Retroviral genomes permanently integrate into the genome of their target cells as a provirus. The fate of the proviral genome becomes entwined with that of the infected cell. To this end, retroviruses have evolved a range of strategies to balance viral replication with the virulent and immunogenic effects of viral expression, enabling lifelong persistence in their host. Representing a deeply integrated form of retroviral persistence, endogenous retroviral (ERVs) genomes are those that have invaded the germline and become plentifully fixed in the host genome. ERVs have persisted throughout vertebrate evolution through interactions with sequence-specific Krüppel-associated box domain zinc finger proteins (KRAB-ZFPs)1. KRAB-ZFPs bind endogenous retroviral genomes and recruit repressive effector complexes to elicit tissue-specific silencing throughout development. These interactions have evolved through successive waves of endogenous retroviral genome invasion, driving the diversification and selection of new KRAB-ZFP genes in the host2. Whereas the infectious retrovirus HIV, which has only been present in humans since the early 20th century, relies on integration in transcriptionally silent regions of the host genome and spreading of heterochromatin for transcriptional latency and persistence in its host3. Although the ancient human deltaretrovirus HTLV-1 shares many similarities in tropism, genomic structure, long terminal repeat (LTR) promoter activity, and viral replication to HIV, viral integration site selection and host epigenetics alone fail to explain HTLV-1 transcriptional latency and persistence4.

Sugata and colleagues sought to identify the molecular features that underpin HTLV-1 transcriptional latency through analysis of provirus chromatin accessibility5. They identified an open chromatin region specific to HTLV-1+ cell lines and ATL patient cells. In luciferase assays, this open chromatin region exerted a silencing effect on the HTLV-1 5′ but not 3′ LTR promoter, suggesting it may act as an intragenic silencing element to regulate sense strand expression. The silencing element harboured binding sites for RUNX1 and GATA3 transcription factors. RUNX1 and GATA3 are critical regulators of haematopoiesis. They are expressed and involved in maintaining the cellular identity and function of HTLV-1’s primary host cell, CD4+ T cells6,7,8,9. RUNX1 and GATA3 both bind the intragenic silencing element of HTLV-1; however, in reporter assays, only RUNX1 enhanced the repressive effects of the silencer. These findings suggest that HTLV-1 has co-opted the critical hematopoietic regulator to dampen sense strand expression from the HTLV-1 provirus, abetting HTLV-1 latency and persistence. Indeed, perturbation of RUNX1 binding to the intragenic silencer resulted in increased expression of viral antigen Tax and increased susceptibility to cytotoxic T lymphocyte (CTL) responses.

The identification of RUNX1 as a trans-acting host factor that restricts HTLV-1 sense strand expression invokes a path towards identifying the mechanisms and upstream regulators of dynamic proviral transcription. HTLV-1 sense strand expression occurs in intermittent bursts in rapid response to cell intrinsic, extrinsic and metabolic signalling10,11,12,13,14. This model is evidenced by the presence of chronically activated HTLV-1-specific CTLs, which are indicative of sustained antigenic exposure15,16. It has been proposed that up to 3% of infected cells express Tax at a given time, leading to the selective expansion of infected cells17. This expansion is offset by an HTLV-I-specific CTL response targeting Tax-expressing cells. Tax expression is then rapidly silenced in surviving cells. While activating stimuli of sense stand transcriptional bursts have been described, the negative regulators that act to curtail such dynamic bursts remain unknown13,18. It is possible that RUNX1 cofactor binding may act as a molecular switch, governing dynamic on/off transcriptional bursts from the sense strand of the provirus (Fig. 1).

Fig. 1: Putative model for RUNX1 cofactor-mediated regulation of sense strand transcriptional bursts from the HTLV-1 provirus.
figure 1

In a steady state in vivo, sense strand expression from the 5′ LTR (yellow, green and blue boxes) is silent in most HTLV-1+ cells (top), and JUND-SP1 heterodimers drive hbz is expression from the 3′ LTR. CTCF binding the provirus insulates 5′ from 3′ LTR transcriptional regulation. Silencing of the 5′ LTR is mediated by repressive RUNX1 cofactors binding to the intragenic silencing element and enforced by cellular effectors, including the PRC1 complex and viral protein HBZ. Upon activating extracellular stimuli, TAX is expressed from the proviral sense strand in an intermittent burst (bottom). This is likely driven by altered stoichiometry and post-translational modification to RUNX1 cofactors, and interactions with both host and viral effectors of sense strand expression. Specific interactions between RUNX1 complex members and known proviral regulators are unknown. 5′ sense strand expression generally occurs in the absence of antisense expression from the 3′LTR; however, in the late stage of a sense strand transcriptional burst, expression of hbz is reinstated, acting as both a functional non-coding RNA and translated HBZ protein, which may co-ordinately help to reinitiate transcriptional latency from the 5′LTR. Select well-characterised positive and negative regulators of HTLV-1 sense strand expression are depicted in black and white. RUNX1 and cofactors are depicted in orange, blue, red and green. Image credit: the authors.

Full size image

To discern effectors of proviral expression that complex with RUNX1, Sugata et al. measured the association of candidate cofactors at the silencing element in both HTLV-1+ cell lines and ATL patient PBMCs. In all cell types, they found co-occupancy of RUNX1 with its core binding partner, CBFβ, which acts to stabilise RUNX1 binding to target sites. In patient PBMCs, the authors showed binding of histone deacetylase HDAC3 and the scaffold protein SIN3A, which present as plausible effectors of RUNX1-mediated transcriptional silencing. Among RUNX1 cofactors, ETS1 stands out as having features characteristic of a molecular switch. ETS1 is a key transcription factor in the development of immune cells, and is highly expressed in adult lymphoid tissues19. ETS1 binds the silencing region and counteracts RUNX1-mediated repression of the 5′ LTR in a dose-dependent manner. ETS1 cooperates with Tax to directly bind and activate Tax response elements in the 5′LTR20. Sugata and colleagues showed that the intragenic silencer and the 5′LTR physically associate when the RUNX1 binding site is intact, indicative of cross-talk between the regulatory elements. Together, these observations suggest that ETS1 binding stoichiometry and activity may act as a molecular switch between RUNX1-mediated repression to activation of the 5’LTR. ETS1 is both positively and negatively regulated by phosphorylation and sumoylation, which would enable rapid responsiveness to cell signalling pathways that converge on proviral expression21,22,23.

ETS1 is upregulated in response to hypoxia and regulates a class of hypoxia-inducible genes, providing a tenable mechanism for hypoxia-induced HTLV-1 transcriptional bursts18,24. More work is needed to understand the interplay between RUNX1 complex binding with characterised positive and negative regulators of HTLV-1 sense strand expression to tease apart the mechanisms of dynamic expression from the HTLV-1 provirus. A key consideration of future work will be to define the chromatin composition of the provirus in single cells. Work presented by Sugata et al. was performed in a heterogeneous population of cells, which would vary with regard to proviral sense and antisense expression. Binding dynamics and stoichiometry of the transcriptional effectors should be resolved in cells with defined sense and antisense expression profiles to elucidate the molecular underpinnings of dynamic proviral expression.

The intragenic silencing element appears to underlie distinct outcomes seen in HIV-1 and HTLV-1 infection. When introduced into the HIV genome, it dampened transcription and suppressed viral production in vitro, suggesting that such proviral-encoded silencers may broadly promote viral persistence. Sugata and colleagues explored this by examining open chromatin regions in cell lines harbouring the deltaretrovirus proviral genomes HTLV-2, simian T-lymphotropic virus 1 (STLV-1) or bovine leukaemia virus (BLV), but found no chromatin accessibility signatures resembling the HTLV-1 silencer. Authors note the possibility of underestimation by using infected cell lines as opposed to primary cells, which raises the tissue-specific regulation of proviral genomes. Early studies in Moloney Murine Leukaemia Virus (M-MuLV) attributed tissue-restricted repression to an intragenic domain that is bound by host factors expressed exclusively in early ectodermal stem cells25. In accordance, Sugata et al. found the HTLV-1 silencer acted strongly in T-cell lineages but not in B cells or other lineages. These analyses highlight the importance of the transcriptional landscape of infected cells in determining the activity of the intragenic silencer and ultimately on proviral activity. These findings raise two key considerations: (1) characterisation of cis-regulators of viral expression will need to consider infectious, resistant and latent cell reservoirs; and (2) that the expression of trans-acting repressors can shape infection outcomes. HTLV-1 binds ubiquitous cellular receptors and infects a range of cell types, yet drives the preferential expansion and transformation of CD4+ T cells. Does the T-cell-specific activity of the intragenic enhancer enable viral persistence in this reservoir? And do HTLV-1+ cells in other lineages rely on host chromatin and integration site for their survival, akin to HIV? Addressing these questions will reveal important aspects of the host-viral relationship that inform the latent reservoir in HTLV-1.

The intragenic silencing element is conserved across all HTLV-1 subtypes, indicating its emergence before subtype divergence tens of thousands of years ago. Despite HTLV-1’s limited capacity for propagation, it has persisted in endemic populations for millenia26,27. Sugata et al. suggest that the silencing element may enable HTLV-1’s long-term persistence by balancing propagation and expansion with the evasion of immune detection. They show that subtle changes in RUNX1 availability, binding, or proviral binding sites can markedly alter 5′ LTR-driven transcription in vitro. Could then changes in RUNX1 and cofactor availability and binding upregulate sense strand expression, and contribute to HTLV-1-associated disease? Supporting this, the RUNX1 cofactor ETS1 is overexpressed in ATL, promoting cell growth, migration, and adhesion28. To this end, authors examined HTLV-1 genomes from ATL and HTLV-1-associated myelopathy (HAM) patients for polymorphisms or deletions at RUNX1 sites. They found limited evidence of mutated or lost RUNX1 binding sites in ATL that could increase 5′ LTR expression. ATL-associated RUNX1 mutations also retained silencing capabilities in reporter assays. These findings suggest that loss of the silencer is not a major contributor to HTLV-1-associated disease, and shift the focus onto the activities of the 3′LTR negative-sense hbz transcripts in supporting pathogenesis. On the contrary, the increased replication and immunogenicity afforded by the impaired silencing function of the 5′LTR may lead to enhanced immune clearance of infected cells.

In vivo loss-of-function studies alongside analysis of larger clinical cohorts will be required to determine the effect of abrogating silencing element function on HTLV-1 pathogenicity and immunogenicity. These investigations will further inform the role of the silencer function on infection outcomes more broadly for both virus and host, including transcriptional latency, persistence, viral replication, host immune response, and viral transmission. Critically, such studies may reveal therapeutic vulnerabilities at various stages of HTLV-1 infection. At present, there are no preventative or curative therapies for HTLV-1-associated disease. Targeting RUNX1-mediated silencing of the provirus in asymptomatic carriers may enable a host-directed “shock and kill”29 approach to clearing the reservoir of expanded and pathogenic HTLV-1-infected CD4+ T-cells, and could represent a means of preventing HTLV-1-associated disease.