The influenza virus HA is synthesized as an inactive precursor that must be cleaved by host proteases to enable fusion with the host endosome. While certain proteases have been identified, uncertainty surrounding which host proteases are responsible for HA cleavage in the human respiratory tract remains. The membrane-bound proteases TMPRSS2, TMPRSS4, and HAT are some of the more recent examples of respiratory tract-localized proteases that have been shown to be capable of cleaving HA in vitro
or in cell culture models (3
). To date, the extracellular proteases known to cleave HA, such as plasmin, urokinase, and mast cell tryptase, generally rely on in vitro
Here we identify matriptase as a human respiratory tract-expressed protease involved in HA cleavage activation. Matriptase has characteristics that are common to each of the HA-cleaving proteases currently identified, being a trypsin-like serine protease that is specific for cleavage after an arginine residue and is highly expressed in the respiratory tract. An interesting feature of matriptase is that the protease is found both on the plasma membrane and in the extracellular space of the respiratory tract. In the case of TMPRSS2 and HAT, however, the majority of the protein was found to exclusively reside on the plasma membrane in vitro
, and furthermore, the TMPRSS2 and HAT that were secreted were shown to be unable to cleave HA (6
). We have determined that secreted matriptase is present in the nasal wash of human subjects; however, it is difficult to compare the abundance of this secreted protease to that of the other proteases relevant for influenza virus cleavage, which are transmembrane proteases and so are not present in a nasal wash. Although it is active for only a short time in vivo
), matriptase does display high catalytic activity compared to other proteases examined by our laboratory (). However, determining if the matriptase identified in the nasal wash samples is in an active form poses a significant challenge (29
It has been a long-standing belief that influenza virus activation is mediated in a paracrine fashion by a secreted, extracellular protease. However, the identification of such a secreted protease in the human respiratory tract has proved elusive. Instead, attention has recently focused on transmembrane proteases known to be expressed in the human respiratory tract (i.e., TMPRSS2, TMPRSS4, and HAT). Here, we provide the first identification of a secreted protease normally present in the human respiratory tract that can cleave and activate influenza virus. Overall, it is likely that both secreted and transmembrane proteases have the ability to activate influenza virus in humans and must be considered candidates for therapeutic development. Within this spectrum of proteases, it should also be considered which proteases are most important in a given situation, such as the localization within the respiratory tract, between individuals, or with individuals with underlying conditions that might affect protease expression, activity, or distribution.
One notable feature of our studies is that matriptase cleavage was limited to the H1 influenza virus subtype, whereas no cleavage was observed with the H2 and H3 subtypes. This finding is distinct from that for other proteases identified in humans, where it is assumed that all human-adapted subtypes are cleaved in an equivalent manner. Our finding that cleavage by matriptase showed a preference for particular strains within the H1 subtype raises the question of whether each human-adapted virus has evolved to be activated by a particular set of host proteases that is distinct from the set used to activate the other subtypes and, furthermore, whether each strain from a given subtype has a preference for particular proteases within that set. One virus of particular interest in this regard is A/WSN/33, which is unusual among H1 viruses in having a bulky aromatic residue (tyrosine) directly N terminal to the cleavage site arginine in the P-2 cleavage site position in place of the highly conserved serine. This change at the P-2 position (S-Y) appears to not be suitable for binding to the matriptase active site and could offer an explanation for the limited cleavage observed for matriptase treatment of the A/WSN/33 HA. Recent data from our laboratory have demonstrated that cleavage of the A/WSN/33 HA by plasmin was enhanced by the P-2 tyrosine substitution, supporting that idea that residues flanking the cleavage site P-1 arginine in the HA cleavage site region can affect proteolytic cleavage (31
). In agreement with HA cleavage data, matriptase efficiently cleaved the peptide mimic of the H1 cleavage site consensus, and a marked decrease in the rate of cleavage was observed with the A/WSN/33 cleavage site peptide mimic compared to that for the H1 consensus peptide (). However, matriptase cleavage was also low for other H1 strains that retain the consensus P-2 serine residue, i.e., A/SC/18 and A/WS/33, and so features of the HA in addition to the P-2 residue cleavage site flanking region must also be involved. Matriptase cleavage of the H2 peptide mimic also resulted in an expected low rate of cleavage compared to that for the H1 consensus (). However, in the case of the H3 HA cleavage site peptide mimic, with matriptase treatment, efficient cleavage that was comparable to that for the H1 consensus peptide was observed. This result is intriguing, since virtually no cleavage was observed by Western blot analysis of each H3 HA examined. In this case, other factors, such as HA glycosylation or secondary structure, must also contribute to the inability of matriptase to cleave the H3 subtype. Overall, the features that allow matriptase cleavage activation of selected H1 viruses remain to be explored.
To conclude, with uncertainty over the host proteases involved in HA cleavage in vivo still remaining, it is important to investigate the diversity of host proteases for their ability to cleave HA, in attempts to determine the critical host protease(s) involved in influenza virus activation in the respiratory tract. Here we show that matriptase cleaved and activated HA from human-adapted, influenza virus subtypes. Importantly, our studies reveal both a subtype and strain specificity for cleavage and fusion activation. Our studies highlight the role of matriptase as a secreted, extracellular protease important for influenza virus infection, alongside transmembrane proteases such as TMPRSS2 and HAT. To our knowledge, matriptase (ST14) is the first example of a member within the matriptase TTSP subfamily having the ability to cleave HA. With renewed effort to develop alternative influenza therapeutics, an improved understanding of the spectrum of proteases critical for activation of the virus in vivo will be essential for the development of effective, protease-based therapeutics. Our finding that matriptase may have both direct and indirect effects via activation of other serum and tissue proteases argues in favor of the development of broad-spectrum protease inhibitors.