4Fl.

AD__ _ _

II

IMMUNOGENIC CRARACTERIZATIOn OF THE DENUE VIRUS SPECIFIEDNONSTRUCTtURAL GLYCOPROTEIN GP48

(lV3, SOLUBLE COMPLEMENT FIXING ANTIGEN)0CD

I m Midterm Report

Jacob J. Schlesinger, N.D.

September 9, 1988

Supported by

U.S. ARMY MEDICAL RESEARCH AND DEVELOPMENT COMMANDFort Detrick, Frederick, Maryland 21701-5012

Contract No. DAND17-87-C-7088

Rochester General Hospital QTICInfectious Disease Unit ECTF

1425 Portland Avenue OE. 2Rochester, NY 14621

DOD DISTRIBUTION STATEMENT

Approved for public release; distribution unlimited

The findings in this report are not to be construed asan official Department of the Army position unless sodesignated by other authorized documents.

S8 12 22 052'

SECURITY CLASSIFICATION OF THIS PA(E

Form ApprovedREPORT DOCUMENTATION PAGE OMBNo.0704-018M

Ia. REPORT SECURITY CLASSIFICATION lb. RESTRICTIVE MARKINGS

Unclassified2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION /AVAILABILITY OF REPORT

2b. DECLASSIFICATION/ DOWNGRADING SCHEDULE Approved for public release;..I E T distribution unlimited

4. PERFORMING ORGANIZATION REPORT NUMBER(S) 5. MONITORING ORGANIZATION REPORT NUMBER(S)

6a. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING ORGANIZATIONRochester General Hospital (If applicable)

6c. ADDRESS (City, State, and ZIP Code) 7b. ADDRESS (City, State, and ZIP Code)

1425 Portland AvenueRochester, NY 14621

8.. NAME OF FUNDING/SPONSORING 8b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBERORGANIZATION U.S. Army Medical (If applicable)

Research & Development Command Contract No. DAMDI7-87-C-7088

St ADDRESS (City, State, and ZIP Code) 10. SOURCE OF FUNDING NUMBERS

PROGRAM PROJECT ITASK WORK UNITFort Detrick ELEMENT NO. NO. 3M1 [NO. ICCESSION NO.Frederick, Maryland 21701-5012 61102A 61102BS13 AA 049

11. TITLE (include Security Classification)

Immunogenic Characterization of the Dengue Virus Specified Nonstructural Glycoprotein GP48(NV3s Soluble Complement Fixinq Anti en,12. PERSONAL AUTHOR(S)

Jacob J. Schlesinger. I-D13.. TYPE OF REPORT I135. TIME COVERED 114. DATE OF REPORT (Year, Month, Day) IS. PAGE COUNTMidterm /FROM 3/9/87 TOjj q"8 1 1988 September 9 3216. SUPPLEMENTARY NOTATION

17. COSATI CODES 18. SUBJECT TERMS (Continue on reverse if necessary and identify by block number)FIELD GROUP SUB-GROUP Characterization of flavivirus NSl; Dengue Fever; RA I;06 03 Infectious Diseases; Monoclonals; Virion; Antigen

I ISTRACT (Continue on reverse if necessary and identify by block number)

8SEfforts."through the midtenm--ndng -period -ave been concerned with mechanismsof the protective immune response to the flavivirus nonstructural protein NSI andidentification of protective NSI domains for incorporation in possible future subunitf lavivirus vaccine ~

1We have reported that immunization with yellow fever (YF) NSI protects mice andmonkeys against lethal YF infection and that mice immunized with dengue (DEN)-2 NSIare similarly protected against this flavivirus. These observations have since beenconfirmed and extended - immunization of mice with a YF NSI recombinant fusion proteinconferred partial protection against the virus (Cane & Gould, 1900) and most recentlyC.J. Lai and coworkers at NIH have demonstrated solid protection after lethal DEN 4challenge in mice actively immunized with recombinant DEN 4 NSl alone or in the form of a . _

20. DISTRIBUTION /AVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION"UNCLASSIFIEDAUNUMITED M SAME AS RPT. 0 DTIC USERS Unclassified

22a. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (Include Area Code) 22c. OFFICE SYMBOL

Mrs. Virainia M. Mille 301/663-7325 SDD Form 1473, JUN 86 Previous editions are obsolete. SECURITY CLASSIFICATION OF THIS PAGE

(19 cont.). /-polyprotein, produced in a baculovirus vector, Vaccinationstud1-es--ifh these recombinants in rhesus monkeys are in progress: to date wedetected cytolytic activity in all animals given these NS1 (but not E)constructs. In collaboration with Ors. Robert Putnak (WRAIR, Washington,D.C.) and Mark Cochran (MicroGeneSys, West Haven, CN) we characterized theImmune response in rabbits to recombinant DEN 2 NS1 produced by similarmethods in a baculovirus vector. Results showed a relatively sluggish low

" titered antibody response compared to that achieved by vaccination withauthentic NS1. Moreover, unlike results with Lai's DEN 4 NS1 recombinants,no cytolytic or complement-fixing activity resulted from immunization with

* the MicroGeneSys NS1 material. The immune response to reduced and alkylatedauthentic DEN 2 NS1 was also examined in order to get some idea of the extentof denaturation, e.g., conformational change, that would still allow genera-tion of a cytolytic (and therefore presumably protective) immune response -

we showed that all twelve cystelne residues of NS1 are involved in intramole-cular bonding. Sera against reduced/alkylated NS1 exhibited marginal CFactivity but no cytolytic activity. Difficulties inherent in the isolatioland purification of adequate amounts of ctll culture-derived authentic NS1encourage us to pursue the recombinant approach as a resource for flaviviralNS1. It is hoped that such material will be forthcoming from work In pro-ress in the laboratories of Drs. Putnak (YF, DEN 2), Henchal (DEN 3) and LalDEN 4).

'Cumulative data from assays which measure the capacity of anti-NS1 mono-

clonal antibodies (Mab) to sensitize cells to complement-mediated lysis sug-gests a correlation between such activity and Mab protective capacity Re-sults of NSI epitope mapping experiments using biotinylated anti-YF Nil Maband rabbit antisera produced against authentic and recombinant 17D YF'NS1(the latter prepared by Dr. Charles Rice) suggest that complement-me-diatedlysis is subserved by selected NS1 domains and that the capacity of an anti-body to sensitize YF infected cells to complement-mediated lysis depends onboth its isotype and epitope specificity.

Attempts to identify NS1 protective epitopes after chemical fragmenta-tion or enzymatic digestion have met with little success. YF NSI fragmentsproduced by cyanogen bromide or N-Chlorosuccinimide fragmentation failed toreact with anti-NS1 Mab by Immunoprecipitation or Western blot assay. How-ever, several partial tryptic peptides of DEN 2 NS1 separated by HPLC ap-peared to react with two monoclonal antibodies, one of which is protective.Unavoidably high peptide loss during collection will require substantiallymore NS1 protein than available from current sources to pursue this approach.It is hoped that the future availability of recombinant NS1 will provide thenecessary material. Using a commercially available kit (PEPSCAN, CambridgeResearch Biochemlcals,.Gsnrbridge, UK), we synthesized all 404 possible over-lapping hexapeptideS of 17D YF NSI as well as a 98 amino acid segment of DEN2 NS1, shown by R. Putnak to be Mab-reactive. Screening of these hexapep-tides with all available anti-YF or DEN 2 NS1 Mab failed to identify anyreactive hexapeptides. Taken together, the data may suggest that protectiveepitopes of NSL are discontinuous and conformationally dependent.

Ifmmune recognition of viral antigen on the surface of infected cells mayprovide a mechanism of host defense and recovery in flavivirus infection Weused Mab and monospecific polyclonal sera against YF NS1 and virion envelopeprotein (E) as probes to detect these antigens on the surface of YF-infected

) \

(19 cont). mouse neuroblastoma (Neuro 2a) and human adenocarcinoma (SW13).Surface NS1 and E were detected on these cells by radiobinding and immuno-fluorescent assays. Of great interest is the observation that monospecificrabbit antl-NS1 serum, but not monospecific rabbit anti-E serum or anti-EMab, sensitized YF-infected Neuro 2a and SW13 cells to complement-mediatedlysis. Productive 17D YF infection was reduced up to 1000-fold in synchro-nously infected Neuro 2a maintained in medium containing an IgG2 a anti-YF NS1Mab (1A5) or monospecific rabbit anti-NS1 serum in the presence, but not ab-sence, of complement. That some anti-NS1 Mab interfere with the in vitrocytolytic activity of protective antl-NS1 Mab raises the possibility of unde-sirable antigen modulation. This will be tested by Mab mixing experimentsin the passive immunization YF mouse challenge model. These results suggesta mcctanism of protection against flavivirus infection in addition to thatprovided by direct virus neutralization.

In anticipation of studies to test the protective capacity of a mixedNS1/E subunit preparation in mice, we purified 17D YF E by immunoaffinitychromatography, using a flavigroup-reactive Mab as ligand. Purified materialwas identified as a 33 Kd protein. Rabbits immunized with E33 produced neu-tralizing antibody to 17D YF and DEN 2 in high titer and mice immunized withthis protein were protected against lethal YF challenge.

Future efforts will focus on the mechanisms of antibody recognition ofplasma membrane NSI and complement activation. Isolation and characteriza-tion by sequencing, virulence testing, etc., of YF escape variants resistantto protective anti-NS1 antibody may provide important information in this re-gard. Similarly, the observation that some anti-NS1 Mab interfere with thelytic activity of protective antl-NS1 antibody by means other than competi-tive binding will be explored further.

Adequate amounts of authentic DEN 2 NS1 have been accumulated for chal-lenge studies in rhesus monkeys in collaboration with Dr. Bruce Innis(AFRIMS). It is hoped that satisfactory recombinant DEN 2 NSL will soon beavailable for parallel study.

Access ion Vow

N S RA&IPCTABU-nunced 0 GO

TJistificatiO GO,

!6

I~ ~ I DV Irbtl-Distribut-ionI

AvailabilitY Codes

Dinit Special

i I I I I

FOREWORD

Cpinions, interpretations, conclusions and recommendations are those of theauthor and are not necessarily endorsed by the U.S. Army.

s~W e re copyrighted material is quoted, permission has been obtained toUJs-uch material. -

_,_tedWhere material from documents designated for limited distribution ist, permission has been obtained to use the material.

Citations of commercial organizations and trade names in this report do& -onstitute an official Department of the Army endorsement or approval ofthe products or services of these organizations.

.* eIn conducting research using animals, the investigator(s) adhered to the1 1 de for the Care and Use of Laboratory Animals," prepared by the Comitteeon Care and Use of Laboratory Animals of the Institute of Laboratory AnimalResources, National Research Council (NIH Publication No. 86-23, Revised 1985).

N6 For the protection of human subjects, the investigator(s) have adheredto policies of applicable Federal Law 45CFR46.

A In conducting research utilizing recombinant DNA technology, theinvestigator(s) adhered to current guidelines promulgated by the NationalInstitutes of Health.

PIS atur-e 5 a e

TABLE OF CONTENTS

Introduction..............................1

Body..................................... 2

Concl usions..............................8

Bibliography.............................10

Table 1................................. 12

Table 2................................. 13

Table 3..................................14

Table 4..................................15

Table 5..................................16

Table 6..................................17

T able 7................................. 18

Table 8..................................19

F igure 1 ................................. 20

Figure 2 ................................. 21

Figure 3 ................................. 22

Figure 4 ................................. 23

Figure 5 ................................. 24

TABLE OF CONTENTS (CONT.)

Figure 6................................ 25

Figure 7................................ 26

Figure 8 .................................27

Figure 9 .................................28

Figure 10 ................................ 29

Figure 11............................... 30

Figure 12 ................................31

Figure 13 ................................32

INTRODUCTION

We have shown that passive transfer of monoclonal anti-bodies against the yellow fever and dengue virus nonstruc-tural protein NS1 protects mice against lethal challengewith the respective viruses. Active immunization withpurified NS1 conferred protection against lethal infectionin rhesus monkeys (yellow fever) and mice (dengue encepha-litis). Most recently, protective immunization of mice withrecombinant yellow fever or dengue virus NS1 has beendemonstrated by others.

Evidence of a protective immune response to flavivirusprotein NS1. Overview.

We have presented evidence of protection in experimen-tal animals against lethal YF and DEN 2 infection by passivetransfer of monoclonal antibodies (Mab) against the flavi-virus nonstructural protein NSI and by active immunizationwith NS1 isolated from YF or DEN 2-infected cells (1,2,3).Although the mechanisms of such protection is presentlyunclear, cumulative data from our laboratory suggests acorrelation between protection and the capacity of anti-NS1antibody to sensitize flavivirus-infected cells tocomplement-mediated cytolysis. These results have now beenconfirmed and extended. Gould and coworkers demonstratedthe protective capacity of antl-NS1 YF Mab (4) and Henchaland coworkers (5) obtained similar results with a panel ofanti-DEN 2 NS1 Mab. Most recently, Cane and Gould (6)showed that mice immunized with a YF NS1 recombinant ex-pressed in E. coli as a galactosidase fusion protein werepartially protected against lethal challenge with the virusand C.J. Lai's group NIH) has prepared immunogenic baculo-virus recombinants expressing DEN 4 gene segments encodingC-prM-E-NS1-ns2a as well as E and NSI individually (7).Mice immunized with each of these constructs were completelyprotected against lethal intracerebral challenge with DEN 4(7, C.J. Lai, personal communication). In anticipation ofchallenge studies, monkeys have been vaccinated with theseconstructs and preliminary assays in our laboratory showcytolytic activity in sera from animals immunized with theNS1 constructs (see below). Taken together, the NS1 dataprovide evidence of an alternative to direct viral neutral-ization in the protective immune response against flavi-virus infection.

-1-

Characterization of monoclonal antl-NS1 Nab.

Table 1 summarizes relevant characteristics of avail-able anti-YF and DEN 2 NS1 Mab produced in our laboratory orin those of our collaborators. More detailed descriptionsof these antibodies have been published or are in press(1,4,5).

Identification of protective NS1 epitopes.

A. Role of complement-mediated lysis in the protectiveimmune response to NS1.

Cumulative data obtained in the course of the presentstudy strongly suggest a correlation between the protectivecapacity of anti-NS1 Mab and their capacity to sensitizecells to complement-mediated lysis. Figures 1 and 2 comparelytic activities (in Neuro 2a or SW13 cells, respectively),and protective capacities among available anti-YF NSI Mab.In general, at least three categories of anti-NS1 Mab acti-vity emerge from these assay results: lytic protective, non-lytic protective, and non-lytic non-protective. That somenon-lytic anti-NS1 Mab are protective (Gould's Mab 492, 863,925, 428, and Henchal's IgG1Mab) suggests that other mecha-nisms of antibody-dependent immunity may be operating.Among these, antibody-dependent cell-mediated cytotoxicity(ADCC) would be expected but, to date, we have been unableto demonstrate ADCC using these antibodies in assays employ-ing primary mouse macrophages or a continuous mouse macro-phage cell line (P388D1) as effectors and YF or DEN-infectedNeuro 2a or SW13 cells as targets.

The capacity of anti-NS1 antibody to sensitizeflavivirus-infected cells to complement-mediated lysis indi-cates that this protein is expressed on the cell surface(discussed further below). As such, recognition by effectorarms of the cellular immune response, e.g., cytotoxic Tcells (Tc), might be expected. Recent reports of H-2restricted T cell killing of flavivirus infected targetsare supportive in this regard, although antigen specificityof the Tc cells was not determined (8) and only T helpercells were reported after T cell cloning (9).

B. NS1 epitope mapping.

It is presently unclear whether complement-mediatedcytolytic activity is subserved by specific NS1 domains orwhether such activity is simply determined by an anti-NS1antibody's complement-fixing capacity without regard to theNS1 domain recognized. Preliminary data from our laboratoryis Inconclusive. Results of competitive binding assays em-ploying biotinylated anti-YF NS1 Mab are shown in Table 2.This assay measures the capacity of protective and nonpro-tective anti-YF NS1 Mab to block binding of biotin-labeled

-2-

complement-fixing, protective anti-NS1 Mab. Binding avidi-ties among this panel of Mab were quite similar (data notshown). Although the assay was limited by denaturation ofthe majority of the Mab after biotinylation, a number ofconclusions are suggested by these data:

1) At least 2 separate epitopes are recognized byprotective cytolytic Mab; 2) a number of nonprotective Mab(4E3, 2G2) recognize a potentially "protective" site(s); 3)two protective antibodies (428, 925) which are not lytic,e.g., complement-mediated, appear to define an epitope dis-tant from at least one which subserves complement-mediatedlysis; 4) two antibodies (428, 492) of an isotype (IgG )expected to fix complement do not compete with the cytOTyticMab and are not cytolytic. Of interest in regard to thelast is failure of rabbit serum prepared against recombinantYF NS1 (provided by Dr. Charles Rice) to sensitize cells tocomplement- mediated lysis or interfere with binding to NS1of biotinylated cytolytic protective Mab above (1A5, 8G4,423, 871). The recombinant YF NS1 was constructed as a TrypE fusion protein and differs from authentic NS1 by deletionof the first 91 amino acids from the NS1 amino terminus. Incontrast, rabbit serum that we prepared to authentic YF NS1was cytolytic and competed with the protective lytic Mab(Figure 3), Both rabbit anti-YF NS1 sera bind to authenticNS1 (the anti-recombinant NS1 less so, Figure 4), immunopre-cipitate NS1, and stain YF-infected SW13 and Vero cells inan indirect immunofluorescent assay. Taken together, theseresults are consistent with the view that complement-mediated lysis is subserved by selected NS1 domains and thatthe capacity of an antibody to sensitize YF-infected cellsto complement- mediated lysis depends both on its isotypeand specificity. It is tempting to speculate that thecytolytic domain(s) are located at the NS1 amino terminus,but masking of other NS1 regions by the Tryp E fusionprotein could also account for the results.

C. Peptide mapping.

Our approach to identification of NS1 immunoreactivedomains has largely assumed that at least some are repre-sented by continuous epitopes or that, if discontinuous,such regions are close enough to each other to allowpreservation of antibody recognition after fragmentation ofthe complete protein. Methods used for fragmentation arecompletely empirical and have included chemical cleavagewith cyanogen bromide (CnBr, methionyl peptide) and N-chlorosuccinimide (NCS), tryptophanyl peptide) or enzymaticdigestion with trypsin (lysinyl, argininyl peptide) (10,11,12). NS1 for this purpose was purified by immunoaffinitychromatography by methods reported earlier from our labor-atory (1-3). Initially, immunoaffinity-purified radioio-dinated YF and DEN 2 NS1 were subjected to CnBr fragmenta-tion followed by immunoprecipitation with representative

-3-

anti-NS1 Nab or monospecific antisera. None of the frag-ments generated were precipitated. The method was modifiedy exposing NS1-containing acrylamide gel slices to CnBrvapors (10) in an effort to minimize formic acid-induceddenaturation. The resulting fragments were electro-elutedinto SDS-acrylamide gels followed by transfer to nitrocel-lulose paper and immunoblotting. Again, no reactivity wasdetected with any of the available monoclonal or monospeci-fic anti-NS1 probes. More recently we have found markedlyheightened sensitivity of detection of intact NS1 usingpolyvinylidene difluoride membranes (Immobilon, Millipore)in Western blots, but use of this support also failed todetect NS1 fragments (CnBr, NCS, tryptic) reactive with anyof the available anti-NS1 antibodies.

In parallel experiments, we have purified DEN 2 NS1 byreverse-phase HPLC (Figure 5) and generated reproducibletryptic peptide maps. Several peptides appeared to reactwith two anti-NS1 Mab, one of which is protective (Figure6). However, these results were inconsistent probably duein large measure to the small amounts of available NS1 andthe marginal sensitivity of the immunoassays at the NSIconcentrations used. Additionally, protein retrieval fromcollection tubes after acetonitrile evaporation was compli-cated by substantial and apparently irreversible binding tothe plastic or glass surface leading to substantial loss.This problem has been commented upon by others and repre-sents an example of the highly empirical nature of thismethodology (14).

The approach of Geysen et al. (15) was used to systema-tically synthesize all possible short (hexa-) peptides of YFas well as part of DEN 2 using sequences provided by Drs.Charles Rice, Vincent Deubel, and Robert Putnak. Thiscomputer-assisted method allowed rapid concurrent synthesisand immunoassay (ELISA) of all 404 possible overlappinghexdpeptides covering the total 409 amino acid sequence of17D YF NS1. Results were disappointing: screening of YF NSIhexapeptides with each available anti-YF Mab or monospecificrabbit and monkey (2) anti-YF NS1 sera failed to demonstratea single convincingly positive peptide on multiple tests.Similarly, no reactivity with any DEN 2 NSI Mab was detectedagainst amino acid residues 261-359 of DEN NS1 shown by R.Putnak to be reactive with several Mab as well as monospeci-fic rabbit anti-NS1 sera in a Western blot analysis ofrecombinant NS1 produced in E. Coli. Correct synthesis ofcontrol peptides included in the "PEPSCAN" kits suggeststhat our failure to detect reactive NS1 peptides was not theresult of error in synthesis. We conclude that possiblecontinuous NS1 epitopes recognized by these antibodies arelikely to be larger than 6-mers. The expense and uncertain-ty of this approach led us to abandon it. Dr. Lew Markoff(NIH) is presently studying overlapping 15-20 mers of DEN 4NSI for immunoreactivity (personal communication) and his

-4-

results are awaited with great interest. However, takentogether our results strongly suggest that the availableprotective YF and DEN 2 NS1 Mab are directed at discontinu-ous and probably conformationally determined epitopes (seebelow).

D. Immunoreactivity/immunogenicity of recombinant DEN2 and 4 NS1.

A problem central to the NS1 fragmentation work hasbeen our inability to isolate NS1 in large enough quantity.Recombinant DEN 2 NS1 (New Guinea C) prepared in a baculo-virus vector by MicroGeneSys (West Haven, CN) and partiallypurified by them was made available to us for furthercharacterization. It was hoped that this material wouldresolve the NS1 supply problem. In preliminary experimentswe compared the anti-NS1 immune responses among rabbitsimmunized with recombinant and authentic DEN 2 NS1. Addi-tionally, the immune response to reduced and alkylatedauthentic DEN 2 NSI was examined in order to get some ideaof the extent of denaturation, e.g., conformational change,that would still allow generation of a cytolytic (and there-fore presumably protective) immune response: all (twelve)cysteine residues are involved in intramolecular bonding(Figure 7). Figure 8 shows the antibody response againstauthentic NS1 among rabbits immunized with either authenticor recombinant NS1 or reduced and alkylated authentic NS1.Antibody titers were measured in an ELISA incorporatingaffinity-purified authentic DEN 2 NS1 in the solid phase. Abrisk antibody response was seen among rabbits immunizedwith native or denatured authentic NS1. However, the re-sponse was delayed and of low titer among rabbits immunizedwith the MicroGeneSys recombinant material. When these serawere tested in parallel against recombinant NS1 (Figure 9),similar high titers developed in rabbits immunized withauthentic NS1 and to a lesser degree after immunization withreduced and alkylated NS1. One rabbit immunized with recom-binant NS1 developed substantial titers by 8 weeks, whereastiters in the other remained on the low side. These resultsare reflected by Western blot analysis shown in Figures 10and 11. The figures show blot results at the time of thefirst and second boost and last bleed at 3,6, and 8 weekspost-prime, respectively. Rabbits immunized with authenticNS1 generated strong specific antibody responses to the 46and 44 Kd moieties of NS1, as did the rabbit immunized withreduced and alkylated NS1. Only weak responsiveness wasseen by 8 weeks in the two rabbits immunized with the recom-binant NS1 preparation. Similarly, sera from rabbits immu-nized with authentic NS1 developed antibody against recombi-nant NSI, as shown by Western blots in Figure 11. Of noteis the somewhat higher molecular weight of recombinant NSI(ca. 52-55 Kd) and its appearance as a triplet which prob-ably reflects glycosylation differences. A consistent posi-tive blotting "smudge" appears at the top of the recombinant

-5-

NS1 strips, which we believe reflects polymerization of therecombinant protein. This is a consistent finding in ourhands. Mab prepared to authentic DEN NS1 did not react withthe recombinant protein in this assay. Of potential impor-tance is the fact that cytolytic antibody developed only inrabbits immunized with authentic NS1: sera from rabbits im-munized with reduced and alkylated authentic NS1 or recom-binant NS1 were not lytic. Similarly, substantialcomplement-fixing activity was present only in sera fromauthentic NS1-immune rabbits (Table 3). These results,though preliminary, are somewhat disappointing, if the pro-position that cytolytic activity correlates with protectivecapacity is correct: critical changes in the conformation ofcandidate small NS1 peptide vaccines may be required if theimmune response to them is to be protective.

Preliminary results with sera from rabbits and monkeysimmunized with DEN 4 baculovirus recombinants from C.J.Lai's laboratory have been more encouraging. Both rabbitand monkey sera against DEN 4 constructs consisting of NS1alone, or as part of a C-)NS1 polyprotein, sensitized DEN4-infected cells to C'-mediated cytolysis (Table 4) andthese constructs (as well as that of E) protected miceaga:nst lethal challenge (C.J. Lai, personal communication).Results of DEN 4 challenge among monkeys immunized withthese recombinants should be available shortly.

Cell surface expression of NS1.

Current knowledge of flavivirus maturation and releasesuggests that exit of these viruses from infected cells islargely by exocytosis of virion-containing vesicles (16),although evidence of budding from the plasma membrane hasalso been presented (17). That anti-YF NS1 Mab sensitizeYF-infected cells to complement-mediated lysis providesstrong evidence that NS1 is cell membrane associated. Table5 examines the cytolytic activities of monospecific rabbitsera prepared to YF NS1 and E using YF-infected Neuro 2a andSW13 target cells. Rabbit sera against the respective unin-fected cells served as positive controls. Anti-NS1, but notanti-E, serum was cytolytic as were the respective controls.These data, along with our failure to demonstrate cytolyticactivity among 11 anti-YF E Mab (9/11 are IgG, 42) sug-gested the absence of detectable plasma membri.. How-ever, in direct biding assays employing radioiodinated anti-NS1 or E Mab and in indirect binding assays with radioiodi-nated Staph. protein A and monospecific rabbit antibodyagainst NS1 or E, both proteins were detected on the sur-face of Neuro 2a and SW13 cells (Table 6). These resultsare consistent with speculation that NS1 is an integralplasma membrane protein, whereas E is recognized in the formof released virion which accumulates on the cell surfaceremaining loosely bound. Washed infected cells held at 4 Cbound substantially less anti-E antibody.

-6-

The effect of anti-NS1 antibody (1A5) and complement onthe replication of 17D YF in Neuro 2a cells was examined aswell (Figure 12). Peak titers of virus were reached by 48hours at which time there was a greater than 100-fold reduc-tion in titrable virus from cells treated with antibody 1A5and complement. At 48 hours post infection no CPE was seenunder any treatment condition, and fewer than 5% of cellstreated with lA5 and complement were IFA-positive, whereas50-75% of cells treated with antibody and inactivated com-plement or myeloma controls were IFA-positive. Monolayerstreated with antibody 1A5 and active complement were intactwith minimal CPE at 70 hours post infection at which timeCPE was complete under all other conditions. In parallelassays, non-complement-fixing anti-NS1 antibodies had noeffect on virus production in the presence or absence ofcomplement. Table 7 summarizes results with anti-NS1 Maband monospecific rabbit serum in both Neuro 2a and SW13cells. In collaboration with Dr. Robert Putnak (WRAIR) wewill attempt to isolate and characterize conditional 17D YFmutants selected by growth in Neuro 2a and SW cells undercomplement and antibody pressure. NS1 sequencing of suchmutants could provide important information about sitescritical to complement-mediated cytolysis and, presumably,regions of NSl that subserve protection.

To our surprise, we have found that selected non-lyticanti-NS1 Mab can substantially interfere or completelyabrogate the cytolytic activity of protective anti-NS1 Mab.In some instances this is the case even though the inter-fering antibody shows no evidence of competition with theprotective antibody for binding to purified NSl or to liveinfected cells. Figure 13 shows representative results ofan experiment comparing the lytic activity of protectiveanti-YF NS1 antibody 1A5 (serially diluted) in the presenceof a single low dilution of control myeloma protein PC5 ornon-protective non-cytolytic anti-YF Mab (2D). In separ-ate assays we determined that these interfering anti-NS1Mab were not anti-complementary, since they had no effect onthe lytic activity of rabbit serum raised against the SWcell membrarne itself. That interfering antibodies alsoreduced the lytic activity of monospecific rabbit serum toNS1 is consistent with the idea that their effect may resultfrom cross-linking and capping of plasma membrane NSI withsubsequent loss of this antigen from the cell surface. Thein vivo significance of this observation demands furtherstudy and will be addressed in future experiments.

Preliminary Immunogenic characterization of 11D YF envelopeprotein E.

In anticipation of studies to test the protective ef-fect of immunization with NS1/E subunit mixtures in mice, wepurified 17D YF E from lysates of infected Vero cells by

-7-

immunoaffinity chromatography using a flavigroup-reactiveMab as the ligand. Purified material could not be identi-fied as intact E but bore antigenic determinants of E asdefined by selective reactivity with anti-E monoclonal anti-bodies. The reactivity correlated with a 33 Kd band, thepredominant component on polyacrylamide gel electrophore-sis. Rabbits hyperimmunized with E33 produced neutralizingantibody to YF in titers similar to those obtained with asingle dose of 17D YF vaccine (Table 8). The rabbit anti-sera had a high degree of cross-neutralizing activityagainst DEN 2, whereas the human antisera, as expected,failed to neutralize dengue. Rabbits immunized with highdose live 17D YF produced higher titers of neutralizingantibody to 17D YF, but cross-neutralizing activity againstdengue was lower than that seen in rabbits immunized withE33. This E protein fragment, then, appears to express agroup-reactive determinant, suggesting the possibility ofheterotypic protection after vaccination.

Mice actively immunized with E33 were protected againstlethal intracerebral challenge with YF: 16/20 E33-immunemice survived vs. 6/33 control ovalbumin-immune mice(p<0.001).

CONCLUSION

Cumulative results from our laboratory and from thoseof our collaborators indicate that the immune response tothe flavivirus nonstructural protein NSI is protective,thereby providing an alternative to direct viral neutraliza-tion in the defense against these viruses. That the immuneenhancement phenomenon should not be an issue with anti-NS1antibody further emphasizes the potential value of denguevaccines incorporating this protein. Cumulative data frommonoclonal antibody screening of peptide fragments of yellowfever and dengue NS1, as well as overlapping hexapeptides ofthis protein, have, in our hands, failed to provide evidencethat protective NS1 epitopes are expressed as continuousdeterminants. To the contrary, it seems highly probablethat NS1 recognition by protective antibody is conformation-ally dependent. Failure of a number of flavivirus E and NS1recombinants, with anticipated conformational variance, toprovide evidence of a protective immune response is consis-tent with this belief.

Further efforts in our laboratory will focus on themechanisms of antibody recognition of plasma membrane NS1and resulting complement activation. In preliminaryexperiments we have found that selected nonprotective YFmonoclonal antibodies can substantially interfere with orentirely abrogate the cytolytic capacity of protectiveanti-NS1 antibodies. Since the interfering antibodies do

-8-

not compete with binding by the cytolytic antibodies, themechanisms of interference is unclear. One possibility isthat they modulate plasma membrane NS1 expression in a man-ner similar to that described for measles virus (18).Should this be true, then appearance of such antibody couldbe undesirable, leading to loss of immune recognition offlavivirus-infected cells and persistence of infection.This possibility will be tested in passive immunizationexperiments. Abrogation of protection conferred by cytoly-tic monoclonal anti-NS1 antibody in mice concurrently im-munized with "interfering" monoclonal anti-NSI antibodywould have obvious implications for the design of NS1immunogens. The importance of complement activation in theprotective response will be further explored by comparingthe protective capacities of univalent (Fab Fab/c), biva-lent (Fab 2 ), and intact IgG NS1 Mab in the 4 NS1 mousemodel.

In parallel experiments we will attempt to select YFmutants resistant to the effect of complement-mediatedcytolytic antibody. Cloning and sequencing of such variants(in collaboration with Dr. Robert Putnak) should provideimportant information about NS1 epitopes involved in immunecytolysis. In essence, sequencing of such variants mayenable us to identify elements of a conformationally deter-mined protective NS1 region which could not be mapped by thefagmentation of "overlapping peptide" appraoches. It isanticipated that such variants would be lethal to micepassively immunized with Mab which would otherwise protectagainst the parent strain of virus.

Enough chromatographically purified DEN 2 NS1 has nowbeen accumulated to permit evaluation of its protectivecapacity in rhesus monkeys and this is currently beingplanned with Dr. Bruce Innis (AFRIMS). This experiment hadbeen delayed in the hope that authentic and recombinant NS1could be tested simultaneously; failure of the MicroGeneSysmaterial to induce a satisfactory immune response in rabbitshas ruled this material out.

-9-

Bibliography

1. Schlesinger JJ, Brandriss MW and Walsh EE. 1985. Protec-tion against 17D yellow fever encephalitis in mice by pas-sive transfer of monoclonal antibodies to the nonstructuralglycoprotein gp48 and by active immunization with gp48.J Immunol 135:2805.

2. Schlesinger JJ, Brandriss MW, Cropp CB and Monath TP. 1986.Protection against yellow fever in monkeys by immunizationwith the yellow fever nonstructural protein NS1. J Virol60:1153.

3. Schlesinger JJ, Brandriss MW and Walsh EE. 1987. Protec-tion of mice against dengue 2 virus encephalitis by immuni-zation with the dengue 2 virus nonstructural glycoproteinNS1. J Gen Virol 68:853.

4. Gould EA, Buckley A, Barrett ADT and Cammack N. 1986.Neutralizing (54K) and nonneutralizing (54K and 48K) mono-clonal antibodies against structural and nonstructuralyellow fever virus proteins confer immunity in mice. J GenVirol 67 :591.

5. Henchal EA, Henchal LS and Schlesinger JJ. 1988. Synergis-tic interactions of anti-NS1 monoclonal antibodies protectpassively immunized mice from lethal challenge with dengue-2virus. J Gen Virol 69:2101.

6. Cane PA and Gould EA. 1988. Reduction of yellow fevervirus mouse neurovirulence by immunization with a bacteri-ally synthesized non-structural protein (NS1) fragment.J Gen Virol 69:1241.

7. Zhang YM, Hayes EP, McCarty TC et al. 1988. Immunizationof mice with dengue structural proteins and nonstructuralprotein NSI expressed by baculovirus recombinant inducesresistance to dengue virus encephalitis. J Virol 62:3027.

8. Kesson AM, Blanden RV and Mullbacher A. 1987. The primaryin vivo murine cytotoxic T cell response to the flavivirusWest Nile. J Gen Virol 68:2001.

9. Wren MF, Doherty PC and Allan JE. 1987. Flavivirus-specific murine L3T4+ T cell clones: Induction, characteri-zation and cross-reactivity. J Gen Virol 68:2655.

10. Zingde SM, Shirsat NV and Gothoskar BP. 1986. Peptide map-ping of proteins in gel bands after partial cleavage withacidic cyanogen bromide vapors. Anal Biochem 155:10.

-10-

11. Lischwe MA and Ochs D. 1982. A new method for partialpeptide mapping using N-Chlorosuccinimide/urea and peptidesilver staining in sodium dodecyl sulfate-polyacrylamidegels. Anal Biochem 127:453.

12. Cleveland DW, Fischer SG, Kirschner MW and Laemmli UK.1977. Peptide mapping by limited proteolysis in sodiumdodecyl sulfate and analysis by gel electrophoresis. J BiolChem 252:1102.

13. Matsudaira P. 1987. Sequence from picomole quantitites ofproteins electroblotted onto polyvinylidene difluoridemembranes. J Biol Chem 262:10035.

14. L'Italien JJ. 1986. Solid phase methods in protein micro-sequence analysis in Methods of Protein Microcharacteriza-tion, JE Shively, Ed., Humana Press, Clifton NJ p. 279.

15. Geysen HM, Meloen RH and Barteling SJ. 1984. Peptidesynthesis used to probe viral antigens for epitopes to aresolution of a single amino acid. PNAS USA 81:3998.

16. Ishak R, Tovey DG and Howard CR. 1988. Morphogenesis ofyellow fever virus 170 in infected cell cultures. J GenVirol 69:325.

17. Hase T, Summers PL, Eckels KH and Baze WB. 1987. Anelectron and immunoelectron microscopic study of dengue 2virus infection of cultured mosquito cells: maturationevents. Arch Virol 92:273.

18. Sissons JG and Oldstone MBA. 1980. Antibody-mediateddestruction of virus-infected cells. Adv Immunol 29:209.

-11-

Table 1. Characteristics of anti-NS1 monoclonal antibodies

DEN 2Immunizing ELISA titer (-log10 )

Clone DEN Strain Ig vs DEN 2 (NGC) NS O Protection

*16-25/3 D80-100 GI < 2*20-1/ o GI < 2*27-12/4 " GI 5.8 +*34-23 " G1 4.8*40-21/9 " G2b 5.6*47-10/10 " G2a 2.7*63-15 " G1 5.6 +*68-5/16 " GI 5.3 +*101-47 " GI 2.3**6A8 NGC M ND +**9A9 " G1 > 5.0**4D11 " G1 3.7+D7-4E9 " G1 > 5.0

17DYF

Immunizing Ig ELISA titer (-Iol1o)Clone YF Strain vs 17D YF Protection

#863 Porterfield G1 > 7.0 +#925 " GI 5.0 +#871 " G2a 6.0 +#979 " G2a 3.0 +#999 " G2a 6.0 +#993 " G2a 6.0 +#917 I G2a 6.0 +#992 " G2a 5.0 +#423 " G2a 6.0 +#492 if G2a < 2.0 +#428 " G2a 5.0 +#924 " M < 2.0 +*1A5 Connaught G2a > 7.0 +*8G4 " G2b 6.0 +*4E3 " GI > 7.0*2G2 " G3 ND*2010 " GI 6.0

* EA Henchal + WRAIR # EA Gould ** Rochester

-12-

Table 2. Competitive Binding Assay

Unlabeled competitoranti-YF NSI monoclonal Isotype Biotinylated anti-YF

antibody IgG NS1 monoclonal antibody

1A5 8G4 423 871

IA5 2a +a + + +8G4 2b + + + +423 2a - +/- + +

Protective, 871 2a + + + +cytolytic 992 2a - -

999 2a - -

979 2a - -

917 2a - - + -

Protective, 428 2a - - -

not cytolytic 925 1 - - -

2D10 1 - -

4E3 1 + + + +Not protective 2G2 3 + + + +not cytolytic 863 1 . . . .

492 2a ....

a competition + = >75%; +/- = 25-50%; - = <1%

-13-

Table 3. Complement fixing activity among sera from rabbitshyperimmunized with authentic, recombinant (MicroGeneSys),

or reduced/alkylated DEN 2 NS1

CF titers against DEN serotype

Rabbit DEN 2 NS1 1 2 3 4 NMB a

1 Authentic 128 1,024 256 64 64

2 Authentic 512 16,384 2,048 128 32

3 Recombinant 16 16 16 16 16

4 Recombinant 8 8 8 8 16

5 Reduced/alkylated 32 64 16 8 16

a normal mouse brain control

-.14-

Table 4. Complement-mediated cytolytic activity amongrhesus monkeys immunized with recombinant DEN 4 protein

Day post-primA1 y immunizaion% specific -Cr release

Immunogen Mk# Day:O 28 35

Control SF9 058D 0 0 0243 0 0 2371D 2 12 3

NS1 003D 0 0 123J8 0 6 204GV 4 1 17

C--0NS1 190D 0 6 11200D 3 2 18819C 0 8 29

E 070D 0 0 041J 3 4 4927C 0 0 0

a All sera (@1/50) assayed at same time in quadruplicate.

-15-

Table 5. Complement-mediated lysis of 17D YF-infected SW13or Neuro 2a cells

% specific 51Cr release from 17D

infected:

Rabbit anti- SW13 Neuro 2a

YF NS1 54 16

YF E 2 2

SW13 34

Neuro 2a -72

Table 6. Binding of radioiodinated monoclonal anti-YF NS1or E antibodies to YF-infected SW13 or Neuro 2a cells

SW13 Neuro 2a

370C 40C 37°C 4°C

*alA5*PC5 2220A214 b 592A106 2108t468 979±203

*1A5+1A5 270±128 90142 121.45 85L11

*2EIOPC5 2100*254 404±122 36590 203*90

*2E10±2E1O 256*78 77±30 96*36 98t37

a *Radioiodinated Mab

b cpm boundtsd; quadruplicate samples

-17-

Table 7. Growth of 17D YF in tissue culture in the presenceof antibody and complement

Cell type

Antibody Complement N2a SW13

PC5 .a 6 .3b 7.0

+ 6.2 7.1

1A5 6.4 7.0

+ 4.1 5.3

NRS 6.2 7. 1

+ 6.1 7.1

R anti-YF NS1 - 6.3 7.P

+ 5.2 6.3

a Complement (-) = heat inactivated, (+) = active

b Log 10 PFU/ml culture supernatant titered in Vero cells

-18-

Table B. Neutralizing antibody titers in rabbits

immunized with 17D YF virion or E p33

Rabbit immunized with:

PRNT

E p33 a 17D YF DEN 2

# 1 256 64

2 64 64

3 128 32

4 512 256

17D virion b

#1 1,024 32

2 2,048 32

a ca 25ug x 3 b 10 6 PFU I.V.

MTB > PLOT Cl C2

%SURVIVE-

- o *

50+

5 0 + - - - - -- - - - - - - - -- - - -. .......% L Y S I S

0.0 5.0 10.0 15.0 20.0

Figure 1. Lyric versus protective capacities among anti-YFNS1 Mab. Mab (Table 1) were ordered by protective capacityas determined by reported (Ref. 1,4) survival inintracerebral mouse protections51ests and cytolytic activitymeasured by per cent specific Cr release from 17D YF-infected mouse neuroblastoma cells (Neuro 2a).

-20-

- r-O.a 7 (O|.O

MTB > plot cl c3

%SURVIVE-- * .3* /,Q"

- ?7975+

_ *999- ,f- * ." ,92.

50+ •

25+ ,;Lc_

- *'fCS

- •*;.2IO

0+---------- +--------------------------+-------------+------ %Cr Rel

-0 16 32 48 64 80

Figure 2. Lytic versus protective capacities among anti-YFNS1 Mab. As in Figure 1, but % lysis determined in 17D YF-infected human adenocarcinoma cells (SW 13).

-21-

- -

* . . , . IH

9

7 . . . . S

74-

3-

2 . I

Fiur 3. Cmeitv bidn betee rabtsr gis

auhni or reobnn Y . .S an bitnyae I i-N

*a 1A or 8G4. Pri n rabi seu sevda acnrl

Bi t nya e 1A 'li: III' Iii auhN 1R bO I o

primun seum % i *

Bitnyae 8G4 vs iikathN1 a 0 eo S,

preimune s rum,

I . . ~ -22 -

*VJ # I

to I I 1 I# 1 1 1t 1 l

6

4

3-'6

* , 6 6. .6 *lL

2; 6 6 g - ,.) 6

Fi ur 4. Reltiv bid n -vdte to auheti 17D *

of ra bi an i-a * .. 1 ( 0 ) an con ecu iv bl e s f o

-2-

390.000

7.

190.000

-io.000a-iO 15 2 25 30 35 40 45 50 5

o 6 10 15 4 Minutes Press RESUME

Figure 5. Fractionation of affinity-purified DEN 2 NSI.Approximately 10 umol DEN 2 NSI loaded onto Vydac C4 columnand separated with a linear 30-80% Buffer B gradient at Iml/min. Buffer A 0.1% TFA. Buffer B 80% acetonitrile inBuffer A.Peaks 1 and 2 reacted with anti-DEN 2 NSl Mab (Western blot)and represent 46 kd (1) and 30 kd (2) moieties of DEN NSI.

-24-

245.000_

117.500

-jO.OOO

0 10 15 20 25 30 35 40 45 50 55MInutes Press RESUME

Figure 6. Practionation and immunoassay of tryptic peptidesderived from dithiothreitol-reduced DEN 2 NS1 (46 Kd). Ca100 ug HPLC-purified NSl in 50 mM ammonium bicarbonate pH8.0 was digested with HPLC-purified trypsin (protein: enzyme50:1) for 2 hours at 37 0 C and loaded onto a Vydac C18 columnfollowed by separation with a 0-100% buffer B gradient. 1ml/min and absorbance at 314 nm. Individual peptide peakswere lyophilized using a Savant Speed-Vac drier, reconsti-tuted in phosphate buffered saline containing 0.1% Triton-Xand tesetd by ELISA for reactivity ( ) with "protective"anti-DEN NSI Mab 65A8 or nonprotective Mab 9A9.

-25-

AB

1 2 12RNR R NR

Figure 7. Acetylation of DEN 2 NS1 by 14C odacetamide.Panel A. Silver stain, 11% SDS pol acrylamide gel.

aDithiothreitol-reduced idane 1, "R") or unreduced (lane 2,"NR") NS1 treated with C iodoacetamide. Panel B. Radio-autogram of Panel A. Alkylation occurred only afterreduction of NSI.

-26-

6- 4

_______________ aSI[1

4A A A aNSI [I.0- aNSI [2]

Log anti-aD2 3*- rNSI [I]NS1

•E3- rNS1 [2]

2o- 00 - R/A NSI[I

1 1 0-.

00 1 2 3 4 5 6 7 8

WEEK

Figure 8. Antibody response (ELISA) against authentic (a)DEN 2 NS1 among rabbits immunized with aDen 2 NS1,recombinant (r) DEN 2 NS1 or reduced and alkylated (R/A)aNS1. Arrows indicate combined intradermal and subcutaneousinjections of NS1 in complete Freund's adjuvant (prime) orincomplete Freund's adjuvant (boost). ca 25 ug prime; ca 10ug booster doses were used.

-27-

6-

5--15,=8

l -- IP @- aNSi [I]4-

Log anti-rD2 rNSi [1NSI -iI rNSi [2]

1-0- rNS I [2

2--13 0- R/A NS1 [I

0 1 2 3 4 5 6 7 8WEEK

Figure 9. Antibody response (ELISA) against recombinantDEN 2 NSI among rabbits immunized with authentic (a),recombinant (r , or reduced and alkylated (R/A) authenticDEN 2 NSI. Route, schedule and dosage as in Figure 8.

-28-

Al rIGlE.N : eI. US|

U.,.,

0- +m

S 3. 3~ 4. 4 8.WJ ANI1 RI A NS41 rNSI f Il

Figure 10. Western blot against authentic DEN 2 NS1.Rabbit sera drawn at 3,6, and 8 weeks following immunizationwith authentic (a), reduced and alkylated (R/A), orrecombinant (r) DEN 2 NS1 (see Figs. 8 and 9). Preimmuneserum (NRS) and anti-NS1 Mab 9A9 were controls.

-29-

10--

Figure 11. Antibody responses against recombinant (r) DEN 2NS1 as in Fig. 10. Note failure of anti-aNSi Mab 9A9 toreact with rNS1.

-30-

0

20-

~qq

Ma 21

-31-

6

0

5

zLU4z Ax

0.

U,

Z E

ILL

o A 1A5+C1-5

2 A 1A5+ AC'* PC5 + C,o PC5 + AC'

10 20 30 40 50 60 70

HOUR POST INFECTION

Figure 13. Effect of antibodies on infection of neuro-blastoma cells. Neuro 2a cell monolayers grown in 22 mmpl stic Costar cluster plates were infected with 17D YF at37 C. After incubation for 60 minutes, residual virus wasremoved by multiple washes and the media replaced withMEM-lO% FCS containing anti-NS1 monoclonal antibody 1A5 at afinal dilution of 10 " and active ( A ) or heat-inactivated( A ) rabbit complement added to a final dilutio,. of 1/20.Myeloma protein PC-5, added with ( a ) and without ( 0 )active complement, served as a control. At 24 hour inter-vals, supernatant media was sampled for infectious virus ina Vero cell plaque assay and active or heat-inactivatedcomplement was replenished in remaining wells to a finaldilution of 1/20.

-32-