Functional Mapping of Brucella abortus Cyclic β-1,2-Glucan Synthase: Identification of the Protein Domain Required for Cyclization

Abstract

Cyclic β-1,2-glucans (CβG) are periplasmic homopolysaccharides that have been shown to play an important role in several symbiotic and pathogenic relationships. Cyclic β-1,2-glucan synthase (Cgs), the enzyme responsible for the synthesis of CβG, is an integral membrane polyfunctional protein that catalyzes the four enzymatic activities (initiation, elongation, phosphorolysis, and cyclization) required for the synthesis of CβG. Recently, we have identified the glycosyltransferase and the β-1,2-glucooligosaccharide phosphorylase domains of Brucella abortus Cgs. In this study, we performed large-scale linker-scanning mutagenesis to gain further insight into the functional domains of Cgs. This analysis allowed us to construct a functional map of the enzyme and led to the identification of the minimal region required for the catalysis of initiation and elongation reactions. In addition, we identified the Cgs region (residues 991 to 1544) as being the protein domain required for cyclization and demonstrated that upon cyclization and releasing of the CβG, one or more glucose residues remain attached to the protein intermediate that serves as a primer for the next round of CβG synthesis. Finally, our results indicate that the overall control of the degree of polymerization of CβG is the result of a balance between elongation, phosphorolysis, and cyclization reactions.

Osmoregulated periplasmic glucans are cyclic, branched cyclic, or branched linear oligosaccharides present in the periplasm of certain gram-negative bacteria. Common features of these oligosaccharides are the presence of glucose as the sole sugar constituent and the regulation of their synthesis by the osmolarity of the growth medium (5). Agrobacterium, Rhizobium, Sinorhizobium, and Brucella species synthesize osmoregulated periplasmic glucans of family II. Glucans of this family have 17- to 25-glucose-residue cyclic β-1,2-backbones substituted with sn-1-phosphoglycerol, succinic acid, methylmalonic acid, or a combination of them (5, 6, 25).

Cyclic β-1,2-glucan synthase (Cgs), the enzyme responsible for the synthesis of cyclic β-1,2-glucans (CβG), is present in a restricted number of symbiotic or pathogenic bacteria, most of them belonging to the alphaproteobacteria, in which CβG are a symbiotic or virulence factor required for successful host interactions (4, 17, 24). In Brucella abortus, the etiological agent of bovine brucellosis, the synthesis and transport of CβG to the periplasmic space are required for the complete expression of virulence (7, 19, 25, 26). Furthermore, CβG plays a major role in circumventing host cell defense. Brucella CβG act on host cell membranes at the level of lipid rafts, controlling vacuole maturation, avoiding lysosome fusion, and allowing Brucella to reach an endoplasmic reticulum-derived vacuole permissive for bacterial replication (4). Thus, CβG is a Brucella virulence factor required for intracellular survival; accordingly, Cgs may be a good target for developing new chemotherapy alternatives against this pathogen.

Cgs catalyzes the four enzymatic activities (initiation, elongation, phosphorolysis, and cyclization) required for the synthesis of CβG (1, 10-12, 19). The initiation activity catalyzes the transference of a glucose residue from UDP-glucose to a not-yet-identified amino acid residue of the same protein. Elongation proceeds by the successive addition of glucose residues from UDP-glucose to the nonreducing end of the linear β-1,2-oligosaccharide protein-linked intermediate, whose length is controlled by the β-1,2-glucooligosaccharide phosphorylase activity. Finally, the polyglucose chain is cyclized, and the cyclic glucan is released from the protein.

Brucella abortus Cgs is a 320-kDa (2,867-amino-acid-residue) polytopic integral inner membrane protein with six transmembrane-spanning segments (TMSs), which define three small periplasmic loops and four large cytoplasmic regions delimited by amino acid residues 1 to 418, 475 to 818, 870 to 938, and 991 to 2867 (13). Cgs is a polyfunctional modular protein in which two regions can be recognized: an N-terminal region (amino acid residues 1 to 1544) and a C-terminal region (residues 1545 to 2867) (11). The N-terminal region constitutes the minimal region required for the catalysis of the initiation, elongation, and cyclization reactions; thus, it is a polyfunctional region by itself. We previously identified the glycosyltransferase domain in this region (12). The C-terminal region displays the β-1,2-glucooligosaccharide phosphorylase activity through which Cgs controls the degree of polymerization (DP) of the β-1,2-glucooligosaccharide protein-linked intermediate (11).

Besides the identification and characterization of the glycosyltransferase and phosphorylase domains, no information is available on the function of the other regions of Cgs. To complement and extend our knowledge of the functional domains of B. abortus Cgs, we performed large-scale linker insertion mutagenesis to generate in-frame insertions of 15 bp at random positions in the cgs gene. This methodology can provide invaluable insight into protein structure-function relationships and the different parts of polyfunctional proteins that are implicated in different activities, especially when no information can be obtained from sequence comparisons with other characterized proteins or from crystallographic data. Analyses of the in vitro and in vivo activities of the resulting mutants allowed us to construct a functional map of Cgs, delimit the minimal region required for the catalysis of initiation and elongation reactions, as well as identify the Cgs region involved in the cyclization reaction. We have also demonstrated that upon cyclization and releasing of the CβG, one or more glucose residues remain attached to the protein intermediate. A revised model for the modular structure of Cgs and the synthesis of CβG is proposed.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Escherichia coli K-12 strains DH5α-F′IQ (Invitrogen) and XL1-Blue MRF′ (Stratagene) were used as host for all plasmids used in this study. Agrobacterium tumefaciens mutant strain A1045 (with a Tn5 insertion in the chvB gene, coding for cyclic β-1,2-glucan synthase) and Brucella abortus mutant strain B2211 (gentamicin-resistant nonpolar mutant in the pgm gene, coding for phosphoglucomutase) were described previously (15, 16, 18, 32).

E. coli and A. tumefaciens strains were grown in Luria-Bertani (LB) broth (27) at 37°C and 28°C, respectively. B. abortus strains were grown at 37°C in Brucella broth (Difco Laboratories). If necessary, media were supplemented with appropriate antibiotics as follows: 100 μg/ml ampicillin for E. coli, 100 μg/ml carbenicillin for A. tumefaciens, 50 μg/ml kanamycin for E. coli and A. tumefaciens, 5 μg/ml nalidixic acid for B. abortus, and 4 μg/ml gentamicin for B. abortus.

Linker-scanning mutagenesis.

Random 15-bp insertions were generated using the linker-scanning mutagenesis GPS-LS kit (New England Biolabs) according to the manufacturer's instructions (3). Mutagenesis was accomplished by in vitro transposition with a Tn7-derived minitransposon with modified ends encoding PmeI restriction enzyme sites. Digestion with PmeI removes nearly the entire transposon, leaving an insertion of 15 bp after recircularization at the original site of transposition. Of this insertion, 10 bp (TGTTTAAACA) were contributed by the residual transposon itself, which code for the unique PmeI site (indicated in boldface type), and 5 bp were contributed by the duplication of the target site. In two out of the three possible frames, an in-frame insertion of 5 amino acids (pentapeptide insertion mutant) was generated. In the remaining frame, the TAA sequence of the PmeI site was read as an in-frame stop codon, resulting in a truncated protein at the site of insertion. Briefly, transposon mutagenesis reactions were performed in vitro with plasmids pBA24 (target DNA containing the B. abortus cgs gene and its own promoter) (11) and pGPS4 (transposon donor plasmid). E. coli XL1-Blue MRF′ cells were transformed by electroporation and selected with chloramphenicol and ampicillin. The resulting colonies were resuspended in LB medium, and total plasmidic DNA was extracted from the pool of colonies (without prior overnight growth in liquid culture to prevent competition during growth and a possible distortion of library representation), yielding mutant library I, in which all plasmids had a full-length transposon inserted. Library I was digested with PmeI, and DNA fragments corresponding to the target plasmid without the bulk of the transposon were religated; XL1-Blue MRF′ cells were transformed by electroporation and selected with ampicillin. The resulting colonies were recovered in LB medium, and total plasmidic DNA was purified directly from the pool of colonies, yielding mutant library II. All plasmids of library II have a 15-bp insertion with the unique PmeI site.

Motility assay and mapping.

Plasmids of library II were introduced in the A. tumefaciens A1045 mutant strain (chvB::Tn5) by electroporation. Individual clones were grown in 96-square-well titer plates (Beckman Coulter) containing 0.4 ml of LB medium per well at 28°C (200 rpm) during 24 h. After that time, 1 ml of LB medium was added to each well, and the plates were further incubated for 24 h. These cultures were used as inocula for the motility assay, which was carried out in yeast extract-mannitol medium (14) with 0.3% agar using a 96-well replica plater.

Plasmidic DNA from selected clones was recovered by triparental mating using E. coli HB101(pRK2013) (29) and C2110 (22) as helper and acceptor strains, respectively. The position of the 15-bp insertion was determined by restriction mapping, using the PmeI site generated at the insertion site, and DNA sequencing.

Construction of truncated mutants by PCR.

The truncated mutants Cgs-1587stop and Cgs-2079stop were constructed by PCR using forward oligonucleotide 5′-CTGCTCAGCCACGATCTT-3′ (p2236Fw) (the underlined sequence indicates a BlpI restriction site) and reverse oligonucleotides 5′-GCTCTAGATGTAGCCCGAACCATTGG-3′ (p4761Rv) (the underlined sequence indicates an XbaI restriction site) and 5′-GCTCTAGACGCCATAACCGTTCCAGA-3′ (p6237Rv) (the underlined sequence indicates an XbaI restriction site). Plasmid pBA24 was used as a template. The amplified DNA fragments of 2,500 and 4,000 bp were digested with BlpI and XbaI and ligated into pBA24 digested with the same enzymes, generating plasmids p1587stop and p2079stop, respectively. The reverse oligonucleotides introduce an in-frame stop codon, TAG (indicated in boldface type above).

Preparation of permeabilized cells and total membranes.

The preparation of permeabilized cells and total membranes of A. tumefaciens and B. abortus strains was carried out as described previously (8). The protein concentration was determined by a method described previously by Lowry et al. (23) using bovine serum albumin as a standard.

Western blotting.

Whole-cell extracts of the corresponding strains were subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (21), followed by electrophoretic transfer onto a nitrocellulose membrane (Inmobilon-NC; Millipore). Immunoblotting was performed as previously described (9) with a specific polyclonal antiserum against the Cgs region comprising amino acid residues 540 to 801 (13).

In vitro activity of Cgs.

Cgs activity was determined as described previously (8, 12). Briefly, permeabilized cells or total membrane fractions of A. tumefaciens and B. abortus strains were incubated with UDP-[¹⁴C]glucose (500,000 cpm; 300 μCi/μmol) in a solution containing 50 mM Tris-Cl buffer (pH 8.2) and 5 mM MgCl₂ at 28°C for 5 or 20 min. Nonidet P-40 (0.1%; Sigma) was added to permeabilized cells in the reaction mixture. For chase experiments, nonradioactive UDP-glucose (5 mM or 10 mM) was added after incubation with UDP-[¹⁴C]glucose, and the reaction mixture was further incubated for the indicated times. The reaction was stopped by the addition of 10% trichloroacetic acid (TCA) to the mixture, and the TCA-insoluble fractions were subjected to SDS-PAGE (21). Proteins were stained with Coomassie blue, and radioactivity was detected by autoradiography.

In vivo activity of Cgs: TLC analysis of CβG.

Cells from 3 ml of stationary-phase cultures (optical density at 600 nm, 3.0) of the A. tumefaciens strains grown for 24 h at 28°C (200 rpm) were harvested by centrifugation. CβG were extracted from cell pellets with ethanol (70% ethanol for 1 h at 37°C) and analyzed by thin-layer chromatography (TLC) as described previously (12).

Phosphorylase enzyme assay.

Total membrane fractions of A. tumefaciens strain A1045 harboring the indicated plasmids were incubated with NaH₂³²PO₄ (222,000 cpm; 1 μCi/μmol), and the radioactive products in the supernatant were analyzed by paper electrophoresis as previously described (11).

Isolation and characterization of glycopeptides.

Total membrane fractions of A. tumefaciens strain A1045 harboring the indicated plasmid were incubated with UDP-[¹⁴C]glucose (500,000 cpm; 300 μCi/μmol) at 28°C for 20 min. The reaction was stopped with 10% TCA. Washed TCA precipitates were treated with type XIV protease from Streptomyces griseus (Sigma), and the [¹⁴C]glucose-labeled glycopeptides were isolated from the supernatant and characterized as described previously (11).

Protein sequence alignment.

BLAST (2) searches were performed to identify close orthologs of B. abortus Cgs (GenBank accession no. AAC34747.2). A multiple sequence alignment was performed by using the ClustalW program (31) with the following protein sequences: GenBank accession numbers NP_533395.1 from A. tumefaciens, P20471 from Sinorhizobium meliloti, NP_108444.1 from Mesorhizobium loti, YP_001368684.1 from Ochrobactrum anthropi, YP_031810.1 from Bartonella quintana, YP_001007373.1 from Yersinia enterocolitica, ZP_01639054.1 from Pseudomonas putida, YP_625114.1 from Burkholderia cenocepacia, YP_001219908.1 from Acidiphilium cryptum, and YP_877436.1 from Clostridium novyi. From the multiple sequence alignment, highly conserved regions were identified, and the percentage of similarity was calculated for these regions from the pair-base alignment generated with the B. abortus Cgs and A. tumefaciens ChvB proteins.

RESULTS

Generation and analysis of linker mutants.

A library of random 15-bp insertion mutants in the cgs gene was constructed using the GPS-LS linker-scanning mutagenesis system (New England Biolabs) (3). By this procedure, two different types of protein mutants were generated: in-frame pentapeptide insertion mutants and truncated mutants. Pentapeptide insertion mutants were named with a number indicating the position of the last amino acid residue of the protein upstream of the pentapeptide insertion; truncated mutants were named with a number indicating the position of the last amino acid residue of Cgs and the word “stop,” indicating that an in-frame stop codon was generated by the 15-bp insertion.

Cyclic β-1,2-glucan synthase of Agrobacterium tumefaciens is encoded by the chvB gene, homologous to the B. abortus cgs gene (16). Besides being inactive for the synthesis of CβG, chvB mutants are also defective in the assembly of flagella having a nonmotile phenotype (16, 24). This phenotype can be restored by heterologous complementation with the B. abortus cgs gene (19). Therefore, we tested the activities of cgs linker mutants for their abilities to complement the motility of an A. tumefaciens chvB::Tn5 mutant strain (see Fig. S1 in the supplemental material). This heterologous expression strategy was used so that we could screen a large number of mutants by simply measuring the motility phenotype, since B. abortus is nonmotile.

A total of 2,464 mutant strains were analyzed, and 343 of them were mapped by restriction analysis and DNA sequencing to determine the exact localization and the reading frame of the insertion (see Tables S1, S2, and S3 in the supplemental material).

Analysis of Cgs functional maps.

With the information obtained from the motility assays and mapping, we constructed two functional maps of the enzyme that correspond to (i) truncated proteins and (ii) in-frame pentapeptide insertion mutants (Fig. 1).

FIG. 1.

Functional maps of Cgs. (A) Distribution and motility phenotypes of truncated mutants. (B) Distribution and motility phenotypes of pentapeptide insertion mutants. The C-terminal region at residues 1545 to 2867 is not to scale. In the Cgs region at residues 475 to 818, the active-site residues (F₅₀₅, D₅₃₈, D₆₃₅xD₆₃₇, E₇₆₉, and R₇₈₁xxR₇₈₄W₇₈₅) of the glycosyltransferase (GT-84) domain are indicated. Gray boxes indicate regions with a high percentage of similarity between Cgs and its orthologs. “1” and “2” indicate the highly conserved subregions of the Cgs region at residues 991 to 1544. Arrows and numbers indicate the positions of the last amino acid residues upstream of the insertion. Numbers below the bars indicate amino acid residues of Cgs. I to VI indicate TMSs. Active, inactive, and partially active indicate motile mutants, nonmotile mutants, and mutants with decreased motility, respectively. Truncated and pentapeptide insertion mutants used for the construction of the functional maps are listed in Tables S2 and S3 in the supplemental material, respectively. Cgs-1587stop and Cgs-2079stop mutants were generated by PCR (see Materials and Methods).

Analysis of the functional maps revealed that all truncated mutants located in the N-terminal region (amino acid residues 1 to 1544) are nonmotile (Fig. 1A), while all the truncated and pentapeptide insertion mutants that map to the C-terminal half of the protein (residues 1545 to 2867) are motile (Fig. 1). We previously showed (11) that truncated mutants in the N-terminal region do not synthesize CβG, while truncated and pentapeptide insertion mutants in the C-terminal half synthesize CβG but with a higher DP than that of the wild-type strain. Taken together, these results confirm the strict correlation between the restoration of motility and the synthesis of CβG and indicate that the lack of control of CβG DP does not affect the capacity to restore the motility phenotype of the A. tumefaciens chvB mutant strain.

Contrary to what was observed with truncated mutants, the pentapeptide insertion functional map revealed that in the N-terminal region (residues 1 to 1544), inactive (nonmotile), active (motile), and partially active (decreased motility) mutants were obtained (Fig. 1B). Furthermore, the distribution of pentapeptide insertion mutants in the N-terminal half of the protein is not uniform; that is, there are regions with clusters of inactive insertions, while other regions are devoid of inactive insertions or have a few active or partially active mutants, such as Cgs regions at residues 1 to 418 and 1120 to 1270. In these two regions, truncated mutants were isolated (Fig. 1A), so the low number of pentapeptide insertions isolated is not due to a target-site selectivity of the transposition reaction but is rather due to a bias introduced by selecting predominantly nonmotile mutants for the analysis (see Table S1 in the supplemental material). These results indicate that the Cgs N terminus contains regions that are sensitive to the insertion of the pentapeptide, such as Cgs regions at residues 475 to 818, 870 to 938, 991 to 1120, and 1270 to 1544, and regions that are insensitive or partially sensitive to the insertion, such as Cgs regions at residues 1 to 418 and 1120 to 1270.

Finally, sequence analysis (see Materials and Methods) revealed that Cgs displays an overall sequence similarity with close orthologs of about 50%; however, some regions were highly conserved, with a percentage of similarity between 60 and 75%, while other regions showed a percentage of sequence similarity of less than 50%. As depicted in Fig. 1B, most of the inactivating pentapeptide insertions fall into highly conserved regions, such as the Cgs regions at residues 475 to 818, 906 to 928, 1010 to 1120, and 1270 to 1530, while noninactivating (active and partially active) insertions fall mostly into less conserved regions, such as regions at residues 1 to 418 and 1120 to 1270, indicating that there is a significant correlation between the effect of pentapeptide insertions and the degree of conservation of the targeted region.

Characterization of truncated mutants.

Based on the analysis of the in vivo activities (TLC analysis of CβG) of several truncated mutants distributed along the protein sequence, we previously determined that truncated proteins at amino acid 1472 and upstream of this position were inactive for the synthesis of CβG (11). Conversely, Cgs truncated at position 1587 and downstream of this position synthesized CβG but with a DP higher than that of the wild-type strain (11).

In this study, we analyzed the in vitro activities of selected truncated Cgs mutants. Western blot analysis revealed that all the truncated proteins displayed the expected molecular weights according to the localization of the stop codon (Fig. 2A and data not shown). The in vitro activity was analyzed incubating permeabilized cells with UDP-[¹⁴C]glucose as described in Materials and Methods. As depicted in Fig. 2B, the truncated mutants Cgs-837stop to Cgs-983stop do not incorporate [¹⁴C]glucose in the protein intermediate; similar results were obtained with the truncated mutants located upstream of position 837 (data not shown). On the other hand, the truncated protein at position 1004 and all the downstream truncated forms incorporated [¹⁴C]glucose in the protein intermediate (Fig. 2B and data not shown). Interestingly, the truncated mutants Cgs-1004stop to Cgs-1472stop incorporated [¹⁴C]glucose in the protein intermediate (Fig. 2B) but failed to synthesize CβG (11). Since cyclization is coupled to the release of CβG from the protein intermediate, this finding suggests that cyclization is impeded in these mutants and that initiation and elongation reactions are uncoupled from cyclization.

FIG. 2.

In vitro characterization of truncated Cgs mutants. (A) Western blot analysis. Whole-cell extracts (100 μg of total proteins) of A. tumefaciens strain A1045 harboring the indicated plasmid were subjected to 10% SDS-PAGE. Proteins were transferred into a nitrocellulose membrane and detected with a specific polyclonal antiserum against the Cgs region comprising amino acid residues 540 to 801. (B) Incorporation of [¹⁴C]glucose into Cgs. Permeabilized cells of A. tumefaciens strain A1045 harboring the indicated plasmids were incubated with UDP-[¹⁴C]glucose during 20 min at 28°C. TCA-insoluble fractions were analyzed by 10% SDS-PAGE, and the radioactivity was detected by autoradiography. pBA24, plasmid expressing wild-type Cgs. A1045, A. tumefaciens A1045 mutant strain without a plasmid carrying the cgs gene from B. abortus. The position of molecular mass standards (in kDa) is indicated on the left. The arrows on the right indicate the position of wild-type Cgs.

In summary, three different classes of truncated mutants can be distinguished based on the in vivo and in vitro activity analysis: (i) inactive mutants for the synthesis of CβG that fails to incorporate [¹⁴C]glucose in the protein intermediate (mutants truncated upstream of position 983), (ii) inactive mutants for the synthesis of CβG that incorporate [¹⁴C]glucose in the protein intermediate (mutants at positions 1004 to 1472), and (iii) truncated mutants that synthesize CβG and incorporate [¹⁴C]glucose in the protein intermediate but produce CβG with a higher DP than that of the wild-type protein (mutants at positions 1587 to 2773).

Taken together, these results indicate that the Cgs region comprising amino acid residues 1 to 1004 represents the minimal region required for the synthesis of the linear β-1,2-glucooligosaccharide protein-linked intermediate and, hence, for the catalysis of the glucosyltransferase reactions (initiation and elongation) and suggest that the Cgs region from amino acid residues 1004 to 1472 may be implicated in the cyclization reaction.

Characterization of pentapeptide insertion mutants.

Several pentapeptide insertion mutants were selected from the different Cgs cytoplasmic regions, and the in vivo and in vitro activities were analyzed (see text in the supplemental material and Fig. S2, S3, and S4 in the supplemental material).

In the Cgs region at residues 991 to 1544, two clusters of nonmotile pentapeptide insertion mutants that fall into the highly conserved regions at residues 1010 to 1120 (subregion 1) and residues 1270 to 1530 (subregion 2) were isolated (Fig. 1B). The in vivo and in vitro activities of several pentapeptide insertions that map to these highly conserved subregions were assayed. The selected mutants failed to synthesize CβG (Fig. 3A) but incorporated [¹⁴C]glucose to the protein intermediate at wild-type levels (Fig. 3B), indicating that glucose residues may be incorporated in the protein through initiation and elongation reactions, but the protein-linked β-1,2-glucooligosaccharide may not be efficiently released through the cyclization reaction. These results are in agreement with those obtained with the truncated mutants that map in the Cgs region at residues 991 to 1544, suggesting again that this region is implicated in the cyclization of the glucan.

FIG. 3.

In vivo and in vitro characterization of pentapeptide insertion mutants of the Cgs region at residues 991 to 1544. (A) TLC analysis of CβG produced by A. tumefaciens strain A1045 harboring the indicated plasmid. * and **, migration of anionic and neutral CβG, respectively. (B) Incorporation of [¹⁴C]glucose into Cgs. Permeabilized cells of A. tumefaciens strain A1045 harboring the indicated plasmids were incubated with UDP-[¹⁴C]glucose during 5 min at 28°C. TCA-insoluble fractions were analyzed by Coomassie blue-staining 10% SDS-PAGE, and the radioactivity was detected by autoradiography. The level of [¹⁴C]glucose incorporation into Cgs and the relative level of protein expression were determined using 1D image analysis software (Kodak Digital Science). [¹⁴C]glucose incorporation into Cgs was normalized by the expression level of the corresponding protein and expressed as a percentage of that of wild-type Cgs. Data represent the means and standard deviations for two separate determinations. “1” and “2” indicate the highly conserved subregions of the Cgs region at residues 991 to 1544 (Fig. 1B). pBA24, plasmid expressing wild-type Cgs.

The Cgs region at residues 991 to 1544 is implicated in the cyclization reaction.

It was previously described that Cgs can be pulse-labeled with UDP-[¹⁴C]glucose and chased with nonradioactive UDP-glucose due to the fact that a linear β-1,2-glucooligosaccharide protein-linked intermediate is formed during the synthesis of CβG and that cyclization is coupled to the release of CβG (8, 19, 36, 37). We reasoned that a mutant defective in the cyclization reaction may not be able to chase the labeled glucose incorporated into the protein. Pulse and chase experiments were performed by incubating permeabilized cells with UDP-[¹⁴C]glucose (pulse), followed by the addition of nonradioactive UDP-glucose (chase). [¹⁴C]glucose incorporation into Cgs was normalized by the expression level of the corresponding protein and the ratio of incorporation of [¹⁴C]glucose at 5 min of pulse/incorporation of [¹⁴C]glucose at 5 min of pulse plus 5 min of chase was calculated. As depicted in Fig. 4, radioactivity incorporated by the mutants was not chased after the addition of nonlabeled UDP-glucose. These results indicate that the Cgs region at residues 991 to 1544 is implicated in the cyclization reaction.

FIG. 4.

Characterization of pentapeptide insertion mutants of the Cgs region at residues 991 to 1544 by pulse and chase assays. (A) Permeabilized cells of A. tumefaciens strain A1045 harboring the indicated plasmids were incubated with UDP-[¹⁴C]glucose during 5 min (pulse [P]) or 10 min, adding nonradioactive UDP-glucose (5 mM) after the first 5 min of incubation (chase [C]). TCA-insoluble fractions were analyzed by Coomassie blue-staining 10% SDS-PAGE (top), and the radioactivity was detected by autoradiography (bottom). The position of the molecular mass standard (in kDa) is indicated on the left. The arrows on the right indicate the position of wild-type Cgs. Only the analysis of three representative mutants is shown. (B) From the analysis described above, the relative levels of [¹⁴C]glucose incorporation (determined by autoradiography) and protein expression (determined by Coomassie blue-stained SDS-PAGE) were determined using 1D image analysis software (Kodak Digital Science). [¹⁴C]glucose incorporation into Cgs was normalized by the expression level of the corresponding protein, and the ratio of incorporation of [¹⁴C]glucose at 5 min of pulse/incorporation of [¹⁴C]glucose at 5 min of pulse plus 5 min of chase (IP/IC ratio) was calculated. Data represent the means and standard deviations for two separate determinations. “1” and “2” indicate the highly conserved subregions of the Cgs region at residues 991 to 1544 (Fig. 1B). pBA24, plasmid expressing wild-type Cgs.

We previously determined that the length of the linear β-1,2-glucooligosaccharide protein-linked intermediate is controlled by that Cgs C-terminal domain that, acting as a β-1,2-glucooligosaccharide phosphorylase, catalyzes the phosphorolysis of the β-1,2-glucosidic bond at the nonreducing end of the protein-linked polyglucose chain, releasing glucose 1-phosphate (11). Total membrane fractions of several pentapeptide insertion mutants that map in the Cgs region at residues 991 to 1544 were isolated and incubated with UDP-[¹⁴C]glucose. All the analyzed mutants incorporated [¹⁴C]glucose to the protein intermediate at wild-type levels but were inactive for the synthesis of CβG (data not shown), confirming previous results obtained with permeabilized cells. Next, to investigate whether these mutants display phosphorylase activity, the obtained total membrane fractions were incubated with ³²P_i in the presence of the sugar-donor substrate UDP-glucose. All the tested mutants released glucose 1-phosphate, indicating that they retained the phosphorylase activity catalyzed by the C-terminal region of the protein (Fig. 5A). Therefore, in these mutants, the cyclization activity is uncoupled from initiation, elongation, and phosphorylase activities. An additional radioactive product, migrating as a UTP standard, was observed (Fig. 5A). The formation of this product depended on the synthesis of the protein-linked β-1,2-glucooligosaccharide regardless of the integrity of the cyclization (Fig. 5A) and phosphorylase domains (11). The pathway leading to the formation of the putative UTP deserves further studies. Furthermore, the fact that these mutants retain the phosphorylase activity, which specifically catalyzes the phosphorolysis of linear β-1,2-glucooligosaccharides (11), demonstrates that the oligosaccharide linked to the protein in the mutants of the Cgs region at residues 991 to 1544 is a linear polyglucose chain with β-1,2-linkages.

FIG. 5.

Determination of phosphorylase activity and analysis of glycopeptides of pentapeptide insertion mutants of the Cgs region at residues 991 to 1544. (A) Determination of phosphorylase activity. Total membrane fractions of A. tumefaciens strain A1045 harboring the indicated plasmids were incubated with ³²P_i, and the radioactive products were analyzed by paper electrophoresis. Radioactivity was detected by autoradiography. Standards: Glc, glucose; Glc 1P, glucose 1-phosphate; UDP-Glc, UDP-glucose; Pi, inorganic phosphate. (B) Analysis of glycopeptides. Total membrane fractions of A. tumefaciens strain A1045 harboring the indicated plasmids were incubated with UDP-[¹⁴C]glucose during 20 min at 28°C. Washed TCA precipitates were subjected to an extensive protease treatment, and [¹⁴C]glucose-labeled glycopeptides were recovered from the supernatant and analyzed by Bio-Gel P4 column chromatography. Radioactivity was quantified by liquid scintillation. V₀, void volume; pBA24, plasmid expressing wild-type Cgs; WT, wild type.

Glycopeptides can be obtained after extensive protease treatment of β-1,2-glucooligosaccharide protein intermediates (37). After this treatment, most of the glycopeptides have only one amino acid attached to the reducing end of the β-1,2-glucooligosaccharide (35); thus, the elution volume from a Bio-Gel P4 column depends on the DP of the polyglucose chain linked to the protein. To characterize the size of the linear β-1,2-glucooligosaccharides linked to the protein, total membrane fractions of the different strains were incubated with UDP-[¹⁴C]glucose. Glucooligosaccharide protein intermediates were subjected to protease treatment, and [¹⁴C]glucose-labeled glycopeptides were analyzed by Bio-Gel P4 column chromatography. Mutants Cgs-1354 and Cgs-1521 yielded glycopeptides with a higher DP than that of the wild type (Fig. 5B) (similar results were obtained with mutants Cgs-1066 and Cgs-1448 [data not shown]), indicating that these mutants accumulate a linear β-1,2-glucooligosaccharide linked to the protein with a higher DP than that of the wild-type strain.

The cyclization reaction does not occur on the first glucose linked to the protein.

The cyclization reaction catalyzed by Cgs may be an intramolecular transglycosylation reaction during which the nonreducing end of the protein-linked oligosaccharide forms a new glucosidic bond with a glucose of the β-1,2-sugar backbone, and concomitantly, the aminoacyl-glucose or a glucose-β-1,2-glucose linkage is cleaved, resulting in the formation and release of CβG.

In order to investigate whether transglycosylation occurs on the first glucose linked to the protein cleaving the aminoacyl-glucose linkage or on an internal glucose cleaving a glucose-β-1,2-glucose linkage, permeabilized cells of B. abortus wild-type strain 2308 (28) and a pgm mutant strain were incubated with UDP-[¹⁴C]glucose (pulse), followed by the addition of nonradioactive UDP-glucose (chase). pgm codes for the enzyme phosphoglucomutase that catalyzes the interconversion of glucose 6-phosphate to glucose 1-phosphate, a key enzymatic step required for the synthesis of UDP-glucose, the sugar donor substrate for the synthesis of CβG. Therefore, pgm mutants do not synthesize in vivo UDP-glucose, and consequently, the Cgs remains nonglucosylated (33). However, permeabilized cells or total membrane fractions from pgm mutants incubated in vitro with UDP-[¹⁴C]glucose are able to transfer glucose residues to the protein intermediate and synthesize CβG (33). When the incubation was performed with the wild-type strain, a complete chase of the labeled glucose residues incorporated into the Cgs protein was observed (Fig. 6). Instead, the incubation of pgm-permeabilized cells resulted in a partial chase of the radioactivity incorporated into the protein even at long incubation times with a high concentration of nonlabeled UDP-glucose (Fig. 6). In the pgm mutant strain, the Cgs protein is recovered nonglucosylated, while in the wild-type strain, it is recovered already glucosylated by nonradioactive UDP-glucose. Hence, the higher radioactive signal observed in the Cgs protein of the pgm mutant strain is due to the fact that it incorporates more labeled glucose residues than does wild-type Cgs. Taken together, these results indicate that during the cyclization reaction, Cgs catalyzes the cleavage of an internal glucose-β-1,2-glucose linkage of the growing glucan chain, leaving one or more glucoses linked to the protein intermediate after the release of the cyclic glucan.

FIG. 6.

The cyclization reaction does not occur on the first glucose linked to the protein. Permeabilized cells of the B. abortus wild-type strain (WT) and the pgm mutant strain were incubated with UDP-[¹⁴C]glucose during 10, 20, or 30 min (pulse [P]). After 10 min of incubation with UDP-[¹⁴C]glucose, nonradioactive UDP-glucose (10 mM) was added, and the reaction mixture was further incubated for the indicated times (chase [C]). TCA-insoluble fractions were analyzed by Coomassie blue-staining 10% SDS-PAGE, and radioactivity was detected by autoradiography. In each lane of the gel, equal amounts of Cgs protein were loaded. The position of the molecular mass standard (in kDa) is indicated on the left. The arrow on the right indicates the position of the Cgs protein.

DISCUSSION

Most studies of protein structure-function rely on the analysis of primary sequence and identification of conserved amino acid residues of the catalytic active site. Recently, utilizing this approach, we have identified and confirmed the active-site residues of the glucosyltransferase and the β-1,2-glucooligosaccharide phosphorylase domains of B. abortus Cgs by site-directed mutagenesis (11, 12). In this study, we performed large-scale linker-scanning mutagenesis of B. abortus Cgs. There is a potential for such linker insertions to impact the overall structure of the target protein, and the effects may not be localized strictly to the site of the insertion. Nonetheless, our data obtained using the linker insertion mutagenesis approach are consistent with those of previous studies and allowed us to identify a novel functional region of cyclic β-1,2-glucan synthases. Several truncated and pentapeptide insertion Cgs mutants were tested for their abilities to complement the motility phenotype of an A. tumefaciens chvB mutant strain as a measure of the overall activity of Cgs. We further characterize these mutants by evaluating the synthesis of CβG and the formation of the β-1,2-glucooligosaccharide protein-linked intermediate in vivo and in vitro. In all the tested mutants, a strict correlation between the restoration of motility and synthesis of CβG was observed, thus validating our approach.

In general, we observed that Cgs regions at residues 1 to 418 and 1120 to 1270 can tolerate pentapeptide insertions without a significant loss of enzymatic activity (Fig. 1B; see also Fig. S2 in the supplemental material). Instead, pentapeptide insertion mutants in the Cgs region at residues 475 to 818, the glycosyltransferase (GT-84) domain, were inactive for the synthesis of CβG, and no incorporation of [¹⁴C]glucose in the protein intermediate was detected, or it was significantly reduced (see Fig. S3 in the supplemental material). These results are in agreement with those previously obtained after a site-directed disruption of the GT-84 domain (12), which resulted in the failure of the protein to catalyze the initiation and/or elongation reaction(s) required for the synthesis of the linear β-1,2-glucoologosaccharide protein-linked intermediate. In the Cgs region at residues 870 to 938, a cluster of inactive mutants was isolated, indicating that this region may be structurally and/or functionally important for the synthesis of CβG (Fig. 1B; see also Fig. S4 in the supplemental material). Finally, our main finding is that the Cgs region at residues 991 to 1544 comprises the protein domain required for cyclization. This activity can be uncoupled from the initiation, elongation, and phosphorolysis activities, indicating that cyclic β-1,2-glucan synthases are polyfunctional proteins comprised of discrete domains.

Amino acid sequence analysis of the Cgs region at residues 991 to 1544 along with protein fold recognition methods do not reveal any significant homology to other proteins in databases (12), not even to those that also catalyze cyclization reactions such as amylomaltases, cyclodextrin glucanotransferases, and potato D enzyme (20, 30). However, two highly conserved subregions between Cgs and its orthologs were identified in this region (Fig. 1B). It is worth remarking that the distribution of inactive pentapeptide insertion mutants correlated very well with this pattern of protein conservation. Pulse and chase experiments (Fig. 4) along with analysis of the in vivo and in vitro activity using either permeabilized cells or total membrane fractions (Fig. 2 and 3 and data not shown) revealed that truncated and pentapeptide insertion mutants of the Cgs region at residues 991 to 1544 failed to cyclize and release the linear β-1,2-glucan chain from the protein. Taken together, these results demonstrate that the integrity of the Cgs region at residues 991 to 1544 is required for cyclization and confirm that cyclization is coupled to the release of CβG. Although truncations and pentapeptide insertions may have an indirect effect on the integrity of the cyclization active site, the fact that these protein mutants retain the glucosyltransferase activities (initiation and elongation) and the phosphorylase activity (Fig. 5A), with conservation of the type of linkage of the protein-linked glucooligosaccharide, strongly suggests that nonsignificant conformational or structural changes in the overall structure of the protein were introduced and reinforces the hypothesis that the cyclization active site may be located in this region of the protein.

It was previously suggested that the size distribution of CβG products, which varies with species, depends on competition between elongation and cyclization reactions (34). B. abortus Cgs produces CβG with a DP ranging from 17 to 22. Recently, we demonstrated that the length of the β-1,2-glucooligosaccharide protein-linked intermediate results from a balance between the opposing activities of elongation and phosphorolysis reactions (11). Pentapeptide insertion mutants in the Cgs region at residues 991 to 1544 accumulate a linear β-1,2-glucooligosaccharide linked to the protein with a higher DP than that of the wild-type strain (Fig. 5B), although phosphorylase activity is not affected (Fig. 5A). This result indicates that the accumulation of a polyglucose chain with a high DP is not due to an indirect abolishment of phosphorylase activity and reveals that when the cyclization reaction is impeded, phosphorolysis is not sufficient to control the DP of the linear β-1,2-glucooligosaccharide intermediate. It may be concluded, therefore, that the overall control of CβG DP is the result of a balance between elongation, phosphorolysis, and cyclization reactions.

Cyclization probably proceeds through an intramolecular transglycosylation reaction, during which part of the donor that has been cleaved off also acts as the acceptor, resulting in the formation of CβG. Pulse and chase experiments carried out with the wild-type and the pgm mutant strains demonstrate that during the cyclization reaction, Cgs catalyzes the cleavage of an internal glucose-β-1,2-glucose bond of the growing glucan chain and not the aminoacyl-glucose linkage, leaving one or more glucoses linked to the protein intermediate after the release of the cyclic glucan (Fig. 6). Therefore, the cyclization reaction catalyzed by Cgs can be described as follows: Cgs − Glc(n)→Cgs − Glc(n − x) + cyclic β-1,2-glucan(x) where n and x indicate the number of glucose residues. The Cgs-Glc (n − x) product then serves as a primer for the elongation reaction, initiating a new round of CβG synthesis in this way.

An important feature of Cgs is the ability to form a covalent intermediate during the synthesis of CβG. The initiation of CβG synthesis consists of the transfer of the first glucose from UDP-glucose to a not-yet-identified amino acid residue of the enzyme. Analyses of the in vitro activities of truncated mutants allowed us to identify the Cgs region at residues 1 to 1004 as being the minimal region required for the catalysis of initiation and elongation reactions, indicating that the amino acid residue to which the first glucose is transferred is located in this region. Work is in progress to identify the amino acid residue and the nature of the linkage between the glucose and the amino acid residue.

On the basis of evidence presented here and from past work, we propose a revised model for the modular structure of Cgs and the synthesis of CβG (Fig. 7). Cgs itself acts as a protein intermediate and catalyzes the four enzymatic reactions required for the synthesis of CβG. The first glucose is transferred from UDP-glucose to a not-yet-identified amino acid residue of the protein (initiation reaction), which is located in the Cgs region at residues 1 to 1004. Successive glucoses are then transferred from UDP-glucose to the protein-bound glucose, thus elongating a linear polyglucose chain (elongation reaction). The elongation and, maybe, the initiation reaction are catalyzed by the glycosyltransferase (GT-84) domain. At some point, probably after the glucan chain length reaches more than 17 glucose residues, the removing activity catalyzed by the β-1,2-glucooligosaccharide phosphorylase (GH-94) domain counteracts Cgs self-glucosylation, controlling the length of the β-1,2-glucooligosaccharide protein-linked intermediate. The cyclization reaction catalyzed by the Cgs region at residues 991 to 1544 puts an end to the balance between the elongation reaction and the opposing action of phosphorylase, the linear β-1,2-glucooligosaccharide protein-linked intermediate is cyclized, and CβG, with the appropriate ring size, are released from the protein, leaving one or more glucoses linked to the protein, which serves as a primer for the next round of CβG synthesis.

FIG. 7.

Proposed model for the modular structure of Cgs and synthesis of CβG. (A) Modular structure of B. abortus Cgs. The cyclization domain (amino acid residues 991 to 1544) identified in this study is indicated. GT-84, glucosyltransferase domain; GH-94, β-1,2-glucooligosaccharide phosphorylase domain. The region at amino acid residues 1 to 1004 is the minimal region required for the catalysis of initiation and elongation reactions; the region at amino acid residues 1 to 1545 is the minimal region required for the catalysis of initiation, elongation, and cyclization reactions. I to VI indicate TMSs. (B) Proposed mechanism for the synthesis of CβG. Cgs itself acts as a protein intermediate and catalyzes the four enzymatic reactions required for the synthesis of CβG: initiation (I), elongation (E), phosphorolysis (P), and cyclization (C). The initiation activity catalyzes the transference of the first glucose from UDP-glucose to a not-yet-identified amino acid residue. Elongation proceeds by the successive addition of glucose residues from UDP-glucose to the nonreducing end of the β-1,2-glucooligosaccharide protein-linked intermediate, whose length is controlled by the β-1,2-glucooligosaccharide phosphorylase activity. Finally, the β-1,2-glucooligosaccharide protein-linked intermediate is cyclized, and CβG is released from the protein, leaving one or more glucoses linked to the protein, which serves as a primer for the next round of CβG synthesis. Light gray circles, glucose; dark gray circles, glucose at the nonreducing end of the polyglucose chain linked to the protein; Pi, inorganic phosphate. The GT-84 domain between TMSs II and III and the C-terminal GH-94 domain are indicated in gray. The cyclization domain is indicated in light gray.

As we previously reported, CβG is a Brucella virulence factor required for proper intracellular trafficking and survival (4, 7, 26). Characterization of the different functional domains of Cgs as well as the identification of key amino acid residues may lead to the rational design of specific inhibitors of the enzyme, for example, of the cyclization reaction, which may be useful for the development of new chemotherapy alternatives against this pathogen.