A Mesorhizobium japonicum quorum sensing circuit that involves three linked genes and an unusual acyl-homoserine lactone signal

ABSTRACT Members of the genus Mesorhizobium, which are core components of the rhizosphere and specific symbionts of legume plants, possess genes for acyl-homoserine lactone (AHL) quorum sensing (QS). Here we show Mesorhizobium japonicum MAFF 303099 (formerly M. loti) synthesizes and responds to N-[(2E, 4E)-2,4-dodecadienoyl] homoserine lactone (2E, 4E-C12:2-HSL). We show that the 2E, 4E-C12:2-HSL QS circuit involves one of four luxR-luxI-type genes found in the sequenced genome of MAFF 303099. We refer to this circuit, which appears to be conserved among Mesorhizobium species, as R1-I1. We show that two other Mesorhizobium strains also produce 2E, 4E-C12:2-HSL. The 2E, 4E-C12:2-HSL is unique among known AHLs in its arrangement of two trans double bonds. The R1 response to 2E, 4E-C12:2-HSL is extremely selective in comparison with other LuxR homologs, and the trans double bonds appear critical for R1 signal recognition. Most well-studied LuxI-like proteins use S-adenosylmethionine and an acyl-acyl carrier protein as substrates for synthesis of AHLs. Others that form a subgroup of LuxI-type proteins use acyl-coenzyme A substrates rather than acyl-acyl carrier proteins. I1 clusters with the acyl-coenzyme A-type AHL synthases. We show that a gene linked to the I1 AHL synthase is involved in the production of the QS signal. The discovery of the unique I1 product enforces the view that further study of acyl-coenzyme A-dependent LuxI homologs will expand our knowledge of AHL diversity. The involvement of an additional enzyme in AHL generation leads us to consider this system a three-component QS circuit. IMPORTANCE We report a Mesorhizobium japonicum quorum sensing (QS) system involving a novel acyl-homoserine lactone (AHL) signal. This system is known to be involved in root nodule symbiosis with host plants. The chemistry of the newly described QS signal indicated that there may be a dedicated cellular enzyme involved in its synthesis in addition to the types known for production of other AHLs. Indeed, we report that an additional gene is required for synthesis of the unique signal, and we propose that this is a three-component QS circuit as opposed to the canonical two-component AHL QS circuits. The signaling system is exquisitely selective. The selectivity may be important when this species resides in the complex microbial communities around host plants and may make this system useful in various synthetic biology applications of QS circuits.

We sought to learn about AHL QS in Mesorhizobium japonicum (formerly Mesorhi zobium loti [16]) by studying a strain called MAFF 303099, which is a symbiont of the model legume Lotus japonicus. This symbiotic relationship has been used to dissect plant-microbe interactions (17,18), and several useful resources exist including the sequenced genomes of both host and symbiont (19,20) and mutant collections (21,22). The M. japonicum MAFF 303099 genome possesses four luxI and luxR homologs. We are unaware of any studies of QS in this strain, but there has been work on QS in other strains and species of Mesorhizobium. Homologs of the MAFF 30399 luxI and luxR genes have been given different names (9,11,23). As a matter of convenience, we refer to the MAFF 303099 QS genes simply as I1-I4 and R1-R4. Table S1 provides information on the different luxI and luxR homologs from Mesorhizobium species studied to date and their homologies to the MAFF 303099 QS genes.
Information on AHLs produced and detected by different Mesorhizobium species and strains is fragmented. AHL synthase activities were sometimes studied by using recombinant Escherichia coli and sometimes by using analytical methods, which can misidentify minor AHLs as the primary AHL signal or not detect abundant AHLs. This can be problematic in the case of some LuxI homologs, which use acyl-coenzyme As (acyl-CoAs) as a substrate for AHL synthesis. Over the last several years, we have learned that there are two subgroups of LuxI homologs, those that use S-adenosylmethionine and fatty acyl-acyl carrier proteins (ACPs) from fatty acid biosynthesis as substrates and those that use S-adenosylmethionine and acyl-CoAs from a variety of different cellular metabolic pathways as substrates. Members of the CoA-utilizing subgroup can be sorted from the acyl-ACP subgroup bioinformatically (24). Because the generation of CoA substrates often requires specific acyl-CoA ligases, results with recombinant E. coli can lead to incorrect identification of the natural signal (24,25). The CoA-dependent AHL synthases are often found in the Alphaproteobacteria (24). The first one to be descri bed is produced by a purple photosynthetic bacterium Rhodopseudomonas palustris, which requires exogenous p-coumarate and a p-coumaroyl-CoA ligase to produce its QS signal p-coumaroyl-HSL (25). Other examples include Prosthecomicrobium hirschii, which uses a CoA-dependent AHL synthase to produce phenylacetyl-HSL (24). Bradyrhizobium japonicum uses a CoA-type AHL synthase to produce isovaleryl-HSL (7), and Bradyrhi zobium strain ORS278 uses its CoA-type AHL synthase to produce cinnamoyl-HSL (26). Two of the M. japonicum I homologs (I1 and I4) sort as CoA-dependent types. It would not be surprising to find these AHL synthases produce yet undescribed AHLs in their native background.
Here we describe our use of a radiotracer analysis of AHLs produced by M. japoni cum MAFF 303099. The radiotracer technique provides information about the relative abundances of AHLs produced under the conditions of the experiment (27). There is one major AHL produced, and it has unique features. We determine which of the four I proteins is responsible for production of this AHL, and we show that the cognate R protein responds selectively to the AHL we identified. We discovered that a gene immediately downstream of, and presumably co-transcribed with, I1 was involved in the synthesis of the unusual AHL. Mining of genome sequence databases indicates that this R-I pair (R1-I1) is conserved in most sequenced Mesorhizobium genomes. This finding suggests that it may be particularly critical to the success of this bacterium in the rhizosphere or in the process of root nodulation.

Production of one major AHL by M. japonicum MAFF 303099
Two common methods to identify AHLs are relaxed-specificity bioassays and mass spectrometry (MS). Bioassays can miss novel AHLs (7), and they can greatly overestimate the relative abundance of minor AHLs, leading to signal misidentification (27). Mass spectrometric approaches for AHL discovery generally rely on the identification of a diagnostic MS2 ion with a mass of 102 (11,28,29), corresponding to the conserved homoserine lactone ring. Mass spectra of most, but not all, AHLs have this MS2 fragment. Novel AHLs that do not exhibit this fragment can be missed by MS. Therefore, we used a 14 C-AHL radiotracer assay (27,30) that detects all AHLs synthesized by cells incubated in the presence of 14 C-methonine regardless of the AHL structure, and it provides informa tion about the relative abundances of the AHLs (7,24). Logarithmic-phase cells were suspended in a buffer with glucose as an energy source and L-[1-14 C]-methionine. The 14 C in this position will be incorporated into the homoserine lactone ring of AHLs synthe sized by a LuxI homolog. After incubation with L-[1-14 C]-methionine, cells were removed by centrifugation, and hydrophobic compounds were extracted from the culture fluid with ethyl acetate. Extracts were fractioned by C 18 -reverse phase high-pressure liquid chromatography (HPLC) (Fig. 1A). Most of the radiolabel was eluted in a single fraction (Fig. 1A). To obtain additional evidence that the radiolabeled material was an AHL, we treated an extract with AiiA lactonase (Fig. 1A), which cleaves the homoserine lactone ring of AHL compounds (31,32). The lactonase treatment essentially eliminated the radioactive peak (Fig. 1A).
Often this type of analysis can lead to a presumptive identification of the radioactive AHL based on HPLC retention time. We were unable to make such an assignment. The two standards that elute most closely to the peak at fraction 64 are 3-oxo-dodecanoyl homoserine lactone (3-oxo-C 12 -HSL) and dodecanoyl homoserine lactone (C 12 -HSL), which elute earlier or later than the radiolabeled material (Fig. 1A). A previous report indicated that C 12 -HSL was the product of a gene homologous to I1 in M. loti NZP 2213 and that one of the other LuxI homologs (I3) directed NZP 2213 to produce shorter chain AHLs (Table S1) (11). We found a very small radioactive peak at the position where C 12 -HSL elutes in our HPLC analysis (Fig. 1A). It might be that the long-chain AHL detected previously was misidentified as C 12 -HSL or that C 12 -HSL is a minor product and a more abundant AHL was overlooked or that although the amino acid sequence of the AHL synthase from the other strain is 95% identical to I1, its reaction product is different than that of I1. We address this issue in the section below.
We focused our attention on the MAFF 303099 I1 gene because it showed homology with the NZP 2213 luxI gene reported to synthesize C 12 -HSL, and the NZP 2213 I1 is required for normal Lotus root nodulation (11). The genome of another Mesorhizobium sp., strain AP09 (36), possesses a single luxR-luxI gene pair, which is homologous to R1-I1 (the I gene product shows 99% identity with the MAFF 303099 I1). We performed a radiotracer experiment with this strain and found that the radioactivity was eluted in the same position as the material found in the analysis of MAFF 303099 (Fig. 1A). This is an indication that the radioactive peak is the product of I1, which is a member of the acyl-CoA subgroup of LuxI homologs (see Materials and Methods).

The M. japonicum MAFF 303099 I1 together with the R1 positively autoregu lates the transcription of I1
In the great majority of LuxR-LuxI-like AHL regulatory circuits, one of the regulated genes is the I gene itself, resulting in AHL synthesis being positively autoregulated (37). In fact, Yang et al. (11) provided evidence to support the idea that the NZP 2213 I1 homolog is positively autoregulated, and we identified a potential R1 DNA-binding sequence upstream of the MAFF 303099 I1 gene (Fig. 1B). We constructed a transcription reporter plasmid (pZS1) containing both R1 and the I1 promoter fused to a promoterless mCherry and introduced this plasmid into P. putida, a heterologous host without its own AHL QS circuit. We have used P. putida to study QS circuits from other bacteria with high GC content genomes (24). When P. putida (pZS1) was grown in the presence of MAFF 303099 culture fluid extracts, mCherry fluorescence was about eight times that of P. putida grown in the absence of MAFF 303099 culture fluid extracts. Furthermore, culture fluid extracts from a MAFF 303099 I1 transposon insertion mutant did not stimulate mCherry expres sion (Fig. 1C). This provides further evidence that I1 is responsible for the synthesis of the AHL produced by cultures of M. japonicum.
To test our hypothesis that the R1 protein was required for a response to the M. japonicum AHL, we constructed pZS2 by deletion of 578 bp of the 732 bp R1 open Research Article mBio reading frame. Addition of MAFF 303099 culture fluid extracts to P. putida (pZS2) did not affect mCherry expression (Fig. 1C). We conclude that I1 is responsible for the production of the AHL, and R1 is a transcription factor that is responsive to the I1-produced AHL.
With the development of an assay for the undefined AHL (Fig. 1C), we then moved forward to purify and elucidate the AHL structure.

Purification and structural determination of the M. japonicum MAFF 303099produced AHL
We used the P. putida (pZS1) response to MAFF 303099 culture fluid extracts for activityguided purification of the AHL. We first showed that the HPLC profile of the bioactive material was similar to that of the radioactive material, and indeed it was eluted in fractions 64 and 65 ( Fig. 2A), whereas the radioactive material was eluted in fraction 64 (Fig. 1A). We then extracted the AHL from 5 L of MAFF 303099 culture fluid with ethyl acetate and subjected the extract to gradient HPLC. Each HPLC fraction was tested for bioactivity by using P. putida (pZS1), and a single peak was eluted in fractions 64 (28,29). Interestingly, such MS2 fragments were not evident in our spectra ( Fig. 2B; Table S2).
To determine the location and stereochemistry of the predicted double bonds in the purified AHL, we performed 1 H, 13 C, and various two-dimensional nuclear magnetic resonance (NMR) analyses (Table S3; Fig. S1 through S5). To improve the resolution for 13 C-NMR analysis, we extracted and purified the AHL from an additional 2 L of the supernatant fluid from cultures grown in a minimal medium with uniformly labeled 13 C-glucose as the sole carbon source (see Materials and Methods). NMR analyses revealed a conjugated system beginning with the carbonyl at the C1 acyl position and trans olefins at the C2 (J in Hz 15) and C4 (J in Hz 15.5) positions, resulting in our determination that the purified compound was N-[(2E, 4E)-2,4-dodecadienoyl] homoserine lactone (2E, 4E-C 12:2 -HSL, Fig. 2C). This molecule was chemically synthesized previously (38) and provided to us for comparison with our purified AHL by Stefan Schultz. The chemically synthesized 2E, 4E-C 12:2 -HSL was active in the P. putida (pZS1) bioassay and had an HPLC elution profile similar to the natural product ( Fig. 2A). Furthermore, the LC-MS/MS profiles of the purified and synthetic AHLs were indistin guishable ( Fig. 2B; Table S2). Because 2E, 4E-C 12:2 -HSL has two conjugated double bonds it should have some UV light absorbance and in fact the chemically synthesized and natural products had similar absorption spectra (two peaks with maximum absorbances at about 210 and 259 nm, Fig. 2D). We conclude that M. japonicum I1 catalyzes the synthesis of 2E, 4E-C 12:2 -HSL.

M. japonicum MAFF 303099 produces 2E, 4E-C 12:2 -HSL in late logarithmic growth phase
To make quantitative determinations of the amount of 2E, 4E-C 12:2 -HSL produced by MAFF 303099, we used the chemically synthesized 2E, 4E-C 12:2 -HSL to create a standard curve based on activity in the P. putida (pZS1) assay (Fig. 3A). The maximum amount of 2E, 4E-C 12:2 -HSL present in culture fluid was about 1.5 µM, a concentration typical for AHL-producing bacteria. Levels of 2E, 4E-C 12:2 -HSL were low in early-and mid-loga rithmic growth phase and then increased rapidly during late logarithmic growth and the transition to the stationary phase (Fig. 3B). This profile is characteristic of positively autoregulated AHL QS systems (39,40). Note 2E, 4E-C 12:2 -HSL levels decreased later in the stationary phase (Fig. 3B). AHLs are unstable at alkaline pH (41); however, the pH of our MAFF 303099 cultures remained at 7. An alternative explanation for the AHL decrease is that MAFF 303099 produces a specific AHL-degrading enzyme(s) in the stationary phase as has been reported for other bacteria (42).

Research Article mBio
As discussed above, Yang et al. showed the NZP 2213 I1 homolog was important for root nodulation, and they presented evidence to suggest the AHL product of I1 was C 12 -HSL (11). The C 12 -HSL assignment was based on a relaxed-specificity bioassay-cou pled thin layer chromatography procedure and MS analysis of compounds with a MS2 fragment of 102 (11). The apparent differences between our findings and those of Yang et al. (11) might be that we have used different Mesorhizobium species; however, the analytical techniques employed previously (11) would have been blind to 2E, 4E-C 12:2 -HSL. We extracted ethyl acetate AHLs from culture fluids of M. loti NZP 2213 as the cells entered the stationary phase and used HPLC to fractionate the extracts. By using our 2E, 4E-C 12:2 -HSL-specific bioassay, we found activity, which was eluted in the position of 2E, 4E-C 12:2 -HSL, and we did not detect any activity with a C 12 -HSL bioassay (Fig. 4). We conclude that the most abundant AHL in the NZP 2213 extracts was 2E, 4E-C 12:2 -HSL. If C 12 -HSL was produced it was below the limit of detection.

The gene adjacent to I1 is involved in 2E, 4E-C 12:2 -HSL production
We believe that the I1 AHL synthase uses S-adenosylmethionine and 2E, 4E-C 12:2 -CoA as reaction substrates. It is not clear as to how the CoA substrate might result from normal cellular metabolism. Our attention was drawn to the gene adjacent to I1. This gene, mlr5639, is separated from I1 by eight base pairs and oriented in the same direction as I1 (Fig. 5A). It seems likely that the two genes are co-transcribed. This gene encodes a polypeptide annotated as a member of the crotonase superfamily (Fig. 5A). Members of Research Article mBio the crotonase superfamily are enzymes that catalyze diverse reactions but share a requirement to stabilize enolate anion intermediates derived from acyl-CoA substrates [reviewed in (44)]. Based on current knowledge about AHL synthases, might the downstream gene be involved together with I1 in 2E, 4E-C 12:2 -HSL production? To address this question, we constructed a mlr5639 deletion mutant and found that it produced little 2E, 4E-C 12:2 -HSL as measured by either 14 C-radiolabeling (Fig. 5B) or the P. putida (pZS1) bioassay (Fig. 5C). The mlr5639 mutation was complemented by providing an intact copy of mlr5639 on a plasmid (pZS3, Fig. 5C). We note that even though we used the I1 promoter to drive mlr5639 expression, cells containing this plasmid grew slowly compared with vector control cells.

The R1-I1-crotonase superfamily gene three-gene element occurs in some other genera of Alphaproteobacteria
The R1-I1 circuit is quite conserved in the genus Mesorhizobium, as homologs with high identity to R1-I1 occur in approximately 90% of the nearly 500 Mesorhizobium genome sequences deposited in the Joint Genome Institute Integrated Microbial Genomes (JGI IMG) repository (46). By using the JGI IMG "top homolog" search feature, we identified I1 homologs (50% amino acid identity or higher) in two genera in addition to Mesorhi zobium. We found genes homologous to the I1-linked crotonase superfamily gene adjacent to I1 (Fig. 5A). Thus, it seems as though this three-gene AHL QS circuit is not restricted to the genus Mesorhizobium. It is of some interest to determine whether these additional bacteria produce an AHL with trans unsaturated double bonds, perhaps 2E, 4E-C 12:2 -HSL itself.
Research Article mBio production have not yet been identified (54). With our discovery that a plant-associated bacterium produces 2E, 4E-C 12:2 -HSL, it is of some interest to determine the isomeric arrangement of the two double bonds in the acyl side chain of the fecal AHL and to identify the bacteria responsible for its production. As discussed above, long-chain AHLs with unsaturated carbons in the acyl tail are common in Alphaproteobacteria. This may provide a clue as to the identity of the species responsible for the 3-oxo-C 12:2 -HSL found in fecal samples. For most AHLs, high-resolution LC-MS/MS shows characteristic homoserine lactone fragments. These fragments were not evident for 2E, 4E-C 12:2 -HSL ( Fig. 2B; Table S2) with the quadrupole time-of-flight (QTOF) MS system we employed. This is also true of some other AHLs with an unsaturation between carbons 2 and 3 (25,26,50), whereas other AHLs with the same unsaturations do show the characteristic 102 fragment (38,51). We note this because LC-MS/MS often relies on identification of characteristic AHL fragments for discovery (11,28,29). Obviously, if we relied on this approach, we would not have found 2E, 4E-C 12:2 -HSL in M. japonicum MAFF 303099 cultures. As an example of the difficulty in relying on MS2 fragmentation approaches, Yang et al. (11) did not detect 2E, 4E-C 12:2 -HSL in M. loti NZP 2213 culture extracts, whereas we did (Fig. 4). Instead they identified C 12 -HSL as the NZP 2213 I1 product (11) likely due to using an AHL reporter sensitive to trace amounts of C 12 -HSL (14) and a reliance on an MS2 102 fragment in MS experiments. We note that Yang et al. (11) found that a M. loti NZP 2213 I1 mutant formed fewer nodules on the roots of L. japonicus than the wild type. Thus, our findings implicate 2E, 4E-C 12:2 -HSL in root nodulation.
We showed that the M. japonicum MAFF 303099 R1 has a robust response to 2E, 4E-C 12:2 -HSL. It responds to 2E, 4E-C 12:2 -HSL concentrations as low as 1 nM (Fig. 3A), but it shows little or no response to the panel of other AHLs we tested. LuxR homologs have varying degrees of specificity. Some respond to a wide variety of AHLs, and others have a more restricted selectivity (43). The R1 response is at the extreme end of the LuxR homolog selectivity spectrum. This selectivity could provide protection against inappropriate responses to AHLs produced by other bacterial species in the rhizosphere, whereas bacteria with more promiscuous LuxR homologs might engage in AHL cross talk (43). It is important to mention that the extreme selectivity of the R1-I1 circuit could be useful in a variety of synthetic biology applications where cross talk can be an obstacle (55).
The unusual nature of the fatty acyl moiety of 2E, 4E-C 12:2 -HSL raised a question as to how might the CoA substrate for the I1 enzyme be produced. Several other CoA-depend ent AHL synthases rely on normal cellular metabolites and CoA ligases to manufacture substrates such as phenylacetyl-CoA or isovaleryl-CoA (24). One interesting case is that of R. palustris, which relies on the exogenous addition of p-coumaric acid and cellular p-coumaroyl-CoA ligase activity to produce p-coumaroyl-HSL (25). Here we report a new twist on the theme. There is a gene adjacent to I1 that we show is critical to I1-dependent 2E, 4E-C 12:2 -HSL synthesis. The product of this gene has been annotated as a member of the crotonase enzyme superfamily. We hypothesize that the gene product is needed for the synthesis of the CoA substrate for the I1 gene product. We further hypothesize that the substrate is 2E, 4E-C 12:2 -CoA. Further biochemical studies are required to establish the role of the downstream gene in AHL production. We note that whole-genome sequencing has revealed similar gene arrangements in a few other bacterial species. This also merits further investigation. Regardless, we have shown that the basic QS circuit in M. japonicum MAFF 303099 involves three genes dedicated to the production of 2E, 4E-C 12:2 -HSL. This is a variation of the canonical two-gene R-I circuits. We consider this to be a circuit involving three linked genes.
We are left with many questions about R1-I1-"crotonase" type QS in the genus Mesorhizobium. In what way does the I1-linked crotonase superfamily gene contribute to 2E, 4E-C 12:2 -HSL production? What genes are controlled by this AHL circuit, and in what way do they influence nodulation of host plants? Under what conditions might the other three AHL QS systems of M. japonicum MAFF 303099 be active? Ramsay and colleagues have shown that in M. japonicum R7A, the R3-I3 homologs are repressed by an epigenetically controlled anti-activator (56,57), does a similar regulatory control exist in MAFF 303099? What is the relationship of this system to the unknown system that results in 3-oxo-C 12:2 -HSL accumulation in human fecal samples?

Identification and classification of Mesorhizobium quorum sensing genes
We identified QS genes in genomes hosted on the JGI IMG website (46) by searching for pfam motifs specific to LuxI (pfam00765) and LuxR (pfam03472 and pfam00196) homologs (64). The JGI IMG database resource also identifies some LuxI homologs as CoA-utilizing enzymes based on the integrated Kyoto Encyclopedia of Genes and Genomes orthology and Enzyme Nomenclature (EC) database (65-67) categories (K18096 and EC 2.3.1.228/229, respectively) as we have done previously (24).

Radiotracer analysis
We used a 14 C-radiotracer assay (27,30) to detect AHLs produced by Mesorhizobium strains. For experiments shown in Fig. 1A, we modified a previous protocol (68) as follows: cells were grown in 5 mL of TY broth to mid-logarithmic phase (OD 600 of 0.7-0.9), harvested by centrifugation, and resuspended in 2 mL of phosphate-buffered saline (69) containing 0.5% (wt/vol) glucose and 5 µCi of L-[1-14 C]-methionine (55 mCi per mmol; American Radiolabeled Chemicals, Inc., St. Louis, MO, USA). Cultures were split, and AiiA lactonase (100 µg/mL final concentration) was added to one of the 1 mL samples (MalE-AiiA enzyme was a gift from N. Smalley, prepared as described in [70]). After further incubation at 30°C for 4-7 h, AHLs were extracted from the cell suspensions with two equal volumes of acidified ethyl acetate (0.1 mL glacial acetic acid per liter), and the extracts were then fractionated by C 18 reverse-phase HPLC in a 10%-100% methanol gradient (0.75 mL/min). One-minute fractions were collected, mixed with 4 mL of 3a70b scintillation fluid (Research Products International, Mount Prospect IL), and radioactivity was determined by liquid scintillation counting. For experiments shown in Fig. 5B, we modified a previous protocol (7), as follows: M. japonicum MAFF 303099 and the mlr5639 crotonase mutant were grown in BVM glucose medium to mid-logarithmic phase (OD 600 of 0.65-0.75). Then 5 µCi of L-[1-14 C]-methionine and 2 µM chemically synthesized 2E, 4E-C 12:2 -HSL (to ensure activation of I1 gene transcription) were added to 5 mL cultures. After 16-18 h at 30°C with shaking culture fluid was extracted with acidified ethyl acetate. Ethyl acetate extracts were fractionated by HPLC and radioactivity was quantitated as described above.

Plasmid and strain construction
We used E. coli DH5α-mediated assembly of PCR fragments (71) to create pZS1, by analogy to the phenylacetyl-HSL reporter plasmid pLL1 (24). The pZS1 construct contains the R1 gene (mlr5637) driven by a gdh constitutive promoter (45) and a I1-mCherry transcriptional fusion. The pZS2 plasmid is pZS1 with 580 bp of the 5′ end of R1 (79% of the 732 bp gene) removed by using a similar protocol. Because there is only 8 bp in the intergenic region between I1 and mlr5639 (Fig. 5A), we presume they are co-transcribed from the I1 promoter. Therefore, we created the crotonase superfamily gene complementation plasmid (pZS3) containing the mlr5639 open reading frame driven by the I1 (mlr5638) promoter and an I1 13-580 deletion. Plasmids were introduced into bacteria (E. coli or P. putida) by electrotransformation, and transformants were selected by plating on gentamicin-containing agar. Plasmid sequences were confirmed by whole-plasmid sequencing by a commercial vendor (Plasmidsaurus, Eugene OR), and the sequences deposited at National Center for Biotechnology Information (Genbank OQ303973, OQ303974, and OQ822817). The mlr5639 knockout construct was created by synthesizing a gBlock (Integrated DNA Technologies, Coralville, IA, USA) comprising 686 bp upstream and 666 bp downstream flanking DNA sequence, which created an in-frame deletion leaving the first 16 and last 5 codons of the mlr5639 open reading frame. This fragment was cloned into the suicide vector pJQ200SK (72), transferred to E. coli S17-1 by transformation, and then mobilized into MAFF 303099 by conjugation. Recombinants were selected by plating on gentamicin and fosfomycin, and double crossovers were selected by growth on 10% sucrose and loss of gentamicin resistance.

AHL bioassays
Detection of 2E, 4E-C 12:2 -HSL from ethyl acetate extracts of culture fluid or HPLC fractions of extracts was by using P. putida (pZS1) as follows: cell culture fluid extracts, HPLC fractions, or chemically synthesized AHLs were added to 13-mm glass tubes, and the solvent removed by evaporation under a stream of N 2 gas. Then 0.3 mL of P. putida (pZS1), diluted 1:100 from an overnight culture into fresh LB broth plus gentamicin, was added to each tube. After 16 h at 30°C with shaking, we transferred 100 or 150 µL samples to wells of 96-well black microtiter dish plates, and mCherry fluorescence (587 nm excitation, 610 nm emission, and gain 100) was measured by using a Synergy H1 plate reader (Biotek Instruments, Winooski, VT, USA). Standard curves were generated by measuring the fluorescence response to synthetic 2E, 4E-C 12:2 -HSL (linear response was between 1 and 10 nM). To detect any C 12 -HSL in NZP 2213 extracts, we utilized the QscR-based reporter strain E. coli (pJNQ pPROBE-P PA1897 ) as described previously (43), except that cells were incubated in 13-mm glass tubes rather than 96-deep well plates, and we used a fluorescence gain setting of 75. Note that E. coli (pJNQ pPROBE-P PA1897 ) can detect 2E, 4E-C 12:2 -HSL but only at micromolar concentrations.

Purification and identification of C 12:2 -HSL
We purified AHLs extracted from culture fluid after the removal of cells by centrifugation. Purification was guided by using the P. putida (pZS1) bioassay. Extraction, preparation, and HPLC fractionation were similar to that described elsewhere (7). We used the 75% and 100% methanol cuts from the C 18 Sep-pak cartridge (Waters, Milford, MA, USA). For HPLC, the flow rate was 0.75 mL/min, and the isocratic separation step was in 67% methanol. LC-MS/MS was performed on an AB Sciex 5600 QTOF (AB Sciex, Framingham, MA, USA) at the Department of Medicinal Chemistry, University of Washington. Samples were run as delivered by LC gradient over an Aquity UPLC HSS T3 column (100 Å, 2.1 × 100 mm; Waters, Milford, MA, USA), with a gradient from 5% methanol in water to 100% methanol (with 0.1% formic acid) over 13 min with a 0.2 mL/min flow rate. NMR analysis was performed at the University of Utah on a Direct Drive 500 MHz instrument (Agilent Technologies, Santa Clara, CA, USA) with a high sensitivity cold probe detection system. Spectroscopy scans were generated by using a DU 800 spectrophotometer (Beckman Coulter, Brea, CA, USA).