Role of a GntR-Family Response Regulator LbrA in Listeria monocytogenes Biofilm Formation

The formation of Listeria monocytogenes biofilms contributes to persistent contamination in food processing facilities. A microarray comparison of L. monocytogenes between the transcriptome of the strong biofilm forming strain (Bfms) Scott A and the weak biofilm forming (Bfmw) strain F2365 was conducted to identify genes potentially involved in biofilm formation. Among 951 genes with significant difference in expression between the two strains, a GntR-family response regulator encoding gene (LMOf2365_0414), designated lbrA, was found to be highly expressed in Scott A relative to F2365. A Scott A lbrA-deletion mutant, designated AW3, formed biofilm to a much lesser extent as compared to the parent strain by a rapid attachment assay and scanning electron microscopy. Complementation with lbrA from Scott A restored the Bfms phenotype in the AW3 derivative. A second microarray assessment using the lbrA deletion mutant AW3 and the wild type Scott A revealed a total of 304 genes with expression significantly different between the two strains, indicating the potential regulatory role of LbrA in L. monocytogenes. A cloned copy of Scott A lbrA was unable to confer enhanced biofilm forming potential in F2365, suggesting that additional factors contributed to weak biofilm formation by F2365.


Introduction
Listeria monocytogenes is the etiological agent for listeriosis, one of the foodborne illnesses with high mortality rate [1]. L. monocytogenes is ubiquitous in nature and associated with a variety of foods, from raw fish, fresh produce to processed foods such as ready-to-eat meat, cheese, smoked fish, and milk products. The formation of L. monocytogenes biofilms is considered an important reason for persistent contamination in the food processing environment [2,3,4].
It is well-established that biofilm formation (Bfm) is a microbial protective mode of living, enabling the microorganisms to survive adverse environmental conditions. Various stimuli, such as nutrients, secondary metabolites and various environmental stresses play important roles in Bfm, from development to detachment. Besides Bfm, additional mechanisms, including but not limited to stress responses, spore formation and horizontal gene transmission (HGT) also play critical roles in the ability of microbes to respond to environmental challenges. Various cellsurface components (cell surface proteins, flagella, pili and fimbriae, etc.), extracellular matrices (polysaccharides, proteins, nucleic acids, and lipids, etc.), and enzymes degrading such compounds (such as alginate lyase, surface protein-releasing enzyme SPRE, extracellular proteases, dispersin B, etc.) directly impact biofilms from formation to detachment [5][6][7], and can be considered as primary biofilm attributes. Microorganisms also effectively utilize and coordinate various regulatory and metabolic pathways to respond to environmental stimuli. For instance, Spo0A is a highly conserved transcriptional regulator that plays a key role in initiating sporulation, but it is also involved in Bfm, persistence in host, and competence in several bacteria [8][9][10][11]. Bacteriocin production, biofilm formation and natural competence are responsive to quorum sensing in several Streptococcus spp. [12][13][14][15]. Quorum sensing as well as other regulatory factors [7] can be considered as secondary biofilm attributes. They indirectly interfere with Bfm through affecting the expression of primary biofilm attributes. Understanding the complex metabolic networks impacting the presence and function of both primary and secondary biofim attributes is critical to design targeted intervention strategies.
The involvement of several attributes in L. monocytogenes Bfm has been demonstrated [16]. Flagella and cell motility were found to be involved in both Bfm as well as cell dissociation from the biofilm community [17,18]. A cell surface protein containing the LPXTG cell wall anchor domain, BapL, affected Bfm on stainless steel and polystyrene surfaces by L. monocytogenes serotype 1/2a. However, although the BapL homolog functions as a virulence factor in S. aureus, this was not the case with BapL of L. monocytogenes [19]. In addition to these primary Bfm attributes, several regulatory factors involved in quorum sensing and stress responses also affect L. monocytogenes Bfm. For instance, the two component system DegS/DegU modulates flagellum formation and pellicle (a biofilm on the surface of broth) formation in Bacillus subtilis [20]. While L. monocytogenes only has the response regulator DegU but not the sensor kinase and DegU regulates its own transcription by directly binding to its promoter region, the degU deletion mutant did not form biofilms and the phenotype could be complemented by a functional copy of degU [20]. As in S. aureus, an AgrD-dependent quorum sensing system in L. monocytogenes has also been shown to play a role in both biofilm development and the expression of inlA [21]. Sigma B (s B ), the SOS response factor YneA, the response regulator HrcA, the chaperone DnaK and the master regulator of virulence PrfA were also found involved in L. monocytogenes Bfm [22,23]. Zhu et al. [24] reported that a putative ABC transporter permease was a negative regulator of L. monocytogenes Bfm. A follow-up study further indicated that the permease affected the expression of several genes including those encoding the cell surface protein Dlt, the cell surface anchor protein SrtA and a GntR family transcriptional regulator (LMOf2365_2274), etc. [25]. It is worth noting that a GntR family regulator was also found to be involved in Bfm in Enterococcus (ebrA, [26]). However, the exact mechanisms of the regulatory factors in Bfm are yet to be revealed.
It is recognized that strains of L. monocytogenes vary in their ability in Bfm. Marsh et al [27] found that unlike strong biofilm forming (Bfm s ) L. monocytogenes strain Scott A, strain F2365 exhibited weak biofilm forming (Bfm w ) phenotype under the same experimental conditions. The observed differences between these two L. monocytogenes strains in Bfm provided an opportunity to examine molecular attributes of importance to Bfm. Thus the objective of this study was to elucidate molecular determinant(s) potentially involved in L. monocytogenes Bfm, initiated by comparing the transcriptome of ScottA and F2365.

Bacterial Strains and Growth Conditions
The strains and vectors used in this study are listed in Table 1 and Table 2, respectively. Frozen L. monocytogenes cultures were activated by incubating in trypticase soy broth supplemented with 0.5% yeast extract (TSBYE) at 37uC for 18 h, and transferred at least once prior to assessments. Besides Brain Heart Infusion media, E. coli strains were also grown in LB broth (lysogeny broth) [28] and L agar plates.

Bfm Assessments
For the rapid crystal violet attachment assay, 50 mL of an overnight culture were inoculated into 4 mL of TSBYE in polypropylene culture tubes and incubated at 37uC for 24 h. After incubation, the broth was removed and 4 mL of 0.1% crystal violet (Sigma) was carefully added to each tube and incubated at room temperature for 30 min. The crystal violet was removed, the culture tubes were washed, and the dye was extracted with 95% ethanol as previously described [27]. The test was performed in three biological replicates with three technical replicates. Tukey's HSD test was run using SPSS 19.0 software (Chicago, IL).
Standard SEM assessment was also performed as describes previously with slight modification [27], using stainless steel sheet (type 316) cut into coupons with K inch diameter (1mm) at the OSU chemistry machine shop. Basically, one disc was placed in each well of a 24-well microtiter plate, filled with 2 mL of fresh TSBYE and inoculated with 100 mL of overnight culture of each strain to be examined. The microtiter plate was incubated at 37uC for 18 h. After dehydration, the microtiter plate was wrapped in parafilm and stored in a dessicator until SEM examination. During the day of SEM assessment, the coupons were sputter coated with gold palladium for 70 sec, and the samples were observed using an ESEM XL-30 (FEI, the Netherlands).

Microarray Assessment
RNA was extracted using the RNeasy mini kit (Qiagen, Valencia, CA) following manufacturer's procedure with slight modification. Briefly, 23.75 ml of fresh TSBYE was inoculated with overnight cultures at 2.5% and grown at 37uC for 6 h until reached at OD 600 of 0.6-0.7 and Log CFU between 8.5 to 9.5. One mL of aliquots were removed and placed in 4uC refrigerator for 15 min, and the cells were collected by centrifugation at 80006g for 5 min. The cells were resuspended in 100 mL of TE buffer and 6 mL of lysozyme (50 mg/mL) followed by incubation at 30uC for 30 min. The rest of the procedures followed manufacturer's instruction.The extracted RNA was used within 30 min of extraction or stored at 280uC until use.
The Standard Operating Procedures for Aminoallyl Labeling of RNA (ftp://ftp.jcvi.org/pub/data/PFGRC/MAIN/pdf_files/ protocols/M007.pdf) and Hybridization of Labeled cDNA probes (ftp://ftp.jcvi.org/pub/data/PFGRC/MAIN/pdf_files/protocols/ M008.pdf) by the J. Craig Venter Institute were followed, using the L. monocytogenes microarray slides (version 3.0,, including 4b F2365, 4b H7858, 1/2a F6854 and EGD-e). Scanned data from dried slides were normalized and statistical analysis was conducted using the TM4 microarray software suite from J. Craig Venter Institute. Briefly, Midas was used to normalize the microarray using LOWESS normalization. The data were then entered into Multi Experiment Viewer (MEV) where T-Tests were performed using a P-value of P = 0.05 [29].

Plasmid Extraction and Transformation
L. monocytogenes plasmid extraction was performed as described by Anderson and McKay [30]. Plasmid extraction from Escherichia coli was conducted using QIAprep Spin Mini and Midi prep kits (Qiagen) following the manufacturer's protocol.
Competent cells of L. monocytogenes were prepared following the protocol of Park and Stewart [31]. For electroporation, 40 ml of the competent cells were mixed gently with 1 mg of plasmid DNA following standard procedures [32]. The cells were plated on multiple TSAYE plates containing the proper antibiotic and incubated at 37uC for two to three days.

Deletion Mutant Construction
An in-frame deletion of the gene lbrA, a GntR-family response regulator was constructed by homologous recombination. The primer pair LbrA mut-1 (59GCGAATTCCAAAGTGACCAAG-GATACAGTGA39) and LbrA mut-2 (59GCGGTACCCA-TAAATGATTCCCCTCTCTCTCTA39) were used to synthesize a PCR fragment A containing the start codon of the lbrA gene and 462 bp, 59 of the start codon, with EcoRI and KpnI restriction sites at the 59 and 39 end, respectively. Primer pair LbrA mut-3 (59GCGGTACCAAATGTTTTCAACACATAAT-GAA39) and LbrA mut-4 (59GCGGATCCAATCAATCGT-CACGGCATAAGA39) were used to synthesize a PCR fragment B containing the last 48 bp of the lbrA gene and 396 bp downstream of the lbrA gene, with KpnI and BamHI sites on the 59 and 39 end, respectively. The in-frame lbrA deletion fragment was amplified by PCR using primers LbrA mut-1 and LbrA mut-4, and the ligation product of the KpnI-digested fragments A and B as the template. The product was cloned into the temperature sensitive plasmid pKSV7 between EcoRI and BamHI sites, resulting pKSV7DlbrA. The recombinant plasmid was transformed into E. coli DH5a, resulting strain AWD2. The pKSV7DlbrA was further electroporated into Scott A. After incubating in BHI+sucrose (0.5M) at 30uC for 2 h, the transformants were recovered on BHI+chloramphenicol (Cam; 10 mg/ mL) plates, at 30uC for 48 h. Chromosomal integration of pKSV7DlbrA was encouraged by 3 consecutive passage of the transformants, each in BHI+Cam broth at 40uC with shaking overnight. The mixture was streaked on BHI+Cam plates at 40uC overnight. Single colonies were inoculated into 2 mL of BHI without cam at 30uC overnight, consecutively passed 5 more times at 30uC for plasmid excision, followed by further passage in BHI at 40uC for seven times without cam to enrich the plasmid-free subpopulation. The cultures were then streaked onto BHI plates and incubated at 37uC for single colonies. Thirty-three recovered single colonies were picked and patched onto BHI+cam and BHI plates. Cam-sensitive colonies were screened by PCR for DlbrA mutants using the primer pair LbrA-mut1 and LbrA-mut4, as well as a third set of chromosomal specific primers LbrA-FPseq (59TTGTTAGGTAATTTTTCAGGTG39) and LbrA-RPseq (59CATATTTTCTTTTACTTTCGTCTC39). One DlbrA mutant, designated AW3, was used in subsequent studies.

Mutant Complementation
Full length lbrA was amplified by PCR using Scott A DNA as the template and primers FPpMSP122 (59GCTCTA-GACTCGGCTTAACAGCTATTGG39) and 514pMSPRP (59GGACTAGTCCAGCATGATAATCACCC39), and cloned into the copy control vector pCC1 (Epicentre Biotechnologies, Madison, WI) to maintain a copy of functional lbrA in E. coli. The functional lbrA insert was removed from the recombinant pCC1lbrA by digesting with BamHI (vector) and SpeI (reverse primer), purified by gel extraction, cloned into the inducible expression vector pMSP3535 [33], and then transformed into E.coli pcn cells. The recombinant plasmid pMSP3535lbrA was extracted from E. coli cells and electroporated into L. monocytogenes F2365 and AW3, resulting in AWF2 and AW5, respectively. To induce the expression of lbrA, cultures were incubated at 37uC for 3 h followed by the addition of nisin to a final concentration of 25 ng/mL as described by [34]. The cultures were then incubated at 37uC for an additional 16 h.

Reverse Transcription PCR
One-step RT-PCR was conducted using superscript transcriptase III (Invitrogen) in accordance to the manufacturer's protocol. The lbrA specific primer pair FPlbrA (59CGTCGCATTTA-TAGGTAAG39) and RPlbrA (59TCATTGCGTTCAT-TATGTG39) was used to synthesize an approximate 200 bp RT-PCR product. The inlA-specific primers (59TCAGTCAA-TAAATTCCCAGC39 and 59CCACTTAAGGCAATTTT-TAATG39, [35]) with an approximate 100 bp RT-PCR product were used as an internal control.

Transcriptome Comparison of L. monocytogenes Scott A and F2365
Strains Scott A and F2365 exhibited similar growth performance in the bacterial media used in the study (data not shown). The cells were harvested during stationary phase, with the logCFU around 9 and 10, and the difference between the two strains within one log. The whole transcriptome comparison between Bfm s Scott A and Bfm w F2365 revealed that 951 genes were differentially expressed at a P value of 0.05. Of those 951 genes, 515 were expressed at higher level and 436 were expressed at lower levels in Scott A than in F2365, respectively (Table S1). One of the genes expressed at a significantly higher level (8 fold increase) in strain Scott A compared to the expression in strain F2365 was LMOf2365_0414, a GntR-family response regulator gene designated as lbrA (Listeria biofilm regulator A), was chosen for further assessments of its potential impact on L. monocytogenes Bfm.

Impact of LbrA on L. monocytogenes Biofilm Formation
In agreement with a previous report [27], Bfm of Scott A was found to be significantly different from that by F2365 (Fig. 1A,  Fig. 2 and Fig. 3). The rapid crystal violet staining assay illustrated that the lbrA deletion mutant strain AW3 exhibited significantly decreased Bfm as compared to the parental strain Scott A (Fig. 1A). SEM assessment confirmed that although structurally the biofilm of the mutant AW3 was similar to that of Scott A, the depth and overall biofilm was less than observed as in the parental strain (Fig. 2). Meanwhile, Bfm was restored in the genetically complemented lbrA deletion mutant AW5 by both rapid crystal violet staining assay (Fig. 1B) and SEM (Fig. 2) in the presence or absence of nisin induction. The results suggested that LbrA has a role in L. monocytogenes Bfm.
L. monocytogenes AWF2, a derivative of F2365 with a functional lbrA in the nisin inducible expression vector pMSP3535, appeared to have slightly increased Bfm by rapid crystal violet staining assay (Fig. 1C) and SEM (Fig. 3) in the presence or absence of nisin, compared to that by the parental strain F2365. However, at pvalue 0.05, the difference by rapid crystal violet staining assay (Fig. 1C) was not significant between F2365 and the derivative. The observation was further validated by SEM, where the attached communities were thin and sporadic (Fig. 3). The attached structure by AWF2, in the presence and absence of nisin, however, seemed to have a higher density than that by vector control AWF1 and the parental strain F2365.
Since the difference in Bfm at the presence and absence of nisin by the lbrA complementing derivative AW5 was minimal, RT-PCR was conducted to examine the expression of the cloned lbrA in AW5. Similar expression of lbrA was observed in strain AW5 with and without nisin (Fig. S1), suggesting that the nisin induction system of pMSP3535 was not functional in this construct.

Transcriptome Assessment of Scott A and the Isogenic Mutant AW3
Since lbrA likely is a transcriptional regulator, a transcriptome analysis of Scott A versus the isogenic mutant AW3 was conducted to identify additional genes including primary Bfm factors potentially affected by lbrA deletion. The results showed that the expression of 304 genes was significantly different between the two strains at a P-value of 0.05. Of those 304 genes, 116 were expressed at higher levels in Scott A. The other 188 genes were expressed at higher values in the mutant strain AW3 (Table S2). An ABC sugar transporter permease protein and multiple hypothetical proteins were among the genes with a higher expression in Scott A than in AW3.

Discussion and Conclusion
Bfm is a specific mode of microbial living resulting from coordination of the metabolic network in responding to environmental challenges. The involvement of several regulatory factors in Bfm has been illustrated in L. monocytogenes. Besides the wellestablished cell-cell communication systems [16], the virulence regulator PrfA also affects Bfm [23]. Interestingly, expression of PrfA had no impact on Bfm in the non-pathogenic L. innocua strain, originally exhibiting sparse small clumps instead of biofilm network [36]. Because PrfA likely serves as a secondary biofilm determinant (regulator), the phenomenon may be due to the lack of the corresponding primary biofilm determinant in the nonpathogenic L. innocua strain. In fact, Travier et al [37] recently reported that the virulence factor ActA, as part of the PrfA regulated virulence gene cluster, is a critical determinant in L. monocytogenes biofilm formation and intestinal colonization. Our study further revealed the positive role of lbrA, a gntR-family response regulator, and potentially the corresponding metabolic segment under its regulation, in L. monocytogenes Bfm.
After comparing the transcriptome of Bfm s strain ScottA and Bfm w F2365, the gntR-family response regulator gene lbrA highly expressed in Scott A was chosen for further investigation, as a positive involvement of a GntR family regulator gene in biofilm formation in Enterococcus EF1809 (enterococcal biofilm regulator ebrA ) was recently reported [26]. The lbrA gene is well-conserved in L. monocytogenes genomes. In F2365, it is followed by coding sequences for a putative protein and components of an ABC transporter and PTS system. This gene organization is observed in multiple L. monocytogenes strains with published sequences, such as 4b LL195, 7 SLCC2482, 1/2b SLCC2755, 4e SLCC2378, 4b L312, 4d ATCC 19117, 3b SLCC2540.
As illustrated in Fig. 1, inactivation of lbrA led to deficiency in biofilm formation by the L. monocytogenes strain Scott A DlbrA mutant AW3. The derivative formed much weaker biofilm compared to the parental strain, with collapsed secondary structure of the biofilm, leaving large, unevenly spaced clumps of cells attached to the stainless steel coupon. Meanwhile, AW5, the isogenic derivative of AW3 complemented with a functional lbrA, was able to form the characteristic honeycomb-biofilm structure. While the density of the biofilm by AW5 was less than that by Scott A, the structures of the biofilm by the two strains were similar. The data suggested that lbrA is involved in L. monocytogenes Bfm.
The nisin inducible expression vector pMSP3535 has been successfully used in several studies involving Gram+ organisms, where the expression of the target gene can be turned on and off in the presence and absence of nisin due to the tight regulation of the nisA promoter [34,[38][39][40]. However, although lbrA was cloned downstream of the nisA promotor into the expression vector pMSP3535, AW5 formed honeycomb biofilm structure with and without nisin. Similar results were obtained by the rapid crystal violet biofilm assessment. RT-PCR results showed that lbrA was expressed in AW5 in both cases, suggesting the nisin-induced promoter was constitutive in L. monocytogenes.
Since lbrA likely served as a regulator instead of primary biofilm attribute, a microarray comparison of Scott A and the isogenic mutant AW3 was conducted. The expression of 304 genes was significantly different between the two strains. Among them, there were many hypothetical proteins, a sugar transport permease protein, and an ABC transporter. Further studies need to be conducted to reveal primary Bfm attributes among the candidates.
Meanwhile, as illustrated in Fig. 1C and Fig. 3, the F2365 derivative with the recombinant pMSP3535lbrA (AWF2) did not  exhibit strong biofilm formation, although its attached structure on stainless steel coupon was more complex than that of the parental strain F2365. The result suggested that LbrA was not sufficient to complement the biofilm defect in F2365. This is consistent with the microarray data showing that 304 genes were found with significant difference in expression between ScottA and its lbrA deletion mutant AW3, roughly one third of the genes with differential expression between Scott A and F2365. Therefore further studies are needed to reveal additional factors involved in Bfm in L. monocytogenes.
It is worth noting that the derived amino acid sequences of lbrA from ScottA and F2365 are 100% identical. In addition, the DNA sequences of the putative lbrA promoter region of Scott A and F2365 are identical. The only differences between the two strains in the immediate regions flanking lbrA are 2 base pairs within the gene itself and 2 downstream of the gene. Despite the fact that introducing an additional copy of lbrA via the recombinant pMSP3535lbrA was not sufficient to complement the Bfm defect in F2365, the observations made with the parent, ScottA, its isogenic mutant and the complemented mutant suggests that lbrA expression can be a critical control point affecting Bfm. It is established that the activities of GntR family regulators can be modulated in response to diverse small molecules, such as histidine (HutC), fatty acids (FarR), sugars (TreR) and alkylphosphonate (PhnF) [26,41]. It may be worthwhile to identify lbrA-responsive small molecules, and to examine their potential impact on Listeria Bfm. Figure S1 Reverse transcription PCR products of L. monocytogenes strains with lbrA and inlA-specific primer pairs. Lanes 1 and 2: Scott A; lanes 3 and 4: AW3; lanes 5 and 6: AW4; lanes 8 and 9: AW5 with nisin induction; lanes 10 and 11: AW5 in the absence of nisin. Lane 7:100 bp ladder, invitrogen. (TIF)