C. difficile has been described as a bacterium with highly mosaic genetic composition and multiple attempts have been made to distinguish between various strains and to correlate them with virulence 18 . We first analysed the variability of the repressor LexA encoding gene sequence among various C. difficile ribotypes (groups characterized by differences in intergenic regions of RNA operon and used worldwide for C. difficile typing) and toxinotypes (characterized by differences in toxin A and B coding region inside the pathogenicity locus called PaLoc) (Additional file 1: Table S1) 19 . Analysis revealed 17 single nucleotide polymorphisms (SNPs) in the lexA gene of 63C. difficile sequences among which four SNPs resulted in missense mutations but none of the mutations modified amino acids in the cleavage or active sites of LexA (Figure 1). Our analysis grouped the investigated strains into three clusters according to the C. difficile LexA (Figure 2). Cluster I encompassed 3 non-toxinogenic strains and strains of toxinotype 0; Cluster II encompassed strains of toxinotypes III, VIII, IX, and X and finally, Cluster III with the highest number of SNPs, was mostly composed of toxinotype V strains. Ribotypes for the above stated toxinotypes can be found in the Additional file 1: Table S1. Previous results showed that strains belonging to the epidemic ribotype 027 form a genome wide clade 20 21 , typically characterised as the toxinotype III (North American pulsed field gel electrophoresis type 1 - NAP1, REA group BI). Interestingly, ribotypes 016, 019, 036, 075, 111, 122, 153, 156, 176, 208 and 273 are closely related to ribotype 027 by comparative genomics 20 21 , and those ribotypes were found to encompass the lexA cluster II. Comparative phylogenomics along with MLST (multilocus sequence typing) and whole genome sequecing has shown that ribotype 078 lineage is different than other C. difficile lineages 22 . Moreover PCR ribotype 078 forms a phylogenetically coherent group with ribotypes 033, 045, 066, 078, 126 and 127 23 – which encompasses lexA cluster III. Genetically distinct strains that belong to ribotypes 078 (V) and 126 (V) clustered together showing the highest number of SNPs in the lexA gene. The phylogenetic tree based on LexA variability reflects similarities to genetic lineages based on ribotype patterns and comparative genomics analysis.

To obtain insight into the LexA regulon genes, we performed in silico genome-wide prediction of LexA binding sites within promoter regions of C. difficile. Using the xFiToM software 24 , we screened genomes of thirty C. difficile strains (Additional file 1: Table S1) for the C. acetobutylicum and C. perfringens consensus operator sequence of LexA 15 16 , allowing for two mismatches in one of the two half sites positioned within 350 bp upstream to 35 bp downstream of a protein coding sequence. Among the thirty genomes, the search yielded at least one putative operator sequence upstream of more than 30 genes involved in a variety of biological processes e.g. DNA repair, transport, virulence and antibiotic resistance (Table 1).

Sequences of putative LexA operators and their positions (according to the start of the gene coding region). Numbers denote strains with the operator identified. NO marks the gene that was identified in the strain but a target LexA site was not found in its promoter region, NG marks that gene was not found in the genome of the strain.

Subsequently, we purified C. difficile LexA and RecA proteins with an N-terminal hexa-histidine tag (Additional file 2: Figure S1) as described for E. coli orthologs 25 . SPR analysis was performed to validate the in silico data and determine the LexA-operator interactions in vitro in real time. Most of the interaction sites were found in putative promoter regions of “common” putative SOS genes for the majority of the genomes tested and of putative LexA regulon genes encoding unusual SOS proteins. Out of 20 DNA fragments tested, the repressor interacted with 16 targets (Figure 3A, Additional file 3: Table S2). We determined interaction with operators in promoter regions of the core SOS response genes: recA, lexA, the genes of the uvrBA operon encoding for components of the UvrABC endonuclease catalyzing nucleotide excision repair and the ruvCA operon genes, encoding the nuclease that resolves Holliday junction intermediates in genetic recombination. In addition, LexA interacts with putative promoter regions of genes involved in sporulation (sspB), regulation of DnaA-dependent initiation of DNA replication (soj), several ABC transporters (potC, oppC, MicroScope:CDR20291_2297) and for homologue of a two-component system regulator of the vancomycin resistance cluster (vanR). The LexA repressor was also found to interact within PaLoc with operator identified 525 base pairs upstream of the toxin A gene (tcdA). While the regulation of toxin production in C. difficile is controlled in response to several environmental signals mediated by pleiotropic regulators (CcpA, CodY, SigD and SigH 26 ), the possible regulation through the SOS system sheds new light on this issue. Furthermore, the subinhibitory concentration of SOS-inducing antibiotic ciprofloxacin was recently shown to increase the Toxin A gene expression in C. difficile 27 . Our SPR analysis revealed that also housekeeping genes required for ribosome function (rplR) and β subunit RNA polymerase (rpoB) belong to the LexA regulon, a feature of the SOS network not yet observed in bacteria. Thus, blockage of LexA self-cleavage could impede pivotal functions in C. difficile and this might provide a new approach to treat C. difficile infections. Moreover, although putative SOS genes are present in most of the analysed genomes, several of these genes encoding for putative cell wall hydrolase, transposase and for two component sensor histidine kinase seem to be regulated by LexA only in the 027 ribotype strains (Table 1). The in silico analysis showed operators in front of several genes upregulated exclusively in ribotype 075 and 027 (celG, vanR, ABC-type transport system). Furthermore, among the analysed genomes, exclusively in the closely related ribotypes 078, 126 and 033, the LexA target site was not found in front of the soj (regulation of DNA replication) and the phnH (phosphonate metabolism protein). Thus the mode of SOS regulation might be related to phylogenetic lineages.

Induction of SOS gene expression is synchronized and the level, timing and duration of expression of the individual LexA regulon genes differs significantly (1). In E. coli, LexA bound to target DNA cannot interact with RecA* and only unbound repressor is proteolytically inactivated 25 . Thus the rate of LexA dissociation from operators controls the precise timing of SOS gene expression following induction. Consequently genes with lower affinity LexA target sites are expressed prior to genes with high affinity operators 1 5 . To follow up on these results, we used SPR to study interaction of the chip-immobilized C. difficile RecA* with LexA interacting with either specific or non-specific DNA. We showed that as in E. coli, the C. difficile LexA repressor interaction with RecA* is prevented by binding to specific DNA targets (Figure 4). In addition, we showed that the key SOS players of E. coli and C. difficile can cross-react in vitro (Figure 4). Hence, our data indicated that the mode of regulation of the C. difficile SOS response resembles the one described for E. coli. Nevertheless, in contrast to the E. coli SOS system, we observed among the investigated C. difficile genes, a slowest LexA dissociation from operators of the core SOS genes, recA, lexA and uvrB (Figure 3A and B, Table 2), implying that these are the last genes upregulated upon SOS induction. For instance, LexA dissociation from the E. coli recA operator is more than 20-times faster than from C. difficile with regard to the dissociation constants of 4.8 ± 2.1 × 10⁻³ s⁻¹ (21) and 1.7 ± 0.5 × 10⁻⁴ s⁻¹, respectively.

We propose that in C. difficile, strong repression of core SOS genes affects the magnitude of the system`s induction. Thus, the low association and non-stable LexA binding to putative regulatory regions of genes encoding the RNA polymerase β subunit (rpoB), 50S ribosomal protein (rplR), spermidine/putrescine permease (potC), vancomycin response regulator (vanR) and putative multidrug-efflux-pump [MicroScope: CDR20291_2297], indicates that LexA contributes to fine-tuning of expression of these genes independently of substantial recA induction (Figure 3).

The paradigm of the SOS system is that DNA repair genes are rapidly induced in the SOS response to deal with DNA lesions 1 2 28 . However, comparison of induction of LexA regulon genes in B. subtilis and E. coli in response to double-strand breaks reveals diversity 29 . After DNA damage, the velocity of assembly of RecA* is similar but in contrast to E. coli, a limited set of LexA-regulated genes are induced early in the response in B. subtilis. Our in vitro results suggest that also in C. difficile, induction of the LexA-regulated DNA repair genes might be induced later in the SOS response as the core SOS gene promoter regions harbour high affinity LexA targets. According to the differences in LexA-operator affinities we predict that upon DNA damage, various biological processes will be derepressed without induction of the SOS DNA repair.

The C. difficile genomes were obtained from an opened access NCBI database 30 and an undisclosed access to MicroScope platform 31 . The strains used for amplification with PCR and sequencing belong to the strain collection of the Institute of Public Health Maribor. The list of strains used for analysis of the LexA variability and regulon is presented in the Additional file 1: Table S1.

Variability of lexA in C. difficile was compared by analysis of alignment and phylogenetic trees of nucleotides and amino acid sequences performed with Vector NTI (Invitrogen) and with the interactive viewer for phylogenetic trees: Dendroscope 32 . Sixty three sequences were analysed in total (NCBI – 9 strains, MicroScope – 44 strains, PCR product of in-house strains – 10). Strains CD196, R20291 and 630 were obtained from both databases. List of strains used for lexA gene variability can be found in Additional file 1: Table S1.

The search for LexA binding sites was performed for 30 genomes (Additional file 1: Table S1). The number of strains covering ribotypes was as follows: ribotype 027 – eight strains; ribotypes: 078, 001, 005 and 012 - three strains from each; ribotypes 075 and 126 two strains from each and one genome from each ribotypes 017, 087, 014, 053. The analysis was performed with xFiToM software 24 . The searched motifs, based on C. acetobutylicum and C. perfringens consensus, were as follows: GAACnnnnGTTT, GAACnnnnGTTC, GAACnnnnnTTT, GAACnnnnnTTC. The default options were used with the limitation to 350 base pairs upstream to 35 bp downstream of a protein coding sequence. An exception was the promoter region of the putative endonuclease/exonuclease/phosphatase (MicroScope: CDR20291_2056) where we found 2 operators positioned approximately 460 upstream of the coding sequence and hence, we included the targets in the analysis. The results were subjected to manual check by extraction of gene sequences along with 1000 base pairs upstream and downstream followed by alignment and re-search of the binding sites.

The C. difficile 630 chromosomal DNA was extracted by the Genomic DNA purification kit (Thermo Scientific) according to the manufacturer`s instructions. The lexA and recA genes were amplified by PCR from the chromosomal DNA using specific primers (DinR_U 5′-GCGCGGATCCAGTGATGTTATGTATTTAGATC-3′ - DinR_D 5′-CGCACGCGTCTATTTAATAACTCTAAATAC-3′) and (RecA_U 5′-GCGCGGATCCAGTGTAGATCAAGAAAAATTAAAAG-3′ - RecA_D 5′-CGCACGCGTTTATTCTTCTACAATTTCTTTTG-3′), respectively. The PCR products were then purified and cut with BamHI and MluI and cloned into pET8c vector digested by the same enzyme to create plasmids pDinRCD and pRecACD for expression of proteins fusion with N-terminal His₆ tag. Large-scale expression of proteins was performed in the E. coli BL21 (DE3) strain and purified from the bacterial cytoplasm by Ni-NTA affinity chromatography as described for the E. coli key SOS proteins 25 . PD10 desalting columns (GE Healthcare) were used for exchange of the buffer. The proteins were stored at −80°C in 20 mM NaH₂PO4 (pH 7.4), 0.2 mM NaCl. Protein concentrations were determined using NanoDrop1000 (Thermo Scientific) and extinction coefficients at 280 nm of 7450 M⁻¹ cm⁻¹ for recombinant LexA and 16055 M⁻¹ cm⁻¹ for recombinant RecA.

C. difficile LexA-operator measurements were performed on a Biacore T100 (GE Healthcare) at 25°C as described 6 . The 3′-biotynilated 5-CGCTCGAGTAGTAAC-TEG-Bio-3′primer was immobilized on the flow cell 2 (Fc2) of the streptavidin sensor chip (GE Healthcare) in SPR buffer containing 20 mM Tris–HCl (pH 7.4), 140 mM NaCl, 0.005% surfactant P20 (GE Healthcare). To prepare double stranded DNA (dsDNA) fragments with the predicted C. difficile LexA operators, complementary pairs of primers presented in Additional file 3: Table S2 were dissolved in 20 mM NaH₂PO₄ (pH 7.4), 0.14 M NaCl and mixed in 1:1.5 (mol : mol) ratio for the longer to shorter primer, respectively. Primers were annealed in temperature gradient from 95°C to 4°C (~ 1.5 h) in PCR machine (Eppendorf). So prepared DNA fragments were approximately 22 bp duplex DNAs with 15-nucleotide overhangs complementary to the chip-immobilized primer. Approximately 44 response units of either DNA fragment were hybridised at 2 μl min⁻¹ to the Fc2. The interaction of C. difficile LexA with the chip-immobilized DNAs was analysed by injecting repressor in SPR buffer in 20 nM concentration across the chip surface at 100 μl min⁻¹ for a minute and dissociation was followed for 9 minutes. The regeneration of the surface was achieved injecting 12 s pulse of 50 mM NaOH at 100 μl min⁻¹. The experiments were performed in triplicates and the representative sensorgrams are shown. Data were fitted to a 1:1 binding model to obtain the dissociation rates constants. Program MEME was used to determine LexA binding motifs 33 .

SPR C. difficile RecA*-LexA interaction measurements were performed on a Biacore X (GE Healthcare) at 25°C as described to study the interaction among the key E. coli SOS proteins 25 . Experiments were performed in SPR_2 buffer (20 mM NaH₂PO₄ (pH 7.4), 150 mM NaCl, 2 mM MgCl₂, 1 mM DTT, 1 mM ATP (Sigma Aldrich), 0.005% surfactant P20 (GE Healthcare). C. diffcile LexA repressor (2.6 μM), interacting with either the 22 bp recA operator DNA fragment or with the 22 bp non-specific DNA fragment derived from the recA operator, was passed over the sensor chip with immobilized RecA* (~2000 response units). LexA specific DNA (recA operator) or non-specific DNA, with 6 nucleotide changed in comparison to the specific DNA, was prepared by hybridising primers (1:1 mol to mol ratio) 5′-CAAGAGAACAAATGTTTGTAGA-3′ and 5′-TCTACAAACATTTGTTCTCTTG-3′or 5′-CAAGACCGGAAATCCTTGTAGA-3′ and 5′-TCTACAAGGATTTCCGGTCTTG-3′, respectively. The RecA*-LexA interaction was assayed at 10 μl/min for 60 s and the dissociation followed for 60 s. The sensor chip was regenerated as described 25 .

Activation of either E. coli or C. difficile RecA (10 μM) nucleoprotein filament was performed on ice for 2 h as described 34 . RecA*-stimulated (~2 μM) cleavage of LexA were performed in 20 mM Tris, pH 7.4, 5 mM MgCl₂, 1 mM ATP-γ-S (Sigma), and 1 mM DTT as described 25 . Samples were resolved on 12% SDS PAGE gels in MOPS running buffer (Invitrogen) and stained by Page blue protein stain (Thermo Scientific). The resolved bands were quantified using a G:Box (Syngene). The integrated optical densities of the LexA monomers were determined. The LexA levels throughout the time course were compared and are presented as the ratio of the density value for the sample at time indicated as 0 min relative to the density value obtained from the samples obtained later in the LexA cleavage reaction. The experiments were performed two times and representative gels are shown.