Viral cis-regulatory elements as sensors of cellular states and environmental cues

Abstract

To withstand a hostile cellular environment and replicate, viruses must sense, interpret, and respond to many internal and external cues. Retroviruses and DNA viruses can intercept these cues impinging on host transcription factors via cis-regulatory elements (CREs) within viral genomes, allowing them to sense and coordinate context-specific responses to varied signals. Here, we explore the characteristics of viral CREs, the classes of signals and host transcription factors that regulate them, and how this informs outcomes of viral replication, immune evasion, and latency. We propose that viral CREs constitute central hubs for signal integration from multiple pathways, and that sequence variation between viral isolates can rapidly rewire sensing mechanisms, contributing to the variability observed in patient outcomes.

Viral sensing strategies

Viruses are obligate pathogens that need viable cells to complete their replication cycle (Figure 1). To leverage the cell machinery, evade antiviral mechanisms, and initiate viral replication, viruses must recognize a variety of cell states and environmental cues [1–3]. Viruses have evolved a variety of sensing strategies according to their specific replication cycles and cellular niches, including those mediated by viral proteins that sense activation states or that interact with host proteins involved in cell cycle, cell death, and stress. Given their small genomes, viruses cannot encode for the diversity of protein sensors needed to detect the myriad cell states they could encounter. One of the main strategies to address this issue involves the sensing of host transcription factors (TFs) by viral cis-regulatory elements (CREs). Eukaryotic cells have a complex system of signal transduction cascades to detect changes in the extracellular and intracellular environment that ultimately impinge on different sets of TFs depending on the trigger signal. In the host, these TFs regulate the activity of promoters and enhancers leading to the upregulation or downregulation of gene expression programs [4]. Viruses can intercept and exploit these host sensing mechanisms by binding different TFs to their own viral CREs (Figure 2). This type of sensing mechanism benefits from the compactness of TF binding sites, the rapid evolutionary gain and loss of binding sites in viral genomes, the complex repertoire of TFs that respond to multiple signals and states, and the direct link between sensing and transcriptional regulation of viral effector proteins [5,6].

Figure 1: Viral replication cycle and transcriptional control.

Schematic of a representative replication cycle for retroviruses and dsDNA viruses. Following initial infection, retroviruses and most dsDNA viruses localize to the nucleus. Here, these viruses may transition to latency through episome formation (e.g. herpesviruses) or via integration into the host genome (e.g. retroviruses and non-enveloped dsDNA viruses). Internal or external cues can promote reactivation from latency or direct entry to the lytic cycle, marked by activation of viral transcription followed by assembly and release of new virions.

Figure 2: Pathways sensed by viral CREs.

Viral CREs sense immune activation, stress, metabolic states, as well as cell proliferation and differentiation by recruiting host TFs downstream of these pathways.

In a collective effort spanning more than four decades of research, the field has identified hundreds of CREs from retroviruses and double-stranded DNA viruses with many studies describing the signal inputs and regulatory mechanisms by which these CREs function. Key contributions in our overall understanding of viral gene regulation include the impact of chromatin states on viral CRE usage [7–11] as well as the downstream mechanisms of viral gene expression, including RNA Pol II pause release and elongation [12,13]. Given the breadth of this field, this review cannot discuss the outstanding work produced by hundreds of laboratories. Instead, we aim to provide a broad and synthesized overview of the current knowledge of viral CREs focusing on their function as sensors of the cellular environment. We describe the main signals sensed and the TFs involved, how viral CREs act as signal integration hubs, and how they can rapidly evolve to acquire or rewire sensing strategies. Finally, we discuss the potential targeting of viral CREs for therapeutic purposes.

Overview of viral CREs

Retroviruses and most dsDNA viruses express their genes using the host transcriptional machinery within the nucleus of infected cells [14,15]. This process is controlled by CREs within the viral genome that activate or repress the transcription of viral genes depending on the stages in the replication cycle as well as states of the infected cells. Given the compactness of viral genomes, most of the identified CREs act as promoter elements; however, some CREs have been shown to act distally as enhancers both upstream or downstream of transcription start sites [16–18]. For example, the origin of latent replication (oriP) region in EBV is known to act as an enhancer to the LMP1 gene that is 10 kbp upstream of the oriP [16].

Viral CREs have been identified and characterized using an array of approaches that determine their location within the viral genome, measure their transcriptional activity under different infection and cellular conditions, and identify the molecular mechanisms involved in their regulation (Box 1). Given the mechanistic parallels between viral and host gene expression, viral CREs often share sequence and functional similarities with host CREs. Indeed, many of the seminal studies in the 1980s on mammalian transcription have used viral CREs, such as the SV40 and BKV enhancer elements and gene promoters from different viruses, as models for CRE structure and regulation processes [19–25]. Similar to host cell CREs, viral CREs also utilize TFs to remodel the chromatin avoiding host repression, recruit RNA Polymerase II, and promote transcriptional elongation to regulate the expression of the viral genes [10,11]. The hijacking of host TFs enables the virus to transcribe viral genes immediately after infection, which are necessary to produce viral proteins such as viral transcriptional regulators (vTRs) that further directly or indirectly impact the expression of additional viral genes [26]. Ultimately, the recruitment of specific sets of TFs and vTRs onto the viral CREs determines the appropriate timing and cellular states in which each viral gene is expressed, leading to productive replication cycles and evasion of antiviral responses.

Box 1. Identification of viral CREs:

Prediction of viral CREs has been performed based on the location of viral coding sequences and by mapping transcription start sites (TSSs) using 5’RACE for individual genes or genome-wide using Cap Analysis of Gene Expression (CAGE), a high-throughput approach to map and quantify TSS activity at single base resolution. This unbiased mapping of TSSs is particularly important for large dsDNA viruses which encode for multiple coding and noncoding genes and that have complex patterns of regulation depending on the stage in the viral replication cycle [36]. To determine whether the identified genomic regions are sufficient to drive transcription, reporter assays are commonly used. Bashing experiments, evaluating either progressive deletions, internal deletions, mutations, or nucleotide replacements, have been instrumental to determine the bounds of CREs and to identify elements within the CREs that activate or repress expression.

Identification of host TFs regulating viral CREs:

To determine the TFs that bind and activate or repress a viral CRE, an array of DNA binding and functional assays is generally used. Direct binding is often identified using electrophoretic mobility shift assays using purified TF or nuclear extracts followed by supershift assays, DNase footprinting assays to identify regions within a CRE that are protected from DNase cleavage using nuclear extracts or purified TFs, or yeast one-hybrid assays, a method that can evaluate in parallel the binding of hundreds of TFs to a DNA element of interest. The functional role of these TFs is generally determined using reporter assays where TF binding sites are mutated or by knocking down TF expression. These studies in non-native contexts are often complemented by chromatin immunoprecipitation and knock down of specific TFs in infected cells to identify the cellular and infection stage contexts in which the TF binds/regulates the viral CRE.

Figure I: Methods to identify and characterize viral CREs.

(A) Viral CREs are often predicted based on their location upstream of protein coding sequences, using 5’RACE or CAGE to identify transcription start sites, and using CRE bashing reporter assay experiments.

(B) Binding of TFs to viral CREs can be determined using yeast one-hybrid assays, ChIP or ChIP-seq, DNase footprinting in the presence of a TF or nuclear extract, and using electrophoretic mobility shift assay (EMSA). The regulatory effect of these TFs is often measured in reporter or functional assays overexpressing or knocking down the TF or mutating TF binding sites.

The number of CREs across viruses often scales with the number of viral genes and the regulatory complexity; however, gene-to-CRE relationships are often not one-to-one [27]. Small viruses, such as polyomavirus and retroviruses, have one major CRE region, which is generally multifunctional and controls the transcription of the entire viral genome. For example, the long terminal repeats (LTRs) of the human immunodeficiency virus (HIV) and human T-lymphotropic virus (HTLV) are involved in the activation and silencing of viral transcription controlling both replication and latency [11,28,29]. In these retroviruses, different genes are produced through alternative splicing of this primary transcript rather than through different CREs [30]. The non-coding control region (NCCR) of the circular polyomavirus also has a dual role, in this case acting as a bidirectional promoter driving the expression of early genes when transcribing in one direction and late genes in the other, depending on the TFs recruited to this CRE [31]. For most other dsDNA viruses, the expression of viral genes is often achieved through independent CREs. For example, human papillomaviruses control the expression of early and late genes through at least two separate promoters [32–34]. Larger dsDNA viruses, such as adenovirus and herpesviruses, have a much more complex CRE landscape, containing tens to hundreds of CREs to control the convoluted cascade of viral gene expression required for different stages in viral replication and latency. For instance, Adenovirus 5 with ~35 kb is known to contain 13 CREs (ref), HSV-1 with ~150 kb contains at least 190 CREs [35], and EBV with ~170 kb contains about 322 CREs [36]. This leads to an estimate of one CRE per 0.5–5 kb depending on the virus, a similar or higher CRE density than the human genome [37] and more biased towards promoter rather than enhancer elements.

Viral CREs as sensors of cellular states

Viruses infect different cellular niches and need to sense and respond to diverse cellular and environmental cues. This is often achieved by recruiting to their CREs an array of TFs that are constitutive, lineage-specific, and those that respond to immune activation, stress, metabolic states, or cell proliferation:

Constitutive and lineage-specific TFs:

Viral replication often leads to high cellular stress and elicits immune responses which may cause apoptosis and early viral clearance. Certain cell differentiation states are less likely to trigger these events and may constitute permissive environments for viral replication. Viruses, therefore, often sense cell differentiation states through the binding of cell- and stage-specific TFs to their CREs to regulate the expression of latent and lytic genes (Figure 2). For example, this is achieved by recruiting XBP1, upregulated during B cell differentiation, to the EBV major lytic gene BZLF1 and the KHSV ORF50 gene, both of which lead to the active viral replication [38,39]. In addition to lineage-specific TFs, constitutive TFs such as SP1, SP3, and YY1 that contribute to maintaining cellular identity, also regulate gene expression in many retroviruses and dsDNA viruses [40,41]. The activity of these TFs is often controlled by interacting cofactors and post-translational modifications, which can be leveraged by many viruses that depend on these TFs for replication. For instance, SP1 dependent viruses such as HSV-1 and HCMV trigger SP1 phosphorylation soon after initial infection, which leads to lower SP1 activity and reduced expression of target HSV-1 Early and Immediately Early genes [42]. This primes the virus for latency and avoids immune detection, while expressing sufficient viral factors necessary for genome integration. YY1, instead, has a dual function in viral gene inhibition and activation controlled by post-translational modifications and interactions with cofactors, which may be present/active in certain cell-stages or conditions [43,44]. This illustrates how constitutive TFs can also lead to context-specific viral gene regulation.

Immune activation:

Sensing of host immune activation states is essential for regulation of entry and exit from latency in many viral species. Depending on the virus, immune activation can either stimulate or repress reactivation which likely reflects different strategies for immune evasion – exit from an immune-stimulated cell versus avoiding detection altogether. Many viral CREs are regulated by the same TFs activated during infection including AP-1, NF-κB, IRFs, NFAT, and STATs (Figure 2), allowing the virus to differentiate between immune signals and potentially coordinate specific evasion responses [45]. For example, AP-1 family members such as FOS, JUN, and ATFs activated by different immune and stress signals regulate the expression of many CREs from adenovirus, herpesvirus, polyomavirus, and retroviruses and can be involved both in latency and reactivation [45]. Similarly, NF-κB, a TF complex that responds to signals transduced via TNF- and Toll-like receptors as well as nucleic acid sensing pathways, is also recruited to many viral CREs. Binding of NF-κB to CREs can have different regulatory outcomes, ranging from promoting to preventing reactivation in polyomavirus and KSHV, respectively [46,47]. This sensing of the upstream and downstream antiviral response pathways may afford viruses the ability to more precisely time their own gene expression to counteract host immunity during exit from latency.

Cellular stress:

Cellular stress often triggers viral reactivation as a way to avoid hostile cell conditions and enable viral dispersal (Figure 2). Indeed, reactivation is induced by hypoxic stress in KSHV and EBV [48,49], fever and UV light in HSV-1 [50], and DNA damage in HCMV [51]. TFs that act downstream of oxidative stress pathways, such as HIF1A and EPAS1, activate CREs of KSHV genes [48,52], including the major regulator of KSHV latency LANA [53], as well RTA and ORF34 both of which are associated with latency switching and viral production [54]. Stress response factors such as MYC and TP53 have overall repressive effects on viral CREs, including the LCR of human papillomavirus (HPV) 16 and 18, the NCRR of BK Polyomavirus, and the X gene promoter of Hepatitis B virus (HBV) [55–57]. In many of these cases, early viral genes involved in reactivation (i.e., HPV E6, Polyomavirus large T-antigen, and HBV X protein) inhibit TP53 function through a variety of mechanisms including degradation and transcriptional repression [55–57]. This suggests that TP53 binding to viral CREs may act as a regulatory switch which, upon repression/degradation by viral factors, leads to full-scale viral reactivation. Some viruses, such as HSV-1 and HPV, sense hormone-mediated stress, for example by binding the glucocorticoid receptor NR3C1 to CREs of immediate early or early genes, promoting viral reactivation [58–60]. Altogether, stress-related TFs can have both positive and negative effects on viral transcription, highlighting the specificity of stress signals sensed by viral CREs.

Metabolic states:

Cellular metabolism is also an essential state that viruses leverage to control their own latency and replication (Figure 2). A hallmark example of this metabolic sensing involves HBV whose main niche are metabolically active liver hepatocytes. HBV can re-enter the lytic phase in response to metabolic stress signals sensed through CREs that recognize metabolic-related nuclear hormone receptors such as PPARA, RXRA, and NR1H4 [61]. In addition to nuclear hormone receptors, viral CREs also respond to multiple TFs that are modulated by metabolic signaling pathways such as Akt, TORC, and ERK, rather than by direct ligand binding, including CREB1 and FOXO1/3 [62,63]. These pathways sense nutrient availability, growth signals, and metabolic stress and impinge on CREs from HBV and multiple herpesviruses.

Cell cycle and proliferation:

Viruses require the host machinery and resources to replicate. These resources, such as nucleotides, amino acids, lipids, and biosynthetic enzymes often fluctuate during the cell cycle and are more readily available in proliferating cells[64]. In addition, replicating during cell division may favor viral production and allow viruses to evade immune sensing mechanisms such as cytosolic DNA sensing[65]. Therefore, among other mechanisms, viruses can sense and initiate viral reactivation by binding cell cycle regulated TFs such as E2F, MYC, TP53, EGR1, and TEADs to their CREs (Figure 2).

Viral CREs integrate cellular and viral signals

Viral gene expression is a muti-step process that, depending on the virus and gene, can involve chromatin remodeling, DNA demethylation, and Pol II recruitment, pause-release and productive elongation (Figure 3A) [8–14]. TFs and vTRs are at the center of all these processes and mediate the integration of different signaling pathways into a finely-tuned transcriptional output. Therefore, viral CREs, similar to many host CREs regulating immune, stress, and cell differentiation genes, function as signal integration hubs by binding TFs that sense different signaling cascades and cellular states [45]. This is achieved through multiple mechanisms that include: DNA binding cooperativity or antagonism between different TFs (and vTRs), cooperative recruitment of cofactors, and recruitment of different cofactors with synergistic or antagonistic transcriptional effects (Figure 3B). Collectively, these processes ensure that the regulation of viral gene expression depends on simultaneous activation or inhibition signals. This not only provides specificity but also leads to failsafe mechanisms that prevent unintended viral gene activation, safeguarding against premature immune detection, unproductive replication, or triggering cell death.

Figure 3: Viral CREs as signal integration hubs.

(A) Consequences of TF and vTR recruitment to viral CREs, including remodeling of local chromatin structure, modification of histone marks or DNA methylation, and alteration of RNA Pol II recruitment, pause-release, and elongation.

(B) Mechanisms for signal integration including cooperative or antagonistic recruitment of TFs to viral CREs, co-recruitment of cofactor by different TFs, and functional synergism or antagonism between cofactors.

Multiple studies have reported cooperative and antagonistic TF binding to viral CREs. Cooperativity is often associated with heterodimeric TFs but also to TFs that do not necessarily interact but facilitate each other’s recruitment [66,67]. This cooperativity can lead to robust viral gene expression when independent signals are present, diminishing the impact of noise in signaling pathway spurious activation. Cooperativity has also been extensively reported between host TFs and vTRs which enables coordination of host cell states with specific stages in viral latency or reactivation cascades [26]. In addition to cooperativity, a few examples of antagonism due to competition for binding sites or TF sequestration have been reported [68–70]. Synergism between TFs and signaling pathways can also occur by the cooperative recruitment of cofactors that remodel the chromatin, recruit RNA Polymerase II, or promote transcriptional elongation (Figure 3B). In addition to cooperative recruitment, TFs that bind a viral CRE may recruit different cofactors that may have additive or synergistic effects on target gene expression by acting on different steps in transcriptional initiation and elongation. Conversely, TFs binding to a CRE may have antagonistic effects if they recruit cofactors with opposing functions, such as a histone acetyltransferase and a histone deacetylase. Both synergistic and antagonistic recruitment of cofactors are observed in the HIV LTR. During reactivation, TFs including NFAT, NF-κB, and AP-1 family members recruit histone acetyl-transferases, increasing local chromatin accessibility and facilitating RNA Pol II recruitment and elongation mediated by P-TEFb [71–73]. On the other hand, SP1 and YY1 can recruit HDACs and methyltransferases such as SUV39H1 to maintain latency [74].

Evolution of viral CREs

The gain of new viral proteins is rare within a particular virus species and evolution of viral protein sequences to rewire sensing mechanisms often requires multiple nucleotide changes. On the contrary, sensing through viral CREs is highly evolvable as single nucleotide variants (SNVs) and indels frequently lead to gain or loss of TF binding sites given their short nucleotide lengths (Figure 4A). This is similar to what is observed with host CREs, where SNVs affecting a single TF binding site can cause severe developmental malformations or drive cancer [75–77], but supercharged by the high mutation rate and large population sizes of viruses. In the case of viral CREs, sequence variants can lead to sensing variability across isolates both within and across individuals, having a large impact on viral dispersal and the course of infection. This includes a variation in the number of binding sites for a TF within the population of viruses which may be responsible for differences in patient outcomes (Figure 4B). For example, variation in the number of NF-κB and NFAT binding sites across HIV-1 clades has been associated with differences in LTR transcriptional activities, in responses to the upstream signaling pathways, and in pathogenesis and transmissibility across isolates [78–81]. Similarly, variation in NFATC3 binding sites in BK polyomavirus NCCR has been associated with differences in the recruitment of RELA and AP-1, in transcriptional activity, in viral production, and in viral reactivation in renal transplant patients [82,83]. Further, viral CREs from closely related strains can not only differ in the number binding sites of a TF but also in their location within the CRE (Figure 4C). This provides an opportunity for novel signal integration outcomes as gain of binding sites at other locations in a CRE can create new cooperative or antagonistic relationships with other TFs and cofactors. Altogether, this leads to high plasticity and variation in viral CREs, which together with the large viral numbers within individuals and across the population and the strong selective pressures viruses face, leads to the selection of variants that can rapidly rewire sensing mechanisms.

Figure 4: Evolution of viral CREs.

(A) Schematic of TF binding variation between CREs from different isolates due to single nucleotide or indel variants.

(B) Scatter plot showing the average number of TF binding sites (TFBS) versus the standard deviation across isolates for TFs that bind a viral CRE.

(C) Viral dendrogram where isolates are colored based on the number TF2 binding sites in a CRE. The heatmap shows the fraction of isolates in each subclade that contains a binding site for TF2 at each position in the CRE.

Viral CREs as therapeutic targets

Latent infections have been a longstanding medical challenge, which often require lifetime treatment to avoid or manage recurrent infections. In HIV-1, which exhibits latent infection, the standard course of antiretroviral therapy is unable to eliminate replication competent provirus residing in CD4+ T cells [84–86]. Exploratory treatment strategies dubbed “shock-and-kill” have been developed to eliminate latently infected cells through reactivation and subsequent immune clearance of the virus using latency reversing agents (LRAs) [87]. LRAs include compounds which promote open chromatin such as HDAC, BRD4, or HMT inhibitors thereby derepressing latent HIV-1 [88]. Other LRAs involve PKC and TLR agonists which reactivate HIV-1 via stimulation of immune signaling pathways, given that the HIV LTR binds immune response TFs such as NF-κB and AP-1 [88,89]. While shock-and-kill can provide a means to target latent viruses, current methods prove challenging in practice as viral reactivation must be balanced with limited host tissue damage. An alternative approach, termed “block-and-lock” seeks to silence HIV-1 using latency promoting agents (LPAs) that insulate the LTR from activating cellular signals for prolonged drug-free remission [87]. However, the efficiency of both shock-and-kill and block-and-lock strategies is limited due to the intrinsic noise in LTR activation, the variation in HIV LTR sequence within and across patients, and the influence of integration sites in reactivation or silencing dynamics. Current research into implementing LRAs and LPAs previously aimed toward retroviral infections have since been adapted to treat latent herpesvirus infections such as EBV and HCMV [90,91]. A recently proposed method for treating latent HCMV involving HDAC inhibition leads to transient expression of lytic immediate early genes without reactivating the whole virus [92]. While overall these avenues of treatment offer promise, broad challenges such as toxicity and applicability across latent viral types remain unsolved. Further, the high evolvability of viral CREs may render shock-and-kill and block-and-lock strategies largely ineffective as pressures imposed by treatment may rapidly select for escape variants.

Targeting individual or select groups of viral CREs that control genes that do not lead to reactivation (e.g., late genes) may prove an effective strategy to signal the immune system for clearance without reactivating the entire virus and initiating lytic replication. Many TFs, such as nuclear hormone receptors, have specific ligands which activate or repress transcriptional effector functions upon DNA binding. Other TFs can also be either directly drugged, their function inhibited through signaling pathway perturbations, or selectively targeted for degradation [93]. Unlike existing methods targeting cofactors, targeting single TFs or small groups of TFs may reduce the overall likelihood of off-target effects. By leveraging the druggability of such TFs and signaling pathways, CREs regulating the expression of highly immunogenic viral proteins which are not involved in lytic replication (e.g., capsid, envelope, etc.) can be selectively targeted for a precision sensitizing rather than shock-and-kill approach. Alternatively, the disruption of negative regulatory loops can be effective to increase viral gene expression to cytotoxic levels, inhibiting viral replication and resulting in a high genetic barrier to resistance [94]. Another approach to prevent viral replication is to interfere with the typical order of events in the lytic signaling cascade, by targeting viral CREs to induce the expression of several lytic viral proteins out of signaling order, or by activating single lytic components known to be dependent on other viral proteins for their activity. Altogether, these precision sensitizing strategies may be able to circumvent some of the issues and limitations of shock-and-kill and block-and-lock approaches.

Concluding remarks

Ongoing research is shedding light into the complex interactions between viruses and hosts and how viruses sense, respond, and adapt to host cellular and organismal states. As discussed in this review, viral CREs are at the center of these sensing and effector mechanisms and constitute promising targets for antiviral development. However, many challenges remain in the path of identifying, characterizing, and targeting viral CREs for therapeutics (see Outstanding questions). High-throughput approaches such as massively parallel reporter assays or STARR-seq can be used for systematic identification of viral CREs and the cellular contexts in which they are active [95,96], whereas high-throughput protein-DNA binding assays may lead to the identification of the TFs and mechanisms involved in viral CRE regulation [97,98]. Leveraging this information for therapeutic targeting will likely prove more challenging given the shared regulatory strategies used by host and viral CREs and the rapid evolution of viral CREs within and across patients which could limit broad applicability and may lead to resistance. Viral CREs studies are also critical for the development of gene therapy as cryptic CREs present in viral vectors may result in off-target payload expression [99]. Future research must therefore prioritize not only the discovery and characterization of viral CREs but also the development of innovative strategies to circumvent the issues of sequence diversity and specificity, ensuring that therapeutic interventions can be both effective and broadly applicable.

Outstanding questions.

What is the full repertoire of CREs for each virus and in which viral stages are they active?

Which cellular and environmental signals impinge on different viral CREs and how are these signals integrated?

What is the impact of different environmental exposures (e.g., drugs, xenobiotics, metabolites, other infections) on viral CRE activity and reactivation?

How variable are viral CREs across isolates and how does this affect viral transmission and the course of infection?

Which CRE-targeting strategies would be most effective to treat latent viral infections?

Highlights.

• Viruses sense cellular states and environmental signals by recruiting host transcription factors to cis-regulatory elements (CREs) within viral genomes.

• Viral CREs act as signal integration hubs that impart specificity to viral gene regulation.

• Viral CREs are highly evolvable due to their high mutation rate and ability to rapidly change sensing mechanisms, affecting infection spread and patient outcomes.

• Latent viral infections can be treated by targeting viral CREs for reactivation or permanent silencing.