Mutants of mouse fibroblasts altered in the synthesis of cell surface glycoproteins. Preliminary evidence for a defect in the acetylation of glucosamine 6-phosphate.

We hve reported the isolation and preliminary biochemical characterization of two mutants of mouse fibroblasts selected for a decrease incell-to-substrate adhesion (Pouysségur, J., and Pastan, I. (1976) Proc. Natl. Acad. Sci. U.S.A. 73, 544-548). We attributed the adhesive defect of these mutants (AD6 and AD8) to the absence of iodinatable cell surface proteins. This study demonstrates that a defect in glycoprotein synthesis is the biochemical basis for the reduction in proteins exposed at the outer surface of the mutant cells. When D-glucosamine or L-fucose was used as radioactive precursor, analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed a decrease in the labeling of all the glycoproteins of both clones. This decrease in glycoprotein labeling was not due to a defect in D-glucosamine uptake, since this precursor was taken up at a higher rate in the mutants than in the wild type. In spite of this high uptake, the rate of D-glucosamine incorporation into macromolecules was decreased by 60% and the carbohydrate content of membranes (mannose, galactose, N-acetylglucosamine, N-acetylgalactosamine, and sialic acid) from clone AD6 was diminished by 40 to 60%. When the various cell lines were labeled for 1 to 3 h with glucosamine and the acid-soluble pool analyzed, the wild type cells were found to accumulate UDP-N-acetylhexosamine as the major component. In contrast, clones AD6 and AD8 accumulated glucosamine 6-phosphate as the major component. This last finding suggests that there is a block in the N-acetylation of glucosamine 6-phosphate in both instants. This suggestion is supported by the finding that feeding the mutant cells 10 mM N-acetylglucosamine reverts them to the wild type phenotype (Pouysségur, J., Willingham, M. and Pastan, I. (1977) Proc. Natl. Acad. Sci. U.S.A., in pressy. In wild type cells all of the iodinatable proteins of the cell surface have the same mobility as the glycoproteins when analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. We conclude that the early block in glycoprotein synthesis accounts for the reduction in iodinatable surface proteins in the mutant cells.

We have reported the isolation and preliminary biochemical characterization of two mutants of mouse fibroblasts selected for a decrease in cell-to-substrate adhesion (Pouyssegur, J., and  Proc. Natl. Acad. Sci. U. S. A. 73,[544][545][546][547][548]. We attributed the adhesive defect of these mutants (AD6 and AD8) to the absence of iodinatable cell surface proteins.
This study demonstrates that a defect in glycoprotein synthesis is the biochemical basis for the reduction in proteins exposed at the outer surface of the mutant cells. When D-glucosamine or L-fucose was used as radioactive precursor, analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed a decrease in the labeling of all the glycoproteins of both c!ones. This decrease in glycoprotein labeling was not due to a defect in D-glucosamine uptake, since this precursor was taken up at a higher rate in the mutants than in the wild type. In spite of this high uptake, the rate of n-glucosamine incorporation into macromolecules was decreased by 60% and the carbohydrate content of membranes (mannose, galactose, N-acetylglucosamine, Nacetylgalactosamine, and sialic acid) from clone AD6 was diminished by 40 to 60%. When the various cell lines were labeled for 1 to 3 h with glucosamine and the acid-soluble pool analyzed, the wild type cells were found to accumulate UDP-N-acetylhexosamine as the major component. In contrast, clones AD6 and AD8 accumulated glucosamine 6-phosphate as the major component. This last finding suggests that there is a block in the N-acetylation of glucosamine 6-phosphate in both mutants. This suggestion is supported by the finding that feeding the mutant cells 10 mM N-acetylglucosamine reverts them to the wild type phenotype (PouyssCgur, J., Willingham, M., and Pastan, I. (1977) Proc. N&l. Acud. Sci. U. S. A., in press).
In wild type cells all of the iodinatable proteins of the cell surface have the same mobility as the glycoproteins when analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
We conclude that the early block in glycoprotein synthesis accounts for the reduction in iodinatable surface proteins in the mutant cells. We have recently reported the isolation of two clones of mutant mouse 3T3 fibroblasts that have a decrease in adhesion to substratum (1). These mutants like many transformed fibroblasts also have a rounded shape and are easily agglutinated by plant lectins (2). However, the mutant cells are not transformed since they still display density-dependent inhibition of growth, will not grow in suspension culture, and are unable to produce tumors in mice.
Using lactoperoxidase-catalyzed radioiodination of the cell surface, (3) we demonstrated a marked alteration in the proteins on the surface of these cells. Two iodinated polypeptides of apparent molecular weights, of 140,000 and 92,000, were missing and a number of others were decreased. On the basis of these data we suggested that the decrease or the absence of outer membrane proteins was responsible for the defect in adhesion and the altered morphology of the mutant cells.
To establish the biochemical basis of these alterations, we have investigated the synthesis of glycoproteins in the mutant and parent cells. We report here that the mutant clones have an overall decrease in glycoprotein synthesis due to a limitation in the rate of formation of N-acetylglucosamine phosphate. A preliminary report of this work has been presented (4).  (6). Sliced gels were solubilized in 0.5 ml of 30% H,O, (7) and counted in a liquid scintillation spectrometer. Dried gels were exposed to RP Royal "X-Omat" film. Cell Surface Radioiodination -Cell monolayers were washed three times with NaCl/P,' at room temperature. Two millihters of NaCl/ Pi containing 5 mM glucose and 400 $Zi of carrier-free Na? were added. Iodination was performed as previously described (1,3). The reaction was started by the addition of lactoperoxidase (20 pg/ml) and glucose oxidase (0.1 unit/ml) and stonued after 10 min at room temperature by washing the celis once \;fth an excess of cold NaI (NaClIP, in which NaCl was replaced bv NaI at the same molaritv) and twice with NaCl/P,.
All washing media contained 2 mM phenilmethylsulfonyl fluoride. Uptake Studies, Soluble Pool, and Acid-insoluble Fraction -For Dglucosamine and 2-deoxyglucose uptake medium from exponentially growing cells (sparse or nearly confluent) was removed by aspiration. Unless otherwise specified, cells in 60-mm dishes were washed three times with 2 ml of glucose and serum-free growth medium and incubated at 37" with 1 ml (60-mm dish) of this medium supplemented with the labeled substrate.
At various time intervals medium was aspirated off and cell monolavers washed four times with 2 ml of glucose and serum-free medium at 37". Cells were then scraped off the dish in 0.4 ml of H,O with a rubber policeman and homogenized.
The cell homogenates were divided intd three aliquots for determination of protein, total radioactivity, and acid-insoluble radioactivity.
For total radioactivity, a O.l-to 0. ice/acetone bath. The membranes were spun down at 100,000 x g for 60 min and the pellet resuspended in H,O. Sialic acid was assayed by the thiobarbituric acid method of Warren (11) after hydrolysis in 1 N HCI for 90 s at 100" (12). Hexosamines were determined with an automatic amino acid analyzer after hydrolysis in 4 N HCI for 6 h at 100".
Protein Analysis -Proteins were determined according to Lowry et al. (13) using bovine serum albumin as a standard.

Cell Surface Glycoproteins
of Mouse 3T3 Fibroblasts-To investigate the glycoproteins present in 3T3 cells, the cells were incubated with [14C]glucosamine and the labeled proteins separated by electrophoresis. The proteins labeled with ['4Clglucosamine did not correspond to the proteins observed by Coomassie blue staining. For example, myosin, a major cell protein, was not labeled. Thus [%lglucosamine is not significantly utilized as a glycolytic substrate in 3T3 cells and as reported by others (14) is a reliable precursor for glycoprotein biosynthesis. Fig. L4 shows a typical glycoprotein pattern of whole cell extract. The main characteristic is the presence of two large peaks (A and B) of molecular weights 100,000 to 110,000 and 45,000 to 55,000. These peaks correspond to the two classes, A and B, of glycoproteins reported by Hunt and Brown (15) in L929 mouse fibroblasts and also found in many other fibroblast cell lines.' The major peak (A) is the main component detected by the periodic acid-Schiff reaction (16,17) but does not stain as a discrete band with Coomassie blue even in a crude membrane fraction where this glycoprotein is enriched. Therefore, Peak A might be composed of a protein with a high carbohydrate content similar to periodic acid-Schiff 1 or glycophorin of human erythrocytes (18). Besides these two main groups of glycoproteins, five other glycoproteins with molecular weights over 90,000 were detected These glycopeptides are membrane-associated since the radioactive pattern does not change when crude plasma membranes were analyzed (Fig. 2).
To establish which of the membrane-associated glycoproteins are exposed on the outer surface of the cell, duplicate dishes were iodinaied with the lactoperoxidase-catalyzed reaction (3). Fig. 1, A (0-O) and B shows the iodinated pattern resolved in the same polyacrylamide slab gel electrophoresis as the [%lglucosamine homogenate. It is noteworthy that the high molecular weight glycoproteins, with the exception of Peak A, correspond to the iodinated bands 1 to 6 (see Table I for molecular weight). Iodinated band 6 had been resolved from Peak A as a separate ['Y!lglycoprotein by increasing the time of electrophoresis. In other experiments band 6 has also been found to bind concanavalin A.:' These data suggest that most of the membrane glycoproteins of 3T3 cells are exposed on the cell surface. This finding is in agreement with the results of Gahmberg (19) who studied the external labeling of human erythrocyte glycoproteins.

Glycoprotein Analysis of Wild Type and Mutant
Cell Homogenate -Previously we reported that the main cell surface alterations of clones AD6 and AD8 were loss of iodinated bands 4 and 6 and a decrease in the intensity of band 5; these three polypeptides have now been identified as glycoproteins (Table  I). To investigate whether or not the mutants synthesize these polypeptides, mutant and wild type cells were grown at different cell densities in the presence of r+C]glucosamine for 48 h. Whole cell homogenates were solubilized in sodium dodecyl sulfate, electrophoresed, and the radioactive glycoproteins analyzed. Fig. 3 shows the autoradiograms of the polyacrylamide to sparse (A) and subconfluent (B) exponentially growing cells and also to nongrowing cells (C) that had been confluent for 2 days. Fig. 4 shows the radioactivity profile of Experiment B. In both mutants the ['*Clglucosamine incorporation was strongly reduced. The reduction involved all of the glycoproteins as well as the complex polysaccharides not entering the gel (see the decrease of the radioactivity at the top of the appropriate gels in Fig. 4). This defect was observed in the three phases of growth but was more marked in nongrowing cells.
A similar experiment, using n-[14C]fucose as a precursor of glycoprotein synthesis and ['*C]glucosamine for comparison, is presented in Table II and Fig. 5. In sparse and confluent cells, Gfucose incorporation into macromolecules was reduced in both mutants; clone AD6 was decreased more than AD8. Cell density affected the incorporation of fucose and glucosamine differently. In confluent cells, the decrease of L-fucose incorporation was less marked, whereas it was more marked for D glucosamine. This difference could result from the fact that Lfucose is mainly incorporated into glycoproteins, whereas D glucosamine is incorporated into both glycoproteins and mucopolysaccharides. Therefore, increased synthesis of mucopolysaccharides at confluency could account for a difference in the incorporation of these two precursors depending upon whether the cells are growing or resting.
These results indicate that not only the formation of bands 4,5, and 6 is depressed in the mutant cells but also that overall glycoprotein synthesis is decreased in both mutants. This statement assumes that the uptake of glucosamine and fucose is not specifically impaired in the mutant clones. This assumption is analyzed in the following section.
Glucosamine Uptake and Rate of Incorporation into Macromolecules -Sparse and nearly confluent cells in the log phase of growth were incubated for various times in a glucose and serum-free medium containing radioactive glucosamine. The A duplicate gel (B) was run longer and exposed to x-ray film. Lane b corresponds to the cell surface-iodinated proteins and-Lane a to the glycopeptide pattern of whole cell homogenate. Apparent molecular weights (x 10m3) were determined with the set of protein standards mentioned in the legend of Table I. time course of total glucosamine uptake and its incorporation into trichloroacetic acid-precipitable material is shown in Fig.  6. The rate of total n-glucosamine uptake was 3 to 4 times higher in both mutants than in the wild type cells, but its incorporation into acid-insoluble material was decreased. This finding was reproduced in three independent experiments with sparse and confluent cells. The difference in total cellassociated counts was not due to a difference in substrate binding to the cell since at least 95% of the glucosamine taken up in 60 min has been metabolized (see analysis of acid-soluble pool, Fig. 9). However, when cells are grown for 2 to 3 days at confluency (stationary phase of the growth) the initial rate of uptake decreases (40-fold decrease in the mutants) to reach about the same rate in the three cell lines. It is thought that Dglucosamine is taken up through the n-glucose carrier because glucose and glucosamine compete with each other, both uptake systems are competitively inhibited by cytochalasin B (20), and both are decreased at high cell densities. 4 We, therefore, measured the 2-deoxyglucose uptake. No significant difference was found in the rate of uptake of the mutants and parental cells.
Despite the high rate of glucosamine uptake in AD6 and AD8, incorporation of glucosamine into macromolecules was decreased by 60% in both mutants (Fig. 6). We also performed this experiment in cells growing in medium containing glucose and serum. (The conditions were similar to those in Fig.  4.) Although the total uptake of glucosamine by AD6 was slightly lower (30%) than in the wild type under these conditions, the rate of incorporation into macromolecules measured for 6 h was still markedly decreased (70%) (Fig. 7). Furthermore, this rate was not limited by the amount of nglucosamine taken up by the cells. Indeed, the amount of ['*C]glucosamine incorporated into macromolecules was un-

FIG. 2. Autoradiograms of sodium dodecyl sulfate gel electrophoresis of plasma membrane (il4) and whole cells ( WHI labeled with ['Y!]glucosamine.
3T3 cells were planted at lo5 cells/lOO-mm dish and grown for 2 days; then the medium was changed and supplemented with ['Vlglucosamine (51 mCi/mmol) at 1.25 &i/ml. Two days later cells were harvested and crude membrane was prepared as follows.
The nuclei were discarded by low speed centrifugation (800 x g, 10 min) and the supernatant centrifuged for 60 min at 100, 000 x g. The 100,000 x g membrane pellet (M) and the whole cell homogenate (WH) were solubilized in sodium dodecyl sulfate and electrophoresed (40 pg of protein) in a 5% polyacrylamide gel. The dried gel was exposed for 5 days to x-ray film.  changed under conditions where the n-glucosamine uptake was increased lo-fold by lowering the glucose concentration from 5 mM to 0.1 mhi.
Similar experiments using L-fucose as a precursor have shown a marked decrease in the rate of its incorporation into acid-insoluble material; a growing culture of AD6 incorporates 15% as much L-fucose as the wild type cells (data not shown).
In parallel experiments we found no difference in the rate of overall protein synthesis among AD6, AD8, and wild type cells using L-leucine as a precursor. The three cell lines also have identical growth rates (1).

Analysis of Acid-soluble Pool after o-Glucosamine Pulse -
The metabolic fate of exogenous glucosamine in eukaryotic cells is outlined in Fig. 8. Exogenous glucosamine is phosphorylated, N-acetylated, and converted after several steps into three nucleotide sugars: UDP-N-acetylglucosamine, UDP-Nacetyl galactosamine, and CMP-sialic acid. These constitute the immediate precursors of heteropolysaccharides, glycolipids, and glycoproteins (21). The hexosamine pathway in mutant and wild type cells was investigated analyzing the radioactive soluble intermediates derived from glucosamine labeling. Fig. 9 shows the identification of the major radioactive peaks after separation by paper chromatography of the acidsoluble extracts. In the three extracts free n-glucosamine represents less than 5% of the total counts. In wild type extracts, the main peak (75% of the radioactive pool) migrates slower than the major component of mutant extracts. The major component in wild type cells was identified as UDP-N-acetylglucosamine because it had the same Rp as the standard compound and was resistant to hydrolysis by alkaline phosphatase (Fig. 9). An identical experiment with a line of SV40 transformed 3T3 cells also gave rise to a single peak of UDP-Nacetylglucosamine. 4 These findings are in agreement with those of Kornfeld and Ginsburg (22) for HeLa cells. In contrast to the normal and SV40 transformed cells, AD6 and AD8 almost exclusively accumulated glucosamine g-phosphate; it constituted 90% of the acid-soluble pool. The identification of this compound is based on the fact that treatment of soluble extracts with alkaline phosphatase converts the labeled product to glucosamine (Fig. 9). The other intermediates of the glucosamine pathway, N-acetylglucosamine 6-phosphate and N-acetylglucosamine l-phosphate, would be converted to Nacetylglucosamine by alkaline phosphatase which has a higher mobility than glucosamine (Fig. 9). Small amounts of N-acetylglucosamine were found when wild type extracts were treated with alkaline phosphatase, indicating that small amounts of N-acetylglucosamine phosphate accumulated in the parent cells.
The acid-soluble pool was also analyzed after 1 h and 3 h in cells growing in the presence of 5 mM glucose and 10% serum. The acid-soluble pool of AD6 was still composed of only glucosamine B-phosphate, whereas wild type cells accumulated UDP-N-acetylhexosamine and N-acetylglucosamine phosphate (data not shown).
These results suggest strongly that the hexosamine pathway of the mutant cells is impaired by a block in the acetylation of glucosamine 6-phosphate (Step 2 of the scheme, Fig. 8).

Fibroblast
Cell Cells were planted and serum and the reaction was started by addition of D-L3Hlglucosamine (0.5 PM, 2.9 &i/ml) in 1 ml of medium without at 105 tells/60-mm dish, medium changed 48 h later, and the experiglucose and serum. The reaction was stopped by removing the mement performed 1 day later: in Experiment A, cells were sparse; in dium and washing four times with glucose and serum-free medium Experiment B, duplicate dishes were assayed 2 days later when the (the four washings took less than 30 s). The cells were then scraped cells were confluent but still in log phase. The uptake studies were off the dish and counted to determine total uptake (left panel) and performed with cells in monolayer, incubated at 37" in a CO, incuba-uptake into the acid-insoluble fraction (right panel general decrease in all of the carbohydrates analyzed: a 56% decrease in mannose, a 66% decrease in galactose, a 58% decrease in N-acetylglucosamine, a 59% decrease in N-acetylgalactosamine, and a 40% decrease in sialic acid. Fucose is also reduced ( Fig. 10) but the radioactivity was too low to calculate an exact percentage. Only the glucose content of the mutant cells was increased, but this could have resulted from a higher glycogen content in AD6.

DISCUSSION
We have previously shown that many of the iodinatable proteins in the plasma membranes of the adhesion-deficient clones (AD6 and AD81 are altered, some are decreased and others are undetectable. This reduction could have been due to decreased synthesis of membrane proteins or to a conformational change in the cell surface resulting in decreased accessibility of these membrane proteins to lactoperoxidase. It is generally believed that many of the proteins exposed on the outer surface of cells and detected by covalent labeling with various nonpenetrating probes (23,24) are glycosylated. This concept has been strengthened by recent studies by Gahmberg (19) in human erythrocytes and Hunt and Brown (15) in mouse L cells. Our data with 3T3 cells showing that the same six polypeptides are labeled by lactoperoxidase iodination and [14C]glucosamine incorporation are in keeping with this concept and furthermore suggested that one possible ex-  in this paper demonstrate that both mutants do have an alteration in glycoprotein synthesis. All of the mutant glycoproteins labeled with n-glucosamine and resolved by sodium dodecyl sulfate-polyacrylamide electrophoresis are decreased. This reduction in the labeling of glycoproteins is not due to a selective impairment in the ability to utilize glucosamine since in both clones (a) the incorporation of L-fucose, a precursor which is utilized by a different pathway of glycoprotein synthesis (251, is also decreased and (b) the uptake of glucosamine is not rate-limiting.
Indeed, instead of being diminished the uptake of glucosamine is 3-to 4-fold higher in the mutants incubated at a low glucosamine concentration and in glucose-free medium. The reason for this increase is not known but it is not associated with an enhancement of glucose uptake. In spite of this increase in uptake, AD6 and AD8 incorporate considerably less glucosamine into macromolecules (30 to 40% of the control rate). These findings are best explained by an overall decrease in glycoprotein synthesis. However, the evaluation of glycoprotein synthesis using an exogenous precursor can be complicated by an aberrant compartmentation of the carbohydrate precursor pools in the mutants. Such a possible artifact was ruled out by direct the morphology, the cell-to-substratum adhesion, and the cell surface protein iodination pattern were restored to normal (2,4). In addition, after the biochemical reversion, the rate of incorporation of L-fucose was restored to normal, whereas the incorporation of n-glucosamine was still impaired (2).
This finding which shows that the block can be bypassed using N-acetylglucosamine supports our suggestion that the acetylation of glucosamine 6-phosphate is limited in both mutants. Apparently such a defect is sufficient to account for all of the abnormal properties of the mutant cells.
A number of important questions remain unanswered. One is whether the defect in acetylation leads to a reduction in the number of glycoprotein molecules synthesized and inserted into the membrane or to the formation of incompletely glycosylated membrane proteins. Another is related to the role of carbohydrates in membrane structure. One of the major iodinated glycoproteins of the surface of 3T3 cells, band 6 of M, = 92,000, is not detected in the mutants. However, a noniodinated protein with a M, of 90,000 accumulates in the plasma membranes of AD6 cells which could be precursor of the 92,000 protein. This preliminary observation suggests that incompletely glycosylated or nonglycosylated proteins are still attached to the membrane of AD6 and that the reduction of the number or size of the oligosaccharide chain reduces the exposure of the protein to the aqueous environment of the cell surface or changes the nature of its insertion into the membrane The nature of the mutation leading to a limitation in the acetylation of glucosamine g-phosphate is not yet established. The enzyme glucosamine 6-phosphate N-acetylase is a likely candidate since the formation of acetylated amino sugars is an obligatory step for their interconversion. This enzyme has been described in yeast, bacteria, and different animal tissues (26-28).
Preliminary attempts to assay this activity in fibroblasts using the calorimetric method of Reissig et al. (29) were unsuccessful because of its low sensitivity. We are currently developing a more sensitive assay using a radioactive substrate. It will be of interest to know whether the regulation or the physical characteristics of this enzyme are modified in AD6. It is clear that the mutants are "leaky" in the acetylation reaction since the glycosylation process is not totally abolished. A further reduction in glycosylation might be lethal for normal cells which require some attachment to substratum to grow. AD6 which displays a more marked alteration than AD8 is vacuolated and grows poorly in low serum or if the medium is not changed regularly.
We have emphasized in our study the alteration of glycoproteins in the mutant cells. Glycoproteins constitute one of the major carbohydrate containing components of the cell surface. However, the early position of the block in the mutant cells should affect the carbohydrate composition and the synthesis of glycolipids and mucopolysaccharides.
An investigation of these specific carbohydrate-containing components is needed to assess to what extent their biosynthesis is altered.
Other mutants with defective cell surface carbohydrates have already been reported. Morphological mutants of Neurospora crassa were found to have an altered carbohydrate composition in their cell wall (30). Further characterization of these mutants demonstrated the existence of mutations in the glycolytic pathway leading to the abnormal surface carbohydrate composition (31, 32). More recently, mammalian cell mutants have been isolated based on their resistance to the toxicity of some plant lectins. Some of these have an altered glycoprotein pattern due to a defect in UDP-N-acetylglucosamine -glycoprotein -N-acetylglucosaminyltransferase (33,34). These mutants also appear to have an altered morphology and a decreased adhesion to substratum. Clones AD6 and AD8 provide another example of cell surface carbohydrate mutants of eukaryotic cells. Such cell lines should be very useful for the studies of membrane structure and to evaluate the role of cell surface proteins and carbohydrates in biological processes.