Research article

ACE2 overexpression inhibits hypoxia-induced collagen production by cardiac fibroblasts

Justin L. Grobe, Shant Der Sarkissian, Jillian M. Stewart, J. Gary Meszaros, Mohan K. Raizada, Michael J. Katovich


Cardiac remodelling is a key risk factor for the development of heart failure in the chronic phase following myocardial infarction. Our previous studies have shown an anti-remodelling role of ACE2 (angiotensin-converting enzyme 2) in vivo during hypertension and that these protective effects are mediated through increased circulating levels of Ang-(1–7) [angiotensin-(1–7)]. In the present study, we have demonstrated that cardiac myocytes have modest ACE2 activity, whereas cardiac fibroblasts do not exhibit any endogenous activity. As fibroblasts are the major cell type found in an infarct zone following a myocardial infarction, we examined the effects of ACE2 gene delivery to cultured cardiac fibroblasts after acute hypoxic exposure. Cardiac fibroblasts from 5-day-old Sprague–Dawley rat hearts were grown to confluence and transduced with a lentiviral vector containing murine ACE2 cDNA under transcriptional control by the EF1α (elongation factor 1α) promoter (lenti-ACE2). Transduction of fibroblasts with lenti-ACE2 resulted in a viral dose-dependent increase in ACE2 activity. This was associated with a significant attenuation of both basal and hypoxia/re-oxygenation-induced collagen production by the fibroblasts. Cytokine production, specifically TGFβ (transforming growth factor β), by these cells was also significantly attenuated by ACE2 expression. Collectively, these results indicate that: (i) endogenous ACE2 activity is observed in cardiac myocytes, but not in cardiac fibroblasts; (ii) ACE2 overexpression in the cardiac fibroblast attenuates collagen production; and (iii) this prevention is probably mediated by decreased expression of cytokines. We conclude that ACE2 expression, limited to cardiac fibroblasts, may represent a novel paradigm for in vivo therapy following acute ischaemia.

  • angiotensin-converting enzyme 2 (ACE2)
  • collagen
  • fibroblast
  • hypoxia
  • remodelling
  • transforming growth factor β (TGFβ)


Acute MI (myocardial infarction) results in local hypoxia, followed by reperfusion, in the infarct zone. During the hypoxic phase of an MI, cardiac myocytes die and formerly quiescent fibroblasts are stimulated by the hypoxic environment to produce large quantities of collagens I and III, and these cells morphologically change into myofibroblasts [13]. These hypoxia-induced changes are tightly controlled by cytokines, including TGFβ (transforming growth factor β) [46].

Although the role of Ang (angiotensin) II in modulating cardiac fibroblast activity is well accepted [711], little is known about the actions of other Ang fragments upon cardiac fibroblasts. Both ACE (Ang-converting enzyme) inhibitors and ARBs (Ang receptor blockers) are used to treat hypertension and heart failure resulting from cardiac remodelling [12,13]. These therapeutic agents, which inhibit the formation and/or actions of AngII, also greatly increase circulating levels of Ang-(1–7), possibly through increased activity of the ACE2 [14]. Zisman et al. reported that ACE2 is the predominant Ang-(1–7)-formation enzyme in the human heart [15], and that this enzyme is up-regulated during heart failure [16]. Studies utilizing ACE2-knockout mice have shown impaired cardiac function in the basal state [17], exaggerated blood pressure responses to exogenous AngII without any morphological cardiac changes [18], and accelerated cardiac dysfunction following transverse aortic constriction [19]. Collectively, these findings are strongly suggestive that the ACE2 and its product may play an important role in cardiac structure and function during pathophysiological states. Additional evidence in support of this concept is our recent finding that lentiviral delivery of the ACE2 gene to the heart of rats prevents both cardiac hypertrophy and myocardial fibrosis in an AngII-infusion model of hypertension [20]. Furthermore, we and others have demonstrated that direct delivery of Ang-(1–7) provides similar cardioprotection in both AngII-dependent [21] and -independent [22,23] models of cardiac remodelling. Iwata et al. [24], using an in vitro model, have demonstrated that collagen production by cultured cardiac fibroblasts is directly attenuated by Ang-(1–7), and that this effect may be mediated by attenuated cytokine signalling. It is well-documented that AngII induces TGFβ expression in cardiac fibroblasts via the AT1R (AngII type 1 receptor) [25], and the effects of AngII on cardiac remodelling have been proposed to be mediated through TGFβ [2632]. Furthermore, inhibitors of the RAS (renin–Ang system; such as ACE inhibitors and AT1R blockers) attenuate extracellular matrix deposition induced by TGFβ [25,3337].

The observation by Averill et al. [38] that Ang-(1–7) is not present in the infarct zone of the Lewis rat following coronary artery ligation, but that Ang-(1–7) is present in high abundance in the surrounding myocardium, led us to hypothesize that normal ACE2 expression, and thereby endogenous Ang-(1–7)-forming ability, is limited to the cardiac myocyte (and is absent in the cardiac fibroblast). Secondly, we hypothesized that ACE2 activity, which would therefore be absent from an infarct zone due to the loss of local myocytes following an MI, could attenuate the (fibrotic) actions of the cardiac fibroblast during and following a hypoxia/reperfusion injury. Thirdly, we hypothesized that ACE2 may mediate its protective effects through inhibiting the production of the pro-remodelling cytokine TGFβ.


Cardiac myocyte isolation and culturing

Neonatal rat cardiac myocytes were isolated using a commercially available kit (Worthington). Briefly, hearts from 5-day-old Sprague–Dawley rats were removed by blunt dissection during halothane anaesthesia, and then digested with trypsin and collagenase. Myocytes were grown in DMEM (Dulbecco's modified Eagle's medium)/F-12 supplemented with 1% penicillin/streptomycin and 10% (v/v) FBS (fetal bovine serum) in a 5% CO2/95% air humidified incubator. This method typically results in cultures containing 93% myocytes and 7% non-myocytes, comprised mostly of fibroblasts and small traces of endothelial and smooth muscle cells [39]. Cells were assayed for ACE2 activity within 1 week, as described below. All procedures were approved by the University of Florida Institutional Care and Use Committee.

Cardiac fibroblast isolation and culturing

Cardiac fibroblasts from neonatal rats were isolated using a method modified from Zhang et al. [40]. Briefly, 5-day-old Sprague–Dawley rats were anaesthetized with halothane before removal of the heart. The heart was then minced, followed by digestion by collagenase and trypsin (Worthington). Cells were cultured in 10% FBS medium [DMEM supplemented with 10% (v/v) FBS, 1% penicillin/streptomycin and 50 μg/ml ascorbic acid]. Cells were split by standard trypsin procedure, and seeded at 3000 cells/cm2. Only passages 1–3 were used for these experiments. The purity of the cultures was >95% fibroblasts, as determined by positive immunocytochemical staining for vimentin (Figure 1) and negative staining for α-SMA (α-smooth muscle actin) and von Willebrand factor, as described previously [4143].

Figure 1 Immunostaining of rat cardiac fibroblasts with vimentin and α-SMA

Cultured neonatal rat cardiac fibroblasts contain positive immunocytochemical staining for the fibroblast marker vimentin, but negative staining for the myofibroblast/vascular smooth muscle cell marker α-SMA. Immunocytochemical staining was also negative for the endothelial marker von Willebrand factor VIII (results not shown).

ACE2 activity assay

ACE2 activity was determined as described previously [44]. Briefly, cells were scraped into reaction buffer [75 mmol/l Tris/HCl (pH 7.5), 1 mol/l NaCl and 0.5 mmol/l ZnCl2]. Cells were then sonicated and protein content was determined using a standard Bradford assay. ACE2 activity was then determined by kinetic assay using the FPS VI {fluorogenic peptide substrate VI [Mca-Tyr-Val-Ala-Asp-Ala-Pro-Lys(Dnp)-OH], where Mca is 7-(methoxycoumarin-4-yl)acetyl and Dnp is 2,4-dinitrophenyl; R&D Systems}. Samples containing ACE2 enzyme (up to 50 μl) were incubated with 100 μmol/l FPS VI, 10 μmol/l captopril (to inhibit ACE activity) and reaction buffer in a final reaction volume of 100 μl at 37°C. In the assay, ACE2 removes the C-terminal dinitrophenyl moiety that quenches the inherent fluorescence of the 7-methoxycoumarin group, resulting in an increase in fluorescence in the presence of ACE2 activity at excitation and emission spectra of 328 and 392 nm respectively.

Lentiviral production and infection

Lenti-ACE2 (lentivirus containing murine ACE2 cDNA) was produced as described previously [20,44]. Cardiac fibroblasts, at approx. 80% confluence, were incubated with serum-free medium containing the lentivirus [at 2.36×105 or 11.82×105 ifu (infection-forming units)/cm2] for 4 h, after which the serum-free medium was replaced with 10% (v/v) FBS medium for 24 h.

Hypoxia/re-oxygenation model

Following infection with lenti-ACE2, cells were serum-deprived for 48 h to synchronize the cell cycle and maintain the cells in a quiescent state, after which 10% FBS medium was returned to the cells. Upon FBS addition, cells were immediately exposed to a 95% N2/5% CO2 mixture (or normoxic control) for 1 h. Following this exposure for 1 h, cells were returned to normoxic conditions.

Soluble collagen assay

Soluble collagen in the cell culture medium was determined 12 h after the end of hypoxia using a kit from Biocolor, according to the manufacturer's instructions. The utility of this dye-based assay was demonstrated recently by Lijnen et al. [45], where the authors demonstrated the fidelity of this simple dye method as equal to that of the more common [3H]proline incorporation assay.

TGFβ assay

Total TGFβ in the cell culture medium from the same cells was measured by ELISA (R&D Systems), according to the manufacturer's instructions. The negative control of unconditioned (never exposed to cells) serum-free medium and the positive control of unconditioned 10% FBS medium were included in the assay. Total TGFβ was undetectable in unconditioned serum-free medium, but unconditioned 10% FBS medium contained approx. 1.0 ng/ml total TGFβ.


Values are means±S.E.M. Soluble collagen and TGFβ results were analysed by two-way ANOVA. Differences were considered significant with P<0.05.


ACE2 activity in cultured cardiac cells

Endogenous ACE2 activities of cultured cardiac myocytes and fibroblasts were examined first. It was determined that while cultured cardiac myocytes had modest, but significant, endogenous ACE2 activity, no such activity was detected in cardiac fibroblasts (Figure 2A). Transduction of cardiac fibroblasts with lenti-ACE2 resulted in a significant increase in ACE2 activity, which represented transgenic ACE2. This increase was viral dose-dependent (Figure 2B). In contrast, we have demonstrated that transduction of cells with control lentiviruses containing a reporter gene, such as GFP (green fluorescent protein) or PLAP (placental alkaline phosphatase; as in Figure 2C), have no effect on ACE2 activity in vivo or in vitro [20,44,46]. From our present results, we chose to use the low (2.36×105 ifu/cm2) dose of lentivirus for subsequent studies, as the level of ACE2 activity induced by this viral concentration (Figure 2B) more closely resembled the ACE2 activity observed in the cultured cardiac myocytes (Figure 2A).

Figure 2 ACE2 activity in cultured cardiac cells

(A) Endogenous ACE2 activity in cultured cardiac myocytes and fibroblasts. Myocytes exhibited a moderate basal ACE2 activity (indicated by a positive slope), whereas fibroblasts do not display any detectable basal ACE2 activity. (B) Lentiviral delivery of the murine ACE2 gene results in dose-dependent increases in ACE2 activity in cultured cardiac fibroblasts. (C) Representative brightfield images showing dose-dependent staining for the PLAP reporter gene in cells infected with lentivirus harbouring an EF1α (elongation factor 1α)–PLAP control transgene. Bars indicate 100 μm. A dose of 2.36×105 ifu/cm2 represents the low viral dose, whereas 11.82×105 ifu/cm2 indicates the high viral dose.

Collagen production

In response to hypoxia/re-oxygenation, cardiac fibroblasts exhibited a significant (approx. 21%; P<0.001) increase in collagen production when compared with normoxic controls (Figure 3A). FBS supplementation during the hypoxia or normoxia treatment was necessary, as no collagen was produced by cells in the absence of FBS (results not shown). Control viruses at varying titres had no effect on collagen production. ACE2 gene transfer, though, resulted in a significant attenuation of basal collagen production (P<0.001) as well as a significant attenuation of the collagen production induced in response to hypoxia/re-oxygenation (P<0.001; Figure 3A). This latter effect was not enhanced by higher doses of ACE2 virus treatment (results not shown). The presence of a significant (P=0.005) interaction between the main effects indicates that ACE2 exhibited more potent antifibrotic effects during hypoxia/re-oxygenation.

Figure 3 Effect of hypoxia/re-oxygenation on (A) soluble collagen and (B) TGFβ production in cultured cardiac fibroblasts

Cultured cardiac fibroblasts were exposed to 1 h of hypoxia and 12 h of re-oxygenation (Hypo). (A) Collagen production was significantly increased after hypoxia/re-oxygenation (main effect, *P<0.001) when compared with normoxic controls (Norm), which was significantly attenuated by ACE2 gene transfer (main effect, P<0.001). A significant (P=0.005) interaction was also observed between the main effects. (B) Medium from these cells exhibit a significantly increased level of total TGFβ protein (main effect, *P=0.05), which was significantly attenuated in cells treated by ACE2 gene transfer (main effect, P<0.001). No significant interaction was observed between the main effects. A viral dose of 2.36×105 ifu/cm2 was used, which corresponds to the ‘low’ dose in Figure 2(B). †P<0.05 compared with Control.

Total TGFβ

Hypoxia/re-oxygenation treatment resulted in a significant increase in total TGFβ protein (main effect, P=0.05). Both baseline and hypoxia/re-oxygenation-induced TGFβ production were significantly (main effect, P<0.001) attenuated by ACE2 gene transfer (Figure 3B). No interaction was observed between main effects (P=0.876), thus indicating that ACE2 decreases TGFβ production to a similar degree in both the basal and hypoxia/re-oxygenation states.


The present series of experiments were designed to evaluate the relative contributions of cardiac cell types to the total ACE2 activity of the myocardium, and to evaluate whether ACE2 gene transfer, specifically to fibroblast cells which typically survive in a hypoxic infarct zone, could alter the production of collagen in cultured fibroblasts and, therefore, probably inhibit the pro-remodelling actions of these cells in vivo.

The role of ACE2 in regulating cardiac structure and function is controversial. Although a majority of studies demonstrate some protective effects of ACE2 on the myocardium [17,19,20], there are reports that disagree with this hypothesis. Gurley et al. [18] recently reported that there were no cardiac abnormalities in the ACE2-knockout mice generated in their laboratories, although this observation, as they demonstrated, may be heavily dependent on the genetic background of the animal. Donoghue et al. [47] observed conduction and rhythmic disturbances in ACE2 transgenic mice. Thus, although the exact effects and mechanisms of action of ACE2 in the myocardium are not fully understood, ACE2 appears to have some regulatory effects on cardiac structure and function.

The observation that cardiac myocytes exhibit a moderate ACE2 activity, whereas cardiac fibroblasts do not show any detectable ACE2 activity, is in strong agreement with the immunohistochemical findings by Averill et al. [38] and Burrell et al. [48]. In the study by Averill et al. [38], it was shown that Ang-(1–7), a major product of ACE2 activity, is found in high abundance in the myocardium, but that this peptide is specifically excluded from an infarct zone, which would contain primarily fibroblasts and myofibroblasts. Burrell et al. [48] showed that ACE2 is localized to the cell membranes of myocytes, infiltrating macrophages and blood vessels in both the infarct zone and the surrounding viable myocardium. Furthermore, Burrell and co-workers [48] demonstrated that, following an MI, ACE2 activity in these cell types is increased in both viable myocardium and infarct zones. Taken together with the results from the present study, these findings suggest an ACE2-mediated paracrine signalling mechanism by which myocytes, macrophages and endothelial and smooth muscle cells may decrease the profibrotic actions of fibroblasts.

It is important to note that the ACE2 activity level observed in cultured myocytes [slope was approx. 0.865 F/F0 (fractional fluorescence) per min; Figure 2A] was more closely approximated by the lowest viral dose used in the cultured fibroblasts (slopes were approx. 2.233 F/F0 per min for the low dose of virus, and 8.695 F/F0 per min for the high dose of virus; Figure 2B). For this reason, we chose to evaluate the effects of ACE2 expression in the fibroblast using the lentivirus at a titre of 2.36×105 ifu/cm2 to elicit a physiological, rather than pharmacological, ACE2 activity. As observed by Donoghue et al. [47], pharmacological levels of ACE2 may actually result in harmful effects in the myocardium. This has also been reported by Oudot et al. [49], who determined that, in an ischaemic rat heart, pharmacological doses of the ACE2 product Ang-(1–7) actually increase NADPH oxidase activity through activation of the AT1R. Clearly more work is required in this area to determine the intricate dose–response effects of AngII and Ang-(1–7) in cardiac tissue. Regardless, the results of the present study support the contention that, at near-physiological levels of activity, ACE2 provides significant protection for cardiac fibroblasts during hypoxia/re-oxygenation.

Taken together, our present study and those of Averill et al. [38] and Burrell et al. [48] suggest an important role for ACE2 in post-MI remodelling; however, none of the studies elucidates the mechanism of ACE2 protection. It seems likely that the degradation of AngII and the production of Ang-(1–7) may both contribute to decreased collagen production by cardiac fibroblasts. AngII is well-established in the literature to cause increased collagen production and thereby cardiac fibrosis [711] and, therefore, any reduction in the local levels of this peptide may lead to decreased collagen production. Conversely, our previous studies have indicated a potent antifibrotic role for Ang-(1–7) in vivo in both AngII-dependent [21] and -independent models [23], and other groups have demonstrated that Ang-(1–7) inhibits remodelling effects in cardiac fibroblasts [24,50] and myocytes [51] in vitro. It is therefore likely that ACE2 can mediate its cardioprotective actions through both the degradation of AngII and the production of Ang-(1–7).

The observation that collagen production during normoxia was significantly reduced by ACE2 overexpression in cardiac fibroblasts was unexpected, and may suggest that baseline production of collagen is tonically inhibited by ACE2 activity. In a pilot study, we placed the ACE2 gene under transcriptional control of the hypoxia-responsive ‘vigilant vector’ (HRE) promoter [52]. Using the same hypoxia/re-oxygenation protocol described in the present study, we observed the same level of reduction in collagen production in response to hypoxia/re-oxygenation, but there was no effect on basal collagen production (results not shown). These results suggest that during hypoxia, when the HRE promoter becomes active and the ACE2 transgene is produced, the transient increase in ACE2 can provide protective effects, whereas during normoxia the promoter is not activated and, thus, normal production of collagen would not be altered. This particular promoter may represent a mechanism by which future studies can differentially examine ACE2-mediated protective effects upon reparative (within the infarct scar) and/or reactive (in the border zone and unaffected myocardium) fibrosis following an MI in vivo. Further studies are needed to investigate this area.

Modulation of TGFβ levels in cardiac cells by ACE2 is a significant finding and suggests a mechanism by which ACE2 can elicit both its antihypertrophic and antifibrotic effects [20]. TGFβ is a well-documented pro-remodelling cytokine [13,34], and expression of the gene for TGFβ is increased in non-myocyte cells of the heart following an MI [31,32,53,54]. Interestingly, 4 weeks after an MI, increased TGFβ gene expression is co-localized with increased collagen expression [31]. Inhibition of this hormone by neutralizing antibodies prevents fibrosis and cardiac function losses in pressure-overloaded animals [55]. Similarly, Okada et al. [6] have recently demonstrated that modulation of this cytokine after an MI can significantly attenuate cardiac remodelling and greatly prolong an animal's survival, and the protective effects of TGFβ inhibition in their study were at least partially mediated by delaying apoptosis of the cardiac fibroblasts. Importantly, the benefits of inhibiting TGFβ appear to be time-dependent, as inhibition of TGFβ by specific neutralizing antibodies at certain time points post-MI can actually cause exacerbated infarct size and greater compensatory hypertrophy [13,56]. This apparent time-dependency may be due to the parallel roles of TGFβ in both reparative and reactive fibrosis, and may actually reflect differential actions of TGFβ in certain cell types and regions of the heart (e.g. inside or outside of the infarct zone).

The results in the present study complement recent in vivo studies from our laboratory and others [2023], which demonstrate that ACE2 and Ang-(1–7) reduce the degree of cardiac fibrosis as well as levels of TGFβ and, thus, suggest future directions of investigation. First, a more direct investigation of the mechanism(s) of ACE2 cardioprotection is required to evaluate the relative contributions of decreased AngII and increased Ang-(1–7). These determinations are typically confounded by the fact that antagonism of either ACE or AT1R is associated with increased Ang-(1–7) formation [14]. Secondly, much work is needed to explore the potential actions of ACE2 and Ang-(1–7) in modulating cardiac redox signalling. The role of reactive oxygen species in the progression of cardiac remodelling and heart failure is a burgeoning field of investigation [57], and the present study supports the contention that ACE2 is a cardioprotective enzyme that may represent a novel therapeutic target for the treatment of cardiovascular disease.

The findings in the present study lead us to the hypothesis that the absence of ACE2-expressing cardiac myocytes within an infarct zone permits overactivity of cardiac fibroblasts and the production of collagen. Furthermore, these results suggest that a therapeutic intervention (such as gene delivery or a pharmacological stimulation) that increases ACE2 activity in or near infarct-zone fibroblasts may provide beneficial anti-remodelling protection following ischaemia. Modulation of ACE2 may also represent a novel mechanism to inhibit collagen deposition in various fibrotic disorders, such as post-operative adhesions or pericarditis. Future investigations should be directed at determining the relative contributions of ACE2-mediated decreases in AngII levels and increases in Ang-(1–7) levels on fibroblast function, investigating the possible role of ACE2 in modulating redox signalling in myocardial cell types and examining the cardiac structural and functional effects of ACE2 gene transfer to an infarct zone in vivo.


J. L. G. was supported by a Pre-Doctoral Fellowship from the American Heart Association: Florida/Puerto Rico Affiliate. S. D. S. was supported by a Post-Doctoral Fellowship from the American Heart Association: Florida/Puerto Rico Affiliate. This work was supported by a grant from the National Institutes of Health (R01- HL 056921-10).

Abbreviations: Ang, angiotensin; ACE, Ang-converting enzyme; AT1R, AngII type 1 receptor; DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; F/F0, fractional fluorescence; FPS VI, fluorogenic peptide substrate VI; ifu, infection-forming units; lenti-ACE2, lentivirus containing murine ACE2 cDNA; MI, myocardial infarction; PLAP, placental alkaline phosphatase; α-SMA, α-smooth muscle actin; TGFβ, transforming growth factor β


View Abstract