(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .
Pertussis toxin suppresses dendritic cell-mediated delivery of B. pertussis into lung-draining lymph nodes [1]
['Nela Klimova', 'Institute Of Microbiology Of The Czech Academy Of Sciences Prague', 'Czech Republic', 'Faculty Of Sciences', 'Charles University', 'Prague', 'Jana Holubova', 'Gaia Streparola', 'Czech Centre For Phenogenomics Biocev', 'Vestec']
Date: 2022-07
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.
Of the three classical Bordetella species causing respiratory infections in mammals, only the human-specialized whooping cough agent B. pertussis produces the pertussis toxin (PT) as its major virulence factor. Human pertussis is an acute respiratory illness and the pleiotropic activities of pertussis toxin account for the characteristic systemic manifestations of the disease, such as hyperleukocytosis, histamine sensitization, hyperinsulinemia, or inflammatory lung pathology. We found that PT activity inhibits the migration of infected dendritic cells from the lungs into the draining mediastinal lymph nodes (mLNs). This prevents mLN infection by bacteria evading from migratory cells and delivery of bacterial antigens into mLNs. As a result, the induction of adaptive serum antibody responses to infection is delayed. We thus propose that PT action serves to create a time window for proliferation of B. pertussis on airway mucosa to facilitate transmission of the pathogen among humans.
Funding: This work was supported by the Czech Science Foundation grant GX19-27630X (P.S.). and GA22-23578S (L.Bu.). N.K. was also supported by the GAUK 507116 project of the Charles University in Prague. The results were acquired at the National Infrastructure for Biological and Medical Imaging at the BioImaging Facility of the Institute of Physiology, CAS - Large RI Project LM2018129 Czech-BioImaging funded by the Ministry of Education, Youth and Sports of the Czech Republic (MEYS CR), and ERDF project No. CZ.02.1.01/0.0/0.0/18_046/0016045, the Czech National Node to the European Infrastructure for Translational Medicine, EATRIS - LM2018133 and the Czech Centre for Phenogenomics LM2018126 funded by MEYS CR, and OP RDE CZ.02.1.01/0.0/0.0/18_046/0015861 CCP Infrastructure Upgrade II by MEYS and ESIF and OP RDE CZ.1.05/2.1.00/19.0395 a CZ.1.05/1.1.00/02.0109. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
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.
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 – 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 – 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 – 68 ].
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 – 55 ].
ACT is produced also by B. parapertussis and B. bronchiseptica and was proposed to play an important role in the early stages of airway colonization [ 17 – 20 ]. The 177 kDa-large ACT protein harbors an N-terminal adenylyl cyclase (AC) enzyme domain (~40 residues) that is fused to a ~1,306 residue-long RTX hemolysin (Hly) moiety [ 21 ]. Hly binds the complement receptor 3 (CR3, α M β 2 integrin CD11b/CD18, or Mac-1) of myeloid phagocytes [ 22 – 26 ], penetrates phagocyte membrane and delivers into cells the AC domain [ 27 ]. The AC enzyme is activated by cytosolic calmodulin and catalyzes a rapid and unregulated conversion of cellular ATP into the key second messenger molecule cAMP [ 28 – 30 ]. Increased cAMP levels deregulate cellular signaling pathways and ablate key innate immunity mechanisms, such as the oxidative burst of neutrophils [ 31 – 34 ], bactericidal NO production of macrophages [ 35 ] and the opsonophagocytic capacities of myeloid phagocytes [ 36 – 40 ]. The ACT-produced cAMP signaling also blocks transition of incoming monocytes into the more bactericidal macrophages and triggers dedifferentiation and apoptosis of tissue-resident macrophages [ 13 , 14 , 41 – 44 ]. Finally, ACT-produced cAMP elevation hampers also the adaptive immune responses, inhibiting dendritic cell (DC) maturation and blocking antigen presentation to T cells [ 45 , 46 ].
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 – 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 – 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 – 16 ].
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 PT-producing 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 wild-type 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.
mLNs of mice infected with 50 μL of bacterial suspension containing 8 x 10 5 CFU of the mScarlet + B. pertussis strains were collected and pooled on days 5 and 14 using 4 mice per condition. (A) Cellular suspensions of mLNs were prepared by enzymatic disruption and cellular subpopulations were stained with a panel of fluorescently-labeled antibodies and analyzed by flow cytometry. The indicated numbers represent mScarlet + cells detected per 2 x 10 6 events on day 5 and 14, except for day 14 of the mLN sample of mice infected by Bp PT – , where the sample of analyzed cells was increased to 6 x 10 6 events. mScarlet + cells associated with Bp bacteria were detected using 585/15 nm emission filter and the gate was set using cellular suspension of mLNs from mice infected with the non-fluorescent control bacteria ( S14 Fig ). 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.
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 events ( S14 Fig ). Upon collection of 10 6 −10 7 events, dozens to hundreds of mScarlet-positive cells were reliably detected, thus representing ~0.002–0.01% of live cells in mLN suspensions ( S14 Fig ). As depicted on Figs 8A and S14 , the mScarlet + cells detected in mLNs from Bp WT-infected mice on day 5 phenotyped as CD19 + MHC-II + B cells (~45%) and Ly6G + CD11b high neutrophils (~37%), with few infected DCs (~4%) detected. In contrast, DCs reproducibly made up the largest share of Bp PT – -infected cells. The mScarlet + Bp PT − bacteria were mostly associated with migratory CD11b – cDC1 cells (43%), migratory CD11b + cDC2 cells (5%), or other dendritic cell (6%) populations. Fluorescence microscopy of mLN cryosections revealed that such Bp PT – -harboring CD11c + cells of dendritic shape were often in contact with CD3 + T-cells (Figs 8B , see S15 for more images). These results indicate that by day 5 of infection, the Bp PT − bacteria were transported from the infected lungs into mLNs mostly by migratory cDC1 cells. Later, by day 14 of infection (Figs 8A and S14 ), ~21% of cells containing mScarlet + bacteria were polymorphonuclear Ly6G + neutrophils, often harboring multiple intracellular Bp PT − bacteria (Figs 8C , see S16 for more images), while a larger fraction (~45%) of mScarlet + Bp PT − bacteria were associated with CD19 + MHC-II + B cells ( Fig 8A ) located in the B (B220 + ) cell zone on mLN cryosections by day 14 of infection (cf. Figs 4A , 8C , S9 and S10 ).
Mice were intranasally infected with 8 × 10 5 CFU of the indicated B. pertussis strains expressing mScarlet fluorescent protein. Collected mLNs were pooled, enzymatically disrupted and the cell suspensions were analyzed by flow cytometry using a panel of monoclonal antibodies and cell counting beads. (A) Representative dot plots of the migratory cDCs (MHC-II high CD11c int ) and resident cDCs (MHC-II int CD11c high ) detected in mLNs of infected mice on day 5 (upper panel) and 14 (lower panel). The indicated conventional dendritic cell (cDC) numbers (lower left corners) were gated-out from 100,000 viable singlets per sample as described in S12 Fig . The percentage (%) of cDC subpopulations are indicated. (B) Total counts of migratory and resident cDC1 (CD11b – ) and cDC2 (CD11b + ) cells in mLNs of infected mice on days 5 and 14. Each symbol represents the value for an individual animal. Data represent mean values and standard deviations for groups of four mice per time point (or 2 mice in Bp AC – PT − group) from two independent experiments. Statistical significance between groups was analyzed by one-way ANOVA followed by Tukey’s multiple comparisons test. * (p < 0.05), ** (p < 0.01), *** p (< 0.001).
Total counts of cells per mLN pool per infected mouse on day 5 and 14. Mice were intranasally infected with 8 × 10 5 CFU of the indicated B. pertussis strains expressing mScarlet fluorescent protein. Collected mLNs were pooled, enzymatically disrupted and the cell suspensions were analyzed by flow cytometry using a panel of monoclonal antibodies and cell counting beads. Each symbol represents the value for an individual animal. Data represent mean values and standard deviations for groups of four mice per time point (or 2 mice in Bp AC – PT − group) from two independent experiments. Statistical significance between groups was analyzed by one-way ANOVA followed by Tukey’s multiple comparisons test. * (p < 0.05), ** (p < 0.01), *** p (< 0.001). moDCs, monocyte-derived dendritic cells; pDCs, plasmacytoid dendritic cells; Mono/Mfs, monocytes/macrophages.
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. B- and 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 – 82 ].
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 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.
(A) Accumulation of mScarlet-producing Bp PT − bacteria in the mLNs of infected mice. 30 μm cryosections of mLNs were examined by confocal microscopy. Scale bar 10 μm. (B) Quantification of bacterial clustering in the mLNs. Violin plots of the number of detected bacteria per site at the indicated time points. 30 randomly selected fluorescent foci containing at least one bacterium were examined per time point. Clusters containing more than 4 individual bacteria were categorized as > 4. Data were obtained from serial cryosections of at least three mLNs in one experiment. (C) Representative mLN cryosections (30 μm) from day 5 and 14 of mice infected with a 1:1 mixture of Bp PT − strains producing mScarlet (magenta) and GFP (green) fluorescent proteins. Images were acquired at 63x magnification using a Leica TCS SPE confocal microscope. (D) Quantification of unicolor and multicolor clusters in mLNs of mice challenged with a 1:1 mixture of Bp PT − strains producing mScarlet and GFP fluorescent proteins. 40 randomly selected bacterial clusters containing at least 2 individual bacterial cells were analyzed per time point. The predominance of mScarlet- over GFP-labeled unicolor clusters was not due to enhanced fitness but rather reflected easier microscopic detection of the brighter mScarlet-producing bacteria. Comparable numbers of magenta and green fluorescent colonies were recovered by plating of mLN homogenates on BG blood agar plates. Data were obtained from serial cryosections of at least six mLNs derived from groups of at least two mice from three independent experiments.
(A) Immunofluorescence microscopy of cryosections of mLNs of infected mice on days 5 (upper panel) and 14 (lower panel). mLNs were fixed with 4% PFA, snap frozen and 10 μm longitudinal cryosections were first labeled with rat anti-mouse CD45R (B220), followed by goat anti-rat Alexa Fluor 488 secondary antibody conjugate and finally by Alexa Fluor 647 rat anti-mouse CD3 antibody conjugate. Nuclei were labeled with DAPI. Stitched images were acquired at 40x magnification using an Olympus IX83 motorized automated microscope. Upper panels show representative images of entire mLN sections (scale bar 1000 μm). The white squares locate the zoomed-in areas shown in lower panels (see S6 , S7 , S8 and S9 Figs for high resolution images). Lower panels depict the details of the indicated areas (scale bar 20 μm). The mScarlet-expressing bacteria are encircled by yellow dotted lines. T cells, B cells, nuclei and bacteria are rendered in white, green, blue and magenta colors, respectively. In total 8 mLNs from three mice per infection group were examined and representative images are shown. (B) Location of bacteria on mLN sections. Individual bacteria were counted across entire mLN sections and location of bacteria in the indicated mLN zones (schematic drawing) was recorded. Mean bacterial counts of 6 mLN sections from 3 mice per infection group are shown.
(A) Tracheal 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. Respiratory tracts were fixed with 4% PFA, snap frozen and 10 μm transversal cryosections of tracheas were labeled with Alexa Fluor 488 streptavidin conjugate. Nuclei were visualized by DAPI staining. Images were acquired using a Leica TCS SPE confocal microscope. 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. (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.
(A) Immunohistochemistry of longitudinal lung sections from infected mice on day 5 (left panel) and day 14 (right panel) after intranasal administration of PBS (control) or 8 × 10 5 CFU of 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.
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 lung-draining 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 ).
(A) B. 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.
To dissect the impact of ACT and PT activities on outmigration of DCs from B. pertussis-infected 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 only ~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 ]).
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 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 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 (~100−101 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 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 ~108 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 [75,77–80]. 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-IIhighCD11cint 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 higher levels. This suggests that PT action inhibited the LOS/TLR4-elicted and ACT-potentiated chemotactic cDC outmigration from the site of infection. Delaying the delivery of bacterial antigens for presentation to B and T cells in the lymph nodes would then likely delay the induction 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 [52,95–97]. 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 aerosol-mediated transmission to new hosts.
[END]
---
[1] Url:
https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010577
Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.
via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/
