l-Fucose for mammalian glycosylation contains an ano- meric carbon atom generating α- and β-l-fucoses. Based on sequence comparison of mouse and human homologs with the prokaryotic fucose mutarotases (FucU) characterized previously, we investigated their function in mammalian cells. By nuclear magnetic resonance (NMR) measurement with saturation difference analysis, the purified mammalian mutarotases were demonstrated to be involved in an interconversion between the two anomeric forms with comparable efficiency as that of the Escherichia coli FucU. The mouse gene was widely expressed in various tissues and cell lines, including kidney, liver, and pancreas, although expression was marginal in muscle and testis. By generating stably expressed cell lines for mutarotase genes in HepG2, it was shown that fucose incorporations into cellular proteins were increased as demonstrated by an incorporation of radiolabeled fucose into the cells. Furthermore, intracellular levels of GDP-l-fucose, measured with high performance liquid chromatography (HPLC), were enhanced by an overproduction of cellular mutarotase, which was reversed by gene silencing of mutarotase based on RNA interference. The results suggest that the mammalian mutarotase is functional in facilitated incorporation of fucose through the salvage pathway.
l-Fucose (6-deoxy-l-galactose) exists in two different forms, α-l-fucose (29.5%) and β-l-fucose (70.5%), among which the β-form is metabolized through the salvage pathway (Ishihara et al. 1968; Ryu, Kim, Park, et al. 2004). This pathway involves the synthesis of β-l-fucose-1-phosphate by l-fucokinase (EC 220.127.116.11, (Ishihara et al. 1968)) and the formation of GDP-l-fucose by GDP-l-fucose pyrophosphorylase (EC 18.104.22.168, (Ishihara and Heath 1968)). The salvage pathway utilizes free fucose obtained from extracellular sources or from intracellular degradation of glycoproteins and glycolipids. On the other hand, the de novo pathway converts GDP-d-mannose into GDP-l-fucose with two enzymes, GDP-mannose 4,6-dehydratase (GMD) and GDP-keto-6-deoxymannose 3,5-epimerase, 4-reductase, also known as the FX protein (Tonetti et al. 1996). GDP-l-fucose formed either by the de novo or salvage pathways is then transported into endoplasmic reticulum (ER), in which it serves as a substrate for glycosylation involved with 13 different fucosyltransferases (Becker and Lowe 2003). l-fucose is a common component of several N- and O-linked glycans and glycolipids present in mammalian cells (McKibbin 1978; Miyoshi et al. 1999; Wang et al. 2001; Smith et al. 2002). Fucosylated structures expressed on cell surfaces or secreted in biological fluids are believed to play a critical role in cell–cell adhesion and recognition processes (Carlow et al. 2001; Homeister et al. 2001; Smith et al. 2002).
Mutarotases are recently characterized group of enzymes that are involved in the anomeric conversions of monosaccharides (Ryu, Kim, Kim, et al. 2004). Although the anomers are subject to interconversion in aqueous environment, the existence of mutarotase activity seems to be justified based on the fact that spontaneous α-to-β conversion of a pyranose (six-membered) ring is not rapid enough (0.015 min−1 for glucose) to support fast-energy generation by an organism (Ryu, Kim, Kim, et al. 2004). The enzyme-catalyzed anomeric conversion may facilitate sugar utilization and incorporation into protein by glycosylation. It was reported using polarimeter that the activities of mutarotases for d-glucose, d-galactose, d-xylose, l-arabinose, and d-fucose were detected in extracts of rat kidney and intestine, which were altered during development (Bailey et al. 1969, 1970). However, their genetic identities have been unknown. Galactose mutarotase (aldose 1-epimerase) that catalyzes the interconversion of α- and β-anomers of hexose such as glucose and galactose was first reported in Escherichia coli (E. coli) (Wallenfels et al. 1965). The human galactose mutarotase was recently characterized based on amino acid sequence similarity to the prokaryotic one to have kcat/Km (M−1 s−1) for galactose and glucose as 340,000 ± 56,000 and 90,000 ± 12,000, respectively as estimated by polarimetry (Timson and Reece 2003), which is similar to that of E. coli enzyme. The importance of the galactose mutarotase (GalM) for d-galactose and d-glucose in the Leloir pathway has also been implicated (Beebe and Frey 1998).
Previously, we characterized three E. coli proteins (RbsD, FucU, and YiiL) as mutarotases for d-ribose, l-fucose, and l-rhamnose, respectively (Ryu, Kim, Park, et al. 2004; Ryu, Kim, Kim, et al. 2004; Ryu et al. 2005). In this study, we report that the mammalian utilization of fucose in the salvage pathway involves fucose mutarotase catalyzing α-to-β anomeric conversions, prior to phosphorylation of the sugar by fucokinase. The functionality of the mammalian fucose mutarotase was demonstrated with nuclear magnetic resonance (NMR) for purified proteins and in vivo by an incorporation of fucose into cellular proteins via GDP-l-fucose, an intracellular glycosylation substrate.
Prediction and expressions of mammalian fucose mutarotase gene
Among three mutarotases of E. coli we have previously characterized (Ryu, Kim, Kim, et al. 2004), only the fucose mutarotase (FucU) has a homolog in mammals. The putative mammalian FucU genes are located on 7qF4 (LOC69064) and 10q26.3 (LOC282969) of the mouse and human chromosomes, respectively (Figure 1A, http://genome.ucsc.edu). Several transcripts of mouse and human FucUs, generated by alternative splicing, have been suspected from cDNA sequences, which are predicted to be translated into variable sizes. However, only the 17 kDa protein from BC028662 was detected in tissue samples by immunoprecipitation (Figure 1B), which is originated from the mouse FucU gene (LOC69064) with six exons included after splicing. The equivalent human transcript AI471699, encoding the exon 1 through 6, was detected from the human hepatoma cell line HepG2 and liver library (data not shown). The human FucU gene also generates the mRNA (BM009909) lacking the exon 2, from which the protein was not expressed perhaps from its instability (not shown). The deduced mouse and human FucU proteins contain 153 and 154 amino acids, respectively, which have 52–54% similarity to that of E. coli (Figure 1C). The His20 residue of E. coli RbsD known as a catalytic residue (Ryu, Kim, Kim, et al. 2004) is also conserved in the mammalian FucUs.
The levels of mouse FucU mRNA and protein expressions were higher in kidney and small intestine than in other tissues (Figure 1B). Endogenous FucU transcripts were also detected in various cell lines including HepG2, COS7, HEK293T, etc. (data not shown). Using antisera raised against the purified mouse FucU protein from E. coli, we detected endogenous levels of FucU protein from various tissues, which were shown as immunoblot after immunoprecipitation of mouse tissues with the same anti-FucU antisera (Figure 1B). With 5 mg of cytoplasmic proteins, FucU was hardly visible from immunoblotting, suggesting that the amount of FucU protein is lower than 0.17 ng/mg of total protein. From this result, we estimated the copy number of FucU protein in highly expressed cells as less than approximately 1.85 × 103 molecules/cell (Rudolph et al. 1999). The relative expressions of mutarotases in various tissues are generally correlated with the levels of transcripts, except for muscle (Figure 1B), that were based on densitometric tracings of the mRNA and protein levels shown on the lower panel.
Protein expression and enzymatic characterization of mammalian fucose mutarotase
We used the 1-dimensional (1-D) saturation difference (SD) NMR technique for analyzing the α-to-β anomeric conversions with substrates in equilibrium between α- and β-anomers (Ryu, Kim, Park, et al. 2004; Ryu, Kim, Kim, et al. 2004). When the α-to-β conversion of sugar is considerably slow, subtracting unsaturated one from the spectrum obtained by a saturation of H1′-peak of α-anomer exhibits only the saturated peak. However, the increase in α-to-β conversion by FucU allowed us to observe not only the saturated peak but also the H1′-peak from β-anomer that was originated by a chemical exchange in α-anomer. An exchange rate of α-to-β conversion of l-fucose in the absence of FucU was extremely slow, so that it was not detected in SD experiment (Figure 2B). The results showed that the mouse and human FucU proteins enhance the exchange rate between the two anomers (Figure 2), although their activities are slightly lower than that of E. coli FucU. The reduction in the activity of mammalian FucU homologs might be due to the property of purified proteins, because the proteins aggregated more often than E. coli FucU during the purification. The aggregation might be due to the C-terminal His-tag attached to the proteins. The mammalian FucU was also observed to form an oligomer, presumably a pentamer as does the E. coli FucU, although the efficiency of oligomerization was lower than 100% (data not shown). Since the fucokinase is known to have a specificity toward β-l-fucose (Butler and Serif 1985), it is likely that intracellular efficiency of the enzyme might be enhanced by the presence of fucose mutarotase.
Physiological effects of the mutarotase expression
In order to generate stably expressed FucU cell line, the mouse and human FucU genes were cloned into pcDNA3.1 plasmid and transfected into HepG2 cells. After G-418 selection, two stable clones were obtained and maintained for nearly 6 months. As a knock-down control, HepG2 transfected with the human FucU RNAi clone was generated (Materials and methods). Western blot analyses of extracts from the stable cells and the knock-down cells (after immunoprecipitation) showed protein bands at 17 kDa position on SDS-PAGE (Figure 3 inset). The latter expressed approximately 67% of the enzyme contained in HepG2, while the overproduced cell lines expressed about more than a hundredfold of HepG2. The cell lines were incubated with radiolabeled l-[5,6-3H]fucose, from which total cellular or secreted proteins were obtained by centrifugation from the lysed cells or from the medium, respectively (Materials and methods). Total cellular incorporations of fucose, presumably including intracellular compounds derived from l-fucose as well as various glycoproteins showed little difference between the control and stable cells (data not shown). However, intracellular (Figure 3, upper panel) and secreted proteins (Figure 3, lower panel) fractionated from the fucose-incorporated cells exhibit considerably enhanced incorporations in the overproduced cell lines. At 72 h of incubation with radiolabeled fucose, the HepG2 and RNAi cells incorporated 25 and 24% of the total radioactivities into intracellular proteins respectively, while 31–33% was found in mFucU and hFucU cells. The remaining radioactivities reside in either protein-free fractions, 69–71% of HepG2/ RNAi and 55–56% of mFucU/hFucU cells, or in secreted proteins, 5–6% (HepG2/RNAi) and 12–13% (mFucU/hFucU).
In order to assess whether anomeric conversion by FucU relates directly to an in vivo production of GDP-l-fucose, we measured metabolites generated from fucose incorporation via mutarotation. After preexposal for at least 2 weeks in Dulbecco's modified Eagle's medium (DMEM) with 10 μM fucose, cells expressing mutarotase were incubated with 1 mM fucose for 2, 10, and 30 min prior to lysis (Materials and methods). The soluble fractions of cells were passed through the column to separate nucleotide sugars, which were then treated with alkaline phosphatase. The enzyme removes the phosphate groups from nucleotides but retains the nucleotide diphosphate sugars intact. The reaction mixtures were applied onto diethyl aminoethyl (DEAE)-sepharose column and analyzed with ion-pair reversed phase high performance liquid chromatography (HPLC). The results show well-separated peaks of intracellular nucleotide sugars (Figure 4), from which the GDP-l-fucose peak was identified either by spiking with the commercially available compound (not shown) or by comparing its retention time (∼28.8 min) with that of the standard (Figure 4, HepG2). In contrast to an undetectable level of GDP-l-fucose (at 2 min of RNAi), the stable cell lines exhibit fairly high levels of GDP-l-fucose 29.97 ± 2.95 (mFucU) and 21.69 ± 1.82 (hFucU) μmole/mg of protein, while HepG2 shows an intermediate level (15.16 ± 1.08 μmole/mg). Furthermore, increases in GDP-l-fucose of the stable cells were more pronounced after 10–30 min of incubation, i.e., 15.04 ± 1.01 to 17.16± 0.38 μmole/mg (HepG2) and 5.24 ± 1.02 to 5.55 ± 1.10 μmole/mg (RNAi), 34.89 ± 0.12 to 41.32 ± 1.16 μmole/mg (mFucU), and 24.74 ±0.04 to 34.16 ± 1.09 μmole/mg (hFucU). The peak for GDP-d-mannose was also identified and designated (*) in Figure 3, which are apparently increased in the stable cells, 3.41 ± 0.93 μmole/mg in HepG2 versus 5.93 ± 1.53 (mFucU) and 6.89 ± 0.77 (hFucU).
The characterization of E. coli FucU has provided an insight into the biological function of the human FucU. Since the mammalian l-fucokinase is specific for β-l-fucose, the mutarotase activity is likely to be critical in regulating fucose metabolism. Quantitative studies of fucose metabolism in HeLa cells indicated that the de novo pathway accounts for the majority of cellular GDP–l-fucose production (Yurchenco and Atkinson 1977). However, several congenital glycosylation disorders related to fucose (Freeze 2001; Becker and Lowe 2003) was shown to be corrected by an exogenous supply of fucose (Marquardt et al. 1999). This suggests that the existence of salvage pathway was effective in complementing the fucosylation defects caused by the leukocyte adhesion deficiency type II (LAD II) disease (Marquardt et al. 1999) and by removing the FX enzyme (Smith et al. 2002). The functional study conducted here for the fucose mutarotase present in various tissues and cells adds further support to the role of salvage pathway in cellular incorporation of fucose, as well as in intracellular recycling through lysosome.
Although the 3-D structure of human FucU protein has not been reported, the sequence homology between bacterial RbsD and FucU proteins allows us to inspect amino acid residues of FucU involved in l-fucose binding, based on the coordinates of Bacillus subtilis RbsD (Kim et al. 2003). The amino acid residues of FucU that are homologous to the residues of RbsD involved in d-ribose binding are highly likely to be involved in the interaction with l-fucose, since overall structures of RbsD and FucU are expected to be similar. The conserved His20 residue of E. coli RbsD was presumed to be a part of the active site recognizing the O-5 position of ribopyranose, since the H20A mutant of E. coli RbsD completely abolished the pyranase activity (Ryu, Kim, Kim, et al. 2004). The mammalian FucU shares the sugar binding residues of H20, D28, Y120, and N122 from B. subtilis RbsD, forming hydrogen bonds with the OH-groups of d-ribopyranose as in l-fucose (Figures 1C and 5). However, the H98 and K102 residues of RbsD are altered in FucU to Arg and Tyr, respectively. The altered residues might play a critical role in anomeric conversion of l-fucose by associating with the OH group at C-1 position of the sugar, since the catalytic mechanisms of FucU (mutarotase) and RbsD (pyranase) are essentially different, i.e., the α-to-β conversion of FucU versus the β-pyran to β-furan conversion of RbsD. The Tyr residue of FucU may play an important role in recognizing the anomeric OH group at C-1 position, for the role of corresponding Tyr18 in l-rhamnose mutarotase was indicated as a crucial residue in mutarotation by forming a hydrogen bond with the OH group at C-1 position (2.8 Å to β-OH and 2.6 Å to α-OH) (Ryu et al. 2005).
The E. coli FucU has both pyranase and mutarotase activities (Ryu, Kim, Kim, et al. 2004), although the human enzyme retains only the mutarotase activity (data not shown). We speculate that the structures of RbsD/FucU family members could easily be adapted to include other monosaccharides as substrates. RbsD is believed to be an ancestor of FucU, as ribose would have been crucial in ancient, RNA environment. Thus, the ancestral FucU protein might then have retained the ancestral pyranase activity, while the specificity of mammalian enzyme became restricted to l-fucose. The presence of an RbsK homolog in human may suggest a possibility that an RbsD activity exists in human cells for efficient ribose utilization. However, we observed that the spontaneous pyran-to-furan conversion of d-ribose was considerably enhanced at 40°C, compared with the α-to-β conversion of l-fucose (Ryu, Kim, Park, et al. 2004), implying that the pyranase activity for d-ribose might not be required in human because the body temperature of human is constant at 36.5°C. The existence and specialization of human FucU toward l-fucose seems relevant in that the incorporation of l-fucose in eukaryote is obligatory.
In mammals, there are nine different nucleotide sugars, CMP-Neu5Ac, UDP-Gal, UDP-Glc, UDP-GlcNAc, UDP-GalNAc, UDP-Xyl, UDP-GlcA, GDP-d-mannose, and GDP-l-fucose, which are transported into the ER or Golgi for glycosylation (Chen et al. 2005). Except for UDP-N-acetyl-d-hexosamine (UDP-HexNAc), i.e., UDP-N-acetyl-d-galactosamine and UDP-N-acetyl-d-glucosamine, being eluted at the same retention time, the other nucleotide sugars were easily separated (Figure 4). CMP-Neu5Ac was missing in the HPLC data because of its early retention time as well as its small quantity. The UDP-galactose (peak #1 in Figure 4S), UDP-glucose (#2) and UDP-xylose (#3) are detected at small quantities in all samples, although the amounts were variable. In contrast, almost equal amounts of UDP-HexNAc were eluted at 11.6 min in all samples. The small peak found at 28.1 min, slightly ahead of GDP-l-fucose, was suspected as GDP-4-keto-6-deoxymannose (GDP-KDM), an intermediate from GDP-d-mannose to GDP-l-fucose, which were observed in bacteria, yeast, and mammalian cells expressing GDP-d-mannose dehydratase (GMD, Rabina et al. 2001). Although we were unable to verify its identity due to an unavailability of GDP-4-keto-6-deoxymannose as standard, the compound was detected at ∼1.5 fold in stable cells than the vector control. Interestingly, the levels of UDP-glucuronic acid (‘∇’ in Figure 4) eluting at 14.4 min with unknown peak at 15.5 min were also varied in the stable cells. It is conceivable that glycosylations of intracellular proteins including various glycosyltransferases and nucleotide sugar dehydrogenases cause the change in intracellular levels of nucleotide sugars, which may need further verification.
Endogenous level of fucose mutarotase in HepG2 was estimated as 0.14 ng/mg protein by immunoprecipitation (Figure 3 inset), while the amounts in human and mouse FucU cells are about 62 ng/mg protein, considering the different efficiencies of antibody recognition. The similar amounts of FucU in these cells are consistent with their similarities in both the fucose incorporation and in the GDP-l-fucose level (Figure 3). It was reported that incorporation of l-fucose in HepG2 was enhanced in concentration-dependent manner from 10 to 50 μM concentrations of sugar (Wiese et al. 1997). Since we used 10 μM of fucose in our experiment, cells might have been able to enhance fucose accumulation with an increased amount of FucU. Indeed, the fucosylation of intracellular protein was accelerated by fucose mutarotases in earlier time points, but the rate of increase remained unchanged with further incubation from 24 to 60 h. Excess fucosylation in intracellular environment may not persist due to an intracellular control of glycosylated proteins, e.g., protein turnover, while an increase in extracellular fucosylated proteins might be tolerated.
Materials and methods
l-Fucose was obtained from Sigma (St. Louis, MO), and UDP-d-galactose, UDP-d-glucose, UDP-N-acetyl-d-galactosamine, UDP-N-acetyl-d-glucosamine, UDP-glucuronic acid, GDP-d-mannose, and GDP-l-fucose from Calbiochem (San Diego, CA). UDP-xylose was purchased from Carbosource Services, and l-[5,6-3H]fucose and Safety-Solve from ICN Biomedicals (Costa Mesa, CA). Cell culture media and fetal bovine serum were obtained from HyClone (Logan, UT). Envi-Carb SPE-columns (250 mg) containing graphitized nonporous carbon were from Supelco Inc. (Bellafonte, PA). DEAE-Sepharose was from Amersham Pharmacia Biotech (Uppsala, Sweden).
Establishment of FucU stable cell lines from HepG2
The human HepG2 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum, 100 U/mL penicillin G, 100 μg/mL streptomycin, and 250 ng/mL amphotericin B. The cells were propagated in Falcon 6-well or 100 × 20 mm tissue culture flasks in an incubator maintained at 37°C with 5% CO2 in humidified air. The mouse and human FucU cDNAs were inserted into pcDNA3.1(+) plasmid using HindIII and XhoI restriction sites, thereby being regulated by the CMV promoter, which were incorporated into HepG2 cells, 10 μg each, by transfection with Lipofectamine (Invitrogen, Carlsbald, CA). Selection was carried out after an addition of 600 μg/mL G-418 (GIBCO-BRL, Rockville, MD). Thirteen positive clones of mFucU, four hFucU clones, and ten vector control clones were randomly selected. The clones (mFucU, hFucU, and HepG2-vec) were used for experiments described here, and experiments with other clones gave similar results (not shown).
Gene silencing for the human FucU
The construct for silencing harbors a 19-nucleotide sequence corresponding to nucleotides 358–376 downstream of the transcription start site (AUGAACGGGCUAAGAAGGC) of human FucU, which is separated by a 9-nucleotide noncomplementary spacer (TTCAAGAGA) from the reverse complement of the same 19-nucleotide sequence. The short hairpin oligonucleotides synthesized by Oligoengine Inc. (Seattle, WA) were annealed and ligated to the pSUPER-basic vector after digestion with BglII and HindIII, which was then transfected to HepG2 cell, together with pGK-puro vector. Selection was carried out with an addition of puromycin (1 μg/mL, Sigma).
Saturation difference (SD) experiment by NMR
The FucU genes lacking their termination codons were cloned into a pET21b vector (Novagen, Darmstadt, Germany) with NdeI and XhoI sites and expressed in E. coli strain (BL21 DE3). The C-terminal His-tagged proteins were purified with a nickel-nitrilotriacetic acid column (Qiagen, Valencia, CA). Protein solutions for NMR assay were prepared in 99% D2O buffer (pH 7.5, 50 mM sodium phosphate and 100 mM sodium chloride) by adding 99% D2O into the lyophilized protein samples. The concentrations of l-fucose and FucU protein were fixed to 25 and 0.1 mM, respectively. The SD NMR experiments were carried out using 600 MHz NMR spectrometer (Varian, UNITY INOVA, Palo Alto, CA) at 25°C, according to the method that was previously reported (Ryu, Kim, Park, et al. 2004; Ryu, Kim, Kim, et al. 2004).
Preparation of anti-FucU antiserum
Four guinea pigs were intracutaneously injected with the His-tagged mouse FucU protein (600 μg) in Freund's complete adjuvant for the first boost, and after 3 weeks the animals were injected 27 times with the protein in Freund's incomplete adjuvant with an interval of 1 week. Blood was drawn after a week of the last injection, from which the antisera were obtained and stored at −70°C.
HPLC analysis of GDP-Lfucose
Growth medium for cell culture was added with 10 μM l-fucose for conditioning cells for at least 2 weeks prior to main experiment. Before lysis, cells were incubated in DMEM containing 1 mM l-fucose for 2, 10, and 30 min. Nucleotide sugars were purified from the cell lysates as described (Rabina et al. 2001) and analyzed by ion-pair reversed-phase HPLC with Prevail C18 column (250 mm × 4.6 mm; Alltech, Deerfield, IL) at a flow rate of 1 mL/min. Isocratic 20 mM triethylamine acetate buffer (pH 6.0) was eluted for 15 min, followed by a linear gradient of 0–2% acetonitrile in 20 mM TEAA buffer for 20 min. Amount of GDP-l-fucose was calculated using nucleotide sugar standard. The column was then washed with 4% acetonitrile in 20 mM TEAA buffer for 7 min. The eluent was monitored with a UV detector at 254 nm. The numbers used in the `Results' were obtained from three independent experiments, each comprising sample preparation to a measurement by HPLC.
l-Fucose incorporation into cellular protein
Cells were grown for 16 h before the actual experiment. Culture medium (DMEM) was modified by an addition of l-[5,6-3H]fucose (1 μCi) and 10 μM unlabeled l-fucose, in which 2 × 105 cells were incubated for 1–72 h, 2.5 mL each, in 60 × 15 mm culture plates. Cells were then washed twice in ice-cold phosphate-buffered saline (pH 7.4) and collected in 1.5 mL water containing the protease inhibitor cocktail (Roche). The cell suspension was sonicated for 10 s on ice, and samples were taken to determine protein concentrations, accumulation of l-[5,6-3H]fucose, and incorporation into cellular proteins. Accumulation (uptake) of l-[5,6-3H]fucose was assessed by taking duplicated aliquots of cell suspensions, adding 4 mL of Safety-Solve (ICN Biomedicals), and measuring the radioactivity in Beckman LS6500 liquid scintillation counter. Incorporations of l-[5,6-3H]fucose into intracellular proteins were determined by treating a 700 μL aliquot of the cell suspension with an equal volume of 20% TCA. After vortexing and allowing the sample to stand on ice for 5 min, the protein was pelleted by centrifugation. Here, the unbound l-[5,6-3H]fucose, lipids, and chemical species such as l-[5,6-3H]fucose 1-phosphate, GDP-l-[5,6-3H]fucose were removed. The pellet was washed with 10% TCA and then transferred to a scintillation vial, followed by an addition of 4 mL Safety-Solve and measurements of their radioactivities. All assays were conducted in triplicates.
In order to assess incorporation of l-[5,6-3H] fucose into secreted protein, cells were incubated in DMEM containing l-[5,6-3H]fucose (1 μCi) and 10 μM unlabeled l-fucose. At the indicated time (Figure 3), the medium was collected, centrifuged (10,000 × g, 1 min), and mixed with 600 μL of 20% TCA in 500 μL aliquot of the cell-free medium plus 100 μL of 1 mg/mL BSA as a carrier protein. After vortexing and incubating on ice for 5 min, the protein was pelleted by centrifugation, which was washed with 10% TCA. The pellet was then transferred to a scintillation vial with an addition of 4 mL Safety-Solve, followed by a measurement of radioactivity. As a control for nonspecific absorption of l-[5,6-3H]fucose by the carrier protein, the medium itself without cells was incubated and treated in the same manner (Wiese et al. 1997).
The sequence data for RbsDs and FucUs were downloaded from UCSC (http://genome.ucsc.edu/) and NCBI (http://www. ncbi.nlm.nih.gov/) databases, which were multiply aligned using Genebee (http://www.genebee.msu.su/) and Pfam (http:// www.sanger.ac.uk/Software/Pfam/) softwares. Based on strong sequence homologies among them, homology-based models were built using the structure of B. subtilis RbsD (1OGD, http://www.rcsb.org/pdb/) as a backbone (http://swissmodel. expasy.org/). The Figure 5 was drawn based on this model.
21C Frontier Microbial Genomics and Application Center Program, Ministry of Science & Technology, Republic of Korea (MG05-0202-2-0 to C.P.).
Conflict of interest statement
bovine serum albumin
cytidine mono- phosphate-N-acetylneuraminic acid
Dulbecco's modified Eagle's medium
high performance liquid chromatography
nuclear magnetic resonance
saturation transfer difference
trichloro- acetic acid