Solo acylhomoserine lactone synthase from predatory myxobacterium suggests beneficial participation in interspecies cross talk

The prototypical intraspecies quorum signaling systems mediated by acylhomoserine lactones are abundant in proteobacteria, and considerable efforts have provided insight into the regulated physiological features impacted by such systems. However, the high occurrence of orphaned AHL receptors present in bacterial species that do not produce cognate AHL signals suggests the involvement of AHL signals in interspecies interactions within polymicrobial communities. The specific benefits of these interactions are mostly unknown. Considered a key taxon in microbial communities, myxobacteria exist as coordinated swarms that utilize an excreted combination of lytic enzymes and specialized metabolites to facilitate predation of numerous microbial phyla. Of all the biosynthetic gene clusters associated with myxobacteria deposited in the antiSMASH database, only one putative acylhomoserine lactone synthase, agpI, was observed in genome data from the myxobacterium Archangium gephyra. Without a cognate AHL receptor, we consider AgpI an orphaned AHL synthase. Herein we report the bioinformatic assessment of AgpI and discovery of a second myxobacterial AHL synthase from Vitiosangium sp. strain GDMCC 1.1324. Heterologous expression of each synthase in Escherichia coli provided detectible quantities of 3 AHL signals including 2 known AHLs, C8-AHL and C9-AHL. The functional, orphaned AHL synthase, AgpI, from the predatory myxobacterium A. gephyra provides unique support for beneficial interspecies crosstalk within polymicrobial communities. Importance The presence of orphaned quorum signal receptors and associated recognition and response to acylhomoserine lactone quorum signals provides evidence for small molecule-mediated interspecies interactions about microbial communities. A solo signal synthase from a predatory myxobacterium provides an alternative perspective on the evolution and benefits of quorum signaling systems within these communities. Ultimately our results support and supplement the hypothetical benefits of interspecies cross talk within diverse microbial communities.

. No features within the proteome of A. gephyra were sufficiently homologous to 1 be considered an autoinducer binding. We next queried the associated Hidden Markov 2 Model (HMM) associated with autoinducer binding domains deposited in Pfam against 3 the proteome of A. gephyra using HMMSEARCH (supplemental data) (45,46). The most 4 significant hit (E-value 0.0015) a PAS domain S-box-containing protein also annotated as 5 a GAF-domain-containing protein (WP_053066299.1) does not include significant 6 sequence homology with LuxR-type, AHL receptors. Interestingly, similar analysis of V. 7 sp. GDMCC 1.1324 provided a highly homologous LuxR-type receptor 8 (WP_108076247.1). While the AHL receptor identified in the genome of V. sp. is not 9 clustered near vitI as is typical of LuxI-LuxR type synthase-receptor pairs, we cannot 10 assume both are unpaired orphans and instead consider VitI might not be a truly solo 11 AHL synthase. From these data we determined AgpI to be an orphaned AHL synthase 12 without any cognate AHL receptor present in the genome of A. gephyra. 13 A. gephyra does not produce AHLs during axenic cultivation. 14 Cultivation of A. gephyra on VY/2 agar plates at 30°C for 21 days provided fully 15 developed, wispy myxobacterial swarms encompassing the entirety of the plate surface. 16 Homogenized agar and cellular contents were extracted using traditional organic phase 17 techniques to provide extracts for LC-MS/MS analysis. The resulting datasets from LC-18 MS/MS analysis of A. gephyra extracts were analyzed against datasets generated from 19 analytical standards for a variety of AHLs including C6-AHL, 3-oxo-C6-AHL, C8-AHL, and 20 C11-AHL to determine the presence of any produced AHL-like metabolites. Data from 21 resulting mass spectra were scrutinized using the Global Natural Products Social 22 Molecular Networking (GNPS) platform to generate molecular networks depicting 23 similarities in detected metabolite scaffolds inferred from ionized fragment commonalities 1 (47). No metabolites that included the diagnostic AHL-associated fragments at 102.0547 2 m/z and 74.0599 m/z associated with the core homoserine lactone moiety were detected 3 in extracts from A. gephyra (48,49). This data supports any one of the following 4 conclusions A. gephyra does not produce AHL-like metabolites when grown axenically 5 but may be active under other growth conditions; metabolites produced by AgpI do not 6 possess structural similarity with typical AHL metabolites; or AgpI is simply nonfunctional. 7

Heterologous expression of AgpI confirms functional production of AHLs. 8
To explore the functionality of both AgpI and VitI and assumed biosynthesis of AHL-like 9 metabolites, inducible codon-optimized constructs of agpI and vitI included in replicating 10 plasmids suitable for expression in Escherichia coli were purchased. Heterologous 11 expression of AgpI and VitI and subsequent extraction, LC-MS/MS analysis, and 12 evaluation of molecular networks rendered by GNPS as previously described, provided a 13 cluster family including 2 of 3 total nodes identified as C8-AHL (228.159 m/z) and C9-14 AHL (242.174 m/z) from internal GNPS public datasets as well as a third AHL metabolite 15 detected at 226.144 m/z ( Figure 4) (47). This cluster family was identical in both 16 heterologous expression experiments suggesting that AgpI and VitI produce the same 3 17 AHL metabolites with similar detected intensities for each AHL. Both C8-AHL and C9-18 AHL were confirmed to be present in AgpI and VitI extracts using analytical standards. 19 Based on associated intensities, C8-AHL was the most abundant and the metabolite 20 detected at 226.144 m/z was the least abundant AHL. No AHL-like entities were detected 21 in control extracts from E. coli containing an empty pET28b expression plasmid. From the 22 mass difference between C8-AHL and the unknown AHL detected at 226.144 m/z (2.015 Da measured vs. 2.01565 theoretical), as well as shared fragmentation patterns, we 1 determined the metabolite detected at 226.144 m/z was likely an unsaturated analog of 2 C8-AHL ( Figure 5). From these experiments we determined that both AgpI and VitI are 3 functional AHL synthases capable of producing the previously characterized AHLs C8-4 AHL and C9-AHL. These results also suggest A. gephyra produces these AHLs and likely 5 requires environmental cues or specific nutrients not present during our axenic cultivation 6 conditions. 7

Discussion 8
Ultimately we conclude that the myxobacteria A. gephyra and V. sp. possess functional 9 AHL synthases that produce the AHL signals C8-AHL and C9-AHL when heterologously 10 expressed in E. coli. Considering the strong precedent for heterologous expression of 11 AHL synthases in E. coli to determine produced AHL metabolites, we suggest that both 12 A. gephyra and V. sp. capably produce one or all of the observed AHL signals and that 13 AgpI is merely silent or cryptic during axenic cultivation of A. gephyra (50-55). However, 14 we should also consider that these synthases could instead utilize an acyl-ACP precursor 15 not available to the heterologous E. coli host, and we are actively exploring cultivation 16 conditions that might induce native AHL production from A. gephyra (54, 55). While 17 numerous bacteria have been observed to possess orphaned LuxR-type AHL receptors, 18 production of AHL metabolites from a solo AHL synthase without any cognate AHL 19 receptor with homology to LuxR also present in the genome of A. gephyra is the first to 20 be reported (19,21,23,56). Although a functional orphaned LuxI-type synthase capable 21 of producing AHLs has been reported from the sponge symbiont Ruegeria sp. KLH11, 22 the strain also harbors 2 pairs of clustered LuxI/LuxR homologues (23, 57). We suggest that production of quorum signals by myxobacteria supports the theoretical benefits of 1 interspecies cross talk similar to functional, solo AHL receptors (21, 58-60). We also 2 propose that the more typical abundance of orphan AHL receptors reported from a variety 3 of bacterial species compared to the seemingly exceptional solo AHL synthase reported 4 here might correlate with the rarity of bacteriovorus micropredators (19,27). The absence 5 of any AHL metabolites during axenic cultivation of A. gephyra suggests an unknown 6 regulatory mechanism independent of the typical LuxR receptor to be involved. However, 7 previously reported eavesdropping by M. xanthus and response to exogenous AHLs 8 despite the absence of any AHL receptor with homology to LuxR suggests myxobacteria 9 may possess an undiscovered, alternative means of AHL detection (17). While the benefit 10 afforded predatory myxobacteria remains unclear, production of AHL signals known to 11 regulate QS-associated physiological functions such as biofilm formation, specialized 12 metabolism, and motility offers some insight (15). Predatory disruption of any one of these 13 functions would likely improve predation of quorum signaling prey. We consider these 14 observations provide a unique perspective and support the continued investigation of 15 small molecule interactions that contribute to microbial community structures and trophic 16 levels. 17

Cultivation of A. gephyra. Archangium gephyra (DSM 2261) initially obtained from 19
German Collection of Microorganisms in Braunschweig was grown on VY/2 agar (5 g/L 20 baker's yeast, 1.36 g/L CaCl2, 0.5 mg/L vitamin B12, 15 g/L agar, pH 7.2). 21 Bioinformatic assessment of AgpI. The amino acid sequence for AgpI 22 (WP_047862734.1) was submitted for blastp analysis and EFI-EST analysis (https://efi.igb.illinois.edu/efi-est/) using the default settings. Results from EFI-EST 1 analysis were visualized using Cytoscape and are provided as supplemental data. 2 Alignments from ClustalW and minimum evolution phylogenetic trees were rendered 3 using MEGA7 (61, 62). 4 Autoinducer binding site search. All 3,014 domains annotated as autoinducer binding 5 domains (PF03472) deposited in Pfam were subjected to blastp analysis against the A. 6 gephyra genome (NZ_CP011509.1). For HMMSEARCH analysis, the raw HHM for 7 autoinducer binding domains was downloaded from Pfam (PF03472) and utilized as input 8 for profile-HMM vs protein sequence database via HMMSEARCH with the taxonomy 9 restrictions set to limit analysis to A. gephyra or V. sp. Results from this analysis are 10 provided as supplemental data. 11

Heterologous expression of AgpI and VitI in E. coli. Constructs of AgpI and VitI codon 12
optimized for expression in E. coli situated in pET28b were purchased from Genscript 13 (Piscataway, NJ). Sequence data for these constructs are provided as supplemental data. 14 Heterologous host E. coli K207-3 was grown at 37°C in LB broth supplemented with 15 50µg/mL kanamycin, induced with 1µM IPTG at OD600=0.6, and grown overnight at 14°C 16 to facilitate heterologous protein expression. controlled with Xcalibur version 2.0.7 and coupled to a Dionex Ultimate 3000 nanoUHPLC 1 system. Samples were loaded onto a PepMap 100 C18 column (0.3 mm × 150 mm, 2 2 μm, Thermo Fisher Scientific). Separation of the samples was performed using mobile 3 phase A (0.1% formic acid in water) and mobile phase B (0.1% formic acid in acetonitrile) 4 at a rate of 6 μL/min. The samples were eluted with a gradient consisting of 5 to 60% 5 solvent B over 15 min, ramped to 95 % B over 2 min, held for 3 min, and then returned to 6 5% B over 3 min and held for 8 min. All data were acquired in positive ion mode. Collision-7 induced dissociation (CID) was used to fragment molecules, with an isolation width of 3 8 m/z units. The spray voltage was set to 3600 volts, and the temperature of the heated 9 capillary was set to 300 °C. In CID mode, full MS scans were acquired from m/z 150 to 10 1200 followed by eight subsequent MS2 scans on the top eight most abundant peaks. 11 The orbitrap resolution for both the MS1 and MS2 scans was 120000. The expected mass 12 accuracy was <3 ppm. 13 GNPS dataset. Generated data were converted to .mzXML files using MS-Convert and 14 mass spectrometry molecular networks were generated using the GNPS platform 15 (http://gnps.ucsd.edu) (47). The corresponding Cytoscape file is provided as 16 supplemental information. LC-MS/MS data for this analysis were also deposited in the 17

Acknowledgements. 19
The authors appreciate funding and support from the National Institute of Allergy and 20