![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
Department of Molecular Biology, The Hebrew University Medical School, P.O. Box 12272, Jerusalem 91120, Israel
Received 29 July 2007/ Accepted 20 September 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
BglF consists of three domains, A, B, and C. The A and B domains are hydrophilic, the latter containing the active-site cysteine. The two hydrophilic domains are connected by the hydrophobic C domain, which presumably forms the sugar translocation channel and at least part of the sugar-binding site (43). The C domain was recently proposed to contain, in addition to eight transmembrane (TM) helices, a reentrant loop and a segment with inherent dynamics, both speculated to be implicated in BglF function(s) (70). Catalysis of the sugar-stimulated functions involves specific interactions between the active-site-containing domain and the membrane domain (12; also S. Yagur-Kroll and O. Amster-Choder, unpublished observations). The use of the same active site for all the (de)phosphorylation reactions that BglF catalyzes suggests that the protein contains different recognition sites for the various substrates, i.e., BglG, P-BglG, and the sugar. A likely scenario is that binding of the sugar shifts the equilibrium by exposing a recognition site for P-BglG. The finding that both the active-site-containing B domain and the membrane C domain are required for BglG dephosphorylation and sugar phosphotransfer, whereas the active-site-containing B domain alone is sufficient for BglG phosphorylation (8), suggests that a site(s) within the membrane C domain is involved in P-BglG dephosphorylation.
To address the question of how the different functions of BglF are coordinated and to study the structural components involved in the opposing functions of BglF, we attempted to isolate BglF mutants that can inhibit BglG activity by phosphorylation but cannot relieve the inhibition by dephosphorylation, i.e., mutants in which the balance is shifted toward BglG phosphorylation. Such mutants were isolated using random mutagenesis and an in vivo screen based on BglG-mediated antitermination of a reporter gene. The mutations, mostly single amino acid substitutions, were mapped to the B and C domains of BglF. As expected, many of the mutants were also impaired in sugar phosphotransfer, an activity which is coupled to dephosphorylation of P-BglG, but some demonstrated normal or almost normal phosphotransfer activity. The residues whose substitution affected just the ability of BglF to dephosphorylate BglG are located in the membrane C domain, either in the TMs flanking the reentrant loop or in the dynamic segment, i.e., in regions suggested to be involved in function, based on biochemical analyses of BglF membrane topology (70). We present biochemical evidence that one of these residues is spatially proximal to the active-site cysteine in the hydrophilic B domain. Bioinformatic analyses propose that this position is part of a novel motif in BglF homologues. This motif resembles a motif found in ion channels and may perhaps be involved in phosphate delivery. Using a second genetic screen, we isolated BglG mutants that suppress the defect in the BglF mutants and restore their ability to dephosphorylate P-BglG.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-32P]ATP (6,000 Ci/mmol) was obtained from PerkinElmer. [32P]PEP was prepared and separated from [32P]ATP as described before (2). Growth media. LB medium and M63 salts minimal medium were prepared essentially as described by Miller (50). M9 salts minimal medium was prepared as described by Miller (50), with the exception that 0.01% thiamine, 0.1 mM CaCl2, and no Casamino Acids were used. MacConkey-lactose plates were prepared from lactose-MacConkey agar (Difco). MacConkey-arbutin and -salicin plates were prepared as described previously (59).
Bacterial strains.
The following E. coli K-12 strains were used: MA152 and MA200-1, both carrying a bgl'-lacZ fusion on their chromosome (
bglR7 bglG' lacZ+ lacY+), but whereas the first is 
bgl, the second is bgl+ and carries a defective bglF gene (49). E. coli BL21(DE3) [F– ompT hsdSB(rB– mB–) gal dcm (DE3)], obtained from Novagen, was used for expression of His-tagged proteins. E. coli MC1061 [(hsdR, mcrB, araD, 139(araABC-leu)7679 lacX74 galU galK rpsL thi] was used for expression of maltose-binding protein (MBP)-tagged proteins. E. coli K38 (HfrC trpR thi ) was used for overexpression of the Bgl proteins used in the in vitro phosphorylation assay and for metabolic labeling of proteins with [35S]methionine. In addition, the Salmonella enterica serovar Typhimurium strain LJ144 (cpd-401 cysA1150/F'198), which contains the pts operon on the E. coli plasmid F'198 and thus produces increased levels of enzyme I (EI), HPr, and EIIAGlc, was used for the in vitro phosphorylation assay (58).
Plasmids. Plasmid pT7-F(NheI) is a derivative of pT7OAC-F (2), which contains an NheI restriction site 13 bp downstream of the first codon of the bglF open reading frame. Briefly, it was constructed by replacing the 215-bp HindIII-NsiI fragment from pT7OAC-F with a similar HindIII-NsiI fragment that contains an NheI restriction site, which was generated by overlap extension with PCR (33).
A plasmid encoding BglF with M22K as a single mutation was constructed by replacing the 370-bp PstI-StuI fragment of pT7OAC-F with a similar fragment that contains the mutation, which was generated by overlap extension with PCR (33).
A plasmid encoding BglF with G64S as a single mutation was constructed by replacing the 266-bp HpaI-SacII fragment of pT7OAC-F with a similar fragment from a derivative of pT7OAC-F encoding BglF with two mutations, G64S and V77M, which were generated by random mutagenesis.
A plasmid encoding BglF with G408D as a single mutation was constructed by replacing the 220-bp AccI-Bpu10I fragment of pT7OAC-F with a similar fragment that contains the single mutation, which was generated by overlap extension with PCR (33).
pBAD18-F is a pBAD18 derivative (28) which carries the bglF gene cloned downstream from the PBAD promoter. In short, a 2,040-bp XbaI fragment from pET15b-F (47) was cloned into the XbaI site in pBAD18. Insertion in the proper orientation was verified. For the ß-glucoside utilization assays, the different mutations were introduced into the bglF gene present on pBAD18-F as follows. The mutations in segment B (Table 1) were transferred by replacing the 237-bp NdeI-StuI fragment of pBAD18-F with the respective fragments from the pT7OAC-F derivatives, obtained by PCR amplification. The mutations in segments C1 and C4 (Table 1) were transferred by replacing the 1,185-bp StuI-Bpu10I fragment of pBAD18-F with the corresponding fragments that contain the different mutations from the pT7OAC-F derivatives.
|
A plasmid encoding BglG with P118L as a single mutation was constructed by replacing the 220-bp AflII-HincII fragment of pZGM-G with a similar fragment from a derivative of pZGM-G encoding BglG with three mutations, P118L, R22Q, and Y50F, which were generated by random mutagenesis.
pZGM-B, used for the salicin phosphotransfer assay, is a pZS*2tet-1MCS1 derivative (48) that expresses the bglB gene from the PLtetO-1 promoter. It was constructed by replacing the 831-bp KpnI-MluI fragment of pZGM-G with a 1,430-bp KpnI-MluI fragment obtained by amplifying the bglB gene from the chromosome of BL21(DE3) by PCR. Induction of the bglB gene was achieved by the addition of 100 ng of anhydrotetracycline hydrochloride to the growth medium.
Plasmids coding for BglF proteins with two cysteines, either G106C and C24 or S146C and C24 (di-Cys mutants), were constructed by replacing a 285-bp StuI-NcoI fragment from pT7SYF26 (encoding Cys24 as a single Cys) (70) with a similar fragment that contained either the G106C or the S146C mutation; the mutations were generated by overlap extension with PCR (33). Plasmids coding for BglF proteins with two cysteines, either G408C and C24 or Q416C and C24 (di-Cys mutants), were constructed similarly by replacing a 579-bp NheI-EcoRV fragment from pT7SYF26 with a similar fragment that contained either the G408C or the Q416C mutation. The mutations introduced or eliminated sites for restriction enzymes.
All mutations were confirmed by sequencing. More details on plasmid constructions are available from the authors upon request.
Construction of BglF and BglG mutant libraries. The bglF mutant library was constructed by random PCR mutagenesis essentially as described by Fromant et al. (18), with some modifications. Five fragments of the bglF gene, spanning nucleotides –52 to 247, 200 to 510, 363 to 785, 741 to 1310, and 1093 to 1440, were amplified under three different error-prone conditions, obtained by varying the MgCl2 and deoxynucleoside triphosphate concentrations. Each reaction mixture (25 µl) contained 2 units of Taq polymerase (Promega), 50 pmol of forward primer, 50 pmol of reverse primer, and template DNA (600 pM) in 10 mM Tris-HCl (pH 9), 50 mM KCl, 0.1% Triton X-100, and 0.5 mM MnCl2. The three reaction mixtures contained MgCl2 at final concentrations of 2.5, 3.32, or 3.28 mM. The final concentrations of the deoxynucleoside triphosphates in the three reaction mixtures were as follows: 0.35 mM dATP, 0.4 mM dCTP, 0.2 mM dGTP, and 1.35 mM dTTP; 0.22 mM dATP, 0.2 mM dCTP, 0.34 mM dGTP, and 2.36 mM dTTP; 0.12 mM dATP, 0.1 mM dCTP, 0.36 mM dGTP, and 2.5 mM dTTP. Fifteen cycles of amplification were performed, each consisting of 60 s of denaturation at 91°C, 60 s of primer annealing at 44°C, and 60 s of primer extension at 72°C. The first PCR product was digested with NheI-StuI and ligated to pT7-F(NheI) digested with the same enzymes. The remaining four PCR products were digested with StuI-SacII, SacII-BsabI, BsabI-AccI, or AccI-Bpu10I and ligated to pT7OAC-F (2) digested similarly. Thus, the five wild-type fragments were replaced by the respective randomly mutated fragments to create five banks of bglF mutants. The amino acid residues encoded by the five fragments are as follows: segment B, A5 to G79; segment C1, L80 to A163; segment C2, G164 to V254; segment C3, I255 to V404; and segment C4, Y405 to P474. Each bank was introduced into the MA200-1 strain by electroporation. Transformed cells were plated on MacConkey-lactose plates and replica plated onto MacConkey-lactose plates containing 7 mM ß-MG. Colonies that were white on both types of plates were isolated, and plasmid DNA was purified by a Promega miniprep kit.
The library of bglG mutants was constructed by amplifying two fragments of the bglG gene, spanning nucleotides –10 to 682 or 297 to 846 under error-prone PCR conditions, essentially as described above for the bglF library. The PCR products were digested with KpnI-HincII and HincII-MluI, respectively, and ligated to pZGM-G digested similarly, thus creating two banks of bglG mutants. The mutants were introduced by electroporation into the MA152 strain that contained a pT7OAC-F derivative encoding BglF with one of the following mutations: S146R, G106D, G408D K466R, or FQ417H. Transformed cells were plated onto MacConkey-lactose plates containing 200 µg/ml ampicillin, 25 µg/ml chloramphenicol, and 30 ng/ml anhydrotetracycline hydrochloride; cells were replica plated onto MacConkey-lactose plates containing 10 mM salicin. Colonies that were white on MacConkey-lactose plates and red or pink on MacConkey-lactose plates with salicin were picked, and plasmid DNA was purified with a Promega miniprep kit. To isolate the pZGM-G derivatives, the pT7OAC-F derivatives were digested with BsaI, which does not cleave the pZGM-G derivatives, and the DNA was used to transform MA152. Only colonies that grew in the presence of chloramphenicol but not in the presence of ampicillin were picked.
All mutations were mapped by sequencing.
ß-Gal assay. Assays for ß-galactosidase (ß-Gal) activity were carried out as described by Miller (50). Cells were grown in minimal medium supplied with 0.4% succinate as a carbon source. When ß-Gal assays were performed with MA152 cells containing pT7OAC-F derivatives and pZGM-G derivatives, growth was in the presence of 2 mM cyclic AMP.
ß-Glucoside utilization assays. To qualitatively estimate the ability of the BglF variants to transfer ß-glucosides into the cells while phosphorylating them, MA200-1 cells expressing the different bglF mutant alleles were plated on MacConkey-salicin and MacConkey-arbutin plates. Utilization of salicin was indicated by the growth of red colonies on MacConkey-salicin plates. Utilization of arbutin was indicated by growth of brown colonies on minimal-arbutin and MacConkey-arbutin plates. The brown coloring results from the hydroquinone generated during arbutin hydrolysis.
To quantitatively estimate the ability of the BglF variants to phosphotransfer ß-glucosides, we measured hydrolysis of p-NPG and salicin. ß-Glucoside hydrolysis directly correlates with ß-glucoside phosphotransfer, because sugar phosphorylation is a prerequisite for efficient sugar transfer. To measure p-NPG hydrolysis, the assay described by Schaefler (59) was carried out with minor modifications. MA200-1 cells expressing the BglF mutants were grown in M63 minimal medium supplemented with 0.4% succinate as a carbon source at 37°. When cells reached an optical density at 600 nm (OD600) of 0.4 to 0.5, they were put on ice for 20 min, and their OD was measured at 600 nm. Cells (1 ml) were pelleted by centrifugation, washed, and resuspended in 0.9 ml of 0.075 M phosphate buffer, pH 7.5, containing 1 mM MgSO4. The reaction was started by adding 0.1 ml of 20 mM p-NPG (final concentration, 2 mM). Incubation was at 37° until the appearance of yellow colonies. The reaction was then stopped by the addition of 0.5 ml of 2 M Na2CO3. Cells were pelleted by centrifugation and the amount of p-nitrophenol was estimated by measuring absorbance at 410 nm. The units of enzyme activity were determined using the following equation: units of activity = (OD410 x 1,000)/(OD600 x V x T), where T is time in minutes and V is the volume of cells in milliliters. To measure salicin hydrolysis, the saligenin assay, which measures specifically the cleavage of phospho-salicin, was carried out using a procedure similar to that described by Schaefler (59). Cells were grown to mid-exponential phase in M9 minimal medium with 0.4% succinate as a carbon source, 0.005% arabinose to induce expression of the bglF derivatives cloned in pBAD18-F, and 100 ng of anhydrotetracycline hydrochloride to induce bglB expression. One milliliter of cells was harvested, washed in 0.8% saline, and resuspended in 0.3 ml of saline. A total of 0.2 ml of 0.1 M salicin was added, and the mixture was incubated at 37°C for 20 min. The reaction was stopped by the addition of 0.5 ml of 2 M Na2CO3. Production of saligenin by cleavage of phospho-salicin was detected by the addition of 0.5 ml of 0.6% 4-amino-antipyrine, followed by the addition of 0.5 ml of 4% K3Fe(CN)6 after 15 min at room temperature. A positive reaction was indicated by the appearance of red colonies and was quantitated by measuring the absorbance at 509 nm. Units of enzyme activity were determined using the following equation: units of activity = (OD509 x 1,000)/(OD600 x V x T).
The above assays depended on expression of the respective enzymes that hydrolyze the different ß-glucosides. BglB, which metabolizes salicin, was expressed from a plasmid, because the chromosomal mutation in the bglF gene of MA200-1 is polar on the downstream bglB gene. Hydrolysis of arbutin and p-NPG is catalyzed by BglA, which is constitutively expressed from the chromosome.
Protein purification. MBP-tagged BglG was expressed in MC1061 and purified as described previously (10). His-tagged IIBbgl was expressed in BL21(DE3) and purified as described before (19).
In vitro phosphorylation. Membrane fractions enriched for the various BglF derivatives were prepared as described previously (2). In vitro phosphorylation of BglF-containing membranes and BglG-containing extracts and dephosphorylation of 32P-labeled BglG were carried out as described before (2). Phosphorylation and dephosphorylation of MBP-BglG were carried out essentially as described previously (21), except that the 32P-IIBbgl used in the phosphorylation reaction was labeled by incubating 54 µg of purified IIBbgl in the presence of [32P]PEP and the soluble fraction of S. enterica serovar Typhimurium LJ144 under the conditions described in Amster-Choder et al. (2). Densitometry analyses of band intensity were performed using ImageJ software (NIH, Bethesda, MD).
In vivo cross-linking. The methods used for growth, induction, and specific labeling of the di-Cys BglF derivatives with [35S]methionine were essentially as described previously (2). In short, E. coli K38 cells containing derivatives of pT7OAC-F, which encode single- or di-Cys BglF proteins cloned under T7 promoter control, and pGP1-2, which encodes the heat-inducible T7 RNA polymerase gene, were grown in minimal medium with 0.4% succinate as the sole carbon source. At an OD600 of 0.3, expression of the T7 RNA polymerase, and hence of BglF, was induced by shifting the culture to 42°C. After 20 min, rifampin at a final concentration of 200 µg/ml was added to inhibit the activity of the E. coli RNA polymerase and, hence, prevent transcription of all genes except for the plasmid-encoded bglF alleles that are transcribed by the T7 RNA polymerase. After pulse-labeling for 5 min, unlabeled methionine was added to a final concentration of 0.5 mg/ml for another 5 min. Eight aliquots of 0.5-ml cultures were pelleted by centrifugation, washed with phosphate-buffered saline (20 mM sodium phosphate buffer and 150 mM NaCl, pH 7), pelleted again, and resuspended in 50 µl of phosphate-buffered saline. Salicin was added to four aliquots to a final concentration of 0.2%, and cells were incubated for 5 min at room temperature. o-PDM, p-PDM, or BMH was added to a final concentration of 1 mM (from stock solutions of 50 mM of the cross-linkers dissolved in N,N-dimethyl-formamide) to three salicin-containing aliquots and three aliquots that did not contain salicin, and cells were further incubated for 45 min at 30°. The reactions were quenched by the addition of 10 mM NEM and 10 mM DTT, followed by incubation for 5 min at room temperature. Electrophoresis sample buffer containing 10 units of endonuclease benzonase was added, and the samples were incubated for 10 min at 30° and for 5 min at 80°. Electrophoresis of proteins was carried out on 7.5% sodium dodecyl sulfate (SDS)-polyacrylamide gels. After electrophoresis, gels were dried and exposed to Kodak BioMax MR film.
Sense analyses. BlastP (http://www.ncbi.nlm.nih.gov/) was used to retrieve sequences of BglF homologues. Alignments of amino acid sequences were performed using the European Bioinformatics Institute ClustalW server (http://www.ebi.ac.uk/clustalw) (65) with adjustments by hand. Consensus sequences for these alignments were derived by the Consensus program (http://coot.embl.de/Alignment/consensus.html). To search for conserved motifs, 29 sequences of BglF homologues, which correspond to residues 100 to 138 of BglF, were submitted to the Block Maker server (http://bioinformatics.weizmann.ac.il/blocks/blockmkr/www/make_blocks.html) (31). The LAMA program (http://bioinformatics.weizmann.ac.il/blocks-bin/LAMA_search_new.sh) (53) was used to search the Blocks Database (30) for protein motifs that show similarity to the blocks identified in BglF homologues by the Block Maker.
Molecular modeling. A model for the three-dimensional (3D) structure of IIBbgl was built by homology modeling, using the structure of IIBglc complexed with IIAglc from E. coli (PDB code 1O2F) (7) as a template. This structure was chosen as the best template for modeling IIBbgl by the threading server 3D Position Specific Scoring Matix (3D-PSSM) (39). The two sequences share 28.6% identity and 41.8% similarity.
A model of BglG open dimer was built by homology modeling using the structure of LicT from Bacillus subtilis (PDB code 1TLV) (26) as a template. This structure was chosen as the best template for modeling BglG open dimer by the threading server 3D-PSSM (39). The two sequences share 39% identity and 69% similarity.
Both models were built by the Modeler program (Accelrys Software Inc., San Diego, CA) using the alignments given by the threading server 3D-PSSM. Figures were depicted using the program DS Modeling, version 1.1 (Accelrys Software Inc., San Diego, CA).
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The amino acid substitutions that impaired the ability of BglF to activate BglG in the presence of ß-MG are listed in Table 1. They map to the active-site-containing domain (segment B) and to two regions in the membrane domain (segments C1 and C4). Nine mutants contained single amino acid substitutions, five mutants contained two substitutions, and the rest contained three amino acid replacements or more. For the two segment B mutants with two and three amino acid substitutions, we identified the replacement that conferred the phenotype and constructed the respective mutants with single amino acid substitutions, i.e., M22K and G64S. These mutants were subsequently analyzed instead of the original isolated mutants. The segment C1 mutants contained single amino acid substitutions except for one double-substitution mutant, which contained a mutation (G119E) that was independently isolated as a single replacement. Of the latter two mutants, only G119E was used for further analyses. Segment C4 mutants contained single (two mutants), double (three mutants), and triple (one mutant) amino acid replacements. The rest contained more substitutions and were not further analyzed. Notably, three positions in segment C4 were replaced by different residues in different mutants [F407(S/I), T418(A/P), and S431(C/N)], suggesting that these positions are important for the stimulated functions of the BglF sensor.
To obtain a quantitative estimate for the ability of the BglF mutants to regulate BglG activity by (de)phosphorylation, the expression of the bgl-lacZ fusion in MA200-1 cells expressing the mutant proteins was measured by performing ß-Gal assays. The results, which are summarized in Table 1, validated the phenotype observed on the indicator plates. In the absence of ß-MG, cells expressing all BglF mutants gave low ß-Gal units (3 to 11 units), which were comparable to the ß-Gal levels in cells expressing wild-type BglF (4 units). Because MA200-1 cells that do not express BglF (Table 1, vector only) produce a high level of ß-Gal (148 units), these results indicate that the isolated BglF mutants negatively regulate the activity of BglG as a transcriptional antiterminator in vivo, implying that they retained the ability to phosphorylate BglG. However, as opposed to the high ß-Gal level (90 units) produced in cells expressing wild-type BglF grown in the presence of ß-MG, low ß-Gal levels (1 to 12 units) were produced in cells expressing all BglF mutants grown under the same conditions. Hence, unlike the stimulatory effect of ß-MG on wild-type BglF, its presence does not stimulate the BglF mutants to relieve the negative inhibition inflicted upon BglG activity. Accordingly, the BglF mutants that we isolated are impaired in just the sugar-stimulated activation of BglG but not in BglG inactivation in the absence of the stimulating sugar, i.e., in their ability to dephosphorylate P-BglG but not in their ability to phosphorylate the regulator. Taken together, the mutations that were isolated in our screen shift the balance toward BglG phosphorylation.
Substitution of single residues in BglF impaired BglG activation but not phosphotransfer of the activating sugar. Dephosphorylation of P-BglG by BglF is stimulated by ß-glucosides; i.e., recognition of the sugar by BglF is a prerequisite for this activity. Therefore, we asked whether failure of the mutants to activate BglG in the presence of ß-glucosides is due to a weaker recognition or lack of sugar recognition by these BglF variants. Since ß-MG was used for the isolation of the BglF mutants, we examined the ability of the mutants to regulate BglG activity in the presence of another ß-glucoside, salicin. To this end, we examined the phenotype of MA200-1 cells producing the BglF mutants on MacConkey-lactose plates containing salicin and also measured the ß-Gal levels produced by these cells in the presence of salicin. The results are given in Table 1. Most BglF mutants failed to activate BglG in the presence of salicin as they did with ß-MG. However, four segment C4 mutants exhibited a different behavior with the two ß-glucoside sugars, as evidenced by the relief of BglG inhibition in the presence of salicin (red colonies on indicator plates and 137 to 183 units of ß-Gal activity) but not in the presence of ß-MG. Similar results were obtained when the ß-glucoside arbutin was included in the growth medium (data not shown). Hence, these four segment C4 mutants are defective in their ability to recognize ß-MG rather than in their BglG phosphatase activity. This suggested to us that at least some of the isolated mutants are not defective exclusively in the recognition of P-BglG but, rather, in the recognition of all or certain ß-glucosides.
To study the ability of the isolated BglF mutants to recognize and phosphotransfer ß-glucosides more directly, we expressed the mutants in cells defective in their chromosomal bglF gene and evaluated the uptake of three ß-glucosides, p-NPG, salicin, and arbutin. This approach relies on direct correlation between the ability of cells to phosphotransfer ß-glucosides and to utilize them, as sugar phosphorylation is a prerequisite for their efficient transfer and hydrolysis. Utilization of salicin and arbutin was indicated by the color of colonies grown on indicator plates. Hydrolysis of p-NPG and of salicin was quantified by assays that monitored their degradation products, as described in Materials and Methods. Utilization of salicin and arbutin was also indicated by growth on minimal medium plates containing each of these sugars as a sole carbon source, and the results were in full agreement with the color phenotype on the indicator plates (data not shown).
The results obtained by the different assays are given in Table 2. The BglF mutants presented several phenotypes of ß-glucoside uptake, which suggested that they differ in their ability to recognize and phosphotransfer ß-glucoside sugars. Two mutants with substitutions in segment B (M22K and Q60L) and one with a substitution in segment C1 (G119E) did not catalyze uptake of any tested ß-glucosides. Two segment B mutants (G56D and G64S) and two segment C1 mutants (G123D and G160R) showed intermediate ability to phosphotransfer p-NPG and salicin and an inability to catalyze arbutin uptake. Hence, all of the above mutants are completely or partially defective in the recognition, phosphorylation, or transport of ß-glucosides. It is reasonable to speculate that the failure of these mutants to shift the balance toward BglG activation in the presence of ß-glucosides is a consequence of their inability or reduced ability to recognize ß-glucosides. A similar explanation holds for the four segment C4 mutants that dephosphorylated BglG in the presence of salicin but not ß-MG (see Table 1); i.e., they are specifically defective in ßMG recognition. The remaining mutants, two in segment C1 (G106D and S146R) and two in segment C4 (G408D K466R and Q417H), presented the most interesting behavior. These four mutants catalyzed uptake of all ß-glucosides tested and yet did not catalyze BglG activation in their presence. Hence, although these mutants recognize, transport, and phosphorylate ß-glucosides, they cannot be stimulated by the ß-glucoside sugars to dephosphorylate BglG, most likely because they are impaired in P-BglG recognition. In view of the intriguing phenotype of this group of mutants, we identified G408D as the replacement that confers the phenotype in the G408D K466R double mutant and constructed the respective mutant with a single amino acid substitution to be used for further analyses.
|
Mutations that specifically impair BglF ability to activate BglG in vivo primarily reduce its capability to dephosphorylate BglG in vitro. To directly examine the (de)phosphorylation capabilities of the BglF mutant proteins, we tested their ability to deliver the phosphate to the ß-glucoside salicin and to reversibly phosphorylate BglG in vitro using inverted cytoplasmic membrane vesicles enriched for the different BglF variants, as previously described (2). When this experimental system is used, it is important to bear in mind that PTS sugar permeases, such as BglF, have periplasmic and cytoplasmic substrate-binding sites with different affinities for the sugar. The in vitro assay, performed with inside-out membranes, examines nonvectorial phosphorylation (phosphorylation of the substrate without transport) catalyzed by the cytoplasmic sugar-binding site, which is usually more permissive than the periplasmic site that catalyzes vectorial phosphorylation (phosphorylation coupled to transport) in intact cells (23, 22). Indeed, all BglF mutant proteins, labeled in the presence of [32P]PEP, EI, and HPr (see Materials and Methods), were dephosphorylated in vitro upon addition of salicin. However, the majority of the mutants that were impaired in salicin uptake in intact cells delivered the phosphate to this sugar in the in vitro system at a much slower rate than wild-type BglF, i.e., after 2 to 5 min, compared to the wild type, which was completely dephosphorylated in less than 30 s (data not shown). Hence, BglF variants impaired in vectorial sugar phosphorylation could still catalyze nonvectorial phosphorylation through their inward-facing sugar-binding site, albeit at a low efficiency. An exception was the M22K mutant, which was impaired in salicin uptake in vivo but underwent dephosphorylation in vitro upon salicin addition at a rate comparable to wild-type BglF, highlighting the different requirements of the vectorial and nonvectorial phosphorylation reactions.
All BglF mutants phosphorylated BglG in vitro at kinetics comparable to wild-type BglF (data not shown). This was expected, mainly because negative regulation of BglG was a requirement in our mutant screen and also because BglG phosphorylation requires only the active-site-containing domain of BglF (8), and none of the mutations affected the phosphorylating residue C24.
To assay the ability of the BglF proteins to dephosphorylate BglG in vitro, we incubated inverted membrane vesicles enriched for wild-type or mutant BglF with 32P-labeled BglG or BglG fused to MBP (MBP-BglG) in the presence of salicin, as previously described (2, 10). As shown in Fig. 1, dephosphorylation by wild-type BglF was rapid. Only 24% of the 32P-MBP-BglG remained after 5 min (Fig. 1) and no 32P-MBP-BglG was detected after 25 min (not shown). In contrast, the capacity of the four BglF mutants that could not activate BglG in vivo (G106D, S146R, G408D, and Q417H) to dephosphorylate 32P-MBP-BglG in vitro was fairly impaired. Between 56 and 96% of the 32P-labeled MBP-BglG was detected after 5 min (Fig. 1), and even after 25 min of incubation 34 to 52% of phosphorylated MBP-BglG could still be detected (data not shown). Very similar results were obtained with 32P-BglG (data not shown). Hence, these residues are important for dephosphorylating BglG not only in vivo but also under conditions that allow easy access to the (de)phosphorylating cysteine from the cytoplasm. As for the other BglF mutants that we isolated, some mutants that were defective in ß-glucoside uptake did not function as BglG phosphatases in vitro (e.g., Q60L) (Fig. 1), whereas others dephosphorylated BglG in vitro, albeit quite inefficiently (data not shown). Evidently, these latter mutants can shift the balance toward P-BglG dephosphorylation, at least to a certain extent, in the in vitro test with inverted membranes. In this test the permease is more accessible to the stimulating sugar, transport of the sugar is not a requirement, and P-BglG is presented at a high concentration compared to tests that involve intact cells.
|
|
The faster-migrating band cannot be explained by cleavage near the cysteine planted at position 106, induced by modification of this residue by the cross-linkers, because the difference in migration between the cross-linked and non-cross-linked products is in the range of a couple of kilodaltons, whereas a truncated protein that results from proteolysis near position 106 is expected to be shorter by 11 to 12 kDa. Also, the additional band was not observed after incubation of the G106C single-Cys variant with the same cross-linkers. Using a cross-linker that can be cleaved by reducing agents and whose range of cross-linking partially overlaps with the other reagents used here, we showed that a pattern of migration similar to that observed with other di-Cys BglF variants containing a cysteine spatially adjacent to C24 can be reverted to the original electrophoretic migration of the non-cross-linked BglF protein upon addition of DTT (S. Yagur-Kroll and O. Amster-Choder, unpublished data). It is noteworthy that the results with the cross-linkers were highly reproducible.
Isolation of BglG mutants that suppress the defects in BglF mutants impaired in P-BglG dephosphorylation. To substantiate our results, we attempted to isolate BglG mutants that compensate for the defect exhibited by these mutants. To this end, we made use of E. coli strain MA152, which has a deletion of the bgl operon and carries a bgl'-lacZ transcriptional fusion (49). Expression of lacZ in MA152 depends on the introduction of a plasmid encoding the BglG antiterminator. Introduction of a second plasmid, which codes for BglF, renders lacZ expression dependent on ß-glucosides. And so transformation of MA152 with the wild-type bglG and bglF alleles results in growth of white colonies on MacConkey-lactose plates in the absence of ß-glucosides and in growth of red colonies in their presence. Expression of each of the BglF variants that is defective merely in the P-BglG dephosphorylation (G106D, S146R, G408D, and Q417H) together with wild-type BglG in MA152 led to growth of white colonies on the indicator plates independent of ß-glucoside addition, in accordance with the phenotype of these mutants in MA200-1. To isolate BglG mutants that suppress the defect in the BglF mutants, we randomly mutagenized the bglG gene, cloned downstream of the tet promoter. We then transformed MA152 cells containing a plasmid that carries one of the above four bglF alleles with the collection of mutated bglG alleles and induced the expression of the BglG mutants by the addition of anhydrotetracycline hydrochloride. Colonies that restored or partially restored the phenotype conferred by coexpression of wild-type bglF and bglG, that is, white on MacConkey-lactose plates without ß-glucosides and red or pink on such plates containing salicin, were isolated.
Between 900 and 2,200 colonies were screened after each of the four BglF mutants was expressed together with the BglG mutant library in strain MA152. Five colonies demonstrated the desired phenotype, which was confirmed to be plasmid conferred. Replacement of H160 of BglG by a leucine suppressed the G106D mutation in BglF. Three BglG mutants—S93G, R22Q Y50F P118L, and S93N P118A—were isolated as suppressors of the G408D BglF mutation. Finally, the H158V BglG mutant suppressed the defect of the Q417H BglF variant. In the case of the R22Q Y50F P118L BglG mutant, we identified P118L as the replacement that suppressed the G408D mutation in BglF and constructed the respective mutant with a single amino acid substitution. Notably, two positions in BglG were replaced by different residues in different mutants [S93(G/N) and P2118(L/A)].
To characterize the nature of the suppressor mutations, each BglG mutant was expressed in MA152 cells together with either one of the four BglF mutants primarily defective in BglG dephosphorylation or with wild-type BglF. Expression of the bgl'-lacZ fusion in these cells was estimated qualitatively by colony color on MacConkey-lactose plates and quantitatively by measuring ß-Gal activity, in both the absence and presence of salicin. There was a good agreement between the color of the colonies and the ß-Gal levels. The results of the ß-Gal assays are documented in Table 3. All the BglG mutants antiterminated transcription efficiently in the absence of BglF, as indicated by the high level of ß-Gal expression (Table 3, no BglF) and by the growth of red colonies on MacConkey-lactose plates. All the BglG mutants were phosphorylated and dephosphorylated by wild-type BglF in the absence and presence of salicin, respectively, as indicated by the low ß-Gal levels in the absence of the inducing sugar and the high levels in its presence (Table 3, wild-type BglF) and by the growth of white and red colonies on the indicator plates lacking and containing salicin, respectively. The BglG mutants suppressed their cognate BglF mutants at different levels; P118L was the best suppressor and A158V was the weakest. Significantly, some of the BglG mutants also suppressed, in addition to the BglF mutant for which they were isolated, other BglF mutants defective in BglG dephosphorylation. The most general suppressing phenotype was demonstrated by the P118L BglG mutant, which suppressed all four BglF mutants. The H160L BglG variant also suppressed all four BglF mutants although suppression was at a lower level than with P118L. The A158V BglG mutant suppressed the S146R and the G408D BglF mutants better than the Q417H BglF mutant, for which it was isolated. Among the BglF mutants, the phenotype exerted by G408D was the most suppressible whereas G106D was the least suppressible.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The activity of several BglG-homologues from gram-positive organisms was shown to be controlled by two opposing phosphorylation reactions; i.e., in addition to phosphorylation by their cognate PTS sugar permease, they are positively regulated by HPr-catalyzed phosphorylation (e.g., see references 40 and 45). The latter reaction was proposed to be part of the carbon catabolite repression mechanism that operates in gram-positive bacteria (reviewed in references 67 and 66). The finding that BglG is phosphorylated in vivo in a PTS-dependent manner in the absence of BglF led to the suggestion that BglG is also phosphorylated by HPr (24, 25). Although direct phosphorylation of BglG by HPr was not detected in sensitive in vitro systems (e.g., see references 2, 10, and 11) and although HPr from E. coli failed to phosphorylate a BglG homologue from B. subtilis, which was efficiently phosphorylated by the B. subtilis HPr (40), it is possible that in vivo there are two ways of phosphorylating BglG. In any case, in the presence of BglF in the in vitro system, BglG is phosphorylated by BglF rather than by HPr, as BglG was not phosphorylated in the presence of EI, HPr, and BglF variants mutated in either one of the phosphorylation sites (10). Direct phosphorylation of a BglG homologue by a PTS permease has also been demonstrated in the case of GlcT, which is phosphorylated by the glucose permease EIIGlc (60). As for P-BglG dephosphorylation, BglF catalyzes this reaction in the absence of EI and HPr and only in the presence of ß-glucosides (2), whereas the HPr was proposed to phosphorylate BglG, rather than dephosphorylate it, in the presence of ß-glucosides. Thus, the phosphorylation and dephosphorylation of BglG that are documented in this study are undoubtedly catalyzed by BglF.
To investigate how the different and opposing functions of BglF are coordinated and to identify structural components involved in these activities, we screened for BglF mutants that inactivated BglG by phosphorylation but were defective in BglG activation by dephosphorylation. Most mutants were also defective in sugar recognition or uptake, but four mutants (G106D, S146R, G408D, and Q417H) recognized all ß-glucosides tested and yet could not activate BglG in their presence, indicating that they are defective exclusively in their interaction with P-BglG. The isolation of BglG mutants that specifically suppress the phenotype of these four BglF mutants corroborated that the different functions of BglF can be uncoupled. Significantly, the four mutations are in the BglF membrane domain, which is required for P-BglG dephosphorylation but not for BglG phosphorylation (8). An interaction between this domain and the active-site-containing domain was suggested to play a role in the ability of BglF to switch between its modes of action, based on the observed ß-glucoside-induced coupling between C24 and a native cysteine in the membrane domain that correlates with catalysis of the sugar-stimulated functions (12). Evidence for a sugar-induced conformational change which leads to spatial reorganization of BglF domains and rearrangement of its membrane domain is accumulating (70; also S. Yagur-Kroll and O. Amster-Choder, unpublished data). The demonstration that one of the residues that we identified as important for P-BglG dephosphorylation is spatially proximal to the active-site cysteine and that this proximity increases upon sugar binding (Fig. 2) supports the idea that a sugar-induced conformational change exposes or creates the site which interacts with P-BglG.
Location and nature of the sites in BglF that are implicated in P-BglG dephosphorylation. A model for BglF membrane topology, which was inferred from biochemical analyses (70), is schematically shown in Fig. S1 in the supplemental material. It proposes that the BglF membrane domain contains a large, alleged cytoplasmic loop that includes the conserved GIXE motif suggested to have a role in PTS carbohydrates phosphotransfer (43, 44). Together with the two TM helices encompassing it, this region forms a structure with an inherent dynamics predicted to play a role in BglF action. The membrane-sidedness of the dynamic segments (see dotted lines in Fig. S1 in the supplemental material), in particular of L409, was shown to be affected by ß-glucosides, suggesting that this region undergoes a conformational change in the presence of the sugar (70). Another region of putative significance is the reentrant loop connecting TM I and TM II (70), an element associated with activity in other membrane proteins (62, 69). The dynamic region and the reentrant loop were shown by in vivo chemical cross-linking to be spatially close to C24 and to each other (S. Yagur-Kroll and O. Amster-Choder, unpublished data). Notably, the four mutations that impaired the ability of BglF to dephosphorylate BglG were mapped to these two regions (see Fig. S1 in the supplemental material): G106 and S146 locate to the TMs that flank the reentrant loop, which may well locate to a water-filled crevice within the membrane (14), and G408 and Q417 map to the dynamic part of the protein (the neighboring positions A400, S406, L409, and S411 were shown to flip between the cytoplasm and the periplasm [70]). In accordance with the proposed spatial proximity of these regions to the active-site cysteine, G106 is shown here to be close to C24, and this proximity becomes more significant upon sugar stimulation. Taken together, positions important for BglG dephosphorylation map to two regions previously suggested to be implicated in BglF activity that are spatially close to each other and to the dephosphorylating cysteine and that are subjected to sugar-induced conformational changes.
The substitutions that impaired the ability of BglF to dephosphorylate BglG were not of a conservative nature (G to D, S to R, and Q to H). A more conservative substitution at these positions by a cysteine residue demolished this activity only in the case of S146, although replacement of a serine by a cysteine is considered fairly conservative. Hence G106, G408, and Q417 are important for dephosphorylation of BglG but can tolerate a moderate replacement, whereas a serine at position 146 is essential for the ability of BglF to dephosphorylate BglG.
The sequences around the four positions whose substitution impaired BglG dephosphorylation were compared to the respective regions in other ß-glucoside phosphotransferases by multiple sequence alignment. The comparison was to ß-glucoside permeases that are associated with either RNA-binding transcriptional antiterminators (BglF homologues) (1) or DNA-binding repressors (AscF homologues) (29). Overall, BglF shows a high degree of homology to proteins from both groups (50 to 99% similarity). There are only a few areas in which the two groups differ, and the region around G106 is an example of such an area. G106 is a conserved residue among all ß-glucoside permeases, but the sequence that surrounds it differs between the two groups (Fig. 3, compare A and B; residues conserved in BglF homologues are highlighted in gray in both groups). Comparison of this region with profiles of other protein families is described and discussed in the next section.
|
|
-helices, so that the positive end of the helix dipole points toward the ion and stabilizes it by electrostatic interactions (16, 17). LAMA also detected a significant similarity between the conserved motif around G106 and a region in type IIa Na+/Pi cotransporters (see Fig. S3 in the supplemental material). These proteins play a key role in the reabsorption of inorganic phosphate by transporting phosphate ions together with sodium ions (52). Structural data are not available yet for these proteins, but topological and structure-function studies implied that the region of the rat Na+/Pi cotransporter, which is similar to the new motif in BglF homologues, is involved in translocation (42, 57). The similarity between the conserved block in BglF homologues and the ion-binding site of ClC channels, on one hand, and the region implicated in phosphate ion translocation in Na+/Pi cotransporters, on the other hand (Fig. 3C), raises the possibility that this region in BglF might be implicated in binding the phosphoryl group. This idea is supported by the spatial proximity of G106 to C24 that accepts the phosphate from P-BglG (Fig. 2). Also, the replacement of G106 by the negatively charged aspartic acid residue demolished the ability of BglF to dephosphorylate BglG, whereas its replacement by a cysteine residue did not, as anticipated if this region is involved in binding or stabilizing the negatively charged phosphate group. The fact that this motif is not conserved in ß-glucoside permeases that are associated with repressors rather than antiterminators, where there is no evidence for regulation via reversible phosphorylation, favors of this idea. The finding that G106D was the least suppressible mutation substantiates the importance of this position in P-BglG dephosphorylation. Previous results (9) support the idea that C24 and a site in the membrane, which is affected by the sugar presence, need to be properly oriented to accept the phosphate from P-BglG. Notably, P-BglG is kept at the membrane and is released to the cytoplasm following dephosphorylation in the presence of the sugar (47).
Different positions in BglF are implicated in recognition of different ß-glucosides. Except for the four mutations discussed above, all the other mutations isolated in our screen abolished or reduced BglF's ability to recognize and/or transport ß-glucosides. Unexpectedly, a number of these BglF mutants were impaired in the recognition and uptake of some ß-glucosides but not others (Table 2). The mutations that impaired ß-glucoside recognition and uptake mapped to the active-site-containing B domain and the membrane C domain.
We modeled the B domain of BglF based on the known structure of its close homologue, the B domain of the glucose permease from E. coli (7) (see Fig. S5 in the supplemental material). The four mutations in this domain, two of them replacements of highly conserved residues (Q60L and G64S) and two of nonconserved amino acids (M22K and G56D), are predicted to be positioned at the base of a trigonal, pyramid-like structure formed by a four-stranded antiparallel ß-sheet that contains the active-site cysteine: M22 is in the same ß strand as C24, Q60 is on a parallel strand, and G56 and G64 are in ß-turns. Except for M22, the other three residues are predicted to be more than 6.5 Å apart from C24, the limiting distance for noncovalent interactions (51), ruling out the possibility of a direct interaction with the active site. Hence, we propose that these residues are involved in composing a recognition site on the face of the B domain which is presented to the incoming sugar.
The mutations in the C domain that affected sugar recognition and uptake mapped to two regions: (i) the reentrant loop and the TM flanking it at the beginning of the C domain or (ii) the dynamic region at the end of this domain. The first group of mutations, all conserved to a certain extent, showed intermediate phosphotransfer activity and/or different levels of activity with different ß-glucosides, whereas the latter were specifically defective in their ability to recognize ß-MG, a nonmetabolizable ß-glucoside that was used in our mutant screen. The different phenotypes presented by the mutants imply that recognition of the various ß-glucosides involves both common and specific elements. Supposedly, whereas some positions recognize the common pyranose ring of these sugars, others handle the aromatic rings, specific for the different compounds.
Lessons from the BglG suppressor mutations. The BglG suppressor mutants that we isolated are not defective proteins, as they all functioned as antiterminators and were properly phosphorylated and dephosphorylated by wild-type BglF (Table 3). Rather, they specifically corrected the ineffective interaction of the BglF mutants with P-BglG to different extents. By and large, they all suppressed the phenotype of more than one BglF mutant, suggesting that the replacements suppressed the BglF mutations by a common strategy, e.g., by making the phosphorylated residue in P-BglG more accessible or by facilitating a conformational shift that leads to BglG activation. This notion is supported by the finding that increasing the level of wild-type BglG expression had a suppressive effect on the phenotype of the BglF mutants, i.e., high-copy-number suppression (data not shown). Still, the isolation of two different replacements for two out of the four positions to which the suppressor mutations were mapped [S93(G/N) and P118(L/A)] emphasizes the importance of these positions for dephosphorylation of P-BglG. The finding that the phenotype exerted by G408D was the most suppressible is in accord with the observation that dephosphorylation of this mutant was the least defective relative to the other mutants (Fig. 1).
BglG represents a growing family of proteins that positively regulate expression of sugar utilization operons by binding to the nascent transcripts of these operons to prevent premature termination of transcription in the presence of the cognate PTS sugars (reviewed in references 1 and 67). They consist of an RNA-binding domain followed by two homologous domains, PRD1 and PRD2 (PTS-regulation domains), each containing t
