Evaluation of B-cell kinetics after acellular pertussis vaccination in four cohorts of different age and priming background

: Pertussis is a vaccine-preventable disease caused by the bacterium Bordetella pertussis. Over the past years, the incidence and mortality of pertussis increased significantly. A possible cause is the switch from whole cell to acellular pertussis vaccines, although other factors may also contribute. To develop future vaccines and improve current vaccination strategies, it is critical to understand factors influencing the generation of immunological memory. We applied high-dimen-sional flow cytometry to investigate changes in B cells in individuals of different ages and distinct priming backgrounds upon administration of an acellular pertussis booster vaccine. These findings were correlated to vaccine-specific plasma cells and serum Ig levels. Expansion and maturation of plasma cells 7 days post-vaccination was the most prominent cellular change in all age groups, and was most pronounced for more mature IgG1+ plasma cells. Cellular responses were stronger in individuals primed with whole cell vaccine than in individuals primed with acellular vaccine. More-over, IgG1+ plasma cell expansion weakly correlated with Prn- and PT- specific serum IgG levels. Our study points at plasma cells as a potential early cellular marker of an immune response and contributes to understanding differences in immune responses between age groups and priming backgrounds.

Immune surveillance data showed that despite high vaccination coverage in many countries, there has been an increase of pertussis in the past decennia [6,7]. This increase is not only seen in aP-using countries, but it has also been reported in countries where primarily wP is used [6,8]. Several possible explanations for this increase have been proposed. First of all, improved awareness, surveillance and diagnostics may increase the detection rate [9]. Furthermore, several new Bp strains have been identified, i.e. that lack antigens present in the aP vaccine (such as FHA or Prn deficient strains), or PtxP3 strains that have adapted to suppress host immunity by producing higher levels of PT [10][11][12][13]. Lastly, there may be increased carriership within the population as well as faster waning of protective immunity in aP-primed individuals. Initial studies comparing efficacy of aP vs wP vaccines showed a similar short-term protection [14,15]. However, later long-term studies showed that protection lasted shorter when using aP vaccines [16][17][18][19]. Also baboon models have shown that aP-induced immunity does not prevent transmission, immunity induced by wP vaccines leads to a faster clearance of bacteria, and immunity generated by infection prevents colonization [20]. These combined data point at the need for mucosal immunity to prevent or reduce colonization and carriership.
An improved vaccine, immunization program, and/or route of administration seem necessary to combat pertussis. This implies a need to first understand the mechanism underlying protection induced by aP and wP vaccines (their differences and similarities). So far, no true correlate of protection (neither serological nor cellular) has been established for pertussis, and this would greatly aid evaluation of newly developed vaccines. This is one of the pillars of the Innovative Medicines Initiative (IMI)2 PERISCOPE consortium (PERtussIS COrrelates of Protection Europe), which aims to increase the scientific understanding of pertussis-related immunity in humans, identify new biomarkers of protection and to generate technology and infrastructure for the future development of improved pertussis vaccines [21].
Several (recent) studies both within and outside the IMI2 PERISCOPE consortium have shown that initial priming against pertussis (aP or wP vaccine) influences protection against disease as well as the immune response to (future) booster vaccinations [16,18,[22][23][24][25]. For example, Hendrikx et al found that in aP-primed children, Ag-specific IgG4 serum levels were higher compared to wP-primed children [23]. Furthermore, Da Silva et al showed that -even after receiving aP booster vaccinations-initial priming (wP or aP vaccine) determined the antigen-specific CD4 T-cell response [24]. Similarly, Lambert et al showed that CD4 T cells isolated from recently aP-boostered individuals could be separated in a PCA view based on priming background, where an aP priming background resulted in a more Th2-related response as compared to a wP priming background [22].
As neither vaccine-induced nor infection-induced immunity leads to life-long protection against pertussis, the use of booster vaccinations later in life is a topic relevant for public health, as people with waned immunity can become carriers and thereby a source of transmission. Moreover, older adults can be more vulnerable to severe disease outcomes [26]. Several studies have shown that aP boosters are effective and well-tolerated in (older) adults [27,28]. Recently, Versteegen et al investigated the specific serological response to an aP booster vaccination in four cohorts of different ages and priming backgrounds in the Netherlands, Finland and the UK (IMI2 PERISCOPE study acronym: BERT; Booster pertussis vaccination study) [29]. Here, they found all age cohorts showed a good response upon booster vaccination, with only limited differences between the different age cohorts for the Bp-specific IgG levels. However, Ag-specific serum IgA (both pre-and post-vaccination) increased with age, likely caused by over-time (mild) exposures to Bp. Previous studies on influenza have shown that up to 80% of the circulating IgG plasma cells 7 days after vaccination can be vaccine-specific [30,31]. This, combined with the low numbers of circulating plasma cells at baseline (median counts <5cells/ul [32]), implies that the plasma cell system is a relatively 'clean' system to monitor. Thus, flow cytometry may serve as a faster and less laborious approach to study vaccination-elicited plasma cells than typically Ag-specific approaches such as ELISpot. Recently, our team used high-dimensional flow cytometry to investigate over-time cellular kinetics in 10 healthy (wP-primed) adults upon aP vaccination. There, we were able to demonstrate a clear expansion and maturation of plasma cells (especially IgG1+) and a strong correlation between IgG1+ memory B-cell expansion and the magnitude of the Ag-specific IgG serum response [33]. Here, we extended our exploratory studies by the analysis of participants of different ages and different priming background after receiving an aP booster (Boostrix-IPV, GlaxoSmithKline (GSK), Wavre, Belgium). We included 48 individuals enrolled in the Dutch cohort used in the IMI2 PERISCOPE-BERT study (periscope-project.eu/patients/study-2-bert/) at pre-defined time points, with the primary objective to describe the kinetics of circulating B-cell populations in four cohorts of different ages and with different priming backgrounds. This may lead to new insights on the impact of priming at infancy with aP or wP vaccines on future immune responses.

Study design and sample collection
This study comprises one of the exploratory objectives of the Dutch 'BERT study', which was initiated by the IMI2 PERISCOPE consortium. It was approved by Medical Research Ethics Committees United (MEC-U, NL60807.100.17-R17.039) and registered at the EU Clinical trial registry (EudraCT number 2016-003678-42). To be eligible for this study, participants had to (1) be generally healthy, (2) have no recent evidence of serious disease, i.e. requiring the use of immunosuppressive or immunomodulating medication, within the 3 months prior to inclusion, (3) received all regular vaccines according to Dutch National Immunization Program (www.rivm.nl/en/national-immunisation-programme) as appropriate for their age. An extensive description of the cohort and exclusion criteria have recently been published by Versteegen and colleagues [29]. A fraction of the participants of this aP vaccination study was subjected to additional exploratory analysis, such as mass cytometry, evaluation of mucosal antibodies, NGS or in-depth flow cytometry. The exploratory sub-study monitoring the fluctuations in circulating B-cell subsets after vaccination is discussed in this manuscript. Here, 48 individuals were selected from 4 cohorts of different age and distinct priming background at infancy (children, 7-10 y/o, aPprimed; adolescents, 11-15 y/o, aP or wP-primed; young adults, 20-34 y/o, wP-primed; older adults 60-70 y/o, in whom vaccination history was unknown (presumably: wPprimed or not vaccinated)).
The study was conducted by the Spaarne Academy (Spaarne Hospital, Hoofddorp, the Netherlands). Written informed consent was obtained at the start of the study. Participants were vaccinated intramuscularly with the Boostrix-IPV vaccine after their first blood donation (baseline). Boostrix-IPV is a reduced-antigen Tdap-IPV booster vaccine, which contains diphtheria toxoid (Diph) (≥2.5 Limit of flocculation (Lf)), tetanus toxoid (Tet) (≥5Lf), three Bordetella pertussis proteins, being PT (8µg), FHA (8µg), Prn (2.5µg) and inactivated poliovirus (Mahoney strain; 40 D-Antigen units (DU), MEF-1 strain; 8 DU, Saukett strain; 32 DU) and aluminum hydroxide as adjuvant [34]. Peripheral blood samples were collected in blood collection tubes using heparin as anticoagulant and in serum collection tubes at baseline, day 7 and 28 after vaccination. An additional peripheral blood sample was taken at day 14 in participants aged 20-34 and aged 60-70. Individuals were excluded and replaced by a new participant if a blood sample at day 0 or 28 could not be obtained.

Evaluation of antigen-specific immunoglobulin levels in serum
Serological analysis was performed in all collected samples. Levels of IgG directed against Tet, PT, FHA, Prn, and Fim2/3, and levels of IgA directed against PT, FHA, Prn and Fim2/3 were determined by multiplex immunoassay (MIA) at the Dutch National Institute for Public Health and the Environment (RIVM, The Netherlands) [35]. The serum antibody responses raised against Bp-antigens during the PERISCOPE-BERT study have been extensively discussed by Versteegen and colleagues [29].
Detection of vaccine-specific antibody-producing plasma cells and memory B cells Analysis of numbers of IgG and IgA producing plasma cells and memory B cells was performed in majority of the samples included in this study. B cells producing IgG directed against PT, FHA, Prn and Tet, and B cells producing IgA direct against PT, FHA and Prn were measured using an Enzyme-Linked Immunospot (ELISpot) assay at the RIVM. This procedure has been described previously [36]. In short, peripheral blood mononuclear cells (PBMCs) were isolated using a density gradient. For antibody producing plasma cells, PBMCs at day 7 post-vaccination were directly transferred to Ag-coated ELISpot filter plates (duplicates). For the detection of vaccine-specific memory B cells, PBMCs collected at day 0 and day 28 were collected and stored at -135°C. Thawed PBMCs were stimulated for 5 days using a culture medium containing CpG, IL-2 and IL-10. Next, cells were transferred to Ag-coated plates (duplicates). Numbers of Ag-specific antibodyproducing cells -appearing as spots-were measured using an ImmunoSpot S6 Ultra-V analyzer (Cellular Technology Limited, Cleveland, OH). Uncoated wells filled with PBS served as negative control and were used to subtract background signal. Wells with a signal below the limit of quantification were set at 0.1 cell/105 PBMCs. Cumulative IgG and IgA spot counts for all antigens measured were used for analyses.

Longitudinal flow cytometric analysis of circulating B-cell subsets
All peripheral blood samples were subjected to high-throughput flow cytometric immunophenotyping of the B-cell compartment. Here, we used the recently developed BIGH-tube: the B-cell and plasma cell tube (BIGH) allows identification of >100 populations of B-and plasma cells distinguished based on their maturation stage and expressed Ig subclasses [32, 37] (Antibody panel and phenotypic description of the identified B-cell subsets: Supplemental Table 1 and 2).
Samples were processed according to the bulk lysis protocol using 10 x 106 cells followed by intracellular staining as described before [33] (protocols available on www.Eu-roFlow.org), with the addition of membrane staining with CD45-AlexaFluor700.
In short, based on the white blood cell count (as determined by an automated hematological analyzer (Sysmex XP-300, Sysmex Europe GmbH, Norderstedt, Germany)), one or multiple tubes were filled with up to 2mL of blood, after which ammonium chloride was added up to a total volume of 50mL. After a 15 min incubation at room temperature on a roller bank to lyse non-nucleated red blood cells, cells were washed, counted on a Sysmex XP-300 and pooled to a total of 10x106 cells. Next, cells were stained with an antibody cocktail directed against surface markers for 30 min in the dark with the BIGH panel (Supplemental Table 1). This was followed by a cytoplasmic staining for intracellular Igs using the Fix & Perm reagent kit (Nordic MUbio, Susteren, The Netherlands) according to manufacturers' protocol. Finally, samples were washed and resuspended in PBS for immediate acquisition (or stored for max ~ 3h at 4°C).
For precise enumeration of cell numbers, we used Perfect-Count Microspheres™ (Cytognos) according to the EuroFlow SOP (protocol available on www.EuroFlow.org). In short, exactly 50ul of well-mixed Perfect-Count Microspheres™ were added to exactly 50ul of peripheral blood. Then, antibodies directed against CD19, CD3 and CD45 were added and the sample was incubated for 30 min in the dark. Next, 500uL of NH4Cl was added and after 10 minutes incubation samples were ready for immediate acquisition. Using this tube, we could identify and quantify total leukocytes and lymphocytes, B-, Tand NK cells in each sample. All samples were acquired at the Flow cytometry Core Facility of LUMC, using a BD FACS LSR Fortessa 4L (BD Biosciences, San Jose, CA, USA) or a BD FACS LSR Fortessa X-20 flow cytometer (BD Biosciences), which were calibrated daily according to EuroFlow guidelines, as previously described [38,39].

Data analysis and statistics
To ensure objective data analysis and minimize operator-induced variability, all data were analyzed using the automated gating and identification (AGI) module of the Infinicyt software (Infinicyt™ Software v2.0, Cytognos). This AGI module makes use of clustering algorithms and comparison with fully annotated reference flow cytometry (FCS) data files of healthy individuals to assign clusters of events to a population [40] . Importantly, when there was no perfect fit for a cluster of events, this was marked as a 'check' population and the software indicated to which populations this cluster may correspond. These check events were assigned manually according to the proposed gating strategies for the BIGH panel (Supplemental Table 2) [32,37].
For visualization and statistical analysis, the GraphPad Prism 8.1.1 software (GraphPad, San Diego, CA, USA) was used. To test longitudinal data within a cohort, the Wilcoxon signed-rank test for paired samples was used. Correlations were determined using Spearman's Ranking Correlation (correlation coefficients of <0.6 or -in case of negative correlation >-0.6 were considered weak correlations). Finally, to compare differences between the four cohorts, the Kruskal-Wallis approach was used, followed by Dunn's test, and, as only BU and BV samples were measured at day 14, the Mann-Whitney test was used instead of Kruskal-Wallis at day 14. To correct for multiple testing, tests were followed by Bonferroni correction when appropriate.

Study cohorts
All subjects enrolled in the study between October 2017 and March 2018. In total, 12 children (age: 7-10, aP-primed, m/f ratio: 6/6), 12 adolescents (age: 11-15, 7 individuals wPprimed, m/f ratio: 2/5, 5 individuals aP-primed, m/f ratio: 4/1), 12 young adults (age 20-34, wP-primed, m/f ratio: 7/5) and 12 older adults (age 60-70, presumably wP-primed or not vaccinated, m/f ratio: 4/7) completed this study (as part of the PERISCOPE BERT study). Three additional children, who were initially enrolled, did not participate beyond the first visit and were replaced by three new participants. From all acquired samples, two baseline B-cell samples were lost due to technical problems (one child and one young adult). Finally, one older adult was excluded due to (potentially) clonal expansion of B cells and replaced by a new participant.
For most participants the leukocyte, lymphocyte, T-cell, B-cell and NK-cell counts at baseline were within the normal age-matched range (Supplemental Table 3, Table 1), or, in case of minor deviations, fell into the normal range at later time points [41,42]. Leukocytes, lymphocytes and T cells remained mostly stable over the time of analysis. Although NK-cell numbers showed a minor decrease at day 28, this was most likely not related to the vaccination response (day 0 vs 28, p≤0.01 and day 7 vs 28, p≤0.05, Supplemental Figure 1). There were no statistically significant differences in absolute leukocyte, lymphocyte, T-cell and NK-cell counts at baseline in-between age cohorts. Thus, regarding the numbers of leukocytes, lymphocytes, B cells, T cells and NK cells, our donors were healthy representatives of the general population.    10 B-cell numbers are known to decrease over time from on average 1400 cells/μl in children 11 <2 years to on average 200 cells/μl in adults [32,41]. This trend was also visible in our data 12 set where children had more B cells than the adult groups (Table 1). This difference was 13 mainly due to high numbers of pre-germinal center (naive) B cells in children. Although 14 memory B-cell numbers were also higher in children than in adults (160 cells/μl in chil-15 dren vs 85.7 cells/μl in young adults, n.s., and 160 cells/μl in children vs 63.8 cells/μl in 16 older adults, p≤0.01), these differences were less prominent, and mainly restricted to 17 IgG1+ and IgG3+ memory B cells. Finally, several plasma cell subsets were significantly 18 more abundant in children than in adults (IgG1+, IgG3+ and IgD+ plasma cells), but, due 19 to their overall low frequencies, this did not have a major impact on total B-cell numbers. 20 Limited differences in B-cell subset numbers were observed between adolescents and 21 adults (  26 We have recently shown that administration of an aP booster vaccination in (wP-primed) 27 adults triggers several cellular changes, of which the expansion of (predominantly) IgG1+ 28 plasma cells at day 7 is most prominent [33]. Now, we set out to determine whether the 29 same type of changes occurs in vaccinated individuals, irrespective of age and primary 30 vaccination background. 31 While total B cells, pre-germinal center B cells and memory B cells remained relatively 32 stable and did not show any consistent fluctuations over time following vaccination, 33 plasma cells in all participants underwent significant expansion between baseline and day 34 7 (p≤0.01 for children, and p≤0.001 for adolescents and the adult cohorts, Figure 1), which 35 returned to baseline at day 14 or day 28. The magnitude of this expansion was highly 36 similar in children, adolescents and young adults (ratio to baseline: 2.6-3.4) and was sig-37 nificantly higher in older adults as compared to children (ratio: 2.6 in children vs 5.7 in 38 older adults, p≤0.01). Total plasma cell numbers returned to baseline at day 14 or -if day 39 14 was not measured-day 28 (p≤0.001 for adolescents, young and older adults, in the chil-40 dren cohort, plasma cell counts were still slightly elevated at day 28). Thus, the expansion 41 of plasma cells 7 days post-vaccination was the most prominent change in all age groups. 42   59 Cell expansion at day 7 was not equally pronounced in all plasma cell subsets. It was the 60 most prominent in IgG1+ plasma cells (ratio to baseline: 5.7-17.6, depending on cohort, 61 p≤0.001 in all cohorts, with a higher increase in older adults compared to children, p<0.05). 62 This was followed by IgG3+ plasma cells (ratio: 3.6-6.5, p≤0.01 in children and adolescents, 63 and p≤0.001 in young and older adults, Figure 2A). IgG4+ plasma cells were significantly 64 increased in children and adolescents (ratio: 5.1 and 7.2 respectively, p≤0.01). IgG2+ 65 plasma cells were significantly increased in adolescents only (ratio:1.8, p≤0.001). IgA1+ 66 plasma cells were significantly increased in both adult cohorts (ratio: 2.1 and 3.9 in 67 younger and older adults, p≤0.01 and p≤0.001, respectively). Although IgM+ plasma cells 68 showed a tendency to expand at day 7 in older adults, there was a large variation in-be-69 tween individuals and the difference was not statistically significant. 70 Despite individual changes in response patterns between the cohorts, IgG1+ plasma cells 71 were the single most expanded subset. They constituted between 43% (young adults) and 72 61% (adolescents) of all plasma cells at the peak of expansion, while only at most 18% at 73 baseline ( Figure 2B). Moreover, although the ratio (to baseline) of plasma cells was often 74 higher in adults than in children, children had higher baseline IgG1+ plasma cell numbers, 75 and also higher IgG1+ plasma cell numbers at day 7 (median cell count at day 7: 7.64 76 cells/µl in children vs 3.81 cells/µl in older adults, ns). Thus, the expansion of IgG1+ 77 plasma cells was the most prominent in all cohorts. Both an increase over baseline (ratio) 78 and in absolute cell numbers were observed.   94 Newly generated plasma cells migrate from germinal centers via the blood stream to be-95 come long-lived antibody-secreting plasma cells in the bone marrow and other peripheral 96 tissues. Over time, they gradually lose expression of CD20 and gain expression of CD138 97 [43] ( Figure 3A). We used this information to divide plasma cells into consecutive matu-98 ration stages and to trace their maturation over time after booster vaccination. 99 Plasma cells representing all maturation stages were expanded at day 7 (Supplemental 100 Figure 2 for total plasma cells, Figure 3 for IgG1+ plasma cells). This expansion was limited 101 in the least mature CD20+CD138-plasma cells (ratio to baseline: up to 3.1 in IgG1+ plasma 102 cells in children, Figure 3B)  Figure 3C). Thus, the expansion of 111 plasma cells at day 7 post-vaccination was accompanied by a shift towards a more mature 112 plasma cell phenotype in all cohorts. 113 As in the adult cohorts an expansion of IgA1+ plasma cells was observed, we evaluated 114 maturation of IgA1+ plasma cells at the peak of expansion as well. In children, no increase 115 in more mature (CD20-CD138-and CD20-CD138+) IgA1+ plasma cells was observed at 116 the peak of expansion. In adolescents, a small increase of most mature (CD20-CD138+) 117 IgA1+ plasma cells was observed, and in both adult cohorts, an increase in both interme-118 diate and most mature (CD20-CD138-and CD20-CD138+) IgA1+ plasma cells was ob-119 served at the peak of plasma cell expansion (adolescent cohort: p<0.05, adult cohorts: 120 p≤0.01). When comparing all cohorts at 7 days post-vaccination, a higher expansion of 121 most mature IgA1+ plasma cells was observed in older adults as compared to children 122 (p≤0.01). Thus, the expansion and maturation of IgA1+ plasma cells seemed to increase 123 with age of the cohort. However, within the oldest cohort, no correlation was found be-124 tween total IgA plasma cells, IgA1+ plasma cells, or vaccine-specific IgA cells and age. Median values for each population were 132 used to construct the plots. Wilcoxon matched pair signed-ranked test followed by Bonferroni correction was used to assess 133 longitudinal differences in percentage of CD20-CD138+ cells in total IgG1+ plasma cells within each cohort. Differences in the 134 percentage CD20-CD138+ cells in total IgG1+ plasma cells between cohorts were assessed using Kruskal-Wallis followed by 135 Dunn's test, but did not yield significant differences. For pediatric cohorts, no blood samples were collected at day 14 147 shown that even minor expansions in circulating memory B cells can strongly correlate 148 with a post-vaccination increase in Bp-specific serum Ig levels [33]. Therefore, we set out 149 to determine whether any quantitative changes can be observed at here-selected time 150 points and whether the same pattern is shared by different cohorts. 151 Neither total memory B cells nor any of the major memory B-cell subsets underwent con-152 sistent quantitative changes over time following vaccination ( Figure 4A), with exception 153 of IgG4+ memory B cells in adolescents, which showed a minor but significant increase at 154 day 28 compared to baseline levels. Moreover, at 14 days post-vaccination, there was a 155 minor but significant difference between the number of IgG3+ memory B cells between 156 young and older adults, although there were no significant longitudinal changes within 157 any of the groups. Within each cohort the distribution of memory B-cell subsets was stable 158 over the time of analysis ( Figure 4B). However, upon further subdivision of memory B 159 cells based on the expression of CD20, CD21, CD24 and CD27, two memory B-cell subsets 160 underwent significant fluctuations over time (Supplemental Figure 3). In adolescents, 161 there was a significant increase in IgG1+ CD20++CD21-CD24+ memory B cells 28 days 162 after vaccination. Moreover, in adolescents and older adults, there was a significant in-163 crease of IgG1+ CD20++CD21-CD24-CD27+ memory B cells 14 or 28 days after vaccination 164 as compared to baseline. Therefore, it is possible that memory B cells specific to Boostrix-165 IPV vaccine reside within these CD20++CD21-memory B cells. Here, the use of an antigen-166 specific approach should lead to additional insights. No significant differences were ob-167 served between the cohorts. Although most of subsets defined within IgG1+ memory B 168 cells were significantly more numerous in children than in both adult cohorts (data not 169 shown), this was mainly due to higher numbers of IgG1+ memory B cells in children at 170 baseline. Thus, except for a few minor fluctuations, no differences in the number of 171 memory B cells were observed after Tdap booster vaccination. Median values for each population were used to construct the plots Wilcoxon matched pair signed-ranked test followed 177 by Bonferroni correction was used to assess longitudinal differences within each cohort. Differences between cohorts were 178 assessed using Kruskal-Wallis followed by Dunn's test, but did not yield significant differences. For pediatric cohorts, no 179 blood samples were collected at day 14. There, the Mann-Whitney test followed by Bonferroni correction was used. For longi-180 tudinal changes, only significant differences compared to baseline are shown. Significant differences between cohorts at the 181 same time point are indicated with #, p<0.05; ##, p≤0.01; ###, p≤0.001. d= days after vaccination.  For memory B cells, no correlation was observed between ELISpot readout and flow cy-214 tometry readout (Supplemental Figure 4). Neither flow cytometry IgG nor IgA memory 215 readout correlated with vaccine-specific serum Igs at day 28 or year 1 (data not shown). 216 These findings indicate that for analysis of memory B cells by flow cytometry an Ag-217 specific approach is required. As differences in cellular changes could be found between 218 the different age groups, we also tested for correlations between cell expansion (ratio over 219 baseline) and vaccine specific serum Ig (IU/mL) within each age group (Supplemental 220 Figure 5). No correlations between cell expansion and serum Ig levels were found. Thus, 221 the expansion of plasma cells as measured by flow cytometry seems reflective of the num-222 bers of Ag-specific plasma cells, and in this regard flow cytometry and ELISpot can pro-223 vide complementary data. This is not observed for memory B cells. Over time changes in total, naive and memory B cells were minor and comparable be-246 tween wP and aP-primed groups ( Figure 6A). In contrast, changes in plasma cells were 247 much more prominent in the group which was primed with the Dutch wP vaccine (ratio 248 to baseline: 4.8 vs 1.5 at day 7 in total plasma cells). From all plasma cells, the differences 249 were the clearest for IgG1+ (ratio: 23.9 vs 4.5 at day 7), IgG3+ (ratio: 19.7 vs 3.7 at day 7) 250 and IgG4+ (ratio: 9.2 vs 3.2 at day 7) plasma cells. Only the difference in IgG1+ plasma cell 251 expansion reached statistical significance (p≤0.01), possibly due to the low number of par-252 ticipants in both groups. Moreover, plasma cell maturation at the peak of expansion was 253 more prominent in wP-primed adolescents in whom most mature CD20-CD138+ plasma 254 cells constituted 46% of IgG1+ plasma cells in contrast to 39% in aP-primed adolescents 255 ( Figure 6A+B). Thus, the type of priming background seems to influence the plasma cell 256 response to later booster vaccinations. In our study, the plasma cell response was stronger 257 and more diverse in wP-primed adolescents. 258 To exclude that differences observed in the adolescent cohort were caused by the different 259 sex distribution between aP-and wP-primed individuals, we assessed the impact of sex 260 on the cell expansion in the young adult cohort and extrapolated this to the adolescent 261 cohort. The young adult cohort was well sex-balanced and had a homogenous priming 262 background. We evaluated the expansion of total B cells, plasma cells, IgG1-3+ plasma 263 cells, and IgA1+ plasma cells in males and females. No significant differences were ob-264 served between male and female responses (Supplemental Figure 6). Therefore, we con-265 cluded that the sex imbalance in the adolescent cohort did not influence the plasma cell 266 expansion as found in this study. 267 The adolescent cohort only consisted of 12 individuals, which is rather small for statistical 268 analysis. Therefore, although we found some age-dependent differences in cellular re-269 sponses, we next grouped all aP-primed (children and adolescents) and all wP-primed 270 individuals (adolescents, younger and the presumably wP-primed older adults) to in-271 crease the size of the study cohorts (Supplemental Figure 7A). Also in this comparison, 272 expansion of total and IgG1+ plasma cells was more pronounced in wP-primed partici-273 pants. Differences in other plasma cell subsets did not reach statistical significance. Fi-274 nally, based on ratio to baseline the (total) plasma cell maturation was more prominent in 275 wP-primed individuals as compared to aP-primed individuals, nevertheless the percent-276 age of most mature cells in the IgG1+ plasma cell population did not significantly differ 277 between the two groups (Supplemental Figure 7). 278 In summary, here we showed that irrespective of the age of vaccinated individuals, the 279 most prominent cellular changes occurred numbers of in circulating plasma cells. Median values for each population were used to con-291 struct the plots. Wilcoxon matched pair signed-ranked test followed by Bonferroni correction was used to assess 292 longitudinal differences in percentage of CD20-CD138+ cells in total IgG1+ plasma cells within each cohort. Dif-293 ferences in the percentage CD20-CD138+ cells in total IgG1+ plasma cells between cohorts were assessed using 294 Kruskal-Wallis followed by Dunn's test, but did not yield significant differences. For longitudinal changes, only 295 significant differences compared to baseline are shown. Significant longitudinal differences within a cohort as 296 indicated with *, p<0.05; **, p≤0.01; ***, p≤0.001. Significant differences between cohorts at the same time point are 297 indicated with #, p<0.05; ##, p≤0.01; ###, p≤0.001. d= days after vaccination; aP = acellular pertussis vaccine; wP = 298 whole cell pertussis vaccine; pre-GC = pre-Germinal Center.

304
In this study, we applied high-dimensional flow cytometry to investigate changes 305 in B cells in individuals of different ages and priming backgrounds upon administration 306 of an aP booster vaccine and correlated these findings with vaccine-specific Ig levels in 307 serum. In all age groups expansion and maturation of plasma cells 7 days post-vaccination 308 was the most prominent cellular change. Although in children the expansion of plasma 309 cells was less prominent than in adults, when expressed as ratio to baseline, they had more 310 plasma cells at peak levels due to their initially high plasma cell numbers. Furthermore, 311 cellular responses were stronger in individuals primed with the Dutch wP vaccine than 312 in individuals who were initially primed with aP vaccines. No consistent over-time 313 memory B-cell fluctuations were observed. No direct correlation between plasma cell ex-314 pansion or memory B-cell expansion with vaccine-specific serum Ig levels was observed. 315 Nevertheless, weak positive correlations were observed between the expansion of IgG1+ 316 plasma cells and Prn-and PT-specific serum IgG levels post-vaccination. Although serol-317 ogy provides insight in Ag-specific Ig levels and function, analysis of circulating immune 318 cells may result in a deeper understanding of the processes induced by the vaccine and 319 the cellular changes preceding Ig production. Our study points at plasma cells as a poten-320 tial cellular marker of an immune response and contributes to a better understanding of 321 differences in immune responses between age groups and primary vaccination back-322 grounds. 323 To ensure objective data analysis, we used the automated gating and identification 324 (AGI) tool in the Infinicyt software, which has been demonstrated to reduce intra-and 325 inter-operator variability and increase reproducibility of the analysis [40, 48-50]. This is 326 especially important for studies with big data from multiple samples, which cannot be 327 analyzed by a single operator within a reasonable timeframe. Irrespective of the new anal-328 ysis strategy, these data corroborated major findings from our previous study, where data 329 were subjected to manual analysis [33]. This automated analysis approach, in combination 330 with the standardized EuroFlow sample processing and acquisition procedures, allows 331 for identification of fluctuations in small populations of cells e.g. different plasma cell 332 maturation stages. 333 Levels of Ag-specific serum Igs are routinely used as readout for vaccine efficacy. In 334 many cases a rise in Ag-specific IgG levels is associated with response to vaccination, and 335 for several vaccines, e.g. against rotavirus, an increase in IgA levels has been indicated as 336 correlate of protection [23,45,[51][52][53]. As Igs are the product of terminally differentiated B 337 cells (plasma cells), the B-cell compartment may harbor new correlates or biomarkers of 338 ongoing immune responses. Indeed, we found that expansion and maturation of circulat-339 ing plasma cells 7 days after booster vaccination was the most prominent cellular change. 340 The generation of mainly IgG1+ plasma cells is in line with previous serology-based stud-341 ies where within Bp-specific serum IgGs mostly IgG1 antibodies were found, with minor 342 contribution of IgG2,-3 and -4 [51,54]. The positive correlation between the numbers of 343 total plasma cells with the vaccine-specific plasma cell numbers supports the assumption 344 that most of the plasma cells at the peak of expansion are vaccine specific. 345 The observed IgA response, mostly in the adult cohorts, is most likely a result of 346 immunological memory generated by previous (subclinical) infection of the respiratory 347 tract, where a mucosal response against Bp was launched. As Bp circulates within the 348 population, causing outbreaks every 2-5 years, the adult cohorts likely encountered Bp 349 multiple times during life, which explains the more prominent IgA1+ plasma cells re-350 sponse in these groups [55][56][57][58]. In contrast, the expansion of IgG4+ plasma cells was mostly 351 seen in the pediatric cohorts, which may be explained by the predominantly aP priming 352 in these cohorts; this has been shown to induce a more Th2-related response as well as 353 increased vaccine-specific serum IgG4 [23,24]. 354 In addition to the expansion of IgG1+ and IgA1+ plasma cells in adults, and the IgG1+ 355 and IgG4+ response in the pediatric cohorts, which are in line with previous (cellular and 356 serology-based) studies, we also observed a prominent increase in IgG3+ plasma cells in 357 all cohorts [23,33,51,54]. A potential explanation for this phenomenon might be that -in 358 addition to the memory B cells-there are naive B cells recognizing the antigens and un-359 dergo first-step IgG3 class switching and affinity maturation [59]. 360 The difference in plasma cell -and thus antibody-production can have consequences 361 for the type and efficacy of the launched immune response. IgG1 and IgG3 antibodies 362 have stronger opsonizing capacities as compared to IgG4 antibodies [60]. The mixed IgG1-363 IgG3-IgG4 response observed in the mostly aP-primed pediatric cohorts may lead to com-364 petition for Bp antigens in future encounters, possibly leading to less efficient bacterial 365 clearance as compared to the IgG1-IgG3 (and IgA1) response observed in the adult cohorts 366 [59,61]. Lastly, the prominent contribution of IgA1+ plasma cells to responses observed 367 in the adult cohorts, which is likely an indicator of previous pertussis encounters, may 368 imply existence of effective mucosal defense mechanisms, and more efficient protection 369 against bacterial translocation in IgA-producing individuals. Comparison of repertoires 370 and reactivities of IgA in mucosa and in circulation could provide better insights into this 371 phenomenon and value of IgA as a biomarker of protection. 372 Maturation of plasma cells (total and IgG1+) was observed irrespective of age and 373 priming background. This clear expansion and maturation of total and IgG1+ plasma cells 374 is in line with our previous findings and may be explained by the prolonged retention of 375 newly generated plasma cells in the periphery as well as the competition for bone marrow 376 niches with pre-existing long lived plasma cells [33,62]. 377 In this study, several differences between the aP and wP-primed cohorts were ob-378 served. Although the size of the age-matched cohort was too limited to reach statistically 379 significant conclusions, major observations were confirmed by analysis of the total cohort, 380 divided based on the priming background. Remarkably, this difference based on priming 381 background was not observed in the overarching part of the BERT study, where the Bp-382 specific Ig responses of 85 Dutch and Finnish adolescents pre-and post-vaccination were 383 evaluated [29]. 384 The formulation of aP and wP vaccines differs with regards to number of antigens 385 and the total antigenic load, with wP vaccines containing the broad variety of pertussis 386 antigens and aP vaccines containing high concentrations of a restricted number of anti-387 gens. In consequence, wP priming is likely to trigger a more diverse antibody response. 388 Since consecutive boosters lead to a more specific, but also more restricted response, this 389 initial broad priming can be beneficial in case of encountering future (mutated) bacterial 390 strains [10,12,13]. Interestingly, in this study we showed that, compared to aP-primed 391 individuals, individuals primed with the Dutch wP vaccine have a stronger response 392 upon aP booster vaccination. It would be of interest to visualize potential differences in 393 breadth of an immune response against Bp antigens. Moreover, since the initial type of 394 priming seems to imprint future responses to given antigens, it should be carefully con-395 sidered in the design of future vaccines and vaccination strategies. This may also hold true 396 for diseases other than pertussis, such as COVID-19. 397 To identify unique and shared patterns in-between groups, we primarily focused on 398 normalized data (represented as the ratio to baseline). However, we also showed that, in 399 line with published studies, children had overall higher leukocyte and B-cell counts com-400 pared to (older) adults [32,41]. Specifically, the cell count of naive B and T cells -and thus 401 the available naive repertoire-is substantially higher in youth [32,48]. Therefore, despite 402 a lower increase in cells expressed as ratio to baseline, children and adolescents may still 403 produce a stronger and more diverse immune responses than adults. 404 In this study, memory B-cell fluctuations were limited. As the frequencies of Ag-spe-405 cific memory B cells are low, as demonstrated by previous studies using ELISpot assays, 406 an increase in only these Ag-specific memory B cells may not have an impact on the total 407 memory B-cell population [63,64]. Indeed, in this study no correlation was found between 408 the memory B-cell fluctuations measured by flow cytometry and the vaccine-specific 409 memory B cells by ELISpot. However, we observed an increase in CD20++CD21-IgG1+ 410 memory B cells in older adults and adolescents at day 14 and 28 after vaccination, respec-411 tively. Interestingly, several studies reported an increased percentage of Ag-specific 412 CD27+CD21-/dim B cells 14 days after influenza vaccination [65,66]. There is no 413 consensus about the exact function of these CD21-/dim B cells, but, it has been suggested 414 that CD21-/dim B cells are exhausted cells or -as described in autoimmunity and chronic 415 infection-are anergic [67,68]. In this context it is not unlikely, that cells which responded 416 to an antigen multiple times would acquire this phenotype. However, Lau and colleagues 417 suggest that CD21-/dim B cells are primed for plasma cell differentiation [65]. Ag-specific 418 flow cytometry studies should give insight into the exact function of this B-cell subset. 419 No correlation was found between expansion of memory B cells and Ag-specific se-420 rum IgG levels at day 28. Previously, we observed a clear correlation between the expan-421 sion of IgG1+ memory B cells and the vaccine-specific IgG levels at day 21 in a cohort of 422 10 healthy adults [33]. Moreover, we found that although in the majority of participants 423 memory B cells showed maximum expansion at 14 days after vaccination, the expansion 424 of memory B cells was not as synchronized in time as the plasma cell expansion, implying 425 that -in some participants-we may not have sampled at the most optimal timepoint [33], 426 especially in the children and adolescent cohorts, where the sampling times were limited 427 to day 0, 7 and 28. This difference in timing of memory B-cell responses might be related 428 to the immune status of each individual at baseline and it makes the use of memory B cells 429 as correlates of protection more difficult. 430 Neither aP nor wP vaccination yields a response that fully mimics natural infection, 431 especially the IgA response seems to be limited upon vaccination and mostly relies upon 432 previous encounters with Bp.  Dashed lines indicate a ratio 455 of 0.67 and 1.5 compared to baseline. Underneath each graph, the baseline cell counts per cohort are 456 presented in cells/µL (median, min-max). To assess longitudinal changes within each cohort, Wil-457 coxon matched pair signed-rank test followed by Bonferroni correction was used. To test differences 458 between cohorts at one timepoint, Kruskal-Wallis followed by Dunn's test was used, with exception 459 of the comparison at day 14. At day 14, only blood samples from the adult cohorts were collected. 460 Here, the Mann-Whitney test followed by Bonferroni correction was used. For longitudinal changes, 461 only significant differences compared to baseline are shown. *, p<0.05; **, p≤0.01.  464 The arrow indicates the direction of changes during maturation. (B) Over-time quantitative changes 465 in plasma cells belonging to different maturation stages, presented as ratio over baseline (median, 466 min-max). Dashed lines indicate a ratio of 0.5 and 2.0 compared to baseline. Underneath each graph, 467 the baseline cell counts per cohort are indicated in cells/µL (median, min-max). (C) Over-time dis-468 tribution of plasma cells representing different maturation stages within total plasma cells. Median 469 values for each population were used to construct the plots. Wilcoxon matched pair signed-ranked 470 test followed by Bonferroni correction was used to assess longitudinal differences in percentage of 471 CD20-CD138+ cells in total plasma cells within each cohort. Differences in the percentage CD20-472 CD138+ cells in total plasma cells between cohorts were assessed using Kruskal-Wallis followed by 473 Dunn's test, but did not yield significant differences. At day 14, only blood samples from the adult 474 cohorts were collected. Here, the Mann-Whitney test followed by Bonferroni correction was used. 475 For longitudinal changes, only significant differences compared to baseline are shown. Significant 476 longitudinal differences within a cohort as indicated with *, p<0.05; **, p≤0.01; ***, p≤0.001. Signifi-477 cant differences between cohorts at the same time point are indicated with #, p<0.05; ##, p≤0.01; ###, 478 p≤0.001.  Figure 5. Correlation between cellular changes and the vaccine-specific serum IgG 497 level post-vaccination as determined by Spearman's Ranking Correlation per age cohort. Per cohort 498 the left plot shows the correlation between the maximum expansion of IgG1 plasma cells (day 7) 499 and vaccine-specific serum IgG (directed against FHA, Prn, PT and Tet) (day 28). The right plots 500 show the correlation between the maximum expansion of IgG1 memory B cells (day 14 or day 28) 501 and vaccine-specific serum IgG (directed against FHA, Prn, PT and Tet) (day 28). MBC = memory B 502 cell; r= Spearman's correlation coefficient; d= days after vaccination. 503 Supplemental Figure 6. Impact of sex on cellular responses after vaccination in the young adult 504 cohort (all wP-primed). Flow cytometry-derived cell numbers (absolute count in cells/ul (A) and 505 ratio over baseline (B)) and their changes over time in an age-matched, wP-primed male (n= 7) and 506 female (n= 5) cohort. Of note, for one male participant, no baseline B-cell data was available. There-507 fore, in the graphs showing the ratio over baseline, data of 6 males are shown, whereas absolute 508 counts include the data of 7 males. Graphs indicate median + range. Dashed line indicates ratio of 509 1.0 (baseline value). plasma cell subsets in aP-primed (12 children + 5 adolescents) and wP-primed (7 adolescents, 12 513 young adults and 12 older adults) donors. (B) Over-time distribution of IgG1+ plasma cells repre-514 senting different maturation stages with total IgG1+ plasma cells. Median values for each popula-515 tion were used to construct the plots. Wilcoxon matched pair signed-ranked test followed by Bon-516 ferroni correction was used to assess longitudinal differences in percentage of CD20-CD138+ cells 517 in total IgG1+ plasma cells within each cohort. Differences in the percentage CD20-CD138+ cells in 518 total IgG1+ plasma cells between cohorts were assessed using Kruskal-Wallis followed by Dunn's 519 test, but did not yield significant differences. For longitudinal changes, only significant differences 520 compared to baseline are shown. Significant longitudinal differences within a cohort as indicated 521 with *, p<0.05; **, p≤0.01; ***, p≤0.001. Significant differences between cohorts at the same time point 522 are indicated with #, p<0.05; ##, p≤0.01; ###, p≤0.001. D= days after vaccination; aP = acellular per-523 tussis vaccine; wP = whole cell pertussis vaccine.