Animal procedures were approved by the National Institute of Diabetes and Digestive and Kidney Diseases Animal Care and Use Committee (ASP K085-MMB-21) and carried out in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Mice were housed in a specific pathogen-free environment under a controlled temperature (23°C) and a 12 h light/dark cycle with free access to food (NIH-31, 14% kcal/fat, 3.0 kcal/g, Teklad Diets) and water.

All the mice had a C57BL/6 background. EPO acts by binding to the cell surface EPO receptor (EPOR) that is expressed at the highest level on erythroid progenitor cells. Mice with targeted deletion of EPOR (Epor−/−) die of severe anemia during embryonic development (Wu et al., 1999; Yu et al., 2001). ΔEPORE mice that are EPOR−/− mice rescued by an erythroid-restricted EPOR transgene (GATA-1 locus hematopoietic regulatory domain-driving mouse EPOR cDNA) (Suzuki et al., 2002) were used as models for the lack of EPOR signaling in non-erythroid cells. ΔEPORE mice exhibit appropriate EPO-stimulated erythropoietic response (Supplementary Figure S1A) and have an obese phenotype with an increased body weight and body fat mass and develop glucose intolerance and insulin resistance (Figures 1A, B; Supplementary Figures S1B–D) (Teng et al., 2011b). For an obese phenotype, we follow previous observations for obese mice on a high-fat diet compared with standard chow that were reported to have a percentage body fat of 36.8% (±12.3%) vs. 15.8 (±5.8%), respectively (Kaiyala et al., 2010). Heterozygous neuronal NOS (nNOS)-knockout (nNOS+/−) mice (Jackson Laboratory, Bar Harbor, ME, United States) were mated with each other to produce nNOS−/−mice.

At 4 months of age, all mice were fed a high-fat diet (60 kcal% fat, 34.9% crude fat, 5.24 kcal/kg, Research Diets Inc.) for 3 weeks to match the ΔEPORE obese phenotype [high-fat diet feeding does not exacerbate heart failure (Brainard et al., 2013)] and treated with or without EPO (3,000 units/kg three times/week of recombinant human EPO, Epogen, Amgen, Thousand Oaks, CA, United States). Body composition was measured using a three-in-one EchoMRI (Echo Medical Systems, Houston, TX, United States). Hematocrit was measured manually before and every week after EPO administration.

Mice were fasted overnight and were then injected intraperitoneally with glucose (2.0 g kg⁻¹ body weight) or insulin (humulin R, 0.75 mU kg⁻¹). Blood glucose levels were measured before (0 min) and up to 120 min (0, 15, 30, 60, and 120 min) after the injection by withdrawal of 3 μL blood from the tail vein. The glucose level was determined using the Elite Glucometer (Bayer, Whippany, NJ, United States), following the procedure of Chen et al. (2005).

Transthoracic echocardiography was carried out using a high-frequency, high-resolution ultrasound system (Vevo2100, FUJIFILM VisualSonics Inc., Toronto, Canada) and a 40-MHz transducer probe (VisualSonics, MS-400). Mice were lightly anesthetized with 1%–2% isoflurane to achieve a target heart rate of >400 beats per minute and positioned over a heated platform with EKG, respiration, and body temperature monitoring. Two-dimensional images were obtained for multiple views or the heart, and M-mode images from the short axis mid-papillary view of the left ventricle (LV) were used for measurements of the systolic and diastolic posterior and anterior wall thicknesses, end systolic, and end diastolic internal LV chamber dimensions (LVIDs and LVIDd), and the LV functional values of fractional shortening (FS) and ejection fraction (%EF) were calculated within the systems analysis software program.

Color and PW Doppler ultrasonography was also performed to assess the pulmonary artery and aortic blood flow. Color flow profiles were obtained for the left ventricular outflow tract (LVOT) from a suprasternal view. For generating the pulmonary artery color flow profiles, the probe was moved slightly cranial and leftward from the parasternal long axis view until the pulmonary artery was observed crossing over the aorta. The color flow images served as a guide for placement of the PW Doppler sample volume to obtain peak aortic and pulmonary artery velocities.

The left bottom of the heart from mice was used for total RNA extraction, and 1 μg of total RNA was reverse-transcribed (Invitrogen, Carlsbad, CA, United States). Quantitative real-time RT-PCR analyses were carried out using gene-specific primers (Supplementary Table S1) and fluorescently labeled SYBR Green dye (Roche Applied Science, IN, United States) in a 7900 Sequence Detector (PE Applied Biosystems, Foster City, CA, United States). Relative mRNA quantification was calculated by the delta–delta Ct method, and Ct values were normalized with RPL13a as an internal control.

Heart tissues were homogenized and lysed in denaturing radioimmunoprecipitation (RIPA) assay buffer supplemented with protease and phosphatase inhibitors (Roche). Proteins were electrophoretically separated with 4%–20% Tris-glycine SDS/PAGE gels, transferred to nitrocellulose membranes using an XCell SureLock Mini-Cell system (Invitrogen), and visualized using protein-specific antibodies. The following antibodies were used: pERK (Cell Signaling Technology; #5726; 1:1000, Danvers, MA, United States), ERK (Cell Signaling Technology; #4695; 1:1000), Nox4 (Novus Biologicals; NB110-58849; 1:1000, Littleton, CO, United States), iNOS (Abcam; Ab3523; 1:1000, Boston, MA), pAkt (Cell Signaling Technology; #9271; 1:1000), Akt (Cell Signaling Technology; #9272; 1:1000), and GAPDH (Cell Signaling Technology; #97166; 1:1000). Quantitative analysis was performed by measuring the integrated density with an NIH ImageJ system and normalized with GAPDH.

The data are expressed as mean ± SEM. Student’s two-tailed non-paired t-test was used, and p values of <0.05 were regarded statistically significant. One-way analysis of variance (ANOVA) adjusted by Fisher LSD was used for multiple comparisons.

All mice at 4 months of age were fed high-fat diet with and without EPO treatment to match the ΔEPORE obese phenotype (Figures 1A, B). EPO treatment for 3 weeks showed comparable erythropoietic response and an increase in hematocrit in male and female WT and ΔEPORE mice (Supplementary Figure S1A). To determine the influence of endogenous EPO activity on heart function, we used echocardiography to assess the cardiac function in WT male and ΔEPORE male mice with EPOR restricted to erythroid tissue. ΔEPORE male mice showed a similar level of PA peak velocity compared to WT male mice (Figure 1C). Without EPO treatment, ΔEPORE male mice showed a decreased LVOT peak velocity, ejection fraction, and fractional shortening compared to WT male mice (Figures 1D–G), suggesting that the endogenous EPO response of non-hematopoietic cells in WT mice protects the cardiac function. ΔEPORE mice also showed an increased systolic volume and systolic diameter compared to WT male mice, suggesting that in WT mice, endogenous EPO in non-hematopoietic tissues might play a role in protecting against hypertrophic cardiomyopathy (Figures 1H, I).

The impact on the heart function of EPO administration to increase hematocrit was determined by echocardiography to assess the cardiac function with EPO or PBS treatment in WT male, WT female, and ΔEPORE male mice. ΔEPORE male mice showed a similar level of PA peak velocity compared to WT male mice (Figure 1C). Although EPO treatment increased hematocrit similarly between WT and ΔEPORE mice (Supplementary Figure S1A), EPO treatment reduced the PA peak velocity of blood flow in male and female WT mice but did not affect the PA peak velocity in ΔEPORE male mice (Figure 1C). EPO treatment reduced the LVOT peak velocity, ejection fraction, and fractional shortening in WT male and female mice but had no effect in ΔEPORE mice (Figures 1D–G), indicating that high-dose exogenous EPO can harm the cardiac capability, that this response is mediated via EPOR expression in non-erythroid tissue and independent of change in hematocrit, and that high-dose EPO cannot modulate the cardiac function in mice that lack EPOR in non-hematopoietic cells. Although the systolic volume and systolic diameter were increased in ΔEPORE male mice compared to WT male mice, EPO treatment did not change the systolic volume and diameter in all mouse groups (Figures 1H, I).

EPOR expression in the cardiovascular system including endothelial cells and cardiomyocytes provides for EPO response. The cardioprotective effect of single bolus EPO administration after myocardial infarction observed in rats has been attributed to neovascularization and EPO-stimulated cardiomyocyte vascular endothelial growth factor production (van der Meer et al., 2005; Westenbrink et al., 2010). Here, we found that EPO treatment over 3 weeks to increase hematocrit adversely affected heart function and that this EPO modulation of heart function in WT mice was mediated by EPOR expression beyond erythroid tissue since EPO treatment had little effect on heart function in ΔEPORE mice (Figure 1). We examined how endogenous EPO activity contributed to the regulation of heart failure-associated gene expression by comparing expression in WT and ΔEPORE male mice. Among the genes examined were Cdk8, for which increased transgenic expression in mice promotes heart failure (Hall et al., 2017), TGFβ that exhibits elevated levels in myocardial infarction (Liu et al., 2017), and Nox4 that is a major source of reactive oxygen species (Gray et al., 2019). ΔEPORE mice showed increased expression of two-fold or more of heart failure-associated genes Cdk8, TGFb2, Nox4, S100A, and SERCA2a (Figure 2A).

Among hypertrophic cardiomyopathy-related genes, high phenotypic expression of Myh7 affects the myocardial mechanical function (de Frutos et al., 2022), Nppa expression reactivated in the ventricles is a marker of heart disease (Horsthuis et al., 2008), and elevated BNP (Nppb) is a hallmark of heart failure (Sangaralingham et al., 2023). ΔEPORE mice also expressed more than two-fold increased expression of hypertrophic cardiomyopathy-related genes, Acta1, Myh7, Nppa, and Nppb (Figure 2B). Sarcomere dysfunction contributes significantly to heart failure and includes, for example, genetic mutations in myomesin 2 (Myom2) located in the M line, middle of the sarcomere, and provides structural integrity, myozenin 2 (Myoz2) located in the Z line and regulates calcium-dependent signaling pathways, and possibly Tcap also in the Z-line (Sequeira et al., 2014; Martin et al., 2021). ΔEPORE male mice also exhibited increased expression of sacromeric genes, Tcap, Myom2, and Myoz2 compared to WT male mice (Figure 2C). Upregulated expression of these genes was consistent with the cardiac dysfunction and hypertrophic cardiomyopathy phenotype in ΔEPORE male mice (Figure 1). The data suggest that endogenous EPO regulates cardiac dysfunction-related genes and sarcomeric genes that are increased with the loss of EPOR in non-erythroid cells including cardiac tissues in ΔEPORE mice (Figure 2).

High-dose EPO treatment increased EPOR in cardiac tissues in WT mice but not in ΔEPORE mice (Figure 2D). High-dose EPO also increased heart failure-associated genes such as Cdk8, Nox4, S100A, and SERCA2a, and hypertrophic cardiomyopathy-related genes such as Acta1, Myh7, Nppa, and Nppb in male WT mice but not in ΔEPORE male mice (Figures 2A, B). Examination by Western blotting showed Nox4 protein was increased in male ΔEPORE mice compared with male WT mice, and EPO treatment in WT male mice increased Nox4 protein, but EPO-treated ΔEPORE mice showed a trend of decreased Nox4 protein (Figure 2E, left). Western blotting demonstrated that changes in Nox4 protein corresponded to changes in gene expression. Extracellular signal-regulated kinase (ERK) contributes to cardiac function, can be stimulated by growth factors to promote proliferation or prevent apoptosis, or, conversely, contributes to cardiac hypertrophy and progression to heart failure (Gallo et al., 2019). We previously observed that in a mouse model of heart injury, acute EPO treatment increased ERK signaling and protected against ischemic-reperfusion injury in mice (Teng et al., 2011a). Here, we found that the cardiac dysfunction and hypertrophic cardiomyopathy phenotype in ΔEPORE male mice was accompanied by enhanced pERK compared to WT mice (Figure 2E, right). In addition, EPO treatment in WT male mice compromised the heart function and increased pERK, but EPO-treated ΔEPORE mice showed minimal changes in pERK (Figure 2E, right). This raises the possibility that pERK signaling plays a critical role for both endogenous and exogenous EPO responses in cardiac modulation.

EPO stimulation of NOS activity contributes importantly to EPO multi-organ response including erythroid, cardiovascular, and brain responses (Burger et al., 2009; Mihov et al., 2009; Chen et al., 2010). In mice, overexpression of eNOS attenuated congestive heart failure and improved survival and EPO protection against ischemia-reperfusion injury requires eNOS activity (Burger et al., 2009; Teng et al., 2011a). Neuronal NOS contributes to both EPO-stimulated erythropoiesis and EPO neuroprotective activity in mice (Chen et al., 2010; Lee et al., 2023). However, EPO-stimulated cardiac dysfunction observed here (Figure 1) maybe related to nNOS activity, as suggested by the protective effect of nNOS knockout in mouse models of diastolic dysfunction (Yoon et al., 2021). Therefore, to investigate if the EPO stimulated cardiac dysfunction is related to nNOS activity, we used nNOS−/− male and female mice. Metabolic and cardiac responses of male and female WT and nNOS−/− mice treated with EPO (3,000 units/kg three times/week) or PBS for 3 weeks on a high-fat diet were determined. Male nNOS−/− mice showed less body weight and body fat composition compared to male WT mice, but body weight and fat mass were not significantly different between female nNOS−/− and WT mice (Figures 3A, B; Supplementary Figures S2A, B). Mice treated with the NOS inhibitor (N-nitro-L-arginine methyl ester) were reported to be protected against body weight gain on a high-fat diet (Tsuchiya et al., 2007; Sansbury and Hill, 2014). Since targeted deletion of eNOS increases susceptibility to diet-induced obesity (Nakata et al., 2008) and mice with targeted deletion of iNOS on high-fat diet gain weight analogous to WT mice but without an increase in insulin resistance (Perreault and Marette, 2001), this suggests that the loss of nNOS may be protective against diet-induced obesity. The minimal change in body weight and fat mass with high-fat diet feeding in male and female nNOS−/− mice is consistent with the anti-obesity effect of NOS inhibition in mice (Sansbury and Hill, 2014) and potentially specific for loss of nNOS. EPO treatment decreased body weight and fat mass accumulation in male WT mice but not in female WT or male and female nNOS−/− mice (Figures 3A, B; Supplementary Figures S2A, B), indicating that unlike WT male mice, EPO does not contribute to regulation of fat mass in male or female nNOS−/− mice. EPO treatment improved glucose tolerance in all groups (Supplementary Figures S2C, E), suggesting that EPO-improved glucose tolerance is independent from EPO protection against body fat gain. EPO enhanced insulin tolerance in male and female WT and male nNOS−/− mice but did not significantly improve insulin tolerance in female nNOS−/− mice (Supplementary Figures S2D, F). EPO increased hematocrit in all groups, male and female WT and nNOS−/− mice, and hematocrit tended to be lower in nNOS−/− mice compared with WT mice, without or with EPO treatment (Supplementary Figures S2G, H) (Lee et al., 2023).

The cardiac function of nNOS−/− mice with or without EPO treatment was determined by echocardiography. Measurements in nNOS−/− mice showed normal cardiac function.

The PA peak velocity for nNOS−/− mice was similar to WT mice (Figure 3C) and not significantly different from ΔEPORE mice (Figure 1C; Supplementary Figure S3A). Assessment of the LVOT peak velocity, ejection fraction, and fractional shortening in nNOS−/− mice was comparable to WT mice (Figures 3D–G) in contrast to the reduced levels observed for ΔEPORE mice (Figures 1D–F; Supplementary Figures S3B–D). The systolic volume and systolic diameter for nNOS−/− mice were also analogous to WT mice (Figures 3H, I). EPO treatment reduced the PA peak velocity of flow in male and female WT and nNOS−/− mice (Figure 3C), in contrast to the PA peak velocity of flow for ΔEPORE mice that was not changed with EPO treatment (Figure 1C; Supplementary Figure S3A). This indicates that modulation of the PA peak velocity by EPO is independent of EPO-stimulated erythropoiesis and requires non-erythroid EPOR expression but does not require nNOS activity. Interestingly, EPO treatment did not change the LVOT peak velocity, ejection fraction, and fractional shortening in male and female nNOS−/− mice, while WT male and female mice showed EPO-induced cardiac dysfunction (Figures 3C–F). These data suggest that endogenous nNOS activity contributes critically to EPO-stimulated cardiac modulation and that the loss of nNOS may be protective for decreased heart function with chronic EPO treatment. EPO did not change the systolic volume and diameter in all groups (Figures 3H, I).

Loss of EPOR in the non-erythroid tissue of ΔEPORE mice resulted in decreased ejection fraction and fractional shortening (Figure 1) and increased expression of heart failure-associated genes (Figure 2). In contrast, the loss of nNOS did not significantly affect ejection fraction and fractional shortening of nNOS−/− mice (Figure 3) or expression of heart failure-associated genes. In nNOS−/− mice, the expression levels of heart failure-associated genes Cdk8, TGFβ2, and NOX4 and hypertrophic cardiomyopathy-related gene Nppa were analogous to levels in WT mice (Figures 4A, B). However, while EPO treatment increased expression of these heart failure-associated genes in WT mice, EPO treatment did not change expression of heart failure-associated genes Cdk8, TGFβ2, and NOX4 and hypertrophic cardiomyopathy-related genes Nppa in nNOS−/− mice (Figures 4A, B). This suggests that the loss of endogenous nNOS activity interferes with EPO-induced upregulation of cardiac function-related genes, which leads to protection against EPO-induced cardiac dysfunction. Expression of Nox4 protein expression that was enhanced by EPO treatment in WT male mice was not significantly different in nNOS−/− male mice with EPO treatment, in agreement with corresponding Nox4 gene expression (Figures 4A, E, left). Although EPOR expression was increased in nNOS−/− mice (Figure 4D), EPO treatment did not significantly affect heart function in nNOS−/− mice (Figure 3). PI3K/Akt signaling participates in myocardial injury, and modulation of PI3K/Akt can protect against ischemia-reperfusion injury or heart failure (Ghafouri-Fard et al., 2022). EPO cardioprotective activity in animal models of ischemia-reperfusion injury in heart was associated with Akt activation (Parsa et al., 2003; Cai and Semenza, 2004; Kobayashi et al., 2008). We found that EPO treatment increased activation of the Akt signaling pathway indicated by increased Akt phosphorylation in WT mice but did not significantly increase phosphorylated Akt relative to total Akt in nNOS−/− mice (Figure 4E, right), suggesting that pAkt signaling might be critical for EPO response in cardiac function mediated by endogenous nNOS. Surprisingly, increased iNOS expression and protein was observed in nNOS−/− mice compared to WT mice (Figures 4C, F), and EPO treatment decreased iNOS protein in nNOS−/− mice (Figure 4F), suggesting modulation of inflammation in nNOS−/− mice without affecting the cardiac function.