Genetic evidence for a novel competence inhibitor in the industrially important Bacillus licheniformis

Natural genetic competence renders bacteria able to take up and, in case there is sufficient homology to the recipient’s chromosome, integrate exogenously supplied DNA. Well studied in Bacillus subtilis, genetic competence is—in several aspects—known to be differently regulated in Bacillus licheniformis. We now report on the identification of a novel, chromosomally encoded homolog of a competence inhibitor in B. licheniformis (ComI) that has hitherto only been described as a plasmid borne trait in the ancestral B. subtilis NCIB3610. Bioinformatical analysis that included 80 Bacillus strains covering 20 different species revealed a ComI encoding gene in all of the examined B. licheniformis representatives, and was identified in few among the other species investigated. The predicted ComI of B. licheniformis is a highly conserved peptide consisting of 28 amino acids. Since deletion of comI in B. licheniformis DSM13 resulted in twofold increased transformation efficiency by genetic competence and overexpression resulted in threefold decreased transformability, the function as a competence inhibitor became evident.


Introduction
Various bacterial species can develop natural genetic competence, a physiological state that enables cells to take up DNA (Dubnau 1999;Johnsborg et al. 2007). The regulatory system governing genetic competence has been studied rather thoroughly in the gram-positive model organism Bacillus subtilis (Dubnau 1999;Hamoen et al. 2003;Spizizen 1958). The development of natural genetic competence in B. subtilis depends on environmental stimuli such as nutritional limitation and/or cell density (Hamoen et al. 2003). The key transcriptional regulator for developing natural genetic competence in B. subtilis is ComK (van Sinderen et al. 1995). Governing cell division, DNA-binding, -uptake, -recombination and -repair, ComK positively controls expression of more than 100 genes; nine genes are negatively affected (Berka et al. 2002;Hamoen 2011).
In contrast to B. subtilis, Bacillus licheniformis DSM13 carries an insertion element within comP rendering ComP, the sensor histidine kinase required for ComX-sensing, inactive (Lapidus et al. 2002). Removing the insertion element (and thereby restoring an active copy of comP) resulted in reduced genetic competence (Hoffmann et al. 2010), which clearly differs from B. subtilis. Further regulatory differences concern ComS action , as the two ComS homologs identified in B. licheniformis did not impact-contrary to B. subtilis-the development of genetic competence.
A competence inhibitor (ComI) was identified in the ancestral B. subtilis strain NCIB3610 . It is encoded on the endogenous 84-kb plasmid pBS32. ComI renders the strain hardly transformable when compared to the frequently used laboratory strain B. subtilis 168 (Nijland et al. 2010). Possibly due to curing, pBS32 is absent in the laboratory strains which descend from B. subtilis NCIB3610, such as B. subtilis 168, B. subtilis PY79 or B. subtilis JH642 McLoon et al. 2011). When pBS32 was cured from the ancestral B. subtilis NCIB3610, transformation efficiencies via genetic competence indeed increased approximately 100-fold; though a similar drastic effect was observed for deletion of comI, the knockout of other plasmid-borne genes positively influenced competence as well .
In this study, we provide evidence for a ComI homolog within the species B. licheniformis. The predicted protein appears to be conserved among B. licheniformis species, only rather seldom ComI homologs could be predicted for other Bacillus species. Deletion of comI has a beneficial effect on the competence mediated transformability of B. licheniformis DSM13, whereas overexpression resulted in a decrease of the transformation efficiency.

Bioinformatical and statistical analysis
Analysis of the primary protein structure of ComI was performed with TMBASE (Hofmann and Stoffel 1993). Sequence analysis was performed with BioEdit 7.0.7.0. Evolutionary history was inferred using the Neighbor-Joining algorithm (Saitou and Nei 1987). The evolutionary distances were computed using the Poisson correction method (Zuckerkandl and Pauling 1965) and are in the units of the number of amino acid substitutions per site. The analysis was conducted using MEGA7 (Kumar et al. 2016). Statistical analysis was performed with GraphPad Prism 7.

Molecular biological techniques
Cloning in E. coli was performed essentially as described in Sambrook and Russel (2001). Genomic DNA from B. licheniformis was isolated as previously described (Nahrstedt et al. 2004)

Vector construction
Primers used in this study were obtained from Eurofins Genomics GmbH (Ebersberg, Germany) and are listed in Additional file 1: Table S1. For disruption of comI in B. licheniformis MW3.1 the flanking regions of comI were amplified; flank A was obtained using the primer pair comI_delA/comI_delA_KpnI and for flank B the primer pair comI_delB/comI_delB_BamHI was applied. For insertion of aphA, the gene was amplified from vector pMB03 using the primer pair KanR_A/KanR_B. For the disruption of comI, the flanks and aphA were fused by SOE-PCR (splicing by overlap extention) (Heckman and Pease 2007), restricted with BamHI and KpnI and cloned into the likewise restricted pUppem vector resulting in plasmid pUE∆comI. For the P comI -GFP fusion the promotor region of comI was amplified using the primer pair comI13f_KpnI/ comI13r_ClaI. The PCR product was subsequently restricted with ClaI and KpnI and ligated into the likewise restricted vector pMUTIN-GFP+, resulting in plasmid pMUTIN-comI.

Transformation
Plasmids were transformed into E. coli using the CaCl 2 mediated method described by Sambrook and Russel (2001) or into B. subtilis SCK6 via a transformation protocol developed by Zhang and Zhang (2011). Sequenced vectors were introduced into B. licheniformis via induced genetic competence (Hoffmann et al. 2010).

GeneBank accession numbers
All primary nucleotide sequences used in this work can be found in the GeneBank sequence database of NCBI. The respective accession numbers are listed in Additional file 1: Table S2.

Bioinformatical identification of ComI within the genus Bacillus
BLAST ® Standalone searches disclosed-contrary to the known plasmid-borne ComI of B. subtilis NCIB3610 (ComI 3610 )-a putative chromosomally encoded homolog in B. licheniformis DSM13 (ComI DSM13 ) (Fig. 1a, first line). We were eager to know, whether such chromosomally located gene is present in other Bacillus strains and species as well. When bioinformatical analyses were performed, including altogether 80 Bacillus strains from 20 different genera (data not shown), a putative comI gene was identified for all 14 B. licheniformis strains included in the survey, whereas it was rather rarely seen in the other Bacillus strains tested (i.e. 4 representatives; see Fig. 1). The predicted ComI of B. licheniformis is a highly conserved protein consisting of 28 aa (VTVSEALQLM-VSFGILVVAILSSNDKKK). Bootstrap analysis revealed three groups of ComI homologs, with ComI DSM13 forming the largest and most conserved group (Fig. 1a, c). Furthermore, a single transmembrane alpha helix could be predicted for ComI DSM13 (Fig. 1b). Exemplarily we studied the function of ComI DSM13 .

Deletion of comI resulted in a twofold increase of transformability
As ComI 3610 was already proven to inhibit genetic competence in B. subtilis , it was tempting to check whether such action is provided by ComI DSM13 as well. We therefore used the suicide plasmid pUE∆comI to replace comI with the kanamycin resistance cassette aphA in the uracil-auxotrophic strain B. licheniformis MW3.1, yielding strain B. licheniformis CM1 (Fig. 2a). The relevant genetic organization of the strain was examined by PCR analysis (Fig. 2b). The possible effect of the comI::aphA substitution on natural genetic competence was tested by comparing strain CM1 with its parental strain MW3.1 in transformation experiments. The transformation frequency in B. licheniformis MW3.1 was arbitrarily set as 100 (Fig. 2c). The deletion of comI had a beneficial effect on the transformability, as CM1 displayed a doubled transformation frequency of 201% ± 4.6, an effect that is nevertheless 50-fold lower than the effect observed in B. subtilis NCIB3610, in which the deletion of comI resulted in an approximately 100-fold increase of the strain's transformability .

Recombinant overexpression of comI resulted in threefold reduced transformation efficiency
Parallel to the comI knockout and the results achieved with strain B. licheniformis CM1, we investigated the effect of comI overexpression. For such purpose the integrative vector pMUTIN-comI was constructed, in which comI is placed under the control of the IPTGinducible promoter P spac . Subsequently the construct was Fig. 1 Alignment of ComI homologs predicted for different Bacillus species. a ComI homologs were identified in 20 strains. Dark blue boxes refer to basic aa, red boxes to acidic aa. Hydrophobic aa are displayed as green-shaded boxes whilst polar, uncharged aa are bluish violet-shaded. Glycine is given in yellow, cysteine in maroon and histidine in pink. b Primary structure of ComI DSM13 . Amino acids are given in single-letter code. The N and C termini are indicated as well as the position inside and outside of the cell. The grey-shaded protein area indicates the predicted transmembrane helix. c Similarity of 23 ComI protein sequences from 20 different Bacillus strains. The reliability of the tree was calculated using the bootstrap test (500 replicates) and is shown next to the branches (Felsenstein 1985). The tree shows three ComI subgroups established in B. licheniformis MW3.1; yielding strain B. licheniformis CM2 (Fig. 3a). The correct integration of the expression vector was verified by PCR analysis and gel electrophoresis (Fig. 3b). Possibly due to the fact, that natural genetic competence only renders at maximum 20% of the cells genetically competent (Turgay et al. 1997), experiments with natural genetic competence and comI overexpression resulted in transformation frequencies too low for allowing reliable evaluation (data not shown). We therefore performed experiments in which genetic competence was induced by overexpression of comK (Hoffmann et al. 2010). Expression of comI was achieved by addition of IPTG to the final concentration of 100 µM. Transformation efficiencies for B. licheniformis MW3.1 were arbitrarily set as 100%. B. licheniformis CM2 yielded only approximately 1/3 (33.06% ± 14.53) of the transformation efficiency compared to B. licheniformis MW3.1 (Fig. 3c).

Discussion
Bacillus licheniformis is a close relative to B. subtilis. Natural genetic competence, which has been examined thoroughly for B. subtilis (Dubnau 1999;Hamoen et al. 2003;Spizizen 1958), has also been reported for B. licheniformis strains (Hoffmann et al. 2010;Jakobs et al. 2014;Leonard et al. 1964;McCuen and Thorne 1971;Thorne and Stull 1966), even though with lower efficiencies than for B. subtilis Waschkau et al. 2008). Despite the close relationship, major differences in the regulation of genetic competence were seen. While ComP is essential for the development of genetic competence in B. subtilis (Weinrauch et al. 1990), B. licheniformis DSM13 carries an insertion element in comP, which renders ComP inactive (Hoffmann et al. 2010). In contrast to B. subtilis, the removal of the insertion element led to lower transformation efficiencies (Hoffmann et al. 2010). Furthermore, it became evident that the two comS homologs found in B. licheniformis (ComS1 and ComS2) did not impact genetic competence .
The existence of a functional chromosomal comI gene is another remarkable difference between the two species. ComI appears as a highly conserved, 28 aa spanning peptide within B. licheniformis species, while it is hardly found in B. subtilis. Indeed, only the plasmid borne, 30 aa peptide-encoding comI gene of B. subtilis NCIB3610 has been reported as a functional competence inhibitor . While a comI locus has been predicted for B. subtilis spizizenii DSM15029 and B. subtilis natto BEST195, it remains to be elucidated whether these loci encode for a functional competence inhibitor.
In B. subtilis ComI is reported as membrane protein containing a single transmembrane domain, that renders the strain hardly transformable . We identified a similar single transmembrane domain in ComI DSM13 . Interestingly, while the N terminus of ComI 3610 is predicted to be intracellular , an extracellular N terminus is suggested for ComI DSM13 . Furthermore, glutamine 12 of ComI 3610 has been described as essential for the protein's competence Fig. 3 Overexpression of comI and its effect on induced genetic competence. a Schematic illustration of the genomic region of the pMUTIN-comI integrant B. licheniformis CM2. Open reading frames are shown as arrows, the direction of which corresponds to the transcriptional orientation. Screening primers are denoted as black triangles, promoters are depicted as angled arrows and the t1t2t0-terminator is shown as a hairpin-structure. comI, encoding the putative competence inhibitor ComI DSM13 ; ftsW, encoding a cell-division protein; gfp, green fluorescent protein gene; lacI, encodes the repressor protein LacI; ori ColE1, origin of replication; bla, encodes ampicillin resistance; ermC, erythromycin resistance gene. b Verification of the pMUTIN-comI insertion via PCR with the screening primers comI13f_KpnI and GFPseqr1 and gel electrophoresis. c Transformation efficiencies for B. licheniformis MW3.1 and B. licheniformis CM2 obtained by induced genetic competence using chromosomal DNA of B. licheniformis ∆spoIV (Hoffmann et al. 2010) to obtain uracil prototrophy (n = 3). Induction of P spac was achieved by addition of IPTG to the cultivation medium to a final concentration of 100 µM. Transformation efficiencies for the wild type (MW3.1) were arbitrarily set as 100%. Data are given as mean ± SD of 3 independent experiments. ***p < 0.001 inhibiting function, as a G12L substitution rendered the protein inactive for competence inhibition . ComI DSM13 possesses a serine residue at position 12. Both glutamine and serine are polar, uncharged amino acids. Konkol and colleagues postulated that competence inhibition might be caused by ComI 3610directly or indirectly-either separating the energy-providing protein from a transmembrane protein involved in DNA uptake or by preventing the separation of the latter two . However, as for ComI 3610 , the mode of competence inhibition needs to be clarified for ComI DSM13 as well.
Our results indicate that ComI DSM13 has an inhibitory effect on genetic competence in B. licheniformis, but does not inhibit competence completely (Hoffmann et al. 2010;Jakobs et al. 2014). The development of genetic competence is a highly sophisticated process, in which the key transcriptional regulator, ComK, controls the expression of competence genes (van Sinderen et al. 1995;Hamoen 2011). The regulation of competence is strictly controlled and, as has been shown before, mainly brought about by deregulation (Hoffmann et al. 2010; and ComI DSM13 appears to be a further peptide that controls the development of genetic competence. As the deletion of comI in B. licheniformis DSM13 doubled the transformation efficiency rather than increasing the efficiency 100-fold, as for ComI 3610 , the intracellular level of ComI DSM13 might be more strictly controlled. However, it must be taken into account that the increased transformation efficiencies described for B. subtilis NCIB3610 resulted from curing of the 84 kb endogenous pBS32 plasmid. Even though the deletion of comI 3610 itself increased transformation efficiencies, Konkol and colleagues (2013) demonstrated that the deletion of other genes and gene clusters also had a beneficial effect on the strain's transformability. pBS32 encodes RapP, a phosphatase that, besides repressing Spo0F activity, also inhibits genetic competence through direct or indirect repression of ComA (Parashar et al. 2013;Omer Bendori et al. 2015). Roughly one-half of the pBS32 located genes encode for phagelike proteins, but the phage-like particles have been shown to be defective and did not kill B. subtilis (Myagmarjav et al. 2016). ComI 3610 , together with RapP and possibly further, hitherto undetected proteins encoded by pBS32 might promote the intracellular persistence of the plasmid as they, through diminution of genetic competence, prevent the uptake of other, possibly competing plasmids into the cell. The task of plasmid persistence might therefore require a much more drastic way of competence inhibition for ComI 3610 than is required for the competence-regulating, but not competence-thwarting ComI DSM13 .
While an interaction with a ComK-induced gene product may prevent ComI DSM13 from performing its competence-inhibiting function, regulation of comI expression, directly or indirectly through ComK, is conceivable as well. Even though the deletion of comI is not crucial for competence development, the deletion greatly improved the transformability and is, thus, a useful tool for enhanced genetic manageability.