Cloning and expression of a novel lactococcal aggregation factor from Lactococcus lactis subsp. lactis BGKP1

Background Aggregation may play a main role in the adhesion of bacteria to the gastrointestinal epithelium and their colonization ability, as well as in probiotic effects through co-aggregation with intestinal pathogens and their subsequent removal. The aggregation phenomenon in lactococci is directly associated with the sex factor and lactose plasmid co-integration event or duplication of the cell wall spanning (CWS) domain of PrtP proteinase. Results Lactococcus lactis subsp. lactis BGKP1 was isolated from artisanal semi-hard homemade cheese and selected due to its strong auto-aggregation phenotype. Subsequently, non-aggregating derivative (Agg-) of BGKP1, designated as BGKP1-20, was isolated, too. Comparative analysis of cell surface proteins of BGKP1 and derivative BGKP1-20 revealed a protein of approximately 200 kDa only in the parental strain BGKP1. The gene involved in aggregation (aggL) was mapped on plasmid pKP1 (16.2 kb), cloned and expressed in homologous and heterologous lactococci and enterococci. This novel lactococcal aggregation protein was shown to be sufficient for cell aggregation in all tested hosts. In addition to the aggL gene, six more ORFs involved in replication (repB and repX), restriction and modification (hsdS), transposition (tnp) and possible interaction with mucin (mbpL) were also located on plasmid pKP1. Conclusion AggL is a new protein belonging to the collagen-binding superfamily of proteins and is sufficient for cell aggregation in lactococci.


Background
The aggregation phenomenon in lactococci has been studied for more than thirty years. It was found to be directly associated with a sex factor and lactose plasmid co-integration event [1] or duplication of the cell wall spanning (CWS) domain of PrtP proteinase [2]. Lactose plasmid conjugation in Lactococcus lactis 712 and in the related strains C2 and ML3, frequently involves plasmid co-integration with a sex factor. Moreover, this phenomenon is often associated with a cell aggregation phenotype and high frequency transfer ability [3][4][5]. The lactococcal sex factor exists integrated in the chromosome [6], although it can be excised as a closed circular form and lost from the cell [1]. Deletion and overexpression experiments confirmed that CluA is the only sex factor component responsible for aggregation in L. lactis. This 136 kDa surface-bound protein, encoded by the chromosomally located sex factor of Lactococcus lactis subsp. cremoris MG1363, is associated with cell aggregation linked to high-frequency transfer [7]. Two domains of CluA involved in distinct functions were determined. The region from D153-I483 is important for promoting cell-to-cell binding (aggregation), whereas K784-K1056 Tra domain is involved in DNA transfer and responsible for high conjugation frequency [8]. Furthermore, the aggregation ability of L. lactis subsp. lactis BGMN1-5 and its cured derivative was dependent on the presence of the plasmid encoded extracellular proteinase, PrtP [2,9]. The PrtP proteinase of BGMN1-5 contains a duplication of the C-terminal cell wall spanning domain (CWS). Experiments in which hybrids of BGMN1-5 PrtP, containing one or more CWS domains were constructed, showed that only cells producing a fusion protein with two or more CWS domains sedimented. Sedimentation resulted from specific interaction between CWS domains [2]. It is interesting that both, CluA protein and PrtP proteinase, have an LPXTG pentapeptide at the carboxy terminus, which is conserved among many cell surface proteins of Grampositive bacteria [10]. In Gram-positive bacteria, these proteins have a multitude of functions, which include binding to host cells and/or tissues or specific immune system components, protein processing, nutrient acquisition and interaction between bacteria during conjugation [11]. Many cell-surface proteins are involved in aggregation and adhesion processes, including the colonization of oral and commensal bacteria [12][13][14] and initiation of infection by pathogens [15][16][17][18][19]. Pathogenic Gram-positive bacteria express cell surface proteins that contribute to virulence [20]. The genes encoding the surface proteins derived from several Enterococcus faecalis plasmids, including pAD1, pPD1 and pCF10 have been sequenced [21][22][23] and over-expressed in different bacteria including Lactococcus lactis [24]. It was found that aggregation substance (AS), a surface protein of E. faecalis, might contribute to virulence [25].
L. lactis subsp. lactis BGKP1 was selected in our laboratory as a specific strain due to the expression of a unique auto-aggregation phenotype. In the present study, we isolated a non-aggregating derivative (Agg -) of BGKP1 and performed comparative analysis. We found that a cell surface protein of high molecular mass, around 200 kDa, is responsible for the aggregation. The gene encoding for aggregation protein (aggL) was mapped on plasmid pKP1 (16.2 kb). The gene was cloned, sequenced and expressed in homologous and heterologous lactococcal and enterococcal hosts, showing that AggL protein is responsible for cell aggregation in lactococci. Therefore, we propose AggL as a novel lactococcal aggregation factor.

Results and Discussion
Aggregation may play the main role in adhesion of bacteria to the gastrointestinal epithelium and their colonization ability, as well as in probiotic effects through coaggregation with intestinal pathogens and their subsequent removal.
Isolation and comparative analyses of Lactococcus lactis subsp. lactis BGKP1 and its non-aggregating derivative BGKP1-20 Considering the importance of aggregation, Lactococcus lactis subsp. lactis BGKP1 was selected during the characterization of microflora of artisanal white semi-hard homemade cheeses manufactured in the village of Rendara (altitude 700 m) on Kopaonik mountain, Serbia. Among 50 lactic acid bacteria (LAB), Lactococcus lactis subsp. lactis BGKP1 was chosen for further study due to its strong auto-aggregation phenotype (Agg + ).
BGKP1 is a lactose positive, bacteriocin and proteinase non-producing strain. The aggregation phenotype may be observed after vigorous mixing of a stationary phase culture, when snowflake-like aggregates become visible ( Figure 1). The aggregates of BGKP1 cells differed in appearance from those of L. lactis subsp. cremoris MG1363 expressing CluA or L. lactis subsp. lactis BGMN1-5. Aggregates rapidly sedimented under resting conditions and more than 95% of BGKP1 cells aggregated in the first minute, as observed by the decrease of cell suspension absorbance (data not shown). BGKP1 cell aggregates resemble those of Lactobacillus paracasei subsp. paracasei BGSJ2-8 [26]. The aggregation ability of BGKP1 was lost spontaneously after transfer of cells from -80°C to 30°C, with a frequency of 5% to 10%, as previously shown for BGSJ2-8 [26]. The resulting nonaggregating derivative (Agg -) of BGKP1 was designated as BGKP1-20. Agg + cells formed smaller and prominent colonies, whereas Aggderivatives showed flat colonies on agar plates. Mutations in genes encoding biofilmassociated proteins were also shown to result in transformation of colony morphology [27]. Since BGKP1 and BGKP1-20 were not able to form biofilms on plastic tissue culture plates, the aggregation phenomenon present in BGKP1 is most probably not linked to biofilm formation. Spontaneous high-frequency loss of the trait indicated a plasmid location of the gene(s) encoding the aggregation phenotype. Comparative analysis of the fitness of BGKP1 and BGKP1-20 in the exponential phase of growth showed differences in generation times between the strains. The doubling time for BGKP1 was 54.4 min (specific growth rate = 1.103/h), while that for BGKP1-20 was 50.2 min (specific growth rate = 1.195/ h). The presence of the aggregation phenotype resulted in a significantly prolonged doubling time for BGKP1 (approximately 8.5%) when compared with that of BGKP1-20. Taking into consideration that bacteria maintain and procure gene coding for the aggregation factor in spite of the energy cost, we could hypothesize that this feature provides some benefit for the cell.

Nature of molecules involved in aggregation
The spontaneous loss of the capacity to aggregate in BGKP1 was tested under various conditions. Aggregation capacity was found to be reversibly lost after repeated washing of BGKP1 cells with bi-distilled water. Nevertheless, when washed BGKP1 cells that had lost the Agg + phenotype were re-suspended in the wash material, they re-gained the ability to aggregate. Obviously, a some molecule(s) with a role in aggregation were washed from the cell wall. However, aggregation was not observed when BGKP1-20 Aggcells were resuspended in wash material from BGKP1 Agg + . To check the nature of molecules involved in the aggregation, BGKP1 Agg + cells were treated with proteinase K prior to washing by water. The wash material of proteinase K-treated cells did not restore the aggregation ability of BGKP1 Aggwashed cells. Results indicated that the aggregation factor is of proteinaceous nature. Since a protein is involved in aggregation, the influence of various pH levels and the concentration of five ions (K + , Na + , Ca ++ , Mg ++ and Fe +++ ) on this phenomenon was examined. It was found that pH did not have as strong impact on the ability of BGKP1 to aggregate as cations did, especially iron. The presence of 1 mM FeCl 3 promoted aggregation of BGKP1 washed cells.
Cell surface protein profiles of BGKP1 and the Aggderivative BGKP1-20 were compared in order to detect any differences between strains. As demonstrated for BGSJ2-8 [26], the SDS-PAGE pattern of cell surface proteins from BGKP1 and BGKP1-20 differed. Thus, Agg + contained an additional ≈200 kDa protein, which was absent from the BGKP1-20 Aggderivative ( Figure 2). This suggested that the aforementioned protein might be responsible for the aggregation. The protein detected and potentially involved in the aggregation of L. lactis subsp. lactis BGKP1 had a slightly smaller molecular mass than that of L. paracasei subsp. paracasei BGSJ2-8.

Localization and cloning of genes linked to the aggregation phenomenon
Plasmid profile analysis (of non-digested and digested plasmids with different restriction enzymes) of parental strain BGKP1 and the Aggderivative BGKP1-20 showed differences in one plasmid designated as pKP1, indicating its potential role in the expression of the aggregation phenotype ( Figure 3).
In order to facilitate cloning and expression of gene (s) responsible for the aggregation phenotype in homologous and heterologous hosts, new lactococcal-E. coli shuttle cloning vectors pAZIL and pAZILcos, based on pACYC184 [28] and pIL253 [29] were constructed [see Additional File 1]. These vectors enabled cloning of large DNA fragments (entire pKP1 -16.2 kb), blue-white selection for the inserted fragments and high stability of the constructs. The plasmid library of pKP1 constructed in pAZIL enabled sequencing and subsequent in silico analysis of the obtained sequence.

Sequence analyses of plasmid pKP1
The complete sequence of plasmid pKP1 was found to consist of 16181 bp, with a G+C content of 35.94%. Within the 4380 bp long nucleotide sequence of pKP1 (region 15394-1-3593), a 99% identity with the pSRQ900 plasmid of Lactococcus lactis (GenBank Accession No. AF001314) was determined. This sequence represented approximately one fourth of the pKP1 nucleotide sequence. This region encompassed the origin of replication, repB gene, repX replication associated gene and putative hsdS gene (Figure 4). The rest of the nucleotide sequence (three quarters of pKP1) did not share identity with pSRQ900 and carried three genes, including two new genes (aggL and mbpL) and one known transposase gene, which implies its novelty. In addition, seven open reading frames (ORF) were revealed in pKP1 by application of the DNA Strider program (Table 1, Figure 4). It is interesting that all ORFs on the plasmid were in the same orientation except the gene encoding for transposase. Two potential ORFs, designated as aggL and mbpL, were additionally analysed. BLAST search revealed that aggL gene showed no similarity with any of the genes from the NCBI BLAST database, while AggL protein shared 51% identity with a hypothetical protein from Oenococcus oeni AWRIB429 (Table 1). The nucleotide sequence of mbpL shared 84% identity with ORF from Leuconostoc citreum KM20 plasmid pLCK1 and 53% amino acid identity with a protein from Enterococcus faecalis TX1322 (Table 1). Motif Scan http://myhits.isb-sib.ch/cgi-bin/motif_scan and DAS Transmembrane Domain [30] programs were used to analyse their potential protein products. It was revealed that AggL included several motifs important for cell adhesion, such as a collagen-binding domain with a jelly-roll fold (C-terminus), CnaB-like domain (Cterminus) as well as serine and threonine-rich domains (N-terminus). MbpL contained a MucBP-like domain and YSIRK-signal. Both AggL and MbpL were predicted to have the Gram-positive cocci cell wall anchoring domain (LPXTG) and two transmembrane domains (by using strict cutoff). Additionally, both proteins had short amino acid repeat regions at the N-terminus, serine and threonine rich regions for AggL and an alanine rich Primary structural analysis of AggL revealed domain organization similar to LPXTG proteins of Gram-positive cocci. The LPXTG motif is a highly conserved part of the C-terminal sorting signal and it plays a role in the covalent linkage of many cell-wall-associated surface proteins to the nascent pentaglycine crossbridge in peptidoglycan [22]. For example, S. aureus is known to express 21 proteins with the LPXTG motif including two clumping factors ClfA and ClfB [20,31]. Another characteristic of AggL primary protein structure is modular architecture and a number of repeat regions that share high mutual identity (98-100%). Previous studies on staphylococcal LPXTG proteins indicated modular architecture and B repeats as their specific characteristics. Such organization could have arisen during evolution through the acquisition of distinct domain-sized polypeptides of which some have expanded by duplication and homologous recombination [31]. Collagenbinding protein B domain (CnaB domain) is the most abundant domain of AggL. Such a structure might mediate bacterial adherence to collagen. Repeated units have been suggested to serve as a 'stalk' that projects the region crucial for adherence to the bacterial surface, thus facilitating bacterial adherence to collagen. Additionally, the N-terminal serine and threonine rich domains of AggL could play a role in aggregation, since it is known that such domains of CD46 protein promote   [32]. Interestingly, the YSIRK domain, another characteristic of staphylococcal LPXTG proteins, was not found in AggL, although it was present within the signal peptide of MbpL. The requirement of a YSIRK motif for efficient secretion implies the existence of a specialized mode of substrate recognition by the secretion pathway of Gram-positive cocci. However, this mechanism is not essential for the surface protein to anchor to the cell wall envelope [33]. Considering the primary protein organization of MbpL, its role in the cell could most likely be interaction with gastrointestinal epithelial cells. Interestingly, the search for lactococcal proteins similar to AggL and MbpL against the NCBI BLAST database revealed that AggL shared identity only within its Nterminal region (encompassing transmembrane domain, serine and threonine rich domains, collagen binding domain and WD repeats). On the other hand, MbpL shared identity within its C-terminal region (encompassing the MucBP-like domain including 36 aa repeats, the transmembrane domain and the G+ anchoring domain).

Homologous and heterologous expression of the aggL gene
Cloning of different fragments of plasmid pKP1 into the pAZIL vector ( Figure 5) and transfer of constructs into three lactococcal strains (homologous Aggderivative BGKP1-20, and heterologous MG1363, BGMN1-596 and Enterococcus faecalis BGZLS10-27) revealed that the aggL gene is sufficient for the aggregation phenotype ( Figure 1). Furthermore, aggregation of enterococcal cells carrying the aggL gene was observed, but the intensity of cell aggregation was lower than that obtained in lactococci (data not shown).
This conclusion was confirmed by transformation of the same lactococci with two types of constructs: pAZIL harboring pKP1 linearized in the aggL gene, that results in the inactivation of this gene (construct pAZIL-KPSl8) and by constructs carrying the DNA fragment of pKP1 containing solely the aggL gene (for example pAZIL-KPPvSc1) ( Figure 5). It was noticed that cell aggregation phenotypes of MG1363 and BGKP1-20 transformants, carrying the aggL gene, were identical to those of the parental strain BGKP1. Transformants of BGMN1-596 showed the aggregation phenotype with slightly different cell aggregates, which were smaller than in BGKP1 (Figure 1).
The location of the gene involved in the aggregation of BGKP1 on plasmid pKP1 potentially enables transfer of this factor through the microbial population.
Experiments with heterologous expression of aggL and/ or mbpL revealed the main role of AggL protein in the aggregation phenomena. According to the morphological characteristics of cell aggregates in heterologous strains, we can assume that even though AggL is crucial for aggregation, some additional protein(s) (like MbpL) might have a modulatory effect on the aggregation phenotype. Additionally, preliminary ex vivo experiments with rat colon sections indicated that AggL is not involved in adhesion to the gastrointestinal epithelium (data not shown).
Further experiments will be focused on studies of AggL and MbpL interactions with human epithelial cells and their role in the adhesion and possible probiotic potential of BGKP1. Moreover, co-aggregation with various pathogenic bacteria will be also tested.

Conclusions
We have demonstrated that in lactococci, a novel aggregation-promoting factor AggL is encoded by the aggL Figure 5 Linear physical map of pKP1 and the scheme of constructed clones in the pAZIL cloning vector used for homologous and heterologous expression of aggregation phenotype. Relevant restriction sites are indicated. Restriction enzymes with a single recognition site are given in bold. Bold arrows indicate the size and orientation of predicted ORFs. + -construct with aggregation ability; --construct with no aggregation ability. gene located on the 16.2 kb pKP1 plasmid. Moreover, functionality of aggL was confirmed by homologous and heterologous expression of different clones containing or lacking this gene in the newly constructed shuttlecloning vector, pAZIL.

Growth kinetics
Cultures of BGKP1 and BGKP1-20 were grown in 10 ml of GM17 to a density of 10 9 cells ml -1 . Approximately 10 6 cells of each strain were added to 10 ml of GM17 and cultures were incubated at 30°C. Aliquots from each culture were taken every hour and plated on solid GM17 medium to calculate generation time for each strain. Experiments were done in triplicate.

Molecular techniques
Molecular cloning techniques like end filling of DNA fragments with the Klenow fragment of DNA polymerase, dephosphorylation, ligation, PCR amplification and agarose gel DNA electrophoresis were carried out essentially as described previously [41]. The miniprep method [42] was used for isolation of large plasmids from lactococci. Plasmids from E. coli were isolated using the QIAprep Spin Miniprep kit according to the manufacturer's recommendations (QIAGEN, Hilden, Germany). The DNA fragments from agarose gels were purified using a QIAqick Gel extraction kit as described by the manufacturer (QIAGEN). Digestion with restriction enzymes was conducted according to the supplier's instructions (Fermentas, Vilnius, Lithuania).
Determination of the effect of ions, pH and proteinase K on aggregation ability of L. lactis subsp. lactis BGKP1 The effect of different ions and pH values on the BGKP1 aggregation phenotype was tested using cells that were three times washed in bi-distilled water until the aggregation phenotype was lost. Cells were then resuspended in buffers of different pH and solutions of various ions. The following buffers were used: KCl (pH 3.0), HCl-glycine (pH 3.0), Na-citrate (pH 4.0 to 6.0), Tris-HCl (pH 7.0 to 10.0) and Tris-NaOH (pH 11.0 to 12.0). The following ions were examined: K + , Na + , Ca ++ , Mg ++ and Fe +++ in concentrations of 0.1, 1, and 10 mM. Proteinase K (1 μg ml -1 ) treatment was done in TE (10 mM Tris, 1mM EDTA, pH8) buffer for 1 h at 37°C. Determination of aggregation phenotype was based on absorption at 600 nm.

Biofilm formation
The ability of BGKP1 and BGKP1-20 to form biofilms was tested as previously described by Christensen and coauthors [43]. Pseudomonas aeruginosa PAO1 and Escherichia coli DH5α were used as the positive and negative control strains respectively. The experiments were done in triplicate.
Analysis of cell surface proteins of L. lactis subsp. lactis BGKP1 and its non-aggregating derivative Cells from overnight culture (250 ml) of strain BGKP1 and its Aggderivative BGKP1-20 were harvested by centrifugation and washed in 50 ml bi-distilled water. Proteins from the wash were precipitated with ammonium sulphate (25% saturation). Precipitated proteins were resuspended in 10 mM Tris-HCl, pH 8.5, and applied on SDS-PAGE (10%). The obtained bands were visualized by Coomassie blue staining.

Construction of shuttle-cloning vectors
The pAZIL shuttle-cloning vector and pAZILcos cosmid vector were constructed in order to perform the molecular analysis of BGKP1 plasmid pKP1 [see Additional File 1]. The tetracycline resistance gene of pACYC184 was replaced with the lacZ gene from the replicative form of M13 mp18 phage using ClaI/NarI and HincII/ AvaII restriction enzymes, resulting in cloning vector pAZ1. In the next step, the chloramphenicol resistance gene from pAZ1 was removed using ScaI and XmnI restriction enzymes and the vector was fused with lactococcal cloning vector pIL253, previously digested with EcoRI-XbaI restriction enzymes and blunted with Klenow enzyme, resulting in shuttle cloning vector pAZIL.
To obtain a cosmid vector for the construction of cosmid libraries of lactococcal genomes, the cos site was introduced into the unique SacII (7697) restriction site of the pAZIL vector. The DNA fragment containing the cos site was obtained by PCR amplification with primers cosF-CATGTTTGACCGCGGATCATCG and cosR-CTAGACACCGCGGAAGCTAGC (SacII restriction sites are underlined). Afterwards, the PCR amplicon was digested with SacII and ligated with SacII-digested pAZIL resulting in the pAZILcos cosmid vector.

DNA sequencing and analysis
For DNA sequencing, pAZIL-KPSl8 and the other constructs aforementioned were isolated from E. coli using a QIAprep Spin Miniprep Kit (QIAGEN) as recommended by the manufacturer. Plasmids were sequenced by primer-walking of both strands in Macrogen's sequencing service (Seoul, Korea). Sequence annotation and the database search for similar sequences were performed using BLAST site programs at the National Center for Biotechnology Information [44]. The DNA Strider program was used for open reading frame (ORF) prediction.

Nucleotide sequence accession number
The nucleotide sequences of the partial 16S rDNA sequence of L. lactis subsp. lactis BGKP1, plasmids pAZILcos and pKP1 were submitted to the EMBL Gen-Bank under accession numbers FR873574, FR872379 and FR872378, respectively.