Activation of extracytoplasmic function sigma factors upon removal of glucolipids and reduction of phosphatidylglycerol content in Bacillus subtilis cells lacking lipoteichoic acid

Edited by Kei Asai * Corresponding author. E-mail: matsuoka@mail.saitama-u.ac.jp † Present address: Department of Applied Chemistry and Biotechnology, Faculty of Engineering, Chiba University, 1-33 Yayoi-cho, Inage-ku, Chiba, Chiba 263-8522, Japan DOI: http://doi.org/10.1266/ggs.18-00046 Activation of extracytoplasmic function sigma factors upon removal of glucolipids and reduction of phosphatidylglycerol content in Bacillus subtilis cells lacking lipoteichoic acid


INTRODUCTION
Bacillus subtilis has seven extracytoplasmic function (ECF) sigma factors, σ M , σ V , σ W , σ X , σ Y , σ YlaC and σ Z , which are a group of bacterial sigma factors that direct transcription of genes involved in tasks such as maintenance of cells upon external stresses.Except σ Z , these are regulated directly by their respective cognate transmembrane anti-σ factors, which sequester the σ factors and keep them inactive.Under stress conditions the σ factors are released from the anti-σ factors, form RNA polymerase holoenzymes and transcribe the respective regulon genes.The genes encoding σ and anti-σ factors form operons whose transcription is autoregulated.Acti-vation of ECF sigma factors can result from a number of stresses on the cell envelope.For example, σ M is activated in response to high salt, cell wall antibiotics, ethanol, heat, acid and superoxide stresses (Thackray and Moir, 2003;Asai et al., 2005;Luo and Helmann, 2012); σ V is strongly and specifically activated in response to lysozyme (Guariglia-Oropeza and Helmann, 2011;Ho et al., 2011); σ W is activated in response to salt, alkali, antibiotics inhibiting cell wall biosynthesis and oxidative stresses (Petersohn et al., 2001;Helmann, 2002;Cao et al., 2003;Asai et al., 2005;Butcher and Helmann, 2006); and σ X is activated by inhibitors of peptidoglycan and wall teichoic acid (WTA) biosynthesis, and serves to alter cell envelope properties to protect against cationic antimicrobial peptides (Helmann, 2002;Cao and Helmann, 2004;Murray and Stanley-Wall, 2010).Among these ECF sigma factors, it is clear that σ V and σ W are activated by proteolysis of anti-sigma factors via regulated intramembrane proteolysis (Schobel et al., 2004;Zellmeier et al., 2006;Heinrich et al., 2009;Hastie et al., 2013Hastie et al., , 2014Hastie et al., , 2016)).
As shown in Fig. 1, B. subtilis membranes contain various lipids, including glucosylated diacylglycerol, monoglucosyldiacylglycerol (MGlcDG), diglucosyldiacylglycerol (DGlcDG) and triglucosyldiacylglycerol (TDGlcDG), and a complex glycerophospholipid, lysylphosphatidylglycerol (LysylPG), in addition to the glycerophospholipids found in Escherichia coli membranes, phosphatidylethanolamine (PE), phosphatidylglycerol (PG) and cardiolipin (CL) (Matsumoto et al., 2015).All genes responsible for synthesis of final lipid products are dispensable, except for pgsA, which encodes phosphatidylglycerophosphate synthase and is responsible for the committed step in PG synthesis.Lipoteichoic acid (LTA), a polymer of glycerol-1-phosphate derived from PG attached to DGlcDG, by which it is anchored on the membrane, is present on the outer leaflet of the membrane (Neuhaus and Baddiley, 2003;Reichmann and Grundling, 2011).
ugtP encodes a glucosyltransferase that processively transfers glucose from UDP-glucose to diacylglycerol (Jorasch et al., 1998).Cells with a deletion of ugtP that are completely lacking glucolipids show abnormal cell morphology and activation of three ECF sigma factors, σ M , σ V and σ X (Salzberg and Helmann, 2008;Matsuoka et al., 2011;Hashimoto et al., 2013).The phenotype is complemented by expression of heterologous glucolipid synthase from Acholeplasma laidlawii (Matsuoka et al., 2016).Moreover, we reported that σ V is activated without proteolysis of RsiV in the ΔugtP cells (Seki et al., 2017).Cells harboring P spac -pgsA in place of the wildtype pgsA allele have a PG content reduced by half (from 28.1% to 14.1%), arrested cell growth, and induction of σ M and σ V at 60 min after removal of inducer (Hashimoto et al., 2009).
Bacillus subtilis has four LTA synthase genes homologous to LtaS in Staphylococcus aureus: ltaS (formerly yflE), yfnI, yqgS and yvgJ (Gründling and Schneewind, 2007a;Schirner et al., 2009).All LTA synthesis genes can be simultaneously deleted and therefore LTA is dispensable for B. subtilis growth (Schirner et al., 2009;Wörmann et al., 2011).Finally, six ECF sigma factors, σ M , σ X , σ W , σ YlaC , σ V and σ Y , are induced in cells lacking LTA because of disruption of all four LTA synthase genes (Hashimoto et al., 2013).However, the exact contributions of these membrane components to the activation of ECF sigma factors in B. subtilis cells are still unclear.
In the case of S. aureus, deletion of ypfP, which is a homolog of B. subtilis ugtP, induces an alteration in the anchor of LTA from DGlcDG to DG (Kiriukhin et al., 2001;Gründling and Schneewind, 2007b).Therefore, it is believed that reduction of PG content and absence of glucolipids in B. subtilis cause abnormalities in LTA, as a result of which some ECF sigma factors are induced in cells lacking glucolipids or having reduced PG content.However, the pattern of activation of ECF sigma factors differs between B. subtilis cells with a deletion of some LTA synthase genes and those without glucolipids or with reduced PG content (Hashimoto et al., 2009(Hashimoto et al., , 2013)).To address this difference, we investigated the effect of PG reduction and absence of glucolipids on LTA profile and activity of ECF sigma factors in this study.Matsumoto et al. (2015).The gene product catalyzing each step is indicated.PG, Phosphatidylglycerol; PE, phosphatidylethanolamine; CL, cardiolipin; lysylPG, lysylphosphatidylglycerol; PA, phosphatidic acid; CDP-DG, CDP-diacylglycerol; MGlcDG, monoglucosyldiacylglycerol; DGlcDG, diglucosyldiacylglycerol; TGlcDG, triglucosyldiacylglycerol; DG, diacylglycerol; LTA, lipoteichoic acid.
Genetic and recombinant DNA procedures These were based on standard methods (Miller, 1992;Sambrook and Russell, 2001).
SDS-PAGE and immunoblot analysis of LTA Cells of 168 (wild-type), YTB019 (ΔugtP) and SLD08 (ΔltaS ΔyfnI ΔyqgS ΔyvgJ) were grown to OD 530 = 0.5 at 37 °C in 5 ml LB medium and then diluted 1/10 into 5 ml of fresh LB medium and grown to OD 530 = 0.5.MHB001 (pgsA::P spac -pgsA) cells were cultivated overnight at 37 °C in 5 ml LB medium containing 1 mM IPTG and then washed three times using fresh LB medium without IPTG and diluted to OD 530 = 0.05 into 5 ml fresh LB medium without IPTG.Cells were grown to the point at which turbidity stopped increasing (OD 530 = 0.5) and 5 ml of culture was collected.Bacterial pellets were suspended in 2 × protein sample buffer normalized for OD 530 readings; that is, 25 μl of 2 × sample buffer was used per ml culture of OD 530 = 0.5.Samples were boiled for 45 min, centrifuged at 19,000 g for 5 min at 4 °C, and 10 μl of the supernatant was loaded on a 15% Extra PAGE One Precast Gel Real-time RT-PCR Cells of 168, YTB019, SLD08, SLD10 (ΔltaS ΔyfnI ΔyqgS ΔyvgJ ΔugtP), MHB001 and SLD12 (ΔltaS ΔyfnI ΔyqgS ΔyvgJ pgsA::Pspac-pgsA) were grown under the same conditions described above for the SDS-PAGE experiments.A 500-μl aliquot of each sample was collected, and RNA was then extracted using a Total RNA Purification Kit (Jena Bioscience) and an RNasefree DNase set (Roche Applied Science) according to the manufacturer's recommendations.The DNase-treated RNA was reverse-transcribed using ReverTra Ace qPCR RT Master Mix with gDNA Remover (TOYOBO) according to the manufacturer's recommendations.Real-time RT-PCR with SYBR Green I was performed using Thunderbird SYBR qPCR Mix (TOYOBO).Real-time RT-PCR was carried out as described previously, using the pairs of primers for sigA, sigM, sigV, sigW, sigX, sigY and ylaC listed in Supplementary Table S1 (Hashimoto et al., 2013).Each real-time RT-PCR was performed in at least quintuplicate with sigA as the internal standard.
β-galactosidase assay in E. coli The E. coli cells with plasmids were grown to mid-log phase and 1-ml aliquots of the culture were withdrawn, and β-galactosidase activity in the cells was assayed as described previously (Wang and Doi, 1984).

RESULTS AND DISCUSSION
Lack of glucolipids affects the structure of LTA To reveal the effect of an absence of glucolipids and reduction of PG content on the LTA profile in B. subtilis, cell extracts from a mid-log phase culture (OD 530 = 0.5) of wild-type cells (168 strain, hereafter called WT cells), cells lacking glucolipids (YTB019 strain, hereafter called ΔGLs cells), cells with a reduced PG content (MHB001 strain, hereafter called PG Red.cells) and cells lacking LTA (SLD08 strain, hereafter called ΔLTA cells) were prepared and LTA content analyzed by Western blot.A signal in the 15-kDa area was detected in WT cells but not in ΔLTA cells (Fig. 2), indicating that it is an LTAspecific signal.In ΔGLs cells the signal was shifted to the 20-kDa area (Fig. 2).This reduced LTA migration on SDS-PAGE observed in B. subtilis cells with a deletion of ugtP suggests that ugtP deletion induces the alteration of the LTA anchor from DGlcDG to DG, as seen in the case of the S. aureus ypfP deletion (Gründling and Schneewind, 2006).However, no change of signal pattern was observed in PG Red.cells.These results suggest that not the reduction of PG, but the absence of DGlcDG affected the LTA profile.In addition, a broad signal  around the 35-kDa area, which is a WTA signal caused by LTA antibody cross-reacting with WTA (Wörmann et al., 2011), was observed in ΔLTA cells (Fig. 2).Although the signal around 35 kDa was not observed in WT cells or PG Red.cells, a weak signal was detected in ΔGLs cells.The change of LTA profile as well as the lack of LTA may reflect an increase of the WTA content to compensate for the LTA abnormality.
Lack of glucolipids activates ECF sigma factors independent of abnormalities in LTA To determine whether or not the lack of glucolipids activates ECF sigma factors independent of the change in LTA profile observed in ΔGLs cells (Fig. 2), we deleted ugtP from ΔLTA cells and analyzed the activation levels of ECF sigma factors.The mRNA level of ECF sigma factors was regarded as the activation level, because the transcription of each ECF sigma factor is autoregulated.The mRNA levels of the genes for six ECF sigma factors in ΔGLs cells, ΔLTA cells and cells lacking both LTA and glucolipids (SLD10 strain, hereafter called ΔLTA ΔGLs cells) were determined by real-time RT-PCR to estimate the activation levels of sigma factors.In ΔGLs cells the mRNA levels of the four ECF sigma factor genes sigM, sigX, sigV and sigY were increased 2.3-, 1.5-, 2.7-and 1.5-fold, respectively, compared with those of WT cells (Fig. 3).In this work a novel activation of σ Y was observed.In ΔLTA cells the mRNA levels of sigM, sigX, sigW, ylaC, sigV and sigY were increased (Fig. 3).In ΔLTA ΔGLs cells the mRNA levels of sigM, sigX, sigV and sigY were increased 4.5-, 1.6-, 4.1-and 3.7-fold, respectively, compared with those of WT cells, but the levels of sigW and ylaC were not changed (Fig. 3).This result shows that the effects of a lack of both glucolipids and LTA on the activation of three sigma factors (sigM, sigV and sigY) were additive in ΔGLs ΔLTA cells, suggesting that the removal of glucolipids activates the three ECF sigma factors σ M , σ V and σ Y independent of the absence of LTA.
To investigate whether the additive increase of activation levels of σ M , σ V and σ Y observed in ΔLTA ΔGLs cells was caused by a secondary effect on the lipid composition induced by simultaneous deletion of the ugtP and LTA synthase genes, the lipid composition of cells lacking glucolipids and/or LTA was examined (Table 2).In ΔLTA cells, CL, CDP-DG and DG were accumulated (2.3-, 6.9and 2.8-fold, respectively), and DGlcDG and LysylPG were reduced (4.0-and1.5-fold) compared with their content in WT cells.In ΔGLs cells, the CL content of ΔGLs cells was increased (2.7-fold) over WT cells and CDP-DG content was decreased (8.0-fold).In ΔLTA ΔGLs cells, the CDP-  ).WT cells, ΔLTA cells, ΔGL cells and ΔLTA ΔGLs cells were cultivated in LB medium to exponential growth phase (OD 530 = 0.5) and total RNA was extracted and used for preparation of cDNA samples.Cultures of PG Red. cells and ΔLTA PG Red. cells were grown in LB medium supplemented with 1 mM IPTG to exponential growth phase, transferred into fresh medium and cultivated without IPTG to the point at which turbidity reached an OD 530 of 0.5, and then total RNA was extracted and used for preparation of cDNA samples.The cDNA samples were then subjected to real-time RT-PCR analysis with the appropriate primer pairs for sigM, sigV, sigW, sigX, sigY and ylaC, with the primer pair for sigA as the internal control.Mean values ± SD, calculated from six measurements excluding the highest and lowest values obtained from eight PCR assays, of the level of each sigma factor mRNA in the mutant cells, relative to those of WT cells (dotted line), are presented.Asterisks indicate significant differences as determined by Student's t-test: *P < 0.01.DG content was also decreased (6.9-fold) compared with that of ΔLTA cells, indicating that ugtP deletion induces the reduction of CDP-DG content.Furthermore, the CL content was decreased (2.0-fold) in ΔLTA ΔGLs cells compared with that in ΔLTA cells.Since CL was accumulated in ΔGLs cells, the decrease of CL content observed in the ΔLTA ΔGLs cells was not caused by ugtP deletion.Thus, the change of CL content in the ΔLTA ΔGLs cells may be caused by a secondary effect of simultaneous deletion of the ugtP and LTA synthase genes.Table 3 shows the relationship between the activated ECF sigma factors and the lipid content, which were changed in each mutant cell.In ΔLTA ΔGLs cells, only glucolipid content was changed.Nevertheless, σ M , σ V and σ Y were additively activated.If the change of CL and CDP-DG content activates σ M , σ V and σ Y , the activation levels of these ECF sigma factors should be decreased in the ΔLTA ΔGLs cells, because these lipids were back to wild-type levels.However, the activation levels of σ M , σ V and σ Y were additively increased in the ΔLTA ΔGLs cells (Fig. 3).Therefore, the change of CL and CDP-DG content is not the cause of the activation of σ M , σ V and σ Y .
Taken together, these results show that a lack of glucolipids activates σ M , σ V and σ Y independent of the changes in LTA structure and other lipid composition caused by deletion of ugtP.The results also suggest that σ M , σ V and σ Y sensed the lack of glucolipids and of LTA as distinct stresses, but that σ X sensed them as a single stress.Although there are no reports that a lack of glucolipids is induced by environmental stresses in B. subtilis, we found that glucolipid content decreased when B. subtilis cells were grown in high salt condition (S.Matsuoka, unpublished results).Therefore, the membrane of cells without glucolipids may be mimicking that of cells exposed to high salt stress.
The activation levels of four ECF sigma factors, except σ X and σ Y , in ΔLTA PG Red. cells corresponded with those in PG Red. cells (Fig. 3).To test whether the correspondence of the activation levels of the four ECF sigma factors (σ M , σ W , σ YlaC and σ V ) between ΔLTA PG Red. cells and PG Red. cells was caused by insufficient reduction of PG content in ΔLTA PG Red.cells, the lipid composition of PG Red. cells and ΔLTA PG Red. cells was examined (Table 2).In PG Red. cells CL and lysylPG, which are positioned downstream of PG in the lipid biosynthesis pathway (Fig. 1), were also reduced (2.2-and 1.7-fold, respectively) in addition to PG (10.6-fold) compared with their levels in WT cells.Moreover, the PE and DG content of PG Red. cells was increased (2.0-and 3.9-fold, respectively).In ΔLTA PG Red. cells PG, CL and lysylPG levels were decreased (3.7-, 2.9-and 1.7fold, respectively) and PE and DG levels were increased (2.4-and 1.8-fold, respectively) compared with those in ΔLTA cells.The reduction level of PG compared to that in WT cells (6.4-fold) in ΔLTA PG Red. cells was comparable with that seen in PG Red.cells.This indicates that the correspondence of the activation levels of the four ECF sigma factors between ΔLTA PG Red. cells and PG Red. cells was not caused by insufficient reduction of the PG content of ΔLTA PG Red.cells.Table 3 shows that in ΔLTA PG Red.cells, the effects of deletion of LTA synthase genes and restriction of pgsA expression on lipid composition were additive.Nevertheless, the effects of deleting LTA synthase genes on activation levels of σ M , σ W , σ YlaC and σ V were masked.Therefore, the correspondence suggests that the activation mechanisms of σ M , σ W , σ YlaC and σ V sensed the reduction of PG content and lack of LTA as the same stress.As in the case of E. coli LacY lactose permease, proper distribution of negative charge on the cell membrane may be required for folding and function of anti-sigma factors (Dowhan and Bogdanov, 2009;Bogdanov et al., 2010).The effect of a lack of LTA on membrane negative charge density may be weak enough to be masked by the reduction of PG content, because LTA, which is localized to the outer leaflet of the cell membrane through its linkage to DGlcDG, is farther from the membrane than PG, which is a component of the membrane.Hence, the effect of the absence of LTA on activities of ECF sigma factors in ΔLTA PG Red. cells may be masked owing to the reduction in PG levels.Since PG is an essential lipid molecule for B. subtilis growth, significant reduction of PG content is not induced by environmental stress.On the other hand, the PG content of a daptomycin-resistant strain is reportedly low, similar to that of PG Red. cells (Hachmann et al., 2011).σ M and σ W are also activated in the daptomycin-resistant strain, suggesting that these ECF sigma factors are activated by a reduction of PG content.
Induction of σ M and σ V in E. coli cells lacking PG Escherichia coli can lack PG due to deletion of pgsA (Kikuchi et al., 2000;Suzuki et al., 2002).To confirm that the activities of σ M , σ V and σ X were altered by the absence of PG, the activation levels of σ M , σ V and σ X were examined using an E. coli heterologous expression system.We deleted the pgsA gene in a strain with a defect in lpp and introduced two plasmids: one was a pBR322 derivative having an operon of an ECF sigma factor and anti-sigma factor (either sigM-yhdLK, sigV-rsiV or sigX-rsiX) and the other an F´ plasmid having a transcriptional fusion of a promoter recognized by an ECF sigma factor and lacZ (P sigM´-lacZ, P sigV -lacZ or P sigX´-lacZ) as a reporter, allowing us to monitor β-galactosidase activity as a proxy for the induction levels of σ M , σ V and σ X .In the ΔpgsA cells the activities of σ M and σ V were significantly increased, by 13.7-and 2.3-fold, respectively, compared with those in WT cells (Fig. 4A), but were not increased in ΔpgsA cells lacking the sigM-yhdLK or sigV-rsiV operons (Fig. 4B), indicating that σ M and σ V are induced by the absence of PG in E. coli cells.These results support our hypothesis that the reduction of PG content activates σ M and σ V independent of the absence of LTA in ΔLTA PG Red. cells of B. subtilis (Fig. 3).Unexpectedly, the sigX-rsiX operon did not work properly in E. coli ΔpgsA cells (Fig. 4A).

CONCLUSION
In this study, we found that the deletion of ugtP additively activated σ M , σ V and σ Y in cells lacking LTA, and that the restriction of pgsA expression also additively activated σ M , σ V and, σ W (Fig. 3).Moreover, σ M and σ V responded directly to a lack of PG in E. coli as well as a lack of glucolipids (Fig. 4A) (see also Seki et al., 2015).Therefore, we concluded that a lack of glucolipids and a reduction of PG content directly activate some ECF sigma factors independently of LTA disorder in B. subtilis.Our data also suggested that σ M and σ V can sense ΔGLs and PG red.-ΔLTA as distinct stresses; σ X can sense ΔGLs-ΔLTA and PG Red.-ΔLTA as distinct stresses; σ Y can sense ΔGLs and ΔLTA as distinct stresses; σ W and σ YlaC can sense PG Red.-ΔLTA; and σ X can sense ΔGLs-ΔLTA (Fig. 5).This implies that these molecules have specific or overlapping functions on the membrane or cell surface.To determine the exact function of each lipid and LTA in the cells, further analysis of the activation mechanisms of each of the ECF sigma factors will be necessary.
We are grateful to K. C. Burtis for carefully reading the manuscript.T. S. is a research fellow of the Japan Society for the Promotion of Science (JSPS).This work was supported in part by JSPS KAKENHI (Grant-in-Aid for JSPS Research Fellows) Grant Numbers JP16J10629 to T. S. and 15K18664 to S. M.   ) were not distinguishable in their activation of σ M and σ V .Therefore, ΔGLs and PG Red.-ΔLTA activate σ M and σ V as distinct stresses.σ Y was activated by ΔGLs and ΔLTA independently but not activated by PG Red. .σ X was not additively activated by ΔGLs and ΔLTA.Hence, these activate σ X as the same stress.ΔLTA and PG Red. were not distinguishable in their activation of σ W and σ YlaC .Hence, these activate σ W and σ YlaC as the same stress. Fig.1

Fig. 5 .
Fig. 5. Summary of relationship between activation of ECF sigma factors and changes of lipid composition.Lack of GLs (ΔGLs) and lack of lipoteichoic acid (ΔLTA) activated σ M and σ V independently, whereas ΔLTA and reduction of PG content (PG Red. ) were not distinguishable in their activation of σ M and σ V .Therefore, ΔGLs and PGRed.-ΔLTA activate σ M and σ V as distinct stresses.σ Y was activated by ΔGLs and ΔLTA independently but not activated by PG Red. .σ X was not additively activated by ΔGLs and ΔLTA.Hence, these activate σ X as the same stress.ΔLTA and PGRed.were not distinguishable in their activation of σ W and σ YlaC .Hence, these activate σ W and σ YlaC as the same stress.

Table 3 .
Relationship between activation of ECF sigma factors and lipid composition

Table 2 .
Lipid compositions of mutants