High-dimensional profiling following third MMR vaccination or mumps virus infection reveals individuals with suboptimal immunity

Abstract

Background

Interindividual variation in vaccine responsiveness can help to identify tailored vaccine strategies for groups with suboptimal responses and potentially higher risk.

Methods

Here we assessed and compared such interindividual variation in humoral and cellular immune responses following a third Mumps-Measles-Rubella (MMR3) immunization and mumps virus infection in young adults with a multiparameter analysis.

Results

We demonstrate that MMR3 boosts (neutralizing) antibody, IgG⁺ B cell and CD4⁺ T cell responses against mumps virus whereas a mumps virus infection induces similar but stronger immune responses, including robust mumps-specific CD8⁺ T cell immunity. Interindividual variation in pre-existing and vaccine-induced immunity unravels subpopulations with low and high mumps-specific responses that are distinct in magnitude, kinetics, longevity and correlation profiles. High-dimensional immunophenotyping subsequently reveals that circulating Th1/Th17 central memory T cells (CD127⁺CD25⁻CD38⁻) correlate with MMR3-induced mumps-specific activated (CD137⁺CD154⁺OX40⁺) CD4⁺ T cell responses, particularly in high-responder groups, while a minor IgG⁺ memory B cell subset (CD38⁺CD21⁺CD95⁺CD24⁻CXCR5⁺CCR6⁺CXCR3⁺) modestly correlates with virus-neutralizing antibody titers. Notably, infected individuals are characterized by elevated frequencies of innate lymphocyte subsets.

Conclusions

Overall, our results indicate that a third MMR immunization boosts humoral and cellular responses in young adults albeit with clear distinct immune response profiles. This may enable the identification and more effective targeting of groups with suboptimal immunity.

Plain language summary

This study examined how young adults’ immune systems respond after receiving a third dose of the Mumps-Measles-Rubella (MMR) vaccine compared to a mumps virus infection. The aim was to understand why some have weaker immune responses and how an additional vaccine dose might boost their protection against disease. Researchers measured antibodies and different types of immune cells in blood samples. Both the third MMR dose and mumps infection increased immunity, although infection led to stronger effects. However, the study also identified groups with higher or lower immune responses to vaccination. Understanding these different immune profiles may help identify individuals who are less protected by vaccination and guide the development of more effective vaccination strategies.

Raeven et al. investigate the immune response to third-round MMR vaccination in comparison to natural mumps virus infection. They find that third-round vaccination boosts mumps-specific responses, though not as strongly as mumps virus infection, and that distinct immune response profiles determine the efficacy of response.

Introduction

Immunization is considered the optimal strategy to prevent most infectious diseases. Most vaccines provide good short-term immunity, and some may even confer long-lasting or possibly lifelong immunity. However, vaccination strategies are also subjected to interindividual variation in immune responses as a result of different factors such as age, genetics, environmental factors and lifestyle^1–3. To assess interindividual variation in immune parameters, its effect on vaccine responsiveness and detection of individuals or groups at risk requires a multiparameter approach⁴ that could be used to optimize and personalize vaccination strategies, such as for the combined mumps-measles-rubella (MMR) vaccine.

The introduction of the live-attenuated MMR vaccine in national immunization programs has led to a high vaccine coverage and a dramatic decrease in the infections caused by these three viruses^5,6. Unfortunately, multiple countries worldwide have also witnessed an increase in mumps cases over the last two decades in individuals, especially young adults, that did receive two childhood MMR vaccinations^7–10. This indicates that the long-term immunity induced by childhood vaccination may not be sufficient in all vaccinees, with multiple plausible underlying explanations such as waning immunity^11,12, mismatch in antigen composition between vaccine and circulating strains¹³ and reduced boosting by natural exposure^14,15. A third MMR immunization, MMR3 hereafter, in this risk group of young adults may boost immunity and consequently decrease virus spread and susceptibility to disease¹⁶. Several studies have demonstrated that MMR3 in young adults is safe and boosts humoral responses^17–21. MMR2 vaccination and natural mumps virus infection also induce T cell^22–25 and B cell^11,26,27 immunity to mumps virus. Moreover, Wang et al. revealed that MMR3 increased mumps-specific B cell responses²⁸. However, more extensive exploration is required to determine mumps-specific cellular immune responses following MMR3, in particular T cell responses, and to assess inter-individual variability as well as the correlation with other aspects of the mumps-specific immune responses.

Here, we first evaluate mumps-specific humoral and cellular (T and B cell) responses in an age- and vaccination-matched cohort of young adults (18–25 years, 2xMMR) following a third MMR immunization. These responses are compared to those induced by a mumps virus infection that may serve as a benchmark for good and long-term immunity²⁹. Subsequently, we identify correlations between serological and cellular immune responses and demonstrate interindividual variation in pre-existing immunity and MMR3 responsiveness, on the basis of which we could classify subpopulations. Finally, we use broad phenotyping high-dimensional flow to profile circulating immune cell subsets over time that characterize either the immune response type after MMR3 or mumps virus infection as well as those distinguishing low- or high-responder groups.

Materials & methods

Cohorts

In the MMR3 study, 147 participants received a third dose of MMR vaccine (MMR VaxPro) in the deltoid muscle of the upper arm^18,30. Inclusion criteria were an age of 18–25 years, past receipt of 2 MMR doses, and no history of mumps. From all 147 participants, 8 mL blood was collected in a tube without anticoagulant to obtain serum for antibody measurements at the same sampling time points, and at 3 years (T3Y) post-MMR3, an additional serum sample was obtained through finger-prick sampling. Additionally, from these 147 participants, 30 individuals were voluntarily selected who donated 80 mL heparinized blood pre (T0), and 1 month (T1M) and 1 year (T1Y) post-MMR3 for cellular immunity measurements. The research protocol was approved by the ethics committee (EudraCT 2016-001104-36 and Netherlands Trial Register [NTR] NTR5911, Registration Date: May 23, 2016) and was performed in accordance with the principles of the Declaration of Helsinki. All participating students provided informed consent before any study handling.

Peripheral blood from six individuals with a laboratory-confirmed mumps virus infection was collected at 1 to 2 months (T1M) and 8 ± 1.5 months post-MVI, labeled as T8M hereafter, after disease onset as part of an observational clinical study VAC-263 (NL37852.094.11). All six individuals received 2 doses of MMR vaccination during childhood and were age-matched (18–25 years) to the MMR3 cohort. Unfortunately, no information on immune parameters prior to infection is available for these individuals.

Serum IgG antibody levels and virus neutralization

An in-house-developed multiplex immunoassay was used to determine IgG antibody levels against mumps, measles and rubella as previously described³¹. Mumps virus–neutralization was determined using a focus-reduction neutralization test (FRNT) using the Jeryl Lynn vaccine strain genotype A as previously described³². Surrogate protective cut-off levels for anti-MuV IgG (IgG concentration > 102 RU/mL) and virus neutralization against the Jeryl Lynn vaccine strain (ND₅₀ titer > 34) were determined previously by Kaaijk et al.¹⁸.

CMV, EBV, and VZV IgG antibody seropositivity

An in-house developed multiplex immunoassay was used to determine antibody levels against CMV, EBV and VZV^31,33. Cutoffs for CMV (≥ 5 RU/ml), EBV (≥ 16 RU/ml) and VZV (1≥ 0.26 IU/ml) were used to determine seropositivity.

Generation (and inactivation of) MuV stocks

MuV vaccine virus (Jeryl Lynn strain) was propagated in Vero cells in DMEM containing 2% FCS. MuV virus was harvested at peak cytopathic effect, and centrifuged (800×g). In order to enrich the virus stock and remove FCS for the B cell ELIspot coating, an additional MuV virus stock was made. For this, cells were cultured in a medium without FCS for the last 48 h. The solution was first centrifuged (10 min, 3000 g) to spin down Vero cell debris. Subsequently, mumps virus was further enriched by ultra-centrifugation (2 h, 4 °C, 141,000 × g)(SW32TI, Beckman Colter) and the pellet containing virus was resuspended in PBS (named UC-enriched MuV hereafter). Supernatant of uninfected Vero cells was used as a mock control after freeze-thawing cycles to lyse cells analogous to MuV-infected cells. Virus and mock stocks were aliquoted and stored at −80 °C until use. Inactivated MuV was prepared by exposing the virus to ultraviolet-C irradiation (UVC) (254 nm, 2 min at 1000 mJ/cm2) (UVC 500, Hoefer) and stored at −80 °C until further use.

T cell ELISpot and activation-induced marker (AIM) assay

Frequencies of mumps-specific interferon-gamma (IFNγ)-producing T cells were determined using a previously developed and described IFNγ ELISpot in which natural killer (NK) cells were depleted from PBMC prior to stimulation with live MuV³⁴.

The activation-induced marker assay (AIM) in combination with intracellular cytokine staining (ICS) was done on PBMCs that were stimulated for 24 h with live MuV JL (MOI (multiplicity of infection) = 2). Subsequently, flow cytometric analysis was performed after staining cells for anti-human CD3, CD4, CD8, CD45RO, CCR7, CD25, CD56, CD69, OX40 and fixable viability stain 780 (Table 1). After fixation and permeabilization, using FoxP3/Transcription Factor Staining Buffer Set (eBioscience, Thermo Fisher Scientific, Waltham, Mass), cells were stained intracellularly for anti-human CD137 and CD154 (Table 1), and cytokines: IL-2, IL-5, IL-10, IL-13, IFNγ and tumor necrosis factor (TNFα). Data were acquired on a FACS Symphony analyzer (BD) and analyzed using FlowJo (V10, Tree Star, Ashland, Ore). In the eventual data analysis, CD69 was excluded because CD69, an early activation marker, proved suboptimal for detecting T cell activation after 24 h stimulation of live MuV in our assay. Next, Boolean gating was used to detect unique single, double and triple populations of activated CD4⁺ T cells. Only activated CD4⁺ T cells that expressed two or three activation markers were further investigated. IL-5, IL-10, and IL-13 were measured but not detected in activated CD4⁺ T cells. CCR7 and CD45RO were used for division in naive (T_N), Central memory (T_CM), Effector memory (T_EM), or T_EMRA phenotype (See Fig S1 for gating strategy T cell AIM assay).

Table 1.

Reagents and software

REAGENT or RESOURCE	SOURCE	IDENTIFIER	Dilution / Conc
T cell AIM-ICS Antibodies
AF700 anti-human CD3 (clone HIT3a)	Biolegend	Cat# 300324	1:50
BUV496 anti-human CD4 (clone SK3)	BD Bioscience	Cat# 612936	1:50
BUV395 anti-human CD8 (clone RPA-T8)	BD Bioscience	Cat# 563795	1:100
PE-Cy7 anti-human CD45RO (clone UCHL1)	BD Bioscience	Cat# 560608	1:100
BV421 anti-human CD69 (clone FN50)	BD Bioscience	Cat# 748764	1:100
BUV805 anti-human OX40 (clone L106)	BD Bioscience	Cat# 749157	1:100
BV605 anti-human CD137 (clone 4B4-1)	Biolegend	Cat# 309822	1:50
PE anti-human CD154 (clone 24-31)	Biolegend	Cat# 310806	1:100
PerCP-Cy5.5 anti-human CCR7 (clone G043H7)	Biolegend	Cat# 353220	1:50
BV785 anti-human IL-2 (clone MQ1-17H12)	Biolegend	Cat# 500348	1:50
APC anti-human IL-5 (clone TRFK5)	BD Bioscience	Cat# 554396	1:50
PE-Dazzle594 anti-human IL-10 (clone JES3-19F1)	Biolegend	Cat# 506812	1:200
APC anti-human IL-13 (clone JES10-5A2)	BD Bioscience	Cat# 561162	1:50
BV711 anti-human TNFα (clone Mab11)	Biolegend	Cat# 502940	1:50
BUV737 anti-human IFNy (clone)	BD Bioscience	Cat# 612845	1:100
Fixable viability stain 780	BD Bioscience	Cat# 565388	1:2000

High-dimensional phenotyping Antibodies
FITC anti-human CD8 (clone RPA-T8)	Biolegend	Cat# 301006	1:800
FITC anti-human IgG (clone G18-145)	BD	Cat# 555786	1:25
PerCP-Cy5.5 anti-human CD14 (clone M5E2)	BD	Cat# 550787	1:100
PerCP-Cy5.5 anti-human CD24 (clone ML5)	Biolegend	Cat# 311116	1:50
RealBlue780 anti-human CD21 (clone B-ly4)	BD	Cat# 755783	1:400
APC anti-human CD127 (clone A019D5)	Biolegend	Cat# 351316	1:50
APC-R700 anti-human CD185 (CXCR5)  (clone RF8B2)	BD	Cat# 565191	1:25
APC-Fire750 CD183 (CXCR3) (clone 1C6/CXCR3)	Biolegend	Cat# 353754	1:25
BV421 anti-human CD279 (PD1) (clone EH12.2H7)	Biolegend	Cat# 351316	1:25
Fixable Viability Dye eFluor506	eBioscience	Cat# 65-0866-14	1:1000
BV605 anti-human CD3 (clone UCHT1)	Biolegend	Cat# 300460	1:100
BV605 anti-human IgD (clone IA6-2)	Biolegend	Cat# 348232	1:100
BV650 anti-human HLA-DR (clone G46-6)	BD	Cat# 564231	1:400
BV711 anti-human CD27 (clone O323)	Biolegend	Cat# 302834	1:50
BV750 anti-human CD194 (CCR4) (clone 1G1)	BD	Cat# 746980	1:25
BV786 anti-human CD16 (clone 3G8)	BD	Cat# 563690	1:800
PE anti-human CCR10 (clone 1B5)	BD	Cat# 563656	1:200
PE-CF594 anti-human CD38 (clone HIT2)	BD	Cat# 562288	1:200
PE-Cy7 anti-human CD25 (M-A251)	BD	Cat# 557741	1:100
BUV395 anti-human CD19 (clone HIB19)	BD	Cat# 740287	1:100
BUV496 anti-human CD4 (clone SK3)	BD	Cat# 612936	1:50
BUV563 anti-human CD56 (clone NCAM16.2)	BD	Cat# 612928	1:200
BUV615 anti-human CD95 (clone DX2)	BD	Cat# 752346	1:50
BUV737 anti-human CD45RA (clone HI100)	BD	Cat# 612846	1:400
BUV805 anti-human CD196 (CCR6) (clone 11A9)	BD	Cat# 749361	1:25
TruStain FcX	Biolegend	Cat# 422304	1:25
TruStain Monocyte Blocker	Biolegend	Cat# 426103	1:25
Brilliant Stain Buffer	BD	Cat# 566349	1:2,5

B cell ELIspot
β mercaptoethanol	Sigma-Aldrich	M3148
Rec human IL-2	BD	554609	10 µg/ml
Rec human IL-10	BD	554611	10 µg/ml
CPG (ODN2006)	Invivogen	tlrl-2006-1	1,2 mg/ml
HRP-conjugated Goat anti-human IgG	Southern Biotech	2040-05	1:5000
PBS	GIBCO	20012-019	NA
FBS	Hyclone	SH30071.03	10%
Aim-V medium containing AlbuMAX, gentamycin, streptomycin and L-glutamine	GIBCO (Fisher Scientific)	31035-025	NA
Anti-human IgG	Southern Biotech	2040-01	1:100 (10 µg/ml)

Software and algorithms
FlowJo v10.10.0
GraphPad Prism 9		https://www.graphpad.com/
Biorender
Excel	Microsoft
Illustrator	Adobe
R version 4.4.1
RADIANT		GitHub - alpercevirgel/RADIANT: R-based Analytical tool for DIfferential ANalysis of cyTometry data (RADIANT)

IgG⁺ B cell ELISpot

For the IgG⁺ B cell ELISpot, inactivated Mumps virus (UC-enriched MuV) was used as antigen coating. PBMCs were cultured for 5 days in AIM-V medium with AlbuMAX (Thermo Fisher, Waltham, MA, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 50 uM β mercaptoethanol, 3 µg/ml CpG (ODN2006), 10 ng/ml recombinant Interleukin (rIL-)2 and 10 ng/ml rIL-10 (Table 1). Wells of ELIspot plate (MSIPS4510, Millipore) were first activated with 35% Ethanol (1 min) and subsequently washed with PBS (4 × 200 µl/well) followed by coating overnight at 4 °C with UC-enriched MuV-JL (10 µg, 200 µg/mL), anti-human IgG (2040-01, Southern Biotech) as positive control, or PBS as negative control. After coating, wells were washed with PBS (4×200 µl/well) and blocked with culture medium (100 µl/well, 30 min, 37 °C). Subsequently, cells were harvested after 5 days of culture and plated at 250.000 cells/well in triplo in pre-coated wells and incubated overnight (37 °C, 5% CO2 and 100% humidity). Plates were washed (4x washing buffer (0,05% Tween20 in PBS), 1x demi water) and subsequently filled with demiwater (5 min, 200 µl/well) to lyse cells. After additional washing (4x washing buffer), wells were incubated for 2 h at 37 °C with Goat Anti-Human IgG-HRP. Next, wells were washed (4x washing buffer) and spots formed by antibody-producing B cells were visualized with TMB substrate (0,22 µm filtered, 50 µl/well, 5–30 min). Results are reported as IgG SFUs per 7,5 × 10⁵ PBMCs.

High-dimensional Flow for immune cell subset profiling

Per donor 500.000 PBMCs were stained for immune profiling by flow cytometry. Cells were first stained in a cocktail containing TruStain FcX and TruStain Monocyte Blocker (10 min, 37 °C), to prevent aspecific antibody binding in the subsequent step. Next, the cells were washed 1x with FACS buffer (PBS + 0.5% BSA) and resuspended in an antibody cocktail containing chemokine receptors CXCR3, CCR4 and CCR6. The cells were incubated (20 min, 37 °C) and washed 2x with 150 μl PBS. For the second staining step the cells were resuspended in a cocktail containing anti-IgG, anti-IgD and Fixable Viability Dye. Incubation was done at (20 min, 4 °C). Finally, the cells were washed 1x with 150 μl FACS buffer and resuspended in an antibody cocktail containing the remaining antibodies for further incubation (20 min, 4 °C). All antibody cocktails contained Brilliant Stain Buffer and TruStain Monocyte Blocker. The cells were acquired on a 5-laser Symphony A3 (BD).

For clustering analysis, we first performed a preliminary step where we exported populations of interest from FlowJo (See Fig. S2 for gating strategy ex vivo phenotyping). We first gated on live cells and from there, myeloid cells and lymphocytes, based on scatter. For both populations, we gated on single cells. From the lymphocytes, we then gated on B cells (CD19⁺), T cells (CD3⁺CD19⁻) and Innate lymphocytes (CD3⁻CD19⁻). Total populations of B cells, CD4⁺ T cells (CD3⁺CD4⁺CD8⁻) and CD8⁺ T cells (CD3⁺CD4⁻CD8⁺) were exported, as well as total Innate lymphocytes. From the myeloid cells, we exported CD3⁻CD19⁻CD56⁻ cells. Finally, since naive subsets for B and T cells comprise ~40–60% of the total population, we decided to separately export naive and non-naive subsets to reach a higher resolution in the non-naive subsets. For the clustering analysis, we used the in-house developed RADIANT (R-based Analytical tool for Differential Analysis of cyTometry data)³⁵. Full documentation and code can be found online (GitHub - alpercevirgel/RADIANT: R-based Analytical tool for Differential ANalysis of cyTometry data (RADIANT)). In brief, FCS files of the different exported subsets were imported into the pipeline and cleaned using the PeacoQC algorithm³⁶. Data was transformed using the Logicle transformation. To correct for batch effect, we used the CytoNorm algorithm, making use of technical controls that were taken along with each batch³⁷. We used histogram overlays of the technical controls and a cross-entropy test on UMAPs before and after CytoNorm to confirm the batch effect had been resolved adequately³⁸. Next, we performed FlowSOM clustering, using 50 metaclusters as final output³⁹. For annotation of immune cell subsets guidelines in literature were used as base^40–43.

Data analysis and statistics and reproducibility

In terms of sample size, 30 individuals were included in the MMR3 immunization study of which PBMCs were isolated for assessment of cellular responses. For the Mumps-infected individuals, we were able to select six individuals who were matched based on age and who had received 2 MMR vaccinations. Data are visualized with boxplots, dots and violin plots. Boxplots display median, interquartile range, and outliers as gray dots. Colored dots represent individual datapoints. Violin plots are trimmed on the range of the data. Univariate hypothesis testing across two or more groups with paired observations was performed with the Wilcoxon signed-rank test. Independent groups were compared with the Wilcoxon rank-sum test (Mann-Whitney U test) or the Kruskal-Wallis test, followed by Dunn’s test for multiple comparisons. p-values for multiple comparisons were adjusted with Bonferroni post-hoc test. Statistical significance is represented with *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001 unless otherwise stated in figure captions. Decay rate (DR) score of mumps-specific responses was calculated for T1Y and T3Y vs. T1M by dividing the median log-transformed values of T1M by the values of T1Y or T3Y, respectively. For multivariable principal coordinate analysis, and given the heterogeneity of the variables’ measurements represented by continuous, count, and percentage data, a pairwise Gower’s distance⁴⁴ between individuals was calculated based on the nine variables displayed in Fig. 1B–G. Percentages of the similarity variance explained were calculated using positive eigenvalues (40% in both cases). All analyses and visualizations were performed in RStudio software (R version 4.4.1, R Core Team), and Adobe Illustrator was used for additional visualization.

Fig. 1. Mumps-specific humoral and cellular responses following a third dose of measles-mumps-rubella vaccine (MMR3) or Mumps virus infection (MVI).

A Experimental study design. B Anti–mumps virus immunoglobulin G (IgG) concentration, C 50% plaque reduction titer (ND50 titer [ie, 50% neutralization dose]) against the mumps virus vaccine strain (JL) before (T0) and 4 weeks (T1M), 1 years (T1Y) and 3 years (T3Y) after MMR3 immunization (blue, n = 30) or 1-2 months (T1M) and 7–10 months (T8M) after mumps virus infection (red, n = 6). Data are visualized with boxplots, dots and violin plots. Boxplots display median, interquartile range, and outliers as gray dots. Colored dots represent individual datapoints. Violin plots are trimmed on the range of the data. Bold black squares represent group medians. Time points are connected with lines per individual (gray) or group median (bold black) illustrating kinetics of responses over time. Red dashed lines indicate the surrogate antibody level cutoff for protection against mumps virus infection¹⁸. D Number of anti-mumps IgG⁺ B cells (antibody producing spots per 7,5x10E5 cells)(MuV-coated ELIspot) with representative wells, E number of IFNy-producing T cells (NK-depleted ELIspot), F activated CD8⁺ T cells expressing CD137⁺, or G activated CD4⁺ T cells expressing CD137⁺OX40⁺, CD137⁺CD154⁺, CD154⁺OX40⁺ or CD137⁺CD154⁺OX40⁺ in young adults at T0, T1M and T1Y post-MMR3 and T1M and T8M post-MVI. H Pie charts illustrate the functionality profile of all four activated CD4⁺ T cell populations over time both after MMR3 immunization and mumps virus infection with the proportion of cells (median values) secreting no, one, or multiple cytokines upon live mumps virus stimulation, as color-indicated. I Spearman correlogram between all immune parameters T1M and T1Y post-MMR3 and T1M and T8M post-MVI. Wilcoxon rank-sum test (Mann-Whitney U test) or pairwise Wilcoxon signed-rank test for univariate hypothesis testing across two independent or paired groups, respectively. For three or more groups, Kruskal-Wallis test followed by Dunn’s test for multiple comparisons or Friedman’s test followed by pairwise Wilcoxon signed-rank test were employed. p-values for multiple comparisons were adjusted with Bonferroni post-hoc test * = significant between time points within MMR3 or MVI. # = significant between MMR3 and MVI at T1M. ^ = significant between MMR3 and MVI at T1Y vs T8M.* p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001 (similar for ^ and #).

Results

MMR3 induces mumps-specific humoral, B cell and CD4⁺ T cell responses

During recent mumps outbreaks, especially young adults were at risk for mumps virus infection. A third dose of MMR vaccine (MMR3) is considered a feasible option to boost their immunity and reduce the risk of mumps virus infection. We previously showed that immunizing young adults with MMR3 is safe and immunogenic in terms of inducing (virus-neutralizing) antibody responses¹⁸. Here, we set out to study whether MMR3 immunization also induces mump-specific T and B cell responses. From 30 out of 147 participants (18–25 years of age) that were enrolled in the study, peripheral blood mononuclear cells (PBMCs) were collected at three time points (pre (T0), 1 month (T1M) and 1 year (T1Y) post-MMR3) to determine MMR3-induced cellular responses. The type and magnitude of these immune responses were additionally compared to those of previously vaccinated persons who recently experienced a mumps virus infection (MVI) that served as a benchmark for strong and long-lasting immunity. For this, PBMCs were collected at two time points (1–2 (T1M) and 7–10 months (T8M) post-MVI) from six individuals who were matched for age (18–25 years) and vaccination status (2xMMR)(Fig.1A).

In the 30 participants, for whom cellular immune responses were measured, we first confirmed that mumps virus–specific IgG antibody concentrations (anti-MuV IgG)(Fig. 1B) and their virus neutralization capacity (VN, ND₅₀ titer, Jeryl Lynn (JL) mumps virus vaccine strain)(Fig.1C) were significantly increased T1M post-MMR3 compared to baseline levels. At T3Y post-MMR3, anti-MuV IgG levels significantly declined compared to T1M and T1Y post-MMR3 but tended to remain higher than pre-immunization levels, although not statistically significant (Fig.1B). In line with this increased humoral response we furthermore demonstrated here that MMR3 also significantly boosted circulating mumps-specific IgG⁺ memory B cell responses (IgG⁺ B cells hereafter) T1M post-immunization, which subsequently contracted at T1Y in the majority of the individuals (Fig.1D). Although humoral and IgG⁺ B cell responses are induced upon MMR3 immunization, levels were significantly lower than those induced after MVI (Fig.1B–D).

MuV-specific T cell immunity may play a pivotal role in protection against mumps virus infection. However, to date, no published studies have specifically investigated MuV-specific T cell responses following MMR3 immunization. To address this gap, we first assessed the frequency of MuV-specific T cells using an IFNγ ELISpot assay. Here, the median MuV-specific T cell frequency tended to be increased T1M post-MMR3, although not significant, yet significantly lower than observed T1M post-MVI (Fig.1E). Subsequently, activation and functionality profiles of mumps-specific CD4⁺ and CD8⁺ T cell responses were assessed using an activation-induced marker (AIM) flow cytometry assay, which employed three activation markers (CD154, OX40, CD137) in combination with intracellular cytokine staining (ICS) for IFNγ, IL-2, TNFα, IL-5, IL-10 and IL-13 (Fig.1F–H). For CD8⁺ T cells, only expression of CD137 and not CD154 or OX40 was found. Therefore, we focused exclusively on CD137⁺CD8⁺ T cells as activated CD8⁺ T cells. The frequencies of activated CD137⁺CD8⁺ T cells did not increase following MMR3; however, markedly higher frequencies were observed at T1M post-MVI (Fig. 1F). For CD4+ T cells, expression of one or more of the three activation markers was detected. Four distinct activated CD4⁺ T cell populations, based on a combination of two or three activation markers, were further investigated. MMR3 led to a significant increase in the frequency of CD4⁺CD137⁺CD154⁺OX40⁺ T cells (triple-AIM-positive CD4⁺ T cells hereafter)(Fig.1G). These activated CD4⁺ T cells mainly displayed an effector memory (EM) phenotype (Fig.S3A). Moreover, these triple-AIM-positive CD4⁺ T cells also showed the highest cytokine production with approximately 50% of cells producing one or more cytokines (IFNγ, TNFα and/or IL-2, while IL-5, IL-10 and IL-13 were not detected). The activated cytokine-producing double-AIM-positive T cell populations, CD4⁺CD137⁺CD154⁺ and CD4⁺CD154⁺OX40⁺, accounted for 33% and 10%, respectively (Fig.1H). No cytokine-producing cells were detected in activated CD4⁺CD137⁺OX40⁺ T cells. Interestingly, while the frequencies of activated CD4⁺ T cells increased as a result of MMR3, the cytokine profiles of all four activated CD4⁺ T cell populations remained stable over time (Fig.1H). One month post-MVI, activated triple-AIM-positive and CD4⁺CD154⁺OX40⁺ T cell frequencies were slightly, but significantly, higher compared to T1M post-MMR3 (Fig.1G). Moreover, among infected individuals, activated CD4⁺ T cell populations generally showed higher proportions of cytokine-producing cells compared to vaccinated individuals. However, this difference was only statistically significant for CD154⁺OX40⁺ and triple-AIM populations (Fig.1H). In the MVI group, activated CD4⁺ T cell numbers contracted at T8M post-MVI to similar levels to those observed at baseline prior to MMR3 immunization (T0) in age-matched individuals with a similar vaccination history, which we used as a proxy for pre-infection levels, given the lack of pre-infection samples. However, this activated CD4⁺ T cell fraction at T8M post-MVI still had a more polyfunctional character, especially the IFNγ⁺IL2⁻TNFα⁺-producing cells, as compared to the same cell fraction at T1Y post-MMR3 (Fig.1H). Although IgG antibody concentrations, VN titers, IgG⁺ B cell, and activated CD8⁺ T cell frequencies also contracted from T1M to T8M post-MVI, levels remained elevated compared to baseline (T0)(Fig.1G). Interestingly, the decay rate of mumps-specific responses between T1M and T1Y is higher for infection-induced (2–7 fold) as compared to immunization-induced responses (1,1-1,5 fold)(Fig. S3B). For MMR3 immunization, the highest decay was observed for IgG⁺ B cells (1,5 fold), while the decay rate within infected individuals was more profound for IgG⁺ B cell, CD4⁺ T cell and CD8⁺ T cell responses (3-7 fold) as compared to IgG and VN antibody responses (2-fold).

Next, correlations between the multiple mumps-specific readouts (antibodies, T cells and IgG⁺ B cells) following MMR3 and MVI were explored (Fig.1I). Within MMR3 immunized individuals, moderate to strong positive correlations were found between the frequencies of various activated CD4⁺ T cell populations T1M and T1Y post-MMR3. Strikingly, these activated CD4⁺ T cell frequencies did not correlate to humoral responses (IgG, VN and IgG⁺ B cells). Notably, weak positive correlations of IgG⁺ B cell responses with IgG and VN levels were found at T1Y post-MMR3. Within the infected individuals, significant positive correlations were observed at T1M post-MVI between (i) activated CD4⁺ T cells and IgG responses, and (ii) B cells and IgG/VN levels, which all disappeared T8M post-MVI (Fig.1I). Overall, the correlations post-MMR3 were weaker as compared to post-MVI.

MMR3 cohort comprises of five subpopulations with distinct MuV-specific immunity profiles

Next, we investigated the multivariate response between MMR3- and MVI-participants and across timepoints. Given the heterogeneity of the readouts, represented by continuous, count and percentage data, Gower’s distance was calculated to account for the inter-individual immune response diversity based on the nine mumps-specific variables presented in Fig.1B-G. The principal coordinate analysis (PCoA) on the Gower’s distance matrix highlighted distinct immune response profiles between MMR3- and MVI-participants, based on both antibody and T cell responses (Fig. 2A). The two variable sets are largely independent, as indicated by their orthogonal orientation, meaning that antibody and T cell responses explain different sources of the variation in participants’ similarity. Additionally, antibody variables showed the highest contribution to the first principal component (PCoA1), which explains a larger proportion of participants´ similarity variance (38.06%) than PCoA2 (17.57%) and clearly distinguishes MMR3 and MVI participants. To gain insights on the potential heterogeneity of immune responses within the MMR3 cohort, we performed an additional principal coordinate analysis on these participants (Fig.2B). This analysis showed that while both sets of variables contribute equally to PCoA1, they separated participants in opposite directions along PCoA2. This suggests that the interindividual variation in mump-specific responses might distinguish different subpopulations of MMR3 responders, distinctly characterized by particularly antibody/VN or T cell responses.

Fig. 2. Interindividual variation in combined mumps-specific datasets following a third dose of measles-mumps-rubella vaccine (MMR3) or mumps virus infection (MVI) leading to classification of five subpopulations with distinct mump-specific responses.

PcoA plot of A combined data for MVI (1/2 months (T1M) and 7/10 months (T8M) post-MVI) and MMR3 (Before (T0), 4 weeks (T1M) and 1 year (T1Y) post-MMR3) and B with only data related to MMR3. For each of the participants (MMR3: n = 30, MVI: n = 6) the three different timepoints are connected. Arrows represent direction of variance for contributing datasets (red = humoral response, green = activated CD4⁺ T cell response). Ellipses cover variance of all individuals belonging to a group and/or time point. C Flow chart with the defining steps to separate the total MMR3 cohort in five subpopulations based on immune characteristics. Boxes were marked green when positive to one or more of the four questions. D Pre-immunization levels (T0) (left panel) and increased ( > 1,5; T1M/T0) fold change (right panel) of anti-MuV IgG, VN (JL) titer and activated CD4⁺ T cells were used to classify individuals in the total MMR3 cohort into five subpopulations (I–V) with distinct mump-specific responses. Red dashed lines represent surrogate protective cut-off levels (102 for anti-MuV IgG, 34 for VN (JL)) and black dashed lines the fold change cut-off at 1,5. Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. p-values for multiple comparisons were adjusted with the Bonferroni post-hoc test. *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

This prompted us to further characterize these subpopulations. For this purpose individuals of the MMR3 cohort were classified in five subpopulations (I-V) based on pre-existing immunity and their MMR3-induced mumps-specific IgG antibody, VN titers and activated CD4⁺ T cell response profiles (Fig.2C, D). All individuals in subpopulation V and some (5/13) in subpopulation I + II had low or absent pre-existing humoral levels as their anti-MuV IgG antibodies were at or below the defined surrogate level for protection of 102 RU/mL before immunization. Post-MMR3, all individuals in subpopulations I & II demonstrated a strong boost (FC > 1,5) in anti-MuV IgG antibody levels, indicating that they benefited from the MMR3 immunization. Subpopulations III and IV consist of individuals who already had high levels of pre-existing anti-MuV IgG antibodies that were not or hardly (FC < 1, 5) boosted post-MMR3. This separation based on pre-existing immunity and post-MMR3 IgG antibody responses could be made in similar proportions in both the total MMR3 cohort (n = 147, Fig.S4A–D) and the subcohort for cellular immunity (n = 30, Fig.2C, D). Furthermore, subpopulation II showcased the highest MMR3-induced increase in VN titers whereas these were not affected or only to a limited extend in subpopulations I, III & IV (Fig.2D). Strikingly, IgG antibody levels (subpopulation I) but particularly VN levels (subpopulations I & II) 1 month post-MMR3 never reached the same magnitude observed in subpopulations III and IV (Fig.3A-B). Additionally, subpopulation II had the highest decay rates for VN titers (median±SD; 1,94 ± 0,41) and IgG⁺ B cell frequencies (1,78 ± 0,47)(Fig.S3B). The lower magnitude of MMR3-induced VN responses observed in subpopulation I and II also led to waning of responses close to the low pre-vaccination levels after 1-3 years (Fig.3A, B) indicating a shorter-lived protective immunity as compared to subpopulation III and IV. In terms of cellular responses, both subpopulations with low pre-existing immunity (I + II) had a boost in CD4⁺ T cell responses (Fig.2D) that were higher in subpopulation I (Fig.3A, B), while IgG and IgG⁺ B cell responses were higher in subpopulation II (Fig.3A, B). The individuals with pre-existing humoral immunity could be divided in a subpopulation that displayed a T cell boost (IV) or not (III)(Figs.2C, D, 3A, B). Interestingly, the CD4⁺ T cell responses observed in subpopulation IV also displayed a higher proportion of functionally-active cytokine-producing cells (Fig.3C). Additionally, IgG⁺ B cell responses were also further boosted in subpopulation IV but not in III (Fig.3A, B). Those individuals that did not have pre-existing immunity nor induced sufficient MuV-specific responses following the MMR3 were considered non/weak responders (subpopulation V) and covered 18 and 10% of the total cohort (n = 147, Fig.S4A–D) and subcohort for cellular immunity (n = 30, Fig.2C, D) of the MMR3 cohort, respectively. Interestingly, the non-responding individuals (subpopulation V) that did not mount any anti-MuV immunity following MMR-immunization did not have hampered responses to the measles and rubella vaccine components as significant induction of post-MMR3 anti-MeV and anti-RuV IgG antibody levels were detected (Fig.S5A, B). No significant differences were observed in the proportion of seropositivity for chronic CMV, EBV or VZV infections in the five subpopulations and therefore this seems no determinant for the distinct MuV-specific responsiveness in this age group (Fig.S5C). Within the four responding subpopulations (I-IV) distinct correlation patterns between the MuV-specific datasets were observed (Fig.3D). Here, subpopulation III showed the best positive correlation between IgG⁺ B cells and IgG responses.

Fig. 3. Interindividual variation and distinct mumps-specific responses of five subpopulations following a third dose of measles-mumps-rubella vaccine (MMR3).

A Interindividual variance and kinetics of mumps-specific responses in young adults before and after MMR3 in all five different subpopulations (I–V; I: n = 7, II: n = 6, III: n = 9, IV: n = 5, V: n = 3). Data are visualized with boxplots that display median, interquartile range, and outliers as gray dots. Colored dots represent individual datapoints. Bold black squares represent group medians. Time points are connected with lines per individual (gray) or group median (bold black) illustrating kinetics of responses over time. Red dashed lines indicate the surrogate antibody level cutoff for protection against mumps virus infection¹⁸. B The relative abundance of all immune parameters was calculated over time in the five subpopulations (I-V) and the total MMR3 cohort (ALL) to illustrate distinct profiles of mump-specific responses. C Pie charts illustrate the functionality profile of the activated CD4⁺CD137⁺CD154⁺OX40⁺ (triple AIM) T cell population over time in all five (I-V) subpopulations after MMR3 with the proportion of cells (median values) secreting no, one, or multiple cytokines upon live mumps virus stimulation, as color-indicated. D Spearman correlogram between all immune parameters T1M post-MMR3 immunization in total cohort and subpopulation 1–4. Subpopulation 5 could not be depicted in a correlogram due to small sample size (n = 3). Friedman’s test followed by pairwise Wilcoxon signed-rank test *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

Overall, based on the full immunization schedule (MMR1-3), the individuals in subpopulation IV are the best responders as they have higher and longer-lasting levels of neutralizing antibodies and MuV-specific IgG⁺ B cells in addition to more activated and functional CD4⁺ T cell responses compared to other subpopulations.

High-dimensional immunophenotyping unravels immune cell subset profiles that characterize MMR3 and mumps virus infection

Next, we used high-dimensional immunophenotyping with flow cytometry to gain insight into changes in proportions and kinetics of circulating immune cell subsets. This allowed us to pinpoint immune cell subsets that changed significantly (i) over time in a cohort (MMR3 or MVI), (ii) between MMR3 and MVI, (iii) over time within a subpopulation of the MMR3 cohort or (iv) between subpopulations of the MMR3 cohort at each timepoint.

Assessment of the median proportions of major immune cell types in PBMCs of the young adults from both cohorts revealed that the composition comprised of approximately 63% T cells, 8% B cells, 16% myeloid cells (including monocytes and dendritic cells), 8% NK cells, and 2% other innate lymphocytes (Fig.4A). Following MMR3, CD4⁺ T_EMRA cells, non-classical monocytes and CD14⁻CD16⁻ myeloid cells significantly increased while total CD8⁺ T cells proportions, particularly the naive subset, as well as IgD⁻CD27⁻ double negative (DN) B cells and intermediate monocytes decreased post-MMR3 (Fig.4A and S6A, B). Proportions of both CD4⁺CD8⁺ T cells and CD4⁻CD8⁻ T cells decreased at T8M compared to T1M post-MVI. Non-classical monocytes and CD56^dimCD16⁻ NK-cells were significantly higher at both T1M and T1Y post-MMR3 compared to post-MVI. Within MMR3 subpopulation III, a significant increase of naive CD4⁺ T cells was observed T1M post-MMR3.

Fig. 4. High-dimensional flow cytometry for Ex vivo phenotyping of immune cell subsets following MMR3 and mumps virus infection.

A A representative pie chart with distribution of 30 general immune cell subsets divided over three layers including the underlying subset name, median proportion (%), kinetics of normalized median over time following MMR3 and mumps virus infection and statistical difference in 12 comparisons. B High dimensional flow cytometry led to dissection of PBMC composition in 264 immune cell subsets divided over CD19⁺ B cell, CD4⁺ T cell, CD8⁺ T cell, myeloid cells and innate lymphocytes. C Distribution of these cell types was depicted for all 264 immune cell subsets and the 174 that were significantly different. Additionally, the number and corresponding immune cell types were depicted for all immune subsets that were identified as significantly different in one of more of the 12 statistical comparisons. Venn diagrams depict the overlap in immune cell subsets per comparison. Wilcoxon rank-sum test (Mann-Whitney U test) or pairwise Wilcoxon signed-rank test for univariate hypothesis testing across two independent or paired groups respectively. For three or more groups, the Kruskal-Wallis test followed by Dunn’s test for multiple comparisons or the Friedman’s test followed by the pairwise Wilcoxon signed-rank test were employed. p-values for multiple comparisons were adjusted with Bonferroni post-hoc test *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

In order to increase the phenotyping resolution we subsequently applied deep ex vivo immunophenotyping using FlowSOM for unsupervised clustering. As approximately 40–60% of both B and T cells are classified as the naive compartment, we decided to separate naive and non-naive subsets to obtain a higher resolution in the non-naive subsets. Overall this provided 264 unique immune cell subsets (Fig.4B) divided over B cells (5 naive + 50 non-naive subsets)(Fig.S7–9), CD4⁺ T cells (3 naive + 50 non-naive subsets)(Fig.S10–12), CD8⁺ T cells (6 naive + 50 non-naive subsets)(Fig.S13–15), myeloid cells (including monocytes and DCs)(50 subsets)(Fig.S16-17) and innate lymphocytes (including NK cells and ILCs)(50 subsets)(Fig.S18–19) for which annotation, proportion, kinetics and various statistical comparisons were determined. In 174 out of the total 264 immune cell subsets identified, one or more of the beforementioned statistical differences were detected (Fig.4C). The response T1M post-MMR3, was characterized by significant changes in the frequencies of 44 immune cell subsets, of which 27 increased and 17 decreased in proportion (Fig.4C and Fig.5 - Green column). In general, the response to MMR3 was characterized by increased circulation of four classical monocyte subsets (Mono-22, 28, 34, 47) and innate lymphocytes (InLym-8, 9, 35) within the innate cellular compartment (Fig.5).

Fig. 5. Kinetics and correlations of significant immune cell subsets following MMR3.

Overview of 44 immune cell subsets that were significantly different between pre immunization (T0) and 1 month post-MMR3 (T1M). Immune cell subsets are listed under major cell types and marked with subset ID, annotation, phenotype marker profile for corresponding markers, and the proportion in the overarching immune cell population, with higher (green) or lower (red) proportions at T1M compared to T0. Kinetics following MMR immunization and mumps virus infection (MVI) is shown based on normalized median. For each immune cell subset, significant differences are depicted for different statistical comparisons. Spearman correlations were determined between immune cell subset frequencies and all mumps-specific datasets at T1M-post MMR3. Wilcoxon rank-sum test (Mann-Whitney U test) or pairwise Wilcoxon signed-rank test for univariate hypothesis testing across two independent or paired groups respectively. For three or more groups, Kruskal-Wallis test followed by Dunn’s test for multiple comparisons or Friedman’s test followed by pairwise Wilcoxon signed-rank test were employed. p-values for multiple comparisons were adjusted with Bonferroni post-hoc test *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

Most notably, at T1M post-MMR3, there was a significant increase in the frequency of circulating CD4⁺ T cells with a CD127⁺CD25⁻CD38⁻Th1/Th17 T_CM phenotype (NNCD4-2) (Figs.5 and S6A, B), which showed the highest increase in the MMR3 high responders, subpopulation IV and was also enhanced post-MVI (Fig.6C). Remarkably, this subset revealed significant positive correlations with most mumps-specific activated CD4⁺ populations (Fig.5), in particular the triple-AIM-positive CD4⁺ T cell population (total MMR3: R = 0,39 p = 0,001; T1M: R = 0,50 p = 0,011) (Fig.6D). The strongest correlations between the triple-AIM-positive CD4⁺ T cell population and this CD4⁺ T_CM subset were observed in MMR3 subpopulations II and IV at T1M post-MMR3 as well as T1M post-MVI (Fig.6E). Furthermore, the frequency of multiple switched memory B cell (MBC) subsets increased one month post-MMR3 (Fig.5). Of these, interestingly, a IgG⁺ switched MBC subset with a CD38⁺CD24⁻CD21⁺CD95⁺CXCR5⁺CCR6⁺CXCR3⁺ phenotype (NNB-47) (Figs. 5 and 6F, G), which also increased the most in MMR3 subpopulation IV and post-MVI (Fig.6H) showed a modest positive correlation (R = 0,41; p = 0,027) with VN(JL) titers T1M post-MMR3 (Figs.5, 6I) particularly in subpopulation II but not to anti-MuV IgG and IgG⁺ B cell responses (Fig.6J). Upon closer examination of immune cell subsets within the various subpopulations, a total of 31 subsets showed significant changes over time following MMR3 immunization; notably, these changes were observed only in subpopulations I and III (Fig.S20A, 21). Of the 9 subsets that changed significantly 1 month post-MMR3 compared to T0 (Fig.4C and S20A) 8 were also found in the 44 immune cell subsets that changed in the total MMR3 cohort (Fig.5). Myeloid cell subsets were particularly increased at T1M in individuals of subpopulation I (Mono-22, 27, 30) and III (Mono-21, 30)(Fig.S20A).

Fig. 6. Immune cell subsets correlating to MMR3-induced immune responses.

A Flow cytometry gating (red dots) and B location in UMAP (dark blue) of NN-CD4+ subset 2 in total non-naive CD4⁺ T cells (gray). C The interindividual kinetics of the NN-CD4⁺ subset 2 over time following MMR3 in total cohort (n = 30) and subpopulations (I–V) (blue) and mumps virus infection (red, n = 6). D Correlation plot between NN-CD4+ subset 2 and activated CD4⁺CD137⁺CD154⁺OX40⁺ cells for three time points post-MMR3 (T0, T1M, T1Y, blue), two timepoints post-MVI (T1M, T8M, red) and all combined (dark gray). Individual time points per individual are connected with a gray line. E Spearman correlations of the NN-CD4⁺ subset 2 against four mumps-specific activated CD4⁺ populations at T1M in total MMR3 cohort, MMR3 subpopulations (I–V) and mumps virus infection cohort. F Flow cytometry gating (red dots) and G location in UMAP (dark blue) of NN-B subset 47 in total non-naive B cells (gray). H The interindividual kinetics of the NN-B subset 47 over time following MMR3 in total cohort and subpopulations (I–V) (blue) and mumps virus infection (red). I Correlation plot between NN-B subset 47 and anti-mumps virus neutralizing titers for three time points post-MMR3 (T0, T1M, T1Y, blue), two timepoints post-MVI (T1M, T8M, red) and all combined (dark gray). Individual time points per individual are connected with a gray line. J Spearman correlations of the NN-B subset 47 against anti-mumps IgG⁺ B cell frequencies, IgG antibody levels and virus neutralizing titers at T1M in total MMR3 cohort, MMR3 subpopulations (I–V) and mumps virus infection cohort. Friedman’s test followed by pairwise Wilcoxon signed-rank test *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

Next we deciphered whether certain immune cell subsets differed between subpopulations at baseline (T0) or after immunization (T1M, T1Y) and detected 12 unique immune cell subsets at one or more time points (Fig.4C and S20B). Here, most striking, the subpopulation with non-responders (V), as well as infected individuals, had higher levels of a CD56^dimCD16⁺CD45RA⁺ (InLym-31) NK cell subset compared to other subpopulations at baseline and post-MMR3 (Fig.S20C). These higher cell frequencies in non-responders (V) only differed significantly compared to the highest MMR responders (IV), which that had the lowest baseline levels of this NK cell subset. This subset may therefore serve as a potential marker for lower MMR responsiveness and increased susceptibility to mumps virus infection.

Unfortunately, the absence of a pre-infection sample (T0) hampered to determine direct changes of immune cell subsets involved during MVI. Only two cell subsets, namely CCR4⁺ classical (Mono-35) and CD45RA⁺⁺HLA_DR⁺CCR6⁺⁺ non-classical (Mono-4) monocytes, significantly changed between T1M and T8M post-MVI, being lower at the latter time point (Fig.4C and Fig.7A). Both cell subsets were also significantly lower in infected as compared to MMR3-vaccinated individuals at both T1M and T1Y. Therefore, we decided to compare differences between MVI and MMR3, which revealed 80 immune cell subsets that significantly differed at T1M between MMR3 and MVI (Figs.4C, 7A - yellow column). With respect to B cells, the frequencies of CD27⁻ASC (NNB-25)(Fig.7A), unswitched MBC (NNB-2) and switched IgG⁻ MBC (NNB-22)(Fig.7A, B) cell subsets, among others, were higher in infected individuals. Moreover, these individuals had higher proportions of CD8⁺ T_CM (NNCD8-47), CD8⁺CD95⁺CD38⁺ T_EM (NNCD8-23) and CD4⁺CD127⁺CCR6⁺ T_CM (NNCD4-13) T cell subsets (Fig.7B). Interestingly, especially in the innate compartment, elevated frequencies were observed at T1M between MMR3-immunized and MVI individuals of two NK cell subsets (InLym-10, InLym-31), two CD3⁻CD4⁺CD56⁻ subsets (InLym-49, InLym-50) and two classical monocyte subsets (Mono-22, Mono-33), which strikingly all remained elevated at T8M post-MVI (Fig.7B).

Fig. 7. Immune cell subsets following mumps virus infection.

A Overview of 80 immune cell subsets that were significantly different between mumps virus infection and MMR3 at T1M. For each immune cell subset, increase (green) or decrease (red) of proportion in MVI vs MMR3 are depicted as well as phenotype based on expression of represented markers, % proportion in overarching immune cell type, kinetics of normalized median and significance per comparison. B The interindividual kinetics of a selection of the immune cell subsets was depicted both for MMR3 (blue) and mumps virus infection (red). Wilcoxon rank-sum test (Mann-Whitney U test) or pairwise Wilcoxon signed-rank test for univariate hypothesis testing across two independent or paired groups respectively. For three or more groups, Kruskal-Wallis test followed by Dunn’s test for multiple comparisons or Friedman’s test followed by pairwise Wilcoxon signed-rank test were employed. p-values for multiple comparisons were adjusted with Bonferroni post-hoc test * = significant between time points within MMR3 or MVI. # = significant between MMR3 and MVI at T1M. ^ = significant between MMR3 and MVI at T1Y vs T8M. *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

Discussion

We applied a multiparameter analysis on a uniform group of young adults after a third Mumps-Measles-Rubella (MMR3) immunization or mumps virus infection (MVI) to evaluate various aspects of cellular and humoral immunity and interindividual variations in immune responses. High-dimensional flow cytometry was additionally used to determine which immune cell subsets characterized responsiveness to MMR3 immunization or MVI.

In this study, we demonstrated that MMR3 immunization clearly boosts MuV-specific IgG⁺ B cell and CD4⁺ T cell responses in young adults on top of the improved IgG and virus-neutralizing (VN) antibody responses that we described before¹⁸. Similar to MMR2 immunization²⁵, MMR3 hardly increased MuV-specific CD8⁺ T cell frequencies. Others have also shown that MuV-specific B cells are induced following MMR vaccination and mumps virus infection^27,28. We additionally demonstrated for MMR3 immunization that while MuV-specific IgG⁺ B cell responses showed a moderate positive correlation with IgG and VN antibody levels, these responses did not correlate to MuV-specific CD4⁺ T cell responses. All responses were measured against whole mumps virus and therefore future opportunities entail to dissect mump-specific humoral and cellular responses on individual antigen level as some indications already point, but are not limited, to Nucleoprotein and Hemagglutinin^45,46. Strikingly, a closer examination of individual variation among vaccinees allowed us to identify five distinct subpopulations, based on their mumps-specific humoral and cellular immunity pre- and post-MMR3. Overall, our results show that ~43% of the young adults in this cohort benefited from a third MMR immunization, as their immunity was clearly boosted, although this increase was transient (subpopulation I and II). Approximately 47% vaccinees already had high pre-existing long-lasting immunity that was only partially boosted by the third dose (subpopulation III and IV). Strikingly, 10% of participants were identified as non-responders (subpopulation V) who did not develop adequate immunity to mumps, even after a third dose of the MMR vaccine.

Pre-existing immunity, especially circulating antibodies and their neutralizing capacity, was the key factor that enabled the separation of two major groups (I + II and III + IV). Subpopulations III+IV were characterized by high pre-existing immunity, while subpopulations I+II had no or low pre-existing immunity, probably due to the waning of responses induced by their MMR2 childhood immunization received at 9 years of age. Waning of immunity after MMR2 has been described before^11,12, but we demonstrate here and previously¹⁸ that, 9–16 years later (prior MMR3 vaccination), the neutralizing IgG levels among individuals aged 18–25 years are highly variable. The additional MMR3 only marginally enhanced humoral immunity in groups with already high pre-existing VN antibody levels (III and IV). This may be caused by their existing antibodies neutralizing the live-attenuated virus in the vaccine, which requires replication to elicit a strong immune response. The strongest enhancement in humoral immunity was observed in individuals with lower antibody levels at baseline (subpopulation I + II), indicating that these groups benefited the most from the MMR3. Strikingly, the levels and especially the neutralizing capacity of antibodies induced in subpopulation I and II did, however, not reach the same levels, not even the baseline levels, as detected in subpopulation III and IV. More importantly, these lower induced levels had mostly declined to baseline levels 3 years post-MMR3, indicating no long-lasting immunity in individuals in subpopulations I and II. This is corresponding to the low pre-existing immunity following MMR2 immunization (T0 levels) in these groups indicating that these groups may be more prone for waning of immunity to non-protective levels. One possible explanation is suboptimal priming by MMR1/2 vaccination⁴⁷; however, antibody data following MMR1 immunization for individuals in this cohort are not available. Lewnard et al. estimated that MMR vaccine effectiveness on average lasts 27 years but importantly also found that in approximately 25% of individuals immune protection lasts only 7 years⁴⁸ indicating faster waning and prompter susceptibility to MVI. Even for MVI-induced immunity, considered the potential gold standard for high and long-lasting immunity²⁶, there is high variability in levels of immunity and protection as reinfections have been observed⁴⁹. The individuals in subpopulation III and IV had longer-lasting humoral and cellular immunity as they were higher 1-3 years post-MMR3 compared to subpopulation I and II. Interestingly, while both subpopulations III and IV showcased similar high pre-existing humoral immunity levels that were not boosted, these groups could be separated as the individuals in subpopulation IV, but not those in subpopulation III, displayed a boost in CD4⁺ T cell (as well as in B cell) responses. Other studies have identified potential predictors of MMR-induced responses, such as specific miRNA profiles in B cells⁵⁰ and early B cell transcriptomic markers associated with measles-specific humoral immunity following a third dose of the MMR vaccine⁵¹. Wang et al. reported on correlations between regulatory T cells and MMR-induced anti-Rubella antibody responses²⁸. Notably, our high-dimensional immunophenotyping strategy revealed a minor CD38⁺CD24⁻CD21⁺CD95⁺CXCR5⁺CCR6⁺CXCR3⁺ IgG⁺ memory B cell (MBC) subset (NNB-47) that correlated moderately positively to virus-neutralizing antibody titers. More striking was the increased frequency of a CD4⁺ subset with a CD127⁺CD25⁻CD38⁻Th1/Th17 T_CM phenotype (NNCD4-2) following MMR immunization that positively correlated, especially in the high T cell responding subpopulation (IV), with activated MuV-specific CD4⁺ responses. One could speculate that the cells of this NNCD4-2 subset and the activated MuV-specific CD4⁺ T cells might be the same cells, as these cells also show Th1 cytokine production. Overall we can conclude that individuals in subpopulation IV could be marked as overall highest responders, whose immune profiles may provide valuable insights for future studies⁵², as they showed longer-lasting and higher levels of IgG antibodies, VN antibody titers, and more prominent B cell and functional CD4⁺ T cell responses. On the contrary, 10% of the participants were considered as non/weak responders (subpopulation V) as they had no or very low MuV-specific humoral and cellular responses. Interestingly, this observation did not coincide with low or absent humoral responses to the measles and rubella components, indicating that this is a mumps-specific issue in these individuals. This percentage of non-responders is in agreement with data reported by Rasheed et al.¹¹. on individuals that did not had circulating mumps-specific memory B cells. These individuals may have suffered from improper MMR1/2 priming⁴⁷ or carry genetic variations^50,53–55 that may influence the responsiveness to MMR immunization. Identification of non-responders is especially important as they may be more susceptible for MVI despite being fully vaccinated⁵⁶ and likely do not benefit from additional MMR boosters⁵⁷. Crooke et al. found that IL-10 and IL-4 were negatively associated to MMR3-induced rubella-specific VN antibody and B cell frequency and could potentially play a role in hampering the induction of antigen-specific responses⁵⁸. However, we did not detect any IL-10-producing mumps-specific CD4⁺ T cells after mumps immunization or infection. Interestingly, non-responders (subpopulation V) and infected individuals had significantly higher frequencies of a CD56^dimCD16⁺ NK cell subset (InLym-31) that might be a potential biomarker of impaired MMR-induced responses. The infected individuals in our study were young adults and fully vaccinated so they matched the criteria to be enrolled in the MMR3 cohort. One could speculate that these individuals may have had similar immunity profiles as subpopulations with no (V) or only low and short-lasting (I and II) immunity, as their MMR2-induced immunity may not have been sufficient to prevent them from a breakthrough infection.

The immune response following mumps virus infection can serve as a benchmark for long-term immunity²⁹. The direct comparison between responses following natural mumps infection and booster vaccination in our study confirms that infection initially induces a more robust immune response, but also leads to a more pronounced decline during the contraction phase. Mump-specific antibody, IgG⁺ B cell and CD4⁺ T cell responses induced by infection were approximately threefold higher than those induced by MMR vaccination. Moreover, more CD8⁺ T cells were activated after natural infection compared to vaccination, as described before²⁵. In terms of duration, the responses following infection were subjected to a higher decay rate as compared to those induced by immunization, yet antibody levels, IgG⁺ B cell and CD8⁺ T cell frequencies remained higher 8 months post-MVI compared to any other timepoint following vaccination, whereas frequencies of activated CD4⁺ T cells declined to pre-immunization levels 8 months post-MVI. Overall, this lower vaccine-induced response is likely attributable to the live-attenuated nature of the vaccine, which relies on limited replication within the host. Attenuation of the vaccine strain, as well as the administration route (intramuscular) and dosing, may account for the lower magnitude of the immune response compared to natural infection. In contrast, during natural infection, the mumps virus enters the respiratory tract and triggers epithelial cells to induce Type-III interferon responses⁵⁹, which is followed by broader innate responses. We demonstrated here that individuals who have experienced a recent mumps virus infection had enhanced systemic circulation of multiple classical monocyte subsets (Mono-22, 33), NK cell subsets (InLym-10, 31) as well as CD56⁻ innate lymphocytes (InLym-49, 50; ILCs). Whereas ILCs⁶⁰ and particularly NK cells⁶¹ are involved in other viral infections, to our knowledge there is no reported data present on involvement of these cells following a mumps virus infection. An important limitation of the present study, especially after mumps virus infection, is that mucosal humoral and cellular responses, which are of relevance given the mucosal port of entry of the virus were not assessed. Interestingly, 1-2 months post-infection, we did observe an increased proportion of an activated switched MBC subset with a CD24⁺CD95⁺CXCR5⁺CCR6⁺CXCR3⁺ phenotype in the IgG-negative B cell compartment (NNB-22) that was not induced post-MMR3. This may represent a migratory IgA-positive MBC subset that homes to the respiratory mucosa via CXCR3⁶², and can migrate to inflamed or mucosal tissues through CCR6 to exert a recall response⁶³. Furthermore, our ex vivo immunophenotyping approach revealed the induction of a CD27⁻CD38⁺CD24⁺ antibody-secreting cell (ASC) subset (NNB-22) following mumps virus infection. This subset shares characteristics with a recently described cell population of circulating CD27⁻ ASC identified by Covens et al., who suggested that these cells represent activated MBCs undergoing differentiation into plasmablasts or plasma cells⁶⁴. Overall, based on the parameters that were analyzed and except for the absence of a prominent CD8⁺ T cell response, MMR immunization induced a similar type of immunity profile, yet lower in magnitude, to that of a natural infection. Even the strongest MMR-responders in subpopulation IV, who exhibited the most robust functional CD4⁺ T cell responses, did not reach the same magnitude of IgG levels, VN titers and IgG⁺ B cell frequencies as observed following MVI.

While immunophenotyping provides valuable information on overall immune cell composition, it should be taken into account that it does not capture antigen-specific responses, especially at later timepoints when such cells are expected to be rare. We are aware that the findings of our ex vivo immunophenotyping are subjected to timing of the snapshots that were taken and that other sampling times could have revealed additional subsets. Furthermore, given the use of different individuals and the large variation within groups, we cannot fully exclude that observed differences may be attributable to inter-individual variability rather than solely to infection versus vaccination response. Although statistically significant correlations were observed between immune cell subset frequencies and mumps-specific immune responses, the modest r values and the low frequencies of certain subsets suggest that the clinical relevance of these relationships is probably limited and should be interpreted with caution. This underscores the complexity of the immune response and the potential contribution of additional factors beyond those measured in this study. Another pitfall of this study is that the eventual group sizes of the detected subpopulations are small, thus decreasing the power of the findings. However, the division of the non-responding subpopulation (V) and separation between high (III + IV) and low (I + II) levels of pre-existing humoral immunity could also be made with a larger group size (n = 147) of the total MMR3 cohort. Future studies could focus on a larger inclusion of specific subpopulations, based on pre-existing immunity levels, to increase the power. In the coming years, it will be particularly interesting to see whether the proportion of individuals with low vaccine-induced immunity levels (subpopulations I and II) increases. This question becomes especially relevant with the recent change in the Dutch National Immunization Program, which moved the MMR2 immunization from age 9 to 3 years, thereby increasing the interval between MMR2 administration and the period of risk in young adulthood.

In summary, our comprehensive immunophenotyping approach enabled the identification of distinct immune cell subsets associated with responses to mumps virus infection and MMR3 vaccination. We observed that natural mumps virus infection elicited markedly stronger and more sustained immune responses compared to the third MMR dose. Nevertheless, MMR3 still enhanced both humoral and cellular immunity in young adults, suggesting it is an effective intervention to boosting immunity. Notably, we observed clear differences between low- and high-responding subpopulations. These insights may inform the development of targeted vaccination strategies by identifying individuals with suboptimal responses who could benefit most from additional immunization.

Author contributions

P.K. conceived and supervised the project. R.H.M.R. and P.K. designed experiments. R.H.M.R., M.E.E., P.M., I.T., H.I.T.H., and M.V. conducted experiments. R.H.M.R. and S.S.H. designed data analysis strategy and tools. R.H.M.R., M.E.E., M.V., and S.S.H. analyzed data. R.H.M.R. and S.S.H. visualized data. R.H.M.R. wrote the first draft of the paper and all other authors critically reviewed it. This work was supported by the Dutch Ministry of Health, Welfare and Sport.

Peer review

Peer review information

Communications Medicine thanks Richard Kennedy and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the Dutch Ministry of Health, Welfare and Sport.

Data availability

The data underlying this manuscript are available in Zenodo via https://doi.org/10.21945/6c125e0d-4f36-42ad-8dbf-730b5ffb3006, but restrictions apply to the availability of these data. Data are available on reasonable request via this link at the Zenodo platform, subject to review by the corresponding author, in accordance with participant consent and institutional guidelines.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s43856-026-01801-5.