structure, vol. 13, 1019–1033, july, 2005, ©2005 elsevier...

Structure, Vol. 13, 1019–1033, July, 2005, ©2005 Elsevier Ltd All rights reserved. DOI 10.1016/j.str.2005.04.011

The Crystal Structure of Coxsackievirus A21and Its Interaction with ICAM-1

Chuan Xiao,1 Carol M. Bator-Kelly,1

Elizabeth Rieder,2 Paul R. Chipman,1 Alister Craig,3

Richard J. Kuhn,1 Eckard Wimmer,2

and Michael G. Rossmann1,*1Department of Biological SciencesPurdue University915 West State StreetWest Lafayette, Indiana 479072Department of Molecular Genetics and MicrobiologySchool of MedicineState University of New York at Stony BrookStony Brook, New York 117943Liverpool School of Tropical MedicinePembroke PlaceLiverpool L3 5QAUnited Kingdom

Summary

CVA21 and polioviruses both belong to the Entero-virus genus in the family of Picornaviridae, whereasrhinoviruses form a distinct picornavirus genus. Nev-ertheless, CVA21 and the major group of human rhino-viruses recognize intercellular adhesion molecule-1(ICAM-1) as their cellular receptor, whereas poliovi-ruses use poliovirus receptor. The crystal structureof CVA21 has been determined to 3.2 Å resolution. Itsstructure has greater similarity to poliovirus struc-tures than to other known picornavirus structures.Cryo-electron microscopy (cryo-EM) was used to de-termine an 8.0 Å resolution structure of CVA21 com-plexed with an ICAM-1 variant, ICAM-1Kilifi. The cryo-EM map was fitted with the crystal structures ofICAM-1 and CVA21. Significant differences in the struc-ture of CVA21 with respect to the poliovirus structuresaccount for the inability of ICAM-1 to bind polioviruses.The interface between CVA21 and ICAM-1 has shapeand electrostatic complementarity with many resi-dues being conserved among those CVAs that bindICAM-1.

Introduction

Coxsackieviruses belong to the family of Picornaviri-dae, characterized by being small, naked, icosahedralanimal viruses with a positive-sense, single-strandedRNA genome (Semler and Wimmer, 2002). The three-dimensional structures of various picornaviruses havebeen determined to near atomic resolution by X-raycrystallography (Rossmann, 2002). Examples includethe major group human rhinovirus (HRV) serotypes 14(Rossmann et al., 1985) and 16 (Hadfield et al., 1997);the minor group human rhinovirus serotypes 1A (Kim etal., 1989) and 2 (Verdaguer et al., 2000); human poliovi-rus (PV) serotypes 1 (Hogle et al., 1985), 2 (Lentz et al.,1997), and 3 (Filman et al., 1989); and coxsackievirus

*Correspondence: [email protected]

serotypes B3 (Muckelbauer et al., 1995) and A9 (Hendryet al., 1999). The external diameter of the viral particlesis approximately 310 Å. Packed inside the virion is theapproximately 7400 base pair long RNA genome with asmall peptide, VPg, covalently linked to the 5# end.These viruses are assembled from 60 protomers, eachcomposed of one copy of the viral proteins VP1–VP4.The major, external capsid proteins VP1, VP2, and VP3are arranged in a pseudo T = 3 icosahedral lattice witheach protein having a “jelly-roll” structure, consisting oftwo antiparallel BIDG and CHEF opposing sheets, inwhich individual β strands are named from A to I alongthe polypeptide chain. VP4, together with the aminotermini of VP1, VP2, and VP3, is situated internally, con-tributing to the protein/RNA interface.

The Picornaviridae family has been divided into ninegenera: Enterovirus, Rhinovirus, Cardiovirus, Aphthovi-rus, Hepatovirus, Parechovirus, Erbovirus, Kobuvirus,and Teschovirus. Coxsackieviruses and poliovirusesbelong to the Enterovirus genus, whereas human rhino-viruses belong to the Rhinovirus genus (Stanway et al.,2002). Based on the different pathogenicity and histo-pathological lesions observed in newborn mice, cox-sackieviruses were initially divided into two groups.Group A contains 23 serotypes, and Group B contains6 serotypes. Group A coxsackieviruses (CVA) induceflaccid paralysis and mainly affect skeletal muscle.Group B coxsackieviruses (CVB) cause spastic paraly-sis and pathological changes in several tissues (Hyypiäet al., 1993). Later, based on sequence homology,coxsackieviruses were reclassified into three humanenterovirus (HEV) clusters, HEV-A, HEV-B, and HEV-C(Stanway et al., 2002), in which CVA21, polioviruses,and another ten CVAs belong to the HEV-C cluster.CVA21 (Figure 1) is closely related to polioviruses, with56.4%, 72.4%, 74.2%, and 75.4% amino acid identityin VP1, VP2, VP3, and VP4 for PV1, respectively. Fornonstructural proteins, such as protease 3C and poly-merase 3D, CVA21 has more than 94% amino acididentity. Similar amino acid identity is seen when com-paring CVA21 with other polioviruses. The overall ge-nome homology between CVA21 and PVs is between81.4% and 82.3% (Hughes et al., 1989).

Picornavirus infection is initiated by the attachmentof the virus to cell surface receptors. Molecular andstructural studies have identified diverse classes of re-ceptors (Rieder and Wimmer, 2002). CVA21, as well asthe major group HRVs, use intercellular adhesion mole-cule-1 (ICAM-1, CD54) as their cellular receptor (Olsonet al., 1993; Shafren et al., 1997a; Staunton et al., 1990;Xiao et al., 2001) and cause common cold-like respira-tory infections. The function of ICAM-1 is as a cell sur-face glycoprotein that facilitates cell adhesion betweenleukocytes and endothelium by binding to lymphocytefunction-associated antigen-1 (LFA-1, CD11a/CD18) (Mar-lin and Springer, 1987) and the plasma protein fibrinogen(Languino et al., 1993). Transgenic mice expressing hu-man ICAM-1 develop classic paralytic poliomyelitis af-ter being infected with CVA21, indicating that CVA21may share PVs’ poliomyelitis potential (Dufresne and

Structure1020

Figure 1. Structure of CVA21 and Its Differences to PV1

(A) Stereo view of a surface-shaded model of CVA21 calculated to 3.2 Å resolution showing the depressed canyon around each 5-fold axis.The radial distance of the surface is indicated by color, from blue (w120 Å) to red (165 Å).(B) Stereo view of a ribbon diagram of one CVA21 icosahedral asymmetric unit. VP1, VP2, VP3, and VP4 are colored blue, green, red, andcyan, respectively. One icosahedral asymmetric unit is outlined in both this panel and in (A) by a black triangle with 5-fold, 3-fold, and 2-foldsymmetry symbols at the corners and along the base of the triangle.(C) Plot of Cα atom distance between PV1 and CVA21 versus the residue number. The higher peaks are labeled with their correspondingsecondary structure.(D) Stereo diagrams of the backbone Cα trace of CVA21 viral proteins (red) superimposed with those of PV1 (blue).

Coxsackievirus A21/Receptor Interaction1021

Table 1. Data Collection and Structure Refinement of CVA21 Crystals

pH 5.6 pH 6.4 pH 7.2

Data Collection and Processing

Beam wavelength (Å) 1.000 1.000 1.000Detector ADSC Quantum4 ADSC Quantum4 ADSC Quantum4Crystal-to-CCD distance (mm) 300 250 300Oscillation angle (°) 0.2 0.4 0.2Mosaicity (°) 0.3 0.3 0.3Number of crystals (number of frames) 1 (300) 2 (230) 2 (480)Exposure time per frame (s) 20 40–45 30Space group (cell dimensions, Å) P4232 (a = 346.4) P4232 (a = 348.0) P4232 (a = 346.6)Resolution range processed (Å) 50.0–3.2 50.0–3.2 50.0–3.2Number of reflections 1,223,590 1,815,084 2,242,144Number of unique reflections 103,249 106,927 110,466Rejection criterion (I/σ(I)) 3.0 3.0 3.0Completeness (%)a 97.4 (87.0) 97.1 (88.9) 99.7 (99.3)Redundancya 11.9 (6.0) 17.0 (11.0) 20.3 (10.8)Rmerge (%)a 9.2 (24.3) 11.9 (23.8) 13.3 (25.3)

Average and Phase Extension

Final correlation coefficienta 0.95 (0.38) 0.96 (0.68) 0.96 (0.66)

Refinement

Number of reflections in working set 99,078 100,039 102,151Number of reflections in test set 11,089 11,161 11,412Rfree (%) 26.7 23.5 27.0Rwork (%) 25.8 22.5 25.4

Final Statistics

Protein modelAtoms 6658Residues 858

Non-protein atomsSolvent (waters) 53Nucleotide 23Myristoyl group 15Pocket factor 16Ion 1

Rms deviations from idealityBond length (Å) 0.009 0.008 0.009Bond angles (°) 1.45 1.41 1.45

a Values in parentheses are for the highest-resolution shell (3.31–3.20 Å).

Gromeier, 2004). These data further support the closerelationship between PVs and CVA21.

CVA21 can also bind to decay-accelerating factor(DAF, CD55) (Newcombe et al., 2003; Shafren et al.,1997b), a complement regulatory protein. DAF containsfour extracellular, short consensus repeats and is an-chored by glycosylphosphatidylinositol on the cell sur-face. DAF is also a cellular receptor for many humanenteroviruses (Bergelson et al., 1994, 1995; Bhella etal., 2004; He et al., 2002; Karnauchow et al., 1996).Binding of DAF does not trigger conformational changeof the CVA21 particles (Shafren, 1998), nor is it suffi-cient to produce an infection, requiring ICAM-1 for thecell entry (Shafren et al., 1997b). The major role of DAFmay be to capture and concentrate infectious virionson the cell membrane (Newcombe et al., 2004). CVA21is perhaps the only known virus that can bind to bothDAF and ICAM-1. Because both of these cell surfacemolecules are overexpressed on malignant melanomacells, CVA21 has been used to selectively lyse tumorcells both in vitro and in vivo (Shafren et al., 2004).

Notwithstanding the high sequence similarity be-tween polioviruses and CVA21, polioviruses use polio-

virus receptor (PVR, CD155) instead of ICAM-1 or DAFas their cellular receptor (Belnap et al., 2000b; He et al.,2000; Xing et al., 2000). Although both ICAM-1 and PVRbind into the “canyons” surrounding each icosahedral5-fold vertex, the orientation of these long molecules isvery different (Xiao et al., 2001). Binding of ICAM-1 orPVR to the corresponding virus surface triggers confor-mational changes (Arita et al., 1998; Shafren et al.,1997a) that release a “pocket factor,” causing destabili-zation of the virus and release of the genome (Ross-mann, 1994). The pocket factor is an unidentified fattyacid-like molecule situated in a hydrophobic pocket un-derneath the floor of the canyon within VP1 (Filman etal., 1989; Hendry et al., 1999; Rossmann, 1994). Hy-drophobic compounds can replace the pocket factorand inhibit the disassembly of virions (Smith et al.,1986).

A natural variant of ICAM-1, ICAM-1Kilifi, was firstfound within certain African populations in whichmalaria is endemic (Fernandez-Reyes et al., 1997).ICAM-1 is used by the malarial parasite Plasmodiumfalciparum-infected erythrocytes as a receptor to at-tach to other cells. The only amino acid sequence dif-

Structure1022

Figure 2. Stereo Diagrams of Densities in Se-lected Regions of CVA21

(A) One putative guanine nucleotide stacksagainst Trp2038.(B) Pocket factor (yellow).(C) Myristylation (yellow) of VP4 N termini.(D) A possible calcium atom (yellow) at theN-terminal region of VP1.

ference between ICAM-1 and ICAM-1Kilifi is Lys29, nbwhich is a Met in ICAM-1Kilifi. Although individuals that

are homozygous for the Met29 allele have increased fsusceptibility to cerebral malaria, they also have a re-duced immune responsiveness because ICAM-1Kilifi re- s

aduces its avidity to LFA-1 and loses its binding abilityto fibrinogen. These observations imply a possible bal- e

mance in selection between pathogen infection and in-flammatory responses (Craig et al., 2000). In vitro ELISA t

Dexperiments and cryo-electron microscopy (cryo-EM)reconstructions have shown that ICAM-1Kilifi, unlike S

ormal ICAM-1, is unable to bind to HRV16. However,oth variants of ICAM-1 can bind to HRV14 with dif-

erent affinity (Xiao et al., 2004).ICAM-1 has five extracellular immunoglobulin-like

uperfamily (IgSF) domains, a transmembrane domain,nd a short carboxyl-terminal cytoplasmic tail. The fivextracellular IgSF domains form a long rod-shapedolecule with a length of approximately 190 Å (Staun-

on et al., 1990). The structures of N-terminal domains1 and D2 (Bella et al., 1998; Casasnovas et al., 1998;himaoka et al., 2003) and of domains D3, D4, and D5


Figure 3. Cryo-EM Map of CVA21/ICAM-1 Complexes

Densities corresponding to CVA21, ICAM-1KilifiFc, and ICAM-1 are colored purple, blue, and green, respectively.(A) Stereo view of CVA21 complexed with ICAM-1KilifiFc calculated to 26 Å.(B) Stereo view of CVA21 complexed with ICAM-1.(C) Stereo diagram of ICAM-1KilifiFc density (blue) around one 2-fold axis superimposed with ICAM-1 density (green). Two ICAM-1Kilifi mole-cules are bent closer to the 2-fold axis than are those in ICAM-1. Cα trace of fitted ICAM-1Kilifi (red) was superimposed with the original X-raystructures (yellow) without changing the elbow angles.(D) One-eighth of the complex cryo-EM map calculated to 26 Å resolution at 0σ (standard deviation of the map) above the average valueshowing a layer of low height densities corresponding to the domain D4, D5, and Fc parts of ICAM-1KilifiFc.(E–H) Cryo-EM densities around one 2-fold axis showing one pair of ICAM-1KilifiFc and part of the virus surface at 26, 16, 11, and 9 Årespectively. The left and right panels show the densities at 0.25 and 1.0σ, respectively, above the average density height.

Structure1024

Table 2. Fitting X-Ray Structures into Cryo-EM Density

Statistics of EMfit Results

X-Ray Structures Residue Rangea Sumf b

ICAM-1Kilifi domain D1 1–83 1.31 (1.36)c

ICAM-1 domain D2 84–185 0.88ICAM-1 domain D3 186–281 0.43ICAM-1 domain D4 282–366 0.22ICAM-1 domain D5 367–450 0.15CVA21 VP1 total 1016–1298 1.68CVA21 VP1 N terminus 1016–1078 0.93CVA21 VP1 C terminus 1282–1298 1.74CVA21 VP2 total 2006–2272 1.76CVA21 VP2 N terminus 2006–2031 1.08CVA21 VP2 C terminus 2263–2272 1.43CVA21 VP3 total 3001–3239 1.75CVA21 VP3 N terminus 3001–3051 1.24CVA21 VP3 C terminus 3224–3239 1.80CVA21 VP4 4001–4069 0.90

Elbow Angles among ICAM-1 Domains

Elbow Angle with Next Domain

ICAM-1 Domain Residues Defining the Long Axis Cryo-EM X-Ray

D1 Pro28, Tyr83 177.6 171.2D2 Pro115, Phe185 149.7 154.6D3 Pro217, Tyr281 148.2 —D4 Phe283, Tyr367 131.7 143.2D5 Pro398, Pro450 — —

Overall Contact Statistics

Comparative Potential Energy (kcal/mol)

Number of Number TotalBuried Putative H Number of of Other Number of H Electrostatic VDWd Total

Complex Surface (Å2) Bonds Ion Pairs Interactions Interactions Bond Interaction Interaction Interaction

CVA21/ICAM-1Kilifi 940.1 19 6 22 41 −30.3 −394.3 62.9 −487.5CVA21/ICAM-1 969.5 22 8 25 47 −32.9 −516.0 75.9 −624.8

a Amino acids in the virus are numbered nxxx, where n is the VP number and xxx is the amino acid sequence number.b Sumf is the normalized mean density over all atoms calculated by the program EMfit (see Experimental Procedures).c The value in parentheses is sumf of ICAM-1 with Lys29 instead of Met29 as in ICAM-1Kilifi.d VDW = van der Waals.

(Yang et al., 2004) of ICAM-1 have been determined by qgX-ray crystallography. Only the first amino-terminal do-

main, domain D1, of ICAM-1 is involved in virus/recep- pator interaction (Kolatkar et al., 1999; Olson et al., 1993;

Xiao et al., 2001) and adhesion with LFA-1 (Shimaoka bcet al., 2003).

An IgSF fold is a β barrel consisting of seven β hdstrands labeled A–G along the polypeptide chain. The

two sides of the barrel are formed by two antiparallel acβ sheets, ABED and GFC, surrounding a hydrophobic

interior. The structure of domain D1 has been classified Daas an “intermediate subset 1” (I1) IgSF fold, in which

there is an extra β strand, A#, in the GFC sheet (Wang aaand Springer, 1998). The B-C, D-E, and F-G loops form

the tip of domain D1 that binds into the viral canyon. vaMet29 in ICAM-1Kilifi is located at the center of the B-C

loop. Domain D2 has been classified as an “intermedi-sate subset 2” (I2) IgSF fold (Casasnovas et al., 1998;

Wang and Springer, 1998), which has an additional C# uestrand in the A#GFC β sheet. Domains D3, D4, and D5

have been classified as having I1, I2, and a distorted I2 EhIgSF fold, respectively (Yang et al., 2004).

ICAM-1 is heavily glycosylated with eight potential sXN-linked glycosylation sites based on amino acid se-

uence analyses (Staunton et al., 1988). Enzymatic di-estion (Tomassini et al., 1989) and X-ray crystallogra-hy (Bella et al., 1998; Casasnovas et al., 1998; Yang etl., 2004) have shown that at least seven of these cane glycosylated. Domain D1 does not have any gly-osylation sites, whereas domain D2 is the mosteavily glycosylated domain, containing four carbohy-rate sites associated with Asn103, Asn118, Asn156,nd Asn175. Domain D3 has two glycosylation sites lo-ated at Asn240 and Asn269. Glycosylation of domain3 may play a role in ICAM-1 dimerization (Reilly etl., 1995). The glycosylation sites have been used asnchors to fit the X-ray structure of ICAM-1 domains D1nd D2 into the low-resolution cryo-EM maps of variousirus/receptor complexes (Kolatkar et al., 1999; Xiao etl., 2001).X-ray crystallography and cryo-EM image recon-

tructions have been combined to study a number ofnstable picornavirus/receptor complexes (Rossmannt al., 2002). One of these was a 26 Å resolution cryo-M map of a complex of CVA21 with ICAM-1, whichad been interpreted by using the homologous PV1tructure to represent the virus and the D1-D2 ICAM-1-ray crystallographic structure of the receptor (Xiao et


Figure 4. Cryo-EM Densities of CVA21/ICAM-1

(A) Stereo diagram of cryo-EM densities within one asymmetricunit, in which densities belonging to ICAM-1Kilifi, VP1, VP2, and VP3are colored yellow, blue, green, and red, respectively.(B) Same densities as in (A), but in transparent cyan and fitted withbackbone Cα trace of ICAM-1Kilifi (yellow), VP1 (blue), VP2 (green),and VP3 (red).(C) Cryo-EM difference densities corresponding to the ICAM-1

It has been shown that the N terminus of VP4 of at

molecule in the CVA21/ICAM-1 complex (26 Å resolution, left panel)or in the CVA21/ICAM-1Kilifi complex (8 Å resolution, right panel).The VP3 backbone is omitted.(D) One-quarter of the central slab (−20 to 20 Å on the z axis) ofcryo-EM density (cyan) fitted with backbone structures of ICAM-1Kilifi (yellow), VP1 (blue), VP2 (green), VP3 (red), and VP4 (black).

al., 2001). Here, we report the crystal structure ofCVA21 determined by X-ray crystallography to a resolu-tion of 3.2 Å, as well as a cryo-EM reconstruction ofCVA21 complexed with ICAM-1Kilifi to a resolution of8.0 Å. These results provide more accurate structuralinformation of the virus/receptor interface.

Results and Discussion

Structure of CVA21The crystals of CVA21 belong to space group P4232,with 346.3 % a % 348.0 Å (Table 1). The Matthews coef-ficient (2.5 Å3/Da) shows that there are two virus par-ticles in the unit cell. Thus, each particle must be situ-ated at a special position with 23 symmetry. Given theicosahedral symmetry of the virus, this implies that thenoncrystallographic symmetry redundancy is 5. The bestR factor, obtained for crystals at pH 6.4, was 0.225 (Ta-ble 1). No significant difference was found for struc-tures determined from crystals grown at pH valuesranging from 5.5 to 7.2.

As expected, among the known structures of picor-naviruses, CVA21 has the greatest similarity to poliovi-ruses (Figure 1). The biggest differences occur at thechain termini and in some loop regions, such as the car-boxyl termini of VP1 and VP3 (forming part of the canyonrim), parts of the internal VP4 protein, the B-C and D-Eloops of VP1 (forming the 5-fold vertex of the virus),and the protruded E-F loop region of VP2 (the “puff”region) (Figure 1). The puff region, which is especiallyvariable among picornaviruses, is one of the large con-tact regions between CVA21 and ICAM-1 and betweenPV and its receptor PVR.

Non-Protein Components that Bindto the CVA21 CapsidAs observed in other picornaviruss, including PV (Fil-man et al., 1989), HRV14 (Arnold and Rossmann, 1988),HRV16 (Hadfield et al., 1997), CVB3 (Muckelbauer et al.,1995), and CVA9 (Hendry et al., 1999), there is a disc-like electron density in the CVA21 map stacked againstthe highly conserved Trp2038 (amino acids are num-bered nxxx, where n is the VP number and xxx is theamino acid sequence number). A guanine nucleotidewas built into this density because tryptophans usuallystack with guanines (Jones et al., 2001) (Figure 2A).

The drug binding pocket in VP1 was found to be filledwith a pocket factor represented by a thin, but long,density. The 2.2 Å resolution map of PV1 (Protein DataBank [PDB] accession number 1HXS) had been inter-preted as having a palmitate molecule in the drug bind-ing pocket. However, the length of the pocket factordensity in CVA21 better resembles a two-carbon-shortermyristate molecule (Figure 2B).

Structure1026

Figure 5. The Receptor Binding into the Canyon

(A) Stereo diagram of the viral surface simulated from the crystal structure coordinates of CVA21 within the icosahedral asymmetric unit. Therectangle indicates the area that is enlarged in Figure 6. The region in orange corresponds to the surface region that makes contact withICAM-1Kilifi.(B) Stereo diagram of the viral surface as in (A), but colored from green to blue based on the radial distance to the center of the virus.Backbone Cα traces of ICAM-1 (pink) bound to HRV16, ICAM-1Kilifi (yellow) bound to CVA21, and PVR (red) bound to PV1 are shown onthe surface.(C) Stereo diagram of domain D1 of ICAM-1Kilifi in contact with the VP1 G-H loop and the VP2 puff region of CVA21. ICAM-1Kilifi is drawn as ayellow ribbon diagram. Backbone Cα traces of CVA21 (red) and PV1 (blue) are superimposed.

least four genera of picornaviruses is myristylated (Chow (oet al., 1987). A myristate-like density can be identified

as being associated with the amino end of VP4 in bpCVA21 (Figure 2C), although the last two carbon atoms

cannot be seen in the electron density. Similar observa- i(tions have been made for the amino ends of VP4 in PV1

(Chow et al., 1987), PV2 (Lentz et al., 1997), PV3 (Filmanaet al., 1989), HRV16 (Hadfield et al., 1997), CVA9 (Hendry

et al., 1999), and CVB3 (Muckelbauer et al., 1995), al- b1though no myristate could be seen in maps of HRV1A

Kim et al., 1989), HRV14 (Arnold and Rossmann, 1990),r HRV2 (Verdaguer et al., 2000). The myristylation haseen postulated to mediate attachment of viral capsidroteins to the cell membrane during virus assembly

nside the cell and during entry from outside the cellBelnap et al., 2000a; Chow et al., 1987).

In the structure of HRVs, divalent cations were foundlong the 5-fold and 3-fold axial cavities, thereby possi-ly increasing the stability of the virions (Zhao et al.,997). No equivalent densities were found in the CVA21


map. However, a high-density peak was found surroundedby polar groups belonging to Ser1021, Thr1022, Ser1024,and Asn1063 (Figure 2D). Thus, based on the coordina-tion geometry (Zhao et al., 1997), a putative metal ion(probably Ca2+, Zn2+, or Mg2+) was built into this den-sity. Although EDTA was used in the virus purificationand was also used at low concentration in the finalcrystallization conditions, the putative metal atom wasburied inside the virus at the N terminus of VP1, pre-venting it from being chelated by EDTA.

Cryo-EM ReconstructionThe cryo-EM reconstruction of CVA21/ICAM-1Kilifi hada resolution of 8.0 Å, a considerable improvement onthe earlier 26 Å resolution results (Xiao et al., 2001). Allfive extracellular domains (D1–D5) of ICAM-1 were visi-ble in the previous study (Figure 3B). Although in thecurrent CVA21/ICAM-1Kilifi work only about three do-mains (D1–D3) are visible, nevertheless, if the resolutionof the data is reduced to 26 Å resolution, all five do-mains, including the Fc fusion component, become vis-ible (Figures 3A and 3D). It would appear that the in-creased flexibility of the molecule at each successiveelbow between sequential domains is the cause of theloss of visibility of the outer ICAM-1 domains. At lowresolution, the molecule looks much the same no matterwhich of the many possible structures are viewed. How-ever, at high resolution, each of the different structureswill be better resolved but present only as a small frac-tion of the total number of molecular conformations,resulting in low and unrecognizable density (Figures3D–3H). This is shown by the decrease in density heightalong the ICAM-1 molecule in which the ratio of therelative heights of the densities for domains D1, D2, D3,D4, and D5 are approximately 1.0, 0.7, 0.3, 0.2, and 0.1,respectively. As the average density height of domainD1 is approximately 0.8 of the viral capsid protein shell(Table 2), presumably only about 80% of the ICAM-1binding sites are occupied.

ICAM-1 DimerizationDimerization enhances ICAM-1’s ability to bind to LFA-1(Reilly et al., 1995), as well as to HRV (Casasnovas andSpringer, 1995). Furthermore, dimeric ICAM-1 more ef-fectively triggers the HRV conformational changes thatinitiate viral infection (Martin et al., 1993). Although sol-uble ICAM-1 is mainly monomeric in vitro, full-lengthICAM-1 molecules form homodimers on cell surfacesin vivo (Miller et al., 1995; Reilly et al., 1995) with inter-molecular contacts between opposing domains D3, D5,and the transmembrane regions (Jun et al., 2001; Milleret al., 1995). The ICAM-1KilifiFc protein dimerizes insolution (Craig et al., 2000), as is also the case withICAM-1 fragments fused with various immunoglobulinmolecules (Martin et al., 1993). Similarly, there are nodirect contacts between ICAM-1 molecules bound tothe surface of CVA21, whereas when ICAM-1KilifiFc wasbound to CVA21, domains D3, D5, and Fc in adjacent2-fold-related molecules are in contact (Figures 3A, 3C,3E, and 3F). Thus, the Fc component appears to act ina manner equivalent to the transmembrane domains orimmunoglobulin fusion fragments at the carboxyl end

of ICAM-1. This similarity suggests that, on the cell sur-face, the virus might bind to a dimer of the full-lengthICAM-1. The contacts between the 2-fold-related D3domains appear to be via the carbohydrate associatedwith Asn240, as has been previously suggested (Reillyet al., 1995).

The Interface between CVA21 and ICAM-1Kilifi

In the previously published 26 Å resolution descriptionof the CVA21/ICAM-1 complex, the density represent-ing ICAM-1 has a smooth surface (Figure 4C, left panel).The current 8.0 Å resolution cryo-EM map providesmore surface features than before (Figure 4C, rightpanel), and these features can be used for an accuratefit of the ICAM-1 structure into the cryo-EM density.

The tip of the ICAM-1 D1 domain (D-E loop, B-C loop,G-F loop) contacts the VP1 βG and βH strands as wellas the VP1 G-H loop. The edge of the ICAM-1 D1 do-main, formed by the βG strand, contacts the VP1 E-Floop and the VP3 G-H loop. The opposite edge of thesame ICAM-1 domain, formed by the βD and βEstrands, contacts the puff region of VP2. However, inPVs, the structure and sequence of the highly variablepuff and the G-H loop are completely different, proba-bly accounting for the inability of ICAM-1 to bind to PVs(Figures 1, 5C, and 7B). Belnap et al. (2000b) had foundthat the pocket factor in poliovirus was missing whenPVR was bound in the canyon. This would verify thehypothesis that receptor attachment into the canyondislodges the pocket factor, thereby destabilizing thevirus. However, the ability to determine the absence orpresence of pocket factor at 8 Å resolution has beenquestioned (Xiao et al., 2004), and that remains true inthe current study.

ICAM-1 binds into the same site within the canyon ofboth CVA21 and HRVs, although the orientation ofICAM-1 relative to the respective viral surfaces is quitedifferent (Figure 5B). The interface between ICAM-1 andCVA21 has 30% less buried surface area and 40%smaller calculated interaction energy than the interfacebetween ICAM-1 and HRV14 (Xiao et al., 2004) (Table2). The shape correlation (Sc, see Experimental Pro-cedures) for ICAM-1 binding to HRV14, CVA21, or HRV16is approximately 0.69, 0.65, and 0.54, respectively, con-sistent with the observed inability of ICAM-1Kilifi bindingto HRV16 (Xiao et al., 2004). The Sc for antibody-antigencomplexes is comparable at about 0.65 (Epa and Col-man, 2001; Lawrence and Colman, 1993).

Theoretical energy calculations showed that electro-static interactions contribute 80% of the contact en-ergy (Table 2). There are four regions of ionic networksin the ICAM-1/CVA21 interface (a, b, c, and d in Figures6D and 7), as has also been observed for the interfacebetween HRVs and ICAM-1 (Xiao et al., 2004). However,for ICAM-1Kilifi, which has Lys instead of Met at residue29 in the interface, there is one less ionic interaction,which suggests that ICAM-1Kilifi may have a weaker in-teraction with CVA21 than ICAM-1, as is also the casefor HRV16 (Xiao et al., 2004).

There are five available sequences of HEV-C classCVAs that bind ICAM-1 (Newcombe et al., 2003). Manyof the CVA21 residues that are involved in forming the

Structure1028


Figure 7. Ionic Networks and Conserved Residues at the CVA21/ICAM-1Kilifi Interface

(A) Stereo diagrams showing details of the four ionic networks (a, b, c, and d) in the CVA21/ICAM-1Kilifi interface. Residues are colored byatom type: carbon (cyan), oxygen (red), and nitrogen (blue). To differentiate residues in CVA21 from those in ICAM-1Kilifi, ICAM-1Kilifi residuesare enclosed in a transparent yellow van der Waals surface. Possible hydrogen bonds are indicated by red dashed lines.(B) Alignment of CVA21, CVA20, CVA18, CVA15, CVA13, PV1, HRV16, and HRV14 sequences in the vicinity of the virus/receptor contact areas.The secondary structural elements are shown above the sequences. Residues in the virus/receptor interface are highlighted based on theresidue property, where basic, acidic, polar, and hydrophobic residues are colored in blue, red, yellow, and green, respectively. Conservedresidues between CVAs in the virus/receptor interface are outlined. Residues involved in forming ionic networks are labeled at the top of thealigned sequence with cyan-colored letters of a, b, c, and d, corresponding to the notation used in (A).

Asp2167) are not conserved, indicating that they may Pro1160, and Pro1162) associated with the ionic site

Figure 6. Road Maps of the ICAM-1Kilifi Contact Area with CVA21

(A) Stereo diagram of enlarged contact area in Figure 5A.(B) The viral (left) and ICAM-1Kilifi (right) residues that are in the virus/receptor interface are shown on a two-dimensional projection of thesame contact area (orange) as in (A). The three contour lines represent virus-receptor separation distances of 2.0 (red), 3.0 (orange), and 4.0(yellow) Å.(C) Stereo diagram of the contact area (white) with surface color based on electrostatic potentials. Positively and negatively charged surfacesare colored blue and red, respectively.(D) Two-dimensional projection of (C) with a roadmap of corresponding residues of CVA21 (left) and ICAM-1Kilifi (right). The four ionic networksites are encircled with green dashed lines and are labeled with green-colored letters of a, b, c, and d, corresponding to the notation used inFigure 7.

interface with ICAM-1 are conserved (Figure 7B). How-ever, some charged residues (Arg1161, Asp1221, and

not be contributing significantly to the virus/receptoraffinity. Four adjacent Pro residues (Pro1156, Pro1157,

Structure1030

o“a” (Figure 7A) are entirely conserved, indicating an in-qvariant conformation of this component of the ICAM-1sbinding surface. Apart from the charge interactions,2

there are large areas of hydrophobic complementarity pin the CVA21/ICAM-1 interface. However, these regions, i

dapart from the interaction of Phe/Tyr1223 with Leu30 oftICAM-1 in the B region, are not well conserved amongtother ICAM-1 binding CVAs, further indicating the im-fportance of the charge interactions.pt

Conclusions crThe availability of atomic resolution structures of bothlvirus (CVA21) and virus receptor (ICAM-1), togethertwith an improved cryo-EM map of their complex, hast

provided accurate information regarding the interfacebetween virus and receptor, indicating which residues

Fare likely to be most important in forming a stable com- Cplex. These results can now be used for functional in- uvestigations with site-directed mutagenesis. c

bOExperimental ProceduresuTSamplesdCVA21 (strain Kuykendall) was prepared by an established lab pro-mtocol (Xiao et al., 2001). The fusion protein, combining ICAM-1Kilifi

oat the amino end with an Fc fragment (consisting of the two humancIgG1 domains linked by four amino acid residues) at the carboxylnend (ICAM-1KilifiFc), was produced as described previously (Craigset al., 2000).msCrystallization of CVA21wAfter CVA21 was purified as described by Xiao et al. (2001), thetvirus was at a concentration of 7 mg/ml in a buffer of 200 mM NaCl,c10 mM Tris (pH 7.2), 5% glycerol, 0.1 mM EDTA. The mother liquorwconsisted of 300 mM NaCl, 200 mM citrate buffer (pH = 5.6, 6.4,r7.2), 5% glycerol. Crystallization trials were set up by mixing 8 �lIvirus sample with 8 �l mother liquor in sitting drops. Sealed crystal-Rlization trays were incubated at 4°C for 2–3 days until the crystalscgrew to a maximum size of 0.2 × 0.2 × 0.1 mm.(In order to provide cryoprotection, glycerol and PEG400 weremadded to the mother liquor to a concentration of 20% and 10%,

respectively, prior to crystal freezing. Several data sets (Table 1)were collected on frozen (100 K) crystals at the Advanced Photon ESource of Argonne National Laboratory. T

The DENZO program (Otwinowski and Minor, 1997) was used for aindexing and integrating the diffraction data, and the SNP program lin the DPS suite (Rossmann and van Beek, 1999) was used for bscaling and merging the data from different frames. Starting phases bwere calculated to 4 Å resolution by using the known PV1 coordi- tnates (PDB accession number 2PLV). Crystallographic and noncrys- atallographic symmetry (NCS) averaging were carried out in real aspace by using the program ENVELOPE (Rossmann et al., 1992). dAfter some cycles of density averaging and solvent flattening,phases were gradually extended to 3.2 Å resolution. The program

EO (Jones et al., 1991) was used for atomic model building betweenAcycles of refinement with the program CNS (Brünger et al., 1998)iwhile constraining the atomic coordinates by the NCS symmetry.tThe final model included 53 water molecules, 1 putative Ca2+ ion,tand 1 ordered guanine nucleotide, but it omitted the disorderedtresidues 1001–1015, 2001–2005, and 3240.maCryo-EMoCVA21 infectious particles, at a final concentration of approxi-tmately 5 mg/ml, were incubated with ICAM-1Kilifi at room temper-wature for 45 min. The viruses were incubated with an excess of

receptor. There were 5.4 receptors for each virus binding site. Afterincubation, the samples were frozen for cryo-EM as described pre- I

Rviously (Xiao et al., 2001). All of the cryo-EM micrographs wererecorded on a Philips FEG300 microscope at a magnification of C

T47,000× using a voltage of 300 kV. Micrographs were then scanned

n a Zeiss Phodis SCSI scanner with a step size of 7 �m. Subse-uently, four pixels were averaged together in order to minimizepace and computing time. The final pixel size corresponded to.98 Å on the sample. The program RobEM was used to box andreprocess the boxed images. There were 33 micrographs, includ-

ng 7 focal pairs (14 micrographs), used for the reconstruction. Theefocus distance ranged from 0.7 to 3.0 �m. The EM reconstruc-

ion procedures, including the contrast transfer function correc-ions and orientation and translation determinations, were per-ormed as described previously (Xiao et al., 2004). A total of 4704articles, selected from 7979 boxed particles based on the correla-

ion coefficient between the raw image and the corresponding cal-ulated projection, were included in the reconstruction. The finalesolution was estimated by finding where the Fourier shell corre-ation fell below 0.5 by using Fourier coefficients based only onhe viral protein shell and D1 of the ICAM-1 density situated be-ween 104 Å and 177 Å radius.

itting the X-Ray Structure into the Cryo-EM Densitiesoordinates of CVA21 determined from X-ray crystallography weresed to calculate an 8.0 Å resolution map. This map was used toalibrate the pixel size of the cryo-EM map of CVA21/ICAM-1Kilifi

y using the program EMfit (Rossmann, 2000; Xiao et al., 2004).nly the density corresponding to atoms in the virus capsid wassed to establish the accurate magnification of the cryo-EM map.he accurate pixel size for the EM map was found to be 2.85 Å. Aifference map was then computed between the cryo-EM complexap and the map based on the atomic structure of the virus tobtain the density corresponding to the 60 bound ICAM-1 mole-ules. The crystal structure of ICAM-1 D1 and D2 (PDB accessionumber 1IAM), with Lys29 in domain D1 modeled as a Met to repre-ent the structure of ICAM-1Kilifi, was then fitted into the differenceap by using the program EMfit (Rossmann et al., 2001) as de-

cribed previously (Xiao et al., 2004). The sumf values (Table 2)ere normalized with respect to the root mean square (rms) devia-

ion from the mean density within the protein shell of the virus. Therystal structure of ICAM-1 D3–D5 (PDB accession number 1P53)as used to fit into a map of the complex calculated to only 16 Å

esolution instead of the final 8 Å, because the density for theCAM-1 D3–D5 is poor when higher resolution data are used (seeesults and Discussion). Nevertheless, the values of sumf were cal-ulated for D3–D5 based on the higher 8 Å resolution cryo-EM mapTable 2) in order to make the values comparable to those of do-

ains D1 and D2.

lbow Angleshe elbow angles between neighboring domains were calculateds described previously (Xiao et al., 2004). Two Cα atoms were se-

ected for each domain (Table 2) to define their long axes. The el-ow angle between two domains was defined as being the angleetween their long axes (Table 2). Although the elbow angle be-ween domains D1 and D2, between D3 and D4, and between D4nd D5, have been characterized crystallographically (Kolatkar etl., 1999; Yang et al., 2004), this is, to our knowledge, the first directetermination of the elbow angle between D2 and D3.

nergy Minimizationlthough in the fitting procedure the number of atoms of ICAM-1

n negative or zero density of the difference map was minimized,he rigid-body fitting still generated a few ICAM-1 D1 side chainshat would be in steric conflict with the virus side chains. Therefore,he CVA21/ICAM-1 complex was subjected to limited energy mini-ization by using the program CHARMM (Brooks et al., 1983). Only

toms within the virus/receptor interface were allowed to move inrder to eliminate steric collisions. After the minimization, the po-ential energies of the interactions between ICAM-1 and CVA21ere also calculated by CHARMM (Table 2).

nterface Analysesesidues in the virus/receptor interface were identified as those inVA21 that had atoms less than 4 Å from any atom in ICAM-1Kilifi.he buried surface area and the number of contacts were calcu-


lated with the program CNS (Brünger et al., 1998) (Table 2). Thesurface electrostatic potential of CVA21 and ICAM-1 was calcu-lated with the program DelPhi (Gilson and Honig, 1988) and wasplotted onto the CVA21 surface. The program SC (Lawrence andColman, 1993) was used to calculate the shape correlation (Sc)among different virus/receptor complexes. Sc measures the geo-metric surface complementarity of protein/protein interfaces.

Acknowledgments

We thank Suchetana Mukhopadhyay and Ricardo A. Bernal for helpwith crystallization and structure determination. We also thank WeiZhang, Xing Zhang, Wen Jiang, and Wah Chiu for help in improvingthe resolution of the cryo-EM map. We are grateful to Jia-huaiWang for providing us with the coordinates of the complete ICAM-1structure, to William I. Whitson for parallel computer support, aswell as to Sharon S. Wilder, Cheryl A. Towell, and Sheryl L. Kelly forhelp in the preparation of the manuscript. The work was supportedby National Institutes of Health grants AI 11219 to M.G.R. and AI15122 and GM 65030 to E.W., a Keck Foundation grant to M.G.R.(1419991440), and a Purdue University reinvestment grant.

Received: March 1, 2005Revised: April 3, 2005Accepted: April 3, 2005Published: July 12, 2005

References

Arita, M., Koike, S., Aoki, J., Horie, H., and Nomoto, A. (1998). In-teraction of poliovirus with its purified receptor and conformationalalteration in the virion. J. Virol. 72, 3578–3586.

Arnold, E., and Rossmann, M.G. (1988). The use of molecular-replacement phases for the refinement of the human rhinovirus 14structure. Acta Crystallogr. A 44, 270–282.

Arnold, E., and Rossmann, M.G. (1990). Analysis of the structure ofa common cold virus, human rhinovirus 14, refined at a resolutionof 3.0 Å. J. Mol. Biol. 211, 763–801.

Bella, J., Kolatkar, P.R., Marlor, C.W., Greve, J.M., and Rossmann,M.G. (1998). The structure of the two amino-terminal domains ofhuman ICAM-1 suggests how it functions as a rhinovirus receptorand as an LFA-1 integrin ligand. Proc. Natl. Acad. Sci. USA 95,4140–4145.

Belnap, D.M., Filman, D.J., Trus, B.L., Cheng, N., Booy, F.P., Con-way, J.F., Curry, S., Hiremath, C.N., Tsang, S.K., Steven, A.C., andHogle, J.M. (2000a). Molecular tectonic model of virus structuraltransitions: the putative cell entry states of poliovirus. J. Virol. 74,1342–1354.

Belnap, D.M., McDermott, B.M., Jr., Filman, D.J., Cheng, N., Trus,B.L., Zuccola, H.J., Racaniello, V.R., Hogle, J.M., and Steven, A.C.(2000b). Three-dimensional structure of poliovirus receptor boundto poliovirus. Proc. Natl. Acad. Sci. USA 97, 73–78.

Bergelson, J.M., Chen, M., Solomon, K.R., St. John, N.F., Lin, H.,and Finberg, R.W. (1994). Decay-accelerating factor (CD55), a gly-cosylphosphatidylinositol-anchored complement regulatory pro-tein, is a receptor for several echoviruses. Proc. Natl. Acad. Sci.USA 91, 6245–6248.

Bergelson, J.M., Mohanty, J.G., Crowell, R.L., St. John, N.F., Lublin,D.M., and Finberg, R.W. (1995). Coxsackievirus B3 adapted togrowth in RD cells binds to decay-accelerating factor (CD55). J.Virol. 69, 1903–1906.

Bhella, D., Goodfellow, I.G., Roversi, P., Pettigrew, D., Chaudhry, Y.,Evans, D.J., and Lea, S.M. (2004). The structure of echovirus type12 bound to a two-domain fragment of its cellular attachment pro-tein decay-accelerating factor (CD 55). J. Biol. Chem. 279, 8325–8332.

Brooks, B.R., Bruccoleri, R.E., Olafson, B.D., States, D.J., Swa-minathan, S., and Karplus, M. (1983). CHARMM: a program for

macromolecular energy, minimization, and dynamics calculations.J. Computat. Chem. 4, 187–217.

Brünger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P.,Grosse-Kunstleve, R.W., Jiang, J.-S., Kuszewski, J., Nilges, M.,Pannu, N.S., et al. (1998). Crystallography & NMR system: a newsoftware suite for macromolecular structure determination. ActaCrystallogr. D Biol. Crystallogr. 54, 905–921.

Casasnovas, J.M., and Springer, T.A. (1995). Kinetics and thermo-dynamics of virus binding to receptor. Studies with rhinovirus, inter-cellular adhesion molecule-1 (ICAM-1), and surface plasmon reso-nance. J. Biol. Chem. 270, 13216–13224.

Casasnovas, J.M., Stehle, T., Liu, J.H., Wang, J.H., and Springer,T.A. (1998). A dimeric crystal structure for the N-terminal two do-mains of intercellular adhesion molecule-1. Proc. Natl. Acad. Sci.USA 95, 4134–4139.

Chow, M., Newman, J.F.E., Filman, D., Hogle, J.M., Rowlands, D.J.,and Brown, F. (1987). Myristylation of picornavirus capsid proteinVP4 and its structural significance. Nature 327, 482–486.

Craig, A., Fernandez-Reyes, D., Mesri, M., McDowall, A., Altieri,D.C., Hogg, N., and Newbold, C. (2000). A functional analysis of anatural variant of intercellular adhesion molecule-1 (ICAM-1Kilifi).Hum. Mol. Genet. 9, 525–530.

Dufresne, A.T., and Gromeier, M. (2004). A nonpolio enterovirus withrespiratory tropism causes poliomyelitis in intercellular adhesionmolecule 1 transgenic mice. Proc. Natl. Acad. Sci. USA 101,13636–13641.

Epa, V.C., and Colman, P.M. (2001). Shape and electrostatic com-plementarity at viral antigen-antibody complexes. Curr. Top. Micro-biol. Immunol. 260, 45–53.

Fernandez-Reyes, D., Craig, A.G., Kyes, S.A., Peshu, N., Snow,R.W., Berendt, A.R., Marsh, K., and Newbold, C.I. (1997). A highfrequency African coding polymorphism in the N-terminal domainof ICAM-1 predisposing to cerebral malaria in Kenya. Hum. Mol.Genet. 6, 1357–1360.

Filman, D.J., Syed, R., Chow, M., Macadam, A.J., Minor, P.D., andHogle, J.M. (1989). Structural factors that control conformationaltransitions and serotype specificity in type 3 poliovirus. EMBO J. 8,1567–1579.

Gilson, M.K., and Honig, B. (1988). Calculation of the total electro-static energy of a macromolecular system: solvation energies,binding energies, and conformational analysis. Proteins 4, 7–18.

Hadfield, A.T., Lee, W., Zhao, R., Oliveira, M.A., Minor, I., Rueckert,R.R., and Rossmann, M.G. (1997). The refined structure of humanrhinovirus 16 at 2.15 Å resolution: implications for the viral life cy-cle. Structure 5, 427–441.

He, Y., Bowman, V.D., Mueller, S., Bator, C.M., Bella, J., Peng, X.,Baker, T.S., Wimmer, E., Kuhn, R.J., and Rossmann, M.G. (2000).Interaction of the poliovirus receptor with poliovirus. Proc. Natl.Acad. Sci. USA 97, 79–84.

He, Y., Lin, F., Chipman, P.R., Bator, C.M., Baker, T.S., Shoham, M.,Kuhn, R.J., Medof, M.E., and Rossmann, M.G. (2002). Structure ofdecay-accelerating factor bound to echovirus 7: a virus-receptorcomplex. Proc. Natl. Acad. Sci. USA 99, 10325–10329.

Hendry, E., Hatanaka, H., Fry, E., Smyth, M., Tate, J., Stanway, G.,Santti, J., Maaronen, M., Hyypiä, T., and Stuart, D. (1999). The crys-tal structure of coxsackievirus A9: new insights into the uncoatingmechanisms of enteroviruses. Structure 7, 1527–1538.

Hogle, J.M., Chow, M., and Filman, D.J. (1985). Three-dimensionalstructure of poliovirus at 2.9 Å resolution. Science 229, 1358–1365.

Hughes, P.J., North, C., Minor, P.D., and Stanway, G. (1989). Thecomplete nucleotide sequence of coxsackievirus A21. J. Gen. Virol.70, 2943–2952.

Hyypiä, T., Kallajoki, M., Maaronen, M., Stanway, G., Kandolf, R.,Auvinen, P., and Kalimo, H. (1993). Pathogenetic differences be-tween coxsackie A and B virus infections in newborn mice. VirusRes. 27, 71–78.

Jones, T.A., Zou, J.-Y., Cowan, S.W., and Kjeldgaard, M. (1991).Improved methods for building protein models in electron density

Structure1032

maps and the location of errors in these models. Acta Crystallogr. RrA 47, 110–119.S

Jones, S., Daley, D.T., Luscombe, N.M., Berman, H.M., and Thorn-6

ton, J.M. (2001). Protein-RNA interactions: a structural analysis. Nu-Rcleic Acids Res. 29, 943–954.3

Jun, C.-D., Carman, C.V., Redick, S.D., Shimaoka, M., Erickson, RH.P., and Springer, T.A. (2001). Ultrastructure and function of di- cmeric, soluble intercellular adhesion molecule-1 (ICAM-1). J. Biol.

RChem. 276, 29019–29027.u

Karnauchow, T.M., Tolson, D.L., Harrison, B.A., Altman, E., Lublin, (D.M., and Dimock, K. (1996). The HeLa cell receptor for enterovirus R70 is decay-accelerating factor (CD55). J. Virol. 70, 5143–5152. C

RKim, S., Smith, T.J., Chapman, M.S., Rossmann, M.C., Pevear, D.C.,GDutko, F.J., Felock, P.J., Diana, G.D., and McKinlay, M.A. (1989).ACrystal structure of human rhinovirus serotype 1A (HRV1A). J. Mol.fBiol. 210, 91–111.1

Kolatkar, P.R., Bella, J., Olson, N.H., Bator, C.M., Baker, T.S., andRRossmann, M.G. (1999). Structural studies of two rhinovirus sero-Ctypes complexed with fragments of their cellular receptor. EMBOsJ. 18, 6249–6259.R

Languino, L.R., Plescia, J., Duperray, A., Brian, A.A., Plow, E.F., Gel- etosky, J.E., and Altieri, D.C. (1993). Fibrinogen mediates leukocyte Sadhesion to vascular endothelium through an ICAM-1-dependent

Rpathway. Cell 73, 1423–1434.

tLawrence, M.C., and Colman, P.M. (1993). Shape complementarity Sat protein/protein interfaces. J. Mol. Biol. 234, 946–950. v

SLentz, K.N., Smith, A.D., Geisler, S.C., Cox, S., Buontempo, P., Skel-aton, A., DeMartino, J., Rozhon, E., Schwartz, J., Girijavallabhan, V.,

et al. (1997). Structure of poliovirus type 2 Lansing complexed with Santiviral agent SCH48973: comparison of the structural and biologi- (cal properties of the three poliovirus serotypes. Structure 5, 961– m978. V

SMarlin, S.D., and Springer, T.A. (1987). Purified intercellular adhe-Rsion molecule-1 (ICAM-1) is a ligand for lymphocyte function-asso-tciated antigen 1 (LFA-1). Cell 51, 813–819.V

Martin, S., Casasnovas, J.M., Staunton, D.E., and Springer, T.A. S(1993). Efficient neutralization and disruption of rhinovirus by chi- Bmeric ICAM-1/immunoglobulin molecules. J. Virol. 67, 3561–3568. S

mMiller, J., Knorr, R., Ferrone, M., Houdei, R., Carron, C.P., and Dus-Rtin, M.L. (1995). Intercellular adhesion molecule-1 dimerization and

its consequences for adhesion mediated by lymphocyte function SMassociated-1. J. Exp. Med. 182, 1231–1241.S

Muckelbauer, J.K., Kremer, M., Minor, I., Diana, G., Dutko, F.J.,a

Groarke, J., Pevear, D.C., and Rossmann, M.G. (1995). The struc-Sture of coxsackievirus B3 at 3.5 Å resolution. Structure 3, 653–667.R

Newcombe, N.G., Andersson, P., Johansson, E.S., Au, G.G., Lind- (berg, A.M., Barry, R.D., and Shafren, D.R. (2003). Cellular receptor ainteractions of C-cluster human group A coxsackieviruses. J. Gen. SVirol. 84, 3041–3050. l

oNewcombe, N.G., Beagley, L.G., Christiansen, D., Loveland, B.E.,iJohansson, E.S., Beagley, K.W., Barry, R.D., and Shafren, D.R.S(2004). Novel role for decay-accelerating factor in coxsackievirusSA21-mediated cell infectivity. J. Virol. 78, 12677–12682.t

Olson, N.H., Kolatkar, P.R., Oliveira, M.A., Cheng, R.H., Greve, J.M., pMcClelland, A., Baker, T.S., and Rossmann, M.G. (1993). Structure

Sof a human rhinovirus complexed with its receptor molecule. Proc.

(Natl. Acad. Sci. USA 90, 507–511.

I2Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffrac-

tion data collected in oscillation mode. Methods Enzymol. 276, T307–326. c

mReilly, P.L., Woska, J.R., Jr., Jeanfavre, D.D., McNally, E., Rothlein,VR., and Bormann, B.J. (1995). The native structure of intercellularnadhesion molecule-1 (ICAM-1) is a dimer. Correlation with binding

to LFA-1. J. Immunol. 155, 529–532. W

ieder, E., and Wimmer, E. (2002). Cellular receptor of picornavi-uses: an overview. In Molecular Biology of Picornaviruses, B.L.emler and E. Wimmer, eds. (Washington, DC: ASM Press), pp.1–70.

ossmann, M.G. (1994). Viral cell recognition and entry. Protein Sci., 1712–1725.

ossmann, M.G. (2000). Fitting atomic models into electron micros-opy maps. Acta Crystallogr. D Biol. Crystallogr. 56, 1341–1349.

ossmann, M.G. (2002). Picornavirus structure overview. In Molec-lar Biology of Picornaviruses, B.L. Semler and E. Wimmer, eds.

Washington, DC: ASM Press), pp. 27–38.

ossmann, M.G., and van Beek, C.G. (1999). Data processing. Actarystallogr. D Biol. Crystallogr. 55, 1631–1640.

ossmann, M.G., Arnold, E., Erickson, J.W., Frankenberger, E.A.,riffith, J.P., Hecht, H.J., Johnson, J.E., Kamer, G., Luo, M., Mosser,.G., et al. (1985). Structure of a human common cold virus and

unctional relationship to other picornaviruses. Nature 317, 145–53.

ossmann, M.G., McKenna, R., Tong, L., Xia, D., Dai, J., Wu, H.,hoi, H.K., and Lynch, R.E. (1992). Molecular replacement real-pace averaging. J. Appl. Crystallogr. 25, 166–180.

ossmann, M.G., Bernal, R., and Pletnev, S.V. (2001). Combininglectron microscopic with X-ray crystallographic structures. J.truct. Biol. 136, 190–200.

ossmann, M.G., He, Y., and Kuhn, R.J. (2002). Picornavirus-recep-or interactions. Trends Microbiol. 10, 324–331.

emler, B.L., and Wimmer, E. (2002). Molecular Biology of Picorna-iruses (Washington, DC: ASM Press).

hafren, D.R. (1998). Viral cell entry induced by cross-linked decay-ccelerating factor. J. Virol. 72, 9407–9412.

hafren, D.R., Dorahy, D.J., Greive, S.J., Burns, G.F., and Barry, R.D.1997a). Mouse cells expressing human intercellular adhesion

olecule-1 are susceptible to infection by coxsackievirus A21. J.irol. 71, 785–789.

hafren, D.R., Dorahy, D.J., Ingham, R.A., Burns, G.F., and Barry,.D. (1997b). Coxsackievirus A21 binds to decay-accelerating fac-

or but requires intercellular adhesion molecule 1 for cell entry. J.irol. 71, 4736–4743.

hafren, D.R., Au, G.G., Nguyen, T., Newcombe, N.G., Haley, E.S.,eagley, L., Johansson, E.S., Hersey, P., and Barry, R.D. (2004).ystemic therapy of malignant human melanoma tumors by a com-on cold-producing enterovirus, coxsackievirus A21. Clin. Canceres. 10, 53–60.

himaoka, M., Xiao, T., Liu, J.-H., Yang, Y., Dong, Y., Jun, C.-D.,cCormack, A., Zhang, R., Joachimiak, A., Takagi, J., et al. (2003).tructures of the αL I domain and its complex with ICAM-1 revealshape-shifting pathway for integrin regulation. Cell 112, 99–111.

mith, T.J., Kremer, M.J., Luo, M., Vriend, G., Arnold, E., Kamer, G.,ossmann, M.G., McKinlay, M.A., Diana, G.D., and Otto, M.J.

1986). The site of attachment in human rhinovirus 14 for antiviralgents that inhibit uncoating. Science 233, 1286–1293.

tanway, G., Hovi, T., Knowles, N.J., and Hyypiä, T. (2002). Molecu-ar and biological basis of picornavirus taxonomy. In Molecular Bi-logy of Picornaviruses, B.L. Semler and E. Wimmer, eds. (Wash-

ngton, DC: ASM Press), pp. 17–24.

taunton, D.E., Marlin, S.D., Stratowa, C., Dustin, M.L., andpringer, T.A. (1988). Primary structure of ICAM-1 demonstrates in-

eraction between members of the immunoglobulin and integrin su-ergene families. Cell 52, 925–933.

taunton, D.E., Dustin, M.L., Erickson, H.P., and Springer, T.A.1990). The arrangement of the immunoglobulin-like domains ofCAM-1 and the binding sites for LFA-1 and rhinovirus. Cell 61,43–254.

omassini, J.E., Maxson, T.R., and Colonno, R.J. (1989). Biochemi-al characterization of a glycoprotein required for rhinovirus attach-ent. J. Biol. Chem. 264, 1656–1662.

erdaguer, N., Blaas, D., and Fita, I. (2000). Structure of human rhi-ovirus serotype 2 (HRV2). J. Mol. Biol. 300, 1179–1194.

ang, J., and Springer, T.A. (1998). Structural specializations of im-


munoglobulin superfamily members for adhesion to integrins andviruses. Immunol. Rev. 163, 197–215.

Xiao, C., Bator, C.M., Bowman, V.D., Rieder, E., He, Y., Hébert, B.,Bella, J., Baker, T.S., Wimmer, E., Kuhn, R.J., and Rossmann, M.G.(2001). Interaction of coxsackievirus A21 with its cellular receptor,ICAM-1. J. Virol. 75, 2444–2451.

Xiao, C., Tuthill, T.J., Bator Kelly, C.M., Challinor, L.J., Chipman,P.R., Killington, R.A., Rowlands, D.J., Craig, A., and Rossmann,M.G. (2004). Discrimination among rhinovirus serotypes for a vari-ant ICAM-1 receptor molecule. J. Virol. 78, 10034–10044.

Xing, L., Tjarnlund, K., Lindqvist, B., Kaplan, G.G., Feigelstock, D.,Cheng, R.H., and Casasnovas, J.M. (2000). Distinct cellular recep-tor interactions in poliovirus and rhinoviruses. EMBO J. 19, 1207–1216.

Yang, Y., Jun, C.-D., Liu, J.-H., Zhang, R., Joachimiak, A., Springer,T.A., and Wang, J.-H. (2004). Structural basis for dimerization ofICAM-1 on the cell surface. Mol. Cell 14, 269–276.

Zhao, R., Hadfield, A.T., Kremer, M.J., and Rossmann, M.G. (1997).Cations in human rhinoviruses. Virology 227, 13–23.

Accession Numbers

The refined CVA21 atomic coordinates have been deposited withthe PDB as accession number 1Z7S. The atomic coordinates of theCVA21/ICAM-1Kilifi complex, which were fitted into the cryo-EMmap and then energy minimized, have also been deposited withPDB accession number 1Z7Z. The cryo-EM map has been depos-ited with the Macromolecular Structure Database at the EuropeanBioinformatic Institute as accession number EMD-1114.

structure, vol. 13, 1019–1033, july, 2005, ©2005 elsevier...

Documents