Resolving the conflict between antibiotic production and rapid growth by recognition of peptidoglycan of susceptible competitors

Microbial communities employ a variety of complex strategies to compete successfully against competitors sharing their niche, with antibiotic production being a common strategy of aggression. Here, by systematic evaluation of all non-ribosomal peptides (NRP) produced by B. subtilis clade, we revealed that they acted either synergistically or additively to effectively eliminate phylogenetically distinct competitors. All four major NRP biosynthetic clusters were also imperative for the survival of B. subtilis in a complex community extracted from the rhizosphere. The production of NRP came with a fitness cost manifested in growth inhibition, rendering NRP synthesis uneconomical when growing in proximity to a phylogenetically close species, carrying resistance against the same antibiotics. To resolve this conflict and ease the fitness cost, NRP production was only induced by the presence of peptidoglycan cue from a sensitive competitor. These results experimentally demonstrate a general ecological concept – closely related communities (“self”) are favoured during competition, due to compatibility in attack and defence mechanisms.


Introduction 31
In terrestrial microenvironments, bacteria are present predominantly in architecturally 32 complex and specialised multispecies communities 1 . In many instances, these 33 communities provide beneficial effects to other organisms, e.g., biocontrol agents form 34 biofilms on the surface of plant roots, thereby preventing the growth of bacterial and 35 fungal pathogens 2 . Soil and plant-associated microbial populations survive in this 36 extremely competitive niche by producing a broad arsenal of antibiotics and evolving 37 complex antibiotic-resistance mechanisms 3,4 . Social structure was shown to mediate 38 competition between populations, suggesting a role for antibiosis in shaping cohesive 39 communities 5 . 40 Bacillus subtilis clade contains several species of Gram-positive, soil dwelling, 41 beneficial bacteria, that employ multiple strategies to compete with neighbour 42 communities 2,6-12 .One major class of antibiotics produced by bacteria are the  Ribosomal Peptides (NRP), which are synthesized by large multi-enzyme complexes 44 of NRP synthases (NRPs) 2,13,14 . B. subtilis encodes four different NRP biosynthetic 45 clusters, which are responsible for the biosynthesis of surfactin, bacillaene, bacilysin 46 and plipastatin 2,15-18 . 47 The most well-characterized NRP is the cyclic lipopeptide surfactin 19,20 21 . Surfactin is 48 a small cyclic lipopeptide induced during the development of genetic competence 22 . 49 The machinery for surfactin synthesis is encoded within the srfAA-AB-AC-AD operon 50 23 . Surfactin is a powerful surfactant with antibacterial 24 and antifungal properties 25 51 and is composed of an amphipathic, cyclic heptapeptide head group that is interlinked 52 with a hydrophobic β-hydroxy fatty acid tail, comprising 12-16 carbon atoms 26-28 . 53 These features enable the surfactin molecule to interact with, and disrupt the integrity 54 of, cellular membranes 29 . 55 Bacillaene and dihydrobacillaene 15,30 , are linear antimicrobial macrolides with two 56 amide bonds and are synthesized by the PksC-R cluster mega-complex, which is 57 composed of 13 PKS and three NRPS modules 30 . The expression of pks genes 58 requires the master regulator for biofilm formation, Spo0A, and promotes the 59 competitiveness of biofilm communities 31 . 60 comparable between self and non-self DNA ( Figures 5B and S11). Thereof, a small 247 molecule pheromone, EPS, proteins or extracellular DNA did not mediate non-self-248 recognition. 249 Peptidoglycan fragments have also been reported to function as signalling molecules 250 that trigger adaptive responses. For instance, low concentration of muropeptides 251 released by actively growing B. subtilis cells act as potent germinant of B. subtilis 252 spores 48 . Therefore, we tested whether non-self peptidoglycan fragments are 253 sufficient to trigger non-self recognition of competing Bacillus species. Indeed, purified 254 peptidoglycan from B. megaterium and B. thuringiensis but not B. subtilis and B. 255 atrophaeus was sufficient to induce the expression of NRPs biosynthetic clusters at 256 concentrations with little or no effect on cell growth ( Figures 5B and S12). 257 We then explored which known regulators of NRPs antibiotics could act downstream 258 of PG signals. In B. subtilis, canonical response to secreted PG is mediated by the PG 259 sensors PrkC kinase 49 , recognizing muramopeptides, MurP 48 for Muramic acid and 260 NagP 48 for N-Acetylglucosamine. However, deleting those genes had no effect on the 261 basal levels or the induction of the transcription from NRPS promoters in the presence 262 of non-self (Figures 5C, S13 and S14). 263 Surfactin and bacilysin are both regulated by the two-component regulatory system 264 ComP/ComA involved in a major quorum response pathway that regulates the 265 development of genetic competence 2 . We found that deletion of ComA was sufficient 266 to eliminate non-self recognition of surfactin and bacillaene promoters. Specifically, for 267 surfactin expression the deletion of ComA eliminated both basal expression and non-268 self recognition, and for Bacillaene the deletion eliminated specifically non-self 269 recognition ( Figure 6A, B and S15). In contrast, deletions of Spo0A and DegU, the 270 master regulators of sporulation and biofilm formation 50 , did not specifically eliminate 271 the response to non-self peptidoglycan. CodY regulated basal expression levels of 272 from bacillaene biosynthesis promoter, but did not trigger non-self recognition (  AmpR 48 seem to be involved in NRPS transcriptional regulation. Although a deletion 277 of a single homologue was insufficient to eliminate non-self PG recognition, a 278 combination of both receptors was sufficient to significantly reduce non-self 279 recognition in competing colonies (Figures 6C and Figure S16). Collectively, these 280 results suggest that the ComA pathway is acting jointly with the PG sensory machinery 281 (YwqM and YkoG) to fine-tune antibiotic production. such as antibiotics, digestive enzymes, quorum sensing inhibitors, low molecular 286 weight compounds like hydrogen peroxide 10, 20-23 , etc. 287 The primary model of our study is the plant biocontrol agent and probiotic bacterium 288 B. subtilis 2,52 . Various wild strains of B. subtilis and related species isolated from the 289 soil are known to defend their host from diverse fungal and bacterial pathogens 21,53 . 290 These probiotic properties are largely mediated by the formation of non-ribosomal 291 peptides 54,55 . 292 In B. subtilis, a boundary is formed between kin and non-kin strains during swarming 11 , 293 and this self-recognition was shown to involve contact dependent inhibition 9 , while in 294 Proteus mirabilis self recognition involved type VI secretion systems 56,57 . It was 295 previously hypothesized that direct cell damage (interference competition) activates 296 bacteriocins and antibiotics 58 . Consistently, antimicrobials were also suggested to play 297 a role in stabilizing communities composed of kin strains in B. subtilis 9 . However, no 298 specific antibiotic was sufficient for boundary formation between related and unrelated 299 strains and the mechanism for non-self recognition remained to be determined. 300 Previous studies demonstrated that during microbial competition, numerous soil 301 bacteria utilizes NRP antibiotics to ward off the invading microbes and thus protect 302 their niche 9,10,24,25 . Although molecular mechanisms of synthesis of the NRPs 303 surfactin, bacillaene, bacilysin and pliplstatin is largely resolved 59 , it remains unknown 304 whether these antibiotics interact. Furthermore, bacteria that produce these antibiotics 305 are subject to a fitness cost lowering their growth rate. Therefore, we used a 306 genetically manipulatable antibiotic producer to study how different antibiotics interact, 307 and to understand their regulation in the light of microbial competition and the conflict 308 between antibiotic production and growth. 309 We found that the production of NRPs was specifically activated during competitions 310 with sensitive species. Our results demonstrated that NRP antibiotics act as 311 synergistic pairs (surfactin and bacillaene) or additively (with plipstatin and bacilysin) 312 during interspecies competition. All antibiotics toxicity was specific towards competing 313 species phylogenetically distant from the Bacillus genus, but sharing the same niche. 314 We therefore asked whether the expression of these biosynthetic clusters results in a 315 deleterious effect on their producers' growth in isolation. Indeed, production and 316 increased production of NRPS came with a cost for the bacterial growth. This trade-317 off generates selective pressure on the regulation of NRP promoters ( Figure 6D), as 318 antibiotic production becomes deleterious when encountering closely related (and 319 resistant) community members. Using transcriptional reporters, we showed that B. 320 subtilis increased the transcription of NRP antibiotics only when sensing a specific 321 secreted signal from sensitive species. This suggests that soil bacteria, such as B. 322 subtilis, can discriminating "self" from "non-self", and reduce the burden of producing 323 antibiotics when encountering resistant clade members. Cell communication and 324 metabolic exchange are essential part of interspecies competition and lead to the 325 regulation of various molecular elements 26,27 . Here we demonstrate how a PG cue for 326 distinguishing of "self" from "non-self" can serve to minimize the cost for antibiotic 327 production. 328 An essential feature of B. subtilis and Gram-positive bacteria in general is their cell 329 wall. It counteracts the high intracellular osmotic pressure, determines the cell 330 morphology, serves as diffusion barrier and serves as first layer of defence against 331 environmental threats 60,61 . Fragments of peptidoglycan, a mesh-like structure in the 332 envelope, were shown to be sensed to induce the germination of spores 49 , virulence 333 genes of the human pathogen P. aeruginosa 62 and antibiotic resistance genes 63 . 334 Furthermore, the soil bacterium Streptomyces coelicolor induces the production of 335 antimicrobials following perception of NAG from similar species 30,45 . Here, we report 336 on a new complementary role for PG recognition: distinguishing self and non-self 337 phylum members competing for the same niche. The normal, unmodified glycan 338 strands of bacterial peptidoglycan consist of alternating residues of β-1,4-linked N-339 acetylmuramic acid and N-acetylglucosamine. Glycan strands become differentially 340 modified by in different species by enzymes responsible for the N-deacetylation, N-341 glycolylation and O-acetylation of the glycan strands 64 . The composition of the 342 interpeptide bridge also differs between different species 64,65 , making PG fragments 343 an appealing source of information regarding potential competitors' identity, 344 Our observation that ComA acts downstream to PG sensing is especially intriguing. 345 ComA is the transcription factor regulating genetic competence, a process in which B. 346 subtilis takes up foreign DNA 66 . It is not required for basal expression from bacillaene 347 biosynthesis promoter ( Figure 6A). However, it was required to regulate antibiotic 348 production genes expression. These results indicate that genetic competence is co-349 regulated with a peak of antibiotics, in the presence of similar but not identical 350 competitors. If so, these networks may indicate a co-evolution between the regulation 351 of antibiotic production and horizontal gene transfer: competitors are sensed by PG 352 receptors, lyzed by the induced antibiotics, and the DNA of these competitors is used 353 to increase genetic diversity. 354 Disguising members of related species and unrelated species is of high importance, 355 as members of the same clade are often resistant to their own self-produced 356 antibiotics, and therefore the production of antibiotics under these conditions is more 357 hazardous than beneficial for the producers. Overall, our findings indicate how 358 multispecies communities of similar genotypes are favoured during competition, due 359 to compatibility in antibiotic production and sensitivity. Coupling antibiotic production 360 and the identification of "non-self" competitors provides a simple new principle that 361 shapes bacterial communities, which may be further used for engineering 362 microbiomes for ecological, medical and agricultural applications. 363 364

365
Strains and media 366 All the strains used in this study are listed in supplementary Table S1.

Strain construction
All deletion mutants were derived from B. subtilis NCIB 3610 WT. Deletions mutants were constructed using standard methods as described in 1,2 . For polymerase chain reactions, primers and plasmids used in this study are listed in supplementary μg/ml chloramphenicol (Amersco) and 10 μg/ml spectinomycin (Tivan biotech).
For the generation of the Luminescence reporter PsrfAA-lux we performed restriction free cloning 5 using the plasmid pBS3C-lux 6 , which contains a functional luciferase operon (luxABCDE). Briefly, promoter region was amplified from B. subtilis chromosomal DNA, using primers listed in supplementary Table S2 and ligated into plasmid. The ligated plasmids were then transformed into E. coli DH5α. Positive reporters were inserted by using double homologous recombination into neutral integration sites (sacA) in the genome of B. subtilis by inducing natural competence 4 . Selective media for cloning purposes were prepared with LB broth or LB-agar using antibiotics at the following final concentrations: 10 μg/ml chloramphenicol (Amersco).

Constructing ofs pPyr-Cm -mKate plasmid
Primer pairs for generating DNA fragments with varying overlapping ends by PCR were designed using SnapGene Viewer program. Briefly, mKate gene was first amplified from the Gibson assembly products were then treated with DpnI digestion the transferred into competent E. coli DH5α cells. Positive reporters were selected and the insertion of mKate was confirmed by PCR. Selective media for cloning purposes were prepared with LB broth or LB-agar using antibiotics 10 μg/ml chloramphenicol (Amersco).

Extraction of DNA, EPS and PG
DNA was extracted using Wizard® Genomic DNA Purification Kit. Spectrophotometric analysis was performed using NanoDrop spectrophotometer (DNA ≈ 100ng/ μL) and DNA was stored at −80 °C for further use.
EPS was extracted from the bacilli biofilms grown at 48 h at 30°C in B4 medium. Biofilm colonies were scrapped and suspended in phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 ), were mildly sonicated, and were then centrifuged to remove the cells. The supernatant was collected and mixed with five