Human Mannose 6-Phosphate-uncovering Enzyme Is Synthesized as a Proenzyme That Is Activated by the Endoprotease Furin*

N-Acetylglucosamine-1-phosphodiester α-N-acetylglucosaminidase, also known as “uncovering” enzyme (UCE), is localized in the trans-Golgi network, where it removes a covering N-acetylglucosamine from the mannose 6-phosphate recognition marker on lysosomal acid hydrolases. Here we show that UCE is synthesized as an inactive proenzyme that is activated by the endoprotease furin, which cleaves an RARLPR↓D sequence to release a 24-amino acid propiece. As furin is localized in the trans-Golgi network, newly synthesized UCE is inactive until it reaches this terminal Golgi compartment. LoVo cells (derived from a human colon adenocarcinoma) lack furin activity and have extremely low UCE activity. Addition of furin to LoVo cell extracts restores UCE activity to normal levels, demonstrating that the UCE proenzyme is stable in this cell type. LoVo cells secrete acid hydrolases with phosphomannose diesters as a consequence of the deficient UCE activity. This demonstrates for the first time that UCE is the only enzyme in these cells capable of efficiently uncovering phosphomannose diesters. UCE also hydrolyzes UDP-GlcNAc, a sugar donor for Golgi N-acetylglucosaminyltransferases. The fact that UCE is not activated until it reaches the trans-Golgi network may ensure that the pool of UDP-GlcNAc in the Golgi stack is not depleted, thereby maintaining proper oligosaccharide assembly.

UCE is a type I transmembrane glycoprotein that is mainly localized to the TGN and cycles between the TGN and the plasma membrane (2). Human UCE is composed of 515 amino acids containing a 24-amino acid signal sequence, a luminal domain of 423 residues, a 27-residue transmembrane region, and a 41-residue cytoplasmic tail (3). The mature enzyme has been shown to be a tetramer composed of two disulfide-linked homodimers (4). UCE purified from bovine liver is missing the signal peptide as well as a 24-residue propeptide (3). Examination of the amino acids upstream of the cleavage site revealed an RARLPR sequence, which is known to serve as a furin cleavage site (5). Furin, a calcium-dependent subtilisinlike serine endoprotease, is a type I integral membrane glycoprotein that, like UCE, is predominantly localized to the TGN (6). This raised the possibility that the propiece of UCE is cleaved by furin in the TGN.
In this study, we provide a variety of evidence that, in fact, furin is the endoprotease that removes the propeptide from UCE. Furthermore, we demonstrate that pro-UCE has little or no enzyme activity, establishing that excision of the propeptide is essential for the generation of an active enzyme. Finally, we present a hypothesis for why it may be advantageous to delay activation of UCE until it reaches the TGN.

Cell Lines
Mouse L cells and L cells expressing wild-type (WT 1-38) or mutant (Y488A) human UCE were obtained from Rosalind Kornfeld (Washington University) and grown as described (2). LoVo human colon adenocarcinoma cells (CCL-229) were purchased from American Type Culture Collection and grown in Kaighn's modification of Ham's F-12 medium with 10% fetal bovine serum. CHO-K1 cells and SF ϩ cells were obtained from American Type Culture Collection and from Protein Sciences Corporation, respectively.

Enzyme Assays
UCE was assayed as previously described using [ 3 H]GlcNAc-␣-Me-P-Man as substrate (8). One unit of activity is defined as 1 nmol of * 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

Plasmid Construction for Expression of rh-UCE in Mammalian Cells
The molecular cloning and expression of wild-type human UCE has been described previously (3). This plasmid, designated as pTriplEx/ UCE, was used as the starting material for the construction of recombinant soluble epitope-tagged UCE.

Stable Expression of Recombinant Soluble Human UCE in CHO Cells
CHO-K1 cells were plated in 10-cm dishes at ϳ60% confluency in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS), and each dish was transfected with 6 g of pKB6 and 18 l of FuGENE 6 reagent (Roche Molecular Biochemicals) and incubated at 37°C in a 5% CO 2 atmosphere. A mock transfection using phosphatebuffered saline was also performed to serve as the negative control plates. On the following day, the mock-and pKB6-transfected cells were trypsinized and plated in 96-well plates at a cell concentration of 1000 cells/well in 100 l of Glasgow's minimal essential medium lacking L-glutamine, but containing 10% FBS. On the 3rd day, 100 l of Glasgow's minimal essential medium containing 10% FBS and 50 M MSX was added to each well to obtain a final MSX concentration of 25 M. These plates were incubated at 37°C in a 5% CO 2 atmosphere until single colonies were detected, and the mock transfectants were abolished. MSX inhibits the endogenous glutamine synthetase; and thus, only cells that had integrated the pKB6 plasmid (containing the glutamine synthetase minigene) into their genome survived. This MSX selection process yielded 11 clones as detected by visual inspection of the 96-well plates using an inverted light microscope (Nikon). These clones were designated as UCE clones MB1-11. The UCE MB clones were expanded and subsequently amplified in 96-well plates at various MSX concentrations (100 -500 M). The clones with amplified UCE expression were initially identified by a sandwich enzyme-linked immunosorbent assay using HPC4 as the capture antibody, a rabbit anti-rh-UCE polyclonal primary antibody, and horseradish peroxidase-conjugated goat anti-rabbit secondary antibodies. These amplified UCE clones were also assayed for UCE activity.
A rapidly growing MSX-amplified UCE clone with high UCE expression (designated as UCE108) was identified from a 96-well plate containing 300 M MSX via enzyme-linked immunosorbent assay and was derived from the unamplified parent clone MB7. Clone UCE108 was expanded and used for the production of rh-UCE.

Plasmid Construction for Expression of rh-UCE in Insect Cells
Plasmid pKB6 was used to construct a plasmid encoding soluble epitope-tagged human UCE in an insect expression system (performed at Protein Sciences, Inc.). PCR was performed using pKB6 and genespecific primers O-1987/O-1989 to generate the N terminus of the mature human UCE region and the C-terminal fragment of the signal sequence from Autographa californica nuclear polyhedrosis virus (Ac-NPV). A parallel PCR was performed to generate the polyhedrin promoter and the AcNPV signal sequence. A third PCR was performed to fuse the N terminus of mature human UCE with the AcNPV promoter/ signal sequence containing a 5Ј-EcoRV restriction site and a 3Ј-BglII site via overlap extension. The resultant PCR product was digested with EcoRV and BglII and ligated into an intermediate plasmid. This intermediate plasmid was then digested with KpnI and XbaI to linearize the plasmid within the UCE coding region. The remaining UCE fragment was obtained by digesting pKB6 with KpnI and XbaI and ligating this fragment into the linearized intermediate plasmid to complete the transfer plasmid designated as pAcD1062. The DNA sequence was verified by automated fluorescence DNA sequencing, and pAcD1062 was used to generate the baculovirus transfer vector.
To generate the baculovirus vector via homologous recombination, the transfer plasmid pAcD1062 was cotransfected with linearized Ac-NPV baculovirus genomic DNA for 3 days. The cotransfected cells were subsequently harvested by centrifugation, and the supernatants were used to grow isolated plaques on plates containing Sf9 insect cells. Several plaques with the clear (versus cloudy) plaque phenotype were identified from this process and used as the viral stocks.

Expression of Recombinant Soluble Human UCE in Insect Cells
The expression of rh-UCE in insect cells was performed at Protein Sciences, Inc. A passage 1 viral stock of recombinant baculovirus was prepared by adding the appropriate plaque to a 5-ml culture of Sf9 insect cells in medium containing 5% FBS for 5 days at 28°C. The infected cells were subsequently harvested by centrifugation, and the supernatant was used to inoculate larger cultures. One milliliter of the passage 1 viral stock was used to inoculate 50-ml cultures of expres SF ϩ (SF ϩ ) cells in serum-free medium at a cell density of 1.5 ϫ 10 6 cells/ml in 100-ml Spinner flasks for 48 h at 28°C. The infected cultures were harvested by low-speed centrifugation, and the supernatants comprising the passage 2 viral stocks were used to seed 500-ml SF ϩ cultures at 1.5 ϫ 10 6 cells/ml in 3-liter Spinner flasks for 48 h at 28°C. The cultures were again harvested, and 1 ml of the supernatant (passage 2 viral stock) was used to seed 500-ml SF ϩ cultures in 3-liter Spinner flasks at the same cell density in serum-free medium for 72 h at 28°C to generate the passage 3 viral stock. The passage 3 viral stock was titered by plaque assay, and a 500-ml SF ϩ culture was infected with the passage 3 viral stock at a multiplicity of infection of 1 plaque-forming unit/cell and incubated at 28°C. One-hundred milliliters of the culture was harvested by low-speed centrifugation after 48, 72, and 96 h, respectively. The supernatants were frozen at Ϫ80°C, and the 96-h post-infection supernatant was sent to Novazyme Pharmaceuticals, Inc. for affinity purification.

Purification of rh-UCE from Mammalian and Insect Expression Systems
Purification of rh-UCE from CHO Cell-conditioned Medium-Clone UCE108 was expanded from six T-150 flasks up to 102 roller bottles in Glasgow's minimal essential medium containing 10% FBS and 25 M MSX and cultured for 10 days at 37°C in a 5% CO 2 atmosphere. Twenty liters of conditioned medium was then collected, clarified, and concentrated ϳ3-fold via a tangential flow filtration system (Millipore Corp.). The concentrated medium was subsequently divided into four aliquots and frozen at Ϫ20°C until purification.
rh-UCE was affinity-purified in four different preparations using a 30-ml HPC4-UltraLink affinity column. Monoclonal antibody HPC4 is calcium-dependent; therefore, CaCl 2 was added to the concentrated medium to a final concentration of 5 mM, and the HPC4-UltraLink column was equilibrated with 150 mM NaCl, 50 mM Tris (pH 7.2), and 5 mM CaCl 2 . The medium was loaded onto the antibody column and washed with 150 mM NaCl, 50 mM Tris (pH 7.2), and 1 mM CaCl 2 . rh-UCE was eluted with 150 mM NaCl, 50 mM Tris (pH 7.2), and 5 mM EDTA; concentrated via a Centriprep Plus-80 concentrator (Millipore Corp.); aliquoted into ϳ25-l fractions; and stored at Ϫ80°C. Each rh-UCE preparation was assayed for UCE activity and characterized by SDS-PAGE and Western blotting using the HPC4 primary antibody, and the protein concentration was determined via absorbance at 280 nm using the Advanced protein assay kit (Cytoskeleton, Inc.). The average purification efficiency and yield of the four rh-UCE preparations from CHO cells are shown in Table I.
Purification of rh-UCE from Infected Insect Cell Medium-Insect cell-derived rh-UCE was affinity-purified as described for CHO cellderived rh-UCE except that only 100 ml of the 96-h post-infected conditioned medium was used with a 5-ml HPC4-UltraLink column. Insect cell-derived rh-UCE was also assayed and characterized as described above. The average purification efficiency and yield of the rh-UCE preparation from insect cells are also shown in Table I.

Electrophoresis and Immunoblotting
Two-hundred nanograms of each PNGase F-digested rh-UCE sample was subjected to SDS-PAGE using 12% polyacrylamide gels (Invitrogen). rh-UCE was transferred to nitrocellulose using a Bio-Rad semidry transfer system. The membrane was subsequently blocked with 5% (w/v) nonfat dry milk in 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), and 1 mM CaCl 2 (blocking buffer) for 1 h at room temperature with rocking. The blot was then incubated with the mouse HPC4 monoclonal primary antibody (1 g/ml) in blocking buffer for 30 min at room temperature with rocking. The blot was washed three times with 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 1 mM CaCl 2 , and 0.05% Tween 20 for 15 min between each wash. The blot was then incubated with horseradish FIG. 2. Distinction of pro-UCE and mature UCE by dot-blot analysis. Three identical nitrocellulose dot blots were prepared using 200 ng of rh-UCE derived from either insect or CHO cells. A rh-UCE molecule engineered to contain an improved furin cleavage site (DB-1 rh-UCE) was also expressed in CHO cells and spotted onto each membrane. A buffer control (150 mM NaCl, 50 mM Tris (pH 7.2), and 5 mM EDTA) was also included in each dot blot. The dot blots were subsequently treated as described for the immunoblots under "Experimental Procedures," except that the blot in A was incubated with rabbit anti-UCE propeptide serum (NZW-HD-9B, diluted 1:50 in blocking buffer), the blot in B was incubated with rabbit preimmune serum, and the blot in C was incubated with protein A-purified rabbit anti-rh-UCE (whole molecule) polyclonal antibody (NZW-HD-1A, diluted 1:5000 in blocking buffer). The detection antibody was horseradish peroxidase-conjugated donkey anti-rabbit antibody (diluted 1:10,000 in blocking buffer; Amersham Biosciences). Leu, Glu 71 29 peroxidase-conjugated sheep anti-mouse secondary antibody (Amersham Biosciences) diluted 1:10,000 in blocking buffer for 30 min at room temperature with rocking. The blot was washed as before and incubated with 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), and 1 mM CaCl 2 for 15 min. rh-UCE was visualized by incubating the blot with 1 ml of ECL chemiluminescent substrate (Amersham Biosciences) for 1 min, wrapped in plastic wrap, and exposed on BioMax x-ray film (Eastman Kodak Co.).

N-terminal Sequencing of rh-UCE
Three micrograms each of purified rh-UCE from insect and CHO cells were subjected to SDS-PAGE (12% polyacrylamide gel) and transferred to polyvinylidene difluoride membrane (Amersham Biosciences) using the Bio-Rad semidry transfer system. The protein bands were visualized by Ponceau S staining, and the rh-UCE bands were excised using a scalpel. The rh-UCE samples were subjected to N-terminal sequencing via the Edman degradation method and high pressure liquid chromatography analysis (performed at the Molecular Biology Resource Facility, University of Oklahoma Health Sciences Center). The amino acid sequence of CHO cell-derived rh-UCE is shown in Table II.

Man-6-P/IGF-II Receptor Affinity Chromatography
␤-Glucuronidase and ␤-hexosaminidase present in cell secretions were fractionated on a Man-6-P/IGF-II receptor affinity column essentially as previously described (9). Briefly, media from LoVo and L cell cultures were collected and applied to the receptor affinity column (1 ϫ 1 cm). The column was washed with 5 ml of phosphate-buffered saline (the flow-through fraction), then with 5 ml of phosphate-buffered saline containing 5 mM Glc-6-P (the retarded fraction), and finally with 5 ml of phosphate-buffered saline containing 5 mM Man-6-P (the bound/eluted fraction). Aliquots of the three fractions were assayed for ␤-glucuronidase and ␤-hexosaminidase activities. To demonstrate the effect of UCE on the secreted hydrolases, 1-ml aliquots of the media were treated with 10 units of rh-UCE for 4 h at 37°C and then applied to the affinity column and fractionated as described above. Similarly, to differentiate between hydrolases with phosphomonoesters versus multiple phosphodiesters, 1-ml aliquots of the media were treated with 50 units of E. coli alkaline phosphatase at pH 8.0 for 12 h at 37°C. The reactions were adjusted to pH 7.4 and then subjected to fractionation on the receptor column.

RESULTS
Pro-UCE Is Inactive-As part of a project to prepare soluble forms of UCE, the cDNA encoding human UCE was mutated to introduce a stop codon just proximal to the transmembrane domain and then expressed in CHO and SF ϩ insect cells. The expression vector for the insect cells was further modified to include an insect viral signal peptide in place of the UCE signal peptide and a deletion of the propeptide (Fig. 1). When rh-UCE secreted from the two cell types was purified and assayed, it was noted that the specific activity of the CHO cell-produced enzyme was only 45% that of the insect cell-secreted enzyme ( Table I). Determination of the N-terminal sequence of rh-UCE produced by the CHO cells revealed that only 35% of the enzyme was the mature form, whereas 65% remained as the proenzyme (Table II). To confirm the presence of the propeptide in a portion of CHO cell-produced rh-UCE, a rabbit antibody was generated to the propeptide sequence. As shown in Fig. 2, this antiserum reacted with CHO cell-produced rh-UCE, but not with rh-UCE secreted by the insect cells. The enzymes from both sources reacted with rabbit antiserum prepared against full-length rh-UCE as expected.
The fact that only 35% of rh-UCE secreted by the CHO cells was the mature form and that this material had only 45% the specific activity of insect cell-derived rh-UCE, which was completely the mature form, suggested that the pro-form of rh-UCE may have little or no enzyme activity. Evidence that this FIG. 3. Effect of furin on the enzyme activity of CHO cellproduced rh-UCE. Aliquots (20 g) of purified CHO cell-derived (Ⅺ and f) or insect cell-derived (E and q) rh-UCE were incubated either with 20 units of furin (f and q) or without furin (Ⅺ and E) at 30°C in a 75-l assay containing 100 mM Hepes (pH 7.5), 0.5% Triton X-100, and 1 mM 2-mercaptoethanol for the times indicated. Two microliters was removed from each sample and diluted 1:100 to 1:1600 with 50 mM Tris-HCl (pH 6.7) and 0.5% Triton X-100 and assayed for UCE activity using the [ 3 H]GlcNAc-␣-Me-P-Man substrate. The UCE activity was calculated for each sample and normalized to the corresponding samples that lacked furin. Data are plotted as the percent increase in UCE activity as a function of incubation time.

FIG. 4. Pro-UCE is cleaved by furin.
Twenty micrograms each of baculovirus expression vector system and CHO cell-produced rh-UCE were treated with 20 units of furin at 30°C, and aliquots were removed after 0, 6, 12, and 24 h. The baculovirus expression vector system and CHO cell-produced enzymes without furin treatment were used as negative controls. The UCE samples were then deglycosylated by incubation with PNGase F, and 200 ng of each sample was subjected to SDS-PAGE, followed by Western blotting using the mouse HPC4 primary antibody and horseradish-conjugated sheep anti-mouse secondary antibody.

TABLE III Effect of furin on UCE activity in cell extracts
Parental mouse L cells, the WT 1-38 cell line, and LoVo cells were grown to 90% confluence in 24-well plates. The media were removed, and 120 l of 100 mM Hepes (pH 7.5) containing 1% Triton X-100, 1 mM CaCl 2 , 1 mM 2-mercaptoethanol, and the indicated amounts of furin were added. After 4 h at 25°C, the cell lysates were assayed for UCE activity. The values are expressed as units of activity/well. is indeed the case was obtained by treating rh-UCE produced by the CHO cells with furin, which was predicted to cleave the propeptide. As shown in Fig. 3, furin treatment resulted in a marked increase in enzyme activity to the level observed with the enzyme expressed by insect cells. CHO cell-derived rh-UCE incubated in the absence of furin exhibited no change in its activity, and furin had no effect on the activity of insect cellderived rh-UCE. Examination of CHO cell-derived rh-UCE by Western blotting using monoclonal antibody HPC4 following deglycosylation with PNGase F revealed two bands corresponding to the pro-form and mature form of the enzyme (Fig. 4). Treatment of this material with furin caused the conversion of the pro-form to the mature form. This did not occur in the absence of furin. Furthermore, furin had no effect on insect cell-derived rh-UCE, demonstrating that furin cleaves only the propeptide of UCE.

Furin Is Required for UCE Activity in LoVo Cell Extracts-
The experiments with CHO cell-derived rh-UCE clearly show that furin is capable of cleaving the propeptide of UCE. However, these results do not establish that furin is the endoprotease that acts on UCE in cells. This is an issue because furin is a member of a large family of proprotein convertases (6). To test whether furin is involved in the activation of UCE in intact cells, we turned to LoVo cells, derived from a lymph node FIG. 5. Furin activates UCE in LoVo cell extracts. LoVo cells and WT 1-38 mouse L cells were grown to 90% confluence in 24-well plates. After removal of the media, 120 l of 100 mM Hepes (pH 7.5) containing 1% Triton X-100, 1 mM CaCl 2 , 1 mM 2-mercaptoethanol, and 2 units of furin was added, and the lysed cell extracts were incubated at 25°C. At the indicated times, the samples were assayed for UCE activity. Data are plotted as a percent of the zero time value. The starting LoVo cell extract contained 0.7 units of UCE activity/mg of protein, whereas the WT 1-38 cell extract contained 70 units of UCE activity/mg of protein. q, LoVo cells; OE, WT 1-38 cells.

FIG. 6. Furin inhibitor I inhibits rh-UCE processing in CHO cells.
A CHO cell line that stably expresses rh-UCE was incubated with furin inhibitor I at 0, 5, 10, or 20 M for 72 h at 37°C. Secreted rh-UCE was subsequently purified via HPC4-Sepharose. Five micrograms of rh-UCE purified from the 20 M furin inhibitor culture was incubated with 10 units of furin for 24 h at 30°C. All samples were then assayed for UCE activity. Each purified rh-UCE sample (3 g) was digested with PNGase F, and 150 ng of each sample was subjected to SDS-PAGE and Western blotting using the HPC4 primary antibody. BVES, baculovirus expression vector system. metastasis of a human colon adenocarcinoma (10). These cells are known to have a total deficiency of furin activity due to mutations in the furin gene (10). Assay of LoVo cell extracts for UCE activity revealed that these cells had less than 10% the activity of mouse L cells (Table III). When LoVo cell extracts were treated with different concentrations of furin for 4 h, the UCE activity increased up to 17-fold to the level observed in the mouse L cell extract (Table III). Furin treatment had no effect on the UCE activity of the mouse L cell extract or an extract of mouse L cells (the WT 1-38 line) that express high levels of rh-UCE (Table III). This demonstrates that the conversion of the pro-form of UCE to the mature enzyme is quite efficient in the L cells. The activation of UCE in the LoVo cell extract by furin was time-dependent, as shown in Fig. 5. These results demonstrate that the LoVo cells contain UCE in an inactive form (presumably the pro-form) that can be activated by furin. Because these cells have a selective deficit in furin activity, we conclude that UCE is specifically activated by furin.
A Furin Inhibitor Blocks Processing of Pro-UCE-We next tested the effect of the furin inhibitor decanoyl-Arg-Val-Lys-Arg-chloromethyl ketone (11) on the maturation of pro-UCE. For this experiment, the CHO cells expressing rh-UCE were incubated with various concentrations of the inhibitor for 72 h, and then secreted rh-UCE was purified and analyzed by SDS-PAGE. As shown in Fig. 6, the inhibitor caused a concentrationdependent shift in secreted rh-UCE from the mature form to the pro-form. This was associated with a decrease in the specific activity of the enzyme. When UCE synthesized in the presence of 20 M furin inhibitor I was treated with furin, all of the pro-form was converted to the mature enzyme along with the restoration of the enzyme activity (Fig. 6, sixth lane). These data provide additional evidence that UCE maturation is mediated by furin.
UCE Cleaves UDP-GlcNAc as Well as GlcNAc-␣-P-Man-In considering why it may be advantageous to activate pro-UCE only after it reaches the TGN, we noted the reports that partially purified detergent-solubilized preparations of UCE cleave GlcNAc from UDP-GlcNAc (12)(13)(14). Because many Golgi N-acetylglucosaminyltransferases utilize UDP-GlcNAc as their sugar donor, it might be detrimental if UCE were active as it passed through the Golgi stack. To confirm this observation with a membrane-bound form of the enzyme, we assayed intact mouse cells expressing human UCE(Y488A) on their surface. This form of UCE contains a mutation in the 488 YHPL internalization signal present in the cytoplasmic tail; and consequently, it accumulates on the cell surface (2). When monolayer cultures of these cells were incubated with medium containing either UDP-[ 3 H]GlcNAc or [ 3 H]GlcNAc-P-Man and the release of [ 3 H]GlcNAc was determined, it was evident that surface UCE acted equally on both substrates (Fig. 7). The non-transfected mouse L cells did not hydrolyze detectable amounts of UDP-[ 3 H]GlcNAc under these assay conditions. This finding FIG. 8. LoVo cells secrete acid hydrolases containing phosphomannose diesters. The medium from a 90% confluent LoVo cell culture was collected, fractionated on a Man-6-P/IGF-II receptor affinity column, and analyzed as described under "Experimental Procedures." Aliquots of the flow-through, retarded, and Man-6-P-eluted fractions were assayed for ␤-glucuronidase (G) and ␤-hexosaminidase (H) activities. The total activity in each fraction is plotted as a percent of the total activity in the three fractions. A, untreated medium; B, medium treated with E. coli alkaline phosphatase; C, medium treated with rh-UCE; D, medium treated with rh-UCE followed by alkaline phosphatase. o, flow-through fraction; Ⅺ, Glc-6-P-eluted fraction; f, Man-6-P-eluted fraction.
FIG. 9. Mouse L cells secrete acid hydrolases containing phosphomannose monoesters. The medium from a 90% confluent L cell culture was fractionated on a Man-6-P/IGF-II receptor affinity column as described under "Experimental Procedures" and in the legend to Fig. 7. A, untreated medium; B, medium treated with E. coli alkaline phosphatase; C, medium treated with rh-UCE; D, medium treated with rh-UCE followed by alkaline phosphatase. G and H refer to ␤-glucuronidase and ␤-hexosaminidase activities, respectively. o, flow-through fraction; Ⅺ, Glc-6-P-eluted fraction; f, Man-6-P-eluted fraction.
confirms that UCE is capable of hydrolyzing UDP-GlcNAc as well as GlcNAc-␣-P-Man.
LoVo Cells Secrete Acid Hydrolases with Phosphomannose Diesters-Because LoVo cells have very little UCE activity, we would expect the cells to secrete acid hydrolases that contain phosphomannose diesters rather than the monoesters found on acid hydrolases secreted by cells that have sufficient UCE activity. To analyze this, culture medium from LoVo cells was passed over a Man-6-P/IGF-II receptor affinity column, and the flow-through, retarded, and Man-6-P-eluted fractions were assayed for ␤-glucuronidase and ␤-hexosaminidase activities. Most of the enzyme activity was in the flow-through fraction, with only 10 -20% in the Man-6-P-eluted fraction (Fig. 8A). The material retained could represent enzyme with phosphomonoester(s) or multiple phosphodiesters because the latter have been shown to be capable of binding to Man-6-P/IGF-II receptor affinity columns (15). To distinguish between these possibilities, an aliquot of the medium was treated with E. coli alkaline phosphatase, which can cleave only the phosphate from phosphomonoesters, and reapplied to the receptor column. This treatment decreased the binding of the two acid hydrolases by ϳ50% (Fig. 8B). Therefore, only 5-10% of the acid hydrolases in the medium contained phosphomonoesters. When the medium was treated with rh-UCE prior to passage over the receptor affinity column, 65% of the ␤-glucuronidase and almost 90% of the ␤-hexosaminidase bound to the column and required Man-6-P for elution (Fig. 8C). This binding was abolished when the medium was subsequently treated with alkaline phosphatase (Fig. 8D). These data show that the acid hydrolases secreted by the LoVo cells contained mostly phosphomannose diesters. In contrast to these results, the majority of the ␤-glucuronidase and ␤-hexosaminidase present in the medium of mouse L cells bound to the receptor affinity column, and this binding was mostly abolished by alkaline phosphatase treatment (Fig. 9, A  and B). Furthermore, incubation with rh-UCE did not enhance binding to the affinity column (Fig. 9C), indicating that cellular UCE had efficiently converted the diesters to monoesters in the L cells. Consistent with this result, treatment with rh-UCE and alkaline phosphatase abolished receptor binding (Fig. 9D). DISCUSSION The findings presented in this study establish that UCE is synthesized as a zymogen that is activated by furin. The strongest evidence that furin serves as the endoprotease that cleaves the propiece of UCE in vivo comes from the studies with LoVo cells. These cells have been documented to be deficient in furin activity due to a frameshift mutation in one allele and a point mutation within the homo B domain in the other allele of the furin gene (10). Furin is a member of the subtilisin-like prototoxin prohormone convertase family, which includes seven members to date (6). It is known to catalyze the maturation of a diverse group of proprotein substrates by cleaving most efficiently and specifically at the C-terminal side of an Arg(P Ϫ4 )-X-Lys/Arg-Arg(P Ϫ1 ) sequence. The arginine residues at positions Ϫ1 and Ϫ4 are essential, and basic residues at positions Ϫ2 and Ϫ6 facilitate efficient cleavage (16 -18). The RARLPR2D cleavage site on pro-UCE shows the RXXR2 arrangement typical for furin cleavage, and the additional arginine at position Ϫ6 probably increases the efficiency of furin-mediated cleavage. Proline at position Ϫ2 has no negative effect on furin cleavage (19). A similar RLPR2E sequence is present on ␤-secretase or BACE (beta-site APP-cleaving enzyme), which cleaves the amyloid precursor protein to release the amyloid peptide, which is the main constituent of the amyloid plaques in the brains of Alzheimer's disease patients (20). This propeptide has been convincingly shown to be processed by furin.
There is also strong evidence that furin processes Pseudomonas exotoxin A that has a proline at position Ϫ2 (21,22).
It is of note that the LoVo cell extracts exhibited a low level of UCE activity and that 5-10% of the secreted acid hydrolases contained phosphomonoesters. There are at least three possible explanations for this. First, the pro-form of UCE may have some activity. At this time, we do not have a homogeneous preparation of the pro-form of UCE to test for possible activity. Second, another proconvertase present in the LoVo cells may have some activity toward pro-UCE. A number of studies have reported that LoVo cells efficiently process a few of the propeptides that furin is known to act upon, indicating that LoVo cells express other proconvertases (23)(24)(25)(26)(27). However, all these substrates have the sequence Arg-X-Lys/Arg-Arg or Lys-X-X-Arg2 at their cleavage site. In contrast, LoVo cells act very inefficiently on Pseudomonas exotoxin and Shiga toxin that have the cleavage sequence Arg-X-X-Arg2, which is present in UCE (28,29). Despite this, we cannot exclude the possibility that one of the proconvertases expressed in the LoVo cells is activating a small proportion of UCE. Finally, it is possible that another enzyme has a slight ability to cleave GlcNAc from GlcNAc-P-Man diesters. Even if this is the case, it is clear that UCE accounts for the vast majority of the uncovering activity.
Although the propeptides of many proteins are necessary for correct folding to form an active protein, this appears not to be the case for UCE. As shown in Table I, a construct expressing a form of UCE that lacks the propiece folds into a fully active enzyme, at least in insect cells grown at 25°C. This indicates that the primary function of the propiece is to inhibit UCE activity until it arrives at the TGN, where it encounters furin. Why is it important to avoid having an active form of UCE in the Golgi stack? One possibility is to prevent the hydrolysis of UDP-GlcNAc in the lumen of the Golgi cisternae. UDP-GlcNAc is the nucleotide sugar donor for the many N-acetylglucosaminyltransferases that are localized in the Golgi stack. Our studies with the cell-surface form of UCE (Fig. 7) show that this enzyme acts on UDP-GlcNAc about as well as it cleaves GlcNAc-P-Man. Similar findings have been reported for the solubilized form of the enzyme (12)(13)(14). Therefore, if UCE were active as it moved through the Golgi on its way to the TGN (or if it resided in the Golgi), it might interfere with the assembly of complex-type N-linked glycans and O-linked glycans by depriving the N-acetylglucosaminyltransferases of their donor substrate, UDP-GlcNAc. By activating UCE only after it arrives in the TGN, where it is localized, this potential problem is avoided.