Contribution of arginine residues in the RP135 peptide derived from the V3 loop of gp120 to its interaction with the Fv fragment of the 0.5beta HIV-1 neutralizing antibody.

The construction, expression, and purification of an active Fv fragment of the 0.5β monoclonal human immunodeficiency virus type 1 (HIV-1) neutralizing antibody is reported. The interaction between the Fv fragment and the RP135 peptide derived from the V3 loop of gp120 from HIV-1IIIB was studied by varying the salt concentration and by mutating arginine residues in the peptide. The mutations R4A, R8A and R11A (which correspond to residues 311, 315, and 318 in gp120 of HIV-1IIIB) reduce the binding free energy by 0.22 (± 0.20), 4.32 (± 0.16), and 1.58 (± 0.17) kcal mol−1, respectively. The salt-dependent components of their contributions to binding are 0.02 (± 0.22), −0.55 (± 0.18), and −0.97 (± 0.19) kcal mol−1, respectively. The magnitudes of the mutational effects and the extent of shielding by 1 M NaCl suggest that Arg-8 is involved in a buried salt bridge in the peptide-Fv fragment complex, whereas Arg-11 is involved in a more solvent-exposed electrostatic interaction.

The 0.5␤ monoclonal antibody was raised (1) against the envelope glycoprotein gp120 (from the strain HIV-1 IIIB ) 1 which is found on the surface of human immunodeficiency virus type 1 (HIV-1) and HIV-1-infected cells. Infection of healthy T-cells is facilitated by the binding of gp120 to the CD4 protein which is present on the surface of helper T-cells. A chimeric monoclonal antibody which contains the variable region of the 0.5␤ antibody and human constant regions was found to protect chimpanzees from HIV-1 infection after passive immunization (2). The 0.5␤ antibody binds to a sequential epitope of gp120 which corresponds to its principal neutralizing determinant. This determinant is within a disulfide-bridged loop in the third hypervariable region (V3) of gp120 (3,4). A 24-amino acid-long peptide, NNTRKSIRIQRGPGRAFVTIGKIG, derived from the principal neutralizing determinant of gp120 of HIV-1 IIIB and designated RP135, is immunogenic by itself and was shown to correspond to the binding site of gp120 to the 0.5␤ antibody (4).
Nuclear magnetic resonance (NMR) studies on the interaction of the RP135 peptide with the Fab fragment of the 0.5␤ antibody have defined a 16-residue epitope from Lys-5 to Ile-20 (5). The recently solved crystal structure of the Fab fragment of a different HIV-1 neutralizing antibody, 50.1, in complex with a 16-residue peptide derived from the V3 loop of gp120 (HIV-1 MN strain) shows that it interacts only with a sequential 7-residue epitope (6). The interaction between the Fab fragment of yet another HIV-1 neutralizing antibody, 59.1, with a 24-residue peptide is similar to that of the 50.1 Fab fragment with respect to the conformation of overlapping residues in the two peptides and the size of the epitope (7).
Recent advances in antibody technology (see, for review, Ref. 8) have made possible direct cloning of antibody genes from hybridomas or lymphocytes into plasmid vectors and their expression in bacteria (see, for review, Ref. 9). In particular, there has been much recent interest in smaller antibody fragments that still retain antigen binding activity. These include Fv fragments (10,11), single-chain Fv fragments (12,13) and Fab fragments (14). Owing to the relatively small size of Fv and single-chain Fv fragments, their structures can be determined by multidimensional NMR techniques (15), and they are expected to have improved pharmacokinetic properties (8). A single-chain Fv of the anti-HIV-1 gp120 antibody, F105, has been constructed previously (16). Here, we report the construction, expression, and purification of an active Fv fragment of the 0.5␤ monoclonal HIV-1 neutralizing antibody (1,17). An accurate assay for measuring the binding of the RP135 peptide to the 0.5␤ Fv fragment was established and used to determine the contributions to binding of arginine residues in the peptide and their salt dependence. We have focused on arginine residues in the peptide since amino acid sequence information and model building (5) suggested that electrostatic interactions are of special importance in this system. Measurement of binding constants in the presence of high salt, which masks electrostatic interactions, facilitates partitioning of the binding energy into ionic and nonionic components. Our long-term goal is to analyze in detail the energetics of this interaction by kinetic and protein engineering methods and to determine the solution and crystal structures of the 0.5␤ Fv fragment-RP135 peptide complex.

EXPERIMENTAL PROCEDURES
Materials-Molecular biology reagents were from New England Biolabs or Promega unless stated otherwise. All Fmoc amino acid derivatives used for peptide synthesis were purchased from Novabiochem, Switzerland. Molecular weight prestained markers were from Bio-Rad. Isopropyl-1-thio-␤-D-galactopyranoside (IPTG) was obtained from Chembridge Corp. All other analytical grade reagents were purchased from Sigma.
Cloning of Amplified DNA and Construction of the 0.5␤ Fv Fragment Expression Vector-The 54ЈCB1 hybridoma cell line producing the 0.5␤ monoclonal antibody was provided by Dr. S. Matsushita (Kumamoto University). Total RNA of 54ЈCB1 cells was purified using an Amer-* This work was supported by Grant 91-00288 from the United States-Israel Binational Science Foundation (BSF), Jerusalem, Israel (to A. H.) and by a grant of the Israel Science Foundation administered by the Israel Academy of Sciences and Humanities (to J. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
sham kit. The design of the PCR primers and the PCR amplification reactions were as before (18). The amplified PCR products were digested using PstI and BstEII for the V H gene and SacI and BglII for the V gene. The digested PCR products were then subcloned into a pUC19 expression vector that previously contained the genes of the V H and V domains of the D1.3 anti-lysozyme antibody, each fused to the pelB signal sequence. In this plasmid, generously donated to us by Dr. G. Winter (Medical Research Council, Cambridge), the V domain is fused at its C terminus to the myc tag peptide. The D1.3 pUC19 vector was digested with PstI and BstEII, purified to remove the DNA coding for the V H domain of the D1.3 antibody, and then religated with the PCR product of the 0.5␤ antibody V H gene. The product of this ligation reaction was then digested with SacI and XhoI, purified to remove the DNA coding for the D1.3 V domain, and then religated with the PCR product of the 0.5␤ antibody V gene. A 0.9-kb EcoRI-HindIII restriction fragment of this plasmid was then subcloned into the pTZ19U vector (19) previously digested with the same enzymes. This vector is designated pT␤. The pTZ19U vector contains an f1 origin of replication that allows production of single-stranded DNA upon infection with helper phage.
Site-directed Mutagenesis-Single-stranded DNA of the plasmid pT␤, harbored in the E. coli strain TG2, was obtained by infecting these cells with helper-phage M13KO7 (Pharmacia Biotech Inc.). Site-directed mutagenesis was carried out using the method of Eckstein (20) and the Amersham kit. The following oligonucleotides were used to correct mutations owing to the PCR oligonucleotides (18) and to reverse an additional mutation Ser-25 3 Phe that occurred for reasons not known. Glu-6(H) 3 Gln: 5Ј-CCCCAGACTG*CTGCAGCT-3Ј, Thr-114(H) 3 Ser: 5Ј-CGGTGACCGA*GGTCCCTT-3Ј, Phe-25(H) 3 Ser: 5Ј-GTGTAGCCAG*AAGCCTTGC-3Ј, Glu-3() 3 Val: 5Ј-GGGT-GAGCA*CGATGTC-3Ј, where H and indicate mutations in the heavy and light chains, respectively. An asterisk follows the mismatched bases. The myc tag peptide was removed by replacing the sequence coding for the first two N-terminal residues of the tag with stop codons using the oligonucleotide: 5Ј-CTGAGATGAGTTTTTGT-TA*T*T*A*TTTGATCTCGAGCTTGG-3Ј. The vector with these five changes is designated pT␤11.
Fv Fragment Expression and Purification-A 5-ml starter culture of Escherichia coli TG2 cells harboring the pT␤11 plasmid was grown overnight at 37°C in 2 ϫ TY medium containing 50 g/ml ampicillin and 0.1% glucose. This culture was used to inoculate 500 ml of 2 ϫ TY medium containing 50 g/ml ampicillin and 0.1% glucose. The cells were grown until their density reached A 600 ϭ 0.6, and then protein expression was induced by the addition of 1 mM IPTG. The cells were grown for another 4 h at 30°C, centrifuged, the supernatant (fraction A) was discarded, and the pellet was resuspended in 30 mM Tris-HCl buffer (pH 8.0) containing 20% sucrose, 1 mM EDTA, 0.6 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 1 g/ml leupeptin. After incubation for 20 min at room temperature, the cells were spun and the supernatant was collected (fraction B). The cell pellets were resuspended in ice-cold water containing 0.5 mM MgCl 2 , 0.6 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 1 g/ml leupeptin and incubated for 5 min on ice. After spinning the cells, the supernatant (fraction C) was collected and combined with fraction B. The combined fractions were loaded onto a Sepharose 4B (Pharmacia) column followed by an affinity column prepared by cross-linking a RP135-related peptide to CNBr-activated Sepharose 4B (Pharmacia), as described by the manufacturer. After extensive washing with PBS and 3 M NaCl, the 0.5␤ Fv fragment was eluted with 0.1 M phosphate-citrate buffer (pH 4.5), and 1-ml fractions were collected in tubes containing 50 l of 3 M Tris-HCl (pH 9.0). The fractions were pooled, concentrated using Centriprep-10 (Amicon), and dialyzed exhaustively against 50 mM phosphate buffer (pH 7.5). The protein was then aliquoted, flash-frozen in liquid nitrogen, and stored at Ϫ70°C. The concentration of the protein was determined using the method of Gill and von Hippel (21) and by quantitative amino acid analysis.
Peptide Synthesis and Purification-The antigenic peptide TRK-SIRIQRGPGRAFVTIGK and variants thereof were synthesized by the solid-phase method using a multiple peptide synthesizer (AMS 422 Abimed Analysen-Technik GmbH, Germany). ␣-Amino functional groups were protected by Fmoc. Side-chain protecting groups were as follows: arginine, N ␥ -2,2,5,7,8-pentamethylchroman-6-sulfonyl (Pmc); serine, O-t-butyl; lysine, N ⑀ -t-butyloxycarbonyl; glutamine, N ␥ -trityl. After the full chain was synthesized, the peptides were cleaved from the polymer using a mixture of trifluoroacetic acid, thioanisole and triethylsilane (9, 0.5, 0.5; v/v, respectively) at room temperature for 2 h. The cleaved peptides were precipitated with ice-cold t-butyl methyl ether and collected by centrifugation (4°C, 2000 rpm). Pellets were twice washed with ether and then centrifuged. The pellets were then dissolved in double-distilled water and lyophilized. Crude peptides were purified by preparative HPLC on a RP-18 column (7 m, Merck). Peptides were eluted from the RP-18 column with a linear acetonitrile gradient from 5 to 65% (v/v) in water with 0.01% trifluoroacetic acid at a flow rate of 4 ml/min. The molecular weights of the peptides were determined using a VG Platform mass spectrometer equipped with an electrospray ion source. 50 pmol/l solutions of the peptides were analyzed by direct injection after calibration with myoglobin. The expected (E) and observed (O) molecular masses (in daltons) were found to be in excellent agreement (wild-type: 2240 (E), 2240.06 Data Analysis-Determination of dissociation constants was achieved by directly fitting fluorescence measurements of the Fv fragment at different peptide concentrations, using Kaleidagraph (version 2.1 Synergy Software (PCS Inc.)), to the following equation for tight binding: where F is the observed fluorescence, F 0 is the fluorescence in the absence of peptide, F ϱ is the fluorescence in the presence of saturating concentrations of peptide, [Fv] T and [P] T are the total Fv and peptide concentrations, and K is the dissociation constant. Estimates (ϮS.E.) of the parameters F ϱ , [Fv] T , and K were obtained from the fits which were carried out using a fixed value of F 0 . In the case of weak binding, Equation 1 is reduced to: where [P] is the concentration of free peptide and all other notations are as before. Determination of dissociation constants in the case of weak binding was achieved by directly fitting fluorescence measurements at different peptide concentrations to Equation 1 using a fixed value for [Fv] T . Identical estimates were obtained by fitting the data to Equation 2. Standard free energies of binding were calculated from dissociation constants, as follows: where R is the gas constant and T is the absolute temperature. The coupling energies between the effects on binding of the mutations and the addition of salt were calculated, as follows: The free energies on the right-hand side of Equation 4 are for binding of wild-type or mutant peptide to the 0.5␤ Fv fragment in the presence of 0 M or 1 M NaCl, as indicated.

RESULTS
Purification of the 0.5␤ Fv Fragment-In our pT␤11 plasmid construct, both the heavy and light chains of the 0.5␤ Fv fragment are fused to the pelB leader sequence and are, therefore, secreted into the periplasmic space. The 0.5␤ Fv was purified from the periplasmic space by osmotic shock. It may be seen in Fig. 1 that most of the purified 0.5␤ Fv was released from the cells upon addition of the sucrose buffer and before the osmotic shock. The 0.5␤ Fv fragment was purified to homogeneity by affinity chromatography using a Sepharose 4B column to which a gp120-derived peptide antigen had been crosslinked. The final yield of the purified 0.5␤ Fv fragment is typically about 1 mg/liter using the expression and purification procedure described here. Correct processing and purification were confirmed by gel electrophoresis and amino acid analysis. The purified Fv fragment is stable as judged by two different assays. A linear relationship was observed between the activity of the Fv fragment and its concentration indicating that dissociation of the heavy and light chains is minimal (data not shown). In addition, repeated freeze and thaw cycles of the Fv fragment did not cause denaturation or dissociation as judged by nondenaturing gel electrophoresis (not shown).
Fluorescence Emission Spectra of the 0.5␤ Fv Fragment and Its Complex with the Peptide Antigen-Fluorescence emission spectra of the 0.5␤ Fv fragment were measured in the absence and in the presence of an excess amount of the RP135 peptide antigen (Fig. 2). As is evident from the spectra in Fig. 2, in the presence of excess antigen there is a blue-shift in max from 337 nm to 333 nm, and there is an enhancement in the fluorescence intensity at wavelengths below 349 nm and quenching above this wavelength. These changes in fluorescence were exploited in order to establish a binding assay for the peptide antigen to the Fv fragment, as described under "Experimental Procedures." Previous model building (5) showed that the framework residue Trp-47(H) is part of the potential antigen-binding site. The observed changes in fluorescence may be due to this antibody residue which may be in contact with the peptide.
Effects of Mutations in the Peptide on Binding-Arginine residues at positions 4, 8, and 11 in the RP135 peptide, which correspond to positions 2, 6, and 9 in the peptides we synthesized and to positions 311, 315, and 318 in gp120, were replaced by alanine. The dissociation constants for the interaction of these peptides with the 0.5␤ Fv fragment were determined by fluorescence enhancement titration as shown in Fig. 3. Free energies of binding were calculated from the measured dissociation constants using Equation 3 (Table I). In the absence of salt, the dissociation constant for the interaction between wildtype peptide and the Fv fragment is about 2 nM. Skinner et al.  (Table I). The changes, upon addition of 1 M NaCl, in the free energies of binding of wild-type peptide and the R4A, R8A, and R11A mutants are 1.76 (Ϯ 0.17), 1.78 (Ϯ 0.13), 1.21 (Ϯ 0.03), and 0.79 (Ϯ 0.07) kcal mol Ϫ1 , respectively. In the presence of 1 M NaCl, the mutations R4A, R8A, and R11A reduce the binding energy to the 0.5␤ Fv fragment by 0.24 (Ϯ 0.09), 3.77 (Ϯ 0.08), and 0.61 (Ϯ 0.09) kcal mol Ϫ1 , respectively. DISCUSSION Amino acid sequence information had suggested that electrostatic interactions are of special importance in the binding of the RP135 peptide antigen to the 0.5␤ Fv antibody fragment. The 20-mer peptide contains 6 positively charged residues (4 arginines and 2 lysines) and no negatively charged residues whereas the antibody's complementarity determining regions contain many negatively charged residues and only a few positively charged residues. In addition, model building (5) showed that a shallow concave groove is formed by the 6 complementarity determining regions of the antibody and by two frame- work residues (Tyr-49(H) and Trp-47(H)), and that this potential antigen-binding site contains many negatively charged side chains. Electrostatic interactions were previously demonstrated to be important in formation of other antibody-antigen complexes (24).
In order to determine the contribution of electrostatic interactions, we established a quantitative and highly accurate binding assay for the interaction of the RP135 peptide with the 0.5␤ Fv fragment. We then analyzed the effects of mutations of arginine residues in the peptide on its interaction with the 0.5␤ Fv fragment both in the absence and in the presence of 1 M NaCl (Table I). All arginine residues in the peptide were replaced except Arg-15 which was not, owing to its role in stabilizing the peptide conformation (7,25). The free energies of binding of the wild-type and all of the mutant RP135 peptides to the 0.5␤ Fv fragment are found to decrease in the presence of 1 M NaCl demonstrating the importance of multiple electrostatic interactions in this system.
The mutations R8A and R11A, but not R4A, are found to have large effects on binding of the RP135 peptide to the 0.5␤ Fv fragment. Our precise measurements are in agreement with the more qualitative findings of Okada et al. (26) that mutation of the corresponding arginine residues in gp120 affect its binding to the 0.5␤ monoclonal antibody. Okada et al. (26) also showed that mutation of these arginine residues affects virus infectivity and syncytium-inducing ability. Our results are also consistent with a previous epitope mapping study by NMR (5) which showed that the antigenic determinant recognized by the Fab fragment of the 0.5␤ antibody consists of 16 residues (Lys-5 to Ile-20). Interestingly, in most HIV isolates there is a deletion in the V3 loop of two residues corresponding to Arg-8 and Gln-7 in RP135 (27) which may explain why the 0.5␤ antibody is strain-specific.
The salt-dependent components of the contributions to the binding energy of the arginine residues mutated in this study were isolated by invoking thermodynamic cycles shown in Fig.  4. By analogy to double-mutant cycles (28), the cycles in Fig. 4 consist of two different steps: (i) a mutation and (ii) transfer from 0 M NaCl to 1 M NaCl. The coupling free energies for such cycles are calculated using Equation 4, and they reflect to what extent the effect of the mutation is salt-dependent. If the coupling energy is zero, then the effect of the mutation is saltindependent. If the coupling energy is different from zero, then the effect of the mutation is salt-dependent. Surprisingly, it may be seen from Fig. 4 that there is no correlation between the magnitude of the salt-dependent contribution to the binding energy of a given residue and its apparent overall contribution.

FIG. 4.
Thermodynamic cycles showing the coupling between effects of high salt and mutations in the peptide antigen on its binding to the 0.5␤ Fv fragment. The Fv fragment and the wild-type peptide antigen are designated by Ab and Ag(wt), respectively. Single-letter notation for amino acids is used. Free energy values are given in kcal mol Ϫ1 (Table I). The coupling free energies, ⌬G int , were calculated using Equation 4. . We can define as the ratio between the salt-dependent contribution, ⌬G int , and the apparent total contribution ( ϭ ⌬G int /⌬G app ). The values for Arg-4, Arg-8, and Arg-11 are Ϫ0.09 (Ϯ 0.99), 0.13 (Ϯ 0.04), and 0.61 (Ϯ 0.14), respectively. There is uncertainty regarding the value for Arg-4 but it is clear that the salt-dependent contribution is more dominant in the case of Arg-11 compared with Arg-8. In principle, two reasons may account for this difference: (i) incomplete shielding by salt, i.e. Arg-11 is more solvent-exposed than Arg-8 in the Fv fragment-peptide complex or (ii) Arg-8 is involved in nonelectrostatic interactions. Further structural and energetic studies now in progress are required in order to distinguish between these possibilities. The strength of exposed and buried salt bridges has been determined in other systems. In T4 lysozyme, a buried salt bridge was found to contribute 3-5 kcal mol Ϫ1 to protein stability (29) whereas solvent-exposed salt bridges in barnase were found to contribute only about 1 kcal mol Ϫ1 to its stability in the absence of high salt which had a strong masking effect on them (30). Here, both the magnitude of the mutational effect (⌬G app ϭ Ϫ4.32 (Ϯ 0.16) kcal mol Ϫ1 ) and the extent of shielding by 1 M NaCl ( ϭ 0.13 (Ϯ 0.04)) suggest that Arg-8 is involved in a buried salt bridge. The magnitude of the mutational effect in the case of Arg-11 (⌬G app ϭ Ϫ1.58 (Ϯ 0.17) kcal mol Ϫ1 ) and the extent of masking by salt ( ϭ 0.61 (Ϯ 0.14)) suggest that this residue is probably involved in a more solvent-exposed electrostatic interaction.