Transcription-independent Induction of ERBB1 through Hypoxia-inducible Factor 2A Provides Cardioprotection during Ischemia and Reperfusion

Myocardial ischemia and reperfusion injury are among the leading causes of death in Western countries¹  and are commonly attributed to coronary artery blockage, which leads to profound hypoxia in the myocardium.²  Local myocardial tissue hypoxia during myocardial ischemia is associated with the stabilization of hypoxia-inducible factors.³  Hypoxia-inducible factors are heterodimeric transcription factors with two major isoforms of the alpha-subunit—hypoxia-inducible factor 1A and hypoxia-inducible factor 2A. During normoxic conditions, both isoforms are immediately degraded through proteasomal degradation. When oxygen levels fall, both isoforms are stabilized and subsequently form a heterodimer with hypoxia-inducible factor 1B that is transcriptionally active and induces hypoxia-inducible factor target genes.^2,4 

Hypoxia-inducible factors are implicated in mediating adaptive responses during conditions of acute hypoxia. They are widely expressed in many tissues and mediate anti-inflammatory effects in a wide range of diseases, including acute kidney injury, lung injury, or myocardial ischemia and reperfusion injury.^4–7  The protective effects of hypoxia-inducible factors during myocardial ischemia are attributed to the induction of hypoxia-inducible factor target genes. For example, hypoxia-inducible factor 1A has been shown to induce enzymes of the glycolytic pathway, thereby enhancing the anaerobic capacity of myocytes to generate adenosine triphosphate, and providing increased myocardial resistance to ischemia.^8,9  The hypoxia-inducible factor pathway can be targeted pharmacologically using compounds that cause normoxic stabilization of hypoxia-inducible factors. For example, previous studies have demonstrated that pharmacologic hypoxia-inducible factor stabilization, using the pharmacologic agent dimethyloxaloylglycine, is associated with protection from ischemia and reperfusion injury (referred to here as cardioprotection).^3,10  Together, these studies provide evidence that hypoxia-inducible factor can be targeted to render the myocardium more resistant to the detrimental effects of ischemia and reperfusion injury. This is of importance for the field of anesthesiology and perioperative medicine, as it is foreseeable that hypoxia-inducible factor activators could be used in surgical patients. For example, pharmacologic hypoxia-inducible factor activators could be given to elective patients undergoing cardiac surgery for a time course of 2 weeks before cardiac surgery and improve outcomes in cardiac surgery patients.^11–13 

Only recently, studies have identified a functional role of hypoxia-inducible factor 2A in providing cardioprotection during ischemia and reperfusion injury.¹⁰  These studies demonstrate that mice with myocyte-specific deletion of Hif2a experience significantly larger infarct sizes compared to control mice or mice with myocyte-specific Hif1a deletion, thereby implicating hypoxia-inducible factor 2A directly in attenuating myocardial ischemia and reperfusion injury. Subsequent studies indicate that myocyte-specific Hif2a provides cardioprotection through induction of the hypoxia-inducible factor 2A target gene amphiregulin.^10,14,15  Gene-targeted mice for Areg experienced larger myocardial infarct sizes, and treatment with recombinant amphiregulin was associated with attenuated myocardial ischemia and reperfusion injury.¹⁰  Amphiregulin is a known ligand for the epidermal growth factor receptor ERBB1,¹⁶  which induces downstream signaling events including the phosphatidylinositol 3-kinase and Akt/Protein kinase B pathways as direct targets.¹⁷  In the current study, we hypothesized that hypoxia-inducible factor 2A–mediated cardioprotection through amphiregulin could include a functional role of hypoxia-inducible factor 2A in controlling ERBB1 expression, thereby enhancing cardioprotection. Surprisingly, our studies provide evidence of a posttranscriptional function for hypoxia-inducible factor 2A in mediating the induction of ERBB1 during myocardial ischemia and provide genetic evidence for a functional role of myocyte-specific ERBB1 signaling in attenuating myocardial infarction.

Human cardiac myocytes (Catalog No. 6200) from ScienCell (USA) were cultured according to the manufacturer instructions with cardiac myocyte medium. For hypoxia experiments, human cardiac myocytes were incubated in a hypoxia incubator (1% O₂ and 5% CO₂) from Coy Laboratory Products (USA) for specific individual time points.

Carrier-free, recombinant human amphiregulin protein (Amphiregulin, Catalog No. 262-AR-100/CF) and mouse amphiregulin protein (AREG, Catalog No. 989-AR-100/CF) were purchased from R&D Systems (USA). After exposure to normoxic or hypoxic conditions, human cardiac myocytes were washed with phosphate-buffered saline, and cardiac myocyte medium with 20 nM amphiregulin was applied to human cardiac myocytes for 10 min under normoxic conditions. For in vivo treatment, murine amphiregulin was prepared and given to mice as described previously.¹⁰ 

Human cardiac myocytes were transfected with lentiviral particles containing target short hairpin RNAs (MISSION short hairpin RNA lentiviral particles, Sigma, USA) as specified: short hairpin RNA control; nontargeting control short hairpin RNA (SHC001), shHIF1A; Sigma’s MISSION pLKO.1 lentiviral vectors targeting hypoxia-inducible factor 1A (TRCN0000003809), shHIF2A; Sigma’s MISSION pLKO.1 lentiviral vectors targeting hypoxia-inducible factor 2A (TRCN0000003807), shRBM4; Sigma’s MISSION pLKO.1 lentiviral vectors targeting RNA-binding protein 4 (TRCN0000164640), shERBB1; Sigma’s MISSON pLKO.1 lentiviral vectors targeting ERBB1 (TRCN0000039633, TRCN0000039634, TRCN0000010329). Before short hairpin RNA transfection into human cardiac myocytes, a puromycin kill curve was determined for human cardiac myocytes, which showed 1.25 µg/ml puromycin concentration as the working concentration range. Human cardiac myocytes were transduced with short hairpin RNAs on passage 2, and experiments were performed on passage 3. Specifically, shERBB1 cardiac myocytes were knocked down with three different targeting short hairpin RNA to enhance the efficacy of knockdown.

Male and female mice, aged 8 to 16 weeks old, were used for these studies. The weights of animals used were 25 ± 5 g. Myosin Cre (B6.FVB[129]-Tg(Myh6-cre/Esr1), Hif2a^loxP/loxP (Epas1tm1Mcs/J) and C57BL/6 mice were purchased from Jackson Laboratories (Bar Harbor, USA). ErbB1^loxP/loxP mice were kindly provided by Maria Sibilia, Ph.D., Medical University of Vienna (Vienna, Austria).¹⁸ Hif2a^loxP/loxP Myosin Cre+, ErbB1^loxP/loxP Myosin Cre+ mice, and Myosin Cre+ mice were injected with tamoxifen for 5 consecutive days (1 mg/day) for Cre-recombinase activation and allowed to recover for 7 days before the start of the experiments. Genotyping was performed by GeneTyper (USA). Animal experiments and procedures were approved by the University of Colorado Denver and the University of Texas Health Science Center at Houston Institutional Animal Care and Use Committee and performed in an Association for Assessment and Accreditation of Laboratory Animal Care–accredited facility.

For myocardial ischemia and reperfusion injury, we followed previously described protocols from our laboratory.^10,19  Per our primary hypothesis, pilot experiments were performed to find the optimal ischemia time. The 45-minute ischemia and 2-h reperfusion model was selected to analyze the difference in infarct size and cardiac troponin I. Experiments were performed in the mouse surgery suite during daytime hours, ranging from 9 am to 5 pm. Mice were anesthetized by pentobarbital at a dosage of 70 mg/kg. To maintain anesthesia, pentobarbital was given at 10 mg/kg/h, as described previously.^20–22  Mice were then intubated and connected to a ventilator (Servo 900C, Siemens from DRE Veterinary, USA). Ventilation was set on pressure control, with a frequency of 110 breaths per minute, positive end-expiratory pressure (PEEP) of 5 mbar, peak inspiratory pressure above PEEP of 10 mbar, and fraction of inspired oxygen (Fio₂) tension of 0.4. The body temperature of mice was maintained at 37°C throughout the experiments using a temperature-controlled surgical table with a feedback loop to a rectal thermometer. A polyethylene 10 tube (tip outer diameter~) [mm/"], tapered > .024 mm/.011") catheter was placed into the carotid artery for fluid replacement, hemodynamic monitoring, or treatment with medications (e.g., reconstitution experiments). A left lateral thoracotomy was performed, and an 8-0 nylon suture was placed around the left coronary artery. The suture was threaded through a small piece of tube, and from both sides weights were applied by Eppendorf tubes for total left coronary artery occlusion, as described previously.²²  Successful occlusion of the left coronary artery was visually confirmed by the immediate change in color of the myocardium from a bright red to white.^10,23,24  After 45 min of ischemia, the weights were removed and reperfusion was initiated. During ischemia, normal saline with an infusion rate of 0.1 ml/hour was applied via the carotid artery catheter to apply proper fluid replacement. During reperfusion, the infusion rate was changed to 0.8 ml/hour to maintain a mean arterial pressure greater than 60 mmHg and to ensure sufficient reperfusion for successful infarct staining. After the end of experiment, blood was obtained from the portal vein and centrifuged at 13,000g for 10 min for serum collection. Serum was stored at −80°C before usage.

To assess the percentage of the infarcted area relative to the area at risk, infarcted tissues and area at risk were measured using staining techniques.¹⁰  For this purpose, the following protocol was used, as described previously.^3,22,25  After applying myocardial ischemia and reperfusion, the heart was flushed with 5 ml of normal saline via the carotid artery catheter and the left coronary artery was re-occluded. Evan’s blue dye (1%; 800 ul) was injected through the carotid artery catheter. The area at risk (area of the myocardium that is perfused by the coronary artery) was determined by the area that was not stained by Evan’s blue. Following the injection of Evan’s blue, the heart was excised and kept at −20°C for 15 min. Hearts were then sliced into 1-mm sections using a heart matrix (1-mm thickness) and incubated with 1% triphenyltetrazolium chloride for 10 min at 37°C. Stained heart slices were placed in 10% neutral-buffered formalin overnight for fixation. Pictures of the heart slices were captured after they were placed between two glass slides with both slide sides being clamped with Nikon D5300 camera at ×16 magnification. Area at risk and infarct size were determined by using ImageJ software. Samples were excluded in instances where an unintentional air bubble existed in the catheter during Evan’s blue injection (thus resulting in an unsuitable area of risk determination). Even so, blood was obtained and used for cardiac troponin I analysis.

In addition to staining for infarct size, myocardial injury was also determined by measuring cardiac troponin via enzyme-linked immunosorbent assay (ELISA), as we have done in previous studies.^22,23  Cardiac troponin I was measured by ELISA using cardiac troponin I ELISA Kit (Catalog No. CTNI-1-HS) from Life Diagnostics (USA) following the manufacturer’s instructions.

To study expression of ERBB1 in human tissues, we examined ERBB1 transcript and protein concentrations in human cardiac tissues. As described previously,^9,10  human cardiac tissues were provided by the University of Colorado at Denver Division of Cardiology biobank, with all specific patient information deidentified. For studying conditions of myocardial ischemia, ischemic heart disease tissues were obtained from the left ventricle of explanted hearts of patients with ischemic cardiomyopathy, undergoing orthotropic heart transplantation. For normal control tissues, control tissues were obtained from the left ventricle of cardiac allografts that were planned for transplantation but could not be utilized for logistic reasons. The use of these tissues was approved by the University of Colorado Denver Colorado Multiple Institutional Review Board.

For studying transcript expression, messenger RNA (mRNA) was extracted from human cardiac myocytes or tissue using RNeasy mini kit (Catalog No. 74106, Qiagen, USA), as described previously.¹⁰  cDNA was synthesized using iScript cDNA synthesis kit (Catalog No. 170-8891, Biorad, USA). Primers were purchased from Qiagen. The sequences of the primers are as follows: murine epidermal growth factor receptor (Catalog No. QT00101584), human epidermal growth factor receptor (Catalog No. QT00085701), human RNA-binding protein 4 (Catalog No. QT00235186), murine β-actin (Catalog No. QT00095242), and human β-actin (Catalog No. QT01680476).

To study the expression of proteins, immunoblotting experiments were performed as described previously.¹⁰  For tissues, murine hearts were first flushed with 5 ml of ice-cold saline and left coronary artery was re-occluded. Evan’s blue dye was injected via carotid artery catheter. The area that was not stained by Evan’s blue was cut out and snap frozen in liquid nitrogen. Human cardiac myocytes were lysed by Lysis buffer (150 mM NaCl, 25 mM Tris, pH 8.0, 5 mM EDTA, 2% Triton X-100). Heart tissues were lysed using T-PER (Catalog No. 78510, Thermo Fisher, USA). Lysed samples were centrifuged at 13,000g for 20 min and the supernatant collected and stored at −80°C. Protein concentration was measured by BCA assay (Catalog No. 23225, Thermo Fisher). Polyacrylamide gels (8% or 10%) were used for sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Protein samples for sodium dodecyl sulfate–polyacrylamide gel electrophoresis were prepared by mixing 4× Laemmli sample buffer (Catalog No. 1610747, Biorad) and 2-mercaptoethanol with equal concentrations of protein per sample followed by 5-min incubation in 95°C. Proteins were transferred to nitrocellulose membranes by Trans-Blot Turbo system (Catalog No. 1704150, Biorad) and were detected using specific primary antibodies as specified: Human ERBB1 (Cell signaling, USA, Catalog No. 4267), murine ERBB1 (Abcam, United Kingdom, Catalog No. ab52894), RNA-binding protein 4 (Santa Cruz, USA, Catalog No. sc373852), phosphorylated AKT (Cell signaling, Catalog No. 9271), total AKT (Cell signaling, Catalog No. 9272), and β-actin (EMD Millipore, USA, Catalog No. cp01). The specificity of ERBB1 antibodies in human and mouse samples were validated by comparing ERBB1 expression in short hairpin RNA control human cardiac myocytes with shERBB1-treated human cardiac myocytes and tamoxifen-induced Myosin Cre+ to ErbB1^loxP/loxP Myosin Cre+ mouse, respectively. Proteins were detected using the ChemiDoc Imaging system (Catalog No. 12003153, Biorad) using chemiluminescent detection. Densitometry was performed using ImageJ software. ERBB1 protein expression was quantified by capturing the whole band, whether it was in a single- or double-band form.

The primary hypothesis is evaluated in the experiment of measuring infarct sizes on Myosin Cre+ and ErbB1^loxP/loxP Myosin Cre+ mice. It was estimated that these two groups would be respectively associated with a mean infarct size difference of 20% with common SD of 8%. Mean comparison should be evaluated by using the two-sided t test. At the 5% significance level, the size of five mice per group provides a power of 97% for detecting a significant difference between Myosin Cre+ and ErbB1^loxP/loxP Myosin Cre+ mice. Myosin Cre+ and ErbB1^loxP/loxP Myosin Cre+ mice were randomly assigned to amphiregulin-treated or phosphate-buffered saline–treated groups. The investigator was not blinded to amphiregulin treatment but was blinded to the genotype of mice to which the treatment was given at the time of surgery. Outliers were detected using the Grubb’s test and removed from the data. In addition, a post hoc analysis was performed without outliers removed, which can be found in Supplemental Digital Content, figure 1 (https://links.lww.com/ALN/C87). All data followed approximately normal distribution and were summarized as mean ± SD. Equal variance was evaluated by F test and Levene’s test for two experimental group setting and ANOVA setting, respectively. Mean comparison between two experimental groups was evaluated by the two-sample equal-variance t test, or the Satterthwaite t test when the equal variance assumption was violated. ERBB1 fold changes collected from six experimental groups at six time points were analyzed by using a one-way ANOVA, followed by comparing the mean at each study time point to the mean at time zero. Experiments involving two or three factors utilized the complete factorial design. All factors were between-subject factors. Data from these experiments were analyzed by ANOVA including main factors and all possible interactions. Pairwise comparisons of study interest were subsequently conducted. Within each ANOVA framework, the Bonferroni method was applied to adjust for multiple pairwise comparisons of study interest. Two-sided P values were reported, and P values less than 0.05 were considered statistically significant. Data reporting and analysis was reviewed by a faculty-level statistician. Statistical analyses were performed using SAS software (version 9.4, SAS Institute, USA) and Prism (version 7, GraphPad Software, Inc., USA).

Previous studies have shown that during myocardial ischemia and reperfusion injury, hypoxia-inducible factors are stabilized and provide cardioprotection.^3,9,11  Recent studies implicate hypoxia-inducible factor 2A in myocardial protection. These studies show that mice with myocyte-specific Hif2a deletion are more vulnerable to ischemia and reperfusion injury. As a mechanism for hypoxia-inducible factor 2A–elicited cardioprotection, they identified that hypoxia-inducible factor 2A provides cardioprotection through the transcriptional induction of epidermal growth factor, amphiregulin.¹⁰  Here, we pursued the hypothesis that hypoxia-inducible factor 2A could also function to provide cardioprotection by enhancing amphiregulin signaling through induction of amphiregulin receptors. In the heart, amphiregulin binds and signals solely through the ERBB1 receptor.^16,26  ERBB1 is highly expressed in cardiac myocytes, whereas its expression is undetectable in other cellular compartments of the heart, such as vascular endothelia.²⁷  Therefore, we pursued studies to assess ERBB1 transcript and protein content during myocardial ischemia. As a first step, we examined the expression of human ERBB1 in left ventricular tissue samples from patients with ischemic heart disease. For this purpose, we measured ERBB1 mRNA and protein via quantitative reverse transcription–polymerase chain reaction and Western blot, respectively. Cardiac tissues were obtained through the University of Colorado at Denver Biobank^9,10  from patients who underwent orthotropic heart transplantation for ischemic heart disease. Control tissues were obtained from the left ventricle of healthy donor hearts that were destined to be cardiac allograft but could not be used for logistic reasons. We found that ERBB1 mRNA levels were not different between the two groups (fig. 1A). In contrast, ERBB1 protein levels were significantly elevated in ischemic heart disease patients as compared with healthy controls (fig. 1, B and C, 1.12 ± 0.24 vs. 13.01 ± 2.20, P < 0.001). Taken together, these findings demonstrate a transcription-independent induction of ERBB1 protein in human cardiac tissues of patients with ischemic heart disease.

After showing that ERBB1 protein expression is elevated in patients experiencing myocardial ischemia, we performed studies to confirm these findings in in vitro and in vivo models. First, we examined ERBB1 transcript and protein levels in human cardiac myocytes exposed to ambient hypoxia (1% oxygen) over a time period of 0 to 24 h. In line with the above studies in human cardiac tissues, we found that ERBB1 mRNA levels were not altered between baseline and different time points in hypoxia (fig. 2A). In contrast, ERBB1 protein levels significantly increased compared with baseline (0 h) after 8 h of hypoxia exposure and remained elevated throughout the time course of this study (fig. 2, B and C). In subsequent studies, we exposed C57BL/6 mice to 45 min of myocardial ischemia by occluding the left coronary artery (fig. 2D) and assessed ERBB1 expression in tissues from the area at risk. Consistent with our findings in human cardiac tissues, ErbB1 mRNA levels were unchanged after myocardial ischemia (fig. 2E), whereas ERBB1 protein levels were significantly elevated within the area at risk (fig. 2, F and G). Of note, there was one experimental result in both groups of fig. 2E which were excluded by Grubb’s test. Our analysis results remained similar both with and without the exclusion of outliers. Taken together, these findings further suggest that cardiac ERBB1 protein expression is elevated during conditions of limited oxygen availability (hypoxia or myocardial ischemia) without concomitant induction of ErbB1 transcript levels, indicating a transcription-independent mechanism of induction.

Based on previous findings implicating hypoxia-inducible factor 2A in enhancing cardiac amphiregulin signaling,¹⁰  we also examined a potential role of hypoxia inducible factor 2A in mediating ERBB1 protein induction. For this purpose, we used previously described cardiac myocyte cell lines with short hairpin RNA-mediated repression of HIF1A or HIF2A.¹⁰  Human cardiac myocytes were exposed to ambient hypoxia (1% oxygen) for 16 h and the ERBB1 transcript and protein were quantified. We observed no changes of ERBB1 mRNA in HIF1A- or HIF2A-deficient human cardiac myocytes or respective controls upon exposure to ambient hypoxia (fig. 3A). Consistent with our previous findings,¹⁰ HIF1A-deficient human cardiac myocytes demonstrated increased ERBB1 protein after 16 h of ambient hypoxia exposure. In contrast, human cardiac myocytes with short hairpin RNA-mediated HIF2A repression did not show significant elevation in ERBB1 protein (fig. 3, B and C). These findings implicate hypoxia-inducible factor 2A in the induction of ERBB1 protein during conditions of ambient hypoxia.

As a next step, we examined the role of hypoxia-inducible factor 2A as an inducer of ERBB1 protein in a murine model of myocardial ischemia. We exposed Myosin Cre+ or Hif2a^loxP/loxP Myosin Cre+ mice to myocardial ischemia. We previously generated and characterized these transgenic mice to allow for deletion of Hif2a in myocytes after tamoxifen treatment.¹⁰  To achieve Hif2a deletion in the myocardium, Hif2a^loxP/loxP Myosin Cre+ mice (or Cre controls) were treated daily with tamoxifen injections for 5 consecutive days followed by a 7-day recovery period. Mice were then subjected to 45 min of ischemia, and ErbB1 expression was assessed in the area at risk. Similar to our in vitro findings above in human cardiac myocytes with hypoxia-inducible factor 2A deletion, we observed that ErbB1 mRNA expression did not change with myocardial ischemia in both Myosin Cre+ and Hif2a^loxP/loxP Myosin Cre+ mice (fig. 3D). Of note, there was one experimental result in the Hif2a^loxP/loxP Myosin Cre+ ischemia (+I) group shown in figure 3D which was excluded by Grubb’s test. The results of analysis remained similar with and without exclusion of outliers. In contrast, ERBB1 protein expression was elevated in Myosin Cre+ mice after ischemia, whereas ERBB1 protein did not increase in Hif2a^loxP/loxP Myosin Cre+ mice (fig. 3, E and F). Of note, there was one experimental result in the Myosin Cre+ no ischemia (−I) group shown in figure 3F which was excluded by Grubb’s test. The results were different when outlier data were excluded versus when outlier data were included in analysis. Taken together, these results implicate hypoxia-inducible factor 2A in the transcription-independent induction of ERBB1 protein during conditions of oxygen deficiency—such as occurs during ambient hypoxia or myocardial ischemia.

In most instances, hypoxia-inducible factor 2A increases target gene expression by binding to hypoxia-response elements located within the gene’s promoter. Recent studies have identified a transcription-independent mechanism of hypoxia-inducible factor 2A–elicited protein induction. Here, these studies demonstrate RNA binding motif protein 4 to recruit hypoxia-inducible factor 2A in hypoxia to the ERBB1 mRNA reverse hypoxia-response element site, whereby the hypoxia-inducible factor 2A–RNA-binding protein 4–eukaryotic translation initiation factor 4E family member 2 complex actively translates ERBB1 mRNA.²⁸  Based on these findings, we hypothesized a functional role of RNA-binding protein 4 in mediating hypoxia-inducible factor 2A–dependent ERBB1 protein induction in the heart. To address this hypothesis, we generated human cardiac myocytes with short hairpin RNA-mediated repression of RNA-binding protein 4. Knockdown of RNA-binding protein 4 was confirmed by reverse transcription–polymerase chain reaction (fig. 4A) and Western blot (fig. 4B). Subsequently, RNA-binding protein 4 knockdown human cardiac myocytes and respected control (nontargeting short hairpin RNA) human cardiac myocytes were exposed to ambient hypoxia (1% oxygen) for 16 h. We noted that increases in ERBB1 protein after hypoxia exposure of control cells were completely abolished in RNA-binding protein 4 knockdown human cardiac myocytes (fig. 4, C and D). These data indicate the likelihood of elevated ERBB1 proteins in human myocytes after hypoxia exposure involves an interaction between hypoxia-inducible factor 2A and RNA-binding protein 4.

After showing that hypoxia-inducible factor 2A coordinates the posttranscriptional induction of ERBB1 protein expression during hypoxia or myocardial ischemia, we performed studies to address the functional role of ERBB1 signaling during myocardial injury. For this purpose, we generated mice with inducible deletion of ErbB1 in cardiac myocytes by crossing previously described ErbB1^loxP/loxP mice¹⁸  with Myosin Cre+ mice. Indeed, ERBB1 protein was repressed in ErbB1^loxP/loxP Myosin Cre+ mice compared with Myosin Cre+ mice after giving 45 min of ischemia (fig. 5, A and B) to 44%. The fact that we did not see a complete reduction of ERBB1 protein in ErbB1^loxP/loxP Myosin Cre+ mice compared with Myosin Cre+ mice may be related to the fact that ERBB1 could also be expressed in tissues other than cardiac myocytes, such as fibroblasts. As next step, we proceeded with studies of myocardial injury induced by ischemia and reperfusion. As shown in figure 5C, Myosin Cre+ or ErbB1^loxP/loxP Myosin Cre+ mice underwent 45 min of ischemia and 2 h of reperfusion, followed by measurements of infarct sizes and circulating cardiac injury markers. Infarct sizes were significantly larger in ErbB1^loxP/loxP Myosin Cre+ as compared with Myosin Cre+ mice (fig. 5, D and E, 22.46 ± 4.06 vs. 46.14 ± 1.81, P < 0.001). Similarly, serum concentrations of the cardiac injury marker troponin I were significantly elevated in ErbB1^loxP/loxP Myosin Cre+ mice compared with controls after myocardial ischemia and reperfusion injury (fig. 5F). Taken together, these data indicate a cardioprotective role of myocyte-specific ERBB1 signaling during myocardial ischemia and reperfusion injury.

Previous studies in myocardial ischemia and reperfusion injury have suggested an important role of phosphatidylinositol 3-kinase and Akt/protein kinase B (PI3K/Akt) signaling²⁹  in providing cardioprotection; and the phosphatidylinositol 3-kinase/Akt pathway has been previously shown to function as a downstream signaling target of ERBB1.³⁰  Based on these findings, we hypothesized that ERBB1 induces cardioprotective effects through activation of phosphatidylinositol 3-kinase/Akt. To address this hypothesis, we first generated ERBB1 knockdown human cardiac myocytes. Human cardiac myocytes were transfected with ERBB1-targeting short hairpin RNA containing lentivirus. For control, nontargeting short hairpin RNA containing lentivirus was transfected into human cardiac myocytes. Reverse transcription–polymerase chain reaction confirmed knockdown of ERBB1 transcript levels, as shown in figure 6A. Subsequently, we exposed human cardiac myocytes to normoxia or hypoxia (1% oxygen). Control human cardiac myocytes showed an increase in ERBB1 protein after 16 h of hypoxia, whereas ERBB1 knockdown human cardiac myocytes showed repression of ERBB1 protein at baseline and following hypoxia exposure (fig. 6B). Next, we assessed phosphorylated Akt in control and ERBB1 knockdown human cardiac myocytes before and after 16 h of hypoxia treatment, in the absence or presence of amphiregulin stimulation. Dose finding studies showed that 10 min of 20 nM recombinant amphiregulin treatment after 16 h of hypoxia in human cardiac myocytes induces the highest phosphorylated Akt levels (Supplemental Digital Content, figure 2A and 2B, https://links.lww.com/ALN/C88). Hypoxia itself did not significantly increase phosphorylation of Akt (fig. 6, C and D) in human cardiac myocytes. However, when 20 nM recombinant amphiregulin was applied to human cardiac myocytes for 10 min following hypoxia exposure, levels of phosphorylated Akt were significantly increased in control human cardiac myocytes. This effect was abolished in ERBB1 knockdown human cardiac myocytes. These data suggest that increased ERBB1 protein during hypoxia enhances amphiregulin-dependent Akt phosphorylation through ERBB1 signaling. To test whether other growth factors could elicit similar results, epidermal growth factor, epiregulin, heparin-binding epidermal growth factor–like growth factor, and transforming growth factor α were used to treat human cardiac myocytes at the same conditions. Whereas epidermal growth factor, epiregulin, heparin-binding epidermal growth factor–like growth factor, and transforming growth factor α significantly increased phosphorylated Akt, treatment with the monoclonal ERBB1 antibody cetuximab resulted in a moderate decrease in pAkt in epidermal growth factor and heparin-binding epidermal growth factor–like growth factor–treated human cardiac myocytes. On the other hand, cetuximab treatment in epiregulin and transforming growth factor α–treated human cardiac myocytes reduced pAkt levels to baseline, similar to amphiregulin (Supplemental Digital Content, figure 2C and 2D, https://links.lww.com/ALN/C88). Taken together, these data suggest that amphiregulin, epiregulin, and transforming growth factor α are the growth factors that significantly increase phosphorylated Akt after hypoxia exclusively via the ERBB1 receptor.

Based on the above in vitro findings implicating ERBB1 signaling in mediating amphiregulin-elicited Akt phosphorylation during hypoxia, we proceeded with in vivo studies to address the role of ERBB1 signaling in amphiregulin-dependent cardioprotection and concomitant Akt phosphorylation. For this purpose, Myosin Cre+ and ErbB1^loxP/loxP Myosin Cre+ mice underwent 45 min of ischemia followed by 2 h of reperfusion after treatment with 10 ug recombinant amphiregulin via a carotid artery catheter or vehicle control. Consistent with previous findings,¹⁰  amphiregulin treatment decreased the infarct size from 29.1% to 10.46% in Myosin Cre+ mice. In contrast, the cardioprotective effect of amphiregulin treatment was abolished in ErbB1^loxP/loxP Myosin Cre+ mice, which showed similar infarct sizes with or without Areg treatment (47.9% to 46.53%, fig. 7A). Similarly, attenuation of plasma troponin I by amphiregulin treatment was abolished in ErbB1^loxP/loxP Myosin Cre+ mice (fig. 7B). Next, we examined phosphorylated Akt in the area at risk after myocardial ischemia and reperfusion. Consistent with our in vitro data above, Myosin Cre+ mice showed increased phosphorylation of Akt after 45 min of ischemia and 2 h of reperfusion when compared with baseline levels. In contrast, phosphorylated Akt in ErbB1^loxP/loxP Myosin Cre+ mice after 45 min of ischemia and 2 h of reperfusion did not increase significantly from baseline (fig. 7, D and E). The slight, but not statistically significant, increase in phosphorylated Akt may be explained by the incomplete knockdown of ERBB1 in myocytes, or by the activation of ERBB1 expressed in different cellular components of the murine heart. Of note, there was one experimental result in the Myosin Cre+ normoxic (−I) group shown in figure 7E which was excluded by Grubb’s test. The analysis results remained similar with and without exclusion of outliers. Taken together, these findings indicate that amphiregulin-ERBB1–mediated signaling induces cardioprotection by increasing Akt phosphorylation.

In the present study, we demonstrate a functional role for hypoxia-inducible factor 2A in mediating cardioprotection through the induction of amphiregulin receptor ERBB1 in human cardiac myocytes and murine heart. Based on our initial studies demonstrating elevated ERBB1 receptor protein expression in patients with ischemic heart disease, we performed in vitro and in vivo studies to examine the role of hypoxia-inducible factor 2A and ERBB1 signaling during myocardial ischemia and reperfusion injury. We found that ERBB1 proteins were also elevated after ambient hypoxia exposure of human myocytes or in the ischemic myocardial tissue of mice. Together, these studies demonstrated that ERBB1 protein was elevated during conditions of limited availability of oxygen without concomitant changes in ERBB1 transcript expression. Subsequent studies with in vitro knockdown of hypoxia-inducible factor 2A in human myocytes or in mice with myocyte-specific deletion of Hif2a indicates a functional role of hypoxia-inducible factor 2A in mediating ERBB1 protein induction. Based on previous reports showing that hypoxia-inducible factor 2A can mediate posttranscriptional induction of ERBB1 during hypoxia,²⁸  we identified a functional role of RNA-binding protein 4 in mediating cardiac induction of ERBB1 protein. Subsequent studies to address the functional role of ERBB1 signaling during myocardial ischemia demonstrated that induced myocyte-specific deletion of ErbB1 is associated with increased myocardial injury during ischemia and reperfusion. Moreover, the cardioprotective effects of the ERBB1 ligand amphiregulin were abrogated in gene-targeted mice for ErbB1. Together with previous studies implicating hypoxia-inducible factor 2A in the induction of amphiregulin during conditions of myocardial ischemia,¹⁰  the present studies highlight a functional role for hypoxia-inducible factor 2A in coordinating enhanced amphiregulin signaling through the transcriptional induction of amphiregulin and posttranscriptional induction of the amphiregulin receptor, ERBB1 (fig. 8).

Several previous studies have highlighted functional roles of hypoxia-inducible factors in providing cardioprotection from ischemia and reperfusion injury. For example, pharmacologic studies with the hypoxia-inducible factor activator dimethyloxalylglycine are associated with cardiac stabilization of hypoxia-inducible factor 1A and attenuated myocardial infarct sizes after ischemia and reperfusion.³  Subsequent studies have highlighted several potential mechanisms on how hypoxia-inducible factor 1A can provide cardioprotection. Hypoxia-inducible factor 1A is known to enhance the production of the extracellular signaling molecule adenosine and the adenosine A2B receptor (Adora2b).^31,32  Adenosine signaling through its receptors is reported to have anti-inflammatory and protective roles during organ injury.³³  In line with these observations, studies of myocardial ischemia and reperfusion injury indicate that cardioprotective effects of hypoxia-inducible factor 1A depend on purinergic signaling events, including enhanced production of extracellular adenosine and Adora2b signaling.^20,21,34  Other studies suggest that hypoxia-inducible factor 1A could provide cardioprotection from ischemia and reperfusion through interaction with the circadian rhythm protein period 2 and thereby enhance myocardial resistance to ischemia by optimizing cardiac carbohydrate metabolism.^9,11 

Only recently has a functional role of hypoxia-inducible factor 2A in cardioprotection from ischemia and reperfusion injury emerged.¹⁰  Mice with small interfering RNA-mediated repression of hypoxia-inducible factor 1A or with myocyte specific deletion of hypoxia-inducible factor 2A show abolished cardioprotection when treated with dimethyloxaloylglycine prior to myocardial ischemia, suggesting that both hypoxia-inducible factor isoforms contribute to cardioprotection.^3,10  Based on our current study, the signaling mechanisms for hypoxia-inducible factor–dependent cardioprotection are likely controlling different signaling pathways. Whereas hypoxia-inducible factor 1A appears to provide cardioprotection through enhancing purinergic signaling pathways,²¹  hypoxia-inducible factor 2A–dependent cardioprotection involves epidermal growth factor signaling, including the transcriptional induction of epidermal growth factor, amphiregulin,¹⁰  and the posttranscriptional induction of its receptor, ERBB1.

Previous studies have identified functional roles of ERBB1 in the heart. The ERBB1 receptor, one of the four subtypes of ErbB receptor tyrosine kinases (ErbB 1–4), is expressed in cardiac tissues, and deletion of ErbB1 in mouse embryonic stem cells results in failure of mice to survive.^18,35  ERBB1 is expressed mainly in cardiac myocytes, but is also found in fibroblasts.³⁶  Although fibroblasts have been reported to contribute to the healing process by expanding and modulating the extracellular matrix, their role during the postischemic inflammatory phase is limited.³⁷  Previous studies showed that increased human ErbB1 (human epidermal growth factor receptor) expression in the mouse myocardium leads to heart hypertrophy and semilunar valve defects.³⁸  Cardiac knockout of ErbB1 affects the phosphorylation and activation of both ERBB1 and ERBB2 receptors, thus resulting in ventricular dilation, hypertrophy, and abnormal cardiac function in the adult mouse heart.³⁹  Deletion of ErbB1 in vascular smooth muscle cells led to dilation of the aorta and coronary artery as well as disruption of homeostasis in the basal vascular smooth muscle cells.⁴⁰  Previous studies also showed functional roles of ERBB1 and its ligands in the heart during ischemia and reperfusion injury. Such studies provide evidence that during reperfusion, bradykinin activates ERBB1 and thus protects the heart by increased phosphorylation of Akt and preservation of mitochondrial membrane potential.⁴¹  Other studies observed increases in epidermal growth factor in the heart during stress-induced injury in mice and showed that it potentially protects the heart through ERBB1 activation.⁴²  Additional studies have searched for a relationship between ERBB1 and the adenosine A1 receptor during ischemic preconditioning and suggest that the cardioprotective effect of ischemic preconditioning may include ERBB1 signaling, because the effects were abolished after treatment with AG-1478, a pharmacologic antagonist of ERBB1 signaling.⁴³  The reason we investigated amphiregulin and its effect on ERBB1 was because both were simultaneously upregulated through hypoxia-inducible factor 2A during hypoxia. Amphiregulin was the highest differentially regulated transcript in the microarray study comparing postischemic heart tissues from Myosin Cre+ and Hif2a^loxP/loxP Myosin-Cre+ mice, and we sought to investigate the effect of amphiregulin-ERBB1 interaction during myocardial ischemia and reperfusion injury.¹⁰ 

Previous studies have examined transcriptional responses of ERBB1 during conditions of limited oxygen availability. For this purpose, the authors compared mRNA expression of ERBB1 in the left ventricle (hypoxic) to the left atrium (normoxic) in biopsies from each patient’s heart undergoing coronary artery bypass operation, which showed downregulation of HER2 and upregulation of ERBB1.⁴⁴  In contrast, the present studies suggest that ERBB1 mRNA levels did not change during hypoxia, whereas ERBB1 protein levels increased, which suggests posttranscriptional mechanism. The discrepancy in results may come from regional differences in where the normoxic control was harvested, as well as methodological differences in how the tissue was processed. Consistent with our findings, previous studies identified hypoxia-inducible factor 2A as a mediator of this response, including hypoxia-inducible factor 2A–dependent increase of ERBB1 in various human cell lines after hypoxia exposure. Here, metabolic labeling techniques showed that increased translation of ERBB1 mRNA was the driving force behind increased ERBB1 expression.⁴⁵  The mechanism was further studied in glioblastoma cells, which showed that hypoxia-inducible factor 2A along with RNA-binding protein 4 and cap-binding eukaryotic translation initiation factor 4E family member 2 creates a complex which binds to the RNA hypoxic response element on the ERBB1 mRNA, thereby initiating selective cap-dependent protein synthesis and escaping the repression of protein synthesis during hypoxia.²⁸ 

Our current study provides evidence for phosphatidylinositol 3-kinase/Akt signaling as an endpoint of ERBB1-mediated cardioprotection. This is consistent with previous studies which demonstrate Akt promotes cellular survival during ischemia-reperfusion injury in the mouse heart.⁴⁶  Akt specifically falls into the “reperfusioninjury salvage kinase” pathway category.⁴⁷  These studies demonstrate that increased phosphorylation of phosphatidylinositol 3-kinase/Akt during reperfusion apparently functions to activate several cardioprotective mechanisms, including phosphorylation of the pro-apoptotic protein Bcl-2-associated death promoter,⁴⁸  inhibition of mitochondrial cytochrome c release,⁴⁹  or inhibition of the opening of mitochondrial permeability transition pore and maintenance of the mitochondrial membrane potential.⁵⁰ 

Our study has several limitations. First, ERBB1 expression in the human heart may vary within different regions, and the spatiotemporal kinetics of ERBB1 upregulation during myocardial ischemia and reperfusion injury needs further investigation. Second, the ERBB1 receptor can potentially bind to ligands other than amphiregulin, such as epiregulin and transforming growth factor α. Although we demonstrate the specific contribution of amphiregulin–ERBB1 interaction in activating the phosphatidylinositol 3-kinase/Akt signaling pathway in this study, other ligands may play a part in this process and contribute to cardioprotection. Future studies may be necessary to investigate the effect of other ligands, where there might be synergistic effects from the combination of multiple growth factors.

In summary, we have identified hypoxia-inducible factor 2A–dependent induction of ERBB1 protein and concomitant cardioprotection through activation of the phosphatidylinositol 3-kinase/Akt signaling pathway as a novel mechanism for hypoxia-inducible factor 2A in providing cardioprotection from ischemia and reperfusion injury. Interestingly, it appears that hypoxia-inducible factor 2A coordinates the enhanced signaling effects of amphiregulin through transcriptional and posttranscriptional mechanisms. Indeed, previous studies from our laboratory implicate hypoxia-inducible factor 2A as a transcriptional regulator of amphiregulin induction during myocardial ischemia,¹⁰  and our current study implicates hypoxia-inducible factor 2A as an enhancer of ERBB1 protein expression. Taken together, these findings indicate that amphiregulin signaling through ERBB1 is controlled by hypoxia-inducible factor 2A during myocardial ischemia and can be targeted for attenuating myocardial injury after ischemia and reperfusion. Clinical studies will have to determine whether these findings can be translated into patients experiencing ischemic myocardial tissue injury, such as during myocardial infarction or during cardiac surgery. Such translational studies could include pharmacologic activators of hypoxia-inducible factors or could directly target the amphiregulin-ERBB1 signaling pathway.

The authors thank Kelley Brodsky, M.S. (Department of Cardiology, University of Colorado School of Medicine, Aurora, Colorado) and Jessica Wang, Ph.D. (Department of Anesthesiology, McGovern Medical School at UTHealth, Houston, Texas) for reviewing the original data. The authors would also like to thank Kelli Wallen, M.P.H. (Department of Anesthesiology, McGovern Medical School at UTHealth, Houston, Texas) for her editorial assistance.

Supported by grant Nos. R01 DK097075, POI-HL114457, R01-HL109233, R01-DK109574, R01-HL119837, and R01-HL133900 from the National Institutes of Health (Bethesda, Maryland; to Dr. Eltzschig); individual research grant No. DFG KO3884/5-1 from Deutsche Forschungsgemeinschaft (German Research Foundation; Bonn, Germany; to Dr. Koeppen); International Anesthesia Research Society Mentored Research Award (San Francisco, California) and grant Nos. P50-CA098258 and DK056338 from the National Institutes of Health (to Dr. Bowser); grant Nos. K08-HL102267 and R01-HL122472 from the National Heart, Lung, and Blood Institute (Bethesda, Maryland; to Dr. Eckle); grant Nos. P18421 and DKW1212 from the Austrian Science Fund (Vienna, Austria), and the Austrian Federal Government’s GENome research in AUstralia (GEN-AU) program “Austromouse” GZ 200.147/1-VI/1a/2006 and 820966 (Vienna, Austria; to Dr. Sibilia).

The authors declare no competing interests.