Human Fibroblasts Prefer Mannose over Glucose as a Source of Mannose for N-Glycosylation

Mannose in N-linked oligosaccharides is assumed to be derived primarily from glucose through phosphomannose isomerase (PMI). The discovery of mammalian mannose-specific transporters that function at physiological concentrations suggested that mannose might directly contribute to oligosaccharide synthesis. To determine the relative contribution of glucose and mannose, human fibroblasts were labeled with either [2-3H]mannose or [1,5,6-3H]glucose at the same specific activity, and theN-linked chains were released by PNGase F digestion. Most of the trichloroacetic acid-precipitable [3H]mannose label was released by this digestion, but only about 10% of the trichloroacetic acid-precipitable material was released from cells labeled with [1,5,6-3H]glucose. Both sugars labeled a similar array of oligosaccharides, and acid hydrolysis of these chains showed that [2-3H]mannose contributed 65–75% of the [3H]mannose in cells labeled for 1 h, despite the 100-fold higher concentration of exogenous glucose. Mannose consumption and [2-3H]mannose utilization were within the range of rates expected for mannose transport via the mannose-specific transporter. About 7–14% of the [2-3H]mannose is used for glycosylation, while the rest (86–93%) is catabolized to3H2O via PMI. Increasing the exogenous mannose concentration beyond mannose transporter saturation results in the conversion of >99% of [2-3H]mannose into3H2O. Long term labeling of cells with [2-3H]mannose showed that the specific activity of mannose in glycoproteins reached 77% of the specific activity of [2-3H]mannose added to the medium. These results show that when fibroblasts are provided with physiological concentrations of mannose, they use the mannose-specific transporter to supply the majority of mannose needed for glycoprotein synthesis. PMI may normally be used to catabolize excess mannose rather than to primarily supply Man-6-P for glycoprotein synthesis.

Phosphomannose isomerase (PMI) 1 (Fru-6-P 7 Man-6-P) plays a pivotal role in the second route, and it has long been assumed that most, if not all, of the mannose in macromolecules is derived from glucose. This assumption is based on the universal distribution of PMI (5), and the fact that this enzyme is essential for the survival of yeast grown in the absence of mannose (6). [2-3 H]Mannose is practically the only isotope used to specifically label newly synthesized N-linked oligosaccharides (3), since its diversion into any other metabolic pathway begins with PMI, which generates 3 H 2 O. This reaction is irreversible in terms of metabolic labeling, since the product is immediately diluted into 55.5 M water (3). In contrast, [ 3 H]glucose can be catabolized through glycolysis and the tricarboxylic acid cycle to generate energy or used to synthesize a wide variety of metabolic intermediates including sugars, amino acids, and fatty acids (7,8). The conspicuous lack of selective incorporation into sugar chains explains why [ 3 H]glucose is rarely used to label them.
Normal blood levels of mannose have been measured at 50 -100 M in a few mammals (9 -12), but its potential contribution to glycoprotein synthesis has not been investigated. This is probably due to the assumption that its entry into cells via the common glucose (GLUT) transporters is competed by the 100-fold higher levels of blood glucose (13). However, we recently identified a mannose-specific transporter with a K uptake of 35-70 M that can supply mannose for glycoprotein synthesis under physiological conditions of 5.0 mM glucose and 50 M mannose (14). This finding prompted us to ask which of these sugars is preferred as a source of mannose for glycoprotein biosynthesis when both are present at their physiological levels.
We find that mannose is highly preferred over the 100-fold greater concentration of glucose for N-linked oligosaccharide synthesis, underscoring the functional significance of the mannose-specific transporter.
Cell Lines-Normal adult fibroblast cultures were obtained from American Type Culture Collection and grown in a medium (CRL 1947 * 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  in Dulbecco's modified Eagle's medium; CRL 1828 in ␣-minimum essential medium) containing 10% heat-inactivated fetal bovine serum and 2 mM glutamine.
In the initial experiments, aliquots of the cell lysates were digested with or without PNGase F, precipitated with trichloroacetic acid, washed, and counted after solubilization in 0.1% SDS as described before (15).
Analysis of Labeled Sugars in N-Linked Oligosaccharides-To analyze the released labeled sugar chains, the cell lysates were precipitated with trichloroacetic acid, solubilized in 0.1% SDS, and chromatographed on a Sephadex G-50 column. Labeled proteins were collected from the void volume, acetone-precipitated, solubilized in 0.1% SDS, and digested with PNGase F to release the oligosaccharides (15). The digests were again run on a G-50 column in 50 mM Tris-HCl buffer, pH 7.5, containing 0.1% SDS to collect the released oligosaccharides. These samples were passed through a C18 cartridge to remove the detergents; or SDS was precipitated with KCl, and the sample was desalted on Biogel P2 column.
To account for the possible loss of radioactivity during sample preparation, all of the steps were also carried out in a single tube. The percentage of radioactivity released from the cell lysate by PNGase F digestion was the same using both methods.
Monosaccharide Analysis-3 H-Labeled oligosaccharides were hydrolyzed in 4 M trifluoroacetic acid at 100°C for 4 h, and the acid was removed by evaporation. Labeled amino sugars (glucosamine plus galactosamine) were measured by binding to a Dowex 50 cation exchange column. The amount of each hexose was determined by treating the sample with specific enzymes followed by QAE-Sephadex anion exchange chromatography to determine the amount of neutral and anionic material. All enzymatic treatments were carried out in a 100-l volume of 100 mM Tris-HCl at 37°C for 2 h. Kinase reactions were done in the presence of 5 mM MgCl 2 and 2 mM ATP. Total hexose content was estimated by incubating the sample with 1 IU of hexokinase (EC 2.7.1.1; type C-300 from Bakers' yeast) at pH 8.0. The amount of glucose was measured by incubation with 1 IU of glucokinase (EC 2.7.1.2 from Bacillus stearothermophilus) at pH 8.5. The amount of galactose was determined as 3 H label that remained neutral after sequential treatments with galactose oxidase (EC 1.1.3.9 from Dactylium dendroides) at pH 7.2 and hexokinase. Standard sugar mixtures were analyzed simultaneously.
Paper Chromatography-The neutral monosaccharides were also analyzed by paper chromatography using Whatman 1 MM paper and developed for 18 h with ethyl acetate/pyridine/butanol/butyric acid/ water (10:10:5:1:5, v/v). Standards were visualized by silver nitrate staining, and strips containing radioactive samples were cut into 1-cm strips and counted (17). 3 H 2 O Determination-An aliquot of the medium used to radiolabel the cells was evaporated to dryness, suspended in water, and counted to determine radioactivity remaining in the medium. The amount of 3 H 2 O formed was taken as the difference between the initial amount of radioactivity in the medium and that remaining after evaporation plus cell lysate radioactivity.
Mannose Assay-Mannose in the labeling medium was determined by the method of Etchison and Freeze (18). Glucose in the medium was assayed using a modification of the glucokinase procedure described in Ref. 18.
Determination of Specific Activity of [2-3 H]Mannose Incorporated into Glycoconjugates-Radiolabeled cells were harvested by scraping and centrifugation at 1000 ϫ g for 5 min. The cells were washed three times with isotonic saline and then lysed in deionized water at a concentra-tion of 1-5 mg/ml. Radiolabeled glycoconjugates were precipitated with 4 volumes of 2-propanol at Ϫ20°C overnight. The precipitate was washed twice with 2-propanol and resuspended in 70% ethanol by sonication. A portion containing 0.1-0.5 mg of protein was dried in vacuo and hydrolyzed with 100 l of 2.5 M trifluoroacetic acid at 100°C for 4 h under nitrogen. The hydrolysate was dried in vacuo, resuspended in 200 l of 0.1 NH 4 HCO 3 , and deionized by passing through a 0.5-ml bed of Dowex 501 mixed bed resin. The effluent containing neutral sugars was dried in vacuo, and a portion was analyzed by HPAEC-PAD on a CarboPac PA1 column by the Glycobiology Core Facility at the University of California, San Diego, Cancer Center. The column effluent was collected, and fractions were assayed by liquid scintillation counting. The specific activity of [ 3 H]mannose from hydrolysates was determined using a standard curve generated by analyzing [ 3 H]mannose standards (10 -250 mCi/mmol) under identical conditions.

Incorporation of [ 3 H]Mannose and [ 3 H]Glucose into Glycoproteins-Mannose and glucose can both be converted into
Man-6-P and then incorporated into sugar chains of glycoproteins ( Fig. 1). Assuming that both labeled sugars enter a common pool of Man-6-P that rapidly equilibrates with the exogenous label, the radioactivity in mannose from each source should be proportional to the contribution from the pathways. To determine which sugar is preferred for N-glycosylation, we incubated identical cultures of human fibroblasts with 5.0 mM glucose and 50 M mannose containing either 1 mCi/ml of [ 3 H]glucose or 10 Ci/ml of [ 3 H]mannose for 1 h. The labeled cells were washed, lysed, and precipitated with trichloroacetic acid either before or after digestion with PNGase F to release the N-linked chains (15). Results from multiple labelings (Table I) show that about 40% of the total cell-associated [ 3 H]mannose is trichloroacetic acid-precipitable and that PNGase F digestion releases 80 -90% of it, in agreement with the results of previous labeling studies using 1 M mannose and 0.5 mM glucose (15). In contrast, only 10% of the label from cells incubated with [ 3 H]glucose is trichloroacetic acid-precipitable, and of this, only 1-2% is released by PNGase F digestion. These results suggested that little of the [ 3 H]glucose was incorporated into any of the sugars found in N-linked chains and that mannose was highly preferred over glucose as a precursor.

Isolation and Analysis of N-Linked Chains Labeled with [ 3 H]Mannose and [ 3 H]
Glucose-Accurate determination of the relative contributions by each labeled precursor required the isolation and analysis of radiolabeled oligosaccharides. A critical assumption for quantitative measurements using these radiolabels is that the intracellular precursor pools have the same specific activity as the exogenous label. To insure equilibration, labeling was extended to 6 and 16 h. The labeled cell lysates were sequentially precipitated with trichloroacetic acid and acetone and then digested with PNGase F to release the N-linked oligosaccharides as described under "Experimental Procedures." Fig. 2 shows representative elution profiles of samples with and without digestion. About 75% of the radioactivity in [ 3 H]mannose-labeled protein was released, but only about 10% of the label was released from [ 3 H]glucose-labeled protein.
The released chains from samples at each time point were analyzed by anion exchange chromatography on QAE-Sephadex (Table II). Essentially all of the N-linked chains labeled for 1 h were neutral. This was consistent with previous results (15,19) showing that they are nearly all high mannose, not complex-type, oligosaccharides. At longer labeling times, a higher proportion of the oligosaccharides labeled with each sugar were anionic, and all of the charge was due to sialic acids, since they were neutralized by digestion with sialidase (data not shown). 3 H-Labeled chains were also analyzed by lectin affinity chromatography on ConA-Sepharose (Table II) and separated into tri-and tetraantennary oligosaccharides (fraction I), primarily biantennary chains (fraction II), or typical high mannose type species (fraction III) (16). At 1 h, the [ 3 H]mannose-labeled chains eluted exclusively in fraction III, which indicates that they are high mannose type and is consistent with their lack of sialic acid. Longer incubations yielded an increasing proportion of chains in fraction II consistent with the appearance of sialylated complex type chains as previously reported (19 (15,19).

Analysis of Monosaccharides Labeled with [ 3 H]Mannose and [ 3 H]Glucose-
The released oligosaccharides were hydrolyzed, and the labeled monosaccharides were quantitatively analyzed by both specific enzymatic treatments and by paper chromatography. Both methods gave similar results. None of the 3 H label derived from either sugar bound to QAE-Sephadex; however, after incubation with hexokinase and ATP, nearly all (ϳ95%) bound, showing that the labels were present entirely in hexoses or hexosamines that had been converted into their 6-phosphate derivatives by hexokinase. Glucokinase analysis (20) showed that only a small amount (3-8%) of [ 3 H]glucose remained as glucose. The amount of label in galactose was determined by first treating the mixture with galactose oxidase and then hexokinase followed by QAE-Sephadex analysis. This treatment specifically oxidizes galactose at the C-6-position, forming an aldehyde that cannot be converted into galactose-6-P. If [ 3 H]galactose is present, this treatment is expected to increase the amount of residual neutral material when the hydrolysate is incubated with hexokinase and ATP. An example of the results of these treatments is shown in Fig. 3. Galactose represented 6 -12% of the radioactivity from chains labeled with [ 3 H]glucose, but none was present in those labeled with [ 3 H]mannose. The amount of label in amino sugars that bound to Dowex-50 cation exchange column was 10 -17%.
After removing the amino sugars, the neutral sugars were separated by paper chromatography. Peaks corresponding to mannose and fucose were present in the hydrolysate labeled with [ 3 H]mannose, and peaks corresponding to the positions of galactose, glucose, mannose, and fucose were evident in the [ 3 H]glucose-labeled material (not shown). The relative distribution of each sugar is shown in Table III.
Utilization of the 3 H-Labeled Sugars-When the exogenous concentration of mannose is 50 M, most of it enters the cell through the high affinity mannose transporter (14). This man-

FIG. 3. 3 H-Monosaccharide analysis using selected enzymatic modifications. PNGase F-released N-linked oligosaccharides labeled with [ 3 H]mannose (A) or [ 3 H]glucose (B)
were hydrolyzed to yield monosaccharides. They were applied to QAE-Sephadex ion exchange columns in Tris base either without any treatment (Ⅺ) or following incubation with hexokinase and ATP (f), with glucokinase and ATP (q), or with galactose oxidase followed by hexokinase and ATP (E) and washed with the same buffer and eluted with 200 mM NaCl (arrows indicate the elution). Results are expressed as percentage of the total dpm in the bound and unbound fraction.

TABLE III Distribution of labeled monosaccharides
PNGase F-released oligosaccharides were hydrolyzed and the amount in each monosaccharide was determined using specific enzyme treatments followed by analysis on QAE-Sephadex and paper chromatography as described under "Experimental Procedures." Labeling sugar 3  IV) and the proportion of that label in mannose (Table III) yields the amount of mannose for each labeling (Table IV). The contribution at each labeling time (  (Table IV). Some of this decrease may result from the depletion of exogenous mannose during the experiment, but a significant portion of radiolabel at 1 h is lost due to conversion of high mannose type chains to complex type oligosaccharides with trimannosyl cores. These results show that direct utilization of labeled mannose is preferred over labeled glucose converted into mannose for N-linked oligosaccharide synthesis. However, glucose may substitute for mannose in fibroblasts if the concentration of the latter is limited. Quantitative Analysis Shows That Mannose Is Highly Preferred over Conversion from Glucose-Radioactivity from [ 3 H]mannose is clearly incorporated into N-linked chains more efficiently than radioactivity from [ 3 H]glucose, suggesting that mannose is preferred over glucose for glycoprotein synthesis. However, this conclusion assumes that the radioactivity contributed by each sugar reflects the molar contribution by each sugar. To prove that this assumption is correct, we used isotope dilution analysis. If mannose is derived exclusively from the exogenous radiolabel, the specific activity of mannose in glycoconjugates will eventually equal that of the precursor in the labeling medium. If other nonradioactive sources contribute to mannose, they will reduce the specific activity of mannose in glycoconjugates in proportion to their contribution. Cells were plated at a low density and allowed to grow for several days in 5 mM glucose and 50 M mannose containing either [ 3 H]mannose at 200 mCi/mmol or [ 3 H]glucose at a specific activity of 50 mCi/mmol. The labeling medium was renewed every day. The proteins were precipitated from cell lysates and hydrolyzed with trifluoroacetic acid, and the specific activity of released mannose was determined as detailed under "Experimental Procedures." The results of two experiments are shown in Fig. 4. As the cells grew, the specific activity of mannose in those labeled with [ 3 H]mannose increased. Extrapolation to infinite labeling time showed that the specific activity (155 mCi/mmol) reached 77% of the specific activity of [ 3 H]mannose in the labeling medium (200 mCi/mmol) (Fig. 4A). This was confirmed in a second experiment in which the cells were incubated along with so- H]glucose in the presence of 50 nmol of D-mannose and 5000 nmol of D-glucose in a volume of 1 ml in Dulbecco's modified Eagle's medium as described under "Experimental Procedures." Utilization of mannose and glucose was measured either by enzymatic assay or by calculation based on the radiolabel left in the medium after evaporation and correction for the amount of secreted glycoproteins.

FIG. 4. Specific activity of [ 3 H]mannose incorporated into cellular macromolecules versus labeling time.
Cells were radiolabeled with [2-3 H]mannose (10 Ci/ml) (q) or [1,5,6-3 H]glucose (250 Ci/ml) (f) in Dulbecco's modified Eagle's medium containing 5 mM glucose and 50 M mannose and 10% fetal bovine serum. Fresh labeling medium was added every 24 h. At the indicated times of labeling, radiolabeled cellular macromolecules were precipitated and hydrolyzed with trifluoroacetic acid, and released monosaccharides were quantified by HPAEC-PAD analysis and liquid scintillation counting of the column effluent as described under "Experimental Procedures" (A). In a second experiment (B) the medium was also supplemented with 1 mM sodium pyruvate, and an excess of radiolabeling medium was added daily (5 ml/25 cm 2 for days 1-3 and 10 ml/25 cm 2 for day 4). These data are shown as the double reciprocal plot for determination of maximal specific activity of incorporated [ 3 H]mannose attainable by extrapolation to infinite labeling time. dium pyruvate (Fig. 4B), since this is a component of most culture media. A double reciprocal plot showed that the specific activity of mannose reached 76% of the [ 3 H]mannose added to the medium, although the time required to reach maximum specific activity was longer in the second experiment. These results show that extracellular mannose is the preferred source of the majority of mannose in total cellular macromolecules even in the presence of a 100-fold molar excess of glucose in the medium, in agreement with results seen for isolated N-linked chains. The specific activity of mannose derived from [ 3 H]glucose did not exceed 23% (11 mCi/mmol) that of the glucose added to the medium (Fig. 4A), suggesting that only labeled mannose or glucose was a source of mannose. It is possible that the 23% could be an underestimate, since cells labeled with 250 Ci/ml [ 3 H]glucose for Ͼ24 h began to show significant radiation-induced cell death. This complication precluded doing longer incubations with glucose. Reducing the specific activity of [ 3 H]glucose in the medium was impractical, since it yields too little radiolabel incorporated into mannose for reliable analysis. Together these labeled sugars account for 99 -100% of the mannose in glycoproteins and indicate that only a very small portion of mannose in glycoproteins could be derived from unlabeled sources such as amino acids or pyruvate via gluconeogenesis.
Cells were also incubated over several days with medium containing 50 M [ 3 H]mannose and 25 mM glucose rather than the physiological concentration of 5 mM. Extrapolation to infinite labeling time shows that the specific activity of mannose in glycoproteins synthesized in the high glucose medium reached only 25% of that seen in 5 mM glucose medium. The decrease probably results from the 3-4-fold greater inhibition of mannose entry via the mannose transporter (14). 2

DISCUSSION
Identifying the source of mannose for oligosaccharide synthesis became a vital question when we found that mannose, but not glucose, corrected glycosylation abnormalities in fibroblasts from children with carbohydrate-deficient glycoprotein syndrome (CDGS) type 1 (15). These CDGS cells made a high proportion of truncated lipid-linked oligosaccharides and underglycosylated their proteins by severalfold (15). This finding raised the possibility that if mannose were directly used for glycoprotein synthesis, it might be a potential therapy for these patients. The case in favor of direct mannose involvement for glycosylation increased when we (14), and others (21-23) also identified a mannose-specific transporter in a variety of mammalian cells that had a K uptake near the concentrations of mannose found in the blood of several mammalian species (9 -12). Since mannose entry via this transporter was not significantly inhibited by glucose at physiological concentration, extracellular mannose might be directly used as a precursor for sugar chain biosynthesis.
The results presented here show that the N-glycosylation pathway preferentially uses free mannose over free glucose to synthesize the same set of sugar chains. The long held assumption that glucose is the major source of mannose for glycoprotein synthesis does not appear to be based on published experiments. There have been no direct comparative studies of the incorporation of mannose and glucose into glycoprotein-bound mannose in mammalian cells. This is undoubtedly due to the inherent difficulties of labeling cells with glucose. Since it is mostly catabolized, it is an unsuitable choice for labeling oligosaccharides. By contrast, [2-3 H]mannose is an ideal specific label for newly synthesized N-linked oligosaccharides, since it can only be catabolized through PMI to generate 3 H 2 O or be converted to [2][3] H]Man-6-P and then to other glycosylation intermediates and finally to macromolecular products (3). These decisive advantages make it practically the universal choice for specific labeling of N-linked chains.
Since the facilitated glucose transporters can also transport mannose at high concentrations (13,24,25), it was assumed that high glucose concentrations would reduce [ 3 H]mannose uptake. Therefore, reducing the glucose concentration should improve mannose labeling efficiency. However, reducing the glucose concentration led to the synthesis of truncated lipidlinked precursor and underglycosylation of many key proteins (26 -35), much like that recently seen in CDGS type I fibroblasts (15). The glycosylation problems associated with the "glucose starvation effect" were reversed by adding either glucose or mannose. However, glucose was always the preferred choice, since it could clearly be converted into the required precursors, while not decreasing the specific activity of [ 3 H]mannose used for labeling.
The interpretation of our results on the relative utilization of [ 3 H]mannose versus [ 3 H]glucose in the analysis of N-linked chains relies on several assumptions. The first is that there is a single pool of Man-6-P that is freely accessible to both Man and Fru-6-P and that subsequent conversions of Man-6-P to Man-1-P and GDP-Man cannot discriminate the origin of Man-6-P. As far as we are aware, there is no evidence for separate pools of these intermediates. Second, the specific activity of the Fru-6-P and Man-6-P pools must quickly attain that of exogenous [ 3 H]mannose and [ 3 H]glucose labels. The Man-1-P, Man-6-P, and GDP-Man pools are very small in cultured cells, rat brain, liver, and kidney (36 -40) and probably equilibrate within a few minutes of adding [ 3 H]mannose (14). The Fru-6-P and Glc-6-P pools have been measured at 0.1-1.0 nmol/mg of protein in various rat organs (38 -40). If these figures are comparable for fibroblasts, all of the relevant precursor pools should equilibrate with exogenous glucose and mannose within a few minutes (14). Given our measured rate of utilization of both labeled mannose and glucose by the cells, it is likely that the internal pools equilibrate with the exogenous radiolabeled medium within seconds or, at most, a few minutes. This is especially important for the 1-h incubations. Continuous labeling of cells with mannose and glucose for 6 and 16 h further ensured equilibration of the label.
The data in Table IV show that mannose consumption measured by direct enzymatic assay and that estimated by accounting for 3 H 2 O and [ 3 H]mannose in the glycosylation pathway closely agree with each other, and mannose uptake rates calculated are comparable with the predicted estimates for mannose transporter (14). As mannose is consumed from the medium, its relative contribution to glycosylation could decrease, but some of this loss is undoubtedly due to N-linked oligosaccharide processing. As long as exogenous mannose remains at physiological levels of 50 M, it contributes the bulk of mannose to glycoproteins synthesized in fibroblasts. Long term labeling experiments require daily renewal of mannose.
The direct demonstration that the specific activity of mannose incorporated into glycoproteins can reach 77% of the specific activity of mannose added to the medium shows that under these physiological conditions, fibroblasts rely primarily on mannose for glycosylation rather than converting glucose into mannose via the well known phosphomannose isomerase-based pathway. The relatively long time needed to achieve maximal labeling with [ 3 H]mannose could reflect a preferential reutilization of mannose salvaged from glycoprotein catabolism. Alternatively, the radiolabel might be self-diluting. A single conversion of [ 3 H]mannose-6-P 3 fructose-6-P ϩ 3 H 2 O by PMI is sufficient to produce a nonlabeled molecule, but the energetically neutral reaction (K eq ϭ 1.03) could sometimes be reversed to generate an unlabeled mannose-6-P that could participate in glycosylation reactions.
The 23% contributed by other sources appears to be primarily from glucose, but a small amount could arise from gluconeogenesis. Including pyruvate in the medium did not affect the ultimate specific activity of [ 3 H]mannose in glycoconjugates, but the calculated time required to reach this level was longer than in its absence. At present we cannot tell whether this is due to culture conditions or other factors.
It is important to point out that the relative contribution of [ 3 H]mannose to glycoprotein synthesis is nearly the same whether measured by isotope dilution analysis or by counting samples labeled with each precursor in short term labelings.
The origin of mannose in the blood is unknown. Some is probably derived from the diet; however, neither the content nor the bioavailability of mannose in foods has been investigated. Mannose may also be derived from normal oligosaccharide processing or from glycoprotein-bound or free oligosaccharide degradation (1,41). Clearly, glucose can be converted into mannose, but the amount may be cell type-or tissue-dependent. If the mannose transporters are major suppliers of mannose for glycoprotein synthesis in mammalian systems, we would expect to find mannose in the blood of all species. Regardless of their specific diets, mammals should also have an intestinal transport system that delivers mannose to the blood. A key factor for supplying mannose is likely to be the efficiency of mannose-specific transporters.
The identification of mannose as a major source for glycosylation in fibroblasts has physiological, nutritional, and medical implications beyond a potential therapy for CDGS type 1 (15,(42)(43)(44). In addition to CDGS patients, sera from chronic alcoholics have underglycosylated glycoproteins and glucosestarved cultured hepatoma cells make underglycosylated glycoproteins that lack entire carbohydrate chains (26,(45)(46)(47)(48). Several reports have underscored the problem of underglycosylation of important glycoproteins (49 -51). Based on the results presented here, mannose may promote more efficient glycosylation of proteins than glucose.