Mutations in sialidosis impair sialidase binding to the lysosomal multienzyme complex.

Sialidosis is an autosomal recessive disease caused by the genetic deficiency of lysosomal sialidase, which catalyzes the catabolism of sialoglycoconjugates. The disease is associated with progressive impaired vision, macular cherry-red spots, and myoclonus (sialidosis type I) or with skeletal dysplasia, Hurler-like phenotype, dysostosis multiplex, mental retardation, and hepatosplenomegaly (sialidosis type II). We analyzed the effect of the missense mutations G68V, S182G, G227R, F260Y, L270F, A298V, G328S, and L363P, which are identified in the sialidosis type I and sialidosis type II patients, on the activity, stability, and intracellular distribution of sialidase. We found that three mutations, F260Y, L270F, and A298V, which are clustered in the same region on the surface of the sialidase molecule, dramatically reduced the enzyme activity and caused a rapid intralysosomal degradation of the expressed protein. We suggested that this region might be involved in sialidase binding with lysosomal cathepsin A and/or beta-galactosidase in the multienzyme lysosomal complex required for the expression of sialidase activity. Transgenic expression of mutants followed by density gradient centrifugation of cellular extracts confirmed this hypothesis and showed that sialidase deficiency in some sialidosis patients results from disruption of the lysosomal multienzyme complex.

Sialidosis (also called mucolipidosis I and cherry-red spot myoclonus syndrome) is an autosomal recessive disease caused by the genetic deficiency of lysosomal sialidase, also called neuraminidase (reviewed in Refs. [1][2][3]. The disease is characterized by tissue accumulation and urinary excretion of sialylated oligosaccharides and glycoproteins (1) and includes two main clinical variants with different ages of onset and degrees of severity. Sialidosis type I or nondysmorphic type is a lateonset mild form, characterized by bilateral macular cherry-red spots, progressive impaired vision, and myoclonus syndrome (4 -8). Sialidosis type II or dysmorphic type is the infantileonset form, which is also associated with skeletal dysplasia, Hurler-like phenotype, dysostosis multiplex, mental retardation, and hepatosplenomegaly (9 -12). A severe form of the disease manifests itself prenatally and is associated with ascites and hydrops fetalis (13)(14)(15). The age of onset and severity of the clinical manifestations correlate with the amount of residual sialidase activity, suggesting the existence of considerable genetic heterogeneity (1)(2)(3).
Although sialidosis was recognized as a deficiency of lysosomal sialidase from the moment of its discovery (16), the molecular mechanism of this disorder was not characterized for the following two decades because the identification and sequencing of sialidase had been hampered by low tissue content and instability of the enzyme. Several works have shown that sialidase is a part of a multienzyme complex containing other lysosomal enzymes such as cathepsin A (protective protein), ␤-galactosidase, and N-acetylgalactosamine-6-sulfate sulfatase (17)(18)(19). Because the functional activity of sialidase completely depends on the integrity of its association with cathepsin A, it was hypothesized that cathepsin A supports catalytically active conformation of this enzyme (18). In addition, the complex protects sialidase and ␤-galactosidase against rapid proteolysis (17,20,21) and may also be important for proper sorting and processing of their precursors (22)(23)(24)(25). In the autosomal recessive disease galactosialidosis, a primary genetic defect of cathepsin A (17,21) results in disruption of the complex and causes the combined deficiency of ␤-galactosidase and sialidase activities. The clinical features and a composition of storage products in galactosialidosis resemble those in sialidosis (8,9,26).
Recently the gene coding for sialidase was cloned, and a series of mutations in sialidosis patients was identified (27)(28)(29)(30)(31). In particular, we have found two frameshift and eight missense mutations in nine sialidosis patients of multiple ethnic origin (28,31). To understand the effect of these mutations on sialidase, we modeled the tertiary structure of the enzyme and localized the identified amino acid substitutions (31). Surprisingly, none of mutations directly affected the deduced active site residues or were found in the central core of the sialidase molecule, but all of them involved residues on the surface of the enzyme. Therefore, in most cases it was unlikely that these mutations would introduce electrostatic or steric clashes in the protein core leading to general folding defects of sialidase and its retention in the endoplasmic reticulum/Golgi compartment, as was observed in most of the mutations affecting cathepsin A (32).
In this paper we show that three sialidase mutants that have amino acid substitutions clustered in one region on the surface of the sialidase molecule were correctly processed and sorted but were not associated with the complex and were rapidly degraded in the lysosome. These results permitted us to conclude that the surface region containing these mutations may be involved in the sialidase binding interface with the lysoso-mal multienzyme complex and that sialidase deficiency in sialidosis patients may be secondary to the disruption of the lysosomal multienzyme complex.

EXPERIMENTAL PROCEDURES
Expression of Sialidase Mutants in COS-7 Cells-Site-directed mutagenesis was performed using a Transformer TM site-directed mutagenesis kit (CLONTECH), the previously described pCMV-SIAL expression vector, mutagenic primers corresponding to mutant sialidase sequences, and a selection primer used to eliminate a unique ScaI restriction site in the vector according to supplier's protocols. All primers were enzymatically phosphorylated, and for each mutant the corresponding mutagenic primer and the selection primer were annealed to a heatdenatured pCMV-SIAL plasmid. After elongation by T4 DNA polymerase, ligation, and primary digestion with ScaI restriction enzyme to linearize all nonmutated DNA, the plasmid pool was used to transform the mutS strain of BMH71-18 bacteria. Plasmid DNA obtained from the pool of ampicillin-resistant transformants was subjected to a second ScaI digestion and transformed into Escherichia coli DH5␣. Positive clones were selected after a final ScaI restriction analysis, and the entire sialidase cDNA was sequenced. Up to 80% of transformants contained the desired mutation. DNA fragments of between 300 and 600 base pairs containing the introduced mutations were obtained from the mutant pCMV-SIAL plasmids by double digestion with either BstEII/NaeI, NaeI/KpnI, or KpnI/EcoRV enzymes and subcloned into the parental pCMV-SIAL plasmid. The final constructs were verified by sequencing.
COS-7 cells seeded in T-25 flasks or 60-mm round dishes were co-transfected with pCMV-SIAL and pCMV-CathA expression vectors (31) using LipofectAMINE Plus reagent (Life Technologies Inc.) in accordance with the manufacturer's protocol. 48 h after transfection, sialidase and control N-acetyl-␤-glucosaminidase activities were assayed in cellular homogenates using the corresponding fluorogenic 4-methylumbelliferyl glycoside substrates as described (33)(34)(35). The cathepsin A activity was determined with benzyloxycarbonyl-Phe-Leu and 3-(2-furyl)acryloyl-Phe-Leu substrates (36). One unit of enzyme activity is defined as the conversion of 1 mol of substrate/min. Proteins were assayed according to Bradford (37) with BSA 1 (Sigma) as the standard. To measure the stability of the expressed sialidase, the cellular homogenate was incubated at 37°C for 0.5, 1, 2, and 3 h before the assay of sialidase activity.
Metabolic Labeling-48 h after transfection with wild-type or mutant sialidase cDNA, COS-7 cells grown to confluency in 60-mm dishes were washed twice with Hank's balanced salt solution, incubated for 2 h in cysteine and methionine-free Dulbecco's modified eagle's medium (Life Technologies, Inc.) supplemented with L-glutamine and sodium pyruvate, then incubated again for 40 min in 5 ml of the same medium supplemented with a mixture of [ 35 S]Cys and [ 35 S]Met (Tran 35 S-label, ICN Pharmaceuticals Inc.; 0.1 mCi/ml medium). The radioactive medium was then removed, and the cells were washed twice with Hank's balanced salt solution and chased at 37°C in Eagle's minimal essential medium containing 20% (v/v) fetal calf serum with and without protease inhibitor leupeptin (5 g/ml).
Immunoprecipitation, Electrophoresis, and Quantitation of Sialidase-1.0 ml of lysate was incubated for 4 h with preimmune serum at a final dilution of 1:20. Then the pellet obtained from 300 l of Pansorbin cells (Calbiochem) was added, and the resulting suspension was incubated for 2 h at 4°C, followed by centrifugation for 10 min at 13,000 ϫ g. Supernatants were incubated overnight with the antisialidase antibodies at a 1:100 final dilution, then for 2 h at 4°C with the pellet from 100 l of Pansorbin cells and precipitated as above. The pellet was washed three times with 1 ml of radioimmunoprecipitation buffer. The antigens were eluted from the pellet by the addition of 100 l of a buffer containing 0.1 M Tris-HCl (pH 6.8), 4% (w/v) SDS, 20% (v/v) glycerol, 0.2 M dithiothreitol, and 0.02% (w/v) bromphenol blue. The proteins were denatured by boiling for 5 min, and 50 l of each sample were subjected to SDS-polyacrylamide gel electrophoresis ac-cording to Laemmli (38). The molecular weights were determined with 14 C-labeled protein markers (Amersham Pharmacia Biotech). The gels were fixed in acetic acid/isopropyl alcohol/water (10:50:40), soaked for 30 min in Amplify™ solution (Amersham Pharmacia Biotech), vacuumdried at 60°C, and analyzed by quantitative fluorometry on a Phospho-rImager SI analysis screen (Molecular Dynamics) using the software supplied by the manufacturer.
Immunofluorescent Microscopy-48 h after transfection with wildtype or mutant sialidase, COS-7 cells were treated for 40 min with 75 nM LysoTracker Red DND-99 (Molecular Probes, Eugene, OR) dye, washed twice with ice-cold PBS, and fixed with 3% paraformaldehyde in PBS for 40 min. Cells were permeabilized by incubating with 0.3% Triton X-100, washed twice with PBS, and stained with rabbit polyclonal anti-sialidase antibodies and fluorescein isothiocyanate-conjugated monoclonal antibodies against rabbit IgG. Alternatively, cells were double-stained with rabbit polyclonal anti-sialidase antibodies and monoclonal antibodies against lysosomal membrane marker LAMP-2 (Washington Biotechnology Inc., Baltimore, MD). Slides were studied on a Zeiss LSM410 inverted confocal microscope (Carl Zeiss Inc., Thornwood, NY).
Density Gradient Centrifugation of Cell Extracts-COS-7 cells grown to confluency in T-25 flasks and harvested 48 h after transfection with wild-type or mutant sialidase were solubilized in 0.2 ml of 0.15 M sodium acetate buffer, pH 5.2, containing 0.5 mg of BSA/ml and 1% (w/v) Zwitterionic TM detergent 3-12 (Calbiochem) as described (19) and centrifuged at 13,000 ϫ g for 15 min. The supernatants were applied on the top of the density gradient of 30% metrizamide (OptiPrep; Nycomed Amersham) preformed by a 2-h ultracentrifugation at 45,000 rpm in a Beckman SW-55 Ti swinging bucket rotor. After application of the sample, the centrifugation was continued for an additional 17 h at the same speed. Immediately after centrifugation, each tube was divided into 10 fractions using a Beckman tube slicer kit. Each fraction was assayed for activities of sialidase, ␤-galactosidase, and cathepsin A as well as for the presence of human sialidase and cathepsin A protein by Western blot as previously described (20). The activity of endogenous N-acetyl-␤-hexosaminidase in fractions was used as an internal control. The molecular masses of proteins were approximated using the following

Expression and Intracellular Targeting of Sialidase Mutants-
The effect of sialidase mutations on enzyme biogenesis was studied by the transient expression of the mutant cDNA. Mutations were generated by site-directed mutagenesis in the pCMV-SIAL vector previously used for the expression of sialidase (28). Short restriction cassettes containing the mutations were then inserted into the parental pCMV-SIAL vector replacing the corresponding fragments of wild-type sialidase cDNA. The inserts and junction regions of the resulting constructs were verified by sequencing to ensure the correct introduction of mutations. Mutant or wild-type sialidase was co-expressed with human cathepsin A, which is necessary for the expression of sialidase activity. 48 h after transfection, the cell lysates were assayed for sialidase, cathepsin A, and control ␤-hexosaminidase activities.
The expression results are shown in Table I. All transfected cells had similar cathepsin A activity, suggesting the same transfection efficiency for all cells. Four of the expressed mutants, G68V, G227R, A298V, and L363P, had very low (Ͻ10% of normal) sialidase activity. The activity of F260Y and L270F mutants was between 10 and 20% of normal, and that of S182G and G328S mutants was between 20 and 40% of normal. Additional experiments showed that F260Y, A298V, and L270F mutants were also significantly less stable than the wild-type sialidase. The half-life of their enzymatic activity in cellular lysates at 37°C was about 30 min as compared with the 2-h half-life of the wild-type enzyme.
Using immunolabeling, we studied the intracellular distribution of the sialidase mutants expressed in COS-7 cells. To identify the lysosomal late endosomal compartment, the COS-7 cells were treated for 40 min with 75 nM LysoTracker Red DND-99 dye prior to fixation and immunostaining with antisialidase antibodies. Alternatively the cells were double-stained with anti-sialidase antibodies and monoclonal antibodies against human LAMP-2. For the wild-type sialidase we have observed the complete co-localization of anti-sialidase immunostaining with lysosomal markers LysoTracker Red (Fig. 1) or LAMP-2 (not shown). The G68V, S182G, F260Y, L270F, A298V, and G328S mutants showed similar localization, suggesting that the mutant protein is able to reach the lysosomes. Although partial co-local-ization of anti-sialidase and LysoTracker staining was also detectable in the cells transfected with the G227R and L363P mutants, the majority of the anti-sialidase antibodies labeled distinct cellular areas, suggesting that in this case the mutant protein is mostly retained in prelysosomal compartments. This finding is consistent with the results of structural modeling of sialidase mutants that suggested general folding defects and retention in the endoplasmic reticulum/Golgi compartment for both G227R and L363P substitutes (31).
Metabolic Labeling of Sialidase Mutants-The results of sialidase activity assay in COS-7 cells expressing sialidase mutants have shown that some of them, i.e. L270F, A298V, and F260Y mutants, had significantly lower stability in cellular homogenates than the wild-type enzyme. To measure the stability of sialidase mutants in the cell, we performed pulse-chase experiments. The 46 -48-kDa polypeptides similar to those previously observed by both immunoprecipitation and Western blotting (20) were precipitated by anti-sialidase antibodies from homogenates of cells transfected with wild-type or mutant sialidase cDNA and pulsed for 40 min (Fig. 2). The intensities of both bands decreased proportionally with the time of chase. By 4 h of chase, normal wild-type sialidase was reduced to ϳ50% of total. In contrast, the degradation rate of F260Y, L270F, and A298V mutants was remarkably increased so that for all these cells 46 -48-kDa sialidase bands were already nearly undetectable after 4 h of chase. The same degradation rate was observed with and without leupeptin, a potent inhibitor of lysosomal serine and thiol proteases, suggesting that they are not involved in the degradation of mutant sialidase.
Association of Sialidase Mutants with the Lysosomal Multienzyme Complex-The ability of sialidase mutants to associate with the lysosomal multienzyme complex was studied by the density gradient ultracentrifugation of the cell extracts (Fig. 3). In the extracts of COS-7 cells co-transfected with FIG. 1. Immunohistochemical localization of sialidase mutants expressed in COS-7 cells. COS-7 cells were probed with 75 nM lysosomal marker LysoTracker Red DND-99 for 30 min at 37°C 48 h after transfection with cathepsin A and wild-type (WT) or mutant sialidase cDNAs as indicated and stained with rabbit polyclonal anti-sialidase antibodies and fluorescein isothiocyanate-conjugated secondary antibodies. Slides were studied on a Zeiss LSM410 inverted confocal microscope. Panels show co-localization of anti-sialidase antibodies (green) and LysoTracker marker (red). Magnification ϫ 600.

Mutation Enzymatic activity Traffic to lysosomes
Short wild-type sialidase and cathepsin A, all sialidase activity was associated with the peak that sedimented before thyroglobulin (M r ϭ 669,000). This peak, which also contained almost all the cathepsin A and the majority of endogenous ␤-galactosidase activities, probably represented the lysosomal multienzyme complex (Fig. 3A). A similar sedimentation profile was observed in the extracts of cells transfected with sialidase G328S or S182G mutants. Although about 3-fold less sialidase activity was detected in the collected fractions as compared with that of the wild-type control, all activity was associated with the high molecular weight fraction. The distribution of sialidase and cathepsin A proteins detected by Western blot (Fig. 3B) followed that of the enzyme activity. In contrast, in the cells transfected with G227R, F260Y, L270F, and A298V mutants, the high molecular weight form of sialidase was not detected. Both sialidase protein and the trace levels of sialidase activity were found in the peak that sedimented together with the low molecular weight marker, BSA (M r ϭ 69,000), suggesting that the mutant enzyme does not associate with the multienzyme complex. Moreover, although a significant amount of sialidase cross-reacting protein was detected in these fractions for F260Y, L270F, and A298V transfected cells, the sialidase activity was ϳ50 -100-fold reduced as compared with wild-type enzyme, which is consistent with the inactivation of sialidase after the dissociation from the complex.

DISCUSSION
Analysis of molecular defects in the sialidase gene in sialidosis patients shows that the spectrum of mutations is very different from that in cathepsin A and ␤-galactosidase, which underlies clinically similar disorders galactosialidosis and G M1 -gangliosidosis, respectively. Most of the sialidosis patients studied so far, 21 of 27, had amino acid substitutions and not frameshift or splicing defects 2 (27,28,30,31,42). The localization of the missense mutations on the sialidase structural model (Fig. 4) suggested that only few of them (shown in blue in Fig. 4) affect active site residues (Y370C) or may interfere with their correct positions (L91R with the active site residue Arg 78 , P80L with Arg 97 , duplication of His 399 and Tyr 400 with Glu 394 , P316S with Arg 280 , and P335Q with Arg 341 ). In addition, the L363P mutation is situated on a ␤-strand adjacent to that containing the active site residue Tyr 370 . The Leu 363 residue is probably necessary to anchor this ␤-strand to the one containing Tyr 370 so that the L363P mutation can potentially also have an effect on the active site. However, in contrast to cathepsin A mutations in galactosialidosis patients, which mostly affect the enzyme central core and cause unfolding of the protein (32), the majority of sialidase mutations involves residues on the surface of the enzyme and is not supposed to result in significant structural change. Moreover, the distribution of mutations on the sialidase surface is uneven. The region that contains the majority of mutations resulting in complete or almost complete inactivation of the enzyme and causing severe sialidosis type II phenotype is easily detectable (shown in red in Fig. 4). In particular this region contains mutations G227R, F260Y, L270F, and A298V (28,31), R294S, L231H, and G218A (30), W240R 2 , and V217M and G243R (42).
We expressed eight sialidase mutants, four of which contained amino acid substitution in the defined surface patch (G227R, F260Y, L270F, and A298V) and four in the other areas of the sialidase molecule (G68V, S182G, G328S, and L363P) in COS-7 cells and studied trafficking, activity, and stability of the produced protein. We found that in two cases (G227R and 2 H. Sakuraba, private communication. 48 h after transfection, cellular extracts were analyzed by density gradient centrifugation as described under "Experimental Procedures." A, the 0.5 ml-fractions were collected and assayed for sialidase activity (Ⅺ, left ordinate), cathepsin A activity (‚, right ordinate), and ␤-galactosidase activity (E, right ordinate). N-Acetyl-␤-glucosaminidase activity in fractions (not shown) was used as an internal control. Each curve represents the average of several independent experiments. The positions of the sedimentation peaks of the M r standards described under "Experimental Procedures" are shown by arrows. B, the indicated fractions with the peak sialidase and cathepsin A activities were analyzed for the presence of sialidase and cathepsin A protein by Western blotting as described. The protein bands are indicated on the right side of the blots: Cath30 and Cath20, 30-and 20-kDa polypeptide chains of cathepsin A; SIAL, sialidase. L363P) the majority of the mutant protein was not sorted to the lysosomes, suggesting that these mutations can cause general folding defects and retention of the mutant in the pre-lysosomal compartments. All other expressed sialidase mutants were targeted to lysosomes and correctly processed.
Subsequent experiments revealed that the sialidase mutants F260Y, L270F, and A298V containing amino acid substitutions in the surface patch of the fifth ␤-sheet have similar properties. First, they had very low or absent sialidase activity. Second, stability of sialidase mutants in cellular homogenates or their half-life in the cell as estimated by metabolic labeling was significantly lower than that of the wild-type enzyme. In addition, previous analysis of COS-7 cells transfected with F260Y, L270F, and A298V mutants by Western blot (31) demonstrated the presence of 37-, 26-, and 24-kDa fragments of sialidase protein similar to those observed in COS-7 cells transfected with wild-type sialidase cDNA in the absence of human cathepsin A. The same pattern of sialidase degradation products was also observed in the cells of a galactosialidosis patient that lacked functional cathepsin A (20). Metabolic labeling studies (20) also demonstrated the dramatically reduced half-life of wild-type sialidase expressed in galactosialidosis cells (30 min versus 2.7 h in normal cells) similar to that observed in this work. Together these data suggest that F260Y, L270F, and A298V mutants are not protected by cathepsin A, although the same high amount of functional human cathepsin A was expressed by COS-7 cells in all cases. Indeed, in the extracts of cells transfected with F260Y, L270F, and A298V mutants we could not detect a high molecular weight complex of sialidase with cathepsin A. Instead we observed that sialidase protein sedimented during the density gradient centrifugation together with low molecular weight marker BSA. Therefore in the case of F260Y, L270F, and A298V mutations, the deficit of sialidase activity resulted from the disruption of normal protein-protein interactions in the lysosome. Further studies should show if this mechanism is unique for sialidase or extends to other lysosomal enzymes.