Interaction of Arylsulfatase A with UDP-N-Acetylglucosamine:Lysosomal Enzyme-N-Acetylglucosamine-1-phosphotransferase*

The critical step in lysosomal targeting of soluble lysosomal enzymes is the recognition by an UDP-N-acetylglucosamine:lysosomal enzyme-N-acetylglucosamine-1-phosphotransferase. The structure of the determinant common to all lysosomal enzymes for proper recognition by the phosphotransferase is not completely understood. Our current knowledge is largely based on the introduction of targeted amino acid substitutions into lysosomal enzymes and analysis of their effects on phosphotransferase recognition. We have investigated the effect of eight anti-arylsulfatase A monoclonal antibodies on the interaction of arylsulfatase A with the lysosomal enzyme phosphotransferase in vitro. We also show that a lysine-rich surface area of arylsulfatases A and B is essential for proper recognition by the phosphotransferase. Monoclonal antibodies bind to at least six different epitopes at different locations on the surface of arylsulfatase A. All antibodies bind outside the lysine-rich recognition area, but nevertheless Fab fragments of these antibodies prevent interaction of arylsulfatase A with the phosphotransferase. Our data support a model in which binding of arylsulfatase A to the phosphotransferase is not restricted to a limited surface area but involves the simultaneous recognition of large parts of arylsulfatase A.

The correct lysosomal targeting of soluble lysosomal enzymes mainly depends on the synthesis of mannose 6-phosphate residues on their N-linked oligosaccharide side chains. These mannose 6-phosphate residues are lysosomal targeting signals. In the late Golgi compartments, lysosomal enzymes bind to mannose 6-phosphate receptors, which mediate the vesicular transport to the lysosomes (for review, see Refs. 1 and 2). The synthesis of mannose 6-phosphate residues is a twostep process: in the first reaction the UDP-N-acetylglucosamine:lysosomal enzyme N-acetylglucosamine-1-phosphotransferase (EC 2.7.8.17) (briefly called phosphotransferase 1 ) transfers N-acetylglucosamine-1-phosphate from UDP-Nacetylglucosamine to mannose residues of high mannose-type oligosaccharide side chains of lysosomal enzymes yielding N-acetylglucosamine-1-phospho-6-mannose residues. In a second reaction the N-acetylglucosamine is removed by a phosphodiesterase generating mannose 6-phosphate residues on the oligosaccharide side chains (1). The phosphotransferase has been purified to homogeneity (3). It is a large enzyme of two disulfide linked homodimers of 166 and 51 kDa and two noncovalently linked 56-kDa subunits. The holoenzyme has a molecular mass of 540 kDa (4).
The specific recognition of soluble lysosomal enzymes by the phosphotransferase is the critical step for proper intracellular targeting of these enzymes. Thus, the phosphotransferase must be able to specifically recognize lysosomal enzymes, but since these represent a heterogeneous family of proteins, it must also be able to interact with structurally very diverse enzymes. No obvious sequence similarities suggesting a common homologous targeting signal can be detected among lysosomal enzymes. Earlier studies demonstrated that the specific recognition of lysosomal enzymes by the phosphotransferase depends on the native conformation of lysosomal enzymes (5). In recent years, various studies tried to characterize the recognition determinants of lysosomal enzymes employing in vitro mutagenesis strategies. In the COOH-terminal lobe of lysosomal protease cathepsin D, two regions have been delineated to be a minimal determinant for recognition by the lysosomal enzyme phosphotransferase, a lysine in position 203 and a stretch of 32 amino acid close to the COOH terminus (6). Crystallographic structure determinations revealed that those critical residues are in close proximity on the surface of cathepsin D (7). However, further studies showed that regions in the NH 2 -terminal lobe of cathepsin D also contribute to efficient recognition by the phosphotransferase (8 -10). These studies demonstrated that various distantly spaced residues may act cooperatively to yield the fully effective sorting signal. Similar conclusions have been reached for aspartylglucosaminidase, in which three lysines distantly spaced on the enzyme surface contribute to the proper recognition by the phosphotransferase (11), and for cathepsin D and L in each of which two lysines approximately 34 Å apart are critical recognition determinants (12).
A number of studies on cathepsin D (6, 8 -10, 12), cathepsin L (12,13), DNase I (14), and aspartylglucosaminidase (11) indicated that recognition of lysosomal enzymes depends on lysines as critical residues. Besides the identification of lysines as a common feature of the lysosomal enzyme recognition determinant, its overall structure remains unknown. Comparisons of three-dimensional structures of recently crystallized lysosomal enzymes suggested that a loose consensus sequence in a ␤-hairpin structure may be common to several lysosomal enzymes (15). Whether this sequence is indeed part of the recognition determinant common to all lysosomal enzymes remains to be proven.
Arylsulfatases A (EC 3.1.6.8) and arylsulfatase B (EC 3.1.6.12) are lysosomal enzymes involved in the degradation of sulfated glycolipids and glycosaminoglycans, respectively. Arylsulfatase A is synthesized as a 62-kDa polypeptide, which bears three high mannose-type oligosaccharide side chains, two of which are phosphorylated by the phosphotransferase (16,17). The position of the recognition determinant in lysosomal arylsulfatase A and homologous sulfatases is unknown. Since our current knowledge on the requirements of the interaction of the lysosomal enzymes with the phosphotransferase is based on the introduction of targeted amino acid substitutions we have chosen to characterize phosphotransferase recognition determinant on arylsulfatase A with monoclonal antibodies. Recently, we reported the generation of four monoclonal antibodies specific for human arylsulfatase A (18). Fab fragments of these antibodies inhibited the synthesis of mannose 6-phosphate residues on arylsulfatase A in an in vitro phosphorylation system. Epitope mapping revealed that these antibodies recognized identical or adjacent epitopes on the enzyme, and it was assumed that the epitope was part of or close to the phosphotransferase recognition determinant. In order to extend this study and to map the recognition determinant more precisely, we have generated eight additional antibodies and examined their ability to interfere with mannose phosphorylation.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media RPMI 1640 for hybridomas and Dulbecco's modified Eagle's medium for all other cell types were from Life Technologies, Inc. Fetal calf serum was obtained from Biochrom. DNA digesting and DNA modifying enzymes were from New England Biolabs or Life Technologies, Inc. [␥-32 P]ATP (3000 Ci/mmol) was from Amersham. Enzymes used in the synthesis of [␤-32 P]UDP-N-acetylglucosamine were purchased from Sigma. All other chemicals were of analytical or molecular biology grade and were from Merck, Serva, Sigma, or Fluka. MAbs 5C6 (19) and NQ2/ 16-2 (20) were used as controls and were provided by Dr. Lemke, Kiel, Germany and Dr. Gordon, Oxford, United Kingdom.
Production of Monoclonal Antibodies-Female BALB/c mice 6 weeks of age were injected three times (day 1, 14, and 28) intraperitoneally with 100 g of arylsulfatase A. Recombinant enzyme was purified by immunoaffinity purification as described earlier (18). Prior to injection the enzyme was adsorbed to aluminum hydroxide. Anti-arylsulfatase A titers were monitored 12 days after the last immunization. Animals scoring positive were boosted with 50 g of arylsulfatase A and sacrificed 3 days later. Spleen cells of mice were fused with X63-Ag8.653 myeloma cells. Fusion was done as described (21). After fusion, cells were plated in 200 wells of 24-well hybridoma plates (Greiner) in RPMI 1640 supplemented with 10% fetal calf serum, penicillin, streptomycine, 100 M hypoxanthine, 0.4 M aminopterin, 16 M thymidine (Boehringer-Mannheim). The medium was further supplemented with 10% conditioned medium from the J774 cell line. Wells containing anti-arylsulfatase A immunoglobuline secreting hybridomas were identified via a radioimmunoassay in which an iodinated secondary antimouse light chain antibody raised in rabbits was used (Dakopatts, Copenhagen). Clones in positive cells were subcloned and reanalyzed by radioimmunoassay. Immunoglobulin subclass determination was performed with a kit from Boehringer-Mannheim. MAbs A3, C, E, and F are IgG 1 ; mAbs A1, A2, and A4 are IgG 2a ; and A5 is IgG 2b . All antibodies have -light chains.
In part of the experiments three previously published mAbs have been used (18). To facilitate reading of this article these antibodies have been renamed, 19C2 is B1, 20D2 is B2, and 11B5 is D.
Purification of Monoclonal Antibodies and Fab Fragments-Hybridomas were cultured in the Hereaus Miniperm system, which in all cases yielded several milligrams of mAbs. IgG 2 mAbs were purified via affinity chromatography on Protein A-Sepharose and IgG 1 on Protein G-Sepharose (Pharmacia). Affinity chromatography was performed according to the manufacturers protocols. For production of Fab frag-ments, purified antibodies were adjusted to a concentration of 1-5 mg in phosphate-buffered saline, 5 mM cysteine. Papain was added at a molar ratio of 1:40 mol of IgG. Depending on the antibody, proteolysis was performed at 37°C for 1 to 4 h. Reaction was terminated by the addition of iodoacetamide to a final concentration of 20 mM. Fab fragments were purified by gel filtration on a Superdex G-75 column (Pharmacia) equilibrated with phosphate-buffered saline.
Immunoprecipitation, SDS-PAGE, and Western Blots-Immunoprecipitation of unlabeled arylsulfatase A under nondenaturing conditions was performed in 50 mM Tris/HCl, pH 7.4, 0.15 M NaCl, 0.05% Triton X-100. Five to 10 milliunits of arylsulfatase A were incubated with 10 g of antibodies for 12-16 h. To enhance precipitation, 10 g of goat anti-mouse affinity purified antibodies (Dianova) were added for another 2 h before Ig were collected by the addition of 60 l of Staphylococcus aureus suspension (Pansorbin, Calbiochem). After 30 min of continuous shaking, bacteria were pelleted for 4 min at 10,000 ϫ g. Pellets were resuspended in immunoprecipitation buffer and washed once. Arylsulfatase A activity was determined in supernatants and pellets.
Monoclonal Antibody Competition Assay-96-well enzyme-linked immunosorbent assay plates were coated with 125 ng of mAbs and subsequently blocked with an excess of casein. Simultaneously, arylsulfatase A was incubated with a 10-fold molar excess of mAb in 20 mM Tris/HCl, pH 7.4, 0.15 M NaCl. MAb-arylsulfatase A complexes generated in liquid phase were then added to the mAb-coated wells and allowed to bind for 2 h at 37°C. Afterward liquid phase was removed, the wells were washed twice, and arylsulfatase A bound to the wells was quantified by activity determination (18). If the antibody coupled to the well and complexed to the arylsulfatase A in the liquid phase compete and thus cannot bind arylsulfatase A simultaneously, the enzyme cannot be retained in the wells.
Detection of Octamerization of Arylsulfatase A-96-well enzymelinked immunosorbent assay plates were coated with 100 ng of arylsulfatase A/well for 16 h at 4°C. Purified arylsulfatase A can be labeled via incubation with [ 35 S]cysteine. 2 35 S-Labeled arylsulfatase A was incubated with a 10-fold molar excess of mAbs. The antibody complexed with [ 35 S]arylsulfatase A was added to the wells at neutral pH when octamerization of arylsulfatase A does not occur. Then pH of the wells was titrated to pH 4.5 and incubation continued for 1 h. Low pH favors octamerization of arylsulfatase A so that arylsulfatase A-antibody complexes should bind to solid phase arylsulfatase A if the antibody does not interfere with octamerization. Liquid phase was removed and wells were washed with 50 mM sodium acetate buffer, pH 4.5. Radioactivity bound to the plates was determined. As a control a duplicate plate was washed with neutral buffer, which reverses association of [ 35 S]arylsulfatase A with arylsulfatase A bound to the wells.
In Vitro Phosphorylation of Arylsulfatase A-Preparation of Golgi membranes, [␤-32 P]UDP-GlcNAc and arylsulfatase A by immunoaffinity chromatography has been described in detail (18,22). Purified arylsulfatase B was kindly provided by M. Evers and C. Peters, Freiburg, Germany. Quantification of radioactivity after SDS-PAGE was done with a Fuji Bioimager.
Chemical Modification of Amino and Carboxyl Residues-To modify carboxyl residues 50 g of arylsulfatase A or arylsulfatase B were dialyzed against 20 mM sodium acetate, pH 5.2, and samples were divided into two parts of 50 l each. One part received no addition and served as a control, the other part was supplemented with 20 mM 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydroxychloride (EDC) and 100 mM ethylendiamide. After incubation for 20 min at 30°C, the reaction was stopped by increasing sodium acetate concentration to 0.66 M. Treated and untreated samples were subsequently dialyzed against Tris-buffered saline.
To modify amino groups of lysines, 50 g of arylsulfatase A or arylsulfatase B were dialyzed against phosphate-buffered saline, pH 8.0, and samples were split into two aliquots. One aliquot received no additions and to the other 17.5 mM sulfo-N-hydroxy-succinimidyl acetate (SNA) was added. Incubation was for 90 min on ice and reaction was terminated by dialysis against Tris-buffered saline. In each case, the success of the modification reaction was monitored by alterations in isoelectric focussing pattern.

Characterization of Anti-arylsulfatase A Monoclonal
Antibodies-Mice were immunized with recombinant human arylsulfatase A and a total of 8 stable hybridomas secreting anti-arylsulfatase A antibodies were obtained. These antibodies precipitated arylsulfatase A under nondenaturing conditions from extracts of LtkϪ cells overexpressing the human arylsulfatase A cDNA. Since the sum of enzyme activity remaining in the supernatant (in all cases less than 10%) and bound to the mAbs in the pellet was in the same range as the activity in the supernatant when a irrelevant control antibody was used, it can be concluded that the antibodies do not interfere with arylsulfatase A activity (data not shown). All mAbs recognize human arylsulfatase A specifically, they do not precipitate rat or murine arylsulfatase A (less than 10% in all cases). They do not cross-react with arylsulfatase B an enzyme highly homologous to arylsulfatase A. None of the mAbs recognizes arylsulfatase A under denaturing conditions as revealed by Western blot analysis (data not shown).
Anti-arylsulfatase A Monoclonal Antibodies Inhibit in Vitro Phosphorylation of Arylsulfatase A-The synthesis of mannose 6-phosphate residues, which represent the specific lysosomal targeting signal can be performed in vitro (18,22). Purified lysosomal enzymes are incubated with a Golgi fraction as a source of lysosomal enzyme phosphotransferase and [␤-32 P]UDP-GlcNAc from which the radioactively labeled 32 P is transferred to the oligosaccharide side chains of the substrate enzymes. Subsequently, phosphorylation is detected by SDS-PAGE analysis and autoradiography. Fig. 1 shows an experiment in which recombinant arylsulfatase A purified to apparent homogeneity was phosphorylated in vitro in the absence or presence of control or anti-arylsulfatase A mAbs. Addition of an irrelevant control mAb has no effect on arylsulfatase A phosphorylation, whereas all mAbs (top panel) or their Fab fragments (bottom panel) inhibited phosphorylation almost completely. Addition of anti-arylsulfatase A mAbs or Fab fragments did not inhibit phosphorylation of arylsulfatase B, which was co-phosphorylated in the same sample, ex-cluding nonspecific inhibition of phosphorylation by the addition of anti-arylsulfatase A mAbs.
Inhibition of arylsulfatase A phosphorylation may be explained in two ways. 1) The antibodies inhibit recognition of arylsulfatase A by the phosphotransferase, because they mask the determinant responsible for lysosomal enzyme recognition. In this case the arylsulfatase A-mAb complexes will not be able to interact with the phosphotransferase. 2) The antibodies may still allow arylsulfatase A to be recognized by the phosphotransferase as a lysosomal enzyme, but after binding, phosphorylation of oligosaccharides is sterically inhibited by the antiarylsulfatase A mAbs. To differentiate between these possibilities, we developed an assay in which arylsulfatase A inhibits the phosphorylation of arylsulfatase B competitively. If arylsulfatase A-mAb complexes can still bind to the phosphotransferase, they should be able to compete arylsulfatase B phosphorylation. If the mAbs prevent arylsulfatase A from interacting with the phosphotransferase, arylsulfatase A-mAb complexes should not compete arylsulfatase B phosphorylation. Fig. 2 shows results of three independent phosphorylation experiments. In these experiments arylsulfatase B and arylsulfatase A were co-phosphorylated in the same reaction. Arylsulfatase A was added in about 10-fold molar excess over arylsulfatase B. Addition of arylsulfatase A inhibits arylsulfatase B phosphorylation by about 40%. In all cases addition of antiarylsulfatase A Fab fragments abolishes the inhibitory effect of arylsulfatase A on arylsulfatase B phosphorylation. Thus, Fab fragments of all mAbs prevent interaction of arylsulfatase A with the phosphotransferase.
Lysines Are Important for Recognition of Arylsulfatase A and Arylsulfatase B by the Phosphotransferase-The importance of lysines for lysosomal enzyme recognition has been documented for several lysosomal enzymes (6, 10 -12). In order to investigate the role of lysines in sulfatases, purified arylsulfatase A and arylsulfatase B were treated with EDC and SNA to modify glutamate/aspartate carboxylic groups and amino groups of lysines, respectively (Fig. 3). EDC in combination with ethylenediamine converts carboxylic groups of glutamate and aspartate to the corresponding amides, SNA causes acetylation of amino groups of lysines.
Modification of carboxyl groups affects arylsulfatase A and arylsulfatase B in vitro phosphorylation only slightly causing a reduction of phosphorylation by about 30 -40% (see Fig. 3). In contrast, lysine modification inhibited phosphorylation of arylsulfatase A and arylsulfatase B completely. In case of modification of carboxylic groups, the enzyme activity of arylsulfatase A was lost, whereas acetylation of lysines left enzyme activity unchanged (data not shown). This excludes that an unspecific denaturing effect of the modifications accounts for the loss of phosphorylation. Thus, lysines are essential for the recognition of arylsulfatase A and arylsulfatase B by the lysosomal enzyme phosphotransferase.
Effect of the Monoclonal Antibodies on pH-dependent Octamerization of Arylsulfatase A-The fact that all mAbs inhibited the interaction of arylsulfatase A with the lysosomal enzyme phosphotransferase raises the question of specificity. The mAbs may interfere unspecifically with any protein-protein interactions of arylsulfatase A simply by size. To get a measure of the specificity of mAb inhibition of phosphorylation we have investigated the influence of the mAbs on another proteinprotein interaction of arylsulfatase A that can be performed in vitro. This is the pH-dependent octamerization of arylsulfatase A. At neutral pH, arylsulfatase A exists as a dimer, which upon acidification associates to form octamers (23). To examine whether the mAbs interfere with octamerization, 100 ng of purified arylsulfatase A was adsorbed to a 96-well microtiter plate. 35 S-Labeled arylsulfatase A was added at neutral pH to the wells. Afterward, pH in the wells was titrated to pH 4.5 under continuous shaking. At low pH, some of the fluid phase 35 S-labeled arylsulfatase A should associate with arylsulfatase A bound to the solid phase via interaction of the octamerization domain. If wells are washed at pH 4.5, labeled arylsulfatase A remains bound to the solid phase (bars Co pH 4.5 in Fig. 4) whereas washing at neutral pH reverses octamerization and removes most of the bound [ 35 S]arylsulfatase A (bars Co pH 7.4 in Fig. 4). To investigate whether mAbs interfere with this pH-dependent octamerization process, mAbs were added either to the wells to bind to solid phase arylsulfatase A before labeled fluid phase [ 35 S]arylsulfatase A was added (light bars in Fig. 4) or mAbs were preincubated with fluid phase [ 35 S]arylsulfatase A before addition to wells (dark bars, Fig. 4). After titration to pH 4.5 and shaking for 1 h, wells were washed at pH 4.5 and counts remaining in the wells were determined. Compared with control mAb, none of the arylsulfatase A-specific mAbs inhibited binding of fluid phase [ 35 S]arylsulfatase A to solid phase enzyme, irrespective whether mAbs were incubated with the fluid phase [ 35 S]arylsulfatase A or unlabeled arylsulfatase A bound to the wells. Radioactivity in all wells could be removed by subsequent washing at neutral pH, which reverses octamerization (data not shown). Control experiments showing that arylsulfatase A-mAb complexes are stable at pH 4.5 have been performed (data not shown).
Simultaneous Binding of Monoclonal Antibodies to Arylsulfatase A-To evaluate how many different epitopes are recognized by the mAbs, we investigated which of the mAbs can bind simultaneously to arylsulfatase A. Wells of a 96-well microtiter plate were precoated with the eight anti-arylsulfatase A mAbs and with a control antibody. Arylsulfatase A was preincubated with a 10-fold molar excess of antibodies for 16 h and then added to the wells containing the immobilized mAbs. In case S-Labeled arylsulfatase A was added to the wells after preincubation with a 10-fold molar excess of antibody indicated at the bottom (dark bars). In a different experiment, antibodies were added to the arylsulfatase A containing wells and 35 S-labeled arylsulfatase A was added after the unbound antibodies have been removed (light bars). After addition of 35 S-labeled arylsulfatase A the pH in the wells was titrated to 4.5 and plates were incubated for 1 h. Plates were washed twice at pH 4.5 to remove unbound arylsulfatase A. Bars represent the radioactivity remaining in the wells. Controls pH 4.5 and pH 7.4 were done without the addition of antibody. Control pH 7.4 was washed at pH 7.4. At this pH octamerization does not occur, and radioactive arylsulfatase A is removed. This represents the background value. Control pH 4.5 was washed at acidic pH and represents the 100% value for octamerization. Absolute values for control pH 7.4 are ϳ200 and ϳ800 cpm for control pH 4.5.
arylsulfatase A-bound mAbs compete binding of the immobilized mAbs arylsulfatase A should not bind to the immobilized mAbs in the wells. After 3 h incubation of arylsulfatase A-mAb complexes in the wells, wells were washed and arylsulfatase A remaining in the wells was quantified by activity determination. Results are shown in Table I. Antibodies A1, A2, A3, A4, and A5 compete when binding to arylsulfatase A, which allows the conclusion that they recognize the same or adjacent epitopes. Antibody C partially competes binding of E, but not of F, and E also competes partially with F. It seems that these antibodies recognize different epitopes. We have performed similar experiments with two mAbs (B1 and B2) we have used in previously published experiments (18). (For alterations in mAb designations see "Experimental procedures.") These antibodies compete with the first group of antibodies, but not with C, E, or F. This demonstrates that the mAbs recognize at least four different epitopes and that at least two mAbs can bind simultaneously to arylsulfatase A.
Epitopes of Anti-arylsulfatase A Monoclonal Antibodies-To localize epitopes, we immunoprecipitated human/murine arylsulfatase A chimeric enzymes, compared mouse (24), rat and human cDNA predicted amino acid sequences (25), and used information available from crystallographic data of arylsulfatase A (23).
We have constructed chimeric arylsulfatase A enzymes, by fusing murine and human arylsulfatase A cDNA sequences. These chimeric cDNAs were expressed in baby hamster kidney cells and arylsulfatase A was immunoprecipitated from the extracts. Arylsulfatase A activity was measured in the supernatants and pellets after immunoprecipitation. Since the mAbs do not recognize the murine enzyme, this approach allows to map the regions to which the mAbs bind. Table II summarizes the results, according to which the mAbs can be divided into four groups based on the different patterns of chimeric arylsulfatase A immunoprecipitation. Since the murine and rat enzyme are not recognized by the antibodies, comparison of rat and murine cDNA predicted amino acid sequences with those of the human cDNA allowed to locate epitopes further to differences in the predicted amino acid sequences. Since the mAbs recognize the native enzyme only, they should react with amino acid residues present on the surface of the enzyme. Surface located residues can be inferred from the three-dimensional structure of the arylsulfatase A enzyme, which we have determined recently (23). Data from the chimeric enzymes, amino acid sequence comparisons, and the criterion of surface location allow the conclusion that mAbs A1, A2, A3, A4, and A5 bind to an epitope between amino acids 18 and 97. MAb C binds to amino acids 165-169, 202-205, 216, or 239 and 240. Epitopes of a mAbs E and F can be less precisely defined, because these mAbs bind to regions in which the number of amino acids, which are different among mouse/rat and human arylsulfatase A, is much higher than in regions to which the other mAbs map. E binds between amino acids 287 and 459 and at least part of the epitope of F is located between amino acids 463 and 507. MAbs D, B1, and B2 have been described previously (18). The latter two show the same chimeric enzyme immunoprecipitation pattern as C, whereas the epitope of D has been mapped to amino acids 256 to 260 (see Table III for summary). The amino acids at which the murine and human cDNA sequences were fused are indicated by position in the one-letter code. The right part lists the mAbs and a polyclonal arylsulfatase A rabbit antiserum (pAS) and their reactivity with the chimeric enzymes. ϩ indicates that more than 90% of enzyme was precipitable, Ϫ that less than 10% of the enzyme was precipitable, and (ϩ) indicates a 50% reduction in immunoprecipitation. Part of the data on antibodies D, B1, and B2 shown at the bottom have been published previously (18).

TABLE I
Competition assay of monoclonal antibody binding to human arylsulfatase A Wells of a 96-well microtiter plate were coated with 125 ng each of the mAbs listed in the left vertical lane. Arylsulfatase A was incubated with a 10-fold molar excess of the mAbs listed in the top horizontal lane. Arylsulfatase A-mAb complexes were then added to the wells and incubated for 2 h. Wells were washed twice and the arylsulfatase A remaining in the wells was quantified by activity determination. 100% is the value of wells to which only arylsulfatase A without prior incubation with antibodies has been added. The residual amount of arylsulfatase A bound when the antibody used for coating the wells and preincubation with arylsulfatase A was identical was substracted as background value. This value was less than 20% for all antibodies. Co designates control antibody. We have generated eight anti-human arylsulfatase A mAbs to study the interaction of this lysosomal enzyme with the lysosomal enzyme phosphotransferase.
Results are based on an in vitro system in which mannose 6-phosphate residues as specific lysosomal targeting signals are synthesized in vitro on oligosaccharide side chains of recombinant arylsulfatase A. All 8 anti-arylsulfatase A mAbs or their Fab fragments, respectively, were able to inhibit in vitro phosphorylation of arylsulfatase A completely. This inhibition may have three different explanations. If all mAbs bind close to the lysosomal enzyme recognition determinant, it can be expected that they prevent interaction of arylsulfatase A with the phosphotransferase. Alternatively, the interacting surface of arylsulfatase A with phosphotransferase may be so large that any antibody will interfere with recognition irrespective of the distance of its epitope to the recognition determinant. Third, Fab fragments may bind to areas outside the recognition site, such that arylsulfatase A-Fab fragment complexes may still be recognized by the phosphotransferase, but due to sterical hindrance phosphate cannot be transferred to the oligosaccharide side chains.
The latter possibility has been excluded by competition experiments. Arylsulfatase A-Fab fragment complexes do not inhibit arylsulfatase B phosphorylation. This allows the conclusion that the Fab fragment complexed arylsulfatase A does not interact with the phosphotransferase. If the effect of the Fab fragments would be limited to sterical hindrance of oligosaccharide phosphorylation, binding to the phosphotransferase via the recognition determinant should have been possible and arylsulfatase A-Fab fragment complexes should inhibit arylsulfatase B phosphorylation.
To distinguish between the remaining two possibilities, we attempted to locate the epitopes more precisely. Data from immunoprecipitation experiments of human/murine chimeric arylsulfatase A and mAb competition experiments suggest that the mAbs, described here and previously (18), recognize at least 6 different epitopes, all of which depend on the native status of the enzyme (see Tables II and III). Table III summarizes the location of the epitopes of anti-arylsulfatase A mAbs and lists possible amino acids being part of these epitopes. Data from these experiments are consistent with those from competition experiments, in which different mAbs were tested for their capability to bind arylsulfatase A simultaneously. MAbs mapping to the same region by chimera immunoprecipitation also compete when binding to arylsulfatase A. Vice versa, mAbs C, E, and F, which map to different parts than A1 to A5 do not compete binding of these mAbs. Similarly, the different epitope mapping pattern of mAbs C, E, and F as revealed by immunoprecipitation of chimeric enzyme is also reflected in competition experiments. C and F do not compete, but at least partial competition can be seen between C/E and E/F. On the other hand, competition experiments also show that antibodies mapping to the same regions according to chimeric arylsulfatase A immunoprecipitation must recognize different epitopes. MAbs B1, B2, and C show the same reactivity pattern toward chimeric arylsulfatase A enzymes (Table II and 3), but B1 and B2 do not compete binding of C proving recognition of different epitopes.
The combination of all data allows location of likely positions of epitopes in the three-dimensional model of arylsulfatase A as presented in Fig. 5. Amino acids determining the epitope of mAbs A1 to A5 (see Table III) are located on one side of the enzyme. Amino acids of B1/B2 and C are found on top and on the opposite side of arylsulfatase A. Antibodies B1 and B2 do not compete with C, although chimeric immunoprecipitation locates them to the same amino acid residues. This is understandable in view of the three-dimensional model: the residues determining the epitopes of these mAbs are found in three distant locations so that simultaneous binding of these antibodies is conceivable. Epitope of D has been determined previously (18) and is found on the same side as those of B1, B2, and C. Epitopes of mAbs E and F can be less precisely defined. However, their epitopes must be located in some distance to those of the other antibodies, because they do not, or in the case of mAb E only partially, compete binding of these antibodies to arylsulfatase A. Thus, as illustrated in Fig. 5 epitopes of mAbs are found on different parts of the surface of arylsulfatase A. Therefore, the inhibition of phosphorylation cannot be explained by binding of all antibodies to a restricted area on the surface of arylsulfatase A.
Modification of lysines specifically inhibits the recognition of arylsulfatase A and arylsulfatase B by the phosphotransferase. In arylsulfatase A only five lysines are clustered on the surface of enzyme (23). These five lysines are not evenly distributed on the enzyme surface, but are located in a single cluster (see Fig.  5). Interestingly, this lysine cluster is in close proximity and partly overlaps the octamerization determinant of arylsulfatase A. The mAbs (except possibly for E and F) have epitopes that are distant from this region, which is functionally supported by the fact that none of the mAbs interferes with octamerization. Thus, there is no evidence that any of the mAbs or Fab fragments bind close to the lysine containing lysosomal enzyme recognition determinant in arylsulfatase A, yet all are able to inhibit binding of arylsulfatase A to the phosphotransferase. The fact that the mAbs do not inhibit octamerization, but phosphorylation suggests that the area interacting with the phosphotransferase is larger than the octamerization area, which covers about 900 Å (23) and thus larger than the lysine cluster necessary for proper recognition.
One may argue that the size relation of arylsulfatase A to Fab fragments (62 versus ϳ50 kDa) is unfavorable, so that an  (18). The antibodies can be divided into six groups recognizing different epitopes according to data from immunoprecipitation of arylsulfatase A chimeric enzymes, comparisons of cDNA predicted murine, rat, and human arylsulfatase A amino acid sequences, surface location of the amino acid residues, and data from antibody binding competition experiments shown in Table I (26). This demonstrates that Fab fragments despite of their size can be used to discern protein-protein interactions even of adjacent small amino acid sequences. Cocrystallization of lysozyme and lysozyme-binding Fab fragments revealed that the binding site of the Fab fragment is a flat structure and the fragment extends from the binding site into the surrounding solution in a rather stretched conformation (27). The Fab fragment leaves most of the surface of the antigen accessible. The size of the interacting area of lysozyme and lysozyme Fab-fragments is 600 Å 2 . This is only a fraction of the total arylsulfatase A surface area. Thus, the ability of the mAbs to inhibit interaction with the phosphotransferase cannot be explained by the assumption that single Fab fragments cover large parts of the arylsulfatase A surface making them inaccessible to other molecules. This is supported by our demonstration that none of the mAbs interferes with enzyme activity or octamerization and that at least two mAbs can bind simultaneously to the enzyme.
Our data support a model in which lysosomal enzyme phosphotransferase interacts with large surface areas of arylsulfatase A, a view that can be brought into line with results of targeted amino acid substitutions in cathepsin D (10,12) and aspartylglucosaminidase (11). In cathepsin D two lysine residues 34 Å apart (203 and 293) cooperate to form part of a phosphotransferase recognition determinant. When these residues are mutated 30% of normal phosphorylation still remains (12), demonstrating the involvement of more residues and thus, most likely a larger surface area in the formation of a fully efficient recognition determinant. In aspartylglucosaminidase three distantly located lysine residues contribute to form a recognition signal. In addition, the minimal regions establishing the cathepsin D recognition determinants were identified as lysine 203 and amino acid residues 265-292. When these residues were introduced into the non-lysosomal homologous protease glycopepsinogen, the chimeric molecule was recognized by the phosphotransferase, but to a much lesser extent than cathepsin D (6). Obviously, residues beyond lysine 203 and amino acids 265-292 contribute to efficient expression of the lysosomal enzyme recognition determinant, and it has been shown that such sequences can be found in the amino-terminal lobe of cathepsin D (8,10). A minimal recognition sequence may be necessary for recognition, but its efficiency can be enhanced by additional even distant areas on the same molecule. The cooperativity of various surface areas in the formation of a fully expressed lysosomal enzyme recognition determinant can be explained by different models in which the distant parts of a lysosomal enzyme bind subsequently or simultaneously to phosphotransferase (10). Phosphotransferase may initially bind lysosomal enzymes through a lysine-dependent recognition determinant and starts to phosphorylate accessible oligosaccharide side chains. After completion of this phosphorylation step, the enzyme dissociates and rotates to get access to another oligosaccharide side chain. However, the lysosomal enzyme remains in the vicinity of the phosphotransferase ensuring a high probability of rebinding even at weaker lysine-dependent or independent sites, present at different locations on the lysosomal enzyme surface. This process may continue until several oligosaccharide side chains have been phosphorylated. If this model is correct, it is surprising that none of the arylsulfatase A-Fab fragment complexes inhibits the phosphorylation of arylsulfatase B. Since the lysine region on arylsulfatase A is not covered by any of the Fab fragments, an initial contact through this region should not be disturbed by Fab fragments and arylsulfatase A-Fab fragment complexes should be able to inhibit phosphorylation of arylsulfatase B competitively. Thus, a model which depends on an initial high affinity binding of restricted area and with subsequent rotation and rebinding at weaker sites seems not likely.
Alternatively, phosphotransferase may contact the surface of A and B, surfacelocated lysines are shown in red, residues involved in octamerization in green. Possible epitopes of mAbs A1 to A5 are depicted in blue, those of B1, B2 and C mAbs are yellow, that of D, gray. Asparagines bearing N-linked oligosaccharide side chains are shown in black or pink. They are labeled Ol1, Ol2, and Ol3 indicating oligosaccharide side chains linked to asparagines 158, 184, and 350, respectively. Ol2 (pink) is not phosphorylated by the lysosomal enzyme phosphotransferase (17). B shows a view of the enzyme after vertical rotation by 180°compared with A. C and D, lysine/octamerization area and oligosaccharide labeling as in A and B. Possible epitopes of mAb E appear dark purple, those of F, tan. Epitopes can be less clearly defined because numerous amino acid residues are different among mouse/rat and human arylsulfatase A in the respective regions. D shows a view of the enzyme after vertical rotation by 180°c ompared with C. lysosomal enzyme at multiple sites simultaneously. The initial contact may depend on a lysine containing recognition determinant and subsequently binding is stabilized via additional contacts. Since phosphotransferase is a large multimeric enzyme of 9 times the molecular weight of an arylsulfatase A monomer, such an explanation seems applicable. Since all Fab fragments irrespective of their binding sites prevent interaction of arylsulfatase A with the phosphotransferase, it may be assumed that lysosomal enzymes enter a large active site pocket covering most of lysosomal enzymes surface. However, since lysosomal enzymes have different shapes it seems unlikely that they all fit tightly into a large inflexible active site of phosphotransferase. Thus, after initial contacts are made perhaps by charge-dependent interactions with lysines a conformational change in the phosphotransferase may be induced which causes embracement of the lysosomal enzyme. This embracement is only stabilized when additional contact points are established, which may then occur at distant points on the surface of a lysosomal enzyme. In this case the arylsulfatase A-Fab fragment complexes may still interact with phosphotransferase but Fab fragments may interfere sterically with embracement and establishment of secondary stabilizing contacts. Since arylsulfatase B phosphorylation cannot be inhibited by arylsulfatase A-Fab fragment complexes, in the embracement model the initial lysine dependent contact must be of low affinity. Our data does not allow distinction between the latter models, but in any case support a model in which large areas of the surface of lysosomal enzymes simultaneously interact with the phosphotransferase.