Peptide Antagonists That Inhibit Sin Nombre Virus and Hantaan Virus Entry through the {beta}3-Integrin Receptor

Specific therapy is not available for the treatment of hantaviruscardiopulmonary syndrome caused by Sin Nombre virus (SNV). Theentry of pathogenic hantaviruses into susceptible human cellsis dependent upon expression of the ${alpha}$ _vß₃ integrin,and transfection of human ß₃ integrin is sufficientto confer infectibility onto CHO (Chinese hamster ovary) cells.Furthermore, pretreatment of susceptible cells with anti-ß₃antibodies such as c7E3 or its Fab fragment ReoPro preventshantavirus entry. By using repeated selection of a cyclic nonamerpeptide phage display library on purified ${alpha}$ _vß₃, weidentified 70 peptides that were competitively eluted with ReoPro.Each of these peptides was examined for its ability to reducethe number of foci of SNV strain SN77734 in a fluorescence-basedfocus reduction assay according to the method of Gavrilovskayaet al. (I. N. Gavrilovskaya, M. Shepley, R. Shaw, M. H. Ginsberg,and E. R. Mackow, Proc. Natl. Acad. Sci. USA 95:7074-7079, 1998).We found that 11 peptides reduced the number of foci to a greaterextent than did 80 µg/ml ReoPro when preincubated withVero E6 cells. In addition, 8 of the 70 peptides had sequencesimilarity to SNV glycoproteins. We compared all 18 peptidesequences (10 most potent, 7 peptides with sequence similarityto hantavirus glycoproteins, and 1 peptide that was in the groupthat displayed the greatest potency and had significant sequencesimilarity) for their abilities to inhibit SNV, Hantaan virus(HTNV), and Prospect Hill virus (PHV) infection. There was amarked trend for the peptides to inhibit SNV and HTNV to a greaterextent than they inhibited PHV, a finding that supports thecontention that SNV and HTNV use ß₃ integrins andPHV uses a different receptor, ß1 integrin. We thenchemically synthesized the four peptides that showed the greatestability to neutralize SNV. These peptides inhibited viral entryin vitro as free peptides outside of the context of a phage.Some combinations of peptides proved more inhibitory than didindividual peptides. In all, we have identified novel peptidesthat inhibit entry by SNV and HTNV via ß₃ integrinsand that can be used as lead compounds for further structuraloptimization and consequent enhancement of activity.

INTRODUCTION

Top

Introduction

The term hantavirus encompasses any of the more than 20 distinctagents within the genus Hantavirus in the family Bunyaviridae.All of the pathogenic hantaviruses are carried by one or twospecies of wild rodents of the family Muridae in any of thesubfamilies Murinae, Arvicolinae (voles), or Sigmodontinae.About half of the known members of the genus Hantavirus arepathogenic; the rest are known through their isolation fromor genetic detection in wild rodent specimens. Hantavirusescause two human diseases: hemorrhagic fever with renal syndromein the Old World and hantavirus cardiopulmonary syndrome inthe New World (34).

All members of the viral family Bunyaviridae consist of envelopedspherical particles with a helical nucleocapsid and use a genomeconsisting of three negative-stranded or ambisense RNAs. Thethree fully negative-sense RNA segments of the hantavirusesare designated L (large; about 6,500 nucleotides [nt]), M (middle;about 3,600 to 3,700 nt) and S (small; 1,700 to 2,100 nt). Theproteins they encode are an RNA-dependent RNA polymerase (RdRp,from the L segment mRNA); a glycoprotein precursor, GPC, thatis processed into G1 and G2 transmembrane glycoproteins fromthe M segment mRNA; and a nucleocapsid protein/RNA chaperone(N), expressed from the S segment mRNA.

In vivo and in vitro, hantaviruses show tropism for the vascularendothelium (2, 6, 12-14, 43, 44). Hantavirus entry into cellsis dependent upon the cell-surface expression of integrins bearingthe ß₃ subunit (10, 11). Perhaps paradoxically forthese rodent-borne viruses, human integrins such as ${alpha}$ _vß₃confer susceptibility to infection to CHO (Chinese hamster ovary)cells, whereas their murine (Mus musculus) counterparts do not.The pathogenic hantaviruses examined thus far, which includeHantaan virus (HTNV), Seoul virus (SEOV), Sin Nombre virus (SNV),New York virus (NYV), and Puumala virus (PUUV), are able toenter cells that have been transfected by human ${alpha}$ _vß₃,whereas two hantaviruses that have not yet been linked conclusivelyto human disease, Tula virus and Prospect Hill virus (PHV),do not (21). The entry of SNV and other pathogenic viruses canbe prevented by treatment with neutralizing antibodies directedeither against the virus or against the integrin receptor; entryis also inhibited by vitronectin but not by fibronectin. Thatantibodies against the ß₃ subunit are capable of preventinghantavirus entry is particularly interesting because one suchhumanized monoclonal antibody, ReoPro, is a Food and Drug Administration-approvedantithrombotic that is already widely available to clinicians.

The viruses that cause hantavirus cardiopulmonary syndrome havean unusual potential applicability as bioweapons because oftheir high lethality, known aerosol route of transmission, andlack of specific therapies (25). This potential threat couldcreate demand for compounds that can be administered eitherprophylactically or therapeutically to victims of a bioweaponattack. Thus, there has been increased interest in the developmentof specific antiviral inhibitors against threat agents. Oneapproach that has proven effective in reducing the morbidityof viral infections has been preventing the virus from enteringinto the host cells through their cell-surface receptors, suchas through prophylactic or therapeutic administration of specificantiviral immunoglobulins. We sought to develop strategies forinhibiting viral entry based upon relatively inexpensive technologiesthat we hoped could be generalized to other threat agents. Oneapproach we investigated was the development of compounds thatmimic the known ability of anti-ß₃ antibodies to preventthe entry of hantaviruses into cells.

MATERIALS AND METHODS

Top

Materials and Methods

Reagents and monoclonal antibodies. A phage display library expressing cysteine-constrained nonapeptideswas purchased from New England Biolabs (Cambridge, MA). Theoctyl-ß-D-glucopyranoside-solubilized formulationof human integrin ${alpha}$ _vß₃ was purchased from ChemiconInternational (Temecula, CA). The purity of ${alpha}$ _vß₃ is>95% as verified by sodium dodecyl sulfate-polyacrylamidegel electrophoresis using silver stain, and the functional integrityof the ${alpha}$ _vß₃ formulation was verified by assays forbinding to biotinylated fibronectin. ReoPro, the Fab fragmentof the chimeric human-murine monoclonal antibody c7E3, was purchasedfrom Eli Lilly and Company (Indianapolis, IN). ReoPro was suspendedin 0.01 M sodium phosphate, 0.15 M sodium chloride, and 0.001%polysorbate 80, pH 7.2. Cysteine-constrained peptides were synthesizedby Biopeptide (San Diego, CA). Peptides were resuspended ata concentration of 10 mM in phosphate-buffered saline (PBS)containing 5% dimethyl sulfoxide, pH 7.4, and stored at –80°Cuntil use. The purity of the peptides was >95% as verifiedby high-performance liquid chromatography. The peptides wereisolated using high-performance liquid chromatography, and thecorrect mass was confirmed by mass spectrometry. Vero E6 cells(ATCC CRL 1586) were purchased from the American Type CultureCollection (Manassas, VA). Fluorescein isothiocyanate (FITC)-conjugatedanti-rabbit immunoglobulin G (IgG) was purchased from BoehringerMannheim (Indianapolis, IN).

Isolation of ${alpha}$ _vß₃ binding phage. Phage display was performed as previously described with minormodifications, using a cyclic nonapeptide phage library (7).For the first round of panning, 0.5 ml of ${alpha}$ _vß₃ at aconcentration of 100 µg/ml in 0.1 M NaHCO₃ (pH 8.6) bufferwas coated onto 60- by 15-mm petri dishes (Nalge Nunc International,Naperville, IL) overnight in a humidified chamber at 4°C.The following day, the plates were blocked for 1 h at room temperaturewith 5 mg/ml bovine serum albumin in 0.1 M NaHCO₃ (pH 8.6) buffer.The plates were washed six times with TBST (50 mM Tris-HCl,pH 7.5, 150 mM NaCl, 0.1% [vol/vol] Tween 20). The phage librarywas added to the plate at a phage concentration of 2 x 10¹¹in 1 ml TBST. The plate was gently rocked for 1 h at room temperature.Unbound phage was removed by washing 10 times with 1 ml of TBST.Bound phage was then eluted by adding 500 µl of ReoPro(42 µM) for 1 h with gentle rocking at room temperature.The eluate containing the bound phage was removed and saved.Phage was amplified in Escherichia coli ER2738 bacteria andpartially purified by polyethylene glycol precipitation. Thebinding, elution, and amplification steps were repeated using100 µg/ml of ${alpha}$ _vß₃. In the third and fourth roundsof panning, 1 ml of ${alpha}$ _vß₃ at a concentration of 50 µg/mland then 2 ml of ${alpha}$ _vß₃ at a concentration of 10 µg/mlwere coated onto the petri dish. In the final four rounds ofpanning, 3 ml of ${alpha}$ _vß₃ at a concentration of 5 µg/mlwas coated onto the petri dish. As a positive control, the phagelibrary was also panned on streptavidin-coated dishes, and apeptide sequence known to bind streptavidin was obtained afterfive rounds [C-G-X-(F/Y/W)-(S/N)-H-P-Q-C] (7).

Nucleotide sequencing and sequence analysis. After selection and amplification of the phage library on ${alpha}$ _vß₃,automated nucleotide sequencing derived the peptide sequenceon the surface of the phage (DNA Research Services, Departmentof Pathology, University of New Mexico, Albuquerque). Double-strandedDNA was prepared according to the manufacturer's specificationsusing the QIAGEN QIAprep spin miniprep kit (Valencia, CA). DNAwas amplified according to the manufacturers' specificationsusing the ABI Prism BigDye Terminator 3.1 kit (Applied Biosystems,Foster City, CA) and the –96 gIII sequencing primer (NewEngland Biolabs, Cambridge, MA). The reactions were purifiedusing Centrisep spin columns (Princeton Separations, Adelphia,NJ). The samples were then dried, resuspended in formamide,and denatured. Sequencing was performed on a Hitachi 3100 geneanalyzer (Applied Biosystems, Foster City, CA).

Sequence alignments of each peptide to SNV strain SN77734 glycoproteins(6), Hantaan glycoprotein (40), and Prospect Hill virus (24)were done in a pairwise fashion. Alignments were performed usingthe Gap program, which is based on the algorithm of Needlemanand Wunsch (23). We employed a Blosum62 scoring matrix whichhas been previously shown to be effective at detecting proteinsequence similarity based on evolutionary rates and analyzedamino acid pairs in blocks of aligned protein segments ratherthan individually as in earlier Dayhoff rule-based programs(17). The gap shift limit for SNV glycoproteins was set at 12and that for the peptide was set at 10. A quality score witha standard deviation was generated. The quality score for thealignment to any point is equal to the sum of the scoring matrixvalues of the matches in that alignment, less the gap-creationpenalty times the number of gaps in that alignment, less thegap-extension penalty times the total length of all gaps inthat alignment. We generated P values by performing 100 randomizedalignments while preserving the amino acid composition (22).The t test was used to generate a P value for each peptide alignmentto SNV.

Focus reduction test. We plated duplicate wells of 10⁵ Vero E6 cells/well in minimalessential medium (MEM) (GIBCO, Grand Island, NY) with 10% fetalbovine serum onto Lab-Tek 16-well chamber slides (Fisher Scientific,Pittsburgh, PA) 24 h before beginning the assay. After washingthe cells once with PBS, we added 10⁸ to 10¹² (serially diluted)transducing units of phage or a ReoPro dilution series in avolume of 50 µl directly to the now-confluent cells for1 h at 37°C in MEM with 2.5% fetal bovine serum and thenadded 50 µl of SNV strain SN77734 (6) (500 to 1,000 focus-formingunits/ml) in MEM-2.5% fetal calf serum and allowed the virusto incubate for 1 h at 37°C at a biosafety level 3 facility,Centers for Disease Control registration number C20041018-0267.We then removed the virus by aspiration, and after two washesin PBS, we replaced the media with MEM supplemented with 2.5%fetal calf serum. After 24 to 36 h, we fixed the cells in 300-µl/wellice-cold methanol/acetone (50:50) and then maintained the slideat 4°C before staining. We washed the cells twice with PBSbefore staining and then overlaid the cells with 100 µlof a 1:10,000 dilution of polyclonal rabbit anti-SNV recombinantN antibody (4) in PBS in a humidified chamber for 1 h at 37°C.We then added 50 µl of 1:500 FITC-conjugated anti-rabbitIgG in PBS, performed three washes in PBS, counterstained with200 µl of 1:10⁵ Evans Blue (wt/vol) for 3 to 5 min, andfollowed this by rinsing in water. After mounting the slides,we counted the foci by fluorescent microscopic examination.We used control wells containing no inhibitor to retrospectivelyverify the number of SNV foci that had been added to each well.

RESULTS

Top

Results

Identification of ${alpha}$ _vß₃ binding peptides that block entry of SNV. Repeated selection of phage from the cyclic nonapeptide phagedisplay library resulted in the enrichment of phage that bound ${alpha}$ _vß₃. We employed the Fab fragment of antibody c7e3(ReoPro) to elute phage from the plate-bound ${alpha}$ _vß₃,since this mAb has previously been shown to bind ${alpha}$ _vß₃and prevent SNV infection, implying that SNV and ReoPro mightbind similar locations on ${alpha}$ _vß₃ (9). As a result, wepostulated that some nonapeptide-bearing phage might competitivelybind the SNV-binding domain of ${alpha}$ _vß₃. In addition, weeluted with mAb for 90 min in order to elute peptides with higheraffinity and slower off rates (i.e., peptides likely to be potentinhibitors).

After six rounds of panning and amplification, the peptide sequencesof individual plaques were deduced by determining the nucleotidesequences of the portion of the phage genome that encoded them(Table 1). The predicted peptide sequences of 72 plaques weredetermined. We arbitrarily sorted the peptides into four groupson the basis of their abilities to inhibit SNV entry (Table1). Two peptide sequences were identified twice as independentclones, and both were in group 3 (CPPGKSSMC and CFPTTSRGC).Each phage was tested for its ability to block SNV infectionin a focus reduction test, and their activity was compared tocontrol media and concentrations of ReoPro that have been shownto result in maximal inhibition (10, 11). Group 4 contained11 different peptide-bearing phages that showed greater peakinhibitory activity than did 80 µg/ml ReoPro (80 to 91%inhibition for the peptides versus 77% inhibition for ReoPro).Peptides with less than 30% inhibition (group 1) were not investigatedfurther.

View this table:
[in this window]
[in a new window]
TABLE 1. Inhibition with peptide-bearing phage

The 12 most potent peptides were further examined for theirabilities to inhibit SNV entry in a dose-dependent manner (Fig.1 and Table 2). The concentration of each peptide-bearing phagethat produces 50% of its maximum potential inhibitory effect(IC₅₀) was determined. IC₅₀s ranged from 8 x 10⁷ to 9 x 10⁸particles/µl (Table 2). ReoPro showed a dose responseover a range from 1 to 25 µg/ml, but its activity wasasymptotic at 80 µg/ml; a control phage bearing the sequenceCQSLPTHNC (from group 1; Table 1) did not inhibit viral entry.

View larger version (26K):
[in this window]
[in a new window]
FIG. 1. Focus reduction assay for Sin Nombre virus entry of Vero E6 cells. Dose-response curves showing the activities of four representative peptide-bearing phages in comparison to those of ReoPro and a phage bearing the control peptide CQSLPTHNC are shown.

View this table:
[in this window]
[in a new window]
TABLE 2. IC₅₀s of peptide sequences

Specificity of peptides homologous with SNV glycoproteins for inhibition of ß₃-integrin-dependent hantavirus infection. In order to determine whether the peptides that bind ${alpha}$ _vß₃might have sequence similarity to SNV glycoproteins, we comparedall of the 70 peptide sequences with those of SNV G1 and G2(Fig. 2). Eight sequences among groups 2, 3, and 4 that hadsignificant similarity were found, with alignment to discontinuousregions of G1 and G2 (P < 0.0001; Table 1; Fig. 2).

View larger version (42K):
[in this window]
[in a new window]
FIG. 2. Alignment of peptide sequences with glycoproteins (G1 and G2) of SNV. Only those peptides showing statistically significant similarity (P $≤$ 0.0001) with SNV G1 or G2 are shown above the glycoprotein sequence. The peptide sequences shown in bold are those that have inhibitory activities greater than that of 80 µg/ml ReoPro (Table 1; group 4). The proteolytic cleavage recognition site, WAASA, between G1 and G2, is boxed. HTNV and PHV sequences are shown below the glycoprotein sequence in areas where peptides have significant homologies (P < 0.0001) with either HTNV or PHV glycoprotein sequences.

We next compared the abilities of the peptides to inhibit SNV,HTNV, or PHV infection in a focus reduction assay. We testedtwo sets of peptides: one set represented peptides that displayedsequence similarity to the glycoproteins of SNV, and the otherrepresented the 11 most potent peptide inhibitors of SNV entry(group 4).

Blocking of SNV entry into Vero E6 cells with the set of eightSNV-related peptides showed results similar to those shown inTable 1, with inhibition increasing from group 2 to group 3and from group 3 to group 4 (Fig. 3A). HTNV infection was alsoinhibited by all eight of the peptides that showed sequencerelatedness to SNV (39% to 85%) (Fig. 3B). The group 2 peptidesinhibited HTNV infectivity (63 to 82%) to a greater degree thanthey inhibited that of SNV (39 to 59%). Two of the group 3 peptides(CPPGKSSMC and CHNLKPPTC) did not inhibit HTNV virus infectivityto the same degree as they inhibited SNV infectivity (77% and79% for SNV and 39% and 60% for HTNV). In contrast to the relativelysimilar inhibition results for HTNV and SNV, PHV infectivitywas not inhibited significantly (>30%) with seven of theeight SNV-related peptides (Fig. 3C). Notably, the peptide fromgroup 2 (CEWTESSMMC) that inhibited PHV does not display significantsimilarity to PHV glycoproteins (data not shown).

View larger version (20K):
[in this window]
[in a new window]
FIG. 3. Focus reduction assay demonstrating specificities of inhibition of ß₃-integrin-dependent hantaviruses with peptide-bearing phage. Peptide sequences that were found to be significantly similar (P < 0.0001) to SNV G1 and G2 were tested for their abilities to block SNV (A), HTNV (B), or PHV (C) infection. Each peptide and its group (Table 1) is indicated. Virus (2,000 PFU) and 1 x 10⁹ phage were added to each well. The degrees of inhibition are indicated, and standard deviations are shown. Each condition was repeated in duplicate. A nonblocking peptide bearing phage from group 1 (CQSLPTHNC) was used as a nonblocking control. ReoPro and fibronectin were used as blocking controls at final concentrations of 80 µg/ml and 40 µg/ml, respectively. Two types of buffer controls are shown: media is phage buffer without phage, and ReoPro buffer is the ReoPro phosphate buffer without ReoPro. The compositions of these buffers are indicated in Materials and Methods.

The 11 peptides from group 4 were also examined for their abilitiesto block SNV, HTNV, and PHV entry (Fig. 4). All 11 peptidesinhibited SNV and HTNV to high levels (>67%) (Fig. 4A and B).In contrast, a generally much lower level of inhibitionwas seen with PHV (6%-71%), although high concentrations of4 of the 11 peptides inhibited >50% of PHV infection (Fig.4C). In both sets of experiments shown in Fig. 3 and 4, SNVand HTNV infections were inhibited by ReoPro, while PHV infectionwas inhibited by fibronectin, but not ReoPro, in agreement withobservations made by others (10).

View larger version (23K):
[in this window]
[in a new window]
FIG. 4. Focus reduction assay demonstrating inhibitory activities of the peptides with the greatest activities against SNV. The 11 peptides (group 4) that were most potent for the inhibition of SNV (A) in a focus reduction assay were tested in parallel for their abilities to inhibit infection by HTNV (B) and PHV (C). Virus (2,000 PFU) and 1 x 10⁹ phage were added to each well. The degrees of inhibition are indicated, and standard deviations are shown. Each condition was repeated in duplicate. ReoPro and fibronectin were used as blocking controls at final concentrations of 80 µg/ml and 40 µg/ml, respectively. Two types of buffer controls are shown: media is phage buffer without phage, and ReoPro buffer is the ReoPro phosphate buffer without ReoPro. The compositions of these buffers are indicated in Materials and Methods.

In order to determine whether any of the eight SNV-related peptideshad significant sequence similarity to HTNV or PHV as well,we compared the sequences of these eight peptides with HTNVand PHV. Only two of the eight peptides had significant sequenceidentity (P < 0.0001) with either HTNV or PHV at the siteanalogous to that in SNV (Fig. 2). No peptide with significantsequence similarity to HTNV and SNV but not PHV was identified.

Potency of synthetic peptides alone and in combination. We next chemically synthesized four cyclic peptides from group4: the three that showed the most inhibition (CPFVKTQLC, CLHKPWSRC,and CRSLTDNQC) and the most potent that had significant sequencesimilarity to SNV surface glycoproteins (CPGHIHRTC) (Table 1).These four peptides were screened for their abilities to inhibitSNV entry in vitro (Fig. 5). At concentrations of 2 mM, thepeptides were able to inhibit infectivity by 51%, 15%, 26%,and 14%, respectively. This degree of inhibition was more thanthose of control phage (CQSLPTHNC) or peptide (CLLRMRSIC). Nonetheless,three of the four peptides displayed quite modest levels ofinhibition (<30% with CLHKPWSRC, CRSLTDNWC, and CPGGHIHRTC).We anticipated that peptides may have a shorter half-life inthe presence of serum due to the proteases in the serum. Forthis reason, we performed the assay both in the presence andin the absence of serum and observed no difference in theirabilities to inhibit SNV infection (data not shown).

View larger version (28K):
[in this window]
[in a new window]
FIG. 5. Focus reduction assay demonstrating the inhibitory activities of peptide-bearing phage and selected synthetic peptides. Phage, peptides, and ReoPro were adsorbed onto confluent monolayers of Vero E6 cells (ATCC CRL 1586) on microscope slides for 1 h before the addition of 50 to 100 immunofluorescence assay foci of SNV strain SN77734; after 24 to 36 h, we stained the monolayers with polyclonal rabbit anti-SNV N antibody (1:1,000) followed by FITC-conjugated anti-rabbit IgG and counted foci. The percentages of inhibition using final concentrations of 1 x 10⁹ phage/µl or 2 mM peptide are shown. Peptides were cyclized by the reduction of flanking cysteines at residues 1 and 9 before addition. Maximal peptide concentrations were chosen based upon their maximal solubilities. ReoPro was used at a final concentration of 80 µg/ml. Three types of buffer controls are shown: media is phage buffer without phage, PBS is phosphate-buffered saline without peptide, and buffer is the ReoPro phosphate buffer without ReoPro. The compositions of these buffers are indicated in Materials and Methods.

We next examined whether using peptides in combination couldhave additive effects (Fig. 5). The relatively high potencyof CPFVKTQLC was not increased when combined with any of theother three peptides. However, combinations of peptides CLHKPWSRCand CRSLTDNWC (50%) or CRSLTDNWC and CPGHIHRTC (40%) showedenhanced inhibition in comparison to each of the three peptidesalone (P < 0.05). The combination of all four in comparisonto CPFVKTWLC alone did not result in a statistically significantincrease in inhibition (P > 0.05). These findings indicatethat among synthesized peptides, CPFVKTQLC is the most potent,but combinations of CLHKPWSRC and CRSLTDNWC or of CRSLTDNWCand CPGHIHRTC may approximate the same level of inhibition asthat of CPFVKTWLC alone.

DISCUSSION

Top

Discussion

The use of combinatorial peptide libraries displayed on thesurfaces of filamentous bacteriophage offers an efficient meansof obtaining a diverse set of peptides that bind a target protein.In this study, we identified peptides that bind ${alpha}$ _vß₃and that inhibit SNV entry. A combinatorial library generated70 different peptide sequences that bind ${alpha}$ _vß₃ and wereeluted by the Fab fragment of c7E3 mAb (ReoPro). Further testingshowed that 11 of the 70 peptide sequences had inhibitory abilitiesthat exceeded that of ReoPro (even when used at 80 µg/ml)when expressed on the phage surface. In addition, even thoughno virus was used to identify the peptides, 8 of the 70 peptidesproved to have significant sequence relatedness to SNV glycoproteins.We compared all 19 peptide sequences (11 with maximum inhibitoryability and 8 homologous peptides) for their abilities to inhibitSNV, HTNV, and PHV infections. There was a distinct differencein the abilities of these peptides to inhibit SNV and HTNV incomparison to PHV, which supports the contention that SNV andHTNV use ß₃ integrins and PHV uses a different receptor,ß₁ integrin, to enter cells (10). We then chemicallysynthesized the four peptides that showed the greatest abilityto neutralize SNV and showed that they could, singly and incombination, inhibit SNV infection when added as free peptidesoutside of the context of a phage. These peptides inhibitedviral entry in vitro, albeit to a lesser degree than did phageexpressing the same peptide. Thus, we have identified novelpeptides that primarily inhibit infection by pathogenic hantaviruses(SNV and HTNV) and that can be used as lead compounds for furtherstructural optimization and consequent enhancement of activity.

All of the evidence that hantaviruses such as SNV or HTNV requireß₃ integrin to enter cells is based upon focus assaysthat measure downstream evidence that entry has occurred (8,9, 31). Gavrilovskaya et al. (8, 9) first demonstrated thatpathogenic hantaviruses appear to require expression of thehuman ß₃ integrin in order to efficiently enter cells.Until the present work, this discovery had not been confirmedin other laboratories. Although it is possible to attributethe requirement for ß₃-integrin expression to a directbinding interaction between SNV or HTNV and ß₃ integrin,it had also been possible that ß₃-integrin expressionprovided an unknown signal to cells that preconditions themin other ways for hantavirus entry via another surface receptor.Since the inhibitory peptides described here were identifiedby competition between a monoclonal antibody and ${alpha}$ _vß₃,our findings support the concept that SNV is interacting directlywith ß₃ integrin.

There were clear differences in the levels of peptide inhibitionamong SNV, HTNV, and PHV, consistent with the concept that HTNVand SNV are using ß₃ integrin for cell entry, whilePHV uses ß₁ integrin. In this context, it is interestingthat some of the peptides block all three viruses. ß₁and ß₃ integrin are highly homologous proteins (44%identity) (1) that have similar tertiary structures (41, 42).Furthermore, many members of the ß₁- and ß₃-integrinfamilies bind identical ligands, although with different affinities(5, 15, 16, 42). Finally, peptide sequences containing RGD sequenceshave been described that block native ligand binding to bothß₁ and ß₃ integrins (18, 26-30, 32, 38).Thus, there is strong evidence that the ß₁ and ß₃receptors are structurally related, can bind similar ligands,and can be inhibited by similar peptide antagonists. It is perhapsnot surprising that viruses from the same group would bind torelated integrin receptors and that binding to different integrinsmay be inhibited in some cases by peptides similar to thoseseen in our study. Nonetheless, understanding these fine structuraldifferences will likely be an important contribution to furtherunderstanding the pathophysiologic differences among hantaviruses.

In this study, we identified peptides that block virus entryby competing phage with a mAb and the virus receptor. We chosethis approach since we had previously used it successfully toidentify inhibitory peptides of other integrin-ligand pairs(35-37). It is unclear what epitope on ß₃ integrinis recognized by ReoPro, and it is likely that antibodies arelarge enough to prevent interactions with regions well beyondthe antibody's own epitope or the virus binding site on ß₃.As a result, when competing with a mAb, it is likely that peptidesare identified that compete with the mAb but not the virus,as seen in our study. In fact, 10 of 11 peptides (group 4) withthe greatest inhibition of SNV infectivity did not have significantsequence relatedness to SNV glycoproteins. Identification ofinhibitory peptides that do not have significant sequence relatednessto native ligands has been described in other systems (35-37),and a large amount of work has indicated that nonhomologousamino acids may mimic the binding activities of each other (reviewedin reference 35). In addition, a comparison of the peptide sequenceshomologous with SNV to HTNV and PHV glycoproteins did not clearlyidentify possible contact points of virus glycoproteins withtheir receptors. The involvement of contact points between SNVglycoproteins and ß₃ would be suggested if the peptideswere found to bear amino acid sequence relationships to bothSNV and HTNV glycoproteins but not to PHV glycoproteins. However,only one of the eight peptides that were homologous with SNVglycoproteins showed significant similarity with HTNV, althoughall eight blocked HTNV infection. Two of the eight peptideshad sequence similarity to PHV, yet neither of these blockedPHV infection. While this analysis does not demonstrate thatany of our eight peptides with sequences similar to those ofSNV glycoproteins represent contact points between SNV and ${alpha}$ _vß₃,the large number of peptides with statistically significanthomology raises the possibility that at least a subset of sequencesmay represent such contact points. Further studies will be neededto determine whether the peptides we have identified can exhibitadditive inhibitory effects on viral entry.

The factors that could contribute towards the inhibitor potentialof a peptide expressed on phage are cyclicity, multivalency,steric hindrance due to size, and the possible conformation-stabilizingand -forming effects of the phage environment. All the peptidesthat we tested in this study were disulfide restrained in acyclic form and were expressed on the pIII protein of the phage,indicating that there may be up to five copies of peptide perphage (8). As a result, the multivalency of the peptide on thephage may contribute to the significant improvement in the maximalinhibitory ability of peptide expressed on phage in comparisonto that of synthesized peptides. Others have observed this phenomenonand have alternatively proposed that the phage environment providesconformation or structural stability to the peptides (19, 20,33). Finally, the large size of the phage could contribute additionalsteric hindrance and cause accentuated inhibition of SNV bindingwith the cell surface receptor on that basis.

Although we identified peptides that in the context of presentationon the surface of phage were substantially more effective atblocking hantavirus entry than was an antibody to ${alpha}$ _vß₃,our findings agree with those of others that inhibitors thatbind ${alpha}$ _vß₃ do not completely block hantavirus entry.In addition, even a combination of four inhibitory peptidesshowed only a 61% inhibition (Fig. 5). These findings are inagreement with those of other investigators, who have shownthat even combinations of anti- ${alpha}$ and anti-ß subunitantibodies cannot further improve the inhibition of SNV entry(10). Although some residual entry of viruses is nonsaturableand likely to be nonspecific, our study as well as observationsmade by others suggests that the glycoproteins of hantavirusesare highly folded and may have multiple discontinuous epitopes(39).

Hantavirus interactions with ${alpha}$ _vß₃ integrins providea potentially contributory mechanism for the observed alterationsin vascular permeability during infection and also provide potentialtargets for therapeutic intervention. ReoPro is a humanizedmouse-derived Fab fragment from the original monoclonal antibodyc7E3, which recognizes both ${alpha}$ _iibß₃ and ${alpha}$ _vß₃integrins and is used therapeutically to inhibit platelet aggregation.However, mAb therapy using even humanized antibodies may havesignificant side effects when compared to the side effects seenin patients after peptide administration (3, 9). This observationmotivated our efforts to identify small peptides that may inhibitSNV infection. The peptides that we identified in this studymay serve as lead compounds for further optimization. The optimizationof peptide antagonist activity through alanine and homologousamino acid substitution has resulted in 2- to 3-log improvementsin inhibitory activity in other systems and may be justifiedin this system as well (35, 37). Finally, structural informationabout peptides may be used as a guide to create orally availablenonpeptide organics.

ACKNOWLEDGMENTS

This work was supported by Public Health Service grants R21AI53334, R21 AI053400, and UO1 AI56618.

FOOTNOTES

* Corresponding author. Mailing address: UNM School of Medicine, 2325 Camino de Salud, CRF 223, Albuquerque, NM 87131. Phone: (505) 272-9762. Fax: (505) 272-5186. E-mail: rlarson{at}salud.unm.edu.

REFERENCES

Top