Distinct roles for type I and type III interferons in virulent human metapneumovirus pathogenesis

Human metapneumovirus (HMPV) is an important cause of acute lower respiratory infection in children and adults worldwide. There are four genetic subgroups of HMPV and both neutralizing antibodies and T cells contribute to protection. However, little is known about mechanisms of pathogenesis and most published work is based on a few extensively passaged, laboratory-adapted strains of HMPV. In this study, we isolated and characterized a panel of low passage HMPV clinical isolates representing all four genetic subgroups. The clinical isolates exhibited lower levels of in vitro replication compared to a lab-adapted strain. We compared disease phenotypes using a well-established mouse model. Several virulent isolates caused severe weight loss, lung pathology, airway dysfunction, and fatal disease in mice, which was confirmed in three inbred mouse strains. Disease severity did not correlate with lung viral titer, as virulent strains exhibited restricted replication in the lower airway. Virulent HMPV isolates were associated with markedly increased proinflammatory cytokine production and neutrophil influx; however, depletion of neutrophils or genetic ablation of inflammasome components did not reverse disease. Virulent clinical isolates induced markedly increased type I and type III interferon (IFN) secretion in vitro and in vivo. STAT1/2-deficient mice lacking both type I and type III IFN signaling showed reduced disease severity and increased lung viral replication. Inhibition of type I IFN signaling using a blocking antibody or genetic ablation of the type I IFN receptor reduced pathology with minimal effect on viral replication. Conversely, blockade of type III IFN signaling with a neutralizing antibody or genetic ablation of the IFN-lambda receptor had no effect on pathogenesis but restored viral replication. Collectively, these results demonstrate distinct roles for type I and type III IFN in HMPV pathogenesis and immunity.

Recent studies of the related respiratory syncytial virus (RSV) showed that RSV clinical isolates induce higher mucin production and more severe respiratory disease in mice, and this virulence mapped to the fusion (F) protein [28][29][30].Another study found that virulent RSV clinical isolates were correlated with a distinct pattern of innate immune activation [31].HMPV, like other viruses, is capable of inhibiting type I interferon (IFN) signaling, though the mechanisms are not fully defined, and multiple viral proteins have been implicated including glycoprotein (G), small hydrophobic (SH), and the M2-2 protein [32][33][34][35][36][37].Nonetheless, type I IFN contributes to disease pathogenesis [38,39].One prior study reported that HMPV induced type III IFN and that treatment of mice with exogenous type III IFN reduced viral titer [40].However, the roles of type I and III IFN during HMPV infection have not been fully elucidated.
In this study, we isolated and characterized disease phenotypes for a panel of HMPV isolates covering all four genetic subgroups.We identified several virulent isolates capable of causing severe and fatal disease in mice.Virulent HMPV isolates were associated with markedly increased proinflammatory cytokine production as well as increased type I and type III IFN (IFN) secretion in vitro and in vivo.Inhibition of type I IFN signaling reduced pathology with minimal effect on viral replication.Conversely, blockade or genetic ablation of type III IFN signaling had no effect on pathogenesis but partially restored viral replication.Collectively, these results demonstrate distinct roles for type I and type III IFN in HMPV pathogenesis and immunity.

PLOS PATHOGENS
Type I and type III interferons in HMPV pathogenesis

Isolation and identification of HMPV clinical isolates
A panel of nine HMPV clinical strains were isolated between 1985-2004 from patients with acute respiratory illness in Nashville, TN, or Cincinnati, OH (Table 1).The clinical isolates were grown in LLC-MK2 cells, maintained below 10 passages, and sucrose purified prior to full genome deep sequencing.There were no sequences from other pathogens in the isolates, and individual testing by real-time RT-PCR was negative for adenovirus; influenza A, B, and C; coronaviruses 229E and OC43; parainfluenzaviruses 1-4; RSV; rhinoviruses; and enterovirus D-68 and all were negative [41][42][43].We performed clustering analysis on the clinical isolates based on their full-length N, F, and G gene sequences, together with the published sequences of reference strains NL/1/00, NL/1/99, and CAN/97-83.The phylogenic trees generated by N, F, and G genes were similar (S1A-S1C Fig) .The genomic sequences of TN/94-49 and TN/94-66 were nearly identical, as were 1501 and 1509; these may represent the same strains infecting multiple patients in the same season.We calculated the nucleotide and amino acid identities of the nine viral proteins.Similar to previous observations for other HMPV sequences [44][45][46][47], the N, P, M, F, M2-1, M2-2 and L genes were highly conserved at both nucleotide and amino acid level (S1 Table ).The SH and G genes were less conserved with amino acid identity even lower than nucleotide identity, suggesting low selection pressure on these regions.
We tested in vitro growth kinetics in LLC-MK2 cells of 4 strains, each representing a unique genetic subgroup.All four isolates exhibited log growth at 1-3 days post inoculation and virus titer decreased in supernatant after 7 days (S1D Fig) .However, the four isolates reached different peak titers; the laboratory strain TN/94-49 replicated to ~10 7 PFU/ml, but clinical isolates C2-202, TN/91-320, and 1501 only reached 10 5 −10 6 PFU/ml.The clinical isolates also exhibited cytopathic effect (CPE) at various degrees.HMPV TN/94-49 cause widespread syncytium formation and cell detachment at 5 days post infection (S1E Fig) .The monolayers of 1501, C2-202, and TN/91-320 infected cells remained intact with minimal syncytia.As previously shown, these isolates had different abilities to form plaques under agarose overlay in LLC-MK2 cells [48].

HMPV clinical isolates cause differential weight loss and disease in mice
We next tested the pathogenesis of the HMPV clinical isolates in an established B6 mouse model for HMPV [26,39].Comparison of different anesthesia and inoculation routes (intratracheal, I.T. vs. intranasal, I.N.) indicated that I.T. yielded modestly but significantly higher lung viral titer and equivalent nasal titer (S2 Fig) .We infected each mouse with 1.0 x 10 6 PFU HMPV or mock LLC-MK2 cell lysate via I.T. route under anesthesia, monitored mice daily for weight loss and survival, and euthanized on day 5 (S3A Fig) .While mice infected by TN/94-49 and TN/94-66 did not lose weight, mice infected by strains 1501, 1509, C2-202, TN/91-320, and TN/96-35 experienced substantial weight loss and up to 60% mortality (Fig 1A and 1B).These mice also showed decreased response to stimuli, hunched posture, huddled behavior, and ruffled fur.Mice infected by strains 42 and C1-334 were moderately ill but recovered.
To exclude the interference of host genetic background, we tested three virulent strains 1501, C2-202, and TN/91-320 together with the avirulent strain TN/94-49 in two other inbred mouse strains BALB/c and DBA/2 (S3B-S3E Fig) .HMPV TN/94-49 caused no weight loss in either inbred mouse strain, while the three virulent strains induced significant weight loss and mortality in both strains.We previously found DBA/2 mice to be more permissive for HMPV replication than other inbred strains [23,27], and all the DBA/2 mice infected by virulent

Virulent HMPV clinical isolates cause pulmonary pathology and respiratory dysfunction in mice
To examine pulmonary histology in mice infected by HMPV clinical isolates, we analyzed mouse lungs at 5 days post-infection.The lungs of mock-infected mice showed clear alveoli, but the lungs of mice infected with clinical isolates exhibited various degrees of pathologic changes.The lungs of mice infected with virulent isolates 1501, 1509, C2-202, TN/91-320, and TN/96-35 exhibited more intense H&E staining (Fig 1C).Higher magnification examination showed massive neutrophil infiltration and thickening of alveolar walls, characteristic of inflammation and viral pneumonia.Mid-virulent strains 42, TN/94-66, and C1-334 induced mild neutrophil infiltration only surrounding the bronchioles, with clear alveoli.In comparison, TN/94-49 induced minimal pathologic changes (Fig 1D).Clinical HMPV isolates also caused airway damage.Total protein concentration was increased in the bronchoalveolar fluid (BALF) of mice infected with C2-202, indicating epithelial damage and leakage of blood proteins into the bronchioles (Fig 1E).Mice infected by HMPV virulent isolates 1501 and C2-202 at sub-lethal dose (2 x 10 5 PFU) survived acute infection but demonstrated altered airway physiology two weeks later (Fig 1F).There was a significant difference in airway resistance (Rn) and tissue resistance (G) between TN/94-49 vs C2-202 and C2-202 vs 1501 at 50 mg/ml dose of methacholine.There was a trend toward increased hysteresis as well as C2-202 being higher than the other groups in tissue elastance (H) though this did not reach statistical significance.

HMPV disease severity does not correlate with lung virus titer
To test the hypothesis that disease severity would correlate with lung virus titer, we measured the lung virus titers of the clinical isolates at 5 days post-infection.The lung virus titers varied widely and did not correlate with virulence of the strain (Fig 2A).For example, the non-virulent strain TN/94-66 reached the highest titer (~10 5 PFU/g).TN/94-49 replicated to 10 3 in this experiment, but in most other experiments it reached titers of 10 5 −10 6 .Virulent strains ranged from limit of detection (TN/96-35) to ~3.5 x 10 3 (1501).The nasal turbinate titers were similar among isolates, showing that the inoculations were effective (Fig 2B).

High pro-inflammatory cytokine levels correlate with more severe lung disease
The limited lung replication of virulent strains and early weight loss kinetics led us to hypothesize that host innate immune response contributed to both disease and viral control.To test this hypothesis, we measured cytokine levels in mouse lungs at 5 days post-infection (

Pathogenesis of HMPV clinical isolates is inflammasome-independent
The correlation between higher pro-inflammatory cytokine levels and disease severity prompted us to hypothesize that the inflammasome pathway plays an important role in the disease pathogenesis.It has been reported that up-regulation of inflammasome-related genes is correlated with more severe disease symptoms in human patients infected with HMPV [49].We first examined the inflammasome activation in HMPV-infected cells by immunofluorescence assay.Bone marrow-derived macrophages (BMDMs) were infected with either TN/94-49 or C2-202 at MOI = 1 for 24 hours, fixed, and stained with antibodies recognizing ASC (Apoptosis-associated Speck-like protein containing CARD) and HMPV-F (S5A Fig) .ASC was evenly distributed in uninfected cells.The formation of inflammasome (as shown by aggregation of ASC) was detected in BMDMs infected by both virus strains, but not in mockinfected or UV-inactivated virus-infected cells (S5B Fig) .This data shows that HMPV infection is able to activate the inflammasome pathway in virus-infected cells.
We next tested the role of the inflammasome in HMPV disease pathogenesis using genetically-engineered mice lacking inflammasome components ASC or nod-like receptor family pyrin domain containing 3 (NLRP3).We infected age-matched female wild type and ASC-KO mice with 1.0 x 10 6 PFU HMPV TN/94-49, C2-202, 1501, and TN/91-320.The deficiency in ASC did not reduce the disease severity in the first five days post-infection: ASC-KO mice infected with virulent clinical isolates C2-202, 1501, and TN/91-320 lost about 30% weight, similar to their wild type companions (S5C Fig) .The more virulent isolates exhibited restricted replication while the less virulent strain TN/94-49 reached about 10 6 PFU/g in lungs of ASC-KO mice (S3D Fig) .We also tested the virulent isolate C2-202 in NLRP3-KO mice.The virulent strain C2-202 was still able to cause about 30% weight loss in NLRP3-KO mice and replicated to low titer, not distinguishable from wild type mice (S5E and S5F Fig) .Collectively, these data show that NLRP3 and ASC inflammasome are not required for disease pathogenesis of HMPV clinical isolates.

HMPV virulent isolate C2-202 infection inhibits dendritic cell recruitment
We next used TN/94-49 (non-virulent) and C2-202 (virulent) as two representative strains to characterize the innate immune cell response to HMPV.The antigen presenting cell (APC) recruitment patterns were analyzed with an APC flow panel shown in S6 Fig.We gated for pulmonary dendritic cells (DCs), alveolar macrophages, and interstitial macrophages with surface markers CD11b, CD11c, and MHC-II based on a previous study [39].To identify APC recruitment patterns, we performed an infection kinetics experiment by harvesting three  In summary, C2-202 infection induced a different APC recruitment pattern compared with TN/94-49, with fewer DC recruitment as the most prominent feature.The role of DCs in C2-202 pathogenesis remains to be assessed.

HMPV C2-202 induces more neutrophil infiltration but neutrophil depletion does not rescue disease severity
Previously we described neutrophil infiltration into mice lung after HMPV infection by histology staining (Fig 1).We thus sought to further characterize the role of neutrophils in HMPV pathogenesis.To quantify neutrophil infiltration, we used flow cytometry by staining with α-CD11b and α-Ly6G antibodies.Neutrophils were defined as CD11b+ Ly6G+ cells in F4/ 80-negative non-lymphocyte populations (Fig 3B).We infected mice with TN/94-49 and C2-202 via I.T. route and analyzed lung neutrophil numbers at 6h, 24h, and 48h post-infection.We also infected mice with LLC-MK2 cell lysates (mock) or UV-inactivated virus and harvested their lungs at 6h post-infection as controls.HMPV infection induced neutrophil infiltration to lung as early as 6 h post infection and the infiltration peaked during the first day after infection (Fig 3C and 3D).Further, HMPV C2-202 induced significantly more neutrophil infiltration to lung compared with TN/94-49 (Fig 3D).The higher neutrophil infiltration correlated with the higher KC cytokine level in C2-202 infection (Fig 3A).
Based on the correlation between elevated neutrophil infiltration and more severe disease, we hypothesized that neutrophils were the cause for the severe pathogenesis of C2-202 infection.To test this hypothesis, we depleted neutrophils with intra-peritoneal (I.P.) injection of Ab targeting Ly6G (clone 1A8) [50].This Ab has been used in multiple previous studies to successfully deplete neutrophils in infection models [51][52][53][54][55][56][57].We optimized the Ab dosage by testing different regimens in C57BL/6 mice infected with C2-202 and verified neutrophil depletion efficiency by flow cytometry.After optimization, we performed the neutrophil depletion and infection experiments according to the scheme shown in

Deficiency in IFN signaling reduces disease severity
To further characterize the role of type I and type III IFN in the pathogenesis of HMPV clinical isolates, we infected genetically engineered STAT1/2-deficient (STAT-KO) mice, which are deficient in both type I and III IFN signaling due to the absence of signal transducers.Deficiency in STAT1/2 partially rescued the C2-202 disease phenotype (Fig 4E).The STAT-KO mice infected by C2-202 experienced less weight loss and were more active than their wild type counterparts.The deficiency in IFN signaling also restored the replication ability of C2-202 in lung.HMPV C2-202 was able to replicate 1000-fold better in STAT-KO mice than in wild type mice, and the titer was similar to that of TN/94-49 (about 10 6 PFU/g) (Fig 4F).HMPV TN/94-49 replicated similarly in STAT-KO mice and wild type mice.Downstream cytokines including IL-6, MCP-1, MIP-1A, MIP-1B, and RANTES induction by C2-202 were lower in IFNAR KO and STAT-KO mice compared to WT (S9 Fig) .Collectively, these data suggest that type I and III IFN signaling control HMPV replication and induce immunopathology.

Deficiency in type I IFN signaling alone rescues disease of HMPV C2-202
We next tested whether type I IFN alone was sufficient to induce anti-viral response and immunopathology using two mouse models that were deficient in type I but competent in type III IFN signaling.In the IFN-α/β receptor (IFNAR) blockade model, we treated the mice with IFNAR blocking Ab or isotype Ab for four doses before and during C2-202 infection (Fig 5A).The IFNAR blockade treatment significantly rescued the acute disease phenotype (Fig 5B).However, there was no change in the restriction of viral replication (Fig 5C).We also tested this hypothesis in mice genetically deficient in IFNAR (IFNAR-KO).Similar phenotypes were observed in the IFNAR-KO mice, with substantial rescue of disease phenotype but no effect on viral replication (Fig 5D and 5E).As previously shown, TN/94-49 replication was not restricted in WT or IFNAR-KO mice.Collectively, these data show that type I IFN signaling pathway mediated the immuno-pathogenesis of HMPV C2-202 infection but had little effect on controlling viral replication.

Type III IFN contributes to virus control but not disease
To dissect the individual roles of type I and III IFN, we used mouse models that we depleted either type I or type III or both.Wildtype C57BL/6J or interferon λ receptor knockout mouse (Ifnlr -/-) were treated with isotype control or IFNAR blocking Ab on day 1 prior to infection with 5x10 5 PFU of virulent strain C2-202 and on day 1 and 3 post-infection.WT B6 and Ifnlr -/-mice treated with isotype control lost >25% of body weight over 5 days, while WT C57BL/6J and Ifnlr -/-mice treated with IFNAR antibody experienced minimal weight loss (Fig 6A).Next, we evaluated how type I and III IFN affected lung viral titer.Ifnlr -/-mice treated with isotype Ab showed a modest but significant 0.8 log 10 increase in lung virus compared to WT B6 treated with isotype Ab (Fig 6B).In comparison, blocking IFNAR alone in B6 showed only 0.4 log 10 on viral shedding, but IFNAR blockade in Ifnlr -/-mice restored viral titer to 4.1 log 10 (Fig 6B).In a separate experiment, we infected WT, IFNAR-KO, and Ifnlr -/-mice and collected lungs for titer at different time points to 14 days (S10 Fig) .WT and Ifnlr -/-mice lost weight, while IFNAR-KO mice did not.Ifnlr -/-mice had higher viral titers on day 5, while all animals cleared by day 10.Collectively, these results suggest that IFN-I contributes primarily to disease, while IFN-III principally functions to control viral replication.

Discussion
Our study reveals two important aspects of HMPV pathogenesis.First, different clinical isolates exhibit markedly differing virulence and distinct inflammatory responses.Moreover, type I and type III IFN exhibit distinct functions in response to HMPV.Many viruses exhibit quite disparate biology, virulence, or even receptor usage between cell-adapted laboratory/vaccine strains and clinical isolates, including RSV [30,31,58]; cytomegalovirus, herpes simplex, and varicella virus [59][60][61]; Dengue virus [62]; and measles virus [63,64].Indeed, a classic method of viral attenuation to generate a vaccine candidate is cell passage.We found a spectrum of in vivo phenotypes between clinical isolates and several lab-adapted strains; moreover, virulence did not correlate with genetic subgroup and importantly was confirmed in multiple inbred mouse strains.Clinical studies have sought but not consistently shown differences in human disease between subgroups [65][66][67][68][69].These data will provide powerful tools to explore mechanisms of HMPV pathogenesis and help elucidate the contribution of specific viral proteins.
Clinical disease and inflammation were not correlated with lung viral titer, consistent with immunopathology caused by both HMPV and RSV [39,[70][71][72][73][74].Virulent HMPV strains were associated with markedly increased proinflammatory cytokine production and neutrophil influx.However, experiments in genetically deficient mice and neutrophil depletion demonstrated that these were not required for pathogenesis.Although neutrophil depletion did not alter disease in these experiments, neutrophils express Ifnlr; future experiments using bone marrow chimeras could confirm whether immune or non-immune cells expressing Ifnlr are responsible for virus control.A recent paper reported that chemical inhibition of NLRP3 reduced disease [75]; however, the effect was more marked in BALB/c mice, which may have more severe disease with some strains of HMPV [25,71].Similarly, depletion of neutrophils enhanced disease in BALB/c mice [51].These differences could be due to different virus strains; the panel of genetically and phenotypically distinct isolates collected herein may enable elucidation of specific viral protein-cell protein interactions that drive these outcomes.
The virulent isolate C2-202 and the avirulent TN/94-49 induced markedly different macrophage and dendritic cell responses, which may contribute to the pro-inflammatory phenotype.The effects of the differential response of antigen-presenting cells warrants future studies of adaptive immune response to different HMPV strains; dendritic cells are particularly important for T cell response to HMPV [76].Virulent C2-202 induced potent production of type I and III IFN, and inhibition of type I IFN signaling reduced pathology with minimal effect on viral replication.Modulation of type I IFN activation by receptor antibody blockade or depletion of key mediators of IFN signaling alleviates disease severity of RSV, influenza virus [77], lymphocytic choriomeningitis virus (LCMV) [78,79], and others.We previously showed that type I IFN signaling controls early HMPV replication but leads to airway dysfunction and inflammation in HMPV infection [39].Here, we demonstrate that type I IFN primarily contributes to disease phenotype.
We found that type III IFN signaling had no effect on pathogenesis but inhibited viral replication.One previous study reported that HMPV induced type III IFN and that treatment of mice with exogenous type III IFN reduced viral titer, but also ameliorated weight loss; however, that study used BALB/c mice [40].Type I IFN have been reported to contribute to antiviral immunity more potently during systemic infections, while type III exhibits a greater role in mucosal surface infections [80,81].HMPV infects airway epithelial cells and does not infect other organs [21,23,82].Thus, modulation of type I or type III IFN might represent a therapeutic intervention against HMPV.
Collectively, these results demonstrate that diverse HMPV clinical isolates exhibit widely divergent virulence in vivo using a mouse model.Moreover, the findings suggest discordant roles for type I and type III IFN.While pathogenesis in mice may differ from that in humans, the use of these clinical isolates will facilitate efforts to dissect viral and host determinants of HMPV virulence.

Ethics statement
The research was approved by the University of Pittsburgh Institutional Animal Care and Use Committee #21028897.

Optimization of infection route for HMPV
Since mice are not natural hosts for HMPV, we first optimized infection routes and anesthesia methods to establish an effective infection procedure for C57BL/6 (B6) mice.We compared the virus yield from intranasal (I.N.) and intratracheal (I.T.) routes.We also compared isoflurane to ketamine/xylazine for anesthesia.Higher lung virus titer was obtained using the intratracheal (I.T) route compared to the intranasal (I.N.) route (S2 Fig) .All the subsequent mouse studies used this I.T. route with isoflurane anesthesia.The intratracheal route has been widely used in drug delivery for local treatment of rhinitis or nasal polyposis, as well as in influenza pathogenesis studies.Under these conditions, most of the inoculum is aspirated into the lungs [83][84][85][86].

Animals
C57BL/6J, C57BL/6T, BALB/cJ, DBA/2J mice were purchased from commercial breeders (Jackson Laboratory or Taconic), and ASC-deficient (ASC -/-) B6 mice were kindly provided by Dr. John Alcorn.Nlrp3-deficient (Nlrp3 -/-) B6 mice were kindly provided by Dr. Jieru Wang.IFNAR IFN-α/β receptor-deficient (IFNAR -/-) B6 mice were kindly provided by Dr. Terence Dermody.All animals were bred and maintained under specific-pathogen-free conditions in accordance with the guidelines of the University of Pittsburgh Institutional Animal Care and Use Committee.Ifnlr -/-mice were generated and backcrossed on a B6 background by the University of Pittsburgh Transgenic Mouse Core.Age-and gender-matched animals were used in all experiments.Mice from different backgrounds were co-housed for two weeks before any experiment.
For all experiments, mice were anesthetized by intraperitoneal injection of ketamine-xylazine or by inhalation isoflurane followed by intranasal or intratracheal administration of sucrose purified HMPV as indicated.After infection, the animals were evaluated daily for weight loss, mortality and symptoms.Animals were euthanized at indicated times postinfection.Nasal turbinates and right lungs were collected for virus titration.For histopathology analysis, a portion of the left lung was inflated with 10% buffered formalin, paraffin embedded, stained, and analyzed using a formal scoring system in a group-blinded fashion by an experienced lung pathologist as previously described.The studies were approved by the University of Pittsburgh Institutional Animal Care and Use Committee #18032376.Animals were euthanized by CO 2 asphyxiation.

IFN-α receptor blockade
C57BL/6J or Ifnlr -/-mice were injected i.p. with 500μg anti mouse IFNAR-1 Ab (clone MAR1-5A3, BioXCell) or isotype control mouse IgG1 in 0.1 ml of PBS 1 day prior to infection or on indicated time post-infection, followed by 250 μg Ab or isotype control every other day.

IFN-λ IFN depletion
To deplete IFN type III IFN, 6 to 12-week-old C57BL/6J or IFNAR KO mice were given 300ug of murine IL28B neutralizing Ab (clone 3C11, InvivoGen) intraperitoneally (i.p.) 1 day prior to infection, and subsequently given 150ug on days 1, and 3 postinfection.Control mice were treated with mice IgG1 isotype control (clone MOPC-21; Bio X Cell) on days −1, 1, and 3 postinfection.Mice were monitored daily for weight change and signs of disease.On Day 5 post infection, mice were euthanized and lung viral titer was determined by plaque assay.

Flow cytometry
Animals were euthanized at indicated times post-infection.The left lung was collected for virus titration and the right lungs were processed into a single cell suspension as previously described [26].Briefly, lungs were rinsed in R10 medium, minced with a scalpel, and then incubated with 2 mg/ml collagenase A (Roche) and 20 μg/ml DNase (Roche) for 1 h at 37˚C.Single-cell suspensions of digested lungs were obtained by pressing lung tissue through a nylon cell strainer (70-μm pore size).Erythrocytes were lysed with red blood cell lysis buffer (Sigma).Surface staining was performed for 1 h at room temperature in phosphate-buffered saline (PBS) containing 2% FBS.For innate immune cell characterization, lung cells were stained with anti-epithelial cell adhesion molecule (EpCAM, clone G8.8; BioLegend), anti-CD11c (clone HL3; BD), anti-CD11b (clone M1/70; Tonbo), anti-major histocompatibility complex (MHC) class II (clone 2G9; eBioscience), anti-Gr-1 (clone RB6-8C5; Biolegend), and anti-Ly6G (clone 1A8; BioLegend) antibodies.Cells were also stained with human anti-HMPV-F Ab 54G10 to detect HMPV antigens.For all cell populations, forward scatter (FSC) and side scatter (SSC) gating was used to obtain cells of the appropriate size and shape and doublets excluded.A Pacific blue Live/Dead staining dye was used to exclude all dead cells.Flow cytometric data were collected using an LSR II (BD Biosciences) and analyzed with FlowJo (Tree Star).

Next generation sequencing of virus genome and sequence analysis
Total RNA was extracted from sucrose-purified HMPV virus stocks using Trizol reagent (Invitrogen).The RNA samples for all HMPV clinical isolates were subjected to ribosomal RNA depletion, fragmentation, library preparation, and sequenced on a 318 chip with an Ion PGM system (ThermoFisher) at the University of Pittsburgh Genomics Research Core.Sequencing reads were assembled and aligned to a reference in CLC Genomics Workbench (QIAGEN) to generate a consensus genome sequence.Genome sequence for HMPV TN/94-49 (JN184400) and NL/1-99 (AY525843) were used as the reference sequence for subgroup A and B, respectively.Phylogenic trees were generated by aligning nucleic acid sequences using the ClustalW algorithm in MacVector 14 (MacVector) [88] and the neighbor-joining method.

Multistep virus growth curves in LLC-MK2 cells
Confluent LLC-MK2 cells in T25 flasks were infected with HMPV clinical isolates at MOI of 0.1 in triplicates.After 1 h of adsorption at 37˚C, the inoculum was removed and the cells were washed three times with PBS.Fresh Opti-MEM containing 0.5μg/ml trypsin-TPCK was added, and the infected cells were incubated at 37˚C in a CO 2 incubator.At indicated time points post-infection, 200 μl supernatant was collected from each well, snap-frozen, and stored at -80C.The virus titer was determined by an immunostaining assay in LLC-MK2 cells as described above.

Pulmonary cytokine/chemokine levels
Mice lung homogenates or BAL fluid were clarified by centrifugation at 13,800 × g for 10 min at 4˚C, and 50uL of the supernatant was used.Cytokines and chemokines from clarified BAL fluid or lung homogenate were measured using Bio-Plex Mouse Cytokine 23-Plex (BioRad) or ELISA (R&D) according to manufacturer's recommendations.

Real-time PCR
Total RNA from 200-μl volume of undiluted lung homogenate from infected or uninfected mice was extracted using RNeasy kit according to manufacturer's recommendations (Qiagen) and stored at −80˚C.Real-time reverse transcription (RT)-PCR was performed with 25-μl reaction mixtures containing 5 μl of extracted RNA on an ABI StepOnePlus real-time PCR system (Life Technologies/Applied Biosystems) with the AgPath-ID One-Step RT-PCR kit (Life Technologies/Applied Biosystems).For IFN α1, IFN β, and hypoxanthine phosphoribosyltransferase (HPRT) gene expression, primers and probes were used according to the manufacturer's instructions (Applied Biosystems/Ambion).All values were normalized to the HPRT expression.Gene expression are reported as fold differences between mock infected and infected by the ΔΔCT method.

Lung mechanics measurements
Following experimental exposures, pulmonary function was assessed by mechanical ventilation of anesthetized (90 mg/kg pentobarbital-NA, i.p.) and tracheotomized mice using a computer-controlled small-animal mechanical ventilator (flexiVent; SCIREQ) as previously described [88,89].Mice were mechanically ventilated at 200 breaths/min with a tidal volume of 0.25 ml and a positive end-expiratory pressure of 3 cm H2O (mimicking spontaneous ventilation).Respiratory mechanics measurements were made prior to and following the administration of the drug methacholine (dose range: 0-50 mg/ml), which causes the smooth muscle surrounding the airways to constrict.Multiple linear regression was used to fit measured pressure and volume in each individual mouse to the linear model of the lung.Model fits that resulted in a coefficient of determination < 0.8 were excluded.

Statistical analysis
Data analysis was performed using Prism version 6 (GraphPad Software).Multiple group comparisons were performed using two-way ANOVA followed by Turkey post-test for individual group comparisons.Error bars in each graph represent SD.
strains died by 3 days post infection (S3E Fig).We confirmed a dose-dependent disease response for C2-202 (S3F Fig).UV-inactivated virus did not cause weight loss or death, confirming that disease required viral replication (S3G Fig).

Fig 1 .
Fig 1. HMPV clinical isolates cause variable disease in mice.(a, b) Seven-week old female B6 mice were infected with 1.0 x 10 6 PFU HMPV clinical isolates or LLC-MK2 cell lysate (mock) diluted in 100 μl PBS via intra-tracheal (I.T.) route.Weight change (a) and survival (b) were monitored daily.Data shown are from 5 mice per group.(c) H&E staining of mice lungs infected by HMPV clinical isolates (200x magnification).(d) Pathology scores of the histology slides in (c), evaluated in a group-blinded fashion by a board-certified pathologist.Data shown are from 5 mice per group.(e) Sevenweek old BALB/c mice were infected with 2 x 10 5 PFU live or UV-inactivated HMPV via I.T. route.Fourteen days postinfection (p.i.), bronchoalveolar fluid (BALF) was obtained by injecting 1 ml PBS into mouse lung through cannulated trachea.About 700 μl BALF was recovered from each mouse.Total protein concentration was determined by Bradford assay.Data shown are from 2-3 mice per group.* P<0.05, ** P<0.01, one way ANOVA with Tukey's multiple comparisons test.(f) Fourteen days p.i., the airway resistance, tissue resistance, and tissue elastance of mice in (e) were analyzed by flexiVent during methacholine challenge.Data shown are from 2-3 mice per group.https://doi.org/10.1371/journal.ppat.1011840.g001
mice per time point (day 2, 3, 4, and 5 post-infection) and counted the number and percentage of each cell population with flow cytometry.As shown in S7A Fig, HMPV live virus infection induced an increasing number of total cells in lung, compared with mock infection or UV-inactivated virus infection.Although an increasing number of DCs were detected in mice infected by PLOS PATHOGENS Type I and type III interferons in HMPV pathogenesis

Fig 3 .
Fig 3. HMPV C2-202 induces pulmonary inflammation and neutrophil infiltration but neutrophil depletion does not rescue disease severity.(A) HMPV pathogenic strains induced higher pro-inflammatory cytokines.C57BL/6 mice were infected with 1.0 x 10 6 PFU HMPV via I.T. route and lungs harvested 5 days p.i. subject to homogenization.Cytokine protein levels were measured with Bio-Rad mouse 23-plex kit.Absolute cytokine concentration was normalized to that of the mock-infected group (5 mice per group) and average fold changes of each group were plotted in a heat-map with cytokine names on the right and virus strain names on top.More virulent strains are grouped to the right of the graph, and less virulent strains to the left.(B) Gating strategy for lung neutrophils.Neutrophils were gated from F4/80-negative non-lymphocytes as CD11b and Ly6G double positive population.Dead cells were excluded by a violet viability dye before further analysis.(C,D) Kinetics for neutrophil infiltration to lung.Mice were infected by 1.0 x 10 6 PFU HMPV I.T. and lungs were harvested at the indicated time point for flow cytometry analysis to count total cells (C) or neutrophils (D).(E) Neutrophil depletion does not rescue virulence by HMPV C2-202.Mice were injected via I.P. route with 250 μg α-Ly6G antibody or rat IgG2a isotype antibody 1 day before infection, following with 100 μg antibody I.P on day 0, 2, and 4 post-infection.(F) Depletion was confirmed by flow cytometry.Mice were infected with HMPV I.T. on day 0 as described before.Weight change (G) and day 5 lung virus titer (H) for neutrophil-depleted mice infected with HMPV.Schematic created with BioRender.com.https://doi.org/10.1371/journal.ppat.1011840.g003 Fig 3E.We confirmed the successful depletion of neutrophils by flow cytometry staining (Fig 3F).However, neutrophil depletion did not rescue the severe pathogenesis of C2-202 as the mice without neutrophils still lost about 30% weight at 5 days post infection (Fig 3G) and the absence of neutrophils did not affect virus replication (Fig 3H).HMPV C2-202 infection induces higher type I and III IFNIFNs are key regulators of the host immune response against viral infection.Type I and III IFN mediate the innate anti-viral and pro-inflammatory responses, and type II mediates the adaptive response.We measured the IFN induction levels in the mouse lungs by extracting total RNA from homogenized lung tissue at 1 day post-infection, and performed qRT-PCR to detect IFNα1, IFNβ, IFNγ, and IFNλ3 expression levels.HMPV infection induced all three types of IFN at 1 day post infection.Interestingly, C2-202 induced a much higher type I (IFNα1, IFNβ) and type III (IFNλ3), but not type II (IFNγ) response compared with TN/94-49 infection (Fig 4A).Measurement of IFN protein level by ELISA confirmed these findings (Fig 4B).IFNα1 mRNA peaked 24h post-infection, while IFNβ was elevated 48h post-infection (Fig 4C).The IFN response was tightly regulated, as type I IFN protein levels dramatically decreased on day 2 post-infection after a transient peak on day 1, while IL-28 exhibited a similar peak on day 1 post-infection and slower decline (Fig 4D).

Fig 4 .
Fig 4. Type I and III IFN are induced by C2-202 and deficiency in IFN signaling reduces disease severity.(A) Relative expression of IFNα1, IFNβ, IFNγ, and IFNλ3 mRNA in mouse lung at 1 day post-infection.Relative expression level was obtained by qRT-PCR and normalized to HPRT expression.Fold changes normalized to mock infection are shown.(B) Protein level of IFNα1, IFNβ, IFNγ, and IL-28 measured by ELISA.(C) Relative expression of IFNα1 and IFNβ mRNA in mouse lung at 6, 24, and 48h post-infection.Relative expression level was obtained by qRT-PCR and normalized to HPRT.Fold changes normalized to mock are shown.* P<0.05, ** P<0.01, *** P<0.001, n.s.P>0.05 by student's t-test.(D) Protein level of IFNα1 and IL-28 measured by ELISA at multiple time points postinfection.(E) Weight change and (F) day 5 lung virus titer of wild type (WT) or STAT1/2-deficient mice (STAT-KO) infected with 5.0 x 10 5 PFU mock, TN/94-49, or C2-202.Solid lines with closed shapes are wild type mice and dashed lines with open shapes are STAT1/2-deficient mice.Data were pooled from a total of 5-6 mice per group from two independent experiments.*** P<0.001, n.s.P>0.05 by student's t-test.https://doi.org/10.1371/journal.ppat.1011840.g004