A role for the essential YycG (WalK) sensor histidine kinase in sensing cell division

Summary

The YycG (WalK) sensor histidine kinase coordinates cell wall remodeling with cell division in Gram-positive bacteria by controlling the transcription of genes for autolysins and their inhibitors. Bacillus subtilis YycG senses cell division and is enzymatically activated by associating with the divisome at the division septum. Here it is shown that the cytoplasmic PAS domain of this multi-domain trans-membrane kinase is a determining factor translocating the kinase to the division septum. Furthermore, translocation to the division septum, per se, is insufficient to activate YycG, indicating that specific interactions and/or ligands produced there are required to stimulate kinase activity. N-terminal truncations of YycG lose negative regulation of their activity inferring that this regulation is accomplished through its transmembrane and extra-membrane domains interacting with the membrane associated YycH and YycI proteins that do not localize to the divisome. The data indicate that YycG activity in non-dividing cells is suppressed by its interaction with YycH and YycI and its activation is coordinated to cell division in dividing cells by specific interactions that occur within the divisome.

Introduction

Two-component signal transduction systems of bacteria have long been thought of as a means of niche adaptation, by regulating the expression of genes for enzymes involved in utilization of specific nutrients, or for expression of virulence factors to name a few. Systems that are embedded in regulating developmental decisions in response to the environment, such as development of genetic competence (Kovacs et al., 2009) or the ability to sporulate (Hoch, 2000) and those that regulate motility (Szurmant and Ordal, 2004) are regulated by multiple signals and are more complex. A limited number of signal transduction systems in the bacteria have proven to be essential for viability and wrapped within the regulation of cell growth and division. One such essential signaling system, the YycFG system (also referred to as WalRK, but named YycFG hereafter, for clarity), is conserved in bacteria of the order firmicutes and is the main subject of this study.

The YycFG system is a prototypical two-component signal transduction system comprised of a multi-domain transmembrane sensor histidine kinase YycG and a response regulator/transcription factor YycF (Fabret and Hoch, 1998; Dubrac et al, 2008; Winkler and Hoch, 2008). The first clues to the role of this system in growth came from depletion studies of the operon that showed the formation of chains of cells with some having lost their cellular contents (Fabret and Hoch, 1998). Overexpression of the response regulator YycF resulted in the formation of mini-cells, which was shown to be associated with elevated levels of FtsZ, the tubulin like master regulator of cell division (Fukuchi et al., 2000). The ftsZ gene was found to be controlled by several promoters and one of them proved to be directly responsive to phosphorylated YycF, making ftsZ the first known gene of the YycFG regulon.

Microarray studies on conditional B. subtilis strains led to the discovery of a consensus binding site for the YycF protein, and to the identification of additional genes of the YycF regulon (Bisicchia et al., 2007; Howell et al., 2003). From these studies it became apparent that YycF is a positive regulator of the expression of a number of autolysins and other enzymes involved in cell wall restructuring and a repressor of the expression of inhibitors of these enzymes. Thus under growth conditions the autolysins required for cell division are induced and their inhibitors repressed. At the end of growth the inhibitors are derepressed to prevent autolysin activity and lysis.

Similar studies of the orthologous systems in Staphylococcus aureus and Streptococcus pneumoniae have confirmed essentiality in these organisms and a general theme for this system in regulating the expression of proteins involved in cell wall restructuring has emerged. Nevertheless, there appears to be diversity in the individual genes that are controlled by YycFG in the different organisms (Dubrac et al., 2007; Howell et al., 2003; Ng et al., 2003; Ng et al., 2005).

While it was clear that the YycFG system was embedded within the regulation of cell wall turnover and perhaps other processes, the factors controlling its activity remained a mystery. In order to answer this question, it was necessary to define the signals to which the YycG kinase responded. In a quest for such signals two additional genes in the yycFG operon, yycH and yycI, were identified by transposon mutagenesis as being involved in the signaling process (Szurmant et al., 2005). Deletion of yycH and/or yycI resulted in strains that failed to reach wild type cell densities in liquid media and showed an enhanced susceptibility to lysis. These phenotypes are known now to result from over activity of the YycG kinase. This demonstrated that the YycFG system performs a homeostatic role, since miss-regulation of the YycF-regulon, both, due to too little or too much phosphorylation has detrimental effects on cellular growth (Szurmant et al., 2007b).

Structures of the YycH and YycI extracytoplasmic membrane anchored proteins (Santelli et al., 2007; Szurmant et al., 2006) allowed structure based site-directed mutagenesis studies of these proteins. These experiments revealed that the single N-terminal transmembrane helices of the two proteins are able to complement for all the phenotypes of the yycH and yycI deletion strains and that they form a transmembrane helix complex with the YycG kinase, a structural model of which could be generated by molecular dynamics simulation and verified by scanning mutagenesis (Szurmant et al., 2008). Thus the YycG kinase is associated in the membrane as a complex with YycH and YycI through interaction of their transmembrane domains.

The results of these studies served to increase the complexity of the regulation of the YycG kinase but did not identify the signals to which the YycFG system and its YycH and YycI auxiliary proteins respond. Deeper insight into this regulation was achieved when immuno-fluorescent localization studies revealed that the YycG kinase co-localized with FtsZ, the tubulin-like master regulator of cell division, to the division septa (Fukushima et al., 2008) joining a growing number of septally localized essential proteins that form the so-called divisome. Transcription studies of YycF regulated genes confirmed that YycG septum localization was required for its activation and further cemented the relationship between the YycFG system and division and cell wall remodeling (Fukushima et al., 2008). Thus the results suggested that localization of YycG to the divisome was a means to ensure coordination of division with cell wall remodeling processes and pointed to this association as source of signals to activate the histidine kinase activity of YycG.

The current study explores the physical and functional connections between the YycG kinase and the cell division apparatus. Two-hybrid studies revealed that of the known cell division proteins, latter stage cell division proteins interact with YycG but are nevertheless individually dispensable for YycG septum localization. Despite accurate septum localization of YycG in some of these mutants, signal activation of YycG was not observed. Truncation studies of the YycG kinase revealed that septum localization is a property of the cytoplasmic domains of the kinase and occurs normally when the transmembrane domains are deleted from YycG. Localization studies on the auxiliary proteins YycH and YycI demonstrated that these regulators do not migrate to the septum along with YycG. Thus one important aspect of YycG activity regulation may be the physical separation of YycG from YycH and YycI during the cell cycle. In their entirety the data is supportive of a model in which the state of cell division is sensed by YycG in an interaction competition mechanism that involves YycH and YycI as negative regulators of YycG activity and the cell division proteins FtsW, FtsL and DivIB as positive regulators of YycG activity.

Results

N-terminal domain truncation of the YycG histidine kinase disconnects a negative regulatory function

YycG is a transmembrane spanning multi-domain histidine kinase, and features domains localized to the cytoplasm as well as the extracytoplasmic space (Fig. 1A). We previously showed that YycG kinase localizes to the division septum and that this localization is dependent on FtsZ (Fukushima et al., 2008). In an FtsZ-depleted strain YycG failed to localize and also failed to activate transcription. Therefore some domain of the YycG protein may contact FtsZ or other proteins of the divisome and this contact is associated with phosphorylation of the YycF response regulator.

Figure 1.

Immunofluorescencent localization of truncated YycG proteins. (A) YycG is a multidomain sensor histidine kinase that features cytoplasmic HAMP (blue), PAS (PAS_CP, pink) and catalytic HisKA and HATPase domains (green), two transmembrane helices (TM1 and TM2, light blue), and a predicted extra-cytoplasmic PAS-like domain (PAS_ECP, red). (B-I) To determine the domains of YycG responsible for septum localization, strains expressing truncated versions of YycG were subjected to immunofluorescence analysis. All strains were grown to an A525nm=0.3, fixed and YycG localization was determined immunologically either with anti-YycG antibody (green) or for myc-tagged constructs with anti-c-myc antibody (red). DNA was stained with DAPI (blue). (B-E) Immunofluorescent pictures of isogenic strains JH25058, JH25060-2, respectively harboring as a single copy in its native locus either (B) full length YycG, (C) YycG_△44-167 deleted for the PAS_ECP domain, (D) YycG_△2-203 truncated for the transmembrane helices and the PAS_ECP domain or (E) YycG_△2-255 truncated for the transmembrane helices, PAS_ECP and the cytoplasmic HAMP domains. For further identification of the YycG domains responsible for septal localization YycG was visualized immunologically with anti-YycG antibody in strains (F) JH25064 and as a negative control (G) JH25033, which feature an IPTG inducible copy of full length YycG expressed ectopically from the amyE locus and either (F) express a C-terminal YycG fragment lacking the catalytic domains from the native yycG locus (JH25064) or (G) were deleted for wild type yycG (JH25033). For these strains, the full-length yycG gene was depleted by exposing these strains in media without IPTG for 3h. Lastly, localization of 3×c-myc tagged constructs of either (H) the catalytic domains of YycG (strain JH25069) or (I) full-length YycG (strain JH25063) was visualized with anti-c-myc antibody in the continuing presence of wild type YycG. The bars indicate 5 μm.

To determine the regulatory role of the individual domains we aimed to replace the wild type copy of yycG with serial truncation mutants, so that the truncated genes represented the only yycG gene copy in these strains. Since YycG is essential for viability this was only possible for YycG truncation constructs that retained sufficient activity to maintain cell viability. For this purpose we modified a previously constructed double cross-over delivery plasmid pJS76 (Szurmant et al., 2008), harboring a DNA fragment with the gene order yycF-yycG-spc^R-P_spac-yycH’ to include various truncated alleles of yycG. Upon transformation of linearized vectors and selecting for spectinomycin resistance, several colonies were screened by colony PCR to identify clones containing the different yycG alleles.

Deletion constructs that could successfully replace the wildtype YycG copy are depicted in Fig. 1C-1E. They either lacked the extra-cytoplasmic PAS-like domain (YycG_△44-167), the extracytoplasmic domain and the transmembrane helices (YycG_△2-203) or the extracytoplasmic domain, the transmembrane helices and the cytoplasmic HAMP domain (YycG_△2-255), respectively. However of more than 40 screened colonies transformed with a construct lacking every domain but the catalytic domains (YycG_△2-373) all transformants retained the wildtype copy of YycG, suggesting that this construct was either unstable or inactive.

Cellular protein levels of the truncated constructs were compared to those of full-length YycG by western blotting utilizing anti-YycG antibody, raised against a cytoplasmic fragment of the kinase. These assays demonstrated that the YycG_△2-203 and YycG_△2-255 constructs were present at much lower levels than intact YycG or YycG_△44-167 (Fig. 2), suggesting a certain robustness of the system in respect to YycG protein levels.

Figure 2.

Characterization of strains harboring genes coding for truncated YycG proteins. (A) Growth and (B) expression of the YycFG dependent yocH-lacZ reporter were assayed in otherwise epigenetic strains JH25058 (wild type YycG, blue diamonds), JH25060 (YycG_△44-167, pink squares), JH25061 (YycG_△2-203, yellow triangles) and JH25062 (YycG_△2-255, purple stars). Phenotypes are compared to the yycHI deletion strain JH25031 (brown circles). A time point of 0h indicates the onset of stationary phase in the wild type strain. (C) The cellular YycG levels in the different strains were visualized immunologically utilizing cell lysate derived from the indicated strains grown to an OD525nm of 0.2 (odd lanes) and 0.8 (even lanes).

The in vivo enzymatic activity of these kinase constructs was assessed with the P_yocH-lacZ reporter construct, which features a promoter under positive regulation by the YycFG system (Howell et al., 2003). Surprisingly, we observed that all strains harboring the various truncation constructs of YycG showed a more than 10 fold increase in yocH expression despite the reduced protein levels found in two of the constructs (Fig. 2). Furthermore, the yocH expression phenotype as well as a growth defect resulting in cell densities less than half of that observed for the wild type strain, were reminiscent of what was previously observed for strains deleted for yycH and/or yycI (Szurmant et al., 2007b). These two genes encode proteins known to reduce the activity of YycG via direct interaction with the transmembrane helices of YycG (Szurmant et al., 2008). The constitutive kinase activity of truncated YycG constructs suggests that the N-terminal domains of YycG serve to link the kinase to negative regulatory processes communicated by YycH and YycI and perhaps other signaling molecules.

Truncated YycG constructs fail to co-immunoprecipitate with the regulatory proteins YycH and YycI

To assess whether the observed lack of negative regulation for all the truncated YycG constructs was due to a failure of these constructs to interact with YycH and/or YycI, we employed an immunoprecipitation assay of cross-linked protein extracts. To this end B. subtilis strains (JH25058 and JH25060-62) harboring either the full-length or the truncated copies of yycG were grown and cross-linked to associated proteins. Detergent solublized, formaldehyde cross-linked whole-cell protein extracts were immunoprecipitated with anti-YycG, anti-YycH or anti-YycI antibody. Precipitates were resuspended and boiled in SDS-sample buffer and subjected to SDS-PAGE. The presence of YycG, and YycI in the precipitate was determined by immunoblotting.

These assays revealed that none of the truncation constructs of YycG retained the ability to interact with YycH or YycI, i.e. YycH and YycI failed to co-precipitate with YycG and vice versa (Fig. 3). But, we observed that YycH and YycI continued to co-precipitate and hence must interact in these strains and similar results were obtained for the YycG depletion strain (JH25033). Thus, YycH-YycI interaction must be independent of their interaction with YycG. Conversely, in a strain deleted for yycI (JH25022), YycH failed to interact with the wild type YycG kinase and in a strain deleted for yycH (JH25021), YycI failed to interact with YycG. This demonstrates that only the complex of YycH and YycI has sufficient affinity for YycG to be detected in immunoprecipitation studies, and thereby explained previous results that showed that yycH, yycI and yycHI deletion strains all have identical phenotypes (Szurmant et al., 2007b; Szurmant et al., 2008). Lastly, in a strain depleted for FtsZ (KP444), all three proteins, YycG, YycH and YycI could be co-immunoprecipitated, suggesting that localization to the septum was not required for YycG-YycH-YycI complex formation to occur.

Figure 3.

YycG interactions with its regulatory proteins YycH and YycI visualized by immunoprecipitation. (A-D) Strains were grown, cross-linked with formaldehyde and the whole cell protein extract was subjected to immunoprecipitation with anti-YycG, anti-YycH or anti-YycI antibody as outlined in Materials and Methods. Precipitates were solublized, subjected to SDS-PAGE and the desired proteins were visualized by western blotting with either anti-YycG (left) or anti-YycI antibody (right). YycG, YycH and YycI interactions were probed in (A) △yycH strain JH25021 (lanes 2) compared to isogenic wild type strain JH25001 (lanes 1), (B) △yycI strain JH25022 (lanes 2) compared to isogenic wildtype strain JH25001 (lanes 1), (C) ftsZ depletion stain KP444 under replete (lanes 1) and depleted (lanes 2) condition and (D) yycG depletion strain JH25033 under replete (lanes 1) and depleted (lanes 2) conditions. (E) To determine whether YycG and YycI retained the ability to interact in strains JH25060-2, which express the indicated truncated YycG constructs, YycI was visualized in immunoprecipitates with anti-YycG, anti-YycH and anti YycI antibody, and co-immunoprecipitation patterns were compared to the isogenic strain JH25058 that expresses full-length YycG.

In conclusion this data suggests that the extracytoplasmic and transmembrane domains of YycG, along with YycH and YycI proteins exist as a complex with reduced YycG kinase activity, and that deletion of any of these domains or proteins results in a hyperactive / deregulated YycG kinase.

YycH and YycI do not localize to the septum

The above immunoprecipitation data along with previously published results demonstrated that YycG, YycH and YycI interact to form a ternary complex with reduced kinase activity. Since YycG is activated at the septum, it could be envisioned that this activation of YycG is either achieved by physical separation from its auxiliary proteins YycH and YycI or alternatively that the ternary complex of all three proteins travels to the septum and serves as a signal receptor in its entirety.

To distinguish between the two possibilities the cellular localization of YycH and YycI was first assessed individually in exponentially growing cells and compared to YycG-localization by immuno-fluorescence utilizing purified rabbit-anti-YycH, anti-YycI or anti-YycG antibodies (Fig. 4A-C)¹. From these images it was observed that—in contrast to YycG—YycH or YycI do not localize to the division septum. Instead, both, YycH and YycI show a spotty localization pattern in the vicinity of the cell membrane, consistent with their transmembrane architecture. As a control, the yycH and yycI deletion strains were assayed with the respective antibodies in the same way, and no protein could be visualized. The observed YycH and YycI localization patterns were not due to detection of cross-reacting material (Fig. 4B and C).

Figure 4.

YycH and YycI do not localize to the septa. To assess cellular localization of YycH and YycI, wildtype strain JH642 or △yycH and △yycI strains were grown exponentially in SM media to an OD₅₂₅ = 0.3, and localization of (A) YycG, (B) YycH and (C) YycI were assessed immunologically utilizing the respective purified rabbit antisera as primary antibodies and Alexa-Fluor 488 labeled secondary antibody (green). For (B) and (C) the upper panels show the wildtype strain and the lower ones the respective deletions strains. (D) To assess localization of YycG and YycH in the same strain, YycH was expressed with a 6×c-myc tag. YycG localization (green) was assessed as in (A), YycH localization (magenta) with mouse anti-c-myc primary, Alexa Fluor 568 labeled goat anti-mouse secondary and Alexa Fluor 568 labeled donkey anti-goat tertiary antibody. DNA was stained with DAPI (blue). The bars indicate 5 μm.

Since all the individual antibodies against the three proteins were raised in rabbits, co-localization of the proteins could not be assayed in the same strain. To circumvent this problem we generated a 6×c-myc-tagged expression plasmids for YycH and a 6×T7-tagged expression plasmid for YycI. The 6×T7-tagged-YycI construct failed to complement the yycI deletion strain, and was not further investigated. The 6×c-myc-tagged YycH construct fully complemented the yycH deletion strain, both in respect to growth and expression of the YycF-dependent yocH-lacZ reporter construct (not shown). Western-blotting with anti-YycH-antibody demonstrated wild type expression levels and when assayed with anti-c-myc-tag-antibody a single intense band could be observed (not shown). Using this strain the cellular localization of both, YycG and 6×c-myc-tagged YycH could be assayed in the same strain. The spotty localization of YycH along the vicinity of the membrane was again observed. YycG in contrast localized mainly to the division septa (Fig. 4D). We conclude that interaction of YycG with YycH and YycI must be transient and that YycG, YycH and YycI are not complexed all the time. The data is most consistent with a model where localization of YycG to the septum results in its physical separation from YycH and YycI.

A cytoplasmic domain of YycG is sufficient for septum localization

In order to determine the YycG domain(s) required for the protein to accumulate at the division septum, the strains JH25060-JH25062, harboring the truncation constructs were subjected to immunofluorescence analysis utilizing purified anti-YycG antibody. All truncated YycG constructs including the shortest one that featured only the cytoplasmic PAS and the two catalytic HisKA and HATPase domains retained the ability to localize to the septum (Fig. 1C-E). YycG localization patterns looked identical to those obtained for the isogenic strain JH25058 expressing a full-length yycG copy (Fig. 1B). We conclude that one of the remaining three cytoplasmic domains (PAS, HisKA or HATPase) had to be responsible for septum localization of YycG.

To further define which of the remaining three cytoplasmic domains of YycG is responsible for FtsZ-dependent septum localization, additional strains expressing the following truncated YycG proteins were constructed: (a) the PAS domain alone, (b) the catalytic core comprised of the HisKA and HATPase domains alone, and (c) a C-terminal truncation devoid of the catalytic domains but encompassing the transmembrane helices, the extra-cytoplasmic domain the HAMP and the PAS domain. Since none of these constructs could complement for the full-length yycG copy, these constructs were expressed as 3×c-myc-tagged versions in the continued presence of the full-length kinase, or they were expressed in a strain in which the wild type yycG copy could be depleted.

The PAS domain construct proved completely unstable and could not be detected in our immunofluorescence assays, when expressed either ectopically from the thrC locus (strain JH25067) or from the multi-copy vector pHT315S (strain JH25068).

The catalytic construct (Fig. 1H) proved unstable as well, however it could be detected with anti-c-myc-antibody when expressed from the multi-copy vector pHT315S (strain JH25069). When expressed from this vector, the catalytic construct proved toxic to wild type strain JH642, and transformants could only be obtained for a strain harboring a chromosomal lacI copy, which rendered the pHT315S P_spac promoter IPTG-inducible. Transformants of this strain (JH25069) grew stably up to an IPTG concentration of 0.2 mM. At higher IPTG concentrations chains of cells with some having lost their cellular content (ghost cells) could be observed. These phenotypes are identical to those described for a yycFG depleted strain and we speculate that the catalytic fragment of YycG inhibits YycF phosphorylation, perhaps by acting as a phosphatase or by sequestering YycF. This is consistent with observations that a cytoplasmic fragment of the S. pneumoniae YycG ortholog exhibits phosphatase activity in vitro (Gutu et al., 2010).

The catalytic domain fragment failed to localize to the septum (Fig. 1H), whereas localization of a full-length 3×c-myc-tagged YycG copy could be observed (Fig. 1I). Furthermore, the C-terminally truncated copy that lacked the catalytic domains but included the PAS domain retained the ability to localize to the septum as assayed utilizing the anti-YycG antibody (Fig. 1F) in a strain where the full-length copy could be depleted (JH25064), No fluorescence signal could be observed in the YycG-depleted control strain (JH25033)(Fig. 1G). These experiments strongly suggest that the cytoplasmic PAS domain is a determinant for directing YycG to the septum.

Two-hybrid analysis suggests YycG interacts with the latter stage cell division proteins DivIB, Pbp2B and FtsL

Previously we showed that FtsZ and YycG proteins could be co-immunoprecipitated from a formaldehyde cross-linked cell extract with anti-YycG antibody, suggesting either they directly interact or they are part of a larger protein complex (Fukushima et al., 2008). In vitro interaction studies with a soluble cytoplasmic YycG construct and FtsZ did not reveal an interaction, implying that the proteins may not interact directly but rather may share common interaction partners. To gain a qualitative understanding of the proteins that do direct YycG to the septum and/or regulate its activity, we employed the bacterial two-hybrid system developed by Karimova et al. (Karimova et al., 1998; Karimova et al., 2000), which was previously utilized to demonstrate interactions between YycG with its regulatory proteins YycH and YycI (Szurmant et al., 2007b) and to demonstrate interactions among cell division proteins (Daniel et al., 2006).

Full-length YycG or a truncated YycG_△211-611 that lacks all cytoplasmic domains were co-expressed with known cell division proteins FtsZ, FtsA, FtsW, DivIB, DivIC, FtsL, Pbp2B, EzrA, ZapA, SepF and MinJ as C-terminal fusions to the individual domains (T18 and T25) of the two-domain Bordatella pertussis adenylate cyclase protein. In these assays interaction of two proteins results in reconstitution of adenylate cyclase activity and in turn expression of β- galactosidase (Karimova et al., 2000). Possible interactions were assayed by co-transforming the adenylate cyclase deficient Escherichia coli strain BHT101 with both plasmids harboring T18 and T25 fusions. Co-transformation reactions were directly spotted on plates containing the β- galactosidase substrate 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside. This is preferable over more quantitative liquid cultures since many of the fusion constructs used here proved toxic to the E. coli cell leading to variable results in liquid media from pressure to lose the toxic plasmids or accumulate mutations in them. To assure specificity of any observed interactions, another B. subtilis kinase, ResE was also probed for interaction with the cell division proteins. The ResE kinase is the closest B. subtilis homolog to YycG with 30 (51) % identical (similar) residues over the full length of the protein and features an identical domain architecture.

When co-expressing all possible combinations of the above mentioned fusion proteins (Fig. 5) six observations can be made. (1) YycG, YycG_△211-611 and ResE homodimerized as expected for sensor kinases. (2) Surprisingly ResE and YycG heterodimerized in one but not the other vector combination. We assume this to be an artifact. (3) YycG and YycG_△211-611 but not ResE interact strongly with FtsL, DivIB and Pbp2B in both vector combinations. (4) Co-expression of MinJ or FtsW with either ResE, YycG or YycG_△211-611 in one but not the other vector combination appears toxic (slow growth) but some color is produced, suggesting that interaction does occur. We assume this to be a non-specific artifact since it occurs with all three kinase constructs and does not occur in the opposite vector combination. (5) Full length YycG but not YycG_△211-611 interacts with FtsW in one vector combination. (6) YycG and YycG_△211-611 interact with EzrA in one but not the other vector combination.

Figure 5.

Interactions between full-length YycG or YycG_△211-611 (YycG*) and cell division proteins probed by the bacterial two-hybrid assay. Adenylate cyclase deficient E. coli strain BTH101 was co-transformed with plasmids expressing YycG as a C-terminal fusion to the adenylate cyclase domain T18 (pUT18C) or T25 (pKT25) and with plasmids expressing the indicated cell division proteins as fusions with the other adenylate cyclase domain (i.e. T18/T25 or T25/T18). Co-transformations were spotted on appropriate indicator plates where the development of a blue color indicates an interaction between the respective proteins (see Methods). For comparison purpose, fusions of the YycG homolog ResE were also probed for their interaction with the cell division proteins.

The data suggests that YycG makes specific interactions with FtsL, DivIB, Pbp2B and interactions with FtsW and EzrA appear also possible. Since all of these interactions except the FtsW interaction are also observed with the YycG_△211-611 fragment, these interactions likely occur via the transmembrane and extra-cytoplasmic domains of YycG. We can’t dismiss the possibility that some of these interactions are indirect, mediated through a common E. coli divisome interaction partner.

Deletion or depletion of later stage cell division proteins does not perturb YycG localization

The two-hybrid assay identified strong interactions between YycG and the cell division proteins DivIB, FtsL and Pbp2B and a possible interaction with FtsW. We previously showed that YycG fails to localize in a strain depleted for FtsZ (Fukushima et al., 2008). To assess whether any other individual proteins were required for YycG localization to the septum and thereby prove the physiological relevance of the two-hybrid results we wanted to probe YycG localization in strains that were either deleted for non-essential cell division proteins or that harbored essential cell division genes under control of the IPTG inducible P_spac or the xylose inducible P_xyl promoter to allow for depletion of the proteins upon removal of the inducer.

Strains tested were (a) a temperature sensitive divIC strain (JH25070), (b) △divIB (JH25072), (c) P_spac-ftsA/△sepF (YK200), (d) P_spac-ftsW (FtsW_P), (e) P_spac-pbpB/P_xyl-ftsL (804) (f) P_spac- pbpB/P_xyl-ftsL/△divIB (JH25073) (for exact genotypes and source of strains see table I). The strains were grown in sporulation medium under conditions that have been reported to induce filamentation and growth cessation. Specifically, filamentation and growth cessation was observed for the divIC and △divIB strains when shifting the growth temperature from 30 to 37°C for 180 min incubation or from 37 to 48°C for 90 min, respectively. The same was observed when growing the 804 strain in the absence of xylose for 180 min to deplete FtsL and when growing the P_spac-ftsW strain in the absence of IPTG. Growth cessation was not observed for the YK200 strain or the 804 strain even after extended periods (up to 360 min) without IPTG, suggesting that under these conditions, ftsA and pbpB transcripts did not deplete. This was confirmed by quantitative RT-PCR that demonstrated equal levels of transcript for these genes upon IPTG replete and deplete conditions (not shown).

Table I.

Strains used in this study

Strain	Genotype	Source or reference
B. subtilis
168	trpC2	D. Ehrlich
JH642	trpC2 pheA1	Laboratory stock
JH25001	trpC2 pheA1amyE::(PyocH-lacZ aph3-A)	(Szurmant et al., 2005)
JH25021	JH25001 ΔyycH1	(Szurmant et al., 2007b)
JH25022	JH25001 ΔyycI1	(Szurmant et al., 2007b)
JH25031	JH25001 ΔyycHI101	(Szurmant et al., 2007b)
JH25032	trpC2 pheA1 amyE::(Pspac-yycG lacI aph3-A)	(Fukushima et al., 2008)
JH25033	JH25032 yycG::cat PyycF-yycHIJ	(Fukushima et al., 2008)
JH25058	JH25001 yycG-spcR-Pspac-yycHIJ	(Szurmant et al., 2008)
JH25060	JH25001 yycGΔ44-167-spcR-Pspac-yycHIJ	pJS78 → JH25001
JH25061	JH25001 yycGΔ2-203-spcR-Pspac-yycHIJ	pJS79 → JH25001
JH25062	JH25001 yycGΔ2-255-spcR-Pspac-yycHIJ	pJS80 → JH25001
JH25063	JH25032 yycG::(yycG-3×myc cat)	pJTF128 → JH25032
JH25064	JH25032 yycG::(yycGΔ374-611cat)	pJTF125 → JH25032
JH25065	trpC2 pheA1 amyE::(Pspac-lacZ lacI aph3-A)	pJM119 → JH642
JH25066	trpC2 pheA1 thrC::(PyycF-3×myc-yycGΔ2-254, Δ374-611spcR)	pJTF122 → JH642
JH25067	JH25033 thrC::(PyycF-3×myc-yycGΔ2-254, Δ374-611 spcR)	pJTF122 → JH25033
JH25068	trpC2 pheA1 pJTF126 (pHT315-Pspac-3×myc-yycGΔ2-254Δ374-611)	pJTF126 → JH25065
JH25069	JH25065 pJTF127 (pHT315-Pspac-3×myc-yycGΔ2-377)	pJTF127 → JH25065
KP444	ftsZ::phleo Pspac-ftsZ	(Beall and Lutkenhaus, 1991)
FtsZp	trpC2 ftsZ::phleo Pspac-ftsZ	KP444 → 168
MDS193	divIC’-spcR–‘divIC (temperature sensitive)	Kit Pogliano
JH25070	trpC2 divIC’-spcR-‘divIC (temperature sensitive)	MDS193 → 168
SU321	trpC2 divIB::cat	(Katis and Wake, 1999)
JH25071	trpC2 divIB::cat	SU321 → 168
JH25072	trpC2 divIB::tetR	pCM::tet → JH25066
804	trpC2 ftsL::pSG441 aphA-3 Pspac-pbpB cat Pxyl-ftsL	(Daniel et al., 1998)
JH25073	trpC2 ftsL::pSG441 aphA-3 Pspac-pbpB cat Pxyl-ftsL divIB::tetR	JH25067 → 804
YK200	trpC2 ftsA::cat aprE::(Pspac-ftsA kanR) sepF::spcR	(Ishikawa et al., 2006)
FtsWp	trpC2 ftsW::km Pspac-ftsW	Daniel, R.A., unpublished
JH25077	JH25021 pJTF131 (pHT315- PyycF-6×c-myc-yycH)	pJTF131 → JH25021
JH25078	JH25022 pJTF133 (pHT315- PyycF-6×T7-yycI)	pJTF133 → JH25022
E. coli
JM10	recA1 Δ(lac-proAB) endA1 gyrA96 thi-1 hsdR17 relA1 supE44 [F’ traD36 proAB+lacIqlacZ ΔM15	Laboratory stock
DH5a	recA1 Δ(lacU169 [Φ80dlacZΔM15]) endA1 gyrA96 thi-1 hsdR17 relA1 supE44	Laboratory stock
C600	supE44 hsrR17 thi-1 thr-1 leuB6 lacY1 tonA21	Laboratory stock
BTH101	F− cya-99 araD139 galE15 galK16 rpsL1 (StrR) hsdR2 mcrA1 mcrB1	(Karimova et al., 1998)

Cellular localization of YycG was probed in the above strains under conditions that did result in growth cessation (with the exception of strain YK200 and 804 without IPTG). In all strains, YycG retained the ability to co-localize with FtsZ at the septum under the above-described conditions (Fig. 6A-H). From these experiments we concluded that DivIB, and SepF are dispensable for YycG septum localization. The results suggest that DivIC, FtsL, and FtsW are dispensable for septum localization although it is in principal possible that residual amounts of protein were still present in the respective conditional or depletion strains under deplete conditions. Since FtsA and Pbp2B failed to deplete in the strains and conditions used here, we cannot exclude the possibility that these genes might be required for septum localization. This however seems unlikely since FtsA failed to interact with YycG in the two-hybrid assay and since the cellular localization of Pbp2B to the membrane and extracytoplasmic space does not coincide with the cytoplasmic fragment of YycG that retained the ability to localize to the septum.

Figure 6.

Co-localization of YycG and FtsZ in divisome depletion strains. All Strains were grown in SM media to an OD525nm=0.2 at which point growth cessation and filamentation was initiated by shifting the growth temperature from 30°C to 37°C (divIC(TS) strain JH25070, panel (A)), from 37°C to 48°C (△divIB strain JH25072, panel (F)) or by washing the cells and incubation in fresh media without xylose (to initiate ftsL depletion in strain 804, panel (E)), without IPTG (to initiate ftsW depletion in strain FtsW_P (B), ftsA depletion in strain YK200 (C) or pbpB depletion in strain 804 (D)). The triple mutant strain JH25073 was incubated in the absence of xylose and IPTG at 37°C and 48°C (panels G and H). Growth cessation was observed for all strain except strains YK200 and strain 804 which continued to grow even in extended absence of IPTG suggesting that ftsA and pbpB did not actually deplete in these strains. YycG (green) and FtsZ (red) localization were visualized immunologically. DNA was visualized with DAPI (blue). The bars indicate 5 μm.

In summary, the only protein whose depletion is known to result in YycG delocalization is FtsZ. This is unlikely to be due to a direct interaction between these two proteins since YycG and FtsZ failed to interact either in vitro or in the two-hybrid assay. We conclude that septum localization of YycG is likely dependent upon a constituent(s) of the divisome but perhaps none of the proteins tested here.

RT-PCR results suggest that signaling involves later stage cell division proteins

We previously demonstrated that in a strain KP444 depleted for FtsZ, the YycG kinase fails to localize to the septum, and that YycF-P dependent transcription is altered in a way similar to what is observed for a YycFG depleted strain (Fukushima et al., 2008). Hence, YycG activation seems to occur at the septum, but whether this is due to its activation signal being produced or accumulated there or because YycG activation requires the contact with the divisome protein complex remained unknown. Since YycG and FtsZ did not appear to directly interact, we wanted to determine whether one of the other members of the divisome, which were not required for YycG localization were still required for YycG activation at the septum, in particular the ones that showed interaction in the two-hybrid assay.

We employed a semi-quantitative RT-PCR assay to assess the effect of the various cell division mutants on YycFG-dependent gene expression of the positively regulated autolysin coding yocH gene, and the negatively regulated peptidoglycan deacetylase coding yjeA gene, which was shown to protect the cell wall from turnover (Fig. 7). As an internal control, yycG transcript level, which is not auto-regulated, was also assayed. As observed previously (Fukushima et al., 2008), when strain JH25033 was depleted for the yycFG operon, yocH expression was down-regulated and yjeA expression was up-regulated. The same was observed for strain FtsZ_P when depleted for ftsZ.

Figure 7.

YycFG-dependent gene-expression in cell division mutant strains. (A) YycF-P initiates expression of yocH and inhibits expression of yjeA. Transcript levels of yocH and yjeA genes and as a control of yycG were assayed in the indicated strains by RT-PCR. RNA was isolated and transcribed into cDNA (see Methods). All strains were grown in SM to an OD525nm=0.2 and depletion of desired genes was initiated by incubating the respective strains for three hours in the presence or absence of IPTG, xylose, and by shifting the growth temperature. For ftsA and pbpB, depletion was not observed in the absence of IPTG. As a control, transcript levels were assayed in wildtype strain 168. 168 genomic DNA was utilized as template to assure the correct transcript size. (B-C) To further define responsibility for the effects on YycFG-dependent transcription observed in the triple mutant strain JH25073, effects of pbpB, ftsL and divIB mutations were assayed individually. (B) Strain 804 was incubated for 180 min in the absence of IPTG, in the absence of Xylose or in the absence of both. RNA was purified from these strains and transcript levels for yocH, yycG, ftsL and pbpB, were assessed and compared to wild type strain 168. (C) yocH and yycG transcripts were also assayed in the △divIB strain JH25072 upon shifting the growth temperature from 37 to 48°C for 180 min. Transcript levels were compared to those in strain 168 grown under the same conditions.

Expanding the assay to the other cell division mutant strains used in this study, we observed a similar phenotype for the ftsW depleted strain FtsW_P and the triple mutant strain JH25073 harboring △divIB/P_spac-pbpB/P_xyl-ftsL genotypes. In contrast no effect on yocH and yjeA was observed for the temperature sensitive divIC strain JH25070 and strain YK200 harboring P_spac-ftsA and △sepF mutations (Fig. 7A).

Under our growth and depletion conditions ftsA and pbpB constructs did not deplete (not shown). All other strains did filament, and depletion could thus be suspected. Since pbpB did not deplete in strain JH25073, only FtsL depletion or DivIB deletion could be envisioned as the culprit of the YycG missregulation phenotype. To further define whether FtsL, DivIB or both were needed to be depleted/deleted in order to observe the YycFG-dependent missregulation phenotype in strain JH25073 we assayed yocH expression in the divIB strain JH25072 and in strain 804 which harbors the P_xyl-ftsL P_spac-pbpB inducible genes. As for the triple mutant strain, we found that pbpB did not deplete even 180 min after IPTG removal. A reduction of ftsL transcript could clearly be observed (Fig. 7B). Nevertheless, yocH expression did not change with respect to the same strain under replete conditions, suggesting that depletion of ftsL alone is not sufficient to alter YycG activity. Similarly, yocH expression did not change in the divIB deletion strain (Fig. 7C). Therefore both, DivIB and FtsL have to be reduced/absent from the septum to prevent YycG kinase activation.

Since in all newly assayed strains, YycG localized to the septum, we can conclude that localization of YycG is not sufficient for its activation. Rather several later stage members of the divisome have to be present for activation to occur.

Discussion

It now seems clear that the YycFG two-component system serves a regulatory link connecting cell division with aspects of cell wall homeostasis (Dubrac et al., 2008; Fukushima et al., 2008). The goal of this study was to discover the regulators of YycG kinase activity and to unravel molecular details responsible for output control in response to the cell division state. In particular we aimed to identify the domains of YycG that are required for septum localization and to discover the cell division proteins that are involved in septum localization and YycG activation. Additionally we wanted to determine the mechanism by which YycH and YycI proteins act in the regulation of the kinase activity of YycG.

In vivo analysis of truncated YycG derivatives provided an initial surprise in the finding that neither the transmembrane nor extracellular domains of YycG were required for localization and that a cytoplasmic domain of the protein was sufficient for septum localization. These studies strongly suggested that the cytoplasmic PAS domain alone can direct YycG to the septum. The experiments did not rule out the existence of other determinants. PAS domains have been identified in more than 30% of all sensor kinases (Szurmant et al., 2007c) and are widespread signaling modules utilized in organisms ranging from bacteria to humans (reviewed in (Moglich et al., 2009)). In sensor kinases only a minute fraction of PAS domains has an identified function in small molecule binding. It has been a long-standing question whether this is the ubiquitous function for all these domains, or whether other functions remain to be identified. Our data suggests that the YycG PAS domain serves as a cell-sorting module required to direct YycG to the septum. PAS domains were shown to be required for localization to specific cellular compartments for a few other histidine kinases (Angelastro et al., 2010; Boyd, 2000; Hallez et al., 2007). This suggests a role in sub-cellular localization as a common function for the PAS domain.

While septum localization of YycG was FtsZ-dependent, a direct interaction between these two proteins could not be identified. A soluble construct of YycG failed to interact with FtsZ in vitro (Fukushima et al., 2008) and our two-hybrid study failed to identify interactions between the two proteins in vivo. The two-hybrid analyses did single out several divisome proteins as possible YycG interaction partners, namely DivIB, FtsL,Pbp2B and perhaps FtsW. FtsL is a short coiled-coil transmembrane spanning proteins and contains a limited N-terminal cytoplasmic and more extensive C-terminal extra-cytoplasmic portion. DivIB and Pbp2B are larger transmembrane spanning proteins with extensive extra-cytoplasmic domains (reviewed in (Errington et al., 2003; Goehring and Beckwith, 2005)). FtsW is an integral membrane protein with 10 transmembrane helices. Of the four proteins, only Pbp2B has a known catalytic activity (transpeptidase), whereas FtsL, DivIB and FtsW are likely scaffolding proteins required for the integrity of the divisome complex, which serves to recruit other proteins, to the septum (Errington et al., 2003; Goehring and Beckwith, 2005). A large body of literature on the B. subtilis proteins and their orthologs in other bacteria (DivIB orthologs are named FtsQ in Gram-negative bacteria) has established multiple interactions, an interdependence for protein stability and septum location, which appear to differ slightly from one organism to another (reviewed in (Errington et al., 2003; Goehring and Beckwith, 2005)). Different roles are also suggestive based on the drastic difference in protein copy numbers determined for some of these proteins in E. coli (25-50 copies FtsQ, 20- 200 copies of FtsL), and B. subtilis (5000-13000 copies DivIB,) (Robichon et al., 2008; Wadsworth et al., 2008), perhaps owing to the difference in cell wall architecture between Gram-positives and Gram negatives.

Our immunofluorescent experiments in strains that were depleted for FtsL, deleted for DivIB, or depleted for FtsW demonstrated that no single one of these interacting proteins was required for YycG localization (Fig. 6). Since the bulk of these proteins is in the membrane or the extra-cytoplasmic space, these results are consistent with our observations that cytoplasmic YycG fragments retained the ability to localize to the septum. Nevertheless interaction with the relatively short cytoplasmic portions of the above proteins could have been envisioned as the culprit of localization of the cytoplasmic YycG fragments. In particular the observed interactions with FtsW in the two-hybrid system only occurred with the full-length YycG protein, but not with a protein truncated for its cytoplasmic domains. The continued ability of YycG to localize to the septum in depletion and deletion strains of the interacting proteins could have two reasons. (1) The numerous interactions made by full-length YycG might give some robustness to YycG localization and delocalization would therefore only be observed in strains where all these interactions are perturbed. To further explore this possibility it will be necessary to study localization of YycG domain fragments in multiple cell division mutant strains. (2) Alternatively the possibility remains that an unidentified protein that localizes to the septum early and features an extensive cytoplasmic domain is responsible for interaction with the PAS domain and YycG septum localization. The two-hybrid analysis ruled out most candidates with these features, namely FtsZ, FtsA, ZapA, SepF and probably MinJ. The assay was inconclusive regarding an interaction with EzrA, where interactions were observed in one but not the other vector combination. EzrA, a membrane-anchored cytoplasmic protein that directly interacts with FtsZ and alters its assembly dynamics (Haeusser et al., 2007; Singh et al., 2007) thus remains as a possible candidate and future experiments will aim at elucidating its potential involvement in the localization and signaling process.

Since localization was not perturbed in the strains that were deleted or depleted for the above proteins a different explanation for the observed interactions between YycG with FtsW, DivIB and FtsL was that they serve to regulate YycG activity. To assess this we assayed YycG activity in the various divisome protein depletion and deletion strains. The RT-PCR results clearly demonstrated that YycG failed to activate in a strain simultaneously deleted for DivIB and depleted for FtsL, and in a strain depleted for FtsW; this result is similar to what could be observed for an FtsZ or a YycFG-depleted strain. Since YycG was found at the septum in some of these strains, we can conclude that septum localization alone is not sufficient for YycG activation and that the identified interactions with the divisome proteins may serve a regulatory rather than a localization function. Elucidation of the exact protein (if there is a single one) for YycG miss-regulation in these strains was complicated by the fact that the individual cell division proteins are known to affect each others abundance (Daniel et al., 2006; Robichon et al., 2008; Wadenpohl and Bramkamp, 2010) and localization to the septum (Daniel et al., 2006; Errington et al., 2003). The conclusion that can be drawn from the results of the interaction and activity experiments is that YycG signaling is relatively robust due to multiple interactions with divisome proteins and only if the divisome complex is significantly perturbed does one observe a phenotype. An elegant in vivo interaction study established that interactions between DivIC and FtsL as well as between DivIB and Pbp2B are strong, suggesting that these protein pairs form sub-complexes, whereas other existing interactions between these proteins might be more moderate (Robichon et al., 2008). In the light of this finding it is interesting that a strain with a member of each of these subcomplexes absent (i.e. DivIB and FtsL) showed a missregulation phenotype, whereas the individual mutants did not. It appears that the multiple interactions between these proteins and YycG are no accident. Rather they hint that YycG responds to a correctly assembled divisome complex.

Previously identified interaction partners of YycG were YycH and YycI, which were shown to serve as negative regulators of YycG activity (Szurmant et al., 2007b; Szurmant et al., 2008). Localization studies on the latter two proteins revealed that these proteins do not travel with YycG to the septum, suggesting that they inhibit YycG when not at the division septum. Instead both proteins were observed in a spotty pattern around the cell membrane. This kind of pattern is often observed for membrane-associated proteins when visualized by immuno-fluorescence and is likely due to loss of membrane regions during fixation [e.g., (Potluri et al., 2010)]. A YycG protein that lacks its N-terminal domains or strains that lack YycH or YycI result in a similarly deregulated, overactive YycG kinase suggesting that the complex of these domains serves to hold the kinase activity in check.

Of note, the S. pneumoniae YycG ortholog does not localize to the division septum according to a recent study (Wayne et al., 2010). This is perhaps not surprising since in the Streptococci the YycG kinase lacks an extracytoplasmic domain and no YycH and YycI orthologs can be identified (Ng and Winkler, 2004). The YycG kinase could thus not be regulated in the same manner as observed here for B. subtilis and likely for all other organisms that feature the YycH and YycI proteins, namely the Bacilli, the Listeria, the Enterococci, the Staphylococci and some Clostridia.

The above data in its entirety is suggestive of a model for YycG activation that involves alternative binding partners to either generate an active or an inactive kinase (Fig. 8). According to this model YycG adopts a state of low kinase activity when in complex with YycH and YycI away from the septum. During cell division, a yet unknown protein directs YycG to the septum, likely by interacting with its cytoplasmic PAS domain. At the septum, YycH and YycI are probably displaced by proteins of the divisome complex, through direct contact of YycG with DivIB, FtsL, Pbp2B and FtsW. Within the divisome context YycG is now active as a kinase. Thus YycG acts to sense the division state of the cell. The known regulon of the YycFG system primarily includes cell wall remodeling enzymes under positive regulation and their inhibitors under negative regulation. It seems logical that under active growth conditions cell wall remodeling is constantly required whereas under non-dividing conditions, cell wall remodeling should be inhibited. Proteins functioning in both states are controlled in a reciprocal manner by the phosphorylation level of YycF and are reflective of the interactions of YycG. The current study demonstrates the complexity of an essential aspect of the bacterial division cycle in B. subtilis and other Gram-positive microorganisms, which connects the essential processes of cell division and cell wall remodeling.

Figure 8.

Molecular model for YycG activity regulation. The present data in conjunction with previous results suggests that YycG (G, colors as in Fig. 1) is inactive when complexed with YycH and YycI (H+I,yellow) and this occurs away from the septum. During active growth, YycG primarily localizes to the septum via interaction between its PAS domain with a yet unidentified cytoplasmic protein (?, white). At the septum YycG complexes with DivIB (B), Pbp2B (P), FtsL (L) and FtsW (W) instead of YycH and YycI. In the context of these proteins, the YycG kinase is activated. Therefore, YycG is primarily in the active form during rapid growth when new septa are constantly formed and in the inactive YycH/YycI-complexed state under non-growing conditions when no new septa are formed. The YycFG system induces expression of genes coding for autolysins required for cell wall restructuring during active growth and represses genes coding for autolysin inhibitors, required during non-growing conditions, to maintain the integrity of the cell wall.

Experimental Procedures

Growth media and conditions

All Escherichia coli strains were grown in lysogeny broth (LB) and all B. subtilis strains were grown in Schaeffers Sporulation Medium (SM), in the presence of appropriate antibiotics whenever necessary. Antibiotic concentrations were 100 μg/ml ampicillin or 30 μg/ml kanamycin for E. coli and 2 μg/ml of kanamycin, 5 μg/ml of chloramphenicol, 100 μg/ml spectinomycin, 5 μg/ml erythromcin/25 μg/ml lincomycin or 1 μg/ml of phleomycin for B. subtilis.

Strain and plasmid construction

E. coli strains used were DH5α and JM109 for cloning and propagation purposes and C600 to prepare plasmids for B. subtilis transformation. For bacterial-two-hybrid assays the adenylate cyclase deficient strain BTH101 was utilized. All B. subtilis strains with mutations in yycG, yycH and yycI genes were derivatives of strain JH642 and all cell division mutants utilized here were in B. subtilis 168 genetic background. Strains, plasmids and oligonucleotides are listed in tables I-III.

Table III.

Oligonucleotides used in this study

Name	Sequence	Plasmid constructed
ON85	5′-GCTAGTATGGAAGATGTCTTTAATC-3′	pJS78-81
ON86	5′-GCTGATCAGCGACTTCTCTACCTG-3′	pJS78
ON87	5′-ATTTTTCATATGAGAACCATTACCCTC-3′	pJS79
ON88	5′-CTTGAACATATGCAGGCGATGACTGAAGG-3′	pJS80
ON89	5′-CAAGAGCATATGGATCAGGAACGCAGAG-3′	pJS81
ON90	5′-GCGGGATCCAGTCCCGTCCGGACCGCC-3′	pJTF116
ON91	5′-TTTTTATCCATATGTTTGATTTTCCTCCTGCCG-3′	pJTF116
ON92	5′-GGAATTCCATATGGGCCGCTCTGAGCAAAAG-3′	pJTF116
ON93	5′-GCGCTCTAGAGCTACTATTAAGATCCTCCTC-3′	pJTF116
ON94	5′-GCCGTCTAGACAGGCGATGACTGAAGGAGA-3′	pJTF119
ON95	5′-GCGCCCGCGGAGATCTTTACATTTTCTCTTGTTCGGTTACATC-3′	pJTF119
ON96	5′-GCCGTCTAGAATGGATCAGGAACGCAGAGAA-3′	pJTF121
ON97	5′-GCGCCCGCGGATCCTCACGCTTCATCCCAATCATC-3′	pJTF121
ON98	5′-GAATGAATTCAGATTGTCCGGTTTATG-3′	pJTF128
ON99	5′-TTATACTCGAGCGCTTCATCCCAATCATCC-3′	pJTF128
ON100	5′-GCCGTCTAGAATGAAGCGTGAAAATATAAAAACG-3′	pJTF117
ON101	5′-GCGCCCGCGGATCCTTATTCCACTCCACTGTTAGCC-3′	pJTF117
ON102	5′-GCCGGCTAGCGAGCAAAAGCTGATCAGCG-3′	pJTF131
ON103	5′-CACGCTTCATTCTAGAGCTAC-3′	pJTF131
ON104	5′-GCCGTCTAGAGTGGAGTGGAATAAGACAAAATC-3′	pJTF118
ON105	5′-GCGCCCGCGGATCCTCATTGATCTGTATCTAAAATCGTA-3′	pJTF118
ON106	5′-CACCATCACCATATGGCTAG-3′	pJTF132
ON107	5′-CAAGCTGGAGATCTTGAACCC-3′	pJTF132
ON108	5′-CATCACCACCATAGATCTATGGCTAGCATG-3′	pJTF132
ON109	5′-CAAGCTGGGGATCTAGAACCC-3′	pJTF132

YycG truncation strains

To construct strains harboring truncated yycG alleles in their native locus, we employed plasmid pJS76, which links the yycG gene to a downstream spc^R marker and places the downstream yycHIJ genes under the P_spac promoter to avoid polarity effects (Szurmant et al., 2008). Plasmid pJS76 was subjected to PCR with inverted primers (ON85-89) followed either by blunt end ligation resulting in pJS78 or NdeI restriction and self-ligation resulting in plasmids pJS79-81.

These vectors harbor yycG alleles deleted for amino acids 44-167, 2-203, 2-255 or 2-373, respectively. These four plasmids were linearized by AatII digest and transformed into strain JH25001 selecting for Spc^R. Spc^R colonies were screened by PCR for correct double-crossover replacement of the wildtype yycG gene. Replacement of the wildtype gene was observed for roughly half the colonies transformed with plasmids pJS78-pJS80, and positive clones were renamed as strains JH25060-JH25062. Transformants of plasmid pJS81 all retained the full-length yycG copy, suggesting that the YycG_△2-373 protein fragment did not retain sufficient activity or stability to afford viability.

To express a gene coding for the cytoplasmic PAS domain of YycG individually, the following strains were constructed, that either expressed the domain in single copy ectopically from the thrC locus under its native promoter or from the multicopy plasmid pHT315S under the P_spac promoter. First, the 3×c-myc-tag coding sequence was amplified by PCR with primers ON92 and ON93 from vector pKL94 introducing 5′ NdeI and 3′ XbaI sites. In a second PCR, the P_yycF promoter was amplified with primers ON90 and ON91 introducing 5′ BamHI and 3′ NdeI sites. Both fragments were digested with NdeI and ligated, and the concatenated fragment was amplified in a third PCR. This fragment was digested with BamHI and XbaI and cloned into the same sites of vector pBluescript KS resulting in plasmid pJTF116. In a fourth PCR, the coding fragment for the YycG PAS domain (amino acids 255-373) was amplified with primers ON94 and ON95, introducing 5′ XbaI and 3′ SacII sites, and the resulting fragment was cloned into the respective sites of vector pJTF116, resulting in pJTF119.

To construct a single copy expression vector for the PAS domain, the P_yycF-3×c-myc-yycG_PAS fragment was excised with enzymes BamHI and BglII from vector pJTF119 and cloned into a BamHI site of thrC integration vector pDG1664, resulting in plasmid pJTF122. This plasmid was linearized with AatII and transformed into wild type strain JH642 and yycG depletion strain JH25033 selecting for Ery^R and resulting in strains JH25066 and JH25067.

For multicopy expression of the YycG PAS domain, the 3×c-myc-yycG_PAS fragment was excised with enzymes HincII and BglII and ligated into SmaI and BamHI sites of multi-copy vector pHT315S, resulting in plasmid pJTF126. This plasmid was transformed into wild type strain JH642 resulting in strains JH25068.

Since a single copy of the catalytic YycG domains could not complement for the absence of the full-length protein a multi-copy expression vector for the YycG catalytic HisKA and HATPase domains was constructed. The fragment coding for amino acids 378-611 was amplified from genomic DNA with primers ON96 and ON97, introducing 5′ XbaI and 3′ SacII sites. This fragment was cloned into the same sites of vector pJTF116 resulting in vector pJTF121. The 3×c-myc-YycG_△2-377-fragment was excised with HincII and BamHI and cloned into SmaI and BamHI sites of vector pHT315S, resulting in pJTF127. When transformed into wildtype strain JH642, no colonies could be obtained suggesting that constitutive expression of this YycG fragment is toxic to the cell. To regulate expression strain JH642 was first transformed with amyE integration vector pJM119, which also harbors a lacI copy, rendering the pHT315S P_spac promoter IPTG inducible (strain JH25065). When this strain was transformed in the absence of IPTG transformants could be obtained (strain JH25069). This strain was viable up to an IPTG concentration of 0.2 mM.

To construct a strain expressing a C-terminally truncated YycG construct devoid of the catalytic HisKA and HATPase domains plasmid pJTF119 was digested with NdeI and XbaI to remove the 3×c-myc-tag, blunt-ended with T4 DNA polymerase and self ligated, resulting in vector pJTF124. The P_yycF-yycG_PAS fragment was excised with enzymes PstI and BglII and cloned into PstI and BamHI sites of vector pJM103, resulting in plasmid pJTF125. This plasmid was transformed into strain JH25032, resulting in strain JH25064, which expresses a C-terminally truncated fragment of YycG from its native locus and a wild type copy of YycG ectopically under IPTG control.

A control strains that expressed 3×c-myc-tagged YycG in single copy was also constructed. For this strain an N-terminally truncated fragment of yycG was PCR amplified with primers ON98 and ON99, introducing 5′ EcoRI and 3′ XhoI sites, and cloned into the same sites of vector pKL94 resulting in plasmid pJTF128. This plasmid was introduced into JH642 via single crossover recombination resulting in strain JH25063, which harbors a yycG-3×c-myc construct in its native locus.

Cell division mutant strains

Since several of the cell division strains used here came from different sources (see table I), genomic DNA was prepared for these strains and transformed into strain 168, to assure an isogenic background for all of these mutations. Strain 804, strain YK200 and strain FtsWp were already in the isogenic 168 background.

To construct a triple mutant strain with conditional expression of ftsL, pbpB and deleted for divIB, the cat cassette in the divIB strain JH25071 was converted to tet^R by transformation with vector pCM::tc resulting in strain JH25072. Strain 804 was transformed with JH25072 genomic DNA, selecting for Tet^R and resulting in strain JH25073.

Tagged-YycH and YycI expression plasmids

The yycH gene was PCR amplified from chromosomal DNA with primers ON100 and ON101, digested with XbaI and SacII and cloned in the same sites of vector pJTF116, resulting in plasmid pJTF117. The P_yycF-3×c-myc-yycH coding fragment was excised by BamHI digest and cloned into vector pDG1664 resulting in vector pJTF129. This vector was digested with EcoRI and EcoRV and the P_yycF-3×c-myc-yycH fragment was subcloned into EcoRI and SmaI sites of vector pHT315 resulting in vector pJTF130. Since the 3×c-myc tagged fragment was not well detectable in western blots the c-myc-tag was expanded to 6×c-myc. This was achieved by PCR amplification of pJTF129 with ON102 and ON103. The resulting fragment was digested with NheI and XbaI and ligated into the XbaI site of pJTF130, resulting in pJTF131. This vector was transformed into the yycH-deleted strain JH25021 resulting in strain JH25077, which expresses a 6×c-myc-tagged YycH protein.

To construct a 6×T7-tagged YycI expression plasmid, the yycI gene was PCR amplified from chromosomal DNA with ON104 and ON105. The amplified fragment was digested with XbaI and SacII and cloned into the same sites of vector pJTF116, resulting in pJTF118. The 3×T7-tag was amplified by PCR on vector pSHU1 with ON106 and ON107 or ON108 and ON109. The two amplified fragments were digested with NdeI and BglII or BglII and XbaI, respectively and ligated with NdeI and XbaI digested pJTF118 in a three-way ligation, resulting in vector pJTF132. The P_yycF-6×T7-yycI coding fragment was excised with PvuII and EcoRI and cloned into SmaI and EcoRI sites of vector pHT315, resulting in pJTF133. This vector was transformed into yycI deleted strain JH25022, resulting in strain JH25078, which expresses 6×T7 tagged YycI protein.

Two hybrid plasmids

Most of the two-hybrid plasmids utilized here have been published previously (Daniel et al., 2006; Szurmant et al., 2007b). All other were generated by amplifying the coding sequence for the respective genes introducing 5′XbaI and 3′KpnI sites, and cloning the fragments into the vectors pKT25 and pUT18C.

Bacterial two-hybrid experiments

Protein interactions were probed utilizing the bacterial two-hybrid assay of Ladant and colleagues (Karimova et al., 1998; Karimova et al., 2000). In this assay, putative interaction partners are fused C-terminally either to Bordatella pertussis adenylate cyclase domain T18 or to domain T25. When interacting T18 fusion proteins and T25-fusion proteins are co-expressed in the adenylate cyclase deficient E. coli strain BHT101, adenylate cyclase activity is reconstituted, which in turn leads to expression of β-galactosidase. β-galactosidase activity was measured on X-gal indicator plates. Since many of the constructs used here were toxic to the E. coli strain, this method is preferable to more quantitative liquid assays. The assay was performed as previously described (Daniel et al., 2006). Briefly, E. coli competent cells were co-transformed with the indicated plasmid combinations and plated directly as spots on an X-gal indicator plate, followed by growth at 30°C for 16 h. X-gal indicator plates were composed of 1% Agar No. 1 (Oxoid), 10 mM ammonium chloride, 1.2 mM ammonium nitrate, 1 mM magnesium sulfate, 0.75 mM sodium sulfite, 0.5 mM potassium dihydrogen phosphate, 0.1 mM manganese (II) chloride, and 4 μM iron (III) chloride. The pH was adjusted to 7 with sodium hydroxide. Then, 0.8% glucose, 0.4% Casamino Acids, 3 μM thiamine, 100 μg of ampicillin/ml, 25 μg of kanamycin/ml, and 0.004% X-Gal were added to the medium prior to pouring the plates. Colonies of strains with non-interacting protein-fusions appear white, whereas strains harboring strongly interacting protein fusions appear dark blue.

Purification of anti-YycH and anti-YycI antibodies for immunofluorescence

The antibodies have been described elsewhere (Szurmant et al., 2008). The antibodies were ammonium sulfate precipitated and resuspended as described (Szurmant et al., 2007a). Antibodies were then purified for use in immunofluoresence experiments by preadsorption with acetone powder derived from yycH or yycI deleted strains, which was generated as described (Harlow and Lane, 1988). The acetone powder was added to the appropriate antibody solutions, the suspensions were shaken for 30 min and cleared by centrifugation. Operationally pure antibodies were obtained following 7 cycles of acetone powder adsorptions.

Immunofluorescence microscopy

The detailed protocol for immunofluorescent detection of YycG and FtsZ has been published (Fukushima et al., 2008; Szurmant et al., 2007a) and was adapted directly for all strains under investigation. However, to achieve depletion of essential genes in conditional strains growth conditions had to be adjusted individually by varying incubation times in the absence of IPTG (for P_spac) or in the absence of xylose (for P_xyl) or by shifting growth temperatures to non permissible temperatures (37°C for divIC and 48°C for divIB strains) as indicated in the text. For visualization of YycH, YycI and YycG (Fig. 4) glutaraldehyde fixation of cells was avoided to reduce background fluorescence. Primary purified antibodies were used at 1:20 (YycG), 1:40 (YycH), 1:300 (YycI) and 1:100 (FtsZ) dilutions. Commercial mouse anti-c-myc antibody (Invitrogen) was used at 1:500 to detect myc-tagged YycG or YycH constructs. Secondary antibodies were Alexa Fluor 488 conjugated goat anti-rabbit IgG (against anti-YycG antibody, anti-YycH antibody and anti-YycI antibody), Alexa Fluor 488 conjugated chicken anti-rabbit IgG (against anti-YycG antibody), Alexa Fluor 546 conjugated donkey anti-sheep IgG (against anti-FtsZ antibody), Alexa Fluor 546 conjugated goat anti-mouse IgG (against anti-c-myc antibody), Alexa Fluor 568 conjugated goat anti-mouse IgG (against anti-c-myc antibody) (Molecular Probes) used at dilutions of 1:500, 1:500, 1:1,000, 1:1,000 and 1:1,000, respectively. Tertiary antibody was Alexa Fluor 568 conjugated donkey anti-goat IgG (Molecular Probes) used at dilution of 1:1,000 for determination of 6×c-myc-YycH localization. DNA was stained with DAPI (final concentration, 1 μg/ml). The samples were visualized either with a Bio-Rad Radianece 2100 Rainbow laser scanning confocal microscope (running under Bio-Rad LaserSharp 2000 software) equipped with blue diode/argon/green HeNe/red diode laser or with a Carl Zeiss Axio Imager M1 microscope (a Plan-APOCHROMAT Fluorite differential interference objective [magnification, ×63; numerical aperture, 1.4], and standard filter sets for DAPI, Alexa Fluor 488 and 568). The sample pictures were taken with a charge-coupled device camera (AxioCam MRm; Carl Zeiss) driven by AxioVision software (version 4.6; Carl Zeiss). Some samples (Fig. 4) were also visualized with a Nikon TE2000-U microscope (60× PlanApo oil-immersion objective; numerical aperture 1.4). A series of optical sections collected at 100 nm intervals in the z-axis was used to create the images. All pictures were processed with Adobe Photoshop software.

Immunoprecipitation experiments

The immunoprecipitation protocol was performed essentially as published (Fukushima et al., 2008). Cells were grown in 100 ml of SM at 37°C until A525nm = 0.6. 5-15 OD525nm units of cells were fixed with formaldehyde solution (1% final concentration) in 32 mM sodium phosphate buffer at pH 7.4 for 20–30 min. Following washing with PBS, cells were suspended in 4.5 mL IP buffer [PBS containing 10 mg/ml of lysozyme, 0.1 M EDTA, 1% Triton X-100 and protease inhibitor cocktail (Roche)], the suspension was kept at 37°C for 1 h before a passage through a French press. Cell debris was removed by centrifugation and the cell extract was completely cleared by gentle shaking for 30 min followed by an additional centrifugation step. 0.9 ml of cross-linked cellular extract was subjected to immunoprecipitation by addition of purified antibody raised against YycG (1:900), or crude antibody raised against YycH (1:90) or YycI (1:2250). After 16 h incubation at room temperature, the mixture was cleared by centrifugation, the supernatant was mixed with 40 μl of recombinant Protein G agarose (Invitrogen) and gently shaken for 4 h at room temperature. The resin was washed with PBS containing 1% Triton X-100 and the immunocomplexes were eluted by boiling with SDS sample buffer. Samples were used in immunoblot analysis in amounts corresponding to cell density A525nm = 0.1.

Immunoblot analysis

Whole cell protein extract was derived, separated by SDS-PAGE and YycG, YycH, YycI and FtsZ were visualized immunologically as described (Fukushima et al., 2008; Szurmant et al., 2007a). Purified antibodies were used at 1:10,000 dilutions for rabbit anti-YycG antibody, at 1:5000 for rabbit anti-YycH or anti YycI antibodies and at 1:10,000 dilutions for sheep anti-FtsZ antibody. Mouse anti-c-myc (Invitrogen) and mouse anti-T7 (Novagen) monoclonal antibodies were utilized at 1:10,000. Secondary horseradish peroxidase-labeled antibodies were used at 1:10,000 dilutions.

RT-PCR assay

The RT-PCR assay to quantify YycFG-dependent gene expression has been described (Fukushima et al., 2008) but needed to be adjusted for the conditional strains utilized here. Generally, cells were grown at 37°C (30°C for the divIC strain) to an OD₅₂₅ of 0.3 (in the presence of 1 mM IPTG, 0.5% xylose), collected by centrifugation, washed and incubated in fresh media with or without 1 mM IPTG or 0.5 % xylose and incubated for an additional 3-6 h. For divIB and divIC strains, growth temperatures where shifted from 37 to 48°C and from 30 to 37°C, respectively. Fifteen to twenty OD525nm units of cells were collected, total RNA isolated, treated with DNase I and quantified as described. RNA (1 μg) was used as template to produce cDNA utilizing SuperScript III Reverse Transcriptase (Invitrogen), following the manufacture’s recommendation in a total volume of 20 μl. 5 μl of the resulting cDNA was used in a standard PCR reaction, and resulting DNA was visualized by agarose gel electrophoresis and ethidium bromide staining. To assure absence of DNA contamination, the same procedure was also performed for each pair of oligonucleotides and mRNA preparation in the absence of reverse transcriptase and, in all experiments, no product was detected (not shown).

Table II.

Plasmids used in this study

Plasmid	Description	Source or reference
pBluescript II (KS+)	cloning vector	Stratagene
pJM103	single cross-over integration vector (bla, cat)	laboratory stock
pJM119	amyE integration plasmid (bla, kanR)	laboratory stock
pDG1664	thrC integration plasmid (bla, eryR)	(Guerout-Fleury et al., 1996)
pJM134	pBluescript II (KS+)-spcR (bla, spcR)	(Szurmant et al., 2008)
pHT315-Pspac	E.coli/B.subtilis shuttle vector (bla, eryR)	(Worner et al., 2006)
pSHU1	pMUTIN2::3×T7-tag	Shu Ishikawa
pKL94	single cross over integration vector, 3xc-myc-tag (bla, cat)	(Perez et al., 2000)
pCM::tc	Vector to exchange cat with tetR marker	(Steinmetz and Richter, 1994)
pKT25	T25 expressing two-hybrid vector; Kanr	Hybrigenics
pUT18C	T18 expressing two-hybrid vector; Ampr	Hybrigenics
pJS76	pJM134-yycF-yycG-spcR-Pspac-yycH’	(Szurmant et al., 2008)
pJS78	pJM134-yycF-yycGΔ44-167-spcR-Pspac-yycH’	this work
pJS79	pJM134-yycF-yycGΔ2-203-spcR-Pspac-yycH’	this work
pJS80	pJM134-yycF-yycGΔ2-255-spcR-Pspac-yycH’	this work
pJS81	pJM134-yycF-yycGΔ2-373-spcR-Pspac-yycH’	this work
pJTF116	pBluescriptII (KS+)-PyycF-3×c-myc	this work
pJTF117	pBluescriptII (KS+)-PyycF-3×c-myc-yycH	this work
pJTF118	pBluescriptII (KS+)-PyycF-3×c-myc-yycI	this work
pJTF119	pBluescriptII (KS+)-PyycF-3×c-myc-yycGΔ2-254,Δ374-611	this work
pJTF121	pBluescriptII (KS+)-PyycF-3×c-myc-yycGΔ2-377	this work
pJTF122	pDG1664-PyycF-3×c-myc-yycGΔ2-254,Δ374-611	this work
pJTF124	pBluescriptII (KS+)-PyycF-yycGΔ2-254,Δ374-611	this work
pJTF125	pJM103-PyycF-yycGΔ2-254,Δ374-611	this work
pJTF126	pHT315-Pspac-3×c-myc-yycGΔ2-254,Δ374-611	this work
pJTF127	pHT315-Pspac-3×c-myc-yycGΔ2-377	this work
pJTF128	pKL94-yycGΔ1-451-3×c-myc	this work
pJTF129	pDG1664- PyycF-3×c-myc-yycH	this work
pJTF130	pHT315- PyycF-3×c-myc-yycH	this work
pJTF131	pHT315- PyycF-6×c-myc-yycH	this work
pJTF132	pBluescriptII (KS+)-PyycF-6×T7-yycI	this work
pJTF133	pHT315- PyycF-6×T7-yycI	this work