bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 Peptide-based quorum sensing systems in polymyxa

2

3 Maya Voichek1, Sandra Maaß2, Tobias Kroniger2, Dörte Becher2, Rotem Sorek1,#

4

5 1 Department of Molecular Genetics, Weizmann Institute of Science, Rehovot 76100, Israel

6 2 Institute of Microbiology, Department of Microbial Proteomics; Center for Functional 7 Genomics of Microbes, University of Greifswald, Germany.

8 # Correspondence: [email protected] bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

9 Abstract

10 is an agriculturally important plant growth-promoting 11 rhizobacterium. Many Paenibacillus species are known to be engaged in complex - 12 bacteria and bacteria-host interactions, which in other species were shown to necessitate 13 quorum sensing communication. However, to date no quorum sensing systems have been 14 described in Paenibacillus. Here we show that the type strain P. polymyxa ATCC 842 15 encodes at least 16 peptide-based communication systems. Each of these systems is 16 comprised of a pro-peptide that is secreted to the growth medium and processed to generate 17 a mature short peptide. Each peptide has a cognate intracellular receptor of the RRNPP 18 family, and we show that external addition of P. polymyxa communication peptides leads 19 to reprogramming of the transcriptional response. We found that these quorum sensing 20 systems are conserved across hundreds of species belonging to the 21 family, with some species encoding more than 25 different peptide-receptor pairs, 22 representing a record number of quorum sensing systems encoded in a single genome. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

23 Introduction

24 Paenibacillaceae is a diverse family of bacteria, many of which are important in 25 agricultural and clinical settings. These include , a pathogen causing 26 the lethal American Foulbrood disease in honeybees1, as well as P. dendritiformis and P. 27 vortex, which are used as models for complex colony pattern formation2,3. Arguably the 28 best studied member of Paenibacillaceae is P. polymyxa, known for its plant-growth 29 promoting traits4–6. P. polymyxa produces a plethora of beneficial compounds, including 30 the polymyxin7 and the phytohormone indole-3-acetic acid (IAA)8, and was 31 shown to protect multiple plant species against pathogens (reviewed in 9). Due to its 32 beneficial traits, P. polymyxa is commercially used as a soil supplement to improve crop 33 growth10,11.

34 Many of the characteristic features associated with the above-mentioned behaviors – 35 production and secretion of antimicrobials, expression of virulence factors, and forming 36 complex colony structures – often require some form of inter-cellular communication such 37 as quorum sensing12. Bacteria use quorum sensing to coordinate gene expression patterns 38 on a population level. For example, the quorum sensing network of Pseudomonas 39 aeruginosa controls the production of secreted toxins as well as the production of 40 rhamnolipids, which are important for its architecture13,14. Quorum sensing 41 regulates the production of bacteriocin antimicrobials in Lactobacilli15 and 42 Streptococci16,17; and in subtilis, quorum sensing systems play a major role in 43 sporulation, biofilm formation and genetic competence18. Although Paenibacillus bacteria 44 are engaged in similar behaviors, no quorum sensing was reported in any Paenibacillus 45 species to date.

46 Gram-positive bacteria frequently use short peptides, termed autoinducer peptides, as their 47 quorum sensing agents12. A widespread group of such peptide-based communication 48 systems involves intracellular peptide sensors (receptors) of the RRNPP family, together 49 with their cognate peptides19. Peptides of these quorum sensing systems are secreted from 50 the cell as pro-peptides and are further processed by extracellular proteases20 to produce a 51 mature short peptide, usually 5-10 amino acids (aa) long19,21. The mature peptides are re- 52 internalized into the cells via the oligopeptide permease (OPP) transporter22,23, and bind bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

53 their intracellular receptors24. These receptors act either as transcription factors or as 54 phosphatases, and peptide binding activates or represses their function eventually leading 55 to modulation of the bacterial transcriptional program. For example, in B. subtilis the Rap 56 receptors function as phosphatases that regulate the major transcription factor Spo0A; upon 57 binding to their cognate Phr peptides, the phosphatase activity of Rap receptors is inhibited, 58 leading to activation of Spo0A that alters the transcriptional program25,26. In B. 59 thuringiensis the PlcR receptor is a transcription factor that becomes activated when bound 60 to its cognate PapR peptide, leading to expression of virulence factors that facilitate 61 infection of its insect larval host27. Such peptide-based communication was recently shown 62 also to control the lysis-lysogeny decision of phages infecting many species of Bacillus28,29.

63 Here we report the identification of a large group of peptide-based quorum sensing systems 64 in P. polymyxa. We found that pro-peptides of these systems are secreted and further 65 processed, and that the mature peptides modulate the transcriptional program of P. 66 polymyxa. We further show that these quorum sensing systems are conserved throughout 67 the Paenibacillaceae family of bacteria, with some bacteria encoding over 25 different 68 peptide-receptor pairs in a single genome.

69 Results

70 Identification of RRNPP-like peptide-receptor pairs in P. polymyxa ATCC 842

71 All members of the RRNPP family of intracellular peptide receptors contain a C-terminal 72 tetratricopeptide repeat (TPR) or TPR-like domain, which forms the peptide-binding 73 pocket in the receptor protein19,30. To search for possible such quorum sensing systems in 74 P. polymyxa we first scanned all annotated protein-coding genes of the type strain P. 75 polymyxa ATCC 84231 for TPR domains using TPRpred32 (Methods). As TPR domains are 76 found in diverse protein types, we manually analyzed the genomic vicinity of the identified 77 TPR-containing proteins to search for a peptide-encoding gene characteristic of quorum 78 sensing systems. We found two TPR domain genes that were immediately followed by a 79 short open reading frame with a predicted N-terminal hydrophobic helix, typical of a signal 80 peptide that targets the pro-peptide for secretion33 (Fig. 1a). Two-gene operons that encode bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

81 the receptor followed by its cognate peptide are a hallmark of many known members of 82 the RRNPP family of quorum sensing systems19.

83 We then used BLAST to search for additional homologs of the two putative TPR-domain 84 receptors in the P. polymyxa ATCC 842 genome (Methods). This search revealed 14 85 additional homologous genes that were not initially recognized by TPRpred, in most cases 86 because their C-terminal TPR domains were divergent and did not pass the default 87 threshold of the TPRpred software. We found that all but three of the homologs had a short 88 ORF (32-50 aa) encoded immediately downstream to them. We denote these putative 89 communication systems as Alo (Autoinducer peptide Locus), with predicted Alo receptor 90 genes designated aloR and the cognate peptide-encoded gene aloP. We number Alo loci in 91 the P. polymyxa ATCC 842 genome from alo1 to alo16 (Fig. 1b).

92 93 Fig. 1: Alo systems identified in P. polymyxa. (a) Schematic representation of aloR-aloP operon 94 organization and putative peptide processing. HTH, helix-turn-helix; TPR, tetratricopeptide repeat. (b) Alo 95 loci identified in P. polymyxa ATCC 842. For the aloR receptor genes, the locus tag in the IMG database34 is 96 specified. Blue sequence in AloP peptides represents the predicted N-terminal signal sequence for 97 secretion, as predicted by Phobius35 (Methods). Asterisks (*) mark cases in which the aloP gene was absent 98 or found as a degenerate sequence. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

99 Alo systems are ubiquitous in P. polymyxa strains

100 Examining the genomes of 13 additional P. polymyxa strains using sequence homology 101 searches revealed that Alo systems are conserved in the P. polymyxa lineage (Fig. 2a; 102 Methods). Some of the Alo receptors (e.g., AloR3, AloR4, AloR7, AloR13, AloR15, and 103 AloR16) are highly conserved, and appeared in all P. polymyxa strains examined. Others 104 appeared variably and were absent from some genomes (Fig. 2b). We also found six 105 additional peptide-receptor pairs that were not present in P. polymyxa ATCC 842 but were 106 detectable in other strains; we denote these Alo17-22 (Fig. 2). Similar observations 107 regarding “core” and “accessory” quorum sensing systems were reported for peptide-based 108 systems in other bacteria: for example, in the case of B. subtilis Rap/Phr genes, RapA/PhrA 109 and RapC/PhrC are considered “core” systems that are conserved among B. subtilis 110 species, while other systems, including RapE/PhrE, RapI/PhrI and RapK/PhrK, occur 111 variably and are often located within mobile genetic elements36.

112 In the vast majority of cases, aloR genes in the various P. polymyxa strains were found just 113 upstream to a peptide-encoding aloP gene with an N-terminal hydrophobic helix predicted 114 to target them for secretion by the Sec system26 (Supplementary Table S1). Exceptions 115 were aloR18, for which we could not find a cognate downstream short ORF, and aloR5 116 and aloR14 in which degenerate peptide sequences were observed. In the Alo12 clade we 117 could identify a conserved short ORF in the majority of P. polymyxa strains (excluding the 118 type strain ATCC 842), yet it lacked the characteristic hydrophobic N-terminal signal 119 sequence marking it for secretion. Orphan receptors, which lack a cognate peptide, have 120 also been observed for RRNPP quorum sensing systems in other bacteria36,37. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

121 122 Fig. 2: Phylogenetic distribution of Alo systems among P. polymyxa strains. (a) Phylogenetic tree of 234 123 AloR homologs found in 14 P. polymyxa strains. Colored segments in the tree circumference define clades 124 of AloR proteins corresponding to the systems found in P. polymyxa ATCC 842 (labeled as arrows). Systems 125 immediately followed by a Spo0E-like protein are indicated. Tree was constructed using IQ-Tree38–40 with 126 1,000 iterations and ultrafast bootstrap support values are presented for major branches. Tree visualization 127 was done with iTOL41. Alo1/8 refers to a system duplicated in P. polymyxa ATCC 842 (where it is represented 128 as Alo1 and Alo8). (b) Frequency and distribution of Alo systems in P. polymyxa strains. The number of 129 copies per each Alo system found within P. polymyxa strains is shown using colored squares. The absence 130 of an Alo system in a specific strain is marked by a white square. Phylogenetic tree of the Alo systems shown 131 on the vertical axis is derived from the tree in panel (a). Phylogenetic tree of the strains shown on the 132 horizontal axis is based on the GyrA protein, with GyrA from B. subtilis 168 as an outgroup. 133 134 To understand how common Alo systems are in bacteria, we performed an exhaustive 135 profile-based search for homologs of AloR proteins in >38,000 bacterial and archaeal 136 genomes (Methods). This search yielded 1,050 homologs in 149 bacteria, essentially all of 137 which belong to the Paenibacillaceae family (Supplementary Table S2). These included 138 species of Paenibacillus, , , Fontibacillus and 139 Gorillibacterium. The largest number of Alo receptors found in a single organism was in 140 Paenibacillus terrae NRRL B-30644, in which 40 such receptors were detected (with 141 cognate peptide-encoding genes detected immediately downstream to 27 of them, 142 Supplementary Table S3). These results suggest that the Alo system is a widespread 143 quorum sensing system in this large family of bacteria, and imply that Paenibacilli are 144 among the most “communicative” bacteria studied to date. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

145 Mass-spectrometry analysis detects secretion and processing of pro-peptides

146 Peptides belonging to members of the RRNPP family of quorum sensing systems are 147 known to be secreted into the growth medium as pro-peptides. Following removal of their 148 N-terminal signal sequence, the peptides are further processed by extracellular proteases 149 to generate the mature communication peptide26,42 (Fig. 1a). To determine whether Alo 150 peptides are similarly secreted, we analyzed growth media taken from P. polymyxa cultures 151 using mass-spectrometry (MS). For this, we grew P. polymyxa ATCC 842 in defined 152 media, removed the bacteria by centrifuging, and took the supernatant (presumably 153 containing the secreted peptides) for MS analysis. Notably, MS was performed without 154 subjecting the peptides to trypsin treatment, to allow detection of peptides in their natural 155 form. We were able to reproducibly detect various fragments of nine out of the 14 Alo 156 peptides predicted in P. polymyxa ATCC 842 (Fig. 3 and Supplementary Table S4). In all 157 cases, the fragments included the C-terminus of the Alo peptide but not the N-terminus, 158 confirming that the N-terminus is cleaved during the pro-peptide secretion, presumably by 159 a signal peptidase associated with the Sec system43. As a control, we repeated the same 160 procedure on cultures of B. subtilis, where the processing patterns of Phr peptides are 161 known26. Indeed, the B. subtilis Phr peptides were also detectable in the MS analysis, with 162 the same patterns of N-terminus removal (Fig 3A and Supplementary Table S4). These 163 results show that Alo peptides are secreted to the medium akin to other cases of RRNPP 164 quorum sensing systems.

165 Following N-terminus removal during the secretion process, B. subtilis Phr peptides are 166 known to be processed further to yield the mature 5 aa peptide. The mature peptide is 167 usually found at the extreme C-terminal end of the pro-peptide, such that it may be released 168 by a single cleavage event (Fig. 3a). Such short peptides are challenging to identify using 169 current MS pipelines44, and indeed our MS analysis of B. subtilis conditioned media did 170 not detect the mature 5 aa Phr peptides but rather revealed the Phr pro-peptides with the 171 known mature peptide missing (Fig. 3a). The absence of a short peptide from the C- 172 terminus of the pro-peptide is therefore a clear signature of the processing event that 173 releases the mature peptide. We were able to detect similar processing events in the P. 174 polymyxa peptides as well (Fig. 3b). These patterns strongly suggest that some of the AloP bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

175 proteins generate 6 aa communication peptides, while others generate presumably longer 176 mature peptides (Fig. 3b).

177 178 Fig. 3: Mass spectrometry-based identification of secreted communication peptides. (a) Phr peptide 179 fragments detected in the growth media of B. subtilis 168. Peptide fragments identified by MS are 180 underlined, with the number of spectra detected indicated. Known mature communication peptides are in 181 red26,45. Blue indicates the predicted N-terminal signal sequence for secretion26. The two most abundantly 182 identified fragments are presented for each protein. (b) AloP peptide fragments detected by MS. Same as 183 (a) for AloP peptide fragments detected in the growth media of P. polymyxa ATCC 842. Blue indicates the 184 N-terminal signal sequence for secretion as predicted by Phobius35. 185

186 Peptide-mediated modulation of the bacterial transcriptional program

187 All Alo receptors contain a helix-turn-helix (HTH) DNA-binding domain at their N- 188 terminus (Fig. 1a), suggesting that they may function as transcriptional regulators similar 189 to many members of the RRNPP peptide receptors family19,24. These regulators become 190 activated or repressed once bound by their cognate peptides, leading to alteration of the 191 transcriptional program in response to the quorum sensing peptides. To examine whether 192 Alo peptides affect the transcriptional program of P. polymyxa, we selected Alo13, one of 193 the systems conserved across all P. polymyxa strains, for further experimental 194 investigation. The sequence of the mature AloP13 peptide, as predicted by the MS analysis, 195 is SHGRGG (Fig. 3b). Aiming to measure the immediate transcriptional response to the 196 peptide, we incubated early log-phase-growing P. polymyxa cells with 5µM of synthetic bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

197 SHGRGG peptide for 10 minutes. RNA-seq analysis of cells harvested after this short 198 incubation with the peptide identified a set of 10 genes whose expression became 199 significantly reduced when the peptide was added (Fig. 4). This transcriptional response 200 was not observed when a scrambled version of the AloP13 pro-peptide (encoding the same 201 amino acids as the pro-peptide but in a scrambled order) was added to the medium, 202 indicating that the transcriptional response is specific to the sequence of AloP13 (Table 203 S5).

204 The most significant reduction in expression was in a gene annotated as Spo0E-like 205 sporulation regulatory protein, whose expression was reduced by 4.5 fold on average in 206 response to the peptide (3.7, 6, and 4 fold reduction in the 3 independent replicates of the 207 experiment) (Fig. 4b). In B. subtilis, Spo0E functions as a phosphatase that regulates 208 Spo0A, the master transcription factor for sporulation, thus inhibiting the initiation of the 209 sporulation pathway46. Interestingly, the spo0E-like gene is encoded directly downstream 210 to the Alo13 locus in P. polymyxa (Fig. 4c-d). These results imply that the immediate effect 211 of the AloP13 quorum sensing peptide involves down regulation of the spo0E-like gene, 212 which likely leads to a further regulatory cascade at later time points. We note that spo0E- 213 like genes occur immediately downstream of 6 Alo systems in P. polymyxa (Fig. 2a) and 214 downstream to 12 of the P. terrae Alo systems. This conserved genomic organization 215 suggests that many Alo systems may exert their effect on the cell by modulating the 216 expression of Spo0E-like regulatory phosphatase proteins.

217 Most of the additional genes whose expression became reduced following 10 minutes' 218 exposure to the AloP13 mature peptide were those encoding other AloP peptides, including 219 aloP1, aloP2, aloP8, aloP10α and aloP10β, as well as the gene encoding the receptor 220 aloR14 that is encoded adjacent to the aloR13 gene (Fig. 4). These results suggest that as 221 in other bacteria37,47, quorum sensing systems can cross-regulate other such systems in P. 222 polymyxa. Combined, these results show that the AloP13 mature peptide affects the 223 transcription of specific target genes, and verifies Alo systems as quorum sensing systems 224 in P. polymyxa. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

225

226 Fig. 4: The mature AloP13 peptide elicits an immediate transcriptional response. (a) Volcano plot depicting 227 P. polymyxa gene expression after incubation with 5µM of the peptide SHGRGG for 10 minutes, as 228 compared to control conditions in which no peptide was added. X-axis, log2 fold expression change 48 229 (adjusted using the lfcShrink function in the DESeq2 R package , Methods). Y-axis, -log10 p-value. Average 230 of 3 independent replicates; each dot represents a single gene. Dots appearing in colors are genes that 231 passed the threshold for statistical significance of differential expression (Methods). Red dots correspond 232 to genes encoded by Alo systems. (b) Differentially expressed genes. Fold change represents the average of 233 the 3 independent replicates. P-adjusted is the p-value after correction for multiple hypothesis testing 234 (Methods). (c) RNA-seq coverage of the Alo13-Alo14 locus (Scaffold: PPTDRAFT_AFOX01000049_1.49), in 235 control conditions (green) or 10 minutes after addition of 5 µM of the SHGRGG peptide (black). RNA-seq 236 coverage is in log scale and was normalized by the number of uniquely mapped reads in each condition. 237 Representative of 3 independent replicates. (d) RNA-seq coverage of the Alo13-Alo14 locus in control 238 conditions (green) or 10 minutes after addition of 5µM of the scrambled AloP13 pro-peptide 239 VSGTRHGSFHSGIGNSGIYIQNT (blue). RNA-seq coverage is in log scale and normalized as in panel (c). 240 Representative of 3 independent replicates. Data for the control sample is the same as in panel (c). bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

241 The AloP13 pro-peptide induces bacterial clumping

242 We noticed that addition of the AloP13 pro-peptide (sequence 243 SYFHNITIGNGSQITVSSHGRGG) to a liquid growth medium resulted in rapid clumping 244 of the bacteria into macroscopic particles (Fig. 5; Supplementary Movie S1). The clumping 245 phenotype repeated when using an AloP13 pro-peptide synthesized by two different 246 vendors. Clumping was not observed when using the synthetic “scrambled” peptide 247 encoding the same amino acids as the AloP13 pro-peptide but in a scrambled order, 248 suggesting that clumping is specifically induced by the AloP13 pro-peptide (Fig. 5). 249 Clumping was previously described to be mediated by quorum sensing in other bacteria, 250 where it was associated with induction of conjugation49, competence50 and virulence51. 251 Surprisingly, addition of the mature 6 aa peptide SHGRGG to the medium did not result in 252 observable clumping, and, conversely, addition of an AloP13 pro-peptide that lacks the last 253 6 aa did induce clumping (Fig. 5b). These results suggest that the AloP13 pro-peptide 254 carries out two separate functions – one is associated with the mature SHGRGG peptide 255 that probably affects transcription by entering the cell and binding its cognate receptor, and 256 the other induces bacterial clumping in a mechanism yet to be explored.

257 258 Fig. 5: The pro-peptide of AloP13 causes cell clumping. (a) Cell clumping is observed in liquid culture of P. 259 polymyxa, 30 minutes after addition of 5µM of the AloP13 pro-peptide SYFHNITIGNGSQITVSSHGRGG. 260 Clumping is not observed in a control sample in which no peptide was added, nor in a sample incubated for 261 30 minutes with 5µM of a scrambled version of the AloP13 pro-peptide (VSGTRHGSFHSGIGNSGIYIQNT). (b) 262 Cultures grown similarly to those in panel (a), as well as cultures incubated with the mature AloP13 263 (SHGRGG) and an AloP13 lacking the last 6 aa (SYFHNITIGNGSQITVS) were plated on LB-agar and grown 264 overnight. Clumping is observed as denser areas in the culture. Image in inverted colors. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

265 Discussion

266 In this work we identified a large family of peptide-based quorum sensing systems 267 employed by Paenibacillaceae bacteria, and characterized these systems in the type strain 268 P. polymyxa ATCC 842. We found that the peptides are secreted to the growth medium 269 and further processed to form mature communication peptides which lead to transcriptional 270 reprogramming. The large number of such systems in a single bacterium (16 in P. polymyxa 271 ATCC 842 and at least 27 in P. terrae) suggest that Paenibacilli may need an unusually 272 large amount of signals to coordinate their complex social traits. Previous studies have 273 predicted that P. polymyxa strains may also utilize AI-2 quorum sensing, based on the 274 presence of components of the AI-2 pathway in its genome, although the production of AI- 275 2 by P. polymyxa was not detected52,53. In addition, P. polymyxa ATCC 842 was found to 276 encode an agr-like gene cassette that is homologous to another peptide-based 277 communication system in staphylococci54. Together with our discovery of the Alo 278 communication systems, to our knowledge, Paenibacilli have the largest number of quorum 279 sensing systems reported in any bacteria to date.

280 In B. subtilis most Phr pro-peptides are processed by a single protease cleavage event, 281 releasing the C-terminal mature peptide. However, in some cases there are two cleavage 282 events (for example PhrH and PhrK), releasing an internal portion of the pro-peptide to 283 form the mature communication peptide26,45. Our MS data of B. subtilis peptides show that 284 these cases may not be detectable in our analyses, but instead may appear as longer mature 285 peptides (see PhrH and PhrK in Fig. 3a). It is therefore conceivable that some of the AloP 286 peptides in P. polymyxa, especially the ones for which the fragment detected in the MS 287 data is followed by a sequence longer than 6 aa, are further processed to yield shorter 288 mature peptides (for example AloP4, P9, P15, and P16 in Fig. 3b).

289 Our data suggest that AloP13 has two separate functions. First, its terminal 6 aa are 290 processed and trigger transcriptional reprogramming similar to other peptides in the 291 RRNPP family of quorum sensing systems. Second, another portion of its sequence (which 292 does not include the last 6 aa) causes the cells to aggregate. There are other known cases 293 of a communication peptide carrying two separate functions: for example, the bacteriocin 294 nisin is known to act both as an antimicrobial peptide and as a signaling molecule that bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

295 induces its own biosynthesis in a cell density-dependent fashion55,56. As for the observed 296 phenotype, cell aggregation is known to be caused by the CSP communication peptide in 297 Streptococcus pneumoniae, in which the peptide leads to cell competence for DNA 298 uptake50. It is therefore possible that AloP13 regulates natural competence in P. polymyxa 299 as well. If this is indeed the case, then using AloP peptides may facilitate DNA 300 transformation and genetic engineering of P. polymyxa, which is currently considered a 301 species challenging to genetically manipulate57–59.

302 We found that addition of AloP13 to growing cells results in immediate down-regulation 303 of a set of genes, including those encoding the Spo0E-like phosphatase and additional Alo 304 peptides. These immediate expression changes are expected to be translated into a signaling 305 cascade that may involve alteration of the phosphorylation state of additional cellular 306 components, ultimately giving rise to a functional phenotype. The phenotype controlled by 307 AloP13, as well as those controlled by other AloP peptides, remain to be elucidated. 308 Revealing the full network of functional phenotypes controlled by peptide communication 309 in P. polymyxa may enable efficient harnessing of the plant-growth promoting traits of this 310 organism for agricultural benefit in the future. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

311 Materials and Methods

312 Identification of Alo systems in P. polymyxa ATCC 842

313 Amino acid (aa) sequences of all protein-coding genes of Paenibacillus polymyxa N.R. 314 Smith 1105, ATCC 842 were downloaded from IMG34 (on 31/1/2017, IMG taxon ID: 315 2547132099, NCBI:txid 1036171) and scanned with TPRpred version 2.832. Genes with 316 TPRpred scores of >75% probability of having TPR/PPR/SLE domains were further 317 inspected. The genome environments of these genes were manually searched for 318 downstream annotated short ORFs with an N-terminal hydrophobic helix as predicted by 319 the Phobius web server35. For two TPR-containing genes, such downstream short ORFs 320 were identified. BlastP60 was then used to search for homologs of these two genes among 321 all P. polymyxa ATCC 842 proteins using an E-value cutoff of 1e-5. Homologs shorter 322 than 250 aa were discarded. The remaining 16 genes were numbered aloR1-aloR16, and 323 their cognate aloP genes were searched as above. In cases no cognate aloP gene could be 324 identified, the last 100 bases of the aloR gene together with the 200 bases immediately 325 downstream of the aloR were searched for short ORFs (30-50 aa) partially overlapping the 326 3' of the aloR gene using Expasy translate tool61. N-terminal hydrophobic helices of these 327 AloP peptides were predicted using the Phobius web server35. Signal sequences were 328 manually inferred in cases where the Phobius score was near the threshold based on 329 homology to other AloP sequences whose Phobius score was above the threshold.

330 Phylogenetic analysis of Alo systems

331 The aa sequences of the 16 AloR proteins identified in P. polymyxa ATCC 842 were used 332 as a query for iterative homology search against 38,167 bacterial and archaeal genomes 333 downloaded from the IMG database34 on October 2017, using the “search” option in the 334 MMseqs2 package62 (release 6-f5a1c) with “--no-preload --max-seqs 1000 --num- 335 iterations 3” parameters. The hits were filtered with an E-value cutoff of <1e-10. Homologs 336 shorter than 250 aa or homologs found in scaffolds shorter than 2,500 nt were discarded.

337 For the analysis appearing in Fig. 2, the 234 AloR homologs found in 14 P. polymyxa 338 strains were aligned using the MAFFT multiple sequence alignment server version 7 with 339 default parameters63. The alignment was used to generate a maximum likelihood bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

340 phylogenetic tree using IQ-TREE38 version 1.6.5 including ultrafast bootstrap analysis39 341 with 1000 alignments (‘‘-bb 1000’’) and using the "-m TEST" parameter to test for the best 342 substitution model40 (Best-fit model: JTT+F+R5 chosen according to BIC). The tree was 343 visualized using the iTOL program41. A similar analysis with sequences of 11 Rap genes 344 from B. subtilis 168 used as an outgroup produced the same tree structure, and was used to 345 define the tree root in Fig. 2a. Clades were manually defined for AloR receptors based on 346 the branching patterns in the phylogenetic tree (Fig. 2a), as well as similarity in the cognate 347 AloP sequence and the immediate genomic environment. For Fig. 2b, the dendrogram of 348 the Alo receptors on the vertical axis of the matrix was generated by collapsing the tree 349 shown in Fig. 2a. The dendrogram of the phylogenetic relationship between the P. 350 polymyxa strains shown on the horizontal axis of Fig. 2b was generated by multiple 351 sequence alignment and IQ-TREE analysis of the conserved GyrA protein of all 14 P. 352 polymyxa strains, including GyrA from B. subtilis as an outgroup. For the data presented 353 in Tables S1 and S3, AloR homologs were searched for their cognate aloP genes as 354 described for P. polymyxa ATCC 842.

355 Bacteria culture and growth conditions

356 P. polymyxa ATCC 842 was obtained from the BGSC (strain 25A2) and stored in -80°C 357 as a glycerol stock. For all experiments, bacteria were first streaked on an LB plate, grown 358 in 30°C at least overnight and colonies were used to inoculate round-bottom ventilated 15 359 mL tubes containing 4-5 mL LB medium, grown at 30°C with 200 RPM shaking.

360 Preparation of growth media for mass spectrometry experiments

361 To search for peptide fragments secreted to the growth medium, overnight 5 mL cultures 362 of P. polymyxa or B. subtilis 168 were first washed by centrifuging for 5 min in 3200 g at 363 room temperature and re-suspended in 5 mL of chemically defined medium lacking any 364 peptides described in reference 64. Washed cultures were then diluted 1:100 into 250 mL 365 Erlenmeyer flasks with 50 mL defined media. The cells were grown in 30°C with 200 RPM 366 shaking for 8 hrs for P. polymyxa log samples (OD ~0.1) and 24 hrs for the stationary 367 samples; and 5 hours for B. subtilis log samples (OD ~0.3) and 8 hours for the stationary 368 samples (OD ~0.9). At the designated time point, 23 mL samples were centrifuged for 5 bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

369 mins at 3200 g in 4°C, and two technical replicates of 10 mL of the supernatant for each 370 sample were transferred to new tubes and flash frozen immediately.

371 Sample preparation for MS-analysis

372 A method based on protein interaction with a solid phase was applied to enrich for peptides 373 present in the supernatants. Prior to usage, a disposable solid phase enrichment (SPE) 374 column (Phenomenex, 8B-S100-AAK), packed with an 8.5 nm pore size, modified styrene- 375 divinylbenzene resin was equilibrated by rinsing with twice 1 mL acetonitrile and twice 1 376 mL water. Subsequently, 10 mL culture supernatant was loaded and unbound, and 377 potentially larger proteins were removed by washing twice with 1 mL water. Finally, the 378 enriched sample fraction was eluted with 0.5 mL 70% acetonitrile (v/v) and evaporated to 379 dryness in a vacuum centrifuge. Samples were resuspended in 0.1% (v/v) acetic acid before 380 MS analysis.

381 Mass spectrometry

382 The enriched peptides were loaded on an EASY-nLC 1000 system (Thermo Fisher 383 Scientific) equipped with an in-house built 20 cm column (inner diameter 100 mm, outer 384 diameter 360 mm) filled with ReproSil-Pur 120 C18-AQ reversed-phase material (3 mm 385 particles, Dr. Maisch GmbH, Germany). Elution of peptides was executed with a nonlinear 386 86 min gradient from 1 to 99% solvent B (0.1% (v/v) acetic acid in acetonitrile) with a flow 387 rate of 300 nL/min and injected online into a QExactive mass spectrometer (Thermo Fisher 388 Scientific). The survey scan at a resolution of R = 70,000 and 3 x 106 automatic gain control 389 target with activated lock mass correction was followed by selection of the 12 most 390 abundant precursor ions for fragmentation. Data-dependent MS/MS scans were performed 391 at a resolution of R = 17,500 and 1 x 105 automatic gain control target with a NCE of 27.5. 392 Singly charged ions as well as ions without detected charge states or charge states higher 393 than six were excluded from MS/MS analysis. Dynamic exclusion for 30 s was activated.

394 Database search of MS results

395 Identification of peptides was carried out by database search using MaxQuant 1.6.3.4 with 396 the implemented Andromeda algorithm65 applying the following parameters: digestion bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

397 mode unspecific; variable modification, methionine oxidation, and maximal number of 5 398 modifications per peptide; activated ‘match-between runs’ feature. The false discovery 399 rates of peptide spectrum match, peptide and protein level were set to 0.01. Only unique 400 peptides were used for identification. Two databases were used for peptide identification: 401 a database for P. polymyxa ATCC 842 proteins downloaded from the IMG database34 and 402 supplemented with the sequences for the Alo systems (5,429 entries); and the reference 403 proteome of B. subtilis 168 downloaded from the Uniprot database (on 19/01/2019) 404 containing 4,264 entries. Common laboratory contaminations and reverse entries were 405 added during MaxQuant search and a peptide length of 5-35 amino acids was specified.

406 Transcriptional response to peptides (RNA-seq experiments)

407 For RNA-seq experiments, synthetic lyophilized peptides were ordered from Peptide 2.0 408 Inc. (Chantilly, Virginia) and Genscript Corp at purity levels of 99% for the mature AloP13 409 (SHGRGG) and 81-93% for the AloP13 pro-peptide (SYFHNITIGNGSQITVSSHGRGG) 410 and its scrambled version (VSGTRHGSFHSGIGNSGIYIQNT). The peptides were 411 dissolved in DDW+2% DMSO to a working stock concentration of 100 µM and kept in 412 aliquots in -20°C. Overnight P. polymyxa cultures grown in LB were diluted 1:100 into 413 500 mL Erlenmeyer flasks with 100 mL defined medium and grown in 30°C with 200 RPM 414 shaking. At OD ~0.1, 45 mL of each culture was washed by centrifuging for 5 min in 3,200 415 g at room temperature and re-suspending the pellet in 45 mL chemically defined medium64. 416 The culture was split between 5 mL tubes containing each of the tested peptides at a final 417 concentration of 5 µM, or a control tube containing a similar amount of DDW+2% DMSO. 418 The cultures were incubated in 30°C with 200 RPM shaking for 10 min , after which 1:10 419 cold stop solution (90% (v/v) ethanol, 10% (v/v) saturated phenol) was added to the 420 samples. The samples were centrifuged for 5 min at 3200 g in 4°C, the supernatant was 421 discarded and the pellets were immediately flash frozen and stored in 80°C until RNA 422 extraction. The experiment was conducted 3 times on different days to produce 423 independent biological replicates.

424 Frozen bacterial pellets were lysed using the Fastprep homogenizer (MP Biomedicals, 425 Santa Ana, California) and RNA was extracted with the FastRNA PRO™ blue kit (MP 426 Biomedicals, 116025050) according to the manufacturer’s instructions. RNA levels and bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

427 integrity were assessed using Qubit® RNA HS Assay Kit (Life technologies, Carlsbad, 428 California, Q10210) and Tapestation (Agilent, Santa Clara, California, 5067-5576), 429 respectively. All RNA samples were treated with TURBO™ DNase (Life technologies, 430 AM2238). Ribosomal RNA depletion and RNA-seq libraries were prepared as described 431 in 66, except that all reaction volumes were reduced by a factor of 4.

432 RNA-seq libraries were sequenced using Illumina NextSeq platform, and sequenced reads 433 were demultiplexed using Illumina bcl2fastq module. Reads were mapped to the reference 434 genome of P. polymyxa ATCC 842 downloaded from IMG (IMG taxon ID: 2547132099, 435 65 contigs), as described in 66. RNA-seq-mapped reads were used to generate genome-wide 436 RNA-seq coverage maps and reads-per-gene counts.

437 Raw counts of reads-per-gene for each of the 3 biological replicates were used as input for 438 DESeq2 package analysis using R 3.6.048, while accounting for batch effect 439 (DESeqDataSet (dds) design model "design= ~batch + treatment"). Genes with normalized 440 mean count <100 were discarded and a significant adjusted p-value (FDR<0.05) and fold 441 change >2 or <0.5 were considered as the threshold for differentially expressed genes in 442 each contrast (mature vs. control and scrambled vs. control). The volcano plot in Fig. 4a 443 was plotted using EnhancedVolcano R package67, based on the results of the DESeqs2 444 analysis.

445 Clumping assay

446 Cultures were incubated with peptides as described above for the RNA-seq experiments 447 for 30 minutes, and 5 µL of culture were then dropped in the center of a dry LB plate and 448 grown overnight in 30°C. The synthetic AloP13 pro-peptide lacking the last 6 aa 449 (SYFHNITIGNGSQITVS) was ordered from Genscript Corp at a purity level of 76.6%. 450 The plate was photographed using Biorad gel imager (Gel Doc XR+) and image colors 451 were inverted for clarity.

452 Data Availability

453 All MS data (Fig. 3 and in Supplementary Table S4) have been deposited to the 454 ProteomeXchange Consortium via the PRIDE68 partner repository with the dataset bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

455 identifier PXD015319. All raw RNA-seq datasets (Fig. 4 and Supplementary Table S5) 456 were deposited in the European Nucleotide Database (ENA), study accession no. 457 PRJEB34369.

458

459 Acknowledgements

460 We thank Yael Helman, Yaara Oppenheimer-Shaanan, Ohad Herches, Alon Savidor, Etai 461 Rotem, Gilad Yaakov, Tabitha Bucher and Jürgen Bartel for their expertise and technical 462 assistance. We thank the Sorek lab members for fruitful discussions and helpful 463 suggestions. R.S. was supported, in part, by the Israel Science Foundation (personal grant 464 1360/16), the European Research Council (ERC) (grant ERC-CoG 681203), the German 465 Research Council (DFG) priority program SPP 2002 (grant SO 1611/1-1), and the Knell 466 Family Center for Microbiology. M.V. is a Clore Scholar and was supported by the Clore 467 Israel Foundation. S.M. and D.B. were supported by the German Research Council (DFG) 468 priority program SPP 2002 (grant BE 3869/5-1).

469

470 Author Contributions

471 MV designed, conducted and analyzed all experiments unless otherwise indicated. SM and 472 DB performed and analyzed the MS experiment, and TK ran the MS instrument. RS 473 overviewed the project and was involved in the study design and data analyses. MV and 474 RS wrote the paper.

475

476 Competing Interests

477 The authors declare no competing interests. bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

478 References 479 1. Genersch, E. American Foulbrood in honeybees and its causative agent, 480 Paenibacillus larvae. J. Invertebr. Pathol. 103, (2010). 481 2. Be’er, A., Strain, S. K., Hernández, R. A., Ben-Jacob, E. & Florin, E.-L. Periodic 482 reversals in Paenibacillus dendritiformis swarming. J. Bacteriol. 195, 2709–17 483 (2013). 484 3. Ingham, C. J. & Ben Jacob, E. Swarming and complex pattern formation in 485 studied by imaging and tracking cells. BMC Microbiol. 8, 36 486 (2008). 487 4. Jeong, H., Choi, S.-K., Ryu, C.-M. & Park, S.-H. Chronicle of a Soil Bacterium: 488 Paenibacillus polymyxa E681 as a Tiny Guardian of Plant and Human Health. 489 Front. Microbiol. 10, 467 (2019). 490 5. Timmusk, S. & Wagner, E. G. The plant-growth-promoting rhizobacterium 491 Paenibacillus polymyxa induces changes in Arabidopsis thaliana gene expression: 492 a possible connection between biotic and abiotic stress responses. Mol. Plant. 493 Microbe. Interact. 12, 951–9 (1999). 494 6. Weselowski, B., Nathoo, N., Eastman, A. W., MacDonald, J. & Yuan, Z. C. 495 Isolation, identification and characterization of Paenibacillus polymyxa CR1 with 496 potentials for biopesticide, biofertilization, biomass degradation and biofuel 497 production. BMC Microbiol. 16, (2016). 498 7. Choi, S. K. et al. Identification of a synthetase gene cluster of 499 Paenibacillus polymyxa and heterologous expression of the gene in Bacillus 500 subtilis. J. Bacteriol. 191, 3350–3358 (2009). 501 8. Lebuhn, M., Heulin, T. & Hartmann, A. Production of auxin and other indolic and 502 phenolic compounds by Paenibacillus polymyxa strains isolated from different 503 proximity to plant roots. FEMS Microbiol. Ecol. 22, 325–334 (2006). 504 9. Grady, E. N., MacDonald, J., Liu, L., Richman, A. & Yuan, Z.-C. Current 505 knowledge and perspectives of Paenibacillus: a review. Microb. Cell Fact. 15, 203 506 (2016). 507 10. Hong, C. E., Kwon, S. Y. & Park, J. M. Biocontrol activity of Paenibacillus 508 polymyxa AC-1 against Pseudomonas syringae and its interaction with 509 Arabidopsis thaliana. Microbiol. Res. 185, 13–21 (2016). 510 11. Luo, Y. et al. Complete genome sequence of industrial biocontrol strain 511 Paenibacillus polymyxa HY96-2 and further analysis of its biocontrol mechanism. 512 Front. Microbiol. 9, (2018). 513 12. Ng, W.-L. & Bassler, B. L. Bacterial Quorum-Sensing Network Architectures. 514 Annu. Rev. Genet. 43, 197–222 (2009). 515 13. Schuster, M. & Greenberg, E. P. A network of networks: Quorum-sensing gene 516 regulation in . Int. J. Med. Microbiol. 296, 73–81 (2006). bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

517 14. Davey, M. E., Caiazza, N. C. & O’Toole, G. A. Rhamnolipid surfactant production 518 affects biofilm architecture in Pseudomonas aeruginosa PAO1. J. Bacteriol. 185, 519 1027–36 (2003). 520 15. Brurberg, M. B., Nes, I. F. & Eijsink, V. G. Pheromone-induced production of 521 antimicrobial peptides in Lactobacillus. Mol. Microbiol. 26, 347–60 (1997). 522 16. Ploeg, J. R. Regulation of bacteriocin production in Streptococcus mutans by the 523 quorum-Sensing system required for development of genetic competence. J. 524 Bacteriol. 187, 3980–3989 (2005). 525 17. Hoover, S. E. et al. A new quorum-sensing system (TprA/PhrA) for Streptococcus 526 pneumoniae D39 that regulates a lantibiotic biosynthesis gene cluster. Mol. 527 Microbiol. 97, 229–43 (2015). 528 18. Grossman, A. D. Genetic networks controlling the initiation of sporulation and the 529 development of genetic competence in . Annu. Rev. Genet. 29, 530 477–508 (1995). 531 19. Neiditch, M. B., Capodagli, G. C., Prehna, G. & Federle, M. J. Genetic and 532 Structural Analyses of RRNPP Intercellular Peptide Signaling of Gram-Positive 533 Bacteria. Annu. Rev. Genet. 51, 311–333 (2017). 534 20. Lanigan-Gerdes, S., Dooley, A. N., Faull, K. F. & Lazazzera, B. A. Identification 535 of subtilisin, Epr and Vpr as enzymes that produce CSF, an extracellular signalling 536 peptide of Bacillus subtilis. Mol. Microbiol. 65, 1321–33 (2007). 537 21. Stephenson, S., Mueller, C., Jiang, M. & Perego, M. Molecular analysis of Phr 538 peptide processing in Bacillus subtilis. J. Bacteriol. 185, 4861–71 (2003). 539 22. Perego, M., Higgins, C. F., Pearce, S. R., Gallagher, M. P. & Hoch, J. A. The 540 oligopeptide transport system of Bacillus subtilis plays a role in the initiation of 541 sporulation. Mol. Microbiol. 5, 173–85 (1991). 542 23. Slamti, L. & Lereclus, D. The oligopeptide ABC-importers are essential 543 communication channels in Gram-positive bacteria. Res. Microbiol. (2019). 544 24. Do, H. & Kumaraswami, M. Structural Mechanisms of Peptide Recognition and 545 Allosteric Modulation of Gene Regulation by the RRNPP Family of Quorum- 546 Sensing Regulators. J. Mol. Biol. 428, 2793–804 (2016). 547 25. Burbulys, D., Trach, K. A. & Hoch, J. A. Initiation of sporulation in B. subtilis is 548 controlled by a multicomponent phosphorelay. Cell 64, 545–52 (1991). 549 26. Pottathil, M. & Lazazzera, B. A. The extracellular Phr peptide-Rap phosphatase 550 signaling circuit of Bacillus subtilis. Front. Biosci. 8, d32-45 (2003). 551 27. Slamti, L. & Lereclus, D. A cell-cell signaling peptide activates the PlcR virulence 552 regulon in bacteria of the Bacillus cereus group. EMBO J. 21, 4550–4559 (2002). 553 28. Erez, Z. et al. Communication between viruses guides lysis–lysogeny decisions. 554 Nature 541, 488–493 (2017). bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

555 29. Stokar-Avihail, A., Tal, N., Erez, Z., Lopatina, A. & Sorek, R. Widespread 556 Utilization of Peptide Communication in Phages Infecting Soil and Pathogenic 557 Bacteria. Cell Host Microbe 25, 746-755.e5 (2019). 558 30. Core, L. & Perego, M. TPR-mediated interaction of RapC with ComA inhibits 559 response regulator-DNA binding for competence development in Bacillus subtilis. 560 Mol. Microbiol. 49, 1509–22 (2003). 561 31. Jeong, H. et al. Draft genome sequence of the Paenibacillus polymyxa type strain 562 (ATCC 842T), a plant growth-promoting bacterium. J. Bacteriol. 193, 5026–7 563 (2011). 564 32. Karpenahalli, M. R., Lupas, A. N. & Söding, J. TPRpred: a tool for prediction of 565 TPR-, PPR- and SEL1-like repeats from protein sequences. BMC Bioinformatics 8, 566 2 (2007). 567 33. Fekkes, P. & Driessen, A. J. Protein targeting to the bacterial cytoplasmic 568 membrane. Microbiol. Mol. Biol. Rev. 63, 161–73 (1999). 569 34. Chen, I.-M. A. et al. IMG/M v.5.0: an integrated data management and 570 comparative analysis system for microbial genomes and microbiomes. Nucleic 571 Acids Res. 47, D666–D677 (2019). 572 35. Käll, L., Krogh, A. & Sonnhammer, E. L. L. L. A combined transmembrane 573 topology and signal peptide prediction method. J. Mol. Biol. 338, 1027–36 (2004). 574 36. Even-Tov, E., Omer Bendori, S., Pollak, S. & Eldar, A. Transient Duplication- 575 Dependent Divergence and Horizontal Transfer Underlie the Evolutionary 576 Dynamics of Bacterial Cell–Cell Signaling. PLOS Biol. 14, e2000330 (2016). 577 37. Perego, M. A peptide export-import control circuit modulating bacterial 578 development regulates protein phosphatases of the phosphorelay. Proc. Natl. Acad. 579 Sci. U. S. A. 94, 8612–8617 (1997). 580 38. Nguyen, L. T., Schmidt, H. A., Von Haeseler, A. & Minh, B. Q. IQ-TREE: A fast 581 and effective stochastic algorithm for estimating maximum-likelihood 582 phylogenies. Mol. Biol. Evol. 32, 268–274 (2015). 583 39. Hoang, D. T., Chernomor, O., Von Haeseler, A., Minh, B. Q. & Vinh, L. S. 584 UFBoot2: Improving the ultrafast bootstrap approximation. Mol. Biol. Evol. 35, 585 518–522 (2018). 586 40. Kalyaanamoorthy, S., Minh, B. Q., Wong, T. K. F., Von Haeseler, A. & Jermiin, 587 L. S. ModelFinder: Fast model selection for accurate phylogenetic estimates. Nat. 588 Methods 14, 587–589 (2017). 589 41. Letunic, I. & Bork, P. Interactive tree of life (iTOL) v3: an online tool for the 590 display and annotation of phylogenetic and other trees. Nucleic Acids Res. 44, 591 W242-5 (2016). 592 42. Lanigan-Gerdes, S., Briceno, G., Dooley, A. N., Faull, K. F. & Lazazzera, B. A. 593 Identification of residues important for cleavage of the extracellular signaling bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

594 peptide CSF of Bacillus subtilis from its precursor protein. J. Bacteriol. 190, 595 6668–75 (2008). 596 43. Dalbey, R. E., Wang, P. & van Dijl, J. M. Membrane Proteases in the Bacterial 597 Protein Secretion and Quality Control Pathway. Microbiol. Mol. Biol. Rev. 76, 598 311–330 (2012). 599 44. Koskinen, V. R., Emery, P. A., Creasy, D. M. & Cottrell, J. S. Hierarchical 600 clustering of shotgun proteomics data. Mol. Cell. Proteomics 10, (2011). 601 45. Mirouze, N., Parashar, V., Baker, M. D., Dubnau, D. A. & Neiditch, M. B. An 602 atypical Phr peptide regulates the developmental switch protein RapH. J. 603 Bacteriol. 193, 6197–6206 (2011). 604 46. Perego, M. & Hoch, J. A. Negative regulation of Bacillus subtilis sporulation by 605 the spo0E gene product. J. Bacteriol. 173, 2514–2520 (1991). 606 47. Parashar, V., Konkol, M. A., Kearns, D. B. & Neiditch, M. B. A plasmid-encoded 607 phosphatase regulates Bacillus subtilis biofilm architecture, sporulation, and 608 genetic competence. J. Bacteriol. 195, 2437–2448 (2013). 609 48. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and 610 dispersion for RNA-seq data with DESeq2. Genome Biol. 15, (2014). 611 49. Dunny, G. M., Brown, B. L. & Clewell, D. B. Induced cell aggregation and mating 612 in Streptococcus faecalis: evidence for a bacterial sex pheromone. Proc. Natl. 613 Acad. Sci. U. S. A. 75, 3479–3483 (1978). 614 50. Havarstein, L. S., Martin, B., Johnsborg, O., Granadel, C. & Claverys, J.-P. New 615 insights into the pneumococcal fratricide: relationship to clumping and 616 identification of a novel immunity factor. Mol. Microbiol. 59, 1297–1037 (2006). 617 51. Guiral, S., Mitchell, T. J., Martin, B. & Claverys, J. P. Competence-programmed 618 predation of noncompetent cells in the human pathogen Streptococcus 619 pneumoniae: Genetic requirements. Proc. Natl. Acad. Sci. U. S. A. 102, 8710–8715 620 (2005). 621 52. Eastman, A. W., Heinrichs, D. E. & Yuan, Z. C. Comparative and genetic analysis 622 of the four sequenced Paenibacillus polymyxa genomes reveals a diverse 623 metabolism and conservation of genes relevant to plant-growth promotion and 624 competitiveness. BMC Genomics 15, 1–22 (2014). 625 53. Qian, Y., Kando, C. K., Thorsen, L., Larsen, N. & Jespersen, L. Production of 626 autoinducer-2 by aerobic endospore-forming bacteria isolated from the West 627 African fermented foods. FEMS Microbiol. Lett. 362, 1–7 (2015). 628 54. Huang, E. & Yousef, A. E. Biosynthesis of paenibacillin, a lantibiotic with N- 629 terminal acetylation, by Paenibacillus polymyxa. Microbiol. Res. 181, 15–21 630 (2015). 631 55. Kleerebezem, M., Quadri, L. E., Kuipers, O. P. & de Vos, W. M. Quorum sensing 632 by peptide pheromones and two-component signal-transduction systems in Gram- bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

633 positive bacteria. Mol. Microbiol. 24, 895–904 (1997). 634 56. Dodd, H. M. et al. Molecular analysis of the regulation of nisin immunity. 635 Microbiology 142, 2385–2392 (1996). 636 57. Timmusk, S., Grantcharova, N. & Wagner, E. G. H. Paenibacillus polymyxa 637 invades plant roots and forms . Appl. Environ. Microbiol. 71, 7292–7300 638 (2005). 639 58. Timmusk, S. et al. Sfp-type PPTase inactivation promotes bacterial biofilm 640 formation and ability to enhance wheat drought tolerance. Front. Microbiol. 6, 641 (2015). 642 59. Kim, S.-B. & Timmusk, S. A Simplified Method for Gene Knockout and Direct 643 Screening of Recombinant Clones for Application in Paenibacillus polymyxa. 644 PLoS One 8, e68092 (2013). 645 60. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local 646 alignment search tool. J. Mol. Biol. 215, 403–410 (1990). 647 61. Gasteiger, E. et al. ExPASy: The proteomics server for in-depth protein knowledge 648 and analysis. Nucleic Acids Res. 31, 3784–8 (2003). 649 62. Steinegger, M. & Söding, J. MMseqs2 enables sensitive protein sequence 650 searching for the analysis of massive data sets. Nat. Biotechnol. 35, 1026–1028 651 (2017). 652 63. Katoh, K., Rozewicki, J. & Yamada, K. D. MAFFT online service: multiple 653 sequence alignment, interactive sequence choice and visualization. Brief. 654 Bioinform. (2017). 655 64. Agata, N., Ohta, M., Mori, M. & Shibayama, K. Growth conditions of and emetic 656 toxin production by Bacillus cereus in a defined medium with amino acids. 657 Microbiol. Immunol. 43, 15–8 (1999). 658 65. Cox, J. et al. Accurate proteome-wide label-free quantification by delayed 659 normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell. 660 Proteomics 13, 2513–26 (2014). 661 66. Dar, D. et al. Term-seq reveals abundant ribo-regulation of resistance in 662 bacteria. Science 352, aad9822 (2016). 663 67. Blighe, K. Publication-ready volcano plots with enhanced colouring and labeling. 664 R-Package version 1.2.0, https://github.com/kevinblighe/EnhancedVolcano 665 (2019). 666 68. Vizcaíno, J. A. et al. 2016 update of the PRIDE database and its related tools. 667 Nucleic Acids Res. 44, D447-56 (2016). 668 bioRxiv preprint doi: https://doi.org/10.1101/767517; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

669 Supplementary Material

670 Table S1: AloR genes and corresponding AloP sequences identified in P. polymyxa 671 strains 672 Table S2: AloR homologs in microbial genomes 673 Table S3: Alo systems in P. terrae NRRL B-30644 674 Table S4: All peptide fragments detected in mass spectrometry analysis 675 Table S5: Differential gene expression analysis at 10 minutes peptide incubation 676 Video S1: P. polymyxa clumping upon exposure to AloP13 pro-peptide