Characterization of the transcriptionally active form of dephosphorylated DctD complexed with dephospho-IIAGlc

ABSTRACT Bacterial enhancer-binding proteins (bEBPs) acquire a transcriptionally active state via phosphorylation. However, transcriptional activation by the dephosphorylated form of bEBP has been observed in DctD, which belongs to Group I bEBP. The formation of a complex between dephosphorylated DctD (d-DctD) and dephosphorylated IIAGlc (d-IIAGlc) is a prerequisite for the transcriptional activity of d-DctD. In the present study, characteristics of the transcriptionally active complex composed of d-IIAGlc and phosphorylation-deficient DctD (DctDD57Q) of Vibrio vulnificus were investigated in its multimeric conformation and DNA-binding ability. DctDD57Q formed a homodimer that could not bind to the DNA. In contrast, when DctDD57Q formed a complex with d-IIAGlc in a 1:1 molar ratio, it produced two conformations: dimer and dodecamer of the complex. Only the dodecameric complex exhibited ATP-hydrolyzing activity and DNA-binding affinity. For successful DNA-binding and transcriptional activation by the dodecameric d-IIAGlc/DctDD57Q complex, extended upstream activator sequences were required, which encompass the nucleotide sequences homologous to the known DctD-binding site and additional nucleotides downstream. This is the first report to demonstrate the molecular characteristics of a dephosphorylated bEBP complexed with another protein to form a transcriptionally active dodecameric complex, which has an affinity for a specific DNA-binding sequence. IMPORTANCE Response regulators belonging to the bacterial two-component regulatory system activate the transcription initiation of their regulons when they are phosphorylated by cognate sensor kinases and oligomerized to the appropriate multimeric states. Recently, it has been shown that a dephosphorylated response regulator, DctD, could activate transcription in a phosphorylation-independent manner in Vibrio vulnificus. The dephosphorylated DctD activated transcription as efficiently as phosphorylated DctD when it formed a complex with dephosphorylated form of IIAGlc, a component of the glucose-phosphotransferase system. Functional mimicry of this complex with the typical form of transcriptionally active phosphorylated DctD led us to study the molecular characteristics of this heterodimeric complex. Through systematic analyses, it was surprisingly determined that a multimer constituted with 12 complexes gained the ability to hydrolyze ATP and recognize specific upstream activator sequences containing a typical inverted-repeat sequence flanked by distinct nucleotides.

ATPase activity to successfully become transcriptional activators (3).DctD, a member of Group I bEBPs, has been initially isolated as the main regulator for the expression of the dicarboxylic acid transporter (DctA) in a nitrogen-fixing Rhizobium species (4).Its sensor kinase DctB, comprising a two-component system (TCS) with the response regulator DctD, senses ambient dicarboxylic acids and phosphorylates DctD, thereby activating the transcription of dctA genes (5).
DctBD systems have also been found in many bacterial species, including the model foodborne pathogen Vibrio vulnificus, in which RpoN-initiated transcription of two distinct gene clusters for exopolysaccharides (EPS) biosynthesis, EPS-II and EPS-III clusters, is activated by DctBD (6).These findings demonstrate that the regulatory roles of DctD may extend beyond the expression of genes related to the uptake and metabolism of dicarboxylic acids.Furthermore, V. vulnificus DctD has been shown to exhibit its in-vivo transcriptional activity even in the absence of DctB ligands such as dicarboxylic acids.The dephosphorylated form of V. vulnificus DctD (d-DctD) transitions into the transcriptionally active state under conditions in which the cellular levels of the dephosphorylated form of the glucose-specific enzyme IIA (d-IIA Glc ) are sufficient to interact with d-DctD (7).In this novel regulatory pathway, the phosphorelay-independ ent activation of DctD occurs through the direct interaction of d-DctD with d-IIA Glc , which forms a transcriptionally active complex of [d-IIA Glc /d-DctD].
Based upon numerous reports regarding the transcriptionally active forms of Group I bEBPs (8), it has been speculated that the complex of [d-IIA Glc /d-DctD] forms multi meric state(s) to bind to the UAS and activate the transcription of EPS-II and EPS-III clusters.Therefore, in the present study, we examined the molecular characteristics of the complex consisting of d-DctD and d-IIA Glc .Among the multimeric states of the [d-IIA Glc /d-DctD] complex, a multimer exhibiting both DNA-binding affinity and ATPase activity was identified.Next, the DNA regions comprising the UAS specifically required for multimeric [d-IIA Glc /d-DctD] were localized, and the consensus binding sequences for multimeric [d-IIA Glc /d-DctD] were proposed.

A phosphorylation-deficient DctD (DctD D57Q ) forms a transcriptionally inactive dimer ([DctD D57Q ] 2 )
NtrC, one of the representative members of response regulators belonging to Group I of bEBPs, forms a dimer when not phosphorylated, and this dimer possesses DNA-binding ability (9).However, the dephosphorylated form of DctD (d-DctD) has been shown to be unable to bind to the UASs of targets such as EPS-II and EPS-III clusters in V. vulnificus (7).Therefore, we examined the oligomeric state of d-DctD using a phosphory lation-deficient mutant DctD D57Q .On gel permeation chromatography (GPC), recombi nant DctD D57Q protein produced a major peak at 13.4 mL in its elution profile (Fig. 1A).When this peak volume was extrapolated to a regression equation derived from standard protein markers (Fig. S1) (10), it was converted to a molecular weight (MW) of 117.1 kDa, which approximated the calculated size of the dimeric DctD D57Q (107.1 kDa).To verify that this dimer did not have DNA-binding ability, as previously shown using the whole fraction of DctD D57Q (7), the fractionated dimeric DctD D57Q ([DctD D57Q ] 2 ) was added to a DNA fragment covering −418 to +62 relative to the RpoN-dependent transcription initiation site (TIS, 6) of the EPS-II cluster.The reaction mixtures were run on two identical gels: one was used to localize the proteins by staining with Coomassie Blue, and the other was used to localize the labeled probes via autoradiography (Fig. 1B).Retarded bands of dimeric DctD D57Q and the DNA probe were not observed on either gel, indicating that [DctD D57Q ] 2 had no DNA-binding ability.Although DctD D57Q is present in a transcriptionally inactive state, it acquires DNA-bind ing ability in the presence of dephosphorylated IIA Glc (d-IIA Glc ), forming a transcription ally active complex (7).In this study, we analyzed the molecular characteristics of the interaction between the two proteins, such as the multimeric state(s) of the d-IIA Glc / DctD D57Q complex and the molar ratio(s) of each protein in the complex.A mixture containing the same concentration (20 µM) of the two recombinant proteins, DctD D57Q and d-IIA Glc , was analyzed by GPC.Resultant elution profile showed two distinct peaks at 8.9 and 12.8 mL (Fig. 2A), which were converted to molecular weights of 905.5 and 151.0 kDa, respectively, by extrapolating each peak volume to a regression equation shown in Fig. S1.These converted values approximated the calculated sizes of the dodecamer (902.6 kDa) and dimer (150.4 kDa) of the d-IIA Glc /DctD D57Q complex.To verify the two multimers' MWs determined via GPC analysis, the same mixture was analyzed by an independent method: a gradient gel electrophoresis alongside the two standards, one covering from 180 to 1,800 kDa and the other covering from 66 to 1,048 kDa (Fig. S2).The mixture was resolved into two bands: a higher MW band at 900 kDa and a lower MW band between 180 and 146 kDa.This observation strongly supported the GPC-mediated estimation of the d-IIA Glc /DctD D57Q complexes.Subsequent analysis of each fraction collected every 1 mL revealed that the protein bands corresponding to the recombinant proteins of DctD D57Q and IIA Glc appeared only in the lanes of a gel run by the fractions covering the two peaks: fractions 9-10 for a dodecamer and fractions 12-13 for a dimer (Fig. 2B).To further examine the compositional characteristics of each complex, fractions 9, 10, 12, and 13 were separated on an SDS-polyacrylamide gel (Fig. 2C).Using densitometric readings of known amounts of each recombinant protein, ranging from 30 to 240 pmol, which were run in the same gel, the standard curves for DctD D57Q (right panel) and IIA Glc (left panel) were plotted (Fig. 2D).The amounts of DctD D57Q and IIA Glc in the four fractions were estimated, indicating that almost the same ratio of the two proteins was present in each fraction.Fractions 9, 10, 12, and 13 contained 27, 63, 57, and 108 pmol of IIA Glc and 27, 57, 48, and 99 pmol of DctD D57Q , respectively.

DctD
To confirm the 1:1 molar ratio of the two proteins in either the dimeric or dodeca meric form of the d-IIA Glc /DctD D57Q complex, mixtures containing different ratios of the two proteins were prepared as follows: 5 µM d-IIA Glc + 20 µM DctD D57Q and 20 µM d-IIA Glc +5 µM DctD D57Q .Two mixtures were subjected to GPC and subsequent SDS-PAGE, as shown in Fig. 2. The mixture containing more DctD D57Q than d-IIA Glc (5 µM d-IIA Glc + 20 µM DctD D57Q ) produced an extra peak at 13.4 mL in the GPC profile in addition to two peaks (8.9 and 12.8 mL) whose heights and areas decreased by 2-to 5-fold compared to those shown in Fig. 2A (Fig. 3A).Fractions 14 and 15, which covered the extra peak at 13.4 mL peak, were revealed to contain only DctD D57Q (Fig. 3B).Therefore, the peak at 13.4 mL contained the dimeric DctD D57Q that was not involved in forming the d-IIA Glc /DctD D57Q complex.In contrast, the mixture containing more d-IIA Glc than DctD D57Q (20 µM d-IIA Glc + 5 µM DctD D57Q ) produced two peaks at 8.9 and 12.8 mL, and their heights and areas were almost the same as those in Fig. 3A (Fig. 3C).However, an SDS-polyacrylamide gel run with each elution showed the presence of a protein band with a MW of IIA Glc in fractions 18 and 19 (Fig. 3D).To prove this band was d-IIA Glc that was not involved in forming the d-IIA Glc /DctD D57Q complex, recombinant d-IIA Glc (20 µM) was passed through GPC and the collected fractions were subjected to SDS-PAGE analysis.Although d-IIA Glc could not be detected using GPC equipped with an UV detector (Fig. 3E), its presence was evident in the 18th and 19th fractions (Fig. 3F).(7), was subjected to GPC (A), as described in Fig. 1A.Each fraction for the peaks at 8.9 and 12.8 mL was run in an SDS-polyacrylamide gel (B).(C and D) Compositional analysis of the d-IIA Glc /DctD D57Q complex in the two peaks.The fractions for the peaks at 8.9 mL (fractions 9 and 10) and 12.8 mL (fractions 12 and 13) were separated in an SDS-polyacrylamide gel (C).For quantitative analysis, the known amounts (30-240 pmol) of each recombinant protein were included in the same gel.Densitometric readings of each band were plotted against the protein amounts to produce the standard curves (D, open circles).
Using the standard curves for d-IIA Glc (left graph) and DctD D57Q (right graph), the amounts of each protein in the four fractions were extrapolated and the resultant values were provided below the graphs.
[d-IIA Glc /DctD D57Q ] 12 has an ability to bind to DNA and hydrolyze ATP The two multimeric forms, a dimer ([d-IIA Glc /DctD D57Q ] 2 ) and a dodecamer ([d-IIA Glc / DctD D57Q ] 12 ), of the complex are composed of d-IIA Glc and DctD D57Q in a 1:1 molar ratio.To identify the multimeric form(s) of the complex capable of binding to DNA, the labeled probe shown in Fig. 1 was mixed with [d-IIA Glc /DctD D57Q ] 2 or [d-IIA Glc /DctD D57Q ] 12 .The mixtures were run on two identical gels, and each gel was stained with Coomassie Blue or exposed to a phosphoimager (Fig. 4A).The mixture containing [d-IIA Glc /DctD D57Q ] 2 did not show any shifted band on either gel, indicating that the dimeric form of the complex had no DNA-binding ability.In contrast, a mixture containing [d-IIA Glc /DctD D57Q ] 12 produced retarded bands of protein and DNA, corresponding to a protein band for the probe-bound [d-IIA Glc /DctD D57Q ] 12 (left panel, Fig. 4A) and a DNA band for the [IIA Glc / DctD D57Q ] 12 -bound probe (right panel, Fig. 4A).This suggests that [IIA Glc /DctD D57Q ] 12 , which exhibits DNA-binding affinity, may be a transcriptionally active form of d-DctD.
It has been reported that transcriptionally active forms of the Group I bEBP have the enzymatic activity to hydrolyze ATP molecules (12).Assays of ATPase activity using  [d-IIA Glc /DctD D57Q ] 12 specifically binds to the sequences homologous to a Rhizobium DctD-binding site It has been commonly observed that many members of the Group I bEBP bind to multiple sites in the upstream region of a target gene to activate its transcription (3).Two sites were localized in the upstream region of dctA of Rhizobium species (14).Addition ally, the DctD-binding sequences, 5′-TGTGCGgaaatCCGCACA-3′, have been identified in Rhizobium meliloti (15).Using these sequences, the putative UASs of the EPS-II and EPS-III clusters in V. vulnificus, which are homologous to the R. meliloti DctD-binding sequences, were searched in their upstream regions.Two sites with homology to the consensus sequences were discernible, which were designated as BS1 remote from TIS-1 (−254 to −237 in EPS-II and −451 to −434 in EPS-III) and BS2 close to TIS-1 (−142 to −125 in EPS-II and −160 to −143 in EPS-III) (Fig. 5A and B).To determine whether [d-IIA Glc /DctD D57Q ] 12 could bind to these putative BSs, DNA probes containing a single BS of each gene cluster (P1 BS1 for the probe including BS1 and P2 BS2 for the probe including BS2) were prepared for electrophoretic mobility shift assay (EMSA).Only the probes containing BS1s were bound by [d-IIA Glc /DctD D57Q ] 12 and produced a shifted band in a concentra tion-dependent manner (Fig. 5C and D).Furthermore, the shifted bands were decreased by the addition of the cold probes but persisted by the addition of the noncompetitive gap DNA, indicating the specific binding of [d-IIA Glc /DctD D57Q ] 12 to the BS1-probes.In contrast, the probes containing BS2s did not produce any retarded band, even in the presence of the highest concentration of [d-IIA Glc /DctD D57Q ] 12 used in this study (up to 100 nM; Fig. 5E and F).These results indicate that [d-IIA Glc /DctD D57Q ] 12 efficiently binds to P1s containing the nucleotide sequences in BS1s.

Interaction of [d-IIA Glc /DctD D57Q ] 12 to DNAs containing the specific binding sites is required for successful transcription of the target gene clusters
Next, to confirm whether the putative BS1s were directly and specifically recognized and bound by [d-IIA Glc /DctD D57Q ] 12 , P1 probes containing the mutagenized BS1s were prepared by substituting the conserved nucleotides (BS1Ms; Fig. 6A and B).An EMSA using P1 probes containing mutagenized BS (P1 BS1M ) in the presence of [d-IIA Glc / DctD D57Q ] 12 did not show any retarded band (Fig. 6C and D).These results indicated that or the mutagenized upstream regions (P1 BS1M -P2 BS2 ) of EPS-II (E) and EPS-III (F) were transferred to the wild type and dctD D57Q (7).Then, V. vulnificus cells were grown in AB-glucose supplemented with tetracycline (3 µg/mL).At an OD 595nm of 0.4, aliquots of bacterial cells were harvested, and their luciferase activities were measured, as described in the Materials and Methods section.The expression of each reporter was presented as normalized values: the relative light unit (RLU) divided by the cell mass (OD 595nm ) of each sample.As a negative control, ΔdctD strain carrying the original reporter was included in each assay.P-values for comparison with the P1 BS1 -P2 BS2 reporter in the wild type or ΔdctD were indicated (**P < 0.001; ns, not significant).
the specific binding of [d-IIA Glc /DctD D57Q ] 12 to P1 was achieved through its interaction with the regions containing the BS1s of the two gene clusters.To determine whether BS1 is the actual cis-element determining the in-vivo transcrip tional activation by [d-IIA Glc /DctD D57Q ] 12 , luxAB-based transcription reporters fused with the original DNA fragments of the gene clusters (P1 BS1 -P2 BS2 ; 18)or DNA fragments containing mutated BS1s (P1 BS1M -P2 BS2 ) were prepared.Their expression was monitored in wild-type and dctD D57Q V. vulnificus grown in AB-glucose medium, in which IIA Glc and DctD were present in dephosphorylated forms, as previously shown by Kang and Lee (7) (Fig. 6E and F).Compared to the original reporters of EPS-II and EPS-III clusters, the expression of P1 BS1M -P2 BS2 -reporters was significantly impaired, which was almost at the same levels as those in ΔdctD V. vulnificus.These results indicate the critical role of BS1 in transcriptional activation by [d-IIA Glc /DctD D57Q ] 12 .

Specific nucleotides in the downstream regions of BS1s (BS1[dn]) are additionally required for successful binding of [d-IIA Glc /DctD D57Q ] 12
Binding sequences for R. meliloti DctD, which comprise 18 nucleotides, have been derived from footprint assays using the truncated DctD mimicking phosphorylated DctD (p-DctD) (19).Considering the molecular size of [d-IIA Glc /DctD D57Q ] 12 , it was plausible to speculate that this complex might interact with an extended region in the target DNA.Therefore, we further tested whether the DNA regions flanking BS1s were required for successful transcriptional activation by [d-IIA Glc /DctD D57Q ] 12 .For this purpose, another luxAB-based transcription reporter of the EPS-II cluster was prepared using DNA fragment, in which the locations of BS1 and BS2 were switched to produce P1 BS2 -P2 BS1 .Then, its expression was monitored in wild-type (Fig. 7A) and dctD D57Q (Fig. 7B) V. vulnificus grown in AB-glucose medium.Although a reporter having P1 BS2 -P2 BS1 contained the intact BS1 sequences, its expression was not induced in both strains of V. vulnificus, compared to the original reporter having P1 BS1 -P2 BS2 .
To elucidate the reason for the lack of induction of the P1 BS2 -P2 BS1 expression, two probes with switched BSs, P1 BS2 and P2 BS1 , were prepared for EMSA in the presence of [d-IIA Glc /DctD D57Q ] 12 .No binding of [d-IIA Glc /DctD D57Q ] 12 to P1 BS2 or P2 BS1 was observed (Fig. 7C).This suggests that in addition to the 18 nucleotide-long BS1, the extra nucleotide sequences in P1 were required for binding by [d-IIA Glc /DctD D57Q ] 12 and the successful activation of EPS-II cluster transcription.

Alignments of BS1s of various DctD-regulons and localization of BS1[dn] for binding of [d-IIA Glc /DctD D57Q ] 12
To draw the consensus-binding sequences for [d-IIA Glc /DctD D57Q ] 12 , putative BS1 and BS1[dn] were localized in the upstream regions of the known DctD-regulons (i.e., dctA genes of various bacterial species) and the tentative DctD-regulons of V. vulnificus (e.g., mlsI, dcuB, and dcuC genes).Alignment of these sequences with those of the EPS-II and EPS-III clusters revealed that 18 nucleotide-long BS1s are relatively well conserved as the following inverted repeat sequences: 5′-TGTG-aa----tt-CACA-3′ (Fig. 8).In addition, the arrays of multiple Ts apparently locate at the 7th or 8th nucleotide after the 3′-end of BS1s of the above genes encoded by three Vibrio species and Escherichia coli (6,20,21).In contrast, T-rich BS1[dn] was not discernible in the upstream regions of dctA genes of Sinorhizobium meliloti (14), R. meliloti (15), and Rhizobium leguminosarum (14): it is noteworthy that genes encoding the components of the typical glucose-PTS system are not present in Rhizobium and related species (22), suggesting the absence of [d-IIA Glc /d-DctD] complex in these bacteria.

DISCUSSION
DctD is a well-known transcription factor belonging to Group I of bEBP, whose regula tory competence in activating RpoN-initiated transcription ( 23) is achieved after it is phosphorylated by its cognate sensor kinase, DctB.Interestingly, the transcriptionally inactive state of dephosphorylated DctD (d-DctD) is converted to its transcriptionally active form when it produces a complex with d-IIA Glc (7).To date, transcriptional activation by dephosphorylated forms of other Group I bEBPs has not been reported.Our results suggest that d-DctD or the d-IIA Glc /d-DctD complex may have unique features that have not been observed in the well-studied members of Group I bEBPs.Therefore, to understand this phosphorylation-independent regulatory mechanism, the molecular characteristics of both d-DctD and the d-IIA Glc /d-DctD complex were investigated in this study using the DctD D57Q and the DctD-regulons identified in V. vulnificus (6).
DctD D57Q , a form of recombinant DctD that mimics dephosphorylated DctD (7), exists in a dimeric conformation (Fig. 1).Although dimeric forms of dephosphorylated Group I bEBPs, such as NtrC and ZraR, were at least able to bind to specific DNA sequences (9,24), dimeric d-DctD ([DctD D57Q ] 2 ) did not show any DNA-binding activity (Fig. 1C).Thus, while a dimeric form of truncated DctD has been shown to have DNA-binding affinity (13,25), it is noteworthy that this recombinant protein was deleted in its regulatory domain and considered to be a constitutively active form.
It is generally considered that the oligomerization process of bEBPs is initiated by the dimerization of two monomers and followed by the formation of a hexamer via direct interaction among three dimers (3).However, a process deviated from the above stepwise oligomerization has been observed in PspF belonging to the Group IV bEBP, which lacks the regulatory domain and requires other proteins to control its ATPase activity and DNA-binding affinity (27).When DctD D57Q was mixed with d-IIA Glc , a hexameric form of the complex was not detected.Thus, it is speculated that [d-IIA Glc / DctD D57Q ] 2 undergoes hexamerization to produce [d-IIA Glc /DctD D57Q ] 12 , in which their ATPase domains can be positioned to attain the ATP hydrolyzing activity.The amino acid residues of DctD and IIA Glc that are involved in the interaction between the two proteins have not yet been identified, and the structural characterization of this complex waits future study.Nonetheless, it has been postulated that d-IIA Glc would interact with the regulatory domain of DctD to mimic the transcriptionally active p-DctD since structural modification of the phosphorylated regulatory domains is critically important for eliciting the ATPase activity of Group I bEBPs (12).Therefore, it is plausible that the arrangement of a dodecamer composed of 12 d-DctDs, to which regulatory domains are adhered by d-IIA Glc , should be different from the arrangements of a higher-order multimer of p-DctD.
The previous studies with the Rhizobium DctD showed that DctD-binding nucleotide sequences were 5′-TGTGCGgnnntCCGCACA-3′ and the multiple (i.e., at least two) binding sites were localized in the regions upstream dctA genes (14,19,28). .These findings led us to putatively localize multiple DctD-binding sites, which were homologous to the above sequences, in the upstream regions of the DctD-regulon, such as the EPS-II and EPS-III clusters.Among the two candidate binding sites, BS1 and BS2 (Fig. 5A and B), BS1s locating relatively far from the RpoN-initiated TISs were specifically bound by [d-IIA Glc / d-DctD] 12 (Fig. 5C through F).Alignment of the putative BS1s in the upstream regions of various DctD-regulons showed the presence of relatively well-conserved inverted repeat sequences of 5′-TGTG-aa----tt-CACA-3′ (Fig. 8).
Structural studies using recombinant NtrC, a representative member of the Group I bEBP, suggest that each monomer in a dimer is directly involved in specifically binding the target DNA sequences consisting of 17 nucleotides (29).Therefore, the binding regions for [d-IIA Glc /DctD D57Q ] 12 may be wider than those determined using a truncated DctD dimer (14,19,28).To test this hypothesis, regions flanking BS1 were examined for their involvement in specific interactions with [d-IIA Glc /DctD D57Q ] 12 .EMSA using various DNA probes, as shown in Fig. 7D and F, clearly demonstrated that both BS1 and its downstream region (BS1[dn]), containing at least three consecutive Ts, were essential for the in vitro binding of [d-IIA Glc /DctD D57Q ] 12 (Fig. 7E, G and H) and the in vivo transcrip tional activation (Fig. 7A and B).
Cellular levels of the dephosphorylated forms of glucose-PTS components are highly increased in bacteria growing in the presence of glucose (30).Thus, cellular levels of [d-IIA Glc /d-DctD] 12 and its transcriptional activity are mainly determined by glucose in the ambient environments.However, complex formation between d-DctD and d-IIA Glc is further regulated by other carbon sources, such as glycerol (7).Glycerol kinase, GlpK, has high affinity to d-IIA Glc (31,32); thus, its presence results in decreased levels of cellular [d-IIA Glc /d-DctD] 12 due to competition between GlpK and d-DctD for binding d-IIA Glc .Since d-IIA Glc is a versatile protein capable of interacting with the other kinases of non-PTS sugars, including maltose, arabinose, melibiose, and raffinose (33), [d-IIA Glc / d-DctD] 12 -mediated regulatory pathways should be under the multilayered control of diverse carbon sources.Furthermore, complex formation with IIA Glc may play an additional role in controlling d-DctD protein stability in vivo.In this case, the formation of [d-IIA Glc /d-DctD] 12 could be a way modulating the appropriate cellular levels of total DctD.
Taken together, DctD, a response regulator comprising a two-component regulatory system, can activate transcription via an alternative pathway that is independent of its cognate sensor kinase, but dependent upon a dephosphorylated component of the glucose-PTS, IIA Glc .When the dodecamer of the d-IIA Glc /d-DctD complex is formed, it specifically binds to a single but extended BS and activates transcription of the target genes.

Purification and gel permeation chromatography of recombinant proteins
For the preparation of d-DctD 2 , pQE30-dctD D57Q (7) was expressed in E. coli JM109 in the presence of 1 mM isopropyl β-D-thiogalactopyranoside. To prepare d-IIA Glc , pQE30-crr was expressed in the presence of glucose, as previously described (36).Each recombinant protein was purified using an Ni + -nitrilotriacetic acid affinity column (Bio-Rad).Next, 500 µL of the recombinant proteins of DctD D57Q , d-IIA Glc and the d-IIA Glc /DctD D57Q complex dissolved in a buffer (50 mM Tris-HCl [pH 8.0], 20 mM KCl, 50 mM MgCl 2 , and 100 mM NaCl) were applied to the AKTA-FPLC system (Amersham Biosciences) equipped with a Superdex 200 Increase 10/300 Gl column (GE Healthcare) (37).Each sample was fractionated using a running buffer (50 mM Tris-HCl [pH 8.0], 20 mM KCl, 50 mM MgCl 2 , and 300 mM NaCl) at a flow rate of 0.4 mL/min.The void volume of the column used in this assay was 7.2 mL, which is consistent with the previously reported value (38).The apparent molecular weights of the oligomeric proteins in the collected fractions were determined using an elution profile derived from the standard proteins (Protein Standard Mix 15-600 kDa, Sigma-Aldrich) as described in Fig. S1.

Site-directed mutagenesis
DctD-binding sites (BS1) in the upstream regions of the EPS-II and EPS-III clusters were mutagenized using the overlap extension method (39) with the following sets of primers carrying the desirably substituted nucleotides: BS1M_II-F and BS1M_II-R for BS1 of EPS-II and BS1M_III-F and BS1M_III-R for BS1 of EPS-III (Table S2).To switch the positions of the BSs of EPS-II, the primer sets of EPS_II_BS1toBS2-F and EPS_II_BS1toBS2-R or EPS_II_BS2toBS1-F and EPS_II_BS2toBS1-R were used to produce P1 BS2 or P2 BS1 , respectively.To mutagenize the nucleotide sequences in the upstream and down stream regions of BS1 of EPS-II, the primer sets of EPS_II_upst-F and EPS_II_upst-R or EPS_II_dnst_F and EPS_II_dnst_R were used to produce P1 BS1-upM or P1 BS1-dnM , respectively.The resultant DNA fragments were cloned into pBlunt-TOPO (MGmed) and the mutagenized nucleotide sequences were verified by DNA sequencing.To construct pQE30-dctD H216R , the internal primers including the altered nucleotides for the 216th arginine of V. vulnificus DctD, H216R-F and H216R-R (13), were utilized as described above (39, Table S2).The resultant mutagenized dctD fragment was digested with BamHI and KpnI and then ligated to pQE30 as previously described (6).7F.DNA fragments were labeled with [γ-32 P] ATP using T4 polynucleotide kinase (TaKaRa), and the resultant labeled probes were incubated in a reaction buffer (50 mM Tris-HCl [pH 8.0], 20 mM KCl, 50 mM MgCl 2 , and 100 mM NaCl) with various concentrations of recombinant proteins of DctD D57Q (7) and d-IIA Glc (36).The reaction mixtures were resolved on 6% or 8% native polyacrylamide gels.

Measurement of transcriptional reporter plasmids fused with luxAB genes
The expression of various luxAB-based transcription reporters was monitored in V. vulnificus strains growing in AB-glucose medium, as previously described (6).At the designated time points, the bacterial culture aliquots were mixed with a luciferase substrate, n-decyl aldehyde (0.006%), and then the resultant light production was measured in a luminometer (GloMax 20/20 luminometer, Promega).Specific biolumines cence was presented by normalizing the RLU with respect to cell mass (OD 595 ), as described (41).

Statistical analyses
Results are expressed as means ± standard deviations of data from at least three independent experiments.Statistical analysis was performed using Student's t-test (Systat Program, SigmaPlot version 9; Systat Software, Inc.).P-values are presented by one asterisk ( * ) or two asterisks ( ** ) when 0.001 < P < 0.01 or P < 0.001, respectively.

FIG 1
FIG 1 Characterization of the multimeric conformation of DctD D57Q (A) Chromatographic profile of DctD D57Q .Recombinant protein of DctD D57Q (7), whose calculated molecular weight is 53.53 kDa, was subjected to GPC as described in the Materials and Methods section.The resultant profile showing a single peak at 13.4 mL was presented with the absorbance at 280 nm (mAu).Its size was determined using a regression equation of the standard proteins, as shown in Fig. S1.The fractions (13 and 14 mL) containing this peak were collected and used for electrophoretic mobility shift assay (EMSA) in Fig. 1B.(B) DNA-binding characteristic of the dimeric DctD D57Q .Approximately 130 nM of the labeled probe from −418 to +62 relative to the transcription initiation site 1 of the EPS-II cluster (6) was incubated with 250 nM dimeric DctD D57Q .Reaction mixtures were run on two identical 8% native polyacrylamide gels; one was stained with Coomassie Blue to localize the proteins (left panel), and the other was observed under a phosphoimager to localize the DNA probes (right panel).The first lanes in each gel were run with the whole fraction of DctD D57Q preparation (left panel) and the labeled probe only (right panel).The second and third lanes in each gel were run with the mixtures containing [DctD D57Q ] 2 in the absence (−) and presence (+) of DNA probe, respectively.

FIG 2
FIG 2 Characterization of the multimeric conformation of the complex composed of dephospho-IIA Glc and DctD D57Q (d-IIA Glc /DctD D57Q complex).(A and B) Chromatographic and electrophoretic analyses of d-IIA Glc /DctD D57Q complex.A mixture of the same concentrations of two recombinant proteins (20 µM each), d-IIA Glc and DctD D57Q(7), was subjected to GPC (A), as described in Fig.1A.Each fraction for the peaks at 8.9 and 12.8 mL was run in an SDS-polyacrylamide gel [DctD D57Q ] 2 , [d-IIA Glc /DctD D57Q ] 2 , and [d-IIA Glc /DctD D57Q ] 12 revealed that only the dodecameric form of the complex exhibited ATPase activity (gray bars, Fig. 4B).To exclude the possibility that the fractions containing [d-IIA Glc /DctD D57Q ] 12 might have been coeluted with some ATPases, an ATP hydrolysis-deficient DctD (13), DctD H216R , was prepared as a dephosphorylated form (d-DctD H216R ) and used for GPC analysis.The fractions containing a dodecamer of the complex composed of d-IIA Glc and d-DctD H216R ([d-IIA Glc /d-DctD H216R ] 12 ) showed insignificant ATPase activity, as [d-DctD H216R ] 2 and [d-IIA Glc /d-DctD H216R ] 2 did (black bars, Fig. 4B).It suggests that the apparent ATP hydrolysis by the fractions of [d-IIA Glc /DctD D57Q ] 12 was not derived from trace contami nation of ATPase activity during GPC fractionation used in this study.Therefore, [d-IIA Glc / DctD D57Q ] 12 , which has both DNA-binding affinity and ATPase activity, is a transcription ally active form of d-DctD.

FIG 3
FIG 3 Verification of the 1:1 binding ratio of IIA Glc and DctD in the d-IIA Glc /DctD D57Q complexes.Mixtures containing 5 µM d-IIA Glc and 20 µM DctD D57Q (A and B) or 20 µM d-IIA Glc and 5 µM DctD D57Q (C and D) were analyzed as described in Fig. 2A and B. Both chromatographic and electrophoretic profiles showed the presence of both proteins in the fractions 9, 10, 12, and 13 mL, as shown in Fig. 2.An extra peak corresponding to dimeric DctD D57Q (A) was formed in a mixture of 5 µM d-IIA Glc and 20 µM DctD D57Q , which was evidenced in an SDS-polyacrylamide gel (B, the fractions 14 and 15 mL were marked with a dashed box).SDS-PAGE analysis of a mixture of 20 µM d-IIA Glc and 5 µM DctD D57Q showed that the fractions 18 and 19 mL contained a protein with the MW of IIA Glc (D, marked with a dashed box).Its identity was verified by profiling d-IIA Glc (20 µM) using chromatographic and electrophoretic analyses (E and F).

FIG 4
FIG 4 Identification of the transcriptionally active form of d-IIA Glc /DctD D57Q complex.(A) DNA-binding ability of the multimeric forms of the d-IIA Glc /DctD D57Q complex.Approximately 100 nM of a labeled DNA probe used in Fig. 1B was incubated with 2-mer ([d-IIA Glc /DctD D57Q ] 2 ) and 12-mer ([d-IIA Glc /DctD D57Q ] 12 ) of the complex (500 nM [d-IIA Glc /DctD D57Q ] 1 -equivalents).Reaction mixtures were run on two identical 8% native polyacrylamide gels: one was for localizing the proteins (left panel) and the other was for localizing the DNA probes (right panel).For comparison of the bands corresponding to the d-IIA Glc / DctD D57Q complex bound to DNA (left panel) with the probe bound by d-IIA Glc /DctD D57Q complex (right panel), a dashed line was drawn parallel to the line positioning the loading wells.The first lanes in each gel were run with the whole fraction of the d-IIA Glc /DctD D57Q complex (left panel) and the labeled probe only (right panel).(B) ATPase activity of the multimeric forms of the d-IIA Glc /DctD D57Q complex.Both [d-IIA Glc /DctD D57Q ] 2 and [d-IIA Glc /DctD D57Q ] 12 were fractionated and aliquots containing 2 µM complexes ([d-IIA Glc /DctD D57Q ] 1 -equivalents) were subjected to the ATPase activity assay (11).One micromolar dimeric DctD D57Q ([DctD D57Q ] 2 ) was included in the assay.As a negative control, an ATPase-deficient mutant DctD, DctD H216R , was purified as a dephosphorylated state, and then its multimeric forms ([DctD H216R ] 2 , [d-IIA Glc /DctD H216R ] 2 , and [d-IIA Glc /DctD H216R ] 12 ) were added to the ATPase reaction mixture (black bars).The activity was presented as μM phosphate produced by μg of proteins per minute.P-values were indicated (**P < 0.001; ns, not significant).

FIG 5
FIG 5 Localization of [d-IIA Glc /DctD D57Q ] 12 -binding sites in the upstream regions of EPS-II and EPS-III clusters.(A and B) Two putative DctD-binding sites in the EPS-II and EPS-III clusters.Two sites (designated as BS1 for the remote site and BS2 for the close site from TIS-1) showing moderate identity to the DctD-binding sequences previously found in R. meliloti (5′-TGTGCGgaaatCCGCACA-3′; 15) were localized in the upstream regions of two gene clusters: the nucleotides positioned at −254 to −237 (BS1) and −142 to −125 (BS2) relative to RpoN-dependent TIS-1 of EPS-II (A) and the nucleotides positioned at −451 to −434 (BS1) and −160 to −143 (BS2) relative to the TIS-1 of EPS-III (B).Homologous nucleotides in each site were marked by boldfaces.In case of the EPS-II cluster (or brp operon), the regulatory sites interacting with the other transcription factors, such as BrpS and BrpT, were marked (16, 17).(C-F) Binding of [d-IIA Glc /DctD D57Q ] 12 to the probes containing either BS1 or BS2.DNA probes containing BS1 of EPS-II (C) or EPS-III (D), which were designated as P1 BS1 (a 210 bp DNA fragment from −418 to −209 of EPS-II and a 272 bp DNA fragment from −602 to −331 of EPS-III), were prepared for EMSA.Similarly, the probes containing BS2 of EPS-II (E) or EPS-III (F) were prepared, which were designated as P2 BS2 (a 270 bp DNA fragment from −208 to +62 of EPS-II and a 366 bp DNA fragment from −330 to +36 of EPS-III).Each probe was labeled and 100 nM was incubated with various concentrations of [d-IIA Glc /DctD D57Q ] 12 up to 100 nM.To verify the specific binding to P1 BS1 (C and D), the identical but unlabeled DNA fragments (cold probes) and the noncompetitive gap DNA were included.The resultant reaction mixtures were subjected to 8% native polyacrylamide gel electrophoresis, and DNA bands corresponding to the unbound or bound probes were indicated by arrows.Lanes 1, probe only; lanes 2, probe with 10 nM of [d-IIA Glc /DctD D57Q ] 12 ; lanes 3, probe with 20 nM of [d-IIA Glc /DctD D57Q ] 12 ; lanes 4, probe with 40 nM of [d-IIA Glc /DctD D57Q ] 12 ; lanes 5, probe with 60 nM of [d-IIA Glc /DctD D57Q ] 12 ; lanes 6, probe with 80 nM of [d-IIA Glc /DctD D57Q ] 12 ; lanes 7, probe with 100 nM of [d-IIA Glc /DctD D57Q ] 12 ; lanes 8, probe with 100 nM of [d-IIA Glc /DctD D57Q ] 12 and 0.5 µM of cold probe; lanes 9, probe with 100 nM of [d-IIA Glc /DctD D57Q ] 12 and 1.0 µM of cold probe; and lanes 10, probe with 100 nM of [d-IIA Glc /DctD D57Q ] 12 and 0.5 µM of gap DNA.

FIG 6
FIG 6 Effect of the mutations in the BS1s on binding by [d-IIA Glc /DctD D57Q ] 12 and transcription of the EPS clusters.(A and B) Mutagenesis of the BS1s of EPS-II and EPS-III clusters.The nucleotides in BS1s of EPS-II (A) and EPS-III (B), which are conserved in the R. meliloti DctD-binding consensus sequences (5′-TGTGCGgnnntCCGCACA-3′; 15), were marked with asterisks.These conserved nucleotides were substituted with other nucleotides, as indicated by arrows, to produce BS1M.(C and D) DNA-binding affinity of [d-IIA Glc /DctD D57Q ] 12 to the probes containing BS1M.DNA fragments of P1 containing BS1 (P1 BS1 ) or BS1M (P1 BS1M ) were prepared from the EPS-II (C) and EPS-III (D) clusters.Then, EMSA was performed as described in Fig. 5. Labeled DNA bands corresponding to the unbound or bound probes, which were resolved in 6% native polyacrylamide gel, were indicated by arrows.(E and F) Expression of the BS1M-containing transcription reporters of EPS-II and EPS-III clusters.The luxAB-based transcription reporters fused with the original upstream regulatory regions (P1 BS1 -P2 BS2 ; 18)