Pharmacovirological Impact of an Integrase Inhibitor on Human Immunodeficiency Virus Type 1 cDNA Species In Vivo

Abstract

Clinical trials of the first approved integrase inhibitor (INI), raltegravir, have demonstrated a drop in the human immunodeficiency virus type 1 (HIV-1) RNA loads of infected patients that was unexpectedly more rapid than that with a potent reverse transcriptase inhibitor, and apparently dose independent. These clinical outcomes are not understood. In tissue culture, although their inhibition of integration is well documented, the effects of INIs on levels of unintegrated HIV-1 cDNAs have been variable. Furthermore, there has been no report to date on an INI's effect on these episomal species in vivo. Here, we show that prophylactic treatment of transgenic rats with the strand transfer INI GSK501015 reduced levels of viral integrants in the spleen by up to 99.7%. Episomal two-long-terminal-repeat (LTR) circles accumulated up to sevenfold in this secondary lymphoid organ, and this inversely correlated with the impact on the proviral burden. Contrasting raltegravir's dose-ranging study with HIV patients, titration of GSK501015 in HIV-infected animals demonstrated dependence of the INI's antiviral effect on its serum concentration. Furthermore, the in vivo 50% effective concentration calculated from these data best matched GSK501015's in vitro potency when serum protein binding was accounted for. Collectively, this study demonstrates a titratable, antipodal impact of an INI on integrated and episomal HIV-1 cDNAs in vivo. Based on these findings and known biological characteristics of viral episomes, we discuss how integrase inhibition may result in additional indirect antiviral effects that contribute to more rapid HIV-1 decay in HIV/AIDS patients.

One of the most exciting recent advances in HIV pharmacotherapy has been the approval of the first human immunodeficiency virus type 1 (HIV-1) integrase inhibitor (INI), raltegravir (MK-0518) (20). INIs are a new class of antiretroviral drugs that inhibit the essential step in the replication cycle of HIV-1 of integrating proviral HIV-1 DNA into the host cell genome. As a direct consequence, a newly infected cell cannot produce viral progeny and HIV-1 cannot persist as a provirus. Integration is a multistep process, the first two of which are catalyzed by HIV-1-encoded integrase (IN). First, IN hydrolyzes a dinucleotide from the 3′ ends of the nascent HIV-1 DNA within the preintegration complex. This is followed by the covalent insertion of the HIV-1 DNA into the cellular DNA in a step referred to as strand transfer. All two-metal-binding INIs, for example, raltegravir and naphthyridinone analogues, such as GSK364735 (14), selectively inhibit this second step (21).

Raltegravir's clinical efficacy was established in multicenter dose-ranging phase II and III studies of HIV-1-infected patients. First, monotherapy resulted in an extensive monophasic decay that was comparable in all dosage groups, with a median decrease of 2.2 log₁₀ HIV RNA copies/ml over 10 days (20, 40, 42). Monotherapy with other INIs (L-870,810, elvitegravir, and GSK364735) has shown similar overall viral-load declines (11, 12). Second, in a combination study of antiretroviral therapy-naïve patients receiving tenofovir and lamivudine as background therapy, individuals also taking raltegravir had a more rapid decline in HIV-1 viremia than those taking the potent nonnucleoside reverse transcriptase inhibitor (RTI) efavirenz (40, 42), although both arms reached the same ultimate decrease in the viral load. Specifically, by week 4 on therapy, 60 to 80% of patients on raltegravir had suppressed the HIV-1 RNA load to less than 50 copies/ml versus only 25% of those treated with efavirenz. This more rapid suppression was unexpected and has triggered a controversy as to the potential underlying mechanism(s) (27, 39, 42, 48).

The in vivo impact of INI treatment on the fate of HIV-1 cDNA species has not been reported previously. In principle, linear HIV-1 cDNA that does not become integrated into the host genome can undergo various processes. It can be a substrate for the host cell nonhomologous-end-joining pathway in the nucleus, which mediates its end-to-end circularization to form two-long-terminal-repeat (LTR) circles. One-LTR circles, which arise through homologous recombination between the two LTRs, or autointegrants, can also form as additional episomal HIV-1 cDNA species (44). In in vitro models of HIV infection, interference with integration has been achieved by genetic (16, 26, 47) or pharmacological (7, 16, 52, 54) inhibition of IN function and through the depletion of cellular cofactors involved in the peri-integrational process (50, 61). In these studies, inhibition of integration resulted in a variable effect on viral episome levels ranging from a 4-fold reduction (54), essentially unchanged levels (7), to a 10-fold enhancement (52) (Table 1) . Similarly, cells infected with lentiviruses encoding a genetically deficient IN displayed considerable variability in the steady-state levels of two-LTR circles (26, 47) (Table 1). In light of this in vitro heterogeneity and the plethora of parameters that can influence the formation and fate of episomal HIV-1 cDNAs (8, 38, 44, 49), the impact of an INI on their in vivo fate is unpredictable. It is important to recognize that episomal viral cDNAs are biologically active: they are capable of expressing HIV-1 proteins (9, 16, 45, 51, 59, 60) with potential relevance for HIV-1 immune recognition. Furthermore, linear, unintegrated episomes can be proapoptotic (38, 53).

TABLE 1.

Literature review and measured changes of integrant and two-LTR circle/episome levels in infected cells following interference with lentiviral integration

Mode of IN interferencea	Cell type	Virus	Integrant fold reduction	Two-LTR circle/episome fold change	Reference
L-731,988	293T	HIV-1NL4-3	>35	Unchangedc	7
L-708,906	293T	HIV-1NL4-3	>35	Unchangedc	7
L-731,988	T cells	HIV-1HXB2	>300	Unchangedc	54
L-731,988	T cells	HIV-1HXB2	>300	4 down	54
L-731,988	CD4+ T cells	HIV-1NL4-3	NAb	4.3 up	16
L-708,906	293T	HIV-1NL4-3	NAb	10 up	52
Raltegravir	HeLa	HIV-1YU-2	NAd	3 up	This study
GSK501015	HeLa	HIV-1YU-2	NAd	Unchangedc	This study
Raltegravir	Rat2	HIV-1YU-2	NAd	4 up	This study
GSK501015	Rat2	HIV-1YU-2	NAd	5 up	This study
IN D66V/D118A	CrFK	FIV	NAb	Unchangedc-2.6 up	47
IN D116N	CD4+ T cells	HIV-1NL4-3	NAb	3.5 up	16
IN D116N	SupT1	HIV-1NL4-3	NAb	10 up	26

^aIntegration interference was achieved by INIs (top 10 rows) or catalytic site mutants of integrase (IN) (bottom 3 rows).

^bNot applicable; inhibition of integration was confirmed indirectly by viral-gene expression reporter assays or HIV p24CA immunofluorescence.

^c“Unchanged” was defined as an alteration of ≤2-fold.

^dNot applicable; inhibition of integration was indirectly validated by p24CA antigen enzyme-linked immunosorbent assay. Levels of p24CA in culture supernatants of INI-treated cells were ≥80% reduced compared to dimethyl sulfoxide-treated controls (n = 2 to 3).

Other aspects of the efficacy of INIs are also not completely understood. Specifically, the in vivo relationship of pharmacokinetics and the virological impact of raltegravir is still poorly defined, as dose-ranging studies of raltegravir showed equal efficacies of viral-load reduction across a range of doses (20, 40, 42). Animal models, such as HIV-susceptible transgenic Sprague-Dawley rats expressing the HIV-1 receptor complex (17, 19, 33, 41), can support the preclinical validation process of antiviral compound candidates, allowing the quantification of their virological impacts and serum concentrations. In these in vivo models, an antiviral compound can be titrated from no effect to full effect, and mechanistic hypotheses can be probed under well-controlled experimental conditions.

To advance the understanding of the molecular consequences of INI treatment in vivo, we studied the impact of GSK501015, a naphthyridinone strand transfer INI, on HIV-1 infection in hCD4/hCCR5 transgenic rats. We focused on characterizing GSK501015's effect on integrated and episomal HIV-1 cDNA species and on establishing the pharmacovirological relationship of the INI and HIV-1 in vivo.

MATERIALS AND METHODS

HIV-susceptible transgenic-rat model.

The generation and initial characterization of hCD4/hCCR5 transgenic rats has been reported previously (17, 33, 41). Animal experiments were conducted according to the German animal welfare act and with authorization of the Regierungspräsidium Karlsruhe (35-9185.81/G-100/02). The experiments were supervised by animal welfare officers of Heidelberg University.

Cell culture and virus stocks.

For experiments conducted at the University of Heidelberg, human 293T cells, TZM-bl cells, and MT-4 T cells, as well as rat-derived Rat2 fibroblasts, have been reported (32, 34). Cultures of primary T lymphocytes and macrophages from transgenic rats and random, HIV-negative human donors were generated and propagated as reported previously (18, 32, 34). The generation and characterization of replication-competent HIV-1_R7/3 YU-2 Env green fluorescent protein (GFP) (57) and HIV-1_YU-2 stocks (33) for in vivo infection studies has been reported previously. For experiments conducted at GlaxoSmithKline, MT-4 cells were maintained as described previously (10). 293T cells were maintained in Dulbecco's modified Eagle's medium/F12 medium containing 10% fetal bovine serum. Peripheral blood mononuclear cells (PBMCs) were isolated from random buffy coats by density gradient centrifugation over lymphocyte separation medium and stimulated by the addition of 5 μg/ml phytohemagglutinin for 24 to 48 h. HIV-1_IIIb was derived from cell-free supernatants of cultures of the chronically infected cell line H93B (H9/HTLV-IIIb). HIV-1_Ba-L was purchased from Advanced Biotechnologies (Columbia, MD) and expanded in phytohemagglutinin-activated PBMCs.

Antiviral drugs.

GSK501015 was synthesized at GlaxoSmithKline, Research Triangle Park, NC. Efavirenz (Sustiva drinking solution; Bristol-Myers Squibb) was purchased as a 30-mg/ml solution and diluted in phosphate-buffered saline. Raltegravir (Isentress tablets; Merck Sharp & Dome) was purchased as 400-mg tablets, dispersed and diluted in H₂O.

In vitro recombinant IN and cellular antiviral assays.

In vitro recombinant HIV IN strand transfer activity (5) and INI binding (14) were measured as previously described. Antiviral HIV activity was measured in human MT-4 T cells as reported previously (14, 25). HIV-1_Ba-L replication in PBMCs was quantitated by measuring reverse transcriptase activity present in the supernatant (25).

Ex vivo efficacy testing of GSK501015.

Primary cells from humans and hCD4/hCCR5 transgenic rats were pretreated with GSK501015 at the indicated concentrations or with 1% dimethyl sulfoxide as a solvent control for 1 h and subsequently challenged with HIV-1_R7/3 YU-2 Env GFP (T cells) or vesicular stomatitis virus (VSV) G HIV-1_NL4-3 GFP (macrophages) reporter virus. On day 3 postinfection, the percentages of GFP-positive T cells and macrophages were scored by flow cytometry on a FACSCalibur using BD CellQuest Pro 4.0.2 software (BD Pharmingen).

In vivo efficacy testing—study design.

Two independent in vivo studies were performed. For the high-dose study, hCD4/hCCR5 transgenic rats (weight range, 236 to 543 g) were treated with GSK501015 at 10 mg/kg of body weight/day by twice-daily oral gavage at 12-h intervals (1% methylcellulose formulation; 5 ml/kg) for 5 days (GSK501015 group, n = 6) or left untreated (control group, n = 6). For the dose titration study, hCD4/hCCR5 transgenic rats (weight range, 262 to 617 g) were treated either with formulated GSK501015 at doses ranging from 0.1 to 10 mg/kg/day or with the RTI efavirenz at 25 mg/kg by once-daily oral gavage. One nontransgenic infected animal, one uninfected animal, and four untreated infected animals served as additional controls (see Fig. 3 for the number of animals assigned to each treatment group and the design of the two in vivo studies). On day 1 of each study, treated rats and the indicated single-transgenic or nontransgenic control rats were anesthetized and challenged by tail vein injection with HIV-1_YU-2 (1.25 × 10⁷ TZM-bl infectious units per rat), in principle as reported previously (17, 33). On days 1 and 3 postinfection, blood was taken at the indicated time points from INI-treated rats in the dose titration study for subsequent pharmacological evaluation. On day 5 postinfection, all animals were sacrificed and their spleens were removed. DNA was prepared from single-cell suspensions of splenocytes by using either a DNeasy tissue kit or the Genomic tip kit (Qiagen) and was analyzed by real-time PCR.

FIG. 3.

Experimental design of the two in vivo studies with GSK501015 in hCD4/hCCR5-transgenic rats. (A) High-dose study. Transgenic rats were treated with GSK501015 at 10 mg/kg by twice-daily oral gavage for 5 days (GSK501015 group; n = 6) or left untreated (control group; n = 6). On day 1, the animals, including one hCD4 single-transgenic rat (ID 940), were challenged with HIV-1_YU-2 by tail vein injection. On day 4 postinfection, the animals were sacrificed and their spleens were removed for the quantitative analysis of HIV-1 cDNA species. (B) Dose titration study. hCD4/hCCR5 transgenic rats were treated with GSK501015 at the indicated doses for 5 days or left untreated. On day 1, the animals, including one nontransgenic rat and one hCCR5 transgenic rat, were challenged with HIV-1_YU-2 by tail vein injection. On day 4 postinfection, the animals were sacrificed and their spleens were removed for the quantitative analysis of HIV-1 cDNA species. Blood samples for concurrent pharmacokinetic analysis were taken from all INI-treated animals on the indicated days at −2 h, +2 h, and +6 h relative to the morning dosing.

GSK501015 pharmacokinetics.

From all INI-treated, HIV-1-infected rats from the dose titration study and two arbitrarily selected INI-treated animals from the high-dose study (rat identifiers [ID] 892 and 903), blood samples were taken from tail veins at the indicated time points on days 1 and 3 postinfection. Serum was harvested by 37°C/4°C temperature shift and centrifugation (3,400 × g; 4°C; 30 min), inactivated at 56°C for 30 min, and stored at −80°C until analysis. Thawed samples were extracted by protein precipitation with ice-cold acetonitrile and analyzed by liquid chromatography-tandem mass spectrometry. The maximum concentration observed in each serum concentration-time profile (C_max) was assessed by observation of the data.

Quantification of HIV-1 DNA species.

Levels of HIV-1 two-LTR circles were measured as reported in detail previously (17). For two-LTR circle standard curves, dilutions of a plasmid encoding the U3-U5 junction region (pU3U5) covering 5 log units were used, supplemented with DNA extracted from uninfected rat cells. To detect and quantify integrated HIV-1 provirus in rat cells, we used a nested real-time PCR assay employing an ID consensus sequence within the rat BC1 RNA gene (35, 36) as the rodent repeat target for the cellular anchor primer pair. A Rat2 cell line containing on average one integrated HIV-1 provirus per cell, termed Rat2^int, served as a standard for species-specific quantitative analyses of provirus formation. Since Rat2^int cells no longer contain significant amounts of unintegrated HIV-1 cDNA species (due to long-term passaging following infection), the absolute number of integrated proviruses per nanogram cellular DNA in Rat2^int could be accurately determined by quantifying the absolute amount of HIV-1 cDNA by real-time PCR (17, 19). DNA was extracted from cells by using Genomic tips (Qiagen), which preferentially extract the chromosomal DNA with high yield. In a first PCR, HIV-1 integrants were amplified by one primer annealing to gag (5′-TTTCAAGTCCCTGTTCGGGCGCCA-3′) and by two outward-facing primers that targeted the highly redundant ID consensus sequence (primers 1734, 5′-GGTAACTGGCACACACAACC-3′, and 1782, 5′-CTGAGCTAAATCCCCAACCC-3′). The conditions for the first PCR were (i) 5 min at 94°C; (ii) 18 cycles of 30 s at 94°C, 30 s at 57°C, and 4 min at 72°C; and (iii) 10 min at 72°C. During the second real-time PCR, a 1:250 dilution of a first-round aliquot was subjected to amplification with primers with the sequences 5′-ATGCCACGTAAGCGAAACTCTGGCTAACTA GGGAACCCACTG-3′ and 5′-TGACTAAAAGGGTCTGAGGGATCT-3′ and the probe 6-carboxyfluorescein-5′-TTAAGCCTCAATAAAGCTTGCCTTGAGTGC-3′- 6-carboxytetramethylrhodamine. The second-round cycling conditions were identical to those used to determine the total numbers of two-LTR circles: (i) 2 min at 50°C, (ii) 10 min at 95°C, and (iii) 18 cycles of 15 s at 95°C and 1 min at 60°C. For each integration PCR run, a control reaction in which the cellular primer pair was omitted was run in parallel. This value was routinely subtracted from the total signal. For standard curves, dilutions of genomic DNA from Rat2^int covering 3.9 log units were used. The lowest detection standard of the PCR was 0.07 HIV-1 integrants per ng rat genomic DNA. Two-LTR circles and integrants were normalized to the amount of cellular DNA, which was quantified in a parallel amplification of the rat GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene (Applied Biosystems). For the latter, dilutions of genomic DNA extracted from uninfected primary rat T cells were used for standard curves. All samples were run in duplicate or triplicate, and each sample was measured two or three times. All quantitative PCR analyses were performed using the ABI 7500 sequence detection system (Applied Biosystems, Foster City, CA), and data analysis was performed with the 7500 System Software (Applied Biosystems).

Statistical analyses.

For in vitro assays measuring activities against recombinant HIV-1 IN and against HIV-1 in cell-based antiviral assays, the concentration-response relationship was modeled by fitting the Hill equation: f(x) = E_min + {(E_max − E_min)/[1 + (x/K)^m]}, where x is the concentration of compound, E_min is the minimum expected response, E_max is the maximum expected response, m is the Hill slope, and K is the value of x that gives a response halfway between E_min and E_max. The concentration giving 50% inhibition (IC₅₀) or efficacy (EC₅₀) was back-calculated from the fitted curve. In vitro IC₅₀s and EC₅₀s were summarized as geometric means. The 95% confidence intervals (CI) were calculated on the log scale using a separate estimate of variance for each assay and compound and exponentiated to give a CI on the nM scale. The EC₅₀s shown in Fig. 1B to E and in Table 2 were calculated using GraphPad software. Integrants and two-LTR circles were analyzed on the log scale. Means are reported on the geometric scale. Group differences are expressed as change with a 95% CI. Means were calculated for serum concentrations of GSK501015 in rats (see Fig. 5). The relationship between log(integrants) and C_max at day 1 was modeled using the Hill equation described above, where x was the C_max at day 1 for each animal. Data from the high-dose and dose titration experiments were combined in this analysis, and experimental differences were accounted for by the addition of a block effect in the statistical model. To estimate in vivo EC₅₀s for integrants, the value of C_max giving a 50% reduction in integrants was back-calculated from the fitted equation, together with a 95% CI. The relationship between log(two-LTR circles) and the administered dose was modeled using an analysis of variance means model and a linear regression on log(dose). Data from both experiments were combined, and experimental differences were accounted for by the addition of a block effect in the statistical model. Statistical comparisons among the 10-mg/kg group, the untreated group, and the group treated with efavirenz were made using t tests calculated using the model-based group mean estimates and standard error. Pearson's correlation was used to assess the strength of the association between log(two-LTR circles) and log(integrants). To adjust for experimental differences, the correlation was calculated after subtracting the mean difference between experiments from the log of each response for animals in the high-dose study. To adjust for the effects on the efficacy of rat or human protein binding, the relationship between in vitro EC₅₀ and percent serum was modeled as follows: log(EC₅₀) = log(α + β × percent serum) = error, where the EC₅₀ is the observed concentration giving 50% efficacy in the presence of a given percentage of rat or human serum, α is the linear model intercept, β is the linear model slope, and “error” is an additive random error with constant variance. The EC₅₀ at 0% and 100% human serum were predicted using the fitted model, along with a 95% CI calculated on the log scale. The expected shift in the EC₅₀ was calculated as the ratio of the 0% and 100% predicted EC₅₀s. For rat serum, the model was fitted to data from a single experiment. For human serum, the model was fitted separately to each of four experiments, and the geometric means of EC₅₀s at 100%, and the shift in the EC₅₀, were reported, along with a 95% CI calculated on the log scale.

FIG. 1.

Chemical structure and antiviral activity of the INI GSK501015. (A) Structure of the two-metal-binding naphthyridinone derivative GSK501015. (B to E) Dose-response curves for the antiviral potency of the INI GSK501015 in ex vivo cultures of primary T cells and macrophages from hCD4/hCCR5 transgenic rats and humans. (B and C) Activated T cells from both species. (D) Spleen-derived macrophages from double-transgenic rats. (E) Monocyte-derived macrophages from human donors were pretreated with the INI and subsequently challenged with HIV-1_R7/3 YU-2 Env GFP (T cells) or VSV-G HIV-1_NL4-3 GFP (macrophages). On day 3 postinfection, the percentage of GFP-positive cells was scored by flow cytometry, and the values obtained for untreated controls were set to 100%. Given are arithmetic means ± standard deviations (n = 3) of one experiment. EC₅₀s were determined by using Prism software (GraphPad, San Diego, CA) and are shown in panels B to E and in Table 3.

TABLE 2.

Potencies of strand transfer INI GSK501015 against recombinant HIV-1 IN in vitro and against HIV-1 in antiviral assays on human cells

Integrase inhibitor	Integrase IC50 (nM)a		Antiviral EC50 (nM)a
	Strand transfer	3H-INI binding	PBMC	MT-4
GSK501015	6 (4.6, 7.9)	19.3 (9.3, 40)	2.7 (1.7, 4.3)	12.4 (9.8, 16)
Raltegravir	7.4 (4.1, 13)	6.5 (3.2, 14)	1.9 (1.3, 2.7)	4.7 (2.5, 9.0)

^aIn vitro recombinant HIV IN strand transfer activity, ³H-INI binding, and antiviral HIV activity in cell-based assays were measured as reported previously (5, 14, 25). The values are geometric means, with 95% CI in parentheses, calculated from at least seven determinations.

FIG. 5.

Concentrations of GSK501015 in sera from HIV-infected hCD4/hCCR5 transgenic rats. Blood samples were taken at the indicated time points relative to the morning dose administration, and GSK501015 concentrations were determined by liquid chromatography-tandem mass spectrometry. The dotted line indicates the in vitro PA-EC₅₀ as an informative reference. p.i., postinfection.

RESULTS

GSK501015 is a potent two-metal-binding strand transfer INI.

While the naphthyridone series of INIs was being developed at GlaxoSmithKline (14, 46), GSK501015 (Fig. 1A) was identified as a low-nanomolar inhibitor of HIV-1 IN. In a strand transfer assay with recombinant HIV-1 IN, GSK501015 displayed an IC₅₀ of 6 nM, which was a potency similar to that of raltegravir (Table 2). GSK501015 inhibited HIV-1 replication in both primary human PBMCs (infected with HIV-1_Ba-L) and MT-4 T cells (infected with HIV-1_IIIb) with EC₅₀s of 2.7 and 12.4 nM, respectively (Table 2). The ratio of 3.2 for the IC₅₀ of GSK501015 in a ³H-radiolabeled ligand binding assay divided by its IC₅₀ in the strand transfer reaction was similar to ratios ranging from 1.3 to 4.8 for GSK364735 and the two-metal-binding INIs reported previously (14). This similarity demonstrated that GSK501015 bound in the same structural pocket of HIV-1 IN as other two-metal-binding inhibitors, and with low-nanomolar potency similar to those of INIs that have been studied in clinical trials.

Cross-species characterization of the in vitro and ex vivo anti-HIV efficacy of GSK501015.

We established several in vitro and ex vivo parameters with relevance for the in vivo evaluation of GSK501015 and to verify the antiviral potency in the laboratory where the in vivo study was to be performed. First, cultured primary T cells and macrophages derived from either hCD4/hCCR5 transgenic rats or healthy human donors were pretreated side by side with GSK501015, challenged with HIV-1 GFP reporter viruses, and analyzed for early viral-gene expression on day 3 postinfection. In these primary target cultures from both species, the relative percentage of HIV-1-infected cells decreased in a concentration-dependent manner, and EC₅₀s were in the nanomolar range and similar for rats and humans (Fig. 1B to E and Table 3).

TABLE 3.

Comparable anti-HIV-1 efficacies of INI GSK501015 in cultured primary cells from hCD4/hCCR5 transgenic rats and humans

Species	Antiviral EC50 (nM)a
	Macrophages	T cells
Rat	7 ± 3 (n = 2)	1 ± 0.05 (n = 3)
Human	2 ± 0.5 (n = 2)	2 ± 1 (n = 4)

^aEC₅₀ values were derived from the dose-response experiments shown in Fig. 1B to E and from experiments conducted in an analogous manner. Shown are the arithmetic means ± standard errors of the mean from primary cell cultures established from two to four donors; each experiment was performed in triplicate.

Second, the impact of potent IN inhibition by either GSK501015 or raltegravir on two-LTR circles in infected human and rat cell lines was examined. As seen in analogous in vitro studies (7, 16, 52, 54), levels of episomal HIV cDNAs were variable, ranging from unchanged to fivefold enhancement relative to solvent-treated control cells (Table 1).

Third, as a critical technical basis for evaluating the impact of GSK501015 on HIV integrants in vivo, a highly sensitive real-time PCR-based assay for the quantitative detection of HIV-1 integrants in the rat genome was developed in a multistep optimization of a previously reported protocol (19) and based on previous studies in human cells (2, 6, 37). This optimized HIV integration PCR protocol (Fig. 2A) (for details, see Materials and Methods) was validated according to previously established standards (6, 19) and achieved an increased intensity of the specific PCR signal for HIV-1 integrants in the rat genome (12.6-fold) (Fig. 2B, left). Furthermore, the optimized protocol resulted in a reduction of the unspecific PCR background (16-fold) (Fig. 2B, middle), which is an inherent feature of nested-PCR-based measurements of HIV-1 integration (6). In combination, this allowed a 40-fold increase in PCR assay sensitivity (Fig. 2B, right). Applying this integrant-specific PCR, GSK501015 was shown to inhibit HIV-1 integrant formation in Rat2 fibroblasts with high potency (EC₅₀ = 0.25 nM) (Fig. 2C).

FIG. 2.

Establishment of a highly sensitive and specific rat integration PCR and of basic pharmacovirological parameters of GSK501015 in rats. (A) Strategy of the “optimized protocol” for the nested rat integration PCR (see Materials and Methods for details). ID, rodent identifier consensus sequence. (B) The sensitivities of detection of HIV-1 integrations in the rat genome in a recently reported nested PCR protocol (standard protocol [19]) and the optimized protocol developed here were compared. Shown are the relative intensities of the specific signals (left), the percentage of unspecific signal (middle), and the lower limit of the standard (right), applying the two PCR protocols to the Rat2^int genomic standard. (C) GSK501015 potently suppresses HIV-1 integrant formation in a rat cell line. Rat2 cells were pretreated with different concentrations of GSK501015 in the presence (1 μM) (negative control) or absence of the RTI efavirenz and challenged with a VSV-G-pseudotyped HIV-1 GFP vector. On day 7 postinfection, DNA was extracted from passaged cells, and the relative levels of HIV-1 integrants were determined as described in Materials and Methods. The number of integrants in the absence of treatment was set to 100%, and the relative percentages in the presence of decreasing concentrations of GSK501015 are depicted. (D) Relationship of oral dose and serum concentration of suspension-formulated GSK501015 in outbred, nontransgenic Sprague-Dawley rats. Data points indicate arithmetic means ± standard deviations (n = 3 animals).

Fourth, to assess the effects of serum proteins on the potencies of anti-HIV drugs in vitro, rat or human serum was titrated in the HIV-1 replication assay on MT-4 T cells in the presence of GSK501015, and the corresponding EC₅₀s for the INI were determined (data not shown). The extrapolated increase in EC₅₀s at 100% rat or human serum was calculated to be 11-fold or 16-fold, respectively. To determine a potency target to estimate the serum concentrations of GSK501015 that would achieve efficacy in vivo, we multiplied the 11-fold shift determined with rat serum by the EC₅₀ in PBMCs (Table 2) to derive a protein-adjusted (PA) EC₅₀ of 31 nM (14 ng/ml).

Finally, a near-linear relationship between the oral dose of GSK501015 and the resulting C_max was established in nontransgenic Sprague-Dawley rats (Fig. 2D). Collectively, these data indicated that GSK501015 was an excellent drug candidate to study the pharmacovirological consequences of an INI in the HIV-susceptible transgenic-rat model.

A high-dose GSK501015 regimen drastically diminishes levels of HIV-1 integrants and enhances levels of episomal two-LTR circles in transgenic rats in vivo.

In an initial study (Fig. 3A), we investigated the effects of a prophylactic high-dose treatment regimen of GSK501015 in HIV-infected transgenic rats on two major HIV-1 cDNA species, integrants in the host cell genome and nonintegrated, episomal two-LTR circles. The latter circularized DNA species is an established quantitative surrogate for all episomal HIV-1 DNA species (3, 22). In the treatment group, GSK501015 was administered to six hCD4/hCCR5 transgenic rats at 10 mg/kg/day by twice-daily oral gavage starting 1 day prior to viral challenge and continuing until the termination of the experiment. The control group consisted of six hCD4/hCCR5 transgenic rats that received no treatment. All rats were challenged with HIV-1_YU-2 via tail vein injection and sacrificed 4 days later. The levels of HIV-1 cDNA species in DNA extracts from splenocytes were determined by real-time PCR.

Oral treatment of hCD4/hCCR5 transgenic rats with GSK501015 resulted in a 66-fold reduction (i.e., reduction by 98.5%) of HIV-1 integrant levels compared to the untreated control group (Fig. 4A, right) (P = 0.026 [Mann-Whitney U test]). Of note, the degree of reduction was quite variable among individual animals from the GSK501015 group. For example, in DNA samples from one INI-treated rat (animal ID 892) (Fig. 4A, left), no integrants could be amplified, while the integrant levels of two other animals (animal ID 846 and 905) (Fig. 4A, left) were in the lower range of the untreated control group.

FIG. 4.

High-dose treatment with GSK501015 inhibits HIV-1 integrant formation and elevates levels of episomal HIV-1 cDNA in vivo. The impact of INI treatment (10 mg/kg per day; the experimental design is shown in Fig. 3A) on HIV-1 DNA species abundance was assessed by determining, by real-time PCR, the load of HIV-1 integrants (A) or episomal two-LTR circles (B) relative to a rat GAPDH standard in DNA extracts from splenocytes. The results given for individual animals (left) are arithmetic means plus the standard error of the mean (SEM). The results given for animal groups (right) are geometric means + SEM. Nonparametric statistical analyses were performed by using the Mann-Whitney U test: integrants, 66-fold reduction (i.e., 98.5% reduction; P = 0.026); two-LTR circles, 7-fold increase (P = 0.03), n.a., not analyzed; ↓, no signal detected. *, statistically significant.

Next, we explored whether unintegrated HIV-1 cDNA species accumulate as a consequence of the failed chromosomal integration of viral DNA in GSK501015-treated rats. Employing a two-LTR circle-specific PCR (17), a sevenfold enhancement in the levels of this episomal HIV-1 cDNA species was observed in splenocyte samples from the treatment group relative to those of untreated controls (Fig. 4B, right) (P = 0.030). As a control of specificity, neither HIV-1 cDNA species could be amplified from samples of a hCD4 single-transgenic rat (animal ID 940) challenged with the identical infectious dose (Fig. 4A and B, left), demonstrating that the amplified HIV cDNA species had been generated de novo following a receptor complex-mediated infection in vivo. Of note, at this time point postchallenge (day 4), the levels of total HIV-1 cDNA exceeded those of the two other DNA species by 2 to 4 orders of magnitude, as reported previously (19), and no significant difference was found between the two groups (data not shown). Thus, in vivo administration of the strand transfer INI reduced the proviral burden in a secondary lymphoid organ and increased the nuclear levels of an abortive, circular HIV-1 cDNA species.

GSK501015 pharmacovirological dose-response study with HIV-infected transgenic rats.

In an extension of the high-dose study, we explored the relationship between the GSK501015 dose, the resulting serum concentrations, and the virological impact in individual HIV-1-susceptible transgenic rats. Information derived from such a dose titration study should for the first time allow a direct comparison of inhibitory concentrations of an INI effective at blocking HIV-1 integration in vitro and in vivo. Furthermore, a refined correlative analysis of the in vivo relationship between HIV-1 integrants and episomal two-LTR circles was anticipated.

In an experimental setup analogous to the high-dose INI study, GSK501015 was administered to hCD4/hCCR5 transgenic rats at oral doses ranging from 0.1 to 10 mg/kg/day (the study design is shown in Fig. 3B). The identical HIV-1_YU-2 stock and infectious titer applied in the high-dose study were used. For a concurrent pharmacological analysis in GSK501015-treated, HIV-1 infected rats, blood samples were taken 2 h prior to the morning dosing (−2 h) and 2 and 6 h after the morning dosing (+2 h and +6 h) on days 1 and 3 postinfection.

Drug concentrations in serum displayed an expected profile and segregated into the different dosing groups (Fig. 5) (the target PA EC₅₀ in vitro value [14 ng/ml] is referenced). A combined analysis of data derived from splenocyte and serum samples from animals in the high-dose and dose titration studies established a decreasing relationship between the levels of HIV-1 integrants in the spleen and the GSK501015 C_max on day 1 (Fig. 6A), demonstrating a clear concentration dependence for the antiviral effect of the INI in vivo. Furthermore, this relationship allowed us to derive for the first time an in vivo EC₅₀ estimate for an INI: 53 nM (24 ng/ml) GSK501015. In this combined-data analysis, transgenic rats receiving the maximum dose of GSK501015 (10 mg/kg) had 300-fold fewer HIV-1 integrants on average (i.e., reduction by 99.7%) than untreated control animals (P < 0.0001). Conversely, the levels of two-LTR circles showed an increasing trend with the administered GSK501015 dose (Fig. 6B). On average, animals dosed with GSK501015 at 10 mg/kg had fivefold-higher levels of this episomal HIV-1 cDNA species than untreated control rats (P = 0.0172). Notably, treatment with efavirenz also drastically diminished the levels of HIV-1 integrants (Fig. 6A, right), but due to the RTI's earlier intervention point in the viral life cycle, it did not enhance episomal HIV-1 cDNA species (Fig. 6B, right).

FIG. 6.

Concentration-dependent efficacy of GSK501015 in HIV-infected hCD4/hCCR5 transgenic rats. These analyses contain results obtained from both the high-dose and the dose titration studies. (A) Number of HIV-1 integrants per ng DNA determined in the spleen versus the day 1 maximum concentration of GSK501015 (C_max) in serum. The dotted curve indicates the fit of the Hill curve. (B) Levels of episomal two-LTR circles relative to the administered GSK501015 dose. The gray boxes indicate the geometric means of two-LTR circles in samples from the untreated control groups and the group treated at 10 mg/kg (P = 0.0172). The dotted line indicates the fitted linear relationship between log(two-LTR circles) and log dose.

Finally, we explored the statistical relationship of levels for integrants and episomal two-LTR circles in splenocyte samples from individual HIV-infected double-transgenic rats from the two GSK501015 in vivo interference studies. The biologically likely sequence of events is that the action of the INI directly blocks HIV-1 integration, and this, as a consequence, favors the formation of circular HIV-1 cDNA species in the nucleus. In line with this interdependence scenario, a significant inverse association between the two virological parameters was found (Fig. 7) (Pearson's correlation, r = −0.47; P = 0.0107).

FIG. 7.

In vivo relationship of HIV-1 integrants and two-LTR circles in a secondary lymphoid organ of GSK501015-treated rats. Data obtained from the two in vivo studies for levels of integrants and two-LTR circles in untreated or GSK501015-treated hCD4/hCCR5 transgenic rats are given (and also presented in part in Fig. 4 and 6). Each circle depicts the result for these two virological parameters for one HIV-infected rat (open circles, high-dose study; filled circles, dose titration study). The fitted line illustrates the inverse correlation determined by Pearson's correlation coefficient between log(integrants) and log(two-LTR circles).

DISCUSSION

The profound clinical efficacy of the new class of HIV antiretrovirals, INIs, and the unexpectedly rapid kinetics of HIV-1 RNA load reduction of raltegravir emphasize the need to investigate basic pharmacovirological parameters of INI treatment in vivo. The current experimental study in an HIV-susceptible small-animal model provides proof of concept that levels of HIV-1 integrants can be drastically reduced by the action of a strand transfer INI. The clear dose-response relationship between GSK501015 and inhibition of HIV-1 integrant formation in the spleens of transgenic rats was established. This allowed us to generate an EC₅₀ for this antiretroviral drug in vivo. Furthermore, our study demonstrates that INI treatment in vivo is accompanied by an elevation of levels of unintegrated cDNA species with potential biological activities. This observation is important for mechanistic hypotheses advanced in respect to the more rapid viral-load decline in HIV-1-infected individuals receiving raltegravir.

When we started this investigation, we reasoned that the pharmacological suitability of outbred Sprague-Dawley rats (1) combined with the robust HIV-1 infection of hCD4/hCCR5 transgenic rats (17) provides a unique experimental setting to assess the pharmacovirological impact of an oral INI. We confirmed several in vitro and ex vivo parameters that are relevant for the evaluation of GSK501015 in the rat-human cross-species context. Importantly, we developed a sensitive PCR protocol for the detection of HIV-1 integrants in the rat genome. This modified assay ensured a dynamic range sufficient to quantify reductions in integrant levels in vivo by over 2 orders of magnitude.

Transgenic rats receiving the maximum dose of GSK501015 had 99.7% fewer HIV-1 integrants than untreated controls in splenocytes 4 days postinfection. Although not unexpected, a reduction of the proviral burden in a lymphatic organ as a consequence of INI oral treatment had until now not been demonstrated. Exploiting unique features of the HIV-susceptible rat model, concurrent pharmacokinetic analysis of infected GSK501015-treated animals allowed a correlative analysis of the INI C_max and its major antiviral effect, the inhibition of integration. First, inhibition of integrant formation depended on both the dose and serum concentration of the INI. Such a correlation is the expected outcome for an enzyme inhibitor yet stands in contrast to raltegravir's dose-ranging studies, which showed equal efficacies for viral decay across a range of four doses (20, 40, 42). While we recognize that this could in principle be related to drug-specific properties, we predict that lower doses of raltegravir would also show concentration dependence in HIV-infected patients. Complementary studies with raltegravir in the transgenic-rat model would also help to address this issue. Regardless, it can be minimally concluded that dose independence of the antiviral activity is not an inherent property of strand transfer inhibitors of HIV IN in vivo.

Second, pharmacovirological analyses showed that inhibitory concentrations of GSK501015 effective at blocking integration in vivo (EC₅₀ [C_max] = 53 nM) and in vitro (PA-EC₅₀ = 31 nM) matched best when the drug's serum protein binding was accounted for (in vitro EC₅₀ alone = 2.7 nM). This finding provides experimental confirmation of the long-held view that the effect of serum protein binding can be critical for prediction of in vivo efficacy. The combined pharmacovirological in vivo analysis could conceivably be extended to other cellular, anatomical, and pharmacological reservoirs (23, 55, 58). The gastrointestinal tract, genital tract, and central nervous system comprise major sanctuaries for HIV and may underlie drug- and compartment-specific pharmacokinetic and pharmacodynamic effects (15, 30, 43). This may be of particular relevance for INI activity, since a superior pharmacokinetic profile of raltegravir for targeting HIV sanctuaries has been hypothesized, though in one study it was not favored due to poor prediction of the clinical data upon modeling (42).

Contrary to the widely prevailing presumption, the consequences of INIs for extrachromosomal HIV-1 cDNA species have been variable, and in one study even antipodal (Table 1). In the current INI in vivo study, levels of HIV-1 integrants and episomal two-LTR circles in the spleens of infected rats were inversely correlated (66-fold reduction versus 7-fold enhancement). Notably, the absolute number of integrants lost exceeded those of accumulating two-LTR circles by ∼10-fold. This apparent discrepancy may be due to the presence of additional unintegrated cDNA species in the nucleus that were not quantified here, including linear HIV-1 cDNA and one-LTR circles (8, 38, 44, 49). In addition, episomal HIV-1 cDNAs may be diluted out by cell division (8) or episome-carrying cells could be eliminated by apoptosis or immunodestruction. It is relevant for the latter scenarios that an accumulation of linear unintegrated HIV-1 cDNAs can promote apoptosis under certain experimental conditions in vitro (28, 38, 53). Also, viral transcription can occur off unintegrated HIV-1 cDNA species in T cells, resulting in the low-level expression of HIV-1 proteins, including Nef, Tat, and Env (9, 16, 45, 51, 59, 60). Processed peptides from these viral gene products could be presented by major histocompatibility complex class I complexes on the surface, potentially tagging these cellular HIV-1 reservoirs for cytotoxic-T-cell-mediated immunodestruction.

Although more far-reaching mechanistic studies were not within the scope of the current study, the documented accumulation of HIV episomes provides a basis for the hypothesis that INIs may impact the survival of cells implicated as HIV-1 reservoirs in vivo. First, long-lived, productively infected macrophages may undergo HIV-1 superinfection in vivo (4, 31, 56). An INI could induce an accumulation of unintegrated HIV-1 cDNA of the superinfecting virus, an event which, in turn, might induce apoptotic cell death. In light of the significant contribution of tissue-resident macrophages to the viral burden in vivo (13, 29, 55), depletion of this macrophage reservoir may accelerate the overall HIV-1 RNA load decrease in the plasma. Second, resting CD4 T cells constitute a major HIV reservoir, in which the viral DNA frequently resides in a state of preintegration latency in the nucleus (24). Upon cellular activation, the linear proviral DNA integrates, allowing these T cells to contribute to the HIV RNA load. Consequently, treatment with an INI, but not an RTI, may affect HIV-1 production from the pool of resting CD4 T cells, as has been discussed by Murray and colleagues (42). Besides this direct effect, INIs may favor episome circularization and viral transcription in newly activated CD4 T cells, resulting in an immunological elimination of these cells. Further clinical studies should directly address these hypotheses.

Collectively, the current study provides proof of concept that the action of an INI potently reduces the HIV-1 proviral burden. It demonstrates for the first time markedly elevated levels of unintegrated HIV-1 cDNAs as a consequence of INI treatment in vivo. Because these DNA species have potential biological activities detrimental to virus production and spread, we propose that the perturbation of HIV-1 cDNA species by this novel class of anti-HIV drugs may account for its more rapid kinetics of viral decay compared to RTIs in HIV/AIDS patients.