Gene-alcohol interactions in birth defects

Abstract

Most human birth defects are thought to result from complex interactions between combinations of genetic and environmental factors. This is true even for conditions that, at face value, may appear simple and straightforward, like fetal alcohol spectrum disorders (FASD). FASD describe the full range of structural and neurological disruptions that result from prenatal alcohol exposure. While FASD require alcohol exposure, evidence from human and animal model studies demonstrate that additional genetic and/or environmental factors can influence the embryo’s susceptibility to alcohol. Only a limited number of alcohol interactions in birth defects have been identified, with many sensitizing genetic and environmental factors likely yet to be identified. Because of this, while unsatisfying, there is no definitively “safe” dose of alcohol for all pregnancies. Determining these other factors, as well as mechanistically characterizing known interactions, is critical for better understanding and preventing FASD and requires combined scrutiny of human and model organism studies.

1. Introduction

Humankind has grappled with the effects of alcohol since its discovery. Alcohol (ethanol) is a teratogen (an agent capable of causing congenital defects). Given the widespread use of alcohol, it is likely the most common environmental cause of human birth defects, and is certainly the most common preventable cause of human birth defects. Fetal alcohol spectrum disorders (FASD) describe the full constellation of congenital abnormalities and neurological deficits associated with embryonic or fetal exposure to alcohol (Riley, Infante, & Warren, 2011; Wozniak, Riley, & Charness, 2019). Prenatal alcohol exposure causes a wide range of embryonic/fetal abnormalities, including malformations of the brain, face, heart, and other organs, as well as short stature, which can cause lifelong, irreparable burdens for those affected (Caputo, Wood, & Jabbour, 2016; Coriale et al., 2013; Denny, Coles, & Blitz, 2017; Wilhoit, Scott, & Simecka, 2017). With recent advances in FASD diagnostic criteria, FASD are now understood to affect a shocking 1–8% of all children born in the United States, making it one of the most common human birth defects (May et al., 2009, 2014, 2020a, 2020b, 2020c).

This high prevalence cannot be explained simply as only a problem of alcoholism, as has been suggested in the past (Ulleland, 1972). Researchers have identified a worrying new trend of increased drinking, including binge drinking, while pregnant (Howard et al., 2022). This trend appears to be driven by a combination of factors, including inadequate or unreliable information from health professionals and/or a disregard of public health recommendations by patients (Hammer & Rapp, 2022). While difficult to truly calculate, it is likely that a majority of embryonic alcohol exposures are unintentional. Around half of all pregnancies are unintended (Finer & Zolna, 2016), and critical stages of embryogenesis, including formation of the brain and face (Sulik, 2005), occur prior to the average gestational age of pregnancy detection, 5.5 weeks (Branum & Ahrens, 2017). For example, a Norwegian study found that 89% of participants reported drinking prior to pregnancy (59% binge drinking), of which 85% changed their drinking behavior once their pregnancy was identified. This suggests that a large percentage of pregnancies were at risk for prenatal alcohol exposure during the critical early weeks of development.

While alcohol exposure is required for FASD, phenotypic outcomes following prenatal alcohol exposure are highly variable. Clearly, not all alcohol exposures result in FASD. This suggests additional factors, either genetic or environmental, can affect embryo susceptibility to alcohol. This also means that a dose of alcohol that is not deleterious for one embryo, could be harmful to another. Thus, there is no single dose of alcohol that is safe for all pregnancies. Instead, multiple lines of evidence from human and model organism studies demonstrate that genetic and environmental factors can alter alcohol toxicity. However, the identities and, moreover, the mechanisms of these myriad interactions with alcohol are mostly unknown, with few notable exceptions. In this review, we examine the history of alcohol-induced birth defects and the evidence for gene-alcohol and multifactorial interactions in alcohol-induced birth defects. We focus this chapter on interactions affecting the face and brain, as most studies on gene-alcohol interactions have focused on these structures.

2. FASD background and prevalence

Homo sapiens have assuredly been consuming alcohol since they first emerged on Earth. There is evidence of humans storing fermented beverages dating as far back as 7000 BCE (McGovern et al., 2004). The association between excessive alcohol consumption and the immediate negative effects on the drinker likely dates back at least equally as far. When humans understood the potential adverse effects of embryonic alcohol exposure is much less clear.

An understanding of the effects of alcohol exposure on a developing embryo in turn requires an understanding of embryogenesis, or development, or at a minimum that such a process happens. In Western writing, many give Aristotle (BCE 384–322) credit for the foundations of developmental biology (Wallingford, 2021). At this same time, or even earlier, Eastern cultures, and potentially cultures of ancient Mexico, also appreciated embryogenesis (Wallingford, 2021). Readers interested in the history of developmental biology are encouraged to read Dr. Wallingford’s detailed review on the subject (Wallingford, 2021). Thus, cultures across the globe had the capacity to associate alcohol consumption and embryonic outcomes over two millennia ago. But, when did they?

Passages from the ancient Greek and Roman as well as the Bible have been used to suggest that people from these ages understood the potential for adverse outcomes following embryonic alcohol exposure. However, careful interpretations taking into account the culture at the time do not particularly support such claims. Rather, these passages relate birth outcomes to the state of being of the parents at the time of conception (Abel, 1997, 1999; Brown, Bland, Jonsson, & Greenshaw, 2019; Obladen, 2021). Notably, the idea that birth outcomes related to the mental state of one or both parents during conception was not specific to alcohol (Wallingford, 2021). Therefore, it is highly unlikely that these statements demonstrate an understanding of alcohol teratogenesis. While we cannot know with 100% certainty, it seems safest to presume that such statements relate to an understanding of the immediate consequences of heavy intoxication and not an effect on the developing embryo.

Part of the difficulty in determining when there was a general recognition that ethanol consumption could be teratogenic is disentangling the social stigma of excessive alcohol consumption from birth outcomes. Authors in the 1700s commented on the deleterious effects of gin (Warren, 2015). Hogarth’s etching Gin Lane has been used by some as evidence for an understanding of the deleterious effects of drinking on offspring (Warner & Rosett, 1975). However, it is worth noting that the companion etching Beer Street also shows a large group of individuals consuming alcohol, but this time beer, with no deleterious effects. The real difference between the two etchings would appear to be class.

There are sporadic manuscripts reporting negative effects of alcohol in pregnant women through the 1800s. These manuscripts are also often rife with sexism, classism and eugenics. Additionally, it is worth bearing in mind, that these manuscripts were published during the growth and peak of the temperance movement, a social movement that promoted moderation or abstinence from alcohol. Thus, many individuals would believe that alcohol was detrimental in its entirety and in all aspects of life. For instance, a report to the British House of Commons in 1834 states “…the highest medical authorities, examined in great numbers before Your Committee, being uniform in their testimony that ardent spirits are absolutely poisonous to the human constitution; that in no case whatever are they necessary, or even useful, to persons in health; that they are always, in every case and to the smallest extent, deleterious, pernicious or destructive, according to the proportions in which they may be taken into the system…” (Great Britain. Parliament. House of commons. Select committee on inquiry into drunkenness & Buckingham, 1834). Thus, it seems imperative to consider whether or not such studies were simply observational.

In the late 1800s numerous publications comment on birth outcomes of mothers with alcoholism. In 1876, Haddon states that intemperate women often had spontaneous abortions (Haddon, 1876). This appears to be purely observational as no comparison group was analyzed and the author attributes the effect on uterine function as opposed to embryogenesis (although it should be noted that uterine function may well play a role in FASD). It is further worth noting that, while the author was discussing pregnancy outcomes in alcoholic women he states, without supporting evidence, that moderate alcohol consumption likely “differs only in degree, and not in kind,” suggesting a temperance bias. In 1888, there appears to be debate as to whether paternal or maternal alcohol consumption has a greater effect on the intellectual outcomes of pregnancy. In his manuscript, Duncan pushes back against the idea that the male has a greater influence saying, “What is the comparative potency of alcoholic disease of the testicles and of the ovaries on population we have at present no definite means of even guessing, but it may be safely judged that that of the ovaries is paramount on account of the much greater part taken in reproduction by the female than by the male” (Duncan, 1888). Here, again, the effect of alcohol appears to be attributed to sex organs as opposed to a direct effect on the embryo. By 1899 it was noted that infant mortality (before age 2) and epilepsy was higher in children born to alcoholic mothers relative to the general population (Sullivan, 2011). This study suffered somewhat from a lack of a truly appropriate comparison group. However, it did conclude that these effects were due to the continued exposure of the developing embryo to alcohol and is perhaps the first strong evidence for what we now term FASD.

The early 1900s provides further careful study indicating the deleterious effects of prenatal alcohol exposure. For instance, in a 1906 conference on infant mortality, two papers reference alcoholism (National Conference on Infantile Mortality, 1906). Dr. Woodhead states that alcohol exerts “poisonous action” on all cells and that the embryo should be particularly sensitive to alcohol due to its proliferative nature. This appears to be the first attempt to gain systematic information on the effects of prenatal alcohol exposure as the author proposes a questionnaire, including questions on the quantity of alcohol consumption as well as hereditary diseases of the parents, a much needed control. However, the only birth outcome examined was suckling. Ballantyne discusses the importance of preventative medicine and prenatal development. In contrast to Dr. Woodhead, Dr. Ballantyne suggests that the effect of alcohol is not on the fetus itself: “…this effect is due rather to the changes in the maternal kidneys and in the placenta than to any direct influence which the drug has upon the fetal tissues themselves.” Thus, while it was becoming clear that alcohol exposure could associate with deleterious birth outcomes at this time, the cause was not understood.

The first descriptions of what we now know of as fetal alcohol syndrome (FAS) would be made in the mid-1900s. Dr. Paul Lemoine is often credited with the first description of what would later be termed FAS in 1968 (translated from French to English in 2003) (Lemoine, Harousseau, Borteyru, & Menuet, 2003). Interestingly, and relevant to whether previous generations had appreciated alcohol as a teratogen, he stated in a later review that it took some time for him to determine that the syndrome he was seeing was due to alcohol, because in his medical training “…we were taught without any doubt that it [alcohol] had no teratogenic effects…” (Lemoine, 1997). It was a staff member at the hospital that informed him that the mothers of two children he was seeing were alcoholics (Lemoine, 1997). In addition to describing the syndrome, Lemoine followed a subset of children up to 16 years of age and noted that they did show catch-up growth but displayed psychomotor deficits and increased aggression (Lemoine et al., 2003). The manuscript is notable also as it out and out rejects the theory that excess alcohol use at the time of conception was the culprit (Lemoine et al., 2003).

While Lemoine, indeed, deserves credit for identifying FAS, he notes that Dr. Jaqueline Rouquette provided the first characterization of what would be FAS in her thesis in 1957 (Rouquette, 1957). While her thesis has been referenced in passing by a handful of reviews on FAS, no details have previously been provided, presumably because it was only published as a thesis and in French. Therefore, we provide some details to put Dr. Rouquette’s work into proper historical perspective. Between the years 1954 and 1957, Dr. Rouquette examined 100 children of confirmed alcoholic parents admitted to the Paul Parquet Infant Hygiene Center, which provided convalescent care to children from 1 month to 3 years of age. She detailed whether it was the mother (n=28), father (n=42) or both parents (n=30) that were alcoholic. In addition to alcoholism alone, she performed analyses of additional comorbidities. She concluded that maternal alcoholism was the predominant factor in preterm birth and congenital birth defects. She also noted a characteristic face in these children. This included a short upper lip, flattened root of the nose, hypertelorism and microcephaly. While she only dealt with young children, she also noted that they exhibited nervous, aggressive and unsociable behaviors. Through the comparison of co-morbidities, she concluded that alcohol was the culprit, as such, Dr. Rouquette’s work appears to be the first clear identification of FAS.

It would be a series of manuscripts from the University of Washington that culminated in the definition and acceptance of FAS. In an 18 month period from the beginning of 1968 to mid-1969, Ulleland found that children born to alcoholic mothers were small for gestational age and failed to thrive (Ulleland, 1972). Notably, she found that over 80% of infants born to alcoholic mothers were below the 10th percentile in birth weight (Ulleland, 1972). In 1973 Jones and colleagues published their findings of eight children across three ethnic groups born to alcoholic mothers that largely shared a set of growth, craniofacial, limb and cardiac defects (Jones, Smith, Ulleland, & Streissguth, 1973). In the same year, Jones and Smith reported three additional cases and formalized the name Fetal Alcohol Syndrome (FAS) (Jones & Smith, 1973).

Despite this, it was not until 1981 that a U.S. surgeon general advisory warning recommended that pregnant women avoid consumption of alcohol. Just a year later, the noted fetal alcohol effects researcher Henry Rosett stated in the Hearing before the Subcommittee on Alcoholism and Drug Abuse of the Committee on Labor and Human Resources, “The credibility of the strong evidence of the dangers to the fetus from heavy drinking and alcoholism is weakened through exaggeration of possible dangers from small amounts of alcohol” (United States. Congress. Senate. Committee on Labor and Human Resources. Subcommittee on Alcoholism and Drug Abuse, 1982). Dr. Rosett provided references for studies failing to find effects of moderate alcohol on birth outcomes (Hingson et al., 1982; Kaminski, Rumeau-Rouquette, & Schwartz, 1976; Tennes & Blackard, 1980).

While true that some studies found no association between moderate alcohol consumption and deleterious outcomes, others did. Indeed, the Tennes and Blackard study referenced by Dr. Rosett did demonstrate a weak but significant association of moderate alcohol consumption and premature delivery (Tennes & Blackard, 1980). Prior to Rosett’s remarks, a study of 163 infants with confirmed alcohol exposure were evaluated and 11 were found to have features consistent with FAS (Hanson, Streissguth, & Smith, 1978). The mothers of six of these children reported consuming between 1 and 2 oz of absolute alcohol/day during pregnancy and only two were noted to be alcoholics (Hanson et al., 1978). Additionally, two manuscripts published side by side in the Lancet in 1980 also associated more moderate alcohol consumption patterns with miscarriage (Harlap & Shiono, 1980; Kline, Shrout, Stein, Susser, & Warburton, 1980). In a very large study of 32,019 women, Harlap and Shiono found that moderate alcohol consumption (1–2 drinks/day) during the second, but interestingly not the first, trimester associated with almost double the risk of miscarriage (Harlap & Shiono, 1980). Kline and colleagues found that consumption of alcohol 2–6 days/week more than doubled the chances of spontaneous abortion (Kline et al., 1980).

While Dr. Rosette’s remarks likely were directed to even lower consumption levels than those in these studies, they do bring up a question of immense importance: What is a safe level of alcohol consumption during pregnancy? The unsatisfying answer is that we simply do not know. Clearly, birth outcomes following exposure vary widely and this is due to a large number of other factors, environmental and genetic, that impact the result of prenatal alcohol exposure. Until we can understand how alcohol can interact with the tens of thousands of genes in our body and the hundreds of thousands of chemicals in our environment, we will not have an answer to this question.

3. What are multifactorial interactions in birth defects?

Despite being the leading cause of infant mortality (Matthews, MacDorman, & Thoma, 2015), the vast majority (80%) of human birth defects do not have a known cause (Feldkamp, Carey, Byrne, Krikov, & Botto, 2017). This has led to the prevailing hypothesis that most birth defects do not have a single cause, but are instead the result of “multiple hits” (reviewed in (Beames & Lipinski, 2020; Brent, 2004; Krauss & Hong, 2016)). In this model, two or more insults intersect to affect embryogenesis. These insults can be either genetic, such as a mutation or polymorphism, or environmental, such as exposure to a teratogen like alcohol. As such, multifactorial interactions fall under three general categories: gene-gene interactions, gene-environment interactions, and co-environmental interactions (co-exposure to two or more toxicants). Thus, for alcohol related birth defects the relevant multifactorial interactions are gene-environment (specifically gene-alcohol) and co-environmental interactions (e.g., alcohol+ another toxicant).

Multifactorial interactions are not always direct. That is, the factors can impinge on the same genetic networks (such as a Sonic Hedgehog (Shh) mutant exposed to a Shh pathway inhibitor), but they can also be indirect, such as two mechanistically distinct perturbations that result in disruptions of common or related biological processes. Multifactorial interactions also can be either additive, where the effect of both factors together equals the cumulative effect of both factors alone, or synergistic, where the effect of both factors is greater than the cumulative effect of both factors alone. These synergistic interactions pose the most concern from the perspective of public health and risk assessment. Most developmental toxicity dose-response assays are performed on single chemicals in non-genetically predisposed subjects, while almost nothing is known about the effect of chemical mixtures or single exposures to genetically predisposed individuals (Krewski et al., 2010; Tice, Austin, Kavlock, & Bucher, 2013). Most chemical exposures occur in mixtures (Koppe et al., 2006; Mitro, Johnson, & Zota, 2015), and each person is born with millions of genetic variants, of which 250–300 have been found to be predicted loss-of-function mutations in annotated genes, with 50–100 of these variants being previously associated with a heritable diseases (Abecasis et al., 2010). Thus, reference doses of chemicals that pass regulatory assessment as single exposures could still be harmful when combined with genetic susceptibility or with other toxicants in mixtures.

4. Evidence for gene-alcohol interactions

While alcohol exposure has been definitely linked to birth defects, what’s less understood is the basis for variability in phenotypic outcomes following an exposure. For example, not all alcohol exposures cause birth defects, even when dose, duration, and timing of exposure are controlled. That is, the same alcohol exposure can cause variable outcomes, with some resulting in malformations while others cause no apparent phenotypes. This apparent discordance could be explained by the presence of genetic and/or environmental modifiers. Several lines of evidence suggest embryonic and maternal genetics can influence embryonic susceptibility or resilience to alcohol. These include human genetic studies, especially on twins and siblings, strain differences in susceptibility in multiple species, and specific gene-environment interactions in both humans and research animals.

4.1. Human studies

One of the first pieces of evidence used to suggest a genetic component in FASD was the observation that children born to mothers who have previously given birth to a child with FASD are more likely to also have FASD. However, this observation could be explained by environment, e.g., a common pattern of drinking during pregnancy, rather than an intrinsic factor like genetics.

4.1.1. Twin studies

The most direct evidence for a genetic component in human FASD is from twin and sibling association studies. In these studies, pairs of twins or siblings with confirmed prenatal alcohol exposure are assessed for FASD, and concordance rates (i.e., whether the twins had the same or different phenotypic outcomes) are calculated. These studies are the gold standard for human association studies, as they enable comparisons between two individuals with near identical embryonic environments and exposures. In addition, comparison of monozygotic twins (100% shared genetic identity) and dizygotic twins (50% shared genetic identity) enables direct comparison of the amount of shared genetic relatedness to concordance in FASD outcomes. Siblings (50% shared genetic identity) and half-siblings (25% shared genetic identity) can also be analyzed, although here exposures are not identical. Together, this provides a sort of “dose-response” for genetic relatedness.

The first of these studies by Streissguth and Dehaene in 1993 provided clear support for a genetic component to the effect of prenatal alcohol exposure (Streissguth & Dehaene, 1993). The authors wondered why FASD outcomes can vary so widely, even in cases of known high alcohol exposure during the critical period for FAS, hypothesizing that genetics can modify embryo susceptibility. The authors compared the rate of concordance of FAS and the milder (antiquated term) fetal alcohol effects (FAE) between 16 pairs of either monozygotic (n=5) or dizygotic (n=11) twins with confirmed heavy alcohol exposure born in the United States or France. The pairwise concordance rate (i.e., FAS diagnosis in both children) was 100% for monozygotic twins, but only 64% (7/11) in dizygotic twins. In addition, the authors noted 4/11 dizygotic twins had markedly opposing phenotypes, two pairs in which one child had FAS and the other FAE, as well as two pairs in which one child had FAE and the other was apparently unaffected (Streissguth & Dehaene, 1993). This provides clear evidence that even when ethanol dose and time are controlled, genetic differences can alter embryo susceptibility.

Building upon this study, a 2019 publication by Hemingway and colleagues supported and elaborated on the findings of Streissguth and Dehaene (Hemingway et al., 2018). In it, the authors included full (n=27) and half (n=9) siblings along with monozygotic (n=9) and dizygotic (n=39) twins, expanding on the effective “genetic relatedness dose-response” to the following: monozygotic twins (100% shared genetic identity), dizygotic twins (50%), full siblings (50%) and half siblings with a common birth mother (25%). As genetic relatedness increased pairwise FASD diagnosis concordance increased between siblings: monozygotic =100%, dizygotic=66%, full sibling=41%, half sibling=22% (Hemingway et al., 2018). Remarkably, this nearly perfectly recapitulates the findings of Streissguth and Dehaene’s foundational work, despite being conducted nearly 30 years later using the new FASD diagnostic criteria, and with a different and much larger study cohort. Notably, the study also highlighted 4/39 twins whose FASD diagnoses were on opposite ends of the severity spectrum (partial fetal alcohol syndrome (pFAS) vs. neurodevelopmental disorder/alcohol exposed), meaning one child had FAS facial phenotypes with severe central nervous system (CNS) dysfunction, but their twin had only moderate CNS dysfunction without facial dysmorphology. No significant sex differences were observed in either phenotypic severity nor pairwise discordance. Using the following formula, the authors then calculated the percent heritability to estimate the importance of genetic factors vs. environmental factors using the following formula: [((dizygotic discordance—monozygotic discordance)/dizygotic discordance) *100%] for both the FASD diagnosis and FAS facial dysmorphology rank, finding both were 100% (Hemingway et al., 2018). This means embryo genotype fully explained the differential outcomes observed between twin pairs. These results finely demonstrate dose and timing of alcohol exposure alone cannot account for phenotypic variability seen in response to alcohol, and strongly support a model in which embryo genetics alters susceptibility to prenatal alcohol toxicity.

4.1.2. Human gene-disease association studies

Association of genotypic variants to FASD phenotypes has been limited but with notable successes. Genome-wide association studies (GWAS) studies on FASD (as with other individual birth defect syndromes) have proven difficult. This is due in large part to the relatively low prevalence of the syndrome, making enrollment of an adequate number of study participants difficult (Khokha, Mitchell, & Wallingford, 2017). This problem is compounded for genes in which the effect size for disease risk is small. Due to the pleiotropic nature of alcohol and the multifactorial gene-alcohol interactions that are predicted to give rise to FASD, study cohorts must be even larger to encompass enough children with FASD from the same etiologies (i.e., with mutations in the same locus).

4.1.2.1. Alcohol metabolizing genes

For these reasons, the majority of FASD modifying genes identified in humans have been candidate genes, with members of the ethanol metabolism pathway being some of the most likely candidates. Alcohol (ethanol) is first metabolized into the toxic metabolite acetaldehyde by alcohol dehydrogenase (ADH) members, which is then metabolized into the non-toxic metabolite acetate by acetaldehyde dehydrogenase (ALDH) members. Both alcohol and acetaldehyde have been demonstrated to act as teratogens, so damage to the embryo can happen until this final metabolic step. In humans, polymorphisms in these gene families alter the rate of metabolism of either ethanol or acetaldehyde, which is predicted to alter the teratogenic potential of identical alcohol consumption patterns (reviewed in (Crabb, Matsumoto, Chang, & You, 2004; Warren & Li, 2005)). However, the results of these studies are inconsistent.

The first study by McCarver et al. examined 243 African American infant-mother pairs with known alcohol exposure during pregnancy for interactions between alcohol and polymorphisms in ADHB1 (previously called ADH2) (McCarver, Thomasson, Martier, Sokol, & Li, 1997). The ADHB1*3 allele encodes a fast-metabolizing allele and is very common in African Americans (25%) (Thomasson, Beard, & Li, 1995). Importantly, this study did not correlate genotypes to FAS facial phenotypes, but instead used as the endpoint the Bayley Scales of Infant Development Mental Index (MDI) at 1 year of age. The authors discovered that the ADHB1*3 allele in both offspring or the mother appeared to protect the child from alcohol-induced deficits.

However, the opposite result was found 5 years later by Stoler et al. The study examined 404 pregnant women and 139 infants (Stoler, Ryan, & Holmes, 2002). The authors found that a higher frequency than expected (46%) of African American women enrolled in the study had the AHDB1*3 allele. In contrast to the previous findings by McCarver et al., Stoler et al. observed maternal ADHB1*3 mutations were over-represented in children with FAS facial phenotypes and growth restriction after adjusting for confounding variables (McCarver et al., 1997; Stoler et al., 2002). While the authors hypothesized this may be due to increased drinking in the mothers with ADHB1*3 polymorphisms, studies have shown that ADH1B*3 polymorphisms are negatively-associated with greater drinking in humans (Zaso et al., 2018). In addition, the fact that infant ADHB1*3 was also over-represented in children with FAS facial phenotypes implies there could be another mechanism, such as increased acetaldehyde accumulation. However, as noted, another major difference between these studies was the endpoint measured—McCarver assessed a neurobehavioral endpoint, while Stoler assessed FAS facial phenotypes.

These differences in study endpoints were reconciled in a follow-up of the McCarver study in 2004. Das et al. photographed the faces of 247 African American infants (173 of which (70%) participated in the original McCarver et al. 1997 study) and their mothers to assess genotype-FAS facial phenotype correlations, as was done by Stoler and colleagues. After selecting only mothers who reported drinking prior to the first prenatal visit, the authors reported that all groups in which the mother or the child had at least one ADH1B*3 had significant (p <0.05), improvements in FAS facial phenotypes compared to infant-mother pairs with no alleles of ADH1B*3 (Das, Cronk, Martier, Simpson, & McCarver, 2004). This result is consistent with ADH1B*3 conferring protection. Thus, the differences between these studies is difficult to ascertain, but could relate to the accumulation of acetaldehyde.

A final study by Jacobson et al. in 2006 also analyzed ADH1B*3 genotype-FAS facial phenotype correlations (Jacobson et al., 2006). 263 African American mother-child pairs were analyzed both as infants at 1 year and at 7.5 years old using a battery of neurobehavioral assays, as well as several morphological endpoints (including head circumference). Maternal ADH1B*3 polymorphism was found to be associated with unaffected head circumference, consistent with a protective effect (Jacobson et al., 2006). In addition, the study confirmed McCarver et al.’s finding that maternal ADH1B*3 is associated with improved Bayley MDI neurobehavior performance at 1 year (Jacobson et al., 2006; McCarver et al., 1997). In contrast, offspring ADH1B*3 polymorphism did not appear protective for Bayley MDI, but did appear protective against reduced head circumference at 1 year (Jacobson et al., 2006).

Together these studies support a role for the fast metabolizing ADH1B*3 polymorphism in protecting against the deleterious effects of prenatal alcohol exposure. However, there are conflicting results that suggest a more complicated or multiple mechanisms. Additionally, the allele is associated with decreased alcohol consumption, which confounds the association. Thus, while most results support a protective function for the maternal ADH1B*3 polymorphism, the mechanism is not known.

Next, a study by Viljoen et al. examined a slow-metabolizing allele of ADHB1, ADH1B*2 (Viljoen et al., 2001). The study included 56 children with FAS, their mothers, and 172 controls from the same region in South Africa. The participants were of a Khoisan-White mixed-ancestry. The authors found the ADH1B*2 mutation was more common in control mothers and offspring, consistent with the mutation conferring protection from alcohol-induced malformations (Viljoen et al., 2001). In addition, ADH1B*3 was examined, but the frequency of this polymorphism was low and not significant between groups. A notable difference between this study and the two previous is the ethnicity of the participants, which explains the low frequency of ADH1B*3 polymorphisms.

One final study by Boyles et al. from 2010 examined the effect of polymorphisms in another ADH gene, ADH1C (formerly ADH3), on orofacial clefting (OFC) (Boyles et al., 2010). This study came on the heels of a clinical trial in France that reported a 1.70 increased odds ratio for OFC in children born to mothers who reported first trimester drinking, which was reduced to 0.71 or 0.63 for children with ADH1C single or dual allele polymorphisms, respectively (Chevrier et al., 2005). The larger study by Boyles et al. included 483 infants with OFC and 503 controls in Norway. The authors had previously examined 573 infants with OFC and 763 controls born in Norway and found an increased odds ratio of 2.2 for cleft lip with or without cleft palate (CL/P) and 2.6 for cleft palate only (CP) in children born to mothers who reported first trimester binge drinking (>5 drinks per setting) (DeRoo, Wilcox, Drevon, & Lie, 2008). Odds ratios were increased to 3.2 for CL/P and 3.0 for CP with three or more first trimester binge episodes. Boyles et al. found an odds ratio for OFC of 2.6 in children born to mothers who reported first trimester binge drinking, but this increased odds ratio was only observed when the mother or child had the ADH1C haplotype associated with reduced alcohol metabolism (odds ratio=3.0). Conversely, when both the mother and child had the ADH1C haplotype associated with increased alcohol metabolism the observed odds ratio for OFC was 0.9, consistent with this rapid metabolizing isozyme being protective against alcohol-induced OFC. This finding helps contextualize the positive association found by DeRoo et al., as most studies examining a connection between maternal drinking and OFC have been negative (meta-analyzed in (Yin, Li, Li, & Zou, 2019)). A high presence of ADH1C slow-metabolizer polymorphisms would explain this apparent discrepancy, as was seen by Boyles and colleagues.

The collective findings on the effects of ADH polymorphisms are somewhat difficult to reconcile. Both slow and fast metabolizing polymorphisms show a protective effect, so clearly the kinetics of ADH is not the entire story. ALDH is required to convert acetaldehyde to acetate, but these studies did not analyze this second enzyme. Additionally, the overall rate of metabolism is also controlled by the level of expression of these enzymes, meaning metabolism rates can vary even between individuals with the same genotype (O’Connor, Morzorati, Christian, & Li, 1998). ADH expression level can be altered by alcohol consumption (Badger et al., 2000). While it seems very likely that enzymes involved in metabolizing alcohol would impact the outcome of an alcohol exposure, the best predictive value likely requires information regarding genotype and epigenetic modifications at multiple loci.

4.1.2.2. Chromosome architecture genes

Two genes involved in chromosome architecture and/or cohesion, SMC2 and MLLT3, have also been associated with alcohol-induced cleft palate (Beaty et al., 2011). Interestingly, several chromosome cohesion complex genes were recently associated with a brain malformation, holoprosencephaly (Kruszka et al., 2019), which shares many phenotypic and etiological characteristics with severe FAS (see Hedgehog Signaling section). The GWAS study included 550 case-parent trios of children with cleft palate to identify associations between cleft palate and SNPs either alone or in combination with one of three common environmental exposures (alcohol, smoking, or multivitamin supplementation). While no genome-wide SNP associations were identified when examined in isolation (i.e., without consideration of maternal exposures), when considering cases with known maternal alcohol exposure, significant SNPs were found in SMC2 and MLLT3 (Beaty et al., 2011). These data support gene-alcohol interactions in the genesis of human birth defects.

Notably, a study in mouse also points to chromosome stability as an important modifier of prenatal alcohol-induced defects. Fancd2 mutations in humans are associated with Fanconi anemia, a genetic disorder that causes genomic instability, leading to developmental defects and other abnormalities (Langevin, Crossan, Rosado, Arends, & Patel, 2011). Langevin et al. generated mice with dual mutations in Fancd2 and the acetaldehyde metabolizing enzyme Aldh2 (Langevin et al., 2011). Mice with mutations in these genes were exposed to 5g/kg alcohol on gestational day (GD)7.5, targeting gastrulation (2–3 weeks in humans). Eye malformations and exencephaly were frequently observed in dual Fancd2; Aldh2 mutants. More rarely, these malformations were also seen in Fancd2 null; Aldh2 heterozygotes. No eye malformations or exencephaly were observed in other genotypes, demonstrating Fancd2 and Aldh2 have protective roles in alcohol pathogenesis. This result suggests that there are likely to be complex multifactorial interactions in the genesis of FASD in humans.

4.2. Animal model studies

4.2.1. Strain differences

The generation of genetically- and phenotypically-distinct animal strains by animal breeders has provided a powerful tool for assessing the contribution of genetics in a disease’s pathogenesis. These studies provide some of the best evidence for a genetic component to alcohol susceptibility, but so far have yielded few specific targets. Nevertheless, these foundational studies help explain the variability observed in FASD by suggesting genetics can strongly influence susceptibility. These include studies in mouse, chicken, and zebrafish.

4.2.1.1. Mouse

Ogawa et al. were the first to examine differences in alcohol susceptibility between two inbred mouse strains, C57BL/6 and DBA/2 (Ogawa, Kuwagata, Ruiz, & Zhou, 2005). These lines were chosen because of their wide use by alcohol researchers. However, the authors do not designate whether these are the 6J or 6N C57Bl6 substrain (an important distinction discussed below). While B6 mice are enthusiastic alcohol drinkers, DBA mice avoid alcohol. This makes comparison of these lines an attractive model for parsing the detrimental effects of alcohol. The authors used whole embryo culture to directly assess the sensitivity of each strain to alcohol-induced defects in conditions where embryo stage as well as ethanol dose and timing were controlled. The authors removed and cultured B6 and DBA embryos at GD8 (targeting primary neurulation, 3 weeks in humans) with or without 400mg/dL of ethanol for 44 h. B6 mice were found to be more sensitive to alcohol-induced malformations than DBA mice. While optic, forebrain, and total defects did not differ between the two strains, alcohol specifically caused caudal neural tube and heart defects in B6 mice, while it specifically caused hindbrain and somite defects in DBA mice.

A follow-up study the same group compared ethanol sensitivity between C57BL/6N (B6), DBA/2 (DBA), and 129S6/SvEvTac (129S6) (Chen, Ozturk, Ni, Goodlett, & Zhou, 2011). The authors again used whole embryo culture to expose stage-matched embryos to a standardized ethanol dose and time of 88mM ethanol for 6h starting at GD8.25. Media was then replaced and embryos were developed for an additional 42h in ethanol-free media. Again, the B6 strain was found to be most susceptible to alcohol-induced defects, with forebrain, midbrain, hindbrain, heart, optic vesical, caudal neural tube, and hindlimbs severely malformed or developmentally delayed. In contrast, only the forebrains and optic vesicles were malformed or delayed in DBA and 129S6 mouse embryos. Interestingly, these differential outcomes corresponded with higher levels of apoptosis (cleaved caspase 3) in B6 compared to DBA and 129S6 embryos.

A similar analysis was conducted to compare the alcohol sensitivity of five inbred mouse lines: Inbred Short-Sleep (ISS), C57BL/6J (B6), C3H/Ibg (C3H), A/Ibg (A), and 129S6/SvEvTac (129S6) (Downing, Balderrama-Durbin, Broncucia, Gilliam, & Johnson, 2009). Pregnant dams of each strain were administered 5.8 g/kg alcohol or control by oral gavage on GD9 (early pharengeal stage, 4 weeks in humans) and fetuses were examined at GD18 (just before birth). As such, this study could be confounded by differences in gestational timing or alcohol metabolism between the strains. In addition, it’s unclear why the GD9 exposure was chosen rather than the more typical gastrulation/early neurulation-stage exposure. Nevertheless, the authors confirmed the previous finding, that 6J mice are most susceptible to alcohol-induced malformations. In addition, they found the 129S6 line was the most resistant to alcohol-induced defects, with the remaining lines, A, ISS, and C3H showing moderate ethanol sensitivity, ordered by decreasing sensitivity.

While C57BL/6 mice appear highly sensitive to ethanol across these studies, the C57BL/6 strain has more sensitive and less sensitive substrains (Green et al., 2007). Two closely related mouse lines, 6J (maintained by Jackson Labs) and 6N (maintained by Envigo (formerly Harlan Labs)) have been maintained separately since 1951, leading to genetic drift and substrain divergence. Pregnant 6J or 6N dams were exposed to 2.9 g/kg ethanol 4h apart at GD7 to specifically target gastrulation and development of the brain and face. The authors found that the 6N substrain was less susceptible to alcohol-induced face defects than 6J embryos. Interestingly, the authors found 6J mice could be protected by pre-exposure to a peripheral benzodiazepine receptor (PBR) agonist PK-11195, which had previously been shown capable of protecting embryos from teratogen-induced defects (Charlap, Donahue, & Knudsen, 2003; O’Hara et al., 2003; O’Hara, Charlap, Craig, & Knudsen, 2002). In contrast, PK-11195 treatment increased ethanol sensitivity of 6N embryos. Thus, important genetic background differences between mouse strains as well as the 6N and 6J substrains must exist to explain these differences in ethanol sensitivity.

Two notable genomic variations have been identified through genome-wide comparison of the 6J and 6N substrains—mutation of Nicotinamide nucleotide transhydrogenase (Nnt) in 6J (Ronchi et al., 2013) and Crumbs Cell Polarity Complex Component 1 (Crb1) in 6N (Mattapallil et al., 2012). Crb1 has a role in eye development, with the specific mutation Rd8 linked to retinal degeneration. Nnt is a ubiquitously expressed gene involved in redox reactions, and therefore may mediate clearance of alcohol-induced reactive oxygen species (ROS). Transcriptomic analyses comparing microdissected anterior poles of GD7 embryos across the 6J and 6N substrains have identified 80 genes with basal differential expression (Boschen, Ptacek, Berginski, Simon, & Parnell, 2021). The authors found 67 genes with higher expression in 6J and 13 with lower expression in 6J. As expected, one gene with lower expression in 6J was Nnt. These analyses confirm the reduction of Nnt expression in the 6J substrain, making differences in ROS clearance a possible mechanism explaining the differential ethanol sensitivity.

ROS have often been implicated in FASD pathogenesis (reviewed in (Brocardo, Gil-Mohapel, & Christie, 2011)). For example, maternal mutations that either increase or decrease superoxide dismutase (SOD) activity are protective or sensitizing in mice, respectively (Wentzel & Eriksson, 2006). Furthermore, interventions that reduce ROS can mitigate alcohol-induced malformations (reviewed in (Zhang, Wang, Li, & Peng, 2018)). For example, administration of the anti-oxidant N-Acetyl Cysteine (NAC) rescued alcohol-induced eye defects in mice exposed to two doses of 2.9 g/kg ethanol on GD7 (Parnell, Sulik, Dehart, & Chen, 2010). While 300mg/kg NAC partially, but not significantly, reduced eye defects, 600mg/kg NAC significantly reduced alcohol-induced eye defects, demonstrating a partial rescue of alcohol teratogenicity.

Our lab has used CRISPR-Cas9 to generate an nnt mutant zebrafish line to directly interrogate the role of nnt in modifying alcohol susceptibility (Mazumdar, unpublished). Consistent with the mouse substrain differences, nnt mutant zebrafish were sensitized to alcohol induced malformations at a normally sub-teratogenic 1% ethanol dose (in media). Unpublished data suggest nnt mutants accumulate more ROS in response to alcohol, leading to increased apoptosis in the brain and pharyngeal arch neural crest cells (Mazumdar, unpublished). In sum, these data support a model in which 6J mice (with Nnt mutation) have increased alcohol sensitivity due to a diminished capacity to tolerate alcohol-induced oxidative stress. The genes that modify ethanol teratogenicity across other mouse strains remain to be determined.

4.2.1.2. Chicken

Debelak and Smith examined 11 chick strains for differences in susceptibility for alcohol-induced neural crest apoptosis (measured by acridine orange uptake) (Debelak & Smith, 2000). The authors demonstrated that cell death was variable based on genetic strain between the lines, with three very sensitive (Babcock ISA, HyLine W98, Babcock B300/Hampshire Red cross (BxHR)), three moderately sensitive (Spafas, HyLine W36, Babcock B300), and five ethanol non-responsive lines (DeKalb White and Black, Shaver White and 2000, DcKalb White/Hampshire Red cross). More detailed interrogation of the two most sensitive lines (W98 and BxHR) compared to the most resistant line (DeKalb White) showed that no differences in embryo staging/timing of alcohol delivery, nor dose of ethanol reaching the embryo explained the differential vulnerability to alcohol (Debelak & Smith, 2000).

These authors published a second analysis in which nine of the 11 chicken strains were assessed for differences in craniofacial outcomes following the difference in cell death they had observed (Su, Debelak, Tessmer, Cartwright, & Smith, 2001). The authors observed three categories of facial malformations in the strains: midfacial flattening (BxHR, ISA-Babcock, HyLine W98, and HyLine W36 strains), facial expansion (Spafas and Babcock B300), or facial hypoplasia (DeKalb White and Black). However, level of cell death was not directly predictive of facial dysmorphology, suggesting more complex compensatory mechanisms exist (Su et al., 2001). Nevertheless, these studies provide excellent evidence for genetic differences being capable of altering an embryo’s sensitivity to alcohol-induced malformations.

Interestingly, similar to mouse, substrains differences exist in chicken. Garic et al. further subtyped the highly sensitive W98 White Leghorn strain, into W98S (from Spencer, IA) and W98D (from Dallas Center, IA) substrains (Garic, Berres, & Smith, 2014). W98D was derived from W98S and separated by at least 10 generations. W98D was selectively bred for egg production. W98S and W98D embryos were exposed to an alcohol dose-response before apoptosis was measured (using acridine orange), confirming the differential sensitivity between the two lines. Next, gene expression differences were examined in the head folds of unexposed W98D and W98S embryos with RNAseq to examine possible genetic mechanisms underlying ethanol vulnerability. These studies identified proteasome and protein synthesis (e.g., ribosome biogenesis) as well as oxidative phosphorylation as genetic networks with differential expression between the two substrains. The authors went on to assess each substrain’s differentially expressed genes following alcohol exposure (Berres, Garic, Flentke, & Smith, 2017), identifying largely the same pathways that were basally differentially expressed between the substrains, including ribosome, oxidative phosphorylation, and spliceosome. Three ribosomal genes were functionally tested using zebrafish. Zebrafish injected with rps3a, rpl5a or rpl11 morpholinos had increased alcohol mediated apoptosis and craniofacial malformations (Berres et al., 2017).

4.2.1.3. Zebrafish

Several analyses of differential sensitivity to ethanol have been performed in zebrafish. Loucks et al. examined three strains of zebrafish for differences in ethanol sensitivity. These included Ekkwill (EK) (ZFIN ID: ZDB-GENO-990520-2), AB (ZFIN ID: ZDB-GENO-960809-7), and Tuebingen (TU) (ZFIN ID: ZDB-GENO-990623-3) strains (Loucks & Carvan, 2004). Embryos were subjected to a dose-response of alcohol to observe differences in cell death (via acridine orange) or survival. The authors found that the EK strain was the most resistant to alcohol-induced embryo lethality, but surprisingly showed the highest cell death in response to alcohol. AB strain embryos showed moderate lethality and cell death. Finally, TU strain embryos had the highest embryo lethality but the lowest cell death. Notably, ethanol uptake between two of these strains, AB and TU has been shown to be the same (Lovely, Nobles, & Eberhart, 2014). However, the differences in cell death observed in this study were modest. For example, at 100mM ethanol, dead cell counts in the head were EK=2.28±0.25, AB=1.70±0.20, and TU=1.27±0.13, deviating by only about 1 dead cell between groups. However, the differences in embryo survival were much more robust, with day six LD50 concentrations of EK=214mM, AB=100mM, and TU=6.5mM, clearly demonstrating differences in embryo susceptibility between these strains. In addition, differences in the specific craniofacial malformations observed differed by strain. Substrain differences in ethanol response in zebrafish have not been carefully analyzed, although such differences are known to exist based on copy number variant analyses (Brown et al., 2012).

Collectively, these data on strain differences in alcohol sensitivity support a model in which genetic factors influence ethanol vulnerability. In some instances, these differences have led to the identification of genes modulating the sensitivity to ethanol teratogenesis. More often, these differences allow researchers to study the effects of ethanol in a sensitized genetic background. The preponderance of our understanding of the genetic modulation of ethanol teratogenesis come from candidate gene and forward genetic approaches.

4.2.2. Forward genetics

In forward genetics, random mutations are generated and individual mutant lines are screened for a phenotype of interest. After identifying interesting phenotypes, the mutation is genetically mapped to identify the causative mutation. Due to the use of random mutagenesis, forward genetics is inherently unbiased, facilitating discovery of novel genes or novel functions of known genes. The downside of forward genetics, at least historically, has been the relatively arduous job of genetic mapping. However, improved methods of genetic mapping using whole genome sequencing and innovative transposable elements that result in fluorescence following mutagenic integrations (Clark et al., 2011; Obholzer et al., 2012), make this approach more time efficient.

Among vertebrate models, zebrafish are highly amenable to forward genetic screens given their external development and fecundity. Given the variety of FASD-related phenotypes, such screens could expand our understanding of the genetics modulating the outcomes of ethanol exposure. To date, one such screen of N-ethyl-N-nitrosourea (ENU)-mutagenized zebrafish has been published (Swartz, Lovely, McCarthy, Kuka, & Eberhart, 2020). From 126 mutagenized families, 5 mutations that disrupted craniofacial development were identified. The majority (4/5) only had abnormal phenotypes in the presence of ethanol. One mutation was genetically mapped and found to be a disruption in an uncharacterized gene (Swartz et al., 2020). This finding supports the utility of forward genetics in our understanding of FASD.

4.2.3. Reverse genetics, candidate-based studies

The other side of genetic screens is reverse genetics. Here, targeted mutagenesis is used to create loss-of-function mutants for known genes of interest. These studies benefit from being faster, but are more biased, as the researcher must already have genes of interest identified. Despite this, reverse genetic approaches have provided important insights into the genetics of ethanol teratogenicity. There are several examples of reverse genetics successfully identifying ethanol-sensitizing mutations, particularly in mouse and zebrafish.

4.2.3.1. Retinoic acid signaling

One of the first major developmental signaling pathways to be linked to modifying alcohol teratogenesis was retinoic acid (RA) signaling. RA signaling is critical in embryogenesis. RA, a metabolite of retinol (Vitamin A), is involved in patterning the anterior-posterior axis of the embryo, and RA signaling is perturbed by prenatal alcohol exposure (reviewed in (Fainsod, Bendelac-Kapon, & Shabtai, 2020; Shabtai & Fainsod, 2018)).

Analyses in both zebrafish and Xenopus, support a role for RA in ethanol teratogenesis. Marrs et al. found that exposing zebrafish embryos to 100mM alcohol from 3 to 24h post fertilization (hpf ) (blastula through pharyngula) resulted in a variety of severe craniofacial defects (Marrs et al., 2010). However, co-treatment of embryos with the same dose of alcohol plus supplemented RA (1nM) resulted in a remarkable rescue of alcohol-induced craniofacial defects. In Xenopus embryos, defects caused by high dose of alcohol (2% in media) could be phenocopied in embryos co-exposed to a low dose of alcohol (1% in media) combined with 20μM 4-diethylaminobenzaldehyde (DEAB), an inhibitor of retinaldehyde dehydrogenase (Raldh). The authors then overexpressed Raldh2 via RNA microinjection and exposed these embryos to high dose alcohol. Consistent with previous work on direct RA supplementation, Raldh2 overexpression rescued alcohol-induced defects (Kot-Leibovich & Fainsod, 2009). Together, these data suggest that susceptibility to alcohol-induced teratogenesis can be influenced by RA signaling status.

4.2.3.2. Hedgehog signaling pathway

Another genetic pathway with ample supportive evidence for a role in modifying ethanol susceptibility is the Sonic Hedgehog (Shh) signaling pathway. Shh signaling is required for brain and face development in vertebrates. Single allele mutations in the gene SHH, which encodes the critically important pathway ligand, or GLI2, encoding a downstream transcriptional activator of the pathway, are the most common causes of a severe midline brain malformation called holoprosencephaly (HPE). Another gene, CDON, is an important SHH co-receptor in the pathway. Interestingly, like FASD, HPE has a complex etiology that is thought to be caused by mixed genetic and environmental interactions (Addissie et al., 2020; Beames & Lipinski, 2020; Grinblat & Lipinski, 2019; Hong & Krauss, 2018; Roessler et al., 2003; Roessler & Muenke, 2010). Furthermore, some of the midline brain and face dysmorphologies observed in HPE are similar to malformations seen in FAS, including agenesis of the corpus callosum (Godin, Dehart, Parnell, O’Leary-Moore, & Sulik, 2011; Hong & Krauss, 2017; O’Leary-Moore, Parnell, Lipinski, & Sulik, 2011; Riley et al., 1995; Wozniak et al., 2009; Yang et al., 2011, 2012). Finally, alcohol has been shown to disrupt Shh signaling (Boschen, Ptacek, et al., 2021; Li et al., 2007). Thus, links between Shh signaling and alcohol in birth defects are extensive.

Similar to work with RA, several studies have rescued the effects of prenatal alcohol exposure by exogenously activating the Shh pathway. This has been accomplished using genetic mutants (e.g., Ptch1 mutants) (Hong & Krauss, 2013), Smoothened Agonist (Burton et al., 2022), cholesterol (which indirectly augments Shh signaling (Li et al., 2007)), and retroviral infection (Ahlgren, Thakur, & Bronner-Fraser, 2002). Care must be taken when interpreting these results, as some approaches more accurately recapitulate normal biology than others. For example, Ptch1 mutants or SAG treatment would be expected to more accurately maintain normal Shh paracrine signaling (Cordero et al., 2004; Eberhart, Swartz, Crump, & Kimmel, 2006; Marcucio, Cordero, Hu, & Helms, 2005). However, other approaches such as retroviral infection ectopically express Shh in the neural crest itself (Ahlgren et al., 2002). Given the pro-survival and growth functions of Shh signaling, such ectopic expression may not accurately model the underlying biology.

Direct genetic evidence for an interaction between Shh and alcohol focused on the genes Shh and Gli2 and comes from analyses in mouse and zebrafish. Heterozygous Shh and Gli2 mice are phenotypically indistinguishable from their wildtype littermates. However, ethanol exposure of Shh and Gli2 heterozygous mice revealed a latent haploinsufficiency, wherein Shh and Gli2 heterozygotes displayed more frequent malformations than siblings exposed in utero to the same ethanol dose (Kietzman, Everson, Sulik, & Lipinski, 2014). Similarly, zebrafish heterozygous for shha mutations (a zebrafish paralog of Shh) are sensitized to ethanol-induced midfacial defects (Everson, Batchu, & Eberhart, 2020).

While there is strong genetic evidence for Shh-alcohol interactions, whether these interactions are direct or not, is unclear. Cdon is a known co-receptor for the Shh pathway. Homozygous Cdon mutations are typically well tolerated and cause no overt malformations, however, Hong et al. discovered Cdon mutants were sensitized to alcohol induced malformations (Hong & Krauss, 2012). Loss of the negative regulator of Shh signaling Patch1 (Ptch1), elevating Shh pathway activity, rescues ethanol-induced defects in Cdon mutants (Hong & Krauss, 2013). More recently, this group has determined that the Cdon-ethanol interaction is through disrupted Nodal signaling, operating upstream of Shh signaling (Hong, Christ, Christa, Willnow, & Krauss, 2020). Among the evidence for an indirect interaction, they found that heterozygosity for Lefty2, a nodal pathway inhibitor, and Tdgf1, the Nodal co-receptor, suppress and enhance the effects of ethanol on Cdon mutants, respectively. Together these series of experiments nicely demonstrate the role of Shh signaling in altering an embryo’s susceptibility to ethanol-induced teratogenesis.

4.2.3.3. Ciliary genes

Phenotypes in FASD can also mirror those in ciliopathies (Schock & Brugmann, 2017), providing additional candidate genes. The ciliary genes Kif3a and Mns1 have each been independently shown to modify alcohol teratogenesis (Boschen, Gong, Murdaugh, & Parnell, 2018). Mns1 is required in motile cilia (Zhou, Yang, Leu, & Wang, 2012). Mns1, heterozygous mice were inbred to produce offspring of wildtype, heterozygous, and null embryos, which were exposed in utero to two doses of 2.9 g/kg ethanol at GD7 (gastrulation, 2–3 weeks in humans), the critical period for severe fetal alcohol syndrome (FAS) (Lipinski et al., 2012; Sulik, 1984; Sulik, Johnston, & Webb, 1981). While alcohol-exposed wildtypes had an incidence of eye defects of 41%, Mns1 heterozygotes and homozygous nulls had incidences of 64% and 92%, respectively. In addition, the most severe defects (scores 5–7 on the authors’ 1–7 severity scale) were only observed in heterozygous or null fetuses (Boschen et al., 2018). These heterozygous and null fetuses with the most severe eye defects also had brain and face malformations linked to FAS, including holoprosencephaly and a long upper lip with diminished upper lip notch. These phenotypes are also remarkably reminiscent of those caused by transient Hedgehog pathway inhibition at a close critical period of development (GD7.75) (Everson et al., 2019; Heyne et al., 2016).

The mechanism of these interactions is hypothesized to be via disrupted cell signaling of pathways like sonic hedgehog or retinoic acid. In the primitive node, motile cilia movements distribute morphogens for these pathways (reviewed in (Ryan & Izpisúa Belmonte, 2000)). Thus, disruption of motile cilia may reduce pathway activity because the morphogens do not reach their proper destinations. Evidence of this mechanism comes from the observation of a high incidence of situs inversus in Mns1 mutants. Situs inversus is a left-right asymmetry defect that can result from disrupted nodal flow (reviewed in (Ryan & Izpisúa Belmonte, 2000)), and has previously been observed in Mns1 mutants (Zhou et al., 2012). However, an alternative or parallel mechanism for the Mns1-alcohol interaction is Mns1 may have a role in primary cilia, which are non-motile cilia that act as hubs of cell signaling (Anvarian, Mykytyn, Mukhopadhyay, Pedersen, & Christensen, 2019). For example, the sonic hedgehog pathway requires the primary cilia for signal transduction (Bangs & Anderson, 2017; Huangfu et al., 2003). Currently, the contribution of each of these mechanisms has not been parsed. In support of a primary cilia model, the same authors more recently showed that mutations in the intra-flagellar transport primary cilia gene Kif3a sensitized embryos to alcohol-induced microcephaly (small head), which was associated with atypical neurobehavior (Boschen, Fish, & Parnell, 2021). The authors concluded that alcohol exposure may cause a “transient ciliopathy,” resulting in abnormal cell signaling and abnormal development.

4.2.3.4. Nitric oxide pathway

Neuronal defects in FASD have led to the Nitric Oxide pathway being explored as a candidate for gene-ethanol interactions. Nitric Oxide (NO) is free radical with many known functions in normal physiology and disease pathology (reviewed in (Bruhwyler, Chleide, Liégeois, & Carreer, 1993)). In mammals, NO is produced from L-arginine by one of three isozymes of nitric oxide synthase (NOS), eNOS, nNOS, and iNOS (reviewed in (Bryan, Bian, & Murad, 2009; Davis & Syapin, 2005)). NO interactions have been described between alcohol and NOS in drinking adults (reviewed in (Davis & Syapin, 2005)). Less is known about the role of NO and NOS during development. However, Karacay et al. demonstrate nNOS null mice exposed to 0, 2.2, or 4.4g/kg alcohol from postnatal days 4–9 (targeting the “brain growth spurt” which occurs during third trimester in humans (Dobbing & Sands, 1979)) display behavioral deficits that are not observed in alcohol-exposed wildtypes (Karacay, Bonthius, Plume, & Bonthius, 2015). Importantly, these behavioral deficits were coincident with neuronal cell death and structural brain malformations (microencephaly, small brain) (Karacay, Mahoney, Plume, & Bonthius, 2015), as had been observed previously (de Licona et al., 2009). Specifically, nNOS null male and female mice exposed to alcohol displayed significant reductions in brain weight at both the 2.2 and 4.4 g/kg dose levels, while wildtype animals showed only a non-significant reduction in brain weight. At the cellular level, combined loss of nNOS and exposure to alcohol resulted in reduced cell numbers in several important brain regions, the cerebral cortex, hippocampus, and dentate gyrus, which was not observed in alcohol-exposed wildtypes (Karacay, Mahoney, et al., 2015).

4.2.3.5. Growth factor and planar cell polarity pathways

The small size and high fecundity of zebrafish allow for a kind of hybrid between forward and reverse genetic approaches, the “shelf screen” or “phenotyping screen.” In this approach, characterized mutant lines are tested for novel phenotypes. In an early shelf screen, five craniofacial mutants were tested for interactions with ethanol that exacerbated the craniofacial defects. Mutations in platelet derived growth factor receptor alpha (pdgfra) were found to sensitize embryos to ethanol-induced malformations (McCarthy et al., 2013). Homozygous pdgfra mutants displayed loss of the ethmoid plate, an analogous structure to the mammalian palate (Swartz, Sheehan-Rooney, Dixon, & Eberhart, 2011). However, pdgfra mutants exposed to a normally sub-teratogenic 1% ethanol dose displayed near complete loss of their facial skeletons (McCarthy et al., 2013). Interestingly, while unexposed pdgfra heterozygotes were indistinguishable from their wildtype siblings, ethanol exposure resulted in 62% of heterozygotes displaying ethmoid plate malformations, demonstrating a clear gene-alcohol interaction. Furthermore, while the defects in unexposed mutants are due to disrupted neural crest cell migration, apoptosis underlies the ethanol-treated phenotype. This highlights the strength of screening approaches, in their capacity to identify strongly synergistic interactions.

A follow-up shelf screen of 20 characterized mutants disrupting normal development was performed to identify gene-ethanol interactions that disrupted facial and/or neural development. Five of these mutations interacted with ethanol: foxi1, hinfp, mars, plk1 and vangl2 (Swartz et al., 2014). Of these, vangl2 displayed midfacial defects ranging from hypotelorism (closely set eyes) to cyclopia (single medial eye). The planar cell polarity gene vangl2 is required for the convergence and extension movements that drive embryogenesis (Yelin et al., 2005) and ethanol has been shown to disrupt convergent extension (Blader & Strahle, 1998). Bioinformatic analysis lead to the finding that this phenotype was due to the mispositioning and downstream perturbation of the Sonic Hedgehog pathway (Sidik et al., 2021).

4.2.4. Non-genetic factors in FASD

A growing body of evidence suggests environmental co-exposures can interact with alcohol to cause malformations at doses that are not hazardous on their own. Many teratogens cause malformations by perturbing important signaling pathways. Given this and the already defined gene-alcohol interactions, it follows that chemicals that perturb the activity of critical pathways or biological processes could also interact with ethanol to cause malformations.

4.2.4.1. Environmental toxicants

Few studies have examined interactions between alcohol and other environmental toxicants. However, the association between alcohol and the Shh pathway has predicted interactions with environmental inhibitors of the pathway. Piperonyl butoxide (PBO) is an EPA-registered pesticide synergist developed in the 1940s that is found in thousands of commercial and household pesticide products, making exposure to PBO nearly ubiquitous (Daiss & Edwards, 2006). Consistent with this, PBO was discovered to be a top 10 contaminant found in indoor dust (Rudel, Camann, Spengler, Korn, & Brody, 2003). PBO was discovered to block hedgehog signaling via smoothened (SMO) antagonism in 2012 (Wang et al., 2012). PBO exposure during early brain and face development dose-dependently caused HPE and associated midfacial defects in mice (Everson et al., 2019). In zebrafish, combined PBO and ethanol co-exposure synergistically caused craniofacial malformations at doses in which either chemical alone did not cause significant malformations (Everson et al., 2020). When this co-exposure was combined with heterozygosity for shha, the interaction was even more dramatic. This directly demonstrates the hazard potential of chemical mixtures, especially in the context of genetic predisposition, and raises the need for examination of chemical mixtures and gene-environment interactions in toxicant safety testing.

4.2.4.2. Cannabinoids

Another co-environmental interaction that unexpectedly appears to center upon the Shh signaling pathway is the interaction between alcohol and cannabinoids. The first direct evidence for an ethanol-cannabinoid interaction came from zebrafish (Boa-Amponsem, Zhang, Burton, Williams, & Cole, 2020). Zebrafish embryos were exposed to alcohol, a cannabinoid 1 receptor (Cb1r) agonist, or both. While single exposures did not alter tank-diving behavior (a fear/anxiety-associated behavior), co-exposure reduced tank diving, indicating increased risk-taking behavior (Boa-Amponsem et al., 2020). Phenotypes were rescued by addition of a Cb1r antagonist.

Around the same time, four cannabinoids, including Δ9-THC and cannabidiol (CBD), were tested and found to cause FASD-like phenotypes which also mimicked Shh mutants in mouse (Fish et al., 2019). Co-exposure of cannabinoids with alcohol caused more frequent malformations than exposure to cannabinoids or alcohol alone. The authors demonstrated that each cannabinoid was capable of inhibiting Shh pathway activity via direct antagonism of Smo (Fish et al., 2019). Cannabinoid-induced defects could be rescued by either addition of a Cb1r antagonist, consistent with the previous study (Boa-Amponsem et al., 2020), or restoration of Shh signaling by injection of Shh mRNA.

Together these studies provide evidence for alcohol interactions, wherein embryos can be chemically predisposed to alcohol-induced defects via co-exposure to Shh disruptors in the form of either cannabinoids or the pesticide synergist PBO.

5. Concluding remarks

What is a safe dose of alcohol? As discussed herein, FASD outcomes can be influenced by each individual’s genetic susceptibility and co-exposures. This susceptibility is determined by a large number of known factors, and probably even more factors that have yet to be discovered. From this, it is clear that it is impossible at this time to define a dose of alcohol that is not potentially hazardous during pregnancy.

Abbreviations

ADH

alcohol dehydrogenase

ALDH

acetaldehyde dehydrogenase

BCE

before common era

CBD

cannabidiol

CL/P

cleft lip with or without cleft palate

CNS

central nervous system

CP

cleft palate

DEAB

4-diethylaminobenzaldehyde

ENU

N-ethyl-N-nitrosourea

EPA

environmental protection agency

FAE

fetal alcohol effects

FAS

fetal alcohol syndrome

FASD

fetal alcohol spectrum disorders

GD

gestational day

GWAS

genome-wide association study

HPE

holoprosencephaly

NAC

N-acetyl Cysteine

NO

nitric oxide

OFC

orofacial clefting

PBO

piperonyl butoxide

pFAS

partial fetal alcohol syndrome

RA

retinoic acid

ROS

reactive oxygen species

SAG

smoothened agonist

Shh

Sonic hedgehog