All animal procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of the University of California, San Diego with approved protocol # S01049.

Ogt^{flox/flox(f/f)} mouse was purchased from Jackson Lab (Stock No:004860 | OGTF). To generate cardiomyocyte-specific OGT KO mice, floxed mice were crossed with cardiac Troponin T (TnT)-Cre transgenic mice [17] to create Ogt^f/y; TnT-Cre (cKO) mice. All mice were of a mixed 129/SvJ and C57BL/6J background. NKX2-5 knockout mouse was described in previous report [18]. Genotypes of mice were confirmed by polymerase chain reaction (PCR) analysis using embryonic yolk sac or tail extracts and Ogt primers (forward: 5’-CATCTCTCCAGCCCCACAAACTG-3’, reverse: 5’-GACGAAGCAGGAGGGGAGAGCAC-3’), Cre primers (forward: 5’-GTTCGCAAGAACCTGATGGACA-3’; reverse: 5’-CTAGAGCCTGTTTTGCACGTTC-3’) and NKX2-5 primers (forward: 5’- CGGCATAGGACCAGAGTGATA-3’, reverse: 5’- TCCCTGAACATGTCCATCAGGTTC-3’).

Histology was performed as previously described[19]. Briefly, mouse embryos were harvested at different stages and fixed in ice-cold phosphate buffered saline (PBS) containing 4% paraformaldehyde overnight at 4°C. Following fixation, embryos were washed with PBS, dehydrated with a series of 50–100% ethanol, embedded in paraffin, and cut into 8-μm sections using a Microtome. Sections were stained with Hematoxylin and Eosin (H&E), mounted, and imaged using a NanoZoomer 2.0HT Slide Scanning System (Hamamatsu).

Protein isolation and western blot analysis were performed as described in our previous publication [19]. Briefly, embryo hearts were harvested and snap-frozen in liquid nitrogen. Frozen heart tissue was homogenized in 80 μL RIPA buffer (50 mM Tris⋅Cl, pH 7.4, 150 mM sodium chloride, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, 1 mM EDTA) using a handheld pestle (Sigma-Aldrich). Protein concentration was determined using a Micro BCA Protein Assay Kit (Thermo Fisher Scientific) before mixing with 4× lithium dodecyl sulfate (LDS) sample buffer (Life Technologies) and incubating for 10 min at 70°C. Proteins were separated on 4–12% SDS-PAGE gels (Life Technologies) and transferred on to nitrocellulose membranes (Biorad) at 4°C at a constant voltage of 35V in transfer buffer (25 mmol/L Tris, 190 mmol/L glycine, 20% methanol, pH 8.3). After blocking for 1–2 hours in blocking buffer (Tris buffered saline containing 0.1% Tween 20 and 5% non-fat dry milk), membranes were incubated overnight at 4°C with the indicated primary antibody in blocking buffer. Blots were washed and incubated with horseradish peroxidase–conjugated secondary antibody for 1 hour at room temperature. Immunoreactive protein bands were visualized using enhanced chemiluminescence reagent (Thermo), analyzed by densitometry, and expressed as pixel density normalized to GAPDH. Primary antibodies include: OGT (1:1000, 61355, Active Motif), RL2 (1:5000, MA1-072, Thermo Fisher), and GAPDH (1:4000; sc-47724, Santa Cruz). Secondary antibodies: anti-mouse (1:4000; Dako), anti-rabbit (1:4000; Dako), NKX2-5 (1:1000, SC8697, Santa Cruz Biotech).

Mouse embryos or embryonic hearts were harvested at different stages and fixed in ice-cold PBS containing 4% paraformaldehyde overnight at 4°C. Fixed embryos were washed with PBS, then saturated in a 5, 10, 15, and 20% sucrose series in PBS, embedded in OCT Tissue-Tek mixed with 20% sucrose (1:1), and cut into 8-μm sections using a Leica CM 3050S cryostat (Leica Microsystems). Sections were permeabilized with wash buffer (PBS with 0.2% Triton 100), then blocked with blocking buffer (PBS with 1% BSA, 5% donkey serum, 0.2%Triton 100) for 1h at room temperature, followed by overnight incubation at 4°C with primary antibody diluted in blocking buffer. Sections were rinsed in wash buffer and incubated with fluorescently conjugated secondary antibodies and DAPI diluted in blocking buffer at room temperature for 1 hour, before rinsing and mounting in mounting buffer (Dako). Sections were imaged using an Olympus Fluoview FV1000 confocal microscope. Primary antibodies included: OGT (1:200, 61355, Active Motif), RL2 (1:300, MA1-072, Thermo Fisher), CD31 (1:150, 550274, BD Biosciences), NKX2-5 (1:50, SC8697, Santa Cruz Biotech), smooth muscle myosin heavy chain (1:200, BT-562, Alfa Aesar) and alpha actinin (1:400; 650931, Sigma), HIF1α (NB100-479, 1:200), TroponinT (Thermo Scientific, clone 13–11, 1:100).

For EdU incorporation assay, pregnant female mice were given IP injection of EdU(Invitrogen) two hours prior to embryo dissection. After cryo-sectioning of embryo heart, EdU positive cells were detected according to manufacture’s introduction, which was followed by normal primary and secondary antibody incubation. Apoptotic cells in E12.5 heart sections were detected using TUNEL assay kit (Roche Cat. No. 11 684 795 910) according the manufacturer's protocol, witch was followed by immunostaining against alpha actinin to detect cardiomyocytes. Three embryos of each genotype were analyzed and at least five fields were analyzed for each section of the embryonic heart.

Total RNA was isolated from homogenized embryonic hearts using TRIzol (Invitrogen) and reverse transcribed to cDNA using a Super Script III cDNA Synthesis Kit (Invitrogen), according to the manufacturer’s instructions. Complementary DNA amplicons were quantified by incorporation of SYBR Green probe (Biorad) into dsDNA. RT-PCR reactions were performed using Sso-Fast EvaGreen Real Time PCR (Bio-Rad) master mix in 96-well low profile PCR plates in the CFX96 Biorad Thermocycler. Samples were compared using the relative (comparative) Ct method after adjusting for 18s RNA (ΔCt). Primers used in this study are listed in S2 Table.

Embryonic hearts at E11.5 were homogenized in TRIzol (Invitrogen) and total RNA was isolated according to the manufacturer’s instructions. cDNA libraries were prepared using an Illumina TruSeq stranded mRNA kit and subjected to deep sequencing. Reads were quality controlled with FastQC and mapped to the mm10 mouse reference genome with TopHat2 and Bowtie2. Differential expression of novel and reference transcripts was performed using Cuffdiff2. Resulting RNA-seq datasets can be accessed at GEO using accession number GSE146962.

Whole-mount in situ hybridization was performed as previously described [15]. Briefly, embryos were fixed in pre-chilled 4% paraformaldehyde in phosphate-buffered saline at 4°C overnight, and subsequently subjected to RNA in situ analysis. The Angiopoietin-1 in situ probe was generated from embryonic heart cDNA using the following primers: forward: TAGAACTAGTGGATCCCTGCCATTCTGACTCACAT, reverse: CGGTATCGATAAGCTTGTCTGTTGGAGAAGTTGCT.

Whole mouse embryos or embryonic hearts were fixed in pre-chilled 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C overnight. After washing with PBS, samples were dehydrated through serial methanol (form 25% to 100%), and then incubated in Dent’s bleach (methanol: dimethyl sulfoxide: 30%H2O2, 4:1:1) for 3 hours, followed by rehydration in serial methanol. After blocked in 3% skimmed milk, samples were then incubated with CD31 antibody (550274, BD Biosciences, dilution from 1:500 to 1:2000 depending on sample), and then followed by secondary antibody incubation. Then secondary antibody was detected by ABC kit (Vector Lab) and visualized using DAB substrates (Vector Lab) according to manufacture’s introduction.

As described previously [20], Hearts were collected from mouse embryos at appropriate stages and dipped into heparinized PBS. Then, ink (Kiwa-Guro, Sailor, Japan) was gently injected from the ascending aorta using a glass micropipette. Samples were fixed in 4% paraformaldehyde solution and dehydrated in sequential ethanol solution. To clear samples, dehydrated samples were immersed in BABB (1:2 benzyl alcohol to benzyl benzoate).

Data are presented as mean ± standard error of the mean (SEM). Statistical analysis was performed using Prism 6.0 software (GraphPad Software, La Jolla, CA), with two-tailed Student’s t test used for comparisons among groups as indicated. P values less than 0.05 were considered significant.

To investigate the expression pattern of OGT in heart at different stages, we performed immunostaining for transverse heart sections at different stages. We used an antibody against NKX2-5, a cardiomyocyte specific transcription factor, to distinguish cardiomyocytes and non-cardiomyocytes. As shown in Fig 1, images were taken from left ventricle compact zone for E12.5 and E16.5 hearts, and from left ventricle free wall myocardium for adult heart (Fig 1). At E12.5, OGT was mostly restricted to the nucleus in cardiomyocytes, while in non-cardiomyocytes OGT had a relatively higher expression in cytosol (Fig 1). At E16.5, OGT was still mostly restricted to the nucleus in cardimyocytes. However, non-cardiomyocytes started to express less OGT compared to cardiomyocytes (Fig 1). At 2 months of age (adult), OGT was almost exclusively localized to the nucleus of cardiomyocytes. In contrast, OGT appeared to be greatly downregulated in non-cardiomyocytes (Fig 1). Our data suggested that OGT is expressed in nucleus of cardiomyocytes through heart development, indicating its function in heart development.

A previous report showed that deletion of OGT using an alpha myosin heavy chain (αMhc) driven Cre affected postnatal viability [10], but there was variability in phenotypic severity. The variability could have arisen from the relatively late expression and low efficiency of αMhc-Cre as described in a recent review paper [21, 22]. To better explore the role of OGT during heart development, we crossed a Troponin-T Cre [17] (TnT-Cre) mouse line, which expresses Cre activity as early as E7.5 and gives high efficiency [17, 19, 22], with an Ogt floxed line to establish an early cardiomyocyte-specific OGT knockout mouse model (hereafter cKO). To confirm the efficiency of Cre-mediated ablation of Ogt, we performed both immunostaining and western blot analyses using a polyclonal antibody against OGT, and a monoclonal antibody against global O-GlcNAc modification (RL2). From immunostaining results, almost all NKX2-5 postive cardiomyocytes showed no signal for OGT in cKO hearts, while cardiomyocytes in control hearts showed strong nuclear staining for OGT. In contrast, NKX2-5 negative non-cardiomyocytes in both control and knockout hearts had similar levels of OGT signal. Similarly, RL2 staining was diminished in knockout myocardium compared to that in control myocardium (Fig 2A,S1 Fig). Western blot analyses using whole heart samples showed that levels of OGT protein and global O-GlcNAc were reduced by approximately 80% in cKO hearts when compared to control hearts (Fig 2B).

Similar to, but more severe than previously reported [10], we did not recover any knockout mouse at weaning stages (Fig 2C and S1 Table). Genotypic analyses revealed that at E18.5 knockout embryos were still viable and recovered at expected Mendelian ratios. At postnatal day 1, although knockout mice were still observed at Mendelian ratios, nearly half of them (6 of 14) had died (Fig 2C and S1 Table). Further survival analysis showed that a majority of cKO mice (11 out 13) died two days after birth, and no knockout mice survived beyond 7 days after birth. These data suggested that ablation of OGT in cardiomyocytes led to postnatal lethality shortly after birth.

When compared to control hearts at E16.5, cKO hearts showed dramatic morphological differences (Fig 3A). Both left and right atria of cKO hearts were smaller than those of control hearts, and showed an uneven surface. The height of knockout heart ventricles was smaller than that of control heart ventricles, whereas the width of knockout heart ventricles was larger than that of control heart ventricles. Histological analysis revealed that loss of OGT in cardiomyocytes caused multiple cardiac developmental defects, including dilated ventricles, atrial septal defects, membranous and muscular ventricular septal defects, reduced ventricular wall thickness, and increased trabecular area and thickness (Fig 3B). Careful quantification analysis confirmed reduced wall thickness, and increased trabecular area (Fig 3B). At the whole mount level, overall morphological differences were apparent by E12.5 (S2 Fig), with histological analyses revealing differences in wall thickness and trabecular thickness as early as E11.5 (S3 Fig). Collectively, those data suggested that OGT was essential for heart development, with loss of OGT in early cardiomyocytes leading to dilated cardiomyopathy with multiple cardiac defects.

From morphological and histological analyses, coronary arteries were barely evident in cKO hearts at E16.5 (Fig 3). To confirm this, angiography based on ink injections from the aorta were performed. This analysis further demonstrated that coronary artery formation was severely affected in cKO hearts (Fig 4A). In control hearts, coronary arteries were well formed and extended from the aorta to the apex, while in cKO hearts, the expansion of the coronary arteries was largely blocked. Larger vessels, which were clearly evident in control hearts, were largely absent in cKO hearts. Instead, there were many more small vessel branches formed in cKO hearts. Furthermore, immunostaining of serial heart sections using antibody against smooth muscle myosin heavy chain (smMHC), a marker of smooth muscle cells in large arterial vessels, confirmed that coronary artery formation was dramatically affected in cKO hearts. In control hearts, smMHC positive cells were detected throughout the ventricular walls. In contrast, in cKO hearts, smMHC positive cells were only detected at the area close to the heart base (Fig 4B). Although whole mount PECAM staining at E16.5 did not show differences in coronary vein development between control and cKO hearts (Fig 3C), whole mount PECAM staining of E12.5 and E13.5 hearts demonstrated that early coronary plexus formation was delayed by loss of OGT (Fig 4D). Our data suggested that loss of OGT in cardiomyocytes alters the development of coronary vessel, specially the development of coronary artery.

As ventricular wall thickness and trabecular thickness were affected when OGT was deleted from early cardiomyocytes, we performed an EdU incorporation assay at E12.5 to investigate whether cardiomyocyte proliferation was affected or not. Consistent with observed phenotypes, proliferation assays showed that cardiomyoctye proliferation in both compact zone and ventricular septum was significantly reduced. In compact zone and ventricular septum of control hearts, the percentage of cardiomyocytes with EdU incorporation following two hours of labeling were 26.98 ± 2.118 and 31.69 ± 1.964, which was reduced to 15.25 ± 0.65 and 14.20 ± 0.511 respectively in cKO hearts (Fig 5A and 5B). However, the EdU incorporation rate was increased in trabeculae of cKO hearts (21.58±0.836%) compared to trabeculae of control hearts (13.21 ± 0.2978%) (Fig 5A and 5B). Furthermore, TUNEL staining did not show significant difference between control and cKO hearts (Fig 5C,S4 Fig). Taken together, those data indicated the decreased cardiomyocyte proliferation rather than abnormal apoptosis induced by loss of OGT contributed to those phenotype observed above.

Since O-GlcNAc has been linked to transcriptional regulation, we did whole transcriptome analysis using RNA sequencing. We chose E11.5 cKO and littermate control hearts for RNA-seq, as E11.5 was the time point when phenotypes were first evident in knockout hearts. RNA-seq revealed that there were a total of 753 dysregulated genes (fold-change greater than 1.5). Among those genes, 413 genes were upregulated, and 340 genes were downregulated (S5A Fig). We also used the bioinformatics database DAVID to do gene ontology analysis. Biological process analysis showed that most genes upregulated consequent to loss of OGT were related to morphogenesis, homeostasis, and muscle contraction (S5B Fig). These results were consistent with cardiac morphological phenotypes observed. Unexpectedly, there were also some categories related to vesicle membrane that were also upregulated. Similar analysis was also performed for downregulated genes. Among downregulated biological process categories, the most highly significant was metabolic processes (S5C Fig). Using quantitative real-time PCR (qRT-PCR), we validated a subset of genes, which were shown to be dysregulated by RNA-seq (S5D Fig). The qRT-PCR results were consistent with RNA-seq analysis. Our RNA-seq analysis suggested that loss of OGT resulted in global alterations in gene transcription.

Examination of the top downregulated genes with significant levels of expression (mean RPKM in control hearts greater than 2) revealed that angiopoietin-1 (Angpt1), an essential gene for heart development [13, 15, 16], was dramatically decreased (Fig 6A). Using qRT-PCR we confirmed that Angpt1 was downregulated approximately 4-fold (Fig 6B). To further confirm downregulation of Angpt1, we performed RNA in situ hybridization at E10.5. Angpt1 was highly expressed in hearts of control embryo. In contrast, the signal for Angpt1 was specifically lost in hearts of cKOs (Fig 6C). Immunostaining of transverse sections with antibody to alpha actinin demonstrated that expression of Angpt1 was mainly restricted to cardiomyocytes, indicating that OGT regulated expression of Angpt1 in a cell-autonomous manner.

Since the oxygen-sensitive transcriptional activator hypoxia-inducible factor-1 (HIF-1) is a key transcriptional mediator of angiogenesis related genes including Angpt1 [23], and cardiomyocyte specific knockout of HIF-1α in mouse leads to embryonic lethality [24], we performed immunostaining to check the exrepssion level and localization of HIF-1α. Our data showed that loss of OGT in cardiomyocytes affected neither the expression nor localization of HIF-1α in cardiomyocytes (S6 Fig). NKX2-5, an essential transcription factor in cardiomyocytes, has been reported to regulate Angpt1 in fetal cardiomyocytes in a NKX2-5 hypomorphic mouse model [25]. And our RNA in situ hybridization also confirmed that Angpt1 was dramatically downregulated in NKX2-5 knockout embryos at E9.0 (S7A Fig). However, we were unable to detect the change of expression level of NKX2-5 in cKO cardiomyocytes (S7B Fig). Also, we could not detect any O-GlcNAcylation of NKX2-5 using immunoprecipitation method (S7C Fig). So our data indicates that OGT is not likely to regulate Angpt1 through HIF-1α or NKX2-5.