GAG-binding variants of tick-borne encephalitis virus

Virology 398 (2010) 262–272

Contents lists available at ScienceDirect

Virology

j ourna l homepage: www.e lsev ie r.com/ locate /yv i ro

GAG-binding variants of tick-borne encephalitis virus

L.I. Kozlovskaya a, D.I. Osolodkin b, A.S. Shevtsova a, L.Iu. Romanova a, Y.V. Rogova a, T.I. Dzhivanian a,V.N. Lyapustin a, G.P. Pivanova a, A.P. Gmyl a, V.A. Palyulin b, G.G. Karganova a,⁎a Chumakov Institute of Poliomyelitis and Viral Encephalitides, Russian Academy of Medical Sciences, Moscow Region 142782, Russiab Moscow State University, Department of Chemistry, Moscow 119991, Russia

⁎ Corresponding author. Fax: +7 4954399321.E-mail address: [email protected] (G.G. Karganova).

0042-6822/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.virol.2009.12.012

a b s t r a c t

Article history:Received 5 November 2009Returned to author for revision25 November 2009Accepted 10 December 2009Available online 12 January 2010

Keywords:Tick-borne encephalitis virusGlycosaminoglycanEnvelope proteinMolecular dynamicsAttenuationHemagglutinationFlavivirusNeuroinvasivenessGAG-binding variants

Previously different authors described various flavivirus mutants with high affinity to cell glycosaminogly-cans and low neuroinvasiveness in mice that were obtained consequently passages in cell cultures or in ticks.In present study the analysis of TBEV isolates has shown existence of GAG-binding variants in natural viruspopulation. Affinity to GAG has been evaluated by sorption on heparin-Sepharose. GAG-binding phenotypecorresponds to such virus properties, like small plaque phenotype in PEK cells, absence of hemagglutinationat pH 6.4, and low neuroinvasiveness in mice. Mutations increasing charge of E protein were necessary butnot sufficient for acquisition of GAG-binding phenotype. Molecular modeling and molecular dynamicssimulation have shown that the flexibility of E protein molecule could bear influence on the phenotypicmanifestation of substitutions increasing charge of the virions.

© 2009 Elsevier Inc. All rights reserved.

Introduction

For wide host range viruses, like arboviruses, the first stages ofvirus–cell interaction remain unclear. It is obvious, that there are atleast three possibilities for such viruses: to use as receptors commonmolecules represented on various cell types, to use distinct receptorson different cell types or to entry cell by receptor-independentmechanism.

Glycosaminoglycans (GAGs) are polydispense mixture of linearpolysaccharides consisting primarily of N-acetylated and N-sulphateddisaccharides arranged mainly in segregated negatively chargeddomains. GAGs could be divided into: hyaluronan, chondroitin sulfateand dermatan sulfate, heparan sulfate and heparin, and keratansulfate (Alberts et al., 2008). Heparan sulphate is presented on the cellsurface in the form of heparan sulphate proteoglycans (HSPGs) ofsyndecan family (Gallagher et al., 1992). All ever studied cells(excluding B-stem cells), and organs express at least one syndecanfamily member (Kim et al., 1994). It has been shown previously thatheparan sulfate GAGs serve as a receptor for various viruses (reviewedin Schneider-Schaulies, 2000). GAGs can serve as a virion concentra-tion initial attachment molecule on the cell surface (Haywood, 1994).

ll rights reserved.

In most cases HSPGs are not the unique receptors involved in viralattachment and entry processes, and virus is still able to infect cellsdeficient in GAG production or after GAGs removal from the cellularmembrane (Byrnes and Griffin, 1998; Kroschewski et al., 2003;Martínez-Barragán and del Angel, 2001). However, in some casesswitch of the receptor molecule (from ICAM-1 to GAG) could changethe course of infection (Khan et al., 2007). Laboratory obtainedvariants with increased affinity to GAGs were shown for severalviruses (Bernard et al., 2000; Heil et al., 2001; Hulst et al., 2000;Klimstra et al., 1998; Sa-Carvalho et al., 1997). These variants wereattenuated for various animals.

Flavivirus genus gathers arthropod-transmitted viruses with virionsurrounded by lipid bilayer, which contains multiple copies ofproteins E and M, but only E protein forms outer surface. Thus, Eglycoprotein serves for viral attachment, entry and other stages ofviral penetration into the cell; besides, it's the main target forneutralizing antibodies (Lindebach et al., 2007). E protein includesectodomain (sE) that is connected to the membrane by stem-anchorregion. The structure of ectodomain was first solved by Rey and co-authors (1995) for tick-borne encephalitis virus (TBEV) and thenwere obtained structures for other genus members (Zhang et al.,2003, 2004; Nybakken et al., 2006; Volk et al., 2006). The sE consists of3 distinct domains: central structural domain I, domain II involvedinto the process of pH-dependent membrane fusion, and receptorattachment domain III (Rey et al., 1995).

263L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

Flaviviruses, like West Nile virus (WNV) and Japanese encephalitisvirus (JEV), incorporate into the target cell by clathrin-mediatedendocytosis (Chu and Ng, 2004a; Nawa et al., 2003). A lot of moleculeswere described as receptors for flaviviruses: integrins (Chu and Ng,2004b; Protopopova et al., 1997), 37/67-kD high affinity lamininreceptor (Protopopova et al., 1997; Thepparit and Smith, 2004),ICAMs (Navarro-Sanchez et al., 2003; Tassaneetrithep et al., 2003),GAGs (Goto et al., 2003; Kroschewski et al., 2003; Lee and Lobigs,2000; Mandl et al., 2001; Martínez-Barragán and del Angel, 2001), etc.It has been shown on TBEV that flaviviruses have 2 types of receptorson the cell surface: high- and low-affinity (Maldov et al., 1992). Therole of each receptor type remains unclear, but GAGs seem to be awidespread low-affinity flavivirus receptor.

GAG-binding variants were described for various Flavivirus genusmembers: Yellow fever virus (YFV), Murray valley encephalitis virus(MVEV), WNV, JEV, and TBEV. These variants were obtained duringlaboratory passages in cell lines: BHK-21 (Goto et al., 2003; Mandl etal., 2001), SW-13 (Lee and Lobigs, 2002; Lee et al., 2004), Neuro-2(Chiou and Chen, 2007), or in ticks (Romanova et al., 2007).Previously, about 2 dozen of increasing charge substitutions in all 3domains of soluble part of E protein molecules of TBEV variants,obtained during cell line passages, leading to the appearance of GAG-binding phenotype and specific properties, were identified (Mandl etal., 2001). The main properties were small plaque phenotype inporcine kidney cell lines and low neuroinvasiveness for laboratorymice. The real impact of such mutations in the E protein structuralchanges has remained unknown. The existence of such variants innatural population of flaviviruses also remains unclear. TBEV strainswithmutations increasing surface charge of the virion were depositedearlier (Ecker et al., 1999; Khasnatinov et al., 2009; Lu et al., 2008), buttheir affinity to cellular GAGs and other characteristics have not beenstudied properly. It is natural to think that the presence of suchmutations leads to appearance of GAG-binding phenotype.

During the present work we tried to solve two main problems: todescribe affinity to GAGs of natural TBEV isolates and to elucidatewhether every mutation increasing charge of E protein leads toappearance of GAG-binding phenotype. First, we analyzed propertiesof 13 pre-existing collection strains isolated in various parts of Russiaand Baltics, including sequences of E protein, plaque phenotype,binding to heparin-Sepharose (HS), etc. Viral characteristics corre-spondedwith increased sorption on heparin-Sepharose–GAG-binding

Table 1TBEV strains used in present work.

TBEV strain/clone Region and year of isolation Origin of isolation

EK-328 Estonia, 1972 imago ticks I. persulcaSofjinKGGb Primorskiy kray, Far-East, Russia, 1937 brain of the patient w205KGGb Khabarovskiy kray, Far-East, Russia, 1973 imago ticks I. persulca80k Sverdlovsk region, Urals, Russia,

before 1970small plaque clone offrom blood of the pat

Absettarovd Leningrad region, European part,Russia, 1951

blood of the patient wAbsettarov 18A clone 18A of strain A256d Belarus, before 1968 imago ticks I. ricinusLesopark Novosibirsk, Siberia, Russia, 1986 imago ticks I. persulcaLK-138 Lithuania, 1972 30 male imago ticks IYuK 4/13 Kemerovo region, Siberia, Russia, 1969 1 imago tick I. persulcDV 936k Primorskiy kray, Far-East, Russia, 1975 imago ticks H. concinPK-36 Primorskiy kray, Far-East, Russia, 1982 1 imago tick I. persulcYa10/89 Yaroslavl region, European part, Russia, 1989 10 imago ticks I. persYa10/89 clone 115 large plaque clone of

strain Ya10/89Ya10/89 clone 125 small plaque clone of

strain Ya10/89

a М—passages in white mice brain; P—passages in PEK cells (Mx—passages made in miceb Sequences of the strains obtained during the present work did not match with previousl

from published previously.c Genotype was determined during the present work.d Sequences of strains 256 and Absettarov obtained during the present work matched w

phenotype–have been identified. It has been shown that GAG-bindingvariants with complex of specific properties do exist among fieldisolates. Second, simulation of E proteins dynamics of 3 viruses hasshown that the phenotypic manifestation of substitution increasingcharge of E protein molecule depends on amino acid context, and notevery such mutation is sufficient for the appearance of the GAG-binding phenotype. Furthermore, we have shown that structuralinflexibility of E protein molecule is highly involved in appearance ofGAG-binding phenotype and complex of relevant properties.

Results

Plaque phenotype of studied TBEV strains

Previously, it was shown that one of the features of GAG-bindingvariants was small plaque phenotype in porcine kidney cell lines(Mandl et al., 2001; Romanova et al., 2007). First we described plaquesize in PEK cells under agar overlay on the 8th day of infection for allcollection strains (Table 1). All results are summarized in Table 2.

Strains Absettarov, 256, LK-138, YuK 4/13, SofjinKGG, 205KGG, DV936k, EK-328, Lesopark, formed large plaques. Strain 80k formedsmall plaques, less than 1mm in diameter. Strain PK-36 formedplaques of middle size, 3–6 mm. The large and middle plaquesappeared on 3–4 days p.i. and reached 7–10 mm in size by the 7–8 days of infection. In contrast, small plaques appeared only on days6–7, and practically did not increase in size during the period ofobservation.

Strain Ya10/89 formed plaques heterogeneous in size. Possibly,this was due to mixing of different viruses from 10 different ticksprepared in one single isolate or true heterogeneity of the isolate. Forfollowing investigations we isolated from plaques clones 115 and 125with plaque sizes of 8 and 1 mm in diameter, respectively.

Strain Absettarov formed plaques from 8 to 10 mm in diameter.From this population was plaque-purified clone 18A. The clone 18Ashowed plaques a bit smaller than parental strain ones: 7.3±0.7 mmvs. 9.5±1 mm (ptb0.0001).

Analysis of TBEV strains and clones sorption on heparin-Sepharose

Small plaque size could be explained by several features, includingvirus inability to reproduce efficiently in the cell culture (low rate or

Passagesa GenBank number Subtype

tus M6P1M2P1-2 DQ486861 Siberianith acute TBE; cloned Mx and P2-5 after cloning GU121963 Far-Easterntus МN20P1 GU121964 Far-Easternstrain 80, isolatedient with acute TBE

Mx and М5P1 after cloning GU121965 Far-Easternc

ith TBE МN20P1 AF091005 Europeanbsettarov P4 after cloning –

МN20P1 AF091014 Europeantus МxM4P1-2 GU121966 Siberianc

. ricinus M2P1-2 GU125720 Europeanc

atus M4P1-2 GU125721 Siberianc

a M2P1-2 GU125722 Far-Easternc

atus M2-6P1 GU121967 Far-Easternc

ulcatus P1-4 – –

heterogeneous P1-2 – Siberianc

heterogeneous P1-2 GU125719 Siberianc

brain by the strain authors before they were received in our laboratory).y GenBank published sequences; we added abbreviation KGG to distinguish our strains

ith GenBank published sequences.

Table 2Biological, immunochemical and virological properties of TBEV strains.

TBEVstrain/clone

Plaque size(mm)

Virions bound to HSa Cathodeprecipitatein RIEc

Amino acidsubstitutionin E proteindNumber of

experimentsSorptionb

Absettarov 9.5±1.0 2 − + −Absettarov 18A 7.3±0.7 4 − + Asp67→Gly256 8.3±0.9 3 − + −LK-138 9.2±0.8 1 − + −EK-328 7.0±0.5 3 − + −Lesopark 8.0±1.5 3 − + −YuK 4/13 11.3±1.8 1 − + −Ya10/89 1.0:2.5

(45:1e)nd nd + nd

Ya10/89 c.115 8.0±0.5 1 − + −Ya10/89 c.125 1.0±0.5 1 + − Glu122→GlySofjinKGG 8.0±0.9 2 − + −205KGG 9.0±1.5 4 − + −80k 1.0±0.5 4 + − Asp67→Asn,

Thr68→AlaDV 936k 10.0±0.5 1 − + −PK-36 4.5±2.0 1 − + −a HS—heparin-Sepharose.b ‘+’—sorption on HS is over 90%, ‘−’—sorption on HS is less than 70%.c RIE—rocket immunoelectrophoresis; ‘+’—presence of cathode-pointed precipitate

of virions with antibodies, ‘−’—absence of cathode-pointed precipitate of virions withantibodies.d Mutation was marked in comparison with relevant genotype consensus, with

parental strain Absettarov for Absettarov 18A, with Ya10/89 clone 115 for Ya10/89clone 125, in comparison with Far-Eastern consensus for strain 80k; ‘—’ is used to markthe absence of such mutation.e The ratio of small plaques (1mm) to bigger ones (2.5mm).

Fig. 1. TBEV strains sorption on heparin-Sepharose. The percentage of sorbed virus wasdeterminedasHS-sorbedvirus to control S-unsorbedvirus ratioevaluatedbyplaqueassay.

264 L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

level of reproduction), virus susceptibility to cell produced IFN,increased virion sorption on the cellular negatively charged proteo-glycans and/or agar sulphopolysaccharide heads, etc. The latter couldbe described like newly formed virions binding to sulphogroups of theneighboring cell GAGs and/or agar that prevent virus diffusion farfrom the primary infected cell.

To evaluate the strains' affinity to heparan sulphate GAGs we usedcommon model of the process—binding to heparin-conjugatedSepharose beads. Virus was incubated with HS and S beads and theamount of unsorbed virus was determined by plaque assay, then theratio HS-sorbed to control S-unsorbed virus was calculated.

Both small plaque forming TBEV variants (80k and Ya10/89 clone125) showed significantly increased affinity to HS (over 90% of sorbedvirions) (Table 2 and Fig. 1) in comparison to large plaque strains (0–25%). Thus, all small plaque virus variants had GAG-binding phenotype.Strain PK-36 with middle plaque size had middle HS sorption activity(about 60%).

Fig. 2. RIE for several TBEV strains: (tracks) 1—strain SofjinKGG; 2—strain Lesopark; 3—strain EK-328; 4—Ya 10/89 clone 125; 5—Ya 10/89 clone 115; 6—strain 205KGG; 7—strain 80k; 8—strain Absettarov; 9—strain 256. Arrows pointed on the rockets of anode-pointed non-virion antigen (oligomers of secreted NS1 protein) and cathode-pointedvirion antigen.

Rocket immunoelectrophoresis (RIE) behavior of studied TBEV strainsand clones

Previously, TBEV virions behavior in RIE was described byLiapustin et al. (1987), Dzhivanian et al. (1991) and Romanova et al.(2007). In RIE buffer system TBEV virions have weak negative charge,and should remain still or slowly drift towards anode in electric field.Upon high endosmosis in agarose virions move towards the cathodeand form cathode-pointed rocket of precipitate with antibodies—virion antigen. The oligomeric forms of secreted NS1 protein in culturefluid have strong negative charge and small size in comparison withthe virion. So, in spite of endosmosis, NS1 forms an anode-pointedrocket—non-virion “soluble” antigen (Fig. 2).

For RIE experiments viruses were normalized by titer (6–7 log10PFU) in infected cell culture fluid. RIE results are summarized in Table2 and Fig. 2. Virions of all large and middle plaque strains formed

cathode-pointed precipitate with antibodies in RIE. Virions of thesmall plaque strains did not produce such precipitate.

Increased affinity to agarose sulphopolysaccharide heads couldexplain the inability of virions of GAG-binding variants to formcathode-pointed rocket in RIE. Precipitation of virions by antibodiesrequires dense arrangement of virions at a distance comparable withthe size of the immunoglobulin molecule, which could be achieved inthe case of dense motion of virions towards the cathode in electricfield. An increase of the charge of virions can promote their binding toagarose sulphogroups in the gel. In this case, virions would move notas a dense front, but as a diffuse smear, which would prevent asufficient concentration of virions in any part of the gel, andsubsequently avert formation of immunoprecipitate.

Hemagglutinating activity (HA) of studied TBEV strains and clones

Previously, described GAG-binding variants did not providenoticeable HA in pH 6.4 or the optimal pH was moved to pH 6.6

Table 4AVirulence of selected TBEV strains in BALB/c mice.

TBEV strain/clone GAG phenotypea Virulence LD50 (log10PFU) after i.p.b

Absettarov − 0.4Absettarov 18A − 2.0256 − −2.0EK-328 − 1.0Lesopark − 1.4Yar10 c.115 − 2.0Yar10 c.125 + 2.9Sofjin KGG − 1.680k + 3.8

a ‘−’—virus without GAG-binding phenotype, ‘+’—virus with GAG-bindingphenotype.b i.p.—upon intraperitoneal inoculation.

Table 4BVirulence of selected TBEV strains in outbred mice.

TBEV strain/clone GAG phenotypea Virulence LD50 (log10PFU)

after i.c.b after i.p.c

Absettarov − −2.7 −1.4Absettarov 18A − 1 2.2EK-328 − 0.7 ndc

Sofjin KGG − −0.8 1.2205KGG − nd 1.680k + 0.4 2.8

a ‘−’—virus without GAG-binding phenotype, ‘+’—virus with GAG-bindingphenotype.b i.c.—upon intracerebral inoculation, i.p.—upon intraperitoneal inoculation.c nd—not done.

265L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

(Goto et al., 2003; Romanova et al., 2007). We carried outhemagglutination of goose erythrocytes with 2-fold dilutions ofinfected cells supernatant aliquots in range pH 5.7–7.0 (pH 5.7, 6.0,6.2, 6.4, 6.6, 6.8, 7.0). Results are summarized in Table 3.

Studied strains differed in pH range with noticeable HA. Thus,strain EK-328 agglutinated erythrocytes at pH 5.7–7.0 and strain205KGG showed HA only in pH 6.2–6.4. Nevertheless, all large andmiddle plaque viruses, excluding clone Absettarov 18A, were able tohemagglutinate at optimal pH 6.4. Clone 18A showed sufficiently highHA, but the optimal pH was moved from 6.4 to acid values (pH 6.0).GAG-binding variants differed from other strains in HA absence at pH6.4. Strain 80k did not provide any noticeable HA in all pH values, andclone 125 of Ya10/89 exhibited light HA at pH 7.0.

Virulence of TBEV strains and clones in mice

Various authors showed that GAG-binding variants of flaviviruseshad low neuroinvasiveness (Goto et al., 2003; Lee and Lobigs, 2002;Lee et al., 2004; Mandl et al., 2001; Romanova et al., 2007).

To evaluate the neuroinvasiveness BALB/c mice were chosen assusceptible to TBEV infection (Pletnev et al., 2000). To estimate thereliability of the neuroinvasiveness test differences for distinct viruseswe had to know the variability of results from experiment toexperiment. For this purpose we chose 2 of investigated viruses—strain Absettarov and strain EK-328 for detailed study and processedobtained results statistically. For strain Absettarov series of 16independent experiments was provided. LD50 after i.p. inoculationwas 0.4±0.2 log10PFU (SD=1.0 log10PFU). For strain EK-328 wereprovided 3 independent experiments. Neuroinvasiveness was 1.0±0.5 log10PFU (SD=0.8 log10PFU). We decided to use 0.5 log10PFU asstandard error for all other obtained results.

Neuroinvasiveness on BALB/c mice differed a lot among studiedTBEV strains from−2 log10PFU for strain 256 to 3.8 log10PFU for strain80k (Table 4A). Neuroinvasiveness of both small plaque GAG-bindingvariants–80k and Ya10/89 clone 125–was significantly lower incomparison with other strains. The difference was the most visible incomparison of viruses of one subtype, like 80k vs. SofjinKGG; Ya10/89clone 125 vs. clone 115, strains EK-328 and Lesopark. The neuroinva-siveness of clone Absettarov 18A was lower than the parental strainone (Absettarov 18A vs. Absettarov).

Low neuroinvasiveness could be due to inability of virus to invadeCNS and/or to kill CNS cells. To elucidate this we obtained i.p. and i.c.values for several strains on outbred mice less susceptible for TBEVthan BALB/c mice (Table 4B). Strains significantly differed inneurovirulence from −2.7 log10 PFU for strain Absettarov to 0.7log10 PFU for strain EK-328. Neuroinvasiveness results in outbred

Table 3Hemagglutinating activity of investigated TBEV strains.

Virus Titer(lgPFU/ml)

Allowed ofpH range

OptimalpH

Titer HA inoptimal pH (log2)

Absettarov 6.7 5.7–6.4 6.4 6Absettarov 18A 7.2 5.7–6.4 6.0 5256 5.2 5.7–6.8 6.4 8LK-138 7.1 5.7–6.4 6.4 4EK-328 7.8 5.7–7.0 6.4 9Ya10/89a 6.9 ndb nd 5Ya10/89 c.115 7.4 6.4–7.0 6.4 3Ya10/89 c.125 7.5 7.0 7.0 1SofjinKGG 6.3 6.2–6.6 6.4 10205KGG 7.7 6.2–6.4 6.2–6.4 380k 6.9 – – –c

DV 936k 8.1 6.2–7.0 6.4–6.8 3PK-36 6.0 5.7–6.6 6.4 8

Titer of HA was determined in culture fluid of infected PEK cells.a HA was determined only in pH6.4.b nd—not done.c Absence of HA in minimal dilution at all pH values.

mice corresponded to the ones in the BALB/c mice. Strain 80k andclone Absettarov 18A were the least virulent for outbred mice uponperipheral inoculation than other investigated strains. Clone 18Avirulence upon i.c. and i.p. inoculation was 5000-fold and 4000-foldlower than parental strain ones, respectively. Thus, clone 18Aneurovirulence and neuroinvasiveness both decreased on the samelevel in comparison to the parental strain. Strain 80k neuroinvasive-ness was 40-fold lower than the one of strain SofjinKGG, andneurovirulence was only 16-fold lower than the one of SofjinKGG.

Sequencing of E protein genes of TBEV strains and clones

We obtained nucleotide sequences of part of viral genomeencoding E glycoprotein (or sE part of the glycoprotein) of collectionTBEV variants. Using these sequences we analyzed amino acidsubstitutions leading to increase of the surface charge of the E proteinmolecule in comparison with consensus sequence of the relevantgenotype.

Investigated TBEV strains represented all three subtypes: Europe-an (Absettarov, 256, LK138), Siberian (EK-328, Lesopark, YuK4/13,clones of Ya10/89), and Far-Eastern (SofjinKGG, 80k, 205KGG,DV936k, PK-36).

Both GAG-binding viruses showed amino acid substitutions increas-ing surface charge of the E protein molecule (Table 2). Strain 80k hadAsp67→Asn, in comparison to the consensus of Far-Eastern genotype.Besides, strain 80k carried additional mutation in Thr68→Ala. Smallplaque clone 125 of strain Ya10/89 carried Glu122→Gly in comparisonwith clone 115 and Siberian genotype consensus. In addition,sequencing revealed 2 nt synonymic substitutions in 1600 nt betweenclones of strain Ya10/98. Moreover, we found out the clone Absettarov18A surprisingly carried Asp67→Gly mutation, similar to strain 80k, incomparison to the parental strain.

All other TBEV variants didn't show any charge increasingsubstitutions in comparison to the genotype's consensus sequences.

266 L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

Homology modeling and molecular dynamics

Finally, we decided to elucidate why clone 18A of strain Absettarovwith substitution increasing charge of the E protein molecule hadn’tchanged its GAG affinity. To solve the problem we chose for furtherinvestigations viruses Absettarov 18A, carrying charge-increasingsubstitution in 67 amino acid position of E protein that didn’t affectthe virus properties, strain 80k, as GAG-binding variant carryinganalog substitution in the same position, and strain SofjinKGG, as astrain of Far-Eastern subtype.

We built models of soluble domain of E protein molecule forAbsettarov 18A, SofjinKGG and 80k on the basis of strain Neudorfl (PDBaccess code 1SVB). The difference between the template and targetproteins was subtle (Fig. 3); no insertions or deletions occurred in thealignment and only local substitutions were present. 12 substitutionsdiffered in Western and Far-Eastern genotype strains (47, 88, 115, 120,178, 206, 260, 267, 277, 317, 331, 363 residues). Strain Absettarovdiffered from strain Neudorfl in only 1 amino acid in position 167 (Ile toVal), and clone 18A differed from parental strain in position 67 (meantabove). 2 amino acids in addition to 2mutationsmarked above differedstrain 80k from strain SofjinKGG (Thr4→ Ile, Ala153→Val). Although,the main difference between strains occurred around position 67.Strains Neudorfl and SofjinKGG contained Asp67, clone 18A carriedGly67, and strain 80k had Asn67, and additionally Thr68→Ala.

The charge distribution for themean structure of modeled proteinsis shown in Fig. 4A. One could easily see that the surface chargedistribution for the protein E of all strains was very similar. It wasobvious for the strains 80k and Absettarov 18A due to the character ofthe amino acid substitutions, but for SofjinKGG such observation wasless expected. Namely, net charge of the protein from strainsAbsettarov 18A and 80k equaled −2, whereas for the protein fromstrain SofjinKGG it was equal to −4. The negative charge of the latterprotein was located mainly at the membrane side of the protein (datanot shown). Although the ridge formed by amino acid side chainsexposed on the external surface across 2 subunits of the E proteindimer slightly differed on the surface charge distribution map. Theseamino acids could be possibly involved into electrostatic interactionswith molecules on cellular membrane, like GAGs.

Nevertheless, static representation of the proteins obtained bymodeling was not sufficient to explain the differences in properties ofAbsettarov 18A and strain 80k.

We collected 10 ns trajectories for TBEV variants under consider-ation—Absettarov 18A, SofjinKGG and 80k. Principal componentanalysis revealed two main components of the protein movementduring molecular dynamics simulation: (1) twist around the longestdimer axis I, and (2) bending around the axis II (Fig. 4B). For all casesthe movement of the protein could be characterized by its bending.According to this observation, we had chosen the bending angle (b.a.)instead of root-mean-square distance (RMSD) as the main variabledescribing the conformation of the protein during dynamics. We hadalso built correlation maps with the aim to understand the possiblecorrelated motions of the protein parts.

The simulation started from the planar conformation (b.a. ca.180°) corresponded to the crystal structure of the template, and thenduring the first nanosecond bending was initiated by the thermalmotion of the molecule. The bending angle dependence on time ispresented in Fig. 4C. There were no significant differences betweenthe strains which could illuminate the distinction between the HS-binding properties of three proteins.

We had built correlation maps (Fig. 4D) with the aim to describecorrelatedmotions of the proteins. Thesemaps were rather similar forall the proteins, but notable differences did exist. The domainsshowed strongly correlated movements inside the domain andnegative correlation between domain II (residues 52–136 and 190–284) and other domains. The movement of domain I (residues 1–51,137–189 and 285–302) possessed moderate correlation with the

movement of domain III (residues 303–395). For clone Absettarov18A and strain SofjinKGG a stronger correlation was observed insidethe domain II compared to the strain 80k. The movement of domain Iwas more distorted for strain 80k, with regions of correlation andanti-correlation being more noticeable than for two other strains,especially in the region of residues 137–189.

Summarizing our simulation studies, we could say that thedynamic behavior of the strain 80k E protein differed from that ofthe clone Absettarov 18A and strain SofjinKGG. The difference wasrather small due to very high identity of amino acid sequences of theseproteins, but it was marked enough to be a working hypothesis forexplanation of the experimental results.

Discussion

Interest for GAG-binding variants derives from several facts: (1)one amino acid substitution corresponds to complex of peculiaritiestypical for GAG-binding phenotype; (2) there could be several sites onthe one protein responsible for GAG-binding phenotype appearance,and position of themutation could influence the intensity of display ofobtained variant properties; (3) change of affinity to GAGs couldmodulate virus cell tropism; (4) low neuroinvasiveness; (5) highpotential for emergence of genetic variability for virulent variants, asset of various mutations compensating increased charge of E proteinand restoring virions low affinity to cellular GAGs and, thus, increasingvirulence, could appear. The existence of such variants for wide rangeof viruses indicates that different viruses could use the samemechanism to change population properties.

Previously, laboratory obtained GAG-binding variants were de-scribed for various flaviviruses, including TBEV (Chiou and Chen,2007; Goto et al., 2003; Lee and Lobigs, 2002; Lee et al., 2004; Mandlet al., 2001; Romanova et al., 2007). During the present study wemonitored 13 collection TBEV strains and found two containing GAG-binding variants. Therefore, we have provided evidence that GAG-binding variants do exist in natural TBEV population using the virus(clone 125 of Ya10/89) with short laboratory passage history andstrain (80k), and could be isolated from different origins, like ticksuspension or blood of patient with acute TBE. In the present work onnatural strains we have supported our data obtained earlier forlaboratory virus variants, that GAG-binding phenotype, i.e. increasedsorption on HS, corresponded to such virus properties, like smallplaques in PEK cells, absence of HA activity at pH 6.4 and cathode-pointed precipitate of virions with antibodies in RIE, low neuroinva-siveness in mice (Romanova et al., 2007). In the case the propertieslike HS-sorption, HA and precipitate forming in the RIE aredetermined only by E protein, and the properties like plaque size orneuroinvasiveness could be due to involvement of some other viralproteins activities. Previously, by methods of reverse genetics wasshown that mutation Glu122-Gly, the same as in Ya10/89 clone 125,and Asp67-Gly, analogous to the one in strain 80k, defined the GAG-binding phenotype (Khasnatinov et al., 2009; Mandl et al., 2001).Summarizing all the data we can consider that the complex ofproperties indicates the GAG-binding variants.

Previously, the set of various mutations increasing virion chargehave been described TBEV during virus propagation in cell culture(Goto et al., 2004; Mandl et al., 2001) or in ticks (Labuda et al., 1994;Romanova et al., 2007). Besides, it has been shown that TBEV can existas stable heterogeneous population containing variants with differentaffinity to GAGs (Romanova et al., 2007). Both GAG-binding variants(strain 80k and Ya10/89 clone 125) described in the present studywere cloned from primary isolate to investigate exactly the small-plaque variants: original strain 80 was described by the author aslarge plaque and parental strain Ya10/89 exhibited heterogeneousplaque phenotype. According to the data presented here andpublished previously of the existence of TBEV strains with E proteincharge increasing mutations (Ecker et al., 1999; Khasnatinov et al.,

Fig. 3. Alignment of E protein ectodomain of studied viruses (strain Absettarov, clone Absettarov 18A, strain SofjinKGG, strain 80k) and strain Neudorfl, used as a template for modeling structures.

267L.I.K

ozlovskayaet

al./Virology

398(2010)

262–272

Fig. 4. (A) Surfaces of the E protein models colored according to electrostatic potential. S—strain SofjinKGG, 80—strain 80k, A—clone Absettarov 18A. The color ramp of electrostaticpotential is in the left; red corresponds to the most positively charged regions, blue—to the most negatively charged. White square encloses residue 67. (B) Overview of the model ofthe E protein for the clone Absettarov 18A. Domain I is red, domain II is yellow, domain III is blue. Axis I is the axis of the twist motion, axis II corresponds to bendingmotion. ResiduesLys64, Ser66, Gly67, Thr68, Lys69, Glu84, Thr90, Lys118 and Glu122 are shown. (C) Plot of bending angle (degrees) dependence on time (ps) for E proteins of 80k (orange),Absettarov 18A (green), and SofjinKGG (purple). (D) MD simulation correlation maps. On the X and Y axes amino acid residues of monomers of E protein dimer are situatedconsequently. The dot on the map symbolizes the motion of X axis residue relative to Y axis residue. The highest correlation corresponds to yellow regions, the lowest—to blue.

268 L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

2009; Lu et al., 2008) or with complex of specific properties describedabove (Chunikhin et al., 1986; Pogodina et al., 1992) we can assumethat GAG-binding variants exist like an admixture in natural TBEVpopulation that is hardly seen.

They showed that GAG-binding variants could appear during TBEVadaptation to ticks: European strain to Ixodes ricinus ticks (Labuda et al.,1994), or Siberian strain to Hyalomma marginatum marginatum ticks(Romanovaet al., 2007), or several strains (including Far-Eastern strain)

269L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

toHyalomma anatolicum ticks (Chunikhin et al., 1986). According to thedata of reproduction rate of GAG-binding variants (Romanova et al.,2007) or of genetically constructed variants with charge increasingmutation in E protein (Khasnatinov et al., 2009) in tickswe can considerthat such variants have selective advantage during reproduction in tickorganism, apparently, regardless of tick species.

Several cell lines, like BHK-21, SW-13, Neuro-2, used for virusisolation, were predisposed to select virus variants with increasedaffinity to cellular GAGs (Lee and Lobigs, 2002; Lee et al., 2004; Mandlet al., 2001). Thus, GAG-binding variants appearing even after short-term passages in these cultures could be novel adaptive or preexist-ing. However, mutants with decreased affinity to cellular GAGs couldappear in population of GAG-binding virus during infection inmammalian host or in PEK cells (Chunikhin et al., 1986; Kaluzová etal., 1994; Liapustin et al., 1987; Romanova et al., 2007). Only onesubstitution has been sufficient to switch the phenotypic character-istics of GAG-binding variants (Romanova et al., 2007). So, passages inlaboratory animals or some cells could lead to practically totalelimination of GAG-binding variants from natural isolate. In thepresent work both GAG-binding variants were propagated in PEK cellsthat allowed us to assume that they preexisted in natural isolatebecause it was shown previously the cell line didn’t lead toappearance of adaptive mutations increasing affinity to GAGs(Romanova et al., 2007). Strain 80k didn’t change through long-term laboratory passaging and never produced plaques of heteroge-neous phenotype (data not shown). Probably, this could be explainedby specific composition of two neighboring mutations (Asp67→Asn,Thr68→Ala) in the E protein that make specific efficient conformationof the molecule, and mutations would be crucial for virus viability.

Amino acid mutations in domain II of E protein could affect on HAand fusion activities changing low-pH-triggered dimer–trimer con-formational switch (Allison et al., 1995; Heinz and Allison, 2000). Inthe present study both GAG-binding variants, carrying substitution inthe domain II of E protein, lost their hemagglutinating activity. Alllarge and middle plaque viruses, excluding clone Absettarov 18A,hemagglutinate erythrocytes at optimal pH 6.2–6.4. Gaining ofmutation in E protein domain II of viruses meant above led them tonarrow the range of allowed pH for HA and drastically moved theoptimal pH. Strain 80k absolutely lost ability to HA, Ya10/89 clone 125was able to agglutinate erythrocytes in minimal dilution at pH 7.0. Incontrast to strain 80k, Absettarov 18Awith analogmutation sustainedHA property, although optimal pH was a bit shifted. Lately the strainYar 46-2 with the same mutation as the one of clone 18A wasdescribed and the strain had no ability to HA (Khasnatinov et al.,2009). It is obvious that HA decreasing or absence associates exactlywith GAG-binding phenotype, rather than with just presence ofmutation increasing charge or hydrophobicity of E protein.

In our work both GAG-binding variants Ya10/89 clone 125 andstrain 80k showed lowed neuroinvasiveness. Previously, we describedthe tick-adapted variant M, obtained from strain EK-328, withincreased sorption on HS (N90%) and low neuroinvasiveness(LD50=5.4 log10 PFU). Ya10/89 clone 125 with the same mutationlike variant M had LD50=2.9 log10 PFU, lower in comparison withlarge plaque clone 115, but significantly higher than variant M's one.On the one hand, this indicates that such substitution is important forTBEV virulence. On the other hand, increased sorption on cellularGAGs decreases neuroinvasiveness, nevertheless, since neuroinva-siveness is a multi-determinative process, the virulence could due toinvolvement of other reactions in vivo determined by other viralproperties. In case of clone 18A with low affinity to HS andsubstitution in position 67 showed drastically decreased neuroviru-lence and neuroinvasiveness in comparison with parental strain, butstill higher than the ones of the GAG-binding variants. It could beexplained that the mutation in Absettarov 18A E protein does notprovide enough impact on the molecule structure to switch the HSaffinity. It could be interpreted that not every E protein charge

increasing substitution is sufficient for HS-affinity switch but isenough for decreasing virulence in mice through influence onsorption on other receptors or other GAG types, for example,chondroitin sulfate on the CNS cells. Besides, the Asn67 in the Eprotein molecule of strain 80k is typical for hemorrhagic flaviviruses(Barker et al., 2009). Therefore, lowed neuroinvasiveness of the viruscould be due to switch of susceptible cell type preference.

The present work provides evidence that not every mutationincreasing net charge of the E protein molecule leads to appearance ofthe new virus with GAG-binding phenotype and all complex ofspecific properties. The analysis of electrostatic potential distributionon the surfaces of the E proteins of Absettarov 18A, 80k and Sofjinrevealed only slight differences. According to Mandl and co-authors(2001) most of the mutations increasing affinity to GAGs lay on thesurface of domains I and II of E protein. Similarly, there is a visibleringe formed by side chains of amino acids of domains I and II of 2connected subunits in dimer on electrostatic potential maps, obtainedduring homology modeling. In case of SofjinKGG this ringe includemore negatively charged sites than in case of Absettarov 18A or 80k.Although, surface charge distribution difference does not explaindistinct properties of clone Absettarov 18A and strain 80k.

Correlation maps based on MD trajectories showed strikingdifferences between E proteins of these 3 viruses. The comparisonof SofjinKGG and Absettarov 18A maps in common shows thegenotype differences of correlated movement inside E proteinmolecule.

Comparison of the correlation map of the 80k with 2 other mapscould explain the dramatic contrast of its properties. The protein ofstrain 80k seems being divided into small grains, whereas domains ofthe protein of strains Absettarov 18A and SofjinKGG move as a whole.Consequently, the protein of strain 80k is less tight than the one oftwo other strains, and highermobility of small parts of the protein canbe the reason for higher HS binding affinity. Because distinct surfaceresidues of 80k E protein move more independently from each otherthat, probably, allows them to interact freely with HS and cellularmembrane molecules.

Previously, all described variants with GAG-binding mutationwere selected by exhibited phenotype, like plaque size. That couldexplain why variants, like Absettarov 18A, remain unnoticed.Although, such variants with pre-existing mutation that in somecircumstances could lead to formation of novel GAG-binding variantsmay store in virus population for long periods of time and throughvirus transmission lifecycle. As far as Absettarov 18A is one of theplaque-purified clones of strain Absettarov, so we can propose it tolong-term maintenance through laboratory passaging of the wholestrain. Thus, such variants could survive silently, hiding its pre-existing mutation, in the virus population.

Thus, GAG-binding variants (with increased affinity to HS) withcomplex of peculiarities, like small plaque phenotype in PEK cells,absence of HA at pH 6.2–6.4, absence of cathode-pointed precipitatein RIE, low neuroinvasiveness, do exist in natural TBEV population.Even though, not every amino acidmutation increasing surface chargeof E protein molecule leads to appearance of GAG-binding phenotype.According to obtained data we can suppose that the flexibility of Eprotein molecule could influence virion–GAG interactions.

Materials and methods

Cells and viruses

Pig embryo kidney (PEK) cell line was maintained at 37 °C inmedium 199 (PIPVE, Russia) supplemented with 5% bovine serum(Furo, Russia).

TBEV strains used in the work are summarized in Table 1. StrainLesopark and the virus containing tick suspension from where strainYa10/89 was isolated were kindly provided by Dr. N.G. Bochkova and

270 L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

Dr. V.V. Pogodina (Chumakov IPVE RAMS). Cloned strain 80k wasreceived from one of strain author Dr. V. Kolotvinov.

Viruses were stored as aliquots of 10% infected mouse brainsuspensions or infected PEK cell culture fluids at −70 °C.

Plaque assay

The procedure of plaque assay was described previously (Roma-nova et al., 2007). Briefly, PEK cells were seed in 6-well plates(Corning) and incubated for 72 h at 37 °C in CO2-incubator. 10-folddilutions of virus samples were prepared on medium 199. Equalaliquots of viral suspensions were added to each well in 1–3 parallels,and incubated in CO2-incubator for 1 h with gentle shaking. Then,each well was overlaid with 5 ml of 1% agar (Difco) on Earle solutioncontaining 7.5% FBS and 0,015% neutral red. After incubation at 37 °Cin CO2-incubator plaques were counted everyday till day 14 ofinfection. The virus titer was calculated and expressed as the log10plaque forming unit (PFU) per ml.

Virus hemagglutination (HA)

Hemagglutination of the TBEV strains was assayed with gooseerythrocytes according to the standard protocol (Clarke and Casals,1958) in 2 parallels for each sample dilution in pH 5.7–7.0.

Virulence in mice

All experiments were carried out in 10- to 12-week-old BALB/c(Stolbovaya, Moscow region, Russia) or outbred (Andreevka, Moscowregion, Russia) mice. TBEV strains were evaluated for both neuro-virulence by intracerebral (i.c.) inoculation (30 μl) and neuroinva-siveness by intraperitoneal (i.p.) inoculation (300 μl). Mice in groupsof six were injected with 10-fold dilutions of virus and observed forclinical symptoms for 21 days after inoculation. LD50 was calculatedaccording to the Kerber method (Lorenz and Bogel, 1973).

Sequencing

Viral RNA was extracted from 10% brain suspensions or from PEKcell culture supernatants by TRIReagent LS (Sigma) according to themanufacturer's instructions. Reverse transcription was carried outwith M-MLV reverse transcriptase (Promega, Madison, WI) accordingto the manufacturer's protocol. Viral genome cDNA was amplified byPCR using overlapping sets of TBEV specific primers (primersequences available upon request). Sequencing was carried outdirectly from PCR DNA in both directions on the ABI PRISM 3730(Applied Biosystems) sequencer using ABI PRISM® BigDye™ Termi-nator v. 3.1. The sequences were aligned with Clustal-X 2.0.11 (Larkinet al., 2007). Genotype was determined according to the phylogenicanalysis.

Virus binding to heparin-Sepharose

The binding assay of TBEV on heparin-Sepharose beads wasperformed as described earlier (Romanova et al., 2007) with somemodifications. Briefly, 0.5 ml of a 50% heparin-Sepharose CL 6B(Pharmacia) suspension in 199 medium, pH 7.6, was spun down.Virus suspension (0.5 ml containing 4–5 log10 PFU) was added to theheparin-Sepharose (HS) pellet, mixed thoroughly, and incubated for1 h at 37 °C with occasional vigorous shaking. Control viruses weretreated similarly with Sepharose CL 6B (Pharmacia) to take intoaccount the matrix sorption. After the binding step, heparin-Sepharose and Sepharose was pelleted by centrifugation, and thesupernatant was collected and tested for presence of unsorbed virususing plaque assay in PEK cells. Binding efficacy was determined as apercentage of sorbed virus (S-unsorbed virus minus HS-unsorbed

virus) to the control virus titer. The viruses were tested for HS bindingin 1–4 separate experiments with 2 parallels in each experiment. Todecrease the variability of plaque assay results titration was providedin 2–4 parallels for each sample dilution.

Rocket immunoelectrophoresis (RIE)

RIE was carried out in 1% high endosmosis agarose (Serva) inbarbiturate-barbital Svendsen buffer, pH 8.8 (Weeke, 1973), with 1.5%horse hyperimmune TBEV gamma-globulin (VNNI VS, Tomsk, Russia)using an immunoelectrophoresis chamber (LKB). Viral antigens wereprepared by precipitation from infected cell culture supernatants with10% PEG-6000 (Dzhivanian et al., 1991; Liapustin et al., 1987).Samples in equal quantity (6–7 log10 PFU in infected cell culture fluid)of each virus were loaded on the anti-TBEV serum enriched agarosegel, resolved at 7 V/cm for 12 h and stained with Coomassie Blue. RIEexperiments were repeated several times for each strain.

Modeling of the protein structures

Homology modeling was performed only for the soluble domain ofTBEV E proteins; stem-anchor domain was excluded from theconsideration. Amino acid sequence of the template protein wasextracted from the structure file from PDB (access code 1SVB (Rey etal., 1995)). Amino acid sequences of the template (strain Neudorfl)and targets were aligned in ClustalX 2.0.11 (Larkin et al., 2007) withdefault parameters.

Homology modeling was performed by means of the programMODELLER 9v5 (Šali and Blundell, 1993). Atomic coordinates of thedimeric template were recreated from crystallographic data withCCP4mg (Potterton et al., 2002). Thirty-five different models of thedimeric soluble domains were built for viruses Absettarov 18A,SofjinKGG, and 80k, and the simulated annealing procedure wasapplied to every model. The N-acetyl-D-glucosamine moiety attachedto Asn154 was kept intact. The best models based on the MODELLERDOPE score and PROCHECK (Laskowski et al., 1993) validation scorewere chosen for subsequent optimization.

Further optimization was performed for the models in SYBYL 8.0(Tripos Inc., 1699 South Hanley Rd., St. Louis, Missouri, 63144, USA).All hydrogen atoms were added, and then 150 steps of the Powellminimization in the Tripos force field were performed. Operations ofthemolecules fitting and comparisonwere performedwith SYBYL 8.0.The final structures of the models were used for molecular dynamicsstudies.

Molecular dynamics (MD) simulation

Molecular dynamics simulation was performed by means of theprogram complex AMBER 10 (Case et al., 2008). For the proteinmolecule force field AMBER ff99SB (Hornak et al., 2006) was used andGLYCAM06 (Kirschner et al., 2008) was used for the carbohydratemoiety. Atomic charges were automatically assigned in LeAP moduleof the AMBER suite. The SHAKE algorithm (Ryckaert et al., 1977) wasapplied to the lengths of the bonds containing hydrogen atoms toreduce computational time. Generalized Born solventmodel (Tsui andCase, 2001) was utilized to mimic solvent effects.

The energy of system was minimized prior to the productiondynamics run with 1500 iterations of steepest descent method and1000 iterations of conjugated gradients method. Production MD runwas performed with sander on 256 processors of the supercomputerSKIFMSU (Moscow State University Research Computing Center). Thefollowing parameters were used: time step–2 fs (femtoseconds),number of iterations–5million, system temperature was kept equal to300 K by Langevin thermostat with collision frequency of 1 ps−1

(picoseconds).

271L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

Visual analysis of trajectories was performed with VMD (Hum-phrey et al., 1996); statistical analysis wasmade in ptraj module of theAMBER suite. The principal component analysis of the trajectories wasalso independently done with DYNAMITE web server (Barrett et al.,2004).

Acknowledgments

The authors gratefully thank Moscow State University ResearchComputing Center for computational time. The free academic licensefor AMBER 10 suite was kindly provided by David A. Case.

The authors thank Dr. V.V. Pogodina and Dr. N.G. Bochkova forallocation of TBEV strains Ya10/89 and Lesopark; Dr. WashingtonCuña for the critical reading of the manuscript; R.S. Blinova, R.V.Daryushina, and V.D. Mitrofanova for technical assistance.

The present work was supported by grant of Russian Foundation ofBasic Research 08-04-01718-а.

References

Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., Walter, P., 2008. Molecular biology ofthe cell, 5th ed. Garland Science, Taylor & Francis Group, USA.

Allison, S.L., Schalich, J., Stiasny, K., Mandl, C.W., Kunz, C., Heinz, F.X., 1995. Oligomericrearrangement of tick-borne encephalitis virus envelope proteins induced by anacidic pH. J. Virol. 69 (2), 695–700.

Barker, W.C., Mazumder, R., Vasudevan, S., Sagripanti, J.L., Wu, C.H., 2009. Sequencesignatures in envelope protein may determine whether flaviviruses producehemorrhagic or encephalitic syndromes. Virus Genes 39 (1), 1–9.

Barrett, C.P., Hall, B.A, Noble, M.E.M., 2004. Dynamite: a simple way to gain insight intoprotein motions. Acta Cryst. 60 (Pt. 12 Pt. 1), 2280–2287.

Bernard, K.A., Klimstra, W.B., Johnston, R.E., 2000. Mutations in the E2 glycoprotein ofVenezuelan equine encephalitis virus confer heparan sulfate interaction, lowmorbidity, and rapid clearance from blood of mice. Virology 276 (1), 93–103.

Byrnes, A.P., Griffin, D.E., 1998. Binding of Sindbis virus to cell surface heparan sulfate. J.Virol. 72 (9), 7349–7356.

Case, D.A., Darden, T.A., Cheatham, T.E., Simmerling, C.L., Wang, J., Duke, R.E., Luo, R.,Crowley, M., Walker, R.C., Zhang, W., Merz, K.M., Wang, B., Hayik, S., Roitberg, A.,Seabra, G., Kolossváry, I., Wong, K.F., Paesani, F., Vanicek, J., Wu, X., Brozell, S.R.,Steinbrecher, T., Gohlke, H., Yang, L., Tan, C., Mongan, J., Hornak, V., Cui, G.,Mathews, D.H., Seetin, M.G., Sagui, C., Babin, V., Kollman, P.A., 2008. AMBER 10.University of California, San Francisco.

Chiou, S.-S., Chen, W.-J., 2007. Phenotypic changes in the Japanese encephalitis virusafter one passage in Neuro-2a cells: Generation of attenuated strains of the virus.Vaccine 26 (1), 15–23.

Chu, J.J.H., Ng, M.L., 2004a. Infectious entry of West Nile Virus occurs through clathrin-mediated endocytic pathway. J. Virol. 78 (19), 10543–10555.

Chu, J.J.H., Ng, M.L., 2004b. Interaction of West Nile Virus with αvβ3 integrin mediatesvirus entry into cells. J. Biol. Chem. 279 (52), 54533–54541.

Chunikhin, S.P., Reshetnikov, I.N., Liapustin, V.N., 1986. Variability of tick-borneencephalitis virus during passage through Ixodes ticks and small mammals. Med.Parazitol. (Mosk) (in Russian) (6), 58–61.

Clarke, D.K., Casals, J., 1958. Techniques for hemagglutination and hemagglutina-tion-inhibition with arthropod-borne viruses. Am. J. Trop. Med. Hyg. 7 (5),561–573.

Dzhivanian, T.I., Karganova, G.G., Sobolev, S.G., Korolev, M.B., Kashtanova, G.M.,Chuprinskaia, M.V., Lashkevich, V.A., 1991. The properties of the particles formedin the reproduction of an acute in vitro infection with the tick-borne encephalitisvirus adapted to H. plumbeum ticks. Vopr. Virusol. (in Russian) 36 (4), 297–300.

Ecker, M., Allison, S.L., Meixner, T., Heinz, F.X., 1999. Sequence analysis and geneticclassification of tick-borne encephalitis viruses from Europe and Asia. J. Gen. Virol.80 (Pt. 1), 179–185.

Gallagher, J.T., Turnbull, J.E., Lyon, M., 1992. Patterns of sulphation in heparan sulphate:polymorphism based on a common structural theme. Int. J. Biochem. 24 (4),553–560.

Goto, A., Hayasaka, D., Yoshii, K., Mizutani, T., Kariwa, H., Takashima, I., 2003. A BHK-21cell culture-adapted tick-borne encephalitis virus mutant is attenuated forneuroinvasiveness. Vaccine 21 (25-26), 4043–4051.

Haywood, A.M., 1994. Virus receptors: binding, adhesion strengthening, and changes inviral structure. J. Virol. 68 (1), 1–5.

Heil, M.L., Albee, A., Strauss, J.H., Kuhn, R.J., 2001. An amino acid substitution in thecoding region of the E2 glycoprotein adapts Ross River virus to utilize heparansulfate as an attachment moiety. J. Virol. 75 (14), 6303–6309.

Heinz, F.X., Allison, S.L., 2000. Structures and mechanisms in flavivirus fusion. Adv.Virus Res. 55, 231–269.

Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., Simmerling, C., 2006.Comparison of multiple amber force fields and development of improved proteinbackbone parameters. PROTEINS: Structure, Function, and Bioinformatics 65 (3),712–725.

Hulst, M.M., van Gennip, H.G., Moormann, R.J., 2000. Passage of classical swine fevervirus in cultured swine kidney cells selects virus variants that bind to heparan

sulfate due to a single amino acid change in envelope protein E(rns). J. Virol. 74(20), 9553–9561.

Humphrey, W., Dalke, A., Schulten, K., 1996. VMD: visual molecular dynamics. J. Molec.Graphics 14 (4), 33–38.

Kaluzová, M., Eleèková, E., Žuffová, E., Pastorek, J., Kaluz, Š., Kouch, O., Labuda, M., 1994.Reverted virulence of attenuated tick-borne encephalitis virus mutant is notaccompanied with the changes in deduced viral envelope protein amino acidsequence. Acta Virol. 38 (3), 133–140.

Khan, A.G., Pichler, J., Rosemann, A., Blaas, D., 2007. Human rhinovirus type 54 infectionvia heparan sulfate is less efficient and strictly dependent on low endosomal pH. J.Virol. 81 (9), 4625–4632.

Khasnatinov, M.A., Ustanikova, K., Frolova, T.V., Pogodina, V.V., Bochkova, N.G., Levina,L.S., Slovak, M., Kazimirova, M., Labuda, M., Klempa, B., Eleckova, E., Gould, E.A.,Gritsun, T.S., 2009. Non-hemagglutinating flaviviruses: molecular mechanisms forthe emergence of new strains via adaptation to European ticks. PLoS One 4 (10),e7295.

Kim, C.W., Goldberger, O.A., Gallo, R.L., Bernfield, M., 1994. Members of the syndecanfamily of heparan sulfate proteoglycans are expressed in distinct cell-, tissue-, anddevelopment-specific patterns. Mol. Biol. Cell. 5 (7), 797–805.

Kirschner, K.N., Yongye, A.B., Tschampel, S.M., Conzález-Outeriño, J., Daniels, C.R., Foley,B.L., Woods, R.J., 2008. GLYCAM06: a generalizable biomolecular force field.Carbohydrates. J. Comp. Chem. 29 (4), 622–655.

Klimstra, W.B., Ryman, K.D., Johnston, R.E., 1998. Adaptation of Sindbis virus to BHKcells selects for use of heparan sulfate as an attachment receptor. J. Virol. 72 (9),7357–7366.

Kroschewski, H., Allison, S.L., Heinz, F.X., Mandl, C.W., 2003. Role of heparan sulfate forattachment of tick-borne encephalitis virus. Virology 308 (1), 92–100.

Labuda, M., Jiang, W.R., Kaluzova, M., Kozuch, O., Nuttall, P.A., Weismann, P., Eleckova,E., Zuffova, E., Gould, E.A., 1994. Change in phenotype of tick-borne encephalitisvirus following passage in Ixodes ricinus ticks and associated amino acidsubstitution in envelope protein. Virus Res. 31 (3), 305–315.

Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam, H.,Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J., Higgins,D.G., 2007. Clustal W and Clustal X version 2.0. Bioinformatics 23 (21), 2947–2948.

Laskowski, R.A., MacArthur, M.W., Moss, D.S., Thornton, J.M., 1993. PROCHECK: aprogram to check the stereochemical quality of protein structures. J. Appl. Cryst. 26,283–291.

Lee, E., Lobigs, M., 2000. Substitutions at the putative receptor-binding site of anencephalitic Flavivirus alter virulence and host cell tropism and reveal a role forglycosaminoglycans in entry. J. Virol. 74 (19), 8867–8875.

Lee, E., Lobigs, M., 2002. Mechanism of virulence attenuation of glycosaminoglycan-binding variants of Japanese encephalitis virus and Murray Valley encephalitisvirus. J. Virol. 76 (10), 4901–4911.

Lee, E., Hall, R.A., Lobigs, M., 2004. Common E protein determinants for attenuation ofglycosaminoglycan-binding variants of Japanese encephalitis and West Nileviruses. J. Virol. 78 (15), 8271–8280.

Liapustin, V.N., Chunikhin, S.P., Reshetnikov, I.A., Lashkevich, V.A., 1987. Changes in thesynthesis of the virion antigen of the tick-borne encephalitis virus after passagethrough ixodid ticks and small mammals. Vopr. Virusol. (in Russian) 32 (4),451–456.

Lindebach, B.D., Thiel, H.-J., Rice, C.M., 2007. Flaviviridae: the viruses and theirreplication, In: Knippe, D.M., Howley, P.M. (Eds.), 4th ed. Fields Virology, Vol.1.Lippincott Williams and Wilkins and Wolters Kluwer Business Publishers,Philadelphia, USA, pp. 1101–1152.

Lu, Z., Bröker, M., Liang, G., 2008. Tick-borne encephalitis in mainland China. VectorBorne Zoonotic Dis. 8 (5), 713–720.

Lorenz, R.J., Bogel, K., 1973. Methods of calculation, In: Kaplan, M.M., Koprowski, H.(Eds.), Laboratory techniques in rabies, 3rd ed. World Health Organization, Geneva,pp. 321–335.

Maldov, D.G., Karganova, G.G., Timofeev, A.V., 1992. Tick-borne encephalitis virusinteraction with the target cells. Arch. Virol. 127 (1-4), 321–325.

Mandl, C.W., Kroschewski, H., Allison, S.L., Kofler, R., Holzmann, H., Meixner, T., Heinz,F.X., 2001. Adaptation of tick-borne encephalitis virus to BHK-21 cells results inthe formation of multiple heparan sulfate binding sites in the envelope proteinand attenuation in vivo. J. Virol. 75 (1), 5627–5637.

Martínez-Barragán, J.J., del Angel, R.M., 2001. Identification of a putative coreceptoron Vero cells that participates in dengue 4 virus infection. J. Virol. 75 (17),7818–7827.

Navarro-Sanchez, E., Altmeyer, R., Amara, A., Schwartz, O., Fieschi, F., Virelizier, J.L.,Arenzana-Seisdedos, F., Després, P., 2003. Dendritic-cell-specific ICAM 3-grabbing-non-integrin is essential for the productive infection of human dendritic cells bymosquito-cell-derived dengue virus. EMBO Rep. 4 (7), 723–728.

Nawa, M., Takasaki, T., Yamada, K.-I., Kurane, I., Akatsuka, T., 2003. Interference inJapanese encephalitis virus infection of vero cells by a cationic amphiphilic drug,chlorpromazine. J. Gen. Virol. 84 (Pt. 7), 1737–1741.

Nybakken, G.E., Nelson, C.A., Chen, B.R., Diamond, M.S., Fremont, D.H., 2006. Crystalstructure of the West Nile virus envelope glycoprotein. J. Virol. 80 (23),11467–11474.

Pletnev, A.G., Karganova, G.G., Dzhivanyan, T.I., Lashkevich, V.A., Bray, M., 2000.Chimeric Langat/Dengue viruses protect mice from heterologous challenge withthe highly virulent strains of tick-borne encephalitis virus. Virology 274 (1), 26–31.

Pogodina, V.V., Bochkova, N.G., Dzhivanian, T.I., Levina, L.S., Karganova, G.G., Riasova,R.A., Sergeeva, V.A., Lashkevich, V.A., 1992. The phenomenon of antigenicdefectiveness in naturally circulating strains of the tick-borne encephalitisvirus and its possible connection to seronegative forms of the disease. Vopr.Virusol. (in Russian). 37 (2), 103–107.

272 L.I. Kozlovskaya et al. / Virology 398 (2010) 262–272

Potterton, E., McNicholas, S., Krissinel, E., Cowtan, K., Noble, M., 2002. The CCP4molecular-graphics project. Acta Crysatallogr. D. Biol. Crystallogr. 58 (Pt 11),1955–1957.

Protopopova, E.V., Konavalova, S.N., Loktev, V.B., 1997. Isolation of a cellular receptorfor tick-borne encephalitis virus using anti-idiotypic antibodies. Vopr. Virusol.(in Russian) 42 (6), 264–268.

Rey, F.A., Heinz, F.X., Mandl, C., Kunz, C., Harrison, S.C., 1995. The envelope glycoproteinfrom tick-borne encephalitis virus at 2 Å resolution. Nature 375 (6529), 291–298.

Romanova, L.Iu., Gmyl, A.P., Dzhivanian, T.I., Bakhmutov, D.V., Lukashev, A.N., Gmyl, L.V.,Rumyantsev, A.A., Burenkova, L.A., Lashkevich, V.A., Karganova, G.G., 2007.Microevolution of tick-borne encephalitis virus in course of host alternation. Virology362 (1), 75–84.

Ryckaert, J.-P., Ciccotti, G., Berendsen, H.J.C., 1977. Numerical integration of theCartesian equations of motion of a system with constraints: molecular dynamics ofn-alkanes. J. Comp. Phys. 23, 327–341.

Sa-Carvalho, D., Rieder, E., Baxt, B., Rodarte, R., Tanuri, A., Mason, P.W., 1997. Tissueculture adaptation of foot-and-mouth disease virus selects viruses that bind toheparin and are attenuated in cattle. J. Virol. 71 (7), 5115–5123.

Šali, A., Blundell, T.L., 1993. Comparative protein modelling by satisfaction of spatialrestraints. J. Mol. Biol. 234 (3), 779–815.

Schneider-Schaulies, J., 2000. Cellular receptor for viruses: links to tropism andpathogenesis. J. Gen. Virol. 81 (Pt. 6), 1413–1429.

Tassaneetrithep, B., Burguess, T.H., Granelli-Piperno, A., Trumpfheller, C., Finke, J.,Wellington, S., Eller, M.A., Pattanapanyasat, K., Sarasombath, S., Birx, D.L., Steinman,R.M., Schlesinger, S., Marovich, M.A., 2003. DC-SIGN (CD209) mediates denguevirus infection of human dendritic cells. J. Exp. Med. 197 (7), 823–829.

Thepparit, C., Smith, D.R., 2004. Serotype-specific entry of dengue virus into liver cells:identification of the 37-kilodalton/67 kilodalton high-affinity laminin receptor asdengue virus serotype 1 receptor. J. Virol. 78 (22), 12647–12656.

Tsui, V., Case, D.A., 2001. Theory and applications of the generalized born solvationmodel in macromolecular simulations. Biopolymers (Nucleic Acid Sci.) 56 (4),275–291.

Volk, D.E., Chavez, L., Beasley, D.W.C., Barret, A.D.T., Holbrook, M.R., Gorenstein, D.G.,2006. Structure of envelope protein domain III of Omsk hemorrhagic fever virus.Virology 351 (1), 188–195.

Weeke, B., 1973. Rocket immunoelectrophoresis. In: Axelse, N.V., Kroll, J., Weeke, B.(Eds.), A manual of quantitative immunoelectrophoresis. Methods and application.Universitetsforlanget, Oslo, Bergen, Tromso, pp. 37–46.

Zhang, Y., Corver, J., Chipman, P.R., Zhang, W., Pletnev, S.V., Sedlak, D., Baker, T.S.,Strauss, J.H., Kuhn, R.J., Rossmann, M.G., 2003. Structures of immature flavivirusparticles. EMBO J 22 (11), 2604–2613.

Zhang, Y., Zhang, W., Ogata, S., Clements, D., Strauss, J.H., Baker, T.S., Kuhn, R.J.,Rossmann, M.G., 2004. Conformational changes of the flavivirus E glycoprotein.Structure 12 (9), 1607–1618.