A trimeric capable gB CMV vaccine provides limited protection against a highly cell associated and epithelial tropic strain of cytomegalovirus in guinea pigs

Abstract

Multiple strains of human cytomegalovirus (HCMV) can cause congenital cytomegalovirus (cCMV) by primary or secondary infection. The viral gB glycoprotein is a leading vaccine candidate, essential for infection of all cell-types, and immunodominant antibody target. Guinea pig cytomegalovirus (GPCMV) is the only small animal model for cCMV. Various gB vaccines have shown efficacy but studies have utilized truncated gB and protection against prototype strain 22122 with preferential tropism to fibroblasts despite encoding a gH-based pentamer complex for non-fibroblast infection. A highly cell-associated novel strain of GPCMV (TAMYC) with 99 % identity in gB sequence to 22122 exhibited preferred tropism to epithelial cells. An adenovirus vaccine encoding full-length gB (AdgB) was highly immunogenic and partially protected against 22122 strain challenge in vaccinated animals but not when challenged with TAMYC strain. GPCMV studies with AdgB vaccine sera on numerous cell-types demonstrated impaired neutralization (NA₅₀) compared to fibroblasts. GPCMV-convalescent sera including pentamer complex antibodies increased virus neutralization on non-fibroblasts and anti-gB depletion from GPCMV-convalescent sera had minimal impact on epithelial cell neutralization. GPCMV(PC+) 22122-convalescent animals challenged with TAMYC exhibited higher protection compared to AdgB vaccine. Overall, results suggest that antibody response to both gB and PC are important components of a GPCMV vaccine.

Data Summary

The authors confirm all supporting data, code and protocols have been provided within the article or through supplementary data files.

Impact Statement.

Human cytomegalovirus (HCMV) is a leading cause of congenital disease. Multiple strains of HCMV can cause congenital CMV (cCMV) by either primary or secondary infection. The guinea pig is the only small animal model for cCMV but studies are limited to a single strain of guinea pig cytomegalovirus (GPCMV), prototype 22122, isolated more than 50 years ago. Among leading CMV vaccine candidates are those based on the immunodominant and essential gB viral glycoprotein. In both HCMV and GPCMV, an additional gH-based viral pentamer complex (PC) is essential for infection of non-fibroblast cells and a neutralizing antibody target. Importantly, the PC is necessary for high-level cCMV in the guinea pig model. A full-length gB vaccine strategy induces a potent antibody response to gB trimer in the virion and is highly effective at neutralizing virus on fibroblasts but more limited on non-fibroblasts where antibodies to the PC enhance virus neutralization. In animals, a gB vaccine (AdgB) has efficacy against 22122 strain GPCMV, which produces a high level of cell-release virus. However, AdgB vaccine is more limited against animal challenge by a novel strain of GPCMV (TAMYC) that is highly cell-associated and preferentially tropic to epithelial cells. In contrast, animals convalescent to GPCMV (22122) with additional antibodies to the PC demonstrate higher protection against TAMYC strain challenge. Results demonstrate the limitation of a gB only based vaccine strategy compared to inclusion of PC and suggest a robust antibody response to both gB and PC antigens are cornerstones for CMV vaccine efficacy and cross-strain protection.

Introduction

Placental transmission of human cytomegalovirus (HCMV) is a leading cause of congenital disease, resulting in cognitive impairment and hearing loss in newborns [1]. Although a vaccine against congenital CMV (cCMV) is a high priority, this is an elusive goal, which is complicated due to reactivation of maternal latent infection or by re-infection with the ability of cCMV to occur either by primary infection during pregnancy or secondary viral infection by a new strain in convalescent individuals [2]. HCMV is species-specific, making direct study of infection, or evaluation of intervention strategies 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) [3, 4]. The guinea pig is the only small animal model for cCMV and the focus of this paper. Both human and guinea pig placentas are haemomonochorial, with a homogenous layer of trophoblast cells separating maternal and foetal circulation [5, 6]. Importantly, congenitally infected newborn pups have similar disease symptoms as humans, e.g. sensorineural hearing loss (SNHL) [7]. Consequently, the guinea pig is potentially well suited for evaluation of intervention strategies against cCMV.

GPCMV encodes functional viral glycoprotein complexes similar to HCMV (gB, gH/gL/gO, gM/gN and gH-based pentamer of gH/gL/GP129/GP131/GP133), which are important for virus cell entry [8–10]. The gB glycoprotein is essential for GPCMV infection of all cell types [8, 9, 11, 12]. The GPCMV viral gH-based pentamer complex (PC) enables virus infection of non-fibroblast cells via an endocytic entry pathway, similar to clinical strains of HCMV, necessary for virus dissemination in animals and congenital infection [9–11]. In both GPCMV and HCMV, infection of fibroblast cells occurs via direct fusion of the cell membrane through interaction of cell-receptor PDGFRA with gH/gL/gO as well as gB [13–15]. In contrast, the PC-dependent pathway is less well understood but various PC candidate receptors have been identified [16]. In HCMV, the viral glycoprotein complexes are important neutralizing antibody targets and GPCMV glycoprotein complexes are similarly immunogenic neutralizing targets [8, 17–19]. The gB glycoprotein is an immunodominant antibody target and forms a trimeric complex in the virion [20]. HCMV gB has multiple antigenic domains (AD1-AD5) that contribute to the antibody response to the protein [21]. AD1 domain is immunodominant and necessary for gB oligomerization and AD1 sequence is conserved in various animal CMV gB [22, 23]. 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 [24–26]. Various gB vaccine strategies have been explored in clinical trials, either as recombinant subunit protein, viral vector, DNA and most recently a mRNA preclinical vaccine strategy [27–32]. Studies with a monomeric gB subunit vaccine and MF59 adjuvant have demonstrated efficacy in seronegative adolescent girls (43 %), preventing CMV infection in seronegative postpartum women (50 %) and reduction in viral load in solid-organ transplant recipients [27, 28, 33]. 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 [34]. However, in clinical trials, a subunit gB vaccine attains at best about 50 % efficacy against maternal infection, potentially due to poor virus neutralization on non-fibroblast cells [33, 35]. In GPCMV guinea pig model studies, the gB antigen has been the most extensively studied vaccine candidate against cCMV [36–39]. 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 strategies attained approximately 50 % efficacy in the guinea pig model [36–39].

In all previous GPCMV gB vaccine studies, virus neutralization on non-fibroblast cells was not evaluated, which is a significant limitation for comprehensive evaluation of vaccine efficacy. We recently demonstrated that a full-length GPCMV gB was a more effective strategy than previous gB approaches, which all utilized various C-terminal truncated gB proteins lacking the ability to form multimeric gB trimer complex and higher-order neutralizing antigens present in the virion of HCMV and GPCMV [12, 40]. This was demonstrated by recombinant defective adenovirus (Ad) vectors encoding full-length codon-optimized gB (AdgB) or truncated GPCMV gB (AdgBTMD) [12]. This approach induced higher anti-gB ELISA antibody titres than previous gB vaccines and sera from AdgB vaccinated animals induced a more potent neutralizing antibody titre on fibroblast cells but gB sera was less effective on epithelial and placental cells [12, 41]. Evaluation of a HCMV gB trimer vaccine strategy in mice improved the neutralizing immune response to HCMV on both fibroblast and epithelial cells. However, the neutralizing titre for the gB trimer was lower than that attained for hyperimmune globulin (HIG) preparation from HCMV convalescent patients that also had antibodies to the PC [40]. Studies with a HCMV gB mRNA vaccine capable of expressing a trimeric gB also evoked a more robust immune neutralizing response compared to the monomeric gB [31]. In GPCMV studies, the trimeric capable AdgB vaccine was relatively effective against challenge with wild-type prototype strain 22122 GPCMV in vaccinated animals and substantially limited virus dissemination to target organs compared to control non-vaccinated animals [12].

A highly significant problem with HCMV is that multiple clinical strains exist and there is potential for re-infection, therefore a vaccine must limit or prevent re-infection. In HCMV, antibodies to the PC are effective at neutralizing virus infection on non-fibroblasts including epithelial, endothelial and trophoblast cells [42–47]. Indeed, much of the protective antibody immunity from HIG preparation is attributed to antibodies directed to gH/gL and PC [48]. However, in other HCMV studies, gB antibody has been suggested to be more important than PC for virus neutralization [49, 50]. Undoubtedly, gB remains an important antibody target and essential for HCMV infection of all cell types. In a recent GPCMV DISC (defective infectious single cycle) vaccine strategy, we demonstrated that inclusion of the PC in a vaccine improved antibody neutralization on non-fibroblast cells and dramatically improved protection against cCMV [14, 17]. All GPCMV challenge studies to date, including the recent DISC vaccine studies, were carried out with prototype strain 22122 (ATCC VR682), which is tropic to both fibroblast and non-fibroblast cells. However, a potential limitation of GPCMV studies is the use of a single strain of GPCMV (22122), which limits the scope of vaccine efficacy and translational impact of GPCMV studies. Additionally, 22122 strain has preferential tropism to fibroblast cells and produces a high level of cell-released virus, unlike clinical HCMV strains.

We recently isolated a novel strain of GPCMV (TAMYC) from the salivary glands of an infected commercial animal [51]. This salivary-gland-derived novel strain would likely be important for horizontal and vertical transmission. The TAMYC strain of GPCMV had 25 % difference in glycoprotein gO amino acid sequence compared to 22122 strain and a similar degree of gO diversity between strains is common in HCMV clinical isolates [51, 52]. In contrast to 22122 strain, TAMYC preferentially grew on non-fibroblast cell lines (renal epithelial and trophoblasts) but grew extremely poorly on fibroblasts unless extensively passaged [51]. Despite the difference in tropism between 22122 and TAMYC, both viral strains had similar dissemination patterns in infected animals [9, 51]. The predicted gB sequences are 99 % identical between 22122 and TAMYC strains, with only two amino acid difference (S59R and D383N) [51], Fig. S1 (available in the online version of this article). A protein blast alignment in UNIprot for GPCMV 22122 gB (https://www.uniprot.org/uniprot/Q69024) with HCMV gB (Towne) shows 43.3 % identity and gB alignment between HCMV and GPCMV has previously been determined [53]. Based on HCMV and GPCMV gB alignment, the equivalent HCMV gB amino acids are 50 and 394 (for altered amino acids in gB GPCMV 59 and 383 respectively). Although antigenic domains have not been evaluated for GPCMV gB, the HCMV gB amino acids 50 and 394 are in AD2 and AD4 regions, respectively [21, 54, 55]. The potential AD2 region in GPCMV has poor conservation with HCMV but the change would likely be minor based on surrounding flanking charged sequences and the change from positive charge to polar (R→S). The change in the potential AD4 region is a conservative substitution without predictable impact in the immunological domain [21, 51, 56]. Indeed, HCMV gB has a D, or N, amino acid at position 394 dependent upon strain isolate (e.g. strains Toledo and Towne, respectively) [51, 56]. Importantly, the immunodominant gB AD1 domain (amino acids 547–620) [22] is identical between GPCMV strains [51]. Overall, the high level of conservation in predicted GPCMV gB sequences between 22122 and TAMYC strains provided an opportunity to determine the ability of a full-length GPCMV AdgB (22122) vaccine strategy to cross protect against multiple strains of GPCMV.

In this current study, we evaluated the ability of a GPCMV AdgB vaccine [12] to provide protection in vaccinated animals against challenge by a novel strain of GPCMV (TAMYC) compared to 22122 strain. Comparative neutralization studies with sera from AdgB vaccinated animals and convalescent (22122) animals infected with GPCMV (PC+) or GPCMV (PC-) indicated that effective virus neutralization on non-fibroblast cells required additional antibodies directed to the PC. In contrast, gB antibodies were more effective in GPCMV neutralization on fibroblast cells than epithelial cells. This suggested that a viral strain such as TAMYC that is highly cell-associated and non-fibroblast cell tropic would be poorly neutralized by a gB vaccine strategy. Unlike previous 22122 strain challenge studies, AdgB vaccinated animals, despite high anti-gB titre, had limited protection against TAMYC strain. The challenge virus (TAMYC) disseminated to all target organs including a substantial viral load in salivary glands and infection was more similar to that in seronegative animals. In contrast, hyperimmune GPCMV (22122) convalescent animals with antibody response to gB and PC had high protection against TAMYC strain challenge compared to AdgB vaccine. Overall, results suggest that an effective CMV glycoprotein complex based vaccine strategy against multiple strains of CMV requires inclusion of PC and gB as antibody targets to increase virus neutralization on non-fibroblast cells and likely provide higher-level protection against cCMV.

Methods

Virus and cells

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 and additionally on PDGFRA knockout fibroblast cells (GPKO) as previously described [9, 11, 14]. Virus stocks for antibody neutralization assays were generated on renal epithelial cells. Virus titres were determined by GPCMV titration on fibroblast cells [57]. TAMYC strain GPCMV was grown on REPI cells [51] and fibroblast adapted virus propagated on GPL cells. TAMYC is an anagram of the institute where the virus was isolated (Texas A & M University) and the investigators that isolated the virus (AM and YC). AdgB, recombinant defective adenovirus (Ad5) vector encoding full-length GPCMV (22122) gB was previously described [12].

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 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 [8]. Animal studies were performed to evaluate protection against challenge by TAMYC strain GPCMV. In a vaccine protection study, two groups of seronegative animals (n=12 per group) were used. Animals in group 1 received three inoculations of AdgB vaccine 4 weeks apart and anti-gB titre 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 TAMYC strain GPCMV (10⁵ p.f.u., SQ) at day 0. At days 4, 8, 12 and 27 days post-infection (days p.i.), three animals per group were euthanized to evaluate viral load in target organs and blood via real-time PCR assay as previously described [14, 17].

Evaluation of TAMYC GPCMV gB sequence from salivary gland virus in vaccinated animals

In order to determine if the AdgB vaccine strategy had selected for escape GPCMV mutants, DNA isolated from salivary gland tissue of vaccinated animals and subsequently challenged with GPCMV (TAMYC) was used to PCR clone the TAMYC GPCMV gB gene following previously described protocol and primers [8, 51]. The predicted amino acid sequence of TAMYC gB was previously reported [51]. DNA sequence of independent bacterial clones for the gB gene was determined and subsequently assembled and analysed in comparison to TAMYC and 22122 GPCMV gB genes by MacVector software. Predicted gB glycoprotein amino acid sequences for 22122 and TAMYC were compared by ClustalW protein alignment by MacVector software. DNA sequencing was carried out by Eurofins Genomics.

GPCMV glycoprotein ELISAs

Specific glycoprotein complex ELISAs (gB, gM/gN, gH/gL and PC) were carried out as previously described [14] using positive coating antigen derived from renal epithelial cell monolayers transduced with recombinant replication defective adenovirus (Ad) vectors expressing specific viral glycoprotein complexes or recombinant Ad vector expressing GFP for negative coating antigen [8, 9, 17]. In the case of gM/gN, codon optimized synthetic genes in mammalian plasmid expression vectors were used instead [8, 14]. 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 0.5 µg of either gB Ag+ or Ag- preparations diluted in carbonate coating buffer overnight at 4 °C, washed in 1× PBST, then blocked with 2 % non-fat dry milk. Test sera were diluted in blocking buffer from 1 : 80 to 1 : 20480 in doubling dilutions, incubated for 2 h at 37 °C and then reacted with anti-Guinea Pig IgG peroxidase antibody (Sigma) diluted (1 : 1000) in blocking buffer for an additional 1 h at 37 °C before reacting with TMB peroxidase substrate (KPL). Net OD (absorbance 450 nm) 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. The described approach is based on similar strategies for glycoprotein complex expression for HCMV and RhCMV and ELISAs [58, 59]. All ELISAs were run a 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. Final titre was determined by inverse of the highest dilution with positive result.

GPCMV Neutralization assays

GPCMV neutralization assays (NA₅₀) were performed on GPL fibroblasts, GPKO fibroblasts, REPI, TEPI and guinea pig aminotic sac epithelial (GPASE) cells with GPCMV(PC+) virus stocks (22122) generated on renal epithelial cells [8, 9] using pooled sera from a specific group as previously described [17]. Pooled sera for GPCMV(PC+) or GPCMV(PC-) infected hyperimmune animals was obtained from a previous study [14]. AdgB pooled sera was from AdgB (full-length gB) vaccinated animals as previously described [12]. Serially diluted sera were incubated with approximately 1×10⁵ p.f.u. GPCMV(PC+) in media containing 1 % rabbit complement (Equitech Bio) for 90 min at 37 °C before infecting GPKO, REPI, TEPI or GPASE cells for 1 h. For neutralization on GPL cells, 1×10³ p.f.u. GPCMV(PC+) was used. Infected cells and supernatant were collected on day four then titrated on GPLs. Final neutralizing antibody titre was the inverse of the highest dilution producing 50 % or greater reduction in plaques compared to virus only control. NA₅₀ assays were performed from each sample three times concurrently with the same virus stocks between groups. Immunohistochemical (IHC) staining of TAMYC GPCMV was carried out with anti-gB mouse monoclonal antibody [18] following previously described protocol [12]. TAMYC GPCMV virus neutralization for IHC assay was carried out at 1 : 40 dilution of anti-gB pooled sera from AdgB vaccinated animals or control sera from GPCMV seronegative animals at matching dilution in presence of excess complement as previously described [12]. NA₅₀ titration on fibroblast cells for TAMYC was carried out with GPL adapted TAMYC strain.

Anti-gB depletion from convalescent sera

Immunodepletion of antibodies to gB from pooled sera of either GPCMV(PC+) or GPCMV(PC-) convalescent animals was carried out as previously described [8]. Briefly, guinea pig cells were transduced with AdgB and at 48 to 72 h p.i., the cells were harvested, washed twice with PBS, and then fixed with a 1 : 1 ratio of an acetone-methanol fixation mixture for 20 min at −20 °C. Fixed cells were pelleted and then resuspended in 500 µl of PBS with 0.1 % Tween 20. An equal volume of serum was used for depletion overnight at 4 °C in a tube rotator. The cells were centrifuged at 10 000 g for 20 min at 4 °C to pellet the cells, and the serum was collected and then stored at −80 °C until needed. Glycoprotein-depleted sera were used for ELISAs and neutralization assays as described above. The starting serum dilution of 1 : 80 was adjusted for the 1 : 2 dilution during the immunodepletion step. Therefore, to reach the starting dilution of 1 : 80, the depleted serum starting dilution was 1 : 40.

Real time PCR

Blood and tissues (lung, liver, spleen) were collected from euthanized guinea pigs to determine the viral load as previously described [8, 17]. For tissue DNA extraction, FastPrep 24 (MP Biomedical) was used to homogenize tissues as a 10 % weight/volume homogenate in Lysing Matrix D (MP Biomedicals). To obtain DNA from whole blood, 200 µl of blood collected in ACD anti-coagulant tubes was used per extraction. DNA was extracted using the QIAcube HT (Qiagen) according to manufacturer’s liquid (blood) or tissue protocol appropriately. 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 25 µl 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 10 min at 40 °C followed by activation at 95 °C for 10 min, then 45 cycles of denaturation at 95 °C for 15 s, annealing at 56 °C for 15 s, elongation at 72 °C for 10 s. Data was collected by ‘single’ acquisition during the extension step. Standard curve was generated using GPCMV GP44 plasmid [60] for quantification and assay sensitivity. The sensitivity of the assay was determined to be five 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.

Statistical analysis

All statistical analyses were conducted with GraphPad Prism (version 7) software. Replicate means were analysed using one-way analysis of variance Tukey’s multiple comparison test with 95 % CI or Student’s t-test (unpaired) with significance taken as a P value of <0.05 or as specified in the figure legends.

Results

Neutralization of GPCMV by AdgB vaccine and sera from GPCMV convalescent animals

We previously demonstrated that full-length gB-based AdgB vaccine strategy evoked a more effective virus neutralizing antibody titre than a truncated gB protein lacking a transmembrane anchor domain, despite both vaccines evoking similar anti-gB titres [12]. In this present study, we compared pooled sera from AdgB (full-length gB) vaccinated animals to pooled sera from separate groups of animals hyper-immune to GPCMV(PC-) or GPCMV(PC+) and evaluated for anti-gB ELISA titre and virus neutralization assay (NA₅₀). Various guinea pig cell types were used in NA₅₀ studies including: guinea pig lung fibroblasts (GPL); renal epithelial (REPI); placental trophoblast (TEPI); amniotic sac membrane (GPASE); PDGFRA KO fibroblast cells (GPKO) [9, 11, 14, 15, 41]. Anti-GPCMV(PC-) and (PC+) sera was historical sera previously described [14] and demonstrated to have additional neutralizing antibodies to gH/gL [GPCMV(PC-)] or gH/gL and PC [GPCMV (PC+)]. Evaluation of anti-gB ELISA titres for each group (Fig. 1a) demonstrated that the AdgB sera (10240) had more than twofold higher titre than GPCMV(PC-) and GPCMV(PC+) sera (4096 and 4680, respectively), which was significant (P<0.05). GPCMV(PC-) and (PC+) convalescent sera had additional antibodies to gH/gL and gM/gN (Fig. 1b) with GPCMV(PC-) sera having a significantly higher ELISA response to gH/gL and gM/gN compared to GPCMV(PC+) convalescent sera (P<0.05). However, GPCMV(PC+) sera had a statistically higher ELISA titre to the PC (Fig. 1b) and GPCMV(PC-) antibodies are directed to gH/gL as the (PC-) virus does not encode the unique PC components (GP129, GP131 and GP133).

Fig. 1.

Comparative anti-glycoprotein ELISA and GPCMV neutralization between AdgB sera and convalescent hyperimmune GPCMV(PC+) and (PC-) sera. Pooled sera from animals at 4 weeks post-final AdgB vaccination was assayed for anti-glycoprotein ELISA titres and neutralizing titre capability of pooled sera measured on different cell types. Historical convalescent sera from separate groups of animals hyperimmune to wtGPCMV(PC+) or wtGPCMV(PC-) were used in comparative ELISA and neutralization (NA₅₀) assays with AdgB pooled sera. (a) Anti-gB ELISA titre between pooled sera collected from animals inoculated with AdgB (blue) and wtGPCMV(PC-) (grey) and wtGPCMV(PC+) (orange). *P<0.05 determined by one-way ANOVA Tukey’s multiple comparison test at 95 % CI or ns=not significant. (b) Anti-glycoprotein (gH/gL, gM/gN, gH-based pentamer complex, PC) ELISA titres of pooled sera from wtGPCMV(PC-) (grey) and wtGPCMV(PC+) (orange). (c) Comparative neutralization of wtGPCMV(PC+) by pooled sera collected from AdgB (blue) or wtGPCMV(PC-) (grey) groups tested on fibroblasts (GPL), epithelial (REPI), trophoblast (TEPI), amniotic sac epithelial (GPASE) and PDGFRA KO fibroblast (GPKO) cells. (d) Comparative virus neutralization by pooled sera collected from AdgB (blue) or wtGPCMV(PC+) (orange) groups on various cell types. (b–d) Statistics determined by unpaired Student's t-test *P<0.05; ns=not significant.

In NA₅₀ GPCMV neutralization assays, the AdgB sera was most effective on GPL fibroblast cells (1920) and least effective on various available non-fibroblast cell lines where entry required PC (range 320–640), including impairment on PDGFRA knockout GPKO fibroblasts (480), which require PC for virus entry via endocytic pathway (Fig. 1c, d) [14, 15]. On GPL fibroblast cells, GPCMV(PC-) and (PC+) sera had the same NA₅₀ value (4267), which was approximately double that of AdgB sera (Fig. 1c, d). However, on non-fibroblasts and GPKO cells, anti-GPCMV(PC-) sera was more similar in NA₅₀ values to AdgB (Fig. 1c). This was presumably because of a lack of specific antibodies to the PC. Anti-GPCMV(PC+) sera was two to threefold more effective at neutralization of virus on non-fibroblast cells and over twofold higher in NA₅₀ value on GPKO cells (Fig. 1d) compared to gB sera as well as GPCMV(PC-) sera (Fig. 1c). Overall, we concluded that despite high antibody titre to gB, neutralization on fibroblasts and non-fibroblasts could be enhanced by antibodies to additional glycoproteins, which supports previous studies using a GPCMV DISC vaccine strategy [14, 17].

On fibroblast cells, enhanced neutralization above that for gB was obtained by inclusion of antibodies to gH/gL, which was the basis for both GPCMV(PC-) and (PC+) sera having similar NA₅₀ values (Fig. 1c, d). However, neutralization of GPCMV on cells lacking PDGFRA are dependent upon the PC for cell entry [14, 15] and were more effectively neutralized by inclusion of PC-specific antibodies as well as gB. In order to demonstrate the limited effect of gB antibodies on virus neutralization on epithelial cells compared to fibroblasts, anti-gB was depleted from GPCMV(PC+) and GPCMV(PC-) convalescent sera (Fig. 2a, b) and subsequently anti-gB depleted sera used in GPCMV NA₅₀ assays on fibroblast and epithelial cells compared to non-depleted sera (Fig. 2c, d). On GPL fibroblast cells, anti-gB depletion had similar effect on both GPCMV(PC-) and (PC+) convalescent sera with over twofold reduction on NA₅₀ (Fig. 2c) compared to non-depleted, native sera. Epithelial cell NA₅₀ titres for both sera were reduced after depletion but GPCMV(PC-) sera NA₅₀ was more significantly reduced compared to GPCMV(PC+) sera (Fig. 2d).

Fig. 2.

Depletion of anti-gB from GPCMV(PC-) and (PC+) convalescent sera and impact on virus neutralization. (a) Anti-gB ELISA of wtGPCMV(PC+) convalescent sera pre- (native, solid orange circle) or post-gB depletion (open orange circle). (b) Anti-gB ELISA of wtGPCMV(PC-) convalescent sera pre- (native, solid grey square) or post-gB depletion (open grey square). Red dash line indicates the negative background level. Antibody neutralization assays (NA₅₀) of hyperimmune wtGPCMV(PC+) (orange) and wtGPCMV(PC-) (grey) sera, prior to (native) or post-gB-antibody depletion (ΔgB) on GPL fibroblasts (c) and REPI epithelial cell lines (d). Statistics determined by unpaired Student's t-test, *P<0.05; ns=not significant.

The comprehensive series of virus neutralization studies of GPCMV on various cell types were carried out on 22122 strain stock virus generated on renal epithelial cells to enable concurrent evaluation of virus neutralization with the same virus stock on all cell types. Since gB is essential for GPCMV infection of all cell types [8, 9, 41] and that gB amino acid sequence is 99 % identical between 22122 and TAMYC strains (changes S59R and D383N)(Fig. S1) [51], there would be high expectation that anti-gB sera would neutralize TAMYC strain on fibroblast and non-fibroblast cells in a similar manner to 2212 strain. However, the initial studies did not directly demonstrate neutralization of TAMYC strain by anti-gB sera. Unlike 22122 strain, TAMYC virus is highly tropic to epithelial cells (Fig. 3a) and grows poorly on fibroblast cells (Fig. 3e) unless extensively passaged on fibroblast cells (Fig. 3g) [51]. Additionally, TAMYC virus stock generated on fibroblasts is impaired for infection of non-fibroblast cells as earlier noted for 22122 strain despite encoding PC [12, 51]. An additional complication of TAMYC is that epithelial derived stock virus is highly cell-associated compared to 22122 (Fig. 3k, l). In contrast, fibroblast adapted TAMYC is more like 22122 strain and produces high levels of cell-release virus (Fig. 3k, m). Since the predicted gB sequence of fibroblast adapted TAMYC was identical to epithelial TAMYC (Fig. S1), a side-by-side NA₅₀ assay was conducted between 22122 and TAMYC on fibroblast cells with AdgB sera. Initially, a Western blot of virus infected cells demonstrated that AdgB sera reacted with gB protein in both 22122 and TAMYC infected cell lysate (Fig. 4a). Next, to demonstrate that gB antibodies in AdgB sera neutralized TAMYC virus infection, a NA₅₀ titration was performed against TAMYC GPL-adapted virus and 22122 virus (Fig. 4b). Results demonstrated that the gB sera was significantly less effective against TAMYC on fibroblasts compared to 22122 strain (NA₅₀ 640 v 1920) (Fig. 4b). In order to demonstrate that gB sera had neutralizing capability against TAMYC on epithelial cells, immunohistochemical staining experiments were performed on REPI cells infected with TAMYC. Prior to infection, TAMYC GPCMV was pretreated with anti-gB sera, or control seronegative sera. Pooled AdgB and control sera were used at a final dilution of 1 : 40 for TAMYC virus neutralization on epithelial (REPI) cells as this dilution of AdgB sera had previously been effective against 22122 strain [12]. Anti-gB sera at a dilution of 1 : 40 blocked virus infection of REPI cells in contrast to the control negative sera (Fig. 4c, f). Mock-infected control cell monolayers failed to detect virus or show any cytopathic effect (Fig. 4e, h). We concluded that anti-gB antibodies had the capacity to block cell entry by TAMYC virus on epithelial cells.

Fig. 3.

TAMYC strain GPCMV is highly cell-associated and tropic for epithelial but not fibroblast cells unless adapted. Comparison of TAMYC GPCMV strain infection on renal epithelial (REPI) cells (a–d) and GPL fibroblast cells (e–j). GPCMV infection of cells by REPI derived TAMYC strain virus (a, b, e and f). GPL fibroblast adapted TAMYC pass 19 on fibroblasts (g–h). Infections imaged at 4 days p.i. Viral gB protein staining of cells infected with TAMYC strain were carried out in six-well plates. Monolayers infected with TAMYC strain GPCMV (moi 1 p.f.u./cell) and mock-infected (MI) monolayers (c, d, i and j) were immunostained with monoclonal anti-gB (GPCMV) primary antibody [18] +anti-mouse IgG-HRP secondary with Vectastain kit (Vector Laboratories) or secondary antibody only (b, d, f, h and j). Individual bright field images are representative of multiple fields (minimum ten per panel). Magnification 10×. Evaluation of cell-associated (CA), cell-released (CR) fractions and total virus titre at 3 days p.i. on REPI or GPL cells from separate wells of six-well plates for TAMYC (k) and 22122 (l). Evaluation on GPL cells for fibroblast adapted TAMYC (m).

Fig. 4.

AdgB sera can neutralize GPCMV TAMYC strain on epithelial cells. (a) Western blot of GPCMV gB with AdgB sera. Lanes: (1) GPCMV (22122) infected cell lysate; (2) GPCMV (TAMYC) infected cell lysate; (3) mock-infected cell lysate on REPI cells. Arrow indicates gB (approximately 150 kDa). (b) Comparative AdgB sera neutralization for 22122 and GPL adapted TAMYC on fibroblast cells. NA₅₀ results for 22122 (blue) v TAMYC (orange). Statistics determined by unpaired student t-test *P<0.05. (c–h) TAMYC neutralization on epithelial cells. Virus neutralization carried out as described in Methods with AdgB sera (1 : 40 dilution) or control sera. IHC staining of GPCMV (TAMYC) neutralization on REPI cells by AdgB sera (c, d) or control sera (f, g). Mock-infected cells (e, h). IHC staining carried out with mouse monoclonal anti-gB (GPCMV) primary antibody [18] +anti-mouse IgG-HRP secondary with Vectastain kit (Vector laboratories) or secondary antibody only (d, g). Images taken at 4 days p.i. Individual bright field images are representative of multiple fields (minimum ten per panel). Magnification ×10.

It should be noted that the neutralization studies were directed against the cell-release virus fraction of GPCMV in our studies. An additional complication of TAMYC strain is that the epithelial derived virus is highly cell-associated unlike 22122 strain (Fig. 3k, l) [51]. Therefore, a limitation of the current assay is that we do not evaluate the impact of antibody on virus cell-cell spread. We consider this beyond the scope of the current study as it would likely require generation of a recombinant GFP-tagged TAMYC strain virus to enable definitive evaluation of antibody effect on virus cell-cell spread. We note that this is a limitation of many HCMV antibody neutralization studies with only a few evaluating the impact of HCMV antibodies on virus cell-to-cell spread. However, these limited studies suggest clinical HCMV adapts to escape antibodies by cell-to-cell spread mechanism [61]. Since TAMYC strain GPCMV infects mainly by cell-to-cell spread then it is likely that this would also be a potential mechanism of gB antibody escape for TAMYC virus compared to 22122 strain.

AdgB vaccine protection against TAMYC GPCMV strain challenge in guinea pigs

In a previous AdgB vaccine study, we demonstrated that the vaccine had high efficacy against challenge by wild-type GPCMV 22122 strain [12]. In this current study, we evaluated the ability of AdgB vaccine to cross-protect against a novel strain of GPCMV (TAMYC), which unlike 22122 virus was preferentially tropic to non-fibroblast cells [51]. Seronegative animals (group 1, n=12) were vaccinated (three dose SQ, 1×10⁸ TDU/injection/animal/every 4 weeks), confirmed for gB seroconversion by gB ELISA (average gB ELISA titre 10240) and subsequently challenged with wild-type GPCMV (TAMYC strain) 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 1×10⁵ p.f.u. SQ of GPCMV (TAMYC), and at 4, 8, 12 and 27 days p.i., 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 days p.i., viral load in salivary gland tissues was additionally evaluated. The viral loads for tissues and blood from each group are shown in Fig. 5a–e. Surprisingly, virus dissemination to target tissue in the gB-vaccinated animals was similar to that of the control seronegative animals in early stages of the study with day 4 viremia and viral load in all tissues showing no significant differences between the groups. Conversely, there was a significantly lower viral load (1–2 logs) of all tissues and blood for the vaccine group at 8 days p.i. (P<0.05). At 12 days p.i., similar viral loads were detected in the lungs of both groups, but viral loads in liver and spleen were only detected in the control group. At 27 days p.i., virus was detected in the salivary gland tissue of both groups with viral load relatively high (>10⁵ copies/mg tissues), and although roughly similar between groups (Fig. 5d), the higher viral load in the control group was statistically significant (P<0.05). This result contrasted with an earlier AdgB vaccine study with 22122 strain wild-type virus challenge that demonstrated higher gB vaccine efficacy [12] (Fig. 6). In this earlier study, 22122 strain virus dissemination to target organs was substantially reduced with detectable virus only at earlier time points at lower levels compared to control group and no detectable virus at 12 and 27 days p.i. [12] (Fig. 6).

Fig. 5.

AdgB vaccine fails to prevent dissemination of GPCMV TAMYC strain to target organs. Viral load in target organs (a–e) of AdgB seropositive animals (blue) and seronegative unvaccinated animals (orange) challenged with TAMYC strain (10⁵ p.f.u., SQ). Target organs: lung (a); liver (b); spleen (c) plotted as genome copies/mg tissue over 4, 8, 12, and 27 days p.i. Salivary gland (d) plotted as genome copies/mg tissue at 27 days p.i. only. Blood viremia at 4, 8, 12 and 27 days p.i. plotted as genome copies/ml blood (e). Statistics determined by unpaired Student's t-test, *P<0.05; ns=not significant. **Viral load in AdgBwt vaccinated samples below level of detection.

Fig. 6.

AdgB vaccine provides protection against GPCMV 22122 strain animal challenge. Viral load in target organs (a–e) of AdgB seropositive animals (yellow) and seronegative unvaccinated animals (purple) challenged with 22 122 strain (10⁵ p.f.u., SQ). Target organs: lung (a); liver (b); spleen (c) plotted as genome copies/mg tissue over 4, 8, 12 and 27 days p.i. Salivary gland (d) plotted as genome copies/mg tissue at 27 days p.i. only. Blood viremia at 4, 8, 12 and 27 days p.i. plotted as genome copies/ml blood (e). Statistics determined by unpaired Student's t-test, *P<0.05; ns=not significant. **Viral load in AdgBwt vaccinated samples below level of detection.

In order to rule out the possibility that the TAMYC virus gB had developed an antibody escape mutant in AdgB vaccinated animals, the GPCMV DNA isolated from the salivary gland tissue was used to PCR clone the GP55 (gB) cDNA and subsequently sequenced. The isolated DNA sequence for GP55 encoded gB with a predicted amino acid sequence identical to previously published TAMYC gB [51] (data not shown). Therefore, the basis for the limited vaccine efficacy on virus dissemination was not associated with a gB escape mutant evasion strategy. Overall, we concluded that the AdgB vaccine strategy was considerably less effective against TAMYC strain GPCMV compared to previous success against 22122 strain, where infection was substantially reduced in AdgB vaccinated animals and absent from salivary gland tissue compared to seronegative control or TAMYC vaccinated animals (Figs 5 and 6) [12]. We presume that this is based on TAMYC virus exhibiting higher cell tropism to non-fibroblasts compared to 22122 strain, where anti-gB neutralizing antibodies are less effective. Importantly, TAMYC is highly cell-associated compared to 22122 strain (Fig. 3) [51], which limits the impact of neutralizing antibodies on cell-free virus. Likely the mechanism for the difference in vaccine efficacy between strains is attributed to both a lack of fibroblast tropism and high level cell-cell spread preventing cell-free virus neutralization.

Convalescent immunity (22122) and heterologous GPCMV (TAMYC) challenge

Although gB immune response alone was insufficient to prevent infection by GPCMV, a recent DISC vaccine (22122) study indicated that inclusion of an immune response to the PC, as well as gB, and other viral antigens resulted in high vaccine efficacy against 22122 strain [14]. Consequently, we hypothesized that hyperimmune GPCMV(PC+) convalescent animals with similar response as a PC+ DISC vaccine would have high-level protection against heterologous TAMYC strain virus challenge. Animal were made hyperimmune to GPCMV(PC+) by three consecutive injections of 22122 GPCMV (10⁵ p.f.u. SQ/ injection) with each injection separated by 4 weeks. At 2 months post-last injection, convalescent animals (n=12) were challenged with TAMYC strain (10⁵ p.f.u. SQ). At 4, 8, 12 and 27 days p.i., animals were evaluated for viral load in target organs and blood. Low-level viral load was detected in various target organs at different times post-TAMYC challenge including relatively high viral load in salivary gland at 27 days p.i. (Fig. 7a). However, comparison of salivary gland between seronegative (Fig. 7c) and 22122 seropositive (Fig. 7a) TAMYC challenge studies showed a difference (2.3×10⁶ v 6×10⁴ genome copies) that was statistically significant (P<0.001) but below the expectation of an effective vaccine. Viral load in target organs was reduced in comparison to seronegative challenged animals (Fig. 7c) and viremia in convalescent animals was below detectable limits (Fig. 7b). Overall, hyperimmune 22122+convalescent animals had better protection against TAMYC challenge virus than AdgB vaccine (Fig. 5) but did not exhibit sterilizing immunity observed for a DISC vaccine against 22122 strain challenge [14], which would be expected for a successful vaccine against cCMV.

Fig. 7.

wtGPCMV (22122) hyperimmune convalescent animals provide limited protection against heterologous TAMYC virus challenge. Dissemination of heterologous GPCMV to target organs in 22122 GPCMV seropositive convalescent animals. Overall, wtGPCMV (22122) seropositive animals were challenged with TAMYC GPCMV (a, b) with 10⁵ p.f.u. virus, SQ (n=12). An equivalent control group of seronegative animals were similarly challenged with TAMYC virus (c, d). At days 4, 8, 12 and 27 days p.i., three animals per group were evaluated for viral load in target organs by real time PCR of tissue extracted DNA. Viral load (lung, liver, and spleen) plotted as viral genome copies/mg tissue (a, c). Salivary gland tissue was only evaluated at day 27 (a, c). Blood viremia at 4, 8, 12 and 27 days p.i. plotted as genome copies/ml blood (b, d). Real-time PCR standard amplification (e, f). A known concentration of GPCMV GP44 plasmid DNA was diluted tenfold and run in triplicate, as described in Methods, to generate the amplification curve (e) and the standard curve (f).

Discussion

What constitutes an effective immune response against HCMV is only partially understood. 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. Consequently, the viral glycoproteins, necessary for cell entry, are important neutralizing target antigens. The gB protein is essential for infection of all cell types and serum antibodies to HCMV gB can constitute 40–70 % of response to HCMV in convalescent patients [62, 63]. However, the majority of these antibodies are non-neutralizing [23]. Importantly, many of the original HCMV studies were based on virus neutralizing titre on fibroblast cells and given the ability of clinical strains to replicate on various cell types, evaluation of neutralizing antibody titre on non-fibroblast cells is important if not fundamental. Various vaccine strategies against cCMV are being investigated at pre-clinical and clinical stages. The gB antigen has been perceived as perhaps the cornerstone of an effective vaccine strategy because of the essential nature of gB and immunodominant antibody response. Therefore, gB as a single target antigen or as part of a cocktail of viral antigens are being widely investigated in various vaccines [25]. The most extensively studied HCMV vaccine is a soluble gB subunit approach but this provided at best 50 % efficacy in the guinea pig or in human clinical trials against cCMV [33, 64] and indicates that other viral targets are also important. HCMV PC is necessary for infection of non-fibroblast cells and also endocytic entry into fibroblast cells. Importantly, much of the protective antibody immunity from HIG preparations is attributed to antibodies directed to gH/gL and PC [48]. Indeed, antibodies to the PC and not gB would seem more effective for virus neutralization on placental cells as well as broad spectrum activity against multiple clinical strains [42–45, 47, 65–67].

A possible contributing factor to the limited efficacy of the gB vaccine in clinical trials is that a C-terminal truncated version of the protein was used that lacks the ability to form a gB trimer found in the virion, and consequently was unable to induce neutralizing antibodies against the higher-order target antigen [12, 40]. In the GPCMV model, we recently demonstrated that a full-length gB expressed by a recombinant Ad vector induces a more potent neutralizing immune response compared to a truncated gB [12]. Indeed, the neutralizing response was higher than that reported for previous GPCMV gB vaccine studies, which all utilized varying lengths of C-terminal truncated gB [38, 64, 68, 69]. However, virus neutralization titre was higher on fibroblasts than epithelial cells despite GPCMV gB being essential for all cell types [8, 9, 11, 14, 41]. Our original GPCMV studies and all GPCMV vaccine studies to date have utilized prototype strain 22122, which is capable of similar growth kinetics on fibroblast and non-fibroblast cells [8, 9, 11, 14, 41]. We recently isolated a new strain of GPCMV (TAMYC) from the salivary gland of an infected commercial animal that would be important for studying both horizontal and vertical transmission of the virus [51]. The TAMYC strain exhibited a preference for growth on various non-fibroblast cells including placental trophoblasts and grew extremely poorly on fibroblasts unless extensively adapted [51]. Potentially, since this new strain has only been passed on epithelial cells, it more realistically retains true wild-type GPCMV cell tropism found in naturally infected animal colonies. This contrasts with 22122 strain of GPCMV, which was passed extensively on fibroblasts (>100) when first isolated in the 1950s [70]. In addition to the preferential cell tropism to non-fibroblast cells, the TAMYC strain is highly cell-associated and produces very little cell-free virus. Consequently, both factors of tropism and lack of cell-free virus could potentially contribute to a mechanism associated with poor AdgB vaccine efficacy against TAMYC strain in contrast to higher efficacy against 22122 strain. However, it might be argued that in general, antibody neutralization of HCMV is ultimately limited as the virus may adapt to escape antibodies by cell-to-cell spread mechanism [61]. Consequently, a PC-based vaccine strategy or indeed gB+PC approach may also have limitations.

Despite reduced ability of AdgB sera to neutralize GPCMV (22122) on epithelial cells compared to fibroblasts, the vaccine strategy demonstrated relatively high efficacy against challenge wild-type virus (22122) in vaccinated animals and substantially reduced viral load in target organs compared to control group [12]. In contrast, the same AdgB vaccine strategy was substantially less effective at preventing dissemination by TAMYC strain challenge virus in vaccinated animals, which resulted in viral loads in target organs more similar to a seronegative control group of animals. Importantly, 22122 and TAMYC strains share 99 % identity in predicted amino acid sequence [51]. Results suggest that the limitation of the gB vaccine-based strategy was associated with preferential virus tropism to non-fibroblast cells, which reduced ability of gB antibodies to neutralize virus. An additional factor is that TAMYC produced limited amounts of cell-free virus for neutralization and infection was by cell-to-cell spread, which would serve to reduce vaccine efficacy. Limitation of the gB vaccine has been demonstrated in RhCMV studies. An initial MVA gB vaccine strategy failed to provide protection in a horizontal transmission model [71]. More recently, a truncated soluble gB subunit vaccine approach induced high antibody titre but poor neutralizing titre and failed to protect against challenge virus and shedding but there was reduction compared to control [72].

In the GPCMV model, the limitation of gB antibodies is further demonstrated by the ability of convalescent GPCMV sera with PC antibodies to be more effective against non-fibroblast infection as well as fibroblasts lacking the PDGFRA viral receptor. Inclusion of the PC in a recent GPCMV DISC vaccine strategy was highly effective in preventing challenge virus (22122 strain) dissemination to target organs with sterilizing immunity and fully protected against cCMV in vaccinated animals compared to a DISC vaccine lacking PC [14, 17]. Although, it should be noted that gB antibodies were important for DISC vaccine sera GPCMV neutralization in culture. Depletion of gB antibodies from DISC vaccine sera impacted GPCMV NA₅₀ values on fibroblast, REPI and TEPI cells [14]. However, only DISC vaccine encoding PC induced a high level of protection against GPCMV and higher titre neutralization on non-fibroblast cells. In this present study, GPCMV convalescent sera lacking PC antibodies was significantly impacted by depletion of anti-gB from the sera compared to depletion of gB from GPCMV(PC+) convalescent sera. Therefore, gB remains important as part of a vaccine strategy, but an effective vaccine likely must utilize additional neutralizing target antigens such as PC. It is interesting to note that gB antibody virus (22122 strain) neutralization on non-fibroblast cells can be enhanced by transient expression of the viral receptor PDGFRA, which binds the viral trimer (gH/gL/gO) to enable 22122 strain preferential entry via direct fusion (PC-independent pathway) [15, 41]. However, as TAMYC strain poorly infects fibroblast cells, transient expression of PDGFRA would likely have no impact on virus tropism or neutralization.

TAMYC is more similar to clinical strains of HCMV where cell–cell spread of the virus is a significant factor [67]. Currently, it is unclear if neutralizing antibodies are as effective in limiting cell-to-cell spread compared to infection by cell-release virus and this is a general limitation in the field for evaluation of CMV vaccines. Limited studies with clinical strains of HCMV would suggest that cell–cell spread virus is more resistant to neutralizing antibodies [61, 67]. Since TAMYC strain GPCMV infects mainly by cell–cell spread then it is likely that this could be a possible mechanism of escape for TAMYC virus compared to 22122 strain in the infected animal. Recent studies have suggested that an added impact associated with a HCMV gB vaccine strategy is non-neutralizing antibodies [34], which would be likely to target virus infected cells. In our previous AdgB study, both full-length and truncated gB (AdgBTMD) produced similar high titre antibodies in vaccinated animals but AdgBTMD induced lower titre neutralizing antibodies and possibly induced more non-neutralizing antibodies [12]. In this previous study, we demonstrated that depletion of complement in vaccinated animals had limited impact on vaccine efficacy for AdgB in a virus challenge experiment but did not rule out potential function of non-neutralizing antibody-dependent cellular cytotoxicity (ADCC), or antibody-dependent cellular phagocytosis (ADCP), which contribute to gB HCMV vaccine efficacy by non-neutralizing antibodies [12, 34].

In conclusion, we demonstrate that although a high titre, potent neutralizing gB based vaccine strategy is possible against GPCMV, there are limitations associated with a gB-only vaccine approach, especially against multiple strains of GPCMV despite conservation of gB sequence. This is the first time this has been evaluated in a CMV vaccine challenge animal model study. Neutralization by gB antibodies on non-fibroblast cells is weaker than neutralization on fibroblasts because of the endocytic PC-dependent route of cell entry for the former. Consequently, limitation of vaccine efficacy is compounded when challenge virus is highly tropic to non-fibroblasts compared to fibroblasts. Additionally, gB vaccine efficacy is perhaps limited against GPCMV if virus is also highly cell-associated and produces very little cell-free virus, as is the case for TAMYC and a common phenotype of clinical HCMV strains. Consequently, inclusion of TAMYC strain in future virus challenge studies will improve the translational impact of CMV vaccine research in the guinea pig model and provide a better indicator for likely vaccine success in clinical trials.