Human Cytomegalovirus mRNA-1647 Vaccine Candidate Elicits Potent and Broad Neutralization and Higher Antibody-Dependent Cellular Cytotoxicity Responses Than the gB/MF59 Vaccine

Abstract

Background

MF59-adjuvanted gB subunit (gB/MF59) vaccine demonstrated approximately 50% efficacy against human cytomegalovirus (HCMV) acquisition in multiple clinical trials, suggesting that efforts to improve this vaccine design might yield a vaccine suitable for licensure.

Methods

A messenger RNA (mRNA)–based vaccine candidate encoding HCMV gB and pentameric complex (PC), mRNA-1647, is currently in late-stage efficacy trials. However, its immunogenicity has not been compared to the partially effective gB/MF59 vaccine. We assessed neutralizing and Fc-mediated immunoglobulin G (IgG) effector antibody responses induced by mRNA-1647 in both HCMV-seropositive and -seronegative vaccinees from a first-in-human clinical trial through 1 year following third vaccination using a systems serology approach. Furthermore, we compared peak anti-gB antibody responses in seronegative mRNA-1647 vaccinees to that of seronegative gB/MF59 vaccine recipients.

Results

mRNA-1647 vaccination elicited and boosted HCMV-specific IgG responses in seronegative and seropositive vaccinees, respectively, including neutralizing and Fc-mediated effector antibody responses. gB-specific IgG responses were lower than PC-specific IgG responses. gB-specific IgG and antibody-dependent cellular phagocytosis responses were lower than those elicited by gB/MF59. However, mRNA-1647 elicited higher neutralization and antibody-dependent cellular cytotoxicity (ADCC) responses.

Conclusions

Overall, mRNA-1647 vaccination induced polyfunctional and durable HCMV-specific antibody responses, with lower gB-specific IgG responses but higher neutralization and ADCC responses compared to the gB/MF59 vaccine.

Clinical Trials Registration

NCT03382405 (mRNA-1647) and NCT00133497 (gB/MF59).

human cytomegalovirus, mRNA-1647 vaccine, pentametric complex, durability, gB/MF59 vaccine

Human cytomegalovirus (HCMV), a ubiquitous herpesvirus, leads to life-threatening diseases in newborns infected in utero [1] and in immunocompromised individuals [2], with severe symptoms such as enteritis, pneumonitis, and neurologic deficits [1, 2], highlighting the need for an effective vaccine to prevent infection and disease [3, 4]. Viral attachment is mediated by major envelope glycoprotein complexes on the virion surface [3, 5], including glycoprotein B (gB), the primary viral fusion protein [6], and gH/gL/UL128/UL130/UL131A pentametric complex (PC), which is essential for entry into epithelial, endothelial, dendritic, and monocytic cells [7]. Both gB and PC are viable targets for antibody-based anti-cytomegalovirus (CMV) interventions [3, 8].

Vaccine development for HCMV has been challenging despite decades of effort, and the gB/MF59 vaccine co-developed by Sanofi and Novartis is the most efficacious HCMV vaccine tested to date, with approximately 50% protection in phase 2 trials in several populations [9–12]. In this experimental vaccine, the Towne gB antigen (gB1 genotype) [13] is transmembrane truncated and furin cleavage site mutated to enhance antigen expression and secretion and then formulated with MF59, an oil-in-water emulsion adjuvant [9–12]. The approximately 50% efficacy achieved was not high enough to advance this candidate to late-stage clinical trials; therefore, efforts have focused on developing a better vaccine that elicits broader protective responses. Interestingly, the gB/MF59 vaccine achieved moderate protection with only limited neutralizing antibody (NAb) responses, suggesting a role for non-NAb responses in preventing HCMV acquisition [9, 14, 15]. However, inducing broad and potent NAb responses might also enhance vaccine efficacy [3].

Frequent reinfection and reactivation in HCMV-exposed individuals represent another challenge to preventing and controlling disease [16]. Vaccination may enhance immunity to prevent reinfection or end-organ disease in affected populations [3]. Although the precise immune correlates of protection against HCMV acquisition, congenital HCMV (cCMV) transmission, or end-organ disease have yet to be established, there is growing evidence that Fc-mediated effector antibody functions are important in preventing cCMV transmission [17, 18]. Adding antigens that induce potent NAb and Fc-mediated effector responses might elicit polyfunctional humoral responses that improve vaccine efficacy [3, 8].

The modified messenger RNA (mRNA) encapsulated in lipid nanoparticle (LNP) vaccine platform has remarkable advantages for both manufacturing ease and safe induction of effective immunity in humans [3, 19]. Thus, Moderna developed an HCMV vaccine candidate, mRNA-1647, employing nucleoside-modified mRNAs encoding wild-type HCMV gB and PC encapsulated in LNP. gB and PC sequences are based on Merlin strain with gB1 genotype (Supplementary Figures 3 and 4) [13, 20]. Preclinical studies demonstrated that mRNA-1647 induced durable and functional antibody responses in mice and nonhuman primates (NHPs) [20, 21]. In this report, we comprehensively assessed the polyfunctional humoral responses elicited by mRNA-1647 in HCMV-seronegative and -seropositive vaccinees and compared the gB-specific immunoglobulin G (IgG) responses with those induced by the partially effective gB/MF59 vaccine.

METHODS

Study Participants and Design

Information about study participants for the mRNA-1647 vaccine cohort is summarized in Table 1 and Supplementary Figure 1. Both mRNA-1647 (phase 1, NCT03382405) and gB/MF59 (phase 2, NCT00133497) vaccines were delivered via intramuscular injection in a 3-dose series at month 0, month 2 (mRNA-1647) or 1 (gB/MF59), and month 6 (Supplementary Figure 2A and 2B). To evaluate durability, we obtained sera samples 12 and 18 months postrecruitment for a partial group of donors (Supplementary Figure 2A; Table 1). Information about study design (Supplementary Figure 2) and products (Supplementary Figures 3 and 4) is detailed in the Supplementary Materials. Institutional review board (IRB) approval was obtained for the gB/MF59 cohort from the Cincinnati Children's Hospital Medical Center, and samples were provided by the National Institute of Allergy and Infectious Diseases Vaccine Treatment and Evaluation Unit (grant number 5R21 AI136556). For the mRNA-1647 study conducted by Moderna, the study protocol, amendments, and informed consent form were reviewed and approved by the Advarra IRB. Written informed consent was obtained from all participants prior to study recruitment and procedures. The study protocol adhered to all applicable national, state, and local laws or regulations, the principles of the International Council for Harmonisation harmonized tripartite guideline E6(R2): Good Clinical Practice, and the Declaration of Helsinki. All mRNA-1647 samples transferred to Duke University Medical School and Weill Cornell Medicine from Moderna were de-identified; the analysis of these were exempt from human volunteer research designation by the IRBs of Duke and Weill Cornell Medicine.

Table 1.

Vaccine Participants in the mRNA-1647 Vaccine Cohort and the Timepoints Evaluated for Humoral Responses in the Current Report

Participant ID	Age Range, y	Sex	Serostatus at Baseline (No. of Participants	Timepoints Analyzed for Humoral Immune Responses
				Month 0	Month 7	Month 12	Month 18
SP-01	16–20	F	Seropositive	Y	Y	Y	Y
SP-02	46–50	M	Seropositive	Y	Y	Y	Y
SP-03	36–40	M	Seropositive	Y	Y	Y	ND
SP-04	41–45	M	Seropositive	Y	Y	Y	ND
SP-05	36–40	F	Seropositive	Y	Y	Y	Y
SP-06	26–30	F	Seropositive	Y	Y	Y	Y
SN-01	41–45	M	Seronegative	Y	Y	Y	ND
SN-02	26–30	F	Seronegative	Y	Y	Y	ND
SN-03	21–25	F	Seronegative	Y	Y	Y	Y
SN-04	36–40	F	Seronegative	Y	Y	Y	Y
SN-05	26–30	M	Seronegative	Y	Y	Y	Y
SN-06	41–45	F	Seronegative	Y	Y	Y	Y
SN-07	16–20	M	Seronegative	Y	Y	Y	ND
SN-08	21–25	F	Seronegative	Y	Y	Y	ND
SN-09	31–35	M	Seronegative	Y	Y	Y	ND
SN-10	21–25	F	Seronegative	Y	Y	Y	ND
SN-11	26–30	F	Seronegative	Y	Y	Y	ND
SN-12	26–30	F	Seronegative	Y	Y	ND	ND
SN-13	46–50	F	Seronegative	Y	Y	Y	ND
SN-14	46–50	F	Seronegative	Y	Y	Y	ND
SN-15	36–40	F	Seronegative	Y	Y	Y	ND
SN-16	41–45	F	Seronegative	Y	Y	ND	Y
SN-17	26–30	F	Seronegative	Y	Y	Y	Y

Abbreviations: F, female; M, male; ND, not done due to unavailability of samples; Y, sample has been analyzed at this particular timepoint.

IgG Binding Assays

Sera IgG binding to HCMV TB40/E virions, Towne gB, Towne gB antigenic domain-2 site (AD2S1) peptide, Towne AD-6 peptide, and VR1814 PC protein was measured by enzyme-linked immunosorbent assay as previously described [9, 13, 17]. Sera IgG binding to gB expressed on the cell surface was measured by flow cytometry as previously described [9, 13, 17]. IgG and subclasses IgG1, IgG2, IgG3 binding to HCMV antigens (gB, gB AD-1, gB AD4, gB AD5, gB AD4 + AD5, PC), Fcγ receptors (FcγR1A, FcγR2A, FcγR2B, FcγR3A), and Fc neonatal receptor (FcRN) were measured by binding antibody multiplex assay on a Bio-Plex 200 system (Bio-Rad) as described elsewhere [9, 13, 17].

Functional Antibody Assays

Neutralization was measured by employing HCMV immediate early-1 (IE-1) gene expression to quantify reduction in infection in fibroblast HFF-1 or epithelial ARPE-19 cells [9, 13, 17] against PC-deficient Towne virus as well as AD169r. 50% inhibitory dose (ID₅₀s) were calculated using nonlinear regression analysis (Sigmoidal, 4PL) using GraphPad Prism software (Supplementary Figure 5). Antibody-dependent cellular cytotoxicity (ADCC) was measured using cell-surface expression of CD107a as a marker for natural killer (NK) cell degranulation, as described previously [13, 17]. Primary human NK cells were incubated with AD169-derivative BadrUL131-Y4/GFP-infected MRC-5 cell monolayers [22]. The frequencies of CD107a⁺ live NK cells were determined by flow cytometry. For antibody-dependent cellular phagocytosis (ADCP) assays, an optimized amount of AD169r virions (1 × 10³ plaque-forming units/well) was conjugated to AF647 N-hydroxysuccinimide ester before incubation with diluted sera (1:100). Virus–antibody immune complexes were spinoculated and incubated with THP-1 cells and stained with Aqua Live/Dead dye, and the percentage of AF647⁺ cells was reported for each sample based on the live, singlet population.

RESULTS

mRNA-1647 Vaccination Elicited Durable HCMV-Specific IgG Responses

HCMV gB-specific IgG responses were significantly increased above baseline at 7 months postvaccination (peak timepoint) in seronegative vaccinees and elevated in seropositive vaccinees (Figure 1A). Similarly, PC-specific peak IgG responses were also significantly increased at month 7 in seronegative vaccinees and elevated in seropositive vaccinees (Figure 1B). Interestingly, seropositive vaccinees had higher gB-specific IgG responses than PC-specific IgG responses at baseline (Supplementary Figure 6A), though PC-specific IgG was boosted to a greater extent (Supplementary Figure 6). Peak gB-specific IgG responses in seronegative vaccinees were comparable to responses induced by natural infection, whereas PC-specific IgG responses were 10-fold higher than those induced by natural infection (Supplementary Figure 7A and 7B). Peak gB- and PC-specific (Supplementary Figure 8A and 8B) IgG1, IgG2, and IgG3 responses were elicited in seronegative vaccinees, while only IgG1 responses were elevated above baseline in seropositive vaccinees (Supplementary Figures 8A and 8B). gB- and PC-specific (Figure 1A and 1B, respectively) IgG responses persisted through 1 year after the last vaccination in seronegative vaccinees. While IgG responses declined between month 7 and month 12, no significant change was observed in vaccine-elicited IgG levels between months 12 and 18 (Figure 1A and 1B). Interestingly, the relative avidity index (RAI) for gB-specific IgG declined significantly between month 7 and 12 only in seronegative vaccinees (Supplementary Figure 9A). In contrast, PC-specific IgG maintained an equivalent RAI at both month 12 and month 18 relative to month 7 in seronegative and seropositive vaccinees (Supplementary Figure 9B).

Figure 1.

mRNA-1647 vaccine–induced long-lasting antigen-specific immunoglobulin G (IgG) antibody responses in human cytomegalovirus-seropositive and -seronegative vaccinees. The soluble glycoprotein B (gB)–specific (A), pentameric complex (PC)–specific (B), and TB40/E virion-bound (C) IgG antibody titers (effective dose 50 [ED₅₀]) were determined via enzyme-linked immunosorbent assay for both seronegative (left panels) and seropositive (right panels) vaccinees, respectively, at the indicated time points including the 1-year follow-up after last vaccination. The dotted line indicates the limit of detection (ED₅₀ = 30). D, IgG-binding activities to Merlin gB-transfected cells were evaluated for seronegative (left panel) and seropositive (right panel) vaccinees at the indicated time points via flow cytometry–based assay. P values from Wilcoxon matched-pairs signed-rank test are reported. *P < .05, ***P < .001.

Significant increase in HCMV-binding IgG levels against whole virion TB40/E (gB1 genotype) [13] was observed in seronegative, but not seropositive individuals after vaccination (Figure 1C), though this was significantly lower than those elicited by natural infection (Supplementary Figure 7C). Virion-binding IgG responses in seronegative vaccinees declined in magnitude (Figure 1C), but not in avidity (Supplementary Figure 9C), between months 7 and 12 and returned to baseline by month 18. In seropositive vaccinees, there was no change in magnitude or avidity of virion-binding IgG after vaccination (Figure 1C; Supplementary Figure 9C),

IgG binding to gB-transfected cells was previously identified as a correlate of protection against HCMV acquisition in the gB/MF59 vaccine trials [9]. Thus, we measured IgG binding to mRNA-1647 vaccine strain–matched Merlin and heterologous gB1 genotype (Towne) gB-transfected cells for both vaccine populations. IgG responses to cell-associated Merlin and Towne gB were elicited in seronegative vaccinees (Figure 1D; Supplementary Figure 10), and Merlin gB-transfected IgG binding responses remained high 1 year after the last vaccination. Only Towne cell-associated gB responses were significantly increased by vaccination in seropositive vaccinees (Supplementary Figure 10), demonstrating that mRNA-1647 was able to elicit broad IgG responses and boost preexisting responses against the cell-associated conformation of gB.

mRNA-1647 Induced Broad and Long-Lasting NAb Responses

We measured NAb responses against homologous (Towne, gB1) and heterologous (AD169r, gB2) gB genotype strains in fibroblasts and epithelial cells. Low-level fibroblast NAb responses against Towne (Figure 2A) and AD169r (Figure 2B) were elicited by month 7 after vaccination in seronegative vaccinees and were increased above baseline in seropositive vaccinees. In contrast, mRNA-1647 elicited potent and durable epithelial NAb responses against AD169r in seronegative vaccinees (Figure 2C, left panel), >400-fold above the fibroblast neutralization titer (Figure 2C and 2B: median ID₅₀, 40 660 vs 99), at peak timepoint. The magnitude of vaccine-elicited epithelial, but not fibroblast, NAb responses was significantly higher than those induced by natural infection (Supplementary Figure 11A–C). This response was increased by 55-fold compared to preexisting responses in seropositive vaccinees and maintained through 1 year after the final vaccination (Figure 2C, right panel).

Figure 2.

Durable neutralizing antibody responses elicited by the mRNA-1647 vaccine in both seropositive and seronegative vaccine recipients. A, Fibroblast neutralization reactivities against Towne virus in HFF-1 cells were compared before (month 0) and after (month 7) vaccination for both seronegative (left panel) and seropositive (right panel) vaccinees. Similarly, fibroblast neutralization responses against AD169r virus in HFF-1 cells (B) and epithelial neutralization antibody responses against AD169r in ARPE-19 cells (C) were monitored before vaccination (month 0) and months 7, 12, and 18 after vaccination for both seronegative (left panels) and seropositive (right panels) vaccinees, respectively, to assess durability. P values from Wilcoxon matched-pairs signed-rank test are reported. *P < .05, **P < .01, ***P < .001. Abbreviation: ID₅₀, 50% inhibitory dose.

IgG responses against gB AD were detectable against AD4 and AD4 + AD5 (Supplementary Figure 12B and 12D) in seronegative vaccinees at month 7 after mRNA-1647 vaccination, revealing an AD4-dominant IgG response that did not correlate with fibroblast NAb responses. In seropositive vaccinees, anti-AD4, AD4 + AD5, and AD2S1 responses were boosted by vaccination (Supplementary Figure 12B, 12D, and 12E), and these responses also did not correlate with fibroblast NAb responses. However, NAb responses against AD169r correlated strongly with AD4- and AD4 + AD5-specific IgG binding in the total vaccine population (Supplementary Table 1). There was no detectable IgG binding to the newly characterized AD6 domain implicated in cell-to-cell spread [23] (Supplementary Figure 12F).

mRNA-1647 Elicited Fc-Mediated Functional Antibody Responses

Recently, growing evidence supports that Fc-mediated effector antibody functionalities play an essential role in preventing cCMV transmission [17, 18]. Hence, we next evaluated vaccine-induced antigen-specific Fc-mediated IgG responses. In all vaccinees, we found significant increases in both gB- and PC-specific (Figure 3A and 3B) IgG responses that bound to all tested Fc receptors (FcγR1A, FcγR2A, FcγR2B, FcγR3A, and FcRN). Both ADCC and ADCP responses were significantly increased in seronegative vaccinees following vaccination but not in seropositive vaccinees (Figure 3C and 3D). ADCC and ADCP responses in seropositive vaccinees at baseline were significantly higher than peak responses for seronegative vaccines, suggesting that these responses are more potently elicited by infection than vaccination (Figure 3C and 3D). AD169r virion-specific ADCC strongly correlated with all tested gB-specific, but not PC-specific, FcR-binding IgG levels in all vaccinees (Supplementary Table 2; Supplementary Figure 13A and 13B), particularly gB-specific FcγR3A binding (Supplementary Figure 13B, r = 0.87, P < .001), which is expressed on NK cells that mediate ADCC [24]. Similarly, we observed positive correlation between AD169r virion-specific ADCP response and all evaluated gB- and PC-specific IgG FcR-binding responses levels in the total vaccinated population (Supplementary Table 2; Supplementary Figure 13C and 13D), in particular for gB-specific FcγR1A (r = 0.93, P < .001) and FcγR2A (r = 0.93, P < .001) (Supplementary Figure 13D).

Figure 3.

Fc receptor (FcR) binding magnitude and immunoglobulin G (IgG) Fc-portion mediated functionalities induced by the mRNA-1647 vaccine in both seropositive and seronegative vaccinees. Glycoprotein B (gB)–specific (A) and pentameric complex–specific (B) IgG Fc antibody (FcγR1A, FcγR2A [H131], FcγR2B, FcγR3A [V158], and FcRN) responses were plotted for both seronegative and seropositive vaccinees before vaccination (month 0 [M0]) and at the peak timepoint (month 7 [M7]). MFI, mean fluorescense intensity. Vaccine-induced IgG antibody Fc-mediated AD169r virion antibody-dependent cellular cytotoxicity (ADCC) (C) and antibody-dependent cellular phagocytosis (ADCP) (D) responses were assessed for both seronegative (S–) and seropositive (S+) vaccinees. P values from Wilcoxon matched-pairs signed-rank test are reported. Open diamond symbols indicate seronegative samples and open circles indicate seropositive samples. *P < .05, **P < .01, ***P < .001.

mRNA-1647 Induced Higher PC- Than gB-Specific IgG Responses

To evaluate if balanced humoral responses were induced for both antigens encoded by the mRNA-1647 formulation, we compared vaccine-elicited PC- and gB-specific IgG responses. We found significantly higher PC- than gB-specific IgG responses in seronegative vaccinees, including total IgG, IgG1, and IgG3 (Figure 4A) responses. There was a trend toward higher PC- than gB-specific total IgG and IgG1 responses in seropositive vaccinees, but only PC-specific IgG3 responses were significantly higher (Figure 4B).

Figure 4.

mRNA-1647 induced higher pentameric complex (PC)–specific immunoglobulin G (IgG) responses than glycoprotein B (gB)–specific responses in the same formulation. The magnitude of mRNA-1647 vaccine–induced total IgG responses and IgG subclass (IgG1, IgG2 and IgG3) responses were compared between gB-specific and PC-specific responses for both seronegative (A) and seropositive (B) vaccinees at month 7. The total IgG responses were measured by enzyme-linked immunosorbent assay, while the IgG subclass (IgG1, IgG2, and IgG3) responses were detected via binding antibody multiplex assay. P values from Wilcoxon matched-pairs signed-rank test are reported. The level of IgG Fc antibody responses (FcγR1A, FcγR2A, FcγR2B, FcγR3A, and FcRN) were compared among gB and PC antigen components in the same vaccine formulation in both seronegative (C) and seropositive (D) vaccinees in the mRNA-1647 vaccine cohort. P values from nonparametric t tests are reported (Mann–Whitney test). *P < .05, **P < .01, ***P < .001. Abbreviations: ED₅₀, 50% effective dose; gB, glycoprotein B; IgG, immunoglobulin G; MFI, mean fluorescense intensity; PC, pentameric complex.

PC-specific FcR-binding IgG responses were also higher in seronegative vaccinees compared to gB-specific responses (Figure 4C), which was also observed when data were normalized to gB- and PC-specific IgG levels (Supplementary Figure 14) with the exception of FcγR1A binding. Further, PC-specific FcγR3A-binding IgG responses also trended higher than gB-specific responses in seropositive vaccinees (Figure 4D). This difference may be underestimated as the PC-specific, but not gB-specific, responses were at the maximum limit of detection (Figure 4D). Although no significant difference was found for FcγR1A, FcγR2A, and FcγR2B IgG binding between PC-specific and gB-specific IgG responses in seropositive vaccinees, these responses were also at the maximum limit of detection (Figure 4D). Higher PC-specific FcRN-binding IgG responses were elicited by month 7 in seropositive vaccinees (Figure 4D).

mRNA-1647 Elicited Broad and Potent NAb and Higher ADCC Responses, but Lower gB-Specific IgG and ADCP Responses Compared to gB/MF59 in Seronegative Vaccinees

We compared peak gB-specific IgG responses in HCMV-seronegative vaccinees at month 7 postvaccination among mRNA-1647 and gB/MF59 vaccine cohorts. The vaccines were delivered intramuscularly on a similar schedule and included a gB1 genotype antigen. gB/MF59 vaccine induced significantly higher gB-specific IgG responses compared to the mRNA-1647 vaccine, yet with similar avidity (Figure 5A). Vaccine-induced gB-specific IgG subclass responses, including IgG1 and IgG3 (Supplementary Figure 15), were also significantly elevated in the gB/MF59 vaccine cohort, in which total IgG and IgG1 was higher than that induced by natural infection (Figure 5A;Supplementary Figure 15A). Interestingly, ratios of gB-specific IgG2/IgG1 and IgG3/IgG1 were higher in the mRNA-1647 versus gB/MF59 vaccine cohort (Supplementary Figure 15B and 15C), suggesting distinct vaccine-induced IgG subclass patterns. Despite the inclusion of 2 major glycoprotein complexes in mRNA-1647 (gB and PC), gB/MF59 vaccination elicited higher virion-binding IgG against TB40/E in terms of both magnitude and avidity (Figure 5B).

Figure 5.

Comparison of glycoprotein B (gB)–specific peak immunoglobulin G (IgG) binding antibody and neutralizing antibody responses. A, Soluble gB-specific total IgG antibody titers (left panel) and relative avidity index (RAI) (right panel) were compared between the 2 vaccine cohorts. B, TB40/E virion-bound IgG titers (left panel) and RAI (right panel) were also compared between the 2 vaccine cohorts. C, Similarly, Towne (left panel) and Merlin gB (right panel) cell-associated IgG antibody responses were compared among the 2 vaccine cohorts. D, The epithelial neutralizing antibody against AD169r in ARPE-19 cells (left panel) and fibroblast neutralizing antibody against AD169r (middle panel) and Towne (right panel) in HFF-1 cells were compared between the mRNA-1647 and gB/MF59 vaccine cohorts. Additionally, the human cytomegalovirus (HCMV)–infected individuals were also included for the comparison. Open blue circles denote the seropositive samples at baseline from the mRNA-1647 vaccine cohort while open purple circles denote the in-house HCMV-seropositive individuals in some assays to increase the statistical power. However, the statistical analysis indicated that the increased sample number did not change the conclusion. P values from nonparametric t tests are reported (Mann–Whitney test). ***P < .001. Abbreviations: ED₅₀, 50% effective dose; gB, glycoprotein B; ID₅₀, 50% inhibitory dose; IgG, immunoglobulin G; MFI, mean fluorescense intensity; S+, human cytomegalovirus seropositive.

gB/MF59 immunization also elicited higher cell-associated autologous Towne gB-specific IgG responses than mRNA-1647 vaccine (Figure 5C). IgG against cell-associated Merlin gB (Figure 5C) trended higher in the gB/MF59 cohort than that induced by mRNA-1647 [23].

Despite the lower gB-specific IgG responses compared to those elicited by gB/MF59, mRNA-1647 vaccine induced higher heterologous virus (AD169r) NAb responses in both epithelial and fibroblast (Figure 5D) cells, including fibroblast NAb responses against the gB/MF59 homologous Towne virus (Figure 5D, right panel), indicating that mRNA-1647 vaccine–induced PC-specific antibody responses may contribute to the fibroblast-tropic NAb responses [25]. Additionally, no differences in gB AD-specific IgG responses were found among the vaccine cohorts with the exception of responses against the newly reported AD-6 epitope, the target of antibodies that can prevent cell-to-cell spread [23], which were higher in the gB/MF59 vaccine cohort (Supplementary Figure 16AF; Supplementary Table 3).

Our data also demonstrated gB/MF59 vaccine induced higher gB-specific IgG FcγR2A, FcγR2B, FcγR3A, and FcRN-binding responses compared to that of mRNA-1647, yet similar levels of binding to FcγR1A (Figure 6A). When normalized to gB-specific IgG responses, the gB/MF59 vaccinees still had higher FcγR2B, FcγR3A, and FcRN, but lower FcγR1A (Supplementary Figure 17C–E and 17A) binding responses and no difference in FcγR2A (Supplementary Figure 17B). Finally, mRNA-1647 vaccine induced higher ADCC (Figure 6B), yet lower ADCP (Figure 6C) responses compared to the gB/MF59 vaccine. No significant correlation was found between tested FcR-binding and ADCC/ADCP responses in the gB/MF59 cohort (Supplementary Table 4). The humoral response comparison between the vaccine cohorts is summarized in Supplementary Table 3.

Figure 6.

The glycoprotein B (gB)/MF59 vaccine elicited higher peak Fc receptors (FcRs) and antibody-dependent cellular cytotoxicity (ADCP) but lower antibody-dependent cellular cytotoxicity (ADCC) functional antibody responses than the mRNA-1647 vaccine candidate. gB-specific FcγR1A, -2A, -2B, -3A, and FcRN antibody responses were compared between the 2 vaccine cohorts indicative of overall higher-level responses except for FcγR1A response in the gB/MF59 vaccine cohort (A); AD169r virion ADCC (B) and ADCP (C) effector antibody responses were compared between 2 vaccine cohorts, suggesting elevated ADCC responses, but lower ADCP responses were found in the mRNA-1647 cohort. Additionally, the human cytomegalovirus–infected individuals (S+) were also included in the comparison. Open blue circle denotes the seropositive samples at baseline from the mRNA-1647 vaccine. P values from nonparametric t tests are reported (Mann–Whitney test). *P < .05, **P < .01, ***P < .001. Abbreviations: ADCC, antibody-dependent cellular cytotoxicity; ADCP, antibody-dependent cellular phagocytosis; MFI, mean fluorescense intensity; S+, human cytomegalovirus seropositive.

DISCUSSION

In this report, we demonstrate that mRNA-1647 HCMV vaccine induced potent and durable HCMV-specific IgG responses in seronegative participants in a first-in-human vaccine trial, mirroring preclinical immunogenicity results [20, 21]. It also boosted preexisting HCMV-specific IgG responses in seropositive vaccinees. mRNA-1647 elicited broad and potent neutralization as well as Fc-mediated effector functions, including robust ADCC responses. These features support its advancement into the ongoing late-stage clinical trial assessing its efficacy for prevention of HCMV acquisition in seronegative women.

Humoral immune correlates of protection against HCMV acquisition or vertical transmission are still being investigated [8]; hence, broad and polyfunctional humoral immunity is desirable. Interestingly, gB-specific IgG responses in mRNA-1647 vaccinees were considerably lower than that of PC-specific IgG responses and those elicited by the gB/MF59 vaccine, indicating potential room for improvement of the immunogenicity of the gB antigen. The lower magnitude of gB- versus PC-specific IgG responses could in part be due to the lower total amount of gB versus PC mRNA in the vaccine formulation (gB vs PC: 30 μg vs 150 μg), suggesting that reformulation of the vaccine could lead to more balanced responses. Alternative approaches could combine mRNA and protein antigens in a co-delivery strategy, similar to that applied in other vaccine platforms [26, 27]. Of note, PC was recently shown to be dispensable for cCMV transmission in NHPs in an intravenous inoculation model [28], though it could still be necessary for mucosal routes of transmission, the most common route for maternal infection. The impact of the addition of PC to an HCMV vaccine on efficacy is yet unclear but will be informed by the ongoing mRNA-1647 trial. It is also important to note that while the gB/MF59 vaccine elicited high IgG responses to gB, its failure to advance to licensure could in part be due to its poor neutralizing ability. While the mRNA vaccine elicits lower gB-specific IgG responses compared to those against PC and elicited by the gB/MF59 vaccine, it elicited potent, durable, and broad neutralization (Supplementary Table 4) and high-magnitude ADCC responses, both of which may be advantageous for protection against CMV transmission. The high neutralizing responses are likely primarily mediated by PC-specific responses compared to gB as preclinical data in mouse and NHP models suggest, where depletion of PC and gH/gL-specific IgG indicate that UL128–UL131A-specific responses are the mediator of the elicited epithelial cell neutralization [20]. This finding is corroborated by anti-CMV neutralizing monoclonal antibodies isolated from HMCV-infected humans, where anti-UL128-UL131A mAbs had greater potency compared to those against gH/gL in endothelial and epithelial cell neutralization [29]. Interestingly, the anti-gH/gL mAbs also mediated neutralization activity in fibroblasts [29].

While elicited IgG responses against cell-associated Towne gB, the immune correlate of protection in the gB/MF59 vaccine trials [9], was lower in the mRNA-1647 group, IgG binding to cell-associated Merlin gB, which is strain-matched to the mRNA-1647 vaccine, was equivalent among the vaccinees (Figure 5C). Furthermore, ADCC responses elicited by mRNA-1647 were more robust than that of the gB/MF59 vaccine, whereas ADCP responses were lower. Both effector responses have recently been associated with reduced cCMV transmission risk [17, 18]. The lower ADCP responses elicited by mRNA-1647 versus gB/MF59 vaccine could be attributable to the distinct gB-specific FcR binding (Figure 6A; Supplementary Figure 17) or IgG subclass ratios (Supplementary Figure 15B and 15C), but also indicate that PC-specific responses may contribute little to this response. Conversely, the higher ADCC response in mRNA-1647 versus gB/MF59 vaccine indicates that the PC is an important target of ADCC. As Vlahava et al. and other colleagues in the field demonstrated, antibodies against gB are considered poor at recognizing gB on the surface of infected cells [30]. Future analysis of antibody binding to gB or PC in the context of infected cells may help address this question, although our previous work indicated that gB-transfected cell binding predicted anti-gB IgG infected cell binding [9].

The expected differences in gB conformation can potentially explain differences in the vaccine-elicited functional responses; gB/MF59 vaccine has a truncated transmembrane domain and a mutated furin cleavage site [9], while the mRNA-1647 encodes an intact gB protein [20, 21] that may be expressed in the context of a cell membrane. Yet, the gB AD-specific IgG responses were not distinct. One report demonstrated that a Towne ectodomain gB antigen induced better immunogenicity than the full-length gB antigen when delivered in the same viral vector [31]. Distinct vaccinee populations may also be a confounding factor for comparison between the cohorts, as previously described in immunity elicited by gB/MF59 [9] and natural infection [32]. The gB/MF59 vaccinees had a lower age range (12–17 years) than that of mRNA-1647 vaccinees (18–49 years) (Supplementary Figure 1) and all were female, though most vaccinees in the mRNA-1647 cohort were also female (76%) (Table 1).

Durable IgG responses elicited by mRNA-based HCMV vaccine mirrors results of the preclinical models including mice, NHPs [20], and rabbits [33]. Yet, it is critical to assess durability at longer intervals given the mRNA-167 vaccine strategy to vaccinate in adolescence for protection against HCMV through reproductive years. Our observation of humoral immune responses lasting >1 year after mRNA-1647 vaccination may stem from long-lived germinal center reactions upon mRNA-LNP vaccination as observed in severe acute respiratory syndrome coronavirus 2 mRNA-LNP-immunized mice [34] and humans [35]. Yet, due to sample availability, we were unable to compare response durability with that of the gB/MF59 vaccine, which is a limitation of this study.

The immune responses required to prevent cCMV transmission may be distinct from those to prevent acquisition. While prevention of mucosal HCMV acquisition may be a challenge, rapid control of viral replication could be effective in eliminating transmission through vaccine-induced Fc-mediated antibody functions such as ADCC [36], ADCP [37], or T-cell responses [3]. Though T-cell responses were not evaluated here, prior studies demonstrated that the mRNA-1647 vaccine is able to elicit potent cell-mediated immunity [20].

The advantages of mRNA-LNP vaccine, such as rapid development and flexible reformulation, provide an opportunity to iteratively redesign vaccine immunogens. The impact of the addition of PC and lower-magnitude gB-specific IgG responses between the mRNA-1647 and the gB/MF59 vaccines on HCMV acquisition will be informed by the ongoing efficacy trial (NCT05085366), an important benchmark for the HCMV field in defining the immunogenicity requirements of a successful HCMV vaccine.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Author contributions. Investigators were X. H., K. P. K., S. H., S. M. V., J. A. J., H. W., I. G. M., M. C., J. P., E. B. W., K. M. E., D. I. B., and S. R. P. Study design was carried out by X. H. and S. R. P. The trial was managed by J. H., M. K., K. M. E., D. I. B., K. W., and A. C. Data analysis and interpretation were carried out by X. H., K. P. K., S. H., S. M. V., J. A. J., H. W., J. P., C. A., L. M. G., and S. R. P. The manuscript was drafted by X. H. and edited by K. P. K. and S. R. P. All authors had full access to the presented data, provided critical input during manuscript preparation, and approved the manuscript for submission.

Acknowledgments. We thank Ms Carolyn Weinbaum and other members of Dr Sallie Permar's laboratory for their support.

Financial support. This work was supported by Moderna, Inc. (SRA project number OA22BEA4) and the National Institutes of Health (grant number 5R21 AI136556).

Author notes

Presented in part: CMV Conference, Virtual, 28 March–1 April 2022 [ePoster # 7.08]; and International Herpesvirus Workshop, Missoula, Montana, 14–19 July 2023 [Abstract # 141].

X. H. and K. P. K. contributed equally to this work.

Potential conflicts of interest. S. R. P. is a consultant to Moderna, Merck, Pfizer, GSK, Dynavax, and Hoopika CMV vaccine programs and leads sponsored programs with Moderna and Merck. S. R. P. also serves on the board of the National CMV Foundation and as an educator on CMV for Medscape. E. B. W. has received funding support from Pfizer, Moderna, Sequiris, Najit Technologies, and Clinetic for the conduct of clinical trials and clinical research. E. B. W. has served as an advisor to Vaxcyte and consultant to ILiAD biotechnologies. J. H., M. K., K. W., L. P., and A. C. have Moderna company stocks. All other authors report no potential conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.