Guinea pig cytomegalovirus trimer complex gH/gL/gO uses PDGFRA as universal receptor for cell fusion and entry

Abstract

Species-specific guinea pig cytomegalovirus (GPCMV) causes congenital CMV and the virus encodes homolog glycoprotein complexes to human CMV, including gH-based trimer (gH/gL/gO) and pentamer-complex (PC). Platelet-derived growth factor receptor alpha (gpPDGFRA), only present on fibroblast cells, was identified via CRISPR as the putative receptor for PC-independent GPCMV infection. Immunoprecipitation assays demonstrated direct interaction of gH/gL/gO with gpPDGFRA but not in absence of gO. Expression of viral gB also resulted in precipitation of gB/gH/gL/gO/gpPDGFRA complex. Cell-cell fusion assays determined that expression of gpPDGFRA and gH/gL/gO in adjacent cells enabled cell fusion, which was not enhanced by gB. N-linked gpPDGFRA glycosylation inhibition had limited effect and blocking tyrosine kinase (TK) transduction had no impact on infection. Ectopically expressed gpPDGFRA or TK-domain mutant in trophoblast or epithelial cells previously non-susceptible to GPCMV(PC−) enabled viral infection. In contrast, transient human PDGFRA expression did not complement GPCMV(PC−) infection, a potential basis for viral species specificity.

Introduction

Human cytomegalovirus (HCMV) is a betaherpesvirus and congenital CMV (cCMV) is a leading cause of sensorineural hearing loss (SNHL) and cognitive impairment in newborns [1]. There is no prevention or intervention strategy currently available and a vaccine against cCMV is a high priority [2]. Although various animal CMV models are available for research, the guinea pig is unique as the only small animal model for cCMV because CMV does not cross the placenta in the mouse or rat [3]. Congenital infection in the guinea pig recapitulates human disease in utero and causes similar symptomology in newborn pups, such as SNHL [4–6]. Thus, the guinea pig CMV model is well-suited for investigating vaccine or interventional strategies against cCMV.

HCMV entry into all cell types requires viral glycoproteins gB and gM/gN [7–9] and entry into fibroblasts via direct fusion also requires the gH/gL/gO complex [10–12]. Additionally, HCMV infection of non-fibroblast cells such as epithelial, endothelial, and myeloid cells is via an endocytic pathway [13, 14] and also requires a second gH-based glycoprotein complex, the pentamer complex (PC). The PC is present in HCMV clinical strains but absent in lab adapted virus [15–17]. Indeed, HCMV isolates from clinical samples have been shown to undergo mutation when cultured on fibroblasts, resulting in loss of an ability to express some or all of the unique components of the PC [18, 19]. Various viral glycoprotein complexes, including gB and the gH-based complexes (gH/gL/gO and PC), are important neutralizing antibody target antigens and consequently, prospective vaccine candidates [20–24]. Studies in the guinea pig model require species specific guinea pig CMV (GPCMV) [25]. Importantly, GPCMV encodes functionally similar glycoprotein complexes, which are essential for cell entry [26–28], including a homologous gH-based PC (gH/gL/GP129/GP131/GP133), which has been demonstrated necessary for infection of guinea pig renal epithelial, trophoblast, amniotic sac and macrophage cells [29–32]. The PC is also critical for GPCMV dissemination in the animal and congenital infection [30, 31].

Although there is a better understanding of viral glycoprotein complexes in HCMV, less is known about the cell receptors for virus entry in different cell types or pathways. Epithelial growth factor receptor (EGFR) and platelet-derived growth factor receptor alpha (PDGFRA) have been reported to facilitate HCMV entry into cells [33, 34]. However, their specific roles have been questioned in subsequent studies [35–37]. PDGFRA was recently identified as the cellular receptor for HCMV infection of fibroblast cells [38], where entry is dependent upon the gH/gL/gO trimer in conjunction with gB [38–40]. Gene knockout of PDGFRA in human fibroblast cells was shown to render the virus dependent upon the pentamer/ endocytic route of cell entry [38, 40]. Similarly, the absence of PDGFRA, or low-level expression, on epithelial cells and other non-fibroblast cell types renders the virus highly dependent on the PC for cell entry. However, PC endocytic cell entry is poorly defined. A recent study has shown that knockdown of EGFR reduces HCMV entry into epithelial cells [38]. However, the presence or absence of EGFR on various cell types (fibroblast, epithelial and endothelial) did not appear to influence HCMV infection in other studies [35, 40]. Additionally, neutralizing antibodies to EGFR did not block HCMV infection of EGFR positive cells [35]. EGFR continues to be cited as important for HCMV cell entry [41–43] following a report that viral gB interacts with cellular EGFR to mediate cell entry [33]. In non-fibroblast cells, additional candidate receptors (e.g. NRP2) have been identified in association with PC endocytic entry pathway [44–48].

Both EGFR and PDGFRA have a tyrosine kinase (TK) domain, and ligand binding to these receptors (EGF and PDGF respectively) results in transduction pathway activation [49]. Previous HCMV studies suggested that viral TK activation by virion interaction was necessary for cell entry [33, 34], but this has been refuted in other studies [35]. It has been reported that HCMV utilizes EGFR and beta-integrins on the surface of monocytes as co-receptors for viral entry, and that the subsequent signaling drives a unique nuclear translocation pattern for the virus in monocytes, a site of latent infection, compared to cells that are permissive for lytic infection such as fibroblasts and endothelial cells [50]. This distinction between cell lines may explain why other studies were unable to determine a role for EGFR in HCMV entry into fibroblast, epithelial, or endothelial cell lines, and that HCMV did not activate the EGFR kinase [35]. Others have suggested that the PDGFRA transduction enhanced HCMV infection [37]. However, this receptor is not present in most epithelial or endothelial cells [39, 40].

We recently identified guinea pig PDGFRA as the likely receptor for guinea pig fibroblast infection by CRISPR/Cas9 gene knockout of PDGFRA [28]. In this report, we extend those findings with a series of studies conducted to evaluate the ability of PDGFRA to interact with GPCMV glycoproteins, function in virus mediated cell-cell fusion, and act as a universal cognate receptor for GPCMV entry and infection in various cell types. Ectopic expression of guinea pig PDGFRA but not human PDGFRA enabled PC-independent infection of various non-fibroblast cells. Additionally, cellular inhibition of PDGFRA N-linked glycosylation only affected GPCMV infection at low multiplicity of infection (moi), whereas blocking cellular TK transduction did not impact on the ability of GPCMV to infect cells. Overall, results continue to demonstrate important similarities to HCMV and the translational relevance of GPCMV studies. Importantly, guinea pig PDGFRA was capable of acting as a universal receptor on various cell types.

Results

GPCMV(PC+) versus GPCMV(PC−) infection in PDGFRA knockout fibroblasts

PDGFRA is expressed by guinea pig fibroblast cells (GPL) but not expressed in guinea pig renal epithelial (REPI), trophoblasts (TEPI), or amniotic sac epithelial cells [28, 32]. A CRISPR/Cas9 strategy generated a PDGFRA knockout cell line (designated GPKO) [28] and GPCMV growth curves demonstrated that GPCMV(PC−) virus infection of fibroblasts was dependent upon PDGFRA, where absence of the receptor resulted in lack of productive (PC−) virus infection [28]. GPCMV(PC−) lacks the unique PC components of GP129, GP131 and GP133 [51]. Preliminary studies did not evaluate (PC−) virus cell entry by staining for viral antigens. To further determine the role of PDGFRA in viral infection, GPKO (PDGFRA−) and GPL (PDGFRA+) fibroblasts in six well plates were infected with wild-type GPCMV(PC+) or GPCMV(PC−) (moi 1 pfu/cell) and subsequently stained for GPCMV immediate early (IE2) protein [52] at 24 h post infection. While GPCMV IE2 protein was detected in GPMCV(PC+) infected GPKO cells, viral antigens failed to be detected in GPKO cells infected with GPCMV(PC−) (Fig 1(i), panels A and B, respectively). In contrast, IE2 protein was detected in GPL cells infected with either (PC+) or (PC−) virus (Fig 1(i), panels G and H, respectively). In a parallel set of experiments, GPKO and GPL monolayers were infected with GPCMV (PC+) or (PC−) virus (moi 1 pfu/cell), and viral growth at 4 days post infection (dpi) was evaluated by titration onto GPL cells in duplicate, see Fig 1(ii–iii). GPCMV(PC+) produced robust infection in GPKO cells, while (PC−) virus was unable to grow (Fig 1(ii)). In comparison, GPCMV (PC+) and (PC−) growth was similar in GPL cells (Fig 1(iii)). Real-time, bright-field images at 4 dpi demonstrated the presence or absence of cytopathic effect (CPE) associated with viral infection (Fig S1). These results demonstrate that fibroblasts have two routes of GPCMV entry: PC-dependent and PC-independent. Epithelial cells are limited to PC-dependent endocytic pathway [13, 30]. PDGFRA is dispensable for GPCMV(PC+) infection, but the receptor is essential for GPCMV(PC−) entry into fibroblasts for viral protein expression, similar to HCMV [38–40].

Fig 1. PDGFRA is essential for GPMCV(PC−) infection of fibroblasts, but not required for GPCMV(PC+) infection.

(i) GPKO and GPL fibroblasts were infected with GPCMV(PC+) or GPCMV(PC−) virus (moi 1 pfu/cell) in separate wells of six well plates. Immunostaining for GPCMV IE2 protein was performed as previously described [30, 32]. Mock infected (MI) GPKO and GPL cell monolayers served as controls. A-C, GPKO stained for GPCMV IE2 (GP122). D-F, Secondary antibody only control stain of GPKO monolayers. G-I, GPL stained for GPCMV IE2. J-L, Secondary antibody only control stain of GPL monolayers. Real-time, bright-field images were taken at 10X magnification. Individual images representative of multiple fields. (ii) Comparative infection (moi 1 pfu/cell) of (PC+) or (PC−) virus on GPKO fibroblasts 4 days post infection (dpi). Red, (PC+). Navy, (PC−). (iii) Growth of (PC+) or (PC−) virus on GPL cells at 4 dpi. Green, (PC+). Blue, (PC−). Viral titers represented as mean ± SD.

Cellular N-glycosylation inhibition and GPCMV(PC−) infection

Cellular and viral proteins have the ability to be post-translationally modified by glycosylation. Potentially, glycosylation state may alter the ability of cellular protein to function as a viral receptor, as noted for various cellular receptors for a number of viruses [53–56]. Similar to human, guinea pig PDGFRA protein is predicted to be post-translationally modified by N-linked glycosylation. Guinea pig PDGFRA protein has six conserved N-linked glycosylation sites also found in human PDGFRA. Inhibition of N-linked glycosylation of host cell glycoproteins including PDGFRA was evaluated with glycosylation inhibitor tunicamycin (TM). The predicted size of guinea pig PDGFRA is 123 kDa. In untreated cells, two molecular weight species of PDGFRA were detected: 180 kDa and 123 kDa. However, in the presence of TM (1 μg/mL), full-length PDGFRA was detected as a single molecular weight species of 123 kDa (Fig 2(i)). We concluded that the higher molecular weight was due to protein glycosylation while the lower molecular weight species was non-glycosylated, as previously noted for GPCMV glycoproteins [26, 57]. Next, GPL cells pretreated (48 h) with various concentrations of TM (0–1.25 μg/mL) were infected with GFP tagged (PC−) virus (moi 1 pfu/cell) [58], and GFP expression was monitored as a real-time marker for viral infection. Mock infected controls were treated with corresponding TM concentrations. All cells of infected groups expressed GFP by 1 dpi (Fig 2(ii)), regardless of the presence or absence of TM. Viral titer at 4 dpi was evaluated for a subset of three TM-treated monolayers (0, 0.31, 1.25 μg/mL), and no significant difference was observed (data not shown). These results suggest that inhibition of N-linked glycosylation of PDGFRA, and potentially other co-receptors, had no impact on GPCMV infection of fibroblasts at moi of 1 pfu/cell.

Fig 2. Impact of PDGFRA glycosylation on cellular susceptibility to GPCMV(PC−).

(i) Western blot of gpPDGFRA expression in GPL cells in the presence (1 μg/mL) or absence of tunicamycin (TM). (ii) GPL cells were pretreated with or without TM as indicated (1.25 μg/mL, 0.63 μg/mL, 0.31 μg/mL, 0.16 μg/mL or 0 μg/mL) for 48 h prior to infection with GFP-tagged GPCMV(PC−) at moi = 1 pfu/cell. (iii) Viral GFP expression of infected GPL cells either pretreated with 1.25 μg/mL TM or without TM for 48 h prior to infection with GFP-tagged GPCMV(PC−) at moi = 0.01, 0.1, or 1.0 pfu/cell. Real-time GFP cell field images images (ii) and (iii) were obtained 36 hpi under fluorescence microscopy at 4X magnification for viral GFP expression. MI = mock infected control monolayer in absence of TM. Individual images are representative of 10 independent fields. Experiments were repeated three times.

Experiments were additionally carried out on six well plates at various moi (1, 0.1 and 0.01 pfu/cell) on cell monolayers pretreated with TM (1.25 μg/mL) or no treatment. Overall, results shown in Fig 2(iii) suggested that inhibition of N-glycosylation of PDGFFRA has a greater impact at lower moi (0.1 and 0.01 pfu/cell compared to 1 pfu/cell) based on comparative cellular GFP tagged GPCMV monolayers in the presence and absence of cellular TM treatment at equivalent moi. The greatest impact on infection was at the lowest moi (0.01 pfu/cell) but infection was also impacted on TM treated cells at 0.1 pfu/cell moi compared to non-treated cells, Fig 2(iii). In HCMV, the impact of TM treatment of cells and virus infection has only been evaluated at low moi (0.05 pfu/cell) on fibroblast cells where there was an impairment of virus infection compared to untreated cells [59]. Impact of inhibition of N-glycosylation on fibroblast infection over a range of different moi awaits further evaluation for HCMV as well as studies on other cell types.

GPCMV gH/gL/gO specifically interacts with PDGFRA

Studies in HCMV have demonstrated that the gH/gL/gO trimer, but not gH/gL, binds to PDGFRA [38, 39]. GPCMV GP75 (gH), GP115 (gL), and GP74 (gO) form homologous gH/gL/gO and gH/gL complexes [26]. Consequently, we evaluated the ability of GPCMV proteins to interact with guinea pig PDGFRA (gpPDGFRA). Specific interactions between gH, gL, gO and PDGFRA were evaluated by transient expression studies in six well plates of GPKO fibroblast cells (PDGFRA-). Recombinant defective adenoviruses encoding GPCMV gH (AdgHGFP, GP75 GFP C-terminal tag) [26] and gL (AdgLmyc, GP115 myc C-terminal tag) [26] were used in conjunction with gO (pGP74TAG8, GP74 FLAG C-terminal tag) [26] and gpPDGFRA (pgpPDGFRA) [28] expression plasmids. GPKO cells were transfected with gpPDGFRA expression plasmid and either transduced with: AdgHGFP only (Fig 3(i)); AdgHGFP and AdgLmyc (Fig 3(ii)); or AdgHGFP and AdgLmyc, and transfected with gOFLAG expression plasmid (pGP74TAG8) (Fig 3(iii)). Immunoprecipitation (IP) was carried out with GFP-Trap (Chromotek) to pulldown all proteins that interact with GFP-tagged gH, as described previously [26]. IP reactions with gHGFP and PDGFRA, or with gHGFP, gL and PDGFRA, failed to pulldown PDGFRA (Fig 3(i–ii)). In contrast, transient expression of all three components of the gH trimer (gHGFP, gL, and gO) resulted in a specific IP of PDGFRA (Fig 3(iii)). These results confirmed that PDGFRA is capable of binding the GPCMV glycoprotein gH/gL/gO trimer and that gH/gL complex by itself is insufficient to interact with PDGFRA in transient expression IP assay. Additional expression of full-length gB [57] with gH, gL, gO and PDGFRA, also resulted in detectable levels of gB protein in a gHGFP IP assay (Fig 3(iv)). However, gB and PDGFRA did not directly interact, as transient expression failed to pull down PDGFRA when immunoprecipitated with anti-gB monoclonal antibody [60] (data not shown). We concluded that gB could indirectly interact with PDGFRA via complex formation with gH trimer, but gB was not necessary for specific interaction with PDGFRA. Importantly, gO was essential for gH/gL interaction with PDGFRA.

Fig 3. GPCMV gH/gL/gO and PDGFRA immunoprecipitation (IP) assays.

All IPs were performed with GFP-Trap (ChromoTek) as described in materials and methods [26] to immunoprecipitate proteins that interacted with gHGFP. (i) Co-expression and IP for GPKO fibroblasts, transfected with gpPDGFRA expression plasmid and transduced with recombinant adenovirus encoding gHGFP. Lanes 1 and 5 are total cell lysate (TCL); 2 and 6, flow through (FT); 3 and 7, IP reactions; 4 and 8, mock cell lysate. (ii) gHGFP, gLmyc and gpPDGFRA co-expression and IP. Western blot lanes 1, 5 and 9 are TCL; 2, 6 and 10, FT; 3, 7 and 11, IP reactions; 4, 8 and 12, mock cell lysate. (iii) gHGFP, gLmyc, gOFLAG and gpPDGFRA co-expression and IP. Western blot lanes 1, 5, 9 and 13 are TCL; 2, 6, 10, and 14, FT; 3, 7, 11, and 15, IP reactions; 4, 8, 12, and 16, mock cell lysate. (iv) gHGFP, gLmyc, gOFLAG, gpPDGFRA, and gB co-expression and IP. Western blot lanes 1 and 5 are TCL; 2 and 6, FT; 3 and 7, IP reactions; 4 and 8, mock cell lysate. Analysis was performed by Western blot using respective primary antibodies: gH (anti-GFP); gL (anti-myc); gO (anti-FLAG); gpPDGFRA (anti-PDGFRA); gB (anti-gB), and appropriate secondary antibody-HRP conjugate as described in methods section.

Transient expression of GPCMV gH/gL/gO and PDGFRA mediates cell-cell fusion

HCMV infection of fibroblast cells is via virion membrane fusion with the cell membrane [61]. To examine the ability of GPCMV gH/gL/gO trimer and guinea pig PDGFRA to induce fusion, we assayed for syncytium formation between cells with different background fluorescent reporter gene expression (GFP or mCherry) and trimer or PDGFRA expression. GPKO cells expressed a cytoplasmic GFP reporter gene (GFP+/PDGFRA-). GPL cells expressed a nuclear mCherry protein (mCherry+/PDGFRA+). GFP and mCherry cell lines were generated as described in Materials and Methods by permanent transfection with plasmids pEGFP-C1 (Clontech) or pmCherryNLS (Addgene) [62]. Fusion between different background fluorescent cells would result in syncytial (fused) cells with GFP cytoplasm and mCherry nuclei. GPL(mCherry+/PDGFRA+) cells were plated in a 1:1 ratio with either untreated control GPKO(GFP+) cells, or GPKO(GFP+) fibroblasts expressing GPCMV gH/gL/gO and incubated 48 h for syncytial formation. Real-time images were obtained via bright-field and fluorescent microscopy at 10X (Fig 4 A–H) and 20X magnification (Fig 4 I–L). Untreated controls (Fig 4 A–D) failed to form multinucleated syncytia, and distinctly separate red and green cells can be seen in overlay merge image Fig 4D. In contrast, syncytial formation between gH/gL/gO+ GPKO(GFP+) cells and GPL(mCherry+/PDGFRA+) cells was visualized as large fused cells with multiple red nuclei and green cytoplasm (Fig 4 E–L). The number of nuclei included in syncytia ranged from 5 to 47 nuclei (mean = 10.5 nuclei) based on sample of 10 fields. Syncytia containing 10 to 30 nuclei/syncytium accounted for 40% of total syncytia. Expression of gH or gH and gL failed to enable cell-cell fusion. Control experiments with GPKO(GFP+) or GPL(mCherry+/PDGFRA+) cells cultured separately demonstrated that there was no cross bleed of fluorescence between channels and no spontaneous syncytia was formed (Fig S2). Overall, these findings show that GPCMV gH/gL/gO has fusogenic activity in the presence of PDGFRA expressed by adjacent cells. In contrast, expression of gB or gB and gH/gL in GPKO(GFP+) cells mixed with GPL(mCherry+/PDGFRA+) cells failed to result in cell fusion (Fig S3). Extended over expression of gB resulted in extensive cellular vacuolization and eventual cell rupture (96 h) but not cell-cell fusion of red and green cells. Cell rupture at high level Ad vector cell transduction was also observed for over expression of pp65 homolog (GP83) expressed by defective Ad vector in transduced GPL cells (data not shown). However, the GP83 protein [63] has no role in glycoprotein mediated cell fusion and so observed gB cell lysis at high Ad vector cell transduction is likely not due to possible fusogenic activity of gB. However, a potential role for gB in cell fusion cannot be entirely ruled out given the essential nature of this glycoprotein and potential ability to interact with other cell receptors. Although, within the context of specific interaction of GPCMV gHgLgO trimer with PDGFRA, it would appear that gB is non-essential. Overall, results indicated that GPCMV gO is required for gH/gL interaction with PDGFRA and cell fusion. The results also explained an earlier finding that GPCMV gO was only essential for (PC−) virus infection of GPL cells, but not (PC+) virus in GPCMV glycoprotein gene mutagenesis studies [26, 30].

Fig 4. Mixture of GPL(mCherry+/PDGFRA+) and gH/gL/gO+ GPKO(GFP+/PDGFRA-) fibroblasts results in cell-to-cell fusion.

GPKO(GFP+) cells were transfected with pGP74TAG8 and transduced with AdgHGFP and AdgLmyc in order to express GPCMV gH/gL/gO trimer prior to co-culture with GPL(mCherry+/PDGFRA+) fibroblasts. Untreated control GPKO(GFP+) cells were co-cultured with GPL(mCherry+/PDGFRA+) fibroblasts in separate wells. A-D, Mock infected GPKO(GFP+) and GPL(mCherry+/PDGFRA+) fibroblasts. E-H, gH/gL/gO+ GPKO(GFP+) cells co-cultured with GPL(mCherry+/PDGFRA+) fibroblasts. Arrows indicate cell-cell fusion demonstrated by multinucleated (red) syncytia with green cytoplasm. I-L, Additional syncytial formation between gH/gL/gO+ GPKO(GFP+) cells and GPL(mCherry+/PDGFRA+) cells at 20X magnification. Real-time fluorescent microscopy images shown are representative of multiple fields.

Ectopic expression of PDGFRA and GPCMV(PC−) infection of non-fibroblasts

Since PDGFRA was absent in cell lines that required PC for cell entry, we decided to test a hypothesis that PDGFRA could act as a universal receptor for GPCMV(PC−) cell entry on various cell types that enabled lytic GPCMV replication by transient expression of PDGFRA prior to GPCMV(PC−) infection to overcome the barrier to (PC−) virus. Western blot analysis of cell lysates of renal epithelial and trophoblast cells confirmed that PDGFRA was not expressed by these cell types (Fig 5(i)). Transient expression of gpPDGFRA by plasmid transfection of renal epithelial (REPI) and trophoblast (TEPI) cells was verified by Western blot of cell lysate for PDGFRA at 24 h post transfection (Fig 5(iii)). Subsequently, REPI, TEPI, GPKO and control GPL cells were transfected with gpPDGFRA plasmid, or pmCherry-C1 control plasmid (Clontech). Replication of GPCMV(PC−) on plasmid transfected monolayers was compared to growth of GPCMV(PC+) on non-transfected cells, as (PC+) virus can infect all cell types (Fig 5(v)) [28, 30, 31]. Cells were transfected with plasmid on day −1 and infected with virus (moi 1 pfu/cell) on day 0. At 4 days post infection, monolayers in six well plates were individually harvested and viral titers determined on GPL cells. Transient expression of gpPDGFRA on REPI, TEPI, and GPKO cells resulted in productive infection of GPCMV(PC−) in cells that were previously non-susceptible to (PC−) virus. There was a minimal increase in progeny virus for PDGFRA plasmid transfected GPL cells compared to control plasmid. Control mCherry vector transfections did not enhance virus infection, and virus was only capable of growth on GPL cells. GPCMV(PC+) was able to infect all cell types without requirement of ectopic PDGFRA expression (Fig 5(v)). In an additional experiment, TEPI cells were transfected with plasmid expressing PDGFRA and mCherry (pLVXPDGFRA-IRESmCherry) and subsequently infected with GFP-tagged GPCMV(PC−) virus [58] (moi 1 pfu/cell), which enabled co-localization of cells expressing PDGFRA (mCherry) (Fig 5(vi) A) and infected with GFP virus (Fig 5(vi) B). Cells expressing mCherry and GFP appear yellow in an overlay image (Fig 5(vi) C). This co-localization demonstrated that ectopic PDGFRA expression was sufficient for GPCMV(PC−) entry into cells that were normally only permissive for (PC+) virus, as control mCherry expression failed to enable GPCMV infection (Figure 5(v)).

Fig 5. Ectopic expression of guinea pig PDGFRA in renal epithelial and trophoblast cells enables GPCMV(PC−) infection.

(i) Western blot analysis of cell lysates for endogenous PDGFRA expression in various guinea pig cell lines. Lanes, total cell lysates: 1, GPL; 2, REPI; 3, TEPI. (ii) Beta-actin control for cell lysate protein loading for (i) above. (iii) Ectopic expression of gpPDGFRA in: 1, GPL; 2, REPI; 3, TEPI cells. (iv) Beta-actin control for protein loading in (iii) above. (v) Complementation of GPCMV(PC−) virus by gpPDGFRA. Cells (REPI, TEPI, GPKO, and GPL fibroblasts) were transfected with transient expression plasmid for gpPDGFRA or a vector only control (pmCherry-C1) 24 h prior to GPCMV(PC−) infection (moi 1 pfu/cell). At 4 dpi, cells were harvested and virus titer determined on GPL cells (represented by mean ± SD). (vi) Co-localization of mCherry-tagged gpPDGFRA and GFP-tagged GPCMV(PC−). TEPI cells were transfected (day −1) with pLVXPDGFRA-IRESmCherry then infected with (PC−) virus (moi 1 pfu/cell) at day 0. Real-time images were taken 48 h post infection at 10X magnification. Cells positive for mCherry or GFP are shown in A and B, with overlay shown in panel C.

In order to further define the receptor requirements for GPCMV(PC−) virus entry, guinea pig PDGFRA constructs lacking the kinase domain (position 593–954), and/or the transmembrane domain (position 525–549), but encoding the extracellular domain (position 1–514) were generated (Fig 6(i)), as described in Materials and Methods. The truncated PDGFRA tyrosine kinase deletion plasmid (pgpPTKD) encoded amino acids (aa) 1–801, and the PDGFRA ectodomain-only plasmid (pgpPED) encoded aa 1–514 (Fig 6(i)). Transient plasmid expression of full-length gpPDGFRA (pgpPDGFRA) and the truncated variants in REPI cells were verified by Western blot (Fig 6(ii)). Ectopic expression of full-length gpPDGFRA, gpPTKD, or gpPED in REPI cells was evaluated for enhancement of GPCMV(PC−) infection. Expression of full-length gpPDGFRA and gpPTKD supported (PC−) infection in REPI cells, but PDGFRA encoding only the ED (pgpPED transfection) failed to complement infection (Fig 6(iv)). GPL fibroblasts, which express endogenous PDGFRA, were used as a control for infection and demonstrated susceptibility to (PC−) virus regardless of the PDGFRA variant used for transient expression (Fig 6(iv)). Next, full-length gpPDGFRA or truncated gpPED were transiently expressed in GPKO cells to evaluate GPCMV(PC−) entry in fibroblasts. GPL control fibroblasts demonstrated robust infection 4 dpi, unlike GPKO untreated controls. This would suggest that although the ED of PDGFRA contains the binding region for the gH/gL/gO trimer, the transmembrane domain may be required for proper trafficking of the receptor to the cell surface to enable presentation and interaction with the viral glycoprotein complex, as recently demonstrated for HCMV [38, 39].

Figure 6. Ectopic expression of PDGFRA external domain and transmembrane domain but not TK domain is required for GPCMV(PC−) infection.

(i) Transient expression plasmids encoding either full-length or modified variants of guinea pig PDGFRA were generated: A, Full-length PDGFRA (gpPDGFRA); B, tyrosine kinase domain mutant (gpPTKD); C, tyrosine kinase and transmembrane domain mutant (gpPED). (ii) Expression of full-length and PDGFRA truncated mutants on REPI cells verified by Western blot. Lanes, cell lysates: 1, gpPDGFRA; 2, gpPTKD; 3, gpPED; 4, empty vector cell lysate. (iii) Beta-actin Western control for cell lysate lane loading in (ii) above. (iv) GPCMV(PC−) infection of GPL or REPI cells expressing full-length PDGFRA or truncated protein mutants. (v) GPCMV(PC−) infection of GPKO fibroblasts transfected with gpPDGFRA or gpPED expression plasmids. GPL fibroblasts and GPKO(PDGFRA-) cells demonstrate infection controls. All infections performed at moi = 1 pfu/cell. Cells harvested 4 dpi and titrated in duplicate.

Since in HCMV, PDGFRA is the cell receptor for viral gH/gL/gO [36, 38–40], we next determined if ectopic expression of human PDGFRA in guinea pig cells could overcome species specificity of the virus. There is high conservation between the mature protein for guinea pig and human PDGFRA (95% identity) [28], and a shared identity of 81% within the ED based on BLAST (https://www.uniprot.org/blast/uniprot/B201910106746803381A1F0E0DB47453E0216320D00B 695Q). A transient expression plasmid encoding the human PDGFRA predicted ORF (huPDGFRA) with a C-terminal His tag (Sino Biological Inc.) was used in these studies. Western blot with anti-PDGFRA antibody verified expression of huPDGFRA protein in plasmid transfected REPI and TEPI cells (Fig 7(i)). Next, REPI, TEPI, GPKO, and GPL cells were transfected with either gpPDGFRA or huPDGFRA expression plasmids. GPCMV(PC−) infection was supported by ectopic expression of gpPDGFRA in various cells, but in contrast, ectopic huPDGFRA was unable to complement (PC−) virus infection (Fig 7(iii)). Control GPL cells, which express endogenous guinea pig PDGFRA, supported GPCMV(PC−) infection under both transfection conditions. An additional experiment was carried out to evaluate surface expression of huPDGFRA in transiently expressed plasmid transfected REPI cells compared to endogenous levels present on human fibroblast (HFF) cells. Flow cytometry analysis was able to detect high level expression of of huPDGFRA on HFF cells (Fig S6A–C) and a lower level of huPDGFRA expressed on the surface of plasmid transfected REPI cells at 36 h post transfection (Fig S6D–F). Although the level of surface huPDGFRA on REPI cells was lower than that of endogenous levels on HFF cells, this should have been sufficient to enable some permissive GPCMV(PC−) virus infection. The complete lack of detectable progeny virus on huPDGFRA plasmid transfected REPI cells compared to results for transient expression of gpPDGFRA on REPI cells (Fig 7) suggests that huPDGFRA is not a receptor for GPCMV. Potentially, this is because of the low conservation in sequence of gO between HCMV and GPCMV preventing trimer interaction with human PDGFRA [26, 64]. Additionally, the variation in amino acid sequence between species in the Ig-like domain of PDGFRA may limit interaction [40]. However, this is outside the scope of the current study and awaits further evaluation.

Fig 7. Ectopic expression of human PDGFRA on various guinea pig cell types does not complement GPCMV(PC−) infection.

Full-length guinea pig PDGFRA (gpPDGFRA) or human PDGFRA (huPDGFRA) expression plasmids were separately transfected in six well plates of various guinea pig cell lines (REPI, TEPI, GPKO and GPL), and GPCMV(PC−) infection was subsequently evaluated at 24 h post transfection. (i) Western blot analysis demonstrating expression of huPDGFRA in plasmid transfected guinea pig epithelial cells. Lanes, total cell lysates: 1, REPI + huPDGFRA; 2, TEPI + huPDGFRA; 3, control REPI lysate. (ii) Beta actin Western control for lane loading in (i) above. (iii) REPI (1), TEPI (2), GPKO (3), and GPL (4), transfected with either gpPDGFRA or huPDGFRA, were infected with GPCMV(PC−) at moi 1 pfu/cell. Infection of GPL cells served as positive control. At 4 dpi, cells were harvested and virus yield titrated in duplicate on GPL cells. Results plotted as mean viral titer ± SD versus cell type for each plasmid transfection.

EGFR and PDGFRA functional requirements for GPCMV infection

The role of EGFR in CMV entry remains controversial because of a possible requirement for EGFR tyrosine kinase (TK) transduction pathway associated with HCMV entry into certain cell types [33, 38]. Therefore, requirement of TK transduction by EGFR or PDGFRA was evaluated. The predicted mature proteins for guinea pig and human EGFR share 89% identity based on BLAST alignment (https://www.uniprot.org/uniprot/H0UT18). The predicted sizes of human and guinea pig EGFR are both approximately 134 kDa. Guinea pig cell lines generated by our laboratory, including REPI cells [30] and trophoblasts [31], together with GPL fibroblasts and control human 293 cells were tested for EGFR protein expression by Western blot of cell lysates. In contrast to guinea pig PDGFRA [28, 32], EGFR was present in all guinea pig cell lines tested, with a protein of approximately 134 kDa observed (Fig 8(i)). Notably, the intensity was approximately five times lower in GPL fibroblasts than in guinea pig renal epithelial cells, trophoblasts, and control human epithelial 293 cells. This likely indicated variable levels of EGFR expression between cell types, since control beta-actin Western blot showed consistent intensity (Fig 8(ii)). In order to assess whether EGFR TK activation is required for GPCMV infection of fibroblasts and epithelial cells, erlotinib HCl (OSI-744, Selleckchem), a potent EGFR inhibitor [65], was used to block the TK signaling transduction pathway. GPL fibroblasts and REPI cells in six-well plates were either untreated (0 nM), or pre- and post-treated with erlotinib HCl (10 nM, 50 nM, 500 nM, 1 μM, 10μM) before and after infection (moi 0.5 pfu/cell). Both treated and untreated cell monolayers were harvested at 72 h post-infection and titrated in duplicate on GPL cells. Erlotinib HCl treatment of GPL and REPI cells had no effect on infection with (PC+) or (PC−) virus (Fig 8(iii–iv)). Because virus entry pathways differ between fibroblasts (direct fusion) and epithelial cells (endocytosis), these results demonstrated that neither route of GPCMV cell entry required the EGFR TK transduction pathway.

Fig 8. Cellular expression of EGFR and impact of inhibition of TK activity on GPCMV infection.

(i) Western blot analysis for endogenous EGFR expression in various guinea pig cells. Lanes, cell lysates: 1, REPI; 2, TEPI; 3, GPL; 4, human 293 cells (positive control). Analysis was performed with rabbit anti-EGFR (Aviva) primary antibody and secondary anti-rabbit HRP conjugate (Sigma). (ii) Beta-actin control Western for cell lysate lane loading in (i) above. (iii) GPL and REPI cells were pretreated in the absence (0 nM) or presence (10 nM, 50 nM, 500 nM, 1 μM, and 10 μM) of erlotinib HCl (OSI-744) for 1.5 h prior to infection with (PC+) virus (0.5 moi). Inoculum was removed, washed, and replaced with media and corresponding concentration of erlotinib from pretreatment. (iv) GPCMV(PC−) infection of GPL cells treated with indicated concentrations of erlotinib HCl as described above (moi 0.5 pfu/cell). (v) GPL and REPI cells were pretreated in the absence (0 nM) or presence (1 nM, 10 nM, 50 nM, 500 nM) of ponatinib (AP24534) for 1.5 h prior to infection with GPCMV(PC+). Inoculum was removed, washed, and replaced with media and ponatinib at concentrations corresponding to pretreatment. (vi) GPCMV(PC−) infection of GPL cells treated with indicated concentrations of ponatinib as described above. All infections performed at moi 0.5 pfu/cell. Virus growth was evaluated 72 h post infection, titrated in duplicate on GPL cells, and represented as mean ± SD.

Potentially, EGFR TK transduction could be substituted by TK signaling from another cell protein during GPCMV infection. To determine whether any TK signaling (including PDGFRA) is required for GPCMV infection, pharmacological inhibition of TK activity was performed using ponatinib (AP24534, Selleckchem), a multi-targeted tyrosine-kinase inhibitor of Abl, PDGFRA, VEGFR2, FGFR1 and Src [66, 67]. GPL and REPI cells were either untreated (0nM), or pre- and post-treated with ponatinib (1, 10, 50, 100, or 500 nM) before and after infection (moi 0.5 pfu/cell). Both treated and untreated control cell monolayers were harvested at 72 h post-infection and titrated in duplicate on GPL cells. Ponatinib treatment on GPL and REPI cells had no effect on GPCMV (PC+) or (PC−) infection (Fig 8(v–vi)). In both drug treatment assays, neither erlotinib or ponatinib had an effect on cell viability based on separate viability assays. GPL and REPI cell viability assays were performed in the presence and absence of drugs at various concentrations in parallel studies for the same duration as infection studies but in the absence of virus. Overall, it was concluded that TK activity is dispensable for GPCMV infection on all cell types that support lytic replication. However, it does not rule out the possibility that EGFR functions as a co-receptor for CMV infection. Verification will likely require CRISPR/Cas9 based gene KO strategy of EGFR for human or guinea pig cells in future HCMV or GPCMV studies.

Discussion

The guinea pig is the only small animal model for congenital CMV (cCMV). Therefore, identification and characterization of receptors for GPCMV dissemination and congenital infection would provide valuable insight into disease pathogenesis as well as help shape vaccine strategies in this preclinical animal model. PDGFRA plays an important role in PC-independent HCMV infection of fibroblasts [38–40], and this is also true for GPCMV [28]. Since PDGFRA is highly conserved across multiple species [68], it is likely an important cognate receptor in various animal CMV aside from GPCMV. Although gH/gL homologs exist in alpha and gammaherpesviruses, gO homologs are unique to betaherpesviruses, which enables specific targeting of PDGFRA as a receptor for one pathway of cell entry. N-linked glycosylation of cell receptor is important in cell entry or infectivity of several viruses, including hepatitis C virus, influenza, and members of the herpesvirus family [53–56]. A study in HCMV [59] indicated that this was similarly true for HCMV infection of fibroblasts (at low moi). In contrast, we found that inhibition of N-linked glycosylation in guinea pig fibroblasts, and specifically PDGFRA, had no impact on GPCMV(PC−) infection at moi 1 pfu/cell compared to HCMV (moi 0.05 pfu/cell) [59]. However, GPCMV infection at lower moi (0.1 and 0.01 pfu/cell were impacted by N-glycosylation inhibition treatment of cells prior to infection. Our results also suggest that glycosylation state of additional receptors/co-receptors potentially impairs cellular susceptibility at low moi but not at higher moi and this is likely true for HCMV but infection at higher moi under inhibitory conditions awaits evaluation for HCMV.

Using guinea pig cells lacking PDGFRA (REPI, TEPI, GPKO) [28, 30] as recipients for ectopically expressed full-length or C-terminal modified guinea pig PDGFRA, this study demonstrated that guinea pig PDGFRA is capable of acting as a universal host cell receptor for GPCMV infection. HCMV exhibits broad cellular tropism, capable of infecting most cell types and organs, but the specific cell receptors involved vary between cell type and pathway of cell entry. Several receptors/co-receptors have been identified for PC-dependent cell entry of HCMV, including neuropilin-2 (NRP2), OR14I1, CD147, CD46, EGFR and beta-integrins [44–47, 69, 70], but the process of PC-dependent cell entry is only partially understood and receptors are likely cell type dependent. The PC of HCMV plays a significant role in cell tropism, virus dissemination and congenital infection [13], which has also been shown for the homolog PC in GPCMV [30, 31]. Identification of specific host cell receptor requirements, such as NRP2 or others, for PC-dependent entry via endocytosis is an important direction of study as it has implications for vaccine design, consequently, identification of receptors for GPCMV is an absolute necessity. However, it was unexpected that cellular PDGFRA expression easily overcomes the requirement for PC for infection of various cell types, and likely suggests that (PC−) virus mutants could emerge in vivo as a possible immune evasion strategy, similar to fibroblast adaptation of clinical strains observed in culture [71, 72].

While HCMV entry into epithelial cells depends on the PC via pH-dependent endocytosis, gH/gL/gO mediates the direct fusion pathway in fibroblasts [73]. In agreement with studies in HCMV [38, 39, 64], we found that gO was essential for the direct interaction between GPCMV gH/gL/gO complex and guinea pig PDGFRA. A recent study in HCMV demonstrated that mutation of gO aa 117–121 interfered with binding of PDGFRA, suggesting that the N-terminus of gO likely mediates interaction with cellular PDGFRA [64]. Since the gO N-terminus is more strongly conserved between HCMV and GPCMV [64], it is likely the implication could be extended to GPCMV. It should be noted that in the current IP studies, gO and PDGFRA alone were not evaluated for interaction in the absence of gH/gL and this is an aspect that has not been evaluated for HCMV gO and is a limitation of the current study. However, gO cannot exist on the outside of the virion without being complexed with gH/gL and so this aspect is of minor importance in the context of vaccine studies as neutralizing antibodies are directed to gH/gL complex. We also demonstrated that expression of gpPDGFRA and GPCMV gH/gL/gO caused guinea pig fibroblasts to fuse, but not gH/gL in absence of gO. It was recently shown that HCMV gH and gB mutants lacking the transmembrane domain did not function in fusion [74], which is also the case with other herpesviruses [75, 76]. Thus, in addition to the requirement of gO, the GPCMV complex likely requires the transmembrane domain of gH for proper trafficking and anchoring in membranes to enable fusion, but this remains to be further evaluated for GPCMV.

GPCMV gB is essential for virus infection of all cell types [26, 30, 32], similar to HCMV [77]. Therefore, identification of potential GPCMV gB target receptor(s) would be important. However, it is not clear if GPCMV gB acts as a viral fusogen without binding cellular receptors or interacts with gH/gL complexes or a host receptor to enable virus entry. It is important to note that GPCMV gB is capable of forming a homotrimer complex [57] similar to HCMV, and this requires the C-terminal transmembrane domain. Therefore, interaction studies would have to utilize a full-length gB similar to that in our present study. A gH/gL/gB complex has recently been identified in HCMV infected cells and virions [78] but a function remains undefined. Potentially, this complex exists in GPCMV, as transient expression of gB with the gH trimer components allowed immunoprecipitation of gB (Fig 3). However, expression of gB alone or with gH/gL did not result in cell fusion (Fig S3). It is possible the pre-fusion conformation of gB [79] is stabilized by gH/gL until receptor binding events, such as gH/gL/gO interactions with PDGFRA, trigger membrane fusion events via post-fusion gB. This would be consistent with findings in HCMV that showed interaction of gB with PDGFRA was only detected when gH/gL/gO is formed [39]. The mechanism by which GPCMV glycoprotein/PDGFRA interaction mediates fusion of infected cells awaits further evaluation, however, this is beyond the scope of this current study.

We demonstrated that transient ectopic expression of full-length guinea pig PDGFRA or C-terminal truncated protein (tyrosine kinase domain deletion mutant) enabled GPCMV(PC−) infection of various guinea pig cells, including amniotic sac epithelial [32], renal epithelial, trophoblasts and GPKO cells. Although deletion of the C-terminal TK domain of PDGFRA in ectopic expression studies did not reduce GPCMV infectivity, deletion of the transmembrane domain prevented entry (Fig 6). In HCMV, transient expression of human PDGFRA enhanced (PC−) HCMV infection, and the extracellular domain of PDGFRA was demonstrated to be necessary for viral entry [40, 80]. It is likely the transmembrane domain is also required for the cell-cell fusion observed in this study, probably because PDGFRA must be anchored in the plasma membrane in order to promote fusion. Although transient ectopic expression of guinea pig PDGFRA enabled GPCMV(PC−) infection in previously resistant cell types (Fig 5) [32], ectopic expression of human PDGFRA in guinea pig epithelial cells, trophoblasts, and GPKO fibroblasts failed to complement GPCMV(PC−) infection (Fig 7). These findings suggest that the entry receptor between species, in addition to gO variability, may contribute to the strict species specificity of HCMV and GPCMV.

In the context of clinical infection of HCMV, it is possible that production of particles with different relative amounts of PC may be due to natural variation in the assembly process. Several HCMV genes have been reported to affect the composition of gH/gL complexes, and thus influence viral cell tropism. For example, UL148 promotes high-level expression of the trimer during infection and may to prevent gO degradation in the ER. In UL148-null infections, gO expression is significantly reduced, trimer expression is low, and infection of epithelial cells is enhanced [81, 82]. Perhaps there is an equivalent gene product in GPCMV, but that remains to be identified. Alternatively, it has been reported that both gO and UL128 interact with gH/gL through the same binding site on gL in a competitive manner to form the gH/gL/gO trimer, or a PC subcomplex gH/gL/UL128 [16], and we have observed this subcomplex formation in GPCMV [30]. The other PC components subsequently interact to form the full pentamer in both HCMV and GPCMV. Thus, propagation of virus in fibroblasts susceptible to (PC−) virus could select for gH/gL/gO dominant virions. This model is consistent with our analysis of viral tropism for GPCMV propagated on fibroblasts or epithelial cells [57].

Observed restriction of virus growth on fibroblasts lacking PDGFRA was previously shown to be specific for GPCMV(PC−), since infection with herpes simplex virus 1 (HSV-1), which does not require PDGFRA for cell entry, resulted in similar viral replication on both GPL and GPKO cells [28]. Susceptibility of GPKO cells to (PC+) infection but refraction to (PC−) virus indicated that two entry pathways of GPCMV exist in guinea pig fibroblasts, similar to HCMV [83]. This enables the use of GPKO fibroblasts to evaluate neutralizing antibody requirements compared to fibroblast and epithelial cells in future vaccine studies. We recently demonstrated the importance of evaluating neutralization capabilities on non-fibroblasts or fibroblasts lacking PDGFRA [32, 57]. Despite the essential nature of gB, anti-gB antibodies were less effective in neutralizing virus infection on fibroblasts lacking PDGFRA compared to PDGFRA+ cells [57], which strongly demonstrates a role for gB in cell entry.

PDGFRA-dependent susceptibility to infection could be critical in the impact of HCMV on specific tissue/ organ infection. Specific tissues express high levels of the receptor, such as the lung, brain, retina, urinary tract, gastrointestinal tract, and ovary [84]. In the context of cCMV and fetal infection, expression of PDGFRA is essential in both embryonic and mature nervous systems, where it is required for neural crest cell development, oligodendrogenesis, and adult neuronal maintenance [85]. The role of PDGFRA in neurological development brings into question the impact of PDGFRA-dependent infection on the cognitive deficits that result from cCMV. Another important step in CMV infection of the developing fetus in utero is transplacental transmission. Both of our established placental guinea pig cell lines (trophoblasts and amniotic sac membrane) lack PDGFRA and require the PC for infection [30, 32]. However, a recent HCMV study found differential expression of PDGFRA in different types of trophoblasts and demonstrated that PDGFRA is important for PC-independent HCMV entry into placental cells that express the receptor [80]. Thus, any cell type that expresses PDGFRA may be more prone to infection, and subsequently lead to more severe disease outcomes in the respective tissues/organs.

Whether EGFR is a direct receptor for HCMV or GPCMV entry remains controversial. Although EGFR was reported as a receptor for HCMV through its interaction with gB [33], it has been refuted by others [35, 37]. Recent studies have shown that integrins bind with the disintegrin-like domain of gB, which promoted CMV entry [69, 86]. It is possible that the contrasting results are due to differential expression of the receptors on different cell types. Surprisingly, in guinea pig cell lines, EGFR was more highly expressed in non-fibroblasts (Fig 8).

Activation of EGFR and other members of the ErbB family upon HCMV infection was reported by Kabanova et al., however, they did not detect direct binding between the PC and any of these receptors [38]. Wu and colleagues demonstrated that EGFR knockdown showed little to no inhibition of HCMV infection [40], suggesting that this receptor is not required for (PC+) virions to enter cells that lack PDGFRA expression. Pharmacological treatment of guinea pig cells with EGFR tyrosine kinase (TK) inhibitors did not modify GPCMV infection of fibroblast or epithelial cells (Fig 8). We concluded that the EGFR transduction pathway is not required for infection in cells that support lytic replication. Indeed, in our studies, TK transduction, in general, was not required for GPCMV infection, which was demonstrated by broad spectrum TK inhibition. Consistent with this finding, it has been shown that the TK activity of PDGFRA is not required for its role in HCMV cell entry [39, 40]. Furthermore, our results suggest that the inhibitors used did not enhance other antiviral cellular targets such as innate immunity. However, it is possible that PDGFRA signaling may influence other intracellular functions that could be important for infection by a variety of viruses, including CMV.

In summary, we have established a method for evaluating host cell receptor requirements for GPCMV infection and specific glycoproteins involved in interaction with PDGFRA cell receptor. Further, we demonstrate that PDGFRA expression not only alters viral tropism, but also modulates the cell entry pathway of GPCMV. Consequently, a successful vaccine will likely need to target the gH-based trimer viral glycoproteins interacting with this receptor, in addition to targeting the PC or gB antigens, to prevent viral dissemination and reduce the risk of cCMV.

Materials and Methods

Cells, viruses, and oligonucleotides

GPCMV (strain 22122, ATCC VR682) was propagated on guinea pig fibroblast lung cells (GPL; ATCC CCL 158) cultured in Dulbecco’s modified Eagle’s medium (DMEM; HyClone) supplemented with 10% fetal bovine serum (FBS; Gibco-BRL), 10,000 IU/liter penicillin, and 10 mg/liter streptomycin (Gibco-BRL) at 37°C/5% CO₂. Human foreskin fibroblast (HFF) cells (ATCC SCRC-1041) were maintained under identical conditions as GPL fibroblast cells. PC negative (PC−) GPCMV was generated by continued passage of virus on fibroblast cells to generate virus deleted in GP129-GP133 locus [30, 58]. GPCMV(GFP+/PC-), designated vAM403, was previously generated by homologous recombination with PC- virus and carries a GFP insertion cassette in a non-essential Hind “N” locus of GPCMV [58]. Additionally, (PC+) virus was propagated on guinea pig renal epithelial (REPI) and trophoblast cells (TEPI) as previously described [30, 31]. HEK293 cells (ATCC CRL-1573) were cultured in DMEM (HyClone) supplemented with 10% FBS (Gibco-BRL), 10,000 IU/liter penicillin, and 10 mg/liter streptomycin (Gibco-BRL) at 37°C/5% CO₂. Guinea pig PDGFRA knock out (GPKO) fibroblast cell line [28] was derived from GPL fibroblasts. Exon 2 of guinea pig PDGFRA was targeted for CRISPR/Cas9 mutagenesis and knockout cell line was verified for loss of PDGFRA by Western blot of cell lysate and sequencing of modified locus [28]. Epithelial cells and GPKO cells were maintained as previously described [28, 30]. pmCherry-C1 mCherry-NLS was constructed through Addgene [62]; eGFP-C1 (Clontech) was used for Cas9 cell lines. GPL(mCherryNLS) cell lines (unpublished 2019, McGregor) were maintained as described for fibroblasts with additional G418 selection at a concentration of 200 μg/mL. Virus growth assays were carried out on six-well plates (Falcon) and titrations were carried out on 12-well plates (Falcon). Plaques were visualized by fluorescence or bright-field microscopy. High titer stock viruses were generated as previously described [26–28, 31]. All oligonucleotides were synthesized by Sigma-Genosys (The Woodlands, TX).

Gene cloning and generation of expression plasmids

A FLAG epitope tagged synthetic cDNA of the predicted guinea pig PDGFRA ORF (XM_003471663.3) was generated (Genscript), cloned into mammalian expression vector pcDNA3.1, and used in anti-PDGFRA antibody verification studies as previously described [28]. The plasmid for full-length gpPDGFRA was designated pgpPDGFRA (codons 1–1089). A truncated gpPDGFRA lacking the tyrosine kinase domain was constructed by SmaI collapse (plasmid pgpPTKD) with the extracellular domain (ectodomain) and the transmembrane domain intact (codons 1–801). A C-terminal FLAG epitope tagged gpPDGFRA ectodomain transient expression plasmid was generated through subcloning as a BamHI fragment using primers FPDGFRABm (5’ATATCAGGATCCAAGCTGGCTAGCGCCGCCACCATG3’) and RPBm514 (5’ATATCAGGATCCCAGATTCTTGGCCAGACACCGCACGGCGATTGTC3’) then cloned in-frame into mammalian expression vector pCMV-Tag 8. This modified plasmid lacking the tyrosine kinase and transmembrane domains was designated pgpPED (codons 1–514). A His-tagged PDGFRA expression plasmid based on the GenBank Ref. ID sequence of the predicted human PDGFRA ORF (NM_006206.4) was generated by Sino Biological Inc., cloned into mammalian expression vector pCMV3-C-His (pCMV3-PDGFRA-His) for expression of human PDGFRA (huPDGFRA). A control construct expressing mCherry was generated by Addgene using the pmCherry-C1 mammalian expression vector backbone (Clontech). Plasmids and PCR products were further verified by sequencing as necessary. Expression vectors for gH, gL, and gO previously described [26]. The gpPDGFRA ORF was additionally subcloned into pLVXIRESmCherry vector (Clontech) to enable gpPDGFRA expression and co-expression of an mCherry reporter gene as bicistronic mRNA in vector pLVXPDGFRA-IRESmCherry [32].

Immunoblot and Immunoprecipitation assays

Western blots were carried out on cell lysates separated by 10% SDS-PAGE and performed as previously described [26, 30]. Immunoprecipitation (IP) assays were carried out using either GFP-trap (ChromoTek), or via protein A Sepharose beads (Life technology), following manufacturer’s protocols with inclusion of protease inhibitor cocktail (Pierce) in cell lysates, as previously described [30]. For Western blots, anti-epitope tag primary antibodies were used at 1/1000: mouse anti-FLAG M2 (Sigma); rabbit anti-GFP (Santa Cruz); mouse anti-Myc-c (Novus Biologicals); goat anti-human PDGFRA (R&D Systems) [28, 30]. gB protein was detected with primary mouse monoclonal antibody to gB [57] (a gift from Dr. Britt, UAB) at 1/1000. Rabbit anti-EGFR (Aviva) and mouse anti-beta-actin (Cell Signaling) were used at 1/500. Secondary antibodies for Western blot analysis: anti-mouse, anti-rabbit, or anti-goat IgG/HRP conjugates (Sigma) were used at 1/2000 [28, 57, 87].

IHC staining of GPCMV infected fibroblasts

GPKO and GPL fibroblasts were infected with GPCMV(PC+) or GPCMV(PC−) at moi 1 pfu/cell. Immunostaining to detect GPCMV IE2 [52] was performed 24 hpi using Vectastain Elite ABC kit (Vector Laboratories) as previously described [30]. GPCMV IE2 was detected with primary rabbit anti-IE2 at 1/500 and a secondary anti rabbit IgG-HRP conjugate at 1/1000.

Cell-cell fusion assays

To express the viral glycoprotein gH/gL/gO complex, GFP-positive GPKO cells were transduced with AdgHGFP [26] and AdgLmyc [26] following transfection with expression plasmid pGP74TAG8 [26], and allowed to incubate for 24 h at 37°C, 5% CO₂. Briefly, 0.5×10⁵ GPL(mcherryNLS) cells were dispensed into each well of coverslip six-well culture plates (Falcon) and overlaid with 0.5×10⁵ GPKO cells per well. Similarly, co-culture was performed with GPL(mcherryNLS) and GPKO cells transduced to express: gB only; gHGFP only; gHGFP and gLmyc; or gB, gHGFP, and gLmyc. Control co-cultures were performed without transduction or transfection (untreated). All co-cultured cells were further incubated an additional 24 h. Fusion between indicator GPL(mcherryNLS) cells and GPKO(GFP+) cells was observed as syncytial formation. Monolayers were visualized by real-time fluorescent microscopy (GFP and mCherry filters). Real-time images at 10X and 20X magnification were obtained.

N-glycosylation inhibitor treatment and GPCMV infection

Stock tunicamycin (Sigma Aldrich) solution was prepared at 5 mg/ml in DMSO. To verify deglycosylation, GPL cells were pretreated with (1 μg/mL) or without Tunicamycin (TM) for 2 h prior to transfection with gpPDGFRA. After transfection, cells were cultured an additional 24 h in fresh media, in the presence (1 μg/mL) or absence of TM. Cell lysates (n=3/group) were then harvested for Western blot analysis to detect gpPDGFRA expression. For infections, duplicate plates of GPL cells were pretreated on day 0 with (1.25 μg/mL, 0.63 μg/mL, 0.31 μg/mL, 0.16 μg/mL) or without (0 μg/mL) tunicamycin (TM) for 24 h at 37°C with 5% CO₂. TM was then removed for 1 h while cells were either infected with GPCMV(PC-/GFP+) at multiplicity of infection (moi) 1, 0.1 or 0.01 pfu/cell or mock infected. Inoculum was removed and cells were incubated in fresh media with or without TM at the same concentration as pretreatment. Fluorescence microscopy was used to monitor viral GFP expression. Cell morphology and CPE were visualized by bright-field microscopy.

EGFR tyrosine kinase inhibitor treatment and GPCMV infection

Erlotinib HCl (OSI-744; SelleckChem) solution was prepared at 20 mg/ml (46.52 mM) in DMSO. On day 0, GPL (or REPI) cells were pretreated with 0 nM (untreated control), 10 nM, 50 nM, 500 nM, 1 μM, or 10μM erlotinib (OSI-744; SelleckChem) for 1.5 h at 37°C with 5% CO₂. Cells were then infected separately with either GPCMV(PC+) or GPCMV(PC−) at moi 0.5 pfu/cell for 1 h. Inoculum was washed, and monolayers were overlaid with fresh media containing the same concentration of inhibitor as pretreatment, respectively. After 72 h incubation at 37°C/5% CO₂, supernatants and monolayers were harvested and titrated in duplicate on GPL cells as previously described [30]. Experiments were carried out in triplicate. Cell viability was assayed by CellTiter-Glo® Luminescent Cell Viability Assay (Promega) following manufacturer’s protocol. Viability assays were carried out on cells in the presence and absence of drugs (at concentrations used in viral assays) as well as control DMSO only in media. During cell viability assays, cells were not infected with virus but incubated in presence of drugs for similar duration as virus studies. Experiments were carried out in triplicate for each assay condition.

Pan-tyrosine kinase inhibitor treatment and GPCMV infection

Ponatinib (AP24534; SelleckChem) stock solution was prepared at 100 mg/ml (187.7 mM) in DMSO. On day 0, GPL (or REPI) cells were pretreated with 0nM (untreated control), 1 nM, 10 nM, 50 nM, 100 nM, or 500 nM ponatinib for 1.5 h at 37°C with 5% CO₂. Cells were infected separately with either GPCMV (PC+) or (PC−) virus at moi 0.5 pfu/cell for 1 h. Inoculum was washed, and monolayers were overlaid with fresh media with the same concentration of inhibitor as pretreatment, respectively. After 72 h incubation at 37°C/5% CO₂, supernatants and monolayers were harvested and titrated in duplicate on GPL cells as previously described [30]. Experiments were carried out in triplicate. Cell viability was assayed by CellTiter-Glo® Luminescent Cell Viability Assay (Promega) following manufacturer’s protocol. Viability assays were carried out on cells in the presence and absence of drugs as described in the above section. Experiments were carried out in triplicate for each assay condition.

Flow cytometry and detection of cell surface expression of huPDGFRA

Flow cytometry antibody detection of human PDGFRA cell surface expression was carried out with mouse anti-human CD140a (PDGFRA)-conjugated with APC (Biolegend) following manufacturer’s protocol. Surface protein detection for endogenous expression of huPDGRFA on human foreskin fibroblast (HFF) cells and plasmid transiently expressed protein on guinea pig REPI cells were tested. Cells were grown on 6 well plates and samples were evaluated in duplicate. For endogenous expression, 1×10⁶ HFF cells were labeled with 5 μl of huPDGRFA-APC antibody in 100 μl of flow buffer (1X PBS + 2% FBS) for 30 mins on ice in the dark. For transient expression, huPDGFRA expression plasmid was transfected onto guinea pig REPI cells and at 36 hr post transfection harvested and 1×10⁶ cells were labeled with huPDGFRA antibody. Unlabeled HFF and REPI cells were included as unstained antibody controls. Flow cytometry was carried out on a FACAria II (BD Life Sciences) and data analyzed by FlowJo software (V10).

Supplementary Material

1

Fig S1. GPKO fibroblasts are resistant to GPCMV(PC-) virus, but GPCMV(PC+) infects cells independently of PDGFRA expression. GPKO and GPL cells were infected with (PC-) or (PC+) virus (moi 1 pfu/cell) in separate wells of six well plates in comparison to mock infected controls. Real-time, bright-field images were taken 4 dpi. (i) Presence or absence of CPE after GPCMV(PC-) virus infection. A-B, GPL; C-D, GPKO. A, C MI; B, D GPCMV(PC-) infected cells. (ii) GPCMV(PC+) infection demonstrating CPE in comparison to MI cells. A-B, GPL; C-D, GPKO. A, C MI; B, D GPCMV(PC+) infected fibroblasts. Magnification 10X. Individual images representative of multiple fields.

Fig S2. Control GFP and mCherry detection in guinea pig fibroblast cell lines.

Bright-field and fluorescent channel real-time images of GPKO(GFP+) fibroblasts and GPL(mCherry+/PDGFRA+) fibroblasts demonstrate no cross bleed of fluorescence between channels. Individual images representative of multiple fields. Magnification 10X.

Fig S3. Individual viral glycoproteins (gH, gB) and gH/gL, gB/gH/gL complexes are not sufficient for GPKO(GFP+) cell fusion with GPL(mCherry+/PDGFRA+) cells.

GPKO(GFP+/PDGFRA-) fibroblasts were transduced to express: A-C, gB; D-F, gH; G-I, gH/gL; and J-L, gB + gH/gL. Each set of monolayers were then co-cultured with GPL(mCherry+/PDGFRA+) fibroblasts for 24 h, and real-time images were obtained at 10X magnification. A, D, G, J show GFP expression; B, E, H, K show mCherry expression; C, F, I, L are the merged images for each group. Individual images are representative of multiple fields.

Fig S4. Co-expression and IP of GPCMV gH/gL/gO trimer and gpPED.

GFP-Trap IP of GPKO fibroblast cells transduced with recombinant adenovirus encoding gHGFP, gLmyc, and plasmid transfected to express gO (pGP74TAG8) and gpPED (pgpPED). Lanes 1, 5 and 9 are total cell lysate; 2, 6, and 10, flow through; 3, 7 and 11, IP reactions; 4, 8, and 12, mock cell lysate. Analysis was performed by Western blot using respective primary antibodies: gH (anti-GFP); gL (anti-myc); gpPED (anti-FLAG). Bands visualized with appropriate secondary antibody-HRP conjugate as described in methods section.

Fig S5. Ectopic expression of either gpPTKD or full-length gpPDGFRA renders REPI cells susceptible to infection by GPCMV(PC-) virus. REPI cells were evaluated for infection with GPCMV(PC-/GFP+) virus in presence of gpPDGFRA (A-C) comparison to GPL cells (D-F). Monolayers of each cell type were transfected to express: A, D empty plasmid vector only; B, E PDGFRA mutant gpPTKD; C, F full-length gpPDGFRA 24 h prior to infection (moi 1 pfu/cell). Real-time fluorescent microscopy images acquired at 3 dpi. Individual images are representative of multiple experiments (6 per panel). Magnification 10X.

FigS6. Cell surface expression of huPDGFRA by FACS analysis.

FACS analysis was performed on human foreskin fibroblast (HFF) cells labeled with anti-huPDGFRA-APC antibody (Biolegend) for endogenous expression of PDGFRA (A-C). Unstained control HFF cells were gated in forward vs side scatter plot (A) then represented in a histogram (B). Histogram of PDGFRA+ cells (C). Guinea pig REPI cells transfected with huPDGFRA expression plasmid at 36 h post transfection were labeled with huPDGFRA-APC antibody for surface expression (D-F). Unstained control REPI cells were gated in forward vs side scatter (D) then represented in a histogram (E). REPI cells transfected with human PDGFRA expression plasmid were labeled with anti-huPDGFRA antibody for detection (F).