Integrins �1�1 and �2�1 are receptorsfor the rotavirus enterotoxinNeung-Seon Seo*†, Carl Q.-Y. Zeng†‡, Joseph M. Hyser‡, Budi Utama‡, Sue E. Crawford‡, Kate J. Kim*, Magnus Hook*§,and Mary K. Estes‡§

*Center for Extracellular Matrix Biology, Institute of Biosciences and Technology, Texas A&M University System Health Science Center, Houston, TX 77030;and ‡Department of Molecular Virology and Microbiology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2007.

Contributed by Mary K. Estes, April 24, 2008 (sent for review March 7, 2008)

Rotavirus NSP4 is a viral enterotoxin capable of causing diarrhea inneonatal mice. This process is initiated by the binding of extracel-lular NSP4 to target molecule(s) on the cell surface that triggers asignaling cascade leading to diarrhea. We now report that theintegrins �1�1 and �2�1 are receptors for NSP4. NSP4 specificallybinds to the �1 and �2 I domains with apparent Kd � 1–2.7 �M.Binding is mediated by the I domain metal ion-dependent adhesionsite motif, requires Mg2� or Mn2�, is abolished with EDTA, and anNSP4 point mutant, E120A, fails to bind �2 integrin I domain. NSP4has two distinct integrin interaction domains. NSP4 amino acids114–130 are essential for binding to the I domain, and NSP4peptide 114–135 blocks binding of the natural ligand, collagen I, tointegrin �2. NSP4 amino acids 131–140 are not associated with theinitial binding to the I domain, but elicit signaling that leads to thespreading of attached C2C12-�2 cells, mouse myoblast cells stablyexpressing the human �2 integrin. NSP4 colocalizes with integrin�2 on the basolateral surface of rotavirus-infected polarized in-testinal epithelial (Caco-2) cells as well as surrounding noninfectedcells. NSP4 mutants that fail to bind or signal through integrin �2are attenuated in diarrhea induction in neonatal mice. Theseresults indicate that NSP4 interaction with integrin �1 and �2 is animportant component of enterotoxin function and rotavirus patho-genesis, further distinguishing this viral virulence factor from othermicrobial enterotoxins.

I domain � diarrhea � signaling � NSP4

Rotaviruses (RVs) are nonenveloped, triple-layered icosahe-dral viruses containing a genome of 11 segments of dsRNA

that code for six structural (VP1, VP2, VP3, VP4, VP6, and VP7)and six nonstructural (NSP1–NSP6) proteins (1). RV is theleading etiologic agent of severe gastroenteritis in infants, youngchildren, and animals worldwide. There is no known treatmentto prevent RV transmission. Although new vaccines have re-cently shown promising results in clinical trials and are nowlicensed (2), vaccine efficacy in developing countries wheredisease prevention is needed most remains unknown. The recentrecognition that RV causes extraintestinal infection in animalsand children (3–6) highlights a continuing need to better un-derstand the mechanisms of RV replication and pathogenesis.Increasing knowledge of the structure of RVs and proteinfunction has supported new ideas about virus pathogenesis, virusassembly, and virus–host interactions (1, 7, 8). However, themolecular details of RV effects on intestinal epithelial cells orextraintestinal cells remain incompletely understood.

NSP4 is the first described viral enterotoxin (9–11). The i.p.delivery of purified recombinant NSP4 or synthetic NSP4 pep-tides from mammalian and avian RVs that lack sequencesimilarity causes age-dependent diarrhea in young mice (9,12–16). The cytoplasmic toxic domain of NSP4 (aa 112–175)is released from infected cells and hypothesized to interactwith receptors on neighboring cells (9, 10, 17), mobilizing Ca2�

by stimulating phospholipase C-mediated inositol 1,4,5-

trisphosphate production (17), and eliciting age-dependent di-arrhea in neonatal mice (9, 18).

Integrins are a family of heterodimeric cell surface receptorscomposed of � and � subunits involved in cell–cell, cell–extracellular matrix, and cell–pathogen interactions (19–25).Vertebrates have nine � subunits that contain a so-called I or Adomain, in which a set of � subunits (�1, �2, �10, and �11) actas collagen receptors (21). The I domain forms a subset of thelarger group of von Willebrand factor A (VWA) domains foundin a wide range of proteins (26), and many appear to be involvedin protein–protein interactions. The x-ray structures of recom-binant �1, �2, �M, and �L I domains have defined a divalentcation-binding motif [metal ion-dependent adhesion site(MIDAS)] that appears to mediate the divalent cation binding ofthe I domains and the I domain-containing integrins to theirligands (27–30).

The discovery that NSP4 functions as an enterotoxin andtriggers epithelial cell responses stimulated a search for targetmolecule(s) that bind extracellular NSP4. Surface plasmonresonance (SPR) analysis isolated soluble HT29 surface com-ponents that bound to (His)6-NSP4FL immobilized onto a Ni2�-nitrilotriacetic acid (Ni2�-NTA)-coated sensor chip. Bound pro-teins were identified by mass spectrometry. Further biochemical,mutagenesis, and functional analyses identified two distinctfunctional domains on NSP4 (lacking known integrin-bindingmotifs) as being responsible for the binding to the MIDAS motifon �1 and �2 I and for eliciting C2C12-�2 spreading, respec-tively, and correlated with inducing diarrhea.

ResultsRV NSP4 Binds to Integrin �1 and �2 I Domains. To identify cellsurface receptors for NSP4, soluble cell surface componentsfrom human intestinal epithelial HT29 cells able to bind (His)6-NSP4 immobilized on a Ni2�-NTA sensor chip were isolated andcharacterized. Mass spectrometry identified an I domain-containing protein as one potential binding protein (data notshown). To confirm the interaction between NSP4 and the I/Adomain, ELISA solid phase-binding assays and SPR were per-formed in the presence of 1 mM MgCl2 or 1 mM EDTA. ByELISA, recombinant RV SA11 full-length NSP4 (NSP4FL)

Author contributions: N.-S.S., C.Q.-Y.Z., J.M.H., S.E.C., M.H., and M.K.E. designed research;N.-S.S., C.Q.-Y.Z., B.U., and K.J.K. performed research; N.-S.S., C.Q.-Y.Z., J.M.H., and S.E.C.contributed new reagents/analytic tools; N.-S.S., C.Q.-Y.Z., J.M.H., B.U., S.E.C., K.J.K., M.H.,and M.K.E. analyzed data; and N.-S.S., C.Q.-Y.Z., J.M.H., S.E.C., M.H., and M.K.E. wrote thepaper.

The authors declare no conflict of interest.

†N.-S.S. and C.Q.-Y.Z contributed equally to this work.

§To whom correspondence should be addressed. E-mail: mhook@ibt.tamhsc.edu ormestes@bcm.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0803934105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0803934105 PNAS � July 1, 2008 � vol. 105 � no. 26 � 8811–8818

MIC

ROBI

OLO

GY

INA

UG

URA

LA

RTIC

LE

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

bound to �2 I in a dose-dependent and saturable manner in thepresence of 1 mM MgCl2 (Fig. 1A, filled circles), but not to BSAunder the same conditions (Fig. 1 A, filled triangles). The bindingof NSP4FL to �2 I was abolished in the presence of 1 mM EDTA(Fig. 1 A, open circles). NSP4FL also bound to �1 I in thepresence of 1 mM MgCl2 [supporting information (SI) Fig. S1 A].

Interactions also were analyzed by SPR using a Biacore 3000and flow cells of a Ni2�-NTA sensor chip immobilized with �2I. NSP4FL bound to �2 I in the presence of MgCl2 (Fig. 1B) orMnCl2 (data not shown), and the binding was abolished in thepresence of CaCl2 or EDTA (data not shown). The apparentdissociation constant (Kd) of NSP4FL for �2 I was 2.7 � 1.9 �M.NSP4FL also bound to �1 I in the presence of MgCl2 with anapparent Kd of 1.1 � 0.5 �M (Fig. S1B).

Pull-down assays were used to determine the physical inter-action between NSP4FL and �2 I. NSP4FL showed a directinteraction specific to �2 I in vitro, and NSP4FL exhibited adose-dependent coprecipitation of (His)6-�2 I (Fig. S2, lanes5–8) in the presence of MgCl2, but not with Ni2�-NTA agarosealone (Fig. S2, lanes d and 4).

To determine the specificity of NSP4 binding to the integrinI domain, binding of NSP4 to A domains of VWA, which do notcontain a MIDAS motif, and to �5�1, �v�3, and �IIb�3 integrinsthat do not contain an I domain, were tested by both ELISA andSPR analysis. NSP4 did not bind to the VWA or to the �5�1,�v�3, and �IIb�3 (data not shown). These data and the require-ment for Mg2� or Mn2� indicate that NSP4 binding to theintegrin I domain is specific and mediated by the MIDAS motif.

The Enterotoxin Region of NSP4 Interacts with the Integrin I Domain.To determine the domain of NSP4 that binds to the integrin Idomain, various purified recombinant NSP4 proteins and syn-thetic NSP4 peptides were analyzed by SPR for their binding to

immobilized �2 I in the presence of MgCl2. Initially, we exam-ined the binding capacity of a recombinant form of the C-terminal domain of SA11 NSP4 (NSP4112–175). NSP4112–175bound to the �2 I in a dose-dependent manner in the presenceof Mg2� (Fig. 2A), indicating that the binding activity is locatedwithin amino acids 112–175. The apparent Kd of NSP4112–175 forthe �2 I was 1.6 � 0.7 �M.

We next determined whether NSP4s from a pair of porcineRV virulent (OSU-v) and attenuated (OSU-a) viruses and adeletion construct, OSU-v NSP4�131–140, were able to bind the �2I by SPR. OSU-v NSP4 induces diarrhea in neonatal mice andmobilizes intracellular Ca2� when added exogenously to culturedcells (31). OSU-a NSP4 contains amino acid changes at aminoacids 136 and 138 in the enterotoxin domain that attenuatediarrhea induction and intracellular Ca2� mobilization (31).These amino acids are deleted in the OSU-v NSP4�131–140construct. OSU-v NSP4FL, OSU-a NSP4FL, and OSU-vNSP4�131–140 each bound to the �2 I (Fig. 2B) with an apparentKd of 1.4 � 0.4 �M, 1.6 � 0.6 �M, and 2.3 � 0.8 �M, respectively.NSP4 from a human RV strain (G1P8) also bound the �2 I witha similar affinity (data not shown). These results indicated thatNSP4 from several RV strains are able to bind the �2 I domain.

The domain on NSP4 responsible for �2 I binding was furthermapped by testing synthetic peptides. SA11 NSP4pep114–135, butnot amino acids 150–175 (data not shown), bound to �2 I in a

Fig. 1. Dose-dependent binding of NSP4 to immobilized �2 I as measured byELISA and SPR. (A) Increasing concentrations of SA11 NSP4FL were incubatedwith immobilized �2 I or BSA with either 1 mM Mg2� or 1 mM EDTA. NSP4FL

binds to �2 I with 1 mM Mg2� (filled circles), whereas it does not with 1 mMEDTA (open circles). NSP4FL does not react with BSA in either 1 mM Mg2� (filledtriangles) or 1 mM EDTA (open triangles). (B) Representative profiles of therelative SPR response of the binding of increasing concentrations of SA11NSP4FL to immobilized �2 I. y axis values are resonance units (RU) normalizedto the maximum resonance units of NSP4FL to �2 I. “Resp. Diff.” refers to“Response Difference.” This is a quantitative calculation in RU that is made bysubtraction of the baseline response from the absolute response of thebinding.

Fig. 2. Localization of the binding region on NSP4 to �2 I determined by SPR.Representative profiles of the relative SPR response of the binding of variousNSP4 proteins and peptides to immobilized �2 I. (A) Increasing concentrationsof SA11 NSP4112–175 were applied to immobilized �2 I. (B) Similar concentra-tions (4.2–4.6 �M) of SA11 NSPFL, OSU-v NSP4FL, OSU-a NSP4FL, OSU-v NSP4�131–

140, and SA11 NSP4E120A were applied to immobilized �2 I. (C) Increasingconcentrations of NSP4pep114–135 were applied to immobilized �2 I. y axisvalues are resonance units (RU).

8812 � www.pnas.org�cgi�doi�10.1073�pnas.0803934105 Seo et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

dose-dependent manner in the presence of Mg2� (Fig. 2C). Apoint mutant, SA11 NSP4 E120A, did not bind to �2 I (Fig. 2B),indicating that E120 plays a key role in this interaction. The farUV CD spectra of NSP4FL and NSP4 E120A were compared torule out the possibility that the lack of binding of NSP4 E120A tothe I domain was due to a change in protein structure. The CDspectrum of NSP4 E120A showed a maximum at 195 nm and adouble minimum at 208 nm and 222 nm (data not shown),consistent with the �-helical content previously reported (32,33). Thus, the point mutation of E120A did not affect itssecondary structure.

C2C12-�2 Cells Attach and Spread on NSP4-Coated Substrates. Thebiological significance of the NSP4–�2 I interaction was exam-ined by using mouse myoblast C2C12 and C2C12-�2 cells. Thesecells are a model for collagen-integrin I domain binding andsignaling (34) and were used to investigate cell attachment andsignaling on plates coated with NSP4, the known adhesiveextracellular matrix protein collagen (positive control), or BSA(negative control). The C2C12-�2 (Fig. 3A), but not the parentalC2C12 (Fig. S3A) cells, bound to NSP4FL- or collagen-coated

wells. C2C12-�2 adhesion increased in a substrate dose-dependent manner for both NSP4FL (Fig. 3A, filled circles) andcollagen (Fig. 3A, open circles). NSP4pep114–135, but notNSP4pep150–175, inhibited C2C12-�2 attachment to collagen, anatural ligand for integrin �2 (Fig. 3C). Finally, C2C12-�2attachment to NSP4 was blocked by both purified �2 I (Fig. 3A,filled squares) and anti-�2 I specific antibody (data not shown),confirming specificity of the interaction. Compared with preim-mune IgG, anti-�2 I IgG inhibited attachment of C2C12-�2 cellsto NSP4 up to 80% in a dose-dependent manner.

Attachment of ligands to integrins can lead to signaling.C2C12-�2 express a functional �2�1 integrin, and ligand-induced signaling is manifested by cell spreading (35). Wedetermined whether attachment of C2C12-�2 to NSP4-coatedwells induced cell spreading. C2C12 and C2C12-�2 were allowedto attach onto microtiter wells coated with collagen, NSP4FL,NSP4112–175, and BSA. NSP4FL and NSP4112–175 inducedC2C12-�2 spreading, as did the positive control collagen (Fig.3B). However, the C2C12-�2 on BSA or parental C2C12 on anysubstrate-coated well did not spread. These results indicate thatNSP4 interacts with �2 I domain to induce cell attachment andpromote cell spreading.

Two Distinct Functional Domains on NSP4 Are Responsible for theIntegrin Binding and Signaling. To understand the relationshipbetween cell adhesion and signaling after C2C12-�2 interactionwith NSP4, wild-type and NSP4 mutants were tested for theability to induce C2C12-�2 spreading (Fig. 4). Specificity of therequirement for the �2 I was shown by the selective attachmentand spreading of C2C12-�2 (Fig. 4), but not parental C2C12 cells

Fig. 3. Cell adhesion and spreading on NSP4FL is mediated by �2�1 integrin.(A) Twenty-four-well plates were coated with increasing concentrations ofcollagen (open circles), SA11 NSP4FL (filled circles), or BSA (open triangles).Then 1.5 � 104 C2C12-�2 cells per well were added, and bound C2C12-�2 cellswere fixed, stained with crystal violet, and quantitated by measuring OD590.Specificity of the binding was shown by using soluble �2 I to block the cellattachment (filled squares). (B) Twenty-four-well plates were coated withcollagen, SA11 NSP4FL, NSP4112–175, and BSA. Then 1.5 � 104 C2C12-�2 or C2C12cells per well were seeded and allowed to spread. C2C12-�2 cells were fixedand stained, and the morphology of spread cells was analyzed by invertedlight microscopy. The morphology of C2C12-�2 cells seeded onto BSA-coatedwells and C2C12 cells seeded onto all substrate-coated wells was directlyobserved by inverted light microscopy without washing, fixation, or staining.(C) A 96-well plate was coated with collagen. Then 5 � 103 C2C12-�2 cells perwell were first incubated with 2.1 nmol of NSP4pep114–135 or NSP4pep150–175 for60 min and then added into collagen-coated wells. The numbers of attachedcells were counted after fixation and staining.

Fig. 4. C2C12-�2 adhesion and spreading on different NSP4 proteins. Ninety-six-well plates were coated with collagen, BSA, or various NSP4 and NSP4mutants. C2C12-�2 cells (5 � 103 cells per well) were applied and allowed toattach and spread. Attached and spread cells were fixed, stained, and countedby using an inverted microscope. (A and B) Attached and spread C2C12-�2 cellson protein-coated wells, respectively.

Seo et al. PNAS � July 1, 2008 � vol. 105 � no. 26 � 8813

MIC

ROBI

OLO

GY

INA

UG

URA

LA

RTIC

LE

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

(Fig. S3 B and C) on collagen-, NSP4FL-, or (His)6-NSP4FL-coated wells (bars 1, 3, and 4). C2C12-�2 did not attach on thenegative control BSA-coated wells (Fig. 4, bar 2), and, as shownpreviously, C2C12-�2 cells did not attach and spread on the(His)6-NSP4 E120A (Fig. 4, bar 5). C2C12-�2 attached andspread on OSU-v NSP4FL-coated wells (Fig. 4, bar 6). Surpris-ingly, C2C12-�2 cells attached but did not spread on OSU-aNSP4FL- or OSU-v NSP4�131–140-coated wells (Fig. 4, bars 7 and8). These results suggested that two different domains in NSP4elicit binding and signaling with the �2 I domain.

To further characterize whether two different functionaldomains in NSP4 elicit C2C12-�2 cell binding and spreading, cellinhibition assays were performed by using rabbit anti-NSP4peptide-specific antibodies. The epitope for anti-NSP4pep114–135antibody maps to amino acid 114–125, whereas the epitope foranti-NSP4pep120–147 antibody maps to amino acids 130–140 (36).Purified rabbit anti-NSP4pep114–135 and anti-NSP4pep120–147 anti-bodies were tested for the ability to inhibit C2C12-�2 cellattachment and spreading on SA11 NSP4FL-coated wells. Thenumbers of attached and spreading cells were counted sepa-rately. Anti-NSP4pep114–135 antibody showed a dose-dependentinhibition of C2C12-�2 attachment on NSP4FL-coated wells (Fig.5A, bars 4 and 5). NSP4pep120–147 antibody had no effect on cellattachment (Fig. 5A, bars 6 and 7), but exhibited a dose-dependent inhibition of C2C12-�2 spreading (Fig. 5B, filledcircles). These results confirm that two separate domainswithin NSP4 are responsible for cell adhesion and eliciting cellspreading.

NSP4 Colocalizes with Integrin �2 on the Surface of RV-Infected Caco-2Cells. Integrins �1 and �2 are basolaterally expressed on differ-entiated intestinal epithelial cells (37). In RV-infected neonatal

mice, NSP4 is present at the basement membrane of bothRV-infected and adjacent uninfected intestinal epithelial cells(38), and a form of NSP4 with enterotoxin activity is secretedfrom RV-infected cells (10). Localization of integrin �2 andNSP4 on RV-infected polarized, human intestinal Caco-2 cellswas determined by confocal microscopy (Fig. 6). RV-infectedcells were identified by intracellular staining for NSP4 (Fig. 6E)and another RV nonstructural protein, NSP5 (data not shown).Integrin �2 localized predominantly to the basolateral mem-brane of control (Fig. 6A) and RV-infected polarized cells (Fig.6D). Intracellular integrin �2 staining was increased in RV-infected cells (Fig. 6D). NSP4, but not other RV proteins (VP6or NSP5) (data not shown) colocalized with integrin �2 at thebasolateral membranes of noninfected cells surrounding theRV-infected cells (Fig. 6F, yellow arrows in XY and Z sections).NSP4 was not detected in noninfected cells distal to the RV-infected cells (Fig. 6F, white arrows in XY and Z sections). Thesedata show that NSP4 released from RV-infected cells binds theintegrin �2 on adjacent uninfected cells.

Integrin Binding and Signaling Correlates with Diarrhea Induction.NSP4-induced diarrhea has been linked to a PLC-mediatedincrease in intracellular calcium concentration (17). PLC-dependent intracellular calcium mobilization is abolished by theaminosteroid U-73122, whereas the analog U-73343 has noinhibitory effect and is useful as a negative control. Neithercompound negatively affected cell attachment regardless of theligand used (Fig. 7A). U-73122 inhibited C2C12-�2 cell spread-ing induced by NSP4 and collagen by 85% and 30%, respectively,whereas the analog U-73343 did not affect cell spreading (Fig.7B). Integrin �2-dependent cell spreading and diarrhea induc-tion by NSP4 both required PLC activation. Analysis of inhib-itors of known integrin signaling pathways found that wortman-nin, an inhibitor of PI3K signaling pathway, inhibited spreadinginduced by NSP4 by 80%, compared with analog U-73343 (Fig.7C), but did not inhibit C2C12-�2 cell adhesion (data notshown).

A possible correlation between NSP4-stimulated C2C12-�2cell spreading and diarrhea induction was tested by examiningthe ability of NSP4 E120A to induce diarrhea in neonatal mice.(His)6-NSP4FL and NSP4FL induced diarrhea in neonatal mice,whereas OSU-a NSP4, the RV capsid glycoprotein VP7 and PBS(negative controls) did not, as reported previously (Table 1) (9,31). The NSP4 E120A point mutant did not interact with �2 I orbind to C2C12-�2 cells and failed to induce diarrhea in neonatalmice. Thus, binding of NSP4 to the integrin I domain correlateswith the ability of NSP4 to induce diarrhea in mice.

DiscussionRotavirus NSP4 is the first described viral enterotoxin and is amultifunctional protein involved in viral morphogenesis andpathogenesis. The majority of known NSP4 functional domainsare located within the cytoplasmic tail of this transmembraneglycoprotein. Studies to determine NSP4 cellular-binding part-ners have identified the extracellular matrix proteins laminin-�3and fibronectin (38) and the intracellular proteins calnexin (39),caveolin (40–42), and tubulin (43).

Other known enterotoxins initiate their effects by binding toa cellular receptor (44, 45). Previous evidence supported theexistence of a cellular receptor involved in NSP4-induced diar-rhea induction and calcium mobilization. For example, theD-amino acid form of the enterotoxic NSP4pep114–135 does notinduce diarrhea (9), and removal of cell surface molecules withtrypsin abrogates NSP4-induced calcium mobilization in HT29cells (17). In the present study, �1 I and �2 I were identified ascellular receptors for NSP4. These interactions, similar to col-lagen require the metal ion Mg2� or Mn2�, are mediated by theMIDAS motif on the I domain, and induce integrin signaling.

Fig. 5. Effect of anti-NSP4 peptide antibodies to C2C12-�2 attachment andspreading on NSP4. (A) Ninety-six-well plate was coated with NSP4FL alone,NSP4FL with preimmune IgG, NSP4FL with anti NSP4pep114–135 antibody, orNSP4FL with anti- NSP4pep120–147 antibody. Then 5 � 103 C2C12-�2 cells per wellwere applied and incubated. Attached cells were fixed, stained, and countedunder a light microscope. Collagen and BSA were used as controls. The y axisindicates the number of attached cells. � and �� represent protein amountsof 10 and 20 �g per well, respectively. (B) The percentage of spread cells versusattached cells on NSP4FL with preimmune IgG (filled triangles) and anti-NSP4pep120–147 antibody (filled circles).

8814 � www.pnas.org�cgi�doi�10.1073�pnas.0803934105 Seo et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

However, unlike collagen, NSP4 lacks the typical �1 or �2binding motifs of many known integrin ligands. NSP4 binding tothe I domain was specific as demonstrated by a lack of bindingto the A domains of VWA, which contains three I/A domainswithout a MIDAS motif (26, 46) and to integrins �5�1, �v�3, and�IIb�3, which lack I domains (47), and NSP4 competed andinhibited C2C12-�2 cell attachment onto collagen I.

Two distinct domains of NSP4 are responsible for integrin Idomain binding and signaling. NSP4 amino acids 114–130 areimportant for binding to the MIDAS motif on the integrin Idomain, and E120 is a key residue possibly coordinating withMg2� or Mn2� for this interaction. Based on the crystal structureof amino acids 95–137 of NSP4, E120 and Q123 coordinate adivalent cation (48). This core metal-binding site (E-X-X-Q) is100% conserved in NSP4 from group A and human groups B andC RV (12–16, 49). The precise function of the E120 residue inthe interaction of NSP4 with integrin I domain remains un-known.

Integrin signaling, measured by C2C12-�2 cell spreading, wasinduced by a second domain on NSP4, amino acids 131–140 assummarized in Table 2. Since this domain was not required forbinding to the I domain, signaling appears to occur throughinteraction of NSP4 aa 131–140 with other domains of �2�1, oranother cell surface molecule. NSP4 constructs that inducedintegrin-mediated signaling correlated with diarrhea inductionin mouse pups and mobilization of intracellular Ca2� (18, 31).Inhibition of Ca2� mobilization by U-73122 blocks NSP4-induced enterotoxin activity (17) and also abolished NSP4-mediated cell spreading, indicating that integrin �2 may be

involved in NSP4-induced diarrhea. Future studies in integrinKO mice rederived into a susceptible genetic background couldprovide direct evidence that interaction with the integrin �2 isthe first step of the enterotoxin action.

Integrins trigger multiple signaling pathways that are involvedin cell migration, proliferation, differentiation, and antiapopto-tic functions (35, 50, 51). RV infection of intestinal cells up-regulates expression of integrin �2�1 (52, 53) by a PI3K-dependent pathway (53). NSP4-induced cell spreading viabinding to integrin �2 was also abolished by the PI3K inhibitorwortmannin. This finding suggests that NSP4 activates at leasttwo signaling pathways (PLC- and PI3K-mediated) through itsinteraction with integrin �2. Future studies are needed toelucidate the significance of NSP4 signaling through I domain-containing integrins. NSP4 signaling through integrin �2�1 maybe a key mediator of RV infection and pathogenesis.

Integrins act as cell surface receptors for cell adhesion toextracellular matrix proteins and for several viral pathogens (21,25, 47). Several viruses, including echovirus I, human herpesvi-rus 8, and RV, interact with integrin �2�1 (54–56). Echovirus 1and some strains of RV bind to the I domain of integrin �2, butthe binding is distinct from the MIDAS motif (57, 58). NSP4 isthe first enterotoxin to bind to an integrin and the bindingmediated by the MIDAS motif, which also binds to the physio-logical ligand collagen. Compared with collagen binding tointegrin, the attachment of NSP4 to the MIDAS motif requiresa second binding to activate integrin �2�1. This result indicatesthat NSP4 has a unique mechanism to bind and activate integrin�2�1. Studies using the integrin �1�1 and �2�1-deficient mice

Fig. 6. Colocalization of NSP4 with integrin �2 in RV-infected polarized Caco2. (A–F) Polarized Caco2 cells grown on Transwell membranes were mock infected(A–C) or RV-infected (D–F), fixed, permeabilized, and stained as described in SI Materials and Methods, Method 4. Integrin �2 is detected in A and D withanti-integrin �2 and Alexa Fluor 488-conjugated antibodies (green). NSP4 is analyzed in B and E with anti-NSP4114–135 and Alexa Fluor 594-conjugated antibodies(red). (C and F) Merged images of A and B or D and E, respectively. (F) Colocalization of NSP4 (red) with integrin �2 (green) on the lateral membrane of noninfectedcells is indicated by a yellow arrow in the XY section, which corresponds to the yellow arrow in the Z section. Lack of NSP4 detection on the lateral membraneof distal noninfected cells is indicated by a white arrow in the XY section, which corresponds to the white arrow in the Z section. (a) Apical surface. (b) Basolateralsurface.

Seo et al. PNAS � July 1, 2008 � vol. 105 � no. 26 � 8815

MIC

ROBI

OLO

GY

INA

UG

URA

LA

RTIC

LE

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

show that these receptors can function in innate immunity,inflammation, and autoimmunity (59–62). The binding of NSP4to integrin �1�1 and �2�1 could affect both induced humoraland cell-mediated immunity.

In the intestine, integrins have a polarized distribution and arelocalized on the basolateral cell surface of enterocytes (19, 63,64), indicating that NSP4 interaction with these integrins wouldoccur at the basolateral surface of enterocytes. In RV-infectedmice, NSP4 is detected at the basolateral surface of noninfectedenterocytes lower on the villus than RV-infected cells (38). Thisobservation suggests that NSP4 is released from infected epi-thelial cells where it can bind to the extracellular matrix com-ponents laminin-�3 and fibronectin. The immunolocalizationstudies reported here confirm that NSP4 localizes to the baso-lateral membrane of both RV-infected as well as adjacentuninfected polarized Caco-2 cells with integrin �2. Futurestudies are needed to determine the form of NSP4 releasedbasolaterally from polarized cells because both full-length andtruncated forms of NSP4 have been found in the media ofRV-infected nonpolarized and polarized cells (10, 65). Together,

these data support the hypothesis that NSP4, released frominfected cells, interacts with integrins on neighboring uninfectedcells to cause diarrhea.

Materials and MethodsCells and Cell Culture. Mouse myoblast C2C12 and C2C12-�2 cells, stablyexpressing the human �2 integrin subunit, were kindly provided by DonaldGullberg (University of Bergen, Bergen, Norway) and maintained in DMEMwith 10% FBS in the absence or presence of 10 �g/ml puromycin, respectively.Methods for cell culture have been described previously (34). HT29 cells werecultured in DMEM supplemented with 10% FBS. Caco2 cells were grown inMEM-based Caco2 medium with 10% FBS.

Expression and Purification of Recombinant Proteins and Peptides. �1 I and �2I were expressed in Escherichia coli and purified as described previously (66,67). �1 I and �2 I were dialyzed against PBS or Hepes-buffered saline beforeuse. SA11 NSP4FL, OSU-v NSP4FL, OSU-a NSP4FL, OSU-v NSP4�131–140, and SA11NSP4112–175 were expressed and purified as published previously (10, 17, 31).Synthetic peptides NSP4pep114 –135, NSP4pep113–149, NSP4pep120 –147, andNSP4pep150–175 were prepared as described previously (32, 68). (His)6-NSP4FL

and (His)6-NSP4 E120A were generated by introducing the E120A gene into theNSP4 gene (SI Materials and Methods, Method 1).

ELISAs to Detect Interactions Between NSP4FL and Integrin I Domains. Microtiterwells were coated with 1 �g of �1 I and �2 I, or BSA in PBS containing 1 mMMgCl2 or 1 mM EDTA, and then blocked with PBS containing 1% BSA (wt/vol)and 0.05% Tween 20. Varying concentrations of the NSP4FL were added to thewells and incubated for 1 h at room temperature. Rabbit anti-NSP4pep120–147

antibody and goat anti-rabbit IgG AP-conjugate (Bio-Rad) were used to detectbound NSP4. Experiments were performed in triplicate. Data are presented asthe mean value � SE of A405 from a representative experiment.

SPR Analysis of Interactions Between NSP4 and �2 I. SPR analysis was performedat 25°C in a BIAcore 3000 using an NTA sensor chip at a flow speed of 5 �l/min.The binding of NSP4 to the integrin I domain was determined by passingvarious concentrations of NSP4FL over a Ni2�-NTA sensor chip, which had beenimmobilized with (His)6-�1 I or (His)6-�2 I (SI Materials and Methods,Method 2).

(His)6-�2 I-Bound NSP4 Complex Pull-Down Assays. (His)6-�2 I (3 �g) wasincubated with increasing amounts of NSP4FL (1.5, 3, 6, and 9 �g) in 100 �l ofPBS containing 1 mM MgCl2 for 1 h at 4°C. Ni-NTA agarose (Qiagen) was addedto the mixture and incubated for 30 min at 4°C. The Ni2�-NTA agarose waswashed in PBS, and bound proteins were eluted by using 20 �l of PBScontaining 100 mM imidazole. The eluted samples were mixed with SDS/PAGEsample buffer without 2-ME and run on SDS/12% PAGE. �2 I-bound NSP4FL wasdetected by Western blotting using rabbit anti-NSP4120–147 antibody and goatanti-rabbit IgG HRP-conjugate (Bio-Rad).

Fig. 7. PLC-specific inhibitor U-73122 and PI3K inhibitor wortmannin do notaffect C2C12-�2 cell adhesion, but disrupt C2C12-�2 cell spreading. Ninety-six-well microtiter plates were coated with NSP4FL or collagen. C2C12-�2 cells(5 � 103 cells per well) were mixed with various amounts of U-73122, its analogU-73343 as control, or wortmannin and then added to the wells and allowedto attach and spread for 120 min. Attached and spread cells were fixed,stained, and counted by using an inverted microscope. (A) Attached cellnumbers. (B) Percentages of spread C2C12-�2 among the attached cells in thevarious concentrations of U-73122 or U-073343, respectively. (C) Percentage ofspread C2C12-�2 in the presence of 3 �M U-73343 or 1 �M wortmannin.

Table 1. SA11(His)6-NSP4E120A does not induce diarrhea in CD1neonatal mice

Inoculum

Dose*No. of mice with diarrhea/

No. of inoculated micenmol �g

Experiment 1SA11 (His)6-NSP4FL WT 0.5 10 10/12

P†

SA11 (His)6-NSP4FL E120A 0.5 10 2/14OSU-a NSP4 0.5 10 2/13(His)6-VP7 0.5 10 2/13PBS (50 �l) 1/11

Experiment 2SA11 (His)6-NSP4FL WT 1.0 20 9/10

P†

SA11 (His)6-NSP4FL E120A 1.0 20 2/12PBS (50 �l) 1/12

*All proteins were dissolved in 50 �l of endotoxin-free PBS.†P � 0.001, � 2.

8816 � www.pnas.org�cgi�doi�10.1073�pnas.0803934105 Seo et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

Cell Adhesion and Spreading Assay. Microtiter wells were coated with variousconcentrations of collagen I (PuroCol), NSP4FL, NSP4 mutants, or BSA and thenblocked with 0.5% BSA in PBS. C2C12 and C2C12-�2 (0.5–1.5 � 104 cells perwell) in 50–100 �l was added and allowed to adhere for different time pointsat 37°C in a 5% CO2 incubator (SI Materials and Methods, Method 3).

Inhibition of Cell Adhesion and Spreading. Rabbit anti-NSP4 peptide antiserawere generated by using synthetic peptides SA11 NSP4pep114 –135 andNSP4pep120–147, respectively. Total IgG was purified by using caprylic acid (69).Microtiter wells were coated with 10 �g of NSP4FL alone, 10 �g of NSP4FL with10 or 20 �g of preimmune IgG, 10 �g of NSP4FL with 10 or 20 �g of anti-NSP4pep114–135 IgG, 10 �g of NSP4FL with 10 or 20 �g of anti-NSP4pep120–147 IgG,or preimmune IgG alone. The 10 �g of collagen I or BSA was coated as controls.Cell attachment and spreading were assessed as described above. To observethe effect of U-73122, analog U-73343, and wortmannin (Sigma–Aldrich) onC2C12-�2 adhesion and spreading, 0.5 � 104 cells per well were mixed withvarious amounts of U-compound or wortmannin and immediately added intowells coated with 1 �g per well of NSP4 and incubated for 2 h at 37°C in a5%CO2 /95% air atmosphere. Cell fixation, staining, and counting were car-ried out as mentioned above.

Immunolocalization. Caco2 cells were maintained on a 0.4-�m Costar Transwellpolycarbonate membrane (Corning) until cells were polarized and then were

infected with RV SA11-C13 at a multiplicity of infection of 1. After 24 h, cellswere fixed, permeabilized, and stained sequentially with mouse anti-integrin�2 I domain and rabbit anti-NSP4pep114–135, followed by anti-mouse AlexaFluor 488- and anti-rabbit Alexa Fluor 594-labeled antibodies. Observationswere carried out on a Carl Zeiss LSM 510 META confocal microscope (SIMaterials and Methods, Method 4).

Diarrhea Induction in Neonatal Mice. Purified SA11 (His)6-NSP4FL, SA11 (His)6-NSP4E120A, OSU-a NSP4, (His)6-VP7, and PBS were inoculated i.p. into 5-day-oldCD1 mice (Charles River Laboratories). Each sample contained 0.5 or 1.0 nmolof protein in 50 �l of endotoxin-free PBS (Gibco). To determine the presenceof diarrhea, each pup was examined every 1.5 h for 9 h after inoculation bygently pressing the abdomen.

ACKNOWLEDGMENTS. We thank Drs. Venkataram Prasad, Andrew Morris,Frank Ramig, and Ulrich Desselberger for helpful suggestions and criticalreading of the manuscript; Drs. Timothy G. Palzkill and Zanna Beharry for helpwith using the BIAcore 3000; Dr. Michael Mancini, Jeannie Zhong, and AntalisKalika for their technical assistance with confocal microscopy; Daniele L.Higgines, Xi-Lei Zeng, and Xiaowen Liang for technical assistance; and Dr.Christi Gendron for recombinant VWA domain. This work was supported, inpart, by National Institutes of Health Grants AI20624 (to M.H.), DK30144 (toM.K.E.), Training Grant in Molecular Virology T32AI007471, and DK56338 (toM.K.E.), which funds the Texas Medical Center Digestive Diseases Center.

1. Jayaram H, Estes MK, Prasad BV (2004) Emerging themes in rotavirus cell entry, genomeorganization, transcription and replication. Virus Res 101:67–81.

2. Glass RI, Parashar UD (2006) The promise of new rotavirus vaccines. N Engl J Med354:75–77.

3. Blutt SE, et al. (2003) Rotavirus antigenaemia and viraemia: A common event? Lancet362:1445–1449.

4. Blutt SE, et al. (2007) Rotavirus antigenemia in children is associated with viremia. PLoSMed 4:e121.

5. Crawford SE, et al. (2006) Rotavirus viremia and extraintestinal viral infection in theneonatal rat model. J Virol 80:4820–4832.

6. Ray P, et al. (2006) Quantitative evaluation of rotaviral antigenemia in children withacute rotaviral diarrhea. J Infect Dis 194:588–593.

7. Taraporewala ZF, Patton JT (2004) Nonstructural proteins involved in genome pack-aging and replication of rotaviruses and other members of the Reoviridae. Virus Res101:57–66.

8. Ciarlet M, Estes MK (2001) Interactions between rotavirus and gastrointestinal cells.Curr Opin Microbiol 4:435–441.

9. Ball JM, Tian P, Zeng CQ, Morris AP, Estes MK (1996) Age-dependent diarrhea inducedby a rotaviral nonstructural glycoprotein. Science 272:101–104.

10. Zhang M, Zeng CQ, Morris AP, Estes MK (2000) A functional NSP4 enterotoxin peptidesecreted from rotavirus-infected cells. J Virol 74:11663–11670.

11. Jagannath MR, et al. (2006) N- and C-terminal cooperation in rotavirus enterotoxin:Novel mechanism of modulation of the properties of a multifunctional protein by astructurally and functionally overlapping conformational domain. J Virol 80:412–425.

12. Mori Y, Borgan MA, Ito N, Sugiyama M, Minamoto N (2002) Diarrhea-inducing activityof avian rotavirus NSP4 glycoproteins, which differ greatly from mammalian rotavirusNSP4 glycoproteins in deduced amino acid sequence in suckling mice. J Virol 76:5829–5834.

13. Guzman E, McCrae MA (2005) Molecular characterization of the rotavirus NSP4 en-terotoxin homologue from group B rotavirus. Virus Res 110:151–160.

14. Ishino M, et al. (2006) Comparison of NSP4 protein between group A and B humanrotaviruses: Detection of novel diarrhea-causing sequences in group B NSP4. Arch Virol151:173–182.

15. Horie Y, Nakagomi T, Oseto M, Masamune O, Nakagomi O (1997) Conserved structuralfeatures of nonstructural glycoprotein NSP4 between group A and group C rotaviruses.Arch Virol 142:1865–1872.

16. Sasaki S, Horie Y, Nakagomi T, Oseto M, Nakagomi O (2001) Group C rotavirus NSP4induces diarrhea in neonatal mice. Arch Virol 146:801–806.

Table 2. Summary of NSP4 structure and function

NSP4 ConstructI domainbinding

C2C12-�2 cells

DiarrheaAttached Spread

SA11 � � � �

OSU-v � � � �

OSU-a � � � �

OSU-v�131–140 � � � �

SA11 His � � � �

SA11 HisE120A � � � �

SA11112–175 � � � �

SA11pep114–135 � NA NA �

SA11pep150–175 � NA NA �*

Names and linear schematic of the NSP4 constructs tested are listed to the left. Two vertical black bars indicateN-linked glycosylation sites, and N-terminal filled black squares indicate a histidine tag. Synthetic peptides aredescribed in the SI Materials and Methods. The � and � indicate a positive or negative event, respectively.NA, not applied.*Unpublished data.

Seo et al. PNAS � July 1, 2008 � vol. 105 � no. 26 � 8817

MIC

ROBI

OLO

GY

INA

UG

URA

LA

RTIC

LE

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

17. Dong Y, Zeng CQ, Ball JM, Estes MK, Morris AP (1997) The rotavirus enterotoxin NSP4mobilizes intracellular calcium in human intestinal cells by stimulating phospholipaseC-mediated inositol1,4,5-trisphosphateproduction.ProcNatlAcadSciUSA94:3960–3965.

18. Morris AP, et al. (1999) NSP4 elicits age-dependent diarrhea and Ca2�-mediated I�

influx into intestinal crypts of CF mice. Am J Physiol 277:G431–G444.19. Hynes RO (1992) Integrins: Versatility, modulation, and signaling in cell adhesion. Cell

69:11–25.20. Shimaoka M, Takagi J, Springer TA (2002) Conformational regulation of integrin

structure and function. Annu Rev Biophys Biomol Struct 31:485–516.21. Hynes RO (2002) Integrins: Bidirectional, allosteric signaling machines. Cell 110:673–

687.22. Isberg RR, Tran Van Nhieu G (1994) Binding and internalization of microorganisms by

integrin receptors. Trends Microbiol 2:10–14.23. Hauck CR, Agerer F, Muenzner P, Schmitter T (2006) Cellular adhesion molecules as

targets for bacterial infection. Eur J Cell Biol 85:235–242.24. Caswell CC, Lukomska E, Seo NS, Hook M, Lukomski S (2007) Scl1-dependent internal-

ization of group A Streptococcus via direct interactions with the �2�1 integrin enhancespathogen survival and re-emergence. Mol Microbiol 64:1319–1331.

25. Stewart PL, Nemerow GR (2007) Cell integrins: Commonly used receptors for diverseviral pathogens. Trends Microbiol 15:500–507.

26. Whittaker CA, Hynes RO (2002) Distribution and evolution of von Willebrand/integrinA domains: Widely dispersed domains with roles in cell adhesion and elsewhere. MolBiol Cell 13:3369–3387.

27. Nolte M, et al. (1999) Crystal structure of the alpha1beta1 integrin I-domain: Insightsinto integrin I-domain function. FEBS Lett 452:379–385.

28. Shimaoka M, et al. (2003) Structures of the alpha L I domain and its complex withICAM-1 reveal a shape-shifting pathway for integrin regulation. Cell 112:99–111.

29. Lu C, Oxvig C, Springer TA (1998) The structure of the �-propeller domain andC-terminal region of the integrin �M subunit. Dependence on � subunit associationand prediction of domains. J Biol Chem 273:15138–15147.

30. Emsley J, Knight CG, Farndale RW, Barnes MJ, Liddington RC (2000) Structural basis ofcollagen recognition by integrin alpha2beta1. Cell 101:47–56.

31. Zhang M, et al. (1998) Mutations in rotavirus nonstructural glycoprotein NSP4 areassociated with altered virus virulence. J Virol 72:3666–3672.

32. Huang H, et al. (2001) Membrane interactions of a novel viral enterotoxin: Rotavirusnonstructural glycoprotein NSP4. Biochemistry 40:4169–4180.

33. Taylor JA, O’Brien JA, Yeager M (1996) The cytoplasmic tail of NSP4, the endoplasmicreticulum-localized non-structural glycoprotein of rotavirus, contains distinct virusbinding and coiled coil domains. EMBO J 15:4469–4476.

34. Tiger CF, Fougerousse F, Grundstrom G, Velling T, Gullberg D (2001) Alpha11beta1integrin is a receptor for interstitial collagens involved in cell migration and collagenreorganization on mesenchymal nonmuscle cells. Dev Biol 237:116–129.

35. Giancotti FG, Ruoslahti E (1999) Integrin signaling. Science 285:1028–1032.36. Hyser JM, Zeng CQ, Beharry Z, Palzkill T, Estes MK (2007) Epitope mapping and use of

epitope-specific antisera to characterize the VP5* binding site in rotavirus SA11 NSP4.Virology 373:211–228.

37. Beaulieu JF (1992) Differential expression of the VLA family of integrins along thecrypt-villus axis in the human small intestine. J Cell Sci 102:427–436.

38. Boshuizen JA, et al. (2004) Rotavirus enterotoxin NSP4 binds to the extracellular matrixproteins laminin-beta3 and fibronectin. J Virol 78:10045–10053.

39. Mirazimi A, Nilsson M, Svensson L (1998) The molecular chaperone calnexin interactswith the NSP4 enterotoxin of rotavirus in vivo and in vitro. J Virol 72:8705–8709.

40. Parr RD, et al. (2006) The rotavirus enterotoxin NSP4 directly interacts with the caveolarstructural protein caveolin-1. J Virol 80:2842–2854.

41. Storey SM, et al. (2007) Full-length, glycosylated NSP4 is localized to plasma membranecaveolae by a novel raft isolation technique. J Virol 81:5472–5483.

42. Mir KD, Parr RD, Schroeder F, Ball JM (2007) Rotavirus NSP4 interacts with both theamino- and carboxyl-termini of caveolin-1. Virus Res 126:106–115.

43. Xu A, Bellamy AR, Taylor JA (2000) Immobilization of the early secretory pathway bya virus glycoprotein that binds to microtubules. EMBO J 19:6465–6474.

44. Morita CT, et al. (2001) Superantigen recognition by gammadelta T cells: SEA recog-nition site for human Vgamma2 T cell receptors. Immunity 14:331–344.

45. Fukuta S, Magnani JL, Twiddy EM, Holmes RK, Ginsburg V (1988) Comparison of thecarbohydrate-binding specificities of cholera toxin and Escherichia coli heat-labileenterotoxins LTh-I, LT-IIa, and LT-IIb. Infect Immun 56:1748–1753.

46. Bienkowska J, Cruz M, Atiemo A, Handin R, Liddington R (1997) The von Willebrandfactor A3 domain does not contain a metal ion-dependent adhesion site motif. J BiolChem 272:25162–25167.

47. Springer TA, Wang JH (2004) The three-dimensional structure of integrins and theirligands, and conformational regulation of cell adhesion. Adv Protein Chem 68:29–63.

48. Bowman GD, et al. (2000) Crystal structure of the oligomerization domain of NSP4 fromrotavirus reveals a core metal-binding site. J Mol Biol 304:861–871.

49. Ito H, Sugiyama M, Masubuchi K, Mori Y, Minamoto N (2001) Complete nucleotidesequence of a group A avian rotavirus genome and a comparison with its counterpartsof mammalian rotaviruses. Virus Res 75:123–138.

50. Howe A, Aplin AE, Alahari SK, Juliano RL (1998) Integrin signaling and cell growthcontrol. Curr Opin Cell Biol 10:220–231.

51. Schwartz MA, Baron V (1999) Interactions between mitogenic stimuli, or, a thousandand one connections. Curr Opin Cell Biol 11:197–202.

52. Cuadras MA, Feigelstock DA, An S, Greenberg HB (2002) Gene expression pattern inCaco-2 cells following rotavirus infection. J Virol 76:4467–4482.

53. Halasz P, Holloway G, Turner SJ, Coulson BS (2008) Rotavirus replication in intestinalcells differentially regulates integrin expression by a phosphatidylinositol 3-kinase-dependent pathway, resulting in increased cell adhesion and virus yield. J Virol82:148–160.

54. Bergelson JM, Shepley MP, Chan BM, Hemler ME, Finberg RW (1992) Identification ofthe integrin VLA-2 as a receptor for echovirus 1. Science 255:1718–1720.

55. Feire AL, Koss H, Compton T (2004) Cellular integrins function as entry receptors forhuman cytomegalovirus via a highly conserved disintegrin-like domain. Proc Natl AcadSci USA 101:15470–15475.

56. Coulson BS, Londrigan SL, Lee DJ (1997) Rotavirus contains integrin ligand sequencesand a disintegrin-like domain that are implicated in virus entry into cells. Proc NatlAcad Sci USA 94:5389–5394.

57. King SL, et al. (1997) Echovirus 1 interaction with the human very late antigen-2(integrin alpha2beta1) I domain. Identification of two independent virus contact sitesdistinct from the metal ion-dependent adhesion site. J Biol Chem 272:28518–28522.

58. Londrigan SL, Graham KL, Takada Y, Halasz P, Coulson BS (2003) Monkey rotavirusbinding to alpha2beta1 integrin requires the alpha2 I domain and is facilitated by thehomologous beta1 subunit. J Virol 77:9486–9501.

59. Edelson BT, et al. (2006) Novel collectin/C1q receptor mediates mast cell activation andinnate immunity. Blood 107:143–150.

60. Edelson BT, Li Z, Pappan LK, Zutter MM (2004) Mast cell-mediated inflammatoryresponses require the alpha 2 beta 1 integrin. Blood 103:2214–2220.

61. de Fougerolles AR, et al. (2000) Regulation of inflammation by collagen-bindingintegrins alpha1beta1 and alpha2beta1 in models of hypersensitivity and arthritis.J Clin Invest 105:721–729.

62. Suzuki K, et al. (2007) Semaphorin 7A initiates T-cell-mediated inflammatory responsesthrough alpha1beta1 integrin. Nature 446:680–684.

63. BeaulieuJF (1999) Integrinsandhumanintestinal cell functions.FrontBiosci4:D310–D321.64. Tafazoli F, Holmstrom A, Forsberg A, Magnusson KE (2000) Apically exposed, tight

junction-associated beta1-integrins allow binding and YopE-mediated perturbationof epithelial barriers by wild-type Yersinia bacteria. Infect Immun 68:5335–5343.

65. Bugarcic A, Taylor JA (2006) Rotavirus nonstructural glycoprotein NSP4 is secreted fromthe apical surfaces of polarized epithelial cells. J Virol 80:12343–12349.

66. Rich RL, et al. (1999) Trench-shaped binding sites promote multiple classes of interac-tions between collagen and the adherence receptors, alpha(1)beta(1) integrin andStaphylococcus aureus cna MSCRAMM. J Biol Chem 274:24906–24913.

67. Xu Y, et al. (2000) Multiple binding sites in collagen type I for the integrins alpha1beta1and alpha2beta1. J Biol Chem 275:38981–38989.

68. Huang H, Schroeder F, Estes MK, McPherson T, Ball JM (2004) Interaction(s) of rotavirusnon-structural protein 4 (NSP4) C-terminal peptides with model membranes. BiochemJ 380:723–733.

69. McKinney MM, Parkinson A (1987) A simple, non-chromatographic procedure to purifyimmunoglobulins from serum and ascites fluid. J Immunol Methods 96:271–278.

8818 � www.pnas.org�cgi�doi�10.1073�pnas.0803934105 Seo et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1