Ethanol decreases Pseudomonas aeruginosa flagellar motility through a cyclic-di-GMP- and stator-dependent pathway

Pseudomonas aeruginosa frequently encounters microbes that produce bioactive metabolites including ethanol. At concentrations that do not affect growth, we found that ethanol reduces P. aeruginosa motility by 30% in a swim agar assay and this decrease is accompanied by a 2.5-fold increase in levels of cyclic diguanylate (c-di-GMP), a second messenger that represses motility, in planktonic cells. A screen of mutants lacking genes involved in c-di-GMP metabolism identified SadC and GcbA as diguanylate cyclases involved in swim zone reduction by ethanol and ethanol-induced c-di-GMP production. The reduction of swimming in response to ethanol also required the stator set, MotAB, two PilZ-domain proteins (FlgZ and PilZ), PilY1-a proposed surface-sensing protein, and PilMNOP, which comprises the pilus alignment complex and these proteins have been previously implicated in the control of motility on agar surfaces. Microscopic analysis of the fraction of quiescent cells in swim medium showed that ethanol decreased the portion of motile cells in the wild type, but had opposite effects in the ∆pilY1, ∆pilMNOP, ∆motAB, and ∆pilZ∆flgZ mutants. Together, these data indicate ethanol induces a regulated change in motility in planktonic cells at concentrations similar to those produced by other microbes. We propose that this ethanol-responsiveness may contribute to the co-localization of P. aeruginosa with ethanol-producing microbes.


Materials and Methods 111
Strains and Media. Strains and plasmids used in this study are listed in Table S3. P. 112 aeruginosa PA14 and E.coli strains were routinely cultured on lysogeny broth (LB) solidified with 113 1.5% agar, or in LB broth at 37 o C with shaking. Gentamicin (Gm) was used at 60 µg/ml and 114 carbenicillin (Cb) at 700 µg/ml for P. aeruginosa. Gm was used at 10 µg/mL for E. coli. For P. aeruginosa strain to generate in-frame deletion mutants using allelic replacement as previously 133 described (39). Exconjugants were selected on solid LB using gentamycin and nalidixic acid 134 8 followed by counterselection on 5% sucrose. PCR amplification and DNA sequencing, using 135 primers that flanked the site of deletion, were used to confirm all resulting mutants. 136 For arabinose-inducible complementation, the gene being complemented was 137 expressed on either pMQ80 (60 µg/ml gentamycin) or pDPM73 (700 µg/ml carbenicillin) plasmid 138 backbones. Confirmed constructs were electroporated into the indicated P. aeruginosa strains, 139 selecting for the appropriate antibiotic resistance marker. Arabinose (0.02 or 0.05%) was added 140 to the medium and complementation was confirmed via the indicated phenotypic assay. 141 142 Swimming motility assays. Swim assays were performed as previously described (18). Briefly, 143 M63 medium without and with 1% (v/v) ethanol and solidified with 0.3% agar (swim agar) was 144 poured into petri plates and allowed to dry at room temperature (~25 o C) for ~4 h prior to 145 inoculation. Sterile tooth picks were used to inoculate bacteria into the center of the agar without 146 touching the bottom of the plate; liquid cultures grown for 8-16 h were used as inoculum. No 147 more than four strains were assayed per plate. Plates were incubated upright at 37 o C in stacks 148 of no more than four plates per stack for 16 h; the swim zone diameter was then measured. P. 149 aeruginosa wild type was included in each experiment so that mutant phenotypes could be 150 assessed despite slight day-to-day variation in swim zone diameter. Each strain was inoculated 151 in four replicates and replicate values were averaged to obtain a final swim zone diameter for 152 each strain. All strains were assessed on at least three separate days. 153 154 Twitching motility assays. Twitching motility assays were performed with T-agar medium (10g 155 tryptone, 5g NaCl, and 15g agar in 1L) without and with 1% ethanol in petri plates that were 156 allowed to dry at room temperature for 24 h prior to inoculation. Sterile toothpicks were used to 157 inoculate into the agar until the toothpick touched the bottom of the petri plate; liquid cultures 158 grown for 16 h were used as inoculum. No more than four strains were analyzed per plate and 9 six replicate plates were included in each experiment. Plates were incubated in inverted stacks 160 of four at 37 o C for 40 h. To visualize the twitch zone, a spatula was used to gently ease the agar 161 out of the petri plates and two mL of 0.1% (w/v) crystal violet in water was added to each plate 162 and allowed to stand for 10 min. The crystal violet was removed and the plates rinsed with water 163 and allowed to air dry. Twitch zone diameter was measure and recorded. All strains were 164 assessed on at least three separate days. 165 166 Swarming motility assays. Swarm assays were performed as previously described (18). 167 Briefly, M8 medium, without and with 1% ethanol, and with 0.5% agar (swarm agar) was poured 168 into 60 x 15 mm plates and allowed to dry at room temperature for ~4 h prior to inoculation. 169 Each plate was inoculated with 0.5 µL of a liquid culture that was grown for 8-16 h, and the 170 plates incubated face-up at 37 o C in stacks of no more than four for 16 h. Each strain was 171 inoculated in four replicates and was assessed on at least three separate days. Images were 172 captured using a Canon EOS Rebel T6i camera and images measured for ethanol-dependent 173 swarm repression. 174 175 Reversal rate measurements. To measure the frequency at which a motile cell changes its 176 direction, we used a modified version of a method that was previously described (25, 40). 177 Briefly, overnight liquid cultures were subcultured 1:100 in five mL M63 medium and incubated 178 at 37 o C for 2 h. Once cultures reached exponential phase, they were then diluted 1:1000 in 179 fresh M63 medium and Ficoll was added to a final concentration of 3% to obtain higher viscosity 180 conditions that slowed the swimming cells sufficiently to allow the monitoring of reversal rates 181 and mimic swimming in soft agar. Cells were then exposed to either control medium or medium 182

Results 231
Ethanol represses P. aeruginosa PA14 swimming motility independently of catabolism 232 and without reducing growth rate. Exogenous ethanol stimulates P. aeruginosa biofilm 233 behaviors, including attachment to glass and plastic, pellicle formation, and microcolony 234 formation on airway cells, in part through stimulation of Pel extracellular matrix production (5, 235 42). Although P. aeruginosa can catabolize ethanol (43, 44), ethanol catabolism is not required 236 for these phenotypes (5) indicating that the ethanol was acting as a signal or stimulus that 237 modulates P. aeruginosa biofilm-related behaviors. 238 To further characterize the response to non-inhibitory concentrations of ethanol, we 239 assessed ethanol effects on flagellar motility using a swim agar assay. We observed that P. 240 aeruginosa strain PA14 wild type had a 33% smaller swim zone diameter in the presence of 1% 241 ethanol when compared to the control cultures (49 mm versus 33 mm; p<0.0001) (Fig. 1A). A 242 ΔflgK mutant that lacks a flagellum is non-motile and served as a reference strain (Fig. 1A). This 243 reduction in swim zone diameter was not a result of differences in growth as P. aeruginosa 244 strain PA14 wild type had similar growth rates in this medium in the absence or presence of 1% 245 ethanol (Fig. S1A). Furthermore, the reduction in swim zone diameter also occurred 246 independently of ethanol catabolism as a ΔexaA mutant, which cannot grow with ethanol as a 247 carbon source (5), still showed motility repression when ethanol was added to the medium ( swimming motility (Fig. 1A), we examined the effects of ethanol on c-di-GMP levels in planktonic 255 13 cells after 1 h and 16 h of growth in medium with 1% ethanol. We found that ethanol caused a 256 2.6-and 1.9-fold increase (p<0.0001) in c-di-GMP at 1 h and 16 h, respectively (Fig. 1B). 257 258 Ethanol-dependent motility repression is not due to increased Pel and alginate matrix 259 production. Increased c-di-GMP signals have been associated with an increase in alginate and 260 Pel matrix production in P. aeruginosa (6, 7). Although ethanol activates WspR-dependent 261 production of Pel polysaccharide matrix (5), neither WspR nor PelA was required for the 262 reduction in swim zone diameter in the presence of ethanol, with a 36.4% and 31.6% 263 (p<0.0001) decrease in their swim zones, respectively (Figs. S2A-B). We did note that the 264 ΔwspR mutant had a slightly larger swim zone diameter in control conditions (Fig. S2B). 265 Alginate was also not required for a reduction in swim zone diameter in the presence of ethanol, 266 as two mutants defective in alginate production, ΔalgD and ΔalgU, also had similar levels of 267 swim zone reduction (26.7% and 32.6% decrease (p<0.0001), respectively, as the wild type 268 primary focus was on mutants that (i) had a swim zone greater than or equal to that of the wild 280 type in control conditions, and (ii) had showed less of a reduction in swim zone diameter when 281 ethanol was present in the medium. Using these criteria, analysis of the data from three 282 independent screens of the mutant collection identified SadC and GcbA as the most promising 283 candidates (Table. S1); data for mutants with swim zone sizes smaller than wild type under 284 control conditions are provided (Table S2), but not pursued as part of these studies. The 285 differences in the magnitude of the effect of ethanol on swim zone size in the ΔgcbA and ΔsadC 286 single mutants compared to the wild type were small ( Fig. 2A and Fig. S4A-B), but could be 287 complemented with the wild-type gcbA and sadC genes, respectively, in trans ( Fig. S4A-B). 288 Both SadC and GcbA have been reported to impact c-di-GMP levels (7, 14, 25). Deletion of 289 wspR in combination with either sadC or gcbA did not enhance the resistance of the effects of 290 ethanol on motility ( Fig. 2A). 291 The effects of SadC and GcbA on changes in motility in response to ethanol were 292 additive as the ΔsadCΔgcbA double mutant showed no significant difference in swim zone 293 diameter between medium without and with ethanol ( Fig. 2A). The ΔsadCΔgcbA mutant also 294 had lower levels of c-di-GMP in planktonic cultures, both in the absence and presence of 295 ethanol, when compared to wild type in the same conditions (Fig. 2B). While the wild type 296 showed 2.1-fold higher levels in c-di-GMP in ethanol-grown cells, the ΔsadCΔgcbA mutant only 297 showed a 1.4-fold difference. The small but significant increase that remained in the 298 ΔsadCΔgcbA mutant upon growth with ethanol suggests that other enzymes may contribute to 299 changes in cellular c-di-GMP pools when ethanol is present. 300 301

Ethanol induced motility repression requires two PilZ-domain proteins, FlgZ and PilZ. 302
Among the c-di-GMP binding effectors in P. aeruginosa are PilZ-domain proteins (47, 48). There 303 are eight known PilZ-domain proteins in P. aeruginosa, and some of these proteins have been 304 shown to mediate changes in motility and/or biofilm formation (22, 48). Given that ethanol 305 stimulates c-di-GMP production and motility regulation, we assessed whether one or more of 306 these PilZ-domain proteins might be involved in ethanol-dependent motility repression. 307 In the absence of ethanol, all eight mutants and the wild type had swim zone diameters 308 that were similar (Fig. 3A). While six of the mutants phenocopied the wild type, two mutants 309 displayed significantly greater swimming motility than that observed for the wild-type strain in In line with the hypothesis that PilZ-domain proteins interact with flagellar stators to 327 reduce motility in the presence of ethanol, the ΔmotAB mutant had no observable change in 328 motility in the presence of ethanol versus control cultures (Fig. 4). Also consistent with previous 329 reports, the ΔmotCD mutant displayed a swim zone diameter that was ~90% less than that of 330 wild-type cells grown in the absence of ethanol (24) (Fig. 4). Overall, these data support the 331 conclusion that the MotAB stator set is required for ethanol-dependent swim repression. In contrast to the wild-type strain, the ΔpilY1 mutant did not show decreased motility when 341 ethanol was added to the medium (Fig. 5B). Instead, the ΔpilY1 mutant showed a reproducible 342 and significant increase in the swim zone diameter when ethanol was added to the medium 343 (Fig. 5B), and a wild-type copy of the pilY1 gene complemented this phenotype (Fig. 5C). 344 Moreover, a mutant lacking pilMNOP showed no ethanol-dependent reduction in swimming 345 motility (Fig. 5B). Though PilY1 and PilMNOP were required for the motility decrease in the 346 presence of ethanol, they were not required for the stimulation of global c-di-GMP levels in 347 planktonic cells (Fig. 5D). These data indicate that PilY1 and PilMNOP are involved in the PilY1 is necessary for T4P activity (55), and thus we sought to determine if PilY1-and 353 PilMNOP-dependent reduction in flagellar motility by ethanol was due to a decrease in T4P 354 activity. Two pieces of evidence argue against a role for the T4P in ethanol-mediated effects on 355 motility. First, a ΔpilA mutant (which lacks pili) still showed the same level of motility and 356 responsiveness to ethanol (Fig. S5A) when compared to wild-type cells. Secondly, ethanol did 357 not reduce twitching motility in wild-type cells (Fig. S5B); rather, a small but significant increase 358 in twitch zone diameter was observed in cultures with ethanol. These data suggest that the 359 ethanol effects on swimming motility are not due to changes in T4P function. Cells from mid-exponential phase cultures were treated with ethanol for 15 min prior to 392 measurement of the reversal frequency. P. aeruginosa strain PA14 wild type showed a 4.8-fold 393 increase in its reversal frequency in the presence of ethanol, going from 4.6 ± 4.1 to 22 ± 9.1 394 reversals/ 10 s (p<0.0001; Fig. S7A). Similarly, the ΔsadCΔgcbA, ΔpilY1, and ΔpilMNOP 395 mutants also showed significant 2.8-fold, 3.4-fold, and 2.7-fold (p<0.0001) increases in reversal 396 frequencies upon the inclusion of ethanol in the medium (Fig. S7A-B). There were no significant 397 differences between the wild type and the mutants in either the control (except for 398 ΔsadCΔgcbA) or ethanol conditions (Fig. S7A-B). These data indicated that in the presence of 399 ethanol, P. aeruginosa had a higher rate of flagellar reversals than in control conditions, but this

Ethanol rapidly increases the sub-population of immobile cells in swim agar, in a PilY1-, 404
PilMNOP-, FlgZ-, PilZ-, and MotAB-dependent manner. We next observed the behavior of 405 single cells in swim agar in the absence and presence of 1% ethanol in order to better 406 understand how ethanol affected the macroscopic swim zone size. To do this experiment, we 407 exposed exponentially growing cells to swim agar without and with ethanol for 30 min followed 408 by the acquisition of 8 s time-lapse movies to visualize cellular behavior as outlined in Fig. 6A. 409 We first noted that when the fraction of motile cells in the control and ethanol-treated 410 samples were compared for each mutant, all except ΔpilMNOP were statistically different (Fig.  411   6B). We also noted that in the control cultures, ΔpilMNOP was significantly higher than wild type 412 and ΔpilY1 in the same condition (Fig. 6B) in the absence and presence of ethanol (58 ± 5.1% and 56 ± 3.6%, respectively; Fig. 6B). 419 Interestingly, the ΔpilY1, ΔpilZΔflgZ, and ΔmotAB cells showed an increase in the fraction of 420 motile cells in the presence of ethanol (Fig. 6B) which also mirrored the observation that the 421 ΔpilY1 strain had a larger swim zone size in the presence of ethanol (Fig. 5B). Of the mutants 422 that were resistant to the effects of ethanol in the macroscopic swim zone assay, only the 423 ΔsadCΔgcbA double mutant was not significantly different from the wild-type strain (46 ± 6.4% 424 motile (control) and 29 ± 6.4% motile (ethanol) ; Fig. 6B). These data suggest that in response to Ethanol inhibits swarming motility. In addition to its effects on planktonic cells and motility in 433 swim agar, ethanol also inhibits flagellum-dependent swarming motility on agar surfaces (Fig.  434 7A) (5). We found that PilY1, the PilMNOP alignment complex, the PilZ and FlgZ proteins, and 435 the MotAB stators, which were all required to increase the fraction of sessile cells in medium 436 with ethanol, were also necessary for full suppression of flagellar-mediated swarming motility on 437 the surface of 0.5% agar in the presence of ethanol (Fig. 7A-D). Furthermore, while the vWA 438 domain of PilY1 has been shown to be important for surface-sensing (23), we found that this 439 domain was not required for ethanol-mediated repression of swarming motility (Fig. 7A). 440 Consistent with our observation that the ΔsadC, ΔgcbA, and ΔsadCΔgcbA mutants were 441 resistant to the effects of ethanol on motility in the swim agar assay, these mutants were also 442 less responsive to ethanol in the surface-associated swarming motility assay (Fig. 7E). Here we present a model (Fig. 8)  Previous studies have highlighted the distinct roles that SadC and GcbA play during the 474 different stages of biofilm formation. GcbA, for example, was implicated in c-di-GMP production 475 only in planktonic cells or cells initiating biofilm formation or dispersing from a mature biofilm 476 (21, 25, 63). SadC, on the other hand, is implicated in biofilm initiation and maturation (7, 21). It 477 is also important to note that the ∆sadC∆gcbA double mutant still showed a significant, albeit 478 reduced, increase in c-di-GMP in the presence of ethanol (Fig. 2B), and thus other c-di-GMP 479 metabolizing enzymes may be involved in ethanol-mediated swimming repression, a finding 480 consistent with our initial genetic screen. In addition, c-di-GMP metabolizing enzymes can affect 481 We also found that P. aeruginosa flagellar reversal frequency was significantly increased 509 in the presence of ethanol, but this response did not depend on any of the proteins that we 510 tested. Previous studies have indicated that an increase in reversal frequency increases the 511 cell's ability to move more efficiently through soft agar (7, 67). Additionally, P. aeruginosa and 512 related bacteria that utilize a run-reverse-turn trajectory, spend equal time going clockwise or 513 counterclockwise with variation in their pause duration in order to turn at different angles to 514 maximize space exploration (32). Since chemotaxis involves the modulation of reversal 515 frequencies (57), these data may suggest that ethanol also affect chemotaxis in P. aeruginosa. 516 More work is required to determine if ethanol influences positive or negative chemotactic 517 pathways. Together, our data suggest that, while ethanol reduces the fraction of motile P. 518 aeruginosa cells within a given time interval, these motile cells can navigate a viscous 519 environment more efficiently in order to remain in the local space of the ethanol-producing 520

microbes. 521
To conclude, our findings indicate that ethanol triggers a complex response that 522 modulates behaviors related to biofilm initiation in order to facilitate the transition from being 523 motile to being sessile. Therefore, the effects of ethanol on microbes at concentrations much  This motility repression requires two PilZ domain proteins, PilZ and FlgZ, as well as flagellar stator set, MotAB. We propose that this signal causes the flagellar machinery to 'brake', resulting in a decrease in the number of cells that are motile. The second response (2) involves a SadC and GcbA c-di-GMP production-dependent change that may also involve other c-di-GMP metabolic proteins. C-di-GMP may further activate the PilZ domain proteins. Together, these responses repress flagellar motility in swim agar conditions and a soft agar that supports swarming motility.