Fasting regulates EGR1 and protects from glucose- and dexamethasone-dependent sensitization to chemotherapy

Fasting reduces glucose levels and protects mice against chemotoxicity, yet drugs that promote hyperglycemia are widely used in cancer treatment. Here, we show that dexamethasone (Dexa) and rapamycin (Rapa), commonly administered to cancer patients, elevate glucose and sensitize cardiomyocytes and mice to the cancer drug doxorubicin (DXR). Such toxicity can be reversed by reducing circulating glucose levels by fasting or insulin. Furthermore, glucose injections alone reversed the fasting-dependent protection against DXR in mice, indicating that elevated glucose mediates, at least in part, the sensitizing effects of rapamycin and dexamethasone. In yeast, glucose activates protein kinase A (PKA) to accelerate aging by inhibiting transcription factors Msn2/4. Here, we show that fasting or glucose restriction (GR) regulate PKA and AMP-activated protein kinase (AMPK) to protect against DXR in part by activating the mammalian Msn2/4 ortholog early growth response protein 1 (EGR1). Increased expression of the EGR1-regulated cardioprotective peptides atrial natriuretic peptide (ANP) and B-type natriuretic peptide (BNP) in heart tissue may also contribute to DXR resistance. Our findings suggest the existence of a glucose–PKA pathway that inactivates conserved zinc finger stress-resistance transcription factors to sensitize cells to toxins conserved from yeast to mammals. Our findings also describe a toxic role for drugs widely used in cancer treatment that promote hyperglycemia and identify dietary interventions that reverse these effects.


Introduction
Despite advances in cancer therapy, the standard of care predominantly includes chemotherapy, radiotherapy, or their combination. These treatments are associated with a multitude of side effects ranging from discomfort to the development of secondary tumors and severe toxicity to multiple systems [1][2][3][4]. Doxorubicin (DXR) is an antineoplastic drug widely used against a variety of human malignancies, including bladder, breast, leukemia, lung, liver, and ovarian cancer, but its clinical use is largely limited by dose-dependent cardiac toxicity [5,6]. To increase the efficacy of cancer treatment or to help with the management of the adverse effects of the therapy, other drugs such as dexamethasone (Dexa), aprepitant, and lorazepam are often used in combination with chemo-and radiotherapy [7,8]. The corticosteroid Dexa is mainly used as a palliative drug but can also be effective in the treatment of multiple myeloma, leukemia, and lymphoma [9][10][11][12]. However, treatment with Dexa can raise blood glucose levels [13,14], and several studies have proposed that it can promote toxicity in the brain [15,16].
Multiple studies have shown the antitumor and antimitotic effect of the mTOR (mammalian target of rapamycin) inhibitor rapamycin (Rapa) in different model organisms [17]. Multiple analogs of Rapa (temsirolimus and everolimus) have been approved by the FDA for the clinical treatment of subependymal giant cell astrocytoma (SEGA), advanced hormone receptor-positive HER2-negative breast cancer, primitive neuroectodermal tumors (PNETs) of pancreatic origin, and for advanced renal cell carcinoma (RCC) [18,19] (see also the National Institutes of Health-National Cancer Institute website). Despite its efficacy in reducing the progression of certain tumors, mTOR inhibitors also promote hyperglycemia, raising the possibility that, similarly to Dexa, they could also increase chemotherapy toxicity to normal tissues [20,21].
Fasting or short-term starvation (STS) and fasting-mimicking diets (FMDs) are effective in improving both the life span and health span of multiple species [22][23][24]. STS can also selectively protect normal (but not cancer) cells against toxins including chemotherapy, a phenomenon we named differential stress resistance (DSR), but the cellular mechanisms responsible for its protective effects are unclear [22,[25][26][27][28][29]. Competing interests: I have read the journal's policy and the authors of this manuscript have the following competing interests: Valter D. Longo is the founder of and has an ownership interest in L-Nutra whose food products are used in human studies of a fasting-mimicking diet. Longo's interest in L-Nutra was disclosed and managed per USC's conflicts of interest policies. USC has an ownership interest in L-Nutra, and the potential to receive royalty payments from L-Nutra. USC's financial interest in the company has been disclosed and managed under USC's institutional conflict of interest policies. We have previously shown that yeast Msn2/4 [27,30] are stress response transcription factors negatively regulated by glucose and protein kinase A (PKA) that mediate part of the fasting-dependent effects on stress resistance and life span extension [25,29]. Mammalian early growth response protein 1 (EGR1), which has high homology to yeast Msn2/4, is a key transcription factor involved in diverse cellular processes such as development, proliferation, DNA repair, apoptosis, and tumor suppression [31][32][33]. In mammals, Egr1 expression is also up-regulated by the master energy sensor AMP-activated protein kinase (AMPK) [9,34,35], which is downstream of and negatively regulated by PKA [36][37][38]. However, the role of mammalian EGR1 in cellular protection and its link to glucose and PKA/AMPK has not been investigated.
Here, we tested the hypothesis that fasting reverses the hyperglycemic effect of Dexa and Rapa to protect against DXR-induced cardiotoxicity [28, 39] by regulating a novel mammalian glucose-PKA-AMPK-EGR1 pathway, analogous to the glucose-PKA-AMPK-Msn2/4 pathway, which regulates aging and stress resistance in yeast.

Short-term starvation protects from DXR-induced cardiotoxicity
We have previously shown that STS effectively reduces the mortality of mice exposed to otherwise lethal doses of DXR (DSR) [28,39]. To investigate whether the protective effects of STS involve reduction of DXR-induced cardiotoxicity, C57BL/6 mice were administered multiple doses of DXR ± STS (48 h; S1A Fig). The dose and schedule of treatment were chosen to provide prolonged exposure to DXR through multiple cycles to model the clinical doses and schedules.
Compared to the control group, DXR-treated mice displayed a reduction in cardiac function and/or functional markers, including diastolic (p-value < 0.0001) and systolic (pvalue < 0.0001) left ventricle (LV) volume and wall thickness (p-value < 0.05) (Fig 1A-1D). STS prior to DXR injection improved the overall response to high-dose chemotherapy, prevented the loss of LV diastolic and/or systolic volumes (p-value < 0.001 and < 0.0001, respectively) (Fig 1A and 1B), and ameliorated LV wall thinning (Fig 1C and 1D).

Hyperglycemia induced by Rapa and Dexa increases cardiotoxicity
STS down-regulates signal transduction proteins, including mTOR and PKA, and reduces circulating IGF-1 and glucose levels [25,28,39]. Because glucose is a key sensitizer of yeast cells, we tested the role of glucose in animals undergoing DXR treatment. Hyperglycemia was induced in C57BL/6 mice by daily injections with Dexa or Rapa-two agents that cause hyperglycemia and insulin resistance [40,41] and are FDA-approved for use in cancer therapy [42, 43]-for 2 wk before DXR treatment (Fig 1E-1J and S1B Fig). Both Dexa and Rapa treatment enhanced the toxicity of DXR in mice (p-value < 0.05) (Fig 1G and 1J). To examine whether the reduced survival was caused by elevated circulating glucose levels, mice pretreated with Dexa and Rapa were also administered with either insulin or underwent STS in order to reduce blood glucose levels ( Fig 1K). STS or insulin administration following Rapa and Dexa prevented hyperglycemia and protected mice from DXR toxicity (Fig 1G, 1J and 1K), suggesting that the hyperglycemia induced by Dexa or Rapa is responsible for the sensitization of mice to DXR-induced chemotoxicity. Notably, STS was more potent in preventing this hyperglycemia-induced sensitization than insulin treatment alone (p-value < 0.0001), indicating that fasting has a broader effect that is partly independent of glucose. Further, restoring hyperglycemia during STS by direct glucose injections (2g/kg IP every 4 h during the 24 h prior to DXR administration, determined based on its clearance rate [44]), restored euglycemia and fully reversed the STS-dependent protection from DXR in mice (Fig 1L and 1M). Because both high levels of insulin (injections) and low levels of insulin (caused by STS) were associated with protection against DXR, these results indicate that reduced glucose and not altered insulin levels are responsible for STS-induced protection.
Glucose restriction protects cardiomyocytes through the PKA/EGR1 pathway The reduction of PKA signaling during fasting promotes anti-aging and stress resistance in yeast and mammalian cells [23,29,45]. In yeast, glucose activates PKA, which in turn sensitizes cells in part by down-regulating stress resistance transcription factors Msn2/4. We therefore tested the involvement of these genes in mediating the protective effect of STS and glucose ). Experiments were repeated twice. Data represent mean ± standard error of the mean (SEM). ANOVA (and Tukey post-analysis) was performed in A, B, C, K, and L; Student's t test was performed in F and I; and log-rank test was performed on E, G, H, J, and M. p-values < 0.05 are considered significant (p-value < 0.05, 0.01, and 0.001 are indicated as *, *, and ***, respectively). Underlying data and method of statistical analysis are provided in S1 Data. restriction (GR) (Fig 2A) [29]. GR protected yeast from oxidative stress, and the deletion of Msn2/4 reduced such resistance by 10-fold ( Fig 2B). Consistent with the yeast data, in vitro against PKA (or a nonspecific control) at day 1, treated with GR or normal medium at day 2, and treated with DXR (1 μM) at day 3. Cytotoxicity is reported. (F) PKA inhibition triggers Egr1 expression. Primary rat cardiomyocytes were treated with PKA inhibitors, H89, or PKI for 1 h. mRNA expression level of Egr1 was measured. (G) Primary and H9c2 rat cardiomyocytes were incubated with GR medium for the indicated time and immunoblotted with an anti-EGR1 antibody. (H) Primary rat cardiomyocytes were incubated with serumfree medium for 1 h and immunoblotted with an anti-EGR1 antibody. (I, J) Egr1 +/+ and Egr1 -/mouse embryonic fibroblasts (MEFs) transiently transfected with EGR1-secreted alkaline phosphatase (SEAP) reporter were incubated with (I) GR or (J) H89. SEAP activity was measured and quantified by using a chemiluminescent assay. (K) H9c2 cardiomyocytes transfected with a Flag-Egr1 expression vector were treated with DXR (1 μM) for 24 h. Cytotoxicity was measured by LDH release. Experiments were repeated three times, and the average for each technical repeat is reported in each graph (n = 3). The significance of the differences between experimental groups was determined by using one-way ANOVA (Tukey postanalysis test). Comparisons between groups were performed with Student's t test. Data represent the mean ± SEM. p-values < 0.05 were considered significant (p-value < 0.05, 0.01, and 0.001 are indicated as *, *, and ***, respectively). Underlying data and method of statistical analysis are provided in S1 Data. STS conditions with reduced glucose (0.5 g/L) and growth factors (1% fetal bovine serum [FBS]) reduced PKA activity (p-value < 0.05) ( Fig 2C) and protected cardiomyocytes against DXR toxicity (p-value < 0.001) ( Fig 2D). PKA knockdown was achieved by using short interfering RNA (siRNA) against Pkaα (also called Prkaca), the main catalytic subunit of PKA [46]. PKA knockdown in H9C2 rat cardiomyocytes significantly protected from DXR toxicity, which was further improved by glucose restriction (GR) (p-value < 0.0001) (Fig 2E). GR also increased levels of Egr1 transcript (S2A Fig), as did treatment with the selective PKA inhibitors H-89 and PKI 14-22 amide (PKI) (Fig 2F). GR, but not serum restriction (FBS), rapidly induced Egr1 expression in cardiomyocytes, indicating that Egr1 expression is regulated by glucose levels and is not strongly affected by growth and other factors contained in the serum (Fig 2G and 2H).
In order to test the action of EGR1 in response to starvation in a different cell type, we examined the effects of GR and PKA inhibition using wild-type and Egr1-null mouse embryonic fibroblasts (MEFs). Similar to our observation in primary cardiomyocytes, EGR1 protein levels in wild-type MEFs were rapidly induced by GR (S2E Fig). To examine whether starvation regulates the transcriptional activity of Egr1, we generated an Egr1-dependent secreted alkaline phosphatase (SEAP) reporter construct containing two copies of an Egr1 binding element. Wild-type and Egr1-null MEFs were transfected with Egr1-SEAP constructs and treated with GR medium. A 48-h GR induced and accumulated Egr1-dependent SEAP expression (p-value < 0.001; Fig 2I). Similarly, the inhibition of PKA using H-89 increased Egr1 activity (p-value < 0.05; Fig 2J). Conversely, the over-expression of Egr1 alone protected cardiomyocytes from DXR ( Fig 2K). Interestingly, PKA inhibition with H-89 was five times less potent than GR in activating Egr1, suggesting that GR can also promote Egr1-dependent transcription activity through pathways independent of PKA. These results indicate that STS and GR protect cells against DXR stress by activating Egr1 by both PKA-dependent and independent mechanisms.

Fasting induces the expression of cardioprotective peptide hormones ANP and BNP in an EGR1-dependent manner
The atrial natriuretic peptide (ANP) and B-type natriuretic peptide (BNP) are members of the family of cardiac peptide hormones that are induced by cardiac stress and heart disease to protect cardiomyocytes [47]. To determine whether STS can regulate the expression levels of ANP and BNP genes, we analyzed heart tissue collected from mice undergoing ad lib feeding or a STS regimen. ANP (p-value < 0.001) and BNP (p-value < 0.001) expression was strongly induced in the heart of mice undergoing STS (Fig 3A and 3B). Circulating ANP levels were not significantly altered, suggesting that local changes in the heart environment (autocrine and/or paracrine) rather than systemic changes (endocrine) could induce cellular protection ( Fig 3C). As previously reported, a low-protein and low-carbohydrate FMD can be as effective as fasting in inducing DSR and in extending life span while maximizing nourishment and minimizing the risk of major adverse effects of complete food restriction [22,24]. We tested the effect of this FMD on the in vivo regulation of ANP and BNP. Three cycles of the FMD regimen significantly induced the expression of ANP in the heart to the same extent observed in mice undergoing STS (p-value < 0.001) (Fig 3D). BNP was also induced by the FMD, although not at the same levels observed in animals undergoing complete fasting (p-value < 0.0001) (Fig 3E). These results indicate that both fasting and a fasting-mimicking diet positively control the expression of the cardioprotective ANP and BNP peptides in heart tissue.
We next tested whether the regulation of the cardiac peptides was mediated by EGR1. The overexpression of Egr1 in cardiomyocytes induced the mRNA levels of the ANP and BNP genes in a dose-dependent manner (p-value < 0.0001) (Fig 3F and 3G). The role of EGR1 in the regulation of ANP and/or BNP was tested by treating cardiomyocytes with Egr1 siRNA. The siRNA-mediated depletion of Egr1 markedly reduced ANP and/or BNP expression levels (p-value < 0.0001; Fig 3H and 3I). Importantly, sequence analysis of the ANP and BNP gene promoters revealed that both contain several putative Egr1 binding domains (Fig 3J). Pretreatment of primary cardiomyocytes with either ANP or BNP (0.1 nM) reduced DXR-induced cytotoxicity (p-value < 0.001 for ANP and p-value < 0.0001 for BNP) (Fig 3K and 3L), indicating that the STS-induced cardioprotection is mediated in part by the EGR1-dependent transcriptional regulation of ANP and BNP.
Glucose restriction protects cardiomyocytes through AMPK/EGR1 activation AMPK is a master cellular energy sensor in most eukaryotic cells and, when activated by starvation, increases cellular protection [48][49][50] and induces Egr1 expression [9,34,35]. Others  6), and data are represented as mean ± SEM. The significance of the differences between experimental groups was determined by using one-way ANOVA (Tukey post-analysis test). Comparisons between groups were performed with Student's t test. p-values < 0.05 were considered significant (pvalue < 0.05, 0.01, and 0.001 are indicated as *, *, and ***, respectively). Underlying data and method of statistical analysis are provided in S1 Data. have also shown that PKA can negatively regulate AMPK [36][37][38]. Here, we show that GR rapidly activates AMPK-as indicated by measuring the level of phosphorylation at Thr172-in rat cardiomyocytes (Fig 4A). To test whether EGR1 activation in response to GR is also mediated by AMPK, we treated cardiomyocytes with the AMPK-specific inhibitor compound C (CC) under normal and GR conditions. EGR1 induction by GR was abolished by CC treatment at the transcript ( Fig 4B) and protein level (Fig 4C). Because CC is not entirely AMPKspecific [51], we also tested the effect of AMPK and Egr1 inhibition using siRNA (S4 Fig). Knockdown of AMPKα1/2 in rat cardiomyocytes significantly inhibited GR-induced EGR1 expression (Fig 4D). Further, siRNA-mediated knockdown of AMPKα1/2 and Egr1 reversed GR-mediated cardiomyocyte protection against DXR, determined by LDH release (Fig 4E) and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction (Fig 4F). The requirement for AMPKα1/2 for the GR-induced protection from DXR in cardiomyocytes was confirmed by trypan blue exclusion (Fig 4G). Conversely, the activation of AMPK by treatment with the pro-longevity diabetes drug metformin (Met) (50 mg/kg/day for 2 wk, intraperitoneal [IP]) significantly reduced blood glucose levels (Fig 4H and S2C Fig), induced Egr1 expression in primary rat cardiomyocytes (Fig 4I and S2B Fig), and caused a nonsignificant trend for the protection of mice against DXR toxicity (Fig 4J). Such STS and/or Metdependent protection against DXR was, in part, reversed by Egr1 knockdown in rat cardiomyocytes (Fig 4K). Met and its analogue phenformin (Phen) both increased We also found that Egr1 is highly expressed at both the RNA and protein level in the heart of FMD-treated animals (p-value < 0.0001) (S3A and S3B Fig). FMD regimens also increased the expression of antioxidant enzymes, including catalase and Mn/CuZnSOD and FOXO1 and FOXO3 (p-value < 0.0001), which are key regulators of stress resistance and adaptive metabolic responses (S3C- S3G Fig) [52]. Because Met and nutrient restriction can induce AMPK activity but also to inhibit PKA activity [23,53], we examined the combined effect of AMPK activation and PKA inhibition on DXR-induced cardiotoxicity. PKA knockdown alone protected cardiomyocytes from DXR toxicity but did not add to the protection provided by Met (Fig 4L), in accord with previous studies suggesting AMPK acts downstream of and is negatively regulated by PKA [36][37][38]. Further investigations will be necessary to determine if the cardioprotective effects of metformin against DXR in vivo are mediated by reduced blood glucose levels, its direct action on the heart, or a combination of both.
These findings suggest that GR-induced stress-resistance is mediated by a conserved pathway that involves PKA, AMPK, and transcription factors Msn2/4 (yeast) and Egr1 (mice) (Fig 4M).

Discussion
DXR is one of the most effective agents for the treatment of human breast, lung, thyroid, and ovarian carcinomas, but its therapeutic applications have been constrained by its dose-limiting side effects and, particularly, its cardiac toxicity [50,37,59]. Here, we provide evidence for the existence of a dexamethasone and glucose-PKA-dependent cellular sensitization conserved from yeast to mammalian cells, which is mediated in part by down-regulation of zinc finger transcription factor Egr1 and its downstream effectors, including cardioprotective peptides ANP and BNP. This study indicates that dexamethasone and chemotherapy, which are widely administered in combination, enhance toxicity in mice, raising the possibility that patients receiving both could also be at risk. In agreement with other studies [36-38], PKA may negatively regulate AMPK. The consequent induction of Egr1 or Msn2/4 triggers the expression of antioxidants, stress response genes (in both yeast and mice), and cardiac peptide hormones ANP/BNP, eventually enhancing cardioprotection (in mice). Technical replicates for the in vitro experiments are reported in B, E-G, K, and L (n = 3 to 4), and data are represented as mean ± SEM. The significance of the differences between experimental groups was determined by using one-way ANOVA (Tukey post-analysis test). Comparisons between groups were performed with Student's t test. p-value < 0.05 were considered significant (p-value < 0.05, 0.01, and 0.001 are indicated as *, *, and ***, respectively). Underlying data and method of statistical analysis are provided in S1 Data. Fasting prevents toxicity of dexamethasone and hyperglycemia The identification of the key players in STS-dependent cellular protection has been particularly challenging given the effect of fasting on a multitude of molecular and physiological changes, including the reduction of IGF-I and circulating glucose levels and the down-regulation of many signal transduction proteins, including mTOR and S6 kinase [28,39]. Moreover, glucose homeostasis is regulated by a complex array of signals and cellular states, making it difficult to extrapolate the role of single components in cellular protection. Dexa and Rapa are given to cancer patients to reduce certain side effects of chemotherapy, enhance the killing of cancer cells, or both [54,55]. However, our results indicate that the hyperglycemia induced by Dexa and Rapa is, at least in part, responsible for the increased mortality following high-dose DXR. These results are in agreement with the finding that high glucose levels in combination with chemotherapy in patients is associated with an increased risk of developing complicated infections and with a significant increase in overall mortality when compared with patients with normal blood glucose [56]. This is particularly concerning because Dexa is often used to overcome only minor side effects of chemotherapy such as nausea. Notably, a simple intervention such as STS, as well as insulin, prevented the hyperglycemia induced by Dexa or Rapa and reversed the sensitization effects in mice.
At the cellular level, GR and Met not only activate AMPK but also inhibit PKA, leading to EGR1-mediated protection of cardiomyocytes, possibly through the expression of antioxidant and stress resistance peptides. Met has been shown to preserve cardiac function through AMPK activation [57], and in fact, AMPK and PKA appear to have opposite effects on cardioprotection. Further investigations would be necessary to determine if Met is cardioprotective in vivo and if its effects against DXR are mediated by reduced blood glucose levels or direct actions on the heart or, possibly, a combination of both. MnSOD has been shown to have Egr1 binding sites in its 5 0 promoter region [58], in agreement with our finding here that EGR1 regulates MnSOD in response to GR, analogously to the effect of the ortholog transcription factors Msn2/4 in the regulation of MnSOD in yeast [59]. Notably, this study provides the first evidence for a pro-aging and/or cellular sensitization glucose-PKA pathway that inhibits zinc finger stress resistance transcription factors and downstream protective enzymes, including superoxide dismutase, conserved from yeast to mammals [23].
Interestingly, the conserved anti-aging forkhead stress resistance transcription factor FOXO, which has been shown to bind the MnSOD promoter and to regulate its expression [60], is not involved in the regulation of MnSOD during GR conditions (S5A- S5F Fig). Thus, the expression of MnSOD is transcriptionally up-regulated in response to GR and PKA inhibition in an EGR1-dependent but FOXO-independent manner.
Oxidative stress and the production of reactive oxygen species (ROS) during chemotherapy treatment can damage normal cells and are implicated in the toxicity of DXR [61]. Measurements of basal mitochondrial oxygen consumption rate (OCR) (as a measure of mitochondrial respiration), extracellular acidification rate (ECAR) (as a measure of glycolysis), and cellular ATP levels in Egr1 +/+ and Egr1 -/-MEFs revealed that EGR1 is preferentially involved in the regulation of mitochondrial respiration without affecting ATP production, possibly suggesting a higher proton conductance across the mitochondrial inner membrane for Egr1 +/+ cells (S5G-S5I Fig). Notably, in addition to activating AMPK, Met is known to inhibit respiration by targeting mitochondrial complex I [53]. These results indicate that EGR1 may contribute to improving mitochondrial efficiency, which may lower intracellular levels of ROS and protect against oxidative stress [62], although further studies are necessary to confirm this hypothesis.
ANP and B-type natriuretic peptide (BNP) are hormones produced in the heart that can regulate the intravascular blood volume. A rise of circulating ANP and BNP has been shown to compensate for heart failure because of their natriuretic, diuretic, and vasodilating actions.
It has also been shown that natriuretic peptides have beneficial effects as therapeutic drugs in heart failure and acute myocardial infarction [63]. ANP and BNP can also act on the heart in an autocrine and/or paracrine fashion, as the heart expresses corresponding receptors [64,65]. Local natriuretic peptides can also prevent cardiac hypertrophy [66,67], which is a known consequence of DXR cardiotoxicity [68,69]. Although fasting did not alter the circulating levels of ANP, it induced the expression of ANP and BNP in the heart. In cardiomyocytes, the expression of ANP and BNP was EGR1-mediated. Further animal studies are needed to confirm the role of EGR-1 and of ANP and/or BNP expression in mediating the cardioprotective effects of fasting against DXR.
Fasting induces a wide range of physiological responses, including nutrient and growth factor alterations [70]. Unlike yeast, mammalian systems utilize hormones, such as insulin, to regulate cellular functions in response to nutrient availability, raising the possibility that some of the effects associated with high glucose levels identified here could be caused by high levels of insulin and not hyperglycemia. However, both conditions in which glucose was low but insulin was either high (injection) or low (fasting) were protective, suggesting that it is the hypoglycemia and not the effects on insulin level that mediate the resistance to chemotherapy treatment. However, factors other than high glucose are likely to also contribute to the increased toxicity of chemotherapy observed after rapamycin or dexamethasone treatment. Further studies to carefully test further the effects of insulin, IGF-1, and other factors on stress resistance are necessary. In fact, our previous results indicate that reduction in IGF-1 levels are also important for protections against chemotherapy toxicity [28,39].
In summary, our data show that the use of Dexa and Rapa, drugs widely used in cancer treatment which promote hyperglycemia, are associated with sensitization of mice and cardiomyocytes to the cytotoxic effects of DXR via the PKA-AMPK-EGR1 pathway. Together with human studies proposing that hyperglycemia may shorten the duration of remission for certain cancers, this study indicates that, unless the effects shown here are ruled out in humans, the clinical use of Dexa and other drugs promoting hyperglycemia in combination with chemotherapy drugs should be discontinued, particularly when it is only administered to prevent relatively minor side effects.

Ethics statement
All of the experiments were approved by University of Southern California's Institutional Animal Care and Use Committee (IACUC) before the experiments were started. All animals were kept in a positively pressurized, HEPA air-filtered animal room with 12-h light/dark cycles. Animals were anesthetized with 2% isoflurane during procedures.

Starvation, fasting, and FMD treatment
For STS in vivo, mice were fasted for 24 to 72 h by complete deprivation of food but with free access to water. In order to avoid dehydration, the mice were fed a low-calorie hydrogel that encouraged water consumption, and body weight was measured immediately before, during, and after fasting. For a cell culture model, STS was performed by glucose and serum restriction. The culture media were supplemented with 0.5 g/L or 2.0 g/L glucose to match blood glucose levels in starved and normally fed mice, respectively [71]. FBS was supplemented at 1% for starvation conditions as compared to the normal 10%. In FMD conditions, mice were provided with less than 50% of the normal calorie intake for 4 d bimonthly, followed by~60%-100% of the normal caloric intake.

Animal studies
C57BL6 female mice were purchased from Charles River Laboratories International, Inc. or The Jackson Laboratory. Mice were maintained in a 12-h light/dark cycle at constant temperature and humidity. For stress resistance experiments, 12-wk-old female C57BL/6 mice were divided in the following experimental groups; ad lib (ad libitum feeding), STS, DXR, STS + DXR, Rapa, and STS + Rapa. It also included the following groups: Dexa, STS + Dexa, Insulin (Ins), Ins + Rapa, and Ins + Dexa.
Doxorubicin HCL was injected intravenously (IV). For the induction of heart toxicity, the animals were injected every 2 wk for a total of four injections with 8 mg/kg/mouse of DXR. For the stress resistance experiment, the mice were injected once with 24 mg/kg DXR. Rapa (1.5 mg/kg), Dexa (1 mg/kg), or Met (50 mg/kg) was administrated daily by intraperitoneal (IP) injection. Human normal insulin was administrated by IP injection twice a day for 2 consecutive days for a total of five injections.
The administration of Rapa and Dexa was performed for a period of 14 d after DXR administration. Insulin was administrated twice a day for a total of 48 h preceding the administration of DXR. Mice in the STS + DXR group were fed a very low-calorie and no-protein FMD for 48 h prior to the injection of DXR. Following DXR injection, the survival was recorded daily.
In order to induce hyperglycemia, glucose (2g/kg) was administrated intraperitoneally during the 24 h prior to DXR injection (Fig 1L and 1M) for a total of six injections. The last glucose injection was performed 1 h prior to DXR injection. Control animals were injected with 150 μl of sterile saline.

Echocardiography
Two-dimensional echocardiography of the left ventricle (LV) on mice was performed using Vevo 770 Ultrasound system (Visualsonics). Mice were anesthetized with isoflurane and placed on a heating table in a supine position with four electrocardiography leads. Warmed ultrasound transmission gel (Parker Laboratories, cat. no. 01-20) was applied to the thorax surface to optimize the visibility of the chambers. The short-axis view of the LV was recorded to evaluate LV structure and diastolic function.

Histopathological studies
Animals were anesthetized with 2% isoflurane, and the heart was collected and perfused with cold PBS. The LVs of heart tissues were extracted from mice and fixed in 4% neutral buffered formaldehyde. Sections of 5 um thickness were stained with hematoxylin (Sigma, cat. no. MHS16) for histological examination. In a blinded fashion, histomorphological evaluation of all the heart sections of mice was performed by a pathologist who was unaware of the groups.

Cell culture
Cardiomyocytes were isolated from 1-3-d-old neonatal Sprague Dawley rats, which were purchased from Jackson. Tissue dissection was performed using Pierce Primary Cardiomyocyte Isolation Kit (Thermo Fisher Scientific, cat. no. 88281). H9c2 rat fetal cardiomyoblasts (ATCC, cat. no. CRL-1446) were grown in DMEM culture medium with high glucose (4.5 g/L; glucose concentration equivalent to~25 mM), which was supplemented with 10% FBS. For glucose restriction studies, the cells were treated with 0.5 g/L glucose and pyruvate-free DMEM medium (Thermo Fisher Scientific, cat. no. A1443001), which was supplemented with 1% FBS for the indicated time.

SEAP activity assay
Wild-type and Egr1-null MEFs were transfected with a reporter plasmid containing two repeats of EGR1-binding consensus sequence GCG GGG GCG driving the expression of SEAP. After 24 h, cells were placed in glucose restriction media or treated with the PKA inhibitor H-89 (Sigma, cat. no. B1427) for the indicated time. The media was collected and tested for SEAP activity using Great EscAPe SEAP Chemiluminescent Kit (Clontech, cat. no. 631736).

Blood glucose measurements
Briefly, the tip of the tail was cut and 5 μl of blood were drawn directly in the glucose strip for the measurement (Precision Xtra, Abbott Diabetes Care Inc.).

Bioenergetic measurements
Cells were seeded onto an XF96 cell culture microplate (Seahorse Bioscience) at 2-3 × 10 4 cells/well. Metabolic rates were measured using an XF96 Extracellular Flux Analyzer (Seahorse Bioscience). XF Cell Mito Stress and XF Glycolysis Stress Test kits were used to measure the key parameters of mitochondrial functions and cellular glycolysis, respectively, according to the manufacturer's instructions (Seahorse Bioscience).

LDH release assay
Primary cardiomyocytes and H9C2 cells were seeded in 96-well culture plates (1 × 106 cells/ well) and transfected with siRNA of Egr1, AMPKα 1/2, and nontargeting control or treated with ANP or BNP peptide. For siRNA-transfected cells, at 24 h posttransfection with siRNA, the cells were placed in glucose restriction media, and at 48 h posttransfection, the cells were treated with 1 uM of doxorubicin (Sigma, cat. no. D1515) for 24 h. For the pretreatment of peptides, ANP or BNP peptide was treated for 24 h. To quantify the release of LDH from the cells, the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, cat. no. G1780) was used according to the manufacturer's instruction. The percent release was normalized to detergent lysis (2% Triton X-100) treated cells for peptide pretreatment and non-DXR treated cells.

MTT assay
H9C2 cells were seeded with culture medium in 96-well culture plates and incubated at 37˚C for 24 h before siRNA transfection. As described above, at 24 h after treatments with doxorubicin, the siRNA-transfected cells were incubated with culture medium containing 0.4 mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Sigma-Aldrich, cat. no. M5655) at 37˚C for 4 h. The reduced formazan crystals were dissolved in SDS and the absorbance was detected at 570 nm against 660 nm.
Trypan blue exclusion assay H9C2 were seeded in 12-well culture plates (0.1 × 10 6 cells/well) and transfected with siRNA of AMPKα 1/2 or nontargeting control for 24 h. At 24 h posttransfection, glucose restriction media was replaced, and at 48 h posttransfection, doxorubicin was treated for 24 h as described above. Cells were trypsinized, and cell viability was determined using a trypan blue dye (0.4% trypan blue solution [Thermo Fisher Scientific, cat. no. 15250061]). The number of viable cells was subsequently counted under light microscope.

PKA kinase activity
PKA kinase activity was assessed in vitro by colorimetric assay (Enzo Biochem, Inc. cat no. ADI-EKS-390A).

Drug administration
Rapamycin. Rapamycin (LC Laboratories) stock solution was prepared by dissolving the compound in DMSO to a final concentration of 50 mg/ml. Single aliquots of the stock solution were stored at -20˚C. For in vivo use, rapamycin 1.5 mg/kg was administrated daily by IP injection. Before injection, rapamycin working solution was freshly made by dissolving the stock solution in 5% PEG, 5% TWEEN 80, and 4% ethanol in sterile saline. For in vitro use, rapamycin stock solution was serially diluted in DMEM cell culture medium (Dulbecco's Modified Eagle's Medium-Life Technologies) at a final concentration of 5 uM for ACN, HTLA-230, NXS2, and SH-SY5Y cells and to 10 uM for 4T1 and MEF cells.
Dexamethasone. Stock solution of dexamethasone (Sigma-Aldrich Co. LLC) was prepared by dissolving the compound in 100% ethanol to a final concentration of 25 mg/ml. The working solution was freshly prepared every day by dissolving the stock solution in saline and administrated intraperitoneally (1 mg/kg).
Metformin. Metformin was purchased from Sigma-Aldrich Co LCC. The powder was dissolved in saline to make a stock solution that was stored at -20˚C. The working solution was freshly made every day, and each animal was injected intraperitoneally with metformin 50 mg/kg (roughly 100 μl per animal) Insulin. Human normal insulin was administrated twice a day for two consecutive days for a total of five injections. Stock solution of insulin was diluted in sterile saline, and 1.5 U/kg per mouse were administrated by IP injection.

Yeast Study
Wild-type and Δmsn2/4 double mutant yeast stains (all derivatives of DBY746 MATα leu2-3, 112, his3Δ, trp1-289, ura3-52, GAL + ) were streaked out from frozen stock onto YPD plates and incubated at 30˚C for 2 d. Next, 3-5 colonies were inoculated in 2 ml of liquid SDC and incubated overnight. 100 μl of the overnight culture was added to four 50-ml flasks containing 10 ml of fresh SDC with different glucose concentrations (i.e., 0.5%, 2%, 5%, or 10%) and incubated at 30˚C for 4 d. Following the incubation period, 1 OD 600 aliquot from each flask was treated with 200 μM menadione, 100 mM H 2 O 2 , or buffer only in 0.1 M KPO 4 buffer and incubated at 30˚C for 1 h. Posttreatment, a 10-fold serial dilution was spotted on YPD plates and incubated at 30˚C for 4 d.

Statistics
ImageJ software was used for the quantification of the signal and GraphPad Prism v5.0c was used for the graphic representation and statistical analysis of the data. The significance of the differences between experimental groups was determined by using Analysis Of Variance (ANOVA) analysis in GraphPad Prism v5.0c. Tukey test was used in the post-analysis, and the differences were considered significant if the p-value was 0.05. Comparisons between groups were done with Student's t test using GraphPad Prism v.5.0c. All the statistical analyses were two-sided, and p-values 0.05 were considered significant. MnSOD-mt-luc, or (F) MnSOD-luc, and treated with GR medium. Luciferase activity was measured 18 h after treatment. (G-I) EGR1 regulates energy metabolism. (G) Basal levels of OCR to ECAR ratios, (H) ATP production, and (I) proton conductance were measured in Egr1 +/+ and Egr1 -/-MEFs by the XF96 extracellular flux analyzer. P-value <0.05 were considered significant (p-value<0.05, 0.01 and 0.001 are indicated as Ã , Ã , and ÃÃÃ , respectively). Underlying data and method of statistical analysis are provided in S1 Data. (TIF) S1 Data. Underlying data and method of statistical analysis. (XLSX)