Molecular Control of HIV and SIV Latency
The HIV latent reservoirs are considered as the main hurdle to viral eradication. Numerous mechanisms lead to the establishment of HIV latency and act at the transcriptional and post-transcriptional levels. A better understanding of latency is needed in order to ultimately achieve a cure for HIV. The mechanisms underlying latency vary between patients, tissues, anatomical compartments, and cell types. From this point of view, simian immunodeficiency virus (SIV) infection and the use of nonhuman primate (NHP) models that recapitulate many aspects of HIV-associated latency establishment and disease progression are essential tools since they allow extensive tissue sampling as well as a control of infection parameters (virus type, dose, route, and time).
HIV latency is a key hurdle to curing HIV. The HIV latent reservoirs are defined as a cell type or anatomical site where a replication-competent form of the virus persists for a longer time than in the main pool of actively replicating virus (Van Lint et al. 2013). This definition mainly restricts the viral reservoirs to latently infected resting CD4+ memory T cells carrying stably integrated, transcriptionally silent but replication-competent proviruses. These cells do not produce virus particles while in resting state, but can give rise to infectious virus following activation by several stimuli, leading to viral rebound when antiretroviral therapy (ART) is stopped (Chun et al. 1995, 1997, 2000; Finzi et al. 1999; Siliciano et al. 2003; Davey et al. 1999). A less conventional, wider definition of HIV reservoirs has also been proposed: all infected cells and tissues containing all forms of HIV persistence that can participate in HIV pathogenesis (Avettand-Fenoel et al. 2016). This definition includes defective proviruses which participate to HIV pathogenesis through viral transcription and synthesis of viral proteins without new virion production. These proteins can induce and maintain immune activation, thus participating in the vicious circle of HIV pathogenesis (Avettand-Fenoel et al. 2016).
The mechanisms conducting to the establishment of HIV latency but also to its maintenance probably vary from one patient to the other, from one tissue or one anatomical compartment to the other, and also from one cell type to the other (Darcis et al. 2017). Therefore, a cure for HIV is unlikely achievable without considering all latent cellular and anatomical reservoirs such as the brain (Kumar et al. 2014).
HIV is divided into HIV type 1 (HIV-1) and HIV type 2 (HIV-2). HIV-1 is responsible for the HIV pandemic and is related to viruses found in chimpanzees and gorillas, while HIV-2 is related to viruses found in primate sooty mangabey. HIV-1 may be further divided into groups (M, N, O, and P) and subtypes within the M group. Simian immunodeficiency virus (SIV) infection and the use of nonhuman primate (NHP) models that recapitulate HIV-associated disease progression are essential tools. Indeed, NHP models of ART-treated macaques infected with the simian immunodeficiency virus of macaques (SIVmac), which is more closely related to HIV-2 in comparison with HIV-1, have been validated and help to characterize the type, establishment, maintenance, and activation of latent viral reservoirs (Deleage et al. 2016). Importantly, unlike nonpathogenic infection in its African natural host, SIVmac induces an AIDS-like disease in Asian rhesus macaque monkeys with similar symptoms and immunological consequences seen in HIV-infected humans. The use of SIV latter in this chapter specifically refers to SIVmac.
NHP infected with SIV provides several significant advantages, including the possibility to perform extensive tissue sampling in animal and elective necropsy. Since the huge majority of viruses persists under ART resides in tissues that are difficult to access in human clinical settings, this is undoubtedly the main benefit.
During the past few years, important progress has been made to characterize the viral reservoirs, to understand the molecular mechanisms underlying HIV/SIV latency, and to better investigate and address the crucial questions of the complexity, diversity, and dynamics of these mechanisms.
In this chapter, we consider our present knowledge of the molecular mechanisms involved in HIV-1 and SIV latency. To begin, we present a brief description of the HIV-1 and SIVmac promoters, which will be of great importance for the subsequent discussion.
2 The HIV-1 and SIVmac Promoters
The strength of the HIV-1 promoter is modulated by cellular factors and its chromatin environment (see below). Indeed, the 5’LTR of HIV-1 contains several DNA-binding sites for various cellular transcription factors (TFs), including Sp1 and NF-κB, that are important for HIV-1 replication, whereas other sites, such as NF-AT, LEF-1, COUP-TF, Ets1, USF, and AP-1 binding sites, enhance transcription without being indispensable (Colin and Van Lint 2009).
In the absence of Tat, critical TFs, such as NF-κB and Sp1, are required for the formation of the pre-initiation complex leading only to the production of short transcripts, while in the presence of Tat, transcription is enhanced and full-length viral transcripts are synthetized. However, in addition to its classically recognized role in the induction of transcriptional elongation and chromatin remodeling, Tat may also influence transcriptional initiation by facilitating the assembly of the pre-initiation complex requiring the Sp1 and NF-κB binding sites (Brady and Kashanchi 2005). Interestingly, in this context, recent studies from Ben Berkhout’s laboratory demonstrate that Tat(HIV) and Tat(SIV) also stimulate HIV-1 or SIV gene expression, respectively, independent of the TAR hairpin, via Sp1 sequence elements in the U3 promoter region (van der Velden et al. 2012; Das et al. 2011).
The three Sp1 binding sites present on the core promoter play a role on HIV-1 transcription recruiting the pre-initiation complex and the transcriptional factor Sp1 that serves as a recruitment platform for modifying chromatin complexes. The TF Sp1 is a ubiquitous factor that can lead to a positive or negative transcriptional effect depending on additional recruited factors. Sp1 bound to the U3 sites can have a negative effect by recruiting histone deacetylases (HDAC1 and HDAC2) to promote histone H3 and H4 deacetylations (Marban et al. 2005, 2007). In microglial cells, the CNS-resident macrophages, this recruitment requires the cofactor CTIP-2 (COUP-TF interacting protein 2). Indeed, the group of Rohr, in collaboration with our laboratory, has demonstrated that Sp1 recruits a multi-enzymatic chromatin-modifying complex including HDAC1, HDAC2, and SUV39H1 to the viral promoter, where CTIP-2 allows deacetylation of the ninth lysine of the N-terminal tail of histone H3 (H3K9), which is a prerequisite for H3K9 trimethylation by SUV39H1 (Marban et al. 2005). This last histone modification allows heterochromatin protein 1 (HP1) binding and polymerization. Interestingly, the Rohr’s group reported displacement of CTIP-2 and subsequent recruitment of CREB-binding protein (CBP) through Sp1 following HIV-1 activation with phorbol esters (Marban et al. 2007). In CD4+ T lymphocytes, another study demonstrated that c-Myc is recruited to the HIV-1 5’LTR by Sp1 and in turn recruits HDAC1 in order to blunt HIV-1 promoter expression (Jiang et al. 2007). Following activation, cellular histone acetyltransferases (HATs), including p300/CBP, PCAF, and Gcn5, are recruited to the promoter region, leading to the acetylation of both H3 and H4 histones via several TFs such as Sp1 (Marsili et al. 2004). Interestingly, HMBA causes the release of P-TEFb from HEXIM1 and triggers CDK9 recruitment to the HIV-1 5’LTR via an unexpected interaction with the transcription factor Sp1 (Choudhary et al. 2008).
Otherwise, NF-κB binding sites are found in the enhancer region of all primate lentiviral LTRs, although their numbers may vary between different groups of SIV and HIV-1. Most subtypes of pandemic HIV-1 group M strains (A, B, D, F, G, H, J, and K) and some SIVs contain two NF-κB binding sites located −104 to −80 bp upstream of the transcriptional start site (Fig. 1a). However, HIV-1 group M subtype C strains, which account for almost 50% of HIV-1 infections worldwide, typically contain three binding sites for NF-κB in their enhancer region (Heusinger and Kirchhoff 2017). In contrast, subtype A/E recombinants of HIV-1 group M, the human immunodeficiency virus type 2 (HIV-2), and several SIV lineages contain just a single NF-κB binding site. Typically, mutations in the NF-κB binding sites of HIV-1 LTRs prevent efficient proviral transcription.
Another well-characterized cellular TF is the C/EBP (CCAAT/enhancer-binding protein) family for which three binding sites have been identified in the HIV-1 LTR and four binding sites in the SIVmac LTR (Ravimohan et al. 2010; Hogan et al. 2003). Functionally, these sites are involved in activation of HIV-1 transcription and are important for viral replication in the monocyte–macrophage lineage, but not in T cell lines. The regulation of HIV-1 transcription and replication in macrophages is mediated primarily by the two isoforms of C/EBPβ, the liver-enriched transcriptional activator protein (LAP) and liver-enriched transcriptional inhibitory protein (LIP) translated from the second and third in-frame AUG, respectively, and in these cells at least one functional C/EBP binding site within the HIV-1 LTR is necessary for basal level transcription and replication (Ravimohan et al. 2012). In the context of SIV, three of the four sites have been shown acting as negative regulators of SIV basal transcription, while the last binding site is associated with positive regulation of basal viral transcription [reviewed in (Liu et al. 2009)]. These differences could be explained by the differential recruitment to the SIV LTR of the C/EBPβ2 isoform (LAP) or the C/EBPβ3 isoform (LIP), which present an activator or repressor activity, respectively (Barber et al. 2006).
HIV-1 and SIV transcriptions are consequently coupled with the cellular activation status and by the abundance of cellular transcription factors that can either induce or repress viral promoter activity depending on the cell types. Interestingly, besides the presence of DNA-binding sites in the HIV-1 promoter region, several ubiquitous and cell-specific TFs have also been shown to be recruited to part of the pol gene coding for the integrase and to have an important impact on viral infectivity [(Goffin et al. 2005), reviewed in (Van Lint et al. 2013)].
Moreover, nucleosome positioning in the HIV-1 promoter appears to be specific and dynamic, supporting a major implication during latency and transcriptional activation. In latent conditions, two nucleosomes (named nuc-0 and nuc-1) are precisely situated at the proviral promoter (Fig. 1a). Nuc-0 is located immediately upstream of the modulatory region and nuc-1 immediately downstream of the viral transcription start site (TSS). The position of those nucleosomes in the 5’LTR appears to be an intrinsic property of the LTR. Indeed, the same positions were observed independently of the integration sites in different cell lines (Van Lint et al. 2013). Notably, during HIV-1 transcriptional activation, the organization of nuc-1 but not of others nucleosomes present on the HIV-1 genome is disrupted (Verdin et al. 1993). To our knowledge, such a precise nucleosome organization of the SIV promoter has not been described yet.
3 Regulation of HIV-1/SIV Transcription
Latency is established and maintained through multiples mechanisms acting in concert and operating mostly at the transcriptional level but also at several post-transcriptional steps. Regarding transcriptional regulation, HIV-1/SIV latency results in a complex and variable combination of multiple elements acting at the initiation and/or at the elongation phases of transcription. This heterogeneous and dynamic combination of transcriptional repression mechanisms impedes the synthesis of the viral trans-activating factor Tat, a viral protein indispensable for profound activation of HIV-1 and SIV transcription.
3.1 Nuclear Topography
Besides the organization of the genetic information itself, the cellular factors associated with transcription, replication, and genomic architecture are structured in sophisticated patterns within the nucleus (Lamond and Sleeman 2003). TFs, chromatin-associated proteins, and RNA-processing factors are confined to precise nuclear areas corresponding to distinctive tasks. In addition, transcription as well as replication occurs at spatially definite nuclear sites (Misteli 2007). Therefore, the nuclear topography of HIV-1/SIV integration may drastically influence its transcriptional level.
The HIV-1 pre-integration complex targets regions of chromatin that are close to the nuclear pore (NP). In contrast, it excludes the internal regions in the nucleus and the peripheral regions associated with the nuclear lamina (Marini et al. 2015). This integration near the NP corresponds to the first open chromatin regions that HIV-1 meets after its entrance into the nucleus (Marini et al. 2015).
The nuclear pore complex (NPC) interacts with specific chromosomal areas, called nucleoporin-associated regions, and contributes to the organization of the three-dimensional nuclear architecture (Capelson et al. 2010). Therefore, the NPC provides a chromatin topology and a nuclear environment favoring HIV-1 transcription. Indeed, the roles of the nucleoporins Tpr and Nup153 have been well demonstrated since the silencing of Tpr and Nup153 leads to a reduction of HIV-1 transcription in infected CD4+ T cells (Marini et al. 2015). Nup153 and Tpr play distinct but complementary roles in the HIV-1 integration process (Lelek et al. 2015). Nup153 is required for HIV-1 nuclear import. Tpr remodels chromatin regions proximal to NPC in a state encouraging HIV-1 transcription (Lelek et al. 2015). Those NPC components are consequently needed for proper HIV-1 genome integration. Following the nucleoporin knockdown, the structure of the chromatin is reformed. This remodeling leads to HIV-1 integration into nuclear regions that are less favorable to an efficient viral gene transcription (Wong et al. 2015).
Transcriptionally silenced but replication-competent HIV-1 proviruses might therefore reside in areas refractory to viral transcription, for instance, in close proximity to promyelocytic leukemia nuclear bodies (PML NB). HIV-1 gene expression inhibition in those nuclear regions is dependent on epigenetic mechanisms. Proviruses typically exhibit transcriptionally inactive heterochromatic marks such G9a-mediated H3K9 dimethylation (Lusic et al. 2013). Other studies suggest that PMLs impede transcriptional elongation through the sequestration of cyclin T1, a subunit of the transcription elongation factor P-TEFb (Marcello et al. 2003; Doucas et al. 1999). The HIV-1 nuclear topography is therefore directly linked to various mechanisms implicated in HIV-1 transcriptional regulation.
3.2 Viral Integration Site
After reverse transcription in the cytoplasm, both viral and cellular proteins associated with the viral cDNA form the pre-integration complex that next migrates into the nucleus. Upon nuclear entry, the viral DNA is integrated into chromatin (see above). This process of HIV-1 integration is a nonrandom process. Indeed, the cellular lens epithelium-derived growth factor (LEDGF/p75) that binds both cellular chromosomal DNA and HIV integrase directs integration preferentially to introns of actively transcribed genes (Wagner et al. 2014). Interestingly, in the absence of LEDGF/p75, integration is still not a random process. Residual integration is then largely facilitated by the Hepatoma-derived growth factor-related protein 2 (HRP-2), another unique cellular protein containing an integrase-binding domain (Schrijvers et al. 2012).
While nothing is known about the SIV nuclear topography, it has been shown that SIVmac integration is also predominant in introns of actively transcribed genes (Crise et al. 2005; Hematti et al. 2004). However, SIV integration sites were determined after in vitro infection of a human lymphoid cell line (Crise et al. 2005) or in rhesus monkeys transplanted for at least 6 months with autologous SIV-transduced CD34+ cells (Hematti et al. 2004). Therefore, additional analyses of the SIV integration sites during natural infection are needed to ensure that SIV and HIV-1 present similar integration site preferences.
Steric hindrance is a phenomenon that may occur when the provirus integrates downstream and in the same transcriptional orientation as the cellular host gene. The “read-through” RNA polymerase transcription from the upstream cellular promoter displaces key transcription factors from the HIV-1 promoter and prevents assembly of the pre-initiation complex on the viral promoter.
Enhancer trapping may occur when the enhancer located in the HIV-1 5’LTR is placed near the promoter of a cellular gene and acts on the transcriptional activity of this cellular promoter, thereby preventing the enhancer action on the viral promoter.
Promoter occlusion occurs when a provirus integrates into the opposite orientation compared to the host gene. This may lead to collisions between the RNA polymerase complexes elongating from the viral and cellular promoters, resulting in a premature termination of transcription from the weaker or from both promoters.
In this context, Chen et al. developed a method called barcoded where HIV ensembles to map the chromosomal locations of thousands of proviruses while tracking their transcriptional activities in an infected cell population (Chen et al. 2017). They showed that HIV-1 expression is strongest close to endogenous enhancers and that the insertion site also affects the response of latent proviruses to reactivation.
Undoubtedly, the insertion context of HIV-1 is a critical determinant of latency and viral response to reactivation therapies (see chapter “LATENCY REACTIVATION AS A STRATEGY TO CURE HIV”). The site of integration may also have another important effect on HIV-1 persistence: it may impact the survival of latently infected cells if HIV-1 integration occurs into genes associated with cell cycle regulation, such as MKL2 or BACH2. In this precise situation, HIV-1 integration thus confers a survival advantage that allows these cells to proliferate and expand despite a potent and suppressive ART (Wagner et al. 2014), and hence propagating HIV-1 without viral replication. The presence of such a clonal extension in the context of SIV infection has never been reported so far.
3.3 Epigenetic Modifications of HIV-1/SIV Promoter
In eukaryotic cells, DNA is wrapped around a nucleosome composed of a histone octamer. The histone tails are subject to multiple post-translational modifications including acetylation, phosphorylation, sumoylation, ubiquitination, and methylation. These reversible epigenetic marks modify gene expression by changing chromatin condensation which dictates the accessibility of DNA to TFs and transcription machineries. Epigenetic modifications are catalyzed by several chromatin-modifying enzymes such as histone acetyltransferases (HAT), histone deacetylases (HDAC), DNA methyltransferases (DNMT), or histone methyltransferases (HMT).
HATs and HDACs influence transcription by selectively acetylating or deacetylating the ϵ-amino group of lysine residues in histone tails, respectively. HATs favor chromatin opening and thus increase the accessibility of TFs to their binding sites. Moreover, histone acetylation marks enable the recruitment of bromodomain-containing proteins, such as chromatin remodeling complexes and TFs, which in turn regulate gene expression. In contrast, deacetylation by HDACs promotes a repressive heterochromatin environment (Yang et al. 2007). HDAC1, HDAC2, and HDAC3 are recruited by transcriptional repressors to the HIV-1 LTR5’ that typically displays deacetylated histones in latent condition. Therefore, deacetylation of the HIV-1 promoter chromatin by these enzymes plays a role in the establishment and maintenance of HIV-1 latency (Colin and Van Lint 2009). Indeed, treatment of infected cells with HDAC inhibitors allowing a global increase of histone acetylation and the remodeling of nuc-1 is coinciding with activation of HIV-1 gene expression (Verdin et al. 1993). Mechanistically, in microglial cells which constitute an important reservoir in the brain (Alexaki et al. 2008), the corepressor CTIP2 (COUP-TF Interacting Protein 2) acts as a recruitment platform for HDAC1 and HDAC2 on the HIV-1 promoter, leading to a heterochromatin environment [reviewed in (Le Douce et al. 2014)].
While histone hypoacetylation is generally associated with transcriptional repression, histone methylation can be either associated with transcriptional repression or activation, depending on the site of modification. H3K9 trimethylation (H3K9me3) and H3K27 trimethylation (H3K27me3) are patterns associated with transcriptional repression and have been shown to be associated with HIV-1 transcriptional silencing in different postintegration latency models (Imai et al. 2010; Friedman et al. 2011).
The HMT enhancer of Zeste homolog 2 (EZH2) is required for H3K27me3. This HMT is present at high levels in the LTR region of silenced HIV-1 proviruses and is rapidly displaced following proviral reactivation (Friedman et al. 2011). EZH2 seems to be of particular importance since the knockdown of this enzyme strongly induced HIV-1 expression compared to the knockdown of SUV39H1, an HMT required for H3K9me3 that has been shown to be recruited by CTIP2 to the HIV-1 LTR in microglial cells (Nguyen et al. 2017). Notably, EZH2 interacts—within the context of the Polycomb repressive complexes 2 and 3 (PRC2/3)—with DNA methyltransferases (DNMTs) and associates itself with DNMT activity in vivo, highlighting a direct connection between two key epigenetic repression systems (Vire et al. 2006). Additionally, the euchromatic histone-lysine N-methyltransferase 2 (EHMT2 also called G9a) is implicated in the control of latency. While both EZH2 and EHMT2 are required to silence HIV-1 proviruses and are recruited to the LTR in latently infected Jurkat T cells, PRC2 is distinctive because it controls the major rate-limiting step restricting proviral reactivation. In contrast, in both the primary cell models and in cells isolated from HIV+-treated patients, PRC2 and EHMT2 are both required to establish and maintain HIV-1 latency (Nguyen et al. 2017). Boehm et al. further performed a systematic small hairpin RNA (shRNA) knockdown of cellular HMT (Boehm et al. 2017). They identified SET and MYND domain-containing protein 2 (SMYD2), a member of the SMYD family of methyltransferases, as a HIV-1 transcriptional repressor. SMYD2 has been previously shown to regulate transcription by methylating H3K36 and H3K4 (Brown et al. 2006; Abu-Farha et al. 2008). Boehm et al. further demonstrated the recruitment of SMYD2 to HIV-1 latent promoter and the presence of H4K20me1 at the 5’LTR. Interestingly, this epigenetic mark allows the recruitment of an MBT (malignant brain tumor) family member, L3MBTL1, a reader protein linked to PRC1 (Boehm et al. 2017).
In addition to histone modifications, DNA methylation at cytosines located in CpG dinucleotides also participates in HIV-1 transcriptional silencing. During latency, the HIV-1 promoter is hypermethylated at two CpG islands surrounding the HIV-1 transcription start site. Methylation of the promoter region is generally associated with gene silencing, either by directly blocking binding of transcription factors to their recognition sequences or indirectly through the recruitment of methyl-CpG-binding domain proteins (MBDs) which in turn interact with HMTs and with HDACs, leading to a repressive chromatin structure (Suzuki and Bird 2008). This link between DNA methylation and histone epigenetic marks is important for our understanding of the establishment of a latent infection (Blazkova et al. 2009; Kauder et al. 2009). In patients’ cells, DNA methylation of the HIV-1 promoter increases progressively during ART treatment, suggesting that this epigenetic mark could participate more to viral persistence than to latency establishment (Trejbalova et al. 2016). Indeed, Trejbalova et al. detected low levels of 5’LTR DNA methylation in resting CD4+ T cells of patients who were ART-treated for up to 3 years. But, after long-term suppressive ART, they observed an accumulation of 5’LTR DNA methylation in the latent reservoir (Trejbalova et al. 2016). The exact mechanism of this DNA methylation accumulation in the latent reservoir of HIV-1-infected individuals remains unclear but has a potential impact on HIV-1 reactivation from latency, one of the most explored cure strategies.
Regarding HIV-1 transcriptional regulation, it has been well demonstrated that a great number of epigenetic modifications participate in the establishment or the maintenance of HIV-1 latency [reviewed in (Van Lint et al. 2013)]. However, much less is known in the context of SIV infection (Fig. 2). It has been shown that acetylation of histone H4 is detected during active/acute SIV replication in the brain (Barber et al. 2006). In contrast, during the asymptomatic infection, when full-length viral transcripts become undetectable, acetylation of histone H4 is lost (Barber et al. 2006). This reduction of acetylation seems linked to the recruitment of LIP to the SIV LTR. Indeed, in contrast to LAP, LIP is unable to interact and recruit to the LTR histone acetyltransferases activity (Barber et al. 2006). In addition, administration of the HDAC inhibitor SAHA (vorinostat) induced an increase in viral expression in ex vivo culture of CD4+ T cells (Ling et al. 2014) or in virally suppressed macaques (Gama et al. 2017), respectively. These data reinforce the importance of histone acetylation and epigenetic modifications in the control of SIV expression and are consistent with data obtained in HIV-1 infection studies.
3.4 Regulation of HIV-1 Transcription by Tat/P-TEFb
The enzyme that transcribes messenger RNA (mRNA) from protein-encoding genes is the RNA polymerase II (Pol II). This protein executes a series of distinct steps: it binds to promoters, initiates RNA synthesis, and then pauses in early transcriptional elongation to allow capping of the neo-synthetized mRNA before resumption of the transcription. In the case of HIV-1 transcription, due to the presence of the nuc-1 nucleosome, the paused Pol II is not able to resume transcription directly, therefore leaving the nascent RNA TAR. Further signals are needed to elicit the transition from the paused Pol II to a productive elongation complex (Adelman and Lis 2012). The switch from promoter-proximal pausing to productive elongation is mediated by the couple Tat/P-TEFb, an essential elongation transcription factor constituted of two subunits: cyclin T1 (CycT1) and the cyclin-dependent kinase 9 (CDK9).
Resting CD4+ T cells are characterized by extremely low levels of cyclin T1 due to actions of specific miRNAs (see below section “POST-TRANSCRIPTIONAL REGULATION OF HIV-1 EXPRESSION”) and of the cellular factor NF-90, which blocks translation of CycT1 mRNA (Budhiraja et al. 2013; Chiang and Rice 2012; Chiang et al. 2012). Moreover, in the absence of Tat, the elongation block is reinforced by the sequestration of P-TEFb within the 7SK small nuclear ribonucleoprotein (snRNP) repressive complex including the 7SK snRNA, the hexamethylene bisacetamide inducible protein 1 (HEXIM1), the 5’methylphosphate capping enzyme (MePCE), and the La-related protein (LARP7), as well as the combined inhibition by the negative elongation factor (NELF) and the 5,6-Dichloro-1-β-d-ribofuranosylbenzimidazole (DRB) sensitivity-inducing factor (DSIF) (Fig. 2).
Upon cellular activation (stress signals) and when Tat is not produced yet, P-TEFb is released from the HEXIM-1/7SK snRNA complex, and associated with the BET bromodomain protein 4 (BRD4), thereby forming the active P-TEFb complex. P-TEFb is then recruited to the HIV-1 LTR via interactions of the BRD4 bromodomains with acetylated histones (Darcis et al. 2015).
Once Tat has been synthesized, on one hand, Tat competes with BRD4 for binding to P-TEFb and on the other hand, Tat is also able to directly disrupt the inactive P-TEFb and to form a stable complex with P-TEFb. Tat then recruits P-TEFb to the HIV-1 promoter through TAR and increases transcription elongation. Tat can also recruit, in addition to P-TEFb, other elongation factors (such as ELL2, AFF4, ENL, and AF9), thereby forming the superelongation complex (SEC). P-TEFb is thereby positioned to phosphorylate the C-terminal domain (CTD) of Pol II, resulting in efficient elongation of viral transcription.
Thus, P-TEFb is present in two forms: a free active form and a 7SK-associated inactive form in which the kinase activity of the CDK9 is repressed. The balance between these two forms controls the activity of P-TEFb and the subsequent viral transcriptional reactivation processes. More clearly, since HIV-1 gene expression critically depends on P-TEFb function, factors that contribute to P-TEFb inactivation will also favor the persistence of latently infected cells.
We have previously observed that the CTIP2 is a corepressor of HIV-1 transcription in microglial cells since it acts as a recruitment platform for HDAC and HMT (Marban et al. 2007). In a separate complex, CTIP2 also associates with the P-TEFb inactive complex and represses P-TEFb functions by inhibiting CDK9 activity (Cherrier et al. 2013). Knocking down CTIP2 increases Tat-dependent transcriptional activity of the HIV-1 promoter. In contrast, overexpression of CTIP2 increases the recruitment of the inactive P-TEFb complex to the HIV-1 core promoter (Cherrier et al. 2013). Interestingly, HMGA1 (High Mobility Group A1), a nonhistone chromatin protein, also participates in the recruitment of the CTIP2-repressed P-TEFb to the HIV-1 core promoter through interaction with the 7SKsnRNA. Thus, HMGA1 and CTIP2 cooperatively repress HIV-1 gene expression by a HMGA1-mediated recruitment of CTIP2-inactivated P-TEFb to the HIV-1 promoter (Eilebrecht et al. 2014).
Another pathway has recently been implicated in HIV-1 latency, or rather in latency reversal, through P-TEFb activity. Besnard et al. (Besnard et al. 2016) showed that knockdown of mammalian target of rapamycin (mTOR) complex subunits or pharmacological inhibition of mTOR activity suppresses reversal of latency in various HIV-1 latency models and HIV-infected patient cells; mTOR inhibitors suppress HIV-1 transcription both through the viral transactivator Tat and via Tat-independent mechanisms. This inhibition occurs at least in part via blocking the phosphorylation of CDK9 (Besnard et al. 2016), further supporting the essential function of the P-TEFb complex in HIV-1 transcription.
3.5 The Role of Cellular Transcription Factors
The NF-κB transcription factor plays a complex role during the replication of primate lentiviruses. On one hand, NF-κB is crucial for induction of efficient proviral gene expression. On the other hand, NF-κB activation also induces expression of genes involved in the innate immune response and the cellular antiviral response.
NF-κB is sequestered in the cytoplasm of unstimulated cells in an inactive form through its interaction with an inhibitory protein from the family of inhibitors of NF-κB (IκB). Following cellular activation, the phosphorylation of IκB by IKK (IκB kinase) leads to its ubiquitination and proteosomal degradation, allowing the translocation of NF-κB into the nucleus and the transcriptional trans-activation of NF-κB-dependent genes. Notably, NF-κB also stimulates HIV-1 transcriptional elongation by interacting with P-TEFb and directs the recruitment of a co-activator complex of HATs to the HIV-1 LTR (Barboric et al. 2001; Perkins et al. 1997).
HIV-1 and SIV have been shown to modulate NF-κB activation through their regulatory and accessory proteins. Indeed, Tat and Nef proteins are able to activate or enhance the NF-κB activation. Later, during the infection, the viral protein Vpu that is only expressed by HIV-1 and its simian precursors suppress NF-κB activation [reviewed in (Heusinger and Kirchhoff 2017)]. Notably, only the Nef proteins of these Vpu-containing viruses are unable to down-modulate the TCR–CD3 complex from the cell surface and render virally infected T cells hyper-responsive to stimulation and increase the induction of NF-κB and NF-AT (Fortin et al. 2004; Schindler et al. 2006). In contrast, efficient down-modulation of TCR–CD3 by the Nef proteins of most SIVs and HIV-2 blocks the responsiveness of CD4+ T cells to stimulation and is associated with low levels of NF-κB and NF-AT activation (Schindler et al. 2008; Khalid et al. 2012). Thus, regulation of T cell activation and the NF-κB and NF-AT pathways by the accessory viral proteins Vpu and Nef may also have implications in viral latency. The importance of the NF-κB pathway in HIV-1 and SIV latency is further demonstrated by the reactivation of viral production upon treatment with NF-κB inducers such as PKC agonists (Gama et al. 2017; Darcis et al. 2015).
Additionally, the capacity of HIV-1 to establish latent infection is partially controlled by a four-nucleotide AP-1 element just upstream of the NF-κB element in the HIV-1 5’LTR (Duverger et al. 2013). Indeed, deletion of this AP-1 site mostly deprived HIV-1 of its ability to establish latent infection. This observation supports the idea that HIV-1 latency is a transcription factor restriction phenomenon (Duverger et al. 2013). The lack of active forms of other key cellular transcription factors (such as NF-AT and STAT5) is another element involved in repression of initiation and elongation of viral transcription in resting CD4+ T cells [reviewed in (Van Lint et al. 2013)]. Abdel-Mohsen et al. also showed that human Galectin-9 (Gal-9) is a potent mediator of HIV-1 transcription and reactivation (Abdel-Mohsen et al. 2016). Recombinant Gal-9 potently reverses HIV-1 latency in vitro and ex vivo through the induction of several HIV-1 transcription factors (NF-κB, AP-1, and NF-AT) expression and the inhibition of several chromatin modification and remodeling factors (including HDAC1, 2, and 3, EZH2, DNMT1) gene expression (Abdel-Mohsen et al. 2016). This pathway well illustrates the impact of different but dynamically linked molecular mechanisms on HIV-1 latency.
4 Post-transcriptional Regulation of HIV-1 Expression
4.1 HIV-1 mRNA Processing and Latency
Transcriptional regulation, described in the previous section, takes place and is influenced by nuclear co-transcriptional processes, including pre-mRNA capping, splicing, and polyadenylation that occur mostly co-transcriptionally (Karn and Stoltzfus 2012). First, pre-mRNAs are capped at their 5’end by capping enzymes: RNA triphosphatase, RNA guanylyltransferase, and RNA (guanine-N7) methyltransferase. Further processing of viral pre-mRNA by the host splicing machinery produces various transcripts and therefore various proteins. Nascent transcripts indeed result in over 40 differently spliced mRNAs (Purcell and Martin 1993), which can be divided into three classes: (i) fully spliced RNA expressing Tat exon1+2, Rev and Nef; (ii) singly spliced RNA encoding Tat exon1, Vif, Vpu-Env, and Vpr; or (iii) unspliced RNA serving as genomic RNA or to produce the Gag and Gag-Pol precursors.
P-TEFb, in addition to its crucial role in transcriptional regulation, also links the co-transcriptional processes of pre-mRNA capping and alternative splicing to transcriptional elongation (Lenasi et al. 2011). P-TEFb therefore facilitates the generation and processing of protein-coding mRNA.
Before the transport to the cytoplasm, the processing of the nascent transcript is completed by polyadenylation. The 3’ processing and polyadenylation of pre-mRNAs involve recognition of the upstream AAUAAA and downstream GU-rich motifs surrounding the cleavage and poly(A) addition site. For this, host cellular proteins such as the cleavage/polyadenylation specificity factor (CPSF), cleavage stimulation factor (CstF), CF1m, CF2m, and poly(A) polymerase are required for endonucleolytic cleavage and polyadenylation of viral pre-mRNA.
The viral Rev protein is involved in the transport of unspliced and partially spliced mRNAs from the nucleus to the cytoplasm, following its interaction with the Rev-responsive element (RRE). Nuclear export occurs upon association of Rev with the nuclear export factor Exportin 1 (Crm-1) and translocation of the Rev/RNA complex to the cytoplasm where it is either translated or packaged into virions [reviewed in (Kula and Marcello 2012)]. Therefore, defects in viral RNA export, which could be due to insufficient levels of either Rev (Huang et al. 2007) or the HIV-1 RNA-binding factors Matrin 3 and PTB (polypyrimidine tract-binding protein)-associated factor PSF (Zolotukhin et al. 2003; Yedavalli and Jeang 2011; Kula et al. 2013) or inhibition of HIV-1 mRNA translation, are also implicated in HIV-1 latency.
4.2 Noncoding mRNAs and HIV-1 Latency
MicroRNAs (miRNAs) are short single-stranded noncoding RNAs of 19–25 nucleotides that mediate post-transcriptional gene silencing. In general, following RNA Pol II transcription, primary miRNA transcripts are sequentially processed via the nuclear RNases III Drosha and Dicer to generate mature miRNAs which interact with a complementary sequence in the 3’ untranslated region of target mRNAs by partial sequence matching, resulting in degradation of the mRNA and/or translational repression. The level of specific mRNA translation can therefore be modulated by miRNAs.
Interestingly, modifications of the miRNA profile have been observed in HIV-1 infected patients (Houzet et al. 2008; Witwer et al. 2012; Bignami et al. 2012). Mechanistically, Tat and Vpr are known to function as RNA silencing suppressors by modulating miRNA expression levels in infected cells (Qian et al. 2009; Coley et al. 2010).
Cellular or viral miRNAs can target either cellular or virally expressed mRNAs. For example, PCAF, a HAT that is involved in chromatin remodeling, is targeted by miR-17/92 and miR-20a, both of which are downregulated in HIV-1 infection (Hayes et al. 2011; Triboulet et al. 2007). Cyclin T1 is repressed by various miRNAs in resting CD4+ T cells (miR-27b, miR-29b, miR-150, and miR-223) (Chiang and Rice 2012; Sung and Rice 2009). Moreover, cellular miRNAs, miR-28, miR-125b, miR-150, miR-223, and miR-382, known to be upregulated in resting CD4+ T cells, recognize the 3’end of HIV-1 mRNAs (Huang et al. 2007) and thus participate in the repression of HIV-1 gene expression.
Several viral miRNAs (vmiRNAs) have also been identified, including TAR-derived miRNA-TAR5p/3p (Klase et al. 2007; Ouellet et al. 2008) and the Nef-derived miR-N367 (Omoto et al. 2004). In contrast, some miRNAs can also have a positive effect on HIV-1 expression, for instance, when targeting HDAC involved in both the regulation of NF-κB and Tat. Indeed, the acetylation of these key factors is needed to allow for their proper action (Darcis et al. 2015).
The role of long noncoding RNAs (lncRNAs) has recently been observed in gene expression regulation, from transcriptional initiation to protein translation and degradation. For instance, the 7SK RNA is a lncRNA involved in the regulation of active P-TEFb levels (see the previous section). Nuclear-enriched abundant transcript 1 (NEAT1) is another example of a lncRNA that is involved in HIV-1 gene expression. This lncRNA is associated with the pathway of HIV mRNA export dependent on Rev and other cellular cofactors (Zhang et al. 2013) and play a crucial role in the post-translational regulation of HIV-1 expression. In addition, expression levels of noncoding repressor of NF-AT (NRON), a lncRNA involved in the HIV-1 latency establishment by targeting Tat for degradation (Imam et al. 2015), was observed to be inversely correlated with levels of HIV mRNA in resting CD4+ T cells (Li et al. 2016).
5 Concluding Remarks
For several years now, intensive efforts have been made by the scientific community to better characterize the HIV-1 latent reservoir and to investigate the molecular mechanisms regulating latency in infected cells. Improved knowledge of these mechanisms of persistence has paved the way for innovative strategies to attempt to eradicate latent HIV-1 but have also highlighted hurdles that should be overcome to reach this goal. One of them is the heterogeneity of latency, resulting from the multiplicity of the molecular mechanisms of HIV-1 transcriptional repression.
Numerous advances in our understanding of viral transmission, pathogenesis, and latency can be attributed to the use of SIV and NHP models. Multiple studies investigated whether these models recapitulate known and newly discovered features of HIV persistence in humans, including molecular mechanisms underlying latency. By many aspects, SIV and NHP models well reflect HIV infection of the human body. However, molecular mechanisms underlying SIV latency have been barely studied. Therefore, it is not excluded that some differences exist between HIV-1 and SIV latency, imposing a prudent analysis of the data obtained from the SIV/NHP models.
This project has received funding from the Belgian Fund for Scientific Research (FRS-FNRS, Belgium), the European Union’s Horizon 2020 research and innovation programme (grant agreement N° 691119 EU4HIVCURE H2020-MSCA-RISE-2015), the ANRS (France Recherche Nord & Sud Sida-HIV Hépatites), the “Fondation Roi Baudouin”, the NEAT Program, the Walloon Region (the Excellence Program “Cibles” and the “Fond de maturation” program), the ARC program (ULB) and the Internationale Brachet Stiftung (IBS). BVD and SB are postdoctoral fellows (ARC program and PDR project from the FRS-FNRS, respectively). CVL is “Directeur de Recherches” of the FRS-FNRS (Belgium).
- Fortin JF, Barat C, Beausejour Y, Barbeau B, Tremblay MJ (2004) Hyper-responsiveness to stimulation of human immunodeficiency virus-infected CD4+ T cells requires Nef and Tat virus gene products and results from higher NFAT, NF-kappaB, and AP-1 induction. J Biol Chem 279:39520–39531CrossRefGoogle Scholar
- Kula A, Marcello A (2012) Dynamic post-transcriptional regulation of HIV-1 gene expression. Biology (Basel) 1:116–133Google Scholar
- Nguyen K, Das B, Dobrowolski C, Karn J (2017) Multiple histone lysine methyltransferases are required for the establishment and maintenance of HIV-1 latency. MBio 8Google Scholar
- Ravimohan S, Gama L, Barber SA, Clements JE (2010) Regulation of SIV mac 239 basal long terminal repeat activity and viral replication in macrophages: functional roles of two CCAAT/enhancer-binding protein beta sites in activation and interferon beta-mediated suppression. J Biol Chem 285:2258–2273CrossRefGoogle Scholar
- Wong RW, Mamede JI, Hope TJ (2015) The impact of nucleoporin mediated chromatin localization and nuclear architecture on HIV integration site selection. J VirolGoogle Scholar