Nucleotide sequence and analysis of pRC12 and pRC18, two theta-replicating plasmids harbored by Lactobacillus curvatus CRL 705

The nucleotide sequences of plasmids pRC12 (12,342 bp; GC 43.99%) and pRC18 (18,664 bp; GC 34.33%), harbored by the bacteriocin-producer Lactobacillus curvatus CRL 705, were determined and analyzed. Plasmids pRC12 and pRC18 share a region with high DNA identity (> 83% identity between RepA, a Type II toxin-antitoxin system and a tyrosine integrase genes) and are stably maintained in their natural host L. curvatus CRL 705. Both plasmids are low copy number and belong to the theta-type replicating group. While pRC12 is a pUCL287-like plasmid that possesses iterons and the repA and repB genes for replication, pRC18 harbors a 168 amino acid replication protein affiliated to RepB, which was named RepB’. Plasmid pRC18 also possesses a pUCL287-like repA gene but it was disrupted by an 11 kb insertion element that contains RepB’, several transposases/IS elements, and the lactocin Lac705 operon. An Escherichia coli / Lactobacillus shuttle vector, named plasmid p3B1, carrying the pRC18 replicon (i.e. repB’ and replication origin), a chloramphenicol resistance gene and a pBluescript backbone, was constructed and used to define the host range of RepB’. Chloramphenicol-resistant transformants were obtained after electroporation of Lactobacillus plantarum CRL 691, Lactobacillus sakei 23K and a plasmid-cured derivative of L. curvatus CRL 705, but not of L. curvatus DSM 20019 or Lactococcus lactis NZ9000. Depending on the host, transformation efficiency ranged from 102 to 107 per μg of DNA; in the new hosts, the plasmid was relatively stable as 29–53% of recombinants kept it after cell growth for 100 generations in the absence of selective pressure. Plasmid p3B1 could therefore be used for cloning and functional studies in several Lactobacillus species.


Competent cells and electroporation
E. coli DH10B competent cells were prepared and transformed after heat shock at 42˚C for 2 min [34]. Preparation of Lactococcus lactis competent cells and transformation were done as previously described [35]. Lactobacillus competent cells were prepared following the procedures described by Berthier et al [14]. Briefly, cells were grown in MRS broth for about 6 h until an optical density between 0.4-0.6. Then, the cells were washed twice with MgCl 2 10 mM and then with a storage solution of sucrose 0.5 M containing 10% glycerol, in which they were suspended and stored at -80˚C. Electrotransformation of L. curvatus CRL 705 cells was optimized using the plasmid pRV566 and varying systematically the electroporation parameters (voltage: 1.8 kV, 2.0 kV and 2.5 kV; resistance: 200 Ω, 600 Ω, 800 Ω and 1000 Ω. The optimized parameters were then used to electroporate L. curvatus and L. plantarum strains, L. sakei 23K cells were electroporated using 1.8 kV, 600 Ω and 25 μF as previously described [14].

Determination of relative plasmid copy number
The relative plasmid copy number was determined using quantitative real time PCR (qPCR), following the procedures described by Lee and coworkers [36]. As a reference, a 151 bp fragment of the rpoD gene was used, which has been identified as a single copy in the chromosome. Two other pairs of primers were used to amplify same-sized fragments of pRC18 and pRC12 plasmids (Table 2). Primers for qPCR were designed using IDT Primer Quest (https:// www.idtdna.com/Primerquest/Home/Index) and synthetized by Sigma Aldrich (Buenos Aires, Argentina). The copy number was calculated according to the formula: N relative = (1+E) -ΔCT [36], where E is the PCR amplification efficiency of the target and reference genes, and ΔCT stands for the difference in the threshold cycle number between the reference and the target genes. The experiment was carried out using three replicates for both, biological and technical repetitions. The results were expressed in average values and the standard errors were calculated from the data with all the repetitions. The plasmid copy number was then classified as low, medium or high when the copies of plasmids per chromosome were 1-10 copies, 11-20 copies or more than 20 copies, respectively.

Host range and stability tests
The host range of the novel replicase (RepB') found in pRC18 was defined using different LAB (Table 1) as transformations hosts with p3B1 vector. The presence of p3B1 was confirmed by plasmid DNA extraction and PCR, using primers 5`-CGGGAAACCGGTGAAATCTA-3`and 5`-CCTACTCCGCAATCGCTAAA-3`, under the following conditions: the first step for denaturalization at 94˚C during 5 min, then 35 cycles for extension including 1 min at 94˚C, 2 min at 58˚C and 1 min at 72˚C followed by a final extension of 5 min at 72˚C. The segregational stability of p3B1 was assayed following the protocol described by Alpert et al [9] with slight modifications: strains were reinoculated in fresh MRS medium every 24 h, using appropriate dilutions, until approximately 100 generations. Cells were then plated on agar media with or without antibiotic and incubated at 30˚C. Colonies forming units (CFU) per mL were determined by biological and technical triplicates, the average values and the standard error were calculated.

Nucleotide sequence accession number
For comparative analyses, the following L. curvatus plasmids were used: pLCUFL03 (NZ_LT841332.

Results
The aim of this study is the analysis and characterization of plasmids pRC12 and pRC18 detected in the draft genome sequence of L. curvatus CRL 705 [23]. While the nucleotide sequence of pRC12 was determined as a single contig (23-fold coverage) by using a wholegenome shotgun (WGS) strategy, the genome sequence of pRC18 was determined by a primer-walking strategy and confirmed by WGS [23]. CDSs of more than 70 nucleotides in length were identified and annotated. The presence of tRNA was not detected. Both plasmids belong to the theta replication group and the relative plasmid copy number of pRC18 and pRC12, as determined by quantitative real time PCR (qPCR), was 6.41 ± 1.20 and 3.60 ± 0.57 copies per cell, respectively. Thus, pRC12 and pRC18 are low-copy number plasmids.

Sequence analysis of pRC12
The DNA sequence of plasmid pRC12 consists of 12,342 bp with 43.99% GC content. The average length of CDSs is about 638 bp and the intergenic average length is 249 bp. Fourteen putative proteins encoding CDSs (designated ORF1 to ORF14) were identified ( Fig 1A,  Table 3). ORF 1 encodes a site-specific tyrosine recombinase (integrase) that belongs to the conserved protein domain family INT_C_like_3 and it is also present in different strains of the genus Lactobacillus (100% identity). Downstream from ORF1 and in the opposite orientation, ORF 2 and ORF 3 encode an antitoxin of the Phd/YefM family and a RelE-like toxin, respectively, a type II toxin-antitoxin stability system. The protein encoded by ORF 4 was predicted to be a hydrolase/deacetylase, and contains a NodB catalytic domain of the carbohydrate esterase 4 superfamily. Two CDSs involved in the replication process were identified: ORF 5 and ORF 6 correspond to RepB and RepA, respectively, two proteins involved in the replication of theta-type plasmids found in other LAB. In particular, the genetic organization of the replication system (ori-repA) of pRC12 plasmid is consistent with that of the well characterized plasmid pUCL287 from Tetragenococcus halophilus ATCC33315 [41,42] (Fig 1B-1D). Indeed, ORF 6 (repA) encodes a 311 amino acids (aa) protein 100% identical to RepA from L. sakei subsp. carnosus DSM15831 (KRL70440.1) while ORF 5 (repB) encodes a 192 aa protein, with 99% identity with the replication protein RepB found in L. curvatus KG6 (ASN62919.1). repA and repB overlap by 8 bp as in pUCL287 (Fig 1D), the untranslated region upstream from repA consists of four 11 bp direct repeats, with two of them being separated from the other two by 2 bp. These four repeats are then separated by 37 bp from a 22 bp motif (iterons) repeated 4.5 times (Fig 1C and S1 Fig). These iterons may constitute the replication origin of pRC12. Between these repeats and the repA start codon, the promoter and ribosome binding sites ( Fig  1C) are located, as reported in [41,42]. ORF 7 is a remnant gene that consists of a putative transposase. ORF 8 encodes a putative metal-transporting ATPase, while ORF 9 and ORF 10 encode conserved proteins of unknown function. pRC12 also harbors two CDSs related to DNA mobilization: ORF 11, which encodes a protein highly similar to the relaxase/mobilization domain found in the plasmid content of L. sakei MFPB19 (SON74348.1), and the protein encoded by ORF 12, which is homologous to MobC of Pediococcus damnosus (KRN47316.1). Finally, a second type II toxin-antitoxin system is harbored by pRC12, divergent from that encoded by ORF2 and ORF3. ORF 13 and ORF 14 encode a putative antitoxin of the lambda repressor-like, Cro/C1-type HTH domain, DNA-binding domain superfamily (COGnitor: COG1396 group) and a HigB-like toxin (COGnitor: COG4679 family), respectively.

Sequence analysis of pRC18
The nucleotide sequence of plasmid pRC18 consists of 18,664 bp, with a GC content of 34.33%. Twenty-four CDSs were identified in this plasmid, with an average CDS length of 557 bp and an average intergenic length of 213 bp (Fig 2A, Table 4); the protein coding density is 65.32%.
ORF 1 encodes a 202 aa protein 95% identical to Xre family transcriptional regulator of Lactobacillus paracollinoides TMW1.1994. As reported above for pRC12, an int-antitoxin-toxin operon is also found in plasmid pRC18. ORF 2 and ORF 3 encode a toxin of the RelE-like and an antitoxin of the Phd/YefM family, respectively. ORF 4 encodes a 195 aa integrase of the INT_C_like_3 DNA breaking-rejoining enzyme superfamily. pRC18 plasmid has a pUCL287-like replication initiator protein RepA which is disrupted in two parts (ORF 6 and ORF 21, the 5' and 3' fragments of repA, respectively) by an 11 kb insertion (Fig 2B). ORF 6 shows 92% identity with the N-terminus of the RepA protein of plasmid pPECL-7 from Pediococcus damnosus LMG 28219 and ORF 21 shows 99% identity with the C-terminus of the RepA protein found in L. paracollinoides DSM 15502. The 11 kb fragment encompasses a transposase-like structure. ORF 8, ORF 9, ORF 10 (frameshifted) and ORF 20 are related to transposases of the IS150 family, whereas ORF 19 and ORF 22 are transposases of the IS1077 and IS30 families, respectively. Moreover, the 11 kb insertion contains the lactocin Lac705 operon (ORFs 11 to 15), a class IIb bacteriocin that relies on the complementary action of two peptides, Lac705α and Lac705β [22,43]. ORF 18 encodes a putative replicase that we named RepB' and is described below. The 11 kb fragment is highly similar to a contig found in the recently sequenced L. sakei CECT 9267 (NZ_OKRC01000015) (S2 Fig). Finally, ORFs 16-17 and ORFs 23-24 did not show significant homology to any protein available in GenBank and Phyre 2 databases. Comparative nucleotide sequence analysis of pRC18 against pRC12 and other plasmids harbored by L. curvatus strains The replication regions, ori and rep genes, of plasmids pRC12 and pRC18 are not homologous. However, as shown in Fig 3, the relE-like toxinphd-yefM type II antitoxin stability systems and the integrase genes of these plasmids shared 100% and 93% identity, respectively. Also, high similarity between the two fragments of the disrupted repA in pRC18 and the complete repA in pRC12 was found. The toxin-antitoxin system and the integrase were also present (identity score above 80%) in other plasmids of L. curvatus (see plasmids pKG6, pIRG2, and pZJUNIT8_3; Fig 4) as well as the pUCL287-like RepA protein was found in eight out of the 20 compared plasmids (Fig 4). It is noteworthy that truncated repA was only found in pRC18.

Replication functions of plasmid pRC18. Host range and stability
The replication region of pRC18 is shown in Fig 5. ORF 18 encodes a putative replicase of 168 aa (annotated as RepB' to differentiate it from the RepB protein of pRC12) that contains a DUF536 superfamily domain (pfam04394) related to theta-replicating plasmids [11,44]. Moreover, its N-terminus contains a potential HTH domain that might contribute to the interaction with DNA for replication. RepB' also has 92% identity with a DUF536-containing protein of plasmid pIRG2 from L. curvatus and 82% of identity with RepB protein of plasmid pLs145-c from L. sakei LK-145. Upstream of RepB' (ORF 18), an A-T rich (60%) non-translated region was found with two direct repeats (DR) of 10 bp and two inverted repeats (IR) with potential formation of a stem-loop structure, one of 16 bp at positions 11,518 and 11,541, and the other of 8 bp at positions 11,497 and 11,563 (Fig 5). This region may be the replication origin ori of pRC18. Putative promoter regions -35 and -10 are located at positions 11,729 and 11,758, respectively; while a ribosome binding site is predicted at position 11,778, 5 bp upstream from the start codon.
To test the functionality of ori-RepB', a 3.4 kb HpaI-EcoRI fragment, between coordinates 10,334-13,806 of pRC18, was cloned in a derivative-pBluescript SKII (+) plasmid, which does not have any ori for Gram positive bacteria and thus cannot replicate in LAB. This vector contains a chloramphenicol resistance cassette from plasmid pC194 for selection. The recombinant plasmid, named p3B1 (Fig 6), was initially successfully used to transform L. curvatus CRL 1890, a plasmid-cured derivative of L. curvatus CRL 705. Transformation efficiency could be improved by optimizing transformation parameters. Indeed the higher efficiencies were obtained using 2.0 kV, 1000 Ω and 2.5 kV, 600 Ω, keeping 25 μF constant. The presence of  (Table 5), the highest frequency being obtained with L. sakei 23K with transformation efficiencies higher than those obtained with other plasmids in this strain [9,11,14]. This gives new alternatives for the development of genetic tools for this well characterized strain.
Furthermore, we have tested the stability of p3B1 in the absence of selective pressure in L. curvatus CRL 1890, L. plantarum CRL 691 and L. sakei 23K transformants. After approximately 100 generations of cells growing in the absence of antibiotic, the percentages of chloramphenicol resistant colonies were 29±6%, 53±12% and 36±8%, respectively.

Discussion
Theta-type replicating plasmids are classified into six classes according to their dependency on three different components: a plasmid-encoded replication protein (Rep), an ori and a DNA polymerase encoded by the host [17]. Plasmid pRC12 belongs to the well-known pUCL287like family described by Benachour et al. [41], and can be clearly assigned to class A. This replication family is characterized by the presence of 3-5 iterons of 20-22 bp in an AT region located upstream the replicase. The presence of mobA and mobC functions in pRC12 suggests it may be a mobilizable plasmid; this is consistent with its size (lower than 15 kb), but not with its low copy number [7]. The co-occurrence of both pRC18 and pRC12 in L. curvatus CRL 705 indicates that the replication systems of both plasmids are compatible. Plasmid-curing experiments showed that pRC12 and pRC18 were very stable at 30˚C (plasmid loss was lower than 1%). A functional replicase RepB' and the presence of an ori on pRC18 plasmid are presented here. The replication RepB' protein of pRC18 (168 aa) has not been previously described; it contains a C-terminal DUF536 domain also found in several bacterial proteins that may be involved in a theta-type replication mechanism. Upstream from the repB' gene, an AT-rich region containing direct and inverted repeats strongly suggest this region could be considered to be the pRC18 replication origin [45]. This structure could act as recognition sites in the initiation of the replication by RepB' or in the regulation of RepB' synthesis [46]. The structural organization of the ori may differ among plasmids families, e.g. pAMβ1 ori consists of a short AT-rich region but does not have a typical structure [47]. In order to classify pRC18 into one of the six classes, further studies addressing the dependency of the ori region and the DNA polymerase of the host are necessary. The genome of Lactobacillus has a complex architecture [5], which also applies to its plasmids [9,48,49]. While plasmid pRC12 exhibits an average GC content (43.99%) similar to that of the L. curvatus CRL 705 chromosome (41.95%), the average GC content of pRC18 (34.33%) is lower than that of its host and suggests horizontal genetic transfer. Indeed, pRC18 shows a mosaic structure. The truncated nonfunctional pUCL287-like RepA in pRC18 is explained due to the presence of an 11 kb insertion rich in transposases and repB' functions. pRC12 and pRC18 shared more than 83% nucleotide sequence identity in some regions (toxin/antitoxin, integrase and repA genes) located outside the 11 kb insertion, suggesting that this fragment was gained by pRC12 after an integration event into its repA gene leading to a second plasmid, pRC18. Although the plasmidome of L. curvatus is heterogeneous in terms of gene and GC contents [17], some regions like the toxin-antitoxin, integrase and RepA are maintained in other L. curvatus plasmids such as pGk6, pIRG2 and pZJUNIT8_3.
Plasmids are dispensable extrachromosomal elements. The persistence of plasmids in their bacterial hosts may result from the presence of plasmid genes encoding benefits for the cells to grow under certain environmental conditions [6,50] or by carrying specialized plasmid stabilization mechanisms such as multimer resolution systems, partition determinants, and post-segregational killing systems (PSK). PSK include toxin-antitoxin systems, restriction modification loci and bacteriocin loci; these systems ensure plasmid maintenance by killing plasmid-free cells [51]. The toxin-antitoxin (TA) systems are a mechanism of plasmid persistence; the antitoxin is more labile and susceptible to degradation than the toxin; thus, in the presence of a plasmid-free cell the toxin induces cell death [52]. Bacterial TA systems are mobile, two-genes modules widely distributed in plasmids and chromosomes, often in multiple copies. TA systems have already been described in lactobacilli [53]; from 1 to 9 TA pair systems have been identified in Lactobacillus genomes (https://bioinfo-mml.sjtu.edu.cn/TADB2/browse_org. php?alpha=L). The genetic organization of 6 type II TA systems detected in several Lactobacillus rhamnosus and Lactobacillus fermentum strains, isolated from various human microbiota, was found to be polymorphic; it was suggested the polymorphisms of the TA systems could be used to the analysis of strains diversity among the Lactobacillus genus [54]. The TA systems detected in the plasmids of strain L. curvatus CRL 705 are type II; both toxins and antitoxins are small proteins and their genes form an operon. Here, RelE-like toxins, which belong to the ParE/RelE superfamily, are combined with antitoxins of different families in the formation of the TA systems. The YefM-RelE-like system is present in both plasmids, pRC12 and pRC18 (100% identity). The YefM antitoxin is from the Phd superfamily. Plasmid pRC12 carries a second TA, the Cro/C1-type HTH domain/HigB-like system [55]. The HigB toxin belongs to the COG4679 family of proteins often found in genomes of phages [55]. The antitoxins to HigB contain a typical HTH domain of the Cro/C1 family and belong to COG1396. Similar to the canonical association is RelB-RelE in E. coli, the RelE-like toxins may function as a ribosomedependent mRNA interferase, degrading mRNA positioned at the ribosomal A-site and causing growth arrest and cell death. The antitoxins would bind tightly to RelE inhibiting its ribonuclease activity. The antitoxin could also function as an autorepressor; they can bind alone or in a complex with RelE-like to the operators in the TA promoter blocking its transcription. In E. coli, TA systems regulate key cellular processes during period of stress; when conditions of stress dissolve, the toxic effect of RelB-RelE is reverse by a trans-translation mechanism, allowing growth cells to resume [56]. As indicated above, both plasmids pRC12 and pRC18 were very stable at 30˚C (plasmid loss lower than 1%). The stability tests of plasmid p3B1, which is free of the toxin-antitoxin systems, revealed a plasmid loss frequency of >47% plasmid-free cells after 100 generations at 30˚C. However, the type II RelE/Phd-YefM system appeared to contribute modestly to plasmid stability of p3B1, since this TA system only slightly improved stability of the recombinant plasmid pRA1TA (plasmid loss frequency of about 65%). These data suggest that the toxin-antitoxin systems present in pRC18 may not contribute by itself to the stability and maintenance of both low copy number plasmids [57]. The presence of the lactocin Lac705 operon, a two-component class IIb bacteriocin [43], in pRC18 could also contribute to the stability of this plasmid by a PSK mechanism.
Results presented here contribute to a better knowledge of Lactobacillus curvatus CRL 705, an important strain used as protective cell culture in industrial applications. The small size of RepB' (168 aa) from pRC18 represents an advantage for constructing vectors that are easier to handle and likely at high transformation efficiencies. In accordance with its mode of replication, pRC18 derivative vector has a narrow host range, as previously described for theta plasmids [8]. Nevertheless, pRC18 has a replicon that could transform L. curvatus CRL 1890, L. plantarum CRL 691 and L. sakei 23K and be potentially useful to construct cloning vectors for Lactobacillus. The vector p3B1 also has suitable unique restriction sites to be used as cloning sites (BmtI, NheI, BlpI, BglII,SacI, BstXI, SacII, AleI, NotI, EagI, XbaI and BamHI). Such new cloning vectors for Lactobacillus strains will be helpful to study the functionality of different genes and gain knowledge about many functional or probiotic properties of lactobacilli important for food fermentation or biopreservation.
Supporting information S1 Fig. Nucleotidic comparison of the replication region of plasmids pRC12 from L. curvatus CRL 705 and pUCL287 from Tetragenococcus halophilus ATCC33315. Alignments performed by Clustal Omega shows that plasmid pRC12 has a homologous replicon (ori-repA) with plasmid pUCL287. The pRC12 plasmid sequence is presented above and the pUCL287 sequence below, the homology is represented by asterisks ( � ), the numeration in the left is arbitrary in terms of localization in the plasmid. (PDF)