Conformational aspects of HIV-1 integrase inhibition by a peptide derived from the enzyme central domain and by antibodies raised against this peptide


S. Fermandjian, Département de Biologie et Pharmacologie Structurales, CNRS UMR 1772, Institut Gustave Roussy, Villejuif 94805 Cedex, France. Tel.: + 33 142114985, Fax: + 33 142115276, E-mail:


Monospecific antibodies were raised against a synthetic peptide K159 (SQGVVESMNKELKKIIGQVRDQAEHLKTA) reproducing the segment 147–175 of HIV-1 integrase (IN). Synthesis of substituted and truncated analogs of K159 led us to identify the functional epitope reacting with antibodies within the C-terminal portion 163–175 of K159. Conformational studies combining secondary structure predictions, CD and NMR spectroscopy together with ELISA assays, showed that the greater is the propensity of the epitope for helix formation the higher is the recognition by anti-K159. Both the antibodies and the antigenic peptide K159 exhibited inhibitory activities against IN. In contrast, neither P159, a Pro-containing analog of K159 that presents a kink around proline but with intact epitope conformation, nor the truncated analogs encompassing the epitope, were inhibitors of IN. While the activity of antibodies is restricted to recognition of the sole epitope portion, that of the antigenic K159 likely requires interactions of the peptide with the whole 147–175 segment in the protein [Sourgen F., Maroun, R.G., Frère, V., Bouziane, A., Auclair, C., Troalen, F. & Fermandjian, S. (1996) Eur. J. Biochem. 240, 765–773 ]. Actually, of all tested peptides only K159 was found to fulfill conditions of minimal number of helical heptads to achieve the formation of a stable coiled-coil structure with the IN 147–175 segment. The binding of antibodies and of the antigenic peptide to this segment of IN hampers the binding of IN to its DNA substrates in filter-binding assays. This appears to be the main effect leading to inhibition of integration. Quantitative analysis of filter-binding assay curves indicates that two antibody molecules react with IN implying that the enzyme is dimeric within these experimental conditions. Together, present data provide an insight into the structure–function relationship for the 147–175 peptide domain of the enzyme. They also strongly suggest that the functional enzyme is dimeric. Results could help to assess models for binding of peptide fragments to IN and to develop stronger inhibitors. Moreover, K159 antibodies when expressed in vivo might exhibit useful inhibitory properties.






dumbbell substrate

The human immunodeficiency virus (HIV) replication cycle is initiated by integration of the viral DNA into the cell genome [1, 2]. Integrase (IN) [2–4] encoded by the virus pol gene catalyses integration sequentially through the deletion of two nucleotides at the conserved 3′ extremity of the viral long-terminal repeats (LTRs) (3′ processing) and the covalent insertion of the recessed 3′ viral DNA into the cell genome (strand transfer) [5–10]. Both steps proceed by a one-step transesterification mechanism [11]. IN has no cellular counterpart and may be therefore considered as a specific target for development of anti-HIV therapy [12].

IN displays three functional domains [13, 14]. The amino-terminal part is highly conserved in retroviral integrases and contains a conserved C,C,H,H motif that binds zinc [15, 16]. This domain is critical for forming a stable complex with viral DNA, its structure is similar to the helix-turn-helix motif of DNA binding proteins and it forms dimers in solution [17]. The carboxy-terminal domain presents a group of basic residues that bind DNA nonspecifically [18–20]. NMR studies performed on fragments of IN have established that the C-terminal domain presents similarities with the SH3 domain and is able to form dimers [21, 22]. Mutational studies have shown that both amino and carboxy-terminal domains are needed for 3′ processing and strand transfer [13, 23, 24]. The central domain is also highly conserved among retroviral INs as well as many transposases; it contains the D,D,E residues forming the catalytic triad of the enzyme [25, 26]. Structural studies of the central domain of HIV-1 and ASV INs have proven the membership of IN to the structurally related superfamily of polynucleotidyl transferases [27–29]. The core of IN displays a dimeric structure both in the crystal and in solution, although it has been proposed that in its active form the entire enzyme is a multimer [6, 27]. Like IN, the central domain IN50–212 is able to catalyze the desintegration reaction [6, 25, 26]. This occurs through liberation of the initial substrate used for strand transfer reaction by cleaving DNA at the CA dinucleotide junction between viral and target DNA [25].

To catalyze the cleavage and strand transfer reaction, IN interacts with viral and target DNA sequences. Several studies have shown that stable complexes can be formed between IN and viral DNA ends in the presence of divalent ions [30, 31]. Two lysine residues have been recently identified near the catalytic site to be involved in the specific interaction with DNA [32]. A synthetic peptide (K159), reproducing the sequence 147–175 of the IN central domain, displays helical heptads and coiled-coil forming properties together with inhibition effects on the integration activity of IN [33, 34]. A mechanism, involving coiled-coil formation between K159 and its counterpart in the protein, was then considered, particularly as several hydrophobic side-chains of the 147–175 helical heptads have been found oriented outwards from the molecule, at the surface of the protein crystal structure [27].

To understand better the functionality of the 147–175 helical domain in the protein and to improve our peptide models for the IN inhibition, monospecific antibodies were prepared from rabbits immunized with the synthetic peptide K159 ( Fig. 1). The antibodies recognize injected K159, the whole IN and the central domain (IN). They also efficiently bind P159 a Pro-containing analog of the antigenic K159, as well as C-terminal peptides, proving that the functional epitope resides in the C-terminal part of K159. Moreover, when tested for their ability to inhibit the catalytic activity and the binding capacity of IN to DNA, the antibodies – likewise the antigenic K159 – alter significantly these two properties, therefore highlighting the importance of the segment 147–175 for the IN functionality. Remarkably, two antibody molecules react with IN in filter-binding assays, indicating that IN is dimeric.

Figure 1.

The three functional domains of HIV-1 IN. Amino acid sequences of the synthetic peptides K159 reproducing the IN segment 147–175, and of the structural analog P159 as well as truncated analogs IN636, IN637, IN638 and IN642. Peptides were synthesized according to a standard Fmoc procedure as previously described [34]. Each peptide carries a tyrosine residue and an amide group at the C-terminal position (not represented).

A conformational analysis was carried out by circular dichroism (CD) and 1H NMR (aqueous and trifluoroethanol (TFE)/H2O mixtures) together with secondary structure predictions [35–39]. Data reveal the propensity of K159 for helix formation over its whole length, i.e. over its four heptad sequences. For P159 the helix propensity is still maintained in the C-terminal part. This peptide is recognized by antibodies to the same extent as K159, indicating that the epitope conformation is not affected by the substitution. However, only K159, which exhibits helicity over its full length and a good propensity for coiled-coil formation, presents the double ability to bind antibodies and to inhibit IN.

Materials and methods


HIV-1 IN was overexpressed in Escherichia coli BL21 and purified as previously described [40].


Peptides. Sequences of synthesized peptides are given in Fig. 1. All peptides were amidated at their C terminus.


Peptides and oligonucleotides synthesis

Peptides were prepared according to a standard Fmoc procedure on an Applied Biosystems model 432A peptide automatic solid-phase synthesizer, as previously described [34]. The DNA strands were synthesized by solid phase using the phosphoroamidite procedure on an Applied Biosystems 318 B automatic apparatus. The resulting single strand DNA was purified by reverse phase HPLC on a C18 Waters µBondapack C18 column, followed by an electrophoresis through 15% denaturating (7 m urea) polyacrylamide gels. The region of the gel corresponding to each oligonucleotide was excised and soaked overnight in 0.3 m sodium acetate. The supernatant fluid from the soaked gel fragment was collected and the DNA desalted on a NAP 10 column (Pharmacia Biotech) before use. Oligonucleotides were 32P-labeled at their 5′ end with T4 polynucleotide kinase (New England Biolabs) and [γ-32P]ATP with specific activity of 4500 Ci·mmol–1 (ICN). Unincorporated radioactive nucleotides were removed by centrifugation through Sephadex G10 columns.

Antibody production and purification

Monospecific antibodies were raised against the synthetic peptide K159, which reproduces the residues 147–175 of the central domain of the HIV-1 IN ( Fig. 1). The peptide was coupled to Keyhole Limpet Hemocyanin (KLH-Pierce) with benzidine and used to immunize two rabbits by injection of 150 µg of synthetic peptide in complete Freund’s adjuvant. Booster injections were given on days 21 and 42. The rabbits were bled on day 49. An enzyme-linked immunosorbent assay (ELISA) was performed against K159 and HIV-1 integrase and permitted to calculate the antisera titers. The antisera presenting high titers against IN were subjected to subsequent purification.

To remove the contaminating lipids and lipoproteins, each antiserum was precipitated by adding an ammonium sulfate solution. IgG immunoglobulins were obtained by an affinity chromatography on a Protein A Sepharose CL4b (Pharmacia) column. The protein concentration of each fraction was measured using the BIORAD protein assay. In order to obtain monospecific antibodies enriched with anti-K159 and to eliminate the anti-KLH antibodies, KLH was immobilized on CNBr activated sepharose 4B (Pharmacia) and the specific IgG were eluted. Eluates were spectrophotometrically measured at 280 nm, fractions containing monospecific IgG were pooled and the final concentration was calculated. An ELISA assay was performed to determine the IgG titers. To study their specificity, the monospecific antibodies were preincubated with either K159, IN1–288 or IN50–212 overnight at 4 °C before coating the ELISA plaque.

Secondary structure predictions

Calculations based on statistical mechanics were carried out using the AGADIR computer program kindly provided by the authors [37–39]. This program considers short-range interactions between residues at different pH and temperatures, and allows calculating the helicity per residue of peptides. Results reflect the behavior of peptides lacking tertiary interactions in solution. Structure predictions more reliable for peptides in the protein environment (amino acid segments) were performed using the GOR algorithm [35, 36], which allows calculating the helix probability per residue.

Western blot analysis

Samples containing purified integrase (250 and 125 ng) or IN50–212 central domain (250 and 125 ng) were subjected to electrophoresis through a 0.1% sodium dodecyl sulfate (SDS)/15% polyacrylamide gel. The separated proteins were transferred to nitrocellulose membranes (Schleicher & Schuell: 0.45 µm reinforced cellulose nitrate membrane). The membrane was blocked overnight at 4 °C then washed and incubated with antisera (2.5 µg·mL–1) for 4 h at 4 °C followed by 1 h incubation at room temperature with horseradish peroxidase-linked donkey anti-(rabbit Ig) (1/5000-Amersham). The antibody binding was detected using the enhanced chemiluminescence (ECL) Western blotting applications (Amersham) followed by autoradiography.

Filter-binding assay

The reaction conditions were described previously [41]. Typically, in a 20-µL volume, HIV-1 IN (5.6 pmol) was incubated with antisera at various IN/antibody molar ratios or P159 and K159 at different peptide concentrations in the reaction buffer (20 m m Tris/HCl, pH 7.5, 1 m mβ-mercaptoethanol, 1 m m dithiothreitol, 1 m m MnCl2, EDTA 0.3 m m) for 15 min at 34 °C then the DNA-labeled substrates (0.6 pmol) were added and incubated to additional 10 min at 34 °C. After incubation, the samples were filtered through a nitrocellulose membrane (see Western blot analysis) in a Dot-blot apparatus. The wells were washed three times with 0.5 mL washing buffer (20 m m Tris/HCl, pH 7.5, 100 m m NaCl) prior to filtration. Only proteins and DNA bound to proteins were retained on the filter. At the end, the wells were washed three times with 0.5 mL washing buffer then the membrane was washed for 15 min in the same buffer and dried for 30 min at 40 °C. Finally, the membrane was autoradiographed and the radioactivity was recorded using a PhosphorImager Scanner (Molecular Dynamics).

HIV integrase assays

To evaluate strand transfer activity [34, 42], IN (280 n m) was preincubated with antisera at different IN/antibody molar ratios and pSP65 DNA (10 ng) in (20 m m 3-(N-morpholino)propanesulfonic acid (Mops), pH 7.5, 10 m m MnCl2, 10% (w/v ) glycerol, 10 m m dithiothreitol, 0.1 mg·mL–1 (BSA) for 30 min at 30 °C, then a double preprocessed stranded oligonucleotide S1/S2 (0.5 ng) was added to the mixture and the reaction was further incubated for 40 min at 30 °C. To stop the reaction, 10 µL of sample loading buffer (300 m m Tris/HCl, pH 6.8, 50% (w/v) glycerol, 0.25% (w/v) bromophenol blue, 0.25% (w/v) xylene cyanole, 10% (w/v) SDS and 50 m m EDTA) were added to the reaction products. These were separated on a 1.2% agarose gel in 0.5 × Tris-Borate-EDTA (TBE), pH 8, and visualized by autoradiography after gel drying.

For 3′ processing and autointegration activities, the double-stranded S3/S2-labeled oligonucleotide was used. The IN preincubation with antisera and incubation with S3/S2 oligonucleotides were carried out at 37 °C in the same buffer as used for the strand transfer assay. The reaction was stopped by the addition of 20 µL of loading buffer (98% formamide, 10 m m EDTA, 0.05% xylene cyanole and 0.05% bromophenol blue). The samples were heated for 3 min at 95 °C then separated in a denaturating (7 m urea) 18% polyacrylamide gels in 1 × TBE, pH 8, and visualized by autoradiography. Autointegration products were observed after overexposure.

Desintegration reactions were carried out using the dumbbell substrate [5]. Briefly, in a 20-µL volume, IN (280 n m) was preincubated with antisera at different IN/antibody molar ratios in the reaction buffer containing (20 m m Tris/HCl, pH 7.5, BSA 0.1 mg·mL–1, glycerol 10% (w/v), Nonidet P-40 0.01% (w/v), 10 m m dithiothreitol, 10 m m MnCl2) for 30 min at 37 °C then the DNA labeled dumbbell substrate (30 n m) was added and the mixture was incubated for another 60 min at 37 °C. The reaction was stopped by the addition of 20 µL loading buffer. The final products were heated for 3 min at 95 °C and subjected to electrophoresis in denaturating (7 m urea) 15% polyacrylamide gels in 1 × TBE, pH 8, and visualized by autoradiography. The quantification of integration products was made with a PhosphorImager Scanner (Molecular Dynamics).

CD spectroscopy

CD spectra were recorded at 25 °C on a Jobin Yvon Mark 6 spectropolarimeter linked to a PC microprocessor. The experimental conditions and the effects of trifluoroethanol addition on α-helix content were carried out as previously described [34]. CD intensities are expressed as Δε ( m–1·cm–1) and the α-helix content is calculated as follows [43]:

Pα = – (Δε222 × 10)/N, where Δε222 is the circular dichroism at 222 nm and N is the number of residues in the peptide.

NMR spectroscopy

Two-dimensional NOESY and TOCSY experiments were recorded at 500 MHz on a Bruker DMX 500 spectrometer and processed with FELIX software. The structural properties of K159 and P159 were analyzed under two experimental conditions: in aqueous solution (90% H2O/10% D2O) and in 20% TFE/80% H2O mixture. Spectra were obtained at 2 m m monomer concentration, pH 3.0, and 20 °C (our unpublished results).


Anti K159 monospecific antibodies recognize IN and its central domain IN50–212

Antibodies were raised against the K159 peptide that reproduces the sequence of the 147–175 segment of HIV-1 IN ( Fig. 1) [34]. Two rabbits were used for immunization, and sera exhibiting high titers against K159 were selected. These were then subjected to purification through elution on a sepharose-protein A column in order to eliminate contaminating lipids and lipoproteins. As purified IgG presented high recognition capacities toward IN, monospecific antibodies were further prepared against K159. These anti-K159 antibodies expressed high titers not only against K159 but also against IN1–288 and IN50–212( Fig. 2). Under the same conditions preimmune sera did not display any binding properties. ELISA experiments showed that the recognition capacity of K159 by antibodies is significantly altered after preincubation with either K159, IN1–288 or IN50–212 (data not shown).

Figure 2.

Relative binding affinity of anti-K159
to IN, IN50–212 and synthetic peptides shown in
Fig. 1. The antigen–antibody interaction was quantified by standard ELISA assays (see Materials and methods). Binding at 100% level represents the maximal absorbancy at 422 nm obtained with immobilized antigen K159 at 90 n m of anti-K159. Presented percentages of bound antibody correspond to averaged values from two experiments.

Specificity of IN recognition by antibodies was established through Western blot assay ( Fig. 3). Both IN1–288 and IN50–212 were separated on a 15% PAGE, transferred to a nitrocellulose membrane and incubated with antibodies. The blot clearly shows that monospecific anti-K159 interact specifically with intact IN and its central core IN50–212. For both IN1–288 and IN50–212 two bands are detected on the blot probably resulting from internal initiation at methionine 22 and 154 or carboxyl-terminal truncation of IN during expression in E. coli as described elsewhere [6, 44]. BSA was used as negative control. Despite the fact that the amount loaded on the gel was eight times more than that of IN1–288 and IN50–212, no binding between anti-K159 and BSA (molecular mass, 66.4 kDa ) was detected on the blot, underlining the good specificity of antibodies to IN. Moreover, a Western blot assay was carried out on cell extracts of E. coli cells expressing IN50–212 and only one band corresponding to 18 kDa was observed on the blot confirming the high specificity of antibodies to IN ( Fig. 4).

Figure 3.

Western blot analysis of IN1–288 and IN50–212 with anti-K159. In lane 1, 2 µg of BSA were loaded on to the gel and used as a negative control. In lanes 2 and 3, 250 ng and 125 ng of IN1–288 were loaded, respectively. In lanes 4 and 5, IN50–212 was loaded at 250 ng and 125 ng, respectively. Molecular mass markers are shown on the left of the blot.

Figure 4.

Western blot assay with anti-K159 on cell extracts ofE. coliexpressing IN50–212. Lane 1, no extracts were loaded; lanes 2–4, extracts were loaded at increasing dilutions. Molecular mass markers are shown on the left of the blot.

Epitope mapping

To specify within antigenic K159 the location of the functional epitopes recognized by the monospecific anti-K159, a series of peptides were synthesized ( Fig. 1). Some of the properties of K159 and of its analog P159 have been previously described [34]. P159 differs from K159 by a Lys159 → Pro substitution that alters the dimerization and inhibitory properties of the peptide. The truncated analog, IN636, reproduces the C-terminal part of K159 and P159, from residues 163 to 175. To obtain IN637, IN636 was elongated by five residues at its N-terminal part, i.e. from residues 162 to 158. IN638 reproduces the N-terminal part of K159 from residues 147 to 163. Interactions between monospecific antibodies and newly synthesized fragments were analyzed by ELISA experiments. Figure 2 shows that K159 and P159 bind antibodies with a similar pattern, i.e. 100% of binding at, for instance, 90 n m of anti-K159. Replacement of Lys159 by the helix-breaking residue Pro at the center of the peptide does not alter the recognition by antibodies, suggesting that the functional epitope is located in either the N or the C-terminal part of K159, but not in the middle. Binding data obtained with IN636, IN637 and IN638 further indicate that antibodies recognize the C-terminal part of K159 by far better than its N-terminal part ( Fig. 2). For instance, at 9 n m of anti-K159 the two C-terminal fragments IN636 and IN637 still exhibit nearly 60% binding against only 15% for the N-terminal fragment IN638.

Helix potential of the functional epitope and peptide recognition by antibodies

To learn of the possible impact of peptide conformation on the epitope recognition by antibodies, we carried out a structural analysis of K159 and analogs. According to X-ray diffraction data, the segment 147–175, corresponding to K159, displays a rather flexible helical structure within the central core IN50–212[27]. Helix stabilization in proteins depends on various factors involving intrahelical and tertiary interactions. The GOR’s program is adapted to prediction of secondary structures for protein segments. While the AGADIR’s program allows to predict the helical potential of peptide fragments in aqueous solution. In that case the helix stability depends mainly on short-range interactions arising between residues. According to AGADIR, peptides IN636, IN637 and IN638 possess, like their parent peptide K159, a small helix content that contrasts with the high content predicted by GOR. The difference reflects the high sensitivity of peptide conformations to medium context and underline the helix stabilizing effect exerted by the protein on the 147–175 segment.

The intrinsic propensity of peptides for helix formation was then assessed by CD spectroscopy in aqueous solution and in TFE/aqueous mixtures ( Fig. 5). In aqueous solution, every peptide shows a CD spectrum characteristic of an unordered conformation and at most a few per cent helix. In the helix promoting solvent, TFE, the proportion of molecules found in helical conformation provides a good image of the intrinsic propensity of the peptide for the helix structure within the protein context. This is well illustrated in Figs 5 and 6 by (a) the set of CD spectra obtained for IN636 and the curves of helix contents as a function of TFE of the various peptides ( Fig. 5 and insert) and (b) the predictions (GOR and AGADIR), NMR and CD data (H2O and 20% TFE) relative to K159 and P159 ( Fig. 6). At peptide concentration of 50 µm, 25 °C, the truncated analogs IN636 and IN637, both containing the epitope, present a weaker helix content compared to K159 and P159 (20% and 25% vs. 70% and 60% at 50% of TFE) (insert in Fig. 5). In comparison, IN642 that derives from IN636 by the replacement in its center of Ala (the best helix former) [45] by the stronger helix-breaking residue Pro, presents a weaker helix content (10% at 50% TFE) (insert in Fig. 5). The ELISA results show that the Ala169 → Pro substitution alters the binding properties of the peptide to antibodies ( Fig. 2) highlighting the influence of helix stabilization on the epitope recognition. As expected, the N-terminal peptide IN638, which displays good helix proportion but lacks the epitope, is devoid of marked antigenicity.

Figure 5.

Trifluoroethanol effect on CD spectra of the epitope peptide IN636. Percentages of TFE within H2O were varied from 0 to 90% by 10% increments. Inset corresponds to the titration curves reflecting the trifluoroethanol effects on helical contents of K159, P159, IN636, IN637, IN638 and IN642. Helical contents were determined by direct reading of ellipticities at 222 nm in CD spectra recorded at 25 °C with 50 µm peptide concentration (see Materials and methods).

Figure 6.

Prediction data from GOR and AGADIR, NMR data from both H2O and 20% TFE, relative to the synthetic peptides K159 and P159. Results are given according to the horizontal bars representation.

There are thus several pieces of evidence indicating that peptide recognition by anti-K159 is related to the intrinsic propensity of the epitope containing peptides for helix formation. Note the good concordance between structure predictions and experimental data. These stipulate for instance, that despite the Lys159 → Pro substitution in K159, the conformation of the functional epitope is maintained intact in the C-terminal part. This explains why antibodies recognize K159 and P159 equally well. Representative NMR spectra of these peptides are shown in Fig. 7. For C-terminal analogs, the gradation obtained from helix contents ( Fig. 5) parallels the gradation obtained from antibody bindings ( Fig. 2). The high affinity shown by antibodies toward IN and IN50–212 ( Figs 2 and 3) suggests that helicity of the epitope within the protein does not differ from that of the antigenic peptide. Note the weaker recognition of IN compared to IN, which can be explained by lesser accessibility of the epitope to antibodies in the entire protein.

Figure 7.

FingerprintdNN NOESY regions for K159 in aqueous solution (A) and 20% TFE (B). Sequential peaks are labeled by residue position in the sequence. Spectra were recorded at 20 °C, pH 3, at 300 ms time, using the WATERGATE solvent suppression [63].

Effects of anti-K159 on DNA binding activity of IN

When cleaving viral and cellular DNA and inserting viral DNA into the cell genome, IN interacts with DNA. The effect of antibodies on IN binding to DNA was assessed by filter-binding assays. These have been already used to study IN–DNA interactions [41] and to screen for IN inhibitors [41, 46, 47]. Here, IN was preincubated with antibodies at different antibodies/IN molar ratios before addition of oligonucleotide substrates, i.e. the S2/S3 DNA used in 3′ processing and autointegration assay, and DS [5] used in desintegration assay. DNA concentration was one-eighth that of IN as in the above biological assays. At this ratio the whole amount of IN is supposed to be fixed to the DNA as shown in previous experiments [48]. The whole mixture was filtered on a nitrocellulose membrane. The curves reported in Fig. 8A and B indicate that antibodies significantly alter the binding of IN to DNA. Antibodies present a high inhibition capacity, i.e. an IC50 value of 0.56 µm with S2/S3 and of 0.61 µm with DS. The effect of P159 and K159 on DNA binding activity of IN was also examined, only K159 exhibited an inhibition capacity (data not shown). The IC50 observed with K159 was in the same range as the one obtained for the integration inhibition with this same peptide [34]. No DNA retention is detected, even at high antibodies concentrations, in the absence of IN (data not shown), establishing that antibodies do not bind DNA and that they interact specifically with the enzyme.

Figure 8.

Curves of filter-binding assays showing the effect of anti-K159 on the binding of HIV-1 IN to DNA (see Materials and methods). The binding percentage was determined at different anti-K159/IN molar ratios. (A) Binding to S2/S3; (B) Binding to DS. In both cases: (▪) IN was preincubated with antibodies before adding DNA (•) IN was preincubated with DNA before adding antibodies (▴) DNA was preincubated with BSA. (C) Binding of IN to RAN DNA (▪).

IN was further preincubated with DNA prior to antibody addition. Figures 8A and B show that inhibition of IN binding to DNA is significantly altered, particularly in the case of the S2/S3 substrate. This strongly suggests that the IN segment 147–175 directly interacts with DNA and becomes thus less accessible to antibodies. The weaker recovery of IN binding to DNA with the DS substrate could be due to a minor affinity exhibited by IN for this DNA substrate. In that case, it can be easier for antibodies to shift IN from DS than S2/S3. Moreover, to test whether DNA binding inhibition is sequence specific, the assay was performed using a DNA random substrate, RAN. Results reported in Fig. 8C indicate that inhibition of IN binding to DNA is not sequence specific. This is not surprising since many studies have already shown that IN associates to nonspecific target DNAs [31, 49, 50]. At the end, BSA was used instead of antibodies as a negative control and data of Fig. 8A and B indicate that in these conditions the binding of IN to DNA remains unchanged.

Binding stoichiometry of anti-K159 to IN

The above experiments were carried out with very close to equal molar ratios of IN and antibody as the IC50 measured in the inhibition of the integration activity by the antibody suggested dissociation constants of the order of only several tens of nanomolar. At the same time the filter-binding assay provided mid points of titration close to 0.5 µm, largely above the Kds. Under such conditions one may assume that all the antibody is bound to integrase.

Curves of filter-binding assays presented in Fig. 8A and B have provided IC50 values of 0.56 µm and 0.61 µm for inhibition of the binding of IN to S2/S3 DNA and DS DNA, respectively, by anti-K159 (see above). Noticeably, preincubation of IN with up to threefold excess of antibody did not affect the IC50 value (data not shown).The IC50 values provide a binding stoichiometry of 2 : 1 (2 anti-K159-1 IN). Interestingly, similar results have been obtained by Müller et al. (1995) [48] by size exclusion chromatography on a complex formed between IN and a selected Fab fragment that inhibits the binding of IN to DNA. Since anti-K159 reacts with a distinct epitope on each molecule, it can be deduced that in the conditions selected for filter-binding assays IN is a homodimer.

Effects of anti-K159 on integration activity

First, we carried out a strand transfer assay in which the S1/S2 DNA substrate and a pSP65 plasmid DNA were preincubated with IN. S1/S2 and pSP65 mimic the viral and cellular genomes, respectively. Preincubation of IN with anti-K159 and pSP65 before adding labeled S1/S2 strongly alters the enzyme activity ( Fig. 9A). The IC50 is obtained at 25 n m of anti-K159. In the test of 3′ processing and autointegration, where the S3/S2 labeled DNA substrate plays the role of both the viral and cellular DNA, preincubation of IN with anti-K159 causes inhibition of both the 3′ processing and the autointegration, even at relatively low anti-K159 concentrations ( Fig. 9B). IC50 values are obtained at 16 n m of anti-K159 for both 3′ processing and autointegration. At the end, the effect of antibodies was tested on the desintegration activity of IN ( Fig. 9C) using a previously described assay [5]. Results indicate that anti-K159 greatly alters the desintegration with an IC50 of 6 n m.

Figure 9.

Effect of anti-K159 on IN-mediated strand transfer to the plasmid pSP65, 3′processing, autointegration and desintegration activities. IN was preincubated with anti-K159 prior to addition of DNA. (A) The gel shows the migration pattern of the plasmid ligated to labeled oligonucleotide. Lane 1, IN incubated with DNA; lane 2, DNA alone; lanes 3–7, IN incubated with DNA and in the presence of increasing anti-K159/IN molar ratios. (B) The positions of DNA substrates (21-mer), 3′ processing (19-mer) and autointegration products are indicated. Lane 1, IN incubated with DNA; lane 2, DNA alone; lanes 3–7, IN incubated with DNA and in presence of increasing anti-K159/IN molar ratios. (C) DNA dumbbell substrate (38-mer) and desintegration product (14-mer) are indicated. Lane 1, IN incubated with DNA; lane 2, DNA alone; lanes 3–7, IN incubated with DNA and in the presence of increasing anti-K159/IN molar ratios.

To demonstrate that antibodies specifically inhibit integrase function, we tried to abrogate the inhibitory effects of these antibodies by adding P159 as a competing peptide. This peptide binds to antibodies ( Fig. 2) but is devoid of inhibitory activity [34]. Fig. 10 clearly shows that P159 depletes the inhibitory activity of antibodies, leading to recovery of enzymatic activity. The specificity of IN inhibition by antibodies was confirmed using anti-insulin antibodies; here, no effect was observed on integration activity (data not shown).

Figure 10.

Recovery of IN inhibition in presence of the competing peptide P159. (A) Integration into pSP65. Lane 1, IN incubated with DNA; lane 2, DNA alone; lanes 3–7, IN incubated with 25 n m of anti-K159 and 25 n m, 250 n m, 2.5 µm, 25 µm and 250 µm of P159, respectively. (B) 3′ processing and autointegration. Lane 1, IN incubated with DNA; lane 2, DNA alone; lanes 3–8, IN incubated with 25 n m of anti-K159 and 25 n m, 250 n m, 2.5 µm, 25 µm, 125 µm and 250 µm of P159, respectively.


IN is required for integration of viral DNA into cellular genome [1, 2]. It is thus an attractive target for anti-HIV therapy, particularly since no cellular counterpart exists for this protein. During the last few years, considerable effort has been made to study functional and structural relationships in HIV-1 IN. It has been proven that IN is a globular protein requiring a multimeric structure to be functional [13, 14]. IN is known to form a highly stable interaction with viral long-terminal repeat substrate during in vitro integration reactions [13, 14]. Moreover, different antibodies have been raised against various epitopes of the protein [51–54] and several antibody fragments expressed in infected cells have been shown to decrease HIV-1 replication [55].

Antibodies raised against the peptide K159, reproducing the segment 147–175 located in the catalytic core of IN [34], present high specific affinity toward not only K159 but also toward IN and the central domain IN50–212. This signifies that K159 or its epitope portion mimics the helical conformation adopted by the segment 147–175 in the protein [27]. Because the antigenic peptide K159 further acts as an inhibitor of IN, the pending question was whether inhibition rested on interaction of the sole epitope portion with the enzyme. This assumption comes from recent observations indicating that proteins such as the human growth hormone–receptor complex [56] and the erythropoietin–receptor complex [57] may interact through small portions of the binding epitopes. As a corollary, a cyclic heptapeptide that mimics a small binding epitope, i.e. the CD4 domain 1CC′ loop, has been shown to block the CD4 biological functions [58]. Thus, it appears that small peptide portions reproducing functional epitopes may interfere with the protein interactions and inhibit biological functions, despite the commonly held view that interactions between proteins generally need large surfaces with numerous intermolecular contacts [59].

Conformational aspects of peptide–antibody recognition

In our previous report [34], we assumed that the inhibitory activity of K159 relied on the propensity of this peptide to adopt an amphiphilic helix and to form a stable coiled-coil structure with its helix counterpart on the protein leading to inhibition. This was suggested by the fact that the segment 147–175 corresponding to K159 displays an α helix structure at the surface of the crystal structure of IN50–212[27] and could be easily contacted by ligands such as antibodies, peptides or peptidomimetics. Here, Western blot binding data show that the surface exposure of the IN/IN50–212 segment 147–175 persists in solution. Another point is that antibodies recognize equally well K159 and its analog P159 with a helix-disrupting residue Pro in its middle. This posed the question of both the location of the functional epitope within the antigenic peptide and of how sensitive are antibodies to conformational changes occurring in the epitope. The use of truncated analogs reproducing the N-terminal part and the C-terminal part of K159 provided evidence that the functional epitope lies in the C-terminal portion, i.e. within residues 163 and 175 (GQVRDQAEHLKTA) of IN. As expected, the substitution of Ala by the helix breaker, Pro, in the middle of the epitope decreases significantly the binding affinity of antibodies, highlighting the importance of the epitope conformation. Actually, NMR data, in specifying that the epitope portion has the same propensity for helix stabilization in K159 and P159, explain why these two peptides are identically recognized by the antibodies. At the same time, the peptide gradation provided by the helix stabilization curves (excluding the N-terminal peptide lacking the epitope) parallels the peptide gradation provided by affinity data. These informations clearly indicate the impact of peptides helicity in the binding process. The stronger is the propensity of the C-terminal peptides to adopt an α-helical structure, the higher is the recognition of these peptides by antibodies. At the same time, the flexibility shown by the helix segment 147–175 in the protein [27] could explain the good cross-reactivity between the antigenic peptide K159 and the entire IN or IN50–212 exhibited by antibodies. Such features have been proven in other systems [60].

Conformational aspects of IN inhibition by peptides and anti-K159

None of the truncated analogs, i.e. neither IN638, which reproduces the N-terminal part of K159, nor IN636 and IN637, which both contain the functional epitopes, are able to block integration. These short peptides are less structured than their counterpart within the full-length peptide K159. Their inability to inhibit integrase could be thus, a priori, assigned to their weaker helix stability. Yet, the inability of P159 – in which the epitope displays same helix stabilization as in K159 – to inhibit IN suggests that the epitope by itself is insufficient to permit good binding to IN.

The ability of K159 both to form stable helix and autoassociate into dimeric structure strongly suggests that inhibition of IN proceeds through formation of a coiled-coil that maintains the peptide fixed to the protein segment 147–175. Actually, a number of at least three or four helical heptads are necessary to permit formation of stable coiled-coil structure, this requiring the cumulative effect of a sufficient number of hydrophobic and electrostatic contacts [61]. Of all the peptides, only antigenic K159 fulfills conditions of dimer formation as it contains the needed four helical heptads, moreover in a correct alignment. In P159, in addition to the local helix destabilization, the Pro residue creates a kink in the middle of the molecule generating a V-shaped conformation rendering P159 unable to line up over its full length and thus inapt to form a stable coiled-coil, as shown by previous cross-linking experiments [34].

Filter-binding assays showed that inhibition of the binding of IN to its DNA substrates occurs at relatively modest anti-K159 concentrations, supporting the proposition that the IN segment 147–175 is implicated in DNA binding. The latter observation conforms to previous findings indicating (a) that a peptide reproducing the IN segment 139–152 cross-links to viral DNA [62] and (b) that residues K156 and K159 play a critical role for binding of IN to DNA [32]. Moreover, single-chain variable antibody fragments (SFvs) raised against the domain 145–185 significantly inhibit the HIV-1 replication in human T-lymphoid cells [54]. It can be therefore assumed that anti-K159 inhibits the IN activity by interfering with the association of the enzyme to substrate DNA. A similar mechanism is expected for the antigenic peptide K159, this exhibiting the capacity, like anti-K159, of inhibiting the binding of IN to DNA. Obviously, binding of anti-K159 or antigenic K159 to the target segment 147–175 could also alter the properties of E152, which belongs to the catalytic triad [D, D, E], including access of E152 to the DNA cleavage site.

Finally, an important feature is the stoichiometrical fixation of anti-K159 to IN, observed in the filter-binding assays with DNA substrates. The derived IC50 values indicate that two anti-K159 molecules react with IN, implying that the enzyme is dimeric in our experimental conditions.

Concluding remarks

The present data demonstrate a relationship between the structure and the function of the peptide domain 147–175 of the catalytic core of IN, critical for integration activity. This domain is exposed at the protein surface, both in crystal and in solution, and encompasses a peptide portion (163–175) that is an important functional epitope. Monospecific antibodies, raised against a synthetic peptide reproducing the peptide domain 147–175, strongly and specifically react with the epitope portion. These antibodies are shown to block the integration/desintegration activity of IN at low concentrations. This is an important result, confirming the importance of the 147–175 domain for the enzyme activity. The peptide 147–175 acts also as an inhibitor of IN, while a truncated peptide mimic of the epitope and a Pro-containing analog of the full peptide, appear too short and distorted, respectively, to insure IN inhibition. Owing to its marked propensity for coiled-coil formation, the full-length peptide is prone to interact with its helical counterpart domain 147–175 of IN according to a coiled-coil structure. This indicates that inhibition of integration proceeds through the binding of the antibody or the antigenic peptide to IN and not to DNA. In filter-binding assays, the complex of IN with the antibody or the peptide is unable to bind DNA, further reflecting the essential role played by the 147–175 domain toward the binding of IN to DNA. Other important information provided by the use of antibodies is that IN displays a dimeric structure in the DNA environment.

In vivo studies on HIV infected cells will tell us whether or not such antibodies are able to react with intracellular IN, the number of IN molecules seemingly being very small. Possibly, antibodies exhibiting high specificity against the 163–175 segment of IN when expressed in vivo will efficiently block the integration process. We also hope that structural information obtained from the conformational analysis of peptides will be helpful for developing binding models and preparation of inhibitors directed against the central domain of HIV-1 IN.


We would like to thank Dr E. Lescot for the synthesis of DNA oligonucleotides, J. P Levillain, M. Le Maout, and Y. Smith for assistance in peptide synthesis and antibody preparation, and A. Deroussent for the ESIMS analysis. We would also like to thank E. Habert (Rhône-Poulenc Rorer, Vitry Sur Seine) for generously providing us with purified HIV-1 integrase. This project was supported by a grant from the Fondation pour la Recherche Médicale (SIDACTION). R.G.M has a fellowship from the Agence Nationale de Recherche sur le SIDA (ANRS) and D.K from SIDACTION.