Isolation of Wheat Germ Agglutinin-resistant Clones of Chinese Hamster Ovary Cells Deficient in Membrane Sialic Acid and Galactose*

Several clones of Chinese hamster ovary cells have been selected for their resistance to the toxic effects of wheat germ agglutinin. The clones do not bind wheat germ agglutinin as well as parent cells and are 5- to 250-fold more resistant to the toxic effects of the lectin. Of three clones studied in detail, all exhibit a decrease in wheat germ agglutinin binding affinity. Two have normal numbers of wheat germ agglutinin binding sites, while one (Clone 13) has a 65% decrease in binding sites. Crude membrane preparations of the clones have a decrease in sialic acid content relative to parent cells, and Clone 13 membranes are also deficient in galactose, while the mannose and hexosamine contents of all three clones are normal. The membrane sugar deficiencies affect both glycoproteins and glycolipids. Sialyl-lactosylceramide is the major glycolipid in parent cells, while Clones 1 and 1021 have lactosylceramide and Clone 13 has glucosylceramide as the predominant glycolipid. Labeling experiments with N-[G-3H]acetylmannosamine suggest that Clone 1021 cells have a block in the transfer of sialic acid from CMP-sialic acid to glycoprotein and glycolipid acceptors. Yet CMP-sialic acid:glycoprotein sialyl-transferase activity in cell lysates of Clone 1021 cells is 80% of normal. While CMP-sialic acid:lactosylceramide sialyl-transferase activity is only 25% of normal, it can be restored to normal or elevated levels by sodium butyrate induction without an associated increase in cellular sialyl-lactosylceramide content. Similarly, the galactose-deficient Clone 13 can synthesize UDP-galactose and has normal levels of UDP-galactose:glycoprotein galactosyltransferase and UDP-galactose:glucosylceramide galactosyltransferase when assayed in vitro. The glycosyltransferases of both these clones can utilize their own glycoproteins as sugar acceptors in in vitro assays. These data suggest that the variant cells fail to carry out specific glycosyltransferase reactions in vivo despite the fact that they possess the appropriate nucleotide sugars, glycoprotein and glycolipid acceptors, and glycosyltransferases.

Several clones of Chinese hamster ovary cells have been selected for their resistance to the toxic effects of wheat germ agglutinin.
The clones do not bind wheat germ agglutinin as well as parent cells and are 5. to 250.fold more resistant to the toxic effects of the lectin. Of three clones studied in detail, all exhibit a decrease in wheat germ agglutinin binding affinity. Two have normal numbers of wheat germ agglutinin binding sites, while one (Clone 13) has a 65% decrease in binding sites. Crude membrane preparations of the clones have a decrease in sialic acid content relative to parent cells, and Clone 13 membranes are also deficient in galactose, while the mannose and hexosamine contents of all three clones are normal. The membrane sugar deficiencies affect both glycoproteins and glycolipids. Sialyl-lactosylceramide is the major glycolipid in parent cells, while Clones 1 and 1021 have lactosylceramide and Clone 13 has glucosylceramide as the predominant glycolipid.
Labeling experiments with N-[G-:'H]acetylmannosamine suggest that Clone 1021 cells have a block in the transfer of sialie acid from CMP-sialic acid to glycoprotein and glycolipid acceptors. Yet CMP-sialie acidzglycoprotein sialyltransferase activity in cell lysates of Clone 1021 cells is 80% of normal. While CMP-sialic acid:lactosylceramide sialyltransferase activity is only 25% of normal, it can be restored to normal or elevated levels by sodium butyrate induction without an associated increase in cellular sialyl-lactosylceramide content. Similarly, the galactose-deficient Clone 13 can synthesize UDP-galactose and has normal levels of UDP-galactose:glycoprotein galactosyltransferase and UDP-galactose:glucosylceramide galactosyltransferase when assayed in vitro. The glycosyltransferases of both these clones can utilize their own glycoproteins as sugar acceptors in in vitro assays. These data suggest that the variant cells fail to carry out specific glycosyltransferase reactions in uiuo despite the fact that they possess the appro-priate nucleotide sugars, glycoprotein and glycolipid acceptors, and glycosyltransferases.
It has recently been demonstrated that plant lectins with cytotoxic properties can be used as selective agents for obtaining tissue culture cells with altered membrane oligosaccharide units (2)(3)(4)(5)(6)(7). Several of these lines are deficient in specific glycosyltransferases which can account for the observed oligosaccharide alterations (3, 5-7). We now describe an entirely different group of variant cells selected for resistance to wheat germ agglutinin.
These variant cells have biosynthetic defects which lead to altered membrane oligosaccharides, but they differ from the previously described variants in that they contain adequate levels of the appropriate glycosyltransferases.

EXPERIMENTAL PROCEDURES
Materials MEM-alpha, name given to a tissue culture developed by C. C. Stewart (8), was obtained from Flow Laboratories, Rockville, Md. Fetal calf serum, heat inactivated fetal calf serum, trypsin, fetuin (Spiro method), penicillin, and streptomycin were purchased from Grand Island Biological Co. Plastic Petri dishes (tissue culture grade) and T-flasks were from Lux or Falcon. Na'*"I (reagent grade) was from Mallinckrodt.
All other radioactive materials were obtained from New England Nuclear Corp. CMP-AJ-acetylneuraminic acid was synthesized by the procedure of Kean and Roseman (9). Pronase and Vibrio duderue neuraminidase were purchased from Calbiochem.
Jack bean P-galactosidase and P-N-acetylglucosaminidase were prepared by the method of Li (10). N-Palmitoyl-1-O-n. glucosyl-sphinganine and N-palmitoyl-1-O.n-lactosylsphinganine were obtained from Miles-Yeda. Purified brain gangliosides were a gift of Dr. R. Burton, Washington University School of Medicine, St. Louis. Cutscum was obtained from Fisher while bovine cardiolipin and ovalbumin were from Sigma. Triton X-100 was purchased from Packard.
All other chemicals were of reagent grade and obtained from commercial sources. Wheat germ agglutinin and the lectins ofRicinus communis (ricin and R. communis agglutinin I), Agaricus bisporus, Lens culinaris, and Phaseolus oulgaris were prepared as previously described (ll-15). Soybean agglutinin was prepared by a slight modification of the affinity chromatography method of Lis and Sharon (16). The lectins were labeled with lzLI by the chloramine-T method using a 10-s exposure to chloramine-T To determine LD,,, replicate numbers of cells (usually 200) were seeded onto 60-mm plastic Petri dishes. After a few hours of incubation to allow adherence to the dish, the medium was removed and replaced with complete medium containing from 0.1 to 50 pg/ml of lectin.
The dishes were then incubated for several days and the number of colonies per dish counted after staining with 0.5% methylene blue in 10% formaldehyde.
The cloning efficiency for the various lines was 75%. The LD,,, is the lectin concentration at which only 50% of the cells (relative to the control) give rise to colonies in complete growth media.

Lectin
Binding Assays The binding reactions were carried out in plastic test tubes that had been presoaked for 1 h with 5 mg/ml of bovine serum albumin. The incubation mixtures contained in 0.25 ml: ""I-labeled lectin (from 0.2 to 25 Kg); 1 mg of bovine serum albumin; 0.9% NaCl, 0.01 M NaHCO,; and 0.5 to 2 x 10" cells. After l-h incubation at room temperature, the cells were washed three times with 2~5 ml of 0.9% NaCl, 0.01 M NaHCO,, and the amount of '"I-labeled lectin bound to the cells was measured in a y scintillation counter. The number of binding sites per cell and the apparent association constant for lectin binding were determined by the method of Scatchard (18). A molecular weight of 23,500 was assumed for wheat germ agglutinin.

Cel l Fractionation
All operations were carried out at 4". The washed cells were either resuspended in 4 to 5 volumes of distilled water and subjected to six cycles of rapid freezing and thawing or resuspended with 4 to 5 volumes of phosphate-buffered saline and disrupted by sonication with the microprobe of a Biosonik II sonicator (Bronwill Scientific, Inc.), using two 10-s pulses at a probe intensity of 35. The cell lysates were centrifuged at 100,000 x g for 90 min to obtain a crude "membrane" (particulate) and a "soluble" fraction. The particulate material was washed once with 6 ml of water or phosphate-buffered saline and resuspended in the same buffer. Total cellular lipids were extracted from either crude membranes or washed whole cells with 20 volumes of chloroform:methanol (2:l) at room temperature.
Non-lipid contaminants were removed by passage through Sephadex G-25-80 as described by Wells and Dittmer (21). Carbohydrate analysis of total cellular lipids was performed as described above for the analysis of crude membranes. Organic phosphate was determined by the method of Ames (22). Radioactive glycolipids were extracted from cells grown in the presence of radioactive sugars as described in the legend to Fig. 3  The nucleotide sugars were separated from free sugars and sugar monophosphates by descending paper chromatography on Whatman No. 3 MM paper in 95% ethanol:1 M ammonium acetate (7:3), and were eluted from the paper with distilled water. CMP-sialic acid was assayed directly by the thiobarbituric acid assay method (23). Zero to 10% hydrolysis of the tracer CMP-sialic acid was observed during the isolation procedure.
For measuring glycosyltransferase activities toward exogenous acceptors, cells were usually disrupted either by freezing and tbawing in water or by sonication in phosphate-buffered saline (see "Cell Fractionation").
Assays were also conducted at least once for each acceptor by adding a suspension of intact cells to the reaction mixtures containing detergents.
Control assays were terminated at 0 min. Reaction mixtures without added asialofetuin were included to detect incorporation into endogenous acceptors. Lipids and detergent were added to the assay tubes in chloroform, which was evaporated at room temperature prior to the addition of the other components. After a 30-min incubation at 37", the lipid products were either extracted with 20 volumes of chloroform:methanol(2:1), followed by removal of non-lipid contaminants (211, or they wore precipitated with 5 ml of cold 1% phosphotungstic acid in 0.5 N HCl and then extracted with 5 ml of chlorofornnmethanol (2:l with authentic GM% on thin layer plates using Solvent A. Similarly, the 3H-labeled product of the UDP-galactose:glucosylceramide galactosyltransferase assays co-chromatographed with lactosylceramide on thin layer plates in Solvent B. Greater than 85% of the labeled product of the UDP-galactose:ovalbumin galactosyltransferase assay was susceptible to degradation by P-galactosidase. The sialyltransferase and galactosyltransferase activities of the cell lystates were totally recovered in the crude membrane fraction (see "Cell Fractionation").
After incubation at 37" for 30 min, the reaction mixtures were assayed for free sialic acid (23).
ARer incubation at 37" for 30 min, the reactions were terminated by the addition of 0.7 ml of 0.2 M Na,CO,.
The mixtures were centrifuged and releasedp-nitrophenol was quantitated by measuring the absorbance of the supernatant at 420 nm.

Selection of Wheat Germ Agglutinin-Resistant
Clones -Resistant cells were selected for their ability to grow in the presence of toxic concentrations of wheat germ agglutinin.
In a typical selection experiment, 2 x lo4 cells were plated onto 60mm Petri dishes, allowed to adhere to the dishes, and then incubated in the presence of 4 pg/ml of lectin in complete growth medium. This amount of lectin killed most of the cells, but usually one or two clones were found per dish after a 2week incubation period. Cells from individual colonies were collected by trypsinization within a stainless steel cloning ring, and subcultured in nonselective medium. No prior mutagenic treatment was employed.
Nine clones were selected for further studies. These clones have growth rates similar to the parent population and have been stable with respect to both lectin resistance and morphological characteristics over many serial passages (in some cases, for nearly 2 years of continuous culture). While all nine clones have individual distinguishing characteristics, the subsequent discussion will focus primarily on three clones which were studied in detail (Clones 1, 13, and 1021).
Morphological Appearance of Clones -Although the parental Chinese hamster ovary cell population is morphologically somewhat heterogeneous, most of the cells give rise to moderately compact colonies of epithelioid cells (Fig. LA) Wheat Germ Agglutinin-resistant Cells the wheat germ agglutinin-resistant clones are morphologically distinguishable from each other and fmm the parent line. Clone 1 cells are extremely rounded in monolayer cultures, and the cells within a colony are not in close contact with each other (Fig. 1B). As the cultures approach confluency, the cells become slightly elongated and flattened. These cells are easily removed from plastic substrata by vigorous agitation, mild trypsinization or EGTA treatment. Clone 1021 cells are more ellipsoidal than the parent cells (Fig. 1C). Neighboring cells adhere tightly to each other and frequently form whorl-like patterns. Clone 13 cells are morphologically indistinguishable from parent cells.
Another morphological variation is exemplified by Clone 5 (Fig. v)). These cells flatten much more than parent cells in monolayer culture, covering a much larger surface area per cell. Unlike most of the other clones, Clone 5 cells do not grow readily in suspension culture. tended to vary in different experiments, the relative resistance of the clones did not change. All of the resistant clones studied do not bind wheat germ agglutinin as well as normal Chinese hamster ovary cells. Typical binding curves are shown in Fig. 2A. When the binding data are analyzed by the method of Scatchard (18), biphasic and perhaps even multiphasic curves are obtained (Fig. 2J3). Nonlinear Scatchard plots of wheat germ agglutinin binding to human erythrocytes has previously been observed (12). The complexity of the curves makes it difficult to calculate meaningful association constants but approximate values can be obtained as well as estimates of the total number of lectin binding sites. In the case of Clones 1 (not shown) and 1021, the decreased binding at low lectin concentrations is due primarily to a decrease in binding affinity (K,) for the lectin rather than to a loss of binding sites. In contrast, Clone 13 exhibits both a decrease in binding affinity and in the total number of binding FIG. 2. Binding of 1251-labeled wheat germ agglutinin to parent and variant cells. The binding reactions were carried out in plastic test tubes (12 x 75 mm) that had been presoaked with 3 ml of 5 mg/ ml bovine serum albumin in phosphate-buffered saline. The reaction mixtures contained, in 250 ~1 of phosphate-buffered saline: 250 pg of bovine serum albumin, 5 x 105 washed cells, and from 0.2 to 25 pg of '*"I-labeled wheat germ agglutinin. After 1 h at room temperature, the cells were washed twice with 2.5 ml of phosphate-buffered saline, and the amount of bound l&in was determined in a Packard y counter. Each  lectin binding data suggest that Clones 1 and 1021 have a selective decrease in membrane sialic acid, since terminal sialic acid residues are known to block the binding of soybean agglutinin, R. communis agglutinin I and ricin (12,25) while facilitating the binding of wheat germ agglutinin (12,261. The data also suggest that Clone 13, by virtue of its decreased ability to bind R. communis agglutinin, is deficient in surface gala&se residues as well as sialic acid residues. The actual carbohydrate content of crude membrane preparations of parent and variant cells is shown in Table II. As predicted, membranes from Clones 1 and 1021 have a selective decrease in sialic acid content while Clone 13 membranes are strikingly deficient in both sialic acid and galactose. Since most of the membrane sialic acid in Chinese hamster ovary cells is linked to gala&se, the reduced sialic acid content of Clone 13 membranes is probably secondary to the gala&se deficiency. Three other clones had a partial deficiency of sialic acid, ranging from 50 to 70% of wild type levels. Further characterization of the membranes of these clones was not performed.
The carbohydrate alterations of all the clones affect both glycolipids and glycoproteins.

Glycolipids of the Variant
Clones -When total cellular lip- Wheat Germ Agglutinin-resistant Cells ids were analyzed for sugar content, Clone 1021 lipids were found to be deficient in sialic acid while Clone 13 lipids were deficient in both sialic acid and galactose (Table III). Fig. 3 depicts the thin layer chromatographic patterns of glycolipids labeled by growing the cells in the presence of [14Clgalactose and N-[G-"Hlacetyl-n-mannosamine.
The major labeled glycolipid in parental Chinese hamster ovary cells is sialyl-lactosylceramide (GM,) while lactosylceramide and glucosylceramide are present as minor components. No higher gangliosides were detected. In contrast, Clone 1021 cells contain primarily lactosylceramide, while Clone 13 cells contain largely glucosylceramide. In all instances, the ['4C]galactose-labeled material gave rise to an unidentified peak, termedX. Acid hydrolysis of this material followed by paper chromatography revealed that the 14C label is neither glucose nor galactose. The material cochromatographs with phosphatidylinositol in Solvent System B. Acid hydrolysis of the other glycolipid peaks yielded only 1111 labeled glucose and galactose. The labeled glycolipid pattern of Clone 1 cells is the same as that of Clone 1021.

Glycoproteins of Variant
Clones-In order to define differ- ences in glycosylation of the membrane glycoproteins, the various cell lines were grown in the presence of ["Hlfucose to label the oligosaccharide units of the glycoproteins. Glycopeptides were then prepared by pronase digestion of isolated crude membrane fractions. As shown in Fig, 4, the major fucosecontaining glycopeptides of Clones 1021 and 13 are significantly smaller than the glycopeptides of parent cells. The estimated molecular weights of the major parental Clone 1021 and Clone 13 glycopeptides are 3500, 2550, and 2000, respectively. Similar results were obtained when the glycopeptides were labeled with ["Hlglucosamine.
Evidence that the glycopeptides from the variant cell lines are deficient in specific sugar residues at their nonreducing termini is presented in Fig. 5. ["H]Glucosamine-labeled glycopeptides from parent and Clone 13 cells were treated with various glycosidases and the release of radioactivity was determined by gel filtration on a Bio-Gel P-2 column. Treatment of parent cell glycopeptides with P-N-acetylglucosaminidase alone resulted in the release of only 15% of the radioactivity, whereas treatment with neuraminidase followed by digestion with both P-galactosidase and P-N-acetylglucosaminidase caused 45% of the label to appear as small molecular weight material (Fig. 5, upper panel). Under these chromatographic conditions, sialic acid elutes with the void volume; all the labeled low molecular weight material is therefore free Nacetylhexosamine.
Treatment of Clone 13 glycopeptides with P-N-acetylglucosaminidase alone resulted in the release of 40% of the radioactivity, and treatment with all three glycosidases did not significantly increase the amount of labeled material released (Fig. 5   is normal with respect to the parent line. The possibility that impaired nucleotide sugar synthesis is responsible for the observed membrane sugar deficiencies can therefore be discounted. The increased CMP-sialic acid content of the sialic acid-deficient clones suggests that a block exists in the utilization of this nucleotide sugar and that therefore the defect in these clones probably involves a biosynthetic step. A similar block in the utilization of UDP-galactose by Clone 13 might not be expected to lead to accumulation of UDP-galactose due to the activity of UDP-glucose-4-epimerase. Sialyltransferase Activities in Parent Chinese Hamster Ovary and Clone 1021 Cells -Sialyltransferase activities of parent and Clone 1021 cells assayed using a variety of acceptors are shown in Table V 5.0 f 0.7 (7) 3.9 ? 0.7 (7) 7.4 (1) 0.17 k 0.05" (8) 0.04 k O.Ol* (8) 0.09 f 0.01~ (4) 0.04 k 0.01" (3) 0.12" (1) 0.18 (1) 0.16 k 0.02 (2) 0.74 k 0.02 (2) 1.9 f 0.6 (3) 1.7 (1) 2.7 k 0.5 (2) 1.6 +-0.1 (2) 1.5 f 0.5 (2) 2.5 2 0.6 (3) Glucosylceramide" 0.25 k 0.01 (2) 0.18 k 0.01 (2) 0.24 T 0.06 (2) " Monolayer cultures were incubated in growth medium containbuffered saline, pH 6.9, for 40 min at 37". ing incorporate similar amounts of IJHlsialic acid into endogenous surface acceptors. That this activity is a surface activity is based on the following criteria: (a) Hydrolysis of the CMP-[G-"HlNeuAc prior to incubation with the cells completely abolished sialic acid incorporation; (b) 95% of the cells excluded trypan blue, suggesting that they were intact; (c) 90% of the incorporated counts are released by neuraminidase treatment of intact cells; (d) disruption of the cells by sonication resulted in reduced transfer of sialic acid to endogenous acceptors; (e) neuraminidase pretreatment of parent cells is required for detectable incorporation but has no effect on incorporation in Clone 1021 cells. Efforts to restore the surface sialic acid content of the sialic acid-deficient clones or neuraminidasetreated parent cells by incubating the cells for extended periods with CMP-NeuAc failed. At most, 2 nmol of sialic acid were incorporated per mg of membrane protein. Replenishment of the CMP-NeuAc-containing medium did not increase incorporation.
CMP-NeuAc:glycoprotein sialyltransferase activity in Clone 1021 cell lysates ranged from 50 to 100% of parent levels. The reason for this fluctuation is not clear. Activity in both the parent and Clone 1021 cells exhibited a broad pH optimum with the peak at pH 6.9. To rule out the possibility that proteases released at the time of cell lysis activated the transferases, the protease inhibitor phenylmethylsulfonylfluoride was added to the cell suspensions prior to cell lysis. The inhibitor had no effect on glycoprotein sialyltransferase in Clone 1021 or parent cells.
We considered the possibility that the sialyltransferase activity assayed in vitro using exogenous glycoprotein acceptors may not be representative of the predominant enzyme which functions in intact cells. Methylation analysis of parent glycopeptides with or without neuraminidase digestion indicates that greater than 80% of the sialic acid residues linked to penultimate galactose residues are linked 2-3 while less than 20% are linked 2+6.' However, Clone 1021 cell lysates are capable of synthesizing both sialyl-(2-3)lactose and sialyl-(2-Q-lactose in amounts equivalent to that observed with parent cell lysates (data not shown).
The CMP-NeuAc:lactosylceramide sialyltransferase (GM, synthetase) activity of Clone 1021 cells is markedly decreased relative to parent cells, while Clone 1 cells exhibit wild type activity (Table V). GM:, synthetase activity in parent cells grown in suspension culture is twice that of cells grown in monolayer, while no such difference was observed in Clone 1021. Clone 1021 extracts exhibit a &fold decrease in V,,,,, relative to parental cell extracts (0.056 nmol/mg of protein/30 min uersus 0.25 nmol/mg of protein/30 min) (Fig. 6). When a mixture of parent and variant cell lysates was assayed, the activity was equal to the sum of the activities observed in the parent and variant cell lysates when assayed separately, suggesting that the reduced activity observed in variant cell lysates was not due to the presence of inhibitors or the absence of necessary cofactors.
The apparent K,,, values for CMP-NeuAc using various ac-2The major glycopeptide fraction was isolated from a pronase digest of chloroform:methanol (2:1)-extracted Chinese hamster ovary cell pellets by chromatography on a Bio-Gel P-6 column (see Fig. 4). This glycopeptide fraction, which contained 87% of the total membrane-bound sialic acid, was subjected to methylation analysis both with and without prior neuraminidase digestion. The glycopeptides were methylated by the method of Hakomori (30). The methylated sugars were identified and quantitated by the gas-liquid chromatography-mass spectrometric system devised by Bjorndal and Lundblad (31). ceptors are similar for parent and Clone 1021 enzymes (see Table VI). Induction of GM, Synthetase by Sodium Butyrate- Simmons et al. (32) have reported that growth of HeLa cells in the presence of millimolar amounts of sodium butyrate increases the specific activity of GM, synthetase 7-to 20-fold within 24 h. This enhanced enzyme activity is associated with a 3-fold increase in GM, content. To determine if there is a similar effect of butyrate on GM, synthetase activity of parent and Clone 1021 cells, and if the GM, content in Clone 1021 cells could be restored to wild type levels, parent and Clone 1021 cells were grown in monolayer cultures in the presence of 1 mM sodium butyrate for 48 h. During this period both parent and Clone 1021 cells grew to confluency or near confluency although their growth rates were slowed relative to controls, and both exhibited marked cellular elongation.
The results of a representative experiment are shown in Table VII. In each of 4 separate experiments, butyrate caused a marked increase in GM, synthetase activity as assayed in vitro in both cell lines but there was no significant effect on cellular GM, content. In parent cells, the limiting factor may be the low amount of endogenous lactosylceramide available in uivo (Fig. 3), but this is not the case for Clone 1021 cells since their major glycolipid is lactosylceramide.
Absolute val- agglutinin can be explained on the basis of their decreased membrane sialic acid content, which results in a decreased affinity of the lectin for membrane oligosaccharides which serve as binding sites. The role of sialic acid residues in wheat germ agglutinin binding has been demonstrated in a number of systems (12,26). However, the 250-fold resistance of Clone 13 to wheat germ agglutinin toxicity cannot be explained on this basis. Since Clone 13 has a 65% decrease in binding sites, it is possible that this clone has lost a special class of "productive" binding sites which are necessary for mediating wheat germ agglutinin toxicity at low lectin concentrations. Stanley et al. have described a variant phenotype of Chinese hamster ovary cells characterized by resistance to wheat germ agglutinin and increased sensitivity to ricin cytotoxicity (36). Although no membrane carbohydrate analyses or lectin binding data have been reported for the clones with this phenotype, their characteristics are very similar to those of Clones 1 and 1021.
The mechanism by which the membrane sugar alterations occur in these variant cells is not clear. It appears that these cells fail to carry out specific glycosyltransferase reactions in uivo despite the fact that they possess the appropriate nucleotide sugars, glycoprotein and glycolipid acceptors, and glycosyltransferases. This situation is in contrast to several other h&in-resistant cell lines where membrane sugar alterations can be accounted for by deficiencies of particular glycosyltransferases (3, 5-7). Also, a number of transformed cell lines have glycolipid changes directed toward simplification with decreased activities in specific glycosyltransferases consistent with the observed change (37). There are, however, a few examples where the in. vitro assays of glycosyltransferase activities do not correlate with the measured changes in membrane carbohydrate composition. In baby hamster kidney cells transformed by a temperature-sensitive polyoma virus mutant, both GM,, content and GM,, synthetase activity are significantly decreased at the permissive temperature.
At the nonpermissive temperature, however, GM,, content returns to normal but GM, synthetase activity remains low (37). In addition, Smith et al. have characterized a yeast mannan mutant which fails to transfer N-acetylglucosamine residues to mannotetraose side chains, yet which exhibits normal N-acetylglucosaminyltransferase activity in cell-free extracts (38). The synthesis of complex oligosaccharide chains of plasma membrane glycoproteins and glycolipids in mammalian cells takes place primarily in the membranes of the endoplasmic reticulum and the Golgi apparatus (39). Glycosylation, therefore, occurs within the two-dimensional framework of a membrane and the process is separated from events occurring in the cytoplasm and in other organelles by that membrane. While little is known about the topographical organization of the multiglycosyltransferase system, Arce et al. (40) have presented indirect evidence that the enzymes and lipid acceptors involved in ganglioside biosynthesis are in a rigid arrangement in the membrane. Winqvist and Dallner (41) have presented evidence that enzymes are arranged in specialized patches in the microsomal membrane. Any defect affecting the organization of the multiglycosyltransferase system, such as an altered subcellular localization of an enzyme, abnormal insertion of a component enzyme in the membrane, or impaired transport of the nucleotide sugar from the cytoplasm to the Golgi apparatus could impair enzyme function and yet be missed by assays involving cell disruption and the use of detergents. It is also possible that glycosyltransferase activity may be subject to tight regulation in uivo. The biochemical defect in the variant clones could affect such regulation so that the glycosyltransferase is permanently in the "off" position in uiuo although the regulatory mechanism may be inoperative under in vitro assay conditions.
It is of interest that wheat germ agglutinin is extremely toxic to Chinese hamster ovary cells, with an LD,, of approximately 1 pg/ml. The mechanism of wheat germ agglutinin toxicity is not known, although two recent reports have shown that it inhibits amino acid transport in tissue culture cells (42,431, suggesting that the lectin may exert its toxic effects by altering plasma membrane function. We have confirmed this finding and are currently studying the mechanism whereby wheat germ agglutinin inhibits amino acid transport in Chinese hamster ovary cells.