of January 5, 2019.This information is current as

Pneumovirus-Infected MiceDelays Acute Lung Injury in Granzyme A- and B-Cluster Deficiency

Rosenberg and Albert P. BosJoseph B. Domachowske, Jan Paul Medema, Helene F. Reinout A. Bem, Job B. M. van Woensel, Rene Lutter,

http://www.jimmunol.org/content/184/2/931doi: 10.4049/jimmunol.0903029December 2009;

2010; 184:931-938; Prepublished online 16J Immunol 

Referenceshttp://www.jimmunol.org/content/184/2/931.full#ref-list-1

, 16 of which you can access for free at: cites 50 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2010 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on January 5, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

The Journal of Immunology

Granzyme A- and B-Cluster Deficiency Delays Acute LungInjury in Pneumovirus-Infected Mice

Reinout A. Bem,* Job B. M. van Woensel,* Rene Lutter,† Joseph B. Domachowske,‡

Jan Paul Medema,x Helene F. Rosenberg,{ and Albert P. Bos*

Lower respiratory tract infection by the human pneumovirus respiratory syncytial virus is a frequent cause of acute lung injury in

children. Severe pneumovirus disease in humans is associated with activation of the granzyme pathway by effector lymphocytes,

which may promote pathology by exaggerating proapoptotic caspase activity and proinflammatory activity. The main goal of

this study was to determine whether granzymes contribute to the development of acute lung injury in pneumovirus-infected mice.

Granzyme-expressing mice and granzyme A- and B-cluster single- and double-knockout mice were inoculated with the rodent

pneumovirus pneumonia virus of mice strain J3666, and were studied for markers of lung inflammation and injury. Expression

of granzyme A and B is detected in effector lymphocytes in mouse lungs in response to pneumovirus infection. Mice deficient for

granzyme A and the granzyme B cluster have unchanged virus titers in the lungs but show a significantly delayed clinical response to

fatal pneumovirus infection, a feature that is associated with delayed neutrophil recruitment, diminished activation of caspase-3,

and reduced lung permeability.We conclude that granzyme A- and B-cluster deficiency delays the acute progression of pneumovirus

disease by reducing alveolar injury. The Journal of Immunology, 2010, 184: 931–938.

Thehuman pneumovirus, respiratory syncytial virus (RSV),is the leading cause of lower respiratory tract illness inyoung children and infants worldwide (1). In Western

countries, the annual incidence rate of RSV-associated hospitali-zation may be as high as 30 per 1000 children under 1 y of age (1,2). RSV disease in these children presents as bronchiolitis orpneumonia. However, the severe end of the RSV disease spectrumincludes the development of acute lung injury and acute re-spiratory distress syndrome (ALI/ARDS), which is characterizedby severe hypoxemia and bilateral lung infiltrates in the setting ofa normal cardiac preload (3). Currently, treatment for severe RSVdisease in normal healthy children is limited to supportive meas-ures, stressing the need to further explore its pathogenesis (4).In general, ALI/ARDS is associated with the development of

diffuse neutrophilic alveolitis and loss of alveolar capillary barrierintegrity as a result of enhanced lung epithelial cell death (5).Interestingly, also in humans with severe RSV disease, a largenumber of neutrophils are recruited to the lungs (6), and caspase-dependent apoptosis of lung epithelial cells is observed (7).Similarly, in the pneumonia virus of mice (PVM) mouse model forsevere RSV disease (8), pulmonary neutrophil recruitment and

caspase-3 activation are associated with enhanced lung perme-ability and clinical signs of illness (8, 9). These findings suggestthat the coordinated actions of proinflammatory and proapoptoticmediators may serve to augment the pathologic responses char-acteristic of severe pneumovirus disease, although the nature ofthe mediators linking these responses are not yet fully understood.Granzymes (Gzms) are effector molecules that may contribute to

both inflammatory and apoptotic features of RSV disease. Gzmsare serine proteases located in granules of CTLs and NK cells andhave traditionally been considered to be crucial to the antiviralimmune response [reviewed in (10–12)]. Gzms are delivered intothe cytosol of target cells by the pore-forming protein perforin, butintracellular trafficking is also known to be facilitated by adeno-virus and bacterial toxins (13, 14). Of the five Gzms found inhumans (GzmA, -B, -H, -K, and -M), the tryptase GzmA andaspartase GzmB are the most thoroughly characterized. It has beenshown that GzmB induces classical (caspase-dependent) apoptosisin target cells, and in vitro micromolar concentrations of GzmA, inthe presence of perforin, can induce caspase-independent DNAdamage (10, 15–17). However, Metkar et al. (18) reported recentlythat GzmA-mediated cytotoxicity at reduced, more physiologicconcentrations may be relatively low (15, 16). Instead, activesoluble GzmA at lower concentrations induces the release ofproinflammatory cytokines such as IL-1b, IL-6, and TNF-a frommonocyte/macrophage, fibroblast, and epithelial cell lines viaa perforin-independent mechanism (18, 19). This raises the pos-sibility that Gzms have a broad functionality in the human im-mune response to viral infection in vivo, including activation of(caspase-dependent) cell death and inflammation.Recently, our group has demonstrated that tracheal aspirates

from mechanically ventilated children with severe RSV infectioncontain high levels of biologically active soluble GzmA and GzmB(20). To investigate whether these Gzms may actively participatein the pathogenesis of RSV disease, specifically promoting pro-gression to ALI/ARDS by inducing proinflammatory and caspase-dependent proapoptotic activity, we explored lung inflammatory

*Pediatric Intensive Care Unit, Emma Children’s Hospital, †Departments of Pulmo-nology and Experimental Immunology, and xLaboratory for Experimental Oncologyand Radiobiology, Academic Medical Center, Amsterdam, The Netherlands; ‡De-partment of Pediatrics, Upstate Medical University, Syracuse, NY 13210; and {Lab-oratory of Allergic Diseases, National Institute of Allergy and Infectious Diseases,National Institutes of Health, Bethesda, MD 20814

Received for publication September 16, 2009. Accepted for publication November 4,2009.

This work was supported in part by Ars Donandi and National Institute of Allergy andInfectious Diseases Division of Intramural Research Grant Z01-AI000943 (to. H.F.R.).

Address correspondence and reprint requests to Reinout A. Bem, Pediatric IntensiveCare Unit, Emma Children’s Hospital, Academic Medical Center, P.O. Box 22660,1100 DD, Amsterdam, The Netherlands. E-mail address: r.a.bem@amc.uva.nl

Abbreviations used in this paper: ALI/ARDS, acute lung injury and acute respiratorydistress syndrome; BALF, bronchoalveolar lavage fluid; Gzm, granzyme; PVM,pneumonia virus of mice; RSV, respiratory syncytial virus.

www.jimmunol.org/cgi/doi/10.4049/jimmunol.0903029

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

and injury markers in acute pneumovirus infection in GzmA and/or GzmB-cluster knockout mice.

Materials and MethodsVirus and mouse strains

The fully pathogenic PVM strain J3666 was originally obtained from Dr. A.J. Easton (University of Warwick, Coventry, U.K.) from virus stocksoriginating at the Rockefeller University. Virulence was maintained bycontinuous passage in mice (21). Virus titer in stock aliquots was 12 3 107

copies of PVM per milliliter, as determined by quantitative PCR (seebelow), and PVM inocula per mouse were as indicated.

Nine- to- 12-week-old female mice were used in all of the experiments.C57BL/6 wild-type mice (CD45.1+) were originally obtained from TheJackson Laboratory (Bar Harbor, ME) and bred in house. The GzmA2/2,GzmB2/2 [generated by the group of Dr. T. Ley (22)], and GzmA2/2

GzmB2/2mice on aC57BL/6 backgroundwere originally obtained fromDr.M. M. Simon (23) and bred in house (24). GzmA+/2GzmB+/2 mice wereused to serve as a primary positive control for the GzmA2/2GzmB2/2mice:GzmA+/2GzmB+/2 mice were generated by crossing GzmA2/2 withGzmB2/2 mice, resulting in a similar mixed C57BL/6 substrain (B6J andB6-III) background as the GzmA2/2GzmB2/2 mice (22, 25). Importantly,the GzmB2/2 (and GzmA2/2GzmB2/2) mice lack the entire GzmB cluster,which includes GzmC, -D, and -F, due to knockdown by a retained phos-phoglycerate kinase I gene promoter (PGK-neo) cassette (26).

Animal protocols

The animal protocols as described belowwere approved by the Animal Careand Experimental Research Committee of the Academic Medical Center,Amsterdam, The Netherlands.

PVM diluted in RPMI 1640 medium (Invitrogen, Paisley, U.K.) in a totalvolume of 80 ml was delivered via intranasal inhalation to isoflurane an-esthetized mice. At day 3, 7, or 8 after inoculation, the mice were eu-thanized with i.p. pentobarbital (120 mg/kg), and the left lung wasremoved and flash-frozen in liquid nitrogen for homogenization. Bron-choalveolar lavage was performed in the right lung by instilling fourseparate 0.5 ml aliquots of 0.9% NaCl containing 0.6 mM EDTA. Thebronchoalveolar lavage fluid (BALF) was maintained at 4˚C and spun at200 3 g for 10 min. The supernatants were stored at 280˚C as individualaliquots, and the cell pellet was processed for total cell counts and dif-ferentials. For histological analysis on day 8 after PVM inoculation, theright lung was fixed with 10% formalin immediately after the bron-choalveolar lavage procedure.

Measurements

Microarray and flow cytometry analysis. The microarray study as describedpreviously (27) was probed to determine the expression profile of GzmAand GzmB transcripts in whole-lung tissue in response to PVM infection.To determine intracellular expression of GzmA and GzmB in effectorlymphocytes, BALF leukocytes in FACS buffer (0.5% BSA and 0.02%potassium-EDTA in PBS) were probed with FITC-labeled anti-CD8,PerCP-Cy5-labeled anti-NK1.1, and allophycocyanin-labeled anti-CD3 (allanti-mouse from eBioscience, San Diego, CA). Next, cells were fixed andpermeabilized with fixation/permeabilization solution (BD Biosciences,Franklin Lakes, NJ) and stained with PE-labeled mouse anti-mouse GzmAmAb (Santa Cruz Biotechnology, Heidelberg, Germany) or with PE-labeledrat anti-mouse GzmB mAb (eBioscience) in 10% normal rat serum to blocknonspecific binding. The FACSCalibur flow cytometer (BD Biosciences)and FlowJo software (Tree Star, Ashland, OR) were used for analysis.

Lung virus titers. Lungs were homogenized in Trizol reagent to extractRNA, according to the manufacturer’s description (Invitrogen). RNA wasresuspended in diethyl pyrocarbonate-treated water. After DNase I treat-ment of 2 mg of RNA (Applied Biosystems, Foster City, CA), cDNAsynthesis was performed using a random hexamer cDNA synthesis kit(Applied Biosystems). Copies of the PVM sh gene (GenBank AY573815)were detected in quantitative real-time PCR reactions containing 1 ml ofcDNA, Taqman PCR Master Mix (Applied Biosystems), 77 nM TAMRAprobe (59-6FAM-CGCTGATAATGGCCTGCAGCATAMRA-39), and 200nM primers (59-GCCTGCATCAACACAGTGTGT-39 and 59-GCCTGAT-GTGGCAGTGCTT-39). The gapdh housekeeping gene was detected incDNA samples using rodent gapdh primers (100 nM) and VIC probe (200nM) (Ambion, Austin, TX; Applied Biosystems). Standard curves withknown concentrations of the full-length sh gene and gapdh decatemplate(Applied Biosystems) were used for quantification. Results are expressedas copies of PVM sh per 109 copies of gapdh (28).

Lung inflammatory markers. Total BALF leukocyte counts were performedon an aliquot of the BALF cells using a Burker bright line countingchamber. Cytospins were stained with Romanovsky (Diff-Quick), anddifferential WBC counts were obtained by counting 200 leukocytes usinga standard light microscope. Lung cytokines in BALF were measured bya multiplex fluorescent bead assay for TNF-a, IFN-g, IL-6, IL-1b, KC, andMIP-2 (Luminex, R&D Systems, Minneapolis, MN)

Lung caspase activation. Caspase-3 immunohistochemistry on 5-mm-thickparaffin-embedded lung sections was performed using rabbit anti-cleaved(active) caspase-3 mAb (Cell Signaling Technology, Danvers, MA) fol-lowed by secondary labeling with anti-rabbit IgG HRP polymer (Immu-nologic, Duiven, The Netherlands). HRP activity was detected witha peroxidase substrate kit (Vector Nova RED, Vector Laboratories, Bur-lingame, CA). Quantification of the overall lung caspase-3 activity in lunghomogenates corrected for lung weight was performed using a caspase-3/CPP32 fluorometric assay kit (Biovision, Mountain View, CA) as de-scribed previously (29).

Lung permeability. Leakageof the highm.w. serumproteina-macroglobulinor IgM into the lungs was measured by immunoassay for mouse a-macro-globulin (Life Diagnostics, West Chester, PA) or mouse IgM (Bethyl Lab-oratories, Montgomery, TX) performed on BALF.

Lung histology.A blinded histology analysis for evidence of lung injury wasperformed on H&E-stained lung sections. A pathological grade for eachlung section was determined according to the following criteria: 1 =presence of areas with cellular alveolitis involving 25% or less of the lungparenchyma; 2 = lesions involving 25–50% of the lung; 3 = lesions in-volving 50% or more of the lung. An additional point was given whenthere were alveolar wall alterations such as thickening or capillary con-gestion outside the aforementioned inflammatory areas.

Clinical response. Total bodyweight andaclinical scoring systempreviouslyvalidated in multiple PVMmouse studies were used: 1 = healthy, no signs ofillness, 2 = subtle ruffled fur, 3 = evident ruffled furwith hunched posture, 4 =evident lethargy with abnormal breathing pattern, 5 = moribund, 6 = death[modified fromCook et al. (30)]. The end point for sacrifice used in this studywas a score of 5 and/or loss of.20% of starting body weight.

Statistical analysis

To detect differences among groups, we performed either unpaired t test (forcomparison of two groups) or one-way ANOVA (for comparison of mul-tiple groups) followed by least significance difference post hoc analysis onlog10-transformed data. A p value of ,0.05 was considered statisticallysignificant.

ResultsExpression of GzmA and GzmB in effector lymphocytes inlungs of PVM-infected mice

Severe pneumovirus (RSV) disease in children is associated withincreased local levels of biologically active GzmA and GzmB andcellular expression of GzmBpredominantly in BALFCTLs (20). Todetermine whether infection with the murine pneumovirus PVMinduces GzmA and GzmB expression in mice, we analyzed thetemporal gene expression profiles of transcripts encoding GzmAand GzmB from a microarray study performed on total lung RNAfrom C57BL/6 mice inoculated with a nonlethal dose of PVM (27).Transcripts encoding both GzmA and GzmB were detected in lungtissue from PVM-infectedmice (Fig. 1A). Levels of bothGzmA andGzmB transcripts reached a peak at day 7 and remained abovebaseline levels until day 14 after inoculation. ImmunoreactiveGzmA was detected predominantly in CD32/NK1.1+ (NK) cells,and GzmB in CD8+/CD3+ (CTL) cells isolated from BALF at day 7of infection (Fig. 1B).

PVM infection in GzmA- and GzmB-cluster knockout mice

On the basis of theGzmgene expression profile,wehypothesized thatiftheGzmresponsecontributeseitherpositivelyornegativelytoseverepneumovirus disease, then this would become apparent at or beyondpeakexpressionatday7.Toaddressthisquestion,weinvestigatedlungvirus titers, caspase-3 activation, and cytokine release in wild-type,GzmAorGzmBsingle-knockout, andGzmAGzmBdouble-knockoutmice on days 7 and 8 after inoculation with PVM (Fig. 2).

932 ROLE OF GRANZYMES IN PNEUMOVIRUS DISEASE

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

Surprisingly, whole-lung virus titers from PVM-infected wild-typemicewere indistinguishable from thosemeasured in lungs fromPVM-infected GzmA2/2, GzmB2/2, and GzmA2/2GzmB2/2

mice, on day 7 and 8 after inoculation as determined by quantitative

real-time PCR (Fig. 2A). However, caspase-3 activity in lung ho-mogenates of the GzmA2/2GzmB2/2 mice was significantly re-duced compared with those of the wild-type and GzmA2/2mice ondays 7 and 8 after PVM inoculation (Fig. 2B). Previously, we have

FIGURE 1. PVM infection induces expression of GzmA and GzmB. A, Gene expression profiles of transcripts encoding GzmA and GzmB in total lung

RNA of C57BL/6 mice (RNA from five mice pooled per time point) at time zero and days 1–7, 10, 14, 21, and 28 after inoculation with PVM J3666 (30

PFU). Peak of GzmA and GzmB expression is at day 7 (green highlight). B, Histogram plots of intracellular GzmA and GzmB expression in BALF CD8+

CD3+ cells (CTLs, blue plot), NK1.1+CD32 (NK cells, green plot), and CD82CD3+ (CD82 T lymphocytes, red plot) of PVM-infected mice at day 7

(BALF leukocytes from three mice pooled).

FIGURE 2. Lung response to PVM infection in GzmA- and GzmB-cluster knockout mice. A, Virus titer in the lung, expressed as number of PVM sh

copies per 109 gapdh copies. B, Lung homogenate caspase-3 activity (arbitrary fluorescence units). C, Concentration of a-macroglobulin in BALF. D,

BALF total polymorphonuclear neutrophil (PMN) counts. E, Concentration of IL-6 (pg/ml) in BALF. F, Concentration of TNF-a (pg/ml) in BALF in the

wild-type (B6 wt), GzmA2/2, GzmB2/2, and GzmA2/2GzmB2/2 mice on day 3 (n = 3 per group), 7 (n = 6 per group), and 8 (n = 4–6 per group) after

PVM inoculation (6 3 103 copies). pp , 0.05. Data are shown as bar graphs depicting mean and SE.

The Journal of Immunology 933

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

identified that reducedwhole-lung caspase-3 activity is associatedwithdecreased leakageof thehighm.w. serumproteina-macroglobulin intothe alveolar spaces in severe PVMdisease inmice (9). In linewith this,GzmB2/2 and GzmA2/2GzmB2/2 mice had reduced concentrationsof a-macroglobulin in BALF, as compared with wild-type andGzmA2/2mice, although this reached statistical significance only forthe GzmB2/2mice (Fig. 2C).PVM infection in mice promotes robust neutrophil influx into the

lungs (8). Neutrophil recruitment, as determined by BALF totalneutrophil counts, was not different among thewild-type,GzmA2/2,and GzmB2/2mice; in contrast, neutrophil recruitment was delayedin the GzmA- and the GzmB-cluster double-knockout mice (Fig.2D). However, this delay was not associated with the release of theproinflammatory cytokines IL-1b (not detected in any of the mice),IL-6 (Fig. 2E), and TNF-a (Fig. 2F) in the lungs.

PVM infection in GzmA2/2GzmB2/2 and GzmA+/2GzmB+/2

mice

The findings thus far suggest that Gzms may alter the response toPVM, including caspase-mediated apoptosis and lung permeability,but do not affect viral clearance or the release of the proin-flammatory cytokines IL-1b, IL-6, and TNF-a. We found a morepronounced phenotype in the double-knockout mice, which isconsistent with previous findings (31). As such, we proceededwith a focus on the GzmA2/2GzmB2/2 mice, and to account forthe differences in the genetic background of the substrains thatmay have affected the results above, we compared the lung re-sponses to PVM infection of GzmA2/2GzmB2/2 mice to those ofGzmA+/2GzmB+/2 mice (22, 25).

Viral titers and caspase activation. Consistent with our findingsabove, whole-lung virus titers remained indistinguishable (Fig. 3A),

although there was a marked reduction in caspase-3 activity in lunghomogenates in the GzmA2/2GzmB2/2 mice as compared with theGzmA+/2GzmB+/2 mice (Fig. 3B). To investigate the lung tissue dis-tribution of active caspase-3 in the mice, we performed immunohis-tochemical localization for cleaved caspase-3 in lung tissue sections ofday 8 after PVM inoculation. The lung tissues of GzmA+/2GzmB+/2

mice showed widespread cleaved caspase-3 staining in alveolar wallcells and among infiltrating leukocytes (Fig. 3C). Interestingly, al-though PVM replication is detected primarily in bronchial epithelialcells (27), cleaved caspase-3 was not evident in these cells.

Lung permeability. As compared with the GzmA+/2GzmB+/2

mice, the GzmA2/2GzmB2/2 mice displayed diminished con-centrations of a-macroglobulin and IgM in BALF, both markersfor lung permeability (9) (Fig. 4A, 4B).

Lung inflammation. As compared with the GzmA+/2GzmB+/2

mice, the GzmA2/2GzmB2/2 mice had decreased total neutrophilcounts in BALF on day 7 after PVM inoculation (Fig. 5A).However, this apparent delay in neutrophil influx was not asso-ciated with the release of cytokines IL-1b (not detected in any ofthe mice), IL-6 (Fig. 5B), TNF-a (Fig. 5C), MIP-2 (Fig. 5E), orKC (Fig. 5F). IFN-g is among the critical regulators of the neu-trophil response in PVM infection (32) and was the only cytokinethat correlated with the delay in neutrophil recruitment in theGzmA2/2GzmB2/2 mice (Fig. 5D).

Lung histology. On day 8 after inoculation with PVM, the lungtissues from all of the mice showed extensive peri-bronchial andperi-vascular cellular infiltrates, as well as areas with acutealveolitis and alveolar wall alterations such as thickening and (mildhemorrhagic) capillary congestion, similar to findings describedpreviously (8). However, cellular alveolitis and capillary conges-tion was more diffuse in the GzmA+/2GzmB+/2 mice, when

FIGURE 3. PVM clearance and apoptosis in GzmA-

and GzmB-cluster knockout mice. A, Virus titer in the

lung, expressed as number of PVM sh copies per 109

gapdh copies. B, Lung homogenate caspase-3 activity

(arbitrary fluorescence units) in the GzmA+/2GzmB+/2

and GzmA2/2GzmB2/2mice on day 7 (n = 6 per group)

and 8 (n = 6 per group) after PVM inoculation (63 103

copies). pp, 0.05.Data are shown as box plots depicting

the median, interquartile range, and full range. C, Rep-

resentative images of cleaved caspase-3 immunohisto-

chemistry in lung tissues of the GzmA+/2GzmB+/2 and

GzmA2/2GzmB2/2 mice on day 8 after PVM in-

oculation. Note the relative positive staining in alveolar

wall cells (arrows) as compared with that in bronchial

epithelial cells in the GzmA+/2GzmB+/2 mice.

934 ROLE OF GRANZYMES IN PNEUMOVIRUS DISEASE

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

compared with GzmA2/2GzmB2/2 mice (Fig. 6): Mean (6SE)histopathology scores were 3.2 (0.3) in the GzmA2/+GzmB2/+

mice and 2.2 (0.2) in the GzmA2/2GzmB2/2 mice (p , 0.05).

Clinical response. Finally, to address whether these Gzm-mediatedalterations of the lung response to PVM affect the clinical phe-notype, we followed clinical scores and body weight up to day 12after inoculation with a lethal inoculum of PVM. The GzmA2/2

GzmB2/2 mice showed a statistically significant delay in reachingthe maximum clinical response to PVM (Fig. 7A). All of theGzmA+/2GzmB+/2 mice reached the defined end point (clinicalscore of 5 and/or body weight loss . 20%) at day 10 after PVMinoculation, whereas at this point this was reached by 67% of theGzmA2/2GzmB2/2 mice (p, 0.05). A delayed response to PVMinfection among the GzmA2/2GzmB2/2 mice was also observedwhen weight loss alone as a sole, purely objective measure of theclinical response was examined (Fig. 7B).

DiscussionThe primary goal of this study was to determine whether Gzmspromote pathophysiologic responses, including acute lung injury,in pneumovirus-infected mice. In this work, we detect both GzmA(predominantly in NK cells) and GzmB (predominantly in CTLs) inassociation with severe respiratory sequelae induced by the mousepneumovirus pathogen, PVM. We found that mice deficient forboth GzmA and the GzmB cluster (GzmA2/2GzmB2/2) havea delayed clinical response to fatal PVM infection, although Gzmexpression has no impact on virus replication or clearance per se.Instead, we determine that the absence of Gzms results in delayedneutrophil recruitment, diminished activation of caspase-3, andreduced lung permeability.The Gzm system exploited by effector lymphocytes plays

a major role in host antiviral immunity by activating cell deathprograms in infected target cells (10–12). Providing the first in vivoevidence for Gzm-mediated viral elimination, Mullbacher et al.(31) showed that survival and clearance of the mouse poxvirus,ectromelia, in mice lacking both GzmA and the GzmB cluster wasmarkedly depressed as compared with wild-type mice. Systematicchallenge of Gzm-deficient mice to different virus families has notbeen undertaken, but replication of cytomegalovirus and herpessimplex virus have also been found to be controlled by Gzm ac-tivity (33, 34). At the same time, there is much evidence thateffector lymphocyte-mediated immunopathogenic mechanismsform the basis of enhanced disease in certain virus infections, andthis includes the development of severe human pneumovirus(RSV) disease (35, 36). For example, both CTLs and NK cells arerecruited to the lungs during primary RSV infection in mice (36–38). Depletion of CTLs in mice reduces clinical signs of RSVdisease, despite delaying viral clearance (36), whereas i.v. transferof high-dose CTLs into CTL-depleted mice causes severe hem-orrhage and neutrophilic inflammation (39). Frey et al. (40)studied the PVM mouse model, which, unlike RSV infection inmice, reproduces several features of severe RSV disease in

FIGURE 4. Lung permeability response to PVM infection in GzmA-

and GzmB-cluster knockout mice. Concentrations of a-macroglobulin (A)

and IgM (B) in BALF of the GzmA+/2GzmB+/2 and GzmA2/2GzmB2/2

mice on day 7 (n = 6 per group) and 8 (n = 6 per group) after PVM in-

oculation (6 3 103 copies). pp , 0.05. Data are shown as box plots de-

picting the median, interquartile range, and full range.

FIGURE 5. Lung inflammatory response to PVM infection in GzmA-

and GzmB-cluster knockout mice. Total polymorphonuclear neutrophil

(PMN) counts (A), and concentrations of IL-6 (B), TNF-a (C), IFN-g (D),

MIP-2 (E), and KC (F) in BALF of the GzmA+/2GzmB+/2 and GzmA2/2

GzmB2/2 mice on day 7 (n = 6 per group) and 8 (n = 6 per group) after

PVM inoculation (63 103 copies). pp, 0.05. Data are shown as box plots

depicting the median, interquartile range, and full range.

FIGURE 6. Lung tissue response to PVM infection in GzmA- and

GzmB-cluster knockout mice. Representative images of H&E-stained lung

tissue from the GzmA+/2GzmB+/2 and GzmA2/2GzmB2/2 mice on day 8

after PVM inoculation (6 3 103 copies). Magnification 3100.

The Journal of Immunology 935

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

humans, including robust viral replication, proinflammatory al-terations in the airways and alveoli, alveolar injury, and, impor-tantly, overt clinical signs of illness (8). They found similarresults, namely, that T cell-deficient mice have delayed clearanceof PVM but reduced pulmonary inflammatory infiltrates andweight loss (40). These findings suggested that effector lympho-cytes may contribute to the immunopathophysiology associatedwith human RSV infection, but whether the Gzm system is in-volved herein remained unclear from those studies.Recently, we found high levels of active GzmA and GzmB in

tracheal aspirates from children with severe RSV disease (20).Surprisingly, in the current study, virus titers in lung tissue werenot altered by Gzm targeting, despite a major reduction in lungcaspase-3 activation in GzmA‑ and GzmB-cluster‑deficient mice.Similarly, Frey et al. (40) found that mice deficient for perforin,which facilitates granzyme trafficking into virus-infected targetcells, did not alter PVM lung titers. These findings suggest thatalternative pathways may play more prominent roles in PVMclearance. Certain viruses, such as adenovirus and parainfluenzavirus type 3, are known to have evolved strategies that specificallyinhibit GzmB (41, 42), and poxvirus can block caspase signaling(11, 43). Interestingly, in our study, we found little to no caspase-3activation in bronchial epithelial cells, clearly the most importantsite for PVM replication (27). However, it is currently unknownwhether PVM directly induces antiapoptotic pathways in infectedbronchial epithelial cells.Similar to previous findings (9), PVM infection was associated

with increased caspase-3 activation in alveolar wall cells, and thiswas markedly reduced in Gzm-deficient mice. As such, thisfinding may explain at least in part the observed decreased lungpermeability in the Gzm-deficient mice. Accumulating evidenceimplicates caspase-dependent apoptosis as an important mecha-nism leading to loss of alveolar capillary barrier integrity andultimately enhanced lung permeability (44). Patients who die withALI/ARDS have increased caspase-3 staining in alveolar wallcells (45), and numerous experimental studies modeling both in-direct and direct ALI/ARDS have shown that enhanced apoptosisof lung epithelial cells is associated with leakage of serum pro-teins into the lungs and histopathological alterations, which can beattenuated by inhibiting proapoptotic pathways (44, 46, 47). Ourfindings suggest that the Gzm proapoptotic pathway can contrib-ute to the development of alveolar injury in severe pneumovirusdisease. Interestingly, both GzmA and GzmB mRNA expression isaugmented in patients with septic ARDS and in mice intra-tracheally instilled with LPS (47, 48), suggesting that the Gzmpathway may also be of relevance to bacteria-induced ALI/ARDS.In addition to proapoptotic functions of Gzms, recent renewed

attention has been given to their potential proinflammatory effects,specifically of GzmA (18, 19). Metkar et al. (18) found that active

human GzmA induces the release of IL-1b, IL-6, and TNF-a frommonocytes, which can be attenuated by the IL-1b convertase (cas-pase-1) inhibitor WEHD-FMK, and similarly that mouse macro-phages produce IL-1b in response to treatment with active mouseGzmA. Interestingly, GzmA‑ and GzmB-cluster‑deficient miceshowed a delayed lung neutrophil response to PVM infection, butthis was not associated with the release of IL-1b, IL-6, TNF-a, KC,andMIP-2. The fact that we did not detect IL-1b in the BALF of anyof the mice may suggest that the above-mentioned mechanism ofGzmA-mediated proinflammatory activity through caspase-1 maynot be directly relevant in the PVM natural infection model. IFN-g,potentially released from activated NK cells, has been identified asa critical regulator of neutrophil influx into the lungs in severe PVMdisease in mice (32) and followed the same pattern as the totalneutrophil counts in the GzmA‑ and GzmB-cluster‑deficient mice.However, it remains to be elucidated whether mouse Gzm may di-rectly mediate the release of IFN-g, as, for example, human GzmAdoes not induce IFN-g specifically in monocytes (18). In addition,one needs to recognize that a number of different events, includingadhesion,migration, and clearance, affect the number of neutrophilsin the lungs. In this light, the potent proteolytic degradation of ex-tracellular matrix proteins by both GzmA and GzmB is of particularinterest, because Gzm-mediated remodeling of the extracellularenvironment has been shown to have important consequences forcellular migration and survival (49).As stated above, in the most current paradigm, GzmA elicits

predominantly proinflammatory activity, whereas GzmB mediatescaspase-dependent apoptosis, based mostly on observations in vitro(15, 16). However, deciphering the specific functions of the in-dividual Gzm family members in vivo has proven to be difficultand not straightforward (11, 16, 18, 31). Interestingly, studies inmice have shown that the combined deficiency for GzmA and theGzmB cluster can result in a more pronounced or paradoxicallydiminished phenotype, as compared with single-gene (GzmA orthe GzmB cluster) deficiency (18, 31), suggesting complex andsynergistic interaction between Gzms. The interpretation of ex-periments involving Gzm-targeted mice is further complicated bythe fact that the GzmB knockout mice lack the entire GzmBcluster, which includes GzmC, -D, and -F (26). In addition, thedifferences in substrain background among wild-type and single-and double-knockout mice may affect host responses to patho-gens. Finally, in vivo, through indirect mechanisms involving in-flammatory cells and extracellular matrix remodeling, Gzms mayinfluence several processes, including cell death pathways (49), towhich they are not directly attributed (e.g., although GzmA doesnot directly cause caspase activation, in our study, GzmA single-knockout mice showed some reduction in lung caspase-3 activity).Taking into account these several issues, we have primarily fo-cused on the effects of combined deficiency for GzmA and the

FIGURE 7. Clinical response to PVM infection in GzmA- and GzmB-cluster knockout mice. A, Kaplan‑Meier curves showing the percentages of GzmA+/2

GzmB+/2 (blackdots,n=9) andGzmA2/2GzmB2/2 (white dots,n=9)mice reaching the endpoint of a clinical score of5 (moribund) and/orweight loss of.20%

after PVM inoculation (13 104 copies). p, 0.05 by log-rank (Mantel‑Cox) test.B, Percentages of baseline (day 0)weight of the GzmA+/2GzmB+/2 (black dots)

and GzmA2/2GzmB2/2 (white dots) mice after PVM inoculation (13 104 copies). pp, 0.05. Bars show the mean.

936 ROLE OF GRANZYMES IN PNEUMOVIRUS DISEASE

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

GzmB cluster, but this obviously limits our interpretation of thefunctions of the individual Gzm family members in acute pneu-movirus infection.Finally, it is important to recognize that mouse and human Gzm

differ in structure, enzyme activity, and inhibitory regulation (50).For example, human GzmB is more cytotoxic than mouse GzmBin vitro (50). In addition, unlike mouse GzmB, the human formrequires Bid cleavage for full caspase-3 activation and is con-trolled by the inhibitor PI-9 (50). Furthermore, the evolution of theGzm system, potentially driven by host-specific pathogens, maycompromise some of the meaning of any study of Gzm responsesto human virus pathogens inoculated in mice. As such, the use ofthe natural mouse pneumovirus pathogen PVM in this work mayrepresent an important advantage, because it permits us to explorethe roles and functions of Gzms in response to a physiologicallyand evolutionarily relevant challenge. However, although PVMand RSVare both pneumoviruses, they are not the same virus, andwe need to stress that there is no one animal model that displaysall of the features of human RSV disease.In conclusion, pneumovirus infection in mice induces expression

of GzmA and GzmB by effector lymphocytes. The combineddeficiency of GzmA and the GzmB cluster results in a delayedprogression of pneumovirus disease by reducing alveolar injury inmice, without altering viral clearance. We speculate that targetingof the Gzm system in humans may be beneficial in diminishingRSV‑ALI/ARDS immunopathogenesis.

AcknowledgmentsWe thank P. J. van de Berg for expert advice on FACS analysis and J. de Jong

for assistance with animal breeding.

DisclosuresThe authors have no financial conflicts of interest.

References1. Hall, C. B., G. A. Weinberg, M. K. Iwane, A. K. Blumkin, K. M. Edwards,

M. A. Staat, P. Auinger, M. R. Griffin, K. A. Poehling, D. Erdman, et al. 2009.The burden of respiratory syncytial virus infection in young children. N. Engl.J. Med. 360: 588–598.

2. Shay, D. K., R. C. Holman, R. D. Newman, L. L. Liu, J. W. Stout, andL. J. Anderson. 1999. Bronchiolitis-associated hospitalizations among US chil-dren, 1980-1996. JAMA 282: 1440–1446.

3. Dahlem, P., W. M. van Aalderen, M. E. Hamaker, M. G. Dijkgraaf, andA. P. Bos. 2003. Incidence and short-term outcome of acute lung injury inmechanically ventilated children. Eur. Respir. J. 22: 980–985.

4. Nokes, J. D., and P. A. Cane. 2008. New strategies for control of respiratorysyncytial virus infection. Curr. Opin. Infect. Dis. 21: 639–643.

5. Ware, L. B., and M. A. Matthay. 2000. The acute respiratory distress syndrome.N. Engl. J. Med. 342: 1334–1349.

6. Wang, S. Z., and K. D. Forsyth. 2000. The interaction of neutrophils with re-spiratory epithelial cells in viral infection. Respirology 5: 1–10.

7. Welliver, T. P., J. L. Reed, and R. C. Welliver Sr. 2008. Respiratory syncytialvirus and influenza virus infections: observations from tissues of fatal infantcases. Pediatr. Infect. Dis. J. 27: S92–S96.

8. Rosenberg, H. F., and J. B. Domachowske. 2008. Pneumonia virus of mice:severe respiratory infection in a natural host. Immunol. Lett. 118: 6–12.

9. Bem, R. A., J. B. van Woensel, A. P. Bos, A. Koski, A. W. Farnand,J. B. Domachowske, H. F. Rosenberg, T. R. Martin, and G. Matute-Bello. 2009.Mechanical ventilation enhances lung inflammation and caspase activity ina model of mouse pneumovirus infection. Am. J. Physiol. Lung Cell. Mol.Physiol. 296: L46–L56.

10. Lieberman, J. 2003. The ABCs of granule-mediated cytotoxicity: new weaponsin the arsenal. Nat. Rev. Immunol. 3: 361–370.

11. Russell, J. H., and T. J. Ley. 2002. Lymphocyte-mediated cytotoxicity. Annu.Rev. Immunol. 20: 323–370.

12. Bots, M., and J. P. Medema. 2006. Granzymes at a glance. J. Cell Sci. 119:5011–5014.

13. Browne, K. A., E. Blink, V. R. Sutton, C. J. Froelich, D. A. Jans, andJ. A. Trapani. 1999. Cytosolic delivery of granzyme B by bacterial toxins: ev-idence that endosomal disruption, in addition to transmembrane pore formation,is an important function of perforin. Mol. Cell. Biol. 19: 8604–8615.

14. Froelich, C. J., K. Orth, J. Turbov, P. Seth, R. Gottlieb, B. Babior, G. M. Shah,R. C. Bleackley, V. M. Dixit, and W. Hanna. 1996. New paradigm for lym-

phocyte granule-mediated cytotoxicity. Target cells bind and internalize gran-zyme B, but an endosomolytic agent is necessary for cytosolic delivery andsubsequent apoptosis. J. Biol. Chem. 271: 29073–29079.

15. Froelich, C. J., J. Pardo, and M. M. Simon. 2009. Granule-associated serineproteases: granzymes might not just be killer proteases. Trends Immunol. 30:117–123.

16. Trapani, J. A., and P. I. Bird. 2008. A renaissance in understanding the multipleand diverse functions of granzymes? Immunity 29: 665–667.

17. Martinvalet, D., D. M. Dykxhoorn, R. Ferrini, and J. Lieberman. 2008. GranzymeA cleaves a mitochondrial complex I protein to initiate caspase-independent celldeath. Cell 133: 681–692.

18. Metkar, S. S., C. Menaa, J. Pardo, B. Wang, R. Wallich, M. Freudenberg, S. Kim,S. M. Raja, L. Shi, M. M. Simon, et al. 2008. Human and mouse granzyme Ainduce a proinflammatory cytokine response. Immunity 29: 720–733.

19. Sower, L. E., G. R. Klimpel, W. Hanna, and C. J. Froelich. 1996. Extracellularactivities of human granzymes. I. Granzyme A induces IL6 and IL8 productionin fibroblast and epithelial cell lines. Cell. Immunol. 171: 159–163.

20. Bem, R. A., A. P. Bos, M. Bots, A. M. Wolbink, S. M. van Ham, J. P. Medema,R. Lutter, and J. B. van Woensel. 2008. Activation of the granzyme pathway inchildrenwith severe respiratory syncytial virus infection.Pediatr. Res. 63: 650–655.

21. Domachowske, J. B., C. A. Bonville, J. L. Gao, P. M. Murphy, A. J. Easton, andH. F. Rosenberg. 2000. The chemokine macrophage-inflammatory protein-1 aand its receptor CCR1 control pulmonary inflammation and antiviral host de-fense in paramyxovirus infection. J. Immunol. 165: 2677–2682.

22. Heusel, J. W., R. L. Wesselschmidt, S. Shresta, J. H. Russell, and T. J. Ley. 1994.Cytotoxic lymphocytes require granzyme B for the rapid induction of DNAfragmentation and apoptosis in allogeneic target cells. Cell 76: 977–987.

23. Simon, M. M., M. Hausmann, T. Tran, K. Ebnet, J. Tschopp, R. ThaHla, andA. Mullbacher. 1997. In vitro- and ex vivo-derived cytolytic leukocytes fromgranzyme A x B double knockout mice are defective in granule-mediated apo-ptosis but not lysis of target cells. J. Exp. Med. 186: 1781–1786.

24. Bots, M., I. G. Kolfschoten, S. A. Bres, M. T. Rademaker, G. M. de Roo,M. Kruse, K. L. Franken, M. Hahne, C. J. Froelich, C. J. Melief, et al. 2005. SPI-CI and SPI-6 cooperate in the protection from effector cell-mediated cytotox-icity. Blood 105: 1153–1161.

25. Ebnet, K., M. Hausmann, F. Lehmann-Grube, A. Mullbacher, M. Kopf,M. Lamers, and M. M. Simon. 1995. Granzyme A-deficient mice retain potentcell-mediated cytotoxicity. EMBO J. 14: 4230–4239.

26. Revell, P. A., W. J. Grossman, D. A. Thomas, X. Cao, R. Behl, J. A. Ratner,Z. H. Lu, and T. J. Ley. 2005. Granzyme B and the downstream granzymes Cand/or F are important for cytotoxic lymphocyte functions. J. Immunol. 174:2124–2131.

27. Bonville, C. A., N. J. Bennett, M. Koehnlein, D. M. Haines, J. A. Ellis,A. M. DelVecchio, H. F. Rosenberg, and J. B. Domachowske. 2006. Respiratorydysfunction and proinflammatory chemokines in the pneumonia virus of mice(PVM) model of viral bronchiolitis. Virology 349: 87–95.

28. Ellis, J. A., B. V. Martin, C. Waldner, K. D. Dyer, J. B. Domachowske, andH. F. Rosenberg. 2007. Mucosal inoculation with an attenuated mouse pneu-movirus strain protects against virulent challenge in wild type and interferon-greceptor deficient mice. Vaccine 25: 1085–1095.

29. Bem, R. A., A. W. Farnand, V. Wong, A. Koski, M. E. Rosenfeld, N. vanRooijen, C. W. Frevert, T. R. Martin, and G. Matute-Bello. 2008. Depletion ofresident alveolar macrophages does not prevent Fas-mediated lung injury inmice. Am. J. Physiol. Lung Cell. Mol. Physiol. 295: L314–L325.

30. Cook, P. M., R. P. Eglin, and A. J. Easton. 1998. Pathogenesis of pneumovirusinfections in mice: detection of pneumonia virus of mice and human respiratorysyncytial virus mRNA in lungs of infected mice by in situ hybridization. J. Gen.Virol. 79: 2411–2417.

31. Mullbacher, A., P. Waring, R. Tha Hla, T. Tran, S. Chin, T. Stehle, C. Museteanu,and M. M. Simon. 1999. Granzymes are the essential downstream effectormolecules for the control of primary virus infections by cytolytic leukocytes.Proc. Natl. Acad. Sci. USA 96: 13950–13955.

32. Bonville, C. A., C. M. Percopo, K. D. Dyer, J. Gao, C. Prussin, B. Foster,H. F. Rosenberg, and J. B. Domachowske. 2009. Interferon-g coordinates CCL3-mediated neutrophil recruitment in vivo. BMC Immunol. 10: 14.

33. Pereira, R. A., M. M. Simon, and A. Simmons. 2000. Granzyme A, a non-cytolytic component of CD8+ cell granules, restricts the spread of herpes simplexvirus in the peripheral nervous systems of experimentally infected mice. J. Virol.74: 1029–1032.

34. Riera, L., M. Gariglio, G. Valente, A. Mullbacher, C. Museteanu, S. Landolfo,and M. M. Simon. 2000. Murine cytomegalovirus replication in salivary glandsis controlled by both perforin and granzymes during acute infection. Eur.J. Immunol. 30: 1350–1355.

35. Domachowske, J. B., and H. F. Rosenberg. 1999. Respiratory syncytial virusinfection: immune response, immunopathogenesis, and treatment. Clin. Micro-biol. Rev. 12: 298–309.

36. Graham, B. S., L. A. Bunton, P. F. Wright, and D. T. Karzon. 1991. Role of Tlymphocyte subsets in the pathogenesis of primary infection and rechallengewith respiratory syncytial virus in mice. J. Clin. Invest. 88: 1026–1033.

37. Hussell, T., and P. J. Openshaw. 1998. Intracellular IFN-g expression in naturalkiller cells precedes lung CD8+ T cell recruitment during respiratory syncytialvirus infection. J. Gen. Virol. 79: 2593–2601.

38. Openshaw, P. J. 1989. Flow cytometric analysis of pulmonary lymphocytes frommice infected with respiratory syncytial virus. Clin. Exp. Immunol. 75: 324–328.

39. Cannon, M. J., P. J. Openshaw, and B. A. Askonas. 1988. Cytotoxic T cells clearvirus but augment lung pathology in mice infected with respiratory syncytialvirus. J. Exp. Med. 168: 1163–1168.

The Journal of Immunology 937

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

40. Frey, S., C. D. Krempl, A. Schmitt-Graff, and S. Ehl. 2008. Role of T cells invirus control and disease after infection with pneumonia virus of mice. J. Virol.82: 11619–11627.

41. Andrade, F., H. G. Bull, N. A. Thornberry, G.W.Ketner, L. A. Casciola-Rosen, andA. Rosen. 2001. Adenovirus L4-100K assembly protein is a granzyme B substratethat potently inhibits granzyme B-mediated cell death. Immunity 14: 751–761.

42. Sieg, S., L. Xia, Y. Huang, and D. Kaplan. 1995. Specific inhibition of granzymeB by parainfluenza virus type 3. J. Virol. 69: 3538–3541.

43. Zhou, Q., S. Snipas, K. Orth, M. Muzio, V. M. Dixit, and G. S. Salvesen. 1997.Target protease specificity of the viral serpin CrmA. Analysis of five caspases. J.Biol. Chem. 272: 7797–7800.

44. Martin, T. R., N. Hagimoto, M. Nakamura, and G. Matute-Bello. 2005. Apo-ptosis and epithelial injury in the lungs. Proc. Am. Thorac. Soc. 2: 214–220.

45. Albertine, K. H., M. F. Soulier, Z. Wang, A. Ishizaka, S. Hashimoto,G. A. Zimmerman, M. A. Matthay, and L. B. Ware. 2002. Fas and fas ligand areup-regulated in pulmonary edema fluid and lung tissue of patients with acutelung injury and the acute respiratory distress syndrome. Am. J. Pathol. 161:1783–1796.

46. Kawasaki, M., K. Kuwano, N. Hagimoto, T. Matsuba, R. Kunitake, T. Tanaka,T. Maeyama, and N. Hara. 2000. Protection from lethal apoptosis in lipopolysac-charide-induced acute lung injury in mice by a caspase inhibitor. Am. J. Pathol. 157:597–603.

47. Kitamura, Y., S. Hashimoto, N. Mizuta, A. Kobayashi, K. Kooguchi, I. Fujiwara,and H. Nakajima. 2001. Fas/FasL-dependent apoptosis of alveolar cells afterlipopolysaccharide-induced lung injury in mice. Am. J. Respir. Crit. Care Med.163: 762–769.

48. Hashimoto, S., A. Kobayashi, K. Kooguchi, Y. Kitamura, H. Onodera, andH. Nakajima. 2000. Upregulation of two death pathways of perforin/granzymeand FasL/Fas in septic acute respiratory distress syndrome. Am. J. Respir. Crit.Care Med. 161: 237–243.

49. Buzza, M. S., and P. I. Bird. 2006. Extracellular granzymes: current perspectives.Biol. Chem. 387: 827–837.

50. Kaiserman, D., C. H. Bird, J. Sun, A. Matthews, K. Ung, J. C. Whisstock,P. E. Thompson, J. A. Trapani, and P. I. Bird. 2006. The major human and mousegranzymes are structurally and functionally divergent. J. Cell Biol. 175: 619–630.

938 ROLE OF GRANZYMES IN PNEUMOVIRUS DISEASE

by guest on January 5, 2019http://w

ww

.jimm

unol.org/D

ownloaded from