Kallistatin Ameliorates Influenza Virus Pathogenesis by Inhibition of Kallikrein-Related Peptidase 1-Mediated Cleavage of Viral Hemagglutinin

Abstract

Proteolytic cleavage of the hemagglutinin (HA) of influenza virus by host trypsin-like proteases is required for viral infectivity. Some serine proteases are capable of cleaving influenza virus HA, whereas some serine protease inhibitors (serpins) inhibit the HA cleavage in various cell types. Kallikrein-related peptidase 1 (KLK1, also known as tissue kallikrein) is a widely distributed serine protease. Kallistatin, a serpin synthesized mainly in the liver and rapidly secreted into the circulation, forms complexes with KLK1 and inhibits its activity. Here, we investigated the roles of KLK1 and kallistatin in influenza virus infection. We show that the levels of KLK1 increased, whereas those of kallistatin decreased, in the lungs of mice during influenza virus infection. KLK1 cleaved H1, H2, and H3 HA molecules and consequently enhanced viral production. In contrast, kallistatin inhibited KLK1-mediated HA cleavage and reduced viral production. Cells transduced with the kallistatin gene secreted kallistatin extracellularly, which rendered them more resistant to influenza virus infection. Furthermore, lentivirus-mediated kallistatin gene delivery protected mice against lethal influenza virus challenge by reducing the viral load, inflammation, and injury in the lung. Taking the data together, we determined that KLK1 and kallistatin contribute to the pathogenesis of influenza virus by affecting the cleavage of the HA peptide and inflammatory responses. This study provides a proof of principle for the potential therapeutic application of kallistatin or other KLK1 inhibitors for influenza. Since proteolytic activation also enhances the infectivity of some other viruses, kallistatin and other kallikrein inhibitors may be explored as antiviral agents against these viruses.

INTRODUCTION

Influenza is an important acute respiratory disease in humans and animals. Influenza epidemics and pandemics are constant threats to human health. Factors implicated in the high morbidity and mortality of influenza virus infection include robust cytokine production (cytokine storm), excessive inflammatory infiltrates, and virus-induced lung tissue destruction. Although pulmonary inflammatory responses may facilitate virus clearance, they often cause severe lung injury. Interactions of influenza virus and host factors are crucial for virus replication. Therefore, understanding the interplay between viral and host factors may provide new targets for antiviral therapy.

Hemagglutinin (HA) present on the envelope of influenza virus is an important determinant for viral virulence and pathogenesis. Infection by influenza virus is initiated by cleavage of HA0 into HA1 and HA2 subunits and leads to the exposure of the hydrophobic N terminus of HA2, the fusion peptide, which facilitates fusion of the viral envelope with the host endosomal membrane (1, 2). Proteolytic cleavage of the precursor HA molecule by host trypsin-like proteases is required for viral infectivity (3). The HA cleavage site is an external loop that links HA1 and HA2. This loop has been shown to be cleaved by host serine proteases in cell culture, such as thrombin (4), plasmin (5), blood-clotting factor Xa (6), acrosin (7), mini-plasmin (8), protease from human respiratory lavage (9), transmembrane protease serine S1 member 2 (TMPRSS2) (10), and human airway trypsin-like protease (HAT) (11). However, the specific proteases responsible for HA cleavage in the human respiratory tract are still unclear. In contrast, serine protease inhibitors (serpins) have been shown to inhibit the cleavage of HA and thus to suppress influenza virus activation (12). They include aprotinin (13), leupeptin (14), pulmonary surfactant (15), an inhibitor of plasminogen activation (16), and human mucus protease inhibitor (17). Aprotinin has therapeutic effects against influenza in humans and mice by suppressing both HA cleavage and inflammation (18, 19).

Human kallikreins are serine proteases that comprise kallikrein-related peptidases (KLKs) and plasma kallikrein (also known as KLKB1). KLKs are composed of 15 members, KLK1 to KLK15, and are involved in proteolytic cascades implicated in pathophysiological processes (20, 21). KLK1 (also known as tissue kallikrein) is responsible for cleavage of kininogen to release kinins (bradykinin-related peptides), which mediate the regulation of smooth muscle contraction, vascular permeability, vascular cell growth, inflammatory cascades, electrolyte balance, neutrophil chemotaxis, and pain induction (22–24). KLK1, which is synthesized as the proenzyme prekallikrein and activated by plasmin or plasma kallikrein, is ubiquitously expressed, and the KLK1-kinin system exerts various biological activities in the cardiovascular, renal, central nervous, and immune systems (22). In addition, KLK1 is expressed in activated resident epithelial cells, alveolar macrophages, and recruited inflammatory cells, such as neutrophils (22).

Kallistatin, first identified in human plasma as a KLK1 binding protein and a specific KLK1 inhibitor (25, 26), can be detected in various tissues, cells, and body fluids (26, 27). Kallistatin has a unique P1 phenylalanine (residue 388), which confers an excellent inhibitory specificity toward KLK1 and inhibits its activity (28). Kallistatin also exhibits pleiotropic effects independently of the KLK1-kinin system, such as reduction of blood pressure, vasodilatation, and inhibition of inflammation, angiogenesis, and tumor growth (29–34). However, the roles of the kallikrein-kinin system and kallistatin in viral infection remain largely unknown. Among the 15 different KLKs, physiological functions have been relatively well established for KLK1, KLK2, and KLK3. KLK1 is ubiquitously expressed, while KLK2 and KLK3 are expressed only in the prostate. The physiological roles of the other KLKs remain largely unknown. KLK1 is the primary kinin-generating enzyme in the lung (35). KLK1 and kinin levels were elevated in the respiratory tracts of patients with asthma, rhinitis, bronchitis, and pneumonia (36, 37). Moreover, experimental rhinovirus infection in individuals with asthma increased KLK1 activation, accompanied by increased interleukin 8 (IL-8) expression (38). In inflammatory responses, the widespread distribution of KLK1 enhances its potential for involvement in kinin generation at multiple sites. KLK1 is moderately expressed in the lung and trachea and also exists in ductal glands, including nasal glands, salivary glands, and small glands in the aerodigestive tracts (35). In addition, KLK1 is expressed in activated resident epithelial cells, alveolar macrophages, and recruited inflammatory cells, such as neutrophils (22). Since KLK1 is involved in many airway diseases, including asthma, nasal allergy, and chronic bronchitis, here, we chose to study the impacts of KLK1 on influenza virus infection. As kallistatin is the naturally occurring KLK1 inhibitor and is ubiquitously expressed, including in the human respiratory system (26, 39), it may also play a role in influenza virus infection, either KLK1 dependent or independent.

The aim of this study was to investigate the potential roles of KLK1 and kallistatin in influenza virus infection. We tested whether KLK1 can cleave and activate HA to promote viral infection and whether kallistatin can inhibit viral production and ameliorate influenza virus pathogenesis. Our results indicate that KLK1, which was upregulated during influenza virus infection, could cleave H1, H2, and H3 HA molecules and enhance viral replication in cultured cells. Such effects could be abrogated by kallistatin treatment. Furthermore, lentivirus-mediated kallistatin gene delivery conferred protection against influenza by inhibiting HA cleavage and alleviating lung inflammation. Collectively, our findings provide insights into the interplay between influenza virus and the host kallikrein/kallistatin system. Our results also imply a potential therapeutic use of kallistatin or other KLK1 inhibitors for influenza.

MATERIALS AND METHODS

Cells, viruses, proteins, and mice.

MDCK cells were obtained from the American Type Culture Collection. The MDCK cells were cultured in Dulbecco's modified Eagle's minimal essential medium (DMEM) supplemented with 10% fetal bovine serum (HyClone, Logan, UT, USA), 2 mM l-glutamine, and gentamicin (50 μg/ml) at 37°C in 5% CO₂. The HBE135-E6E7 human bronchial epithelial cell line, obtained from P. L. Kuo, which was originally from the American Type Culture Collection (40), was cultured in keratinocyte serum-free medium (Invitrogen, Carlsbad, CA, USA) containing 5 ng/ml human recombinant epidermal growth factor and 50 ng/ml bovine pituitary extract and supplemented with 5 ng/ml insulin and 500 ng/ml hydrocortisone. Influenza A viruses were propagated and titrated as described previously (41). To produce stocks of influenza A/WSN/33 (H1N1) and A/Taiwan/N2723/06 (H3N2) viruses containing uncleaved HA0, MDCK cells were infected with influenza virus at a multiplicity of infection (MOI) of 1 in serum-free DMEM containing trypsin (1 μg/ml), which is capable of cleavage activation of HA, for 4 h and washed three times with phosphate-buffered saline (PBS), and the cultures were replenished with serum-free DMEM in the absence of trypsin for 24 h. The supernatants of cell cultures were cleared by centrifugation, aliquoted, and stored at −80°C as virus stocks until use. HA can be cleaved by plasmin (42), which is converted from plasminogen by plasminogen activators present in calf serum (43). To rule out the possibility of having serum plasminogen from the culture medium, we washed the MDCK cells extensively with PBS and cultured the cells with serum-free medium in the absence of trypsin in all the in vitro experiments. The lentiviral vector pSin-null was derived from pSin-EF2-Oct4-Pur (Addgene plasmid 16570; Addgene, Cambridge, MA, USA); the coding region of Oct4 was removed by digestion with SpeI and BamHI, and the resulting large fragment of the plasmid was filled in with T4 DNA polymerase and subsequently self-ligated by T4 DNA ligase. The recombinant lentiviruses LV-KS, LV-GFP, and LV-Null, which encode human kallistatin or green fluorescent protein (GFP) or carry no transgene, were produced by the transient transfection of 293T cells with pWPXL-Kallistatin, pWPXL, and pSin-null, respectively, along with the packaging construct psPAX2 and the vesicular stomatitis virus G protein (VSV-G) expression plasmid pMD2G (33). All work on influenza viruses and recombinant lentiviral vectors was carried out in biosafety level 2 laboratories.

Recombinant rat KLK1 proteins were produced and purified as previously described (44). Recombinant human kallistatin proteins were purchased from Abcam (Cambridge, United Kingdom). Female 5- to 6-week-old C57BL/6 mice were obtained from the Laboratory Animal Center of National Cheng Kung University. The experimental protocols adhered to the rules of the Animal Protection Act of Taiwan and were approved by the Animal Care and Use Committee of National Cheng Kung University (IACUC number 104088).

Analysis of HA cleavage.

Recombinant HA proteins (100 ng in 20 μl saline) derived from influenza A/WSN/33 (H1N1), A/California/04/09 (H1N1), A/Japan/305/57 (H2N2), and A/Brisbane/10/07 (H3N2) viruses were purchased from Sino Biological Inc. (Beijing, China). They were incubated with KLK1 at concentrations ranging from 0.1 to 1 μg/ml for 3 or 4 h at 37°C. To detect cleavage of influenza A virus per se by KLK1, influenza A/WSN/33 and A/Taiwan/N2723/06 viruses (2 × 10⁵ PFU in 5 μl serum-free DMEM) were incubated with KLK1 at a concentration of 2.5, 5, or 15 μg/ml for 4 h at 37°C. After incubation, the patterns of HA cleavage were analyzed by immunoblot analysis with antibodies against different subtypes of HA.

To investigate whether kallistatin could inhibit KLK1-mediated HA cleavage, recombinant HA proteins (100 ng in 20 μl saline) derived from influenza A/WSN/33, A/California/04/09, and A/Japan/305/57 viruses were incubated with kallistatin (12.5, 25, or 50 μg/ml) and/or KLK1 (0.5 or 1 μg/ml), and influenza A/WSN/33 and A/Taiwan/N2723/06 virus particles (2 × 10⁵ PFU in 5 μl serum-free DMEM) were incubated with kallistatin (6, 12, or 24 μg/ml) and/or KLK1 (2.5 μg/ml) for 4 h at 37°C. After incubation, the patterns of HA cleavage were analyzed by immunoblot analysis with antibodies against different subtypes of HA.

Immunoblot analysis.

Immunoblot analysis was performed to detect KLK1, kallistatin, and influenza virus HA using standard methods. The primary antibodies used for immunoblotting included mouse monoclonal anti-prekallikrein/kallikrein antibody (Abcam), mouse anti-human kallistatin monoclonal antibody (ascitic fluid) (44), goat anti-influenza A virus (H1N1) antiserum (Virostat, Portland, ME, USA), mouse anti-influenza A H1N1 virus HA monoclonal antibody (C-102; Santa Cruz Biotechnology, Dallas, TX, USA), rabbit anti-influenza A virus H2N2 HA antibody (Sino Biological Inc.), rabbit anti-influenza A virus H3N2 HA antibody (Sino Biological Inc.), and mouse monoclonal anti-β-actin peroxidase antibody (Sigma-Aldrich, St. Louis, MO, USA). Horseradish peroxidase-conjugated donkey anti-goat, goat anti-mouse, or goat anti-rabbit IgG (Jackson, West Grove, PA, USA) was used as the secondary antibody where appropriate. Protein-antibody complexes were detected by the ECL system (Millipore, Bedford, MA, USA) and visualized with the BioSpectrum imaging system (UVP Inc., Upland, CA, USA) or Kodak X-AR film (Eastman Kodak Co., Rochester, NY, USA). The relative intensities of protein bands were quantified using the public-domain image analysis software package ImageJ (U.S. National Institutes of Health) or the UVP BioSpectrum Imaging System.

RT-PCR analysis.

Total RNA was isolated with the TRIzol reagent (Invitrogen). A Reverse-It first-strand synthesis kit (Applied Biosystems, Foster City, CA, USA) was used for cDNA synthesis. The following primers were used for reverse transcription (RT)-PCR analysis: mouse KLK1, 5′-AGATGTTGTGAAGCCCATCGACCT-3′ (forward) and 5′-GGGCTTTGGCACAGTCCTCATTAC-3′ (reverse); mouse kallistatin, 5′-CCTGACAACACATCCAACCAAAC-3′ (forward) and 5′-TCTTGAATGGACCTGAAACCCTC-3′ (reverse); and mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-GTTGTCTCCTGCGACTTCAACA-3′ (forward) and 5′-TTGCTGTAGCCGTATTCATTGTC-3′ (reverse). The PCR conditions for KLK1 and GAPDH were 28 cycles of 60 s at 94°C, 45 s at 56°C, and 30 s at 72°C. The PCR conditions for kallistatin were 35 cycles of 60 s at 94°C, 45 s at 56°C, and 30 s at 72°C. The GAPDH gene was used as an internal control gene to normalize the expression of genes of interest. The relative intensities of the bands were quantified using the UVP BioSpectrum Imaging System.

Cytotoxicity and plaque assays.

Confluent MDCK cells (5 × 10⁴) cultured in 48-well plates in triplicate were infected with influenza virus in the presence or absence of KLK1 and kallistatin for 48 or 72 h. Moreover, MDCK cells (8 × 10⁶) grown in 10-cm dishes were infected with LV-KS or LV-Null at an MOI of 5 or left uninfected for 24 h in DMEM containing 2% cosmic calf serum (HyClone, Logan, UT, USA). These cells were then seeded at 2.5 × 10⁴ cells/well into 48-well plates and cultured in 0.3 ml DMEM containing 10% cosmic calf serum for 24 h. The cultures were replenished with serum-free DMEM for an additional 24 h. Subsequently, the cells were infected with influenza A/WSN/33 virus at an MOI of 0.05 or mock infected in the presence or absence of KLK1 (2 μg/ml) in serum-free DMEM and cultured for 48 h. Viable cells were stained with crystal violet, the wells were scanned, and the crystal violet staining was quantified using ImageJ software. Titers of influenza virus were quantified by plaque assay. Briefly, confluent MDCK cells (4 × 10⁵/well) grown in 12-well plates were incubated with 0.5 ml of 10-fold serially diluted virus samples in serum-free DMEM at 37°C in 5% CO₂. After 2 h, the cells were washed with PBS and replenished with 1 ml of DMEM containing 2% cosmic calf serum, trypsin (1 μg/ml), and 0.5% agarose (Lonza, Rockland, ME, USA) for 48 h. The cells were then fixed with 0.1% crystal violet containing formalin for 24 h, and the plaques were counted. Alternatively, viral titers were determined by the 50% tissue culture infective dose (TCID₅₀) method.

Animal studies.

To determine the levels of kallistatin and KLK1 in mice following influenza virus infection, C57BL/6 mice were intranasally inoculated with 10⁷ PFU of influenza A/WSN/33 virus or saline, and their lung tissues were examined for kallistatin and KLK1 expression by immunoblot and RT-PCR analyses at days 3 and 6 postinfection (p.i.). In kallistatin gene delivery, C57BL/6 mice were intranasally inoculated with 10⁷ transduction units (TU) of LV-KS or LV-GFP or with saline for three consecutive days. One day after treatment with lentiviral vectors, mice were intranasally inoculated with 7 × 10⁶ PFU of influenza A/WSN/33 virus. The mice were monitored daily for illness and death for 13 days after viral infection. In a similar experiment, mice that had been treated with LV-KS or LV-GFP were challenged with influenza A virus (5 × 10⁶ PFU) and sacrificed at day 6 p.i. for histological analysis.

Histological analysis.

Mouse lungs were prepared and fixed in 4% buffered formalin for 48 h, dehydrated in graded alcohol, embedded in paraffin, and sectioned. The 5-μm sections were mounted onto glass slides and stained with hematoxylin and eosin (H&E) (Dako, Carpinteria, CA, USA). Inflammatory changes on the basis of numbers of inflammatory cells and tissue damage in the lungs were examined histologically from H&E-stained longitudinal cross sections and graded as no change (score = 0), mild (score = 1), moderate (score = 2), or severe (score = 3), as previously described (41, 45). The sections were reviewed and scored in a blind fashion by a pathologist (C.-C.L.).

ELISA and nitric oxide (NO) detection.

Confluent HBE135-E6E7 cells (5 × 10⁴ cells/well) grown in 48-well plates were infected with 1.5 × 10⁴, 3 × 10⁴, and 6 × 10⁴ PFU of influenza A/WSN/33 virus, equivalent to MOI of 2.5, 5, and 10, respectively. After 24 h, the conditioned medium was assessed for kallistatin and KLK1 contents by enzyme-linked immunosorbent assay (ELISA) as described previously (27, 46). The detection ranges for human kallistatin and KLK1 by ELISA are 0.4 to 25 ng/ml. Bronchoalveolar lavage (BAL) was performed by two intratracheal injections of 1 ml of saline into the alveolar space of influenza virus-infected mice at day 3 p.i., and BAL fluid (total, 2 ml) was collected by gentle suction. It was centrifuged, and the cell-free supernatant was collected and stored at −80°C for subsequent analysis.

The levels of mouse IL-1β, tumor necrosis factor alpha (TNF-α), IL-6, gamma interferon (IFN-γ), monocyte chemotactic protein 1 (MCP-1), and macrophage inflammatory protein 1α (MIP-1α) in the BAL fluid or serum were detected with ELISA kits (R&D Systems, Minneapolis, MN, USA). Serum NO levels were estimated by measuring NO₂⁻ accumulation using the Griess reagent.

Statistical analysis.

Data are expressed as means and standard deviations (SD). Differences in the efficiency of HA cleavage were compared by two-way analysis of variance (ANOVA) with repeated measures. The survival analysis was performed using the Kaplan-Meier survival curve and log-rank test. The remaining data were analyzed by one-way ANOVA with Bonferroni post hoc test or Student's t test. The differences were considered significant if P values were <0.05. Statistical tests were performed using GraphPad Prism (version 6.0; GraphPad Software, San Diego, CA, USA).

RESULTS

KLK1 expression is increased, whereas kallistatin expression is reduced, in mice during acute influenza virus infection.

To investigate whether changes in KLK1 and kallistatin levels occur during acute influenza virus infection, we determined protein and mRNA levels of KLK1 and kallistatin in the lungs of mice intranasally infected with influenza A/WSN/33 (H1N1) virus. Immunoblot analysis showed that KLK1 was upregulated, whereas expression levels of kallistatin remained similar in the lungs of mice at days 3 and 6 p.i. (Fig. 1A and B). RT-PCR analysis revealed that expression of KLK1 mRNA progressively increased over time, whereas kallistatin mRNA levels markedly decreased at day 6 p.i. in the lungs of mice infected with influenza virus (Fig. 1C and D). Figure 1E shows that human bronchial epithelial (HBE135-E6E7) cells, when infected with influenza virus at an MOI of 10, significantly secreted smaller amounts of kallistatin than mock-infected cells. However, the KLK1 contents were below the detection limit of the assay (0.4 ng/ml) in the supernatant of either mock-infected or infected cells under our ELISA conditions (data not shown). Collectively, these results indicate that both mRNA and protein levels of KLK1 are increased, whereas kallistatin mRNA levels are decreased, in the lungs of influenza virus-infected mice.

FIG 1.

KLK1 expression is increased, whereas kallistatin expression is reduced, during acute influenza virus infection. (A to C) C57BL/6 mice were intranasally inoculated with 10⁷ PFU of influenza A/WSN/33 (H1N1) virus or saline, and lung tissues were collected for immunoblot and RT-PCR analysis at days 3 and 6 p.i. (A) Detection of protein levels of KLK1 and kallistatin (KS). Each lane represents samples from individual mouse lungs. (B) Expression levels of KLK1 and kallistatin proteins after normalization to β-actin levels. The ratios between the intensities of the bands corresponding to KLK1 or kallistatin and those corresponding to β-actin were calculated. The values shown are means and SD (n = 4 to 8). The results are representative of three independent experiments. (C) Detection of mRNA levels of KLK1 and kallistatin. Each lane represents samples from individual mouse lungs. (D) Expression levels of KLK1 and kallistatin mRNAs after normalization to GAPDH levels. The ratios between the intensities of the bands corresponding to KLK1 or kallistatin and those corresponding to GAPDH were calculated. The values shown are means and SD (n = 5; P = 0.003; one-way ANOVA). N.D., not done. The results are representative of three independent experiments. (E) Detection of kallistatin secreted from human bronchial epithelial cells after influenza virus infection. Confluent HBE135-E6E7 cells (10⁵ cells/well) grown in 48-well plates were infected with influenza A/WSN/33 virus at the indicated MOIs for 24 h, and the conditioned medium was assessed for kallistatin amounts by ELISA (n = 3; P = 0.0029 by one-way ANOVA). The results are representative of two independent experiments.

KLK1 cleaves HA of dominant human subtypes (H1, H2, and H3) of influenza virus in the contexts of HA proteins and virus particles.

Given that KLK1 is a serine protease and is upregulated in the lungs of mice during influenza virus infection, we investigated whether KLK1 is capable of cleaving H1, H2, and H3 HA molecules in the context of purified proteins. Figure 2A shows that KLK1 dose-dependently cleaved recombinant HA proteins of influenza A/WSN/33 (H1N1), A/California/04/09 (H1N1), A/Japan/305/57 (H2N2), and A/Brisbane/10/07 (H3N2) viruses into HA1 and HA2 subunits. Furthermore, the cleavage efficiencies appeared to be different among various strains of influenza A virus. To compare the cleavage efficiencies of KLK1, we used 1 μg/ml of KLK1 to cleave recombinant HA proteins derived from different subtypes for 3 or 4 h and analyzed the cleavage profiles by immunoblotting with the respective anti-HA antibodies. As shown in Fig. 2B, KLK1 efficiently cleaved H1 and H2 HA molecules from three viral strains tested. However, H3 HA molecules from influenza A/Brisbane/10/07 virus were cleaved at a very low level. We determined the percentage of HA cleavage by densitometric analysis of the immunoblots shown in Fig. 2B. Quantitative analysis revealed that KLK1 cleaved the HA proteins of influenza A/WSN/33 and A/California/04/09 viruses more efficiently than the HA proteins of influenza A/Brisbane/10/07 virus (Fig. 2C). We further examined whether KLK1 can cleave HA on the surfaces of virus particles. Figure 2D shows that treatment with KLK1 increased proteolytic cleavage of viral HA on the surface of influenza A/WSN/33 virus in a dose-dependent manner. However, the cleavage of viral HA on the surface of influenza A/Taiwan/N2723/06 (H3N2) virus by KLK1 was less efficient. Compared to the cleavage of purified HA proteins, higher concentrations of KLK1 were required to cleave HA in the context of whole virus particles. Taken together, these results indicate that KLK1 can cleave the HA of dominant human subtypes (H1, H2, and H3) of influenza A virus.

FIG 2.

KLK1 cleaves HA of human subtypes of influenza A virus with different efficiencies. (A) Cleavage of H1, H2, and H3 HA molecules by KLK1, ranging from 0.1 to 1 μg/ml for 4 h. Recombinant HA proteins derived from influenza A/WSN/33 (H1N1), A/California/04/09 (H1N1), A/Japan/305/57 (H2N2), and A/Brisbane/10/07 (H3N2) viruses (100 ng in 20 μl saline) were incubated with KLK1 for 4 h at 37°C. The cleavage patterns of the HA proteins were analyzed by immunoblotting. To facilitate visualization of all the HA bands in the blots, films with multiple exposures were analyzed to ensure that the signals for HA1 and HA2 could be detected. The exposure time was longer for HA1 and HA2 bands than for HA0 bands to increase sensitivity. The bands were quantified by densitometric analysis, and the ratios (HA1 plus HA2 divided by HA1 plus HA2 plus HA0) representing the HA cleavage efficiency are shown at the bottom of the blots. The results are representative of three independent experiments. (B) Cleavage of H1, H2, and H3 HA molecules by KLK1 (1 μg/ml) for 3 or 4 h. Note that the lanes shown for influenza A/California/04/09 and A/Japan/305/57 virus HA proteins were cut from the same individual blot. To facilitate comparison of the cleavage efficiencies of different HA subtypes, the individual exposure times for HA0, HA1, or HA2 among HA proteins derived from different influenza A virus strains were identical. Since H1 subtypes were cleaved more quickly and easily by KLK1 than H2 and H3 subtypes, HA1 of the H1 subtype was partially degraded, resulting in weaker staining of HA1 than HA2. However, this seemingly unusual result did not occur when 0.5 μg/ml or less KLK1 was used. The results are representative of three independent experiments. (C) Efficiency of HA cleavage of KLK1. The percentage of HA cleavage was determined by densitometry of the immunoblot analysis shown in panel B. The values shown are the means and SD from three independent experiments. (D) Cleavage of influenza A virus by KLK1. Influenza A/WSN/33 (H1N1) and A/Taiwan/N2723/06 (H3N2) viruses (2 × 10⁵ PFU in 5 μl serum-free medium) were incubated with various concentrations of KLK1 or 0.25 μg/ml of trypsin (serving as the positive control) for 4 h at 37°C. In the H1N1 subtype, HA1 was stained more weakly than HA2 when 5 or 15 μg/ml of KLK1 was used, which resulted from partial degradation of HA1. The bands were quantified as described for panel A. The results are representative of three independent experiments.

Treatment with KLK1 enhances influenza virus production in vitro.

To further study whether the cleavage of viral HA by KLK1 correlates with the extent of the cytopathic effect (CPE) in the infected cells, MDCK cells were infected with either influenza A/WSN/33 or A/Taiwan/N2723/06 virus and treated with various concentrations of KLK1, followed by examination of CPE with crystal violet staining at day 3 p.i. The culture wells were scanned, and crystal violet staining was quantified with image analysis software. As shown in Fig. 3A, treatment with KLK1 enhanced virus-induced CPE (left) and resulted in the reduction of viable cells stained by crystal violet (right) in cells infected with influenza A/WSN/33 virus. Accordingly, cells treated with this strain of virus in the presence of KLK1 produced higher viral yields than those without KLK1 treatment (Fig. 3B). In the case of influenza A/Taiwan/N2723/06 virus, a weak enhancement of virus-induced CPE was seen only at much higher KLK1 doses (Fig. 3C), which resulted in higher viral production in the infected cells (Fig. 3D). These data are consistent with those in Fig. 2, showing that KLK1 was more efficient in cleaving the H1 HA molecule of influenza A/WSN/33 virus than the H3 HA molecules of influenza A/Taiwan/N2723/06 virus. Taken together, these results indicate that cells infected with influenza virus that is susceptible to HA cleavage in the presence of KLK1 produce higher yields of influenza virus than with those without KLK1 treatment.

FIG 3.

KLK1 enhances influenza virus production. (A) CPE in cells infected with influenza A/WSN/33 (H1N1) virus in the presence of KLK1. MDCK cells were infected with influenza A/WSN/33 virus at an MOI of 5 or mock infected in the presence of various concentrations of KLK1 for 72 h. (Left) The cells were stained with crystal violet to monitor the CPE. (Right) The wells were scanned, and crystal violet staining was quantified to determine cell survival. The values, which are shown as relative intensity levels, with the levels in the mock-infected cells arbitrarily set to 100, represent means and SD (n = 8; P = 0.0001; one-way ANOVA). The results are representative of three independent experiments. (B) Viral yield produced from influenza A/WSN/33 virus-infected cells in the presence of KLK1 (12 μg/ml) for 72 h. MDCK cells were infected with influenza A/WSN/33 virus at an MOI of 5 in the presence or absence of KLK1 (12 μg/ml) for 72 h. The viral titers in the culture supernatants were quantified by plaque assay. Each symbol represents an individual value; the horizontal bars represent means ± SD (n = 8). The results are representative of two independent experiments. (C and D) CPE (C) and viral yield (D) in cells infected with influenza A/Taiwan/N2723/06 (H3N2) virus in the presence of KLK1. The experiments were performed as described for panels A and B, except that influenza A/Taiwan/N2723/06 (H3N2) virus was used. The values shown are means and SD (n = 3; P = 0.0045; one-way ANOVA).

Exogenously added kallistatin inhibits HA cleavage and reduces influenza virus production in vitro.

As KLK1 can enhance influenza virus production by the cleavage activation of HA, we next determined whether kallistatin, a naturally occurring KLK1 inhibitor, can inhibit KLK1-mediated HA cleavage and thereby reduce viral production. Treatment with recombinant kallistatin proteins dose-dependently inhibited KLK1-mediated cleavage of recombinant HA proteins of influenza A/WSN/33, A/California/04/09, and A/Japan/305/57 viruses (Fig. 4A). Importantly, kallistatin also inhibited the HA cleavage induced by KLK1 on the surfaces of virus particles (Fig. 4B). As shown in Fig. 4C, MDCK cells treated with kallistatin were protected from influenza A/WSN/33 virus-induced CPE. Notably, such treatments also abrogated the effect of KLK1 on potentiating virus-induced CPE. Accordingly, cells infected with influenza A/WSN/33 virus in the presence of kallistatin reduced viral yields by 70% compared to the untreated cells (Fig. 4D). Furthermore, treatment with kallistatin also significantly alleviated KLK1-induced elevation of viral titers (Fig. 4D). Collectively, these results demonstrate that exogenous treatment with kallistatin proteins can inhibit KLK1-mediated HA cleavage and reduce virus production.

FIG 4.

Exogenous treatment with kallistatin proteins inhibits HA cleavage and reduces influenza virus production. (A and B) Cleavage of purified HA proteins and influenza A virus particles by KLK1 in the presence of kallistatin. Recombinant HA proteins derived from influenza A/WSN/33 (H1N1), A/California/04/09 (H1N1), and A/Japan/305/57 (H2N2) viruses (100 ng in 20 μl saline) (A) and influenza A/WSN/33 (H1N1) and A/Taiwan/N2723/06 (H3N2) virus particles (2 × 10⁵ PFU in 5 μl serum-free DMEM) (B) were incubated with kallistatin and/or KLK1 for 4 h at 37°C. After incubation, the cleavage patterns of the HA proteins were analyzed by immunoblotting. To facilitate visualization of all HA bands, films with multiple exposures were analyzed to ensure that the signals for HA1 and HA2 could be detected. The exposure time was longer for HA1 and HA2 bands than for HA0 bands to increase sensitivity. The bands shown in panels A and B were quantified by densitometric analysis, and the ratios (HA1 plus HA2 divided by HA1 plus HA2 plus HA0) representing the HA cleavage efficiency are shown at the bottom of the blots. The results are representative of three independent experiments. (C) CPE in cells infected with influenza virus in the presence of KLK1 and/or kallistatin. MDCK cells were infected with influenza A/WSN/33 virus at an MOI of 5 in the presence of KLK1 (6 μg/ml) and/or kallistatin (13.3 μg/ml) or mock infected for 48 h. (Top) The cells were stained with crystal violet to monitor the CPE. (Bottom) The wells were scanned, and the crystal violet staining was quantified to determine cell survival. The values, which are shown as relative intensity levels, with the levels in the mock-infected cells arbitrarily set to 100, represent means and SD (P < 0.0001; one-way ANOVA). The results are representative of three independent experiments. (D) Quantification of influenza virus produced from virus-infected cells in the presence of KLK1 and/or kallistatin. The viral titers in the culture supernatants of the treated cells shown in panel C were quantified by plaque assay. Each symbol represents an individual value; the horizontal bars represent means ± SD (n = 3; P = 0.0007; one-way ANOVA). The results are representative of two independent experiments.

Cells transduced with the kallistatin gene are more resistant to influenza virus infection.

In addition to using exogenously added kallistatin, we also used endogenously overexpressed kallistatin to demonstrate its anti-influenza virus activity. As shown in Fig. 5A, cell viability was significantly higher in MDCK cells transduced with LV-KS than in those transduced with the control vector, LV-Null, following influenza virus infection in either the presence or absence of KLK1. Secretion of kallistatin in the conditioned medium of MDCK cells transduced with LV-KS was verified by ELISA (Fig. 5B). We estimated that 7.7 ng of kallistatin was secreted by 5 × 10⁴ MDCK cells that had been transduced with the kallistatin gene. Furthermore, LV-KS-transduced cells produced significantly less virus than did LV-Null-transduced cells following influenza virus infection in the presence of KLK1 (Fig. 5C). Taken together, these results demonstrate that endogenous overexpression of kallistatin can reduce virus-induced CPE and virus production.

FIG 5.

Overexpression of kallistatin in MDCK cells decreases influenza virus-induced CPE and reduces viral production. MDCK cells that had been transduced with LV-KS or LV-Null or left untransduced were infected with influenza A/WSN/33 virus at an MOI of 0.05 in the presence or absence of KLK1 (2 μg/ml) and cultured for 48 h. (A) (Top) Examination of CPE by crystal violet staining. (Bottom) The wells were scanned, and the crystal violet staining was quantified to determine cell survival (P < 0.0001; one-way ANOVA). (B) Detection of kallistatin contents in the conditioned medium by ELISA. The values shown are means and SD (n = 3; P < 0.0001; one-way ANOVA). N.D., not detectable. (C) Quantification of influenza virus in the conditioned medium by TCID₅₀ assay. Each symbol represents an individual value; the horizontal bars represent means ± SD (n = 3; P = 0.028; one-way ANOVA). The results shown in panels A and B are representative of three independent experiments.

Treatment with LV-KS protects mice against lethal influenza virus challenge and alleviates lung inflammation.

Intranasal inoculation of lentiviral vectors has been reported to target airway epithelial cells, type II alveolar cells, and the lung endothelium of mice (47). To test whether the anti-influenza virus activity of kallistatin seen in vitro is also manifested in vivo, we treated mice intranasally with LV-KS or LV-GFP for 3 consecutive days. On the next day, the mice were challenged with a lethal dose of influenza virus. Expression of human kallistatin in the lungs of mice receiving LV-KS was confirmed (Fig. 6A). Treatment with LV-KS significantly protected the mice against lethal influenza virus challenge compared to treatment with LV-GFP or saline (Fig. 6B). Viral titers were also significantly reduced in the BAL fluid of the infected mice pretreated with LV-KS compared with those pretreated with LV-GFP (Fig. 6C). Because kallistatin can regulate NO secretion and modulate the release of cytokines, including TNF-α, MCP-1, and intercellular adhesion molecule 1 (ICAM-1) (48), we examined whether LV-KS gene delivery reduces the levels of proinflammatory cytokines and chemokines in the BAL fluid. Mice pretreated with LV-KS produced lower levels of IL-1β in the BAL fluid at day 3 p.i. than those pretreated with LV-GFP (Fig. 6D). Nevertheless, no statistical differences in the levels of TNF-α, IL-6, IFN-γ, MCP-1, and MIP-1α in the BAL fluid at day 3 p.i. were found between LV-KS- and LV-GFP-treated mice (data not shown). Similarly, serum TNF-α, IL-6, and NO levels in LV-KS- and LV-GFP-treated mice at 6 and 12 h p.i. were not significantly different (data not shown). In terms of lung pathology, kallistatin gene delivery reduced lung injury and decreased infiltrating inflammatory cells in mice infected with influenza virus (Fig. 6E). Furthermore, lung histology scores were lower in LV-KS-treated mice than in LV-GFP- or saline-treated control mice (Fig. 6F). Collectively, these results demonstrate that lentivirus-mediated kallistatin gene delivery significantly reduces viral production in the respiratory tract, ameliorates lung pathology, and protects mice against lethal influenza virus infection.

FIG 6.

Kallistatin gene delivery enhances survival and decreases viral loads, IL-1β levels, and the severity of lung pathology in influenza virus-infected mice. (A) Kallistatin levels in the lung after lentivirus-mediated gene transfer. C57BL/6 mice were intranasally inoculated with 10⁷ TU of LV-KS or LV-GFP for 3 consecutive days. After 72 h, the levels of kallistatin in the lung were detected by ELISA (means and SD; n = 5). The results are representative of three independent experiments. N.D., not detectable. (B) Kaplan-Meier survival curve. C57BL/6 mice were intranasally inoculated with 10⁷ TU of LV-KS or LV-GFP or with saline at days −3, −2, and −1, followed by intranasal inoculation with influenza A/WSN/33 virus (7 × 10⁶ PFU) at day 0. The survival curves of the treated mice are shown (n = 10 to 12). The data were pooled from two independent experiments. (C) Viral titers in the BAL fluid of the treated mice at day 3 p.i. were quantified by plaque assay (means ± SD; n = 4 or 5). The results are representative of two independent experiments. (D) IL-1β levels in the BAL fluid of the treated mice were determined at day 3 p.i. by ELISA (means and SD; n = 8 to 10). The data were pooled from two independent experiments. (E) H&E-stained lung sections collected at day 6 p.i. (original magnification, ×200; scale bar, 100 μm). Lungs from uninfected mice served as the control. The results are representative of three independent experiments. (F) Lung histology scores. H&E-stained longitudinal cross sections were graded as no changes (score = 0), mild (score = 1), moderate (score = 2), or severe (score = 3) based on the severity of inflammation and tissue damage (means and SD; n = 4 to 6; P = 0.0008; one-way ANOVA). The results are representative of three independent experiments.

DISCUSSION

Aberrant kallikrein-kinin signaling plays important roles in a wide range of pathological processes, including inflammation. Prekallikreins, inactive precursors of kallikreins, are widely distributed and can be activated by inflammatory conditions. In a mouse model of dual infection with influenza virus and Serratia marcescens, administration of a 56-kDa protease of Serratia enhanced influenza virus replication and generated larger amounts of plasmin activity in the lung, which contributed to enhanced proteolytic cleavage of HA (49). In a ferret model of influenza, lysyl-bradykinin and bradykinin were produced in ferret nasal secretions at ratios similar to those seen in humans during influenza virus infection, suggesting that ferret kallikreins were induced and functional in the respiratory tracts of ferrets with influenza (50). There has been limited information so far on the host proteases involved in influenza virus activation. Apart from KLK1, which was upregulated in mice during influenza virus infection as demonstrated in the present study, KLK5 and KLK12, which are present in the respiratory tracts of healthy individuals, have been reported to cleave and activate HA (51). Although KLK5 and KLK7 have been shown to be involved in skin inflammation and atopic dermatitis, their roles in influenza virus infection have yet to be elucidated. In the lung and trachea, KLK12 is most abundant, whereas KLK1 is moderately expressed (52). Kallistatin has been shown to display high specificity toward KLKs, being a strong inhibitor of KLK1 and KLK7 and a weak inhibitor of KLK14 (25, 53). It was shown that KLK5 and KLK12 can be inhibited by α2-antiplasmin, whereas KLK5 can also be inhibited by proteinase C inhibitor (54). Whether these two natural inhibitors are physiologically relevant to the regulation of KLK5 and KLK12 in healthy and diseased states, such as infection and inflammation, awaits clarification. In addition to KLK1, it is worthwhile to examine whether other kallikreins, such as plasma kallikrein and other KLK members, are also upregulated during influenza virus infection and their roles in the cleavage activation of HA.

Influenza A virus undergoes antigenic variation, allowing the virus to evade host immune responses, which may result in virus replication to select for resistance to currently available anti-influenza drugs. The increasing emergence of drug-resistant influenza virus strains reinforces the need for identifying new therapeutic targets. To develop effective antiviral agents against influenza virus, cellular factors or signaling pathways that are essential for the viral replication cycle may be more favored targets than viral factors. Our in vitro and animal studies demonstrated that kallistatin exerted anti-influenza virus activity, presumably through binding to KLK1 and thus counteracting KLK1-mediated cleavage activation of HA. Endogenous human and rat kallistatin-KLK1 complexes have been demonstrated in plasma and various body fluids (25, 55). Amelioration of lung inflammation in virus-infected mice may be attributable to both anti-influenza virus and anti-inflammatory activities of kallistatin. Since kallistatin exerts pleiotropic effects, either dependent on or independent of KLK1, we presume that not only does kallistatin inhibit the cleavage activation of HA through binding to KLK1, but it may also attenuate inflammation in both KLK1-dependent and- independent manners in influenza virus infection. Our findings provide impetus for developing anti-influenza virus agents based on kallistatin or other KLK inhibitors to target the kallikrein-kinin system. We have used recombinant human kallistatin to treat influenza virus-infected mice via the intratracheal route. As the treatment doses and schedules were not optimized and the sample sizes may not have been large enough, differences in body weight loss and survival time were not significantly different between the kallistatin-treated and the control mice. Nevertheless, the kallistatin-treated mice appeared to exhibit less body weight loss and to have longer survival times (see Fig. S1 in the supplemental material). These results suggest that local administration of kallistatin protein might be effective in ameliorating influenza pathology. Moreover, identification of novel small-molecule inhibitors of KLK1 by high-throughput screening may deserve exploration for developing anti-influenza drugs.

Proteolytic cleavage of HA is determined by the susceptibility of the HA molecules of each influenza virus strain and the substrate specificity of host proteases. HA cleavage occurs at the C terminus of a single arginine or lysine residue that is located adjacent to the glycine residue (3). In the present study, we used two H1, one H2, and one H3 HA recombinant proteins to analyze their susceptibility to KLK1 cleavage. We found that the H1 HA molecule of influenza A/WSN/33 virus, a well-studied neurotropic H1N1 strain, could be cleaved most efficiently by KLK1 among the four virus strains. Amino acid substitutions near the HA cleavage site affect the cleavability by host proteases (5). The consensus sequence of the H1 cleavage site is IQSRG, with cleavage occurring between the arginine residue in the P1 position and the glycine residue in the P1′ position. However, influenza A/WSN/33 virus contains tyrosine (IQYRG) at the P2 position, which allows efficient HA cleavage by plasmin and is independent of trypsin (5). Thus, the replacement of the conserved serine residue with a tyrosine residue in the P2 position in influenza A/WSN/33 virus may account, in part, for its efficient cleavage by KLK1, similar to plasmin. In the cleavage activation of HA, the efficiency varies widely among HA subtypes and is dependent on the protease, implying that characteristics of proteases may influence the host range and adaptation of influenza virus. The cleavage specificity and efficiency of KLK1 on various HA subtypes requires further investigation with large sample sizes.

In clinical settings, lower levels of plasma and intestinal kallistatin were detected in patients with severe community-acquired pneumonia (56) and inflammatory bowel disease (57), respectively. Elevation of both circulating and synovial levels of KLK1 and kallistatin was found in patients with rheumatoid arthritis (58). In our murine model of acute influenza virus infection, expression of KLK1 was increased, whereas expression of kallistatin was decreased, in the lung, suggesting that KLK1 and kallistatin may play roles in influenza virus infection. To verify the importance of endogenous KLK1 and kallistatin in host defense against influenza virus infection, gene knockout mice would be useful tools to address these questions.

Host inflammatory responses play important roles in defense against influenza, but these responses may sometimes contribute to immunopathology. Studies on IL-1 receptor knockout mice revealed that IL-1 mediates acute lung inflammatory pathology but enhances survival of influenza virus-infected mice (59). In the present study, we showed that influenza virus infection was accompanied by production of IL-1β, TNF-α, IL-6, IFN-γ, MCP-1, and MIP-1α in the BAL fluid. Influenza virus infection induces NLRP3 inflammasome activation and produces IL-1β and IL-18 in human macrophages by a caspase 1-dependent pathway (60). Virus infection activates several transcription factors involved in the induction of cytokine and chemokine gene expression. Influenza virus infection can activate NF-κB, which is important in virus-induced cytokine production (61). Upon inflammation, bradykinins are rapidly generated and can induce IL-1β gene expression through activation of NF-κB (62). IL-1β production by kinin-stimulated cells may also exert a positive effect on the expression of bradykinin receptors, which mediate cellular effects of kinins (63). Thus, cross talk between cytokines and the kallikrein-kinin system may facilitate and amplify inflammatory responses upon influenza virus infection.

Identification of KLK1 and kallistatin contributions to the pathogenesis of influenza virus aids our understanding of the roles of host serine proteases and their inhibitors in virus infection. Our results provide evidence for the first time that KLK1 and kallistatin have impacts on influenza virus infection by affecting the cleavage activation of HA and inflammatory responses. These findings suggest that inhibition of KLK1 activity by kallistatin or other KLK1 inhibitors may be further explored for controlling influenza. Since ELISA kits for quantifying mouse KLK1 and kallistatin levels are not commercially available at this time, we did not have appropriate materials to quantify their basal and induced levels after influenza virus infection in mice. As results obtained from mice would not necessarily resemble those from patients infected with influenza virus, it will be of great interest to further analyze the levels of KLK1 and kallistatin in the BAL fluid and serum in patients with acute influenza. Further studies will be necessary to determine whether the KLK1-to-kallistatin ratio might serve as an early biomarker for acute influenza virus infection. Since proteolytic activation is also essential for the infectious cycle of some viruses besides influenza virus, such as severe acute respiratory syndrome coronavirus (SARS-CoV) and respiratory parainfluenza virus (64, 65), whether KLK1 and kallistatin have impacts on infection by these viruses warrants further investigation.