Neutralizing antibodies to gB based CMV vaccine requires full length antigen but reduced virus neutralization on non-fibroblast cells limits vaccine efficacy in the guinea pig model

Abstract

Cytomegalovirus is a leading cause of congenital disease and a vaccine is a high priority. The viral gB glycoprotein is essential for infection on all cell types. The guinea pig is the only small animal model for congenital CMV (cCMV), but requires guinea pig cytomegalovirus (GPCMV). Various GPCMV gB vaccine strategies have been investigated but not a full length protein. Previous GPCMV gB vaccines have failed to fully protect against cCMV, with approximately 50% efficacy. In an effort to define the basis of GPCMV gB based vaccine failure, we evaluated recombinant defective Ad vectors encoding GPCMV gB full length (gBwt), or truncated protein lacking transmembrane domain (gBTMD). Both candidate vaccines evoked high anti-gB titers and neutralized virus infection on fibroblast cells but had varying weaker results on non-fibroblasts (renal epithelial and placental trophoblasts). Non-fibroblast cells are dependent upon the viral pentamer complex (PC) for endocytic pathway cell entry unlike fibroblasts cells that express the viral receptor platelet derived growth factor receptor alpha (PDGFRA) to enable entry by direct cell fusion independent of the PC. Anti-gBwt sera was approximately 2-fold (renal epithelial) to 3-fold (fibroblasts) more effective at neutralizing virus compared to anti-gBTMD sera. Both gB vaccines were weakest against virus neutralization on trophoblasts. Knockout of PDGFRA cell receptor on fibroblast cells (GPKO) rendered virus dependent upon the PC pathway for cell entry and anti-gB GPCMV NA₅₀ was more similar to epithelial cells. In a gBwt vaccine protection study, vaccination of animals significantly reduced, but did not prevent dissemination of wild type GPCMV challenge virus to target organs. Depletion of complement in vivo had limited impact on vaccine efficacy. Overall, a full length gB antigen has the potential to improve neutralizing antibody titer but fails to fully prevent virus dissemination and likely congenital infection.

Introduction

Human cytomegalovirus (HCMV) is a ubiquitous pathogen and a leading cause of congenital disease, occurring in approximately 1% of newborns in the US [1]. In congenital HCMV infection, the greatest risk is to mothers who acquire a primary infection during pregnancy and consequently a vaccine against congenital CMV (cCMV) is a high priority. Any intervention strategy, or therapy, for HCMV should ideally be evaluated in a pre-clinical animal model. HCMV is species-specific, making direct study of infection in animal models untenable. Species specific animal CMV crosses the placenta in both rhesus macaque (rhesus cytomegalovirus virus, RhCMV) and guinea pig (guinea pig cytomegalovirus, GPCMV) [2, 3]. The guinea pig is the only small animal model for cCMV and the focus of this paper. Both human and guinea pig placentas are hemomonochorial, containing a homogenous layer of trophoblast cells separating maternal and fetal circulation [4, 5]. The guinea pig gestation period (approximately 65 days) is divided into three trimesters similar to human pregnancy. Furthermore, congenitally infected newborn pups have similar disease symptoms as humans, eg. sensorineural hearing loss (SNHL) [6]. Consequently, the guinea pig is potentially well suited for the testing of intervention and vaccine strategies against cCMV.

The GPCMV genome has been sequenced and the virus encodes conserved genes with HCMV [7, 8]. Importantly, GPCMV encodes functional viral glycoprotein complexes to HCMV (gB, gH/gL/gO, gM/gN and gH-based pentamer), which are important for virus cell entry [9–11]. The GPCMV viral gH based pentamer, or pentamer complex (PC), was demonstrated to be necessary for virus infection of non-fibroblast cells, via an endocytic entry pathway similar to clinical strains of HCMV, virus dissemination in animals and congenital infection [10–12]. In HCMV, the viral glycoprotein complexes are important neutralizing antibody targets and GPCMV glycoprotein complexes are similarly immunogenic [9, 13–15]. Although various vaccine strategies and target antigens have been evaluated against HCMV, the gB glycoprotein remains a significant focus in various vaccine approaches, either as a standalone antigen, or in conjunction with other target antigens [16]. A renewed interest in gB-based HCMV vaccine is also due to novel insight of action of non-neutralizing antibodies against gB that potentially enhances vaccine protection against HCMV [17]. In clinical trials, a subunit gB vaccine attains at best about 50% efficacy against maternal infection [18]. GPCMV gB is a neutralizing target antigen [14, 19] and essential for virus growth on all cell types [9, 10]. In the guinea pig model, the gB antigen has been the most extensively studied vaccine antigen against cCMV. Either as a subunit vaccine approach [19], or expressed in recombinant vaccine vector delivery platforms: replication defective adenovirus (Ad) [20]; replication defective lymphocytic choriomeningitis virus, LCMV [21]; and attenuated vaccinia vector [22]. These studies demonstrated that the gB viral antigen induced antibodies that neutralized virus infection on fibroblast cells. However, in congenital protection studies, the various gB vaccine studies attained approximately 50% efficacy in the guinea pig model [19–22]. In both HCMV and GPCMV, non-fibroblast cells are dependent upon the viral PC for endocytic pathway cell entry unlike fibroblasts cells that express the viral receptor platelet derived growth factor receptor alpha (PDGFRA) to enable entry by direct cell fusion independent of the PC [23, 24]. Therefore, despite the essential nature of gB for all cell types, potentially neutralizing antibodies directed to the PC might constitute a more effective vaccine target [25, 26].

In HCMV studies, the basis for the approximately 50% efficacy of the gB subunit based vaccine [27] has been in part attributed to an inability of antibodies to effectively neutralize virus infection on epithelial cells. Additionally, because the full length gB protein was not used in vaccine studies there was a potential lack of antibodies directed towards a trimeric form of gB found in the virion [28]. Our laboratory has recently generated novel non-fibroblast guinea pig cell lines (renal epithelial and trophoblast cells) [10, 12]. Consequently, we sought to determine if GPCMV gB antibodies had equivalent, or weaker neutralizing capability, on non-fibroblast cells compared to fibroblasts. Both a truncated (lacking transmembrane anchor domain) and a full length gB protein were investigated as vaccine candidates to additionally determine if the previous use of truncated versions of GPCMV gB in this model impacted on the generation of neutralizing antibodies against GPCMV. Candidate gB proteins were expressed in separate recombinant defective Ad vectors. A gB vaccine protection study in animals demonstrated some efficacy against challenge virus but despite high neutralizing antibody titer the vaccine failed to prevent virus dissemination in guinea pigs. Additionally, depletion of complement in vivo had minimal impact on vaccine efficacy.

Materials and Methods

Virus, cells, synthetic genes and oligonucleotides.

Wild type GPCMV (strain 22122, ATCC VR682) was propagated on guinea pig fibroblast lung cells (GPL; ATCC CCL 158), renal epithelial (REPI) or placental trophoblast (TEPI) cell lines as previously described [10, 12]. Additionally, a fibroblast cell line (GPKO) that lacked the GPCMV cell receptor PDFRA was also used in virus neutralization studies [24]. Virus stocks for antibody neutralization assays were generated on renal epithelial cells. Virus titers were determined by GPCMV titration on fibroblast cells [29]. Synthetic GPCMV GP55 codon optimized genes (Genscript) were generated: (1) full length gB (codons 1–901), designated gBwt; (2) truncated gB lacking transmembrane domain (deletion codons 711–757), designated gBTMD. Recombinant defective adenovirus (Ad5) vectors were generated by insertion of gB ORF into the E1 locus of the Ad vector by Welgen Inc (MA) to express either gBwt or gBTMD under HCMV IE enhancer promoter control with a 3’ SV40 polyA sequence. High titer CsCl gradient purified recombinant defective adenovirus virus stocks (10¹² TDU/ml) were propagated by Welgen Inc. (MA) after preliminary characterization of primary virus stock by our laboratory. Recombinant Ad vectors: full length gB Ad vector was designated AdgBwt; truncated gB vector was designated AdgBTMD.

Animal study.

Guinea pig (Hartley) animal studies were performed under IACUC (Texas A&M University) permit 2017–0227. All study procedures were carried out in strict accordance with the recommendations in the “Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.” Animals were observed daily by trained animal care staff, and animals that required care were referred to the attending veterinarian for immediate care or euthanasia. Terminal euthanasia was carried out by lethal CO₂ overdose followed by cervical dislocation in accordance with IACUC protocol and NIH guidelines. Animals purchased from Charles River Laboratories were verified as seronegative for GPCMV by anti-GPCMV ELISA of sera collected by toenail clip bleed as previously described [9]. Animal studies were performed to evaluate 1) immune response to recombinant defective Ad vectors encoding full length (AdgBwt) or truncated gB (AdgBTMD); 2) AdgBwt vaccine protection against virus dissemination.

Two groups of GPCMV seronegative animals (n=3 per group) were vaccinated (SQ, 1 x 10⁸ TDU) on three separate occasion at 4 week intervals. Ad vector vaccine dosage was based on Xing et al., [30]. Group 1 received AdgBwt and group 2 received AdgBTMD. At 3 weeks post last vaccination, animals were bled and serum from individual animals within each group was pooled for further study. Pooled anti-GPCMV sera was previously described [13] and generated by SQ infection of guinea pigs with wild type GPCMV.

In a vaccine protection study, two groups of seronegative animals (n = 12 per group) were used. Animals in Group 1 received 3 inoculations of AdgBWT vaccine 4 weeks apart and anti-gB titer was determined 4 weeks after the last boost. A second group of seronegative animals (Group 2, n=12) served as a non-vaccine control. Animals in both groups were challenged with wtGPCMV (10⁵ pfu, SQ). At days 4, 8, 12 and 27 dpi, 3 animals per group were euthanized to evaluate viral load in target organs and blood via DNA extraction from tissue and blood and real time PCR assay as previously described [13, 24].

An additional group of animals (Group 3, n=9) were vaccinated with AdgBwt as previously described and verified for gB seroconversion. Prior to challenge with wild type virus, animals were treated with 100 μg/kg of recombinant cobra venom factor (CVF) (Millipore Sigma) [31, 32]. Recombinant CVF was delivered by i/p injection and treatment started two days prior to challenge (D-2) and given again at 1 day before virus challenge (D-1). At day 0, animals were challenged with wild type virus (10^5 pfu/ SQ). Subsequently, animals were given additional 100μg/kg CVF from 1 day post virus challenge and every other day until tissues were harvested. Prior to CVF treatment, serum was collected from all animals in the group and also at each tissue harvest time point to test for presence of complement C3. Serum collected on 0, 4, and 8 days post infection (dpi) had only background levels of C3 (Table 1). Levels spiked back to nearly normal concentrations by day 12 (Table 1), presumably as a result of an immune response against CVF. An evaluation of viral load in target organs and blood of group 3 animals was carried out at days 4, 8 and 12 post infection. Three animals per group were euthanized to evaluate viral load in target organs and blood via DNA extraction from tissue and blood and real time PCR assay as previously described [13, 24].

Table 1.

Guinea pig C3 in sera evaluated by ELISA for complement level in AdgBwt seropositive animals treated with cobra venom factor.

	D-2	D0	D4	D8	D12
OD	2.65	0.14	0.13	0.14	2.42
ng/ul	80.34	1.15	1.12	1.16	53.46

^*FBS negative control OD = 0; C3 ELISA kit level of detection 1.38ng/ml

GPCMV Glycoprotein anti-gB ELISA.

Specific anti-gB glycoprotein complex ELISA was carried out as previously described using positive coating antigen derived from renal epithelial cell monolayers transduced with recombinant replication defective adenovirus (Ad) vectors expressing AdgBwt or recombinant Ad vector expressing GFP for negative coating antigen [9, 10, 13]. Harvested cells were washed with PBS and cell pellets fixed prior to processing as coating antigen. Protein concentration was normalized by Bradford assay. MaxiSorp ELISA plates (NUNC) were coated with 0.5μg of either gB Ag+ or Ag− preparations diluted in carbonate coating buffer overnight at 4°C, washed in 1X PBST then blocked with 2% nonfat dry milk. Test sera were diluted in blocking buffer from 1:80 to 1:20480 in doubling dilutions, incubated for 2 hours at 37°C and then reacted with anti-Guinea Pig IgG peroxidase antibody (Sigma) diluted (1:1000) in blocking buffer for an additional 1 hour at 37°C before reacting with TMB membrane peroxidase substrate (KPL). Net OD (absorbance 450nm) was attained by subtracting OD of Ag− from OD of Ag+. All anti-gB ELISAs described in this report were carried out with the same batch of coating antigen. Described approach is based on similar strategies for glycoprotein complex expression for HCMV and RhCMV and ELISAs [33, 34]. All ELISAs were run minimum of three times in duplicates. ELISA reactivity was considered positive if the net OD was greater than, or equal to 0.2, as determined by GPCMV negative serum.

Guinea pig complement C3 ELISA.

Complement C3 level in guinea pig sera was determined using the Complement C3 Guinea Pig ELISA kit (Abcam) according to manufacturer’s instructions. Briefly, serum samples collected at different time points before and after CVF treatment during the pathogenicity study were diluted 1:10,000 before adding 100 μl of the diluted sera to the precoated wells for 10 minute incubation at room temperature. All wells were washed with provided wash buffer four times before 100 μl of 1X enzyme-antibody conjugate was added and incubated for 10 minutes at room temperature. Wells were washed four times with wash buffer before 100 μl TMB substrate solution was added to each well for 5 minutes then reaction was stopped by adding 100 μl of stop solution before absorbance (450nm) was determined. To determine the concentration of complement C3 present in each time point serum sample, a standard curve with provided guinea pig complement C3 calibrator was generated according to protocol. A two-fold dilution starting at 200 ng/ml of control was prepared to 3.125 ng/ml. A logarithmic graph was plotted using known concentration versus OD. Fetal bovine serum was used as negative control which gave an OD at 450nm reading of 0.0 abs. The assay sensitivity was 1.3083 ng/ml. All samples and standards were carried out in duplicates.

GPCMV Neutralization assays.

GPCMV neutralization assays (NA₅₀) were performed on GPL fibroblasts, GPKO fibroblasts, renal epithelial and trophoblast cells with PC+ GPCMV virus stocks generated on renal epithelial cells [9, 10] using pooled sera from a specific group as previously described [13]. Serially diluted sera were incubated with approximately 1 x10⁵ pfu PC+ GPCMV in media containing 1% rabbit complement (Equitech Bio) for 90 minutes at 37°C before infecting GPKO, REPI or TEPI cells for 1 hour. For neutralization on GPL cells, 1 x 10³ pfu PC+ GPCMV was used. Infected cells and supernatant were collected on day 4 then titrated on GPLs. Final neutralizing antibody titer was the inverse of the highest dilution producing 50% or greater reduction in plaques compared to virus only control. NA₅₀ were performed from each sample three times concurrently with the same virus stocks between groups.

Western blot.

Western blots were carried out on denatured cell lysates separated by 10% SDS-PAGE gels (BioRad) following previously described protocol [9]. For non-reducing gel analysis of recombinant Ad vector expressed gB, cell monolayers from separate 6 well plates were washed in PBS and lysed in Laemmli buffer in the absence of β-Mercaptoethanol. Samples were heated to 37°C for 5 minutes prior to loading onto 3–8% gradient Tris-Acetate Nu-PAGE gel (Invitrogen) with Native-PAGE (Invitrogen) running buffer in absence of SDS. Protein size determined by HiMark pre-stained protein standard ladder (Life Technologies). Detection of gB protein was carried out with primary mouse monoclonal antibody to gB (a gift from Dr. Britt, UAB) [14].

Real time PCR.

Blood and tissues (lung, liver, spleen) were collected from euthanized guinea pigs to determine the viral load as previously described [9, 13]. Briefly, for tissue DNA extraction, FastPrep 24 (MP Biomedical) was used to homogenize tissues as a 20% weight/volume homogenate in Lysing Matrix D (MP Biomedicals). To obtain DNA from whole blood, blood was collected into tubes containing ACD anticoagulant and 200μl of blood was subsequently used per extraction. DNA was extracted using the QIAcube (Qiagen) according to manufacturer’s liquid (blood) or tissue protocol instructions. Viral load was determined by real time PCR on LightCycler 480 (Roche Applied Science) using primers and hydrolysis probe to amplify a product from the GPCMV GP44 gene. PCR master mix contained LightCycler ProbesMaster (Roche Life Science), 0.4 μM primers and 0.1 μM probe, 0.4U uracil N-glycosylase (UNG) in 25μl total reaction volume including 10 μl of DNA per reaction. Standard controls and no template controls (NTC) were run with each assay for quantification. Lightcycler480 amplification parameters were: UNG step for 10minutes at 40°C followed by activation at 95°C for 10minutes, then 45 cycles of denaturation at 95°C for 15s, annealing at 56°C for 15s, elongation at 72°C for 10s. Data was collected by ‘single’ acquisition during the extension step. Standard curve was generated using GPCMV GP44 plasmid (McGregor 2011 virology) for quantification and assay sensitivity. The sensitivity of the assay was determined to be 5 copies/reaction. Viral load was expressed as copy number/ml of blood or copy number/mg tissue. Results calculated were a mean value of triplicate PCR runs per sample.

Results

Virus tropism is dependent upon the cell type used to generate virus stock

GPCMV encodes a gH-based pentamer complex, gH/gL/GP129/GP131/GP133, [10] and this complex is necessary for virus infection of non-fibroblast cells including epithelial, trophoblast and macrophage [10, 12, 35]. Consequently, viral mutants lacking all, or part of the unique PC genes (GP129, GP131 and GP133), are unable to replicate on non-fibroblast cells. However, propagation of GPCMV virus stock on fibroblast cells as opposed to epithelial cells alters the ability of the PC to be incorporated into the virion by an undetermined process similar to HCMV [36]. Virus stock generated on epithelial cells easily infects all cell types but GPCMV stock generated on fibroblast cells is highly restricted and only efficiently infects fibroblast cells with low level infection of non-fibroblast cells. This is demonstrated in Figure 1, where non-recombinant wild type GPCMV virus stock, generated on renal epithelial cells, can infect fibroblasts, epithelial and trophoblast cells with similar ability based on progeny virus titrations post infection. In contrast, epithelial virus stock subsequently passaged once on fibroblast cells and new fibroblast based virus stock generated can only efficiently infect fibroblast cells (Figure 1) despite similar multiplicity of infection (1 pfu/cell) and viral genome encoding the PC. Results were based on titration of progeny virus from infected cell monolayers at 4 days post infection with virus titration carried out on fibroblast cells. Thus, in the context of infection, similar input virus from different stocks generated on different cell types have potentially altered outcome for virus infection/growth on various cell types. Infection of PDGFRA gene knockout fibroblasts (GPKO) is similarly affected as it is dependent upon the PC endocytic route of entry [24]. We concluded that stock virus cell tropism range is an important factor when evaluating antibody neutralizing capability. Therefore as part of a standardization of GPCMV studies, we suggest that vaccine antibody neutralization assays should be performed with virus stock with tropism range for both epithelial and fibroblast cells. This would be more representative of a clinical strain in the host and consequently have greater translational relevance. We have adopted the use of epithelial cell generated virus stock in this current study and in other on-going GPCMV vaccine studies.

Figure 1. Wild-type virus stock generated on epithelial cells infects all cell types unlike GPCMV stock grown on fibroblast cells.

Virus stock was designated based on cell line used to generate stock renal epithelial (REPI) for REPI stock and fibroblast (GPL) for GPL virus stock. Each virus stock was used to infect various guinea pig cell lines in 6 well plates including: (1) renal epithelial cells (REPI, red); (2) trophoblasts (TEPI, yellow); (3) fibroblasts (GPL, blue). MOI = 1 pfu/cell. Samples harvested 4 days post infection for titration of progeny virus on fibroblast cells. Viral titers are represented as mean +/− SD.

Recombinant AdgB vector expression and immune response in animals

Analysis of the predicted GPCMV gB protein sequence from the GPCMV genome, strain 22122, [9, 37] had previously identified that the protein had a leader peptide signal sequence and a number of glycosylation sites. Further analysis with an online program for prediction of transmembrane helices in proteins (TMHMM Server v. 2.0, http://www.cbs.dtu.dk/services/TMHMM/) identified a C-terminal transmembrane domain, aa 711–757, (Figure 2A) similar in length and location to that of HCMV gB (https://www.uniprot.org/uniprot/P06473). Two synthetic GPCMV GP55 genes encoding the gB ORF were generated (Genscript): (1) full length gB, gBwt, (codons 1–901); (2) gB lacking transmembrane anchor domain, gBTMD, (deletion of codons 711–757). Recombinant replication defective adenovirus (Ad) vectors encoding either gBwt (AdgBwt) or gBTMD (AdgBTMD) were subsequently created as previously described for other recombinant Ad vectors [10]. Verification of gB expression by recombinant Ad vectors was demonstrated by Ad vector transduction of fibroblast cells and western blot of transduced cell lysates with mouse monoclonal antibody to GPCMV gB [14], which identified a protein of approximately 150 kDa in size (Figure 2B). The predicted size of full length GPCMV gB was approximately 102 kDa [9] and so the size was larger than predicted. However, GPCMV gB is predicted to encode a number of glycosylation sites (32 O-linked and 15 N-linked) [9]. This would likely increase the size of the gB glycoprotein as observed for other glycosylated GPCMV glycoproteins such as gO and gH [9]. Original GPCMV gB studies had indicated that full length protein would be approximately 150 kDa in size and if non-glycosylated the size was 102 kDa [14]. The TMD region of gB is not predicted to encode any glycosylation sites, which likely accounts for inability to easily detect a difference in size between gBwt and gBTMD. However, a TMD truncated GPCMV gB expressed in a recombinant LCMV gB vector also had a molecular size of approximately 150 kDa [21].

Figure 2. Analysis of the GPCMV gB protein for transmembrane domain and expression of full length and truncated gB by recombinant Ad vectors.

(A) Predicted GPCMV gB protein sequence was analysed by TMHMM Server v. 2.0, http://www.cbs.dtu.dk/services/TMHMM/ for location of a potential transmembrane domain. Results show identified domain region based on normal analysis parameters and highlighted in red. (B) Western blot of gB protein in Ad transduced cells. Expression of full length gB (gBwt) or truncated protein (gBTMD) by recombinant Ad vector infected cell monolayers (MOI=50 TDU/cell). At 24 h post infection, total cell lysates of transduced GPL cells were analyzed by 10% SDS-PAGE and western blot for detection of gB by anti-gB mouse monoclonal antibody and secondary anti-mouse IgG/HRP conjugate as described in materials and methods. Lanes: (1) mock infected GPL cell lysate; (2) AdgBwt infected GPL; (3) loading dye; (4) AdgBTMD infected GPL. (C) Non-reducing western blot analysis. GPL cells tranduced with AdgBwt or AdgBTMD were harvested 24 hpi and run on 3–8% Tris-Acetate gels under Native-PAGE conditions and detected for gB by anti-gB mouse monoclonal antibody. Lanes: (1) mock infected GPL; (2) AdgBwt infected GPL; (3) AdgBTMD infected GPL. (D) Western blot assay for detection of soluble recombinant gB protein released into tissue culture by AdgBTMD or AdgBwt infected cells. GPL cells in 6 well plates were transduced (100 TDU/cell) with AdgBwt or AdgBTMD in separate wells for approximately 48 hours. Tissue culture supernatants of Ad transduced cells or mock infected wells were collected and processed as described in materials & methods. Corresponding cell monolayers of Ad transduced and mock infected wells of a 6 well dish were also harvested for analysis. Samples were analyzed by 10% SDS-PAGE and western blot for detection of gB by anti-gB mouse monoclonal antibody. Blot lanes: (1) mock infected GPL cell lysate; (2) mock infected tissue culture supernatant; (3) AdgBTMD infected GPL cell lysate; (4) AdgBTMD tissue culture supernatant; (5) loading dye; (6) AdgBwt infected GPL cell lysate; (7) AdgBwt tissue culture supernatant.

Since GPCMV gB has the potential to form a multimeric triplex complex similar to full length HCMV gB [28]. The ability of both full length and truncated recombinant GPCMV gB proteins to multimerize was evaluated. Western blot analysis of AdgB transduced cell lysate, separated by 3–8% non-reducing tris-acetate PAGE demonstrated that gBwt was capable of forming a triplex with protein complex of approximately 450 kDa was detected as well as 150 kDa monomer (Figure 2C). In contrast, the gBTMD protein would appear to form mainly gB dimers (300 kDa) as well as monomer (Figure 2C). Both gB proteins had the potential to multimerize as they encoded the oligomerization AD-1 domain homolog region (amino acids 547–620) [38]. Potentially, gBTMD was more immunogenic than gBwt as it was expected to be secreted from the cell because of a lack of a transmembrane anchor. The release of soluble recombinant gB protein in vitro by AdgBTMD, or AdgBwt transduced fibroblast cells was accessed by western blot of cell lysate, or culture media supernatant, of transduced cell monolayers (Figure 2D). Glycoprotein B was identified in the supernatant and cell lysate of AdgBTMD transduced fibroblast cells. The full length gBwt was detected in the cell lysate at similar level to AdgBTMD but was not detected in the culture media supernatant of AdgBwt transduced cells. We concluded that gBTMD was more soluble and readily secreted from the cell unlike gBwt.

Next, female guinea pigs seronegative for GPCMV were randomly assigned to AdgBwt or AdgBTMD groups and inoculated subcutaneously three times with a specific AdgB vector once every 4 weeks (SQ, 1x10⁸ TDU/inoculation). At 21 days post last inoculation, animals were bled and sera within each group pooled for further study. Based on gB ELISA results, both gBwt and gBTMD were equally immunogenic and evoked similar high titer anti-gB ELISA results (Figure 3A). However, virus neutralization assays of both gB antisera demonstrated that antibodies to the full length gB construct was significantly more effective at neutralizing virus infection on fibroblast and epithelial cells compared to antibodies against gBTMD (Figure 3B). On fibroblast cells, the NA₅₀ titer (1:1707) for gBwt was almost 3 times that of the neutralizing titer for gBTMD (1:640). Neutralizing capability was lower on non-fibroblast cells but antibodies to the full length gB was approximately 2-fold more effective on renal epithelial (REPI) cells (NA₅₀ 1:640) compared to gBTMD antibodies (NA₅₀ 1:320). On trophoblast cells, the gB antibodies had further reduced neutralizing NA₅₀ titers (1:320 for gBwt and 1:240 for gBTMD) compared to that of renal cells. However, unlike NA₅₀ titers on fibroblast and epithelial cells, there was no statistical difference between gBwt and gBTMD NA₅₀ titers. Overall, results indicate that despite similar anti-gB ELISA titers for AdgBwt and AdgBTMD, the antibody response to the full length gB had higher neutralizing titer on fibroblast and renal epithelial cells compared to anti-gBTMD. However, gB antibodies for gBwt appear to have most virus neutralizing activity on fibroblast cells. On non-fibroblast cells, the gBwt antibody neutralizing titer is approximately 3 (renal epithelial) to 5-fold lower (trophoblasts) compared to normal fibroblasts. Neutralization assays were all carried out in the presence of complement as it had previously been demonstrated that antibody neutralization of GPCMV was complement dependent and is a standard approach in most GPCMV studies [14]. However, a comparative set of neutralization assays were carried out in the presence and absence of complement for gB antibodies (AdgBTMD). Neutralization assay carried out in the absence of complement reduced NA₅₀ on fibroblasts from 1:640 (+complement) to 1:160 (no complement). This confirmed the dependency on complement for maximal virus neutralization by gB antibodies.

Figure 3. AdgBwt and AdgBTMD vaccines induce similar anti-gB titers but with different virus neutralization capability.

Animals were inoculated four time approximately 4 weeks part with either AdgBwt (n=3) or AdgBTMD (n=3). Sera collected 4 weeks post final vaccination from each group were pooled and assayed for anti-gB by ELISA or its neutralizing capability measured on different cell types. (A) Anti-gB ELISA titer was compared between pooled sera collected from animals inoculated with AdgBwt (blue) and AdgBTMD (orange). Student t-test showed no significant difference (ns) between the two groups. (B) Neutralization of wtGPCMV PC+ from sera collected from animals inoculated with AdgBwt (blue) or AdgBTMD (orange) were tested on fibroblasts (GLP), epithelial (REPI) or trophoblast (TEPI) cells. Student t-test * P < 0.05; ns = not significant.

In order to demonstrate that anti-gB sera blocked cell entry, immunohistochemical staining experiment was performed to detect viral IE2 immediate early antigen expression in virus infected cells with GPCMV pretreated with anti-gB sera, or control seronegative sera. Pooled AdgBwt sera at a final dilution of 1:40 was used to neutralize GPCMV PC+ virus on fibroblast (GPL) and epithelial (REPI) cells. Anti-gB antibodies at a dilution of 1:40 completely blocked virus infection of both GPL (Figure 4iA) and REPI (Figure 4iiA) cells. However, GPCMV negative control serum failed to block virus with positive IE2 staining present in GPL (Figure 4iB) and REPI cells (Figure 4iiB). No IE2 staining was detected in control mock infected GPL (Figure 4iC), or REPI cells (Figure 4iiC). Secondary antibody only staining in both cell types regardless of presence, or absence of virus did not detect any IE2 antigen (Figure 4iD–E and 4iiD–E). We concluded that anti-gB antibodies had the capacity to block cell entry for both fibroblast and epithelial cells but neutralizing titers differed between cell types.

Figure 4. Assay for GPCMV IE2 antigen cellular expression demonstrates that anti-gBwt antibodies prevents GPCMV infection on both GPL fibroblast and REPI epithelial cells.

Pooled sera (1:40 dilution) from AdgBwt inoculated animals was used to neutralize GPCMV infection on GPL (i) and REPI (ii) cells. GPCMV infected monolayers A, B & D (MOI=0.5 pfu/ml) Viral antigen detected by immunohistochemical staining for IE2 viral antigen with primary rabbit anti-GPCMV IE2 antibody and biotinylated goat anti-rabbit IgG secondary as described in materials and methods. Cell monolayers shown in panels stained for IE2: (A) AdgBwt serum treated GPCMV infected cells; (B) Control negative sera treated GPCMV infected cells; (C) Mock infected cells; (D) Control GPCMV infected cells stained with secondary antibody only; (E) Control mock infected cells stained with secondary antibody only. Cells fixed at >48hpi. Monolayers were visualized under bright-field at 20X magnification.

Loss of CMV cell receptor PDGFRA impacts on the ability of anti-gB to effectively neutralize GPCMV cell infection

We recently identified guinea pig PDGFRA, via gene CRISPR knockout strategy, as the cell receptor for GPCMV infection of fibroblast cells and that this protein was absent in other cell lines that are dependent upon the PC for the endocytic pathway of cell entry [24]. Hence, unlike other cell types, fibroblast cells have two routes of virus cell entry and knockout of PDGFRA in guinea pig fibroblasts (GPKO cells) restricted GPCMV cell entry to the endocytic PC dependent pathway rather than by direct cell membrane fusion that occurs in PDGFRA positive cells. Although gB is required for entry into all cell types, the restricted route of entry on the GPKO cells potentially altered the neutralizing titer of anti-gB on these fibroblast cells compared to PDGFRA positive fibroblasts (GPL cells). This was illustrated in a side by side comparison of GPCMV neutralization on GPL and GPKO cells using anti-gBwt sera. The NA₅₀ results in Figure 5 demonstrated that anti-gB NA₅₀ for GPKO cells (1:480) is more than 3 times lower than that of GPL fibroblasts (1:1707) and the value is more similar to renal epithelial cells, or trophoblasts (1:640 and 1:320 respectively, Figure 3). These results suggest that evaluation of vaccine efficacy by NA₅₀ assays should consider alternate pathways of virus entry. Therefore, for fibroblasts, which uniquely has two pathways of CMV entry, the use of both PDGFRA+ and PDGFRA knockout cells should be routinely used in neutralization studies to separately evaluate the two pathways of entry into cells as PC+ virus growth kinetics are unaffected by the loss of PDGFRA receptor on fibroblasts [24].

Figure 5. Presence or absence of cell receptor PDGFRA impacts on the ability of anti-gB to neutralize GPCMV infection.

Comparative neutralization of wtGPCMV PC+ virus by pooled sera collected from animals inoculated with AdgBwt (n = 3) evaluated on fibroblasts (GLP, black) or PDGFRA knockout GPL (GPKO, diagonal lines) cells. Student t-test * P = 0.02.

Recombinant gB vaccine protection against GPCMV challenge in guinea pigs

Since the full length gB antigen appeared to provide better NA₅₀ values compared to gBTMD, only the AdgBwt was evaluated in a vaccine protection study. GPCMV seronegative animals (Group 1, n=12) were vaccinated (3 dose SQ, 1x10⁸ TDU/injection/animal/every 4 weeks), confirmed for gB seroconversion by ELISA and subsequently challenged with wild type GPCMV at 4 weeks post last vaccination. A control group of seronegative animals (Group 2, n=12) were also challenged with wild type GPCMV. Both groups received 1x10⁵ pfu of GPCMV, SQ and at 4, 8, 12 and 27 days post infection (dpi) animals from each group (3/group/time point) were evaluated for viral load in target organs (liver, lung and spleen) and blood as described in materials and methods. At 27 dpi, viral load in salivary gland tissues was additionally evaluated for each group. The viral loads for tissues and blood from each group are shown in Figure 6A–D. The gB vaccine group 1 animals had more limited viral load in target organs and viremia compared to the control group 2 animals (Figure 6C–D). In vaccinated animals, GPCMV failed to be detected in organs and blood at 12 and 27 dpi (Figure 6A–B). The control no vaccine group 2 animals had dissemination of virus to all target tissues and viremia with detectable virus in salivary gland at 27 dpi (Figure 6C–D). Unlike the control group, vaccinated animals lacked detectable virus at 27 dpi in salivary gland tissue (Figure 6A and 6C).

Figure 6. Dissemination of GPCMV to target organs in gBwt seropositive and seronegative animals.

(A-B) AdgBwt seropositive animals and (C-D) seronegative animals (n = 12 per group) were infected with wtGPCMV (10⁵ pfu). At various days (4, 8, 12 and 27 days post infect infection, DPI), 3 animals per group were evaluated for viral load in target organs by real time PCR of tissue extracted DNA (A, C). Viral load plotted as viral genome copies/mg tissue. Salivary gland tissue was only evaluated at day 27. Blood viremia at 4, 8, 12 and 27 DPI plotted as genome copies/ml blood (B, D). (E-F) Detection of GPCMV to target organs and blood in gBwt seropositive animals depleted of complement. At days 4, 8, and 12 DPI, 3 animals per group were evaluated for viral load in target organs by real time PCR of tissue extracted DNA (E). Viral load plotted as viral genome copies/mg tissue. Blood viremia at 4, 8, and 12 DPI plotted as genome copies/ ml blood (F). Target organs: lung (blue); liver (orange); spleen (gray); blood (black).

Recent studies with anti-gB subunit vaccine sera suggested that non-neutralizing antibodies to gB enhanced vaccine efficacy but these antibodies were highly dependent upon complement. In order to evaluate the effect of complement on GPCMV anti-gB vaccine in vivo, we temporally depleted complement from AdgBwt vaccinated guinea pigs prior to challenge with GPCMV. C3 depletion in guinea pigs was carried out by the use of recombinant CVF, which mimics activated C3. This strategy has been used to successfully deplete complement in guinea pigs in other disease models [31, 32]. Unlike C3 non-depleted AdgBwt animals, virus was detectable in the lung at 4 and 8 dpi at a significantly higher level in CVF treated animals (Figure 6E). There were no detectable levels of virus in target organs at day 12 in both groups 1 and 3. Viremia was detected only on day 4 blood samples (Figure 6F) compared to 4 and 8 dpi for the non-CVF treated group 1 gB vaccinated animals. Based on C3 values pre and post CVF treatment (Table 1), depletion of C3 complement was relatively effective. However, base levels may have been sufficient for function and we concluded that depletion of C3 complement had minimal effect on gB vaccine efficacy. Overall, the gB vaccine was highly effective in reducing virus dissemination and viremia in the animal model.

DISCUSSION

HCMV is a leading cause of congenital disease, which can result in neurodevelopmental and hearing problems in surviving newborn babies. Given the added complexity of testing vaccine strategies against cCMV in human clinical trials, the use of preclinical animal models for evaluation of potential vaccine candidates against cCMV is an important preliminary approach. What constitutes an effective immune response against HCMV is only partially understood as patients convalescent for HCMV mount both antibody and cell mediated responses to various target antigens. Potentially, in the context of protection against cCMV, the antibody response might be of greater significance. The viral glycoproteins are important neutralizing target antigens. Antibodies to HCMV gB can constitute 40–70% total neutralizing activity to HCMV in convalescent patients [39, 40]. The most extensively studied gB vaccine is the Chiron HCMV gB subunit vaccine, which is based on a gB lacking a furin cleavage site and the C-terminal transmembrane anchor domain [18]. Although this gB vaccine produced 50% efficacy in phase II clinical trials, it could be argued that a failure of this vaccine was because the recombinant protein lacked a furin cleavage site as well as transmembrane domain, which resulted in an impaired immune response to a protein dissimilar to that present in HCMV virions. In this report, we evaluated neutralizing antibody response to GPCMV gB in the guinea pig model using recombinant defective Ad vectors encoding full length gB, or gB lacking the transmembrane anchor domain. In both cases, the gB furin cleavage site was intact. In a preliminary study, we tested a full length GPCMV gB lacking a furin cleavage site and found the expressed protein was poorly immunogenic compared to other Ad vector expressed proteins (McGregor and Choi, unpublished data). Although the full length AdgBwt and truncated AdgBTMD induced similar anti-gB ELISA antibody titers, the sera collected from AdgBwt animals were superior at neutralizing wt GPCMV not only on fibroblast cells but also non-fibroblast cells, epithelial and trophoblasts (Figure 3). In AdgBwt vaccinated animals, GPCMV challenge virus had more limited dissemination to target organs compared to non-vaccine control group.

Prior GPCMV single antigen vaccine studies employed gB lacking both transmembrane anchor domain and C-terminal tail domain as part of a subunit vaccine, recombinant defective adenovirus, or Vaccinia vector [19, 20, 22]. In recombinant defective LCMV vector vaccine studies, two slightly different GPCMV gB ORFs were used: (1) gB lacking the transmembrane domain but intact for the C-terminal tail [21]; (2) gB with transmembrane domain but deleted for C-terminal tail [41]. However, no GPCMV gB strategy to date has employed full length gB in vaccine studies, except this current report, or vaccines based on attenuated GPCMV vaccine strains. Unfortunately, except for this paper, GPCMV gB vaccine studies have not been tested for neutralizing antibody capability on non-fibroblast cells, which limits comparisons to other studies. However, our results with AdgBTMD sera would suggest that other published GPCMV gB vaccines would be weakly effective on these cell types and also fibroblasts lacking PDGFRA. HCMV gB has 5 antigenic domains (AD1–5). In GPCMV gB, the homolog AD-1 (aa 547–620) was encoded in all the various GPCMV gB vaccines tested to date and represents the dominant antigenic domain in HCMV gB for neutralizing antibody response [38]. The HCMV gB AD3 domain is in the C-terminus tail of the protein and potentially contributes to immune response against gB [42]. Although an AD3 domain has not been identified in GPCMV gB, an equivalent region would only be expressed by full length gB described in this report and potentially in a partial truncated GPCMV gB expressed in a LCMV vaccine [21]. The use of a candidate full length HCMV gB vaccine in mice identified higher order gB trimer antigens which might be similar to that present on virions [28]. Similar higher order antigens related to multimeric protein would also potentially exist in GPCMV gB but would only be present in the full length gB described in this report as demonstrated in Figure 2C. Potentially, the additional target epitopes within the full length gB may be the basis for a higher neutralizing titer for AdgBwt compared to AdgBTMD in this report and previously described GPCMV gB candidate vaccines. Full length HCMV gB evoked antibodies in mice that had better neutralization titers on both fibroblast and epithelial cells compared to truncated gB [28]. Similarly, the ability of full length GPCMV gB encoded by AdgBwt might also be the basis for AdgBwt sera having slightly better neutralizing activity on renal epithelial cells compared to AdgBTMD. Curiously, the anti-gB neutralizing titer was not improved on trophoblasts compared to the response for the truncated GPCMV gB. Consequently, it would be of potential interest to evaluate HCMV anti-gB on trophoblast cells. The neutralizing antibody titers evoked by the recombinant AdgB vectors in this report are substantially higher than that reported for previous GPCMV gB vaccine strategies. The gB ORFs used in our studies were synthetic genes that were optimized for efficient translation unlike previous GPCMV gB studies. However, it is most likely that the severe truncations found on other GPCMV gB candidates (most terminate at 692 aa) impaired immunogenicity and possibly protein stability. Based on higher neutralizing antibody titers, it is likely that the full length gB encoded by the AdgBwt vaccine would have better efficacy against cCMV than gB vaccine strategies previously studied and consequently would be worthy of further evaluation. More especially, since in HCMV, gB is a CD8 and CD4 cytolytic T cell target [43] and Ad based vector vaccine strategies are capable of inducing T cell responses [44]. Consequently, vaccine efficacy could also be potentially enhanced by T cell response to gB in the guinea pig model but this remains to be determined and will require development of GPCMV gB specific T cell assays. An additional factor that could also potentially contribute to the enhancement of the gB vaccine efficacy are non-neutralizing antibodies [17]. However, this aspect was not evaluated in the present study, or previous gB vaccine strategies. Since all previous GPCMV gB vaccines tested had relatively poor neutralizing antibody capability, it is likely that non-neutralizing antibodies also contributed to vaccine efficacy but this aspect awaits further study. Partial depletion of complement in animals with potent neutralizing anti-gB antibodies, as described in this paper, would seem to have minimal impact in protection in the animal model based on our studies. The lack of a C3 knockout animal prevents comprehensive evaluation long term in the guinea pig model.

Despite the current focus on the CMV pentamer complex as a basis for future neutralizing antibody CMV vaccines and endocytic pathway of cell entry in non-fibroblast cells, the gB protein remains an important viral antigen because of the essential requirement for infection of all cell types. Potentially, gB should be considered a corner stone of any candidate vaccine strategy against cCMV. Our studies with a full length gB demonstrate the importance of the protective immune response against gB. Theoretically, inclusion of gB and viral pentamer in future vaccine strategies will enhance the neutralizing antibody capability of candidate vaccines on various cell types, efficacy in animal models and ultimately successful approach in clinical trials. Antibodies directed to the PC are noted to have a broad neutralizing range across a number of HCMV strains [26] and effective against trophoblast infection [45], which are prerequisites of an effective vaccine against cCMV. The characterization of the various GPCMV viral glycoprotein complexes and availability of novel non-fibroblast cell lines increases the translational relevance of guinea pig based CMV vaccine studies. A potential criticism of the current model is the focus only on the single 22122 strain of GPCMV, which sharply contrasts with HCMV studies. Overall, the flexibility and high throughput capability of the guinea pig model potentially enables future characterization of the most effective vaccine strategy and candidate antigens. Therefore more extensive studies on cCMV in the guinea pig model could provide a shorter route to a successful vaccine against cCMV.