Human cytomegalovirus vaccine development: Immune responses to look into vaccine strategy

ABSTRACT

Human cytomegalovirus (HCMV) causes considerable morbidity and disability in high risk, immunocompromised populations including recipients of solid organ transplants, and fetuses whose immune systems are not yet mature. Vaccines aimed at ameliorating the severity of disease and preventing HCMV infection can be categorized into two main approaches of vaccine design, with one focusing on virus modification and the other on individual antigens. However, no candidates in either class have been successful in achieving durable and protective immunity. Recent studies on the natural immune response provide new insight into HCMV vaccine strategy. In particular, studies have demonstrated that the incorporation of a pentameric complex is necessary for a vaccine to generate the potent neutralizing antibodies often seen in seropositive individuals. This review summarizes recent findings in the development of HCMV vaccines and key considerations that should be taken into vaccine design based on improved understanding of natural HCMV immunity.

KEYWORDS:

• human cytomegalovirus
• vaccine
• immune response
• neutralizing antibodies
• T cells

Introduction

Human cytomegalovirus (HCMV) is a prototypical β-herpesvirus that commonly infects humans but rarely causes disease in healthy subjects. HCMV infection can have serious consequences in fetuses and immunocompromised individuals^1–4 such as recipients of solid organ transplants (SOT) or hematopoietic stem cell transplant (HSCT) with immune suppression. Consequences of HCMV infection in transplant recipients can include pneumonitis, encephalitis, and hepatitis. HCMV infection remains a risk factor for transplantation-related mortality and graft failure.^5–7 Despite antiviral therapy, the incidence of HCMV infection is still high, ranging from 20–70% in the first year post transplantation.^4,8,9 In addition, pregnant women are susceptible to HCMV infection through contact with infected body fluids such as urine or saliva.^2,10,11 It is estimated that there are 20,000 to 40,000 children are born with congenital HCMV infection in the United States each year. Of these cases, there are about 100–200 deaths as a consequence of symptomatic infections, and about 4000–8000 children suffer permanent auditory, cognitive, and neurologic disabilities each year. These sequelae bring high economic burden on healthcare system.^12–14 Thus, an effective HCMV vaccine would address an unmet medical need. Efforts to develop a vaccine for the prevention of HCMV congenital infection has been assigned to the category of the highest priority by the Institute of Medicine since 1999.^15,16

Natural history and experimental intervention studies support the hypothesis that natural immunity acquired through HCMV infection in healthy subjects is protective against secondary infection or congenital transmission. From analysis of women with infants shedding virus, Adler and coworkers showed that over 47% HCMV seronegative mothers (9 out of 19 subjects) were infected with HCMV, while only 7% seropositive mothers (3 out 42) acquired new infection from their children.¹⁷ Serological responses in this study were measured by enzyme immunoassay (EIA). A titer was defined as the highest dilution of serum that produced an OD of ≧ 0.1 (OD = 0.1 is lower limit of detection, LLOD). In addition, Fowler and coworkers showed that maternal seropositive status before conception was associated with reduced rate of congenital transmission.^18,19 To prevent congenital transmission in women with primary HCMV infection during pregnancy, therapy includes administration of high-titer anti-HCMV hyperimmune globulin (CMVIG), usually prepared from the plasma of seropositive donors with high titers of antibody against HCMV. Administration of CMVIG providing passive immunity has been evaluated as an intervention. In a seminal study by Negro et al, 6/37 women who received hyperimmune globulin during pregnancy had infants with congenital HCMV infection, compared with 19/47 women who did not receive the high-titer HCMV globulin.²⁰ CMVIG has also been studied in post-transplantation patients; prophylactic administration of CMVIG in kidney transplant patients showed that CMVIG has a beneficial effect on total survival, reduced HCMV disease, and HCMV-associated deaths.^7,21 HCMV natural immunity was determined by measuring HCMV-specific IgG antibodies in serological EIA assay. Besides CMVIG, early reconstitution of HCMV natural immunity was also significantly associated with a lower risk of HCMV infection in transplant patients. The strongest immune correlates with protection against HCMV viremia in HSCT recipients are reconstitution of HCMV-specific T cell memory responses, measured by lymphoproliferative assays.²²

Natural immunity demonstrates efficacy against HCMV infection and congenital transmission. The rational approach for HCMV vaccine design is to focus on the immune response seen in natural immunity, including potent neutralizing antibodies and HCMV-specific cellular responses such as HCMV-specific T cell memory responses. A number of experimental vaccines have been evaluated in clinical trials.¹⁵ This review summarizes the current state of HCMV vaccine development (Table 1) and discusses related quality and quantity of the immune responses by induced by vaccination.

Table 1. Summary of HCMV vaccine candidates.

Category	Type	Candidate	Status	Comments
Modified Virus Vaccine	Live, attenuated vaccine	^aAD169 vaccine	Phase I	Passage-attenuated in human fibroblast cells, no longer in clinical trials^80
		^aTowne vaccine	Phase II	Lack of efficacy for HCMV infection;^85 reduced HCMV disease in renal transplant recipients; failed to protect congenital HCMV infection^17
		^bTowne/Toledo chimeric vaccines	Phase I	Favorable safety profile; No efficacy data available^93
		^cDense body vaccine	Preclinical	Tested in small animals for immunogenicity^126
		^dreplication defective vaccine with restored gH complex	Phase I	induce lasting and potent neutralizing antibodies and T cell response in multiple animal species and non-human primates^119
Individual antigen vaccine	Subunit protein vaccine	^egB/MF59	Phase II	Expressed in CHO cells, showed 50% efficacy in seronegative women^12 and controlling viremia in ^hR^−/D^+ solid organ transplant recipients^127
	Subunit DNA vectored vaccine	^fgB/pp65	Phase II	Efficacious in reducing viremia in hematopoietic stem cell transplant recipients^109
		gB/pp65/IE1	Phase I	Prime-boost regimen with Towne vaccine^112
	Subunit viral vectored vaccine	^gAlphavirus VEE replicon particle (VRP)	Phase I	Engineered using replication-deficient alphavirus technology; VRP was produced and purified from Vero cells^128
		Alphavirus VRP for gH/gL	Preclinical	A single VRP expressing full length gH and gL, aimed at inducting neutralizing antibodies;^129 tested in small animals for immunogenicity
		ALVAC vector for gB	Phase I	Recombinant canarypox virus expressing gB, either tested alone or in combination with Towne vaccine,^113 or in a combination regimen with gB/MF59 vaccine^114
		ALVAC vector for pp65	Phase I	Recombinant canarypox virus expressing pp65 tested for induction of CTL responses; No efficacy data available^130
		Modified Vaccinia Virus Ankara (MVA)	Preclinical	Multiple MVA vectors expressing soluble gB, IE1, pp65, or a fusion of pp65 and IE1; Excellent immunogenicity (humoral and cellular responses) in mice^131
		UL130, UL131 peptide conjugate vaccines	Preclinical	Synthetic peptides from UL130 and UL131 conjugated to KLH carrier and attempt to induced neutralizing antibodies in small animals^132

^a AD169 vaccine and Towne vaccine are live attenuated virus, passage-attenuated in human fibroblast cells. They failed to prevent HCMV infection, due to the lack of pentameric gH complex in genome.

^b Four recombinant chimeric viruses between Towne and Toledo strains, manufactured in human fibroblast cells. It is currently under clinical evaluation in HCMV seronegative subjects (NCT01195571).

^c Noninfectious viral particles enriched from culture supernatant of human fibroblast cells infected with AD169 or Towne virus.

^d Engineered to be conditionally replication-defective, and produced in human retinal pigment epithelial cells. Currently in Phase I (NCT01986010).

^e Soluble gB expressed in CHO cells, with the furin-cleavage site mutated and transmembrane domain deleted, and formulated with MF59 adjuvant.

^f Two DNA constructs encoding a soluble form of gB and a modified pp65, formulated with a poloxamer (CRL1005).

^g VEE, Venezuelan equine encephalitis virus; VRP, VEE replicon particle; Two VRP constructs expressing a soluble form of gB and a pp65/IEI fusion protein.

^h R⁻/D⁺, a HCMV seronegative recipient with SOT from a seropositive donor.

HCMV viral particles and cell tropism in HCMV neutralization

HCMV is one of the largest and most complex enveloped viruses. It has a 230-kb double stranded linear DNA genome surrounded by a cellular lipid layer and a number of viral glycoproteins.^23,24 It is capable of encoding over 165 open reading frames (ORFs).^25–30 The roles of these components in infection and replication were not completely understood, due to the difficulty of viral gene expression and the lack of HCMV animal model.³¹ HCMV can infect a broad range of cell types. Epithelial cells, endothelial cells, smooth muscle cells, and fibroblast cells are the cell types most commonly infected for virus replication.^23,32 The mature virions are ∼200 nm in diameter and contain over 50 viral proteins, including viral capsid proteins, tegument proteins, and envelope glycoproteins.²⁴ Many viral structural proteins, such as tegument proteins pp65, pp150, and pp50/52, are targets of host cellular immune responses.³³ The phospholipid envelope contains 6 encoded glycoproteins: gpUL55 (gB), gpUL73 (gN), gpUL74 (gO), gpUL75 (gH), UL100 (gM), and gpUL115 (gL). These glycoproteins are essential in virus entry, replication, and spreading.^24,34 The protein products of these six major glycoproteins form three associate complexes: gCI (gB), gCII (gM/gN), and gCIII (gH/gL/gO).³⁵ These complexes are highly conserved in the herpesvirus family. They are assembled for a variety of experimental vaccines being evaluated in clinical trials (Table 2). All major viral glycoprotein complexes are capable of eliciting neutralizing antibodies in naturally infected individuals or in vaccinated experimental animals. However, the majority of neutralizing antibody activity in naturally acquired immunity is directed at the pentameric gH complex consisting of gH, gL, pUL128, pUL130, and pUL131^36,37 (Table 2). Sha Ha and coworkers expressed and purified soluble gH/gL/pUL128–131 pentameric complex and gH/gL to better understand the neutralizing mechanism. They proved that the pentameric gH complex, but not gH/gL, presents dominant native neutralizing epitopes. Antibodies that bind only to the pentameric complex are expected to be more potent neutralizers than those that bind both the pentameric complex and gH/gL.³⁸

Table 2. Potential CMV proteins in vaccine design and their roles in immune response.

Type	Candidate proteins	Role in immune response
Glycoproteins	gB complex	Induce weak NAbs^a; targets of CTLs^b
	gH/gO/gL complex	Induce weak NAbs^a; targets of CTLs^b
	gM/gN complex	Induce weak neutralizing response
	pentameric gH complex	Major targets of potent NAbs^a; targets of CTLs^b
Structural proteins	pp65 (UL83)	Major targets of CTLs^b; antibody response
	pp150 (UL32), pp28 (UL99)	Targets of CTLs^b
	pp71 (UL82)	Antibody response and CD8^+ T cell response
Regulatory protein	IE1 (UL123)	Important target of CTLs^b; antibody response

^a NAbs, Neutralizing antibodies; weak NAbs, neutralizing antibodies with weak potency; potent NAbs, neutralizing antibodies with high potency

^b CTLs: Cytotoxic T lymphocytes

HCMV was isolated for the first time in 1956 in human fibroblasts.³⁹ In early vaccine development, researchers established two clinical isolates AD169⁴⁰ and Towne⁴¹ from fibroblasts. Since then, neutralizing antibodies have been evaluated in this cell system. These laboratory strains were developed as attenuated vaccines by serial passages cultured exclusively in fibroblasts. Scientists found that these vaccines failed to infect epithelial and endothelial cells, which are permissive for virus replication.^42,43 In a panel of antibodies from human sera with high neutralizing titers against the HCMV clinical isolates, a striking difference was observed. The neutralizing antibodies showed higher blocking ability in endothelial or epithelial cells than was measured in fibroblast cells. Hence, conventional determination of neutralizing activity in fibroblasts is misleading.^44,45 More recently, the UL131–128 locus of the HCMV genome has been shown to be indispensable for the infection of endothelial cells and epithelial cells; its mutation will cause loss of the pentameric gH complex.^46–48 This complex, which is a determinant for viral tropism to endothelial and epithelial cells, comprises the gH/gL scaffold and UL131, UL130, UL128 proteins.⁴⁹ When the UL131 mutation was restored in the AD169 parental virus, the revertant virus induced 10-fold higher neutralizing antibody titers than an attenuated AD169 virus. Further, the peak neutralizing titers post vaccination in rabbits and monkeys were 2-4-fold higher than the levels determined in HCMV seropositive subjects.⁵⁰ These results reveal the importance of the pentameric gH complex as a target of potent neutralizing antibodies in vaccinated animals or seropositive humans.³⁷

Defining the efficacy for HCMV vaccines in target populations

Despite significant progress in HCMV vaccine development, no product is yet under consideration for licensing. One challenge is how to define vaccine effects in clinical trials. Recently, a multidisciplinary meeting addressed priorities on the relevant and practical endpoints for assessing vaccine efficacy for prevention of HCMV disease in target populations.⁵¹ The first endpoint is that a universal childhood HCMV vaccine should rapidly reduce congenital HCMV disease, as infected children are sources of viral transmission to seronegative and seropositive mothers.⁵¹ The second is that clinical trials of HCMV vaccines in women should be evaluated for an endpoint of prevention of congenital HCMV infection. This is a more practical and acceptable endpoint for assessing vaccine effects on maternal-fetal transmission.⁵¹ Recipients of allogeneic HSCT and SOT are at high risk for HCMV disease, pneumonitis, enteritis, retinitis, and viremia. HCMV serological status is associated with clinical outcomes including transplant-related mortality, re-infections, and failure of graft survival.⁶ The third endpoint is that employment of vaccines should be of value in decreasing incidence and severity of HCMV disease, as well as the other complications of transplantation.⁵¹ Such vaccines would be expected to be capable of eliciting humoral and cellular response after administration to either the transplant recipients or to the HSCT or solid organ donor prior to transplantation.

Natural immunity provides new insights for HCMV vaccine efficiency

HCMV natural immunity provides protection against the most devastating forms of congenital infection. Studies from Fowler and coworkers showed the presence of maternal antibody before conception provides substantial protection against congenital HCMV infection in newborns. Serologic assays were conducted in EIA using HCMV-specific IgG antibodies.^18,19 Following 197 newborns with congenital infection for 4.7 years, the study showed that 13% of infants whose mothers had primary infection during pregnancy had mental impairment, as compared with no symptoms for those whose mothers had recurrent HCMV infections.¹⁹ There was a marked difference between the infants who acquired immunity vertically transmitted from their mothers and those who did not. Another line of evidence indicated that naturally acquired immunity resulted in a 69% reduction in the risk of congenital HCMV infection in future pregnancies.¹⁸ Fowler and coworkers also concluded that protection against congenital HCMV infection was enhanced by longer time intervals between pregnancies.⁵² An emphasis should be placed on HCMV vaccine design for which adaptive immune responses after vaccination mimic those components observed in the natural human immune response. The immune response may be imitated in both humoral and cellular immunity, including functional antibodies and T cells that have a significant impact on preventing HCMV infections.

Neutralizing antibodies in humoral response

Before clinical trials, HCMV vaccine strategies must be evaluated in animal models. One strategy that has been explored in the murine cytomegalovirus (MCMV) disease model is that of live attenuated vaccination. Studies of this strategy revealed that the attenuated mutants used as candidate vaccines were capable of inducing both humoral and cellular immune responses.⁵³ In addition, several investigations have examined the efficacy of subunit vaccines^54,55 and DNA vaccines⁵⁶ in MCMV models. A prime-boost immunization strategy used in the MCMV model supported further study of prime-boost approaches for CMV vaccination in other animal models, potentially, for human clinical trials.⁵⁷ The MCMV is similar to HCMV in virion structure, genome organization, gene expression, tissue tropism and latency. MCMV M123 (IE-1) is the homologous gene of HCMV UL123 (IE-1), and is expressed at the IE phase of MCMV replication as a non-structural protein pp89. M84 is expressed in E phase of replication as a non-structural protein of 65 kD (p65), and its homologous protein pp65 in HCMV is the main antigen for cytotoxic T lymphocyte (CTL) response against HCMV infection in humans. The m04 gene is only present in MCMV, and is expressed in E phase of virus replication as gp34 protein and plays an important role in immune escape of MCMV. M105 is expressed as DNA helicase of MCMV, and its HCMV homolog UL105 is highly conserved. M55 gene and its HCMV counterpart UL55 gene both encode the envelope glycoprotein B, the target antigen recognized by HCMV specific antibodies. The five DNA vaccines were constructed by Ze Chen's group. They found that the DNA vaccines, especially m04, M84 and IE-1, could effectively reduce the virus loads in salivary glands and spleens of mice, but they couldn't completely clear the residual virus. Immunization with M55 or M105 DNA at four doses offered mice only 62.5% survival rate after the lethal challenge.⁵⁸ Evidence from several animal models of CMV infection indicates that a variety of vaccine strategies are capable of inducing immune responses sufficient to protect against CMV-associated illness following viral challenge.⁵⁹ However, we still cannot study the virus–host interactions in any animal model for HCMV. HCMV infection is restricted to its natural host, and human pathologies caused by HCMV cannot be easily emulated in animal models.

Therefore, studies of naturally infected humans are the best basis for us to interpret and define the effectiveness of HCMV vaccine. But the hurdle is that we are limited to characterize the immune responses accurately by in vitro methods, for example investigation of epithelial or endothelial tropism deficiency of laboratory strains. To address this issue, a panel of antibodies from human sera exhibiting ≥ 128-fold higher neutralizing potency were measured with respect to their ability to block virus infection in endothelial and epithelial cells, but not tested in fibroblasts during early primary infection.⁴⁵ Results of this study suggest that induction of robust epithelial/endothelial specific neutralizing activities would be necessary for an effective HCMV vaccine.^60,61 In humans, the primary HCMV infection route is infection first of epithelial cells. Then endothelial cells and leukocytes disseminate virus into blood stream, resulting in viral infection of organ and tissue-specific cells.^62,63 As most HCMV infections are transmitted orally, epithelial specific neutralizing antibodies have the potential to block viral transmission by preventing virus entry into mucosal epithelial cells.⁶⁴ This may explain why Towne and gB/MF59 vaccines failed to induce high levels of neutralizing titers against virus epithelia entry to sustain durable protection in congenitally infected women.⁶⁵ On the other hand, virus or immunity vertical transmission between maternal and fetal is known to spread via placenta.^66,67 Studies showed that pregnant women developing antibodies with high avidity early after the onset of infection appeared to be at a lower risk of vertical transmission.⁶⁸ Furthermore, the children born to HCMV seropositive mothers were less likely to develop congenital HCMV disease than those born to mothers with primary HCMV infection.^69,70 To dissect the antibody response to HCMV glycoproteins in transmitter and non-transmitter pregnant women, 23 pregnant women were analyzed for the presence of neutralizing antibodies against different glycoproteins and glycoprotein complexes. The neutralizing antibodies were detected using ARPE-19 cells (human retinal pigment epithelial cells) and HELF cells (human lungs fibroblast cells) in a neutralization assay. This study demonstrated that neutralizing antibodies targeting the pentamer gH/gL/pUL128-131 complex were predominant, and that the early presence of neutralizing antibodies directed to multiple sites on the pentamer was associated with a reduced risk of HCMV vertical transmission.⁷¹

Functional T cells in cellular response

HCMV infections are characterized by a dynamic, life-long interaction in which host immune responses, particularly of T cells, restrain viral replication and prevent disease but do not eliminate the virus or preclude transmission. From a study of cytokine flow cytometry screening, scientists found that 151 HCMV ORFs were immunogenic for CD4⁺ or CD8⁺ T cells, and that ORF immunogenicity was only modestly influenced by ORF expression, kinetics and function. They also reported that total HCMV-specific T cell responses in seropositive subjects were enormous, comprising on average 10% of both the CD4⁺ and CD8⁺ memory compartments in blood.³³ HCMV antigen-specific T cell responses involving both CD4⁺ and CD8⁺ T cells were further confirmed in HCMV seronegative vaccine recipients for clinical protective efficacy.⁷² Studies in MCMV model revealed that the adoptive transfer of murine CMV specific CD8⁺ cytotoxic T cells to immunodeficient mice conferred protection from MCMV disease.^73,74 Further research supported this approach, showing that the recovery of CD4⁺ and CD8⁺ HCMV specific T cell responses in BMT (bone marrow transplant) patients who were HCMV seropositive was strongly correlated with protection from HCMV disease.^75–77 To investigate the therapeutic application of HCMV specific T cell lines, Hermann and colleagues adoptively transferred donor-derived HCMV-specific T cell lines into 8 stem cell transplant recipients lacking HCMV-specific T cell proliferation. They found that at a median of 11 days after transfer, T cells proliferation were detected in 6 of them, a significant increase of HCMV-specific CD4⁺ T cells in 5 patients. At a median of 13 days, 1.12 to 41 HCMV specific CD8⁺ T cells/μL blood were detected after transfer. In conclusion, their results demonstrated that anti-HCMV cellular therapy represents a therapeutic option in viremic patients after stem cell transplantation.⁷⁸ Together, HCMV-specific CD4⁺ and CD8⁺ T cells are the dominant compartments for HCMV natural infected response or adoptive derived response, also the golden measurement for vaccine efficiency.⁷⁹

Experiences in HMCV vaccine development and related immune responses

There is no experimental vaccine approach with imminent licensure in the pharmaceutical market. There are main reasons proposed for the failure to achieve the goal. First, the immune correlation for HCMV vaccine is not yet established due to deficiency of animal models. The ideal target protein capable of eliciting durable immune responses that closely mimic those seen in HCMV seropositive subjects is not fully characterized. This section will provide the strategy for development of vaccines in preclinical and clinical trials, and immune response induced by these vaccines (Table 1). In summary, there are two categories of vaccine candidates, one involves modification of attenuated or replication-defective virus to maintain safe efficiency in humans. This category is referred as modified virus vaccine (MVV). The second approach is to present targeted viral particles in the form of recombinant protein antigen through a DNA vector. This approach is referred as individual antigen vaccine (IAV).

Live attenuated HCMV vaccines

Early efforts to produce a HCMV vaccine were directed at developing a live, attenuated vaccine that presents all relevant antigens to the immune system in a conformation that is the same as what would be presented during natural HCMV infection. Some examples are the Towne and AD169 viruses, Towne/Toledo chimeric viruses, and dense body (DB) vaccines. The first candidate tested in human was developed from the AD169 laboratory strain by serial passages in fibroblasts, which were isolated from the adenoids of a child. This vaccine was found to be safe and well tolerated when administered in HCMV seronegative adults.^40,80,81 Subsequently, another Towne strain was established, derived from the virus isolated from the urine of an infant with congenital HCMV infection.^82,83 In the initial human trial, 11 HCMV seronegative adults were inoculated with Towne vaccine. None of them developed an augmented antibody response. In another human test, those inoculated developed only mild local reactions two weeks after vaccination. AD169 yielded similar results to those observed in the Towne trials.⁸⁴ The efficacy of Towne vaccine was evaluated in a series of studies in recipients of renal transplants.⁸⁵ Each study indicated that although Towne failed to prevent HCMV infection,^86,87 it did provide a degree of protection comparable with that conferred by natural infection.^88–90 Towne was also assessed in a placebo controlled study involving seronegative mothers who had children in daycare. It failed to protect women from HCMV infection transmitted from their children. In contrast, women with preexisting naturally infected immunity were protected from re-infection with a new strain being transmitted in the daycare.¹⁷ Furthermore, it was shown that Towne was not capable of establishing persistent or latent infection in humans.⁹¹ The inability of Towne vaccine to convey protection in high risk settings might be due to over-attenuation of non-invasive vaccines. To test this possibility, chimeric viruses between Towne and Toledo strains were constructed in the late 1990s and tested for safety in HCMV seropositive volunteers.^92,93 Researchers realized that the more important explanation is its genomic difference from clinical isolates. The failure of Towne to elicit neutralizing antibodies comparable to those induced by natural infection,^94–96 the fact that Towne is insufficiently immunogenic, is due to the lack of pentameric gH complex in its composition and the related to the loss of ability to infect endothelial or epithelial cells.

Subunit vaccines

Subunit vaccines are usually combined with an adjuvant to present defined viral antigens in the form of recombinant protein or delivered as a DNA vaccine or through a viral vector, then attempt to engender immune response recognized by specific immunogenic viral proteins to protect against infection or disease.⁹⁷ In clinical application, subunit vaccines focus primarily on gB, pp65, and IE1 as individual antigen vaccines.

Subunit glycoprotein B vaccine

The most fully characterized HCMV protein is the glycoprotein complex I (gCI) consisting of gB (UL55), which is based on Towne strain sequence and expressed in Chinese hamster ovary (CHO) cell as secreted protein. Reports assert that this protein is the dominant target of virus neutralizing antibody response in natural infection.^98–100 Formulated with an oil-in-water adjuvant, MF59, the vaccine was well tolerated and capable of inducing gB-specific antibodies in HCMV seronegative subjects.¹⁰¹ After administration of the third dose of gB/MF59 vaccine, the levels of gB-specific antibodies and neutralizing activity exceeded those seen in HCMV seropositive controls.¹⁰² The immunogenicity and safety of gB/MF59 vaccine has also been studied in a limited number of toddlers.¹⁰³ A Phase II trial was implemented with gB/MF59 vaccine in cohort of HCMV seronegative, postpartum women, who are at high risk of primary HCMV infection following delivery.¹⁰⁴ Unfortunately, these trials finally failed; the antibody response and neutralizing titers were transient, and returned to baseline within several months of vaccine immunization.⁷⁹ In contrast to Towne vaccine, gB/MF59 vaccine could boost gB-specific antibodies and T cells in HCMV seropositive volunteers,¹⁰⁵ but immune sera of the gB vaccine did not show any level of neutralizing titers against viral infection in epithelial cells.⁶⁵ To identify whether gB is the dominant antigen in the natural immunity to HCMV, German scientists isolated antibody repertoire against gB from different seropositive individuals in unbiased fashion. Their study revealed that most of the anti-gB antibodies produced during infection failed to neutralize cell-free virus, and most neutralizing antibodies were found to bind to epitopes that were not located within the previously characterized antigenic domains (AD) of gB.¹⁰⁶ This study will be useful tool to provide an immunogenic map of important proteins, showing that gB is not the prominent antigenic particle for neutralizing antibodies. Depletion of gB-specific antibodies in human polyclonal IgG would not affect its neutralizing activity in epithelial cells.³⁷ However, it is an essential viral protein that is involved in the early events of infection.

Subunit vectored vaccine

The purpose of the subunit vectored vaccine approach is to induce cellular and humoral response in a safe way. Protein pp65 (UL83) was defined as a significant target antigen for CD8⁺ class I major histocompatibility complex (MHC)-restricted HCMV-specific cytotoxic T lymphocytes (CTL). Protein pp65 was employed as a subunit vectored vaccine to develop HCMV-specific T cell response.^107,108 The TransVax vaccine consisted of plasmid DNAs encoding pp65 and gB formulated with poloxamer adjuvant. Clinical studies showed that TransVax was well-tolerated and significantly reduced viremia and reoccurrence in the HSCT setting. The reported safety and efficacy outcomes supported further development in a phase III trial, notwithstanding a lack of significant reduction in the use of HCMV-specific antiviral therapy compared with placebo in this phase II trial.¹⁰⁹ HCMV IE1 gene is also an important target of the CD8⁺ T cell response to HCMV infection.¹¹⁰ A trivalent DNA vaccine targeting gB, pp65, and IE1 was under evaluation in a Phase I trial involving a total of forty healthy adult subjects.^111,112

Poxvirus vectors and alphavirus vectors have both been studied for their potential as HCMV vaccine candidates to stimulate CTL responses. Another formulation of gB vaccine evaluated in clinical trial is based on canarypox vector ALVAC, an attenuated poxvirus that replicates abortively in mammalian cells. ALVAC-gB vaccine was modestly immunogenic. The vaccine was usually boosted with a live, attenuated Towne vaccine or gB/MF59, being otherwise unable to develop binding and neutralizing antibody titers comparable to those observed in naturally seropositive subjects.^113,114 Although these concomitant vaccine approaches induced high titer antibodies and lymphoproliferative response, it appears to be more like an augmented gB-specific response, with no benefit detected for prime-boost strategy.

Novel strategy facing the immunological challenge

Advances in studies of natural immunity inform development of strategy and definition for future vaccine candidates in the clinic. An effective HCMV vaccine should be designed to induce broad neutralizing antibodies capable of protecting a variety of cell types and balanced CD4⁺ and CD8⁺ T cell responses, ideally comparable to those seen in naturally infected individuals. There are two main points that should be taken into consideration. First, the immunogenic target for both arms of adaptive immune responses should be incorporated into vaccine formation. Second, we should choose an appropriate way to establish and express vaccine candidates challenging modalities for development.

Primary target in HCMV vaccine design

Prior knowledge of neutralizing antibodies of natural immunity to HCMV are limited to those targeting gB, gM/gN, and gH/gL/gO complexes, which are essential for virus entry in fibroblast cells.⁴⁹ More recent studies revealed that gB and the gH/gL dimer are necessary for fibroblast entry. In contrast, entry into endothelial or epithelial cells requires three additional viral proteins, UL128, UL130, and UL131, which interact with gH/gL to form a pentameric complex.^{43,47,48,60,115} Researchers measured the neutralizing activities of HCMV seropositive sera during pregnancy on fibroblasts and epithelial/endothelial cells. Results showed that the neutralizing antibody response measured in epithelial or endothelial cells was potent, occurred very early, and was directed mostly against combinations of two or three gene products of the UL131-128 locus. On the contrary, neutralizing antibody measured in fibroblasts appeared late, was relatively weak, and was directed against gH and gB.¹¹⁶

To identify the neutralizing component of CMVIG, serial depletions of CMVIG on cell-surface-expressed HCMV antigens as well as purified antigens was studied. The results indicated that major neutralizing antibodies for epithelial cell entry in CMVIG were directed against the gH/gL/pUL128/pUL130/pUL131 complex. The gH/gL antibodies in CMVIG were shown to have a dominant role in inhibition of viral entry into fibroblasts; anti-gB antibodies played little role in CMVIG neutralization.³⁷ This conclusion was also supported by the clinical data associated with the Towne vaccine and the gB/MF59subunit vaccine, which induced epithelial entry-specific neutralizing activities that were on average 28-fold (Towne) and 15-fold (gB/MF59) lower than those observed in natural infection. These results suggest that HCMV vaccine efficacy may be enhanced by induction of epithelial entry-specific neutralizing antibodies.⁶⁵

Subsequent studies further proved that the most potent neutralizing epitopes were target the pentameric gH complex. Macagno and coworkers isolated 27 human monoclonal antibodies with extraordinary high potency in neutralization from immortalized memory B cells of HCMV-immune donors. Only one of them recognized conserved epitope of UL128, all others recognized conformational epitopes that required expression of two or more proteins of the gH/gL/pUL128-131 complex.¹¹⁷ In another report, Potzcsch and coworkers isolated over 600 gB-specific antibodies from six human donors and found the vast majority of these antibodies had no neutralizing activity.¹⁰⁶ Tong-ming Fu and his group isolated a panel of 45 monoclonal antibodies from a rabbit immunized with an experimental vaccine virus in which the expression of the pentameric gH complex was restored. Their studies revealed that over one-half (25 of 45) of the antibodies had high titer neutralizing activity. These results clearly suggest that the pentameric gH complex is the primary target for potent neutralization in humoral immunity to natural infection.¹¹⁸ Recently, this group developed a whole-virus vaccine candidate from the live attenuated AD169 strain, with genetic modifications to improve its immunogenicity and attenuation. This vaccine restored the expression of the pentameric gH/gL/pUL128-131 protein complex, a major target for neutralizing antibodies in natural immunity.¹¹⁹

Alternative expression options in vaccine design

HCMV vaccine candidates are generally grouped into two categories. In one category are modified virus vaccines, which were designed as live attenuated viruses, include Towne and AD169 viruses and Towne/Toledo chimeric viruses. In the other category are dense body (DB) vaccines consisting of defined quantities of noninfectious virus particles. Individual antigen vaccines are designed to deliver immunogenic antigen to induce durable immune responses. Delivery is achieved through a viral or DNA vector such as is used for subunit vaccines gB/MF59, pp65, and IE1. Modified virus vaccines must be safe for human use,⁶⁰ and a unique manufacturing process would have to be developed. In contrast to modified virus vaccines, individual antigen vaccines are developed based on optimized antigens which potentially combine the immunogenicity of a killed vaccine with safety or replication-defective viral vectors (e.g., pox, adenovirus, alphavirus, and others) expressing subunits or multi-subunit complexes. Because of the defined target, the dose potency and quality could be detected in a straightforward manner. The correlation between immunogenicity, protection, and vaccine would be easier to define and standardize. The key remaining question is whether the antigen composition in individual antigen vaccines is sufficient for protection and prevention.

As previously described, pentameric gH complex is unique for epithelial or endothelial cell entry. It also has a potent ability to generate neutralizing antibodies either in the form of virus vaccine or in live virus during natural infection. Glycoprotein gB is required for virus entry, cell-cell spread, and driving fusion of the viral envelope with the target cell membrane.¹²⁰ The answer to whether more immunogenic forms of virus are effective for vaccine development remains elusive. Considering this, the modified virus vaccine approach is preferred, as it targets both arms of the adaptive immune response. The incorporation of the pentameric gH complex into vaccines will improve the vaccine efficacy, while other forms of the virus could be retained to maintain virus integrity and to be able to mimic the immune response observed in naturally infected individuals. This approach was achieved using an experimental vaccine, an epithelial tropism revertant virus of AD169 origin in which the pentameric gH complex was restored.^50,119 Use of this experimental vaccine in an animal model elicited high levels of neutralizing antibodies titers, significantly better than those achieved by attenuated AD169/Towne vaccine and recombinant gB vaccines. The neutralizing titers after revertant virus vaccine immunization were comparable to those seen in HCMV-seropositive human sera.¹¹⁸

Lessons from herpesvirus family vaccines

HCMV is a member of the herpesvirus family. The development of other herpesvirus family vaccines provides us clues suggesting that the vaccine design should be based on improved understanding of natural immunity. We may focus determination of whether a candidate vaccine is capable of eliciting neutralizing antibodies for disease prevention and treatment and what antigens are important for eliciting immune responses from protective antibodies and diverse and polyfunctional T cells. Based on this strategy, a promising herpes zoster subunit vaccine, Shingrix, was developed, with vaccine efficacy against Herpes Zoster (HZ) 97.2% for all age groups,¹²¹ which consists of varicella zoster virus (VZV) glycoprotein E and AS01B adjuvant was assessed in Phase III. Of noted, the combination of Shingrix antigen containing T and B cell epitopes and an adjuvant was able to stimulate both VZV-specific antibody and T cells and restore age-related decline in T cell immunity. Besides Shingrix, the most studied subunit herpesvirus vaccines are herpes simplex virus type 2 (HSV-2) vaccines. However, to date, no HSV-2 vaccines have conferred protective immune responses against infections in clinical trials.¹²² The combination of HSV-2 gD and gB with MF59, Chiron vaccine, was shown to be immunogenic in humans, but it was ineffective in the treatment of recurrent genital herpes.¹²³ Both antibodies and T cells can contribute to control of viral replication in the genital tract. This study suggests that rather than focusing on induction of only antibodies or T cell responses, vaccines that lead to combined immune responses against HSV-2 may have increased efficacy. Hence, live attenuated vaccines may have a distinct advantage over subunit and inactivated vaccines, primarily because replication of the pathogen allows for the entire repertoire of pathogen-specific antigen expression. The licensed vaccines Varivax¹²⁴ and Zostavax¹²⁵ are classic examples. Varivax is a vaccine for varicella and Zostavax for zoster. These live attenuated vaccines have been used extensively in prophylactic and therapeutic approaches to combat primary and recurrent zoster infection. For HCMV vaccine, the whole-virus vaccine candidate from the live attenuated AD169 strain, with genetic modifications to improve its immunogenicity and attenuation, might also be a more practical vaccine design approach.¹¹⁹

Conclusions

Although several experimental HCMV vaccines have advanced to clinical evaluation (Table 1), a major barrier remains in the progress of vaccine development: the optimal HCMV vaccine strategy must be suitable for the population of potential patients. However, better understanding of the correlates of protective immunity to natural HCMV infection has been achieved recently. In particular, the correlates of protective immunity involved in epithelial/endothelial cell entry have been elucidated as areas for future vaccine strategy. The subunit expression of the gH/gL/pUL128-131 proteins would be exploited in vaccine design, which also merits consideration for vaccine efficiency evaluation. The studies on preclinical and clinical analysis of vaccine candidates will substantially accelerate the pace of successful vaccine emergence and solve this public health urgency.

Abbreviations

AD	=	antigenic domains
ARPE-19 cells	=	human retinal pigment epithelial cells
BMT	=	bone marrow transplant
CHO	=	Chinese hamster ovary
CMVIG	=	HCMV hyperimmune globulin
CTL	=	cytotoxic T lymphocytes
DB	=	dense body
EIA	=	enzyme immunoassay
HCMV	=	human cytomegalovirus
HELF	=	cellshuman lungs fibroblast cells
HSCT	=	hematopoietic stem cell transplant
HSV-2	=	herpes simplex virus type 2
HZ	=	herpes zoster
IAV	=	individual antigen vaccine
LLOD	=	lower limit of detection
MCMV	=	murine cytomegalovirus
MHC	=	major histocompatibility complex
MVV	=	modified virus vaccine
ORFs	=	open reading frames
R−/D+	=	a HCMV seronegative recipient with SOT from a seropositive donor
SOT	=	solid organ transplants
VEE	=	Venezuelan equine encephalitis virus
VRP	=	alphavirus VEE replicon particles
VZV	=	varicella zoster virus

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr. Georgina Salazar at the University of Texas Health Science Center at Houston for her careful and critical reading of the manuscript.

Funding

This work was supported in part by the Welch Foundation (Grant AU-0042-20030616 to Z.A.) and National Natural Science Foundation of China Grant 31670927 (W.L.).