Lysosomal Enzyme Phosphorylation - Semantic Scholar

THEJOURNAL OF BIOLOGICAL CHEMISTRY 1992 by The American Society for Biochemistry ’ and Molecular Biology, Inc

VOl. 267, No. 32, Issue of November 15, PP. 23342-23348, 1992 Printed in U.S.A.

(c?

Lysosomal Enzyme Phosphorylation I. PROTEIN RECOGNITION DETERMINANTSINBOTH LOBES OF PROCATHEPSIN D MEDIATE ITS INTERACTION WITH UDP-G1cNAc:LYSOSOMAL ENZYME N-ACETYLGLUCOSAMINE-1PHOSPHOTRANSFERASE* (Received for publication, June 25, 1992)

Thomas J. Baranski$, Alan B. Cantor$, and Stuart KornfeldQ From the Department of Medicine, Washington University School of Medicine, St. Louis,Missouri 631 10

We have investigated the nature of a protein domain soma1 compartment where the hydrolases are subsequently that is shared among lysosomal hydrolases andis rec- packaged into lysosomes. Phosphomannosyl residues are synognized by UDP-G1cNAc:lysosomal enzyme N-acetyl- thesized by the concerted action of two enzymes (1, 2). First, glucosamine- 1-phosphotransferase, the initial enzymeUDP-G1cNAc:lysosomal enzyme N-acetylglucosaminylphosin the biosynthesis of mannose 6-phosphate residues. photransferase (phosphotransferase) selectively transfers N Previously, elements of this recognition domain were acetylglucosamine 1-phosphate to mannose residues on lysoidentified using a chimeric protein approach. comThe somal enzymes to form phosphodiester intermediates. Then, bined substitution of two regions (amino acids 188- the N-acetylglucosamine residues are removed by N-acetyl230, particularly lysine 203, and 265-292) from the to carboxyl lobe of the lysosomal hydrolase cathepsin D glucosamine-l-phosphodiester-cu-N-acetylglucosaminidase into the homologous positions of the related secretory produce Man-6-P monoesters. The efficient targeting of acid protein glycopepsinogen was sufficient to confer rec- hydrolases to lysosomes is dependent on the ability of the ognition by phosphotransferase and subsequent phos- first enzyme, phosphotransferase, to recognize and bind with phorylation of the oligosaccharides whenthis chimeric high affinity to a protein determinant that is common to protein was expressed inXenopus oocytes. (Baranski, lysosomal enzymes but absent from non-lysosomal glycoproT. J., Faust, P. L., and Kornfeld, S. (1990) Cell 63, teins. The catalytic site of phosphotransferase then specifi281-291). The current study demonstrates thatwhen cally phosphorylates the high mannose oligosaccharides of these two regions are replaced in cathepsin D by the the lysosomal enzymes (3-5). Cathepsin D is an aspartyl protease that is phosphorylated homologous glycopepsinogen amino acids,the resultant chimeric molecule is poorly phosphorylated. However, and targeted to lysosomes in several systems (2, 6, 7), includwhen eitherof these regionsis substituted individually, ing Xenopus oocytes (8). Pepsinogen is a well-studied secrethe chimeric molecules are well phosphorylated. The tory aspartyl protease that shares 45% identity in amino acid is sequence with cathepsin D. A glycosylated form of human phosphorylation of these latter chimeric proteins dependent on the presence of procathepsin D amino pepsinogen that contains two N-linked glycosylation sites lobe elements. By analyzing a series of chimeric pro- mutated into positions homologous to thetwo sites in cathepteins that contain all eightcombinations of three con- sin D was secreted and not detectably phosphorylated when secutive segments of the entire amino lobe of procathepsin D, it was found that multiple regions of the expressed in Xenopus oocytes (9). Cathepsin D and pepsinoamino lobe of cathepsin D enhance phosphorylation of gen are predicted to have very similar structures since the of the bilobed asparthe chimeric proteins. These elementsmay be part of overall secondary and tertiary structures an extended carboxyllobe recognition domain or com- tyl proteases are remarkably conserved in the fungal and mammalian enzymes that have been crystallized (10). prise a second independent recognition domain. To identify the protein determinants of cathepsin D that comprise the phosphotransferase recognition domain, we previously generated cDNA constructs that encode a number of In many cell types, lysosomal hydrolases acquire phos- chimeric proteins in which regions of human cathepsin Dare phomannosyl residues that mediate their binding to mannose substituted into the backbone of human glycopepsinogen. 6-phosphate (Man-6-P)’ receptors and targeting to an endo- When these chimeric molecules were expressed in Xenopus oocytes and analyzed for phosphorylation, we found that two * This work was supported in partby United States Public Health noncontinuous primary sequences from the carboxyl lobe of Service Grant CA08759. The costs of publication of this article were cathepsin D (amino acids 188-230, particularly lysine 203, defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 and 265-292)’ were sufficient to generate an efficient phosU.S.C. Section 1734 solely to indicate this fact. photransferase recognition domain in glycopepsinogen (9). $ Supported by National Institutes of Health Research Award GM- When localized to the homologous position in the crystal 07200. Medical Scientist from the National Institutes of General structure of porcine pepsinogen, these two sequences were Medical Sciences. To whom correspondence should be addressed Div. of Hematol- found to be in directapposition on the surface of the molecule. ogy-Oncology, Washington University School of Medicine, 660 S. Moreover, this recognition domain on the carboxyl lobe of cathepsin D is sufficient to direct not only the phosphorylaEuclid Ave., Box 8125, St. Louis, MO 63110. The abbreviations used are: Man-6-P, mannose 6-phosphate; CP, tion of the oligosaccharide on the carboxyl lobe, but also the chimericprotein; CI-MPR,cation-independent mannose6-phosphosphorylation of the oligosaccharide on the amino lobe of phate/IGF-I1 receptor; phosphotransferase, UDP-G1cNAc:lysosomal the molecule, although to a lesser extent (11).However, the enzyme N-acetylglucosaminylphosphotransferase; SDS-PAGE, SOdium dodecyl sulfate-polyacrylamide gel electrophoresis; endo H; endo-8-N-acetylglucosaminidaseH.

23342

Human cathepsin D numbering is used throughout (15).

Lysosomal Enzyme Phosphorylation

23343

jected (-10 ng/oocyte) into stage V-VI Xenopusoocytes(50-60 oocytes/construct determination)as previously described (8). For labeling with [35S]methionine/cysteine, oocyteswere incubated in groups of five to eight at 19 “C for 72 h (note that incubation past 20 h constitutes chase a in this system (17))with 10 pl/oocyte of modified Barth’s saline containing 5% fetal bovine serum and 1 mCi/ml [35S] methionine/cysteine. At the termination of the incubation, oocytes and media were collected, rapidly frozen on dry ice, and stored at -20 “C. [3H]Mannose-labeling of the expressed proteins was performed as described elsewhere (11).Briefly, 12 pCi of GDP-[3,4-3H] mannose were dried under a stream of nitrogen gas a t 4 “C, resuspended in 8 r l of the appropriate mRNA (-200 ng/pl), and microinjected into 120 Xenopus oocytes/construct. After 72 h of incubation at 19 “C,oocytes and media were collected, rapidly frozen on dry ice, and stored at -20 “C. Man-6-P/IGF-lI Receptor Affinity Chromatography-Chromatography of oocyte detergent extracts and media samples on Man-6-P/ IGF-I1 receptor columns has been previously described (8). Low density (1.75 ml of 0.4 mg receptor/ml resin) MPR-Affi-Gel affinity EXPERIMENTAL PROCEDURES columns were used for the initial experiments and were later replaced by high density (1.5 ml of 2.2 mg receptor/ml resin) MPR-Affi-Prep Materials-The preparation of rabbit antisera to human cathepsin D was described previously (12). Rabbit antiserato human pepsinogen affinity columns. Column buffer was 50 mM imidazole/HCl, pH 6.5, was a generous gift of Dr. I. M. Samloff (UCLA School of Medicine, 150mM NaC1,0.05% Triton X-100,5 mM sodium @-glycerophosphate, Los Angeles, CA). Immobilized bovine mannose 6-phosphate/IGF-II and 5 mM EDTA. Fractions corresponding to two oocyte equivalents receptor (MPR) was prepared as described previously (13). Mannose- of column run-through and Man-6-P eluted material were immuno6-P was prepared as described (14). All oligonucleotides were synthe- precipitated with either anti-pepsinogen antisera alone or in combisized on an Applied Biosystems 380A solid-phase synthesizer. Adult nation with anti-cathepsin D antisera and subjected to SDS-polyfemale Xenopus frogs were obtained from NASCO (Fort Atkinson, acrylamide gel electrophoresis (SDS-PAGE) (18) as described (8). WI). The pSP64 plasmid vector was purchased from Promega, Inc., The gels were fluorographed and theappropriate regions of the dried and pGBT plasmid vector was from Gold Biotechnologies, Inc. Re- gel were excised, solubilizedwith 90% PROTOSOL, and radioactivity striction endonucleases, RNA polymerases, T4 DNA ligase, and T4 measured by liquid scintillation counting according to the manufacpolynucleotide kinase were purchased from New England Biolabs or turer’s instructions. The total radioactivity recovered for an individPromega, Inc. Muta-Gene M13 in vitro mutagenesis kit was from ual determination ranged from 2,000 cpm to 360,000cpm with a Bio-Rad. The deoxy- and dideoxynucleoside triphosphates and T7 median value of 13,000 cpm. The “high density” MPR-affinity colDNA polymerase were purchased from Pharmacia LKB Biotechnol- umns bound up to 25% more of the expressed proteins compared with ogy Inc. Geneclean Kit was purchased from BIO 101, Inc. Endo-@- the binding to the“low density” columns. OligosaccharideAnalysis-The procedures used to separate and N-acetylglucosaminidase H (endo H) was purchased from Boehringer Mannheim. EN3HANCE and PROTOSOL solubilizer were from Du analyze the amino and carboxyl N-linked oligosaccharides are dePont-New England Nuclear. Protein A-Sepharose as well as other scribed elsewhere (1l ) . biochemicals were from Sigma, unless otherwise stated. EXPRE35S35S (L-methionine, [35S]:L-cysteine, [35S]),>lo00 Ci/mmol and GDPRESULTS AND DISCUSSION [3,4-3H]mannose, 15-35 Ci/mmol, were purchased from Du PontReplacement of the Lysine 203 and Loop Regions in CathepNew England Nuclear. [o(-%3]dATP, 650 Ci/mmol was from Amersham Corp. sin D Decreases Phosphorylation-A series of constructs enPlasmids-The cDNA clone of human cathepsin D has been pre- coding CPs in which human cathepsinD sequences have been viously described (15). A cDNA clone for human pepsinogen was substituted with the analogous glycopepsinogen sequences generously provided by Dr. J. Tang (Laboratory of Protein Studies, Oklahoma Medical Research Foundation). The cDNAs for glycopep- were prepared to determine if the lysine 203 region (amino sinogen (mPeplZ), CPs 1,3,5, and17 have been previously described acids 188-230) and the loop region (265-292) of cathepsin D (9). ScaI, TaqI, HinfI, HincII, PuuII, PstI, and BamHI restriction are necessary for the efficient phosphorylation of cathepsin endonuclease sites are shared by the cathepsin Dand glycopepsinogen D. These cDNAs were used to generate mRNAs by in vitro cDNAs at the positions 15, 132, 188, 230, 265, 293, and 319, respec- transcription, and the mRNAs were injected into Xenopus tively, in the corresponding cathepsin D aminoacid sequence. Standard recombinant DNA methods (16) were utilized to prepare the oocytes which were then incubated with [“S]methionine/ cysteine for 72 h. The extent of phosphorylation of the excDNAs that encoded CPs 51-68. Briefly, the appropriate restriction endonuclease fragments were isolated from agarose gels by the pressed proteins was determined by their ability to bind to a Geneclean method according tothe manufacturer’s instructions. MPR affinity column. This affinity procedure detects most of These were ligated with T4 DNA ligase, and themixture was used to the phosphorylated molecules as confirmed by direct analysis transform Escherichia coli JM109. Plasmids were isolated and the of [3H]mannose-labeled oligosaccharides of expressed glycoconstructions confirmed by fine restriction mapping. A poly(A) tail was attached to most of the constructs to improve the levels of proteins (9). The expressed proteins in the receptor column run-through and the Man-6-P eluate fractions were immuexpression in Xenopus oocytes. Muta-Gene M13 in vitro mutagenesis kit was utilized to introduce noprecipitated with anti-cathepsin D and/or anti-pepsinogen mutations into the cDNA of cathepsin Dto change arginine 202 and/ antisera and analyzed by SDS-PAGE. or lysine 203 to alanines in the encoded proteins. Single-stranded The chimeric proteins CP51 and CP52 (amino acids 188phage were obtained from a pGBT plasmid containing full-length 230 and either265-319 or 265-292 of cathepsin D substituted cathepsin D cDNA in E. coli CJ236 by the addition of helper phage M13K07. Isolation of single-stranded DNA and the mutagenesis with glycopepsinogen sequences, respectively) were phoswere performed according to manufacturer’s instructions. The mis- phorylated poorly (0.8 and 2.2% receptor binding, respecmatched oligonucleotides introduced mutations that changed the tively, Fig. 1, lanes 5-12). This is to be compared to 94% encoded arginine 202 and/or lysine 203 to alanines, and removed a MPR binding for cathepsin D (Fig. 1, lanes 1-4). In fact, StuI restriction endonuclease site from the cathepsin D cDNA. The CP51 and CP52 were phosphorylated at levels comparable to plasmids were screened by restriction enzyme digestion and agarose CP 1, which contains the amino lobe of cathepsin D and the electrophoresis to detect the mutations and thensubjected to doublestranded DNA sequencing over a region that was subsequently sub- entire carboxyl lobe of glycopepsinogen (0.8 and 2.2%receptor binding versus 1.6% receptor binding, respectively, Fig. 1). cloned into the cathepsin D cDNA in a pSP64 vector. Oocyte Expression-RNA transcripts were synthesized in vitro with These results indicate that when both the lysine 203 region SP6 polymerase from cDNA constructs in pSP64 vectors and microin- andthe loop region of cathepsin D aresubstituted with

chimeric protein containing these two regions was not phosphorylated as well as cathepsin D, indicating that cathepsin D contains yet other elements that contribute to the optimal interaction with phosphotransferase. In the currentstudy, we have replaced the lysine 203 region and amino acids 265-292 (loop region) of cathepsin D with the homologous glycopepsinogen sequences to determine if these regions are necessary for recognition by phosphotransferase. Our experiments have revealed that both regions are required for maximal phosphorylation of procathepsin D. However, replacement of either region individually results in only a partial loss of phosphorylation, indicating that additional regions of procathepsin D contribute to the phosphotransferase recognition domain. These additionalelements have been localized to theamino lobe of procathepsin D.


23344

expressed, CP54 was phosphorylated to the same extent as CP53 (22% receptor binding, Fig. 1) indicating that residues in thelysine 203 region (amino acids 188-230 of cathepsin D) 9 95 72 94 t 2 other than thepositively charged arginine 202 and lysine 203 90 28 05 0820.6 contribute to the phosphotransferase recognition domain. 3.2 2.0 22220 77 We next tested the consequence of separately replacing the 18 22 t 2 60 29 cathepsin D lysine 203 region or the region encompassed by 40 22 21 22 amino acids265-348 with pepsinogen sequence. When amino 1621.2 1.5 82 20 acids 188-230 of procathepsin D were substituted with glycopepsinogen sequence,the resultantchimeric protein (CP55) was phosphorylated 39%, which is less well than cathepsinD, but considerably better than CP51 which lacks both of the critical regions (Fig. 2, lanes 1-4). CP56, which has amino acids 188-265 of procathepsin D substituted with glycopepsinogen residues, was only phosphorylated 12% (Fig. 2, lanes 5-8), indicating that amino acids 230-265 of cathepsin D may 201 2 3 6 5 6 7 8 0101112 13ldIS16 play a role in the recognition domain. However, this region is FIG. 1. Binding of [36S]methionine/cysteine-labeledpro- not sufficient for phosphorylation even in combination with teins to the Man-6-P receptor affinity column. Thevarious the amino lobe of procathepsin D as illustrated by the poor chimeric proteins tested are shown schematically with the open bar representing cathepsin D (CATH D)sequence and the line represent- phosphorylation of CP51 and CP52 (Fig. 1). On the other ing pepsinogen sequence. The arrowheads denote the glycosylation hand, CP55 and CP56, which contain the entire amino lobe sites, and the vertical line represents the cleavage site of the propep- of procathepsin D, were phosphorylated about 10-fold better tide. Xenopus oocytes were microinjected with the various construct than therespective chimeric proteins that contain the amino mRNA and then incubated with [""S]methionine/cysteinefor 72 h. lobe of glycopepsinogen (CP3 and CP5, summarized in Fig. 3; Oocyte detergent homogenates (C) and media ( M )were applied to 39% versus 4.1% receptor binding and 12% versus 1.4% recepCI-MPR affinity columns, and an equivalent of two oocytes of the tor binding, respectively). Therefore, the amino lobe of procolumn run-through ( R T )and Man-6-P eluates (M6P) were immuthe phosphorylnoprecipitated and analyzed by SDS-PAGE and fluorography. Ap- cathepsin D appears to contribute to efficient propriate regions of the dried gels were excised and the radioactivity ation of CP55 and CP56. measured by liquid scintillation counting. The results for each chiCP57 was tested to determine theeffect of replacing amino meric protein are summarized on the figure and representthe average acids 265-319 of procathepsin D with glycopepsinogen seof two experiments with the exception of CP54 which is a single quence. When expressed, CP57 was also phosphorylated less determination. The data for cathepsin D,CP1, and CP17 are from CP51 (30% Ref. 9. Total percent bound to CI-MPR is calculated by combining well thancathepsin D butmuchbetterthan the results of CI-MPR binding of oocyte and media material and is receptor binding, Fig. 2, lanes 9-12). The results obtained with CP57 and CP55 demonstrate that either thelysine 203 reported & the standard error of the mean. region or the loop region of cathepsin D can be separately glycopepsinogen sequence, no other regions of cathepsin D replaced with glycopepsinogen sequence without severely impairing the ability of procathepsin D to interact with phosare able to generate an efficient recognition domain. The poor phosphorylation of CP51 and CP52 is unlikely to photransferase. However, these chimeric proteins are phosbe a consequence of improper folding of the chimeric mole- phorylated 2-3-fold less well than procathepsin D (30 and cules since 90% of CP51 and 77% of CP52 were secreted into 39% receptor binding versus 94% receptor binding, Fig. 2). CP58 was constructed to examine the role of the COOHthe medium, in contrast misfolded to proteins that are usually (19). The decreased retained in the endoplasmic reticulum x I .eomd CI-MPR mobility of the secretedglycoproteins on SDS-PAGE (Fig. 1, Cells Medm lolo1 Serresd lanes 7 and 11 ) is consistent with the oligosaccharides having 9 95 72 94 2 2 CATH I been processed to complex-type forms.:' I5 50 3929 60 CP55 Since the substitution of lysine 203, in combination with 32 A4 12.5 72 CP56 amino acids265-348 of cathepsin D, into theglycopepsinogen 40 11 30.3 36 CP57 backbone results in a chimeric protein (CPl7, Fig. 1) that is 69 22 462 5 46 CP58 phosphorylated aswell as a chimeric molecule containing the 8, I 86 61 80+6 20 CP59 entire cathepsin D sequence 188-230 (9), we next analyzed A 76 80 22 67 CP60 the consequence of replacing lysine 203 of cathepsin D with 68 86 IO 86 CP61 an alanineresidue. When this was done, in combination with the replacement of amino acids 265-319, the resultant chi87 76 66 CP62 53 meric protein(CP53)was phosphorylated considerably better than CP51 and CP52, but less well than cathepsin D (22% receptor binding, Fig. 1, lanes 13-16). The finding that CP53 is phosphorylated better thanCP51 and CP52 indicates that other residuesin the lysine 203 region contributetothe recognition domain. One candidate residue is the positively charged arginine a t position 202 in cathepsin D. To test this 29possibility, we constructed CP54 which contains alanine resI 2 3 1 S I 7 8 0101l12 I3l1IS16 idues a t positions 202 and 203 and amino acids 265-319 of FIG. 2. Regional substitution of glycopepsinogen into cacathepsin D substituted withglycopepsinogen residues. When thepsin D. The schematic representation of the various chimeric proteins and the details of the sample analysis are described in the 'The secreted glycoproteinsCP51 and CP52 were mostly resistant legend to Fig. 1.The resultsfor each chimeric protein are summarized to endoglycosidase H digestion (T. Baranski, unpublished observaand represent the average of two to four separate experiments with tion). the exception of CPs 60-62 which are single determinations. a

S.cr.,.d

25

C1.1.

I6

%B o v d 10 CI-MPR Madm k o l

31

20 t 4

to

Phosphorylation Lysosomal Enzyme

23345

protein and its retention in the cell, although a number of other exceptions were noted. Most of these were chimeric proteins thatwere poorly phosphorylated butonly secreted to 14 47 CP 5 16 I 2 a limited extent (such as CP64 and CP63, Fig. 3). The most 4.4 40 3.9 5 2 CP63 likely explanation for this phenotype is that the protein is I2 5 6 22!20 24 CP64 impaired in folding and consequently is retained in the en73 9 2 2.5 4.4f33 CP65 doplasmic reticulum. 74 7 4 CP66 26 12 ? 6 Analysis of Amino Lobe Elements of Procathepsin D That 24 09 1750.1 52 CP67 Contribute to the Recognition Domain-The finding that CP55 49 CP68 22 36 I4 t 6 isphosphorylated 10-fold betterthan CP3 indicates that CP55 40 I 5 5 0 39 * 9 elements of the amino lobe of procathepsin D contribute to the efficient phosphorylation of theprotein.To begin to identify these elements, constructs were prepared that systematicallysubstitutedthree consecutivesegments of the amino lobe of procathepsin D, in combination with amino acids 230-348 of cathepsin D, into glycopepsinogen. CPs 6365 contain individual regions of procathepsin D (amino acids 29-44 to 14, 15-131, or 132-187) substituted into the amino I 2 3 1 S b 7 I 9101112 I3ldISIb lobe of CP3 or CP5 (see Fig. 3). When expressed, the chimeric FIG. 3. Substitution of consecutive segments of the amino lobe of procathepsin D into a glycopepsinogen/cathepsinD proteins were all poorly phosphorylated (4.4, 2.2, and 4.4% receptor binding, respectively, Fig. 3). Therefore, no individchimera. Theschematicrepresentation of thevariouschimeric proteins and the details of the sample analysis are described in the ual segment of the amino lobe functions aswell as the entire legend to Fig. 1.The results for each chimeric protein are summarized amino lobe of procathepsin D in CP55. and represent the average of two to four separate experiments with CPs 66-68 contain combinationsof the individual segments the exceptionof CP5 and CP63 which are single determinations. The of procathepsin D. When expressed, CPs 66 and 68 were data for CP3 and CP5 include determinations from Ref. 9. phosphorylated at intermediate levels relative to CP3 and CP55 (12 and 14% uersus 4.1 and 39%, receptorbinding, terminal 28 amino acids of cathepsin D. In the crystallorespectively, Fig. 3). Therefore, the amino acids 15-131 of graphic structure of porcine pepsinogen, this region is buried cathepsin D in combination with either amino acids -44 to internally in the molecule, except for aturn of /J"loop structure 14 or aminoacids 132-187 of procathepsin D are sufficient to that is accessible on the surface and in close proximity to mediate increased levels of phosphorylation. However, CP67, CP55 a n arginine which contains amino acids -44 to 14 and 132-187 in comlysine 203. It was possible,therefore, that in that occurs a t position 339 intheturn of the p-loop of bination, was poorly phosphorylated (1.7% receptor binding, cathepsin D could function in the recognition domain when Fig. 3). These results are consistentwith two interpretations. the lysine 203 is absent. InCP58, the glycopepsinogen COOH First, multiple regions of the amino lobe of procathepsin D terminus, which contains a glutamine a t position 339, was may contribute specifically to therecognition domain. Second, substituted for the COOH terminus of cathepsin D. When the critical determinantsfor the recognition domain may be expressed, CP58 was phosphorylated to a similar extent as located within the segment that contains amino acids 15-131, CP55 (Fig. 2, 46 uersus 39% receptor binding, respectively), but the other amino lobe segments of procathepsin D are excluding a critical role for arginine 339. required for this segment to achieve a proper conformation. CP59 and CP60 were prepared to determine whether the Oligosaccharide Analysis on CP55"There are several possingle replacement of lysine 203 or the replacement of arginine sible mechanisms by which the aminolobe of procathepsin D 202 and lysine 203 with alanineresidues would have anyeffect could enhance the phosphorylation of chimeric proteins.First, on the phosphorylation of procathepsin D. When expressed, the amino lobe could directly contribute to the increased both chimeric proteinswere well phosphorylated, although a t phosphorylation of the molecule by specific regions serving as somewhat diminished levels compared with procathepsin D (80 and 74% uersus 93% receptor binding, respectively, Fig. direct contact sites for the binding of phosphotransferase. 2). These results demonstrate that, despite its critical role in Alternatively these elements could act to position the amino it a better the generationof a minimal carboxyl lobe recognition domain, lobe oligosaccharide in sucha manner that becomes lysine 203 can bereplaced by alanineinprocathepsin D substrate for phosphotransferase. Another possible role for without a major deleterious effect on the phosphorylation of the aminolobe is toimprove the overall folding of the carboxyl lobe and thereby generate a more potent carboxyl lobe recthe resultant molecule. T o examine theroles of NH,- and COOH-terminalresidues ognition domain. In this case, the effect of the amino lobe would be an indirectone. One approach distinguish to between inprocathepsin D, CP61 and CP62 were prepared.When oligosaccharide siteexpressed, these chimeric proteins were well phosphorylated thesemechanismsistocomparethe specific phosphorylation of CP55 with that obtained with (84 and 76% receptor binding, Fig. 2). Thus, aminoacids -44 cathepsin D and a number of the otherchimeric proteins (11). to 15 and 319-348 of procathepsin D are not necessary to generate anefficient phosphotransferase recognition domain. In cathepsin D, the amino lobe oligosaccharide is 76% phosHowever, shared residues between procathepsin D and gly- phorylated whereas the carboxyl lobe oligosaccharide is 96% copepsinogen in these regions may contribute to therecogni- phosphorylated (11).On the other hand, in CP2 which conlobe of glycopepsinogen joined to thecarboxyl tion domain, since they would fail to be identified by this tains the amino lobe of cathepsin D, the amino lobe oligosaccharide is only approach. It should be noted thatCP62 was secreted toa much greater 37% phosphorylated while the carboxyl lobe is 93% phosextent than the other chimeric proteins with a similar degree phorylated (11).Therefore, in CP2, the patternof phosphorylof phosphorylation (compareCP62 to CP61 for example). The ation at the two glycosylation sites appears to reflect the reason for this is not clear. Overall, there was a correlation position of the phosphotransferaserecognition domain in the between the extent of phosphorylation of a given chimeric molecule. If the role of the amino lobe of procathepsin D in %

S.r?.t.d

CP 3

51

C.11,

%BoundloCI-MPR M d o lolol

51

3 9

41'11


23346

CP55 is to directly interact with phosphotransferase, then a greater percentage of the amino lobe oligosaccharides might be phosphorylated relative to the carboxyl lobe oligosaccharides. Conversely, if the role of the aminolobe of procathepsin C-COOH 800

,

I

P-NHZ C-NHP

I

D in CP55is to promote a more properly folded carboxyl lobe recognition domain, then the carboxyl lobe oligosaccharides would be phosphorylated to a greater extent than the amino lobe oligosaccharides, as is the case for CP2. In an attempt to differentiate between these two possibilities, an analysis of the extent of phosphorylation of the N-linked oligosaccharides at the two glycosylation sites of CP55 was performed. Xenopus oocytes were coinjected with GDP-[3,4-'H]mannose and CP55 RNA transcripts, and then incubated for 72 h. The cellular and secreted CP55 molecules were immunoprecipitated with anti-cathepsin D serum and digested with chymotrypsin to generate glycopeptides. These were fractionated by reversed-phase liquid chromatography under conditions where the aminolobe glycopeptide is separated from the carboxyl lobe glycopeptide (11). The elution profiles of the glycopeptides are shown in Fig. 4. As expected, the amino lobe glycopeptide of CP55 fractionated in theposition of the amino lobe glycopeptide of procathepsin D whereas the carboxyl lobe glycopeptides of CP55 fractionated in theposition corresponding to thecarboxyl lobe of glycopepsinogen. Each glycopeptide was incubated with endo H to release the high mannose oligosaccharides which were then analyzed for the presence of phosphomannosyl residues by QAE-Sephadex chromatography (20). As shown in Fig. 5, the oligosaccharides of CP55contained species that bound to QAESephadex and eluted atpositions characteristic for oligosaccharides with one phosphomonoester (70 mM NaC1) or two phosphornonoesters (140 mM NaCl). The endo H-resistant material was subjected to concanavalinA-Sepharose chromatography to determine the amount of complex-type units for each glycopeptide (21). The results of these oligosaccharide analyses are summarized inTable I. Theamino lobe andthe carboxyl lobe oligosaccharides of CP55 were equally phosphorylated (48 and 53%, respectively). This pattern of phosphorylation is similar to that observed withprocathepsinD and issignificantly differentfrom that observed withCP2. These results are consistent with the proposal that the procathepsin D amino lobe elements in CP55 contribute directly to the interaction with phosphotransferase rather than just serving to enhance the folding of the carboxyl lobe elements.

P-COOH

I

I

I

I

Media

Retention Time (min) FIG. 4. Fractionation of glycopeptides by reversed-phase liquid chromatography. Xenopus oocytes were coinjected with GDP-[3,4-3H]mannoseand the CP55 mRNA transcript and the oocytes were incubated for 72 h. The cellular homogenates and media samples of CP55 were immunoprecipitated, and glycopeptides were generated by chymotryptic digestion and then fractionated by reversed-phase liquid chromatography, as described (11). The glycopeptides of cathepsin D and glycopepsinogen expressed in the same experiment served as controls for the identification of each glycopeptide, and thepositions of their elution are shown by the arrows at the top of the figure. C-NH,, cathepsin D amino lobe glycopeptide; CCOOH, cathepsin D carboxyl lobe glycopeptide; P-NH,, glycopepsinogen amino lobe glycopeptide; P-COOH, glycopepsinogen carboxyl lobe glycopeptide.

CONCLUSIONS

Previously we identified two noncontinuous primary sequences (amino acids 188-230 and 265-292) in the carboxyl

B

A

600 -

20mM

I

FIG. 5. QAE-Sephadex analysis of CP55 oligosaccharides. The fractionated glycopeptides (Fig. 4 ) were treated with endo H to release the high mannose type-oligosaccharides whichwere then separated from endo H-resistant glycopeptides by reversed-phase chromatography, as described (11). Cell-associated ( A and B ) and medium ( C and D ) high mannose type-oligosaccharides derived from the amino lobe ( A and C ) or the carboxyl lobe ( B and D ) of CP55 were applied to analytical QAE-Sephadex columns and eluted with either 20, 70, 100, or 140, mM NaCl.

z

70mM

I

60 -

l0OW 140mM NaCl

I

I

400 -

40

0

io

0

I

100 mM 140 mM NaCl

I

I

OO

s-

I

20

10

30

40

50

D-

C

z a

70 mM

-\

20 1

-

60

400 -

40

200 -

20 -

600

20 mM

I

io

Fraction

-

OO-YO

-*go

Fr

Phosphorylation Lysosomal Enzyme

23347

TABLE I Analysis of [3H]mannose-labeled oligosaccharides The data obtained from the QAE-Sephadex analysis of the endo H-released oligosaccharides (Fig. 5) are summarized.The endo H-resistant glycopeptides from the various samples were analyzed by concanavalin A-Sepharose chromatography to determine the content of complextype units (21). NHM, neutral high mannose type; PM, phosphomonoester. OligosaccharideSpecies (9% of total) Protein

CP 55

Source

Oocyte

9% of total"

51

NH, lobe COOH lobe Media NH2 lobe COOH lobe

32 43

Total cpmb

Complex NHM+2PM NHM+lPM NHM

24 20 21 35 41 28

32 33 30 8 8 8

38 44 50 41 53

6.5 2 11 8.5 5 12

3,650 3,200 450 4,050 3,600 450

The data are reportedas molar percents by assuming that the high mannose oligosaccharidescontain 5.1 mannoses (11) and the complextype oligosaccharides contain three mannose residues. * As described (ll), the recoveries of the carboxyl lobe hydrophobicglycopeptides were consistently lower than the recoveries of the amino lobe glycopeptides, thus less material was available for subsequent analysis.

lobe of cathepsin D that, when substituted together into the homologous regions of glycopepsinogen, were sufficient to generate a phosphotransferase recognition domain (9). The data presented in the current study demonstratethat replacement of these two regions in procathepsin D with the homologous glycopepsinogen sequences results in a chimeric protein that is very poorly phosphorylated, thereby establishing that these two carboxyl lobe elements are also necessary for the expression of the phosphotransferase recognition domain in procathepsin D. However, when these regions were replaced individually, the resultant chimeric proteins were well phosphorylated (although not to the same extent as procathepsin D). The ability of these latter chimeric proteins to serve as substrates for phosphotransferase was found to be dependent on the presence of amino lobe elements of cathepsin D. These resultscan be explained in two ways. The first possibility is that procathepsin D contains two independent phosphotransferase recognition sites, one oneach lobe of the molecule, with the carboxyl lobe elements being more potent than the amino lobe elements. Secondly, there could be a single extended recognition site that includes the carboxyl lobe elements as well as regions of the amino lobe. In addition, the amino lobe elements of cathepsin D may improve the overall folding of the carboxyl lobe and thereby enhance the expression of the carboxyl lobe recognition domain. The independent site model is an interesting possibility since the aspartylproteases arethoughtto have arisen by a gene duplication event (22). However, when the homologous lysine 203 and loop regions in the amino lobe of procathepsin D are localized on thecrystal structure of porcine pepsinogen, they are predicted to be buried at theinterface of the two lobes of procathepsin D. Therefore, a second independent phosphotransferase recognition site would have to be located elsewhere on the amino lobe. The single extendedsite modelwould necessitate that phosphotransferase bind avery large area on the surface of procathepsin D, since molecular modeling predicts the surface area of the carboxyl lobe recognition domain to be about 1,600 square angstroms (23). This would appear to be possible since protein-protein surface interactions up to 5,000 square angstroms have been reported, primarily in the association of subunits in multimeric proteins (24-26). To differentiate between these various models, the elements in the amino lobe that lead to enhanced phosphorylation will have to be more precisely defined. Any model of phosphotransferase action must explain how this enzyme can interact with 40-50 different lysosomal hydrolases while displaying low affinity for hundreds of other glycoproteins with identical high mannose-type oligosaccha-

rides. One possible mechanism that we have previously suggested is that phosphotransferase binding may involve multiple contacts extendingover a broad surface of the lysosomal hydrolase, with only a portion of these required to generate a productive interaction (9). In thisway the different lysosomal hydrolases need only express some components of the entire recognition marker to be able to bind to phosphotransferase. The single extended site model is compatible with this view. An alternative possibility is that phosphotransferase contains binding sites for a variety of protein motifs. This would be analogous to the interaction of human growth hormone with its receptor (27). In this system a single molecule of growth hormone binds and brings together two molecules of the receptor. DeVos et al. (27) have found that both receptor molecules are involved in the growth hormone binding and utilize essentially the same residues to interact with the ligand. However, the two binding sites on growth hormone have no structural similarity. These data show that thesingle binding site onthe growth hormone receptor is able to interact with two different structural elements. In a similar manner, phosphotransferase could potentially bind to a number of different protein motifs that are present on the various lysosomal hydrolases. Consequently, there would be no need for all lysosomal hydrolases to express the same determinants. We are attempting to distinguish between these possibilities by examining the phosphotransferase binding determinant on a second lysosomal hydrolase. Acknowledgments-We thank I. Michael Samloff and Beth Westlund for generously providing human pepsinogen antisera and Walter Gregory for providing human cathepsin D antisera, mannose 6phosphate, and MPR affinity columns. REFERENCES 1. Kornfeld, S. (1986) J. Clin. Inuest. 77,1-6 2. Hasilik, A., Waheed, A., and von Figura, K. (1981) Biochern. Biophys. Res. Cornmun. 9 8 , 761-767 3. Reitman, M. L., and Kornfeld, S. (1981) J . Bid. Chem. 256 11977-11980 4. Waheed, A., Hasilik, A., and von Figura, K. (1982) J. Bioi. Chern. 2 5 7 , 12322-12331 5. Lang, L., Reitman, M., Tang, J., Roberts, R. M., and Kornfeld, S. (1984) J. Biol. Chern. 2 6 9 , 14663-14671 6. Rosenfeld, M. G., Kreibich, G., Popov, D., Kato, K., and Sabatini, D. D. (1982) J. Cell Bid. 9 3 , 135-143 7. Gieselmann, V., Pohlmann, R., Hasilik, A., and von Figura, K. (1983) J. Cell Biol. 9 7 , 1-5 8. Faust, P. L., Wall, D. A. Perara E., Lingappa,V. R., and Kornfeld, S. (1987) J. Cell. Biol. 106, 1937-i945 9. Baranski, T. J., Faust, P. L., and Kornfeld, S. (1990) Cell 63,281-291 10. Davies, D. R. (1990) Annu. Rev. Biophys. Biophys. Chem. 19, 189-215 11. Cantor, A. B., Baranski, T. J., and Kornfeld, S. (1992) J. Biol. Chem. 2 6 7 , 23349-23356 12. Takahashi, T., and Tang, J. (1981) Methods Enzyrnol. 80,565-581 13. Varki, A., and Kornfeld, S. (1983) J. Biol. Chern. 2 5 8 , 2808-2818 14. Stein, M. (1957) Methods Enzyrnol. 3 , 154-157 15. Faust, P. L., Kornfeld, S., and Chirgwin, J. M. (1985) Proc. Natl. Acad. Sci. U. S. A . 82,4910-4914

23348


16. Maniatis, T.,Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning; a Laboratory Manmi, Cold Spring HarborLaboratory, Cold Spring Harbor,

NY

17. Colman, A. (1984) in Translation of Eukaryotic Messenger RNA in Xenopus oocytes (Hames, B. D., and Higgins, S. J., eds) pp. 271- 302, IRL Press, Oxford 18. Laemrnli, U. K. (1970) Nature 227,680-685 19. Lodish, H.(1988) J. Biol. Chem. 263,2107-2110 20. Varki. A,. and Kornfeld. S. (1980) J. B~ol.Chem. 265. 10847-10858 21. Cummings, R.D.,and Kornfeld, S. (1982) J. Biol. Chem. 257,11235-11240

22. Tang, J., James, M. N. C., Hsu, I. N., Jenkins, J. A,, and Blundell, T.L. (1978) Nature 271,618-621 23. Baranski, T.J., Koelsch, G., Hartsuck, J. A,, and Kornfeld, S. (1991) J. Biol. Chem. 266,23365-23372 24. Miller, S., Lesk, A. M., Janin, J., and Chothia,C. (1987) Nature 328,834836 25. Argos, P. (1988) Protein Eng. 2 , 101-113 26. Janin, J., Miller, S., and Chothia, C. H. (1988) J. Mol. Eiol. 2 0 4 , 155-164 27. DeVos. A. M.. Ultsch. M.. and Kossiakoff. A. A. (1992) Science 255.306312