Pertussis toxin suppresses dendritic cell-mediated delivery of B. pertussis into lung-draining lymph nodes

The adenylate cyclase (ACT) and the pertussis (PT) toxins of Bordetella pertussis exert potent immunomodulatory activities that synergize to suppress host defense in the course of whooping cough pathogenesis. We compared the mouse lung infection capacities of B. pertussis (Bp) mutants (Bp AC− or Bp PT–) producing enzymatically inactive toxoids and confirm that ACT action is required for maximal bacterial proliferation in the first days of infection, whereas PT action is crucial for persistence of B. pertussis in mouse lungs. Despite accelerated and near complete clearance from the lungs by day 14 of infection, the PT− bacteria accumulated within the lymphoid tissue of lung-draining mediastinal lymph nodes (mLNs). In contrast, the wild type or AC− bacteria colonized the lungs but did not enter into mLNs. Lung infection by the PT− mutant triggered an early arrival of migratory conventional dendritic cells with associated bacteria into mLNs, where the PT− bacteria entered the T cell-rich paracortex of mLNs by day 5 and proliferated in clusters within the B-cell zone (cortex) of mLNs by day 14, being eventually phagocytosed by infiltrating neutrophils. Finally, only infection by the PT− bacteria triggered an early production of anti-B. pertussis serum IgG antibodies already within 14 days of infection. These results reveal that action of the pertussis toxin blocks DC-mediated delivery of B. pertussis bacteria into mLNs and prevents bacterial colonization of mLNs, thus hampering early adaptive immune response to B. pertussis infection.

Introduction Pertussis, or whooping cough, is an acute respiratory illness caused by Bordetella pertussis (Bp) that used to be the leading cause of infant mortality in the pre-vaccine era [1][2][3]. Despite global vaccine coverage, pertussis burden remains high and it is estimated that~24 million whooping cough cases and~160,000 pertussis-related deaths occur annually world-wide [4]. Moreover, pertussis is on the rise in developed countries using the acellular pertussis (aP) vaccines that confer a short-lasting protection and fail to prevent B. pertussis transmission by vaccinated individuals [5,6]. Indeed, data from animal models show that the aP vaccine does not prevent nasopharynx infection by B. pertussis [7][8][9][10][11][12]. B. pertussis attaches to ciliated epithelia of the airways by means of several adhesins, evades the first line of host innate defense by deploying several complement resistance factors and subverts host immunity by the synergy of pertussis toxin (PT) and adenylate cyclase toxin-hemolysin (ACT, AC-Hly or CyaA) activities [3]. PT and ACT deliver their cytotoxic enzyme subunits into an array of immune cells and blunt the bactericidal functions of phagocytes and hamper induction of adaptive immune responses through hijacking of cellular signaling pathways [13][14][15][16].
In contrast to ACT, PT is only produced by B. pertussis and represents its major virulence factor, exerting both local and systemic immunomodulatory effects [16]. Systemic effects of PT action account for the characteristic manifestations of pertussis, such as hyperleukocytosis due to leukocyte proliferation, lymphocyte egress from lymphoid organs and bone marrow into circulation and inhibition of leukocyte extravasation [47]. PT action leads to formation of mixed lymphocyte aggregates in arterioles, triggering pulmonary hypertension and eventual heart failure [48,49], as well as histamine sensitization [50]. Moreover, PT action plays a key role in immune evasion of the bacterium, delaying neutrophil recruitment onto the infected mucosa and interfering with induction of adaptive immune responses [17,[51][52][53][54][55].
PT is formed by six subunits (S1:S2:S3:2S4:S5) that assemble into a typical AB 5 toxin oligomer, with an ADP-ribosyl transferase enzyme subunit A (S1) capping a pentameric B 5 subunit. The B 5 subunit mediates toxin binding to surface sialoglycoproteins and itself triggers mitogenic signaling and DC maturation [56][57][58][59]. PT enters cells by receptor-mediated endocytosis and upon retrograde transport reaches the endoplasmic reticulum, where the enzymatic A subunit dissociates and translocates into the cell cytosol [60]. Using cytosolic NAD as a substrate, the A subunit ADP-ribosylates and inactivates the inhibitory G i/o α subunits of heterotrimeric G proteins, thereby disrupting the inhibitory GPCR signaling and relieving the inhibition of endogenous adenylyl cyclase enzymes by G i/o α [60][61][62]. The resulting increase of cellular cAMP levels dysregulates protein kinase A-activated signaling pathways depending on various GPCRs and the activities of Ca 2+ and K + channels in various cell types [63,64]. Since most chemokine receptors are coupled to G i/o proteins [65], one action of PT is to deregulate the chemotaxis of macrophages, lymphocytes and neutrophils [51,52,[66][67][68].
Previously, PT activity was found to reduce CCR7-mediated ex vivo migration of human monocyte-derived dendritic cells (MDDCs) towards the lymphoid chemokine CCL21 [69]. Moreover, an enhanced migration of CD11c + cells from mouse lungs into the lung-draining mediastinal lymph nodes (mLNs) was observed upon lung infection with a B. pertussis mutant lacking genes for PT production [67]. On the other hand, we have previously observed that the cAMP-elevating activity of ACT enhanced migration of TLR-activated mouse and human DC towards lymph node-homing chemokines CCL19 and CCL21 [46]. Therefore, we tested here the hypothesis that PT activity delays ACT-triggered migration of B. pertussis antigen-loaded DCs from infected lungs into mLNs and show that early DC-mediated delivery of PT-deficient bacteria enables B. pertussis proliferation in mLNs.

Pertussis toxin prevents prolonged B. pertussis infection of lung-draining lymph nodes
To dissect the impact of ACT and PT activities on outmigration of DCs from B. pertussisinfected lungs into mLNs, we used Tohama I-derived mutants (Table 1) producing enzymatically inactive toxoids of ACT or PT (Bp AC − or Bp PT -) [44]. The used bacteria further produced the fluorescent mScarlet protein from a plasmid [70]. This did not affect growth or secretion of the AC − and PT − toxoids by the bacteria in vitro and enabled B. pertussis imaging in the tissues of infected mice for up to 14 days without the need for antibiotic selection (S1 Fig and below). In line with intact fitness in vitro, the Bp WT bacteria (AC + PT + ) inoculated intranasally (8 x 10 5 CFU in 50 μL) proliferated in the lungs of adult BALB/c mice by~1.5 order of magnitude within 5 days (to~2.7 x 10 7 CFU) and persisted after a progressive decline at~10 5 CFU/lung on day 14 ( Fig 1A). Under the same conditions, the Bp AC − bacteria producing the ACT toxoid and active PT (AC -PT + ) proliferated less and reached an order of magnitude lower CFU level in the lungs, while reproducibly persisting at a comparably high (~10 5 CFU/lung) level as Bp WT by day 14. In contrast, the Bp PT − bacteria, secreting the PT toxoid and producing active ACT (AC + PT -), proliferated as rapidly as Bp WT bacteria for the first 3 days of infection, but their CFU counts in the lungs rapidly decreased thereafter to onlỹ 10 3 CFU/lung on day 14 ( Fig 1A). These results corroborate the previous conclusion, reached under different experimental conditions, that ACT plays an important role in the early phases of B. pertussis infection, whereas PT activity is important for sustained lung infection [17,19,20,55,67,71]. In fact, we have repeatedly found that the PT-producing Bp AC − bacteria tended to persist in mouse lungs at detectable levels weeks after the wild-type or PT − bacteria have already been cleared (S2 Fig and [71]). Intriguingly, at a lower level of lung damage (S3 Fig) and at a comparable bacterial load in the lungs (cf. Fig 1A), the animals infected by the Bp PT − bacteria had strikingly larger lungdraining lymph nodes on day 5 than the animals infected by the Bp AC − strain ( Fig 1B). The size of mLNs of Bp PT --infected mice was comparable to that of animals infected by the Bp WT strain that by day 5 of infection proliferated to an order of magnitude higher CFU level in mouse lungs (cf. Fig 1A). Moreover, the difference of mLN size between animals infected by the Bp PT − and Bp AC − strains persisted through to day 14 ( Fig 1B), even though the CFU counts of the Bp AC − bacteria in the lungs remained two orders of magnitude higher than those of the Bp PT − bacteria (cf. Fig 1A). Unexpectedly, an opposite trend was reproducibly observed when viable bacteria present in the mLNs were enumerated by plating of mLN homogenates (Figs 1C and S4). The ACT-producing Bp PT − bacteria (AC + PT -) reached the mLNs at higher numbers than Bp WT (AC + PT + ) already on day 3 of infection and persisted at numbers increased to~4 x 10 3 CFUs in the mLNs on day 14 of infection, thus exceeding the Bp PT − CFU counts in the lungs (cf. Fig 1A). In contrast, the number of viable Bp WT bacteria in the mLNs mirrored their CFU counts in the lungs, peaking at~10 3 CFUs on day 7 and decreasing to~10 2 CFUs in mLNs by day 14. Moreover, the Bp AC − (AC -PT + ) bacteria were detected in the mLNs only at an order of magnitude lower level of~10 2 CFU over the 14 days of infection, despite persisting in the lungs (cf. Fig 1A). Finally, the Bp AC -PT − bacteria producing the combination of the two toxoids were unable to proliferate in the lungs and were found in mLNs at very low numbers (~10 CFU) (Fig 1A and 1C). Hence, in the absence of PT activity, the production of active ACT favored early arrival of the Bp PT − mutant into mLNs, where the bacteria persisted at increased numbers by day 14, but did not disseminate further into spleen or other lymphatic tissue (Fig 1D).

PT-deficient B. pertussis invades the lymphoid tissue and proliferates in mLNs
Following proliferation in the lungs for the first 3 days, the Bp PT − mutant was rapidly cleared (cf. Fig 1A) and caused milder inflammatory damage of lung tissue (S3 Fig), thus being unlikely to be drained from the parenchyma into the mLNs in higher numbers than Bp WT bacteria. Indeed, higher numbers of Bp WT than Bp PT − bacteria were visualized within lung parenchyma on day 5 of infection by immunohistochemistry and immunofluorescence microscopy (Fig 2A and 2B). Further, no difference to uninfected control airways was identified when observing penetration of intranasally applied NHS-biotin [72] into lung parenchyma (Fig 2B), or the lamina propria of tracheas (Fig 3A and 3B) in mice infected by the Bp WT or the Bp PT − and Bp AC − strains. We thus performed confocal fluorescence microscopy of mLN cryosections to assess whether the mScarlet-producing B. pertussis bacteria arrived into the lymphoid tissue of mLNs of infected animals (Figs 4 and S5-S10). Compared to the tiny mLNs of uninfected animals, mLN hypertrophy was observed on day 5 in animals infected by the Bp WT and Bp PT − bacteria producing active ACT (Fig 4A, top), with striking mLN enlargement observed by day 14 (Fig 4A, bottom). The mScarlet-producing coccobacilli were observed on the cryosections as very bright fluorescent objects, likely corresponding to intact and viable bacteria present in the mLN tissue (Figs 4, S5 and S10). Bacterial cells were detected on day 5 sections of mLNs of all infected animals, but only the PT − bacteria were abundantly present inside of the lymphoid tissue in all collected mLN samples (Figs 4A top, 4B, S7 and S10). In contrast, the PT-producing Bp WT and Bp AC − bacteria were only detected outside of the mLNs on day 5, remaining attached to the mLN capsule, or being trapped within the surrounding non-lymphoid tissue, such as lymphatic vessels (Figs 4A, top, 4B, S6, S10). Hence, the PT-producing bacteria were drained from inflamed lungs by the lymphatics but were not delivered into the lymphoid tissue of the mLNs. Moreover, staining of the lymph node compartments for the CD3 (T-cell) and B220 (B-cell) markers revealed that the Bp PT − bacteria localized mainly within the T-cell zone (paracortex) on day 5 (Figs 4A, top, 4B, S7 and S10), whereas on day 14, the Bp PT − bacteria were mostly found in close contact with B-cells in the cortex (Figs 4A, bottom, 4B, S9 and S10). Furthermore, the Bp PT − bacteria were not only found in deeper regions of the mLNs, but were also increasingly found in clusters by day 14 of infection, suggesting proliferation of the Bp PT − bacteria within mLNs (Figs 4A, S9, S10, 5A, 5B and S11A). While single Bp PT − cells were mostly observed in mLNs on days 1 and 3 after infection, groups of 2-3 bacterial cells, or even clusters of several Bp PT − bacteria were found in mLNs on days 5 and 7. Abundant clusters of fluorescent Bp PT − coccobacilli were then observed on all examined mLN sections from day 14 (Fig 5A and 5B), in line with the increase of CFU counts detected in mLN homogenates between days 7 and 14 (cf. Fig  1C). Single bacteria were still observed on day 14, but many cells with clusters of bacteria in the vicinity of their nuclei, as well as groups of extracellular bacteria scattered in the lymphoid tissue, were also observed (S11A Fig). To test if the clusters of Bp PT − bacteria arose by replication of individual founder bacteria, or through sequential phagocytic events, mice were infected with a 1:1 mixture of pertussis colonization of lungs after intranasal administration of the wild-type (Bp WT) and mutant B. pertussis strains producing the catalytically-inactive pertussis toxin (PT -), the catalytically-inactive adenylate cyclase toxin (AC -) or a combination of both toxoids (AC -PT -). Four-week-old BALB/c mice were intranasally challenged with 8 × 10 5 CFU (in 50 μl) of the bacteria expressing the mScarlet fluorescent protein. The total number of bacteria in the lungs at the indicated time points was determined by CFU counting upon plating of lung homogenates on BG blood agar plates. Data represent the mean values with standard deviation obtained from groups of at least three mice per time point in three independent experiments. Two-way ANOVA followed by Dunnett's multiple comparisons test was used to analyze statistical significance between groups. � (p < 0.05), �� (p < 0.01), ��� p (< 0.001), ���� (p < 0.0001). (B) Examination of mediastinal lymph nodes (mLNs) on day 5 and 14 post infection. The location of the mLNs, thymus, and heart is indicated by arrows. (C) B. pertussis colonization of mLNs. The mLNs of a mouse sacrificed at the indicated time point were pooled and the homogenate was plated on BG blood agar for subsequent CFU counting. Data represent the means with standard deviations obtained from at least three mice per group and time point in three independent experiments. Two-way ANOVA followed by Dunnett's multiple comparisons test was used to analyze statistical significance between groups. � (p < 0.05), �� (p < 0.01), ��� p (< 0.001), ���� (p < 0.0001). (D) B. pertussis colonization of lungs and lymphoid organs on day 14 after intranasal administration of the B. pertussis PT − strain. The number of bacteria in the lungs and lymphoid organs on day 14 postinfection was determined by plating of organ homogenates on BG blood agar. LNs from each mouse were pooled. https://doi.org/10.1371/journal.ppat.1010577.g001

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response Bp PT − bacteria expressing mScarlet (magenta) or GFP (green) fluorescent proteins (S1A, S5C, S5D and S11B Figs). As shown in Figs 5C, 5D and S11B, mixed groups of green and magenta fluorescent bacteria were observed 5 days after infection, likely resulting from sequential phagocytosis. However, no mixed color bacterial clusters were observed by day 14 anymore and unicolor fluorescent clusters were observed instead on a number of lymph node sections from 3 independent experiments, thus indicating clonal proliferation of the bacteria inside mLNs.

PT-deficient B. pertussis is delivered from lungs into mLNs by migratory dendritic cells
Since draining of wild-type B. pertussis from inflamed lung parenchyma by the lymphatics did not cause their accumulation in the lymphoid tissue of mLNs, we reasoned that Bp PT − bacteria might be specifically delivered into mLNs by some migratory phagocytes known to traffic antigens for presentation in lymph nodes. Therefore, we analyzed the leukocyte composition of mLNs by multicolor flow cytometry. In line with the observed hypertrophy (cf. Figs 1B and 4A), the mLNs of mice infected by Bp WT and Bp PT − bacteria contained significantly higher total numbers of cells than mLNs of Bp AC --infected or uninfected animals (cf. Fig 6, top, see S12 for gating strategy). A generalized increase in numbers of all types of immune cells (e.g. Band T-lymphocytes, neutrophils, eosinophils, Ly6c high monocyte-derived DCs (moDCs) and CD11b + monocytes and macrophages) was detected in the mLNs on days 5 and 14 of Bp WT and Bp PT − infection, as compared to infection by the Bp AC − bacteria producing active PT but inactive ACT (Fig 6). Intriguingly, lung infection by the Bp PT − bacteria triggered a significant early increase of MHC-II high CD11c int migratory conventional dendritic cells (cDC) in mLNs by day 5 of infection ( Fig 7A and 7B). In turn, Bp WT bacteria, proliferating to much higher counts in the lungs, elicited a delayed increase of cDC numbers in mLNs by day 14 of infection. Furthermore,~65% of all cDCs found in the mLNs on day 5 of Bp PT − infection were migratory cDCs ( Fig 7A), with roughly equal numbers of CD11b -cDC1 and CD11b + cDC2 migratory cells (Fig 7B). These results revealed that PT activity blocked early outmigration of cDCs from B. pertussis-infected lungs into mLNs and suggested that Bp PT − bacteria may be delivered into mLNs by migratory cDCs capable of transporting various microorganisms [73][74][75][76][77][78][79][80][81][82].
To better identify the cells harboring the mScarlet-expressing bacteria, flow cytometry of mLN suspensions was used. This was technically challenging, as only~15% of the low numbers of viable bacteria detected in the mLNs (cf. Fig 1C) were associated with pelleted cells upon low speed (300 x g) centrifugation of mLN homogenates and the rest of the bacteria appeared to be extracellular (S13 Fig). Hence, a very low proportion of mLN cells was expected to contain fluorescent bacteria. Therefore, autofluorescence and unspecific antibody staining had to be rigorously controlled by parallel processing of mLN cells from animals infected by corresponding non-fluorescent B. pertussis strains. Such comparative cytometric analysis revealed that mScarlet + events were about ten-times more frequent than unspecific the indicated B. pertussis mScarlet-producing strains. Lungs were fixed with 4% PFA, embedded in paraffin and examined upon immunohistochemical staining of 2 μm sections with polyclonal rabbit anti-B. pertussis serum. Details of bronchial epithelium (I.) and lung parenchyma (II.) are indicated. Data represent representative images of groups of three mice analyzed in two independent experiments. Scale bar 100 μm. (B) Epithelial lining of infected mouse lungs on day 5 (left panel) and day 14 (right panel). At indicated time points, the epithelial lining of the airways of mice was labeled in vivo by intranasal administration of 50 μl NHS biotin (1 mg/ml) for 5 minutes before the animals were euthanized. Lungs were fixed with 4% PFA, snap frozen and 10 μm longitudinal cryosections of the left lobes were labeled with Alexa Fluor 488 streptavidin conjugate. Images were acquired using a Leica TCS SPE confocal microscope. Nuclei were visualized by DAPI staining. Details of bronchial epithelium (I.) and lung parenchyma (II.) are indicated. Biotin, nuclei and bacteria are rendered in green, blue, and magenta colors, respectively. The images are representative of one experiment performed in groups of three mice. Scale bar 100 μm. https://doi.org/10.1371/journal.ppat.1010577.g002

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response

PT-deficient B. pertussis infection triggers a very early serum antibody response
The above results indicated that PT action prevents the delivery of B. pertussis antigens into mLNs by migratory DCs and this might delay induction of the adaptive immune response to infection. We thus compared the levels of Bp-specific IgG antibodies in sera collected from the infected mice as early as 14 days after bacterial inoculation (Fig 9). Intriguingly, the sera of Bp PT --challenged mice contained a substantial level of B. pertussis-specific IgG antibodies, exhibiting a titer of~2,000 already 14 days after infection. In contrast, no anti-B. pertussis IgG was detectable in sera collected 14 days after infection from mice inoculated by the PTproducing wild-type or Bp AC − mutant bacteria that proliferated in mouse lungs to substantially higher levels and persisted longer than the Bp PT − bacteria. In line with these antibody responses, already on day 5 of infection the splenocytes of mice infected with Bp PT − responded to antigenic restimulation by higher production of inflammatory cytokines (e.g. IFNγ, TNF-α, IL-6 and IL-12p70) than splenocytes of mice infected by the PT producing wildtype bacteria (S17 Fig). Intriguingly, some B. pertussis-specific IgG were also detectable in sera of mice infected by the Bp AC -PT − double mutant despite its rapid clearance from the lungs.

Discussion
We report that PT activity is critical for prevention of migratory dendritic cell-mediated delivery of B. pertussis bacteria from the infected lung parenchyma into the lymphoid tissue of lung-draining mediastinal lymph nodes (mLNs). The PT-producing bacteria largely failed to enter the mLNs and remained trapped within the associated tissue outside of mLNs, or localized to the capsule or the subcapsular sinus, as expected for bacteria being simply shuttled from the infected lung parenchyma by the lymph. In contrast, mLN cryosection microscopy revealed that in the absence of PT activity the cDC-associated Bp PT − bacteria were delivered green, blue, and magenta colors, respectively. The images are representative of one experiment performed in groups of three mice. Scale bar 100 μm. (B) Bacterial load in the trachea of infected mice on day 5 (full circles) and day 14 (open circles). Each symbol represents an individual animal, and the lines indicate the means. Data were pooled from two independent experiments. https://doi.org/10.1371/journal.ppat.1010577.g003

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response deep inside the T-cell zone (paracortex) of the LN, which is a zone accessed by migratory DCs delivering antigens for presentation to T cells. This indicates that the PT − bacteria were readily transported by migratory cDCs into deeper regions of mLNs. While being almost cleared from

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response the infected lungs by day 14, the ACT-producing PT − bacteria proliferated in mLNs by that time to higher counts than found in the lungs and relocated to the B-cell zone (cortex) of mLNs, where they often formed clusters. This suggests that the capacity of ACT to ablate the oxidative burst of phagocytes, block killing of phagocytosed non-opsonized bacteria and trigger apoptosis by cAMP-mediated signaling [34,35,41] enabled the release of live bacteria into the lymphoid tissue of mLNs from the apoptosing cDCs [83]. Indeed, when mice were infected with a mixture of green and magenta fluorescent Bp PT − bacteria, dual-labeled groups of phagocytosed bacteria were still observed in the LNs on day 5. However, on day 14 only unicolor bacterial foci were observed, revealing that each bacterial cluster originated by proliferation of a single founder bacterium inside the mLNs. Nevertheless, the PT − bacteria did not disseminate beyond mLNs to other organs within 14 days of infection and colonized the mLNs only transiently. By day 21 (cf. S4 Fig), the CFU counts in mLNs started to decline and viable PT − bacteria were nearly absent from the mLNs 5 weeks after infection (~10 0 −10 1 CFU), likely due to onset of antibody-dependent opsonophagocytic killing.
The inhibition of cDC-mediated delivery of PT-producing bacteria into the mLNs would be in line with PT-mediated perturbation of G i/o -associated chemokine receptor signaling (e.g. CCR7) and with the in vitro observed PT-dependent inhibition of dendritic cell migration towards the lymphoid chemokine CCL21 [69]. Inhibition of delivery of B. pertussis bacteria

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response into the LNs by PT would also suggest why the bacterium generally does not cause disseminated systemic infections in immunocompetent humans [84,85] despite being able to proliferate in the infected upper airways of human infants to as high levels as up to~10 8 B. pertussis CFU/ml of undiluted nasal aspirates [86]. Moreover, a striking influx of neutrophils into mLNs of Bp WT-infected animals was observed on day 14, suggesting a role of neutrophils in clearance of B. pertussis from the lymphatic system (cf . Figs 6 and 8).
In contrast, the B. bronchiseptica species, which shares a common ancestor with B. pertussis and produces ACT but not PT [87], was previously found to infect the NALT and mLNs of experimentally infected mice [88] and was reported to cause disseminated infections in immunocompromised humans [89,90]. Indeed, B. bronchiseptica triggers an early migration of DCs from the infected airways into the mLNs [88]. Moreover, B. holmesii that does not produce any of the classical virulence factors (PT, ACT, Type III secretion systems, pertactin or fimbriae) is still able to cause a disseminated infection and pertussis-like illness with a much higher capacity of invasiveness into other tissues than B. pertussis [91].
The delivery of PT-deficient bacteria into mLNs indicates that one of the biological roles of PT consists in restriction of B. pertussis dissemination beyond the mucosa by phagocytic cells, so as to limit the early immune response to infection. In this respect, B. pertussis differs importantly from many other pathogens that exploit cDCs as a Trojan horse for delivery of live microorganisms into mLNs to cause systemic infections that are part of their natural life cycle [75,[77][78][79][80]. In the case of B. pertussis, such hijacking of cDCs by the Bp PT − bacteria devoid of PT activity appears as counterproductive. Delivery of PT-deficient Bp PT − bacteria into mLNs ramped up the immune response to Bp PT − infection. Indeed, it has previously been shown for M. tuberculosis that the mere presence of bacteria in close proximity to lymphocytes within the LNs can boost antigen presentation and the overall immune response to infection [92]. It will be of interest to examine if also the PT − B. pertussis BPZE1 bacteria are trafficked into mucosa-draining LNs and whether this enhances the immunogenicity of the intranasally applied live attenuated BPZE1 pertussis vaccine, which has already passed phase IIb clinical trials in adults and advances into trials in school age children (https://www.iliadbio.com/ clinical.html).
We hypothesized previously that the cAMP signaling activity of ACT might drive outmigration of immature intraepithelial DCs into draining LNs in the course of B. pertussis infection, as ACT action was found to hamper maturation of DCs while increasing their LPS-stimulated chemotactic migration in vitro [46]. Indeed, on day 5 of lung infection with the ACT-producing Bp PT − bacteria we observed a significant increase in the absolute numbers, as well as in the relative proportion, of migratory conventional DCs (both CD11b -cDC1 and CD11b + cDC2) in the lung-draining mLNs (cf. Fig 7A and 7B). These CD11b -cDC1 likely corresponded to the intraepithelial DC population [93,94] and the most prominent cell type associated with the Bp PT − bacteria on day 5 in the mLNs was phenotyped as MHC-II high CD11c int migratory cDC1 (CD11b -) cells. However, such cDC number increase in mLNs was not observed upon lung infection by the PT-producing wild-type bacteria that proliferated in the lungs to importantly

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response of an early adaptive immune response to infection. Indeed, others have observed in mouse models that anti-B. pertussis antibody responses appear in only~3-4 weeks, or even later upon infection by PT-producing wild-type B. pertussis, later when the bacteria are already being cleared from the infected lungs through antibody-dependent phagocytosis by neutrophils [52,[95][96][97]. We found that upon infection by the Bp PT − mutant a robust serum IgG antibody response to B. pertussis infection was detected as early, as on day 14 after infection. It is, hence, plausible to assume that the PT-provoked delay in delivery of bacterial antigen into LNs serves to delay the induction of antibody response to infection, in line with a previous observation that PT action suppressed serum antibody responses to immunodominant B. pertussis antigens [53]. This would enable B. pertussis to proliferate to the observed high levels on the mucosa of the upper airway during the catarrhal phase of infection and would support its efficient aerosolmediated transmission to new hosts.

Bp-associated cells were visualized in simple dot plots from down-sampled live mScarlet + events (left panels): MHC-II + CD11c + dendritic cells (DC, red dots), CD19 + CD3 -B cells (blue dots, S14 Fig) and
CD11b + Ly6G + neutrophils (green dots). Pie charts on the right-hand panels show the distribution of subpopulations of mScarlet + cells determined by in depth phenotyping (see S12 Fig for gating strategy). Data from one representative experiment out of 3 (Bp WT) or 4 (Bp PT -) performed are shown. (B) and (C) In parallel to cytometric analysis, cryosections of mLNs from Bp PT --infected mice were prepared on days 5 and 14 and mScarlet + cells were visualized by immunofluorescence microscopy. (B) CD11c + dendritic cells were detected on day 5 with biotin-conjugated anti-CD11c followed by AF488-conjugated streptavidin (green). T cells were detected with AF647-labeled anti-CD3 (white). (C) on day 14, neutrophils were detected with biotin-conjugated rat Ly6G antibody followed by AF488-conjugated streptavidin (green) and B cells were detected with rat anti-B220 antibody followed by goat anti-rat AF488-labeled secondary antibody (green). Images were acquired at 63x magnification using Leica TCS SPE confocal microscope. Scale bar 10 μm. https://doi.org/10.1371/journal.ppat.1010577.g008

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response

Antibodies and reagents
The antibodies used in this study are listed in S1 Table. Rabbit anti-B. pertussis serum was kindly provided by Branislav Vecerek (Institute of Microbiology, Prague, Czech Republic). Anti-ACT (clone 9D4) monoclonal antibody was provided by Erik Hewlett (University of Virginia, USA). The monoclonal antibody against the S1 subunit of pertussis toxin (clone 63.1G9) was purchased from Santa Cruz Biotechnology (USA). EZ-Link Sulfo-NHS-Biotin was from Thermo Fisher Scientific (USA).

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response days after challenge), mice were sacrificed by cervical dislocation, and the lungs, trachea, spleen and thymus were aseptically removed and homogenized in PBS using a tissue grinder (Heidolph RZR 2020, Merck, Germany). The removed mediastinal lymph nodes (mLN), cervical lymph nodes (cLN), and inguinal lymph nodes (iLN) were gently homogenized in cold PBS by passage through a Corning 70 μm cell strainer (Corning, USA). The numbers of viable bacteria in mouse organs were determined as colony-forming units (CFU) upon plating of tissue homogenates on BG agar plates. The percentage of non-fluorescent bacteria in the organs was determined by counting of fluorescent and non-fluorescent colonies using a G:Box Chemi XRQ gel doc system (Syngene, UK).

Flow cytometry
Lymph nodes were mechanically disrupted by needles and homogenized in high glucose Dulbecco's Modified Eagle's medium (DMEM, Sigma) by enzymatic digestion with 2 mg/ml of collagenase D and 40 μg/ml of DNase I (Roche) at 37˚C for 30 min. EDTA (14 mM) was added and cell suspensions were filtered through a 50 μm CellTrics strainer (Sysmex Partec, Germany). Prior to flow cytometric analysis, the cells were blocked with 10% BALB/c mouse serum and anti-mouse CD16/CD32 monoclonal antibody (12.5 μg/ml; eBioscience) in wash buffer for 30 min on ice and stained with fluorochrome-labeled monoclonal antibodies (S1 Table) and fixable viability dye eFluor 780 (eBioscience) for 30 min on ice in the dark. Where appropriate, CountBright Absolute Counting Beads (Invitrogen, USA) were added for determination of absolute cell counts. Cytometric data were acquired on a BD LSRII flow cytometer (BD Biosciences, USA) and analyzed using FlowJo v10 software (Tree Star, USA). The compensation matrix was calculated using single stain controls. To detect fluorescent bacteria associated with cells, live cells from tissues of mice infected with fluorescent or nonfluorescent bacteria were downsampled, using a DownSample FlowJo plugin to obtain identical numbers of cells, and then gated based on mScarlet fluorescence (585/15 nm emission filter).

Histology and immunohistochemistry
Mice were sacrificed at indicated time points, the tracheas were cannulated with a 26-gauge needle and the lungs were inflated with 1 ml of 4% (w/v) PBS-buffered formaldehyde (PFA) solution. The entire respiratory tract and the mLNs were excised, placed in 4% PFA for 24 hours, transferred into 70% (vol/vol) denatured ethanol and dehydrated with ethanol series using a Leica ASP 6025 tissue processor (Leica Biosystems, Germany). The samples were embedded in paraffin using a Leica EG1150 C embedding system (Leica Biosystems, Germany) and cut to 2 μm-thick sections using a Leica RM2255 rotary microtome (Leica Biosystems, Germany). Hematoxylin and eosin staining (HE) was performed using an automated Leica ST5020 multistainer (Leica Biosystems, Germany). For immunohistochemical (IHC) staining of B. pertussis cells, tissue sections were rehydrated and antigen was retrieved at 95˚C for 15 min at pH 6.0 with antigen retrieval solution (Zytomed Systems, Germany). Endogenous peroxidase was blocked with 3% (v/v) hydrogen peroxide in methanol. Sections were incubated with a rabbit anti-B. pertussis serum diluted 1:1,000 overnight at 4˚C, followed by a 30-min incubation with the secondary anti-rabbit antibody conjugated to horseradish peroxidase (Zytomed Systems, Germany) at 25˚C. Signal was developed in 3-amino-9-ethylcarbazole (AEC) solution (Zytomed Systems, Germany) for 10 min at 25˚C. Sections were counterstained with hematoxylin (Sigma, USA), mounted with Aquatex mounting medium (Merck-Millipore, Germany), and scanned with an AxioScan.Z1 automatic slide scanner at 40x magnification (Carl Zeiss, Germany). Images were processed using ZEN software (Carl Zeiss, Germany).

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response

Immunofluorescence microscopy of frozen tissue sections
Immunofluorescence analysis of lungs, trachea, and lymph nodes was performed at indicated time intervals upon intranasal challenge of mice with PBS (control) or B. pertussis. For in vivo epithelial barrier permeability assessment, 50 μl of NHS-biotin in PBS (1 mg/ml) was intranasally applied to anesthetized mice for 5 min. Mice were euthanized by cervical dislocation, their lungs were inflated with 1 ml of 4% PFA, excised, and placed into 4% PFA for 24 h at 4˚C. The PFA-fixed tissues were transferred into PBS-buffered 30% sucrose for 24 h at 4˚C, embedded in a Tissue-Tek OCT cryo-embedding medium (Sakura Finetek, USA) and frozen in liquid nitrogen-cooled isopentane. The cryomolds were cut into 10 or 30 μm thick sections using a Leica CM1950 cryotome (Leica Biosystems, Germany) and were mounted on SuperFrost Plus adhesion slides (Thermo Fisher Scientific, USA) and stored at -20˚C. Before staining, the slides were thawed at 25˚C for 1 h and washed with PBS-T (PBS with 0.05% Tween 20). After blocking (5% [w/v] BSA in PBS or 5% [v/v] goat serum in PBS for goat secondary antibodies) and permeabilization (PBS supplemented with 0.3% [v/v] Triton X-100), the cryosections were stained overnight at 4˚C with unconjugated, fluorochrome-or biotin-conjugated primary antibodies (antibody details in S1 Table) diluted in PBS containing 0.2% (w/v) BSA, followed by incubation at 20˚C with the corresponding fluorochrome-conjugated secondary antibodies and Alexa Fluor-conjugated streptavidin (Thermo Fisher Scientific, USA) for 1 h and 30 min, respectively. In vivo biotinylation of tracheal epithelial cells was visualized directly by incubating the cryosections with Alexa Fluor 488-conjugated streptavidin (Thermo Fisher Scientific, USA). B. pertussis cells were labeled with rabbit anti-B. pertussis serum (diluted 1:5,000) followed by goat anti-rabbit Alexa Fluor 647-conjugated secondary antibody. The cell nuclei were stained with DAPI. Specificity of staining was verified by using the appropriate immunoglobulin isotype controls. Specimens were mounted in Vectashield Antifade Mounting Medium (Vector Laboratories, USA) and imaged using a Leica TCS SPE confocal microscope (Leica Biosystems, Germany). Stitched images of entire LN tissue sections were acquired at 40x magnification using an IX83 motorized and automated fluorescence microscope (Olympus, Japan) with cellSens software (Olympus, Japan). The images were processed using Fiji [102]. Where appropriate, background inhomogeneity (shading) was corrected using the plugin BaSiC [103].

Immunodetection of secreted proteins
Proteins secreted into B. pertussis culture supernatants (13,000 x g, 5 min) were probed in immunoblots with appropriate monoclonal antibodies recognizing the ACT and PT proteins, decorated with horseradish peroxidase-conjugated goat anti-mouse secondary antibody (GE Healthcare, USA) and detected using the SuperSignal West Femto substrate (Thermo Fisher Scientific, USA) in a G:Box Chemi XRQ Gel Doc System (Syngene, UK).

IgG level determination
Serum samples were prepared by centrifugation of whole blood at 5,000 × g at 4˚C for 15 min after retro-orbital bleeding of anesthetized mice. Total IgG levels against B. pertussis antigens were determined using a whole bacterial cell ELISA as previously described [11] and the antibody titers were calculated as inflection points of titration curves.

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response the bronchopneumonia observed by day 5 in Bp WT, AC − and PT --infected mice was characterized by peribronchial interstitial edema with abnormal bronchial dilation, and by localized peribronchial, perivascular and alveolar inflammation, as compared to the healthy mouse lungs with normal bronchial and alveolar structure. The Bp WT strain provoked the highest degree of inflammatory damage. The low-colonizing double mutant Bp AC -PT --infected mice showed only mild alveolar inflammation by day 5, characterized by sparse interstitial infiltrates. Pseudostratified bronchial epithelium was not damaged in any group by day 5. The highest pathology was observed in Bp WT-infected mice on day 14. The massive inflammation affected a large portion of the parenchyma, causing a lobar pneumonia. Bronchi were dilated and scarred, surrounded by diffuse peribronchial inflammation. The Bp AC − strain elicited a more pronounced bronchopneumonia on day 14 compared to day 5 and compared to the Bp  S1 Table for details). Only single cell events from stable flow rate over time were selected, followed by dead cell elimination using a Fixable Viability Dye (FVD) eFluor 780 live/dead stain. After the gating of neutrophils (CD11b + Ly6G + ), plasmacytoid DCs (pDCs) were identified as Ly6C high CD11band further confirmed to be CD11c + MHC-II low . T-cells were separated using a CD3 marker. Dendritic cells (DCs) were identified on the basis of their autofluorescence in V500 channel together with CD11c and MHC-II positivity.
Conventional DCs (cDCs) were left after a population of monocyte-derived DCs (moDCs), highly expressing Ly6C and CD11b markers, was gated out. Conventional DCs were then separated into migratory and resident cDCs according to Sheng et al., who showed that the MHC-II high CD11c int population had a phenotype of migratory cells, confirmed by their high expression of CCR7 and ability to be labeled by intranasally-applied CFSE [107]. Both migratory and resident cDCs were further separated on the basis of their CD11b expression into CD11b low cDC1 (which typically express CD8 in lymphoid organs and CD103 in non-lymphoid organs) and CD11b + cDC2 (CD11b + DCs). Non-T and non-DC cells were further gated into MHC-II + CD19 + B-cells, SSC high CD11b + eosinophils and other SSC low CD11b + myeloid cell populations defined as monocytes/ macrophages (Mono/Mfs). The remaining gate contained some NK cells and erythrocytes. (TIF) S13 Fig. Majority of B. pertussis isolated from mLNs were associated with cells. Mice were intranasally inoculated with 50 μl of bacterial suspension containing 8 x 10 5 CFU of the indicated B. pertussis strains (mScarlet + ). Pools of mLNs from individual mice were collected. mLN homogenates were prepared by pressing the mLNs through a 70 μm cell strainer in cold PBS. To separate the cell-associated and free bacteria in mLN suspensions, LN homogenates were kept on ice to prevent phagocytosis and bacteria were separated from mouse cells by lowspeed centrifugation (300 × g, 5 min, 4˚C). No antibiotic treatment was applied to kill extracellular bacteria. Pellets resuspended in PBS and supernatants were then plated separately on BG agar plates. A portion of CFU recovered from pellets was identified as cell-associated bacteria.  Fig 8C. Neutrophils on cryosections of mLNs from Bp PT --infected mice on day 14 were stained with biotin-conjugated rat Ly6G antibody followed by AF488-conjugated streptavidin. Orthogonal views of Bp PT − mScarlet-infected neutrophils clearly show intracellular localization of bacteria in neutrophils. Z-stack images were acquired using a confocal microscope. Neutrophils, nuclei and bacteria are rendered in green, blue, and magenta colors, respectively. Scale bar 10 μm. (TIF) S17 Fig. Cytokine secretion by B. pertussis antigen-restimulated splenocytes from mice on days 5 and 14 after infection. Mice were intranasally inoculated with 50 μl of bacterial suspension containing 8 x 10 5 CFU of the indicated B. pertussis strains (mScarlet + ) and on day 5 and 14 spleens were collected. Splenocytes were isolated as previously described [11] and restimulated for 48 h by heat-killed B. pertussis WT (Tohama I), or by PBS and PMA/ionomycin (eBiosciences) used as negative and positive controls, respectively. IFN-γ, IL-17A, IL-2, IL-6, IL-10, TNF-α and IL-12p70 concentrations were determined in the culture supernatants of splenocytes using a custom-made ProcartaPlex cytokine bead assay (ThermoFisher Scientific, USA) on a Bio-Plex 200 instrument (Bio-Rad, USA). The analysis was done once (n = 3 mice/ group

PLOS PATHOGENS
Pertussis toxin blocks infection of lymph nodes by Bordetella pertussis and early immune response