WO1994018232A1

WO1994018232A1 - Methods for generating broadly neutralizing anti-hiv antibodies and antigens capable of eliciting same

Info

Publication number: WO1994018232A1
Application number: PCT/US1994/001458
Authority: WO
Inventors: Albert T. Profy; Ian A. Wilson
Original assignee: Repligen Corp; Scripps Research Institute
Current assignee: Repligen Corp; Scripps Research Institute
Priority date: 1993-02-12
Filing date: 1994-02-09
Publication date: 1994-08-18
Anticipated expiration: 1995-08-12
Also published as: AU6136694A

Abstract

This invention features methods for identifying molecules which will act as antigens capable of eliciting broadly neutralizing anti-HIV antibodies, and methods for producing recombinant, broadly neutralizing anti-HIV antibodies.

Description

METHODS FOR GENERATING BROADLY NEUTRALIZING ANTI-HIV ANTIBODIES AND ANTIGENS CAPABLE OF ELICITING SAME

Background of the invention

Human Immunodeficiency Virus (HIV), the etiologic agent of Acquired Immunodeficiency Syndrome (AIDS) and related disorders, is a retrovirus which infects certain immune system cells, including T4 lymphocytes and CD4⁺ cells of the monocyte/macrophage lineage. In the absence of effective treatment, the mortality rate for AIDS patients approaches 100% (Fauci, Science 239:617, 1988).

Well over 100 HIV variants have been identified. The amino acid sequence of the HIV envelope glycoprotein gp120 is particularly variable; its amino acid sequence can vary by 20-25% from one strain to the next. In addition to strain to strain variability, there is a more subtle variation in genome sequence probably caused by the high error rate of reverse transcriptase.

Consequently any particular viral isolate consists of a cohort of quasi-species. Further, the diversity and number of quasi-species apparently differs from one HIV variant to another. There is substantial evidence that these quasi-species evolve in vivo . For example,

successive viral isolates from an infected individual reveal substantial temporal fluctuations in the

proportion of various quasi-species (Meyehans, Cell

58:901, 1989). Infected individuals initially mount a humoral and cellular immune response against HIV, and there is reason to believe that an infected individual's immune response may actually encourage viral spread and the emergence of more resistant variants (McCune et al., Cell 64:351, 1991). This extreme heterogeneity makes it difficult to generate vaccines or antibodies that will be effective against a wide range of HIV strains.

Antibodies that have neutralizing activity against the HIV virus have been proposed for treatment of HIV infection. The primary targets for neutralizing anti-HIV antibodies are within gp120, the envelope glycoprotein. The loop structure within the third variable (V3) domain of gp120 is believed to be the principal neutralization domain (PND) of gp120. Because of the extreme sequence heterogeneity of gp120 among HIV isolates, the V3 loop elicits predominately strain-specific neutralizing antibodies. Nevertheless anti-V3 loop antibodies that recognize short, highly conserved sub-sequences of the loop and are capable of neutralizing a broad range of HIV isolates have been identified (Scott et al., PCT

Publication No. WO 90/15078; Javaherian et al., Science 250:1590, 1990). Such antibodies are referred to as broadly neutralizing antibodies.

Given the importance of attacking a wide range of HIV strains, it would be very desirable to have the ability to generate a variety of broadly neutralizing antibodies and variety of antigens capable of eliciting such antibodies.

Summary of the Invention

In general, the invention features methods for identifying molecules, preferably organic molecules, which will act as antigens capable of binding to or eliciting broadly neutralizing anti-HIV antibodies. The invention also features methods for designing and

producing recombinant, broadly neutralizing anti-HIV antibodies. Both methods rely on the use of precise structural information derived from x-ray

crystallographic studies of the Fab portion two different broadly neutralizing anti-HIV antibodies, 58.2 and 59.1, each complexed with a peptide antigen, and described herein.

Specifically, these data permit the identification of the atoms in each antigen that are important for antigen-antibody binding. More importantly, the data described herein permit one to describe the precise three-dimensional arrangement of these important contact atoms. Other molecules which include atoms having a three-dimensional arrangement similar to some or all of these contact atoms are likely to be capable of binding to and eliciting broadly neutralizing anti-HIV

antibodies. The use of the structural data presented herein permits the identification of such structurally similar molecules using any of a number of commercially available computer programs.

The data presented herein also permits the identification of the atoms in antibodies 58.2 and 59.1 that are important for antigen binding. This information permits one to generate broadly neutralizing recombinant anti-HIV antibodies by engineering a suitable antibody to include these important contact atoms.

Accordingly, the invention features a method for identifying molecules capable of binding to or eliciting anti-HIV antibodies (preferably broadly neutralizing antibodies), involving (1) selecting an anti-HIV

antibody-contacting pharmacophore within the peptide antigen RP142; (2) comparing the pharmacophore to three- dimensional molecular structures in a structural database using a computer program; (3) identifying candidate molecules within the database predicted to include the pharmacophore; and (4) screen candidate molecules to identify one or more capable of binding to or eliciting HIV antibodies (preferably broadly neutralizing

antibodies).

The same method can be used employing antibody- contacting pharmacophores within the cyclic peptide AS.

In another embodiment, the invention features a method for identifying molecules capable of binding to or eliciting anti-HIV antibodies (preferably broadly neutralizing antibodies), involving: (1) selecting a region of three to seven contiguous amino acids within the central domain of the peptide antigen RP152; (2) comparing the atomic coordinates with the backbone of the selected region with the atomic coordinates of the backbones of three to seven amino acid polypeptide sequences in a protein structure database using a

computer program; (3) identifying a candidate three to seven amino acid polypeptide sequence present in the database in which the root mean square difference between the backbone atomic coordinates of the candidate peptide and the backbone atomic coordinates of the selected region of RP142 is less than about 0.5 Å; (4) in a model, sequentially replacing each amino acid side chain of the candidate peptide with an alternative amino acid side chain to create a set of substitute selected peptides; and (5) identifying preferred substituted selected peptides present in the set of substituted peptides so that the root mean square difference between all of the atomic coordinates of each of the preferred substitute selected peptide and all atomic coordinates of the central domain of RP142 is less than about 0.3Å. The preferred peptides are then screened as described herein.

In a preferred embodiment of the last-recited process, step (5) involves, in a model, sequentially replacing each amino acid side chain of the candidate peptide with an alternative amino acid side chain, and then rotating each replaced amino acid side chain of each candidate peptide on its alpha covalent bond to identify a minimum energy position.

In another aspect, the invention features a method for identifying molecules capable of eliciting anti-HIV antibodies (preferably broadly neutralizing antibodies), involving (1) selecting a region of three to seven contiguous amino acids within cyclic peptide AS; (2) comparing the atomic coordinates of the backbone of the selected region with the atomic coordinates of the backbones of throe to seven amino acid polypeptide sequences in a protein structure database using a

computer program; (3) identifying a candidate three to seven amino acid polypeptide sequence present in the database wherein the root mean square difference between the backbone atomic coordinates of the candidate peptide and the backbone atomic coordinates of the selected region of cyclic peptide AS is less than about 0.5 Å; (4) in a model, sequentially replacing each amino acid side chain of the candidate peptide with an alternative amino acid side chain to create a set of substituted selected peptides; and (5) identifying preferred substituted selected peptides present in the set of substituted peptides wherein the root mean square difference between all the atomic coordinates of each preferred substituted selected peptide and all atomic coordinates of the cyclic peptide AS is less than about 0.3A. Candidates are screened as described herein.

In a preferred embodiment, step (5) of the just mentioned aspect of the invention involves the steps of: (a) in a model, sequentially replacing each amino acid side chain of the candidate peptide with an alternative amino acid side; and (b) rotating each replaced amino acid side chain of each candidate peptide on its alpha carbon bond to identify a minimum energy position.

In another aspect, the invention features a recombinant, broadly neutralizing anti-HIV antibody, composed of a recombinant kappa light chain framework subgroup IV and human heavy chain framework subgroup II, wherein at least five CDR amino acids selected from the group consisting of Kabat light chain amino acids 27D, 92, 93, 94, 96 and Kabat heavy chain amino acids 33, 50 52, 53, 56, 57, 95 and 1001 are identical to the

corresponding 58.2 contact amino acids.

In a further aspect, the invention features a recombinant broadly neutralizing anti-HIV antibody, composed of a recombinant kappa light chain framework subgroup IV and human heavy chain framework subgroup II, wherein at least five CDR amino acids selected from the group consisting of Kabat light chain amino acids 27D, 28, 91, 92, 93, 94, and Kabat heavy chain amino acids 33, 35, 50, 52, 53, 54, 97 and 98 are identical to the corresponding 59.1 contact amino acids.

Other features and advantages of the invention will be apparent from the following description of the preferred embodiments and from the claims.

Detailed Description

Described herein is the precise molecular structure assumed by RP142, an antigenic peptide, when it is bound to the Fab of antibody 59.1, a broadly

neutralizing anti-HIV antibody, and the precise molecular structure assumed by a second peptide, cyclic peptide AS, when it is bound to the Fab of antibody 58.2, a second broadly neutralizing antibody. The structures of these peptides are unexpected based on sequence-based

structural predictions. Importantly, the atomic

structure of the antigen-antibody complexes are

described, and the atoms important for antigen-antibody binding are identified. This information permits one to improve identified antigens so that they elicit a

stronger immune response, identify other molecules capable of eliciting broadly neutralizing antibodies, improve existing broadly neutralizing antibodies, and design new broadly neutralizing antibodies.

For any given antigen-antibody interaction it is desirable to generate variant antigens which bind the antibody with higher affinity, since such variant antigens are expected to be more antigenic. Likewise it is considered desirable to generate tighter binding antibodies, since such antibodies εre thought to be more effective (e.g. have increased HIV neutralization

activity). Rational design of new antigens and

antibodies is dependent on accurate structural

information which can be used for modeling antigen- antibody interaction. Once an antigen-antibody structure has been modeled, one can predict what molecular changes are consistent with improved (or unchanged) binding.

The flexibility of peptides makes it

extraordinarily difficult to predict the preferred structure of a peptide composed of more than three or four amino acids. It is even more difficult to predict the structure assumed by a bioactive peptide when it binds to its receptor, since receptor-peptide

interactions are thought to significantly influence peptide structure. By deducing the structures assumed by two different antigens when bound to two different broadly neutralizing antibodies we have provided

important structural information which will permit the design of structurally related peptide and non-peptide antigens which are capable of eliciting broadly

neutralizing anti-HIV antibodies.

Recently developed computer-based methods which permit the identification of compounds with a desired molecular structure are currently being used to identify compounds whose structure is similar all or a part of a compound of interest. These computer-based methods fall into two broad classes: database methods and de novo design methods. In database methods the compound of interest is compared to all compounds present in a database of chemical structures and compounds whose structure is in some way similar to the compound of interest are identified. The structures in the database are based on either experimental data, generated by NMR or x-ray crystallography, or modeled three-dimensional structures based on two-dimensional data. In de novo design methods models of compounds whose structure is in some way similar to the compound of interest are

generated by a computer program using information derived from known structures and/or theoretical rules.

The success of both database and de novo methods in identifying compounds with activities similar to the compound of interest depends on the identification of the functionally relevant portion of the compound of

interest. For drugs, the functionally relevant portion is referred to as a pharmacophore. A pharmacophore then is an arrangement of structural features and functional groups important for biological activity. Similarly, one can identify one or more pharmacophores for a given antigen. In this case the pharmacophore is a group of atoms that play an important role in antibody binding (and therefore elicitation of antibodies).

The data provided herein concerning the structures of two antigen-antibody complexes permits the

identification of pharmacophores important for the elicitation of broadly neutralizing antibodies and removes the barrier to identifying other molecules capable of eliciting broadly neutralizing antibodies.

Programs suitable for generating predicted three- dimensional structures from two-dimensional data include: Concord (Tripos Associates, St. Louis. MO), 3-D Builder (Chemical Design Ltd., Oxford, U.K.), Catalyst (Bio-CAD Corp., Mountain View, CA), Daylight (Abbott Laboratories, Abbott Park, IL). Programs suitable for searching three- dimensional databases to identify molecules bearing a desired pharmacophore include: MACCS-3D and ISIS/3D

(Molecular Design Ltd., San Leandro, CA), ChemDBS-3D (Chemical Design Ltd., Oxford, U.K.), Sybyl/3DB Unity (Tripos Associates, St. Louis. MO). Programs suitable for pharmacophore selection and design include: DISCO (Abbott Laboratories, Abbott Park, IL), Catalyst (Bio-CAD Corp., Mountain View, CA), and ChemDBS-3D (Chemical

Design Ltd., Oxford, U.K.). Databases of chemical structures are available from Cambridge Crystallographic Data Centre (Cambridge, U.K.) and Chemical Abstracts Service (Columbus, OH). De novo design programs include Ludi (Biosym Technologies Inc., San Diego, CA) and

Aladdin (Daylight Chemical Information Systems, Irvine CA).

Structural Data

The coordinates for antibody 58.2 bound to peptide AS are presented in Appendix A in standard Brookhaven database format; Peptide AS is a cyclic peptide having the sequence JSIGPGRAFGZC where J and Z are connected by a hydrazone linkage and are represented by:

The light chain atoms are listed first starting with the amino terminal Asp (amino acid 1001) and ending with the carboxyl terminal Glu (amino acid 1215). The heavy chain atoms are then listed beginning with the amino terminal Asp (amino acid 2001) and ending with the carboxyl terminal Arg (amino acid 2226). The peptide atoms are listed next beginning with Gly (amino acid 3002) and ending with Gly (amino acid 3012).

The coordinates for antibody 59.1 bound to peptide RP142 (HIGPGRAFYT) are presented in Appendix B in standard Brookhaven database format. The light chain atoms are listed first starting with the amino terminal Asp (amino acid 1) and ending with th*3 carboxyl terminal Arg (amino acid 215). The heavy chain atoms are then listed beginning with the amino terminal Gin (amino acid 1001) and ending with the carboxyl terminal Thr (amino acid 1221). The peptide atoms are listed next beginning with His (amino acid 2002) and ending with Thr (amino acid 2011).

The contact data for antibody 58.2 and the AS peptide are presented in Table 1 in standard Brookhaven database notation.

The contact data for antibody 59.1 and RP142 peptide are presented in Table 2 in standard Brookhaven database notation. Thus the first column indicates the bond type (van der Waals, hydrogen bond or short hydrogen bond). The second column indicates the antibody atom involved. The amino acid are indicated first by their single letter code where the prefix "1" indicates the light chain and the prefix "2" indicates the heavy chain. The amino acid number (according to the numbering scheme above) is indicated next followed by the identification of the atom. The third column indicates the peptide amino acid involved by its single letter code (with the prefix "3" followed in the next column by the amino acid number (according to the numbering scheme above) followed by the identification of the atom involved. The last column indicates the bond length in angstroms. Thus, for the first contact listed, there is a van der Waals interaction between the gamma oxygen of serine 31 of the light chain and the alpha carbon of glycine 4 of the peptide, and the bond is 3.4δA long.

58.2 Structure

The data presented in Appendix A for antibody 59.1 bound to cyclic peptide AS show that the peptide adopts a relatively flat conformation. The GPGR sequence adopts a type I turn conformation, and additional turns are evident. The peptide is seen to lie flat in the antibody combining site with the Arg side chain buried deeply in the binding pocket. Light chain CDRs L1 and L3 are near the IGPGR sequence in the peptide, making several

hydrophobic contacts. The Phe residue of the peptide is only in contact with CDR H2, and in particular is stacked with the imidazole ring of His-53. CDRs H1 and H3 are in contact with the ends of the peptide. The buried Arg side chain is contacted by several residues including Asp-98 and Leu-100 of the light chain and Glu-99, Tyr-33, and His-36 of the heavy chain. There are thus two positive and two negative charges in the binding pocket when the peptide is bound. This structure is consistent with immunological data which suggest that the epitope for this antibody is HIGPGRAF. in which many different amino acids are tolerated at the underlined positions. The specificity for the Phe residue may be eliminated by mutating CDR H2 and the specificity for the His residue may be eliminated by mutating CDR H3.

Table 1 lists the contacts between cyclic peptide AS and antibody 58.2. Any subset of at least five and preferably at least seven contacting atoms, up to 10-15 atoms, can constitute a pharmacophore which can be used to identify structurally related compounds using any of the modeling programs listed herein or any similar structure modeling program. By a contact between atoms on the antibody and antigen is meant a hydrogen bond or a Van der Waals interaction; only contacting atoms pairs are listed in Table 2.

Once one or more suitable compounds have been identified, their structures can be modeled and docked with a computer model of antibody 58.2 (or 59.1) to look for the presence of unwanted repulsive interactions or distortions. Molecular mechanics can then be used to calculate and minimize conformational energy and maximize attractive interactions.

59.1 Structure

The data presented in Appendix B for antibody 58.2 bound to peptide RP142 show that the peptide adopts a relatively flat conformation. Only the IGPGRAFY sequence of RP142 is visible in the structure; presumably because the rest of the peptide is floppy. While this peptide, like the cyclic AS peptide, adopts several turns after the GPGR sequence, its backbone conformation differs from that of the cyclic AS peptide. The Arg side chain is buried, making contacts with Asp-98 and Asn-95 of the light chain and His-102 of the heavy chain, as well as the CDR L3 backbone. The is a unusual disulfide linkage between Cys-35 of CDR H1 and Cys-54 of CDR H2. This linkage is in close contact with the Phe side chain of the peptide.

Immunochemical studies suggest that antibody 59.1 recognizes the sequence GPGRAF. This is consistent with the present structural data. The data suggest an

explanation for the fact that 59.1 binds the HIV-III_B sequence more strongly than the HIV-MN sequence. In HIV- III_B a Arg replaces the lle of HIV-MN and it appears the positively charged Arg side chain would be in close contact with Glu-97 of CDR L3 providing an electrostatic contact not present in a complex between 59.1 and an HIV- MN peptide.

Table 2 lists the contacts between RP142 and antibody 59.1. Any subset of at least five and

preferably at least seven atoms can constitute a

pharmacophore which can be used to identify structurally related compounds using any of the modeling programs listed herein or any similar structure modeling program. Once one or more suitable compounds have been identified, their structures can be modeled and docked with a computer model of antibody 59.1 (or 58.2) to look for the presence of unwanted repulsive interactions or distortions. Molecular mechanics can then be used to calculate and minimize conformational energy and maximize attractive interactions.

Generation and Screening of Broadly Neutralizing

Antibodies

Novel prospective antigens identified using the methods described herein can be used to generate broadly neutralizing anti-HIV antibodies using standard

techniques. Described below are methods for generating antibodies using the novel antigens and screening the antibodies so produced to identify those with

particularly significant broadly neutralizing activity. Methods for preparing and analyzing antibodies directed towards the V3 loop of a particular HIV-1 isolates (HIV- MN or an HIV-MN viral variant), are also described in U.S. Application No. 07/665,306, filed March 6, 1991 and in PCT Publication No. WO 91/15078, assigned to the same assignee and hereby incorporated by reference.

Molecules to be used as immunogens can be either unconjugated or conjugated to an immunogenic carrier, e.g., keyhole limpet hemocyanin (KLH) or ovalbumin, using succinyl maleimidomethyl cyclohexanylcarboxylate (SMCC) as a conjugation agent using standard techniques. The molecules can be prepared for immunization by

emulsification in complete Freund's adjuvant or other similar methods according to standard techniques.

Antibodies may be prepared by intraperitoneal immunization of mouse strains (e.g., Balb/c, C57BL/6, A.SW, B10.BR, or B10.A, Jackson Labs., Bar Harbor, ME). Immunized mice can be given booster immunizations of the immunogen, either in an emulsification of incomplete Freund's adjuvant or in soluble form (e.g, two to three times at two to four week intervals) following the initial immunization. Mice were bled and the sera assayed for the presence of antibodies reactive with the immunogen or RP142. Mice showing a strong serological response are boosted, and (3-5 days later) spleen cells from these mice are fused with, for example, NS-1

(American Type Culture Collection, Rockville, MD,

Accession No. TIB18), SP2-0 (ATCC No. CRL8287, CRL8006), or P3.X63.AG8.653 myeloma cells incapable of secreting both heavy and light immunoglobulin chains (Kearney et al., J. Immunol . 123:1548, 1979) by standard procedures based on the method of Kohler and Milstein, (Nature 256:495, 1975).

Supernatants from hybridomas which appeared 6-21 days after fusion are screened for production of

antibodies by an ELISA screening assay using the

immunizing antigen or RP142 using standard techniques.

For example, if RP142 is used in the screening, each well of a 96-well Costar flat-bottom microtiter plate is coated with the peptide by placing a 50 μl aliquot of a PBS solution containing the peptide at a final concentration of 0.1-10 μg/ml in each well. The peptide solution is then aspirated and replaced with PBS + 0.5% BSA. Following incubation, the wells are

aspirated, washed, and 50 μl of hybridoma supernatant is added. Following incubation, the wells are washed 3 times with PBS, and then incubated with 50 μl of an appropriate dilution of goat anti-mouse immunoglobulin conjugated with horseradish peroxidase (HRP, Zymed

Laboratories, San Francisco, CA). The wells are washed again 3 times with PBS and 50 μl of ImM ABTS (2,2 azino- bis (3-ethylbenzthiazoline-6-sulfonic acid) in 0.1M Na- Citrate, pH 4.2, to which a 1:1000 dilution of 30% H₂O₂ had been added), the substrate for HRP, is added to detect bound antibody. HRP activity is monitored by measuring the absorbance at 410nm.

A peptide titration assay can be used as an initial screen to predict if a given anti-V3 loop

antibody will have strong neutralization activity. In this assay, the antibody is tested for its ability to prevent syncytia formation among gp120 expressing CD4+ cells in the presence of competitor peptide whose

sequence is derived from a V3 loop sequence. This assay can be used to test for potential neutralization activity of any anti-V3 loop antibody towards any HIV isolate by using a peptide derived from the V3 loop from the HIV isolate of interest as the competitor.

Syncytia formation is measured in the presence of an anti-V3 monoclonal antibody mixed with one or more test peptides representing V3 loop sequences of a variety of HIV isolates.

In the peptide titration assay, the test peptide, at a series of concentrations ranging from 10μM to

0.10μM, is added to anti-V3 loop antibody (at 5 times the concentration required for the 90% endpoint in an Std. SN assay, described below), incubated for 30' at 37° and then added to CEM-ss CD4+ cells expressing HIV-MN gp120. These cells express gpl20 because they are infected with a recombinant vaccinia virus that encodes the HIV-MN env gene.

The HIV gp120 envelope protein produced and presented on the surface of these cells enables them to bind to the CD4 receptor on other cells, resulting in cell fusion and the formation of syncytia. Antibodies that bind to the V3 loop of the gp120-expressing cells (anti-V3 loop antibodies) can inhibit syncytia formation. If the test peptide competes with the gp120 epitope recognized by the antibody for binding with the anti-V3 loop antibody, syncytia formation occurs. Conversely, if the test peptide does not compete with the cell surface epitope recognized by the antibody for binding with the anti-V3 loop antibody, syncytia formation in the presence of peptide is inhibited relative to syncytia formation in the absence of the peptide.

Construction of a recombinant vaccinia virus capable of expressing the full-length HIV envelope gene from a vaccinia virus promoter is described in EP

Publication No. 0 243 029, hereby incorporated by

reference. The recombinant vector pSC25, containing the HIV env gene and the lacZ gene of E. coli expressed from a second vaccinia virus promoter, and flanked by vaccinia viral sequences which together encode thymidine kinase (TK), was used to produce the recombinant virus.

A recombinant vector that contains DNA encoding an envelope gene having the specificity of the HIV-MN variant was prepared by removing a 570 bp Bglll fragment (encoding 180 amino acids) from the HIV-III_B env gene which spans the region of the VS loop in pSC25, and replacing it with the analogous Bg1ll fragment from the HIV-MN env gene. The resulting plasmid, pSCR2502, contained a hybrid envelope gene which encoded an

envelope protein having the principal neutralizing domain of the MN virus and the remainder of the env gene

sequence from the HIV-III_B envelope.

A smaller region of the HIV-MN gp160 protein can be used in place of the 180 amino acid replacement just described; e.g., DNA encoding the 36 amino acid V3 loop from any HIV strain can be inserted into the envelope- encoding DNA in place of the corresponding III_B DNA sequence. Alternatively, a recombinant could be used which contains the complete HIV-MN env gene. Multiple HIV envelope expressing strains are useful for assessing the specificity of an antibody. The recombinant vector pSCR2502 was transfected into CV-1 host cells that had been pre-infected with vaccinia virus containing an intact TK gene. The HIV envelope gene was integrated into the viral DNA by homologous recombination between the TK sequences on the vector and the TK sequences within the viral genome.

Recombinants containing the HIV envelope gene were selected by infection of TK- cells and plating on media containing bromodeoxyuridine (BUdR) and X-gal. BUdR is toxic to TK⁺ cells and thus selects for TK- recombinants; X-gal is a chromogenic substrate cleaved by the product of the lacZ gene which results in the production of blue plaques where the lacZ gene is expressed and further identifies the recombinant virus which also contains the HIV-env gene.

Two biological assays, the standard serial neutralization assay (Std. SN) and the expanded serial neutralization assay (Ex. SN), can be used to as a more refined measurement HIV neutralizing activity. These assays use reverse transcriptase (RT) activity as a measurement of viral activity. The reduction in reverse transcriptase activity under a given set of conditions is a measure of viral neutralization. A reduction in reverse transcriptase activity of 90% under a given set of conditions is a standard measure of viral

neutra1ization.

As described in more detail below, the Std. SN assay measures RT activity at a single time point 7 days post-infection. As a result, it is possible to compare a number of conditions with relatively few assays.

However, since each viral isolate has a characteristic time course of infection, the 7 day time point used in the Std. SN assay may not include the period of optimal viral replication. As result, in some instances, the Std. SN assay will not permit accurate determination of the effectiveness of the added anti-HIV agents. The Ex. SN assay measures RT activity at several timepoints out to 15 (or 20) days post-infection and thus is more likely to include the period of optimal viral activity for any given viral isolate. Therefore, the Ex. SN assay is preferred.

Three HIV isolates are commonly used in the Std. SN and Ex. SN assays: HIV-MN, HIV-IIIB and HIV-Ala. HIV- Ala is considered a relevant field isolate because it has had low passage in CEM cells. (The sequence of the HIV- Ala V3 loop has been reported to be the most

representative of North American HIV isolates.) The viruses used in this assay can be propagated in H9 cells (ATCC, Rockville, MD; or AIDS Research and Reference Reagent Program, Rockville, MD) for 15-30 days to

establish a chronic cell line. Newly formed virions are harvested from the supernatant of infected cells and used to infect test cultures as described below.

Serial two-fold dilutions of a anti-V3 loop antibody are incubated with 64 infectious units of HIV virus for 30 minutes at 37°, and then added to CEM-ss cells (50,000 cells per well in 96-well plates). After 7 days, cell free supernatants are harvested and assayed for RT activity using the method of Willey et al. (J .

Virol . 62:139, 1988). For determination of 90% and 50% endpoints (i.e., reduction in RT activity by 90% or 50%), densitometry readings of autoradiographs can be generated at 410 nm using a Molecular Devices microplate reader.

The Ex. SN assay is identical to the Std. SN assay except that media is replenished twice during the course of the assay (at day 7 post-infection and at day 12 post- infection). Aliquots are assayed for RT activity (as described above) at 7, 12 and 15 days post-infection.

An Infectivity Reduction Assay (IRA), which measures the difference between the infectious dose of a virus in the presence and absence of a standard dilution of an anti-HIV agent, can be used as an even more

stringent test of neutralization activity. In contrast to the Std. SN and Ex. SN assays described above, IRA conditions promote cell division and virus replication, and thus it is apt to more closely predict neutralization potential in vivo .

Both laboratory HIV strains and HIV field isolates are tested in the IRA using either CEM-ss cells or PBMC (peripheral blood mononuclear cells, prepared by standard methods), which are more susceptible to infection thought to be more like in vivo situations. Briefly, the IRA is performed as follows. Viral isolates are serially diluted 10-fold in RPMI containing 10% fetal bovine serum. For each dilution of virus, approximately 20 μg of antibody is then inoculated onto 1 x 10⁶ CEM-ss or PBMC in a 24-well plate. The cultures are maintained for the appropriate number of days (calibrated for the virus used, for example, HIV-MN is 21 days and 14 days for Duke 6587-5. Cultures are split 3 times a week (if PBMCs are used, they are supplemented with IL-2 three times a week to maintain optimal virus replication conditions). Virus replication (infectious units) is monitored by reverse transcriptase activity as described above. The

infectious titer of virus plus the antibody is compared to the infectious titer of virus plus media.

Engineered Antibodies

The data described herein can be used to generate recombinant antibodies simply by engineering a

recombinant antibody to have a suitable framework

(preferably subgroup IV or III for the kappa light chain and subgroup II for the heavy chain) and CDRs that

include, at the relevant positions some or all of the contact amino acids identified in antibody 58.2 or 59.1. Tables 3 and 4 list the Kabat amino acid numbers corresponding to the identified contact amino acids in antibodies 58.2 and 59.1 respectively. This information can be used to identify vthich amino acids should be exchanged for contact amino acids.

Since, for the most part, monoclonal antibodies are produced in species other than humans, they are often immunogenic to humans. In order to successfully use antibodies in the treatment of humans, it may be

necessary to create chimeric antibody molecules wherein the antigen binding portion (the variable region) is derived from one species, and the portion involved with providing structural stability and other biological functions (the constant region) is derived from a human antibody. Methods for producing chimeric antibodies in which the variable domain is derived from one species and the constant domain is derived from a second species are well known to those skilled in the art. See, for

example, Neuberger et al., WO Publication No. 86/01533, priority September 3, 1984; Morrison et al., EP

Publication No. 0,173,494, priority August 27, 1984. An alternative method, in which an antibody is produced by replacing only the complementarity determining regions (CDRs) of the variable region with the CDRs from an immunoglobulin of the desired antigenic specificity, is described by Winter (GB Publication No. 2,188,638, priority March 27, 1986). Murine monoclonals can be made compatible with human therapeutic use by producing an antibody containing a human Fc portion (Morrison, Science 229:1202, 1985). Single polypeptide chain antibodies are also more easily produced by recombinant means than are conventional antibodies. Ladner et al. (U.S. Patent No. 4,946,778) describes methods for producing single

polypeptide chain antibodies and these methods may be adapted to produce antibodies useful in the methods and compositions of the invention. Established procedures would allow construction, expression, and purification of such a hybrid monoclonal antibody. Quadromas can be used to generate bispecific antibodies (Reading et al., U.S. Patent Nos. 4,474,893 and 4,714,681). The patents and publications referred to herein are hereby incorporated by reference.

TABLE 3: CONVERSION FOR 58.2 NUMBERS TO KABAT NUMBERS

Antibody Amino Acid Kabat Number

TABLE 4: CONVERSION FOR 59.1 NUMBERS TO KABAT NUMBERS Antibody Amino Acid Kabat Amino Acid Number

Claims

1. A method for identifying molecules capable of binding to or eliciting anti-HIV antibodies, said method comprising:

(1) selecting an antibody-contacting pharmacophore within peptide antigen RP142;

(2) comparing said pharmacophore to three- dimensional molecular structures in a structural database using a computer program;

(3) identifying candidate molecules present within said database predicted to include said pharmacophore; and

(4) screening said candidate molecules to identify one or more said molecules capable of binding to or eliciting anti-HIV antibodies.

2. The molecules identified by the method of claim 1.

3. A method for identifying molecules capable of binding to or eliciting anti-HIV antibodies, said method comprising:

(1) selecting an anti-HIV antibody-contacting pharmacophore within cyclic peptide AS;

(3) identifying candidate structures present within said database predicted to include said

pharmacophore; and

4. The molecules identified by the method of claim 3.

5. A method for identifying molecules capable of binding to or eliciting anti-HIV antibodies, said method comprising:

(1) selecting a region of three to seven contiguous amino acids within the central domain of peptide antigen RP142;

(2) comparing the atomic coordinates of the backbone of said selected region of RP142 with the atomic coordinates of the backbones of three to seven amino acid polypeptide sequences in a protein structure database using a computer program;

(3) identifying a candidate three to seven amino acid polypeptide sequence present in said database wherein the root mean square difference between said backbone atomic coordinates of said candidate peptide and said backbone atomic coordinates of said selected region of RP142 is less than about 0.5 Å;

(4) in a model, sequentially replacing each amino acid side chain of said candidate peptide with an

alternative amino acid side chain to create a set of substituted selected peptides; and

(5) identifying preferred substituted selected peptides present in said set of substituted peptides wherein the root mean square difference between all said atomic coordinates of each said preferred substituted selected peptide and all atomic coordinates of said central domain of RP142 is less than about 0.3 Å.

6. The method of claim 5 wherein step (5) comprises the steps of: (a) in a model, sequentially replacing each amino acid side chain of said candidate peptide with an alternative amino acid side chain; and

(b) rotating each replaced amino acid side chain of each said candidate peptide on its alpha carbon bond to identify a minimum energy position.

7. The molecules identified by the method of claim 5.

8. A method for identifying molecules capable of binding to or eliciting anti-HIV antibodies, said method comprising:

(1) selecting a region of three to seven contiguous amino acids within cyclic peptide AS;

(2) comparing the atomic coordinates of the backbone of said selected region of cyclic peptide AS with the atomic coordinates of the backbones of three to seven amino acid polypeptide sequences in a protein structure databases using a computer program;

(3) identifying a candidate three to seven amino acid polypeptide sequence present in said database wherein the root mean square difference between said backbone atomic coordinates of said candidate peptide and said backbone atomic coordinates of said selected region of cyclic peptide AS is less than about 0.5Å;

(5) identifying preferred substituted selected peptides present in said set of substituted peptides wherein the root mean square difference between all said atomic coordinates of each said preferred substituted selected peptide and all atomic coordinates of said cyclic peptide AS is less than about 0.3Å.

9. The method of claim 8 wherein step (5) comprises the steps of:

(a) in a model, sequentially replacing each amino acid side chain of said candidate peptide with an alternative amino acid side; and

10. The molecules identified by the method of claim 8.

11. A recombinant broadly neutralizing anti-HIV antibody, comprising a recombinant kappa light chain framework subgroup IV and human heavy chain framework subgroup II, wherein at least five CDR amino acids selected from the group consisting of Kabat light chain amino acids 27D, 92, 93, 94, 96 and Kabat heavy chain amino acids 33, 50, 52, 53, 56, 57, 95 and 1001 are identical to the corresponding 58.2 contact amino acids.

12. A recombinant broadly neutralizing anti-HIV antibody, comprising a recombinant kappa light chain framework subgroup IV and human heavy chain framework subgroup II, wherein at least five CDR amino acids selected from the group consisting of Kabat light chain amino acids 27D, 28, 91, 92, 93, 94, and Kabat heavy chain amino acids 33, 35, 50, 52, 53, 54, 97 and 98 are identical to the corresponding 59.1 contact amino acids.