Characterization of Plasma Immunoglobulin G Responses in Elite Neutralizers of Human Cytomegalovirus

Abstract

Background

Human cytomegalovirus (HCMV) is the most common infectious complication of organ transplantation and cause of birth defects worldwide. There are limited therapeutic options and no licensed vaccine to prevent HCMV infection or disease. To inform development of HCMV antibody-based interventions, a previous study identified individuals with potent and broad plasma HCMV-neutralizing activity, termed elite neutralizers (ENs), from a cohort of HCMV-seropositive (SP) blood donors. However, the specificities and functions of plasma antibodies associated with EN status remained undefined.

Methods

We sought to determine the plasma antibody specificities, breadth, and Fc-mediated antibody effector functions associated with the most potent HCMV-neutralizing responses in plasma from ENs (n = 25) relative to that from SP donors (n = 19). We measured antibody binding against various HCMV strains and glycoprotein targets and evaluated Fc-mediated effector functions, antibody-dependent cellular cytotoxicity (ADCC), and antibody-dependent cellular phagocytosis (ADCP).

Results

We demonstrate that ENs have elevated immunoglobulin G binding responses against multiple viral glycoproteins, relative to SP donors. Our study also revealed potent HCMV-specific antibody-dependent cellular cytotoxicity and antibody-dependent cellular phagocytosis activity of plasma from ENs.

Conclusions

We conclude that antibody responses against multiple glycoprotein specificities may be needed to achieve potent plasma neutralization and that potently HCMV elite-neutralizing plasma antibodies can also mediate polyfunctional responses.

Elite neutralizers of human cytomegalovirus (HCMV) exhibit elevated immunoglobulin G binding responses against multiple viral glycoproteins as well as potent HCMV-specific antibody-dependent cellular cytotoxicity and antibody-dependent cellular phagocytosis responses, demonstrating that HCMV elite-neutralizing plasma antibodies can mediate polyfunctional responses.

Human cytomegalovirus (HCMV) is a ubiquitous β-herpesvirus that persists as a lifelong infection in human hosts [1–3]. While primary HCMV infection is usually asymptomatic in healthy people, it is the most common viral cause of birth defects worldwide and can cause life-threatening complications in immunocompromised individuals [1, 4–6]. Despite ongoing research efforts, no licensed vaccine exists to prevent HCMV infection, and current antiviral treatment options remain suboptimal due to long-term toxicity.

Cellular immunity is considered the main factor responsible for controlling HCMV infection, but there is evidence that humoral immunity also serves a critical role in protection [7–9]. Neutralizing antibodies are thought to be especially important for preventing viral entry into host cells because they bind cell-free virus and block interactions between viral glycoproteins and host receptors [10]. More recently, nonneutralizing effector functions, such as antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent cellular cytotoxicity (ADCC), have been highlighted as potentially protective against HCMV infection [11, 12].

Glycoprotein B (gB) is a central focus of HCMV vaccine efforts because of its high abundance, immunogenicity, and critical role in facilitating viral entry into host cells. Five distinct genotypes of gB have been described in the literature to date, with approximately 95% conserved amino acid identity [13]. The 2 main sites of sequence variation between gB genotypes are the furin cleavage site and gB antigenic domain 2 (AD2) [14, 15]. Within AD2, there are 2 binding sites; site 1 (AD2S1) is highly conserved and targeted by neutralizing antibodies and site 2 is a variable, nonneutralizing site [14].

While gB is known to induce a strong antibody response, the most potently neutralizing monoclonal antibodies isolated from HCMV-infected individuals target the pentameric complex (PC) [16, 17]. The PC is required for viral entry into epithelial, endothelial, and myeloid lineage cells, but not for entry into fibroblasts [18]. A dimer of glycoproteins H and L (gH/gL) exists as part of the PC but also associates with other glycoproteins, including glycoprotein O (gO) and gB [19, 20]. Antibodies that are specific for the gH/gL and gH/gL/gO complexes can block HCMV infection in vitro, suggesting that they may also be protective [21]. It is unknown, however, whether individuals with potent HCMV-neutralizing responses have a dominant neutralizing B-cell lineage response against a single glycoprotein complex, like those that develop against neutralizing epitopes on the human immunodeficiency virus (HIV) envelope in HIV-infected individuals with broad neutralizing activity [22], or a combined response against multiple glycoprotein specificities.

Importantly, high titers of HCMV-specific antibodies elicited by natural infection alone are insufficient for preventing reinfection and transmission. HCMV hyperimmune globulin (HIG) is a concentrated antibody product containing pooled plasma from HCMV-seropositive adults with high-binding HCMV-specific immunoglobulin (Ig) G [16, 23]. HCMV-HIG has been evaluated for its ability to prevent congenital transmission and infection but has not been established as efficacious in randomized clinical trials [24–27]. Yet the most efficacious HCMV vaccine to date (gB/MF59) provided approximately 50% protection against HCMV acquisition in adolescent and postpartum women, as well as a reduction in the duration of HCMV viremia and treatment with antiviral therapies in transplant recipients [6, 28–30]. This vaccine elicited high gB-binding IgG titers and nonneutralizing responses, such as virion phagocytosis, but elicited poor neutralization responses and antibodies with limited ability to neutralize heterologous HCMV strains [11, 31]. Moreover, IgG binding to gB expressed on the cell surface was identified as a potential immune correlate of protection associated with the risk of infection across 2 gB/MF59 vaccine trials, indicating the importance of glycoprotein conformation in effective antibody immunity [32]. There is considerable interest in advancing our understanding of protective anti-HCMV antibody specificities in natural infection to guide vaccine and therapeutic antibody development.

With the goal of improving antibody-based therapeutics, a previous study identified “elite neutralizers” (ENs) of HCMV from seropositive (SP) blood donors [33]. Plasma from these individuals was selected based on significantly enhanced capacity to neutralize HCMV in a strain-independent manner against infection of fibroblasts and endothelial cells, relative to HCMV-HIG and plasma from other SP individuals [33]. The differences in HCMV neutralization capacity between ENs and SP individuals prompted our investigation into the anti-HCMV antibody glycoprotein binding strength, breadth, specificities, and ability to mediate nonneutralizing effector functions in individuals with the most potent neutralizing responses in plasma. The goal of our study was to identify the specificities and characteristics of plasma antibody responses associated with elite-neutralizing capacity to guide HCMV vaccine design and the development of more efficacious therapeutic antibody products.

METHODS

Study Population

We obtained 57 plasma samples from previously defined HCMV EN [33, 34] (n = 25), non-EN SP (n = 19), and HCMV-seronegative (n = 13) deidentified blood donors. Samples were provided by C. S. from Ulm University in Germany (Duke University Institutional Review Board; Pro00105640) and were previously obtained from the German Red Cross Blood-Transfusion Service, Baden-Württemberg and Hessen, with informed consent (Ethical Board of Ulm University; vote no. 53/14) [33].

HCMV Neutralization Assay

The 50% neutralizing titer was determined for EN and SP donors against 7 strains of HCMV, as described elsewhere [33]. HCMV Towne, AD169, TB40/E, VHL/E, Merlin, VR1814, and Toledo strains were used in this assay [34].

Binding Antibody Multiplex Assay

Antibody binding to HCMV antigens (gB domains I, II, and I+II, gB ectodomain, PC, gH/gL, and gH/gL/gO) were measured by means of multiplex enzyme-linked immunosorbent assay (ELISA), as described elsewhere [32, 35, 36]. Antigens were covalently coupled to fluorescent polystyrene beads (Luminex) and incubated with plasma samples. HCMV-specific IgG was detected using phycoerythrin (PE)–conjugated goat anti-human IgG secondary antibody (2 µg/mL; Southern Biotech). Results were acquired on a Bio-Plex 200 system (Bio-Rad) and reported as mean fluorescence intensity.

IgG Binding and Avidity ELISAs

Plasma IgG and IgM binding to whole HCMV virions, 5 gB genotypes (gB1, gB2, gB3, gB4, and gB5), and gB AD2S1 were measured by ELISA, as described elsewhere [36]. Briefly, 384-well plates were coated with virus at an optimized concentration of plaque-forming units (PFUs) per well (TB40/E, 100 PFUs; AD169r, 2700 PFUs; Towne, 360 PFUs) or with 2 μg/mL of protein per well. Data were collected via SpectraMax plate reader (Molecular Devices) and reported as the area under the curve because full sigmoidal curves were not achieved by all samples. Avidity was reported as the relative avidity index, calculated as the ratio of the optical density at 450 nm of wells treated for 5 minutes with 7-mol/L urea to that of paired wells treated with 1× phosphate-buffered saline.

gB-Transfected Cell Binding Assay

Plasma IgG binding to gB expressed on the cell surface was measured as described elsewhere [32]. Human embryonic kidney–293T cells were cotransfected with DNA plasmids expressing green fluorescent protein (GFP) and gB open reading frame (HCMV Towne) and incubated at 37°C with diluted plasma samples. Cells were stained with Live/Dead Fixable Near-IR Dead Cell Stain, followed by PE-conjugated goat-anti-human IgG Fc, and fixed with 10% formalin before acquisition on the flow cytometer (BD) using the high-throughput sampler. The frequency of PE⁺ cells was reported for each sample based on the live, singlet, GFP⁺ population.

Whole HCMV Virion Phagocytosis (ADCP)

AD169r virions (1 × 10⁶ PFUs) were conjugated to AF647 NHS ester before incubation with diluted plasma samples (1:100). Virus-antibody immune complexes were then added to THP-1 cells for spinoculation and incubation. Cells were stained with Aqua Live/Dead, fixed with 10% formalin, and washed before acquisition on the flow cytometer (BD) using the high-throughput sampler. The percentage of AF647⁺ cells was reported for each sample based on the live, singlet population.

Natural Killer Cell CD107a Degranulation ADCC Assay

Cell-surface expression of CD107a was used as a marker for natural killer (NK) cell degranulation [37, 38], similar to what was described elsewhere [11, 39]. Live primary human NK cells were added to wells containing AD169-derivative BadrUL131-Y4-GFP–infected MRC-5 cell monolayers. Diluted plasma samples were added with brefeldin A (GolgiPlug, 1 μL/mL; BD), monensin (GolgiStop, 4 μL/6 mL; BD), and anti-CD107a–fluorescein isothiocyanate (clone H4A3; BD). After a 6-hour incubation, NK cells were washed and stained with a viability dye, anti–CD56-PE/cyanine 7 (clone NCAM16.2; BD), and anti-CD16-PacBlue (clone 3G8; BD) [11]. The frequencies of CD107a⁺ live NK cells were determined by flow cytometry. Final data represent specific activity, determined by subtraction of nonspecific activity observed in assays performed with mock-infected cells.

Statistical Analysis

The statistical analysis plan was created before analysis. Raw data were internally reviewed before analysis, based on established quality control criteria for each assay. Differences in the antibody-binding responses between groups were analyzed using Fisher exact or Wilcoxon rank sum tests. For analyses where a cutoff was established, data below the cutoff (calculated as the mean of the HCMV-seronegative donors plus 3 standard deviations) was set to the cutoff before analysis. For the gB-binding breadth analysis, established cutoffs were used to determine binding for each sample to each genotype. Differences in the number of gB genotypes bound per sample between EN and SP groups were determined by means of Fisher exact test.

In the analysis of HCMV neutralization, we first log-transformed the data and fit a Tobit regression model [40, 41] (a censored linear regression model allowing for data at lower limit of detection) while adjusting for HCMV-specific IgG to assess differences between EN and SP groups. For ADCP, we logit-transformed the data before analysis. Binding responses were converted to antibody concentration (in micrograms per milliliter) using GraphPad Prism software version 9.4.1. Plasma IgG binding responses were then normalized to HCMV-specific IgG concentration, and Wilcoxon rank sum tests were performed to assess differences between EN and SP groups. Multiple testing correction was performed per set of analysis using Benjamini-Hochberg false discovery rate (FDR) adjustment [42]. Statistical significance was predefined at P < .05, with an FDR-adjusted P value <.2, reflecting the hypothesis-generating nature of this study [32, 43]. All statistical analyses were performed using R software.

RESULTS

Plasma IgG and IgM Binding Responses to HCMV in EN and SP Donors

To determine differences between antibody responses against HCMV in EN and SP donors, plasma IgG binding to TB40/E, AD169r, and Towne whole virus was measured with ELISA. These viruses were selected based on gB genotype and differences in PC expression (listed in Table 1). Plasma from EN donors exhibited significantly higher IgG binding to all tested strains, relative to plasma from SP controls (all P < .001) (Figure 1A). Next, we explored whether EN and SP individuals differed in terms of time since primary infection or reinfection by measuring HCMV-specific plasma IgG avidity and IgM binding responses. ENs exhibited a significant increase in IgG avidity for AD169r (P = .004), but not for TB40/E (P = .09) or Towne (P = .07), relative to SP donors (Figure 1B). In addition, while the proportion of ENs with detectable plasma TB40E-specific IgM was higher than that of SP donors (78.6% vs 21.4%, respectively), this difference was not statistically significant, as determined by Fisher exact test (FDR-adjusted P value = .07) (Figure 1C). We also determined that the IgM response magnitude is marginally correlated with HCMV-specific IgG binding magnitude but not with neutralization (Supplementary Figure 1).

Table 1.

Human Cytomegalovirus (HCMV) Neutralization Potency After Adjustment for HCMV-Specific Immunoglobulin G Concentration^a

HCMV Strain	gB Genotypeb	PC	Log Differencec	P Value	FDR-Adjusted P Value
Toledo	gB3	…	−3.64	<.001	<.001
TB40/E	gB1	Intact	−3.47	<.001	<.001
VR1814	gB3	Intact	−3.66	<.001	<.001
VHL-E	gB1	Intact	−1.87	<.001	<.001
Towne	gB1	…	−1.25	<.001	<.001
AD169	gB2	…	−2.29	<.001	<.001
Merlin	gB1	Intact	−1.74	<.001	<.001

Abbreviations: FDR, false discovery rate; gB, glycoprotein B; HCMV, human cytomegalovirus; PC, pentameric complex;

^aResults from the Tobit regression are shown in log scale as non–elite neutralizing seropositive controls compared with elite neutralizers. Statistical significance was defined as P < .05 and a multiple-testing-adjusted P value < .2.

^bGenotype descriptions are available elsewhere [13, 36].

^cEstimates of the log difference between groups, adjusting for immunoglobulin G concentration if the outcome was uncensored.

Figure 1.

Plasma immunoglobulin (Ig) G and IgM binding responses to whole human cytomegalovirus (HCMV) virions in samples from HCMV elite neutralizer (EN; n = 25), HMCV-seropositive (SP; n = 19), and HCMV-seronegative (SN; n = 13) blood donors. A, Area under the curve (AUC) for IgG binding to TB40/E (circles; median AUC, 10.21 for EN and 6.89 for SP donors), AD169r (squares; 6.02 and 2.77, respectively), and Towne (triangles; 6.22 and 3.35) whole virus, as measured by enzyme-linked immunosorbent assay (ELISA). Median AUCs are indicated by horizontal lines. B, Relative IgG binding avidity index (RAI) to TB40/E (median RAI, 97.3 for EN vs 93.5 for SP donors; P = .09), AD169r (79.0 vs 71.6, respectively; P = .004), and Towne (84.9 vs 85.8; P = .07) whole virus, as measured by ELISA. Median RAIs are indicated by horizontal lines. C, Optical density at 450 nm (OD₄₅₀) showing plasma from EN and SP donors with TB40/E-specific IgM binding. All samples were assayed in duplicate. Wilcoxon rank sum test (A, B) and Fisher exact test (C) were used to calculate statistical differences between groups; statistical significance was defined as P < .05 and a multiple-testing–adjusted P value <.2. All reported P values are false discovery rate adjusted. Dotted lines represent the negative cutoff, determined by the average SN binding response plus 3 times the standard deviation.

Plasma IgG Binding to HCMV Surface Glycoproteins and Peptides in EN and SP Donors

To investigate plasma IgG binding responses to known HCMV glycoprotein targets of anti-HCMV antibodies, we used a binding antibody multiplex assay. Plasma IgG from ENs bound with significantly higher magnitudes to all glycoprotein complexes tested (all P < .001), including gH/gL dimer, gH/gL/gO trimeric complex, PC, and gB ectodomain, relative to plasma IgG from SP donors (Figure 2A and 2B). A similar trend was observed for plasma IgG binding to gB domains among ENs, including gB domain I, domain II, domains I + II, cell-associated gB, and gB AD2S1, relative to SP donors (all P < .001) (Figure 2B-D).

Figure 2.

Human cytomegalovirus (HCMV) glycoprotein conformational and linear epitope immunoglobulin (Ig) G binding in samples from HCMV elite neutralizer (EN; n = 25), HCMV-seropositive (SP; n = 19), and HCMV-seronegative (SN; n = 13) blood donors. A, Mean fluorescence intensity (MFI) of plasma IgG binding to HCMV surface glycoproteins, including gH/gL (median log₁₀ MFI, 3.02 for EN and 1.53 for SP donors), gH/gL/gO (3.97 and 2.29, respectively), and PC (3.92 and 2.33). B, MFI of plasma IgG binding to HCMV glycoprotein B (gB) domains, including gB domain I (median log₁₀ MFI, 1.64 for EN and 0.763 for SP donors), gB domain II (3.79 and 2.33, respectively), gB domains I + II (3.44 and 1.73), and gB ectodomain (4.01 and 2.61). C, Frequency of plasma IgG binding to cell-associated gB (median percentage of binding, 26.4% for EN and 3.38% for SP donors). D, Area under curve (AUC) for plasma IgG binding to known neutralizing epitope, gB antigenic domain 2, site 1 (AD2S1) (median AUC, 7.30 for EN and 2.18 for SP donors). Samples were assayed in duplicate. Wilcoxon rank sum test was used to calculate statistical differences between groups; statistical significance was defined as P < .05 and a multiple-testing–adjusted P value <.2. All reported P values are false discovery rate adjusted. Dotted lines indicate the negative cutoff, determined by the average SN binding response plus 3 times the standard deviation.

Plasma IgG Binding Magnitude and Breadth Across HCMV gB Genotypes in EN and SP Donors

HCMV gB variants are classified into 5 common gB genotypes (gB1–gB5). Previous studies have reported elevated risk of HCMV disease among individuals with multi–gB genotype HCMV infections, demonstrating potential implications for future vaccine candidates that incorporate multiple gB genotypes, especially given the moderate efficacy of a single-valent gB protein antigen vaccine [6, 13, 29, 30, 44]. Thus, we wanted to determine whether there were detectable differences in the breadth of gB-specific IgG binding responses between plasma from ENs and plasma from SP donors. We define binding breadth as the ability of antibodies from a single donor to bind multiple genotypes of the same glycoprotein antigen; in this case we evaluated plasma IgG binding across 5 discrete gB genotypes by ELISA. ENs exhibited a significantly higher magnitude of binding to all 5 gB genotypes (all P < .001) relative to SP controls (Figure 3A and 3B). We then assessed plasma IgG binding breadth by calculating the number of gB genotypes bound by each plasma sample using established cutoffs. There was no significant difference in IgG binding breadth across gB genotypes between EN and SP groups (P = .07).

Figure 3.

Plasma immunoglobulin (Ig) G binding to multiple glycoprotein B (gB) genotypes in samples from human cytomegalovirus (HCMV) elite neutralizer (n = 25), HCMV-seropositive (n = 19), and HCMV-seronegative (n = 13) blood donors. A, Area under the curve (AUC) for IgG binding to multiple gB genotypes, gB1, gB2, gB3, gB4, and gB5, as measured by enzyme-linked immunosorbent assay. B, Wilcoxon rank sum test was used to test for statistically significant differences between the magnitude of IgG binding to each gB genotype across groups, and Fisher exact test was used to calculate differences between groups in gB binding breadth (no. of gB genotypes with value above the seronegative cutoff). Significance was defined as P < .05 and a multiple-testing–adjusted P value <.2. Abbreviation: FDR, false discovery rate.

Relationship Between Total HCMV–Specific IgG Levels, Neutralization Responses, and Glycoprotein-Specific IgG Binding

Previous work by Falk et al [34] demonstrated that plasma anti–HCMV IgG titer does not determine HCMV neutralizing status. To follow up on this finding in the present cohort, we first interpolated the HCMV-specific plasma IgG concentration for each sample based on the HCMV-HIG standard binding curve against TB40/E. A Tobit regression model was applied to the neutralization data from this cohort to adjust for the impact of HCMV-specific IgG concentration on the neutralization capacity of plasma from EN and SP donors. The model confirmed that plasma neutralization response magnitude against all tested HCMV strains remained associated with EN and SP status (Table 1), after adjustment for total HCMV–specific plasma IgG concentration. This indicates that the quality of the antibody response, and not just the magnitude of the total HCMV–specific IgG response, is important in determining plasma neutralization potency. Next, we normalized the glycoprotein-specific plasma IgG binding responses to the total HCMV–specific IgG concentration and performed a Wilcoxon rank sum test on the normalized outputs. After normalization, the magnitude of plasma HCMV glycoprotein–specific IgG binding remained significantly higher among ENs relative to SP donors for all measured targets except gB genotype 2 (P = .36) (Table 2).

Table 2.

Human Cytomegalovirus (HCMV) Glycoprotein–Specific Immunoglobulin (Ig) G Concentration Normalized to Total HCMV–Specific IgG Concentration

Glycoprotein	Plasma Samples From EN Donors (n = 25)	Plasma Samples From SP Donors (n = 19)	P Valuea
gB ectodomain
Median (range), µg/mL Data missing, no. (%) b	0.0411 (0.0213–0.166) 0 (0)	0.0170 (0.00187–0.0408) 1 (5.3)	<.001 …
gH/gL
Median (range), µg/mL	0.0489 (0.00509–0.135)	0.0145 (0.00304–0.673)	.005
Data missing, no. (%) b	0 (0)	6 (31.6)	…
gH/gL/gO
Median (range), µg/mL	0.0428 (0.00556–0.139)	0.0160 (0.00192–0.0519)	.002
Data missing, no. (%) b	0 (0)	7 (36.8)	…
PC
Median (range), µg/mL	0.0545 (0.00787–0.108)	0.0148 (0.00987–0.0411)	<.001
Data missing, no. (%) b	0 (0)	8 (42.1)	…
gB protein, median (range), µg/mL
gB AD2 site 1	0.810 (0.00346–12.9)	0.0814 (0.0114–1.44)	.008
gB1	0.850 (0.116–1.99)	0.575 (0.196–1.26)	.04
gB2	0.846 (0.160–1.76)	0.565 (0.184–1.88)	.36
gB3	4.28 (1.60–14.8)	2.13 (0.815–4.22)	<.001
gB4	3.65 (1.85–10.3)	2.21 (0.944–3.43)	<.001
gB5	0.914 (0.262–1.88)	0.461 (0.190–0.955)	<.001

Abbreviations: AD2, antigenic domain 2; EN, human cytomegalovirus (HCMV) elite neutralizer; gB, glycoprotein B; gH, glycoprotein H; gL, glycoprotein L; gO, glycoprotein O; PC, pentameric complex; SP, HCMV-seropositive.

^a P value from Wilcoxon rank sum test.

^bData are missing for cases where the IgG concentration could not be accurately interpolated off of the standard curve.

Investigation of Nonneutralizing Antibody Effector Functions in Plasma from EN and SP Donors

Aside from neutralizing responses, Fc-mediated antibody effector functions also potentially contribute to protection from HCMV infection [39]. Thus, we compared the ability of plasma from EN and SP donors to promote HCMV-specific ADCP and ADCC in flow-based assays using whole HCMV virions and HCMV-infected fibroblasts, respectively. Using a linear regression model, we determined that plasma from ENs exhibited significantly enhanced ability to promote ADCP of the whole HCMV virion and ADCC responses against infected cells, relative to SP donors (P < .001 and P = .01, respectively) (Figure 4A and 4B). However, after adjustment for plasma total HCMV–specific IgG concentration in the linear regression model, the ADCC and ADCP responses did not differ between EN and SP groups (Table 3).

Figure 4.

Plasma immunoglobulin (Ig) G nonneutralizing effector function activity in samples from human cytomegalovirus (HCMV) elite neutralizer (EN; n = 25), seropositive (SP; n = 19), and seronegative (SN; n = 5) blood donors. A, The percentage of AF647⁺ THP-1 cells was measured to quantify antibody-mediated phagocytosis of TB40/E (median percentage of antibody-dependent cellular phagocytosis (ADCP), 75.1% for EN and 42.1% for SP donors), AD169r (56.3% and 34.2%, respectively), and Towne (78.3% and 55.8%) whole HCMV virions. B, The percentage of CD107a⁺ natural killer (NK) cells was measured to quantify plasma antibody-mediated cellular cytotoxicity (ADCC) against HCMV-infected cells (median percentage of CD107a⁺ NK cells, 13.7% for EN and 8.65% for SP donors).

Table 3.

Nonneutralizing Human Cytomegalovirus (HCMV) Antibody Responses After Adjustment for HCMV-Specific Immunoglobulin G Concentration^a

HCMV Strain	gB Genotypeb	PC	Log Differencec	P Value
TB40/E	gB1	Intact	−0.306	.23
AD169r	gB2	Repaired	−0.158	.51
Towne	gB1	…	0.959	.74

Abbreviations: gB, glycoprotein B; HCMV, human cytomegalovirus; PC, pentameric complex.

^aResults from linear regression after adjustment for total HCMV–specific immunoglobulin (Ig) G concentration are shown in log scale as non–elite neutralizing seropositive controls compared with elite neutralizers. Statistical significance was defined as P < .05 and a multiple-testing–adjusted P value <.2.

^bGenotype descriptions are available elsewhere [13].

^cEstimates of the log difference between groups with adjustment for IgG concentration.

DISCUSSION

In the current study, we found that plasma from ENs can mediate potent HCMV-specific polyfunctional responses, including neutralizing and nonneutralizing effector functions. We observed significantly elevated HCMV-specific IgG binding responses to 3 strains of HCMV and higher avidity for AD169r in plasma from ENs, relative to that from SP donors. Moreover, we found that plasma from ENs exhibited significantly higher magnitudes of IgG binding to all tested HCMV glycoproteins known to be targets of neutralizing antibodies, namely, gB, PC, gH/gL, and gH/gL/gO. These findings suggest that robust IgG responses to multiple glycoproteins may be critical to achieving potently neutralizing antibody responses against HCMV, as opposed to single-specificity IgG potency driving EN status. This is notably distinct from what is reported in studies of potent HIV-specific neutralizing responses in HIV-infected individuals, wherein certain plasma IgG epitope specificities, such as the HIV envelope CD4 binding site, are dominant in ENs [22, 45]. Thus, a vaccine that presents multiple HCMV glycoprotein immunogens, either as distinct protein or encoded antigens or in the context of a virion, may be required to elicit the most potent plasma neutralization across cell types that could contribute to protection against HCMV infection.

We observed that plasma IgG from ENs bound with significantly higher magnitudes to gB domain II and AD2S1, known neutralizing epitopes, and to gB domain I, which is targeted by neutralizing and nonneutralizing antibodies [31, 46]. Relative to plasma from SP donors, plasma from ENs also exhibited significantly increased IgG binding magnitudes across the 5 gB genotypes. However, there were no significant differences in IgG binding breadth across gB genotypes between groups. After normalizing HCMV glycoprotein–specific IgG concentrations by total HCMV–specific IgG concentration, we observed that plasma from ENs maintained significantly elevated binding responses against all targets, except for gB genotype 2, relative to plasma from SP donors. Moreover, the differences observed between these groups in binding to cell-associated gB could indicate that plasma from ENs contains antibodies specific for unique epitopes of gB that are exposed when gB is associated with the cell surface, which was previously identified as a correlate of protection in the gB/MF59 vaccine trials [32]. Overall, our findings indicate that incorporating multiple gB genotypes into future vaccine formulations may not be necessary to confer broad and potent neutralizing responses to HCMV, whereas including multiple glycoprotein antigens with appropriate conformations to allow antibody access to neutralizing epitopes could have more impact on elicitation of broad and potent neutralizing responses.

Finally, we observed that plasma from ENs exhibits significantly elevated neutralization against HCMV relative to plasma from SP donors, even after adjustment for total HCMV–specific IgG concentration. Thus, neutralizing status does not appear to be determined solely by the magnitude of the HCMV-specific antibody titer but rather by the quality and specificities of HCMV-specific antibodies generated in response to infection. However, while plasma from ENs was also capable of mediating significantly higher magnitudes of ADCP and ADCC relative to plasma from SP donors, these responses did not distinguish between groups after adjustment for HCMV-specific IgG in a linear regression model. This suggests that unlike the distinct quality of HCMV-neutralizing responses between EN and SP individuals, differences in ADCP and ADCC response magnitudes between groups are due to the higher HCMV-specific IgG concentration in plasma from ENs, not a distinction in antibody response quality.

The current study has several limitations. First, the sample size is relatively small, and all plasma samples were selected from deidentified blood donors at a single time point. This study was designed as a cross-sectional study, so our results cannot definitively confirm acute versus chronic infection within the cohort. Interestingly, while the proportion of HCMV-specific IgM responses was higher in ENs, the difference between groups was not statistically significant. We cannot rule out the possibility that IgM responses due to a recent primary HCMV infection or reinfection contribute to EN status. We observed that IgM binding is marginally correlated with IgG binding responses (Supplementary Figure 1). However, HCMV-specific IgM levels in plasma are not highly correlated with neutralization responses, suggesting that they are not a major contributor to plasma neutralization in individuals with EN status (Supplementary Figure 1). Therefore, the higher neutralization capacity observed in plasma from EN donors is likely not solely attributable to a recent primary HCMV infection. This aligns with previous findings that elite neutralization capacity remains stable over several years in EN individuals [33]. In addition, owing to limited sample volume, we were unable to assess the relative contributions of domain-specific IgG responses toward the potent and broad neutralizing capacity of plasma from ENs through glycoprotein-specific IgG depletion studies. This project also did not investigate the ability of plasma antibodies in ENs to prevent cell-cell spread of HCMV. Further investigation is required to understand these aspects of plasma IgG responses in ENs.

This study represents the first characterization of antibody responses to multiple HCMV surface glycoproteins, including multiple gB glycoproteins and conformations, elicited by natural infection that contribute to elite neutralization of HCMV. Overall, our data reveal that plasma or monoclonal antibodies from EN individuals can mediate strong polyfunctional antibody responses against HCMV, suggesting that the potency and efficacy of currently available HCMV-antibody therapeutics, such as HCMV-HIG, may be improved by selecting for EN plasma donors. HCMV is capable of infecting various cell types via distinct entry mechanisms and uses multiple glycoprotein complexes to do so, making it highly unlikely that targeting a single epitope or even a single glycoprotein will be sufficient to elicit a protective response across multiple strains of HCMV. Thus, our results further support the development of vaccines that incorporate multiple glycoprotein complexes needed for infection of multiple cell types to elicit broad and potent neutralizing antibodies and polyfunctional antibody responses against HCMV.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Jessica Falk, who initiated this collaboration together with S. R. P. at the International Herpesvirus Workshop 2018 in Vancouver, British Columbia. We also acknowledge support from the Biostatistics, Epidemiology and Research Design Methods Core at Duke University.

Financial support. This work was supported by the National Center for Advancing Translational Sciences, National Institutes of Health (grant UL1TR002553 to the Biostatistics, Epidemiology and Research Design Methods Core at Duke University), the Translating Duke Health Initiative (award to R. B. and C. C.), the Duke Center for Human Systems Immunology, the Else Kröner-Fresenius Foundation (award 2016-A126 to C. S.), a Medearis CMV Scholar Award from Duke University School of Medicine (to M. J. H.), and the National Institute of Allergy and Infectious Diseases, National Institutes of Health (grant 3P01AI129859 to S. R. P. and T32 training grant 2T32AI052077-16A1 to M. J. H.).