Diving into the world of alcohol teratogenesis: A review of zebrafish models of fetal alcohol spectrum disorders

Abstract

Fetal alcohol spectrum disorders (FASD) refer to the entire suite of deleterious outcomes resulting from embryonic alcohol exposure. Along with other reviews in this edition, we provide insight into how animal models, specifically the zebrafish; have informed our understanding of FASD. We first provide a brief introduction to FASD. We discuss the zebrafish as a model organism and its strengths for alcohol research. We detail how zebrafish has been used to model some of the major defects present in FASD. These include behavioral defects, such as social behavior as well as learning and memory, and structural defects, disrupting organs such as the brain, sensory organs, heart and craniofacial skeleton. We provide insights into how zebrafish research has aided in our understanding of the mechanisms of ethanol teratogenesis. We end by providing some relatively recent advances that zebrafish has provided in characterizing gene-ethanol interactions that may underlie FASD.

Introduction

The World Health Organization reports that globally, one of the top five risk factors for disease, disability and mortality is harmful alcohol (ethanol, ethyl alcohol) use (Com 2014). Though classic and biblical texts alluded to the ill effects of alcohol consumption during pregnancy on the developing fetus (see review by Calhoun and Warren 2007), a scientific report on this subject did not arrive until centuries later (Stratton et al. 1996). In 1968, a French team published a set of recurring features found in children born to alcoholic mothers (Lemoine et al. 2003). A few years later in America, Jones and Smith coined the term fetal alcohol syndrome (FAS) to define the characteristic traits exhibited by infants with heavy prenatal alcohol exposure (Jones and Smith 1973). These FAS characteristics include: (a) facial abnormalities, (b) growth retardation and (c) central nervous system defects (Stratton et al. 1996).

Since 1973, it has become clear that FAS is on the severe end of a spectrum of deleterious outcomes resulting from embryonic ethanol exposure. Other described birth defects within this spectrum include: partial FAS, alcohol related birth defects, alcohol-related neurodevelopment disorders, static encephalopathy (alcohol exposed) and neurodevelopmental disorder (alcohol exposed) (Dörrie et al. 2014). Each classification differs based on the type of abnormalities (e.g. structural versus CNS) and the combination of deficits present (fully reviewed by Dörrie et al. 2014). Collectively, all the negative outcomes associated with prenatal alcohol exposure are defined as fetal alcohol spectrum disorders (FASD) (Calhoun and Warren 2007; Elliott et al. 2008; May et al. 2009; Riley et al. 2011; Sokol 2003).

It is likely that prenatal ethanol exposure is the leading contributor to birth defects. A recent meta-analysis suggests worldwide prevalence of FAS is between 0.09% and 0.23% (Popova et al. 2017). These authors estimated that 1 in 67 prenatal alcohol exposures results in FAS (Popova et al. 2017). The prevalence of FASD is strikingly higher. Recently data from the US, South Africa and Italy suggest that FASD has a prevalence rate that ranges between 2 and 5 %(May et al. 2009; 2014). In areas where heavy binge drinking is common, such as rural South Africa, these rates can be as high as 20–28% (May et al. 2017). For comparison, the prevalence of Autism Spectrum Disorder is estimated to be 1.4 to 1.5% in the U.S. (Christensen et al. 2016).

While high, the estimated prevalence of FASD may still be an underestimation. In 2011, in the US, 45% of pregnancies were unintended (Finer and Zolna 2016). Moreover, between the years 2006–2010, 51.5% of women between the ages of 18 and 44 reported consuming at least one drink during the previous 30 days (Centers for Disease Control and Prevention (CDC) 2012). Thus, the combination of a large portion of women of childbearing age consuming alcohol paired with a high volume of unplanned pregnancies could cause difficulties in ascertaining if an exposure occurred (Flak et al. 2013). Other factors such as physicians frequently failing to correctly identify FASD (Rojmahamongkol and Cheema-Hasan 2015); the reluctance to enter a diagnosis of FAS into the official medical records due to its stigma (Sampson et al. 1997); as well as the reliance on self-reports of maternal alcohol during pregnancy may all contribute to an underestimate of FASD prevalence (Lange et al. 2014). While it has been difficult to establish the true prevalence rate of FASD, the negative impact that prenatal alcohol exposure has on individuals is unquestionable.

Individuals with FASD can demonstrate a myriad of physical aberrations. Low birth weight for gestational age, decelerating weight not due to nutrition or disproportional weight to height ratios, are diagnostic growth defects for FAS (Stratton et al. 1996). Craniofacial malformations associated with FASD include: short palpebral fissures, midfacial hypoplasia, smooth philtrum, a thin upper lip vermilion, epicanthal folds, low nasal bridge, minor ear anomalies, short noses, micrognathia and cleft palate (Sampson et al. 1997; Swayze et al. 1997). Numerous other organs are also impacted by prenatal alcohol exposure, including the heart, kidney, liver, gastrointestinal track and the endocrine systems (Caputo et al. 2016). The precise organ systems that are impaired in individuals with FASD are likely due to ethanol exposures that are coincident with important developmental events within that organ.

The brain is the first organ system that develops and, particularly in humans, its development continues well after birth. Not surprisingly then, among the organs known to be affected by prenatal alcohol exposure, the brain is typically the most severely impacted (Caputo et al. 2016). The most commonly reported brain structure that is altered by gestational alcohol exposure is the corpus callosum (Caputo et al. 2016), which contains the fibers connecting the right and left cerebral hemispheres. The cerebellum is also frequently disrupted in FASD (Nuñez et al. 2011). Other alcohol-induced brain malformations include smaller whole brain volumes (Donald et al. 2015), lower cortical volumes (Lebel et al. 2012), impaired neurochemistry and functional connectivity (Donald et al. 2015). Collectively, it is clear that prenatal alcohol exposure can cause wide-ranging defects to the developing nervous system.

Due to the neural defects caused by alcohol, it is unsurprising that prenatal alcohol exposure is a leading cause of mental retardation (Sokol 2003). Individuals with FASD frequently have impaired behavior and/or cognition (Kodituwakku 2007). For example, individuals with FASD commonly fail to account for the consequences of their actions, are indifferent to social cues or lack the appropriate initiative to form and maintain reciprocal friendships (Roebuck et al. 1999; Streissguth et al. 1991). Not only do these social issues directly impact the ability to form social relationships or obtain employment, but they may lead to trouble with the law, inappropriate sexual behavior, suicide and depression (Kelly et al. 2000; Kully-Martens et al. 2011). Moreover, craniofacial malformations, which can be present in FASD, have also been linked to anxiety, low self-esteem as well as economic and relationship difficulty (Bradbury, 2012). Additionally, people with FASD also show deficits across a number of cognitive functions, including learning and memory (please see Kodituwakku 2007 for full list). Understanding the root cause of the various symptoms of FASD is of great importance, a goal that can be aided by animal models.

A variety of animal models (round worms, fruit flies, rats, mice, guinea pigs, chickens, pigs, sheep, frogs, nonhuman primates and zebrafish) have been used to study alcohol teratogenesis (Patten et al. 2014; Wilson and Cudd 2011). While no animal model is likely to recapitulate all aspects of human FASD, concerted efforts across species can aid in a mechanistic understanding of FASD. Mammalian, avian and non-vertebrate models of FASD have contributed a vast amount of knowledge to the field and are reviewed elsewhere (Comeau et al. 2015; Kelly et al. 2009; Kiecker 2016; Patten et al. 2014; Sulik 2005). Research as early as 1910 showed that embryonic alcohol exposure disrupted the development of the brain, face and eyes of the salt-water minnow, fundulus heteroclitus (Stockard 1910), a species commonly used in toxicological research. Despite this early pioneering work, aquatic models of FASD have trailed mammalian and avian models for the latter part of the century. In the current paper, we will review the contribution of a relatively new animal model for FASD research – the zebrafish.

Making fish great again: zebrafish as a vertebrate model of developmental genetics

Beginning in the 1970s at the University of Oregon George Streisinger, along with Charles Kimmel, established the zebrafish as a model for vertebrate genetics and development. George Streisinger was drawn to the zebrafish due to several characteristics (Streisinger et al. 1981). First, zebrafish have a short generation time, with recent advances in rearing resulting in times of approximately 60 days (Aoyama et al. 2015; Mahabir et al. 2013). Second, zebrafish are highly fecund in the laboratory, with a single female being able to lay hundreds of eggs once per week. Third, eggs are externally fertilized, which is particularly relevant for alcohol research as we discuss below. Fourth, the adults are small and hardy. Fifth, the embryos develop rapidly to stages resembling the adult. Collectively, these attributes made Streisinger realize that zebrafish were ideal for mutant screens (Streisinger et al. 1981). The accuracy of Streisinger’s vision culminated in the entire December 1996 edition of the journal Development being dedicated to the zebrafish “big screen”, characterizing thousands of mutants from the Nusslein-Volhard, Driever and Fishman labs.

In addition to traits making zebrafish ideal for genetics, these early investigators found characteristics ideal for studies of development. The clearest advantage for developmental biologists was the optical clarity of the zebrafish embryo and larvae. This made it possible, for instance, to view the axons of Mauthner’s neurons in living zebrafish (Kimmel et al. 1981). As the use of fluorescent tracers came to light, these were rapidly applied to analyses in zebrafish for generating fate maps (e.g Kimmel and Law 1985) and tracking axonal outgrowth in smaller neurons (Myers et al. 1986). The first transgenic incorporation of foreign DNA into zebrafish was characterized in the lab of Monte Westerfield (Stuart et al. 1988). While originally described as a method of mutagenesis, following the characterization of GFP as a transgenic marker (Chalfie et al. 1994) using GFP in generating stable transgenic zebrafish swiftly followed (Amsterdam et al. 1995). Today, tools such as the Tol2kit, originally generated by the Chien lab (Kwan et al. 2007), make the construction of transgenic zebrafish straightforward and simple. This, combined with the zebrafish’s genetic attributes makes it ideal for studies of FASD.

Why use zebrafish to model FASD?

Knowing precisely when an ethanol exposure occurred is a major challenge in understanding FASD. This challenge is virtually insurmountable in humans and is one reason that animal models have been extremely useful in FASD research. Due to external fertilization, the zebrafish, along with Xenopus laevis, is particularly well suited for understanding how the timing and dosage of ethanol exposure relates to teratogenic outcomes. The embryos develop externally, free from any maternal or placental degradation of ethanol. Ethanol is added directly to the fish water, resulting in an equal dose across all embryos, reducing the variation between alcohol-treated subjects (Fernandes et al. 2015a). Therefore, the deleterious effects of ethanol on the organ systems of an embryo can be precisely measured.

Multiple organ systems can be disrupted as an outcome of FASD, notably the brain, face, sense organs and heart. All of these organ systems evolved at, or before, the divergence of vertebrate and invertebrate animals. In vertebrates, all of these organs are also forming during the phylotypic stage, an embryonic stage that is highly constrained and resistant to evolutionary change. Thus, while there are clearly differences between the organs in fish and humans, the building blocks of these organs are conserved at a time when embryos are most sensitive to the teratogenic effects of ethanol. Even beyond the striking early similarities, the structure and function of these important organ systems are well conserved between zebrafish and humans. For instance, zebrafish share a basic brain layout and neurochemistry with mammals (Chatterjee and Gerlai 2009; Tropepe and Sive 2003). Due to this conservation, zebrafish have been successfully used to model human diseases in all of the organs noted above (Blanco-Sánchez et al. 2017; Hall et al. 2014; Houk and Yelon, 2016; Swartz et al. 2011).

In addition to structural similarities, zebrafish display a wide repertoire of complex behaviors relevant to FASD. Individuals with FASD frequently have impairments in social behavior, learning and memory (Kodituwakku 2007; Kully-Martens et al. 2011). Both in the wild and the lab, zebrafish from social groups, a behavior not common among other animal models (Norton and Bally-Cuif 2010). Moreover, zebrafish demonstrate multiple forms of learning which are reviewed elsewhere (Blaser and Vira 2014). They display various forms of memory, including short-term memory (Fernandes et al. 2015d; Jia et al. 2014), relational memory (Gerlai 2017), and episodic-like memory (Hamilton et al. 2016). Zebrafish can thus, provide mechanistic understanding of the neurobehavioral defects in FASD due to the ability to combine behavioral, imaging and genetic analyses.

FASD researchers are increasingly seeking to understand the genetic risk and resilience factors that underlie ethanol teratogenicity. Here again, the zebrafish is a powerful model. Over 70% of human genes have at least one zebrafish homolog (Howe et al. 2013). For human genes known to cause disease, this percentage increases to 82% (Howe et al. 2013). Homologs for several human ligands, such as LIF, OSM and IL-6, cannot be identified via sequence homology, while their receptors have been identified (Howe et al. 2013). This suggests that the predicted level of gene conservation is an underestimate, due to sequence divergence preventing the identification of some homologs. It is worth noting that the teleost lineage, of which zebrafish is a part, has undergone a whole genome duplication, which can result in zebrafish having two paralogs for each human gene (Chen et al. 2009). These paralogs can have redundant or subcompartmentalized functions, which can frequently necessitate examining both genes (Force et al. 1999). All in all, the genetic commonalities between humans and zebrafish makes it possible to use the fish to study human-disease related gene function (Jones and Norton, 2015).

There are numerous genetic tools available in zebrafish to characterize FASD risk and resilience (Driever et al. 1994; Rasooly et al. 2003). For researchers interested in examining specific genes, there are a vast number of mutants are easily available from Zebrafish International Resource Center (ZIRC) (Eisen 1996; Rasooly et al. 2003). The zebrafish mutation project continues to add additional alleles to this list of available mutants. Additionally, the ease of CRISPR-Cas9 in zebrafish makes it possible to mutate virtually any gene of interest (Talbot and Amacher 2014). For those interested in specific phenotypic outcomes, forward genetic screens remain a mainstay of zebrafish genetics (Grunwald and Eisen 2002). Thus, the combination of available resources for genetic analysis and homology with humans make the zebrafish a valuable tool for FASD research.

How much does a fish “drink”?

A major challenge in establishing zebrafish as an accepted model of FASD was determining the concentration of ethanol in the tissue. Most zebrafish FASD studies use levels of ethanol in the fish water that would be lethal if converted directly to a blood alcohol concentration in humans. This suggested that either the physiology of zebrafish was vastly different than human or that the zebrafish embryo could exclude ethanol. Numerous studies have examined tissue levels post alcohol exposure and the initial results of these studies were highly variable (see Lovely et al. 2014), with some studies suggesting only a miniscule fraction of ethanol enters the zebrafish (Ali et al. 2011a; Li et al. 2007; Lockwood et al. 2004).

There is now consensus that the tissue levels of ethanol are approximately 25–35% of the media level, at least for embryos ≤ 24 hours postfertilization (hpf). Using an enzyme-based assay, a level of 31–35% was initially reported by Reimers and colleagues (2004) and later Zhang et al (2011) reported a level of 32%. Flentke et al (2014) used the Analox assay and found tissue levels of ethanol were approximately 35% of the media. Using Headspace Gas Chromatography our group found tissue levels to be approximately 24–37% (Lovely et al. 2014). Tissue levels varied based on strain and age, with older embryos having lower tissue levels (Lovely et al. 2014). Because these studies used three different assays to determine ethanol levels, the type of assay does not explain the differences in the previous reports.

Physiologic and structural differences also seem unlikely to explain the previous discrepancies. Strain dependent effects on ethanol teratogenesis exist in zebrafish (Loucks and Carvan 2004; Mahabir et al. 2013), which could suggest physiological differences in ethanol accumulation. Our group examined tissue levels of alcohol in AB, Tubigen or fli1:EGFP zebrafish backgrounds. We found a small, insignificant difference, with the AB strain of zebrafish appearing to accumulate ethanol to a slightly lower extent than the others (Lovely et al. 2014), at 24 hpf approximately 25% for AB and 35% for the other two. It is worth noting that Reimers et al. (2004) and Zhang et al. (2011) also used the AB strain. The small differences between their results and ours could relate to methodological differences or divergence of AB substrains. In our study, strain specific differences disappear at 48 hpf, with all strains having levels of approximately 15% (Lovely et al. 2014). None of these differences are adequate to explain the large discrepancies in the zebrafish literature, however. The zebrafish develops within a chorion, which could also serve as a barrier to ethanol, generating variability in published results. However, recent results show the chorion has either no (Flentke et al. 2014) or very little (Lovely et al. 2014) ability to exclude ethanol.

A likely reason for the early disagreement between reports is procedural. The tissue levels of ethanol equilibrate to steady state within 5 minutes of adding or removing ethanol (Lovely et al. 2014). Therefore, extended or multiple washes or any other procedural events that lengthen the time between ethanol removal and tissue processing could result in an underestimation of the ethanol concentration. We conclude that, for labs that have not tested their own strains for ethanol uptake, a safe estimate for the partitioning of ethanol into young embryos is 30%.

Zebrafish studies of FASD phenotypes

Behavior and cognition

Due to their social nature, several zebrafish assays have been generated to test the effects of alcohol exposure on social behavior. A live shoal and a test subject can be placed on opposite sides of a transparent divider to measure shoaling behavior as time spent near shoal mates. Zebrafish exposed to 0.12% ethanol from 2 to 9 days postfertilization (dpf) shoal significantly less than unexposed fish (Parker et al. 2014). Zebrafish will also rapidly approach and maintain proximity to a computer-animated shoal (Fernandes et al. 2015c; Qin et al. 2014). Exposure to 1% alcohol from 24 to 26 hours postfertilization (hpf) impairs this behavior in wildtype AB strain of zebrafish at 4 (Fernandes and Gerlai 2009), 8 (Fernandes et al. 2015a) and even 24 months of age (Fernandes et al. 2015b). These studies reveal a dose-dependent decrease in adult social behavior, with even 0.5 and 0.75% alcohol exposures having some effect on shoaling (Fernandes et al. 2015b; 2015a; Fernandes and Gerlai 2009).

Researchers have examined shoal cohesion as an alternative measure of social behavior (Buske and Gerlai 2011; Parker et al. 2014). In these paradigms the distance between members of an individual shoal is quantified. Similar to the results above adult zebrafish of the AB strain exposed to 1% ethanol from 24–26 hpf maintain greater inter-individual distance within a shoal (Buske and Gerlai 2011). Likewise, wild-type zebrafish of the Tuebingen strain exposed to 0.12% ethanol for 7 days cluster significantly less compared to controls (Parker et al. 2014) Collectively, these results demonstrate that either very short exposures or very low concentrations (0.12% would correlate with a blood alcohol concentration well below the legal limit) of ethanol disrupt social behavior across at least two strains of zebrafish.

While FASD is associated with impairments in social behavior and in learning and memory (Kodituwakku 2007), zebrafish work suggests the two can be temporally separated. Embryos from the AB strain of zebrafish embryos exposed to 0.25, 0.5, 0.75 and 1% from 16 to 18 hpf, but not 24–26 hpf, fail to form an association between a visual cue and a food reward (Fernandes et al. 2014). Zebrafish embryos exposed to 0.06% or 0.18% alcohol from 0 to 24 hpf need significantly more trials than control fish to learn that a food reward will be presented on the opposite side from the previous reward (Carvan et al. 2004). Bailey et al found exposing embryos of the AB strain to either 1 or 3% from 8 to 10 hpf or 24 to 27 hpf, overlapping the time when social defects are generated, does not affect habituation learning or spatial discrimination (Bailey et al. 2015). To assess habituation learning an automated tap was delivered to the bottom of cylindrical arenas housing individual fish and the total distance moved 5 seconds before and after the tap was measured (Bailey et al. 2015). Furthermore, a tank with a central starting chamber with openings to a left and right chamber, was used to train fish to select the non-preferred chamber and evaluate spatial discrimination (Bailey et al. 2015). Collectively these results suggest that early neurulation is a sensitive time window for ethanol-induced learning defects.

While the effect of ethanol on social behavior and learning seems consistent, much disagreement is present with regard to ethanol-induced anxiety. The novel tank diving test is a popular test of anxiety in zebrafish (Stewart et al. 2012). In an unfamiliar environment zebrafish initially dive to the bottom but this response dissipates, presumably as the fish become more comfortable and begin exploration (Levin et al. 2007). Parker et al (2014) reported that zebrafish exposed to 0.12% ethanol for 7 days spent significantly more time at the bottom of the tank compared to untreated controls. In contrast, adult fish exposed to 3% ethanol between 24 to 27 hpf were significantly higher in the water column compared to controls (Bailey et al. 2015) suggesting an anxiolytic effect (Baiamonte et al. 2016; Bailey et al. 2015). Other groups have reported that exposure to 1% alcohol from 24 to 26 hours (hpf) had no effect on any measure of behavior of zebrafish to an animated predator (Seguin et al. 2016). It is unclear whether differences in timing, dosage or protocol underlie these different behavioral findings on stress.

Biological responses to stress involve the cortisol system. Embryonic exposure to either 0.12% or 0.3% ethanol from 1 to 9 days postfertilization (dpf) led to a reduction in cortisol levels in response to a stressor, netting and exposure to air (Baiamonte et al. 2015). Given that embryonic alcohol exposure led to a decrease in cortisol levels, these researchers measured several anxiety-related behaviors: (a) the time spent at the edge of the arena, thigmotaxis; (b) the time spent in the bright side of the arena, scototaxis and (c) novel tank diving (Baiamonte et al. 2016). They found that compared to controls, alcohol-exposed fish spent less time in the periphery, more time at the top of the tank and less time in the dark side of the tank (Baiamonte et al. 2016). They discovered that adult fish acutely exposed to anxiolytic Diazepam also spent time in the periphery and more time at the top of the tank, light preference was not tested (Baiamonte et al. 2016). Currently, it is unclear why the results from Parker et al (2014) are in disagreement with aforementioned studies, particularly Baiamonte et al (2016) given that the same strain, and alcohol concentrations were used between the studies. One possible reason is timing; analysis of embryos treated from 1 dpf to 2 dpf would be insightful.

In addition to adult behaviors, young zebrafish (larvae) can be used to examine the effects of gestational alcohol exposure has on behavior. For example, Ali et al exposed embryos of the AB strain to 10% alcohol for one hour and found that exposures initiated at either 31 hpf or 48 hpf led to hypoactivity (Ali et al. 2011a). Furthermore, the c-start is a rapid escape / startle response that very young (2 day old) zebrafish display (Roberts et al. 2011). Exposure to 0.6% ethanol from 5.25 hpf to 10.75 hpf caused an increase in abnormal c-starts and tail speed, suggesting hyperactivity (Shan et al. 2015). Mauthner neurons mediate the c-start and following treatment with 0.6% these neurons have smaller cell bodies and axons (Shan et al. 2015). Future study is needed to explain the opposing outcomes of these studies. However, a likely cause is alcohol concentration, given that relatively low concentrations of alcohol induce hyperactivity while higher concentrations reduce activity (Gerlai et al. 2000; Lockwood et al. 2004).

Collectively, the behavioral findings outlined above demonstrate that low or short exposures can result in altered behavior or learning and memory. This has important implications for our understanding of what a “safe” level of alcohol consumption may be during pregnancy.

Ocular deficits, FASD and Zebrafish

Up to 90% of children with a history of prenatal alcohol exposure have eye irregularities (Strömland 2004), reviewed in detail by (Strömland and Pinazo-Durán 2002). The zebrafish has been used extensively as a model for vertebrate eye development because the eye develops in 2.5 days and by day 3 larvae can see (Dlugos and Rabin 2007). The short development time of the zebrafish visual systems among other characteristics make the zebrafish well-suited for visualizing the effects of prenatal alcohol exposure on the eyes.

Cyclopia is one of the eye defects found in zebrafish embryonically exposed to alcohol. After varying the length and developmental stage of alcohol exposure, Blader & Strahle (1998) found that exposure to 2.4% alcohol during a time window encompassing the late blastula and early gastrula stages causes cyclopia. More recently, Santos-Ledo et al. (2011) used an alcohol dose of 1.5% from 4 – 24 hpf to induce cyclopia in AB zebrafish. Furthermore, Arenzana et al (2006) examined the effect various alcohol doses (0%, 2.4%, 1.5% and 1.0%) had on the eyes of multiple zebrafish strains (AB, EK, GL, and TL). These researchers began the alcohol exposure at the same developmental stage (dome/30% epiboly) as Blader & Strähle (1998), but exposed embryos to alcohol for approximately 19.5 hours instead of 3 hours. Embryonic exposure to 2.4% alcohol caused cyclopia in all strains, while exposure to 1.5% alcohol only caused cyclopia in AB and EK strains (Arenzana et al. 2006). Thus, alcohol-induce eye defects appear to vary based on genetic background.

Microphthalmia or reduced eye size, is considered to be indicative of alcohol insult to the eye (Strömland and Pinazo-Durán 2002). Multiple groups have found that exposure to 1.5% alcohol between 24 and 48 hpf causes microphthalmia in zebrafish (Bilotta et al. 2004; 2002; Kashyap et al. 2007; 2011). Alternatively, Dlugos and Rabin (2007) reported microphthalmia in fish exposed to 0.4% alcohol from fertilization to 3 days post fertilization, while Ali et al. (2011a) reported stage-dependent microphthalmia in 5 dpf fish exposed to 10% ethanol. The difference between these findings is unclear. However, genetic background and dosage are likely explanations.

Besides cyclopia and microphthalmia embryonic alcohol exposure can affect retinal morphology, retinotectal projections and visual function. Ethanol disrupts photoreceptor development in a dose-dependent manner (from 1% to 2% ethanol), and causes hypoplasia of the optic nerve at 2% ethanol (Matsui et al. 2006). Consistent with this finding, exposure to 0.5% alcohol causes a decrease in retinotectal projection areas (Cowden et al. 2012). In addition to these structural defects, Bilotta et al (2002) and Matsui et al. (2006) found that embryonic alcohol exposure affects retinal activity using electroretinograms. There are behavioral consequences to these defects. The optokinetic response and optomotor response track eye and swimming movements in response to a rotating stimulus, respectively (Brockerhoff 2006). Both behaviors are impaired in fish embryonically exposed to alcohol (Matsui et al. 2006; Bilotta et al. 2002). Thus, there is broad agreement across these zebrafish studies that ethanol negatively effects development and function of the visual system.

Cardiac defects

Humans prenatally exposed to alcohol have various congenital heart defects (Sarmah and Marrs 2013). Furthermore, it is estimated that 54% of patients with FAS have alcohol related-heart deformities (Abel 1996), making the heart a common organ disrupted by embryonic alcohol exposure.

Numerous zebrafish studies have described heart defects caused by embryonic ethanol exposure. Li et al (2007) exposed zebrafish to various alcohol doses (0, 0.25, 0.5, 1.0, 1.5 and 2.0% v/v) during two separate developmental windows. Embryonic alcohol exposure between 4.25 and 10.25 hpf modestly increased embryo mortality, but resulted in dose-dependent abnormalities. Embryos exposed to the highest alcohol concentrations had defects including pericardial edema (Li et al. 2007). Alcohol exposure beginning at the 1 to 2 cell stages, for 3 hours, was found to be uniformly lethal, except at the 0.25% concentration (Li et al. 2007). This stands somewhat in contrast to findings of pericardial edema and enlarged hearts in AB strain zebrafish exposed to 0.5% alcohol from 0 to 3 dpf (Dlugos and Rabin 2010). Other reported heart irregularities included increased heart volumes, decreased thickness of the ventricular wall, and decreased basal heart rate (Dlugos and Rabin 2010).

More recent studies have specifically examined the effects of ethanol exposure during specific stages of heart development. Embryos from multiple zebrafish strains, including AB and TL, were exposed to 0.6% or 0.9% alcohol throughout all stages of cardiogenesis or during specific stages of heart development (Sarmah and Marrs 2013). Ethanol exposures during specification of the cardiac mesoderm, from the 2 cell to 20 somite stage, results in upregulation of hand2 and gata5, critical transcription factors in cardiac specification. These alterations were apparently transient, as later heart development appeared normal. Extending the exposure to when the bilateral cardiac precursors fuse into a heart progenitor resulted in disruption of this fusion. Exposures during the period of cardiac patterning, 20 somites to 36 hpf, resulted in defective cardiac chambers and cushions. Even later exposures during cardiac growth, 40 to 56 hpf, resulted in smaller but otherwise normal appearing hearts (Sarmah et al. 2016). Consistent with previous reports, cardiac defects were more severe after chronic exposure (2–48 hpf) compared to the shorter exposure (Sarmah et al. 2016; Sarmah and Marrs 2013). Together these studies demonstrate that the zebrafish heart is vulnerable to embryonic alcohol exposure, as in human beings.

Growth and craniofacial defects, FASD and Zebrafish

Facial abnormalities and growth retardation are common across FASD, and are a requirement for a FAS diagnosis (Stratton et al. 1996). A number of zebrafish models of FASD have demonstrated that embryonic alcohol exposure causes similar physical defects. Ali et al (2011a) found that branchial arch skeletal elements were sensitive to 10% alcohol exposure. Carvan III et al (2004) exposed zebrafish embryos to various alcohol concentrations from 4 hpf to 6 dpf. They found that a number of craniofacial and neurocranial skeletal elements were affected, with the most sensitive elements being eye structures, the otic capsule, ethmoid plate and head width. Interestingly, they did not find a strict dose-dependent relationship between alcohol and affected structures (Carvan et al. 2004). Alcohol-induced craniofacial defects vary based on zebrafish strain. When exposed to various alcohol concentrations from 2 hpf to 6 dpf, AB larvae had significantly more ethanol-induced abnormalities in the dorsal and lateral structures compared to Ekkwill (EK), however EK had more abnormalities in ventral structures (Loucks and Carvan 2004). Together these results demonstrate that the facial defects caused by alcohol in zebrafish vary based on the concentration, timing and genetic background.

Furthermore, embryonic alcohol exposure has also been shown to affect body length in zebrafish. Bilotta et al. (2004) report that exposure to 1.5% alcohol from 6 to 24 hpf lead to a decrease in body length. They also report that the distance between eyes was larger in alcohol treated fish than controls (Bilotta et al. 2004). Similarly, other groups have reported that embryonic alcohol exposure causes a dose dependent decrease in body length (Ali et al. 2011a; Loucks and Ahlgren 2012; 2009). These findings demonstrate that two core FAS symptoms, facial abnormalities and growth retardations, can be recapitulated in zebrafish.

Brain defects

Prenatal alcohol exposure can alter the central nervous system of humans (Cole et al. 2012). Numerous experiments show that the zebrafish CNS is also vulnerable to embryonic alcohol exposure. For example, Buske and Gerlai (2011) as well as Fernandes et al (2015a) found that in adult zebrafish embryonic alcohol exposure led to decreases in whole brain levels of dopamine and its metabolite 3,4-dihydroxyphenylacetic acid (DOPAC). Interestingly, the decrease in dopamine and DOPAC was caused by ethanol treatments that coincided with those impairing social behavior (Buske and Gerlai 2011; Fernandes et al. 2015a). Besides the dopaminergic system, research has also shown that embryonic alcohol exposure causes a decrease in serotonin and its metabolite 5-hydroxyindoleacetic acid (5HIAA) (Buske and Gerlai 2011; Mahabir et al. 2013). More recently Baggio et al (2017) found that embryonic alcohol to 0.1, 0.25, 0.5 and 1% alcohol for 2 hours at 24 hpf decreased glutamate transport in a dose-dependent fashion. For at least serotonin, the ethanol-induced neurotransmitter changes are strain dependent, with alcohol-induced neurotransmitter defects in the AB, but not the Tubigen, strain of zebrafish (Mahabir et al. 2013). Thus, several major neurotransmitter systems are altered by embryonic alcohol exposure, suggesting widespread neural dysfunction.

Motor neuron dysfunction is also common in FASD. Using the motor neuron labeling transgenic line isl1:GFP, Joya et al. (2014) determined that 1% alcohol from 10 to 24 hpf caused defects in motor neuron axon branches. Sylvain et al. (2010) exposed embryos from 8 to 24 hpf to various alcohol concentrations. They found that fish exposed to 1.5% alcohol or higher were smaller and displayed impaired swimming responses after being touched (Sylvain et al. 2010). The morphology of the ventral and dorsal neuron axons in these fish had abnormal curving, branching and angles of projection (Sylvain et al. 2010). Fish treated with 2 or 2.5% alcohol had significant aberrations in both primary and secondary motor neurons, labeled with znp-1 and zn-8, respectively (Sylvain et al. 2010). These researchers also determined that muscle fiber length; width and angle were affected by embryonic exposure to 1.5, 2 or 2.5% alcohol (Sylvain et al. 2010). Based on patch clamp analyses, however, only fish exposed to 2.5% alcohol displayed altered neurophysiology (Sylvain et al. 2011).

Transgenic analyses have helped characterize how ethanol disrupts the CNS. In both the central and peripheral nervous systems, elavl3:Kaede and isl3:GFP mark differentiated neurons and sensory neurons, respectively (Kim et al. 1996). Exposure to 1% ethanol from 10 to 24 hpf does not disrupt the pattern of neuronal differentiation along the spinal cord, but decreases the number of differentiated neurons and sensory neurons (Joya et al. 2014). Proliferating neuronal progenitor cells express the transcription factor, Neurogenin1 (neurog1) (Ma et al. 1996). Ethanol exposures disrupting the levels of differentiated neurons, did not appear to alter the expression of neurog1:GFP, marking proliferating neural progenitor cells (Joya et al. 2014). Consistent with this, 1% ethanol did not change the expression levels of neurog1 detected by qPCR (Yin et al. 2014). However, the levels of sox2 were increased and neurog1 levels were increased at 2% ethanol (Yin et al. 2014). These findings suggest that ethanol disrupts the terminal differentiation of neurons (Joya et al. 2014).

Numerous transgenics and markers also exist for analysis of specific brain regions. The gsc:EGFP transgenic labels a number of areas, including the preoptic area. The isl1:EGFP and pax2a:EGFP transgenic lines label neuronal populations in the forebrain and the mid-hindbrain boundary, respectively. Oxytocin is associated with social behaviors and the oxt:EGFP transgenic line labels the preoptic area and pituitary (Coffey et al. 2013). Coffey et al. used these transgenics to determine stage specific sensitivities to ethanol. Exposure to 2% alcohol from approximately 8–32 hpf eliminates isl1:EGFP expressing neurons (Coffey et al. 2013). Expression of gsc:EGFP is most sensitive from 24–48 hpf. Similarly, pax2a:EGFP is most sensitive to ethanol exposure from 16–40 or 24–48 hpf. This alcohol-induced defect of the mid-hindbrain is in agreement with other groups using zebrafish (Burton et al. 2017; Zhang et al. 2015; 2013). However, recent studies using a similar ethanol exposure (2% from 6–24 hpf), found relatively minor alterations to the pattern of molecular markers of the brain (Zhang et al. 2017). The patterning of the forebrain, (based on otx2 and pax6a expression) midbrain (based on fgf8a, wnt1 and pax2a), and hindbrain (based on pax6a, epha4 and, krox20) appeared largely normal, with the exception of the expression of epha4a in the rhombomere 1 boundary (Zhang et al. 2017). The expression of oxt:EGFP sensitive time window is from 4–5 dpf. These results suggest that different neuronal populations have different windows of ethanol sensitivity.

Mechanistic insights into FASD from zebrafish

It is unlikely that a single mechanism will account for the teratogenic effects of ethanol given that is a small molecule with many potential pleiotropic effects. Data from multiple animal models, including zebrafish suggest that several mechanisms are involved in ethanol teratogenicity and are conserved across species.

A common mechanism across mammalian and avian models of FASD for alcohol-induced damage is apoptosis. Flentke and colleagues (2014) compared the effects of embryonic alcohol exposure on craniofacial elements of zebrafish and chickens. Alcohol concentrations that induce apoptosis of neural crest cells in mouse and chicken also cause apoptosis in zebrafish. Furthermore, ethanol-induced cell death is calcium and CamKII-dependent, and conserved between zebrafish and chicken (Flentke et al. 2014). This data suggests that a conversed mechanism regulating cell apoptosis exists for the facial abnormalities caused by embryonic alcohol exposure across vertebrates.

Ethanol has also been shown to alter cell adhesion and migration. Blader and Strähle (1998) were among the first groups to report that embryos exposed to 2.4% alcohol during a short developmental window caused cyclopia in zebrafish, and that these changes were due to disruption of the migration of prechordal plate mesoderm during gastrulation. Zhang et al (2010) report that 3% alcohol exposure during similar developmental stages caused a split body axis, often associated with cyclopia. The Wnt/PCP pathway is critically involved in gastrulation events. Not only do mutations in genes acting in the Wnt/PCP pathway such as wnt11 lead to cyclopic phenotypes in zebrafish, but others genes like vangl2 are ethanol sensitive. For example, exposure to 1% ethanol produces cyclopia in vangl2 mutants (Swartz et al. 2013). Furthermore, ethanol exposure in Xenopus, has shown that gastrulation is disrupted by gestational alcohol exposure (Yelin et al. 2005). Therefore, once again a conserved mechanism of ethanol teratogenesis appears to impinge upon cell movements.

In addition to cell movements, ethanol has been widely implicated in disrupting two critical embryonic signaling pathways, retinoic acid (RA) and Sonic Hedgehog (Shh). The normal development of many organ systems is dependent on RA, a Vitamin A metabolite (Zachman and Grummer 1998). Alcohol is believed to act by prohibiting retinaldehyde dehydrogenases from catalyzing the synthesis of RA from retinaldehyde (Duester 1998). Multiple animal models, have shown that embryonic alcohol exposure reduces RA levels and/or ethanol-induced phenotypes can be rescued or partially rescued by RA treatment (Kashyap et al. 2011). RA supplementation protects against alcohol-induced defects to mid-hindbrain development (Zhang et al. 2015). In contrast, Napoli (2011) reports that ethanol can upregulate RA signaling. In zebrafish treated with 0.6% ethanol from 3 to 24 hpf, RA supplementation rescues facial defects (Marrs et al. 2010). However, microphthalmia induced by ethanol exposure is not rescued by RA supplementation (Kashyap et al. 2011; Zhang et al. 2015), although optic nerve and photoreceptor differentiation defects are rescued (Muralidharan et al. 2015). Differences in methodology make direct comparisons between these studies difficult but the mixed results suggest that the mechanism by which RA modulates ethanol teratogenesis is complex. One potential mechanism could be through crosstalk an RA target, Shh, (Zhang et al. 2015).

Work in multiple animal models including zebrafish demonstrate that embryonic alcohol exposure decreases Shh signaling leading to increased cranial neural crest apoptosis, disrupted midline development as well as eye development (Ahlgren et al. 2002; Aoto et al. 2008; Brennan and Giles 2013; Li et al. 2007; Zhang et al. 2013). In support of a role for Shh in ethanol teratogenesis, injection of Shh mRNA was found to rescue ethanol-induced alterations to body length, somite shape and cyclopia (Loucks and Ahlgren 2009). More recently Shh signaling has been shown to be a key contributor to alcohol-induced changes in anxiety and risk-taking behavior (Burton et al. 2017). Mechanistically, it has been shown in zebrafish that embryonic alcohol exposure can attenuate Shh via disruption of cholesterol modification of Shh (Li et al. 2007). Here again, there is consensus across animal models supporting the Shh pathway as a conserved mechanism of ethanol teratogenesis.

Zebrafish and the identification of gene-ethanol interactions

In humans monozygotic twins are 100% concordant for FAS while dizygotic twins are only 64% concordant (Streissguth and Dehaene 1993). Furthermore, strain dependent defects have been reported in mice, chick and zebrafish (Cavieres and Smith 2000; Chen et al. 2000; Dlugos and Rabin 2003; Loucks and Carvan 2004; Su et al. 2001). While, there is strong and growing evidence for genetic risk factors for FASD, only a small number of gene-ethanol interactions have been identified (See Eberhart and Parnell 2016). Due to their genetic amenability (discussed above), the zebrafish is an ideal organism to identify and characterize gene-ethanol interactions.

Due to the evidence that ethanol disrupted Shh signaling, this pathway has been carefully scrutinized for gene-ethanol interactions. Agrin is a heparan sulfate proteoglycan involved in eye development. Using morpholino oligonucleotides (MO) against agrin, Zhang et al (2011) demonstrated that a partial knockdown of agrin expression and low dose ethanol exposure caused microphthalmia, which could be rescued by shh mRNA injection. A 1% ethanol dose also interacts with morpholinos against agrin or shha to reduce markers for GABAergic and glutamatergic neurons, a defect that can be rescued by Shh mRNA injection (Zhang et al. 2013) When combined with studies in other species, these findings provide strong evidence for an evolutionarily conserved interaction between ethanol and the Shh pathway.

While FASD phenotypes have primarily been used as guides for the study of gene-ethanol interactions, the number of available zebrafish mutants allows for an alternative approach. In our lab, we screened five zebrafish mutant lines: smoothened (a Shh pathway member), cyp26b1 (RA catabolizing enzyme), smad5 (Bone Morphogenetic Protein (BMP) pathway member), gata3 (transcription factor) and pdgfra (platelet-derived growth factor receptor) for gene-ethanol interactions in craniofacial development. We found that pdgfra mutants had massive reductions of the entire craniofacial skeleton when treated with 1% ethanol, and over two-thirds of the heterozygous embryos had craniofacial malformations (McCarthy et al. 2013). Our data suggest that these defects were in part at least due to ethanol-induced neural crest apoptosis (McCarthy et al. 2013). This interaction appears to be through the phosphoinositide 3 kinase (PI3K) / mechanistic target of rapamycin (mTOR) pathway (McCarthy et al. 2013). Pdgfra function is conserved between zebrafish, mice and humans (Eberhart et al. 2008; Soriano 1997; Tallquist and Soriano 2013) and analysis of human data identified gene-ethanol interactions with single nucleotide polymorphisms (SNPs) in PDGFRA in human (McCarthy et al. 2013). Therefore, our data demonstrate that zebrafish gene-ethanol interactions may be conserved in humans.

A follow up screen involved 20 mutant lines, obtained from the Zebrafish International Resource Center (ZIRC). We identified five additional ethanol-sensitive mutant lines that had facial and other ethanol-induced defects (Swartz et al. 2013). The identified genes affected by embryonic alcohol exposure covered a wide variety of cellular functions. For example, Mars, a methionine-tRNA synthetase has important biosynthetic function and is crucial for proper translation of mRNA to protein. Mild viscerocranial defects are present in untreated mutants, however these defects are intensified by developmental alcohol exposure (Swartz et al. 2013). Polo-like kinase 1 (Plk1) and histone nuclear factor p (Hinfp) are both involved in various aspects of cell cycle regulation. An entire loss of the viscerocranium, malformations of the neurocranium, reduction in axonal projections and increased apoptosis are evident in untreated plk1 mutants (Swartz et al. 2013). In these mutants, ethanol exposure leads to severe growth retardation, extensive cell death throughout the embryo and complete loss of the craniofacial skeleton as well as axon projections (Swartz et al. 2013). Reduced head and eye size as well as reduction other craniofacial skeletal elements are present in untreated hinfp mutants and these phenotypes are exacerbated by alcohol (Swartz et al. 2013). The correct development of a number of skeletal elements are dependent on Foxi1(Solomon et al. 2003). In untreated foxi1 mutants, a number of craniofacial elements are deformed (Swartz et al. 2013). Embryonic alcohol exposure caused a loss in a number of cartilage elements, aberrant axonal projection and reduced ear size (Swartz et al. 2013). Finally, Vangl2, is necessary for convergent/extension of the body (Marlow, 1998). Shortened body axis, and synophthalmia (at very low penetrance) were evident in untreated vangl2 mutants. Ethanol treated vangl2 mutants had disrupted axon projections, loss of the posterior viscerocranial elements, narrowing of the palate and severe synopthalmia, which were also present in a small percentage of heterozygotes (Swartz et al. 2013). It remains to be seen if the vangl2 or any of the other gene-ethanol interactions are conserved in humans.

Concluding remarks

Prenatal alcohol exposure continues to have a large effect on society, impacting millions of lives. Even though FASD is thought to impact 1 in 100 children born, this rate may still be an underestimate (Abel and Sokol 1991; Guerri et al. 2009; May et al. 2014; Sampson et al. 1997). Furthermore, the deficits caused by prenatal alcohol exposure are life long and severely reduce the quality of life for the affected individuals. Therefore, a complete understanding of the biological mechanisms of FASD is critical to human health.

Multiple animal models have revealed biochemical, molecular and genetic events that lead to FASD (Ali et al. 2011b; Kelly et al. 2009; Patten et al. 2014; S. M. Smith et al. 2014; Sulik 2005). In order to identify potential therapeutic approaches for the devastating effects of prenatal alcohol exposure, studies will need to link animal models to human outcomes. The zebrafish has a high degree of genetic conservation with humans (Howe et al. 2013) and are very amenable for drug screening. Furthermore their high fecundity, external fertilization, embryo transparency, rapid development time and genetic tractability make them a useful model for the study of FASD phenotypes and associated mechanisms, thus this species should greatly accelerate our understanding of FASD(Swartz et al. 2011; Titus et al. 2006; Woods et al. 2005).