CpG Oligodeoxynucleotides Activate HIV Replication in Latently Infected Human T Cells*

CpG oligodeoxynucleotides (CpG ODNs) stimulate immune cells via the Toll-like receptor 9 (TLR9). In this study, we have investigated the effects of CpG ODNs on latent human immunodeficiency virus (HIV) infection in human T cells. Treatment of the latently infected T cell line ACH-2 with CpG ODNs 2006 or 2040 stimulated HIV replication, whereas no effects were evident when ODNs without the CpG motif were used. CpG-induced virus reactivation was blocked by chloroquine, indicating the involvement of TLR9. In contrast to the responsiveness of ACH-2 cells, CpG ODNs failed to activate HIV provirus in the latently infected Jurkat clone J1.1. We also studied the effects of CpG ODNs on productive HIV infection and found enhancement of viral replication in A3.01 T cells, whereas again no stimulating effects were observed in Jurkat T cells. CpG ODN treatment activated NF-κB in ACH-2 cells, which was similarly triggered in uninfected A3.01 T cells following exposure to CpG ODNs, indicating that TLR9-induced signal transduction was not dependent on proviral infection. Our study demonstrates that CpG ODNs directly trigger the activation of NF-κB and reactivation of latent HIV in human T cells. Our results point to a novel role for CpG ODNs as stimulators of HIV replication and open new avenues to eradicate the latent viral reservoirs in HIV-infected patients treated with antiretroviral therapy.


CpG oligodeoxynucleotides (CpG ODNs) stimulate immune cells via the Toll-like receptor 9 (TLR9). In this study, we have investigated the effects of CpG ODNs on latent human immunodeficiency virus (HIV) infection in human T cells. Treatment of the latently infected T cell line ACH-2 with CpG ODNs 2006 or 2040 stimulated HIV replication, whereas no effects were evident when ODNs without the CpG motif were used. CpG-induced virus reactivation was blocked by chloroquine, indicating the involvement of TLR9. In contrast to the responsiveness of ACH-2 cells, CpG ODNs failed to activate HIV provirus in the latently infected Jurkat clone J1.1. We also studied the effects of CpG ODNs on productive HIV infection and found enhancement of viral replication in A3.01 T cells, whereas again no stimulating effects were observed in Jurkat T cells. CpG ODN treatment activated NF-B in ACH-2 cells, which was similarly triggered in uninfected A3.01 T cells following exposure to CpG ODNs, indicating that TLR9-induced signal transduction was not dependent on proviral infection. Our study demonstrates that CpG ODNs directly trigger the activation of NF-B and reactivation of latent HIV in human T cells. Our results point to a novel role for CpG ODNs as stimulators of HIV replication and open new avenues to eradicate the latent viral reservoirs in HIV-infected patients treated with antiretroviral therapy.
The immunostimulatory activity of bacterial DNA was first reported in 1984 (1). This effect is mediated by unmethylated CpG motifs within the DNA and can also be observed with synthetic oligodeoxynucleotides (ODNs) 1 containing these CpG sites (2). The ability of CpG sequences to stimulate the immune system is mainly due to their effects on antigen-pre-senting cells. (reviewed in Refs. 3 and 4). CpG ODNs have also been shown to activate T cells via direct (5,6) and indirect mechanisms (7,8).
CpG ODNs act through binding to Toll-like receptor 9 (TLR9) (9), one of 10 members of the Toll-like receptor (TLR) family identified in humans (10). TLRs belong to the superfamily of pattern recognition receptors, which are involved in generating innate immune responses. TLR9 is expressed primarily in B cells and plasmacytoid dentritic cells (9,(11)(12)(13), but a weak TLR9 expression has also been observed in T cells (12).
The immunostimulatory activity of CpG ODNs can be used therapeutically to enhance specific immune responses both in vaccinations and in acute infections and has led to promising results for the treatment of various pathogens, including retroviruses (14 -19). Stimulating the immune system during HIV infection may not only be useful in enhancing antiviral immune responses but may also resolve the problem of viral latency in antiviral treatment. Highly active antiretroviral therapy has had impressive results in the past few years, effectively suppressing virus replication in many patients and reducing virus titers to below clinical detection limits (20 -22). However, as soon as antiviral therapy is interrupted, the virus load increases again, requiring HIV patients to receive life-long maintenance therapy (23)(24)(25). Latent reservoirs, such as infected memory T cells, contribute to this viral reemergence (26), and it is estimated that up to 60 years would be required to eliminate these reservoirs by use of highly active antiretroviral therapy alone (27,28).
It has been proposed that treatment with a combination of cytokines could stimulate the latent reservoirs and thus render them vulnerable to antiviral therapy (29). Clinical studies have been conducted using IL-2 or OKT3 to restimulate HIV from its latent reservoirs (30,31).
Here we show that CpG oligodeoxynucleotides directly stimulate HIV replication in latently infected human T cells. The results of our study help to clarify the role of TLR9 in HIV replication and provide a new potential approach to eradicate latent HIV in infected patients treated with antiretroviral therapy.
Flow Cytometry-To determine expression of HIV p24 and TLR9, cells were fixed with 4% formalin in phosphate-buffered saline and permeabilized with phosphate-buffered saline containing 5% bovine serum albumin and 0.5% saponin. Expression of HIV-p24 was detected with the mouse anti-HIVp24 mAb 183-H12-5C (AIDS Research and Reference Reagent Program). Expression of TLR9 was detected with the mouse anti-TLR9 mAb 26C593 (BioCarta, Hamburg, Germany). As a second antibody, we used a phycoerythrin-labeled anti-mouse IgG antibody (DAKO, Hamburg, Germany). Cells were analyzed by flow cytometry using a FACScan flow cytometer (BD Biosciences). Markers were set according to staining with an isotype-matched control antibody (DAKO).
NF-B EMSA-Following stimulation, cells were washed in ice-cold phosphate-buffered saline, and whole cell extracts were prepared by lysing in a high salt, Nonidet P-40 buffer on ice for 15 min (20 mM HEPES, pH 7.8, 20% glycerol, 350 mM KCl, 1 mM MgCl 2 , 0.5 mM EDTA, 0.1 mM EGTA, 1% Nonidet P-40, 1 mM dithiothreitol, 1ϫ Complete protease inhibitor (Roche Applied Science)). Extracts were cleared by centrifugation at 15,000 ϫ g and stored at Ϫ80°C. Protein concentration was determined by the Bio-Rad microassay. 10 g of total protein was incubated with 0.5 ng of radiolabeled, double-stranded NF-B consensus oligonucleotide sc-2505 (Santa Cruz Biotechnology, Heidelberg, Germany). The probe was end-labeled with [␥ 32 P]ATP (Amersham Biosciences) using T 4 polynucleotide kinase (MBI Fermentas GmbH, St. Leon-Rot, Germany) and purified on a G-50 Sephadex quick spin column (Roche Applied Science). The binding reaction was carried out at room temperature for 20 min in a 20-l volume (15 mM Tris, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 5% glycerol, 5 g of bovine serum albumin, 2 g of poly(dI-dC), Complete protease inhibitor). DNA-binding complexes were resolved on a 5% polyacrylamide (29:1) gel containing 0.4ϫ Tris-borate-EDTA, dried and visualized by autoradiography using Kodak BioMax film. For specificity analysis, 100-fold excess of the unlabeled oligonucleotide or the mutant NF-B oligonucleotide sc-2511 (Santa Cruz Biotechnology) was added to the binding reaction 10 min before the addition of the radiolabeled probe. To confirm the composition of the DNA-binding complexes, antibodies against Rel A (sc-109X, Santa Cruz Biotechnology) and p50 (sc-1190X, Santa Cruz Biotechnology) were preincubated with the extracts for 30 min on ice before the addition of the radiolabeled probe. For densitometric quantification of NF-kB activation, the intensity of NF-B bands as compared with background was determined with the software Quantity One (Bio-Rad).

RESULTS
In this study, we have analyzed the effects of CpG ODNs on latent HIV infection in human T cells. We have exposed latently infected ACH-2 cells to different CpG and non-CpG phosphothioate ODNs and measured HIV-p24 production by intracellular staining and flow cytometry. Under normal conditions, ACH-2 cell cultures produce only low amounts of virus with the majority of cells remaining in a latently infected state. Following an appropriate stimulus, transcription of the provirus begins, and cells start to produce HIV (32,36).
Treatment of ACH-2 cells for 24 h with CpG ODN 2006 resulted in a significant increase of cells producing HIV (Fig.  1A). Similar results were obtained with CpG ODN 2040. Non-CpG sequences, such as ODN 2041 or a mutated analog of ODN 2006, in which all cytidine residues are exchanged for thymidine (ODN 2006 4xTG), did not stimulate HIV replication, indicating that the observed effects were sequence-specific for CpG sites and not related to phosphothioate modification of the ODNs. CpG-triggered reactivation of HIV was dose-dependent as shown for CpG ODN 2006 (Fig. 1B). The kinetics of CpG ODN 2006-mediated HIV expression is depicted in Fig. 1C. Since CpG ODNs have been reported to act via TLR9 (9), we confirmed expression of TLR9 in ACH-2 cells by reverse transcriptase-PCR (data not shown) and by flow cytometry (Fig. 2).
In contrast to ACH-2 cells, the latently infected Jurkat clone J1.1 did not show enhancement of virus replication following CpG treatment (Fig. 3). As a positive control, we have used TPA/ionomycin to show that the provirus in J1.1 cells can be activated. Our findings are consistent with previous studies in which it was reported that Jurkat T cells are insensitive toward CpG signaling, such as activation of NF-B (11). Since J1.1 cells were positive for TLR9 mRNA (data not shown) and express TLR9 protein (Fig. 2B), we suggest either that a signaling defect is located downstream of the receptor or that additional factors (e.g. costimulatory events) are necessary to observe TLR9 signaling in these cells.
Binding of CpG ODNs to TLR9 occurs after cellular uptake of the DNA and requires endosomal acidification (37). Chloroquine, an inhibitor of endosomal acidification, blocks TLR9 signaling and is therefore being used to study TLR9 activity (37)(38)(39)(40). To further confirm involvement of TLR9 in CpG-induced reactivation of HIV, we analyzed the effects of chloroquine on CpG-triggered virus reactivation in ACH-2 cells. As depicted in Fig. 4, chloroquine completely inhibited CpG-mediated virus reactivation at concentrations to inhibit endosomal acidification. Even at maximum chloroquine concentrations, no cell death was observed (according to the FSC/SSC criteria in flow cytometry, data not shown), indicating that inhibition of virus reactivation was not related to any cytotoxic effects.
To analyze whether CpG sequences stimulate HIV replication in cells that are productively infected with HIV, we infected A3.01 T cells with HIV and measured HIV p24 expression by flow cytometry. The CpG ODN 2006 enhanced viral replication as compared with untreated cells, whereas no effects could be observed with the non-CpG sequence 2006 4xTG (Fig. 5). Similar to the situation with J1.1 cells, productively infected Jurkat T cells were insensitive toward CpG-induced enhancement of virus replication (Fig. 5B).
NF-B is an important transcription factor for replication of HIV (41). Moreover, TLR9 signaling is known to activate NF-B in antigen-presenting cells (42,43). To address the role of NF-B in TLR9-induced reactivation of HIV, we analyzed cellular extracts of CpG-treated ACH-2 cells by EMSA. Treat- ment with CpG ODN 2006 activated the p65/p50 NF-B heterodimer. Increased DNA binding was detectable at 4 h, reaching a maximum at 8 h and declining thereafter (Fig. 6, B and C). In parallel to the NF-B studies, we determined HIV expression by flow cytometry. As shown in Fig. 6A, NF-B induction preceded HIV expression, suggesting that CpG-induced HIV replication was initiated by NF-B. It had been shown previously that CpG ODNs do not activate NF-B in Jurkat cells, which correlates nicely with the inability of CpG ODNs to reactivate HIV in these cells (11). In contrast, CpG ODNs activated NF-B in the uninfected A3.01 T cell line, indicating that CpG-triggered signal transduction in these T cells is not dependent on proviral infection but can also be observed in uninfected cells (Fig. 7). A further characterization of the NF-B complexes showed that both the p50 homodimer and the p50/p65 heterodimer were formed in A3.01 T cells following CpG treatment (Fig. 7A). The same NF-B complexes have also been found in CpG-stimulated B cells (44), indicating that the proximal signaling pathways triggered by CpG ODNs in T cells and "classical" CpG target cells are similar. DISCUSSION Here we show that CpG ODNs directly activate HIV replication in latently infected human T cells in vitro. The potential effects of CpG ODNs on HIV replication are currently a matter of debate. Equils et al. (45) recently observed a stimulatory effect of CpG ODNs on HIV replication in HIV-transgenic mouse spleen cells in vitro . It remained unknown, however, whether the increase in virus replication was caused by a direct stimulation of spleen cells by CpG sequences or mediated by indirect effects. Our findings now demonstrate that CpG ODNs can directly trigger signaling events in human T cells that lead to activation of NF-B and reactivation of latent HIV in vitro. There is evidence that these results have some relevance for HIV reactivation in vivo. It has been reported that purified human T cells (mixed CD4ϩ and CD8ϩ T cells) express TLR9 (12). Despite their TLR9 expression effector and memory, CD8ϩ T cells did not respond to CpG ODN treatment (12). However, CD4ϩ T cells, which are the main target cells for HIV, have not been investigated in this study. Our current findings together with some anecdotal evidence from HIV patients suggest that CpG ODN might be able to directly stimulate at least certain subpopulations of CD4ϩ T cells in vivo. For example, it was observed that viral loads increase in HIV- infected patients with opportunistic bacterial infections (46,47). Further evidence that the observed effects could be operative in vivo came from Agrawal and Martin (48), who recently suggested that the unexpected increase in viral loads observed in HIV-infected patients treated with HIV-gag antisense ODNs could have been due to CpG motifs within the administered sequences.
We have recently shown that CpG ODNs are protective in post-exposure treatment of retrovirus-induced disease in the Friend virus mouse model (19). Interestingly, the timing of treatment was a critical factor in treatment efficiency: only post-exposure treatment was protective, whereas preinfection treatment resulted in an accelerated development of virusinduced erythroleukemia (49). These observations indicate that CpG treatment, depending on the time point of inoculation, is able to enhance retroviral replication in vivo, and further studies will have to reveal whether mechanisms similar to those described in this manuscript are responsible for the observed effects.
The biological effects of CpG ODNs on the human immune system are currently a matter of great scientific interest, and our study contributes to the understanding of the biological activities of these substances. Our data point to a novel role for CpG ODNs as stimulators of HIV replication. This effect might be used in the future to eradicate the latent viral reservoirs in HIV-infected patients treated with antiretroviral therapy. To this end, the ability of CpG ODNs to reactivate the HIV provirus in vivo needs to be urgently explored.