Neutralization sensitivity and accessibility of continuous B cell epitopes of the feline immunodeficiency virus envelope

Transcription

1 Journal of General Virology (1996), 77, Printed in Great Britain 759 Neutralization sensitivity and accessibility of continuous B cell epitopes of the feline immunodeficiency virus envelope J. Richardson, ~ I. Fossati,~t A. Moraillon, 2 S. Castelot, ~ P. Sonigo ~ and G. Pancino 1. 1 G~ndtique des Virus (ICGM-CNRS UPR 0415), Institut Cochin de Gdndtique Moldculaire, 22 rue Mdchain, Paris and 2 Gdn~tique moldculaire gdndtique virale Institut National de la Recherche Agronomique, Ecole Nationale Vdtdrinaire d'alfort, Maisons-Alfort Cedex, France Antibodies elicited during natural infection of domestic cats by the feline immunodeficiency virus (FIV) recognize continuous epitopes in nine domains of the virus envelope glycoproteins. Whereas antibodies directed against the V3 envelope region can neutralize laboratoryadapted virus, nentralization of FIV has been shown to depend upon cellular substrate, and virus adaptation to laboratory cell lines may alter sensitivity to neutralizing antibodies. We therefore undertook a systematic analysis of the continuous B cell epitopes of the envelope of a primary FIV isolate, Wo. The capacity of feline antisera elicited against nine envelope domains to neutralize primary and laboratory-adapted virus was evaluated in feline peripheral blood mononuclear cells (PBMC). The laboratory-adapted strain Petaluma was used to compare neutralization in PBMC and CrandeU feline kidney cells (CrFK). Antibodies specific for the V3 region neutralized both primary and laboratory-adapted virus whether residual infectivity was measured in CrFK or in feline PBMC. However, a large discrepancy in the efficiency of neutralization was observed in these ex vivo models of infection, perhaps reflecting diversity in the interaction between virus and different cellular targets. We also examined the accessibility of epitopes on the functional oligomeric envelope complex of FIV. Most of the epitopes were poorly exposed on native envelope glycoproteins at the surface of live infected cells. The most accessible domain was the only domain sensitive to neutralizing antibodies. These results suggest that inaccessibility on oligomeric envelope glycoproteins may frequently underlie the insensitivity of diverse lentivirus B cell epitopes to neutralization. Introduction Primary lentivirus infection gives rise to a vigorous humoral response followed by a reduction in plasma viraemia. Nevertheless, the immune response does not ordinarily resolve the infection and persistent infection ensues. The degree of benefit derived by the host from the humoral response to lentiviruses, whether in contributing to the control of acute primary infection or to the maintenance of a prolonged asymptomatic state, is unknown. Certain observations made on infection with the human, simian and feline immunodeficiency viruses (HIV, SIV and FIV) have indicated that antibodies may afford protection. Thus, naive primates have been protected from challenge with HIV-1, HIV-2 and SIV by the administration of purified antibodies or the transfer of sera from animals immunized with virus subunit vaccines or with attenuated virus strains (reviewed in * Author for correspondence. Fax fr Present address: Department of Microbiology, H-325 Health Sciences Center, SC-42, Seattle, WA 98195, USA Norrby & Matthews, 1993). Similarly, cats have resisted challenge with FIV after receiving the sera of immunized cats (Hohdatsu et al., 1993) and maternal antibodies from either vaccinated or infected queens protected kittens from FIV infection (Pu et al., 1995). Nevertheless, certain subunit vaccines of FIV have augmented or accelerated infection after challenge (Hosie et al., 1992) and the demonstration that such enhancement may be extended to naive animals upon passive transfer of plasma from immunized cats strongly suggests that antibodies mediated the enhanced infection observed in vivo (Siebelink et al., 1995a). Various ex vivo models of infection have been used to evaluate the functional characteristics of antibodies directed against immunodeficiency-inducing lentiviruses. Antibodies elicited during natural infection are frequently observed to neutralize infection of cultured cells (Tozzini et al., 1993; Osborne et al., 1994; reviewed for HIV in Nara et al., 1991); however, where appropriate cellular systems exist to assess enhancing antibodies, these have also been found (Takeda et al., 1988; Robinson et al., 1988; Homsy et al., 1988). The measurement of neutralizing antibodies depends upon, SGM

2 760 J. Richardson and others amongst other parameters, both virus and cellular substrate (Baldinotti et al., 1994). In particular, adaptation of HIV-1 to growth in cultured cell lines seems to select virus strains which are inordinately sensitive to neutralization (Moore et al., 1995; Wrin et al., 1995; Sullivan et al., 1995). The use of ex vivo models has shown that most if not all neutralizing activity in serum of humans infected with HIV is specific for the envelope glycoproteins and has permitted the localization of several B cell epitopes sensitive to neutralizing antibodies in both the surface (SU) and transmembrane (TM) components of the virus envelope (reviewed in Nara et ai., 1991). Among B cell epitopes of the envelope gtycoproteins recognized during natural HIV-1 infection, some, the so-called continuous or linear epitopes, are preserved in short peptides or fragments of envelope glycoproteins divested of native structure, while others are discontinuous or conformationally sensitive (Moore & Ho, 1992). Although the humoral response directed against epitopes dependent on higher order structure, such as those overlapping the receptor-binding domain of HIV-1, may play a significant role in neutralization (Ho et al., 1991; Earl et al., 1994), a considerable proportion of antibodies neutralizing HIV-1 recognize linear epitopes in a putative looped structure in the third variable region (V3) of the SU subunit (Vogel et al., 1994). A systematic approach to the identification of linear B cell epitopes of the FIV envelope has shown that antibodies elicited during natural infection with FIV recognize nine domains of the virus envelope glycoproteins (Pancino et al., 1993). One such domain has been observed to induce antibodies which neutralize the infectivity of laboratory-adapted strains of FIV for Crandell feline kidney cells (CrFK) (Lombardi et al., 1993; de Ronde et al., 1994). In the present work we describe an exhaustive analysis of the immune response of the natural host against all the previously defined linear B cell epitopes expressed in Escherichia coli as 3040 amino acid insertions in bacterial fusion proteins. In particular, we have analysed the functional relevance of the response to each envelope domain with regard to neutralization of primary and laboratory-adapted strains in two different ex vivo models of infection, CrFK and feline peripheral blood mononuclear cells (PBMC). We have also considered the accessibility of these epitopes on functional oligomeric envelope glycoproteins in relation to sensitivity to neutralizing antibodies. The most accessible linear domain was observed to elicit antibodies neutralizing both primary and laboratory-adapted virus, whether residual infectivity was measured in CrFK or in feline PBMC. However, a large discrepancy in the apparent efficiency of neutralization was observed in these different cellular contexts. Methods Tissue culture. The HO6T1/2 subline of CrFK was generously provided by R. Osborne (MRC Retrovirus Research Laboratory, University of Glasgow, UK) and corresponds to the CrFK/ID10/R clone (Osborne et al, 1994). A chronically infected CrFK cell line (CrFK-Petaluma) was established after infection of HO6T1/2 cells with the Petaluma isolate of FIV (FIVp~tamm~). The CrFK fibroblasts were cultivated in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum (FCS), 2 mm-glutamine, 100 IU/ml penicillin and 100 gg/ml streptomycin (complete DMEM). The feline T lymphoid cell lines, Fet-1 and FL-4, were a gift from J. Yamamoto (University of Florida, Fla., USA). The FL-4 cell line is chronically infected with FIVp~t~um~ and is a derivative of the uninfected cell line Fet-I (Yamamoto et al., 1991). The FL-4 cells were cultivated in RPMI-1640 with FCS, glutamine, penicillin and streptomycin as above (complete RPMI), while the Fet-1 cells were cultivated in complete RPMI supplemented with 100 U/ml interleukin 2 (IL-2), 50 ~tm-2-mercaptoethanol (2-ME) and 10 mm-hepes. Feline PBMC were isolated from the blood of specific pathogen-free (SPF) cats by density gradient centrifugation and activated for 3 days in complete RPMI containing 5 lag/ml concanavalin A, 50 IaM-2-ME and 10 mm-hepes. For subsequent culture of PBMC, 100U/ml of recombinant human IL-2 was added to the medium. Virus. The infectious molecular clone pfiv34tf10 (Talbott et al., 1989) was obtained from J. Elder through the AIDS Research and Reference Reagent Program (NIAID, NIH, Bethesda, Md., USA). Stocks of FIVpet~h~m~ and the Wo isolate (Moraillon et al., 1992) of FIV (FIVw~,) were derived from the supernatant of the FL-4 celt line or infected PBMC, respectively. FIVwo has been passaged a limited number of times and only in feline PBMC. Viral titres were expressed as TCID~0 calculated for infection of CrFK-HO6T 1/2 or PBMC under conditions used for neutralization. Preparation of fl-galactosiduse [usion proteins. The nine linear antibody-binding domains of the SU (SU1 to SU5) and TM envelope proteins (TM 1 to TM4) of FIVwo (Pancino et al., 1993) were expressed as 13-galactosidase fusion proteins from 2gtl 1 bacteriophage clones in lysogenic E. coli Y1089 as described (Huynh et al., 1985). The peptides of FIVwo incorporated in fusion proteins (Pancino et al., 1993) and corresponding peptides of FIVp~t~ ~... are shown in Fig. 1. flgalactosidase fusion proteins were purified from lysates of 2gtll lysogens by immunoaffinity chromatography using immobilized mouse anti-//-galactosidase monoclonal antibody (ProtoSorb lacz Immunoaffinity Adsorbent; Promega) according to the manufacturer's instructions. Since small amounts of fusion proteins containing the SU4, SU5, TM1 and TM4 domains were produced by 2gtl 1, the EcoRI FIV DNA inserts from these clones (and of the clone containing the SU3 domain) were subcloned in the C-terminal part of the lacz gene in ptex, which expresses high levels of a cro-fl-galactosidase protein (Stanley & Luzio, 1984). Expression and purification of ptex fusion proteins were performed conventionally (Stanley, 1988). Fusion proteins from both 2gtll and ptex were subjected to SDS-PAGE to evaluate the proportion of contaminating proteins and quantified using the Micro BCA method (Pierce). The immunological reactivity of the fusion proteins was tested either in ELISA (fusion proteins expressed in 2gtl 1) or by immunoblotting (Pancino et al., 1993). Peptides. The TM2 domain includes a 6 amino acid sequence delimited by two cysteines that are likely to be bound in a disulphide linkage (Pancino et al., 1995a). This structure corresponds to the principal immunodominant B cell epitope of the TM during FIV infection (Pancino et al, 1993; Fontenot et al., 1992). We therefore immunized cats with a chemically cyclized synthetic peptide (PTM2c) comprising amino acids (QELGCNQNQFFCKV) covatently

3 FlY neutralization 761 SUI 253 ~ W ~ / x U I L Y N I S V I V P D Y L ~ A 3 R V D I W 290 K... S... V... SU2 387 ~ ~TFA K ~--I--P-... SU3 460 ~ I ~ I S P A N ~ C [ L - q M Y A N ~ 4 Y b I 2 ~ / ~. S SD-... N-P... S-K... SU4 498 C S L ~ L I P ~ W N K T K A ~ ( ~ N ~ S C V- - -M... SU5 572 YNPSIAGLY{QSLEKYQVVKQPDYL~RR Q-... L--E... TMI 595 w~vmn<p TVLq~C~OTCATAI(~rv'ILTYQQVrAq}{ 647 L--E... I... H-... TM2674 VIGLKVFAV~<FLYTAFAM~FgX]<VPSALWERYNVm3N~rI~ I-LE--T... TM3743 RFg~I]2431E~GL~LQEWEI3WV(TWI6~TIF~G[LGG 788 K... T-I... K-... R... TM4 810 IRNCIHKILGYT97AMPE~'~.:~,:~:~:[Q~Z~~ G-... Fig. 1. Amino acid sequence of antibody-binding domains of the envelope glycoproteins of FIV. Sequences of antibody-binding domains of FIVwo (upper line) and the corresponding domains of the 34F10 molecular clone of FIVp~t~a... (lower line) are shown, numbered from the N-terminal methionine of the Wo envelope. The antibody-binding domains of FIVwo were those used, in the form of fusion proteins, for immunization. coupled to keyhole limpet haemocyanin (KLH). PTM2c was purchased from Neosystem Laboratory (Strasbourg, France). Three overlapping peptides were used to analyse the humoral response to the SU2 domain: SU2-e, RAISSWKQRNRWEWRPD (amino acids ); SU2-fl, WEWRPDFESEK (398408); and SU2-7, WEWRPDFESEKVKVSLQC ( ). The last two peptides contain a residue thought to be involved in FIV tropism (Verschoor et al., 1995) and a group-specific epitope (Pancino et al., 1993). An unrelated 16 amino acid peptide (RRPARLLPLREHRALC) derived from the fl~-adrenergic receptor (fl3-ar) was generously supplied by M. Hattab (this laboratory). Immunization. Two outbred SPF cats were immunized with each immunogen. Cats were injected intradermally with fusion proteins in physiological saline solution ( gg 2gtl 1 or 200 gg ptex fusion proteins) emulsified in Freund's complete adjuvant (FCA). Cats received 3-5 boosts with the same amount of fusion protein in Freund's incomplete adjuvant (FIA). The immune response against the immunogen was provisionally assessed by ELISA on fl-galactosidase protein (Sigma), After two or three boosts all immunized cat sera showed antifl-galactosidase titres above ~. Two cats were injected with the KLH-coupled PTM2c (120 gg per injection) in FCA. The booster inoculations were administered in FIA. Antibody response was monitored by immunoprecipitation and flow cytometry and, in the case of anti-su2 and anti-ptm2c, by ELISA on adsorbed peptides. Radioimmunoprecipitation. ass-labelled products of in vitro transcription and translation of the env gene of FIVwo were immunoprecipitated as previously described (Pancino et al., 1993). Fully glycosylated virus envelopes derived from metabolically labelled FL-4 cells and activated feline PBMC infected with FIVwo were immunoprecipitated as described elsewhere (Pancino et al, 1995b). Sera were used at a 1 : 50 dilution. Flow cytometry. Flow cytometry was performed with an EPICS Elite flow cytometer (Coulter) after immunostaining of the chronically infected FL-4 or CrFK-Petaluma cell lines and the uninfected Fet-1 or CrFK cell lines as respective negative controls. Suspensions of CrFK cells were prepared by treating monolayers with 0-02 % EDTA. For flow cytometry of cells fixed prior to antibody binding, cells were fixed in 2 % paraformaldehyde for 10 min at room temperature (rt) before resuspension in acetone:methanol (1 : 1 v/v) for 3 rain at -20 C. Cells were subsequently incubated in PBS with 1% BSA for 30 min at rt. Immunostaining was performed by conventional methods using feline sera and fluoroisothiocyanate-conjugated goat anti-cat IgG (Kirkegaard & Perry). Flow cytometry of live cells (fixed subsequent to antibody binding) was carried out similarly; cells were generally resuspended in 1% paraformaldehyde prior to analysis. Incubations and washes of live cells were generally performed at 4 C, although reaction of primary antibody (feline sera diluted to 1:25) was sometimes carried out at 37 C to examine the effect of temperature on the accessibility of selected epitopes. Peptide ELISA and Pepscan. The reactivity of feline sera raised against the SU2 and TM2 domains was tested in ELISA with three overlapping peptides (SU2-cq SU2-fl and SU2-7) and one cyclic peptide (PTM2e) corresponding to sequences of the SU2 and TM2 domains, respectively (see above). ELISA was performed on adsorbed peptide as previously described (Avram~as et al, 1993). Pin-linked overlapping nonapeptides spanning amino acids 467 to 496 of the FIVwo envelope were purchased from Cambridge Research Biochemicals. They were used to map epitopes of the SU3 antibodybinding domain by Pepscan analysis as described elsewhere (Pancino et al. 1993). The peptides were screened with sera from two cats infected with FIVwo and two cats infected with the Envnip and Me strains of FlY (Hurtret et al, 1992; Moraillon et al., 1992). Absorbance values higher than the mean absorbance plus 1 SD were considered positive. Neutralization assay All sera used in neutralization assays were heat-inactivated at 56 C for 30 min and sterilized by filtration (0'22 lain), (i) CrFK cells'. The detection of antibodies neutralizing infection of CrFK by FIV was performed essentially as described by Osborne et al. (1994). CrFK-HO6TI/2 fibroblasts were resnspended at I x l05 cells/mt in complete DMEM and 0'5 ml was placed in each well of 48- well tissue culture plates (5 104 cells per well). The next day heatinactivated feline sera were adjusted to an appropriate dilution in complete DMEM and sterilized by filtration (0-22 ~tm). Supernatant from FL-4 cells, used as a source of FIV~,t~l~m ~, was diluted in complete DMEM. Then, 0-4 ml of diluted virus containing approximately 100 TCIDso was incubated with an equal volume of diluted serum for 1 h at 37 C. Medium was removed from adherent CrFK in duplicate wells and replaced with 350 ~tl of virus treated with antiserum. Virus and antiserum were incubated with cells tbr I h and 15 rain. Virus and antiserum were then removed and the cells were washed once with 0"5 ml complete DMEM. Three days following infection cell supernatants were removed. Cells were fixed with acetone:methanol (50:50 v/v) prior to immunocytochemistry, (ii) PBMC. Dilutions of serum and virus stock were prepared in complete RPMI with 2-ME and HEPES; 100 TCID~o of virus in 0.25 ml was mixed with an equal volume of appropriately diluted serum in duplicate wells of 48-well tissue culture plates and incubated for 1 h at 37 C. Feline PBMC were adjusted to cells/ml in complete RPMI with 2-ME, HEPES and 200 U/ml IL-2, and 0.5 ml of the suspension (1 x 106 cells) was added to the wells. Half the medium was replaced 5 days later and duplicate aliquots of 10 ~tl were ordinarily removed after 7 days for analysis of reverse transcriptase (RT) activity. In order to analyse the degree of neutralization occurring over time, duplicate aliquots of supernatant were removed daily from 5 to 10 days after initiation of the cultures.

4 762 J. Richardson and others Immunohistochemical detection of p24. Infected cells were detected essentially as described by Fevereiro et al. (1991) using the anti-p24 monoclonal antibody 1E2, biotinylated goat anti-mouse immunoglobulin (Dako) and peroxidase-conjugated streptavidin (Dako). Detection of RT activity. RT activity in cell supernatants was assayed conventionally. Quantitative densitometry of autoradiography film (Amersham) was performed with the software program NIH hnage V1.54 (NIH, Bethesda, Md., USA). Inhibition of neutralization in CrFK with synthetic peptides. Neutralization of the Petaluma strain of FIV was performed in the presence of overlapping synthetic peptides from the SU2 domain in order to localize the epitope(s) sensitive to neutralizing antibodies. Sera from immunized or naturally infected cats were diluted so as to reduce the number of infected cells by approximately 95 % and pre-incubated with various concentrations of SU2 domain or irrelevant peptides in a volume of 0-4 ml for 1 h at 37 C. An equal volume of virus (100 TCIDs0 ) was added and incubated for 1 h at 37 C. Subsequent steps conformed to the ordinary neutralization assay. Results Characterization of feline antisera raised against the antibody-binding domains of FIVwo envelope glycoproteins Antisera were raised in SPF cats against nine domains of the envelope glycoproteins of FIV eliciting antibodies during natural infection. The induction of antibodies recognizing the virus envelope was ascertained by immunoprecipitation of radiolabelled envelope glyco- proteins, either the product of in vitro translation or biosynthetic labelling of infected cells, and by flow cytometric analysis of fixed, infected cells (Table 1). All sera, with the exception of those directed against the SU3 domain, efficiently immunoprecipitated the envelope precursor derived from in vitro translation. Regarding the immunoprecipitation of fully processed virus envelope derived from infected cells, sera raised against three of five antibody-binding domains (SU2, SU4 and SU5) in the SU subunit immunoprecipitated the envelope glycoprotein of both FIVretal,~m a and FIVwo from lysates of chronically infected FL-4 cells and infected PBMC, respectively. Sera raised against two of four TM domains (TM3 and TM4) precipitated virus envelope fi'om infected cells, although we observed precipitation of only the FIVwo glycoprotein by the serum specific for TM3. In flow cytometry all sera raised against the TM subunit and sera raised against three of five SU domains (SU1, SU4 and SU5) bound to the majority of fixed FL-4 cells. Thus, with the exception of anti-su3, all sera were demonstrated to recognize mature envelope glycoproteins produced by infected cells, whether by immunoprecipitation, flow cytometry or both. It may be noted that sera specific for the SU1, TM1 and TM2 antigenic sites bound to fixed and permeabilized FL-4 cells but did not immunoprecipitate virus envelope derived from infected cells. The treatment of Table 1. Characterization of feline antisera raised against antibody-binding domains of FIVwo envelope glycoproteins Immunoprecipitation'~ Biosynthetic labelling Flow cytometry~ In vitro Antiserum* translation FL-4 PBMC FL-4 Fet-1 SU SU '7 0.5 SU3 _ SU SU TM TM TM TM PTM2c NT NT NT NT Control Control "5 * Data are given for one representative feline antiserum of each specificity. t Immunoprecipitation was performed on radiolabelled envelope as the product of in vitro translation of the entire env gene or biosynthetic labelling of FL-4 and PBMC cells infected with FIVpet~u,n ~ and FIVwo, respectively. Immunoprecipitation, as assessed by the autoradiographic intensity of precipitated bands, was classified as strong (+ + ), moderate (+), weak (_+) or negative (--). NT, Not tested. Flow cytometry was performed on fixed FL-4 and Fet-1 cells. The results are expressed as the percentage of cells contained in the selected window of fluorescence intensity, The positive control sera were those of cats experimentally infected with FIVwo and FIVre~lum ~, for immunoprecipitation and flow cytometry, respectively. The negative control consisted of autologous preimmune feline sera or pooled non-immune feline sera for immunoprecipitation and flow cytometry, respectively.

5 FIV neutralization 763 (a) 1-2 (b) ]-o 0"8 -* 0" "2,! !!!!!! 0"0 I t!! I I I I I i (e) 1.2 i i i i w l i I1 12!!!! I I! I cells with acetone:methanol prior to flow cytometry presumably disrupted higher order structure of the envelope glycoproteins and permitted binding to continuous epitopes, while the mild solubilization technique with non-ionic detergent used prior to immunoprecipitation may have resulted in the retention of elements of secondary and tertiary structure. This would thereby block access to epitopes in the interior of the native structure or prevent the unfolding of sites that are recognized only in linear form. Since feline antisera raised against the SU3 domain expressed as an insertion in the fl-galactosidase protein encoded by 2gtll had poorly recognized the FIV envelope, two additional cats were immunized with the SU3 domain expressed as a fusion protein by the ptex vector. Sera from these cats equally failed to immunoprecipitate homologous and heterologous envelope glycoprotein from infected PBMC and FL-4 cells, respectively (data not shown). We concluded that the SU3 domain, expressed as these fusion proteins, is unable to induce high titres of antibody specific for epitopes present on the full-length glycoprotein. To investigate the immunogenicity of the SU3 domain of FIVwo in greater detail and to localize the epitope sequence, we examined this region by Pepscan (Fig. 2). Neither sera from cats infected with heterologous isolates (Me and Envnip) nor sera from SPF cats recognized any peptides. Two sera from cats infected with FIVwo that strongly bound the SU3 domain expressed as a fusion protein in phage blot assays (Pancino et al., 1993) reacted with peptides located at both the N and C termini of the sequence. However, while linear B cell epitopes should be represented in at least two consecutive nonameric sequences, positive absorbance values were obtained for only two non-overlapping peptides, thus precluding the unambiguous identification of a linear epitope. It seems likely that particular amino acid residues in the two bound peptides participate in the formation of a discontinuous epitope or epitopes whose structure may be only partially reconstituted by short linear sequences. The inability of the fusion proteins to generate antibodies which recognize the native protein is also consistent with the contribution of higher order structure to the immunogenicity of the SU3 domain Peptides Fig. 2. Pepscan analysis of the SU3 domain of FIVwo. Nonapeptides staggered by two amino acids and spanning amino acids 467 to 496 of the FIVwo envelope sequence were screened with sera from cats infected with the Wo (a), Envnip (b) and Me strains of FIV or with serum from an uninfected SPF cat (c). Absorbances above the dotted line in (a), denoting one SD above the mean absorbance, were considered positive. Positive reactions were not observed in a heterologous reaction with anti-envnip and anti-me sera, Pepscan with anti-me serum is not shown. The amino acid sequence of the region ofwo SU3 used for Pepscan and the corresponding sequences of Envnip and Me were as follows: Wo 467 NTSLIDTCGETQNVSRANPVDCTMYANRMY Envnip... N--K--G... S-K-- Me -I... TNP--TG... K--T--

6 764 J. Richardson and others Table 2. Neutralization of FIVpe, t,z~m ~ during infection of CrFK cells' Antiserum* Infection? PercentageS: SU1 A 75.5 (19-1) 103 B 60.0 (13-6) 82 SU2 7-0 (4-1) 10 SU3 A 69.3 (21-4) 95 B 90.8 (8-4) 124 SU4 A 55.5 (17-7) 76 B 83"5 (13'8) 114 SU5 A 57.5 (13-9) 79 B 55'0 (12'5) 75 TMI A 44.5 (15.3) 61 B 50"3 (15'5) 69 TM2 A 56"5 (17.4) 77 B 65-8 (11"4) 90 TM4 A 67"0 (29-8) 92 B 60"3 (12.1) 82 PTM2c Cyclic 78.3 (20.3) 107 FIVvet~... ]1 0 (0) 0 Non-immune A 66'3 (16"9) B 80'3 (5.7) * Data are given for two sera (A and B) of each specificity, except anti-su2 and anti-ptm2c, where data for one serum are shown. The serum of the second cat immunized with the SU2 fusion protein could not be used for comparison since the cat was lost to follow-up after the second booster inoculation. Serum from a third cat immunized with the SU2 fusion protein was tested in a separate set of experiments: residual infection of 1% was obtained at a serum dilution of 1/20. Anti-TM3, tested in separate experiments, did not neutralize infection of CrFK by FIVr,~t~m a (data not shown). t Mean (SteM) of number of infected cells in quadruplicate wells in the presence of immune or non-immune feline sera at dilutions of 1/20. :~ Infection in the presence of immune sera expressed as a percentage of infection occurring in the presence of non-immune sera. 11 Serum of cat experimentally infected with FIV~,~t~um ~. The specificity of feline antisera raised against the two envelope domains most frequently recognized in natural infection, SU2 and TM2 (Pancino et al., 1993; de Ronde et ai., 1994), was further analysed by ELISA on synthetic peptides. The serum raised against the SU2 fusion protein bound three overlapping peptides derived from the SU2 region (SU2-~, SU2-fl and SU2-7) at a titre of 1 : However, the sera raised against the TM2 fusion protein reacted at low titre (1:100-1:270) with its cognate TM2 peptide and a chemically cyclized peptide, PTM2c, designed to mimic the presumed structure of the conserved immunodominant epitope in the TM2 domain. We therefore immunized cats with the PTM2c peptide covalently coupled to carrier protein. The sera thus obtained contained high titres of antibodies recognizing the cyclized peptide (~< 10-~). It is worthy of note that the sera generated against the PTM2c peptide, unlike those raised against the TM2 domain fused to flgalactosidase, immunoprecipitated the envelope glycoprotein precursor from infected cells (Table 1), thus suggesting that in the fusion protein the natural immunodominant epitope is not efficiently presented. Virus neutralization in CrFK cells Feline sera raised against the nine antibody-binding domains of the envelope glycoproteins of FIV were assayed for antibodies neutralizing infection of CrFK by FIVpet~l~ma. Neutralization was initially examined at a serum dilution of 1:20. Only one sermn, that directed against the SU2 domain of FIVwo, reduced the number of infected cells to a considerable extent (Table 2). Serial dilution of the anti-su2 serum revealed a modest titre of neutralizing antibodies in comparison with neutralization mediated by a serum from a cat experimentally infected with FIVpet~um~: the last doubling dilution reducing infection by at least 50 % was 1:160 for anti- SU2 but more than 1 : for infected serum (Fig. 3 a). The substantially higher efficiency of neutralization mediated by polyspecific sera, in relation to that mediated by anti-su2 serum, was confirmed using several sera from naturally and experimentally infected cats (data not shown). Inhibition q]" virus neutralization in CrFK cells by synthetic peptides In order to localize epitopes involved in neutralization, we measured the degree of residual neutralization mediated by feline antibodies, whether present in anti- SU2 serum or in homologous infected cat serum, in the presence of different concentrations of three peptides derived from the SU2 domain of FIVwo The peptides were observed to have no direct effect on infection of CrFK by FIVpeta 1... at the concentrations employed (data not shown). As shown in Table 3, a high degree of neutralization ranging from 89 to 99 % was observed for monospecific antibodies directed against the SU2 domain, as well as for polyspecific antibodies from an infected cat, in the presence of peptides representing the C terminus of the SU2 domain (SU2-fl and SU2-?0 and an irrelevant peptide (fl3-ar). The peptide SU2-0~, however, considerably inhibited neutralization mediated by monospecific antibodies directed against the SU2 domain. This 17 amino acid peptide derived from the FIVwo sequence is perfectly conserved in the 22 amino acid FIVpet~l... peptide (V3-3) used to induce neutralizing feline sera (Lombardi et al., 1994) and monoclonal antibodies (Lombardi et al., 1995). While inhibition of anti-su2 was marked even at a concentration of 0.25 gm (residual neutralization of 16%), neutralization mediated by serum from an experimentally infected cat was only modestly inhibited (residual neutralization of 80 %) even at the highest concentration (25 gm) of SU2-~. It is unlikely that the relative inefficiency of inhibition of neutralization mediated by a highly neutralizing serum can be attributed to an extremely elevated level of

7 FIV neutralization 765 (a) ,...,![ i (b) (c) O0,,, 1 O0,,,, ! I ~ ~'~ 80~.~o L I ~ 250 ~ 0 80, Time post-infection (days) , '1 8o I(n) Reciprocal of serum dilution Fig. 3. Neutralization of infectivity of FIVp~t~mm ~ for CrFK and PBMC cells. (a) Titration in CrFK of sera from cats immunized against the SU2 domain (i) or experimentally infected with FIVv~t~m~ ~ (ii). The number of infected CrFK cells following infection by FIVF~t~ ~... preincubated with serial dilutions of immune (solid bars) or non-immune serum (hatched bars) is compared, (b) Titration in PBMC of sera from cats immunized against the SU2 domain (i) or experimentally infected with FIVpet~l,,m ~ (ii). Percentage infection represents RT activity occurring in cell supernatants in the presence of immune sera as a percentage of activity detected in the presence of autologous preimmune serum (i) or pooled preimmune serum (ii). (e) The influence of the incubation period on the apparent degree of neutralization mediated by feline anti-su2 domain serum (hatched bars) and serum of a cat experimentally infected with FIVF~t~ ~... (solid bars). Anti-SU2 and infected cat serum were applied at dilutions of 1/10 and 1/80, respectively. Percentage infection is expressed relative to pooled preimmune serum. Table 3. Inhibition of neutralization of FIVp~az~m. during infection of CrFK cells Peptide* Serum ~tm SU2-c~ SU2-fl SU2-y fl3ar SU2t _5.9 (38%) 6-3_+_5-0 (91%) (93%) (96%) _+17.0 (12%) 1.8_+1.5 (97%) 7.3_+5-1 (89%) 4-3+_3.9 (94%) _ (16%) 3"3+_4-0 (95%) (94%) (92%) FIV~,~t,lum,~ (80%) 1.8_+1-7 (99%) (99%) 3-3+_3-4 (98%) (93%) 2" (98%) (97%) (98%) _2.9 (97%) 3-8+_3-9 (97%) (98%) (99%) Non-immune Non-immune No serum * Residual neutralization is expressed as the average number of infected ceils in quadruplicate wells ±SEM, and in parentheses, as the percentage neutralization, where 100 % infection is defined as that occurring in the presence of the same dilution of preimmune serum. Anti-SU2 was used at a dilution of 1/20, corresponding to approximately 95 % neutralization. :~ Serum from a cat experimentally infected with FIWpetamrn~ was used at a dilution of 1/6000, corresponding to approximately 95 % neutralization. Dilution 1/20. Dilution 1/6000.

8 766 J. Richardson and others neutralizing antibodies directed against SU2. First, the two sera were observed to contain very similar titles of antibodies directed against the SU2 region, as measured by ELISA on synthetic peptides (data not shown). Second, to achieve a degree of neutralization of approximately 95 %, the infected cat serum had to be diluted much further than the monospecific anti-su2 serum: 1 : and 1:40, respectively. The simplest explanation for the high degree of neutralization mediated by infected serum despite the presence of elevated concentrations of V3 region peptides is the existence of neutralizing antibodies directed against another epitope(s). We have in fact observed that certain highly neutralizing sera do not recognize the SU2 domain in Western blots (C. Brito, unpublished observation). Virus neutralization in PBMC We then examined virus neutralization in a second model of infection, feline PBMC, which permitted the study of a primary FIV strain unadapted to infection of CrFK cells, the French isolate Wo, whose envelope was used for induction of antibodies. In developing the neutralization test in PBMC, we examined whether neutralization was more efficient when the cells were washed following incubation with virus and neutralizing antibody and then resuspended in the absence of neutralizing antibody or when virus and neutralizing antibody were left in contact with the cells. As reported by Baldinotti et at. (1994) for neutralization assays performed with the lymphoid cell line MBM, neutralization was more effective when virus and neutralizing antibody remained in the culture supernatant. Feline sera raised against antibody-binding domains of the envelope glycoproteins of FIVwo were assayed for antibodies neutralizing infection of PBMC by FlVpetalnma and FIVwo. Neutralization was initially examined at a serum dilution of 1:10. In relation to a pool of nonimmune sera, the serum directed against the SU2 domain considerably reduced the RT activity in the supernatant of cells infected with FIV~eta a... and to a lesser degree with FIVwo (Table 4). Conversely, RT activity was augmented in the presence of most of the other monospecific sera. It is worthy of note that in this experiment homologous infected cat serum neutralized FIVpetal~m~ but not FIVwo. We have in fact observed neutralization of FIVwo with a few homologous infected sera, while FIVveta~ma was wholly or partially neutralized by most homologous infected sera. It is possible that FIVwo, having been passaged a limited number of times and only in feline PBMC, has retained the resistance to neutralizing antibodies which may be characteristic of primary isolates of HIV-1 (Moore et al, 1995; Wrin et al., 1995; Baldinotti et al., 1994). Table 4. Neutralization of infection of PBMC by FIVwt,,,,~, and FIVwo Infectiont Antiserum* FlVpetaluma FIVwo SU1 A B 111 NT SU SU3 A B SU4 A B SU5 A I B TM1 A B TM2 A B ptm2c A NT 242 B NT 171 TM4 A B FIVpet~h~m~ ~ 10 NT FlVwo NT 122 Non-immune * Data are given for two sera (A and B) of each specificity. The serum of the second cat immunized with the SU2 fusion protein could not be used for comparison since the cat was lost to follow-up after the second booster inoculation. t Infection as percentage of RT activity in the presence of a pool of non-immune sera. Data represent the mean of duplicate wells at serum dilutions of 1/10, :ST, Not tested. :~ Serum of cat experimentally infected with FIVpetaluma. Serum of cat experimentally infected with FIVwo. Pooled non-immune feline serum. Slight variation in infection was observed when individual non-immune sera were used independently. We then compared neutralization of FIVp~t~ ~... mediated by serial dilutions of the anti-su2 sera with that mediated by its corresponding preimmune sera. As measured by reduction of RT activity, feline antibodies directed against the SU2 domain reduced infection relative to autologous preimmune serum (Fig. 3b). Neutralization of FIVretaluma during infection of PBMC was also confirmed by reduction of p24 in the celt supernatant (data not shown). The two means of measuring virus neutralization provided similar estimations of the titre of neutralizing antibodies: the last doubling dilution of anti-su2 reducing infection by at least 50% was 1: 10. This degree of neutralization seemed extremely modest in comparison with neutralization mediated by a serum from a cat experimentally infected with FIVFet~ 1... for which a dilution of 1:640 reduced infection by more than 50 % (Fig. 3b). The observation made by others that feline sera and monoclonal antibodies elicited by a V3 peptide did not neutralize virus infection of the MBM cell line (Lombardi et ai., 1995) may reflect differences in cellular substrate or assay sensitivity.

9 FIV neutralization 767 SU1 55 I SU2 55,,, SU3 55,,, 0 I SU SU5 55,,, 0f A TM1 55,,, 0 I... I = TM2 55 I I i 0 I I TM4 55 fi... 0f PTM2c TM3 [. I 0l t ~ I 10 o o c~-pet Negative I i t k. I I I 55,,, 0f 505 f.. [ l Fluorescence intensity (loglo) Fig. 4. Flow cytometry of live CrFK cells with feline antisera raised against antibody-binding domains of FIVwo envelope glycoproteins. Histograms represent analyses of uninfected CrFK cells (filled histogram) and CrFK cells chronically infected with FIVpet~ 1... (unfilled histogram). Cell count is plotted against fluorescence intensity. Feline antisera directed against envelope domains were diluted to 1/25. Analyses performed with the serum of a cat experimentally infected with FIVp~t~luma (a-pet) diluted to 1/100 and uninfected cat serum (Negative) diluted to 1/25 are included. Sera of two uninfected cats were tested; the serum yielding the highest background is shown. We also examined, for a given dilution of anti-su2, the degree of neutralization apparent over time. As shown in Fig. 3 (c), the degree of neutralization mediated by antibodies directed against the SU2 domain gradually diminished from 6 to 10 days, while that afforded by the serum of an experimentally infected cat remained stable over this period. These results may be interpreted to mean that sterilizing neutralization is difficult to achieve in PBMC. We have in fact observed that when the serum of the experimentally infected cat was not left in contact with the cells for the duration of the assay the degree of neutralization diminished over time. It is thus likely that neutralizing antibodies in infected serum controlled virus dissemination after an initial limited degree of infection. The neutralizing anti-su2 antibodies were less effective in controlling virus released after replicative cycles in PBMC cultures, perhaps as a result of greater sensitivity to an unfavourable ratio of antibody to virus or the necessity for extended contact with virus to mediate neutralization.

10 768 J. Richardson and others Epitope accessibility on the oligomeric envelope complex In order to address the insensitivity of most continuous B cell epitopes to the neutralizing action of antibodies, we examined the accessibility of such epitopes on the functional oligomeric envelope complex. By flow cytometry most antibody-binding domains appeared to be virtually inaccessible to monospecific antibodies on envelope glycoproteins of FIVpet~ 1... expressed at the plasma membrane of CrFK-Petaluma cells, as judged by comparing the binding of monospecific sera to infected and uninfected live cells (Fig. 4). However, such comparison showed that antisera directed against the SU2 domain provided a shift in fluorescence intensity which clearly exceeded the small shift observed with an uninfected cat serum (Fig. 4). Similar results were obtained with FL-4 cells (data not shown). The accessibility of this domain is likely to be only partial: of antibodies elicited by a V3 peptide, polyclonal feline antibodies but not monoclonal antibodies labelled intact infected cells (Lombardi et al., 1993, 1995). Temperature did not appear to influence the apparent degree of epitope exposure, as incubation of FL-4 cells with primary antibody at 4 C or 37 C had little effect on fluorescence intensity (data not shown). The extent of structural analogy between the envelope glycoproteins on membranes of infected cells and virions is unknown. However, the virus envelope expressed at the cell surface is functional, in so far as it is capable of inducing fusion with cells bearing appropriate receptors, and therefore should be similar in many respects to virion envelope. Thus, the accessibility of envelope epitopes on infected cells is likely to provide an approximation of epitope accessibility on the surface of virions. Discussion The suitability of ex vivo models of virus infection for the evaluation of the functional nature of antiviral antibodies is likely to depend upon both virus preparation and cellular substrate. In the present study we have analysed the neutralizing capacity of antibodies elicited by nine envelope domains for a primary isolate and a laboratory-adapted strain of FIV, in feline PBMC and CrFK fibroblasts, respectively. Moreover, neutralization of the laboratory-adapted strain was compared in the two cell types. Neutralization of primary and laboratory-adapted virus The systematic analysis of the continuous B cell epitopes of the envelope glycoproteins of primary and laboratory-adapted FIV revealed a single domain sensitive to neutralizing antibodies. This domain, SU2, is located in the V3 region of the SU glycoprotein, which has previously been shown to be sensitive to neutralizing antibodies, as assessed with laboratory-adapted strains during infection of CrFK fibroblasts (Lombardi et al., 1993; de Ronde et al., 1994) but not a lymphoid cell line (Lombardi et al., 1995). We confirm and extend these results in demonstrating that antibodies directed against this region also neutralize the infectivity of primary virus for a more natural cellular target, activated feline PBMC. Primary lentivirus isolates may differ from laboratoryadapted strains; adaptation to passage in continuously cultivated cell lines has been shown to result in phenotypic change, notably in sensitivity to neutralizing antibodies of diverse specifcity (Wrin et al., 1995; Moore et al., 1995; Sullivan et al., 1995). We have observed that neutralization of the laboratory-adapted Petaluma strain appeared to be more efficient than neutralization of the primary Wo isolate, when neutralization mediated by sera of cats infected with the homologous strain was considered (Table 4 and unpublished observations). However, we did not note large differences in the efficiency of neutralization mediated by monospecific antibodies directed against the SU2 domain. In fact, the SU2 subdomain that is the target of neutralizing antibodies, as defined by neutralization with sera directed against overlapping SU2 domain peptides and by inhibition of neutralization with synthetic peptides (Table 3), is rather well conserved among FIV isolates (Sodora et al., 1994). Antibodies which recognize this subdomain neutralize the infectivity of homologous and heterologous virus for infection of CrFK (Lombardi et al., 1994) and PBMC (Table 4). In this regard the SU2 domain of the V3 region of FIV differs from the V3 loop of HIV-I: most neutralizing monoclonal antibodies induced against the V3 loop of HIV-1 have been described as being type-specific and immunity induced by V3 loop peptides has generally afforded homologous but not heterologous protection (Girard et al., 1991; Fultz et al., 1992). Unlike the V3 loop, the ratio of synonymous to non-synonymous amino acid substitutions observed for the V3 region of FIV does not suggest that immune selection for change is operative (Pancino et al., 1993; Rigby et al., 1993). This degree of sequence conservation in an epitope relevant to protection in vivo would seem contradictory. Indeed, immunization with a V3 region peptide, although inducing high titres of antibodies neutralizing infectivity for CrFK, failed to protect against challenge with homologous virus (Lombardi et al., 1994). Functional analogies between the V3 loop of HIV-1 and the V3 region may not, however, be limited to sensitivity to neutralizing antibodies. Indeed, genetic determinants of tropism have been mapped to the V3 loop of HIV-1 (Westervelt et al., 1991; Hwang et al., 1991) and more recently to the V3 region of FIV

11 FlY neutralization 769 (Verschoor et at., 1995). The FIV determinants lie adjacent to the epitope(s) sensitive to neutralizing antibodies, delimited by inhibition of neutralization with synthetic peptides (Table 3). We have in fact observed that a monoclonal antibody elicited against a peptide (SU2-~,) containing an amino acid residue of the V3 region implicated in the determination of tropism did not neutralize FIV infection (P. Sibille & J. Richardson, unpublished observations). Neutralization and cellular context The apparent efficiency of neutralization of diverse animal viruses has been observed to depend upon cellular context (Dimmock, 1993). Regarding neutralization of FIV, neutralizing antibodies in sera of infected cats have been described as broadly cross-reactive and of high titre or isolate-specific and of low titre when residual infectivity was measured in CrFK or in a T lymphoid cell line and primary feline blasts, respectively (Tozzini et al., 1993; Baldinotti et al., 1994; Osborne et al., 1994). However, the degree of antigenic conservation in epitopes sensitive to neutralizing antibodies during infection by CrFK has probably been overestimated due to the selection of antigenicalty similar isolates for heterologous neutralization assays. Indeed, the results of Osborne et al. (1994) attest to a certain degree of antigenic diversity in neutralization epitopes among four members of the FIV env subtype A, as defined by Sodora et al. (1994), and thus to the existence of neutralization subtypes even within the same env subtype. Nevertheless, interpretation of results obtained in different cultured cells has been obscured since different virus preparations had to be used to infect the different cellular substrates. We have observed a large discrepancy in the efficiency with which the same preparation of FIVpetaluma was neutralized in CrFK and PBMC, whether neutralization was mediated by serum from an infected cat or monospecific antibodies directed against the SU2 domain. We can therefore unequivocally attribute the observed discrepancy to differences associated with neutralization in different cell types. Although the two assays employed to assess neutralization are not easily comparable, it seems unlikely that methodology alone could account for the large difference in efficiency of neutralization observed in different cultured cells. Rather, the observed disparity may reflect diversity in the mechanism of infection in different cell types, such as the use of distinct cellular receptors or accessory molecules for virus entry in different cellular contexts. In this case we might expect different envelope glycoprotein domains to be critical to infection of different cell types and thus vulnerable to neutralizing antibodies. However, monospecific antibodies directed against the SU2 domain neutralized the infectivity of FIV irrespective of the cellular substrate, albeit with greater efficiency when residual infectivity was assessed in CrFK. Reduced vulnerability of the SU2 domain during infection of PBMC could be attributable to functional redundancy in these cells. Alternatively, the degree of infection observed in the presence of antibodies may be the sum of the contributions of neutralizing and enhancing activities of antibodies, and the contribution of enhancing antibodies may be expected to depend upon cellular substrate. In fact, several monospecific sera augmented infection of FIV by two- or threefold (Table 4). The enhancement of virus infection in the presence of antibodies specific for envelope glycoproteins has previously been observed for FIV (Baldinotti et al., 1994; Lombardi et al., 1994; Siebelink et al., 1995b) as well as for numerous other viruses including HIV (reviewed in Mascola et al., 1993) and several mechanisms have been proposed (Robinson et al., 1990; Takeda et al., 1988; Homsy et al., 1989). It is noteworthy that we did not observe this phenomenon in CrFK fibroblasts, which may lack functional Fc receptors or other molecules required for antibody-mediated enhancement of infection. Further experiments in more appropriate ex vivo systems, such as in cultured feline macrophages, are needed to analyse this phenomenon. Accessibility of continuous B cell epitopes Whereas feline antibodies elicited by linear B cell epitopes of the FIV envelope generally recognized envelope glycoproteins under conditions which disrupt higher order structure (immunoprecipitation of detergent lysates and flow cytometry of cells fixed prior to antibody binding), we have found that such antibodies bound poorly or not at all to the oligomeric envelope complex at the surface of infected cells. We conclude that most linear B cell epitopes of the FIV envelope are poorly exposed on the oligomeric form of envelope glycoproteins. The induction in large quantity of feline antibodies directed against these linear epitopes during natural infection may be in part the result of a humoral response to envelope monomers released from infected cells. It is worthy of note that only the most accessible of linear antibody domains was sensitive to neutralizing antibodies. The inaccessibility of linear epitopes on envelope otigomers present at the surface of virions may underlie the failure of the remaining antibody-binding domains to elicit antibodies that neutralized virus infectivity. Recently, the neutralizing capacity of antibodies specific for the HIV-1 envelope has been correlated with their affinity for oligomeric envelope glycoproteins (Sullivan et al., 1995; Sattentau & Moore, 1995). In conclusion, most linear B cell epitopes of the FIV

12 770 J. Richardson and others envelope recognized during natural infection are inaccessible on functional oligomeric envelope glycoproteins and therefore do not contribute to virus neutralization. Moreover, antibodies directed against the SU2 domain, the sole domain containing linear epitopes sensitive to neutralizing antibodies, may only account for a small fraction of neutralizing activity in infected feline serum. These results suggest that linear B cell epitopes cannot play a major role in protection during natural infection by FIV. Similar conclusions have been drawn for infection by HIV-1 (Moore & Ho, 1992; Broder et at., 1994). Certain arguments exist, however, for the role of the SU3 antibody-binding domain, located in the fourth variable region (V4) of the SU glycoprotein, in the induction of a protective immune response. This domain is subjected to selective pressure for change (Pancino et al., 1993; Rigby et al., 1993) and a single amino acid substitution in the V4 region (position 483) was shown to confer resistance to virus neutralization (Siebelink et al., 1995 b). Pepscan analysis of the SU3 domain (Fig. 2) has suggested that recognition of this domain requires higher order structure. It remains to be seen whether conformational epitopes exposed on the virion and infected cell surface provide better targets for protective humoral responses. Confirmation of this hypothesis would have critical implications for the design of vaccines based on FIV envelope glycoproteins: strategies of expression and delivery which preserve the native conformation of the virus envelope could elicit a more effective immune response. We would like to thank R. Osborne and J. Yamamoto for providing the CrFK-HO6T1/2 and FL-4 cell lines, respectively. We are grateful to the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NtH, for providing the infectious molecular clone pfiv34tf10 contributed by J. Elder. We also thank B. Hurtrel for the gift of infected cat serum and M. Hattab for the fla-ar peptide. We thank A. Avramras for performing peptide ELISAs, K. Rosenthal- Gautho for excellent technical assistance and L. Duarte, C. Fournier and C. Melin for assisting with veterinary procedures. We are also indebted to A. Sequeira and I. Bouchaert for guidance with flow cytometric analyses. Finally, we thank M. Alizon for critically reviewing the manuscript. G.P. and J.R. were supported by grants from the Agence Nationale de Recherches sur le SIDA, the ' Ensemble contre le SIDA' Association and the Fondation pour la Recherche Mrdicale. References AVRAMI~AS, A., STROSBERG, A.D., MORAILLON, A., SONIGO, P. PANCINO, G. (1993). Serological diagnosis of feline immunodeficiency virus (FIV) infection based on synthetic peptides from Env glycoproteins. Research in Virology 144, BALDINOTTI, F., MATTEUCCI, D., MAZZETT1, P., GIANNELLI, C., BANDECCHI, P., TozztNt, F. & BENDINELLI, M. (1994). Serum neutralization of feline immunodeficiency virus is markedly dependent on passage history of the virus and host system. Journal of Virology 68, BRODER, C. C., EARL, P.L., LONG, D., ABEDON, S.T., Moss, B. & Dogs, R.W. (1994). Antigenic implications of human immunodeficiency virus type 1 envelope quaternary structure: oligomer- specific and -sensitive monoclonal antibodies. Proceedings qf the National Academy of Sciences, USA 91, DE RONDE, A., STAM, J. G., BOERS, P., LANGEDIJK, H., MELOEN, R., HESSELINK, W., LIAN, C.E., KELDERMANS, J.M., VAN VLIET, A., VERSCHOOR, E.J., HORZIN~K, M.C. & EGBERINK, H.F. (1994). Antibody response in cats to the envelope proteins of feline immunodeficiency virus: identification of an immunodominant neutralization domain. Virology 198, DIMMOCK, N.J. (t993). Neutralization of animal viruses. Current Topics in ~Iierobiology and Immunology 183, I 149. EARL, P.L., BRODER, C.C., LONG, D., LEE, S.A., PETERSON, J., CHAKRABARTI, S., DOMS, R. W. MOSS, B. (1994). Native oligomeric human immunodeficiency virus type 1 envelope glycoprotein elicits diverse monoclonal antibody reactivities. Journal of t~qrology 68, FEVEREIRO, M., RONEKER, C., LAUFS, A., TAVARES, L. ~; DE NORONHA, F. (1991). Characterization of two monoclonat antibodies against feline immunodeficiency virus gag gene products and their application in an assay to evaluate neutralizing antibody activity. Journal qf General Virology 72, FONTENOT, J. D., HOOVER, E. A., ELDER, J. H. & MONTELARO, R. C. (1992). Evaluation of feline immunodeficiency virus and feline leukemia virus transmembrane peptides for serological analysis. Journal of Clinical Microbiology 30, FULTZ, P. N., NARA, P., BARRE-SINOUSSI, F., CHAPUT, A., GREENBURG, M. L., MUCHMORE, E., KIENY, M.-P. & GIRARD, M. (1992). Vaccine protection of chimpanzees against challenge with HIV-1 infected peripheral blood mononuclear cell. Science 256, GIRARD, M., KIENY, M.-P., PINTER, A., BARRE-SINOUSSI, F., NARA, P., KOLBE, H., KUSUMt, K., CHAPUT, m., REIN}tART, T., MUCHMORE, E., RONCO, J., KACZOREK, M., GOMARD, E,, GLUCKMAN, J.-C. & FULTZ, P.N. (1991). Immunization of chimpanzees confers protection against challenge with human immunodeficiency virus. Proceedings of the National Academy of Sciences, USA 88, Ho, D. D., McKEATING, J. A., LI, X. L., MOUDGIL, T,, DAAR, E. S., SUN, N.-C. & ROBINSON, J. E. (1991). Conformational epitope on gp120 important in CD4 binding and human immunodeficiency virus type-1 neutralization identified by a human monoclonal antibody. Journal of Virology 65, HOHDATSU, T., PU, R., TORRES, B. A., TRUJILLO, S., GARDNER, M. B. & YAMAMOTO, J.K. (1993). Passive antibody protection of cats against feline immunodeficiency virus infection. Journal of Virology 67, HOMSY, J., TATENO, M. & LEVY, J.A. (1988). Antibody-dependent enhancement of HIV infection. Lancet i, HOMSY, J., MEYER, M, TATENO, M., CLARKSON, S. & LEVY, J. A. (1989). The Fc and not CD4 receptor mediates antibody enhancement of HIV infection in human cells. Science 244, HOSlE, M. G., OSBORNE, R., REID, G., NEIL, J. C. & JARRETT, O. (1992)o Enhancement after feline immunodeficiency virus vaccination. Veterinary Immunology and Immunopathology 35, HURTREL, M., GANI~RE, J.-P., GUELFI, J.-F., CHAKRABARTI, L., MAIRE, M.-A., GRAY, F., MONTAGNIER, L. & HURTREL, B. (1992). Comparison of early and late feline immunodeficiency virus encephalopathies. AIDS 6, HUYNH, T. V., YOUNG, R. A. & DAVIS, R. W. (1985). Constructing and Screening cdna Libraries in 2gtlO and 2gt11, pp Oxford: IRL Press. HWANG, S. S., BOYLE, T. J., LVERLY, H. K. & CULLEN, B. R. (1991). Identification of the envelope V3 loop as the primary determinant of cell tropism in HIV-I. Science 253, LOMBARDI, S., GARZELL1, C., LA ROSA, C., ZACCARO, L., SPECTER, S., MALVALDI, G., TOZZINI, F., ESPOSITO, F. & BENDrNELLI, M. (1993). Identification of a linear neutralization site within the third variable region of the feline immunodeficiency virus envelope. Journal of l~trology 67, LOMBARDI, S., GARZELLI, C., PISTELLO, M., MASSI, C., MATTEUCCI, D., BALDINOTTI, F., CAMMAROTA, G., DA PRATO, L., BANDECCHI, P., TOZZINI, F. & BENDINELLI, M. (1994). A neutralizing antibodyinducing peptide of the V3 domain of feline immunodeficiency virus envelope glycoprotein does not induce protective immunity. Journal of Virology 68,

13 HV neutralization 771 LOMBARDI, S., MASSI, C., TOZZINI, F., ZACCARO, L., BAZZICHI, A., BANDECCHI, P., LA ROSA, C., BENDINELLI, M. & GARZELLI, C. (1995). Epitope mapping of the V3 domain of feline immunodeficiency virus envelope glycoprotein by monoclonal antibodies. Journal of General Virology 76, MASCOLA, J.R., MATIflESON, B.J., ZACK, P.M., WALKER, M. C., HALSTEAD, S. B. & BURKE, D. S. (1993). Summary report: workshop on the potential risks of antibody-dependent enhancement in human HIV vaccine trials. AIDS Research and Human Retroviruses 9, MOORE, J.P. & HO, D.D. (1992). Antibodies to discontinuous or conformationally sensitive epitopes on the gpl20 glycoprotein of human immtmodeficiency virus type 1 are highly prevalent in sera of infected humans. Journal of Virology 67, MOORE, J. P., CAO, Y., QING, L., SATTENTAU, Q. J., PYATI, J., KODURI, R., ROBINSON, J., BARBAS, C. F., III, BURTON, D. R. & HO, D. D. (1995). Primary isolates of human immunodeficiency virus type 1 are relatively resistant to neutralization by monoclonal antibodies to gpl20, and their neutralization is not predicted by studies with monomeric gpl20. Journal of Virology 69, MORAILLON, A., BARRE-SINOUSSI, F., PARODI, A., MORAILLON, R. & DAUGtrET, C. (1992). In vitro properties and experimental pathogenic effect of three feline immunodeficiency viruses (FIV) isolated from cats with terminal disease. Veterinary Microbiology 31, NARA, P. L., GARRITY, R. R. & GOUDSMIT, J. (1991). Neutralization of HIV-I: a paradox of humoral proportions. FASEB Journal 5, NORRBY, E. & MATTHEWS, T. (1993). B-cell antigenic sites in the envelope proteins of primate lentiviruses and their role in vaccine development. AIDS 7, supplementum l, OSBORNE, R., RIGBY, M., SIEBELINK, K., NEIL, J. C. & JARRETT, O. (1994). Virus neutralization reveals antigenic variation among feline immunodeficiency virus isolates. Journal of General Virology 75, PANCINO, G., CHAPPEY, C., SAURIN, W. & SONIGO, P. (1993). B epitopes and selection pressures in feline immunodeficiency virus envelope glycoproteins. Journal of Virology 67, PANCINO, G., CAMOIN, L. & SONIC, O, P. (1995a). Structural analysis of the principal immunodominant domain of the feline immunodeficiency virus transmembrane glycoprotein. Journal of Virology 69, PANCINO, G., CASTELOT, S. & SONIGO, P. (1995b). Differences in feline immuuodeficiency virus (FIV) host cell range correlate with envelope fusogenic properties. Virology 206, Pu, R., OKADA, S., LITTt.E, E. R., XU, B., STOFFS, W. V. & YAMAMOTO, J. K. (1995). Protection of neonatal kittens against feline immunodeficiency virus infection with passive maternal antiviral antibodies. AIDS 9, RIGBY, M. A., HOLMES, E. C., PISTELLO, M., MACKAY, A., BROWN, A. J. L. & NEIL, J. C. (1993). Evolution of structural proteins of feline immunodeficiency virus: molecular epidemiology and evidence of selection for change. Journal of General Virology 74, ROBINSON, W.E., MONI~3FIORI, D, C. & MITCHELL, W.M. (1988). Antibody-dependent enhancement of human immunodeficiency virus type 1 infection. Lancet i, ROBINSON, W.E., MONTEFIORI, D.C. & MITCHELL, W.M. (1990). Complement-mediated antibody-dependent enhancement of HIV-1 infection requires CD4 and complement receptors. Virology 175, SATTENTAU, Q.J. & MOORE, J. P. (1995). Human immunodeficiency virus type 1 neutralization is determined by epitope exposure on the gpl20 oligomer. Journal of Experimental Medicine 182, SIEBELINK, K. H. J., THHAAR, E., HUISMAN, R. C., HUISMAN, W., DE RONDE, A., DARBY, I. H., FRANCIS, M. J., RIMIVlELZWAAN, G. F. & OSTFa~HAUS, A. D. M. E. (1995a). Enhancement of feline immunodeficiency virus infection after immunization with envelope glycoprotein subunit vaccines. Journal of Virology 69, SIEBELINK, K. H. J., HUISMAN, W., KARLAS, J. A., RIMMELZVv\AAN, G. F., BOSCH, M. L. & OSa~RHAUS, A. D. M. E. (1995b). Neutralization of feline immunodeficiency virus by polyclonal feline antibody: simultaneous involvement of hypervariable regions 4 and 5 of the surface glycoprotein. Journal of Virology 69, 5t SODORA, D. L., SHPAER, E. G., KITCHELL, B., DOW, S. W., HOOVER, E. A. & MULLINS, J. I. (1994). Identification of three feline immunodeficiency virus (FIV) env gene subtypes and comparison of the FlY and human immunodeficiency virus type 1 evolutionary patterns. Journal of Virology 68, STANLEY, K. K. (1988). Producing antibodies of predetermined specificity from E. coli hybrid proteins. Methods in Molecular Biology 4, STANLEY, K. K. & LUZIO, J. P. (1984). Construction of a new family of high efficiency bacterial expression vectors: identification of edna clones coding for human liver proteins. EMBO Journal 3, SULLIVAN, N., SUN, Y., LI, J., HOFMAN, W. SODROSKI, J. (1995). Replicative function and neutralization sensitivity of envelope glycoproteins from primary and T-cell line-passaged human immunodeficiency virus type 1 isolates. Journal of Virology 69, TAKEDA, A., TUAZON, C. U. & ENNIS, F. A. (1988). Antibody-enhanced infection by HIV-1 via Fc receptor-mediated entry. Science 242, TALBOTT, R.L., SPARC3ER, E.E., LOVELACE, K.M., FITCH, W.M., PEDERSEN, N. C., LUCIW, P. A. & ELDER, J. H. (1989). Nucleotide sequence and genomic organization of feline immunodeficiency virus. Proceedings of the National Academy of Sciences, USA 86, TozzINI, F., MATTEUCCI, D., BANDECCHI, P., BALDINOTTI, F., SIEBE- LINK, K., OST~RHAUS, A. & BENOINELLI, M. (1993). Neutralizing antibodies in cats infected with feline immunodeficiency virus. Journal of Clinical Microbiology 31, VERSCHOOR, E.J., BOVEN, L. A., BLAAK, H. & VAN VLIET, A. L.W. (1995). A single mutation within the V3 envelope neutralization domain of feline immunodeficiency virus determines its tropism for CRFK cells. Journal of Virology 69, VOGEL, T., KURTH, R. & NORLEY, S. (1994). The majority of neutralizing Abs in HIV-1-infected patients recognize linear V3 loop sequences. Journal of Immunology 153, WESTER~,%LT, P., GENDELMAN, H. E. & RATNER, L. (1991). Identification of a determinant within the human itnmunodeficiency virus 1 surface envelope gtycoprotein criticm for productive infection of primary monocytes. Proceedings of the National Academy of Sciences, USA 88, WRIN, T., LOH, T. P., VENNARI, J. C., SCHUITEMAKER, H. & NUNBERG, J.H. (1995). Adaptation to persistent growth in the H9 cell line renders a primary isolate of human inamunodeficiency virus type 1 sensitive to neutralization by vaccine sera. Journal of Virology 69, YAMAMOTO, J.K., ACKLEY, C.D., ZOCHL1NSKI, H., LOUIE, H., PEMBROKE, E., TORTen, M., HANSEN, H., MUNN, R. & OKUDA, T. (1991). Development of IL-2-independent feline lymphoid cell lines chronically infected with feline immunodeficiency virus: importance for diagnostic reagents and vaccines. Intervirology 32, (Received 7 September 1995 ; Accepted 14 November 1995)