ABSTRACT
Rhamnose catabolism in Rhizobium leguminosarum was found to be necessary for the ability of the organism to compete for nodule occupancy. Characterization of the locus necessary for the catabolism of rhamnose showed that the transport of rhamnose was dependent upon a carbohydrate uptake transporter 2 (CUT2) ABC transporter encoded by rhaSTPQ and on the presence of RhaK, a protein known to have sugar kinase activity. A linker-scanning mutagenesis analysis of rhaK showed that the kinase and transport activities of RhaK could be separated genetically. More specifically, two pentapeptide insertions defined by the alleles rhaK72 and rhaK73 were able to uncouple the transport and kinase activities of RhaK, such that the kinase activity was retained, but cells carrying these alleles did not have measurable rhamnose transport rates. These linker-scanning alleles were localized to the C terminus and N terminus of RhaK, respectively. Taken together, the data led to the hypothesis that RhaK might interact either directly or indirectly with the ABC transporter defined by rhaSTPQ. In this work, we show that both N- and C-terminal fragments of RhaK are capable of interacting with the N-terminal fragment of the ABC protein RhaT using a 2-hybrid system. Moreover, if RhaK fragments carrying either the rhaK72 or rhaK73 allele were used, this interaction was abolished. Phylogenetic and bioinformatic analysis of the RhaK fragments suggested that a conserved region in the N terminus of RhaK may represent a putative binding domain. Alanine-scanning mutagenesis of this region followed by 2-hybrid analysis revealed that a substitution of any of the conserved residues greatly affected the interaction between RhaT and RhaK fragments, suggesting that the sugar kinase RhaK and the ABC protein RhaT interact directly.
IMPORTANCE ABC transporters involved in the transport of carbohydrates help define the overall physiological fitness of bacteria. The two largest groups of transporters are the carbohydrate uptake transporter classes 1 and 2 (CUT1 and CUT2, respectively). This work provides the first evidence that a kinase that is necessary for the catabolism of a sugar can directly interact with a domain from the ABC protein that is necessary for its transport.
INTRODUCTION
Cells are separated from the external environment by a selectively permeable membrane. A means to transport physiologically relevant substrates across these membranes is required (1–3). The largest and most widely distributed family of transporters is the ATP binding cassette (ABC) transporters (1). ABC transporters are responsible for the transport of a very diverse set of substrates and can be broken into two main classes, importers and exporters (1, 3–5). Much of our understanding of the structure, mechanism, and kinetics of these systems has been derived from the study of relatively few model transporters (6).
Gram-negative ABC importers typically consist of a permease component that is made up of two membrane-spanning proteins (encoded by either one or two genes), a periplasmic substrate binding protein, and two copies of the ABC-type ATPase protein (7–10). Carbohydrate uptake transporters (CUT), however, are divided into two main groups: CUT1 and CUT2. This division is based on the ATPase protein. The CUT2 ABC protein is typically about 200 amino acids larger than the CUT1 family counterpart and appears to have arisen from a fusion of two ABC domains, with the ABC protein containing two ATP binding sites (9). However, a functional substitution of a key lysine residue in the C-terminal Walker A motif likely leads to the loss of ATP hydrolysis function in this second domain (9). The CUT1 class maltose transporter from Escherichia coli has served as a model for Gram-negative ABC-type importers (8, 9, 11–14). Unlike with the genetic and biochemical evidence that exists for the CUT1 class of transporters (6, 10), relatively little evidence is available for the CUT2 class of ABC transporters.
The catabolism of carbon and the transport of substrates are important aspects that come into play during the interaction of bacteria with plants (15, 16). In Rhizobium leguminosarum, the catabolism of rhamnose was shown to be necessary for competition for nodule occupancy (17). The genes encoding the determinants necessary for the transport and catabolism of rhamnose are arranged in two divergently transcribed transcripts that are regulated by a negative regulator (RhaR) (18). One transcript contains the rhaD and rhaI genes, whereas the other contains the genes rhaRSTPQUK (18). The catabolism is carried out by a mutarotase (RhaU), an isomerase (RhaI), a kinase (RhaK), and a dehydrogenase/aldolase (RhaD) (19, 20). The transport of rhamnose is dependent on a CUT2-type ABC transporter system consisting of RhaSTPQ (18, 19, 21). Functional characterization of this locus revealed that the transport of rhamnose by the ABC transporter encoded by rhaSTPQ was affected by the loss of the gene encoding the sugar kinase (rhaK). The absence of rhaK did not affect either the transcription or the translation/membrane localization of the components of the ABC transporter (19). Using an in-frame insertional mutagenesis strategy, it was shown that the function of RhaK as a sugar kinase and its role in transport could be genetically separated (21).
Homology models of RhaK suggest that it is arranged in a typical sugar kinase arrangement containing N- and C-terminal lobes separated by an active-site cleft. The two insertional alleles capable of uncoupling transport function from kinase function mapped to either the N- or C-terminal lobe, and based on these models, the inserts appear to be localized to the same protein surface (21). Taken together, this suggests that RhaK is capable of interacting with another protein, and this interaction affects the ability of RhaSTPQ to transport rhamnose (21). In this work, we address the hypothesis that RhaK directly interacts with the ABC protein RhaT.
MATERIALS AND METHODS
Bacterial strains, plasmids, and media.The bacterial strains and plasmids used and generated in this work are listed in Table 1. Bacterial strains used were routinely grown at 30°C on complex (either LB or tryptone-yeast extract [TY]) or defined (Vincent minimal) medium (22–24). Vincent minimal medium contained (per liter) 1 g of K2HPO4, 1 g of K2HPO4, and 0.6 g of KNO3. Following autoclaving, MgSO4, CaCl2, and FeCl3 were added to final concentrations of 1 mM, 0.5 mM, and 50 μM, respectively. In addition, biotin, thiamine, and pantothenic acid were also added to a final concentration of 1 μg ml−1 each. The addition of carbon to the defined medium was carried out as previously described (17). E. coli strains were routinely grown at 30°C when in broth cultures and at 37°C on agar medium. When required, antibiotics were used at concentrations of 100 μg ml−1 ampicillin (Amp) and 50 μg ml−1 kanamycin (Kan). Growth was routinely monitored spectrophotometrically at 600 nm. X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) was used at a concentration of 20 μg ml−1. To induce expression from vectors, 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was used as necessary.
Strains, plasmids, and primers
Cloning and genetic manipulations.Standard techniques were used for the amplification of DNA by PCRs, DNA isolation, restriction enzyme digestion, ligations, transformations, and agarose gel electrophoresis (24). Nucleotide sequencing reactions were carried out by cycle sequencing using the BigDye version 3.1 kit and analyzed using an ABI 3130 sequencer, as previously described (21).
To construct a plasmid overexpressing Sinorhizobium meliloti RhaK, the broad-host-range Gateway-compatible destination vector pCO37 was used (25). Briefly, the S. meliloti ORFeome was utilized (26, 27), and the entry clone corresponding to the rhaK gene was recombined into pCO37, as previously described (28, 29).
Bacterial two-hybrid analyses.Bacterial two-hybrid analyses were carried out using an adenylated cyclase-based system, essentially as described previously (30). Fragments from rhaT and rhaK were amplified using the primers listed in Table 1 and cloned into the pUT18 and pTK25 vectors, respectively. The final constructs were sequenced to ensure that they were in frame with the T18 and T25 fragments of adenylate cyclase in pUT18 and pTK25, respectively.
Generation of motif variants.Variants of rhaK F1 were designed and purchased from GenScript, Inc., NJ. The variants were supplied in pUC57. The inserts were each subsequently recloned into pKT25 to carry out two-hybrid assays (Table 1).
Beta-galactosidase assays.Bacterial strains carrying plasmids to be assayed for beta-galactosidase activity were generally grown overnight in the appropriate medium and subcultured. The assays were carried out on log-phase cells, essentially as described previously (17).
Phylogenetic and bioinformatic analyses.Basic protein alignments of RhaK were carried out using both PRALINE and ClustalX (31, 32). RhaK homology models were constructed using the Phyre2 server (33), using gluconate kinase from Lactobacillus acidophilus (Protein Data Bank [PDB] file 3GBT), as previously described (21). Phylogenetic analysis was carried out using the Phylogeny.fr server with default settings (http://www.phylogeny.fr) (34). Briefly, protein sequences were aligned using MUSCLE version 3.7 (35) and subsequently curated using gBlocks to eliminate poorly aligned sequences (36). Phylogenies were constructed using PhyML version 3.0, and trees were rendered using TreeDyn (37). Consensus sequences of putative protein motifs were generated using WebLogo (38). Final figures were constructed using Adobe Illustrator, as previously described (39).
RESULTS
Two-hybrid analysis of RhaK and RhaT.In an effort to demonstrate that RhaK might interact with components of the rhamnose transporter, we carried out Western blotting of RhaK in a wild-type background and in a genetic background in which the components of the transporter were not expressed (21). The results showed that a small proportion (7%) of RhaK was able to interact with the cell membrane, that the protein could be removed using a 150 mM NaCl wash, and that this interaction was not dependent upon the presence of the transporter (21). The way these experiments were carried out included the overexpression of rhaK; therefore, it was not clear if the result was physiologically relevant or if it adequately addressed the hypothesis that RhaK interacts directly with components of the rhamnose transporter. Since the basic architecture of an ABC transporter is well characterized (1, 40–42), it is likely that RhaT is the only component physically available to interact directly with RhaK. To investigate this further, we wished to address the possibility of direct in vivo interaction by using a bacterial two-hybrid system.
Briefly, the bacterial two-hybrid system is based on interaction-mediated reconstitution of adenylate cyclase activity in E. coli using the Bordetella pertussis adenylate cyclase (30, 43). Adenylate cyclase consists of two domains (T18 and T25) that do not generate enzymatic activity when physically separated on two plasmids. However, if the two domains are brought into close proximity because of protein-protein interaction due to in-frame translational fusions that have been added to these fragments, it results in functional complementation, resulting in cyclic AMP (cAMP) synthesis. The cAMP produced by the reconstituted chimeric enzyme regulates expression of the lac operon in a nonreverting adenylate cyclase-deficient (cya mutant) E. coli strain, thus allowing for direct screening of interactions using X-Gal and quantification of the relative strength of the interaction using beta-galactosidase assays (43).
Initially, the full lengths of RhaK and RhaT were translationally fused to the T18 and T25 fragments in pUT18 and pKT25, respectively. No evidence of interaction was detected. Since the fusion of proteins to these domains can cause folding, stability, or physical inability of interacting proteins to interact (44), a strategy to subdivide both RhaK and RhaT and to subclone the portions of these genes into the pUT18 and pKT25 vectors to assay for interactions was devised.
RhaK was divided into 4 segments using the secondary structure, as predicted by the Phyre2 server and homology models that we had previously generated, such that each predicted alpha helix and beta sheet were contained fully intact in at least one fragment (21). RhaK was divided into 4 fragments, with the largest being 120 amino acids. Primers were generated to amplify the DNA that encoded each of these segments. Each was subsequently cloned into pKT25, yielding plasmids pTM10, pTM13, pTM15, and pTM17 (Table 1 and Fig. 1). Similarly, using a predicted secondary structure and what is known about the functional domains associated with ABC proteins, five fragments representing the entire length of RhaT were generated and cloned into pUT18, yielding plasmids pDR201, pDR202, pDR204, pDR206, and pDR211 (Table 1 and Fig. 1).
Diagram indicating which residues make up each RhaK and RhaT fragment (F1 to F5) translationally fused to Cya fragments T25 and T18. The stretch of amino acid residues included in each fragment is indicated at the bottom, with the N termini and C termini of the wild-type proteins labeled. The sites of the insertions that define the rhaK72 and rhaK73 alleles are indicated. Note that RhaK is presented C to N terminus, whereas RhaT is presented N to C terminus.
To test for an interaction between fragments of RhaK and RhaT, all possible combinations of the RhaK fragments in pUT18 and RhaT fragments in pKT25 were cotransformed into BTH101 and incubated at both 37°C and 30°C, as suggested (30). The results showed that blue colonies were present only when the plasmids containing genes encoding either the N-terminal or the C-terminal fragment of RhaK fused to the T25 fragment of adenylate cyclase (pTM10 and pTM17, respectively) were cotransformed with pDR201, which contains genes encoding the N-terminal portion of RhaT translationally fused to the T18 fragment of adenylate cyclase. To ensure that these were bona fide positive interactions, the plasmids were subsequently isolated, retransformed, and resequenced. To quantitate this interaction, three representative colonies from each cotransformation were assayed (Fig. 2). The results show that each putative positive interaction yielded values of approximately 600 Miller units, whereas the typical background value for fragments that did not interact was approximately 50 Miller units (Fig. 2).
Beta-galactosidase activity of BTH101 cells cotransformed with the RhaK and RhaT fragments, as indicated. RhaKF1-F4 and RhaTF1-F5 are as defined in the legend of Fig. 1. The growth conditions and assays were carried out as indicated in Materials and Methods. The data represent the means of the results from three biological replicates, with activity reported as Miller units. The error bars represent the standard deviations.
rhaK72 and rhaK73 alleles disrupt two-hybrid interactions.Two 5-amino-acid in-frame insertional rhaK alleles, rhaK72 and rhaK73, which affect rhamnose transport but not kinase activity, were previously isolated (21). Interestingly, the mutations that define rhaK73 and rhaK72 map within the segments, which correspond to the wild-type N- and C-terminal fragments found in pTM10 and pTM17, respectively. Since we had previously hypothesized that it was possible that these alleles defined either a direct or an indirect point of interaction with RhaT, it was decided to directly test these alleles in the two-hybrid system. Since the primers that were used to generate the N- and C-terminal fragments of RhaK flanked the rhaK73 and rhaK72 alleles, these alleles were cloned into pKT25 by using pDR74 and pDR73, respectively, as the templates. The resultant constructs, pDR209 (RhaK fragment 4 carrying the insert that defines rhaK72) and pDR210 (RhaK fragment 1 carrying the insert that defines rhaK73), were transformed into and assayed in E. coli BTH101 strains carrying each of the rhaT fragments in pUT18 (pDR201 to pDR206).
The results show that when the plasmid carrying either the N-terminal or the C-terminal fragment of RhaK (pTM10 or pTM17, respectively) was transformed into BTH101 carrying the N-terminal region of RhaT (pDR201), these colonies were phenotypically blue on medium that contained X-Gal. However, when plasmids expressing fragments of RhaK that contain either the insertion defined by rhaK72 or rhaK73 (pDR209 or pDR210, respectively) were transformed into BTH101 carrying the N-terminal region of RhaT, the colonies were phenotypically white on medium containing X-Gal. Quantification of these interactions using beta-galactosidase assays showed that, similarly to what was seen previously, the N-terminal and C-terminal RhaK constructs in pKT25 yielded between 400 and 600 Miller units of activity (Fig. 3). When either pDR209 or pDR210 containing the rhaK72 or rhaK73 allele was introduced into BTH101 containing the N-terminal RhaT region, these yielded 132 ± 23 and 63 ± 7 Miller units, respectively (Fig. 3). Whereas the rhaK73 value was not different from the background values that were attained with other negative interactions (between 45 and 78 units), the rhaK72 value was significantly greater (Fig. 3). It is noteworthy that the rhaK72 allele was also capable of conferring a very slow growth phenotype on defined medium, with rhamnose as a sole carbon source, when it was introduced into a strain carrying a mutation in rhaK. This was correlated with marginal rhamnose transport, whereas the rhaK73 allele did not have this phenotype (21).
Beta-galactosidase activity of BTH101 cells cotransformed with the RhaK and RhaT fragments, as indicated. The RhaKF1, RhaKF4, and RhaTF1-F5 fragments are as defined in the legend of Fig. 1. RhaK72 or RhaK73 listed below the fragment number indicates that the insertion that defines RhaK72 or RhaK73 is present in the corresponding fragment. The growth conditions and assays were carried out as indicated in Materials and Methods, with activity reported in Miller units. The data represent the means of the results from three biological replicates, each assayed in triplicate, with the error bars representing the standard deviations of results from all replicates. The asterisk indicates that this value was found to be statistically different from all other negative values by Student's t test (P = 0.05).
Identification of putative motifs in the C- and N-terminal fragments of RhaK.The initial characterization of RhaK demonstrated that it had rhamnose kinase activity and that a functional protein was necessary for the transport of rhamnose (19). The initial bioinformatics analysis that was carried out did not identify any motifs that may play a role in interactions with the rhamnose ABC transporter (19). Phylogenetic analysis did, however, identify other putative orthologs that suggested that this was not a unique protein (19). Because the number of sequenced bacteria has increased since the initial characterization, it was reasoned that a reanalysis coupled with the functional data that we had generated might be useful in identifying regions that potentially participate in an interaction with RhaT.
To define an RhaK data set, the amino acid sequence of RhaK was used as a BLAST query sequence against sequences in the Integrated Microbial Genomes (IMG) database of complete, draft, and permanent draft genome sequences. To refine the data set, it was assumed that proteins with similar biochemical properties would be found to have similar operon structures. Carbohydrate kinases that were found in operons that were not syntenic or for which the sequence coverage did not cover the entire operon were omitted. One hundred twenty-nine BLAST hits with an E value of <le−150 were retained. Additionally, hits that came from environmental samples that were not identified to the species level were also discarded. Finally, each hit was manually curated to eliminate multiple identical representatives from the same species. If the RhaK orthologs were identical at the amino acid level within a species, only one representative was included in the analysis. The final data set consisted of 52 RhaK-like sequences.
Phylogenetic analysis of the sequences was carried out using the E. coli rhamnulose kinase (RhuK, encoded by rhaB) as an outgroup. The data support the hypothesis that the RhaK-like sequences separated into 3 clades (Fig. 4). Whereas the clade that contained the Mesorhizobium species is less well supported, a clade that contained the Sinorhizobium (Ensifer) species and another that predominantly contained the Rhizobium, Agrobacterium, Ochrobactrum, and Brucella genera and a few other organisms was strongly supported (Fig. 4).
Phylogenetic tree of RhaK-like proteins from alphaproteobacteria. Phylogenetic analysis was carried out using the Phylogeny.fr server. Support for the major nodes is indicated, whereas nodes with <50% support have been collapsed. Rlt100 and S. meliloti Rm1021 are indicated with an arrow.
RhaK from R. leguminosarum strain Rlt100 clustered with other R. leguminosarum strains, Rhizobium etli, Rhizobium tropici, Agrobacterium rhizogenes, and Agrobacterium radiobacter. Agrobacterium tumefaciens and Rhizobium lupini were separated from R. leguminosarum; however, all of these were within a single clade (Fig. 4). It is of note that, whereas the original published analysis had grouped the S. meliloti and A. tumefaciens strains together (19), this analysis clearly shows that there are differences among all of these strains.
Since there was divergence between the S. meliloti and R. leguminosarum RhaK proteins, we wished to determine if the S. meliloti RhaK could functionally replace the R. leguminosarum RhaK, reasoning that if it could, an alignment of the C-terminal and N-terminal RhaK fragments, coupled with previous functional genetic characterization, might give some insight into regions that interact with RhaT. To carry this out, we utilized the S. meliloti ORFeome platform that contains each of the annotated open reading frames from S. meliloti Rm1021 in a Gateway-compatible vector (26, 45). The rhaK open reading frame was recombined into the destination vector pCO37 (25), yielding pDR35. The introduction of pDR35 into R. leguminosarum strain Rlt144 (rhaK50::Tn5) resulted in functional complementation, suggesting that S. meliloti RhaK can carry out the same sugar kinase function and role in rhamnose transport as the RhaK from Rlt100.
Identification of a putative interaction domain.Since the S. meliloti strain Rm1021 RhaK homolog was able to complement Rlt144 (Rlt100 carrying an rhaK mutation), it was assumed that both of these proteins must be able to interact with the R. leguminosarum rhamnose transporter component RhaT. In an effort to identify putative interacting motifs, an analysis that included only the 22 RhaK protein sequences from groupings that contained S. meliloti and R. leguminosarum was carried out. The analysis consisted of first aligning the proteins on the basis of local regions of identity using PRALINE (32). This was then supplemented with a secondary-structure analysis using the Phyre2 server, biological information from previously characterized rhaK variants (21), and functional residue predictions based on glycerol and xylulose kinase (46, 47). An overview of the C- and N-terminal fragments that were positive in the two-hybrid screens was generated (Fig. 5).
Amino acid conservation in the clades containing Rlt100 and Rm1021 RhaK proteins. The regions corresponding to RhaKF1 (A) and RhaKF4 (B) are displayed, with only the amino acid sequence from R. leguminosarum shown. A conservation score (listed below each amino acid) was calculated using PRALINE, with the scoring ranging from 0 to 10, with invariant residues marked by an asterisk. The amino acid residues are correspondingly colored from red (invariant) to blue (least conserved). The secondary structure of these regions, as predicted by the Phyre2 server, is indicated above the amino acid sequence. Regions predicted to be β sheets are indicated by blue arrows, while regions predicted to form a helix are indicated by red cartoon helices. The sites of the previously generated insertional mutations are indicated by triangles, with their corresponding allele numbers listed above (21). rhaK72 and rhaK73 are highlighted. The colored bar below the conservation number depicts the predicted function of a region. This was determined by comparison to the homologous regions of the E. coli glycerol and xylulose kinases. Regions of this bar that are gray indicate no predicted function. The predicted functions of the colored regions are as indicated in the legend. One region that is highly conserved with no known function is highlighted in purple.
The results show that many regions of high conservation were identified (Fig. 5). These include regions identified as being involved in kinase dimerization, ATP binding, substrate binding, and magnesium binding. Two areas of conservation were shown to be present in the regions surrounding the rhaK72 and rhaK73 insertion sites. Although the region surrounding the rhaK72 insertion site within the C-terminal fragment showed conservation, this region may also be associated with ATP binding (Fig. 5B). However, in the N-terminal fragment, a conserved amino acid region with no predicted function, proximal to the insertion site of the RhaK73 variant, was identified (Fig. 5A). Based on previous modeling, this region was predicted to be a surface-exposed alpha helix that was disrupted in the RhaK73 variant (21).
To visually represent the region that makes up the putative motif, a sequence logo of the aligned regions was generated (Fig. 6A). The motif consists of 12 amino acids, 7 of which are invariant. If this sequence is used as a query for a BLASTP search of the NCBI database, the top 100 hits that are returned are sequences of carbohydrate kinases from the alphaproteobacteria. The highest-scoring hits either are predicted to be involved in rhamnose catabolism or are found in operons that show synteny to the R. leguminosarum rhamnose catabolic operon. Within the rhizobia, this region does not appear to be conserved in other kinases outside those predicted to be involved in rhamnose catabolism. However, this region does not seem to be conserved in all kinases involved in rhamnose catabolism. For example, the E. coli RhuK protein does not share conservation in this region (data not shown). Taken together, the data support the hypothesis that this region may constitute a motif involved in a physical interaction with RhaT.
(A) Consensus sequences of the putative binding motif as generated using WebLogo (38). (B) Beta-galactosidase activity of BTH101 cells cotransformed with each RhaKF1 variant fragment and RhaTF1. Variants are identified by the amino acid substitution that they contain. RhaKF1* contains a substitution of 6 of the 7 identified invariant amino acids from panel A (Y, P, H, D, L, W, and F). Activity is presented in Miller units. The data represent the averages from three biological replicates, with the error bars representing the standard deviations.
Characterization of the N-terminal motif.To systematically determine the relative importance of each residue in the proposed binding motif, alanine-scanning mutagenesis of the region was carried out. Each of the amino acids was converted to an alanine, except for amino acid 74, which was already an alanine. To ensure that the interaction was disrupted, a variant in which each of the conserved residues was converted to an alanine was also constructed. These variants were cloned into the appropriate vector and assayed for interaction with RhaTF1.
The results show that all of the substitutions made to the predicted binding domain appeared to significantly affect the ability of the two fragments to interact (Fig. 6A). The RhaKF1 variant containing multiple substitutions, RhaKF1* (containing a substitution of 5 of the conserved residues) resulted in beta-galactosidase activity reduced to background levels. This variant, however, contained multiple substitutions that by themselves nearly abolished beta-galactosidase activity. The substitution that affected this interaction the least was E73A. This substitution, however, still resulted in an approximate 50% reduction in beta-galactosidase activity (Fig. 6). The region in the middle of this sequence appears to be more tolerant to an alanine substitution, whereas the four N-terminal-most substitutions (Y67A, P68A, H69A, and Y70A) and the 2 C-terminal-most substitutions (S77A and F78A) resulted in the most severe reduction in beta-galactosidase activity. The four N-terminal-most substitutions are predicted to make up a beta sheet. This beta sheet leads to a predicted surface-exposed helix. Part of this beta sheet itself also appears to be exposed at the surface (Fig. 6) (21). The five least severe substitutions appear to make up the initial residues of the surface-exposed alpha helix. The two C-terminal-most substitutions (S77A and F78A) localize to a predicted surface-exposed alpha helix. Changing either of these two residues (S77 and F78) abolishes beta-galactosidase activity. Of note, the site of the linker-scanning insertion sequence in the rhaK73 allele is localized between these two amino acids (S77 and F78) (Fig. 6).
DISCUSSION
The data show that both the N-terminal and C-terminal fragments of RhaK, RhaKF1, and RhaKF4 can interact with an N-terminal fragment of the ABC protein RhaT (Fig. 2). Linker-scanning mutagenesis of RhaK was previously utilized to uncouple the in vivo ability of RhaK to affect whole-cell rhamnose transport from its ability to function as a kinase (21). However, no evidence was found that RhaK that lacked kinase activity could still transport rhamnose (19, 21). Since the two alleles (rhaK72 and rhaK73) that were previously isolated as being able to uncouple transport and kinase were localized within the N- and C-terminal regions defined by RhaKF1 and RhaKF4, these alleles were directly tested and found to disrupt the ability of these fragments to interact with the RhaT N-terminal fragment (Fig. 3). Taken together, these data strongly suggest that the interactions found in the two-hybrid assays are correlated with physiologically relevant phenotypes and are unlikely to be due to spurious interactions due to the nature of the assay(53).
It is also worth noting that the rhaK72 allele when in the F4 fragment has a statistically higher beta-galactosidase activity than that of the rhaK73 allele when it is in the F1 fragment (132 Miller units versus 65 Miller units, respectively) (Fig. 3). This correlates well with previously published transport data that show that when the rhaK72 allele is used to complement a chromosomal rhaK mutation in trans, it is capable of conferring very slow growth with some residual labeled rhamnose accumulation, whereas this was not observed with the rhaK73 allele (21). Although this may imply that the strength of the interaction is correlated with the ability to confer transport, more work will be needed to support this hypothesis.
Based on predictive modeling using the Phyre2 server, it was previously proposed that the rhaK72 and rhaK73 alleles encoded proteins with surface variations (21). Coupling this modeling approach with the experimental evidence generated in this work and with previous linker scanning results allowed us to delineate a conserved region within the N-terminal fragment of RhaK that had no predicted role. Changing any of the residues in this region resulted in a ≥50% loss in its ability to interact with the N-terminal fragment of RhaT in a bacterial 2-hybrid assay. Together, the data support the hypothesis that RhaK directly interacts with RhaT. At present, our data do not allow us to predict how this interaction might occur or what role the region proximal to the RhaK72 insertion in the C-terminal fragment of RhaK may play.
Phylogenetic analysis was previously used in an attempt to identify other RhaK-like proteins that may have the ability to interact with ABC transporters (19). Although the number of sequences was limited, putative orthologs were identified (19). Since a greater number of sequences are now available, repeating the analysis showed that RhaK-like sequences can be divided into separate clades, with Mesorhizobium, Sinorhizobium, and Rhizobium RhaK sequences all grouping separately. Although the rhaK genes that encode these proteins are found in syntenic operons predicted to be used for rhamnose catabolism, it is of note that the RhaK protein sequences show the greatest degree of divergence. For example, the RhaK proteins from R. leguminosarum Rlt100 and S. meliloti share only 55% identity and 67% similarity, whereas the associated RhaI proteins share 78% identity and about 90% similarity. The physiological/biochemical ramifications of these differences are unclear at this time, since rhaK from S. meliloti expressed from a plasmid can complement an rhaK mutation in R. leguminosarum for both transport and biochemical activity.
In light of this work, it is clear that the nature of the interaction of RhaK with RhaT should be further characterized. Whereas the physical characterization of the kinase may allow a more robust modeling of the interacting domains, it may be possible to use an analogous linker-scanning approach to find sites on RhaT that allow the transport of rhamnose to be RhaK independent. Moreover, since RhaK appears to be the most divergent component involved in the transport and catabolism of rhamnose, the characterization of orthologs may provide insight into some of the nuances of this interaction.
ACKNOWLEDGMENTS
This work was funded by an NSERC Discovery grant awarded to I.J.O. D.M.R.R. gratefully acknowledges partial support in the form of a University of Manitoba Faculty of Science Award.
FOOTNOTES
- Received 25 June 2015.
- Accepted 24 September 2015.
- Accepted manuscript posted online 28 September 2015.
- Address correspondence to Ivan J. Oresnik, ivan.oresnik{at}umanitoba.ca.
Citation Rivers DMR, Oresnik IJ. 2015. The sugar kinase that is necessary for the catabolism of rhamnose in Rhizobium leguminosarum directly interacts with the ABC transporter necessary for rhamnose transport. J Bacteriol 197:3812–3821. doi:10.1128/JB.00510-15.
REFERENCES
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
