Transport of Sugar Nucleotides into Rat Liver Golgi A NEW GOLGI MARKER ACTIVITY*

Following incubation of sealed, "right side out," rat liver Golgi-derived vesicles with a mixture of [ 3 H]GDP-fucose and GDP-[' 4 C]fucose, the difference in the 3H to 14C ratio between the incubation medium and the lumen of the vesicles was less than 11%, suggesting that the sugar nucleotide was transported intact into the vesicles. Transport of GDP-fucose was temperature-de- pendent, saturable, with an apparent Km of 7.5 jM and a V,, of 14 pmol/mg of protein/10 min and inhibited by substrate analogues. Pretreatment of intact Golgi vesicles with pronase inhibited transport by 80% under conditions in which sialyltransferases (a lumenal marker) were not inhibited. This result is consistent with a sugar nucleotide carrier protein, portions of which face the cytoplasmic side of Golgi vesicles. Previous studies from this laboratory activity (2% yield), and 2.7-fold in 5' nucleotidase activity (5% endoplasmic reticulum-derived vesicles enriched 3.4-fold over homogenate in glucose-6-phosphatase activity (12% yield of total homogenate activity), 0.35-fold in sialyltransferase activity (1% yield), and 1.3-fold in 5' nucleotidase activity (5% and endoplasmic reticulum-derived as determined by latency of mannose-6-phosphatase

and is maintained during subsequent glycosylation in the Golgi (8)(9). It has been shown that if the permeability barrier of the Golgi membrane is disrupted, the glycosyltransferases catalyzing the transfer of sialic acid and galactose to proteins and lipids are inactivated by proteases; this suggests that the active sites of these glycosyltransferases are facing the lumen of the Golgi (8)(9)(10). Studies in vivo have shown that CMP-NeuAc', a substrate for sialyltransferases, is present in the lumen of mouse liver microsome mixtures of vesicles derived from the smooth and rough endoplasmic reticulum and Golgi (11). Since CMP-NeuAc and other sugar nucleotides are not synthesized in the Golgi lumen (21), the question arises of how such substrates become available to glycosyltransferases and glycoprotein acceptors which face the Golgi lumen. Previous studies in vitro from this laboratory (11) have demonstrated that mouse liver microsomes can transfer CMP-NeuAc from the incubation medium into a lumenal compartment in a manner suggesting carrier-mediated transport. The purposes of this study were to refine the location of the CMP-NeuAc transport process from a general microsomal location to a specific organelle and to obtain evidence for similar transport of other sugar nucleotides.

Synthesis and Purification of P3HGDP-fucose and [ 4 CJfucose
1-Phosphate GDP-fucose labeled in the guanosine moiety was prepared by enzymatic synthesis according to a procedure kindly obtained from Drs. J. A. Munro and H. Schachter (University of Toronto). The reaction mixture contained the following in a final volume of 7.5 ml: fucose -phosphate, prepared as described below (5 jamol); Tris-HC1, pH 8.0 (250 tmol); KF (62.5 Dpmol); MgCI 2 (31.25 mol); [ 3 H]GTP (7.5 pmol, 1,875 pICi); and GDP-fucose pyrophosphorylase, prepared as described below (4 ml). Following incubation for 4 h at 37 C, the reaction was stopped by addition of 2 volumes of 95% ethanol. The mixture was centrifuged at 23,000 x g and the supernatant solution was evaporated under reduced pressure. The residue was dissolved in water (100 ml), GDP-[' 4 C]fucose (0.7 Ci) added as marker, and the mixture purified on a Dowex 1-X2 chloride column (2.5 x 30 cm) equilibrated with 5 mM Tris-HCI, pH 7.4. The column was eluted with a linear gradient of 1 liter of 5 mM Tris-HCl, pH 7.4, and 1 liter of buffer containing 1.2 M KCI. The peak containing GDP-fucose was 'The abbreviations used are: CMP-NeuAc, CMP-N-acetylneuraminic acid; RER, rough endoplasmic reticulum; SER, smooth endoplasmic reticulum. pooled, evaporated, and desalted on a Sephadex G-10 column (1.5 x 45 cm), equilibrated in 50 mM NH 4 HCO 3 followed by preparative ascending chromatography on Whatman No. 3MM paper using ethanol:ammonium acetate (1 M, pH 7.5) (3:2) as solvent. One symmetrical peak was obtained which co-migrated with standard GDP-[' 4 C]fucose in the following systems: Whatman No. 3MM paper chromatography in ethanol:ammonium acetate (1 M, pH 7.5) (3:2); thin layer chromatography on polyethyleneimine (Brinkman Polygram CEL 300 PEI) in LiCI (1 M). The overall yield was approximately 12%. Upon digestion of the sugar nucleotide with nucleotide pyrophosphatase from Crotalus adamanteus all the radioactivity co-migrated with standard GMP in the above thin layer system. GDP-fucose pyrophosphorylase was purified from pork liver according to a procedure kindly obtained from Drs. H. J. Munro and H. Schachter (University of Toronto). Fresh liver (250 g) was homogenized in 0.25 M sucrose (1 liter, pH 7.4) and centrifuged at 23,000 X g for 20 min. The supernatant solution was made 30% saturation with ammonium sulfate. Following centrifugation at 23,000 x g for 20 min, the resulting supernatant solution was made 50% saturation with ammonium sulfate. The resulting pellet was dissolved in 0.03 M sodium phosphate, pH 7.4, containing 1 mM dithiothreitol, and was applied to a Sephadex G-100 column (5.0 x 90 cm) equilibrated in the same buffer. The column was eluted with the same buffer and fractions containing 12.5 ml were obtained. The color sequence of the eluent tubes was the following: tubes 1-30, clear, colorless; tubes 31-42, cloudy, dark reddish brown; tubes 43-56, clearing to light reddish brown; tubes 57-64, light straw yellow (GDP-fucose pyrophosphorylase region); tubes 65-72, reddish brown; tubes 73 and on, clearing through light yellow. The enzyme-rich fraction was used immediately.
[' 4 C]Fucose 1-phosphate was prepared by enzymatic cleavage of GDP-[' 4 C]fucose by C. adamanteus nucleotide pyrophosphatase. The product was purified using DEAE-paper chromatography in ethanol:ammonium acetate (0.2 , pH 7.5) (3:7). One symmetrical peak was obtained which co-migrated with standard fucose 1-phosphate. Upon digestion of this material with alkaline phosphatase all radioactivity co-migrated with fucose in the above system. The yield was 70%. Purity was also assayed using chromatography on Dowex-1formate columns. Fucose 1-phosphate was prepared from GDP-fucose (a generous gift to Drs. M. Hayes and R. Barker, Cornell University) as described above for the radioactive material.

Isolation and Topography of Vesicles Derived from Rat Liver Smooth and Rough Endoplasmic Reticulum and Golgi
The above three fractions were obtained as previously described by Carey and Hirschberg (12) using a modification of the subcellular fractionation procedure originally described by Fleischer and Kervina (13). Smooth endoplasmic reticulum-derived vesicles were enriched 3.5-fold over homogenate in glucose-6-phosphatase activity (6% yield of total homogenate activity), 1.2-fold in sialyltransferase activity (2% yield), and 2.7-fold in 5' nucleotidase activity (5% yield). Rough endoplasmic reticulum-derived vesicles were enriched 3.4fold over homogenate in glucose-6-phosphatase activity (12% yield of total homogenate activity), 0.35-fold in sialyltransferase activity (1% yield), and 1.3-fold in 5' nucleotidase activity (5% yield). Integrity of smooth and rough endoplasmic reticulum-derived vesicles, as determined by latency of mannose-6-phosphatase (17), was found to be 92% and 88%, respectively.

Transport Assay of Sugar Nucleotides
CMP-N-Acetyineuraminic Acid-Solutes whose penetration (transport) was to be measured were dried under a stream of nitrogen and, subsequently, dissolved in sufficient TKM buffer (10 mM Tris-HCI + 150 mM KCI + 1 mM MgCI2, pH 7.5) to give concentrations of 1-3 X 105 cpm/0.8 mi. Aliquots of 0.8 ml were then transferred into Ti 50 centrifuge tubes. Incubations were begun following the addition/ tube of 0.2-ml aliquots containing 1-2 mg of protein (in 0.25 M sucrose, 10 mM Tris-HCI, pH 7.0) from vesicles derived from rough or smooth endoplasmic reticulum or Golgi. Reactions were stopped by placing samples on ice. These were then centrifuged for 1 h at 40,000 rpm (100,000 x g) at 4 °C. Aliquots of the supernatant solution (0.1 ml) were removed for calculation of [Sm] (see below) or for further analyses of radioactive species in the incubation medium. Pellets were surface-washed three times, each with 1.5 ml of ice-cold TKM buffer.
Pellets were then suspended in water (0.5 ml) and sonicated for 30 min in a Heat Systems Ultrasonics sonicator at room temperature. Perchloric acid (8%, 0.5 ml) was added to the pellet, vortexed, and the mixture was allowed to remain on ice for 15 min. The suspension was centrifuged at 15,000 rpm for 15 min at 4 C. A 0.5-ml aliquot was removed for determination of St (see below). When analyses of soluble radioactive species within pellets were to be performed, ethanol was used instead of percholoric acid. For determination of perchloric acidinsoluble radioactivity in the pellet, the pellets were surface-washed three times, each with 1.5 ml of 4% perchloric acid. Following sonication for 1 h in 2 mil of 4% perchloric acid, the samples were centrifuged at 15,000 rpm for 15 min at 4 C. The supernatant solution was removed and the pellet surface-washed once with 1.5 ml of water. NaOH (1 N, 1 ml) was added to dissolve the pellet. Following neutralization, all samples were counted in Aquasol-2 (New England Nuclear).
GDP-Fucose-Transport of this sugar nucleotide was measured as described for CMP-NeuAc with the following modifications: 1) the incubation mixture contained 0.5 mM 2,3-dimercaptopropanol (an inhibitor of nucleotide pyrophosphatase (22)) in TKM buffer; 2) the reaction was stopped by adding 5'-AMP (4 mM, 0.25 ml) in TKM buffer containing 0.625 mM dimercaptopropanol; and 3) perchloric acid (8%) was used when analyzing soluble radioactive species in the pellet. Calculations-The calculations used in these experiments are given below.
[Sm] = concentration of solute in the incubation medium (M) = (cpm/ml of solute in the (supernatant)/(specific activity of solute, expressed as cpm/nmol).
St = soluble radioactivity/mg of protein in pellet (pmol) = (total soluble radioactivity associated with pellet, expressed as cpm/mg of protein)/(specific activity of solute, expressed in cpm/pmol). Vt = total pellet volume (pl) = volume outside + inside vesicles = (cpm/g of protein in pellet for deoxyglucose)/(cpm/pl of supernatant for deoxyglucose).

Transport of GDP-Fucose into Golgi
Vesicles-It has previously been shown in this laboratory that CMP-NeuAc can penetrate liver microsomes in vitro in a manner suggesting carrier-mediated transport (11). CMP-NeuAc is the onlysugar nucleotide with a monophosphate group; all other sugar nu-cleotides contain diphosphate linkages. It was, therefore, of interest to determine whether a similar transport system existed for these sugar diphosphate nucleotides.
Studies in vitro and in vivo have shown that fucose is transferred to glycoproteins in the Golgi (19,20). This transfer is thought to occur in the lumen. GDP-fucose, which is synthesized in the cytoplasm (21), could be expected to be transported into the Golgi lumen via a carrier protein system similar to that suggested for CMP-NeuAc. The following experiments were designed to substantiate this hypothesis. A fraction highly enriched in Golgi-derived vesicles, as determined by enrichment in marker enzymes (see "Materials and Methods") was used for subsequent studies. At least 90% of these vesicles were sealed and of the same membrane topography as in vivo. This was determined by measuring latency of neuraminidase catalyzed removal of labeled sialic acid from Golgi vesicles prelabeled with radioactive CMP-NeuAc (11; see "Materials and Methods").
Incubation of GDP-[1 4 Cfucose with Golgi vesicles resulted in the accumulation of radioactive solutes within such vesicles ( Fig. 1). The total amount of solutes inside the vesicles, the penetration index S, was determined by subtracting the total solutes outside vesicles in the pellet, SO, from the total solutes (inside and outside) of the vesicle pellet, St. Detailed calculations are described under "Materials and Methods." This accumulation was linear with time and protein (see below) and was saturable with an apparent Km of 7.5 zm (based on the initial concentration of GDP-fucose in the incubation medium) and a Vma, of 14 pmol/10 min/mg of protein. Transport at 0 °C was less than 10% of that at 23 °C (not shown; see Table V).
Chemical analyses of the radioactive solutes in the incubation medium and Golgi-vesicle pellet, following a 10-mm incubation, showed that both compartments contained GDP-[ 14 C]fucose, [1 4 C]fucose, and [ 4 C]fucose 1-phosphate, although each component was present in strikingly different ratios. The major radioactive species in the reaction medium was fucose 1-phosphate (72% of total); however, it was only 15% of the soluble radioactive species in the Golgi pellet. The major soluble radioactive species of the Golgi pellet was free fucose (80% of total); however, it was only 3% of the radioactive species in the reaction medium. GDP-fucose was 25% of the radioactivity in the reaction medium and 5% of the Golgi pellet.
For each of the above solutes of the Golgi pellet, we determined their penetration index, Si, and their concentration within Golgi vesicles and the reaction medium (see "Calculations" under "Materials and Methods"). As shown in Table I, Experiment 1, only fucose was accumulated to a significant extent within Golgi vesicles (over 100-fold). Radioactive fucose 1-phosphate and GDP-fucose in the Golgi pellet could be accounted for as solutes trapped between vesicles. The accumulation of fucose within Golgi vesicles could have been the result of transport of GDP-fucose from the medium and subsequent breakdown within vesicles; however, it could also have arisen from transport of free fucose or fucose 1phosphate, following exogenous breakdown of GDP-fucose.
To determine to what extent this latter event was contributing to the accumulation of free fucose within vesicles, Golgi preparations were incubated with fucose or fucose 1-phosphate at concentrations comparable to those found in the incubation medium at the beginning and end of the reaction described in Table I, Experiment 1. Table I, Experiment 2, shows that no large increase in concentration of radioactive solutes occurred within Golgi vesicles, compared to the incubation medium, regardless of the initial concentrations of fucose and fucose 1phosphate in the medium. This strongly suggests that the

Transport of GDP-[ 4 C]fucose into Golgi vesicles: concentration of radioactive solutes within vesicles following a 10-min incubation
Golgi vesicles (1-2 mg of protein/ml) were incubated with the above solutes for 10 min at 23 C as described under "Materials and Methods" and the picomoles of radioactive solute inside the Golgi vesicles determined. Radioactive species in Experiment were separated using Dowex-2 formate columns.

Transport of a mixture of [H]GDP-fucose and GDP-[' 4 C]fucose or [ 4 C]GMP and [H]fucose into Golgi vesicles: isotope ratios inside and outside vesicles following a 10-min incubation
Golgi-derived vesicles were incubated for 10 min at 23 C as described under "Materials and Methods." Counting efficiencies were determined using internal standards of radioactive toluene (New England Nuclear). Numbers obtained for a typical incubation as in Experiment 1 were: medium following incubation = 52,930 dpm of 8 H and 10,940 dpm of 4 C; Golgi pellet = 9,940 dpm of 3 H and 1,810 dpm of 4 C. Numbers obtained for a typical incubation as in Experiment 2 were: medium following incubation = 6,090 dpm of ' 4  C]fucose into Golgi vesicles: concentration of radioactive solutes within vesicles following a 10-min incubation Incubations using Golgi-derived vesicles (2 mg/ml of protein) were done as described under "Materials and Methods." Soluble radioactive species were separated using Dowex-2 formate columns.
Radioactive solutes following 10-Concentration of min incubation, Per cent of total of radioactive soleach radionuclide, mean + S.  (Table III). Their enrichment in concentration within the Golgi vesicles relative to the medium were not equal because their concentrations in the medium following a 10min incubation are different, and a significant portion of the fucose has become covalently bound to Golgi proteins (see below).
To determine whether the free fucose and GMP within the Golgi were due to the transport of free fucose and GMP fro'h the medium (following exogenous breakdown of GDP-fucose), GMP, and fucose, labeled with different isotopes, were incubated with Golgi vesicles at concentrations similar to the double labeled sugar nucleotide used in Table II, Experiment 1. As can be seen in Table II, Experiment 2, the isotope ratio of the Golgi pellet was strikingly different from that of the medium following a 10-min incubation (-12-fold). This result also suggests that GMP can penetrate Golgi vesicles at considerably higher rates than fucose. This is consistent with preliminary evidence for carrier-mediated transport by Golgi vesicles of nucleoside monophosphates (30). The mechanism for penetration of fucose into Golgi is (as yet) unknown.
GMP and free fucose in the Golgi lumen (Table III) most likely arise from a combination of 1) transfer of the fucose moiety from GDP-fucose to glycoprotein and glycolipid acceptors, 2) transfer of the sugar to water, and 3) a lumenal nucleoside diphosphatase (24) and phosphatase.
The above results, together with the previous ones showing lack of significant transport by Golgi vesicles of free fucose or fucose 1-phosphate (Table I), strongly suggest that GDPfucose is transported intact by Golgi vesicles, even though the intact sugar nucleotide could not be detected within such vesicles. Although the possibility of cleavage of the sugar nucleotide on the cytoplasmic surface of the Golgi vesicles followed by a simultaneous translocation of GMP and fucose (perhaps as an energy-rich intermediate) across the Golgi membrane cannot be ruled out, such translocations would have to be occurring at similar rates in order to maintain a H to 4 C ratio in the medium within 11% of that of the Golgi vesicles (Table II, Experiment 1). This appears to be unlikely.
In addition to the soluble fucose derived from GDP-fucose within the lumen of the vesicles, a portion of fucose has been transferred covalently to glycoproteins. In a representative experiment, 59% of the radioactivity in the pellet (derived from GDP-[ 14 C]fucose) was perchloric acid-insoluble following a 10-min incubation at 23 C. Evidence suggesting a lumenal orientation of this macromolecular-bound radioactivity was obtained by the failure of pronase to solubilize the radioactivity unless the permeability barrier of the membrane has been disrupted with detergents (8, not shown). Mammalian and bacterial fucosidases (kindly obtained from Dr. D. Aminoff, University of Michigan) were inactive in solubilizing macromolecular bound fucose.
On the assumption that a cytoplasmic-facing carrier protein is involved in GDP-fucose transport, intact Golgi vesicles were pretreated with pronase and sugar nucleotide transport measured. Under conditions in which transport was inhibited by more than 79%, the activity of sialyltransferase (presumably a lumenal protein marker (8-10)) was not decreased (Table  IV). Higher concentrations of pronase inactivated both activities (not shown). While this experiment suggests that at least a portion of the putative carrier protein faces the cytoplasm, it does not rule out the possibility that both activities being assayed have the same membrane topography but different susceptibilities to proteases. Transport of GDP-fucose was inhibited by GMP (Table VII).  Transport of CMP-N-Acetylneuraminic Acid into Golgi Vesicles-Previous studies from this laboratory showed that CMP-NeuAc could penetrate liver microsomes in vitro (11). These microsomes contained a mixture of vesicles derived from the smooth and rough endoplasmic reticulum and Golgi. It was, therefore, important to determine whether transport of CMP-NeuAc was occurring in one or more types of the above mentioned vesicles. Fractions highly enriched in vesicles from the above organelles were obtained (see "Materials and Methods"). At least 90% of each of the three types of vesicles used in subsequent transport studies were sealed and of the same membrane topography as in vivo. This was determined by measuring latency of mannose-6-phosphatase for smooth and rough endoplasmic reticulum-derived vesicles and latency of neuraminidase catalyzed removal of labeled sialic acid from Golgi vesicles prelabeled with radioactive CMP-NeuAc (see "Materials and Methods").
The above three types of vesicles were then incubated separately with CMP-NeuAc for 10 min at 0 and 23 C and reisolated. As shown in Table V, columns 2 and 3, the penetration indices at 23 and 0 C were approximately 30 times higher for Golgi than for SER-derived vesicles at the same temperature. The penetration index of SER vesicles was in turn four times higher than that of the RER-derived vesicles at 23 C. From the penetration index and the volume within vesicles, the concentrations of solutes within vesicles were calculated. As shown in Table V, columns 4 and 5, only Golgiderived vesicles were able to concentrate solutes from the incubation medium, and this occurred to a much larger extent (over 90%) at 23 than at 0 C. The specific transport activity has, therefore, been defined as the difference between penetration indices at 23 and 0 C (Table V, column 6). Golgi vesicles were enriched approximately 25-fold in this activity over SER-derived vesicles; RER-derived vesicles were inactive. The total transport activity/organelle without correction for contamination with other organelles was then calculated (Table V, column 7). Eighty-seven per cent of the uncorrected CMP-NeuAc transport activity of the total homogenate was in the Golgi and 13% in the combined SER-and RER-derived fraction. The question of whether the transport activity in these latter fractions could be accounted for by contamination with Golgi was then addressed. Based on the units of sialyltransferase activity (a Golgi marker) contaminating the isolated SER-and RER-derived fractions, these fractions would be expected to contain 4 pmol/10 min of CMP-NeuAc transport activity. This is within experimental error of the 3.4 pmol/10 min of CMP-NeuAc transport activity which was actually measured in these fractions (Table V, column 7). These results strongly suggest that all the CMP-NeuAc transport activity is in the Golgi-derived vesicles and that the activity detected in SER-and RER-derived vesicles can be fully accounted for by Golgi contamination of these organelles (Table V,

column 8).
A comparison of the specific and total activities of sialyltransferases and CMP-NeuAc transport among the three types of vesicles is shown in Table VI. The relative enrichment of Golgi vesicles in both activities was very similar. As these values were obtained independently from one another, these results strongly suggest that, for this particular subcellular fractionation procedure, both activities were copurifying in

Transport of CMP-NeuAc into rat liver vesicles derived from the smooth and rough endoplasmic reticulum and Golgi
Rat liver (20 g) was subfractionated into vesicles derived from RER, SER, and Golgi according to the procedure of Carey and Hirschberg (12). Integrity and topography of the vesicles, as well as marker enzyme activities, were determined as described under "Materials and Methods." Calculations and determinations of penetration indices were done by incubating subfractions (1 mg/ml of protein) with CMP-NeuAc as described under "Materials and Methods." Values obtained by multiplying the specific transport activity of each fraction by the mg of protein recovered of each (column 1) and dividing by the yield of the corresponding marker enzyme activity recovered for each fraction (18.1% of total homogenate glucose-6phosphatase activity for SER-and RER-derived vesicles and 13.6% of total homogenate sialyltransferase activity for Golgi.) d Means of two separate determinations.

TABLE VI
Relative activities of sialyltransferase and CMP-NeuAc transport in rat liver vesicles derived from the smooth and rough endoplasmic reticulum and Golgi Rat liver vesicles were prepared according to the method of Carey and Hirschberg (12) and analyzed for their sialyltransferase and CMP-NeuAc transport activities as described under "Materials and Methods." Specific activities shown are the averages of two separate determinations. Total activities were obtained by multiplying specific activities by milligrams of protein recovered in each fraction and dividing by the yield of corresponding marker enzyme in each fraction.   (14) were also highly active in CMP-NeuAc transport, as shown in Table V, column 4. Characterization of CMP-NeuAc Transport into Golgi Vesicles-The above results showed that transport of CMP-NeuAc was occurring solely into Golgi-derived vesicles and not into vesicles derived from the smooth or rough endoplasmic reticulum. Transport of CMP-NeuAc into Golgi at 0 °C was less than 10% of that measured at 23 C (Table V); therefore, transport in subsequent experiments was measured only at this latter temperature since the correction for 0 C was very small. Whether the penetration index at 0 C represents true transport, adsorption, or a combination of both events is not clear. Transport of CMP-NeuAc into Golgi vesicles was linear with time for at least 10 min and was linear with protein concentrations between 0.4 and 4.0 mg/ml (not shown).
Transport of CMP-NeuAc into Golgi was saturable with increasing concentrations of CMP-NeuAc in the incubation medium (not shown); the apparent Km was 2.4 pM (based on the initial concentration of CMP-NeuAc in the incubation medium) and the Vma~ was 150 pmol/10 min/mg of protein.
As with GDP-fucose, the transport of CMP-NeuAc was inhibited (60%) by pronase pretreatment of the Golgi vesicles while sialyltransferase activity was not affected (Table IV)) thus suggesting the involvement of a cytoplasmic-facing carrier protein in CMP-NeuAc transport across the Golgi membrane. Transport of CMP-NeuAc was inhibited by substrate analogues such as 5'-CMP (Table VII, Experiment 1).

DISCUSSION
The results of this paper demonstrate that Golgi membranes can transport CMP-NeuAc and GDP-fucose from the incubation medium into their lumen. As the vesicles used in this study are of the same topographical orientation as in vivo (8) and the above sugar nucleotides are not synthesized in the Golgi (21), the in vitro transport system described here is postulated to be the same system that translocates the sugar nucleotides in vivo.
Th e transport of CMP-NeuAc into a mixture of vesicles derived from the RER, SER, and Golgi has been previously described (11); in these studies the transport of CMP-NeuAc from the medium into RER and SER vesicles followed by the fusion of these endoplasmic reticulum-derived vesicles with existing Golgi elements could not be ruled out. The results shown in Tables V and VI strongly suggest that the transport activity for CMP-NeuAc is contained only in Golgi vesicles; therefore, the measurements of such transport activity can be used as a marker for Golgi vesicles. It had also been previously shown, using CMP-NeuAc labeled in both the sugar and nucleotide moieties, that the plasma membrane was impermeable to this sugar nucleotide (34).
It must be emphasized, however, that for the above Golgi transport measurements to be valid, one must first prove that the vesicle population being assayed is sealed and "right side out"; i.e. it must have the same topography as in vivo. Studies in this laboratory (8) as well as those of Howell et al. (23) have previously shown that freezing of such vesicles prior to transport assays can lead to vesicle breakage unless precautions, e.g. bovine serum albumin (10 mg/ml) in the buffer, are taken.
The Golgi specificity and the observations that transport is temperature-dependent, saturable, and inhibited by substrate analogues and proteases (under conditions which presumably only cleave cytoplasmic proteins in the Golgi vesicles (8)) all strongly suggest that the transport has a protein component, e.g. sugar nucleotide carrier protein(s).
The apparent K, (2.4 M) and Vma,, (150 pmol/10 min/mg of protein) obtained for CMP-NeuAc transport into Golgi vesicles are in good agreement with those values previously obtained using liver microsomes, a mixture of vesicles derived from the endoplasmic reticulum and Golgi (11); in these vesicles the apparent K, for CMP-NeuAc transport was 0.6 gM and the Vmax was 12 pmol/10 mmin/mg of protein. Since approximately 10% of this microsomal protein was due to Golgi, one would expect a Vma for a Golgi-enriched fraction to be in the range of 120 pmol/10 min/mg of protein. This value is in close agreement with the 150 pmol/10 min/mg of protein found here.
It has been previously shown in this laboratory, using the sugar nucleotide labeled in both the sugar and nucleotide moiety, that CMP-NeuAc was transported intact into microsomes (11). This fact, in light of the current findings that only Golgi vesicles within microsomes are active in transport, strongly suggests that CMP-NeuAc is transported as a unit into Golgi vesicles.
Previous studies with Golgi have shown that a portion of the neuraminic acid (derived from CMP-NeuAc) was transferred to glycoproteins and that the covalently linked Nacetylneuraminic acid was oriented toward the lumen of such vesicles (8). In a representative experiment with Golgi vesicles, after a 10-min incubation at 23 C, 83% of the radioactivity in the pellet was covalently bound, the remainder being soluble. Previous studies also suggested that the active sites of sialyltransferases were oriented toward the lumen of Golgi vesicles (8)(9)(10).
In the present studies with Golgi, as well as in previous ones with microsomes, it was found that a portion of CMP-NeuAc (up to 40%) had broken down in the reaction medium to CMP and N-acetylneuraminic acid (11). In both microsomes and Golgi vesicles the soluble species derived from CMP-NeuAc within vesicles were CMP-NeuAc, cytidine, and NeuAc (11). These latter two compounds most likely arise from a combination of 1) transfer of the sugar moiety from the sugar nucleotide to appropriate glycoprotein and glycolipid acceptors, 2) transfer of the sugar to water (hydrolysis), and 3) a lumenal phosphatase activity. The Golgi membrane is impermeable to both N-acetylneuraminic acid and cytidine (11).
It has also been shown in this study that GDP-fucose enters Golgi vesicles via a transport mechanism which is temperature-dependent, saturable, and inhibited by substrate ana-logues and protease pretreatment as described above. The studies with GDP-fucose labeled in both the sugar and nucleotide moieties also strongly suggest that the sugar nucleotide is transported intact into the Golgi lumen in a manner analogous to CMP-NeuAc. Following transport, a portion of the sugar moiety is transferred to glycoproteins while the remainder of the sugar nucleotide is hydrolyzed to GMP and free fucose. It is not known at this time whether this latter reaction is catalyzed by a pyrophosphatase and phosphatase or by a fucosyltransferase acting as a hydrolase in the absence of appropriate acceptors. As shown in Table II, the enrichment of GMP and fucose within the Golgi lumen over that in the incubation medium cannot be accounted for by transport of GMP, fucose, or fucose 1-phosphate.
Concentrations of GDP-fucose up to 250 gM did not inhibit CMP-NeuAc transport (at 2 pM), suggesting that the two sugar nucelotides are not being transported by the same carrier protein. In addition, these two carrier proteins must have recognition entities facing the cytoplasmic side of the Golgi membrane; prior to this, only proteins with recognition sites facing the Golgi lumen (glycosyltransferases (8-10) and thiamine and nucleoside diphosphatase (24-26)) had been described.
In contrast to CMP-NeuAc, significant transport of GDPfucose into microsomes containing a mixture of vesicles derived from the RER, SER, and Golgi was not detected (data not shown), probably because the Vm~ for GDP-fucose transport into Golgi is approximately 10-fold less than that of CMP-NeuAc. Kuhn and White (27,28) and Kuhn et al. (29) have previously presented indirect evidence for transport of UDP-galactose by rat mammary gland Golgi membranes. These investigators, as well as Brandan and Fleischer (30), have also shown transport of 5'-UMP by Golgi vesicles.
The above studies should encourage further investigation into the localization of other sugar nucleotide transport activities in functional regions of the Golgi. Although highly speculative, the carrier proteins for UDP-galactose and UDP-GlcNAc could logically be envisioned in Golgi regions closer to the forming face (cis), while the carrier proteins for the terminal sugars GDP-fucose and CMP-NeuAc could be closer to the mature face (trans). Bretz et al. (31) have found 3-fold differences in specific activities of glycosyltransferases in different Golgi vesicles whose isolation was based on differences in density; however, total activity measurements of the glycosyltransferases showed no differences within the different vesicle populations. Another interesting question raised by these studies is whether the transport into endoplasmic reticulum-derived vesicles is required for UDP-glucose, GDP-mannose, and UDP-GIcNAc, sugars which are added via dolichol to asparagine-linked nascent polypeptide chains. Hanover and Lennarz (32) have presented evidence showing that in rough microsomes chitobiosyl-PP-dolichol is localized toward the lumen. One possible explanation for such a topographic arrangement would be that RER membranes transport UDP-GlcNAc into their lumen. It is not clear at this time whether GDP-mannose and UDP-glucose must also enter the lumen of the RER prior to sugar transfer to dolichol oligosaccharides. Some of the enzymes involved in such transfers can be inactivated with proteases without apparent disruption of the permeability barrier of the vesicles (33), suggesting that their active sites may be facing the cytoplasm. In such case, no intralumenal transport for the above sugar nucleotides would be necessary. Clarification of this problem as well as further characterization of the above described transport activities are needed.