Isovaleryl-homoserine lactone, an unusual branched-chain quorum-sensing signal from the soybean symbiont Bradyrhizobium japonicum

Abstract

Many species of Proteobacteria communicate by using LuxI-LuxR–type quorum-sensing systems that produce and detect acyl-homoserine lactone (acyl-HSL) signals. Most of the known signals are straight-chain fatty acyl-HSLs, and evidence indicates that LuxI homologs prefer fatty acid-acyl carrier protein (ACP) over fatty acyl-CoA as the acyl substrate for signal synthesis. Two related LuxI homologs, RpaI and BtaI from Rhodopseudomonas palustris and photosynthetic stem-nodulating bradyrhizobia, direct production of the aryl-HSLs p-coumaroyl-HSL and cinnamoyl-HSL, respectively. Here we report that BjaI from the soybean symbiont Bradyrhizobium japonicum USDA110 is closely related to RpaI and BtaI and catalyzes the synthesis of isovaleryl-HSL (IV-HSL), a branched-chain fatty acyl-HSL. We show that IV-HSL induces expression of bjaI, and in this way IV-HSL functions like many other acyl-HSL quorum-sensing signals. Purified histidine-tagged BjaI was an IV-HSL synthase, which was active with isovaleryl-CoA but not detectably so with isovaleryl-ACP. This suggests that the RpaI-BtaI-BjaI subfamily of acyl-HSL synthases may use CoA- rather than ACP-linked substrates for acyl-HSL synthesis. The bjaI-linked bjaR₁ gene is involved in the response to IV-HSL, and BjaR₁ is sensitive to IV-HSL at concentrations as low as 10 pM. Low but sufficient levels of IV-HSL (about 5 nM) accumulate in B. japonicum culture fluid. The low levels of IV-HSL synthesis have likely contributed to the fact that the quorum-sensing signal from this bacterium has not been described elsewhere.

Quorum sensing (QS) allows bacteria to perceive and respond to population density and coordinate group behaviors (1, 2). Over 100 species of Proteobacteria use small diffusible N-acyl-homoserine lactones (acyl-HSLs) as QS signals. Acyl-HSLs are synthesized by LuxI protein homologs from S-adenosylmethionine (SAM) and acyl-acyl carrier proteins (acyl-ACPs) (3–5). Specific acyl-HSL signals bind to cognate LuxR transcriptional regulators to activate expression of QS-dependent genes. Signal specificity is conferred by the acyl side chain (2, 6–8).

Acyl-HSL QS is crucial for some bacteria-host interactions (8, 9) and has been described for many plant-associated bacteria, including members of the genera Rhizobium, Sinorhizobium, Mesorhizobium, and Bradyrhizobium. QS controls a variety of processes in these bacteria, including motility, exopolysaccharide synthesis, plasmid transfer, root nodulation efficiencies, and nitrogen-fixation efficiencies (10–14). Identification of the LuxI-family–produced acyl-HSLs has proven difficult in some plant-associated bacteria. A prominent example is the Bradyrhizobium japonicum strain USDA110. B. japonicum is the nitrogen-fixing symbiont of soybean, and strain USDA110 has been studied intensively due to its superior nitrogen-fixing abilities (15, 16). There is a chromosomal luxI homolog (bjaI, blr1063) with an adjacent luxR homolog (bjaR₁, blr1062) in strain USDA110 (17) but we, and others, have failed in attempts to detect an acyl-HSL product of BjaI by using assays based on heterologous acyl-HSL bioreporter strains (18, 19). Perhaps bjaI is not expressed under laboratory growth conditions or perhaps the BjaI-BjaR₁ QS system is nonfunctional. However, previous transcriptome analyses have shown that the genes are in fact expressed in laboratory culture and during symbiosis (20–22). An explanation that appealed to us was that B. japonicum USDA110 produces a unique acyl-HSL signal.

Whereas most LuxI homologs studied to date are fatty acyl-HSL synthases that produce straight-chain fatty acyl-HSLs, we have recently described two aryl-HSL synthases. The product of RpaI from the photosynthetic bacterium Rhodopseudomonas palustris is p-coumaroyl-HSL (23), and the product of BtaI from photosynthetic stem-nodulating bradyrhizobia is cinnamoyl-HSL (24). Both of these aryl-HSL signals serve as coactivators of gene expression together with their respective LuxR homologs. The BjaI polypeptide sequence is closely related to BtaI and RpaI (Fig. 1). Thus, it seemed possible that BjaI produces a unique aryl-HSL, which has eluded detection by heterologous bioreporters.

Fig. 1.

Phylogenetic tree of LuxI family members in selected α- and γ-proteobacteria. The scale bar indicates the number of substitutions per residue. Bootstrap values as the percentage of 500 samplings are shown for nodes with values of 50% or greater. The subfamily tree of signal synthases containing RpaI, BraI, and BjaI is highlighted in red. Abbreviations for genus and species are followed by the name of the LuxI homolog: Agrobacterium vitis S4 (Av), A. tumefaciens C58 (At), R. palustris (Rp), Sinorhizobium meliloti 1021 (Sm), Mesorhizobium opportunistum WSM2075 (Mo), stem-nodulating Bradyrhizobium (B), Methylobacterium extorquens AM-1 (Me), B. japonicum USDA110 (Bj), Vibrio fischeri ES114 (Vf), P. aeruginosa PAO1 (Pa), and Yersinia pseudotuberculosis YP111 (Yp). Note that the genomes of R. palustris BisB18 and A. vitis S4 contain more than one luxI homolog as indicated. Both organisms have one LuxI homolog that falls within the RpaI-BraI-BjaI cluster and at least one LuxI homolog that falls outside of the cluster.

Here we show that the B. japonicum USDA110 QS signal is in fact not an aryl-HSL, but a branched-chain fatty acyl-HSL, isovaleryl-HSL (IV-HSL). Fatty acyl-HSL synthases that have been studied show a strong preference for acyl-ACPs over acyl-CoAs as activated fatty acid substrates (4, 5). In contrast, we show that BjaI is active with isovaleryl-CoA (IV-CoA), but we did not detect activity with isovaleryl-acyl carrier protein (IV-ACP). We suggest that there is a subfamily of acyl-HSL synthases, which includes RpaI, BtaI, and BjaI. We propose that a hallmark of this subfamily is an acyl-CoA substrate preference over acyl-ACP. We also show that IV-HSL is produced at relatively low (nM) concentrations and that it is active at exceedingly low (pM) concentrations.

Results

B. japonicum USDA110 Produces a Branched-Chain Acyl-HSL.

Our approach to identifying the B. japonicum USDA110 acyl-HSL capitalized on the fact that many luxI family members are positively autoregulated by their cognate LuxR homolog and their acyl-HSL product (2, 6–8). There are inverted repeat elements in promoter regions of genes controlled by LuxR homologs, and these elements serve as binding sites for these transcription factors (1, 8). The bjaR₁ gene is immediately upstream and transcribed in the same direction as bjaI (Fig. 2A). Quantitative RT-PCR revealed that bjaI and bjaR₁ are transcribed independently. Inspection of the intergenic region between bjaR₁ and bjaI revealed the presence of an inverted repeat centered 55 bases upstream of the predicted bjaI translational start codon (Fig. 2 A and B). We hypothesized that this region serves as a binding site for BjaR₁ and that the bjaI promoter is activated by BjaR₁ and the putative acyl-HSL signal. Thus, we constructed a B. japonicum reporter strain with lacZ fused to and insertionally inactivating the chromosomal bjaI (bjaI-lacZ). Because bjaI is insertionally inactivated by lacZ, this strain cannot produce the putative BjaI product.

Fig. 2.

B. japonicum bjaR₁-bjaI gene organization. (A) Map of the bjaR₁-bjaI region. Numbers above refer to the nucleotide positions in the genome, and gene designations are indicated. The sequence and location of an inverted repeat that might serve as a BjaR₁-binding site are indicated. The inverted repeat is centered 55 bp from the predicted bjaI start codon and 22 bp from the BjaR₁ translational stop codon. RP, predicted regulatory protein; HP, hypothetical protein; AA-T, predicted amino acid transport membrane protein. (B) Sequence alignment of the putative BjaR₁-binding site with inverted repeats found in promoter regions of other luxI-type family members. Ralstonia solanacearum (solI), P. aeruginosa PAO1 (rhlI), V. fischeri ES114 (luxI), R. palustris CGA009 (rpaI).

We found that bjaI-lacZ expression in our reporter strain was activated by ethyl acetate extracts of cell-free fluid from the late-exponential growth phase B. japonicum cultures. When we separated the extracts by HPLC and followed biological activity, a single peak of activity was observed. Extracts of a bjaI deletion mutant had no activity (Fig. 3A), indicating that production of the bioactive compound was dependent on BjaI. To gain evidence that the bioactive compound was an acyl-HSL, we provided cells with [carboxy-¹⁴C]-methionine, a portion of which is incorporated into SAM via SAM synthetase. LuxI-type synthases incorporate ¹⁴C-SAM into the acyl-HSL homoserine lactone ring. This radiolabeling approach allows detection of acyl-HSL synthesis regardless of the nature of the acyl side chain (25). A single peak of radioactivity was detected, and the retention time of this peak was the same as that of the bioactive compound (Fig. S1). This result indicates that the bioactive material was an acyl-HSL, and we did not detect other acyl-HSLs produced by B. japonicum USDA110. The HPLC retention time of the bioactive material did not match those of known fatty acyl-HSL or aryl-HSL signaling molecules.

Fig. 3.

Evidence supporting the conclusion that IV-HSL is the B. japonicum USDA110 quorum-sensing signal. (A) HPLC elution profile of B. japonicum wild-type culture extracts (■) and culture extracts of the bjaI deletion mutant (○). HPLC fractions were assayed with the bjaI-lacZ bioreporter strain. (B) Mass spectrum of the signal extracted and purified from cell-free B. japonicum culture fluid. (C) Acyl-HSL dose–responses with the bjaI-lacZ reporter. IV-HSL (●), C5-HSL (▲), C4-HSL (○), C6-HSL (◇), and PIV-HSL (◆). Values are the means of at least five independent experiments, and the bars show the ranges. (D) Mass spectrum of synthetic IV-HSL. (E) The proposed B. japonicum QS molecule IV-HSL.

To determine the chemical composition of the bioactive compound, we analyzed active HPLC fractions of purified material by high-resolution mass spectrometry (MS). The parent ion of the purified material had a M+1 of 186.1140, which is consistent with a chemical formula of C₉H₁₅NO₃ (Fig. 3B). The spectrum also showed a fragment with a mass of 102, which is characteristic of the homoserine lactone fragment of acyl-HSLs (26). The deduced chemical composition suggested three possible molecules: valeryl-HSL (C5-HSL), IV-HSL, and pivalyl-HSL (PIV-HSL). When we compared the HPLC elution profiles of synthetic IV-HSL, PIV-HSL, and C5-HSL with that of the natural product, only IV-HSL was eluted in the same fractions as the material isolated from B. japonicum culture fluid (both gradient and isocratic HPLC). Moreover, IV-HSL induced the expression of the bjaI-lacZ reporter, whereas PIV-HSL had no detectable activity. C5-HSL showed some activity, but only at much higher concentrations than the concentrations required for IV-HSL activity (Fig. 3C). Finally, the MS spectra of synthetic IV-HSL and the natural product were indistinguishable (Fig. 3 B and D). Thus, we conclude that the BjaI B. japonicum USDA110 quorum-sensing signal is IV-HSL (Fig. 3E).

IV-HSL Is Active at Low Concentrations and Functions with BjaR₁.

We hypothesized that, as is the case for other acyl-HSL quorum sensing systems, the genetically linked luxR homolog would be involved in the response to the acyl-HSL signal. Therefore, we constructed a bjaR₁ deletion strain and showed by quantitative RT-PCR that, in the midlogarithmic growth phase, bjaI expression in this mutant was about 15% of the level in wild-type cells and IV-HSL was <25% of wild-type levels.

When we monitored β-galactosidase in the bjaI-lacZ reporter strain, we found that the bjaI promoter responded to IV-HSL concentrations as low as 10 pM, with a maximal response at about 1 nM (Fig. 3C). This is unusually low. Most acyl-HSL–dependent promoters respond to signal that is in the range of 1–100 nM (for a discussion, see ref. 24). The bjaI promoter responded to C5-HSL and C4-HSL, but with much lower sensitivity than to IV-HSL. C6-HSL had detectable activity at a relatively high concentration of 100 nM (Fig. 3C). These data show that, although the B. japonicum USDA110 quorum-sensing system is specific for IV-HSL, it also responds to straight-chain acyl-HSLs at concentrations that are physiologically appropriate for other bacteria. The maximum amount of IV-HSL produced by B. japonicum was also low relative to other bacteria—about 2–10 nM (5 nM in the experiment shown in Fig. 4). Acyl-HSLs produced by other bacteria typically reach much higher concentrations in laboratory media (0.1–10 μM). However, because 5 nM IV-HSL exceeds the concentration required for maximal activation of bjaI-lacZ expression in the reporter strain, we conclude that the amount of IV-HSL produced is sufficient to establish a QS circuit. As is characteristic of quorum sensing, the accumulation of IV-HSL in culture fluid lagged behind increases in cell density in the early part of growth and then accelerated (Fig. 4). This is consistent with positive autoregulation of acyl-HSL production. Notably, signal concentrations rapidly decreased in stationary phase. Alkaline conditions enhance spontaneous degradation of acyl-HSL molecules. However, because the pH of the culture was slightly acidic in stationary phase, a more likely explanation for disappearance of the signal is that it is degraded by one or more enzymes produced by B. japonicum.

Fig. 4.

Cell density-dependent production of IV-HSL. A B. japonicum culture was sampled at the indicated times, samples were extracted with acidified ethyl acetate, and bioactive material in the extracts was measured by using the B. japonicum bjaI-lacZ bioassay (○). Cell growth was measured as optical density at 600 nm (■). The results show a representative experiment. In five independent experiments, the peak concentration of IV-HSL measured ranged from 1.5 to 8 nM.

BjaI Is an IV-HSL Synthase That Uses IV-CoA as an Acyl Donor.

Our genetic analysis showed bjaI is required for IV-HSL synthesis (Fig. 3A). To more directly show that BjaI is an IV-HSL synthase, we purified histidine-tagged BjaI (H₆-BjaI) and showed that the purified protein is an IV-HSL synthase. H₆-BjaI requires SAM and IV-CoA as substrates for IV-HSL synthesis. We calculated a specific activity of 33 nmol min⁻¹ mg⁻¹ H₆-BjaI. If either substrate was omitted from the reaction mixture, or if IV-ACP was used as a substitute for IV-CoA, enzyme activity was below the level of detection (6 nmol min⁻¹ mg⁻¹ H₆-BjaI). To confirm that the bioactive product of H₆-BjaI was IV-HSL, we showed that it had a methanol-gradient HPLC retention time identical to that of synthetic IV-HSL. For comparison, our BjaI preparation is about 30 times more active with IV-CoA than the LuxI is with its acyl-ACP donor and has about the same specific activity as TraI with its fatty acyl-ACP substrate. Purified RhlI from Pseudomonas aeruginosa is about 20 times more active than BjaI with its preferred substrate, butyryl-ACP (3–5).

Discussion

We have determined that the soybean symbiont B. japonicum synthesizes IV-HSL, a branched-chain fatty acid HSL. Because B. japonicum has a luxI homolog, it was expected to produce an acyl-HSL, but the identity of the compound was elusive. We found that B. japonicum USDA110 produces on the order of 5 nM IV-HSL, compared with 0.1–10 μM concentrations of acyl-HSLs produced by most bacteria known to use acyl-HSL QS systems (23, 24, 27, 28). This very low concentration of IV-HSL could explain why it might not be detected in bioreporter screens (18, 19) or by liquid chromatography MS/MS methods that have come into recent use (26, 29).

Because B. japonicum BjaI is closely related to the R. palustris RpaI and stem-nodulating bradyrhizobia BraI proteins that direct the synthesis of p-coumaroyl-HSL and cinnamoyl-HSL, respectively, we expected that B. japonicum might also produce an aryl-HSL. That BjaI is not an aryl-HSL synthase suggests that RpaI, BraI, and BjaI have some other feature in common, leading us to hypothesize that they might prefer CoA- rather than ACP-linked substrates. In all cases studied, LuxI-type proteins generate acyl-HSLs preferentially from acyl-ACPs and SAM rather than from acyl-CoAs and SAM (3–5). ACP and CoA have similar chemical characteristics, and both form adducts with carboxylated substrates and present these substrates to enzymes. ACP proteins typically function in biosynthesis and in transferring fatty acid substrates, whereas CoA is found linked to more structurally diverse carboxylated substrates. CoA-modified compounds are intermediates in biosynthetic pathways as well as in biodegradation pathways.

We previously reported the synthesis of p-coumaroyl-HSL from p-coumaroyl-CoA and SAM by RpaI, but the activity was rather low and we were unable to synthesize soluble p-coumaroyl-ACP as an alternative substrate (23). Here we were able to generate both IV-CoA and IV-ACP for use in comparative studies. We found that purified His-tagged BjaI could use IV-CoA as an acyl donor for IV-HSL synthesis, but we did not detect activity with IV-ACP, suggesting that IV-CoA is the natural substrate. Although this is unlike the fatty acyl-HSL synthases that have been studied, it makes metabolic sense. IV-CoA is a common cellular metabolite and is formed as an intermediate during the metabolism of leucine (30, 31). IV-ACP was prepared with Escherichia coli ACP rather than B. japonicum ACP. However, E. coli ACP was also used to prepare butyryl-ACP for studies of the P. aeruginosa RhlI protein, and RhlI showed a strong preference for butyryl-ACP over butyryl-CoA (4).

One other bacterium has been reported to produce branched-chain acyl-HSLs. Aeromonas culicicola produces small amounts of C-7 and C-9 branched-chain acyl-HSLs (29). The A. culicicola genome has not been sequenced, and we do not have information on luxI homologs in this species. Therefore, we do not yet know whether the long branched-chain acyl-HSLs are derived from an ACP-bound intermediate of branched-chain fatty acid biosynthesis or whether A. culicicola might have a BjaI-like acyl-HSL synthase that prefers a CoA-modified substrate.

The B. japonicum BjaI polypeptide clusters with RpaI and BraI in our phylogenetic analysis of LuxI homologs (Fig. 1). We know that R. palustris can enzymatically synthesize p-coumaroyl-CoA (23), and we are unaware of any work showing that bacteria can synthesize p-coumaroyl-ACP. Thus, we propose that the subfamily of LuxI homologs including BjaI, RpaI, BraI, and other LuxI homologs may have diverged from other subfamilies in that it uses only acyl-CoAs for acyl-HSL synthesis. It will be of interest to study other members of this subfamily.

The B. japonicum USDA110 bjaI-lacZ bioreporter is ultrasensitive with lacZ induction requiring as little as 10 pM synthetic IV-HSL. Such a high sensitivity for a QS system is rather unusual as bacteria typically require nanomolar signal concentrations to respond (23, 27, 28). We have recently reported that photosynthetic stem-nodulating bradyrhizobia have ultrasensitive cinnamoyl-HSL QS systems (24). In addition to being very sensitive to signal, the photosynthetic bradyrhizobia and strain USDA110 have relaxed signal receptor specificity. They respond to short-chain acyl-HSLs, including C4-HSL and C5-HSL in the case of USDA110, at concentrations in the low nanomolar range. The basal level of bjaI-lacZ expression in B. japonicum is low, and this reporter is induced about 20-fold by exogenous addition of IV-HSL. An especially high affinity of BjaR₁ for IV-HSL could account for its ultrasensitivity. Notably, these picomolar signal-response systems that can also respond to nanomolar levels of other acyl-HSLs might provide a means to both avoid detection by other species and detect other species in a local environment.

Ultrahigh signal sensitivity and low amounts of acyl-HSL production do not appear to be general features of the noncanonical QS systems. R. palustris produces and responds to p-coumaroyl-HSL at concentrations typical of other QS systems. In addition, we found that another R. palustris strain, strain BisA53, which encodes a LuxI synthase with 84% sequence similarity to BjaI, also produces IV-HSL, but at 10–100 times higher concentrations than B. japonicum USDA110.

In contrast to p-coumaroyl-HSL production by R. palustris, neither the photosynthetic bradyrhizobia nor B. japonicum USDA110 require exogenous substrate to produce cinnamoyl-HSL or IV-HSL. Thus, it is unlikely that these plant symbionts depend on their host to produce their QS signals. However, QS has been shown to influence the interactions of other rhizobia with their plant hosts (32–36). The B. japonicum USDA110-soybean symbiosis represents an excellent opportunity to study the responses of both host and symbiont to a QS signal. The symbiosis can be experimentally manipulated, the genomes of each organism have been sequenced, and tools are available to assess the transcriptional responses of both B. japonicum and soybean to IV-HSL.

Materials and Methods

Bacterial Strains and Growth Conditions.

The bacterial strains and plasmids used are listed in Table S1. We used strain B. japonicum USDA110spc4, which carries a spectinomycin resistance cassette at a neutral genetic locus (37). E. coli was grown in Luria-Bertani broth with aeration at 37 °C. B. japonicum strains were grown at 30 °C in a modified Vincent's minimal medium (VMM) (38, 39) with shaking. Antibiotics were used at the following concentrations (μg/mL): for E. coli, ampicillin (200), kanamycin (30), and tetracycline (10); and for B. japonicum, spectinomycin (100), streptomycin (50), kanamycin (100), and tetracycline (50 for agar and 25 for broth).

Plasmid and Strain Construction.

For construction of the B. japonicum bjaI-lacZ reporter strain AL05 and the bjaI deletion strain AL17, the regions upstream and downstream of bjaI were PCR-amplified with appropriate primer pairs, verified by sequencing, and cloned head-to-tail into the suicide plasmid pSUP202pol4, resulting in plasmid pAL04. The upstream region was a 715-bp EcoRI-PstI fragment extending 15 bp downstream of the annotated bjaI translational start site. The downstream region was a 698-bp PstI-NotI fragment, which included the last 18 bp of bjaI. A 5.3-kb PstI-fragment that included a promoterless lacZ gene and a kanamycin resistance marker from pHRP314 was cloned into the PstI site of pAL04 to generate pAL05. To construct pAL17, we inserted the Ω cassette from pBSL15Ω into the PstI site of pAL04. The suicide plasmids pAL05 and pAL17 were mobilized into B. japonicum USDA110spc4 by conjugation from E. coli S17-1. Homologous recombination into the chromosome with appropriate antibiotic selection resulted in strains AL05 and AL17, respectively. For construction of the bjaR₁ deletion strain 6485, the suicide plasmid pRJ6485 was generated by amplification and cloning of upstream and downstream regions of bjaR₁ into pSUP202pol4. The 762-bp XbaI-KpnI upstream region contained 62 bases of bjaR₁; in the 655-bp KpnI-PstI downstream region, 6 bp of bjaR₁ were retained. A kanamycin cassette was inserted into the KpnI site. To construct a gene coding for BjaI with an N-terminal six-histidine tag, we PCR-amplified bjaI by using the primers listed in Table S1. The amplified 711-bp fragment was digested with PciI and NdeI and cloned directionally into NcoI- and NdeI-digested pET16b, thereby replacing a 62-bp fragment of pET16b that encodes the His₁₀ tag. The resulting plasmid, pET16b-BjaI, codes for a polypeptide with a His₆ tag fused to the second amino acid residue of the BjaI protein.

Detection of Acyl-HSLs.

Acyl-HSL bioassays were performed with the bjaI-lacZ bioreporter strain AL05 as follows. Whole bacterial cultures or cell-free culture fluid prepared by removing cells by centrifugation (22,000 × g, 20 min, 4 °C) were extracted twice with an equal volume of acidified ethyl acetate (0.1% glacial acetic acid). Ethyl acetate extracted material was added to 2-mL Eppendorf tubes, and the ethyl acetate removed by evaporation under a gentle stream of N₂ gas. A culture of the bioreporter strain AL05 (bjaI-lacZ) (early growth phase culture at an OD₆₀₀ of 0.1–0.2) was diluted to an OD₆₀₀ of 0.05 and was added to the ethyl acetate extract-containing tubes (0.5 mL vol). The tubes were then incubated for 16 h at 30 °C with shaking, and then β-galactosidase activity was measured as described elsewhere (40). Synthetic acyl-HSLs were used to prepare standard curves. We also used [carboxy-¹⁴C]-methionine in radiotracer acyl-HSL assays. The radiolabeling and HPLC separation of radiolabled material were as described elsewhere (25, 41), except that the labeling time was extended to 6 h at 30 °C.

Purification and Identification of IV-HSL.

Material for MS was extracted from 4 L of late-exponential growth phase B. japonicum cultures. Cell-free culture fluid was extracted twice with acidified ethyl acetate and then dried, dissolved in 50% methanol in water, and passed through a C₁₈ Sep-Pak cartridge (Waters) from which activity was eluted in 10–30% methanol. This active material was separated by C₁₈ reverse-phase HPLC (10–100% methanol, 1 mL/min, with collection of 70 1-mL fractions). The bioactive fractions (18, 19) were then subjected to isocratic HPLC (10% methanol). Fractions (1-mL) were collected, and the bulk of the bioactive material was eluted in fractions 31 and 32, which were pooled and concentrated under a gentle stream of N₂ gas. MS analysis was performed with a Waters Quadrupole/Triwave/Orthogonal Acceleration Time-of-Flight Tandem Hybrid Mass Spectrometer MS/IMS/MS with a collision energy of 6 eV and a cone of 35 V (Waters).

Acyl-HSLs.

IV-HSL and PIV-HSL were prepared on a 20-mg scale by using reported solution- and solid-phase synthesis methods (42, 43). All reagents and solvents were purchased from commercial sources and used without further purification, with the exception of dichloromethane, which was distilled over calcium hydride. Solid-phase synthesis was performed by using polystyrene amine resin (Biotage; loading 1.1–1.5 mmol/g). Full details of the instrumentation and analytical methods can be found in previous reports (42, 43). C4-, C5-, and C6-HSLs were purchased from commercial sources.

Production and Purification of H₆-BjaI.

For overproduction of His-tagged BjaI, E. coli Rosetta (DE3) pLysS carrying pET16b-BjaI was grown at 37 °C in 100 mL of medium containing ampicillin. When cultures reached an OD₆₀₀ of 0.4–0.5, H₆-BjaI production was induced by addition of 1 mM isopropyl thiogalactoside (IPTG). After overnight incubation with IPTG at 16 °C, cells were harvested by centrifugation and suspended in 5 mL of binding buffer [20 mM Tris⋅HCl (pH 7.9), 500 mM NaCl, 5 mM imidazole] plus 50 μL Halt Protease Inhibitor Mixture (Thermo Scientific). Cells were lysed by sonication on ice. The lysate was cleared by centrifugation (22,000 × g, 20 min, 4 °C), and the supernatant fluid was further clarified by ultracentrifugation (125,000 × g for 1 h, 4 °C) to remove cell walls and membranes. We affinity-purified H₆-BjaI by nickel-nitrilotriacetic acid (Ni-NTA) agarose column chromatography (Qiagen). The 0.5-mL Ni-NTA column was pre-equilibrated with binding buffer. Crude extracts were loaded, and the column was washed consecutively with 5 and 50 mM imidazole before H₆-BjaI was eluted with 300 mM imidazole. Protein purity was assessed by SDS/PAGE and judged to be >95% pure. Protein concentrations were determined by using the Bio-Rad assay.

Activity of Purified H₆-BjaI.

Activity assays were modeled on those described for assessing the P. aeruginosa RhlI rates of C4-HSL synthesis (4) with the following modifications. Reactions were performed in 100 μL volumes at 37 °C. The reaction mixtures contained buffer [20 mM Tris (pH 7.5), 10 mM MgCl₂, 1 mM DTT] and SAM (60 μM), IV-CoA (40 μM), or IV-ACP (40 μM). Reactions were initiated by addition of 10 ng H₆-BjaI. Synthesis of IV-HSL with SAM, IV-CoA, and 10 ng of protein was linear over 20 min. Our standard assay time was 10 min. Reactions were stopped by addition of 100 μL of acidified ethyl acetate into which acyl-HSLs were extracted. After a second extraction with 100 μL of acidified ethyl acetate, the extracts were combined and a portion was used to determine the concentration of IV-HSL with the bioreporter strain AL05 as described above. IV-CoA and SAM were purchased from Sigma-Aldrich. IV-ACP was synthesized as described below.

Synthesis of IV-ACP.

Apo-ACP was expressed in and purified from E. coli DK574 (pJT93) as described elsewhere by using a modified ion exchange protocol (44). After lysis, incubation, precipitation, and dialysis overnight against 25 mM 2-(N-morpholino)ethanesulfonic acid, MES (pH 6.1), apo-ACP was purified on a 1.8-mL POROS HQ 20 column on an AKTA purifier 10 in homologous buffer with a 0–1 M lithium chloride gradient. Fractions containing apo-ACP were pooled, precipitated, and analyzed by conformation-sensitive native-gel electrophoresis on a 20% gel with 2.5 mM urea. Apo-ACP was dialyzed against a storage buffer containing 25 mM MES, 5 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP), and 10% glycerol overnight before flash freezing and storing at −80 °C. The molecular mass was verified by electrospray MS. IV-ACP was synthesized using Bacillus subtilis phosphopantetheinyl transferase (Sfp) (45). Sfp expression and purification was as described (46). The reaction mixture contained 100 mM Tris⋅HCl (pH 8.0), 10 mM MgCl₂, 10 mM TCEP, 5 μM Sfp, 3 mM IV-CoA, and 0.3 mM apo-ACP. The reaction was allowed to proceed for 2 h at 37 °C after which the reaction mixture was dialyzed overnight against 10 mM MES (pH 6.1). Native PAGE was used to confirm conversion of Apo-ACP to IV-ACP. IV-ACP was purified by ion exchange as described for apo-ACP. Pure fractions were pooled and precipitated with 2 vol of acetone overnight at −20 °C. The precipitate was centrifuged at 46,000 × g for 20 min and suspended in 50 mM Tris⋅HCl and 5 mM TCEP (pH 8.0). IV-ACP was dialyzed against storage buffer before being flash-frozen and stored at −80 °C. The molecular mass was verified by electrospray MS as described (46).

Quantitative RT-PCR.

Cells for comparative analysis of gene expression were harvested at an OD₆₀₀ of 0.5–0.6. Cell harvest, RNA extraction, and RNA purification were as described previously (20, 47). cDNA was synthesized with SuperScript II reverse transcriptase from DNase-treated RNA by using random hexamer primers as recommended by the manufacturer (Invitrogen). Quantitative RT-PCR was performed in a Bio-Rad CFX96 thermocycler. Each PCR contained 10 μL 2× SsoFast EvaGreen Supermix (Bio-Rad), 0.4 μM of individual primers, and appropriate dilutions of cDNA in a total volume of 20 μL. Gene expression fold changes were calculated as described elsewhere (48). Transcript levels of sigA were used to normalize bjaI transcript levels.

LuxI Family Tree Construction.

Amino acid sequences of selected LuxI family members were obtained from the Integrated Microbial Genome Database (http://img.jgi.doe.gov/cgi-bin/w/main.cgi) and aligned with ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2). The family tree was constructed with the distance matrix-based Fitch algorithm in the PHYLIP software package (http://bioweb2.pasteur.fr/phylogeny/intro-en.html) and visualized by using Phylodendron TreePrint (http://www.es.embnet.org/Doc/phylodendron/treeprint-form.html).