The development of the brain is highly orchestrated, coordinated by proper gene expression in response to developmental cues and in response to the environment. It is also a remarkably vulnerable target for a variety of drugs of abuse, particularly ethanol
[32, 33]. This vulnerability leads to changes in brain structure and function, resulting in the long-term behavioral and cognitive changes that characterize FASD
[1, 3, 34–36]. The manifestation and severity of specific FASD phenotypes, however, varies widely across individuals prenatally exposed to ethanol for reasons that are not entirely clear. Further, while it is known that developmental ethanol exposure leads to changes in neural molecular architecture, including changes to genome-wide gene expression
[11, 12, 15]), how these alterations are initiated, maintained over time, and relate to specific FASD-relevant neurological phenotypes remains unknown. The results included in this report focus on the pattern of long-term changes in brain gene expression in adult B6 mice following ethanol exposure at three neurodevelopmental times corresponding to the approximate human equivalent of trimesters one, two, and three.
In a previous experiment, we have determined that the exposure paradigms used in this study (binge-like treatments at developmental days representing early, mid, and late human gestational neurodevelopmental exposure) result in behavioral phenotypes that are relevant to the abnormalities observed in individuals with FASD
. This study seeks to extend these findings by evaluating the genes and their associated pathways that show long-term disruption due to prenatal alcohol exposure. We do acknowledge that the neurodevelopmental timelines in rodents and humans are similar but not identical, particularly the period of synaptogenesis that occurs during the third trimester and extends postnatally in humans but occurs primarily postnatally in mice
[19, 37]. Also, due to these differences, two of the treatment models (ethanol exposures at E8/11 and E14/16) represent maternal treatments whereas mouse pups were administered ethanol in the trimester three model (exposure at P4/7). This may result in differences in total blood alcohol levels achieved, which is a caveat of these methods, although there is evidence that ethanol rapidly crosses across the placental barrier and can appear within the embryo within 5 minutes after maternal treatment
. Regardless of these potential differences in blood alcohol levels achieved, the results obtained using this mouse model may be an effective means to parse the long-term consequences of fetal alcohol exposure to the developing brain
Ethanol-induced long-term changes to gene expression are subtle, multifactorial, and timing-dependent
The clustering of gene expression signal intensities of the arrays representing the control and ethanol-treated samples (Figure
1) suggested that alcohol exposure during neurodevelopment significantly altered the transcriptome of the adult brain, regardless of timing. Maternal care differences were not assessed in this study and although we and others have previously observed that minor, if any, alterations to maternal care occur following gestational alcohol exposure
[40–42], these finding are predominantly based on voluntary maternal consumption paradigms and not binge-like exposures. It is possible, then, that maternal care differences between ethanol-treated and control dams may contribute to the observed gene expression differences. However, there was very little overlap between the genes identified by each of the ethanol-treatment times (Table
2), suggesting that a major contributor to the set of genes affected was the timing of exposure. This may be attributed to the considerable changes that occur to the brain transcriptome across development, resulting in a different repertoire of processes on which ethanol may act and leading to a very different footprint of genes that remain altered into adulthood.
Genes that remain altered by neurodevelopmental alcohol exposure may reflect predominant biological processes occurring at the time of exposure
The results included in Table
1 and Figure
2 show that only a small number of genes were altered by ethanol by multiple treatment models. In general, these genes represent pleiotropically-acting molecules associated with cell death, regulation of gene expression, and, interestingly, methylation. Such results may argue that the most common and timing-independent response to alcohol is cellular stress, potentially leading to apoptosis at the time of exposure. The long-term changes to these genes may suggest that the surviving cells retain a ‘memory’ of exposure and may show altered response to subsequent exposures. This is supported by evidence that prenatal ethanol exposure can result in heightened ‘sensitization’ of adult neurons to ethanol
[43, 44], which may be a contributing factor to fetal alcohol exposure and increased risk for alcohol abuse
. These genes, however, represent only a small proportion of the genes detected, which suggests that the effect of developmental alcohol exposure may be dictated by the interaction between ethanol and the specific biological processes occurring at the time of exposure.
Ethanol exposure during trimester one alters genes associated with tissue morphology
We performed GO analysis on the gene lists generated by each treatment model (E8/11, E14/16, and P4/7) in order to identify the overrepresented biological functions altered in each paradigm. Interestingly, genes affected in each of the three treatments showed relevance to predominant biological processes occurring at the developmental stage of exposure. For example, E8/11 exposure was associated with alterations to genes involved in cellular assembly, proliferation, differentiation, cell death, and tissue morphology (Table
2), even in the adult brain. Developmentally, ethanol-induced disruptions in decisions regarding the ‘accumulation of cells’ and ‘quantity of interneurons’ during early gestation could lead to alterations in the number of cells that exist in certain brain regions, perhaps through inhibition of mitosis or by promoting inappropriate apoptosis, but it is unclear what effect the alteration of these genes may have in the adult brain. The significant functions annotations for trimester one exposure appear to be driven primarily by the altered expression of Neurotrophin 3 (Ntf3). Disruptions in Ntf3 expression during development are detrimental given its canonical role as a neuronal survival factor
[46, 47] and regulator of axonal growth and guidance, synaptic structure, and synaptic plasticity
[48–51]. In addition to having developmental roles, reduced Ntf3 expression in the adult brain has been shown to be associated with deficits in spatial learning, long-term potentiation impairments, and increased anxiety-related traits
[52–54]. Canonical pathways analysis indicated alterations to genes involved in endoplasmic reticulum (ER) stress response and such as the downregulation of DnaJ (Hsp40) homolog, subfamily C, member 3 (Dnajc3), also known as p58IPK. Cells showing decreased expression of this molecular chaperone protein are more prone to ER stress-induced apoptosis
[55, 56]. These results suggest that, beyond inducing cell death upon immediate exposure, early gestational ethanol treatment may lead to increased cellular vulnerability to other insults in surviving adult neurons.
Ethanol exposure during trimester two affects genes involved in cellular differentiation and migration
In contrast to E8/11 exposure, GO analysis of genes affected by ethanol treatment at E14/16 revealed the alteration of genes developmentally involved with cell migration, differentiation, and morphology (Table
2), processes that occur rapidly during this developmental stage in many brain regions
[20, 39, 57]. Physiologically, ethanol exposure during trimester two has been associated with abnormal cell division and increased the appearance of radial-glia-like precursors which dictate neural stem cell migration from the ventricular zone (VZ) to the subventricular zone (SVZ) during early differentiation
[58, 59]. This leads to the depletion of certain susceptible cell types in the VZ and a corresponding abnormal increase of cells in the SVZ and may lead to the cortical heterotopias associated with FASD. Interestingly, the homeobox transcription factors distal-less homeobox 1 and 2 (Dlx1 and Dlx2) as well as doublecortin (Dcx) were identified as altered in the adult brain following E14/16 exposure. Dcx has been previously implicated in alcohol’s effects on both the developing and the adult brain
[60–62], while Dlx1 and Dlx2 are the earliest Dlx genes to be expressed in the SVZ and are critical regulators of interneuron differentiation and migration in the developing telencephalon
[63–65]. They are also responsible for the majority of GABAergic neurons in the mammalian neocortex and are therefore crucial for the regulation of the development of inhibitory neocortical circuitry
[66, 67]. IPA canonical pathway analysis also associated E14/16 exposure with changes to genes involved in serotonin signaling (Table
3). Indeed, three serotonin receptor subunits were identified as altered: 5-hydroxytryptamine (serotonin) receptor 5A (Htr5a), Htr3a, and Htr6. Reduced density of serotonin neurons have been previously associated with prenatal alcohol exposure
[68–70]. Further, low serotonin signaling, including that caused by neurodevelopmental alcohol exposure, has been implicated in a number of phenotypic outcomes relevant to FASD such as increased impulsivity, aggression, hopelessness, exploratory activity, and risk for stress and anxiety-related traits due to altered hypothalamic-pituitary-adrenal axis function
Early postnatal ethanol exposure altered genes associated with synaptic function in the adult brain
Spanning approximately the first 2 weeks of postnatal mouse development, the ‘brain growth spurt’ is a period of extensive synaptogenesis and growth within the cortex, hippocampus, and corpus callosum
[18, 20]. This stage of development, equivalent to approximately the third trimester of human neurodevelopment
[18, 19], determines much of the extent of the brain’s neural circuitry, including the maintenance or pruning of synaptic connectivity in multiple brain regions. Disruptions to these processes, such as by ethanol, may result in inappropriate synapse formation and interfere with synaptic transmission, potentiation, and plasticity in adulthood
[76–79]. GO analysis of P4/7 ethanol exposure suggested that many genes altered in the adult brain by this treatment model were involved in biological processes relevant to synaptic function. A number of glutamate receptor subunits were identified, known to be critical in the formation and maintenance of synapses, synaptic plasticity, and long-term potentiation (LTP)
[80–85]. Also of note is the downregulation of Eph receptor B1 and B2 (Ephb1 and Ephb2), known to have classic functions in axon guidance and brain region boundary formation
 as well as potentially controlling many other aspects of excitatory synaptic transmission and plasticity
[87–89]. Interestingly, ethanol exposure at P4/7 also resulted in altered steroid secretion due to the downregulation of both apolipoprotein E (Apoe) and proopiomelanocortin (Pomc), molecules that are essential for proper hypothalamic-pituitary-adrenal (HPA) axis function. Apoe deficiency is associated with age-related neurodegeneration and cognitive decline
[90, 91]. Phenotypically, Apoe null mice show increased anxiety in the elevated plus maze but reduced activity in a novel open field environment, which coincides with increased plasma corticosterone levels and impaired performance in spatial learning tests
[90, 92]. Contrastingly, Pomc has been shown to be essential for postnatal adrenal maturation, with Pomc null mice showing no production of corticosterone
. Given this, it is certainly possible that long-term alterations in the expression of these genes may contribute to phenotypes such as cognitive abnormalities and increased stress reactivity observed individuals with FASD. The IPA canonical pathway analysis reflected the results of the GO analysis, with genes associated with glutamate receptor signaling and ephrin receptor significantly represented. Genes associated with retinoic acid-mediated apoptosis signaling were also altered, a finding which is supported by other studies that report that prenatal ethanol exposure can lower retinoic ethanol receptor function and elevate retinoic acid levels beyond normal physiological levels
[94–96]. Folate metabolism has also been implicated in FASD phenotypes as folic acid supplementation has been shown to ameliorate some of the effects of prenatal alcohol consumption
[97, 98]. Finally, although it is known that the hypothalamus, including the control of circadian ‘clock genes’, is disrupted by neurodevelopmental alcohol exposure
[14, 99, 100], the identification of circadian rhythm signaling is interesting given that disturbed sleep patterns is a reported but relatively unexplored consequence of fetal alcohol exposure. This may be due to the disruption of corticotrophin-releasing hormone neurotransmission due to abnormal synaptic tract formation, but this potential consequence associated with trimester three ethanol exposure requires further investigation.
Common molecular pathways may mediate the effect of ethanol on neurodevelopment despite execution through trimester-specific gene sets
As a final part of this study, we conducted IPA network analysis initially to evaluate further differences to adult brain gene expression following ethanol treatment at various neurodevelopmental stages. Evaluation of these gene sets, however, revealed numerous similarities between gene networks. While each treatment was associated with somewhat different significant IPA networks, the merged gene networks for each model revealed similar sets of gene ‘hubs’ (Figures
6). While these ‘hubs’ were not necessarily altered themselves, they linked dysregulated genes suggesting that there may be some common processes that remain altered into adulthood following developmental ethanol exposure, regardless of timing. Specifically, glutamate receptor subunits appeared as core molecules in all treatment paradigms, as did many neurotrophic molecules. Also, the developmentally-essential gene Htt was associated with the altered gene networks of each ethanol treatment model. This is, as far as we are aware, unreported in prenatal alcohol exposure literature; however, it is known that Htt has roles in cell survival, proliferation, and migration
[101, 102]. Further, wild-type Htt binds repressor proteins allowing the expression of critical neuronal genes such as BDNF. Despite this and its well-established role in neuropathology, its relationship with neurodevelopmental alcohol exposure remains an unexplored avenue, and though its consistent appearance as a hub here is intriguing, an explanation would only be speculative.
Neurodevelopmental ethanol exposure and long-term gene expression changes
Given that this study examined the long-term effects of neurodevelopmental exposure to ethanol, it may be pertinent to ask what initiates and maintains these gene expression changes over time. A clear answer to these questions is not straightforward and may involve some combination of the primary (acute) and secondary effects of ethanol exposure. Factors that instigate these changes in gene expression may include ethanol-induced apoptosis of susceptible cell types, leading to an overall change in the cellular composition of the brain and, subsequently, the overall pattern of brain gene expression
[27, 32]. The authors acknowledge that the use of whole brain tissue makes this a possible reason for the results obtained by this study. Also, it is important to note that, while the genes identified may represent a broad response by the neural transcriptome to ethanol, it would be expected that various brain regions respond differently to ethanol exposure given that developmental processes do not occur uniformly and each region would have its own gene expression profile.
Another contributing mechanism may be the ability of ethanol to disrupt developmental processes that are highly reliant on external cues such as cell proliferation, migration, or differentiation
[39, 58, 104]. These mechanisms would represent immediate effects of alcohol that, during adulthood, do not directly alter gene expression per se but have altered cellular identity or physiology of the brain such that the appropriate balance of neural gene expression is not maintained. This hypothesis has been suggested for a number of spectrum disorders, and has been referred to as a neurodevelopmental ‘footprint’ of teratogen exposure
[15, 105, 106].
Finally, it is possible that there may be some factors that maintain inappropriate gene expression at the molecular level. Recent reports suggest that developmental ethanol exposure can affect epigenetic patterning, particularly DNA methylation and microRNA expression, which can produce long-term and relatively stable changes to the expression of a number of genes
[107–111]. How these changes may occur and how they are maintained is unknown, but this hypothesis is certainly worth further investigating as it would have implications towards the persistence of FASD phenotypes throughout the life of an individual.