Cadherin-related family member 3, a childhood asthmasusceptibility gene product, mediates rhinovirus Cbinding and replicationYury A. Bochkova,1, Kelly Wattersb, Shamaila Ashrafa,2, Theodor F. Griggsa, Mark K. Devriesa, Daniel J. Jacksona,Ann C. Palmenbergb,3, and James E. Gerna,c,3

aDepartment of Pediatrics and cDepartment of Medicine, University of Wisconsin–Madison, Madison, WI 53792; and bInstitute for Molecular Virology,University of Wisconsin–Madison, Madison, WI 53706

Edited by Robert A. Lamb, Northwestern University, Evanston, IL, and approved March 17, 2015 (received for review November 5, 2014)

Members of rhinovirus C (RV-C) species are more likely to causewheezing illnesses and asthma exacerbations compared withother rhinoviruses. The cellular receptor for these viruses washeretofore unknown. We report here that expression of humancadherin-related family member 3 (CDHR3) enables the cellsnormally unsusceptible to RV-C infection to support both virusbinding and replication. A coding single nucleotide polymorphism(rs6967330, C529Y) was previously linked to greater cell-surfaceexpression of CDHR3 protein, and an increased risk of wheezingillnesses and hospitalizations for childhood asthma. Comparedwith wild-type CDHR3, cells transfected with the CDHR3-Y529 var-iant had about 10-fold increases in RV-C binding and progenyyields. We developed a transduced HeLa cell line (HeLa-E8) stablyexpressing CDHR3-Y529 that supports RV-C propagation in vitro.Modeling of CDHR3 structure identified potential binding sitesthat could impact the virus surface in regions that are highly con-served among all RV-C types. Our findings identify that the asthmasusceptibility gene product CDHR3 mediates RV-C entry into hostcells, and suggest that rs6967330 mutation could be a risk factorfor RV-C wheezing illnesses.

rhinovirus C | CDHR3 | receptor

More than 160 types of rhinoviruses (RVs) are classified intothree species (A, B, and C) of enteroviruses in the family

picornaviridae. Viruses belonging to the RV-C species were onlydiscovered in 2006 (1, 2), some 50 y after the initial identificationof other RVs (3), because these viruses are not detectable bystandard tissue-culture techniques (4–7). RV-Cs are of specialclinical interest because they can cause more severe illnessesrequiring hospitalization in infants and children compared withthe RV-A or RV-B, and are closely associated with acute exac-erbations of asthma (8–10).After multiple attempts to grow RV-C in established cell lines

were unsuccessful, we described the use of organ culture of hu-man sinus mucosa to propagate a few RV-C clinical isolates invitro (11). Subsequently, we and others documented RV-Cgrowth in cultures of epithelial cells isolated from airway tissuespecimens that had then become redifferentiated under air–liquidinterface (ALI) conditions (12, 13). Notably, only fully differ-entiated ALI cultures supported RV-C replication, whereasundifferentiated airway epithelial cell monolayers did not. RV-Ccan only be produced using reverse genetics, as the full-lengthviral RNA transcripts synthesized in vitro are infectious whentransfected into human cell lines (for example, HeLa, WisL, andHEK293T) normally not susceptible to viral infection (11).With these tools, it was possible to show that the RV-C re-

ceptor is distinct from the intercellular adhesion molecule 1(ICAM-1) and low-density lipoprotein receptor (LDLR) familymembers used by the other species of RV (14, 15). However, theexisting RV-C culture systems have relatively low throughputand are labor intensive. Failure to identify the cellular receptorand the inability to propagate RV-C in convenient cell lines has

been a major obstacle to the study of virus-specific characteristicsthat could lead to effective antiviral strategies for this commonand important respiratory pathogen. We now report that humancadherin-related family member 3 (CDHR3), a member of thecadherin family of transmembrane proteins, mediates RV-Centry into host cells, and an asthma-related mutation in this geneis associated with enhanced viral binding and increased progenyyields in vitro.

ResultsIn Silico Identification of Candidate RV-C Receptors. To obtain aworking list of potential cellular receptors, we measured geneexpression on microarray chips (Human Gene 1.0 ST Array,Affymetrix) in two series of experiments involving cells that wereeither susceptible or not susceptible to RV-C infection (Fig. 1and Fig. S1). In one experimental series, susceptible cells in-cluded whole sinus mucosal tissue specimens (n = 5), epithelialcell suspension from sinus tissue, and nasal epithelium obtainedvia brushing, and unsusceptible cells, including monolayers ofprimary undifferentiated epithelial cells and transformed celllines (n = 5, including HeLa). In a second experimental series,we compared three pairs of undifferentiated (monolayers) andfully differentiated (ALI) sinus epithelial cell cultures.

Significance

The rhinovirus C (RV-C) species was first identified in 2006 andis a major cause of acute respiratory illnesses in children andhospitalizations for exacerbations of asthma. In this study, wediscovered that expression of human cadherin-related familymember 3 (CDHR3), a transmembrane protein with yet un-known biological function, enables RV-C binding and replica-tion in normally unsusceptible host cells. Intriguingly, wefound that a coding SNP (rs6967330, C529Y) in CDHR3, pre-viously linked to wheezing illnesses and hospitalizations forchildhood asthma by genetic analysis, also mediates enhancedRV-C binding and increased progeny yields in vitro. Finally,using structural modeling, we identified potential binding sitesin CDHR3 domains 1 and 2 interacting with viral capsid surfaceregions that are highly conserved among RV-C types.

Author contributions: Y.A.B., A.C.P., and J.E.G. designed research; Y.A.B., K.W., S.A.,T.F.G., M.K.D., and A.C.P. performed research; Y.A.B., K.W., D.J.J., A.C.P., and J.E.G. ana-lyzed data; and Y.A.B., A.C.P., and J.E.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE61396).1To whom correspondence should be addressed. Email: yabochkov@wisc.edu.2Present address: ATCC Microbiology Systems, Manassas, VA 20110.3A.C.P. and J.E.G. contributed equally to this work.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1421178112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1421178112 PNAS | April 28, 2015 | vol. 112 | no. 17 | 5485–5490

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 27

, 202

1

By in silico analysis of expression profiles, we identified a totalof 415 and 347 genes, preferentially expressed (>eightfold dif-ference, adjusted P < 0.05) in virus-susceptible cells in the first

and second experiments, respectively. We then performed ad-ditional filtering steps to narrow the candidate gene lists on thebasis of available Gene Ontology information (membrane lo-calization, receptor activity) and expression levels of the knownrhinovirus receptor genes (Fig. 1A and Table S1). We identifieda total of 12 common genes (represented by 14 probe sets)encoding proteins localized to plasma membrane or with pre-dicted or functionally demonstrated receptor activity, includingmembers of the human MHC class II, stomatin, guanine nucle-otide-binding, type I cytokine, and atypical chemokine receptorand cadherin protein families (Fig. 1B).

CDHR3 Expression in HeLa Cells Enables RV-C Binding and Replication.We next selected a subset of candidate genes [interleukin 5 re-ceptor, alpha (IL5RA); chemokine (C-C motif) receptor-like 1(CCRL1); cadherin-related family member 3 (CDHR3); low den-sity lipoprotein receptor class A domain containing 1 (LDLRAD1);calponin homology domain containing 2 (CHDC2); membrane-spanning 4-domains, subfamily A, member 8 (MS4A8)] for func-tional validation (Fig. 1B and Fig. S2). We transfected HeLa cellswith plasmid DNAs encoding the identified genes under control ofthe CMV promoter. The cells were then exposed to a reportervirus (RV-C15-GFP) engineered to express GFP during replication(Fig. 2A) to facilitate the analysis. This reporter virus replicatedwell in susceptible ALI cultures of airway epithelial cells (Fig. S3).We repeatedly detected some GFP expression only in cells trans-fected with CDHR3-expressing cDNA (Fig. 2B).To confirm that this result was not unique to the GFP-expressing

virus, we then infected CDHR3-expressing HeLa cells with wild-type RV-C15. Relative to untransfected control cells, we againobserved enhanced virus binding and evidence of vigorous rep-lication (about 2-log increase in RV-C RNA compared with in-put) (Fig. 2C). Therefore, transfection-induced expression ofCDHR3 was sufficient to convert HeLa cells normally unsusceptibleto RV-C infection into targets that supported virus binding andsubsequent replication.

Mutation in CDHR3 (Cys529→Tyr) Increases RV-C Binding and ProgenyYields. CDHR3 is a related member of the cadherin family oftransmembrane proteins. Typically, cadherins are involved inhomologous cell adhesion processes that are important for epi-thelial polarity, cell–cell interactions, and tissue differentiation(16–18). Four alleles of CDHR3 gene are described, of whichone representing a single nucleotide polymorphism (G→A) thatconverts residue cysteine to tyrosine at position 529 (Cys529→Tyr, rs6967330), was recently linked with a much greater risk ofasthma hospitalizations and severe exacerbations in young chil-dren in a genome-wide association study (19). It has been shownthat this point mutation in cadherin-repeat domain 5 near acalcium binding site leads to a marked increase in cell surfaceexpression of the CDHR3 protein, presumably by altering theprotein conformation.We engineered this single amino acid change into CDHR3

cDNA, inserted the recombinant FLAG tag (DYKDDDDK)between the signal sequence and domain 1 to facilitate detectionof CDHR3 surface expression, and then expressed it in HeLacells. Fluorescent microscopy and Western blot analysis con-firmed enhanced surface expression of Y529 (Fig. 2D), but alsosimilar levels of overall cellular expression of C529 (wild-type)and Y529 (risk allele) variant (Fig. 2E). Notably, we found ahigher (≥7-fold) number of cells expressing GFP after infectionwith reporter C15 virus (Fig. 2F) and about 10-fold greater RV-C15 binding and subsequent progeny yield, relative to the wild-type CDHR3 (Fig. 2G). The FLAG tag did not affect virusbinding or replication (Fig. 2G). Parallel experiments with an-other human cell line (HEK293T) again showed that CDHR3expression conferred susceptibility to RV-C infection (Fig. S4).

+3.7 +13.4

STOML3DNAAF1KLHL6CCRL1CERKLENPP3IL5RAPROM1CDHR3HLA-DRAHLA-DRAHLA-DRARGS22GNA14

Wis

L

Cel

l sus

pN

asal

epi

Sin

us #

1S

inus

#2

Sin

us #

3S

inus

#4

Sin

us #

5

NC

I-H35

8

PB

E

PS

E

HeL

a

Sin

us #

1S

inus

#2

Sin

us #

3

Sin

us #

6S

inus

#7

Sin

us #

8

monolayers whole tissue ALI

RV-C susceptible (n = 10)RV-C unsusceptible (n = 8)

log2 expression

A

B

Plasma membrane n = 66 genes

Receptor n = 65 genes

Exp #2

Exp #1

Common n = 25

Exp #1

Exp #2

Common n = 23

Gene Ontology terms

Expression level: RV-C susceptible > 7 log2 > RV-C unsusceptible

n = 38 genes n = 26 genes

Exp #2

Exp #1

Common n = 11

Exp #2

Exp #1

Common n = 11

Common n = 8

3 3

12 genes 14 probe sets

Plasma membrane Receptor

Fig. 1. Identification of candidate RV-C receptors by gene expression analysis.(A) Venn diagrams showing the common receptor candidate genes identified intwo series of microarray experiments after the filtering steps. The differentiallyexpressed genes were filtered stepwise on the basis of the indicated criteria andthe diagrams were generated by ArrayStar software v4.0 (DNASTAR). (B) Heatmap showing clustering analysis of expression profiles of the 12 differentiallyexpressed genes in primary cells, cell lines, and whole-tissue samples. Candidategenes selected for validation of RV-C receptor activity are shown in red. Ex-pression intensity values were analyzed by hierarchical clustering of samplesand genes using the Euclidean distance metric with centroid linkage methodusing ArrayStar software v4.0 (DNASTAR). Color bar represents gene expressionintensity (log2 scale). See also Figs. S1 and S2 and Table S1.

5486 | www.pnas.org/cgi/doi/10.1073/pnas.1421178112 Bochkov et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 27

, 202

1

Development and Characterization of Transduced HeLa Cell LineStably Expressing CDHR3-Y529 Variant. To confirm RV-C multi-cycle replication and enable large-scale virus propagation, weestablished (by lentiviral transduction) a human epithelial cellline HeLa-E8 stably expressing CDHR3-Y529. CDHR3 is expressedin these cells as a GFP-fusion protein that is cotranslationallycleaved because of the presence of a 2A peptide derived fromporcine teschovirus-1 to facilitate clonal selection of transducedcells. Similarly to transfected cells, RV-C15 efficiently replicated(about 2-log increase in viral RNA) in cells stably expressing

CDHR3 (Fig. 3A). The phenomenon is not unique to the C15 virustype, because HeLa-E8 cells were also susceptible to C2 and C41infection (Fig. 3A). Quantitative RT-PCR (RT-qPCR), Westernblot analysis, and flow cytometry confirmed higher CDHR3mRNA (over 400-fold increase) and protein expression intransduced HeLa-E8 cells compared with parental HeLa(control) cells (Fig. 3 B and C).The RV-C15 growth kinetics in HeLa-E8 monolayers was

overall similar to that of representative clinical isolates of A andB species showing continuous increase in viral RNA over thetime of infection (Fig. 3D). High expression levels of the ICAM-1 receptor used by the A16 and B52 strains to enter host cellsmight be responsible for about 1-log higher levels of theirbinding (2 h postinfection, hpi) and progeny yields (16–42 hpi),compared with that of C15. Analysis of the binding properties offluorescently labeled RV-C15 (RV-C15-APC) by flow cytometryconsistently showed increased binding to HeLa-E8 (22.1% ofpositively stained cells) compared with control HeLa (3.72%)cells (Fig. 3E).

An Extended Model of CDHR3 Structure. The CDHR3 protein se-quence (885 aa) has a short signal peptide (19 aa) and six tan-demly repeated cadherin domains (∼100 aa each), followed bytransmembrane (63 aa) and C-terminal cytoplasmic (150 aa)domains (Fig. 4A). The structure of CDHR3 has not yet beenexperimentally determined, but an approximation for domains2–6, as modeled on the mouse N-cadherin ectodomain structure(PDB ID code 3Q2W), is described elsewhere (19). We extendedthis model to include CDHR3 domain 1 (Fig. S5), using I-Tasser(20) and Robetta (21) algorithms to template the domain 1 and2 sequences on multiple linked-domain cadherin analogs (e.g.,PDB ID codes 1FF5A, 1Q5CA, and 2QVIA).The consensus output was joined to the domains 2–6 model

coordinates using the align function of PyMol (Molecular GraphicsSystem, v1.6). Multiple iterations of this process, specifying arange of templates, did not alter the general configuration ofthe final, full-domain structure (average RMSD: 1.632). Theextended model (Fig. 4B) provides predictions of residuesdisplayed on various repeat unit faces, including the Cys529→Tyr mutation at the domain 5–6 junction that defines the“risk” allele.

Docking the CDHR3 to a Two-Protomer Model of the RV-C15 Capsid.On cell surfaces or between cells, cadherins self-dimerize throughreciprocal domain 1 pairings mediated by cysteine, tryptophan, orhydrophobic interactions (22). There can be multiple glycosyla-tion sites dispersed along the ectodomain. The dimer format ofCDHR3 is unknown, but there is a single tryptophan (Trp76)residue in domain 1. According to the prediction algorithms,there is also a single putative asparagine-linked glycosylation site(Asn186) in domain 2. Trp76 and Asn186 position as solvent ex-posed on opposite sides of their respective domains (Fig. 4B).Assuming interactions similar to other picornaviruses, the RV-Care expected to access distal rather than proximal locations rel-ative to the cellular membrane.To test the feasibility of virus–receptor interactions, we docked

the CDHR3 domains 1, 2, and 3 to a two-protomer model of RV-C15 capsid (23). Preliminary mutagenesis experiments suggestedthat Asn186, but not Trp86, was participating in virus binding;therefore, both queried algorithms [Gramm-X (24) and HAD-DOCK (25)] were given only two constraints: (i) CDHR3 Asn186should be at or near the interface and (ii) the solvent exposed(outer) surface of the capsid proteins should be involved. Thepreferred complexes docked the receptor across the adjacent pro-tomers, impacting the virus where capsid proteins VP1, -2, and -3abut in a fivefold to twofold orientation (Fig. 4 C and D). In thisorientation, Asn186 of domain 2 is buried in VP1 contacts and do-main 1 extends across the twofold axis with VP1 and VP2 contacts,

eGFP

-LISSA / GPS-x2

A

B

2h 48h

Viru

s tit

re (P

FUe)

106

107

105

104

103

C

2h 48hVi

rus

titre

(PFU

e)

106

107

105

104

103

D

RV-C15-GFP

E

F 108G

surfa

cecy

topl

asm

Control CDHR3-C529 CDHR3-Y529

C529 Y529ControlCDHR3

RV-C15-GFP

RV-C15-GFP

RV-C15

RV-C15

5’-UTR 1A 1B 1C 1D 2A 2B 2C 3A3B

3C 3D 3’-UTRA(n)

120100

80

60

50

kDa

Fig. 2. RV-C15 replication in HeLa cells transfected with CDHR3 cDNA.(A) Genome structure of the RV-C15 infectious clone expressing GFP. Twoviral 2A protease cleavage sites permitting the release of GFP after poly-protein translation are indicated. (B) GFP expression 48 h after C15-GFP in-fection of HeLa cells transfected with wild-type CDHR3 for 48 h. (Scale bar,100 μm.) Representative image of three independent experiments. (C) RV-C15 binding (2 hpi) and replication (48 hpi) in HeLa cells transfected withwild-type CDHR3 (o), or Lipofectamine 2000 control (Δ) for 48 h (n = 5, dataare means ± SD). (D) Fluorescent microscopy images of cells transfected withFLAG-tagged versions of CDHR3 wild-type (C529) or Cys529→Tyr variant (Y529).Permeabilized (cytoplasm) or nonpermeabilized (surface) HeLa cells werestained with the α-CDHR3 or α-FLAG antibodies, respectively, to visualizeintracellular or surface expression. (Scale bar, 100 μm.) (E) Western blotanalysis of CDHR3 expression in HeLa cells transfected with wild-type (C529)or Cys529→Tyr variant (Y529) for 24 h and probed with α-CDHR3 polyclonalantibodies and α-tubulin antibodies (loading control). (F) GFP expression48 h after C15-GFP infection of HeLa cells transfected with CDHR3 Cys529→Tyrvariant for 48 h. (Scale bar, 100 μm.) Representative image of three independentexperiments. (G) RV-C15 binding (2 hpi) and replication (48 hpi) in HeLa cellstransfected with CDHR3 Cys529→Tyr variant (□), FLAG-tagged CDHR3 Cys529→Tyr(o), or Lipofectamine 2000 control (Δ) for 48 h (n = 5, data are means ± SD). Seealso Figs. S3 and S4.

Bochkov et al. PNAS | April 28, 2015 | vol. 112 | no. 17 | 5487

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 27

, 202

1

whereas domain 3 is free of the virus and Trp86 is on a nondockedface (Fig. 4 D and E).The RV-C as a species conserves <75% capsid protein iden-

tity, but the RV-C15 structure model predicts a single sharedreceptor (23). The modeled CHDR3 footprint covers nearlyevery surface residue of RV-C15 that is conserved within thespecies (Fig. 4E). Moreover, although modeled as a monomerunit, this particular receptor orientation would easily accom-modate domain 1-mediated dimer contacts across the twofoldaxis of the virus, or even other multimers around the fivefold.

DiscussionThe cure for the common cold has been elusive, in part becauseof incomplete information on the molecular biology of RV-C,which is an important cause of both upper and lower respiratoryillnesses as well as acute exacerbations of chronic childhoodasthma. Toward this goal, we performed genome-wide gene-expression analysis of cells that were either susceptible or notsusceptible to RV-C infection, selected and functionally vali-dated a subset of candidate receptor genes by expression inHeLa cells followed by infection with GFP-expressing RV-C15,and established that human CDHR3 confers susceptibility toRV-C infection to normally unsusceptible cells. Furthermore,replication of multiple RV-C types was demonstrated in HeLa-E8 cells with stable expression of CDHR3-Y529. Finally, wemodeled the complete CDHR3 structure, and identified a pu-tative docking site for CDHR3 monomers and dimers to bindvirus in a region with multiple residues conserved among all RV-Ctypes. These findings have great potential utility for RV re-search. Cell lines such as HeLa-E8 will enable production oflarge quantities of virus that can be used to develop RV-Cvaccines and to determine the actual, rather than modeled, vi-rion crystal structure. In addition, the HeLa-E8 cells could beused to develop RV-C infectivity assays for antiviral drugscreening, for detection of neutralizing antibody response toRV-C in clinical studies, and to grow RV-C clinical isolatesfrom samples of respiratory fluids.

CDHR3 is a member of a cadherin family of transmembraneproteins with a yet unknown biological function that is highlyexpressed in human lung tissue (26), bronchial epithelium (19),and during mucociliary differentiation in human airway epithe-lial cells in vitro (27). Members of this family are responsible forcommunication between identical cells through calcium-dependentinteractions. Cadherins on the same cell surface can self-associateinto cis-dimers (lateral dimers) and cell adhesion is based on in-teractions between identical cadherins on neighboring cell surfacesto form transdimers (28). Our structure modeling of the CDHR3-virus complex identified a potential binding site for monomers anddimers in a region of the RV-C15 capsid that is highly conservedamong different RV-C types. The docking exercise further suggeststhat receptor domains 1 and 2 may be key to virus interactions, andthat glycosylation, particularly at Asn186, may be contributory. Suchmodels of proteins and their interactions will help guide studiesusing mutational analysis to accurately map virus binding domainsin CDHR3.Notably, our findings demonstrate that a coding SNP (rs6967330)

in CDHR3, which was previously associated with wheezing illnessesand hospitalizations in childhood asthma by genetic analysis (19),is also associated with increased RV-C binding and progenyyields in vitro. If the same relationship is true in vivo, rs6967330and the surface overexpression of CDHR3 could also be a spe-cific risk factor for more severe RV-C illnesses. Another point toconsider is that RV wheezing illnesses in infancy are a strong riskfactor for subsequent childhood asthma and decreased lungfunction (29–31), but whether the viral infections serve a causalrole is unknown (32). These findings suggest the possibility thatthe rs6967330 SNP could promote RV-C wheezing illnesses ininfancy, which could in turn adversely affect the developing lungto increase the risk of asthma.Expression of CDHR3 increased RV-C binding and enabled

replication in normally unsusceptible host cells. These findingssuggest that CDHR3 is a functional receptor for RV-C. Onelimitation of our results is that anti-CDHR3 antibodies toextracellular domains or recombinant CDHR3 (such as theCDHR3–Fc fusion protein) are not currently available to test for

A B

C

Viru

s tit

re (P

FUe)

CDHR3

Tubulin

HeLa-E8

HeLacontrol

Viru

s tit

re (P

FUe)

D

SS

C-A

0

1

10

100

1000

HeLa-E8

HeLacontrol

mRNA Protein

Rel

ativ

e ex

pres

sion

2h 8h 16h 24h 42h

RV-C15RV-A16RV-B52

106

107

105

104

109

108

2h 24h

RV-C15RV-C41RV-C2

106

107

105

104

0

50K

100K

150K

200K

250KHeLa-controlHeLa-E8

Control HeLa - mock Control HeLa - RV-C15-APC

HeLa-E8 - mock HeLa-E8 - RV-C15-APC

E

CDHR3-GFP

RV

-C15

-AP

C

0.02% 0.03%

0.24%99.7%

CDHR3-GFP

3.72% 0%

0.01%96.3%

RV

-C15

-AP

C

0.02% 0.05%

99.9%0%

0.01% 22.1%

77.8%0.08%

104

105

103

1020

104

105

103

1020

104 1051031020 104 1051031020

104 1051031020 104 1051031020

104

105

103

1020

104

105

103

1020

104 105103-103 0CDHR3

Fig. 3. RV-C binding and replication in HeLa-E8 cells stably expressing CDHR3-Y529. (A) Binding (2 hpi) and replication (24 hpi) of RV-C clinical isolates in HeLa-E8 cells. (B) CDHR3 mRNA and protein expression in transduced (HeLa-E8) compared with control HeLa cells assessed by RT-qPCR and Western blot analysis.For mRNA expression, each bar represents five different RNA preparations (n = 5) and data are presented as fold-change relative to control cells. For proteinanalysis, whole-cell lysates were probed with α-CDHR3 polyclonal antibodies. (C) Control and transduced (HeLa-E8) cells were fixed, permeabilized, stainedwith monoclonal anti-CDHR3 antibody (Abcam, ab56549), and analyzed by flow cytometry. Data for control and transduced cells were acquired separatelyand combined. Data are representative of three independent experiments. (D) Growth curve of RV-C15, RV-A16 and RV-B52 clinical isolates in HeLa-E8 cells(n = 3, data are means ± SD). (E) Binding characteristics of fluorescently labeled RV-C15-APC in control and transduced (HeLa-E8) cells analyzed by flowcytometry. The percentages shown in the top two quadrants represent the percent of gated events that are negative for GFP fluorescence but positive forAPC fluorescence (Left) or positive for both GFP and APC fluorescence (Right). Data are representative of three independent experiments.

5488 | www.pnas.org/cgi/doi/10.1073/pnas.1421178112 Bochkov et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 27

, 202

1

direct interaction between the virus and putative receptor, whichwill be necessary to further prove its receptor function. Anotherpoint to consider is that our studies show some binding of RV-Cto control HeLa cells. This low-level binding could be non-specific, or could indicate that CDHR3 is expressed at a very lowlevel in control HeLa cells. Alternatively, there could be acoreceptor for RV-C that interacts with CDHR3, or a cofactorthat facilitates binding and entry. For example, bacterial surfacepolysaccharides stabilize polioviruses and facilitate cell attach-ment by increasing binding to the viral receptor (33, 34). In anycase, experiments in our laboratory and elsewhere demonstratedthat the RV-C interaction with native HeLa cells or other celllines (e.g., A549) is not sufficient for efficient virus entry andreplication (5, 6, 35).In conclusion, no vaccines or effective antivirals are currently

available for RV (36, 37). Blocking interactions between majorreceptor group RVs and ICAM-1 inhibits infection, airway in-flammation, and illness (38, 39), but is ineffective against replicationof minor receptor group viruses or RV-C types. Anti-CDHR3 an-tibodies, soluble CDHR3, or short peptides specifically targetingvirus-binding sites could serve as inhibitors of RV-C infection andprovide a new therapeutic approach for RV-C–induced illnesses.Development of RV-C–specific antiviral drugs may be especiallyimportant for treatment of infants and children with the rs6967330asthma risk allele in CDHR3.

Materials and MethodsExpanded methods are presented in SI Materials and Methods.

Cell Cultures. Sinus epithelial tissue samples were obtained from residualsurgical specimens from individuals undergoing sinus surgery and cultured as

described previously (11). The protocol was approved by the University ofWisconsin–Madison Human Subjects Committee. Primary airway epithelialcells were cultured submerged (undifferentiated monolayers) or at the ALI(fully differentiated), as previously described (13, 40).

Viruses and Infection. Recombinant RV-C15, RV-C15-GFP, RV-C2, RV-C41, RV-A16, and RV-B52 were produced by transfecting viral RNA synthesized in vitrofrom linearized infectious cDNA clones intoWisL cells, as previously described(11, 40). Cells grown in 12-well plates (monolayers) or in Transwell polycarbonateinserts (0.4-μm pore size; Corning) (ALI cultures) were inoculated with RV at106 PFUe per well or insert (unless other dose indicated), incubated for 2–4 hat 34 °C, and washed three times with PBS to remove unattached input virus.

RNA Extraction and RT-qPCR. Total RNA was extracted from sinus tissue orcultured cells using the RNeasy Mini kit (Qiagen). Viral RNA concentrationswere determined by RT-qPCR using Power SYBR Green PCR mix (Life Tech-nologies), as previously described (11). Relative expression of CDHR3 mRNAwas determined by RT-qPCR using the CDHR3-qf and CDHR3-qr primers(Table S2).

Gene Expression Analysis. Total RNA samples were treated with RQ1 DNase,labeled, fragmented and hybridized to Human Gene 1.0 ST arrays (Affy-metrix). Raw data (.cel files) were uploaded into ArrayStar software v4.0(DNASTAR) for normalization and statistical analysis. Gene-expression datahave been deposited to Gene Expression Omnibus (GEO) database underaccession no. GSE61396.

Expression Plasmids and Transfection. Plasmids for expression of receptorcandidate genes CCRL1 (Missouri S&T cDNA Resource Center), MS4A8(OriGene), CHDC2, and CDHR3 (TransOmic) were purchased. IL5RA andLDLRAD1 ORFs were PCR-amplified from a cDNA sample obtained fromdifferentiated airway epithelial cells using the corresponding primers(Table S2). The mutation in domain 5 (C529Y) of CDHR3 was engineered bytwo-step PCR using the flanking (CDHR3-f3 and CDHR3-r3) and internal

A

B C D

E

Fig. 4. Structure modeling of RV-C15 binding to CDHR3 receptor. (A) Schematic map of the CDHR3 protein with indicated domains, predicted glycosylationsites, and engineered point mutation and FLAG tag. (B) Structural model of the complete CDHR3 protein. Domain 1 was added to the published model ofdomains 2–6 (19) after I-Tasser modeling relative to known cadherin structures. Key residues are highlighted. Gray spheres are interdomain calcium ions fromPDB ID code 3Q2W. (C) Surface model of domains 1–3 docked to two RV-C15 protomers (model), projected onto pentamer coordinates. VP1 (blue), VP2(green), and VP3 (red) are shown following standard colors. Model was then duplicated across twofold axis to show putative paired orientations. Trianglemarks the icosahedral subunit. Biological subunit would include clockwise VP3. (D) Same as C, rotated x = 90°, y = 90°. (E) Similar to C with highlights (white)showing all RV-C15 surface amino acid residues conserved with >90% identity among all known RV-C isolates. See also Fig. S5.

Bochkov et al. PNAS | April 28, 2015 | vol. 112 | no. 17 | 5489

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 27

, 202

1

(CDHR3-C529Y-f and CDHR3-C529Y-r) primers. The plasmid DNA was pre-pared by Plasmid Maxi kit (Qiagen) and transfected into monolayers of HeLaor HEK293T cells using Lipofectamine 2000 (Life Technologies) according tothe manufacturer’s instructions.

Fluorescent Microscopy. HeLa cells plated on glass coverslips were transfectedwith 1 μg of pCDHR3-FLAG DNA using Lipofectamine 2000 (Life Technolo-gies) and fixed 24 h posttransfection. For detection of cell surface expressionof CDHR3, nonpermeabilized fixed cells were washed (two times) with PBS,blocked, and reacted with rabbit monoclonal anti-FLAG primary antibody(Sigma, F2555). Cells were then washed (three times) and treated with AlexaFluor 594 anti-rabbit antibody (Life Technologies). Next, for detection oftotal cellular CDHR3 expression, cells were permeabilized, washed (threetimes), reblocked, and stained with rabbit polyclonal anti-CDHR3 (Sigma,HPA011218). After wash (three times) with PBS, cells were treated withAlexa Fluor 488 anti-rabbit antibodies (Life Technologies).

Generation of Stable HeLa Cell Line Expressing CDHR3. The mutation in domain5 (C529Y) of CDHR3 was engineered in lentiviral vector pLX304 containingwild-type CDHR3 sequence (TransOmic) by subcloning from pCDHR3-C529Y.We then added a 2A peptide sequence derived from porcine teschovirus-1(41) and the GFP sequence to the 3′-end of CDHR3 using synthetic genefragments (gBlocks, Integrated DNA Technologies) to encode the CDHR3–GFP fusion protein, which is cotranslationally cleaved to facilitate clonalselection of transduced cells by direct fluorescent microscopy. The resultingplasmid, pLX304-CDHR3-C529Y-NPGP-GFP, was cotransfected with the mix-ture of packaging plasmids (psPAX2 and pMD2.G) into the 293T cells using

Lipofectamine 2000 (Life Technologies) to produce lentivirus particles. HeLacells were transduced, selected with blasticidin (5 μg/mL) and cloned bylimiting dilution in 96-well plates. The HeLa-E8 clone, showing the highestRV-C replication levels (over 2-log), was selected for further experiments.

Flow Cytometry. Control or transduced cells grown in suspension werewashed, stained with Ghost 780 (Tonbo) exclusion dye, fixed, and per-meabilized. Cells were then blocked [10% (vol/vol) FBS, 0.05% Tween-20in PBS], washed, and reacted with anti-CDHR3 mAbs (Abcam, ab56549).After wash (three times) with PBS, cells were reacted with Alexa Fluor647-conjugated donkey anti-mouse secondary antibody (Life Technolo-gies), washed again (three times), and analyzed by flow cytometry.

Fluorescent Labeling of RV-C15 and Virus Binding Assay. The purified C15 viruswas labeled with NHS ester fluorescent probe DyLight 650 (Thermo Scientific)following the manufacturer’s recommendations. Unincorporated dye wasremoved by fluorescent dye removal columns (Thermo Scientific). Monolayersof HeLa cells were trypsinized, incubated with labeled virus (multiplicityof infection of 10) for 1.5 h at 25 °C, washed (three times) before fixationwith 2% (wt/vol) paraformaldehyde, and analyzed by flow cytometry.

ACKNOWLEDGMENTS. We thank Shakher Sijapati, Marchel Hill, MariaLeighton, and Rebecca Brockman-Schneider (University of Wisconsin–Madison)for their technical assistance, and Alar Aab (University of Tartu) for adviceregarding the virus-labeling procedure. This study was supported by the Na-tional Institutes of Health–National Institute of Allergy and Infectious Dis-eases Grant U19-AI0104317 and by the National Institutes of Health–NationalHeart, Lung, and Blood Institute Grant P01 HL070831.

1. Lamson D, et al. (2006) MassTag polymerase-chain-reaction detection of respiratorypathogens, including a new rhinovirus genotype, that caused influenza-like illness inNew York State during 2004-2005. J Infect Dis 194(10):1398–1402.

2. Arden KE, McErlean P, Nissen MD, Sloots TP, Mackay IM (2006) Frequent detection ofhuman rhinoviruses, paramyxoviruses, coronaviruses, and bocavirus during acute re-spiratory tract infections. J Med Virol 78(9):1232–1240.

3. Andrewes CH, Chaproniere DM, Gompels AE, Pereira HG, Roden AT (1953) Propa-gation of common-cold virus in tissue cultures. Lancet 265(6785):546–547.

4. Lee WM, et al. (2007) A diverse group of previously unrecognized human rhinovirusesare common causes of respiratory illnesses in infants. PLoS ONE 2(10):e966.

5. Lau SK, et al. (2007) Clinical features and complete genome characterization of adistinct human rhinovirus (HRV) genetic cluster, probably representing a previouslyundetected HRV species, HRV-C, associated with acute respiratory illness in children.J Clin Microbiol 45(11):3655–3664.

6. Arden KE, Mackay IM (2010) Newly identified human rhinoviruses: Molecular meth-ods heat up the cold viruses. Rev Med Virol 20(3):156–176.

7. Arden KE, et al. (2010) Molecular characterization and distinguishing features of anovel human rhinovirus (HRV) C, HRVC-QCE, detected in children with fever, coughand wheeze during 2003. J Clin Virol 47(3):219–223.

8. Bizzintino J, et al. (2011) Association between human rhinovirus C and severity ofacute asthma in children. Eur Respir J 37(5):1037–1042.

9. Miller EK, et al. (2009) Human rhinovirus C associated with wheezing in hospitalisedchildren in the Middle East. J Clin Virol 46(1):85–89.

10. Drysdale SB, et al. (2014) Respiratory outcome of prematurely born infants followinghuman rhinovirus A and C infections. Eur J Pediatr 173(7):913–919.

11. Bochkov YA, et al. (2011) Molecular modeling, organ culture and reverse genetics fora newly identified human rhinovirus C. Nat Med 17(5):627–632.

12. Hao W, et al. (2012) Infection and propagation of human rhinovirus C in humanairway epithelial cells. J Virol 86(24):13524–13532.

13. Ashraf S, Brockman-Schneider R, Bochkov YA, Pasic TR, Gern JE (2013) Biologicalcharacteristics and propagation of human rhinovirus-C in differentiated sinus epi-thelial cells. Virology 436(1):143–149.

14. Greve JM, et al. (1989) The major human rhinovirus receptor is ICAM-1. Cell 56(5):839–847.

15. Hofer F, et al. (1994) Members of the low density lipoprotein receptor family mediatecell entry of a minor-group common cold virus. Proc Natl Acad Sci USA 91(5):1839–1842.

16. Kowalczyk AP, Nanes BA (2012) Adherens junction turnover: Regulating adhesionthrough cadherin endocytosis, degradation, and recycling. Subcell Biochem 60:197–222.

17. Leckband D, Sivasankar S (2012) Cadherin recognition and adhesion. Curr Opin CellBiol 24(5):620–627.

18. Nelson WJ, Dickinson DJ, Weis WI (2013) Roles of cadherins and catenins in cell-celladhesion and epithelial cell polarity. Prog Mol Biol Transl Sci 116:3–23.

19. Bønnelykke K, et al. (2014) A genome-wide association study identifies CDHR3 as asusceptibility locus for early childhood asthma with severe exacerbations. Nat Genet46(1):51–55.

20. Roy A, Kucukural A, Zhang Y (2010) I-TASSER: A unified platform for automatedprotein structure and function prediction. Nat Protoc 5(4):725–738.

21. Kim DE, Chivian D, Baker D (2004) Protein structure prediction and analysis using theRobetta server. Nucleic Acids Res 32(Web Server issue):W526–W531.

22. Harrison OJ, et al. (2010) Two-step adhesive binding by classical cadherins. Nat StructMol Biol 17(3):348–357.

23. Basta HA, Sgro JY, Palmenberg AC (2014) Modeling of the human rhinovirus C capsidsuggests a novel topography with insights on receptor preference and immunoge-nicity. Virology 448:176–184.

24. Tovchigrechko A, Vakser IA (2006) GRAMM-X public web server for protein-proteindocking. Nucleic Acids Res 34(Web Server issue):W310–W314.

25. de Vries SJ, van Dijk M, Bonvin AM (2010) The HADDOCK web server for data-drivenbiomolecular docking. Nat Protoc 5(5):883–897.

26. Yanai I, et al. (2005) Genome-wide midrange transcription profiles reveal expressionlevel relationships in human tissue specification. Bioinformatics 21(5):650–659.

27. Ross AJ, Dailey LA, Brighton LE, Devlin RB (2007) Transcriptional profiling of muco-ciliary differentiation in human airway epithelial cells. Am J Respir Cell Mol Biol 37(2):169–185.

28. Boggon TJ, et al. (2002) C-cadherin ectodomain structure and implications for celladhesion mechanisms. Science 296(5571):1308–1313.

29. Kotaniemi-Syrjänen A, et al. (2003) Rhinovirus-induced wheezing in infancy—The firstsign of childhood asthma? J Allergy Clin Immunol 111(1):66–71.

30. Jackson DJ, et al. (2008) Wheezing rhinovirus illnesses in early life predict asthmadevelopment in high-risk children. Am J Respir Crit Care Med 178(7):667–672.

31. Guilbert TW, et al. (2011) Decreased lung function after preschool wheezing rhino-virus illnesses in children at risk to develop asthma. J Allergy Clin Immunol 128(3):532–538, e1–e10.

32. Jackson DJ, Lemanske RF, Jr (2010) The role of respiratory virus infections in childhoodasthma inception. Immunol Allergy Clin North Am 30(4):513–522, vi.

33. Robinson CM, Jesudhasan PR, Pfeiffer JK (2014) Bacterial lipopolysaccharide bindingenhances virion stability and promotes environmental fitness of an enteric virus. CellHost Microbe 15(1):36–46.

34. Kuss SK, et al. (2011) Intestinal microbiota promote enteric virus replication andsystemic pathogenesis. Science 334(6053):249–252.

35. McErlean P, et al. (2008) Distinguishing molecular features and clinical characteristicsof a putative new rhinovirus species, human rhinovirus C (HRV C). PLoS One 3(4):e1847.

36. Patick AK (2006) Rhinovirus chemotherapy. Antiviral Res 71(2-3):391–396.37. Privolizzi R, Solari R, Johnston SL, McLean GR (2014) The application of prophylactic

antibodies for rhinovirus infections. Antivir Chem Chemother 23(5):173–177.38. Turner RB, et al. (1999) Efficacy of tremacamra, a soluble intercellular adhesion

molecule 1, for experimental rhinovirus infection: a randomized clinical trial. JAMA281(19):1797–1804.

39. Traub S, et al. (2013) An anti-human ICAM-1 antibody inhibits rhinovirus-inducedexacerbations of lung inflammation. PLoS Pathog 9(8):e1003520.

40. Nakagome K, et al. (2014) Effects of rhinovirus species on viral replication and cyto-kine production. J Allergy Clin Immunol 134(2):332–341.

41. Kim JH, et al. (2011) High cleavage efficiency of a 2A peptide derived from porcineteschovirus-1 in human cell lines, zebrafish and mice. PLoS ONE 6(4):e18556.

5490 | www.pnas.org/cgi/doi/10.1073/pnas.1421178112 Bochkov et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 27

, 202

1