Biosynthesis of the Human Sucrase-Isomaltase Complex DIFFERENTIAL 0-GLYCOSYLATION OF THE SUCRASE SUBUNIT CORRELATES WITH ITS POSITION WITHIN THE ENZYME COMPLEX*

The biosynthesis and maturation of human sucrase- isomaltase (SI, EC 3.2.1.48-lo), was studied in cul- tured small intestinal biopsy specimens and mucosa explants. Pulse-chase experiments with [35S]methio-nine revealed one high mannose intermediate of M, = 210,000 (pro-SIh) which was processed at a slow rate to an endo H-resistant, mature form of M, = 245,000 (pro-SI,). The fully core-glycosylated form (M, = 212,000) was detected only when 1-deoxynojirimycin was added to the culture medium, thus indicating that the core sugars undergo rapid processing by rough endoplasmic reticulum membrane-bound glycosidases. The data presented showed that trypsin specifically and instantaneously (within 1 min) cleaves pro-SI, to two subunits I, (Mr = 145,000) and S, (Mr = 130,000). Elastase and chymotrypsin are not effective. Enzymic and of SI with endo-&N-acetylglucosaminidase F/glycopeptidase F probing for the binding capacity of SI to Helixpomatia lectin demonstrated that pro-SI,, I,, and S. are N- low reproducible results carrier protein added reaction immunoprecipitated precooled


Biosynthesis of the Human Sucrase-Isomaltase Complex
DIFFERENTIAL 0-GLYCOSYLATION OF THE SUCRASE SUBUNIT CORRELATES WITH ITS POSITION WITHIN THE ENZYME COMPLEX* (Received for publication, December 7, 1987) Hassan Y. Naim, Erwin E. Sterchi, and Michael

J. Lentze
From the Department of Gastroenterology, Children's Hospitat of the University of Bern, Freiburgstrasse 15,Switzerlnnd The biosynthesis and maturation of human sucraseisomaltase (SI, EC 3.2.1.48-lo), was studied in cultured small intestinal biopsy specimens and mucosa explants. Pulse-chase experiments with [35S]methionine revealed one high mannose intermediate of M, = 210,000 (pro-SIh) which was processed at a slow rate to an endo H-resistant, mature form of M, = 245,000 (pro-SI,). The fully core-glycosylated form (M, = 212,000) was detected only when 1-deoxynojirimycin was added to the culture medium, thus indicating that the core sugars undergo rapid processing by rough endoplasmic reticulum membrane-bound glycosidases. The data presented showed that trypsin specifically and instantaneously (within 1 min) cleaves pro-SI, to two subunits I, (Mr = 145,000) and S , (Mr = 130,000).
Elastase and chymotrypsin are not effective.
Enzymic and chemical deglycosylations of SI with endo-&N-acetylglucosaminidase F/glycopeptidase F and trifluoromethanesulfonic acid (TFMS) as well as probing for the binding capacity of SI to Helixpomatia lectin demonstrated that pro-SI,, I,, and S. are Nand 0-glycosylated. Furthermore, the results were indicative of a posttranslational 0-glycosylation of pro-SI, since ( i ) the earliest detectable precursor form, pro-SIh, did not bind to H. pomatia lectin and (ii) its deglycosylation products with both endo-8-N-acetylglucosaminidase H and TFMS were identical.
Both the S, and I, subunits contain eight N-linked glycan units, at least one of which is of the high mannose type and found on s,. Finally, s,, but not I,, was shown to display at least four populations varying in their content of 0-linked glycans. The heterogeneous 0-glycosylation pattern of S, could be correlated with the distal position of this subunit (and its O-glycosylation sites) within the pro-SI molecule, thus affecting the extent of 0-linked oligosaccharide processing and their subsequent presentation on the mature molecule.
Much of our knowledge of the biogenesis of plasma membrane proteins has evolved from studies in cultured epithelial cells, such as the Madin-Darby canine kidney cell infected with enveloped viruses (for a review see Ref. 1). These studies revealed the existence of highly organized pathways for viral glycoproteins to enter the host cell, follow distinct sequential routes through the cellular compartments, and segregate in a * This work was supported by Grants 3.838-0.86 and 3.810-0.86 from the Swiss National Science Foundation for Scientific Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. polarized fashion to the basolateral and apical domains. To which extent the viral model applies to endogenous membrane proteins remains, however, to be determined. This is mainly because cytopathic effects that accompany infections might influence the polarized surface expression of viral glycopro-

teins.
The epithelial cells of the human small intestine (enterocytes) provide a good experimental model for studies aimed at elucidating the molecular events involved in the biogenesis and trafficing of membrane proteins. These cells are characterized by a striking polarization of their plasma membrane in two distinct structural and functional domains: the brush border or microvillar membrane and the basolateral membrane. The microvillar membrane is endowed with a number of glycoproteins which are involved in the terminal steps of digestion of micromolecular nutrients, and most of them have been characterized (2)(3)(4). Sucrase-isomaltase (SI, , an enzyme complex responsible for the final steps in starch and glycogen digestion, is a particularly attractive microvillar membrane glycoprotein to study, as it displays several interesting structural features (5)(6)(7)(8)(9)(10)(11)(12)(13). SI is a heterodimeric molecule comprising two subunits of unequal size, sucrase and isomaltase. It is anchored to the brush border (microvillar) membrane by a hydrophobic NH2-terminal segment of the isomaltase subunit. The sucrase subunit lacks a membrane anchor and is associated with the isomaltase subunit on the luminal side of the brush border membrane by noncovalent, strong ionic interactions. Recent structural data for rabbit SI deduced from sequenced cDNA revealed striking homologies between both subunits, so that the consensus is now emerging that the enzyme complex has evolved by partial gene duplication (14). Furthermore, these results provided additional details about the membrane anchoring of SI. Thus it became apparent that SI spans the membrane once, having the amino terminus exposed on the cytoplasmic side, and, conclusively, SI is synthesized without a cleavable signal sequence in a fashion similar to the invariant chain of class I1 histocompatibility antigens (15)(16)(17)(18), the transferrin receptor (19), the asialoglycoprotein receptor (20)(21)(22), and the neuraminidase of influenza virus (23). Several studies have shown that sucrase-isomaltase is synthesized as a single chain precursor protein (denoted pro-SI) (7, 24), which is cleaved, once it reaches the microvillar membrane, by pancreatic se-cretions to its two enzymatically active subunits. The protease responsible for this event in the rat small intestine has been shown to be elastase (7). However, this finding was not extended to or confirmed in other species.
Despite extensive work on the structure and topology of SI in many species, several aspects of the intracellular processing have not been explored in great detail. In particular, thorough analyses of the events occurring at the rough endoplasmic reticulum (RER) and the Golgi apparatus are as yet not clearly assessed. For instance, it is not clear whether SI, as is generally believed for other membrane glycoproteins in many cellular systems, undergoes early modification of the carbohydrate residues in the RER (25) and whether this type of modification is crucial for the transport kinetics of the molecule from the RER to t h e Golgi. Furthermore, although data were presented which were suggestive of an 0-glycosylation of SI (26,27), unequivocal determination of the type (0linked/ZV-linked) and the size of the carbohydrate chains on the mature precursor molecule and on both subunits, as well as the correlation of the glycosylation events to the positioning of the subunits within the large precursor molecule, are still lacking. In addition, provided that it takes place, the question as to whether the attachment of 0-linked oligosaccharides to the polypeptide backbone is a cotranslational or a posttranslational event has not yet been answered. Finally, the immense body of data on the structure of SI and its implication for the biosynthesis of the molecule has not been extended t o or investigated in great detail in the human species in normal small intestinal epithelial cells (28). Results from such investigations should provide a better knowledge of the overall importance of the expression of sucrase-isomaltase. This is particularly crucial in studies aimed at elucidating the molecular mechanisms underlying the synthesis and processing of sucrase-isomaltase in a number of developmental and pathological situations such as t h e transient expression of sucrase-isomaltase in the human fetal colon (29), its expression in some human colon cancers (30), or its impaired expression in congenital sucrase-isomaltase deficiency (28,31). These most interesting cases provide good experimental tools for dissecting the sequence of events during synthesis and intracellular transport of proteins in general at the molecular level.
In this paper, we investigated the biosynthesis of SI in organ culture of human small intestinal epithelial cells, identified precursor and mature forms of SI, analyzed in detail the posttranslational modifications at the RER and in the Golgi, and assessed the type and size of the carbohydrate moieties of both subunits and of the precursor molecules. We show that SI is synthesized as a single polypeptide high mannose precursor that has undergone rapid trimming in the RER, slowly transported to the Golgi where complex glycosylation takes place, and finally is cleaved by trypsin after translocation across the brush border membrane. We present data which demonstrate that both subunits are Nand 0glycosylated and provide evidence for the existence of several populations of differently glycosylated brush border forms of the sucrase subunit.

Biological Materials
Human small intestinal biopsy specimens (approximately 5-10 mg wet weight) were obtained for routine diagnostic purposes by suction with a pediatric Watson capsule. They appeared normal when examined by light microscopy and expressed normal levels of brush border disaccharidase activities (sucrase, and lactase,determined according to Asp et al. (32)). Another source for the tissue organ culture material was the small intestine of kidney donors taken directly after the respiratory support system has been switched off. The tissue was transferred to the laboratory within minutes where it was rapidly rinsed with 0.9% cold saline. Intestinal explants were prepared by dissection of the mucosa from the proximal jejunum. They were approximately 2 X 5-mm in size and had normal morphology and disaccharidase activities. The usage of tissue from both sources was in accordance with the rules of the ethical committee of the hospitals of the university.

Immunochemical Reagents
Monoclonal antibodies against the human small intestinal brush border membrane hydrolases were produced according to established hybridoma techniques and described in detail elsewhere (33). The mouse anti-sucrase-isomaltase (anti-SI) monoclonal antibody was a product of hybridoma HBB 2/219/88 and used in ascites form prepared from hybridoma-bearing Pristane-primed Balb/c mice. For immunoprecipitation, anti-SI monoclonal antibody was partially purified from the ascites fluid by two successive precipitations with 45% ammonium sulfate. The precipitates were dissolved in 0.1 M sodium bicarbonate (pH 8.3) containing 0.5 M sodium chloride, extensively dialyzed against the same buffer, and conjugated to CNBr-activated Sepharose 4B according to the manufacturer's instructions. About 4-6 mg of the immunoglobulin fraction was usually coupled to 1 ml of Sepharose.

Biosynthetic Labeling of Biopsy Specimens (Continuous Puke)--
Biopsy specimens were washed three times with RPMI 1640 medium supplemented with streptomycin (100 pg/ml), penicillin (100 units/ ml), and 10% dialyzed fetal calf serum (designated complete medium) and placed on stainless steel grids in tissue culture dishes essentially as described by Browning and Trier (34). They were then incubated in methionine-deficient RPMI 1640 medium containing 10% fetal calf serum and antibiotics as above (denoted Met-free medium) at 1 ml/biopsy specimen for 2 h a t 37 'C in a CO, + O2 (595, v/v) incubator prior to pulse labeling with 150 pCi of [35S]methionine. Continuous labeling was performed for 15 min, 4 h, and 18 h. When used, tunicamycin (5 pg/ml) and dNM (5 mM) were present during the preincubations of Met-free medium and during pulse labeling. After the labelingperiods, the biopsy specimens were chilled to 4 "C, washed three times with ice-cold phosphate-buffered saline, and homogenized with a Teflon-glass homogenizer in 1 ml of 25 mM Tris-HC1 (pH 8.1), 50 mM NaCl, and a mixture of protease inhibitors containing 1 mM PMSF, 1 pg of pepstatin, 5 pg of leupeptin, 17.4 pg of benzamidine, and 1 pg of aprotonin (homogenization buffer). The homogenates were either further directly processed for immunoprecipitation or kept frozen at -20 "C until used. For studies aimed at examining the effect of pancreatic proteases on cleavage of the polypeptide precursor of sucrase-isomaltase, the protease inhibitor mixture was omitted from the homogenization buffer.
Treatment of Biopsy Homogenutes with Pancreatic Proteases-Biopsy specimens were continuously labeled with [35S]methionine for 4 h, homogenized in 4 ml of the homogenization buffer, divided into 4 aliquots, and the effect of three pancreatic proteases was investigated by modifying the procedure of Hauri et al. (7). The aliquots were treated with 0.5 mg/ml bovine trypsin, 0.5 mg/ml bovine achymotrypsin, or 60 units/ml porcine elastase. The treated aliquots and an untreated control were incubated for 15 min at 37 "C. The enzyme reaction was stopped by adding soybean trypsin inhibitor at 1 mg/ml in the case of trypsin or PMSF at 10 mM in the case of achymotrypsin or elastase. The reaction mixture was rapidly chilled to 4 "C and stored frozen (-20 "C) or directly processed for immunoprecipitation. The data obtained from this experiment have shown that trypsin, but not elastase or a-chymotrypsin, was effective in generating sucrase and isomaltase from the precursor molecule prosucrase-isomaltase (see "Results," below). The kinetics of this cleavage were examined as follows. Biopsy specimens labeled with methionine for 4 h were homogenized and divided into several aliquots each containing 1 ml of homogenate equivalent to one biopsy specimen. The aliquots were prewarmed to 37 "C, and trypsin was added to a final concentration of 0.5 mg/ml. Incubations were carried out for 1, 2, and 5 min at 37 "C. Thereafter, the aliquots were further processed as described above. The control sample (zero time) constituted an aliquot to which trypsin and soybean trypsin inhibitor were rapidly and successively added. The reaction mixture in this case was immediately cooled to 4 "C by brief cooling in a slurry of Dry Ice in ethanol.
Biosynthetic Labeling of Mucosa Explants (Pulse-Chose)-Mucosa explants were prepared for biosynthetic labeling as described above €or biopsy specimens. Following a pulse period of 10 min with 200 pCi of [36S]methionine, the explants were rapidly washed with icecold complete medium and incubated for 2 min at room temperature in the same medium containing 0.1 mg/ml cycloheximide. Samples were chased for various times with complete medium supplemented with 2.5 mM methionine. At each time point an explant was washed in ice-cold phosphate-buffered saline, suspended in the homogenization buffer as described for biopsy specimens, and prepared for immunoprecipitation. When used, monensin (1 PM) was present during the preincubations of Met-free medium, pulse labeling, and chase.
Preparation of Brush Border Membrane Vesicles (BBM) from Small Zntestinul Mucosa-All procedures were carried out at 4 "C unless otherwise stated. Mucosa was scraped off with the back of a scalpel blade from a 30-cm piece of frozen and thawed human small intestine, and a 2% homogenate was made in 2 m M Tris-HC1 (pH 7.1) containing 50 mM mannitol and 40 pg/ml PMSF. Schmitz et ai. (35) and as modified by Sterchi and Woodley (36). In A brush border membrane fraction was prepared according to brief, solid CaCL was added to the homogenized mucosa to a final concentration of 10 mM. The homogenate was left on ice for 30 min with gentle stirring and then centrifuged at 2,000 X g for 20 min to yield a pellet (PI) and a supernatant (SI). SI was centrifuged at 25,000 X g for 30 min. The pellet (Pz) was resuspended and homogenized in 2 mM Tris-HC1 (pH 7.1) and 50 mM mannitol and layered onto a discontinuous (37,40,42,45, and 60%) glycerol density gradient. The gradient was centrifuged at 63,000 X g for 15 min, and five fractions were collected (FI-Fv). The brush border membranes were recovered in fraction I1 (Fir). They had a specific activity for sucrase of 2 units/ mg protein, representing a 25-fold purification as compared with the homogenate.
Preparation of BBM and Intracellular Membranes (IM) from BWsynthetically Z.abeled Biopsy Specimens-The procedure for subcellular fractionation of mucosa homogenates (35) was modified and scaled down to biosynthetically labeled biopsy specimens. All procedures were done at 4 "C. In short, 1 ml of homogenate was centrifuged at 1,000 X g for 30 min to remove debris and nuclei. The supernatant was incubated with 10 mM CaClz on ice for 30 min with occasional mixing. Following centrifugation at 5,000 X g for 15 min, the pellet (PI) was further washed and homogenized with the homogenization buffer containing 10 mM CaCl,, whereas the supernatant (SI) was kept on ice. Washed P, was similarly centrifuged to yield Sz and PP.
This procedure was repeated four times in total. The pellet following the fourth wash (Pa) was homogenized in the homogenization buffer without CaCl, and centrifuged at 15,000 X g for 15 min. The IM fraction was recovered from the pellet, the BBM fraction from the supernatant and the preceding washings (S1-S6) after centrifugation at 20,000 X g for 45 min.
Zmmumprecipitation-Homogenates or purified BBM and IM from 35S-labeled biopsy specimens and mucosa explants were solubilized with Triton X-100 and sodium deoxycholate, 0.5% final concentrations, by stirring on ice for 30 min. BBM from nonlabeled tissue (FII fraction) was similarly solubilized at 2 mg of protein/ml buffer. The solubilized material was centrifuged at 100,000 X g for 1 h at 4 "C and immunoprecipitated with anti-SI-antibodies conjugated to Sepharose as describedpreviously for lactase-phlorizin hydrolase (37).

Endo-@-N-acetylglucosaminidase H (Endo H)
and Endo-@-N-acetylglucosaminidase F/Glycopeptidase F (Endo FIGF) Treatments-Digestion of %-labeled immunoprecipitates with endo H and endo F (containing glycopeptidase F) were performed as described previously (37). Endo H and endo F/GF concentrations used were 3 mIU and 1 unit, respectively. Digested proteins were recovered by precipitation with trichloroacetic acid, 15% (w/v) final concentration, and the pellets were washed twice with ice-cold acetone and kept at -20 'C.
Dose-dependent deglycosylation of SI with endo F/GF was carried out as follows. Mucosa explants were biosynthetically labeled for 4 h with [%]methionine, homogenized, and centrifuged at 1000 X g for 15 min. The membranes contained in the supernatant were permeabilized for 3 min with 0.1% Triton X-100, so that the pro-SI molecules in the inside-out oriented vesicles could be made accessible to trypsin. Subsequently, trypsin was added at 0.5 mg/ml, and the reaction was allowed to proceed for 5 min at 37 "C. The homogenates were further solubilized with Triton X-100 and deoxycholate, 0.5% final concentrations and the detergent extracts were divided into equal parts each corresponding to one mucosa explant. SI was immunoprecipitated and the precipitates were subjected to endo F/GF treatment with varying concentrations of the endoglycosidase (2-100 milliunits).

Elution of Proteins from Fixed and Dried Polyacrylamide Gels and
H. pomatia Lectin Chromatography-Proteins from purified brush border membranes (FII) were identified by Coomassie Blue staining of polyacrylamide slab gels, excised, and eluted essentially as described by Luscher et al. (40) with slight modifications. Briefly, the protein bands (approximately 7-10 pg each) were swollen in water, homogenized in 1 ml of 100 mM Tris-HCl (pH 7.5), 1% SDS, and 0.1% 2-mercaptoethanol in Eppendorf tubes, and incubated overnight at 56 "C.
The eluted proteins were precipitated with trichloroacetic acid to a final concentration of 15% (w/v). The pellets were washed three times in ice-cold acetone at -20 "C. The proteins were either dissolved in phosphate-buffered saline containing 1% Triton X-100 (buffer A, 1 ml/protein band) or additionally treated with endo F/GF followed by precipitation with trichloroacetic acid and washing of the protein pellets with ice-cold acetone. The digestion products were finally solubilized in 1 ml/protein band in buffer A. Both types of preparation were subjected to chromatography on 0.5 ml of H. pomatia lectin. The bound material was eluted with buffer A containing 10 mM N-acetylacid, washed twice with acetone, and subjected to SDS-PAGE after D-galactOSamine, concentrated by precipitation with trichloroacetic solubilization in sample buffer.
Lectin-Sepharose Affinity Chromatographies-Biopsy specimens labeled for 4 h with [35S]methionine, or nonlabeled brush border membranes (Fir, 2 mg/ml) were homogenized and solubilized in buffer A supplemented with the protease inhibitor mixture (see above) and centrifuged at 100,000 X g for 45 min at 4 "C. The supernatant (1 ml corresponding to one biopsy specimen or to 2 mg of F d was run through a 1-ml lectin-Sepharose column equilibrated in buffer A. The adsorbed material was eluted in buffer A containing 50 mM methyla-D-mannopyranoside for the lentil lectin columns and 10 mM Nacetyl-D-galactosamine for the H. pomatia lectin columns. Aliquots with the highest radioactivity (%3-labeled material) or highest protein content (F,,) were pooled, dialyzed against the homogenization buffer containing 0.5% deoxycholate and 0.5% Triton X-100, and subjected to immunoprecipitation with anti-SI-antibodies.
Deglycosylatwn of Sucrase-Zsomaltase by TFMS-%-Labeled molecules were subjected to TFMS according to Edge et al. (41). Due to the low amount of protein in the sample, reproducible results were obtained only when a glycosylated carrier protein was added to the reaction mixture. Sucrase-isomaltase was immunoprecipitated with anti-SI-Sepharose beads, and the beads were eluted in 4% SDS in water by boiling for 5 min. To the eluates, 10 pg of ovalbumin was added, and the mixture was precipitated in 3 volumes of ice-cold acetone for at least 6 h a t -20 "C. After centrifugation, the protein pellet was dried at 37 "C and cooled on ice. To the pellet, 30 pl of a precooled mixture of TFMS and anisole (2:1, v/v) were added. The vial was capped after bubbling N2 through the solution and left for 2.5 h at 0 "C. The reaction was terminated by the addition of 125 pl of pyridine/water (4:1, v/v) in 10-pl portions. During this treatment the vial was immersed in a slurry of acetone/Dry Ice. The solution was precipitated with 3 volumes of ice-cold acetone and washed twice with acetone. A similar treatment was carried out on "S-labeled molecules which were digested with endo F/GF. Proteins were further analyzed by SDS-PAGE.

Identification of the Precursor and Mature Forms of Human Intestinal SI
To identify and to assess the size of the precursor and This polypeptide was susceptible to endo H digestion, since treatment caused a substantial decrease in its apparent molecular weight to 185,000 (Fig. LA, lane b). A similar molecular species was revealed upon endo F/GF treatment (Fig. lA, lane  c). Addition of tunicamycin, which inhibits the co-translational N-glycosylation of proteins (44), to the culture medium before and during labeling of biopsy specimens resulted in the identification of the 185,000 protein by anti-SI antibodies (Fig. lB, lane b). Taken together, these results indicate that the M, = 210,000 protein has N-linked oligosaccharides exclusively of the high mannose type and represents therefore the cotranslationally glycosylated high mannose precursor of SI. To identify the molecular species of SI in the BBM and IM, biopsies metabolically labeled with [3sSS]methionine for 4 h were homogenized, and cellular fractionation was performed by the Ca2+ precipitation method. Both fractions were solubilized and immunoprecipitated with anti-SI antibodies. Fig.   2 shows that the M, = 245,000 protein was the only component of SI detected in BBM ( l a n e b). The IM fraction contained both the high mannose precursor, i.e. the M, = 210,000 as well as the M, = 245,000 proteins (Fig. 2, lane a). The M, = 245,000 protein is therefore the mature, complex-glycosylated, brush border form of SI revealed in biopsy specimen biosynthetically labeled in organ culture. In brush border membrane preparations from nonlabeled mucosa, SI is revealed as three polypeptides of M, = 245,000, 145,000, and 130,000 (Fig. 2, lane c). In analogy with other systems (4,(7)(8)(9), these polypeptides correspond to the brush border forms of pro-SI and the subunits isomaltase and sucrase, respectively. That the latter two polypeptides were not detected in biosynthetically labeled biopsies is consistent with the concept of the extracellular cleavage by intraluminal proteases of pro-SI after maturation and insertion into the membrane. However, a considerable proportion of the enzyme persists in the noncleaved precursor form. Densitometric scannings of Coomassie Blue-stained gels of electrophoretically analyzed SI showed that the M , = 245,000 amounts to approximately 20% of the total protein immunoprecipitated from nonlabeled tissue (mean of five preparations, lane c in Fig. 2 is representative for these purifications). isomaltase and the single-chain SI precursors respectively, without special preference), SI (total sucrase-isomaltase).

Cleavage of Pro-SI by Trypsin
While the role of the pancreatic secretions in the cleavage of pro-SI to both its active subunits sucrase (S) and isomaltase (I) is well documented (7,9), the protease responsible for this process in man is not known. Hauri et al. (7) have shown, in the rat small intestine, that pro-SI is cleaved by elastase. In the course of our studies on the posttranslational processing of SI and assessment of pre-and post-Golgi transit times, we attempted to make use of cleavage after maturation of pro-SI for the determination of the transport kinetics from the Golgi to the cell surface by correlating these to the detection of cleaved pro-SI molecules? We therefore investigated the effect of trypsin, elastase, and chymotrypsin on pro-SI.
Homogenates from biopsy specimens labeled for 4 h were treated with elastase, trypsin, and chymotrypsin, solubilized, and immunoprecipitated with anti-SI antibodies. Fig. 3A shows that only trypsin treatment generated two polypeptides which migrated on the gel as authentic isomaltase (I,) and sucrase (S,) (lane b). Elastase and chymotrypsin did not result in significant cleavage of the pro-SI, molecule (lanes c and d).
In addition, the time course of appearance of the two polypeptides after various incubations of biopsy homogenates with trypsin was determined. Fig. 3B shows that already after 1 min at 37 "C, isomaltase and sucrase became visible (lane b).
* H. Naim, manuscript in preparation. However, more pro-SI, molecules were cleaved after 2 min of incubation with trypsin as assessed by comparing the labeling intensity of the products by densitometry (not shown). Essentially, similar results were obtained after 5 min of incubation. These data demonstrate that the cleavage of pro-SI, by trypsin is specific, unique to the generation of the two subunits, and instantaneous, since it occurs within 1 min.

Nand 0-Glycosylutwn of pro-SI and the Subunits S and I
To determine the size, nature (high mannose/complex type), and mode of linkage (N-1inkedlO-linked) of the glycan units present on pro-SI and the two subunits the following approaches were undertaken : 1)  Biopsy specimens were biosynthetically labeled for 4 h with [3sS] methionine, homogenized, and subsequently brush border and intracellular membranes were prepared by the Ca"-precipitation procedure (35). SI was immunoprecipitated from total homogenates (H) (

lanes a, b ) , intracellular membranes (ZM) (lanes c, d ) and brush border membranes ( B B M ) (lanes e, f ) .
Portions of each immunoprecipitate were treated with endo F /GF (lanes b, d, f ) . Endo F/GFtreated and nontreated samples were subjected to SDS-gel electrophoresis and fluorography. Assuming that pro-SI, carries exclusively N-linked oligosaccharides, then deglycosylation of pro-SIh as well as pro-SI, with endo F/GF should generate polypeptides of similar apparent molecular weights (MI = 185,000). Since this was not the case, we conclude that pro-SI, contains in addition to Nlinked carbohydrates endo F/GF-resistant glycan units which are most likely 0-linked to serine or threonine residues.
To test this hypothesis, we treated SI purified from labeled biopsy samples (15 min or 4 h) with TFMS. This reagent, introduced by Edge et al. (41), is efficient in cleaving 0glycosyl bonds. It is therefore suitable for investigating whether further deglycosylation of endo F/GF-resistant forms of glycoproteins can be achieved. SI was treated with TFMS prior to or after endo F/GF digestion. In both cases, similar results were obtained. Fig. 5 shows that digestion of the pro-SI), with endo H (lane 6) and TFMS (lane c) gave digestion products of similar Mr. The polypeptide of MI = 185,000 was also revealed when SI isolated from 4-h labeled biopsy samples (containing pro-SI, and pro-SIh) (lane d ) was treated with TFMS (lane e ) . These results indicate that pro-SIh is not 0glycosylated and bears exclusively N-linked carbohydrate residues, since similar digestion products were obtained with endo H, endo F/GF (see also Fig. l), and TFMS, whereas the mature species, pro-SI,, is Nas well as 0-glycosylated. The difference in M, between the endo F/GF-and TFMS-treated pro-SI, (M, = 205,000 uersus 185,000) can therefore be attributed to the presence of approximately 20 kDa of 0-linked sugars on serine or threonine residues of the mature molecule.
To assess the size and nature of glycan units of the sucrase The partial sensitivity of pro-SI, to endo H is therefore accounted for by the presence of at least one high mannose chain on the S, molecule. Digestion with 1 unit of endo F/GF converted the I, and S, subunits to polypeptides of M, = 124,000, 109,000, 104,000, 99,000, and 94,000 (Fig. 6A, lane d ) . To determine which of these correspond to the endo F/GF forms of I, and S, , we performed endo F/GF digestions on the individual subunits after excision from Coomassie Blue-stained gels. Thus, endo F/GF treatment of I, generated the MI = 124,000 species, and S, was converted to polypeptides ranging in M, between 109,000 and 94,000 (the results of the electrophoretic analysis were essentially similar to those shown in Fig. 9B (see below)).
By contrast, one deglycosylated form of S, was revealed when SI was treated with TFMS. In fact, the M, = 94,000 species was the only polypeptide3 among the components of endo Ftreated S, detected, indicating that the other three endo F/ GF forms were converted to this form (Fig. 6B, lanes c and  d ) . The I, subunit was shifted by TFMS to M, = 111,000 (Fig.  6B, lanes c and d )   Samples were analyzed by SDS-gel electrophoresis and fluorography.
The slight reduction in the apparent molecular weight of S, upon endo H treatment is denoted by the open triangle in lane c. E, %Shomogenates from biosynthetically labeled biopsy specimens (4-h labeling) were treated with trypsin. SI was immunoprecipitated from the detergent extracts and treated with endo H ( l a n e b ) , TFMS ( l a n e c), and TFMS after endo F/GF digestion ( l a n e d). Treated and untreated ( l a n e a ) samples were analyzed by SDS-gel electrophoresis and fluorography.
F/GF was not required to achieve complete deglycosylation with TFMS, as shown by the similar digestion patterns of endo F/GF-treated or nontreated SI (Fig. 6B, lanes c and d).
In summary, these data show that the I, as well as the S, In an attempt to assess the number of carbohydrate side chains N-linked to sucrase and isomaltase, immunoprecipitated molecules from permeabilized and trypsin-treated ' 73-  Table I.
Binding of SI to Lectim-To confirm the deglycosylation data and to substantiate these by an independent approach, we examined the binding of pro-SIh and pro-SI, to lentil lectin and H. pomatia lectin.
It is now well established that glycan units linked to the polypeptide backbone through serine or threonine residues, contain N-acetylgalactosamine, galactose, and sialic acid (45). ' We preferred to use in these experiments 35SS-labeled homogenates rather than nonlabeled tissue, as dose-dependent deglycosylation of nonlabeled SI has imposed several technical problems related to the inadequate amount of the molecule remaining after digestion with endo F and precipitation with trichloroacetic acid.  Crucial for the binding of glycoproteins containing these sugar types is removal of sialic acid to expose galactosyl/N-acetylgalactosaminyl residues. However, sucrase-isomaltase isolated from the adult small intestine has been shown not to contain sialic acids (26) and neuraminidase-treated and nontreated SI bind equally to peanut agglutinin and H. pomatia lectin columns? Lentil lectin is known to bind mannose-rich domains in carbohydrate side chains (47).
Homogenates from biopsy specimens labeled for 4 h were solubilized and subjected to lentil lectin-Sepharose or H.
pomatia lectin-Sepharose. The glycoproteins bound to lentil lectin were eluted with methyl-a-D-mannopyranoside and those bound to H . pomatia lectin with N-acetyl-D-galactosamine. The eluates were further immunoprecipitated with anti-SI-antibodies. Fig. 8 shows that pro-SIh was precipitated from the lentil lectin eluates ( l a n e a ) but not from those of H. ( l a n e b). Pro-SI, was found in both eluates (lanes a and b). These data confirm that pro-SIh is rich in ' H. Naim, unpublished experiments.
mannose residues and show that pro-SIh does not contain cotranslationally added N-acetyl-D-galactosamine. In contrast, since it bound to H. pomutia lectin, pro-SI, is 0-glycosylated and this type of glycosylation occurs later in the biosynthesis, most likely in an early region of the Golgi apparatus.
To obtain more information on the existence of O-glycosylation sites on both subunits of SI, we investigated the binding of SI to H. pomatia lectin. For this purpose brush border membranes prepared from nonradioactive tissue (FII) were solubilized and run on H. pomatia lectin-Sepharose. The bound glycoproteins were eluted with N-acetyl-D-galactosamine, immunoprecipitated with anti-SI-antibodies, and the precipitates were analyzed by SDS gel electrophoresis. Fig.   9A, lane b, shows that SI bound to H. pomatia lectin since it was detected in the eluates, thus providing further evidence for the occurrence of 0-linked carbohydrates on the SI molecule. To show that the 0-linked glycans are alone responsible for the binding of SI to H. pomatia lectin and to rule out any possible nonspecific binding or weak interaction which may originate from galactose residues of the N-linked glycan units, SI was purified from brush border membranes, digested with endo F/GF, and run on H. pomatia lectin-Sepharose. The molecules retained by the column revealed molecular species identical to those of endo F-treated SI used as control, indicating that endo F/GF-treated SI specifically reacted with H. pomatia lectin (Fig. 9A, lanes c and d).
The strong association between sucrase and isomaltase is well documented (7-9). This is also confirmed by the ability of monoclonal antibodies directed against different epitopes on either the sucrase or isomaltase molecules to immunoprecipitate similar molecular species (33). It is therefore reasonable to assume that, for binding of SI to H. pomatia lectin to take place, the presence of 0-linked sugars on one of the two molecules is sufficient. To corroborate, however, that both S, and I, contain sites reactive with H. pomatia lectin, thc individual subunits of SI were passed through an H. pomati lectin column after or without endo F/GF treatment. Electro phoretic analysis of the bound material demonstrated by silve staining of the gel that I,, S,, as well as their correspondin endo F/GF forms bound to the lectin, indicating that bot subunits bear 0-linked oligosaccharides (Fig. 9B). Finall: TFMS treatment of I, and S, abolished completely the bindin capacity of these molecules to H. pomatia lectin, demonstrr ting that 0-glycosidically linked glycans are responsible fc the binding of SI to the lectin (data not shown).

Posttranslational Processing of SI
To identify more clearly the core-glycosylated intermed ates of SI, biopsy specimens were labeled for 10 min wi1 [35S]methionine and simultaneously treated with 5 mM dNP a specific inhibitor of endoplasmic reticulum-bound glucoz dase I (48). SI was immunopurified and a portion of tl immunoprecipitate was treated with endo H. Under the conditions, the core-glycosylated precursor of SI was detect as a species a t M , = 212,000 (Fig. 10, lane a); thus slight higher than the polypeptide observed after short pulse labeli in the absence of the inhibitor (Fig. 10, lane b) brush border membranes were purified from mucosa homogenates and solubilized with 1% Triton X-100. SI was immunoprecipitated from the detergent extracts ( l a n e a ) or from eluates of H. pomutia lectin-Sepharose chromatography of identical amount of detergent extracts ( l a n e b). SI immunoprecipitated from detergent extracts was treated with endo F/GF ( l a n e e), or treated with endo F/GF, run through H. pomatia lectin-Sepharose, and eluted with N-acetyl-Dgalactosamine ( l a n e d ) . Samples were subjected to SDS-gel electrophoresis and the gel was stained with silver nitrate. E, brush border membranes (FIT) were prepared from nonlabeled mucosa homogenates. SI was immunoprecipitated from the detergent extracts and analyzed by SDS-gel electrophoresis and Coomassie Blue staining.
The bands corresponding to L and S, were identified, and the middle of each band was excised, eluted, and part of the eluate treated with endo F/GF. Treated and untreated samples were passed through an H. pomutia lectin column and the eluates analyzed by SDS-PAGE and silver staining. In another experiment, the proteins were subjected to SDS-PAGE after endo F/GF treatment. Electrophoretic patterns essentially similar to those shown in this figure were obtained.
of SI, pulse-chase experiments with [35S]methionine combined with endo H treatments of immunoprecipitated SI were performed. The endo H-sensitive high mannose form of SI (pro-SIh) already appeared as a strong band after 10 min of pulse (Fig. 11, lane u). It could still be detected clearly after 240 min into the chase (Fig. 11, lane 0 ) . The fact that the M, = 212,000 form was not detected during the early stages of chase is suggestive of a rapid trimming of glucose residues from the core sugars on the precursor molecule. After 60 min of chase, a faint band corresponding to pro-SI, was revealed ( l a n e i). The intensity of this band became stronger upon prolonged chase periods indicating that more high mannose precursors were processed. However, complete conversion to The treated and nontreated samples were further analyzed by SDSgel electrophoresis and fluorography.
this mature form was not achieved after 4 h ( l a n e 0 ) or even after 18 h of chase (not shown). The rate of transport from the rough endoplasmic reticulum to the Golgi was probed by the conversion of the processed molecule to endo H resistance. Thus, about 105-110 min were required for conversion of half of the high mannose precursor to the complex glycosylated, endo H-insensitive form (Fig.  11). To demonstrate that this processing takes place in the Golgi vesicles, pulse-chase experiments were undertaken in the presence of the carboxylic ionophore monensin, whose site of action is the Golgi apparatus (49). Fig. 12 depicts the results of this experiment. As expected, in the presence of monensin, pro-SIh was not converted to pro-SI, after 120 min of chase ( l a n e h), whereas in the absence of monensin it did ( l a n e g). Similar results were obtained after 180 min of chase. Later into the chase, a broad band was detected of M, = 210,000-230,000 indicating that some processing of pro-SIh has occurred but complete conversion to pro-SI, was inhibited by monensin. These data show that monensin greatly inter- feres with the processing of pro-SIh to pro-SI, and indicates that the acquisition of complex carbohydrates by SI occurs exclusively in the Golgi membranes.

DISCUSSION
The data presented in this report provide further information on the identity of the sucrase-isomaltase complex and extend these to the characterization of the molecular forms of the enzyme in the epithelial cells of the human small intestine. Furthermore, our results demonstrate that: (i) the mature forms of sucrase (S,) and isomaltase (IE) contain 0and N-linked oligosaccharides; (ii) 0-glycosylation of the precursor molecule (pro-SI) is a posttranslational event; and (iii) the S, subunit displays a heterogeneous pattern of 0glycosylation which correlates with its position within the precursor molecule (pro-SI).
Consistent with previous results obtained on the biosynthesis of sucrase-isomaltase in organ culture of pig small intestinal explants (24) and in a colorectal carcinoma cell line, Caco-2 (33), sucrase-isomaltase is synthesized as a single polypeptide chain. Despite its unusually large size, precursor forms can be detected within 10 min of pulse labeling ( M , = 210,000), similar to the molecule in Caco-2 cells but unlike that in the pig small intestine. Several lines of evidence indicate that the M , = 210,000 form represents the high mannose precursor (pro-SI,) which has undergone cotranslational N-glycosylation at the site of synthesis in the RER. Hence, it may be converted into the M , = 185,000 component by endo H treatment. Moreover, it was still precipitable after inhibition of N-glycosylation with tunicamycin, as an M , = 185,000 polypeptide. Finally, when the processing of newly synthesized glycoproteins was blocked by the use of dNM, which inhibits trimming of glucose residues in the endoplasmic reticulum (46), a shift in the apparent M, of the precursor to 212,000 was obtained.
Subsequent processing of Pro-SIh comprises modification of oligosaccharides to the complex, endo H-resistant type which is characterized by the production of pro-SI, ( M , = 245,000). This latter step occurs in the Golgi complex, as monensin, whose site of action is the Golgi membranes (49), largely inhibited the appearance of pro-SL. Consistent with current concepts of biosynthesis and of plasma membrane proteins (25,50,51), pro-SIh therefore undergoes membrane-associated transport to the Golgi apparatus where trimming of the high mannose chains by mannosidases I and I1 and addition of complex sugars takes place.
Using two independent procedures we demonstrated that pro-SI, contains 0-linked as well as N-linked glycans. Hence, deglycosylation of pro-SI, with endo F/GF and TFMS revealed the existence of approximately 20 kDa of 0-linked oligosaccharides. Furthermore, both pro-SI, as well as its endo F/GF product bound the N-acetyl-D-galactosamine-specific lectin of H. pomatia. The addition of 0-linked oligosaccharides to pro-SI and the larger size of the complex glycan units uersus high mannose chains appear to explain the unusual increase in apparent molecular weight of the mature molecule in comparison to the high mannose precursor form (from M , = 210,000 (pro-SIh) to M , = 245,000 (pro-SI,)).
The following observations have corroborated that O-glycosylation of SI is not a co-translational event and occurs later in the biosynthesis, presumably in the Golgi apparatus.
Hence, the earliest detectable form of SI, pro-SIh, failed to bind to H. pomatia lectin but bound strongly to lentil lectin confirming that the core-sugars of this early form are mannose-rich and indicating that they do not contain D-galactose-@( 1-3)-N-acetyl-D-galactosamine structures. Furthermore, the lack of detectable 0-linked ~-galactose-/3(1-3)-N-acetyl-D-galactosamine residues on pro-SIh is manifested by the identical species generated upon treatment of this polypeptide with endo H and endo F/GF, which cleave N-linked structures, and TFMS, which removes in addition 0-glycosidic linkages (MI = 185,000). Finally, tunicamycin, which inhibits the addition of Glc3Man9GlcNAc2 to acceptor asparagine residues, but not of 0-linked oligosaccharides to serine/threonine, generated the M , = 185,000 form which did not react with H. pomatiu lectin or lentil lectin (not shown).
The conclusion that 0-linked oligosaccharides are added postranslationally to the SI molecule is consistent with results which emerged from several recent studies. Hanover et al. Being associated with the brush border membrane, the pro-SI, molecule is cleaved by proteases of pancreatic origin to the subunits sucrase (S,) and isomaltase (L). In the organ culture, these proteases are not present, and consequently the molecule persists to be found in its noncleaved precursor form. The current data have demonstrated that trypsin, but not elastase or chymotrypsin, specifically and rapidly (within 1 min) generates sucrase and isomaltase from pro-SI,. Hauri et al. (7) have shown that the rat enzyme is cleaved by elastase, whereas trypsin and chymotrypsin were not effective. Apparent structural differences between the molecule in the two species, at least in the region where cleavage takes place, may account for their different susceptibilities to the two proteases. Alternatively, variations in the glycosylation patterns of the rat and human SI might influence the final conformation and spatial orientation of the molecule in the two species rendering one of them cleavable by one protease but not by the other. The transport of pro-SIh from the RER to the Golgi was found to occur at a slow rate as probed by the conversion of the transferred and further processed molecule to endo H resistance. This is also demonstrated by the unusual existence of a significant proportion of pro-SIh even after 4 and 18 h (not shown) of chase. The transport rate from the RER to the Golgi does not seem to be influenced by early modifications of the co-translationally added sugar chains. That dNM treatment of cultured biopsy samples produced precursor molecule of slightly higher apparent molecular weight (MI = 212,000) than pro-SI,, whereas in the absence of the reagent and during the very early stages in the pulse-chase experiments similar molecules were not detected, is indicative of rapid processing of the core oligosaccharides. Similar data were obtained when the biosynthesis of lactase-phlorizin hydrolase, another disaccharidase of the intestinal brush border membrane, was investigated (37). Thus, the pulse-chase experiments with lactase-phlorizin hydrolase revealed slow conversion to the final endo H-resistant forms which occurred within 2 h of chase. In experiments not shown: the transport kinetics of two brush border hydrolases dipeptidylpeptidase IV and aminopeptidase N were found to proceed 3-and 4-fold faster than those of SI (and also lactase-phlorizin hydrolase) from the RER to the Golgi. In spite of slight differences in the absolute transit times obtained, our data are consistent with those of Hauri et al. (33) in regard to the general concept of the occurrence of two classes of brush border membrane molecules which migrate at different rates within the small intestinal epithelial cells. A comparison with the conversion rates of a large number of membrane and secretory glycoproteins in a variety of cell systems favors classification of SI (and also lactase-phlorizin hydrolase) into a category of slow moving proteins. In accordance with the receptor-mediated transport mechanism (55, 56), SI may not bind avidly to the putative transport receptor and likely moves in a bulk-phase fashion in the lumen of transport vesicles. This might explain why a considerable proportion of the SI precursor remained in an endo H-sensitive form as a consequence of transport arrest and accumulation at the RER. It cannot be the large size of the molecule which caused the partial processing of the high mannose precursor; otherwise aminopeptidase N, also a large brush border membrane molecule, would have been processed at a rate similar to SI, and that was not the case6 (33).
The movement of the sucrase-isomaltase molecule within the Golgi vesicles and the accompanying covalent modifications appear to take place rapidly since intermediates of the processing of N-linked and 0-linked carbohydrates are difficult to detect by pulse-chase analysis. This is suggestive of a rapid elongation of 0-linked side chains concomitant with or immediately followed by the processing of the N-linked glycans, resulting in the identification of the fully complex glycosylated pro-SI,.
The results reported here have shown that both subunits of the enzyme complex, sucrase and isomaltase, contain Nlinked as well as O-linked oligosaccharides. From the deglycosylation data with endo F/GF and TFMS, we propose that the mature sucrase (S,) subunit exists as a number of populations (at least four) which bear similar numbers of N-linked glycans (about eight) but differ in their content of 0-linked oligosaccharides. In contrast, the mature isomaltase species (IJ does not show a heterogeneous pattern of glycosylation. The variability in the 0-glycosylation of the sucrase subunit and the uniformity of that of isomaltase can be explained in two ways. Either by virtue of its location within the pro-SI molecule, all the 0-glycosylation sites of isomaltase are clustered or distributed in such a way that they are more available e. H. Naim, unpublished observations. than those of the sucrase molecule for the addition of 0-linked glycans, or the heterogeneous cell population in the biopsy specimen gives rise to different levels of N-acetylgalactosaminykransferase which will add N-acetyl-D-galactosamine mainly to the most accessible 0-glycosylation sites, and these are likely found on the isomaltase subunit. The first possibility offers, however, a more plausible explanation of our results, since it can be better interpreted in terms of the structural model of SI evolved from a number of studies which dealt with the association of the molecule with the membrane and the implication of this association on its biosynthesis. It has become apparent from these studies that sucrase-isomaltase is anchored to the microvillar membrane by a hydrophobic NHz-terminal segment of the isomaltase subunit (12,13). The synthesis of the precursor molecule starts with the isomaltase domain and ends with the sucrase subunit, implying that the carboxyl terminus of the pro-SI molecule is that of sucrase. More recently, Hunziker et al. (14) were able to predict the primary structure of rabbit pro-SI from the sequence of a nearly full length cDNA. Interestingly, a serine/threonine stretch in the isomaltase region may be located in close proximity to the membrane. From the sequence of pro-SI and the glycosylation data presented here, it can be hypothesized that the position of the 0-glycosylation sites with respect to the carboxyl terminus is of paramount importance in the processing of 0-linked glycan units and their subsequent presentation in the mature glycoprotein. Thus, some potential 0-glycosylation sites may only circumstantially be made available during the concomitant and rapid processing of N-and 0-linked carbohydrates. The 0-glycosylation sites which do not always receive 0-linked oligosaccharides are likely located in distal regions in the molecule or clustered at the COOH terminus. In accordance with the structural model proposed by Hunziker et al. (14), and extrapolating to human intestinal pro-SI, processing of 0-linked sugars in the isomaltase region may occur first and involve all potential 0-glycosylation sites, whereas the 0-glycosylation sites of the sucrase subunit are located farther towards the COOH terminus and are not always accessible to the addition of 0-linked sugars. This would explain why the mature sucrase subunit, in contrast to isomaltase, displays a heterogeneous pattern of O-glycosylation. The positioning of sucrase within the enzyme complex might also explain why at least one chain of the N-linked high mannose type remains unprocessed to the complex form. The validity of these interpretations will ultimately need detailed analyses of the carbohydrate structures and assessment of the location of the glycosylation sites on sucrase and isomaltase. We are currently trying to do this.
Consistent with the results shown in this paper, the sequence of events which the initial translation product undergoes during the biosynthesis can be dissected into: (i) rapid cotranslational glycosylation and trimming at the site of synthesis in the RER; (ii) slow transport to the Golgi where processing of N-linked and addition of 0-linked glycans takes place; (iii) translocation after maturation to the brush border membrane and rapid extracellular cleavage by trypsin to two subunits. Since it is heavily 0and N-glycosylated, sucraseisomaltase provides a good model for studying the temporal relationship between 0-glycosylation and other well defined processing events involved in the biosynthesis of membrane proteins. The most interesting aspect in this regard is the differential 0-glycosylation of two regions within a large precursor molecule. clonal anti-human sucrase-isomaltase antibody (HBB 2/219/88) and for critical reading of the manuscript.