INTRODUCTION

Glycoprotein 120 (gp120)¹ is the envelope protein on the surface of HIV that binds to CD4 cell surface receptors in the first step of a cascade of events leading to HIV infection (1, 2, 3). The region of gp120 that is involved in binding to CD4 has been located to the C4 domain in the parent gp120 (4, 5), but several other areas have been implicated as well (6, 7, 8). Peptides derived from the C4 domain have not been shown to bind CD4 in a way that competes with the binding of gp120 to CD4. Thus, the notion that CD4 is binding to a single locus on gp120 is an unsettled controversy.

In a recent publication, we reported the results of an investigation in which we tested the ability of a helical immunogen derived from C4 of gp120 to induce an immune response against gp120 and we compared that with the response against the unconstrained immunogen (9). The results from that study indicated that the C4 probably exists as an amphipathic -helix in the intact gp120. Although the possibility of C4 being a helix was mentioned in 1987 by Cease et al. (10), there appeared to be little mention of the association of the helix with CD4-binding ability. The key exception to this is the extensive work of Reed and Kinzel and others from that lab who have been working on C4 peptides that were not conformationally constrained in physiological buffer systems (11, 12, 13, 14). These efforts have been useful in providing an understanding of the behavior of C4-derived peptides in various defined conformation-inducing environments.

For the present work, we studied the ability of the amphipathic -helical constructs, used as immunogens previously (9), for their ability to bind CD4. In this conformation, we found the CD4 binding to be comparable with intact gp120 and we found that a C4 peptide could block CD4 binding to gp120 provided the peptide was in a helical conformation. Thus, these findings will provide an understanding of the molecular mechanism used by gp120 to bind to CD4, and the information will be useful in designing therapeutics and vaccines to block HIV infection in vivo.

MATERIALS AND METHODS

RsCD4 and rgp120_IIIB were gifts from Genentech, Inc. (South San Francisco, CA). The chemicals used for peptide synthesis were from Applied Biosystems, a Division of Perkin-Elmer, Inc., Foster City, CA. Peptomers and peptides were synthesized and analyzed for helical content as described previously (9). Rabbit antipeptomer antibodies were produced as described previously (9). The concentrations of CD4, biotinylated CD4, gp120, and all of the peptides and peptomers used in this study were determined by quantitative amino acid analyses using the Picotag^® amino acid analysis system (Waters Associates, Millford, MA). A bovine serum albumin conjugate of peptide-(419-436) was synthesized as described previously for making conjugates of bovine serum albumin from chloroacetyl-derivatized peptides (32). Quantitation of the amount of S-carboxymethylcysteine obtained from amino acid analysis of the conjugate gave a value of 14 for the number of moles of peptide covalently linked to 1 mol of bovine serum albumin.

CD4 and gp120 were denatured by treating the individual proteins at 1 mg/ml with 10 µg of dithiothreitol in TBS (0.01 M Tris, pH 7.4, 0.15 M NaCl) for 15 min at 100 °C. 20 µg of iodoacetamide then was added to block all of the available free thiols from oxidizing, and the reaction solution was allowed to stand for 1 h at 25 °C. The proteins then were dialyzed against TBS at 4 °C and used without further treatment.

rsCD4 at a concentration of 1 mg/ml in TBS was dialyzed against 0.1 M NaHCO₃ at 4 °C. To the rsCD4 then was added 12.3 µl of a solution of 1.8 mg of biotin succinimide ester (Pierce) in 180 µl of N,N-dimethylformamide. The solution was rotated for 1 h at 4 °C, dialyzed at 4 °C into TBS, and stored initially at 4 °C until use. Under these storage conditions, it appeared, however, that the activity of the CD4 was lost after about 1.5-2 weeks. Thus, biotinylated CD4 was stored in small aliquots at 70 °C and used once after thawing. The amount of biotin that was covalently linked to the rsCD4 could not be determined because, at present, there are no methods for accurately making these types of measurements. However, 15 molecules of biotin is the maximum amount of biotin that could be linked to one molecule of rsCD4, because the rsCD4 was labeled with a 15-fold molar excess of biotin succinimide ester. We have learned that using a 50-fold molar excess of the biotin succinimide ester to label rsCD4 caused the rsCD4 to precipitate and it was unable to bind to gp120.

Peptomers or peptides in H₂O were spotted onto nitrocellulose paper at the concentrations indicated in the legend to Fig. 3. The nitrocellulose then was blocked with 3% commercial dried milk in TBS for 1 h and then treated with 1 µg/ml biotinylated CD4 for 3 h in TBS-Brij (TBS containing 0.03% Brij 35). The nitrocellulose sheets then were treated with a 1:750 dilution of streptavidin-horseradish peroxidase and developed with 0.3 mg/ml 4-chloro-1-naphthol substrate (Pierce), 0.3% H₂O₂ in TBS.

The ability of biotinylated CD4 to bind to the HIV-derived peptides, peptomers, or gp120 was evaluated using a solid-phase assay with the peptides, peptomers, or the gp120 immobilized to the polystyrene surface of 96-well microtiter plates (Nunc).

For the CD4-peptomer binding assay, 96-well polystyrene microtiter plates were coated for 1 h at 25 °C with 10 µg/well peptomer in 0.01% formic acid. The peptomer solution was made by dissolving 1 mg of the peptomer in 100 µl of 88% formic acid and then adding this solution to 19.9 ml of deionized H₂O. 200 µl of this peptomer solution was used to coat each well. The plates then were washed with TBS-Brij. Biotinylated rsCD4 in TBS-Brij then was diluted serially in the wells across each row at 100 µl/well with the peptomer and incubated for 1 h. The plates were washed three times with TBS-Brij, and each well was reacted with 1:1000 dilutions of streptavidin-alkaline phosphatase (Tago, Inc., Burlingame, CA) conjugate in TBS-Brij for 1 h. Plates were developed with p-nitrophenyl phosphate in pH 9.8 NaHCO₃ buffer.

The same assay was used for the competition assay using rsCD4 to compete with the biotinylated rsCD4, except the rsCD4 was serially diluted in each well across the rows of the peptomer-coated plate first, and this was followed by the addition of 100 µl/well biotinylated rsCD4 at 1.0 µg/ml. The plate was then incubated and developed as above.

For the binding assay involving immobilized gp120, the stock solution of 2.6 mg/ml rgp 120 in TBS was diluted to 2.6 µg/ml, and 200 µl of this solution was used to coat each of the wells of the microtiter plates. After 16 h at 25 °C, the wells were washed three times with TBS-Brij, and the unoccupied sites on the plate were blocked by treating the wells for 1 h with 3% commercial dried milk in TBS. The biotinylated rsCD4 was used thereafter, and the various experiments are described separately below.

For the inhibition assays, the microtiter plates were coated for 1 h with 10 µg/well peptomer-(419-436) as described above. The plates were then washed and blocked with 3% commercial dried milk in TBS. In separate 1.5-ml polypropylene tubes, stock solutions of gp120 or denatured gp120 were diluted serially at 100 µl/tube, and to each tube containing the diluted amounts of gp120 was added 2 µg of biotinylated rsCD4 in 100 µl of TBS-Brij. The total volume of the biotinylated CD4 with varying amounts of gp120 was 200 µl in each tube. Following a 1-h incubation period at 25 °C, 200 µl of a 3% solution of commercial dried milk in TBS was added to each tube, 200 µl of this was then transferred to the peptomer-coated wells of the ELISA plate, and the mixtures were incubated for an additional hour. The plates were washed and developed as above.

For the assay to test the ability of the various C4 constructs to block CD4-gp120 binding, ELISA plates were coated with 100 µl of rgp120 in TBS (520 ng/well) as detailed above. Serial dilutions of peptomer were incubated in separate 1.5-ml polypropylene tubes with 2 µg of biotinylated CD4 in a total volume of 200 µl of TBS-Brij. After a 1-h incubation period, 200 µl of 3% commercial dried milk in the same buffer then was added to each tube, and 200 µl of the mixture was added to the separate wells of the ELISA plate. Following a 2-h incubation period, 1:1000 streptavidin-alkaline phosphatase was added for 1 h, and the were plates developed with p-nitrophenyl phosphate in pH 9.8 Na₂HPO₄ buffer.

CD spectra were recorded from 280 to 185 nm on a Jasco model J-500A/DP-501N CD spectropolarimeter in Hellma QS cells with a 1-mm path length at room temperature. Peptide concentrations were 36 µM in H₂O, pH 5.3, or in 10 mM, pH 7.2, sodium phosphate buffer. The conformations of the peptide and the peptomer were estimated using Provencher's spectral deconvolution program (15) using the experimental CD data.

With the exception of the inhibition data of the peptide-(419-436) constructs, all of the data could be fit with the classical equation for binding based upon Scatchard analysis using the equation,

(Eq. 1)

where X is the fraction of the binding sites occupied as a function of free ligand in a binding experiment or is the fraction of binding sites unbound in a competition experiment. K_a is the intrinsic association constant (K_d = 1/K_a) on a molar concentration scale, and C is the molar concentration of the free ligand (16).

For reasons we do not understand, the above equation did not give a satisfactory fit to the inhibition data obtained with the peptide-(419-436) constructs. Therefore, we used an alternative fitting function. The K_a values were determined by using a curve fitting function (16) developed with the idea of optimally fitting data to a sigmoid-shaped curve and calculating the 50% dissociating point (which gives the K_d),

(Eq. 2)

where X is expressed as a decimal fraction and is determined from the experimental data; C is the free ligand concentration; C(0.5) is the concentration of free ligand at 50% binding, and n is a parameter that controls the sharpness of the sigmoidal transition. The value of C(0.5) then is used as the concentration of ligand that is found to inhibit the binding by 50% (K_d).

K_d is equal to the IC₅₀ when we make the explicit assumption of a single binding site, which we are doing here. This is a reasonable assumption to make, since we do not know either the size of the receptor molecule or how many receptor sites there are per receptor molecule; therefore, we can assume a single site per receptor molecule of unknown size.

RESULTS

The C4 region of gp120 is highly conserved among all of the HIV-1 strains but not between HIV-1 and HIV-2 (Table I), and the amino acid sequence identity between the C4s from HIV-1 and HIV-2/simian immunodeficiency virus (SIV) is only 56%. Although conformational identity between HIV-1 and HIV-2/SIV has not been reported, conformational identity is predicted when the C4 regions are presented in a comparative helical wheel (17) as shown in Fig. 1. Whereas the hydrophilic surfaces of the amphipathic helices are quite different, the hydrophobic surfaces are highly conserved. The single difference in the hydrophobic surface between Met⁴³⁴ of HIV-1 and Val⁴²⁷ of HIV-2 is quite conservative, with both amino acids being hydrophobic. The conformational homology as shown in Fig. 1 provides a compelling argument to synthesize and test helical peptide-based derivatives of the C4 domains in binding and competition studies involving CD4 and gp120.

Table I. Sequences from CD4 binding region of gp120 Data are taken from the Los Alamos HIV-1 Data Base. From HIV-1 sequences Consensus 1 R I K Q I I N M W Q E V G K A M Y A (393 -410) HIVMN K I K Q I I N M W Q E V G K A M Y A (419 -436) Consensus 2 R I K Q I I N M W Q ? V G K A M Y A (369 -386) From HIV-2 sequencesa Consensus H I R Q I I N T W H K V G K N V Y (374 -390) HIV-2-ISYR H I E Q I I N T W H K V G K N V Y L (412 -429) a  Boldface letters in the HIV-2 sequences indicate homology with HIV-1.

Peptomers are polymers composed of specifically cross-linked synthetic peptides (18, 19, 20, 21, 22), and the synthesis of the peptomers used in this study, starting with the N-chloroacetyl, C-cysteine-containing peptide, has been published (9). A key finding about the peptomers from the C4 region of gp120 is that the peptide conformation in the peptomer is -helical in aqueous solutions at physiological pH, and this is noteworthy when the monomeric peptide could be helical, in theory (see above), but has no helical conformation in aqueous solutions at neutral pH. Such is the case for peptide-(419-436) from HIV-1_MN used here and in Ref. 9. Thus, for the amphipathic peptide-(419-436) from HIV-1_MN and peptide-(412-429) from HIV-2_ISYR, the helical conformation of the peptide, as presented in the peptomer, may be more representative of the peptide as it resides in intact gp120 than as the free monomeric peptide. A picture of the peptide going to the peptomer is displayed in Fig. 2, and it serves to assist readers in conceptualizing the conformational transition that takes place in the peptide upon polymerization. Proof that the peptide, as a component of a peptomer, mimics the conformation of the peptide in the intact protein would be provided, in part, if the peptomers had CD4-binding specificities that were comparable with intact gp120.

The first set of experiments was designed to evaluate the ability of biotinylated rsCD4 to bind to the helical peptide constructs, and comparisons were made with binding to the peptide having no conformational constraints. It is clear from the experiments run in a dot blot format that biotinylated CD4 would recognize only those peptides from C4 having helical conformations, whereas those not having helical conformations were not able to bind CD4 (Fig. 3).

Additional peptides tested in the dot blot experiment, but not shown in Fig. 3, included a scrambled peptomer (amino acid sequence IMWKEAAKYQVGQMNIKIC-NH₂) and peptomer-(308-322), a peptomer from the V3 loop of gp120_MN. Neither of these were recognized by biotinylated CD4. Peptomer-(419-436)-CHO-Trp⁴²⁷, a polymeric presentation in which Trp⁴²⁷ was formylated and which displayed no helical conformation (9), did not bind CD4. Likewise, the monomeric peptide-(419-436) did not bind CD4 in this dot blot assay (data not shown).

An ELISA-type format then was used for an assay to generate quantitative comparisons of the binding of CD4 to the helical constructs with CD4 binding to gp120 from HIV-1. In addition, it was of importance to test these materials in inhibition-type assays to draw conclusions about the specificities of the CD4 binding to a helical construct. For these experiments a molecular weight of 41,607 that was based on the amino acid composition of rsCD4, was used.²

When biotinylated CD4 was titrated down a series of peptomer-(419-436)-coated microtiter wells, a K_d of 8.59 nM was obtained (Fig. 4). In addition, the immobilized peptomer from the C4 of HIV-2_ISYR showed a K_d of 14.59 nM (Fig. 4). No other peptides or conjugates of peptide-(419-436), when immobilized to the polystyrene surface of the microtiter plate, bound biotinylated CD4. A single exception to this was the albumin-peptide-(419-436) conjugate, which was able to bind CD4, albeit with a much lower affinity than that observed for the immobilized peptomers (Fig. 4).

As a first test of whether nonbiotinylated CD4 would block biotinylated CD4 binding to peptomer-(419-436), thereby indicating a degree of specificity for the biotinylated CD4 as compared with nonbiotinylated CD4, a similar set of experiments was set up in 96-well microtiter plates. Using serial dilutions of the rsCD4 with a constant 5 µg/ml biotinylated rsCD4, competition experiments with decreasing concentrations of nonbiotinylated CD4 clearly indicated that there was equal competition with the nonbiotinylated CD4 (Fig. 5A). The K_d obtained from this experiment is 9.88 nM, a value in the range of that obtained for the affinity obtained for CD4 binding to gp120, which was reported to be 4 nM (4). Denatured CD4 did not block the binding of biotinylated CD4 to the peptomer-(419-436) in the experiments reported here.

The K_d for CD4 blocking the binding of biotinylated CD4 to the HIV-2-derived peptomer-(412-428) was found to be approximately 28.27 nM (data not shown). This value also is parallel to that reported previously in which it was demonstrated that gp120 from HIV-2 bound to CD4 with less affinity than gp120 from HIV-1 (23).

Fig. 5B shows that rgp120 blocks the binding of biotinylated CD4 to peptomer-(419-436), and the reduced-carboxamidomethylated gp120 did not (data not shown). This experiment clearly indicates, first, that gp120 is blocking CD4 binding to the immobilized peptomer by perhaps binding to the same region of CD4 that is recognizing the peptomer and, second, due to the fact that the denatured gp120 was not active here, the gp120 conformation is important for CD4 binding. A K_d of 8.08 nM was obtained for the inhibition of biotinylated CD4 binding to immobilized peptomer-(419-436) by rgp120. This value is close to the value of 4 nM obtained by Lasky et al. (4) using a different assay to study gp120 binding to CD4.

Although peptide-(419-436) did not bind CD4 in the dot blot or ELISA-type assays when it was immobilized to nitrocellulose or polystyrene, we did find that peptide-(419-436) could block CD4 binding to immobilized peptomer-(419-436) (data not shown) and to immobilized rgp120 (Fig. 6). This result was unexpected; however, when we examined the CD spectrum of peptide-(419-436) in a solution containing 0.03% Brij 35, the same concentration of detergent used in the ELISA-type binding assay, we found that the peptide displayed 17 ± 1% -helix (Fig. 7). When compared with the same peptide in detergent-free buffer, this conformation difference could account for its ability to bind CD4 and block the binding of CD4 to gp120. From these experiments, we obtained a K_d of 42 µM for the peptide-(419-436) blocking the binding of biotinylated CD4 to immobilized gp120 in TBS-Brij (Fig. 6).

Very similar results were obtained for the N-acetylated peptide-(419-436), first solubilized in Me₂SO, and for peptomer-(419-436), also solubilized first in Me₂SO (data not shown). Me₂SO did not influence the binding of CD4 to gp120, but it did assist in solubilizing the N-acetylated peptide-(419-436) and peptomer-(419-436) at the high starting concentrations (~1 mg/ml) (data not shown) used in these experiments. A peptide that was identical to peptide-(419-436) but had glycine at amino acid 427 instead of tryptophan did not block CD4 binding to gp120, and it did not display any helical content in the Brij 35-containing solution (data not shown).

Shorter segments of peptide-(419-436) were synthesized, but they required the N-terminal lysine for solubility in aqueous solutions. Shorter peptides composed of KIKQIINMW and QEVGKAMYA did not bind CD4; nor did they block the binding of CD4 to immobilized gp120 in the Brij-containing buffer (data not shown). Clearly, a separate study is necessary to determine which amino acids would be required for binding to CD4 and for optimizing the necessary conditions to mimic the gp120 C4 region with nonpeptide-based molecules.

Although the antibodies that were produced against the peptomer-(419-436) from HIV-1 did react with the peptomer-(419-436) from HIV-1 and with native and recombinant gp120 from HIV-1 (9), they did not block the binding of CD4 to peptomer-(419-436) or to gp120. In addition, the same antibodies did not cross-react with peptomer-(412-429) from HIV-2 (data not shown). Since both peptomer-(419-436) from HIV-1 and peptomer-(412-429) from HIV-2 bind CD4 and both share the same hydrophobic surfaces on their predicted -helices (Fig. 1), we conclude that CD4 is probably binding to the hydrophobic surfaces of the helices. Antibodies against these peptomers, then, could be binding to the hydrophilic surfaces of the helices because such an explanation would account for the inability of the antibodies to cross-react between HIV-1 and HIV-2 peptomers and block CD4 binding.

DISCUSSION

In a recent publication, we showed that an amphipathic -helical immunogen from C4 produced antibodies that reacted with native and recombinant gp120 (9). The antibodies reported in the previous study (9) did not react with denatured gp120 in either an ELISA or in the commercial Western blot (data not shown). Our conclusion, therefore, based on immunological data, is that C4 is an amphipathic -helix in the intact gp120. Since C4 has been implicated as being the CD4 binding site in gp120 (4, 5), it was reasonable for us to evaluate the amphipathic -helical constructs for their ability to bind CD4, and we showed that in this study.

In this report we show that CD4 can bind to a C4-derived peptomer from HIV-1_MN with a K_d of 8.59 nM and to a C4-derived peptomer from HIV-2_ISYR with a K_d of 14.59 nM. These values are well within the range of affinities reported for the binding of CD4 to HIV-1 or HIV-2 (23). The binding requires that C4 be presented as an amphipathic -helix, which, as it appears from the theoretical helical wheel shown in Fig. 1, contains both a hydrophilic surface and a hydrophobic surface. The high affinity indicates that much of the binding could be attributed to the hydrophobic surface of the helix, and this conclusion is supported by the many papers that describe hydrophobic molecules blocking HIV-1 infection (24, 25, 26, 27, 28, 29, 30, 31). Hydrophobic interactions in an aqueous environment could explain the nM affinities that have been observed for gp120 binding to CD4. In addition, the homology shown on the theoretical wheels (Fig. 1) supports the hypothesis that it is the hydrophobic surface of C4 that is being recognized by CD4.

It has been shown that the CC" ridge in CD4 contains the major locus for binding gp120 (33). The height of the ridge contains a phenylalanine residue flanked by leucine, another hydrophobic amino acid, and is exposed with an extended hydrophobic core supporting the CC" loop (34, 35). The proper exposure of the hydrophobic surface on the C4 amphipathic helix in the intact gp120 could be sufficient to fit into an extended CC" ridge from CD4.

In this paper, we have built on the early work of C4's involvement in binding CD4 (4) and proposed at the start of our efforts that conformational considerations regarding C4 itself probably play a role in the ability of C4 to bind CD4. In addition, though, the other regions of gp120 that are outside of C4 but are thought to play a role in gp120 binding CD4 (6, 7, 8) actually may influence the surface expression of C4 and not be involved directly with CD4 binding. Drugs and therapeutics that are rationally designed to block the binding of gp120 to CD4 could address this last point; designing a CD4-binding material to block gp120 binding to CD4 as an anti-HIV therapeutic may have undesirable consequences.