Participants were symptomatic, laboratory-confirmed pertussis (ex-)patients and healthy controls who declared to have no clinical history of infectious diseases for at least one year. These participants were part of a Dutch controlled longitudinal observational study started in 2015 (Immfact). This study was approved by the accredited review board METC UMC Utrecht (NL46795.094.13). All participants or parents/guardians of minor participants provided written informed consent. This study was conducted according to the principles described in the Declaration of Helsinki. For the collection of fresh human neutrophils and subsequent analyses, blood sample were obtained from a mini-donor system and all blood donors provided written informed consent. Blood samples were processed anonymously and the research goal, primary cell isolation, required no review by an accredited Medical Research Ethics Committee, as determined by the Dutch Central Committee on Research involving human subjects.

This study involves twenty pertussis (ex-)patients (age at inclusion 9.2 – 78.1 years, median 48.2 years, male/female ratio: 0.15/0.85) and 21 age-matched controls (age at inclusion 9.2 – 77.3 years, median 46.7 years, male/female ratio: 0.43/0.57) representing a wide spectrum of age groups. All participants were vaccinated according to the Dutch national immunization program, except for those born prior to implementation of the pertussis vaccination (1957) in the Netherlands. Coagulated and anti-coagulated peripheral blood was collected for serum and plasma isolation, respectively, and samples were stored at -20°C until use. The pertussis (ex-)patients were sampled in a longitudinal fashion at <3, 9 ± 1, 18 ± 2 and 36 ± 3 months after diagnosis of pertussis and controls were sampled once. For this study a total of 99 plasma samples were used which were tested for IgG and IgA antibody levels against nine B. pertussis antigenic specificities (Pieren et al, manuscript in preparation), using a previously described multiplex immune assay (van Twillert et al., 2017), extended to include in addition to pertussis toxin (Ptx), filamentous hemagglutinin (FHA), pertactin (Prn), outer membrane vesicles (OMV), and lipooligosaccharide (LOS), also fimbriae 2 (Fim2), fimbriae 3 (Fim3), virulence associated gene 8 (Vag8) and BrkA (Supplementary Table 1). For two pertussis (ex-)patients no samples were available at <3 months after diagnosis (Figure 1). For one pertussis (ex-)patient, serological data for IgA specific for Ptx, OMV and LOS at 18 months after diagnosis were missing as well as all IgA data from one healthy control participant. Healthy controls for this study were prescreened for the absence of a plasma IgG-Ptx serological response (< 20 IU/mL) (de Greeff et al., 2010; van Twillert et al., 2017).

In this study the naturally circulating B1917 B. pertussis strain, isolated from a Dutch pertussis patient in 2000, was used. To ensure consistency between experiments, flash freeze vials (FFVs) were prepared as previously described (Kroes et al., 2019). Before using the FFVs, they were spun down for 10 min at 16,000 x g and for the ROS assay resuspended in Hanks’ balanced salt solution (HBSS; Sigma-Aldrich) + 0.1% human serum albumin (HSA; Sanquin) and 50μM luminol (Sigma-Aldrich). For the killing assay the bacteria were resuspended in Iscove Modified Dulbecco Media (IMDM; Gibco) with 10% heat inactivated (HI) fetal calf serum (FCS; Hyclone).

Isolation of primary neutrophils was performed using fresh blood from multiple healthy donors. Neutrophils were isolated by using a Ficoll-Histopaque gradient method. In short, heparinized blood was diluted with an equal volume of PBS and layered onto a gradient of Ficoll-Paque (GE Healthcare Life Sciences) and Histopaque-1119 (Sigma-Aldrich). After centrifugation at 396g, for 20 minutes, neutrophils were collected from the Histopaque layer. After washing with HBSS containing 0.1% HSA, the neutrophils were subjected to hyperosmotic shock with ice-cold water for 30 seconds to lyse erythrocytes. Isolated neutrophils were washed again with HBSS + 0.1% HSA. Viability of neutrophils was verified after isolation by using the live-dead staining BD Horizon Fixable Viability Stain 780 (BD Biosciences) and was shown to be >95% for all experiments. Samples were acquired on the FACS Symphony A3 (BD Biosciences) flow cytometer and data analysis was done using FlowJo software (BD Biosciences).

The serum samples used for opsonization of B. pertussis were collected at <3, 9 ± 1, 18 ± 2 and 36 ± 3 months after pertussis diagnosis. Sera from un-infected age matched controls were also included. All serum samples were heat inactivated (HI) at 56°C for 30 min. For antibody-mediated opsonization, bacteria were incubated with 10% HI serum for 30 min at 37°C under gentle agitation. For normalization purposes, a HI reference serum sample from a positive pertussis patient was included. This reference serum was kindly provided by the Centre for Diagnostics and Laboratory Surveillance (IDS) at the National Institute for Public Health and the Environment.

Neutrophils were suspended in HBSS + 0.1% HSA in the presence of 50 μM luminol. A total of 200,000 neutrophils were seeded into a white 96-well microplate (Sigma-Aldrich). Cells were incubated with non-opsonized B. pertussis; HI Ctrl- or BP-serum only; or opsonized B. pertussis at multiplicity of infection (MOI) 10 in HBSS + 0.1% HSA. To inhibit ROS production, neutrophils were incubated with 5 µM VAS3947 NADPH oxidase (NOX) inhibitor VIII (Millipore) for 30 min prior to incubation with live B. pertussis. The concentration of the ROS inhibitor used was selected from previous titration experiments showing optimal ROS inhibition and neutrophil viability (data not shown). As a positive control for ROS production by neutrophils, cells were incubated with 5 ng/ml PMA (Supplementary Figure 1). For normalizing purposes, a reference serum was included in every plate when measuring ROS production to correct for experiment, plate and neutrophil donor variability. The production of ROS was determined in real-time by measuring chemiluminescence with a TriStar² LB 942 multimode microplate reader (Berthold Technologies). Light emission was recorded in relative light units (RLU’s) for 30 minutes at 37°C. For representation purposes area under the curve (AUC) values were determined using the area under curve function in GraphPad prism 9.1.0 using standard parameters.

Opsonized or non-opsonized B. pertussis at MOI 10 were incubated in the presence or absence of 200,000 neutrophils in IMDM with 10% HI FCS (Hyclone) for 4 hours at 37°C. When indicated, neutrophils were incubated with 5 nM VAS2870 NOX inhibitor III (Millipore) for 30 min prior to incubation with live B. pertussis to inhibit ROS production. After 4 hour incubation the samples were treated with sterile water at pH 11 for 5 minutes to lyse neutrophils and release surviving bacteria (Decleva et al., 2006). Lysates were further diluted and each dilution was plated on Bordet Gengou (BG) agar plates containing 15% sheep blood (BD Bioscience) and incubated at 35°C and 5% CO₂ for 5 days after which bacterial colonies were counted using a ProtoCOL 3 colony counting system (Symbiosis). Only bacterial counts between 15 and 305 colonies were included for further analysis. To correct for the different neutrophil donors used in the various experiments, the data shown (% CFU) is relative to the bacterial counts when using neutrophils incubated with non-opsonized B. pertussis.

Statistical analysis was performed as previously described (Kroes et al., 2021). Briefly, the permutation based exact Wilcoxon-Mann-Whitney test was used (wilcox_test function from rstatix R package). p-values were corrected for multiple testing by the Benjamini-Hochberg method.

Effect sizes (ES), indicating how big the differences between two groups are, were calculated as Cliff’s delta. The established thresholds are: small, <0.28; medium, <0.43; large, <0.7 and very large >= 7 (Vargha and Delaney, 2000). Comparisons with both padj <0.05 and ES either medium or higher, as well as padj <0.1 and ES either large or very large are considered significantly different.

When measuring ROS production using the control sera, one serum sample induced 8 times more ROS than the average of the group. The presence of outliers was explored for this group by Grubbs’ test (GraphPad Prism (version 9.1.0)), which qualified that unique sample as outlier, and was consequently eliminated from the analysis.

For correlation analysis the rcorr function from the Hmisc R package was used to calculate the spearman correlations between ROS production and antibody levels. Spearman’s Rho rank (r_s) and approximate p-values were provided. The corrplot function from corrplot R package was used to represent the above mentioned correlations. p-values were corrected for multiple testing by the Benjamini-Hochberg method. A correlation was considered significant at an adjusted p-value <0.05. If not significant, the corresponding cell in the plot appeared empty.

To investigate ROS production by neutrophils in response to B. pertussis we isolated neutrophils from the blood of healthy individuals and incubated these with live B. pertussis in the presence of luminol, allowing the measurement of ROS production. In the absence of antibodies, we found no evidence that neutrophils produce additional ROS upon encountering B. pertussis (Figure 2A). Previous studies in mice highlighted a critical role for B. pertussis-specific antibodies in the control of B. pertussis infection by neutrophils (Andreasen and Carbonetti, 2009). Therefore, we set out to elucidate the role of B. pertussis-specific antibodies in the production of ROS by human neutrophils. To this end, we opsonized live B. pertussis with HI serum from laboratory-confirmed pertussis patients longitudinally collected <3 and 9, 18 and 36 months after the diagnosis of symptomatic pertussis (BP-serum) or HI serum of healthy age-matched control individuals (Ctrl-serum; Figure 1). Bordetella pertussis opsonized with BP-serum collected at any timepoint after diagnosis significantly induced ROS production by human neutrophils (Figure 2A) which did not wane up to 3 years after the acute stage of the disease. While B. pertussis opsonized with Ctrl-serum also induced ROS production, this ROS production was significantly lower compared to B. pertussis opsonized with BP-serum (Figure 2A). Incubation of neutrophils with serum alone did not induce any ROS production above background levels (Figure 2A). To have an indication of which antibodies are most responsible for the increased ROS production against the BP-serum opsonized bacteria, we made use of a dataset of IgG and IgA antibody levels targeting Ptx, FHA, Prn, Fim2, Fim3, Bp-OMV, LOS, Vag8 and BrkA in corresponding plasma samples (Supplementary Table 1). Correlation analysis was performed between the individual antigen-specific antibody levels and the neutrophil ROS production induced by BP-serum only or by B. pertussis opsonized with the BP-serum. Findings indicate a positive correlation between ROS production by neutrophils and IgG antibodies targeting Fim3 followed by Prn and BrkA (Figure 2B). These correlations were strong and highly significant (r_s >0.5, p-value <1.05E-06). Weaker correlations were found for all other antigen-specific antibody levels, except those targeting Ptx for which no correlation at all was found. For the antigen-specific IgA antibodies, no positive correlations were found with the neutrophil ROS production (data not shown).

We then investigated whether primary human neutrophils were able to kill B. pertussis and the role for ROS in this process. Figure 3 shows the % CFU for live B. pertussis opsonized with either BP or Ctrl sera in the presence or absence of neutrophils. The % CFU was calculated relative to the CFU values obtained when using non-opsonized bacteria incubated with neutrophils from the corresponding donor (dotted line). Results indicate that in the presence of neutrophils a significant enhanced killing of B. pertussis (less % CFU) was observed when the bacteria were opsonized with BP-sera (green) (p = 0.075, large effect size (ES) = 0.605) compared to the opsonized B. pertussis in the absence of neutrophils. No significant difference was observed when the bacteria were opsonized with Ctrl-sera (purple) and incubated with neutrophils compared to the opsonized bacteria only

To determine the contribution of ROS production in bacterial killing, we incubated the neutrophils with the VAS287 NOX inhibitor III which blocks both intra- and extracellular ROS production. Validation of the inhibitory effect of the VAS287 NOX inhibitor III on ROS production by human neutrophils was performed (Supplementary Figure 1). Figure 3 shows a reduction of bacterial killing in the presence of the NOX inhibitor, as indicated by an increased trend in % CFU which value is comparable to that obtained when the opsonized bacteria were cultured in the absence of neutrophils. These findings suggest an important role for neutrophil derived ROS in killing of B. pertussis opsonized with sera from convalescent pertussis patients.