Glucuronyltransferase Activity of KfiC from Escherichia coli Strain K5 Requires Association of KfiA

Heparan sulfate is a ubiquitous glycosaminoglycan in the extracellular matrix of most animals. It interacts with various molecules and exhibits important biological functions. K5 antigen produced by Escherichia coli strain K5 is a linear polysaccharide N-acetylheparosan consisting of GlcUA β1–4 and GlcNAc α1–4 repeating disaccharide, which forms the backbone of heparan sulfate. Region 2, located in the center of the K5-specific gene cluster, encodes four proteins, KfiA, KfiB, KfiC, and KfiD, for the biosynthesis of the K5 polysaccharide. Here, we expressed and purified the recombinant KfiA and KfiC proteins and then characterized these enzymes. Whereas the recombinant KfiC alone exhibited no GlcUA transferase activity, it did exhibit GlcUA transferase and polymerization activities in the presence of KfiA. In contrast, KfiA had GlcNAc transferase activity itself, which was unaffected by the presence of KfiC. The GlcNAc and GlcUA transferase activities were analyzed with various truncated and point mutants of KfiA and KfiC. The point mutants replacing aspartic acid of a DXD motif and lysine and glutamic acid of an ionic amino acid cluster, and the truncated mutants deleting the C-terminal and N-terminal sites, revealed the essential regions for GlcNAc and GlcUA transferase activity of KfiC and KfiA, respectively. The interaction of KfiC with KfiA is necessary for the GlcUA transferase activity of KfiC but not for the enzyme activity of KfiA. Together, these results indicate that the complex of KfiA and KfiC has polymerase activity to synthesize N-acetylheparosan, providing a useful tool toward bioengineering of defined heparan sulfate chains.

acid (IdoUA)) and D-glucosamine (GlcN) residues carrying sulfogroups at various sites of sugar residues. Usually, HS chains are covalently attached to a core protein in the form of proteoglycans and are present ubiquitously on the cell surface and in the extracellular matrices of animals. HS chains interact with cytokines, growth factors, coagulation factors, proteases and their inhibitors, and other molecules and contribute to several biological processes, including development, morphogenesis, cell proliferation and differentiation, and cancer cell invasion (1,2).
The biosynthesis of HS begins with the synthesis of the linkage tetrasaccharide, GlcUA-Gal-Gal-Xyl, on the serine residues of core proteins. The backbone of HS polysaccharide is then synthesized onto the linkage tetrasaccharide by alternating addition of monosaccharide units of GlcNAc and GlcUA with ␣1-4 and ␤1-4 bonds, respectively, using UDP-sugar donors with desired GlcNAc transferase (GlcNAc-T) and GlcUA transferase (GlcA-T). The elongation reaction is performed by HS co-polymerases, EXT1 and EXT2, in the Golgi apparatus (3,4).  (5,6).
Capsular polysaccharides are a Gram-negative bacteria coat on the outer membrane of bacterial cells. They play important roles in virulence and survival to protect against nonspecific host defense (7). K5 antigen produced by Escherichia coli strain K5 is a linear capsular polysaccharide consisting of GlcUA ␤1-4 and GlcNAc ␣1-4 repeating disaccharide, which is an analogue of the unsulfated and unepimerized HS backbone (N-acetylheparosan, HPR) (8). The HPR chain of K5 antigen is synthesized onto a phospholipid initiator by the alternate addition of the two monosaccharides on the cytoplasmic surface of the plasma membrane (9). The nascent polysaccharide moves across the plasma membrane, periplasmic space, and the outer membrane and then forms the capsule on the extracellular layer.
A gene cluster of K5 antigen consists of three functional regions termed 1, 2, and 3. Region 2, located in the middle, is serotype-specific, and it encodes four proteins, KfiA, KfiB, KfiC, and KfiD (Fig. 1A), required for synthesis of the K5 polysaccharide (10). KfiD is a UDPglucose dehydrogenase that catalyzes the formation of UDP-GlcUA from UDP-glucose (11), and KfiB is thought to be a binding protein. In the previous study using crude cell lysate and mutagenesis assays (12,13), KfiA and KfiC were suggested to possess GlcNAc-T and GlcA-T activity, respectively. Another study showed that recombinant KfiA protein exhibits GlcNAc-T activity (14). However, the enzyme activity of purified KfiC has not been determined yet.
In this study, we expressed and purified both KfiA and KfiC recombinant proteins at a large scale using an E. coli expression system, and we characterized them and found that KfiA interacts with KfiC and activates the GlcA-T activity of KfiC. The complex of KfiA and KfiC has GlcNAc-T and GlcA-T activity and exhibits polymerase activity to synthesize the HPR polysaccharide. Our results provide mechanisms of K5 antigen biosynthesis and lead to bioengineering of HS chains.
The kfiA gene was cloned into the pET15b ϩ expression vector (Novagen) to yield the N-terminal His 6 -tagged protein. The resulting expression plasmids (pETHisKfiA) were transformed into E. coli BL21(DE3)pGro7 cells (Novagen). The transformants were cultured in 2ϫ YT medium containing 50 g/ml ampicillin and 20 g/ml chloramphenicol for 3 h at 37°C. The culture was cooled to 22°C and then supplemented with isopropyl ␤-D-thiogalactopyranoside (1 mM) and arabinose (1 mg/ml) for induction of KfiA-tagged protein and chaperone, respectively. After 14 h of culture, the cells were harvested by centrifugation and lysed by sonication in 50 mM Tris-HCl, pH 7.4, and 0.5 M NaCl (buffer A).
The kfiC gene was cloned into the pENTR11 (Invitrogen), and the resulting plasmid was named pENTR-KfiC. To construct the E. coli expression plasmid pColdDEST, the Gateway DEST cassette was amplified from pDEST38 (Invitrogen) by PCR with the primers 5Ј-CCGATGGTGATGACCTGACC-GTCGGG-3Ј and 5Ј-ACAAGTTTGTACAAAAAAGCTG-3Ј and digested with XbaI. The resulting fragment was introduced into a Ecl36II-XbaI site of pColdII (Takara Bio). The E. coli trigger factor (TF) sequence was amplified from DH10B cells (Invitrogen) by PCR with the primers 5Ј-GGAATTCCATAT-GCAAGTTTCAGTTGAAAC-3Ј and 5Ј-GGAATTCCATAT-GCGCCTGCTGGTTCATCA-3Ј, followed by NdeI digestion, and cloned into an NdeI site of pColdDEST. The resulting plasmid was named pColdTFDEST. Then the construct pColdTFDEST was subjected to the Gateway LR reaction with entry vector, pENTR-KfiC, using LR Clonase TM enzyme mixture (Invitrogen) according to the protocols recommended in the manufacturer's manual. The resulting expression plasmid named pColdTFKfiCDEST was transformed into E. coli BL21(DE3) Codonplus RIL cells (Stratagene) to yield the N-terminal His-tagged TF fusion protein.
The transformants were cultured in LB medium containing 100 g/ml ampicillin for 8 h at 37°C. The culture was cooled to 15°C, and then supplemented with isopropyl ␤-D-thiogalactopyranoside (0.5 mM) for induction of KfiC fusion protein and chaperone, prospectively. After 14 h of culture, the cells were harvested by centrifugation and lysed by sonication in buffer A.
The cell lysate containing His-tagged protein was applied to nickel-nitrilotriacetic acid-agarose gel, and His-tagged protein was eluted with buffer A containing 250 mM imidazole. The eluted proteins were dialyzed against 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 2 mM 2-mercaptoethanol.
Preparation of HPR Oligosaccharides-HPR oligosaccharides were prepared as described previously (16), with a slight modification. Briefly, HPR polysaccharide (10 mg) was digested with heparitinase I (10 milliunits) in 0.1 M sodium acetate, pH 7, containing 5 mM calcium acetate at 30°C for 16 h, and the digest was heated in boiling water for 5 min. After centrifugation, the even-numbered oligosaccharides containing unsaturated GlcUA at the nonreducing termini were desalted with a Superdex 30 HiLoad 16/60 column and then treated with 70 mM mercury(II) acetate solution, pH 5.0, at room temperature for 10 min to remove the unsaturated GlcUA residue. The oddnumbered oligosaccharides containing GlcNAc residue at the nonreducing termini were directly passed through a column (10-ml bed volume) of Dowex 50W-X8 (H ϩ form) to remove mercury ions and then separated by gel filtration chromatography with a Superdex 30 HiLoad 16/60 column. Some fractions contained HPR heptasaccharide (HPR-7). The structures of the oligosaccharides were determined by MALDI-TOF mass spectrometry and electrospray ionization-mass spectrometry analyses (17). The amount of HPR oligo-and polysaccharides was determined by carbazole assay method with GlcUA monosaccharide as standard (18).
Treatment of Nonreducing Terminus of HPR Polysaccharide-HPR polysaccharide (7 mg) was treated by ␤-glucosidase (5.0 units) in 0.1 M NaOAc, pH 5.2, containing 0.15 M NaCl (500 l) at 37°C for 18 h. The reaction mixture was heated in boiling water for 3 min and chilled to room temperature. The polysaccharide-containing GlcNAc residue at the nonreducing terminus (GlcNAc-HPR, 5.3 mg) was separated by gel filtration chromatography on a Superdex 6 HR10/30 column.
Measurement of Enzyme Activities Using Radioisotope Donor Substrates-GlcNAc-T, GlcA-T, and HPR polymerase activities of the recombinant enzymes were measured using radioisotope donor substrates. For the standard assay of GlcNAc-T, a 50-l mixture containing 50 mM BisTris-HCl, pH 6.5, 2 mM MnCl 2 , UDP-[ 3 H]GlcNAc (0.1 Ci, 300 pmol) as the donor substrate, and HPR polysaccharide (300 pmol) as the acceptor substrate was incubated with the recombinant enzyme (60 pmol) at 30°C for 60 min. The final concentration of NaCl to be responsible for the enzyme solution was ϳ5 mM. For the standard assay of GlcA-T, a reaction mixture of UDP-[ 14 C]GlcUA (0.1 Ci, 300 pmol) and GlcNAc-HPR (300 pmol) in 50 mM BisTris-HCl, pH 6.5, 2 mM MnCl 2 was incubated with the recombinant enzymes (either 60 pmol) at 30°C for 60 min. The final concentration of NaCl to be responsible for the enzyme solutions was ϳ10 mM. For the standard assay of the polymerase, a reaction mixture of HPR-7 (100 pmol) as the acceptor substrate and a pair of UDP-[ 14 C]GlcUA (0.1 Ci, 3.3 nmol) and UDP-GlcNAc (3.3 nmol) or that of UDP-[ 3 H]GlcNAc (0.1 C, 3.3 nmol) and UDP-GlcUA (3.3 nmol) as the donor substrates in 50 mM BisTris-HCl, pH 6.5, 2 mM MnCl 2 were incubated with the recombinant enzymes (either at 60 pmol) at 30°C for 18 h. The final concentration of NaCl to be responsible for the enzyme solutions was ϳ10 mM. After each reaction mixture was heated in boiling water for 1 min, the radiolabeled product was separated by a Superdex Peptide HR10/30 column and measured by liquid scintillation counting. The enzyme activities were determined by calculating the amount of the incorporated radioactive sugars.
SDS-PAGE and Western Blotting-SDS-PAGE was carried out on a 10 and 15% polyacrylamide gel by the method of Laemmli (19). Proteins were detected by Coomassie Brilliant Blue staining. For Western blotting, proteins on the SDS-polyacrylamide gel were transferred onto a polyvinylidene difluoride transfer membrane (Immobilon-P, Millipore). The membrane was treated with 50 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 0.05% Tween 20 (TBS-T) containing 5% skim milk for 1 h and then with an anti-His antibody. After washing with TBS-T, the membrane was treated with a horseradish peroxidase-conjugated secondary antibody, and the signal was detected using the enhanced chemiluminescence detection system (Western Lightning Plus, PerkinElmer Life Sciences) and a lumino-image analyzer (LAS-4000 mini, Fujifilm). Protein content was determined using a micro BCA protein assay kit (Pierce) with bovine serum albumin as standard.
Measurements of Enzyme Activities of KfiA and KfiC- Table  1 shows the summary of enzyme activities of the recombinant K5 proteins. His-KfiA exhibited GlcNAc-T activity, using HPR polysaccharide as acceptor and UDP-[ 3 H]GlcNAc as donor, consistent with the previous report (14). In contrast, TF-KfiC showed no GlcA-T activity, using HPR polysaccharide acceptor and UDP-[ 14 C]GlcUA donor. His-KfiA exhibited no GlcA-T activity, and TF-KfiC had no GlcNAc-T activity (data not shown). Interestingly, a mixture of TF-KfiC and His-KfiA showed certain levels of [ 14 C]GlcUA incorporation into HPR polysaccharide, which were elevated when reacted for 18 h with both UDP-[ 14 C]GlcUA and UDP-GlcNAc as donors, suggesting chain polymerization.
Next, we examined the GlcA-T activity of the enzyme mixture, using GlcNAc-HPR prepared with glucuronidase as acceptor substrate. His-KfiA did not transfer GlcNAc to the acceptor substrate, confirming the absence of GlcUA at the nonreducing terminus of HPR. TF-KfiC alone did not show GlcA-T activity, even using GlcNAc-HPR as acceptor sub-strate. In contrast, the mixture of TF-KfiC and His-KfiA transferred [ 14 C]GlcUA onto GlcNAc-HPR, exhibiting GlcA-T activity. The activity (8.41 Ϯ 0.50 pmol of GlcUA/min/pmol of each enzyme) was comparable or even higher than that of GlcNAc-T of KfiA (7.54 Ϯ 0.37 pmol of GlcNAc/min/pmol of enzyme). However, when native HPR polymer was used as acceptor, the enzyme mixture showed lower GlcA-T activity (1.46 Ϯ 0.24 pmol of GlcUA/min/pmol of enzyme), suggesting that biosynthesis of HPR polysaccharide by E. coli strain K5 is prone to terminate when GlcUA residue was transferred at the nonreducing end.
When both UDP-[ 3 H]GlcNAc and UDP-GlcUA were used as donor substrates, the enzyme mixture certainly incorporated Assay Conditions for GlcNAc-T and GlcA-T Activities-We measured GlcA-T activity of the enzyme mixture at different ratios, adjusting the final NaCl concentration to 50 mM ( Fig.  2A). GlcA-T activity increased as the ratio of KfiA/KfiC increased. Addition of KfiA at a 2-fold molar ratio to that of KfiC showed ϳ1.5-fold GlcA-T activity compared with an equimolar ratio of KfiA and KfiC. The activity of a 5-fold addition of KfiA was similar to that of the 2-fold addition. KfiA and KfiC alone showed no GlcA-T activity. Addition of KfiC had no effect on the GlcNAc-T activity of KfiA (data not shown).
We then examined the effects of NaCl concentrations on GlcNAc-T and GlcA-T activities of His-KfiA and the mixture of TF-KfiC and His-KfiA (Fig. 2B). A maximal activity of both transferases was obtained in the reaction buffer containing 50 mM BisTris-HCl, pH 6.5, 2 mM MnCl 2 and the lowest concentration of NaCl (5-10 mM). Although GlcNAc-T activity did not change at different NaCl concentrations up to 205 mM, GlcA-T activity was significantly attenuated by increasing NaCl concentrations. The GlcA-T activity was ϳ50 and ϳ25% at 60 and 160 mM NaCl, respectively, compared with that under the lowest NaCl concentration (10 mM). Addition of KCl in place of NaCl showed similar effects (data not shown). The GlcNAc-T (Fig. 2C) and GlcA-T (Fig. 2D) activities were optimal at pH 7.0 and 6.5, respectively. pH conditions examined in a range of pH 5.5-8.5 affected GlcNAc-T activity little, but it did affect GlcA-T activity. The relative activity of GlcNAc-T was 90% optimal at pH 6.5 for GlcA-T in BisTris-HCl buffer, whereas that of GlcA-T was ϳ66% optimal at pH 7.0 for GlcNAc-T in Tris-HCl buffer. The relative activity of GlcNAc-T and GlcA-T was 90 and 66% at pH 5.5 in BisTris-HCl buffer and 88 and 14% at pH 8.5 in Tris-HCl buffer, respectively, compared with that under the optimal conditions. Thus, all the following enzymatic reactions were carried out in 50 mM BisTris-HCl buffer, pH 6.5.
Next, we examined the effects of divalent metal ions on the GlcNAc-T and GlcA-T activities of His-KfiA and the mixture of TF-KfiC and His-KfiA (Table 2). Maximal activity was obtained in the presence of Mn 2ϩ ion (2 mM) for both transferase activities. The presence of Co 2ϩ ion showed 56 and 3% that of GlcNAc-T and GlcA-T activity. Mg 2ϩ and Fe 2ϩ showed 10 -20% that of both activities, as compared with that of Mn 2ϩ ion. Ca 2ϩ . Ba 2ϩ , Cd 2ϩ , Cu 2ϩ , and EDTA showed little activity (0 ϳ 3%). We then examined the effects of Mn 2ϩ concentrations on GlcNAc-T and GlcA-T activities of His-KfiA and the mixture of TF-KfiC and His-KfiA (Table 3). A maximal activity was obtained at 2 mM MnCl 2 in a 60-min reaction for both GlcNAc-T and GlcA-T. Both activities were attenuated by decreasing MnCl 2 concentrations. Although GlcA-T activity at 20 mM MnCl 2 decreased to ϳ67% that at 2 mM MnCl 2 , GlcNAc-T activity was affected only a little.
Enzyme Activities of the Point Mutants of KfiA and KfiC-The fact that the enzyme mixture of KfiA and KfiC indicated GlcA-T activity implies that KfiC exhibits GlcA-T activity in the presence of KfiA or that KfiA acquires GlcA-T activity in the presence of KfiC. To determine which enzyme transfers GlcUA, we measured activity using the enzymes with point mutations.
The DXD motif that binds the UDP-sugar donor substrate with Mn 2ϩ ion is essential for glycosyltransferase activity. Both KfiA and KfiC have a DXD motif in the glycosyltransferase active site. We constructed point mutants replacing an aspartic acid residue with an alanine residue at position 79 DDD of KfiA and 352 DAD of KfiC to 79 ADD and 352 AAD, respectively. The recombinant proteins, His-KfiA D79A and TF-KfiC D352A, were purified by nickel affinity chromatography and dialyzed.
His-KfiA D79A point mutant showed no GlcNAc-T activity (Table 4 and Fig. 3A). The mixture of intact TF-KfiC and His-KfiA D79A represented the GlcA-T activity comparable with a mixture of intact enzymes. In contrast, TF-KfiC D352A mutant showed no GlcA-T activity with intact or mutant His-KfiA.

TABLE 2 Effects of divalent cations and chelate reagent on the GlcUA transferase activity of KfiA and HPR polymerase activity of KfiC and KfiA mixture
The enzyme activities were measured as described under "Experimental Procedures." The reaction mixture (50 l) containing 2 mM divalent metal ion (Mn 2ϩ , Mg 2ϩ , Ca 2ϩ , Ba 2ϩ , Cd 2ϩ , Co 2ϩ , Cu 2ϩ , or Fe 2ϩ ) or 20 mM EDTA substituted for the metal ion was incubated at 30°C for 60 min with the enzymes, acceptor, and donor substrates. The data in the table are the relative percentages of the radioactivity incorporated in comparison with the assay containing Mn 2ϩ ion (100%).

Metal ion or reagent
Relative activity These results clearly demonstrate that KfiC has the GlcA-T activity in the enzyme mixture. A mixture of either mutant or both mutants at the DXD motifs of KfiA and KfiC did not show polymerase activity when HPR-7 acceptor, radiolabeled UDP-GlcNAc, and nonlabeled UDP-GlcNAc were used as donor substrates. This indicates that the polymerization occurs by alternate transfer of GlcNAc by KfiA and GlcUA by KfiC.
Enzyme Activities of the Truncated Mutants of KfiA and KfiC-To determine regions responsible for enzymatic activity, we used various truncated mutants. A truncated mutant of KfiA that lacked the C-terminal 28 amino acid residues (His-KfiA ⌬211-238) exhibited no GlcNAc-T activity (Fig. 3A). The C-terminal truncated mutant of KfiA did not activate GlcA-T activity of KfiC (Fig. 3B). The C-terminal sequence may be essential for both activities. Two synthetic peptides consisting of 25 amino acid residues at the C-terminal region of KfiA (KKWPLDIIKETQAIAGYSKLNLELV and LDI-IKETQAIAGYSKLNLELVYNVG) were added to the GlcA-T assay system with or without KfiA. The peptides showed neither inhibition nor activation effects on GlcA-T up to 1 mM (data not shown).
N-terminal truncated mutants of KfiC were used to measure GlcA-T activity with intact His-KfiA (Fig. 3C). TF-KfiC ⌬8 -14, TF-KfiC ⌬8 -39, and TF-KfiC ⌬8 -49 indicated ϳ1.2, 0.6, and 2.3 times the GlcA-T activity of intact TF-KfiC, respectively. In comparison, further truncated mutants TF-KfiC ⌬8 -75 showed a little GlcA-T activity (ϳ3%), and TF-KfiC ⌬8 -150, and TF-KfiC ⌬8 -225 had practically no activity of GlcA-T. The H]GlcNAc radioactivity incorporation to HPR and that of the reaction containing the wild type His-KfiA (100%). B, GlcA-T activity of the wild type TF-KfiC with the mutants of KfiA was measured. The data are shown as a percent of the [ 14 C]GlcUA radioactivity incorporation to GlcNAc-HPR and that of the reaction containing wild type of His-KfiA and TF-KfiC (100%). C, GlcA-T activity of the mutants of TF-KfiC with the wild type His-KfiA was measured. The data are shown as a percent of the radioactivity incorporation to GlcNAc-HPR and that of the reaction containing wild type of TF-KfiC and His-KfiA (100%).

TABLE 3 GlcNAc and GlcUA transferase activity of KfiA and equimolar mixture of KfiC and KfiA at various concentrations of Mn 2؉ ions
The enzyme activities were measured as described under "Experimental Procedures." The reaction mixtures containing 0 -20 mM MnCl 2 was incubated at 30°C for 60 min. The data in the table are the relative percentages of the radioactivity incorporated in comparison with the assay containing 2.0 mM MnCl 2 (100%).  single reaction of all KfiC mutants that we examined showed no GlcA-T activity without KfiA (data not shown), same as intact KfiC.
Polymerization Reaction of KfiA and KfiC-We further investigated the polymerase activity of the enzyme mixture, using HPR-7 (GlcNAc-(GlcUA-GlcNAc) 3 ) oligosaccharide as an acceptor substrate and UDP-GlcUA and UDP-GlcNAc as donor substrates, either of which was radioisotope-labeled. An equimolar enzyme mixture of TF-KfiC and His-KfiA produced ϳ10 and 20 kDa of [ 3 H]GlcNAc-or [ 14 C]GlcUA-incorporated polymers in 8-and 18-h reactions (Fig. 4). Heparitinase I digested the incorporated products completely. When these digests were analyzed by high pressure liquid chromatography for glycosaminoglycan disaccharide component assay, only unsaturated GlcUA-GlcNAc disaccharide was detected (data not shown). Chondroitinase ABC, Streptomyces hyaluronidase, and sheep testicular hyaluronidase did not digest the polysaccharide products (data not shown). These results indicate that both products were HPR polysaccharide chains.

DISCUSSION
The K5 capsular polysaccharide synthesized by E. coli strain K5 consists of disaccharide repeating units of (-4 GclA ␤1-4 GlcNAc ␣1-), a backbone of HS. A gene cluster responsible for its synthesis encodes the following four proteins: KfiA, KfiB, KfiC, and KfiD. Of them, KfiA and KfiC were supposed to be GlcNAc-T and GlcA-T, respectively. Whereas recombinant KfiA was expressed, purified, and identified to be GlcNAc-T (14), the enzyme activity of purified KfiC had yet to be determined. Here, we have successfully expressed KfiC as a fusion protein with TF for the first time, and we have shown that it exerts GlcA-T activity in the presence of KfiA. Furthermore, we have demonstrated that both KfiA and KfiC form a complex, which exhibits polymerase activity to generate HPR chains. Our results provide mechanisms underlying biosynthesis of HPR in E. coli strain K5 and would lead to bioengineering of HS chains.
Although KfiC was speculated to have GlcA-T activity, the recombinant fusion protein TF-KfiC, by itself, had no GlcA-T activity. Addition of KfiA to the TF-KfiC reaction attained GlcA-T activity. Analysis using various mutants of DXD motifs confirmed the sequence of 79 DDD of KfiA as responsible for GlcNAc-T activity and that of 352 DAD of KfiC as responsible for GlcA-T activity in the crude cell lysate assays (12,13). Whereas a mixture of intact KfiA with D352A KfiC mutant did not show GlcA-T activity, that of intact KfiC and D79A KfiA mutant did, indicating the GlcNAc-T catalytic site of KfiA is not necessary for the GlcA-T activity of KfiC. By interacting with KfiA, KfiC may change its conformation and acquire GlcA-T activity. In the presence of both UDP-GlcUA and UDP-GlcNAc as donors, the enzyme mixture efficiently polymerized HPR chains onto HPR oligo-and polysaccharides, suggesting that these enzymes are present in a complex.
His-KfiA ⌬211-238 truncated mutant that lacks the C-terminal 28 amino acids had neither its own GlcNAc-T nor KfiC activation activities; however, the peptides at the C-terminal region indicated neither activation nor inhibition effects on GlcA-T. The interaction site of KfiA with KfiC is not limited to the C-terminal region, although it must be important for both GlcNAc-T and GlcA-T activities.
Four positions in the sequence of KfiA containing ionic amino acid clusters that may be concerned in the interaction were selected. The point mutants at N-terminal sites (K13A/ K14A/E15A and E38A/E39A) exhibited GlcNAc-T activity as well as intact KfiA, although the mutants at C-terminal sites (E188A/K189A and K209A/K210A) had lower activity of GlcNAc-T than intact KfiA. These results suggested that the C-terminal site is responsible for the transferase activity of KfiA. In regard to activation activity for GlcA-T, the activity of three mutants (K13A/K14A/K15A, E38A/E39A, and E188A/ K189A) was increased 1.5-2.2 times, and the activity of the point mutant replaced at the most C-terminal site (K209A/ K210A) was decreased slightly. These positions were suggested not to be very important for demonstrating the activation of GlcA-T activity.
The shorter truncated KfiC mutants (⌬8 -14, ⌬8 -39, and ⌬8 -49) showed GlcA-T activity with intact KfiA; however, the longer truncated mutants (⌬8 -75, ⌬8 -150, and ⌬8 -225) hardly had any activity. This strongly suggested that the amino acid sequence between 49 and 75 is essential for GlcA-T activity. However, it is not yet known whether the sequence in KfiC is necessary for its own GlcA-T activity or for the interaction with KfiA.
The lack of enzymatic activity of KfiC may be due to the presence of an inhibitory domain within the molecule, and the interaction of KfiC with KfiA may cancel out the action of such an inhibitory domain. However, all truncated mutants of KfiC that we examined represented no GlcA-T activity without the aid of KfiA. There is little possibility of this hypothesis.
GlcA-T activity shown by the interaction of KfiA and KfiC was reduced by increasing NaCl concentrations and abrogated at physiological concentrations. These observations indicate that KfiA and KfiC have a tendency to interact weakly with each other and that their stable association under physiological conditions requires other molecules. Previous studies revealed that KfiA, KfiB, and KfiC were localized in the cytoplasmic membrane and that KfiB is required for the association of KfiC with the membrane (13). Other K5 capsular molecules (KpsC, KpsD, KpsE, KpsF, KpsM, KpsS, KpsT, and KpsU) involved in the synthesis and attachment of phosphatidyl-KDO and exportation of the polysaccharide chains across the plasma membrane are known to be required for association of the K5 synthetic proteins (KfiA-D) with the membrane (7,20). They may be involved in the stabilization and localization of the enzyme complex.
GlcA-T activity of the enzyme mixture at different ratios increased with additional KfiA up to a 1:1 molar ratio of KfiA and KfiC. It also represented an efficient HPR polymerization reaction of alternating GlcUA and GlcNAc transfer to HPR oligo-and polysaccharides. These results suggest that the two enzymes interact at an equimolar ratio. In contrast, reactions with different amounts of KfiC showed a similar level of GlcNAc-T activity of KfiA, consistent with the previous report (14).
Although the GlcNAc-T activity of KfiA was little affected by various pH and salt concentrations, the GlcA-T activity of the complex was largely affected. The presence of Mn 2ϩ ion (2 mM) in the reaction buffer showed the highest activity of the GlcNAc-T and GlcA-T of KfiA and KfiC, like other glycosyltransferases (21). A higher concentration (20 mM) of MnCl 2 reduced the GlcA-T activity but not the GlcNAc-T activity, presumably due in part to instability of UDP-GlcUA at high MnCl 2 concentrations (22). The effects of various divalent metal ions on the enzyme activity were different between GlcNAc-T and GlcA-T, which may be due to differences of glycosylation reaction (␣1-4, retaining reaction, or ␤1-4, inverting reaction) and substrate structure (GlcNAc or GlcUA) (23).
Many elongation enzymes for glycosaminoglycan biosynthesis are bifunctional glycosyltransferases. For example, animal species have EXT-1 and -2 for HS chain elongation (3), CSS-1/ ChSy-1, CSS-2/ChPF, and CSS-3/ChSy-2 for chondroitin chain elongation (24,25), and HAS-1, -2, and -3 for hyaluronan synthesis (26). Many types of bacteria also have dual action enzymes (27), such as HAS-A (Streptococcus sp.), pmHS1, -2, pmCS, and pmHAS (Pasteurella multocida), and K4CP (KfoC, E. coli strain K4) (28). The majority of the enzymes consist of two independent glycosyltransferase sites containing DXD motifs, except for HAS-1, -2, and -3 and HAS-A (class 1 hyaluronan synthase) (29). The crystal structure of K4CP revealed two distant active sites (30). PmHS-1 and -2 contain two glycosyltransferase sites at the N-and C-terminal regions similar to the active sites of KfiC and KfiA (31), and they have dual transfer activity to synthesize the HPR polymer (32). These dual action enzymes might be evolutionally derived by gene fusion from a pair of genes encoding single-action glycosyltransferases, gaining efficient polymerizing activity of heteropolysaccharides. It is intriguing that KfoC in the gene cluster of E. coli strain K4 contains both GlcA-T and N-acetylgalactosaminyltransferase sites and polymerizes a backbone of chondroitin chains itself (15), whereas in strain K5, the complex of KfiA and KfiC polymerizes HPR. Further studies remain to be performed to understand the molecular basis for their distinct catalytic mechanisms.
We revealed that HPR polymer synthesized by E. coli strain K5 serves as a good substrate for GlcNAc-T assay but not for GlcA-T assay. This was constantly observed in five different HPR samples purified by individual lots from E. coli K5 culture. Both GlcNAc-HPR substrates, prepared by removing a GlcUA residue at the nonreducing terminus of HPR with glucuronidase and by adding a GlcNAc residue at the nonreducing terminus of HPR with KfiA, were useful for GlcA-T analysis.
Our results clearly demonstrate that the complex of KfiA and KfiC is the core machinery of HPR biosynthesis. The threedimensional structure of these glycosyltransferases and their complex remains to be determined. Further characterization of the enzyme complex containing other molecules such as KfiB would reveal detailed mechanisms of HPR polymerization. This study provides a useful tool to generate expected chain length of HPR, and it leads to preparation of different HS chains in length and structure by modification with HS sulfotransferases and HS C5-epimerase. Such defined HS chains would enable a detailed analysis on functions of HS and, in the future, could be used for clinical applications.