Long-term analysis of pertussis vaccine immunity uncovers a memory B cell response to whole cell pertussis immunization that is absent from acellular immunized mice

Over two decades ago acellular pertussis vaccines (aP) replaced whole cell pertussis vaccines (wP) in several countries. Since then, a resurgence in pertussis has been observed, which is hypothesized to be linked to waning immunity. To better understand why waning immunity occurs, we developed a long-term outbred CD1 mouse model to conduct the longest murine pertussis vaccine studies to date, spanning out to 532 days post primary immunization. Vaccine-induced memory results from follicular responses and germinal center formation; therefore, cell populations and cytokines involved with memory were measured alongside protection from challenge. Both aP and wP immunization elicit protection from intranasal challenge and generation of pertussis specific antibody responses in mice. Responses to wP vaccination were characterized by a significant increase in T follicular helper cells in the draining lymph nodes and CXCL13 levels in sera compared to aP mice. In addition, a population of B. pertussis+ memory B cells was found to be unique to wP vaccinated mice. This population peaked post-boost, and was measurable out to day 365 post-vaccination. Anti-B. pertussis and anti-pertussis toxoid antibody secreting cells increased one day after boost and remained high at day 532. The data suggest that follicular responses, and in particular CXCL13 levels in sera, should be monitored in pre-clinical and clinical studies for the development of the next-generation pertussis vaccines.


INTRODUCTION 46
Pertussis (whooping cough) is a vaccine-preventable respiratory disease caused by the Gram-negative bacterium Bordetella pertussis 1,2 . Whole cell pertussis vaccines (DTP/wP) 48 were first developed and implemented in 1914, but did not become widely available for 49 distribution until the 1940s 1,3 . After their implementation in the United States, DTP 50 vaccines dramatically reduced pertussis disease from ~200,000 cases a year in the 1930s 51 to ~2,000 in the 1970s 1,4,5 . However, safety concerns arose that led to the development 52 of acellular pertussis vaccines (aP) in the United States and Europe in the late 1990's 6-8 . 53 Unlike wPs, which are composed of inactivated B. pertussis, aPs contain 1-5 pertussis 54 antigens adsorbed to aluminum hydroxide 1,9 . While aP vaccines protect against 55 symptomatic disease, they do not prevent transmission or asymptomatic carriage of 56 pertussis 10 . Pertussis outbreaks have been increasing at an alarming rate despite high 57 vaccine coverage since wP vaccines were replaced with aP vaccines [11][12][13] . During the 58 2012 pertussis outbreak in the US, there were ~50,000 cases and 20 deaths attributed to 59 this disease, the worst incidence of pertussis observed since the 1950s 14 . Fluctuation of 60 pertussis incidence is hypothesized to be in part due to waning in immunity elicited by the 61 current aP vaccines, with as low as 10% efficacy in between boosters during 62 adolescence [15][16][17] . The pertussis field has several hypotheses for the resurgence of 63 pertussis including (in no particular order): 1) more sensitive PCR-based testing, 2) 64 increased reporting, 3) bacterial evolution, and 4) lack of duration of the protective 65 immune response 6,18-21 .

RESULTS 295
Use of a long-term vaccine memory model using outbred CD1 mice 296 Immune memory against pertussis varies greatly depending on the vaccine used. It is 297 estimated that the duration of protection conferred by wP vaccines lasts 4-12 years 52 . For 298 aP vaccines, immunity wanes much more quickly, as observed and underscored by the 299 outbreaks in California and other locations 53 . In our study, we set out to establish a long-300 term pertussis vaccine efficacy model to evaluate the duration of immunity and identify 301 additional factors that contribute to either wP efficacy or the inadequate responses of aPs. 302 To model the human outbred population, we selected outbred CD1 mice that are also 303 known to have a long lifespan. We aimed to compare the DTaP vaccine (Infanrix; GSK) 304 to a prototype whole cell pertussis vaccine. DTP vaccines are no longer available in the 305 US; therefore, we used the NIBSC whole cell pertussis antigen vaccine for comparison. 306 Unlike most combination DTP vaccines, the NIBSC whole cell pertussis vaccine is not 307 adjuvanted with alum nor combined with diphtheria and tetanus toxins. Female CD1 308 mice were mock-immunized (PBS), immunized with 1/10 th human dose DTaP, or 1/10 th 309 human dose of NIBSC wP vaccine at six weeks of age by intramuscular administration 310 and subsequently boosted three weeks later. We performed our analysis on mice at day To design the next generation of pertussis vaccines, it is important to understand the 319 underlying immunological cause of the relatively short-term immunity provided by aP 320 vaccines. aP vaccines were originally developed and tested in mouse models that 321 provided limited information regarding the longevity of protection. Furthermore, it remains 322 to be determined if this model can mimic the waning immunity observed in humans. We 323 hypothesized that mice immunized with 1/10 th the human dose of wP and aP would be 324 protected from challenge early on, but that protection would decrease over time in aP 325 immunized mice. Mice were intranasally challenged with 2x10 7 CFU/dose of B. pertussis 326 at multiple time points between day 35 and day 532 post-vaccination. Mice were then 327 euthanized three days post-infection to study vaccine-mediated protection. The bacterial 328 burdens in the lung (Fig 1B), trachea (Fig 1C), and nasal wash (Fig 1D)  observed that bacterial burden in the lung (Fig 1B), trachea (Fig 1C), and nasal wash 336 (Fig 1D) was significantly decreased in immunized mice, regardless of the vaccine 337 administered, compared to the vehicle control group at days 35, 60, and 90. Protection 338 remained significant at day 532 in the trachea of wP immunized animals compared to 339 16 mock-vaccinated animals (Fig 1D). At this dose, both aP and wP vaccines protected mice 340 from intranasal challenge. The data illustrate unique susceptibility of CD1 mice to B. 341 pertussis over time and the importance of longitudinal studies to identify the optimal 342 timeframe to study vaccine efficacy. 343 Pertussis specific antibody responses to aP and wP immunization persist as far as day 344 532 post-prime 345 To gain insights into the differences between aP and wP immunological responses, the 346 model described above was applied to study pre-and post-boost responses, and their 347 evolution over time. The subsequent studies were performed in non-challenged animals 348 to clearly separate response to vaccination from response to challenge. 349 Antibodies play a major role in vaccine-mediated protection against numerous pathogens 350 and are a correlate of protection used to evaluate or predict vaccine efficacy 6,54 . DTP/wP 351 and DTaP vaccines induce opsonizing antibodies that contribute to bacterial clearance 352 and induce anti-pertussis toxin (PT) antibodies that neutralize toxin function 55,56 . 353 However, the value of some of these antibodies in protection against pertussis disease 354 or B. pertussis infection is highly debated 57 . Here, we hypothesized that antibody 355 responses to B. pertussis antigens in aP immunized mice would decrease over time 356 compared to wP immunized mice. We first measured anti-B. pertussis and anti-FHA IgG 357 antibodies in the serum of immunized mice over time to determine the levels of surface-358 binding antibodies (Fig 2A, 2B). Both aP and wP immunization elicited significant 359 production of anti-B. pertussis and anti-FHA IgG antibodies compared to vehicle-control 360 immunized mice (Table S1). Anti-B. pertussis antibody levels were not statistically 361 different between aP and wP vaccinated groups, and they peaked after boost and 362 remained elevated out to day 532 post-prime (Fig 2A, Table S1). Anti-FHA antibodies 363 also increased after boost and remained elevated at day 532 post-prime (Fig 2B). 364 In addition to opsonizing antibodies, it is imperative that pertussis immunization stimulates 365 the production of PT-neutralizing antibodies that can prevent symptoms, disease 366 manifestation, and in the case of infants, death 58 . Anti-PT antibodies have been proposed 367 as a correlate of protection against pertussis; however, this is highly debated 57 . PT is an 368 AB5 exotoxin that plays a key role in the pathogenesis of pertussis by triggering ADP-369 ribosylation which inhibits G protein-coupled signaling 59-64 . PT activity leads to a decrease 370 in airway macrophages, induction of leukocytosis, and modulation of adaptive immune 371 responses to B. pertussis. Unlike opsonizing antibodies, significant anti-PT IgG antibody 372 responses were only observed in aP vaccinated animals compared to both vehicle control 373 and wP immunized mice (Fig 2C, Table S1). Anti-PT antibodies peaked at day 60 post 374 vaccination and remained elevated until day 532. To note, no anti-PT IgG antibodies were 375 detected in wP immunized animals, consistent with the lack of PT in the NIBSC wP 376 formulation due to manufacturing practices (Fig 2C). 377 DTaP is a combination vaccine also containing diphtheria toxoid and tetanus toxoid. 378 Unlike the immunity to pertussis conferred by DTaP and Tdap immunization, immunity 379 against diphtheria and tetanus does not wane overtime 65 . We observed that aP 380 immunization elicits significant anti-diphtheria toxoid (Fig S1A) and anti-tetanus toxoid 381  (Table S2). In the right inguinal lymph node (Fig 3A) immunization with wP 402 induced significant TFH responses compared to aP and PBS immunized mice 20 days 403 post-vaccination. Although no significant differences were observed between aP and wP 404 immunized animals at other time points, TFH cells in the draining lymph node at day 35 405 and in the spleen at days 20 and 22 were more numerous in the wP group (Fig 3B). 406 Consistent with previous studies, a greater number of TFH cells was only detected at the 407 earliest time point studied (Fig 3A). 408 To facilitate their function in the GCs, TFH cells express the chemokine CXCL13 which is 409 a signaling molecule that plays a crucial role in B cell recruitment and GC organization 410 through binding to CXCR5 69 . While CXCL13 can be found locally in GCs it can also be 411 measured in the serum as a biomarker of GC activity 45 . Therefore, we measured CXCL13 412 levels in the serum of immunized mice during the course of this experiment (Fig 3C). 413 Levels of CXCL13 in the serum of vehicle control and aP-immunized mice were low (Fig  414   3C). In contrast, we observed that only wP immunization elicits significant production of 415 To study MBCs, we incubated splenocytes with fluorescently labeled B. pertussis. We 429 then separated MBCs from the rest of the splenocytes using a proprietary kit (Miltenyi), 430 followed by labeling with CD3e, CD45R, IgG, CD38, and CD80 75,7 . We analyzed MBC 431 populations based on labeling with B. pertussis (antigen-specific) and further defined the 432 B cell populations based on CD38 and CD80 surface expression 75-77 (Fig S2). We observed significant expansion of B. pertussis + MBCs in wP immunized mice 439 compared to both mock and aP immunized mice (Fig 4A, Table S2, Fig S2). In the wP 440 group, this population peaked at day 35 post-boost, and although not statistically different 441 from the mock vaccinated control group, B. pertussis + MBCs were measurable at days 442 152 and 365 post-prime in wP vaccinated mice (Fig 4A, Fig 4B). We observed that in 443 vaccinated animals, B. pertussis + MBCs tend to be double positive for CD38 + CD80 + , while 444 B. pertussis -MBCs are mainly CD38 + (Fig 4C, D, E, F). This is likely important as IgG + 445 CD80 + MBCs have been shown to differentiate into antibody secreting cells with the 446 capacity to seed GCs 82,83 . Interestingly, single-labeled CD80 + MBCs were low in both B. 447 pertussis + and B. pertussispopulations. Overall, the data show that significant production 448 of B. pertussis + MBCs is most prevalent in wP immunized mice, and that B. pertussis + 449 MBCs are characterized by a unique combination of cell surface marker profiles. 450

Immunization against B. pertussis elicits production of antibody secreting cells in mice 451
In the GC, naïve B cells that receive TFH cell help and undergo affinity maturation can 452 differentiate into plasma cells. These cells migrate to the bone marrow and function to 453 secrete antibodies and protect from infection. Long-lived plasma cells are terminally 454 differentiated cells that can survive and continue to secrete antibodies for years, which 455 has been demonstrated in both humans and mice 84 . We hypothesized that wP 456 immunization would induce greater production of pertussis-specific long-lived plasma 457 cells compared to aP immunization, mimicking the waning immunity observed in the 458 human population. Therefore, we isolated bone marrow cells and quantified the number 459 of total long-lived plasma cells and the number of antigen-specific antibody secreting 460

cells. 461
Using flow cytometry, we first observed that there was no difference in the number of total 462 long-lived plasma cells (CD19 -CD45R + CD138 + ) 85 in the bone marrow in any of the groups 463 at any of the time points studied (data not shown). To determine the number of antigen-464 specific antibody secreting cells, we performed B cell ELISPOT on bone marrow samples 465 (Fig 5A-F). We only observed a significant increase in the number of anti-B. pertussis 466 antibody secreting cells in wP immunized mice one day post-boost compared to both 467 mock and aP vaccinated animals (Fig 5B). Although there were no significant differences 468 in antibody secreting cells at days 20 and 532 post-prime (Fig 5A, C), wP immunized 469 animals had higher numbers of spots overall (Fig 5A-C). This contrasts with what was 470 observed in the serological studies in which anti-B. pertussis titers remained high during 471 the course of the study (Fig 2A). 472

22
In this study, we also observed that anti-pertussis toxoid antibody secreting cells were 473 present in greater numbers in aP mice compared to vehicle or wP mice, and persisted 474 out to day 532 post-vaccination (Fig 5F). Notably, the number of anti-PT antibody 475 secreting cells was significantly greater at day 532 post-prime in aP immunized animals 476 (Fig 5F). This observation is consistent with the anti-PT titers detected in the serological 477 study (Fig 2C) and with the fact that the wP vaccine used in this study contains minimal 478 pertussis toxin. The data highlight important differences between serological detection of 479

B. pertussis antibodies and quantification of antibody-producing cells in the bone marrow. 480
Altogether, the data described in this study highlight important differences in the 481 immunological signatures triggered by aP and wP vaccination in mice (Fig 6). wP 482 vaccination was characterized by stronger TFH cell responses, CXCL13 production, B. 483 pertussis + MBCs, and anti-B. pertussis antibody secreting cells (Fig 6, Fig 5), compared 484 to aP vaccination. These markers correlate with the stimulation of, or result from GC 485 reactions, suggesting that wP vaccination triggers stronger follicular responses and 486 vaccine-induced memory (Fig 6B). During the development of acellular pertussis vaccines, immunogenicity of candidate 506 vaccines was assessed in animals and humans 93 . Antibodies to PT, FHA, fimbriae, 507 pertactin, DT, and TT were measured in immunized infants, along with toxin neutralization 508 assays to determine levels of agglutinating antibodies 93 . Currently, immunoglobulin levels 509 are measured and provide an approximation of vaccine efficacy; however, these metrics 510 do not predict the duration of immunological memory and protection. Unfortunately, rarely 511 in pertussis studies is the timing and longevity of protection considered due to the obvious 512 challenges. This study addresses this gap in knowledge by determining long-term aP and 513 wP vaccine-mediated protection out to day 532 post-vaccination, which we think is the 514 longest-lasting pertussis vaccine study in mice performed to date. Laboratory shutdown 515 during the COVID-19 pandemic increased our experimental time-frame but maybe for the 516 better. In this study we used numerous approaches to characterize the follicular response 517 24 to pertussis vaccination including antibody titers to vaccine antigens, CXCL13 levels in 518 sera, TFH cells, B. pertussis specific MBCs, and B. pertussis/PT specific bone marrow 519 antibody secreting cells (Fig 6). We also identified serum levels of CXCL13 and B. 520 pertussis-specific MBCs as potential biomarkers of pertussis vaccine-induced immune 521 memory. 522 One of the challenges associated with monitoring the longevity of vaccine-mediated 523 protection is that the model used must remain susceptible to infection over time. This was 524 illustrated with the infant baboon model, which allows vaccine schedules to be studied 525 and recapitulates human disease, but in which adults are not susceptible to pertussis 526 infection (~15 months of age). Therefore, we first studied the susceptibility of mice to B. 527 pertussis over time. We assessed bacterial burden in the respiratory system of mice 528 (lungs, trachea, and nasal wash) and found that susceptibility appears to change 529 overtime, as seen in humans (Fig 1B, C, D). Neonates and unvaccinated infants are 530 highly susceptible to pertussis infection; however, susceptibility decreases as they age 531 toward adulthood. Furthermore, adults over 65 years of age are more likely to be 532 hospitalized for pertussis than younger adults 94,95 . We observed a similar pattern of 533 susceptibility in our vehicle control mice, in which mice between days 35 and 90 post-534 vaccination were susceptible to infection, but bacterial burden was below the limit of 535 detection by day 152 post-vaccination. Bacterial burden was again detectable by day 365 536 post-vaccination. 537 One potential explanation behind the fluctuation in susceptibility to B. pertussis in mice, 538 is that we suspect that like pigs, mice have differential expression over time of some 539 25 inhibitory factors such as beta-defensin 1 that may render mice no longer susceptible to 540 B. pertussis 96,97 . Alternatively, specific microbiota in the airway might out compete the 541 challenge dose 98 . These data highlight the importance of conducting these types of 542 studies over a long period of time, as intermediate lengths of studies may not allow 543 measurement of vaccine-mediated efficacy due to low susceptibility. This is also 544 important as susceptibility to B. pertussis in humans and in particular, death associated 545 with pertussis, varies over time. Combinations of neonatal models with long-term models 546 may be able to better evaluate pertussis vaccine efficacy. Additionally, inbred mice may 547 have different windows of susceptibility that should be considered and studied further. In humans, studies clearly show that antibody titers against pertussis decay over 556 time 100,101 . For example, human serum titers against PT decay as quickly as 6-12 months 557 after vaccination whereas anti-tetanus serum titers last up to 19 years 102,103 . Similar to 558 humans, BALB/c mice immunized with a low dose of wP or aP elicited high serum IgG 559 antibody titers that increased rapidly after vaccination, but were undetectable by 6-9 560 months 92,99 . However, in our model and at 1/10 th the human dose of the vaccines, we did 561 not observe antibody decay over time. In fact, antibody levels peaked after boost and 562 26 remained elevated out to day 532 post-vaccination (Fig 2, Fig S1). These observations 563 could be due, in part, to the relatively high dose of vaccine used here, and correlate with 564 the sustained protection provided by both aP and wP vaccines over time. 565 To further investigate the humoral immune response to B. pertussis vaccination, we 566 measured B. pertussis + MBCs in the spleen of immunized mice (Fig 4, Fig S2) and the 567 presence of antibody-secreting cells in the bone marrow. We observed that wP 568 immunization elicited significant B. pertussis-specific MBC responses compared to both 569 PBS and aP immunization (Fig 4A, B). Additionally, CD38 + CD80 + cells were associated 570 with B. pertussis + MBCs but not B. pertussis -MBCs (Fig 4C, D, E, and F) Detection of antigen-specific MBCs from the spleen and antibody secreting cells in the 576 bone marrow provides insight into the differences between wP and aP vaccine-induced 577 immunity. Unfortunately, the protocols described here to detect antigen-specific MBCs 578 are time consuming and technically challenging due to the rarity of these populations. 579 Therefore, the implementation of antigen specific MBC analysis for clinical evaluation of 580 pertussis vaccines in humans is unlikely. In addition, detection of antibody secreting cells 581 requires invasive procedures to obtain samples for analysis, making its implementation 582 unfavorable at the clinical level. 583

27
To identify additional markers of vaccine-induced memory, we measured CXCL13 in 584 immunized mice as it has previously been measured in sera from humans and would be 585 more feasible to monitor in clinical studies 104 . CXCL13 levels in the serum of immunized 586 (non-challenged) mice peaked one day post-boost in wP-, but not in aP-immunized mice 587 (Fig 3C), again highlighting another difference between both vaccines (Fig 6A). CXCL13 588 levels were significant both pre-boost and post-boost as far as day 60 post-vaccination 589 (Fig 3C). Additional studies are needed to monitor the production of CXCL13 early on 590 after vaccine priming. There appears to be a narrow window of CXCL13 production in 591 mice, likely consistent with GC formation in response to vaccination. CXCL13 is a reliable 592 plasma biomarker of GC activity in both humans and nonhuman primates 45 . Therefore, 593 measuring CXCL13 levels post-vaccination using a minimally invasive blood draw could 594 be utilized during clinical studies when testing candidate pertussis vaccines in humans. 595 There are obvious caveats to using CXCL13 as a biomarker for the longevity of pertussis 596 vaccine-induced memory that need to be considered when designing clinical trials. The 597 first is that CXCL13 is not antigen-specific. Another caveat is that CXCL13 expression is 598 altered in response to infection, in cancer, systemic lupus erythematosus, rheumatoid 599 arthritis, and other diseases involving germinal center, or ectopic lymphoid structure 600 formation 105,106 . This should be taken in account when establishing exclusion criteria for 601 clinical trials. 602 From this work, we propose that CXCL13, circulating B. pertussis + MBCs, and pertussis 603 specific antibody titers should be measured together in the blood of patients enrolled in 604 clinical trial vaccine trials to inform the longevity of vaccine-mediated protection (Fig 6). 605 28 Future studies need to address if follicular responses induced by aP vaccination can be 606 improved to levels similar to that induced by wP vaccination. Addition of adjuvants known 607 to stimulate GC formation could enhance the longevity of aP vaccines and prevent waning 608 immunity. This work strongly suggests that GC quantification, size, and composition 609 should be evaluated in response to vaccination to identify formulations of the next 610 generation pertussis vaccine that provide long-term protection. Single Cell Core Facility for her support. 627