Different effects of knockouts in ALDH2 and ACSS2 on embryonic stem cell differentiation

Abstract

BACKGROUND:

Ethanol (EtOH) is a teratogen that causes severe birth defects, but the mechanisms by which EtOH affects stem cell differentiation are unclear. Our goal here is to examine the effects of EtOH and its metabolites, acetaldehyde (AcH) and acetate, on embryonic stem cell (ESC) differentiation.

METHODS:

We designed ESC lines in which aldehyde dehydrogenase (ALDH2, NCBI#11669) and acetyl CoA synthetase short chain family member 2 (ACSS2, NCBI#60525) were knocked out by CRISPR-Cas9 technology. We selected these genes because of their key roles in EtOH oxidation in order to dissect the effects of EtOH metabolism on differentiation.

RESULTS:

By using kinetics assays we confirmed that AcH is primarily oxidized by ALDH2 rather than ALDH1A2. We found increases in mRNAs of differentiation-associated genes (Hoxa1, Cyp26a1, and RARβ2) upon EtOH treatment of WT and Acss2^−/− ESCs, but not Aldh2^−/− ESCs. The absence of ALDH2 reduced mRNAs of some pluripotency factors (Nanog, Sox2, and Klf4). Treatment of WT ESCs with AcH or 4-hydroxynonenal (4-HNE), another substrate of ALDH2, increased differentiation-associated transcripts compared to levels in untreated cells. mRNAs of genes involved in retinoic acid (RA) synthesis (Stra6 and Rdh10) were also increased by EtOH, acetaldehyde, and 4-HNE treatment. Retinoic acid receptor-γ (RARγ) is required for both EtOH- and AcH-mediated increases in Hoxa1 and Stra6, demonstrating the critical role of RA:RARγ signaling in AcH-induced ESC differentiation.

CONCLUSIONS:

ACSS2 knockouts showed no changes in differentiation phenotype while pluripotency-related transcripts were decreased in ALDH2 knockout ESCs. We demonstrate that AcH increases differentiation-associated mRNAs in ESCs via RARγ.

INTRODUCTION

Alcohol use disorders affect millions of adults in the United States every year, and are responsible for approximately 4% of the global disease burden each year (Connor, 2016). Binge drinking is defined as acute intoxication from ethanol (EtOH) ingestion that causes blood concentrations of EtOH to exceed 0.08 g/dL (Kanny et al., 2015). The Centers for Disease Control and Prevention have established that there is no safe level for alcohol intake in pregnancy, yet binge drinking occurs most frequently in adults between the ages of 18–34 (Kanny et al., 2015) when women are in prime child bearing age. EtOH diffuses into the placenta in concentrations similar to those that are present in maternal blood (Brien et al.,1983; Thomas and Riley, 1998), which causes a variety of teratogenic effects in embryos, leading to birth defects under the category of fetal alcohol spectrum disorders (FASD) (Olney, 2004). Therefore, we used embryonic stem cells (ESCs), which represent the most primordial stage of development, to serve as a model system for investigating whether EtOH or its metabolites can alter differentiation capacity.

Alcohol is metabolized in a two-step oxidation process (Zimatkin and Bubin, 2007). First, EtOH is oxidized to acetaldehyde (AcH) by members of the alcohol dehydrogenase (ADH) family of enzymes, CYP4E1, or catalase; subsequently, AcH is metabolized to acetate by aldehyde dehydrogenase (ALDH) family members (Figure 1A) (Zimatkin and Bubin, 2007; Vasiliou et al., 2004). Acetate can then either exit the cell through monocarboxylate (MCT1/4, SLC16A1–3) transporters (Halestrap, 2012) or serve as a substrate for acetyl CoA formation, primarily via the enzymatic actions of cytoplasmic Acyl-CoA synthetase short chain member 2(ACSS2/ACECS1) (Wolfe, 2005). Acetyl CoA is a central metabolite in several physiologic processes, including anabolic production of fatty acids and triglycerides (lipogenesis), protein and histone acetylation, and catabolic processes to generate energy, such as its entry into the tricarboxylic acid cycle (Figure 1A) (Huang et al., 2015; Ochocki and Simon, 2013; Shi and Tu, 2015). AcH is a highly reactive intermediate (Lopachin et al., 2009) that causes damage to nucleic acids, proteins, lipid membranes, and can disrupt the functions of organelles, such as mitochondria (Seitz and Stickel, 2007). The major toxic effects of AcH result from formation of covalent modifications with several biomacromolecules to generate adducts that interfere with their physiologic functions, resulting in protein inactivation, and DNA damage, and mutations in many cell types, including stem cells (Lopachin et al., 2009; Garaycoechea et al., 2018).

Figure 1: Aldh2 and Acss2 were targeted by CRISPR/Cas9 to generate knockout cell lines.

A, Schematic of the ethanol (EtOH) oxidation pathway. EtOH is first oxidized to acetaldehyde (AcH) by an alcohol dehydrogenase family member, Catalase, or CYP4E1 using NAD⁺ as a cofactor. AcH can then generate reactive carbonyl species (RCS) or reactive oxygen species (ROS), which damage macromolecules and mitochondria. ALDH2 metabolizes AcH to acetate while reducing another molecule of NAD⁺. Acetate is then either exported by a transporter in the SLC16A family or converted to acetyl CoA by ACSS2. Acetyl CoA can then enter into a number of physiologic pathways, including lipogenesis, pathways of energy production, and protein acetylation. B-C, Western blotting of ALDH2 (B) and ACSS2 (C) in WT and CRISPR/Cas9-generated mutant clones. β-actin was used as a loading control in both experiments. D-E, Gene sequences for WT and mutated alleles for a selected clone of Aldh2^−/− (28 and 9 base pair deletions) (D) and Acss2^−/− (28 and 11 base pair deletions) (E) ESCs.

ALDH2, with a Km for AcH of 0.2 μM (Klyosov et al., 1996), plays a critical role in AcH metabolism (Vasiliou et al., 2004). However, there is no consensus as to how AcH is metabolized in ESCs, which express both ALDH2 and ALDH1A2, an enzyme also required for retinaldehyde conversion to retinoic acid (RA). Data from some previous studies have suggested that EtOH can be metabolized by ALDH1A2 at the expense of RA formation at some early stages of embryonic development (Kot-Leibovich, 2009; Shabtai et al, 2018).

In this research, we show that AcH is primarily metabolized by ALDH2 in ESCs. EtOH, AcH, and 4-hydroxynonenal (4-HNE), an endogenous aldehyde that requires ALDH2 for metabolism (Siems and Grune, 2003), induce transcripts of genes associated with differentiation, such as Hoxa1, Cyp26a1, and Stra6, in WT ESCs. Furthermore, we show that the loss of ALDH2 partially inhibits the EtOH-mediated induction of these differentiation-associated genes, while ACSS2 ablation does not. Deletion of Aldh2 additionally results in decreases in pluripotency-related mRNAs, whereas we did not observe reductions of these transcripts in Acss2^−/− ESCs. Importantly, we show that both EtOH- and AcH-mediated induction of differentiation-associated transcripts require retinoic acid receptor-γ (RARγ).

MATERIALS AND METHODS

Cell culture and reagents:

Cells from the WT ESC line, CCE, and mutant cells derived from this cell line, were cultured as previously described (Kashyap et al., 2011). Cells were treated with 40 mM EtOH, 1 mM acetaldehyde (AcH) (Calbiochem, San Diego, CA), 2 mM N-acetylcysteine (NAc) (Sigma, St. Louis, MO) or 1 μM 4-hydroxynonenal (4-HNE) (Sigma). AcH was aliquoted from a freshly opened bottle and tubes were stored at −20°C for no more than 2 months. Each aliquot was prepared for a single use. NAc was made fresh in water and pH-adjusted to pH=7.4 for each dosage administered and replaced in the medium every 12 hours. For experiments in which only EtOH was used as a reagent, we administered EtOH to fresh medium twice daily, every 12 hours, for 48 hours, with the final reagent change completed 8 hours prior to harvest. For experiments in which AcH or 4-HNE were used, reagents were changed three times daily, 6–8 hours apart. 1 × 10³ units/ml of LIF were added to the medium for all experiments. ESCs were seeded in 6-well plates to collect mRNA or protein, and 24 well plates for proliferation assays. Cell numbers were counted using an electron particle counter (Coulter Z1, Beckman Coulter, Brea, CA).

Western blotting:

CCE WT, Aldh2^−/−, and Acss2^−/− ESCs were harvested in 4% SDS lysis buffer, boiled, resolved on SDS-PAGE gels, and transferred onto nitrocellulose membranes. Antibodies were applied to membranes overnight with shaking at 4°C using the following dilutions: ALDH2 (1:2000, Abcam; ab108306; Lot GR97098–10), ACSS2 (1:1000, Cell Signaling; 3658S; Lot 2) and actin (1:40,000 or 1:50,000, Millipore, MAB1501; Lot 2665057). We then incubated the membranes with secondary antibody at room temperature with shaking for one hour and recorded chemiluminescence using a gel imaging station (Bio-Rad ChemiDoc). We analyzed images using Image Lab software (Bio-Rad).

Generation of Acss2^−/− and Aldh2^−/− embryonic stem cells:

Guide (g)RNAs targeting the sequence GAAGTCGCCGTCGATGGGAA (A) in the sense strand of exon 5 and the sequence TATACCCGCCATGAGCCTGT (B) in the antisense strand of the Aldh2 gene were cloned into the BbsI sites of disparate pX461-hSpCas9n(BB)-2A-GFP vectors. We digested vector A with XbaI overnight, dephosphorylated with shrimp alkaline phosphatase, ran the DNA on an agarose gel, excised, and purified the DNA. We amplified vector B using primers targeting the hU6 promoter (Fwd: 5’-TTTGCTAGCGAGGGCCTATTTCCCATGAT −3’) and a downstream CRISPR sequence (Rev: 5’-GGTACCGCTAGCGCCATTTGTCTGC-3’). A 400 bp product was excised from an agarose gel, purified, and ligated into vector B. We transformed clones into DH5α E. coli. Following XbaI and PciI double digestion, clones positive for both gRNAs exhibited an 850 bp band, 400 bp greater than clones that failed to incorporate vector B. Following transfection, we harvested colonies, amplified their DNA by quantitative polymerase chain reaction (PCR) (Fwd: 5’-TGAGCATGGCTGACCCCAAGT-3’/Rev: 5’-AGCCAAATGCCAGGGTTGTTGC-3’, 272 bp product), and digested the DNA with BccI to genotype CRISPR-edited clones. We digested clones with a band size difference between the two alleles that was separated on a 2–3% agarose gel following PCR amplification and then purified DNA from each band. We utilized a TOPO-TA kit (ThermoFisher) to isolate DNA from disparate alleles in additional clones in which separation could not be achieved on a gel. We sequenced clones lacking the restriction site on both alleles via the Cornell University Institute of Biotechnology DNA sequencing Core Facility. Following confirmation of biallelic homozygous deletion by DNA sequencing, we expanded these knockout clones in culture.

The same procedure was followed for the Acss2 gene, using gRNAs CAGCAATGTTCTCCGTAAAC (A) and GAGTTCACGGTATGTGATCT (B) targeting exon 3, and EcoNI for genotyping clones. The primers used for PCR amplification are as follows: Fwd: 5’-GTTGGAATTTTGTGACTGCTCCTG-3’ Rev: 5’-CCTGTTACCAGATCCATCCATTTC-3’, 283 bp product.

Kinetic assessment of wild type, Aldh2^−/−, and Aldh1a2^−/− embryonic stem cell lines:

We cultured ESCs in either 6-well plates or 10 cm dishes and harvested in TEN buffer pH=7.2 (50 mM Tris, 150 mM NaCl, and 1 mM EDTA) to retain enzymatic activity of cells. We then centrifuged the lysates, resuspended them in 250 mM Tris pH=7.8 with vortexing, and lysed them by performing snap thawing after freezing at −70°C. The ESC lysates were further concentrated using 10K concentrators (Amicon Ultra, Sigma), and by performing two additional concentration steps using 250 mM Tris pH=7.8. Kinetics assays were conducted to determine the rate of AcH metabolism in various cell lines using the SpectraMAX 340PC plate reader (Molecular Devices, San Jose, CA). 0.99 mg/ml of protein lysate was incubated in PBS buffer (pH 7.4) supplemented with NAD⁺ 3 mM, Alda1 (Sigma) 15 μM, or diethylbenzaldehyde (DEAB) (Sigma) 1 μM. After three minutes, the assay was initiated by the addition of 2.7 mM AcH. We monitored the absorbance every 60 seconds at 340 nm for NADH formation and at 600 nm for background determination. The averages of three technical replicates for each sample were used to quantitate rate values in nmol•min⁻¹•g⁻¹ protein units, and two biological repeats were performed. Rates were calculated using the slopes of linear regression lines generated for the average of each sample in Microsoft Excel.

RNA isolation and real-time PCR:

We performed RNA extraction using TRI Reagent (Sigma) according to the manufacturer’s instructions. We quantified 1 μg RNA and reverse transcribed to make complementary DNA using the qScript cDNA synthesis kit (Quanta Biosciences, Gaithersburg, MD). All cDNA was diluted five-fold. We used SYBR Green quantitative PCR Supermix in a 15 μl reaction mix to conduct reactions on a Bio-Rad iCycler using 3 μl of cDNA. We performed quantification using the Ct method and generated standard curves for all runs to assess efficiency. The levels of all mRNA transcripts were normalized using a 36b4 internal control. Table 1 shows a list of primers used.

Table 1:

List of primer sequences used for qPCR with predicted product size.

Target	Forward (5’ → 3’)	Reverse (5’ → 3’)	Product size (bp)
Col4a1	CCAGCCTGGAGCTAAGGGAGA	TCCAGGTGTACCGGGAATGC	700
Cyp26a1	GAAACATTGCAGATGGTGCTTCAG	CGGCTGAAGGCCTGCATAATCAC	272
Hoxa1	TTCCCACTCGAGTTGTGGTCCAAGC	TTCTCCAGCTCTGTGAGCTGCTTGGTGG	220
Klf4	GCACACCTGCGAACTCACAC	CCGTCCCAGTCACAGTGGTAA	53
Nanog	ATGCCTGCAGI 1 1 1ICATCC	GAGCTTTTGTTTGGGACTGG	153
Oct4	GAGGAGTCCCAGGACATGAA	AGATGGTGGTCTGGCTGAAC	154
RARβ	GATCCTGGATTTCTACACCG	CACTGACGCCATAGTGGTA	247
Rdh10	TCTGGACATCACCTTCTGGAATG	CTCAACTCCAGCAGTGCTGAAC	177
Sox2	GAGTGGAAACTTTTGTCCGAGA	CTCCGGGAAGCGTGTACTTA	156
Stra6	GTTCAGGTCTGGCAGAAAGC	CAGGAATCCAAGACCCAGAA	102
36b4	AGAACAACCCAGCTCTGGAGAAA	ACACCCTCCAGAAAGCGAGAGT	448

Statistical treatment of the data:

Statistical analysis on qPCR data was conducted on at least three, independent biological replicates for each experiment using Graph Pad Prism 7.0 software, and determined the means ± SEM. We used one-way ANOVA to determine statistical significance within sets of 3 or more groups, using multiple comparisons analyses comparing the means of each column with the means of either a control or pre-selected group, followed by a post-hoc Bonferroni’s test. Student’s t-tests were used to compare two independent populations. A two-tailed p-value < 0.05 was considered statistically significant.

RESULTS

We generated ALDH2- and ACSS2-knockout embryonic stem cell lines by CRISPR/Cas9.

We sought to ablate ALDH2 activity in CCE embryonic stem cells (ESCs) using CRISPR/Cas9 targeted to two sequences in exon 5 of the Aldh2 gene, which has low sequence conservation with other ALDH family genes and is upstream of the catalytic site. Likewise, we targeted two sequences in exon 3 of the Acss2 gene for CRISPR/Cas9-mediated deletion. We identified five clones from guide (g)RNA-Aldh2-treated ESCs that lacked functional ALDH2 protein by Western blotting (Figure 1B). Two clones from the gRNA-Acss2-treated cells were negative for protein expression (Figure 1C). We selected one clone from each knockout line for sequencing on both alleles to ensure loss of both alleles in the targeted exon of the Aldh2 (Figure 1D) and Acss2 (Figure 1E) genes.

Acetaldehyde is primarily metabolized by ALDH2 in embryonic stem cells.

To determine if ALDH2 is responsible for most of the AcH oxidation in ESCs, we compared the kinetics of NADH production following AcH consumption in protein extracts of wild type (WT) ESCs (Figure 2A). ALDH1A2, instead of ALDH2, has been reported to metabolize AcH (Kot-Leibovich et al., 2009; Shabtai et al., 2018). Therefore, we also used an ALDH1A2-knockout ESC line previously generated in our lab (Serio et al., 2019) to probe the extent of AcH oxidation by ALDH2 versus ALDH1A2 in ESCs.

Figure 2: Acetaldehyde is primarily metabolized by ALDH2 in embryonic stem cells.

A, Diagram of experimental design for kinetics assays. ESCs were lysed and their protein extracts were concentrated and incubated with 2.7 mM AcH and 3 mM NAD⁺, ±15 μM Alda1, ±1 μM diethylaminobenzaldehyde (DEAB) prior to measuring absorbance at 340 nm (λ for NADH formation) in a plate reader. B-D, Enzyme assay measuring NADH accumulation from the reaction of embryonic stem cell lysate with AcH and NAD⁺. B, Kinetic analysis of rate of NADH accumulation in WT (black), Aldh1a2^−/− (blue), and Aldh2^−/− (red) ESCs over 5 hours. C-D, Kinetic analysis of rate of NADH accumulation in WT (C) and Aldh2^−/− (D) ESCs treated with AcH + NAD (black), AcH + NAD + 1μM ALDH2 inhibitor, DEAB (red), and AcH + NAD + 15 μM ALDH2 agonist, Alda1 (blue). A representative experiment of n=2 biological replicates is shown.

The cofactor NAD⁺ is reduced to NADH upon catalyzed conversion of AcH to acetate by members of the aldehyde dehydrogenase family of enzymes, making detection of NADH production observed at 340 nm, which is convenient and sensitive for enzyme activity measurements (Yan et al., 1987). We determined the rate of 2.7 mM AcH oxidation by WT ESC extracts to be 71.5 nmol•min⁻¹•mg⁻¹ protein (Table 2). The rate at which Aldh2^−/− protein extracts oxidized AcH was decreased by 91.9% (5.8 nmol•min⁻¹•g⁻¹ protein) compared to WT ESC extracts, whereas we observed only a 19% reduction in the rate of AcH oxidation in Aldh1a2^−/− (57.9 nmol•min⁻¹•g⁻¹ protein) compared to WT protein extracts (Figure 2B). These results confirm that AcH is primarily metabolized by ALDH2, and not by ALDH1A2, in ESCs.

Table 2:

Kinetic rates of acetaldehyde oxidation in wild-type and mutant cell lines.

	nmol • min−1 • g−1 protein
	WT	Aldh2−/−	Aldh1a2−/−
+ AcH	71.5 ± 6.1	5.8 ± 1.5	57.9 ± 6.9
+ AcH + Alda 1	101.4 ± 5.3	8.2 ± 2.6	-
+ ACH + DEAB	34.4 ± 2.8	5.2 ± 2.1	-

To further confirm that NADH was primarily being produced by AcH conversion to acetate, we measured the activity of the cell lysates in the presence of 1 μM diethylaminobenzaldehyde (DEAB), a pharmacological inhibitor of ALDH2, or 15 μM Alda1, an ALDH2 agonist, respectively. Alda1 increased (41.8%, 101.4 vs. 71.5 nmol•min⁻¹•g⁻¹ protein) and DEAB decreased (51.8%, 34.4 vs. 71.5 nmol•min⁻¹•g⁻¹ protein) the rate of AcH oxidation in the WT ESC lysates, respectively (Figure 2C). Moreover, the rate of AcH oxidation (5.8 nmol•min⁻¹•g⁻¹ protein) in Aldh2^−/− ESC extracts is similar to that in the presence of Alda1 (8.2 nmol•min⁻¹•g⁻¹ protein) or DEAB (5.2 nmol•min⁻¹•g⁻¹ protein). These data suggest that the role of additional ALDH family enzymes in the oxidation of AcH in ESCs is minimal (Figure 2D, Table 2).

Ethanol treatment results in differentiation of wild-type embryonic stem cells, and knockout of ALDH2 prevents EtOH-associated differentiation of embryonic stem cells.

We next evaluated whether EtOH affected cell proliferation by performing cell counts of WT and Aldh2^−/− ESCs ± 40 mM EtOH at 72 hours. We found no changes in proliferation rates between the EtOH-treated and untreated cells in either WT (Figure 3A) or ALDH2-null (Figure 3B) ESCs. To further probe the phenotypes of ESCs that lack either ALDH2 or ACSS2, we measured cell proliferation of Aldh2^−/− and Acss2^−/− compared to WT ESCs. Aldh2^−/− ESCs displayed a 56.5% lower proliferation rate than WT cells (p<0.0001), while Acss2^−/− ESCs exhibited no obvious change in proliferation rate (Figure 3C).

Figure 3: Ethanol induces differentiation-associated transcripts, and ALDH2 ablation inhibits differentiation associated with ethanol.

A-B, Proliferation rate of WT and Aldh2^−/− ESCs ± 40 mM EtOH at 72 hours. C, Proliferation rate of WT, Aldh2^−/−, and Acss2^−/− ESCs at 72 hours. D, Fold changes in mRNA levels of pluripotency markers in untreated WT, Aldh2^−/−, and Acss2^−/− ESCs. E, Fold changes in mRNA levels by EtOH in WT and Aldh2^−/− ESCs. F, Fold changes in transcript levels of the late differentiation marker Col4a1 by EtOH in WT and Aldh2^−/− ESCs. G, Fold changes in mRNA levels of Hoxa1 by EtOH, 2 mM NAc, and EtOH + NAc. All qPCR experiments were conducted at 48 hours. n/s, not significant. Y-axes vary with samples being analyzed, and mRNA levels are shown in arbitrary units. Error bars represent standard errors of independent experiments where n = 3 biological repeats. *, p≤0.05, **, p≤0.01, ***, p≤0.001, ****, p≤0.0001.

Next, we assessed how loss of either ALDH2 or ACSS2 affects stem cell markers.

Loss of ALDH2 caused a reduction in three out of four pluripotency-related transcripts that we tested, including Nanog (−72.3 ± 0.09%, p=0.001), Sox2 (−33.7 ± 0.05%, p=0.003), and Klf4 (−64.2 ± 0.14%, p=0.011), but not Oct4 (112.2 ± 0.08%, p=0.203) (Figure 3D). However, these pluripotency markers exhibited transcript levels that were either unchanged (Sox2, Klf4, Oct4) or increased (Nanog, 2.30 ± 0.42-fold, p=0.036) in Acss2^−/− ESCs (Figure 3D). These results suggest that loss of ALDH2, but not ACSS2, is sufficient to disrupt the core pluripotency network in ESCs, which is critical in that loss of pluripotency results in ESC differentiation (Ng and Surani, 2011).

To elucidate the role of ALDH2 and its substrate AcH in stem cell differentiation, we compared the transcript levels of Hoxa1, Cyp26a1, and RARβ2 in WT and Aldh2^−/− ESCs ± EtOH. In WT ESCs, EtOH treatment increased differentiation-associated mRNAs compared to the mRNA levels in untreated cells (2.2 ± 0.3-fold, p=0.018, Hoxa1; 4.0 ± 1.0-fold, p=0.027, Cyp26a1; 1.6 ± 0.2-fold, p=0.044, RARβ2). mRNA levels of Hoxa1, Cyp26a1, and RARβ2 trended upward or were increased (2.3 ± 0.3-fold, p=0.007, Cyp26a1) in untreated Aldh2^−/− versus untreated WT ESCs (Figure 3E). Levels of these transcripts were not further increased by EtOH compared to the transcript levels in untreated Aldh2^−/− ESCs cells (Figure 3E). Furthermore, mRNA levels of the terminal differentiation marker Col4a1 were increased by EtOH in WT ESCs (2.37 ± 0.45-fold, p=0.038) but only trended upward in untreated Aldh2^−/− ESCs (Figure 3F).

To examine whether the attenuated, EtOH-mediated induction of differentiation-associated mRNAs in Aldh2^−/− ESCs was the result of specific increases in oxidative stress secondary to a defect in AcH clearance, we treated WT and Aldh2^−/− ESCs with the free radical scavenger, N-acetylcysteine (NAc), which is used at concentrations of 0.75–2 mM in various cell culture models to induce potent antioxidant activity without the deleterious effects on viability and morphology observed at doses greater that 5 mM (Shi et al, 2009; Hu et al., 2016). Addition of 2 mM NAc to EtOH-treated cells failed to restore the EtOH-mediated mRNA increases in Hoxa1 in Aldh2^−/− ESCs, ruling out a general oxidative stress signature in preventing Hoxa1 induction (Figure 3G). This series of experiments demonstrate that Aldh2^−/− ESCs do not efficiently respond to EtOH-induced differentiation signals.

ACSS2 is not required for induction of differentiation-associated transcripts by ethanol.

To analyze the roles of ACSS2 in the induction of differentiation-associated transcripts by EtOH, we measured transcript levels of Hoxa1, Cyp26a1, and RARβ2 ± 40 mM EtOH for 48 hours in WT and Acss2^−/− ESCs, respectively. EtOH increased these mRNAs in WT (3.7 ± 0.1-fold, p<0.0001, Hoxa1; 4.4 ± 0.7-fold, p=0.009, Cyp26a1; 2.0 ± 0.1-fold, p=0.007, RARβ2) and Acss2^−/− (2.5 ± 0.3-fold, p=0.03, Hoxa1; 3.5 ± 0.8-fold, p=0.049, Cyp26a1; 1.97 ± 0.3-fold, p=0.031, RARβ2) ESCs compared to the levels in untreated ESCs (Figure 4). Therefore, lack of ACSS2 does not reduce the ability of ESCs to induce differentiation-associated transcripts upon EtOH treatment.

Figure 4: Ablation of ACSS2 does not impair ethanol-mediated induction of differentiation-associated transcripts.

Fold changes in mRNA levels by EtOH in WT and Acss2^−/− ESCs at 48 hours. Y-axes vary with samples being analyzed, and mRNA levels are shown in arbitrary units. Error bars represent standard errors of independent experiments where n = 3 biological repeats. *, p≤0.05, **, p≤0.01, ***, p≤0.001, ****, p≤0.0001.

Differentiation-associated transcripts are increased by ethanol (EtOH), acetaldehyde (AcH), and 4-hydroxynonenal (4-HNE).

Both AcH and 4-HNE are reactive aldehyde species (RAS) that are substrates of the enzyme ALDH2 (Siems and Grune, 2003). We have previously observed increases in differentiation-associated transcripts upon AcH treatment (Serio et al., 2019). If the induction of differentiation-related mRNAs by AcH results from aldehyde damage, similar increases are expected in these mRNAs by other aldehyde substrates of ALDH2. We therefore tested whether treating WT ESCs with 1 μM 4-HNE or 1 mM AcH for 48 hours recapitulated the increases in Hoxa1, Cyp26a1, and RARβ2, compared to untreated ESCs. In WT ESCs, both EtOH and AcH similarly increased Hoxa1 (EtOH: 1.7 ± 0.2-fold, p=0.003; AcH: 1.8 ± 0.4-fold, p=0.044) and Cyp26a1 (EtOH: 2.8 ± 0.4-fold, p=0.003; AcH: 2.2 ± 0.5-fold, p=0.048) transcripts compared to untreated cells. Treatment with 4-HNE caused increases in Cyp26a1 mRNA (1.6 ± 0.2-fold, p=0.010) (Figure 5A). The genes Hoxa1, Cyp26a1, and RARβ2 are targets of retinoic acid-mediated transcription via retinoic acid receptor-γ (RARγ) (Langston et al.,1997; Kashyap et al., 2011). Furthermore, EtOH mediates ESC differentiation via retinol (ROL) uptake from media and subsequent metabolism (Serio et al., 2019). Therefore, we measured mRNAs of genes that encode proteins in ROL import and oxidation: the ROL transporter stimulated by retinoic acid 6 (Stra6) and the enzyme, retinol dehydrogenase 10 (Rdh10). Addition of EtOH, AcH, or 4-HNE increased transcripts of both Stra6 (EtOH: 1.7 ± 0.1-fold, p=0.0003; AcH: 1.4 ± 0.1-fold, p=0.016; 4-HNE: 1.5 ± 0.1-fold, p=0.006) and Rdh10 (EtOH: 2.5 ± 0.3-fold, p=0.008; AcH: 2.4 ± 0.5-fold, p=0.044; 4-HNE: 1.7 ± 0.1, p=0.002) compared to levels in untreated WT ESCs (Figure 5B). Therefore, the aldehyde substrates of ALDH2, AcH and 4-HNE, increase mRNA levels of differentiation-associated genes related to retinoid signaling.

Figure 5: Differentiation-associated transcripts are increased by acetaldehyde and 4-hydroxynonenal.

A, Fold changes in mRNA levels of genes involved in differentiation by 40 mM EtOH, 1 mM AcH, and 1μM 4-HNE in WT ESCs. B, Fold changes in mRNA levels of genes involved in RA synthesis by EtOH, AcH, and 4-HNE in WT ESCs. All qPCR experiments were conducted at 48 hours. Y-axes vary with samples being analyzed, and mRNA levels are shown in arbitrary units. Error bars represent standard errors of independent experiments where n = at least 3 biological repeats. *, p≤0.05, **, p≤0.01, ***, p≤0.001, ****, p≤0.0001.

RARγ is required for acetaldehyde-dependent increases in differentiation-associated transcripts.

To investigate whether Aldh2^−/− ESCs respond to differentiation signals, we differentiated WT and Aldh2^−/− ESCs along an extraembryonic endoderm lineage using 1 μM retinoic acid (RA) (Chen and Gudas, 1996). Hoxa1 and Stra6 were selected as readout genes, as Hoxa1 is a primary response gene that is critical for the initiation of differentiation via RARγ (Boylan et al., 1993), and Stra6 is required for the EtOH-induced uptake of ROL from the medium for generating RA (Serio et al., 2019). RA increased transcript levels of Hoxa1 and Stra6 in WT (16.2 ± 3.2-fold, p=0.003, Hoxa1; 60.5 ± 15.7-fold, p=0.019, Stra6) and Aldh2^−/− (10.96 ± 1.5-fold, p=0.0004, Hoxa1, 52.5 ± 14.0-fold, p=0.023, Stra6) ESCs (Figure 6A). Therefore, Aldh2^−/− ESCs adequately respond to RA-induced differentiation signals.

Figure 6: RARγ is required for acetaldehyde-mediated increases in differentiation-associated transcripts.

A, Fold changes in Hoxa1 and Stra6 mRNAs by 1 μM RA in WT and ALDH2^−/− ESCs. B, Fold changes in Hoxa1 and Stra6 mRNAs by 40 mM EtOH and 1 mM AcH in WT, Aldh2^−/−, and RARγ^−/− ESCs. qPCR experiments were conducted at 48 hours. Y-axes vary with samples being analyzed, and mRNA levels are shown in arbitrary units. Error bars represent standard errors of independent experiments where n = at least 3 biological repeats. *, p≤0.05, **, p≤0.01, ***, p≤0.001. C, Model for AcH regulation of ESC differentiation. EtOH is converted to AcH, which is oxidized by ALDH2 to acetate in ESCs. One fate of acetate is its metabolism to acetyl CoA by ACSS2. Whereas ablation of ACSS2 to decrease acetate conversion to acetyl CoA does not cause increased transcript levels of differentiation-associated genes ± EtOH, AcH induces transcript levels of the same genes. These AcH-mediated increases are dependent on RARγ to mediate ESC differentiation along an extraembryonic endoderm lineage. Although EtOH-mediated differentiation via RARγ likely occurs via its conversion to AcH, partial activation of RARγ-dependent transcription by EtOH via alternate mechanisms cannot be entirely ruled out (dotted arrow).

We have previously shown that EtOH-mediated differentiation of ESCs requires RA:RARγ-dependent transcription (Serio et al., 2019). To determine whether the effects of AcH on ESC differentiation are likewise dependent on RA signaling, we used RARγ^−/− ESCs previously generated in our lab (Laursen and Gudas, 2018). We treated WT, Aldh2^−/−, and RARγ^−/− ESCs with either 40 mM EtOH or 1 mM AcH to determine whether AcH-dependent increases in differentiation-associated transcripts proceed via RARγ and are thus dependent on RA signaling. In WT ESCs, both EtOH and AcH increased Hoxa1 (EtOH: 1.7 ± 0.1-fold, p=0.005; AcH: 1.6 ± 0.2-fold, p=0.041) and Stra6 (EtOH: 2.1 ± 0.04-fold, p=0.001; AcH: 3.8 ± 0.1-fold, p=0.002) mRNA levels compared to untreated WT ESCs (Figure 6B). In untreated Aldh2^−/− ESCs, mRNA levels of both Hoxa1 (1.4 ± 0.1-fold, p=0.024) and Stra6 (1.6 ± 0.09-fold, p=0.026) were increased compared to those in untreated WT ESCs, and were not further increased by EtOH (Figure 6B). In contrast, Hoxa1 and Stra6 transcript levels were lower in untreated RARγ^−/− versus WT ESCs (−3.5 ± 0.07-fold, p=0.0004, Hoxa1; −3.0 ± 0.06-fold, p=0.007, Stra6), and neither EtOH nor AcH increased these transcripts in RARγ^−/− ESCs (Figure 6B). Thus, we conclude that RARγ is required for the AcH-mediated increases in Hoxa1 and Stra6 mRNA in WT ESCs (Figure 6C, Model).

DISCUSSION

Acetaldehyde is primarily metabolized by ALDH2 in embryonic stem cells.

Using knockout models of ALDH2 and ALDH1A2, we showed that AcH is preferentially metabolized by ALDH2 in ESCs (Figure 2B–D). ALDH2, with a Km of 0.2 μM, metabolizes AcH when it is adequately expressed in cells (Klyosov et al., 1996; Vasiliou et al., 2004). In comparison, ALDH1A2, which is also expressed in ESCs and is the first retinoid-oxidizing aldehyde dehydrogenase enzyme expressed in the embryo (Niederrether et al., 2002), has a low affinity for AcH (Km = 650 μM) (Wang et al., 1996), thus is unlikely to efficiently metabolize AcH if ALDH2 is expressed in the same cell.

The importance of ALDH2 in AcH detoxification in humans is best illustrated by analyzing the phenotypes of those individuals who lack its enzymatic activity in at least one allele (ALDH2*2). A large population of East Asian descent is genetically predisposed to toxic accumulation of AcH from alcohol consumption from harboring this dominant genetic variant (Goedde et al. 1992), which has been directly linked to increased risk of cancers (Muto et al., 2000; Peng et al., 2007) and neurologic sequelae (Wang et al., 2008). The ALDH2*2 variant is a glutamate to lysine (E487K) substitution that inactivates ALDH2 by causing conformational changes to its active site and greatly weakening its affinity for the NAD⁺ cofactor (Steinmetz et al., 1997). The severe phenotype caused by the ALDH2*2 variant (Goedde et al., 1992; Muto et al., 2000; Wang et al., 2008) upon alcohol exposure demonstrates that ALDH2 is a critical catalyst for EtOH detoxification in humans.

Despite the important role of ALDH2 in EtOH detoxification, some studies have shown that in certain situations, ALDH1A2 can also metabolize AcH (Kot-Leibovich, 2009; Shabtai et al., 2018). For example, Xenopus embryos do not express Aldh2 during gastrulation and EtOH could instead be metabolized by ALDH1A2 during this window of time (Shabtai et al., 2018). In ESCs, we showed that without ALDH2, most AcH oxidation was absent, while deletion of ALDH1A2 caused little change compared to the oxidation in WT extracts (Table 2, Figure 2B). Therefore, AcH is primarily metabolized by ALDH2 in our ESC model and thus ALDH2 is a salient protein to be targeted for ablation in our strategy to dissect the functional consequences of EtOH metabolism in ESCs. While the ALDH2 agonist Alda1 clearly induced the activity and the ALDH2 antagonist efficiently inhibited the activity of AcH metabolism (Figure 2C), the same pharmacological agents had no marked effects in Aldh2^−/− cells (Figure 2D), which further confirms the role of ALDH2 in EtOH metabolism.

Loss of ALDH2, but not ACSS2, results in a reduction of pluripotency-related mRNAs and inhibits the response of embryonic stem cells to ethanol.

Because EtOH has strong teratogenic potential, understanding its effects in stem cell differentiation is crucial. As we have previously shown that both EtOH and AcH increase transcript levels of differentiation-associated genes (Serio et al., 2019), we disrupted AcH oxidation to acetate by knocking out both alleles of the Aldh2 gene (Figure 1). Several pluripotency factors exhibited reduced mRNA levels in untreated Aldh2^−/− compared to untreated WT ESCs. Of the four pluripotency-related genes we selected for assessment in the ALDH2-KO cell line, the genes that displayed the greatest reductions in transcript levels in Aldh2^−/− ESCs were Nanog and Klf4 (Figure 3D). Interestingly, these genes were also decreased by 40 mM EtOH in WT ESCs (Serio et al., 2019). Oct4 remained high in both EtOH-treated WT (Serio et al., 2019) and untreated Aldh2^−/− ESCs (Figure 3D). This is consistent with data reported by Ogony and colleagues demonstrating that EtOH alters the ratio of Oct4/Sox2 by decreasing Sox2 but not Oct4 expression (Ogony et al., 2013). An increased Oct4/Sox2 ratio was associated with a redirection of differentiation away from a neuroectodermal cell fate by EtOH (Ogony et al, 2013).

In contrast to the marked reductions in several pluripotency markers in untreated Aldh2^−/− ESCs, of the four differentiation-associated genes that we initially assessed (Hoxa1, Cyp26a1, RARβ2, and Col4a1), only Cyp26a1 transcript levels were increased in the untreated Aldh2^−/− ESCs (Figure 3E–F). Our results suggest that although Aldh2^−/− ESCs display decreases in mRNAs of some pluripotency factors, they are not fully differentiated in the absence of added EtOH.

Deletion of both alleles of ACSS2 had no effect on proliferation (Figure 3C), loss of pluripotency-related mRNAs (Figure 3D), or mRNA increases of Hoxa1, Cyp26a1, and RARβ2 (Figure 4). We have previously demonstrated that addition of acetate in the medium of cultured WT ESCs does not increase differentiation-associated transcripts (Serio et al., 2019). In congruence with those findings, we demonstrated here that ACSS2 ablation does not result in decreased pluripotency-related transcripts, and instead caused Nanog levels to increase by 2.3-fold (Figure 3D). Our results therefore indicate that AcH, not acetate or a downstream metabolite of acetate, causes decreases in pluripotency-related mRNAs.

Context-dependent changes in pluripotency and differentiation-related factors have been demonstrated by using EtOH to investigate effects of stem cell differentiation (Vandevoort et al., 2011; Arzumnayan et al., 2009; Ogony et al., 2013; Veazey et al., 2013). For example, treatment of ESCs with a wide range of EtOH doses, from 0.1%−1% (17.4 mM-174 mM), for two weeks caused a loss of pluripotency and spontaneous differentiation, as measured by alkaline phosphatase and TRA-1–81 staining, despite some pluripotency-related genes remaining highly expressed (VandeVoort et al., 2011). In contrast to these findings, a 48-hour treatment with 100 mM EtOH followed by 6 days of differentiation via Leukemia inhibitory factor (LIF) removal caused a delay in the loss of pluripotency factor-related mRNAs, suggesting that differentiation was also being delayed by EtOH (Arzumnayan et al., 2009). Furthermore, EtOH can interfere with directed differentiation toward specified cell lineages by diverting differentiation away from neuroectodermal (Ogony et al., 2013; Sánchez-Alvarez, 2013), hepatic (Gao et al., 2014), or cardiac (Wang et al., 2017) fates, indicating a broad range of defects resulting from EtOH exposure during differentiation.

We showed here that differentiation-related transcripts were induced in ESCs upon exposure to EtOH within 48 hours (Figure 3E–G). These results are consistent with a model of EtOH stimulating stem cell differentiation (VandeVoort et al., 2011; Serio et al., 2019). The delay in differentiation observed by Arzumnayan et al. (Arzumnayan et al., 2009) may have been unique to the method of differentiation being utilized, via depletion of LIF in cell culture medium. Our lab has shown several transcriptional differences between ESC differentiation by LIF removal compared with differentiation by adding exogenous RA (Lane et al., 1999; Martinez-Ceballos and Gudas, 2008). Depletion of LIF from culture medium causes ESCs to express the primitive ectoderm marker, Fgf5 (Hébert et al., 1990; Martinez-Ceballos and Gudas, 2008), which increases over time to drive cells along a neuroectodermal fate (Shimozaki et al., 2003). In contrast, RA treatment of ESCs suppresses Fgf5 mRNAs and differentiates cells along an extraembryonic endodermal, or epithelial, lineage (Chen and Gudas, 1996; Martinez-Ceballos and Gudas, 2008). EtOH has been demonstrated to inhibit neuroectodermal differentiation (Ogony et al., 2013; Sánchez-Alvarez et al., 2013; Vangipuram and Lyman, 2012). Thus, it is expected that ESC differentiation secondary to LIF depletion would also be inhibited by EtOH.

Aldehydes, including acetaldehyde and 4-hydroxynonenal, recapitulate the ethanol-mediated differentiation phenotype of embryonic stem cells.

Reactive aldehyde species (RAS), such as AcH, are becoming increasingly appreciated for their roles in initiating signaling responses (Raza and John, 2006; Poganik et al., 2018). Adducts formed from covalent modifications with the electrophilic groups may not only act as toxic stimuli but also serve as relevant post-translational modifications for proteins to affect downstream signaling pathways (Poganik et al., 2018). What effects these changes have on stem cell biology are still largely unknown at a mechanistic level.

As a reactive aldehyde species, 4-HNE, like AcH, serves as a substrate of ALDH2, and 4-HNE is associated with differentiation in stem cell and cancer cell models (Moneypenny and Gallagher, 2005; Rinaldi et al., 2001). We show here that both exogenous AcH and 4-HNE can increase the mRNA levels of several differentiation-related genes (Cyp26a1, Stra6, and Rdh10) compared to the levels in untreated WT ESCs (Figure 5).

The differentiation-associated transcript induction following both AcH and 4-HNE treatment implies that reactive aldehyde species (RAS) such as AcH and 4-HNE may disrupt stem cell function and predispose ESCs to precocious differentiation. We did not find increases in differentiation-associated genes upon treatment of Aldh2^−/− ESCs with EtOH compared to untreated Aldh2^−/− cells (Figures 3E, 6B). However, basal levels of these transcripts displayed marked variability and often trended higher in untreated Aldh2^−/− ESCs (Figures 3E, 6B). Because cells produce and may accumulate 4-HNE, AcH, and other RAS endogenously (Liu et al., 2018; Lopachin et al., 2009), it is tempting to speculate that ESCs that lack ALDH2 may be vulnerable to differentiation by endogenous aldehydes and therefore do not respond as robustly to external RAS insults as stem cells containing functional ALDH2. In this manner, a ceiling effect is possible whereby administering additional RAS does not increase transcript levels of differentiation-association genes beyond those levels that are induced by endogenous RAS in the absence of ALDH2-dependent metabolism.

Reactive aldehyde species, including AcH and 4-HNE, cause secondary accumulation of reactive oxygen species (ROS) in cells that affects a wide range of additional non-specific, toxic events (Raza and John, 2006; Seitz and Stickel, 2007). However, we showed that the effects on differentiation-associated mRNAs are unrelated to the ability of AcH to increase ROS in ESCs, as N-acetylcysteine, a radical scavenger, did not prevent the EtOH-mediated increases in Hoxa1 transcript levels, nor did it affect mRNA levels in Aldh2^−/− ESCs (Figure 3G). This suggests that the EtOH-dependent induction of differentiation-related transcripts does not require secondary responses resulting from increased ROS in cells.

RARγ is required for both ethanol- and acetaldehyde-dependent differentiation.

AcH-induced cellular stress has been linked to chromosomal damage in stem cells (Garaycoechea et al., 2018) and embryonic lethality and impaired hematopoiesis in Aldh2^−/− mice (Oberbeck et al., 2014). However, the effects of AcH on RA:RAR signaling have not been previously characterized, and elucidation of this potential interaction would provide opportunities to further scrutinize the mechanisms underlying EtOH-related toxicities in individuals harboring the ALDH2*2 gene variant.

The RA signaling pathway is critical to stem cell differentiation and embryogenesis (Gudas and Wagner, 2011). We previously showed that EtOH differentiates ESCs along an extraembryonic endoderm lineage via RA:RARγ-dependent transcriptional signaling (Serio et al., 2019). Therefore, we aimed to test whether loss of RARγ^−/− is sufficient to prevent AcH-mediated differentiation. We demonstrate here that ablation of RARγ prevents both AcH-and EtOH-mediated increases in the differentiation-related transcripts Hoxa1 and Stra6. These findings are consistent with studies in animal models demonstrating that AcH recapitulates the developmental defects caused by EtOH (Fort et al., 2003; Lee et al., 2005; Reimers et al., 2004). To our knowledge, this is the first evidence of AcH altering RARγ-dependent differentiation signals.

Importantly, the STRA6 transporter protein is required for retinol import into several cell types (Bouillet et al., 1997; Laursen et al., 2015), and without activity of the Stra6 RARE retinol is not imported into ESCs to subsequently generate RA (Serio et al., 2019). How AcH interacts with STRA6 remains unknown. The structure of the STRA6 transporter has recently been elucidated (Chen et al., 2016), and may provide insights into the regulation of this critical “gatekeeper” protein. STRA6 was crystallized bound to two calmodulin molecules, hinting at potential calcium-dependent regulation of STRA6 function (Chen et al., 2016; Tidow and Nissen, 2013). Further studies elucidating the structure-function relationship of the STRA6 transporter and its regulation may help to illuminate a potential connection between protein and DNA damage downstream of aldehyde stress and STRA6 activation.

CONCLUSIONS

Our findings illustrate the role of EtOH on stem cell differentiation through its conversion to AcH and subsequent metabolism by ALDH2. Both EtOH and AcH stimulate increases in differentiation-associated genes, including Hoxa1 and Stra6. This effect was recapitulated with 4-HNE, which, like AcH, is a substrate of ALDH2. Aldh2^−/− ESCs display decreases in pluripotency-related transcripts compared to WT ESCs. The response of differentiation-related transcripts to EtOH in Aldh2^−/− ESCs was also inhibited, although this finding may be related to higher basal levels of some of these mRNAs in untreated Aldh2^−/− versus WT cells. In contrast, the pluripotency-related phenotype of Acss2^−/− ESCs was largely unchanged, and these cells responded to EtOH similarly to WT ESCs, by increasing mRNA levels of Hoxa1, Cyp26a1, and RARβ2. Importantly, the induction of differentiation-associated transcripts by both EtOH and AcH require RARγ. We conclude that AcH mediates ESC differentiation via activation of RARγ and transcription of RARγ-target genes.

LIST OF ABBREVIATIONS:

AcH

acetaldehyde

DMSO

dimethyl sulfoxide

ESC

embryonic stem cell

EtOH

ethanol

gRNA

guide RNA

NAc

N-acetylcysteine

qPCR

quantitative polymerase chain reaction

RA

retinoic acid

RAS

reactive aldehyde species

ROL

retinol

ROS

reactive oxygen species

WT

wild type

4-HNE

4-hydroxynonenal