INTRODUCTION
Coronaviruses are enveloped RNA viruses which cause enteric, respiratory, and central nervous system diseases in a variety of animals and humans (
1). The coronaviruses NL63, 229E, and OC43 are adapted to spread in humans, and infection is usually associated with mild respiratory symptoms (
2–8). In contrast, the zoonotic transmission of animal coronaviruses to humans can result in novel, severe diseases. The severe acute respiratory syndrome coronavirus (SARS-CoV), which is believed to have been transmitted from bats via an intermediate host to humans (
9–11), is the causative agent of the respiratory disease SARS, which claimed more than 700 lives in 2002-2003 (
12). Similarly, the recently emerged Middle East respiratory syndrome coronavirus (MERS-CoV) induces a severe, SARS-related respiratory disease, and its spread is at present responsible for 64 deaths (
13,
14). The elucidation of the molecular processes underlying the spread and pathogenesis of highly pathogenic coronaviruses is required to devise effective antiviral strategies and is therefore the focus of current research efforts.
The coronavirus surface protein spike (S) mediates entry into target cells by binding to a cellular receptor and by subsequently fusing the viral envelope with a host cell membrane (
15,
16). The receptor binding activity of the S proteins is located within the S1 subunit, while the S2 subunit harbors the functional elements required for membrane fusion (
15,
16). The SARS-CoV S protein (SARS-S) utilizes angiotensin converting enzyme 2 (ACE2) as a receptor for host cell entry (
17,
18). ACE2, a metallopeptidase, is expressed on major viral target cells, type II pneumocytes and enterocytes (
19–22), and its catalytic domain binds to SARS-S with high affinity (
17,
23). Binding of SARS-S to ACE2 triggers subtle conformational rearrangements in SARS-S, which are believed to increase the sensitivity of the S protein to proteolytic digest at the border between the S1 and S2 subunits (
24,
25). Cleavage of the S protein by host cell proteases is essential for viral infectivity (
15), and the responsible enzymes constitute potential targets for intervention.
The SARS-CoV can hijack two cellular proteolytic systems to ensure the adequate processing of its S protein. Cleavage of SARS-S can be facilitated by cathepsin L, a pH-dependent endo-/lysosomal host cell protease, upon uptake of virions into target cell endosomes (
25). Alternatively, the type II transmembrane serine proteases (TTSPs) TMPRSS2 and HAT can activate SARS-S, presumably by cleavage of SARS-S at or close to the cell surface, and activation of SARS-S by TMPRSS2 allows for cathepsin L-independent cellular entry (
26–28). Both TMPRSS2 and HAT are expressed in ACE2-positive cells in the human lung (
27,
29), and results obtained with surrogate cell culture systems suggest that TMPRSS2 might play a significant role in SARS-CoV spread in the human respiratory tract (
30). Notably, TMPRSS2 and HAT also activate influenza viruses bearing a hemagglutinin with a monobasic cleavage site (
31,
32) and TMPRSS2 was shown to cleave and activate the F protein of human metapneumovirus (
33), indicating that several human respiratory viruses hijack TTSPs to promote their spread.
The role of host cell proteases in SARS-CoV infection is not limited to cleavage of the S protein: two studies suggest that ACE2 is proteolytically processed by host cell proteases and that processing might play an important role in SARS-CoV entry and pathogenesis. It was shown that SARS-S binding to ACE2 triggers processing of ACE2 by a disintegrin and metallopeptidase domain 17 (ADAM17)/tumor necrosis factor α-converting enzyme (TACE), and evidence was provided that this process, which facilitates shedding of ACE2 into the extracellular space, promotes uptake of SARS-CoV into cells (
34,
35). However, it is disputed whether the increased uptake translates into increased infection efficiency (
34,
36). Irrespective of its role in entry, the SARS-S-induced shedding of ACE2 might be integral to the development of SARS. Thus, ACE2 expression was shown to protect against experimentally induced lung injury in a mouse model, and evidence for a decreased ACE2 expression in the context of SARS-CoV infection was obtained (
37,
38). It is therefore conceivable that S protein-induced, ADAM17-mediated shedding of ACE2 might promote SARS pathogenesis. A more recent study demonstrated that ACE2 is also processed by TMPRSS2 and HAT, and it was suggested that ACE2 cleavage increases SARS-S-mediated entry (
28). However, the mechanism underlying augmentation of infection is unclear and the role of ACE2 proteolysis in TMPRSS2/HAT-dependent SARS-S activation is unknown. Similarly, the potential interplay between ACE2 processing by TMPRSS2/HAT and ADAM17 and its consequences for SARS-CoV entry have not been examined.
Here, we show that ACE2 proteolysis by TMPRSS2/HAT accounts for the ability of these proteases to augment SARS-S-driven entry but is dispensable for SARS-S activation. In addition, we provide evidence that increased SARS-S-mediated entry into TMPRSS2/HAT-expressing cells might be due to augmented viral uptake. Finally, we show that TMPRSS2 and ADAM17 compete for ACE2 cleavage and that only processing by TMPRSS2 promotes SARS-S-driven entry.
DISCUSSION
TMPRSS2 and HAT, members of the TTSP family, cleave and activate SARS-S for host cell entry (
26–28,
43,
53). A recent study indicated that TMPRSS2 and HAT also process the SARS-CoV receptor ACE2 and that expression of these proteases increases viral entry into host cells (
28). However, the molecular mechanisms underlying protease-augmented cellular entry and the potential contribution of ACE2 cleavage to SARS-S activation were unknown. Here, we show that arginine and lysine residues within ACE2 amino acids 697 to 716 are essential for ACE2 cleavage by TMPRSS2 and HAT and that ACE2 processing is required for augmentation of SARS-S-driven entry but not for SARS-S activation. Moreover, we demonstrate that ADAM17, an ACE2 sheddase, requires arginine and lysine residues within ACE2 amino acids 652 to 659 for receptor cleavage and competes with TMPRSS2 for ACE2 processing. However, ADAM17 activity did not modulate SARS-S-driven entry. In sum, these results and previously published work (
26–28) indicate that TMPRSS2 facilitates SARS-CoV infection via two independent mechanisms, cleavage of ACE2, which might promote viral uptake, and cleavage of SARS-S, which activates the S protein for membrane fusion.
Several coronaviruses use peptidases as receptors for host cell entry: the novel coronavirus MERS binds to CD26 (
54), most alphacoronaviruses use CD13, and SARS-CoV and the human coronavirus NL63 engage the carboxypeptidase ACE2 (
17,
41). ACE2 is an integral component of the renin-angiotensin system (RAS), which controls blood pressure as well as fluid and salt balance (
55). In addition, ACE2 expression protects against acute respiratory distress syndrome (
37,
38). ACE2 exerts its regulatory activities by facilitating the generation of the heptapeptide Ang 1-7 (
55), which modulates RAS activity by signaling via the G-protein-coupled receptor MAS (
56). Thus, ACE2 is intimately involved in several physiological and pathophysiological processes, and cellular factors modulating ACE2 expression, receptor function, and enzymatic activity might afford novel strategies for therapeutic intervention.
TMPRSS2 and HAT processed ACE2 in an identical fashion, with a short C-terminal ACE2 fragment of approximately 13 kDa being consistently detectable in lysates of cells coexpressing ACE2 and protease but not ACE2 alone. Additional fragments of slightly higher molecular weights were observed upon expression of small amounts of protease and likely constitute cleavage intermediates. Shulla and colleagues previously reported identical processing of ACE2 by TMPRSS2 and HAT but noted the production of a 20-kDa fragment (
28). The reasons for this discrepancy are at present unclear but might relate to batch-specific differences in the 293T cells employed, the use of antigenically tagged ACE2 in the published but not the present study, and most importantly, the amount of ACE2 and protease expressed. The production of the 13-kDa ACE2 cleavage fragment was also observed upon expression of animal orthologs of TMPRSS2 and HAT, indicating that ACE2 cleavage might be conserved between humans and animals. Finally, human hepsin, a TTSP expressed in kidney, pancreas, lung, and other tissues (
57,
58), was found to process ACE2 and SARS-S (not shown). These observations demonstrate that all SARS-S-processing TTSPs identified so far also cleave ACE2 and raise the question how ACE2 cleavage impacts SARS-S-driven entry.
Shulla and colleagues suggested that ACE2 cleavage is required for TMPRSS2 and HAT-mediated augmentation of SARS-S-driven entry (
28), but formal proof was lacking. In addition, it was not investigated whether ACE2 cleavage is required for TMPRSS2-dependent activation of SARS-S for cathepsin L-independent entry. Answering these questions requires the identification of the protease cleavage site in ACE2. ACE2 cleavage by HAT was readily detectable upon analysis of recombinant proteins and mass spectrometric analysis identified R621 as the cleavage site. However, mutation of this residue in the context of cellular ACE2 did not interfere with processing by TMPRSS2 and HAT. Differences in the folding and/or accessibility to cleavage between the recombinant and cellular proteins might account for these differential results. Instead, arginine and lysine residues within ACE2 amino acids 697 to 716 were critical for ACE2 cleavage by TMPRSS2 and HAT, as demonstrated by mutagenic analysis of potential cleavage sites in the membrane-proximal region of ACE2. Residual ACE2 cleavage, occurring after mutation of arginine and lysine residues within 697 to 716 (ACE2 mutant C4), was largely abrogated when R621 was also mutated, suggesting a minor role of this residue in ACE2 processing. Collectively, these results demonstrate that arginine and lysine residues within ACE2 amino acids 697 to 716 are essential for ACE2 cleavage and, based on the molecular weight of the C-terminal cleavage product, might represent the actual cleavage site.
The cleavage-resistant ACE2 mutant C4 was robustly expressed and facilitated SARS-S-driven entry into target cells with efficiency similar to that of the ACE2 wt. Therefore, mutant C4 was used as a tool to investigate the role of ACE2 cleavage in SARS-S-driven entry. The analysis of TMPRSS2/HAT-dependent augmentation of SARS-S-mediated entry clearly revealed that ACE2 processing is a prerequisite to this process. Thus, expression of these proteases augmented entry into target cells expressing the ACE2 wt but not ACE2 mutant C4. In contrast, TMPRSS2 facilitated efficient cathepsin L-independent entry into cells expressing the ACE2 wt and mutant C4, demonstrating that ACE2 cleavage is dispensable for SARS-S activation. This observation is noteworthy, since ACE2 binding is believed to trigger conformational changes in SARS-S which alter the susceptibility of SARS-S to proteolytic activation by trypsin (
25,
59). Thus, one could have assumed that cleavage by TMPRSS2 slightly alters ACE2 conformation and that only SARS-S bound to cleaved ACE2 is appropriately presented for processing by TMPRSS2, a possibility disproved by the present study. In sum, our observations indicate that TMPRSS2 and HAT impact SARS-S-driven entry via two independent mechanisms: ACE2 cleavage by these proteases increases entry efficiency, while SARS-S cleavage by TMPRSS2 activates the S protein for cathepsin L-independent host cell entry (
Fig. 9).
How does ACE2 cleavage by TMPRSS2 and HAT augment SARS-S-driven entry? We did not observe substantial differences in SARS-S binding to cells expressing ACE2 or coexpressing ACE2 and TMPRSS2, demonstrating that augmentation of entry is not due to increased capture of SARS-S. Instead, we found that cells coexpressing TMPRSS2 and ACE2 internalize SARS-S more efficiently than cells expressing ACE2 alone, indicating that TMPRSS2 might promote particle uptake into receptor-positive cells. Particles taken up via cleaved ACE2 might then enter the cathepsin L-dependent pathway and fuse with the endosomal membrane (
Fig. 9). Alternatively, they might be activated by TMPRSS2 and fuse with a vesicle membrane immediately after particle uptake. The molecular mechanism responsible for increased uptake is unclear at present, but we speculate that the ACE2 cleavage fragment harbors signals for internalization which might be parasitized by SARS-CoV to promote particle uptake into target cells.
The involvement of proteases in SARS-S entry is not limited to TTSPs and cathepsin L. ACE2 is cleaved by the metalloprotease ADAM17, which results in shedding of the ACE2 ectodomain (
49–51). Haga and colleagues provided evidence that SARS-S also stimulates ADAM17-dependent ACE2 cleavage and that cleavage promotes uptake of authentic, infectious SARS-CoV into target cells (
34,
35) (
Fig. 9). In contrast, our previous analysis failed to detect evidence for an important contribution of ADAM17 to SARS-S-mediated cellular entry (
36). In the present study, we revisited the role of ADAM17 in SARS-S-mediated entry.
We first asked whether TMPRSS2/HAT and ADAM17 cleave ACE2 at overlapping sites, which would suggest that the two proteases might regulate SARS-S-driven entry in a similar fashion. However, the comparison of ACE2 cleavage products revealed striking differences. The ACE2 ectodomain was shed into culture supernatants upon cleavage by ADAM17 but not TMPRSS2, although we cannot formally exclude the possibility that TMPRSS2-expressing cells released into the culture supernatants an ACE2 cleavage product which was unstable and/or not detectable with the particular antibody used. In contrast, a C-terminal, intracellular cleavage fragment was observed only upon ACE2 processing by TMPRSS2, not ADAM17. The latter finding is in agreement with previous reports suggesting that the intracellular portion released from ACE2 upon cleavage by ADAM17 is unstable (
34,
51,
58). The reason for the differential shedding of the ACE2 ectodomain by ADAM17 and TMPRSS2 is at present unknown. However, the differential fate of the ACE2 cleavage products clearly indicated that these proteases cleave ACE2 at different sites. Indeed, mutagenic analysis revealed that arginine and lysine residues within ACE2 amino acids 652 to 659 are essential for ACE2 shedding by ADAM17 but do not impact ACE2 processing by TMPRSS2/HAT. Previous studies suggested that ADAM17 cleaves ACE2 at residues 708 to 709 (
60) or 716 to 741 (
51), respectively. Therefore, residues 652 to 659 identified in our study might not constitute the cleavage site itself but might determine whether a downstream cleavage site is recognized by ADAM17. Such a scenario is supported by previous studies suggesting that the structure of the juxtamembrane region might be more important for shedding than the sequence of the actual cleavage site (
49,
61,
62).
The ACE2 mutant resistant to ADAM17 cleavage might be useful to further dissect the role of ADAM17 in SARS-S entry, although the reduced expression or stability and receptor function of this mutant will likely complicate such endeavors. In order to commence such analyses, we first addressed if a role of ADAM17 in SARS-S-driven entry can be detected under optimized conditions, in which particles are bound to cells, unbound particles are removed, and ADAM17 is immediately activated or inhibited. However, modulation of ADAM17 activity did not impact SARS-S-driven transduction, arguing against a contribution of this protease to SARS-S-mediated entry, at least under the conditions examined. On the other hand, the ACE2 mutant resistant to processing by ADAM17 failed to robustly facilitate SARS-S-driven entry even when similar amounts of ACE2 mutant and wt protein were expressed on the cell surface (due to titration of the plasmids used for transfection [data not shown]), and the nature of this defect requires further investigation.
Our study provides evidence that cleavage of ACE2 by TMPRSS2, HAT, and potentially other TTSPs could increase uptake of viral particles and is essential for protease-mediated augmentation of SARS-S-driven entry. In contrast, ACE2 processing by TMPRSS2 is dispensable for activation of SARS-S for cathepsin L-independent entry. Although some of these results await confirmation with authentic SARS-CoV, the identification of the ACE2 site(s) controlling cleavage by TTSPs and ADAM17 reveals important insights into ACE2 biology and might afford novel therapeutic strategies for treatment of lung disease. Finally, our results should stimulate efforts to determine whether receptor proteolysis impacts infection by hCoV-229E and MERS-CoV, which, like SARS-CoV, use peptidases as receptors (
54,
63) and are activated by TTSPs (
64,
65).