Introduction
Human immunodeficiency virus type 1 (HIV-1) is a single-stranded positive-sense RNA virus that replicates in CD4
+ human immune cells. Prerequisite for productive replication is the integration of the reverse transcribed viral DNA into the genome of infected target cells. Integrated HIV-1-derived double-stranded DNA is termed provirus and serves as a template for all viral-derived replication components [
1]. In the vast majority of cases, acute infection induces replication-associated cytopathic effects that lead to destruction of the target cells. A small percentage of infected cells however enter a state of latent infection. Such latent HIV-1 infection is characterized by a largely transcriptionally silenced provirus and by the absence of detectable mature viral gene products [
2].
Combination antiretroviral therapy (cART) has been developed as a highly efficient treatment to suppress HIV-1 replication at various stages of the viral replication cycle. However, because latency is a non-productive state of infection, latently infected cells escape targeting by cART. They thus contribute to the so-called HIV-1 reservoir, which is responsible for the observed viral rebound in patients upon interruption of antiretroviral treatment. Once infected, individuals can hence never fully clear the virus [
3]. This phenomenon has been termed HIV persistence and leads to a state of chronic HIV-1 infection. Around 20 million people worldwide are currently considered chronically HIV-1 infected [
4].
HIV-1 proviral DNA in chronic infection shows a number of characteristics. First, an estimated fraction of only around 3% of proviral sequences found in infected individuals on cART is intact and replication competent, as assayed by quantitative viral outgrowth assays [
5,
6]. Thus, the vast majority of proviral DNA in chronic infection consists of defective sequences [
5,
7‐
9]. These are thought to arise through the activity of host restriction factors, errors in reverse transcription or RNA splicing and recombination events post integration. While defective proviral sequences do not contribute to the reservoir hurdling HIV-1 cure, they could play a role in pathogenesis of chronic infection. Aberrant transcripts that often harbour translational competence have been reported from proviruses with various defects [
7,
9,
10]. These transcripts represent abnormal HIV-RNA species detectable in infected individuals on cART that result from transcription and alternative splicing of integrated proviral DNA with defects such as hypermutations, small internal deletions or mutated major splice donor sites. Although these aberrant transcripts do not support a replicative viral life cycle, they can be translated to viral proteins such as gag or give rise to novel chimeric HIV-protein species. These proteins have been shown to be capable of eliciting cytotoxic T-lymphocytic (CTL) responses and thus postulated to induce immune activation [
9,
10]. Chronic HIV-1 infection is therefore characterized by presence of a proviral-derived DNA burden integrated within the genomes of infected target cells. This burden is made up of a percentage of replication-competent intact proviruses as well as a large fraction of defective HIV-1-derived sequences.
The pathogenic impact of HIV-1 proviral-derived DNA burden is likely dependent on the regulation of the viral sequences. Thus, we need to gain a comprehensive understanding of (i) molecular pathways that regulate HIV-1-derived DNA and (ii) the mechanisms that influence genomic crosstalk between proviral DNA and the human target genome. This review will focus on the epigenetic aspects involved in these processes.
The term epigenetics has originally been introduced to describe heritable features in cell identity and physiology that are not linked to alteration in genetic sequence composition [
11]. Today, the field of epigenetics investigates a broad variety of mechanisms that collectively affect DNA-dependent activities [
12]. Of predominant interest have been epigenetic mechanisms that determine transcriptional state and activities of gene expression. Chromatin, i.e. the totality of DNA-associated proteins and RNAs, is considered the platform through which epigenetic regulation is exerted [
13,
14]. This regulation is driven through chemical modifications of DNA or chromatin components or alterations in chromatin composition, which lead to structural changes that influence accessibility of regulatory DNA regions to for example transcriptional regulators or effector proteins. Furthermore, these modifications often serve as recruitment platforms for binding of specific downstream effectors. The most studied and first described epigenetic chromatin features are covalent modifications of DNA bases and histone proteins. These include cytosine methylation of CpG dinucleotides (5mCpG) in DNA and post-translational modifications (PTMs) of mainly N-terminal histone residues, such as methylation (me), acetylation (ac) or phosphorylation (phospho) [
15,
16]. It has become evident that certain epigenetic features are usually associated with either a transcriptionally active and open or transcriptionally silent and condensed chromatin state, termed euchromatin or heterochromatin, respectively [
13,
14]. Unmethylated DNA, hyper-acetylated and hypo-methylated histones located on promoter
cis-regulatory elements are generally markers for euchromatin, whereas methylated DNA, hypo-acetylated and hyper-methylated histones are considered heterochromatin features. This is however a very simplistic representation, since the position of one and the same PTM within a histone can be associated with opposing activity states. For example, trimethylation of lysine 4 in histone 3 (H3K4me3) generally marks active promoter regions, while trimethylation of lysine 9 or trimethylation of lysine 27 in histone 3 (H3K9me3, H3K27me3) demarcates a transcriptionally repressed promoter. A large pool of epigenetic effector proteins has been described, which are responsible for catalysing or indeed removing specific modifications, such as DNA methyltransferases (DNMTs), histone methyltransferases (HMTs) or histone acetyltransferases (HATs) for example. Overall, it is today evident that epigenetic regulation is mediated by a dynamic and highly complex interplay of different epigenetic marks and pathways that involve DNA, chromatin components and in addition also higher-level features, such as for example 3D nuclear organization [
13,
14,
17,
18].
A multitude of studies focusing on epigenetic mechanisms in HIV-1 biology and in particular HIV-1 persistence was initiated following the observation that HIV-1 latency contributes to the establishment of a reservoir that prohibits viral elimination and thus a cure for HIV-1 infection. When reviewing findings from these and ongoing studies, two points should be noted. First, our current understanding of how proviral activity is epigenetically regulated is largely based on analysis of cellular HIV-1 infection models. This is due to the relative paucity, lack of molecular markers and heterogeneous identity of latently HIV-1 infected cells in individuals on cART, which largely prevents analysis of regulatory mechanisms in vivo or on patient-derived materials. Secondly, also as a result of this technical hurdle, work was so far mainly focused on understanding how epigenetic mechanisms influence integrated HIV-1 proviral DNA sequences. There are few studies to date that have investigated epigenetic effects of acute or latent HIV-1 infection on the host genome as a result of proviral/human DNA crosstalk.
Changes of the cellular epigenome as a result of HIV-1 infection
HIV-1 gene expression, as previously described, is actively controlled through usage of the epigenetic machinery of infected host cells. While much data has been gathered during the past decades to decipher details of this interaction, we are currently only beginning to understand converse effects of HIV-1 infection, i.e. how host cell epigenomes and the epigenomes of bystander immune cells are altered in response to HIV-1 infection.
There are a number of possibilities for how such alterations could be mediated, as examples of different viral infections have shown [
113]. First, viral proteins might inherently possess enzymatic activities of epigenetic modifiers. This has for example been demonstrated for vSET, a protein of
Paramecium bursaria chlorella virus (PBCV-1), which codes for a viral SET domain enzyme that catalyses deposition of the repressive H3K27me3 mark on host cell chromatin [
114]. Second, viral-encoded proteins could interact with epigenetic players of infected cells and thereby alter their activity. This results in a changed epigenomic profile of infected cells, which could, for example, promote viral replication, a state of viral latency or indeed alter the proliferative behaviour of targeted cells. Examples for such scenario are found during infection with human gamma herpesviruses Kaposi’s sarcoma-associated virus (KSHV) and Epstein-Barr virus (EBV) [
113]. The KSHV-encoded latency-associated nuclear antigen (LANA) interacts with SUV39H1 and DNMT3A to induce transcriptional repression of a range of host genes [
115,
116]. These changes have been proposed to induce epigenetic reprogramming of infected cells, which can be associated with transition towards a transformed phenotype and explain KSHV-associated tumour development [
113]. In a similar manner, various EBV-encoded nuclear antigen (EBNA) proteins have been shown to functionally interact with the cellular polycomb epigenetic repression complex to induce transcriptional downregulation of tumour suppressor genes in infected cells [
113,
117]. As a third scenario, cellular epigenetic profiles could be altered as an indirect consequence of viral infection. Innate cellular mechanisms sense viral infection and initiate signalling cascades that eventually result in epigenomic restructuring. In this case, epigenetic changes might not only be seen in infected target cells but bystander cells might equally be affected through infection-induced cytokine signalling and an altered microenvironment. Through this mechanism, chronic inflammation induced by hepatitis viruses HBV and HCV has been proposed to result in aberrant DNA methylation signatures in hepatocytes of infected livers [
118,
119].
In the case of HIV-1 infection, few studies have so far focused on epigenetic changes on the host genome and further mechanistic insights are yet to be uncovered. These studies primarily addressed the cellular methylome, i.e. alterations in the extent and pattern of 5mCpG in cellular genes in response to HIV-1 infection. Notably, the analysis was mainly undertaken in CD4
+ T cells or peripheral blood mononuclear cells (PBMCs) of HIV-1-infected individuals on ART, which included infected, but to a large extent also bystander, non-infected cells. At least in part, observed effects are thus likely indirect effects of HIV-1 infection. A pioneering report two decades ago also showed that HIV-1 infection results in increased DNMT activity and de novo methylation of a single CpG in the gamma interferon (IFN-γ) promoter [
120]. This provoked transcriptional downregulation of the cytokine as important player for different immune functions [
120]. The same team further showed that HIV-1 infection was associated with hypermethylation and reduced expression of p16
INK4A, a tumour suppressor gene [
121]. These findings brought about the idea that aberrant DNA methylation might be a conserved mechanism of HIV-1 pathogenesis. Indeed, more recently, the use of array-based genome-wide techniques for methylome analysis has revealed that blood cells of HIV-1-infected individuals on ART are epigenetically altered in a characteristic way, linking HIV-1 infection to premature ageing and abnormal immune responses [
122‐
124]. One study compared methylation patterns at over 26,000 genome-wide CpG sites validated as ageing markers and came to the conclusion that HIV-1 infected individuals on cART showed an average epigenetic ageing advancement of 4.9 years compared with healthy controls [
124]. The authors furthermore observed global deregulation of the methylome across over 80,000 CpG sites, which in addition to changes reminiscent of advanced age, also showed local abnormalities specific for HIV-1 infection. These include hypomethylation at the human leukocyte antigen (HLA) locus, which indeed could suggest epigenetic regulation in innate immune responses involved in HIV-1 infection control [
124]. Interestingly, it appears that antiretroviral therapy can alter observed methylome changes and thus might influence premature ageing as well as onset and progression of comorbidities in HIV-1 infected individuals [
123].
Although much remains to be analysed, these studies collectively support the early finding that HIV-1 infection profoundly alters the cellular methylome. Based on our understanding of epigenetic regulation as a complex interplay of different features, it would be astonishing if methylome changes would not be accompanied by genome-wide changes of further epigenetic marks. Indeed, there is preliminary data that HIV-1 infection also alters levels of several histone PTMs [
125,
126]. One study showed that HIV-1 infection ex vivo is accompanied by strong fluctuations in histone PTM levels as demonstrated by mass spectrometry and transcriptional profiling of PTM-associated enzymes [
126]. In accordance with the findings on increased CpG methylation upon infection, a second report suggested that global repressive histone marks, such as H3K9me3 and H3K27me3 increase [
125]. Notably, these studies focused on acute infection ex vivo—data on global histone PTM changes in chronically HIV-infected individuals has to our knowledge not yet been reported. This will likely change in the near future with current advances in sequencing technologies and the fast evolving protocols for genome-wide histone PTM analyses on low cell numbers in primary cells.
In conclusion, current pioneer works indicate that HIV-1 infection leads to long-term epigenetic reprogramming of target and bystander immune cells. While underlying molecular mechanisms remain to be uncovered, this reprogramming could play an important role not only in promoting HIV-1 persistence but also in the development of chronic HIV-1 disease and associated comorbidities.
Epigenetic targets in clinical approaches to HIV-1 disease
The realization that cART, although efficiently controlling viral replication, could not eliminate HIV-1 from infected individuals has initiated a now decade-long search for clinical strategies to achieve an HIV-1 cure. Efforts have mainly been focused on targeting the latent HIV-1 reservoir responsible for viral persistence. In addition, much work has been done to find approaches to strengthen immunological defences to HIV-1, such as for example through use of anti-HIV-1 broadly neutralizing antibodies (bNAbs) or chimeric antigen receptor (CAR) T cells targeted to HIV-1-infected cells [
127]. A second avenue of research has focused on counteracting the state of HIV-1 latency. This ‘shock and kill’ approach has been based on the idea that forced reversal of proviral transcriptional repression (‘shock’) would lead to fast depletion of viral reservoir cells, while ongoing cART would prevent new reservoir seeding and immunological defences would clear reactivated reservoir cells (‘kill’) [
1]. Much effort has therefore been spent on finding so-called latency reversing agents (LRAs), i.e. compounds that have the capacity to induce proviral transcription from silenced HIV-1 5′ LTR [
128].
The first LRA compounds tested were the powerful immune-activating interleukin-2 (IL-2) [
129] and anti-CD3 antibodies [
130], based on the observation that engagement of the T cell receptor consistently activated HIV-1 production in latently infected CD4
+ T cells [
35]. However, cART discontinuation after treatment with these antibodies resulted in rapid plasma viral rebound in all patients [
131]. Thanks to our improved understanding of molecular mechanisms underlying HIV-1 latency, a variety of different LRA classes has since been developed (reviewed in detail in [
128]). Of particular clinical interest are a range of so-called epi-LRAs, i.e. agents that reverse proviral latency through direct interference with epigenetic silencing mechanisms at the 5′ LTR [
35]. These include inhibitors of histone deacetylases (HDACi), e.g. Vorinostat and Panobinostat, inhibitors of histone methyltransferases (HMTi), e.g. Chaetocin, and inhibitors of DNA methylation (DNMTi), e.g. 5-aza-2′-deoxycytidine [
35]. Since epi-LRAs performed well in activation of latent HIV-1 ex vivo and importantly in a number of cases, these compounds have already been FDA-approved for use in clinical practice in the context of anti-cancer regimens, several trials have been undertaken to investigate their potential in purging the HIV-1 reservoir in chronically infected individuals. However, although transient HIV-1 production has been observed, no trials using individual LRA have so far succeeded in significantly reducing HIV-1 reservoir size [
132‐
134]. One possible reason for these findings is the growing evidence that HIV-reservoirs are of highly heterogeneous nature, not only regarding cellular identities but also concerning cellular activation state and tissue type-dependent microenvironment [
135‐
137]. Therefore, combination of epi-LRAs with LRAs targeting different cellular pathways, such as for example protein kinase C (PKC) agonists or positive elongation factor B (P-TEFb)-releasing agents will likely be necessary for optimal HIV-1 latency reversal in vivo [
128]. A possibly even greater challenge in the ‘shock and kill’ approach is the achievement of a sufficient ‘kill’ of targeted reservoir cells. It has become evident that intricate adjuvant immunotherapies will be required to eliminate newly activated cells and prevent proliferation of the reservoir [
138,
139]. These hurdles have currently somewhat halted the surge for ‘shock and kill’ in HIV cure research and have allowed for alternative concepts to be brought forward.
One such concept is the ‘block and lock’ strategy—an approach which promotes the idea of disarming HIV-1 reservoir cells through blocking HIV-1 transcriptional activity and locking the proviral promoter in a state of deep latency [
140]. This concept not only opposes the strategy of ‘shock and kill’, it furthermore also complies with a growing perception in the field that full elimination of HIV-1 might clinically not be achievable. Instead, therapeutic efforts should support a sustainable remission, i.e. a state in which HIV-1-infected individuals are able to control the viral burden without need for continuous cART. Since at its core, transcriptional activity of the HIV-1 provirus is regulated through epigenetic mechanisms, strategies for deep latency will need to target the proviral epigenetic landscape for long-term, heritable silencing. This has indeed been found for the most promising ‘block and lock’ agent so far reported, the Tat inhibitor didehydro-cortistatin A (dCA) [
141‐
144]. dCA is an analog of the natural steroidal alkaloid cortistatin A, which prevents Tat/TAR interaction and thus Tat-mediated transactivation of HIV-1 promoter through binding the TAR-binding domain of Tat. In cellular HIV-1 latency models, treatment with dCA promotes heterochromatinization of the HIV-1 5′ LTR, with an increase of deacetylated H3 at nuc-1 and the recruitment of repressive chromatin-modifying complexes to the HIV-1 promoter [
143]. In CD4
+, T cells of HIV-1 positive individuals dCA thus appear to induce a state of persistent latency—HIV-1 transcriptional activity is blocked and increasingly becomes refractory to reactivation by LRAs [
144]. This finding has been mimicked in a mouse model for HIV latency, where dCA treatment reduced tissue HIV-1 RNA and although viral rebound upon discontinuation of ART was still observed, rebound was delayed and quantitatively reduced [
141]. Clinical trials with dCA in humans have so far not been reported and several questions, including potential viral mutation-based drug resistances remain to be addressed [
145,
146]. Nevertheless, the findings on dCA show that inhibitors of Tat or indeed agents that mediate heterochromatinization of the proviral LTR might in future be important components of clinical approaches for sustainable remission in HIV-1 disease.
In this context, a second class of compounds, the so-called LEDGINs, should be noted. LEDGINs are small molecules that inhibit lens epithelial-derived growth factor (LEDGF)/p75 cofactor for HIV-1 proviral integration [
147]. LEDGINs bind dimers of HIV-1 integrase and inhibit the interaction between integrase and LEDGF/p75, which results in reduced catalytic integrase activity and relocation of residual proviral integrants out of transcription units, towards the inner nuclear component into chromatin regions increasingly associated with epigenetic marks of transcriptional silencing (e.g. H3K9me3, H3K27me3) [
148,
149]. This relocation propagates a latent proviral phenotype which shows reduced activation potential by LRAs. Interestingly, LEDGINs also appear to have an inhibitory effect on late events in the HIV-1 replication cycle: Viral particles produced in the presence of the inhibitors show aberrant integrase multimerization, which leads to an impaired infection potential at several levels [
148,
150]. As for dCA, future studies including trials in humans will be necessary to evaluate the potential of LEDGINs in the quest for HIV-1 remission.
Finally, observed changes in the genome-wide epigenetic profile of HIV-1-infected and bystander immune cells represent a so far unexplored but possibly interesting target for clinical approaches. This might, in particular, be the case for alterations in the cellular methylome, which have already been associated with a phenotype of ageing that might promote HIV-1 associated comorbidities [
124]. A more detailed mechanistic understanding of these alterations will, however, be required, before clinical strategies, similar to the use of epi-LRAs in the ‘shock and kill’ approach, can be followed. In general, epigenetic therapies will likely play a role in future innovative approaches to HIV-1 disease, but more work will be needed to circumvent drawbacks of current epigenetic drugs, such as for example toxicity effect due to the lack of specificity [
151].
Conclusions
The cellular epigenetic machinery plays an important role in chronic infection with HIV-1. On the one hand, epigenetic mechanisms are heavily involved in regulating transcriptional silencing of the proviral-derived DNA burden. This regulation is decisive particularly since transcriptional activity of intact proviral genomes in reservoir cells results in viremic rebound with grave clinical consequences. On the other hand, infection with HIV-1 also appears to change the epigenomic landscape of infected and bystander immune cells. This signature of infection, whether directly or indirectly linked to the proviral burden, could be an important cofactor in developing HIV-1 disease-associated morbidities. Many questions remain. It will in future be necessary to broaden epigenetic studies on HIV-1 disease increasingly to primary cells and tissues of affected individuals. Improvements also need to be done on mechanistic aspects of epigenetic crosstalk, in particular in understanding how infection leads to reprogramming of the human epigenome. Furthermore, it is to date unclear whether and how epigenetic mechanisms might play a role in observed phenomena of integration site recurrence and clonal proliferation of infected cells. In view of the potential clinical importance of these phenomena, in particular for the control of reservoir size and inducibility, this aspect certainly deserves future interrogations. Despite these open questions, our current understanding of the epigenetic regulation in chronic HIV-1 infection already holds a strong indication that pharmacological agents able to interfere and modify these regulatory pathways are promising candidates in future clinical strategies for sustainable remission in HIV-1 infection.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.