key: cord-0893694-gsf8ad65
authors: Jarvis, Donald L.; Butel, Janet S.
title: Modification of simian virus 40 large tumor antigen by glycosylation
date: 1985-03-31
journal: Virology
DOI: 10.1016/0042-6822(85)90250-8
sha: 2bb60c5dc8bdb5328cfb611ae328786e9e76433b
doc_id: 893694
cord_uid: gsf8ad65

Abstract The SV40-encoded transforming protein, large tumor antigen (T-ag), is multifunctional. Chemical modifications of the T-ag polypeptide may be important for its multifunctional capacity. T-ag is additionally modified by glycosylation. T-ag was metabolically labeled in SV40-infected cells with tritiated galactose or glucosamine, but not with mannose or fucose. The identity of glycosylated T-ag was established by immunoprecipitation with a variety of T-ag-specific antisera, including monoclonal antibodies. Incorporation of labeled sugar into T-ag was inhibited in the presence of excess unlabeled sugars, but not in the presence of excess unlabeled amino acids. Labeled monosaccharides could be preferentially removed from T-ag with a mixture of glycosidic enzymes. In addition, galactose was removed from purified T-ag by acid hydrolysis and identified as such by thin-layer chromatography. T-ag oligosaccharides were resistant to treatment with EndoH, and glycosylation was not inhibited by tunicamycin. Together, these data strongly suggest that T-ag is glycosylated. Several characteristics, including lack of mannose labeling, EndoH resistance, and tunicamycin resistance, suggest that T-ag is not an N-linked glycoprotein. Rather, these properties are more consistent with the identification of T-ag as an O-linked glycoprotein.

Simian virus 40 (SV40) encodes the synthesis of two early proteins, designated the small tumor antigen (t-ag) and the large tumor antigen (T-ag), that exhibit apparent molecular weights (MW) of about 20,900 and 90,000, respectively. Both are chemically modified polypeptides, and their entire amino acid sequences are known (Fiers et al, 1978; Reddy et &, 1978) . It is thought that T-ag mediates most of the events culminating in cellular transformation by SV40. Accordingly, a vast array of biochemical and biological functions has been attributed to this polypeptide (reviewed by Martin, 1981; Tooze, 1981; Rigby and Lane, 1983) . However, the molecular basis for its multifunctional capacity remains obscure.

One possible explanation invokes the known chemical modifications of the T-ag polypeptide. Theoretically, those modifi- ' Author to whom reprint requests should be addressed.

cations could generate distinct forms of T-ag that might perform different functions in the host cell. The modifications reported for T-ag include phosphorylation (Tegtmeyer et al, 1977) , N-terminal acetylation (Mellor and Smith, 19'78) , poly-ADP-ribosylation (Goldman et cd, 1981) , fatty acid acylation (Klockmann and Deppert, 1983) , and glycosylation (Schmidt-Ullrich et cd, 1977 . However, the evidence suggesting that T-ag possesses the latter modification (i.e., glycosylation) is both limited and indirect. Therefore, we have addressed this question by performing detailed biochemical studies. The results of these analyses establish that Tag is glycosylated and that its oligosaccharide moiety (or moieties) contains galactose and glucosamine and/or their derivatives. The oligosaccharides probably do not contain mannose or fucose and are resistant to treatment with endo+N-acetylglucosaminidase H (EndoH). In addition, glycosylation of T-ag is not inhibited by tunicamycin (TM), suggesting that it is not an N-linked, but rather, is probably an O-linked glycoprotein.

Glycosylation of viral transformation proteins has been recognized in only a few retrovirus systems and has not previously been conclusively demonstrated with any DNA tumor virus. All of the known retroviral transformation glycoproteins are N-glycosylated (Dresler et aL, 1979; Hayman et aL, 1983; Privalsky et cd, 1983) ; none has been shown to be Oglycosylated. Only a single viral glycoprotein, the El matrix glycoprotein of certain coronaviruses, is believed to be exclusively 0-glycosylated (Holmes et d, 1981; Niemann and Klenk, 1981) . Thus, the demonstration of 0-glycosylation of SV40 Tag provides a novel basis for further investigation of T-ag multifunctionality, particularly with respect to the significance of this modification. In addition, Tag may be a useful model for the study of O-linked glycoproteins in general, which remain relatively poorly characterized. MATERIALS AND METHODS CeZ.Zs and wirus. TC-7 cells (Robb and Huebner, 1973) were propagated in enriched Eagle's minimum essential medium (E-MEM; GIBCO, Grand Island, N. Y.; Noonan et al, 1976) . MmFjmt/cl cells (Owens and Hackett, 1972) , mammary tumor cells that produce mouse mammary tumor virus (MMTV), were propagated in Dulbecco's modified minimum essential medium (D-MEM, GIBCO; Slagle et d, 1984) . Wild-type SV40 was passaged at a low multiplicity of infection (m.o.i.) and plaque assayed in TC-7 cells (Noonan and Butel, 1978) . Vesicular stomatitis virus (VSV, Indiana strain) was passaged and titrated as described for SV40, except that it was not subjected to freeze-thawing. C3H mouse mammary tumor virus [(C3H)-MMTV] concentrates were provided by the Biological Carcinogenesis Branch, Division of Cancer Cause and Prevention, National Cancer Institute. Antisera Normal hamster serum (NHS), hamster ascites fluid containing antibodies against T-ag (HAF), normal rabbit serum (NRS), and rabbit antiserum against purified T-ag (RacuT) or disrupted (C3H)-MMTV (cYMMTV) have been previously described (Lanford and Butel, 1979; Slagle et aL, 1984) . Rabbit antiserum against purified VSV (RbaVSV) was generously provided by Dr. Trudy Morrison. T-ag or cellular protein p53-specific monoclonal antibodies used in this study included PAb 100, 101, and 122 (Gurney et cd, 1980) , PAb 204 Hoeffler, 1980), and PAb 402, 405, 414, 416, 419, 421, 423, and 430 (Harlow et aL, 1981) . Hybridomas were cultured and antibodies were precipitated from culture supernatants as previously described (Harlow et cd, 1981; Santos and Butel, 1984) .

Iqfeiticm and metabohc labeling. TC-7 monolayers were mock infected with E-MEM or infected with SV40 or VSV at an m.0.i. of 50 plaque-forming units (PFU)/cell (Jarvis et o& 1984) . For VSV infections, 1.5 pg/ml actinomycin D (Sigma Chemical Co., St. Louis, MO.) was included with the inoculum and maintained throughout the infection. At appropriate times postinfection (p.i.), monolayers were washed with Tris-buffered saline (TBS) and starved in glucose-free E-MEM for 30 min (Martineau et cd, 1972) . Cells were then labeled with ["S]methionine, D-[l-3H]galactose, ~-[l-~Hlglucosamine, D-[2-3H]mannose, or L-[~-~H]fucose (Amersham, Arlington Heights, Ill.). Labeling times and radioisotope concentrations are specified in the figure legends. Mm5mt/cl cells were labeled as previously described (Slagle et cd, 1984) .

Special labeling corditions. In some experiments, SV40-infected cells were labeled with pS]methionine or 3H-monosaccharides in the presence of excess unlabeled amino acids, excess unlabeled sugars, or TM. Normal (1X) amino acid concentrations in E-MEM were approximately: Trp, 0.1 mM; Met, 0.3 mM; His, 6.4 mM; Gly and Phe, 0.5 mM; Ser and Arg, 0.6 mM; Leu, Ile, Thr, and Val, 1.0 mM; Lys, 1.4 d;

and Gln, 3.0 mM. E-MEM and glucose-free E-MEM containing 5 times (5X), 10 times (10X), or 50 times (50X) the normal concentration of amino acids were prepared. For labeling in the presence of excess amino acids, infected cells were treated with E-MEM containing different amino acid concentrations be-ginning at 2 hr p.i. Cells were starved from 20.5 to 21 hr p.i. using glucose-free E-MEM supplemented with the same amino acid concentrations and were labeled for 3 hr using 25 &i/ml

[%S]methionine or 100 rCi/ml PHlgalactose in the same medium. E-MEM normally contained 1 mg/ml glucose. For labeling in the presence of excess unlabeled sugars, E-MEM also contained 1 mg/ml galactose for 1X sugar-enriched medium or higher concentrations of both sugars for 5X, 10X, and 50X sugar-enriched medium. Infected cells were starved in glucose-free E-MEM from 20.5 to 21 hr p.i., then labeled for 3 hr using 25 &i/ml [?S]methionine or 100 &i/ml rH]galactose in E-MEM containing different sugar concentrations. Finally, for TM experiments, E-MEM and glucose-free E-MEM containing various concentrations of TM were prepared. Treatment was initiated at 15 hr p.i. and maintained throughout the 30-min starvation and 3-hr labeling periods (20.5-24 h p.i.).

Extmctk, immuvwprecipitation, and gel electrophoresis. Labeled monolayers were washed, detergent-extracted in the presence of leupeptin (Jarvis et al, 1984) , and extracts were clarified. After incubation with an appropriate antiserum, formalinfixed Staphybcoccus aureus strain Cowan I (SACI) (Kessler, 19'75) was added to adsorb immune complexes. If monoclonal antibody was used, a goat anti-mouse IgG bridge was included. Immunoprecipitates were then washed three times with wash buffer (WB; 50 mM Tris, pH 8.0, 100 mM NaCl, 1% NP-40,1% sodium deoxycholate, and 0.1% SDS), disrupted, and analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE; Soule and Butel, 1979) . Gels were impregnated with Autofluor (National Diagnostics, Somerville, N. J.), dried, and exposed to X-ray film at -70".

Treatment of T-ag or MMTV poll/pep tides with gl~cosiduses. SV40-infected TC-7 cells or MmSmt/cl cells were glucose starved, doubly labeled with [%]methionine and PHlglucosamine, and extracts were immunoprecipitated with HAF or crMMTV and SACI. Immune complexes were washed three times with WB, then twice with TBS. Pellets were resuspended in TBS supplemented with 200 fl leupeptin (Sigma) and 1000 kIU Trasylol (Mobay Chemical Co., New York, N. Y.), mixed, and triplicate samples were removed for determination of pretreatment acid-precipitable 3H and ?S radioactivity. TBS, EndoH (Miles Laboratories, Inc., Elkhart, Ind.), or Char&a lampas mixed glycosidases (Miles) were added, and mixtures were incubated with shaking at 37" for 15 hr. Triplicate samples were then removed for determination of post-treatment acid-precipitable 3H and % radioactivity, and the remainder was processed for SDS-PAGE analysis. Pre-and posttreatment samples were disrupted, clarified, and spotted onto glass microfiber filters (Whatman, Clifton, N. J.). Filters were dried, then washed once with 10% trichloroacetic acid (TCA), twice with 5% TCA, and once with 95% ethanol. Filters were redried, then assayed for 3H and ?S radioactivity by liquid scintillation spectroscopy.

Acid hydrolysis of T-ag and analysis of monosaccharides. Glucose-starved, SV40infected TC-7 cells were doubly labeled with p5S]methionine and PHlgalactose, and T-ag was isolated by immunoprecipitation and SDS-PAGE. T-ag was excised, eluted from the gel, concentrated, and desalted on a Sephadex G25M column (Pharmacia Fine Chemicals, Uppsala, Sweden). The sample was then reconcentrated and SDS was extracted (Henderson et ul, 1979) . Carrier galactose was added, and the mixture was suspended in 2.0 N HzS04 and sealed under nitrogen in a glass ampule. After hydrolysis for 4 hr at lOO", the reaction mixture was diluted to 0.1 N H2S04 and applied to a Dowex 5OW-X8 cation-exchange column. The flowthrough was neutralized, clarified, and lyophilized. This material was then applied to a silica gel G thin-layer plate (Fisher Scientific Co., Pittsburg, Pa.) and chromatographed with n-propanol:water (71, v/v) in a closed chamber (Gal, 1968) . The plate was dried and stained with 30% aqueous ammonium bisulfate for visualization of the carrier and standard mono-saccharides. After measurement of Rfvalues, the lanes in which hydrolysates had been chromatographed were fractionated, and fractions were assayed for radioactivity by liquid scintillation spectroscopy.

Cells with Radioactive Mcmomcchurides

In initial experiments designed to determine if T-ag is glycosylated, we utilized metabolic labeling of SV40-infected cells with [%]methionine or tritiated monosaccharides.

A polypeptide of about 88,000 (88K) in apparent MW was detected by immunoprecipitation of pS]methionine-labeled, SV40-infected cell extracts with HAF (Fig.  lA, lane 3) . This polypeptide was identified as T-ag because it was not detected in mock-infected cell extracts immunoprecipitated with NHS or HAF or in infected cell extracts immunoprecipitated with NHS (lanes 1, 2, and 4). A corn&rating polypeptide, presumably T-ag, was also detected in PHlgalactose-(lane 8) and mlucosamine-labeled (lane 16), infected cell extracts immunoprecipitated with HAF; it was absent in the corresponding mock-infected or NHS-treated controls (lanes 5-7 and 13-15). Labeled T-ag was not recovered from cells pulsed with rH]mannose (lanes 9-12) or rH]fucose (lanes 17-20), although unlabeled T-ag was immunoprecipitated from infected cells with HAF as evidenced by Coomassie blue staining. Incorporation of rH]galactose into T-ag was enhanced when infected cells were pretreated with glucosefree E-MEM and the isotope was added in this same medium. This can be seen by comparing lanes 1 and 2 in Confluent TC-7 cell monolayers were washed with warm TBS, then mock infected with E-MEM (A: lanes l-2, 5-6, 9-10, 13-14, and 17-18) or infected with 50 PFWcell SV40 (A: lanes 3-4, '7-8, 11-12, 15-16, and 19-20; B: lanes 1 and 2). After adsorption for 2 hr at 37", the cells were fed with E-MEM and returned to 37". Except for the sample in B, lane 1, cells were starved prior to labeling for 30 min in glucose-free E-MEM. Cells were then labeled from 21 to 24 hr p.i. with 100 &i/ml 4,5, 7, 9.11,13, 15, 17, and 19) or HAF (A: lanes 2, 3, 6, 8.10, 12, 14, 16, 18, and 20; B: lanes 1 and, 2). Immune complexes were adsorbed using SACI, washed thoroughly, then disrupted and analyzed by SDS-PAGE on 8% acrylamide gels. Numbers to the left of the figure indicate the positions of molecular weight standards, designated by their molecular weights x lo-*. galactose and glucosamine were incorporated into T-ag, suggesting that its putative oligosaccharide moiety(ies) contains those sugars and/or their derivatives.

It was possible that the relatively short (3 hr) labeling time used in the initial experiments had not permitted equilibration of intracellular sugar pools with each input labeled monosaccharide, producing false negative results. Therefore, SV40infected cells were labeled for 3, 6, or 18 hr using tritiated galactose, glucosamine, or mannose, and samples were processed as before. Radioactive T-ag was detected in all of the galactose-labeled samples and in the 18-hr glucosamine-labeled sample, but was not detected in any of the mannose-labeled samples (Fig. 2) . These observations support the interpretation that putative T-ag oligosaccharide(s) contain galactose and glucosamine and/or their derivatives but lack mannose. It was not firmly established that the oligosaccharide(s) lack fucose, since that sugar was labeling of SV40 T-ag with various monosaccharides. Confluent TC-7 monolayers were infected with SV40 as described in the legend to Fig. 1 . Cells were labeled without prior glucose starvation from 6 to 24 hr p.i. (18 hr; lanes l-3), 18-24 hr p.i. (6 hr; lanes 4-6), or 21-24 hr p.i. (3 hr; lanes 7-9) with 20 &i/ml Djl-%Jgalactose (lanes 1, 4, 7), D-Cl-%Jglucosamine (lanes 2, 5, 8), or D-f% 'Hlmannose (lanes 3, 6, 9). Samples were then processed and analyzed as described in the legend to Fig. 1 , using HAF for immunoprecipitation and 8% acrylamide gels for analysis. Molecular weight markers are shown on the left. not included in this experiment. However, except for the 18-hr galactose-labeled sample, T-ag was not as intensely labeled in this experiment as in the experiment shown in Fig. 1A . Thus, shorter pulses with higher concentrations of radioactive sugars generally provided better incorporation than longer pulses with lower concentrations, suggesting that adequate uptake of the input sugars had occurred under those labeling conditions. Preliminary experiments using [35S]methionine labeling had established the maximal time of T-ag synthesis under our conditions of infection to be about 20-24 hr pi. (data not shown). This time period was chosen for the experiments described above since a large amount of T-ag would be available for labeling and the detection of monosaccharide-labeled T-ag would be optimized. However, if glycosylated T-ag represents a discrete subpopulation of Tag, then its maximal time of synthesis might differ from that of overall T-ag synthesis. Therefore, we examined the absolute and relative incorporation of [?S]methionine and [3Hlgalactose into Tag at various times p.i. The absolute incorporation of both isotopes was greatest at 21-24 hr p.i., suggesting that maximal glycosylation occurred at the same time as maximal overall T-ag synthesis (Table  1) . However, the relative incorporation of [3Hlgalactose compared to [35S]methionine was highest very early in infection, as shown by 3H:35S ratios (Table 1 ). The ratio decreased steadily until about 12 hr p.i., then remained approximately constant until 48 hr p.i. This result might indicate that relatively more T-ag molecules or more sites on T-ag are glycosylated early in infection. Although a large degree of variability in the 3H:35S ratios was observed among different experiments, the ratio repeatedly decreased with time after infection.

Positive Ident~catiim of the 88K Glvcoprotein as T-ag

In the experiments described above, an 88K glycoprotein was identified as T-ag because it was immunoprecipitated from T-ag was immunoprecipitated from clarified cell extracts, gel purified, and eluted from the gel. Concentrated eluents were then quantitated for "S and 'H content by double-label liquid scintillation spectroscopy.

bRaw counts per minute (cpm) were determined using two channels on a Beckman model LS-250 spectrometer, preadjusted to detect "S or %I. Raw cpm were corrected for relative efficiency of counting on those channels as compared to an open channel and, in the case of %I cpm, for a small amount of contamination with "S cpm. Correction factors were determined by counting singly labeled samples under conditions identical to those used for the experimental samples.

infected cell lysates with HAF and was absent in the corresponding mock-infected and NHS controls. However, since HAF is a polyclonal antiserum, exhibiting multiple immunologic reactivities, it was necessary to more conclusively identify this 88K polypeptide.

Several different T-ag-specific and control antisera were used for immunopre-cipitation of galactose-labeled, infected cell extracts (Fig. 3) . HAF, RaaT, and all 10 of 10 different monoclonal antibodies against T-ag immunoprecipitated the galactose-labeled 88K polypeptide. This polypeptide was not immunoprecipitated with NHS, NRS, or a control monoclonal antibody directed against human IgG. These results identify the 88K glycoprotein as T-ag, since it is extremely unlikely that another protein would possess that many cross-reactive antigenic determinants.

Mmosaccharicle Label

An important consideration in the interpretation of metabolic labeling experiments, particularly those involving monosaccharides, is the possibility that randomization of label may occur. The input monosaccharide might become metabolically converted to another form in the cell, such as amino acids, which could then be incorporated into the protein of interest. The following experiments were performed to evaluate this possibility.

Monoclonal antibody PAb 419 recognizes an antigenic determinant on the amino end of T-ag (Harlow et aL, 1981) . Since large T-ag and small t-ag share common amino terminal sequences, that antibody would coimmunoprecipitate both polypeptides from [35S]methionine-or [3HJgalactose-labeled infected cell extracts. Both should be labeled with galactose if randomization were occurring. In contrast, small t-ag would not be labeled if randomization of label did not occur and the glycosylation site(s) on large T-ag were located on a portion of the polypeptide distal to those sequences shared with small t-ag. As expected, PAb 419 immunoprecipitated both polypeptides from [35S]methionine-labeled infected cell extracts (Fig. 4) . Coomassie blue staining revealed that both large T-ag -and small t-ag were also immunoprecipitated from galactose-labeled infected cell extracts (data not shown). However, only large Tag was detectably labeled with PHIgalactose. This observation, although in- FIG. 3 . Immunoprecipitation of galactose-labeled SV40 Tag with various antisera. TC-7 cells were mock infected (-) or SV40 infected (+), glucose starved, and labeled with D-[l-?@&Wb3e as described in the legend to Fig. 1 . Cells were extracted, then clarified extracts were immunoprecipitated with different antisera. Immune complexes were adsorbed, washed, disrupted, and analyzed on 8% acrylamide gels. The headings at the top of the figure indicate the antisera used, numbers refer to PAb designations for T-ag and p53-specific monoclonal antibodies, CM refers to a control monoclonal antibody directed against human IgG. Other antisera abbreviations are described under Materials and Methods. Molecular weight markers are shown on the left. direct, suggests that randomization of the input sugar label had not occurred to a significant degree. In addition, it is likely that the glycosylation sites on large T-ag are not located within the region of the polypeptide encoded from nucleotides 5163 to 4917, since that is the region in common with small t-ag.

A more direct approach to the potential problem of randomization involved metabolic labeling of T-ag with [?S]methionine or ['Hlgalactose in the presence of excess unlabeled amino acids or sugars. The incorporation of pS]methionine should be inhibited by the presence of unlabeled amino acids but unaffected by the presence of unlabeled sugars. Conversely, if malactose were incorporated into T-ag in the form of sugar, that incorporation should be unaffected by unlabeled amino acids but inhibited by unlabeled sugars. The results with ["S]methionine labeling were as predicted; 63 and 82% inhibitions were observed when labeling was performed in the presence of 5X and 10X amino acids, respectively (Fig.  5A, Table 2 ). No inhibition was observed with up to 10X sugars; instead, [35S]methionine incorporation was slightly enhanced with increasing sugar concentration (Fig.  5B, Table 2 ). At 50X sugars, methionine incorporation was strongly inhibited, for unknown reasons. Perhaps uptake of metabolites from the growth medium was inhibited at this sugar concentration. Coomassie blue staining revealed little difference in the amount of T-ag synthesized in cells treated with various sugar concentrations (data not shown). For FHkalactose incorporation, excess unlabeled amino acids had little effect, except for a slight inhibition at the 10X level (Fig. 5A, Table 2 ). Coomassie blue staining revealed a slight reduction in the amount of T-ag immunoprecipitated from cells treated with 10X amino acids (data not shown). In addition, cells treated with higher (50X) levels of amino acids were killed. Thus, it is likely that the slight inhibition of PHJgalactose incorporation observed in the presence of 10X amino acids was due to minor toxicity and reflected a decrease in overall T-ag synthesis. Incorporation of PHlgalactose into T-ag was significantly inhibited in the presence of unlabeled sugars (Fig. 5B) ; 90, 95, 96, and 86% inhibitions were observed for 1X, 5X, 10X, and 50X sugars, respectively (Table 2). These results strongly suggest that T-ag was specifically labeled with PHJgalactose due to glycosylation events, rather than simply due to randomization of the input sugar label.

Another possibility not addressed in . After extraction and clarification, PAb 419 was used for immunoprecipitation. Immune complexes were harvested, washed, disrupted and analyzed on 12% acrylamide gels. T and t on the right-hand side mark the positions of large and small tumor antigens, respectively. Molecular weight markers are shown on the left. the randomization control experiments described above is that [3H14galactose could have been metabolically converted to labeled nucleotides. In fact, galactose can be converted to ribose through the phosphogluconate pathway and could then be incorporated into T-ag as poly-ADP-ribose or RNA oligonucleotides. These products are found in covalent or tight association, respectively, with T-ag (Goldman et a& 1981; Khandjian et oL, 1982) . However, this possibility was precluded by using galactose tritiated in the number 1 position as the input sugar. That carbon and its hydrogens are lost in the early steps of the phosphogluconate pathway before ribose is produced. Thus, nucleotides con-taining ribose generated from D#-~H]galactose would not be radioactive, and labeled T-ag would not be detected as a result of such a conversion.

Another approach undertaken to substantiate that T-ag is modified by glycosylation involved treatment with a mixture of glycosidic enzymes. T-ag was immunoprecipitated from doubly labeled, 1 and 4) , 5X (lanes 2 and 5), or 10X (lanes 3 and 6) amino acids, as described under Materials and Methods. Cells were labeled with 25 &i/ml pS]methionine (lanes l-3) or 100 &i/ml PI-Dgalactose (lanes 4-6), extracted, and Tag was immunoprecipitated from clarified extracts with HAF. Disrupted immunoprecipitates were analyzed on 8% acrylamide gels. (B) Glucose-starved, SV40-infected TC-'7 cells were labeled in the presence of OX (lanes 1 and 6), 1X (lanes 2 and 7), 5X (lanes 3 and 8), 10X (lanes 4 and 9), or 50X (lanes 5 and 10) glucose-and galactose-containing E-MEM. Cells were labeled with 25 &i/ml pS]methionine (lanes l-5) or 100 &i/ml [sHlgalactose (lanes 6-lo), then processed and analyzed as described for A. Molecular weight markers are shown on the left. "T-ag bands were excised from each of the gel lanes shown in Fig. 5 . Gel slices were solubilized by the method of Mahin and Lofberg (1966) infected cell extracts, and the immune complexes were adsorbed to SACI, washed, and treated with TBS or various amounts of mixed glycosidases, as described under Materials and Methods. Aliquots of the reaction mixture were removed before and after treatment and assayed for acidprecipitable 3H and Y!! radioactivity ( Table 3 ). A small decrease in acid-precipitable 3H and 35S radioactivity was observed in the control samples after incubation with TBS for 15 hr at 37". This was probably due to nonspecific degradation or some autoproteolytic activity on the part of Tag, as proposed by Seif (1982) . Incubation in the presence of increasing amounts of mixed glycosidases resulted in a small loss of acid-precipitable as radioactivity, probably reflecting contamination of the enzyme preparation with proteases. However, we observed a much more marked reduction in the recovery of acid-precipitable 3H radioactivity from the same samples; the decrease was greater than 60% relative to the TBS control at the lowest concentration of enzyme.

This result suggests that the input 3H label was incorporated into T-ag as a monosaccharide, since it was preferentially removed from T-ag with glycosidic enzymes, and supports the interpretation that T-ag is glycosylated.

Rewwval and Recovery of Gahctose from T-ag The most direct evidence for glycosylation would require removal of sugar from "SV40-infected TC-7 cells were glucose starved, then labeled from 21 to 24 hr p.i. with 100 &i/ml each of CJ6Sjmethionine and n-[l-%Ilglucosamine.

T-ag was immunoprecipitated with HAF and SACI, and immunoprecipitates were washed with WB and TBS. SACI-bound immune complexes were then resuspended in TBS and treated with 0, 0.5, 1.0, 1.5, or 2.0 mg of mixed giycosidaaes for 15 hr at 37O. Aliquots of each sample were removed before and after treatment, TCA precipitated, and analyzed for ?S and eH radioactivity.

basS and 'H cpm were determined as described in the legend to Table 1. purified T-ag, followed by recovery and identification of that sugar. Again, T-a@; was doubly labeled, then purified as described under Materials and Methods and hydrolyzed in 2.0 N HzS04 for 4 hr at 100" in the presence of carrier galactose. This treatment was expected to preferentially release neutral monosaccharides from T-ag with minimal hydrolysis of the polypeptide chain. The hydrolysate was then subjected to cation-exchange chromatography, and the flow-through was neutralized and lyophilized. Analysis of the sample before and after cation exchange revealed that most of the ?3labeled material had been removed from the hydrolysate (Table 4 ). This was expected, since positively charged %-labeled protein, peptides, and amino acids would bind to the column matrix in the presence of acid. The flow-through was comprised primarily of %-labeled material, assumed to be galactose since it was the input 3H label and would not be expected to bind to the column matrix.

This assumption was verified by thinlayer chromatography of the flow-through material. Standard galactose reproducibly migrated to a position corresponding to fraction 28 in this thin-layer system (Fig.  6) . The carrier galactose present in the hydrolysate also migrated to fraction 28, as revealed by ammonium bisulfate staining. The distribution of 3H radioactivity after chromatography of the hydrolysate is shown in Fig. 6 . Two peaks were observed: a smaller peak at the point of sample application, and a larger peak which coincided with the position of galactose (fraction 28). The small amount of ?3-labeled material remaining in the hydrolysate was found at the origin; no %3 radioactivity comigrated with the galactose carrier. These results suggested that the majority of the 3H-labeled material recovered from purified T-ag was gala&se and provided direct evidence that T-ag is glycosylated. We noted, however, that different neutral monosaccharides were not particularly well-resolved in this TLC T-ag was isolated by immunoprecipitation and SDS-PAGE, then excised and eluted from the gel. After concentration, the sample was desalted by gel filtration, SDS was extracted, and acid hydrolysis was performed in the presence of carrier galactose, as described under Materials and Methods. The sample was diluted, aliquote were removed for liquid scintillation spectroscopy, and the remainder was applied to a small column (about 5 ml packed volume) of Dowex 5OW-X8. The flow-through was collected, and aliquots were again removed for liquid scintillation spectroscopy.

*&S and ?I-I cpm were determined as described in the legend to Table 1 . Posthydrolysis = samples taken before application to the Dowex column; post-ion exchange = samples taken from column flow-through. was purified, acid hydrolyzed, and subjected to cation-exchange chromatography as described in the legend to Table  4 . The flow-through was neutralized, concentrated, and analyzed by thin-layer chromatography as described under Materials and Methods. The carrier and standard sugars (denoted by arrows in the figure) were visualized by staining with 30% aqueous ammonium bisulfate. The radioactivity profile was determined by dividing the lane containing the hydrolysate into 0.5-cm fractions and analyzing each fraction by liquid scintillation spectroscopy.

system. Since the peak of radioactivity in that area was somewhat broad, we cannot discount completely the possibility that A 0 5 10 15 67-galactose may have been converted to other sugars that were then incorporated into T-ag.

Experiments were designed to attempt to generate nonglycosylated T-ag that could be used to assess the role of glycosylation in T-ag function. TM and EndoH treatments were chosen because their effects are well-characterized and both reagents are readily available in purified form. The success of such treatments depended, of course, upon the structural characteristics of the glycoprotein in question.

To analyze the effect of EndoH, SV40infected cells were labeled with [?3]methionine and T-ag was immunoprecipitated with HAF and SACI. Immune complexes were washed, treated with TBS or various amounts of EndoH, disrupted, and analyzed by SDS-PAGE (Fig. 7A) . No change in the electrophoretic mobility of T-ag was observed after EndoH treatment, suggesting that it was resistant to the effects of this endoglycosidase. Control experiments utilizing a known EndoHsensitive glycoprotein, gp52 of MMTV (Dickson and Atterwill, 1980) that the treatment protocol was effective, even at the lowest concentration of enzyme used (Fig. 7B) . It should be noted that EndoH treatment did not alter the electrophoretic mobility of the EndoH-resistant MMTV glycoprotein, gp36, or the nonglycosylated MMTV protein, ~28.

To determine the effect of TM treatment, SV40-infected cells were treated with E-MEM or various concentrations of TM starting at 15 hr pi., then were glucose-starved and labeled with pS]methionine or THlglucosamine from 21 to 24 hr p.i. After extraction, T-ag was immunoprecipitated and analyzed by SDS-PAGE (Fig. 8A) . The mobility and intensity of labeling of T-ag with either isotope was unchanged in the presence of TM, suggesting that the glycosylation events i., then E-MEM containing 0, 0.5, 1.0, or 1.5 rg/ml TM was added, and the cells were returned to 3'7". After 30 min starvation in glucose-free E-MEM, cells were labeled from 21 to 24 hr p.i. with 100 pCi/ml [86Sjmethionine or ~-[l-*HJglucosamine. All solutions were supplemented with TM during starvation and labeling. Cells were then extracted, and extracts were clarified, immunoprecipitated with HAF and SACI, disrupted, and analyzed by SDS-PAGE on 8% acrylamide gels. (B) VSV-infected TC-7 cells were TM-treated, glucosestarved, and labeled as in A, except that treatment began at 1 hr p.i., labeling was performed from 7 to 10 hr p.i., and VSV-infected cell lysates were immunoprecipitated using RhoVSV. were resistant to this inhibitor. Control experiments utilizing the TM-sensitive G glycoprotein of VSV (Leavitt et ak, 1977) established that TM treatment was effective in TC-7 cells even at the lowest concentration of inhibitor used (Fig. 8B ). In addition, incorporation of [SHjglucosamine into total cellular protein was effectively inhibited by TM treatment (54-61%), with minimal effect evident upon overall protein synthesis (O-18%; Table 5 ).

This study has established that T-ag synthesized in SV40-infected cells is modified by glycosylation. Metabolic-labeling experiments revealed that galactose and glucosamine could be incorporated into Tag while mannose and fucose could not. The identity of T-ag was confirmed by its reactivity with a variety of T-ag-specific antisera. It should be noted, however, that T-ag was not strongly labeled with either galactose or glucosamine. Many of our fluorograms required exposure times of l-4 weeks, and those were obtained only after we had optimized the conditions used for infections and labeling. It was crucial to infect cells at a high m.o.i., to prestarve cells with glucose-free medium, and to label with a high concentration of tritiated sugar added in the same medium at the maximal time of T-ag synthesis. In view of these limitations and of the resistance of glycosylated T-ag to treatments with EndoH and TM (see below), the possibility of randomization of input labeled sugar posed a difficult problem in the interpretation of these results. We therefore rigorously explored that possibility.

Several different lines of evidence substantiated that randomization alone could not account for the monosaccharide labeling of T-a@;. First, only two of the four monosaccharides tested were incorporated into T-ag. Mannose, which was not detectably incorporated, can enter glycolysis after conversion to fructose-6-phosphate, and could have generated labeled amino acids through the tricarboxylic acid cycle (Lehninger, 1982) . Second, only large Tag was labeled with monosaccharides; small t-ag synthesized in the same cells was not. More direct evidence was obtained by labeling T-ag with THlgalactose in the presence of excess unlabeled amino acids or sugars. Incorporation of the label was clearly inhibited by the presence of unlabeled sugars but was unaffected by the presence of unlabeled amino acids. In addition, treatment of doubly labeled Tag with a mixture of glycosidases resulted in the preferential removal of 3H radioactivity, suggesting that 3H had been incorporated in the original input form of sugar. Finally, we were able to remove and identify the sugar from purified Tag. After acid hydrolysis of [3HJgalactoselabeled T-ag, the major peak of %I radioactivity cochromatographed with carrier galactose. Taken together, these results strongly refute the possibility that T-ag was labeled with monosaccharides only after their metabolic conversion to some other form.

Thus, other explanations must be entertained for the relatively low level of incorporation of galactose or glucosamine into T-ag. One is probably the low specific activity of the tritiated monosaccharides we used for labeling. Those were commer-cially available at only 2.4-25 Ci/mmol, as compared to 1000-1500 Ci/mmol for [%3]methionine. Another possible explanation could be that only a specific subpopulation of T-ag is glycosylated. However, preliminary experiments from our laboratory suggest that this is not the case; rather, glycosylation appears to be a characteristic of T-ag in general (Jarvis and Butel, unpublished observations) . We are currently investigating other possible explanations. For example, the number of oligosaccharides attached to T-ag, or the number of sugar residues per oligosaccharide, or both could be quite small.

The effects of EndoH and TM on glycosylation of T-ag were analyzed for preliminary characterization of the glycosylation events involved. EndoH cleaves highmannose N-linked oligosaccharides distal to the linkage sugar, N-acetylglucosamine (Tarentino and Maley, 1974) . TM is known to inhibit the initial glycosylation event in N-linked glycoproteins (Takatsuki et a& 1975) . Thus, these agents would exert their effects on T-ag only if it were an Nlinked glycoprotein. This was possible, since there is a consensus N-glycosylation site (Asn-Arg-Thr; Hunt and Dayhoff, 1970) at amino acid residues 156-158 of T-ag. However, treatment with neither agent changed the electrophoretic mobility of [%]methionine-labeled Tag, even though control experiments verified the efficacy of both treatment protocols. Because inhibition of glycosylation or removal of oligosaccharides might not be accompanied by a detectable change in electrophoretic mobility, TM experiments were also performed using PHfglucosamine-labeled T-ag. No change in the intensity of labeling of T-ag was observed. These results suggest that T-ag is not an N-linked glycoprotein. This is consistent with our inability to label T-ag using mannose, since that sugar is a common constituent of N-linked glycoproteins. Thus, T-ag provides another example of a protein in which a consensus N-glycosylation site is not used, suggesting that the site is necessary, but not sufficient, for Nglycosylation (Hunt and Dayhoff, 1970) .

T-ag is more probably an O-linked gly-coprotein. This interpretation is compatible with the apparent absence of mannose, with EndoH resistance and with TM resistance. It is also compatible with the low level of monosaccharide incorporation, as discussed above, as the oligosaccharide moieties of many O-linked glycoproteins are small. For example, one of the oligosaccharides of the El glycoprotein of mouse hepatitis virus contains only a single residue each of sialic acid, galactose, and N-acetylgalactosamine; the other has only an additional sialic acid residue (Niemann et c& 1984). We have attempted to verify the identification of T-ag as an Olinked glycoprotein by mild alkaline /3elimination (Spiro, 1966) . However, this approach has not been successful because the T-ag polypeptide appears to be degraded under those conditions (0.05M NaOH, 1.0 M NaBH4, 15 hr at 45"; Jarvis and Butel, unpublished observations). In any case, some controversy exists over the use of this method to distinguish between N-and O-linked glycoproteins (Rasilo and Renkonen, 1981; Ogata and Lloyd, 1982) . A better method would involve treatment of T-ag with acetylgalactosamine oligosaccharidase, which should cleave O-linked oligosaccharides. Unfortunately, this enzyme is no longer commercially available. One of the most interesting characteristics of T-ag is its ability to mediate a large number of different functions in infected and transformed cells (Tooze, 1980; Martin, 1981; Rigby and Lane, 1983) . While it remains difficult to explain its multifunctional nature at a molecular level, the structural characteristics of Tag reveal some intriguing possibilities. One explanation involves the presence of several unique structural domains on the polypeptide, with each domain performing one or more distinct functions. In fact, different T-ag functions have been localized to specific portions of the polypeptide (Rigby and Lane, 1983) . Another possibility emphasizes the localization of T-ag to distinct subcellular compartments in the host cell. Although the majority of T-ag is localized within the nucleus (Pope and Rowe, 1964, Rapp et d, 1964) , small amounts are associated with the plasma membrane (Tevethia et a& 1965; Soule and Butel, 1979; Deppert et aL, 1980; Chandrasekaran et c& 1981; Santos and Butel, 1982; Soule et d, 1982) and mitochondria 19'77) . T-ag species localized within different subcellular compartments might exhibit unique activities, due to putative structural differences and/or the influence of their local environments. A third model invokes the ability of T-ag to form different supramolecular complexes (Prives et a& 1979; McCormick and Harlow, 1980; Fanning et ale, 1981) . These complexes could represent distinct subpopulations of T-ag capable of mediating different functions in the host cell. It is clear that some supramolecular forms exhibit distinct DNA-binding and ATPase properties (Bradley et o& 1982; Fanning et aL, 1982; Gidoni et ok, 1982) . The existence of discrete T-ag subpopulations is also suggested by the finding that certain monoclonal antibodies react with only a subset of the total T-ag present in host cells (Gurney et aL, 1980; Scheller et cd, 1982) .

Thus, the functional diversity of T-ag might be explained by its structural diversity. Structural variability might, in turn, be mediated by combinations of different chemical modifications. Glycosylation could play an important role in generating structurally and functionally discrete forms of T-ag. The study of glycoproteins from other systems has established that glycosylation influences their structure and/or function. For example, N-glycosylation of fibronectin and acetylcholine receptor is crucial to their structural integrity (Olden et al, 1982) . In the absence of glycosylation, those proteins were more susceptible to proteolytic degradation. In another example, the specific nuclear matrix association of certain high mobility group proteins was precluded in the absence of glycosylation (Weintraub et ak, 1983) . This finding is particularly interesting, since some T-ag normally resides in association with the nuclear matrix of host cells (Staufenbiel and Deppert, 1983) . Further, it has been shown that glycosylation is required for the proper supramolecular assembly of thyroid-stimulating hormone (Reeves and Chang, 1983) , suggesting that glycosylation might be important for the supramolecular assembly of T-ag subpopulations.

It should be noted that O-linked glycoproteins remain relatively poorly characterized. In contrast to N-linked glycoproteins, little is known of their biosynthesis or function. It is possible that T-ag will provide a good model for the study of this class of macromolecules.

Relationship of oligomerization to enzymatic and DNA-binding properties of the SV40 large T antigen

Surface proteins of simian-virus-40-transformed cells

Simian virus 40 T-antigen-related cell surface antigen: Serological demonstration on simian virus 40-transformed monolayer cells in situ

Structure and processing of the mouse mammary tumor virus glycoprotein precursor Pr73

Glycoprotein encoded by the Friend spleen focus-forming virus

Detection and characterization of multiple forms of simian virus 40 large T antigen

Subclasses of simian virus 40 large T antigen: Differential binding of two subclasses of T antigen from productively infected cells to viral and cellular DNA

Complete nucleotide sequence of SV40 DNA

Separation and identification of monosaccharides from biological materials by thinlayer chromatography

Different forms of simian virus 40 large tumor antigen varying in their affinities for DNA

Modification of SV40 T antigen by poly-ADPribosylation

Monoclonal antibodies against simian virus 40 T antigens: Evidence for distinct subclasses of large T antigen and for similarities among nonviral T antigens

Monoclonal antibodies specific for simian virus 40 tumor antigens

Identification and characterization of the avian erythroblastosis virus e&B gene product as a membrane glycoprotein

A micromethod for complete removal of dodecyl sulfate from proteins by ion-pair extraction

Tunicamycin resistant glycosylation of a coronavirus glycoprotein: Demonstration of a novel type of viral glycoprotein. fi-115

The occurrence in proteins of the tripeptides Asn-X-Ser and Asn-X-Thr and of bound carbohydrate

Structural comparisons of wild-type and nuclear transport-defective simian virus 40 large tumor antigens. Fir-134

Rapid isolation of antigens from cells with a staphylococcal protein A-antibody adsorbent: Parameters of the interaction of antibody-antigen complexes with protein A

Simian virus 40 large tumor antigen: A "RNA binding protein

Acylation: A new post-translational modification specific for plasma membrane-associated simian virus 40 large T-antigen

SV40 large T shares an antigenic determinant with a cellular protein of molecular weight 68,000

Antigenic relationship of SV40 early proteins to purified large T polypeptide. Fir-97

Tunicamycin inhibits glycosylation and multiplication of Sindbis and vesicular stomatitis viruses

Principles of Biochemistry

A simplified method of sample preparation for determination of tritium, carbon-14, or sulfur-35 in blood or tissue by liquid scintillation counting

The transformation of ceil growth and transmogrification of DNA synthesis by simian virus 40

Enhancement of hexose entry into chick fibroblasts by starvation: Differential effect on gala&se and glucose

Association of a murine 53,000 dalton phosphoprotein with simian virus 40 large T antigen in transformed cells

Characterization of the amino-terminal tryptic peptide of simian virus 40 small-t and large-T antigens

The carbohydrates of mouse hepatitis virus (MHV) A59 Structures of the 0-glycosidically linked oligosaccharides of glycoprotein El

Coronavirus glycoprotein El, a new type of viral glycoprotein

Characterization of simian cells transformed by temperature-sensitive mutants of simian virus 40

Temperaturesensitive mutants of simian virus 40. I. Isolation and preliminary characterization of B/C gene mutants

Mild alkaline borohydride treatment of glycoproteins-A method for liberating both N-and O-linked carbohydrate chains

Function of the carbohydrate moieties of glycoproteins

Tissue culture studies of mouse mammary tumor cells and associated viruses

Detection of specific antigen in SV40-transformed cells by immunofluorescence

The product of the avian erythroblastosis virus e&B locus is a glycoprotein

DNA binding and sedimentation properties of SV40 T antigens synthesized in viro and in vitro

Virus-induced intranuclear antigen in cells transformed by papovavirus SV40. Prm Sot Exp

Mild alkaline borohydride treatment liberates N-acetylglucosamine-linked oligosaccharide chains of glycoproteins

The genome of simian virus 40

Investigations of the possible functions for glycosylation in the high mobility group proteins. Evidence for a role in nuclear matrix association

Structure and function of simian virus 40 large tumor antigen

Effect of cell chromosome number on simian virus 40 replication

Association of SV40 large tumor antigen and cellular proteins on the surface of SV40-transformed mouse cells

Antigenic structure of simian virus 40 large tumor antigen and association with cellular protein p53 on the surfaces of simian virus IO-infected and -transformed cells

A small subclass of SV40 T antigen binds to the viral origin of replication

Antigenic distinctions of glycoproteins in plasma and mitochondrial membranes of lymphoid cells neoplastically transformed by simian virus 40

New properties of simian virus 40 large T antigen

Expression of mammary tumor virus proteins in preneoplastic outgrowth lines and mammary tumors of BALB/cV mice

Subcellular localization of simian virus 40 large tumor antigen

Detection of simian virus 40 surface-associated large tumor antigen by enzyme-catalyzed radioiodination

Characterization of carbohydrate unite of glycoproteins

Different structural systems of the nucleus are targets for SV40 large T antigen

Inhibition of biosynthesis of polyisoprenol sugars in chick embryo microsomes by tunicamycin

Purification and properties of an endo-&N-acetylglucosaminidase from Streptomyces griseus

Modification of simian virus 40 protein A

New surface antigen in cells transformed by simian papovavirus SV40

Molecular Biology of Tumor Viruses

Glycosylation of thyroid-stimulating hormone in pituitary tumor cells: Influence of highmannose oligosaccharide units on subunit aggregation, combination, and intracellular degradation

This investigation was supported in part by Research Grant CA 22555 and by National Research Service Award CA 09197 from the National Institutes of Health. The authors thank Dr. Marion Steiner for helpful advice on carbohydrate chemistry.