Myxococcus xanthus, a Gram-negative developmental bacterium, contains a large number of protein Ser/Thr kinases (PSTKs). Among these PSTKs, Pkn4 has been shown to be 6-phosphofructokinase (PFK) kinase. PFK associates with the regulatory domain of Pkn4 (Pkn4RD) and is activated by Pkn4-mediated phosphorylation. The activation of PFK is required to consume glycogen accumulated during early development and is essential for efficient sporulation. Using the yeast two-hybrid screen, we identified three new factors, MkapA, MkapB and MkapC, that interact with Pkn4 and each contains well-known protein–protein interaction domains. MkapB contains eight tandem repeats of the TPR (tetratrico peptide repeat) domain and its interaction with Pkn4RD was phosphorylation-dependent. MkapB remained associated with Pkn4RD. As a result, Pkn4 did not interact with PFK and its activation was inhibited. While deletion of the pfk-pkn4 operon did not inhibit fruiting body formation, the spore yield was low. In contrast, a mkapB deletion mutant exhibited a 24 h delay in fruiting body formation, accumulated less glycogen in the stationary phase and gave rise to 3.2% spore formation as opposed to 100% attained with DZF1. In addition to Pkn4, MkapA associated with other membrane-associated PSTKs, Pkn1, Pkn2, Pkn8 and Pkn9, while MkapB associated with Pkn8 and Pkn9, and MkapC with Pkn8. These results indicate that there are complex PSTK networks in M. xanthus that share common modulating factors.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Myxococcus xanthus is a Gram-negative soil bacterium that exhibits a communal lifestyle during vegetative growth and multicellular development. Upon nutritional starvation, M. xanthus cells begin to migrate cooperatively towards aggregation centres by gliding on the solid surface, forming a multicellular fruiting body which holds 105 cells. About 10% of the starved cells within fruiting bodies can differentiate to myxospores, which are dormant and resistant to various environmental stresses. The developmental process is achieved by a series of sophisticated intercellular signalling pathways that regulate the expression of a specific set of genes (reviewed by Dworkin, 1996). For this purpose, M. xanthus has developed numerous signal transduction pathways involving a receptor-type histidine (His) kinase (HK) and eukaryotic serine/threonine (Ser/Thr) kinase.
In bacteria, the two-component signal transduction system is predominant. This involves a receptor-type HK that is autophosphorylated at a His residue upon signal binding and the phosphoryl group is subsequently transferred to an aspartic acid (Asp) residue in a response regulator. This phosphorylation induces a conformational change in the regulatory domain (RD) of the response regulator resulting in the activation of the effector domain. The activated effector domain, equipped with a DNA-binding motif, regulates gene expression in response to environmental signals (reviewed by Stock et al., 2000). In M. xanthus, HKs that are essential in development have been investigated: asgA (Plamann et al., 1995), asgD (Cho and Zusman, 1999a), espA (Cho and Zusman, 1999b), sdeK (Pollack and Singer, 2001), and mrpA (Sun and Shi, 2001). SdeK probably monitors the accumulation of (p)ppGpp, whereas AsgA and AsgD appear to be involved in both environmental sensing and intercellular signalling. mrpA and its cognate response regulator gene, mrpB, are required for the synthesis of an essential transcription factor, MrpC. MrpC is a transcriptional activator of the fruA gene, which encodes an essential response regulator for fruiting body development (Ueki and Inouye, 2003).
On the other hand, in eukaryotes, protein Ser/Thr kinase (PSTK) and Tyr kinase (PTK) are known to play a major role in signal transduction pathways that modulate cellular responses and adaptations to various environmental conditions. Upon ligand binding, PSTK and PTK are phosphorylated, activated and recruit a variety of proteins to form signalling complexes in a specific signalling pathway. These signalling complexes are assembled by protein–protein interactions via protein interacting domains that function in phosphorylation-dependent manner (reviewed by Hunter, 2000). Recent advances in the sequencing of whole prokaryotic genomes revealed that PSTKs exist in a wide variety organisms such as the pathogenic bacteria Mycobacterium tuberculosis and Mycobacterium leprae, and the developmental bacteria, M. xanthus, Streptomyces sp. and Synechocyctis sp. However, only a few studies have demonstrated their functional roles at a molecular level. In the case of Streptomyces coelicolor A3(2), AfsK has been shown to phosphorylate AfsR, a transcription activator for afsS involved in secondary metabolism (Lee et al., 2002). AfsR is also phosphorylated by other AfsK-like protein, PkaG and AfsL, suggesting that the regulatory networks involving PSTKs are diverse and versatile in S. coelicolor A3(2) (Sawai et al., 2004). The crystal structure of the catalytic domain (CD) of PknB from M. tuberculosis showed remarkable conservation with the PSTK superfamily in eukaryotes (Ortiz-Lombardia et al., 2003). Furthermore it has been suggested that PknB and its phosphatase, PstP, may function as a pair in vivo to control mycobacterial cell growth (Boitel et al., 2003). However, protein substrates of PknB have yet to be identified.
The genomic analysis (http://tigrblast.tigr.org) revealed that M. xanthus contains at least 99 PSTKs, many of which have a hydrophobic sequence between the kinase CD and the RD, suggesting that they may function as a receptor-type kinase. Among 99 PSTKs, Pkn1 (Munoz-Dorado et al., 1991), Pkn2 (Udo et al., 1995), Pkn5 and Pkn6 (Zhang et al., 1996), Pkn9 (Hanlon et al., 1997), Pkn4 (Nariya and Inouye, 2002; 2003) and MasK (Thomasson et al., 2002) have been characterized for their roles in the M. xanthus life cycle. Pkn1, Pkn2, Pkn6, Pkn9 and MasK possess a hydrophobic sequence, and Pkn2, Pkn6 and Pkn9 were shown to be transmembrane kinases. Pkn4 is the membrane-embedded 6-phosphofructokinase (PFK) kinase exposed both CD and RD in the cytoplasm. Pkn4 activates PFK by phosphorylation at Thr-226 during fruiting body development and sporulation. The activation of PFK by Pkn4 is required for glycogen consumption and for effective myxospore formation (Nariya and Inouye, 2002; 2003).
In order to elucidate factor(s) involved in the regulatory mechanism of PFK activation by Pkn4 during vegetative growth and fruiting body development, a yeast two-hybrid screen was performed using Pkn4 as bait and the M. xanthus genomic library. Three factors, MkapA, MkapB and MkapC, containing well-characterized protein–protein interaction domains, were found to associate with Pkn4. Among these factors, MkapB associated with Pkn4 in a phosphorylation-dependent manner and this association inhibited PFK activation by interfering with its interaction with Pkn4. These results indicate that MkapB functions as a modulator for PFK activation by affecting glycogen metabolism. Furthermore, using membrane-associated PSTKs, Pkn1, Pkn2, Pkn8 and Pkn9, as bait, MkapB was also found to associate with Pkn8 and Pkn9. MkapA associated with all PSTKs tested while MkapC associated with only Pkn8 in addition to Pkn4. As deletion of mkapB showed a significant delay in fruiting body formation, these multi-kinase associated proteins (Mkaps) may play important roles in modulating signal transduction through membrane-associated PSTKs in M. xanthus.
Identification of Pkn4-associating factors
We reported earlier that Pkn4 activates PFK by phosphorylation to control glycogen metabolism for effective sporulation during fruiting body development (Nariya and Inouye, 2003). Interestingly, while the expression levels of PFK and of Pkn4 remain similar in vegetative growth as well as in the stationary phase, the glycogen content in the cell is drastically increased in the stationary phase. This suggests that there may exist a modulating factor(s) for Pkn4 activity. In the present work, we applied a yeast two-hybrid screen using Pkn4 (Fig. 1A) as bait to identify this putative factor(s). Three Pkn4-associating factors were identified. The functional roles of these factors have not been previously described. Thus we designated them multi-kinase associated protein: (MkapA, MkapB and MkapC) (Table 1 and Fig. 1B).
Table 1. The positive clones identified by yeast two-hybrid screen using the membrane PSTKs as bait from Myxococcus xanthus genomic library.
MkapA consists of 151 residues with a C2H2-type Zn-finger-like motif, CX3CX9HX3H, and a blast search did not reveal homology with any other proteins (http://www.ncbi.nlm.nih.gov). In eukaryotes, many Zn-finger proteins including the C2H2-type Zn-finger have been shown to associate with PSTKs and PTKs by yeast two-hybrid screens (Wolf and Rohrschneider, 1999). Thus, the Zn-finger-like motif of MkapA most probably functions as the association domain for Pkn4. MkapB consists of 271 residues containing eight tandem repeats of the tetratrico peptide repeat (TPR) domain, which is also known to be a protein–protein interaction domain (Das et al., 1998). MkapC is a large protein consisting of 1167 residues with three tandem repeats of fibronectin type 3 (FN3)-like domain from residues 252–507 which is satisfactory for association with Pkn4 based on the DNA sequences. The N-terminal region from the FN3-like domain has no homology to any other proteins and the C-terminal region has a weak similarity to heparinase II from Flavobacterium heparinus (Su et al., 1996). MkapA, B and C are likely to be cytoplasmic proteins because they contain neither a putative hydrophobic sequence nor a signal peptide [as predicted by the tmhmm and signalp programs (http://www.cbs.dtu.dk)].
Pkn4 domain(s) for association with Mkaps A, B and C
To determine the domain(s) in Pkn4 that associate with MkapA, B or C in the yeast two-hybrid screen, the Pkn4-kinase catalytic domain (Pkn4CD) and the Pkn4-regulatory domain (Pkn4RD) shown in Fig. 1A were cloned in pGBD-NdeI and used as bait. Yeast strain PJ69-4A harbouring both bait and prey plasmids was grown to mid-log phase and subjected to a spot test as described by Ohta and Newton (2003). As shown in Fig. 2A, the cells containing Pkn4 with MkapA or B or C formed colonies within 2 days while the cells harbouring Pkn4RD formed colonies only with MkapB and C but not with MkapA which associated with Pkn4CD. After 7 days of incubation (Fig. 2B), the cells containing Pkn4RD and MkapA did form colonies, indicating that their association was not strong. No difference of growth was observed for all combinations in SD-W–L– medium (data not shown).
To quantify these interactions, β-galactosidase activity was assayed as shown in Table 2. MkapA associated with Pkn4 and Pkn4CD with high efficiently exhibiting β-galactosidase activities of 30.5 and 31.8 U respectively. On the other hand, the association of MkapB and MkapC with either Pkn4 or Pkn4RD were similar and at lesser efficiency (approximately 10 U). MkapA association with Pkn4RD was the weakest (2.1 U). All other combination displayed the same β-galactosidase activity as that of the empty prey plasmid. The results from the β-galactosidase assays correlated well with the growth profiles of the serially diluted spots (Fig. 2). Thus, these results indicate that Pkn4 associates with MkapB and MkapC through the RD domain, whereas Pkn4 associates with MkapA through the CD domain.
Table 2. Interaction of Mkaps with Pkn4, Pkn4CD and Pkn4RD measured by β-galactocidase activity in yeast two-hybrid screen.
β-Galactocidase assay was performed by the method described by Miller (1972) on the same cultures used in Fig. 2 and three independent cultures of each strain.
Although we previously showed that Pkn4 and Pkn4RD co-precipitated with PFK using anti-PFK IgG (Nariya and Inouye, 2002), PFK was not detected using Pkn4 as bait (Table 1). Therefore, the association of Pkn4 with PFK was also examined using Pkn4, Pkn4CD and Pkn4RD as bait. PFK was able to associate with Pkn4RD, but not with the entire Pkn4 protein (data not shown), indicating that the in vivo association between Pkn4 and PFK is subtle in the presence of ATP as described later.
Phosphorylation of MkapB by Pkn4 inhibits PFK phosphorylation
To investigate the possible roles of the Mkaps in the Pkn4-PFK cascade, we first examined the effect of MkapB on PFK phosphorylation by Pkn4. MkapB was purified as described in Experimental procedures and was found to form a dimer on the basis of gel filtration analysis (data not shown). The phosphorylations of PFK and MkapB by Pkn4 were performed as described previously (Nariya and Inouye, 2002). When PFK (1 µg) and MkapB (1 µg) were separately added to the phosphorylation mixture containing Pkn4 (0.1 µg) and incubated for 30 min in P-buffer [50 mM Tris-HCl (pH 8.0), 10 mM MgCl2, 2 mM DTT] with 200 µM ATP and 1.2 µCi of [γ-32P]-ATP, PFK and MkapB were each phosphorylated as shown in lanes 1 and 2 in Fig. 3A respectively. Interestingly, as shown in lane 3, when both PFK and MkapB were added to the same reaction mixture, MkapB phosphorylation was observed at the same level as in the case of MkapB alone (lane 2), whereas PFK phosphorylation was decreased to 20%. To further examine the effect of MkapB on PFK phosphorylation, PFK (1 µg) was added to the mixture after MkapB (1 µg) was pre-incubated with Pkn4 (0.1 µg) for 5 (lane 4) and 30 (lane 5) min. The reaction mixtures were incubated for another 30 min. As shown in lanes 4 and 5, PFK phosphorylation by Pkn4 was completely abolished under these conditions. MkapB did not affect Pkn4 autophosphorylation, however. As interaction between PFK and MkapB was not detected in the yeast two-hybrid screen (data not shown), MkapB-mediated inhibition of PFK phosphorylation did not result from its binding to PFK. Instead, as both MkapB and PFK can associate with Pkn4RD (Fig. 2), these results suggested that Pkn4 association with MkapB, even after its phosphorylation, inhibited subsequent association between Pkn4 and PFK. On the other hand, the reverse inhibition was not observed. Pre-incubation with Pkn4 and PFK did not inhibit MkapB phosphorylation by Pkn4. Both PFK and MkapB were phosphorylated under these conditions (lane 6).
Phosphorylation-dependent association of MkapB with Pkn4
To further elucidate the MkapB inhibition on PFK phosphorylation, the associations of MkapB and PFK with Pkn4 were examined by immunoprecipitation analysis using anti-Pkn4RD IgG as described previously (Nariya and Inouye, 2002). Pkn4 or Pkn4RD were incubated with MkapB in the presence or absence of ATP, and immunoprecipitates were than analysed by SDS-PAGE. As shown in Fig. 4A, MkapB co-precipitated with Pkn4 in the presence of ATP (lane 2), but not in its absence (lane 3). The MkapB band in lane 2 was broad and its position was slightly up-shifted compared with that of purified MkapB (lane 1) probably due to phosphorylation by Pkn4.
To clarify whether phosphorylation of MkapB is required for its association with Pkn4, Pkn4 was first autophosphorylated in P-buffer with ATP. The buffer was then changed to P-buffer without ATP using BIOMAX-10K (Millipore). MkapB was added to phosphorylated Pkn4 (Pkn4-P) and immunoprecipitation was performed. As shown in lane 1 of Fig. 4B, MkapB was not detected. On the other hand, MkapB did co-immunoprecipitate with Pkn4 when ATP was added to the mixture at a final concentration of 200 µM (lane 2). These results indicate that autophosphorylation of Pkn4 is not sufficient, and that phosphorylation of MkapB by Pkn4 is absolutely required for their association. Although the interaction of MkapB and Pkn4RD was observed in the yeast two-hybrid screen (Fig. 2 and Table 2), MkapB did not co-immunoprecipitate with Pkn4RD in the presence or absence of ATP (lanes 4 and 5 in Fig. 4A). It is possible that MkapB might be phosphorylated by a PSTK in yeast cells resulting in the interaction of phosphorylated MkapB with Pkn4RD.
To elucidate how MkapB interferes with PFK phosphorylation by Pkn4, the associations of PFK and MkapB with Pkn4 were examined in the presence or absence of ATP. As shown in Fig. 4C, when PFK (1 µg) was added to P-buffer containing Pkn4 (250 ng) without ATP, they formed a complex as detected by immunoprecipitation using anti-Pkn4RD IgG (lane 2) as reported previously (Nariya and Inouye, 2002). On the other hand, the amount of PFK bound to Pkn4 decreased in the presence of ATP (lane 1), indicating that once phosphorylated, PFK dissociated from Pkn4. To examine the effect of MkapB on PFK and Pkn4 association, MkapB (2.5 µg) and Pkn4 (250 ng) were pre-incubated for 30 min with or without ATP and then PFK was added to the mixture. After further incubation for 60 min, Pkn4 complexes were immunoprecipitated by anti-Pkn4RD IgG. Surprisingly, only MkapB was detected with Pkn4 in the presence of ATP (lane 3), while only PFK was detected with Pkn4 without ATP (lane 4). These results indicate that PFK is not able to associate with Pkn4 in the presence of phosphorylated MkapB. Therefore, the inhibition by MkapB on PFK phosphorylation results from the binding of phosphorylated MkapB to Pkn4 that interferes with Pkn4 binding to PFK.
The phospho-amino acids of MkapB that are phosphorylated by Pkn4 were analysed as described in Experimental procedures. The band corresponding to MkapB in lane 2 in Fig. 3A was excised from the PVDF membrane and analysed to determine the phospho-amino acids. As shown in Fig. 3B, MkapB was found to be phosphorylated at a Thr residue by Pkn4.
mkapB gene expression during M. xanthus life cycle
As expression of the pfk-pkn4 operon was previously reported to be constant throughout vegetative growth and development (Nariya and Inouye, 2003), it was important to similarly examine mkapB expression. To determine a transcriptional initiation site(s) used during vegetative growth and development, primer extension analysis was carried out using total RNAs extracted from the mid-log (ML) and the early stationary (ES) phase cells during vegetative growth (Fig. 5C), and 12 and 24 h developmental cells of DZF1 as shown in Fig. 5A. The transcriptional initiation site was G for both vegetative growth and development (Fig. 5B). To characterize the pattern of mkapB gene expression, an mkapB–lacZ fusion was constructed with the mkapB promoter region from −331 to +202 with respect to the transcriptional initiation site fused to the lacZ gene as reporter (Fig. 5B). The mkapBP–lacZ/DZF1 strain was obtained by integrating the pmkapBP–lacZ into the original location of the DZF1 chromosome. β-Galactosidase activity of mkapBP–lacZ/DZF1 was low until the late-log (LL) phase in vegetative growth and then was greatly increased at the stationary phase (Fig. 5D). When the ES phase cells were spotted on TM agar plates as described in Experimental procedures, β-galactosidase activity remained at almost the same level until 6 h and then steadily decreased until 24 h as development progressed.
In contrast to the expression of the pfk-pkn4 operon, mkapB expression was upregulated when cells entered the ES phase and stayed at a high level until 12 h of development. Therefore, it appears that MkapB that is expressed in the ES phase is likely phosphorylated by Pkn4 thereby inhibiting PFK activation by interfering its association with Pkn4. This then results in glycogen accumulation at the ES phase (Nariya and Inouye, 2003).
Effect of MkapB on glycogen metabolism
Previously, we demonstrated that glycogen accumulates at the stationary phase and that the accumulated glycogen is consumed for effective myxospore formation during development. The deletion of pfk-pkn4 strain (ΔPFKN) caused a higher accumulation of glycogen at the ES phase relative to DZF1 and the accumulated glycogen was not utilized during development resulting in low spore yield (Nariya and Inouye, 2003). To elucidate the effect of MkapB on glycogen metabolism, an mkapB deletion strain, ΔmkapB, was constructed by replacing the 0.7 kb SacI–BstBI fragment of the mkapB gene with the 1.3 kb kanamycin-resistant gene (Fig. 6A). ΔmkapB grew normally up to the LL phase similar to DZF1 in CYE medium (Fig. 5C), but then entered into the stationary phase earlier than DZF1 and did not reach maximum growth. The ΔmkapB cells began to lyse 12 h earlier than DZF1.
To elucidate the effect of MkapB on glycogen metabolism during vegetative growth and at 48 h of development, the levels of glycogen were measured for ΔmkapB, ΔPFKN and DZF1. As shown in Fig. 6B, the glycogen amounts of the three strains were similarly low at the ML and LL phases. At the ES and MS phases, the glycogen levels were greatly increased in both DZF1 (0.63 unit) and ΔPFKN (1.23 unit). On the other hand, the levels in ΔmkapB were about 24% and 36% of DZF1 at the ES and MS phases respectively. After 48 h of development initiated by spotting the ES phase cells on TM agar plates, the glycogen amounts were 0.12, 1.23 and 0.14 units in DZF1, ΔPFKN and ΔmkapB respectively. The accumulated glycogen in DZF1 and ΔmkapB were consumed during early development, but the level of glycogen of ΔPFKN stayed unchanged as reported previously (Nariya and Inouye, 2003). Therefore, the Pkn4-mediated activation of PFK that is required for glycolysis was inhibited by MkapB whose expression was upregulated in the ES phase and glycogen thus accumulated. On the other hand, in ΔmkapB glycogen accumulation was less than in DZF1 at the ES phase because MkapB was not present to inhibit PFK activation of PFK by Pkn4 and thus glycolysis proceeded.
It was demonstrated previously that the glycogen accumulated at the ES phase in DZF1 was almost consumed at 30 h during development leading to a high spore yield, whereas no glycogen consumption was observed in ΔPFKN resulting in a low spore yield (Nariya and Inouye, 2003). The spore yields of DZF1, ΔPFKN and ΔmkapB at 48 h of development were 7.1%, 0.12% and 0.23% of the initial cell numbers used for development respectively. The spore yield of ΔmkapB was 3.2% compared with that of DZF1 (100%), and was double that of ΔPFKN (1.7%). Therefore, these results strongly support the previous conclusion that the consumption of the accumulated glycogen in early development is essential for effective spore formation.
Effect of MkapB on fruiting body formation
Given that MkapB also associates with Pkn8 and Pkn9 (Table 1), the role of mkapB in fruiting body formation was investigated. As shown in Fig. 6C, ΔmkapB showed a 24 h delay in fruiting body formation compared with DZF1 on TM agar plates spotted with the ES phase cells, whereas ΔPFKN did not show any delay in fruiting body formation (Nariya and Inouye, 2003). ΔmkapB did not exhibit any defective phenotype in either A or S motilities (data not shown). MkapB seems to have pleiotropic effects on fruiting body formation and sporulation through interacting with Pkn8 and Pkn9, and probably with other PSTKs.
Association of MkapA and MkapC with Pkn4
As shown in Fig. 2, MkapA and MkapC were found to associate with Pkn4CD and Pkn4RD by the yeast two-hybrid screen respectively. In order to demonstrate their interaction, we performed pull-down analysis using haemagglutinin epitope (HA)-tagged MkapA (HA-MkapA) and the FN3 region in MkapC (HA-FN3) as described in Experimental procedures. Histidine-tagged Pkn4CD or Pkn4RD was added to the soluble fractions of E. coli expressing HA-MkapA or HA-FN3. The complexes formed were then isolated using Ni-NTA resin and analysed by Western blot using anti-HA IgG (Sigma). As shown in Fig. 7A, MkapA formed a complex with Pkn4CD (lanes 1 and 3), but not with Pkn4RD (lanes 2 and 4). While a weak association of MkapA with Pkn4RD was detected by the yeast two-hybrid screen (Fig. 2 and Table 2), their association may not be strong enough to detect by pull-down analysis. On the other hand, the FN3 region in MkapC formed a complex with Pkn4RD as shown in Fig. 7B (lanes 2 and 4). As MkapA was found to associate with Pkn1m, a kinase-defective mutant, the interaction of MkapA with Pkn4CD may not be dependent on phosphorylation. These results were consistent with those obtained by the yeast two-hybrid screen.
Association of MkapA, B and C with Pkn1, Pkn2, Pkn8 and Pkn9
In order to examine whether the association with MkapA, B and C was unique to Pkn4, we carried out a yeast two-hybrid screen of the M. xanthus genomic library with four other membrane-associated PSTKs, Pkn1, Pkn2, Pkn8 and Pkn9, as bait (Fig. 1A). Interestingly, all PSTKs tested were found to associate with MkapA, while Pkn8 and Pkn9 associated with MkapB, and Pkn8 with MkapC (Table 1). Pkn8 contains two hydrophobic sequences probably functioning as a transmembrane segment in between CD and RD. Therefore, Pkn8 seems to expose both domains in the cytoplasm. Based on the analysis using the Prosite database (http://hits.isb-sib.ch/cgi-bin/PFSCAN), the RD of Pkn8 shows similarity with the kinesin light chain (KLC) and is predicted to contain five tandem repeats of TPR. The TPR domain is known to interact with TPR domains in other proteins (D’Andrea and Regan, 2003), thus the Pkn8RD and MkapB association was examined using the yeast two-hybrid screen. Pkn8RD was found to interact with MkapB (data not shown). Furthermore, Pkn8 was found to interact with Pkn14, a new member of the PSTK family that seems to be a cytoplasmic as it does not contain the predicted hydrophobic sequence in the RD (Fig. 1B).
Pkn9 was also found to interact with another interesting protein, termed Pkn9associated protein 1 (K9ap1). Pkn9 is a transmembrane PSTK having a CD at the N-terminus in the cytoplasm and a C-terminal RD in the periplasmic space. Deletion of the pkn9 gene causes severe reduction in development progression resulting in 40% of the spore yield of the parent strain, DZF1 (Hanlon et al., 1997). K9ap1 contains a fork-head associated (FHA) domain (Durocher et al., 2000) at the N-terminal end followed by a putative hydrophobic sequence (Fig. 1B), suggesting that K9ap1 is a transmembrane protein having the N-terminal FHA domain in the cytoplasm and the C-terminal region in the periplasmic space.
Cellular signalling requires constant assembly and disassembly of protein complexes depending on the environmental conditions. In Bacillus subtilis, entry into sporulation is tightly controlled by the multicomponent His-Asp phosphorelay system involving multiple histidine kinases, KinA, B, C and D, to Spo0F, Spo0B and finally the response regulator Spo0A (reviewed by Burkholder and Grossman, 2000). When initiation of DNA replication is perturbed, the onset of sporulation is prevented at the first step in the relay by inhibiting the accumulation of phosphorylated KinA through Sda (Burkholder et al., 2001). As Sda binds to the KinA autokinase domain, Sda forms a molecular barricade that prevents interdomain communication between the phosphorylatable His residues and the ATP binding site of KinA (Rowland et al., 2004). On the other hand, in eukaryotes, it is known that PSTKs and PTKs regulate various cellular functions via protein–protein interactions in a phosphorylation-dependent manner (reviewed by Hunter, 2000). Although M. xanthus contains 99 PSTKs, such adaptor/modulator proteins for PSTKs had not yet been identified. To find a factor(s) that binds and modulates PSTK function, we performed yeast two-hybrid screens using membrane-associated PSTKs as bait. When the yeast two-hybrid screen was performed using the M. xanthus genomic library with Pkn4 as bait, three factors, MkapA, MkapB and MkapC, containing well-characterized protein–protein interaction domains were identified. In this study, MkapB was demonstrated to associate with the Pkn4RD upon phosphorylation by Pkn4 and to remain (Figs 2–4). As a result, the activation of PFK was inhibited based on interference with PFK and Pkn4 interaction. Therefore, MkapB functioned as a modulator for glycogen accumulation by suppressing the activation of PFK in a phosphorylation-dependent manner in the stationary phase and in early development, periods with exhibiting mkapB expression. As mkapB expression decreased at the appearance of sonication-resistant myxospores (Fig. 6A), PFK was activated by phosphorylation with Pkn4 and consequently, accumulated glycogen was consumed for effective sporulation (Nariya and Inouye, 2003). This is the first demonstration of a PSTK cascade in prokaryotes, which modulates a specific biological function.
MkapB contains eight tandem repeats of TPR and was found to interact with Pkn8 and Pkn9. The TPR domain exists in a variety of organisms, from bacteria to human, and is involved in various cellular functions (D’Andrea and Regan, 2003). Comparison of TPRs from a variety of proteins reveals eight loosely conserved consensus residues which form the hydrophobic surfaces on two α-helices of TPR (Lamb et al., 1995). Based on the three-dimensional structure of several TPRs determined by NMR and X-ray crystallography (see review by D’Andrea and Regan, 2003), TPR domains are known to interact with TPR proteins and also non-TPR proteins. TPR–TPR interactions are mediated by the hydrophobic surfaces formed by the eight consensus residues. TPR–non-TPR protein interactions are not limited to the consensus residues in the α-helices but can also involve interactions with residues in the turns between two helices of a single TPR and between adjacent TPRs. Although Pkn4, Pkn8 and Pkn9 associated with MkapB, there is no detectable similarity in their RDs. The KLC-like domain of Pkn8RD containing five tandem repeats of TPR associated with MkapB. Therefore, it is most likely that Pkn8 and MkapB interaction is mediated by their TPR domains. On the other hand, the P-rich sequence was found in the Pro, Ala, Thr and Gly (PATG)-rich region (from 384 to 480) of Pkn4 and in the PA-rich region (from 427 to 554) of Pkn9, downstream of their CDs. P-rich motives are known to play critical roles in the assembly and regulation of many signalling complexes by binding SH3 and WW domains (Lu et al., 1999; Zarrinpar et al., 2003). Therefore, P-rich sequences of Pkn4 and Pkn9 may play a role in the interaction with MkapB. Note that Pkn1 and Pkn2 that did not interact with MkapB contain neither a TPR nor a PA-rich region.
The Zn-finger of MkapA consists of CX3CX9HX3H with nine residues in a finger loop. The Zn-fingers function as molecular recognition elements and are extremely common protein domains as perhaps 1% of all mammalian genes encode Zn-finger proteins (Mackay and Crossley, 1998). The typical C2H2, C3H and GATA-like C4 fingers have been known to interact with DNA and RNA in a sequence-specific manner and also to associate with proteins (Mackay and Crossley, 1998). MkapA was found to associate with the CDs of Pkn4 (Figs 2 and 7) and Pkn8 (data not shown) using the yeast two-hybrid screen. As MkapA is a small protein consisting of 151 residues with a Zn-finger, the association of MKapA with the CDs of PSTKs is most probably mediated by its Zn-finger.
On the other hand, MkapC contains three repeats of the FN3-like domain. Repeats of this domain mediate fibronectin polymerization. The FN3-like domain is a ubiquitous one in mammalian proteins that include cell adhesion molecules, cell surface hormone and cytokine receptors, chaperonins, and carbohydrate-binding domains (Harpaz and Chothia, 1994). Although three repeats of FN3-like domain in MkapC were sufficient for the association with Pkn4RD and Pkn8RD (data not shown), the association domains of Pkn4 and Pkn8 remain to be demonstrated. Besides the FN3-like domains, the sequence of the C-terminal region of MkapC has weak similarity to hepalinase II. Therefore, it is tempting to speculate that the hepalinase II-like activity of MkapC may be regulated through its phosphorylation by Pkn4 or Pkn8.
In addition to Mkaps, Pkn9 was found to associate with a putative transmembrane protein, K9ap1, which has an FHA domain at the N-terminal region. This domain has been shown to be a recognition module for phospho-Thr and Tyr motifs and to be involved in numerous processes including intracellular signal transduction, transcription, protein transport, DNA repair and protein degradation in both eukaryotes and prokaryotes (Durocher et al., 2000). Although FHA domains consisting of roughly 75 residues do not exhibit extensive sequence similarity, they do share similar secondary and tertiary structures (Li et al., 2000). Therefore, it is likely that Pkn9 with the phospho-Thr motif may form a heterodimer with K9ap1 via the FHA domain when Pkn9 is activated by ligand binding, and thus function as a modulator or an adapter molecule in the Pkn9 signalling pathway.
Interestingly, Pkn8 was found to associate with a new member of PSTK, Pkn14, that was predicted to be a cytoplasmic PSTK from its sequence (Fig. 1B). As Pkn8 was found to phosphorylate a kinase-deficient mutant of Pkn14 in vitro (data not shown), it is possible that Pkn14 is a substrate of Pkn8 and is the PSTK downstream of Pkn8. This would indicate that Pkn8 and Pkn14 form a kinase cascade in the Pkn8 signalling pathway similar to that of a receptor-type PTK cascade in eukaryotes (see review by Schlessinger, 2000). Beside Pkn14, Pkn8 was also found to associate with three other targets (Table 1). Based on the blast search, K8ap1 was found to be identical to FruD, which is involved in cell division (Akiyama et al., 2003). K8ap2 and K8ap3 were homologous to the Enzyme I component in the phosphotransferase system (PTS) and E. coli KpsF that is required for polysialic acid capsule expression (Cieslewicz and Vimr, 1997) respectively.
Based on the results of this study, we propose that MkapA, B and C probably modulate PSTK functions by forming a network as depicted in Fig. 8. As MkapA associated with the CDs of Pkn4 and Pkn8, MkapA may function as a regulator for the auto-kinase activity of PSTKs or as an adapter molecule for recruiting downstream factors in their signalling pathways. MkapB, consisting of eight tandem repeats of TPR domain, also functions as an adapter molecule for recruiting factors in Pkn8 and Pkn9 signalling pathway. The elucidation of the roles of MkapB in the phosphorylation of Pkn14 and K9ap1 by Pkn8 and Pkn9 and the identification of downstream factors by yeast two-hybrid screens will provide further insights into the complex PSTK networks along with their roles in the M. xanthus life cycle.
Myxococcus xanthus libraries for yeast two-hybrid screen
The genomic DNA (200 µg) prepared from M. xanthus DZF1 was partially digested with AciI, HinP1I and MspI, and fragments in length of 0.5–2.5 kb for each digestion were isolated with use of 0.7% agarose gel electrophoresis. The isolated fragments were ligated with pGAD-C1, -C2 and -C3 (James et al., 1996) at the ClaI site and nine ligated mixtures were electroporated into E. coli DH 10B (Gibco-BRL). Transformation efficiency for each mixture was about 1 × 107 colonies per 1 µg of DNA. More than 95% of the transformants were found to contain the M. xanthus genomic DNA fragments (data not shown). Because the genome size of M. xanthus is approximately 9140 kb (http://tigrblast.tigr.org), 1 × 107 colonies from each transformation were harvested. Three cultures with the same vector (e.g. AciI, HinP1I or MspI fragments in pGAD-C1) were pooled together. Plasmids were isolated from the pooled cultures by the method described by Birnboim and Doly (1979) with addition of the polyethlene glycol treatment and the resulting libraries of the three vectors were referred to as pGAD-Mx1, pGAD-Mx2 and pGAD-Mx3.
Constructions of pGBD-PSTKs and its expression in the yeast PJ69-4A strain
The yeast GAL4 fusion vectors, pGAD-NdeI and pGBD-NdeI were constructed because it makes it easy to shuttle the genes used as bait from pET vector in which the pkn1, pkn2, pkn4, pkn9 and pfk genes had been cloned in the NdeI–BamHI or NotI sites. The multicloning sites between EcoRI and BglII of pGAD-C1 and pGBD-C1 were modified to NdeI–ClaI–BamHI–NotI by introducing NdeI-linker listed in Table 4 at the EcoRI and BglII sites of both plasmids. As a result, the NdeI and BamHI or NotI sites of pGAD-NdeI and pGBD-NdeI can be used to construct the fusion genes with GAL4–AD or GAL4–BD. The NdeI–BamHI or NotI fragments of pkn1 (Munoz-Dorado et al., 1991), pkn2 (Udo et al., 1995), pkn9 (Hanlon et al., 1997), and pkn4, pkn4CD, pkn4RD and pfk (Nariya and Inouye, 2002) genes from pET vector were cloned into pGBD-NdeI or pGAD-NdeI resulting in pGBD-Pkn1, -Pkn2, -Pkn9, -Pkn4, -Pkn4CD, -Pkn4RD, -PFK and pGAD-PFK respectively. pGBD-Pkn8, pGAD-MkapA, -MkapB and MkapC were constructed using NdeI–BamHI or NotI fragments amplified by polymerase chain reaction (PCR) using the primers, Pkn8-N and Pkn8-C, MkapA-N and MkapA-C, MkapB-N and MkapB-C, and MkapC-N and MkapC-C, respectively (Table 4), and the chromosomal DNA as a template. All of the PCR-amplified fragments were confirmed by DNA sequence before they were used.
Table 4. The linker and primers used in study.
Bold letters indicate the translational initiation and stop codons. Underlined sequences represent restriction enzyme sites.
A yeast two-hybrid screen was performed by the method described by James et al. (1996). The yeast strain, PJ69-4A, contains three reporter genes, GAL1–HIS3, GAL2–ADE2 and GAL7–lacZ (Table 3). PJ69-4A harbouring pGBD-PSTK was grown to 5 × 108 cells ml−1 in SD-W– medium. Twenty micrograms of each pGAD-Mx1, pGAD-Mx2 and pGAD-Mx3 were transformed into 2.5 × 1010 cells and the cells were plated on SD-W–L– plates. Approximately 107 transformants were obtained for each set of library. After 2 days, approximately 2.5 × 105 transformants growing on SD-W–L– per plate were replica-printed on SD-W–L–Ade– plates. The primary positive colonies were further screened on SD-W–L–His– plates. pGAD plasmid containing positive fragments was isolated and the sequences of the fragment were determined by an ABI Genetic Analyzer 310 using the methods provided by a company. As a control experiment, PJ69-4A harbouring pGBD-PSTK was transformed with an empty vector, pGAD-NdeI. No positive colony was detected in these screens. The expressions of GAL4–BD fusion proteins in yeast strain, PJ69-4A were confirmed by immunoblotting analysis using GAL4–BD antibody (Santa Cruz Biotechnology). As PJ69-4A harbouring pGBD-Pkn1 could not form colonies, a kinase-defective mutant, Pkn1m, that had a substitution of N at invariable K-88 in catalytic subdomain II (Hanks et al., 1988) was used as bait. PJ69-4A harbouring pGBD-NdeI and pGBD-bait were cultured in SD-W– medium (James et al., 1996) to Klett unit 200. The cells harvested were resuspended in SDS-PAGE sample application buffer to be 10 Klett units µl−1, and with the addition of glass beads (425-600 microns, Sigma) boiled in vigorous vortexes for 3 min. After centrifugation at 10 000 g for 1 min, supernatants were applied onto SDS-PAGE and fusion proteins were detected by immunoblotting analysis as previously described (Nariya and Inouye, 2003).
Deletion of 3.0 kb SmaI fragment containing Mx8-attP from pZKAT
IPTG-inducible expression vector with Kmr
N-terminal HA-tagged protein expression vector constructed by insertion of HA-linker at XbaI and NdeI site in pINIII(NdeI)Km
5439 bp BamHI–EcoRI fragment in pUC19
5581 bp EcoRI–SalI fragment in pUC19
7642 bp EcoRI–SphI fragment in pUC19
4304 bp BamHI–SalI fragment in pUC19
1578 bp AciI fragment obtained by Y2H using pGBD-Pkn9 as bait
2.1 kb NdeI–BamHI fragment from pET-Pkn1m in pGBD-NdeI
1.9 kb NdeI–NotI fragment from pET-Pkn4 in pGBD-NdeI
1.1 kb NdeI–NotI fragment from pET-Pkn4CD in pGBD-NdeI
0.8 kb NdeI–NotI fragment from pET-Pkn4RD in pGBD-NdeI
3.1 kb NdeI–BamHI fragment in pGAD-NdeI
2.3 kb NdeI–BamHI fragment from pET-Pkn9 in pGBD-NdeI
1.1 kb NdeI–NotI fragment from pET-PFK in pGAD-NdeI
0.45 kb NdeI–BamHI fragment in pGAD-NdeI
0.8 kb NdeI–NotI fragment from pET-MkapB in pGAD-NdeI
3.5 kb NdeI–BamHI fragment in pGAD-NdeI
0.9 kb Nde–BamHI fragment in pGAD-NdeI
2607 bp NruI–BsaAI fragment from pGMkapB at the SmaI
site of pUC19
0.7 kb BstBI–SacI fragment in pMkapB was replaced with Kmr
500 bp HindIII–BamHI fragment of the mkapB promoter region in pZK
MkapB expression vector containing 0.8 kb NdeI–NotI fragment from pGBD-MkapB in pET16b
0.45 kb NdeI–BamHI fragment from pGAD-MkapA in pINK-HA
0.9 kb NdeI–BamHI fragment from pGAD-FN3 in pINK-HA
Cloning the genomic DNA fragments containing the mkapA, B, C and pkn14 genes, and their nucleotide sequence accession numbers
The genomic DNA fragments containing mkapA, B, C and pkn14 were isolated from restricted genomic DNA fragments by Southern blot hybridization using the positive clones of yeast two-hybrid screen as probes. The probes were labelled by DIG-Chem-Link labelling and Detection Set following the procedure described in the kit (Boehringer Mannheim). The fragments harbouring mkapA, B, C and pkn14 genes were 5439 bp of BamHI–EcoRI, 5581 bp of EcoRI–SalI, 7642 bp of EcoRI–SphI and 4304 bp of BamHI–SalI fragments respectively. These were cloned into pUC19 resulting in pGMkapA, pGMkapB, pGMkapC and pGPkn14. Their DNA sequences have been determined and submitted to the GenBank DNA sequence database under Accession No. AF377338, AF377339, AF377340 and AF377337 respectively. The DNA sequence of the positive clone containing the k9ap1 gene is also submitted as AY569568.
Construction of a ΔmkapB strain
As shown in Fig. 6A, the 0.7 kb BstBI–SacI region in pMkapB was replaced with the 1.3 kb HincII fragment containing the kamamycin-resistant gene (Kmr) from pUC7Km(P-) (Norioka et al., 1995) in an opposite direction against the mkapB transcription. The resulting plasmid is named pMkapB-KR. pMkapB consists of the 2607 bp NruI–BsaAI fragment from pGMkapB at the SmaI site of pUC19. After digestion with ScaI, pMkapB-KR was introduced into DZF1 by electroporation, and the kanamycin-resistant colonies were screened for double crossing-over by colony hybridization using the pUC19 DNA as a probe and the deletion strains were confirmed by Southern blot analysis as described previously (Nariya and Inouye, 2003). The resulting strain was termed ΔmkapB.
Construction of a mkapBP–lacZ/DZF1 strain
The promoter region of mkapB from −331 to +202 was amplified by PCR using pGMkapB and the primers, MkapB-P5 and MkapB-P3 shown in Fig. 5B and Table 4. The PCR product digested with HindIII and BamHI was cloned into pZK (Table 3) resulting in the lacZ transcriptional fusion plasmid, pmkapBP–lacZ. pmkapBP–lacZ was electroporated into DZF1 and integrated at the original location by a single crossing-over. The recombination site of pmkapBP–lacZ was confirmed by PCR using MkapB-5 located 33 bases upstream of MkapB-P5 in the mkapB promoter region and LacZ-3FSC in pZK as primers (Fig. 5B and Table 4). The mkapBP–lacZ/DZF1 is the lacZ expression strain under the control of the promoter of mkapB in DZF1.
Purification of N-terminal His-tagged MkapB
All column chromatography was performed at 4°C using FPLC system LCC-500 Plus (Amersham Pharmacia Biotech). The columns used are purchased from Amersham Pharmacia Biotech. pET-MkapB, the expression vector of MkapB, was constructed by cloning the 0.8 kb NdeI–NotI fragment from pGBD-MkapB into pET16(+) (Novagene). MkapB was purified from the soluble fraction of E. coli BL21 (DE3) harbouring pET-MkapB by the method described for Pkn4 purification (Nariya and Inouye, 2002). The MkapB fraction eluted with 300 mM imidazole from Ni-NTA SUPER FLOW resin (Qiagen) was loaded on the HiTrap Q column and was eluted at 370 mM NaCl with a linear gradient from 0 to 1.0 M. The MkapB fraction was dialysed against buffer A [50 mM Tris-HCl, (pH 8.0) containing 100 mM NaCl, 2 mM DTT and 1 mM MgCl2] and after dilution to 0.5 mg ml−1 with buffer A, stored at −80°C.
An HA-tagged protein expression vector, pINK-HA, was constructed by inserting the HA-linker (Table 4) between the XbaI and NdeI sites in pINIII(NdeI)Km in which the NdeI site was created at the translation initiate site of pINIIIKm (Ghrayeb et al., 1984). The DNA fragment coding FN3 region from 208 to 516 in Fig. 1B was amplified by PCR using primers FN3-N and FN3-C (Table 4). The PCR fragments digested with NdeI and BamHI and cloned into pGAD-NdeI. The resulting plasmid was named pGAD-FN3. The NdeI–BamHI fragments of pGAD-MkapA and pGAD-FN3 were isolated and cloned into pINK-HA resulting in the plasmids, pINK-HA-MkapA and pINK-HA-FN3 respectively. HA-tagged MkapA (HA-MkapA) and FN3 (HA-FN3) were expressed in E. coli SB221 (Nakamura et al., 1982) by addition of 2 mM IPTG for 4 h. The cells were suspended at 1000 Klett ml−1 in Pull-down buffer [50 mM Na-phosphate (pH 8.0) containing 10% (v/v) glycerol, 100 mM NaCl, 40 mM imidazole and 2 mM 2-mercaptoethanol]. After the cells were disrupted by sonication, the soluble fractions were prepared by centrifugation at 105 000 g for 30 min. After 10 µg of purified His-tagged Pkn4CD or Pkn4RD (Nariya and Inouye, 2002) was added to 1 ml of the soluble fractions, the mixtures were incubated for 2 h at 4°C. Ten microlitres of Ni-NTA SUPER FLOW resin were added to each mixture and incubated for 1 h. The resin containing the complexes was washed three times with 1 ml of Pull-down buffer then boiled in 100 µl of 1× SDS buffer for 1 min. Twenty microlitres and 1 µl of samples were separated on a 10% SDS-PAGE, and visualized by CBB stain and Western blot analysis using anti-HA IgG (Sigma). When the soluble fraction from cells harbouring pINK-HA was treated with Pkn4CD or Pkn4RD, no specific proteins was detected.
Microbial strains, growth conditions and plasmids
Microbial strains and plasmids used in this study are listed in Table 3. M. xanthus strains were cultured in CYE medium (Campos et al., 1978) at 30°C and supplemented with 80 µg ml−1 kanamycin when necessary. Development of M. xanthus cells was induced on TM [10 mM Tris-HCl, (pH 7.6), 8 mM MgSO4] agar plates as described by Nariya and Inouye (2003). E. coli were grown in LB (Miller, 1972) or M9 (Maniatis et al., 1989) medium supplemented with 100 µg ml−1 ampicillin or 40 µg ml−1 kanamycin when necessary.
Phosphorylation and immunoprecipitation of Pkn4, PFK and MkapB were performed in P-buffer [50 mM Tris-HCl (pH 8.0), 10 mM MgCl2, 2 mM DTT] with and without 200 µM ATP and 1.2 µCi of [γ-32P]-ATP by the methods described previously (Nariya and Inouye, 2002; 2003). Phosphoamino acid analysis was carried out by the method described by Munoz-Dorado et al. (1991). The primer extension analysis was carried out using MkapB-AS as a primer by the method described previously (Nariya and Inouye, 2003). β-Galactosidase assays were carried out by the methods described by Kroos et al. (1986). The clear lysates were obtained from the cells that were harvested at indicated time points during vegetative growth and development.
We are grateful to M. Inouye for discussions and critical reading of this manuscript. We thank M. Inouye, T. Ueki and L. Vales for critical reading of the manuscript. This work was supported by the Foundation of Medicine and Dentistry of New Jersey.