Previous studies demonstrated that in utero caffeine treatment at embryonic day (E) 8.5 alters DNA methylation patterns, gene expression, and cardiac function in adult mice. To provide insight into the mechanisms, we examined cardiac gene and microRNA (miRNA) expression in cardiomyocytes shortly after exposure to physiologically relevant doses of caffeine. In HL-1 and primary embryonic cardiomyocytes, caffeine treatment for 48 h significantly altered the expression of cardiac structural genes (Myh6, Myh7, Myh7b, Tnni3), hormonal genes (Anp and BnP), cardiac transcription factors (Gata4, Mef2c, Mef2d, Nfatc1), and microRNAs (miRNAs; miR208a, miR208b, miR499). In addition, expressions of these genes were significantly altered in embryonic hearts exposed to in utero caffeine. For in utero experiments, pregnant CD-1 dams were treated with 20–60 mg/kg of caffeine, which resulted in maternal circulation levels of 37.3–65.3 μM 2 h after treatment. RNA sequencing was performed on embryonic ventricles treated with vehicle or 20 mg/kg of caffeine daily from E6.5-9.5. Differential expression (DE) analysis revealed that 124 genes and 849 transcripts were significantly altered, and differential exon usage (DEU) analysis identified 597 exons that were changed in response to prenatal caffeine exposure. Among the DE genes identified by RNA sequencing were several cardiac structural genes and genes that control DNA methylation and histone modification. Pathway analysis revealed that pathways related to cardiovascular development and diseases were significantly affected by caffeine. In addition, global cardiac DNA methylation was reduced in caffeine-treated cardiomyocytes. Collectively, these data demonstrate that caffeine exposure alters gene expression and DNA methylation in embryonic cardiomyocytes.
- cardiac development
- differential gene expression
- differential exon usage
caffeine is a stimulant widely consumed by humans (23). In addition to adults and children, fetuses are frequently exposed to caffeine (6). Caffeine is consumed during the first month of pregnancy by 60% of American women, and 16% of pregnant mothers consume 150 mg or more per day (6). Caffeine freely crosses the placenta, and caffeine levels in embryo and fetus reach 90% of maternal levels (6, 24). The half-life of caffeine is also much longer in the fetus than in the adult (6, 24).
There are conflicting data as to whether in utero caffeine exposure is harmful to the developing human fetus (4, 7, 13, 45, 49). However, reports from large cohorts of human subjects indicate a correlation between prenatal caffeine consumption and decreased birth weight (42). Although a number of compounds to which fetuses are exposed are associated with long-term effects in adulthood (41), little is known of the long-term consequence of prenatal caffeine exposure in humans.
Recently, we found that in utero caffeine treatment leads to adverse effects on embryonic and adult hearts (8, 9, 38, 47, 48). Caffeine exposure during early gestation of mice leads to reduced embryonic myocardium mass (47) and exerts long-term adverse effects on adult cardiac morphology and function (9, 47). The adult offspring of caffeine-exposed dams manifest increased left ventricular mass, decreased cardiac contractility, and reduced cardiac output, which are features of concentric cardiac hypertrophy (9).
In the adult myocardium, cardiac DNA methylation is reduced by embryonic caffeine exposure, and DNA methylation patterns and expression of specific cardiac genes, including myocyte enhancer factor 2c (Mef2c), myosin heavy chain 6 (Myh6) and 7 (Myh7), are altered (9). Others have found that prenatal caffeine exposure leads to persistent activation of renin angiotensin system, leading to hypertension and adverse cardiac remodeling in adult offspring (43). Collectively, these data suggest that fetal caffeine exposure may lead to long-term cardiovascular abnormalities in adulthood, but the mechanisms resulting in these changes remain unclear.
Caffeine acts as a nonselective antagonist of adenosine receptors (ARs) (24). A1ARs are the dominant AR subtype in transducing the prenatal effects of adenosine (9). In embryonic hearts, A2aARs are also present, whereas A2bARs and A3ARs are difficult to detect (8). Selective AR antagonists include the A1AR antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX) and the A2aAR antagonist SCH58261.
Cardiac development and disease progression is governed by a network of specific genes (28, 52). Cardiac hypertrophy is associated with the altered expression of cardiac structural genes (Myh6 and Myh7), cardiac hormones [atrial natriuretic peptide (ANP) and B-type natriuretic peptide (BNP)], and transcription factors [GATA-binding protein 4 (Gata4), Mef2c, myocyte enhancer factor 2d (Mef2d), nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 1 (Nfatc1)] (28). The ratio of Myh6 and Myh7 influences cardiac function, and increased Myh7 expression in the adult heart is a common feature of cardiac hypertrophy and heart failure (11, 25). Troponin is integral for cardiac muscle contraction, and subtypes troponin I and T are diagnostic markers of cardiac muscle injury (3, 35).
Several microRNAs (miRNAs) are located within Myh6 (miR208a, intron 29), Myh7 (miR208b, intron 31), and Myh7b (miR499, intron19) and are coexpressed with the corresponding myosin genes (36). Upregulation of miR208a is found in patients with cardiomyopathy, cardiac fibrosis, and heart failure (36). miR208a stimulates the expression of Myh7 and Myh7b (36).
Although caffeine is consumed by millions of pregnant women, little is known about its effects on cardiomyocytes at the molecular level during development. To provide new insights in caffeine action, we assessed if caffeine alters the expression of critical transcription factors (Gata4, Mef2c, Nfatc1) and regulatory miRNAs (miR208a, miR208b, miR499) in cardiomyocytes, which could lead to abnormal expression of cardiac genes including Myh6, Myh7, Myh7b, troponin I type 3 (cardiac) (Tnni3), ANP, and BNP.
MATERIALS AND METHODS
Caffeine, 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), and SCH58261 were purchased from Sigma-Aldrich (St. Louis, MO). Antibodies MYH6 (K-13, sc-168676), MYH7 (A4.951, sc-53090), goat anti-mouse IgG-horseradish peroxidase (sc-2005), and donkey anti-goat IgG-horseradish peroxidase (sc-2020) were purchased from Santa Cruz Biotechnology (Dallas, TX). Antibodies for monoclonal anti-α-actinin (sarcomeric) and β-actin were purchased from Sigma-Aldrich. Alexa Fluor 546 phalloidin, 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI), and goat anti-mouse IgG 488 were ordered from Life Technologies (Grand Island, NY).
Animals and treatment.
All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Florida. CD-1 mice were purchased from Charles Rivers Laboratories International (Wilmington, MA). Timed mating was performed, and the day a vaginal plug was observed was designated as embryonic day 0.5 (E0.5). For isolation of primary cardiomyocytes, embryonic hearts were collected at E13.5.
For studies of gene expression and DNA methylation, pregnant dams were injected intraperitoneally daily from E6.5 to 9.5 or from E6.5 to 10.5 with 0.9% NaCl (vehicle), 10, 20, or 60 mg/kg of caffeine. On E10.5, embryos were dissected and counted. For dams treated from E6.5 to 9.5, ventricles were collected 24 h after the last caffeine treatment. For dams treated from E6.5 to 10.5, ventricles were collected 2 h after the last caffeine treatment on E10.5. Although the half-life of caffeine is longer in embryo (12–24 h) than in adult due to absence of the metabolic enzyme CYP1A2 for caffeine in placenta and fetus (1, 6, 24), the level of caffeine drops over time in the embryos. These two treatment regimens in pregnant dams were used to determine how long the effects of caffeine have on gene expression. Gross morphology of the embryos was observed under a dissecting microscope. Ventricles were isolated from each embryo and all ventricles from the same litter were pooled for total RNA isolation.
HL-1 cell culture and treatment.
HL-1 cardiomyocytes were obtained from Dr. William C. Claycomb (12) and cultured with Claycomb medium (Sigma-Aldrich) supplemented with 10% fetal bovine serum (Sigma-Aldrich), 100 U/ml penicillin-streptomycin (Sigma-Aldrich), 2 mM l-glutamine (Sigma-Aldrich), and 0.1 mM norepinephrine (Sigma-Aldrich). Cells were treated with drugs or vehicle [dimethyl sulfoxide (DMSO) or phosphate-buffered saline (PBS)] for 48 h, with daily changes of medium supplemented with drugs or vehicle.
Isolation, culture, and treatment of primary embryonic cardiomyocytes.
Cardiomyocytes were isolated from mouse E13.5 embryonic ventricles, as previously described (40). Briefly, hearts were dissected from embryos and atria were removed. The remaining ventricular tissue from each litter of embryos was pooled and digested with papain solution that contained 4 mg/ml papain (Sigma-Aldrich), 1.1 mM EDTA, 5.5 mM l-cystein, and 67 μM β-mercaptoethanol dissolved in the Earle's Balanced Salt Solution (EBSS; Sigma-Aldrich). Digestion was stopped with 1 mg/ml soybean trypsin inhibitor solution (dissolved in EBSS; Sigma-Aldrich). Cells were suspended in the Dulbecco's modified Eagles media (DMEM; Life Technologies) supplemented with 10% inactivated fetal bovine serum (Life Technologies/GIBCO), 2 mM l-glutamine, and antibiotic-antimycotic solution (Life Technologies/GIBCO). Cells were placed in flasks coated with 2% gelatin for 1 h to allow differential attachment to occur. The nonattached cells, which are predominantly cardiomyocytes, were transferred to new flasks or plates coated with 2% gelatin.
After overnight incubation, the culture medium was changed and drugs or vehicle (DMSO or PBS) were added. Medium with drugs or vehicle was changed in 24 h, and the experiments were stopped at 48 h. For each treatment group, three to five batches of cardiomyocytes isolated from different litters of embryos were used. For total RNA isolation, cells were cultured in 12.5-mm2 flasks, trypsinized, and collected in lysis buffers.
Quantitative real-time PCR (qPCR) analysis.
Total RNA was isolated with the RNeasy Plus Mini Kit (Qiagen, Valencia, CA), according to the manufacturer's protocol. cDNA was synthesized using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). qPCR primer pairs are listed in Table 1. Myh6 (PPM04500A) and Myh7 (PPM67019A) primers were designed and synthesized by Qiagen SABiosciences. β-Actin primers were used as an internal control. SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) was used to perform qPCR analysis with a GeneAmp 7300 Real Time PCR System (Applied Biosystems) (19, 20). Each sample was measured in two separate reactions on the same plate. Amplification efficiencies of the target gene and β-actin primer pairs were tested, and there were no statistical differences in their efficiencies. The 2−ΔΔCT method was used for relative quantification.
qPCR analysis of miRNA expression.
Total RNA including small RNA was isolated with the mirVana miRNA Isolation Kit (Life Technologies). miRNA was reverse transcribed to cDNA with the miScript II RT Kit (Qiagen). Predesigned primers for the mature miRNAs, miR208a, miR208b, miR499, and RNU6-2 were purchased from Qiagen. Small nuclear RNA (snRNA) RNU6-2 was used as an internal miRNA control. qPCR was performed with the miScript SYBR Green PCR Kit (Qiagen) in a GeneAmp 7300 Real Time PCR System (Applied Biosystems). Each sample was measured in two separate reactions on the same plate, and data were averaged before analysis (44, 54). The 2−ΔΔCT method was used for relative quantification.
Cells were homogenized in RIPA buffer (Thermo Scientific, Rockford, IL) supplemented with 1× Complete Protease Inhibitor Cocktail Tablets (Roche, Basel, Switzerland) by a Sonic Dismembrator (Thermo Scientific). Protein was quantitated with the Pierce BCA Protein Assay kit (Thermo Scientific). Protein (40 μg/lane) was separated on a Criterion Tris·HCl Precast Gel (7.5% polyacrylamide; Bio-Rad) followed by transfer to a nitrocellulose membrane (Bio-Rad) using the Bio-Rad electrotransfer system (27, 50). Blots were probed with antisera to MYH6 or MYH7 and against β-actin, as a control for sample loading. After detection with the Pierce ECL Plus Western Blotting Substrate (Thermo Scientific) in a ChemiDoc XRS+ Imaging System (Bio-Rad), band densitometry was quantitated with the Image Lab software (Bio-Rad).
Cardiomyocytes cultured on coverslips were fixed overnight in 4% paraformaldehyde (PFA) and washed in PBS three times. Cells were permeabilized with 0.5% Triton X-100 in PBS, blocked with 2% bovine serum albumin-2% goat serum, and incubated overnight with α-actinin antibody (55). After washing with blocking buffer was completed, goat anti-mouse IgG 488, Alexa Fluor 546 Phalloidin, and DAPI were added to the cells. After being washed with PBS, coverslips were mounted onto glass slides with Fluoromount G (Electron Microscopy Sciences, Hatfield, PA). Cells were imaged with an Axio Vert.A1 inverted microscope (Zeiss, Jena, Germany).
Cell viability assay.
HL-1 cells were seeded in 96-well plates at a density of 20,000 cells/well and treated with various concentrations of caffeine for 48 h. Drugs were changed daily. Cell viability was measured with a CellTiter-Glo Luminescent Cell Viability Assay kit (Promega, Madison, WI) (53). This experiment was repeated three times.
Caffeine levels were measured in the serum of dams by an ELISA assay (Neogen, Lexington, KY), as described (47). CD-1 female mice were treated with 10, 20, and 60 mg/kg ip caffeine at E10.5, and blood serum was collected 2 h later.
Global DNA methylation.
Genomic DNA was isolated with the DNeasy Blood & Tissue Easy Kit (Qiagen) according to manufacter's instructions. DNA was treated with RNAase to remove RNA contaminants. The MethyFlash Methylated DNA Quantification Kit (Epigentek, Farmingdale, NY) was used to quantitate the percentage of methylated cytosine in genomic DNA samples (14, 18). After the colorimetric reaction, the methylated cytosine was measured with a Synergy HT Multi-Mode Microplate Reader (BioTek, Winooski, VT).
Illumina transcriptomic RNA sequencing (RNA-seq).
Pregnant dams received daily intraperitoneal injection from E6.5 to 9.5 with 0.9% NaCl (vehicle) or 20 mg/kg caffeine. On E10.5, ventricles were dissected from the embryos, and total RNA was isolated with RNeasy Plus Mini Kit (Qiagen) according to the manufacturer's protocol. cDNA library preparation and Illumina sequencing were performed by the Yale Center for Genome Analysis (Yale University, New Haven, CT). Briefly, total RNA was quantitated using the Qubit RNA assay and Qubit 2.0 Flourometer (Life Technologies). RNA integrity was confirmed using the Agilent RNA Pico assay reagents and Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA). One microgram of total RNA was used for the Illumina TruSeq v2 high throughput library construction procedure (Illumina, San Diego, CA). Briefly, mRNA was enriched using oligo-dT magnetic beads, fragmented, and reverse transcribed to cDNA. The resulting DNA was prepared for sequencing by blunt end repair, 3′ adenylation, multiplex compatible adapter ligation (containing TruSeq indexes), and PCR amplification. Library validation was performed using the Agilent Bioanalyzer 2100 followed by quantitation using the Qubit HS DNA assay (Life Technologies) and qPCR kit for Illumina (Kapa Biosystems, Wilmington, MA).
Five Illumina-adapted libraries, which included two vehicle-treated and three caffeine-treated samples, were pooled at equal molar ratio and clustered via cBot (Illumina) on a single lane of a TruSeq PE Cluster Kit v3 flowcell (Illumina). Single-ended 1 × 75 bp sequencing was carried out on an Illumina HiSeq2000 sequencer with TruSeq SBS Kit v3 chemistry (Illumina). All RNA-seq data were uploaded to the Gene Expression Omnibus (GEO) and can be accessed via http://www.ncbi.nlm.nih.gov/geo/; accession number GSE56902.
RNA sequencing data analysis for differential gene expression.
RNA sequencing (RNA-seq) data were uploaded to the galaxy platform https://main.g2.bx.psu.edu/ (5). RNA-seq reads were trimmed, filtered, and mapped to the mouse genome (mm10) with the Tophat for Illumina tool. The resulting Binary Sequence Alignment/Map (BAM) files were analyzed with Partek Genomics Suite (GS) version 6.11 (Partek, St. Louis, MO) for read counts. Refseq Transcripts (2013-05-10) and Ensembl Transcripts release 71 databases were used for gene and transcript annotation. Pearson test, principle component analysis, and dendrogram clustering were performed to verify the correlation of samples.
Differential expression (DE) of gene and transcript reads between treatments was analyzed with R package EdgeR (39). Low expressed genes/transcripts were filtered, and the remaining data were normalized with the Trimmed Mean of M-values (TMM) method. DE was determined using the exact test. Genes/transcripts with false discovery rate (FDR) <0.05 and absolute fold change >1.5 were considered as significant. DE genes were defined as genes with altered expression at either gene or transcript level. Unique DE genes were identified by combining the results from the Refseq and Ensembl annotations. Functional ontology was conducted with The Database for Annotation, Visualization and Integrated Discovery (DAVID) http://david.abcc.ncifcrf.gov/ (29), MetaCore Data-mining and Pathway Analysis (Thomson Reuters), and Ingenuity Pathway Analysis (IPA, Qiagen). Gene networks were generated with IPA.
RNA-seq data analysis for differential exon usage.
RNA-seq data were trimmed, filtered, and mapped to the mouse genome (Ensembl GRCm38.75.dna) with the Tophat2 tool in the galaxy platform https://main.g2.bx.psu.edu/. Differential exon usage (DEU) analysis was performed with R package DEXSeq to identify changes in the relative usage of exons caused by in utero caffeine treatment (2). Functional ontology was conducted with DAVID.
All experiments were performed at least three times. Results were analyzed using GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA). Data are presented as means ± SE. Statistical differences between treatment groups were determined using Student's t-test (P < 0.05) or one-way ANOVA followed by Newman-Keuls post hoc test (P < 0.05). For qPCR analysis, statistical differences between treatments were determined on the linearized 2−ΔCT values. Change in gene and miRNA expression was calculated with the 2−ΔΔCT method. The half-maximal inhibitory concentration (IC50) for cell survival by caffeine was calculated by using nonlinear regression to fit the data to the log(inhibitor) versus response (variable slope) curve.
Caffeine increases the expression of cardiac structural genes, transcription factors, and miRNAs in HL-1 cells.
We first examined caffeine effects on cardiac gene and miRNA expression in HL-1 cardiomyocytes. HL-1 cells are an adult murine cardiac cell line that expresses ARs (12). HL-1 cells were treated with increasing concentrations of caffeine for 48 h. Caffeine treatment increased mRNA expression of the cardiac structural genes Myh6, Myh7, Myh7b, and Tnni3 in a dose-dependent manner (Fig. 1A), but it had no effects on Tnnt2 and Tnnc1 expression. Caffeine significantly increased the ratio of Myh6 to Myh7 expression (Fig. 1D).
Caffeine treatment also altered the expression of cardiac hormone genes, with decreased expression of Anp and increased expression of Bnp (Fig. 1A). In addition, caffeine significantly increased the expression of cardiac transcription factors Gata4, Mef2c, Mef2d, Nfatc1-b, and Nfatc1-c (Fig. 1B) and cardiac miRNAs, including miR208a, miR208b, and miR499 (Fig. 1C).
Western blotting results showed that caffeine treatment increased the protein level of MYH6 (Fig. 1, E and F). In contrast, protein expression of MYH7 was not detectable in HL-1 cells.
Caffeine alters the morphology of cardiomyocytes.
HL-1 cells and primary embryonic cardiomyocytes were treated with increasing concentrations of caffeine and stained with phalloidin, anti-sarcomeric α-actinin, and DAPI to determine morphological responses to caffeine treatment in vitro. Using the CellTiter-Glo Luminescent Cell Viability Assay kit, we determined the IC50 of caffeine for HL-1 cell survival to be 3.145 mM. In addition to a reduction in cell numbers, microscopy showed that caffeine treatment led to a change in cardiac cell morphology (Figs. 2 and 3).
Caffeine alters expression of cardiac structural genes, transcription factors, and miRNAs in primary embryonic cardiomyocytes.
After studies of HL-1 cells, we examined the influence of caffeine on gene expression in mouse primary embryonic cardiomyocytes. Our isolation protocol yielded about 600,000 cells per E13.5 embryonic heart, and immunostaining with α-actinin antibody indicated that 86% of the cells were cardiomyocytes.
Similar to HL-1 cells, caffeine treatment of primary cardiomyocytes increased mRNA expression of cardiac structural genes Myh6, Myh7, Myh7b, and Tnni3, cardiac hormones Anp and Bnp, and transcription factors Gata4 and Mef2c in a dose-dependent manner (Fig. 4, A and B). The Myh6-to-Myh7 gene expression ratio was significantly increased after caffeine treatment (Fig. 4D). Caffeine treatment significantly reduced the expression of miR208a, whereas it increased the expression of miR208b and miR499 (Fig. 4C), similar to that observed in HL-1 cells.
Caffeine alters expression of cardiac structural genes, transcription factors, and miRNAs in embryonic heart ventricles.
To complement cell-based studies, we next examined gene expression in isolated embryonic hearts. Pregnant CD-1 mice were given daily injection of caffeine from E6.5-9.5 or from E6.5-10.5. The doses used were 10, 20, and 60 mg/kg. These doses of caffeine resulted in maternal circulation levels of 37.3 ± 8.3, 41.6 ± 10.7, and 65.3 ± 7.8 μM (n = 3) 2 h after injection, respectively. These levels are equivalent to circulating levels of caffeine following consumption of 2 to 8 cups of coffee in humans (24). These doses and timing of treatment were selected because long-term adverse effects of caffeine on adult heart function have been observed with a single intraperitoneal injection of 20 mg/kg of caffeine at E8.5 (9, 47). In addition, this treatment window (E6.5-10.5) corresponds with a critical period of cardiac development (47).
Pregnant dams were treated from E6.5-9.5 or from E6.5-10.5. Embryonic ventricles were isolated on E10.5. We examined gene expression at either 2 or 24 h after the last caffeine dose. Caffeine treatment during early gestation (E6.5-9.5) or (E6.5-10.5) in CD-1 mice did not alter the number of embryos (13 ± 3 pups/litter, n = 5–7), embryo reabsorption rates (2.4 ± 3.6%, n = 5–7), or gross embryonic morphology.
Similar to in vitro, caffeine treatment from E6.5-10.5 (2 h after last caffeine treatment) significantly increased the expression of structural and hormonal genes (Myh6, Myh7, Tnni3, Anp, and Bnp; Fig. 5A), transcription factors (Gata4, Mef2c, Mef2d, Nfatc1-a, Nfatc1-c; Fig. 5B), and miRNA (miR208a; Fig. 5C). In comparison, caffeine treatment decreased the expression of Myh7b (Fig. 5A), Nfatc1-b (Fig. 5B), miR208b, and miR499 (Fig. 5C), which is differed from in vitro results.
At 24 h after the last caffeine dose, transcriptional changes of some genes persisted. Specifically, the expression of Myh6, Myh7, Tnni3, Bnp (Fig. 5D), and miR208a (Fig. 5F) was consistently upregulated, and the expression of Myh7b (Fig. 5D) and miR499 (Fig. 5F) was downregulated in the embryonic hearts.
Expression of some genes returned to basal level or changed in an opposite direction at 24 h after the last caffeine treatment. The expression of Anp, Gata4, and Mef2c was significantly downregulated in vivo (Fig. 5, D and E). Caffeine treatment from E6.5-9.5 did not change the expression of Mef2d, Nfatc1, or miR208b in vivo on E10.5 (Fig. 5, E and F). These data indicate that some changes in gene expression by caffeine are transient, whereas others persist. A summary of caffeine-induced transcriptional changes in HL-1 cells, primary cells, and embryonic hearts is given in Table 2.
Effects on myosin gene expression by adenosine receptor drugs.
To determine whether specific ARs transduce caffeine action in cardiomyocytes, we next measured myosin gene expression after treating HL-1 cells with an A1AR-specific antagonist DPCPX or an A2aAR-specific antagonist SCH58261 for 48 h. DPCPX significantly increased Myh6 expression but had no effects on the expression of Myh7 or Myh7b (Fig. 6A). In contrast, SCH58261 treatment had no effects on Myh6 expression but significantly increased the expression of Myh7 and Myh7b (Fig. 6B), indicating that caffeine may be affecting cardiac gene expression through both receptors.
Differential gene expression in embryonic ventricles after in utero caffeine treatment.
We next performed transcriptomic mRNA sequencing on the embryonic ventricles to identify other cardiac genes affected by caffeine. Pregnant mice were injected with 20 mg/kg of caffeine daily from E6.5-9.5. At E10.5 (24 h after last caffeine treatment), total RNA was isolated from embryonic ventricles for transcriptomic RNA-seq with Illumina HiSeq2000. The mapping efficiency was about 78.7% (Table 3) and the samples were clustered by treatments (Fig. 7, A and B). Differential expression analysis revealed that 59 genes were significantly upregulated, and 65 genes were significantly downregulated by prenatal caffeine treatment (absolute fold change >1.5; P value with FDR<0.05; Fig. 7A, Table 4, Supplementary Table 1).
Because almost all genes have multiple transcript isoforms (46), we determined changes in expression at the transcript level. We identified 451 upregulated transcripts and 398 downregulated transcripts following caffeine treatment (absolute fold change >1.5; P value with FDR<0.05; Fig. 7B, Table 4, and Supplementary Table 1). Differentially expressed genes and transcripts included cardiac structural genes, transcription factors, DNA methylation genes, and histone modification genes (Tables 5 and 6). Gene expression changes detected by RNA-seq were similar to the results from qPCR (Fig. 8). qPCR results also verified that expressions of DNA methylation enzymes were altered in embryonic ventricles by caffeine treatment, with more significant changes observed in 2 h after last caffeine treatment than those changes in 24 h later (Fig. 9, A and B). Global cardiac DNA methylation was significantly increased by caffeine (Fig. 9C).
In total, 900 unique genes were identified to have altered expression either at the gene or transcription level, and these genes were designated as DE genes (Fig. 7C). Functional enrichment analysis by MetaCore on DE genes revealed that molecular pathways that influence cardiac development (Fig. 10A) and cardiac diseases (Fig. 10B) were significantly enriched. The gene networks of factors promoting cardiogenesis and role of NFAT in cardiac hypertrophy were activated (Fig. 7, D and E). In addition, we identified other cardiovascular-related pathways and genes by functional enrichment analysis with DAVID (Tables 5 and 7).
DEU in embryonic ventricles after in utero caffeine treatment.
To study caffeine effects on alternative splicing, as indicated by different levels of exon usage, further analysis by DEXseq identified 597 DEU exons with absolute fold change greater than 1.5 and FDR adjusted P value <0.05 in embryonic ventricles after caffeine treatment (Supplementary Table 2). Expression was upregulated in 465 exons and downregulated in 132 exons. These exons belong to 590 unique genes. qPCR confirmed DEU in exon 7 of troponin I type 1 (skeletal, slow) (Tnni1) and exon 27 of myosin, light polypeptide 2, regulatory, cardiac, slow (Myl2) (Fig. 11). DAVID analysis identified the DEU genes that are involved in the pathways related to cardiac hypertrophy (Table 8). Representative cardiac genes with DEU are shown in Fig. 12.
Our previous studies demonstrated that disruption of adenosine signaling by caffeine during early embryogenesis resulted in features of cardiac concentric hypertrophy in adulthood (9, 47). We also found altered gene expression and DNA methylation patterns in adult heart ventricular tissue of mice with prenatal caffeine exposure (9). To investigate the mechanisms that could lead to these long-lasting consequences of prenatal caffeine exposure, we assessed the acute effects of caffeine exposure on the transcriptome of adult (HL-1 cells) and embryonic cardiomyocytes. The caffeine doses we used resulted in maternal circulating caffeine levels in mice that are equivalent to circulating levels of caffeine following consumption of 2–8 cups of coffee in humans. The pregnant mice were treated during E6.5-10.5, which is a critical stage for cardiogenesis and genome-wide DNA remethylation in mouse embryos. The equivalent stage in humans is between 3 and 4 wk of gestation, a time before many women know they are pregnant. DNA methylation in this period is vulnerable to environmental and chemical insults, thus changes in DNA methylation at this stage may influence gene expression long term. We now report that caffeine alters the expression and exon usage of important cardiac genes. In addition, we show that caffeine affects the expression of genes involved in epigenetic mechanisms, including DNA methylation, histone modification, and miRNA regulation. These epigenetic effects thus may represent a previously unknown mechanism by which in utero caffeine treatment can have long-lasting effects in adult hearts.
We previously found altered gene expression and abnormal DNA methylation patterns in Mef2c, Myh6, and Myh7 in hearts of adult animals exposed to caffeine in utero (9). Furthermore, in utero caffeine exposure changed the DNA methylation patterns of cardiac genes and miRNAs including Tnni3, Tnnt2, Tnnc1, Anp, Gata4, Mef2d, miR208b, and miR499 (9). We thus examined the expression of these genes as well as other pertinent genes in HL-1 cardiomyocytes treated with caffeine. We found that caffeine significantly increased the expression of many of these genes that are also increased during cardiac hypertrophy (28).
In addition to HL-1 cells, we examined the expression of these target genes in primary cardiomyocytes isolated from mouse embryonic ventricles. We found changes of gene expression in embryonic cardiomyocytes similar to those observed in HL-1 cells. We also observed that primary cardiomyocytes were more sensitive to the effects of caffeine, with alterations seen at lower concentration (100 μM) than in HL-1 cells. In addition, Anp was upregulated and miR208a was downregulated in primary cells, which was opposite of the changes observed in HL-1 cells. Overall, changes in gene expression in HL-1 cells and embryonic cardiomyocytes were similar.
We also observed similar changes in gene expression following caffeine exposure in vivo. We treated pregnant dams with clinically relevant doses of caffeine daily from E6.5 to 10.5, and we examined gene expression changes 2 h after the last caffeine treatment. The transcriptional changes of target genes in vivo were similar to those observed in cultured cardiomyocytes. At 24 h after caffeine treatment in vivo, the expression of transcription factors was different from the aforementioned treatment, with downregulation of Gata4 and Mef2c and no changes in Mef2d and Nfatc1, suggesting recovery or feedback downregulation in response to caffeine exposure. These data indicate that the in vitro models are representive of acute in vivo caffeine treatment.
In all models, caffeine treatment consistently altered the expression of cardiac-specific myosin genes. It has been reported that caffeine increases Myh6 expression in the adult heart (30). Consistent with this observation, we found that expression of Myh6 and Myh7 was increased by caffeine in embryonic hearts. Upregulation of Myh6 and Myh7 has also been observed in adult hearts treated with prenatal caffeine (9), indicating that the changed expression of Myh6 and Myh7 during early development was maintained into adulthood. During development, Myh6 and Myh7 are expressed differently with Myh7 as the predominant fetal isoform and Myh6 as the dominant adult isoform in mice (25, 33). Increased Myh7 expression is a common feature of cardiac hypertrophy (11, 25). Therefore, persistent upregulation of Myh6 and Myh7 may be responsible for the long-term caffeine effects on cardiac morphology and function.
Expression of Myh6 and Myh7 is activated by transcription factors including Gata4, Mef2c, Mef2d, and Nfatc1 (15, 32, 34). In addition, miR208a stimulates Myh7 expression (36). Our results revealed that caffeine increased the expression of Gata4, Mef2c, Mef2d, Nfatc1, and miR208a, leading to downstream upregulation of Myh6 and Myh7 (Fig. 13). These transcription factors also regulate transcription of Tnni3, Anp, and BnP (16), and we observed differential expression of these effector genes following caffeine treatment as well. Increased expression of Tnni3 indicates myocardial injury, and upregulation of Anp and Bnp exerts protective effects by dilating blood vessels during cardiac hypertrophy and heart failure (17). Elevated levels of Anp and Bnp after caffeine exposure may therefore be important to reverse the activation of adrenergic system by caffeine.
To examine the effects of caffeine on the whole transcriptome, we performed RNA-seq analysis on the embryonic ventricular tissues after caffeine treatment. RNA-seq data were consistent with the qPCR. With RNA-seq analyses, we observed that caffeine affected the transcription or splicing of more cardiac genes that are associated with heart disease including cardiac hypertrophy. These data indicate that caffeine impairs heart development by altering the expression of a large battery of genes.
RNA-seq data also indicate that DNA methylation, histone modification, and alternative splicing pathways were compromised by caffeine treatment. In mice, DNA methylation patterns are reestablished from E3.5–10.5 and are sensitive to methylation modifying agents in the intrauterine environment (26). These established patterns of genomic DNA methylation are maintained into adulthood (26).
Previous studies found that caffeine alters gene-specific DNA methylation patterns (9, 37, 51). The present study is the first to show that expression of enzymes responsible for DNA methylation and demethylation were affected by prenatal caffeine treatment. The DNA methylation genes DNA methyltransferase 1 (Dnmt1), 3a (Dnmt3a), and 3b (Dnmt3b) were downregulated. Expression of Tet1, 2, and 3 was downregulated at the gene level, and Tet1–004 was upregulated at the transcript level. The consequence of these changes is global hypermethylation in the caffeine-treated embryonic hearts.
When we compared the DE genes identified in the current study with the differentially methylated (DM) genes identified in the adult hearts exposed to caffeine in utero (treated with 20 mg/kg caffeine at E8.5) (9), we found that transcriptional changes of 146 DE genes in embryonic ventricles inversely correlated with the direction of changes in the DM genes of adult ventricles, whereas transcriptional changes of 140 DE genes positively correlated with the changes in DM genes (Supplementary Table 3). The genes with both DE and DM regions were annotated by DAVID and we found that 17 of them were related to cardiac morphology and function (Table 9). Therefore, DNA methylation may play a role in altering gene expression following caffeine treatment.
Changes in DNA methylation are closely linked to changes in histone modification (10), and we identified expressional changes of histone deacetylases Hdac2, Hdac3, and Hdac10. Some histone changes are inheritable during cell division (31) and, therefore, they may acutely and chronically influence gene expression.
Caffeine also affected factors involved in alternative splicing, which may explain why we observed changes in exon usage in 590 genes. Alternative splicing is a major mechanism to generate proteomic diversity in higher eukaryotes (46). Although the functional significance of alternative splicing is only partially understood, abnormal alternative splicing, also known as DEU, is often observed in many disease statuses including cardiac hypertrophy (46). In the current study, we found that in utero caffeine treatment affects exon usage in many genes. In particular, caffeine affects the splicing of cardiac genes troponin T type 2 (cardiac) (Tnnt2), Tnni1, Tnni3, Myh6, Myh7, and Myh7b, which may contribute to development of cardiac hypertrophy (46).
Previously, we found that embryonic caffeine exposure acts via A1AR to alter cardiac function and DNA methylation in adult offspring (9). In this study, we found that blockades of A1AR and A2AR by selective antagonists caused similar effects as caffeine on myosin gene expression, but in-depth mechanism studies are needed to confirm the causal relationship between AR antagonism and cardiac gene expression. Moreover, caffeine elevates cytosolic calcium level via activation of ryanodine receptor, and it is known that calcium controls numerous cardiomyocyte activities, including gene transcription (21). Therefore, further studies are needed to examine the interaction between caffeine-induced cytosolic calcium levels and cardiac gene expression.
Perspectives and Significance
Overall, this is one of the first studies to comprehensively examine the effects of in utero caffeine treatment on the transcriptome of cardiomyocytes at early developmental stages. We show that caffeine exposure alters gene expression and exon usage in embryonic cardiomyocytes and affects DNA methylation. The biological influence of these changes on acute and long-term adult cardiac function is currently unknown. Although previous publications indicate that in utero caffeine exposure can affect cardiac function in adult mice (9, 47), the long-term effects of caffeine on human cardiac function are unclear. Thus further animal and human studies are needed to evaluate the safety of caffeine exposure during human pregnancy before recommendations can be made.
This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development at the National Institutes of Health (R01 HD058086 to S. A. Rivkees and C. C. Wendler).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: X.F., S.A.R., and C.C.W. conception and design of research; X.F. performed experiments; X.F., W.M., and W.B.B. analyzed data; X.F., S.A.R., and C.C.W. interpreted results of experiments; X.F. prepared figures; X.F. drafted manuscript; X.F., S.A.R., and C.C.W. edited and revised manuscript; X.F., W.M., W.B.B., S.A.R., and C.C.W. approved final version of manuscript.
We thank Dr. William C. Claycomb for providing the HL-1 cardiomyocytes. We thank Ryan Poulsen, Daniel Freeman, Olivia Shi, and Olivia Donnelly for technical assistance.
- Copyright © 2014 the American Physiological Society