G

J

Gm

JJa

b

c

d

e

a

ARRA

KHFAPG

1

pfal

mp

2f

1h

ARTICLE IN PRESS Model

CV-2843; No. of Pages 7

Journal of Clinical Virology xxx (2013) xxx– xxx

Contents lists available at ScienceDirect

Journal of Clinical Virology

j ourna l h om epage: www.elsev ier .com/ locate / j cv

enetic diversity and molecular evolution of the major humanetapneumovirus surface glycoproteins over a decade

esse Papenburga,b, Julie Carbonneaua, Sandra Isabela, Michel G. Bergerona,ohn V. Williamsc,d, Gaston De Serrese, Marie-Ève Hamelina, Guy Boivina,∗

Centre de recherche en infectiologie de l’Université Laval, Centre de recherche du CHUQ, Quebec City, Quebec, CanadaMontreal Children’s Hospital, McGill University Health Centre, Montreal, Quebec, CanadaDepartment of Pediatrics, Vanderbilt University School of Medicine, Nashville, TN, United StatesDepartment of Pathology, Microbiology and Immunology, Vanderbilt University School of Medicine, Nashville, TN, United StatesInstitut national de santé publique du Québec, Quebec, Canada

r t i c l e i n f o

rticle history:eceived 26 June 2013eceived in revised form 19 August 2013ccepted 26 August 2013

eywords:uman metapneumovirususion (F) proteinttachment (G) proteinhylogenyenetic selection

a b s t r a c t

Background: Human metapneumovirus (HMPV) is a recently discovered paramyxovirus that is a majorcause of respiratory infections worldwide.Objectives: We aim to describe the molecular evolution of the HMPV F (fusion) and G (attachment) surfaceglycoproteins because they are targets for vaccines, monoclonal antibodies and antivirals currently indevelopment.Study setting: Nasopharyngeal aspirates were collected in children <3 years old with acute respiratoryinfection in Quebec City during 2001–2010. HMPV-positive samples (n = 163) underwent HMPV-F and -Ggene sequencing. Furthermore, HMPV-F (n = 124) and -G (n = 217) sequences were obtained from GenBankand other studies. Evolutionary analyses (phylogenetic reconstruction, sequence identity, detection ofrecombination and adaptive evolution) were computed.Results: Sequences clustered into 5 genetic lineages (A1, A2a, A2b, B1 and B2). Multiple lineages circulatedeach year in Quebec City. With the exception of B1, each of the 5 subgroups was the predominant lineageduring ≥1 season. The A1 lineage was not detected since 2002–2003 in our local cohort. There wasno evidence of inter- or intragenic recombination. HMPV-F was highly conserved, whereas HMPV-G

exhibited greater diversity. HMPV-F demonstrated strong evidence of purifying selection, both overalland in an abundance of negatively selected amino acid sites. In contrast, sites under diversifying selectionwere detected in all HMPV-G lineages (range, 4–15), all of which were located in the ectodomain.Conclusions: Predominant circulating HMPV lineages vary by year. HMPV-F is highly constrained andundergoes significant purifying selection. Given its high genetic variability, we found a modest number

s in H

of positively selected site

. Background

Human metapneumovirus (HMPV) is a recently discoveredaramyxovirus [1]. Evolutionary analyses suggest that it emerged

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

rom the closely related avian metapneumovirus subgroup Cpproximately 200 years ago and comprises at least four geneticineages in two major subgroups (A1, A2, B1, and B2) [2–5]. HMPV

Abbreviations: CHUQ, Centre Hospitalier Universitaire de Québec; HMPV, humanetapneumovirus; RTI, respiratory tract infection; RT-PCR, reverse-transcriptase

olymerase chain reaction.∗ Corresponding author at: Centre hospitalier de l’Université Laval, Room RC-709,

705 blvd Laurier, Quebec City, Quebec, Canada G1V 4G2. Tel.: +1 418 654 2705;ax: +1 418 654 2715.

E-mail address: Guy.Boivin@crchul.ulaval.ca (G. Boivin).

386-6532/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.jcv.2013.08.029

MPV-G.© 2013 Elsevier B.V. All rights reserved.

is now recognized as an important cause of upper and lower respi-ratory tract infection (RTI) across all ages in both healthy andimmunocompromised hosts worldwide, producing seasonal epi-demics in temperate climates [6,7]. Reinfections are common overan individual’s lifespan.

The fusion (F) and attachment (G) proteins are the major HMPVsurface glycoproteins. HMPV-F is a class 1 viral fusion proteinthat mediates virus entry through both attachment [8] and fusionwith the host cell membrane [9]. Furthermore, it is the primaryHMPV antigenic determinant, eliciting protective neutralizing anti-bodies [10,11]. HMPV-G is a predicted type II transmembraneprotein that may contribute to attachment via non-specific inter-

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

actions with cellular glycosaminoglycans [12]. Although it is notessential to viral replication in vitro, viruses in which the G genehas been deleted are highly attenuated in vivo [13]. HMPV-Gis only weakly immunogenic [10], yet exhibits several potential

ING Model

J

2 Clinica

gs

2

ifUogdFa

3

3

saRpfCscP(g

3

ttMRootPrcDCIeA

3

wcda

d(t8AF

ARTICLECV-2843; No. of Pages 7

J. Papenburg et al. / Journal of

lycosylation sites and significant genetic variability, suggestingelective immune pressure [14].

. Objectives

Because of their importance in HMPV replication and hostmmune response, the F and G proteins are potential targetsor vaccines [11], monoclonal antibodies [15] and antivirals [16].nderstanding their molecular evolution is essential to the designf such intervention strategies. Thus, we aimed to analyze theenetic diversity of HMPV strains identified in Quebec City, Canada,uring 2001–2010, and to compare these sequences with available

and G gene sequences worldwide for evidence of recombinationnd adaptive evolution.

. Study design

.1. Study subjects and clinical samples

Nasopharyngeal aspirates (NPA) for HMPV-F and -G geneequencing were obtained during three prospective cohort studiespproved by the Centre Hospitalier Universitaire de Québec (CHUQ)esearch Ethics Board. Subjects were children 0–35 months hos-italized at the CHUQ pediatric center or seen at a pediatric clinicor acute RTI over a total of seven winter seasons in Quebec City,anada [17–19]. A NPA was collected for clinical purposes at pre-entation and tested for HMPV by molecular diagnostic assays:onventional RT-PCR (2001–2003) [20], multiplex real-time RT-CR (2004–2005) [21], or multiplex PCR/DNA hybridization assay2006–2010) [22]. All HMPV-positive samples were included in ourene sequencing study.

.2. PCR amplification and gene sequencing

HMPV-F and -G gene sequencing was performed directly onhe nasopharyngeal secretions. RNA was extracted from 200 �L ofhawed specimen using the QIAmp Viral RNA Mini Kit (QIAGEN,

ississauga, Ontario) according to manufacturer’s instructions.everse transcription was performed on 10 �L of RNA using 1 �Lf either 50 ng/�L random primers (Amersham, Piscataway, NJ)r HMPV specific primers and the Superscript II RT Kit (Invi-rogen, Carlsbad, CA) as per manufacturer’s recommendations.rimer sequences and thermocycling conditions are available uponequest. HMPV-F and G gene-specific primers [3,23] were used foronventional PCR amplification performed with the Platinum TaqNA Polymerase (Life Technologies – Invitrogen, Burlington, ON,anada) or the Pfu Turbo DNA Polymerase (Agilent Technologies

nc, Burlington, ON, Canada). Nested PCR was performed when nec-ssary. Both strands of each amplicon were sequenced using anBI3730xl DNA Analyzer (Applied Biosystems, Foster City, CA).

.3. Sequence analyses

BioEdit 7.0.5 (http://www.mbio.ncsu.edu/bioedit/bioedit.html)as used to visually inspect chromatograms and prepare

ontigs. Newly-generated F and G gene sequences wereeposited in GenBank (accession nos. KF192711–KF192803nd KF178952–KF179045).

With the aim of compiling comprehensive nucleotide sequenceatasets for phylogenetic analysis, we performed BLAST searcheshttp://blast.ncbi.nlm.nih.gov/Blast.cgi) using the F and G genes of

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

he respective HMPV-A and HMPV-B reference strains, CAN97-3 (accession no. AY297749.1) and CAN98-75 (accession no.Y297748.1), as the query sequences. We then retrieved all HMPV-

gene sequences containing positions 154–1506 (CAN97-83 and

PRESSl Virology xxx (2013) xxx– xxx

CAN98-75 numbering), inclusively, and all complete HMPV-G genesequences originating from clinical specimens available in Gen-Bank by January 15, 2012. In addition, we searched PubMed for allpublished articles reporting HMPV-F or -G gene analyses. We con-tacted authors of articles for which sequences were not availablein GenBank to obtain them. For each gene, redundant, i.e., identi-cal, nucleotide sequences were then removed. Finally, one avianmetapneumovirus serotype-C sequence was added to the F genedataset as outgroup for rooting the phylogram (GenBank accessionno. AY579780.1). Because of the high genetic diversity of pneu-movirus G genes and consequent lack of homology between species[24], no outgroup was used for HMPV-G.

Multiple-sequence alignment for each gene was per-formed using the PRANK algorithm on the GUIDANCE server(http://guidance.tau.ac.il/) [25]. PRANK is a recently developedalignment method reported to be more accurate than MUSCLEor ClustalW in terms of gap placement [26]. The GUIDANCE toolassigns a confidence score (from 0 to 1) for each sequence positionaligned. We used the default cutoff score of 0.93 for the removal ofunreliably aligned positions.

Evolutionary history and phylogenetic reconstructions of eachgene were inferred using the Maximum Composite Likelihood dis-tance method in MEGA 5.0 [27]. One thousand bootstrap analyseswere performed to estimate the robustness of the phylogeneticinference.

Intragenic recombination was assessed in the F and G genesusing the RDP3 suite of programs [28]. Nucleotide substitutionmodel selection and analysis of selection pressures, i.e., adaptiveevolution, for each major lineage in each gene (A, B, A1, A2a, A2b,B1, B2) were performed using HyPhy software on the Datamonkeyweb interface http://www.datamonkey.org [29]. Prior to adaptiveevolution analyses, multiple sequence alignment was performedagain for HMPV-G, now separately for each lineage, with the aimof improving reliability scores and permitting analyses of lengthiersequences by aligning strains that were more closely geneticallyrelated. Overall and site-specific selection pressures were quan-tified as the ratio of non-synonymouns (dN) to synonymous (dS)substitutions. Sites under negative (i.e., purifying [dN/dS < 1]) orpositive (i.e., diversifying [dN/dS > 1]) selective pressure were iden-tified using the Fixed Effects Likelihood (FEL) [30] and Mixed EffectsModel of Evolution (MEME) [31] algorithms, respectively. P-Values<0.05 were considered significant.

4. Results

4.1. Newly-generated sequences

Our Quebec City studies identified 163 children <3 years oldwith acute HMPV RTI over 7 winter seasons from 2001 to 2010.All NPA specimens were available for gene sequencing. Over-all, we generated a sequence for HMPV-F and/or -G in 151/163(92.6%) specimens. We sequenced a 1352 bp region of the F gene(nucleotide positions 154–1506 in the prototype CAN97-83 andCAN98-75 genes) among 146/163 (89.6%) samples. Of these, 93were unique, i.e., non-identical, sequences. Regarding the G gene,we sequenced its entirety in 148/163 (90.7%) of the specimens,yielding 94 unique sequences. HMPV-G open reading frames variedin length from 654 to 723 bp among Quebec City strains.

4.2. Sequences from GenBank and other investigators

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

After removal of redundant sequences and sequences withambiguous sites, 100 unique F gene sequences (nt 154–1506)and 130 unique G gene sequences obtained from GenBank wereretained for analysis. In addition, other investigators provided us

ARTICLE IN PRESSG Model

JCV-2843; No. of Pages 7

J. Papenburg et al. / Journal of Clinical Virology xxx (2013) xxx– xxx 3

F netic

p

wsN[Gt

4

tsGarGwr

4

Hbadg

TP

HN

ig. 1. HMPV-F and -G segregate into two main subgroups (A and B) and 5 main geer the indicated horizontal distance.

ith unique sequences (24 F gene and 87 G gene) originating fromtudies of clinical respiratory samples from hospitalized children inew York and Tennessee [32], pediatric outpatients in Tennessee

33] and children in Freiburg, Germany [34,35]. Sequences fromenBank and other authors were from specimens dating from 1981

o 2007.

.3. Multiple sequence alignment

For HMPV-F, a total of 217 unique sequences underwent mul-iple sequence alignment. The GUIDANCE confidence score for allequences and all positions was maximal (1.0). Regarding HMPV-, a total of 311 unique sequences underwent multiple sequencelignment. As expected, the high genetic variability of HMPV-Gesulted in some uncertainty regarding its alignment. The overallUIDANCE alignment score was 0.939. After removal of residuesith a confidence score of <0.93, a total of 540 nucleotide positions

emained in the dataset.

.4. Phylogenetic analyses

Phylogenetic analysis demonstrated that 121/217 (55.8%)MPV-F sequences and 176/311 (56.6%) HMPV-G sequences

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

elonged to subgroup A. Among 125 specimens from Quebec Citylso tested by multiplex PCR/DNA hybridization assay capable ofistinguishing HMPV-A and HMPV-B [19,22], concordance of sub-rouping results was 100% when compared with sequence analysis.

able 1attern of circulation of HMPV lineages in Quebec City, Canada, during seven winter seas

Winter season HMPV-positive specimens by lineageN (% of HMPV strains that season)

A1 A2a A2b

2001–2002 6 (54.5) 0 (0.0) 1 (92002–2003 7 (53.4) 3 (23.0) 3 (22004–2005 0 (0.0) 0 (0.0) 1 (22006–2007 0 (0.0) 5 (38.5) 3 (22007–2008 0 (0.0) 0 (0.0) 5 (12008–2009 0 (0.0) 17 (53.1) 4 (12009–2010 0 (0.0) 2 (5.7) 26

Total 13 27 43

MPV, human metapneumovirus.ote: Based on analysis of F and G genes; strains always grouped to the same lineage by b

lineages (A1, A2a, A2b, B1 and B2). Scale bar represents substitutions per basepair

Phylograms of all HMPV-F and HMPV-G sequences (Fig. 1) showthat each major subgroup can be divided into 2 minor subgroups(A1, A2, B1 and B2). For both genes, bootstrap values of >70%suggested that the A2 minor subgroup is itself composed of twolineages, A2a and A2b.

4.5. Pattern of HMPV genotype circulation in Quebec City,2001–2010 (Table 1)

Multiple lineages circulated concurrently during each of theseven winter seasons (2001–2002 to 2009–2010) in our QuebecCity studies. With the exception of B1, each of the 5 lineages wasthe predominant lineage during at least 1 season. The A1 lineagewas not detected in our cohorts since 2002–2003. Furthermore,only two strains in our entire dataset have grouped to A1 since2004 (GER04-034 and CHN06-03).

4.6. Nucleotide and amino acid sequence identities (Table 2)

Overall, the mean nucleotide identities for HMPV-F and -G were89.1% and 57.9%, respectively. The respective mean amino acididentities were 97.4% and 48.6%. HMPV-F exhibited greater diver-sity at the nucleotide compared to amino acid level, whereas the

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

opposite was true for HMPV-G. Within major subgroups A andB, HMPV-G nucleotide and amino acid identity (>85% and >78%,respectively) were significantly higher compared to HMPV-G over-all.

ons.

Total

B1 B2

.1) 0 (0.0) 4 (36.4) 11 (100.0)3.0) 0 (0.0) 0 (0.0) 13 (100.0)5.0) 1 (25.0) 2 (50.0) 4 (100.0)3.0) 0 (0.0) 5 (38.5) 13 (100.0)1.6) 0 (0.0) 38 (88.4) 43 (100.0)2.5) 5 (15.6) 6 (18.8) 32 (100.0)(74.3) 1 (2.9) 6 (17.1) 35 (100.0)

7 61 151 (100.0)

oth genes. Bold face indicates the predominant lineage for that winter season.

ARTICLE ING Model

JCV-2843; No. of Pages 7

4 J. Papenburg et al. / Journal of Clinica

Table 2Estimated nucleotide and amino acid sequence identities, HMPV-F and HMPV-G.

Gene/protein Lineage Mean identity (%)

Nucleotide Amino acid

HMPV-F Overall 89.1 97.4Within A 96.3 99.3Within A1 98.8 99.7Within A2 97.4 99.4Within A2a 98.3 99.5Within A2b 98.4 99.7Within B 96.9 99.0Within B1 98.4 99.4Within B2 98.2 99.3

HMPV-G Overall 57.9 48.6Within A 85.5 78.3Within A1 95.9 94.1Within A2 91.6 87.6Within A2a 94.1 91.6Within A2b 93.8 90.3Within B 89.6 84.7Within B1 94.4 91.8Within B2 94.6 91.7

Mean nucleotide and amino acid identities were estimated by calculating pairwisedmH

4

tbwAi

dbmm0els

TM

a

istances in MEGA5 software using the maximum-composite likelihood and JTTatrix models, respectively.MPV, human metapneumovirus.

.7. Assessment of recombination and adaptive evolution

Visual inspection of the two phylograms (Fig. 2) for each of thewo major surface glycoproteins did not suggest intergenic recom-ination between HMPV strains. All strains for which sequencesere available for both genes segregated to the same lineage (A1,2a, A2b, B1 or B2) in both the HMPV-F and -G trees. No evidence of

ntragenic recombination was observed using the RDP3 program.Regarding adaptive evolution (Table 3), both the F and G genes

emonstrated overall evidence of purifying selection (P < 0.0001)y the codon-based Z-test of selection using the Nei-Gojoboriethod. For each major lineage (A, B, A1, A2a, A2b, B1, B2),ean dN/dS for HMPV-F (range, 0.049–0.066) and HMPV-G (range,

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

.496–0.664) is presented in Table 3. The F gene showed strongvidence of multiple sites under negative selective pressure in allineages (range, 12–110) and only one site undergoing positiveelection in subgroup A and another in subgroup B. The G gene

able 3ean dN/dS values, positively and negatively selected sites of HMPV-F and HMPV-G.

Gene Lineage Codons analyzed Mean dN/dS Positively select

F HMPV-A 52–502a 0.060 1 (a.a. 271)

HMPV-A1 52–502b 0.055 0

HMPV-A2a 52–502a 0.049 0

HMPV-A2b 52–502c 0.064 1 (a.a 271)

HMPV-B 52–502d 0.065 1 (a.a 223)

HMPV-B1 52–502e 0.053 0

HMPV-B2 52–502d 0.066 1 (a.a 223)

G HMPV-A 1–218a 0.577 15 (a.a. 33, 65, 9HMPV-A1 1–236b 0.588 6 (a.a. 106, 143HMPV-A2a 1–218a 0.496 7 (a.a 33, 112, 1HMPV-A2b 1–218c 0.612 4 (a.a. 93, 109,

HMPV-B 1–224e 0.625 14 (a.a. 55, 56, 5HMPV-B1 1–231e 0.597 4 (a.a 56, 174, 1HMPV-B2 1–226d 0.664 7 (a.a. 55, 58, 8

.a., amino acid.a CAN97-83 reference strain numbering.b NL00-01 reference strain numbering.c JPS03-240 reference strain numbering.d CAN98-75 reference strain numbering.e CAN97-82 reference strain numbering.

PRESSl Virology xxx (2013) xxx– xxx

showed fewer sites under purifying selection (range, 15–36). More-over, sites under diversifying selection were detected in all lineages(range, 4–15).

5. Discussion

We explored the evolution of HMPV by sequencing the F andG genes of >150 strains isolated in Quebec City during 2001–2010and by analyzing them within large datasets of sequences reportedglobally. We identified 5 genetic lineages (A1, A2a, A2b, B1 and B2),confirming that A2 comprises two distinct lineages (A2a and A2b),as first proposed by Huck et al. [34]. In Quebec City, we observedtemporal co-circulation of multiple lineages and annual variabil-ity of the predominant lineage. Since 2003, we did not identifythe A1 lineage locally, and in only two specimens globally. Otherinvestigators studying partial HMPV-N, -F or -G gene sequences(consequently, not in our dataset) also did not detect A1 strains inAustria [36], Argentina [37], New York State [38] or Cambodia [39]after 2006. These data suggest that A1 may circulate less extensivelythan previously.

The phylogeny proposed herein, including the distinction of theA2a and A2b lineages, is not simply a function of the relatively novelalignment method used (PRANK). HMPV-F phylogenetic recons-tructions based on alignments produced with other tools (MUSCLEand ClustalW, data not shown) were identical to those producedwith our PRANK alignment. Also, our strains always segregated tothe same lineage in both the HMPV-F and -G trees, further validat-ing the PRANK-based phylogeny.

Recombination is an important mechanism of evolutionarychange and genetic diversity in RNA viruses [40]. However, its fre-quency varies widely between viruses and is smallest among thosewith single-stranded negative-sense genomes [41], like HMPV. Itis therefore not surprising that we did not find evidence of eitherinter- or intragenic recombination.

HMPV-F demonstrated strong evidence of purifying selection,both overall and in an abundance of negatively selected sites.Similarly to previous reports [4], we estimated the mean dN/dSto be 10-fold smaller in F compared to G. Also, we identified a

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

paucity of positively selected sites in HMPV-F. Our selection pres-sure analysis, in combination with the fact that HMPV-F’s aminoacid identity is greater than its nucleotide identity, suggests thatthis highly conserved fusion protein is under strong structural and

ed sites Negativelyselected sites

110103731711236

3, 121, 127, 133, 143, 154, 156, 158, 164, 165, 170, 182, 190) 36, 154, 155, 158, 170) 1645, 146, 151, 165, 170) 15

127, 165) 198, 85, 88, 139, 162, 174, 180, 191, 192, 193, 220, 221) 2892, 220) 16

5, 88, 93, 139, 162) 36

ARTICLE IN PRESSG Model

JCV-2843; No. of Pages 7

J. Papenburg et al. / Journal of Clinical Virology xxx (2013) xxx– xxx 5

Fig. 2. Phylogeny of 217 near-full-length unique HMPV-F gene sequences (nt 154–1506) and of 311 unique full-length HMPV-G gene sequences. Panels A and B are detailedphylograms of HMPV subgroup A for the F and G genes, respectively. Panels C and D are detailed phylograms of HMPV subgroup B for the F and G genes, respectively.Phylogeny was inferred using the Neighbor-Joining method with evolutionary distances computed using the Maximum Composite Likelihood model in MEGA 5.0 software.The topologic accuracy of the tree was evaluated by using 1000 bootstrap replicates. Only bootstrap values >70% are shown. Scale bar represents substitutions per basepairper the indicated horizontal distance. For newly-generated sequences from Quebec City (indicated by a triangle) the clinical origin (hospitalized [H] [17,19], clinic [C] [19]or palivizumab study patient [SYN] [18]) is noted, followed by the year of sample collection, the specimen identifier, and the result of HMPV subgroup testing (HMPV-A [A],HMPV-B [B]) by multiplex PCR/DNA hybridization assay [22]. For other strains, GenBank accession number, geographic origin and year of isolation are noted when available.A y; JP,

U

ffataMtd

eted4ib

RG, Argentina; AUS, Australia; CAN, Canada; CHN, China; FL, Finland; GER, GermanK, United Kingdom; UY, Uruguay.

unctional constraints that limit its mutational ability. It there-ore represents an interesting target for vaccines [11], monoclonalntibodies (MAb) [15] and specific antivirals [16]. In particular,he anti-HMPV-F MAb338 has shown both prophylactic and ther-peutic benefits in a mouse model [15]. MAb-resistant mutants toAb338 generated in vitro harbored a I241R amino acid substi-

ution [42]. Of note, we found no I241R mutations in our globalataset.

Given the high genetic variability of HMPV-G, we found a mod-st amount of positively selected sites, ranging in number from 4o 15 across lineages. The nature of these sites varied between lin-ages, but all were in the putative ectodomain. In an earlier study by

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

e Graaf et al. [4], lineages A1, A2 and B2 respectively demonstrated, 2, and 1 sites under diversifying selection, several of which were

dentical to those we detected. We likely identified a greater num-er of sites because we used the recently developed mixed effects

Japan; NL, Netherlands; NY, New York; TN, Tennessee; SIN, Singapore; TW, Taiwan;

model of evolution (MEME), capable of capturing diversifying selec-tion that is both pervasive (affecting all branches) and episodic(affecting only a subset of branches) since it allows dN/dS distribu-tion to vary not only from site to site, but also from branch to branch[31]. Furthermore, we also used a more complete dataset and adifferent algorithm for multiple sequence alignment (PRANK), thelatter ensuring that unreliably aligned sites, often due to insertionsand deletions, were not analyzed [25,26]. The detection of adap-tive evolution is particularly sensitive to alignment accuracy; thus,removing unreliable regions improves the performance of positiveselection inference [43].

In addition to the season to season variation of the predominant

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

circulating genetic lineage in a community, it has been proposedthat HMPV-G genetic diversity may provide another mechanism forimmune escape [3]. The presence of positively selected sites in theG ectodomain, as our data suggest, would support this hypothesis.

ING Model

J

6 Clinica

HnHFitip

Cbiafrr

F

RMCtgIip

C

mfaeb

E

d

A

f

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

ARTICLECV-2843; No. of Pages 7

J. Papenburg et al. / Journal of

owever, in vivo, the G protein is only weakly immunogenic andot protective [10,44]. Indeed, the evaluation of a novel candidateMPV virus-like particle vaccine demonstrated that incorporating

alone in the vaccine was sufficient to induce neutralizing antibod-es in mice, whereas co-incorporation of F and G did not increasehe yield of protective antibodies [45]. Thus, the role of HMPV-G inmmune evasion remains unclear, although the presence of severalotential glycosylation sites may contribute to such a function [14].

Our evolutionary analyses of HMPV strains isolated in Quebecity during 2001–2010 and globally over a period of 25 years shoulde informative for the design of preventive and therapeutic modal-

ties against this virus, an important respiratory pathogen in allge groups. Given HMPV’s ability to re-infect persons over time,urther work evaluating the breadth and duration of host immuneesponses will be important to the development of vaccines thatemain effective against all strains over several winters.

unding

This work was supported by the Canadian Institutes of Healthesearch [grant number 151504 to G.B.] and a research grant fromedImmune Inc. [to G.B.]. G.B. is the holder of the Canada Research

hair on Emerging viruses and antiviral resistance. JP receivedraining bursaries from the Royal College of Physicians and Sur-eons Canada and the Association for Medical Microbiology andnfectious Diseases Canada for this work. The funders had no rolen study design, data collection and analysis, decision to publish, orreparation of the manuscript.

ompeting interests

G.B. has received research funds from GlaxoSmithKline, Hoff-ann La Roche, and Abbott. G.D.S. has received research funds

rom GlaxoSmithKline and Sanofi-Pasteur and has participated inn ad hoc GlaxoSmithKline advisory board meeting for which travelxpenses were reimbursed. J.V.W. serves on the scientific advisoryoard for Quidel. Other authors declare no conflicts of interest.

thical approval

This study was approved by the Centre Hospitalier Universitairee Québec (CHUQ) Research Ethics Board.

cknowledgements

We thank Valeria Falcone, Universitätsklinik Freiburg, Germany,or sharing unpublished sequences used in our analyses.

eferences

[1] van den Hoogen BG, de Jong JC, Groen J, Kuiken T, de Groot R, Fouchier RA, et al.A newly discovered human pneumovirus isolated from young children withrespiratory tract disease. Nat Med 2001;7:719–24.

[2] Peret TC, Abed Y, Anderson LJ, Erdman DD, Boivin G. Sequence polymor-phism of the predicted human metapneumovirus G glycoprotein. J Gen Virol2004;85:679–86.

[3] van den Hoogen BG, Herfst S, Sprong L, Cane PA, Forleo-Neto E, de Swart RL,et al. Antigenic and genetic variability of human metapneumoviruses. EmergInfect Dis 2004;10:658–66.

[4] de Graaf M, Osterhaus AD, Fouchier RA, Holmes EC. Evolutionary dynamics ofhuman and avian metapneumoviruses. J Gen Virol 2008;89:2933–42.

[5] Boivin G, Mackay I, Sloots TP, Madhi S, Freymuth F, Wolf D, et al. Globalgenetic diversity of human metapneumovirus fusion gene. Emerg Infect Dis2004;10:1154–7.

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

[6] Papenburg J, Boivin G. The distinguishing features of human metapneumovirusand respiratory syncytial virus. Rev Med Virol 2010;20:245–60.

[7] Edwards KM, Zhu Y, Griffin MR, Weinberg GA, Hall CB, Szilagyi PG, et al. Bur-den of human metapneumovirus infection in young children. N Engl J Med2013;368:633–43.

[

PRESSl Virology xxx (2013) xxx– xxx

[8] Cseke G, Maginnis MS, Cox RG, Tollefson SJ, Podsiad AB, Wright DW, et al. Integ-rin alphavbeta1 promotes infection by human metapneumovirus. Proc NatlAcad Sci USA 2009;106:1566–71.

[9] Cox RG, Williams JV. Breaking in: human metapneumovirus fusion and entry.Viruses 2013;5:192–210.

10] Skiadopoulos MH, Biacchesi S, Buchholz UJ, Amaro-Carambot E, Surman SR,Collins PL, et al. Individual contributions of the human metapneumovirus F G,and SH surface glycoproteins to the induction of neutralizing antibodies andprotective immunity. Virology 2006;345:492–501.

11] Herfst S, Fouchier RA. Vaccination approaches to combat human metapneu-movirus lower respiratory tract infections. J Clin Virol 2008;41:49–52.

12] Thammawat S, Sadlon TA, Hallsworth PG, Gordon DL. Role of cellularglycosaminoglycans and charged regions of viral G protein in human metap-neumovirus infection. J Virol 2008;82:11767–74.

13] Biacchesi S, Skiadopoulos MH, Yang L, Lamirande EW, Tran KC, Murphy BR,et al. Recombinant human Metapneumovirus lacking the small hydrophobicSH and/or attachment G glycoprotein: deletion of G yields a promising vaccinecandidate. J Virol 2004;78:12877–87.

14] Yang CF, Wang CK, Tollefson SJ, Lintao LD, Liem A, Chu M, et al. Human metap-neumovirus G protein is highly conserved within but not between geneticlineages. Arch Virol 2013;158:1245–52.

15] Hamelin ME, Gagnon C, Prince GA, Kiener P, Suzich J, Ulbrandt N, et al.Prophylactic and therapeutic benefits of a monoclonal antibody against thefusion protein of human metapneumovirus in a mouse model. Antiviral Res2010;88:31–7.

16] Deffrasnes C, Hamelin ME, Prince GA, Boivin G. Identification and evaluationof a highly effective fusion inhibitor for human metapneumovirus. AntimicrobAgents Chemother 2008;52:279–87.

17] Gilca R, De Serres G, Tremblay M, Vachon ML, Leblanc E, Bergeron MG, et al.Distribution and clinical impact of human respiratory syncytial virus genotypesin hospitalized children over 2 winter seasons. J Infect Dis 2006;193:54–8.

18] Boivin G, Caouette G, Frenette L, Carbonneau J, Ouakki M, De Serres G.Human respiratory syncytial virus and other viral infections in infants receivingpalivizumab. J Clin Virol 2008;42:52–7.

19] Papenburg J, Hamelin ME, Ouhoummane N, Carbonneau J, Ouakki M, RaymondF, et al. Comparison of risk factors for human metapneumovirus and respiratorysyncytial virus disease severity in young children. J Infect Dis 2012;206:178–89.

20] Boivin G, De Serres G, Cote S, Gilca R, Abed Y, Rochette L, et al. Human metapneu-movirus infections in hospitalized children. Emerg Infect Dis 2003;9:634–40.

21] Hamelin ME, Cote S, Laforge J, Lampron N, Bourbeau J, Weiss K, et al. Humanmetapneumovirus infection in adults with community-acquired pneumoniaand exacerbation of chronic obstructive pulmonary disease. Clin Infect Dis2005;41:498–502.

22] Raymond F, Carbonneau J, Boucher N, Robitaille L, Boisvert S, Wu WK, et al.Comparison of automated microarray detection with real-time PCR assays fordetection of respiratory viruses in specimens obtained from children. J ClinMicrobiol 2009;47:743–50.

23] Pizzorno A, Masner M, Medici C, Sarachaga MJ, Rubio I, Mirazo S, et al. Moleculardetection and genetic variability of human metapneumovirus in Uruguay. J MedVirol 2010;82:861–5.

24] van den Hoogen BG, Bestebroer TM, Osterhaus AD, Fouchier RA. Analysis of thegenomic sequence of a human metapneumovirus. Virology 2002;295:119–32.

25] Penn O, Privman E, Ashkenazy H, Landan G, Graur D, Pupko T, et al. a web serverfor assessing alignment confidence scores. Nucleic Acids Res 2010;38:W23–8.

26] Loytynoja A, Goldman N. Phylogeny-aware gap placement prevents errors insequence alignment and evolutionary analysis. Science 2008;320:1632–5.

27] Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5: molec-ular evolutionary genetics analysis using maximum likelihood, evolutionarydistance, and maximum parsimony methods. Mol Biol Evol 2011;28:2731–9.

28] Martin DP, Lemey P, Lott M, Moulton V, Posada D, Lefeuvre P. RDP3: a flex-ible and fast computer program for analyzing recombination. Bioinformatics2010;26:2462–3.

29] Delport W, Poon AF, Frost SD, Kosakovsky Pond SL. Datamonkey 2010: asuite of phylogenetic analysis tools for evolutionary biology. Bioinformatics2010;26:2455–7.

30] Kosakovsky Pond SL, Frost SD. Not so different after all: a comparisonof methods for detecting amino acid sites under selection. Mol Biol Evol2005;22:1208–22.

31] Murrell B, Wertheim JO, Moola S, Weighill T, Scheffler K, Kosakovsky Pond SL.Detecting individual sites subject to episodic diversifying selection. PLoS Genet2013;8:e1002764.

32] Williams JV, Edwards KM, Weinberg GA, Griffin MR, Hall CB, Zhu Y, et al.Population-based incidence of human metapneumovirus infection among hos-pitalized children. J Infect Dis 2010;201:1890–8.

33] Williams JV, Wang CK, Yang CF, Tollefson SJ, House FS, Heck JM, et al. The roleof human metapneumovirus in upper respiratory tract infections in children:a 20-year experience. J Infect Dis 2006;193:387–95.

34] Huck B, Scharf G, Neumann-Haefelin D, Puppe W, Weigl J, Falcone V. Novelhuman metapneumovirus sublineage. Emerg Infect Dis 2006;12:147–50.

35] Huck B, Egger M, Bertz H, Peyerl-Hoffman G, Kern WV, Neumann-Haefelin D,et al. Human metapneumovirus infection in a hematopoietic stem cell trans-

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

plant recipient with relapsed multiple myeloma and rapidly progressing lungcancer. J Clin Microbiol 2006;44:2300–3.

36] Aberle JH, Aberle SW, Redlberger-Fritz M, Sandhofer MJ, Popow-Kraupp T.Human metapneumovirus subgroup changes and seasonality during epi-demics. Pediatr Infect Dis J 2010;29:1016–8.

ING Model

J

Clinica

[

[

[

[

[

[

[

[nant human metapneumovirus G protein is immunogenic but not protective.Vaccine 2010;28:4145–52.

ARTICLECV-2843; No. of Pages 7

J. Papenburg et al. / Journal of

37] Velez Rueda AJ, Mistchenko AS, Viegas M. Phylogenetic and phylodynamic anal-yses of human metapneumovirus in Buenos Aires (Argentina) for a three-yearperiod (2009–2011). PLoS ONE 2013;8:e63070.

38] Lamson DM, Griesemer S, Fuschino M, St George K. Phylogenetic analysisof human metapneumovirus from New York State patients during Februarythrough April 2010. J Clin Virol 2012;53:256–8.

39] Arnott A, Vong S, Sek M, Naughtin M, Beaute J, Rith S, et al. Genetic variabil-ity of human metapneumovirus amongst an all ages population in Cambodiabetween 2007 and 2009. Infect Genet Evol 2013;15:43–52.

Please cite this article in press as: Papenburg J, et al. Genetic diversity andglycoproteins over a decade. J Clin Virol (2013), http://dx.doi.org/10.1016/

40] Holmes EC. The mechanisms of RNA virus evolution. In: The evolution andemergence of RNA viruses. New York: Oxford University Press; 2009. p. 37–86.

41] Chare ER, Gould EA, Holmes EC. Phylogenetic analysis reveals a low rateof homologous recombination in negative-sense RNA viruses. J Gen Virol2003;84:2691–703.

[

PRESSl Virology xxx (2013) xxx– xxx 7

42] Ulbrandt ND, Ji H, Patel NK, Barnes AS, Wilson S, Kiener PA, et al. Identificationof antibody neutralization epitopes on the fusion protein of human metapneu-movirus. J Gen Virol 2008;89:3113–8.

43] Privman E, Penn O, Pupko T. Improving the performance of positive selec-tion inference by filtering unreliable alignment regions. Mol Biol Evol 2012;29:1–5.

44] Ryder AB, Tollefson SJ, Podsiad AB, Johnson JE, Williams JV. Soluble recombi-

molecular evolution of the major human metapneumovirus surfacej.jcv.2013.08.029

45] Levy C, Aerts L, Hamelin ME, Granier C, Szecsi J, Lavillette D, et al. Virus-likeparticle vaccine induces cross-protection against human metapneumovirusinfections in mice. Vaccine 2013;31:2778–85.