing factor, CRF1R = CRF1 receptor.

Evidence from manipulations of drug self-administration also points to sex differences in CRF regulation of addiction-related behaviors. CRF overexpression in the nucleus accumbens (NAc) increased nicotine self-administration in female rats to a larger extent than male rats (Uribe et al., 2019), and this effect was absent in ovariectomized female rats. Female rats were also more sensitive than males to CRF-induced reinstatement of cocaine seeking (Buffalari et al., 2012). Several studies with CRF or CRF1 knockout mice have failed to find evidence for sex differences in reductions in alcohol consumption or sensitization as compared with WT mice (Pastor et al., 2011, 2012; Kaur et al., 2012; Giardino and Ryabinin, 2013). However, these manipulations affected all brain regions equally and thus may have obscured sex differences that would have emerged had the manipulation occurred specifically within a sexually dimorphic region such as the PVN or CeA. Together, a picture of the female CRF system is emerging that indicates hyper-activity in response to alcohol when compared with the male CRF system and a lack of adaptations to stress/alcohol interactions that are evident in males. Importantly, of the two clinical trials conducted for CRF1 antagonists in the treatment of AUD, one was conducted exclusively in female participants and failed to find any benefit of CRF1 antagonism on alcohol craving and relapse (Schwandt et al., 2016). It is possible that the disproportionate focus on male animals in preclinical research failed to appreciate sex differences in CRF regulation of alcohol drinking and dependence (Peltier et al., 2019). These findings are summarized in Table 1.

Table 1.

Sex in the effects of CRF and alcohol.

Reference	Species	Sex	Age	Housing	Exposure	Region	Results
Ogilvie and Rivier (1997)	Rat	F + M	Adult	Isolation	EtOH Inj. 3 g/kg i.p.	–	EtOH ↑ CORT F > M
Pronko et al. (2010)	Mouse	F + M	PND 70–100	Isolation	EtOH inj. 2.5 g/kg i.p.	–	EtOH ↑ CORT F > M
Retson et al. (2015a)	Rat	F + M	Adult	Isolation	EtOH liquid diet 12–14 g/kg/day	LC	EtOH ↑ cytoplasmic CRF1 M EtOH ↑ membrane CRF1 F
Silva and Madeira (2012)	Rat	F + M	PND 60–240	Group	EtOH liquid diet F 11 g/kg/day M 8 g/kg/day	PVN	EtOH blunts LPS CORT ↑ F > M EtOH blunts LPS Crf mRNA ↑ F only
Nentwig et al. (2019)	Mouse	F + M	Adult	Isolation	Binge EtOH drinking	CeA	EtOH normalizes ↑ Crf mRNA from transgene F only
Retson et al., 2015b	Rat	F + M	Adult	Isolation	EtOH liquid diet 12–14 g/kg/day	CeA	EtOH ↑ activity/stress sensitivity of CRF neurons M only
Uribe et al. (2019)	Rat	F + M	PND 45	Pair	Operant SA of Nic	NAc	CRF overexpression ↑ nicotine SA F > M
Buffalari et al. (2012)	Rat	F + M	Not specified	Isolation	Operant SA of Coc	–	CRF induces reinstatement of cocaine seeking F > M
Pastor et al. (2012)	Mouse	F + M	PND 50–115	Group	EtOH inj. 2.5 g/kg	–	CRF knockout blocks EtOH sensitization No sex differences
Kaur et al. (2012)	Mouse	F + M	PND 49–91	Isolation	Binge EtOH drinking	–	CRF/CRF1 knockout ↓ EtOH drinking No sex differences
Pastor et al. (2011)	Mouse	F + M	PND 68–127	Group	Two-bottle choice EtOH drinking	–	CRF1 knockout ↓EtOH drinking No sex differences
Giardino and Ryabinin (2013)	Mouse	F + M	PND 56–98	Isolation	Binge EtOH drinking	–	CRF1 knockout ↓ EtOH drinking No sex differences

F = female. M = male. - = condition not tested. PND = post-natal day. RA = restricted access. SA = self-administration. Coc = cocaine. Nic = nicotine. Adol = adolescence.

3. Age

Along with sex, developmental stage is a major determinant of the effects of alcohol on the brain and behavior. Exposure to alcohol during prenatal development can cause Fetal Alcohol Spectrum Disorder (FASD), a constellation of physical and behavioral impairments including increased lifetime risk for developing AUD (Streissguth et al., 2004). Risky drinking in older adults (age 55+) has been linked to more severe health consequences of alcohol use (Oslin, 2004). One of the periods of highest vulnerability to the enduring effects of alcohol use is adolescence, a distinct developmental period that occurs as an organism matures from a juvenile into an adult. This section will focus on the interaction of age and alcohol on the CRF system in adolescents, since CRF has not been as extensively investigated in the context of prenatal alcohol exposure or drinking in older adults. A role for CRF in the development of FASD and the consequences of alcohol in older adults is an important avenue for future work.

3.1. Adolescent development, alcohol, and stress

In humans, adolescence spans the onset of puberty through the complete development of the brain, roughly the early teens to the mid-twenties (Crone and Dahl, 2012). Adolescence is conserved across mammalian species (Stevens and Vaccarino, 2015), which allows the use of animal models to assess the effects of alcohol on the developing brain. In rodents, adolescence ranges from the early post-pubertal period through physiological maturity, conservatively estimated to occur between postnatal days (PND) 28–42 (Schneider, 2013). However, adolescent-typical behaviors such as novelty seeking and reward sensitivity are still evident through PND 43–60 in males and females (Vetter-O’Hagen et al., 2009; Vetter-O’Hagen and Spear, 2012). Some researchers therefore describe this period as late adolescence or emerging adulthood, roughly analogous to the 18–25 year developmental epoch in humans (Varlinskaya et al., 2013; Spear, 2015).

Adolescence has emerged as a critical developmental period in the study of alcohol and AUD because alcohol exposure during this time appears to have uniquely harmful consequences. The majority of adults with AUD first begin drinking during adolescence (Hingson et al., 2006) and adolescent exposure to alcohol is associated with much higher lifetime rates of AUD than alcohol exposure during adulthood (Dawson et al., 2008). Across mammalian species, male and female adolescents consume greater amounts of drugs of abuse, including alcohol, as compared with adults when adjustments are made for body weight (Spear, 2000; Schwandt et al., 2010). Adolescent exposure to alcohol also increases subsequent alcohol consumption in male and female adult rodents (Maldonado-Devincci et al., 2010). Adolescent males and females have been shown to be less sensitive to the aversive properties of alcohol in a variety of measures, including sedative/hypnotic effects of high-dose alcohol (Silveri and Spear, 2000), locomotor impairments after intoxicating doses of alcohol (Ramirez and Spear, 2010), conditioned taste aversion (Vetter-O’Hagen et al., 2009; Holstein et al., 2011; Anderson et al., 2013), social impairment (Varlinskaya and Spear, 2007), and anxiety behavior related to acute alcohol withdrawal (Doremus et al., 2003). In contrast, male and female adolescent rats are more sensitive to alcohol facilitation of social behavior and the reinforcing properties of alcohol (Varlinskaya and Spear, 2006; Pautassi et al., 2008). Additionally, exposure to alcohol during adolescence, but not adulthood, produces memory impairments in adult male rats (White et al., 2000). This pattern of blunted sensitivity to the negative consequences of alcohol and enhanced sensitivity to the rewarding properties of alcohol has been suggested to underlie age differences in alcohol consumption (Doremus-Fitzwater and Spear, 2016), but the neurobiological mechanisms that underlie these behavioral differences are not yet clear. Developmental alterations in stress signaling, particularly CRF activity, may represent one such mechanism.

The maturation of stress circuitry also occurs during childhood and adolescence, leading to some unique features of the stress response in adolescent versus adult rodents. Hormonal responses to stressors develops during adolescence, with levels of corticosterone reaching adult levels between PND 30–40 and ACTH between PND 50–60 in male rats (Foilb et al., 2011). Crf mRNA is already present during embryonic development in mice (Keegan et al., 1994) and rats (Bugnon et al., 1982). In both male and female rodents, genes for both the CRF peptide and CRF1 receptor undergo developmental changes in expression from birth through the juvenile period in several affective brain regions including the hippocampus, hypothalamus, and cortex (Vazquez et al., 2006; Korosi and Baram, 2008; Lukkes et al., 2016). Notably, Crf mRNA in the amygdala appears to take longer to reach adult levels than in the cortex or hippocampus in male but not female rats (Avishai-Eliner et al., 1996; Viau et al., 2005), which may indicate that the amygdala is an important brain region for the developmental regulation of CRF activity in males. Multiple lines of evidence also suggest that adolescence is generally characterized by stress hyper-reactivity (McCormick and Mathews, 2010; Klein and Romeo, 2013), and the effects of stress exposure during adolescence can last longer and impact behavior more significantly than adult stress exposure (Jankord et al., 2011). Some of this increased sensitivity to stress may be mediated by CRF signaling (Bingham et al., 2011). Together, these findings indicate that by the time rodents reach adolescence, the building blocks of the CRF circuitry are in place within brain regions that are relevant to AUD and a state of heightened sensitivity to the effects of stress has developed.

3.2. Adolescent alcohol exposure and CRF

Population studies of human AUD patients and adolescent drinkers with high risk for AUD point to a role for the CRF system in regulating adolescent drinking patterns. CRHR1, the human gene for the CRF1 receptor, has been associated with vulnerability to alcoholism in male and female adults (Treutlein et al., 2006). This gene also appears to regulate the P3 peak of event-related potential, a feature of electroencephalography recordings that queries cognitive impairment and a characteristic of several psychiatric illnesses including AUD and neurodevelopmental disorders such as ADHD (Enoch et al., 2008; Chen et al., 2010). The Mannheim Study of Children at Risk has yielded insights of particular relevance to the relationship between genetic variation at CRHR1 and adolescent alcohol use. Single nucleotide polymorphisms (SNPs) in the CRHR1 gene were associated with increased rates of binge drinking, lifetime alcohol intake, and lifetime episodes of drunkenness in male and female adolescents (Treutlein et al., 2006). CRHR1 SNPs were also associated with greater amounts of maximum alcohol intake per occasion and higher lifetime rates of heavy drinking following stressful events in a separate sample of male and female adolescents (Blomeyer et al., 2008). A third study found associations between age at first drink (which was associated with higher alcohol consumption) and CRHR1 SNPs in male and female participants with a high number of stressful life events prior to drinking onset (Schmid et al., 2010). Most recently, evidence from the Michigan Longitudinal Study of families with parents experiencing AUD has indicated a potential functional role for SNPs at CRHR1 in the associated risk for adolescent alcohol use in males and females. Functional MRI indicated that individuals with the protective G allele, which was associated with a reduced propensity to binge drink and fewer negative consequences of alcohol drinking, displayed greater ventrolateral prefrontal cortex (VLPFC) activation during negative emotional word processing and reported less negative emotionality (Glaser et al., 2014). These associative studies in human adolescents strongly suggest a relationship between the CRF system and the effects of adolescent alcohol exposure, and work in rodent models has allowed the assessment of a causal link between alcohol and the CRF system in the developing brain.

Among the many effects of alcohol on the developing rodent brain, the CRF system appears to be particularly sensitive to adolescent alcohol exposure in a brain region-dependent manner. In the PVN, adolescent binge alcohol consumption increases Crf mRNA in adult male animals under baseline conditions (Przybycien-Szymanska et al., 2011). However, adolescent alcohol exposure reduces Crf mRNA when subjects are challenged with acute alcohol during early adulthood (PND 61–62) in males (Allen et al., 2011a) but increases Crf mRNA when challenged in later adulthood (PND 70–71) in males but not females (Logrip et al., 2013). Adolescent intermittent ethanol exposure increases Crf mRNA in the prefrontal cortex (PFC) but reduces Crf mRNA in the NAc in males (Boutros et al., 2018). The effects of alcohol drinking on amygdala CRF have been mixed, reflecting the diversity of amygdalar subnuclei, exposure procedures, and developmental time points that have been investigated. Adolescent binge operant alcohol self-administration in males and females (Allen et al., 2011b; Gilpin et al., 2012; Karanikas et al., 2013) and adolescent exposure to forced alcohol consumption via diet in males (Wills et al., 2010) reduce CRF immunoreactive (IR) positive cells in the CeA of adolescent rats. Additionally, Karanikas et al. (2013) found that the effects of alcohol exposure were not dependent upon estrous phase in female rats. In contrast, ethanol gavage does not appear to alter Crf mRNA in the male CeA (Boutros et al., 2018). In adult male rats exposed to chronic ethanol vapor during adolescence, Crf mRNA in the CeA was increased (Boutros et al., 2016), indicating that the effects of adolescent alcohol on CRF do persist into adulthood. In the basolateral amygdala (BLA), male adolescents are less sensitive than adults to ethanol-induced reductions in Crf mRNA (Falco et al., 2009), highlighting the importance of differentiating amygdalar nuclei in the effects of alcohol. Together, these findings (summarized in Table 2) indicate that the adolescent CRF system is particularly responsive to the effects of alcohol.

Table 2.

Age in the effects of CRF and alcohol.

Reference	Species	Sex	Age	Housing	Exposure	Region	Results
Treutlein et al. (2006)	Human	F + M	15 years	-	-	-	CRHR1 SNP associated w/binge drinking + drunkenness
Blomeyer et al., 2008^	Human	F + M	15 years	-	-	-	CRHR1 SNP associated w/drinking
Schmid et al. (2010)	Human	F + M	19 years	-	-	-	CRHR1 SNP associated w/age at first drink
Glaser et al. (2014)	Human	F + M	19 years	-	-	-	CRHR1 SNP assoc. w/binge drinking and neg. emotionality
Przybycien-Szymanska et al. (2011)	Rat	M	PND 37–68	Pair	EtOH inj. 3.0 g/kg i.p.	PVN	Adol. EtOH ↑ Crf mRNA in adulthood
Logrip et al. (2013)	Rat	F + M	PND 28–42	Adol. group, Adult isolation	Intermittent EtOH vapor	PVN	Adol. EtOH ↑ Crf mRNA to EtOH challenge in adulthood M only
Allen et al. (2011a)	Rat	M	PND28–42, 61–62*	Group	Daily EtOH vapor/EtOH gavage 4.5 g/kg	PVN	Adol EtOH ↓ Crf mRNA
Boutros et al. (2018)	Rat	M	PND 28–53	Pair	EtOH gavage 5.0 g/kg	PFC NAc CeA	Adol. EtOH ↑ Crf Mrna Adol. EtOH ↓ Crf mRNA No effects of adol. EtOH
Allen et al. (2011b)	Rat	M	PND 25–53	Not specified	Operant SA of EtOH + supersac	CeA	Adol. EtOH ↓ CRF-IR cells
Gilpin et al. (2012)	Rat	M	PND 28–42, 196*	Isolation	Binge-like operant SA of EtOH	CeA	Adol. EtOH ↓ CRF-IR cells in adulthood
Karanikas et al. (2013)	Rat	F + M	PND 28–42	Group	Binge-like operant SA of EtOH	CeA	Adol. EtOH ↓ CRF-IR cells
Wills et al. (2010)	Rat	M	PND 26–45	Isolation	Intermittent 2.5% EtOH liquid diet	CeA	Adol. EtOH ↓ CRF-IR cells. Blocked by repeated stress
Boutros et al. (2016)	Rat	M	PND 28–42, 182*	Pair	Intermittent EtOH vapor	CeA	EtOH ↑ Crf mRNA
Falco et al. (2009)	Rat	M	PND 28–46, 106* PND 80–98t, 158	Isolation	RA EtOH drinking adol. 2.5, adult 2.3 g/kg/day	BLA	EtOH ↓ Crf mRNA Adults only
Burke et al. (2016)	Rat	M	PND 35–44, 65–77*	Pair	Operant SA of cocaine	VTA	CRF1 antagonist blocks social defeat-induced ↑ in coc. SA
Wills et al. (2009)	Rat	M	PND 26–45	Isolation	Intermittent 2.5% EtOH liquid diet	-	CRF1 antagonist blocks EtOH-induced ↓ in social interaction
Brielmaier et al. (2012)	Rat	M	PND 27–31	Group	Nicotine inj. 0.4 mg/kg s.c.	-	CRF1 antagonist blocks shock-induced ↑ in nic. CPP

^*age at which the same cohort was tested during adulthood

^tage of separate cohort of adults.

^{^}also referenced in Table 3.

- = condition not tested. F = female. M = male. PND = post-natal day. RA = restricted access. SA = self-administration. Inj = injection. Coc = cocaine. Nic = nicotine. Adol = adolescence. W/ = with. Neg = negative.

3.3. Potential functional roles for CRF in the consequences of adolescent alcohol exposure

Although the evidence for sensitivity of the CRF system to the effects of alcohol during adolescence is strong, a functional role for CRF signaling in the actions of alcohol on the adolescent brain or alcohol-related behaviors has received less investigation. There are several potential biobehavioral mechanisms by which the activity of CRF/CRF1 signaling may regulate adolescent sensitivity to alcohol and alcohol consumption. Evidence suggests a potential role for CRF in the mediation of drug aversion and reward, the interaction between stress and alcohol consumption, and the regulation of neural pruning and synaptic plasticity. These three functions of CRF may represent mechanisms by which age differences in CRF activity may contribute to age differences in alcohol intake and vulnerability to the enduring consequences of adolescent alcohol exposure (Fig. 2).

Fig. 2. Potential functional roles of CRF in the effects of adolescent alcohol exposure.

Schematic of the rodent and human adolescent brain under the influence of alcohol. Alcohol drinking activates CRF signaling in the adolescent brain, which may influence drug reward/aversion, stress signaling, and synaptic plasticity in the developing brain. CRF activity in these three domains may drive increased consumption of alcohol by adolescents.

CRF1 antagonists have been repeatedly shown to reduce alcohol consumption in rodent models using adult subjects (Zorrilla et al., 2014), but to date, age differences in sensitivity to CRF1 inhibition have not been assessed in adolescent consumption models. Findings in other behavioral tests and with other drugs of abuse suggest the hypothesis that activity of CRF/CRF1 may specifically regulate adolescent alcohol consumption and contribute to age differences in alcohol drinking. Pretreatment with a CRF1 antagonist prevents the reduction in social interaction induced by withdrawal from chronic alcohol in adolescent male rats (Wills et al., 2009), suggesting a role for the CRF1 system in some of the adolescent-specific interactions between alcohol and social behavior that may contribute to adolescent motivation to drink. Further work to determine whether CRF1 mediates other aversive properties of alcohol withdrawal in adolescents, such as withdrawal-related anxiety, would help to clarify the contribution of CRF signaling to alcohol drinking that is motivated by a desire to avoid the negative consequences of withdrawal. CRF1 signaling may also play a role in drug reward and reinforcement in adolescents. In male adolescent rats, CRF1 activity regulated the ability of a pre-trial foot shock stressor to facilitate the development of conditioned place preference (CPP) to nicotine (Brielmaier et al., 2012). CRF1 inhibition specifically within the ventral tegmental area also blocked the ability of social defeat stress during adolescence to potentiate cocaine self-administration in adulthood in male rats (Burke et al., 2016). In male mice, the loss of environmental enrichment during adolescence enhanced CPP for cocaine and increased Crf mRNA in the BNST and the increase in cocaine CPP was blocked by the CRF1 antagonist antalarmin (Nader et al., 2012). Similar experiments in alcohol CPP or self-administration models would offer clarity on the role of CRF1 activity in adolescent alcohol drinking that may be motivated by enhanced sensitivity to alcohol reward.

Another potential neurobiological mechanism for the role of CRF in adolescent alcohol exposure is mediation of the effects of stress exposure on alcohol consumption. As both stress and alcohol appear to alter the expression of CRF during adolescence, this is a hypothesis worth investigating. The majority of studies in human adolescents have found that stress interacts with CRHR1 genotype to impact alcohol risk (Blomeyer et al., 2008; Schmid et al., 2010; Glaser et al., 2014). Glaser et al. (2014) proposed a mechanism for CRHR1 to modulate VL-PFC activation during negative emotional processing and predicted that the indirect effects of genotype on alcohol risk were mediated by childhood stress. Preclinical research has also pointed to a relationship between stress and CRF in the effects of alcohol on adolescent rodents. Adolescent male rodents have prolonged corticosterone responses to both acute and chronic stressors (Gomez et al., 2002; Romeo et al., 2006), and adolescent exposure to alcohol blunts tolerance to ethanol-induced ACTH increases in adult male rats (Lee and Rivier, 2003). Initiating alcohol use during adolescence enhances vulnerability to stress-facilitated increases in adult alcohol consumption in male and female rats (Siegmund et al., 2005; Fullgrabe et al., 2007). Stress during adolescence, but not adulthood, has also been shown to selectively increase alcohol CPP (Song et al., 2007) and self-administration (Chester et al., 2008) in both male and female mice. Adolescent intermittent alcohol exposure increased brain stimulation reward thresholds following social defeat stress in adult male rats, indicating that adolescent alcohol exposure had increased sensitivity to some of the negative affective results of stress (Boutros et al., 2018). This increased sensitivity may be related to the enduring increases in CeA CRF which the same research group reported in adult male rats following adolescent alcohol vapor exposure (Boutros et al., 2016), although a functional role for CRF in this behavior has not yet been established. These findings raise the strong possibility of CRF as a mediator of stress effects on adolescent alcohol exposure.

CRF may also influence the lasting consequences of adolescent alcohol exposure via regulation of synaptic plasticity and neural pruning. Adolescent brain development in humans, primates, and rodents is characterized by decreases in cortical gray matter and increases in white matter, reflecting increased myelination and removal of synaptic connections (Luo and O’Leary, 2005; Whitford et al., 2007; Hoftman and Lewis, 2011). Neural pruning is experience-dependent, with less active synapses being targeted for removal (Huttenlocher, 1984), and is thought to underlie the increase in functional connectivity and coordination between brain regions that occurs during adolescence (Stevens et al., 2009; Hwang et al., 2010). Work in adult animal models has indicated that alcohol exposure can alter dendritic spine morphology (Lescaudron et al., 1989; Tarelo-Acuna et al., 2000), a process that may be driven by ethanol engagement of synaptic plasticity (Carpenter-Hyland and Chandler, 2006). Alcohol exposure during adolescence reduces dendritic spine number and biases spine morphology towards immature structures in addiction-associated brain regions in male rodents, including the hippocampus (Risher et al., 2015; Mulholland et al., 2018), CeA (Pandey et al., 2015), BLA, and PFC (Jury et al., 2017). Excessive pruning of synapses during adolescence has been suggested as a major component of other psychiatric illnesses, such as schizophrenia (McGlashan and Hoffman, 2000). . Importantly, stress also alters dendritic spine morphology and neural pruning in male rodents (Cook and Wellman, 2004; Brown et al., 2005; Radley et al., 2006) and a role for CRF in the regulation of stress effects on spine dynamics is emerging. Early life administration of CRF reproduces the effects of stress on hippocampal spine regression in male mice (Brunson et al., 2001), and blockade of the CRF1 receptor via knockout or antagonists promotes dendritic branching, increases dendritic length, and produces more complex dendritic trees in mice (Chen et al., 2004). Further, CRF1 antagonism prevents the ability of early life stress to increase adolescent neural pruning in the hippocampus of male mice (Liu et al., 2016). Thus, CRF may play a role in the deleterious effects of stress on dendritic morphology and plasticity in the developing brain (Bennett, 2008). Together, these findings suggest the hypothesis that CRF may act as a molecular mechanism regulating the effects of adolescent alcohol on dendritic spine morphology and alcohol-induced synaptic plasticity. This hypothesis has not yet been tested, but it would be interesting to see whether CRF1 antagonism can prevent some of the alcohol-induced dendritic morphological changes in the adolescent brain in the same way that it protects against the effects of stress.

4. Environmental factors

4.1. Interaction of genetic variations in CRF and environmental factors

Sex and age are important biological variables that can influence drinking behavior and the effects of alcohol on the brain, but life experiences and environmental influences can also impact drinking and the development of AUD. Only a portion of individuals who drink alcohol suffer from alcohol use disorder (SAMHSA, 2017). It is widely understood that an individual’s risk for developing psychiatric disorders, including AUD, is a combination of genetic and environmental factors. As discussed in section 3.2, some of these genetic risk factors for AUD include variation in the CRF system. Importantly, there is evidence for an interaction between CRF and environmental influences, as individuals with more negative stressful life experiences and a SNP in CRHR1 drank more heavily and drank more alcohol per occasion (Blomeyer et al., 2008). Furthermore, male adult Marchigian-Sardinian (msP) rats, which have been selectively bred for high alcohol preference, also have genetic variants in the promoter region of Crhr1, which results in increased Crhr1 mRNA and CRF1 protein expression in many areas of the brain, including the amygdala and hippocampus (Hansson et al., 2006). Additionally, antagonizing the CRF1 receptor with antalarmin dose-dependently suppressed alcohol self-administration and blocked stress-induced reinstatement of alcohol seeking in adult male msP rats, but not Wistar rats (Hansson et al., 2006). Maternal separation of preweanlings, which is a particularly negative life experience for rodents, induces ethanol-mediated locomotion of male and female rat pups (Arias et al., 2010), and increases voluntary intake of ethanol in juvenile (PND 22–29) male rats (Odeon and Acosta, 2019). Maternal separation also increases Crf mRNA in the adult PVN (Francis et al., 2002) and increases Crf mRNA and protein in the adult hippocampus (Wang et al., 2014). Antagonizing the CRF1 receptor with CP154,526 reversed maternal separation-induced impairment in hippocampal long-term potentiation (LTP) and reduces deficits in a memory based task in adult rats (Wang et al., 2014). CP154,526 treatment also attenuated ethanol-induced locomotion in maternally separated and isolated male and female rat pups (Arias et al., 2010). Recent evidence from the lateral habenula (LHb) also indicates a role for CRF in the effects of severe negative stressors. Maternal deprivation for 24 h at PND9 enhanced LHb neuronal excitability and blunted the response of LHb neurons to acute CRF application in juvenile (PND 21–28) male rats (Authement et al., 2018). These data suggest that individual differences in genetic and environmental factors can increase the susceptibility to alcohol use disorder, particularly alterations in the CRF system.

These previous examples focused on negative life experiences, and indeed more negative life experiences are associated with increased susceptibility to psychiatric disorders. However, mildly stressful experiences may be inoculating against future stressors or increase resilience to psychiatric disorders. Examining why individuals who drink large amounts of alcohol do not develop alcohol dependence or AUD may result in a better understanding of protective mechanisms from AUD. Examining an animal model of this resilience to, or protection from, addiction may be an effective way of identifying protective factors.

4.2. Animal model of resiliency to psychiatric disorders: Environmental enrichment

An animal model of resilience to psychiatric disorders via inoculation stress is environmental enrichment (Crofton et al., 2015). Environmental enrichment involves maximizing novelty, social contact, and exercise by group housing conspecifics of the same sex in a large cage with plastic objects that are replaced and rearranged daily (specific protocols vary, see Crofton et al., 2015 for discussion). Without using pharmacological, surgical, or genetic manipulations, environmental enrichment results in many beneficial effects due to repeated mild stress inoculation (Crofton et al., 2015). Environmentally enriched adult rats show resiliency to depression-like and addiction-related behaviors including lower self-administration of amphetamine, cocaine, methylphenidate, and nicotine (Bardo et al., 2001; Green et al., 2002, 2010; Sáenz et al., 2006; Stairs et al., 2006; Brenes et al., 2008; Alvers et al., 2012; Venebra-Muñoz et al., 2014; Gomez et al., 2015). These effects are seen when animals are reared in the enriched environment for 30 days from weaning and also when animals are enriched after exposure to drugs or stress. Enrichment after drug exposure reduces cocaine locomotor activation, cocaine CPP, and cocaine seeking behaviors as well as self-administration of methamphetamine, heroin, and nicotine (Solinas et al., 2008; Thiel et al., 2010; Ranaldi et al., 2011; Chauvet et al., 2012; Sikora et al., 2018).

Enriched animals are best compared to isolated animals, which lack social contact, novelty, and exercise while pair-housed or group-housed animals without novelty (plastic objects) constitute an intermediate phenotype (Solinas et al., 2010; Crofton et al., 2015). Isolated animals have not been exposed to repeated mild stress or adapted to it like the enriched animals, and therefore do not cope well with stress or drug exposure. Although humans are not isolated per se, social isolation may influence alcohol drinking, and the isolated condition is a model for individuals that do not cope as well with stress, such as those exposed to more negative life experiences who drank more alcohol (Blomeyer et al., 2008). Research examining this animal model of resilience has already resulted in a better understanding of the molecular determinants of the resilience to cocaine addiction and depression in the nucleus accumbens (Green et al., 2010; Fan et al., 2013a, 2013b; Pavlovsky et al., 2013; Lichti et al., 2014; Zhang et al., 2014, 2016a, 2016b, 2019; Crofton et al., 2017; Scala et al., 2018). Environmental enrichment has been shown to globally alter the neuronal structure of the brain such as dendritic arborization in the cortex (Renner and Rosenzweig, 1987; Rosenzweig and Bennett, 1996). As discussed in section 3.3, alcohol during adolescence alters dendritic spine morphology possibly with a role for CRF (Chen et al., 2004). Therefore, examining environmental enrichment in the context of CRF and alcohol may also result in a better understanding of the molecular determinants of individual differences in susceptibility to alcohol dependence and AUD.

4.3. Environmental enrichment in adolescence produces a protective phenotype for ethanol

In comparison to previous results of the protective effects of environmental enrichment on self-administration of addictive drugs, some of the first studies of the effects of enrichment on alcohol drinking found that enriched male rats drank more alcohol compared to isolated male rats (Rockman et al., 1986, 1989). However, subsequent studies found that environment bi-directionally influences alcohol drinking; isolated animals are susceptible to alcohol drinking behaviors while enriched animals are resilient. Provided the choice of water, ethanol, or sweetened ethanol, adult male and female rats housed in an enriched environment for 25 days from 2 months of age consumed less sweetened ethanol as well as total ethanol compared to group housed or isolated rats (Kulkosky et al., 1980). In operant self-administration of ethanol, male rats isolated from weaning respond more and show a clear preference for the ethanol-associated lever compared to group housed or isolated rats (Deehan Jr et al., 2007). Male rats group housed from weaning also show lower operant self-administration of ethanol compared to isolated rats (McCool and Chappell, 2009). In a model of binge drinking, male mice reared in an enriched condition have the lowest preference for ethanol, followed by group housed, and isolated mice have a high preference for ethanol (Holgate et al., 2017). Switching environments for three days from isolated to an enriched condition reduces ethanol preference while isolating enriched male mice increases ethanol preference (Holgate et al., 2017). Male mice enriched from weaning drink less ethanol compared to group housed mice (Rodriguez-Ortega et al., 2018). Housing in isolation post weaning in male rats causes increased drinking compared to pair housed (Hall et al., 1998) or group housed rats (Schenk et al., 1990; McCool and Chappell, 2009) when provided a choice between water and ethanol. In a two-bottle choice paradigm, female rats enriched post weaning also drink less compared to group housed rats (de Carvalho et al., 2010) and female mice group housed from weaning also drink less ethanol compared to isolated (Lopez et al., 2011). Thus, environmental enrichment produces a resilient phenotype in both male and female mice and rats to ethanol regardless of ethanol exposure paradigm, while isolation housing produces a susceptible phenotype for ethanol.

Interestingly, Schenk et al. only found a significant difference in alcohol drinking when male rats were reared in isolation or in a group from weaning, and not when differentially housed as adults (1990). Lopez et al. also found that male and female mice reared in isolation from weaning drank significantly more ethanol in a two bottle choice paradigm compared to group housed mice but male mice isolated in adulthood did not show increased ethanol intake (2011). Additionally, female mice overall consumed more ethanol compared to males and female mice isolated in adulthood rather than immediately post weaning, showed a small but statistically significant decrease in ethanol intake (Lopez et al., 2011). Environmental enrichment in adult female mice during abstinence from ethanol drinking is able to reduce depressive-like behavior compared to isolated animals (Pang et al., 2013). However, male mice housed in an enriched environment in adulthood showed blunted ethanol consumption after restraint stress compared to group housed (Marianno et al., 2017). When paired housed adult male rats drink socially (physical divider during drinking days), they have initially higher ethanol intake and preference compared to isolated males and show an ethanol deprivation effect (Scott et al., 2020). Female adult rats do not show a difference in drinking in this paradigm and estrous phase does not affect alcohol intake or preference although female adult rats, both pair housed and isolated, show an ethanol deprivation effect (Scott et al., 2020). Therefore, males and females appear to respond similarly to environmental enrichment in adolescence, but differently in adulthood.

These results also suggest that the adolescent period may be particularly sensitive to alcohol exposure compared to other drugs of abuse, since environmental enrichment in adulthood can reduce drug taking and seeking behaviors for many other drugs of abuse, but not alcohol (Schenk et al., 1990; Solinas et al., 2008; Thiel et al., 2010; Lopez et al., 2011; Ranaldi et al., 2011; Chauvet et al., 2012; Sikora et al., 2018). Additionally, environmental enrichment during adolescence increases dendritic arborization (Renner and Rosenzweig, 1987) and increases survival of hippocampal progenitor cells, but fetal alcohol exposure completely blocks this increased survival of progenitor cells in male and female mice (Choi et al., 2005). Thus it is critical to report the specific age of rodents utilized in a study, not only when the rodents are acquired, but also during testing. Additionally, examining the molecular effects of enrichment in adulthood compared to adolescence and between males and females may be useful in better understanding sex and age differences in susceptibility and resilience to AUD.

4.4. Anxiety, environmental enrichment, and CRF

Anxiety and alcohol use disorder are often comorbid in humans (SAMHSA, 2017), but the effects of environmental enrichment on anxiety have been somewhat mixed. Enriched animals spend more time in the center in the open field test, spend more time on the open arms in the elevated plus maze (EPM), and display lower amounts of defensive burying (Roy et al., 2001; Friske and Gammie, 2005; Leal-Galicia et al., 2007; Sztainberg et al., 2010; Urakawa et al., 2013). Isolated male rats showed increased anxiety-like behavior in the EPM, but not in the light-dark box test (McCool and Chappell, 2009). However, in other tests not related to locomotor activity in a novel environment, such as the sucrose neophobia test, cold-stress induced defecation, and latency to ejaculation, enriched rats show increased anxiety-like behaviors (Green et al., 2010; Urakawa et al., 2014; Crofton et al., 2015). Regardless of this discrepancy, there is some evidence that in mice, environmental enrichment is anxiolytic and this effect is driven by altered CRF in the amygdala (Sztainberg et al., 2010). Female mice reared in an enriched environment showed less anxiety-like behaviors in the light-dark test, EPM, and open field test (Sztainberg et al., 2010). Enriched female mice had significantly lower expression of Crhr1 in the BLA after 4 and 10 weeks of differential housing and significantly lower expression of Crhr1 in the BNST after 10 weeks of differential housing (Sztainberg et al., 2010). In a separate group of male mice, specific knockdown of CRF1 in the BLA resulted in reduced anxiety in the light-dark test, EPM, and open field test (Sztainberg et al., 2010). These data suggest a role for amygdalar CRF in the anxiolytic effect of environmental enrichment. Sztainberg et al. highlighted the role of amygdalar CRF in enrichment; however the knockdown study was only conducted in male mice, and the enrichment study was only conducted in female mice. Knockdown of CRF1 was conducted in male mice to avoid any estrous cycle effects, although estrous phases were not determined, while the enrichment experiment was conducted in females since male mice are more aggressive compared to male rats. Therefore, investigations of the role of CRF in environmental enrichment in both sexes is still needed, particularly in the BLA.

4.5. Alcohol, environmental enrichment, and CRF

Since amygdalar CRF1 is associated with the anxiolytic effects of environmental enrichment, the CRF system may also be involved in the protective alcohol phenotype of environmental enrichment. To date, no thorough investigation has linked CRF signaling underlying environmental enrichment and reduced ethanol drinking. Circumstantial evidence does suggest that this may be an interesting avenue. For example, rats bred for alcohol preference have increased genetic variants in the Crhr1 promoter as well as increased Crf mRNA expression in the amygdala (central, basolateral, medial), hippocampus (CA1, CA3, CA4), and nucleus accumbens (Hansson et al., 2006). Male alcohol-preferring rats reared in an enriched condition showed significantly lower ethanol consumption, ethanol preference, and self-administration of ethanol (Deehan Jr et al., 2011). Antalarmin (a CRF1 receptor antagonist) and diazepam decreased anxiety-like behaviors and decreased acquisition of ethanol self-administration in another line of alcohol-preferring rats that were isolation-reared (Lodge and Lawrence, 2003). Only CRF1 receptor antagonism was able to reduce established ethanol consumption and ethanol preference in isolated rats (Lodge and Lawrence, 2003). Interestingly, adult female mice during abstinence from ethanol show increased depressive-like behaviors while mice housed in an enriched condition during the abstinence period do not show increased depressive-like behaviors (Pang et al., 2013). Female mice housed in isolation and those in an enriched condition during abstinence from alcohol both show a significant decrease in hypothalamic Crf mRNA expression (Pang et al., 2013). Switching environments from enriched to group housed has been shown to increase Crf mRNA levels in the BNST and increases the rewarding effects of cocaine (Nader et al., 2012). Treatment with antalarmin for one week during the loss of environmental enrichment prevented the enhancement of the rewarding effects of cocaine (Nader et al., 2012). Thus, the protective effects of environmental enrichment on ethanol administration may be due to modulating CRF signaling in the brain.

Enrichment may be protective against ethanol drinking due to alterations in CRF signaling in the BLA. As described above, enrichment results in lower Crhr1 mRNA in the BLA and after a longer time, a decrease in Crhr1 mRNA in the BNST (Sztainberg et al., 2010), a downstream target of the BLA. The loss of enrichment increases Crf mRNA, but not Crhr1 mRNA, in the BNST (Nader et al., 2012), and specifically knocking down CRF1 in the BLA mimics the anxiolytic-like effects of environmental enrichment (Sztainberg et al., 2010). Other evidence shows that although enrichment can alter behavior, in contrast to the BLA, enrichment does not overcome molecular alterations in the CRF system in other regions. Maternal separation increases ethanol drinking in juveniles and environmental enrichment reduces drinking in the same rats as adults (Odeon and Acosta, 2019). Repeated maternal separation causes increased corticosterone and increased Crf mRNA in the PVN and enrichment post weaning blocks this increase in corticosterone, but not the increase in Crf mRNA in the PVN (Francis et al., 2002). Enrichment decreases depression-like behavior induced by alcohol abstinence, but does not mitigate the increase in hypothalamic Crf mRNA induced by alcohol abstinence (Pang et al., 2013). Wang et al. found that maternal separation epigenetically upregulates CRF, increases hippocampal CRF mRNA and protein, impairs hippocampal long-term potentiation (LTP), and impairs memory which is all prevented by environmental enrichment (2014). Further work is needed to confirm the mechanistic role of enrichment on the CRF system in the BLA and determine the importance of enrichment modulating CRF in the hippocampus on ethanol drinking. A better understanding of the protective role of reduced CRF signaling in environmental enrichment may also shed light on the failure of translation from the preclinical work on CRF to human populations. A proposed role for BLA CRF in the protective effects of enrichment on alcohol drinking behavior is depicted in Fig. 3.

Fig. 3. Potential mechanistic role of CRF in the BLA in the protective alcohol drinking phenotype of environmental enrichment.

Schematic of the continuum of isolation to enrichment housing during adolescence. Enrichment reduces Crf mRNA in the basolateral amygdala (BLA), which may underlie the protective effects of enrichment. Enriched rodents drink less ethanol compared to group or pair housed rodents and isolated rodents drink the highest about of ethanol.

5. Conclusions

Although work examining the role of sex, age, and housing environment in the interactions between alcohol and the CRF system has been relatively sparse, important tentative conclusions can already be drawn based on current findings. Importantly, the influence of stress on the CRF circuitry appears to unmask sex, age, and housing environment differences in the structure and function of this signaling system. Although males and females exhibit some brain-region specific differences in CRF/CRF1 mRNA and protein expression under non-stressed conditions, the response of the CRF system to stressors and its regulation of anxiety-related behaviors appears to be more sexually dimorphic. The CRF system in female rodents also appears to be more sensitive to alcohol-induced changes, but lacks the adaptations to stress-alcohol interactions that are exhibited by the male CRF system. The impact of stress on alcohol-related behaviors and the CRF system also appears to be more pronounced in adolescents as compared with adults. The ontogeny of CRF/CRF1 in different brain regions during adolescent development may be related to the divergent effects of stress and alcohol on this age group. The protective effect of environmental enrichment on alcohol consumption appears to be related to decreases in anxiety-like behavior. Together, these observations support the conclusion that the greatest therapeutic benefit from CRF-directed treatments for AUD may be in contexts of higher stress, such as stress-related relapse and comorbid anxiety disorders.

Another important conclusion is evident in the diversity of ages, housing conditions, and alcohol exposure procedures employed by the studies reviewed here: it is difficult to determine which variable(s) might have contributed to different findings from similar studies. Overall, however, despite this consideration there are strikingly similar findings across diverse study designs. For instance, the increased sensitivity of females to alcohol-induced changes in corticosterone has been consistent across time, species, and differing doses of alcohol. Evidence for increased sensitivity of female rodents to alcohol’s effects on the CRF system has also come from multiple exposure models, species, and housing conditions. The ability of adolescent alcohol exposure to increase PVN Crf mRNA but decrease CeA Crf mRNA has also been shown repeatedly using multiple exposure procedures, age ranges, and housing conditions (albeit mostly in males). Despite a variety of alcohol exposures and different enrichment protocols including comparing between enriched vs group housed, group housed vs isolated, pair housed vs isolated, and even enriched vs socially housed vs isolated, environmental enrichment clearly shows a protective effect on alcohol addiction-related behaviors. Furthermore, there is consistency across species as well. In this light, a multiplicity of experimental procedures can be a barrier to replication of results, but it can also help scientists determine which effects are pronounced enough to emerge across differing manipulations of variables and even species. The most robust effects should therefore be the top candidates for translation.

5.1. Interactions

The role of biological variables becomes more complex, and more interesting, when interactions between multiple variables are considered. The intersection of age and sex has received little investigation in the context of alcohol’s effects on CRF (Table 2), but the available evidence suggests that interactions between these two variables are likely important. The developmental trajectory of CRF maturation appears to differ between sexes, with male but not female rats showing increases in amygdala Crf mRNA during adolescence (Viau et al., 2005). The effects of CRF overexpression in the PFC during early life development on anxiety-like behavior also appear to differ between sexes (Toth et al., 2014, 2016). More research is needed to determine whether sex differences in CRF expression or activity in other brain regions emerge pre- or post-pubertally and are subject to changes during adolescence. Intriguingly, adolescent intermittent ethanol blunts the effect of alcohol challenge in adulthood to increase Crf mRNA in the PVN of male but not female rats (Logrip et al., 2013), suggesting that these sex differences are relevant to the effects of alcohol.

The interaction of sex and housing environment is exemplified by the role of social hierarchy on plasma hormones and behavior. In non-human primates (Sapolsky, 2005), rats (Monder et al., 1994), and mice (van den Berg et al., 2015), group housing produces social ranks with dominant and subordinate individuals, particularly in males. These ranks are commonly associated with changes in serum levels of corticosterone and testosterone (Williamson et al., 2017). Rank in social hierarchy has been shown to affect alcohol intake in multiple species (Blanchard et al., 1987; Kudryavtseva et al., 1991; McKenzie-Quirk and Miczek, 2008), with subordinate animals typically displaying higher corticosterone, lower testosterone, and greater alcohol drinking. Social rank has also been shown to influence Crf gene expression in the CeA (Albeck et al., 1997), and the CRF1 receptor functionally regulates the expression of anxiety behavior following social defeat in male rats (Liebsch et al., 1995). As depicted in Table 3, group housing and environmental enrichment produce molecular and behavioral changes in both male and female animals, but there is some evidence that the effects of alcohol on male and female enriched animals are mixed, particularly with respect to age. These pieces of evidence suggest the potential for group housing to cause disparate effects on male and female CRF regulation of alcohol effects, such that even identical housing conditions may lead to variation on the basis of sex.

Table 3.

Environment in the effects of CRF and alcohol.

Reference	Species	Sex	Age	Housing	Exposure	Region	Results
Blomeyer et al., 2008^	Human	F + M	15 years	-	-	-	Neg life experiences and CRHR1 SNP assoc. w/drinking
Hansson et al. (2006)	Rat (msP vs Wistar)	M	Adult (350–400g)	Not specified	- Operant SA of EtOH	Various (Amy, Hipp) -	Alcohol preferring line ↑ variation in Crhr1 promoter, ↑ Crf mRNA CRF1 antagonist ↓ SA of EtOH + blocks shock-induced reinstatement of EtOH only in alcohol preferring line
Kulkosky et al. (1980)	Rat	F + M	PND 60–106	EE vs group or isolated	Three-bottle choice EtOH drinking	–	EE rats drink less EtOH vs group or isolated
Deehan Jr. et al., 2007	Rat	M	PND 21–111	EE vs pair or isolated	Operant SA of EtOH	–	Isolated rats ↑ SA of EtOH compared to grouped or EE
McCool & Chappell (2009)	Rat	M	PND 21–42	Group vs isolated	Operant SA of EtOH, two-bottle choice EtOH drinking	–	Group housed rats drink less and ↓ SA of EtOH vs isolated
Holgate et al. (2017)	Mouse	M	PND 35–40, 40–75	EE vs group or isolated	Binge EtOH drinking	–	EE very low EtOH preference followed by group, isolated highest. Changing housing conditions alters preference
Rodriguez-Ortega et al. (2018)	Mouse	M	PND 28–140	EE vs group	Binge EtOH drinking	–	EE in adolescence ↓ EtOH drinking as adults
de Carvalho et al. (2010)	Rat	F	PND 21–90	EE vs group	Two-bottle choice EtOH drinking, EtOH CPP	–	EE rats drink less EtOH and reduced EtOH CPP vs group
Lopez et al. (2011)	Mouse	F + M	PND 21–65, PND 60–105t	Group vs isolated	Two-bottle choice EtOH drinking	–	Group housing during adolescence, but not adulthood, ↓ EtOH drinking. CVS ↑ EtOH drinking in group housed
Schenk et al. (1990)	Rat	M	PND 21–84, PND 65–149t	Group vs isolated	Two-bottle choice EtOH drinking	–	Group housing during adolescence, but not adulthood, ↓ EtOH drinking
Marianno et al. (2017)	Mouse	M	PND 56–70	EE vs group	Two-bottle choice EtOH drinking	–	EE in adulthood blunts 24h EtOH consumption after restraint stress
Pang et al. (2013)	Mouse	F	PND 56–112	EE vs group	Two-bottle choice EtOH drinking	– Hyp	EE reduces depression-like behavior induced by alcohol abstinence Alcohol abstinence ↑ Crf mRNA, not reduced by EE
Sztainberg et al. (2010)	Mouse	F	PND 28–98	EE vs group	–	– BLA, BNST	EE ↓ anxiety-like behaviors EE ↓ Crhr1 mRNA
		M	PND 56–70	Group	–	BLA	CRF1 knockdown ↓ anxiety-like behaviors
Deehan Jr. et al., 2011	Rat (P vs NP)	M	PND 21–81	EE vs pair or isolated	Two-bottle choice EtOH drinking, operant SA of EtOH	–	EE alcohol preferring rats drink less, prefer less, and ↓ SA of EtOH vs pair and isolated
Lodge and Lawrence (2003)	Rat (FH)	–	PND 105–146	Isolated from weaning	Two-bottle choice EtOH drinking		CRF1 antagonism ↓ acquisition and maintenance of EtOH drinking
Nader et al., 2012	Mouse	M	PND 21–81,82–171	EE vs group	Loss of enrichment in adulthood, coc CPP	BNST –	Loss of EE ↑ Crf mRNA, no change in Crhr1 mRNA CRF1 antagonism blocks increase in coc CPP when EE is lost
Francis et al. (2002)	Rat	-	PND 1–14, 110a	EE (PND 22–70) vs pair housed	Repeated MS	PVN	MS ↑ Crf mRNA, not reversed by EE in adolescence
Wang et al. (2014)	Rat	-	PND 1–10, > 77a	6h/day EE (8 wks) vs group	Repeated MS	Hipp	MS ↑ CRF mRNA & protein, impairs hipp LTP but blocked by CRF1 antagonism or EE. MS epigenetically upregulates CRF and memory dysfunction, reversed by EE

^aAge at which the same cohort was tested during adulthood,

^tage of separate cohort of adults,

^{^}also referenced in Table 2.

- = condition not tested. F = female. M = male. PND = postnatal day. RA = restricted access. SA = self-administration. Coc = cocaine. Nic = nicotine. Adol = adolescence. W/ = with. Neg = negative. MS = maternal separation. EE = environmental enrichment. Wks = weeks. LTP = long-term potentiation. CPP = conditioned place preference. CVS = chronic variable stress. FH = Fawn-Hooded rats.

Finally, sex, age, and environment have been shown to interact in the context of the effects of shipping animals from large vendors to individual labs. This is a common practice in neuroscience research, and evidence suggests that shipping may have unique effects on physiology and behavior as compared with breeding rodents in-house (Landi et al., 1982). More recent reports have shown that shipping specifically during adolescence appears to alter sensitivity to several psychoactive drugs in a sex-dependent manner (Wiley and Evans, 2009). As adolescents have been shown to have heightened sensitivity to the influence of handling stress on alcohol effects as compared with adults (Ristuccia et al., 2007), it is possible that shipping stress, sex, and age could interact to produce more variability in the effects of ethanol on the stress circuitry, including CRF signaling. These examples are not exhaustive, but have been chosen to highlight the risks of neglecting biological variables in experimental design.

5.2. Recommendations

The first and most critical step in adequately accounting for biological variables is to specifically report their status in the methods section of every manuscript. A 2011 meta-analysis (Beery and Zucker) reported an alarmingly high number of research papers that failed to indicate the sex of the animal under study, and it is still commonplace to give only limited information about subject age and housing conditions over the lifespan. Furthermore, methods should include specifics about the age of the animals when obtained and age of animals during testing. Of equal importance, an awareness of the role of biological variables should inform experimental design. Within the same study or between studies that may reasonably be compared (such as work from the same research group, within the same animal line, or replications of prior work) every effort should be made to keep biological variables as similar as possible. The use of broad age ranges (i.e. 6–12 weeks, 2–6 months), particularly those that include different developmental epochs, should be avoided. Variations in housing conditions, such as group versus isolation housing, unless otherwise justified by experimental needs, should also be avoided. When necessity requires changes in housing, these differences should be clearly stated in the methods and explored in the results and discussion as appropriate. And in accordance with NIH policy, the majority of preclinical CRF research should include both male and female subjects. As suggested by Shansky (2019) and others, the directive to consider sex as a biological variable does not necessitate the inclusion of five times as many subjects in order to have equal representation of females throughout the estrous cycle in the design. An initial approach might simply be to run mixed cohorts of males and females and determine whether there is any evidence of sex differences in the measures under observation. If so, increasing subject numbers to allow for the analysis of sex is warranted, but if not, the study can continue with mixed cohorts and collapse across the sex variable. In a similar vein, if female data are no more variable than male data in a given experiment, this can be interpreted as evidence that estrous cycle variations are not a major contributor to variation and controlling for estrous phase may not be necessary. A similar approach is well-suited to the consideration of age as a biological variable; if cohorts are mixed, differences between age groups (particularly adolescent and adult animals, as these two stages are most easily conflated in neuroscience experimental design) should be assessed and if age differences are evident, group sizes should be increased to allow for age analyses. The role of housing is particularly important in the context of CRF research, as an unenriched environment is a condition that may alter CRF activity. When possible, housing environments should include toys and nesting material and the presence of conspecifics for social interaction. If experimental conditions require isolation housing or limited enrichment, it is worthwhile to determine whether this housing environment changes basal CRF when compared with typical rodent housing. These suggestions do require some additional planning and analysis on the part of the research community, but their impact on overall animal numbers will often be small, whereas the potential increase in rigor and reproducibility of research is large.

Despite a wealth of strong preclinical evidence for a functional role of CRF signaling in the development of alcohol dependence as well as other psychiatric illnesses with stress components, as a group CRF1 antagonists have not been successful when research moved from the lab to the clinic. The failure of two clinical trials for CRF1 antagonists as a treatment for AUD, in addition to failures in treating depression and several anxiety disorders, has been a subject of frustration and debate within the field. A number of potential explanations for these outcomes have been suggested, including a revised understanding of CRF1 antagonism as less efficacious in conditions where stress is a chronic rather than intermittent component (Koob and Zorrilla, 2012), suboptimal binding kinetics of the CRF1 antagonists that have been used in most clinical trials (Fleck et al., 2012; Spierling and Zorrilla, 2017), and overemphasis on CRF1 that neglects the contribution of the CRF2 receptor or the CRF binding protein CRF-BP (Albrechet-Souza et al., 2015; Haass-Koffler et al., 2016). But so many inconsistencies between preclinical research using rodents and clinical outcomes in human participants does demand closer scrutiny of the animal models that informed the trials. Spierling and Zorrilla (2017) have pointed out that several benchmarks for clinical efficacy of CRF1 antagonists for AUD were also absent in animal models (no effect on cue or substance-induced reinstatement of alcohol seeking, no reduction of alcohol drinking in models other than chronic intermittent exposure to vapor). The failure of the research to translate may therefore be viewed not as a failure of preclinical models of AUD, but of emphasis on the wrong evidence. However, these models overwhelmingly used adult male rodents in relatively unenriched environments, making the chances of successfully translating the preclinical findings to a wider human population even lower. The International Stress and Behavior Society (ISBS) Strategic Task Force on anxiolytic drugs emphasized the need for including women and both children and older adults in preclinical models to improve the validity and generalizability of animal models for anxiety disorders (Stewart et al., 2015). We believe that similar considerations are warranted in the pursuit of new treatments for AUD. Like many anxiety disorders, AUD often manifests differently in women than it does in men. For instance, women develop AUD more rapidly, and at lower levels of alcohol intake, than do men (Erol and Karpyak, 2015) and may present with different symptoms (van der Walde et al., 2011). These sex differences limit the relevance of preclinical animal models developed solely based on disease presentation in males. These unique characteristics of AUD in women may also indicate important new avenues for research that can suggest novel therapeutic strategies for both sexes that are not obvious in work with males alone. Additionally, despite the fact that the majority of AUD patients began their alcohol use during adolescence (Grant et al., 2015), most preclinical animal research is conducted in adult rodents with no prior history of alcohol exposure. This is a significant departure from the human condition that may impede the discovery of new targets for pharmacological intervention. The intersection of sex and age is also important, as women are at increased risk for the development of AUD after menopause (Milic et al., 2018). Housing conditions likewise have profound effects on alcohol consumption, and pharmacological intervention may have different effects in enriched versus isolated rodents. Designing thoughtful experiments that better replicate the human condition via careful consideration of biological variables may improve the translatability of alcohol research and point us towards the best avenues for therapeutic intervention.

HIGHLIGHTS.

• Corticotropin-releasing factor (CRF) is involved in stress, anxiety, and addiction.

• The biological variables sex, age, and environmental enrichment interact with CRF.

• CRF, age, sex, and environment contribute to cellular/behavioral alcohol effects.

• How CRF and these biological variables interact in anxiety and alcohol is unclear.

• An integrated perspective on biological variables in CRF and alcohol is needed.