Pharmacological targeting of host chaperones protects from pertussis toxin in vitro and in vivo

Whooping cough is caused by Bordetella pertussis that releases pertussis toxin (PT) which comprises enzyme A-subunit PTS1 and binding/transport B-subunit. After receptor-mediated endocytosis, PT reaches the endoplasmic reticulum from where unfolded PTS1 is transported to the cytosol. PTS1 ADP-ribosylates G-protein α-subunits resulting in increased cAMP signaling. Here, a role of target cell chaperones Hsp90, Hsp70, cyclophilins and FK506-binding proteins for cytosolic PTS1-uptake is demonstrated. PTS1 specifically and directly interacts with chaperones in vitro and in cells. Specific pharmacological chaperone inhibition protects CHO-K1, human primary airway basal cells and a fully differentiated airway epithelium from PT-intoxication by reducing intracellular PTS1-amounts without affecting cell binding or enzyme activity. PT is internalized by human airway epithelium secretory but not ciliated cells and leads to increase of apical surface liquid. Cyclophilin-inhibitors reduced leukocytosis in infant mouse model of pertussis, indicating their promising potential for developing novel therapeutic strategies against whooping cough.


Specific pharmacological inhibition of chaperone and PPIase activities protects cells from intoxication with PT.
A previously established morphology based assay of cellular intoxication was used to determine the functional role of host cell chaperones and PPIases during PT-intoxication 29,35 . CHO-K1 cells show a characteristic clustering morphology if treated with PT ( Fig. 1). CHO-K1 cells were pre-incubated prior to PT-treatment with radicicol (Rad) to block Hsp90-activity, and FK506 or CsA to inhibit FKBP or Cyp activity, respectively. Presence of Rad, CsA or FK506 robustly reduced CHO-K1 cell clustering compared to cells treated with PT (Fig. 1a). BFA was used as an inhibitor of PT-induced CHO-K1 cell clustering 17,29,36 . Additionally, fewer cells were observed in wells treated with PT (Fig. 1a). This effect was also inhibited by Rad, CsA and FK506. Hsp70 inhibitors VER-155008 (VER) and HA9 37 both inhibited PT-intoxication (Fig. 1b). The inhibitors alone had no effect on cell morphology or cell number (Supplemental Fig. 1a,b). Taken together, the results indicate a functional role of Hsp90, Hsp70, Cyps and FKBPs for PT-intoxication.

Inhibition of chaperone/PPIase activities leads to reduced ADP-ribosylated Giα without interfering with enzyme activity of PTS1 or binding of PT to cells. After incubation of CHO-K1 cells
with PT in the presence or absence of the chaperone and PPIase inhibitors, cells were lysed and incubated with fresh PTS1 in the presence of biotin-labeled NAD + . During this incubation, the portion of Giα, which has not been modified by PTS1 during the previous treatment of living cells with PT, was ADP-ribosylated in vitro 29 . This results in biotin-labeling of Giα due to the covalent transfer of the biotin-labeled ADP-ribose moiety from NAD + . Biotin-labeled i.e. ADP-ribosylated Giα was then detected by Western blotting. A strong signal was obtained from control samples, whereas a weaker signal was observed in the presence of PT, showing that in PT-treated cells a substantial amount of Giα was already ADP-ribosylated and could not serve as substrate in the subsequent in vitro ADP-ribosylation reaction. Pre-treatment of cells with chaperone/PPIase inhibitors reduced the ADP-ribosylation of Giα compared to cells treated with PT alone (Fig. 2a). Treatment of cells with the inhibitors only did not affect the subsequent in vitro ADP-ribosylation of Giα by PTS1 (Supplemental Fig. 1c).
Inhibition of Cyps and FKBPs prevented PT-mediated cAMP increase in the iGIST bioassay (Fig. 2b). This assay is based on HEK293 cells that express the Giα-coupled somatostatin receptor 2 (SSTR2) as well as a luminescent cAMP probe 38  www.nature.com/scientificreports/ octreotide to activate SSTR2 which inhibits adenylate cyclase activity show a moderate increase in luminescence signal, i.e. intracellular cAMP (Fig. 2b). In contrast, in samples treated with PT, an enhanced increase in cAMP levels was detected because SSTR2-mediated inhibition of the adenylate cyclase through Giα was impaired. If cells were treated only with forskolin, a comparable increase in cAMP levels was detected with or without PTtreatment (Supplemental Fig. 2b). CsA and FK506 prevented the PT-mediated increase while only marginally affecting cAMP levels in the absence of PT (Fig. 2b). Rad, VER and BFA reduced cAMP levels in the absence of PT (Supplemental Fig. 2a). An inhibitory effect of Rad, VER and BFA on PT-mediated cAMP increase could therefore not be detected with this assay. Solvents of inhibitors had no effect on cAMP levels (Supplemental Fig. 2a). Taken together, these results suggest that inhibition of chaperones/PPIases reduced PTS1-mediated cytosolic ADP-ribosyltransferase activity. To explain this, we have developed competing hypotheses: (1) the inhibitors reduce the enzyme activity of PTS1 or (2) the inhibitors interfere with the transport of PTS1 into the cytosol.
None of the inhibitors impaired the ADP-ribosylation of Giα by PTS1 in vitro (Fig. 2c, Supplemental Fig. 2c), suggesting that the inhibitors prevent uptake of PTS1 into the cytosol. Binding of PT was not inhibited by were pre-incubated with 10 µM BFA, Rad, FK506 or CsA or left untreated for control. After 30 min 10 ng/ml PT were added. 1.5 h later the culture medium was removed, and cells were further incubated at 37 °C and 5% CO 2 in fresh medium that did not contain PT or any inhibitor. Pictures were taken after 18 h. A quantitative analysis of total cell number of CHO-K1 cells is shown, values are normalized on control cells (n = 3, mean ± SD). (b) Pharmacological inhibition of Hsp70 activity protects cells from PT-intoxication. CHO-K1 cells were treated with VER (30 μM) or HA9 (20 µM) for 30 min and then intoxicated with PT (10 ng/ml) for 18 h. Pictures were taken, and cell numbers were determined as described in A. Significance was tested by one-way ANOVA with Dunnett's multiple comparisons test and refers to samples treated with PT only (*p < 0.05, **p < 0.01, ***p < 0.001). Scale bar = 50 µm. www.nature.com/scientificreports/ chaperone/PPIase inhibitors (Fig. 2d). This implies that another step of toxin uptake, such as the intracellular membrane transport of PTS1 into the cytosol is the target of these inhibitors.

In the presence of chaperone/PPIase inhibitors, less free PTS1 is detectable in cells.
Previously we used fluorescence microscopy to demonstrate that a monoclonal PTS1-antibody recognizes preferably PTS1 when it is detached from the B-subunit pentamer, i.e. cytosolic PTS1 29 . Detection of free PTS1 is decreased following incubation with chaperone/PPIase inhibitors compared to cells challenged with PT only (Fig. 3, Supplemental Fig. 3). This suggests that the inhibitors interfere with uptake of PTS1 into the cytosol. The reason for that might be that endocytosis, retrograde transport to the ER or translocation from the ER to the cytosol are affected. Clustering of cells was still observed in some samples (Rad-and CsA-treated), which might be caused by a small amount of PTS1 that still reached the cytosol in these cells. Notably, the PT-concentration needed for fluorescence experiments is higher than for intoxication experiments in Fig. 1. This might explain why protection by inhibitors against clustering is not as strong as shown in Fig. 1. A reduced PTS1 signal was also observed with BFA, which prevents transport of PT from the Golgi apparatus to the ER and the results are in line with our earlier findings that CsA inhibits uptake of PTS1 into the cytosol 29 . Hence, the reduction of the PTS1 signal in the presence of chaperone/PPIase inhibitors suggests that chaperones/PPIases are involved in uptake of PTS1 to the cytosol.

PT invades secretory but not ciliated cells in a human airway epithelium.
To determine if the phenomenon of chaperone/PPIase requirement for PT uptake is recapitulated in a model of primary human bronchial airway epithelium, we first demonstrated that PT intoxicates human primary basal cells from airway epithelium in a concentration-dependent manner (Supplemental Fig. 7). Although clustering was not observed in these cells upon PT-treatment, impairment of cell morphology was noted (Supplemental Fig. 7) suggesting some cell damage induced by PT. Application of Rad or CsA reduced ADP-ribosylation of Giα in PT-treated basal cells (Fig. 5a), as analyzed by sequential ADP-ribosylation. Inhibitors alone had no significant effect on ADP-ribosylation of Giα (Supplemental Fig. 8). Mechanistic studies performed in CHO-K1 cells demonstrated that Rad and CsA do not influence enzyme activity or receptor binding but reduce the cytosolic amount of PTS1 in target cells, suggesting that Hsp90 and Cyps are involved in the transport of PTS1 into the cytosol of these cells.
Basal cells that represent one cell type of human bronchial airway epithelial cells (hBAECs) can be differentiated to a functional airway epithelium at air-liquid interface conditions, containing ciliated and secretory cells. By analyzing the effect of PT on the differentiated cell layer by fluorescence microscopy, it became evident that PTS1 was selectively detected in secretory (CC10 or MUC5B positive) cells 41 , but not in ciliated cells (β-IV-tubulin positive) (Fig. 5b). Moreover, CsA but not Rad or FK506 reduced the amount of PTS1 detected by fluorescence microscopy in this functional airway epithelium (Fig. 6). Noteworthy, the PTS1-antibody preferentially recognizes PTS1 that is not bound to the B-oligomer, which is most likely located in the cytosol. Rad, CsA or FK506 alone had no adverse effects on morphology of the airway epithelium or tight junctions (Supplemental Fig. 9a). However, VER and BFA revealed strong adverse effects on the airway epithelium showing a reduced cell number for VER-treated cells and nearly no cells were detected after BFA treatment (Supplemental Fig. 9b). . PBS and C3 toxin from C. botulinum were used as negative controls. Membrane was cut and overlay with His-PTS1 (200 ng/ml) or PBST was performed. After extensive washing, bound PTS1 was detected with a specific antibody. PTS1-and PBST-overlayed membranes were detected on the same X-ray film and images were cropped for display purposes only. www.nature.com/scientificreports/  www.nature.com/scientificreports/ Next, effects of PT on functional aspects of the human bronchial airway epithelium were investigated. PT showed no significant effect on trans-epithelial electrical resistance (TEER) of the human bronchial airway epithelium (Fig. 7a). However, PT significantly increased the apical surface liquid (ASL) of the functional human airway epithelium compared to untreated controls (Fig. 7b). This effect was not impaired by the chaperone inhibitors suggesting that other mechanisms might play a role that do not require PTS1 activity in the cytosol such as binding of the B-pentamer to the cell surface. Chaperone/PPIase inhibitors alone led to no significant effects on ASL or TEER values of treated samples compared to control (Fig. 7, right panels).

Cyp inhibitors reduce leukocytosis in an infant model of pertussis disease. Finally, effects of
CsA and its non-immunosuppressive derivative NIM811 42 were investigated in an infant mouse model of pertussis. This model has recently been established and recapitulates several hallmarks of severe disease observed in humans such as leukocytosis and death, which are both PT-dependent 8 . Here, 7-day old mice were infected  www.nature.com/scientificreports/ with a PT-producing wild type strain of B. pertussis via aerosol and then treated intranasally with either CsA or NIM811 or vehicle. Treatment with CsA or NIM811 had no effect on the colony forming units detected from lung homogenates (Fig. 8). However, PT-induced leukocytosis was significantly reduced by CsA or NIM811 treatment in infant mice implicating a protective effect upon Cyp inhibition for the first time in vivo.

Discussion
Pertussis disease is a reemerging public health crisis for which there is currently no effective treatment. The current standard of care, treatment with macrolide antibiotics is only associated with improved symptoms when administered during the catarrhal stage of disease 43 . This is challenging as it is before the onset of the characteristic "whooping" cough. B. pertussis mediates disease through its ADP-ribosylating toxin (ADP-RT) PT. The ability of ADP-RTs to translocate their enzyme subunits to target cell cytosols has been the subject of much research (for review see 44,45 ). We have shown other membrane translocating ADP-RTs, such as Corynebacterium diphtheriae diphtheria toxin, C. botulinum C2 toxin, C. perfringens iota toxin, and C. difficile CDT, require the chaperone activity of heat shock proteins Hsp90 and Hsp70, Cyps as well as PPIases from the FKBP family to transfer their enzymatic subunits from vesicular compartments to the cytosol of target cells 37,40,[46][47][48][49][50][51][52][53] . Identifying commonalities in these toxins mechanisms of action may allow for the development of treatments which function for across diverse bacterial species. Our previous work identified Cyp inhibitors as potent inhibitors of PT intoxication of mammalian cells 29 . Previously, we determined that Cyp isoforms CypA and Cyp40 interact with PTS1 in vitro and are required for uptake of PTS1 into the cytosol. This study revealed host Hsp90, Hsp70, Cyps and FKBPs are required for uptake of PTS1 into the cytosol. A direct and specific interaction of PTS1 with Hsp90, Hsp70, CypA, Cyp40, FKBP51 and FKBP52 was detected in vitro and interaction with FKBPs was partly mediated by the PPIase domain of these proteins. Human bronchial derived airway epithelial cells differentiated under air-liquid interface conditions allowed us to identify secretory cells as targets for PTS1 intoxication. Further, pharmacological inhibitors of Cyps protected against leukocytosis in an in vivo murine model of infant pertussis disease demonstrating the clinical potential of these molecules.
ADP-ribosylation of GTP-binding regulatory protein Giα by PT 12 induces distinct morphological changes in CHO-K1 cells in a sensitive and reproducible manner 35 . This phenomenon is the basis of routine assays to determine monoclonal antibody function and to measure antitoxin responses 54,55 . Here, we utilized this CHO-K1 assay and pharmacological inhibitors of heat shock proteins, Cyps and FKBPs to show PT interactions with Hsp90, Hsp70, FKBP and Cyps are required for PT-intoxication of CHO-K1 cells. Further, we determined inhibition of these host molecules prevents ADP-ribosylation of Giα protein and intracellular cAMP accumulation of CHO-K1 by preventing uptake of PT into the cytosol, not by inhibiting PT enzymatic activity. Immobilization of the studied host chaperones and folding helper proteins on a nitrocellulose membrane and incubation with PTS1 identified Hsp90, Hsp70 and Hsc70, CypA, Cyp40, FKBP51 and FKBP52 as binding partners of PTS1 in a specific and concentration-dependent manner. Interestingly, Hsp90 requirement for translocation of PTS1 from the ER into the cytosol was shown in a previous study. In this study, PTS1 was expressed directly in the ER of transfected CHO cells and it was demonstrated that Hsp90-inhibition in these cells prevented translocation of PTS1 from the ER to the cytosol 56 . The effect of CsA, FK506 and VER on translocation of PTS1 from the ER to the cytosol was not investigated in such an isolated manner and therefore we cannot fully exclude an inhibiting effect of CsA, FK506 or VER on endocytosis or retrograde transport to the ER. Nevertheless, assistance of Cyps, FKBPs and Hsp70 during translocation into the cytosol as demonstrated for Hsp90 is conceivable.
Cyp40, FKBP51 and FKBP52 contain three TPR domains through which they bind to Hsp90 32,57 forming a Hsp90-multichaperone complex 30,31 . The Hsp90-multichaperone complex facilitates the stabilization, folding and activation of more than 200 mammalian proteins including steroid hormone receptors 30,58-60 . For folding and activation of steroid hormone receptors, activity of chaperones is required in a concerted and sequenced manner 30 . First, Hsp70 binds the unfolded protein and transfers it to an Hsp90 dimer. Subsequent binding of FKBPs or Cyps through TPR domains influences Hsp90 activity, in particular its ability to bind and hydrolyze ATP. ATP hydrolysis leads to conformational changes in Hsp90, which are transferred onto the unfolded protein resulting in a native folded protein. Since the same composition of cellular chaperones is required for the cytosolic delivery of ADP-RTs including PT it is conceivable that also in this case the chaperones act in a concerted manner.
Recently, it has been shown that cholera toxin requires Hsc/Hsp70 but not Cyps in addition to Hsp90 for its cellular uptake 61 . However, interaction of Hsc/Hsp70 and Hsp90 with CTA1, the enzyme domain of cholera toxin, occurs independently since both chaperones can bind simultaneously to CTA1, and binding occurs in the absence of the adapter protein Hop. This suggests an alternative interaction mechanism between cholera toxin and Hsp90/Hsc/Hsp70 that differs from the described Hsp90-multichaperone machinery.
We showed that interaction between Cyp40 and PTS1 was detected after 48 h after PT incubation. This suggests that Cyp40 not only facilitates the uptake of PTS1 into the cytosol but is required to stabilize PTS1 in an active conformation. A comparable mechanism is described for the cholera toxin, which also harbors ADPribosyltransferase activity and belongs to the group of long trip toxins. Hsp90 facilitates translocation of CTA1 to the cytosol, refolds CTA1 into an active conformation in an ATP-dependent manner and continues to bind to refolded CTA1 62,63 . The finding that CTA1 and PTS1 require the assistance of chaperones even after translocation and refolding might be due to the fact that both enzyme subunits are thermally unstable. This means that unfolding of CTA1 and PTS1 occurs at 37 °C after dissociation from the B-subunit. Therefore, releasing of toxin subunits from chaperones after translocation/refolding could again result in unfolding. Recently, two Hsp90 binding motifs in CTA1, which both are necessary for efficient translocation of CTA1 into the cytosol were identified 56 . One of these binding motifs was also identified in other ER-translocating ADP-ribosylating toxins including PT and Hsp90 requirement for PT uptake into the cytosol was demonstrated confirming and www.nature.com/scientificreports/ supporting the findings in the present study. Interestingly, the identified Hsp90-binding motif was absent in ADP-ribosylating toxins that translocate from endosomes suggesting an interaction mechanism distinct to some extent between short trip and long trip ADP-ribosylating toxins. B. pertussis colonizes its host through attachment to the ciliated cells of the airways. It is known that PT can bind to various sialic acid-containing glycoproteins and various receptors have been found 11,13 . However, so far it is not clear which cell types are the target for PT in humans. Human bronchial airway epithelial basal cells differentiated in air-liquid interface conditions recapitulate the characteristics of the airway epithelial barrier. These epithelial cells differentiate into pseudostratified, polarized cells including ciliated and goblet (secretory) cells. From this, we determined that CC10 or MUC5B positive secretory cells, and not β-IV-tubulin positive ciliated cells were the target of PTS1 intoxication. It was determined that pharmacological inhibition of Cyps, but not Hsp90 or FKBP reduced detectable cytosolic PTS1 in these cells. Moreover, PT had no effect on the trans-epithelial resistance but increases the ASL of the differentiated human airway epithelium. These data suggest that PT does not impact gross epithelial barrier integrity but rather affects vectorial ion or fluid transport. The ASL shields the mucosal surface of lung epithelia and forms the first line of defense against airborne noxae and pathogens. A major task of lung epithelial cells is therefore to tightly regulate the ASL volume to maintain appropriate lung function and to ensure mucociliary clearance. ASL volume homeostasis is regulated by balancing water secretion and resorption. Secretion is driven by apical chloride channels (e.g. CFTR), whereas the epithelial Na + channel (ENaC) mediates resorption. Hence, our observation that PT disturbs ASL homeostasis might reflect an impact on vectorial ion transport, either activating apical chloride channels or inhibiting ENaC. It has been demonstrated that PT intoxication results in increased intracellular cAMP levels, and CFTR is activated by intracellular cAMP 64 . Yet, future studies, including Ussing chamber measurements, will be required to decipher the impact of PT intoxication on transepithelial ion/water transport in more detail.
Results from another previous study in mice showed an upregulation of the epithelial anion exchanger pendrin in the lung after infection with B. pertussis 65 . Pendrin knockout mice showed reduced lung inflammatory pathology but even higher bacterial loads during infection suggesting a role of pendrin in the pathology. Pendrin exports bicarbonate to the ASL and thereby increases pH, which possibly contributes to promoting the inflammatory pathology. These results indicate that manipulation of ASL parameters like volume or pH might be mediated by PT and play a role for the pathology and course of disease.
The mouse model of pertussis disease recapitulates the age-dependent nature of severe pertussis illness. Infant mice succumb to lethal infection at doses tolerated by adult mice. Leukocytosis, the rapid increase in vascular leukocytes, correlates with disease severity and lethality and is PT-mediated 10,66 . Infant mice, like humans, show robust increases in circulating leukocytes following infection. Here, we showed that inhibition of Cyp activity by CsA, a licensed drug with well-known pharmacokinetics and safety profiles 67,68 , and its non-immunosuppressive derivative NIM811 42 led to significantly decreased leukocytosis upon infection with PT-producing B. pertussis strains. To our knowledge, this is the first time that Cyps were pharmacologically targeted to successfully prevent cytotoxic effects of a bacterial AB-type toxin in an in vivo model. Together with our detailed mechanistic studies in CHO-K1 cells and in vitro differentiated human airway epithelial model, these results provide a promising basis for the development of novel therapeutic strategies to prevent severe symptoms like leukocytosis caused by PT during B. pertussis infection.

Protein-protein interaction studies via dot blot system. Purified recombinant chaperones, PPIases
and PPIase fragments were vacuum aspirated onto a nitrocellulose membrane as serial dilution starting with 1 µg/ml using the dot blot system (Bio-Rad, Feldkirchen, Germany). Recombinant C3 toxin from C. botulinum (C3bot) was chosen as a random protein to exclude unspecific protein-protein interactions. Successful transfer was confirmed by Ponceau S staining. Membrane was blocked with 5% skim milk powder in PBS-Tween (PBST) and then cut and probed with PTS1 (Aviva Systems, San Diego, California, USA) or for control with PBST. After extensive washing, both membranes were incubated with anti-PTS1 antibody (Santa Cruz, Heidelberg, Germany) to detect bound PTS1 using HRP-coupled secondary antibody (Santa Cruz) in combination with ECL (enhanced chemiluminescence, Millipore, Merck, Darmstadt, Germany) system. Signals for both membranes were detected on the same X-ray film. Pictures were cropped for display purposes only.

Cell culture and intoxication experiments.
Cell culture materials were obtained from Gibco unless indicated otherwise. Chinese hamster ovary cells strain K1 (CHO-K1, from DSMZ, Braunschweig, Germany) were cultivated in DMEM and HAM's F12 containing 5% heat-inactivated fetal calf serum (Invitrogen, Thermo Fisher Scientific, Waltham, Massachusetts, USA), 1 mM sodium-pyruvate and Penicillin-Streptomycin (Pen-Strep) (1:100) (Thermo Fisher Scientific). Cells were grown at 37 °C and 5% CO 2 as described before 29 . Cells were trypsinized and reseeded every two to three days for at most 15-20 times. For intoxication experiments 29,46,48 , cells were seeded in culture dishes and the specific pharmacological inhibitors radicicol (inhibitor of ATPbinding site of Hsp90, Sigma-Aldrich, Merck, Darmstadt, Germany), cyclosporine A (inhibitor of Cyp activity, Sigma-Aldrich, Merck), FK506 (inhibitor of FKBP activity, Sigma-Aldrich, Merck), VER (inhibitor of ATPbinding site of Hsp70, Hsc70 and Grp78, Tocris Bioscience, Wiesbaden-Nordenstadt, Germany), HA9 (inhibitor of the substrate binding domain of Hsp70) 37  www.nature.com/scientificreports/ further incubation for 18 h at 37 °C or PT was added for 18 h. Morphology of cells was documented by using a Zeiss Axiovert 40CFL microscope with a Jenoptik ProgRes C10 CCD camera. At least 3 images per treatment were taken and analyzed and the experiment was repeated independently at least 3 times. To quantify toxininduced effects on CHO-K1 cells total number of cells per picture was manually counted using ImageJ (National Institutes of Health, Bethesda). Untreated control cells were set as 100% and values of other samples were calculated accordingly. Materials for cell culture experiments were obtained from TPP Techno Plastic Products (Trasadingen, Schweiz). iGIST bioassay. The iGIST bioassay was performed as previously described 38  Toxin binding assay. CHO-K1 cells were pre-incubated with 10 µM Rad, CsA, FK506 or 30 µM VER. The assay was performed as previously described 29,46 . Cells were cooled down to 4 °C for 15 min to prevent endocytosis. 500 ng/ml PT were added, and cells were incubated for 30 min at 4 °C. Subsequently, culture medium was removed, and cells were washed with PBS to remove unbound PT. Then cells were scraped with boiling SDSsample buffer and heated for 10 min at 95 °C. SDS-PAGE and Western Blotting was performed. To detect cellbound PT an antibody against the S1 subunit of PT and a peroxidase-coupled secondary antibody were used.
Protein interaction analysis in cultured cells by Duolink using proximity ligation assay (PLA). CHO-K1 cells were incubated with PT for indicated periods of time. Cells were fixed with 4% PFA, permeabilized and blocked with 10% NGS and 1% BSA in PBST. Subsequently, cells were incubated with mouse anti-PTS1 (Santa Cruz or Abcam) and rabbit anti-Cyp40 (Thermo Fisher Scientific), rabbit anti-Hsp90 (Thermo Fisher Scientific), rabbit anti-Hsp70 (Enzo Life Sciences) or rabbit anti-FKBP51 (Santa Cruz) antibodies for 1 h at 37 °C. PLA assay was performed according to the manufacturer's protocol (Duolink using PLA technology, Sigma-Aldrich, Merck) and as described before 46,47 . Measurement of the transepithelial electrical resistance (TEER). TEER was analyzed by impedance spectrometry using the cellZscope (NanoAnalytics, Münster, Germany). For measurements, the basal electrode was overlaid by 500 µl equilibrated DMEM-H medium filters inserted and 250 µl ediuDMEM-H medium was added to the apical side of the filter. Measurements were performed immediately after positioning of apical electrodes. Data were acquired and analyzed using the software provided with the instrument (NanoAnalytics).
Measurement of apical surface liquid (ASL) volume. ASL measurements were performed using the D 2 O dilution method as described previously 72 . Filters with confluent cell layers were placed in 500 µl of differentiation medium containing respective inhibitors and pre-incubated for 30 min at 37 °C. 25 µl of isotonic NaCl, containing respective inhibitors, was added to the apical compartment. Silicon sealed control filters loaded with 25 µl isotonic NaCl solution were randomly distributed throughout the plate to estimate volume changes caused by evaporation. After 72 h apical fluid volumes were collected with 25 μl D 2 O containing 0.9% (w/v) NaCl for analysis. Water concentrations were measured by attenuated total reflexion mid-infrared spectroscopy on A Vertex 70 FT-IR spectrometer, equipped with a BioATRCell-II unit and a liquid nitrogen cooled MCT detector (Bruker Optics, Fällanden, Switzerland), to calculate ASL volumes. Data acquisition and processing was performed using OPUS 6.5 (Bruker Optics). Before each measurement a background spectrum of the empty ATR-unit was collected and a calibration series (0%, 15%, 25%, 40%, 50%, 65% (v/v) of isotone NaCl solution in H 2 O and 0.9% NaCl solution in D 2 O) was measured. Areas below absorption bands were blotted against water concentration and linear regression was calculated through data points to obtain the slope (m) and y-interception (y 0 ), to calculate water concentrations as described in detail before 72 . The apical volume (V api ) was calculated according to: with V D 2 O being the volume (25 μl) of D 2 O in 0.9% NaCl. Changes in apical volume (�V api ) were calculated by subtracting the remaining volume from the initial volume (25 µl isotonic NaCl solution). Changes in apical volumes in experimental filters were corrected for changes in evaporation controls.
Infant mouse studies. 7-day old C57BL/6 mice were used in accordance with the University of Maryland, Baltimore Institutional Animal Care and Use Committee. A streptomycin resistant derivative of the Tohama I "wild-type" B. pertussis strain was grown on Bordet-Gengou (BG) agar plates supplemented with defibrinated sheep blood (10%, Lampire Biological Products, Pipersville, Pennsylvania, USA) and 200 μg/ml streptomycin (Sigma-Aldrich, Merck). All comparisons were made between littermates pre-designated as receiving vehicle or one of the inhibitors. When dividing mice into groups attempts were made to match bodyweights between groups instead of true randomization. Tails were marked with colored pens daily to distinguish experimental groups within a litter. Only litters between sized between 5 and 8 pups were used to minimize variability associated with litter size. Animals were challenged via aerosol administered by a nebulizer system (Pari Vios) for twenty minutes. Lungs were excised post-euthanasia, homogenized mechanically (Omni International, Tulsa, Oklahoma, USA) and bacterial burden determined by plating on BG agar. Animals were intranasally administered vehicle (Cremophor:Ethanol:Saline 10:1:89), CsA (25 mg/kg, Sigma-Aldrich, Merck) or NIM811 (25 mg/ www.nature.com/scientificreports/ kg, MedChemExpress, Monmouth, New Jersey, USA). To measure leukocytosis blood was harvested 7 days post-infection by cardiac puncture into ethylenediaminetetraacetic acid (EDTA) containing 1.7 ml tubes (Fisher Scientific) and treated with ammonium-chloride-potassium (ACK) lysis buffer to lyse red blood cells. The investigator performing the harvest then assigned the samples numbers unrelated to their group. White blood cells were then counted using a hemocytometer by an investigator blind to sample identifications. Inhibitor treated mice were compared to vehicle treated animals by ANOVA with Dunnett's multiple comparisons test using GraphPad Prism. Studies were performed in both male and female mice, but exact numbers of each gender could not be determined at time of infection (7-days-old).
Ethics statement. Animals were used in accordance with the University of Maryland Institutional Animal Care and Use Committee protocol 0417005 (University of Maryland, Baltimore, MD) following approval by the Office of Animal Welfare Assurance. Animals were euthanized by CO 2 asphyxiation followed by cervical dislocation and thoracotomy. All methods were performed in accordance with the relevant guidelines and regulations.
The study was carried out in compliance with the ARRIVE guidelines.