Intracellular autoactivation of TMPRSS11A, an airway epithelial transmembrane serine protease.

Type II transmembrane serine proteases (TTSPs) are a group of enzymes participating in diverse biological processes. Some members of the TTSP family are implicated in viral infection. TMPRSS11A is a TTSP expressed on the surface of airway epithelial cells, which has been shown to cleave and activate spike proteins of the severe acute respiratory syndrome (SARS) and the Middle East respiratory syndrome coronaviruses (CoVs). In this study, we examined the mechanism underlying the activation cleavage of TMPRSS11A that converts the one-chain zymogen to a two-chain enzyme. By expression in human embryonic kidney 293, esophageal EC9706, and lung epithelial A549 and 16HBE cells, Western blotting, and site-directed mutagenesis, we found that the activation cleavage of human TMPRSS11A was mediated by autocatalysis. Moreover, we found that TMPRSS11A activation cleavage occurred before the protein reached the cell surface, as indicated by studies with trypsin digestion to remove cell surface proteins, treatment with cell organelle-disturbing agents to block intracellular protein trafficking, and analysis of a soluble form of TMPRSS11A without the transmembrane domain. We also showed that TMPRSS11A was able to cleave the SARS-CoV-2 spike protein. These results reveal an intracellular autocleavage mechanism in TMPRSS11A zymogen activation, which differs from the extracellular zymogen activation reported in other TTSPs. These findings provide new insights into the diverse mechanisms in regulating TTSP activation.

Type II transmembrane serine proteases (TTSPs) are a group of enzymes with a similar domain structural arrangement, including a short N-terminal cytoplasmic tail, a single-span transmembrane domain, and an extended extracellular region with various modules and a C-terminal trypsin-like protease domain (1, 2). TTSPs act in diverse tissues to participate in physiological and pathological processes, including iron metabolism (3, 4), liver metabolism (5), blood pressure regulation (6, 7), food digestion (8), auditory function (9, 10), vascular remodeling (11), epithelial development (12, 13), intestinal barrier function (14, 15), and cancer development (16, 17).

In recent years, TTSPs have been implicated in coronavirus (CoV) infection. In particular, TTSPs expressed in the human respiratory tract, including human airway trypsin-like protease (HAT) (18), the TTSP mosaic serine protease large form (19), differentially expressed in squamous cell carcinoma 1 (DESC1) (19), transmembrane protease serine 2 (TMPRSS2) (18, 20–26), and TMPRSS11A (27, 28), were shown to cleave the severe acute respiratory syndrome (SARS) CoV, the Middle East respiratory syndrome (MERS) CoV or SARS-CoV-2 spike (S) proteins in cell-based studies. As the S proteins are major determinants for receptor binding and membrane fusion in host cells (29), it appears that human airway TTSPs have been exploited by the CoVs to enhance their infectivity.

TTSPs are synthesized in a precursor or zymogen form with little catalytic activity. Proteolytic cleavage at a conserved activation site converts the zymogen to an active enzyme. To date, how CoV-activating TTSPs are activated in cells is not well understood. In this study, we analyzed the activation cleavage of TMPRSS11A, which is expressed in airway epithelial cells (28, 30) and activates SARS and MERS CoV S proteins (27, 28). By immunostaining, flow cytometry, Western blotting, protease digestion, and site-directed mutagenesis, we show that TMPRSS11A is autoactivated inside the cell before reaching the cell surface. This mechanism of intracellular activation cleavage differs from the extracellular activation cleavage reported in other TTSPs. Moreover, we found that TMPRSS11A exhibited the activity in cleaving the SARS-CoV-2 S protein.

Results

Cleavage of the conserved activation site in TMPRSS11A

Human TMPRSS11A consists of 418 amino acids. Fig. 1A shows the domain structure of TMPRSS11A, including an N-terminal cytoplasmic tail, a transmembrane domain (TM), and an extracellular region containing a SEA (sea urchin sperm protein/enteropeptidase/agrin) domain and a C-terminal serine protease domain. The conserved activation cleavage site is at Arg186–Ile187 (Fig. 1A and Fig. S1). There is a disulfide bond (Cys175–Cys292) linking the protease domain to the propepide region after the cleavage at the Arg186–Ile187 (Fig. 1A).

Figure 1.

TMPRSS11A expression and cleavage in HEK293 cells. A, human TMPRSS11A protein consists of an N-terminal cytoplasmic tail, a TM domain, a SEA domain, and a C-terminal protease domain. The activation cleavage site is at Arg186–Ile187 (arrowhead). Catalytic residues in the protease domain and a disulfide bond (s-s) connecting the propeptide region and the protease domain are indicated. A V5 tag (v) is at the C terminus. Expected molecular masses for TMPRSS11A zymogen and the cleaved protease domain fragment are shown. B, flow cytometric analysis of TMPRSS11A on the surface of HEK293 cells transfected with a TMPRSS11A (11A)-expressing plasmid or a vector. Percentages of TMPRSS11A-positive cells are indicated. C, immunostaining of TMPRSS11A in nonpermeabilized and permeabilized HEK293 cells transfected with a TMPRSS11A (11A)-expressing plasmid or vector. Arrows indicate TMPRSS11A-positive cells. Scale bars, 5 μm. D and E, Western blotting of TMPRSS11A proteins in lysates (D) and biotin-labeled surface proteins (E) from HEK293 cells transfected with a TMPRSS11A (11A)-expressing plasmid or a vector. Experiments were done under nonreducing (NR) or reducing (R) conditions with an anti-V5 antibody. TMPRSS11A fragments are indicated by arrowheads. GAPDH and Coomassie Blue (CB)-stained nonspecific bands were used as protein loading controls in the respective experiments. Data are representative of at least three experiments.

To study TMPRSS11A, we expressed human TMPRSS11A with a C-terminal V5 tag in transfected human embryonic kidney 293 (HEK293) cells. In flow cytometry, we found TMPRSS11A on the cell surface (Fig. 1B). In immunostaining, we detected TMPRSS11A on the surface of nonpermeabilized cells (Fig. 1C). The immunofluorescent signal was stronger when the cells were permeabilized, which allows staining both the cell surface and intracellular TMPRSS11A protein (Fig. 1C). These results are consistent with the predicted membrane topology of TMPRSS11A being a TTSP.

In Western blotting of lysates from the transfected HEK293 cells, we detected a single ∼57-kDa band under nonreducing conditions (Fig. 1D, left panel). When Western blotting was done under reducing conditions, three bands of ∼57, ∼37, and ∼28 kDa, respectively, were detected (Fig. 1D, right panel). In Western blotting with cell surface-labeled proteins, a single band (∼57 kDa) and two bands (∼57 and ∼37 kDa) were detected under nonreducing and reducing conditions, respectively (Fig. 1E). Based on the calculated molecular mass, the ∼57-kDa band represents the one-chain TMPRSS11A zymogen, whereas the ∼37-kDa band represents the protease domain fragment cleaved at the conserved activation site (Fig. 1A). Because the V5 tag was at the C terminus, the cleaved N-terminal fragment was not detected by the anti-V5 antibody in Western blotting. These results indicate that TMPRSS11A is activated and present on the surface of the transfected HEK293 cells.

Identification of another cleavage site in the protease domain

The identity of the ∼28-kDa band detected in Western blotting (Fig. 1D, right panel) was unclear. This band was observed under reducing conditions in cell lysates, but not among cell surface-labeled proteins (Fig. 1E, right panel), indicating that this fragment remained inside the cells. Human TMPRSS11A contains two N-glycosylation sites: one at Asn153 in the SEA domain and the other at Asn303 in the protease domain (Fig. 2A). To exclude the possibility that the ∼28-kDa band was an unglycosylated fragment, we treated HEK293 cell lysates with PNGase F to remove N-glycans on proteins. In Western blotting under reducing conditions, the ∼57-, ∼37-, and ∼28-kDa TMPRSS11A bands all migrated faster compared with those in untreated samples (Fig. 2B), indicating that the ∼28-kDa band may be a proteolytically cleaved fragment but not an unglycosylated fragment.

Figure 2.

TMPRSS11A A, illustration of N-glycosylation sites (Y shaped symbols) in human TMPRSS11A. B, Western blotting of TMPRSS11A fragments in cell lysates treated with control buffer or PNGase F. Western blotting was done under reducing (R) conditions with an anti-V5 antibody. Shifted bands after PNGase F treatment are indicated by double arrowheads. C, illustration of a potential cleavage site at Arg265–Ser266 (red arrowhead) in the TMPRSS11A protease domain. D, Western blotting of TMPRSS11A fragments in lysates from HEK293 cells transfected with a vector or plasmids expressing TMPRSS11A WT and the mutant R265A. Experiments were done under reducing (R) conditions with an anti-V5 antibody. TMPRSS11A fragments are indicated by arrowheads. GAPDH was used as a protein loading control. Data are representative of at least three experiments.

Based on the calculated molecular mass, the ∼28-kDa band could be generated from a cleavage at Arg265 in the protease domain (Fig. 2C). To test this hypothesis, we made a plasmid expressing the mutant R265A, in which Arg265 in TMPRSS11A was replaced by Ala. In Western blotting with lysates from transfected HEK293 cells, the ∼28-kDa band was detected in TMPRSS11A WT but not the mutant R265A (Fig. 2D), indicating that the ∼28-kDa band is created by proteolytic cleavage at Arg265 in the protease domain.

Analysis of TMPRSS11A proteins with mutated activation cleavage and catalytic sites

In a recent study in transfected 293T cells, more than seven major TMPRRSS11A fragments were detected by Western blotting (28). It is difficult to know if those fragments were derived by TMPRSS11A autocatalysis or unknown protease(s) or both in 293T cells. To circumvent this problem, we made plasmids expressing mutants R186A and S368A, in which the activation cleavage site at Arg186 and the catalytic Ser368 were mutated to Ala, respectively (Fig. 3A). In Western blotting of lysates from HEK293 cells expressing the R186A mutant, only the ∼57-kDa zymogen band was detected (Fig. 3B), indicating that the ∼37-kDa band was derived from cleavage at the conserved activation site Arg186 and that the single-chain TMPRSS11A was incapable of cleaving at Arg265. Similarly, Western blotting of lysates from HEK293 cells expressing the S368A mutant showed the ∼57-kDa band only (Fig. 3B), indicating that cleavages at Arg186 (generating the ∼37-kDa band) and Arg265 (generating the ∼28-kDa band) depended on the catalytic activity of TMPRSS11A.

Figure 3.

Analysis of inactive TMPRSS11A mutants R186A and S368A. A, illustration of TMPRSS11A WT and inactive mutants R168A (zymogen activation cleavage site) and S368A (active site). B and C, Western blotting of TMPRSS11A fragments in lysates from HEK293 (B) and EC9706 (C) cells transfected with a vector or plasmids expressing TMPRSS11A WT and mutants R168A and S368A. Western blotting was done under reducing (R) conditions with an anti-V5 antibody. TMPRSS11A fragments are indicated by arrowheads. GAPDH was used as a protein loading control. Data are representative of at least three experiments.

TMPRSS11A, also called ECRG1 (esophageal cancer-related gene 1), was first identified in human esophageal cancers (31, 32). To verify our results, we expressed TMPRSS11A WT and mutants R186A and S368A in EC9706 cells, a human esophageal cancer cell line (33). In Western blotting of lysates from transfected EC9706 cells, we detected three bands of ∼57, ∼37, and ∼28 kDa, respectively, in TMPRSS11A WT, but a single ∼57-kDa band in mutants R186A and S386A (Fig. 3C). Similar results were observed in additional experiments with human bronchial (16HBE) and alveolar basal (A549) epithelial cells (Fig. S2). These results are consistent, indicating that TMPRSS11A undergoes autoactivation at Arg186 and subsequent autocleavage at Arg265 in the protease domain.

Intracellular cleavage of TMPRSS11A

To understand if the detected TMPRSS11A autoactivation cleavage occurred intracellularly or on the cell surface, we expressed TMPRSS11A WT in HEK293 cells and treated the cells with trypsin to remove surface proteins. In flow cytometry, TMPRSS11A was detected on the surface of the transfected HEK293 cells (Fig. 4A). The expression was reduced to the background level in the cells treated with trypsin (Fig. 4A). When the cells were lysed and lysates were analyzed by Western blotting, we observed the ∼57-, ∼37-, and ∼28-kDa bands in the cells without or with trypsin treatment (Fig. 4B). These results indicate that TMPRSS11A activation cleavage occurred intracellularly.

Figure 4.

Analysis of TMPRSS11A and corin in HEK293 cells treated with trypsin, BFA, or monensin. A, HEK293 cells were transfected with a control vector and the plasmid expressing TMPRSS11A. TMPRSS11A-expressing cells were treated without or with trypsin. Cell surface TMPRSS11A expression was analyzed by flow cytometry with an anti-V5 antibody. Numbers indicate the percentages of TMPRSS11A-positive cells. Quantitative data of five experiments are shown in the bar graph with p values. B, HEK293 cells expressing TMPRSS11A were treated with buffer (control) (−) or trypsin (+) before being lysed for Western blotting under reducing (R) conditions. C, Western blotting of TMPRSS11A fragments in lysates from HEK293 cells treated with buffer (−) or BFA (left panel) or monensin (right panel) (+). Western blotting was done under reducing (R) conditions with an anti-V5 antibody. D, Western blotting of corin protein in HEK293 cells treated without (−) or with (+) trypsin (left panel), BFA (middle panel), or monensin (right panel). TMPRSS11A and corin fragments are indicated by arrowheads. Data are representative of at least three experiments.

We next treated HEK293 cells expressing TMPRSS11A with brefeldin A (BFA) and monensin, which inhibit protein trafficking in the endoplasmic reticulum (ER) and the Golgi (34). In Western blotting, we found the ∼57-, ∼37-, and ∼28-kDa bands in TMPRSS11A-expressing cells without or with BFA or monensin treatment (Fig. 4C). In these studies, we did parallel control experiments with corin (Fig. S3), a TTSP known to be activated on the cell surface but not intracellularly (6). In Western blotting, the ∼40-kDa corin protease domain fragment from activation cleavage was detected in the cells without, but not with, trypsin, BFA, or monensin treatment (Fig. 4D). These results indicate that, unlike corin, TMPRSS11A is activated intracellularly before reaching the cell surface.

Intracellular cleavage of soluble TMPRSS11A

To examine if the transmembrane domain in TMPRSS11A is required for activation cleavage, we made a plasmid expressing a soluble form of TMPRSS11A, in which the cytoplasmic tail and the transmembrane domain were replaced with the signal peptide of IgK (Fig. 5A) (35). In addition, we made another plasmid expressing an inactive soluble TMPRSS11A mutant (soluble S368A), in which the catalytic Ser368 was replaced by Ala (Fig. 5A). As expected, TMPRSS11A WT, but not soluble TMPRSS11A (s11A), was found on the surface of transfected HEK293 cells in flow cytometry (Fig. 5B). In Western blotting, we found all three bands of ∼57/53, ∼37, and ∼28 kDa, respectively, in lysates from HEK293 cells expressing TMPRSS11A WT and s11A (Fig. 5C), indicating that soluble TMPRSS11A undergoes similar intracellular activation cleavage and that the transmembrane domain is not required for TMPRSS11A autoactivation. We next incubated the conditioned medium containing s11A with a recombinant SARS–CoV-2 S protein fragment corresponding to the nearly entire extracellular region (residues 16-1213) produced from insect cells. We detected a ∼60-kDa band (Fig. 5D), which is close to the calculated molecular mass of the cleaved S2 fragment (59 kDa). The ∼60-kDa band was not detected in samples from vector-transfected cells or cells expressing the inactive soluble TMPRSS11A S368A (Fig. 5D). In a positive control, a similar ∼60-kDa band was observed when we used the conditioned medium from cells expressing a soluble form of TMPRSS2, which cleaves SARS–CoV-2 S protein (22) (Fig. 5D). We did not detect any SARS–CoV-2 S protein fragments of ∼45 kDa, which is the calculated molecular mass of the S2' fragment. These results indicate that TMPRSS11A is capable of cleaving SARS–CoV-2 S protein at least in vitro.

Figure 5.

Expression and cleavage of a soluble form of TMPRSS11A. A, illustration of soluble forms of TMPRSS11A (s11A) and the mutant TMPRSS11A S368A (sS368A) with a signal peptide from IgK and the entire extracellular region. B, flow cytometry was done with an anti-V5 antibody to analyze TMPRSS11A on the surface of HEK293 cells transfected with a vector or plasmids expressing TMPRSS11A WT and s11A. Numbers indicate the percentages of TMPRSS11A-positive cells. Quantitative data from five experiments are shown in the bar graph with p values. C, Western blotting of TMPRSS11A fragments in lysates from HEK293 cells transfected with a vector or plasmids expressing TMPRSS11A WT and s11A. Western blotting was done under reducing (R) conditions with an anti-V5 antibody. TMPRSS11A fragments are indicated by arrowheads. D, recombinant SARS–CoV-2 S-ECD was incubated with the conditioned medium from HEK293 cells transfected with a vector or plasmids expressing s11A, sS368A, and a soluble form of TMPRSS2 (sTMPRSS2). Western blotting was done under reducing conditions with an anti-FLAG antibody. The uncleaved S-ECD fragment (filled arrowhead) and the cleaved ∼60-kDa fragment (open arrowhead) are indicated. GAPDH and Coomassie Blue (CB)-stained nonspecific bands were used as protein loading controls in respective experiments. Data are representative of three experiments.

Intermolecular cleavage of TMPRSS11A

The results described above support TMPRSS11A autoactivation. It was unclear if the autoactivation cleavage of TMPRSS11A occurs in cis (intramolecular) or in trans (intermolecular). To address this question, we further analyzed the soluble S368A mutant (sS368A) (Fig. 5A). In transfected HEK293 cells, sS368A was detected in the conditioned medium, as expected (Fig. 6A). On Western blots, the sS368A fragment in the conditioned medium had a higher molecular mass than that in lysates (∼57 versus ∼53 kDa) (Fig. S4). When the samples were treated with PNGase F, the sS368A fragment from the conditioned medium and lysates migrated faster at ∼53 and ∼51 kDa, respectively (Fig. S4). The results suggest that other conformational changes or post-translational modifications may account for the higher molecular mass observed in the sS368A fragment from the conditioned medium.

Figure 6.

Expression and cleavage of the soluble TMPRSS11A mutant S368A. A, immunoprecipitation and Western blotting of TMPRSS11A protein in the conditioned medium from HEK293 cells transfected with a vector or plasmids expressing the sS368A mutant and TMPRSS11A WT. B, immunoprecipitation and Western blotting of TMPRSS11A protein in the conditioned medium from HEK293 cells transfected with a vector or the plasmid expressing the sS368A mutant together with a vector or plasmids expressing TMPRSS11A WT and mutants R186A and S368A. In Western blotting, an anti-V5 antibody was used. Coomassie Blue (CB)-stained nonspecific bands were used as a protein loading control. Data are representative of at least three experiments.

We next transfected HEK293 cells with the plasmid expressing the sS368A mutant together with plasmids expressing TMPRSS11A WT and mutants R186A and S368A or a control vector. In Western blotting, the ∼37-kDa band derived from activation cleavage was observed in the conditioned medium from HEK293 cells co-expressing TMPRSS11A WT, but not mutants R186A and S368A (Fig. 6B), indicating that intermolecular cleavage of TMPRSS11A did occur under our experimental conditions. We then did another experiment, in which plasmid expressing the sS368A mutant was co-transfected with plasmids expressing TMPRSS11A, hepsin, TMPRSS2, and corin in HEK293 cells. In Western blotting of the conditioned medium, the ∼37-kDa band from the sS368A mutant was detected in samples from TMPRSS11A-, hepsin- and TMPRSS2-expressing, but not corin-expressing, cells. The levels of this band, however, were much lower in samples from hepsin- and TMPRSS2-expressing cells than that in TMPRSS11A-expressing cells (Fig. S5). These results support the idea of TMPRSS11A autoactivation, although the possibility that other TTSP-mediated transactivation cleavage may occur cannot be excluded.

Effects of HAI-1 and HAI-2 on TMPRSS11A activation cleavage

Hepatocyte growth factor activator inhibitors 1 and 2 (HAI-1 and HAI-2) are structurally related type I transmembrane serine protease inhibitors (36). Recently, HAI-1 was shown to inhibit HAT and DESC1, but not TMPRSS11A, activity in cell-based studies (28). To examine the effect of HAI-1 and HAI-2 on TMPRSS11A activation cleavage, we co-transfected HEK293 cells with plasmids expressing TMPRSS11A WT and human HAI-1 or HAI-2. In Western blotting of lysates from the transfected cells, we detected the ∼37-kDa TMPRSS11A band in cells co-expressing HAI-1 (Fig. 7, top panel). However, the level of the ∼37-kDa band was lower than that in control vector co-transfected cells. In cells co-expressing TMPRSS11A and HAI-2, the ∼37-kDa band was barely visible (Fig. 7, top panel). The ∼28-kDa TMPRSS11A band was not detected in cells co-expressing HAI-1 or HAI-2 (Fig. 7, top panel). These results indicate that HAI-2 is more potent than HAI-1 in inhibiting TMPRSS11A activity when co-expressed with TMPRSS11A in cells.

Figure 7.

Effects of HAI-1 and HAI-2 expression on TMPRSS11A cleavage. HEK293 cells were transfected with a vector or plasmid expressing TMPRSS11A (11A) together with a vector or plasmids expressing human HAI-1 and HAI-2. TMPRSS11A fragments in cell lysates were analyzed by Western blotting under reducing (R) conditions with an anti-V5 antibody (top panel). GAPDH was used as a protein loading control (second panel). HAI-1 (third panel), and HAI-2 (bottom panel) protein expression in the transfected cells was verified in parallel Western blotting experiments using aliquots of the same lysate samples and an anti-FLAG antibody. Data are representative of at least three experiments.

Discussion

TMPRSS11A is an airway epithelial TTSP implicated in SARS and MERS CoV infection (27, 28). For trypsin-like serine proteases, zymogen activation is essential for biological function (37). In this study, we examined the activation cleavage of human TMPRSS11A in HEK293 cells, esophageal cancer EC9706 cells, and lung epithelial 16HBE and A549 cells. Our results indicate that TMPRSS11A activation cleavage is mediated primarily by autocatalysis. This conclusion is supported by the findings that no activation cleavage was detected in catalytically inactive mutants R186A and S368A. In addition, TMPRSS11A WT, but not mutants R186A and S368A, cleaved the soluble TMPRSS11A mutant S368A, indicating that there were no other endogenous proteases capable of cleaving TMPRSS11A in the cells tested in our study. Moreover, our results show that TMPRSS11A activation cleavage occurs before the protease reaches the cell surface, as indicated by trypsin digestion, BFA, and monensin treatment, and intracellular cleavage of soluble TMPRSS11A. Together, our results reveal an intracellular autoactivation mechanism in converting one-chain TMPRSS11A zymogen into a two-chain enzyme.

Zymogen activation has been studied in TTSPs. To date, several distinct mechanisms have been identified. For example, hepsin and matriptase 2 (hepatic TTSPs) are autoactivated on the cell surface but not inside the cell (38–40). Corin (a cardiac TTSP) is activated in the extracellular space by proprotein convertase subtilisin/kexin-6 (6, 41). Enteropeptidase (an intestinal TTSP) is activated extracellularly and in a reciprocal manner in the presence of its substrate, trypsinogen (42). Similarly, epithelial matriptase activation is mediated by autocleavage, probably on the cell membrane (43), and by reciprocal activation with prostasin, a glycosylphosphatidylinositol-anchored serine protease on the cell surface (44–46). More recently, activation cleavage of HAT and DESC1 (airway TTSPs) was observed in transfected cells (28); however, the responsible proteases (HAT and DESC1 versus unknown proteases) and the location of the cleavage (intracellular versus cell surface) are not defined. Together with our new findings of intracellular autoactivation cleavage in TMPRSS11A, these data show that molecular and cellular mechanisms in zymogen activation vary considerably among TTSPs.

TMPRSS11A was identified as a putative tumor suppressor in human esophageal cancers (32, 47). TMPRSS11A variants are linked to the risk of oral and esophageal squamous cell carcinomas (31, 48, 49). To date, the physiological function of TMPRSS11A remains unclear. In mice, the Tmprss11a gene is co-localized with six other Tmprss11 genes in a locus on chromosome 5E1 (50). Most of the Tmprss11 genes are expressed in multiple epithelial tissues, including the trachea, esophagus, stomach, bladder, and skin (51, 52). Tmprss11a-deficient mice, however, did not exhibit noticeable defects in embryonic development and postnatal survival (51, 53). In humans, TMPRSS11A is expressed in the trachea (27, 28). In single-cell analysis, TMPRSS11A expression was detected in basal and secretory cells in human and mouse airway epithelia (30). In Tmprss11a-deficient mice, no apparent abnormalities were observed in trachea sections (51). These results indicate that TMPRSS11A is unnecessary for normal trachea structures but may have a regulatory function in the airway epithelium, which is yet identified.

Human airway epithelial proteases are important in CoV pathogenicity. By cleaving CoV S proteins between the S1/S2 and the S2' sites, the proteases enhance viral entry and cell-cell fusion in airway tissues (22, 54, 55). In purified protein-based experiments, TMPRSS11A cleaved SARS–CoV S protein at both the S1/S2 and S2' sites (27). Similar TMPRSS11A-mediated cleavages of MERS–CoV S protein were observed in cell-based studies (28). In our study, we showed that TMPRSS11A cleaved the recombinant SARS–CoV-2 S protein extracellular fragment, producing an ∼60-kDa band. A similar ∼60-kDa band was also detected when sTMPRSS2 was used. Previously, the full-length SARS–CoV-2 S protein expressed in HEK293T cells was shown to be cleaved by furin at the S1/S2 site, generating an ∼90-kDa band, consistent with the calculated molecular mass of a glycosylated S2 subunit (24). In that study, the S2' fragment with an expected molecular mass of 68 kDa (with glycosylation) was not detected. Together with previous reports (27, 28), the results from our study support a potential role of TMPRSS11A in CoV S protein cleavage. Additional studies will be important to verify the precise cleavage site in the SARS–CoV-2 S protein by TMPRSS11A. It is known that CoV S protein cleavage can occur at different locations in host cells: on the cell surface during virus entry, inside the cell during virus replication and S protein synthesis, and upon release from the cell (29, 54). In our study, we found activation-cleaved two-chain TMPRSS11A on the cell membrane and inside the cell. Further studies are required to determine whether and how TMPRSS11A enhances CoV infectivity in host cells.

Physiologically, protease activities are tightly regulated. In a previous study (28), TMPRSS11A-mediated cleavage of influenza A hemagglutinin was inhibited by a small molecule serine protease inhibitor, but not HAI-1, which inhibits several epithelial TTSPs (36). Moreover, co-expression of TMPRSS11A and HAI-1 in 293T cells reduced, but did not abolish, the cleavage of TMPRSS11A (28). Consistently, we found that HAI-1 was less effective than HAI-2 in blocking TMPRSS11A autoactivation cleavage when co-expressed in HEK293 cells. In single-cell analysis of the human airway epithelium, levels of SPINT2 (encoding HAI-2) expression were ∼2-fold higher than that of SPINT1 (encoding HAI-1) (30). These results are consistent with our findings, indicating that HAI-2 is more important than HAI-1 in regulating TMPRSS11A activity in the respiratory track.

In summary, we examined the mechanism underlying TMPRSS11A zymogen activation cleavage. We show that TMPRSS11A activation cleavage is mediated primarily by autocatalysis that occurs before the protease reaches the cell surface. This intracellular autoactivation mechanism differs from the extracellular activation mechanism reported in other TTSPs. We also show that TMPRSS11A cleaved SARS–CoV-2 S protein in cell culture-derived medium. The results indicate that, in addition to TMPRSS2, other TTSPs in the respiratory system may also participate in SARS–CoV-2 infection. These findings should encourage more studies to understand the role of airway epithelial TTSPs in CoV infection.

Experimental procedures

Cell culture

HEK293 cells (ATCC, CRL-1573, authenticated by STR profiling) were cultured in Dulbecco's modified Eagle's medium (Corning, 10-0130CVRC) in the presence of 10% fetal bovine serum (FBS) (Gibco, 16000-044). Human esophageal squamous cell carcinoma EC9706 cells were from Shanghai Fuxiang Biological Technology (Shanghai, China) and cultured in RPMI 1640 medium (Hyclone, sh30255) with 10% FBS. SV40-transfected human bronchial epithelial 16HBE cells and adenocarcinomic human alveolar basal epithelial A549 cells were from Ningbo Mingzhou Biological Technology (Ningbo, China; STR profiled) and cultured in minimum essential medium (Hyclone, SH300024) and RPMI 1640 medium, respectively, with 10% FBS. All cells were grown at 37°C with 5% CO2.

Expression plasmids

Plasmids expressing human corin and hepsin were pcDNA 3.1/V5 (Thermo Fisher, K4800-01) based and encoded a C-terminal V5 tag, as described previously (40, 56). Full-length cDNAs encoding human TMPRSS11A and TMPRSS2 were amplified from myeloma (made in this study) and prostate (Clontech 636743) cDNA libraries, respectively, and inserted into the pcDNA 3.1/V5 plasmid to express TMPRSS11A and TMPRSS2. PCR-based site-directed mutagenesis (ClonExpress One Step Cloning kit) was carried out to make constructs expressing TMPRSS11A mutants R186A, R265A, and S368A. To express soluble forms of TMPRSS11A (sTMPRSS11A and sTMPRSS11A S368A), cDNA sequences encoding the extracellular regions of TMPRSS11A WT and the mutant S368A were amplified by PCR and inserted into pSecTag/FRT/V5 plasmid (Thermo Fisher, K6025–01) encoding an N-terminal Igκ signal peptide (35). Another pSecTag-based plasmid was made to express a soluble form of TMPRSS2 with the extracellular region (residues 106-492). All expressed TMPRSS11A proteins contained a C-terminal V5 tag. Human HAI-1 cDNA (106-1542 bp) and HAI-2 cDNA (84-759 bp) were amplified from a HEK293 cell-derived library and cloned into pcDNA 3.1 plasmid with an inserted 5' sequence encoding the CD33 signal peptide and a FLAG tag. A list of expression plasmids used in this study is included in Table S1.

Cell transfection

HEK293, EC9706, 16HBE, and A549 cells were cultured under the conditions described above. When the cells were at ∼80% confluent, expression and control plasmids were transfected into the cells using PolyJet reagents (SignaGen Laboratories, SL100688) at 37 °C, based on the manufacturer's instructions. After 6 h of incubation, the cells were switched to fresh medium and incubated for 24-72 h before being used for further studies described below.

Flow cytometry

To verify cell surface expression of TMPRSS11A proteins, transfected HEK293 cells were detached from culture plates with 0.02% (w/v) EDTA without or with 0.25% (w/v) trypsin (Gibco, 25200). After being washed with serum-free medium, the cells were incubated with an anti-V5 antibody (Thermo Fisher, R96025, 1:1000) at 37 °C for 1 h. After washing with PBS, an Alexa Fluor 488-labeled secondary antibody (Invitrogen, A21202, 1:500) was added to the cells and incubated at room temperature in the dark for 1 h. After washing, the cells were analyzed with a flow cytometer (Gallios, Beckman). Pyridinium iodide (Sigma) was used for life gating. Data were analyzed with Kaluza software.

Immunostaining

HEK293 cells were cultured in 12-well–plates with glass coverslips (20 mm in diameter) and transfected with plasmids using PolyJet reagents at 37 °C, as described above. After 6 h, the cells were switched to fresh medium. After 24 h, the cells were fixed with pre-cooled acetone (cell membrane permeabilized) or paraformaldehyde (4% v/v) (cell membrane not permeabilized) at room temperature for 5 min and incubated with 5% (w/v) BSA in PBS at 37 °C for 1 h. After being washed with PBS, the cells were incubated with the anti-V5 antibody (described above) at 37 °C for 1 h, followed by incubation with the Alexa Fluor 488-labeled secondary antibody (described above) at 37 °C for 1 h. After washing, the coverslips were mounted with the 4′,6-diamidino-2-phenylindole (DAPI) solution (Southern Biotech, 0100-20) and the cells were examined under a confocal microscope (Olympus, FV1000).

Western blotting

HEK293, EC9706, 16HBE, and A549 cells transfected with plasmids expressing TMPRSS11A proteins or a control vector were cultured, as described above. When the cell culture reached confluence, the conditioned medium was collected. TMPRSS11A proteins in the conditioned medium were immunoprecipitated using the anti-V5 antibody. The cells were lysed in a buffer containing 1% (v/v) Triton X-100, 50 mm Tris-HCl (pH 8.0), 150 mm NaCl, and a protease inhibitor mixture (1:100, Roche Applied Science, 04693116001). The proteins from the conditioned medium and cell lysates were mixed with a Laemmli sample buffer (Bio-Rad) with (reducing) or without (nonreducing) β-mercaptoethanol (2.5% v/v), and analyzed by SDS-PAGE and Western blotting using a horseradish peroxidase (HRP)-conjugated anti-V5 antibody (1:5000, Thermo Fisher, R96125). After incubation with a solution with an enhanced chemiluminescent substrate (NcmECL Ultra) (NCM Biotech, P10100), Western blots were exposed to a chemiluminescent imager (Amersham Biosciences Imager 600).

Biotin labeling of cell surface proteins

To label cell surface proteins, sulfo-NHS-biotin (0.25 mg/ml, Thermo Fisher) was added to cultured cells. After 4 min on ice, a glycine solution (100 mm) was added to stop the reaction. After 30 min, the cells were lysed, as described above. The biotin-labeled proteins were precipitated with NeutrAvidin beads (Thermo Fisher, 29201) at 4 °C. After 16 h, proteins on the beads were eluted with the Laemmli buffer with or without β-mercaptoethanol (2.5%, v/v), and analyzed by SDS-PAGE and Western blotting, as described above.

PNGase F digestion

To examine N-linked glycans on TMPRSS1A, cell lysates or protein precipitates in a denaturing buffer (0.5% SDS and 40 mm DTT) were boiled for 10 min and treated with PNGase F (30 units, New England Biolabs, P0704), which removes N-linked oligosaccharides. After 3 h at 37 °C, the treated protein samples were analyzed by Western blotting using an anti-V5 antibody under reducing conditions, as described above.

Trypsin digestion of cell surface proteins

To examine proteins on HEK293 cell surface, transfected cells expressing TMPRSS11A and control cells expressing corin were treated with 0.25% (w/v) trypsin and 0.02% EDTA (w/v) (Gibco, 25200) at 37 °C for 30 s. Dulbecco's modified Eagle's medium with 10% FBS was added to neutralize trypsin activity. After washing with PBS, the cells were lysed in the lysis buffer as described above. Proteins were separated by SDS-PAGE under reducing conditions with 2.5% (v/v) β-mercaptoethanol in the Laemmli buffer. Western blotting was done using an HRP-conjugated anti-V5 antibody (1:5000, Thermo Fisher, R96125), as described above.

Brefeldin A and monensin treatment

To study the subcellular location of TMPRSS11A activation, HEK293 cells expressing TMPRSS11A and control cells expressing corin were treated with BFA (1 μm) (Sigma, 203729-1MG) or monensin (0.3 μm) (Sigma, 475897-100MG) to inhibit protein trafficking in the ER and the Golgi, respectively (34). After 20 h at 37 °C, the cells were washed and lysed. TMPRSS11A and corin proteins in the lysates were analyzed with Western blotting, as described above.

Cleavage of SARS–CoV-2 S protein extracellular fragment

To test the activity of TMPRSS11A forward SARS–CoV-2 S protein, sTMPRSS11A, sTMPRSS11A S368A (negative control), and sTMPRSS2 (positive control) were expressed in HEK293 cells. The conditioned media from the cells or vector-transfected control cells were collected and concentrated ∼20-fold with a filter device. The concentrated media were incubated with a SARS–CoV-2 S protein extracellular domain (S-ECD) fragment (residues 16-1213) produced in insect cells and with a C-terminal FLAG tag (8 μg/ml, Bioword Technology, NCP0030P). After 12 h at 37 °C, SARS–CoV-2 S-ECD and derived fragments were analyzed by Western blotting using an HRP-conjugated anti-FLAG antibody (Sigma, A8592, 1:10000).

Statistical analysis

Data were analyzed with Prism 6 software (Graphpad). Analysis of variance followed by Tukey's multiple comparison test was used to analyze data from three or more groups. p Values of <0.05 were considered to be statistically significant.

Data availability

All the data described in this study are contained within the article and accompanying supporting information.