Glucose inhibits cardiac muscle maturation through nucleotide biosynthesis

The heart switches its energy substrate from glucose to fatty acids at birth, and maternal hyperglycemia is associated with congenital heart disease. However, little is known about how blood glucose impacts heart formation. Using a chemically defined human pluripotent stem-cell-derived cardiomyocyte differentiation system, we found that high glucose inhibits the maturation of cardiomyocytes at genetic, structural, metabolic, electrophysiological, and biomechanical levels by promoting nucleotide biosynthesis through the pentose phosphate pathway. Blood glucose level in embryos is stable in utero during normal pregnancy, but glucose uptake by fetal cardiac tissue is drastically reduced in late gestational stages. In a murine model of diabetic pregnancy, fetal hearts showed cardiomyopathy with increased mitotic activity and decreased maturity. These data suggest that high glucose suppresses cardiac maturation, providing a possible mechanistic basis for congenital heart disease in diabetic pregnancy.


Introduction
Congenital heart disease (CHD) is the most common type of birth defect affecting 0.8% of human live births (Fahed et al., 2013). Although genetic factors play a significant role in the development of CHD, current genomic technologies, including exome sequencing and SNP arrays, have provided a genetic diagnosis for only 11% of the probands , highlighting the crucial role of non-genetic contributors.
Among the non-genetic factors that influence the fetal heart, maternal hyperglycemia is the most common medical condition, associated with a 2-5-fold increase in CHD independent of genetic contributors (Centers for Disease Control, 1990;Simeone et al., 2015;Yogev and Visser, 2009). Diabetic pregnancy is often accompanied by maternal complications including vasculopathy, neuropathy, and insulin resistance, which potentially affect fetal cardiac formation indirectly. These systemic complications are often subclinical, hindering the dissection of the pathomechanism of CHD in diabetic pregnancy. Thus, despite the established association between maternal hyperglycemia and malformation of the fetal heart, little is known about how glucose levels impact cardiomyocyte development and how hyperglycemia affects heart formation in diabetic pregnancy (Gaspar et al., 2014).
The metabolic environment is one potential non-genetic determinant of cell proliferation and differentiation. Cells display distinct metabolic characteristics depending on their differentiation stage (Carey et al., 2015;Tohyama et al., 2016;Wang et al., 2009), and the fuel type used by the cells serves not merely as a source of energy but also as a critical regulator of self-renewal and differentiation of stem or progenitor cells (Harris et al., 2013;Oburoglu et al., 2014;Shiraki et al., 2014;Shyh-Chang et al., 2013). However, little is known about the mechanism. Cardiomyocytes shift their energy substrate during late embryonic and neonatal stages (Makinde et al., 1998). Glucose is the major energy source during the early developmental stages. Oxidative phosphorylation is low until E10.5 of developing rodent hearts and rapidly increases between E10.5 and E14.5 (Cox and Gunberg, 1972). This coincides with the rapid maturation of the mitochondrial structure in embryonic cardiomyocytes (Mackler et al., 1971). Shortly after birth, fatty acid oxidation becomes the predominant source of ATP production to meet the high energy demand of the maturing heart (Warshaw and Terry, 1970). These metabolic changes occur as a consequence of changes in the expression of metabolic enzymes and transporters. However, it remains unclear whether and how these metabolic changes, in turn, regulate the cardiac differentiation program.
Here, we describe the use of in vitro human embryonic stem cell-derived cardiomyocytes (hESC-CMs) and an in vivo murine diabetic model to show that glucose not only induces cardiomyocyte proliferation but also inhibits cardiomyocyte maturation. Glucose can be metabolized in multiple catabolic and anabolic pathways, including glycolysis, oxidative phosphorylation, the pentose phosphate pathway (PPP) and the hexosamine biosynthesis pathway. Our chemical screening revealed that the pro-mitotic and anti-maturation effect of high glucose is regulated by glucose-derived deoxynucleotide biosynthesis through PPP. In vivo measurement of 18 F-FDG accumulation revealed that glucose uptake is drastically suppressed during the late gestational and early postnatal stages. Exposure to high blood glucose in a murine model of diabetic pregnancy resulted in higher mitosis and delayed maturation of fetal cardiomyocytes in vivo. Together, our data uncover how the dynamics of glucose metabolism impact late embryonic cardiogenesis.

Results
Glucose reduction promotes hESC-CM differentiation hESC-CMs were differentiated in monolayer in a chemically defined condition, reproducibly yielding~90% of MF20 + cardiomyocytes at day 14 with multiple cell lines including WA09 (H9) and UCLA4 hESCs ( Figure 1A-C) (Arshi et al., 2013;Minami et al., 2012). hESC-CMs start to beat synchronously at around day 6-7 in our system. To characterize the differentiation stages, mRNA expression profiles from H9 hESC-CMs were serially examined by RNA-seq at five distinct stages (GSE84815): undifferentiated hESC (day 0); mesodermal precursor stage (hMP, day 2); cardiac progenitor stage (hCP, day 5); immature cardiomyocyte (hCM14); and hESC-CMS differentiated for 14 additional days (hCM28). The expression data were analyzed using signatures collected from MSigDB, a body atlas and primary cell atlas (Mabbott et al., 2013;Su et al., 2004;Subramanian et al., 2005). As expected, the stem cell signature decreases during these five stages, while signatures associated with heart and smooth muscle increase, further suggesting that our protocol leads to a high degree of enrichment for cardiomyocytes ( Figure 1D and E). This differentiation course is comparable to that reported in previous publications (Paige et al., 2012;Wamstad et al., 2012).
To examine the impact of glucose levels on cardiac differentiation, hESC-CMs were cultured in media containing various concentration of glucose, starting at the hCM14 stage when cells are already differentiated to immature cardiomyocytes ( Figure 2A, Video 1 and 2). The basal differentiation medium contains 25 mM glucose, 0.9 mM pyruvate, essential and nonessential amino acids, and human albumin (G25 medium). Interestingly, glucose dose-dependently suppressed the expression of TNNT2 (a cardiac marker), NKX2-5 (a cardiac marker), and PPARGC1A (a mitochondrial marker) ( Figure 2B). Gene expression profiling by RNA-seq revealed that genes that are related to cardiac muscle and function are enriched in hESC-CMs in low glucose medium, and that genes that are associated with mitosis and cell cycle are enriched in the high-glucose group genome-wide ( Figure 2C, Figure 2-figure supplement 1B; GSE84814). These data suggest that low glucose after day 14 induces the differentiation and suppresses the cell cycle of hESC-CMs.
To validate these results, hESC-CM proliferation was analyzed by pH3 staining and EdU flow cytometry analysis. Low glucose decreased mitotic activity at day 28 without affecting the viability of hESC-CMs ( Figure 2D,E). In addition, hESC-CMs in low glucose medium showed more robust staining of a-actinin, although the sarcomere length did not significantly change ( Figure 2F, Figure 2figure supplement 1B). MitoTracker staining and flow cytometry analyses revealed that hESC-CMs cultured in low glucose media have increased mitochondrial contents and inter-myofibrillar distribution of mitochondria, characteristic of differentiated cardiomyocytes ( Figure 2F and G). Addition of 2-DG (2-deoxy-D-glucose), a competitive inhibitor of glucose phosphorylation, induced higher levels eLife digest Congenital heart disease is the most common type of birth defect, affecting nearly 1 in 100 children born. It can involve a weak heart, narrowed arteries, narrowed heart valves, or the main arteries of the heart switching places. These conditions can be fatal if untreated and often need surgery to correct.
The mother's blood sugar levels during pregnancy can have a large effect on how likely the baby is to have congenital heart disease. If a pregnant woman has poorly controlled diabetes with rapidly fluctuating sugar levels, she may be at a higher risk of having a child with the condition. High sugar levels in the mother's blood make the baby up to five times more likely to have congenital heart disease. It has been difficult to find out exactly how sugar levels interfere with heart development because diabetes can affect the fetus in many ways.
Nakano et al. used stem cells and experiments in pregnant mice with diabetes to hone in on how high sugar levels affect the fetus's heart development. First, heart cells were grown from human stem cells, and exposed to high levels of glucose in a dish. This revealed a new mechanism for how high sugar levels affect heart formation: the cells created too many nucleotides, the building blocks of molecules such as DNA. It turns out that high glucose levels boosted a chemical process in the cell known as the pentose phosphate pathway. Some of the products of this pathway are nucleotides. This made the cells divide rapidly, but did not allow them to mature well compared with cells exposed to normal levels of sugar. In another experiment, Nakano et al. found similar results in pregnant diabetic mice. The heart cells in mouse fetuses also divided quickly but matured slowly when exposed to high sugar levels.
An estimated 60 million women at an age to have children have diabetes. These new findings help us to understand why and how these women are more likely to have children with congenital heart disease, and further study will hopefully lead to a better way to prevent this condition.
of MitoTracker and MF20 expression even in the presence of 5 mM or 25 mM glucose (Figure 2figure supplement 1C), suggesting that the effect is specific to glucose and not to changes in osmotic pressure. Consistently, flow cytometry showed a significant increase in cell size under glucose-restricted conditions ( Figure 2I). Together, these results demonstrate that glucose dosedependently suppresses the maturation of cardiomyocyte cellular architecture and the upregulation of cardiac genes in hESC-CMs.

Glucose reduction promotes functional maturation of hESC-CMs
We next compared the metabolic and functional maturity of hESC-CMs cultured in the presence and absence of glucose by six methods. First, hESC-CMs were stained with JC-1, a green fluorescent dye that generates red fluorescence upon formation of aggregates in active mitochondria. The level of red fluorescence is often used as an indicator of mitochondrial membrane potential and, therefore, mitochondrial activity. Immunofluorescent staining revealed that mitochondria in glucosereduced hESC-CMs are more elongated ( Figure 3A), and flow cytometry analysis revealed that the level of JC-1 aggregation is significantly higher in glucose-reduced hESC-CMs cultured in both    Figure 3B). These data suggest that high glucose inhibits the functional maturation of mitochondria.
Second, intracellular lactate levels of hESC-CMs were measured using Laconic, a fluorescence resonance energy transfer (FRET)-based lactate nanosensor (San Martín et al., 2013). After glucose deprivation, the Laconic construct was introduced into hESC-CMs via adenovirus as described previously (John et al., 2008). In hESC-CMs differentiated in standard 25 mM glucose, bath-applied lactate (4 mM) caused a rapid increase in intracellular lactate (measured as a decrease FRET ratio), demonstrating the efficacy of the probe. Subsequent addition of pyruvate evoked a similar but smaller elevation of intracellular lactate. Under these conditions, inhibition of mitochondrial respiration with sodium cyanide (NaCN) had only a minor effect on the intracellular lactate level. This result suggests that hESC-CMs that are differentiated in standard 25 mM glucose do not actively metabolize pyruvate ( Figure 3C, upper panel). By contrast, addition of 4 mM pyruvate to glucose-reduced hESC-CMs did not cause intracellular lactate accumulation, and addition of NaCN in the presence of pyruvate resulted in a substantial increase in lactate level. This result is consistent with pyruvate utilization by mitochondria ( Figure 3C, lower panel). Together, these data suggest that mitochondria metabolize pyruvate in hESC-CMs that are cultured in glucose-reduced conditions, but not in the presence of glucose.
Third, we assessed cellular respiration of hESC-CM using the XF24 Extracellular Flux Analyzer (Seahorse Bioscience), in which oxygen consumption rate (OCR) was measured in real time in a basal state and in response to oligomycin (an ATP synthase inhibitor), FCCP (carbonilcyanide ptriflouromethoxyphenylhydrazone, a mitochondrial uncoupler), and rotenone or myxothiazol (complex I or III inhibitors, respectively) ( Figure 3D). Although base-line mitochondrial respiration was not changed ( Figure 3E), ATP-linked respiration was elevated in the no glucose condition ( Figure 3F and G). Glucose-reduced hESC-CMs also demonstrated substantially larger maximum respiration capacity, as indicated by the response to FCCP ( Figure 3H). These results corroborate the increased capacity of cellular respiration in hESC-CMs cultured under low glucose conditions. Fourth, the Ca 2+ kinetics of hESC-CMs were assessed using a Ca 2+ transient assay (Shimizu et al., 2015). Although the peak amplitude of the transient (DF/F0) did not show a significant difference, the maximum upstroke (V max ) was significantly faster and the time to 50% decay was significantly shorter in the glucose-reduced group ( Figure 3I). This pattern is consistent with the previous report demonstrating the role of thyroid hormone on hiPSC-CM maturation (Yang et al., 2014), and suggestive of an inhibitory role of high glucose on hESC-CM maturation.
Fifth, taking advantage of our monolayer culture system, we examined electrophysiological properties, using a multi-electrode array (MEA) culture plate as reported previously (Zhu et al., 2017). Maximum upstroke velocity (dV/dt max ) of field potential is a reliable parameter for approximating the electrophysiological maturity of cardiomyocytes derived from pluripotent stem cells (Haase et al., 2009;Ma et al., 2011;Zhang et al., 2009). Compared with hESC-CMs cultured in 25 mM glucose, those hESC-CMs cultured in the absence of glucose displayed a significant increase in dV/dt max ( Finally, the monolayer culture method allowed us to measure cell contractility by digital image correlation using the MotionGUI program  (Huebsch et al., 2015). Both average maximum contraction speed and average maximum relaxation speed were higher in those hESC-CMs cultured in 0 mM glucose ( Figure 3J). Together, these data suggest that glucose reduction promotes the functional maturation of hESC-CMs at the metabolic, electrophysiological, and biomechanical levels.
Glucose blocks cardiac maturation through the pentose phosphate pathway Having determined that glucose reduction induces cardiac maturation at morphological, genetic metabolic, and functional levels, we next sought to analyze the mechanism by which glucose blocks cardiac maturation. Glucose is metabolized by multiple pathways involving both catabolic reactions (anaerobic glycolysis and the aerobic TCA cycle) and anabolic reactions (PPP, the hexosamine pathway, etc.). We first examined the impact of glucose reduction on the global metabolomics signature. Mass spectrometry revealed that glucose deprivation resulted in a significant decrease in the levels of the metabolites in purine metabolism, pyrimidine metabolism, the PPP, the hexosamine pathway, and glycolysis, whereas lipid precursors, amino acids, glutamine, and glutamate, as well as urea cycle metabolites, were not significantly affected ( Figure 4A and Figure 4-figure supplement 1). ATP levels were not significantly different under glucose-restricted conditions ( Figure 4B), neither were any specific stress pathways significantly increased when examined by RNA-seq, suggesting that the cells were not energy-starved in the absence of glucose in the cell culture media.
To identify the metabolic pathway responsible for the improved cardiac maturation by glucose reduction, we conducted a systematic screening using chemical inhibitors for the various glucose metabolic pathways in the monolayer 384-well format ( Figure Figure 4-figure supplement 3, our chemical inhibitor screening suggests that the PPP plays a critical role in the inhibition of cardiac maturation and that blocking this pathway by either use of chemical inhibitors or glucose deprivation induces cardiac maturation. Although mitochondria are a major source of reactive oxygen species (ROS) and physiological levels of ROS promote cellular differentiation (Crespo et al., 2010), the level of ROS measured by DCFDA (dichlorodihydrofluorescein diacetate) did not increase in the absence of glucose and neither did ROS inhibition had a significant impact on TNNT2 expression level (Figure 4-figure supplement 3F), suggesting that the increase in ROS is not responsible for the induction of cardiac maturation.

Nucleotide metabolism regulates cardiomyocyte maturation
The oxidative arm of the PPP generates two major products: reducing power in the form of NADPH and 5-carbon sugars that supply the backbone for nucleotide biosynthesis. To test whether glucose level impacts cardiac maturation by affecting nucleotide biosynthesis, we rescued nucleotide synthesis by adding uridine to hESC-CMs cultured in low glucose media. Under glucose starvation, supplementation of uridine is known to rescue the growth of bacteria, yeast and malignant cells (Linker et al., 1985). In our hESC-CM culture system, uridine restored cell proliferation even in lowglucose conditions ( Figure 5A and B). Interestingly, uridine dose-dependently reduced the level of TNNT2 even in glucose-deprived conditions ( Figure 5C and Figure 5-figure supplement 1A), suggesting that glucose-mediated inhibition of cardiac maturation is dependent on the supply of nucleotides and not of NADPH.
In order to test whether nucleotides are necessary for the glucose-dependent inhibition of cardiac maturation, nucleotide biosynthesis was blocked by multiple methods. Addition of an excess amount of thymidine (unlike uridine) blocks the synthesis of DNA by inhibiting the formation of deoxycytidine (i.e., the thymidine block method), which is commonly used to synchronize the cell cycle (Reichard et al., 1960;Xeros, 1962). When excess thymidine was added to hESC-CMs, the expression of TNNT2 and NKX2-5 were increased ( Figure 5D and . Together, these gain-and loss-offunction data suggest that nucleotide biosynthesis is a key regulatory pathway of the pro-mitoticanti-maturation effect of glucose.

Nucleotide deprivation, not cell cycle block, induces cardiomyocyte maturation
Nucleotide synthesis is a key step in DNA replication and thus cell cycling. Therefore, it is not clear whether the maturation of hESC-CMs that results from deprivation of glucose is due to the cell-cycle block in general or to the effects of nucleotides themselves. To examine whether cell-cycle arrest in general is an essential trigger for cardiac maturation, we blocked the mitotic activity of hESC-CMs by a CDK4/6 inhibitor and paclitaxel (Taxol; an inhibitor of microtubule breakdown), both of which block the cell cycle without directly inhibiting the nucleotide kinetics. Interestingly, neither CDK4/6 inhibitor nor paclitaxel induced cardiac maturation ( Figure 5-figure supplement 2A-B). Together, these data suggest that cell-cycle arrest by itself is not crucial for the promotion of cardiac maturation. Rather, nucleotide deprivation is a key mechanism for cardiac maturation.  Glucose uptake is progressively suppressed during physiological cardiogenesis in utero These in vitro data suggest that glucose reduction promotes cardiac maturation while inhibiting cardiomyocyte proliferation. An intriguing possibility is that the same mechanism underlies cardiac maturation in the in vivo natural counterpart. During normal embryogenesis, however, the blood glucose level is primarily regulated by maternal metabolism and stays relatively stable in utero, leading us to hypothesize that cellular glucose uptake becomes restricted during late fetal stages. To test this possibility, we measured the glucose uptake in the fetal hearts using 18 F-labeled 2-deoxy-2-fluoroglucose (FDG). 18 F-labeled FDG was injected intravenously via the maternal tail vein at E10.5, E12.5, and E15.5 or intraperitoneally into P1 and P7 pups. After 2 hr, the mice were imaged by PET/CT, the hearts were dissected, and cardiac accumulation was measured quantitatively. Interestingly, the normalized cardiac accumulation progressively and rapidly decreased from E10.5 to P7, with 0.11% and 0.05% 18 F-FDG uptake in P1 and P7 hearts, respectively, compared to E10.5 hearts ( Figure 6). These data suggest that cardiac glucose uptake becomes significantly restricted during late gestational and early postnatal stages, creating an intracellular glucose deprivation condition during natural in vivo development.

Hyperglycemia promotes the proliferation and inhibits the maturation of cardiomyocytes in utero
We next tested whether hyperglycemia promotes the proliferation and inhibits the maturation of cardiomyocytes in vivo using fetuses and neonates from diabetic pregnancy. Akita heterozygous mice carry a single amino-acid substitution in the Ins2 gene and exhibit multiple disorders associated with maturity-onset diabetes of the young (MODY) (Barber et al., 2005;Fujita et al., 2001;Wang et al., 1999;Yaguchi et al., 2003;Yoshioka et al., 1997). By crossing an Akita female with a wild-type male, we created a diabetic pregnancy condition in which wild-type fetuses (half of the pups in the litters) and their littermates are exposed to hyperglycemia ( Figure 7A). In the C57BL/6 background, the average blood-glucose levels of the Akita mothers that we used were significantly higher than those of sex-matched control littermates (215 ± 84 vs 71 ± 12 mg/dl, respectively; p<0.0005). Fetal and neonatal Tnnt2 + cardiomyocytes from wild-type hearts from wild-type mothers and wild-type hearts from Akita mothers were examined for mitotic activity in an in vivo EdU incorporation assay at E16.5 and P0 stages, when cardiomyocytes are not yet multinucleated or multiploidic. As shown in Figure 7B and C, the number of cardiomyocytes in S phase was significantly higher at both E16.5 and P0 in the diabetes group. Histological analyses showed that the number of phosphorylated histone H3-positive cardiomyocytes (pH3 + /Tnnt2 + ) is higher at E16.5 in the embryos from diabetic pregnancies ( Figure 7D and E). These data suggest that fetal cardiomyocytes are more mitotic when exposed to maternal hyperglycemia.
To examine whether hyperglycemia inhibits the maturation of fetal cardiomyocytes in vivo, we analyzed the fetal and neonatal hearts from diabetic pregnancies. The level of Tnnt2 expression was significantly lower in the hearts from the diabetic pregnancies ( Figure 7F). A hallmark of the congenital heart disease associated with diabetic pregnancy is asymmetric cardiac hypertrophy. Although heart weight/body weight ratio did not show a difference in our mouse model, the thickness of left and right ventricular free walls was significantly increased in the hearts from diabetic pregnancies at P1 ( Figure 7G). Consistently, the cardiomyocyte size measured by flow cytometry was significantly smaller in the diabetic pregnancy group ( Figure 7H). These data suggest that overproliferation and/ or delayed maturation underlie the pathological mechanism of cardiomyopathy associated with diabetic pregnancy.

Discussion
In summary, we investigated the role of glucose in cardiac formation, and discovered (1) that glucose dose-dependently inhibits cardiac maturation in hESC-CMs, (2) that the maturationinhibitory effect is dependent on nucleotide biosynthesis via the PPP, (3) that the developing heart accomplishes intracellular glucose starvation by limiting glucose uptake in late gestational stages during normal embryogenesis, and (4) that perturbation of the environmental glucose level in diabetic pregnancy affects natural cardiomyocyte maturation in vivo.
Cardiomyocytes switch their main energy substrate from glucose (or other carbohydrates) to fatty acids shortly after birth. This metabolic switch has long been believed to be an adaptation of cardiomyocytes to facilitate more efficient production of ATP. However, our study has revealed that a drastic suppression of glucose uptake occurs during gestational stages, long before the metabolic switch after birth ( Figure 5). Our in vitro hESC-CM glucose-deprivation experiments mimic this in vivo glucose starvation. The results suggest that glucose deprivation induces cardiac maturation at genetic, morphological, metabolic, electrophysiological and biomechanical levels (Figures 2 and  3), suggesting that glucose is a negative regulator of the maturation of fetal cardiomyocytes in vitro and in vivo, as well as a positive regulator of the mitotic activity of these cells. Perturbation of the natural glucose starvation results in higher mitotic activity and lower maturity of cardiomyocytes in vivo (Figure 7). Together, our results suggest that the metabolic switch during perinatal stages is necessary not only to meet the energy demand but also to induce the genetic program that facilitates the maturation of the cardiomyocytes in vivo. An important question that is yet to be answered is how the drastic suppression of intracellular glucose uptake is achieved in the fetal heart. As the fetal glucose environment is primarily determined by maternal metabolism and kept relatively constant in utero, one possibility is that the glucose uptake is limited at the glucose transporter level in fetal cardiomyocytes. In fact, fetal heart switches its glucose transporter isoform at around this stage. Understanding how glucose regulates the genetic program and how the glucose uptake is regulated at the genetic level will be key to further dissecting the cross-talk between the genetic and non-genetic factors governing heart formation.

Nucleotide biosynthesis via the PPP as a key mechanism in balancing proliferation and maturation
Glucose is the most fundamental and commonly available nutrient for the cells. Hence, the activity of the glucose metabolic pathways is tightly regulated in cells. Glucose is broken down to extract energy through the glycolysis pathway and also shunts to supply 5-carbon sugars and NADPH through the PPP. In our study, chemical inhibition of glucose metabolic pathways in hESC-CMs revealed that it is not the catabolic breakdown of glucose to extract energy but rather the anabolic use of glucose to build nucleotides that is responsible for the glucose-dependent inhibition of cardiac maturation. Most of the proliferating cells synthesize nucleotides de novo from glucose, glutamine, and CO 2 . In our hESC-CM experiments, blocking the PPP and nucleotide biosynthesis inhibited the glucose-mediated induction of mitosis and suppression of maturation, and  (n = 7, each; data shown are mean ± SD, p-value by t-test). (D,E) Double immunostaining for phospho-histone H3 (pH3, green) and a-actinin (red) of the heart from normal and diabetic pregnancies at E16.5, P1 and P7 (D). %pH3 + cells within a-actinin + cardiomyocytes (CM) from normal (WT) and diabetic (Akita) pregnancies at E16.5 and P1 (E). At least 10,000 cardiomyocytes were counted for each of five hearts. (n = 7, each; mean ± SD, p-value by t-test.) (F) qPCR analysis for Tnnt2 expression in hearts from normal and diabetic pregnancies. The expression level is normalized to control at each stage (n = 3, each; p-value by t-test.) (G) Histological analysis of P1 hearts from normal (WT) and diabetic (Akita) pregnancies. Ventricular wall thickness (RV, right ventricle; LV, left ventricle) and interventricular septum (IVS) thickness were analyzed histologically. Scale bar = 200 mm. (n = 5 and 4 for WT and Akita, respectively; data shown are mean ± SD, p-value by t-test). (H) Cell-size analysis of the cells isolated from P1 hearts from normal (WT) and diabetic (Akita) pregnancies using forward scatter by flow cytometry (FSC). At least 25,000 cells were measured per sample. Representative histogram from three flow cytometry measurements for each group (left) and the geometrical means of FSC (right). (n = 3, p<0.05 by t-test). DOI: https://doi.org/10.7554/eLife.29330.018 supplementation of nucleotides was sufficient to recapitulate the effect of glucose (Figures 4 and  5). These data suggest that nucleotide biosynthesis via the PPP is the key regulator of the promitotic/anti-maturation effect of glucose. Interestingly, cardiomyocyte maturation was not fully induced by blocking of the cell cycle with a CDK4/6 inhibitor or paclitaxel, neither of which directly impact the nucleotide kinetics. Therefore, it is not the cell cycle in general but the nucleotides themselves that blocks the maturation ( Figure 5-figure supplement 2). It is well-documented that there is generally an inverse correlation between cell proliferation and differentiation during developmental stages (Ruijtenberg and van den Heuvel, 2016). Our data raise an intriguing possibility that nucleotide biosynthesis serves as a nodal point balancing cell proliferation and differentiation during development.
Hyperglycemia as a potential teratogen in the fetal heart Clinically, maternal diabetes can accompany multiple complications including neuropathy, microvasculopathy, nephropathy, and insulin resistance. Although meta-analysis predicts that hyperglycemia itself is a major teratogen during diabetic pregnancy (Reece et al., 1996), it is often difficult to dissect the impacts of maternal complications on CHD as they are often subclinical. To our knowledge, our in vitro study is the first to demonstrate that environmental glucose itself, if excessive, directly impacts cardiac differentiation. The formation of the heart is regulated by both genetic and nongenetic factors, with the latter playing important roles particularly during late-stage cardiogenesis. An interesting aspect of the interaction between genetic and non-genetic mechanisms is that they seem to reinforce each other mutually. Our data suggest that the glucose metabolic environment is, on the one hand, a consequence of changes in cardiac genetic program, and on the other hand, a cause of the changes in cardiac gene expression.

Potential application to the translation
Understanding the metabolic signature of hESC-CMs will potentially open new methods for purifying these cells (Tohyama et al., 2016;Tohyama et al., 2013) or inducing their maturation (Drawnel et al., 2014). Considering that the inhibitors of the PPP and nucleotide biosynthesis have entered clinical trials for cancer treatment (Tennant et al., 2010;Vander Heiden, 2011), our data raise the possibility that manipulating this pathway may allow us to control the proliferation and maturation of cardiomyocytes for regenerative medicine.
With the advances in fetal diagnosis and surgical techniques, the number of CHD patients who survive childhood (and so have adult CHD) is growing rapidly by nearly 5% per year (Brickner et al., 2000). Maternal hyperglycemia is a common medical condition associated with 2-5-fold increase in CHD (Centers for Disease Control, 1990;Simeone et al., 2015;Yogev and Visser, 2009). Currently, 60 million women of reproductive age (18-44 years old) worldwide and approximately 3 million in the U.S. have diabetes mellitus. This number is predicted to double by 2030, posing a huge medical and economic burden (Gabbay-Benziv et al., 2015). Our findings will lay a foundation for understanding how the glucose environment regulates cardiogenesis and how disturbance of nongenetic factors affects the genetic program during the pathological development of the heart.

Mouse and cell lines
Wild-type and Akita mice were maintained on the C57BL/6 background according to the Guide for the Care and Use of Laboratory Animals published by the US National Institute of Health (NIH Publication No. 85-23, revised 1996). Housing and experiments were performed according to the Institutional Approval for Appropriate Care and Use of Laboratory Animals by the UCLA Institutional Animal Care and Use Committee (Protocol #2008-127-07). H9 (WA09) and UCLA4 (UCLA stem cell core) hESC lines were maintained as described before (Arshi et al., 2013). Authentication of hESCs was achieved by confirming the expression of pluripotency genes and protein markers. hESCs were routinely verified as mycoplasma-free using a PCR-based assay. hESCs were grown and differentiated in a chemically defined condition (Minami et al., 2012;Young et al., 2016;Zhu et al., 2017). Usage of all the human embryonic stem cell lines is approved by the UCLA Embryonic Stem Cell Research Oversight (ESCRO) Committee and the Institutional Review Boards (IRB) (approval #2009-006-04).

RNA-seq and data analyses
For the RNA-seq analyses shown in Figures 1D, E and 2C, RNA was extracted from hESC, hMP, hCP, hCM14, hCM28, hCM28 (in 25 mM glucose) and hCM28 (in 0 mM glucose) using TRIZOL (TheroFisher) and RNeasy kit (QIAGEN). 500 ng of DNaseI-treated RNA was used as input material for library preparation using the Illumina TruSeq mRNA kit (Illumina, RS-122-2001), according to manufacturer's instructions. Final libraries were sequenced as single-end 50 bp on the Illumina HiSeq2000 platform (GSE84814). Libraries for RNA-Seq shown in Figures 1D, E and 2C were prepared with the KAPA Stranded RNA-Seq Kit. The workflow consists of mRNA enrichment, cDNA generation, end repair, A-tailing, adaptor ligation, strand selection and PCR amplification. Different adaptors were used for multiplexing samples in one lane. Sequencing was performed on an Illumina HiSeq 3000 for a paired end 2 Â 150 run (GSE84815). A data quality check was done on an Illumina SAV. De-multiplexing was performed with the Illumina Bcl2fastq2 v 2.17 program. Short read sequences generated from an Illumina Sequencer were aligned to the UCSC human reference genome hg19 downloaded from support.Illumina.com (http://support.illumina.com/sequencing/ sequencing_software/igenome.html) using TopHat from the Tuxedo Tools. The average overall read mapping rate reached over 82 percent (average = 82.43 percent). The output was in the form of BAM (Binary Sequence Alignment/Map format) files. These outputs contain information for assigning location and quantifying the short-read alignments obtained from the RNA-seq samples. This is necessary for downstream analyses such as annotation, transcript abundance comparison and polymorphism detection. Counts or gene expression matrices were generated using HTSeq, which quantified the reads per transcript. The expression matrices were log transformed and normalized using the rlog function in the Deseq2 package. The normalized gene expression matrices were used as input for SaVanT (Signature Visualization Tools), which allowed for the visualization of molecular signatures directly related to heart development as seen in Figures 1D, E and 2C.

Immunocytochemistry and morphological analysis
Cells were fixed with 4% paraformaldehyde, blocked for an hour with 5% normal goat serum, and incubated with mouse alpha actinin antibody (Sigma) followed by Alexa fluor 488-conjugated secondary antibody (ThermoFisher). Images were taken with Zeiss LSM780 confocal microscopy. Sarcomere lengths were analyzed using Zeiss Zen software.

mtDNA-to-nDNA ratio analysis
Total DNA including mtDNA was extracted from cells using the PureLink DNA kit (ThermoFisher), and DNA purity and quantity were determined using a spectrophotometer. To determine the ratio between mitochondrial and nuclear DNA, qRT-PCR was performed on a Roche Lightcycler 480 using SYBR Green dye. Mitochondrial gene expression was corrected for nuclear gene expression values, and normalized to the value of the control group for each experiment as described before. Forward and reverse primer sequences are as follows: UUR forward, CAC CCA AGA ACA GGG TTT GT; UUR reverse, TGG CCA TGG GTA TGT TGT TA for mt DNA; B2-microglobulin forward, TGC TGT CTC CAT GTT TGA TGT ATC T; and B2-microglobulin reverse, TCT CTG CTC CCC ACC TCT AAG T.

Ca 2+ transient assay
Ca 2+ transient was measured as described (Shimizu et al., 2015). Briefly, hESC-CMs cultured in the presence or absence of glucose were loaded with 5 mM fluo-4 AM and imaged in Tyrode buffer containing 138.2 mM NaCl, 4.6 mM KCl, 1.2 mM MgCl, 15 mM glucose and 20 mM HEPES according to the manufacturer's instruction. Images were recorded on a Zeiss LSM 780 confocal microscope. Data analysis was carried out using the Zeiss Zen and ImageJ.

In vitro contractility assay
Contractility assessments were performed by utilizing a video-based technique with the UCSF Gladstone-developed Matlab program MotionGUI (Huebsch et al., 2015). The videos were converted from .mts to .avi format at native resolution using commercially available software and loaded into the MotionGUI program. The conversion between pixels and real distance was performed within the MotionGUI program using a reference image with unit divisions of 100 um, taken under the same objective and video zoom settings as the cell videos, to yield a pixel size of 0.681125. This pixel size was used for all contractility assessments. Motion vectors were calculated and the data were evaluated upon completion. All samples were subjected to the same post-processing procedures in order to ensure consistency during comparative analysis. Each video sample was post-processed using neighbor-based cleaning with the vector-based cleaning criterion within the program. The threshold for this post-processing method was set to two for all samples and was adequate for improving the signal-to-noise ratio enough to identify peaks clearly corresponding to beating events in most samples. A small number of videos suffering from significant noise issues were separately subjected to fast Fourier transform (FFT) frequency domain cleaning with a cut-off frequency of 1 Hz. Only one post-processing method was applied to one video at one time. All other parameters of the Motion-GUI program not outlined here were set to their respective default values.

Measurement of intracellular lactate level by Laconic
The lactate biosensor Laconic was a gift from Dr. Barros (San Martín et al., 2013). Overexpression of Laconic in hESC-CM was achieved using engineered adenoviruses encoding the construct. Expression of the construct was sufficiently high after 36-48 hr for FRET experiments or microscopy imaging. All cells were imaged live without fixation. Images (16 bit) were acquired using a microscope (Eclipse TE300; Nikon) fitted with a 60Â (1.4 NA) oil immersion lens (Nikon) and equipped with a filter cube comprising a CFP bandpass excitation filter, 436/20b, together with a longpass dichroic mirror (455DCLP; Chroma Technology Corp). Two LEDs (Philips Lumileds), one emitting at 455 ± 20 nm (royal blue) and the other emitting at 505 ± 15 nm (cyan) were used as light sources. Ratiometric FRET measurements were obtained from the YFP and CFP images acquired simultaneously using a Dual View image splitter (Optical Insights) equipped with a 505 nm longpass dichroic filter to separate the CFP and YFP signals, a CFP emission filter (480/30), and a YFP emission filter (535/40) (John et al., 2008). Images were captured with a Cascade 512B digital camera (Photometrics). Figure 3C were added and followed by washing.

XF24 extracellular flux analyzer
hESC-CMs were seeded onto a matrigel-coated XF24 Cell Culture Microplate (Seahorse Bioscience) at 2-7.5 10 4 cells/well with or without glucose (25 mM glucose of cardiac differentiation media). Oxygen consumption rate (OCR) was measured using an XF24 Extracellular Flux Analyser (Seahorse Bioscience) in unbuffered DMEM assay medium supplemented with 1 mM pyruvate, 2 mM glutamine and with or without 25 mM glucose. OCR was measured before and after the sequential addition of 0.75 mM oligomycin, 0.5 mM FCCP and 0.75 mM of rotenone/myxothiazol. OCR was normalized to protein concentration using a Bradford assay (Bio-Rad). Mitochondrial respiration was calculated as the difference between total and rotenone/myxothiazol rates. Maximal respiration was the response to FCCP. ATP-linked respiration was represented by the oligomycin-sensitive respiration rate, whereas uncoupled respiration was represented by the difference between oligomycin and rotenone/myxothiazol rates.

Multi-electrode array
hESC-CMs at the stage of hCM14 were plated on microelectrode arrays (MEAs) containing 120 integrated TiN electrodes (30 mm diameter, 200 mm interelectrode spacing). The MEAs were placed in an incubator with a temperature of 37˚C and 5% CO 2 . Two days were given to allow the cardiomyocytes to well attach the MEAs before recording started. Local field potentials at each electrode were collected over a period of 5 min every day in total with a sampling rate of 1 KHz using the MEA2100-HS120 system (Multichannel systems, Reutlingen, Germany). Data analysis was carried out using the MC_DataTool (Multichannel Systems), Origin (OriginLab Corporation) and Matlab (Math-Works). Data shown are based on three independent hESC-CM preparations.

Mass spectrometry-based metabolic measurements
The experiments were performed as described (Krall et al., 2016). Briefly, cells were seeded in 6well plates, so that the final cell count at the time of metabolite extraction was about 7*10 5 and this was even across all cell lines. To extract intracellular metabolites, cells were briefly rinsed with cold 150 mM ammonium acetate (pH 7.3), followed by the addition of 1 ml cold 80% MeOH on dry ice. Cell scrapers were used to detach cells, and the cell suspension was transferred into Eppendorf tubes. Extracted metabolites were transferred into glass vials and dried down under vacuum. For the LC-MS-based analysis, the samples were resuspended in 70% acetonitrile and 50 ml were injected onto a Luna NH2 column (150 mm x 2 mm, Phenomenex). Separation was achieved using A) 5 mM NH 4 AcO (pH 9.9) and B) ACN. The gradient started with 15% A) going to 90% A) over 18 min, followed by an isocratic step for 9 min and reversal to the initial 15% A) for 7 min. Metabolites were quantified with TraceFinder 3.3 using accurate mass measurements ( 3 ppm) and retention times of pure standards. Data analysis was performed using the statistical language R.
Gene expression analysis by quantitative reverse-transcriptase PCR RNA was extracted from the tissue or the cells cultured with a specific concentration of glucose together with or without titrated metabolic pathway inhibitors using the Direct-zol RNA mini prep kit (Zymo research). RNA was reverse-transcribed into complementary DNA using the qScript cDNA synthesis kit (Quanta Biosciences). Quantitative reverse-transcriptase PCR was performed using Viia7 (Applied Biosystems/ThermoFisher). In Figures 1B, 2B  duplicates with at least three independent wells, each of which is triplicated for qPCR reaction. The relative mRNA level is normalized to the expression level of 25 mM glucose without any chemicals. Forward and reverse primer sequences are as follows: GAPDH forward, TTGAGGTCAATGAAGGGGTC; GAPDH reverse, GAAGGTGAAGGTCGGAGTCA; TNNT2 forward, CAGAGCGGAAAAGTGGGAAGA; TNNT2 reverse, TCGTTGATCCTGTTTCGGAGA; NKX2-5 forward, GTTGTCCGCCTCTGTCTTCT;NKX2-5 reverse, TCTATCCACGTGCCTACAGC; PPARGC1A forward, GGTGCCTTCAGTTCACTCTCA; and PPARGC1A reverse, AACCAGAGCAG-CACACTCGAT.
18 F-FDG measurement by g-counter 18 F-FDG was obtained from the UCLA Department of Nuclear Medicine. Warmed pregnant mice or pups were injected intravenously or intraperitoneally, respectively, with~90 microCi (~3.33 MBq) of 18 F-FDG. After 2 hr, the mice were sacrificed. Preliminary experiments suggested that 2 hr was sufficiently long for 18 F-FDG to reach maximum accumulation in each organ and embryo. Fetal or neonatal hearts were separated from the other tissue (carcass), and the mass and radioactivity in both the hearts and the carcasses were measured using a standard balance and a Wizard 3' automatic gamma counter (Perkin Elmer), respectively. The radioactivity levels in the pup carcasses were higher than the detection limit of the gamma counter, so the expected gamma counter values for the pup carcasses were calculated on the basis of the decay-corrected injected dose of 18 F-FDG and known conversion values between microCi and CPM on the gamma counter. To calculate the 'Normalized FDG accumulation', radioactive accumulation in each heart was divided by heart weight and then further divided by the total radioactivity in each embryo or pup. This last normalization is to account for differences in 18 F-FDG injected dose and accessibility to the embryos and pups. Averages and standard errors of the mean (SEM) were calculated, and the values were normalized such that E10.5 embryo FDG accumulation was set to 100.

Immunostaining
Mouse embryos were isolated in cold PBS and fixed in 4% PFA for 1~2 hr, followed by equilibration in 30% sucrose in PBS solution overnight. The tissues were placed in 1:1 30% sucrose/OCT (Tissue-Tek, Electron Microscopy Sciences) solution for 1-2 hr, then in 100% OCT compound for 1 hr at 4˚C, and embedded in 100% OCT compound, carefully oriented in Cryomolds (Ted Pella). The blocks were immediately frozen on dry ice with isopropanol and stored at À80˚C. The sections were cut 5 mm with a Leica CM3050 S cryostat. The following primary and secondary antibodies were used: aactinin (mouse, 1:200, Sigma-Aldrich), Phospho-Histone 3 (rabbit, 1:250, Millipore), as well as Alexa Fluor 488 (green)-and Alexa Fluor 594 (red)-conjugated secondary antibodies specific to the appropriate species, which were used (1:500; ThermoFisher) for fluorescent staining. Sections were mounted with antifade mounting medium with DAPI (ThermoFisher), and analyzed by using AxioImager D1 (Carl Zeiss Microimaging, Inc).

Statistical analysis
ANOVA and Student's t-test were used to determine whether any statistically significant difference exists among independent groups.