Glucocorticoid Receptor and Cell Cycle Regulator (E2F2) Cooperatively Transactivate a Cis-Regulatory Module in the HSV-1 Infected Cell Protein 0 (ICP0) Promoter

TL;DR

The paper shows how the glucocorticoid receptor and E2F2 work together to activate a specific region of the HSV-1 ICP0 promoter, potentially boosting viral replication.

Contribution

The novel finding is that GR and E2F2 cooperatively transactivate the ICP0 CRM-C, which was not previously known.

Findings

GR and E2F2, but not other E2F variants, cooperatively transactivate the ICP0 CRM-C promoter region.

Mutating the GRE or E2F binding sites in CRM-C reduces transactivation by GR and E2F2.

E2F2 binds to the CRM-C region during productive HSV-1 infection.

Abstract

Human alpha-herpesvirus 1 (HSV-1) acute infection culminates in life-long latency in sensory neurons in trigeminal ganglia and certain neurons in the central nervous system. Previously, E2F family members and glucocorticoid receptor (GR) were shown to stimulate HSV-1 and bovine herpesvirus 1 (BoHV-1) replication. Consequently, we hypothesized GR and E2F family members activate certain HSV-l promoters. To test this hypothesis, we determined if four HSV-1 ICP0 cis-regulatory modules (CRM) upstream of the ICP0 promoter were activated by E2F. GR and E2F2, but not E2F1, E2F3a, or E2F3b, cooperatively transactivate the ICP0 CRM-C, but not CRM-A, -B, or -D fragments upstream of a minimal promoter in a luciferase reporter construct. CRM-C sequences contain two E2F consensus binding sites, a GC-rich motif that E2F2 can bind, and a consensus ½ GR response element (GRE) adjacent to the consensus E2F #2 binding site. Mutating the ½ GRE or the 3 E2F binding sites significantly reduced GR- and E2F2-mediated transactivation. Chromatin immunoprecipitation studies revealed E2F2 occupied ICP0 CRM-C sequences during productive infection and mutating the E2F binding sites prevented E2F2 binding. These studies reveal GR and E2F2 transactivate ICP0-promoter activity, which may enhance viral replication in certain cell types.

Figures

Full text

1. Introduction

HSV-1 acute infection of craniofacial mucosal surfaces culminates in life-long latency in certain neurons located in the trigeminal ganglia (TG), brainstem, and other regions in the central nervous system [1,2,3]. In contrast to lytic infection of permissive cells, viral protein expression in neurons is not readily detected during latency because the genome exists as a silent heterochromatin [4,5,6]. Periodically, HSV-1 reactivates from latency, which increases virus transmission and recurrent disease. Recurrent HSV-1 ocular disease, including herpetic stromal keratitis [7,8], and HSV-induced encephalitis [9,10] are the direct result of reactivation from latency.

E2F family members regulate the mammalian cell cycle [11,12]. These regulatory proteins contain a well-conserved DNA-binding motif; certain family members interact with the tumor suppressor Retinoblastoma (Rb) protein and contain transcriptional activation domains [12]. Rb family members are phosphorylated by cyclin-dependent kinase-cyclin complexes, which release E2F family members. E2F1, E2F2, and E2F3a generally activate transcription whereas E2F3b and E2F4-E2F8 family members impair transcription or have little to no effect on transcription [13,14,15,16]. Consensus E2F binding sites are present in promoters of many genes that regulate cell-cycle progression [13,14]. Previous studies reveal E2F family members mediate replication and gene expression of BoHV-1 [17] and HSV-1 replication in cultured cells [18].

Stress (acute or chronic), UV light, and heat stress increase the incidence of reactivation from latency in humans and mouse models of infection [19,20,21,22,23,24,25,26,27]. Stress generally leads to glucocorticoid receptor (GR) activation, reviewed in [28]. For example, UV light induces GR phosphorylation and transcriptional activation via ligand-independent mechanisms [29,30]. Heat stress also increases cortisol and activates GR [31]. Notably, GR activation correlates with increased HSV-1 gene expression in latently infected neurons [32,33]. Approximately 50% of TG neurons express GR in a rat [34]. Understanding the mechanism by which reactivation from latency occurs is important because acyclovir does not dramatically reduce the episodes of reactivation from latency and there are no HSV-1 vaccines.

The goals of this study are to test whether E2F1, E2F2, or E2F3 transactivate cis-regulatory motifs (CRM) derived from the HSV-1 ICP0 promoter, and if GR enhances E2F-mediated transactivation.

2. Materials and Methods

2.1. Cells and Virus

Murine neuroblastoma (Neuro-2A) was grown in Minimal Essential Media (MEM; Corning, Glendale, AZ, USA) supplemented with 10% fetal bovine serum (FBS). All media contained penicillin (10 U/mL) and streptomycin (100 µg/mL). The HSV-1 McKrae strain was obtained from the late Dr. Steven Wechsler (University of California, Irvine, CA, USA), and stock cultures were prepared in monkey kidney cells (Vero; ATCC, Manassas, VA, USA).

2.2. Plasmids

The respective ICP0 CRM constructs were synthesized by Genscript and inserted into pGL4.23[luc2/minP] (Promega; Madison, WI, USA) at SacI and XhoI unique restriction enzyme sites. The GR-α expression construct was obtained from Dr. John Cidlowski (NIEHS, Research Triangle Park, NC, USA).

Plasmids expressing E2F1 and E2F2 (pCMV-E2F1 and pCMVE2F2, respectively) were obtained from Dr. J. R. Nevins (Duke University, Durham, NC, USA). E2F3a (Plasmid 3790) and E2F3b (Plasmid 37975) were obtained from Addgene (Watertown, MA, USA) via Dr. Jacqueline Lees [35]. All plasmids were prepared from bacterial cultures by alkaline lysis and 2 rounds of cesium chloride centrifugation.

2.3. Transfection and Dual-Luciferase Reporter Assay

Neuro-2A (~6 × 10^5^) were seeded into 60 mm dishes containing MEM with 10% FCS at 24 h prior to transfection. Cells were cotransfected with the designated ICP0 CRM constructs (0.5 ug plasmid DNA) and a plasmid-encoding Renilla luciferase under the control of a minimal herpesvirus thymidine kinase (TK) promoter (50 ng DNA). To maintain equal plasmid amounts in the transfection mixtures, empty expression vector (pCDNA3.1) was added as needed. Neuro-2A cells were incubated in MEM containing 2% charcoal-stripped FBS after transfection. At 24 h after transfection, cell cultures were treated with water-soluble DEX (10 µM; D2915; Sigma; St Louis, Missouri, USA). At 48 h after transfection, cells were harvested, and protein extracts subjected to a dual-luciferase assay using a commercially available kit (E1910; Promega). Luminescence was measured by using a GloMax 20/20 luminometer (E5331; Promega).

2.4. Chromatin Immunoprecipitation

ChIP studies were performed using standard procedures as previously described [1,2,3]. In brief, Neuro-2A cells were grown in 100 mm dishes until ~80% confluency. Cells were infected with HSV-1 (MOI of 1) or transfected with designated ICP0 CRM-C constructs (1.5 μg DNA) and a plasmid that expresses GR-α (3 μg DNA) and/or E2F2 (1.5 μg DNA) using Lipofectamine 3000 (Invitrogen; L3000015; Carlsbad, CA, USA) according to the manufacturer’s instructions. After transfection or infection, MEM with 2% stripped FBS was added to cultures. At 40 h post-transfection or 0, 4 and 8 h post-infection, cells were cross-linked with paraformaldehyde and harvested for ChIP analysis. Cleared samples were immunoprecipitated and incubated with a GR antibody (Cell Signaling, Cat# 3660S; Boston, MA, USA), E2F2 antibody (Invitrogen, Cat# PA5-41473), non-specific isotype control rabbit IgG (Abcam; Cat# ab171870; Cambridge, UK) in radioimmunoprecipitation assay (RIPA) buffer. DNA was purified using phenol-chloroform-isoamyl and amplified by PCR primers ICP0 CRM-C [AACCCCGGTATTCCCCGCCT (F) and AACAGGGGCTTGGCCTGCGT (R)] and ICP0-R0 [CGCTTCCCGGTATGGTAATTAGAAAC (F) and CGTGTGTTCCGCCAAAAAAGCAATTAGC (R)]. ICP0 R0 primer pairs, previously described [36,37], were designed using Strain 17 sequences, which are the same in McKrae. Following agarose gel electrophoresis, DNA bands were quantified using Image Lab software 6.1 for Mac and presented as percent of the input sample.

2.5. Statistical Analysis

All graphs and analyses were made using Microsoft Excel 365 for Mac. Student’s t-test with two-tailed and two-sample unequal variance was performed on the data sets. An * denotes significant difference in construct alone compared to its own cotransfected samples. An # denotes a significant difference in mutant samples compared to the relevant wt CRM cconstruct. The */# denotes a significant difference in p < 0.05, and **/## denotes a significant difference in p < 0.005, as determined by Student’s t-test. NS denotes a non-significant difference.

3. Results

3.1. GR and E2F2 Cooperatively Transactivate the ICP0 CRM-C Fragment

The HSV-1 ICP0 gene is present in the two repeats: hence, two copies of the ICP0 gene are in the viral genome (Figure 1A). An E2F1 siRNA was previously shown to impair BoHV-1 and HSV-1 replication in cultured cells [17,18]. Initial studies revealed E2F2 consistently had a modest level of transactivation on the full-length ICP0 promoter. Since the full-length ICP0 promoter contains numerous binding sites for cellular transcription sites, we predicted this may have dampened the effect of E2F2. Consequently, we tested whether ICP0 CRM fragments (Figure 1B) are transactivated by a plasmid that expresses GR-α and/or E2F1, E2F2, E2F3a, E2F3b in mouse neuroblastoma cells (Neuro-2A). E2F1, E2F2, and E2F3a strongly stimulate transcription compared to the other E2F family members, reviewed in [11,12]. E2F3b was used as a negative control because it is a negative regulator of E2F-dependent transcription, reviewed in [38]. Neuro-2A cells were used for this study because they are readily transfected, have certain neuronal features, and can be differentiated into dopamine-like neurons [39].

Initial studies tested whether ICP0 CRM fragments (Figure 2) were transactivated by E2F1- and/or GR-α-expression plasmids. Twenty-four h after transfection, certain cultures were treated with the synthetic corticosteroid DEX. Following 24h of DEX treatment, cells were collected and luciferase activity measured. Consistent with a previous study [40], GR-α and DEX cooperatively transactivate the ICP0 CRM A and B constructs but E2F1 reduces transcription relative to basal transcriptional activity of the CRM-A and CRM-B (Figure 2A,B). GR and DEX treatment reduced basal promoter activity of ICP0 CRM-C (Figure 2C) and CRM-D constructs (Figure 2D). In summary, E2F-1 impaired GR- and DEX-mediated transactivation of all 4 ICP0 CRM constructs.

Additional studies tested whether E2F2 transactivates the ICP0 CRM constructs. When the GR-α expression plasmid was transfected with E2F2, ICP0 CRM-A and ICP0 CRM-B promoter activity was reduced regardless of DEX treatment (Figure 3A,B). Interestingly, E2F2 transactivated the ICP0 CRM-C construct more than 20-fold but DEX had no effect (Figure 3C). Furthermore, E2F2 and GR-α had an additive effect on the transcriptional activation of the ICP0 CRM-C construct. The ICP0 CRM-D construct was not transactivated by E2F2 regardless of transfection with GR or DEX treatment (Figure 3D).

3.2. E2F3 Isoforms Do Not Transactivate the CRM-C Construct

The E2F3 gene encodes two isoforms, E2f3a and E2f3b, that differ in their N-termini [35]. Although these isoforms have overlapping sequences, E2F3a partially compensates for the loss of E2F1 and E2F3 whereas E2F3b impairs transcription [35,38]. E2F3a or E2F3b did not transactivate the ICP0 CRM-C construct in Neuro-2A cells (Figure 4) or the other ICP0 CRM constructs (data not shown).

3.3. GR-α and/or E2F2 Isoforms Do Not Transactivate the CRM-C Construct in NIH-3T3

Since Neuro-2A cells are mouse neuroblastoma cells and can be differentiated into dopamine-like neurons (36), we tested whether an immortalized mouse fibroblast cell line (NIH-3T3) supported transactivation of the CRM-C construct by GR-α and/or E2F2 (Figure 5). As expected, the CRM-C construct exhibited significantly higher levels of promoter activity relative to the empty vector (EV). When the CRM-C construct was cotransfected with the GR-α expression construct regardless of DEX treatment, promoter activity was reduced in NIH-3T3 cells. When E2F2 was cotransfected with the CRM-C construct, there was a significant reduction in promoter activity regardless of DEX treatment of NIH3T3 cells. Cotransfection of the CRM-C construct with GR and E2F2 regardless of DEX treatment also led to significantly reduced compared to basal CRM-C promoter activity. Consistent with results in Neuro-2A cells, the CRM-A, CRM-B, and CRM-D constructs were not transactivated by E2F2 and GR-α regardless of DEX treatment in NIH-3T3 cells (data not shown). In summary, this study suggests Neuro-2A-specific factors were important for transactivating the CRM-C fragment by GR and E2F2.

3.4. Influence of E2F Binding Sites and ½ GRE on GR- and E2F2-Mediated Transactivation of CRM-C

Examination of CRM-C revealed two consensus E2F binding sites; denoted as #1 and #2 (Figure 6A–C) [41,42]. E2F binding site #3 is a GC-rich motif that contains a consensus Sp1 binding site that was reported to interact with E2F2 [43,44]. A consensus ½ GRE [45] was also adjacent to the E2F2 #2 binding site. A GR monomer was reported to bind this ½ GRE in other promoters and is transactivated by increased stress [45].

Four mutant CRM-C fragments were prepared and tested whether these mutations influenced GR-α and E2F2 mediated transactivation (see Figure 6C–E for the mutants made in the E2F binding sites and ½-GRE). Surprisingly, mutating just the ½ E construct (CRM-C∆1/2GRE) did not significantly reduce basal transcriptional relative to the wt CRM-C construct; however, E2F2-mediated transactivation was significantly reduced (Figure 6F). Mutating all E2F binding sites (CRM-C∆1-3), two binding sites (CRM-C∆1,2) or just the #3 E2F2 binding site (CRM-C∆3) was transactivated by E2F2 significantly less than the wt CRM-C (Figure 6G). Notably, basal transcription was not significantly different between CRM-C∆1-3, CRM-C∆1,2, and CRM-C∆3. This study reveals transactivation by GR and E2F2 is complex and all E2F2 binding sites and the ½ GRE were crucial for transactivation by GR and E2F2.

3.5. E2F2 and GR Occupy the CRM C Region of the ICP0 Promoter During Productive Infection

Chromatin immunoprecipitation (ChIP) studies were performed to determine if GR and E2F2 occupied the ICP0 promoter in productively infected Neuro-2A cells. Within 60 min of incubating virus inoculum with Neuro-2A cells at 37 °C, GR bound to the ICP0 promoter was not significantly different than the isotype-specific antibody used as a negative control using ICP0 CRM-C primers (Figure 7B). However, significantly higher levels of E2F2 occupied the ICP0 promoter versus the isotype-specific antibody using the CRM-C primers. We suggest the infectious virus rapidly entered Neuro-2A cells, uncoating occurred, and low levels of viral DNA were present in the nucleus within 60 min. Consequently, low levels of GR and E2F2 occupied the ICP0 promoter that contains CRM-C sequences. Consequently, significantly higher levels of E2F2 occupying ICP0 promoter sequences were detected 60 min after the virus was incubated with Neuro-2A cells. Notably, significant differences between GR binding to the ICP0 promoter using the CRM-C primers were not observed. R0 primers span ICP0 promoter sequences that include CRM-C sequences, which we suggest is why significantly higher E2F2 and GR occupied the ICP0 promoter relative to the isotype-specific antibody at 0 h after infection (Figure 7D). It was also clear that GR and E2F2 levels bound to ICP0 CRM-C sequences increased at 4 and 8 h after infection, regardless of whether the R0 or CRM-C primers were used for ChIP.

3.6. Comparison of E2F2 and GR Occupying the wt ICP0 CRM-C Construct Versus Mutant ICP0 CRM-C Constructs

Neuro-2A cells were transfected with the ICP0 CRM-C, CRM-C ½GRE, or the CRM-C∆1-3 construct. At 48 h after transfection, cells were collected and ChIP studies performed to compare occupancy of E2F2 and GR in the wt or mutant CRM-C fragments. Binding of GR or E2F2 was significantly higher than the isotype antibody even when E2F2 or GR-α was not over-expressed (Figure 8A). As expected, over-expression of E2F2 lead to significantly higher levels of E2F2 and GR occupying the wt CRM-C fragment (Figure 8B). Over-expressing GR and E2F2 were not significantly different than just over-expressing E2F2 (Figure 8C). Comparing E2F2- and GR-binding wt CRM-C relative to the CRM-C ½ GRE reveals the E2F2 and GR binding to the mutant was significantly reduced relative to the wt CRM-C fragment (Figure 8E). A similar trend was observed when E2F2 and GR were over-expressed and E2F2 or GR occupancy was significantly reduced in the CRM-C ½ GRE construct (Figure 8F).

Comparing the wt CRM-C construct to the CRM-C∆1-3 construct, there was a dramatic reduction in GR and E2F2 occupying the CRM-C∆1-3 construct (Figure 7G). Over-expressing E2F2 (Figure 8H) or GR+E2F2 increased binding of E2F2 or GR to the wt CRM-C construct but had no effect on the CRM-C∆1-3 mutant construct. In fact, binding of E2F2 or GR to the CRM-C∆1-3 construct was not observed. This study reveals binding of E2F2 or GR to the CRM-C∆1-3 construct was not detected, whereas there were low levels of binding to the CRM-C ½GRE construct. Notably, promoter activations of these two mutant constructs by E2F2 and GR were not significantly different.

4. Discussion

The finding that GR and E2F2, but not E2F1, E2F3a, or E2F3b cooperatively transactivated the ICP0 CRM-C construct was unexpected because two consensus E2F binding sites are present in this fragment [41,42] (Figure 6). Furthermore, E2F2 #3 matches a GC-rich motif that E2F2 binds to and stimulates transcription [43,44]. We predicted that E2F1 and E2F3a would have transactivated the ICP0 CRM-C construct because these two E2F family members typically activate transcription [11,12,35]. Mutating the ½ GRE adjacent to the 2nd E2F2 consensus binding site reduced GR- and E2F2-mediated transactivation as efficiently as the construct where all three E2F2 binding sites were mutated. This result implies GR played a pivotal role in E2F2-mediated transactivation. In contrast, mutating one or all E2F2 binding sites resulted in GR and E2F2 binding not being detected (Figure 8G–I), whereas low levels of GR and E2F2 binding were detected when just the ½ GR site was mutated (Figure 8D–F). We suggest that mutating one E2F2 binding site impairs binding of E2F2 to the other E2F2 binding sites and GR is only recruited to the ½ GRE when all E2F2 binding sites are intact. In contrast to the ICP0 promoter, promoters or CRM constructs that drive ICP4, ICP27, or VP16 expression were not transactivated by GR and/or E2F2 regardless of DEX treatment in Neuro-2A cells (unpublished studies, Jayathilake and Jones). Furthermore, these promoters do not contain E2F consensus binding sites. It is reasonable to predict that HSV-1 can replicate in actively growing cells because they express higher levels of E2F2. This prediction can be directly tested because E2F2 knockout mice are available.

Notably, GR and E2F2, but not E2F1, E2F3a, or E2F3b cooperatively transactivate the bovine alphaherpesvirus 1 (BoHV-1) immediate early transcription unit 1 (IETu1) promoter that drives immediate early ICP0 and ICP4 expression [46]. The IEtu1 promoter contains two “whole” GREs [47], and mutating these GREs dramatically reduces promoter activity, including GR- and E2F2-mediated transactivation [46,48]. A 222 bp fragment upstream of the two GREs in the IEtu1 promoter contains four consensus Sp1 binding sites. Mutating these Sp1 binding sites significantly reduced GR- and E2F2-mediated transactivation; however, the effect was not as dramatic compared to mutating the two GREs. Although the CRM-C fragment contains numerous GC or CG motifs that resemble consensus Sp1 binding sites, there are no consensus Sp1 binding sites. Finally, E2F1 transactivates the HSV-1 thymidine kinase (TK) promoter, and transactivation is reduced when the GC-rich motif, not a consensus E2F binding site, is mutated [49]. Unfortunately, E2F2 was not tested for whether it transactivates the TK promoter. The inability of E2F2 and GR-α to transactivate the ICP0 CRM-C construct suggests that a neuronal-specific transcription factor is expressed in Neuro-2 cells, but not NIH 3T3 cells. It is also possible that negative regulators present in NIH-3T3 cells are possible, or novel chromatinization of the CRM-C construct impaired transactivation. Furthermore, it will be interesting to develop a mutant virus that lacks the ½ GRE and E2F2 binding sites in the CRM-C region and test whether this impairs viral replication, viral pathogenesis in a mouse model of infection, and/or the latency reactivation cycle.

Despite having overlapping functions as the E2F activators (E2F1 and E2F3), E2F2 has novel functions that are important for GR-mediated transactivation of the CRM-C fragment and the IEtu1 promoter. For example, E2F1 and E2F2 over-expression in the absence of growth factors will drive cells into the S phase; however, E2F2 does not induce apoptosis in certain cell types, reviewed in [50]. Conversely, E2F1 over-expression induces apoptosis in most cells in the absence of growth factors [13,51]. E2F2, but not E2F1 or E2F3, is required for normal cell-cycle progression from G_o_ to G_1_, because in-mouse embryonic cells and T-cells lacking E2F2 prematurely enter the G_1_ and S phase [52]. In keeping with this finding, an independent study concluded E2F2 knockout mice contained more proliferating activated lymphocytes [53]. Tumors develop in certain cell types in the absence of E2F1 and E2F2 expression, suggesting these genes exhibit tumor suppressor activity under certain circumstances [54]. Finally, E2F2 impairs Myc-mediated proliferation and tumor formation [55], confirming E2F2 has tumor suppressor functions.

E2F2 has also been reported to regulate the PI3K/Akt/NF-kB signaling axis, which can induce expression of pro-inflammatory mediators [56,57]. Interestingly, the PI3K/Akt signaling axis is linked to the HSV-1 latency reactivation cycle, reviewed in [58], and appears to play a role in the BoHV-1 latency reactivation cycle [59,60]. Since ICP0 can transactivate all viral promoters [61], enhance chromatin remodeling to stimulate viral gene expression [62,63], and impair innate immune responses [64,65], it is not surprising that ICP0 triggers reactivation from latency of TG neurons [66]. Furthermore, ICP0 has intrinsic E3 ubiquitin ligase activity that is reported to reduce p53 protein levels [67] and other proteins, including promyelocytic leukemia (PML) bodies [67,68,69] that impair viral replication. In summary, transactivation of CRM-C by GR and E2F2 is predicted to stimulate ICP0 promoter activity: consequently, virus replication and reactivation from latency are predicted to be increased.

5. Conclusions

This study reveals that only E2F2 transactivates the ICP0-CRM-C. Surprisingly, E2F1, E2F3a, and E2F3b had no effects on any of the ICP0 CRM fragments. This study confirmed and enhanced our understanding that certain E2F family members, for example E2F2, have novel properties relative to E2F1, E2F3a, and E2F3b. Furthermore, this study reveals GR and E2F2 have an enhanced transactivation of CRM-C, in part because this fragment contains a functional ½ GRE. Despite the ability of HSV-1 to replicate in quiescent cells, cell-cycle factors can stimulate viral gene expression in certain cells, which enhances viral spread in the host.