(Received for publication, June 24, 1996, and in revised form, December 2, 1996)
From the Department of Biochemistry and Biophysics, University of
North Carolina, Chapel Hill, North Carolina 27599 and the
Department of Cell and Tumor Biology, Division of
Surgery, City of Hope National Medical Center,
Duarte, California 91010
A single point mutation in the lysosomal proenzyme receptor-inhibiting sequence near the N terminus of mouse procathepsin L can result in glycosylation of a normally cryptic site near its C terminus. When alanine replaced His36, Arg38, or Tyr40, the nascent chain of the mutant protein cotranslationally acquired a high mannose oligosaccharide chain at Asn268. In contrast, when alanine replaced Ser34, Arg37, or Leu39, this second carbohydrate chain was not added. This alternating pattern of abnormal glycosylation suggested that propeptide residues 36-40 normally assume an extended conformation having the side chains of residues 36, 38, and 40 facing in the same direction. When tyrosine conservatively replaced His36 or lysine replaced Arg38, Asn268 was not glycosylated. But the procathepsin L mutant having phenylalanine in place of Tyr40 was glycosylated at Asn268, which indicates that the hydrogen bond between the hydroxyl group of Tyr40 and the carboxylate group of Asp82 is necessary for normal folding of the nascent proenzyme chain.
Mutation of the adjacent 2p (ERININ) helix of the propeptide or
addition of a C-terminal epitope tag sequence to procathepsin L also
induced misfolding of the proenzyme, as indicated by addition of the
second oligosaccharide chain. In contrast, the propeptide mutation
KAKK99-102AAAA had no effect on carbohydrate modification even though
it reduced the positive charge of the proenzyme.
Misfolded mutant mouse procathepsin L was not efficiently targeted to lysosomes on expression in human HeLa cells, even though it acquired phosphate on mannose residues. The majority of the mutant protein was secreted after undergoing modification with complex sugars. Similarly, epitope-tagged mouse procathepsin L was not targeted to lysosomes in homologous mouse cells but was efficiently secreted. Since production of mature endogenous protease was not reduced in cells expressing the tagged protein, the tagged protein did not compete with endogenous procathepsin L for targeting to lysosomes.
The lysosomal proenzyme receptor (LPR)1 is a 43-kDa integral membrane protein that binds mouse procatL at pH 5 to microsomal membranes from mouse fibroblasts (1). A short synthetic peptide containing a 9-residue sequence from mouse procathepsin L (procatL) inhibits the binding of the procatL to the LPR, but a scrambled version of this peptide does not (2). To study the physiological effects of inhibiting pH-dependent membrane association, we made over a dozen mutants of mouse procatL having from one to four substitutions in the LPR-inhibiting sequence (LIS, Lys-Ser-Thr-His-Arg-Arg-Leu-Tyr-Gly, residues 33-41, preprocatL numbering (2)), which begins 16 residues from the N terminus of procatL.
During expression of these mouse mutants in human HeLa cells, we found that certain LIS residues were critical for normal glycosylation of the proenzyme. Mouse procatL contains two potential sites (Asn221 and Asn268) for cotranslational attachment of an asparagine-linked high mannose oligosaccharide chain (3), but it is normally glycosylated only at Asn221 (4). We have found that mutation of any one of several LIS residues, which are located in the N-terminal 10% of the 317-residue procatL chain, sufficiently changes the folding of the nascent chain during translation that it is also glycosylated at Asn268, a normally cryptic site located in the C-terminal 20% of the procatL chain. Secretion of a protein is often used to indicate that the protein is correctly folded because grossly misfolded proteins are normally degraded in the endoplasmic reticulum (ER) (5). Yet protein misfolding may be missed if secretion is used as the only indicator of correct protein conformation. In this study, the addition of an oligosaccharide chain to a normally cryptic site serves as an indicator that a subtle perturbation of the normal pathway of protein folding has occurred.
Reagents used include the following:
[35S]methionine (Trans35S-label; 850-950
Ci/mmol) and [32P]orthophosphate (400-800 mCi/ml) from
ICN Biomedicals, Costa Mesa, CA; Protein-A Sepharose CL4B from
Pharmacia Biotech Inc.; Dulbecco's modified Eagle's medium, Opti-MEM,
LipofectAMINE, and fetal bovine serum from Life Technologies, Inc.;
aprotinin, dithiothreitol, endo--N-acetylglucosaminidase
H (EndoH), N-glycosidase F, HEPES, pepstatin, and
phenylmethanesulfonyl fluoride from Boehringer Mannheim; enhanced
chemiluminescence Western blotting detection reagents and donkey
anti-rabbit and goat anti-mouse IgG conjugated to horseradish
peroxidase from Amersham Corp.; Immobilon-P from Millipore, Bedford,
MA; brefeldin A from Epicentre Technologies, Madison, WI; Muta-Gene
in vitro mutagenesis kit and prestained molecular mass
marker proteins from Bio-Rad; eukaryotic expression vector pSG5 from
Stratagene, La Jolla, CA; prokaryotic expression vector pET-27b and
HSV-Tag monoclonal antibody from Novagen, Madison, WI; plasmid
purification columns from Qiagen, Chatsworth, CA; oligonucleotides from
Lineberger Nucleic Acids Core Facility, University of North Carolina,
Chapel Hill, NC; restriction endonucleases, Vent polymerase,
Taq polymerase, and DNA modifying enzymes from New England
Biolabs, Beverly, MA, Life Technologies, Inc., or Promega, Madison, WI;
and Sequenase V.2.0 and nucleotides from U. S. Biochemical Corp. The
cysteine protease inhibitor E-64, MEGA-8, and all other reagents were
the highest quality available and were obtained from either
Sigma or Fisher.
Mouse KNIH cells (Kirsten sarcoma virus-transformed NIH 3T3 fibroblasts) were a gift of Dr. C. Scher, University of Pennsylvania. Human HeLa and mouse NIH 3T3 cells were obtained from the American Type Culture Collection. Rockville, MD. Cells were grown in Dulbecco's modified Eagle's medium containing 4.5 g/liter glucose, 10% fetal bovine serum, 0.005% gentamycin, 0.0025% kanamycin, and 200 mM glutamine.
Construction of Mutant cDNAsMutations were introduced into the propeptide of mouse procatL by the method of Kunkel et al. (6) using the Bio-Rad Muta-Gene kit according to the manufacturer's instructions. When possible, codons were chosen to introduce restriction sites that facilitated identification of the mutants. Mutations were confirmed by sequencing relevant regions of the modified single-stranded M13/mp19 phage DNA using Sequenase T7 DNA polymerase according to the manufacturer's instructions. The M13 replicative form was purified on a Qiagen midi-column, digested with BamHI, and the resulting fragments were resolved on an agarose gel. A 1161-base pair fragment encoding mouse procatL was excised and extracted from the agarose using a 0.45 µM filter (7). The fragment was ligated into dephosphorylated, BamHI-digested pSG5 vector and transformed into competent MC1061 cells (8) using standard molecular biology techniques (9). The plasmid pSK1.proCL.Q204 contains a mutation that converts the wild-type Asn268 to Gln268 (4). This plasmid and procatL/pSG5 were both digested with BstXI and SacII. The 630-base pair inserts were exchanged by subcloning as described above.
Construction of ProcatL-HSV Chimeric cDNATo
distinguish recombinant mouse procatL from the endogenous protein in
mouse cells, a 26-amino acid C-terminal epitope tag was added to the
recombinant protein. To construct the plasmid encoding this chimeric
protein, the sequence encoding procatL was first subcloned into the
prokaryotic expression vector pET-27b using the polymerase chain
reaction (PCR). The 5-oligonucleotide primer
(CAGCAGCAGCATATGACTCCAAAATTTGATCAAACCTTT) consisted of bases
encoding an NdeI restriction site followed by the bases encoding the first eight residues of the single chain enzyme. The 3
primer (GTCGTCCTCGAGATTGACCACAGGATAGCT) consisted of bases encoding an
XhoI site followed by bases complementary to the C-terminal 6 residues of procatL. The cDNA was synthesized in a 100-µl PCR reaction mix consisting of 10 ng of template (procatL/pSG5), 200 pM each primer, 2.5 mM deoxynucleotide
triphosphates, 2 mM MgSO4, and 3 units of Vent
DNA polymerase in 1 × Vent reaction buffer. This reaction mixture
was subjected to 1 cycle of 94 °C for 1.0 min, 5 cycles of 94 °C
for 30 s, 58 °C for 30 s, and 72 °C for 1.5 min, and 25 cycles of 94 °C for 30 s and then 72 °C for 1.5 min in a
model PTC-150 thermocycler from M. J. Research, Watertown, MA. The
terminal nucleotides of the PCR product were removed with NdeI and XhoI, and the product was ligated into
the pET-27b vector that had been previously digested with these same
enzymes.
The cDNA encoding procatL and an adjacent 78-base sequence
including bases encoding the herpes simplex virus (HSV) peptide, -SQPELAPEDPED-, and six adjacent 6 histidine residues was removed from
proL/pET-27b and subcloned back into the eukaryotic expression vector
pSG5 using PCR. The 5 primer (GCCAGATCTTCAGTGGTGGTGGTG) contained
bases encoding a BglII site followed by bases complementary to the stop codon and those encoding 4 histidine residues, and the 3
primer (CAGTCTTCTGTGCGTGGCCTTCCACTGGTGCCACTC) was a 36-base pair
sequence homologous to the procatL proregion. This PCR product and
procatL/pSG5 were both digested with BstXI and
BglII, and the 740-base pair inserts were exchanged. The
insert was sequenced in denatured pSG5 using Sequenase T7 DNA
polymerase according to the manufacturer's instructions to confirm
that no mutations were introduced by PCR during subcloning.
Expression plasmid DNA encoding mouse procatL was introduced into HeLa cells or NIH 3T3 fibroblasts using LipofectAMINE. Plasmid DNA (6 µg) and LipofectAMINE (30 µl) were diluted individually with Opti-MEM (300 µl) and then combined and incubated for 45 min at room temperature. Cells (60-mm dishes, 80% confluent) were washed with phosphate-buffered saline (pH 7.2, PBS) and incubated with Opti-MEM (2.5 ml). The DNA-lipofectAMINE complex was then added to the cells, which were incubated for 5 h before addition of Dulbecco's modified Eagle's medium containing 20% fetal bovine serum (3 ml). On day 2, this medium was replaced with cell growth medium or serum-free medium supplemented with insulin (5 µg/ml), transferrin (5 µg/ml), and sodium selenite (5 ng/ml). On day 3, the cells were either harvested for Western blot analysis or were radiolabeled prior to immunoprecipitation of procatL. When specified, control cells were mock-transfected with LipofectAMINE alone.
RadiolabelingProcatL was radiolabeled by washing cell monolayers (60-mm dishes) twice with PBS, starving for 1 h in cell starvation medium (Dulbecco's modified Eagle's medium lacking methionine but containing 4.5 g/liter glucose, 8% dialyzed fetal bovine serum, 0.005% gentamycin, 0.0025% kanamycin, and 200 mM glutamine), and pulsing in cell starvation medium supplemented with [35S]methionine (500 µCi/ml) for the times indicated. For experiments including chase periods, the radiolabeled medium was removed, and the cells were washed twice with PBS and incubated with cell growth medium supplemented with unlabeled methionine (5 mM) for the chase period indicated. When specified, cells were incubated with brefeldin A (20 µg/ml) for 1 h prior to the addition of medium containing the radiolabel and fresh brefeldin A. For radiolabeling with phosphate, the cells were washed, starved for phosphate, and then incubated for 4 h in phosphate-free medium containing [32P]orthophosphate (500 µCi/ml).
ImmunoprecipitationPolyclonal rabbit antiserum was prepared as described previously (3) against mouse procatL expressed in bacterial cells as a fusion protein. A sample of pelleted membranes or supernatant was adjusted to contain 0.6% SDS, 50 mM Tris buffer (pH 9), 100 mM NaCl, and 2 mM EDTA, boiled for 3 min, and centrifuged in a microcentrifuge for 3-5 min. In a clean tube, the supernatant was adjusted to contain 8 mM iodoacetamide, 4% (w/v) Triton X-100, and 26 µg/ml aprotinin. Polyclonal anti-cathepsin L serum (10 µl) was added, and the sample was rotated at 4 °C for 16 h. Antigen-antibody complexes were bound to Protein A-Sepharose (40 µl of a 1:1 resin/water slurry per 10 µl of antiserum) for 3 h at room temperature. The Sepharose beads were washed four times with Tris buffer (150 mM (pH 8)) containing 150 mM NaCl, 5 mM EDTA, and 0.1% SDS. Antigen-antibody complexes were eluted from the beads by incubation at 100 °C for 3 min in PAGE sample buffer (3.3% SDS, 80 mM Tris (pH 7), 20% sucrose, 0.008% bromphenol blue, 17 mM EDTA) containing dithiothreitol (17 mM). Iodoacetamide (83 mM) was added to alkylate the thiol groups. The immunoprecipitate was analyzed by SDS-PAGE on 12% polyacrylamide gels (10) and visualized by fluorography (11). Indicated sizes of proteins were calculated based on the relative positions of commercial radiolabeled and prestained marker proteins.
EndoH TreatmentFollowing immunoprecipitation, samples to
be digested with endo--N-acetylglucosaminidase H (EndoH)
were eluted from the Protein A-Sepharose by boiling for 5 min in 50 mM Tris buffer (pH 8.0, 50 µl) containing 1% SDS and 50 mM dithiothreitol. The eluate was transferred to a clean
tube. The Protein A-Sepharose was washed with 0.3 M citrate
buffer (pH 5.5, 120 µl). Part (100 µl) of this wash was added to
the eluate, EndoH (2 milliunits) was added, and the mixture was
incubated overnight at 37 °C. Proteins were precipitated with 20%
trichloroacetic acid containing yeast tRNA (25 µg/ml) for 1 h on
ice and resolved by SDS-PAGE. For EndoH treatment of nonradioactive
samples (in the absence of immunoprecipitation), cells were scraped
into 0.5% SDS (1.0 ml/60-mm dish), heated in a boiling water bath for
5 min, and sonicated to disrupt DNA. Samples were split in half,
diluted 1:10 with 50 mM NaH2PO4 (pH 5.5), and MEGA-8 was added to a final concentration of 0.5%. EndoH (40 milliunits) was added to one of the two duplicate samples, and both
tubes were incubated at 37 °C for 16 h. The proteins were then
precipitated with trichloroacetic acid as described above. For EndoH
treatment of secreted proteins, cells were incubated overnight in
Dulbecco's modified Eagle's medium containing insulin (5 µg/ml),
transferrin (5 µg/ml), and sodium selenite (5 ng/ml) (3 ml/60-mm
dish). Proteins were precipitated with trichloroacetic acid, and the
precipitate was resuspended and incubated with EndoH as were the cell
samples. Samples treated with N-glycosidase F (600 milliunits) were prepared similarly, except the buffer was 20 mM NaH2PO4 (pH 7.2), 50 mM EDTA, 1%
-mercaptoethanol, and 0.5% MEGA-8.
Cells were washed twice with PBS and harvested by scraping into PAGE sample buffer (0.5 ml) (10). The lysate was sonicated for 10 s and boiled for 5 min. Secreted proteins in cell culture medium were precipitated with 20% trichloroacetic acid containing yeast tRNA (25 µg/ml) and resuspended in PAGE sample buffer containing dithiothreitol (17 mM). Solubilized proteins were resolved on 12% polyacrylamide gels and transferred to Immobilon-P (presoaked in 0.1 M CAPS (pH 11), in 10% methanol) by using a semi-dry apparatus (Integrated Separation Systems, Natick, MA) for 2 h at 1mA/cm2. Nonspecific binding sites were blocked for 30 min with 0.5% non-fat dry milk (Carnation) in Tris buffer (10 mM, pH 7.5) containing NaCl (150 mM) and Tween 20 (0.05%). Rabbit antiserum specific for mouse cathepsin L (diluted 1:15,000 with Tris buffer) or mouse monoclonal antiserum specific for the HSV sequence (diluted 1:20,000 with Tris buffer) was incubated with the blot for 2 h at 37 °C. The blots were washed and incubated in the secondary antibody (horseradish peroxidase-conjugated donkey anti-rabbit or goat anti-mouse IgG diluted 1:20,000 with Tris buffer) for 30 min at room temperature and then washed. Proteins were detected with enhanced chemiluminescence reagent (diluted 2:3 with distilled water) and exposed to film for 1-60 min.
In preparation for a study of the physiological role of the pH-dependent membrane association of procatL, the region of the procatL cDNA that encodes the LPR-inhibiting sequence (LIS) (2) was altered by site-specific mutagenesis (Table I). The various forms of mouse procatL expressed transiently in human HeLa cells were detected in cell extracts and in the cell culture medium either by immunoprecipitation after radiolabeling of the newly synthesized forms or by Western blotting of the forms present under steady-state conditions.
|
Expressed wild-type mouse procatL migrated as a 38-kDa protein (Fig.
1, lanes 4-8), similar to the endogenous
proenzyme of mouse KNIH fibroblasts (Fig. 4). This mass, which was
calculated using both commercial prestained and radiolabeled standards,
is slightly higher than the mass of 36-kDa we reported previously based
on standards radiolabeled in our own lab (12).
The endogenous human procatL of HeLa cells was distinguished from the expressed mouse proenzyme by its molecular mass. The human proenzyme migrated on SDS-PAGE as a 44-kDa protein that was barely detectable on most gels due to its weak cross-reactivity with antiserum raised against SDS-denatured mouse proenzyme (Fig. 1, lanes 1-3, arrow). The reason for this difference in migration of mouse and human procatL remains unclear (13).
Mutant ProcatL Is Larger than Wild-type ProcatLThe first procatL mutant studied had 4 adjacent residues in the middle of the LIS (His36-Arg37-Arg38-Leu39) replaced by 4 alanine residues. HRRL36-39AAAA lacks 2 wild-type arginine residues that are positively charged at pH 5-7 and a wild-type histidine residue that is positively charged at pH 5. When this propeptide mutant protein was expressed in HeLa cells, a 38-kDa form was detected that comigrated with wild-type mouse procatL (Fig. 1, lane 10 versus lane 5). In addition, a 41-kDa form was present that was absent in HeLa cells expressing the wild-type mouse protein. The mutant protein was secreted, although more slowly than the wild-type protein. After a 1-h chase, the 38-kDa form of procatL was detected in the culture medium of cells expressing the wild-type protein but was not detectable in the culture medium of cells expressing the mutant protein (Fig. 1, lanes 7 versus lane 12). After a 4-h chase, however, three forms of the mutant protein were present in the cell culture medium, a major 38-kDa form that comigrated with the secreted wild-type protein, a minor 41-kDa form, and a major 45-kDa form (Fig. 1, lane 13).
Increased Mass of the Mutant Is Due to CarbohydrateTo
determine if the larger forms of this mutant protein were due to
glycosylation of the polypeptide chain, they were treated with EndoH,
an enzyme that removes high mannose but not complex oligosaccharide
chains. EndoH treatment of the wild-type forms of procatL secreted by
transfected HeLa cells decreased the mass of the 38-kDa mouse form by
2.5 kDa, as expected for removal of a single high mannose
oligosaccharide chain (Fig. 2, lane 1 versus lane
2). Thus a complex oligosaccharide chain was not present on this
protein. Similar results (not shown) were obtained for endogenous
procatL secreted by mouse KNIH fibroblasts.
EndoH treatment of the 38- and 41-kDa intracellular forms of the HRRL36-39AAAA mutant produced a single protein band at 35.5 kDa, as expected for removal of all oligosaccharide chains (Fig. 2, lane 3 versus lane 5). Thus the difference in mass between the 38- and 41-kDa forms is due to the presence of a second EndoH-sensitive asparagine-linked high mannose oligosaccharide chain on the 41-kDa form. In contrast, EndoH treatment of the 45-kDa secreted form of this mutant protein only decreased its mass to 42.5 kDa, as expected for removal of a single high mannose oligosaccharide chain (Fig. 2, lane 4 versus lane 6). Thus the difference in mass between the 42.5- and 35.5-kDa forms produced by EndoH treatment is due to the presence of an EndoH-resistant complex oligosaccharide chain on the 42.5-kDa form. The 45-kDa secreted form of the HRRL36-39AAAA mutant of mouse procatL therefore contains two asparagine-linked oligosaccharide chains, one a high mannose chain and the other a complex oligosaccharide chain.
Mutants Are Glycosylated at Asn268, a Normally Cryptic SiteThe 38-kDa form of endogenous mouse procatL has a single high mannose oligosaccharide chain at Asn221 (4), but endoglycosidase treatments revealed that the 41-kDa form of the HRRL36-39AAAA mutant protein had a second high mannose oligosaccharide chain. Since these proteins have only two potential sites for the attachment of an asparagine-linked oligosaccharide chain (Asn221-Asp-Thr, Asn268-Cys-Ser), the second oligosaccharide chain on the 41-kDa mutant proenzyme must be attached at Asn268.
To confirm that the 41- and 45-kDa forms of mutant mouse procatL
contained an oligosaccharide chain at Asn268, this residue
was replaced by mutagenesis. The N268Q mutant of mouse procatL cannot
acquire a second asparagine-linked oligosaccharide chain because it
contains a glutamine residue in place of Asn268 (4). The
Arg38 residue of this mutant was replaced by alanine to
obtain the R38A/N268Q double mutant. Like the HRRL36-39AAAA quadruple
mutant, the R38A single mutant secreted the 41- and 45-kDa forms of the proenzyme (Fig. 3, lane 4). When the
R38A/N268Q double mutant was expressed in HeLa cells, however, the 41- and 45-kDa forms were not detected either within the cells or in the
cell culture medium (Fig. 3, lanes 7 and 8).
Since they were present when Asn268 was present but were
absent when Asn268 was conservatively replaced by
glutamine, the 43- and 45-kDa forms of procatL must contain an
oligosaccharide chain at Asn268, the normally cryptic
glycosylation site.
As expected, EndoH treatment of the intracellular forms of the mutants R38A and R38A/N268Q generated the 35.5-kDa form of the proenzyme, which lacks oligosaccharide (Fig. 3, lanes 5 and 9). EndoH treatment of the secreted forms of the R38A/N268Q mutant also generated only the 35.5-kDa form and no larger EndoH-resistant form (Fig. 3, lane 10). But EndoH treatment of the secreted R38A mutant also produced the larger EndoH-resistant, complex oligosaccharide form (Fig. 3, lane 6). Thus replacing Arg38 by alanine induced addition at Asn268 of a high mannose oligosaccharide that is later modified to complex oligosaccharide.
When wild-type procatL is synthesized in an excessive amount in transformed or transfected cells, a minor amount of the doubly glycosylated 41-kDa form of the proenzyme is detected (Fig. 1, lanes 4, 5, and 8). Thus cotranslational misfolding and glycosylation at Asn268 might also be due in part to the limited availability of a chaperone protein needed for normal folding of wild-type procatL.
Glycosylation at Asn268 Does Not Correlate With Decreased Positive Charge of the Propeptide RegionGlycosylation of the HRRL36-39AAAA mutant at Asn268 might be due to the decreased positive charge of its propeptide region (residues 18-113) resulting from replacement of two positively charged arginine residues (Arg37 and Arg38) by uncharged alanine residues. This possibility was explored by studying the KAKK99-102AAAA triple mutant, which has three positively charged residues (Lys99, Lys101, and Lys102) near the C terminus of the propeptide region replaced by alanines. When this mutant was expressed in HeLa cells, the 41- and 45-kDa forms of procatL were not detected (Fig. 4, lanes 3-7). Like recombinant wild-type mouse procatL, the KAKK99-102AAAA mutant protein comigrated with endogenous procatL that had been immunoprecipitated from radiolabeled mouse KNIH fibroblasts (Fig. 4, lane 1). Also like the recombinant wild-type mouse protein (Fig. 1, lane 7), the KAKK99-102AAAA mutant was detected in the culture medium of transfected HeLa cells after a 1-h chase (Fig. 4, lane 6).
We also prepared WKST32-35AAAA, a propeptide mutant that lacks only one wild-type residue (Lys33) that is positively charged at pH 7. When it was expressed in HeLa cells, both the 38- and 41-kDa forms of the procatL mutant were prominent within the cells. The 38-kDa forms of the KAKK99-102AAAA mutant and the WKST32-35AAAA mutant were present within the cells in comparable amounts after a 30-min pulse with [35S]methionine (Fig. 4, lane 3 versus lane 8). After a 2-h chase, however, the 38-kDa form was present in the culture medium of cells expressing the KAKK99-102AAAA mutant but was still absent from the culture medium of cells expressing the WKST32-35AAAA mutant (Fig. 4, lane 7 versus lane 12). Thus mutation of certain residues in the propeptide region of mouse procatL affects oligosaccharide modification of the protein. Merely decreasing the positive charge is not sufficient to promote glycosylation at Asn268.
ProcatL Bearing Complex Oligosaccharide Is SecretedThe
45-kDa form of the mouse procatL mutants was not detected within HeLa
cells during pulse-chase experiments. When secretion was blocked by
incubating the transfected cells with the fungal metabolite brefeldin
A, all three forms of the WKST32-35AAAA mutant were detected within
the cells (Fig. 5, lane 7) but were absent in
the cell culture medium (lane 8). Thus most of the 41-kDa
form of this mutant, which bears high mannose oligosaccharide chains at
both Asn221 and Asn268, appears to be converted
into the 45-kDa form by modification of one of its high mannose chains
with complex oligosaccharide. The 45-kDa form is secreted rather than
accumulating in cells.
Not All Mutants Are Glycosylated at Asn268
Other residues in the N terminus of procatL were replaced by alanine, and the resulting mutants were examined for their ability to undergo glycosylation at Asn268 (Table I).
Unlike the single mutant R38A (Fig. 3), the single mutants S34A and
R37A and the corresponding double mutant S34A/R37A were not
glycosylated at Asn268 (Fig. 6). Substantial
secreted procatL was detected in the cell culture medium 1 h after
a 30-min pulse (Fig. 6, lanes 4, 9, and 14),
indicating that each of these three mutants moved through the secretory
pathway at a normal rate. After a 3-h chase two forms of the mature
enzyme were detected in cells, the 31-kDa single-chain form of
cathepsin L and the 25-kDa heavy chain of the two-chain form (Fig. 6).
Thus a portion of each of these mutants was targeted to lysosomes and
underwent normal maturation by lysosomal proteolysis. Therefore, the
hydroxyl group of Ser34 and the guanidinium group of
Arg37 are not needed for transport of procatL to
lysosomes.
In contrast, the single mutants H36A and R38A (Fig. 7,
lanes 1-2 and 5-6) and the corresponding double
mutant H36A/R38A (data not shown) did undergo glycosylation at
Asn268. When expressed in HeLa cells, each of these LPR
inhibiting-sequence mutants was present within the cells in the two
high mannose forms, the 38-kDa singly glycosylated form and the 41-kDa
doubly glycosylated form. In addition, each of these mutants was
present in the cell culture medium in the doubly glycosylated 41- and
45-kDa forms.
The two large hydrophobic residues in the LIS of mouse procathepsin L, Leu39 and Tyr40, were also replaced by the small hydrophobic residue alanine. The Y40A mutant was glycosylated at Asn268 (Fig. 7, lanes 11-12), but the L39A mutant was not efficiently glycosylated at Asn268 (Fig. 7, lanes 9-10).
Lack of a Hydroxyl Group Can Result in Glycosylation at Asn268When His36, Arg38, or Tyr40 was replaced by alanine, the resulting mutant proenzyme was cotranslationally glycosylated at Asn268 (Table I and Fig. 7). Each of these residues was next individually replaced with a more conservative residue. When expressed in HeLa cells, the H36Y mutant having the hydroxyphenyl group of tyrosine in place of the imidazole group of His36 was not efficiently glycosylated at Asn268 (Fig. 7, lanes 3-4). Thus, unlike the small side chain of alanine, the large flat phenolic ring of tyrosine sufficiently resembles the large flat imidazole ring of histidine so that the nascent chain of the H36Y mutant folds normally during translation and only undergoes the normal glycosylation at Asn221. Similarly, the R38K mutant having the positively charged methylammonium group of lysine instead of the positively charged guanidinium group of Arg38 was not appreciably glycosylated at Asn268 (Fig. 7, lanes 7-8).
In contrast, the Y40F mutant having the benzene ring of phenylalanine in place of the hydroxybenzene ring of tyrosine did undergo glycosylation at Asn268 (Fig. 7, lanes 13-14). This result is surprising because wild-type mouse procatL, which is not glycosylated at Asn268, contains just one more atom than the Y40F proenzyme mutant, which is glycosylated at Asn268. The crucial atom is the hydroxyl oxygen of Tyr40. These results suggest that the hydroxyl group of Tyr40 participates in at least one hydrogen bond that allows the nascent chain of wild-type procathepsin L to fold during translation in a manner that prevents glycosylation at Asn268.
An ERININ-Helix Mutant Is Glycosylated at Asn268The LIS (residues 33-41) precedes the ERININ
region of procatL (residues 44-63), a discontinuous pattern of six
residues that is conserved in several cysteine proteases (consensus:
E ... R ... F ... N ... I ... N) and folds
into an helix (14, 15). Structural changes in the ERININ helix
might also change the cotranslational folding of the nascent procatL
chain sufficiently to allow the normally cryptic Asn268 to
undergo glycosylation. To test this possibility, a double mutant was
prepared in an effort to retain the conformation of the ERININ
helix but change its surface charge. The positively charged
Arg48 was replaced by a negatively charged glutamate
residue, which should increase the surface charge of the ERININ
helix by +2, and the large aromatic hydrophobe Trp52 was
replaced by the large aliphatic hydrophobe leucine, which should
promote formation of an
helix. When expressed in HeLa cells, the
resulting R48E/W52L double mutant was indeed glycosylated at
Asn268 (Fig. 7, lanes 15 and 16).
Secretion of this mutant form was reduced, however, suggesting that a
significant portion of the protein may be degraded in the ER.
To determine if the
conformational determinant recognized by the UDP-GlcNAc:lysosomal
enzyme N-acetylglucosamine 1-phosphotransferase (16) is
preserved in the propeptide mutants, three of these mutants were
expressed in HeLa cells in the presence of
[32P]orthophosphate, and the secreted forms of procatL
were examined for the presence of the radiolabel. The 38-kDa forms of
wild-type mouse procatL (Fig. 8, lane 1), the
single mutants S34A (lane 2) and R38A (lane 3),
and the double mutant R38A/N268Q (lane 4) were each labeled
with [32P]phosphate. The 41- and 45-kDa doubly
glycosylated secreted forms of the R38A mutant (lane 3) were
also phosphorylated. In each case, the radiolabel was removed by
treatment with EndoH (data not shown), indicating that the
[32P]phosphate was attached to a high mannose
oligosaccharide chain.
The R38A/N268Q double mutant must be folded sufficiently correctly for the phosphotransferase to recognize its single high mannose oligosaccharide chain at Asn221 and to attach N-acetylglucosamine 1-phosphate to one or more of its mannose residues. By analogy, it is likely that the R38A single mutant is also phosphorylated on its high mannose oligosaccharide chain at Asn221, so that the EndoH-resistant complex oligosaccharide chain of its 45-kDa secreted form would be located at Asn268.
Addition of a C-terminal Epitope Tag Induces Glycosylation at Asn268 and Inhibits Lysosomal TargetingPropeptide mutants that did not acquire the second carbohydrate chain were targeted to lysosomes as revealed by the presence of mature forms of the enzyme in cells (Fig. 6). In contrast, mutants that were misfolded, as revealed by addition of the second oligosaccharide chain to at least a proportion of the expressed protein molecules, did not reach lysosomes. However, even targeting of wild-type mouse procatL was relatively inefficient in human HeLa cells, making comparison of the relative targeting efficiency of the various mutants difficult. To determine if the efficiency of lysosomal targeting of the expressed mouse mutants might be greater in homologous mouse cells than in human HeLa cells, a mouse procatL mutant cDNA was constructed that encoded a 26-residue epitope tag, containing a 12-residue segment of a herpes simplex virus (HSV) protein, attached to the C terminus of the proenzyme. The epitope-tagged mouse proenzyme was expressed in mouse NIH fibroblasts, and the recombinant and endogenous forms of procatL were visualized under steady-state conditions by consecutive Western blotting with an antiserum that specifically recognizes the tag on the recombinant proenzyme, followed by an antiserum that recognizes both the recombinant and the endogenous mouse procatL.
The anti-HSV serum specific for the recombinant protein detected 41-kDa
cellular and secreted forms of procatL, larger than the endogenous
protein due to the presence of the extra C-terminal amino acid residues
(Fig. 9, panel A). In addition, a 43-kDa form of tagged procatL was detected in the transfected cells (Fig. 9,
lanes 3 and 5), and a 46-kDa form of the protein
was present in the cell culture medium (Fig. 9, lanes 4 and
6). No mature forms of the protease were detected in cells,
and no cross-reacting protein was present in the mock-transfected
control cells treated with LipofectAMINE alone (Fig. 9, lanes
1-2). Subsequent reaction of the same blot with antiserum
specific for mouse procatL detected, in addition to the above forms of
expressed procatL, the endogenous forms expected (panel B),
38-kDa procatL and the 30-kDa single-chain and 23-kDa heavy-chain forms
of the mature protease.
Failure to detect mature forms of procatL with the HSV-specific antiserum indicated that the expressed protein was not being targeted to lysosomes, even in the homologous mouse cells. To improve the chances of detecting mature forms of cathepsin L, cells were treated with leupeptin which slows conversion of the single-chain protein to the two-chain protein (Fig. 9, panel B, lanes 7 and 9 versus lane 11). However, the single-chain form of the epitope-tagged protein was still not detectable (Fig. 9, panel A, lane 5). The 25-kDa heavy chain of the two-chain form of cathepsin L cannot be detected by the HSV serum because it is derived from the N terminus of the single-chain protease and thus lacks the tag. Although it is derived from the C terminus of the single-chain protease and thus bears the HSV tag, the small light chain of the two-chain form of cathepsin L was not detected by the anti-HSV serum or the procatL antiserum because it is not retained by the 12% polyacrylamide gel used in this experiment.
The larger forms of epitope-tagged procatL appeared similar to the
mutant forms of the proenzyme that acquire a second oligosaccharide chain. Treatment with EndoH converted the epitope-tagged 41- and 43-kDa
cellular proteins (Fig. 10, lane 1) to a
single 39-kDa protein (Fig. 10, lane 3), confirming that the
41-kDa form possessed one and the 43-kDa form possessed two high
mannose oligosaccharide chains. The EndoH resistance (Fig. 10,
lane 4) and N-glycosidase F sensitivity (Fig. 10,
lane 6) of the 46-kDa secreted proenzyme indicated that this
protein had been modified with complex carbohydrate. A portion of the
secreted 46-kDa protein was completely EndoH-resistant, indicating both
oligosaccharide chains had acquired complex carbohydrate and therefore
should lack mannose 6-phosphate. The protein migrating as the middle
band (Fig. 10, lane 4) apparently possessed one high mannose
oligosaccharide chain and one oligosaccharide chain modified with
EndoH-resistant complex sugars. As for several point mutations in the
propeptide, the presence of the C-terminal tag sequence alters protein
folding sufficiently to permit a significant portion of the expressed
protein to acquire the second oligosaccharide chain at
Asn268. The epitope-tagged protein is secreted but is not
sorted to lysosomes, which suggests that even the portion lacking the
second oligosaccharide chain is misfolded.
Our previous
studies indicated that the mannose 6-phosphate-independent association
of mouse procatL with microsomal membranes at acidic pH requires the
propeptide of the proenzyme (1). A synthetic peptide containing the LIS
(residues 33-41) inhibits the binding of procatL to membranes in
vitro, suggesting that the LIS is involved in this membrane
association (2). The recently published three-dimensional structure of
human procatL (15) reveals that the LIS forms a solvent-exposed elbow
between helices 1p and
2p, which is compatible with the LIS
playing a role in the binding of procatL to the LPR. Before this
structure became available, we explored the physiological function of
this mannose 6-phosphate-independent membrane association by examining
the effect of changing one or more LIS residues on procatL targeting. Our studies revealed that several wild-type LIS residues are critical for normal folding and glycosylation of procatL, which has complicated a study of the role of the LIS in the cellular targeting of
procatL.
We found that substantial glycosylation of the normally cryptic Asn268 serves as an indicator of abnormal folding of mouse procatL. Four single LIS mutants (H36A, R38A, Y40A, Y40F) underwent substantial glycosylation at Asn268 but five others (S34A, H36Y, R37A, R38K, L39A) did not. Alanine scanning of LIS residues 36 through 40 revealed the following pattern:
![]() |
![]() |
Conservatively replacing His36 by tyrosine or Arg38 by lysine maintains Asn268 in its normal cryptic state. But the presence of Tyr40 is clearly necessary to avoid cotranslational misfolding and glycosylation at Asn268 because it cannot be replaced by phenylalanine. This suggested that the hydroxyl group of Tyr40 forms a hydrogen bond that is necessary for normal folding of the nascent proenzyme. Indeed, the three-dimensional structure of human procatL (15) reveals that the hydroxyl group of Tyr40 forms a hydrogen bond to the carboxylate group of Asp82. The absence of this single hydrogen bond evidently alters the interaction of the N-terminal globular domain with the rest of the nascent procatL sufficiently to allow glycosylation at Asn268.
Phosphorylation of Mannose ResiduesThe enzyme that transfers phosphate to the high mannose oligosaccharide chain of wild-type procatL recognizes conformational determinants on the proenzyme (17) that are apparently preserved in representative N-terminal mutants of mouse procatL. We expected that both high mannose oligosaccharide chains of the 41-kDa form of a procatL mutant would be phosphorylated. Assay of the phosphorylation of high mannose oligosaccharide chains introduced at novel sites on cathepsin D has shown that the phosphotransferase can modify chains attached to a variety of sites on the proenzyme (18). Our results suggest, however, that only one of the two oligosaccharide chains on mutant procatL acquires phosphate. Most of the 41-kDa mutant form is transformed into the 45-kDa form having one EndoH-sensitive, high mannose oligosaccharide chain and one EndoH-resistant, complex oligosaccharide chain. Since a phosphorylated high mannose oligosaccharide chain cannot be converted into a complex oligosaccharide chain (19), only one of the two high mannose oligosaccharide chains of the 41-kDa mutant form is phosphorylated during passage through the early Golgi compartments. The normal glycosylation and mannose phosphorylation site at Asn221 is on the opposite side of procatL from the cryptic glycosylation site at Asn268 (15). Evidently the phosphotransferase cannot efficiently modify the high mannose oligosaccharide chain at Asn268.
Proenzyme SecretionThe 41-kDa form of mouse procatL mutants,
which has high mannose oligosaccharide chains at both
Asn221 and Asn268, is detected to a significant
extent within the transfected HeLa cells and to a minor extent in the
cell culture medium. The majority of the secreted mutant protein is the
45-kDa form, which is produced in the trans Golgi by conversion of the
high mannose oligosaccharide at Asn268 of the 41-kDa form
into an EndoH-resistant, complex oligosaccharide. Unless its secretion
is artificially blocked by brefeldin A, the 45-kDa form is rapidly
secreted and is not detected within the cells by pulse-chase assays or
Western blots. Similarly, the plasma proteins
1-proteinase inhibitor and
2-macroglobulin are rapidly secreted from hepatocytes
once they acquire complex carbohydrate (20). Thus the protein
misfolding detected by addition of the second oligosaccharide chain
does not significantly alter passage through the secretory pathway, but
it may hinder targeting to lysosomes.
Targeting of mouse procatL to lysosomes was relatively inefficient in human HeLa cells even though the proenzyme contained mannose 6-phosphate. Mature forms of the mutants were readily detected only when a mutant that did not alter protein folding, such as S34A or R37A, was expressed in a large amount. After adjusting for the number of methionine residues, 70% of newly synthesized procatL was secreted, and only about 20% was processed to mature forms. Similarly, procathepsin D targeting has been reported to be very inefficient in heterologous cells (21). This difference in lysosomal sorting between the two species is difficult to understand if only mannose 6-phosphate receptors are involved.
In an attempt to increase the targeting efficiency of mutant proteins, the mouse enzyme was expressed in homologous mouse cells instead of in heterologous human cells. To distinguish the recombinant protease from the endogenous protein, an epitope tag sequence was added to the C terminus of the recombinant protein. As seen for several of our N-terminal propeptide mutants, this modification resulted in cotranslational misfolding and addition of a second oligosaccharide chain. Similarly, we have previously observed that Asn268 is glycosylated when the conformation of mouse procatL is altered by insertion of a 21-residue tag sequence just before the C terminus (4). Evidently, addition of the high mannose oligosaccharide at Asn268 does not occur cotranslationally as the Asn268 site of procatL emerges from the ER membrane but after at least part of the C-terminal tag sequence has emerged from the ER membrane, when the synthesis of procatL is essentially complete. Similarly, folding of the yeast vacuolar enzyme procarboxypeptidase Y can precede glycosylation at Asn (22).
The epitope-tagged protein acquired complex carbohydrate and was secreted. Even in homologous cells it was not efficiently targeted to lysosomes, as judged by the absence of mature forms of cathepsin L. Since that part of the tagged protein which lacks the oligosaccharide at Asn268 also fails to target to lysosomes, it is unlikely that this second oligosaccharide chain prevents lysosomal targeting of the tagged protein. Thus either the tag itself or a conformational change it induces may prevent interaction with a sorting mediator.
The presence of the second oligosaccharide chain at Asn268 indicates that a mutant or tagged procatL form is misfolded. Two observations suggest that misfolding of a mutant procatL is sufficient to alter its targeting. First, the procatL mutants of procatL that were glycosylated at both Asn221 and Asn268 accumulated in the culture medium more slowly than the singly glycosylated wild-type recombinant proenzyme. After a 1-h chase, the proportion of mutant forms in the cell culture medium was clearly less than that for the wild-type forms. But after a 4-h chase, a substantial proportion of the mutant forms was detected outside the cells. In contrast, grossly misfolded proteins are not secreted but are degraded in the ER by unknown mechanisms (5). For example, deletion of the complete propeptide region of mouse procatL results in turnover of the misfolded protein in the ER (23). Second, on cell lysis the recombinant wild-type procatL remains soluble, but the recombinant proenzyme mutants that acquire the second oligosaccharide chain aggregate to a significant extent (data not shown). This result, coupled with their slower accumulation in the cell culture medium and the addition of the second oligosaccharide chain, suggests that these mutant proteins remain misfolded to a significant degree after exit from the ER whether or not they have acquired the second carbohydrate chain. This misfolding might interfere with their ability to interact with a mediator of lysosomal targeting.
Expression of the tagged procatL might be expected to produce a dominant-negative phenotype characterized by excessive secretion of the endogenous procatL. Lysosomal targeting of endogenous procatL, however, was not altered when the misfolded tagged protein was transiently expressed. Evidently the misfolded tagged protein does not efficiently compete with the normally folded endogenous protein for interaction with the sorting mediator.
Roles of the LIS ResiduesThis study has produced four single LIS mutants (H36A, R38A, Y40A, and Y40F) of mouse procatL that undergo cotranslational misfolding, substantial glycosylation at Asn268, and subsequent secretion. Four other propeptide mutants (WKST32-35AAAA, HRRL36-39AAAA, H36A/R38A, and R48E/W52L) also exhibit this altered phenotype. Clearly it is inappropriate to use any of these phenotypically altered mutant proteins to study the role of the LIS in the lysosomal targeting of mouse procatL.
The three-dimensional structure of human procatL (15) indicates that four LIS residues common to the human and mouse proenzymes (Lys33, His36, Arg38, and Tyr40) are structurally important because their side chains contribute to the hydrophobic core of the N-terminal globular domain. Some of them are likely to be important for the normal folding of this domain, which is probably the earliest event in the cotranslational folding of procatL. Indeed, the altered phenotype of four mouse procatL mutants (H36A, R38A, Y40A, and Y40F) is consistent with three of these LIS residues (His36, Arg38, and Tyr40) being important for normal folding of this domain and its later interactions with the rest of the proenzyme. It is well established that a propeptide can play this chaperone role during the folding of a nascent proenzyme chain (24, 25).
In contrast, the four intervening LIS residues of mouse procatL
(Ser34, Thr35, Arg37, and
Leu39) have their side chains exposed to solvent near the
end of the 1p helix. Some of these external LIS residues might be
involved in the binding of the full-length procatL to the LPR. Also,
the interaction of the LIS with the LPR may change as the protein moves
from the ER, where at pH 7 the imidazole ring of His36
should be predominantly uncharged, to the acidic prelysosome, where at
pH 5 this imidazole ring should be positively charged. Further studies
of these mutants should reveal the role of these LIS residues in the
lysosomal targeting of mouse procatL.
We thank Leah Mechanic for assistance with construction of the R38A/N268Q mutant.