Identification of a Novel Virulence Factor in Recombinant ...

JOURNAL OF VIROLOGY, Sept. 2007, p. 9490–9501 Vol. 81, No. 170022-538X/07/$08.00�0 doi:10.1128/JVI.00364-07Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Identification of a Novel Virulence Factor in Recombinant PneumoniaVirus of Mice�

Christine D. Krempl,1,2* Anna Wnekowicz,1 Elaine W. Lamirande,2 Giw Nayebagha,1Peter L. Collins,2 and Ursula J. Buchholz2

Department of Virology, Institute for Medical Microbiology and Hygiene, University of Freiburg, Freiburg, Germany,1 andLaboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases,

Bethesda, Maryland 20892-80072

Received 20 February 2007/Accepted 6 June 2007

Pneumonia virus of mice (PVM) is a murine relative of human respiratory syncytial virus (HRSV). Here wedeveloped a reverse genetics system for PVM based on a consensus sequence for virulent strain 15. Recom-binant PVM and a version engineered to express green fluorescent protein replicated as efficiently as thebiological parent in vitro but were 4- and 12.5-fold attenuated in vivo, respectively. The G proteins of HRSV andPVM have been suggested to contribute to viral pathogenesis, but this had not been possible to study in adefined manner in a fully permissive host. As a first step, we evaluated recombinant mutants bearing a deletionof the entire G gene (�G) or expressing a G protein lacking its cytoplasmic tail (Gt). Both G mutants replicatedas efficiently in vitro as their recombinant parent, but both were nonpathogenic in mice at doses that wouldotherwise be lethal. We could not detect replication of the �G mutant in mice, indicating that its attenuationis based on a severe reduction in the virus load. In contrast, the Gt mutant appeared to replicate as efficientlyin mice as its recombinant parent. Thus, the reduction in virulence associated with the Gt mutant could notbe accounted for by a reduction in viral replication. These results identified the cytoplasmic tail of G as avirulence factor whose effect is not mediated solely by the viral load. In addition to its intrinsic interest, arecombinant virus that replicates with wild-type-like efficiency but does not cause disease defines optimalproperties for vaccine development.

Human and bovine respiratory syncytial viruses (HRSV andBRSV, respectively) are the most extensively characterizedrepresentatives of the genus Pneumovirus, subfamily Pneumo-virinae, family Paramyxoviridae. HRSV and BRSV are impor-tant agents of severe lower respiratory tract infections in theirrespective hosts. Reverse genetics systems are available forHRSV and BRSV (9, 10) and have been useful for determiningthe functions of viral proteins and RNA signals on a molecularlevel. However, in vivo characterization of the contribution ofthese factors to viral pathogenesis has been hampered by thedifficulty of performing defined experiments in the naturalhost. Mice, cotton rats, and monkeys are frequently used asexperimental animal models of HRSV infection, but viral rep-lication is only semipermissive and disease is either lacking orminimal (29). Chimpanzees approach humans with regard tothe magnitude of HRSV replication and disease but are incon-venient, expensive, and scarce (45). BRSV has been used as asurrogate model of HRSV infection. The genome sequence isclosely related to that of HRSV (9), and infection of calves canresult in a disease resembling that caused by HRSV in humans(reviewed in reference 29). However, the virulence of BRSV inexperimental infections can be variable, and there are limita-tions to the manageability of this large-animal model and theavailability of immunological tools (38, 43).

Pneumonia virus of mice (PVM), a murine member of the

genus Pneumovirus, is another potential surrogate for HRSV(15). PVM was first isolated by Horsfall and Hahn in the 1930swhen mice were used during an attempt to isolate new humanrespiratory viruses (21, 22). Mice were inoculated with naso-pharyngeal aspirates from patients with acute noninfluenzarespiratory tract disease, followed by serial passage of mouselung homogenates from animal to animal. Following severalpassages, the mice frequently developed morbidity and fatalpneumonia. PVM was isolated and identified as a potentialmurine pathogen rather than a human pathogen. Thus, thenatural history of PVM is unclear, and reports of naturallyoccurring infection and disease are rare and limited to immu-nocompromised animals (36, 44). Infection of immunocompe-tent animals is thought to be inapparent or latent. However,PVM infections of other rodents have been described, andantibodies specific to PVM or a serologically related virus havebeen detected in many rodent species as well as in other mam-mals, including humans (16, 20, 28, 31, 33).

Two laboratory strains, strain 15 and strain J3666, have beendescribed for PVM. An isolate of strain 15 has been used as aprototype in several experiments (13, 34), although this nowappears to be a nonpathogenic mutant that was attenuated bypassage in primate cell culture. However, we recently showedthat the isolate of strain 15 that is available from the AmericanType Culture Collection (ATCC) is a virulent virus that thusresembles the original strain 15 isolate described by Horsfalland Hahn (21–23). The second laboratory strain of PVM,strain J3666, is virulent in mice and has been maintained en-tirely by passage in mice (12, 34). Presumably it was isolated atapproximately the same time and place as strain 15 (42). How-

* Corresponding author. Present address: Institute of Virology andImmunobiology, University of Wurzburg, Versbacher Strasse 7,D-97078 Wurzburg, Germany. Phone: (49) 931-201-49586. Fax: (49)931-201-49553. E-mail: [email protected].

� Published ahead of print on 13 June 2007.

9490

on February 12, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

ever, there are no reports of the isolation in the literature, andthus, its origin is somewhat unclear. Strain J3666 has been usedfor in vivo studies of pathogenicity.

Recently, complete genomic sequences were reported forthe virulent strain 15, its nonpathogenic variant, and strainJ3666 (24, 42). The prototypic virulent PVM strain 15 has anonsegmented negative-sense RNA genome of 14,886 nucleo-tides (nt) (24). The genome contains 10 genes that encode 12proteins. Eleven PVM proteins appear to correspond to thefollowing HRSV proteins: the nucleocapsid protein N; thephosphoprotein P; the large polymerase protein L; the M2-1and M2-2 proteins, which are encoded by overlapping openreading frames (ORFs) in the M2 mRNA and are involved inRNA synthesis (3, 11, 18); the matrix protein M; the nonstruc-tural NS1 and NS2 proteins, involved in inhibiting the hostinterferon system (7, 27, 37, 39, 43); and three transmembranesurface glycoproteins, namely, the G protein, which is involvedin attachment, the F protein, which is involved in membranefusion, and the small hydrophobic protein, which might be anantiapoptotic factor (46). The 12th PVM protein does not havea counterpart in HRSV or BRSV; it is a 137-amino-acid pro-tein of unknown function that is encoded by a second ORF inthe P mRNA (2).

Studies of PVM pathogenesis to date have compared strainJ3666 as the virulent strain with the nonpathogenic version ofstrain 15 as the nonpathogenic strain (see, e.g., references 13,34, and 42). The recent sequence analysis mentioned aboveshowed that these two viruses differ by 59 nucleotide substitu-tions involving 37 amino acid substitutions, complicating theidentification of the relevant differences. In addition, the Ggene of the nonpathogenic variant of PVM strain 15 appears tohave sustained a single-nucleotide insertion in a run of four Uresidues at gene positions 169 to 172 (GenBank accessionnumber AY729016), which immediately follows codon 29 inthe G ORF. This would shift translation into a second framethat terminates after six additional codons. However, 5 ntupstream of this termination site, there is a second AUG in theG ORF (codon 34), and initiation there would yield a G pro-tein that lacks the first 33 amino acids at its N terminus. Thepneumovirus G protein is a type II glycoprotein: in the case ofPVM, amino acids 1 to 34 are predicted to constitute its cyto-plasmic tail, amino acids 35 to 59 would constitute its signal/anchor sequence, and the remainder of the molecule would bethe ectodomain. Thus, initiation at Met-34 would eliminateessentially the entire cytoplasmic tail. This mutation was yetanother candidate to be involved in the attenuation phenotypeof the nonpathogenic version of strain 15, but it was necessaryto evaluate this in a defined genetic background.

In the present study, we developed a reverse genetic systemfor the virulent PVM strain 15 and used this system to studythe contribution of the G protein, and in particular its cyto-plasmic tail, to virulence. We constructed two G mutants thatwere found to be highly attenuated, but for strikingly differentreasons. Specifically, a virus lacking the complete G proteinwas nonpathogenic due to drastic decreases in replication andviral load in vivo. In contrast, a virus lacking only the cytoplas-mic tail of G was nonpathogenic despite a level of in vivoreplication indistinguishable from that of its wild-type parent.These mutants demonstrate a major role of the cytoplasmic tailof G in pathogenicity in vivo and illustrate the utility of this

reverse genetic system for the study of pneumovirus virulencefactors in a susceptible in vivo model.

MATERIALS AND METHODS

Viruses and cells. PVM virulent strain 15 and its recombinant derivatives werepropagated in BHK-21 (ATCC CCL-10) cells as described previously (24).BHK-21 cells and the BHK-derived BSR-T7/5 cells (9) were maintained inGlasgow minimal essential medium (MEM) (Invitrogen GIBCO), and Vero cells(ATCC CCL-81) were maintained in Eagle’s MEM (Biochrom), all supple-mented with 10% fetal calf serum. Virus titers were determined by a plaque assayon Vero cells under 0.8% methylcellulose for 5 days at 32°C, followed by fixationwith 80% methanol. Plaques were visualized by incubation with a polyclonalrabbit serum raised against the PVM G protein, followed by incubation withbiotinylated goat anti-rabbit immunoglobulin G (IgG) (Pierce) as a second an-tibody. Bound antibodies were detected by incubation with streptavidin coupledto horseradish peroxidase (HRP) (BD Pharmingen), followed by color develop-ment by incubation with 3,3�-diaminobenzidine tetrahydrochloride (DAB) sub-strate (Merck). Growth curves were performed in replicate wells. At the indi-cated time points, duplicate wells were harvested by scraping the cells into theoverlying medium. The cells were pelleted by low-speed centrifugation, subjectedto three cycles of freeze-thawing in a small volume of medium, and centrifugeda second time. The supernatant was harvested, combined with that from the firstcentrifugation, flash frozen, and stored for titration. Each aliquot was titrated induplicate, and the mean was calculated.

RNA isolation, RT-PCR, and nucleotide sequencing. Total cellular RNA orvirion-associated RNA (vRNA) was isolated as described previously (24). Forreverse transcription (RT), RNA was mixed with 5 pmol of a specific primer,incubated for 10 min at 70°C, and assembled for an RT reaction that wasperformed using Superscript II reverse transcriptase (Invitrogen) at 50°C for 1 h,followed by digestion of DNA-RNA hybrids with RNase H at 37°C for 20 min.PCR was performed with Pfx polymerase (Invitrogen) at 68°C. Nucleotide se-quence analysis was performed using an ABI 3100 sequencer with the BigDyeTerminator ready reaction kit, version 1.1 (Applied Biosystems).

Construction of PVM cDNAs. All cDNAs are based on the consensus se-quence of the virulent PVM strain 15 (GenBank accession number AY729016)(24). For construction of support plasmids expressing the nucleoprotein N, thephosphoprotein P, and the putative transcription factor M2-1, the respectiveORFs were amplified from vRNA by RT-PCR using specific primers and werecloned into the NcoI restriction site of the T7 expression vector pTM1 such thatthe ATG of the NcoI site became the translational start site of the ORF (17). Theexpression plasmid carrying the ORF of the RNA polymerase was assembledfrom two overlapping cDNA fragments derived by RT-PCR such that it couldalso be used as a shuttle vector for construction of the full-length plasmid (Fig.1A). Fragment 1, of 2,895 nt (fragment 4 of the full-length construct), encom-passed a naturally occurring BamHI site, which contained the last nucleotide ofthe translation initiation codon of the L ORF, and an XhoI site (nt 8475 to 8,480and 11374 to 11,379, respectively, of PVM 15). This fragment was cloned into theBamHI-XhoI window of pT7 HMPV L, a plasmid encoding the L protein ofhuman metapneumovirus (HMPV) (5), thereby replacing the HMPV sequence.Fragment 2, of 3,541 nt (fragment 5 of the full-length construct), encompassedthe residual part of the L gene including the XhoI site and the L gene end(GE), as well as the complete PVM 15 trailer sequence and 35 nt of thesequence of the hepatitis D virus (HDV) ribozyme. This fragment was clonedinto a separate pTM1 vector. After confirmation of the sequence, the secondL fragment was transferred into the XhoI-StuI window of pTM pvmL1, thuscreating pTM pvmL.

For construction of the plasmid encoding the complete antigenome of PVM 15(Fig. 1), plasmid pKS II 3/7 HDV-T7ter, containing the HDV ribozyme followedby a terminator for the T7 RNA polymerase (14), was modified to contain apolylinker with NheI, AgeI, and BstBI sites, as well as the first 17 nt of the PVML gene, including the gene start (GS) signal and the naturally occurring BamHIsite. This polylinker was cloned into the XmaI-RsrII window of p3/7 HDV-T7terto accept the BamHI-RsrII fragment of pTM pvmL and three overlappingsubgenomic cDNA fragments, described below, that were generated by RT-PCRfrom vRNA. Fragment 1 (2,727 nt) comprised the promoter for the T7 RNApolymerase, three G residues to enhance promoter activity, the PVM leader,NS1, NS2, N, and 441 nt of the P gene, flanked on the upstream end by an XmaIsite and on the downstream end by a naturally occurring NheI site at position2702 (relative to the PVM genome). Fragment 2 (1,807 nt) overlapped with theNheI site on the upstream end and contained the residual 472 nt of the P geneand the complete M and SH genes. An AgeI marker restriction site that wouldbe located in the intergenic region between the SH and G genes in the final

VOL. 81, 2007 RECOMBINANT PNEUMONIA VIRUS OF MICE 9491


http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

plasmid was added on the downstream site of fragment 2 (Fig. 1B). Fragment 3encompassed the complete G, F, and M2 genes. On the upstream end, thefragment overlapped with the added AgeI site, whereas a BstBI restriction sitethat would be located in the M2–L intergenic region was added to the down-stream end. The subgenomic fragments were cloned and the sequences con-firmed. Confirmed fragments were added to the corresponding restriction sites ofthe vector described above containing the PVM L gene, thus creating pPVM.The sequence of the complete assembled antigenome was confirmed in its en-tirety.

To construct a variant encoding green fluorescent protein (GFP), pPVM-GFP7, the pPVM plasmid was modified by addition of a transcription cassettecontaining the ORF for enhanced GFP (Clontech, Inc.) to the AgeI restrictionsite in the SH–G intergenic region (Fig. 1C). To create the GFP insertioncassette, the GFP ORF was amplified by PCR to add an AgeI site, followed bythe GS sequence of the PVM N gene (which is identical to that of the P gene),and then the Kozak sequence, on the upstream side and the GE sequence of thePVM F gene, followed by an AgeI site, on the downstream end (Fig. 1C). Thestructure of the GFP transcription cassette was confirmed by sequencing.

To construct a mutant in which the G gene was deleted (pPVM-GFP-�G), the

F-M2 portion of pPVM-GFP7 was amplified with a forward primer that added anAgeI site at the upstream side of the F GS and the same reverse primer that wasused to add the BstBI site to the downstream end of the M2 gene, as shown withfragment 3 of pPVM1 (Fig. 1A). The PCR product was digested with AgeI andBstBI and then cloned into the AgeI-BstBI window of pPVM-GFP7. This ineffect replaced the AgeI-BstBI fragment containing the G, F, and M2 genes (Fig.1A) with an AgeI-BstBI fragment containing only the F and M2 genes, thusdeleting the G gene.

To construct a mutant in which the cytoplasmic tail of G was deleted (pPVM-GFP-Gt), the pPVM-GFP7 cDNA was subjected to PCR using a forward primerthat contained the AgeI restriction site and G GS sequence, followed by 22 nt ofthe G ORF starting at nt 178 of the G gene. PCR amplification using this forwardprimer and the same reverse primer as for the generation of fragment 3 resultedin a cDNA containing a truncated G gene with the GS sequence fused to nt 178of the G gene (thereby deleting the intervening167 nt), the complete naturallyoccurring G gene downstream of nt 178, and the G–F intergenic region, followedby the complete downstream part of fragment 3 of pPVM1. The insert was thendigested with AgeI and BstBI and cloned into the corresponding window ofpPVM, resulting in pPVM-Gt. The antigenomic cDNAs of pPVM-�G and

FIG. 1. Construction of plasmids expressing the antigenomes of rPVM and rPVM-GFP7. (A) Map of the antigenome of rPVM showing thelocations of the viral genes and, below, the five overlapping cDNA fragments (numbered 1 to 5) used to construct its cDNA. PT7, T7 RNApolymerase promoter; �, hepatitis delta virus ribozyme; TT7, T7 RNA polymerase terminator. Unique restriction enzyme sites used to constructthe cDNA are given, with their nucleotide positions in parentheses; the AgeI and BstBI sites that were created during construction are underlined.(B) Nucleotide changes introduced into rPVM to create AgeI (upper panel) and BstBI (lower panel) markers in the SH–G and M2–L intergenicregions, respectively. Predicted transcription GS and GE signals are boldfaced, intergenic sequences are roman, restriction enzyme sites areunderlined, and nucleotide substitutions and additions in rPMV are italicized, while deletions are indicated by gaps. (C) Insertion of a transcriptioncassette encoding enhanced GFP into the AgeI site of rPVM, placing GFP as the seventh gene in the 3�-to-5� order. The GFP ORF (openrectangle, not drawn to scale) was modified by PCR to add a GS sequence derived from the N gene (which also is identical in the P gene) on theupstream side of the ORF and a GE sequence derived from the F gene on the downstream side (each underlined). The fragment was flanked byAgeI restriction sites whose positions in the final cDNA are indicated.

9492 KREMPL ET AL. J. VIROL.


http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

pPVM-Gt were further modified by insertion of a GFP transcription cassette intothe AgeI site as described for the generation of pPVM GFP7, thereby generatingpPVM-GFP-�G and pPVM-GFP-Gt, respectively. All regions of the final full-length cDNA clones that had been amplified by PCR were confirmed completelyby sequence analysis.

Recovery of recombinant viruses. Plasmids were transfected into BSR T7/5cells as described previously for recovery of recombinant BRSV (rBRSV) andrHMPV (5, 9), with minor modifications to adjust for optimal growth conditionsfor PVM. Briefly, BSR T7/5 cells in 35-mm-diameter wells were transfected with5 �g of plasmid pPVM or a derivative, 2 �g each of pTM pvmN and pTM pvmP,and 1 �g each of pTM pvmM2-1 and pTM pvmL using Lipofectamine 2000(Invitrogen) according to the manufacturer’s instructions. After incubation at37°C overnight, the transfection medium was replaced with Glasgow MEMsupplemented with 5% fetal calf serum, and the culture was maintained at 32°C.Four days after transfection, cells were trypsinized, cocultivated with an equalnumber of BHK-21 cells, and incubated for as long as 6 days to yield passage 1virus.

Analysis of secreted proteins. To isolate soluble proteins, two-thirds of themedium overlying an infected-cell monolayer was harvested and subjected tolow-speed centrifugation to remove cell debris, followed by ultracentrifugation at100,000 � g for 90 min to remove virus particles. Soluble proteins contained inthe resulting supernatant were concentrated 50-fold using an Amicon centrifugalfilter device (Millipore). After the addition of an equal volume of 2� sodiumdodecyl sulfate (SDS)/Tris buffer (nonreducing), the samples were heat dena-tured, subjected to SDS-polyacrylamide gel electrophoresis (PAGE) (10% gel),and transferred to supported nitrocellulose membranes.

To purify virus particles, the infected cells from the same culture were scrapedinto the remaining one-third of the overlying medium and processed by freeze-thawing as described under “Viruses and cells” above. Virus particles werepelleted from the supernatant by ultracentrifugation through a 30% sucrosecushion (at 100,000 � g for 90 min). Following lysis with 1� SDS/Tris buffer(nonreducing), the samples were heat denatured and subjected to SDS-PAGE asdescribed above.

The membranes were incubated with one of the following antibody prepara-tions for detection of viral proteins: biotinylated polyclonal HRSV-specific goatantibodies (Biogenesis, Poole, United Kingdom), followed by detection withstreptavidin-conjugated HRP; rabbit serum raised against the PVM G protein,followed by detection with biotinylated goat anti-rabbit IgG (Pierce) and strepta-vidin-conjugated HRP; or serum from a mouse that had recovered from infectionwith PVM, followed by detection with HRP-conjugated rabbit anti-mouse IgG(DAKO). Bound antibodies were visualized and documented by the chemilumi-nescence detection system ChemiDoc XRS (Bio-Rad).

Mouse studies. For infection with PVM, BALB/c mice were anesthetized witha mixture of ketamine and xylazine (100 mg and 5 mg per kg of body weight,respectively, given intraperitoneally) and inoculated intranasally with the indi-cated viruses diluted with phosphate-buffered saline to a final volume of 80 �l.For determination of virulence, mice were weighed daily and monitored by visualinspection twice per day. The main visual disease signs were ruffled fur, hunch-ing, and listless behavior. Mice were sacrificed by cervical dislocation if theweight loss exceeded 25% of the weight on day 0 or if the animals were obviouslyin extremis. To determine the extent of virus replication, mice were sacrificed onthe indicated day, and lungs and nasal turbinates were removed separately forvirus quantification by a plaque assay. All experiments involving animals wereapproved by the animal ethics committee of the Regierungsprasidium Freiburg.

RESULTS

Reverse genetic system for virulent PVM strain 15. A plas-mid encoding the complete antigenomic RNA of virulent PVMstrain 15 (pPVM) was constructed from five overlapping sub-genomic RT-PCR fragments as illustrated in Fig. 1A. As mark-ers, a unique AgeI site was added to the SH–G intergenicregion by the substitution of 1 and addition of 3 nt, increasingits length from 2 nt to 5 nt, and a unique BstBI site was addedto the M2–L intergenic region by the addition of 2 and deletionof 1 nt, increasing its length from 9 nt to 10 nt. The encodedrPVM antigenome would be 14,890 nt long, compared to14,886 nt for its biological parent. Apart from these markers,the sequence of the rPVM cDNA was identical to the consen-

sus sequence of its biological parent (GenBank accession num-ber AY729016) (24).

To facilitate virus detection, we also made a version ofpPVM containing a gene encoding GFP. PCR was used toconstruct a transcription cassette containing the ORF of en-hanced GFP under the control of the GS signal of N (whichalso is identical to that of the P gene) and the GE signal of F,flanked by AgeI sites (Fig. 1C). This cassette was inserted inthe AgeI site of pPVM. Thus, the resulting plasmid, pPVM-GFP7, contained 11 genes, with the added GFP gene located atposition 7 between the SH and G genes. The insert increasedthe genome length to 15,642 nt.

Recovery of recombinant PVM. Plasmid pPVM or pPVM-GFP7 was transfected, together with T7 expression plasmidsencoding the N, P, M2-1, and L support proteins, into BSR-T7/5 cells that constitutively express T7 RNA polymerase (5,9). In cultures transfected with pPVM-GFP7, individual GFP-expressing cells were detected 2 days after transfection. Duringthe following days, foci of infected cells formed, followed byspreading over the residual monolayer. Final titers of bothrecombinant viruses reached more than 107 PFU/ml after thesecond passage. To confirm the recombinant origin of therecovered viruses, total intracellular RNA was isolated frominfected BHK-21 cells and RT-PCR was used to amplify afragment spanning nt 3312 (in the N gene) to 5981 (in the Fgene) (positions relative to PVM15), which included the AgeImarker restriction site and the added GFP gene. Detection ofthe predicted PCR products depended on the presence ofreverse transcriptase, indicating that the products representedRNA (data not shown). Restriction digestion confirmed thepresence of the AgeI site in rPVM and the presence of theGFP insert in rPVM-GFP7, whereas neither was present inthe biologically derived parent (Fig. 2A).

The in vitro replication efficiencies of rPVM and rPVM-GFP7 were compared to that of biologically derived PVMstrain 15 (PVM15) in a multicycle time course in BHK-21 cells.Cell monolayers were infected in duplicate at a multiplicity ofinfection (MOI) of 0.01 PFU/cell, and cells and medium su-pernatants were harvested at 24-h intervals and stored forsubsequent viral titration by a plaque assay as described inMaterials and Methods. As shown in Fig. 2B, the kinetics andviral yields of PVM15, rPVM, and rPVM-GFP7 were essen-tially the same, reaching titers of more than 107 PFU/ml by day7. Monitoring of the expression of GFP by fluorescent micros-copy revealed a continuous spread of rPVM-GFP7 up to day 4,at which time almost 100% of the cells were expressing GFP(Fig. 2C). Interestingly, the monolayer stayed viable up to theend of the experiment at day 7, continuing to produce infec-tious particles and with no sign of syncytium formation. Thiswas also observed for BHK-21 monolayers infected withPVM15 and rPVM and for Vero cell monolayers infected witheach of the three viruses (data not shown).

Replication and virulence of rPVM and rPVM-GFP7 inmice. To investigate the virulence of the recombinant viruses,BALB/c mice were infected intranasally with graded doses ofbiologically derived PVM15 (500 PFU and 50 PFU), rPVM(5,000 PFU, 500 PFU, and 50 PFU), or rPVM-GFP7 (5,000PFU and 500 PFU). The mice were observed closely andweighed daily.

All of the mice that received 500 PFU of PVM15 or rPVM



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

died or were sacrificed in extremis by day 8 and day 10, re-spectively (Fig. 3A). However, the onset of symptomatic dis-ease (day 5 versus day 6) (Fig. 3B) and occurrence of firstfatalities (day 7 versus day 8) (Fig. 3A) were delayed 1 to 2 daysfor rPVM, indicating a modest degree of attenuation of rPVMcompared to its biological parent. This attenuation was morepronounced in infections involving 50 PFU. In this case, all ofthe animals infected with rPVM survived (Fig. 3A), althoughsubstantial disease was evident, as indicated by weight loss(Fig. 3B). In contrast, following infection with 50 PFU of thebiological wild-type virus, most of the mice had to be sacrificedin extremis, and just one mouse survived. The median lethaldose was calculated to be 40 PFU for PVM15 and 160 PFU forrPVM, indicating a fourfold attenuation for rPVM (35).

Although the insertion of GFP into the genome of rPVMdid not have a detectable effect on replication in vitro, it had amodest attenuating effect in vivo. Fifty percent of the miceinfected with 500 PFU of rPVM-GFP7 survived (Fig. 3A),although they experienced substantial weight loss (Fig. 3B). Inthis comparison, the onset of symptomatic disease, as indicated

by ruffled fur and weight loss, was delayed 1 day for rPVM-GFP7 compared to rPVM (Fig. 3B and data not shown). Basedon this 50% lethal dose (LD50) of 500 PFU, rPVM-GFP7 was3-fold and 12.5-fold attenuated compared to its recombinantand biological parents, respectively (Fig. 3B). At a dose of5,000 PFU of rPVM-GFP7, the kinetics of weight loss anddeath were essentially the same as those for the same dose ofrPVM (Fig. 3B).

In order to determine the efficiency of replication of therecombinant viruses in vivo, BALB/c mice were inoculatedintranasally with 500 PFU of rPVM, rPVM-GFP7, or the bio-logical virus PVM15. Mice were sacrificed on days 3 and 6 afterinfection, lungs and nasal turbinates were harvested, and virustiters were determined by a plaque assay. For biologically de-rived PVM15, the mean titer in the lungs was approximately1 � 105 PFU/g of tissue on day 3 (Fig. 4A) and approximately4 � 106 PFU/g on day 6 (Fig. 4B). In comparison to PVM15,the mean titer of rPVM in the lungs was reduced threefold(approximately 4 � 104 PFU/g) and twofold (2 � 106 PFU/g)on days 3 and 6, respectively, reflecting the delayed onset of

FIG. 2. In vitro analysis of recovered rPVM and rPVM-GFP7. (A) RT-PCR and restriction analysis of viral RNA from BHK-21 cells infectedwith the biological wild-type virus PVM15 (lanes 1 and 2), the wild-type virus rPVM (lanes 3 and 4), or rPVM-GFP7 (lanes 5 and 6). The RT-PCRproducts were made using primers that hybridized within the N and F genes, spanning PVM positions 3312 to 5981. The amplified fragments weredigested with AgeI (lanes marked �) or left untreated (lanes marked �) and electrophoresed on a 1% agarose gel. The predicted sizes of thefragments are given on the left. The last lane represents a 1-kbp DNA ladder. (B) Multistep growth kinetics of biological and recombinant PVMsin vitro. Replicate monolayers of BHK-21 cells were infected with the biological parental virus PVM15, rPVM, or rPVM-GFP7 at an input MOIof 0.01 PFU per cell. At the indicated time points, the cells and medium were harvested, freeze-thawed, clarified, and flash frozen. Virus titers weredetermined by a plaque assay. Error bars, standard errors of the means. (C) Cells from the experiment for which results are shown in panel B wereexamined for GFP expression on days 1 to 4 using an inverted fluorescence microscope.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

disease and lethality observed in Fig. 3. In the nasal turbinateson day 3, virus was detectable in three out of five mice infectedwith PVM15 and in just one mouse infected with rPVM (Fig.4C), likewise indicating a delayed replication of rPVM. Incontrast, on day 6, both viruses replicated to approximately thesame titer in the nasal turbinates, reaching average titers ofapproximately 2 � 105 PFU/g of tissue (Fig. 4D).

In animals infected with rPVM-GFP7, virus was detected onday 3 in the lungs of only half of the animals and was notdetected in the nasal turbinates of any of the animals (Fig. 4Aand B). On day 6, the mean titer of rPVM-GFP7 in the lungswas approximately sevenfold reduced compared to that for itsparent, rPVM (Fig. 4B), whereas there was no significant dif-ference in the nasal turbinates. In a second experiment, wecompared the replication of rPVM and rPVM-GFP7 followingintranasal infection with a higher dose, 5,000 PFU (data notshown). At this higher dose, virus was detectable in the nasalturbinates and lungs of all animals on days 3 and 6: the meanviral titer of rPVM-GFP7 was threefold reduced compared tothat of rPVM on day 3, whereas the titers were essentially thesame on day 6. Thus, these data confirm that rPVM-GFP7 ismodestly more attenuated than its parent, rPVM, which in turnis modestly more attenuated than its biological parent,PVM15. These differences were greater on day 3 than on day 6.

Construction and replication of rPVM-GFP-�G and pPVM-GFP-Gt. We constructed two derivatives of pPVM-GFP7 inwhich either the complete G gene was deleted (�G) or thesequence encoding the G cytoplasmic tail was deleted (Gt)(Fig. 5B). The second construct was designed based loosely onthe predicted G gene of the nonpathogenic isolate of strain 15(Fig. 5A). However, instead of introducing a nucleotide insertand frameshift, we deleted 167 nt spanning the beginning ofthe ORF so as to move the AUG that normally is located atcodon 34 (nt 182 to 184) into position as the first AUG in theG ORF. The resulting viruses, rPVM-GFP-�G and rPVM-GFP-Gt, were readily recovered, and the deletions were con-firmed in the viral RNA of the final virus stocks by RT-PCRamplification and sequencing. In addition, an antiserum spe-cific for the PVM G protein confirmed the expression of G byrPVM-GFP-Gt and the lack of G expression in cells infectedwith rPVM-GFP-�G (Fig. 5C).

In a multistep growth study with BHK-21 cells, the kineticsof replication and the final virus yield for the �G and Gtmutants were essentially indistinguishable from those of theirimmediate parent, rPVM-GFP7, and also were very similar tothose of rPVM (Fig. 5D). Thus, the G protein is completelydispensable for PVM replication in BHK-21 cells. In addition,the N-terminal truncation did not detectably interfere withviral replication, which is consistent with previous findings forsimilar HRSV mutants in Vero cells (40, 41).

Analysis of G protein expression by rPVM. To investigatewhether the G protein of PVM is expressed in a secreted formanalogous to that of HRSV, BHK-21 cells were infected withrPVM-GFP7, rPVM-GFP-Gt, or rPVM-GFP-�G. As a posi-tive control for detecting a soluble G protein, HEp-2 cells wereinfected with HRSV and treated in parallel. Virus particles andsoluble proteins in the medium supernatants were separatedand subjected to SDS-PAGE and Western blot analysis asdescribed in Materials and Methods. HRSV proteins weredetected using polyclonal antibodies directed against the com-plete virus (Fig. 6C). In the gel pattern representing purifiedHRSV, this antiserum reacted with most of the major struc-tural proteins, including the G protein (Fig. 6C, lane V). In thepattern representing the medium supernatant from which thevirus had been removed by centrifugation, this antiserum re-acted with a single band that migrated slightly more rapidlythan virion-associated G protein and thus represented secreted

FIG. 3. Virulence of wild-type and recombinant PVM. Six- toeight-week-old BALB/c mice were infected intranasally with the indi-cated doses of PVM15, rPVM, and rPVM-GFP7 in an inoculum vol-ume of 80 �l. Mice were observed closely and weighed daily. Mice thatmet the end point regulations of the German animal protection boardwere euthanized by cervical dislocation. The results of two indepen-dent experiments, each including at least four animals per group, werecombined. (A) Survival after infection with 500 PFU (PVM15, rPVM,and rPVM-GFP7) or 50 PFU (PVM15 and rPVM) of the indicatedviruses. The total numbers of animals are given beneath the doses.(B) Body weight (relative to that on day 0, taken as 100%) followinginfection with the indicated viruses at the indicated doses.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

HRSV G protein (Fig. 6C, lane S). The absence of otherHRSV structural proteins in this lane confirmed that this bandwas not virion associated. The PVM G protein was detectedusing a G-specific rabbit antiserum (Fig. 6A) and an antiserumfrom a mouse that had recovered from PVM infection (Fig.6B). In the pattern representing purified virus from rPVM-GFP, the G-specific rabbit serum detected a pair of G proteinbands (Fig. 6A) that appear to correspond to the two G-specific bands described by Ling and Pringle (25, 26). Theconvalescent-phase mouse antiserum detected the same twobands as well as several other structural proteins (Fig. 6B).Similar species were detected for the Gt mutant, for which thepair of G-related species migrated slightly more rapidly thanthat of rPVM-GFP7. The two species of G protein were absentfrom purified �G virus, confirming their identities. However,we were not able to detect the PVM G protein in virus-de-pleted medium from cells infected with the parental virus orwith rPVM-GFP-Gt. Thus, in contrast to HRSV, the PVM Gprotein does not appear to be expressed significantly in asecreted form by wild-type virus, and deletion of the cytoplas-mic tail of G also did not result in the significant production ofa secreted form.

Replication and virulence of rPVM-GFP-�G and rPVM-GFP-Gt in vivo. In order to evaluate the effect of deletion ortruncation of the G protein on PVM replication in vivo,

BALB/c mice in groups of five were infected intranasally withrPVM-GFP-�G or rPVM-GFP-Gt at 500 PFU or 5,000 PFUper mouse. For comparison, one group of three mice received500 PFU of the parental virus rPVM-GFP7, a dose corre-sponding to the LD50. The mice were sacrificed on day 6, thelungs and nasal turbinates were removed, and virus titers weredetermined by a plaque assay (Table 1). In order to determinethe virulence of the G mutants, BALB/c mice were infectedwith doses identical to those described above and were ob-served closely and weighed daily (Fig. 7).

Although deletion of the complete G gene had no effect onvirus replication in vitro, no virus was detected in the nasalturbinates and lungs of mice 6 days after infection with 500PFU or 5,000 PFU of the �G virus (Table 1). In contrast, theGt mutant replicated with an efficiency that was indistinguish-able from that of its parent, rPVM-GFP7, resulting in virustiters of 106.1 PFU/g and 106.4 PFU/g in the lungs of miceinfected with 500 PFU and 5,000 PFU of rPVM-GFP-Gt, re-spectively, compared to 106.0 PFU/g after infection with 500PFU of rPVM-GFP7 (Table 1). Similarly, the Gt virus repli-cated to the same high titer as its parent, rPVM-GFP7, in thenasal turbinates (Table 1).

Whereas infection with 500 PFU of rPVM-GFP7 resulted indramatic weight loss and 50% mortality (Fig. 3B and 7A), andinfection with 5,000 PFU of rPVM-GFP7 was uniformly lethal

FIG. 4. Replication of rPVM and rPVM-GFP7 in the upper and lower respiratory tracts of BALB/c mice. Mice were inoculated intranasallywith 500 PFU of PVM15, rPVM, or rPVM-GFP7. Mice were sacrificed on day 3 (A and C) or day 6 (B and D) after infection, lungs (A and B)and nasal turbinates (NT) (C and D) were removed and homogenized, and viral titers were determined by a plaque assay. The mean virus titersand the percentages of animals with detectable virus are given below the corresponding graphs. Neighboring groups (indicated with brackets) werecompared using the Mann-Whitney test. Asterisks indicate statistically significant differences (P � 0.05). The results of two independentexperiments were combined in each graph.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

FIG. 5. rPVM-GFP7 viruses with mutations in the G gene. (A) Schematic diagrams (not drawn to scale) of the G genes of the biologically derived virulentand nonpathogenic isolates of PVM strain 15 (23, 34). Open rectangles, ORFs. The amino acid length of the encoded protein is given in each rectangle.Translational start (filled triangles pointing right) or stop (filled vertical bars) codons are identified by their nucleotide positions in the respective G gene. TheGS and GE signals are indicated, and nontranslated gene regions are shown as thin horizontal lines. The nucleotide sequence (negative sense) of positions 166to 190 of the G gene is given, and the inserted U residue that results in a frameshift is shown in parentheses. The complement of the translational stop codon(nucleotides 188 to 190) that is accessed by the frameshift is italicized, and the complement to the alternative start codon at nucleotides 183 to 185 (Met-34) isboldfaced. (B) Schematic representations (not drawn to scale) of rPVM-GFP-�G and rPVM-GFP-Gt. The diagram in the middle represents the rPVM-GFP7genome (filled box, GFP gene). The box above the diagram illustrates the deletion of the entire G gene to create rPVM-GFP-�G. The nucleotide sequence(negative sense) shows the GFP/F gene junction and the AgeI site (italicized) left following the 1,350-nt deletion. The sequence is numbered according to thesequence of rPVM-GFP7, and the GFP GE and F GS signals are underlined. The box below the diagram illustrates the deletion of the cytoplasmic tail of Gto create rPVM-GFP-Gt. The unmodified G gene is depicted as in panel A, with the nucleotide positions of its ends numbered according to the completerPVM-GFP7 sequence. The two potential translational start codons at positions 83 (Met-1) and 182 (Met-34) relative to the G gene sequence are indicated. Thenucleotide sequence (negative sense) shows the G GS signal (underlined), the 167-nt deletion that deletes the cytoplasmic tail, and the UAC triplet at positions182 to 184 that is the complement of the new translational initiation site (Met-34 in the original G ORF). (C) Characterization of rPVM-GFP7, rPVM-GFP-Gt,and rPVM-GFP-�G with respect to the expression of G protein. Vero cell monolayers in six-well plates were infected with serial dilutions of the indicated virusesand incubated at 32°C under a methylcellulose overlay for 5 days. (Upper panels) GFP-expressing foci were photographed without further treatment by usingan inverted fluorescence microscope. (Lower panels) This was followed by fixation and staining using a G-specific antiserum as described for the plaque assayprocedure. Each pair of upper and lower panels depicts approximately the same field of view. (D) Multistep growth kinetics of rPVM, rPVM-GFP7,rPVM-GFP-�G, and rPVM-GFP-Gt in BHK-21 cells. Duplicate monolayers were infected at an input MOI of 0.01 PFU per cell. At the indicated dayspostinfection, the medium supernatants were harvested and flash frozen, and fresh medium was added. Viral titers were determined by a plaque assay. Meansfrom two independent experiments were used to generate the diagram. Error bars, standard errors of the means.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

(Fig. 3B), the same doses of the �G or Gt mutant appeared tobe benign. Specifically, infection with 500 PFU of the �G or Gtmutant did not result in any loss of body weight (Fig. 7A) or inthe appearance of other disease signs (data not shown). Thelack of virulence associated with �G was not surprising giventhe lack of detectable virus replication, but the lack of viru-lence associated with Gt was unexpected given that it appeared

to replicate as efficiently as its parent, rPVM-GFP7. Afterinfection with 5,000 PFU of the �G or Gt mutant, the micealso appeared to be mostly disease free (Fig. 7B). However, inthe case of the Gt mutant, there appeared to be a minor weightloss of less than 10% on days 7 and 8 after infection (Fig. 7B).To further investigate possible residual virulence, additionalmice in groups of five were infected with 5,000 or 50,000 PFUof the Gt mutant (Fig. 7C). This confirmed that infection with5,000 PFU resulted in a slight dip in body weight, occurring atday 7 in this case, whereas the 50,000-PFU dose resulted in

FIG. 6. (A and B) Western blot analysis of virus-associated andsoluble secreted viral proteins. BHK-21 cells were infected with rPVM-GFP7, rPVM-GFP-Gt, or rPVM-GFP-�G. Virus particles (V) andsoluble proteins in the medium overlying the infected cells (S) werecollected separately, subjected to SDS-PAGE under nonreducing con-ditions, and transferred to nitrocellulose membranes. PVM proteinswere detected using a G-specific rabbit antiserum (A) or serum froma convalescent mouse (B). (C) As a positive control, HEp-2 cells wereinfected with HRSV strain A2. Virus particles and soluble proteinswere separated in parallel to those of PVM, and proteins were de-tected using a commercial polyclonal HRSV-specific antibody. V, virusparticles; S, supernatant; M, supernatant from mock-infected cells.

TABLE 1. Replication of rPVM-GFP-�G and rPVM-GFP-Gt inthe upper and lower respiratory tracts of BALB/c mice

VirusDosea

(PFU/animal)

No. ofanimals

% withdetectable

virus

Mean virus titerb (log10PFU/g of tissue SDc)

URT LRT

rPVM-GFP7 500 3 100 5.3 0.8 (A) 6.0 0.1 (A)rPVM-GFP-�G 500 5 0 �2.1 (B) �2.8 (B)

5,000 5 0 �2.1 (B) �2.8 (B)rPVM-GFP-Gt 500 5 100 5.8 0.7 (A) 6.1 0.2 (A)

5,000 5 100 5.2 0.3 (A) 6.4 0.2 (C)

a BALB/c mice were inoculated intranasally with 500 or 5,000 PFU of theindicated virus per mouse on day 0.

b Animals were sacrificed on day 6 postinoculation, and virus titers in the lungsand nasal turbinates were determined by a plaque assay. For infection with 500PFU, data from one of two independent experiments providing comparableresults are shown. URT, upper respiratory tract; LRT, lower respiratory tract.

c The Tukey-Kramer test was used to assign mean virus titers to similar groups(A, B, and C). Mean titers with different letters are statistically different (P �0.01) within each column.

FIG. 7. Virulence of rPVM-GFP-�G and rPVM-GFP-Gt in mice,measured by changes in body weight. Six- to eight-week-old BALB/cmice were infected intranasally with 500 PFU of rPVM-GFP7 (eightmice), rPVM-GFP-�G (nine mice), or rPVM-GFP-Gt (nine mice)(A), 5,000 PFU of rPVM-GFP-�G or rPVM-GFP-Gt (five mice pergroup) (B), or 5,000 or 50,000 PFU of rPVM-GFP-Gt (five mice pergroup), as indicated (C). Mice were observed closely and weigheddaily. Weights are expressed relative to the weight at day 0, taken as100%. When necessary, mice that met the end point regulations of theGerman animal protection board were euthanized by cervical disloca-tion. The percentage of dead animals per group is given in parenthesesin each graph. For panel A, two independent experiments involvingfour and five animals per group, respectively, were combined.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

significant weight loss that started on day 5 after infection andreached a maximum loss in excess of 20% on day 9. The miceshowed ruffled fur and reduced activity, although all of themice recovered.

To estimate the difference in virulence between rPVM-GFP7 and its Gt derivative, we noted that 500 PFU of theformer resulted in weight loss approaching 30%, with only 50%of the mice surviving (Fig. 3 and 7A), whereas 50,000 PFU ofthe latter caused weight loss approaching 30% but did notresult in any deaths. This comparison indicates that the Gtmutant was more than 100-fold attenuated compared to itsimmediate parent, rPVM-GFP7. This was not associated withany difference in replication efficiency or virus load.

DISCUSSION

We recently showed that the isolate of PVM strain 15 that isavailable from the ATCC is highly virulent in mice and thusresembles the description of the original strain 15 isolate ofHorsfall and Hahn (21–23). We determined the complete con-sensus sequence of the genome of this virus (24) and, in thepresent study, used it to develop a reverse genetic system. Thissystem was used to investigate the contribution of the viral Gprotein to replication and virulence in the BALB/c mouse, aconvenient experimental animal that is thought to be a naturalhost and is highly permissive to PVM replication and disease.

Although rPVM retained a high level of virulence in mice, itwas slightly attenuated compared to its biological parent withrespect to replication and virulence. Specifically, rPVM wasreduced twofold in replication in the lower respiratory tractand was fourfold less virulent as measured by the LD50. How-ever, it should be noted that all of the fatalities after infectionwith 50 PFU of PVM15 were euthanized in response to severeweight loss as required by the German regulations for animalrights, and thus, none of the animals died as a direct conse-quence of infection. If nature had been allowed to take itscourse, some of these animals might have survived, and thedifference between the biological and recombinant versionsmight thus have been reduced (although there still would be adifference).

There are several possible explanations for the modest re-duction in the virulence of rPVM compared to its biologicalparent. The recombinant virus differed from its biological par-ent by the addition of two marker restriction sites, an AgeI sitein the SH–G intergenic region and a BstBI site in the M2–Lintergenic region, which in each case changed the length of theintergenic region in addition to introducing nucleotide substi-tutions. While the pneumovirus intergenic regions usually ap-pear to be nonspecific spacers, modest effects on virus geneexpression and replication due to manipulation of these se-quences cannot be ruled out. A second possibility is that theconsensus sequence that we determined might be a compositeof more than one virus present in the preparation rather thana single predominant virus, a composite that is slightly subop-timal. It is also possible that this preparation of rPVM acquiredone or more adventitious mutations during recovery and pas-sage, although we note that the virus preparation used in thisstudy had been propagated only three times in BHK-21 cellsand thus was not a high-passage stock. In any event, the dif-

ference in virulence was marginal, and further work will helpdetermine the basis for this modest difference.

The reverse genetics system was used to generate rPVM-GFP7, which expresses GFP. Insertion of this additional gene,which increased the length of the genome by 752 nt (5%),resulted in sevenfold and threefold attenuation of in vivo rep-lication and virulence, respectively, compared to its immediateparent, rPVM. A comparable GFP insert in HRSV was notattenuating in vitro (47), and multiple gene insertions intoHMPV that increased its size by 30% only moderately atten-uated replication efficiency (5). However, the replication effi-ciencies of those viruses have not been evaluated in vivo. It wassomewhat surprising to find that the addition of another geneto PVM had such a small effect on replication in a highlypermissive host. The GFP backbone will represent our “wild-type” virus in future studies, permitting direct fluorescencetracking and histological characterization of pulmonary infec-tion in the natural host.

Subsequently, we used the GFP-bearing backbone to re-cover rPVM lacking the G protein altogether or encoding a Gprotein lacking its N-terminal cytoplasmic tail. The latter mu-tation was based on the G protein of the cell culture-adaptednonpathogenic variant of PVM strain 15 described byRhandhawa and colleagues (34). The G ORF of this biologi-cally derived variant contained one added nucleotide, causinga frameshift such that the main ORF presumably initiated atthe internal AUG at position 183 (Met-34). In that study,expression of a truncated protein of the predicted size wasconfirmed in a cell-free translation system programmed withsynthetic mRNA, and surface expression was confirmed byimmunofluorescence of cells expressing a cDNA of the gene(34). In the present study, we simplified the situation by mak-ing a deletion that placed Met-34 as the first AUG in thesequence.

Neither mutation—deletion of the entire G gene or deletionof the cytoplasmic tail of G—affected viral replication in cellculture. This is consistent with previous studies in which dele-tion of the G protein of HMPV had no effect on replication invitro (6). Somewhat different results were obtained with avianmetapneumovirus, in which deletion of the G gene (in combi-nation with the SH gene) attenuated the virus in vitro (30), andHRSV, in which deletion of the G gene was attenuating inhuman HEp-2 cells but not in African green monkey Vero cells(41).

In contrast to the situation in vitro, rPVM-GFP-�G wasseverely attenuated for replication in mice, such that no infec-tious virus was detected in the lungs or nasal turbinates 6 daysafter intranasal infection with 500 PFU or 5,000 PFU. How-ever, following infection with the latter (but not the former)dose, mice were protected against challenge 28 days later witha lethal dose of biological wild-type PVM (data not shown).This suggests that a low level of replication occurred, since wehave previously observed that inoculation of rodents with 5.7log10 50% tissue culture infective doses of HMPV was notimmunogenic unless the virus was capable of replication (8).The high degree of attenuation due to the deletion of G iscomparable to that observed with HRSV �G in mice (41),whereas a comparable HMPV �G mutant appeared to be lessattenuated in rodent and nonhuman primate models (4). How-



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

ever, the latter models involved infection with significantly (upto 103-fold) higher doses, making a direct comparison difficult.

The Gt virus appeared to replicate as efficiently as its parent,rPVM-GFP7, both in the nasal turbinates and in the lungs,reaching titers in excess of 5.0 log10 and 6.0 log10 PFU/g in therespective tissues. Despite this efficient growth, infection withthe Gt virus was asymptomatic at doses that were highly patho-genic and often lethal with its parent, rPVM-GFP7. The Gtvirus appeared to be at least 100-fold attenuated compared toits parent, rPVM-GFP7, with respect to weight loss; attenua-tion based on the LD50 was not available, because none of theanimals infected with this virus died under the conditions ofthese experiments. Thus, the magnitude of virulence waslargely dissociated from the magnitude of replication. Thisdissociation was not complete, since increasing the dose of theGt virus to 50,000 PFU resulted in weight loss, and it seemslikely that a higher dose would result in deaths. Nonetheless,whereas the parent, rPVM-GFP7, induced weight loss at dosesof 500 and 5,000 PFU and was 50% and 100% lethal, respec-tively, at these doses, its Gt derivative was essentially non-pathogenic at the same doses. To our knowledge, this is thefirst example of an attenuating mutation in a pneumoviruswhose effect is not mediated primarily by reduced viral repli-cation. This is offered with the caveat that we did not deter-mine virus titers on every day between inoculation and day 10,when deaths began to occur with the rPVM-GFP7 virus. None-theless, the equivalency between rPVM-GFP-Gt and pPVM-GFP7 at the peak of replication on day 6 suggests that theformer virus replicated with an efficiency similar to that of itsparent.

In the present study, we were not able to detect a secretedform of the PVM G protein in the supernatants of cells in-fected with PVM expressing the wild-type or truncated G pro-tein. This is consistent with the surface expression of truncatedG observed by Rhandhwa and colleagues, as mentioned above(34), although that study did not address whether a proportionof wild-type or truncated G protein might also be secreted.This is in contrast with HRSV, for which a substantial propor-tion of G protein expressed from the wild-type gene is se-creted. In the case of HRSV, secretion is dependent on trans-lational initiation at the second AUG in the ORF (Met-48),which is located within the transmembrane anchor, followed byproteolytic trimming that completely removes the remainingtransmembrane domain and creates a new N terminus at po-sition 66 (19). In the case of PVM, the G ORF contains asecond AUG at codon 34, but it is in a sequence context thatis not consistent with efficient translational initiation due to thepresence of a C in the �3 position, and it is not known whethersignificant translation occurs. In addition, as already noted, thesecond AUG in the PVM G ORF is located at the inner faceof the transmembrane domain rather than in the middle of thisdomain. In any event, no secreted form of PVM G was ob-served. Thus, while the immunomodulatory function of theHRSV G protein appears to depend in large part on its secre-tion (32), the effects of PVM G presumably are not mediatedby a comparable species.

These findings suggest that the cytoplasmic tail of G hascharacteristics of a true virulence factor, that is, a factor whosepathogenic effect can be dissociated from pathogen load. Themechanism of attenuation for the Gt virus remains unknown.

Comparisons between the biologically derived nonpathogenicversion of strain 15 and the virulent strain J3666 indicated thatthe pathogenesis observed with the latter was associated withenhanced pulmonary inflammation that was paralleled by in-creased induction of proinflammatory cytokines, includingmacrophage inflammatory protein 1, macrophage chemoat-tractant peptide-1, eotaxin, and others (13). Thus, the viru-lence of strain J3666 appears to be mediated, at least in part,by an overly robust inflammatory response. As already noted,strains 15 and J3666 contain numerous differences involving anumber of proteins and potential cis-acting elements, includingresidues that are conserved within the Pneumovirus genus (24,42). Therefore, while it seems likely that the truncation of theG protein will prove to be an important factor in the differ-ences between the pathogenic and nonpathogenic phenotypescharacterized in the published comparisons between strains 15and J3666 (42), this will need to be directly evaluated. If thecytoplasmic tail of G indeed is a major contributor to thedifference between an “inflammatory” and a “noninflamma-tory” phenotype, it will be interesting to determine the mech-anism for this effect. One possibility is that the cytoplasmic taildirectly affects intracellular signaling. As an inexact precedent,the cytoplasmic tail of pathogenic Hantaan virus recently wasfound to inhibit the induction of type I interferon involving theRIG-I helicase, whereas the cytoplasmic tail of a nonpatho-genic strain lacked this ability (1). In the case of PVM, onepossibility is that the cytoplasmic tail somehow stimulates ac-tivation of NF-�B, leading to the expression of proinflammatorycytokines, and that its deletion ablates this ability. However, fur-ther studies are necessary to investigate the mechanistic basis forthese effects.

Deletion of the cytoplasmic tail provides a novel attenuatingmutation for the pneumoviruses. If the attenuation phenotypeof the Gt virus can be reproduced in those pneumoviruses forwhich a vaccine is needed (HRSV, HMPV, avian metapneu-movirus), it would be ideal for developing a live vaccine. Aparticular problem with live vaccines is that attenuation usuallyis based on decreased virus replication, which in turn reducesimmunogenicity. A viral mutant that replicates efficiently with-out causing disease would be an ideal vaccine candidate, onethat combines safety with a high level of immunogenicity.

In conclusion, we have developed a reverse genetic systemfor PVM that largely reproduces the virulent phenotype of thebiological wild-type virus. The utility of the system was shownby construction of recombinant viruses expressing GFP asmarker protein, as well as by construction of gene deletion/truncation mutants that investigated the contribution of the Gprotein to virulence. The capability of engineering recombi-nant PVM provides a convenient model for characterizingfactors involved in pneumovirus virulence in a natural host.

ACKNOWLEDGMENTS

We thank Jurgen Brandel for his expertise in photography andRenate Mielke for technical assistance.

This work was supported in part by the NIAID Intramural ResearchProgram and by a grant from the Ministry of Science, Research and theArts of Baden-Wurttemberg, Germany (Az: 21-655.042-1-1/1), toC.D.K.

REFERENCES

1. Alff, P. J., I. N. Gavrilovskaya, E. Gorbunova, K. Endriss, Y. Chong, E.Geimonen, N. Sen, N. C. Reich, and E. R. Mackow. 2006. The pathogenic



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

NY-1 hantavirus G1 cytoplasmic tail inhibits RIG-I- and TBK-1-directedinterferon responses. J. Virol. 80:9676–9686.

2. Barr, J., P. Chambers, P. Harriott, C. R. Pringle, and A. J. Easton. 1994.Sequence of the phosphoprotein gene of pneumonia virus of mice: expres-sion of multiple proteins from two overlapping reading frames. J. Virol.68:5330–5334.

3. Bermingham, A., and P. L. Collins. 1999. The M2-2 protein of humanrespiratory syncytial virus is a regulatory factor involved in the balancebetween RNA replication and transcription. Proc. Natl. Acad. Sci. USA96:11259–11264.

4. Biacchesi, S., Q. N. Pham, M. H. Skiadopoulos, B. R. Murphy, P. L. Collins,and U. J. Buchholz. 2005. Infection of nonhuman primates with recombinanthuman metapneumovirus lacking the SH, G, or M2-2 protein categorizeseach as a nonessential accessory protein and identifies vaccine candidates.J. Virol. 79:12608–12613.

5. Biacchesi, S., M. H. Skiadopoulos, K. C. Tran, B. R. Murphy, P. L. Collins,and U. J. Buchholz. 2004. Recovery of human metapneumovirus fromcDNA: optimization of growth in vitro and expression of additional genes.Virology 321:247–259.

6. Biacchesi, S., M. H. Skiadopoulos, L. Yang, E. W. Lamirande, K. C. Tran,B. R. Murphy, P. L. Collins, and U. J. Buchholz. 2004. Recombinant humanmetapneumovirus lacking the small hydrophobic SH and/or attachment Gglycoprotein: deletion of G yields a promising vaccine candidate. J. Virol.78:12877–12887.

7. Bossert, B., and K. K. Conzelmann. 2002. Respiratory syncytial virus (RSV)nonstructural (NS) proteins as host range determinants: a chimeric bovineRSV with NS genes from human RSV is attenuated in interferon-competentbovine cells. J. Virol. 76:4287–4293.

8. Buchholz, U. J., S. Biacchesi, Q. N. Pham, K. C. Tran, L. Yang, C. L. Luongo,M. H. Skiadopoulos, B. R. Murphy, and P. L. Collins. 2005. Deletion of M2gene open reading frames 1 and 2 of human metapneumovirus: effects onRNA synthesis, attenuation, and immunogenicity. J. Virol. 79:6588–6597.

9. Buchholz, U. J., S. Finke, and K. K. Conzelmann. 1999. Generation ofbovine respiratory syncytial virus (BRSV) from cDNA: BRSV NS2 is notessential for virus replication in tissue culture, and the human RSV leaderregion acts as a functional BRSV genome promoter. J. Virol. 73:251–259.

10. Collins, P. L., M. G. Hill, E. Camargo, H. Grosfeld, R. M. Chanock, andB. R. Murphy. 1995. Production of infectious human respiratory syncytialvirus from cloned cDNA confirms an essential role for the transcriptionelongation factor from the 5� proximal open reading frame of the M2 mRNAin gene expression and provides a capability for vaccine development. Proc.Natl. Acad. Sci. USA 92:11563–11567.

11. Collins, P. L., M. G. Hill, J. Cristina, and H. Grosfeld. 1996. Transcriptionelongation factor of respiratory syncytial virus, a nonsegmented negative-strand RNA virus. Proc. Natl. Acad. Sci. USA 93:81–85.

12. Cook, P. M., R. P. Eglin, and A. J. Easton. 1998. Pathogenesis of pneumo-virus infections in mice: detection of pneumonia virus of mice and humanrespiratory syncytial virus mRNA in lungs of infected mice by in situ hybrid-ization. J. Gen. Virol. 79:2411–2417.

13. Domachowske, J. B., C. A. Bonville, A. J. Easton, and H. F. Rosenberg. 2002.Differential expression of proinflammatory cytokine genes in vivo in re-sponse to pathogenic and nonpathogenic pneumovirus infections. J. Infect.Dis. 186:8–14.

14. Durbin, A. P., S. L. Hall, J. W. Siew, S. S. Whitehead, P. L. Collins, and B. R.Murphy. 1997. Recovery of infectious human parainfluenza virus type 3 fromcDNA. Virology 235:323–332.

15. Easton, A. J., J. B. Domachowske, and H. F. Rosenberg. 2004. Animalpneumoviruses: molecular genetics and pathogenesis. Clin. Microbiol. Rev.17:390–412.

16. Eaton, M. D., and W. van Herick. 1944. Demonstration in cotton rats andrabbits of a latent virus related to pneumonia virus of mice. Proc. Soc. Exp.Biol. Med. 54:89–92.

17. Elroy-Stein, O., T. R. Fuerst, and B. Moss. 1989. Cap-independent transla-tion of mRNA conferred by encephalomyocarditis virus 5� sequence im-proves the performance of the vaccinia virus/bacteriophage T7 hybridexpression system. Proc. Natl. Acad. Sci. USA 86:6126–6130.

18. Fearns, R., and P. L. Collins. 1999. Role of the M2-1 transcription antiter-mination protein of respiratory syncytial virus in sequential transcription.J. Virol. 73:5852–5864.

19. Hendricks, D. A., K. McIntosh, and J. L. Patterson. 1988. Further charac-terization of the soluble form of the G glycoprotein of respiratory syncytialvirus. J. Virol. 62:2228–2233.

20. Horsfall, F. L., and E. C. Curnen. 1946. Studies on pneumonia virus of mice.II. Immunological evidence of latent infection with the virus in numerousmammalian species. J. Exp. Med. 83:43–64.

21. Horsfall, F. L., and R. G. Hahn. 1940. A latent virus in normal mice capableof producing pneumonia in its natural host. J. Exp. Med. 71:391–408.

22. Horsfall, F. L., and R. G. Hahn. 1939. A pneumonia virus of Swiss mice.Proc. Soc. Exp. Biol. Med. 40:684–686.

23. Krempl, C. D., and P. L. Collins. 2004. Reevaluation of the virulence ofprototypic strain 15 of pneumonia virus of mice. J. Virol. 78:13362–13365.

24. Krempl, C. D., E. W. Lamirande, and P. L. Collins. 2005. Complete se-quence of the RNA genome of pneumonia virus of mice (PVM). VirusGenes 30:237–249.

25. Ling, R., and C. R. Pringle. 1989. Polypeptides of pneumonia virus of mice.I. Immunological cross-reactions and post-translational modifications.J. Gen. Virol. 70:1427–1440.

26. Ling, R., and C. R. Pringle. 1989. Polypeptides of pneumonia virus of mice.II. Characterization of the glycoproteins. J. Gen. Virol. 70:1441–1452.

27. Lo, M. S., R. M. Brazas, and M. J. Holtzman. 2005. Respiratory syncytialvirus nonstructural proteins NS1 and NS2 mediate inhibition of Stat2 ex-pression and alpha/beta interferon responsiveness. J. Virol. 79:9315–9319.

28. Miyata, H., M. Kishikawa, H. Kondo, C. Kai, Y. Watanabe, K. Ohsawa, andH. Sato. 1995. New isolates of pneumonia virus of mice (PVM) from Japa-nese rat colonies and their characterization. Exp. Anim. 44:95–104.

29. Moore, M. L., and R. S. Peebles, Jr. 2006. Respiratory syncytial virus diseasemechanisms implicated by human, animal model, and in vitro data facilitatevaccine strategies and new therapeutics. Pharmacol. Ther. 112:405–424.

30. Naylor, C. J., P. A. Brown, N. Edworthy, R. Ling, R. C. Jones, C. E. Savage,and A. J. Easton. 2004. Development of a reverse-genetics system for Avianpneumovirus demonstrates that the small hydrophobic (SH) and attachment(G) genes are not essential for virus viability. J. Gen. Virol. 85:3219–3227.

31. Pearson, H. E., and M. D. Eaton. 1940. A virus pneumonia of Syrian ham-sters. Proc. Soc. Exp. Biol. Med. 45:677–679.

32. Polack, F. P., P. M. Irusta, S. J. Hoffman, M. P. Schiatti, G. A. Melendi,M. F. Delgado, F. R. Laham, B. Thumar, R. M. Hendry, J. A. Melero, R. A.Karron, P. L. Collins, and S. R. Kleeberger. 2005. The cysteine-rich regionof respiratory syncytial virus attachment protein inhibits innate immunityelicited by the virus and endotoxin. Proc. Natl. Acad. Sci. USA 102:8996–9001.

33. Pringle, C. R., and R. P. Eglin. 1986. Murine pneumonia virus: seroepide-miological evidence of widespread human infection. J. Gen. Virol. 67:975–982.

34. Randhawa, J. S., P. Chambers, C. R. Pringle, and A. J. Easton. 1995.Nucleotide sequences of the genes encoding the putative attachment glyco-protein (G) of mouse and tissue culture-passaged strains of pneumonia virusof mice. Virology 207:240–245.

35. Reed, L. J., and H. Muench. 1938. A simple method of estimating fiftypercent endpoints. Am. J. Hyg. 27:493–497.

36. Richter, C. B., J. E. Thigpen, C. S. Richter, and J. M. Mackenzie, Jr. 1988.Fatal pneumonia with terminal emaciation in nude mice caused by pneumo-nia virus of mice. Lab. Anim. Sci. 38:255–261.

37. Schlender, J., B. Bossert, U. Buchholz, and K. K. Conzelmann. 2000. Bovinerespiratory syncytial virus nonstructural proteins NS1 and NS2 cooperativelyantagonize alpha/beta interferon-induced antiviral response. J. Virol. 74:8234–8242.

38. Schmidt, U., J. Beyer, U. Polster, L. J. Gershwin, and U. J. Buchholz. 2002.Mucosal immunization with live recombinant bovine respiratory syncytialvirus (BRSV) and recombinant BRSV lacking the envelope glycoprotein Gprotects against challenge with wild-type BRSV. J. Virol. 76:12355–12359.

39. Spann, K. M., K. C. Tran, B. Chi, R. L. Rabin, and P. L. Collins. 2004.Suppression of the induction of alpha, beta, and lambda interferons by theNS1 and NS2 proteins of human respiratory syncytial virus in human epi-thelial cells and macrophages. J. Virol. 78:4363–4369. (Erratum, 78:6705.)

40. Techaarpornkul, S., N. Barretto, and M. E. Peeples. 2001. Functional anal-ysis of recombinant respiratory syncytial virus deletion mutants lacking thesmall hydrophobic and/or attachment glycoprotein gene. J. Virol. 75:6825–6834.

41. Teng, M. N., S. S. Whitehead, and P. L. Collins. 2001. Contribution of therespiratory syncytial virus G glycoprotein and its secreted and membrane-bound forms to virus replication in vitro and in vivo. Virology 289:283–296.

42. Thorpe, L. C., and A. J. Easton. 2005. Genome sequence of the non-patho-genic strain 15 of pneumonia virus of mice and comparison with the genomeof the pathogenic strain J3666. J. Gen. Virol. 86:159–169.

43. Valarcher, J. F., J. Furze, S. Wyld, R. Cook, K. K. Conzelmann, and G.Taylor. 2003. Role of alpha/beta interferons in the attenuation and immu-nogenicity of recombinant bovine respiratory syncytial viruses lacking NSproteins. J. Virol. 77:8426–8439.

44. Weir, E. C., D. G. Brownstein, A. L. Smith, and E. A. Johnson. 1988.Respiratory disease and wasting in athymic mice infected with pneumoniavirus of mice. Lab. Anim. Sci. 38:133–137.

45. Whitehead, S. S., A. Bukreyev, M. N. Teng, C. Y. Firestone, M. St Claire,W. R. Elkins, P. L. Collins, and B. R. Murphy. 1999. Recombinant respira-tory syncytial virus bearing a deletion of either the NS2 or SH gene isattenuated in chimpanzees. J. Virol. 73:3438–3442.

46. Wilson, R. L., S. M. Fuentes, P. Wang, E. C. Taddeo, A. Klatt, A. J. Henderson,and B. He. 2006. Function of small hydrophobic proteins of paramyxovirus.J. Virol. 80:1700–1709.

47. Zhang, L., M. E. Peeples, R. C. Boucher, P. L. Collins, and R. J. Pickles.2002. Respiratory syncytial virus infection of human airway epithelial cells ispolarized, specific to ciliated cells, and without obvious cytopathology. J. Vi-rol. 76:5654–5666.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

Identification of a Novel Virulence Factor in Recombinant ...

Documents

Transcript of Identification of a Novel Virulence Factor in Recombinant ...