Sinorhizobium meliloti Fur-Like (Mur) Protein Binds a Fur Box-Like Sequence Present in the mntA Promoter in a Manganese-Responsive Manner

In Sinorhizobium meliloti, the Mur_Sm protein, a homologue ofthe ferric uptake regulator (Fur), mediates manganese-dependentregulation of the MntABCD manganese uptake system. In this study,we analyzed Mur_Sm binding to the promoter region of the S. melilotimntA gene. We demonstrated that Mur_Sm protein binds with highaffinity to the promoter region of mntA gene in a manganese-responsivemanner. Moreover, the results presented here indicate that twomonomers, or one dimer, of Mur_Sm binds the DNA. The bindingregion was identified by DNase I footprinting analysis and coversa region of about 30 bp long that contains a palindromic sequence.The Mur_Sm binding site, present in the mntA promoter region,is similar to a Fur box; however, manganese-activated Mur_Smbinds a canonical Fur box with very low affinity. Furthermore,the data obtained indicate that Mur_Sm responds to physiologicalconcentrations of manganese.

INTRODUCTION

Top

INTRODUCTION

Sinorhizobium meliloti belongs to the Rhizobiales group of bacteria.Rhizobia comprise different genera of Alphaproteobacteria whichhave in common the ability to fix nitrogen in association withleguminous plants. These bacteria can be found as free-livingorganisms in soil and in the rhizosphere or as endosymbiontwithin their legume host. In rhizobia, the homeostasis controlof some transition metals presents substantial differences fromthat of the well-studied group of Gammaproteobacteria (18, 33).Indeed, recent in silico studies lead to speculation that, ingeneral, the Alphaproteobacteria have a particular regulatorynetwork for iron and manganese homeostasis (32).

Manganese is essential for almost all organisms due to its participationin the detoxification of reactive oxygen species and enzymaticfunctions (23, 29). Two major classes of manganese transportsystems in bacteria have been described: the natural resistance-associatedmacrophage protein (NRAMP)-type transporters (mntH) and ABC-typetransporters such as mntABC and sitABC (29). S. meliloti 1021genome has a mntH orthologue gene (SMa1115) and the sitABCDoperon. Platero et al. (31) have previously shown that in S.meliloti, the SitABCD transport system is involved mainly inmanganese uptake and, accordingly, the operon was renamed mntABCD(30).

Metal homeostasis is maintained essentially through the controlof metal transport across the membranes. Different manganeseregulators that repress or activate transcription of genes involvedin manganese transport in response to the presence or absenceof manganese, respectively, have been identified. The MntR regulator,a member of the DtxR family of regulators, is the primary transcriptionalmanganese-regulatory protein identified so far. In addition,in some bacteria, the Fur (ferric uptake regulator) proteinand two regulators involved in defense against reactive oxygenspecies (PerR and OxyR) could also participate in the transcriptionalregulation of manganese uptake (21, 22). Interestingly, in S.meliloti and Rhizobium leguminosarum, a fur orthologue codesfor the primary manganese-responsive regulator of the mntABCD/sitABCDmanganese transport operon (8, 13, 30); accordingly, it hasbeen renamed Mur (13). Computational analysis revealed thatthe S. meliloti Mur protein (Mur_Sm) is included in a compactbranch comprising Fur homologs of the conventional Fur regulatorfound in almost all Alphaproteobacteria (32). In Escherichiacoli and other Gammaproteobacteria, the Fur protein is a globalregulator of iron metabolism (2). Microarray studies showedthat in S. meliloti, about 30 genes exhibited altered expressionin a mur_Sm mutant but, surprisingly, those of the mntABCD operonwere the only ones directly implicated in metal uptake (8).In S. meliloti and R. leguminosarum, a different protein, RirA,which belongs to the Rrf2 superfamily of regulators, is responsiblefor the regulation of most genes involved in iron uptake (9,34).

The DNA target for the metal-activated ferric uptake regulatorFur is called Fur box (12, 15). The canonical Fur box sequenceconsists of a palindromic region 19 or 21 bp long, dependingon the model of interpretation (TGATAATGATAATCATTATCA) (3, 12).The iron-activated E. coli Fur protein can bind the Fur boxas a monomer, as a dimer, as two dimers and, eventually, asan oligomer (10, 15, 25). There are different models for theinterpretation of Fur binding to the Fur box. It can be viewedas a 9-1-9 palindromic sequence, as three imperfect direct repeathexamers, as two direct and one inverted hexamer, or as twooverlapping 7-1-7 inverted repeats (3, 12, 15, 24). Fur interactionwith transition metals is also a controversial aspect. Althoughin vivo the classical Fur protein is a biosensor highly specificfor iron, in vitro assays show that different metals, such asMn²⁺, Cu²⁺, Cd²⁺, Co²⁺, and Zn²⁺, facilitate DNA binding (1,27). Interestingly, in Bradyrhizobium japonicum and R. leguminosarum,the Fur orthologues—Fur_Bj and Mur_Rl—are able torescue the phenotype of an E. coli fur mutant. Moreover, theseproteins bind in vitro to a canonical Fur box sequence (19,36). However, Fur_Bj and Mur_Rl differ markedly from the classicalFur protein. For instance, Fur_Bj recognizes and binds a DNAsequence different from the Fur box consensus sequence, whilethe E. coli Fur protein does not bind this region (17, 18).The Mur_Rl protein binds two sequences, named Mur-responsivesequence 1 (MRS1) (TGCAATT-N₇-AATTGCA) and MRS2 (TGCAAAT-N₇-AATCGCA),which are located upstream of the sitA gene in R. leguminosarum,in response to manganese and not to iron (13). The MRS motifis highly similar to a palindromic sequence found in the mntApromoter region in S. meliloti (30). In this study, we analyzedMur_Sm binding to the putative regulatory region of the S. melilotimntA gene. The recombinant protein was purified, and its activitywas tested. The Mur_Sm box was identified. Furthermore, we analyzeddifferent factors that influence DNA binding and determinedthe stoichiometry of the protein-DNA complex.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Bacteria and media.
The S. meliloti 1021 wild-type and mur_Sm mutant Mf1 strains(30) were grown at 30°C in M3 minimal defined medium plus37 µM FeCl₃ (5). E. coli strain DH5 ${alpha}$ was used for the propagationof plasmids, and E. coli strain BL21DE3(pLysS) was used forMur_Sm protein overexpression. E. coli strains were grown at37°C in Luria-Bertani medium with appropriate antibiotics.

Overexpression and purification of the Mur_Sm protein.
The mur_Sm gene was amplified from pFur_Sm, a plasmid derivedfrom pBSK carrying the mur_Sm gene and flanking regions (30),by PCR using Optimase polymerase (Transgenomic), a forward primer(5'-GCAATTCCATATGAGCCAGAGCAAG) encoding an NdeI restrictionsite and a reverse primer (5'-CGGGATCCTCAGTCCTTGCGC) encodinga BamHI restriction site for cloning in pET14b. Plasmid pET14bcontaining the mur_Sm gene was used to transform E. coli strainBL21(DE3)(pLysS). Overexpression and purification of the Mur_Smprotein was performed as described by Friedman and O'Brian (17).Fractions containing the Mur_Sm protein were combined and equilibratedwith 50 mM phosphate buffer, pH 8.0, containing 300 mM NaCland 5 mM imidazole by gel filtration through a Sephadex G-25PD-10 column (Amersham Biosciences). The His tag present inthe recombinant protein was removed by using a Thrombin CleanCleavekit (Sigma, St. Louis, MO). The His tag and undigested His-Mur_Smwere afterward removed by chromatography with a Ni-nitrilotriaceticacid-agarose (Ni-NTA) column. The digested protein, withoutthe His tag, was eluted with a gradient of imidazole from 20to 50 mM. Under these conditions, the His tag and the undigestedprotein were retained on the column. Purity of the digestedprotein was assessed by reverse-phase high-performance liquidchromatography (HPLC) by using a C-4 column, and the identitywas confirmed by matrix-assisted laser desorption ionization-timeof flight mass spectrometry. Total protein was quantified byLowry assay (26) and by densitometric measurements of bandsby using sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE) with bovine serum albumin (BSA) as a standard. Aliquotsof the digested recombinant Mur_Sm protein were stored at –80°Cin the presence of 10% glycerol.

DNase I footprinting experiment.
A 227-bp fragment containing the mntA-mur_Sm intergenic regionwas PCR amplified using primers DF1 (5'-GCTCTAGAGCAGGCCATCGCCG)and DF2 (5'-GCTCTAGACCCGCAGGATTCCTTCC), Taq polymerase, andpFur_Sm as template. The fragment obtained was digested withXbaI and cloned in both orientations in the unique XbaI siteof pUC18. Fragment orientation and DNA sequence were checkedby restriction analysis and DNA sequencing using M13/pUC universalprimers. Both EcoRI/PstI fragments from the obtained plasmidswere gel purified and end labeled at the EcoRI end with [ ${alpha}$ -³²P]dATPby using the Klenow fragment of DNA polymerase. For DNase Iprotection assays, labeled DNA fragments were diluted to approximately5 nM in 20 µl of electrophoretic mobility shift assay(EMSA) binding buffer supplemented with 2 µg BSA, 5 mMspermidine, and 1 µg of the nonspecific competitor poly(dI:dC)·(dI:dC)(Promega) and with or without 100 µM MnCl₂. Differentamounts of Mur_Sm protein diluted in the same buffer were added,and mixtures were incubated for 30 min at 4°C. DNA was digestedwith 0.5 U of DNase I for 5 min at 37°C, and digestionswere stopped by the addition of 5 mM EDTA, ethanol precipitated,dried, and suspended in 8 µl of formamide loading buffer(98% formamide, 10 mM EDTA [pH 8.0], 1 mg/ml xylene cyanol FF,1 mg/ml bromophenol blue). Digested probe products were separatedon a 6% denaturing polyacrylamide gel containing 7 M urea in1x Tris-borate-EDTA (pH 8.3) electrophoresis buffer. Protectedsequences were located by comparison with a G+A sequencing ladderobtained from a known sequence, as described previously (6).

EMSA.
The ability of the recombinant Mur_Sm protein to bind to thepromoter region of the mntABCD operon was determined by EMSA.For that purpose, an 82-bp fragment present in the mntA-mur_Smintergenic region covering the identified Mur-box and flankingsequences was PCR amplified with Optimase polymerase using 5'-AGGCGAATGGTCTGCATAAGTas the forward primer (EM1) and 5'-GACACTAGCCAAGGGGACAC as thereverse primer (EM2) (Fig. 1). The PCR product was purifiedfrom agarose gel and cloned in the SmaI site of pBSK. A 104-bpEcoRI/BamHI fragment from the plasmid obtained was gel purifiedand end labeled with [ ${alpha}$ -³²P]dATP using the Klenow fragment ofDNA polymerase. Binding reactions were performed with approximately1 nM probe corresponding to 2,000 cpm. Labeled probe was incubatedwith a solution of Mur_Sm protein and/or with soluble cell extracts,2 µg BSA, 5 mM spermidine, and 1 µg poly(dI:dC)·(dI:dC)in EMSA binding buffer (10 mM Tris-HCl, 40 mM KCl, 10 mM MgCl₂,1 mM dithiothreitol, 0.1% of the nonionic detergent nonylphenyl-polyethyleneglycol [NP-40] [pH 7.5]) at 4°C for 30 min. The Mur_Sm proteinstock solution was diluted in EMSA binding buffer prior to beingincluded in EMSA experiments. Cell extracts were prepared fromsaturated liquid cultures of S. meliloti strain 1021 or Mf1grown on M3 minimal defined medium supplemented with 37 µMFeCl₃. Washed cells were resuspended in 50 mM Tris-HCl (pH 8)buffer, and supernatants were obtained by passage of the cellsthree times through a French pressure cell at 25,000 lb/in²and were clarified by centrifugation at 10,000 x g. Metal saltswere freshly prepared as 1-M solutions and diluted in EMSA bindingbuffer to 1-mM working solutions.

View larger version (16K):
[in this window]
[in a new window]

FIG. 1. Region of the S. meliloti 1021 genome containing the DNA fragment used in this study. (A) Genetic organization of mur_Sm and mntABCD genes. (B) Sequences of the DNA fragments used in EMSA and footprinting assays. Primers used for EMSA and DNase I footprinting assays are underlined and double underlined, respectively. The region containing the palindromic Fur box-like sequence is highlighted in gray.

Samples were run at 20 mA in 6% nondenaturing polyacrylamidegels at 4°C in 0.5x Tris-borate (pH 8.3) buffer. Gels wereprerun for 30 min prior to the loading of samples. Subsequently,gels were dried by hot vacuum and subjected to autoradiography.Autoradiograms were developed on BioMax film (Eastman KodakCo.) and scanned, and signal intensities were quantified usingKodak 1D image analysis software. To determine the apparentdissociation constant (K_D) of Mur_Sm-DNA binding, the percentageof DNA bound to total labeled DNA was plotted against increasingMur_Sm concentrations. Apparent K_D was defined as the concentrationof protein at which 50% of the DNA is bound.

Ferguson analysis of the Mur_Sm-DNA complex.
A modification of the Ferguson analysis (28) was used to determinethe molecular weight of the Mur_Sm-DNA complex as described byFriedman and O'Brian (18). Binding reactions were done as describedabove, with or without 100 nM Mur_Sm, and electrophoresed alongwith protein markers in a series of native polyacrylamide gelscontaining 0.25x Tris-borate buffer in a range of 6 to 10% acrylamidegels (29:1 ratio of acrylamide:bisacrylamide). Lanes containingradiolabeled samples were dried and subjected to autoradiography.The rest of the gel was stained with Coomassie blue. The logarithmof the relative mobility (R_f) for each species (native proteins,DNA, and DNA-protein complexes) was plotted against acrylamideconcentration. Then, the negative slope of R_f for each proteinstandard was plotted against its molecular weight to generatea standard curve from which the estimated molecular weight correspondingto the EMSA complex was interpolated. Since the molecular weightof the DNA probe was known, this value was subtracted from theestimated molecular weight of the complex and divided by 16,435(the molecular weight of the Mur_Sm monomer). In this way, thenumber of Mur_Sm monomers present in the gel-shifted band wasestimated.

RESULTS

Top

RESULTS

Overexpression and purification of recombinant Mur_Sm protein.
In order to improve the yield of Mur_Sm, we decided to overexpressthe protein in E. coli strain BL21(DE3)(pLysS) by followingstandard procedures (see Materials and Methods). Results ofSDS-PAGE of fractions obtained after expression and purificationof the Mur protein from S. meliloti 1021 are shown in Fig. 2A.According to the methodology employed, the recombinant proteinincludes a six-histidine tag at the N-terminal end that canbe removed by enzymatic digestion with thrombin. As shown inFig. 2B, more than 95% of the Mur_Sm-His protein was digestedby thrombin. The identity of the digested recombinant proteinwas confirmed by matrix-assisted laser desorption ionization-timeof flight mass spectrometry. The molecular mass was determinedto be 16,435 Da, which is in agreement with the calculated massof Mur_Sm—16,183 Da—plus three amino acids (glycine,serine, and histidine) introduced at the N-terminal end duringthe cloning procedure. The yield of the purification processof recombinant Mur_Sm protein was calculated as about 15 to 20mg of protein per liter of cell culture, with a purity aboveor over 95%, according to SDS-PAGE and HPLC analyses.

View larger version (61K):
[in this window]
[in a new window]

FIG. 2. Purification of the recombinant Mur_Sm protein. (A) SDS-PAGE of different fractions obtained in the Mur_Sm expression and purification process. Lane 1, soluble extract of IPTG (isopropyl-ß-D-thiogalactopyranoside)-induced E. coli (pET14b) cells; lane 2, eluted fraction from the Ni-NTA column loaded with soluble extract of E. coli (pET14b) cells; lane 3, soluble extract of uninduced E. coli (pETMur_Sm) cells; lanes 4 and 5, soluble extract of IPTG-induced E. coli (pETMur_Sm) cells; lane 6, purified recombinant Mur_Sm-His protein eluted from the Ni-NTA column. (B) His tag removal from the recombinant Mur_Sm-His protein. Lane 1, molecular weight markers (weights [in thousands] are shown at left); lane 2, recombinant Mur_Sm-His protein; lane 3, recombinant Mur_Sm protein after thrombin digestion.

Manganese-activated Mur_Sm binds to a DNA sequence similar to a Fur box-like sequence.
We and others have previously shown that Mur_Sm mediates manganese-dependentrepression of mntA in vivo (8, 30). Moreover, in the mntA presumptivepromoter region, a putative Fur box was identified (13, 30).In an attempt to define the Mur_Sm binding site, protein-DNAinteraction was analyzed by DNase I footprinting. The DNA probecontained the mntA-mur_Sm intergenic sequence and the first 50bases of the mntA and mur_Sm genes and included the putativeFur box (Fig. 1). A protected region was detected in both strandswith manganese-activated Mur_Sm. No additional protected regioncould be detected even when incubation mixtures were done with500 nM Mur_Sm, suggesting that no additional interactions wereformed in these conditions (Fig. 3A). The region protected fromDNase I nicking was 31 bp long (CTAGTTGCAAATGCTTCTCATTTGCATTGAC)and contains a palindromic region (underlined). These featuresare similar to ones found in the classical Fur box and the Murbox identified for R. leguminosarum (32).

View larger version (122K):
[in this window]
[in a new window]

FIG. 3. (A) DNase I footprinting analysis of the mntA promoter region protected by Mur_Sm. Lane 1 and 2, coding strand (CS) without Mur_Sm protein; lane 3, CS plus 5 nM Mur_Sm plus 100 µM MnCl₂; lane 4, CS plus 50 nM Mur_Sm and 100 µM MnCl₂; lane 5, CS plus 100 nM Mur_Sm and 100 µM MnCl_2; lane 6, CS plus 500 nM Mur_Sm and 100 µM MnCl₂; lanes 7 and 8, noncoding strand (NCS; without protein) plus 100 µM MnCl₂; lane 9, NCS plus 5 nM Mur_Sm and 100 µM MnCl₂; lane 10, NCS plus 50 nM Mur_Sm and 100 µM MnCl₂; and lane 11, NCS plus 5 nM Mur_Sm without manganese addition. A+G is the sequencing ladder used to locate the sites of protection. (B) Comparison between DNA sequences of the canonical Fur box (above) and the Mur_Sm protected region (below). Conserved bases are shown in gray type in the Fur box.

Mur_Sm protein binds the promoter region of the mntA gene in a manganese-responsive manner.
In order to determine the functional role of the Mur_Sm box identified,we performed EMSA with a fragment containing the Mur_Sm box (Fig.1) and the recombinant Mur_Sm protein. Figure 4 shows the gelmobility shift assay when incubations were done with differentconcentrations of the Mur_Sm protein and in the presence or absenceof 100 µM MnCl₂. Under the conditions used, the presenceof one complex could be detected either with or without MnCl₂;nonetheless, lower protein concentrations were required forMur_Sm-DNA complex formation in the presence of metal (Fig. 4A).The K_D was determined by plotting bound DNA as a function ofprotein concentration (Fig. 4B). The K_D of the Mur_Sm-DNA complexformed in the presence of 100 µM MnCl₂ was calculatedas 1 nM. This binding affinity is higher than the reported forMur_Rl binding to MRSs for R. leguminosarum (7, 13) and is similarto the Fur_Bj affinity to the irr promoter in B. japonicum (18).Meanwhile, in the absence of metal, the affinity of Mur_Sm forits target DNA was nearly 60-fold lower than with manganese.These results clearly indicate the metal dependence of Mur_Smbinding.

View larger version (89K):
[in this window]
[in a new window]

FIG. 4. Binding of the Mur_Sm recombinant protein to the mntA promoter region. (A) EMSAs were carried out using 1 nM of ${alpha}$ ³²-P-labeled DNA probe and increasing amounts of purified recombinant Mur_Sm protein (indicated below each lane) plus 100 µM MnCl₂ or in the absence of added metals. (B) Affinity of Mur_Sm for DNA was calculated by plotting the percentage of shifted probe versus protein concentration in the presence or absence of MnCl₂. The K_D was defined as the protein concentration required to shift 50% of the probe.

To further investigate the role of other divalent metals inthis interaction, EMSAs were done in the presence of iron andzinc. Although the affinity of Mur_Sm for DNA was higher in presenceof zinc or iron than in the absence of metal, K_Ds for each ofthose metals were still tenfold higher than for manganese (Fig.5). These results suggest that other divalent metals could alsoincrease the Mur_Sm binding activity in vitro, albeit less efficientlythan Mn²⁺.

View larger version (14K):
[in this window]
[in a new window]

FIG. 5. Effect of different metals on Mur_Sm-DNA binding. The Mur_Sm affinity for DNA in the presence of 100 µM Fe²⁺, 100 µM Zn²⁺, or 100 µM Mn²⁺ was compared by plotting the percentage of shifted probe versus protein concentration in the presence of the different metals.

To determine metal concentrations sensed by Mur_Sm, we performedan EMSA with 1 nM Mur_Sm and 5, 15, 25, 50, or 100 µM Mn²⁺.The formation of a protein-DNA complex could be detected inthe presence of at least 25 µM Mn²⁺, whereas a band shiftcould not be detected at lower metal concentrations (data notshown).

Since the amino acid sequences of Mur_Sm and E. coli Fur proteinsare highly similar (42% identity), we asked if the Mur_Sm proteinwas able to recognize a "classical" Fur box. For this purpose,we performed an EMSA using a radiolabeled probe containing theE. coli aerobactin promoter. The promoter region of the aerobactinoperon has been used as a model for studies of DNA-Fur interaction(11, 12, 15). We found that in the presence of Mn²⁺ ions, aMur_Sm concentration of about 500 nM was necessary to shift 50%of the probe, yet a concentration as high as 1 µM wasnot able to entirely shift the probe (Fig. 6). These resultsprovide evidence that Mur_Sm binds the canonical Fur box withlow affinity.

View larger version (8K):
[in this window]
[in a new window]

FIG. 6. Mur_Sm binding to the E. coli aerobactin promoter. EMSAs were carried out by using 1 nM of ${alpha}$ -³²P-labeled DNA probe and increasing amounts of purified recombinant Mur_Sm protein in the presence or absence of 100 µM MnCl₂. The affinity of Mur_Sm for DNA was calculated by plotting the percentage of shifted probe versus protein concentration in the presence of MnCl₂. The K_D was defined as the protein concentration required to shift 50% of the probe.

Stoichiometry of the Mur_Sm-DNA complex.
EMSAs done for B. japonicum and R. leguminosarum showed thatFur_Bj and Mur_Rl proteins form both high-mobility and low-mobilitycomplexes with their target DNA. According to the apparent massesof protein-DNA complexes, it has been suggested that the high-mobilitycomplex corresponds to a dimer bound to DNA, whereas a tetrameror two dimers are present in the low-mobility complex (7, 18).We wanted to test if this was the case for the Mur_Sm-DNA interaction.However, under the conditions assayed, only one band shift wasdetected, even at a Mur_Sm concentration as high as 300 nM (Fig.4A). In order to determine the stoichiometry of the Mur_Sm-DNAcomplex, we measured the apparent mass by native PAGE usinggels containing different acrylamide concentrations (Fig. 7A).Compared with protein standards, the molecular mass of the complexwas estimated as 106 kDa (Fig. 7B). Taking into account thatthe molecular mass of the 104-bp target DNA used correspondsto about 68 kDa, the molecular mass obtained for the complexindicates that Mur_Sm is bound to the DNA as two monomers oras a dimer.

View larger version (13K):
[in this window]
[in a new window]

FIG. 7. Stoichiometry of the Mur_Sm-DNA complex. (A) Plot of the logarithm of the R_f of standard proteins and Mur_Sm-DNA complex against the percentage of polyacrylamide gel concentration. (B) The slope for each species (retardation coefficient) was plotted against molecular masses (MM) of proteins. The molecular weight of the Mur_Sm-DNA complex was estimated from the curve. Protein standards used were chicken egg albumin (circles; 45 kDa), bovine serum albumin monomer (diamonds; 66 kDa), E. coli His-aconitase A (triangles; 98 kDa), and E. coli glutathione S-transferase-aconitase B (squares; 130 kDa). Closed circles represent the Mur_Sm-DNA complex.

Auxiliary proteins are not involved in native Mur_Sm-DNA binding.
In order to analyze if native Mur_Sm protein had DNA bindingactivity, EMSAs were done using soluble cell extract of S. melilotistrain 1021. Cell extract obtained from a mur_Sm knock out mutant,S. meliloti strain Mf1 (30), was used as a negative control.As shown in Fig. 8, incubation of DNA with soluble S. meliloti1021 extract was sufficient to form a complex (lane 8), whereasa band shift could not be detected in the presence of solubleMf1 extract (lane 9). These results clearly indicate that theMur_Sm protein expressed in S. meliloti 1021 cells in vivo isinvolved in the formation of the mntA promoter complex. Moreover,a similar migration pattern was found for the complex formedby the recombinant protein and the cell extract.

View larger version (54K):
[in this window]
[in a new window]

FIG. 8. Binding of the native Mur_Sm protein to DNA. EMSAs were performed in the presence of soluble cell extracts of S. meliloti strain 1021 (wild type) or soluble cell extracts of the mur_Sm mutant strain (Mf1). Lane 1, free probe; lane 2, 0.3 nM Mur_Sm; lane 3, 1 nM Mur_Sm; lane 4, 0.3 nM Mur_Sm plus wild-type cell extract; lane 5, 1 nM Mur_Sm plus wild-type cell extract; lane 6, 0.3 nM Mur_Sm plus Mf1 cell extract; lane 7, 1 nM Mur_Sm plus Mf1 cell extract; lane 8, wild-type cell extract (without added recombinant Mur_Sm protein); and lane 9, Mf1 cell extract (without added recombinant Mur_Sm protein).

To test the involvement of auxiliary protein in the native complex,the target DNA was incubated with cell extracts of S. melilotistrains 1021 and Mf1 supplemented with the recombinant Mur_Smprotein. As shown in Fig. 8, a similar migration shift was obtainedfor the DNA incubated with the pure protein (lanes 2 and 3),the pure protein plus wild-type strain cell extract (lanes 4and 5), and the pure protein plus the Mf1 cell extract (lanes6 and 7), suggesting that auxiliary proteins are not presentin the Mur_Sm-DNA complex obtained in vivo.

DISCUSSION

Top

DISCUSSION

Control of manganese uptake is complex, and different effectorsand regulators have been reported to be involved (21). Usually,transcriptional control of the sitABCD/mntABCD operon is repressedmainly by manganese through MntR (a member of the DtxR familyof regulators). Interestingly, in Salmonella enterica serovarTyphimurium, both the "classical" Fur and the MntR proteinscan regulate sitABCD in response to iron or manganese, respectively,even though both regulators can use alternative metals to partiallyrepress transcription (22). MntR orthologs are not found inS. meliloti and R. leguminosarum; in these bacteria, manganeseuptake is controlled by Mur, while high-affinity iron transportis regulated mainly by RirA. In this study, we demonstratedthat Mur_Sm binds a DNA region present in the S. meliloti mntApromoter. In the presence of metals, Mur_Sm binds its targetDNA with high affinity. Mur_Sm-DNA binding was also detectedin the absence of metals, although it occurred less efficiently,indicating that the binding is metal dependent. These resultsare in agreement with the classical model of action describedfor this family of regulators.

In S. meliloti and R. leguminosarum, physiological repressionof mntA/sitA genes by Mur is highly selective for manganese(13, 30). However, Bellini and Hemmings (7) found in vitro alack of metal discrimination for Mur_Rl binding to its targetDNA, an observation that has also been reported for E. coliFur binding to the Fur box (27). In our system, a preferencefor manganese could be determined (K_D value of ca. 1 nM), althoughthe other metals tested also improved Mur-binding activity (K_Dvalue of ca. 10 nM). We do not know if these differences aresufficient to explain the selectivity observed in vivo. In thiswork, we have also provided evidence for the absence of auxiliaryproteins in the complex formed with the cell extract. However,we cannot exclude the involvement of accessory factors, suchas metallochaperones, that could facilitate metal transfer orspecial associations in cell sites, improving the metal accessibilityof Mur_Sm.

We determined that a manganese concentration of at least 25µM is required for the DNA complex formation in the presenceof 1 nM Mur_Sm. It was previously reported that total intracellularmanganese concentration in bacteria is maintained in the rangeof 10 to 100 µM (16). In the hypothetical situation thatmost of this metal is free or accessible to Mur_Sm, our interpretationof the data is that when the intracellular manganese concentrationis higher than ca. 25 µM, a manganese high-affinity uptakesystem is not required; therefore, repression is activated throughMur_Sm binding to the regulatory response element.

It has been suggested that the Mur_Rl protein binds as a dimeror as two dimers in two regions present in the promoter regionof R. leguminosarum sitABCD, termed the MRS (13). Although MRS1and MRS2 were first described as DNA binding targets atypicalof the Fur superfamily of regulators (14), further computationalanalysis demonstrated that the defined MRS motif was similarto known Fur-binding motifs (32). Here, we found that even inthe presence of 300 nM manganese-activated Mur_Sm protein, onlyone complex could be detected by using EMSA. Moreover, the resultspresented here indicated that this band corresponds to a dimer(or two monomers) bound to DNA. By using DNase I footprinting,we identified an approximately 30-bp-long protected sequencethat included the predicted Mur box sequence. Fourteen nucleotidespresent in the palindromic region are conserved within the 21-bpFur box sequence (Fig. 3B). It has been suggested that at least11 conserved base pairs should be present for a sequence tobe considered a Fur box-like sequence (4, 18, 35); therefore,according this criterion, we can conclude that the Mur_Sm boxis a Fur box-like sequence.

The high amino acid sequence similarity found between Mur andFur proteins could suggest similar physiological roles (32,33). However, our study indicates that subtle differences betweenthese proteins and their targets specify distinct cellular responsesin transition metal homeostasis. It is also worth noting thatsome key residues for metal binding and protein folding, suchas histidines and cysteines, are not conserved in comparisonsof the different protein compositions (20). For instance, R.leguminosarum and S. meliloti Mur proteins have 7 histidinesand only 1 cysteine, B. japonicum Fur contains 6 histidinesand 3 cysteines, and E. coli Fur has 12 histidines and 4 cysteines.Besides, in P. aeruginosa Fur and S. meliloti Mur proteins,the positions of the unique cysteine are markedly different.According to currently available data, it is not possible topredict a physiological role for Fur/Mur based only on proteinsequence comparisons.

Taken together, our observations strengthen the idea that auniversal model for predicting the Fur function based on studiesperformed with only few species cannot be extrapolated to allspecies of bacteria and reinforce the idea that there is morediversity among regulators involved in manganese homeostasisthan previously considered.

ACKNOWLEDGMENTS

This work was partially supported by PEDECIBA Química/Biologíaand by Fondo "Prof. Clemente Estable"-PDT-10088, Uruguay.

We are very grateful to Paul R. Gill for critical readings ofthe manuscript and to Florencia Pratto for her kind help withEMSA and footprinting experiments.

FOOTNOTES

* Corresponding author. Mailing address: Laboratorio de Ecología Microbiana, IIBCE, Av. Italia 3318, Montevideo 11600, Uruguay. Phone: 598 (2) 4871616, ext. 146. Fax: 598 (2) 4875548. E-mail: efabiano{at}iibce.edu.uy

Published ahead of print on 8 June 2007.

REFERENCES

Top