Rescue of Pentamer-Null Strains of Human Cytomegalovirus in Epithelial Cells by Use of Histone Deacetylase Inhibitors Reveals an Additional Postentry Function for the Pentamer Complex

ABSTRACT

Human cytomegalovirus (HCMV) tropism for epithelial cells is determined by the pentameric glycoprotein complex found on the viral envelope. Laboratory-adapted strains, such as AD169, typically develop loss-of-function mutations for the pentamer, thus losing the ability to efficiently initiate lytic replication in epithelial cells. Using our human salivary gland-derived epithelial (hSGE) cell model, we observed that 3 chemically distinct histone deacetylase (HDAC) inhibitors can rescue infection in hSGE cells using pentamer-null strains of HCMV. Additionally, infection in ARPE-19 epithelial cells was rescued in a similar manner. We isolated nuclei from AD169-infected cells, quantified viral genomes by quantitative PCR (qPCR), and discovered that while HDAC inhibitors increased immediate early (IE) gene expression, they did not increase the amount of viral DNA in the nucleus. Using immunofluorescence microscopy, we observed that pentamer-null strains showed punctate patterning of pp71 in proximity to the nucleus of infected cells, while pp71 was localized to the nucleus after infection with pentamer-containing strains. Upon treatment with HDAC inhibitors, these punctae remained perinuclear, while more cells displayed entry into the lytic cycle, noted by increased IE-positive nuclei. Taken together, our data indicate that HCMV pentamer-null viruses are able to infect epithelial cells (albeit less efficiently than pentamer-positive viruses) and traffic to the nucleus but fail to initiate lytic gene expression once there. These studies reveal a novel postentry function of the pentamer in addition to the recognized role of pentamer in mediating entry.

IMPORTANCE Human cytomegalovirus has a wide cellular tropism, which is driven by one of its glycoprotein complexes, the pentamer. Laboratory-adapted strains continuously passaged on fibroblasts readily lose pentamer function and thus lose their ability to infect diverse cell types such as epithelial cells. Pentamer has been attributed an entry function during infection, but mechanistic details as to how this is achieved have not been definitely demonstrated. In this study, we investigate how pharmacological rescue of pentamer-null strains during epithelial infection by histone deacetylase inhibitors implicates a novel role for the pentamer downstream of entry. This work expands on potential functions of the pentamer, will drive future studies to understand mechanistically how it affects tropism, and provides a new target for future therapeutics.

INTRODUCTION

Human cytomegalovirus (HCMV) is found ubiquitously throughout the world, and while estimates vary, the virus likely infects 50 to 90% of the global population (1). Immunocompetent hosts rarely develop serious disease from a primary HCMV infection; however, immunocompromised individuals can develop life-threatening illnesses. AIDS patients can develop deadly viral gastroenteritis, and organ transplant patients can suffer organ rejection and associated morbidity (2–4). HCMV can also cross the placental barrier, leading to congenital infections of developing fetuses, which can result in developmental birth defects (5). There remains no approved vaccine for HCMV, and it is critical to understand the basic features of its pathogenesis to aid in the development of efficacious vaccines and novel antiviral therapeutics.

HCMV has a diverse cellular tropism, being competent to infect fibroblasts, myeloid cells, endothelial cells, and epithelial cells derived from diverse tissues such as the salivary gland (6–9). Persistent shedding of HCMV from salivary epithelial cells into salivary secretions is one of the mechanisms by which it can be spread and infect new hosts (10). An understanding of the mechanisms underlying HCMV infection in the salivary epithelium and how it differs from infection of other cell types and in other tissues is not known, and salivary replication is modeled primarily using an in vivo murine cytomegalovirus (MCMV) model (11, 12). More recently, our group developed a model using primary cells derived from human salivary glands to better understand the specific virulence factors and mechanisms used by HCMV for infection and replication within salivary epithelial cells (9).

One factor that leads to the wide cellular tropism exhibited by HCMV is variation in the mechanisms used for viral entry. HCMV particles are enveloped in a lipid membrane, which needs to fuse with a host membrane to allow viral particle entry and subsequent infection to proceed. This can occur at the level of the plasma membrane via direct fusion or can occur after the endocytic uptake of viral particles into the cell (13, 14). In the well-studied fibroblast systems, direct fusion with the plasma membrane is thought to deliver virions into the host cell; however, more recent work suggested that other potential entry mechanisms in addition to direct fusion may exist (14, 15). In the case of epithelial cells, internalization of virions into endosomes followed by vesicle acidification and fusion with endosomal membranes has been demonstrated to be an important mechanism for entry (8, 13). Other studies in epithelial cells have demonstrated that some HCMV strains can use a macropinocytosis pathway involving low-pH-dependent fusion (15–17). The cellular receptors involved in each of these pathways remain unclear, although PDGFRα, NRP2, and OR14I1 have been implicated to various degrees (18–20). Taken together, the mechanisms of entry and the cellular receptors utilized are likely to vary significantly depending on the cellular phenotype and viral strain used and remain an active area of investigation.

HCMV encodes multiple glycoproteins that assemble into distinct complexes on the viral envelope. These glycoprotein complexes are involved in processes such as initial binding to a host cell via heparin sulfate proteoglycans or mediating fusion between the viral envelope and cellular membranes. The gH and gL proteins form a disulfide-linked heterodimer that serves as the basis for 2 different complexes dependent upon which other viral proteins they associate with. If gH/gL assembles with gO, a trimeric complex (trimer) is formed, which binds PDGFRα and potentially mediates fusion through the gB homotrimer in all cell types expressing PDGFRα (21–23). If the gH/gL heterodimer assembles with UL128/UL130/UL131, a pentameric complex (pentamer) is formed, which is required for HCMV to infect most nonfibroblast cell types, such as epithelial cells, and as such is a major tropism determinant for the virus (8). There has been much speculation on the mechanism of function for the pentamer during epithelial infection, including a recent study that identifies CD147 as a potential coreceptor for the pentamer to indirectly mediate fusion during epithelial and endothelial infection (20). However, the precise mechanism of pentamer-mediated viral entry into different cell types remains poorly understood, likely complicated by multiple cellular receptors binding to the trimer and pentamer with additional coreceptors forming large or variable complexes (16, 18, 19, 24, 25). The variability of viral particles produced from different cell types also likely influences the entry mechanisms utilized by HCMV, possibly due to fluctuations in the amount of pentamer incorporated into the viral envelope (22, 26).

Following viral entry, the virus enters into either a lytic program followed by the new synthesis of viral particles (fibroblasts, epithelial cells, and endothelial cells) or a latent program in which new particles are not produced (myeloid progenitors). The initiation of the lytic program is typically carried out following the production of the immediate early (IE) products IE1 and IE2. The expression of IE1/2 is controlled by the major immediate early promotor (MIEP), which is regulated by numerous host factors such as NF-κB and can be autoregulated via IE2 (27–29). Shortly after the viral genome has entered the nucleus, it is thought that the host cell recognizes this foreign DNA and utilizes histones to compact the DNA alongside transcriptional silencing machinery containing histone deacetylases (HDACs) (30). Daxx is a cellular protein that is involved in the recruitment of HDAC-containing complexes such as the NuRD complex, which has been shown to lead to the transcriptional repression of the HCMV MIEP (31). The HCMV tegument protein pp71, packaged into mature virions, traffics to the nucleus where it leads to the proteasome-dependent, ubiquitin-independent degradation of Daxx, ultimately resulting in the derepression of the MIEP and the initiation of viral lytic protein production (32–34). Entry into the lytic cycle can lead to the activation of the host immune response, which can be detrimental to the establishment of HCMV latency in cells such as CD34⁺ hematopoietic progenitor cells (35). To control this, pp71 is excluded from the nucleus during latency where it is unable to interact with cellular Daxx until the virus is triggered to initiate lytic replication (17, 36). Pharmacological inhibition of host HDACs can reverse this process in myeloid progenitor cells, leading to the reinitiation of the viral lytic program, but leads to incomplete progression through the lytic cycle (37).

In the current study, we have used our primary salivary cell model alongside the established retinal pigment epithelial (ARPE-19) cellular model and characterized the ability of HDAC inhibitors to enhance HCMV replication in epithelial cells. These data reveal an additional, previously unappreciated postentry role for the pentamer complex that is important for driving the lytic phase in epithelial cells.

RESULTS

Valproic acid rescues the epithelial-defective strain AD169 in salivary gland-derived epithelial cells.

Valproic acid (VPA) has previously been shown to facilitate HCMV reactivation and entrance into the lytic cycle in latently infected monocytes (37). Due to its robust histone deacetylase (HDAC) activity, VPA is believed to relieve the repression of immediate early (IE) gene expression, promoting a transition to the lytic phase. HDAC activity and its regulation of the acetylation/deacetylation of the IE promoter is an important switch governing the entry into the lytic phase in cell types such as fibroblasts or entry into a latent/persistent phase in cell types such as hematopoietic progenitor cells (HPCs) and monocytes. In cell types destined to enter the full lytic phase, trafficking of the tegument protein pp71 into the nucleus interacts with cellular regulators such as Daxx to inhibit HDAC activity, thus enabling IE promoter acetylation and entry into the lytic phase (32).

In the course of characterizing our newly established primary human salivary gland-derived epithelial (hSGE) cell model, we reported that clinical isolates expressing pentamer, such as TB40E and MOLD, exhibited efficient infection and entry into the lytic phase (9). As expected, laboratory isolates such as AD169, lacking the pentamer, failed to establish a lytic infection and demonstrated little to no production of IE gene products. This effect was observed in cells derived from either parotid or submandibular glands. Interestingly, we also observed that while the salivary cells infected with TB40E or MOLD supported entry into the lytic phase and produced infectious virions, the virus failed to robustly spread through the salivary epithelial monolayer as efficiently as it does in fibroblasts. The failure of HCMV to robustly spread throughout the salivary epithelial cells led us to hypothesize that perhaps the IE promoter was not fully activated and that lower levels of IE gene expression may contribute to inefficient viral spread in these cells. We therefore investigated the effect of the HDAC inhibitor VPA on HCMV’s ability to infect and initiate IE gene expression in our primary hSGE cell model. To this end, we infected cells derived from parotid salivary glands at a multiplicity of infection (MOI) of 2 (i.e., 2 PFU per cell quantitatively determined by determining the titer of viral stocks on human foreskin fibroblasts [HFFs]), with or without VPA added at the time of infection. At 2 days postinfection, we harvested cells and analyzed IE expression by flow cytometry. Surprisingly, the epithelial-“defective” strain AD169 showed a significant enhancement of cells in which virus had entered the lytic cycle (4% versus 27%) (Fig. 1). The epitheliotropic clinical isolate MOLD and the intermediately epitheliotropic isolate FIX showed a noticeable, but far less dramatic, enhancement of cells with virus entering the lytic cycle. Quantification of the fold enhancement demonstrated that AD169 exhibited a strong 8.7-fold increase in IE-positive cells upon treatment with VPA compared to 1.4-fold and 2.1-fold increases for MOLD and FIX, respectively. Examining the literature, we discovered that another group had found a similar enhancing effect when using VPA or another HDAC inhibitor, trichostatin A (TSA), to pretreat primary retinal pigment epithelial (RPE) cells (38). Moreover, other experiments using the fusogen polyethylene glycol indicated that AD169 infection could be rescued by forcing entry into epithelial cells (8). While the pentamer (which is absent in AD169) is clearly important for viral entry, these results suggested that the pentamer may have additional roles that are broader than simply facilitating viral entry into epithelial cells, as the HDAC inhibitor, which presumably would function several steps distal to viral entry, is able to overcome the pentamer deficiency in hSGE cells. We therefore set out to explore the mechanisms underlying how HDAC inhibitors rescue AD169 and other pentamer-deficient strains for epithelial cell infection and perhaps glean additional information regarding why AD169 fails to efficiently establish lytic replication in epithelial cells like those in the salivary gland.

FIG 1.

The HDAC inhibitor valproic acid rescues infection by the epithelial-defective strain AD169 in salivary gland-derived epithelial cells. Primary human salivary gland-derived epithelial cells from 2 to 3 different human parotid cultures were infected at an MOI of 2 with different strains of HCMV. Cells were left untreated or treated with 1 mM valproic acid (VPA) at the time of infection. At 2 days postinfection, cells were analyzed for the accumulation of the IE1/2 viral proteins by flow cytometry. *, P ≤ 0.05 as determined by Student’s unpaired t test. The results are representative of data from 4 to 5 independent experiments performed in 2 to 3 different primary salivary epithelial lines. The average fold increase by VPA is displayed above each virus. ns, not significant.

Chemically distinct classes of HDAC inhibitors recapitulate the rescue of AD169 in epithelial cells.

Previous studies have indicated that the epithelial-defective pentamer-null HCMV strain AD169 can still be internalized into epithelial cells using a mechanism independent of direct pentamer/receptor interaction on the plasma membrane (8). In this scenario, the internalized virus may ultimately be fated for lysosomal degradation, thereby failing to traffic to the nucleus and failing to enter the lytic phase. Related studies have also demonstrated that pentamer-positive HCMV will enter epithelial cells using a mechanism involving low-pH-dependent endocytosis (8, 13, 39). Since both VPA and TSA contain acidic moieties, we first investigated whether the AD169 rescue that we observed with VPA could be due to the acidification of endosomes forcing the release of virions from the endosomal compartment and not due to the actual HDAC-inhibitory activity exhibited by these two compounds. To test this, we investigated additional HDAC inhibitors (listed in Fig. 2D) to determine if they shared the ability to rescue AD169 entry into the lytic phase in our primary salivary epithelial cells. MC1568 contains the same hydroxamic acid moiety as that of TSA; however, unlike TSA, which is a paninhibitor of HDACs, MC1568 shows specificity for only the class II HDACs (40). Entinostat (also known as MS-275) belongs to the benzamide class of HDAC inhibitors, distinct from VPA, TSA, and MC1568. It does not contain any inherent acidic groups and shows dose dependence specificity for the class I HDACs HDAC1 and HDAC3 (41, 42). Primary hSGE cells from either submandibular or parotid glands were then treated with these additional inhibitors, and we repeated the experiments in Fig. 1 by analyzing the entry of AD169 into the lytic phase in cells treated with or without the inhibitors (Fig. 2A and B). As described above, VPA mediated a significant increase in cells expressing IE1/2 albeit with slightly reduced efficacy in parotid epithelial cells in comparison to submandibular epithelial cells. Most notably, both TSA and entinostat strongly stimulated AD169 entry into the lytic phase as measured by IE expression. In contrast, the class II HDAC inhibitor MC1568 was unable to significantly enhance IE expression except mildly in the case of parotid gland-derived cells. Taken together, given that the rescue of AD169 entry into the lytic phase can be mediated by multiple classes of HDAC inhibitors regardless of the presence of acidic moieties that could potentially acidify endosomes, these data suggest that the AD169 rescue observed is likely mediated by the authentic HDAC activity of the drugs rather than an unrelated off-target effect. Moreover, since the class I-selective inhibitor entinostat is able to strongly rescue AD169 lytic-phase entry, while the class II-selective inhibitor MC1568 does not, our data suggest that the inhibition of class I HDACs such as HDAC1 and HDAC3 may be responsible for the AD169 rescue in the hSGE cells.

FIG 2.

Chemically distinct HDAC inhibitors reveal that the rescue of AD169 replication in primary salivary and established epithelial cell lines is due to inhibition of class I HDACs. (A to C) Epithelial cells derived from 4 independent human submandibular salivary glands (A) or 4 independent parotid salivary glands (B) or cells of the ARPE-19 cell line (C) were infected with AD169 at an MOI of 2. At the time of infection, cells were treated with DMSO (vehicle) (1%), VPA (1 mM), TSA (1 μM), MC1568 (10 μM), or entinostat (10 μM). Cells were analyzed for the accumulation of the IE1/2 protein by flow cytometry. (D) Specificities of each of the HDAC inhibitors for the various HDAC classes. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are representative of data from 3 to 5 independent experiments performed in duplicate.

We next wanted to determine if the rescue of AD169 entry into the lytic phase in epithelial cells was specific to our salivary cells or indicative of a more generalized epithelial phenotype. We tested the ability of these HDAC inhibitors to rescue AD169 in ARPE-19 cells as they are a commonly utilized epithelial model for HCMV infection. We repeated the experiments described above and found a similar response. Most notably, the class I inhibitors TSA and entinostat both strongly induced entry into the lytic phase, while the class II inhibitor MC1568 failed to significantly enhance AD169’s ability to initiate IE gene expression (Fig. 2C). Taken together with our experiments in primary salivary epithelial cells, our observations that HDAC inhibition rescued AD169 infection in both primary cells and established cell lines indicate that the effect of HDAC inhibition and promotion of the lytic phase by epithelial-defective strains such as AD169 is a generalized effect exhibited by epithelial cells.

The HDAC inhibitors TSA and entinostat exhibit enhancement of lytic gene expression following epithelial infection with a broad range of HCMV strains.

We infected hSGE cells derived from submandibular tissue with TB40E, FIX, and the low-passage-number clinical isolate MOLD to evaluate how the most potent of the HDAC inhibitors, TSA and entinostat, affected different strains of HCMV in our primary cell system (Fig. 3A to C). FIX is poor in infecting hSGE cells (likely due to the notable serine-to-proline mutation in pentamer), while TB40E maintains moderate hSGE infectivity, and MOLD exhibits strong hSGE infectivity (43). Both FIX and TB40E exhibited significant increases in cells expressing IE with treatment with TSA or entinostat (Fig. 3A and B). We further analyzed the fold increase over the vehicle control and found that the degree of enhancement correlates with how epitheliotropic each strain is, with weakly epitheliotropic strains such as FIX exhibiting a >10-fold enhancement with HDAC inhibitors but strongly epitheliotropic strains such as MOLD exhibiting a <2-fold enhancement with HDAC inhibitors (Fig. 3D).

FIG 3.

Clinical and laboratory isolates of HCMV with various epithelial tropisms all exhibit enhanced lytic replication in epithelial cells treated with HDAC inhibitors. (A to C) Submandibular gland-derived hSGE cells were infected at an MOI of 2 with strains displaying different inherent abilities to infect and replicate in epithelial cells (MOLD > TB40E > FIX). Cells were analyzed at 2 days postinfection by flow cytometry for the accumulation of IE proteins or the expression of green fluorescent protein (GFP). (D) Data from panels A to C were further analyzed by comparing the fold changes of TSA- or entinostat-treated cells to the DMSO control. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are derived from 3 independent experiments performed in duplicate.

As MOLD maintains high infectivity of epithelial cells (Fig. 3C and D), we wanted to more closely examine the effects of HDAC inhibitor enhancement to determine if MOLD’s high baseline level of epithelial infection masks the potent enhancing effect that we observe with HDAC inhibition and AD169 infection. At the same time, we also wanted to know whether these inhibitors were capable of enhancing fibroblast infections where pentamer is not required. Therefore, we infected either HS68 HFFs or ARPE cells at a relatively low MOI of 0.2 using the AD169, TB40E, or MOLD strain of HCMV and treated them with the vehicle control, TSA, or entinostat at the time of infection. In the case of HS68 fibroblast infection, while there is a general slight enhancing trend with the HDAC inhibitors, this enhancement was between 1.3-fold and 1.5-fold and was not statistically significant (Fig. 4A and B, left). In contrast, we saw that the inhibitors significantly enhanced each strain in ARPE cells when the cells were infected at a relatively low MOI of 0.2 (Fig. 4A, right). Given that each strain exhibited enhancement following HDAC inhibition, we wanted to carefully examine the magnitude of enhancement as this might reveal mechanistic insight regarding the mode of enhancement. Therefore, we calculated the fold change under either TSA or entinostat treatment conditions compared to the dimethyl sulfoxide (DMSO) controls (Fig. 4B, right). In this way, we see that the weakly epitheliotropic AD169 strain displayed the most drastic enhancement, being >25-fold (28- to 35-fold) over the vehicle control. In contrast, the strongly epitheliotropic strain MOLD displayed a slight level of enhancement, being <5-fold (4- to 4.5-fold) over the vehicle control. Additionally, given that HS68 and ARPE cells were infected side by side, with the same virus stocks and at the same MOI, but result in drastically different observable effects with the HDAC inhibitors, these findings underscore the fact that these strains contain distinctly different tropism factors, which significantly define the outcome of lytic-phase entrance during epithelial infection.

FIG 4.

HDAC inhibitors potently enhance AD169 initiation into the lytic cycle in epithelial cells but not fibroblasts. (A) HS68 or ARPE cells were infected at an MOI of 0.2 with HCMV strain AD169, TB40E, or MOLD and then dosed with either DMSO, TSA, or entinostat at the time of infection. At 2 days postinfection, fixed cells were analyzed by flow cytometry for the proportion of cells that were IE positive. (B) The fold changes of TSA- or entinostat-treated cells from panel A were compared to the DMSO control. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are representative of data from 3 independent experiments.

HDAC inhibitors overcome the lack of pentamer during epithelial infection.

The magnitude and ability of HDAC inhibitors to enhance HCMV infection in epithelial cells in our experiments are inversely correlated with how inherently epitheliotropic each strain is. Weakly epitheliotropic strains are strongly enhanced by HDAC inhibitors, while highly epitheliotropic strains are weakly enhanced by HDAC inhibitors. The most prominent viral factor involved in epithelial tropism is the pentamer (6–8). AD169 has been well characterized for its inability to infect epithelial cells due to a frameshift mutation in UL131, a component of the pentamer, resulting in the functional loss of pentamer expression (7, 8). FIX contains a substitution in the UL130 gene resulting in a serine-to-proline change, affecting the total level of pentamer incorporation into virions and, potentially, pentamer function (43). TB40E has a point mutation in intron 1 of the UL128 gene, which results in inefficient splicing and reduced pentamer production (43). No sequences of MOLD are available, but given its propensity for epithelial infection, we predict not only that it has no loss-of-function mutations in the pentamer locus but also that virions of MOLD likely have fairly high levels of pentamer, facilitating successful epithelial infection, although other factors cannot be excluded.

Since our data reveal that the degree of enhancement due to HDAC inhibition is inversely correlated with pentamer expression in that pentamer-minus strains exhibit strong enhancement by HDAC inhibition (∼25-fold), while pentamer-plus strains exhibit only weak enhancement by HDAC inhibition (∼3- to 5-fold), we therefore sought to directly investigate this link using a genetically tractable system to modulate pentamer gene expression in an otherwise genetically identical virus strain. We utilized a recombinant strain of Merlin, which has a tetracycline-repressive element upstream of the UL131 gene (44). When virus stocks are prepared in a fibroblast line (Tet-Tert) that expresses the Tet repressor protein, the UL131 gene is repressed, and progeny viruses are assembled without pentamer but retain the expression of the trimer glycoprotein. When virus stocks are prepared in normal fibroblasts, the UL131 genes are derepressed, and the resulting progeny are highly enriched in pentamer, although they have been shown to have decreased trimer levels, potentially resulting in a loss of infectivity in all cell types (21). We produced virus stocks of this Merlin recombinant under repressive conditions in the HFF-tet-tert cell line to generate a pentamer-negative population and harvested virions when approximately 60 to 80% of the cells displayed cytopathic effects (CPE). We postulated that if we relieved repression by the addition of doxycycline when most of the cells were infected and virions were still under assembly, the expression of pentamer might result in virions that contain suitable levels of both pentamer and trimer. We then used these Merlin stocks (minus or plus pentamer) to infect ARPE cells and again treated cells with the vehicle control, TSA, or entinostat. As expected, under repressive conditions, the pentamer-negative virions display very low epithelial tropism but can be highly enhanced upon HDAC inhibitor treatment (Fig. 5A and B). In contrast, as predicted, the unrepressed pentamer-positive virions are highly epitheliotropic (Fig. 5A and B). We examined the fold change for the inhibitors compared to the vehicle control and found that virions generated under the repressive conditions without pentamer expression displayed the most dramatic enhancement (9.3- to 19-fold), while virions that were highly epitheliotropic as a result of pentamer expression displayed modest enhancing effects (1.4- to 3.3-fold) (Fig. 5C).

FIG 5.

HDAC inhibitors overcome the lack of pentamer for infection of ARPE epithelial cells. (A and B) ARPE cells were infected at an MOI of 0.2 or 2 with Merlin (TetO-UL131) virus stocks prepared under repressive (UL128-131-negative) or nonrepressive (UL128-131-positive) conditions. At the time of infection, cells were treated with DMSO (vehicle) (1%), TSA (1 μM), or entinostat (10 μM). Cells were analyzed at 2 days postinfection by flow cytometry for the accumulation of IE proteins. (C) Data from panels A and B were further analyzed by comparing the fold changes of TSA- or entinostat-treated cells to the DMSO control. (D and E) ARPE cells were infected as described above for panels A and B but using the pentamer-negative BADwt strain or the pentamer-containing UL131-repaired BADrUL131 mutant. (F) Data from panels D and E were further analyzed by comparing the fold changes of TSA- or entinostat-treated cells to the DMSO control. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are derived from 3 independent experiments performed in duplicate.

We also examined the parallel roles of pentamer and HDAC inhibition utilizing a mutant of AD169 where the UL131 frameshift mutation has been repaired (BADrUL131). For these experiments, the corresponding AD169 bacterial artificial chromosome (BAC) clone from which this repair mutant was derived (BADwt) was used as a control. This particular line of investigation also controls for the lack of the ULb′ region as the AD169 genome does not contain ULb′ and thus enables us to interpret the results of pentamer and/or HDAC rescue of epithelial infection using a virus that lacks the other viral proteins expressed from the ULb′ region (45, 46). We repeated the experiments in the same way as we had performed with the Merlin viruses. Similar to our above-described experiments, we found that BADwt was weakly epitheliotropic and that TSA and entinostat treatment leads to a strong enhancement of pentamer-null BADwt virus entrance into the lytic phase, while the pentamer-positive virus was highly epitheliotropic and exhibited only a modest enhancement with the HDAC inhibitors (Fig. 5D and E). Examination of the fold change for the inhibitors revealed that the strong enhancement of pentamer-null BADwt by HDAC inhibition was 8- to 14-fold, while the modest enhancement of pentamer-positive BADrUL131 by HDAC inhibition was 1.5- to 3.0-fold.

Taken together, these data demonstrate that while HDAC inhibitors can enhance all strains of HCMV for epithelial infection, they are most proficient at enhancing the initiation of lytic replication for strains that have lost the expression of pentamer (AD169 and Merlin UL128-131 negative). These HDAC inhibitors also failed to enhance infection in fibroblasts, where pentamer is not required for infection. It has been reported previously by others that at least 24 h of pretreatment using VPA before infection of fibroblasts results in a slight enhancement of lytic replication; however, their experimental design differs from ours as they pretreated cells for at least 24 h, while we added the inhibitors at the time of infection (47). Altogether, the enhancing effect of HDAC inhibitors on epithelial infection and their ability to strongly enhance and rescue pentamer-null virus strains suggest that the pentamer plays an extended role in epithelial infection corresponding to the inhibition of HDAC activity subsequent to the initial steps of infection.

Infection of primary hSGE cells similarly demonstrates a requirement for pentamer, which can be overcome by HDAC inhibition.

We wanted to confirm that the strong ability of HDAC inhibitors to enhance epithelial cell lytic infection by pentamer-null viruses was not restricted to the ARPE cell line, so we once again utilized our primary salivary cell system. We infected hSGE cells derived from parotid or submandibular glands at MOIs of 0.2 and 2 using both pentamer-positive and -negative Merlin virions and then treated the cells with the same HDAC inhibitors or the vehicle control as in Fig. 5. We analyzed these cells in the same manner, and consistent with our results in the ARPE model, both TSA and entinostat greatly restored the pentamer-negative virus’s ability to initiate lytic replication (Fig. 6A and B). We again further calculated the fold change for each HDAC inhibitor’s ability to enhance the pentamer-negative virions relative to the vehicle control and found an ∼13- to 60-fold enhancing effect of the HDAC inhibitors (Fig. 6C). In contrast to what we observed in the ARPE model, pentamer-containing virions demonstrated little to no increase in active lytic replication upon HDAC inhibitor treatment. Taken together with the data in Fig. 5, these results demonstrate that HDAC inhibitors act most profoundly by rescuing a defect specific to pentamer-null viruses.

FIG 6.

Pentamer is required for primary salivary infection, and pentamer-null virus can be rescued by HDAC inhibition. (A and B) Primary salivary cells (derived from 1 submandibular and 1 parotid gland) were infected with Merlin (TetO-UL131) virus prepared under repressive (UL128-131-negative) or nonrepressive (UL128-131-positive) conditions at an MOI of 0.2 or 2. Cells were treated with DMSO, TSA, or entinostat at the time of infection and then analyzed for the accumulation of IE protein at 2 days postinfection by flow cytometry. (C) Data from panels A and B were further analyzed by comparing the fold changes in infection after TSA or entinostat treatment compared to the DMSO control. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are derived from 3 independent experiments. dox, doxycycline.

HDAC inhibition rescues the ability of AD169 to progress through early and late stages and leads to viral envelopment.

Based on the findings that HDAC inhibitors can strongly rescue AD169’s ability to initiate lytic infection as determined by the expression of the IE1/2 proteins, we next asked if this rescue was limited to IE expression or if AD169 could proceed through other stages of the lytic phase marked by the expression of early and late proteins. We infected ARPE cells with TB40E as a control or AD169 with and without TSA treatment. At 24 h postinfection, cells were fed with fresh medium to remove excess TSA. Cells were harvested at 1, 4, or 7 days postinfection and probed by Western blotting for markers of IE (IE1/2), early (UL44), or late (pp65) viral proteins, with actin as a cellular loading control. After TSA treatment, AD169 proceeds into early and late protein production similarly to TB40E, with increased UL44 and pp65 protein accumulation over time (Fig. 7A). This indicates that TSA restores AD169’s ability to undergo lytic replication up until the late stages and possibly to full virion production.

FIG 7.

HDAC inhibition restores early and late viral gene expression and results in increased virus production following HCMV AD169 infection. (A) Representative images from 4 independent experiments where ARPE cells were infected with TB40E or AD169 with (+) or without (−) TSA treatment at an MOI of 2. Cell extracts were harvested at the indicated time points and probed for anti-CMV IE1/2, UL44, or pp65 as surrogate markers of IE, early, and late protein production. β-Actin was included as a loading control. (B) ARPE cells were infected at an MOI of 2 using either TB40E or AD169 with DMSO, TSA, or entinostat added at the time of infection. The supernatants were harvested at the indicated days postinfection and used to infect HS68 HFFs to determine viral titers. After 2 days of infection, HS68 cells were analyzed by flow cytometry using an anti-IE1/2 antibody to determine the proportion of cells infected, which was used to calculate the viral titer.

To determine if fully infectious AD169 viral particles were produced following HDAC inhibition, we infected ARPE cells with TB40E or AD169 at a high MOI of 2 and then treated cells with DMSO, TSA, or entinostat. Cells were washed the next day to remove residual drug and virus, and the supernatant was then harvested over a 14-day time course. The supernatant was used to infect human HS68 fibroblasts for viral titer determination. TB40E was used as a control as it maintains the ability to replicate in epithelial cells (Fig. 7B, left). Interestingly, and consistent with the early and late protein expression data, TSA and entinostat each enhanced the amount of AD169 virus produced from these cells by up to 10-fold over the time course (Fig. 7B, right). These data reveal that treatment with HDAC inhibitors not only allows pentamer-deficient AD169 to enter into the lytic cycle but also allows the virus to proceed efficiently through all stages of lytic replication, including the assembly and release of progeny virus. Consistent with other data, TB40E showed increased viral yields as a result of treatment with HDAC inhibitors (Fig. 7B, left), but the degree of enhancement was not as potent as in the case of AD169, which is consistent with previous data.

HDAC inhibitors do not increase the entry of AD169 into epithelial cells or facilitate virion trafficking to the nucleus.

AD169’s inability to infect and initiate lytic replication in epithelial cells has been attributed to a defect in the ability of the virus to escape from endocytic vesicles (8). If the lack of the pentamer in AD169 results in the virus being unable to exit endocytic vesicles, then under normal conditions, we would expect to find less viral DNA in the nucleus than in the whole cell. However, since many HDACs canonically function within the nucleus, it seems plausible that the HDAC inhibitors in this circumstance are mediating a postentry rescue such that we would see similar levels of AD169 viral DNA in the nucleus in the absence or presence of the inhibitor. We therefore sought to directly determine if these inhibitors could be mediating enhanced entry of the virus into the nucleus or if they might truly be acting at a postentry step.

To determine if viral DNA from AD169 was able to traffic to and accumulate in the nucleus, we utilized a nuclear isolation assay. We validated our assay by blotting cytoplasmic and nuclear fractions and confirmed that clean fractionation occurred (Fig. 8A), as proteins found abundantly in the cytoplasm (β-actin and the endosomal marker EEA1) and nucleus (lamin A/C) were properly localized to their predicted fraction. Cells treated with the vehicle control or entinostat were then infected with AD169 and fractionated, and DNA was isolated. Each sample was split into 2, with one set undergoing purification of DNA from unfractionated whole cells and the other set undergoing purification of DNA from the isolated nuclei. We then performed quantitative PCR (qPCR) to quantify viral genome copies in either unfractionated whole cells or isolated nuclei. Viral genome copies were determined by the presence of the viral gene UL54, normalized against a host gene (RPPH1) to determine how many copies of the viral genome had entered the cells and trafficked to the nucleus at the time point of 48 h postinfection. In vehicle-treated AD169-infected cells, virtually all the viral DNA was located in the nucleus (Fig. 8B). In entinostat-treated cells, we similarly observed that the AD169 viral DNA was able to localize to the nucleus, and the amounts of nucleus-localized viral DNA were comparable between the vehicle control- and entinostat-treated cells. These data indicate that entinostat treatment does not lead to enhanced trafficking of AD169 viral DNA to the nucleus and suggest that the effect of entinostat on promoting entry into the lytic phase is subsequent to viral DNA localizing to the nucleus, consistent with its known role as an HDAC inhibitor.

FIG 8.

HDAC inhibitors do not increase entry or nuclear trafficking of AD169 in epithelial cells. (A) Representative Western blot of 3 independent experiments utilizing nuclear purification protocols. Twenty micrograms of the protein extract was loaded onto a Tris-glycine gel, transferred to nitrocellulose, and probed for markers of cytosolic proteins (actin and EEA1, an endosomal marker) or nuclear lamin A/C. WC, whole cell; Cy, cytoplasm; Nu, nucleus. (B and C) ARPE cells were infected at an MOI of 2 using AD169 with or without the specified inhibitors. At 2 days postinfection, cells were harvested, and DNA was purified from either whole cells or isolated nuclei. Twenty nanograms of DNA from each sample was used for qPCR using primers against the viral UL54 locus (Viral Pol.) or the host RPPH1 gene (Cell). Data are presented as the copy numbers of viral genomes relative to 2 copies of host RPPH1 from the same sample over 4 independent experiments. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle controls. **, P ≤ 0.01; *, P ≤ 0.05.

The escape of HCMV from endosomes requires endosomal acidification in epithelial infections. Lysosomotropic agents that block this acidification greatly inhibit HCMV infection of epithelial cells (39). To ensure that entinostat is functioning at a step subsequent to release from endosomes and trafficking to the nucleus, we used bafilomycin A1 (BafA1), a bacterial toxin that selectively inhibits the vacuolar ATPases, which mediate endosomal acidification (48). BafA1 treatment prior to infection significantly inhibited AD169 trafficking to the nucleus, confirming that AD169 in our experiments is indeed entering ARPE cells via an endosomal pathway involving acidification for release (Fig. 8C). Treatment with entinostat did not overcome the effects of BafA1, again supporting the conclusion that HDAC inhibition is functioning at a step after entry to facilitate HCMV lytic-phase entry (Fig. 8C).

A small portion of the samples from these nuclear isolation assays was saved for subsequent staining for the presence of IE by flow cytometry analysis to compare IE levels to viral DNA levels in the nucleus (Fig. 9A). In this way, we can confirm that BafA1 not only blocks the endosomal entry of AD169 into ARPE cells but also blocks entinostat rescue of the lytic phase. As expected, entinostat increased the percentage of ARPE cells harboring a lytic infection from 5 to 33% (comparing the vehicle to entinostat), while preventing the acidification of endosomes with BafA1 completely abrogated this effect (Fig. 9A). We performed similar experiments in HS68 fibroblasts and demonstrated that neither entinostat nor BafA1 had much of an effect on AD169 entry into the lytic phase (Fig. 9B). These experiments highlight the differences between viral entry and progression to the lytic phase in fibroblasts and epithelial cells. In fibroblasts, the virus enters via a pathway independent of endosome acidification and initiates lytic gene expression using viral mechanisms independent of pentamer. In contrast, in epithelial cells like ARPE cells, the virus enters via an endosomal pathway involving the acidification of the endosomes and initiates lytic gene expression using a novel mechanism involving pentamer and HDAC inhibition, which can be overcome by pharmacological inhibition of HDACs in the case of pentamer-minus virus strains like AD169.

FIG 9.

HCMV AD169 lytic infection in epithelial cells requires the release of virions from endocytic vesicles and is distinct from AD169 lytic infection in fibroblasts. (A) Cells from nuclear isolation experiments in Fig. 5 were used for flow cytometry-based detection of viral IE products to confirm and compare active infections relative to genomic copy number assays. (B) HS68 cells were infected with AD169 and underwent treatments similar to the ones described above for panel A. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are derived from 3 to 4 independent experiments performed in duplicate.

Kinetics of HDAC inhibition gives insight into the mechanism of AD169 rescue.

To gain additional insight into the mechanism underlying HDAC inhibition and the rescue of AD169 infection and the initiation of lytic replication in epithelial cells, we examined the timing of drug addition to determine when it was predominantly acting during infection. In previous experiments, we added the HDAC inhibitor at the time of infection; however, we posited that if the effect of the HDAC inhibitor was to prevent the deacetylation of HCMV IE promoters, the rescuing effect of HDAC inhibition would be present if we added the inhibitors for a period of time postinfection. To start, we infected ARPE-19 cells with AD169 or MOLD and then added TSA at the time of infection or 2, 7, 10, 24, or 36 h postinfection. AD169 could be rescued when the drug was added up to 10 h postinfection; however, by 24 h, TSA was unable to overcome AD169’s defect (Fig. 10A). Similarly, TSA began to show a decrease in its enhancement of MOLD infection starting at 24 h postinfection, which was entirely lost by 36 h postinfection (Fig. 10B). Similar results were seen when using entinostat (data not shown). These results indicate that HDAC inhibition must be in place during the IE phase after infection, or the inhibitors do not result in the rescue of AD169 lytic replication. Moreover, these results argue against a model whereby HDAC inhibition results in the upregulation of viral receptors or vesicle trafficking regulators rescuing infection by pentamer-null viruses. We have also examined the expression of the HCMV trimer receptor PDGFRα and several markers of intracellular trafficking such as Rabs 5, 7, and 11 and have not observed a global upregulation of genes that may allow infection via a trimer-mediated route (data not shown). Taken together, these results suggest that inhibition of histone deacetylation during the IE phase after HCMV infection is critical for the initiation of the lytic phase in epithelial cells and that this inhibition of HDAC activity can be accomplished by the presence of pharmacological HDAC inhibitors or pentamer.

FIG 10.

Kinetic analysis indicates that HDAC inhibition must be intact during the IE phase (0 to 10 h postinfection) to rescue AD169 replication. (A and B) ARPE cells were infected using AD169 (A) or MOLD (B) at an MOI of 2. Cells were either left untreated or given DMSO (vehicle) or TSA at the indicated times postinfection. After 48 h postinfection (hpi), cells were analyzed for the accumulation of IE1/2 proteins by flow cytometry. (C) Schematic of the time course presented in panels A and B. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are representative of data from 3 to 4 experiments.

Our experiments presented thus far have focused on the treatment of cells with HDAC inhibitors at the time of infection or added at 2 to 36 h postinfection. Other work has shown that these inhibitors had an effect when treating the cells for several days before infection (38). We therefore wanted to understand to what extent the HDAC inhibitors affect HCMV entry into the lytic phase when added prior to infection and removed at the time of infection (see Fig. 11B for a diagram demonstrating when the HDAC inhibitors were added). Moreover, we wanted to determine how pretreatment with HDAC inhibitors affected H3-K9 acetylation within the treated cells. Cells were pretreated with entinostat or the vehicle 2 days before infection or pretreated with entinostat 2 or 5 h before infection. Cells were washed once with Dulbecco’s phosphate-buffered saline (DPBS) to remove residual extracellular drug and then infected. One set of samples was treated with either DMSO or entinostat at the time of infection to facilitate comparison with the data from the experiments described above. Pretreatment of the cells for 2 days prior to infection allows the rescue of AD169 entry into the lytic phase at a level comparable to that when entinostat is added at the time of infection. Pretreatment with entinostat for 2 to 5 h resulted in an intermediate but significantly reduced level of lytic entry compared to the 2-day pretreatment (Fig. 11A). These data suggest that entinostat allows rescue after its removal if sufficient time is given for it to act on the cell and inhibit target HDACs.

FIG 11.

Rescue of HCMV lytic replication in epithelial cells correlates with increased H3 acetylation (Ac.). (A) ARPE cells were plated 48 h prior to infection with AD169 at an MOI of 2. Cells were either pretreated at the times indicated with DMSO or entinostat or left alone until infection where they received either DMSO or entinostat treatment. (B) Schematic of the time course presented in panel A. All samples were harvested at 2 days postinfection and analyzed by flow cytometry for the accumulation of viral IE protein. (C) ARPE cells were plated and treated with DMSO or entinostat at the indicated times prior to harvest of the cell lysate for Western analysis. (D) Semiquantification of Western blots from panel C to quantify the relative proportion of total H3 that is acetylated at the K9 position. Results were analyzed by unpaired Student’s t tests comparing each drug to the vehicle control. **, P ≤ 0.005; *, P ≤ 0.05. The results are derived from 3 independent experiments.

As these drugs are known to cause hyperacetylation of histones, we wanted to ascertain the timing of this hyperacetylation of histone H3 and how long this is maintained after treatment. Our previous data indicated that HDAC inhibition in the IE phase is required for the rescue of AD169 entry into the lytic phase; therefore, we investigated if HDAC inhibitor pretreatment resulted in residual inhibition of HDAC and hyperacetylation of histone H3. We treated ARPE cells in a fashion similar to that in the experiment described above in the absence of infection to determine how entinostat affected histone acetylation. Cellular lysates were probed by Western blotting using a pan-H3 antibody for total histone input or with an antibody that recognizes acetylated H3-K9. As expected, entinostat pretreatment for 24 to 48 h led to highly acetylated H3-K9 compared to the vehicle control, while pretreatment for 2 to 5 h led to an intermediate response but not nearly as strong as that in the 24- to 48-h samples (Fig. 11C). We quantified the intensity of each sample for H3-K9 acetylation and normalized it to the total H3 present (Fig. 11D). The amount of acetylation with increasing time with the drug on the cells correlates with the infectivity seen in pretreated cells in Fig. 11A. Taken together, both HDAC inhibition up to 10 h postinfection and HDAC inhibition prior to infection are capable of rescuing AD169 entry into the lytic phase. Therefore, these data suggest that pentamer-minus viruses can be rescued for entry into the lytic phase by ensuring high-level HDAC inhibition during the immediate early phase of infection.

The viral tegument protein pp71 fails to localize to the nucleus in epithelial cells infected with the epithelial-defective HCMV strain AD169.

The HCMV tegument protein pp71 is packaged with mature virions and plays multiple roles throughout infection. One important role for pp71 is to control the regulation of the major immediate early promoter (MIEP) by coordinating the proteasomal degradation of Daxx, a cellular protein that cooperates with a larger protein complex containing HDACs to silence DNA associated with promyelocytic leukemia nuclear bodies (PML-NBs) (32). In this scenario, pp71 functions to prevent HDACs from silencing the MIEP, thus favoring lytic gene expression. In fibroblasts or differentiated myeloid cells that proceed directly to the lytic phase, pp71 localizes to the nucleus following infection, while in undifferentiated hematopoietic cells destined to enter a latent or nonlytic state, pp71 localizes to the cytosolic face of the nucleus and fails to promote MIEP activity (36). Since we have shown that AD169 DNA enters the nucleus but fails to efficiently enter the IE phase and drive MIEP activity in the absence of pharmacological intervention with HDAC inhibitors, we postulated that pp71 may not properly traffic to the nucleus in AD169-infected cells, thus failing to drive lytic gene expression.

We infected ARPE cells with AD169, TB40E, or MOLD and analyzed pp71 subcellular localization at 24 h postinfection by immunofluorescence. Cells were fixed and permeabilized prior to pp71 staining and counterstained with 4′,6-diamidino-2-phenylindole (DAPI) to visualize nuclei. MOLD and TB40E showed localization of pp71 almost exclusively to the nucleus, consistent with the ability of these viral strains to efficiently initiate lytic-phase gene expression (Fig. 12). Infection with AD169 demonstrated a punctate pattern of pp71 in proximity to but not within the nucleus after 24 h of infection. Thus, AD169-delivered pp71 fails to localize to the nucleus where it could function as an HDAC inhibitor. Taken with our nuclear DNA assay (Fig. 8), this demonstrates that AD169 DNA is competent for translocation to the nucleus, but pp71 fails to translocate. This suggests that the pentamer has a role in mediating a pathway that ensures pp71 localization to the nucleus, although it cannot be formally excluded that these punctae are viral particles that are still trapped inside endosomal vesicles and that our assay is not sensitive enough to detect low levels of pp71 in the nucleus for virions that have fully entered.

FIG 12.

HCMV AD169 infection in epithelial cells fails to induce pp71 localization to the nucleus. ARPE-19 cells were infected with AD169, MOLD, or TB40E and stained for pp71 at 24 h postinfection. Cells were counterstained with DAPI to visualize nuclei. Arrows indicate the location of pp71 in the cell. The results are representative of data from 3 to 4 independent experiments, each with similar results.

While we were confident that the HDAC inhibitors are not affecting viral entry, we wanted to confirm that they had no effect on these pp71 punctae that we had observed. We repeated our immunofluorescence microscopy experiment using AD169 and treating cells at the time of infection with the DMSO control or entinostat. When we again image the cells at 24 h postinfection, we see that there are no discernible changes in the amount or localization of pp71 punctae in infected cells (Fig. 13), although some nuclei do display very high levels of pp71, likely as a result of entinostat increasing the early expression of the pp71 gene, consistent with Western blot data where HDAC inhibitors potently increase viral protein expression and cycloheximide treatment prevents nascent pp71 expression (Fig. 7A and data not shown). Taken together, these results indicate that HDAC inhibition does not lead to nuclear trafficking of pp71 and is functioning to relieve HDAC activity at the level of viral promoters.

FIG 13.

Entinostat overrides the requirement for pp71 movement into the nucleus prior to the initiation of lytic replication. ARPE-19 cells were infected with AD169 and then fixed and stained at 24 h postinfection for pp71 and IE. Nuclei were counterstained using DAPI. Arrows indicate punctae containing pp71. The results are representative of data from 3 independent experiments.

DISCUSSION

Efficient HCMV infection of epithelial cells requires the presence of the pentamer glycoprotein complex consisting of gH/gL/UL128-131 (6, 7). The long-term passage of “laboratory-adapted” strains in fibroblasts leads to a loss of the pentamer complex due to the increased fitness of strains containing the gH/gL/gO trimer complex, which is used for viral entry into fibroblasts. Pentamer-null strains have been “rescued” for their ability to infect epithelial cells either by the use of polyethylene glycol to force the fusion of the viral envelope with cellular plasma membranes or by the reconstitution/repair of the UL128-131 locus in the genomes of pentamer-null strains (7, 8, 21). Our studies reveal, to our knowledge, the first pharmacological agent capable of rescuing pentamer-null strains for epithelial infection. Thus, pharmacological inhibition of histone deacetylases provides a new tool for characterizing the role of the pentamer and further understanding mechanistically how it functions to establish infection in various epithelial cells. Additionally, there is clinical significance to these findings, as HDAC inhibitors are currently in use or being investigated for use as therapeutics to treat diseases such as epilepsy or cancer, and their use in these patients may lead to undesirable side effects via their ability to enhance HCMV replication or potentially other DNA viruses (47, 49).

To successfully initiate lytic replication in an epithelial cell, a particle of HCMV must overcome at least 2 major barriers: (i) it must gain access to the cytoplasm so that capsid can uncoat and viral DNA can translocate to the nucleus, and (ii) the viral MIEP must become activated to allow the expression of immediate early genes, which initiates the subsequent cascade of viral protein expression and produces new viral particles.

With regard to the first point, the pentamer has routinely been described as an entry mediator, due in part to the ability of polyethylene glycol to rescue pentamer-null strains in epithelial infection (8). The exact mechanism as to how pentamer mediates entry remains unclear. Early studies indicated that it was not required for internalization into the cell, and fusion assays designated gH/gL and gB the minimal viral proteins required for membrane fusion (8, 50). Moreover, it has been speculated by Zhou and colleagues that the gH/gL/gO trimer is chiefly responsible for fusion based on studies in which a Merlin mutant that expresses low levels of trimer displayed broad defects for infection in all cell types, a defect which was reversed upon treatment with polyethylene glycol (21). However, a recent genetic screen identified CD147 as playing an important role in HCMV infection of epithelial and endothelial cells, and these studies revealed that both the pentamer and gB were competent for mediating fusion with HeLa cells expressing CD147, suggesting that pentamer could, in fact, be mediating fusion using CD147 as a coreceptor or via some other indirect mechanism, as the pentamer does not appear to directly interact with CD147 itself (20). Taken together, whether pentamer can truly function as a fusion mediator and its precise role as an entry mediator remain unclear.

By examining viral DNA accumulation in the nucleus (Fig. 7B), we found that AD169 is competent for nuclear entry to some extent in epithelial cells although likely not as efficiently as pentamer-containing strains. Moreover, the amounts of viral DNA accumulating in the nucleus are comparable between the vehicle- and entinostat-treated cells, yet the difference in IE is dramatically different such that while approximately one-third of the cells harbor viral DNA, they fail to initiate lytic replication unless entinostat is present (Fig. 9A). Furthermore, entinostat does not appear to be acting to mediate entry or fusion for virions trapped in endosomes as evidenced by entinostat being unable to overcome the block presented by BafA1 treatment (Fig. 7C). In these data, we see that BafA1 clearly reduces the trafficking of AD169 DNA to the nucleus and that entinostat cannot overcome this block and subsequently does not result in increased initiation of lytic replication. Validation of our fractionation approach (Fig. 8A) supports these conclusions and argues against the possibility that sedimentation of endocytic membranes containing viral DNA is what is being detected by the nuclear PCR in this assay, although we cannot formally exclude this possibility. Taken together, these data reveal that HDAC inhibitors are acting to rescue infection and lytic replication at a step after entry, as AD169 infection can translocate sufficient viral genomes to the nucleus but fails to initiate lytic replication once there.

Regarding the second point, after a virus has overcome the entry barrier and viral DNA has successfully translocated to the nucleus, it must then succeed in MIEP expression to initiate the lytic cascade. Viral DNA that makes it to the nucleus is typically sequestered in PML-NBs where it acquires histones and is targeted for genetic silencing via cellular Daxx-mediated recruitment of HDAC-containing complexes (30). HCMV overcomes this block in many cells via the tegument protein pp71, which prevents HDAC recruitment by mediating the degradation of Daxx, which then prevents subsequent HDAC recruitment (32). By immunofluorescence microscopy, we observe that pp71 accumulates as nonnuclear punctae in AD169-infected epithelial cells. Given that viral DNA can translocate to the nucleus (Fig. 7B) and that these cells subsequently enter into the lytic phase after HDAC inhibition (Fig. 8A), a plausible argument is that AD169 is inefficient at driving pp71 translocation to the nucleus, which results in the silencing of the viral DNA. Further studies examining these punctae to determine if they colocalize with viral capsids, early or late endosomes, or the trans-Golgi network may clarify if the pp71 protein is indeed free from the rest of the virion or part of intact virions that have failed to fully enter these cells but have not been degraded.

Taken together, our data are consistent with a model where pentamer-null strains can successfully traffic viral DNA to the nucleus yet fail to translocate pp71, thus being unable to overcome MIEP repression by cellular HDACs. Why this is the case may involve a unique internalization and intracellular trafficking pathway utilized by pentamer-null virions. A study that examined the role of PDGFRα overexpression during epithelial infection uncovered that virions were internalized differently depending on whether the virion had only trimer or both trimer and pentamer (51). In the case of trimer-only virions, they were internalized via dynamin-dependent endocytosis with subsequent entry via a low-pH-independent pathway; in contrast, pentamer-containing virions appeared to be internalized via macropinocytosis in a low-pH-dependent manner. Those studies contrasted with ours in that they used ARPE cells ectopically overexpressing PDGFRα, which is typically expressed at low levels in ARPE cells; however, that study underscores that internalization into epithelial cells can be driven by either trimer or pentamer, doing so by different mechanisms (22, 51). It should also be noted that those studies identified that trimer-only virions could enter PDGFRα-overexpressing cells in a low-pH-independent manner, while in our study, the trimer-only virions appear to utilize a low-pH-dependent mechanism of entry as revealed by BafA1 inhibition. This may suggest that in our studies, trimer-only virions are internalized by a third pathway that may be largely PDGFRα independent. It may be that each internalization pathway utilizes distinct intracellular trafficking networks whereby pentamer-mediated entry results in nuclear translocation and lytic gene expression, while trimer-mediated entry is capable of nuclear translocation only.

Consistently throughout this study, we have shown that HDAC inhibitors belonging to 3 chemically distinct classes result in a dramatic increase (20- to 30-fold) in the initiation of lytic replication for strains lacking pentamer. We also noted that strains that contain pentamer displayed low levels of enhanced initiation of the lytic cycle upon HDAC inhibitor treatment. At first, this may seem counter to the argument that HDAC inhibitors are overcoming the lack of pentamer as even pentamer-competent strains show some level of enhancement. While it is true that HDAC inhibition can show some enhancing effects on pentamer-positive strains, they are much less dramatic enhancements as demonstrated by the fold changes that we have calculated throughout this study. Furthermore, the degree of enhancement correlates strongly with each different strain’s tropism for epithelial cells; MOLD, which displays a high ability to infect epithelial cells, showed the lowest enhancing effect, only reaching as high as a 2- to 4-fold increase in our studies, while strains less epitheliotropic but still pentamer competent to various degrees (such as FIX and TB40E) displayed intermediate enhancing effects of 2- to 10-fold. Completely pentamer-null AD169 showed the highest fold change upon HDAC inhibitor treatment, being upwards of 20- to 35-fold depending on the MOI and inhibitor used. All of these points are nicely recapitulated in our experiments involving the Merlin (TetO-UL131) or the BADwt/BADrUL131 mutants where the pentamer-null virions display the most significant and highest fold changes when treated with HDAC inhibitors compared to their pentamer-positive counterparts.

The reason for these differences in HDAC enhancement between pentamer-null, pentamer-low, and pentamer-high strains is likely dependent on multiple factors. First, the degree of pentamer incorporation of different strains likely plays a role in the tropism of the nascent viral particle, and it has previously been shown that different strains incorporate different amounts of pentamer and trimer into their envelope (21, 52). How this leads to changes in tropism for different cell types is still unclear. It has been suggested that newly made HCMV stocks are comprised of a heterogeneous population of virions containing small or large amounts of pentamer (26). One study utilizing flow virometry showed that this may not be the case, however, as those researchers detected only one homogeneous population when examining viral particles for pentamer incorporation (53). It should be noted that their study examined only a UL131-repaired mutant of AD169 and clinical isolates but no other clonally derived strains such as TB40E, FIX, and Merlin, etc., which may be able to incorporate pentamer to different degrees. Additionally, all of their virus stocks tested were derived from ARPE-19 cells, and it has been speculated that progeny virus derived from epithelial/endothelial cells may be more homogeneous and contain less pentamer than fibroblast-derived virus, resulting in the lower epithelial/endothelial tropism observed than with HFF-derived virions (26). Alternatively, regardless of whether viral populations are heterogeneous or homogeneous, other factors, such as the sequence in which the trimer or pentamer binds to cognate receptors or the ratio of trimer-receptor- versus pentamer-receptor-bound complexes, might drive different internalization and intracellular trafficking pathways leading to different outcomes during epithelial infection.

The other confounding factor likely involved is the amount of pp71 incorporation into HCMV particles. A recent analysis demonstrated that nascent HCMV particles vary in how much pp71 is packaged into the tegument such that there are particles with small and large amounts of pp71 (54). It was proposed that this is a mechanism by which HCMV can push the probability of a nascent viral population toward some virions being better suited to establishing latency (pp71-low) and some being more efficient at establishing lytic replication (pp71-high). Under this assumption, it is plausible that in the case of a strain such as MOLD, which displays high tropism for epithelial cells, HDAC inhibitors can still enhance MOLD lytic replication by augmenting IE expression in cells infected by viral particles that have low pp71 levels and are thus less likely to overcome the block of HDAC recruitment by Daxx.

Taken together, our studies reveal an additional role for pentamer in promoting epithelial cell infection with HCMV. While it remains clear that pentamer binding to its cellular receptor(s) is an important step in promoting viral entry and/or fusion, it appears that pentamer plays an expanded role in that it is also capable of profoundly enhancing lytic gene expression subsequent to viral entry. Whether this is via a direct role of pentamer-induced signaling or the consequence of a particular pentamer-driven internalization pathway is yet to be determined. The continued exploration of the mechanisms of HCMV entry into epithelial cells is likely to reveal additional roles for this enigmatic glycoprotein complex and reinforce the notion that HCMV uses a broad array of processes to enter and initiate lytic replication in the diverse cell types that it is able to infect.

MATERIALS AND METHODS

Cell lines used and isolation of primary salivary epithelial cells.

HS68 and ARPE-19 cells were acquired from the American Type Culture Collection (ATCC). HS68 cells are a normal human foreskin fibroblast (HFF) cell line derived from infant fibroblasts. HS68 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Corning) supplemented with 10% FetalClone III serum (HyClone) and 1% penicillin-streptomycin (P/S). ARPE-19 cells are a retinal pigment epithelial line derived from the human retina. ARPE-19 cells were maintained in DMEM–F-12 medium (50:50) (Corning) with 10% fetal bovine serum (FBS) (Corning), 1% P/S, and 0.5% amphotericin B. HFF-tet-tert cells (Richard Stanton, Cardiff University) are a line of HFF cells that have been transduced to express the repressor protein TetR and were grown in the same medium as the one used for HS68 cells. Primary human submandibular and parotid salivary gland-derived epithelial (hSGE) cells originated from patient salivary tissue at the University of Cincinnati’s Department of Otolaryngology (9, 55). This study has been reviewed and approved by the Institutional Review Board at the University of Cincinnati (Federal-wide assurance number 00003152, IRB protocol 2016-4183). Salivary cells were cultured in BEGM growth medium (containing the BEGM bullet kit) provided by Lonza and supplemented with 4% charcoal-stripped FBS (Gibco).

Viral strains used and production of viral stocks.

HCMV strains used in this study include AD169 (ATCC), TB40E (TB40E-mCherry-3×FLAGUS28) and RV-FIX (FIX-GFP) (Christine O’Connor, Cleveland Clinic), MOLD (Chris Benedict, LAIA), BADwt and BADrUL131 (Andrew Yurochko, Louisiana State University, and Brent Ryckman, University of Montana), and Merlin (TetO-UL131) (Richard Stanton, Cardiff University). Stocks were made by coculturing HS68 cells with previously infected HS68 cells (approximately 1% infected cells) and were fed every 2 to 3 days in T-175 flasks until it was observed that cells displaying cytopathic effects (CPE) reached 100%. The supernatant was snap-frozen and stored at −80°C, and harvests were typically done between days 10 and 13 postinfection. To obtain pentamer-null stocks of Merlin, HFF-tet-tert cells were infected at approximately 1%, and the virus-containing supernatant was harvested at around 70 to 80% observable CPE. These cells were then washed with Dulbecco’s phosphate-buffered saline (DPBS) to remove residual virus and fed medium containing 2 μg/mL doxycycline to relieve repression at the pentamer locus. The supernatant was harvested 3 days later to obtain pentamer-positive virions.

Pharmacological agents.

Valproic acid (Sigma-Aldrich) was dissolved in distilled water (dH₂O) at a final concentration of 100 mM, filter sterilized, and stored at −20°C. Stocks of trichostatin A (Sigma-Aldrich) were prepared at a final concentration of 0.5 mM in DMSO and stored at −20°C. MC1568 (Sigma-Aldrich) stocks were prepared at a final concentration of 1 mM in DMSO and stored at 4°C. Entinostat (Sigma-Aldrich) was prepared at a final concentration of 1 mM in DMSO and stored at −20°C. Bafilomycin A1 (Sigma-Aldrich) was prepared at a final concentration of 35 μM in DMSO and stored at −20°C.

Flow cytometry analysis of infected cells.

In a 12-well plate, 10⁵ cells were seeded for infection or left uninfected. Cells were harvested at the indicated time points using trypsin and then neutralized with complete medium. Cells were centrifuged at 800 relative centrifugal force (rcf) for 5 min and then resuspended with 70% ice-cold ethanol for at least 30 min to fix the cells. Cells were then again pelleted at 800 rcf and resuspended in permeabilization buffer (DPBS with 0.5% bovine serum albumin [BSA] and 0.5% Tween 20) for 10 min at 4°C, pelleted once again, and then stained with an anti-IE1/2 antibody (mAb810-Alexa Fluor 488) diluted 1:250 in DPBS with 0.5% BSA for approximately 2 h in the dark. Cells were then resuspended in DPBS and analyzed on the FL1 channel of a FACSCalibur flow cytometer (BD Biosciences) or on the FITC-A channel of a CytoFLEX S instrument (Beckman Coulter).

Protein extract preparation and Western blot analysis.

In a 12-well plate, 2 × 10⁵ ARPE-19 cells were seeded and infected. Cells were washed once with DPBS prior to collection and then, using 3× Laemmli sample buffer, harvested by cell scraping. Samples were sonicated on ice 10 s at a time for 20 s total and boiled at 100°C for 3 min. The samples were then separated using a 4 to 20% Tris-glycine gel (Invitrogen) and then transferred to nitrocellulose. Samples were probed using antibodies against IE1/2 (1:1,000) (Chemicon), UL44 (1:50) (John Shanley), pp65 (1:1,000) (Virusys Corporation), β-actin conjugated to horseradish peroxidase (HRP) (1:1,000) (Cell Signaling Technologies), pp71 (Thomas Shenk) (1:100), Eea1 (1:1,000) (Cell Signaling Technologies), Daxx (1:4,000) (Millipore), lamin A/C (1:1,000) (Cell Signaling Technologies), histone H3-pan (1:10,000) (Millipore), and histone H3-K9Ac (1:5,000) (Millipore). Appropriate secondary anti-rabbit or anti-mouse antibodies conjugated to HRP were used, followed by incubation with the SuperSignal West Pico Plus chemiluminescent substrate (Thermo), and the signal was detected by the use of a C-DiGit blot scanner (Li-Cor).

Nuclear isolation, DNA extraction, and quantitative PCR.

ARPE-19 cells were seeded into 100-mm² dishes at a density of 2 × 10⁶ cells per dish. The next day, after adhering, select cells were incubated with 35 nM BafA1 for 1 h before infection at 37°C. Next, DMSO (vehicle control) or entinostat was added to a final concentration of 1% DMSO or 10 μM entinostat to appropriate plates; next, all cells were infected at an MOI of 2 and placed back at 37°C until the next day. The next day, all cells were washed once with 10 mL DPBS and given fresh medium without supplemented drugs. After 2 days of infection, cells were harvested using trypsin, neutralized with complete medium, and centrifuged at 500 rcf for 5 min. Cell pellets were resuspended in DPBS and split into 3 samples: approximately 45% for whole-cell DNA extraction, approximately 45% for nuclear isolation followed by DNA extraction, and approximately 10% to be stained for IE1/2 flow cytometry analysis as described above. Nuclear isolation was achieved by utilizing the nuclear extract kit (Active Motif) and the accompanying protocol. In brief, cells were resuspended in 1× hypotonic buffer and incubated on ice for 15 min. Cell lysis was achieved by the addition of the provided detergent and 10 s of vigorous vortexing. Complete lysis of cells, and now free nuclei, was confirmed via microscopy. Nuclei were pelleted at 14,000 rcf for 30 s, the buffer was aspirated, and the pellet was washed once with 1 mL DPBS followed by pelleting at 14,000 rcf for 30 s. Harvested whole cells or purified nuclei were then subjected to DNA purification using the GeneJet genomic DNA purification kit (Thermo Scientific) according to the manufacturer’s protocol. Quantitative PCR (qPCR) was performed on 50 ng of the sample in triplicate using PowerUp SYBR green master mix (Applied Biosystems). DNA from ARPE-19 cells was used to generate a standard curve to calculate the copy number of host genomes. DNA from the pHB5 bacmid was used to generate a standard curve to calculate copy numbers of viral genomes. Amplification of RPPH1 was achieved by repeating 40 cycles of 10 s at 95°C, annealing for 15 s at 56°C, and extension for 20 s at 72°C using the primer pair CTAACAGGGCTCTCCCTGAG and ACCTCACCTCAGCCATTGAAC. Amplification of the viral UL54 sequence was achieved by 40 cycles of 10 s at 95°C, annealing for 15 s at 66°C, and extension for 25 s at 72°C using the primer pair GGGCACAGCGGCGGTAGAGATGAT and CATTAGCCACGAAACAACGCGGGA.

Viral growth analysis.

ARPE cells were seeded at 5 × 10⁵ cells in a 6-well plate; dosed with DMSO (1%), TSA (1 μM), or entinostat (10 μM); and then infected at an MOI of 2. The following morning, medium was aspirated, and cells were washed using 2 to 3 mL DPBS 3 times before being fed 1 mL medium. At 1, 2, 4, 7, 10, or 14 days postinfection, 500 μL of the supernatant was harvested, and cells were replenished back to a 1-mL total volume with fresh medium. The supernatant was flash-frozen in a dry-ice–ethanol bath and stored at −80°C until further use. The viral titer was determined by thawing the frozen supernatant at 37°C and infecting 12-well plates seeded with 10⁵ HS68 cells per well using 2 to 3 different volumes of the supernatant. Cells were harvested at 2 days postinfection, and the amount of positively infected cells was determined using flow cytometry as described above to calculate the viral titer in the supernatant.

Immunofluorescence microscopy.

Sterile coverslips were coated in a 6-well plate with rat tail collagen for 30 min at room temperature and then washed once with DPBS. Coated coverslips were then seeded with 4 × 10⁵ ARPE cells and then infected at an MOI of 2 or left uninfected. At 24 h postinfection, coverslips were washed once with DPBS and then fixed using 4% paraformaldehyde for 10 min at room temperature. Coverslips were then washed 3 times using cold DPBS. Cells were permeabilized using 0.3% Triton X-100 (Sigma) diluted in DPBS for 10 min at room temperature and then washed 3 times with DPBS for 15 min total. Coverslips were incubated with DPBS containing 0.1% Tween 20, 1% BSA, and 22.52 mg/mL glycine for 30 min at room temperature. Coverslips were sequentially stained first with anti-pp71 (1:100) followed by anti-mouse antibody conjugated to Alexa Fluor 594 (1:250) (BioLegend) and anti-IE1/2 antibody conjugated to Alexa Fluor 488 (1:250). Each incubation was performed for either 2 h in the dark at room temperature or overnight in the dark at 4°C. In between each antibody stain, coverslips were washed 3 times using 2 mL DPBS for 15 min total. Cells were counterstained with DAPI (1:1,000) for 5 min and then washed for 5 min with 2 mL DPBS. Coverslips were then mounted onto slides using ProLong glass antifade mountant (Invitrogen). Images were taken using an IXplore fluorescence confocal microscope (Olympus-Lifescience).