Independent and simultaneous translocation of two substrates by a nucleotide sugar transporter

Abstract

Nucleotide sugar transporters play an essential role in protein and lipid glycosylation, and mutations can result in developmental phenotypes. We have characterized a transporter of UDP-N-acetylglucosamine and UDP-N-acetylgalactosamine encoded by the Caenorhabditis elegans gene C03H5.2. Surprisingly, translocation of these substrates occurs in an independent and simultaneous manner that is neither a competitive nor a symport transport. Incubations of Golgi apparatus vesicles of Saccharomyces cerevisiae expressing C03H5.2 protein with these nucleotide sugars labeled with ³H and ¹⁴C in their sugars showed that both substrates enter the lumen to the same extent, whether or not they are incubated alone or in the presence of a 10-fold excess of the other nucleotide sugar. Vesicles containing a deletion mutant of the C03H5.2 protein transport UDP-N-acetylglucosamine at rates comparable with that of wild-type transporter, whereas transport of UDP-N-acetylgalactosamine was decreased by 85–90%, resulting in an asymmetrical loss of substrate transport.

In eukaryotes, ≈80% of secreted or membrane-bound proteins are glycosylated in the lumen of the Golgi apparatus. The glycan portion of these proteins has been shown to play a pivotal role in the structure and/or activity of proteins as well as in proteoglycans and glycolipids. The sugar donors in the biosynthesis of these glycoconjugates are nucleotide sugars. These must be translocated from the cytosol, where most are synthesized, into the Golgi apparatus lumen by nucleotide sugar transporters. In the lumen, enzymes transfer sugars from nucleotide sugars to the appropriate acceptors. Transporters of nucleotide sugars are hypothesized to regulate the availability of their substrates in the Golgi apparatus lumen and thereby the quantity and quality of their corresponding reaction products (1, 2).

Nucleotide sugar transporters are very hydrophobic proteins with 6–10 transmembrane spanning domains and appear to be homodimers within the Golgi apparatus membrane (3, 4). Primary sequence of these transporters has not proven to be a reliable indicator of substrate specificity (5). Transporters with 40–50% amino acid sequence identity may have very different substrate specificities, i.e., the mammalian transporters for CMP-sialic acid and UDP-galactose (6).

Mutants of nucleotide sugar transporters have developmental phenotypes and have been described in Caenorhabditis elegans (7, 8), yeast (9), Leishmania donovani (10), Drosophila melanogaster (11, 12), cattle (13), and humans (14–16). Mutations in the transporters of UDP-N-acetylglucosamine (UDP-GlcNAc) in Holstein cows and GDP-fucose in humans give rise to complex vertebral malformation and leukocyte adhesion deficiency syndrome II, respectively. Developmental phenotypes are often accompanied by deficiencies in specific sugars bound to proteins, proteoglycans, and lipids.

We have recently begun in-depth studies on the biochemistry and biology of nucleotide sugar transporters in C. elegans. So far, this has been done with only two such transporters, SQV-7 and SRF-3. SQV-7, the first nucleotide sugar transporter characterized in this organism, was identified in a screen of nematodes with a squashed vulva (Sqv) phenotype. It transports UDP-galactose, UDP-glucuronic acid, and UDP-N-acetylgalactosamine (UDP-GalNAc). Until then, we had found that nucleotide sugar transporters were specific for only one substrate. Transport of each of the above three nucleotide sugars by SQV-7 was saturable, and transport of one of the nucleotide sugars was inhibited by the other two in a competitive, noncooperative manner. The two mutants in SQV-7 had a concomitant decrease in all three transport activities, consistent with the above competitive transport kinetics (7). Analyses of glycans deficient in SQV-7 mutants showed that both chondroitin and heparan sulfate were diminished, as predicted by the substrate specificity of SQV-7 (17). The expression pattern of SQV-7 was tissue-specific (18).

SRF-3, the other nucleotide sugar transporter characterized in C. elegans so far, transports UDP-galactose and UDP-GlcNAc. Although mutants deficient in SRF-3 activity do not show morphological phenotypes when grown on E. coli, they are resistant to infection by the bacterial pathogens Microbacterium nematophilum and Yersinia pseudotuberculosis, whereas the wild-type nematodes are not. SRF-3 has a narrow tissue expression pattern, and mutants have a decrease in O- and N-linked galactose-containing glycoconjugates (8, 19).

We have now characterized a transporter of C. elegans, encoded by the CO3H5.2 gene that transports UDP-GlcNAc and UDP-GalNAc independently and simultaneously. This is a noncompetitive, nonsymport mechanism for a multisubstrate transporter.

Results

CO3H5.2 Transports UDP-GlcNAc and UDP-GalNAc.

The genome of C. elegans encodes 18 putative nucleotide sugar transporters based on amino acid sequence homology to characterized nucleotide sugar transporters. So far, only two of these transporters, SQV-7 (7) and SRF-3 (8), have been characterized for their substrate specificity. Although both transporters transport multiple substrates, it is only for SQV-7, the first described multisubstrate nucleotide sugar transporter, that competition among these substrates has been demonstrated. In all other cases, including mammals (20, 21), insects (11, 12), plants (22), and protista (23), where multiple substrates have been described for a given nucleotide sugar transporter, a competitive transport mechanism has been assumed but not shown.

We chose to study the substrate specificity of the CO3H5.2 gene product, a putative nucleotide sugar transporter of C. elegans, because of its high amino acid sequence homology to SRF-3 (37.8%) and relatively low homology to SQV-7 (17.4%). Although CO3H5.2 has relatively high amino acid sequence homology to the human UDP-GlcNAc (55.4%) and UDP-galactose (37.6%) transporters (see Fig. 4, which is published as supporting information on the PNAS web site), previous studies from our and other laboratories have shown that sequence similarity is not sufficient to predict substrate specificity of these transporters (5).

We therefore determined the ability of a FLAG-C03H5.2 fusion protein, expressed in Saccharomyces cerevisiae, to transport different nucleotide sugars. Previously we had shown that microsomal preparations of this yeast have only two endogenous nucleotide sugar transport activities, those for UDP-glucose and GDP-mannose (24, 25). This facilitates assays of nucleotide sugar transport when a putative nucleotide sugar transporter from a different species is expressed in S. cerevisiae. This approach has been used by us and others to determine substrate specificity of transporters from mammals (26), C. elegans (7, 8), plants (22), and protists (23). As shown in Fig. 1B, S. cerevisiae transformed with the vector encoding the CO3H5.2 fusion protein express this protein in their membrane fraction, whereas yeast transformed with the vector alone do not. The mobility of the fusion protein in SDS/PAGE was somewhat higher than that predicted for a 37.2-kDa protein. A similar observation was previously made in our and other laboratories when nucleotide sugar transporter encoding genes from a heterologous species were expressed in S. cerevisiae.

Fig. 1.

Substrate specificity of the CO3H5.2 transporter. (A) CO3H5.2-encoded nucleotide sugar transport into Golgi vesicles from S. cerevisiae. Transport activity is shown as the amount of nucleotide sugar within vesicles at 25°C minus the corresponding value at 0°C. S. cerevisiae was transformed with pG426 (gray bars) or pG426-FLAG-CO3H5.2 (black bars). Results are the average of two independent determinations and standard error. (B) Western blot analyses of proteins derived from Golgi vesicles after SDS/PAGE. S. cerevisiae was transformed with the empty vector, pG426, or the vector encoding FLAG-CO3H5.2 wild-type and mutations V192F and G267E. Proteins were visualized with an anti-FLAG monoclonal antibody. (C) Nucleotide sugar transport with UDP-[³H]GalNAc (Left), UDP-[³H]GlcNAc (Center), and UDP-[³H]Glc (Right) into Golgi vesicles of cells transformed with wild-type CO3H5.2 and the two above-described point mutants. Concentration of nucleotide sugar was 1 μM. Results are expressed as averages of two independent determinations and the standard error. (D) Separation of K. lactis by FACS after cell-surface labeling with fluorescent lectin from G. simplicifolia II. K. lactis Golgi apparatus UDP-GlcNAc transporter mutants KL3 were transformed with pE4 or pE4-CO3H5.2. Cells were then incubated with the fluorescent lectin and separated in a FACS. (Upper Left) KL8 wild type. (Upper Right) KL3 mutant. (Lower Left) KL3 transfected with empty plasmid pE4. (Lower Right) KL3 transfected with pE4-C03H5.2.

Golgi apparatus-enriched vesicles from S. cerevisiae transformed with the vector encoding the above fusion protein or the vector alone were next isolated. Vesicle integrity was measured by their ability to transport UDP-glucose, an endogenous activity. As can be seen in Fig. 1A, vesicles transformed with the empty vector or the vector encoding the FLAG-CO3H5.2 fusion protein have comparable UDP-glucose transport activities, suggesting that both preparations are of the same quality. Both vesicle preparations were then assayed for their ability to transport different nucleotide sugars (Fig. 1A). Transport of UDP-GlcNAc and UDP-GalNAc into vesicles expressing CO3H5.2 was 3- to 4-fold higher than into those transformed with the empty vector. The low transport signal observed for UDP-galactose was comparable in both preparations, suggesting that CO3H5.2 was not a transporter for this substrate (Fig. 1A). In a separate experiment, we determined that neither UDP-glucuronic acid nor GDP-fucose were translocated via CO3H5.2 (data not shown). We then studied the kinetics of UDP-GlcNAc and UDP-GalNAc transport. Transport of UDP-GlcNAc and UDP-GalNAc were saturable with apparent K_m values of 25.2 and 24.9 μM, respectively. These results are in close agreement with previously described apparent K_m values for other nucleotide sugar transporters (8).

To further demonstrate that CO3H5.2 indeed encodes a nucleotide sugar transporter, we introduced in the CO3H5.2 protein encoding region two different point mutations resembling missense mutations of transporters of UDP-GlcNAc from other organisms. One of the mutations (V192F) affects the bovine UDP-GlcNAc transporter causing complex vertebral malformation and leads to abnormal protein glycosylation in tissues of affected animals (13). The other mutation (G267E) in the SRF-3 transporter of C. elegans, which transports UDP-galactose and UDP-GlcNAc, causes loss of both nucleotide sugar transport activities and a pattern of abnormal glycosylation (8). Both of the above point mutations are within conserved amino acid sequence regions of CO3H5.2 (see Fig. 4). We therefore introduced such mutations, separately, in the CO3H5.2 gene and expressed the appropriate fusion proteins in S. cerevisiae. After isolation of Golgi apparatus-enriched vesicles, these, as well as those from cells that had been transformed with the wild-type gene, were assayed for their ability to transport UDP-GlcNAc and UDP-GalNAc (Fig. 1C). As can be seen in Fig. 1B, although expression of the two transporter proteins with the point mutations and the wild-type construct was comparable, transport activity was significantly lower into vesicles expressing the mutated transporters compared with those expressing the wild-type one (Fig. 1C). Transport of UDP-glucose, as expected, was not affected (Fig. 1C). This strongly suggests that transport of UDP-GlcNAc and UDP-GalNAc, mediated by the CO3H5.2 protein, depended on an active transporter protein rather than protein expression per se. The results also demonstrate that the mutated residues are within conserved regions affecting transport activity.

To demonstrate that the above-described UDP-GlcNAc in vitro transport activity also occurs in vivo, we took advantage of a previously characterized mutant of the yeast K. lactis that is deficient in transport of UDP-GlcNAc into its Golgi apparatus lumen. This mutant lacks terminal N-acetylglucosamine in its mannan chains and has significantly reduced binding to its surface by the lectin Griffonia simplicifolia II (9). We had also shown that the wild-type UDP-GlcNAc transporter from K. lactis, when expressed in mutant cells, restored the lectin binding phenotype (27). The same experimental approach was used to demonstrate that the bovine UDP-GlcNAc transporter was indeed impaired in complex vertebral malformation in bovines (13). We therefore transformed mutant K. lactis cells with the vector pE4 encoding the CO3H5.2 protein or the vector alone. Phenotypic correction was monitored after cell-surface labeling with fluorescein isothiocyanate conjugated to G. simplicifolia II lectin and a fluorescent activated cell scanner. As can be seen in Fig. 1D, ≈50% of cells, expressing the CO3H5 transporter, showed lectin binding with levels comparable with wild-type cells. This strongly suggests that the UDP-GlcNAc transport activity of CO3H5.2 can restore addition of N-acetylglucosamine to K. lactis proteins in vivo.

CO3H5.2 Transports UDP-GlcNAc and UDP-GalNAc Independently and Simultaneously.

We had previously determined that UDP-glucuronic acid, UDP-GalNAc, and UDP-galactose are competitive substrates for the C. elegans nucleotide sugar transporter SQV-7 (7). We initially assumed that the same mechanism occurred with transporter CO3H5.2. If this was the case, one would expect a decrease in transport of UDP-[³H]GlcNAc upon addition of nonradioactive UDP-GalNAc to the reaction mixture and vice versa. This phenomenon would cause an affinity shift to the right (higher K_m) of the corresponding saturation curve. However, as can be seen in Fig. 2A, this did not occur. Addition of 20 μM nonradioactive UDP-GalNAc to the reaction mixture does not cause a right shift of the UDP-GlcNAc transport saturation curve but a left one (higher V_max; Fig. 2A). This result suggests stimulation or cooperation among the substrates by an unknown mechanism. In any case, UDP-GalNAc does not inhibit UDP-GlcNAc transport. In a similar experiment, the radioactive UDP-GalNAc transport saturation curve was compared with that obtained when 20 μM nonradioactive UDP-GlcNAc was added at each point of the curve. In this experiment, no right shift of the saturation curve, as expected for inhibition (decrease in affinity), was observed either (Fig. 2B). These results show that UDP-GalNAc and UDP-GlcNAc do not inhibit entry of each other and suggest that both substrates are translocated independently into the Golgi apparatus lumen via transporter CO3H5.2.

Fig. 2.

Kinetics of substrate transport of the CO3H5.2 transporter. (A) Transport into S. cerevisiae Golgi vesicles from cells transformed with CO3H5.2 showing transport of UDP-GlcNAc versus concentration without (filled circles) or in the presence of 20 μM nonradioactive UDP-GalNAc (open circles). (B) Transport of UDP-GalNAc versus concentration without (filled circles) or in the presence (open circles) of 20 μM nonradioactive UDP-GlcNAc. (C) Double radiolabeling experiment. Shown are results of incubations of Golgi vesicles from S. cerevisiae encoding CO3H5.2 with 2 μM UDP-[¹⁴C]GlcNAc (Left) and 20 μM UDP-[³H]GalNAc (Right). (D) Vesicles encoding CO3H5.2 were incubated with 0.5 μM UDP-[³H]GlcNAc in the presence of variable concentration of nonradiolabeled UDP-GalNAc and UDP-Gal. Transport is expressed as the percentage of the initial rate of UDP-GlcNAc transport in the absence of the nonradioactive nucleotide sugars.

Direct evidence that UDP-GlcNAc and UDP-GalNAc are transported independently and simultaneously was obtained from an uptake experiment into S. cerevisiae Golgi-enriched vesicles encoding the CO3H5.2 transporter in which each substrate was labeled in their sugar with a different radionuclide. After incubations, the amount of each radionuclide transported into the lumen of the vesicles was determined. As can be seen in Fig. 2C Left, the amount of UDP-[¹⁴C]GlcNAc inside the vesicles was the same when the nucleotide sugar was incubated alone or in the presence of a 10-fold excess of UDP-[³H]GalNAc. Fig. 2C Right shows that the amount of UDP-[³H]GalNAc entering the vesicles was also independent of the presence of UDP-[¹⁴C]GlcNAc in the reaction mixture. These results are consistent with our hypothesis that nucleotide sugars enter the Golgi apparatus lumen via the CO3H5.5 protein without competitive transport. As a control, an autocompetition double labeling experiment was performed by using UDP-[¹⁴C]GlcNAc and UDP-[³H]GlcNAc. The addition of a 10-fold increase of UDP-[³H]GlcNAc to the reaction mixture containing 2 μM UDP-[¹⁴C]GlcNAc caused a 65% decrease in the amount of UDP-[¹⁴C]GlcNAc entering the vesicles (data not shown).

Additional support for the above hypothesis is shown in Fig. 2D. Golgi apparatus-enriched vesicles, expressing the CO3H5.2 protein, were incubated with 0.5 μM UDP-[³H]GlcNAc in the presence of increasing concentrations of the other substrate, UDP-GalNAc and of UDP-Gal, a nonsubstrate. The amount of radiolabeled UDP-GlcNAc entering the vesicles was not affected at any concentrations of the other nucleotide sugars, regardless of whether they are substrates for this transporter. Together, these approaches and results strongly suggest that protein CO3H5.2 transports its substrates in a simultaneous, independent manner that is different from the competitive substrate transport mechanism previously observed for SQV-7.

A 16-aa Deletion in Protein CO3H5.2 Causes an Asymmetrical Loss of Transport Activity.

Based on the transmembrane domains predicted by the TMHMM model (see Fig. 4), a conserved region of 16 aa was deleted in the loop between TM2 and TM3 of protein CO3H5.2. Although we expected this mutant to be inactive in the transport of both nucleotide sugar substrates, as found for the point mutation described in Fig. 1C, the results, shown below, were surprising. The deletion mutant was cloned as a FLAG-fusion protein and expressed in S. cerevisiae. After isolation of a Golgi apparatus-enriched vesicle fraction, transport of UDP-GalNAc and UDP-GlcNAc into vesicles was measured and compared with that into vesicles obtained from yeast transformed with the wild-type FLAG-CO3H5.2 fusion construct. Levels of expression of each transporter construct were monitored by immunoblots. Depending on the particular experiment, the levels of expression of the deletion mutant construct were found to be similar or somewhat lower than that of the wild-type construct (Fig. 3A). As shown in Fig. 3B, whereas transport of UDP-GalNAc was severely impaired in the mutant construct, transport of UDP-GlcNAc was virtually not affected. Clearly an asymmetrical loss of transport activity of each substrate had occurred.

Fig. 3.

Kinetics of substrate transport of deletion mutant of the CO3H5.2 transporter. (A) Immunoblot of proteins subjected to SDS/PAGE from vesicles of S. cerevisiae expressing wild-type FLAG-CO3H5.2 or the 16-aa deletion mutant of CO3H5.2. (B) Transport of 2 μM UDP-GalNAc (Left) and 2 μM UDP-GlcNAc (Right) into Golgi vesicles expressing the wild-type or deletion construct. The results are expressed as averages of two independent determinations and the standard error. (C) Transport of UDP-GalNAc. (D) UDP-GlcNAc. Filled triangles, vesicles containing the wild-type construct; open circles, vesicles from cells transformed with empty vector; filled circles, vesicles from cells encoding the deletion mutant transporter. (E) TSR. Mixtures of 10 μM UDP-[³H]GalNAc and 10 μM UDP-[¹⁴C]GlcNAc were prepared in such a manner that the sum of their concentrations remained constant at 10 μM. Golgi vesicles from S. cerevisiae expressing the wild-type or deletion mutant construct of CO3H5.2 were incubated by using the nucleotide sugars in ratios of 3:7, 5:5, and 7:3 as described in Materials and Methods, and TSR was measured. TSR = (³H dpm/¹⁴C dpm) × ([UDP-GlcNAc]/[UDP-GalNAc]). Filled squares, wild-type CO3H5.2; open squares, deletion mutant.

To further rule out that the decreased transport activity for UDP-GalNAc in the deletion mutant was a consequence of reduced protein expression levels, two different experimental approaches were pursued. In one case, shown in Fig. 3C, the rate of transport of UDP-GalNAc and UDP-GlcNAc was measured at different concentrations of substrate. The saturation profile for UDP-GalNAc with the deletion mutant transporter appears to be virtually independent of the concentration of the substrate and is very similar to the values measured by using vesicles of yeast transformed with the empty vector (Fig. 3C). On the other hand, as shown in Fig. 3D, the deletion mutant is still able to transport UDP-GlcNAc in a saturable manner, even though the K_m and V_max values are slightly higher than those of the wild-type transporter (Fig. 3D). We don't know the reason for this.

We also calculated the “transport specificity ratio” (TSR) for each substrate nucleotide sugar with the above wild-type and deletion constructs of CO2H5.2. This method detects changes in intrinsic substrate binding energy that are reflected as perturbations in substrate specificity (k_cat/K_m). TSR constitutes an index of the different specificities by a protein toward a pair of substrates. The TSR value is constant and independent of the substrate concentrations used and of the expression level of the particular transporter protein in the membrane (28). TSR is calculated from a double radiolabel transport assay in which UDP-[¹⁴C]GlcNAc and UDP-[³H]GalNAc are mixed in 3:7, 5:5, and 7:3 ratios with a total constant substrate concentration of 10 μM. Results are expressed as TSR = (³H dpm/¹⁴C dpm) × ([UDP-GlcNAc]/[UDP-GalNAc]). As expected, TSR values calculated for the wild-type and deletion mutant CO2H5.2 transporter constructs remain constant with the different mixtures of substrates. However, the TSR value for the deletion mutant transporter is considerably lower than that calculated for the wild-type transporter, suggesting a reduction in substrate specificity toward UDP-GalNAc (Fig. 3E). These results, therefore, strongly support our hypothesis that the deletion in CO3H5.2 causes an asymmetrical loss of UDP-GalNAc transport activity relative to UDP-GlcNAc and suggest that the amino acids involved in determining substrate specificity for both substrates are different.

Discussion

We have characterized a new nucleotide sugar transporter of C. elegans encoded by the gene CO3H5.2. The transporter appears to translocate substrates via a previously uncharacterized mechanism for transporters in general.

Expression of CO3H5.2 in S. cerevisiae and subsequent isolation of Golgi apparatus-enriched vesicles showed that CO3H5.2 is a nucleotide sugar transporter that translocates UDP-GlcNAc and UDP-GalNAc in a saturable manner. CO3H5.2 can also correct the cell-surface phenotype of a mutant of the yeast K. lactis that is specifically defective in transport of UDP-GlcNAc into its Golgi apparatus. Although the nucleotide sugars are epimers, we had previously shown that nucleotide sugar transporters may be epimer-specific (1, 7).

We had characterized SQV-7 as a multisubstrate nucleotide sugar transporter in which substrate translocation occurred via a competitive, noncooperative mechanism (7). Since then, multisubstrate transporters of nucleotide sugars have been described in several organisms, including L. donovani (29), D. melanogaster (12), humans (20), E. histolytica (23), etc. A competitive substrate translocation mechanism has been since then assumed but not proven in any other study.

We have now obtained three separate lines of evidence strongly suggesting that CO3H5.2 translocates its two nucleotide sugars simultaneously and independently of each other. The first line of evidence comes from the experiment shown in Fig. 2A in which transport of UDP-GlcNAc via CO3H5.2 was not inhibited, but actually somewhat stimulated, upon addition of 20 μM UDP-GalNAc into the reaction medium. The same lack of inhibition was observed when transport of UDP-GalNAc was measured in the presence of 20 μM UDP-GlcNAc (Fig. 2B). These results are very different from the significant inhibition of transport observed via SQV-7 when UDP-Gal transport was measured in the presence of excess UDP-GlcA or UDP-GalNAc.

The second line of evidence directly demonstrates that both nucleotide sugars are translocated into the lumen of yeast vesicles via CO3H5.2 in a simultaneous and noncompetitive manner. Incubations of vesicles with UDP-GlcNAc and UDP-GalNAc, labeled with different radionuclides in their sugars, showed that both substrates enter the lumen to the same extent, whether they are incubated alone or in the presence of a 10-fold excess of the other nucleotide sugar (Fig. 2C). When the same experimental design was done with vesicles expressing the SQV-7 transporter instead of CO3H5.2, it was found that a 10-fold excess of UDP-[¹⁴C]GlcA decreases the amount of entry of UDP-[³H]GalNAc by ≈80%, in complete agreement with the substrate competitive inhibition curves previously described for this transporter (data not shown).

The third line of evidence supporting our hypothesis that CO3H5.2 translocates its substrates in a simultaneous and independent manner comes from the asymmetrical loss of transport activity resulting from the deletion of the putative third loop of the CO3H5.2 transporter. Although, as mentioned previously, this finding was fortuitous, because we expected to find loss of both substrate translocation activities, the results strongly support the hypothesis that translocation of both substrates is mediated by different amino acids of the CO3H5.2 protein. In the case of SQV-7, the two point mutations showed a parallel decrease of all three translocation activities in contrast to the results shown in Fig. 3B for CO3H5.2.

Although we are not aware of another transporter that can translocate its solutes in a simultaneous and independent manner, a chimera of the mammalian nucleotide sugar transporters for CMP-sialic acid and UDP-galactose, constructed with the aim of determining the translocation sites for each nucleotide sugar, was able to transport both substrates (6). These observations lead us to speculate that perhaps CO3H5.2 in the membrane has two separate translocation sites, although other possible translocation mechanisms cannot be ruled out until detailed structural studies on this transporter are performed. Because nucleotide sugar transporters have been found to be active as homodimers in the membrane, we do not know whether the translocation sites require amino acids from both subunits or whether each subunit has its own site.

Interestingly, a phylogenetic tree of putative nucleotide sugar transporters of C. elegans shows that SRF-3 and CO3H5.2 are closely related, whereas SQV-7 is not. This suggests that different translocation mechanisms may have evolved for these transporters. The results presented in this study suggest that other multisubstrate transporters with simultaneous and independent translocation sites may exist in other organisms including mammals.

Materials and Methods

Strain Maintenance and Genetics.

S. cerevisiae strain PRY225 (ura3-52, lys2-801am, ade2-1020c, his3, leu2, trpl-1Δ1) was grown at 30°C in liquid yeast extract/peptone/dextrose or on solid yeast extract/peptone/dextrose media containing 2% Bacto-agar. Strains derived from PRY225 transformed with URA plasmids were grown at 30°C in synthetic complete medium lacking uracil prepared by using SCM-URA (Sigma). For transformation with pG426 or pG426-FLAG-C03H5.2, a lithium acetate/polyethylene glycol method was used (30). The following K. lactis strains were used: KL3 (Mat a, uraA, mnn2–2, arg K⁺, pKD1⁺) (9) and MW103-1C (Mat a, uraA lysA, K⁺, pKD1⁺) (31). K. lactis was grown at 30°C in synthetic complete medium lacking uracil prepared by using SCM-URA (Sigma).

Molecular Biology.

Standard molecular biology protocols were used as described by Sambrook et al. (32) unless otherwise noted. The C. elegans C03H5.2 cDNA was purchased from Open Biosystems and amplified by PCR using BIO-X-ACT Short Mix DNA polymerase (BIOLINE). The primers C03H5.2FLAGfor (forward) and C03H5.2rev (reverse) were used to clone C03H5.2 cDNA into the S. cerevisiae pG426 vector. The sequence of these oligonucleotide primers were CGGGCCACTAGTATGGATTACAAGGATGACGAC GATAAGAATCGAGCTAACGACA (C03H5.2FLAGfor) and CGCCGCTCGAGTTATCAGGCATTATGAGCTTCG (C03H5.2rev). The isolated PCR product sequence, fused with an 8-aa FLAG tag coding sequence, was digested with SpeI/XhoI and ligated to the p426GPD vector (digested with SpeI/XhoI) to obtain plasmid pG426-FLAG-C03H5.2 for the expression in yeast (33). The identity of the cloned fragment was confirmed by DNA sequencing. Similarly, the open C03H5.2 ORF was into the K. lactis pE4 vector.

Radioactive Substrates.

The following radioactive substrates used were all purchased from American Radiolabeled Chemicals: UDP-[³H]Glc [60 Ci/mmol (1 Ci = 37 GBq)], UDP-[³H]Gal (20 Ci/mmol), UDP-[³H]GlcNAc (60 Ci/mmol), UDP-[³H]GalNAc (60 Ci/mmol), and UDP-[¹⁴C]GlcNAc (300 mCi/mmol). GDP-[³H]Fuc was purchased from PerkinElmer.

Western Blot Analysis.

Total membrane fractions, from 3 ml of liquid yeast cultures (AB_600B = 3.0), were prepared by glass bead disruption of the cells in membrane buffer [0.8 M sorbitol/1 mM EDTA/protease inhibitor mixture (Roche Applied Science)/10 mM triethanolamine/acetic acid, pH 7.2, at 4°C] plus 2 mM DTT after centrifugation at 134,000 × g for 1 h, total membranes were resuspended in membrane buffer, and proteins were electrophoresed on NuPAGE 4–12% Bis-Tris gels (Invitrogen) and subsequently electrotransferred onto Immobilon-P membranes (Millipore). Nonspecific binding was blocked by incubating the membranes for 1 h at room temperature in 5% nonfat dry milk in phosphate-buffered saline buffer. Membranes were incubated for 1 h with 1 μg/ml mouse anti-FLAG M2 antibody (Sigma) in phosphate-buffered saline. Protein detection was performed by using horseradish peroxidase-conjugated mouse IgG (Promega) followed by chemiluminescence using ECLP Western blotting detection reagents (Amersham Biosciences).

Subcellular Fractionation.

S. cerevisiae transformed with pG426 or pG426-FLAG-C03H5.2 were grown in SCM-URA liquid medium to OD₆₀₀ = 3. The culture was centrifuged and converted to spheroplasts as previously described by using a total of 1 mg of Zymolyase 100 T (Seikagaku America) per g of cells. The spheroplast suspension was centrifuged at 2,000 × g for 10 min. Cells were broken by suspending the pellet in 1.5× volume of membrane buffer and by drawing the cells rapidly several times into a narrow-bore serological pipette. The suspension was centrifuged successively for 10 min at 2,000 × g, for 8 min at 5,000 × g, and finally for 45 min at 125,000 × g to obtain a pellet fraction enriched in endoplasmic reticulum- and Golgi apparatus-derived vesicles.

Nucleotide Sugar Transport Assay.

The theoretical basis for the translocation assay of nucleotide sugars into Golgi apparatus-enriched vesicles has been described previously. Transport assays of C. elegans nucleotide sugar transporter expressed in S. cerevisiae and analyses of the samples was carried out as described in ref. 7. Radioactivity was detected by using a liquid scintillation spectrometer.

Abbreviations

TSR

transport specificity ratio

UDP-GalNAc

UDP-N-acetylgalactosamine

UDP-GlcNAc

UDP-N-acetylglucosamine.