STING facilitates nuclear import of herpesvirus genome during infection

Significance

Human cytomegalovirus (HCMV) establishes lifelong latent infection in 60 to 90% of the population worldwide and constitutes a serious global health burden. For a successful infection, HCMV must overcome physical barriers: the plasma membrane and the nuclear envelope. Here, we report that the immune adaptor protein stimulator of interferon genes (STING), typically located in the endoplasmic reticulum, is a crucial host factor involved in the nuclear entry of HCMV. Without STING, the interaction between viral capsid and nuclear pore complex is defective, and HCMV fails to deliver the viral genome into the nucleus. In monocytes, STING deficiency prevents a successful establishment of HCMV persistence. This study offers a perspective for understanding HCMV uncoating and for developing interventions against herpesvirus infection.

Abstract

Once inside the host cell, DNA viruses must overcome the physical barrier posed by the nuclear envelope to establish a successful infection. The mechanism underlying this process remains unclear. Here, we show that the herpesvirus exploits the immune adaptor stimulator of interferon genes (STING) to facilitate nuclear import of the viral genome. Following the entry of the viral capsid into the cell, STING binds the viral capsid, mediates capsid docking to the nuclear pore complex via physical interaction, and subsequently enables accumulation of the viral genome in the nucleus. Silencing STING in human cytomegalovirus (HCMV)-susceptible cells inhibited nuclear import of the viral genome and reduced the ensuing viral gene expression. Overexpressing STING increased the host cell’s susceptibility to HCMV and herpes simplex virus 1 by improving the nuclear delivery of viral DNA at the early stage of infection. These observations suggest that the proviral activity of STING is conserved and exploited by the herpesvirus family. Intriguingly, in monocytes, which act as latent reservoirs of HCMV, STING deficiency negatively regulated the establishment of HCMV latency and reactivation. Our findings identify STING as a proviral host factor regulating latency and reactivation of herpesviruses.

Human cytomegalovirus (HCMV) belongs to the β-herpesvirus subfamily and establishes lifelong latent infection in 60 to 90% of the population worldwide. The inability of the immune response to clear the virus and the absence of a vaccine renders HCMV a serious global health burden (1). HCMV is the most common cause of congenital viral infection during pregnancy and also increases the morbidity and mortality among immunosuppressed transplant patients and AIDS patients (2, 3).

DNA viruses must sequentially overcome two physical barriers, the plasma membrane and the nuclear envelope, to replicate in the nucleus (4). The herpesvirus double-stranded DNA (dsDNA) genome is tightly packed inside an icosahedral capsid structure measuring 130 nm in diameter (5, 6). The nuclear pore complex (NPC) allows the transport of macromolecules up to 39 nm in diameter; thus, nuclear entry of herpesviruses with a capsid diameter that is larger than this diffusion limit is a critical step for successful viral infection (7). Herpesviruses employ an uncoating strategy whereby the capsid undergoes conformational change without disruption of the capsid structure, open the portal, and release the viral DNA into the nucleus via NPC. Several host cell surface receptors determining HCMV cell tropism and permissiveness have been identified (8–14); however, the host factor(s) involved in capsid uncoating and nuclear genome import is (are) yet to be uncovered.

In the event of improper translocation of the viral genome, the genome is exposed to the cytoplasm. Then, host factors recognize viral DNA as a foreign antigen and elicit antiviral immune response resulting in degradation of the viral genome by cytosolic nucleases (15). Pressure-driven viral DNA release into the host nucleus accounts for only 50% of DNA ejection into the cellular environment, and small polycationic molecules can block viral genome ejection from the capsid, according to recent studies (16, 17). These findings offer two interesting perspectives: the presence of unknown host factors that regulate uncoating and the possibility of developing antiviral agents that target this uncoating process.

Although a primary lytic HCMV infection induces robust innate and adaptive immune responses, HCMV is not completely cleared in vivo but, rather, persists in the host cell, becoming a lifelong potential threat (18). HCMV has contrived a strategy to establish latency in the host cell, such as CD34⁺ hematopoietic progenitor cell (HPC) and CD14⁺ monocyte, to survive. Although several entry receptors in fibroblasts and endothelial and epithelial cells have been identified (8–14), only a few host genes or signaling pathways in monocytic cells required for HCMV infection have been researched. For example, EGFR and c-Src signaling must be active to lead to cytoplasmic trafficking, nuclear translocation, and infection in monocytes (19, 20). However, host factors that regulate the HCMV latency and reactivation after HCMV enters the monocytes are largely unknown.

Stimulator of interferon (IFN) genes (STING) is an endoplasmic reticulum–resident membrane protein ubiquitously expressed in both immune cells and nonimmune cells (21, 22). Recognition of cytosolic dsDNA leads to the synthesis of cyclic GMP–AMPs (cGAMPs) by activating a cyclic GMP–AMP synthase (cGAS). cGAMPs bind to STING, inducing activation of TBK1 and IRF3 signaling cascades (23). The canonical IFN-dependent role of STING in antiviral immunity has been extensively investigated. Interestingly, several recent reports suggest a noncanonical proviral function of STING in viral infection (24, 25). A potential proviral role of STING in HCMV infection is largely unknown.

Here, we demonstrate a proviral function of STING during herpesvirus infection. We show that ectopic expression of STING in STING-deficient cells converts an otherwise insusceptible cell to an HCMV- and herpes simplex virus 1 (HSV-1)–susceptible cell, implying that this STING function is evolutionarily conserved. Furthermore, loss of STING significantly represses the establishment of HCMV latency and viral reactivation in monocytic cells, suggesting that STING is a host factor that regulates the herpesvirus life cycle.

Results

STING Is Critical for HCMV Gene Expression during the Immediate–Early Phase of Infection.

HCMV infection activates the cGAS–STING–TBK1 signaling pathway in human endothelial cells, monocyte-derived dendritic cells, and macrophages (26, 27). These cells are far less permissive to infection than human fibroblasts. We speculated that the relatively higher permissiveness of fibroblasts might be associated with a defective cGAS–STING antiviral signaling axis in these cells. We tested whether the cGAS–STING pathway is functional in fibroblasts by transiently silencing the expression of cGAS and STING using small interfering RNA (siRNA) and then infecting the cells with HCMV. The silencing did not affect cell viability (SI Appendix, Fig. S1A). HCMV infection induced the IFN-β messenger RNA (mRNA) levels in the control cells, while cGAS silencing resulted in a dramatically decreased induction of the IFN-β mRNA levels (Fig. 1A). Similarly, knockdown of STING led to a reduction of IFN-β mRNA levels in HCMV infection (Fig. 1A). These findings indicate that the cGAS–STING axis is active in fibroblasts. Since both cGAS and STING are antiviral proteins, we expected that the silencing of either of these molecules would increase viral gene expression. Indeed, in cGAS-silenced cells, mRNA and protein levels of the HCMV immediate–early gene IE1, a hallmark of lytic infection, were increased (Fig. 1B and C). However, unexpectedly, silencing of STING resulted in diminished IE1 mRNA and IE1 protein levels (Fig. 1 B and C). We established STING-silenced stable immortalized human foreskin fibroblasts (HFF) (hereafter referred to as HFF-TEL cells) using lentiviral short hairpin RNA (shRNA) specific to STING. The shRNA-mediated silencing of STING did not affect cell viability (SI Appendix, Fig. S1B). Similar to siRNA-treated cells, stable knockdown of STING resulted in decreases in IE1 protein levels (Fig. 1D). Next, we generated STING-depleted fibroblasts using the lentiviral CRISPR-Cas9 system. Since we utilized primary fibroblasts, we used a transient mixed population of STING-knockout (KO) cells without single-cell clone selection, after confirming efficient depletion of STING by immunoblotting, as described elsewhere (28). KO of STING in the primary fibroblasts did not significantly affect cell viability (SI Appendix, Fig. S1C). We observed a substantial decrease in both IE1 mRNA and IE1 protein levels that proceeded in a guide RNA sequence-independent manner (Fig. 1 E and F).

Fig. 1.

STING proviral activity in HCMV gene expression during immediate–early phase infection. (A) RT-qPCR analysis of primary HFF cells treated with scrambled control RNA, cGAS, or STING siRNA, harvested at 6 hpi after HCMV infection (MOI = 3) using human IFN-β–specific (Left) and target gene–specific primers (Right). (B) Immunoblot of HCMV IE1 protein under 3 MOI infection at 6 hpi in transiently silenced HFF cells in A (Left). The relative band intensity of IE1 normalized by GAPDH was quantified by ImageJ software (Right) (C) RT-qPCR analysis of primary HFF cells treated with scrambled control RNA, cGAS, or STING siRNA, harvested at 6 hpi after 3 MOI HCMV infection using HCMV IE1-specific (Left) and UL36-specific primers (Right), related to A. (D) Immunoblot of HCMV IE1 protein under 3 MOI infection at 6 hpi in stably silenced HFF-TEL cells. (E) RT-qPCR analysis of primary HFF cells knocked out by the CRISPR-Cas9 system, harvested at 6 hpi after 0.5 MOI HCMV infection using HCMV IE1–specific primer. (F) Immunoblot of HCMV IE1 protein under 0.5 MOI infection at 6 hpi in STING-KO primary HFF cells. (G) Immunoblot of HCMV IE1 protein under 3 MOI infection at 6 hpi in transiently silenced U373MG cells. (H) Immunoblot of HCMV IE1 protein under 3 MOI infection at 6 hpi in STING-KO U373MG cells (Left). The relative band intensity of IE1 normalized by GAPDH was quantified by ImageJ software (Right). qPCR data were normalized to GAPDH mRNA. shLuc; luciferase shRNA, shSTING; STING shRNA. For immunoblot, one of three biological replicates is representatively shown. Mean ± SEM, n = 3; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (one-way ANOVA with Tukey’s multiple comparisons test or two-tailed unpaired Student’s t test).

To test if STING is also critical for HCMV IE1 gene expression in other cell types, we used siRNA to transiently silence STING in the U373MG cell line, an epithelial cell line susceptible to HCMV. The siRNA treatment did not affect the viability of U373MG cells (SI Appendix, Fig. S1D). Similar to what we observed with fibroblasts, IE1 protein levels were reduced in HCMV-infected STING-silenced U373MG cells (Fig. 1G). In addition, we generated a STING-KO U373MG cell line using lentiviral CRISPR-Cas9 technology. After puromycin selection of STING-KO cells, we confirmed no differences in the viability of control and STING-knockout U373MG cell lines (SI Appendix, Fig. S1E). Upon HCMV infection, IE1 protein levels were dramatically reduced in STING-KO U373MG cells, suggesting that the proviral activity of STING during the immediate–early phase of HCMV lytic infection is not cell-type specific but a conserved phenomenon (Fig. 1H)

STING Does Not Impact Viral Entry into the Cell.

To successfully infect the host cell, a virus must first cross the plasma membrane barrier in a process called the “viral entry.” After HCMV entry into the cell via receptor-mediated macropinocytosis or endocytosis, the tegumented viral capsid is released in the cytoplasm and ready to transverse the cytoplasm to reach the nucleus. pp65/pUL83, a major tegument protein of HCMV, is the most abundant viral protein and is localized in both the cytoplasm and the nucleus after infection (29). Consequently, viral entry into a cell is usually quantified based on the amount of pp65 protein accumulated in the cytoplasm and the nucleus of the infected cell. To investigate whether STING silencing influences viral entry, we infected the STING-silenced primary HFF-TEL with HCMV and harvested whole-cell lysates after 30 min for pp65 immunoblot analysis; siControl and sicGAS treatments served as negative controls. Neither siRNA treatment affected the viral entry (Fig. 2A). We then immunostained these cells with pp65-specific antibody and quantified the amount of pp65 in cells 2 h postinfection (hpi) (Fig. 2B). We quantified two different parameters: the average intensity of the pp65-positive signal and the number of pp65 foci per cell. Similar to the result of the pp65 immunoblot assay, no major differences were apparent between the siControl-, sicGAS-, and siSTING-treated fibroblasts (Fig. 2C and D). In addition, we extracted total genomic DNA from whole-cell lysates, harvested 2 h after HCMV infection, and determined the viral genome levels by qPCR using HCMV genomic DNA-specific primers. The amount of intracellular viral genomic DNA was the same in the three samples (Fig. 2E). These observations indicate that the increased IE1 levels in cGAS-silenced cells and decreased IE1 levels in STING-silenced cells are not caused by differences in the viral entry into the cell but, rather, are linked to a factor acting at a subsequent stage of infection.

Fig. 2.

STING does not influence viral entry into the cell. (A) Immunoblot analysis of tegument pp65 using whole-cell lysate of primary HFF cells treated with control, cGAS, or STING siRNA, harvested at 30 min after HCMV infection (MOI = 3). (B–D) pp65 was immunostained with anti-pp65 in primary HFF cells treated with control, cGAS, or STING siRNA, harvested at 2 hpi after HCMV infection (MOI = 3). Representative images of each sample are displayed (B). Quantification of the average intensity of pp65 signal and the number of pp65 foci per cell was processed as described in Materials and Methods (C and D). (E) Genomic DNA (gDNA)-qPCR analysis of the total fraction of primary HFF cells treated with control, cGAS, or STING siRNA, harvested at 2 hpi after HCMV infection (MOI = 3). (F) Immunoblot of tegument pp65 in shSTING-HFF-TEL cells infected with HCMV (MOI = 0.5 and 3) and harvested 30 min later. (G) Immunoblot of tegument pp65 in STING-KO primary HFF cells infected with HCMV (MOI = 0.5) and harvested at 2 hpi. ns; not significant.

We further tested the notion that the efficacy of viral entry is independent of multiplicity of infectivity (MOI) in both control and STING-silenced cells (Fig. 2F). To rule out the possibility that viruses binding to the cell surface that had not yet entered the cell are being detected in the pp65 immunoblot assay, we analyzed the amount of pp65 after HCMV infection over time. The amount of pp65 protein that had entered the cells was similar in the control and STING-silenced cells and remained similarly constant until 6 hpi (SI Appendix, Fig. S2A). The detected pp65 protein was not de novo synthesized because it was also detected after cycloheximide (CHX) treatment. Further, comparable amounts of tegument pp65 protein were detected in cells, with or without STING, infected with ultraviolet (UV)-inactivated or wild-type (WT) viruses (SI Appendix, Fig. S2B). Since the UV-inactivated virus only loses the ability to express de novo synthesized viral genes (SI Appendix, Fig. S2B) but contains functionally active glycoproteins, teguments, and capsid proteins, these observations indicate that silencing of STING does not affect the viral entry.

We also confirmed that the KO of STING in primary HFF-TEL does not affect the viral entry, as indicated by the same band intensity in the pp65 immunoblot assay (Fig. 2G). The viral entry into STING-KO U373MG cells was not defective, as compared to WT U373MG cells, regardless of the MOI, as determined by pp65 immunoblot and qPCR of total viral genome levels (SI Appendix, Fig. S2 C and D). In conclusion, loss of STING does not affect the viral entry.

STING Plays a Role in Nuclear Import of the HCMV Genome.

STING silencing did not affect the initial entry of the virus into the cell (Fig. 2). Hence, we set out to identify the step at which STING influences HCMV infection. Using transmission electron microscopy (TEM), we visualized and compared the efficacy of viral genomic DNA release into the nucleus of shLuc- and shSTING-HFF-TEL cells. Using TEM, unreleased viral genomes are observed as a dark mass in a hexagonal capsid, as described previously (30, 31), enabling us to distinguish empty from full capsids (Fig. 3A). We prepared TEM samples 4 hpi using a modified procedure from a previous report (31).

Fig. 3.

Silencing of STING abrogates nuclear import of the HCMV genome. (A) Representative images of an empty capsid in the shLuc (yellow arrowhead) and a full capsid in shSTING-HFF-TEL cells (red arrowhead). (B) Schematic for the distribution of cytoplasmic/perinuclear and full/empty capsids in TEM images. (C and D) Quantitative analysis of viral capsids in TEM images of the 4-hpi sample infected with HCMV at 50 MOI. Related data are presented in SI Appendix, Table S1. One of two independent TEM analyses is representatively shown. (E) Genomic DNA extracted from the subcellular fractions was used for qPCR analysis (Left). Immunoblot of fractionated cell lysates used for gDNA extraction with indicated antibodies (Right). For immunoblot, one of three biological replicates is representatively shown. C, cytoplasm; M, membrane; N, nucleus. Mean ± SEM, n = 3; ****P < 0.0001, ns; not significant (two-tailed unpaired Student’s t test).

We categorized intracellular capsids into two groups, cytoplasmic and perinuclear, considering the 130-nm diameter of the HCMV capsid (Fig. 3B and SI Appendix, Fig. S3A). We analyzed ∼200 capsids in each sample and calculated the ratio of empty to full capsids in the cytoplasmic and perinuclear regions (SI Appendix, Table S1). The overall proportion of capsids in the perinuclear and cytoplasmic regions did not differ in the shLuc and shSTING cells (Fig. 3C), suggesting that STING deficiency does not affect the intracellular transport of the capsid. However, the ratio of empty to full capsids in the perinuclear fraction decreased nearly twofold in STING-silenced cells compared to the control cells (Fig. 3 D, Left). In the cytoplasmic region, no significant differences were observed in the capsid ratios between the control and STING-silenced cells (Fig. 3 D, Right). To confirm the finding that STING silencing reduces the efficiency of nuclear delivery of the viral genome by a quantitative biochemical assay, we infected the cells with the same viral load, harvested them 4 hpi for subcellular fractionation, and analyzed the viral genome levels by qPCR. As expected, the viral genome levels were lower in the nuclear fraction of STING-silenced cells than in the control cells (Fig. 3E).

Next, we examined whether depletion of STING in U373MG cells similarly influences nuclear import of the viral genome. We found that the ratio of empty to full capsids in the perinuclear fraction was considerably reduced in STING-KO cells compared to the control cells, whereas the ratio of full to empty capsids in that fraction increased in STING-depleted cells (SI Appendix, Fig. S3B and Table S2). Quantification of HCMV genomic DNA in the nuclear fraction of HCMV-infected cells revealed that the KO of STING was associated with an impaired nuclear import of the viral genome (SI Appendix, Fig. S3C). These findings indicate that STING participates in capsid uncoating and subsequent delivery of the viral genome into the nucleus.

STING-Facilitated Binding of the Viral Capsid to Nup358 of NPC Is Crucial for HCMV Infection.

Nup358 and Nup214 are components of the cytoplasmic filaments of NPC involved in HSV-1 capsid binding to the nuclear envelope pore (30–32). Since the main architectural features of the HCMV and HSV-1 capsids are similar, we examined whether both Nup358 and Nup214 are necessary for HCMV uncoating. Knockdown of Nup358 and Nup214 did not influence STING expression or viral entry (SI Appendix, Fig. S4 A and B). However, silencing of Nup358 resulted in a decreased expression of IE1, whereas the effect of Nup214 on viral gene expression appeared to be minimal (SI Appendix, Fig. S4A).

Coimmunoprecipitation analysis revealed that Nup358, but not Nup214, interacted with STING in endogenous as well as ectopic expression systems (Fig. 4A and SI Appendix, Fig. S4 C–E). These observations indicate that both STING and Nup358 might cooperate in the uncoating of HCMV. Accordingly, we investigated the mechanisms of the STING role in the uncoating process by asking which viral capsid protein is capable of interacting with both STING and Nup358. We cloned the genes of the five viral capsid proteins that compose the virion (5) (SI Appendix, Fig. S5A). The individually overexpressed viral capsid proteins were located in the cytoplasm or in both the nucleus and the cytoplasm. None of these proteins was exclusively located in the nucleus (SI Appendix, Fig. S5B). Interestingly, all capsid proteins except for UL85 were able to bind to both Nup358 and STING (SI Appendix, Fig. S5 C and D). Furthermore, we found that the interaction between the major capsid protein UL86 and Nup358 was enhanced in a STING-dependent manner (Fig. 4B). Together with the notion that STING can be localized in the nuclear membrane (SI Appendix, Fig. S6), these findings suggest that STING acts as a physical platform, which is implemental for capsid docking onto NPC in the perinuclear region.

Fig. 4.

STING-mediated interaction of viral capsid with Nup358 enhances HCMV infection. (A) Coimmunoprecipitation using control IgG or anti-Flag antibody in stable STING-Flag–expressing HFF-TEL cells after siRNA-mediated Nup358 knockdown. (B) Coimmunoprecipitation using antibody to endogenous Nup358 in HEK293T cells expressing UL86-HA and STING-Flag either alone or in combination as depicted. (C) Proximal ligation assay (PLA) using anti-MCP (UL86) and anti-STING to detect the interaction between the capsid and STING at 3 hpi, after HCMV infection (MOI =10). (D) Proximal ligation assay (PLA) using anti-MCP (UL86) and anti-Nup358 to detect colocalization between the viral capsid and Nup358 at 3 hpi after HCMV infection (MOI = 20) (Upper). Data are represented as mean ± SD of all counted cells. Representative images for each sample used for analyzing PLA (Lower). (E) Schematic of the structure of truncated mutant forms of STING. (F and G) Immunoblot analysis of IE1 protein expression in shSTING-HFF-TEL cells reconstituted with (F) STING-FL and -ΔTM or (G) STING-FL and chimeric mutants, harvested at 6 hpi after HCMV infection (MOI = 0.5). For immunoblots, one of two biological replicates is representatively shown. ****P < 0.0001, ns; not significant (one-way ANOVA with Šidák’s multiple comparison test).

To confirm the tripartite interaction between STING, Nup358, and the viral capsid observed by coimmunoprecipitation in a transient overexpression assay, we performed an in situ proximal ligation assay (PLA), an antibody-based detection of protein–protein interactions. Compared with the control, we detected a specific interaction between UL86 (major capsid protein, MCP) and STING in the HCMV-infected fibroblast (Fig. 4C and SI Appendix, Fig. S5G). The physical interaction between UL86 and Nup358 was also readily detected in virus-infected cells 3 hpi (Fig. 4D, lane 3, and SI Appendix, Fig. S5H). STING depletion reduced the interaction between UL86 and Nup358 (Fig. 4D, lane 4, and SI Appendix, Fig. S5H), supporting the notion that STING quantitatively stabilizes the interaction between the capsid and the nuclear pore to promote viral genome delivery.

To identify the STING domain that is responsible for mediating the tripartite interaction with Nup358 and UL86 during an infection, we generated two truncated STING variants: a ΔCT variant, with a missing carboxyl terminus, and a ΔTM variant, missing a transmembrane domain (Fig. 4E and SI Appendix, Fig. S7A). Both variants exhibited a much weaker affinity for Nup358 and UL86 than full-length STING (SI Appendix, Fig. S7 B and C). We also found that reconstituting shSTING cells with the ΔTM variant failed to restore HCMV IE1 expression (Fig. 4F). To further verify if the inability of the ΔTM variant to restore IE1 expression is due to its mislocalization, we additionally constructed two chimeric STING variants, DGAT2(TM)+STING(CT) and STING(TM)+GFP, which have the transmembrane domain for ER localization but are unable to activate the STING–TBK1–IRF3 signaling cascade (Fig. 4E). Unlike the ΔTM variant, which was localized in both the cytoplasm and the nucleus, the other three variants were exclusively localized in the cytoplasm (SI Appendix, Fig. S7A). The binding of DGAT2(TM)+STING(CT) variant to Nup358 and UL86 was similar to that of STING-FL, whereas the STING(TM)+GFP chimeric variant showed a much weaker binding affinity for both (SI Appendix, Fig. S7 D and E). This suggests that the carboxyl terminus of STING is important for binding to Nup358 and UL86. Furthermore, these observations indicate that a correct ER localization of STING is required for the tripartite interaction between STING, Nup358, and UL86. Expression of DGAT2(TM)+STING(CT) or STING(TM)+GFP proteins in shSTING cells failed to fully restore IE1 expression (Fig. 4G), indicating that both the TM domain and carboxyl termini of STING are critical for promoting viral infection.

Ectopic Expression of STING Increases Host Cell Susceptibility to Herpesvirus Infection.

Since DNA viruses must overcome the nuclear envelope barrier to infect the host cell, we proposed that STING might act as a cytoplasmic platform for the capsid and a determinant of cell susceptibility to HCMV. Accordingly, we detected a positive correlation between cell susceptibility and endogenous STING levels in a panel of cell lines, supporting this notion (Fig. 5A). High HCMV IE1 levels were dependent on STING levels in the different cell lines, whereas there was no significant difference in the amount of pp65 tegument delivery (i.e., capacity for cell entry; Fig. 2) between the cell lines (Fig. 5 B and C). These observations support the notion that STING is a host factor regulating a postentry step during HCMV infection. We already showed that the loss of STING resulted in a decreased expression of viral IE1 proteins in a cell-type–independent manner (Fig. 1). Consistent with our hypothesis that STING confers susceptibility to HCMV, ectopic expression of FLAG-tagged STING in naive U373MG cells resulted in increased IE1 protein levels (Fig. 5D). Further, A549 (human lung carcinoma) cells did not harbor detectable levels of endogenous STING, which may account for their lack of susceptibility to HCMV infection (Fig. 5 A–C). We coexpressed STING and platelet-derived growth factor receptor alpha (PDGFR-α) in A549 cells. The latter is required for HCMV viral translation upon phosphorylation after PI3K–Akt activation (8) (Fig. 5E and SI Appendix, Fig. S8 A and B). Although HCMV entered the A549 cells independently of PDGFR-α and STING expression (SI Appendix, Fig. S8C), the levels of viral IE1 protein were strongly induced only in cells expressing both PDGFR-α and STING (Fig. 5E, lane 2). In line with the findings in fibroblasts (Fig. 3E), HCMV genomic DNA was enriched in the nuclear fraction of STING-expressing A549 cells within 4 hpi (Fig. 5F). Hence, STING can render originally HCMV-insusceptible cells susceptible to the virus.

Fig. 5.

STING renders insusceptible cells becoming susceptible to herpesvirus infection. (A) Immunoblot of endogenous STING, cGAS, TBK1, and GAPDH in different cell lines. (B) Immunoblot of the HCMV major tegument protein, pp65, using the lysate obtained from HCMV-infected cell lines (MOI = 1), which was harvested after 30 min. (C) Immunoblot of cell lysates after HCMV infection (MOI = 1), harvested at 6 hpi. (D) Immunoblot of IE1 expression in transient STING-Flag–expressing U373MG cells after 1 MOI infection, harvested at 6 hpi. (E) Immunoblot of IE1 expression in PDGFRα-HA and GFP or STING-Flag–expressing A549 cells after HCMV infection (MOI = 1), harvested at 6 hpi. (F) gDNA-qPCR analysis of nuclear fraction of stably GFP- or STING-Flag–expressing A549 cells after HCMV infection (MOI = 1), harvested at 4 hpi. qPCR data were normalized to human MDM2 DNA in the host genome. Mean ± SEM, n = 3; *P < 0.05 (two-tailed unpaired Student’s t test).

Next, we determined whether the proviral activity of STING extends to HSV-1 infection. HSV-1 is another member of the herpesvirus family. STING requirement for HSV-1 replication has been implied, although the mechanism remains unknown (33). We confirmed that the KO of STING in HepG2 cells decreased HSV-1 gene expression (SI Appendix, Fig. S8 D and E). We also found that the amount of HSV-1 viral DNA in the nuclear fraction of STING-KO HepG2 cells was reduced by 50% compared with that in WT cells (SI Appendix, Fig. S8F), which reflected our earlier observations (Figs. 3E and 5F). In addition, HSV-1 capsid proteins VP5 and UL6, homologs of HCMV UL86 and UL104, respectively (34) (SI Appendix, Fig. S5 A and B), were able to bind to Nup358 and STING (SI Appendix, Fig. S5 E and F). These observations suggest that the proviral function of STING during capsid uncoating is evolutionarily conserved, at least when considering the herpesvirus family.

Establishment of Latency and Reactivation of HCMV in Monocytes Is STING Dependent.

After primary HCMV infection of predominantly endothelial and epithelial cells in vivo, HCMV can establish a lifelong latent infection in CD14⁺ monocytes and CD34⁺ HPCs, which are generally far less susceptible to viruses than other cell types. We considered that the proviral function of STING in promoting nuclear entry of viral DNA might be more relevant in cell types in which the cell entry is less efficient, such as human monocytic cell lines.

The human monocytic cell line THP-1 is used as a representative experimental model to study HCMV latency and reactivation. To investigate whether STING in THP-1 cells also engages in nuclear import of HCMV genomic DNA, we generated stably shRNA-expressing THP-1 cells to silence STING. Consistent with our earlier observations, STING silencing did not affect the initial viral entry into THP-1 cells (Fig. 6A, Left). By contrast, the amount of nuclear-imported viral genome was twofold lower in STING-silenced cells 4 hpi than in the control cells (Fig. 6A, Middle). Silencing of STING in U937 cells, another human monocytic cell line, also led to a decrease of viral DNA accumulation in the nucleus, while the total amount of intracellular viral DNA did not differ from that in the control cells (Fig. 6B). Hence, STING mediates nuclear import of the viral genome also in monocytic cells.

Fig. 6.

Decreased viral genome delivery in STING-silenced monocytes leads to impaired establishment of latency and reactivation. (A and B) gDNA-qPCR analysis of total and nuclear fraction of shRNA-THP-1 cells (A) or U937 cells (B) after HCMV infection (MOI = 2), harvested at 1 hpi (total fraction) or 4 hpi (nuclear fraction) (Left). Immunoblot of fractionated cell lysates used for gDNA extraction with indicated antibodies (Right). C, cytoplasm; M, membrane; N, nucleus. (C) Schematic of the time flow for HCMV latency/reactivation in THP-1 model cells (Upper). RT-PCR analysis to detect the expression of viral transcripts during latency and reactivation by PMA treatment (Lower). (D) Total genomic DNA-qPCR analysis extracted from each shRNA-THP-1 cells infected with 2 MOI HCMV and harvested at each time point. (E) Immunoblot of viral IE1/2 proteins during latency and following reactivation upon PMA treatment. qPCR data were normalized to human MDM2 DNA in the host genome. Mean ± SEM, n = 3; *P < 0.05, **P < 0.01, ns; not significant (two-tailed unpaired Student’s t test). For immunoblot and RT-PCR, one of three biological replicates is representatively shown.

Treatment with phorbol 12-myrisate 13-acetate (PMA) can induce reactivation of latent HCMV infection, and PMA induces cell differentiation into macrophage-like cells (35, 36). To examine whether the decreased efficacy of nuclear import of the viral genome in STING-silenced THP-1 cell lines leads to an unsuccessful establishment of latency and/or reactivation, we followed an approach described in previous studies (37–39). We first analyzed the expression of several marker genes to test the HCMV latency and reactivation model in THP-1 cells. Expression of the HCMV UL138 gene, whose product is involved in inhibiting viral replication and promoting latency (40), was inhibited by PMA treatment, as anticipated (Fig. 6C). This was independent of the presence of STING, although UL138 expression in STING-depleted cells was much lower than that in control shLuc-THP-1 cells during the infection. HcmvIL-10 is a reactivation marker that is expressed only when the cells are treated with PMA. As expected, it was also expressed in the absence of STING, albeit at delayed and lower levels than those in control cells (Fig. 6C). Finally, LUNA mRNA was detected throughout the course of infection in shLuc cells but not in shSTING cells, confirming that LUNA is a latency-associated transcript (Fig. 6C) (41). The expression analysis thus confirmed that the HCMV latency and reactivation can be modeled in THP-1 cells. The amount of viral genomic DNA accumulated in the nucleus of STING-depleted THP-1 cells was ∼50% that in the shLuc control cells but remained stable until 9 d after the infection (Fig. 6D). In the shLuc control cells, the levels of IE1 and IE2, a master regulator of HCMV lytic infection, gradually decreased over 5 d after HCMV infection, which was consistent with the establishment of latency. By contrast, IE1 and IE2 protein levels were decreased in the STING-silenced cells at 1 dpi and until 5 dpi (Fig. 6E, Left), consistent with the reduced levels of nuclear viral DNA (Fig. 6A). When differentiation and reactivation were induced by PMA, the IE1 levels rapidly increased in shLuc control cells but were negligible in the shSTING cells (Fig. 6E, Right). These observations suggest that STING silencing reduces the ability of HCMV to establish latency. Collectively, our findings indicate that STING is a crucial host factor that is important for productive HCMV infection not only in the fibroblast and epithelial cell models but also in monocytes, which are physiological reservoirs for herpesvirus persistence.

Discussion

Here, we showed that in addition to its well-known antiviral effect, STING has a proviral effect affecting the nuclear entry of the HCMV and HSV-1 genome in the infection. The two effects of STING are distinct and separable.

We identified STING as a host factor involved in the regulation of herpesvirus infection. Indeed, the expression pattern of STING dictated herpesvirus cell susceptibility. Immortalized HFF-TEL cells showed increased expression of the STING protein, correlating with increased cell susceptibility to HCMV compared with the parental HFF-TELs (Fig. 5 and SI Appendix, Fig. S8). Overexpression of STING in A549 cells also enhanced cellular susceptibility to HCMV, while the KO of STING decreased HSV-1 viral gene expression (Fig. 5), indicating that STING determines cell viral susceptibility. Increased cellular susceptibility or permissiveness to HCMV has been detected previously using certain experimental approaches. Harvey-ras (H-ras) or adenovirus E1A/E1B-transformed cells are more permissive to HCMV than their parental nontransformed cells (42–44). Chemicals, such as 5-iodo-2′-deoxyuridine and 12-O-tetradecanoyl-phorbol-13-acetate, or lipopolysaccharide treatment also convert nonpermissive cells into HCMV permissive cells (45–47). Furthermore, p53 KO decreases the ability of HCMV to replicate in fibroblasts (48). The detailed mechanisms of how these stimuli influence susceptibility or permissiveness to HCMV are still unclear, but their effect on the cellular levels of STING should be investigated.

STING is an ancient protein that has been conserved for over 600 million years (49). This implies that STING and the viruses have coevolved so that the viruses employ STING to efficiently infect and replicate in the host cells, while STING detects infection and represses viral replication. The dual proviral and antiviral role of STING can be an outcome of the arms race between STING and viruses during evolution. In support of the proviral role of STING at the early stage of herpesvirus infection, viral transcript levels upon murine cytomegalovirus infection are reduced in STING-KO mice within 6 hpi (24) According to another study, STING is a proviral host factor for human rhinoviruses independent of its canonical innate immune functions (25), although the rhinoviruses are RNA viruses and thus, likely, a different mechanism may be involved from that described in the current study for HCMV.

Here, we have uncovered a role of the nuclear membrane-localized STING in HCMV infection. STING is not only a cytoplasmic protein but also a nuclear membrane protein. Actually, there have been several reports on tentative STING location to the outer nuclear membrane (50–52). We observed a complete coimmunostaining of endogenous STING with Nesprin-1 (outer nuclear membrane protein) and Lamin A/C (inner surface protein of the nuclear lamina) at the nuclear lamina, similar to Calnexin (an ER membrane protein) (SI Appendix, Fig. S6). According to a recent study, STING can locate to the inner nuclear membrane, which was supported by Lamin B1 costaining and immunoelectron microscopy (53). Taken together with our findings that the viral capsid and Nup358 directly bind to each other and that STING mediates their interaction, promoting viral nuclear entry, nuclear membrane–localized STING plays a proviral role at the postentry stage of HCMV infection. Whether another cofactor or modifications, such as ubiquitination, phosphorylation, or SUMOylation, is required for the tripartite interaction and the following nuclear import of the viral genome should be further investigated.

Studying the molecular mechanism of HCMV infection in myeloid cells, where it establishes latency is important in the physiological context. The postentry phase of HCMV infection in monocytes is much longer than that in fibroblasts (54). UL82 (pp71) is abundant in the monocyte cytoplasm and may contribute to the avoidance of STING activation and the subsequent DNA sensing of the incoming latent genomes (55–58). A recent study describing a probabilistic bet hedging strategy in HCMV latency emphasized the central role of pp71 in viral silencing and establishment of latency (59). In line with this, we suggest that HCMV virions have evolved to actively exploit STING as a platform for entering the nucleus in an opportunistic manner while evading the innate immune response via cytoplasmic pp71. In the current study, we found that STING is involved in regulating HCMV latency and reactivation. However, we have not been able to determine whether this function of STING is directly related to STING-mediated nuclear import of the HCMV genome or its immune function. Nonetheless, in this study, a proviral role of STING in the immediate–early phase of herpesvirus infection, which is cell-type independent, has been definitely identified. We expect that this finding will offer a perspective for understanding herpesvirus uncoating mechanism and developing therapeutics against herpesvirus infection.

Materials and Methods

Cells and Viruses.

Primary HFF, HFF-TEL, human embryonic kidney (HEK) 293T, U373MG, A549, HepG2, U937, and THP-1 cells were obtained from the American Type Culture Collection and cultured in complete Dulbecco’s modified Eagle’s medium (DMEM) or Roswell Park Memorial Institute Medium (RPMI 1640) supplemented with 10% fetal bovine serum (FBS), 2 mM GlutaMAX-I, 50 U/mL penicillin, and 50 μg/mL streptomycin at 37 °C in the presence of 5% CO₂. HFF cells were trypsinized, resuspended, and plated at near 90 to 95% confluence for infection. All infections were performed with the HCMV Towne_long (Towne varL) strain with various MOIs as indicated. Infectious virus particles of HCMV were generated from primary HFF cells. About 2 wk after 0.01 to 0.05 MOI infection showing a 100% cytopathic effect, cell culture supernatants were harvested and centrifuged to remove cell debris. To obtain cell-associated virions, the cell pellet was sonicated, centrifuged, and the additional supernatants were summed with previously collected supernatants. The aliquots of viral stocks were stored at −80 °C. The viral stocks were titrated by immunofluorescence analysis using IE1/2-specific antibody. Complete media was used for mock infection. After virus absorption for 30 min, virus-containing medium was immediately replaced by complete DMEM. U937 and THP-1 cells were infected with 2 MOI, and cell suspension was mixed by rocking every 30 min for 4 h, followed immediately by centrifugation at 450 × g for 20 to 30 min at room temperature. For subcellular fractionation, U937 and THP-1 cells were harvested 4 h later, washed with 1×PBS twice, and fractionated as depicted protocol. During the 5-d latency period, infected cells proliferated. THP-1 cells were treated with 100 nM PMA for reactivation after PBS wash and changing with complete media. Cells were harvested at the indicated postinfection times after washing with 1×PBS at least twice.

HFFs and U373MG cells were transfected with each gene-targeting siRNA using Dharmafect 1 transfection reagent with 30 nM final concentration. To generate stable knockdown or KO cell lines, we transduced lentiviruses harvested from the culture supernatant of HEK293T cells transfected with pLKO1-puro/neo-shSTING or LentriCRISPR.v2-puro-sgSTING vectors, psPAX2, and pMD2.G using polyethylenimine (PEI) reagent. Detailed procedures are described in SI Appendix, SI Materials and Methods.

Other Materials and Methods.

Other materials and methods used in this study are described in SI Appendix, SI Materials and Methods.

Data Availability

All study data are included in the article and/or SI Appendix.