Dual Regulation of Bacillus subtilis kinB Gene Encoding a Sporulation Trigger by SinR through Transcription Repression and Positive Stringent Transcription Control

It is known that transcription of kinB encoding a trigger for Bacillus subtilis sporulation is under repression by SinR, a master repressor of biofilm formation, and under positive stringent transcription control depending on the adenine species at the transcription initiation nucleotide (nt). Deletion and base substitution analyses of the kinB promoter (PkinB) region using lacZ fusions indicated that either a 5-nt deletion (Δ5, nt -61/-57, +1 is the transcription initiation nt) or the substitution of G at nt -45 with A (G-45A) relieved kinB repression. Thus, we found a pair of SinR-binding consensus sequences (GTTCTYT; Y is T or C) in an inverted orientation (SinR-1) between nt -57/-42, which is most likely a SinR-binding site for kinB repression. This relief from SinR repression likely requires SinI, an antagonist of SinR. Surprisingly, we found that SinR is essential for positive stringent transcription control of PkinB. Electrophoretic mobility shift assay (EMSA) analysis indicated that SinR bound not only to SinR-1 but also to SinR-2 (nt -29/-8) consisting of another pair of SinR consensus sequences in a tandem repeat arrangement; the two sequences partially overlap the ‘-35’ and ‘-10’ regions of PkinB. Introduction of base substitutions (T-27C C-26T) in the upstream consensus sequence of SinR-2 affected positive stringent transcription control of PkinB, suggesting that SinR binding to SinR-2 likely causes this positive control. EMSA also implied that RNA polymerase and SinR are possibly bound together to SinR-2 to form a transcription initiation complex for kinB transcription. Thus, it was suggested in this work that derepression of kinB from SinR repression by SinI induced by Spo0A∼P and occurrence of SinR-dependent positive stringent transcription control of kinB might induce effective sporulation cooperatively, implying an intimate interplay by stringent response, sporulation, and biofilm formation.


INTRODUCTION
In Bacillus subtilis, entry into the sporulation pathway is governed by a member of the response regulator family of transcription factors known as Spo0A (Hoch, 1993). Spo0A is indirectly phosphorylated by a multicomponent phosphorelay system involving at least two kinases called KinA and KinB (Stephenson and Hoch, 2002). An increased level of phosphorylated Spo0A (Spo0A∼P) results in repression of abrB transcription (Strauch et al., 1990), leading to derepression of transcription of the σ H (spo0H) gene encoding σ H . kinA is transcribed by RNA polymerase (RNAP) possessing σ H (Predich et al., 1992), but kinB is transcribed by RNAP possessing σ A (Trach and Hoch, 1993;Dartois et al., 1996). Hence, kinB transcribed by σ A -RNAP is supposed to be a trigger gene for sporulation rather than kinA.
Expression of the kinA and kinB genes is under positive stringent transcription control (Tojo et al., 2013). Their expression is induced upon amino acid starvation through GDP 3 -diphosphate (ppGpp) inhibition of GMP kinase (Kriel et al., 2012) or by the addition of decoyinine, a GMP synthase inhibitor (Mitani et al., 1977;Tojo et al., 2013), resulting in the reciprocal change of a GTP decrease and an ATP increase (Ochi et al., 1981;Tojo et al., 2010). The transcription initiation nucleotide (nt) of stringent promoters P kinA , P kinB and P ilvB (P ilv−leu ) under positive stringent transcription control is the adenine species; ilvB is the first gene of the ilv-leu operon for branched-chain amino acid synthesis (Krásný et al., 2008;Tojo et al., 2008Tojo et al., , 2013. In contrast, the transcription initiation nt of stringent genes such as ptsG and pdhA for glucose catabolism under negative stringent transcription control is the guanine species (Tojo et al., 2010). It is likely that occurrence of both the positive and negative stringent transcription controls causes the B. subtilis cell to enter the sporulation phase (Fujita et al., 2012;Tojo et al., 2013).
The sinR gene was originally isolated as a sporulation inhibition (sin) gene in multiple copies (Gaur et al., 1986). SinR represses transcription from the Spo0A∼P-dependent promoters of sporulation genes such as spoIIA and spoIIG (Cervin et al., 1998). Moreover, transcription of kinB was found to be repressed by SinR on lacZ-fusion analysis (Dartois et al., 1996). Furthermore, the SinR repressor is the master regulator of the formation of a biofilm, a natural lifestyle for most bacteria formed on natural and artificial surfaces (Kearns et al., 2005;Stewart and Franklin, 2008). The wild-type B. subtilis secretes exopolysaccharides (EPSs) and proteins to form an extracellular matrix for building the biofilm (Stewart and Franklin, 2008;Vlamakis et al., 2013). The extracellular matrices are composed of EPSs synthesized from the gene products of the 15-gene epsA-O operon, TasA protein fibers, and the BslA surface layer protein (Vlamakis et al., 2013). SinR is one of the major regulators of the genes required for biofilm formation. SinR binds to the promoter regions of the epsA-O and tapA-sipW-tasA operons to repress their expression (Kearns et al., 2005;Chu et al., 2006). The consensus DNA binding sequence for SinR comprises a 7bp pyrimidine-rich sequence (GTTCTYT, with Y representing an unspecified pyrimidine base), which can be found in an inverted and tandem repeat orientation/arrangement and in a monomer state at SinR operator sites (Kearns et al., 2005;Chu et al., 2006;Colledge et al., 2011). The direct interaction of amino acid residues of SinR with bases of its consensus sequences in an inverted repeat orientation was visualized in the crystal structure of the complex of SinR with operator DNA of the eps promoter (Newman et al., 2013). SlrR is a protein homologous to SinR. SlrR binding to SinR inhibits the DNA-binding activity of SinR, and slrR expression itself is repressed by SinR (Kobayashi, 2008;Chai et al., 2010). Thus, these proteins form a double-negative feedback loop. The SinR antagonist SinI determines which protein is dominant in this loop through protein-protein interaction with SinR (Bai et al., 1993;Chai et al., 2008;Chu et al., 2008). sinI expression is transcriptionally induced by Spo0A∼P (Shafikhani et al., 2002), which is a master regulator of sporulation Lopez et al., 2009). It was recently reported that post-transcriptionally regulated heterogeneous expression of SinR is important for the differentiation of cells present in a biofilm (Ogura, 2016).
In this work, we identified a pair of SinR consensus sequences in an inverted orientation (SinR-1) between nt −57/−42 (+1 is transcription initiation nt) as a SinR-binding site for kinB repression. Unexpectedly, we found that SinR is essential for positive stringent transcription control of P kinB . Electrophoretic mobility shift assay (EMSA) analysis indicated that SinR bound not only to SinR-1 but also to SinR-2 consisting of another pair of SinR consensus sequences in a tandem repeat arrangement (nt −29/−8) that partially overlap the '−35' and '−10' regions, respectively, which is likely involved in positive stringent transcription control of P kinB .

Bacterial Strains and Their Construction
The B. subtilis strains used in this work are listed in Table 1.
Strain FU1204 ( sinR::erm) was constructed as follows. The regions upstream and downstream of the sinR gene were firstly amplified by PCR using DNA of strain 168 as a template, and primer pairs F04a/F04b and F04e/F04f, respectively. The erm cassette was amplified by PCR using DNA of plasmid pMUTIN2 (Yoshida et al., 2000) as a template, and primer pair F04c/F04d. Secondly, recombinant PCR involving primer pair F04a/F04f and three PCR fragments resulted in a PCR product covering the region upstream of sinR, the erm gene, and the region downstream of sinR. The resultant recombinant PCR product was used to transform strain 168 to erythromycin-resistance (0.3 µg/ml) on TBABG plates to produce strain FU1204. Strains FU1206, FU1210, FU1218, FU1219, and FU1224, which carry sinR::erm and each of the lacZ fusions, were obtained by transformation of FU1115, FU1191, FU1216, FU1217, and FU1195 with DNA of strain FU1204 to erythromycin-resistance, respectively.

Cell Cultivation and β-Galactosidase (β-Gal) Assaying
The lacZ-fusion strains were grown at 30 • C overnight on TBABG plates containing the appropriate antibiotic(s); chloramphenicol (5 µg/ml), erythromycin (0.3 µg/ml), spectinomycin (60 µg/ml), and (or) tetracycline (10 µg/ml). The cells were inoculated with an optical density at 600 nm (OD 600 ) of 0.1 in 50 ml of a nutrient sporulation medium (NSMP) (Fujita and Freese, 1981), and then cultivated. Then, 1 ml aliquots of the culture were withdrawn at 1-h intervals, and the β-Gal activity in crude cell extracts was measured spectrophotometrically, as described previously (Yoshida et al., 2000). The cells were also inoculated into 50 ml of a minimal sporulation medium containing 25 mM glucose and 50 µg/ml tryptophan (S6) (Fujita and Freese, 1981). (In the case of the inoculation of the sinR, sinI, and slrR strains into S6 medium, the cells were first cultivated in LB medium before inoculation.) When the cells reached an OD 600 of 0.5, 15 ml each culture was distributed into two flasks, and decoyinine was added to one flask to give a final concentration of 500 µg/ml (18 mM). Before and after decoyinine addition, 1-ml aliquots of the culture were withdrawn at 30-min intervals, and the β-Gal activity was measured.

Sporulation Percentage Measurement
The titers of viable cells (V) and spores (S) that were heatresistant (75 • C for 20 min), for the cultures of strains 168 and FU1204 ( sinR), were measured to obtain the sporulation percentages (S/V x 100) at T0 and T20 (0 and 20 h after entry into the stationary cell phase during sporulation in NSMP). The sporulation percentages for S6 cultures at 0 and 10 h after decoyinine addition (T0 and T10) were also measured.

EMSA Analysis
The PCR primers and template DNA used for preparing biotinylated probes are shown in Supplementary Tables S1, S2-2. Site-directed mutagenesis of the probes was performed using an oligonucleotide-based PCR method as described previously (Ogura and Tanaka, 1996). For EMSA, appropriate amounts of SinR and (or) RNAP were incubated for 15 min at 28 • C with a probe (20 fmol) in 16 µl of a reaction mixture (15 mM Tris-Cl, 4 mM MOPS-KOH, 15 mM KCl, 50 mM NaCl, 0.8 mM MgCl 2 , 0.6 mM DTT, and 12.5% glycerol, pH 7.8) containing 1 µg of poly(dI-dC) (GE Healthcare). After the addition of 2 µl of loading buffer [40% glycerol, 1× TBE (89 mM Tris-borate, and 2 mM EDTA, pH 8), 2 µg/ml bromophenol blue], the samples were applied onto a polyacrylamide gel, and electrophoresis was performed in 0.1× TBE buffer at 4 • C. The method used for the detection of biotin-labeled DNA was described previously (Ogura and Tanaka, 1996).
Most EMSAs were performed with the gradient of the SinR concentration. Not a few critical EMSAs were duplicated.

kinB Transcription and Its Regulation
The kinB gene encoding one of the two major sensor kinases (KinA and KinB) of the phosphorelay system that phosphorylates Spo0A was identified, and its transcription was examined (Trach and Hoch, 1993). The kinB gene is transcribed from the σ A -dependent promoter, which starts from adenine (nt +1) (Trach and Hoch, 1993) (Figure 1). It is co-transcribed with kapB encoding a lipoprotein involved in autophosphorylation of KinB and phosphorylation of Spo0F (Dartois et al., 1997). An ρ-independent transcription terminator was found downstream of kapB, which presumably results in the kinB-kapB transcript. The patB-encoding aminotransferase is located immediately upstream of kinB. Another ρ-independent transcription terminator was found downstream of the patB gene, suggesting that the read-through of patB transcription is blocked. It was communicated in SubtiWiki 2.0 1 (Michna et al., 2016) that the efficient blockage at the transcription terminator actually occurred. kinB transcription was reported to be repressed by SinR (Dartois et al., 1996). It was reported FIGURE 1 | Schematic representation of the patB-kinB-kapB region of B. subtilis. Promoters (P patB and P kinB ) and two stem-loop structures are shown. The kinB gene is cotranscribed with the kapB gene encoding an activator of KinB. This kinB-kapB transcription terminates at the stem-loop structure (ρ-independent terminator) downstream of kapB. patB transcription from P patB terminates at the stem-loop structure downstream of patB, which is indicated by two blue horizontal arrows. SinR-binding site 1 (SinR-1) consisting a pair of SinR consensus sequences (GTTCTYT) in an inverted orientation (C1-1, nt -57/-51 and iC1-2, nt -48/-42, boxed in red), which carries one mismatch and two mismatches to the consensus sequence (the mismatched nt are underlined), respectively. SinR-1 is involved in kinB repression. The 5 deletion (boxed in black, nt -61/-57) and G-45A substitution suppressed the repression. The '-35' and '-10' regions of P kinB are doubly underlined. SinR-binding site 2 (SinR-2) consisting of a pair of SinR consensus sequences in a tandem orientation (C2-1, nt -29/-23 and C2-2, nt -14/-8), both carrying three-mismatches to the consensus sequence. The T-27C C-26T substitution partially affected positive stringent transcription control of P kinB . SinR-2 is likely involved in positive stringent transcription control. The adenine species at transcription initiation nt (+1) is required for positive stringent transcription control to occur (Krásný et al., 2008;Tojo et al., 2010Tojo et al., , 2013. to be presumably repressed by AbrB (Strauch, 1995) and CodY (Molle et al., 2003). Recently, kinB expression was found to be under positive stringent transcription control (Tojo et al., 2013), that is, it is positively regulated upon stringent conditions such as amino acid starvation or on the addition of decoyinine, an inhibitor of GMP synthase, which induces stringent transcription control as well as sporulation. The positive stringent transcription control is strictly dependent on the adenine species at the transcription initiation nt, as described for kinB transcription (Tojo et al., 2013). However, kinB expression was not regulated by CodY or AbrB, at least as observed when examined by use of an lacZ fusion with the P kinB region (nt −55/+10) (Tojo et al., 2013) . To determine if the CodY-or AbrB-binding site is located outside of this region, we attempted to fuse a larger P kinB region with lacZ to yield the largest P kinB -lacZ fusion carrying P kinB (nt −95/+120); the larger fragment including the patB gene upstream of kinB could not be cloned to plasmid pCRE-test2, presumably because patB is harmful in its multiple copy state in E. coli. No significant difference in lacZ expression by the largest lacZ-fusion strain was observed in the wild-type, codY, and abrB genetic backgrounds, on cultivation in NSMP or S6 medium with and without decoyinine (data not shown), suggesting that the CodY-and AbrB-binding sites that affect P kinB are unlikely to be located in the P kinB region (nt −95/+120). This finding implied that kinB expression might not be directly regulated by AbrB and CodY.
In addition, it was notable that the positive stringent transcription control only partially contributed to enhancement of kinB transcription for sporulation in NSMP in contrast to Frontiers in Microbiology | www.frontiersin.org FIGURE 2 | Requirement of the adenine species at the transcription initiation base for positive stringent transcription control of P kinB . The P kinB regions of nt -75/+10 and -55/+10 were fused with lacZ to yield strains FU1191 P kinB (-75/+10) and FU1115 P kinB (-55/+10). The adenine at nt +1 was replaced with a guanine to yield strains FU1193 P kinB (-75/+10 A+1G) and FU1116 P kinB (-55/+10 A+1G). The synthesis of β-Gal encoded by lacZ in strains FU1191 and FU1115, and strains FU1193 and FU1116 was monitored during sporulation in a nutrient sporulation medium, NSMP (A,B), and after addition of decoyinine to the culture in minimal medium, S6 (C,D). Circles and squares indicate the P kinB with adenine and guanine at nt +1, respectively. β-Gal synthesis during sporulation in NSMP was indicated by closed symbols. In the case of S6 medium, closed and open symbols indicate with and without addition of decoyinine, respectively. Large and small symbols denote β-Gal activity and OD 600 , respectively. In all Figures of β-Gal monitoring, the standard deviations of the average β-Gal activity values from the multiple replicates are indicated by error bars (one experiment gives two activity values at an indicated time); tiny error bars are invisible due to their overlap with the symbols. In the case of β-Gal monitoring shown in (A,B), the experiments were performed with triple replicates. a large contribution to it for decoyinine-induced sporulation in S6.

Truncation and Deletion Analysis of the P kinB Region to Identify a Repressor-Binding Site
To localize a repressor-binding site in the P kinB region (nt −95/−55), we constructed a successive series of lacZ-fused P kinB truncation derivatives [P kinB (−95/+10), P kinB (−85/+10), P kinB (−75/+10), P kinB (−65/+10), and P kinB (−55/+10)]. When β-Gal synthesis in these truncation derivatives was monitored during sporulation in NSMP (Figure 3A, left), the P kinB (−55/+10)-lacZ derivative exhibited a higher level of β-Gal synthesis than the other truncation derivatives [P kinB (−95/+10), P kinB (−85/+10), P kinB (−75/+10), and P kinB (−65/+10)], which showed similar levels. When it was monitored upon decoyinine addition to the S6 cultures ( Figure 3A, right), the basal level of β-Gal synthesis by the P kinB (−55/+10)-lacZ derivative before decoyinine addition was higher in comparison with those by the other derivatives. However, the positive stringent transcription control of P kinB was observed to be nearly the same level, approx.1.5-fold increase, for all the truncation derivatives, suggesting that the relief from kinB repression is not involved in this positive stringent transcription control. These overall results suggest that a binding site of a repressor or part of it is likely located in the P kinB region (nt −65/−55), which is responsible for kinB repression but not involved in positive stringent transcription control of P kinB .
is not involved in positive stringent transcription control of P kinB .
Identification of a Putative Binding Site of SinR for kinB Repression, and Involvement of SinR in Positive Stringent Transcription Control of P kinB The sinR gene was isolated as a sporulation inhibition gene in multiple copies (Gaur et al., 1986). At first, we determined the sporulation percentages (%) during cultivation in NSMP medium and during cultivation in S6 after decoyinine addition. The sporulation percentages for strains 168 and FU1204 ( sinR) in NSMP were < 5 × 10 −5 % at T0, and 80 and 100% at T20, respectively. The sporulation percentages for strains 168 and FU1204 in decoyinine-induced sporulation were 0.4% and 1.5% at T0 (at decoyinine addition time) and 40% and 98% at T10. (The sporulation experiments were repeated at least three times. Representative values were presented. The standard deviations were less than 15% of the values shown.) Hence, the sinR deletion tended to promote the sporulation, especially on cultivation in S6 medium with decoyinine.
A previous study involving a lacZ-fusion with the P kinB region (Dartois et al., 1996) suggested that kinB expression is repressed by SinR, and the substitution of guanine at nt −45 in the P kinB region with adenine resulted in relief from SinR repression. Thus, we constructed four strains each carrying P kinB (−75/+10), P kinB (−75/+10 5), P kinB (−75/+10 G-45A), and P kinB (−75/+10 5 G-45A), in the wild-type (sinR + ) and sinR genetic backgrounds. In the sinR + strains cultivated in NSMP medium, the introduction of the inner deletion of 5 or the base substitution (G-45A) greatly and equally relieved the severe repression of lacZ expression observed in strain [P kinB (−75/+10)] without the deletion or substitution ( Figure 4A). Moreover, the introduction of both 5 and G-45A gave further relief from the repression. In the sinR strains cultivated in NSMP, the severe repression of the strain without 5 and G-45A as well as the residual repression observed in the 5 or G-45A strain were well relieved on the introduction of sinR ( Figure 4B).
For decoyinine-induced sporulation of the sinR + background strains in S6 medium, 5 or G-45A equally well relieved the severe repression in strain P kinB (−75/+10) carrying no deletion or base substitution ( Figure 4C). Also, it was completely relieved in strain P kinB (−75/+10) carrying both 5 and G-45A. In sinR strains cultivated in S6 (Figure 4D), the levels of lacZ expression before decoyinine addition were nearly the same in strains P kinB (−75/+10) with and without 5 and (or) G-45A, indicating that the repression observed in strain P kinB (−75/+10) was well relieved on the introduction of sinR. Surprisingly, positive stringent transcription control of P kinB , which is inducible through the addition of decoyinine, did not occur in any sinR strain with and without 5 and (or) G-45A at all ( Figure 4D). However, significant repression of lacZ expression was remained even in the genetic background of sinR, as observed in Figure 4D as well as in Figure 4B. These results indicated that SinR is involved in positive stringent transcription control of P kinB .
A mild plateau of β-Gal synthesis around T0 and a clear decrease in it after T1 were observed in all sinR strains for sporulation in NSMP ( Figure 4B). We could not explain this sporulation phase-dependent variation of β-Gal synthesis without SinR regulation, because SinR repression of kinB transcription and positive stringent transcription control of P kinB did not occur in the sinR strains during sporulation under the cultivation conditions (Figures 4B,D).
The consensus DNA binding sequence for SinR comprises a sequence (GTTCTYT) that can be found in inverted and tandem repeat arrangement/orientation and in a monomer state at SinR operator sites (Kearns et al., 2005;Chu et al., 2006;Colledge et al., 2011). Examination of the P kinB sequence around the 5 deletion and the G-45A substitution allowed us to identify a pair of SinR consensus sequences [C1-1 (nt −57/−51) and iC1-2 (−48/−42)] in an inverted repeat orientation (SinR-1 site), each consensus unit containing part of the 5 deletion or the G-45A substitution [Figures 1, 4 (Top)]. Therefore, SinR-1 is most likely a SinR-binding site for kinB repression.
As described above, SinR is essential for positive stringent transcription control of P kinB. Examination of the sequences around the '−35' and '−10' regions revealed another pair of SinR consensus sequences [C2-1 (nt −29/−23) and C2-2 (−14/−8)] in a tandem repeat arrangement (SinR-2 site) [Figures 1, 4 (Top)]. The C2-1 and C2-2 sequences partially overlap the '−10' and '−35' regions of P kinB , which might be possibly a SinR-binding site for positive stringent transcription control of P kinB . We attempted to isolate mutants of strain FU1115 [P kinB (−55/+10)-lacZ] that are defective in positive stringent transcription control of P kinB . We arbitrarily introduced nine base substitutions to the SinR-2 sequence (nt −29/−8) , and examined if the strength of each mutant P kinB is comparable to that of the wild-type, and if β-Gal synthesis under the control of the mutant P kinB is positively regulated after decoyinine addition. Thus, we found that only one mutant, FU1249 P kinB (−55/+10 C-26T T-27C) carrying the substitution in the C2-1 consensus sequence of SinR-2, synthesized β-Gal almost at the same level as wild-type strain FU1115, and exhibited partially impaired positive stringent transcription control in comparison with strain FU1115 (Figure 5). Although the other eight substitutions affected the P kinB strength, they did not affect positive stringent transcription control significantly (Supplementary Figure S1). The T-15C, G-14A, and (G-16A T-15C) mutations abolished the P kinB activity. The A-17G, (T-20C T-19), and (T-18C A-17G) mutations decreased it by severalfold, but did not affect positive stringent transcription control. The G-16A and (C-26T T-25C) mutations considerably enhanced the P kinB activity, but they did not affect positive stringent transcription control significantly. These results suggested that the C2-1 consensus sequence of SinR-2, where the C-26T T-27C substitution only affecting positive stringent transcription control of P kinB is located, is likely involved in its positive stringent transcription control.
The overall results indicated that SinI deficiency causes stronger SinR-dependent repression and reduces derepression, but SlrR is not involved in the repression, and also indicated that SinR is involved in positive stringent transcription control of P kinB , but SinI and SlrR are not.

EMSA Analysis of SinR Binding with Probes Carrying Deletion and Base-Substitution That Affect kinB Regulation in Vivo
On lacZ fusion analysis using sinR as well as 5, G-45A, and C-26T T-27C, SinR was found to be responsible not only for kinB repression involving SinR-1 consisting of C1-1 and iC1-2 (Figure 4), but also for positive stringent transcription control of P kinB probably involving SinR-2 consisting of C2-1 and C2-2 (Figures 4, 5). On EMSA analysis using the probes carrying 5 and G-45A, and C-26T T-27C, we found that these mutations actually affected in vitro SinR binding to SinR-1 and to SinR-2, respectively, as follows (Figure 7).
As shown in Figure 7A-1, the wild-type P kinB (−75/+10) probe gave the two closely located bands on EMSA, which likely resulted from SinR binding to SinR-1 and SinR-2. The upper band is invisible at 12.5 nM SinR, and visible at 25 nM with the P kinB (−75/+10) probe carrying the G-45A substitution, likely resulting from approximately 2-fold less binding affinity to SinR-1 (Figure 7A-3). This band disappeared with the probe carrying the 5 deletion or 5 and G-45A (Figures 7A-2,-4), suggesting that SinR cannot bind to SinR-1 if part of the C1-1 sequence is deleted by 5.
As described above, Figure 4C shows that in the wild-type cells with P kinB (−75/+10) during cultivation in S6 medium, as well as in those with P kinB (−75/+10) possessing 5 (and G-45A), approximately 1.5-fold positive stringent transcription control of P kinB steadily occurred, regardless of the level of kinB repression before decoyinine addition. The same level of positive stringent transcription control was also observed in the cells of a series of truncation and deletion derivatives of the P kinB region (nt −95/+10 to −55/+10) that exhibited different levels of kinB repression (Figure 3). These results suggested that SinR simultaneously binds to both SinR-1 and SinR-2 to form a larger complex than that on SinR binding to SinR-1 or SinR-2. Nevertheless, a more slowly migrating band other than the two closely located bands did not exist ( Figure 7A). Thus, the closely located upper and lower bands were considered to probably result from simultaneous SinR binding to SinR-1 and -2, and from SinR binding to SinR-1 or SinR-2, respectively.
It should be noted that EMSA analyses involving the probes of P kinB (−75/+10) and P kinB (−75/−7) carrying 5 G-45A gave an apparent equilibrium dissociation constant (K d ) of approximately 10 nM for SinR binding to SinR-2 ( Figure 7A and Supplementary Figure S2-1), but EMSA involving P kinB (−55/+10), P kinB (−31/+104), and P kinB (−39/+104) (Fm0) probes gave K d of more than 100 nM for SinR binding to SinR-2 ( Figure 7B-1 and Supplementary Figure S3-2). This finding implies that an unidentified sequence upstream of nt −55 might function to enhance SinR binding to SinR-2 without its binding to SinR-1. This unknown enhancement of SinR binding to SinR-2 remains to be studied. Figure 9 summarizes the results of EMSA analyses involving a series of three-base substituted PCR probes to determine which parts of SinR-1 and SinR-2 sequences are necessary for SinR binding. For EMSA to determine which part of the SinR-1 is necessary for SinR binding, the mutant probes (Rm6, Rm5, Rm4, Rm3, Rm2, and Rm1) and the wild-type one (Rm0) were used, as illustrated on the left side of Figure 9. The EMSA results as to Rm0 and Rm1 to Rm6 are shown in the upper panel of Supplementary Figure S3-1. The relative densities of the shifted bands of the mutant probes [++, +, +/−, and ND (not detected)] to that of the wild-type (++) as to their binding to SinR-1 are arbitrary given in the lower left column from their vision (Supplementary Figure S3-1). The base-substitutions within C1-1 and iC1-2 of SinR-1 (Rm4, Rm3, Rm2, and Rm1) almost completely abolished the shifted band, whereas those upstream of C1-1 (Rm6 and Rm5) did not diminish the band density. The results indicated that both C1-1 and iC1-2 are essential for SinR-binding to SinR-1.
[2] EMSA results using the P kinB (-55/+10) probe carrying the T-27C C-26T substitution in the C2-1 consensus sequence. DNA probes were prepared by use of primer pair and template DNA listed in Supplementary Tables S1, S2-2.
shifted bands of the mutant probes to that of the wild-type as to their binding to SinR-1 [++, +, +/−, and ND] are given in the lower right column. The base-substitutions in C2-1 (Fm2) only partially affected SinR binding to SinR-2, but those in C2-2 (Fm5) considerably affected it. Besides, the substitution (Fm4) immediately upstream of C2-2 partially affected it.
EMSA analysis with the mutant probes (Figure 9) indicated that SinR-binding to SinR-2 only partially requires C2-1 but it well requires C2-2. The EMSA results with deleted probes as described above (Figure 8 and Supplementary Figure S2) suggested that both C2-1and C2-2 are likely essential for SinR binding to SinR-2. This inconsistency might reflect the difference between the three-base substitution in C-2-1 (Figure 9) and its complete elimination (Figure 8). However, the role of AGT just upstream of C2-2 in SinR binding to SinR-2 is unknown. These EMSA results suggested that both C2-1 and C2-2 are likely necessary for SinR binding to SinR-2, although C2-1 might not be so strictly required in comparison with C2-2. Thus, the SinR binding site (SinR-2) likely comprises the two SinR consensus of C2-1 and C2-2 sequences in a tandem arrangement, which partially overlap the '−35' and '−10' regions of P kinB , respectively.
The overall EMSA analyses clearly indicated that SinR binds to SinR-1 consisting of C1-1 and iC1-2 for transcription repression of P kinB and it binds to SinR-2 consisting of C2-1 and C2-2 for its positive stringent transcription control.
EMSA Analysis of the Binding of SinR and RNA Polymerase (RNAP) to the P kinB Region We found that SinR-2 consists of C2-1 and C2-2 in tandem arrangement, which is likely involved in positive stringent transcription control of P kinB . C2-1 and C2-2 partially overlap the '−35' and '−10' regions of P kinB (Figure 1), so it was expected that SinR might bind to SinR-2 to form a transcription initiation complex of SinR, RNAP, and P kinB to exert its positive stringent transcription control. As shown in Figure 10, the electrophoretic band of a complex of the P kinB probe and RNAP appeared to shift to a slightly slower position when SinR was further added. This implies that a positively regulated stringent promoter such as P kinB might form a transcription initiation complex with SinR and RNAP for its positive stringent transcription control.

DISCUSSION
ppGpp is synthesized by the RelA protein associated with ribosomes upon amino acid starvation (Fujita et al., 2012). In the case of E. coli, the target of ppGpp is RNAP, stringent genes being regulated positively and negatively, depending on their specific promoter sequences. In contrast, the ppGpp target is GMP kinase in B. subtilis, the in vivo GTP concentration being reduced (Kriel et al., 2012). The GTP concentration also decreased upon addition of decoyinine, an inhibitor of GMP synthase (Tojo FIGURE 8 | Deletion analysis to localize SinR-binding sites (SinR-1 and SinR-2) by EMSA. (Top) The nt sequence of the P kinB region (nt -75/+10), where SinR-1 (C1-1 and iC1-2) and SinR-2 (C2-1 and C2-2) are localized, is shown. The EMSA probes cover various P kinB regions indicated by thick blue bars where SinR consensus sequences (red boxes) are shown; X means a defective consensus sequence. et al., 2008). The GTP decrease reciprocally results in an ATP increase via feedback regulation (Tojo et al., 2010(Tojo et al., , 2013. The transcription initiation of negative stringent control genes such as rrn, ptsG, and pdhA, whose transcription initiation base is guanine, is reduced upon a GTP decrease, whereas that of positive stringent control genes such as ilvB, pycA, kinB, and kinA, whose transcription initiation base is adenine, is enhanced upon an ATP increase (Krásný and Gourse, 2004;Krásný et al., 2008;Tojo et al., 2008Tojo et al., , 2010Tojo et al., , 2013. Decoyinine induces sporulation of B. subtilis cells exponentially growing in the presence of rapidly metabolizable carbon, nitrogen, and phosphate sources (Mitani et al., 1977). It is known that the stringent response also induces sporulation (Ochi et al., 1981(Ochi et al., , 1982. Recently, decoyinine was found to induce positive stringent transcription control of the kinB gene encoding a trigger of sporulation (Tojo et al., 2013), which might be the reason why decoyinine induces sporulation. The lacZ-fusion analysis using the mutant cells carrying the A+1G substitution (Figure 2) disclosed that positive stringent transcription control has a larger contribution to kinB expression in decoyinine-induced sporulation in minimal S6 medium than in sporulation in nutrient NSMP medium. Both kinB repression by SinR and SinR-dependent positive stringent transcription control of P kinB simultaneously occur, as inferred from that approx. 1.5-fold positive stringent transcription control of P kinB was constantly and steadily observed after decoyinine addition to the S6 culture, regardless of the level of kinB repression before decoynine addition (Figures 3, 4).
The sinR strain exhibits the sporulation-deficient phenotype when present in multiple copies (Gaur et al., 1986). The sinR strain sporulated a little bit better than the wild-type strain. The kinB gene was repressed by SinR (Dartois et al., 1996) (Figure 4). SinR was also involved in positive stringent transcription control of P kinB (Figures 4, 5). SinI is an antagonist of SinR (Bai et al., 1993;Chai et al., 2008;Chu et al., 2008), which is induced by Spo0A∼P (Shafikhani et al., 2002;Lopez et al., 2009). SinI induced during sporulation initiation eventually inhibits SinR, leading to relief of kinB repression through SinR detachment from its binding site. Thus, SinI deficiency resulted in stronger SinR-dependent repression and reduced derepression (Figure 6). SrlR, a protein homologous to SinR (Kobayashi, 2008), was unlikely involved in the relief from this SinR repression. Furthermore, neither SinI nor SrlR was involved in its positive stringent transcription control (Figure 6).
Thus, in the wild-type SinR + cells, the limited level of derepression of kinB from SinR repression by SinI induced by Spo0A∼P and significant induction of SinR-dependent positive stringent transcription control of P kinB upon stringent response cooperatively induce effective sporulation. It is inferred from the results (Figures 4, 6) that the level of kinB expression on sporulation of the wild-type strain is likely lower than that on sporulation of the sinR strain even if positive stringent transcription control is blocked by sinR. This might be the reason why the sinR strain sporulated a little bit better than the wild-type strain.
Examination of the sequence of the P kinB region revealed two SinR binding sites (SinR-1 and SinR-2), i. e. a pair of SinR consensus sequences (C1-1 and iC1-2) in an inverted orientation, and another pair of SinR ones (C2-1 and C2-2) in a tandem arrangement, respectively (Figure 1). Such SinR-binding motifs consisting of a pair of SinR consensus sequences in an inverted orientation and a tandem arrangement are often observed in the promoter regions of the operons involved in biofilm formation such as espA-O (Kearns et al., 2005) and tapA-sipW-tasA (Chu et al., 2006). In vivo deletion and base substitution analyses of SinR-1 for kinB repression (Figures 3, 4) and EMSA using various deleted and mutated probes (Figures 7-9) revealed that both C1-1 and iC1-2 are necessary for kinB repression and SinR binding to SinR-1. Moreover, the base substitution in C2-1, which was involved FIGURE 10 | EMSA analysis of the binding of SinR and RNA polymerase (RNAP) to the P kinB region. The P kinB (-75/+10 5 and G-45A) probe was used for EMSA. The slightly lower and upper arrowheads denote the positions of the complexes of SinR and the probe, and of RNAP, SinR and the probe, respectively.
The sinR deletion ( sinR) abolished positive stringent transcription control of P kinB (Figure 4). lacZ-fusion analysis of the other stringently-controlled promoters (unpublished observation by S. Nii and Y. Fujita) indicated that sinR also abolished the positive stringent transcription control of P ilvB , P pycA, and P kinA. Interestingly, sinR did not affect the negative stringent transcription control of P ptsG and P pdhA . This observation suggested that positive stringent transcription control involves SinR, but negative stringent transcription control does not involve it. EMSA analyses (Figures 7-9) showed that the P kinB region actually possesses an SinR-binding site (SinR-2), i. e. a pair of C2-1 and C2-2 sequences partially overlapping the '−35' and '−10' regions, respectively, which is likely involved in positive stringent transcription control of P kinB (Figure 5). Furthermore, EMSA indicated that a complex of RNAP, SinR, and P kinB for transcription initiation is likely formed, implying that SinR might be involved in transcription initiation of positively controlled stringent genes (Figure 10). Detailed investigation of the molecular mechanism involving SinR underlying positive stringent transcription control is in progress.

AUTHOR CONTRIBUTIONS
YF, SN, and KH performed in vivo study of kinB regulation by SinR in B. subtilis from April 2015 to March 2017 in