1. Introduction
Human cytomegalovirus (HCMV) is the most common congenital viral infection globally, and is a major cause of neurodevelopmental delays and long-term disability [
1,
2]. HCMV is also a leading cause of nongenetic sensorineural hearing loss worldwide [
3,
4]. Development of a vaccine against congenital CMV (cCMV) is a major public health priority, but, to date, there is no licensed vaccine available. A major challenge in vaccine development is the fact that the protective correlate(s) of immunity, essential for preventing infection of the developing fetus, remain unknown [
5,
6]. Moreover, it is not clear if a vaccine platform should be based on a live, attenuated strain of HCMV; disabled, replication-deficient HCMV genomic variants; or subunit vaccines delivering specific viral immunogens (either as adjuvanted purified proteins, or vectored through heterologous expression systems) that are believed to be important in protective immunity. Several HCMV vaccines based on these, and other, strategies are currently in various phases of preclinical and/or clinical trial evaluation [
7].
Of the three HCMV subunit vaccines that have been evaluated in Phase II clinical trials [
8], the HCMV envelope glycoprotein B (gB), administered with a squalene-based, oil-in-water emulsion adjuvant known as MF59, has received the most attention [
9,
10]. This vaccine is the only platform to date that has demonstrated efficacy in prevention of primary HCMV infection in women of potential child-bearing age [
9,
10]. When administered with MF59, the gB vaccine demonstrated 43–50% efficacy in phase II clinical trials in populations of seronegative postpartum [
10,
11] and adolescent women [
12]. Surprisingly, recent evidence suggests that the correlate of protection elicited by the gB/MF59 vaccine was not a virus-neutralizing antibody, but a non-neutralizing function: antibody-dependent cellular phagocytosis (ADCP) [
13,
14]. These observations have stimulated interest in efforts aimed at further refining gB subunit vaccines. Strategies include the inclusion of other immunogenic HCMV antigens, in addition to gB; the exploration of more potent adjuvants; and the development of genetic and biochemical approaches that might optimize immune responses to the most immunogenic conformations and epitopes of the gB molecule [
15,
16].
Several additional HCMV glycoproteins and glycoprotein complexes have been proposed, as additions and/or alternatives to gB, for prophylactic subunit vaccination to protect against cCMV transmission. HCMV utilizes different glycoprotein complexes to infect different cell types: entry into fibroblasts involves glycoprotein homotrimers gB and gH/gL/gO [
17,
18,
19], whereas entry into epithelial and endothelial cells requires gB, gH/gL/gO, and the pentameric complex (PC, gH/gL/UL128/130/UL131A) [
20,
21,
22,
23]. The discovery that high potency anti-PC antibodies neutralize HCMV infectivity [
24,
25] has rekindled interest in the potential clinical utility of high-titer anti-HCMV immune globulins. Recent studies have demonstrated that the majority of the virus-neutralizing activity in Cytogam
®, (CSL Behring LLC, King of Prussia, PA, USA), a commercially prepared, pooled, high-titer human convalescent sera, targets the PC, and not gB [
26]. These observations have, in turn, driven interest in devising expression strategies targeting the PC in subunit vaccine design. Strategies include the expression of the PC as a component of a disabled, infectious, single-cycle (DISC) HCMV vaccine [
27]; the purification of recombinant PC protein from CHO cells [
28]; and the expression of PC in a vectored vaccine, using recombinant modified vaccinia virus Ankara (MVA; [
29]). PC-based vaccines are in several stages of development, but none have yet been tested in clinical trials. An initial report suggested an association between rapidly-developing high-titer, anti-PC antibodies and improved pregnancy outcomes [
21]. Thus, the rapidity of induction of anti-PC antibody may be a critical requirement for such a vaccine. However, not all studies report a protective benefit of anti-PC antibodies against cCMV transmission. For example, in a recent report from Brazil, the titer of anti-PC and anti-gH/gL/gO antibodies did not correlate with protection against cCMV in a high-seroprevalence population of women of childbearing age [
30].
Given the uncertainty about which glycoprotein complex(es) represent the ideal immunogen(s) for inclusion in a subunit cCMV vaccine, it would be desirable to compare the protective efficacy of gB- and PC-based vaccines in an animal model of cCMV transmission. This information would help inform and direct future clinical trials. Since CMVs are highly species-specific, HCMV vaccines cannot be evaluated for protection against cCMV in any animal model. However, both rhesus macaque CMV (RhCMV) and guinea pig CMV (GPCMV) provide useful models for vaccine study [
31,
32]. Moreover, both viruses encode homologs of the PC. In addition to gH and gL, Rh157.5, Rh157.4, and Rh157.6 encode proteins that are homologs of HCMV UL128, UL130, and UL131 [
33]. As with HCMV, the RhCMV PC is essential for entry into epithelial cells [
34,
35]. In rhesus macaques vaccinated with a bacterial artificial chromosome (BAC)-derived MVA vector co-expressing all five RhCMV pentameric subunits, potent RhCMV-specific neutralizing antibody responses were elicited that were capable of blocking infection of epithelial cells and fibroblasts. Reduced RhCMV plasma viral loads were also observed following vaccination and subsequent RhCMV challenge [
36]. Notably, GPCMV also encodes homologs of the HCMV PC ORFs [
37,
38,
39,
40], including gH, gL, and proteins designated GP129/131/133 (homologs of HCMV UL128/130/131a). Although the GPCMV PC elicits antibody responses in the context of GPCMV infection [
39], no high-potency neutralizing monoclonal antibodies, specifically targeting conformational epitopes, have been described [
41], and, to date, there has been no examination of a PC-based subunit vaccine to test for protection against congenital GPCMV transmission. Thus, this study was undertaken to compare the protective efficacy of a gB-based vaccine with gH/gL and PC-based vaccines, each generated using an MVA-vectored approach, in the context of the GPCMV congenital infection model.
4. Discussion
Against the backdrop of the urgent need to generate a vaccine capable of preventing the disabilities caused by congenital HCMV infection, there is considerable interest in developing a vaccine platform that targets the HCMV PC (gH/gL/UL128/130/131A). This interest is driven by the discovery that the PC is essential for viral entry into epithelial and endothelial cells [
17,
18,
19,
20], and the observation that antibodies to PC are capable of potently neutralizing HCMV, both in the context of natural infection and following subunit immunization of mice using HCMV PC vaccine constructs [
21,
22,
23,
24,
25,
26]. Accordingly, multiple investigators and vaccine manufacturers are pursuing strategies that focus on targeting the PC as a key immunogen [
27,
28,
29]. Since the GPCMV recapitulates much of the biology of HCMV, including the propensity of the virus for trans-placental transmission to the developing fetus [
32], we undertook these studies in order to compare the efficacy of GPCMV, PC and gB-based subunit vaccines against congenital infection and disease.
Like HCMV, the GPCMV encodes a PC, consisting of proteins gH, gL, and UL129/131/133 [
37,
38,
39,
40]. We elected to express the GPCMV PC as an MVA-vectored vaccine. This approach has been successfully employed to generate both RhCMV and HCMV PC-based vaccines [
29,
36], both of which have been shown to be immunogenic and capable of directing the synthesis of a functional pentamer. Generation of the multi-subunit vaccine was facilitated by the use of 2A peptides [
44]. This approach also had the advantage of generating a PC-expressing construct that could be directly compared to other MVA vaccines (hence the use of the same expression platform), based on additional key immunogens of interest in CMV vaccine design, such as gB. Previous work with an MVA-vectored gB vaccine in the GPCMV model demonstrated a vaccine-mediated reduction in pup mortality and reduced congenital GPCMV transmission [
42]. Since the GPCMV PC has been demonstrated to elicit antibody responses in the context of GPCMV infection [
39], this experiment provided an opportunity to compare the relative protective effect provided by the gB and PC vaccinations, in this relevant preclinical model.
Interestingly, the comparative analysis demonstrated that the gB subunit vaccine was superior in stimulating correlates of protective immunity by varying degrees, compared to either the PC vaccine or the gH/gL vaccine (engineered to express only the gH/gL proteins, and not the GPCMV PC). The MVA-gpgB vaccine resulted in a statistically significant reduction, compared to controls, in maternal DNAemia at day 14, following an early third trimester challenge of pregnant dams with virulent SG-adapted GPCMV. The gB vaccine was also significantly associated with reduced congenital transmission, and lower visceral organ viral loads in pups (
Figure 4b). However, although not significantly associated with reduced congenital GPCMV transmission, both the MVA-gp75/gL and MVA-gpPC vaccines were also associated with improved pup survival (
Table 4) and improved pup birth weight (
Figure 5). These observations may be relevant to HCMV vaccines. Although the often-stated metric of success for the licensure of a vaccine against congenital HCMV infection is prevention of congenital infection [
56], a vaccine that reduces disease in the newborn infant, even if transmission occurs, could be valuable. Although controversial [
57], there are lines of evidence that suggest that congenital infection occurring in the context of recurrent maternal infection (re-infection in the face of pre-existing maternal immunity) may be less likely to produce sequelae in infants [
58] than transmission occurring in the context of primary maternal infection during pregnancy.
The key effector of protection in this study may have been the antibody response, in particular the neutralizing antibody response to vaccination. We observed that the MVA-gpgB vaccine elicited both an enhanced ELISA response, as well as a significantly higher neutralizing response, compared to the MVA-gp75/gL and MVA-gpPC vaccines. We confirmed that the eGFP-tagged reporter virus used for the neutralization studies, vJZ848, encoded wild-type sequence in the GPCMV PC locus (data not shown). Accordingly, any differences in neutralizing titers in GPL cell lines was not due to a failure of the reporter virus to express the PC proteins.
One potential shortcoming of this study is that the GPCMV PC may not completely represent a bona fide homolog of the HCMV PC, in terms of pathogenesis and immune response. The HCMV PC plays a key role in viral entrance into epithelial and endothelial cells, but is dispensable for its entry into fibroblasts. The GPCMV PC, in contrast, does not appear to confer the same cell-type specific tropism-directing functions for epithelial and endothelial cells as the HCMV PC [
38]. Indeed, the GPCMV PC appears to play a critical role in entry into many, diverse cell types, since a GP129-133 deletion mutant demonstrates defects in both endothelial cell and fibroblast cell entry [
38]. The GPCMV PC, in particular the GP129, also plays a role in macrophage-mediated dissemination of virus in vivo [
59], possibly mediated through a putative CC chemokine function [
54]. The GPCMV PC is also required for pathogenesis in experimentally challenged, non-pregnant animals [
60]. Future work will be required to more fully characterize whether vaccination against the GPCMV PC exerts a beneficial effect by eliciting virus-neutralizing activity, or perhaps instead by inducing antibody responses that target an immunomodulatory or chemokine-like activity, conferred by the GP129 protein. It is also notable that, while there is a correlation between neutralization titres and protection in this study, this may not completely represent the mechanistic correlate of protection, since non-neutralizing functions, such as ADCP, have recently been described that appear to mediate the modest protection conferred by gB/MF59 in HCMV vaccine studies [
13,
14,
15]. Such non-neutralizing effectors of protective immunity would be fruitful areas for future evaluation of the GPCMV congenital infection model.
Recently, a disabled, single-cycle (DISC) vaccine against GPCMV was described [
61], attenuated by virtue of a mutation introduced into the
GP85 gene. This vaccine elicited a broad repertoire of immune responses, including both humoral and cellular responses (in particular, CD4+ T cell responses to the GPCMV homolog of HCMV pp65). The vaccine provided excellent protection against congenital GPCMV transmission. This DISC vaccine had an intact GPCMV PC sequence, but the dependence of anti-PC responses on the magnitude of vaccine-mediated protection was difficult to gauge, since the PC responses occurred in the context of an additional, broad repertoire of other antibody and cellular responses to many viral antigens. Furthermore, clear evidence that would suggest the production of potent neutralizing antibodies against the GPCMV PC (similar to the previously isolated human antibodies that predominantly target conformational epitopes of the UL128/130/131A subunits of the HCMV PC) was absent. Even more importantly, the protective studies were compared to historical controls that were based on responses engendered against a PC-negative “first generation” DISC vaccine [
62] using a different dose regimen. The authors noted that, although both two- and three-dose regimens were examined for immunogenicity with the DISC vaccine, protection studies for the first-generation vaccine were completed using only those dams immunized with the suboptimal, two-dose vaccine regimen [
62]. These were nonetheless directly compared to pregnancy outcome results using the second-generation DISC vaccine which were obtained using an optimized three-dose regimen [
61]. Thus, the two studies are not strictly comparable. A contemporaneous, side-by-side, three-dose comparison of PC-intact- and PC-deficient DISC vaccines would be required to definitively assess the importance of the PC for protection in the GPCMV model.
In summary, the data from this study support the continued exploration of vaccine-mediated anti-PC responses, possibly in combination with T-cell targets, in the GPCMV model using MVA vectors. The neutralization of CMV-encoded immunomodulatory gene products is also emerging as a novel future strategy for vaccination [
63]. In addition to DISC vaccine studies, subunit vaccine approaches, combining glycoprotein targets and GPCMV immune modulation genes, may merit future consideration in the GPCMV model, and could help inform and direct HCMV vaccine strategies.