Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Nov 4;12(568):eabb3611.
doi: 10.1126/scitranslmed.abb3611.

Antibody binding to native cytomegalovirus glycoprotein B predicts efficacy of the gB/MF59 vaccine in humans

Jennifer A Jenks  1 , Cody S Nelson  1   2 , Hunter K Roark  1 , Matthew L Goodwin  1 , Robert F Pass  3 , David I Bernstein  4 , Emmanuel B Walter  1 , Kathryn M Edwards  5 , Dai Wang  6 , Tong-Ming Fu  7 , Zhiqiang An  7 , Cliburn Chan  8 , Sallie R Permar  9   10
Affiliations

Antibody binding to native cytomegalovirus glycoprotein B predicts efficacy of the gB/MF59 vaccine in humans

Jennifer A Jenks et al. Sci Transl Med. .

Abstract

Human cytomegalovirus (CMV) is the most common infectious cause of infant brain damage and posttransplant complications worldwide. Despite the high global burden of disease, vaccine development to prevent infection remains hampered by challenges in generating protective immunity. The most efficacious CMV vaccine candidate tested to date is a soluble glycoprotein B (gB) subunit vaccine with MF59 adjuvant (gB/MF59), which achieved 50% protection in multiple historical phase 2 clinical trials. The vaccine-elicited immune responses that conferred this protection have remained unclear. We investigated the humoral immune correlates of protection from CMV acquisition in populations of CMV-seronegative adolescent and postpartum women who received the gB/MF59 vaccine. We found that gB/MF59 immunization elicited distinct CMV-specific immunoglobulin G (IgG)-binding profiles and IgG-mediated functional responses in adolescent and postpartum vaccinees, with heterologous CMV strain neutralization observed primarily in adolescent vaccinees. Using penalized multiple logistic regression analysis, we determined that protection against primary CMV infection in both cohorts was associated with serum IgG binding to gB present on a cell surface but not binding to the soluble vaccine antigen, suggesting that IgG binding to cell-associated gB is an immune correlate of vaccine efficacy. Supporting this, we identified gB-specific monoclonal antibodies that differentially recognized soluble or cell-associated gB, revealing that there are structural differences in cell-associated and soluble gB are relevant to the generation of protective immunity. Our results highlight the importance of the native, cell-associated gB conformation in future CMV vaccine design.

PubMed Disclaimer

Conflict of interest statement

Competing interests: The authors declare no competing interests.

Figures

Fig. 1.
Fig. 1.. Sera from adolescent and postpartum vaccine cohorts exhibited distinct CMV and gB-specific IgG binding responses.
(A) Log area under the curve (Log AUC) for IgG binding to full-length gB by enzyme-linked immunosorbent assay (ELISA) in AV and PV. Log AUC median is indicated (n = 39 AV, n = 33 PV). (B) Relative IgG binding avidity to gB by ELISA, calculated as the ratio of antibody binding in the presence or absence of urea. Log AUC median is indicated (n = 39 AV, n = 33 PV). (C) Frequency of IgG binding to gB-transfected HEK293T cells. HEK293T cells were cotransfected with separate plasmids encoding gB and GFP. Cells positive for IgG binding were detected by flow cytometry and were defined as the % of live singlet cells positive for both GFP and anti-human IgG Fc PE. Dotted line indicates the threshold for detection of gB-transfected cell IgG binding, defined as 1% using the average binding of CMV-seronegative sera (n = 11) + 2x the standard deviation. Median % binding to gB-transfected cells is indicated (n = 39 AV, n = 32 PV). (D) Log mean fluorescence intensity (MFI) of IgG binding to full-length gB, gB ectodomain, AD-1, AD-2, AD-4, AD-5, and AD-4 + AD-5 as detected by binding antigen multiplex assay (BAMA). Dotted line indicates the threshold for positivity, defined by the average of CMV-seronegative samples (n = 28) + 2x the standard deviation. Log MFI median is indicated (n = 39 AV, n = 33 PV). (E) Log AUC for IgG binding to TB40/E strain whole virus by ELISA. LogAUC median (n = 39 AV, n = 33 PV). (F) Relative IgG binding avidity to whole TB40/E virions by ELISA, calculated as the ratio of antibody binding in the presence or absence of urea. Log AUC median is indicated (n = 39 AV, n = 33 PV). For all assays, samples were run in duplicate, and statistical significance was determined by Mann Whitney U test. n.s., not significant; red triangles, adolescent vaccinees (AV); blue circles, postpartum vaccinees (PV).
Fig. 2.
Fig. 2.. Properties of vaccine-elicited IgG Fc regions were distinct between the adolescent and postpartum cohorts.
(A) Log MFI for gB binding by IgG1, IgG2, IgG3, and IgG4 subclass antibodies, as measured by BAMA. Log MFI median is indicated (n = 39 AV, n = 33 PV). (B) Log MFI for gB-specific IgG engagement by FcγR2a (H131), FcγR1a, FcγR2b, and FcγR3a (V158) as measured by BAMA. Log MFI median is indicated (n = 39 AV, n = 33 PV). All samples were run in duplicate, and dotted lines indicate the thresholds for positivity, defined as 100 MFI. Statistical significance was determined by Mann Whitney U test; those lacking P values were determined not significantly different. Red triangles, adolescent vaccinees (AV); blue circles, postpartum vacinees (PV).
Fig. 3.
Fig. 3.. Vaccine-elicited functional antibody responses were distinct between the adolescent and postpartum cohorts.
(A) Antibody-dependent cellular phagocytosis (ADCP) by THP-1 monocytes of fluorescently labeled AD169r-GFP strain or TB40/E strain virus. Dotted line indicates the threshold for positivity, as defined by the average of CMV-seronegative samples run simultaneously (n = 6) + 2x the standard deviation. Median % phagocytosis is indicated (n = 39 AV, n = 33 PV). (B) Autologous neutralization of Towne strain virus, reported as the log-transformed dilution that prevented 50% infection of MRC-5 fibroblasts by the virus (LogID50). For all neutralization assays, dotted lines indicate the threshold of detection (starting sera dilution). LogID50 median is indicated (n = 39 AV, n = 33 PV). (C) Neutralization of TB40/E strain virus on BJ5T-a fibroblasts in the presence or absence of rabbit complement. LogID50 median is indicated (n = 39 AV, n = 33 PV). (D) Neutralization of TB40/E strain virus on ARPE epithelial cells in the presence or absence of rabbit complement. LogID50 median is indicated (n = 39 AV, n = 33 PV). (E) Neutralization of AD169r-GFP strain virus on BJ5T-a fibroblasts in the presence or absence of rabbit complement. LogID50 median is indicated (n = 39 AV, n = 33 PV). (F) Neutralization of AD169r-GFP strain virus on ARPE epithelial cells in the presence or absence of rabbit complement. LogID50 median is indicated (n = 39 AV, n = 33 PV). For all assays, samples were run in duplicate. Statistical significance was determined by Mann Whitney U test. n.s., not significant; red triangles, adolescent vaccinees (AV); blue circles, postpartum vacinees (PV).
Fig. 4.
Fig. 4.. Principal component analysis (PCA) was used to visualize humoral immunogenicity variables by vaccination cohort and outcome.
Principal components 1 and 2 (PC1 and PC2), along which the antibody immune responses have the largest variance (n = 39 AV, n = 33 PV, including n = 24 infected, n = 48 uninfected vaccinees).
Fig. 5.
Fig. 5.. Immune correlates of protection of primary CMV acquisition elicited by the gB/MF59 vaccine.
(A) Frequency of IgG binding to gB-transfected HEK293T cells by flow cytometry. Dotted line indicates the threshold for gB-transfected cell IgG binding, as defined by the average of CMV-seronegative samples run simultaneously (n = 11) + 2x the standard deviation. Median % binding to gB-transfected cells is indicated (n = 23 infected vaccinees, n = 48 uninfected vaccinees). Statistical significance was determined by univariate logistic regression analysis with Benjamini-Hochberg correction. (B) Frequency of IgG binding to MRC-5 fibroblasts infected with TB40/E strain virus, measured by flow cytometry. Dotted line indicates the threshold for CMV-infected cell binding by the average of CMV-seronegative samples run simultaneously (n = 8) + 2x the standard deviation. Median % binding to CMV-infected cells is indicated (n = 20 infected vaccinees, n = 44 uninfected vaccinees). Statistical significance was determined by Mann Whitney U test.
Fig. 6.
Fig. 6.. CMV gB-specific mAbs isolated from naturally CMV-infected individuals bind with similar strength to full-length gB yet bind with a bimodal distribution to cell-associated gB.
Binding by gB-specific mAbs (5 μg/mL) to gB-transfected cells, measured by flow cytometry, as compared to binding to full-length gB, measured by ELISA. n = 26 anti-gB mAbs recombinantly produced from gB-specific memory B cells isolated from three naturally infected CMV-seropositive (SP) individuals. Samples were run in duplicate.
Fig. 7.
Fig. 7.. To predict the infection outcome of combined vaccine cohorts, in-sample and out-of-sample receiver operator characteristic curves for a penalized multiple logistic regression model were generated using humoral immunogenicity features.
(A) Logistic regression coefficients from a LASSO (least absolute shrinkage and selection operator) regression analysis. LASSO regression analysis was performed to generate the feature coefficients and was run after exclusion of heterologous neutralization results and collinear variables [variance inflation factor (VIF) > 10]. (B) Frequency of feature selection by LASSO techniques, which was applied to 1000 bootstrapped samples to evaluate the robustness of feature selection. Stability selection plot indicates the number of times that each feature had non-zero coefficients from performing LASSO on each bootstrapped sample. (C) Receiver operator characteristic (ROC) curve for in-sample data and the corresponding area under the curve (AUC), with 1 indicating perfect discriminatory value and 0.5 or less indicating no discriminatory value. (D) ROC curves and the corresponding AUCs for out-of-sample data. Out-of-sample results were obtained by 3-fold stratified cross-validation, including a nested cross-validation within each fold for selection of the inverse regularization strength hyperparameter. ROC ± 1 SD are displayed.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Manicklal S, Emery VC, Lazzarotto T, Boppana SB, Gupta RK, The “silent” global burden of congenital cytomegalovirus. Clin Microbiol Rev 26, 86–102 (2013). - PMC - PubMed
    1. Ramanan P, Razonable RR, Cytomegalovirus infections in solid organ transplantation: a review. Infect Chemother 45, 260–271 (2013). - PMC - PubMed
    1. Johnson J, Anderson B, Pass RF, Prevention of maternal and congenital cytomegalovirus infection. Clin Obstet Gynecol 55, 521–530 (2012). - PMC - PubMed
    1. Fowler KB et al., The outcome of congenital cytomegalovirus infection in relation to maternal antibody status. N Engl J Med 326, 663–667 (1992). - PubMed
    1. Boppana SB, Pass RF, Britt WJ, Stagno S, Alford CA, Symptomatic congenital cytomegalovirus infection: neonatal morbidity and mortality. Pediatr Infect Dis J 11, 93–99 (1992). - PubMed

Publication types

LinkOut - more resources