The UL133-138 locus
Work from a variety of groups have revealed that the UL133-138 locus is quite important to the regulation of latency and reactivation (reviewed in [
99]). UL138 is required for the establishment of latency, and in fact, abrogation of this open reading frame (ORF) results in an infection that favors lytic replication, leading to increased virion production [
100‐
102]. More recently, Buehler et al. showed UL138 regulates EGFR signaling, which downstream, upregulates PI3K/Akt signaling [
80,
103]. Pharmacological inhibition of PI3K/Akt favors reactivation, though this phenotype is most significant when combined with reactivation stimulating cytokines, suggesting other factors regulate this pathway [
80]. These data also highlight that viral manipulation of host cell signaling is likely not as simple as a direct on/off state, but instead reflects a mechanism that is more fine-tuned. In fact, UL138 actually regulates its own expression during HPC latent infection, which it accomplishes through the upregulation of the EGFR-regulated transcription factor, EGR-1. This host protein binds the viral genome, thereby driving
UL138 transcription in HPCs, as well as fibroblasts [
80], in turn creating a feedback loop regulating UL138-mediated events. Supporting the role of this feedback loop towards latency, disruption of the EGR-1 binding site upstream of UL138 in the viral genome results in the inability for the virus to establish/maintain latency in HPCs [
80].
The importance of the UL133-138 locus does not end with viral latency. While UL138 helps maintain latency, UL135 is critical for efficient reactivation. When given the proper reactivation cues, UL135 counters UL138-mediated functions by targeting EGFR [
80]. Indeed, disruption of the UL135 ORF results in a virus that fails to efficiently reactivate in CD34
+ HPCs [
104]. To ensure EGFR signaling and its downstream effectors are appropriately regulated, HCMV additionally encodes an miRNA, cmv-miR-US22, which targets EGR1 [
105]. UL135’s role during reactivation does not end with countering UL138’s functions. UL135 interacts with host adapter proteins, Abelson-interactor (Abi)-1 and Cbl-interaction protein (CIN) 85/ CD2 associated protein (CD2AP), which in turn regulate EGFR on the cell surface. Hence, in the absence of UL135 and its interactions with these host proteins, EGFR is increased on the cell surface of HPCs, thereby amplifying signaling and favoring latency. In line with this, inhibiting EGFR or its downstream pathways leads to reactivation when coupled with reactivation stimuli and rescues the reactivation defect observed for UL135 mutant viral infections [
80,
103]. Collectively, both UL135 and cmv-miR-US22 antagonize UL138-mediated EGFR regulation, thereby creating a cellular environment more amenable to reactivation. These data also illuminate the critical nature of EGFR signaling during latency, for which the virus has devised several means to regulate this pathway.
Manipulation of host cell signaling impacts MIE enhancer/promoter activity
A key step in establishing and maintaining HCMV latent infection is silencing of the MIE locus, which is likely initiated by chromatin remodeling (reviewed in [
29]). The MIE enhancer/promoter is thought of as the “lytic switch”, acting like what may more accurately be described as a rheostat to skew the infection towards one that is latent versus one that is lytic (reviewed in [
29]). Long regarded as a silencing factor of the MIE enhancer/promoter [
106,
107], Poole and colleagues recently confirmed the requirement for yin yang 1 (YY1) transcription factor binding for maintaining latency [
108]. Perhaps unsurprisingly, HCMV regulates YY1 by manipulating host cell signaling. Host-encoded transmembrane serine/threonine kinase, bone morphogenetic protein receptor 2 (BMPR2) signaling induces SMAD6, a SMAD family member that negatively regulates BMP and transforming growth factor β (TGFβ) signaling (reviewed in [
109]). In the context of latent infection, SMAD6 upregulation restricts the activity of TGFβ receptor (TGFβR) [
108]. This is critical, as latently infected induced pluripotent stem cells (iPSCs) or CD34
+ HPCs display a significant upregulation of TGFβ [
108,
110,
111], mediated at least in part by cmv-miR-US5-2 attenuation of NGFI-A binding protein 1 (NAB1) [
112]. Since NAB1 is a transcriptional repressor of EGR-1 [
113,
114], this represents an additional mechanism by which HCMV ensures EGR-1 transcription and subsequent TGFβ production. Additionally, cmv-miR-UL22A also targets TGFβ signaling, and in fact deletion of the pre-miR-UL22A sequence within the viral genome results in a viral mutant that is less efficient at reactivation [
110]. This is consistent with the finding that increased TGFβ signaling leads to an increase in the host-encoded miRNA, hsa-miR-29, which ultimately targets YY1. In turn, recruitment of YY1 to the MIE enhancer/promoter is decreased, which relieves the repression of the viral promoter and leads to reactivation [
108]. Collectively, these findings reveal not only the importance of TGFβ signaling to latency and reactivation, but the critical nature of this pathway towards regulating a central transcription factor that contributes to the balance between the active and repressive states of the MIE enhancer/promoter. This latter point is amplified by the multiple biological mechanisms HCMV has devised to regulate cell signaling that culminates at YY1 regulation.
A region rich in transcription factor binding sites, the MIE enhancer/promoter locus is studded with multiple binding sites for those which activate this very strong promoter region. Thus, just as the virus must evolve strategies to only recruit silencing transcription factors like YY1 during latency, HCMV has converse tactics to recruit transcription factors that activate the MIE enhancer/promoter as the virus reactivates. Investigators have shown that several of these transcription factors are regulated by viral manipulation of host cell signaling. For example, Keller an colleagues found transcription from the MIE locus was derepressed in quiescently infected NTera2-derived neuronal cells treated with forskolin, a compound that phosphorylates CREB [
115,
116]. Indeed, this was reliant upon the CRE-binding sites located within the MIE distal enhancer [
115]. More recently, Kew et al. showed phosphorylated CREB binding to the MIE enhancer/promoter aids in viral reactivation in DCs, which is dependent upon the activation of the ERK-MSK signaling axis. Consistent with this, deletion of the CREB binding sites in the MIE enhancer/promoter region results in a mutant virus unable to reactivate in DCs, though both CD14
+ monocytes and immature DCs maintained viral genomes. It is also important to point out that in this context, CREB not only acts as a canonical transcription factor, but it also promotes the phosphorylation of histone H3, which aids in chromatin remodeling of the MIEP, facilitating reactivation [
73]. More recently, we showed a parallel mechanism for regulating CREB activity and recruitment to the MIE enhance/promoter. Consistent with previous findings in other cell types (e.g. COS-7, fibroblasts; [
117,
118]), we found signaling via the viral GPCR, UL33, activates CREB [
63]. Furthermore, UL33-mediated signaling facilitates recruitment of phospho-CREB to the MIE locus during reactivation. Indeed, disruption of the entire UL33 ORF or UL33’s G-protein coupling motif (the ‘DRY’ motif) results in a failure to reactivate from latency following infection of CD34
+ HPCs, despite the ability of each mutant virus to maintain viral genomes. While phospho-CREB was recruited to the MIE locus in latently infected Kasumi-3 hematopoietic cells treated with 12-O-tetradecanoylphorbol-13-acetate (TPA) to induce reactivation [
63,
119], this was significantly reduced in parallel cultures infected with either UL33 mutant [
63]. Since cellular GPCRs coupled to Gα
o activate CREB via p38 MAPK [
120], it is plausible UL33 uses a similar mechanism. However, inhibition of p38 MAPK in monocytes had little impact on phospho-CREB binding to the MIE locus [
73]. Thus, additional work is needed to comprehensively understand the upstream mechanisms underlying CREB regulation.
NF-κB and AP-1 host transcription factors also function to activate the MIE enhancer/promoter (reviewed in [
29]). Latently infected Kasumi-3 hematopoietic cells stimulated with tumor necrosis factor (TNF) α to induce reactivation [
119] and treated simultaneously with curaxins to inhibit NF-κB, results in a significant decrease in
UL123 transcription, when compared to cultures treated with TNFα alone [
121]. Furthermore, the HCMV-encoded GPCR, US28 attenuates NF-κB during latent infection [
82], consistent with the requirement of US28 expression and signaling for viral latency, discussed in detail below [
25,
57,
62,
82,
122‐
126]. In fact, pharmacological inhibition of NF-κB in monocytes infected with a US28-deletion viral mutant resulted in an infection that favored latency rather than the lytic-like phenotype infection with this mutant usually observed [
82]. With four binding sites within the MIE enhancer/promoter (reviewed in [
29]), it is likely NF-κB’s role during reactivation is key. Further work elucidating the exact mechanisms by which US28, for example, modulates host signaling to regulate this important transcription factor is warranted and may reveal hematopoietic-specific signaling cascades critical for viral reactivation.
AP-1 is a heterodimeric transcription factor, comprised of c-fos and c-jun subunits [
127]. We have shown previously that both c-fos [
62] and c-jun [
128] are attenuated during latency, thereby limiting their heterodimerization. The balance of AP-1 binding to the MIE enhancer/promoter is key to its regulation; while the absence of AP-1 binding aids in keeping the MIE enhancer/promoter silenced [
62], its binding to the promoter proximal site is required for viral reactivation [
129]. Despite a requirement of this transcription factor for reactivation, however, AP-1 binding is dispensable for lytic replication in fibroblast or epithelial cells [
72,
129]. The upstream signaling events regulating fos and jun are currently under investigation, and while we have shown US28-induced signaling targets fos [
62], the viral and/or cellular factors manipulating jun are unknown. As discussed in more detail below, the signaling cascade US28 hijacks to attenuate c-fos remains to be elucidated, but it is likely that the virus balances the activation and attenuation of signaling cascades to skew the cellular milieu towards one that favors latency versus one that aids in reactivation. Thus, viral proteins, like US28, are likely antagonized to “switch” their functions during reactivation, similar to the relationship between UL138 and UL135.
A recent study detailed the involvement of Kruppel-associated box domain-associated protein (KAP)-1/ tripartite motif-containing (TRIM) 28 and mammalian target of rapamycin (mTOR) during latency and reactivation in CD34
+ HPCs [
130]. KAP-1 co-regulates transcription, as it recruits SET domain bifurcated (SETDB) 1 and HP1α, which facilitate H3K9me3. This histone modification is a repressive chromatin mark, and during HCMV latency, represses the MIE locus after SETDB1 and HP1α recruitment (reviewed in [
29]). As a result, these factors silence the MIE locus throughout latency. However, when mTOR is activated, it phosphorylates KAP-1, relieving chromatinization of the MIE locus, leading to activation of lytic gene transcription and the production of viral particles, suggesting a role for this pathway in both latency and reactivation [
130]. Supporting this, and as mentioned above, work from Buehler et al. reveal treatment with either an Akt or PI3K pharmacological inhibitor stimulates lytic replication in CD34
+ HPCs cultured under latent conditions [
80]. Additionally, we have shown HCMV stimulates mTOR activity 24 h post-latent infection of monocytes [
17,
18], though this activity was not sufficient to drive active replication [
18]. This could reflect differences in cell type specificity or cell environment at distinctive times during latent infection (e.g. early [24hpi] vs. later [7dpi] events). Alternatively, this supports the notion that mTOR is regulated in a rheostat-like fashion, where a threshold of activation has to be met or has not been reached. While the mechanism(s) through which this pathway is regulated remain unknown, rapamycin, an mTOR inhibitor, administered to transplant recipients suppresses viral reactivation [
131‐
134]. Whether this is due to a direct impact on the virus or the immune response is debated [
135], since rapamycin failed to impact
UL123 expression in LPS-stimulated DCs [
136]. MAPK and Akt signaling axes regulate downstream mTOR signaling, all of which are implicated in entry and maintenance of CMV in cells supporting latency [
137]. Akt is activated rapidly following latent infection of monocytes [
16,
18,
138] and CD34
+ HPCs [
80], though it is attenuated by 72hpi in monocytes [
16,
138] and minimally sustained in CD34
+ HPCs [
80]. Similarly, mTOR signaling is also rapidly upregulated within 24hpi of monocytes [
17,
18], which is attenuated during latency maintenance [
130]. Additionally, sustained pharmacological inhibition of Akt activity results in reactivation of WT, latent virus in CD34
+ HPCs [
80], suggesting completely abrogating Akt activity for prolonged times alters the cellular environment, such that it no longer is amenable to supporting HCMV latency.
Manipulation of MAPK signaling
The importance of MAPK signaling to HCMV latency and reactivation has become increasingly clear over the past decade. Several studies have shown MAPK signaling promotes HCMV reactivation in monocyte-derived DCs [
73,
77,
139]. However, this regulation is not binary; like Akt, low-levels of MEK and ERK phosphorylation are maintained during HCMV latent infection [
80], arguing activity of these MAPK proteins is fine-tuned. Additionally, such subtle differences may reflect tissue- or cell type-specificity. In support of this, MAPK activation promotes reactivation in a cell type specific manner [
139,
140], suggesting that not all cells that harbor CMV latently do so similarly (reviewed in [
141]). Indeed, IL-6 mediated activation of MAPK signaling facilitates viral reactivation in monocyte-derived-DCs and monocyte-derived-Langerhans-like cells (LCs), although viral reactivation was not coupled with activation of IL-6-mediated MAPK signaling in LCs [
77,
140], suggesting involvement of other viral or host factors. As mentioned above, SFKs, specifically Src and Hck, play important roles in this cell type-specific signaling (reviewed in [
141]). Both of these SFKs display upregulated expression during reactivation in monocyte-derived-DCs, and while MAPK activity impacts histone phosphorylation at the MIEP, chromatinization is regulated in parallel in an SFK-dependent fashion via the recruitment of MOZ histone acetyltransferase (HAT) [
77]. Upstream of SFK signaling are various receptors capable of potentiating signals, one of which is the receptor tyrosine kinase, Fms related receptor tyrosine kinase 3 receptor (FLT-3R), which downstream, regulates a cascades such as Ras and ERK/MAPK (reviewed in [
142]). Crawford et al. recently identified pUL7 as a novel, secreted ligand for the FLT-3R using HEK293T cells. This ligand-receptor interaction indeed leads to cellular signaling, and in bone marrow lymphoblast cells, PI3K/Akt and MAPK signaling cascades are activated. In turn, pUL7 induces differentiation of both HPCs and monocytes. Supporting these data, the investigators found that pUL7 required for reactivation [
143].Thus, pUL7 activation of MAPK signaling may represent another mechanism HCMV has devised to ensure viral reactivation. Whether Src and/or Hck aid in regulating the pUL7-FLT-3R cascade to impact downstream MAPK signaling remains unknown, but it is attractive to hypothesize these factors coordinate their functions to ensure proper MAPK activity during viral reactivation. Furthermore, such regulation may in fact be cell type dependent, underscoring the need to interrogate such pathways across hematopoietic cell model systems. To this point, the addition of MEK or ERK inhibitors in combination with reactivation stimuli significantly increases viral reactivation compared to reactivated CD34
+ HPCs in the absence of the inhibitors [
80]. These data reveal: 1) MEK/ERK inhibition alone is not sufficient to drive reactivation, 2) significant increases in MEK/ERK phosphorylation tip the balance towards reactivation, and 3) MAPK activity as it pertains to HCMV latency and reactivation may depend on cell type. Collectively, these data reveal MAPK signaling is a key pathway usurped by HCMV during latency and reactivation, and like Akt, is likely fine-tuned by the virus to skew the host cell environment to favor a specific phase of infection.
US28 has long been considered a latency-associated transcript, as early as 1998 when Patterson and colleagues showed this viral GPCR was detected in the peripheral blood mononuclear cells of infected individuals [
144]. Shortly thereafter, Beisser et al. were the first to demonstrate
US28 was transcribed during latent infection of THP-1 monocytic cells [
145]. Until recently, however, a role for US28 during this phase of infection had not been described. We were the first to demonstrate the requirement for US28 during viral latency [
57], a finding subsequently confirmed independently by several groups [
25,
82,
122‐
124,
126]. US28 is a potent signaling molecule [
146], thus it is unsurprising that US28-mediated signaling helps establish and maintain viral latency [
57,
62,
82,
147]. Incorporation of US28 into the mature viral particle [
57,
62] allows for its immediate expression, facilitating silencing of the MIE promoter/enhancer as early as 2 days post-infection of hematopoietic cells [
62]. To this end, US28 attenuates MAPK and NF-κB signaling [
82], as well as fos expression downstream [
62]. That US28 regulates the MAPK pathway is consistent with previous studies showing the upregulation of MAPK signaling promotes HCMV reactivation in monocyte-derived DCs [
73,
77,
139], detailed above. US28-mediated attenuation of MAPK signaling is also consistent with downstream suppression of fos, which ultimately prevents the AP-1 transcription factor from binding and activating the MIE promoter/enhancer [
62]. As detailed above, preventing recruitment of AP-1 to this promoter region is critical for successful latent infection, and conversely, its binding is essential for viral reactivation [
129]. As an active signaling protein expressed during latency, it is perhaps not surprising that US28 regulates host factors that ultimately impact the activity of the MIE enhancer/promoter, since this region is so crucial to balancing latency and reactivation. Recently, Elder et al. demonstrated US28 also regulates CCCTC-binding factor (CTCF) binding to the MIE enhancer/promoter. CTCF binding to this region increases during latency, thereby suppressing transcription from the MIE locus. Indeed, this is dependent upon US28-mediated signaling [
122]. Furthermore, these new data reveal that the neutrophil chemoattractants, S100A8 and S100A9, which are downregulated during latency [
148], are in fact regulated, at least in part, by US28-mediated recruitment of CTCF to their promoter [
122], revealing yet another means by which US28 manipulates cellular proteins, rendering the host cell more amenable to viral latency.
As much as we have learned collectively as a field, there are many outstanding questions surrounding US28’s function(s) during latency that remain. For the signaling pathways US28 regulates that are identified to-date, it is clear from that US28 alters the cellular milieu during latency in a ligand- [
57,
62,
82,
126,
147] and G protein-coupling-dependent fashion [
57,
62,
82,
126]. US28 binds a variety of cellular chemokines and couples to various G proteins (reviewed in [
56]), thus understanding the key, cellular proteins US28 usurps to its advantage will inform potential treatment strategies. Latently infected granulocyte macrophage progenitors (GMPs) display an increase in the expression of the CC chemokine, monocyte chemotactic protein-1 (MCP-1) [
149], a known US28 ligand (reviewed in [
56]). Pharmacological inhibition of G
α proteins with pertussis toxin or PI3K with wortmannin attenuated MCP-1 transcript levels [
149], suggesting this is regulated via GPCR-mediated signaling. Though the molecular mechanism(s) underpinning MCP-1’s regulation during latency is yet to be elucidated, it is certainly plausible the viral-encoded GPCRs expressed during latency [
57,
63,
145,
150,
151], such as US28, may leverage MCP-1’s upregulation to its own advantage, perhaps promoting dissemination in the initial stages of viral infection.
While the numerous host cell chemokines are obvious potential ligands for US28, it is equally plausible US28 interacts with a viral-encoded chemokine. HCMV encodes viral chemokines and cytokines (reviewed in [
152]), thus, this could represent a novel mechanism by which US28 regulates host cell signaling in hematopoietic cells, thereby retaining the virus in its latent state until given the proper cues. While US28 is not required for viral reactivation in hematopoietic cells [
25,
82,
122‐
124,
126], Crawford et al. published expression of the complete, functional US28 ORF has no impact on maintaining latency, but is required for viral reactivation in CD34
+ progenitor cells isolated from fetal liver. [
147]. This difference is possibly explained by tissue origin of the cells (fetal liver-derived vs. hematopoietic-derived). Nonetheless, since this viral GPCR is expressed throughout all stages of infection, some host or viral factor(s) likely influence US28 during reactivation in hematopoietic cells to either overcome its strong “pro-latent” signaling, or “switch” its signaling to favor lytic infection.
miRNA regulation of latency and reactivation
Both host cell- and viral-encoded miRNAs have functions during latency and reactivation (reviewed in [
153]). The known functions for cmv-US5-2, cmv-miR-UL22A, and cmv-miR-US22 are discussed above. Several other CMV-encoded miRNAs also regulate cell signaling pathways during latency and reactivation. For example, cmv-miR-US25-1 targets RhoA, and disruption of this viral-encoded miRNA increases the proliferation of CD34
+ HPCs [
154]. As part of the Rho family of GTPases, RhoA acts as a switch for a variety of signaling cascades as it cycles between its inactive GDP-bound and active GTP-bound states (reviewed in [
155]). How RhoA might be manipulated and coopted by viral signaling proteins and other factors during latency in hematopoietic cells is unclear, though Diggins et al. hypothesize a role for TGFβ signaling [
154], which regulates the RhoA pathway [
156‐
161]. If true, this would reveal yet another means by which HCMV attenuates the TGFβ cascade during latency. As discussed above, cmv-miR-UL22A targets SMAD3 to prevent robust TGFβ signaling during latent infection of CD34
+ HPCs [
110]. Thus, it is possible that cmv-miR-US25-1 and cmv-miR-UL22a function cooperatively to ensure this host cell signaling pathways is dampened during latent infection. Similarly, cmv-miR-UL148D, which is robustly expressed during viral latency, targets the activin signaling axis in monocytes, by directly suppressing the activin A receptor type (ACVR) 1B cellular receptor, which in turn limits the secretion of IL-6 [
162]. This represents a possible mechanism by which the virus subverts immune detection. Additionally, cmv-miR-UL148 targets the cellular immediate early response gene 5 (IER5). Repression of IER5 results in an increase in host-encoded cell division cycle 25B (CDC25B) expression, which aids in suppressing
UL123 transcription while simultaneously increasing cyclin-dependent kinase 1 (CDK1). Thus, cmv-miR-UL148-mediated regulation of the IER5-CDC25B axis is important for latent infection of Kasumi-3 and primary CD34
+ cells [
163]. cmv-miR-US5-1 and cmv-miR-UL112 also function to alter host cell signaling pathways during latency. Hancock and colleagues recently found these two viral-encoded miRNAs downregulate host cell Forkhead box O3a (FOXO3a). While both miRNAs protect CD34
+ HPCs from apoptosis [
164], whether their expression and targeting of FOXO3a is required for latency remains outstanding. FOXO3a binds and drives transcription from the MIE internal promoter 2 (iP2) [
165], a promoter that aids in reactivation of latent virus in primary CD34
+ HPCs [
166] and Kasumi-3 cells [
129]. Furthermore, mutation of the FOXO binding sites within the MIE promoter/enhancer locus leads to inefficient viral reactivation following stimulation of latently infected CD34
+ HPCs [
165]. Thus, it seems plausible that cmv-miR-US5-1 and cmv-miR-UL112 target FOXO3a to limit sufficient quantities of this protein, such that it cannot transactivate the MIE locus. cmv-miR-UL112 may indeed have dual functions in this regard, as Lau et al. showed this miRNA targets IE72 in both monocytes and THP-1 monocytic cells to aid in the maintenance of latency [
167], consistent with earlier work [
168]. HCMV also modulates host cell miRNAs to suppress the MIE-encoded proteins. Indeed, hsa-miR-200 family members target IE86. Mutation of the seed sequence in the
UL122 3’ untranslated region (UTR) results in a virus that fails to undergo latency in Kasumi-3 cells. Expression of this family of host-encoded miRNAs are upregulated in cells that favor latency (e.g. Kasumi-3, CD34
+, and monocyte cells) [
169], thus it is attractive to speculate that they may also target proteins involved in signal transduction networks. Collectively, both viral and host miRNAs are pivotal to latency and reactivation. Many of the changes these non-coding RNAs impart are small, yet significant. This again highlights that the regulation of factors involved in cellular signaling are indeed fine-tuned.