Abnormal Glycosylation of Human Fibronectin Secreted in the Presence of Monensin*

Detailed studies of the effects of the ionophore mo- nensin upon the glycosylation of secreted fibronectin have been carried out. Human fibroblasts in culture were incubated in 1 monensin for several hours, following which radiolabeled glucosamine or mannose was added to the cultures. Parallel incubation and la- beling of control cultures were done. Labeled fibronectin was isolated from the culture media by gelatin- Sepharose chromatography, from cell surfaces by urea extraction, and from intracellular locations by cell lysis followed by immunoprecipitation. Detailed comparison of the glycopeptides released from fibronectin by pro- nase and of the oligosaccharides liberated by hydrazinolysis was carried out, particularly focusing on the secreted fibronectin, using gel filtration, high performance liquid chromatography, and concanavalin A chro- matography, in conjunction with the use of endoglyco-sidase H and specific exoglycosidases. We demonstrate that fibronectin in the medium of monensin-treated cultures differs in its glycosylation pattern from the control fibronectin. High mannose oligosaccharides are abundant in the monensin-derived fibronectin, whereas the control protein contains primarily complex oligosaccharides. Monensin apparently does not alter the initial glycosylation of fibronectin since the high mannose oligosaccharides are present on both control and monensin-treated intracellular fibronectin. We suggest, therefore, that monensin,

Detailed studies of the effects of the ionophore monensin upon the glycosylation of secreted fibronectin have been carried out. Human fibroblasts in culture were incubated in 1 monensin for several hours, following which radiolabeled glucosamine or mannose was added to the cultures. Parallel incubation and labeling of control cultures were done. Labeled fibronectin was isolated from the culture media by gelatin-Sepharose chromatography, from cell surfaces by urea extraction, and from intracellular locations by cell lysis followed by immunoprecipitation. Detailed comparison of the glycopeptides released from fibronectin by pronase and of the oligosaccharides liberated by hydrazinolysis was carried out, particularly focusing on the secreted fibronectin, using gel filtration, high performance liquid chromatography, and concanavalin A chromatography, in conjunction with the use of endoglycosidase H and specific exoglycosidases. We demonstrate that fibronectin in the medium of monensin-treated cultures differs in its glycosylation pattern from the control fibronectin. High mannose oligosaccharides are abundant in the monensin-derived fibronectin, whereas the control protein contains primarily complex oligosaccharides. Monensin apparently does not alter the initial glycosylation of fibronectin since the high mannose oligosaccharides are present on both control and monensin-treated intracellular fibronectin. We suggest, therefore, that monensin, by impairing intracellular translocation through the Golgi region, allows incompletely processed forms of fibronectin to reach the cell surface and to be released into the culture medium.
The monovalent ionophore monensin has been shown to inhibit the secretion of macromolecules from several cell types (1-6). Ultrastructural evidence (1,3,5), autoradiography (l), and the kinetics of secretion in inhibited cells (4) suggest that the primary site of action of monensin is within the Golgi complex. The block on secretion results in the intracellular accumulation of secretory products (1,5) but is not absolute, and some portion of synthesized material is released from a This research was supported by Grant AM-17220 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18   11 To whom requests for reprints should be addressed. inhibited cells. Passage of these molecules through the monensin-affected cells has been shown to be accompanied by abnormalities in their post-translational modification. This effect has been most completely characterized in the case of dramatic undersulfation of proteoglycans secreted from chicken embryo chondrocytes in the presence of monensin (6, 7). The reduced sialylation of lymphoid cell surface glycoproteins (8) and lowered incorporation of radiolabeled galactose and fucose into secreted immunoglobulin M molecules (9) in monensin-treated cells have suggested that the intracellular processing of N-asparagine-linked oligosaccharides is also altered in the presence of monensin. These data (8,9) are consistent with a monensin effect on the relatively "late" addition of sugars such as sialic acid, galactose, and fucose to oligosaccharides (for a review of this subject, see Ref. 10). In addition, our preliminary results have suggested that fibronectin and procollagen secreted from human fibroblasts in the presence of monensin exhibit a greater incorporation of mannose than do control molecules (11). Such results suggest a more complex and possibly earlier action of monensin. In view of the present interest relating the structure of N-asparaginelinked oligosaccharides to the subcellular sites of their processing, we have further investigated the structure of oligosaccharides present on human fibroblast fibronectin which has been secreted in the presence of monensin. The kinetics of secretion and the intracellular accumulation of this glycoprotein in monensin-treated cells have been described previously (4, 5, 12).

EXPERIMENTAL PROCEDURES
Cell Culture and Metabolic Labeling-Normal adult human skin fibroblasts (American Type Culture Collection, CRL 1220) were maintained in Dulbecco's modified Eagle's medium (Gibco) containing 50 pg/ml of ascorbate and 10% horse serum (Gibco). For experiments, fully confluent cell layers were split 12 and used 3-4 days later when each dish was overconfluent and contained about 2 X lo6 cells. Preincubation of cultures was in 3 ml of Dulbecco's modified Eagle's medium containing 2% dialyzed horse serum with or without monensin at 1 p~. After 3-4 h, this medium was exchanged for Dulbecco's modified Eagle's medium containing 10% of the normal glucose concentration, 2% dialyzed horse serum, 1 ~L M monensin as required, and radiolabeled sugars. The latter, obtained from ICN and New England Nuclear, were [6-3H]glucosamine/HC1 (20 Ci/mmol), [I-"C]glucosamine/HCl (10 mCi/mmol), [2-3H]mannose (10 Ci/mmol), or [1-"C] mannose (50 mCi/mmol). Tritiated compounds were used at 15 pCi/ ml and '%-labeled sugars at 5 pCi/ml. To permit the eventual mixing of oligosaccharides derived from control and treated cultures, controls were usually incubated with the 14C-labeled sugar and treated cultures with the tritiated sugar. Affinity-purified fibronectins were then mixed, prior to enzymatic or chemicai digestion, such that the "H counts per min were at least twice as high as "C counts per min. In one series of experiments, labeled medium was removed from control and monensin-treated cultures under sterile conditions, mixed, and incubated at 37 "C for 8 h. Fibronectin was then isolated from this mixture and parallel preparations of unmixed culture media.
Isolation of Fibronectin-Medium was collected after 12-h incu-547 of Human Fibronectin bation, chilled, and brought to 1 mM phenylmethylsulfonyl fluoride, 5 mM N-ethylmaleimide. and 10 mM EDTA (=protease inhibitors). After centrifugation of the medium (10,OOO X R. 10 min). fibronectin was isolated by gelatin affinity chromatography as described elsewhere (13). Intracellular fibronectin was obtained as follows. Cell layers were rinsed once in ice-cold Ca2'/Mg''-free phosphatebuffered saline and 1 ml of 0.25%. trypsin in phosphate-buffered saline was added to each dish. Plates were gently shaken for 5 min at room temperature by which time cells could be suspended by pipetting. Cells thus prepared were rinsed several times in ice-cold phosphatebuffered saline containing soybean trypsin inhibitor (0.5 mg/ml). The cell pellet was subsequently treated in one of two ways. For immunoprecipitation, the cells were lysed in 200 p1 of NET buffer at pH 11.0 (0.15 M NaCI, 0.05 M Tris, 0.5% Nonidet P-40.0.28 bovine serum albumin, and protease inhibitors). After 3 min a t 4 "C, 800 pI of NET buffer, pH 7.4, were added, and the total lysate was centrifuged (10,OOO X g, 10 min) 20 pI of rabbit antifibronectin antiserum (13) were added to the supernatant and, after 1 h a t 4 "C, 100 pl of a 10% suspension of Staphylococcus aureus (IgGsorb, Enzyme Center) were added. After 1 h a t 4 "C, the suspension was pelleted, rinsed twice in NET, pH 7.4, twice in NET, pH 7.4, containing 0.5 M NaCI, and twice in 0.05 M Tris-CI, pH 7.5. Cell surface fibronectin was obtained by the urea extraction method of Yamada et al. (14) as described in Ref. 13. Aliquots of immunoprecipitated or affinity-purified fibronectins were checked for purity by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (15) ( Fig. 1).
Enzymatic a n d Chemical Treatment of Fibronectin a n d Oligosaccharides-Glycopeptides were generated by pronase digestion of affinity-purified or immunoprecipitated fibronectin. Pronase (Calbiochem) was preincubated for 1 h at 50 "C in 0.1 M Tris-CI, pH 8.0, 2 mM CaC12, and then digestion was carried out in this buffer for 18 h a t 50 "C with pronase at 2 mg/ml under toluene (16).
Oligosaccharides were produced by hydrazinolysis of whole fibronectin or glycopeptide fractions. Starting material was dried in vacuum over HSO., for 12 h in hydrolysis tubes before 400p1 of anhydrous hydrazine (Pierce Chemical Co.) were added. Sealed tubes were heated a t 100 "C for 10 h (17,18). After cooling, hydrazine was removed under vacuum and the residue was dried down three times from toluene. Finally, the residue was dissolved in 400 p1 of saturated NaHCOTr and reacetylated with acetic anhydride (19).
Desialylation of oligosaccharides was effected by treatment with 0.01 M HCI for 60 min a t 80 "C under nitrogen (20).
Endo HI (Streptomycesplicatus, Health Research Inc., NY) digestion of glycopeptides was performed by incubation a t 37 "C for 24 h with 5 milliunits of enzyme in 50 p1 of 0.05 M Na citrate, pH 5.0.
Fractions of 0.65 ml were collected.
Affinity chromatography on Sepharose-bound ConA (Pharmacia) was as described by Kornfeld et al. samples were injected in 100 pl of 12.5 mM KHJ' O.,, pH 4.0. Elution at 1 ml/min was continued with this buffer for 15 min followed by an increasing linear gradient to 500 mM KHJ'O,, pH 4.0, a t 45 min. Oligosaccharides which had been desialylated by mild acid treatment were separated on Micro Pak AX-IO with a mobile phase of acetonitrile/H20 (24). Samples were injected in a maximum value of 100 pl of the initial solvent, acetonitrile/H?O (6535). Elution was continued a t 1 ml/min and the H 2 0 content was increased to 46% over 90 min and then to 70% over a further 10 min. For the KHzPO, system, fractions of 0.5 ml were collected; for acetonitrile/HnO, fractions were 1.0 ml.
Standards for gel filtration and HPLC were prepared by partial acid hydrolysis of dextran (25) and chitin (26). Their elution positions were determined by chemical analysis of reducing groups in the collected fractions (27). Dextran oligomers were periodically run on the analytical Bio-Gel P4 column; variation between runs was never greater than *one fraction (*I%).
The radioactive content of column fractions was determined by scintillation counting in Hydrofluor or Liquiscint (Nuclear Diagnostics). For double label experiments, data were corrected for "C entry into the .'H channel as described (28).

Total Glycopeptide and Oligosaccharide Preparations-
Gel filtration of GlcN-labeled control glycopeptides derived by pronase digestion of medium fibronectin yielded a broad peak with leading and trailing shoulders (Fig. 2a). Co-chromatography of ["'ClGlcN-labeled control glycopeptides with ["HIGlcN-labeled glycopeptides from monensin-treated cells revealed that the latter consistently eluted after most of the control glycopeptides (Fig. 2a). This difference in elution position was also found when control and monensin glycopeptides, both labeled with ["HIGlcN, were sequentially run on the same column. Oligosaccharides released from fibronectin by hydrazinolysis also showed a size difference between samples derived from control and treated cells (see below).
The sensitivity of pronase-released glycopeptides to the endoglycosidase, endo H, was also examined (Fig. 26). Only monensin oligosaccharides showed a change in elution profile indicative of endo H sensitivity. Control oligosaccharides were not cleaved by endo H.
Analysis by HPLC showed further differences between control and monensin-derived oligosaccharides. Separation of intact oligosaccharides in phosphate buffer demonstrated their relative degree of sialylation ( Fig. 3) (23). Integration of the data indicated that in controls, 56% of GlcN label was present in nonsialylated oligosaccharides, 28% in monosialylated, and 15% in disialylated forms. Corresponding values for oligosaccharides from monensin-treated cells were 73% nonsialylated, 20% monosialylated and 6% disialylated. These values varied between preparations, but the relative undersialylation of oligosaccharides from monensin-treated cells was consistently seen. After desialylation by mild acid hydrolysis, all oligosaccharides behaved as nonsialylated species (not shown). HPLC of these desialylated oligosaccharides in the acetonitrile/HzO system resolved the control samples into three major peaks (Fig. 4). Under the same conditions, monensin samples were resolved into numerous components (Fig.  4).
A further difference between control and monensin oligosaccharides was revealed by chromatography on ConA-Sepharose (Fig. 5). Control samples of oligosaccharides derived from medium fibronectin contained some components (ConA-I) that did not bind to ConA and some that could be eluted with 10 m~ a-methyl-D-glucoside (ConA-11). Samples from monensin-treated cells contained an additional component that was eluted with 0.5 M a-methyl-D-mannoside (ConA-111). Furthermore, in these samples, the elution with 10 mM amethyl-D-glucoside consistently produced a peak that slowly trailed off, unlike the sharp peak observed in controls. For both control and monensin samples, the nonbinding components did not bind to the column when reapplied. Recovery  for control and experimental samples averaged 90%, and no additional label could be eluted with 1.0 M NaC1. Pronaseproduced glycopeptides and desialylated or intact oligosaccharides from the same preparation all produced similar elution patterns on ConA-Sepharose. Also, the differences between control and monensin samples were always found in such experiments, confirming the results in Fig. 5.

Abnormal Glycosylation of Human Fibronectin
" . \-..  The ConA and Bio-Gel P-6 elution profiles of glycopeptides from cell surface and medium fibronectins were identical. Also, the monensin-induced differences were found in oligosaccharides from both sources.
Monensin is known to affect the release of lysosomal hydrolases from fibroblasts (29), and so the monensin-induced differences in glycopeptide structure might result from the action of glycosidases released into the culture medium. However, the behavior of glycopeptides on Bio-Gel P-6 and ConA was not altered by the incubation of a mixture of conditioned media from control and monensin-treated cultures. Elution profiles of control samples were identical, and the monensininduced differences were evident whether or not mixed incubation of samples was performed (data not shown).
Enzymatic Analysis of C o d Fractions-ConA chromatography demonstrated that oligosaccharide preparations of medium fibronectin were mixtures of several species. Also, the degree of sialylation of oligosaccharides varied slightly between preparations (see above). Thus, to minimize potential heterogeneity and to facilitate subsequent analysis, oligosaccharides were desialylated by mild acid treatment before fractionation on ConA. The total oligosaccharide preparations showed differences (Fig. 6a) in the size distribution of control and monensin samples when chromatographed on Bio-Gel P-4. Unlike the glycopeptide preparations (Fig. 2a), these differences were free of possible influences of the peptide portion and thus were entirely due to carbohydrate differences.
The ConA pools were desalted on Bio-Gel P-6 and analyzed by chromatography on Bio-Gel P-4 prior to and after various enzymatic treatments. On Bio-Gel P-4, ConA-I, the fraction that did not bind to ConA, presented a complex elution profiie including oligosaccharides up to 15 glucose equivalents in sue. Although differences between control ConA-I and monensin ConA-I were seen, this fraction was not evaluated further (see "Discussion").
Control ConA-I1 oligosaccharides (Fig. 6b) eluted as a major peak centered around the elution position corresponding to 13 glucose equivalents. Monensin ConA-11 oligosaccharides overlapped this peak but also included later eluting components (Fig. 6b). The results of treating GlcN-labeled ConA-I1 oligosaccharides with exoglycosidases are shown in Fig. 6, C -f . Digestion with P-galactosidase (Fig. 6c) resulted in a change of the elution profiie of both control and monensin oligosaccharides and diminished the difference in their size distributions. Exo-P-N-acetylglucosaminidase alone had little effect on control ConA-11, but altered the peak shape of monensin oligosaccharides (Fig. 6d) and caused 16% of the 'H label to be released. This eluted at the position of a GlcNAc standard, which coincided with the elution position of 2 glucose equivalents (Fig. 6d) (30). Treatment of ConA-I1 with P-N-acetylglucosaminidase after P-galactosidase digestion (Fig. 6e) caused label to be released from both monensin and control samples and, in the case of controls, 48% of the total counts eluted at the position of 2 glucose equivalents (Fig. 6e). A smaller proportion (36% of total) was released from monensin ConA-11. Treatment of GlcN-labeled samples with cu-mannos- idase (Fig. S f ) had little effect on control oligosaccharides, but the monensin ConA-I1 fraction showed a markedly altered elution profile and yielded a prominent new peak eluting at -5 glucose equivalents. Some GlcN-labeled ConA-I1 was digested with /I-galactosidase, /3-N-acetylglucosaminidase, and a-mannosidase to potent i d y produce the Man /I+GlcNAc a(fFuc a)-+GlcNAc core. Material pooled from the expected elution position of this structure was then rechromatographed with or without prior cy-fucosidase digestion (Fig. 7 ) . Without digestion, control material eluted with a size difference equivalent to one saccharide larger than monensin oligosaccharides (Fig. 7a). This difference was abolished by a-fucosidase treatment (Fig. 7b).
ConA fractions of desialylated oligosaccharides were also obtained from preparations labeled with mannose. The ConA-I1 fraction (Fig. 8a) showed a different size distribution for monensin and control oligosaccharides. Differences between the elution profiles of Man- (Fig. 8a) and GlcN- (Fig. 66) labeled oligosaccharides were attributed to the relative abundance of labeled saccharides in particular oligosaccharides.
Using the Man-labeled ConA-I1 fraction, sensitivity to amannosidase (Fig. 86) was illustrated by the release of label, which eluted at the one glucose position, as did a mannose standard. Release of label from control ConA-I1 was negligible (Fig. 8b).
Digestions were also performed on ConA-111 fractions, which were unique to monensin-treated cells (Fig. 5). Digestion of Man-labeled ConA-111 (Fig. 8c) with a-mannosidase ( Fig. 8 4 caused 80% of the label to elute at the one glucose position. GlcN-labeled ConA-I11 (Fig. 9a) ran as a narrow peak centered around the elution position of 12 glucose equivalents. Digestion with /I-N-acetylglucosaminidase caused the appearance of a peak which eluted in the position of GlcNAc but contained only 12% of the radioactivity (Fig. 9b). An identical elution profile was generated by treatment with /I-N-acetylglucosaminidase after ,&galactosidase digestion (not shown). a-Mannosidase drastically altered the elution profile of ConA-I11 and left a major GlcN-labeled peak at 5 glucose equivalents (Fig. 9c).  The endo H sensitivity detected in monensin glycopeptides (Fig. 2b) was further investigated in ConA-fractionated oligosaccharides. GlcN-labeled monensin ConA-I1 showed some endo H sensitivity (Fig. loa), as was evidenced by the appearance of label at the 2-glucose position, corresponding to cleavage of a single GlcNAc residue from the reducing end of the oligosaccharide (31). A distinct peak also appeared at about the 9-glucose position. A greater effect was seen in the endo H digest of GlcN-labeled ConA-111 (Fig. lob). Label appeared at the 2-glucose position, and the remaining oligosaccharide peak was almost entirely shifted from its previous position (about 12 glucose equivalents, Fig. 9a) to between 9 and 10 glucose equivalents. Endo H sensitivity was also detected in the Man-labeled ConA fractions I and I1 (not shown).   Figs. 66 and 9a, respectively. ---, I4C, Control; -, 'H, monensin.

DISCUSSION
The carbohydrate content of fibronectins from several tissues and species has been described. These glycoproteins contain from about 5 (20, 32) to 9.5% (33) carbohydrate in an estimated three (34) to five (35) N-linked oligosaccharides/ fibronectin monomer. The most abundant oligosaccharide, described for several fibronectins (17, 20, 30, 35), has the structure (+.SA+Gal&+GlcNAcP+Manai+) eManp+ GlcNAcfLGlcNAc. Variations have been noted in the peripheral Gal/I+GlcNAc linkages (30) and, more commonly, in the abundance and linkage of terminal SA residues (30, 37). The presence of fucose linked to the proximal GlcNAc of the chitobiose unit is variable and may (33) or may not (21, 34, 36) be present on human fibronectins. Thus, the major fibronectin oligosaccharide is a typical complex biantennary structure similar to that found on transferrin (34). The occurrence on fibronectin of oligosaccharides of greater complexity has also been suggested (34, 35, 37). The intracellular elaboration of fibronectin oligosaccharides appears to follow a known pathway via high mannose intermediates (38).
A major component of control oligosaccharides derived from medium fibronectin is the low affinity ConA-I1 fraction that is rapidly eluted by 10 mM a-methylglucoside. This behavior (39) and the results of the enzymatic digestions are entirely compatible with the complex biantennary structure cited above. As did Fisher and Laine (34) and Wagner et al. (37), we find an oligosaccharide subfraction that does not bind to ConA. This property (39) and the elution position on Bio-Gel P-4 (data not shown) suggest that this fraction, present in both control and monensin oligosaccharides, may consist of more highly branched complex oligosaccharides. However, due to the lack of more definitive structural information, we have not evaluated this fraction further.
That monensin can have effects on the glycosylation of fibronectin was initially demonstrated by comparison of glycopeptide elution profiles on Bio-Gel P-6 ( Fig. Za). The mean size of glycopeptides derived from medium fibronectin was diminished in the presence of monensin, and a substantial proportion of them were endo H-sensitive, in contrast to the endo H-resistant control glycopeptides (Fig. 26). This enzyme sensitivity showed that the size difference required the presence of novel oligosaccharides and did not simply reflect a differential abundance of normal oligosaccharides. This conclusion was reinforced by ConA chromatography and by HPLC of desialylated oligosaccharides; in monensin samples, unique peaks were present (Figs. 4 and 5).
Assumptions concerning the probable structure of some of these novel forms were made, based on the structure of normal fibronectin oligosaccharides (as summarized above) as well as the sequence of events involved in the synthesis of typical complex oligosaccharides (10). Analysis was facilitated by the initial fractionation of desialylated oligosaccharides on ConA. The ConA-I1 fraction, which contained the typical fibronectin oligosaccharides, also contained some abnormal forms in the monensin sample. Thus, the ability of P-N-acetylglucosaminidase to remove significant amounts of GlcN label without prior treatment by ,&galactosidase (Fig. 6d) is indicative of the incomplete galactosylation of some monensin oligosaccharides. Cleavages from control ConA-I1 fractions by this enzyme were minimal. However, after initial P-galactosidase exposure, the 48% GlcN label released from controls by P-Nacetylglucosaminidase is close to the theoretical value of 50% (Fig. 6e). Fucosylation of ConA-I1 oligosaccharides was investigated by digesting them to a presumptive Manp-GlcNAcP-(fFuccr-*)GlcNAc core. The resulting mixture eluted from Bio-Gel P-4 with a size difference of one saccharide between control and monensin samples (Fig. 7a). This difference was abolished by a-fucosidase treatment (Fig. 76), suggesting that in control fibronectin, the majority of oligosaccharides bear an a-fucose on the core structure, whereas in the presence of monensin, fucose is absent or severely reduced. Impaired fucosylation, galactosylation, and the general reduction of sialylation (Fig. 3) seen in the presence of monensin are all indicative of abnormalities in the "later" events involved in complex oligosaccharide formation. The conclusion that complex oligosaccharides with incompletely formed branches are present on fibronectin secreted into the medium of Human Fibronectin 553 in the presence of monensin is supported by the greater abundance of smaller oligosaccharides in the monensin ConA-I1 fraction (Fig. 66). Also, we note that Baenziger and Fiete (39) have shown that the sequential removal of peripheral sugars from complex biantennary oligosaccharides results in a progressive increase in their affinity for ConA. This could explain the trailing of the monensin ConA-I1 peak, in contrast to the sharply eluted control ConA-I1 (Fig. 5). If they are present at all, structures as small as (Man-)+ MankGlcNAcj?-+GlcNAc cannot be abundant, as only a small proportion of monensin oligosaccharides elute from Bio-Gel P-4 at the -7-glucose position expected for such structures (Fig. 66). Digestion with a-mannosidase (Figs. 6f and 86) shows, nevertheless, that the monensin ConA-I1 fraction does contain structures with terminal a-linked mannose residues. However, at least some of these are probably larger high mannose structures because this same fraction does contain a component that is sensitive to endo H and which yields a structure eluting at the 9-glucose position after cleavage with that enzyme (Fig. loa) (40).
A far greater proportion of such high mannose forms are present in the ConA-I11 fraction, which is unique to monensin oligosaccharides. Digestion of this fraction with a-mannosidase reduces most Man label to the single saccharide (Fig. 8d) or the GlcN label to a structure eluting at the 5-glucose position expected for Man,LI+GlcNAc,LI+GlcNAc (Fig. 9c).
Consistent with this is the endo H sensitivity of almost the entire Cod-I11 fraction (Fig. 106). The elution position of ConA-111 on Bio-Gel P-4 suggests that structures as large as (Man);.-" (GICNAC)~ may be present. The occurrence of a small amount of terminal (P-N-acetylglucosaminidase-releasable) GlcNAc in the GlcN-labeled ConA-HI (Fig. 96) is probably due to spillover of some ConA-I1 components. This possibility was avoided in the Man-labeled preparation by extending the elution with 10 mM a-methylglucoside. Further support for the high mannose nature of ConA-I11 is its high affinity for ConA, which is characteristic of such oligosaccharides (41,42). Also, ConA chromatography of oligosaccharides from intracellular fibronectin yielded high affinity ConA-I11 peaks for both control and monensin samples (not shown). This is consistent with the expected presence of high mannose intermediates on intracellular control fibronectin (38). Why some apparently high mannose oligosaccharides should be present in the monensin ConA-I1 fraction is unclear. ConA chromatography may not provide absolute separation of identical forms, or subtle differences in similar (but not identical) high mannose structures may modulate their affinity for ConA.
We conclude that fibroblast fibronectin secreted in the presence of monensin bears abnormal oligosaccharides which fall into two types. The fist are those in which the final stages of complex oligosaccharide formation are incomplete and so are deficient in their sialylation, galactosylation, and fucosylation. This more rigorously illustrates, at the structural level, the monensin effects inferred by Tartakoff and Vassalli (9) from studies of immunoglobulin M. We have also explained our previous observation that monensin apparently increased the incorporation of mannose into secreted fibronectin (11) by the demonstration that high mannose oligosaccharides, usud y restricted to intracellular fibronectin, are found on fibronectin secreted in the presence of monensin. Other investigators have described the intracellular accumulation of high mannose glycoproteins under such conditions (43), but no information was given about their secretion. The release of molecules bearing high mannose oligosaccharides resembles the situation seen with the alkaloid, swainsonine (44). This compound has, however, been shown to specifically inhibit an a-mannosidase involved in oligosaccharide processing (45).
The pleiotropic effects of monensin can be viewed differently; uiz. the monensin-induced disruption of Golgi elements described in many cell types (1, 3, 5) may be a manifestation of incorrect fission-fusion of membrane systems, of which reduced secretion is a prominent effect.
At another level, one can invoke current paradigms of glycoprotein biosynthesis to explain our observations. These are the sequential nature of the multienzyme systems involved in oligosaccharide processing (46) and the concept of functionally and topologically distinct Golgi subcompartments (47,48). Thus, membrane-bound packages of secretory molecules emerging from the endoplasmic reticulum may be denied access to the complete range of oligosaccharide-processing enzymes in the monensin-treated cell. However, whether the incompletely processed fibronectin molecules completely bypass particular compartments or whether they pass through functionally incompetent compartments remains to be determined.