In most firmicutes expression of the mannitol operon is regulated by MtlR. This transcription activator is controlled via phosphorylation of its regulatory domains by components of the phosphoenolpyruvate : carbohydrate phosphotransferase system (PTS). We found that activation of Bacillus subtilis MtlR also requires an interaction with the EIIBMtl domain of the mannitol permease MtlA (EIICBMtl). The constitutive expression of the mtlAFD operon in an mtlF mutant was prevented when entire mtlA or only its 3′ part (EIIBMtl) were deleted. Yeast two-hybrid experiments revealed a direct interaction of the EIIBMtl domain with the two C-terminal domains of MtlR. Complementation of the Δ3′-mtlA ΔmtlF or ΔmtlAFD mutants with mtlA restored constitutive MtlR activity, whereas complementation with only 3′-mtlA had no effect. Moreover, synthesis of EIIBMtl in strains producing constitutively active MtlR caused MtlR inactivation. Interestingly, EIIBMtl fused to the trans-membrane protein YwqC restored constitutive MtlR activity in the above mutants. Replacing the phosphorylatable Cys with Asp in MtlA or soluble EIIBMtl lowered MtlR activation, indicating that MtlR does not interact with phosphorylatyed EIIBMtl. Induction of the B. subtilismtl operon therefore follows a novel regulation mechanism where the transcription activator needs to be sequestered to the membrane by unphosphorylated EIICBMtl in order to be functional.
D-Mannitol is a hexitol widely used as low-calorie sweetener in food industry. It is present in large quantities in fungi, algae and plants (Oddo et al., 2002). Many enterobacteria (Jacobson et al., 1983) and firmicutes (Reiche et al., 1988; Deutscher et al., 1994; Henstra et al., 1996) developed the capacity to use this polyol as carbon and energy source. Bacteria usually take up mannitol via the phosphoenolpyruvate (PEP) : carbohydrate phosphotransferase system (PTS), which phosphorylates mannitol to mannitol-1-P during the transport process. Mannitol-1-P dehydrogenase, which is encoded by the mtlD gene, converts intracellular mannitol-1-P into fructose-6-P, which is subsequently metabolized via glycolysis.
The PTS is usually composed of three sugar-specific domains or proteins (EIIA, B and C) and the two general components enzyme I (EI) and HPr (Deutscher et al., 2006). The two latter proteins form a phosphorylation cascade together with the EIIA and EIIB components. In the first step, EI autophosphorylates with PEP and transfers the phosphoryl group to His-15 in HPr. In firmicutes, P∼His-HPr is the only phosphoryl donor for usually several sugar-specific EIIAs. The EIIAs transfer their phosphoryl group to an EIIB with the same sugar specificity and P∼EIIB phosphorylates the carbohydrate bound to the corresponding membrane-integral EIIC. In the last step the phosphorylated carbohydrate is released into the cytoplasm.
In firmicutes, the mannitol-specific PTS is usually composed of a distinct EIIAMtl and a fused EIIC–EIIB component (Reiche et al., 1988; Reizer et al., 1992). Accordingly, the mtl operon contains the mtlA, mtlF and mtlD genes, with mtlA encoding EIICBMtl and mtlF the EIIAMtl protein. In Escherichia coli and other proteobacteria mtlA (encodes an EIICBAMtl) and mtlD are followed by mtlR, which encodes a small protein of about 20 kDa proposed to function as co-repressor (Tan et al., 2009). However, it was recently reported that expression of Vibrio cholerae mtlA is regulated by a small RNA (MtlS sRNA) transcribed antisense to the 5′ untranslated region of the mtl operon (Mustachio et al., 2012). MtlR is thought to regulate the amount of MtlS in response to the absence or presence of mannitol. Firmicutes contain a different mtlR gene that is usually located between mtlA and mtlF. Bacillus subtilis mtlR is an exception, because it is located about 15 kb downstream from the mtl operon (Watanabe et al., 2003). MtlR of firmicutes is a transcription activator of about 75 kDa that binds to an operator site upstream from the mtl promoter (Henstra et al., 1999). It is a member of a large group of regulatory proteins which are composed of an N-terminal DNA-binding domain followed usually by four regulatory domains (Deutscher et al., 2006). MtlR contains two PTS regulation domains (PRDs), an EIIBGat- (Gat = galactitol) and an EIIAMtl-like domain. The activity of MtlR is regulated via phosphorylation by PTS proteins. MtlR of Geobacillus stearothermophilus was reported to be phosphorylated in vitro by EI and HPr at conserved histidyl residues in PRD1 and PRD2 (Henstra et al., 1999). Phosphorylation in PRD2 increased and phosphorylation in PRD1 lowered the affinity of MtlR for the mtl operator site, which precedes the promoter and which is conserved in B. subtilis (Henstra et al., 2000). Phosphorylation of the EIIAMtl-like domain required in addition to EI and HPr also EIIAMtl and the EIIBMtl domain of MtlA. This phosphorylation lowered the affinity of MtlR for its DNA target (Henstra et al., 1999). Although MtlR of B. subtilis exhibits 41% sequence identity to G. stearothermophilus MtlR it is quite differently regulated. Phosphorylation by EI and HPr at His-342 in PRD2 (Fig. 1) stimulates B. subtilis MtlR activity. Its absence during growth on repressing sugars probably serves as a catabolite control protein A (CcpA)-independent carbon catabolite repression mechanism (Joyet et al., 2010). B. subtilis MtlR is also phosphorylated in vitro by EI, HPr, EIIAMtl and the EIIBMtl domain of MtlA at His-599 in the C-terminal EIIAMtl-like domain; this phosphorylation has only a slight negative effect on MtlR. The main negative effect is exerted by the EI-, HPr- and EIIAMtl-requiring phosphorylation of cystein 419 in the EIIBGat-like domain. The absence of phosphorylation at Cys-419 leads to induction of the mtl operon. In the presence of mannitol P∼EIIAMtl donates its phosphoryl group primarily to the EIIBMtl domain, which in turn phosphorylates mannitol and Cys-419 in the EIIBGat-like domain is therefore only poorly phosphorylated. The importance of these His or Cys phosphorylation sites for B. subtilis MtlR function was confirmed by measuring the effect of their replacement with Ala or Asp residues, which prevents phosphorylation or has a phosphomimetic effect, respectively, on the expression of a Pmtl–lacZ fusion (Joyet et al., 2010; Heravi et al., 2011). MtlR activity in mutants carrying deletions of various PTS components has also been studied. Deletion of ptsI (encodes EI) (Joyet et al., 2010) or replacement of the phosphorylatable His-15 in HPr with Ala (Heravi et al., 2011) rendered MtlR inactive, confirming that phosphorylation of His-342 in PRD2 by P∼His-HPr is essential. In contrast, deletion of mtlF (encodes EIIAMtl) caused strong constitutive expression of a Pmtl–lacZ fusion (Joyet et al., 2010), confirming the inhibition of MtlR function by P∼EIIAMtl-mediated phosphorylation of Cys-419 in the EIIBGat-like domain.
We here report that the simultaneous deletion of mtlA and mtlF or of mtlF and only the 3′ part of mtlA encoding the EIIBMtl domain (Fig. 1) leads to inactivation of MtlR. EIIBMtl activates MtlR by interacting with the two C-terminal domains. Our results establish that EIIBMtl-mediated membrane sequestration of MtlR is essential for its transcription activator function.
Deletion of B. subtilismtlAFD inhibits MtlR activity
MtlR has previously been shown to become phosphorylated by P∼EIIAMtl at Cys-419 in the EIIBGat-like domain and this phosphorylation inhibits MtlR activity. Deletion of the B. subtilis mtlF gene, which encodes EIIAMtl, therefore leads to strong constitutive, but glucose-repressible expression of a Pmtl–lacZ fusion integrated in the chromosome at the amyE locus (Joyet et al., 2010). In in vitro assays His-599 in the EIIAMtl-like domain of MtlR becomes also slowly phosphorylated by the P∼EIIBMtl domain of MtlA. This phosphorylation has a slight negative effect on MtlR activity because the His-599-Ala replacement caused weak constitutive expression of the Pmtl–lacZ fusion (Joyet et al., 2010). In order to study whether the EIIBMtl domain of MtlA has an additional role in MtlR regulation we constructed a mutant carrying a deletion of the entire mtlAFD operon. Based on the above results, we expected that the mtlAFD deletion mutant would exhibit strong constitutive expression of a Pmtl–lacZ fusion, because similar as in the mtlF mutant phosphorylation at both negative regulatory sites is prevented. Unexpectedly, MtlR activity in the triple deletion mutant was very low when grown in the presence or absence of mannitol (Fig. 2A and B).
Deletion of the EIIB domain of MtlA also inhibits B. subtilis MtlR activity
We suspected that the loss of constitutive lacZ expression owing to the additional deletion of mtlA and mtlD in the ΔmtlF mutant was probably due to the loss of the EIIBMtl or EIICMtl domain of MtlA. In order to distinguish between these two possibilities we deleted the 3′ part of the mtlA gene encoding the EIIBMtl domain in the previously described ΔmtlF mutant PJ027 (Joyet et al., 2010). In contrast to the parental strain, which expresses the Pmtl–lacZ fusion constitutively, the resulting Δ3′-mtlA ΔmtlF double mutant exhibited very low β-galactosidase activity independently of whether it was grown in the presence or absence of mannitol (Fig. 2A). In order to confirm the results of the β-galactosidase assays we also carried out quantitative RT-PCR (qRT-PCR) experiments with the wild-type strain and the mutants. As expected, the wild-type strain grown in LB medium in the absence of mannitol produced low levels of lacZ mRNA, which strongly increased when mannitol was included in the LB medium (Fig. 2B). The ΔmtlF mutant grown in the absence of mannitol also produced high levels of lacZ mRNA. In contrast, little lacZ mRNA was detected in the Δ3′-mtlA ΔmtlF double mutant and the ΔmtlAFD triple mutant (Fig. 2B). These results suggested that in order to be active MtlR needs a functional EIIBMtl domain, which stimulates the expression from the Pmtl promoter and does not affect a post-transcriptional regulatory step. In the ΔmtlF mutant the EIIBMtl domain is probably not phosphorylated owing to the absence of EIIAMtl. It is therefore likely that MtlR activation requires the unphosphorylated form of the EIIBMtl domain, although an interaction with phosphorylated EIIBMtl cannot be excluded.
We also constructed a strain, in which only the EIIBMtl domain was deleted. In this mutant, MtlR is most likely phosphorylated by EIIAMtl at the EIIBGat domain and thus inactivated (Joyet et al., 2010). In addition, induction by mannitol was not possible owing to the absence of EIIBMtl. Indeed, the Δ3′-mtlA mutant exhibited very low β-galactosidase activity in the absence and presence of mannitol (Fig. 2A).
The EIIBMtl domain interacts with the two fused C-terminal domains of MtlR
To test whether the stimulatory effect of the EIIBMtl domain on MtlR function was mediated by direct interaction of the two proteins we carried out yeast two-hybrid experiments. The entire B. subtilis mtlR gene and the 3′ part of mtlA encoding the EIIBMtl domain were cloned into plasmids pGAD and pGBDU and interaction studies were subsequently carried out by screening for growth on selective media (see Experimental procedures). A positive growth signal was obtained for the pGAD-mtlR/pGBDU-mtlR combination (Fig. 3A), indicating that the transcription regulator MtlR interacts with itself by forming oligomers, most likely dimers, as reported for other PRD-containing regulators. A positive signal was also obtained for the pGAD-mtlR/pGBDU-3′-mtlA and pGAD-3′-mtlA/pGBDU-mtlR combinations, suggesting a direct interaction between MtlR and the EIIBMtl domain of MtlA (Fig. 3A). In order to identify the domains of MtlR that interact with EIIBMtl, we either cloned each regulatory domain of MtlR separately into both yeast two-hybrid vectors or fused them to their upstream or downstream regulatory domain. Subsequent protein interaction studies revealed that the EIIBMtl domain of MtlA specifically binds to the fused C-terminal EIIBGat-like/EIIAMtl-like domains of MtlR (Fig. 3B), but not to the single domains. MtlR also interacts with the two fused domains, suggesting that they contribute to MtlR oligomerization (Fig. 3).
Complementation of the ΔmtlAFD and the Δ3′-mtlA ΔmtlF double mutant with EIICBMtl and EIIBMtl
To confirm the importance of the EIIBMtl domain for MtlR function we cloned the B. subtilis xyl promoter and the Shine–Dalgarno (SD) box of xylA (Gärtner et al., 1992) as well as the entire mtlA gene into the integrative plasmid pDG1664 (Guérout-Fleury et al., 1996). In the resulting construct mtlA is expressed under control of the xyl promoter. The ectopic integration of this plasmid into the genome of the Δ3′-mtlA ΔmtlF double mutant or the mtlAFD deletion mutant indeed restored constitutive expression of the Pmtl–lacZ fusion observed in the ΔmtlF mutant (Fig. 4). Integration of an identical plasmid carrying an allele encoding non-phosphorylatable Cys396Ala mutant MtlA led to similar MtlR activity, whereas the presumed phosphomimetic replacement of Cys-396 with an Asp significantly lowered β-galactosidase activity (Fig. 4). Surprisingly, when an identical integration was carried out with a pDG1664-derived plasmid carrying only the 3′-mtlA fragment encoding the EIIBMtl domain β-galactosidase activity in the ΔmtlAFD mutant remained very low. The same result was obtained when the 3′-mtlA-containing plasmid was integrated into the chromosome of the Δ3′-mtlA ΔmtlF double mutant (Fig. 5). Even more surprising, when this plasmid was integrated into the chromosome of strain PJ001, in which MtlR and the mannitol operon are intact, it prevented mannitol-mediated induction of the expression of the Pmtl–lacZ fusion (Fig. 5). Moreover, the constitutive synthesis of β-galactosidase in strains PJ007 (mtlRH342D), PJ031 (ΔmtlF, mtlRC419A) (Joyet et al., 2010) and HB41 (ΔmtlF), was prevented when these mutants were transformed with the pDG1664PxylA-3′-mtlA plasmid and grown in the presence of xylose, which leads to the synthesis of soluble, cytoplasmique EIIBMtl (Fig. 5). In conclusion, the above results establish that only EIIBMtl fused to EIICMtl can stimulate MtlR activity and that in contrast soluble EIIBMtl inhibits it.
EIIBMtl fused to the integral membrane protein YwqC restores MtlR activity
One possible explanation for these unexpected results was that EIIBMtl needs to sequester MtlR to the membrane in order to render it active. This activation might be prevented by the overproduction of soluble EIIBMtl, which probably competes with the membrane-attached EIIBMtl domain of MtlA for MtlR binding. In order to test this hypothesis, we cloned the B. subtilis ywqC gene lacking the 3′ part encoding the last 34 amino acids, which functions as modulator of the tyrosine kinase YwqD (PtkA) (Mijakovic et al., 2003), into the B. subtilis replicative plasmid pBQ200 (Martin-Verstraete et al., 1994). The cloned DNA fragment codes for a truncated YwqC protein composed of its two trans-membrane helices and a C-terminal cytoplasmic tail composed of 15 amino acids. By using the His-tag encoded by the pQE30 plasmid as linker the 3′-mtlA fragment was fused to the truncated ywqC gene and the resulting plasmid was used to transform the 3′-mtlA mtlF double-deletion mutant. Expression of the ywqC::3′-mtlA fusion from the constitutive PdegQ36 promoter in pBQ200 indeed restored full constitutive expression of the Pmtl–lacZ fusion in the Δ3′-mtlA ΔmtlF double mutant (Fig. 6). In contrast, very low MtlR activity was observed when the Δ3′-mtlA ΔmtlF double mutant was transformed with pBQ200 or pBQ200 carrying only the DNA fragment coding for the EIIBMtl domain. In order to test whether phosphorylation of the EIIBMtl domain would have an influence on MtlR activation the phosphorylatable Cys-8 of the EIIBMtl domain (corresponds to Cys-396 in MtlA) was replaced with an Ala or Asp residue in the YwqC–EIIBMtl fusion protein. β-Galactosidase activity similar to that observed for wild-type EIIBMtl fused to YwqC was obtained when Cys-8 in the EIIBMtl domain was replaced with an Ala. In contrast, the Δ3′-mtlA ΔmtlF mutant transformed with the pBQ200-derived plasmid carrying the Cys-8-Asp mutant allele exhibited 3.2-fold lower MtlR activity (Fig. 6).
Expression of the ywqC::3′-mtlA fusion in the Δ3′-mtlA ΔmtlF mutant probably also restores the synthesis of the membrane-integrated EIICMtl domain of MtlA. It was therefore possible that the EIICMtl domain was necessary for MtlR activation. In order to test this possibility, the pBQ200-derived plasmid carrying the ywqC::3′-mtlA fusion was also used to transform the ΔmtlAFD mutant. In the triple mutant lacking also the EIICMtl domain of MtlA synthesis of the YwqC/EIIBMtl or the YwqC/EIIBMtlCys8Ala fusion protein restored expression of the Pmtl–lacZ fusion to more than half the level observed for the complemented Δ3′-mtlA ΔmtlF double mutant (Fig. 6). In conclusion, these results suggest that the EIICMtl domain plays only a minor role in B. subtilis MtlR activation. Most of the 15- to 20-fold activation of MtlR in the YwqC/EIIBMtl complementation experiments owes to sequestration of the transcription regulator to the cytoplasmic membrane.
Phosphorylation of EIIBMtl reduces MtlR activation
The reduced activity of MtlR observed in strains producing EIICBMtl(Cys396Asp) (Fig. 4) or YwqC/EIICBMtl(Cys8Asp) (Fig. 6) suggested that phosphorylation of the transport protein prevents MtlR activation. In order to further support this concept we replaced the phosphorylatable Cys in the EIIBMtl domain with a serine. The resulting mutant EIIBMtl is phosphorylated at this seryl residue by EI, HPr and EIIAMtl (Otten et al., 2004; Opacić et al., 2010). To test whether cytoplasmic EIIBMtl(Cys8Ser) is also able to inhibit MtlR we cloned the corresponding allele into pDG1644-Pxyl. The resulting plasmid was used to transform strain ΔmtlF, in which EIIBMtl(Cys8Ser) is not phosphorylated and thus inhibits MtlR activity similar to soluble wild-type EIIBMtl (Fig. 7). To test whether seryl-phosphorylated EIIBMtl(Cys8Ser) would also bind to and inhibit the transcription activator, we transformed the wild-type strain with pDG1644-Pxyl carrying the 3′-mtlA(Cys8Ser) allele. Indeed, while wild-type EIIBMtl strongly inhibited mannitol-induced induction of MtlR activity, EIIBMtl(Cys8Ser) had no detectable effect and the Pmtl–lacZ fusion was as strongly induced as in the non-transformed wild-type strain (Fig. 7). In the wild-type strain producing soluble EIIBMtl(Cys8Ser) mannitol is probably mostly phosphorylated by the EIIB domain of EIICBMtl and owing to its low phosphoryl group transfer potential phosphorylated EIIBMtl(Cys8Ser) accumulates in the cytoplasm. Apparently, P-EIIBMtl(Cys8Ser) is not able to bind to MtlR, which is therefore not trapped in the cytoplasm, but binds to EIICBMtl which allows its activation. These results strongly suggest that phosphorylation of the EIIBMtl domain prevents MtlR activation.
In order to control the expression of genes encoding either PTS components or catabolic enzymes, many bacteria use regulatory proteins containing PRDs. MtlR of firmicutes is a PRD-containing transcription activator required for the expression of the mannitol operon. Phosphorylation of MtlR from G. stearothermophilus (Henstra et al., 1999) and B. subtilis (Joyet et al., 2010) by PTS components and the effects of these phosphorylations on MtlR activity have been extensively studied. We here report an additional new mode of regulation of MtlR from B. subtilis. We observed that the constitutive activity of MtlR in an mtlF mutant (Joyet et al., 2010) was prevented when in addition either the entire EIICBMtl or only its EIIBMtl domain was deleted. MtlR apparently needs EIIBMtl in order to be functional. We could demonstrate that MtlR is activated by direct interaction with EIIBMtl. Yeast two-hybrid experiments revealed that EIIBMtl binds to the C-terminal part of MtlR encompassing both the EIIBGat- and the EIIAMtl-like domains. Interactions between PTS components for phosphoryl group transfer usually last only for very short time periods and they are therefore not detectable by the yeast two-hybrid method, as was shown for EI and HPr of B. subtilis (Poncet et al., 2009). In addition, interaction between the E. coli EIIAMtl and EIIBMtl components did not require additional domains (Suh et al., 2006). It is therefore likely that the interaction between the EIIBMtl domain of MtlA and MtlR observed in the yeast two-hybrid experiments differs from the interaction between these two proteins required for MtlR phosphorylation at His-599 in the EIIAMtl-like domain.
The results obtained by the various complementation assays of the Δ3′-mtlA ΔmtlF and ΔmtlAFD mutant with the EIIBMtl domain revealed that the interaction of MtlR with cytoplasmic EIIBMtl does not stimulate the activity of the transcription regulator but rather inhibits it. Apparently, in order to activate MtlR EIIBMtl needs to be fused to a membrane protein. Under normal conditions the EIIBMtl domain is fused to the membrane-integrated EIICMtl domain. We could demonstrate that fusion of the EIIBMtl domain to another transmembrane protein, the protein kinase modulator YwqC, also restores MtlR activity in the ΔmtlAFD and Δ3′-mtlA mtlF mutants. EIIBMtl-mediated MtlR activation does not depend on the presence of the EIICMtl domain, because strong stimulation of MtlR activity by the YwqC–EIIBMtl fusion occurred also when EIICMtl was lacking.
Dephosphorylation of MtlR at Cys-419 in the EIIBGat-like domain owing to the presence of mannitol in the growth medium was the first induction mechanism established for the B. subtilis mtl operon (Joyet et al., 2010). EIIBMtl-mediated MtlR sequestration represents a second induction mechanism. Our results suggest that P∼EIIBMtl, which prevails when mannitol is absent, does not sequester MtlR and stimulate its activity. The question arises why does B. subtilis use two independent induction mechanisms for the mtl operon? Possibly B. subtilis wants to lock mtl operon expression very tightly. In order to be active, MtlR needs to be phosphorylated at His-342 and dephosphorylated at Cys-419 (Fig. 8). Nevertheless, even in the absence of mannitol Cys-419 in some MtlR molecules remains unphosphorylated and the mtl operon will be slightly expressed. By introducing membrane sequestration by the EIIBMtl domain as a third condition for MtlR activation, even fewer MtlR molecules will be active in the absence of mannitol and consequently the mtl promoter will be almost entirely shut down.
The detailed mechanism of EIIBMtl-mediated membrane sequestration and activation of MtlR is not yet understood. Approaching MtlR to the hydrophobic environment of the cytoplasmic membrane might be sufficient for its activation (Fig. 8). EIIAs of the fructose/mannitol PTS family form dimers (van Montfort et al., 1998) and this is probably also true for the C-terminal EIIAMtl-like domain of MtlR. It is possible that the interaction with the EIIBMtl domain of MtlA breaks up the dimer structure thereby exposing part of the hydrophobic interface, which might interact with hydrophobic parts of the membrane or a yet unknown membrane protein. The resulting structural rearrangements might be transmitted via the EIIBGat-like domain, PRD2 and PRD1 to the N-terminal DNA-binding domain and induce conformational changes leading to elevated affinity of MtlR for its DNA target site (Fig. 8).
Regulation of a PRD-containing protein by membrane sequestration has previously been reported. The E. coli BglG antiterminator, which controls the expression of the β-glucoside operon bgl, interacts with the EIIBBgl domain of the PTS permease BglF (Lopian et al., 2003). However, the mechanism of EIIB-mediated BglG regulation is quite different from that reported here for B. subtilis MtlR. First, BglG sequestration by EIIBBgl leads to inactivation of the antiterminator and not to its activation. Second, in contrast to the mannitol system, the presence of a β-glucoside transported by BglF, such as salicin, prevents the interaction of the EIIBBgl domain with BglG, which is consequently deliberated into the cytoplasm and thereby regains its antitermination activity. Salicin-induced release of BglG from the membrane has been observed after fusing BglG to GFP (Lopian et al., 2003).
A similar membrane sequestration mechanism mediated by the phosphorylation state of an EIIB domain has been reported for E. coli Mlc (Lee et al., 2000; Tanaka et al., 2000; Nam et al., 2001), which is a member of the ROK repressor family and does not contain a PRD. Mlc inhibits the expression of several genes and operons including ptsHI-crr, ptsG and manXYZ. Its activity is regulated by interaction with the EIIBGlc domain of the glucose-specific PTS transporter PtsG. In the presence of glucose, the unphosphorylated form of EIIBGlc prevails. It interacts with Mlc and leads to its membrane sequestration, which inhibits the repressor function of Mlc. Mlc-controlled transcription units are therefore expressed when glucose is present. Similar to MtlR activation, it is the EIIBGlc-mediated membrane localization and not the interaction with the EIIBGlc domain that inactivates Mlc (Tanaka et al., 2004). In the absence of glucose the EIIBGlc domain of PtsG is present mainly in phosphorylated form, which cannot interact with Mlc. The transcription regulator is therefore released into the cytoplasm and represses the expression of its target operons. In the case of maltotriose utilization by E. coli, the ABC transporter MalFGK2 sequesters MalT, the PRD-less activator of the maltose regulon, at the membrane in the absence of its substrate (Richet et al., 2012). This leads to inactivation of MalT and therefore prevents expression of the mal regulon. When transporting its substrate maltotriose the ABC transporter does not sequester MalT.
Finally, some recent findings suggest that the PRD-containing regulator ManR, which functions as activator of the operon encoding the main glucose/mannose PTS transporter EIIAB/EIIC/EIIDMan of Listeria monocytogenes, requires the presence of EIIBMpo (Mpo = mannose permease one) of another glucose/mannose type PTS in order to be functional (Aké et al., 2011). ManR is also controlled by phosphorylation at two of its four regulatory domains (Xue and Miller, 2007). The inhibitory phosphorylation at the PRD2 domain is probably catalysed by EIIBMpo. Deletion of EIIAMpo therefore caused constitutive ManR activity. However, deletion of EIIBMpo or both PTS components caused a loss of ManR function, clearly establishing that ManR requires unphosphorylated EIIBMpo for its activity. Complementation of the ΔmpoA ΔmpoB double mutant with wild-type mpoB or the mpoB(His14Ala) allele expressed from a plasmid restored ManR activity (Aké et al., 2011). L. monocytogenes EIIBMpo is a soluble, cytoplasmic protein. ManR is therefore not activated by membrane sequestration but rather by interaction with cytoplasmic EIIBMpo. In conclusion, the above examples clearly demonstrate that there exists a great variability of EIIB-mediated regulation of transcription activators and repressors.
Bacterial strains and growth conditions
The B. subtilis strains used in this study are listed in Table 1. They were grown in Luria–Bertani (LB) medium or LB medium containing 1% mannitol and appropriate antibiotics. B. subtilis strains containing pDG1644-Pxyl-derived plasmids were grown in LB medium with or without 0.2% xylose. LB medium containing appropriate antibiotics (5 μg ml−1 kanamycin, 4 μg ml−1 chloramphenicol, 100 μg ml−1 spectinomycin or 0.5 μg ml−1 erythromycin) was also used for selection of transformants obtained by electroporation with a Gene Pulser apparatus (Bio-Rad Laboratories). E. coli strain NM522 (Gough and Murray, 1983) was used for cloning experiments with the various plasmids constructed during this study. Selection of E. coli transformants was carried out by using 100 μg ml−1 ampicillin or 50 μg ml−1 kanamycin.
Construction of the B. subtilis Δ3′-mtlA, Δ3′-mtlA/ΔmtlF and ΔmtlAFD mutants
Introduction of the mtlF deletion in strain PJ002 has previously been described (Joyet et al., 2010). An identical approach was used to introduce the mtlF deletion, the 3′-mtlA mtlF double deletion and the mtlAFD triple deletion in strain PJ001. DNA fragments of 0.8 kb corresponding to the upstream and downstream regions of 3′-mtlA/mtlF and mtlAFD were amplified by using B. subtilis BSB168 chromosomal DNA as template and appropriate primers (Table 2). The PCR fragments were inserted upstream and downstream from the kanamycin resistance cassette present in the vector pGEM-T(aphA3), the construction of which has previously been described (Joyet et al., 2010). The resulting plasmids and the previously constructed plasmid pGEM-TmtlFup(aphA3)mtlFdown were used to transform strain PJ001, which contains the gene for wild-type MtlR and a Pmtl–lacZ fusion inserted at the amyE locus (Joyet et al., 2010) providing the ΔmtlF (HB41), Δ3′-mtlA ΔmtlF (HB40) and ΔmtlAFD (HB42) mutant (Table 1). In order to construct a 3′-mtlA (EIIBMtl domain) deletion mutant the region upstream from 3′-mtlA and the entire mtlF gene together with the 5′ part of mtlD were amplified by PCR using genomic DNA as template and the primer pairs BamHIDirEIIC and RevEIICjoncBglIIMtlF as well as DirEIICjoncBglIIMtlF and RevHindIIIMtlD (Table 2). The two PCR products served as template in a second PCR with only the distal primers BamHIDirEIIC and RevHindIIIMtlD. The resulting DNA fragment was cloned into pMADΔHindIII (Joyet et al., 2010) and the pMAD-derived plasmid was used to transform the Δ3′-mtlA ΔmtlF double mutant HB40 thus providing the Δ3′-mtlA mutant. In all mutants, deletion of the correct DNA fragments was verified by PCR followed by DNA sequencing.
Construction of vectors used for complementation assays
The integrative plasmid pDG1664 (Guérout-Fleury et al., 1996), which contains two sequences for insertion at the threonine locus of the B. subtilis genome, was used for complementation of the ΔmtlF/Δ3′-mtlA and ΔmtlAFD mutants with entire MtlA or with only the EIIBMtl domain of MtlA. In a first step, the B. subtilis xyl promoter together with the Shine Dalgarno box preceding the xylA gene were amplified by PCR using chromosomal DNA of strain BSB168 as template and appropriate primers (Table 2). The entire mtlA gene or its 3′ part coding for the EIIBMtl domain were subsequently amplified by PCR using chromosomal DNA of strain BSB168 as template and primers introducing an NdeI site at the 5′-end and an EcoRI site at the 3′-end (Table 2). The start codon of the EIIBMtl domain was chosen in such a way that the phosphorylatable Cys was located in position 8, which corresponds to its usual position in EIIB proteins of the lactose PTS family (Finkeldei and Hengstenberg, 1991). The two amplified DNA fragments were incubated with DNA ligase and the two external oligos were subsequently used to amplify the ligated DNA fragment. The 3′-mtlA(Cys8Ser) allele was amplified by using the corresponding mutagenic primer together with oligonucleotide r-EIIBEco (Table 2) and chromosomal DNA of B. subtilis. The mutant allele was fused to the xyl promoter by ligation as described for the wild-type gene. The three resulting DNA fragments were cut with BamHI and EcoRI and cloned into pDG1664 cut with the same enzymes providing pDG1664-Pxyl-mtlA, pDG1664-Pxyl-3′mtlA and pDG1664-Pxyl-3′mtlA(Cys/Ser). In the resulting plasmids mtlA or the 3′-mtlA alleles are expressed from the xyl promoter. In order to obtain the mtlACys396Ala or mtlACys396Asp alleles we made use of a BglI restriction site located next to the Cys codon. We first needed to delete a BglI site in the vector pDG1644. For that purpose the pDG1664-Pxyl-mtlA plasmid was cut with MscI/EcoRV deleting also the bla gene and the origin of replication. We subsequently complemented MscI/EcoRV-cut pDG1664-Pxyl-mtlA with the origin of replication of pACYC184, which was obtained by digesting this vector with PvuII and EcoRV. The resulting plasmid was used as template for the amplification of a DNA fragment with the primers XylAproBam and r-EIIB-alaBglI or r-EIIB-aspBglI, which allowed the Cys396Ala or Cys396Asp replacements in the EIIBMtl domain of MtlA. The amplified fragments were cut with BamHI and BglI and used to replace the corresponding region in the above plasmid thus providing pDG1664-Pxyl-mtlACys396Ala and pDG1664-Pxyl-mtlACys396Asp. The pDG1664-derived plasmids carrying the various 3′-mtlA and mtlA alleles were integrated by double cross-over into the thrC locus of the ΔmtlF/Δ3′-mtlA and ΔmtlAFD mutants. The pDG1664-derived plasmid containing the wild-type 3′ part of mtlA was also integrated into strain HB41, PJ001, PJ007 and PJ031, which each contains the Pmtl–lacZ fusion inserted at the amyE locus and either wild-type mtlR, mtlRH342D or mtlRC419A; ΔmtlRC419A also contains a deletion of mtlF (Joyet et al., 2010). The correct insertion was confirmed by PCR amplification.
Construction of the ywqC::3′-mtlA fusion
The B. subtilis replicative plasmid pBQ200, which is a pHT315 derivative (Arantes and Lereclus, 1991) carrying the constitutive promoter PdegQ36 (Martin-Verstraete et al., 1994) was used for the construction of the ywqC::3′-mtlA fusion. In a first step a DNA fragment containing a truncated ywqC gene (lacking the last 34 codons) together with its Shine Dalgarno box was amplified by PCR using chromosomal DNA of B. subtilis BSB168 as template and primers YwqCForBam and YwqCRevSal (Table 2) introducing BamHI and SalI restriction sites at the 5′- and 3′-ends respectively. The amplified DNA was cut with the corresponding restriction enzymes and cloned into pBQ200 digested with the same enzymes providing plasmid pBQ200-ywqC. In the second step the 3′-mtlA fragment was amplified by using genomic DNA of B. subtilis BSB168 as template and the primer pair EIIBBam and r-EIIBsalI. In order to obtain the 3′-mtlA(Cys-8-Ala) and (Cys-8-Asp) alleles (Cys-8 corresponds to Cys-396 in entire MtlA) the mutagenic primers EIIBalaBam or EIIBaspBam, respectively, were used instead of EIIBBam. The three amplified DNA fragments were cloned into pQE30 cut with the corresponding restriction enzymes. The pQE30 inserts were subsequently amplified by PCR by using primers EIIBMtlForSal and EIIBMtlRevSph (Table 2) introducing SalI and SphI restriction sites at the 5′- and 3′-ends respectively. The resulting PCR fragment contained the entire His-tag encoded by pQE30, which was kept in order to function as a spacer between YwqC and the EIIBMtl domain. The amplified DNA fragments were cut with the corresponding restriction enzymes and cloned into pBQ200-ywqC digested with the same enzymes providing the three plasmids pBQ200-ywqC-3′-mtlA, pBQ200-ywqC-3′-mtlA(Cys/Ala) and pBQ200-ywqC-3′-mtlA(Cys/Asp). The correct sequence and in frame fusion of the inserts were verified and the plasmids were used to transform the ΔmtlF/Δ3′-mtlA and ΔmtlAFD mutants. For a control experiment the 3′-mtlA fragment preceded by the Shine Dalgarno box of xylA was amplified by using plasmid pDG1664-xyl-3′-mtlA as template and the amplified DNA fragment was cloned into pBQ200 without ywqC.
Yeast two-hybrid experiments
The entire B. subtilis mtlR gene, the DNA fragments encoding the four regulatory domains of MtlR and each regulatory domain fused to its upstream or downstream domain as well as the 3′-mtlA DNA fragment encoding the EIIBMtl domain were PCR-amplified by using appropriate primers (Table 2). All obtained PCR products contained a BamHI restriction site at the 5′-end and a SalI site at the 3′-end. They were cloned into both, the bait vector pGBDU (URA3) and the prey vector pGAD (LEU2), which were cut with the same enzymes. The bait constructs were introduced in the Saccharomyces cerevisiae strain PJ69-4a and the prey constructs in strain PJ69-4α (James et al., 1996). The correct DNA sequence of all cloned fragments was confirmed by DNA sequencing. The plasmids were used to transform the two yeast strains, which for mating were grown in rich YEPD medium. Resulting diploids were selected for an interaction phenotype by growing them in synthetic complete medium (SC) lacking the appropriate amino acids (Leu, Trp, His) or nucleotides (Ade, Ura) (Guthrie and Fink, 1991). An interaction of the two proteins encoded by the inserts of the pGBDU and pGAD vectors was indicated by growth of the corresponding S. cerevisiae strain. False-positive interactions generated by the yeast two-hybrid system were eliminated experimentally as previously described (Noirot-Gros et al., 2002).
RNA isolation and qRT-PCR
The B. subtilis wild-type strain was grown to exponential phase in LB medium with and without mannitol and the ΔmtlF and Δ3′-mtlA ΔmtlF mutants were grown only in LB medium. Cells from 30 ml of cultures were collected by centrifugation and RNA was extracted using the RNeasy mini Kit (Qiagen) according to the manufacturer's protocol. Residual DNA was removed using the Turbo DNA-free kit (Ambion). The quality of the RNA preparation was tested with an Agilent Bioanalyser before RT reactions were performed using random hexamers (GE Healthcare) according to the manufacturer's recommendations. qRT-PCR experiments were carried out with the LightCycler (Roche) using primers specific for the lacZ gene and the SuperScript III Reverse Transcriptase (Invitrogen) by following the protocol of LightCycler Faststart DNA Master SYBR Green I Kit (Roche). Data were processed with the Roche Molecular Biochemicals LightCycler Software. In order to normalize transcript levels the constitutively expressed rpoB gene was used as a reference (Milohanic et al., 2003). Statistical analyses of the data were carried out with the Wilcoxon–Mann–Whitney test.
β-Galactosidase activity in toluenized B. subtilis cells was measured as previously described (Miller, 1972). In short, 3 ml of LB medium containing appropriate antibiotics and either no carbohydrate, 1% mannitol or 0.2% xylose (for strains containing mtlA or 3′-mtlA fused to Pxyl) were inoculated with 200 μl of an overnight culture and grown for 4 h at 37°C before an aliquot of 1 ml was withdrawn. Cells were harvested by centrifugation, resuspended in 500 μl of Z buffer and permeabilized with toluene. β-Galactosidase assays were performed with o-nitrophenyl-β-d-galactopyranoside as substrate and the change of the absorption at 420 nm (A420) was continuously followed with a Kontron Bio-Tek spectrophotometer using the ‘Autorate’ program. The specific activity is expressed as ΔA420 per min and μg protein. Protein concentrations were determined by using the protein-dye binding method with Coomassie Brilliant Blue (Bradford, 1976).
We thank Meriem Derkaoui and Dominique Le Coq for useful discussions.