The PAS domains of the major sporulation kinase in Bacillus subtilis play a role in tetramer formation that is essential for the autokinase activity

Abstract Sporulation in Bacillus subtilis is induced upon starvation. In a widely accepted model, an N‐terminal “sensor” domain of the major sporulation kinase KinA recognizes a hypothetical starvation signal(s) and autophosphorylates a histidine residue to activate the master regulator Spo0A via a multicomponent phosphorelay. However, to date no confirmed signal has been found. Here, we demonstrated that PAS‐A, the most N‐terminal of the three PAS domains (PAS‐ABC), is dispensable for the activity, contrary to a previous report. Our data indicated that the autokinase activity is dependent on the formation of a functional tetramer, which is mediated by, at least, PAS‐B and PAS‐C. Additionally, we ruled out the previously proposed notion that NAD +/NADH ratio controls KinA activity through the PAS‐A domain by demonstrating that the cofactors show no effects on the kinase activity in vitro. In support of these data, we found that the cofactors exist in approximately 1000‐fold excess of KinA in the cell and the cofactors’ ratio does not change significantly during growth and sporulation, suggesting that changes in the cofactor ratio might not play a role in controlling KinA activity. These data may refute the widely‐held belief that the activity of KinA is regulated in response to an unknown starvation signal(s).

bacterial two-component systems (Stock, Robinson, & Goudreau, 2000). However, the detailed mechanisms of the kinase activation, including how the putative "sensor" domain recognizes the starvation signal(s) and how the autokinase activity is then triggered, remain elusive because the molecular identity of the starvation signal(s) is as yet unclear.
To date, multiple pieces of evidence have suggested the importance of Spo0A activation in regulating the initiation of sporulation upon starvation. First, the cellular level of Spo0A~P increases gradually over the course of starvation in the cell . Second, when the cellular concentration of Spo0A~P reaches a certain level (threshold), the position of the cell division site switches from medial to polar, resulting in the arrest of vegetative growth (Fujita, Gonzalez-Pastor, & Losick, 2005). Third, after asymmetric cell division, the level of Spo0A~P continues to increase only in the mother cell (larger compartment), while Spo0A disappears in the forespore (smaller compartment) (Fujita & Losick, 2003). Fourth, this spatiotemporal increase in Spo0A activity is essential for proper cell-fate determination (Kovacs, 2016;Vishnoi et al., 2013).
While the starvation signal(s) is unknown, recent studies have attempted to address the mechanisms of KinA activation. Some key findings are that, first, there is no significant difference in the level of kinA transcripts between growing and sporulating cells . Second, both the protein level and activity of KinA increase slightly and gradually over the course of nutrient starvation (Eswaramoorthy, Dinh, Duan, Igoshin, & Fujita, 2010;. Third, a certain threshold level of KinA protein is necessary and sufficient to trigger phosphorelay signaling and sporulation Narula et al., 2016). And fourth, the N-terminal domain of KinA plays a role in the formation of a stable tetramer complex, but may not contribute a signal sensing function to activate the enzyme's C-terminal catalytic domain Eswaramoorthy, Guo, & Fujita, 2009;Eswaramoorthy et al., 2011).
Thus, based on these observations, the threshold level of KinA appears to be a primary regulator to produce Spo0A~P beyond a critical level to direct the expression of the Spo0A-controlled sporulation genes.
Prior in vivo evidence suggests that the most N-terminal PAS domain of KinA, PAS-A, is dispensable for the enzyme's autokinase activity as well as for sporulation Eswaramoorthy et al., 2009). These results are in agreement with an in vitro study reported by Winnen, Anderson, Cole, King, & Rowland (2013). In contrast, several of the previous studies have indicated that PAS-A is essential for the autokinase activity (Kolodkin-Gal et al., 2013;Lee et al., 2008;Wang et al., 2001). Among them, Lee et al., (2008) reported that the purified PAS-A domain forms a dimer by itself. However, studies by the Hoch group indicate that PAS-A is monomeric and possesses ATP-binding and nucleoside diphosphate kinase-like activities . These two groups also demonstrated that the PAS-A domain is essential for the autokinase activity (Lee et al., 2008; Stephenson Wang et al., 2001). In support of their findings, a more recent paper by Kolodkin-Gal et al., (2013) reported that the removal of PAS-A from KinA impairs sporulation, although the key data were not shown. Furthermore, they proposed that KinA activity is inhibited by high NAD + /NADH ratio through the PAS-A domain (Kolodkin-Gal et al., 2013). It is noted that the PAS-A deletion construct (aa 151-606) used in the study by Lee et al. (2008) featured an extra 8aa deletion which extended into the PAS-B domain (aa 143-150), giving it a structure more similar to our autokinase-deficient PAS-AB mutant (aa 259-606) (Eswaramoorthy et al., 2009). These facts suggest that the Lee et al. strain may not be truly representative of a PAS-A deletion mutant. Multiple distinct models have been proposed to describe potential roles for the PAS-A domain by individual research group as a result of the above discrepancies.
Nevertheless, based mainly on our in vitro and in vivo data Eswaramoorthy et al., 2009;Eswaramoorthy et al., 2011;Narula et al., 2016), we have proposed a model in which a threshold level of the major kinase KinA acts as a molecular switch to determine entry into sporulation, which may rule out the possibility that KinA is activated by sensing an as yet unidentified starvation signal(s) with the N-terminal PAS domains. Our model is supported by more recent data, in which cell growth slows down upon starvation, leading to a relative increase in the cellular concentration of KinA as its production rate remains constant (Narula et al., 2016). Thus, when the KinA threshold is reached, the flow of phosphoryl groups through the phosphorelay proceeds toward the production of the phosphorylated Spo0A sufficient to activate transcription of the high-threshold genes required for sporulation.
In this paper, to provide direct supporting evidence for our proposed model, we further explored the role of the N-terminal PAS domains of KinA in its enzymatic activity. We first investigated whether the PAS-A domain is dispensable for the autokinase activity using an in vitro assay. Second, we tested whether KinA activity is dependent on the tetramer formation mediated by the N-terminal PAS domains in vivo. Finally, with in vivo and in vitro biochemical characterizations, we attempted to rule out the previously proposed possibility, in which KinA activity is controlled by NAD + /NADH ratio through the PAS-A domain (Kolodkin-Gal et al., 2013).

| Strains, plasmids, and oligonucleotides
All B. subtilis strains were derived from the prototrophic strain PY79 (Youngman, Perkins, & Losick, 1984). Details of the constructions are available upon request. All plasmid constructions were performed in Escherichia coli DH5α using standard methods. The E. coli BL21(DE3) pET vector system (Novagen) was used for protein overexpression.
The strains and plasmids used in this study are listed in Tables S1 and S2, respectively. Oligonucleotides used for plasmid construction are listed in Table S3.

| Media, culture conditions, protein expression, and protein purification
To induce protein synthesis in B. subtilis cells, a gene of interest was placed under the control of an isopropylβ-D-thiogalactopyranoside (IPTG)-inducible hyper-spank promoter (Phy-spank) in a plasmid that integrates stably into the amyE locus of the chromosomal DNA. After plasmid integration into the host genome, the desired concentration of IPTG was added to Luria-Bertani (LB) cultures during the exponential growth phase (optical density at 600 nm [OD600], 0.5) at 37°C as the rich medium conditions. To induce protein synthesis in E. coli, the E. coli BL21(DE3) pET vector system was used according to manufacturer's protocol (Novagen). For his-tagged protein purification, the cells were harvested after 2 hr of induction and the induced proteins were purified by Ni-NTA beads according to the manufacturer's protocol (Invitrogen).

| Immunoblot analysis
Whole-cell lysates for immunoblot analysis were prepared by sonication. Protein concentration was measured by the Bradford method (Pierce). Total proteins were subjected to SDS-PAGE and transferred to a nitrocellulose filter. Immunoblot analysis was done with polyclonal anti-GFP antibodies (a gift from David Rudner). Alkaline phosphatasecoupled secondary antibodies (Anti-Rabbit IgG, Promega) were used for recognizing the primary antibodies and detected using substrate nitro blue tetrazolium-5-bromo-4-chloro-3-indolyl phosphate (NBT-BCIP) (Promega).

| Protein cross-linking
Protein cross-linking with bis-maleimidohexane (BMH; Pierce) was performed as described previously (Eswaramoorthy et al., 2009). A regression line of the log molecular mass versus the relative migration distance was generated using the HiMark Protein Standard (Thermo

| Blue native polyacrylamide gel electrophoresis
The blue native PAGE (BN-PAGE) assay was performed according to the protocol provided by Novex (Thermo Fisher Scientific).
In brief, the C-terminal hexahistidine-tagged GFP fusion proteins of were expressed under the control of an IPTG-inducible Phy-spank promoter in B. subtilis. To avoid the possible PAS domains-mediated heterocomplex formation between the native KinA (i.e., non-his tagged) and the his-tagged protein, a gene for the native KinA was deleted (ΔkinA::tet). The proteins examined were purified using Ni-NTA beads (Promega). The sample eluates were subjected to a 4-16% gradient BN-PAGE gel (Novex, Thermo Fisher Scientific), followed by immunoblot analysis with polyclonal anti-GFP antibodies (Fujita & Losick, 2002). A regression line of the log molecular mass versus the relative migration distance was generated using the NativeMark Unstained Protein Standard (Thermo Fisher Scientific).  (Sterlini & Mandelstam, 1969). Sporulation efficiency was measured by counting cells expressing GFP signals under the control of the foresporespecific spoIIQ promoter (PspoIIQ).
Proteins used for the phosphorylation assays were 0.2 μmol L −1 KinA, 0.2 μmol L −1 Spo0B, and 2 μmol L −1 Spo0A. The reaction was initiated by the addition of ATP to a final concentration of 0.4 mmol L −1 containing 1 μCi of [γ-32 P] ATP (purchased from Perkin-Elmer, 3000 Ci/mmol, 10 mCi/ml). Reaction mixtures were incubated for 30 min at 30°C and then stopped by adding SDS-loading buffer (Laemmli, 1970). Samples were subjected to electrophoresis through an 18% SDS-polyacrylamide gel, and radioactive proteins were visualized by autoradiography.

| Determination of cellular NAD + and NADH concentrations
Cellular NAD + and NADH concentrations were determined in B. subtilis PY79 using the NAD/NADH Quantification Kit provided by Sigma-Aldrich (Catalog Number MAK037, St Louis, MO). In brief, growing vegetative cells were cultured in LB for 4 hr at 37°C and sporulating cells were prepared in DSM for 5 hr (2 hr after onset of sporulation) at 37°C. Cell numbers were determined as colony-forming unit (CFU) with serial dilution and plate counts. One aliquot of cells (approximately 1 × 10 7 cells) was lysed with two freeze-thaw cycles in the NAD/NADH extraction buffer, and then total extracts were filtered through 3-kD cutoff filters (Amicon Ultra-5, Sigma-Aldrich) to recover only the unbound free forms of NAD + and NADH. Half of the flow-through lysate was used to determine total NAD concentrations (NAD + and NADH). The other half was heated to 60°C for 5 min and used to determine NADH concentrations. NADH standards were prepared as 0 (blank), 20, 40, 60, 80, and 100 pmole/reaction. A reaction mixture containing cycling enzyme provided by the kit was added to each sample and allowed to convert NAD + to NADH at 25°C for 5 min.
NADH developer supplied in the kit was then added to the sample reactions and the NADH standards were incubated for 15 min at 25°C and read at 450 nm. Finally, the NAD + concentration per reaction was determined by subtracting NADH concentration per reaction from total NAD (NAD + and NADH) concentration per reaction. Intracellular concentrations were calculated by taking into account the number of cells per reaction (determined with CFU) and cell volume (1 × 10 -15 L).
E. coli DH5α was used as a control (Bennett et al., 2009).

| The most amino-terminal PAS-A domain is dispensable for autokinase activity of KinA
To conclude the controversial debate on whether the PAS-A domain is required for the autokinase activity (Eswaramoorthy et  KinA ΔPAS-A is active similarly in the wild-type KinA, we performed an in vitro autophosphorylation assay under standard conditions (Fujita & Losick, 2003). We found that the purified KinA and KinA ΔPAS-A were both phosphorylated in the presence of radiolabeled ATP, in agreement with a previous report by Winnen et al., (2013) (Figure 2a The C-terminal domain is subdivided into DHp (dimerization and histidine phosphotransfer) domain and CA (catalytic ATP-binding) domain (Eswaramoorthy et al., 2009). Domain boundaries indicate the amino acid residue numbers from N-terminus. Positions of cross-linkable cysteine residues with bis-maleimidohexane (BMH) are indicated in the N-terminal domain. The histidine phosphorylation site is located at His405 as indicated. As a summary of this study, autokinase activity, tetramer (4-mer) formation, and reverse phosphotransfer activity (reverse reaction from Spo0F~P to KinA C ) are scored with a plus sign or a minus sign to denote detectable or not, respectively of KinA ΔPAS-A , similar to wild-type KinA. Taken together, these results indicated that the purified KinA ΔPAS-A displayed similar activity and kinetics to those of the wild-type enzyme, indicating that the PAS-A domain is dispensable for the autokinase activity in an in vitro assay, as it was demonstrated to be in an in vivo assay Eswaramoorthy et al., 2009Eswaramoorthy et al., , 2011. irrespective of nutrient availability (Eswaramoorthy et al., 2009(Eswaramoorthy et al., , 2011.

| KinA
Conversely, most other studies argue that KinA forms a homo-dimer as a functional unit (Dago et al., 2012;Lee et al., 2008;Szurmant & Hoch, 2013;Wang et al., 2001;Winnen et al., 2013). However, these reports used heterologous E. coli overexpression systems for preparation of a series of truncated KinA (e.g., C-terminal domain, N-terminal domain, and PAS-A domain), which might not be physiologically relevant and therefore may not reflect the reactions that take place under physiological conditions in B. subtilis. To further provide solid supportive data to clarify the above discrepancies, we examined com- as previously reported (Eswaramoorthy et al., 2009;. The cell extracts were then incubated in the presence or absence of bis-maleimidohexane (BMH), a protein cross-linker specific for a free sulfhydryl group that was shown to be effective in detection of the wild-type KinA tetramer (Eswaramoorthy et al., 2009). We note that KinA has five cysteine residues, one in PAS-A, that two cysteine residues in GFP (GFPmut2) (Cormack, Valdivia, & Falkow, 1996) used as a tag are not involved in the complex formation. In Figure 3b

| Removal of the N-terminal domain from KinA converts the autokinase into a phosphotransferase
It was demonstrated that the C-terminal domain of KinA is phosphorylated by a reverse phosphotransfer reaction from the phosphorylated form of Spo0F under both in vitro (Wang et al., 2001) and in vivo conditions (Eswaramoorthy et al., 2009). On the basis of these notions and the data in the previous sections in this study, we hypothesized KinA and KinB has been reported with the systematic yeast twohybrid screening using KinA or KinB as a bait (Fukushima, Yoshimura, Chibazakura, Sato, & Yoshikawa, 2006). Therefore, if the sporulation efficiency of cells expressing a certain domain is lower than that of the parental ΔkinA strain, it would suggest that the domain possesses reverse phosphotransfer activity (Eswaramoorthy et al., 2009). Using this "reverse phosphorylation" assay system, we systematically examined KinA, KinA ΔPAS-A , KinA ΔPAS-AB , KinA N , and KinA C for reverse phosphorylation activity ( Figure 5). To detect sporulation, we constructed a strain harboring a GFP reporter system for the forespore-specific spoIIQ gene promoter controlled by σ F -RNA polymerase. First, we verified that 6% of the parental ΔkinA cells showed GFP signals in the forespore compartment (Figure 5a, panel 1). When the wild-type KinA (Figure 5a,

| KinA activity is not regulated by the change of the NAD + /NADH ratio
A recent study by Kolodkin-Gal et al., (2013) reports that, when KinA is purified under sporulation conditions, NAD + binds to the PAS-A domain of KinA. Based on these results and other assumptions, they propose that impaired oxidative phosphorylation leads to a drop in NAD + levels, resulting in a conversion of the kinase from an inactive NAD + -bound form to an active free form which can participate in biofilm formation (Kolodkin-Gal et al., 2013). Another study reported by Gyan, Shiohira, Sato, Takeuchi, & Sato (2006) indicates that the NADH levels are relatively higher than those of NAD + during growth and stationary phases. In general, NAD + and NADH are ubiquitous and essential biomolecules, and thus these cellular concentrations are relatively high, ranging from 80 μmol L −1 (NADH) to 3 mmol L -1 (NAD + ) in E. coli (Bennett et al., 2009). We have previously reported that the KinA concentration is comparatively low (0.2-2 μmol L −1 ) during sporulation , while, to the best of our knowledge, the absolute cellular concentrations of the cofactors have not been reported in B. subtilis. Based on the available facts, we reasoned that, at any given moment during culture, the concentrations of NAD + and NADH could be substantially higher than that of KinA. Under such conditions, if NAD + specifically inhibits KinA with direct binding, there might be low-affinity, but specific interactions between NAD + and KinA.
To clarify the above possibilities and provide direct evidence, we measured autokinase activities of the purified KinA and KinA ΔPAS-A from LB culture, in the presence of varying concentrations of NAD + and NADH using an in vitro assay. As a control, KinC ΔTM1+2 , the soluble and functional form KinC that is mainly involved in biofilm formation (Devi, Vishnoi, et al., 2015), was used. We found that autokinase activities of KinA and KinA ΔPAS-A , as well as KinC ΔTM1+2 , are constant in the presence of excess concentrations of NAD + and NADH (up to 1000fold excess, similar to the in vivo conditions, Figure 6, lanes 3 and 5).
Finally, we directly determined the cellular concentrations of NAD + and NADH during growth and sporulation (Table 1). To prevent undesired enzymatic degradation of NAD + and NADH and also to recover the protein-unbound free forms, the cell lysates were filtrated through a 3 kDa cutoff filter. In cells growing in rich medium (LB), the concentrations of NAD + and NADH were approximately 3 mmol L −1 and 1 mmol L −1 , respectively. In cells under sporulation conditions (DSM), the same conditions demonstrated by Kolodkin-Gal, those concentrations were found to be approximately 1 mmol L −1 and 0.4 mmol L −1 , respectively. As a control for the accuracy of this assay, we also measured the concentrations of the cofactors for E. coli cells grown in LB and DSM media. The assay returned values of approximately 2 mmol L −1 and 0.4 mmol L −1 for NAD + and NADH, respectively, which are similar to those previously reported for E. coli under the same conditions (Bennett et al., 2009). These results indicate that NAD + and NADH exist in approximately 1000-fold excess of KinA (0.2-2 μmol L −1 ). Thus, these results suggest that the NAD + and NADH ratio does not change significantly during growth and sporulation and, under such conditions, changes in the ratio between the two forms of the cofactors might not play a significant role in regulation of KinA activity, even if NAD + does bind to the PAS-A domain of the wild-type KinA as reported (Kolodkin-Gal et al., 2013).

| DISCUSSION
It has long been debated how the major sporulation kinase is activated upon starvation. A general model, like a simple ligand-receptor model (Henry & Crosson, 2011) or a traditional bacterial two-component F I G U R E 6 Effects of NAD + and NADH on the autophosphorylation of KinA, KinA ΔPAS-A , and KinC ΔTM1+2 . In vitro autophosphorylation assays were performed with each of the purified proteins (0.2 μmol L −1 ) in the presence of the indicated concentrations of NAD + or NADH. Reaction mixtures were analyzed by SDS-PAGE followed by autoradiography  (Stock et al., 2000), has been proposed based on speculation that an as yet unidentified signaling molecule(s) produced only under starvation conditions binds directly to the N-terminal "sensor" PAS domains, leading to the C-terminus autokinase activity (Hoch, 1993(Hoch, , 2000Stephenson & Hoch, 2002;Wang et al., 2001). However, to date, no ligand that directly interacts with the PAS domains and activates the autokinase activity has been reported. Alternatively, an as yet unidentified signaling molecule(s) produced only under nutrientrich conditions could bind directly to the N-terminal PAS domains, interfering with the C-terminus autokinase activity, in a similar fashion to bacterial chemotaxis (Hazelbauer, Falke, & Parkinson, 2008).
Recently, it has also been suggested that NAD + directly binds to the PAS-A domain (Kolodkin-Gal et al., 2013). While no direct supporting evidence is provided in that report, it is proposed that KinA activity is directly inhibited by the NAD + binding to PAS-A when the relative cellular concentration of the cofactors (NAD + /NADH) is high during growth in the presence of oxygen (Kolodkin-Gal et al., 2013).
Furthermore, the removal of the PAS-A domain reportedly lowers the autokinase activity (Kolodkin-Gal et al., 2013;Lee et al., 2008;Wang et al., 2001), contrary to the other studies Eswaramoorthy et al., 2009;Winnen et al., 2013). In support of the "sensor" model, the previous studies propose that PAS-A acts as an important regulator of autokinase activity in response to environmental change (Kolodkin-Gal et al., 2013;Lee et al., 2008;Wang et al., 2001).
In an attempt to conclude these debates, we have provided direct evidence in this study that (1)  Our data in this study may rule out the possibility of NAD + as an inhibitory ligand for KinA (Kolodkin-Gal et al., 2013). It is known that NAD + and NADH are involved in hundreds of reactions in cells, including not only central metabolic pathways (Holms, 1996) but also RNA metabolism and regulation (Jaschke, Hofer, Nubel, & Frindert, 2016).
Thus, to maintain essential cellular functions, concentrations of the cofactors are generally high, within millimolar ranges (Bennett et al., 2009). In support of these notions, in growing and sporulating B. subtilis, cellular concentrations of free NAD + and NADH (millimolar range) are approximately 1000-fold higher than that of KinA (micromolar range as a monomer, but note that KinA is a tetramer) . Furthermore, the ratio between two cofactors' concentrations does not significantly change between growth and sporulation conditions. In addition, our in vitro data indicate that the kinase activity is not inhibited in the presence of excess amounts of each of the cofactors. Therefore, the results in this study appear to Recent studies using a combination of empirical and mathematical modeling suggest that cells' growth rates decrease in response to nutrient starvation, leading to an increased cellular concentration of KinA and thus increased activation of Spo0A through the phosphorelay (Narula et al., 2016). In support of this model, it is important to note that the autophosphorylation activity of KinA is independent of enzyme concentration and also an as yet unidentified starvation signal(s), suggesting that the subunit assembly into a functional and constitutively active KinA tetramer is independent of the concentration of the individual protomers and the unknown starvation signal(s) (Eswaramoorthy et al., 2011). Thus, both empirical and modeling results suggest that the cellular concentration of the constitutively active KinA increases as its dilution slows down when cell growth ceases in response to nutrient starvation, thus, in turn, allowing the increase in KinA level to act as a trigger factor for phosphorelay activation.
Furthermore, other studies suggest that Spo0A is activated in a pulsatile manner every cell cycle, resulting in a gradual increase in Spo0A activity over the course of multiple cell cycles (Levine, Fontes, Dworkin, & Elowitz, 2012). These pulsatile activations are controlled by the chromosomal arrangement of two phosphorelay genes, one (spo0F) located at the origin-proximal part of the chromosome and the other (kinA) located near the terminus. During growth, replication of circular DNA starts at the origin, moves around the chromosome in both directions, and ends when the two replication forks meet each other on the opposite side of the chromosome. After replication starts but prior to cell division, a transient imbalance in gene dosage between originproximal (spo0F) and terminus-proximal (kinA) genes occurs, resulting in a temporary ratio of spo0F:kinA = 2:1. As a result, substrate inhibition of KinA by Spo0F occurs, leading to a delay in Spo0A activation through the phosphorelay (Narula et al., 2015).
Taken together, these recent findings suggest that the starvation event cannot be defined by the level of any single metabolite, but rather nutrients are consumed by cells, resulting in the production of a complex set of metabolites and the slowdown of cell growth. Thus, growth rate change in response to the available nutrients is a reasonable and accurate means of monitoring the availability of a variety of nutrients in the environment. Using this strategy, B. subtilis cells can integrate multiple sources of complicated environmental information into sporulation regulation by monitoring the cellular levels of KinA at the top of phosphorelay.