Functional expression of the murine Golgi CMP-sialic acid transporter in saccharomyces cerevisiae.

We have functionally expressed the murine Golgi putative CMP-sialic acid transporter in Saccharomyces cerevisiae. Using a galactose-inducible expression system, S. cerevisiae vesicles were able to transport CMP-sialic acid. Transport was dependent on galactose induction and was temperature-dependent and saturable with an apparent Km of 2.9 microM. Transport was inhibited by CMP, and upon vesicle disruption with Triton X-100 parameters were very similar to the previously described CMP-sialic acid transport characteristics observed with mammalian Golgi vesicles. CMP-sialic acid transport induction was specific as no transport of UDP-galactose was observed even though the latter putative transporter has a high degree of amino acid sequence identity with the CMP-sialic acid transporter. Together, the above results demonstrate that the previously described cDNA encoding the putative CMP-sialic acid transporter encodes the transporter protein per se and suggests that this heterologous expression system may be used for further structural and functional studies of other Golgi membrane transporter proteins.

Transporters for nucleotide sugars, nucleotide sulfate, and ATP of the Golgi apparatus membrane are required for the translocation of these solutes from the cytosol into the lumen of this organelle (1,2). Within this compartment these nucleotide derivatives and ATP are substrates for glycosylation, sulfation, and phosphorylation of secretory and membrane-bound proteins as well as lipids (1,2). Studies in vitro and in vivo have shown these transporters to be antiporters with the corresponding nucleoside monophosphates (1,2). These transport activities have been detected in Golgi vesicles from mammals (1,2), yeast (1,2), protozoa (3), and plants (4). Mutants defective in transport activities of CMP-sialic acid (1,2), UDPgalactose (1,2), UDP-N-acetylglucosamine (1,2), and GDPmannose (3,5) have been described in these organisms and have a block of glycosylation of proteins and lipids in vivo. These phenotypes have been used as selection for expression cloning of nucleic acids, which encode proteins that correct the phenotype of the yeast UDP-GlcNAc transporter (6), the murine CMP-sialic acid transporter (7), the Leishmania donovani GDP-mannose transporter (5), and the human UDP-galactose transporter (8). So far in every instance, a highly hydrophobic multitransmembrane spanning domain protein was found to correct the mutant phenotype, suggesting that the protein is the corresponding Golgi membrane nucleotide sugar transporter. In some studies, the protein was localized to the Golgi apparatus (3,7), while in other cases Golgi vesicles from the transfected mutant cells were shown to have recovered the ability to transport the nucleotide sugar, which the corresponding mutant cell line vesicle lacked (3,6).
We have expressed the murine Golgi apparatus membrane putative CMP-sialic acid transporter in Saccharomyces cerevisiae for the following reasons. (a) Because yeast cells do not synthesize sialoglycoconjugates and do not have such transporters, expression of transport activity would provide strong evidence that the putative CMP-sialic acid transporter is, indeed, the transporter protein per se and not a protein regulating the transporter in a mammalian cell. (b) Expression would allow a rapid and easy manner to obtain relatively large amounts of the transporter protein to be used for future structural as well as organellar targeting studies.

MATERIALS AND METHODS
Construction of pPB11-The cDNA for the CMP-sialic acid transporter tagged in the C terminus with 33 nucleotides coding for the hemagglutinin (HA) 1 epitope (1.1 kilobase pairs) was obtained by digesting pME8-HA (7) with NcoI, filled in with Klenow DNA polymerase, and then digested with XbaI and isolated from agarose using Spin-X filters (Costar), followed by phenol/chloroform extraction and ethanol precipitation. pYES2 (Invitrogen, Inc.) was digested with SacI, filled in with Klenow DNA polymerase, and then digested with XbaI and ligated to the 1.1-kilobase pair fragment. Following transformation of DH5␣ Escherichia coli, plasmid DNA obtained from transformants was analyzed by restriction digestion. One of the plasmids, pPB11, was used for further studies.
Expression of the CMP-Sialic Acid Transporter in S. cerevisiae-S. cerevisiae INVSc1 (MAT␣, his3-d1, Leu2, trp1-289, ura3-52) (Invitrogen, Inc.) cells were made competent for electroporation by extensive washing in ice-cold sterile water, resuspended in 10% glycerol, and electroporated with 1 g of pYES2 or pPB11 using a Bio-Rad gene pulser (2.5 kV, 25 microfarads, 200 ohms). Cells were incubated for 1 h at 30°C in YPD medium before plating in SD agar plates containing 30 mg/liter L-leucine, 2 mg/liter L-histidine, and 20 mg/liter L-tryptophan. Transformants grown at 30°C for 2 days were isolated and grown in liquid selective medium (0.67% Bacto-yeast nitrogen base, 2% dextrose, 2 mg/liter L-histidine, 20 mg/liter L-tryptophan, 30 mg/liter L-leucine). For induction experiments, transformants were grown in liquid selective medium in which dextrose was replaced by 4% raffinose, until they reached an A 600 of 2.0. Samples were taken uninduced, and the remaining cells were spun at 2,000 ϫ g for 10 min to remove the medium. Cells were then resuspended in the same volume of selective medium in which raffinose had been replaced by 2% galactose and allowed to grow at 30°C for the times indicated in each case.
Western Blots-To determine expression of the CMP-sialic acid transporter protein, samples of uninduced and induced cells were taken at different times (A 600 of 1.0 each), and total extracts were prepared as described previously (9). Total extracts were fractionated in 12% SDSpolyacrylamide gel electrophoresis gels, electrotransferred to polyvi-* This work was supported by National Institutes of Health Grant GM R37-30365. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

RESULTS
To express the mammalian putative CMP-sialic transporter in S. cerevisiae, a yeast expression vector with an inducible promoter was chosen, since this would allow the separation of the growth phase under uninduced conditions from the expression phase in the presence of the inducer. This approach may also circumvent the possibility that high levels of expressions of this heterologous protein may be toxic or growth inhibitory. For these reasons, pYES2, a 2-m derived plasmid containing the strong galactose 1 promoter was chosen.
The putative CMP-sialic acid transporter cDNA, tagged with an HA epitope to facilitate the detection of the expressed protein, was cloned into pYES2 giving rise to pPB11. S. cerevisiae INVSc1 was transformed with pYES2 (vector alone) and pPB11. Total cell extracts were prepared from the transformants grown under uninduced conditions (4% raffinose) and following different times of induction with 2% galactose. As shown in Fig. 1, cells transformed with pPB11 when induced by galactose expressed a protein of approximately 30 kDa that strongly reacted with anti-HA antibody. This protein was not detected in the same cells prior to galactose induction and in cells transformed with the vector alone regardless of the presence of galactose. The mobility of the above immunoreactive protein (ϳ30 kDa) is lower than that of the predicted molecular mass of the HA-tagged mammalian CMP-sialic acid transporter (39 kDa). A total extract of COS-1 cells transfected with pME8-HA and subjected to SDS-polyacrylamide gel electrophoresis in parallel with the yeast samples showed a strong immunoreactive protein of the same mobility as in yeast (ϳ30 kDa). The difference between predicted molecular mass and the apparent one, based on mobility, is thus also observed when the protein is expressed in mammalian cells.
To determine whether the mammalian putative CMP-sialic acid transporter expressed in S. cerevisiae was functional, Golgi-enriched vesicles from uninduced and induced pPB11transformed cells were prepared, and transport of CMP-[ 3 H]sialic acid was measured into these vesicles. Table I shows that vesicles from induced cells transport CMP-sialic acid in a temperature-dependent manner with a 50-fold higher rate at 30°C than at 0°C; in contrast transport into vesicles derived from uninduced cells was not significantly different at both temperatures. Transport of CMP-sialic acid in the induced cells was dependent on vesicle integrity because Triton X-100 caused virtually complete absence. Transport of CMP-sialic acid was also inhibited by 5Ј-CMP as previously reported (12).
To determine the specificity of the CMP-sialic acid transport measurements including the substrate specificity of the CMPsialic acid transporter per se, transport of UDP-galactose into the same vesicles as described above was also measured. S. cerevisiae do not have galactose in their mannan chains and do not transport UDP-galactose into the Golgi lumen. As shown in Table I, the putative UDP-galactose transport signal was not dependent on temperature or vesicle integrity, both very important controls for true transport into vesicles. In addition, transport of CMP-sialic acid was not affected by the presence of 15 M UDP-galactose in the incubation medium (Table I) demonstrating that UDP-galactose was not competing with CMPsialic acid for entering into the vesicles.
Finally, it was important to determine whether transport of CMP-sialic acid into yeast vesicles was saturable, as previously found in mammalian Golgi vesicles. Fig. 2 shows that transport was saturable within an apparent K m of 2.9 M.

DISCUSSION
The following criteria provide strong evidence that we have expressed the murine Golgi membrane CMP-sialic acid transporter protein in S. cerevisiae. (a) A protein of apparent mobility of 30 kDa was expressed in S. cerevisiae only as a result of induction of the galactose 1 promoter. Induction of cells with plasmids without the cDNA did not result in expression of the above protein. The murine putative Golgi CMP-sialic acid transporter expressed in yeast cells has the same mobility as COS-1 cells that were transfected with pME8HA and expressed. (b) Upon isolation of a Golgi-enriched fraction from S. cerevisiae and measurements of CMP-sialic acid transport into these vesicles, the following characteristics of this transport activity were obtained. (i) Transport of CMP-sialic acid was temperature-dependent with the value at 30°C being 50-fold that at 0°C. (ii) Transport was dependent on time and protein concentration and was saturable with an apparent K m of 2.9 M, a value very similar to that previously described for rat liver Golgi membrane CMP-sialic acid transport (1). (iii) Transport was inhibited by 20 M of CMP in a manner analogous to the previously described transport of CMP-sialic acid into rat liver Golgi-derived vesicles (1); (iv) transport was specific for CMP-sialic acid, and S. cerevisiae Golgi-enriched vesicles from induced and uninduced cultures were unable to transport UDP-galactose. (v) Treatment of vesicles with Triton X-100 abolished transport. (vi) Following transport of CMP-sialic acid into the Golgi lumen, the concentration of solutes derived from this nucleotide sugar was concentrated 8 -10-fold in relation to the concentration in the reaction medium. Together, these results provide strong evidence that the protein encoded by the plasmid pMEB-HA that corrected the phenotype of the Chinese hamster ovary Lec2 mutant of mammalian cells (7) does indeed encode the Golgi membrane CMP-sialic acid transporter protein. The likelihood of the protein being an activator of the Golgi membrane CMP-sialic acid transporter protein is remote.
This heterologous expression also opens the possibility of expressing and purifying large amounts of the CMP-sialic acid transporter in other yeast and reconstituting this protein into proteoliposomes (12) as we have also done previously with the transporters for UDP-galactose (13), UDP-glucuronic acid (13), and adenosine 3Ј-phosphate,5Ј-phosphosulfate (12,14). Although reconstitution is a powerful direct proof that the protein is the transporter per se this approach has such problems that insertion of all the transporter protein molecules into liposomes FIG. 1. Expression of HA-tagged CMP-sialic acid transporter  by yeasts transformants and COS-1 transfectants. Extracts of transformants carrying pYES2 (Ϫ insert) or pPB11 (ϩ insert) grown in raffinose (uninduced (U)) and shifted to galactose for 4 and 8 h (4G and 8G) and extracts from pMEB-HA-transfected COS-1 cells (C, 1 g of protein; CЈ, 5 g of protein) were analyzed by immunoblotting using a monoclonal anti-HA antibody, as described under "Materials and Methods." may not be of the same topographical orientation as the protein in vivo; some molecules may be facing the liposomes "rightside-out" while others may be "inside-out," as was found to occur with sialyltransferases (12). Some transporter proteins may also be intercalated into the lipid bilayer without particular functionality. It is for these reasons that the in vivo insertion of the CMP-sialic acid transporter protein into S. cerevisiae vesicles, a yeast which neither has the transporter nor transports CMP-sialic acid naturally, is of major importance in organellar targeting and structure-function studies.
Subcellular fractions of galactose-induced cultures suggest that the majority of the CMP-sialic acid transporter protein is not in a Golgi vesicle fraction but in a fraction containing a mixture of vacuolar and endoplasmic reticulum-derived vesicles (not shown). This is not surprising in view of other studies in which mammalian Golgi proteins have been expressed in yeast and found to be functional even though the majority of the protein was not targeted to the predicted organelle (see below). We do not know whether the only functional expression of the CMP-sialic acid transporter occurred in the Golgi vesicles or also other organelles.
Mammalian Golgi proteins have been expressed in an active form in S. cerevisiae; so far active expression and correct targeting has apparently required the construction of chimeras between amino-terminal regions of either the rat liver ␣-1,6sialyltransferase attached to mature invertase (15) or the ami-no-terminal region from the yeast ␣-1,2-mannosyltransferase attached to the luminal domain of the human ␤-1,4-galactosyltransferase (16). In many instances, the transmembrane domain of mammalian and yeast Golgi glycosyltransferases appears to contain the necessary information for correct Golgi targeting of non-Golgi membrane proteins (17)(18)(19). Nevertheless, there are many exceptions to this in mammals and yeast (19,20). When the intact full-length human ␤-1,4-galactosyltransferase or ␣-2,3-sialyltransferase were expressed in S. cerevisiae, the active enzymes appeared to be retained in the endoplasmic reticulum (21). While the former studies (15,16) suggest that mammalian Golgi-targeting domains can be used in S. cerevisiae, the latter do not (21).
The apparent concentration within the lumen of S. cerevisiae vesicles of CMP-sialic acid, relative to its concentration in the reaction medium, deserves some attention. Assuming this observation is correct, where does the energy for concentration within the lumen come from? In mammals, studies in vitro have strongly suggested that the CMP-sialic acid transporter is an antiporter with CMP, the reaction product following sialylation of proteins and lipids (1,12). CMP-sialic acid is unique as a nucleotide sugar in that it does not need a nucleoside diphosphatase in the Golgi lumen to generate the nucleoside monophosphate as antiporter solute (1). In S. cerevisiae, where CMPsialic acid is not a donor in sialylation, the mechanism for CMP-sialic acid concentration in the vesicle lumen remains to be determined.
Is there an amino acid sequence motif for targeting these multitransmembrane proteins to the Golgi membrane in S. cerevisiae and mammals? To date, answers to this question are not available and the topography of such proteins is not known in the Golgi membrane. The expression system described here should be helpful in providing answers to these questions as well as determining the nucleotide sugar and nucleoside monophosphate binding sites in these proteins.  2. Rate of solute transport into S. cerevisiae vesicles versus CMP-sialic acid concentration in the incubation medium. A P 3 vesicle fraction (1 mg of protein) was incubated at 30°C for 3 min at different concentrations of CMP-[ 3 H]sialic acid in a final volume of 1 ml. Transport was measured (Si, solutes within vesicles) and calculated as described (11).