Inhibition of Human Cytomegalovirus Entry into Mucosal Epithelial Cells

Abstract

Human cytomegalovirus (CMV) causes serious developmental disabilities in newborns infected in utero following oral acquisition by the mother. Thus, neutralizing antibodies in maternal saliva have potential to prevent maternal infection and, consequently, fetal transmission and disease. Based on standard cell culture models, CMV entry mediators (and hence neutralizing targets) are cell type-dependent: entry into fibroblasts requires glycoprotein B (gB) and a trimeric complex (TC) of glycoproteins H, L, and O, whereas endothelial and epithelial cell entry additionally requires a pentameric complex (PC) of glycoproteins H and L with UL128, UL130, and UL131A. However, as the mediators of mucosal cell entry and the potential impact of cellular differentiation remained unclear, the present studies utilized mutant viruses, neutralizing antibodies, and soluble TC-receptor to determine the entry mediators required for infection of mucocutaneus cell lines and primary tonsil epithelial cells. Entry into undifferentiated cells was largely PC-dependent, but PC-independent entry could be induced by differentiation. TC-independent entry was also observed and varied by cell line and differentiation. Infection of primary tonsil cells from some donors was entirely TC-independent. In contrast, an antibody to gB or disruption of virion attachment using heparin blocked entry into all cells. These findings indicate that CMV entry into the spectrum of cell types encountered in vivo is likely to be more complex than has been suggested by standard cell culture models and may be influenced by the relative abundance of virion envelope glycoprotein complexes as well as by cell type, tissue of origin, and state of differentiation.

1. Introduction

Human cytomegalovirus (CMV) is a ubiquitous betaherpesvirus that is prevalent in the human population worldwide. CMV infections are associated with significant morbidity and mortality in immunocompromised patients, including transplant recipients and AIDS patients (Munawwar and Singh, 2016; Soderberg-Naucler, 2006). Congenital infections also occur when CMV is transmitted vertically from mother to fetus during pregnancy. In the US, congenital CMV infections occur in about 0.45% of all live births (Fowler and Boppana, 2018). Approximately 15-20% of congenitally infected newborns are symptomatic, with often severe and sometimes life-threatening sequelae. The remainder appear asymptomatic at birth but 17-20% will develop significant sequelae later in life. Sensorineural hearing loss is a common disability associated with congenital CMV infections, impacting 50% of symptomatic and 10-15% of asymptomatic newborns (Bartlett et al., 2017; Dollard et al., 2007).

Preventive or therapeutic options prior to or during pregnancy are limited. There are no vaccines to prevent maternal infections, and toxicity and potential teratogenicity preclude the use of current CMV antivirals in pregnant women. However, acyclovir, which has weak activity against CMV but a good safety profile during pregnancy, was recently shown to reduce the rate of vertical transmission when used at a high dose (Amir et al., 2023; Chatzakis et al., 2023; D'Antonio et al., 2023; De Santis et al., 2020; Egloff et al., 2023; Faure-Bardon et al., 2021; Leruez-Ville et al., 2016; Shahar-Nissan et al., 2020; Zammarchi et al., 2023). Similar safety considerations make antibodies attractive. CMV hyperimmune globulin (HIG, polyclonal IgG isolated from the blood of CMV-seropositive donors) is beneficial in solid organ transplant patients (Snydman, 1993; Snydman et al., 1987) but its efficacy in the context of congenital CMV infections is less clear. While some studies have failed to show a benefit (Hughes et al., 2023; Revello et al., 2014), rapid administration of HIG after diagnosis of maternal infection, at higher dose, and biweekly (vs monthly) was shown to significantly reduce fetal transmission rates compared to historical controls (Kagan et al., 2021; Kagan et al., 2019). Therapeutic monoclonal antibodies (mAbs) that neutralize viral infectivity by targeting specific CMV glycoprotein complexes have also been generated. While efficacy has been observed in renal transplant settings (Ishida et al., 2017), mAbs have not as yet been evaluated in the context of congenital CMV. Thus, there is considerable need for either active immunization or passive delivery of prophylactic or therapeutic antibodies to prevent maternal infections, fetal transmission, or fetal disease following congenital infections.

The established CMV entry mechanisms are complex and utilize different glycoprotein complexes in the virion envelope for entry into different cell types (Fig. 1A). The process is thought to begin with attachment of the virion to cell surface heparan sulfate proteoglycans (HSPGs) via interactions with the virion envelope-associated heterodimer of glycoproteins M and N (gM/gN) (Compton, 2004; Compton et al., 1993; Kari and Gehrz, 1992). Subsequent steps depend on the target cell type. If the target cell is a fibroblast, binding of a trimeric complex (TC) comprised of glycoproteins H, L, and O in the virion envelope to its cellular receptor, the platelet-derived growth factor-alpha receptor (PDGFRα), triggers membrane fusion mediated by a homotrimeric complex of glycoprotein B (gB), an immunodominant class III fusion glycoprotein that is among the most conserved across the Herpesvirdae (Chandramouli et al., 2017; Heldwein, 2016; Kabanova et al., 2016; Si et al., 2018; Wu et al., 2018). These events occur at the cell surface and do not require endocytosis or endosomal acidification. This contrasts with entry into epithelial or endothelial cells, in which the virion is first endocytosed and endosomal acidification is then needed to initiate gB-mediated membrane fusion (Ryckman et al., 2006). The TC is also required for the endocytic entry pathway (Wille et al., 2010), although the apparent absence of PDGFRα from epithelial and endothelial cells suggests it may occur via a different receptor (Vanarsdall et al., 2012). In addition to the TC, a pentameric complex (PC) is needed for endocytic entry. The PC also contains gH/gL complexed with the UL128, UL130, and UL131A proteins instead of with gO (Heldwein, 2016; Martinez-Martin et al., 2018; Ryckman et al., 2006; Wang and Shenk, 2005b; Wille et al., 2010; Wille et al., 2013; Zhou et al., 2015). While several PC receptors have been reported, including neuropilin 2 and the olfactory receptor OR14I1 (E et al., 2019; Martinez-Martin et al., 2018), the mechanistic role of this complex in the endocytic entry pathway remains unclear.

Figure 1. CMV virion envelope glycoprotein complexes required for entry into different cell types.

The CMV virion surface is illustrated with glycoprotein complexes gM/gN, gB, TC, and PC embedded in the virion envelope. The importance of each complex for efficient entry into each cell type are indicated as required (✔), dispensable (✘), or partially required (✔✘), based on genetic mutations (PC) or sensitivity to neutralizing antibodies (gB, PC), PDGFRα-Fc (TC), or heparin (gM/gN). (A) Summary of findings from the literature; (B) Results from the present studies.

These intricacies of cell type-dependent entry pathways complicate the functionality of neutralizing antibodies to CMV. While antibodies or inhibitors of gM/gN, gB, gH/gL, or gO can block CMV entry into a broad range of cells, antibodies to epitopes unique to the PC (e.g., dependent on UL128, UL130, or UL131A) neutralize entry into epithelial and endothelial cells but have no activity against fibroblast entry (Gerna et al., 2005; Hahn et al., 2004; Ishibashi et al., 2011; Ryckman et al., 2008; Song et al., 2001; Stegmann et al., 2017; Wang and Shenk, 2005b; Zhou et al., 2015). While such PC-specific mAbs are often far more potent at neutralizing epithelial or endothelial cell entry than mAbs targeting gB or gH/gL (Cui et al., 2017b; Gerna et al., 2016; Macagno et al., 2010), optimal therapeutic strategies may require complementation from antibodies that neutralize entry into a broader range of cell types (albeit less potently). For example, a Moderna mRNA-1647 vaccine currently undergoing phase 3 evaluation encodes both full-length gB and the five PC subunits (Fierro et al., 2024; John et al., 2018; Wu et al., 2024), and in a recent phase 2 trial, post-transplant prophylaxis with a bivalent mAb mixture (one targeting a gH epitope common to the TC and PC and one specific to the PC) significantly extended the time to appearance of the first CMV viremia and reduced the incidence of CMV disease in renal transplant patients (Ishida et al., 2017).

Prevention of maternal CMV infection should indirectly avert congenital infection. Maternal infections primarily occur through oral contact with bodily fluids such as urine or saliva containing infectious CMV (Knipe and Howley, 2013). Children in daycare have a high prevalence of CMV shedding lasting for a year or more (Lidehall et al., 2013), and pregnant women are at high risk if they are exposed to secretions from asymptomatic children through kissing, diaper changing, or toilet training (Cannon et al., 2011). Consequently, the oronasal cavity is considered the primary route of maternal CMV acquisition, and passive or active vaccination strategies that deliver neutralizing antibodies to the oral cavity may be effective in preventing CMV from infecting the mother and subsequently reaching the developing fetus. However, many congenital CMV infections occur in mothers who had CMV infections prior to conception. The source of virus in such non-primary infections, whether from maternal reinfection or reactivation of maternal latent virus, remains uncertain (Britt, 2020). While fetal infections from reactivation presumably bypass the oral cavity, reinfections are believed to occur upon virus acquisition through the oral cavity in a manner similar to primary infections.

The mechanistic details of endocytic epithelial/endothelial cell entry and non-endocytic fibroblast entry described above were ascertained primarily from in vitro studies using foreskin or lung fibroblasts to model entry into mesenchymal lineage cells, aortic or umbilical vein endothelial cells to model endothelial cell entry, and retinal pigment epithelium cells to model epithelial cell entry (Fig. 1A) (Kuppermann et al., 1993; Offit and Moser, 2011). In vivo, however, CMV exhibits a broad cellular tropism (Detrick et al., 1996; Knowles, 1976), and although PC-dependent entry has been demonstrated for epithelial cells derived from a number of tissues (Detrick et al., 1996; Shimamura et al., 2010; Twite et al., 2014; Wang and Shenk, 2005a), the organ of origin can also dictate specific entry requirements. For example, CMV entry into trophoblast progenitor cells does not conform to the established paradigm of epithelial cell entry described above as infection is neutralized by gB- but not by PC-specific antibodies (Zydek et al., 2014). As CMV acquisition is believed to involve epithelial cells of the oronasal mucosa, and as CMV entry into these cells has not been characterized, the viral targets of neutralizing antibodies capable of disrupting oral acquisition are not known. Consequently, understanding the specific glycoprotein complexes used by CMV to enter oronasal epithelial cells may be critical to develop vaccines or therapeutic antibodies.

To investigate the potential for neutralizing antibodies or other entry inhibitors in preventing oral CMV acquisition, we used neutralizing mAbs, soluble PDGFRα, heparin, and viruses with PC-disrupting mutations to characterize the viral entry mediators required for CMV entry into mucocutaneous epithelial cell lines under both undifferentiated and differentiated conditions, as well as into primary tonsil epithelial cells (PTEC) isolated from the palatine tonsil of different donors (Table 1). Entry into PTECs and into undifferentiated mucocutaneous epithelial cell lines was predominantly PC-dependent, but changes induced by differentiation greatly enhanced entry by a PC-independent mechanism. Importantly, TC-independent entry into mucocutaneous epithelial cell lines and into PTECs was prevalent and influenced by virus strain, cell line, and PTEC donor. Complete inhibition of entry could only be attained by simultaneously blocking both the PC and the TC, suggesting that in the context of mucocutaneous epithelial cell entry, these complexes serve redundant functions. In contrast, targeting virion attachment using heparin or targeting gB using a gB-specific neutralizing mAb fully inhibited entry into all cell types under all conditions tested.

Table 1.

Properties of cells used in these studies

type	name	tissue origin	modificationsa	references
fibroblast	HFF	foreskin	none	(Prichard et al., 1996)
	MRC-5	fetal lung	none	(Jacobs et al., 1970)
endothelial	HAEC	aorta	none	Lonza, CC-2535
epithelial	ARPE-19	retinal pigment epithelium	none	(Sparrow et al., 2000)
	N/TERT	foreskin	tert	(Dickson et al., 2000)
	NOK	gingiva	tert	(Krisanaprakornkit et al., 1998; Piboonniyom et al., 2002)
	HAK	adenoid	tert	(Farwell et al., 2000)
	HTE	palatine tonsil	tert p16	(Kimple et al., 2013)
	PTEC	palatine tonsil	none	(Spanos et al., 2008)

^atert, transduced with human telomerase reverse transcriptase; p16, transduced with shRNA to p16

2. Materials and Methods

2.1. Cells

Table 1 lists the cells used, the tissues they were derived from, and genetic features. Human foreskin fibroblast (HFF) cells (Prichard et al., 1996) were a gift from E. S. Mocarski (Emory University, Atlanta, GA). Human MRC-5 lung fibroblasts (ATCC CCL-171) (Jacobs et al., 1970) and ARPE-19 retinal pigment epithelium epithelial cells (ATCC CRL-2302) (Sparrow et al., 2000) were obtained from ATCC and were used between passage 35 and 45. HFF, MRC-5, and ARPE-19 cells were cultured in Dulbecco’s modified Eagle’s Minimum Essential Medium (Gibco BRL, Gaithersburg, MD, USA) with 10% fetal bovine serum (Gemini, West Sacramento, CA, USA) and 1% Penicillin Streptomycin Glutamine (Gibco BRL, Gaithersburg, MD, USA). Human aortic endothelial cells (HAEC, Lonza CC-2535) were a gift from Dong Yu (Washington University in St. Louis, St. Louis, MO) and were cultured using EGM-2 Bullet Kit endothelial cell growth medium (Lonza, Walkersville, MD, USA). Primary human tonsil epithelial cells (PTEC), human tonsil epithelial cells (HTE), and human adenoid keratinocytes (HAK) were isolated as described previously (Farwell et al., 2000; Kimple et al., 2013; Spanos et al., 2008). Neonatal foreskin keratinocytes (N/TERT) were a gift from Peter Howley (Harvard Medical School, Boston, MA, USA) and have been described previously (Dickson et al., 2000). Normal oral keratinocytes (NOK) (Krisanaprakornkit et al., 1998; Piboonniyom et al., 2002) were a gift from Karl Munger (Tufts University, Medford, MA). HTE, HAK, N/TERT, and NOK cell lines were immortalized through retroviral transduction of the human telomerase reverse transcriptase gene, while HTE were additionally transduced to express an shRNA targeting p16 (Blanton et al., 1991; Dickson et al., 2000; Farwell et al., 2000; Hahn et al., 1999; Kimple et al., 2013; Krisanaprakornkit et al., 1998; Piboonniyom et al., 2002). N/TERT, NOK, HTE, HAK, and PTEC were cultured in Keratinocyte serum-free medium (Invitrogen, Grand Island, NY, USA) supplemented with 50 μg/mL bovine pituitary extract, 5 ng/mL human recombinant epidermal growth factor, and 1% penicillin-streptomycin (Gibco BRL, Sanborn, NY, USA) (KSFM). PTEC were used between passage four and six. For maintenance in an undifferentiated state, cells were cultured in KSFM with 0.3 mM CaCl₂ (Sigma, St. Louis, MD, USA) as described (Bristol et al., 2020), and are referred to as uN/TERT, uNOK, uHAK, or uHTE.

2.2. Differentiation

Prolonged elevation of intracellular calcium has been shown to promote keratinocyte differentiation (Deucher et al., 2002; Hohl et al., 1991; Pillai et al., 1990; Rubin et al., 1989; Su et al., 1994; Yuspa et al., 1989). We therefore used a previously reported protocol of CaCl₂ modulation (Regan and Laimins, 2013) to differentiate N/TERT, NOK, HTE, and HAK cells. Cells were seeded in 96-well plates or in 16-well glass chamber slides (Nunc^™ Lab-Tek^™) coated with 10% rat tail collagen (Advanced BioMatix) and cultured overnight in KSFM containing 0.3 mM CaCl₂. The next day (day one) the medium was changed to KSFM containing 0.03 mM CaCl₂, and on day two it was changed to KSFM containing 1.5 mM CaCl₂ but without bovine pituitary extract, recombinant epidermal growth factor, or 1% Penicillin-Streptomycin. Cells were maintained in this medium from days three through eight, with medium changes every other day. Cells that had completed differentiation, referred to as dN/TERT, dNOK, dHAK, or dHTE, were infected on day nine. When visualized using bright field microscopy, all four cell lines exhibited a notable change in appearance upon differentiation (not shown). As an increase in involucrin levels is a commonly used marker for keratinocyte differentiation (Watt, 1983), undifferentiated and differentiated cultures were evaluated for involucrin expression by western blot analyses as described previously (Prabhakar et al., 2022). Increased involucrin levels were observed in dN/TERT, dNOK, and dHAK cells relative to uN/TERT, uNOK, and uHAK cells, respectively, while levels were similar in dHTE vs uHTE and also comparable to those in dN/TERT, dNOK, dHAK cells (He et al., manuscript in preparation). Thus, at least with respect to involucrin expression, uHTE cells may constitutively be partially differentiated.

2.3. Viruses

Table 2 lists the virus strains and variants used and their genetic features. Virus BADrUL131-Y4 (BADr, a gift from Thomas Shenk and Dai Wang) is a variant of CMV strain AD169 derived from a bacterial artificial chromosome (BAC) clone modified to encode a green fluorescent protein (GFP) marker cassette and a functional PC through repair of a mutation in UL131A (Wang et al., 2004; Wang and Shenk, 2005a). Virus TS15-rN is a variant of CMV strain Towne-varS derived from a BAC clone modified to encode a GFP marker cassette and a functional PC by passage on ARPE-19 cells, resulting in spontaneous repair of a frameshift mutation in UL130 (Cui et al., 2012; Cui et al., 2013). BAC clones pAL1119 and p1128 of CMV strain Merlin (Stanton et al., 2010) were gifts from Qiyi Tang and Richard Stanton. BAC clones GT1c (a gift from Barbara Adler and Irene Görzer) and KL7 (a gift from Christian Sinzger) represent different variants of CMV strain TB40/E. GT1c is a variant of TB40-BAC4 (Sinzger et al., 2008) modified to encode firefly luciferase (Scrivano et al., 2011). KL7-FS is a variant of BAC KL7 (Sampaio et al., 2017) modified to contain a frame-shift mutation disrupting the RL13 gene (Ourahmane and McVoy, unpublished).

Table 2.

Properties of virus strains and variants used in these studies

strain	variant	marker	PCa	references
AD169	AD169varUK	none	UL131AFS	(Hahn et al., 2004)
	BADr	GFP	wt	(Wang and Shenk, 2005a)
Towne	RC2626	luciferase	UL130FS	(McVoy and Mocarski, 1999)
	TS15-NR	GFP	wt	(Cui et al., 2012; Cui et al., 2013)
Merlin	pAL1128	none	wt	(Stanton et al., 2010)
	pAL1119		UL128FS
	pAL1128-ΔPC		ΔUL131A-128ex2	this study
	pAL1119-ΔPC		ΔUL131A-128ex2
TB40/E	TB40/EF	none	UL130FS, UL128FS	(Vo et al., 2020)
	TB40/EE	none	wt
	GT1c	luciferase	UL128splice	(Kalser et al., 2017)
	KL7-FS	none	wt	(Sampaio et al., 2017)
	GT1c-ΔPC	luciferase	ΔUL131A-128ex2	this study
	KL7-FS-ΔPC	none	ΔUL131A-128ex2
Uxc	UxcAp14	none	wt	(Corcoran et al., 2017; Cui et al., 2017a; Qi et al., 2020)
	UxcA66	GFP	wt
	ABV	GFP	wt

^awt, wild type; FS, frame shift; splice, inefficient intron excision; Δ, engineered deletion

A mixed stock of CMV strain TB40/E (Sinzger et al., 1999), a gift from Christian Sinzger, was passaged twice on HFFs to produce a stock designated TB40/EF and subsequently passaged five times on ARPE-19 cells to generate a stock designated TB40/EE (Al Qaffas et al., 2020; Vo et al., 2020). CMV strain Uxc was cultured from the urine of a newborn using ARPE-19 cells and subsequently passaged 14 times exclusively on ARPE-19 cells to produce virus stock UxcAp14 (Cui et al., 2017a). This stock was further amplified through one additional passage in ARPE-19 cells to produce a stock designated UxcAp15. BAC-clones ABV and UxcAp66 were derived from the UxcAp14 stock and both encode a GFP marker cassette. ABV contains a spontaneous deletion removing UL146 to UL150 while UxcAp66 contains a spontaneous deletion removing IRS1 to US28 (Cui et al., 2017a; Qi et al., 2020).

Virus stocks were derived from infected cell culture supernatants, adjusted to 0.2 M sucrose, and stored in liquid nitrogen. Viral titers were determined by simultaneously infecting MRC-5, ARPE-19, and undifferentiated or differentiated NOK, HAK, N/TERT, and HTE as described (Cui et al., 2012).

2.4. BAC Recombineering

Galactokinase-mediated recombineering (Warming et al., 2005) was used as described previously (McVoy et al., 2016) to replace all of UL130 and UL131A and the first and second exons of UL128 with a galK cassette encoding galactokinase. The galK cassette was inserted into BACs GT1c, KL7-FS, pAL1119, or pAL1128 by PCR amplification of plasmid pgalK using the oligonucleotide pairs UL128-131-galK-FW (CTCTGTCTTACTCTCCCATAGGCTGTAAGGCCCTCGAGGAAGAGACTTACACGACTCACTATAGGGCGAATTGG) and UL128-131-galK-RV (TACTATGTGTATGATGTCTCATAATAAAGCTTTCTTTCTCAGTCTGCAACGCTATGACCATGATTACGCCAAGC), transformation of the PCR product into E. coli strain SW102 cells containing target BACs, colony selection on Gal-positive selection plates, and verification of a clone containing the expected galK insertion by PCR and targeted sequencing (eurofins), as described previously (Schleiss et al., 2015). BAC-derived viruses were reconstituted by transfection of BAC DNA into human fibroblasts using effectene transfection reagent (Qiagen, Hilden, Germany), as previously described (Brait et al., 2020; Cui et al., 2012).

2.5. Inhibition of CMV entry

TC-dependent entry was assessed using recombinant chimeric human PDGFRα-Fc (R&D Systems, Minneapolis, MN, USA) dissolved in PBS at stock concentration of 500 μg/mL. PC-dependent entry was assessed using the PC-specific human mAb 2-25, which was isolated and cloned from cultured memory B-cells of a healthy CMV-seropositive donor, or PC-specific rabbit mAb 57.4, which was isolated from a rabbit immunized with virions of a PC+ variant of CMV strain AD169 (Ha et al., 2017). Requirement for gB in viral entry was determined using mAb TRL345 (Trellis Bioscience), a human mAb targeting the AD-2 (site I) epitope of gB (Kauvar et al., 2015). Requirement for virion interactions with cell surface heparan sulfate proteoglycans was assessed using heparin sodium (ACROS organics^™) dissolved in PBS (Fisher Scientific, Fair Lawn, NJ, USA) at a stock concentration of 300 mg/mL.

Inhibitors were diluted at various concentrations in culture medium appropriate for each cell type, then mixed with matching aliquots of each virus; control mixtures contained virus with medium alone. Following incubation for one or two hours at 37°C, mixtures were transferred in triplicate to 96-well plates containing monolayer cultures of different cell types. Cultures were incubated in 37°C for two or three days, then immunofluorescent (IF) staining was used to detect nuclear CMV immediate early 1 and 2 (IE1/2) proteins indicative of viral infection. Cultures were fixed with 3% formaldehyde (Sigma) in PBS at room temperature for 30 min, permeabilized in 0.5% Triton-X (Fisher Scientific) in PBS on ice for 20 min, and blocked with 20% FBS in PBS (blocking buffer) for 30 min. Fixed cultures were incubated for one hour at room temperature with MAB810 (ThermoFIsher), an antibody recognizing an epitope common to both the IE1 and IE2 proteins, diluted 1:400 in blocking buffer, then washed with blocking buffer and incubated for one hour with Alexa Fluor^™ 594-conjugated goat anti-mouse IgG (ThermoFisher) diluted 1:200 in blocking buffer. After washing with blocking buffer, the numbers of IE1/2-positive nuclei per well were manually counted using a Nikon Eclipse TS100 fluorescence microscope. Percent inhibition was calculated by dividing the number of IE1/2-positive nuclei in wells incubated with treated viruses by the number of IE1/2-positive nuclei in wells incubated with medium-treated viruses (control). Images were captured using NIS-Elements version 4.0 and contrast and brightness adjusted using ImageJ (NIH).

2.6. Inhibition of CMV virion attachment to cell surfaces

Virus TB40/EE (1.9 x 10⁵ PFU) was incubated at 37°C for two hours with 900 ng/mL PDGFRα-Fc or for one hour with mAbs to gH/gL (124.4, 100 μg/mL), gH (223.4, 100 μg/mL), the PC (2-25, 80 μg/mL), gB (TRL345, 160 μg/mL), or to an irrelevant control (TRL308, 160 μg/mL) (Ha et al., 2017; Kauvar et al., 2015). Heparin sulfate interactions were disrupted by pre-incubating cell cultures with 300 μg/mL heparin sodium for one hour at 37 °C before adding virus TB40/EE. Mixtures were then chilled to 4°C and added to replicate wells of 16-well chamber slides containing HFF or ARPE-19 cells pre-chilled to 4°C. Cultures were incubated for one hour at 4°C to allow virion attachment, then washed three times with DMEM chilled to 4°C to remove unattached virions. One replicate was immediately fixed and stained as described above except using mAb 3A12 (Virusys Corporation, CA003-100) to detect the CMV pp65 lower matrix protein. The second replicate was incubated at 37°C for four hours to allow entry of attached virions prior to fixing and staining for pp65 using mAb 3A12.

2.7. Statistical Analyses

The 50% inhibitory concentration (IC₅₀) values were determined using Prism 5 (GraphPad Software, Inc., San Diego, CA, USA) as the infection points of four-parameter curves fitted by non-linear regression to plots of mean percentage of neutralization (from triplicate wells) vs. Log₁₀ (antibody concentration), as described previously (He et al., 2022). Statistical analyses were performed by using Prism 5 software. Paired t-tests were used to compare percent inhibitions as noted in the figure legends.

3. Results

3.1. NOK differentiation induces changes that permit PC-independent entry

Initial entry neutralization experiments were performed using ARPE-19 cells, or NOK cell monolayers that were either undifferentiated (uNOK) or differentiated (dNOK), then infected with PC-positive variants of laboratory strains AD169 (BADr) or Towne (TS15-rN) (Table 2). Viruses were incubated with serial dilutions of mAb TRL345, which is specific for a linear epitope in gB (Kauvar et al., 2015), or mAb 2-25, specific for an epitope unique to the PC (Ha et al., 2017). The percent neutralization for each mAb concentration was assessed after counting the number of cells expressing the IE1/2 proteins at day two (ARPE-19) or three (uNOK and dNOK) post-infection.

The gB-specific mAb TRL345 fully neutralized entry of both viruses into all three cell types with similar neutralizing potencies (Fig. 2 and Table 3). At the highest TRL345 concentrations tested the residual infectivity was < 4.3%. The PC-specific mAb 2-25 similarly inhibited entry of BADr into all three cell types with only 2.9% residual infectivity in dNOK cells. In contrast, while mAb 2-25 effectively neutralized TS15-rN entry into uNOK cells with only 2.7% residual infectivity, high concentrations mAb 2-25 failed to neutralize 31.5% of TS15-rN entry events into dNOK cells and 12.7% of entry events into ARPE-19 cells.

Figure 2. NOK differentiation induces resistance to CMV entry neutralization by the PC mAb 2-25.

Viruses BADr (strain AD169) or TS15-rN (strain Towne) were incubated for one hour at 37°C with gB mAb TRL345 or PC mAb 2-25, then added to 96-well plates containing monolayers of ARPE-19, uNOK, or dNOK cells. The number of infected cells in each well was determined by staining for the IE1/2 proteins at day two (ARPE-19) or three (NOK) post-infection and normalized as percentages of the maximum number of IE1/2-positive nuclei. Data are means of triplicate wells ± SEM.

Table 3.

Residual infectivity (%) after mAb neutralization

virus	mAb	ARPE-19	uNOK	dNOK
BADr	TRL345	0	0	0.8
	2-25	0	0	2.9
TS15-NR	TRL345	4.3	3.2	1.7
	2-25	13	2.7	32
TB40/EE	2-25	2.1	0	0
TB40/EF	2-25	13	4.6	13
GT1c	2-25	33	3.0	26
KL-7 FS	2-25	0	0	0
UxcAp14	2-25	0	0	3.0
UxcAp66	2-25	0	0	3.6
ABV	2-25	0	0	0

These results suggest that specific circumstances exist that enable CMV entry into epithelial cells by a PC-independent mechanism that is insensitive to neutralization by a PC-specific mAb. Such circumstances are likely dictated by a combination of cellular factors present at low prevalence in ARPE-19 cells, absent from uNOK cells, and induced to higher prevalence upon uNOK differentiation and of viral factors such as the relative abundance of PC and TC molecules carried on the envelope of BADr vs TS15-rN.

3.2. Reliance on PC-independent entry may be determined by both viral genetic and non-genetic factors that influence levels of functional PC in virions

The striking difference between BADr and TS15-rN regarding sensitivity to neutralization by the PC-specific mAb 2-25 using cells in a differentiated state suggested that strain-specific genetic factors may modulate the ability to access PC-independent epithelial cell entry. Comparison of BADr and TS15-rN whole genome sequences revealed that while both have deletions removing a large block of genes in the U_L/b’ region, the deletion in BADr is larger and removes the UL146, UL147, UL147A, and UL148 ORFs, which are retained in TS15-rN.

To determine if genes in this region contribute to PC-independent epithelial cell entry, three variants of the CMV Uxc strain were evaluated for resistance to neutralization by mAb 2-25. UxcAp14, a mixed stock initially isolated from urine by culture on ARPE-19 cells and subsequently passaged 14 times on ARPE-19 cells (Cui et al., 2017a), contains an intact U_L/b’ region (Corcoran et al., 2017). UxcAp66 and ABV are two independent BAC clones derived from the UxcAp14 stock. ABV has a 10.3-kb deletion that removes UL133 to UL147A but retains UL148, whereas UxcAp66 has a 31.2-kb deletion that removes IRS1 to US28 but retains an intact U_L/b’ region (Cui et al., 2017a; Qi et al., 2020). When assayed for neutralization by mAb 2-25, all three viruses were effectively neutralized for entry into ARPE-19, uNOK, and dNOK cells (Fig. 3A, Table 3). These results were somewhat surprising, as UxcAp14, which has an intact U_L/b’ region, was unable to utilize PC-independent epithelial cell entry.

Figure 3. Intra-strain variation of differentiation-induced resistance to neutralization by PC mAb 2-25.

Sensitivity to neutralization by mAb 2-25 was assessed as in Fig. 2 using (A) variants of CMV strain Uxc that retain (UxcAp14 and UxcAp66) or lack (ABV) the UL146-UL148 region, or (B) variants of strain TB40/E that express the PC at low (TB40/EF, GT1c) or high (TB40/EE, KL7-FS) levels.

Noting that UxcAp14, UxcAp66, and ABV were initially isolated and passaged on ARPE-19 cells to maintain epithelial tropism and PC expression, we reasoned that viruses with low PC expression or stocks in which functional PC levels have declined, perhaps due to prolonged cryogenic storage, may rely more heavily on the TC for entry. Therefore, a higher proportion of their entry events would be insensitive to PC inactivation by mAb 2-25.

To address this hypothesis, we evaluated four variants of the TB40/E strain, having low or high PC expression, for their ability to utilize PC-independent epithelial cell entry. In previous studies (Al Qaffas et al., 2021; Vo et al., 2020) we passaged an endothelial cell-adapted stock of strain TB40/E twice on fibroblasts to generate a stock designated TB40/EF that exhibited mutations cumulatively disrupting either UL128 or UL130 in 83% of genomes. Five subsequent passages on ARPE-19 cells resulted in a stock designated TB40/EE that exhibited no evidence for mutations disrupting PC subunit genes (Al Qaffas et al., 2021; Vo et al., 2020). The results of mAb 2-25 neutralization assays supported our hypothesis, in that mAb 2-25 fully neutralized TB40/EE entry into uNOK and dNOK cells with only 2.1% residual infectivity of ARPE-19 cells, while the TB40/EF stock exhibited much higher residual infectivity in ARPE-19 (12.6%) and dNOK (13.1%) cells, but not in uNOK cells (4.6%) (Fig. 3B, Table 3). The same experiment was then conducted using two BAC-cloned variants of the TB40/E strain, GT1c and KL7-FS. Virus GT1c is derived from the TB40-BAC4 clone, which has a unique single-nucleotide substitution within exon 2 of UL128 that does not alter the UL128 protein coding sequence but impairs mRNA splicing, resulting in reduced levels of PC expression and poor epithelial cell entry efficiency (Murrell et al., 2013). Virus KL7-FS, derived from the TB40-KL7 BAC clone, lacks the UL128 exon 2 mutation found in TB40-BAC4 and infects ARPE-19 cells efficiently (Sampaio et al., 2017). Similar to the TB40/EE stock, the KL7-FS virus, which lacks mutations impacting PC expression, was fully neutralized by mAb 2-25 (Fig. 3B, Table 3), while GT1c, expressing low PC levels, exhibited high residual infectivity of ARPE-19 and dNOK cells, but its entry into uNOK cells was effectively neutralized (Fig. 3B, Table 3). Thus, variants of the same CMV strain exhibited considerable variation in their utilization of PC-independent epithelial cell entry, and this correlated with their ability to express a functional PC.

The GT1c stock tested above was produced after multiple fibroblast passages and had been stored under liquid nitrogen for three years. To determine if serial passage or prolonged cryogenic storage impacts reliance on PC-independent entry, a fresh GT1c stock, designated GT1c/F, was produced by transfection of GT1c BAC DNA into HFFs followed by one passage on HFF, and storage in liquid nitrogen for either one week or nine months. In contrast to the results for the original GT1c stock shown in Fig. 3B, mAb 2-25 effectively neutralized infectivity of the GT1c/F stock for entry into ARPE-19, uNOK, and dNOK cells after either short- (one week) or long-term (nine months) storage in liquid nitrogen (Fig. S1A). In addition, mAb 57.4, which recognizes a PC epitope different than that of 2-25 (Ha et al., 2017), effectively neutralized entry of the original GT1c stock into uNOKs but not dNOKs (Fig. S1B), indicating that resistance is not unique to mAb 2-25 but extends to other PC-specific antibodies.

3.3. PC-independent entry is supported by different epithelial cells but not by aortic endothelial cells

To determine if additional cell types can support PC-independent entry, the ability of mAb 2-25 to fully neutralize entry of CMV viruses BADr, TS15-rN, TB40/EE, or TB40/EF into HAEC (aortic endothelial cells) or into uN/TERT, dN/TERT, uHAK, dHAK, uHTE, or dHTE cells was assessed (note: as it is not clear whether N/TERT cells are mucosal or cutaneous, the four epithelial cell lines are hereafter collectively referred to as “mucocutaneous”).

Entry of all four CMV stocks into HAEC was completely neutralized by mAb 2-25, indicating that this cell type does not support PC-independent entry (Fig. 4A, Table 4). As observed for uNOKs, mAb 2-25 effectively neutralized entry of all four viruses into uN/TERT and uHAK cells, whereas at high mAb 2-25 concentrations, BADr, TS15-rN, and TB40/EF exhibited varying amounts of residual infectivity for dN/TERT and dHAK cells (Fig. 4B-C, Table 4). In contrast, TS15-rN, TB40/EE, and TB40/EF exhibited residual entry into both uHTE and dHTE, and unlike NOK, N/TERT, and HAK cells, differentiation of HTE cells resulted in decreased levels of residual infectivity (Fig. 4D, Table 4). High levels of residual infectivity for uHTE cells were also observed for stocks GT1c (70%) and GT1c/F (35%) (Fig. S1C). Lastly, to confirm that HTE cell entry is selectively resistant to neutralization by PC-specific mAbs and that resistance does not extend to neutralizing antibodies targeting other glycoprotein complexes, entry into uHTE cells of TB40/EF and TB40/EE neutralized with gB-specific mAb TRL345 was assessed. Both TB40/EF and TB40/EE were effectively neutralized by mAb TRL345 (Fig. S1D).

Figure 4. Endothelial cell infection is not resistant to mAb 2-25 and resistance is variable for mucocutaneous epithelial cell entry.

Sensitivity to neutralization by mAb 2-25 was assessed as in Fig. 2 using the indicated viruses and cells.

Table 4.

Residual infectivity (%) after neutralization with mAb 2-25

virus	HAEC	uN/TERT	dN/TERT	uHAK	dHAK	uHTE	dHTE
BADr	2.6	2.2	28	0.8	4.2	9.9	15
TS15-rN	1.7	1.2	17	0	7.2	30	11
TB40/EE	0.3	0	1.1	0	0	12	1.4
TB40/EF	2.3	0	5.5	1.1	13	78	42

These cumulative results suggest that differentiation enhances PC-independent entry into NOK, N/TERT, and HAK cells, whereas uHTE cells constitutively support relatively high levels of PC-independent entry that is partially reduced by differentiation.

3.4. Entry of CMVs with PC mutations is enhanced by differentiation of NOK, HAK, and N/TERT cells but is constitutively high in HTE cells

In the studies described above, the extent to which different CMV strains utilized PC-independent entry to infect dNOK, dN/TERT, dHAK, and HTE cells varied considerably, not only by strain but also by different variants of the TB40/E strain or by different stocks of the GT1c variant. This suggests that utilization of PC-independent entry might vary depending on the levels of functional PCs present in the virion envelope, which in turn could be impacted by viral genetics as well as cell culture or virus storage conditions. To eliminate PC expression as a variable, viruses in which PC expression is precluded by genetic disruptions in PC subunit genes were assessed for entry into undifferentiated or differentiated epithelial cells.

For the strain AD169 background we used AD169varUK, which has a frame-shift insertion in UL131A; for strain Towne we used RC2626, which has a frame-shift insertion in UL130; and for strain Merlin we used virus pAL1119, which has an in-frame premature stop codon in UL128 (Table 2). Each of these mutations are known to produce defects in PC formation that severely impair infectivity in epithelial cells and endothelial cells (Cha et al., 1996; Ho et al., 1984; McVoy and Mocarski, 1999; Murphy et al., 2003; Sinzger et al., 1999; Stanton et al., 2010). MRC-5, ARPE-19, as well as undifferentiated or differentiated NOK, HAK, HTE, and N/TERT cultures were simultaneously infected by each of these three viruses and entry efficiencies were determined by staining for the IE1/2 proteins.

Compared to MRC-5 entry, entry of PC-deficient mutants into ARPE-19 cells was decreased by 1.9 logs (81-fold) for AD169, 3.2 logs (1770-fold) for Towne, and 2.6 logs (414-fold) for Merlin (Fig. 5A, Table 5). For all three viruses, entry into differentiated NOK, HAK, HTE and N/TERT cells was similar to ARPE-19 entry, while entry into undifferentiated cells was dramatically lower with the exception of uHTE cells (Fig. 5A, Table 5). These results are consistent with previous data suggesting that differentiation of NOK, HAK, and N/TERT cells induces factors involved in supporting PC-independent entry, while in HTE cells such factors are constitutively present regardless of the differentiation state.

Figure 5. Differentiation enhances PC-independent entry into NOK, HAK, and N/TERT cells.

(A) Viruses with point mutations that disrupt PC subunit genes in strains AD169 (AD169varUK, UL131A frame shift), Towne (RC2626, UL130 frame shift), or Merlin (pAL1119, UL128 in-frame stop codon) were added to 96-well cultures containing monolayers of the indicated cells. IE1/2-positive nuclei were detected by IF staining at day two (MRC-5 and ARPE-19) or three (all others) post-infection. (B) Schematic illustration of the galK cassette (purple box) insertion site resulting in deletion of PC subunit genes, including the entire UL130 and UL131A ORF, and exons 1 and 2 of the UL128 ORF. (C) ΔPC mutants of the indicated viruses were analyzed as in Fig. 5A. Increases in viral entry efficiencies into differentiated versus undifferentiated cells were statistically significant for NOK, HAK, and N/TERT cells (p < 0.0001, paired t-test) but not for HTE cells.

Table 5.

Impact of mucocutaneous epithelial cell differentiation on entry efficiency^a of PC-deficient viruses

virus	ARPE-19	uNOK	dNOK	uHTE	dHTE	uHAK	dHAK	uN/TERT	dN/TERT
AD169varUK	−1.9	−4.6	−2.3	−1.0	−2.0	−4.8	−1.9	−4.5	−1.8
RC2626	−3.2	−4.6	−2.4	−2.6	−2.4	ndb	−3.5	ndb	−3.6
pAL1119	−2.6	−4.0	−2.4	−2.3	−2.2	ndb	−3.0	ndb	−3.2
GT1c ΔPC	−1.9	−4.8	−2.6	−1.6	−2.5	−4.3	−2.9	−4.2	−2.3
KL7-FS ΔPC	−1.7	ndb	−3.3	−2.6	−2.8	ndb	−3.0	nde	−3.2
pAL1119 ΔPC	−2.0	ndb	−2.7	−2.3	−2.7	−4.4	−2.6	−4.4	−3.1
pAL1128 ΔPC	−2.5	ndb	−3.9	−3.2	−2.6	ndb	−3.6	ndb	−3.0

^aLog₁₀ (#IE1/2-positive nuclei in epithelial cells/#IE1/2-positive nuclei in MRC-5 cells)

^bnd, not detected

To eliminate any possibility that entry by these PC-mutant viruses was mediated by truncated subunits or partial complexes, the sequences encoding all of UL130 and UL131A and the first and second exons of UL128 were deleted and replaced by a galK cassette (Warming et al., 2005) in the strain TB40/E (GT1c and KL7) or Merlin backgrounds (pAL1119 and pAL1128) (Fig. 5B, Table 2). The resulting viruses were designated “ΔPC”, and loss of a functional PC was confirmed by failure to form foci on ARPE-19 cell monolayers (not shown).

Compared to MRC-5 cells, entry efficacy of ΔPC viruses was dramatically reduced in all epithelial cell lines, and similar to the results obtained above using viruses with point mutations, ΔPC viruses entered HTE cells at comparable levels regardless of differentiation status, while entry into uNOK, uHAK, or uN/TERT cells was low or undetectable but increased by several logs upon differentiation (Fig. 5C, Table 5). These results confirmed that CMV can utilize a PC-independent mechanism for entry into dNOK, dHAK, and dN/TERT cells, whereas HTE cells constitutively support PC-independent entry regardless differentiation. In contrast, PC-independent entry was poorly supported by uNOK, uHAK, or uN/TERT cells.

3.5. CMV entry into mucocutaneous epithelial cells requires either the TC or the PC, but not both

Next, we wondered whether the PC-independent entry observed with viruses lacking the PC was TC-dependent. Studies have shown that PDGFRα is the TC receptor on fibroblasts and that pretreatment of CMV virions with a soluble chimeric PDGFRα-Fc recombinant protein inhibits CMV entry into fibroblasts and endothelial cells (Stegmann et al., 2017). Other studies have shown that PDGFRα-Fc neutralizes infectivity in a manner similar to that of an anti-gO antibody, i.e. by interacting with gH/gL/gO trimers in the virion envelope (Wu et al., 2017). Therefore, recombinant PDGFRα-Fc was incubated with ΔPC CMVs for two hours prior to infection and entry efficacies were compared to mock-treated controls. As seen previously, ΔPC CMVs were unable to enter uNOK, uHAK, or uN/TERT cells; however, entry into MRC-5, ARPE-19, dNOK, dHAK, dN/TERT, uHTE, or dHTE cells was neutralized by recombinant PDGFRα-Fc with inhibition rates from 91% to 100% (Table S1), indicating that the TC is essential for PC-independent entry into mucocutaneous epithelial cells.

To determine if PC-expressing CMVs also require the TC for entry into mucocutaneous epithelial cells, variants of strain TB40/E expressing low (GT1c) or high (KL7-FS) PC levels were incubated with recombinant PDGFRα-Fc and entry inhibition was assessed as above. Recombinant PDGFRα-Fc inhibited entry of both viruses into MRC-5 or ARPE-19 cells by 95% to 100% (Fig. 6A, Table 6). In contrast, PDGFRα-Fc only partially inhibited entry into undifferentiated mucocutaneous cells (20-81%), suggesting that under some circumstances CMV can enter these cells by a TC-independent mechanism. Moreover, differentiation increased TC-dependence of GT1c entry into all four mucocutaneous cell types (Fig. 6A, Table 6). This suggests that PC-high viruses such as KL7-FS can efficiently utilize a PC-mediated entry pathway that is not influenced by differentiation, whereas PC-low viruses such as GT1c use PC-mediated entry inefficiently and are therefore more reliant on a PC-independent/TC-dependent pathway that is induced by differentiation.

Figure 6. CMV can enter mucocutaneous epithelial cells using either the TC or the PC.

(A) PC-low (GT1c) or PC-high (KL7-FS) variants of strain TB40/E were treated with medium alone or with medium containing 0.9 μg/mL PDGFRα-Fc for two hours at 37°C. Viruses were then added to the indicated cell monolayers and the numbers of infected cells were determined as described in Fig. 5A. Inhibition of entry was expressed as the percentage of the infected cells observed in mock-treated cultures. (B) Viruses GT1c or TS15-rN were treated with medium alone or incubated with medium containing 0.9 μg/mL PDGFRα-Fc for two hours, 80 μg/mL mAb 2-25 for one hour, or with both sequentially, then added to the indicated cell monolayers and analyzed as above. Data in (A) and (B) are means of triplicate wells ± SEM. Paired t-test p-values indicate significance of differences in PDGFRα-Fc inhibition between undifferentiated and differentiated cells.

Table 6.

TC-dependent entry^a by PC-expressing viruses

virus	MRC-5	ARPE-19	uNOK	dNOK	uHTE	dHTE	uHAK	dHAK	uN/TERT	dN/TERT
GT1c/F	100	96	57	93	81	96	55	96	77	99
KL7-FS	99	95	54	56	53	52	48	61	20	47

^a% inhibition after treatment with 900 ng/mL PDGFRα-Fc for two hours

Lastly, to determine if TC-independent entry into mucocutaneous epithelial cells is PC-dependent, viruses GT1c and TS15-rN were incubated with mAb 2-25 or PDGFRα-Fc separately or in combination and inhibition of entry was assessed as above. PDGFRα-Fc strongly inhibited entry of both viruses into MRC-5 and ARPE-19 cells but only partially inhibited entry into uNOK, uHAK, and uN/TERT cells (71-77% for GT1c, 34-56% for TS15-rN), while mAb 2-25 inhibited entry into NOK, HAK, and N/TERT cells (85-100%) (Fig. 6B, Table 7). Importantly, the combination of 2-25 and PDGFRα-Fc completely blocked entry of each virus into all epithelial cell types (Fig. 6B, Table 7), suggesting that functions necessary for entry into these cell types can be provided by either TC or PC.

Table 7.

Impact of cell type and differentiation on utilization of TC- and PC-dependent entry

virus	entrya	MRC-5	ARPE-19	uNOK	dNOK	uHTE	dHTE	uHAK	dHAK	uN/TERT	dN/TERT
TS15-rN	TCD	98	95	34	39	36	78	52	60	56	74
	PCD	−	96	98	92	72	65	99	99	99	86
	TCD + PCD	−	100	99	98	100	99	100	100	99	100
GT1c	TCD	99	97	71	90	97	96	72	94	77	92
	PCD	−	77	100	84	10	22	97	96	99	90
	TCD+ PCD	−	100	99	98	100	99	99	100	100	100

^aTCD, TC-dependent entry (% inhibition after treatment with 900 ng/mL PDGFRα-Fc for 2 h); PCD, PC-dependent entry (% inhibition after treatment with 80 μg mAb 2-25 for 1 h);

“-”, not determined

These results indicate that during mucocutaneous epithelial cell entry the TC and PC can often serve redundant functions, and that the proportion of entry events governed by each complex depends on the cell type and differentiation status. For example, only 10% of GT1c entry events into uHTE cells are PC-dependent (blocked by mAb 2-25) and 97% are TC-dependent (blocked by PDGFRα-Fc), while in contrast, TS15-rN entry events into uNOK cells are 98% PC-dependent and 34% TC-dependent (Table 7). As noted above, viruses with high PC expression (TS15-rN) are better able to utilize PC-dependent entry pathways and are less reliant on the TC, whereas those with low PC expression (GT1c) use PC-dependent entry inefficiently and are consequently more reliant on the TC.

3.6. PDGFRα-Fc and neutralizing antibodies to gB or the PC block CMV virion entry but not attachment

For many viruses, infection begins with attachment of the virion to target cell surfaces through interactions between virion envelope glycoproteins and cell surface HSPGs (Aquino and Park, 2016; Cagno et al., 2019; Song et al., 2001). Consequently, heparin, a heparan sulfate mimetic, blocks virion attachment and subsequent infection by binding to virion envelope glycoproteins and competitively preventing virion-HSPG interactions (Song et al., 2001). CMV infection of fibroblasts requires HSPG interactions (Compton et al., 1993) and heparin effectively blocks CMV infection of fibroblasts and ARPE-19 epithelial cells (Shoup et al., 2020; Zoepfl et al., 2021). To determine if HSPG dependence extends also to mucocutaneous epithelial cells, culture medium lacking or containing heparin was added to mucocutaneous cell cultures prior to addition of viruses KL7-FS or TS15-rN and infection was assessed as described above. Infection of all cell types by both viruses was effectively inhibited by heparin treatment with inhibition greater than 97% (Fig. S2).

To further define the mechanisms by which heparin, PDGFRα-Fc, or neutralizing mAbs to gB or the PC prevent CMV infection, TB40/EE virions were incubated with medium alone or with medium containing heparin, PDGFRα-Fc, or mAbs TRL345 (gB), 2-25 (PC), or TRL308 (control). The mixtures were then chilled to 4°C and added to prechilled MRC-5 or ARPE-19 monolayers prior to incubation for one hour at 4°C to allow virion attachment but prevent membrane fusion. Cells were then fixed and stained for the abundant virion-associated tegument protein pp65.

As shown in Fig. S3, virion attachment was evidenced by punctate signals on cells exposed to virions treated with medium or control mAb TRL308, as well as with PDGFRα-Fc, TRL345, or 2-25, while no punctate signals were detected on cells exposed to virions treated with heparin or medium that did not contain CMV virions. Identical replicate cultures exposed to virions for one hour at 4°C were shifted to 37°C to allow membrane fusion and after an additional four hours at 37°C cells were fixed and stained for pp65. In cells exposed to virions treated with medium or control mAb TRL308, nuclear accumulation of pp65 was evident as following membrane fusion virion-associated pp65 was released into the cytoplasm and was subsequently translocated to the nucleus (Fig. S3). In contrast, no pp65-positive nuclei were observed in MRC-5 or ARPE-19 cells following treatment with heparin, TRL345, or PDGFRα-Fc (Fig. S3). As the PC is not required for fibroblast entry, PC mAb 2-25 neutralizes CMV infection of ARPE-19 epithelial cells but has no neutralizing activity against fibroblast entry (Cui et al., 2017b; Ha et al., 2017). Consistent with this, pp65-positive nuclei were not observed in ARPE-19 cells exposed to 2-25-treated virions for four hours at 37°C but were observed in similarly treated MRC-5 fibroblasts (Fig. S3). Taken together, these results indicate that heparin blocks virion attachment, while TRL345, 2-25, and PDGFRα-Fc do not block attachment, but rather, block one or more subsequent entry steps.

3.7. The PC, but not the TC, is required for CMV infection of PTECs

The previously described experiments were performed using mucocutaneous-derived epithelial cell lines that were genetically manipulated to bypass certain cell-cycle restriction points. To evaluate CMV infection of primary mucosal cells, PTECs were isolated from the palatine tonsil of four different donors. As PTECs have limited capacity to divide in culture and efforts to differentiate PTECs resulted in loss of the monolayer, CMV infection was characterized using undifferentiated PTECs passaged only two or three times after isolation. Viruses TS15-rN or GT1c were incubated with PDGFRα-Fc, mAb 2-25, or heparin prior to addition to PTEC monolayers and their inhibitory effects were used to determine the relative importance of the TC, PC, or HSPGs for PTEC infection.

Heparin and mAb 2-25 effectively blocked entry of both viruses into ARPE-19 cells and into PTECs 51806A and 3503, and while 0.9 μg/mL PDGFRα-Fc also impaired entry of both viruses into ARPE-19 cells and entry of GT1c into both PTECs, it had no impact on PTEC entry by TS15-rN (Fig. 7A). To ensure that sufficient amounts of PDGFRα-Fc were used, this experiment was repeated with 0.9, 3, or 10 μg/mL PDGFRα-Fc, with similar results (Fig. 7B). Images of TS15-rN-infected cells stained for the IE1/2 proteins showed profound decreases in IE1/2-positive nuclei following treatment with neutralizing mAbs 2-25 (PC) or TRL345 (gB) or heparin, but no impact of PDGFRα-Fc treatment on the number of IE1/2-positive nuclei following TS15-rN infection of three PTEC cultures (Fig. 7C). These, results suggest that PTEC entry by GT1c requires both the PC and the TC, whereas PTEC entry by TS15-rN is PC-dependent but entirely TC-independent.

Figure 7. The TC is not required for efficient entry into PTECs.

(A) PC-high (TS15-rN) or PC-low (GT1c) viruses were treated with medium alone or with medium containing 0.9 μg/mL PDGFRα-Fc, 80 μg/mL mAb 2-25, or both sequentially, prior to addition to ARPE-19 or PTEC monolayers. Alternatively, mock-treated viruses were added to cell cultures containing 300 μg/mL heparin. The numbers of infected cells were determined as described in Fig. 5; inhibition of entry was expressed as the percentage of infected cells observed in mock-treated cultures. (B) The experiment in (A) was repeated using 0.9, 3, or 10 μg/mL of PDGFRα-Fc. Data in (A) and (B) are means of triplicate wells ± SEM. Paired t-test p-values indicate significance of differences in inhibition between 0.9 μg/mL PDGFRα-Fc and 0.9 μg/mL PDGFRα-Fc+2.25 (A) or between 10 μg/mL PDGFRα-Fc and 10 μg/mL PDGFRα-Fc+2-25 (B). (C) Representative images of TS15-rN-infected cultures stained for IE1/2 following treatment medium alone (φ) or medium containing 0.9 μg/mL PDGFRα-Fc, 80 μg/mL mAb 2-25, 160 μg/mL mAb TRL345, or 300 μg/mL heparin. Original magnification: 100x.

As viruses TS15-rN and GT1c exhibited significant differences in their reliance on TC for PTEC entry, similar experiments were conducted using additional CMV strains, including PC-positive variants of strains AD169 (BADr), Uxc (UxcAp15), and TB40/E (KL-7 FS and TB40/EE). Heparin and neutralizing mAbs 2-25 (PC) and TRL345 (gB) were highly effective in blocking infection of these viruses into ARPE-19 cells and two PTECs, but inhibition of PTEC entry by PDGFRα-Fc was variable, with PTEC entry by UxcAp15 inhibited only 3-22%, TB40/EE inhibited 29-73%, and BADr and KL7-FS inhibited 49-77% (Fig. 8).

Figure 8. Utilization of TC-independent entry into PTECs varies with CMV strain.

Variants of strains AD169 (BADr), Uxc (UxcAp15), and TB40/E (KL7-FS, TB40EE) were analyzed as described in Fig. 7. Data are means of triplicate wells ± SEM. Paired t-test p-values indicate significance of differences in inhibition between 0.9 μg/mL PDGFRα-Fc and 0.9 μg/mL PDGFRα-Fc+2-25.

This suggested that genetic features of UxcAp15, which is more genetically authentic than BADr or TB40/E, might exist that enhance its ability to utilize TC-independent entry. We thus explored the ability of ABV and UxcAp66 (BAC-cloned viruses derived from UxcAp14) to utilize TC-independent entry. PDGFRα-Fc effectively blocked (>95%) entry of all three variants into MRC-5 fibroblasts (Fig. 9) but, as previously shown, was relatively ineffective at blocking UxcAp15 entry into both PTEC and ARPE-19 cells, and while inhibition was significantly higher for ABV and UxcAp66, it was still under 80% (Fig. 9). Thus, it appears that loss of sequences within regions deleted in both BAC-derived viruses (US6 to US28 for UxcAp66, UL146 to UL150 for ABV) may result in decreased ability to utilize TC-independent entry.

Figure 9. Utilization of TC-independent entry into PTECs varies between variants of strain Uxc.

PDGFRα-Fc inhibition of entry in to PTECs, MRC-5, or ARPE-19 cells was determined using the UxcAp15 stock and two BAC-cloned viruses, ABV and UxcAp66, that were derived from the UxcAp14 stock. (A) Strain UxcAp15, ABV, or UxcAp66 were incubated with medium alone (φ) or medium containing 0.9 μg/mL PDGFRα-Fc and analyzed as described in Fig. 6A. Data are means of triplicate wells ± SEM. Paired t-test p-values indicate significance of differences in inhibition of strain UxcAp15 versus ABV viruses. (B) Representative images of cultures from the experiment in panel (A). Original magnification: 100x.

4. Discussion

Current strategies to prevent or treat CMV infections remain suboptimal. Toxicities and resistance associated with CMV antivirals complicate the management of immunocompromised patients, and while HIG or high dose acyclovir can partially mitigate the risk of secondary fetal infection following primary maternal infection (Nigro et al., 2023), screening in early pregnancy is required to identify primary maternal infections and strategies to identify pregnancies at risk of secondary infection in the context of non-primary maternal infections remain elusive. Consequently, a need remains for effective therapeutics that are safe for use during pregnancy, as well as for prophylactic measures either through active immunization or by passive administration of antibodies to pregnant women during the first trimester.

Prior and current efforts to develop CMV vaccines or therapeutic mAbs have largely focused on neutralizing antibodies against envelope glycoproteins involved in viral entry. However, CMV entry mechanisms are complex and utilize different virion complexes for entry into different cell types (Fig. 1A). During fibroblast entry, TC interaction with PDGFRα on the cell surface triggers conformational changes in gB that drive fusion of the virion envelope with the plasma membrane (Kabanova et al., 2016; Soroceanu et al., 2008; Stegmann et al., 2017; Wu et al., 2018; Wu et al., 2017). In contrast, entry into endothelial and epithelial cells requires endocytosis, and fusion occurs between the virion envelope and the endosomal membrane in a process requiring the PC in addition to gB and the TC (E et al., 2019; Martinez-Martin et al., 2018). Thus, depending on the cell type, antibodies can have vastly different neutralizing activities. At present these activities are defined using a limited set of established cell culture models; for example, HFF or MRC-5 fibroblasts are used to assess neutralizing activity against non-endocytic gB/TC-dependent entry, while ARPE-19 cells are used to assess neutralizing activity against endocytic gB/TC/PC-dependent entry (Fig. 1A). However, the accuracy whereby these cell lines reflect the mechanisms used by CMV to infect the broad spectrum of cell types that it encounters in vivo remains largely unexplored. Consequently, although gB, the TC, and the PC are considered key immunogens for induction of humoral immunity and potential targets for therapeutic mAbs, the ability of antibodies targeting each complex to protect against infection of key cell types in vivo remains unclear. Even so, partial efficacy of experimental vaccines and antibody therapeutics suggests that improved efficacy can be achieved through optimization, and that this can be informed by a better understanding of the mechanisms and glycoprotein complexes required for CMV entry into cell types that are relevant in vivo.

The present studies explored inhibition of CMV entry into epithelial cells of mucocutaneous origin, motivated by the premise that maternal infections largely occur through oral exposure to CMV shed in the urine or saliva of young children. Therefore, antibodies capable of neutralizing CMV entry into mucosal epithelial cells could potentially prevent virus acquisition if present in maternal saliva. In addition, as mucosal layers represent a range of differentiation stages that can be modeled in vitro (Andrei et al., 2005), and as cell differentiation plays an important role in the life cycle and natural history of some viruses (Hong and Laimins, 2013; Johnson et al., 2005; Nawandar et al., 2015; Temple et al., 2014), we sought to determine if differentiation influences the mechanisms whereby CMV enters mucosal epithelial cells.

Using four cell lines we found that the efficiency of CMV entry into mucocutaneous cells is not significantly influenced by differentiation. However, for three of the four cell lines (NOK, HAK, N/TERT), differentiation significantly enhanced resistance to neutralization by a PC-specific mAb, indicating an enhanced use of PC-independent entry. In contrast, undifferentiated HTE cells exhibited a significant level of PC-mAb resistance that was not enhanced by differentiation. Similar results were obtained using genetic disruption of PC expression to confirm that CMV viruses lacking the PC can infect mucocutaneous epithelial cells, and that for NOK, HAK, and N/TERT cells, PC-independent entry is significantly enhanced by differentiation, presumably through alterations in the expression of cellular factors.

The use of soluble PDGFRα-Fc to block TC on PC-null viruses further determined that in the absence of PC, CMV entry into these cells is entirely TC-dependent. Similarly, mucocutaneous cell entry of PC+ viruses could only be fully blocked by combining PDGFRα-Fc with a PC-specific neutralizing mAb. Thus, in the context of mucocutaneous cells, TC and PC appear to be frequently redundant, such that one or the other is required for CMV entry, but not both. Consistent with this, the PC-high virus TS15-rN exhibited less reliance on TC-dependent entry than PC-low viruses derived from TB40-BAC4, suggesting that the PC/TC ratio in the virion may influence the extent to which TC-independent versus PC-independent mechanisms are utilized for mucocutaneous cell entry. As these results were observed using immortalized cell lines, we further sought to validate findings using primary mucosal cells. While there was variability associated with both virus strain and cell donor, in general CMV entry into low passage PTEC cultures was predominantly PC-dependent and to a large extent TC-independent. Fig. 1B summarizes these findings for glycoprotein complex requirements for CMV entry into the mucocutaneous cells studied.

Overall, these studies suggest that in vivo CMV entry mechanisms are more complex than the established ARPE-19- and fibroblast-based cell culture models suggest, and may depend not only on the cell type but also on the tissue of origin and the state of cell differentiation. Moreover, given that the key cell types and tissues that CMV infects to ultimately cause fetal infection and disease are not known, the relative importance of antibodies targeting the TC vs the PC remains difficult to surmise. At least with respect to mucosal immunity, simultaneously targeting both complexes may be most advantageous.

A number of animal studies have examined the gH/gL dimer as a potential vaccine immunogen (Cui et al., 2020; Cui et al., 2019; Loomis et al., 2013; Wen et al., 2014; Wussow et al., 2014). The TC has received considerably less attention, in large part due to the highly polymorphic nature of gO (Suarez et al., 2019), although neutralizing antibodies to gO have been reported (Chin et al., 2022; Gerna et al., 2016). More recent vaccine strategies have favored the PC over gH/gL, as it contains gH/gL but also epitopes unique to the complex that induce potent epithelial-specific neutralizing repsonses. Thus, for example, the Moderna mRNA-1647 vaccine combines the PC with gB (Fierro et al., 2024; John et al., 2018; Wu et al., 2024). Similarly, therapeutic mAbs targeting gH, the PC, and gB are in development (Ye et al., 2022) a number of gH-specific mAbs have been evalulated, alone or in bivalent combinations with PC-specific mAbs, for treatment or prevention of CMV infections in immune-compromised patients (Nuevalos et al., 2023). Two gB-specific mAbs have recently transitioned to clinical phase of development. TRL345, the gB mAb used in our studies, is currently undergoing phase 1 evaluation for safety (NCT05408091), while NPC-21 has completed phase 1 (Furihata et al., 2022) and is currently in a phase 2 study for prevention of CMV infections in kidney transplant recipients (NCT04225923).

Throughout our studies the gB mAb TRL345 fully blocked entry of all CMV strains and variants into all cell types, and this activity was not affected by differentiation. Thus, cumulative evidence supports the current consensus that gB is essential for entry into all cell types and that antibodies to gB should thus neutralize CMV entry into all cell types in vivo. In addition to neutralization, antibodies to gB can promote NK cell-mediated anti-CMV activities (Jenks et al., 2019) that are thought to contribute the partial efficacy of a gB subunit vaccine (Nelson et al., 2018). Recent elucidation of the prefusion conformation of CMV gB (Liu et al., 2021) may provide new opportunities to enhance the efficacy of gB-based vaccines through strategies to stabilize gB in the prefusion conformation, or to discover therapeutic mAbs specific to prefusion epitopes.

As heparin was similarly able to block CMV infection of all cell types, targeting initial virion attachment via cell surface HSPGs may also provide broad spectrum neutralizing activities. While gM/gN has been implicated in mediating virion attachment, this function was largely inferred from its capacity to bind heparin (Kari and Gehrz, 1992, 1993). Additional studies are thus warranted to determine the molecular events and players involved in virion attachment, and to further explore gM/gN as a vaccine immunogen or therapeutic antibody target.

Supplementary Material

1

Figure S1. Differentiation-induced PC-independent entry is stock-dependent. (A) A freshly produced stock of virus GT1c (GT1c/F) was stored under liquid nitrogen for one week or nine months, then incubated for one hour at 37°C with PC mAb 2-25 and added to 96-well plates containing monolayers of ARPE-19, uNOK, or dNOK cells. (B) The original stock of GT1c was analyzed as in panel (A) but using PC mAb 57.4 instead of 2-25. (C) The original GT1c stock or the freshly produced GT1c/F stock were incubated for one hour at 37°C with PC mAb 2-25, then added to 96-well cultures containing uHTE cells. (D) Virus stocks TB40/EE or TB40/EF were incubated for one hour at 37°C with gB mAb TRL345, then added to 96-well plates containing monolayers of uHTE cells. For all panels, number of infected cells in each well was determined by staining for the IE1/2 proteins at day two (ARPE-19) or three (uNOK, dNOK, uHTE) post-infection and normalized as percentages of the maximum number of IE1/2-positive nuclei; data shown are means of triplicate wells ± SEM.

2

Figure S2. Heparin fully blocks CMV entry into MRC-5, ARPE-19, or mucocutaneous epithelial cells. Cell monolayers were pre-incubated with medium alone or with medium containing 300 μg/mL heparin for one hour at 37°C before addition of viruses KL7-FS or TS15-rN. Inhibition of infection was determined as described in Fig. S1. Differences in heparin inhibiton of entry into undifferentiated versus differentiated cells were not statistically significant for any cell type (p >0.12, paired t-test).

3

Figure S3. Heparin blocks CMV virion attachment, whereas PDGFRα-Fc and neutralizing antibodies to gB or to the PC block entry but not attachment. Virus TB40/EE was incubated at 37°C with medium alone or with medium plus 160 μg/mL mAb TRL308 (control mAb), 160 μg/mL gB mAb TRL345, or 80 μg/mL PC mAb 2-25 for one hour, or with 900 ng/mL PDGFRα-Fc for two hours. The mixtures were then cooled to 4°C and added to HFF (left) or to ARPE-19 (right) cell monolayers in 16-well chamber slides that were pre-cooled to 4°C. Replicate cultures were mock-infected (no virus) or were incubated for one hour at 37°C with 300 μg/mL heparin prior to cooling to 4°C and subsequent infection. Cultures were maintained at 4°C for one hour followed by washing three times with cold DMEM. One set of cultures was fixed immediately and stained for the CMV tegument protein pp65 (blue arrows). A replicate set of cultures was shifted to 37°C for four hours before staining for pp65 (red arrows). Representative images are shown. Original magnification: 100x.

Highlights.

• Mucocutaneous cells do not fully adhere to the paradigms established by standard cell culture models

• CMV entry into cells encountered in vivo is more complex than has been suggested by standard cell culture models

• These findings are important for the development of vaccines, immunotherapeutics, or other inhibitors that target virion entry

Data availability

Data will be made available on request.