Genome sequencing-assisted identification and the first functional validation of N-acyl-homoserine-lactone synthases from the Sphingomonadaceae family

Bacterial strains, growth media and biosensor strains

Novosphingobium subterraneum DSM 12447 (previously Sphingomonas subterraneum DSM12447) was provided by Andreas Stolz (Institut fur Mikrobiologie at the Universitat Stuttgart, Stuttgart, Germany). This strain was isolated from terrestrial subsurface and was shown to catabolize a variety of natural recalcitrant and anthropogenic compounds including naphthalene, toluene, biphenyl, dibenzothiophene and fluorine (Balkwill et al., 1997; Takeuchi, Hamana & Hiraishi, 2001). The bacterial strains and plasmids used in this work are listed in Table 1.

Novosphingobium subterraneum DSM 12447 strain was grown in tryptone soy broth (TSB), potato dextrose (PD) or R2A medium (Difco Laboratories, Detroit, MI) at 28 °C. Agrobacteria AB minimal media (Chilton et al., 1974) at 28 °C was used to grow AHL-dependent biosensor strains Agrobacterium NTL4 and A136. For AHL signal induction bioassays, E. coli JM109 and Agrobacterium NTL4 strains harboring empty vector pSRKKm or pSRKKm with cloned novI genes were grown on Luria-Bertani broth (LB) medium at 37 °C and 28 °C and supplemented with kanamycin at 25 and 50 µg/ml, respectively. For AHL signal detection bioassays, Agrobacterium tumefaciens NTL4 (pZLR4) and A136 (pCF218, pMV26) were grown in AB medium supplemented with 0.2% (w/v) dextrose and 0.01% (w/v) yeast extract and gentamycin (10 µg/ml) for NTL4 (pZLR4) (Cha et al., 1998) and kanamycin (25 µg/ml) and tetracycline (5 µg/ml) for A136 (pCF218, pMV26) (Sokol et al., 2003) E. coli-based biosensors JM109 (pSB401), JM109 (pSB1075) and JM109 (pSB536) were grown in LB media with appropriate antibiotic for plasmid maintenance (Winson et al., 1998). Chromobacterium violaceum CV026 biosensor was growth in tryptone yeast extract/potato dextrose (1:1) agar media for T-streak bioassays (McClean et al., 1997).

Each AHL-dependent bacterial biosensor strain used in this work along with its AHL receptor protein and cognate AHL signal is listed in Table 1. All media and growth conditions for AHL detection bioassays are as previously described by our laboratory (Gan et al., 2009; Lowe et al., 2009).

Biosensor detection

Reverse-phase (RP) one-dimensional (1-D) TLC plates were used to determine AHL signal profiles. Concentrated acidified ethyl acetate (aEtOAc) extracts were spotted on to the C18 RP-TLC plate origin in 2-µL volumes and representing from 0.5 to 2-mL supernatant equivalents (EMD Chemicals Inc., Gibbstown, NJ). Plates were developed in a 70:30 (v/v) methanol:water mobile phase, dried and AHLs were detected as described (Scott et al., 2006). Bioluminescence produced by the A. tumefaciens A136 traR, PtraI::luxCDABE-based biosensor strain overlaid on the chromatograms was detected with a Bio-Rad charge-coupled device (ccd) ChemiDoc MP system at two different sensitivity settings AHL signals were identified with appropriate reference compounds. This involves determining and comparing retardation factors (Rf) of unknown samples to AHL reference compounds (Shaw et al., 1997).

Development of two-dimensional (2-D) thin layer chromatography for AHLs

The AHL extract was initially spotted onto the bottom left corner of the C18 RP-TLC plate The amount needed was estimated based on the AHL signal strength obtained from multiple independent 1-D RP-TLC runs. The spotted TLC plate was eluted with 70:30 (v/v) methanol: water as the first mobile-phase in a glass tank. The mobile-phase was allowed to rise until the top of the TLC plate before removing the plate to dry overnight. Then, the TLC plate was rotated 90° counterclockwise, placed into a tank with 25:75 (v/v) 2-propanol: water as the second mobile-phase until it reached the top of the TLC plate. After drying, the TLC plate was overlaid with TraR-dependent Agrobacterium biosensor strain A136 using the same procedure as used for 1-D TLCs.

AHL identification and quantification by LC-MS/MS

Whole genome sequencing of N. subterraneum DSM 12447

Genomic DNA of strain DSM12447 was extracted using the GenElute™ (Sigma-Aldrich, St. Louis, MO, USA) and converted into next generation sequencing library using Nextera XT (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. Whole genome sequencing was performed using the MiSeq (Illumina, San Diego, CA, USA) at the Monash University Malaysia Genomics Facility. The raw data for each bacterium were error-corrected and assembled using Spades v2.5 (default setting) (Bankevich et al., 2012). The generated contigs were scaffolded and gap-closed using SSPACE and GAPFiller, respectively (Boetzer et al., 2011; Boetzer & Pirovano, 2012). Genome annotation was performed using Prokka and InterProScan5 (Jones et al., 2014; Seemann, 2014).

Identification and phylogenetic analyses of LuxI homologs

The whole genome of strain DSM 12447 was submitted to Anti-SMASH server (Weber et al., 2015) for the identification of biosynthetic gene cluster(s) (including AHL synthase cluster). In-silico validation of the identified LuxI homologs was performed through protein alignment with bona fide LuxI homologs and manual inspection of the alignment for conserved LuxI homologs amino acid residues. Visualization of the gene organization was performed with EasyFig (Sullivan, Petty & Beatson, 2011) using NCBI annotated sequence as input. Protein alignment was done using MAFFT-LINSI (Katoh & Standley, 2014) and the alignment was trimmed with trimal (-gappyout) to retain as much site informative as possible (Capella-Gutierrez, Silla-Martinez & Gabaldon, 2009). Subsequently, a phylogenetic tree was inferred using phylobayes (-cat –gtr –ncat 4) (Lartillot et al., 2013). A total of four independent chains were run for 10,000 generations each. The first 1,000 trees were discarded as burn-in and a consensus tree was built based on the 50% majority rule. Mesquite was used for tree visualization and editing (Maddison & Maddison, 2015).

Amplification and cloning of 3 putative AHL synthase genes

Primers and plasmids used in this study are listed in Table 1. Amplification of the luxI homologs was performed using Q5 polymerase mastermix (New England Biolabs, Ipswich, MA, USA) according to the manufacturer’s instructions. Approximately 150 ng of the purified PCR amplicons were mixed with 50 ng of pSRKKm vector (Khan et al., 2008) and double digested with NheI and NdeI (New England Biolabs, Ipswich, MA, USA) for 1 h. After heat inactivation, the digested products were purified using magnetic beads (Omega Biotek, Norcross, GA, USA) and ligated with Electroligase (New England Biolabs, Ipswich, MA, USA) for 30 min. The ligated products were transformed into A. tumefaciens NTL4 and Escherichia coli JM109 using electroporation.

Inducible expression of luxI homologs and detection of AHLs from solid media

Bacterial culture (A. tumefaciens NTL4 and Escherichia coli JM109 grown for 96- and 48 h, respectively) supernatants were resuspended from LB plates supplemented with antibiotic kanamycin and filter-sterilized IPTG at 0, 10, 100 and 1000 µM containing pSRKKm with and without the cloned luxI homologs (Table 1) were extracted with acidified ethyl acetate (aEtOAc) (1 mL of glacial acetic acid per 200 mL of ethyl acetate) for 60 min with shaking (150 r.p.m.). The extracts were then centrifuged to separate the aqueous and ethyl acetate phases. The ethyl acetate phase was recovered and dried in a Savant Speed Vac. Twenty-, thirty- or fifty-fold concentrated extracts were prepared and used in AHL detection bioassays.

Acyl-homoserine lactone extractions for induction assays

NTL4 (pNsub1,2 or 3) strains were grown in 20 mL of LB (50 µg/ml kanamycin) supplemented with different amount of inducer isopropyl β-D-1-thiogalactopyranoside (IPTG) inducer to final concentration of 0,10,100, or 1,000 µM and incubated overnight in a shaking incubator at 28 °C. The next day, 20 ml of ethyl acetate was added to each of the twelve tubes and shaken at room temperature for 2 h. The tubes were then centrifuged at 5,000 rpm for 10 min to separate the liquid layers and then the top layer of ethyl acetate was aspirated off and stored. This layer was separated into several 1.5 ml micro centrifuge tubes and the ethyl acetate evaporated off using a speed-vac. The residue in the tube was then resuspended in 75 µl of fresh ethyl acetate to bring all samples to a 20×-volume equivalent extract concentration.

N. subterraneum DSM 12447 produces multiple distinct AHL signals

Culture extracts prepared from N. subterraneum DSM 12447 activated three of the five AHL-dependent whole cell bacterial biosensors tested (Table 3). One-dimensional (1-D) RP-TLC separation of the culture extract followed by AHL detection using TraR-based bioluminescence biosensor led to the confident detection of three putative AHL signals (Fig. 1A). Given the lack of signal resolution presumably due to high AHL signal diversity, two-dimensional (2-D) RP-TLC was developed in this work to improve AHL separation and detection. The additional separation of AHL using 25% 2-propanol as the second mobile phase coupled with detection using a luminescence-based reporter rendered significant improvement in AHL signal detection and identification. Based on 2-D RP-TLC of extracts prepared from N. subterraneum DSM 12447 strain followed by AHL detection using A. tumefaciens A136 overlay, six distinct putative AHL signals were identified (Figs. 1B–1C).

Whole genome sequencing of N. subterraneum DSM 12447 identified 3 putative luxI homologs (novI) that share a common novRnovIphyH gene synteny

The draft genome of strain DSM 12447 has a GC content of 63.2% and consists of 54 contigs with a total genome length of 4,885,942 bp (N₅₀ of 181,386 bp). Anti-SMASH analysis (Weber et al., 2015) revealed three LuxI-type AHL synthase genes that are separately located in three different contigs (Fig. 2A). Protein alignment of the putative LuxI homologs with known LuxI homologs shows that these homologs contain the highly conserved amino acid signatures which are crucial for the function of AHL synthesis (Asterisk signs in Fig. 2B). We propose the names, NovI_Nsub1, NovI_Nsub2 and NovI_Nsub3 for locus tags NJ75_2841, NJ75_2498 and NJ75_4146, respectively. Among autoinducer proteins within the genus Novosphingobium, NovI_Nsub1, NovI_Nsub2 and NovI_Nsub3 show 62.4%, 51.1% and 62.2% protein identity to LuxI homologs of N. sp AP12 (PMI02_00996), N. sp. Leaf2 (ASE49_1606) and N. sp. AAP1 (IP65_14795), respectively. Beyong the genus Novosphingobium, NovI_Nsub1, NovI_Nsub2 and NovI_Nsub3 show 61.5%, 100% and 60.8% protein identity to the LuxI homologs of Sphingobium sp AP49 (PMI04_04262), Sphingopyxis sp. H050 (ATE67_10720, and Sphingobium japonicum UT26S (SJA_C1-29990)), Analysis of the gene neighbourhood of all three novI genes reveals a conserved novR-novI-phyH arrangement (Fig. 2A). The gene organization of novI_Nsub1 and novI_Nsub3 differ slightly in that novI_Nsub3 contains an additional convergently oriented gene coding for GntR-like transcriptional regulator directly downstream of phyH along with the same other genes of similar composition as in novI_Nsub1 but in the opposite orientation. It is also worth noting that several transposase-coding genes were tightly clustered upstream of novI_Nsub3 suggesting that novI_Nsub3 maybe a result of replicative transposition (Fig. 2B). In addition, a luxR solo was also identified in contig15 based on the presence of several signature domains associated with the canonical LuxR protein IPR005143 (Autoinducer binding), IPR016032 (Signal transduction response regulator, C-terminal effector), IPR011991 (Winged helix-turn-helix DNA-binding domain), and IPR000792 (Transcription regulator LuxR, C-terminal) in the translated protein (Fig. 2B).

Phylogenetic analysis of all functionally validated acyl-homoserine lactone synthases reveals new insight into their evolutionary relatedness

By rooting MAG-14 homologs as the outgroup, Bayesian inference of the newly identified NovI proteins do not exhibit close evolutionary relatedness to any of the selected LuxI homologs and instead occupy a very basal position in the phylogenetic tree. NovI_Nsub1 andNovI_Nsub3 formed a monophyletic clade among themselves with moderate strong posterior probability support (pp = 0.87) (Fig. 3). The relatedness of NovI_Nsub1 and NovI_Nsub3 as observed in the phylogenetic tree is further supported by the conservation in their gene organization (Figs. 3 and 2A). A majority of the functionally validated LuxI homologs isolated from metagenomic libraries did not demonstrate novel phylogenetic position and instead formed monophyletic clustering with known LuxI homologs with strong posterior probability support. CviI from Chromobacterium violaceum shared the most common ancestor with metagenome-derived AubI with maximal posterior probability support while LasI from Pseudomonas aeruginosa is sister taxa to the clades containing AusI, QS6-1 and QS10-1 (pp = 0.91). One notable exception is QS10-S (accession number: ACH69675) that formed a weakly supported (pp = 0.67) monophyletic cluster with the clade containing LuxI homologs from the genera Rhodopseudomonas, Bradyrhizobium and Methylobacterium which also include BjaI, an unusual isovaleryl- HSL synthase.

First functional validation of three LuxI homologs from the family sphingomonadaceae

All three identified NovI homologs led to the accumulation of AHL signals in culture medium when they were heterologously expressed in Agrobacterium tumefaciens strain NTL4 (Fig. 4). Using an inducible expression, the effect of low-, medium- and high-expression on AHL accumulation pattern i.e., detection of additional AHL signals previously not observed in the wild type, can be better studied. In this system, the addition of inducer, IPTG, to the culture medium results in the de-repression of the cloned genes within the cells of the population (Khan et al., 2008). Out of the 20 screened AHL signals, a total of 7 AHL unsubstituted and OH-substituted signals can be identified at the highest IPTG induction of three cloned novI. The signals detected include C8, C8-OH and C10-OH for NovI_nsub1; C8, C10, C12 for NovI_nsub2 ; and C8, C8-OH, C9-OH, C10-OH and C12-OH for Nov_nsub3. It is worth noting that, in the absence of IPTG inducer, basal levels of AHL signals were detected in the growth medium, suggesting suboptimal gene repression by the lacR repressor in Agrobacterium tumefaciens host. In comparison to the 2-D RP-TLC analysis (Fig. 1), LC-MS/MS analysis corroborated the presence of six AHLs and extended it to seven AHLs through inducible heterologous expression of the three individual NovI homologs of N. subterraneum DSM 12447 (Fig. 4).

Dissimilar amount and ratio of major AHLs accumulation in liquid media during the heterologous expression of NovI_Nsub1 and NovI_Nsub3

To quantify the major AHLs (C8, C8-OH and C10-OH) produced by the heterologously expressed NovI proteins, culture extract samples from the three novI homologs cloned in pSRKKm and harboured in A. tumefaciens NTL4 were analyzed by LC-MS/MS alongside prepared samples for a six-point calibration curves ranging from 0.5 µM–20 µM for C8, C8-OH, and C10-OH. AHL signals. C8, C8-OH, and C10-OH, are the signals present in the highest concentrations in culture extracts. This approach identified and quantified the three main signals produced by the three NovI homologs showing that NovI_Nsub2 mainly produces C8 while C8-OH and C10-OH are the major AHLs synthesized by NovI_Nsub1 and NovI_Nsub3. The ratio of C8-OH to C10-OH consistently differ by at least 2-fold in NovI_Nsub1 and NovI_Nsub3 (6.6 vs 2.8) at various IPTG induction concentrations (Fig. 5B). The strong overlap of AHL accumulation profile between NovI_Nsub1 and NovI_Nsub3 provides additional evidence supporting their close evolutionary relatedness.