Identification of trypsin I as a candidate for influenza A virus and Sendai virus envelope glycoprotein processing protease in rat brain

Extracellular cleavage of virus envelope fusion glycoprotein hemagglutinin (HA0) by host trypsin-like proteases is a prerequisite for the infectivity and pathogenicity of human influenza A viruses and Sendai virus. The common epidemic influenza A viruses are pneumotropic, but occasionally cause encephalopathy or encephalitis, although the HA0 processing enzyme in the brain has not been identified. In searching for the brain processing proteases, we identified a processing enzyme in rat brain that was inducible by infection with these viruses. The purified enzyme exhibited an apparent molecular mass of approximately 22 kDa on SDS-PAGE and the N-terminal amino acid sequence was consistent with that of rat pancreatic trypsin I. Its substrate specificities and inhibition profiles were the same as those of pancreatic trypsin I. In situ hybridization and immunohistochemical studies on trypsin I distribution revealed heavy deposits in the brain capillaries, particularly in the allocortex, as well as in clustered neuronal cells of the hippocampus. The purified enzyme efficiently processed the HA0 of human influenza A virus and the fusion glycoprotein precursor of Sendai virus. Our results suggest that trypsin I in the brain potentiates virus multiplication in the pathogenesis and progression of influenza-associated encephalopathy or encephalitis.     

Mast cell tryptase from pig lungs triggers infection by pneumotropic Sendai and influenza A viruses. Purification and characterization.

A novel trypsin-type serine proteinase, which processes the precursors of the envelope fusion glycoproteins of pneumotropic Sendai and human influenza A viruses, was purified to homogeneity from pig lungs. On SDS/PAGE, the purified enzyme gave a protein band corresponding to about 32 kDa, and has an apparent molecular mass of 120 kDa, as determined by gel permeation chromatography. Immunohistochemical staining with antibodies against this enzyme revealed that the enzyme is located in pig lung mast cells. The N-terminal 44-amino-acid sequence of the enzyme exhibits about 80% identity with those of mast cell tryptases from other species. Of the inhibitors tested, di-isopropyl fluorophosphate, antipain, leupeptin, benzamidine and a few proteinaceous inhibitors, such as mucus protease inhibitor and aprotinin, inhibited this enzyme activity. Heparin stabilized the enzyme, but high-ionic-strength conditions did not, unlike for human mast cell tryptase. The purified enzyme efficiently processed the fusion glycoprotein precursor of Sendai virus and slowly processed hemagglutinin of human influenza A virus, and triggered the infectivity of Sendai virus in a dose-dependent manner, although human mast cell tryptase beta and rat mast cell tryptase (rat MCP-7) from lungs did not process these fusion glycoproteins at all. These results suggest that mast cell tryptase in pig lungs is the possible trigger of the pneumotropic virus infections.     

Mini-plasmin found in the epithelial cells of bronchioles triggers infection by broad-spectrum influenza A viruses and Sendai virus.

Extracellular cleavage of virus envelope fusion glycoproteins by host cellular proteases is a prerequisite for the infectivity of mammalian and nonpathogenic avian influenza viruses, and Sendai virus. Here we report a protease present in the airway that, like tryptase Clara, can process influenza A virus haemagglutinin and Sendai virus envelope fusion glycoprotein. This protease was extracted from the membrane fraction of rat lungs, purified and then identified as a mini-plasmin. Mini-plasmin was distributed predominantly in the epithelial cells of the upward divisions of bronchioles and potentiated the replication of broad-spectrum influenza A viruses and Sendai virus, even that of the plasmin-insensitive influenza A virus strain. In comparison with plasmin, its increased hydrophobicity, leading to its higher local concentrations on membranes, and decreased molecular mass may enable mini-plasmin to gain ready access to the cleavage sites of various haemagglutinins and fusion glycoproteins after expression of these viral proteins on the cell surface. These findings suggest that mini-plasmin in the airway may play a pivotal role in the spread of viruses and their pathogenicity.     

Accumulation of mini-plasmin in the cerebral capillaries causes vascular invasion of the murine brain by a pneumotropic influenza A virus: implications for influenza-associated encephalopathy

The infectivity and pathogenicity of influenza virus are primarily determined by host cellular trypsin-type processing proteases which cleave the viral membrane fusion glycoprotein hemagglutinin (HA). Therefore the distribution of the processing protease is a major determinant of the infectious organ tropism. The common epidemic human influenza A virus is pneumotropic and the HA processing proteases tryptase Clara, mini-plasmin, tryptase TC30 and ectopic anionic trypsin have all been isolated from mammalian airways. However, the pneumotropic influenza virus occasionally causes severe brain edema, particularly in children presenting with Reye's syndrome treated with aspirin, or in children with influenza-associated encephalopathy without antipyretic treatment. We have observed that, after influenza virus infection, the accumulation of mini-plasmin in the cerebral capillaries in mice with a congenital or acquired abnormality of mitochondrial beta-oxidation mimicking the pathological findings of Reye's syndrome, causes an invasion and multiplication of the pneumotropic influenza virus at these same locations. From these findings, we hypothesize that the accumulated mini-plasmin modifies the brain capillaries from a non-permissive to a permissive state, thereby allowing multiplication of pneumotropic influenza virus. In addition, mini-plasmin proteolytically destroys the blood-brain barrier. These pathologic findings, consistent with encephalopathy in mice with a systemic impairment of beta-oxidation, may have implications for human influenza encephalopathy.     

Cellular proteinases trigger the infectivity of the influenza A and Sendai viruses.

It has been proposed that the pathogenicity of the influenza and Sendai virus is primarily determined by host cellular proteases that activate viral infectivity. We isolated trypsin-type serine proteases from rat lungs, candidates for the processing proteases of viral envelope glycoproteins, such as tryptase Clara localized in the Clara cells of the bronchial epithelium and mini-plasmin. These enzymes specifically cleave the precursor of fusion glycoprotein HA of influenza virus at Arg325, and the F0 of Sendai virus at Arg116 in the consensus cleavage motif, Gln(Glu)-X-Arg, resulting in the induction of infectivity of these viruses. Proteolytic activation of viruses by these enzymes occurs extracellularly, probably on the surface and/or in the lumen of the respiratory tract. On the other hand, we isolated two compounds from human bronchial lavage, which inhibit the activity of tryptase Clara. One was a mucus protease inhibitor and the other was a pulmonary surfactant. These compounds inhibited multiple cycles of virus replication in vitro and in vivo, but did not themselves affect the hemagglutination and the infectivity of the virus. Administration of these compounds in the airway may be useful for preventing and treating infection with influenza virus and Sendai virus.     

Tryptase Clara, an activating protease for Sendai virus in rat lungs, is involved in pneumopathogenicity.

Tryptase Clara is an arginine-specific serine protease localized exclusively in and secreted from Clara cells of the bronchial epithelium of rats (H. Kido, Y. Yokogoshi, K. Sakai, M. Tashiro, Y. Kishino, A. Fukutomi, and N. Katunuma, J. Biol. Chem. 267:13573-13579, 1992). The purified protease was shown in vitro to behave similarly to trypsin, cleaving the precursor glycoprotein F of Sendai virus at residue Arg-116 and activating viral infectivity in a dose-dependent manner. Anti-tryptase Clara antibody inhibited viral activation by the protease in vitro in lung block cultures and in vivo in infected rats. When the enzyme-specific antibody was administered intranasally to rats that were also infected intranasally with Sendai virus, activation of progeny virus in the lungs was significantly inhibited. Thus, multiple cycles of viral replication were suppressed, resulting in a reduction in lung lesions and in the mortality rate. These findings indicate that tryptase Clara is an activating protease for Sendai virus in rat lungs and is therefore involved in pulmonary pathogenicity of the virus in rats.     

A novel influenza A virus activating enzyme from porcine lung: purification and characterization

Proteolytic activation of hemagglutinin, an envelope glycoprotein of the influenza virus, by host proteases is essential for infection and proliferation of the virus. However, there is no well-defined, inherent source of host proteases in man or swine, both of which are natural hosts for human influenza viruses. We have recently isolated a 32 kDa protein in a high salt extract from porcine lungs, which possess the hemagglutinin processing activity. In this study, we attempted to purify another hemagglutinin processing enzyme from porcine lung. The purified enzyme, named tryptase TC30, exhibited a molecular mass of about 30 kDa by SDS-PAGE and 28.5 kDa by gel filtration chromatography, suggesting that it is a monomer. Tryptase TC30 cleaved peptide substrates with Arg at the P1 position, and preferentially substrates with the Ser-Ile-Gin-Ser-Arg sequence corresponding to the HA cleavage site sequence of the A/PR/8/34 influenza virus. Among various inhibitors tested, trypsin-type serine protease inhibitors, such as aprotinin, antipain, benzamidine and leupeptin, efficiently inhibited the proteolytic activity of the enzyme. The N-terminal 40 amino acid sequence of tryptase TC30 exhibits more than 60% homology to mast cell tryptases from mice MCP-6 and human tryptase-alpha and -beta. These data indicate that tryptase TC30, the 30 kDa enzyme from porcine lung, is a novel hemagglutinin-cleaving enzyme.     

Isolation and characterization of a novel trypsin-like protease found in rat bronchiolar epithelial Clara cells. A possible activator of the viral fusion glycoprotein.

A novel trypsin-like protease associated with rat bronchiolar epithelial Clara cells, named Tryptase Clara, was purified to homogeneity from rat lung by a series of standard chromatographic procedures. The enzyme has apparent molecular masses of 180 +/- 16 kDa on gel filtration and 30 +/- 1.5 kDa on sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing conditions. Its isoelectric point is pH 4.75. Studies with model peptide substrates showed that the enzyme preferentially recognizes a single arginine cleavage site, cleaving Boc-Gln-Ala-Arg-4-methylcoumaryl-7-amide most efficiently and having a pH optimum of 7.5 with this substrate. The enzyme is strongly inhibited by aprotinin, diisopropylfluorophosphate, antipain, leupeptin, and Kunitz-type soybean trypsin inhibitor, but inhibited only slightly by Bowman-Birk soybean trypsin inhibitor, benzamidine, and alpha 1-antitrypsin. Immunohistochemical studies indicated that the enzyme is located exclusively in the bronchiolar epithelial Clara cells and colocalized with surfactant. An immunoreactive protein with a molecular mass of 28.5 kDa was also detected in airway secretions by Western blotting analyses, suggesting that the 30-kDa protease in Clara cells is processed before or after its secretion. Proteolytic cleavage of the hemagglutinin of influenza virus is a prerequisite for the virus to become infectious. Tryptase Clara was shown to cleave the hemagglutinin and activate infectivity of influenza A virus in a dose-dependent way. These results suggest that the enzyme is a possible activator of inactive viral fusion glycoprotein in the respiratory tract and thus responsible for pneumopathogenicity of the virus.     

Trypsin action on the growth of Sendai virus in tissue culture cells. V. An activating enzyme for Sendai virus in the chorioallantoic fluid of the embryonated chicken egg

A trypsin-like protease which is responsible for activation of Sendai virus was found in the chorioallantoic fluid (CAF) of embryonated chicken eggs. Treatment of the inactive form of Sendai virus, grown in LLC-MK2 cells, with CAF enhanced both hemolytic activity and infectivity for the cells. Soybean trypsin inhibitor restrained the enhancing activity of CAF. These results indicate that CAF contains a trypsin-like protease which activates the inactive form of Sendai virus. The activation was strongly inhibited by phenylmethylsulfonylfluoride, ethylenediaminetetraacetate, antipain, and leupeptin but not by tosyllysylchloromethylketone, suggesting that the activating enzyme in CAF is a protease similar to but not identical with trypsin. The inactive form of the virion was produced in ovo when the seed virus was inoculated along with antipain or leupeptin. In deembryonated chicken eggs in which CAF was substituted for a culture medium, multiple cycle growth occurred, but not when soybean trypsin inhibitor was present. These observations indicate that some activating enzyme, possibly the same one as found in CAF, was secreted from the chorioallantoic membrane.     

Pneumopathogenicity in mice of a Sendai virus mutant, TSrev-58, is accompanied by in vitro activation with trypsin.

The pneumopathogenicity of a trypsin-sensitive revertant of Sendai virus, TSrev-58, which was derived from a trypsin-resistant mutant, TR-5, was examined in mice. In comparison with TR-5, the revertant had a single amino acid substitution at residue 116 (Ile----Arg) on F protein, which was the cleavage site, and had the same trypsin sensitivity as the wild-type virus. However, TSrev-58 still had a single amino acid difference from the wild-type virus at residue 109 (Asn----Asp) (M. Itoh, H. Shibuta, and M. Homma, J. Gen. Virol. 68:2939-2943, 1987). Nevertheless, the present study revealed that TSrev-58 had the same pneumopathogenicity in mice as the wild-type virus. This result indicates that the activating protease of Sendai virus present in the lungs of mice is quite similar to trypsin and also that the in vitro trypsin sensitivity of Sendai virus can be a good marker of pneumopathogenicity in mice.     

Difference of acrosomal serine protease system between mouse and other rodent sperm

A fraction of acrosomal proteins dispersed during calcium ionophore A23187‐induced acrosome reaction was prepared from cauda epididymal sperm of wild‐type and acrosin‐deficient mice, rat, and hamster. The acrosome‐reacted sperm were further extracted by Nonidet P‐40 to obtain the detergent‐soluble protein fraction. Activities of serine proteases in the two protein fractions were examined by sodium dodecyl sulfate‐polyacrylamide gel electrophoresis in the presence of gelatin. A mixture of 42‐ and 41‐kDa gelatin‐hydrolyzing proteases was found in both fractions of the wild‐type mouse sperm, whereas the acrosin‐deficient mouse sperm contained the active 42‐kDa protease and apparently lacked the activity of the 41‐kDa protease. However, exogenous bovine pancreatic trypsin compensated for the absence of acrosin in the protein fractions of the mutant mouse sperm; the gelatin‐hydrolyzing activity of the 41‐kDa protease appeared when the sperm proteins of the mutant mice were treated with pancreatic trypsin. Two‐dimensional polyacrylamide gel electrophoresis revealed that the 42‐ and 41‐kDa proteases were distinguished from acrosin by the isoelectric point and immunoreactivity with affinity‐purified antibody against an oligopeptide corresponding to the N‐terminal amino acid sequence of mouse proacrosin. Moreover, the gelatin‐hydrolyzing proteins corresponding to these two proteases were not detected in rat and hamster sperm, in spite of the treatment of the sperm extracts with pancreatic trypsin, and the total amount of gelatin‐hydrolyzing activities in mouse was much smaller than those in rat and hamster. These results may reflect the difference of the serine protease system for the sperm penetration through the egg zona pellucida between mouse and other rodent animals, possibly explaining why the acrosin‐deficient mouse sperm are capable of penetrating the zona pellucida. Dev. Genet. 25:115–122, 1999. © 1999 Wiley‐Liss, Inc.     

Sequence specificity of furin, a proprotein-processing endoprotease, for the hemagglutinin of a virulent avian influenza virus.

The virulence of avian influenza viruses correlates with the sensitivity of their hemagglutinin (HA) to cellular proteases. Furin, a proprotein-processing subtilisin-related endoprotease, is a leading candidate for the enzyme that cleaves the HA of virulent avian viruses. We therefore compared the specificity of furin with those of proteases in a variety of cultured cells and in a rat Golgi fraction, using the HA cleavage mutants of a virulent avian influenza virus, A/Turkey/Ireland/1378/85 (H5N8). The results indicated similar sequence specificities among the endoproteases when purified furin was used. In experiments with the vaccinia virus expression system, overexpressed furin cleaved mutant HAs that were not recognized by the endogenous proteases, resulting in an apparent broader specificity of furin. These findings authenticate the proposed role of furin as an HA-activating protease in vivo and caution against the use of expression vectors to study protease sequence specificity.     

Fusion and infection of influenza and Sendai viruses as modulated by dextran sulfate: a comparative study

We have directly compared the effect of two types of dextran sulfate with distinct molecular weights (500 kDa and 5 kDa) on the fusion activity and infectivity of both Sendai and influenza viruses, two lipid-enveloped viruses that differ in their routes of entry into target cells. To correlate membrane merging and infectivity MDCK cells were used as targets for the viruses in both approaches. In either case pronounced inhibition of virus–cell interactions by dextran sulfate was only observed at low pH, even though Sendai virus fuses maximally at pH 7.4. Although membrane merging could not be fully abolished, the inhibitory effect was always greater when the higher molecular weight dextran sulfate was used. The presence of this residual fusion activity, that could not be reduced even with high concentrations of agent, suggests that a limited number of binding sites for dextran sulfate may exist on the viral envelopes. The compounds also inhibited fusion of bound virions, and all results could be reproduced using erythrocyte ghosts as target membranes in the fusion assay, instead of MDCK cells. In agreement with these observations only the infectivity of influenza virus (which requires a low pH-dependent step to enter target cells) was affected by dextran sulfate, again the higher molecular weight compound showing a more pronounced inhibitory effect.     

流感病毒血凝素蛋白裂解机制的研究进展

血凝素(Hemagglutinin,HA)是流感病毒的主要表面抗原之一,诱导机体产生中和抗体,介导病毒囊膜与靶细胞膜融合,从而启动病毒对宿主细胞的感染过程。HA蛋白以前体形式合成,需经宿主蛋白酶水解为HA1、HA2两个亚单位,并以二硫键连接,病毒才获得感染性。研究表明宿主蛋白酶的分布与流感病毒感染后的致病力和组织嗜性有直接关系。潜在的裂解酶及其抑制因子的发现为流感的防治提供了新的思路,成为干预治疗的新潜在靶点。就当前国内外关于流感病毒血凝素的结构与功能、裂解机制及其应用的研究进展进行综述。     

Purification and cDNA cloning of a novel protease inhibitor secreted into culture supernatant by MDCK cells.

The infectivity of influenza viruses to host cells depends on the activation of the viral glycoprotein hemagglutinin (HA) by proteases. Starting from the observation that influenza virus replication in MDCK (Madin Darby canine kidney) cells was impaired by inactivation of trypsin in the culture fluids, we demonstrated that the inhibitory activity was resolved into two Trypsin-inactivating factors (TF), TF A (15 kDa) and TF B (11 kDa). N-terminal protein sequences of the factors revealed that TF A was a known Submandibular Protease Inhibitor (SPI) secreted in dog saliva, while TF B was a novel protein (renamed CKPI; canine kidney protease inhibitor). Following peptide mapping and protein sequencing of CKPI we obtained a 390 bp cDNA encoding a 130-amino-acid protein from MDCK cell total RNA. Protein sequence comparison showed a 63.8% identity with human secretory leukocyte protease inhibitor (SLPI), the molecule containing two conserved whey acidic protein (WAP) motifs, and we suggest that CKPI is thought to be the canine analogue of human SLPI. These results suggest that the inhibitory factors are secreted from MDCK cells, which are involved in prevention of virus replication, and applicable to the protection of host cells from virus infection.     

Estimation by radiation inactivation of the size of functional units governing Sendai and influenza virus fusion

The target sizes associated with fusion and hemolysis carried out by Sendai virus envelope glycoproteins were determined by radiation inactivation analysis. The target size for influenza virus mediated fusion with erythrocyte ghosts at pH 5.0 was also determined for comparison; a value of 57 +/- 15 kDa was found, indistinguishable from that reported previously for influenza-mediated fusion of cardiolipin liposomes [Gibson, S., Jung, C. Y., Takahashi, M., & Lenard, J. (1986) Biochemistry 25, 6264-6268]. Sendai-mediated fusion with erythrocyte ghosts at pH 7.0 was likewise inactivated exponentially with increasing radiation dose, yielding a target size of 60 +/- 6 kDa, a value consistent with the molecular weight of a single F-protein molecule. The inactivation curve for Sendai-mediated fusion with cardiolipin liposomes at pH 7.0, however, was more complex. Assuming a "multiple target-single hit" model, the target consisted of 2-3 units of ca. 60 kDa each. A similar target was seen if the liposomes contained 10% gangliosides or if the reaction was measured at pH 5.0, suggesting that fusion occurred by the same mechanism at high and low pH. A target size of 261 +/- 48 kDa was found for Sendai-induced hemolysis, in contrast with influenza, which had a more complex target size for this activity (Gibson et al., 1986). Sendai virus fusion thus occurs by different mechanisms depending upon the nature of the target membrane, since it is mediated by different functional units. Hemolysis is mediated by a functional unit different from that associated with erythrocyte ghost fusion or with cardiolipin liposome fusion.     

Cleavage of influenza A virus H1 hemagglutinin by swine respiratory bacterial proteases.

Cleavage of influenza A virus hemagglutinin (HA) is required for expression of fusion activity and virus entry into cells. Extracellular proteases are responsible for the proteolytic cleavage activation of avirulent avian and mammalian influenza viruses and contribute to pathogenicity and tissue tropism. The relative contributions of host and microbial proteases to cleavage activation in natural infection remain to be established. We examined 23 respiratory bacterial pathogens and 150 aerobic bacterial isolates cultured from the nasal cavities of pigs for proteolytic activity. No evidence of secreted proteases was found for the bacterial pathogens, including Haemophilus parasuis, Pasteurella multocida, Actinobacillus pleuropneumoniae, Bordetella bronchiseptica, and Streptococcus suis. Proteolytic bacteria were isolated from 7 of 11 swine nasal samples and included Staphylococcus chromogenes, Staphylococcus hyicus, Aeromonas caviae, Pseudomonas aeruginosa, Stenotrophomonas maltophilia, and Enterococcus sp. Only P. aeruginosa secreted a protease, elastase, that cleaved influenza virus HA. However, compared to trypsin, the site of cleavage by elastase was shifted one amino acid in the carboxy-terminal direction and resulted in inactivation of the virus. Under the conditions of this study, we identified several bacterial isolates from the respiratory tracts of pigs that secrete proteases in vitro. However, none of these proteolytic isolates demonstrated direct cleavage activation of influenza virus HA.     

Genetic Engineering Techniques: Recent Developments : Edited by P.C. Huang,T.T. Kuo and R. Wu Academic Press; New York, 1982 360 pages. $28.50

Extracts from white croaker skeletal muscle showed two alkaline proteases and a trypsin inhibitor when they were chromatographed in DEAE-Sephacel. The activity against azocasein was maximal at pH 8.5 and 9.1 for proteases I and II, respectively. Both enzymes showed optimum activity at 60° C. The molecular masses were found to be 132 kDa for protease 1,363 kDa for protease II, and 65 kDa for the inhibitor. Protease I showed the characteristics of a trypsin-like enzyme, and protease II those of a SH-enzyme. These proteins may play important roles in mechanisms of cellular proteolysis.     

Functional comparison of homologous members of three groups of Kunitz-type enzyme inhibitors from potato tubers ( Solanum tuberosum L.)

For functional studies, nine cDNAs encoding Kunitz-type enzyme inhibitors from potato tubers were expressed as GST (glutathione S transferase)-tagged fusion proteins in the fission yeast Schizosaccharomyces pombe. The inhibitors represented the three major homology groups A, B and C found in tubers. Members of the same homology group were at least 90% identical in sequence. The purified GST fusion proteins were tested for their ability to inhibit the proteases trypsin, alpha-chymotrypsin, subtilisin, papain and aspergillopepsin I, and for inhibition of the growth of fungi. Fusion proteins belonging to the same and different homology groups were found to exhibit distinct protease inhibition profiles. Removal of the GST tag by cleavage with enterokinase did not change the inhibition profile but increased the inhibitory activity. Group A and B inhibitors affected the proteases to different extents, whereas group C inhibitors showed only weak or no protease inhibition. One fusion protein completely inhibited aspergillopepsin I. One fusion protein each of groups A and B strongly inhibited mycelial growth of the fungus Fusarium moniliforme. The results suggest functional polymorphism among closely related members of the Kunitz-type inhibitor family.     

Role of proteases in the release of porcine epidemic diarrhea virus from infected cells

Porcine epidemic diarrhea virus (PEDV), a causative agent of pig diarrhea, requires a protease(s) for multicycle replication in cultured cells. However, the potential role of proteases in the infection process remains unclear. In order to explore this, we used two different approaches: we infected either Vero cells in the presence of trypsin or Vero cells that constitutively express the membrane-associated protease TMPRSS2 (Vero/TMPRSS2 cells). We found that PEDV infection was enhanced, and viruses were efficiently released into the culture fluid, from Vero cells infected in the presence of protease, while in cells without protease, the virus grew, but its release into the culture fluid was strongly hampered. Cell-to-cell fusion of PEDV-infected cells and cleavage of the spike (S) protein were observed in cells with protease. When infected Vero cells were cultured for 3 days in the absence of trypsin but were then treated transiently with trypsin, infectious viruses were immediately released from infected cells. In addition, treatment of infected Vero/TMPRSS2 cells with the protease inhibitor leupeptin strongly blocked the release of virus into the culture fluid. Under electron microscopy, PEDV-infected Vero cells, as well as PEDV-infected Vero/TMPRSS2 cells treated with leupeptin, retained huge clusters of virions on their surfaces, while such clusters were rarely seen in the presence of trypsin and the absence of leupeptin in Vero and Vero/TMPRSS2 cells, respectively. Thus, the present study indicates that proteases play an important role in the release of PEDV virions clustered on cells after replication occurs. This unique observation in coronavirus infection suggests that the actions of proteases are reminiscent of that of the influenza virus neuraminidase protein.