Bacteria communicate with each other through a phenomenon called quorum sensing (QS), which enables them to detect the cellular density of the population, thus directing proper genotypic and phenotypic adaptations accordingly (Bassler and Losick, 2006). This is achieved through the synthesis, release, detection, and response to small diffusible signaling molecules called the autoinducers (Schauder and Bassler, 2001). A diverse range of phenotypes has been discovered to be regulated by QS, such as biofilm formation, plasmid conjugation, virulence, motility and root nodulation (Williams et al., 2007).

N-acyl homoserine lactone (AHL) is the first autoinducer being studied and to date, it is also the most commonly studied type of autoinducer (Eberhard et al., 1981). The structure of AHL molecules varies, hence, providing signal specificity. However, all of them possess an invariant homoserine lactone ring. The length of the acyl side chain ranging from four to eighteen carbon atoms, the saturation level of the side chain, as well as the substitution group at the third carbon atom by oxo or hydroxyl group contribute to the variation in AHL structures (Whitehead et al., 2001; Williams et al., 2007).

Synthesis of AHL relies on the enzymatic activity of LuxI-type protein, which was first isolated from Aliivibrio fischeri (Engebrecht and Silverman, 1984). This enzyme, in A. fischeri, catalyzes the formation of N-(3-oxohexanoyl)homoserine lactone (OC6-HSL) and hexanoyl homoserine lactone (C6-HSL) using S-adenosylmethionine (SAM) and an acylated acyl carrier protein (ACP) from fatty acid biosynthesis pathway. SAM first binds to the active site of LuxI protein, and the acyl group from acylated ACP is then transferred to this protein. LuxI then catalyzes the formation of an amide bond between the acyl group with the amino group of SAM. Final lactonization results in the production of AHL and the by-product, 5′-methylthioadenosine. The affinity of the LuxI protein to specific acylated ACP gives rise to the product specificity of the enzyme. Interestingly, some LuxI homologs can react with more than one type of acylated ACP, resulting in the synthesis of more than one type of AHL molecules (Schaefer et al., 1996; Parsek et al., 1999).

Similar to other members of the genus Cedecea, Cedecea neteri is lipase positive and resistant to cephalothin and colistin but it does not hydrolyze gelatin or DNA. It differs from other species of Cedecea in that C. neteri is negative for ornithine decarboxylase (Moeller’s), fermentation of raffinose and melibiose, but positive for fermentation of sucrose, D-sorbitol, D-xylose, and malonate utilization, and it grows in media without thiamine. Originally known as “Cedecea species 4,” its name was coined when its role as a human pathogen was recognized in 1982 (Farmer et al., 1982). It was 13 years later that another case of bacteremia caused by C. neteri was reported. As these cases involved old and immunocompromised patients, C. neteri is posed as an opportunistic pathogen (Aguilera et al., 1995). However, the incidence of its isolation is infrequent and its role in human disease is mostly not well understood.

C. neteri has also been isolated from non-human sources such as vegetables and insects (Jang and Nishijima, 1990; Osterblad et al., 1999). In the previous study, we have reported the isolation of a strain of C. neteri, SSMD04, from pickled mackerel sashimi (Chan et al., 2014). There are several characteristics of this bacterium that drew our attention. First of all, it is capable of degrading lipid and thus, presenting potential biotechnological uses (Berman, 2012; Tan et al., 2015). Studies have shown that lipid-degrading microorganisms could be exploited as additive in detergent, lipid modification in the food industry and removal of the pitch from pulp in paper making (Sharma et al., 2001). This phenotype in strain SSMD04, coupled with its isolation from a food source, makes it a potential food spoilage agent. However, this speculation requires further evidence. Nevertheless, limited knowledge on this bacterium makes it an interesting subject of study. Apart from this, C. neteri was reported to exhibit resistance toward a wide range of antimicrobial agents, such as amoxicillin, cephalosporin, aminoglycosides, ampicillin, and others (Dalamaga et al., 2008; Mawardi et al., 2010; Abate et al., 2011). In fact, examination of the genome of strain SSMD04 using Rapid Annotations using Subsystems Technology (RAST) reveals multiple antimicrobial resistance genes toward fluoroquinolones, fosfomycin and β-lactam antibiotics (Chan et al., 2014). Despite all the characteristics discussed, it was when this bacterium exhibits QS activity that spiked our interest for further study as bacterial pathogenicity has been shown to be associated with QS (Rutherford and Bassler, 2012).

In our previous work, we demonstrated that C. neteri SSMD04 exhibits QS activity using biosensors and thus, hypothesized that QS potentially plays a role in the regulation of its virulence and lipolytic pathways. A pair of luxIR homolog was identified from its genome sequences (Tan et al., 2015). In this study, we have cloned the luxI homolog, cneI, and subjected to overexpression in Escherichia coli host and verified that this gene is responsible for the synthesis of the signaling molecule, N-butyryl-homoserine lactone (C4-HSL). The current work facilitates the understanding of the QS mechanism in the bacterium, thus shedding light into its biology.

The protein sequence of CneI was searched against the NCBI database using BLAST and it was found to have a high degree of similarity with several protein sequences. It shows the highest similarity (94%) with LuxI homolog from Klebsiella michiganensis. These LuxI homologs are known to share high homology, particularly the 10 invariant residues, characteristics of LuxI homologs (Fuqua et al., 2001) (Figure 1).

A phylogenetic tree was constructed from the 10 LuxI homologs that share the highest sequence similarity with CneI. As was shown in Figure 2, CneI is most closely related to LuxI homologs of K. michiganensis and Enterobacter sp. Ag1 instead of Citrobacter rodentium as previously thought (Tan et al., 2015). LuxI homologs from the Citrobacter genus seem to belong to a group of its own as they form a cluster, suggesting conservation of LuxI homolog sequence in this genus.

The 636 bp cneI was identified from the whole genome sequence (Chan et al., 2014) and amplified from the genomic DNA of strain SSMD04 by PCR (Figure 3) and cloned into an expression vector, pET28a. This recombinant plasmid was named pET28a_cneI. This gene was cloned upstream of a 6x His-tag sequence under the control of a T7 promoter. Expression of cneI in E. coli BL21(DE3)pLysS was induced by IPTG. Two concentrations of IPTG, 0.5 and 1.0 mM, were used for induction but 1.0 mM of IPTG was later chosen as the optimum concentration (data not shown).

E. coli BL21(DE3)pLysS harboring pET-28a_cneI was shown to induce purple pigmentation in the biosensor C. violaceum CV026, while E. coli BL21(DE3)pLysS harboring pET-28a alone did not activate violacein production in CV026. This strongly suggests the role of cneI in the synthesis of short-chain AHL (Figure 4).

Triple quadrupole LC/MS was then used to determine the AHL profile of E. coli harboring pET28a_cneI after induction with IPTG. The cells were induced with 0.5 and 1.0 mM of IPTG and grown for 8 h after induction at 25°C. The extracted-ion chromatogram (EIC) showed a peak with m/z 172.0 at the same retention time (0.546 min) with that of the synthetic N-butyryl-homoserine lactone (C4-HSL), suggesting the presence of C4-HSL in the organic extract of the induced E. coli. The same result was observed when two different concentrations of IPTG (0.5 and 1.0 mM) were used for induction. However, no AHL was detected in the spent culture supernatant of E. coli harboring pET28a alone (Figure 5). The structure of the AHL produced was later confirmed through the examination of the product ions of the peak found in EIC. The presence of the peak (m/z 102.100) indicates the presence of a homoserine lactone ring moiety, the core structure of AHL molecule.

The recombinant His-tagged CneI protein was later purified and examined by SDS-PAGE (Figure 6). Despite overexpression, the protein profile did not show a very apparent CneI band (Lane 4) compared to the ones without it (Lanes 1–3, and Lane 5). Nevertheless, subsequent purification using nickel-based affinity column showed that the recombinant protein of CneI has a molecular weight of approximately 25 kDa.

N-acyl homoserine lactone-type QS was first discovered in a marine bacterium, A. fischeri in the regulation of bioluminescence (Hastings and Nealson, 1977; Nealson and Hastings, 1979). Decades of research have found a link between QS and many other phenotypes. Virulence and food spoilage traits (Thomson et al., 2000; Weeks et al., 2010; Skandamis and Nychas, 2012) are common QS-controlled phenotypes found in Pseudomonas aeruginosa (Whiteley et al., 1999), Serratia proteamaculans (Christensen et al., 2003), S. liquefaciens (Riedel et al., 2001) and E. carotovora (Pirhonen et al., 1993). A well-exemplified characteristic is the formation of biofilm. Many microbial species encase their colonies within an extracellular matrix called biofilm, which provides resistance toward antimicrobial substances as well as cleaning agents, making them hard to be eradicated (Costerton et al., 1995; Skandamis and Nychas, 2012), thus elevating their virulence and food spoilage potential. It is known that many bacteria such as P. aeruginosa, Aeromonas hydrophila, Burkholderia cepacia, S. liquefaciens, and Pantoea stewartii employ AHL-type QS as a mean of regulation of biofilm formation (Huber et al., 2001; Lynch et al., 2002; Labbate et al., 2004; Koutsoudis et al., 2006).

Besides biofilm formation, many genes related to pathogenicity and food spoilage are cell density-dependent in their expressions, such as virulence determinants, proteolytic or lipolytic enzymes, and biosurfactant. These genes share a similarity in that they synthesize exo-products, which benefit from cell density-dependent regulation. This is because a timed expression of a gene leads to a high production of such products, which inadvertently amplifies the effects brought by the secreted products. In addition, a favorable external environment facilitates the activation of QS circuit, hence, the secretion of exo-products. However, there was an alternative explanation for bacterial cell-to-cell signaling based on this foundation, despite not being the most commonly accepted. This hypothesis, known as diffusion sensing, suggests that bacterial cells use the concentration of signaling molecules to gauge the rate of diffusion of exo-products instead of population density. Therefore, a high extracellular concentration of signaling molecules signifies a favorable environment for the secretion of exo-products as the rate of diffusion is low (Redfield, 2002).

Similar to strain SSMD04, some aforementioned bacteria produce C4-HSL as the signaling molecule as well. A. hydrophila, for example, was reported to possess a luxI homolog termed ahyI, which synthesizes C4-HSL. It has been found that C4-HSL in this food poisoning species is critical in its biofilm formation and extracellular protease production (Swift et al., 1999; Lynch et al., 2002). Both of these traits are relevant to not only to its pathogenicity but also its food spoilage potential. Although QS in A. hydrophila has not been shown explicitly to affect its food spoilage trait, we believe it is likely due to the fact that C4-HSL from this bacterium has frequently been detected in food sources such as aerobically-chilled stored ground beef and commercial bean sprouts. It was reported that C4-HSL was believed to be responsible for food spoilage (Skandamis and Nychas, 2012). Therefore, the discovery of AHL-type QS in strain SSMD04 may suggest the same association in this bacterium.

Apart from that, a close relative of the Cedecea genus, S. liquefaciens, also employs C4-HSL as its signaling molecule. Mutant deficient of QS regulation in S. liquefaciens was found to lose the ability to produce biosurfactant, which is responsible for the swarming motility in Serratia spp. (Lindum et al., 1998). However, QS-regulated genes are not necessarily genes that code for the exo-products. In S. liquefaciens, QS also positively regulates the expression of lipBCD, which encode Lip exporter, a type I secretion protein that mediates the secretion of metalloprotease, lipase, and S-layer protein. A loss in QS regulation led to reduced secretion of metalloprotease and S-layer protein (Riedel et al., 2001). We, therefore, hypothesize that QS in strain SSMD04 is involved in the regulation of exo-products expression or secretion, which is very likely to cause an altered phenotype in its virulence or food spoilage traits. However, this hypothesis is yet to be validated.

The LuxI homolog found in strain SSMD04 was previously thought to be a homolog of CroI found in C. rodentium (Coulthurst et al., 2007) through BLAST search against NCBI database. However, the addition of new genome sequences to the GenBank database changed our perspective. We found another sequence from K. michiganensis which showed a much higher sequence similarity to CneI. K. michiganensis is a new member of the genus Klebsiella first reported by Saha et al. (2013). However, it was not until 2 years later that the genome was sequenced and enabled us to find its LuxI homolog.

The multiple sequence alignment and the phylogenetic tree constructed (Figures 1 and 2) illustrated a high degree of homology and conserved regions among several members of LuxI family. All the strains shared the 10 invariant amino acids which are characteristics of LuxI homologs. This strongly indicates a low rate of random mutation for these autoinducer synthase genes even though they are from different genera. It also shows that these proteobacteria share similar basic QS mechanism and gene regulation in AHL synthesis although they are responsible for different target genes. In fact, random and site-directed mutagenesis demonstrated that the amino acids are important in the binding of acyl substrates to QS signal generators (Parsek et al., 1997). Out of the 10 LuxI homologs that share the highest sequence similarities with CneI, only a few of the QS circuits have known regulatory functions. C. rodentium, for example, uses QS to regulate its surface attachment as its QS mutant adheres less well compared to its wild type counterpart (Coulthurst et al., 2007). On the other hand, gall-forming Pantoea agglomerans utilizes QS system to regulate its gall formation on plants, as the loss of QS regulation led to reduction in the gall size (Chalupowicz et al., 2008). Interestingly, LuxI homologs in both C. rodentium and P. agglomerans produce C4-HSL as the major AHL. Other bacterial species have been reported to possess LuxIR homologs but their regulatory role remains unknown.

The study of QS has been of great interest in that QS has been shown to be a suitable approach in the interruption of the regulated phenotype such as virulence (Rutherford and Bassler, 2012; Singh, 2015). Over the years, a large number of LuxI homologs have been found (Case et al., 2008), but it is not possible to predict the nature of AHL(s) produced by a given LuxI homolog based on its sequence alone. Expressing the luxI homolog in E. coli or comparing the wild type parent against a corresponding luxI mutant has been the only option in order to elucidate the type of AHL(s) produced by a given LuxI (Ortori et al., 2007). In this study, we have successfully cloned cneI and expressed its protein, CneI, in an E. coli host. E. coli does not produce AHL despite possessing an orphan LuxR, the SdiA (Kanamaru et al., 2000), therefore making it an ideal host to characterize CneI. As many LuxI type proteins synthesize more than one type of AHLs (Parsek et al., 1999), it is worth noting that only C4-HSL was produced by strain SSMD04, indicating the specificity of a single AHL in regulating physiological processes and virulence of the bacterium.

Expression of cneI in E. coli leads to the synthesis of C4-HSL, strongly suggests that CneI is indeed the AHL synthase of strain SSMD04. The expression level of cneI under the control of T7 promoter was high such that sufficient C4-HSL was produced to activate the production of purple violacein by the biosensor C. violaceum CV026 without IPTG induction (Figure 4). Purification of CneI and subsequent SDS-PAGE analysis showed that CneI was highly expressed with a molecular weight of approximately 25 kDa, in agreement with the prediction by bioinformatics tools. The expected molecular weight of CneI is 24.9 kDa as predicted from Expasy software². Purification of CneI using Ni-NTA column also enables the protein to be found in soluble fraction, hence, a possible large scale purification of CneI for future work in protein characterization.

This study is the first documentation of cloning and molecular characterization of luxI homolog from C. neteri. The study on the autoinducer synthase gene, cneI, helps in providing further information in the elucidation of QS in C. neteri, as well as paving the way for future studies on the regulatory role of QS in this bacterium. This gene is also a suitable candidate for the interception of QS in this bacterium, providing ways of further research on the regulatory network of QS as well as diversifying the knowledge of the possible roles played by QS. On a final note, this work makes it possible to perform a genome-wide comparative transcriptomics by knocking out cneI for further investigations.

K-HT, K-YH, J-YT, and W-FY performed the experiments and analyzed the data. K-HT and K-YH prepared the draft and K-GC approved the final draft. K-GC conceived the ideas, supervised, applied for funding and monitored the entire project.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.