New Approach To Targeting The Hidden Reservoir Of HIV
ScienceDaily (Oct. 4, 2009) - The drugs used to treat individuals infected with HIV-1 keep the virus under control and dramatically improve prognosis, but they do not eliminate the virus from the body completely, some remains hidden in immune cells known as resting CD4+ T cells. There are currently no clinically acceptable strategies for eliminating this reservoir of HIV-1.
However, Robert Siliciano and colleagues, at Johns Hopkins University School of Medicine, Baltimore, have developed an in vitro system that mimics the situation in people and used it to identify a compound that can get at this hidden HIV-1 and eliminate it from the CD4+ T cells. The researchers report their results in the Journal of Clinical Investigation.
Importantly, this compound does not cause global T cell activation, the side effect that has made other approaches to eliminating the HIV-1 reservoir clinically unacceptable. However, it has other potential toxicity issues that are likely to preclude its use in the clinic.
The authors therefore hope to use their in vitro system to screen more compounds so that a clinically acceptable drug that eliminates HIV-1 can be developed.
1. Hung-Chih Yang, Sifei Xing, Liang Shan, Karen O%u2019Connell, Jason Dinoso, Anding Shen, Yan Zhou, Cynthia K. Shrum, Yefei Han, Jun O. Liu, Hao Zhang, Joseph B. Margolick, Robert F. Siliciano. Small-molecule screening using a human primary cell model of HIV latency identifies compounds that reverse latency without cellular activation. Journal of Clinical Investigation, 2009
Small-molecule screening using a human primary cell model of HIV latency identifies compounds that reverse latency without cellular activation
Hung-Chih Yang1, Sifei Xing1,2, Liang Shan1,2, Karen O'Connell1, Jason Dinoso1,2, Anding Shen3, Yan Zhou1, Cynthia K. Shrum1, Yefei Han1, Jun O. Liu2, Hao Zhang4, Joseph B. Margolick4 and Robert F. Siliciano1,5
1Department of Medicine and
2Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA.
3Department of Biology, Calvin College, Grand Rapids, Michigan, USA.
4Department of Molecular Microbiology and Immunology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA.
5Howard Hughes Medical Institute, Baltimore, Maryland, USA.
Address correspondence to: Robert F. Siliciano, Broadway Research Building, Room 871, 733 N. Broadway, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. Phone: (410) 955-2958; Fax: (443) 287-6218; E-mail: email@example.com.
Published October 1, 2009
The development of highly active antiretroviral therapy (HAART) to treat individuals infected with HIV-1 has dramatically improved patient outcomes, but HAART still fails to cure the infection. The latent viral reservoir in resting CD4+ T cells is a major barrier to virus eradication. Elimination of this reservoir requires reactivation of the latent virus. However, strategies for reactivating HIV-1 through nonspecific T cell activation have clinically unacceptable toxicities. We describe here the development of what we believe to be a novel in vitro model of HIV-1 latency that we used to search for compounds that can reverse latency. Human primary CD4+ T cells were transduced with the prosurvival molecule Bcl-2, and the resulting cells were shown to recapitulate the quiescent state of resting CD4+ T cells in vivo. Using this model system, we screened small-molecule libraries and identified a compound that reactivated latent HIV-1 without inducing global T cell activation, 5-hydroxynaphthalene-1,4-dione (5HN). Unlike previously described latency-reversing agents, 5HN activated latent HIV-1 through ROS and NF-κB without affecting nuclear factor of activated T cells (NFAT) and PKC, demonstrating that TCR pathways can be dissected and utilized to purge latent virus. Our study expands the number of classes of latency-reversing therapeutics and demonstrates the utility of this in vitro model for finding strategies to eradicate HIV-1 infection.
Highly active antiretroviral therapy (HAART) can suppress HIV-1 levels in plasma to below the limit of detection of clinical assays (<50 copies/ml) and reduce the morbidity and mortality of HIV-1 infection. However, HAART alone fails to cure the infection. In particular, HAART leaves latent integrated proviruses unaffected (1, 2). Latent viral genomes reside in a small pool of infected resting memory CD4+ T cells that constitute a stable viral reservoir. In these cells, the provirus remains transcriptionally silent as long as the host cells are in a quiescent state (3-5). This allows the virus to evade host immune surveillance and rebound quickly following discontinuation of HAART. The remarkable stability of the latent viral reservoir necessitates lifelong HAART (6). Given the potential for toxicity and resistance, elimination of the latent reservoir has recently been proposed as a goal worthy of a major scientific effort (7).
Novel therapies targeting the latent reservoir generally involve reactivation of latent virus (2, 7-13). Expression of viral genes renders infected cells susceptible to viral cytopathic effects and immune clearance. Along with HAART, this reactivation strategy could ultimately purge latent virus from infected individuals. Latent viruses respond to T cell activation signals (10, 12, 14-17). However, initial attempts to deplete the latent reservoir through TCR stimulation using anti-CD3 antibodies proved toxic (18, 19). The toxicity likely resulted from nonspecific T cell activation and release of proinflammatory cytokines. Therefore, an ideal treatment should reactivate latent HIV-1 but avoid global T cell activation.
In the search for activators of latent HIV-1, an important tool would be an in vitro model that mimics the latent state of HIV-1 in primary resting CD4+ T cells and allows for high-throughput screening. Some in vitro latency models have been established in transformed T cell lines (20-22). Although useful, these cell-line models are fundamentally different from the resting CD4+ T cells because of their proliferating nature and aberrant signaling pathways. Some primary cell models have been recently developed in thymocytes or CD4+ T cells (23-26). However, a model that yields cells in a truly quiescent state in quantities sufficient for high-throughput screening is lacking. Given that HIV-1 preferentially infects activated CD4+ T cells and that latent viral genomes are mainly found in resting memory CD4+ T cells (27, 28), it is likely that HIV-1 latency is established when infected lymphoblasts transit back to a resting state and persist as memory T cells. The low frequency of the latently infected cells in vivo (5) indicates that the establishment of latency is inefficient. This is due in part to the fact that only a small fraction of lymphoblasts normally survive to become memory cells. Viral cytopathic effects and host immune clearance may also contribute to the loss of infected cells prior to the establishment of latency. The inefficiency with which latency is established in vivo makes the development of an in vitro model especially challenging.
Certain cytokines play a role in T cell survival (29). IL-7 controls the generation and maintenance of memory CD4+ T cells in vivo (30, 31). However, IL-7 also reactivates latent HIV-1 (12, 32), and thus using IL-7 to promote the establishment of latency in vitro has been difficult. The antiapoptotic protein Bcl-2 is a downstream effector of IL-7 signaling and plays an essential role in counteracting the proapoptotic effects of cytokine withdrawal and maintaining the survival of resting memory CD4+ T cells (29, 33, 34). The prosurvival effect via Bcl-2 is the key homeostatic function of IL-7. Overexpression of Bcl-2 largely restores peripheral T cell homeostasis in the absence of IL-7 receptor (35, 36), indicating that the prosurvival ability of IL-7 can be decoupled from the proliferative and metabolic effects of IL-7 (37, 38).
Here, we report a strategy for generation of an in vitro HIV-1 latency model in primary CD4+ T cells that takes advantage of the prosurvival effects of Bcl-2. We developed a method for establishing latent HIV-1 infection in the Bcl-2-transduced primary human CD4+ T cells. We screened libraries of small molecules and identified what we believe to be a novel compound that efficiently reactivates latent HIV-1 but does not induce global T cell activation. Our study reveals a new strategy for reactivating latent HIV-1 and illustrates the application of this in vitro primary cell model in finding novel agents for the cure of HIV-1 infection.
The latent reservoir for HIV-1 in resting memory CD4+ T cells remains a major barrier to virus eradication. The lack of a convenient in vitro latency model in primary resting CD4+ T cells has hampered progress in exploring ways to overcome HIV-1 latency. In this study, we took advantage of the genetically modified primary CD4+ T cells to establish an in vitro HIV-1 latency system. We demonstrated that Bcl-2-transduced cells can reach a profoundly quiescent state in which HIV-1 latency can be established. More importantly, using this model, we performed small-molecule screening and identified 5HN as an activator of latent HIV-1 that, unlike previously described activators, does not induce global T cell activation. These results illustrate the application of this approach to the discovery of novel agents for the eradication of latent HIV-1.
Overexpression of Bcl-2 allows the primary CD4+ T cells to survive in a quiescent state, a key step to establishing HIV-1 latency in vitro. Although overexpression of Bcl-2 may alter the physiology of CD4+ T cells, our extensive analysis revealed that the Bcl-2-transduced cells are similar to freshly isolated CD4+ T cells in both the activated and resting states. Bcl-2 is a downstream antiapoptotic effector of IL-7, an essential cytokine in maintaining the survival of resting naive and memory T cells in vivo. Prior studies have shown that the Bcl-2 transgene can partly compensate for deprivation of IL-7, indicating the important role of Bcl-2 in the homeostasis of resting T cells (35, 36). Because IL-7 also activates latent HIV-1 (12, 32), decoupling Bcl-2 from other IL-7 signaling pathways provided us an opportunity to measure the HIV-1 LTR activity in cells that are in a profoundly quiescent state.
Another concern is that a modified HIV-1 vector was used instead of wild-type virus. Although the modifications increased the yield of latently infected cells, there is the possibility that modified HIV-1 may behave differently than wild-type virus. We have preserved portions of the HIV-1 sequence relevant to the regulation of viral gene expression (LTR, tat, and rev). Therefore, this model is most useful in studying the upregulation of HIV-1 gene expression in latently infected cells, as we have done here. We have shown that the signaling pathways that lead to reactivation of latent HIV-1 are well preserved in this in vitro model. We tested an array of small molecules and cytokines and observed responses very similar to those seen in other primary cell models and latently infected CD4+ T cells from patients on HAART (10, 12, 53, 54). This result justifies the use of this model in investigating the reactivation of latent HIV-1. In order to increase the yield of latently infected cells, we inactivated genes whose products have been associated with cytopathicity. For this reason, the model may not be ideal for studying the fate of cells reactivated from latency or the efficiency with which latency is established. Importantly, we have recently shown that latently infected cells can be obtained using the same approach with a vector in which all of the HIV-1 ORFs except env are functional (L. Shan et al., unpublished observations). However, as expected, the yield of latently infected cells is substantially lower.
Although several in vitro HIV-1 latency models in primary CD4+ T cells have been described recently, certain factors limit their utility in high-throughput screening for activators of latent HIV-1. Several studies have demonstrated that HIV-1 can establish latency by direct infection of resting CD4+ T cells in vitro (40, 67, 68). This strategy does have some advantages because it avoids the activation of T cells and utilizes wild-type HIV-1. However, the efficiency with which latency is established may be lower than that of the activation-dependent pathway described here, and this system is still subject to the short life span of resting CD4+ T cells in an in vitro culture system. Bosque et al. infected activated CD4+ T cells and maintained them in the presence of IL-2 (25). Because IL-2 also activates latent HIV-1, it is uncertain whether IL-2 disturbs the quiescent state of these cells. Some models were created by infection of thymocytes with HIV-1 (24, 69). However, the majority of the resulting cells were resting naive CD4+ T cells rather than memory CD4+ T cells, which harbor latent HIV-1 in vivo. Marini et al. used low doses of IL-7 to generate and maintain memory CD4+ T cells in vitro up to 2 months and established viral latency among these cells (26). Although interesting, this model only yields approximately 20% viable cells and therefore may not produce enough latently infected cells for high-throughput screening. In addition, IL-7 can activate latent HIV-1 (12). This may also confound the results of studies examining the reactivation of latent HIV-1 by other agents. Therefore, our primary cell HIV-1 latency model is the only one in which cells are maintained in vitro in the absence of activating cytokines.
The system described here can serve as an efficient platform to screen for activators of latent HIV-1 that do not induce global T cell activation. Bcl-2 transduction allows us to generate sufficient numbers of latently infected cells for high-throughput screening. In addition, the resting phenotype of the resulting cells recapitulates the quiescent state of primary resting CD4+ T cells. By flow cytometry analysis, we can easily detect changes in parameters that measure T cell activation, for instance, cell size. We illustrated the potential application of this in vitro system by identifying 5HN as an activator of latent HIV-1 that does not induce global T cell activation. 5HN can strongly reactivate latent HIV-1 with an efficiency similar to that of TCR stimulants. High reactivation efficiency is essential for eradicating HIV-1 because even a small number of viruses released from any residual latently infected cells can rebound exponentially after discontinuation of HAART.
Our data suggest that 5HN activates latent HIV-1 through the ROS and NF- κB signaling pathways. The role of ROS in the activation of NF-κB was first described by Schreck et al. (59). Some investigators have proposed a central role for ROS in mediating the ability of extrinsic stimuli to activate NF-κB, although even after more than a decade of studies, there is still uncertainty about the target protein that senses ROS and transduces the signals to activate NF-κB (58). The potential toxicity of ROS also creates concern about the therapeutic application of 5HN. Previous studies revealed differential cellular responses to the different levels of ROS. In response to low levels of ROS, cells upregulate the antioxidant genes, while high levels of ROS induce apoptosis. Interestingly, intermediate levels can stimulate some inflammatory genes, such as NF-κB and AP-1 (58). This may explain why 5HN has a narrow therapeutic window. In addition, 5HN is chemically reactive and affects several cellular proteins (70). Although we showed that 5HN activates latent HIV-1 independent of Pin1, it is possible that other mechanisms are involved. Further studies are required to determine whether we can decouple the toxicity of 5HN from its ability to activate latent HIV-1. One intriguing finding is that 5HN activates latent HIV-1 without causing nonspecific T cell activation. Given that NF-κB is a key transcription factor in controlling innate and adaptive immunity, it is surprising that 5HN stimulates latent HIV-1 by activating NF-κB without inducing global T cell activation. This can be explained by our observation that 5HN activates latent HIV-1 independent of NFAT and PKCΘ. PKCΘ is a master regulator of T cell activation and activates multiple signaling pathways, including the NF-κB, AP-1, and ERK1/2 pathways (62). The importance of NFAT proteins in T cell activation is also well documented (71). Bypassing PKCΘ and NFAT may allow 5HN to avoid full-blown T cell activation. This finding also suggests that reactivation of latent HIV-1 does not necessarily require T cell activation and that the pathways leading to the activation of latent HIV-1 can be separated from those involving in T cell activation.
Because of the high cost and potential toxicities of long-term HAART and the disappointing results from the clinical trials of HIV-1 vaccines and microbicides (72, 73), there is still a pressing need for pursuing the goal of eradication. To cure HIV-1 infection is exceptionally challenging and will likely require combining HAART with agents that can purge latent virus. The identification of 5HN not only expands the number of classes of latency-reversing agents but also demonstrates the possibility of utilizing pathway(s) further downstream of TCR stimulation to avoid global T cell activation. Although the toxicities of 5HN raise concerns for its clinical application, this is a proof of concept for this approach to finding novel strategies to reactivate latent HIV-1 without inducing global T cell activation.
Transduction of primary human CD4+ T cells with Bcl-2. To allow long-term in vitro survival, we transduced primary CD4+ T cells with Bcl-2 using a lentiviral vector, EB-FLV, in which Bcl-2 expression is driven by the constitutively active promoter of elongation factor 1 α (EF1α) (Figure 1, A and B). Freshly isolated primary human CD4+ T cells were costimulated with anti-CD3 and anti-CD28 and then transduced with the Bcl-2-expressing lentiviral vector. Transduced cells were maintained in the absence of TCR stimulants and exogenous cytokines for more than 3 weeks. Viable cells were then isolated using Ficoll-Hypaque density gradient centrifugation. As shown in Supplemental Figure 1A (supplemental material available online with this article; doi: 10.1172/JCI39199DS1), more than 80% of the activated cells are lost during the first 3 weeks of cytokine withdrawal. However, the remaining viable cells have greatly improved survival (see below).
Activated primary T cells die quickly in culture without trophic cytokines (29). Therefore, only Bcl-2-transduced cells survive under the above culture conditions. Bcl-2 was overexpressed in the vast majority of viable cells present after 4 weeks of culture (Figure 1C). In control experiments, cells transduced with the control lentiviral vector that did not express Bcl-2 died within 2 weeks. To further confirm the longevity of the Bcl-2-transduced cells and the stability of Bcl-2 expression, we quantified viability and Bcl-2 expression weekly in cultures of isolated Bcl-2-transduced cells maintained without supplemental cytokines. We found less than 20% cell loss (Supplemental Figure 1B) and stable Bcl-2 expression (Supplemental Figure 1C) over a 4-week follow-up period.
Bcl-2-transduced cells reach a quiescent state similar to that of freshly isolated primary resting CD4+ T cells. Because resting memory CD4+ T cells are a major reservoir for latent HIV-1 in vivo, an in vitro latency model should consist of infected cells in a similarly quiescent state. After 4 weeks in culture without TCR stimulation or cytokines, surviving Bcl-2-transduced cells were characterized with regard to properties unique to primary resting T cells. These include small cell size, absence of cell proliferation and cytokine production, and lack of activation markers.
Resting Bcl-2-transduced cells exhibited small cell size (Figure 2A) and scanty cytoplasm (Supplemental Figure 2A). Using DNA/RNA staining (39-41), we found that the majority of resting Bcl-2-transduced cells were in the G0/G1a phase of the cell cycle, as is the case with freshly isolated primary resting CD4+ T cells (Figure 2B). Both cell populations readily entered the cell cycle following activation with anti-CD3 and anti-CD28 (Figure 2B). Surface expression of classic T cell activation markers such as CD25, CD69, and HLA-DR was very low on resting Bcl-2-transduced CD4+ T cells (Figure 2C). Less than 5% of these cells expressed CD25, and expression of HLA-DR was similarly low. Approximately 15%-20% of these cells had low but detectable levels of CD69. However, following 3 days of activation by anti-CD3 and anti-CD28, more than 90% of cells expressed high levels of CD25 and CD69 and more than 25% of cells expressed HLA-DR (Figure 2C). The resting status of the Bcl-2-transduced CD4+ T cells was further confirmed by measurement of cytokine mRNA levels using real-time RT-PCR (Figure 2D). Both the IL-2 and IFN-γ transcripts were present only at very low levels in the resting Bcl-2-transduced cells, while transcript levels were increased 8-fold and 90-fold, respectively, following activation by anti-CD3 and anti-CD28.
The host transcription factor NF-κB is critical for HIV-1 replication (14-16, 42). Lack of NF-κB activation in resting CD4+ T cells is a major factor in HIV-1 latency (43). We therefore examined the activation state of NF-κB in resting Bcl-2-transduced cells by ELISA measurement of the nuclear levels of NF-κB p65. As Bcl-2-transduced cells returned to a quiescent state, nuclear levels of NF-κB fell to below those seen in freshly isolated resting CD4+ T cells (Figure 2E). Upon stimulation with anti-CD3 and anti-CD28, nuclear NF-κB was readily induced to high levels equivalent to those of activated primary CD4+ T cells (Figure 2E).
An additional characteristic of resting CD4+ T cells is their resistance to productive HIV-1 infection. Using a recombinant HIV-1 vector capable of expressing GFP, we showed that resting Bcl-2-transduced CD4+ T cells were resistant to HIV-1 infection regardless of whether the viruses were pseudotyped with an X4 HIV-1 envelope (Figure 3A) or VSV-G (Supplemental Figure 2B). However, T cell activation overcame this resistance (Figure 3A and Supplemental Figure 2B). Taken together, the above data demonstrate that the Bcl-2-transduced cells reach a resting state similar to that of primary resting CD4+ T cells while retaining the capability to be fully activated.
Because latent HIV-1 primarily resides in memory CD4+ T cells (44), we further characterized Bcl-2-transduced cells for expression of memory T cell markers. Memory T cells are CD45RO positive. Central memory T cells (TCM) constitutively express CCR7 and CD62L, while effector memory T cells (TEM) show heterogeneous expression of CD62L but not CCR7 (45). A large portion of Bcl-2-transduced cells (∼85%) expressed CD45RO (Figure 3B). Interestingly, approximately 30% of cells expressed both CD45RA and CD45RO. Only a few cells (∼6.5%) expressed CCR7. Therefore, we concluded that resting Bcl-2-transduced cells are more like TEM than TCM. Of note, a recent study reported that CD4+ TCM and transition memory T cells (TTM) are 2 major cellular reservoirs for HIV-1 in vivo (46). Determining whether the latently infected cells generated in our model exhibit properties similar to those of TTM cells requires further investigation
Establishment of HIV-1 latency in Bcl-2-transduced resting CD4+ T cells. Latently infected cells are rare in vivo (47) because most infected cells die from viral cytopathic effects or host cytolytic responses before they can revert to a resting state in which HIV-1 gene expression is shut off. Preliminary results confirmed the previously reported cytopathic effects of HIV-1 gene products (48, 49). To increase the efficiency with which HIV-1 latency was established, we mutated the gag, vif, vpr, vpu, and nef genes of the reference strain NL4-3. Point mutations were introduced to produce premature stop codons in each gene except for nef, which was truncated (Figure 4A). Our primary goal was to study the reactivation of latent HIV-1, and thus the rev and tat genes were left unchanged because their products are crucial for the expression of viral genes. To keep splicing sites intact and mimic the processing of viral transcripts, the overall genome structure of HIV-1 was preserved. Destabilized EGFP was introduced to track the viral infection. Compared with NL4-3, this modified viral vector, pNL4-3-∇6-drEGFP, gave dramatically improved viability of infected Bcl-2-transduced cells. As shown below, this approach made it possible to obtain large numbers of latently infected primary resting CD4+ T cells that could be used to study upregulation of HIV-1 gene expression. The ultimate yield of latently infected cells was much lower (approximately 5- to 10-fold) when Bcl-2-transduced cells were infected with an HIV-1 construct in which all the reading frames except env were open. For this reason, subsequent experiments were performed with the NL4-3∇6-drEGFP virus.
To establish HIV-1 latency, Bcl-2-transduced cells were activated and then infected with pseudotyped NL4-3-∇6-drEGFP. Because the env gene is disrupted by insertion of EGFP, an X4 envelope was provided in trans to achieve single-round infection. The infection rate, based on the percentage of GFP+ cells 3 days after infection, was 5% to 10%. Following more than 4 weeks of culture in the absence of cytokines and other activating stimuli, approximately 20%-30% of cells that initially expressed GFP became GFP negative. We then isolated the GFP-negative cells by sorting. To determine whether latency had been established, the GFP-negative cells were activated and analyzed for GFP expression by flow cytometry. Of the sorted GFP-negative cells, 1% to 3% could be induced to express GFP by activating stimuli such as anti-CD3 plus anti-CD28 or PMA (Figure 4, B and C). In control infected cultures that were not stimulated, the fraction of GFP-positive cells remained less than 0.2%. Taken together, these results indicate that latent infection was established in this system. We could further increase the frequency of latently infected cells by enriching the GFP-positive cells via sorting following the initial infection (Supplemental Figure 3A). The frequency of latently infected cells using this strategy could reach 15%-50% (Supplemental Figure 3, A and B). Although this enrichment increased the frequency of latently infected cells, it did not increase their absolute number. Our preliminary data showed that populations containing 1%-3% latently infected cells generated without enrichment were adequate for high-throughput screening. We therefore performed subsequent experiments without enrichment (Figure 4B).
An important feature of the stable reservoir for HIV-1 in vivo is that the viral genome is integrated into a host-cell chromosome (27, 47). To prove that the latently infected cells generated in this system contain an integrated form of the viral genome, we activated the cells in the presence of an HIV-1 integrase inhibitor, raltegravir. The concentration used was sufficient to block 99% of integration events (50). Raltegravir did not block the reactivation of latent HIV-1 (Figure 4D). This result confirms that this in vitro system is a model for postintegration latency.
Even after prolonged culture in the absence of activating stimuli, a small fraction of GFP-negative cells still expressed low levels of activation markers such as CD25, CD69, and HLA-DR. To exclude the possibility that latent HIV-1 might reside in the cells with some extent of activation, we removed these CD25+, CD69+, or HLA-DR+ cells by negative depletion. A very similar frequency of cells with inducible GFP expression was found in the CD25-CD69-HLA-DR- cells, excluding the possibility that latent HIV-1 resided in the cells with low levels of activation. We also evaluated the cell-cycle status of latently infected cells using DNA/RNA staining. As shown in Figure 4E, the majority of latently infected cells were in G0/G1a phase. Taken together, these observations demonstrate the establishment of HIV-1 latency in vitro in resting Bcl-2-transduced primary CD4+ T cells.
To investigate whether there was any residual integrated provirus that did not respond to T cell activation signals in this system, we sorted the GFP- cells following activation of latently infected cells by anti-CD3 and anti-CD28 for 3 days. The frequency of cells with proviral DNA among the GFP- cells determined using real-time PCR was similar to the percentage of GFP+ cells following activation. This suggests that approximately 50% of integrated proviruses were unresponsive to T cell activation (refer to Supplemental Material for details). Whether these viruses can be reactivated by subsequent rounds of activation requires future exploration.
The signaling pathways leading to the reactivation of latent HIV-1 remain intact in this in vitro model. To validate that the signaling pathways known to induce reactivation of latent HIV-1 remain intact in this in vitro model, we tested known HIV-1 activators, including TCR agonists, mitogens, cytokines, and small molecules alone or in combination. Representative flow cytometry plots are shown in Supplemental Figure 4, A and B. The results are summarized in Figure 5, A and B. PMA, prostratin, and 12-deoxyphorbol 13-phenylacetate (DPP), known activators of PKC, all strongly activated latent HIV-1. Ionomycin, an activator of nuclear factor of activated T cells (NFAT), reactivated a smaller fraction of latent HIV-1. Interestingly, valproic acid (VA), a known histone deacetylase (HDAC) inhibitor and HIV-1 activator (13), barely activated latent virus, even at concentrations as high as 5 mM. However, trichostatin A (TsA), a more potent HDAC inhibitor, activated a greater fraction of the cells at a concentration of 200 nM. Hexamethylene bisacetamide (HMBA), which activates latent HIV-1 in some cell line systems through regulation of active and inactive forms of positive transcription elongation factor b (pTEFb) (51, 52), only caused slight activation of latent HIV-1 at a concentration of 5 mM in our assay.
We also tested some cytokines that are known to activate latent HIV-1 (12, 53). Cytokines were used at concentrations previously reported to induce activation of latent HIV-1 in various primary cell systems. IL-7 alone strongly induced reactivation of latent HIV-1 as previously reported (12, 32), while TNF-α, a well-known activator of latent HIV-1 in transformed T cell lines (16, 21), activated only a small fraction of latent virus (Figure 5B). Consistent with this observation, previous studies using the resting CD4+ T cells isolated from patients on HAART also revealed the low activity of TNF- α in reactivation of latent HIV-1 (53). However, a combination of IL-2, IL-6, and TNF-α dramatically increased the reactivation of latent HIV-1 in our system. This response is also similar to that reported in resting CD4+ T cells isolated from patients on HAART (53).
Overall, the responses of the latently infected cells generated in this system to known activators were very similar to those previously reported in other primary cell systems or in resting CD4+ T cells isolated from patients on HAART (10, 12, 53, 54). These results strongly suggest that this in vitro HIV-1 latency model can mimic the latently infected CD4+ T cells present in vivo.
Screening of small-molecule libraries for compounds that reactivate latent HIV-1. Using the system described above, we screened a library of 2000 drugs and natural products and another set of more than 2400 compounds from the Johns Hopkins Drug Library (JHDL) (55). Representative results are shown in Figure 6A. In total, 9 and 8 compounds that could activate latent HIV-1 were discovered in the former and latter libraries, respectively. For further study, we selected 1 of these compounds, 5-hydroxynaphthalene-1,4-dione (5HN), because it was detected in both libraries and displayed the highest capacity to reactivate latent HIV-1. 5HN is a natural quinone (Figure 6B) found in the leaves, roots, and bark of the black walnut tree. 5HN was comparable to the combination of anti-CD3 and anti-CD28 antibodies in its ability to reactivate latent HIV-1 (Figure 6A and Supplemental Figure 5A).
To confirm the effect of 5HN on latent HIV-1, we used real-time RT-PCR to examine levels of HIV-1 transcripts before and after stimulation with 5HN. Viral transcription was induced up to 8.8-fold in a dose-dependent manner by 5HN (Figure 6C). The ED50 of 5HN in reactivating latent HIV-1 was 0.5 μM (Supplemental Figure 5B), while the LD50 measurements for freshly isolated and Bcl-2-transduced primary CD4+ T cells were 9.9 μM and 7.7 μM, respectively (Supplemental Figure 5C). These results indicate the narrow therapeutic range of 5HN.
We also examined the effect of 5HN in a transformed cell line model of HIV-1 latency, the J-Lat system (22). We measured the expression of viral marker GFP and viral transcripts. 5HN also strongly activated latent HIV-1 in J-Lat cells (Supplemental Figure 6, A and B). Interestingly, although the potency of 5HN to reactivate latent HIV-1 was comparable to that of PMA in our primary cell model, 5HN was less effective than PMA in J-Lat cells. These results highlight the importance of using primary cells to screen for compounds that target latent HIV-1.
5HN does not cause global T cell activation. Agents that induce latent HIV-1 by causing global T cell activation are likely to be too toxic for clinical use. Therefore, we determined whether 5HN activates T cells. Unlike cells stimulated with PMA or anti-CD3 and anti-CD28 antibodies, the 5HN-treated latently infected cells retained the same small size as untreated cells as determined by flow cytometric analysis (Figure 6D). We further showed that 5HN does not cause upregulation of classic T cell activation markers such as CD69, CD25, or HLA-DR on freshly isolated resting CD4+ T cells (Figure 7A). Since the toxicity induced by T cell-activating agents is likely to be due to cytokine release, we determined whether 5HN induced production of T cell cytokines. IL-2 expression was not induced by 5HN, while IFN-γ transcripts were upregulated by approximately 10-fold. However, this level of IFN-γ activation is much less than that induced by costimulation with anti-CD3 and anti-CD28, which caused a greater than 1000-fold upregulation (Figure 7B). Importantly, at concentrations that induce reactivation of latent HIV-1, 5HN did not render uninfected resting CD4+ T cells susceptible to HIV-1 infection (Figure 7C). Finally, 5HN did not stimulate the proliferation of CD4+ T cells. This was evidenced by measurement of DNA/RNA staining (Figure 7D) or the dilution of CFSE in the 5HN-treated cells (Supplemental Figure 7). The above results demonstrate that 5HN did not cause global T cell activation at concentrations that activated latent HIV-1.
ROS are involved in the reactivation of latent HIV-1 by 5HN. 5HN is a quinone that can be reduced to a semiquinone radical by enzymes such as NADPH oxidoreductase. Under aerobic conditions, the semiquinone radical then generates superoxide anion (O2-) and hydrogen peroxide (H2O2), which induce oxidative stress (56, 57). Because ROS can indirectly activate the host transcription factor NF-κB that can in turn activate latent HIV-1 (58, 59), we hypothesized that induction of oxidative stress by 5HN may play a role in the reactivation of latent virus. To test this hypothesis, we first assessed the effects of 5HN on the production of ROS in freshly isolated primary resting CD4+ T cells using dihydrorhodamine 123 (DHR123). DHR123, a nonfluorogenic dye, is oxidized intracellularly by ROS into the fluorescent dye rhodamine 123 (60). Production of ROS in primary CD4+ T cells was detected within 2 hours of addition of concentrations of 5HN known to reactivate latent HIV-1 (Figure 8A). Because ROS can indirectly activate NF-κB, we next assessed the activation of NF-κB by measuring the nuclear levels of NF-κB using an ELISA-based assay. 5HN activated NF-κB in a dose-dependent manner (Figure 8B). To ensure that 5HN activates NF-κB in primary CD4+ T cells, we also quantified the transcripts of IκBα, an NF-κB-responsive gene, and showed that 5HN stimulates the expression of IκBα (Figure 8C). Antioxidants like N-acetylcysteine (NAC) or pyrrolidine dithiocarbamate (PDTC) suppress the effects of ROS on the activation of NF-κB (59). Consistent with this observation, preincubation of the latently infected cells with NAC or PDTC dramatically suppressed the stimulatory effect of 5HN on the expression of latent HIV-1 (Figure 8D). This result supports the role of ROS in the activation of latent HIV-1 by 5HN.
To further confirm that NF-κB is required for the activation of latent HIV-1 by 5HN, we mutated the 2 tandem NF-κB-binding sites in the enhancer region of 3' LTR of pNL4-3-∇6-drEGFP (mκ2-LTR- NL4-3-∇6-drEGFP). Following reverse transcription and integration into the genome of infected cells, the 5' LTR of integrated reporter virus is derived from the 3' LTR that contains mutated NF-κB-binding sites (61). The response to 5HN in the cells latently infected with mκ2-LTR NL4-3-∇6-drEGFP was much lower than that in cells infected with wild-type NL4-3-∇6-drEGFP (Figure 8E), indicating the essential role of NF-κB in the signaling pathways downstream of 5HN. Interestingly, at higher concentrations of 5HN (5 μM), there was partial activation of the mκ2-LTR NL4-3-∇6-drEGFP reporter virus, suggesting the possibility of NF-κB-independent pathways for the reactivation of latent HIV-1 by 5HN.
PKCΘ is a master regulator in TCR signaling pathways (62). Activation of PKC leads to activation of key transcription factors in the immune response, including NF-κB and activator protein 1 (AP-1). Because 5HN activates latent HIV-1 via NF-κB, we determined whether the effects of 5HN depend on PKC. Using the pan PKC inhibitor Go6983, we found that inhibition of PKC completely suppressed PMA-induced activation of latent HIV-1 and reduced by approximately 50% the response to anti-CD3 and anti-CD28 costimulation. However, it almost had no effect on the activation of latent HIV-1 by 5HN. This suggests that 5HN acts downstream of PKC in the NF-κB signaling pathway leading to the activation of latent HIV-1 (Figure 9A). NFAT, another essential transcription factor for T cell activation, is also known to promote HIV-1 gene expression (63). To exclude the possibility that NFAT plays a role downstream of 5HN, we showed that cyclosporin A (CsA), an inhibitor of the NFAT signaling pathway, does not suppress the effect of 5HN on the reactivation of latent HIV-1, while it inhibits the effect of ionomycin (Figure 9B).
5HN is also known to inhibit Pin1, a peptidyl-prolyl isomerase that plays an important role in transcription regulation (64, 65). To determine whether inhibition of Pin1 is involved in reactivation of latent HIV-1, we knocked down Pin1 expression in the latently infected Bcl-2-transduced cells using Pin1-specific RNAi. Downregulation of Pin1 did not activate latent HIV-1 (data now shown). We also demonstrated that fredericamycin A, a Pin1 inhibitor (66), failed to induce latent HIV-1 in our in vitro primary cell model (data not shown). Together, these results suggest that inhibition of Pin1 by 5HN is not responsible for reactivation of latent HIV-1.