Toxoflavin secreted by Pseudomonas alcaliphila inhibits growth of Legionella pneumophila and its host Vermamoeba vermiformis

Legionella pneumophila is a natural inhabitant of water systems. From there, it can be transmitted to humans by aerosolization resulting in severe pneumonia. Most large outbreaks are caused by cooling towers contaminated with L. pneumophila. The resident microbiota of the cooling tower is a key determinant for the colonization and growth of L. pneumophila. The genus Pseudomonas correlates negatively with the presence of L. pneumophila, but it is not clear which species is responsible. Therefore, we identified the Pseudomonas species inhabiting 14 cooling towers using a Pseudomonas-specific 16S rRNA amplicon sequencing strategy. Cooling towers free of L. pneumophila contained a high relative abundance of members from the Pseudomonas alcaliphila/oleovorans phylogenetic cluster. In vitro, P. alcaliphila JCM 10630 inhibited the growth of L. pneumophila on agar plates. Analysis of the P. alcaliphila genome revealed the presence of a genes cluster predicted to produce toxoflavin. L. pneumophila growth was inhibited by pure toxoflavin and by extract from P. alcaliphila culture found to contain toxoflavin by LC-ESI-MS. In addition, toxoflavin inhibits growth of Vermameoba vermiformis, a host cell of L. pneumophila. Our study indicates that P. alcaliphila may be important to restrict growth of L. pneumophila in water systems through the production of toxoflavin. A sufficiently high concentration is likely not achieved in the bulk water but might have a local inhibitory effect such as in biofilm.

Not long after the discovery of Legionella, it was established that it is transmitted to human by 55 inhalation of aerosols containing L. pneumophila that are generated by engineered water systems 56 (EWS) (Meyer, 1983). Several type of EWS can shed L. pneumophila, including water distribution 57 systems (showers and faucets), spas, fountains, and cooling towers (Heijnsbergen et al., 2015). The biofilm-formation ability of P. alcaliphila was investigated by inoculating media with 185 bacterial. Briefly, 1 mL of trypticase soy broth, King's B, or R2-A broth was added to center four 186 wells of a 24-well plate. Surrounding wells were filled with sterile water to prevent desiccation. 187 Then, 20 µL of P. alcaliphila in Fraquil (OD600nm = 0.1) was added to each well and incubated at 188 room temperature with or without shaking at 150 rpm. After a week, images of the wells were 189 taken. When incubated without shaking, plates were shaken at 150 rpm for 1 hour to determine if 190 pellicles could be formed. 191

Vermamoeba vermiformis inhibition assay 192
10 The sensitivity of L. pneumophila host Vermamoeba vermiformis to toxoflavin was determined by 193 monitoring its growth in the presence of toxoflavin. V. vermiformis were grown at room 194 temperature in 75 cm 2 cell culture flasks (Sarstedt) in modified PYNFH medium (ATCC medium 195 1034) and passaged when confluence was reached. The amoebas were passaged 3 days prior to 196 exposure by adding 5 mL of culture to 20 mL of fresh modified PYNFH. Cell concentration was 197 with a Guava EasyCyte flow cytometer. To prepare samples for flow cytometer, 400 µL of culture 198 was centrifuged at 5000 g for 2 min, the supernatant discarded, and the pellet resuspended in 400 199 µL of phosphate buffered saline (PBS). The stock culture was diluted to 5 × 10 4 cells/mL in fresh 200 modified PYNFH and 900 µL was added to the wells of a 24-well plate. Then 100 µL of different 201 toxoflavin solutions were added to wells to give final toxoflavin concentrations of 0, 10, 25, 50 or 202 100 µg/mL. The plate was incubated at room temperature without shaking. After 2 and 4 days, 400 203 µL samples were taken from each well to measure cell concentration with flow cytometer. Each 204 condition was performed in triplicate. Results were analyzed using two-way ANOVA, with time 205 and toxoflavin concentration as factors, and Tukey's test correction for mutltiple comparison was 206 used to access significance between conditions. 207 208 3. RESULTS 209

Profiling of Pseudomonas species in cooling towers 210
Pseudomonas-specific 16S rRNA amplicon sequencing was performed on triplicate samples from 211 14 cooling towers. The microbiota of these cooling towers was previosuly studied using 16S rRNA 212 sequencing and 18S rRNA sequencing (Paranjape et al., 2020b(Paranjape et al., , 2020a control and DNA extracted from a blank carthridge were also included. As can be seen in Figure  216 2, these 2 control samples contained very few sequences passing quality control, indicating that 217 the amplicons from the cooling tower samples are not contaminated by spurious sequences. The 218 cooling tower samples contained a minimum of 51089 sequences passing quality control. The data 219 set was rarefied to 50000 sequences per cooling tower. All sequences were assigned to 220 Considering all the cooling towers, 34 Pseudomonas species were found among which 14 can be 229 described as major or abundant species and the other 20 as minor species, collectively representing 230 less than 5% of the population. Of note, this method is unable to differentiate closely related 231 species, such as P. alcaliphila and P. oleovorans. Such species are therefore grouped in clusters. 232 The highest diversity of Pseudomonas species was observed in cooling towers Cdq1 and Est1 233 containing 26 and 13 different species, respectively, followed by CT CN3, CN4 and MTL5 ( Figure  234 2A). Gobally, the top three most abundant Pseudomonas species in the studied cooling towers 235 were P. alcaliphila/oleovorans, P. monteilii and P. alcaligenes. P. alcaliphila/oleovorans was 236 observed in nearly 100% of cooling towers in various proportions, but was the largely dominant 237 Pseudomonas species in several cooling towers including CN2, Out1, Out2, Mont1, MTL1, 238 MTL4, MTL7 and MTL8 ( Figure 2A). The human pathogen P. aeruginosa was detected only in 239 towers CdQ1 and Est2 at a low abundance of 0.01. Next, we calculated the abundance of each 240 Pseudomonas species as a fraction of relative bacterial abundance (Paranjape et al., 2020b). As 241 can be seen in Figure 2B  Our results reavealed that a member of the P. alcaliphila/oleovorans cluster seems to be the main 293 inhibitor of L. pneumophila colonisation in the cooling towers we studied. Therefore, we 294 investigated if an isolate of that cluster, P. alcaliphila strain JCM 10630, can inhibit L. 295 pneumophila growth in vitro (Yumoto et al., 2001). 296 We carried out L. pneumophila inihibition assay at three different temperatures: 25 °C, 30 °C and 297 37 °C. After three days of incubation, P. alcaliphila inhibited the growth of L. pneumophila at 25 298 °C and 30 °C but not at 37 °C ( Figure 5). A one-way ANOVA with a Tukey correction for mutltiple 299 comparison was used to access significance between conditions. The diameter of inhibition was 300 significantly larger at 25 °C than at 30 °C for both strains (P < 0.001). The two strain tested showed 301 similar inhibition zones at each temperature tested (P > 0.6). The size of the colony of P. 302 alcaliphila was significantly larger at 25 °C than at 30 °C (P < 0.001). The strain of L. pneumophila 303 seems to influence slighlty the growth of P. alcaliphila at 25 °C as the colony was slightly larger 304 (14.3 mm) when grown with the Quebce strain than with the Philadelphia-1 strain (13 mm, P = 305 0.02). There were no difference in colony size at 30 °C. The P. alcaliphila strain JCM 10630 genome was retrieved from RefSeq (GCF_900101755.1) and 316 was analysed to identify clues as to the cause of the inhibition of L. pneumophila growth. We first 317 used antiSMASH (Blin et al., 2019) to identify putative biosynthetic gene clusters (BCGs). Five 318 20 clusters were found but none showed similarity higher than 50% with known clusters (Table 1). 319 Next, we used the Blast KOALA function of the Kyoto Encyclopedia of Genes and Genomes to 320 assign Kegg orthology annotation to the genes and predict metabolic pathways present in this 321 genome (Kanehisa et al., 2016). A cluster of genes homologous to toxoflavin synthesis cluster was 322 detected. Toxoflavin is an improtant virulence factor of the plant pathogen B. glumae (Suzuki et 323 al., 2004). Toxoflavin is also produced by Pseudomonas protegens Pf-5 (Philmus et al., 2015). alcaliphila genes compared to B. glumae varries between 69% identity to 36% identity (Table 2). 327 Our in silico analysis suggests that the inhibition of L. pneumophila growth by P. alcaliphila could 328 be due to the production of toxoflavin, another compound, or a mixture of several molecules. 329

Toxoflavin is secreted by P. alcaliphila on CYE agar plate 346
In order to confirm that P. alcaliphila produces toxoflavin, we performed chloroform extraction 347 from CYE plate inoculated with a pure culture of P. alcaliphila. Controls included an extract from 348 a sterile CYE plate and the methanol carrier alone. L. pneumophila growth was inhibited by the 349 extract from plates inoculated with P. alcaliphila ( Figure 7C) slightly more than sterile CYE and 350 methanol ( Figure 7A and B), with zone of inhibitions of 12, 10 and 10 mm, respectively. In order 351 to confirm that toxoflavin was present, the extracts were then subjected to LC-ESI/MS. Pure 352 toxoflavin solution produced a strong peak at m/z= 194.0 ( Figure 7D). The same strong peak 353 appeared in extract from P. alcaliphila plate extracts ( Figure 7F Wozniak, 2012). Therefore, we investigated the ability of P. alcaliphila JCM 10630 to form these 368 structures. First, we tested the production of attached biofilm in R2A, King's B, and in trypticase 369 soy broth at room temperature under shaking. After one week, no attached biofilm was seen, 370 however a filamentous floating mass of cells could be seen in both R2A and trypticase soy broth 371 ( Figure 8A). We then tested the production of a pellicle by incubating P. alcaliphila in the same 372 three media but without shaking. No pellicle was formed in any of these media ( Figure 8B). 373 Nevertheless, we could see a mat at the bottom of the well in all case. Shaking for 1 h dislodge the 374 mat produced in trypticase soy broth and R2A, resulting in a floating mat similar to what was seen 375 after incubation with shaking ( Figure 8A). We therefore concluded that P. alcaliphila may 376 produces floating biofilm mats in cooling towers. Since P. alcaliphila was also negatively correlated with the presence of host cells in cooling towers 388 (Paranjape et al., 2020a), we next hypothesize that toxoflavin might be toxic for amoebas typically 389 25 found in water systems. Therefore, we monitored the growth of V. vermiformis when exposed to 390 toxoflavin (Figure 9). Within four days, cells unexposed to toxoflavin grew by 7.6-fold. In contrast, 391 cells exposed to 10 µg/ml and 25 µg/ml, grew much less, by a factor of 3.7 and 3.3 respectively. 392 PYFNH to a concentration of 50,000 cells/mL and exposed to 10 µg/ml, 25 µg/ml, 50 µg/ml and 399 100 µg/ml of toxoflavin. Cultures without toxoflavin served as a control. Cell concentration was 400 determined using flow cytometry at day 0, 2 and 4. A two-way ANOVA with a Tukey correction 401 for mutltiple comparison was used to access significance of each test conditions compared to the 402 control (ns, non-significant, * P < 0.05, ** P < 0.01, *** P < 0. Unfortunately, the method used is not able to differentiate between members of the P. 429 alcaliphila/oleovorans cluster as the region targeted is identical. In addition to P. alcaliphila and 430 P. oleovorans, this cluster also contain P. chengduensis (Pereira et al., 2018). These three species 431 are associated with various water environment (Peix et al., 2018;Tao et al., 2014). It is possible 432 that the cooling towers studied here contain a diversity of species belonging to this cluster. The 433 water of the cooling towers included in this study was typically between 20-25 °C and pH 8 434 (Paranjape et al., 2020b). This falls within the conditions that P. alcaliphila JCM 10630 can thrive 435 in, having been isolated from sea water and shown to be alkali-tolerant and psychrophilic, growing 436 best at temperature between 4 and 30 °C (Yumoto et al., 2001). 437

438
In this study, we found that P. alcaliphila was able to inhibit L. pneumophila growth was, at least 439 in part, through toxoflavin production. We found that 0.5 µg of toxoflavin directly inhibit L. 440 pneumophila growth on plates, and that a concentration of 25 µg/mL inhibits the growth of L. 441 pneumophila host V. vermiformis. Genomic analysis revealed that P. alcaliphila also contains a 442 homologue of the toxoflavin biosynthetic cluster and the presence of toxoflavin was confirmed in 443 corresponding organic extracts. Inhibition appeared to be temperature-dependent since P.