Primary Cell Model for Activation-Inducible Human Immunodeficiency Virus

Abstract

Quiescent T lymphocytes containing latent human immunodeficiency virus (HIV) provide a long-lived viral reservoir. This reservoir may be the source of active infection that is reinitiated following the cessation of antiretroviral therapy. Therefore, it is important to understand the mechanisms involved in latent infection to develop new strategies to eliminate the latent HIV reservoir. We have previously demonstrated that latently infected quiescent lymphocytes can be generated during thymopoiesis in vivo in the SCID-hu mouse system. However, there is still a pressing need for an in vitro model of HIV latency in primary human cells. Here, we present a novel in vitro model that recapitulates key aspects of dormant HIV infection. Using an enhanced green fluorescent protein-luciferase fusion protein-containing reporter virus, we have generated a stable infection in primary human CD4⁺ CD8⁺ thymocytes in the absence of viral gene expression. T-cell activation induces a >200-fold induction of reporter activity. The induced reporter activity originates from a fully reverse-transcribed and integrated genome. We further demonstrate that this model can be useful to study long terminal repeat regulation, as previously characterized NF-κB response element mutations decrease the activation of viral gene expression. This model can therefore be used to study intricate molecular aspects of activation-inducible HIV infection in primary cells.

Many human immunodeficiency virus (HIV)-infected patients undergoing highly active antiretroviral therapy (HAART) maintain undetectable viral loads (<50 copies of RNA/ml) for extended periods of time. However, if these patients stop therapy, a rapid rebound in plasma viremia occurs (10, 14, 18, 19, 43, 52, 54). It has been demonstrated that patients taking HAART harbor quiescent memory CD4⁺ T cells that are latently infected with replication-competent HIV, and it is believed that this latent reservoir could be the source of viral rebound that occurs upon the cessation of therapy (2, 7, 12, 22, 40). Currently available antiretroviral agents cannot target latently infected cells, and it is not possible to completely eradicate the infection until all latently infected cells are eliminated. Therefore, the study of HIV latency is crucial to the development of new approaches for eradicating HIV from infected individuals.

Proposed strategies for targeting the latent reservoir generally involve reactivating latent virus in order to destroy the infected cells by the cytopathic effects of the virus, clearance by immune effector cells, or the addition of antiviral agents such as HIV-specific immunotoxins (11, 29-33, 45, 51). These strategies would ideally induce the expression of latent HIV without causing general activation or other undesirable and potentially damaging effects upon uninfected cells. Such strategies therefore rely on a more thorough understanding of the molecular mechanisms governing latent HIV infection. Ideally, latently infected quiescent memory CD4⁺ T cells from HIV type 1-infected patients would be used to study these mechanisms. However, the generation of latently infected CD4⁺ T cells in vivo is rare, and their isolation is difficult because they express no known surface markers to distinguish them from noninfected cells (6, 9, 13). Furthermore, comparison studies of specific genetic variants of HIV are not feasible when working with patient samples because the genotype of the latent provirus cannot be manipulated.

Due to the above-mentioned complications of using patient samples, most research studying the molecular aspects of latent HIV infection has come from cell lines such as ACH-2 (20), U1 (21), or the more recently described J-Lat (23) cell line. Although the information from these studies has been enlightening to the field, the clinical relevance is controversial due to the transformed phenotype of the infected cells. The activation state of the virus is intimately related to that of the infected cell, so it is important to study latent HIV infection in primary human cells and not constitutively active cell lines that contain forms of HIV that may be mutated or aberrantly integrated to sustain latency (15, 16).

Our laboratory has previously developed an in vivo model for studying HIV latency in primary human thymocytes derived from the SCID-hu mouse. In this model, latent infection is thought to occur by the infection of transcriptionally active immature CD4⁺ CD8⁺ thymocytes before these cells convert to mature resting CD4⁺ or CD8⁺ thymocytes during normal thymocyte differentiation. We have utilized the SCID-hu mouse model to characterize latently infected cells, perform small-scale screening for activators of latent HIV, and identify signaling pathways involved in the stimulation of latent HIV in primary cells (3, 5, 6). Further studies have used this model to characterize activation/elimination strategies aimed at depleting the latent reservoir (4).

Although the SCID-hu model of HIV latency has many benefits, it also has some limitations that make it suboptimal for investigating certain aspects of latent HIV infection. In particular, a direct comparison of different HIV variants bearing specific mutations that affect long terminal repeat (LTR)-mediated transcription cannot be performed since these mutations may also affect replication rates, preventing proper quantitative analysis. Therefore, appropriate comparison studies on the effects of specific mutations must be done with a single round of infection. However, current vector-producing techniques do not yield titers high enough to be used for single-round infection in the context of in vivo models such as the SCID-hu mouse. In addition, the SCID-hu mouse model shares negative aspects of many other in vivo models in that it is relatively expensive, is technically complex, and has much longer experimental turnaround times than most in vitro approaches. To circumvent these limitations, we have developed an in vitro model of activation-inducible HIV in primary cells based on the established SCID-hu mouse model. Herein, we extensively characterize the model and demonstrate its utility in studying the contribution of the NF-κB cellular transcription factor on viral activation from quiescence.

MATERIALS AND METHODS

Vector construction.

NLEGFPLuc was derived by replacing enhanced green fluorescent protein (EGFP) from the parental vector DAEGFP (1) with an EGFP-luciferase fusion protein from pEGFPLuc (BD Biosciences) by using the restriction sites AgeI and MluI. NLEGFPLucΔNFκB was generated by mutating the NF-κB sites in the 3′ LTR of NLEGFPLuc by using the QuikChange II XL site-directed mutagenesis kit (Stratagene). First, the 3′ LTR region of NLEGFPLuc was amplified outside of the MluI and AatII sites and cloned into a Zero Blunt TOPO PCR cloning vector (Invitrogen) to produce the subclone LTR-TOPO. The two NF-κB sites were then sequentially mutated using the following primer pairs: 5′-GACATCGAGCTTGCTACAACTCACTTTCCGCTGGGGACTTTC-3′ and 5′-GAAAGTCCCCAGCGGAAAGTGAGTTGTAGCAAGCTCGATGTC-3′ for the upstream NF-κB site and 5′-CAACTCACTTTCCGCTGCTCACTTTCCAGGGAGGCG-3′ and 5′-CGCCTCCCTGGAAAGTGAGCAGCGGAAAGTGAGTTG-3′ for the downstream NF-κB site (underlined sequences represent mutated bases). These modifications have previously been shown to abolish NF-κB binding within this region of the HIV genome (36). The wild-type LTR in NLEGFPLuc was then replaced with this modified LTR using MluI and AatII restriction sites to produce NLEGFPLucΔNFκB.

Vector production and concentration.

Vesicular stomatitis virus G protein-pseudotyped reporter viruses were generated by the transfection of 293FT cells using Lipofectamine 2000 reagent (Invitrogen). 293FT cells were seeded at 5 × 10⁶ cells/10-cm² plate and then transfected the following day with 2 μg of pHCMVG (1) for pseudotyping with vesicular stomatitis virus glycoprotein, 5 μg of pCMVΔ8.2DVPR (1) for packaging, and 5 μg of either vector NLEGFPLuc or vector NLEGFPLucΔNFκB. Integrase-defective virus was generated by substituting the pCMVΔ8.2DVPR packaging plasmid with the integrase-defective packaging plasmid pCMVR8.2DVPR D64E (41), which contains the class 1 D64E mutation in integrase. Cells were incubated in transfection medium for 20 h and then incubated with fresh medium for an additional 30 h before the viral supernatant was harvested. Cellular debris was removed from the supernatant by centrifugation at 1,500 rpm for 5 min in a Sorvall T6000D centrifuge (Sorvall), followed by filtration using a 40-μm cell strainer. Reporter virus was concentrated approximately 100-fold by ultracentrifugation at 50,000 × g for 90 min at 4°C, and the viral pellet was resuspended in phosphate-buffered saline. Titers of viruses were determined by seeding 5 × 10⁴ 293T cells in a 24-well plate and infecting the cells in quadruplicate 24 h later for 2 h at 37°C in the presence of 8 μg/ml polybrene (Sigma). After the infections, fresh medium was added, and the cells were cultured for 3 days before being harvested. The viral titer was then calculated based on the percentage of EGFP-positive cells determined by flow cytometric analysis. The viral input of integrase-defective NLEGFPLuc and integrase-positive NLEGFPLuc was normalized by adding the same amounts of p24, determined in the two viral stocks in parallel.

Cell culture.

CD4⁺ CD8⁺ thymocytes were isolated from human fetal thymic tissues, which ranged in gestational age from 18 to 24 weeks. These tissues were obtained from Advanced Bioscience Resources with appropriate institutional review board approval. Thymic tissue was minced against a sterile stainless steel screen in Iscove's modified Dulbecco medium (Irvine Scientific) with 10% fetal calf serum plus 1% penicillin-streptomycin to obtain a single-cell suspension. The cells were passed through a 40-μm filter before centrifugation at 1,100 rpm for 5 min. Cells were resuspended in magnetically activated cell sorting running buffer (Miltenyi Biotech), and CD4⁺ CD8⁺ thymocytes were isolated by magnetically activated cell sorting using immunomagnetic beads and an AutoMACs cell sorter (Miltenyi Biotech) according to the manufacturer's instructions. CD8⁺ cells were first isolated by positive selection using the human CD8 multisort kit (Miltenyi Biotech), and the magnetic beads were then removed. CD4⁺ CD8⁺ cells were isolated by subjecting the CD8⁺ cells to positive selection using human CD4 microbeads (Miltenyi Biotech). The purity of CD4⁺ CD8⁺ cells after this procedure was typically >98% as assessed by flow cytometry. Cells were mock infected or infected with reporter virus for 2 h at a multiplicity of infection (MOI) of 0.1 in the presence of polybrene (8 μg/ml). Infection medium was removed, and the thymocytes were cultured in Iscove's modified Dulbecco medium containing α-monothioglycerol, l-glutamine, 10% human AB serum, 1% penicillin-streptomycin, 20 U/ml recombinant human interleukin-2 (IL-2) (R&D Systems, Inc.), and 20 ng/ml recombinant human IL-4 (R&D Systems, Inc.). Cells were costimulated for 3 days with soluble anti-CD28 and plate-bound anti-CD3 antibodies in 24-well plates (unless noted otherwise) as described previously (5).

Isolation of quiescent CD4⁺ T cells.

Human peripheral blood mononuclear cells were incubated with a cocktail of mouse monoclonal antibodies against human lymphocyte markers to remove activated CD4⁺ T cells (CD25, CD38, CD69, and HLA-DR [BD Biosciences, San Jose, CA]) and unwanted cell lineages such as CD8 T cells (CD8 [BD Biosciences]), macrophages (CD14 and CD16 [BD Biosciences]), B cells (CD19 [BD Biosciences]), granulocytes (CD123 [Beckman Coulter, Fullerton, CA]), and NK cells (CD56 [BD Biosciences]). Cells stained with the above-described antibodies were removed after incubation with magnetic beads coated with goat antibodies against mouse immunoglobulin G (Miltenyi Biotech, Auburn, CA) and separated using an AutoMACS (Miltenyi Biotech) cell sorter.

Flow cytometric analysis.

To assess the purity, maturation, and EGFP reporter expression of CD4⁺ CD8⁺ thymocytes, 10⁶ cells were stained with CD4-phycoerythrin (Coulter) and CD8-peridinin-chlorophyll-protein complex (Becton Dickinson), and 10,000 events were acquired using a FACSCalibur flow cytometer (Becton Dickinson). Data were analyzed using the CellQuestPro program (Becton Dickinson), and live cells were gated using forward-versus-side-scatter dot plots. The purity of the quiescent T-cell population was determined by staining 5 × 10⁴ cells for the markers listed above in cell culture and isolating a quiescent CD4⁺ T-cell section using monoclonal antibodies conjugated to phycoerythrin or fluorescein isothiocyanate (Coulter). Data acquisition and analysis were conducted as mentioned above. Quiescent cells contained less than 1% contaminating cell populations.

Alu-PCR-based integration assay.

Detection and quantitation of integrated viral DNA were assessed by Alu-PCR as previously described (37). Briefly, we performed two nested PCR amplification steps. In the first preamplification step, we used primers to Alu and gag. This was followed by real-time PCR using internal primers and probes to the LTR-gag junction. To control for any nonintegrated viral DNA, we performed the preamplification step with only the gag primer (linear preamplification of DNA) or without any primers (no preamplification of DNA). Experimental samples were also compared with samples exposed to integrase-defective virus in parallel. A standard curve representing integrated HIV sequences was generated from cells infected with a nonspreading HIV-based reporter vector. A beta-globin standard was used to determine the number of proviruses per cell.

Luciferase assay.

To quantify vector expression, thymocytes were taken from culture and processed according to the Luciferase assay system (Promega) protocol. Cells were washed with phosphate-buffered saline, resuspended in 1× cell lysis buffer (Promega), vortexed, and stored at −70°C. For analysis, samples were thawed and mixed with 100 μl of Luciferase assay reagent (Promega), and the relative light units of the samples were then determined using the Monolight 2010 luminometer (Analytical Luminescence Laboratories) set at a 10-s exposure.

Electrophoretic mobility shift assay (EMSA).

Nuclear extracts were prepared from thymocytes using the NP-40 method for cell membrane lysis (24). Protein levels were quantified by Bradford assay (2a), and 2.5 μg nuclear protein was used for each binding reaction. Proteins were incubated with a double-stranded probe corresponding to the NF-κB binding sites (underlined) of the LTR 5′ labeled with [γ-³²P]ATP: 5′ sense GAGCTTTCTACAAGGGACTTTCCGCTGGGGACTTTCCAGGGAGGT.

In addition, an NF-κB mutant probe was used as a control in which the NF-κB sites were mutated as shown in Fig. 5A. Proteins and probe (approximately 1,000 cpm/sample) were incubated in 1× gel shift binding buffer (Promega) for 20 min at room temperature before resolution on a 5% native polyacrylamide gel in 1× Tris-borate-EDTA. For supershift experiments, antibodies against the p65 (SC-109X; Santa Cruz) and the p50 (SC-114X; Santa Cruz) subunits of NF-κB were preincubated with nuclear extract for 30 min at room temperature before the addition of the probe.

FIG. 5.

NF-κB response elements within the LTR of HIV greatly impact viral activation. (A) Sequence of the two NF-κB response elements within the LTR of HIV. Nucleotides mutated in the NLEGFPLucΔNFkB LTR are highlighted in boldface type. (B) EMSA using either the wild-type LTR probe or the NF-κB mutant probe on cellular extracts from nonstimulated (−) or costimulated (+) CD4⁺ CD8⁺ thymocytes from day 10. Three complexes induced by costimulation and disrupted by preincubation with antibodies to NF-κB subunits (“supershift”) or by NF-κB binding site mutations are indicated by arrows. One smaller complex unaffected by NF-κB antibodies or NF-κB binding site mutations is indicated by an asterisk. (C) Luciferase activity in cells infected with either wild-type or NF-κB mutant reporter virus and cultured as described in the legend of Fig. 1B. Results with the costimulated conditions are plotted as a percentage of expression from wild-type NLEGFPLuc. Results are a representation of three different experiments done in quadruplicate, and the error bars represent standard deviations.

RESULTS

In vitro model of activation-inducible HIV.

Previous work performed in our laboratory has established that HIV can generate a latent infection during thymopoiesis in vivo (5). In this model, transcriptionally active immature CD4⁺ CD8⁺ thymocytes become infected with HIV, and a latent infection develops as these cells convert to the less transcriptionally active mature CD4⁺ or CD8⁺ thymocytes. We have shown that this differentiation process can be recapitulated in vitro using the appropriate cytokines and culture conditions (28). We therefore speculated that infection of CD4⁺ CD8⁺ thymocytes with HIV in vitro could lead to the generation of a latent infection in a manner similar to that which occurs in vivo. In order to accurately quantify the viral expression levels in this model, the reporter vector NLEGFPLuc was constructed (Fig. 1A). This vector contains a gene encoding an EGFP-luciferase fusion protein in place of nef that is expressed under the control of the viral LTR. To focus on the activation of the LTR without confounding effects from other regions of the HIV genome, all of the structural genes and most of the accessory genes of HIV were deleted or attenuated in this reporter virus. Furthermore, by using a replication-incompetent virus, the effects of specific mutations on viral activation could be directly compared in a single round of infection.

FIG. 1.

In vitro model of activation-inducible HIV in primary cells. (A) Schematic of the HIV-based reporter virus NLEGFPLuc used in our model. The vector contains two functional LTRs and is highly attenuated, with the only functional genes being tat, rev, vpu, and EGFP-luciferase. EGFP and luciferase activities were assessed to measure gene expression driven from the HIV LTR. SD and SA, splice donor and acceptor sites, respectively; Ψ, packaging signal; Δgag, truncated gag sequence; RRE, rev-response element. (B) Schematic of the in vitro model used to study activation-inducible HIV infection. CD4⁺ CD8⁺ thymocytes were isolated and infected with NLEGFPLuc on day 0 and then cultured for 7 days to allow sufficient maturation. After 7 days in culture, the cells were divided into nonstimulated or costimulated (anti-CD3 and anti-CD28) conditions for an additional 3 days before the cells were harvested and reporter activity was assessed.

Primary human CD4⁺ CD8⁺ thymocytes were isolated as described in Materials and Methods and then mock infected or infected with NLEGFPLuc at an MOI of 0.1. The cells were then cultured for 7 days in the presence of IL-2 (20 U/ml) and IL-4 (20 ng/ml) to allow sufficient differentiation of the CD4⁺ CD8⁺ thymocytes into mature CD4 or CD8 thymocytes (28). After 7 days, cells were divided into nonstimulated conditions (maintained in IL-2 and IL-4) and costimulated conditions (incubated with anti-CD3 and anti-CD28 antibodies to mimic T-cell receptor engagement) for an additional 3 days before being harvested, and the reporter activity was determined (Fig. 1B).

Viral expression levels.

Reporter activity was assessed after various times in culture to determine the level of vector expression in CD4⁺ CD8⁺ thymocytes infected with NLEGFPLuc. Baseline luciferase activity within infected cells was comparable to that of noninfected cells at all time points prior to stimulation. When the infected thymocytes cultured for 7 days were costimulated for an additional 3 days, the luciferase activity increased more than 200-fold over that of the nonstimulated cells (Fig. 2A). The luciferase activity of the infected nonstimulated cells remained comparable to that of noninfected cells throughout the time course. These results indicate that the virus is maintained in culture with minimal gene expression for at least 10 days and that the virus remains competent to achieve robust levels of expression following cell activation.

FIG. 2.

Assessment of viral gene expression. (A) Luciferase activity of CD4⁺ CD8⁺ thymocytes cultured for 10 days as described in the legend of Fig. 1B. Cells (10⁵) were harvested from noninfected and infected cultures on the days indicated, and the luciferase activity at each time point was determined. Results are representative of two separate experiments. (B) EGFP expression in CD4⁺ CD8⁺ thymocytes infected and cultured for 10 days as described above. The left column shows the cells under phase-contrast microscopy, and the right column shows the cells expressing EGFP by using florescence microscopy. Results are representative of four separate experiments. (C) CD4⁺ CD8⁺ thymocytes infected with NLEGFPLuc and cultured in parallel with noninfected cells as described above were harvested and assayed for EGFP and CD25 expression by flow cytometry on day 10 p.i. Results are representative of two separate experiments.

The same trend in reporter expression was observed when EGFP expression from the infected CD4⁺ CD8⁺ thymocytes was detected. By fluorescent microscopy at day 10 postinfection (p.i.), it was determined that only infected cells that were costimulated expressed EGFP (Fig. 2B). Flow cytometric analysis of infected cells throughout the culture period also confirmed this observation. The EGFP-expressing cells were also brightly positive for the thymocyte activation marker CD25 (Fig. 2C). This approach permitted the percentage and total number of infected cells expressing the reporter virus to be determined. Results from the luciferase and EGFP expression analyses allow us to conclude that infection of CD4⁺ CD8⁺ thymocytes with NLEGFPLuc followed by differentiation in vitro effectively generates a dormant infection. Furthermore, the dual reporter virus used in this system can accurately measure the degree of induced expression, quantity, and phenotype of cells that express HIV after activation.

Effects of thymocyte maturation on the ability to induce expression of dormant HIV.

It has been previously shown that quiescent CD4⁺ cells, which make up the bulk of latently infected cells in vivo, do not support a productive infection by HIV unless the cells are stimulated within a few days of infection (50, 53, 56). The nonpermissive nature of quiescent CD4⁺ cells is a major obstacle to developing an in vitro model of latent HIV infection with this cell type. To highlight the utility of infected CD4⁺ CD8⁺ thymocytes in vitro, we determined how the duration of cell culture after the time of infection affected the ability to induce the expression of HIV. CD4⁺ CD8⁺ thymocytes were infected with NLEGFPLuc, cultured as described above, and then costimulated at various times after infection. In sharp contrast to the previously documented behavior of infection in quiescent CD4⁺ cells (50, 53, 56), CD4⁺ CD8⁺ thymocytes infected with NLEGFPLuc did not generate detectable levels of reporter activity if stimulated prior to 4 days p.i. However, reporter activity could be detected once the thymocytes were cultured for at least 4 days p.i. prior to costimulation, and this signal increased with costimulation at later time points (Fig. 3B). The ability to rescue reporter activity from infected CD4⁺ CD8⁺ thymocytes by costimulation correlated with the development of mature CD4⁺ and CD8⁺ thymocytes in culture, suggesting that these cells are the source of persistent virus (Fig. 3C). In accordance with these results, it has been shown that immature CD4⁺ CD8⁺ thymocytes that receive strong stimulatory signals, such as costimulation, undergo apoptosis (26, 35, 46). The effects of cellular maturation on the ability to induce HIV was further addressed in the EGFP-expressing thymocytes that were infected and cultured as described in the legend of Fig. 1B. Thymocytes that were costimulated following 7 days in culture showed two distinct populations of EGFP-expressing cells, EGFP bright and EGFP dim, and the CD4 and CD8 profiles of these two EGFP-expressing populations were compared, along with total infected thymocytes. (Fig. 3C). The thymocytes expressing bright levels of EGFP exhibited a more mature phenotype including CD4 cells, whereas the cells expressing dim levels of EGFP were less mature double-positive cells. Furthermore, the mature single-positive population of EGFP bright-expressing cells is most likely underrepresented, as these bright EGFP-expressing cells have been shown to up-regulate CD4 and CD8 upon costimulation, so they appear more immature (27). Together, these results indicate that infected CD4⁺ CD8⁺ thymocytes require sufficient time in culture to allow for the development of mature cells capable of responding to costimulation and highlight the unique nature of this system.

FIG. 3.

Effects of thymocyte maturation on the ability to induce expression of dormant HIV. (A) Luciferase activity of CD4⁺ CD8⁺ thymocytes infected with NLEGFPLuc at an MOI of 0.1 and cultured for the indicated times before 5 × 10⁵ cells were placed under nonstimulated or costimulated conditions for an additional 3 days in a 96-well format. Noninfected cells were cultured in parallel with infected cells. At the end of culture under each condition, the cells were harvested, and the luciferase activity was determined. Results are representative of two separate experiments performed in quadruplicate, and the error bars represent standard deviations. (B) CD4 and CD8 expression profiles of the thymocytes from over 7 days in culture. Results are representative of two separate experiments. (C) Maturation state of cells expressing virus. CD4 and CD8 expression profiles of total thymocytes were compared to those of cells expressing bright versus dim EGFP levels following costimulation. The panel on the left indicates the gating used in the EGFP-expressing panels to the right. Results are representative of two separate experiments. C illustrates the same experiment as that shown in Fig. 2C.

Reverse transcription and integration.

Infection of CD4⁺ CD8⁺ thymocytes in the in vitro model system described above produces a stable activation-inducible HIV infection. The behavior of HIV in this system is much different than that of other primary cells, such as quiescent CD4⁺ T cells, which leads to a labile RNA intermediate that is lost 4 days p.i. (53, 56). To establish that the reporter activity detected in the infected CD4⁺ CD8⁺ thymocytes after stimulation is not derived from a partial reverse transcription intermediate, the reverse transcription inhibitor zidovudine (AZT) was added to infected cells at various time points throughout culture. Following 7 days in culture, the different AZT-treated conditions were divided into nonstimulated and costimulated conditions for an additional 3 days before the cells were harvested and the reporter activity was compared. The addition of AZT to the infected cultures was effective only when it was added immediately after the 2-h infection (Fig. 4), resulting in a 70 to 80% decrease in reporter activity. The addition of AZT at later time points (24 h p.i. through 7 days p.i.) had little or no effect compared with the reporter activity of infected cultures without AZT. These results indicate that CD4⁺ CD8⁺ thymocytes cultured in this assay can be infected without a block in reverse transcription that occurs in nonpermissive resting CD4⁺ T cells and that NLEGFPLuc completes reverse transcription within 24 h after infection.

FIG. 4.

Reverse transcription and integration of NLEGFPLuc in infected CD4⁺ CD8⁺ thymocytes. (A) Luciferase assay of cells infected and cultured for 10 days as described in the legend of Fig. 1B. AZT was added at the indicated time periods following infection. Results are a representation of the average of two separate experiments, and the error bars represent the ranges. (B) Alu-PCR indicating levels of integration in thymocytes infected with NLEGFPLuc or an integrase-defective form of NLEGFPLuc on day 7 p.i. prior to costimulation. The results are shown as the number of integrated copies of HIV per 10⁵ cells assessed in triplicate, and the error bars represent standard deviations. (C) Luciferase activity of cells from B that have been cultured for an additional 3 days under nonstimulated or costimulated conditions. Results are representation of two separate experiments assessed in quadruplicate, and the error bars represent standard deviations. Results shown in B and C are derived from the same experiment illustrated in Fig. 2C and 3C.

CD4⁺ CD8⁺ thymocytes infected with NLEGFPLuc and cultured as described in the legend of Fig. 1B were analyzed using Alu-PCR techniques (37, 50) to determine the integration status of the virus on day 7 prior to costimulation. As a control, CD4⁺ CD8⁺ thymocytes were infected in parallel with an integrase-defective form of NLEGFPLuc containing the class 1 D64E mutation in integrase (17, 34, 41). Integrated NLEGFPLuc was detected only in thymocytes infected with the integrase-positive virus (Fig. 4B). Copies of integrated virus correlated well with the rescue of reporter activity following costimulation, as an average of 0.16% of integrated copies seen on day 7 p.i. (Fig. 4B) correlated with the 0.14% of infected thymocytes able to express EGFP on day 10 after costimulation of the same cells (Fig. 2C and 3C). In addition, luciferase activity assayed from the costimulated thymocytes was detected only in the cells infected with the integrase-positive forms of NLEGFPLuc and not the integrase-defective virus on day 10 p.i. (Fig. 4C). These data demonstrate that integration prior to cell activation is required in this model.

Impact of NF-κB on HIV activation.

Induction of the viral LTR is tightly regulated in our in vitro model system. We therefore determined if our model could be used to assess the importance of various LTR-binding transcription factors on the reactivation of HIV. Since replication-incompetent reporter viruses are used in this assay, the effect of specific LTR mutations on viral expression levels can be clearly determined in a single round of infection. This approach allows us to focus on how the induced mutations affect reactivation without the complications caused by additional effects on replication. NF-κB has long been established as an important regulator of HIV expression (36). Therefore, we chose to determine the level of involvement of NF-κB response elements in HIV reactivation in our primary cell model. The NF-κB sites within the enhancer region of the 3′ LTR of NLEGFPLuc were abolished by mutating three nucleotides within each NF-κB response element to previously described sequences (36) that inhibit NF-κB binding (Fig. 5A). The mutations were induced in the U3 region of the 3′ LTR because this region becomes the promoter of the virus after reverse transcription is completed. To confirm that NF-κB binding was abolished, we conducted EMSAs using nuclear extracts from CD4⁺ CD8⁺ thymocytes cultured in vitro for 10 days. As expected, costimulation led to an increase in protein binding to a probe spanning the NF-κB response elements (Fig. 5B, compare lanes 2 and 3). Antibody supershifts confirmed that these DNA binding complexes contained NF-κB subunits p65 (Fig. 5B, lane 4) and p50 (lane 5). However, the NF-κB response element mutations effectively abolished the binding of these complexes in nonstimulated and costimulated cell extracts (Fig. 5B, compare lanes 2 and 3 with 7 and 8). These results show that NF-κB is not abundant in the nonstimulated thymocytes and that NF-κB binding is induced upon stimulation. The NF-κB binding activity in the nucleus correlated with the amount of wild-type reporter virus expression observed in cells subjected to different stimulation conditions. In addition, the mutations induced in the viral vector effectively block NF-κB binding.

We then investigated how the NF-κB mutant virus responds to costimulation in our in vitro assay. Equal numbers of CD4⁺ CD8⁺ thymocytes were mock infected or infected with either NLEGFPLucΔNFkB or the wild-type reporter virus NLEGFPLuc at an MOI of 0.1 and cultured as described in the legend of Fig. 1B. It was found that after costimulation, the luciferase activity from the NF-κB response element mutant virus was consistently reduced by 70% compared to the luciferase activity of the wild-type virus (Fig. 5C). These results indicate that the interaction of NF-κB with the HIV LTR is required for the majority of HIV activation by costimulation of these primary cells. The residual reporter activity from the NF-κB response element mutant virus demonstrates that other factors are also involved in reactivation, and ongoing experiments aim to determine the factors involved.

DISCUSSION

Since the implementation of HAART and the discovery of an extremely stable latent reservoir of HIV that is HAART resistant, there has been a strong emphasis on understanding the mechanisms involved in latent HIV infection. The experiments described here represent the first extensively characterized in vitro model of activation-inducible HIV in primary human thymocytes and the utilization of this model to study the involvement of transcription factor binding sites on HIV reactivation. This model was based upon previous in vivo work performed in our laboratory in which we have shown that immature CD4⁺ CD8⁺ thymocytes are receptive to infection by HIV because of their enhanced activation state compared to more mature CD4⁺ or CD8⁺ thymocytes (5). Here, we show CD4⁺ CD8⁺ thymocytes permit the completion of reverse transcription of HIV-based reporter viruses in vitro within 24 h of infection (Fig. 4A). However, the infected thymocytes do not permit significant expression of the reporter at any time throughout culture without stimulation, which was demonstrated by similar levels of luciferase reporter expression from infected and noninfected cells (Fig. 2A). As the CD4⁺ CD8⁺ thymocytes mature and differentiate into less active CD4⁺ or CD8⁺ cells, they become receptive to stimulatory signals, which in turn allows the activation of the integrated virus. This behavior is exemplified by the increasing levels of reporter expression that can be induced by costimulation when the cells are cultured and allowed to differentiate for a longer duration (Fig. 3A). The EGFP-luciferase fusion protein containing the dual reporter virus NLEGFPLuc used in this assay allows the identification of cells that are expressing virus and provides the ability to quantify levels of viral gene expression from activated cells (Fig. 2). Luciferase activity can be used to quantify the induction of viral expression levels by comparing the >200-fold induction of expression of infected cells that are stimulated to those of infected nonstimulated cells and noninfected cells. The large induction of luciferase expression in infected cells that are stimulated allows extremely accurate comparisons of different parameters. For example, the degree of involvement that specific response elements may have on HIV reactivation can be determined. Here, we show that the NF-κB sites of HIV are responsible for 70% of expression when HIV is reactivated by costimulation (Fig. 5C). We intend to use this in vitro model to determine the involvement of other LTR response elements in the activation of HIV in future experiments.

There has not been an in vitro model to study the mechanisms of latent HIV infection in primary human cells until recently because of the inability to isolate latently infected cells from patient samples and the difficulties of developing culture conditions that allow the study of HIV latency in primary human cells. Most of the information regarding the mechanisms involved in latent HIV infection has come from studies using chronically infected cell lines, which are constitutively active and differ greatly from the quiescent phenotype of the infected CD4⁺ T cells that make up the majority of the latent reservoir. The drastic difference in the cellular phenotype of cell lines compared to in vivo targets makes the relevance of these findings questionable. Recent efforts by Swiggard et al. (47) and Sahu et al. (44) have produced in vitro assays that can generate latent HIV infection in primary human CD4⁺ T cells using spinoculation or coculture techniques, respectively. Although both of these new models of HIV latency using primary human cells are potentially useful for studying different aspects of latent HIV infection, neither one shows the dramatic induction and high sensitivity of reporter expression seen here. Also, the ability to effectively compare mutant viruses is not clear using the aforementioned models.

Latent HIV infection has been shown to exist as two classical forms in patient samples: a preintegration form and a postintegration form. The preintegration form of HIV is common in recently infected resting CD4⁺ T cells where the virus does not receive sufficient cellular factors required to complete reverse transcription (8, 53). This form of latency is usually abortive, since reverse transcription is inefficient unless the cells become activated within approximately 1 to 3 days of infection (50, 53).

Postintegration latency allows the generation of a stable reservoir of latently infected resting CD4⁺ cells that can persist for long periods of time in the presence of HAART. How HIV develops an integrated latent infection in resting CD4⁺ T cells is still largely unknown. It is possible that the in vivo microenvironment of lymphoid organs may yield more permissive conditions to allow viral integration as the cells transiently pass though the region (25, 38, 42, 48, 49, 55). The favored theory for the generation of latent HIV infection is that an activated CD4⁺ T cell becomes infected while reverting back to a resting memory state before the cytotoxic effects of the virus or immune clearance destroys the infected cell (39). However, previous work done in our laboratory has shown that HIV can also generate an integrated form of latent infection in vivo when the infection occurs during thymopoiesis (5). Here, we show that integrated forms of latent HIV infection generated during thymopoiesis can be generated in vitro.

Since latent forms of HIV remain a major barrier to the eradication of infection, new therapeutic approaches targeting the elimination of the latent reservoir are being developed (11, 29-33, 45, 51). The activation/elimination strategy aims to reactivate latent HIV to destroy the latently infected cells while the drug regimen disables newly produced virus. Successful techniques capable of purging the latent reservoir could ultimately lead to the eradication of infection. One of the most potentially useful applications for our in vitro model is to screen for novel activators of latent HIV. Furthermore, this model can be used to study intricate molecular aspects of activation-inducible HIV infection and possibly lead the discovery of new therapeutic agents.