Embryonic ethanol exposure alters expression of sox2 and other early transcripts in zebrafish, producing gastrulation defects

Ethanol exposure during prenatal development causes fetal alcohol spectrum disorder (FASD), the most frequent preventable birth defect and neurodevelopmental disability syndrome. The molecular targets of ethanol toxicity during development are poorly understood. Developmental stages surrounding gastrulation are very sensitive to ethanol exposure. To understand the effects of ethanol on early transcripts during embryogenesis, we treated zebrafish embryos with ethanol during pre-gastrulation period and examined the transcripts by Affymetrix GeneChip microarray before gastrulation. We identified 521 significantly dysregulated genes, including 61 transcription factors in ethanol-exposed embryos. Sox2, the key regulator of pluripotency and early development was significantly reduced. Functional annotation analysis showed enrichment in transcription regulation, embryonic axes patterning, and signaling pathways, including Wnt, Notch and retinoic acid. We identified all potential genomic targets of 25 dysregulated transcription factors and compared their interactions with the ethanol-dysregulated genes. This analysis predicted that Sox2 targeted a large number of ethanol-dysregulated genes. A gene regulatory network analysis showed that many of the dysregulated genes are targeted by multiple transcription factors. Injection of sox2 mRNA partially rescued ethanol-induced gene expression, epiboly and gastrulation defects. Additional studies of this ethanol dysregulated network may identify therapeutic targets that coordinately regulate early development.


Ethanol exposure during blastula period perturbs gene expression patterns prior to gastrulation.
Affymetrix GeneChip microarray analysis comparing control embryos to those treated with ethanol from 2 to 4.5 hpf (cleavage and pre-gastrulation stages) showed statistically significant changes of expression of many genes critical for embryogenesis (Supplementary Table S1) including sox genes, Notch ligands and Hairy/E(spl)-related (her) genes. To validate microarray results, a subset of genes were examined by either qPCR or by in situ hybridization. Downregulation of sox2 (array fold change −1.99, p < 0.0001), dlc (array fold change −1.82, p = 0.007) and her7 (array fold change −2.51, p < 0.001) genes at 4.5 hpf after ethanol exposure was confirmed by qPCR ( Fig. 1A). In situ hybridization showed reduced staining for sox2, dlc and dld (array fold change −1.53, p < 0.001) at 4.5 hpf in ethanol-exposed embryos compared to control embryos ( Fig. 1B-G).
There were significant changes in the expression of 651 probes, (absolute changes ≥ 1.25, FDR 0.15, p < 0.03) due to ethanol exposure (Supplementary Table S1). Out of those 651 probes, we were able to map Ensembl IDs for 534 probes, which correspond to 521 genes. Functional annotation analysis of ethanol dysregulated genes was done using DAVID that identified genes enriched in cellular processes, including transcription regulation and gene expression; DNA recombination; cell division and microtubule-based movement; cell-cell adhesion; and carbohydrate metabolic processes. Genes enriched in developmental processes, including dorso-ventral and anterior-posterior axes formation, cerebellum, somite, and optic fissure development were also detected in DAVID analysis. Dysregulated genes were enriched in Wnt, Notch, and retinoic acid signaling pathways (Table 1).
Among the 521 dysregulated genes, we identified 61 transcription factors (Table 2), including Sox2, a critical transcription factor. The expression of sox2 was significantly reduced after ethanol exposure. To identify Sox2 targets across the zebrafish genome, position weight matrixes for Sox2 were mapped within 2 kb upstream of transcription start sites of genes using find individual motif occurrences software 31 . Possible Sox2 targets were compared with the ethanol-dysregulated genes, which showed that 52 genes were common in both datasets ( Fig. 2A). Transcriptome changes caused by SoxB1 knockdown (quadruple knockdown: sox2/3/19a/19b) at 30% epiboly (~4.7 hpf) were reported previously 29 . Results of this study were compared with ethanol dysregulated genes (4.5 hpf). We found 98 genes common in between SoxB1 knockdown dysregulated genes and ethanol dysregulated genes. Comparison of all three datasets showed 11 common genes in all these datasets ( Fig. 2A). These data indicate that ethanol affects the expression of Sox2 and several Sox2 transcriptional targets.
To predict the possible binding sites of other ethanol-dysregulated transcription factors across the zebrafish genome, we explored the available position weight matrixes (TRANSFAC). Position weight matrixes were found (2020) 10:3951 | https://doi.org/10.1038/s41598-020-59043-x www.nature.com/scientificreports www.nature.com/scientificreports/ for 24 of the 60 other ethanol-dysregulated transcription factors. Target genes of these 24 transcription factors were predicted by mapping position weight matrixes within 2 kb upstream of the start site of genes across zebrafish genome. Predicted targets of the dysregulated transcription factors show that many ethanol-dysregulated genes are targets of these transcription factors. We compared the predicted targets of these dysregulated transcription factors and examined for the enrichment of ethanol dysregulated genes by computing hypergeometric probability. This transcription factor-target gene interaction analysis identified 827 interactions that include 25 transcription factors targeting 423 dysregulated genes. Individual interaction counts for each transcription factor is listed in Table 3. The enrichment of ethanol-dysregulated targets over all possible genomic targets for a given www.nature.com/scientificreports www.nature.com/scientificreports/ transcription factor plotted as a bar graph is shown in Fig. 2B. A network analysis was done using cytoscape software to visualize the interactions between the dysregulated transcription factors and the dysregulated targets. This analysis showed that 25 transcription factors target many of the same ethanol dysregulated genes, which identifies a potential ethanol-induced transcription factor-target gene regulatory network in the early embryo (Fig. 2C). As many of the genes dysregulated by ethanol exposure were targeted by multiple dysregulated transcription factors, these factors could produce synergistic ethanol dysregulation effects during early zebrafish development.
To mark the axial mesendoderm, germ band (mesodermal cells at the leading-edge during epiboly progression) and the dorsal forerunner cells (a group of cells that migrate at the leading-edge of shield during gastrulation but do not involute) in control and treated embryos, ntl in situ hybridization was performed. Control embryos had dorsal forerunner cells closely associated with the germ ring. Dorsal forerunner cells were dissociated from one another and from the germ band in ethanol-treated embryos, which was partially rescued by sox2 mRNA injection (Fig. 3F-I). Functional annotation analysis detected dysregulation of genes involved in dorsal/ ventral and anterior/posterior axes formation (Table 1). ntl staining confirmed that ethanol-exposed embryos had reduced convergence-extension of the axial mesendoderm cells compared to control embryos, producing shorter www.nature.com/scientificreports www.nature.com/scientificreports/ and wider axes. The convergence-extension defect was partially rescued by sox2 mRNA injection in the ethanol treated embryos (Fig. 3F-I).
The effects of sox2 mRNA injection on the expression of a few ethanol-dysregulated genes were analyzed. The expression level of dld was restored in sox2 mRNA injected, ethanol-exposed embryos as seen by dld in situ hybridization (Fig. 4A-D). Quantitative PCR was done to analyze the expression of sox2, her7, and dlc. The sox2 mRNA injection restored expression of sox2 that was downregulated in ethanol-treated embryos (Fig. 4E). The reduced expression of her7 and dlc in ethanol-treated embryos was significantly restored by sox2 mRNA injection in ethanol-treated embryos (ethanol-treated vs. sox2 mRNA + ethanol-treated: dlc; p < 0.01, her7; p < 0.01) (Fig. 4E). However, the expression levels of these genes in sox2 mRNA injected + ethanol-treated embryos were higher compared to control embryos (control vs. sox2 mRNA injected + ethanol-treated: dlc; p < 0.01, her7; p < 0.01) (Fig. 4E). Overall, sox2 mRNA injection partially rescued ethanol-induced early developmental defects.

Discussion
Mouse and rat studies showed the association between prenatal ethanol exposure and gene expression changes in postnatal and adult stages [34][35][36][37] . This study examined ethanol-induced gene expression changes during embryogenesis before gastrulation. This is the first animal model study that identified the effects of ethanol on a master regulator, Sox2 that orchestrates embryogenesis, self-renewal, and pluripotency 27,29,32 . Ethanol exposure altered the expression of a large number of genes, which include other critical regulators of development. The differentially expressed genes are involved in various functions ranging from cellular processes, embryonic development to signaling pathways. This indicates possible multifactorial effects, which may include the alteration of epigenome by ethanol exposure, causing the changes in expression of critical genes.
Functional annotation analysis of the dysregulated genes identified enrichment of genes involved in mitotic nuclear division and microtubule-based movements. In fact, our previous study identified large multinucleated enveloping layer cells in 8 hpf ethanol-treated embryos and fragmentation of yolk microtubules 18 , which support current findings. The cell adhesion defects that was observed previously at 8 hpf 18 were detected at 4.5 hpf ethanol-treated embryos, suggesting continuous defects in cell-to-cell communication and cell movements in those embryos. Interestingly, ethanol-sensitive signaling pathways detected during early embryogenesis were Wnt, Notch, and retinoic acid. Ethanol-induced dysregulation of retinoic acid signaling pathway was reported earlier 16,17,19,[38][39][40][41] . Our studies examining the heart and the eye in ethanol-treated embryos identified disruption of Wnt, Notch, retinoic acid and Bmp signaling pathways during organogenesis [17][18][19][20] . This study showed that those signaling defects initiate early in ethanol-exposed embryos and continue to have their detrimental effects at later stages of development.
This microarray analysis detected a reduction of sox2 expression. Ethanol-induced effects on stoichiometry of Sox2 and Oct4 was reported during differentiation of mouse embryonic stem cell 42 . Sox2 and other Sox B1 type transcriptional regulators control a wide range of developmental effectors, including pcdh18a (gastrulation movement), neurog1, hesx1, and zic1 (neural differentiation), oep, and shh (neural patterning) 29 . Our previous Affymetrix GeneChip microarray study (GEO accession: GSE48380) comparing genes in control and ethanol-exposed embryos during mid-gastrulation identified significant dysregulation of all these genes after ethanol exposure at 8 hpf 15,22 . Reduction of Sox2 prior to gastrulation might be responsible for the dysregulation of these genes, interfering with gastrulation and other developmental events. Notably, convergence-extension defects seen in the ethanol-exposed embryos was also reported in the SoxB1 knockdown embryos, which display similar wedge-shaped ntl expression pattern. Presence of precise amount of sox2 transcripts is essential for normal development of the embryo. Injection of more than 30 pg of sox2 mRNA into the embryos caused delayed epiboly progression and other developmental defects, which caused the death of injected embryos (data not shown). Injection of 25 pg of sox2 mRNA partially rescued ethanol-induced epiboly defects. Injection of less than 25 pg of sox2 mRNA gave weaker rescue effects. sox2 mRNA injection raised the her7 and dlc transcript levels in ethanol-exposed embryos. However, the her7 and dlc transcript levels in sox2 mRNA-injected + ethanol-exposed embryos were significantly higher than the transcript levels in control embryos. Although sox2 mRNA injection did not fully rescue her7 and dlc expression in ethanol-exposed embryos, the results support their transcriptional regulation by Sox2 and ethanol. The transcription factor-target gene regulatory network (Fig. 2C) shows that her7 is regulated not only by Sox2, but also by Mnt, Cebpg3, and Atf7. Similarly, the expression of dlc is regulated by Sox2, Elf3, and Maz.
Transcription factor-target gene interaction analysis showed that the number of potential targets among the ethanol-induced dysregulated genes are more for Foxc1a and Foxa2 (members of forkhead transcription factors), Elf3 (a member of the E26 transformation specific family of transcription factors), and Cdx4 (a member of caudal www.nature.com/scientificreports www.nature.com/scientificreports/ gene family) than for Sox2. However, ethanol-induced gene expression change for sox2 was higher (and possibly playing a more significant role) than foxc1a, foxa2, elf3 and cdx4. Foxc1a plays crucial role in somitogenesis, cardiovascular and retina development during embryogenesis [43][44][45][46][47][48] . Foxa transcription factors are essential for developmental of dorsal axis structures including prechordal plate, notochord, hypochord, and floor plate [49][50][51] . A growing body of research shows that Elf3 plays significant roles in the development of cancer [52][53][54][55][56] . Although poorly understood, there is evidence that Elf3 is important during development. A null mutation of Elf3 caused the death of about 30% of mice in utero 57,58 . Another Sox family gene sox9b was also detected as the ethanol-dysregulated gene in our study. Sox9 is involved in many developmental processes including craniofacial, heart, brain and retinal development in mammal 59 . Zebrafish has two homologues of sox9, sox9a and sox9b. Both Sox9a and Sox9b play roles on neural, cardiac, and cartilage development Zebrafish [60][61][62][63][64] . Network analysis showed that many of these dysregulated transcription factors interact with each other, which suggests that the ethanol-induced developmental defects were the combined and perhaps synergistic effects of multiple regulators and that manipulation of one gene would not lead to complete rescue of ethanol-induced defects. Future study is needed to analyze the potential interaction between combinations of genes coordinately disrupt early development, contributing to ethanol-induced defects.

Methods
Zebrafish husbandry. Zebrafish (Danio rerio) ABTL strain was raised and maintained under standard laboratory conditions 65 following Indiana University Policy on Animal Care and Use. The use of zebrafish adults for breeding, embryo collection and embryo experiments were approved by the campus animal care and use ethics committee: IUPUI School of Science Institutional Animal Care and Use Committee (IACUC).
Ethanol treatment. Embryos were kept in embryo medium after fertilization. At 2 hpf, embryos were divided into two groups. One group was transferred to embryo medium containing 100 mM ethanol (E100) and the other group was kept in embryo medium without ethanol (control). Standard protocol for the Affymetrix 3′IVT Express kit (Affymetrix, Santa Clara, CA, USA) was followed to label the samples starting with 100 ng of total RNA. The 12 samples were each hybridized to a Zebrafish Genome Array (Affymetrix) for 17 h, washed, stained and scanned following the standard protocol; all 12 were handled in parallel as a single batch. Arrays were visually scanned for abnormalities or defects; none were found.
Affymetrix gene expression console software was used to generate MAS5 66 signals and detection calls; arrays were scaled to a target of 1000. Only those probe sets that had a MAS5 signal fraction present ≥ 0.50 in at least one of the two treatments were analyzed 67 . MAS5 signals were imported into Partek Genomics Suite (Partek, Inc., St Louis, MO, USA) and log 2 transformed. These log 2 transformed signals were used for principal components analysis, hierarchical clustering and signal histograms to determine if there were any outlier arrays, and no outliers were detected. A 2-way ANOVA with factors for treatment (ethanol vs., control) and independent experiment (random effect) was used to analyze log 2 transformed signals. This analysis indicated that the embryo batch was indeed significant. The False Discovery Rate (FDR) was calculated using the Storey qvalue method 68 . Microarray data were deposited in the NCBI GEO database, accession number GSE145574.
A subset of the differentially expressed probes with absolute changes ≥ 1.25 were included in downstream analysis (FDR 0.15; p < 0.03). We performed function annotation analysis of these ethanol dysregulated probes using DAVID 69 . Additionally, we annotated the transcription factors for these differentially expressed probe IDs using AnimalTFDB database 70 . For these annotated ethanol-dysregulated transcription factors, we obtained the position weight matrices available in TRANSFAC 71 and searched for their occurrence using "find individual motif occurrences" algorithm 31 to predict the target site within 2 kb upstream of start site of every gene in zebrafish genome (Zv11). Find individual motif occurrences computes a log-likelihood ratio score for the occurrence of each motif in a specific input sequence and hence enables the discovery of recognition sequences of transcription factors in the upstream regions of gene starts. Find individual motif occurrences computes converts these scores into p-values using dynamic programming generated at random genomic loci with user defined background frequencies for genomic alphabet (A, T, G, C) with false discovery rates. It provides a ranked list of motif occurrences per position weight matrices, each with an associated log-likelihood ratio score, p-value and other statistical metrics. Find individual motif occurrences computes based predicted targets of these dysregulated transcription factors were compared and examined for the enrichment of ethanol induced dysregulated targets by computing hypergeometric probability for the gene set enrichment. Resulting data was represented as networks using cytoscape 72 network visualization software.  www.nature.com/scientificreports www.nature.com/scientificreports/ Quantitative PCR analysis. One µg of total RNA extracted from control and E100 embryos was reverse transcribed to cDNA using M-MLV reverse transcriptase (Promega, Madison, WI, USA), and cDNA was diluted tenfold with RNase free water. Each 20 µl PCR reaction was performed with 1-4 µl of cDNA using Power SYBR In situ hybridization. Whole-mount in situ hybridization of zebrafish embryos was performed using digoxigenin-labeled riboprobes for ntl, dlc, and dld. The riboprobes were synthesized using DIG RNA Labeling Kit (Roche, Indianapolis, IN, USA) according to manufacturer's recommendations. For sox2 riboprobe, pCBA3-zf-sox2 plasmid was cut using BamHI restriction enzyme, and the Dig RNA probe was synthesized using T7 RNA polymerase. Images were collected using a Leica MZ12 microscope equipped with Leica DFC290 camera. www.nature.com/scientificreports www.nature.com/scientificreports/ sox2 rescue experiments. mRNA was synthesized from a pCBA3-zf-sox2 vector 29 using a SP6 mMessage mMachine kit (Ambion, Austin, TX, USA). Synthetic mRNA (25 pg/embryo) was injected into the embryos at the 2-cell stage. Injected and uninjected embryos were treated with or without 100 mM ethanol until analyzed. For epiboly measurement, embryos were fixed at 8 hpf, dechorionated and imaged focusing on enveloping cell layer at the embryo margin. Percent epiboly progression was calculated using Image J software. For gen expression analyses, embryos were dechorionated and total RNA was extracted at 4.3 hpf, and quantitative PCR was performed. One-way ANOVA and post hoc Tukey HSD for individual comparisons were used for analyses in rescue experiments.