Glucose promotes cell growth by suppressing branched-chain amino acid degradation

Glucose and branched-chain amino acids (BCAAs) are essential nutrients and key determinants of cell growth and stress responses. High BCAA level inhibits glucose metabolism but reciprocal regulation of BCAA metabolism by glucose has not been demonstrated. Here we show that glucose suppresses BCAA catabolism in cardiomyocytes to promote hypertrophic response. High glucose inhibits CREB stimulated KLF15 transcription resulting in downregulation of enzymes in the BCAA catabolism pathway. Accumulation of BCAA through the glucose-KLF15-BCAA degradation axis is required for the activation of mTOR signaling during the hypertrophic growth of cardiomyocytes. Restoration of KLF15 prevents cardiac hypertrophy in response to pressure overload in wildtype mice but not in mutant mice deficient of BCAA degradation gene. Thus, regulation of KLF15 transcription by glucose is critical for the glucose-BCAA circuit which controls a cascade of obligatory metabolic responses previously unrecognized for cell growth. Hypertrophic cardiomyocytes switch their metabolism from fatty acid oxidation to glucose use, but the functional role of this change is unclear. Here the authors show that high intracellular glucose inhibits the degradation of branched-chain amino acids, which is required for the activation of pro-growth mTOR signaling.

Glucose and branched-chain amino acids (BCAAs) are essential nutrients and key determinants of cell growth and stress responses. High BCAA level inhibits glucose metabolism but reciprocal regulation of BCAA metabolism by glucose has not been demonstrated. Here we show that glucose suppresses BCAA catabolism in cardiomyocytes to promote hypertrophic response. High glucose inhibits CREB stimulated KLF15 transcription resulting in downregulation of enzymes in the BCAA catabolism pathway. Accumulation of BCAA through the glucose-KLF15-BCAA degradation axis is required for the activation of mTOR signaling during the hypertrophic growth of cardiomyocytes. Restoration of KLF15 prevents cardiac hypertrophy in response to pressure overload in wildtype mice but not in mutant mice deficient of BCAA degradation gene. Thus, regulation of KLF15 transcription by glucose is critical for the glucose-BCAA circuit which controls a cascade of obligatory metabolic responses previously unrecognized for cell growth.
C ell proliferation, such as during fetal development or cancerous growth, is highly dependent on glucose metabolism which not only provides a source of energy but also supplies metabolites for biosynthesis of membrane lipids and nucleic acids, as well as contributes to epigenetic and posttranslational modifications [1][2][3][4] . The growth of post-mitotic cells, i. e, increases of cell mass, is also associated with increased glucose utilization [5][6][7][8] . The hypertrophic growth of the heart under stress conditions is a hallmark of pathological remodeling which ultimately leads to heart failure 9,10 . Cardiac pathological hypertrophy is accompanied by a switch of substrate metabolism from fatty acid oxidation to glucose use, resulting in a fetal like metabolic profile 5,11,12 . While such a metabolic reprogramming has been shown maladaptive for sustaining energy supply 13,14 , its role in cardiomyocyte growth is poorly understood. It has been suggested that metabolites of glucose, in particular, glucose-6-phosphate, play an important role in cell growth by activating the mechanistic target of rapamycin (mTOR) complex 15,16 . However, enhancing glucose utilization alone does not drive cell growth suggesting that it is unlikely that glucose or its metabolites directly stimulates mTOR 17,18 .
The Krüppel-like factors (KLFs) are a family of gene regulatory proteins that govern the transcriptional regulation of cell proliferation, differentiation and metabolism 19,20 . KLF15 has been shown as a negative regulator of cardiomyocyte hypertrophy, and KLF15 expression is negatively correlated with cell growth in proliferating and non-proliferating cells 21,22 . Despite the close association, molecular mediator(s) connecting the downstream of KLF15 and the growth signals have been elusive. The driver or suppressor of KLF15 expression at different developmental stages is also unknown. Interestingly, the expression profile of KLF15 during developmental and postnatal phases predicts a negative relationship with glucose utilization 23,24 .
Using an unbiased transcriptome analysis, we here identified glucose as a negative regulator of KLF15 expression. Increases of intracellular glucose, via downregulation of KLF15, suppressed the expression of BCAA degradation enzymes resulting in BCAA accumulation. While glucose did not directly modify the upstream signaling of cell growth, e.g., phosphatidylinositide-3kinase (PI3K) or mitogen-activated protein kinase (MAPK), activation of glucose-KLF15-BCAAs axis was required for the signal transduction through the mTOR complex. Furthermore, upregulation of KLF15 in an in vivo model of cardiac pressure overload prevented the development of cardiomyocyte hypertrophy in wildtype mice but not in mice deficient of BCAA degradation genes. Collectively, our results identified a regulatory circuit between glucose and BCAA, and unveiled a previously unrecognized pathway for metabolic regulation of cell growth.

Results
Glucose negatively regulates BCAA degradation. Previously we have demonstrated that overexpression of an insulin independent glucose transporter Glut1 in cardiomyocytes (Glut1-TG) increased glucose uptake and utilization but did not alter cardiac function or survival of the mice under unstressed condition 17,25 . RNA microarray and gene ontology analysis of the Glut1-TG hearts revealed that downregulation of the "branched-chain amino acids (BCAAs) degradation" pathway was among the most enriched terms (Fig. 1a), which was confirmed by the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways analysis (Supplementary Table 1). A total of 26 out of 46 genes in the BCAA degradation pathway was downregulated in the Glut1-TG heart ( Supplementary Fig. 1a, b), and the changes of the key enzymes were confirmed by real-time PCR (Fig. 1b, c) 26 . The mRNA levels of branched-chain amino acid transaminase 2 (BCAT2) and branched-chain alpha-keto acid dehydrogenase (BCKDH) complex, two enzymes that catalyzed the initial and committed steps of BCAA degradation to branched-chain acyl-coA, were significantly downregulated, so were multiple enzymes involved in downstream reactions for converting the CoA products into the TCA cycle (Fig. 1b, c). The mRNA level of mitochondrial targeted 2C-type serine/threonine protein phosphatase (PP2Cm), that dephosphorylates and activates BCKDH complex, was also decreased (Fig. 1b, c). The protein levels of the BCAT2, branched-chain keto acid dehydrogenase E1 Alpha Polypeptide (BCKDHA), and PP2Cm were also significantly decreased in Glut1-TG hearts (Fig. 1d). No reduction was seen in the mRNA and protein expression of BCKD kinase (BCKDK) in Glut1-TG hearts ( Supplementary Fig. 1c, d).
Consistent with the in vivo observations in mice, increasing glucose uptake by either overexpression of Glut1 or high glucose medium (HG, 25 mM) reduced the expressions of BCAT2, BCKDHA, and PP2Cm, the three key proteins in BCAA degradation pathway, in both primary cells and proliferating cell lines (Fig. 1e, Supplementary Fig. 1e-h). Taken together, these data indicate that high glucose suppresses the expression of BCAA degradation enzymes.
We subsequently performed targeted metabolomics analysis to determine whether the downregulation of gene expression in Glut1-TG hearts affected the BCAA degradation in vivo. Increased glucose uptake in Glut1-TG hearts led to a higher level of glycolytic metabolites ( Supplementary Fig. 1i, j), as expected. Importantly, the level of BCAAs was also increased while the downstream BCAA metabolites, e.g., α-keto-βmethylvalerate (KMV) and α-ketoisocaproate (KIC) were decreased in Glut1-TG hearts (Fig. 1f). Notably, levels of other amino acids exhibited the opposing trend compared to that of the BCAAs ( Supplementary Fig. 1i), suggesting that high glucose selectively inhibited BCAA degradation. Independent biochemical assays showed a 2.5-fold increase of BCAA level in Glut1-TG hearts (Fig. 1g). In addition, high glucose medium or Glut1 overexpression also increased the BCAA content in both myocyte and non-myocyte cell types (Fig. 1h, Supplementary Fig. 1k). Taken together, these observations strongly suggest that glucose negatively regulates intracellular BCAA degradation.
KLF15 is the target of high intracellular glucose. We next searched for the upstream regulatory mechanisms mediating the transcriptional suppression of BCAA degradation pathway. The microarray data showed~2-fold downregulation of KLF15 in Glut1-TG hearts ( Supplementary Fig. 2a), which was confirmed by RT-PCR and immunoblot (Fig. 2a, b). Furthermore, both high glucose medium and Glut1 overexpression reduced the KLF15 mRNA and protein levels in neonatal rat cardiomyocytes (NRCMs) (Fig. 2c, d, Supplementary Fig. 2b). Similar observations were also made in other cell lines including H9C2 and HEK-293 ( Supplementary Fig. 2c, d). Previously, KLF15 has been shown to mediate the transcription of genes for BCAA degradation in multiple organs, including skeletal muscle, liver, and heart 27,28 . We found that overexpression of KLF15 in the presence of high glucose is sufficient to normalize both mRNA and protein expression of BCAA degradation enzymes (Fig. 2e, Supplementary Fig. 2e, f), suggesting that glucose negatively regulates BCAAs degradation through a KLF15 dependent mechanism.
To investigate the mechanisms by which high glucose suppressed KLF15 expression, the KLF15 promoter region containing~1 kb upstream sequence from the transcription start site or its serially truncated segments was cloned into a luciferase vector and the luciferase activity was examined (Fig. 3a). High glucose medium or Glut1 overexpression significantly reduced the luciferase activity containing 1 kb promoter sequence and the combination of the two showed synergetic effects ( Supplementary  Fig. 2g). However, the suppression effect on luciferase activity by high glucose was no longer observed in smaller promoter constructs (Fig. 3a), indicating that the glucose response element(s) is located between the −1068 to −541 bp regions. We found two potential binding elements (CREI and CREII) for the cAMP response element binding protein (CREB) in this region (http://jaspar.genereg.net/); both exhibited high similarity with CRE consensus sequence and were conserved across species (Fig. 3b). The chromatin immunoprecipitation (ChIP) revealed that endogenous CREB bound to both CREI and CREII sites ( Fig. 3b). Importantly, high glucose significantly reduced CREB binding occupancy of CREI and CREII on the KLF15 promoter (Fig. 3c). To further test if high glucose suppressed the CREB transcriptional activity, cells expressing a luciferase vector containing three CREB consensus elements (CRE-luc) were subjected to high glucose medium or Glut1 overexpression. Both measures significantly decreased CRE luciferase activity (Fig. 3d). We also observed that the phosphorylation of CREB at Ser 133, critical for CREB transcriptional activation 29 , was significantly decreased in cells cultured in high glucose medium or overexpressing Glut1 (Fig. 3e) Physiological stimuli for growth that caused increased glucose uptake e.g., insulin or insulin like growth factor 1 (IGF-1) also induced similar changes of KLF15 and BCAA degradation enzymes ( Supplementary Fig. 3e-g) 31 . While overexpression of Glut1 did not further increase myocyte cell size in response to PE stimulation ( Supplementary Fig. 3h), reducing glucose uptake by knocking down Glut1 (sh-Glut1, Supplementary Fig. 3i, j) was sufficient to attenuate the downregulation of KLF15 and BCAA Data shown as mean ± s.e.m. P values were determined using unpaired Student's t-test (a, b, c, d) or one-way ANOVA followed by Newman-Keuls comparison test (e) Fig. 1 Glucose negatively regulates BCAA degradation. a GO enrichment analysis of differentially regulated genes in Glut1-TG hearts using DAVID. Top 5 clusters of either downregulated or upregulated genes are shown. b qRT-PCR analyses of selected BCAA degradation genes in Glut1-TG and WT mouse hearts. The expression was normalized to 18S rRNA and reported as fold change over WT (*p < 0.05 vs. WT, n = 6 degradation enzymes in PE stimulated cells (Fig. 4a, b). Furthermore, the increases in cell size and atrial natriuretic factor (ANF) promoter activity was significantly reduced by knocking down Glut1 (sh-Glut1) in the presence of PE (Fig. 4c, d).
Knockdown of KLF15 reversed the anti-hypertrophy effect by sh-Glut1 in PE treated groups (Fig. 4c, d, Supplementary Fig. 3k), suggesting that KLF15 was an important mediator of glucosedependent cell growth. We also found that overexpression of KLF15 prevented the downregulation of BCAA degradation enzymes in PE treated cells, in parallel with the reductions in cell . Taken together, these data reveal a mechanism that links glucose reliance in cell growth to BCAA degradation pathway through KLF15.
Glucose modulates mTOR in an energy independent manner.
In searching for the end effector of glucose mediated cell growth, we found that knockdown of Glut1 attenuated PE induced mTOR activation in NRCMs ( Supplementary Fig. 4a). Consistent with previous observations, inhibition of mTOR activity by rapamycin treatment is sufficient to prevent PE induced cardiomyocyte growth ( Supplementary Fig. 4b, c). As glucose is an important energy substrate and the mTOR activity is highly sensitive to cellular energy status, we investigated whether glucose is required to maintain cellular energy balance during the growth. Replacing glucose with either pyruvate (12 mM) or lactate (12 mM) in the culture medium attenuated NRCMs' growth in response to hypertrophic stimulus ( Fig. 5a) with no change in viability or intracellular ATP level compared with cells cultured with glucose ( Supplementary Fig. 4d, e). Moreover, only NRCMs cultured with glucose showed mTOR activation in response to PE treatment ( Fig. 5b, Supplementary Fig. 4f). However, we observed that NRCMs cultured with pyruvate or lactate exhibited increased phosphorylation of AMP-activated protein kinase (AMPK) indicating that AMPK was activated ( Supplementary Fig. 4f). AMPK is an energy sensor and a negative regulator of mTOR activity and cell growth 32 . To determine whether restricting glucose inhibited mTOR through activation of AMPK, we prevented AMPK activation by including both lactate (6 mM) and acetate (6 mM) as the replacement of glucose in the culture (Supplementary Fig. 4d, e, Fig. 5b). NRCMs cultured with lactate and acetate, nevertheless, remained resistant to mTOR activation and hypertrophic growth by PE (Fig. 5a, b), suggesting that AMPK activation is not required for mTOR inhibition under these conditions. Similarly, rat adult cardiomyocytes cultured with pyruvate or lactate showed no activation of AMPK but showed attenuated mTOR activation and reduced hypertrophic growth during PE stimulation ( Supplementary Fig. 4g, h). Taken together, these observations show that independent of its function in energy provision, glucose plays an essential role in mediating mTOR activation in response to growth stimulation.  Fig. 4j) and restored mTOR activation by PE in KLF15 overexpressing cells (Fig. 6c, Supplementary  Fig. 4k). Similarly, the downregulation of BCAA degradation enzymes by PE was abolished in cardiomyocytes cultured with non-glucose medium ( Supplementary Fig. 4l). Knocking down PP2Cm or BCAT2 was sufficient to reactivate mTOR during PE treatment in both neonatal and adult rat cardiomyocytes cultured even in non-glucose medium and restored the growth response to the same extent as in cells cultured with glucose ( Fig. 6d, e, Supplementary Fig. 4m-p). Taking together, these data suggest that suppression of BCAA degradation pathway by glucose is required for mTOR activation during cell growth.
Crosstalk between growth stimulus and metabolic response. Although glucose is required for cell growth, high glucose alone did not stimulate mTOR activation in cardiomyocytes in the absence of hypertrophic stimuli ( Supplementary Fig. 5a). The Glut1-TG hearts did not develop hypertrophy without stress 17 , and high glucose medium did not increase cell size in NRCMs ( Supplementary  Fig. 5b). These findings suggest that BCAA accumulation induced by increased glucose reliance is required but not sufficient to stimulate cell growth, i.e., the activation of mTOR requires additional input even in the presence of metabolic reprogramming. We thus investigated the cross talk between glucose and the two upstream regulators of mTOR signaling: PI3K and MAPK pathways. As expected, PE treatment stimulated both PI3K and MAPK pathways ( Supplementary Fig. 5c), while high glucose had no effect on either ( Supplementary Fig. 5d). Similarly, removing glucose or overexpression of KLF15 had no effect on activation of either PI3K or MAPK in response to PE treatment (Fig. 7a, b‚ Supplementary  Fig. 5e), but strongly suppressed mTOR activation and cardiomyocyte growth in response to PE (Figs. 4f, 5a, b, 6c, Supplementary Fig. 4i). In addition, inhibition of PI3K pathway by wortmannin or inhibition of MAPK by U0126 was sufficient to inhibit mTOR activation during PE stimulation but had no effect on BCAA degradation pathway (Fig. 8a, Supplementary Fig. 5f).  Knockdown of KLF15 or PP2Cm in cells subjected to PE stimulation, which effectively suppressed BCAA degradation pathway, failed to restore mTOR activation or hypertrophic growth when PI3K or MAPK was inhibited (Fig. 8b, c, Supplementary Fig. 5g-l), indicating that PE induced mTOR activation also required intact PI3K or MAPK signaling. Taken together, these data suggest that the activation of mTOR during a growth response requires input signal via the PI3K and/or MAPK pathway, as well as the metabolic response via increased intracellular glucose (Fig. 8d). Either component is indispensable, and they act cooperatively to sustain a growth response.
Glucose-BCAA circuit regulates cardiac hypertrophy in mice.
Increased glucose reliance, downregulation of KLF15 and impaired BCAA degradation have been demonstrated in hearts with pathological hypertrophy 6,21,28 , but their role as a regulatory circuit for hypertrophic response has never been tested. If our hypothesis is correct, the anti-hypertrophy effect of KLF15 will not be observed in mice deficient of BCAA degradation such as PP2Cm KO. We thus sought to increase KLF15 expression in the heart of WT and PP2Cm KO mice, via retro-orbital injection of an adeno-associated virus serotype 9 vector carrying KLF15 and directed by the cardiac-specific cTNT promoter (AAV9-KLF15) 35 .   This approach showed high efficiency and specificity of gene expression in the heart (Supplementary Fig. 6a, b). As high level overexpression of KLF15 in transgenic mice causes arrhythmia due to a regulatory role of KLF15 on potassium channel 36 , we titrated the dosage of AAV9-KLF15 to limit the KLF15 overexpression to~2-fold (Fig. 9a, Supplementary Fig. 6c). The expression level of KLF15 was stable for up to 5 weeks and did not cause any overt phenotype in WT mice. Importantly, AAV9-KLF15 administration was sufficient to increase multiple enzymes involved in BCAA degradation pathway (Fig. 9a).
We found that the increases in heart weight and myocyte cross-sectional area, induced by TAC, were significantly reduced by KLF15 overexpression in WT but not in PP2Cm KO mice (Fig. 9b, c). Immunoblot analysis demonstrated that AAV9-KLF15 treatment significantly reduced the phosphorylation of p70 S6K and mTOR after TAC in WT mice but not in PP2Cm KO mice (Fig. 9d). Importantly, PI3K and MAPK signaling was activated to a similar extent in all the groups subjected to TAC compared with sham operated mice ( Supplementary Fig. 6c). In addition, molecular markers for pathological cardiac hypertrophy ANP and BNP, were markedly elevated in WT-TAC and a greater increase observed in PP2Cm KO-TAC hearts (Fig. 9e). AAV9-KLF15 treatment significantly reduced the expression of ANP and BNP in WT but not in PP2Cm KO hearts after TAC (Fig. 9e). Echocardiographic measurements demonstrated that TAC induced cardiac dysfunction in both WT and PP2Cm KO mice compared with sham operated mice, which was rescued by AAV9-KLF15 in WT only (Fig. 9f, Supplementary Table 2). Collectively, these data demonstrate that suppression of KLF15-BCAA degradation pathway is required for activation of mTOR and development of pathological hypertrophy of adult hearts in vivo.

Discussion
The study reveals a regulatory circuit between glucose and BCAA that serves as a metabolic determinant of growth signaling. In a variety of cell types, we observed that increased intracellular glucose inhibits the expression of KLF15 and its downstream genes resulting in decreased BCAA degradation. Such a response is required for sustaining the activation of mTOR during growth stimulation even though energy provision by glucose is dispensable. Importantly, disruption of this regulatory cascade alleviates cardiomyocyte hypertrophy during pressure overload in mice demonstrating that this mechanism plays a critical role in cellular growth response in disease.
The mTOR complex is a nutrient sensor and a signaling hub where various inputs are vetted and relayed cumulating in changes of protein synthesis, cell growth, and survival. Here we demonstrate that activation of mTOR requires growth stimuli that act in concert with increased glucose metabolism. This is consistent with previous observations that mTOR activation by insulin stimulation or increased hemodynamic load in the heart requires increased glucose metabolism 15,16 . However, prior hypothesis that mTOR activity is modulated by a direct interaction of glucose metabolite(s) with protein(s) in the mTOR complex has not been supported by direct evidence. Here we have identified a cascade of mechanisms linking glucose metabolism to the mTOR activity through the inhibition of BCAA degradation. Previous studies have shown that mTOR activity is regulated by nutrients, mostly in the form of amino acids, especially leucine, the most abundant BCAA [37][38][39][40] . Here we demonstrate that increased glucose metabolism in growing cells modulates the amino acid levels by suppressing the BCAA degradation pathway, thus, reveal a missing link in the coordination of cell metabolism and intracellular nutrient environment for cell growth. Furthermore, the identification of the glucose-BCAA circuit reveals a fundamental mechanism in metabolic regulation that will likely advance research on metabolic signaling beyond cell growth.
Similar to our observations in this study, activation of mTOR by amino acids also requires the presence of upstream signals, such as growth factors 39 . These observations indicate that metabolic responses elicited by growth signals, instead of independently stimulating cell growth, play an obligatory role in setting the intracellular environment in favor of growth signal transduction. Understanding the molecular mechanisms underlying the metabolic determinant of mTOR activity, therefore, presents new avenue to target cell growth independent of growth stimulation.
We find that restoration of mTOR activity is sufficient to promote the growth of both neonatal and adult rat cardiomyocytes in the absence of glucose. This seemingly surprising observation suggests that glycolysis is not required in the growth of post-mitotic cells once mTOR is activated. Substitution of glucose with pyruvate, lactate and/or, acetate is sufficient to meet the energy demand and prevent AMPK activation. However, glucose metabolites also fuel several accessary pathways, e.g., hexosamine biosynthetic pathway or pentose phosphate pathway. Since gluconeogenesis is normally not present in cardiomyocytes, long term culture with non-glucose substrates could deplete metabolites produced by these pathways which have been shown critical for cell growth 3,41 . It is possible that these metabolites are Fig. 4 Suppression of BCAA degradation by high glucose is required for cardiomyocyte growth in response to PE. a, b NRCMs transduced with indicated adenovirus for 72 h were treated with phenylephrine (PE, 100 μM) or vehicle for 6 h. a Representative immunoblots of cytosolic and nuclear fractions (left) and statistical analysis of densitomeric measurement of KLF15 (right) are shown (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. sh-con/PE, n = 4). b Representative immunoblots of cell lysates (left) and statistical analyses of densitomeric measurements of BCAT2 and BCKDHA (right) are shown (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. sh-con/PE, n = 5). c NRCMs transduced with indicated adenovirus were treated with phenylephrine (PE, 100 μM) or vehicle for 48 h. Cellular surface area in each group was quantified and expressed relative to the control (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. shcon/PE, & p < 0.05 vs. sh-Glut1/PE, n = 4). Scale bar, 25 μm. d NRCMs transfected with 1 μg ANF promoter luciferase reporter (ANF-luc) were transduced with indicated adenovirus for 36 h and further incubated with phenylephrine (PE, 100 μM) or vehicle. The luciferase activity was measured after 36 h of PE (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. sh-con/PE, & p < 0.05 vs. sh-Glut1/PE, n = 4). e NRCMs transduced with indicated adenovirus were treated with phenylephrine (PE, 100 μM) or vehicle for 6 h. Representative immunoblots of cell lysates (upper) and statistical analyses of densitomeric measurements of BCAT2, BCKDHA and PP2Cm (lower) are shown (*p < 0.05 vs. LacZ/Vehicle, # p < 0.05 vs. LacZ/PE, n = 4). f NRCMs transduced with indicated adenovirus were treated with phenylephrine (PE, 100 μM) or vehicle for 48 h. Cellular surface area in each group was quantified and expressed relative to the control (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. sh-con/PE, & p < 0.05 vs. KLF15/PE, n = 4). Scale bar, 25 μm. g NRCMs transfected with 1 μg ANF-luc were transduced with indicated adenovirus followed by incubation with phenylephrine (PE, 100 μM) or vehicle for 36 h. The luciferase activity was measured (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. sh-con/PE, & p < 0.05 vs. KLF15/PE, n = 4). Data shown as mean ± s.e.m. P values were determined using one-way ANOVA followed by Newman-Keuls comparison test (a, b, g) or Kruskal-Wallis test followed by Dunn's comparison test (c, d, e, f) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05362-7 not sufficiently depleted during the short-term experiments in the present study hence not affecting cell growth. It also raises the intriguing question whether other mechanisms are present to compensate for the loss of glycolysis in those cells during hypertrophic growth. Further investigations along this line may lead to discovery of novel metabolites and pathways that contribute to the growth of post-mitotic cells.
A key finding of this study is the identification of KLF15 as a critical node connecting metabolism and cell growth. KLF15 has long been recognized as a pleiotropic transcriptional factor 20,42 . KLF15-mediated transcriptional mechanisms regulate cell differentiation, circadian rhythm, ion channel activity, and multiple aspects of metabolism 23,27,36,[43][44][45] . Despite clear evidence for a role of KLF15 in determining the metabolic phenotype of differentiated cells, the upstream mechanism(s) that drive KLF15 expression have not been defined. Interestingly, KLF15 expression is substantially suppressed during development and in proliferating cells when basal glucose uptake and utilization is high [21][22][23]46 . Results from this study reveal a regulatory mechanism by glucose through which immature cells suppress KLF15 to sustain a metabolic profile favorable for cell growth. Consistent with the notion, exponential increases of KLF15 expression occur at the end of postnatal growth period which coincide with the switch to insulin regulated glucose disposal in the heart coupled with increased fatty acid oxidation and amino acid catabolism 23,24 . When the metabolic pattern returned to the fetal form in hypertrophic hearts, KLF15 level is reduced 21 . Indeed, the present study confirmed prior observation that both physiological and pathological growth stimuli, known to stimulate glucose uptake, reduced the KLF15 expression 21,47 . Here we have identified one glucose responsive mechanism that suppresses KLF15 transcription through reducing the binding of CREB on its promoter region. KLF15 expression is also responsive to multiple hormonal regulations such as glucocorticoids or during fasting of which the molecular mechanisms are unknown. The role of intracellular glucose in these responses warrants investigation.
Several observations have been made that the expression of KLF15 and BCAA degradation enzymes are downregulated in failing hearts 21,28 . However, Glut1-TG mice was able to maintain cardiac function at baseline and protected against pressure overload induced cardiac dysfunction even though the KLF15-BCAA pathway was inhibited. We speculate that the seemingly contradictory outcome observed in Glut1-TG hearts reflects the complex role of substrate metabolism in modulating myocardial energetics and signaling transduction. In Glut1-TG mice, the tremendous capacity of glucose supply allows glucose to become the predominant fuel and is able to maintain cardiac energetics during pressure overload, a condition when downregulation of fatty acid oxidation (FAO) compromises myocardial energy supply while increased glucose utilization through endogenous mechanisms is not sufficient to maintain myocardial energetics 17,48 . The study by Sun et al. suggests that accumulation of BCAA metabolites, BCKAs, contribute to the development of TAC induced cardiac dysfunction in PP2Cm KO hearts 28 . We did not observe BCKAs accumulation in Glut1-TG hearts (Fig. 1f), since the entire BCAA degradation pathway was downregulated including the enzyme converting BCAAs to BCKAs. Therefore, improved energy metabolism in Glut1-TG hearts likely contributes to delayed progression to heart failure. However, compared to other interventions of sustaining myocardial energetics, such as preventing the downregulation of FAO, improved function in Glut1-TG after TAC was not accompanied by the expected reduction of cardiac hypertrophy 49 . These observations are consistent with the role of glucose in regulating cell growth independent of its role in ATP production, as shown in the present study.   The identification of the glucose-KLF15-BCAA regulatory circuit has potential implications in the cardiac remodeling of patients with diabetes or metabolic disorders. BCAAs catabolic enzyme gene expression is reduced in individuals with the metabolic syndrome and obesity 50,51 . Metabolomics studies has demonstrated a positive correlation between circulating BCAA levels and insulin resistance in obesity, and a high blood BCAA level is a risk factor for future development of diabetes 52,53 . These observations suggest a potential feedback regulation between glucose and BCAA metabolism 54 . A further understanding of the molecular mechanisms holds promise of revealing novel therapeutic targets for metabolic disorders.
In conclusion, our study demonstrates that glucose negatively regulates BCAAs degradation pathway through a KLF15 dependent mechanism. Mobilization of this pathway is required for mTOR activation and the development of cardiac hypertrophy. Thus, the study uncovers a molecular link between the metabolic programming and the growth signaling in the post-mitotic cells. Importantly, identification of the glucose and BCAA circuit adds to the fundamental principle of metabolic regulation, which has significant implications beyond cell growth.

Methods
Animal model. A transgenic mouse expressing human Glut1 driving by the alphamyosin heavy chain promoter (designated as Glut1-TG) was generated on an FVB background 17 . PP2Cm systematic knock out mice (designated as PP2Cm KO) was achieved by gene targeting and maintained on C57BL/6 background 55 . All the mice were housed at 22°C with a 12-h light, 12-h dark cycle with free access to water and standard chow. The experiments included in this study were performed with male mice. All protocols concerning animal use were approved by the Institutional Animal Care and Use Committee at University of Washington.
Transverse aortic constriction (TAC) surgery. PP2Cm KO mice and their littermate controls at the age of 10-12 weeks underwent TAC or sham surgery 56 . Briefly, mice were anesthetized with an intraperitoneal injection of 130 mg/kg ketamine and 8.8 mg/kg xylazine in saline. Mice were intubated with 20 G cannula and ventilated 140 breaths per minute by small animal TOPO ventilator (Kent Scientific). The aortic arch was exposed via a left thoracotomy and by carefully separating the thymus. A constriction of the transverse aorta was generated by tying a 6-0 Ethilon ligature against a 27-gauge blunt needle around the aorta between the brachiocephalic and left common carotid arteries. Promptly the needle * * * * * *  Fig. 8 Activation of mTOR requires both growth stimulus and metabolic checkpoint. a NRCMs with preincubation of wortmannin (2 μM, 1 h) or vehicle were treated with phenylephrine (PE, 100 μM) or vehicle for 6 h. Immunoblots of total cell lysates (left) and statistical analyses of densitomeric measurements of BCAT2, BCKDHA, and PP2Cm (right) are shown (*p < 0.05 vs. Vehicle, n = 5). b, c NRCMs transduced with indicated adenovirus were preincubated with 2 μM wortmannin for 1 h and subsequently treated with phenylephrine (PE, 100 μM) or vehicle. b Immunoblots of total cell lysates (left) and statistical analyses of densitomeric measurements of p-p70 S6K (Thr 389) and p-mTOR (Ser 2448) (right) are shown (*p < 0.05 vs. sh-con/Vehicle, # p < 0.05 vs. sh-con/PE, n = 4). c 48 h after PE treatment, myocytes were fixed and stained with anti-Troponin T. Cell surface area in each group was quantified and expressed relative to the control (*p < 0.05 vs. sh-con/Vehicle, #p < 0.05 vs. sh-con/PE, n = 4). Scale bar, 25 μm. d Schematic illustration of the working hypothesis on how glucose regulates cell growth through modulating KLF15 mediated transcriptional control of BCAA degradation. Data shown as mean ± s.e.m. P values were determined using one-way ANOVA followed by Newman- Keuls comparison test (a, b) or Kruskal-Wallis test followed by Dunn's comparison test (c) was removed and the chest and skin were closed by 5-0 polypropylene suture. The animal was removed from ventilation and kept on a heating blanket during recovery from anesthesia. SR buprenorphine (0.05 mg/kg) was administered subcutaneously for analgesia. Sham operated mice underwent all the same procedures as TAC mice excluding the constriction of the aorta. All mice were monitored every 12 h during the first 72 h post-surgery, followed by daily visits over the next 4 weeks.
Transthoracic echocardiography. The mice were anesthetized and maintained with 1-2% isofluorane in 95% oxygen. Trans-thoracic echocardiography was conducted at 2 and 4 weeks post TAC surgery with Vevo 2100 high-frequency, high-resolution digital imaging system (VisualSonics) equipped with a MS400 MicroScan Transducer. A parasternal short axis view was used to obtain M-mode images for analysis of fractional shortening, ejection fraction, and other cardiac functional parameters.
Bioinformatics analysis. Glut1-TG and WT hearts were subjected to microarray analysis (Affymetrix). The preselected genes (Fold change > 1.25, p < 0.05 vs. WT) were analyzed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) (http://david.abcc.ncifcrf.gov) 57,58 . The lists of genes showing either down-regulation (806 genes) or up-regulation (1248 genes) in GLUT1-TG hearts were separately entered into the DAVID and subjected to Functional  were seeded onto 96-well plates. On second day, the cells were changed to serum free medium for 24 h followed by incubation with either DMEM or glucose-free medium containing non-glucose substrates for 48 h before the CTB assay.
Construction of adenoviral expression vectors. Briefly, pBHGloxΔE1,3Cre (Microbix), including the ΔE1 adenoviral genome, was co-transfected with pDC shuttle vector containing the gene of interest into 293 cells using Lipofectamine 2000 (Invitrogen). Through homologous recombination, the test genes were integrated into the E1-deleted adenoviral genome. The viruses were propagated in 293 cells as described 60 . The cDNA clones for Glut1 and KLF15 were purchased from Origene. Adenovirus harboring β-galactosidase (Ad-LacZ) was used as a control.
Generation and administration of AAV (serotype 9). KLF15 cDNA and GFP were separately cloned into ITR-containing AAV plasmid harboring the chicken cardiac TNT promoter to yield constructs pAAV9.cTnT-KLF15 and pAAV9.cTnT-GFP, respectively. AAV was packaged in 293 T cells using AAV9:Rep-Cap and pAd:deltaF6 (Penn Vector Core); then purified and concentrated by gradient centrifugation 61,62 . AAV9 titer was determined by Droplet Digital PCR (ddPCR). AAV9 virus (1 × 10 12 virus genome per animal) was injected into PP2Cm KO mice and their control littermates by retro-orbital injection.
Cryosectioning and immunofluorescence of GFP. To visualize transgene distribution after retro-orbital injection, hearts from mice injected with AAV9-GFP or PBS were harvested and fixed for four hours at 4°C in PBS containing 0.1% glutaraldehyde/1.5% paraformaldehyde/20% sucrose, embedded in Tissue-Plus O. T.C Compound (Fisher Scientific) and quickly frozen in liquid nitrogen. Cryosections of 5 µm thickness were cut and GFP expression was detected by anti-GFP antibody (1: 400, A-11122, Invitrogen). Slides were mounted with Prolong Diamond Antifade Mountant containing DAPI (Life Technology).
Targeted metabolite profiling. Targeted reverse phase (RP) liquid chromatography (LC)-mass spectrometry (MS) metabolite analysis was performed at the University of Washington Northwest Metabolomics Research Center 63 . Briefly, frozen heart tissues (~20 mg) were homogenized in 2 mL 80:20 methanol:water. The soluble extracts were collected and then dried at 30°C in a Speed-Vac. Dried extracts were reconstituted in 200 μL of 5 mM ammonium acetate in 95% water/ 5% acetonitrile + 0.3% acetic acid (pH = 3.5) and filtered prior to LC-MS analysis. Targeted LC-MS analysis was performed in RP chromatography mode on two parallel C18 analytical columns, one for positive and the other for negative MS ionization mode, respectively. LC system was composed of two Agilent 1260 binary pumps, an Agilent 1260 auto-sampler and an Agilent 1290 column compartment containing a column-switching valve (Agilent Technologies). The chromatography was performed in RP mode using solvents A (95% water/5% acetonitrile + 0.3% acetic acid) and B (5% water/95% acetonitrile + 0.3% acetic acid): 5% B for 2 min, 5% B to 80% B in 3 min, 80% B for 3 min, 80%B to 5% B in 3 min, and 5% B for 7 min (18 min total for each ionization mode). After the chromatographic separation, MS ionization and data acquisition was performed using AB Sciex QTrap 5500 mass spectrometer (AB Sciex) equipped with electrospray ionization (ESI) source. Multiplereaction-monitoring (MRM) mode was used for targeted data acquisition. Total 137 metabolites were measured as relative MRM peak areas using MultiQuant 2.1 software (AB Sciex).
Transient transfection and luciferase assays. Neonatal cardiomyocytes were plated on 12-well plates and transfected with 1 μg of the indicated luciferase vectors with or without overexpression plasmids using Fugene 6. Six hours after transfection, the cells were washed and transduced with the indicated adenovirus. Then the cells were lysed with Passive Lysis Buffer, and the transcriptional activity was determined using a luciferase assay system (Promega). For each construct, more than three independent experiments were performed in triplicate.
Chromatin immunoprecipitation (ChIP). Neonatal cardiomyocytes were treated with 1% formaldehyde for 10 min to cross-link proteins and chromatin. The reaction was stopped by adding 0.125 M Glycine for 5 min. Cells were washed with cold PBS twice. Cells were then resuspended in 1 ml ChIP lysis buffer (20 mM Tris-HCl (pH 8.0), 85 mM KCl, 0.5% NP-40) for 10 min at 4°C and centrifuged at 2600×g to pellet the nuclei. The cell nuclei were resuspended in 400 μL nuclei lysis buffer (50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS) and then subjected to sonication 5 times for 30 s with 30 s intervals. Purified chromatin was analyzed on a 1% agarose gel to determine the shearing efficiency. As a control, normal IgG (sc-2027, Santa Cruz Biotechnology) was used as a replacement for the CREB antibody (sc-186×, Santa Cruz Biotechnology). The ChIP procedure was performed as in the supplier's protocol (Active Motif). The primer information is listed in Supplementary Table 3.
Histological analyses. Mouse heart tissues were rinsed with PBS and fixed in 10% neutral buffered formalin overnight. Fixed samples were dehydrated, embedded in paraffin wax and sectioned for Wheat Germ Agglutinin staining. The outline of cross-sectional areas for 100-150 myocytes was traced and quantified with ImageJ software in each section. Suitable cross sections of the myocardium were defined as having nearly circular capillary profiles and circular-to-oval myocyte sections.
RNA extraction and quantitative real-time PCR. Total RNA was extracted from cells using Quick-RNA MiniPrep Kit (Zymo Research) and from frozen LV tissue using the RNeasy Fibrous Tissue Mini Kit (Qiagen) according to the manufacturer's instructions. Total RNA was reverse-transcribed into the first-strand cDNA using the Superscript First-Strand Synthesis Kit (Invitrogen). cDNA transcripts were quantified by Rotor Gene Real-Time PCR System (Qiagen) using SYBR Green (Biorad). Results of mRNA levels were normalized to 18S rRNA levels and reported as fold-change over control. The primer information is listed in Supplementary Table 3.
Statistical analysis. The numbers of independent experiments are specified in the relevant figure legends. Data are expressed as mean ± standard error of the mean (SEM). Statistical analysis was performed with Prism 7.0 software (GraphPad). Normal distribution of the data was assessed using Shapiro-Wilk normality test. Statistical comparisons between groups were conducted by unpaired Student's t-test or one-way ANOVA followed by a Newman-Keuls comparison test for normal distribution. Otherwise, Mann-Whitney test or Kruskal-Wallis test followed by Dunn's multiple comparison test were applied. (No assumption was made that variance was similar between the groups being statistically compared). The value of p < 0.05 was considered to be significant. No statistical method was used to predetermine samples size. There was a 10-15% surgical failure rate with the mouse TAC surgery, and it was pre-established that these mice would be excluded from the analysis. Animals within each cohort were randomly assigned to various groups based upon experimental design. Investigators were blinded for experiments including AAV injection, TAC surgery and echocardiography. Otherwise, the experiments were not randomized and investigators were not blinded to allocation during experiments and outcome assessments.
Data availability. Microarray data displayed in Fig. 1a, Supplementary Fig.1a