Bovine herpesvirus 1 (BoHV-1), an important bovine pathogen, establishes life-long latency in sensory neurons within trigeminal ganglia (TG). Stress, as mimicked by the synthetic corticosteroid dexamethasone, consistently induces reactivation in calves latently infected with BoHV-1. Dexamethasone induces expression of several transcription factors in TG neurons during early stages of reactivation, including Krüppel-like transcription factors (KLF): KLF4, KLF6, KLF15, and promyelocytic leukemia zinc finger. Furthermore, the glucocorticoid receptor (GR) is consistently detected in TG neurons expressing viral regulatory proteins during reactivation from latency. The viral immediate early transcription unit 1 (IEtu1) promoter that drives expression of two viral transcription factors (bICP0 and bICP4) contains two GR response elements (GREs) and is stimulated by DEX. KLF15 and the GR form a feed forward transcription loop that synergistically stimulates productive infection and IEtu1 promoter activity. New studies demonstrate the GR and KLF6 synergistically stimulate productive infection and IEtu1 promoter activity if the GREs are intact. Furthermore, the GR and KLF6 interact with wild-type GREs within the IEtu1 promoter, but not GRE mutants. These studies suggest that certain KLF family members and the GR can convert a silent viral genome in latently infected neurons into an actively transcribing genome during reactivation from latency.
- bovine herpesvirus 1
- immediate early transcription
- reactivation from latency
- sensory neurons
- glucocorticoid receptors
- Krüppel-like transcription factors
Bovine herpesvirus 1 (BoHV-1), an alpha-herpesvirinae subfamily member, is an important bovine pathogen that causes conjunctivitis and/or upper respiratory tract disease [1, 2, 3]. BoHV-1 is a significant cofactor in the polymicrobial disease referred to as bovine respiratory disease complex (BRDC), which is the most important disease of cattle. BoHV-1 infection impairs cell-mediated immunity, CD8+ T cell recognition of infected cells, and induces apoptosis in CD4+ T cells . Viral proteins, VP8, bICP0 and bICP27, inhibit interferon dependent transcription [4, 5, 6, 7, 8]. Infection also erodes mucosal surfaces of the upper respiratory tract, which promotes establishment of the bacterial pathogen, Mannheimia haemolytica (MH) in the lower respiratory tract . BoHV-1 productive infection increases neutrophil adhesion and activation , thus amplifying the pathogenic potential of MH. MH, a gram negative bacterium, exists as normal flora within the upper respiratory tract of healthy ruminants . Stress and/or co-infections disrupt this commensal relationship; consequently MH becomes the predominant organism that causes life-threatening bronchopneumonia in BRDC cases . BRDC is the most important disease in cattle because it costs the US cattle industry more than one billion dollars in losses each year [9, 12, 13]. A BoHV-1 entry protein is a BRDC susceptibility gene for Holstein calves  confirming BoHV-1 is a significant BRDC cofactor.
Like most alpha-herpesvirinae subfamily members, including human herpes simplex virus 1 (HSV-1) and HSV2, BoHV-1 initiates acute infection on mucosal surfaces [1, 2, 3]. High levels of infectious virus are produced; consequently BoHV-1, HSV-1, or HSV-2, spread to the peripheral nervous system via cell-to-cell spread. Latency is subsequently established in sensory neurons, but periodically reactivates from latency, and thus is widespread in cattle throughout the world. Reactivation of the virus from the latent state is initiated by external stimuli (e.g. stress and immunosuppression). During reactivation, viral gene expression is stimulated and infectious virus is produced and transported back to mucosal surfaces. The ability of alpha-herpesvirinae subfamily members to reactivate from the latent state is critical for virus transmission. Regulation of the complex virus host interactions controlling the latency-reactivation cycle is not well understood, which hinders developing therapeutic strategies that prevent reactivation from latency.
BoHV-1 is an excellent model to study these events because the natural host can be used and the synthetic corticosteroid dexamethasone (DEX) consistently initiates reactivation from latency in infected calves . We have used experimentally infected calves treated with DEX to initiate reactivation from latency in order to identify virus-host interactions important for the latency-reactivation cycle. These studies identified host cellular factors and pathways that may be crucial for latency maintenance  and reactivation . The following discussion focuses on the mechanisms by which BoHV-1 “escapes” a latent infection following a stressful stimulus and subsequently successfully reactivates from latency. Certain steps during BoHV-1 reactivation from latency are likely to be similar during reactivation of latency of other alpha-herpesvirinae subfamily members.
2. Acute infection leads to a life-long latent infection in sensory neurons
Acute infection of calves induces programmed cell death, inflammation and high levels of virus production [1, 2, 3]. BoHV-1 genes are expressed in three distinct phases during acute infection or productive infection of cultured cells: immediate early (IE), early (E), or late (L) [1, 2, 3]. IE gene expression is specifically stimulated by viral protein 16 (VP16), a tegument protein. IE transcription unit 1 (IEtu1) encodes two transcriptional regulatory proteins, BoHV-1 infected cell protein 0 (bICP0) and bICP4, because a single IE transcript is differentially spliced and subsequently translated into bICP0 or bICP4 (Figure 1). The bICP0 protein is also translated from an E mRNA (E2.6) because a separate E promoter drives expression of the bICP0 E transcript.
During acute infection of calves, infectious virus particles enter the peripheral nervous system via cell–cell spread. If infection is initiated within the oral, nasal, or ocular cavity, the primary site for latency is sensory neurons located in trigeminal ganglia (TG) [1, 2, 3]. Viral gene expression and infectious virus are detected in TG from 2 to 6 days after infection. Lytic cycle viral gene expression is then extinguished, a significant number of infected neurons survive, and these neurons harbor viral genomes, which is operationally defined as the establishment of latency. Abundant expression of the BoHV-1 encoded latency related (LR) gene occurs in latently infected neurons, but infectious virus is not detected (maintenance of latency) [1, 2, 3]. LR-RNA is anti-sense to and overlaps the BoHV-1 infected cell protein 0 (bICP0) gene. The LR gene has two open reading frames (ORF1 and ORF2), and two reading frames lacking an initiating methionine (RF-B and RF-C). In addition, the LR gene encodes two micro-RNAs that interfere with bICP0 expression in transfected cells . A LR mutant virus strain with three stop codons at the N-terminus of ORF2 exhibits diminished clinical symptoms, and reduced virus shedding from the eye, TG, or tonsils of infected calves [1, 2, 3]. ORF1, ORF2, and RF-C are expressed when bovine cells are infected with wild-type or the LR-rescued virus, but these proteins have reduced or no expression following infection with the LR mutant virus [1, 2, 3]. Wild-type (wt) BoHV-1, but not the LR mutant virus, reactivates from latency after treatment with the synthetic corticosteroid DEX. The anti-apoptosis activity of ORF2 is predicted to increase the survival of infected neurons and thus would be important for the latency-reactivation cycle [1, 2, 3].
Recent studies demonstrated that during latency, the canonical Wnt/β-catenin signaling is active and ORF2 appears to be important for maintaining this important signaling pathway [15, 18]. Although dysregulation of the Wnt/β-catenin signaling is frequently associated with many types of cancer [19, 20], this signaling pathway has the potential to promote the establishment and maintenance of latency in sensory neurons because it enhances cell survival, axonal growth, and directs axons to their proper synaptic targets [21, 22, 23, 24, 25].
3. Stress-induced reactivation from latency
Increased corticosteroid levels, due to increased stress, correlates with increasing the incidence of BoHV-1 reactivation from latency [1, 2, 3]. DEX can also stimulate productive infection , and initiate reactivation from latency in calves or rabbits latently infected with BoHV-1 [1, 2, 3]. Six hours after DEX treatment lytic cycle viral RNA expression is detected in neurons of latently infected calves [27, 28]. Certain lytic cycle viral proteins, bICP0 and VP16 for example, are readily detected in TG neurons within hours after DEX treatment [29, 30]. The glucocorticoid receptor (GR) and mineralocorticoid receptor (MR), which are present in sub-populations of sensory neurons [31, 32], are activated by interacting with corticosteroids. The GR is frequently detected in TG neurons that express bICP0 or VP16 [31, 32]. IEtu1 promoter activity is stimulated by the GR and the synthetic corticosteroid DEX because there are two consensus GREs in the promoter  suggesting this promoter is activated by the GR and/or MR following stressful stimuli. Since the IEtu1 promoter drives expression of two viral transcriptional regulatory proteins (bICP0 and bICP4; Figure 1), activation of this promoter may stimulate productive infection in certain latently infected neurons.
DEX treatment of latently infected calves induces apoptosis of T cells that persist in TG after infection . T cells also persist in TG of humans or mice latently infected with HSV-1 and may promote maintenance of latency [33, 34, 35, 36, 37]. Within 3 h after DEX treatment, 11 cellular genes are induced more than ten fold in TG . Pentraxin 3, a regulator of innate immunity and neuro-degeneration, is stimulated 35–63 fold at 3 or 6 h after DEX treatment. Furthermore, expression of a soluble Wnt antagonist, Dickkopf-1 is induced more than 10 fold [15, 16]. Dickkopf-1 is responsible for stress-induced neuronal death [38, 39] indicating there is a correlation between disrupting the Wnt signaling pathway and activation of lytic cycle viral gene expression during reactivation. Two transcription factors, promyelocytic leukemia zinc finger (PLZF) and Slug are induced more than 15-fold 3 h after DEX treatment. PLZF or Slug stimulates BoHV-1 productive infection 20-fold or 5-fold respectively, and Slug stimulates the late glycoprotein C promoter more than 10-fold. Additional DEX induced transcription factors, SPDEF (Sam-pointed domain containing Ets transcription factor), Kruppel-like transcription factor 15 (KLF15), KLF4, KLF6, and GATA6, stimulate productive infection and certain key viral promoters.
The finding that four KLF family members (KLF4, KLF6, KLF15, and PLZF) are stimulated during DEX induced reactivation from latency is intriguing because KLF family members resemble the SP1 transcription factor family and both family of transcription factors interact with guanine-cytosine (GC) rich motifs, reviewed in [40, 41]. Genomes of alpha-herpesvirinae subfamily members, including BoHV-1, are GC rich and many viral promoters contain Sp1 consensus binding sites as well as other GC rich motifs . In fact, HSV-1 gene expression is activated by Sp1 . HSV-1 and probably BoHV-1 genomes exist as silent chromatin during latency, : however, HSV-1 DNA is associated with unstable chromatin during productive infection [44, 45, 46]. Regardless of the stimulus that initiates reactivation from latency, silent viral heterochromatin must be converted into an actively transcribing template for reactivation from latency to be successful suggesting cellular transcription factors initially stimulate lytic cycle viral gene expression.
To test whether the GR and certain stress-induced transcription factors can cooperate to stimulate viral transcription, the IEtu1 promoter and BoHV-1 DNA fragments (less than 400 bp) containing potential GR and KLF binding sites were identified and examined for transcriptional activation by stress-induced transcription factors. The rational for testing intergenic regions of the BoHV-1 genome is the viral genome contains more than 100 putative GRE binding sites  and a subset of GREs in cellular chromatin can activate transcription from greater than 5 kb to the nearest promoter . KLF15 cooperated with the GR to stimulate the IEtu1 promoter activity and productive infection . Furthermore, intergenic regions within the unique long 52 gene (UL-52; component of DNA primase/helicase complex), bICP4, IEtu2 that expresses the regulatory protein (bICP22), and unique short region were stimulated by KLF15 and the GR. In contrast to KLF15, the other stress-induced transcription factors only have a modest effect on IEtu1 promoter activity. The GR and KLF15 interact with sequences within wild-type IEtu1 GREs and UL-52 fragment, but not GRE mutants. Co-immunoprecipitation studies indicated that KLF15 and the GR are stably associated with each other. Interestingly, the GR and KLF15 can synergistically regulate gene expression by a feed-forward transcription loop [49, 50, 51]. Hallmarks of a feed-forward loop are a primary factor (GR in this example) induces expression of a second factor, KLF15 [16, 49, 50, 51, 52, 53, 54], and the two factors synergistically activate expression of genes in a specific pathway. Adipogenesis  and amino acid metabolizing enzymes are also synergistically regulated by the GR and KLF15 [50, 51]. In summary, these studies suggest that activation of BoHV-1 gene expression during DEX induced reactivation from latency is, in part, regulated by a feed-forward transcription loop containing the GR and KLF15.
4. The GR and KLF6 cooperate to stimulate productive infection
To test whether KLF6 and the activated GR have a cooperative effect on productive infection, a mouse neuroblastoma cell line (Neuro-2A) was cotransfected with gCblue genomic DNA and KLF6 and/or the GR. The gCblue virus grows to similar titers as the wt parental virus and expresses the Lac Z gene from the gC locus during productive infection (Figure 2A). Neuro-2A cells were used for these studies because they have neuronal like properties , can be readily transfected, and are semi-permissive for BoHV-1 . Neuro-2A cells were transfected with gCblue DNA instead of infecting cells because VP16 and other viral regulatory proteins in the virion particle can diminish the stimulatory effects of DEX on productive infection (data not shown). KLF6 and the GR plus DEX treatment increased the number of β-Gal+ Neuro-2A cells more than 4-fold, which was significantly higher than GR + DEX and the GR or KLF6 alone (Figure 2A and B). Cotransfection of gCblue and the GR + KLF6 stimulated productive infection 2-fold even when cultures were not treated with DEX, which was similar to the effects observed when gCblue genomic DNA was cotransfected with the GR and DEX treatment.
5. KLF6 and GR synergistically trans-activates the IEtu1 promoter
Transient transfection studies were performed in Neuro-2A cells to test whether KLF6 and the GR synergistically trans-activate the IEtu1 promoter because this promoter contains two consensus GR binding sites (Figure 3A) required for DEX mediated transactivation . The IEtu1 promoter drives IE expression of bICP0 and bICP4, the most important viral transcriptional regulatory proteins encoded by BoHV-1 [58, 59, 60] (Figure 1). The IEtu1 collapsed promoter construct (Figure 3A) was initially used to test whether sequences adjacent to the GREs were trans-activated by KLF6 and the GR. The full-length IEtu1 promoter construct contains extensive sequences downstream from the start site of transcription and has sequences between the TATA box and the GREs that are important for KLF trans-activation : consequently the collapsed IEtu1 collapsed promoter construct was used for these studies. We have consistently found that the GR+ DEX stimulated promoter activity more than 15 fold and GR + KLF6 + DEX stimulated promoter activity more than 50 fold (Figure 3B). RU486 antagonizes corticosteroid/GR signaling [61, 62] and as expected RU486 significantly reduced the ability of KLF6 and GR to transactivate the IEtu1 collapsed construct.
A 280 bp fragment containing both GREs within the IEtu1 promoter and flanking sequences was cloned upstream of the minimal SV40 early promoter and designated IEtu1 GREs  (Figure 3A). This construct was examined for its ability to be activated by KLF6 and the GRE as a comparison to the IEtu1 collapsed promoter construct. KLF6 and the GR consistently stimulated the IEtu1 GREs construct approximately 16-fold whereas the GR + DEX stimulated this construct only 6-fold (Figure 3B). RU486 also significantly reduced the ability of KLF6 and GR to transactivate the IEtu1 GREs construct. Although the IEtu1 collapsed construct was trans-activated more by the GR + KLF6+ DEX relative to the IEtu1 GREs construct, the overall trends were similar.
6. The GREs are necessary for transactivation by the GR and KLF6
To identify sequences in the IEtu1 GREs that mediate transactivation by KLF6 and the GR, constructs containing site-specific mutations in GRE#1, GRE#2, and KLF like binding sites were compared to the wt IEtu1 GREs (Figure 4A–C) . Mutagenesis of GRE1 (∆GRE1) or both GREs and the KLF binding sites (∆2xGRE∆KLF) significantly reduced cooperative activation by KLF6 and the GR when DEX was added to the cultures (Figure 4D). Mutagenesis of the 2 putative KLF sites (∆KLF) had no effect on trans-activation by KLF6 and the GR when cultures were treated with DEX. As previously reported  and shown in Figure 4D, the effect of DEX and the GR was significantly reduced when GRE#1 (∆GRE1) was mutated and abolished when both GREs and putative KLF sites (∆2xGRE∆KLF) were mutated. In summary, mutagenesis of the GRE#1 significantly reduced synergistic transactivation by KLF6 and the GR when cultures were treated with DEX.
7. KLF6 and the GR interact with sequences located in the IEtu1 GREs
To test whether KLF6 and the GR interact with sequences located in the IEtu1 GREs, chromatin immuno-precipitation (ChIP) studies were performed in Neuro-2A cells. Cells were transfected with the promoter construct containing the IEtu1 GREs followed by treatment with Vehicle or DEX. As shown in Figure 5A, ChIP studies demonstrated that the GR and KLF6 occupied the GRE region of the IEtu1 GREs (lanes 2–4). No specific PCR product was amplified from ChIPs of cells transfected with the IEtu1 GREs from IPs using the control IgG (IgG C Panel) or cells transfected with the ∆2XGRE∆KLF construct (Figure 5C and D). Treatment with DEX had little effect on the levels of GR bound to IEtu1 GRE sequences (Figure 5A and B); however, we detected an increase in KLF6 bound to the IEtu1 GREs when cotransfected with KLF6 and GR in the absence of DEX when compared to DEX treatment. At least three reasons may have led to this unexpected result. First, we suggest that low levels of corticosteroids in media containing 2% stripped fetal bovine serum may be a reason why the GR was associated with the IEtu1 GREs in the absence of DEX. Secondly, independent studies concluded that the GR can be associated with GREs in the absence of corticosteroids [61, 63]. Thirdly, treatment of cells with DEX reduces GR levels and the availability of GR to bind to DNA [26, 64]. All input samples (whole lysate prior to IP) yielded the specific 107 bp PCR product except Neuro-2A cells not transfected with the IEtu1 GREs construct (Figure 5A, Input panel, lane 1). In summary, the GR and KLF6 were specifically recruited to IEtu1 GRE sequences when the GREs were intact.
8. The GR does not stably interact with KLF6
Co-immunoprecipitation (co-IP) studies were used to test whether GR and KLF6 physically interact. Neuro-2A cells were cotransfected with plasmids that express KLF6 and the GR. Following IP with the GR antibody, we were unable to detect KLF6 in the immunoprecipitate regardless of DEX treatment (Figure 6). As expected, both proteins were detected in whole cell lysate (input). Furthermore, the GR was detected in the immunoprecipitate after IP was performed with the GR antibody. When KLF6 was used to IP whole cell lysate, the GR was not detected in the immunoprecipitate (data not shown). The same experimental conditions revealed that KLF15 and the GR were stably associated in transfected Neuro-2A cells . In summary, co-IP studies suggested KLF6 and the GR were not stably associated with each other.
9. Discussion and summary
In this study, we provided evidence that KLF6 and the GR synergistically stimulate productive infection and IEtu1 promoter activity. The IEtu1 promoter must be activated for productive infection because it encodes two viral transcriptional regulators, bICP0 and bICP4 (Figure 7A) . During reactivation from latency, stress, as mimicked by the synthetic corticosteroid DEX, activates the GR and induces expression of several KLF family members (KLF4, KLF6, KLF15, and PLZF) . A previous study demonstrated that KLF15, but not KLF4, and the GR synergistically stimulate IEtu1 promoter activity . In contrast to KLF6, KLF15 stably interacts with the GR to establish a feed-forward transcriptional loop [48, 51, 53, 65, 66] (Figure 7B). Although KLF6 and KLF15 can both positively regulate promoter activity, they also can repress transcription in a promoter-specific manner [67, 68]. One study concluded there is a synergistic effect between the GR and transcriptional factors that recognize CACCC motifs , a known KLF6 binding site [70, 71]. There are no CACCC motifs on the positive strand of the IEtu1 GREs fragment; however, there are 2 CACCC motifs on the negative strand (KLF-1 like; Figure 7B). When these motifs were mutated (∆KLF mutant), there was no difference in KLF6 and GR mediated trans-activation suggesting there may be KLF binding sites located between GRE#2 and GRE#1.
Relative to GRE#2, mutating GRE#1 was more important for GR mediated trans-activation [26, 48]. To ablate DEX induction of the IEtu1 promoter or the IEtu1 GREs, both GREs must be mutated [26, 48]. This is consistent with the results demonstrating there are cooperative effects between KLF15  or KLF6 and the GR. ChIP results demonstrated that mutagenesis of both GREs interfered with KLF6 binding to sequences spanning the IEtu1 GREs, suggesting: 1) an unknown GR or KLF6 coactivator functions as a bridge between the GR and KLF6, which allows interactions between these two transcription factors (Figure 7C; left scenario at GRE#2), or 2) GR interactions with GRE#1 and/or GRE#2 influence adjacent sequences that are necessary for KLF6 to bind DNA Figure 7B; right scenario at GRE#1). Since KLF family members can bind to several GC or CA rich motifs, it is difficult to predict which sequences adjacent to GRE#1 or GRE#2 are important for interacting with KLF6 and/or KLF15.
The BoHV-1 genome contains approximately 100 putative GREs . We identified 13 intergenic regions in the viral genome that contain at least 2 putative GREs and potential KLF binding sites within 400 base pairs. KLF15 and the GR significantly transactivate fragments present in unique long (UL)-52, bICP4, IEtu2, and Us fragments when DEX was added to cultures . In contrast, KLF6 and the GR were unable to transactivate these intergenic fragments in the presence or absence of DEX (data not shown) confirming KLF15 has novel properties relative to KLF6.
KLF4, KLF6, and KLF15 expression are induced in TG neurons of calves that are latently infected with BoHV-1 during early stages of DEX induced reactivation from latency . Cellular, not viral encoded, transcription factors are predicted to be crucial for initiating viral transcription during initial stages of reactivation from latency because lytic cycle viral gene expression is not readily detected in TG of latently infected calves [29, 30]. Thus, activation of the IEtu1 promoter by the GR and DEX-induced transcription factors, KLF6 and KLF15 for example, may be sufficient to trigger lytic cycle viral gene expression in a subset of latently infected neurons following a stressful stimulus, as shown in Figure 7B and C.
This research was supported by grants from the USDA-NIFA Competitive Grants Program (13-01041 and 16-09370), funds derived from the Sitlington Endowment, and support from the Oklahoma Center for Respiratory and Infectious Diseases (National Institutes of Health Centers for Biomedical Research Excellence Grant # P20GM103648). Research reported in this publication was also partially supported by the National Institute Of Neurological Disorders And Stroke of the National Institutes of Health under Award Number R21NS102290. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Fouad S. El-mayet was supported by a fellowship from the Egyptian Ministry of Higher Education, Mission Sector (JS-3541).