Transcriptional Downregulation of a Type III Secretion System under Reducing Conditions in Bordetella pertussis

Bordetella pertussis uses a type III secretion system (T3SS) to inject virulence proteins into host cells. Although the B. pertussis T3SS was presumed to be involved in host colonization, the efficient secretion of type III secreted proteins from B. pertussis has not been observed. To investigate the roles of type III secreted proteins during infection, we attempted to optimize culture conditions for the production and secretion of a type III secreted protein, BteA, in B. pertussis. We observed that B. pertussis efficiently secretes BteA in ascorbic acid-depleted (AsA−) medium. When L2 cells, a rat lung epithelial cell line, were infected with B. pertussis cultured in the AsA− medium, BteA-dependent cytotoxicity was observed. We also performed an immunofluorescence assay of L2 cells infected with B. pertussis. The clear fluorescence signals of Bsp22, a needle structure of T3SS, were detected on the bacterial surface of B. pertussis cultured in the AsA− medium. Since ascorbic acid is known as a reducing agent, we cultured B. pertussis in liquid medium containing other reducing agents such as 2-mercaptoethanol and dithioerythritol. Under these reducing conditions, the production of type III secreted proteins was repressed. These results suggest that in B. pertussis, the production and secretion of type III secreted proteins are downregulated under reducing conditions. IMPORTANCE The type III secretion system (T3SS) of Bordetella pertussis forms a needle-like structure that protrudes from the bacterial cell surface. B. pertussis uses T3SS to translocate virulence proteins called effectors into host cells. The culture conditions for effector production in B. pertussis have not been investigated. We attempted to optimize culture medium compositions for producing and secreting type III secreted proteins. We found that B. pertussis secretes type III secreted proteins in reducing agent-deprived liquid medium, and that BteA-secreting B. pertussis provokes cytotoxicity against cultured mammalian cells. These results suggest that redox signaling is involved in the regulation of B. pertussis T3SS.


Introduction 51
The genus Bordetella is made up of Gram-negative bacteria and has been 52 subdivided into 16 subspecies to date (1). Bordetella pertussis is a causative agent 53 of whooping cough, also known as pertussis. B. pertussis strictly adapts to humans, 54 and it lacks an environmental reservoir (2). A variety of virulence factors have been 55 identified in Bordetella, e.g., filamentous hemagglutinin (3), adenyl cyclase toxin (4) 56 and pertactin (5). These virulence factors are regulated by a Bordetella two-57 component regulatory system, BvgAS. In addition to these virulence factors, the type 58 8 Anti-BteA, anti-BopB, and anti-Bsp22 antibodies were purified from rabbit serum in 163 our previous study (10,18,29). To detect filamentous hemagglutinin (FHA) signals, 164 we used mouse anti-FHA serum (30). Mouse anti-CyaA monoclonal antibody was 165 purchased from Santa Cruz Biotechnology (Santa Cruz, CA). To prepare the anti-166 BtrS antibody, the peptide corresponding to the C-terminus region of BtrS 167 (CALREALRERGYDSVP) were conjugated with hemocyanin from keyhole limpets 168 (Sigma) by using 3-maleimidobenzoic acid N-hydroxysuccinimide ester (Sigma). 169 These cross-linked peptides were used to immunize rabbits, and resulting anti-sera 170 were incubated with peptide immobilized on epoxy-activated sepharose 6B 171 (Amersham) to obtain specific Ig-fractions. 172 173

Quantitative reverse transcription-PCR (qRT-PCR) 174
The amounts of mRNA were measured by qRT-PCR. Bacterial total RNA was 175 prepared using a Trizol Max Bacterial RNA Isolation Kit (Invitrogen), an RNeasy Mini 176 Kit (Qiagen, Hilden, Germany) and an RNase-free DNase-free Kit (Qiagen). The 177 RNA sample was reverse transcribed with Transcriptor Universal cDNA Master 178 (Roche Diagnostics, Indianapolis, IN) and a T100 Thermal Cycler (Bio-Rad, Hercules, 179 CA). The resulting cDNA was amplified by FastStart Essential DNA Probes Master 180 (Roche) using the following primer pairs: 5-recA and 3-recA for recA; 5-bteA and 3-181 bteA for bteA; 5-bopB and 3-bopB for bopB (Table 3). The results were calculated as 182 described in the Roche manual. The amount of recA mRNA was used as an internal 183 control. The mRNA amounts are presented herein as relative to the amounts in the 184 Tohama I cultured in SS medium, which was defined as 1. 185

LDH Assays 187
To examine the release of lactate dehydrogenase (LDH) from B. pertussis-infected 188 cells, 5.0×10 4 cells/well of L2 cells seeded in 24-well plates were infected with 189 bacteria at the multiplicity of infection (MOI) of 125, 250, or 500. The plates were 190 centrifuged at 900 g for 5 min and incubated for 3 hr at 37ºC under a 5% CO2 191 atmosphere. The amounts of LDH were measured spectrophotometrically using a 192 Cyto-Tox 96 Non-radioactive Cytotoxicity Assay Kit (Promega, Madison, WI). We 193 added 50 µL of 10 %Triton X-100 to the 950 µL of extracellular medium of L2 cells in 194 24-well plate. We then mixed the extracellular medium by pipetting. We subtracted 195 the LDH value obtained from extracellular medium of uninfected cells from the value 196 obtained from the Triton X-100-treated cells and used the resulting value as 100%. 197

Immunofluorescent staining 199
For the immunofluorescent staining assay, 2.5×10 5 cells/well of L2 cells were 200 seeded on coverslips in six-well plates and incubated overnight, and then infected 201 with B. pertussis at an MOI of 125. The plates were centrifuged at 900 g for 5 min 202 and incubated for 3 hr at 37ºC under a 5% CO2 atmosphere. The infected L2 cells 203 were then immunostained as described previously (12). Briefly, the cells were 204 treated with 4% paraformaldehyde, 50 mM NH4Cl, 0.2% Triton-X, and 4% bovine 205 serum albumin (BSA). After blocking, the cells were stained by anti-Bsp22 antibody 206 (29). As a secondary antibody, Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen) was 207 used. F-actin was stained with Rhodamine Phalloidin (Molecular Probes, Eugene, 208 OR). Bsp22 signals per one cell were counted under a fluorescence microscope. 209 210

Statistical analyses 211
The statistical analyses were performed using the nonparametric unpaired t-test with 212 a one-tailed p-value with Prism ver. 5.0f software (Graphpad, La Jolla, CA Cyclodextrin solid medium (CSM) is reported as a synthetic medium for clinical 219 isolation of B. pertussis (26). We sought to determine whether B. pertussis grown on 220 this agar secretes type III effectors, and we prepared a liquid medium, designated 221 "low casamino acids (LCA)" medium, by removing the agar from CSM contents 222 (Table 2). B. pertussis was cultured in SS or LCA medium, and then the whole cell 223 lysates (WCL) and supernatant fractions (Sup) samples were subjected to western 224 blotting with antibodies against BteA (an effector, a type III secreted protein) and 225 BopB (a translocator, a type III secreted protein). 226 BteA and BopB signals were detected in WCL and Sup of the wild-type 227 Tohama I cultured in LCA medium (Fig. 1). As reported previously, BteA forms SDS-228 resistant multimers, and in our present study we also detected the signals at around Although the ingredients of the LCA and SS media are the same, the amounts of 238 ascorbic acid and casamino acids in LCA medium are lower than those in SS 239 medium (Table 2). To determine which substance is involved in the production and 240 secretion of type III secreted protein, we cultured B. pertussis in SS medium, 241 ascorbic acid-deprived SS medium (SS_AsA − ), casamino acid-deprived SS medium 242 (SS_CaA − ), or both ascorbic acid-and casamino acid-deprived medium 243 (SS_AsA − _CaA − ). WCL and Sup samples were subjected to western blotting with 244 antibodies against BteA and BopB (Fig. 2). 245 The BteA multimer signal intensities in the WCL samples of the SS_AsA − , SS_CaA − 246 and SS_AsA − _CaA − media were stronger than that in the SS medium (Fig. 2). A faint 247 BteA signal intensity in the Sup samples of SS was detected (Fig. 2). The BteA 248 signal intensities in the Sup samples were evident in the SS_AsA − and 249 SS_AsA − _CaA − media (Fig. 2). Again, BopB signal intensities in both WCL and Sup 250 samples were evident in the SS_AsA − and SS_AsA − _CaA − media (Fig. 2). 251 We also investigated the potential involvement of ascorbic acid or casamino 252 acids in Bvg-regulated virulence factors, e.g., FHA and CyaA. WCL and Sup 253 samples were subjected to western blotting with antibodies against FHA and CyaA 254 (Fig. 2). The results revealed that there were no significant differences in the FHA or 255 CyaA signal intensities in the WCL and Sup samples among the media used here 256 (Fig. 2). Collectively, these results suggest that ascorbic acid has a specific influence 257 on the production and secretion of BteA and BopB in B. pertussis.

12
Standard SS medium contains ascorbic acid at a final concentration of 2270 µM. To 259 further explore whether ascorbic acid affects the production and secretion of type III 260 secreted proteins, we cultured B. pertussis in SS medium in the presence of ascorbic 261 acid at the final concentrations of 91, 454, or 2270 µM. WCL and Sup samples were 262 prepared from each bacterial culture and analyzed by western blotting (Fig. 3). The 263 BteA signal was detected in both WCL and Sup samples of SS medium at 0-454 µM 264 ascorbic acid, but the BteA signal of the SS medium at 2270 µM ascorbic acid was 265 faint or absent (Fig. 3). The BopB signals showed a similar pattern (Fig. 3). These

B. pertussis induces cytotoxicity against cultured mammalian cells 299
A B. pertussis BspR-deficient strain was reported to induce cytotoxicity against 300 cultured mammalian cells (15). In the present study, we identified better culture 301 conditions for the production and secretion of BteA in B. pertussis than those 302 provided with standard SS medium. We thus investigated whether wild-type 303 B. pertussis provokes cytotoxicity against cultured mammalian cells under these 304 conditions. L2 cells, a rat lung epithelial cell line, were infected at an MOI of 500 for 3 305 hr in several modified SS media shown in Fig. 6. The amounts of ascorbic acid 306 and/or casamino acids was reduced to one-twentieth of and/or to one-half of those in 307 SS medium. We also used the SS media without ascorbic acid and/or casamino 308

acids. 309
The amounts of lactate dehydrogenase (LDH) released into the extracellular 310 medium were then measured. 311 The results revealed that released LDH was detected from L2 cells infected with 312 Tohama I in the SS_1/20AsA or SS_1/20AsA_1/2CaA medium (Fig. 6). The 313 SS_1/20AsA medium and SS_1/20AsA_1/2CaA medium showed no cytotoxicity. On 314 the other hand, the L2 cells incubated in the SS_AsA − or SS_1/20AsA_CaA − 315 medium were injured by the medium itself. Therefore, these media were not useful 316 for this assay. Taken together, these results demonstrate that we successfully 317 The results showed that the amounts of Bsp22 signals (Fig. 7A, green) on L2 cells 328 infected with Tohama I and BP350 cultured in SS_1/20AsA_1/2CaA medium were 329 greater than those in SS medium (Fig. 7). Thus, we demonstrated that B. pertussis 330

T3SS is activated during the infection of cultured mammalian cells under appropriate 331
conditions. 332 As a result, LDH was not detected in the medium of L2 cells infected with MGKU2 343 (Fig. 8A,B). The amount of LDH detected in the medium of L2 cells infected with 344 MGKU3 was significantly lower than that in the medium of L2 cells infected with 345 MGKU1 (Fig. 8A,B). The amount of LDH detected in the medium of L2 cells infected 346 with MGKU4 was significantly greater than that in the medium of L2 cells infected 347 with MGKU1 at an MOI of 125 (Fig. 8A), but not at an MOI of 250 (Fig. 8B). 348 WCL and Sup samples were prepared from each strain and separated by SDS-349 PAGE, then analyzed by western blotting with anti-BteA antibody (Fig. 8C). BteA 350 signals of the WCL samples were detected in MGKU1, MGKU2, and MGKU4, but 351 not in MGKU3 (Fig. 8C). BteA signals of the Sup samples were detected in MGKU1 352 and MGKU4, but not in MGKU2 or MGKU3 (Fig. 8C). These results suggest that the 16 cytotoxicity induced by wild-type B. pertussis (Fig. 6)  anti-FHA antibodies (Fig. 9). The BteA and BopB signal intensities of both WCL and 365 Sup samples were decreased in the presence of each of the reducing agents (Fig. 9). In this study, B. pertussis increased the production and secretion of type III secreted 375 proteins, BteA and BopB, when cultured in ascorbic acid-deprived SS medium 376 (SS_AsA − ). We successfully produced cytotoxicity against mammalian cells by the 377 infection of wild-type Tohama I cultured in SS 1/20AsA_1/2CaA medium. 378 B. pertussis decreased the production and secretion of type III secreted proteins 379 when cultured in SS_AsA − medium containing reducing agents such as 2-ME or DTT. 380 These results suggest that redox signaling is involved in regulation of the T3SS in 381

B. pertussis. 382
It has been reported that no secretion of the type III secreted proteins BteA 383 and Bsp22 was detected from Tohama I (31), and that the secretion of the type III 384 secreted proteins BteA and BopD was detected from a non-vaccine-type strain, 385 BP157. However, BP157 showed no cytotoxicity against L2 cells, J774 mouse 386 macrophage-like cells, and HeLa cells (32). In another report, glutamate limitation 387 upregulates production of the type III secreted proteins (33). However, the 388 requirement of a type III secretion system (T3SS) for virulence in B. pertussis thus 389 remains unclear. On the other hand, a BscN-deficient mutant in which the T3SS was 390 inactive colonized mouse lungs to a significantly lower degree compared to the wild-391 type (31). In addition, cytotoxicity against HeLa cells was observed by infection with 392 a BspR (a negative regulator for type III secreted proteins)-deficient B. pertussis 393 strain (15). These results suggest that the T3SS has a significant role in the 394 virulence of B. pertussis. 395 The production and the relative amounts of bopB mRNA of Tohama I cultured 396 in SS AsA − medium were significantly increased compared to those of Tohama 397 cultured in SS medium, but this was not the case for BteA mRNA (Figs. 2, 5). It was 398 reported that bteA and bopB genes are upregulated by BtrS (a sigma factor for type 399 III secreted proteins) and downregulated by BspR (13, 15). It has also been reported 400 that the amount of bteA mRNA was increased 2-fold in a BspR-deficient strain 401 compared to the wild-type, whereas the amount of bopB mRNA was increased by 402 18-fold compared to the wild-type by RNA-seq (15). The bopB gene is localized in 403 the T3SS apparatus locus (bsc locus) adjacent to the btrS gene. Since the bteA 404 gene is separated from the bsc locus by >2.5 Mb (9), BtrS would preferentially 405 regulate transcription of the bopB gene over that of the bteA gene. 406 The cytotoxicity against L2 cells was observed by the infection of Tohama I 407 cultured in SS_1/20AsA _1/2CaA medium (Fig. 6). Before infection, we exchanged 408 Ham's F-12K medium for the SS_1/20AsA _1/2CaA medium. When L2 cells were 409 infected with Tohama I in Ham's F-12K medium, the cytotoxicity against L2 cells was 410 not observed (Fig. S1). In order to establish oxidizing conditions in the tissue culture 411 media, we used several oxidizing agents, i.e., hydrogen peroxide (H2O2), potassium 412 ferricyanide(III) (K3[Fe(CN)6]), methyl 3-nitro-2-pyridinesulfenate (Npys-OMe), and 413 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB). We also used another tissue culture 414 medium, i.e., Minimum Essential Medium Eagle (MEM), because MEM contains no 415 reducing agents, while F-12K, which is the medium for the L2 cells culture, contains 416 a reducing agent, i.e., cysteine. Even when L2 cells were infected in F-12K medium 417 containing the oxidizing agent or MEM, B. pertussis showed no cytotoxicity (Fig. S1). 418 We therefore speculate that a substance which inhibits the secretion of type III 419 secreted proteins from Tohama I was present in the Ham's F-12K medium, or a 420 substance that is necessary for the secretion of type III secreted proteins from 421 Tohama I was absent in the Ham's F-12K medium. We found that infection of the 422 human alveolar epithelium cell line A549 with wild-type B. pertussis cultured in 423 SS_1/20AsA _1/2CaA medium did not result in cytotoxicity (data not shown). For the 424 observation of cytotoxicity against A549 cells, it is necessary to optimize the 425 conditions for the production and secretion of type III secreted proteins in 426

B. pertussis. 427
A previous study showed that cytotoxicity against the cell line J774A.1 was 428 achieved by infection with wild-type B. pertussis 18323, but not by infection with a 429 CyaA-deficient strain (34). This finding suggested that CyaA has the potential to 430 provoke the cytotoxicity. However, we confirmed herein that the production and 431 secretion of CyaA in a BteA-deficient strain (MGKU3) were not significantly different 432 from those of the wild-type (data not shown). In our results, the amounts of LDH 433 released from L2 cells infected with MGKU3 were greater than those of the T3SS-434 inactive strain (MGKU2), and no LDH release was detected from L2 cells infected 435 with MGKU2 (Fig. 8). These results raised the possibility that an unknown type III 436 effector other than BteA is involved in the cytotoxicity of B. pertussis. 437 Although we observed no significant difference in the growth rate of Tohama I 438 between SS medium and SS_AsA − medium (data not shown), the production and 439 secretion of type III secreted proteins of Tohama I cultured in the SS_AsA − medium 440 were increased compared to those in the SS medium (Fig. 2). It had not been 441 determined whether the expression of type III secreted proteins is regulated by 442 reducing agents at the molecular level, and in this study, the production and 443 secretion of CyaA and FHA in Tohama I cultured in SS medium were not significantly 444 different from those in SS_AsA − medium (Fig. 2). It was reported that there is no 445 The funders had no role in the study design, data collection or analysis, the decision 620 to publish, or the preparation of the manuscript.      Restriction enzyme recognition sites were underlined.