ACE2: Structure, Expression, Tissue Distribution, and Function

Structure of ACE2
ACE2 is a 40 kb gene and it is positioned on chromosome Xp22, differently from ACE gene that is located on chromosome 17. The 18 exons of ACE2 are remarkably similar to ACE exons. The ACE2 gene depicts a large polymorphism and several novel polymorphisms of ACE2, with specific geographical distribution, have been described and associated with susceptibility to hypertension and cardiovascular disease (Burrell et al., 2013; Patel et al., 2014; Pinheiro et al., 2019).
The ACE2 gene codifies for a typical zinc-metallopeptidase of 805 amino acids (120 kDa), with a unique catalytic domain. Despite the high resemblance of ACE and ACE2, considerable differences exist in their substrates and products. While ACE acts as dipeptidase, ACE2 removes only a single amino acid from its substrates. Therefore, ACE2 is not active in transforming angiotensin I to angiotensin II and in inactivating bradykinin; moreover, ACE2 is insensitive to ACE inhibitors, like lisinopril and captopril (Tipnis et al., 2000; Rice et al., 2004; Turner et al., 2004).
These differences depend on variances in the three-dimensional structure (3D) of the two enzymes. Comparative homology modeling and crystallography contributed to shed light on ACE2 3D structure (Figure 1). Prabakaran et al. (2004) clarified the major characteristic of ACE2, which is a deep channel on the summit of the protein, hosting the catalytic domain. Specific loops, like the long loop N210-Q221 that is exclusive of ACE2, α helices and a portion of β-sheet are located around the catalytic channel. The negative charge of the channel and the presence of distinct hydrophobic regions contribute to the specificity of the binding site (Prabakaran et al., 2004). The determination of the crystal structure of the extracellular domain to 2.2-3-A resolution from Towler et al. (2004) and the model from Guy et al. (2003) showed that the catalytic domain of ACE and ACE2 are very conserved and have similar mechanisms of action. The main difference stems from the smaller ACE2 pocket, thereby lodging only a single amino acid: the crucial substitution of the Gln281 in ACE binding pocket with Arg273 in ACE2 is likely to be responsible for the steric conflict (Guy et al., 2003; Towler et al., 2004). Another surprise of the ACE2 structure was its C-terminal domain, which—differently from ACE—revealed high homology with collectrin, a renal protein, which is involved in amino acids trafficking through the membrane (Yang et al., 2017).
FIGURE 1  Crystal structure of ACE2. The peptidase domain (PD) is in green, whereas the collectrin homology domain is enclosed in the light cyan square. The active zinc ion is showed enclosed in a red circle, whereas the glycosylation moieties are showed as cyan cubes and denoted by dashed arrows. The structures have been drawn from PDB 1R42 (Towler et al., 2004) by Mol on the PDB website.

Transcriptional, Post-transcriptional, and Post-translational Regulation of ACE2
The location of the ACE2 gene on the X chromosome questions whether one of the two ACE2 is silenced in females, to balance female/male expression dosage (X chromosome inactivation or XCI), or otherwise belongs to the class of “escape genes” which are transcribed on both chromosomes. Interestingly, a wide survey of XCI in several individuals and tissues showed that ACE2 is a heterogeneous escape gene, because it has a tissue-dependent sex bias (Tukiainen et al., 2017).
A growing number of recent findings point to an important role of epigenetic mechanisms associated with several human diseases (Surguchov et al., 2017). In this context, several authors highlighted the regulatory role of 17β-estradiol (E2), a primary female sex steroid, in the expression of ACE2 in a tissue-dependent fashion. Liu et al. (2010) and Stelzig et al. (2020) found out that E2 downregulates ACE2 in kidney and differentiated airway epithelial cells, respectively. The latter result is particularly important, as the male-bias of ACE2 expression in the lung could account for the alleged higher susceptibility of males to COVID-19 symptoms following ACE2-dependent SARS-CoV-2 infection (section Links Between ACE2 and COVID-19) (Jin et al., 2020). Yet, Bukowska et al. (2017) observed that E2 increases ACE2 transcription and expression in human atrial tissue, while at the same time depressing the level of ACE protein. This mechanism attenuates the renin-angiotensin system and, in tandem with anti-inflammatory and anti-oxidative effects, enables a stronger response to myocardial stress and contributes to antiarhythmic effects. The upregulation of ACE2 (and downregulation of ACE) was clearly linked to binding of E2 to Estrogen Receptor alpha (ERα) (Bukowska et al., 2017). The E2-ERα complex might migrate to the nucleus to bind to estrogen response elements, although the actual mechanism is still obscure and should likely involve other co-factors to take into account the observed tissue bias of E2 regulation. Nonetheless, it was demonstrated that the Estrogen Related Receptor alpha (ERRα), which likewise ERα recognizes the estrogen response element in target genes, binds to ACE2 promoter to repress transcription (Tremblay et al., 2010; Lee et al., 2012). Hopefully, future studies will shed more light on the intriguing role of estrogens in ACE2 regulation.
A recent analysis of public genomic and transcriptomic data outlined the role of histone methylation, a classical epigenetic mark, to regulate ACE2 transcription. Indeed, Li Y. et al. (2020) showed that transcription of ACE2 was significantly upregulated when the histone mutant H3K27M was overexpressed to inhibit H3K27me3. Conversely, overexpression of mutant H3K4/9/36M did not change ACE2 transcription. Trimethylation of K27 on H3 is catalyzed by the polycomb groups (PcG), a group of conserved transcriptional gene repressors (Schuettengruber et al., 2017). PcG proteins assemble into two major complexes: PRC1 and PRC2. The simplest model of PcG activity involves trimethylation of H3 by PRC2 at target gene promoters (Blackledge et al., 2015). These epigenetic marks recruit PRC1 on DNA, which in turn acts as E3-ligase and ubiquitinates nearby H2A histones (Storti et al., 2019), triggering silencing of gene transcription by local and reversible compaction of chromatin (Illingworth, 2019). The catalytic subunit of PRC2 is constituted by the EZH2 protein. In agreement with the inverse correlation between ACE2 level and H3K27me3, ACE2 expression in human ESCs was upregulated following EZH2 knock-out (Li Y. et al., 2020). On the other side, recovery of EZH2 restored basal ACE2 levels. Chromatin immunoprecipitation sequencing (ChIP-seq) showed that EZH2 depletion induced H3K27me3 decrease, with concomitant H3K27ac increase, at ACE2 promoter in human ESCs (Li Y. et al., 2020). The role of H3 methylation and acetylation in the epigenetic regulation of ACE2 was also hypothesized by Pinto et al., who demonstrated that co-morbidities such as hypertension, diabetes, and chronic obstructive lung disease increase ACE2 transcription in the lung (Pinto et al., 2020).
Histone methylation does not appear to exhaust the epigenetic regulation of ACE2. Notably, the NAD+-dependent deacetylase SIRT1 binds to ACE2 promoter favoring its transcription during cellular energy stress (Clarke et al., 2014). Two recent unrefereed preprints highlighted other epigenetic mechanisms at play. Corley et al.1 pointed out that DNA methylation across three CpG islands in the ACE2 promoter was lower in lung epithelial cells compared to other cell types, suggesting high transcription in lung tissue. These findings are in excellent agreement with the reported inverse correlation between ACE2 transcription and promoter methylation in tumors, which will be discussed in section ACE2 and Other Diseases. This correlation is also supported by the observation that in children ACE2 is normally hypermethylated and poorly expressed either in the lung and in other organs (Pruimboom, 2020). Glinsky (2020) addressed the epigenetic role of Vitamin D on ACE2 expression, showing by gene set enrichment analysis that the Vitamin D receptor (VDR) should be involved in a set of regulatory pathways conveying on ACE2. More specifically, VDR activation would downregulate ACE2, thus affording a potential reason for the alleged beneficial role of Vitamin D in COVID-19 (section ACE2 and the Inflammatory Response to Sars-CoV-2).
Finally, the strict homeostatic balance of ACE/ACE2 activities suggests transcriptional co-regulation of both proteins. Remarkably, Yang et al. (2016) have demonstrated that a subtle regulatory mechanism acts in cardiac endothelial cells, where the Brg1 chromatin remodeler and the FoxM1 transcription factor cooperate to determine the ACE2/ACE expression ratio, particularly under cardiac hypertrophy of the heart.
Further regulation occurs at the mRNA level. From putative microRNA-binding sites identified in vitro, Lambert et al. (2014) demonstrated that miR-421 downregulates ACE2. According to the hypothesized mechanism, miR-421 modulates ACE2 expression by hampering translation rather than by degradation of mRNA transcripts.
Beside undergoing post-translational modifications by glycosylation and phosphorylation, ACE2 is also post-translationally regulated by shedding from cell membrane through the action of the metalloproteinase ADAM17. The proteolysis of ACE2 releases a soluble, enzymatically active form which corresponds to the ACE2 ectodomain (Jia et al., 2009; Xiao et al., 2014; Conrad et al., 2016). The function, if any, of soluble ACE2 is still obscure, but the shedding mechanism is under strict molecular control. Lambert et al. (2008) highlighted that calmodulin associates with ACE2 to prevent its shedding, while calmodulin inhibitors increase the cellular release of ACE2. Patel identified a positive feedback in the RAAS: Ang II activates ADAM17, thereby increasing the release of ACE2, its negative regulator (Patel et al., 2016b). It is worth noting that high levels of plasma-soluble ACE2 have been associated with myocardial dysfunction (Epelman et al., 2009). The potential pathophysiological role of ADAM17 is further discussed in paragraph ACE2 and the Inflammatory Response to Sars-CoV-2.
Figure 2 summarizes the known transcriptional, post-transcriptional, and post translational regulation pathways of ACE2.
FIGURE 2  Regulation pathways of transcription, translation, and post-translational shedding of ACE2. Red text: non-molecular factors. Black text: molecular factors.

Tissue Distribution of ACE2
Detectable quantities of ACE2 protein have been found almost ubiquitously in tissues across mammalian species, using immunostaining methods. ACE2 is predominantly located in the cardiovascular system and kidney, where it probably plays a role in the maintenance of hydro electrolyte homeostasis (section “Mechanism of Viral Entry Mediated by the S Protein”). In fact, ACE2 is pervasively expressed throughout the vasculature, at the level of the arteries and veins, mainly in smooth muscle cells of the media and in the endothelium (Hamming et al., 2004; Burrell et al., 2005). Such signal from vessels also delivers part of the expression detected in specific organs. Indeed, ACE2 is evident in: coronary vessels and myocardial capillaries (Wiener et al., 2007; Garabelli et al., 2008); lung microvascular endothelial cells (Wiener et al., 2007; Chen et al., 2013); kidney interlobular arteries (Lely et al., 2004); endothelial and smooth muscle cells in the brain (Hamming et al., 2004; Kar et al., 2010). Notably, the mesangium and glomerular endothelium in the kidney, and the endothelial lining of the sinusoids in the liver are allegedly negative for ACE2 (Hamming et al., 2004). On the contrary, ACE2 is virtually absent from the lymphatic system, and human hemato-lymphoid organs (i.e., spleen, lymph nodes, and bone marrow) (Hamming et al., 2004; Li et al., 2007). In blood cells, it has been observed in platelets and macrophages, but not in B and T lymphocytes (Hamming et al., 2004; Fraga-Silva et al., 2011).
Expression of ACE2 was originally identified in rodent heart (Donoghue et al., 2000a), where it was observed to occur in both atrium an ventricle (Gembardt et al., 2005), and, cellularly, in cardiomyocytes and in specialized cells of the sinoatrial node (Burrell et al., 2005; Garabelli et al., 2008; Ferreira et al., 2011; Wang et al., 2017). In human heart, ACE2 has been found in the stromal area in spongiosa layer in aortic valves (Peltonen et al., 2011), where it is expressed in myofibroblasts and fibroblasts (Guy et al., 2008).
High levels of ACE2 protein expression have been detected in mammalian, including human, kidney (Gembardt et al., 2005; Koka et al., 2008; Reich et al., 2008; Giani et al., 2012; Mitani et al., 2014; Grobe et al., 2015; Shi et al., 2015; Larouche-Lebel et al., 2019; Alawi et al., 2020). Strong signals were reported in the brush border of the proximal tubular cells, whereas weak to moderate signals could be found in the glomeruli, Henle’s loop, distal tubules, and collecting duct (Hamming et al., 2004; Lely et al., 2004; Kamilic et al., 2010; Giani et al., 2012; Bae et al., 2015; Cao et al., 2017; Errarte et al., 2017).
In the respiratory tract of primates, positive labeling for ACE2 has been reported at multiple sites, from the nasal and oral mucosa, to the larynx, trachea, bronchi and lung (Liu et al., 2011). Whether ACE2 is expressed in human nasal and oral epithelium remains unclear, as contradictory results have been reported by studies using immunohistochemistry (Hamming et al., 2004; Bertram et al., 2012), in face of positive single-cell RNA sequencing findings (Sungnak et al., 2020). This point is of great interest to understand the role of those tissues in SARS-CoV-2 initial infection, spread and clearance (section “Links Between ACE2 and COVID-19”). In the upper respiratory tract ACE2 is expressed in the epithelial lining and lamina propria, in some muscle cells and in the salivary gland duct epithelium. In the lung, an intense signal for ACE2 protein has been consistently observed in type I and II pneumocytes in several species, including mouse, rat, cat, ferret, monkey and human (Wiener et al., 2007; van den Brand et al., 2008; Liu et al., 2011; Wong et al., 2012; Chen et al., 2013; Zhang B. N. et al., 2019). Data from rodents suggest an age- and gender-dependent pattern of expression, with a more rapid decline with age in males as compared to females (Xie et al., 2006).
Although some ACE2 signal has been observed in the liver, it appears to mainly come from small vessel endothelium, and occasionally bile duct epithelial cells, while negligible expression is observed in hepatocytes (Hamming et al., 2004; Paizis et al., 2005; Guan et al., 2020). ACE2 protein is abundantly expressed in the brush border of enterocytes of all parts of the small intestine, including the duodenum, jejunum, and ileum, but not in enterocytes of the colon. Other organs of the digestive tract, such as the stomach and colon, did not show brush border staining, but rather a positive signal in the muscolaris mucosae and the muscolaris propria (Hamming et al., 2004). In rodents, ACE2 is also expressed in both exocrine and endocrine pancreatic tissue, particularly in the islets of Langerhans (Niu et al., 2008; Fang and Yang, 2010).
ACE2 distribution is widespread in the mouse brain, from the telencephalon to the medulla. As expected, ACE2 is found in brain areas involved in the regulation of cardiovascular function and fluid balance, such as the vascular organ of lamina terminalis, subfornical organ, magnocellular neurons in the hypothalamic paraventricular nucleus, area postrema, nucleus of the solitary tract, dorsal motor nucleus of the vagus, nucleus ambiguous, and rostral ventrolateral medulla (Doobay et al., 2007). However, significant expression had also been reported in brain areas not engaged in the classical functions of the RAAS, namely the piriform cortex, hippocampus, caudate putamen, hypoglossal nucleus and primary motor cortex (Doobay et al., 2007; Lin et al., 2008; Liu et al., 2014). ACE2 immunostaining was identified in neurons as well as astrocytes (Gallagher et al., 2006; Doobay et al., 2007; Yamazato et al., 2007). Furthermore, ACE2 has been documented in the retina, predominantly in the inner nuclear layer but also in photoreceptors (Tikellis et al., 2004; Senanayake et al., 2007).
With regard to the endocrine system, ACE2 expression was found in both male and female reproductive systems. In human testis, ACE2 was localized to the Leydig and Sertoli cells, and might be involved in testicular function (Douglas et al., 2004). At present, no data about ACE2 protein expression is human ovaries is available, although evidence of expression in stroma, theca, and granulosa cells has been reported in other species (Tonellotto dos Santos et al., 2012; Barreta et al., 2015; Pereira et al., 2015).
In rodent bone, ACE2 is expressed in osteoblasts and osteoclasts, as well as in epithelial cells and fibroblasts. However, a similar expression in human samples still awaits clarification (Queiroz-Junior et al., 2019).
In human skin, ACE2 was present in the basal cell layer of the epidermis extending to the basal cell layer of hair follicles. Smooth muscle cells surrounding the sebaceous glands were also positive for ACE2. Weak cytoplasmic staining was observed in sebaceous gland cells. A strong granular staining pattern for ACE2 was seen in cells of the eccrine glands. Positive staining for ACE2 was also noted in the membrane of human fat cells in various organs, including the epicardial adipose tissue (Hamming et al., 2004; Patel et al., 2016a). Globally, ACE2 is chiefly bound to cell membranes, while negligible levels can be detected in the circulation.

ACE2, the RAAS System and Cardioprotection
After the initial discovery of ACE2 in the heart and kidney, it is now clear that it is widely distributed in tissues (section Tissue Distribution of ACE2), where it exerts many physiological effects and may be involved in pathophysiological events (Turner, 2015). The effect of ACE2 which has been more extensively investigated is the regulation of the RAAS system, where ACE2 counter-balances ACE, limiting the potent vasoconstrictive effect of angiotensin II (Ang-II). The first evidence that ACE2 was involved in RAAS control came from the transgenic knockout mouse model (ACE2–/–), which was characterized by severe reduction of cardiac contractility and thinning of the left ventricular wall. Interestingly, in this knockout model disruption of the ACE pathway could rescue the myocardial phenotype (Crackower et al., 2002). In another study, a selective ACE2 knockout model showed high blood pressure, worsened by the infusion of Ang-II (Gurley et al., 2006).
As a matter of fact, ACE2 displays its carboxypeptidase activity converting Ang-II to a heptapeptide, namely Ang1–7 (Turner et al., 2004). ACE2 can also convert angiotensin I (Ang-I) to the non apeptide Ang1–9, which is in turn converted into Ang1–7 by ACE, competing with Ang-I and thus further decreasing Ang-II (Arendse et al., 2019; Figure 3). Ang1–7 has been demonstrated to bind to the MasR receptor, which was initially regarded as an orphan receptor, since the use of a MasR antagonist caused inhibition of Ang1–7 effects (Alenina et al., 2008).
FIGURE 3  ACE2 signalling pathways: ACE2 displays its carboxypeptidase activity converting Ang-II (ANG II) to ANG1–7 and can also convert angiotensin I (ANG-I) to the nonapeptide Ang1–9, which is in turn converted into ANG1–7 by ACE. ANG1-7 binds the MasR receptor to exert its effects on target organs (primarily heart, vessels, lungs). ACE2 might also act via the bradykinin-DABK/BKB1R axis: the increased activity of the DABK axis triggers a proinflammatory cytokine storm. In the last years the ACE2/Ang1–7/MasR system has been intensively studied: physiological effects on cardiomyocytes include modulation of Ca++ signaling and cytokine production, stimulation of cardiomyocytes progenitors and prevention of uncontrolled cell growth (Grobe et al., 2006; Flores-Muñoz et al., 2012; Souza et al., 2013; Chang et al., 2016). Through the Ang1–7 pathway ACE2 produces endothelial antithrombotic effects, vasodilation, nitric oxide release, and inhibits vascular smooth muscle cells (VSMC) proliferation (Loot et al., 2002; Sampaio et al., 2007a,b; Fraga-Silva et al., 2008). In preclinical studies Ang1–7 displayed antifibrotic effects, protecting from deleterious myocardial hypertrophy and modulating left ventricle remodeling after myocardial infarction (MI). In animal models, Ang1–7 also showed an antiarrhythmic action (Ferreira et al., 2001; Santos et al., 2004; Grobe et al., 2006; Gomes et al., 2010), while compensatory ACE2 upregulation has been observed in explanted human hearts, in patients affected by ischemic or dilated cardiomyopathy (Goulter et al., 2004; Burrell et al., 2005).
Additional players contribute to ACE2-mediated cardio-protection. Another substrate of ACE2 carboxypeptidase activity is Angiotensin A (Ang-A), identified in 2008, which has a similar, although less potent, vasopressor effect as AngII (Jankowski et al., 2007). The product of the enzymatic reaction catalyzed by ACE2 is the peptide Ala-Ang (1–7) also known as alamandine (Lautner et al., 2013; Figure 3). Alamandine receptor is a Mas-related G-protein coupled receptor, known as MrgD, which is expressed in cardiomyocytes and blood vessels. MrgD knockout animals develop a severe cardiopathy and alamandine showed a cardioprotective effect in a model of sepsis (Santos et al., 2019).
Unlike ACE, ACE2 is not active on bradykinin, but it can degrade the active bradykinin metabolite des-Arg9 bradykinin (DABK), blocking the signaling pathway of the B1 bradykinin receptor (Vickers et al., 2002; Sodhi et al., 2018).
Other substrates of ACE2 include apelin-13/17, neurotensin, dynorphin A (1–13), and ghrelin (Turner, 2015). Interestingly peptides of apelin family have been demonstrated to upregulate ACE2 expression in physiological condition and more actively during heart failure in in vivo experiments (Sato et al., 2013). ACE2 in turn is able to regulate apelin bioavailability, establishing a negative feedback loop, and the crosstalk between RAAS, ACE2 and Apelin system might play a significant role in the pathophysiology of hypertension (Kalea and Batlle, 2010; Chen et al., 2015).
On the whole these findings suggest that ACE2 might exert antihypertensive and/or cardioprotective effects by different mechanisms, namely: (i) limiting the availability of ACE substrates, (ii) degrading Ang-II, (iii) activating the Ang1-7/MasR and/or Alamandine/MrgD pathways, (iv) interfering with other substrates, such as DABK and apelins (Figure 3).

ACE2 and Lung Injury
ACE2 is highly expressed in type I and II alveolar epithelial cells and in pulmonary small vessels (either endothelial cells or VSMC). The hypothesis that the ACE2 arm of the RAAS system could be of benefit in lung disease derives from the observation that ACE and angiotensin II are upregulated in acute lung injury (ALI), pulmonary fibrosis, pulmonary hypertension, and acute respiratory distress syndrome (ARDS) (Imai et al., 2005, 2010).
ARDS is an overly aggressive form of ALI and it is the final mechanism of lung injury in many diseases, including sepsis, acid aspiration, pancreatitis, anthrax and virus infections (Spanish flu, H5N1 avian flu and SARS). Imai et al. developed three ACE2 knockout mice models with severe ARDS induced by acid aspiration, endotoxin administration or peritoneal sepsis. They showed that ARDS was accompanied by increased vessel permeability, lung oedema, and infiltration of inflammation cells, with consequent impairment of respiratory function (Imai et al., 2005, 2007; Kuba et al., 2006). The phenotype was dramatically improved and rescued either administering ACE2 recombinant analogs or AT1Ra inhibitors (Imai et al., 2005, 2008).
ACE2 may also reduce lung inflammation via Ang1–7/MasR, since treatment with recombinant Ang1–7 in a model of allergic asthma reduced eosinophil mobilization, peri-bronchial inflammation, fibrosis and goblet cells metaplasia (El-Hashim et al., 2012). An anti-inflammatory effect with reduction of airway remodeling has also been demonstrated in another model of chronic asthma, after administration of Ang1–7 analog (Rodrigues-Machado et al., 2013). The underlying mechanism seems to be the modulation of the so-called cytokine storm and particularly the inhibition of transforming growth factor β (TGF- β) and NFkB signaling pathways (Li et al., 2015; Meng et al., 2015).
ACE2 might also modulate lung inflammation via the bradykinin-DABK/BKB1R axis. In fact, decreased ACE2 function in mouse lungs caused increased activity of the DABK axis and triggered a proinflammatory cytokine storm (CXCL5, MIP2, KC, and TNF-α), leading to pulmonary collapse (Sodhi et al., 2018).
In human, ACE/ACE2 imbalance may be related to genetic factors. In particular, a specific polymorphism of the ACE gene, namely ACE D, which determines increased ACE activity and decreased ACE2 activity has been correlated to ARDS susceptibility and mortality rate (Marshall et al., 2002). More recently, high levels of angiotensin II have been reported in patients infected with avian influenza viruses H5N1 and H7N9, and they were strongly predictive of a poor outcome (Huang et al., 2014; Zou et al., 2014).