Mechanisms of Human Immunodeficiency Virus Type 1 Concerted Integration Related to Strand Transfer Inhibition and Drug Resistance

Abstract

The “strand transfer inhibitors” of human immunodeficiency virus type-1 (HIV-1) integrase (IN), so named because of their pronounced selectivity for inhibiting strand transfer over 3′ OH processing, block virus replication in vivo and ex vivo and prevent concerted integration in vitro. We explored the kinetics of product formation and strand transfer inhibition within reconstituted synaptic complexes capable of concerted integration. Synaptic complexes were formed with viral DNA donors containing either two blunt ends, two 3′-OH-processed ends, or one of each. We determined that one blunt end within a synaptic complex is a sufficient condition for low-nanomolar-range strand transfer inhibition with naphthyridine carboxamide inhibitors L-870,810 and L-870,812. We further explored the catalytic properties and drug resistance profiles of a set of clinically relevant strand transfer inhibitor-resistant HIV-1 IN mutants. The diketo acids and naphthyridine carboxamides, mechanistically similar but structurally distinct strand transfer inhibitors, each select for a distinct set of drug resistance mutations ex vivo. The S153Y and N155S IN resistance mutants were selected with the diketo acid L-841,411, and the N155H mutant was selected with L-810,812. Each mutant exhibited some degree of catalytic impairment relative to the activity of wild type IN, although the N155H mutant displayed near-wild-type IN activities. The resistance profiles indicated that the S153Y mutation potentiates susceptibility to L-870,810 and L-870,812, while the N155S mutation confers resistance to L-870,810 and L-870,812. The N155H mutation confers resistance to L-870,810 and potentiates susceptibility to L-841,411. This study illuminates the interrelated mechanisms of concerted integration, strand transfer inhibition, and resistance to strand transfer inhibitors.

The replication of human immunodeficiency virus type 1 (HIV-1) is contingent upon a highly conserved catalytic event in which retroviral integrase (IN) integrates the viral cDNA genome into the host chromosome (2, 4). IN associates with and juxtaposes the U3 and U5 termini of the linear blunt-ended viral genome, forming a cytoplasmic nucleoprotein complex termed the preintegration complex (PIC). IN, in an event termed 3′-OH processing, catalyzes the removal of dinucleotides from the 3′ end of each long terminal repeat (LTR) terminus. Following nuclear localization, the PIC associates with host chromosomes and coordinates a nucleophilic attack in which the 3′ hydroxyl moieties of the CA nucleotide attack the phosphodiester backbone of the host chromosomal DNA. In the case of HIV-1, this results in a 5-bp stagger between the sites of covalency. The components of the PIC dissociate from the virus-host junction, allowing the host-mediated removal of the 5′ overhangs and subsequent host-mediated filling in of the 5-bp gaps (5). Concerted integration is readily reconstructed in vitro (Fig. 1) through the use of recombinant IN, synthetic LTR substrates (0.5 to 4 kbp), and supercoiled target DNA, which represents the genomic target (12, 14, 15, 18, 19, 21-24).

FIG. 1.

Schematic representation of concerted integration in vitro. Preincubation of IN and the LTR donor at 14°C followed by further incubation at 37°C results in the assembly of an SC, within which 3′-OH processing occurs. The SC subsequently associates with the supercoiled DNA target and covalently attaches the two viral LTR ends (red) to opposite complements of the target (circle outlined in black), preferentially along the major groove, producing the STC. Deproteinization of the STC produces the FS or concerted DNA product. In addition, CHS, D-D, and Y-type structures are also formed, presumably as a result of malformed or misaligned nucleoprotein complexes.

Chen and Engelman have shown that, within the HIV-1 PIC, the 3′-OH processing of the U3 end of the viral genome occurs in a manner independent of the functionality of the U5 end, and vice versa (3). This was in contrast to previous results with Moloney murine leukemia virus showing that 3′-OH processing of either end within the PIC required that both DNA ends be functional (17). In addition, the report on HIV-1 suggested that within the PIC, 3′-OH processing necessarily occurs in the context of a higher-order nucleoprotein complex, that is, after end synapsis (3). We previously showed, using native agarose gel electrophoresis, that the population of early transient synaptic complexes (SC) formed with a blunt-ended DNA substrate and HIV-1 IN contained ∼85% blunt ends, indicating that 3′-OH processing occurs gradually within the SC (18), while strand transfer occurs rapidly following 3′-OH processing of both DNA ends, producing the strand transfer complex (STC) (15, 18). The juxtaposed viral DNA ends within the SC appear constrained from strand transfer activity (18), similar to what has been observed in the PIC (3), until the ultimate processing of both DNA ends.

The critical role of IN in the HIV-1 replication cycle, coupled with the lack of analogous host proteins, strongly implicates IN as a viable chemotherapeutic target. There is a strong impetus for the identification and development of clinically amenable IN inhibitors (20). The first such promising compounds belonged to the diketo acid (DKA) class (1, 6, 11, 20). The inhibitors of this class possess a DKA moiety, which purportedly coordinates Mg²⁺ in the active site of the catalytic core domain and thereby precludes the possibility of binding to a genomic target (7). The DKAs exhibit pronounced selectivity for inhibition of strand transfer over inhibition of 3′-OH processing and therefore have been called “strand transfer inhibitors” (10). It has been shown recently that these inhibitors show a ∼10- to 20-fold preference for inhibition of in vitro reactions proceeding with blunt-ended substrates over those with 3′-OH-processed substrates, suggesting that in vivo, inhibition occurs within the cytoplasmic PIC (18). We further explore this hypothesis by testing whether only one blunt end in the SC is required for effective inhibition of concerted integration at low nanomolar concentrations of the naphthyridine carboxamide (NCA) strand transfer inhibitors L-870,810 and L-870,812 (9, 11).

The DKA inhibitor L-841,411 and the NCA inhibitor L-870,810 give rise to different resistance mutations ex vivo (9). The two classes of mutations mapped to two different regions of the IN catalytic core domain, each distal to the active site. It was further shown that mutants resistant to L-841,411 retained susceptibility to L-870,810, and vice versa (9). The only mutation violating this trend and conferring cross-resistance was the L-841,411-selected mutation N155S. The DKA L-841,411 selects ex vivo for S153Y, a non-cross-resistant mutation, and N155S, which has been shown to confer cross-resistance to the NCA L-870,810 ex vivo (9). In another study, under selective pressure in vivo with virus-infected macaques, the NCA L-870,812 gave rise to the N155H mutation (11). In this study, we investigate the concerted-integration activities of wild-type (wt) III-B, N155H, N155S, and S153Y INs and their responses to L-870,810, L-870,812, and L-841,411. We conclude that N155S is unique in its ability to confer resistance to all three inhibitors investigated here, and we further show that the S153Y mutation, which confers resistance to L-841,411, potentiates susceptibility to L-870,810 and L-870,812. The L-870,812-selected mutation N155H confers cross-resistance to L-870,810 but potentiates susceptibility to L-841,411.

MATERIALS AND METHODS

Viral DNA substrates.

DNA substrates containing either blunt-ended or preprocessed (recessed) wt HIV-1 U5 LTR termini were prepared as described elsewhere (18, 19).

Purification of HIV-1 IN.

wt IN from strain NY (21, 22) and wt, N155H, N155S, and S153Y INs derived from strain III-B (9, 11) were expressed in Escherichia coli BL21(DE3) cells and purified to near-homogeneity (21). Sequence analysis of wt NY and III-B INs revealed six differences in their amino acid contents. The residue changes between the NY and III-B INs were at positions 10 (E to D), 113 (V to I), 124 (T to A), 151 (I to V), 232 (D to N), and 234 (V to L). The level of synthesis of full-site (FS) products by wt III-B IN was approximately 60% of that routinely observed with wt NY IN.

Integration assay.

The integration assay was performed as described elsewhere (18, 21). Briefly, IN was preincubated with the donor substrate (0.5 nM) in the presence of 20 mM HEPES (pH 7.0), 5 mM dithiothreitol, 10 mM MgCl₂, 25 μM ZnCl₂, 100 mM NaCl, and 10% polyethylene glycol (6,000 Da) at 14°C for 15 min. The IN and donor concentrations are given below for each experiment. The standard reaction volume was 100 μl. The reactions were initiated by the addition of supercoiled target DNA (1.5 nM) and subsequent incubation for 2 h at 37°C. pBSK2Δ-Zeo (2.69 kbp) (16) was used as the target. Aliquots were taken for native analysis (see below) prior to deproteinization. The remaining reactions were stopped by the addition of EDTA, sodium dodecyl sulfate, and proteinase K the final concentrations of 25 mM, 0.5% and 1 mg/ml, respectively. Equivalent quantities of each sample (∼15,000 cpm) were subjected to either 0.7% or 1.5% agarose gel electrophoresis in Tris-borate-EDTA buffer for 10 or 16 h, respectively, at 100 V. Note that the circular half-site (CHS) product (Fig. 1) exhibits different migration rates on 0.7% and 1.5% gels, relative to other products (18). The gels were dried and exposed to a PhosphorImager screen, and the DNA products were quantified using a Storm 860 imaging system (Amersham Biosciences).

Analysis of native samples on agarose gels.

In addition to the analysis of deproteinized samples, aliquots were taken out of the reaction mixture for native analysis prior to deproteinization. The reactions were stopped by adding EDTA to 25 mM, and the products were subjected to electrophoresis on 0.7% agarose gels in Tris-borate-EDTA buffer at 4°C (15, 18, 19).

IN inhibitors.

The strand transfer inhibitors L-870,810, L-870,812, and L-841,411 have been described previously (9, 11). Stocks (10 mM) of each inhibitor were made in 100% dimethyl sulfoxide and stored in small aliquots at −70°C. After preincubation of the IN-DNA complexes at 14°C, the target and then the inhibitors were added. The inhibitors were added ∼20 s after the addition of the target. The samples were immediately incubated for 2 h at 37°C. As indicated in the reports of some experiments, L-870,810 was added after assembly of IN-DNA complexes at 14°C for 15 min. The inhibitor was further incubated at this temperature for either 10 or 20 min prior to addition of the target and incubation at 37°C.

RESULTS

Synthesis and visualization of intermediate and terminal nucleoprotein complexes in the concerted-integration pathway.

With in vitro assays that contain wt NY IN and a 1.6-kbp blunt-ended U5 LTR substrate (Fig. 1), the synthesis of integration products proceeded in a time-dependent fashion and was essentially complete after 2 h of incubation at 37°C, with a maximum FS-product yield of 23% (Fig. 2B, lane 7). It is possible to monitor the course of these reactions by arresting enzymatic activities at various time points through the addition of EDTA. These samples can then be either directly subjected to agarose gel electrophoresis or first deproteinized and then subjected to electrophoresis (referred to here as native and deproteinized samples, respectively) (Fig. 2A and B). In the former case, at 15 to 45 min of incubation at 37°C, the transient appearance of the SC migrating at ∼3.5 kbp was evident (Fig. 1 and 2A). It has been shown through 2-dimensional electrophoresis that the SC consists of free donors under noncovalent, IN-mediated juxtaposition (18). With further incubation, the SC disappeared upon association with target DNA and underwent conversion to the STC (Fig. 2A, lanes 4 to 7), the terminal nucleoprotein product of concerted integration (15). Slower-migrating, higher-order complexes of the SC, designated H-SC, and of the STC were also evident (18). Upon deproteinization, IN dissociated from the STC, leaving the linear FS product (Fig. 2B) (15, 18, 19). This implicates the SC as a precursor to the STC.

FIG. 2.

Visualization of the concerted-integration pathway in vitro by agarose gel electrophoresis of native and deproteinized samples. (A) NY IN (20 nM) was preincubated with the 1.6-kbp ³²P-labeled 5′-U5 blunt-ended donor (0.5 nM) at 14°C for 15 min, followed by addition of the target (1.5 nM) and further incubation at 37°C for various times (given above the gel). The reactions were arrested by the addition of EDTA, and the reaction products were subjected to native 0.7% agarose gel electrophoresis. Lane 1, 1-kbp ladder (Promega); lane 2, control (C) without IN; lanes 3 through 7, individual reactions in which IN has been arrested with EDTA (25 mM) at increasing times of incubation. The SC, STC, and H-SC are identified on the left. The filled circle on the right indicates slower-migrating forms of the SC and STC (18). (B) Aliquots of the same samples for which results are shown in panel A were deproteinized and electrophoresed on a 0.7% agarose gel. The DNA products are identified on the right. At 120 min (lane 7), the percentages of the donor incorporated into FS, CHS, and D-D products were 23, 5, and 20%, respectively.

In addition to the intermediates and products in the concerted-integration pathway, there are products that result from side reactions arising from spurious nucleoprotein complexes (14, 15, 21, 22). The CHS product presumably arises when IN complexed with a single donor end attacks one complement strand of the target (Fig. 1 and 2A and B). Donor-donor (D-D) and Y-type structures are thought to arise from misaligned SC-like structures that integrate one LTR end into another within the same complex (Fig. 1 and 2B) (19).

Differential rates of formation of the FS product associated with an SC containing either two blunt ends, two recessed ends, or one of each.

Previous results show dissimilar kinetics of FS product formation, depending on whether the donor substrate possesses blunt or 3′-OH-processed ends (14). This led us to ask several questions regarding the kinetics of FS product formation in a reaction mixture containing a mixed population of blunt-ended and 3′-OH recessed-ended donor substrates. It should be noted that in the STC, both DNA ends have undergone 3′-OH processing and strand transfer, and our nomenclature here refers to the original identity of the substrate. In order to resolve the possible FS products, we selected recessed (R) and blunt-ended (B) 1.1-kbp and 1.6-kbp donor substrates, respectively. This scheme permits the resolution of deproteinized products on a 1.5% agarose gel: B-B, B-R, and R-R FS products necessarily arise from their respective B-B, B-R, and R-R SC precursors (Fig. 3) (18).

FIG. 3.

Differential rates of FS product formation with SC containing either two blunt-ended donors (B), two recessed-ended donors (R), or one of each (B+R). (A) Three reaction mixtures were prepared: one containing the 1.6-kbp U5 blunt-ended donor at 0.5 nM (lanes 2, 5, 8, 11, and 14), one containing the 1.1-kbp U5 recessed-ended donor at 0.5 nM (lanes 3, 6, 9, 12, and 15), and one containing 0.25 nM blunt-ended donor and 0.25 nM recessed-ended donor (lanes 4, 7, 10, 13, and 16). Lane 1 contains molecular size markers; lanes 2 to 16 contain 20 nM NY IN. Lanes 2 to 4, control reactions at 14°C for 15 min, stopped prior to incubation at 37°C; lanes 5 to 7, 8 to 10, 11 to 13, and 14 to 16, reactions stopped with 25 mM EDTA after 20, 40, 60, and 120 min of incubation, respectively. The samples were deproteinized and subjected to 0.7% agarose gel electrophoresis. The various products are indicated. (B) The percentages of the donors incorporated into FS products for the single-donor reactions (with either blunt-ended or recessed-ended donors) are plotted against time. (C) The percentages of the donors incorporated into B-B, B-R, and R-R FS products in the mixed-donor reactions (panel A, lanes 4, 7, 10, 13, and 16) are plotted against time.

We carried out three reactions, the first with 0.5 nM blunt-ended 1.6-kbp substrate, the second with 0.5 nM 1.1-kb recessed-ended substrate, and the third with 0.25 nM each (Fig. 3A, lanes 2 to 4). All reaction mixtures contained 15 nM NY IN, and reactions were stopped sequentially at 0, 20, 40, 60, and 120 min. The deproteinized samples were run on a 1.5% agarose gel (Fig. 3A). The formation of a B-B FS product in the single-donor reaction with a blunt-ended substrate (Fig. 3A, lanes 2, 5, 8, 11, and 14) followed the expected course, beginning between 20 and 40 min at 37°C and approaching saturation at 120 min (Fig. 3B). In the single-donor reaction with a blunt-ended substrate, the final level of donor incorporation into the FS product after 120 min of incubation was 16% (Fig. 3B). In the single-donor reaction with a recessed-ended substrate (Fig. 3A, lanes 3, 6, 9, 12, and 15, and B) and in the mixed-donor reaction (Fig. 3A, lanes 4, 7, 10, 13, and 16, and C), the formation of an R-R FS product proceeded immediately, occurring much more rapidly than that of a B-B FS product (Fig. 3C). In the reaction containing only the recessed-ended donor, the final level of incorporation into the FS product after 120 min was 8% of all the products produced (Fig. 3B). The formation of a B-R FS product in the mixed reaction followed an intermediate course, appearing more rapidly than the B-B FS product but more slowly than the R-R FS product (Fig. 3A, lanes 4, 7, 10, 13, and 16, and C). Ultimately, at 120 min, the quantities of the B-B and B-R FS products in the mixed reaction were nearly equivalent (Fig. 3C). The identities of the B-B, B-R, and R-R linear FS products from the mixed reactions were confirmed by 2-dimensional electrophoresis (data not shown) (18). We repeated the mixed-donor experiment using a 1:1.5 ratio of the blunt-ended to the recessed-ended donor and observed the same ordering in the rates of formation of the various possible FS products (data not shown). This indicates that the differential rates of formation do not arise solely from differences in the abundances of the various possible complexes. The results suggest that (i) the processing of DNA ends is the rate-limiting step in the FS pathway; (ii) processing occurs primarily in a sequential fashion within the SC; and (iii) concerted integration is contingent upon the ultimate processing of both DNA ends within the SC.

Differential susceptibilities to strand transfer inhibition for blunt-ended and preprocessed substrates when each is present in a single SC.

In an earlier report, we showed that the potency of strand transfer inhibitors is highly dependent on the presence of blunt ends within an SC (18). In reactions containing a recessed-ended substrate and wt NY IN, the 50% inhibitory concentrations (IC₅₀s) associated with concerted integration were ∼10- to 20-fold higher than those in reactions containing a blunt-ended substrate. This observation led us to two conclusions: (i) strand transfer inhibition requires a transient structural intermediate that arises either before or during the 3′-OH-processing step; and (ii) the action of these inhibitors is restricted primarily to the cytoplasmic PIC, since this is where most processing occurs (3). This information led us to ask if effective inhibition by a strand transfer inhibitor, at a low nanomolar concentration, requires the presence of one or two blunt ends within an SC capable of concerted integration.

To address this question, we carried out titration with L-870,810 between 30 nM and 1,000 nM among aliquots of a reaction mixture containing 20 nM NY IN and 0.25 nM each 1.1-kbp recessed-ended and 1.6-kbp blunt-ended donors (Fig. 4). The deproteinized samples were run on a 1.5% agarose gel in order to facilitate the resolution of strand transfer products as shown in Fig. 3A. The control reaction mixtures contained 0.5 nM blunt-ended 1.6-kbp substrate or 0.5 nM 1.1-kbp recessed-ended substrate (Fig. 4A, lanes 3 and 4, respectively). In the control reaction with the blunt-ended substrate, after 120 min, the level of incorporation into the FS product reached 13.1%, while in the control reaction with the recessed-ended substrate, it reached 5.8%. In the mixed control reaction containing 0.25 nM each blunt-ended and recessed-ended substrate with no L-870,810 (Fig. 4A, lane 5), the levels of B-B, B-R, and R-R incorporation into FS products reached 3.0%, 3.1%, and 2.1%, respectively, while the levels of incorporation into CHS products for the reactions with blunt- and recessed-ended substrates were 5.8% and 2.8%, respectively. The inhibition profiles for these three complexes (Fig. 4B) indicate that for the B-B, B-R, and R-R FS products, the IC₅₀s were ∼32 nM, ∼60 nM, and >1,000 nM, respectively. For the B-CHS and R-CHS products, the IC₅₀s were 230 nM and >1,000 nM, respectively. The B-B FS reaction was subject to inhibition, as previously described (18). The B-R FS reaction was slightly less sensitive to L-870,810 than the B-B reaction; however, it still fell in the effective low nanomolar range. We observed similar trends in identical experiments substituting L-870,812 for L-870,810. The IC₅₀s of L-870,812 for the B-B FS, B-R FS, and B-CHS products were ∼85 nM, ∼368 nM, and ∼1,350 nM, respectively (data not shown). The R-R FS and R-CHS products in the mixed-donor reactions were ineffectively inhibited, with IC₅₀s of >1,000 nM. In summary, it appears that a single blunt end within an SC is a sufficient condition for effective strand transfer inhibition of concerted integration in vitro and likely in vivo.

FIG. 4.

Differential susceptibilities to strand transfer inhibition with SC containing either two blunt-ended donors (B-B), two recessed-ended donors (R-R), or one of each (B-R). (A) Lane 1 contains molecular weight markers. Lane 2, designated “C,” contains 0.25 nM each 1.6-kbp U5 blunt-ended donor and 1.1-kbp U5 recessed-ended donor in the absence of IN. Lane 3, designated “B,” contains 0.5 nM blunt-ended donor, and lane 4, designated “R,” contains 0.5 nM recessed-ended donor. Both samples (lanes 3 and 4) contained 20 nM NY IN and 1.5 nM target and were incubated for 2 h at 37°C. Lanes 5 to 12, with 20 nM IN, contain 0.25 nM each blunt-ended and recessed-ended donors, as well as 0, 30, 50, 100, 300, 500, 750, or 1,000 nM L-870,810, respectively. Incubation was performed for 2 h at 37°C. The B-B, B-R, and R-R FS products are indicated on the right, as are the B-CHS and R-CHS products. (B) Percentages of inhibition of the indicated FS and CHS products upon addition of L-870,810. Error bars indicate the standard deviations about the means from data obtained in three independent experiments. A lack of error bars at a particular inhibitor concentration indicates that only one experiment was performed at that particular concentration.

We determined that the order of addition of the inhibitor to the reaction mixture did not affect the IC₅₀s obtained with L-870,810. In the experiment described above, the inhibitor was added immediately after the addition of the target (Fig. 4) (18). In independent experiments, IN-DNA complexes were assembled for 15 min at 14°C using either recessed- or blunt-ended donor substrates. Varying concentrations of L-870,810 were added and incubated for an additional 10 or 20 min at 14°C prior to the addition of the target and subsequent strand transfer analysis at 37°C. The same IC₅₀s were obtained upon the addition of the inhibitor to assembled IN-DNA complexes (data not shown) as were observed for both blunt- and recessed-ended substrates when the inhibitor was added immediately after the target.

Strand transfer capacities of selected strand transfer inhibitor-resistant IN mutants.

Using methodologies described previously, we set out to determine the catalytic properties and drug resistance profiles of a set of inhibitor-resistant mutants of HIV-1 strain III-B IN (9-11). We standardized our reactions by using wt IN and N155H, N155S, and S153Y IN mutants with the 1.6-kbp U5 blunt-ended substrate and analyzing strand transfer products on 0.7% agarose gels. We found that increasing the incubation time from the standard 120 min to 240 min resulted in a marked increase in the level of synthesis of FS products for the N155H, N155S, and S153Y mutants but markedly less synthesis for wt IN (Fig. 5A and B). The final levels of incorporation into FS products after 240 min were 11.8%, 10.2%, 6.5%, and 4% for the wt and the N155H, N155S, and S153Y mutants, respectively (Fig. 5C). SC and H-SC formation associated with the N155S and S153Y mutants was decreased and delayed relative to that observed with the wt and the N155H mutant on native agarose gels, suggesting an assembly defect with the blunt-ended substrate (data not shown).

FIG. 5.

Kinetics of FS product formation using IN from wt III-B or the N155H, N155S, or S153Y mutant. (A) Reaction mixtures contained 0.5 nM 1.6-kbp U5 blunt-ended donor, 1.5 nM target, and 20 nM wt III-B or N155H IN. Reactions were stopped with 25 mM EDTA at various time points up to 240 min in lanes 3 to 8 and 9 to 14, respectively. The reactions were stopped at 10, 20, 30, 60, 120, and 240 min. The samples were deproteinized. The FS, CHS, and D-D products are indicated on the left. Lane 1 contains markers; lane 2, designated “C,” lacks IN. (B) Same as panel A except that N155S (lanes 1 to 6) and S153Y (lanes 7 to 12) INs were analyzed at the same time points. (C) Graphical representation of the percentage of the donor incorporated into the FS product at each time point for wt and IN mutants for which results are shown in panels A and B.

To further illuminate the cause of the varying degrees of delayed synthesis, we carried out an identical time course experiment using the 1.1-kbp recessed-ended substrate (Fig. 6A). In all cases, the formation of the FS product proceeded with greater rapidity than it did in reactions with the blunt-ended substrate, with the maximum percentage of incorporation into the FS product (Fig. 6B). The maximum percentages of incorporation after 240 min were 4.0%, 4.7%, 5.0%, and 3.5% for the wt, N155H, N155S, and S153Y INs, respectively (Fig. 6B). It should be noted that with the recessed-ended substrate, the activities associated with incorporation of all the enzymes into FS and CHS products are nearly equivalent. This suggests the existence of a defect in 3′-OH processing in each mutant analyzed here, resulting in a delay rather than an incapacitation of processing (Fig. 5).

FIG. 6.

Kinetics of FS product formation with reactions containing the 3′-OH recessed-ended donor using wt III-B, N155H, N155S, or S153Y IN. (A) Reaction mixtures contained 0.5 nM 1.1-kbp U5 recessed-ended donor, 1.5 nM target, and 20 nM wt III-B, N155H, N155S, or S153Y IN. Reactions were stopped with 25 mM EDTA at various time points up to 240 min (lanes 2 to 6, 7 to 12, 13 to 18, and 19 to 24, respectively). The wt reaction was stopped at 10, 20, 30, 120, and 240 min. All other reactions were stopped at 10, 20, 30, 60, 120, and 240 min. The FS, CHS, and D-D/Y-type products are indicated. Lane 1, designated “C,” contains the recessed donor without IN. (B) Graphical representation of the percentage of the donor incorporated into the FS product at various time points with wt or mutant IN.

Resistance profiles for IN mutants resistant to strand transfer inhibitors.

The IC₅₀s for the concerted integration of wt III-B and its inhibitor-resistant mutants are summarized in Table 1. All of the assays were performed with the 1.6-kbp blunt-ended U5 LTR substrate. We carried out at least three titration experiments for each such combination and averaged the percentages of inhibition obtained for each concentration of inhibitor, taking the error at each concentration to be the standard deviation among all repeated measurements at that concentration. We used SigmaPlot to determine the best-fit dose-response curves, from which we determined the IC₅₀s. From the spread around the best-fit curve, SigmaPlot calculates and, upon command, displays the 95% confidence interval associated with the best-fit nonlinear regression. From this interval, around the IC₅₀s, we calculated the error in the IC₅₀ within the 95% confidence interval. This error was then used to calculate the error in the level of resistance. The IC₅₀s for each combination of mutant and inhibitor are summarized in Table 1. The level of resistance or susceptibility of each combination to a given inhibitor is summarized in graphical format (Fig. 7).

TABLE 1.

IC₅₀s of various strand transfer inhibitors for the concerted-integration activities of wt IN and inhibitor-resistant IN mutants^a

Inhibitor	IC50 (nM) for IN
	wt III-B	N155H mutant	N155S mutant	S153Y mutant
L-870,810	142 ± 10	378 ± 14	228 ± 39	93 ± 11
L-841,411	227 ± 25	122 ± 12	338 ± 32	288 ± 32
L-870,812	268 ± 46	2,134 ± 262	878 ± 115	167 ± 25

^aIC₅₀s for wt IN and each IN mutant were obtained from dose-response curves generated by averaging the percentage of inhibition of FS product formation at a given concentration of inhibitor from at least three experiments for each possible permutation of mutant and inhibitor. Reaction conditions using a blunt-ended donor and IN are described in the legend to Fig. 3. For each independent experiment, the percentage of inhibition at a given concentration of inhibitor was taken to be the percentage of the decrease in FS product formation relative to that in the absence of inhibitor. Best-fit nonlinear, logistic, dose-response curves were generated using SigmaPlot.

FIG. 7.

In vitro resistance or potentiation of susceptibility to L-870,810, L-841,411, and L-870,812 associated with the N155H, N155S, and S153Y IN mutations. The resistance (n-fold) for a given mutant-inhibitor combination is calculated relative to the IC₅₀ of the inhibitor for wt III-B. For inhibitor-resistant mutants, resistance (n-fold) is calculated as (IC₅₀ for the mutant)/(IC₅₀ for the wt). For inhibitor-susceptible mutants, susceptibility (n-fold) is calculated as (IC₅₀ for the wt)/IC₅₀ for the mutant).

The resistance profiles show that the N155S mutation is unique in its ability to confer resistance to both NCA and DKA inhibitors. In all other cases explored here, a mutation conferring resistance to one class of strand transfer inhibitor potentiates susceptibility to the other. It should be noted that the N155S mutation is not a naturally occurring polymorphism (13), probably due to negative selection, and is therefore found less frequently in patients.

DISCUSSION

Processing of LTR ends within the SC occurs in a sequential fashion.

We use a time course experiment in which reaction mixtures contain equimolar concentrations of blunt-ended and recessed-ended substrates to reveal the differential rates of formation of the FS product associated with various possible SC: R-R products were formed at the highest rate, followed by B-R and then B-B products. The R-R SC associates rapidly with the target, proceeding immediately to the FS product (Fig. 3). The B-B SC forms a relatively long-lived, intermediate SC within which processing occurs gradually (∼16% of ends are processed in 20 min) (18) (Fig. 2 and 3), and only upon the ultimate processing of both blunt ends does the complex proceed to the FS product. The B-R SC proceeds to the FS product with greater rapidity than the B-B SC but more slowly than the R-R SC (Fig. 3). This result poses the question: why does an SC containing one donor each with blunt and recessed ends proceed to the FS product more rapidly than complexes containing two blunt ends? The simplest and most compelling answer is that in the mixed B-R SC, concerted integration is contingent upon the processing of just one end, while in the B-B SC, both ends must undergo 3′-OH processing. If processing occurred in a simultaneous fashion within the B-B SC, one would expect the kinetics of FS product formation to be identical with that associated with the B-R SC, since in each case, one processing event and a subsequent concerted-integration event takes place. However, this is not what occurs. The difference in the early rates of formation between B-B and B-R FS products suggests the opposite: processing within the B-B SC occurs in a sequential fashion.

Sequential processing of U3 and U5 ends would give rise to a complex containing one potentially reactive end and one nonreactive end. It would be advantageous to the virus to somehow constrain the activity of the single reactive end until both DNA ends undergo 3′-OH processing, so as to prevent suicidal autointegration or single-end integration (3). In fact, evidence of such a constraint would serve to support the idea that processing occurs sequentially within the HIV-1 PIC (3). Our results indicate that such a constraint does exist. Without such a constraint, the recessed end within the B-R SC would readily incorporate into a CHS product. The assays carried out here employ only U5 substrates, and it is possible that the presence of a U3 LTR within an SC may affect the outcome. However, we have not observed marked differences in the behavior of our in vitro assays when we used solely U3 substrates or a combination of U3 and U5 substrates (18; also unpublished data).

Sufficient conditions for low-nanomolar-range strand transfer inhibition.

To investigate the conditions sufficient for low-nanomolar-range strand transfer inhibition, we carried out titrations with L-870,810 in reaction mixtures containing wt NY IN and an equimolar, mixed population of blunt and recessed ends. The control reaction, in the absence of inhibitor, resulted in the expected B-B, B-R, and R-R FS products (Fig. 4, lane 5). Upon the addition of increasing concentrations of L-870,810, there appeared a rapid decline in the formation of all products arising from the blunt-ended substrate, and ineffective inhibition of R-R FS and R-CHS products arising from the recessed-ended substrate (Fig. 4A, lanes 6 to 12). The potency of L-870,810 in inhibiting FS product formation was highest for B-B products, intermediate for B-R products, and lowest for R-R products (Fig. 4B). The B-B and B-R SC are comparable in their sensitivities to L-870,810, with IC₅₀s of ∼32 nM and ∼60 nM, respectively. One possible reason for the difference in sensitivity between the B-B and B-R complexes is that an SC containing one blunt and one recessed end is simply less accommodating to a molecule of L-870,810. Another possibility is as follows. Under the assumptions that processing occurs sequentially within the SC and that strand transfer inhibition is contingent upon an intermediate that arises during processing (20), it is plausible that within the B-B SC there are twice as many processing events and therefore twice the number of opportunities for the inhibitor to bind. In support of this idea, there is almost a twofold difference between the IC₅₀s for the B-B and B-R FS products. These results, taken together, suggest that the presence of one blunt end within an SC is a sufficient condition for effective strand transfer inhibition in vitro and potentially within the PIC in vivo (18).

The catalytic properties of clinically relevant HIV-1 strand transfer inhibitor-resistant IN mutants reveal a potential dual mechanism of drug resistance.

In our preliminary analyses of inhibitor-resistant N155H, N155S, and S153Y III-B IN mutants, it appeared that their activities were by far insufficient to permit valid evaluation, except for the N155H mutant, which displayed ∼60% of wt activities for FS and CHS products after 120 min with a blunt-ended U5 substrate (Fig. 5). For all mutants analyzed here, incubation for 240 min at 37°C eventually resulted in progression to the STC and a corresponding ∼2- to 4-fold increase in the synthesis of the FS product (Fig. 5C). This observation raises questions regarding the reason behind the delay in FS product synthesis. Does the defect conferred by this set of mutations reside in 3′-OH processing, strand transfer, or both? To address this question, we carried out a time course experiment using a preprocessed substrate. For each mutant, we observed a marked augmentation in the rapidity of FS product formation, with essential saturation in FS product formation after an additional 120-min incubation at 37°C (Fig. 6). These results suggest indirectly that the defect(s) conferred by the mutations resides both in assembly and likely in processing using blunt-ended substrates, bringing about a delay rather than an incapacitation. We attempted to directly assess the 3′-OH-processing activities of wt, N155H, N155S, and S153Y INs by filling in a recessed-ended substrate with ³²P-labeled nucleotides, but we found the results difficult to interpret due to minor nuclease contamination in the mutant IN preparations.

The IC₅₀s of L-870,810, L-870,812, and L-870,411 for concerted integration of wt III-B IN at 20 nM (Table 1) are all at least twice as high as that obtained for wt NY IN (18). The reason for this difference in IC₅₀s is unknown, but the IC₅₀ may be influenced both by the differences in six residues between these two HIV-1 IN proteins, resulting in a less-active III-B IN, and by optimal assay conditions for the two proteins. We had also demonstrated previously that the determination of the IC₅₀ of L-870,810 for NY IN is influenced by the concentration of IN in the assay mixture (18).

So, what does our analysis mean within the context of drug resistance? It is generally accepted that a drug resistance mutation confers resistance by altering the conformation of the enzyme in question so as to reduce its affinity for the inhibitor while retaining its intrinsic catalytic activities. It seems possible that, in the case of HIV-1 IN, a resistance mutation might modify catalytic activity so as to circumvent the action of the inhibitor in addition to reducing the affinity for the inhibitor. We showed in a previous report that effective strand transfer inhibition requires the presence of blunt ends within a wt SC and that strand transfer inhibitors recognize a transient structural intermediate that arises before or during 3′-OH processing (18). Assuming that inhibitor binding occurs during processing, which implies that the inhibitor should be in association with complexes that have undergone 3′-OH processing and possess recessed DNA ends, it seems reasonable that a delay in processing could give rise to a lack of inhibition. When an inhibitor-treated cell is infected with a virus bearing a resistant IN mutation, the PIC would be more likely than that of the wt virus to retain blunt ends during its traversing of the cytoplasm. In this state, assuming, for the sake of argument, that the binding of the inhibitor requires the processing step, the PIC would remain nonsusceptible to the presence of the inhibitor during its cytoplasmic voyage. When processing ultimately occurred, after nuclear transport of the PIC, the resulting complex would be in the vicinity of its genomic target and could perhaps carry out concerted integration untrammeled.

Resistance profiles associated with the strand transfer inhibitor-resistant mutants.

As previously discussed, an earlier report described discordance in resistance between two sets of inhibitor-resistant IN mutants (9). The resistance profiles were evaluated through the monitoring of inhibition of either wt or mutant virus replication ex vivo in the presence of a given inhibitor (9, 11). In order to confirm that the resistance profiles observed ex vivo are in fact due to the phenotypic alterations imparted by the mutation, we used our in vitro reconstitution system focusing on the inhibition of concerted integration.

Except for the virus bearing the N155S IN mutation, mutant viruses arising under L-841,411 selective pressure remained susceptible to L-870,810 inhibition, and vice versa (9). Here we show that the N155H mutation, which arises under selective pressure from the NCA L-870,812 in vivo in rhesus macaques (11), unlike the N155S mutation, does not confer cross-resistance to DKAs but weakly potentiates susceptibility to L-841,411 (Fig. 7). Similarly, S153Y, an L-841,411-selected resistance mutation, potentiates susceptibility to the NCA inhibitors L-870,810 and L-870,812. This is significant because it provides further proof-of-concept that mutants exhibiting resistance to a strand transfer inhibitor of a given class may retain or even potentiate suceptibility to a strand transfer inhibitor of another class, highlighting the possibility of employing combinational drug therapy in combating HIV-1 infections resistant to a given class of inhibitor (9).

The FDA recently approved raltegravir (Isentress), a hydroxypyrimidone carboxamide strand transfer inhibitor, for use on treatment-experienced patients who harbor strains of HIV-1 resistant to other antiretroviral agents (8). Significant resistance to raltegravir can arise through two distinct genetic pathways: one involving an initial N155H mutation in the IN gene, the other beginning with a Q148H or Q148H G140S IN mutations (D. J. Hazuda, M. D. Miller, B. Y. Nguyen, and J. Zhao, presented at the 16th International HIV Drug Resistance Workshop, Barbados, West Indies, 12 to 16 June 2007). Following the appearance of one of these initial resistant IN mutants, the subsequent accumulation of additional mutations, a unique set corresponding to each respective pathway, culminates in significant drug resistance associated with virologic failure (Hazuda et al., 16th International HIV Drug Resistance Workshop). The N155H mutation is a natural HIV-1 polymorphism found in circulation in treatment-naïve patients (13). This implicates the appearance of N155H as a pervasive obstacle that will likely prevent the effective long-term treatment of HIV-1 infections with raltegravir alone. Our results show, however, that for at least one member of the DKA class of strand transfer inhibitors, L-841,411, the N155H mutation actually potentiates susceptibility to the inhibitor, indicating that combinational therapy utilizing several classes of strand transfer inhibitors may be an effective means of combating the development of drug resistance associated with the administration of a single inhibitor (9).