Zidovudine, Lamivudine, and Abacavir Have Different Effects on Resting Cells Infected with Human Immunodeficiency Virus In Vitro

Abstract

We have previously described an in vitro model for the evaluation of the effects of different immunomodulatory agents and immunotoxins (ITs) on cells latently infected with human immunodeficiency virus (HIV). We demonstrated that latently infected, replication-competent cells can be generated in vitro after eliminating CD25⁺ cells with an IT. Thus, by selectively killing the productively infected cells with an anti-CD25 IT we can generate a population of latently infected cells. CD25⁻ cells generated in this manner were treated with nucleoside analog reverse transcriptase inhibitors and subsequently activated with phytohemagglutinin in the presence of the drugs. The antiviral activities of zidovudine (ZDV), lamivudine (3TC), and abacavir (ABC) were evaluated by using this model. 3TC and ABC demonstrated significant activity in decreasing HIV production from recently infected resting cells following their activation, whereas the effect of ZDV was more modest. These results suggest that the differences in antiviral activity of nucleoside analogs on resting cells should be considered when designing drug combinations for the treatment of HIV infection. The model presented here offers a convenient alternative for evaluating the mechanism of action of new antiretroviral agents (J. Saavedra, C. Johnson, J. Koester, M. St. Claire, E. Vitteta, O. Ramilo, 37th Intersci. Conf. Antimicrob. Agents Chemother., abstr. I-59, 1997).

Human immunodeficiency virus (HIV) replication in vivo involves the rapid turnover of CD4⁺ lymphocytes and the release of virions into the plasma (24, 42). The majority of HIV detected in plasma is derived from these newly infected, short-lived, HIV-producing CD4⁺ T cells. Another population of HIV-infected cells consists of latently infected cells. These cells have become a major target of HIV research, since they are an important reservoir of virus and may be key to the persistence of infection (4, 6-8, 10, 11, 16-18, 23, 30, 32, 37-39, 43, 46). The success of highly active antiretroviral therapy in reducing circulating HIV in plasma to levels below the limit of detection in many infected individuals has encouraged investigators to initiate studies aimed at eradicating HIV. However, replication-competent HIV has been isolated from infected individuals after prolonged treatment with highly active antiretroviral therapy (7-11, 16, 17, 23, 30, 37, 43, 45), and several studies have demonstrated that resting CD4⁺ T cells are a long-lived latent reservoir of the virus (4, 6, 9, 10, 14, 16, 23, 38, 46). The virus may remain viable in nonproducing resting cells in a latent form either as integrated (8, 9, 11, 14, 16, 18, 30, 38) or preintegrated DNA with the ability to integrate into host DNA after cell activation (4, 6, 8, 9, 18, 38, 39, 44).

The persistence of HIV infection in patients with undetectable virus may also be secondary to ongoing low-level viral replication (7, 8, 11, 14, 18, 21, 32, 35, 38, 45, 46). This remaining viral replication could be due to protected sanctuaries and to variable tissue concentrations of antiretroviral drugs (7, 18, 21, 32, 38, 45, 46). Findings from two recent studies which used very sensitive techniques showed that, at least in aviremic individuals, latently infected cells do not allow viral replication (10, 23).

Several groups have underscored the need to design therapeutic interventions which target latently infected cells (7, 11, 16, 17, 30, 38, 43, 45, 46). One approach would be to use drugs that are more active in resting cells. If antiretroviral drugs (ARV) can be incorporated into resting cells and remain active for a period of time, these agents may reduce the dissemination of HIV to these cells following reactivation. As new ARV become available, it is therefore important to determine whether they are active in both productively infected and resting cells. This information should facilitate the design of drug combinations capable of suppressing viral replication in different cell populations and perhaps reduce the expansion of the pool of latently infected cells.

An in vitro system to evaluate the effect of ARV on resting cells was previously described (1-3, 5, 27, 31, 34). Thus, by selectively eliminating the activated cells with an anti-CD25 immunotoxin (IT), we obtained a population of resting CD25⁻ cells, some of which are latently infected, that can be cultured in the presence of ARV and subsequently activated to induce viral production. Using a similar in vitro model, the present study was designed to compare the in vitro antiviral activity of three different nucleoside analogs, zidovudine (ZDV), lamivudine (3TC), and abacavir (ABC) (12, 15), on unfractionated cells (containing both productively infected and resting cells) versus that on CD25⁻ resting cells by measuring their capacity for viral p24 production following activation.

MATERIALS AND METHODS

Drugs and IT.

The nucleoside analogs ZDV, 3TC, and ABC were provided by GlaxoSmithKline, Inc. (Research Triangle Park, N.C.). Stocks were prepared in 100% dimethyl sulfoxide at a 1 mM concentration and stored in 1-ml aliquots at −70°C. Drug dilutions were freshly prepared for each experiment. Initial dilutions were based on the 50% inhibitory concentrations for viral isolates, and based on preliminary experiments, we used 0.01 to 0.15 μM ZDV, 0.25 to 5 μM 3TC, and 2 to 15 μM ABC. The anti-CD25 IT, RFT5-dgA, was prepared and purified as previously described (1, 3, 27).

Virus and PBMCs.

Virus stocks and peripheral blood mononuclear cells (PBMCs) were prepared and infections were induced as described previously (3, 27).

Activation with PHA.

The protocol for activation with phytohemagglutinin (PHA) is shown in Fig. 1. PBMCs were isolated from HIV-negative individuals and incubated with HIV. Cells were divided into two groups containing either complete medium (CM) (RPMI, glutamine, 15% fetal bovine serum, 10% interleukin-2, and penicillin-streptomycin) or CM plus 10 nM IT and cultured in 24-well tissue culture plates at a concentration of 10⁶ cells/ml per well. After 3 days in culture, the media were replaced (either CM or CM plus IT). Three days later, cells were washed twice and the study drugs were added at different concentrations. The cells were cultured for 3 days. On the 10th day, both the media and the drugs were replaced and the cells were stimulated by adding PHA (1 μg/ml). Three days later, media were again replaced. Finally, on day 15 after infection, cell-free supernatants were collected and assayed for levels of p24. Cells from treatment groups were cultured in triplicate wells, and all experiments were repeated at least three times.

FIG. 1.

Experimental protocol. PBMCs were isolated from HIV-negative individuals and incubated with HIV. They were divided into two groups and cultured in 24-well plates at 10⁶ cells/well in CM or CM plus 10 nM IT. After 3 days in culture, the media were replaced (either CM or CM plus IT). Three days later, cells were washed twice and the study drugs (ZDV, 3TC, and ABC) were added at different concentrations. Cells were cultured for 3 days, both media and the drugs were replaced, and cells were stimulated by the addition of PHA (1 μg/ml). Three days later, media were again replaced. Finally, on day 15 after infection, cell-free supernatants were collected and assayed for p24 concentrations.

p24 antigen assay.

Concentrations of p24 antigen in cell-free supernatants were measured by using a commercially available enzyme-linked immunosorbent assay kit (NEN, New Life Science Products, Inc., Boston, Mass.).

Comparative effect of nucleosides on unfractionated (CD25⁺and CD25⁻) versus resting (CD25⁻) cells.

We analyzed the activity of these drugs in two different settings: (i) unfractionated cells containing both activated and latently infected cells (CM-nucleoside analog reverse transcriptase inhibitor [NRTI]) and (ii) resting cells (IT-NRTI). To assess the effect of treatment with IT alone, we included two groups of cells without NRTIs but incubated with either CM or CM plus IT. p24 production was determined after PHA activation, since CD25⁻ cells do not produce virus unless they are activated (1, 5, 31). Nevertheless, we compared each culture of resting CD25⁻ cells plus NRTI (IT-NRTI) with a control culture of CD25⁻ cells without NRTIs (IT alone). Hence, the additional reduction of p24 production observed in the IT-NRTI groups reflects the activity of the NRTIs on resting cells and their ability to suppress viral production from these cells following activation. Therefore, in each experiment, we measured the reductions of p24 production induced by the NRTIs at different concentrations under two conditions (unfractionated and resting cells) and calculated the subsequent ratio by comparing the effect in both settings (Fig. 2). If the effect of the NRTI on resting cells is superior to that observed on unfractionated cells, it would demonstrate an additional effect at that particular drug concentration, since the NRTIs in the IT treatment groups were added to the cultures after the CD25⁻ resting cells were isolated (2, 27, 31).

FIG. 2.

Effect of NRTIs in HIV-infected resting PBMCs. The decline of p24 in both the control group (first set of columns) and the NRTI-treated groups (second set of columns) were compared. The resulting value was taken as the effect of that drug at that concentration on HIV-infected resting cells.

Statistical analysis.

Because values did not follow a normal distribution, data are also presented with medians with ranges. Statistical analysis was performed by using Kruskal-Wallis one-way analysis of variance (ANOVA) on ranks for nonparametric data with the SigmaStat 2000 software package.

RESULTS

The effects of the different nucleosides on unfractionated (CM treated) and CD25⁻ (IT treated) cells were evaluated following the addition of PHA to the cultures to activate the cells. The concentrations of ARV were selected on the basis of earlier in vitro studies, and additional concentrations above and below the ranges (0.01 to 0.15 μM for ZDV, 0.25 to 5 μM for 3TC, 2 to 15 μM for ABC) were included.

IT alone reduces p24 concentrations in the controls without NRTIs (CM versus IT).

In all experiments, there was a reduction in p24 production when CM- and IT-treated cells were compared within controls (without NRTIs) due to the effect of the IT on activated cells with a corresponding decrease in virus production. The mean (± standard error of the mean) reduction of viral production following treatment with the IT was 58 ± 10% (Fig. 3). Viability studies demonstrated a mean percentage of viable cells of 89.5% in cells treated with CM for 6 days compared to 88% after treatment with the anti-CD25 IT. These results confirmed the small percentage of activated PBMCs usually present in fresh samples.

FIG. 3.

Reduction of p24 production following treatment with different concentrations of NRTIs. Black bars express cells cultured with CM. Gray bars show cells treated with anti-CD25 IT. The pair of columns on the left in each panel corresponds to the control group (no NRTI treatment). Values represent percentages of p24 production compared to cells cultured only with CM (100%). *, P < 0.05 (compared to controls).

NRTIs reduce p24 production following activation of both unfractionated and resting HIV-infected cells (CM versus CM-NRTI and IT-NRTI).

As expected, every concentration of NRTI evaluated on unfractionated cells (CM-NRTI) resulted in a significant reduction of p24 production compared to the untreated group (CM). The suppression of viral production was dose dependent and increased as drug concentrations were augmented (Fig. 3). The overall reduction of p24 production was 15 to 89.8% with ZDV, 22 to 96.9% with 3TC, and 32 to 94.6% with ABC, depending on the concentration of ARV used. The decrease of p24 production in the NRTI-treated cells was greater when cells were cultured with IT, reflecting the additive effect of both the NRTI on viral replication and the IT in eliminating activated cells. The reductions of viral production with the combination treatment (NRTI plus IT) compared with untreated cells (CM) were 85 to 98.4% with ZDV, 69 to >99% with 3TC, and 54 to >99% with ABC (Table 1).

TABLE 1.

Comparison of effect of each drug concentration on p24 production in resting cells^a

Drug	Concn (μM)	Median (range)b	Mean ± SEMb	Pc
ZDV	0.01	1.03 (0.77-2.64)	1.46 ± 0.35	NS
	0.06	2.57 (1.29-3.64)	2.51 ± 0.48	NS
	0.1	2.73 (0.73-3.47)	2.36 ± 0.60	NS
	0.15	1.50 (1.42-1.62)	1.51 ± 0.06	NS
3TC	0.25	2.26 (0.74-5.68)	2.82 ± 0.90	NS
	2	33.92 (18.34-50.90)	34.27 ± 8.50	<0.05
	4	16.93 (10.85-20.23)	16.24 ± 2.00	<0.05
	5	19.00 (8.52-44.13)	24.01 ± 7.75	<0.05
ABC	2	0.80 (0.52-1.15)	0.81 ± 0.11	NS
	5	3.49 (0.72-6.75)	3.62 ± 0.97	NS
	10	30.00 (18.2-59.00)	33.63 ± 6.66	<0.05
	15	13.05 (6.2-33.00)	16.33 ± 4.12	<0.05

^aIn each experiment, the effect in p24 reduction achieved by NRTI-IT versus NRTI-CM with each drug concentration was compared to the effect of the control group (CM versus IT). The value for the control group was always considered 1 (effect obtained with IT treatment, in each experiment). Values that are >1 indicate activity of that drug concentration on resting infected cells (see text).

^bMedians and means were obtained from a number of repeated experiments (three to six for each drug concentration).

^cThe P value was calculated by Kruskal-Wallis one-way ANOVA on ranks (for medians). NS, not significant.

Effect of different concentrations of NRTIs on HIV-infected resting cells, as assessed following cell activation.

To assess the effect of NRTIs on HIV-infected resting cells, we first compared the activities of NRTIs on unfractionated versus resting cells (CM-NRTI versus IT-NRTI). Then we calculated the difference in p24 production between unfractionated and resting cells without NRTIs (CM versus IT; control group). Finally, we adjusted the differences in p24 production with NRTI (CM-NRTI versus IT-NRTI) with the reduction observed with IT alone in the control group (CM versus IT). This resulting value was calculated for each drug concentration in each experiment as shown in Fig. 2. The median and mean of this ratio were calculated (Table 1). This value represents the effect of NRTI at the given concentration in HIV-infected resting cells following activation, since the drug was added after the CD25⁺ cells had been depleted with the IT. After the PBMCs are treated with the IT, only resting cells remain (2, 31), and therefore, any additional effect obtained with the NRTIs on this treatment group compared with the control group would correspond to their effects on resting cells. This value would express an effect not induced by either the IT itself or the NRTIs alone. Therefore, it indicates the additive effect of the corresponding NRTI and the IT in reducing p24 production after cell activation compared with the control (IT treated).

Significant increases in the mean of this value were observed in the 3TC and ABC treatment groups (Table 1). The maximum additional effect compared with the IT-induced reduction in levels of p24 observed with each NRTI was 2.5 with ZDV at 0.06 μM, 34.3 with 3TC at 2 μM, and 33.6 with ABC at 10 μM. To determine whether this additional effect was significantly different from the effect of the IT alone, we compared the medians of the values of each concentration of NRTI with the control (always considered to be 1, since it represents the effect of the IT alone) by Kruskal-Wallis one-way ANOVA. 3TC at concentrations of 2, 4, and 5 μM and ABC at concentrations of 10 and 15 μM showed significant differences, whereas the additional effect observed with the other concentrations of the NRTIs evaluated were not significant (Table 1).

ZDV did not show a significant effect on the reduction of p24 production in resting cells infected with HIV compared with its effect on unfractionated cells. However, both 3TC and ABC had a significant effect in decreasing p24 production in HIV-infected resting cells after activation. The most significant effects were obtained with 3TC at 2 μM and with ABC at 10 μM. Since latently HIV-infected cells are unlikely to be affected by NRTIs, the mechanism by which some of these drugs can decrease the infection in these cells is unclear. These agents may work by concentrating or maintaining a high rate of activity in resting cells and then preventing HIV infection of resting cells either before or after activation.

DISCUSSION

The importance of latently infected cells in the pathogenesis of HIV infection has been underscored by several studies demonstrating the presence of HIV DNA and viable virus in purified resting memory T cells obtained from HIV type 1-infected individuals at different stages of the disease, including individuals with undetectable plasma viremia (4, 6-11, 14, 16, 17, 23, 30, 35, 37, 38, 43, 45). Such studies have characterized a population of CD4⁺ DR⁻ CD45RO⁺ T lymphocytes which contain integrated (8, 9, 11, 14, 16, 18, 30, 38) and/or unintegrated HIV provirus (4, 6, 8, 18, 38, 39, 44).

In our in vitro model of acute infection, the CD4⁺ CD25⁻ latently infected cells contained predominantly incomplete and unintegrated HIV provirus but also small amounts of full-length HIV DNA (2, 5). Thus, there are differences in the predominant species of HIV provirus present in in vitro latently infected cells compared with latently infected cells purified from HIV-infected individuals. Regardless of the predominant HIV DNA species, the consistent and clear differences observed among the different NRTIs evaluated indicate distinct antiviral activity of certain NRTIs on HIV-infected resting cells.

Our in vitro model has facilitated the direct assessment of the activity of nucleoside analogs on HIV-infected resting cells by directly isolating these cells after treatment with an IT. These cells were isolated, incubated with NRTIs, and then activated to induce virus production. The major findings to emerge from this study are as follows. (i) There are significant differences in antiviral activity among nucleoside analogs in resting cells infected with HIV in vitro. (ii) Among the three agents evaluated, ZDV did not show activity on resting cells infected in vitro, whereas 3TC and ABC had significant antiviral activity in these cells.

Few studies have examined the activity of antiretroviral agents on latently infected cells. Shirasaka et al. analyzed the effect of different NRTIs on resting and activated cells. Although they assessed antiviral activity by stimulating PBMCs and measuring p24 production, they used fresh PBMCs obtained from HIV-negative individuals not previously stimulated as resting cells. Thus, the activated CD25⁺ cells present in fresh PBMCs (3, 31) probably influenced their results. Our results are in agreement with their findings that some NRTIs had better anti-HIV activity on resting cells (ddC and ddI) than others (ZDV and stavudine [d4T]) (36). Gao et al. (19, 20) studied the mechanisms of phosphorylation by using different dideoxynucleoside analogs (ddNs) and classified them into two groups: (i) cell activation-dependent ddNs, such as ZDV and d4T, which were preferentially phosphorylated and exhibited more potent anti-HIV activity in activated cells, and (ii) cell activation-independent ddNs, such as ddI, ddC, and 3TC, which showed more potent anti-HIV activity in resting cells. They reported that in resting cells some NRTIs reached higher ratios of triphosphate nucleoside analog (dideoxynucleoside triphosphate [ddNTP]) compared to the triphosphate nucleoside (deoxynucleoside triphosphate [dNTP]). The efficiency of anabolism to the active triphosphate that is one of the most important limiting steps in the efficacy of NRTIs in controlling virus replication could be due to the difference in the ratio of concentrations between the ddNTP and the dNTP (ddNTP/dNTP ratio). This ratio was highly favorable in some NRTIs compared to others in resting cells, mainly with ddI, ddC, and 3TC. They concluded that the antiviral activity of these ddNs should be substantially different in resting versus activated cells, but direct testing was not performed. In another study, Watson and Wilburn (41) isolated resting cells and analyzed the effect of NRTIs on the synthesis of proviral HIV DNA in these cells. They found that the synthesis of proviral DNA was inhibited by some NRTIs, especially ddC. Similarly, Davis et al. (13) used PCR to analyze the effects of several NRTIs on the inhibition of HIV replication in both activated and resting lymphocytes. They also observed that certain NRTIs, such as ddI, ddC, 3TC, or ABC, had a greater effect in these cells compared with others such as ZDV or d4T.

Some authors have also measured 3TC (29), ABC, or their metabolites (26) in PBMCs from HIV-infected individuals. Moore et al. determined that the diphosphate nucleoside analog (dideoxynucleoside diphosphate) of 3TC reached the highest concentration in PBMCs, whereas ddNTP had the lowest concentrations in these cells. They suggested that the limiting step on 3TC activity may be the phosphorylation of dideoxynucleoside diphosphate (29). Nevertheless, they did not specifically analyze resting cells. Although the intracellular concentrations of NRTIs may be important in controlling HIV infection in vivo, the aim of our study was to analyze differences in antiviral activity between various NRTIs in HIV-infected resting PBMCs in vitro. Thus, we maintained the concentrations of all three NRTIs evaluated close to their 50% inhibitory concentrations to obtain a maximum effect. By doing so, drug concentrations should not be a limiting factor when evaluating their activities in resting cells.

This in vitro model was previously used to characterize the anti-HIV activity of cyclosporine and other immunomodulatory agents that had a clear effect on latently infected cells (1, 3). Others have used p24 production to evaluate different NRTIs in unfractionated in vitro HIV-infected PBMCs (36), but this is the first study to examine the activity of nucleoside analogs based on virus production.

Taken together, these results demonstrate important differences in the antiviral activities of ZDV, 3TC, and ABC on in vitro HIV-infected CD25⁻ cells. The results with ZDV and 3TC were expected based on their cell type-specific pattern of phosphorylation previously described by others (19, 20). It was predicted that ZDV would be less active than 3TC in resting cells. The potent activity of ABC in HIV-infected resting cells was somewhat unanticipated, since this is a relatively new agent and it is not as well characterized, but this result confirms the observations made by Davis et al. with PCR analysis (13). ABC is phosphorylated by a novel enzymatic pathway (15) which might contribute to its activity in resting cells. In the initial clinical trials, ABC appeared more potent than other nucleoside analogs (33, 40), and subsequent clinical experience corroborates the earlier data (22, 25, 28). Whether this in vivo activity is related to its activity in resting (or latently infected) cells is an intriguing possibility which will require additional studies.

Nucleoside analogues are reverse transcriptase (RT) inhibitors, active primarily in newly infected cells when the viral RNA initiates reverse transcription. Therefore, one would predict that HIV-infected cells containing full-length integrated viral DNA would be resistant to the effects of RT inhibitors. Our model is one of acute infection, and previous studies with this model have demonstrated that latently infected cells contain predominantly incomplete HIV DNA transcripts which could be susceptible to the antiviral effects of RT inhibitors (5, 41). Although this may be somewhat different from the situation in vivo, the model is useful for comparing the effects of different NRTIs in resting cells. We analyzed two different population of cells, one with unfractionated PBMCs and another with only resting cells. If the NRTIs used in this study were able to efficiently inhibit HIV in resting cells as well as in activated cells, the effect of each NRTI in both settings would have been similar. Since the effect of reducing HIV production from resting cells following activation with either ABC or 3TC was clearly superior to that of ZDV, we hypothesize that there is a different mechanism underlying this additional effect. Whether it is related to intracellular concentrations, metabolism, or another factor has yet to be determined.

Recent studies have demonstrated the possibility of ongoing HIV replication in patients with viremia below the limit of quantitation (11, 16, 18, 21, 35, 38, 46) and the ability of HIV to infect resting cells (35, 38, 39, 44, 47). If 3TC and ABC metabolites accumulate in resting uninfected cells for significant periods of time, they may prevent, or at least decrease, the spread of the infection by blocking reverse transcription.

Other classes of ARV with good antiviral activity have not been carefully studied with resting cells, and it is not known whether their clinical efficacy in HIV infection may be related to their activity in this cell population. This is an in vitro study and has its limitations. Thus, many other factors may influence the activity of different ARV, such as their intracellular half-lives. For instance, it is not clear whether the large volume of distribution of non-NRTIs (compared to a close in vitro system) or the protein binding and activity of the P-glycoprotein drug transporter, when analyzing both non-NRTIs and protease inhibitors, could interfere with the analysis of these other ARV in vitro. Future studies comparing acutely infected cells with cells obtained from HIV-infected individuals will be required to address these questions. Nevertheless, our results suggest that in designing combinations of ARV, in addition to potency, different patterns of resistance, and the pharmacokinetic interactions among several agents, we could also consider combining drugs with activity in both productively (active) and latently (resting) infected cells, as suggested previously by other authors (19, 20, 36), to optimize the suppression of viral replication.