Lysosomal Acid �-Glucosidase Consists of Four Different PeptidesProcessed from a Single Chain Precursor*

Received for publication, April 12, 2004, and in revised form, October 20, 2004Published, JBC Papers in Press, November 1, 2004, DOI 10.1074/jbc.M404008200

Rodney J. Moreland‡, Xiaoying Jin§, X. Kate Zhang§, Roger W. Decker‡, Karen L. Albee§,Karen L. Lee§, Robert D. Cauthron‡, Kevin Brewer‡, Tim Edmunds§, and William M. Canfield‡¶

From the ‡Genzyme Corporation, Oklahoma City, Oklahoma 73104 and the §Genzyme Corporation,Framingham, Massachusetts 01701

Pompe’s disease is caused by a deficiency of the lyso-somal enzyme acid �-glucosidase (GAA). GAA is synthe-sized as a 110-kDa precursor containing N-linked carbo-hydrates modified with mannose 6-phosphate groups.Following trafficking to the lysosome, presumably viathe mannose 6-phosphate receptor, the 110-kDa precur-sor undergoes a series of complex proteolytic and N-glycan processing events, yielding major species of 76and 70 kDa. During a detailed characterization of hu-man placental and recombinant human GAA, we foundthat the peptides released during proteolytic processingremained tightly associated with the major species. The76-kDa form (amino acids (aa) 122–782) of GAA is asso-ciated with peptides of 3.9 kDa (aa 78–113) and 19.4 kDa(aa 792–952). The 70-kDa form (aa 204–782) contains the3.9- and 19.4-kDa peptide species as well as a 10.3-kDaspecies (aa 122–199). A similar set of proteolytic frag-ments has been identified in hamster GAA, suggestingthat the multicomponent character is a general phenom-enon. Rabbit anti-peptide antibodies have been gener-ated against sequences in the proteolytic fragments andused to demonstrate the time course of uptake and proc-essing of the recombinant GAA precursor in Pompe’sdisease fibroblasts. The results indicate that the ob-served fragments are produced intracellularly in thelysosome and not as a result of nonspecific proteolysisduring purification. These data demonstrate that themature forms of GAA characterized by polypeptides of76 or 70 kDa are in fact larger molecular mass multicom-ponent enzyme complexes.

Lysosomal acid �-glucosidase (GAA1; EC 3.2.1.3) is an exo-1,4- and -1,6-�-glucosidase that hydrolyzes glycogen to glucose.The cDNA for GAA encodes a protein of 952 amino acids witha predicted molecular mass of 105 kDa (1). The newly synthe-sized precursor has an amino-terminal signal peptide for co-translational transport into the lumen of the endoplasmic re-

ticulum, where it is N-glycosylated at seven glycosylation sites,resulting in a glycosylated precursor with an apparent molec-ular mass of 110 kDa.

The intracellular processing of GAA has been investigatedpreviously (2, 3). It was proposed that, after transport throughthe Golgi complex and targeting to the endosome/lysosome, the110-kDa precursor is proteolytically processed at the aminoterminus, resulting in a 95-kDa intermediate with a sequencebeginning at amino acid 122. Prior to this study, the 95-kDaintermediate was proposed to be proteolytically processed to a76-kDa form, which was believed to occur between amino acids816 and 881 (3). The 76-kDa form is then proteolytically pro-cessed at the amino terminus at amino acid 204 to give the70-kDa mature form (3). The nomenclature used for the pro-cessed forms of GAA is based on apparent molecular mass asdetermined by SDS-PAGE.

The identities of the proteases involved in the maturation ofGAA have never been established. GAA has been purified frommany different tissues such as bovine testis (4), rat liver (5), pigliver (6), human liver (7), rabbit muscle (8), human heart (9),human urine (10), and human placenta (2, 11). The predomi-nant species observed are the 76- and 70-kDa mature forms.The fate of the cleaved fragments is unknown. Several reportshave noted the presence of small polypeptides in highly puri-fied preparations of the 76- and 70-kDa forms (2, 10, 14). It wassuggested that the small polypeptides might be contaminants,degradation products, or a previously described GAA activationprotein (11).

Proteolytic processing appears to be required for optimal ac-tivity toward the natural substrate glycogen. There is a 7–10-foldincrease in the affinity of the 76/70-kDa species for glycogencompared with the 110-kDa precursor (3, 17). In addition to theproteolytic maturation of the GAA peptide backbone, there isextensive processing of the carbohydrate chains. GAA is targetedto the lysosomes by the mannose 6-phosphate receptor, but anal-ysis of the carbohydrate chains from purified 76/70-kDa GAAfrom human placenta revealed the absence of mannose 6-phos-phate and additional carbohydrate processing (18).

A deficiency of acid �-glucosidase causes Pompe’s disease,which results in the accumulation of glycogen in lysosomes.Pompe’s disease is an autosomal recessive disorder that variesfrom a fatal infantile form to a more slowly debilitating adult-onset form (reviewed in Ref. 19). Efforts are under way todevelop enzyme replacement therapy for Pompe’s disease. Sev-eral similar but distinct transgenic (2, 3) or recombinant (17,18) GAA preparations have been developed for this purpose.Several reports have demonstrated that enzyme replacementtherapy using the 110-kDa precursor form of GAA degradeslysosomal glycogen in cultured Pompe’s disease fibroblasts andGAA knockout mice (2, 3, 18). Although the 110-kDa form of

* The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked“advertisement” in accordance with 18 U.S.C. Section 1734 solely toindicate this fact.

¶ To whom correspondence should be addressed: Genzyme Corp., 800Research Parkway, Suite 200, Oklahoma City, OK 73104. Tel.: 405-271-8144; Fax: 405-271-8188; E-mail: william.canfield@genzyme.com.

1 The abbreviations used are: GAA, acid �-glucosidase; BisTris, 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol; HPLC, highpressure liquid chromatography; rhGAA, recombinant human acid �-glu-cosidase; CHO, Chinese hamster ovary; HP-rhGAA, highly phosphoryl-ated recombinant human acid �-glucosidase; PVDF, polyvinylidene diflu-oride; MALDI-TOF-MS, matrix-assisted laser desorption ionization time-of-flight mass spectrometry; LC/MS, liquid chromatography/massspectrometry; MS/MS, tandem mass spectrometry; CAPS, 3-(cyclohexy-lamino)propanesulfonic acid; aa, amino acids.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 280, No. 8, Issue of February 25, pp. 6780–6791, 2005© 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org6780

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

GAA has been administered therapeutically, it is the matureforms that are found in the lysosome and that are most activetoward the glycogen substrate.

In this study, we demonstrate that proteolytically processedGAA purified from human placenta consists of four differentpeptides processed from a single chain precursor. We demon-strate that the recombinant 110-kDa GAA precursor is inter-nalized by Pompe’s disease fibroblasts via the mannose 6-phos-phate receptor and undergoes proteolytic cleavage duringwhich several of the cleaved fragments remain associated.

EXPERIMENTAL PROCEDURES

Materials

Concanavalin A, DEAE-Sepharose FF, and Superdex 200 prep gradewere obtained from Amersham Biosciences. �-Methyl glucoside, benza-midine, methionine sulfoximine, and 4-methylumbelliferyl �-D-gluco-side were obtained from Sigma. Other chemicals were reagent grade orbetter and were from standard suppliers. SDS-polyacrylamide gelswere obtained from Invitrogen. Antibodies were made by WashingtonBiotechnology (Baltimore, MD). Roller bottles were obtained from Corn-ing. Dulbecco’s modified Eagle’s medium and fetal bovine serum wereobtained from JRH Biosciences. Dialyzed phosphate-buffered salinewas obtained from Hyclone Laboratories.

Acid �-Glucosidase Activity and Protein Assay

Acid �-glucosidase was assayed fluorometrically in a microtiter plateusing 4-methylumbelliferyl �-D-glucoside as a substrate in 25 mM so-dium acetate (pH 4.8) (10). Protein concentration was estimated byabsorbance at 280 nm assuming E1% � 10 or using the micro BCA assay(Pierce) standardized with bovine serum albumin (20).

SDS-PAGE

Reduced and nonreduced samples and molecular mass markers (Am-ersham Biosciences) were applied to a 4–20 or 10% Tris/glycine/SDS-polyacrylamide gel (Invitrogen). Electrophoresis was performed at 150V for 1.5 h, and proteins were visualized with either Coomassie Blue orsilver stain (21).

Buffers

Buffer A contained 25 mM Tris-HCl (pH 6.5), 2 mM phenylmethylsul-fonyl fluoride, 2 mM benzamidine, 50 mM NaCl, and 0.25% Triton X-100.Buffer B contained 20 mM Tris-HCl (pH 6.5), 1 mM CaCl2, 1 mM MnCl2,and 500 mM NaCl. Buffer C contained 20 mM BisTris-HCl (pH 5.5).Buffer D contained 20 mM BisTris-HCl (pH 5.5) and 500 mM NaCl.Buffer E contained 3 M (NH4)2SO4 and 75 mM Tris-HCl (pH 6.5). BufferF contained 1 M (NH4)2SO4 and 25 mM Tris (pH 6.5). Buffer G contained50 mM NaH2PO4 (pH 6.5) and 20 mM NaCl. Buffer H contained 40 mM

Tris-HCl (pH 8.3), 39 mM glycine, 20% MeOH, and 0.04% SDS. Buffer Icontained 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM CaCl2, and0.05% Tween 20. Buffer J contained 20 mM NaH2PO4 (pH 6.5).

Cell Lines

CHO-K1 cells were obtained from American Type Culture Collection(Manassas, VA). The Pompe’s disease fibroblast cell line GM00248 wasobtained from Coriell Cell Repositories (Camden, NJ). This cell line ishomozygous for the common African-American R854X mutation.

Purification of Hamster Acid �-Glucosidase fromChinese Hamster Ovary Cells

Step 1: Harvesting Cells and Cell Lysis—CHO-K1 cells were ex-panded to 500 1700-cm2 roller bottles in medium containing Dulbecco’smodified Eagle’s medium/1� glutamine synthetase supplement, 10%dialyzed fetal bovine serum, 25 �M methionine sulfoximine, and 5% CO2

at 37 °C. Each bottle was rinsed with 40 ml of dialyzed phosphate-buffered saline and then rolled for 30 min at 37 °C in the presence of 40ml of buffer A. The cell lysate was clarified with a 10-inch 0.2-�mOpticap filter (Millipore Corp.) and stored at �20 °C.

Step 2: Concanavalin A—The clarified cell lysate was adjusted to 1mM MnCl2 and 1 mM CaCl2 and applied to a concanavalin A column(5.0 � 18 cm) equilibrated with buffer B at 50 ml/min. The column waswashed with �7 column volumes of buffer B and step-eluted with bufferB containing 0.25 M �-methyl glucoside. Three 1-liter fractions werecollected and pooled.

Step 3: Affinity Chromatography on Superdex 200—GAA is retardedbecause of its affinity for the dextran backbone of the Superdex matrix.

Solid (NH4)2SO4 (0.39 g/ml) was added to the concanavalin A eluatewith stirring at 4 °C. The suspension was stirred overnight at 4 °C. Theprecipitate was collected by centrifugation at 38,400 � g for 30 min ina JLA 16.250 rotor and then dissolved in a final volume of 30 ml ofbuffer C. The sample was applied to a Superdex 200 column (5 � 90 cm)pre-equilibrated with buffer D at 10 ml/min. Fractions of 40 ml werecollected and assayed for acid �-glucosidase activity and protein.

Step 4: DEAE-Sepharose FF—Two activity peaks eluted from theSuperdex 200 column. The first activity peak contained 25% of therecovered activity. Because no additional purity was gained in the firstactivity peak, we could not isolate GAA from that peak. However, thesecond activity peak to elute, which contained 75% of the recoveredactivity, was significantly enriched for GAA. The second peak waspooled and diluted with 2 volumes of buffer C and applied to a DEAE-Sepharose FF column (1.6 � 10 cm) equilibrated with buffer C at 10ml/min. After washing with 5 column volumes, the column was devel-oped with a 30-column volume gradient from 0.0 to 0.3 M NaCl in bufferC. Fractions of 10 ml were collected and assayed for acid �-glucosidaseactivity and protein.

Step 5: Reverse-phase Chromatography—Reverse-phase narrow boreHPLC was performed using a Michrom Bioresources ultra-fast microprotein analyzer. An aliquot (�50 �g) of Chinese hamster acid �-glu-cosidase (DEAE fraction) was diluted 1:1 with “magic mixture” dena-turant (4.0 M guanidine hydrochloride, 4.0 M urea, 7.5% acetonitrile,0.15% trifluoroacetic acid, and 0.2% Zwittergent ZC-8 (Calbiochem)) (4)and applied to a 2.1 � 50-mm PLRP-S reverse-phase HPLC column(4000-Å pore size, 8-�m particle size; Michrom Bioresources, Inc.) pre-equilibrated at 54 °C with 98% eluent A (2% acetonitrile, 0.1% triflu-oroacetic acid, and 0.03% Zwittergent ZC-8) and 2% eluent B (90%acetonitrile, 0.09% trifluoroacetic acid, and 0.03% Zwittergent ZC-8).The column was rapidly stepped to 20% eluent B and developed at 0.18ml/min with a 7.2-ml linear gradient from 20 to 95% eluent B. Fractionsof �0.18 ml were collected.

Purification of Mature Acid �-Glucosidase from Human Placenta

The purification of human placental acid �-glucosidase was based onpreviously reported methods (14, 18).

Step 1: Homogenization of Tissue—Three human placentas werediced after the connective tissue had been removed and added to aWaring blender. An equal weight of ice-cold 25 mM NaCl was added,and the tissue was homogenized for 3 min. The homogenate was thencentrifuged at 38,400 � g for 30 min at 4 °C. The supernatant wasclarified with a 4-inch 0.22-�m Opticap filter (Millipore Corp.).

Step 2: Concanavalin A—The clarified supernatant was adjusted to 1mM MnCl2 and 1 mM CaCl2 and applied to a concanavalin A column(2.6 � 20 cm) equilibrated with buffer B at 9 ml/min. The column waswashed with �10 column volumes of buffer B and step-eluted withbuffer B containing 0.25 M �-methyl glucoside. Three 0.25-liter fractionswere collected and pooled.

Step 3: Phenyl-Sepharose 6 FF—The concanavalin A eluate wasadjusted to 1 M (NH4)2SO4 with buffer E and applied to a phenyl-Sepharose 6 FF column (1.6 � 20 cm) equilibrated with buffer F at 6.5ml/min. After washing the column with 5 column volumes of buffer F,the column was washed with buffer F containing 0.25 M (NH4)2SO4.Placental GAA was eluted with a 5-column volume gradient from 0.25to 0 M (NH4)2SO4 in buffer F. Fractions of 15 ml were collected andassayed for acid �-glucosidase activity and protein.

Step 4: Affinity Chromatography on Superdex 200—Peak fractionsfrom the phenyl-Sepharose elution were concentrated to 10 ml in a50-ml stirred ultrafiltration cell (Amicon, Inc.) using a polyethersulfonemembrane (nominal molecular weight limit of 30,000; Millipore Corp.).The sample was then applied to a Superdex 200 column (2.6 � 65 cm)pre-equilibrated with buffer G at 2.5 ml/min. Fractions of 6 ml werecollected and assayed for acid �-glucosidase activity and protein.

Step 5: Superose 6—A peak fraction from the Superdex 200 elutionwas applied to a Superose 6 column (10 � 30 cm) equilibrated withbuffer G at 2 ml/min. Fractions of 2 ml were collected and assayed foracid �-glucosidase activity and protein.

Recombinant Human GAA (rhGAA) Precursor and Intermediate

The rhGAA precursor and intermediate were produced using Chi-nese hamster ovary (CHO) cells transfected with a vector containing afull-length human GAA cDNA following the method described previ-ously (22). rhGAA was purified from the CHO-conditioned medium asfollows. The CHO cell culture harvest was clarified by sequential directflow filtration using a polyethersulfone pre-filter, followed by a What-man Polycap 75TC 0.2-�m PES membrane filter. The clarified harvest

Acid �-Glucosidase Consists of Four Peptides 6781

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

was concentrated 10-fold by ultrafiltration with a 30-kDa spiral woundultrafiltration membrane (S10Y30, Millipore Corp.). The concentratewas diafiltered into 25 mM sodium phosphate buffer (pH 7.0). Thediafiltered harvest was loaded onto a Q-Sepharose column equilibratedwith diafiltration buffer at 2.5 mg of rhGAA/ml of resin. The columnwas washed with equilibration buffer, followed by 50 mM sodium ace-tate buffer (pH 5.0), eluting some of the mature forms of GAA. Theenriched precursor form of rhGAA was eluted with 25 mM sodiumphosphate and 0.17 M sodium chloride (pH 7.0). The Q-Sepharose eluatewas adjusted to 0.75 M ammonium sulfate and loaded onto a butyl-Toyopearl 650C column equilibrated with 25 mM sodium phosphate and0.75 M ammonium sulfate (pH 6.5) at 8 mg of GAA/ml of resin. Thecolumn was washed with equilibration buffer, and the rhGAA proteinwas eluted with 15 mM sodium phosphate buffer (pH 6.5).

Preparation of Recombinant Human GlcNAc-phosphotransferase

A detailed description of the preparation of GlcNAc-phosphotrans-ferase will be reported elsewhere.2 Briefly, a production cell line secret-ing recombinant soluble human GlcNAc-phosphotransferase was gen-erated by transfection of CHO cells with a “double gene” expressionvector in pEE14.1/pEE6.1 (Lonza Biologics, Portsmouth, NH) contain-ing modified human GlcNAc-phosphotransferase �/�-subunit se-quences and the native �-subunit sequence. The �/�-subunit cDNAcontained four modifications of the native sequence. First, the amino-terminal signal anchor sequence was replaced with a signal peptidederived from the mouse immunoglobulin � light chain. Second, thecarboxyl-terminal transmembrane sequence was deleted by truncationfollowing amino acid 1209. Third, the sequence at the �/�-cleavage site(amino acids 924–929) was replaced with the sequence RARYKR, in-troducing a recognition site for the processing enzyme furin. Fourth, a12-amino acid epitope sequence recognized by monoclonal antibodyHPC4 was added following the signal peptide cleavage site. The �-sub-unit sequence was unmodified from the native sequence.

To prepare recombinant soluble GlcNAc-phosphotransferase, theproduction cell line was cultured in Dulbecco’s modified Eagle’s mediumcontaining 10% fetal bovine serum in roller bottles incubated at 37 °C ina 5% CO2 atmosphere. Conditioned medium was chromatographed on acolumn of monoclonal antibody HPC4 immobilized at 5 mg/ml on NHS-activated Sepharose 4 fast flow (Amersham Biosciences). Soluble Glc-NAc-phosphotransferase was eluted with 100 mM NaCl and 50 mM

Tris-HCl (pH 7.2) containing 5 mM EGTA and 15 mM MgCl2. Followingelution and buffer exchange into 150 mM NaCl and 50 mM NaAc (pH6.5) containing 15 mM MgCl2, GlcNAc-phosphotransferase was storedat �80 °C until used. The purified GlcNAc-phosphotransferase had aspecific activity of �10 �mol/mg/h. A detailed description has beenpublished (32).

Purification of Highly Phosphorylated rhGAA (HP-rhGAA)

A detailed method for the production of HP-rhGAA will be describedelsewhere.2 Briefly, rhGAA was expressed in CHO cells grown in thepresence of kifunensine at 1 mg/liter. Kifunensine selectively inhibits�-mannosidase I and causes the accumulation of Man7–9-GlcNAc2 oli-gosaccharides on glycoproteins. A preparation containing the 110-, 95-,and 76-kDa forms was purified from conditioned medium, and theglycans containing terminal mannose were phosphorylated by treat-ment with GlcNAc-phosphotransferase. The transfer of GlcNAc phos-phate to terminal mannoses was performed at a GlcNAc-phosphotrans-ferase to GAA ratio of 1:10 (w/w). UDP-GlcNAc was added to thereaction at 3 mM. The reaction was followed by the addition of 1% (w/w)N-acetylglucosamine-1-phosphodiester �-N-acetylglucosaminidase (un-covering enzyme). The uncovering enzyme removes the GlcNAc to givemannose 6-phosphate. The 95/76-kDa intermediates were separatedfrom the 110-kDa precursor by Superdex 200 affinity chromatography.

Western Blot Analysis

Samples were boiled in 4� sample buffer (NuPAGE, Invitrogen),applied to a 4–20% Tris/glycine/SDS-polyacrylamide precast gel(Novex, Invitrogen), and run at 180 V for �60 min in 25 mM Tris and0.2 M glycine. Proteins were transferred to polyvinylidene difluoride(PVDF) membrane (Pierce) with a semidry blot apparatus (Bio-Rad) inbuffer H at 20 V for 60 min. The membrane was blocked in 3% nonfatdry milk in buffer I for 30 min and then incubated with the designatedantibody at 1 �g/ml in 3% milk in buffer I for 1 h at room temperature.The membrane was washed five times with buffer I, and then thehorseradish peroxidase-conjugated anti-mouse, anti-goat, or anti-rabbit

secondary antibody was applied at 1 �g/ml in 3% nonfat dry milk inbuffer I for 1 h at room temperature. The membrane was washed fourtimes with buffer I and then four times with buffer I without Tween 20.Acid �-glucosidase was visualized with the SuperSignal detectionkit (Pierce).

Antibodies

Goat polyclonal antibody to human GAA was produced by FerrellFarms (McLoud, OK). Goats were immunized with highly purifiedplacental GAA that had been mixed with Freund’s adjuvant. Rabbitanti-peptide polyclonal antibodies to GAA were produced by Washing-ton Biotechnology. Rabbits were immunized with synthesized peptidescoupled to keyhole limpet hemocyanin. The sequence for each peptidewas derived from sequence within each GAA fragment as follows: anti-GAA-(57–74) (QQGASRPGPRDAQAHPGR), anti-GAA-(78–94) (VPTQ-CDVPPNSRFDCAP), and anti-GAA-(183–200) (IKDPANRRYEVPLET-PRV). Monoclonal antibody to GAA (GAA1) was prepared by immuniz-ing mice with highly purified placental GAA that had been mixed withFreund’s adjuvant. Isolation of the GAA1 hybridoma was performedfollowing standard procedures.

Reduction, Alkylation, and Tryptic Digestion of Proteins

Fifty �g of protein samples was denatured and reduced with 150 �lof 6 M guanidine hydrochloride and 0.1 M Tris (pH 8.5) and 2.5 �l of 2 M

dithiothreitol. The samples were overlaid with nitrogen and incubatedat 55 °C for 1 h in darkness. After cooling, 12.5 �l of 10% 4-vinylpyri-dine, 6 M guanidine hydrochloride, and 0.1 M Tris (pH 8.5) was added tothe protein samples. The samples were then overlaid with nitrogen andincubated at room temperature for 2 h in darkness. The reaction wasquenched with 12.5 �l of 2 M dithiothreitol. After reduction and alky-lation, the samples were dialyzed overnight with a Slide-A-Lyzer cas-sette (Mr 10,000 cutoff; Pierce) against 50 mM Tris (pH 8.5). Trypsin wasadded to the protein sample at a ratio of 1:50 (w/w). After an 18-hincubation at 37 °C, the digestion was quenched with 0.1% trifluoro-acetic acid and stored at �20 °C for future analysis.

Matrix-assisted Laser Desorption Ionization Time-of-Flight MassSpectrometry (MALDI-TOF-MS) Analysis of Reduced and

Nonreduced Protein Samples

Protein samples were mixed at a 1:1 ratio with saturated sinapinicacid matrix solution (50% of 0.1% trifluoroacetic acid and H2O and 50%acetonitrile), and 1 �l of the mixture was applied to the sample target.The data were acquired in positive linear mode on a Voyager DE PROMALDI-TOF mass spectrometer (Applied Biosystems). Singly and dou-bly charged molecular ions of bovine serum albumin (Sigma) were usedto calibrate for the mass range of 20–140 kDa. The singly chargedmolecular ions of insulin and apomyoglobin from Calibration Mixture 3(Applied Biosystems) were used to calibrate for the mass range of2–20 kDa.

Capillary Liquid Chromatography/Mass Spectrometry (LC/MS)Analysis of Reduced and Nonreduced Protein Samples

Capillary LC/MS was performed on a QSTAR QQ-TOF mass spec-trometer (Applied Biosystems) interfaced with an Ultimate capillaryLC system (LC Packings). Protein separations were performed on aVydac C4 reverse-phase column (320 �m �50 cm; Microtech Scientific)with a mobile phase of 1% formic acid in water (A) and acetonitrile (B).The flow rate was 4 �l/min. The mass spectra were acquired in thepositive mode at a range of m/z 800–3000. Protein reduction was per-formed by adding 1 �l of 10% �-mercaptoethanol to every 10 �l ofprotein sample. The samples were sonicated for 25 min and heated at100 °C for 1 min before MS analysis.

Capillary LC/Tandem Mass Spectrometry (MS/MS) PeptideMapping of Protein Tryptic Digests

On-line capillary LC/MS/MS was performed on both Esquire LC(Bruker) and QSTAR QQ-TOF systems. Proteins were separated on aVydac C18 column (320 �m � 15 cm; Microtech Scientific) using thechromatography conditions described above. On both instruments, thetwo most intensive peaks with m/z �350 were chosen for on-line MS/MS.

Amino-terminal Microsequencing

Samples of reduced and nonreduced rhGAA and placental GAA wereapplied to a 4–20 or 6% Tris/glycine/SDS-polyacrylamide gel (Invitro-gen). Electrophoresis was performed at 150 V for 1.2 h. The proteinswere transferred to a PVDF membrane using 10 mM CAPS (pH 11.0).Transfer was performed for 1 h at 100 V. Protein bands were visualized2 M. Kudo and W. M. Canfield, manuscript in preparation.

Acid �-Glucosidase Consists of Four Peptides6782

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

with Coomassie Blue. Fifteen cycles of automated Edman chemistrywere performed on protein bands excised from PVDF electroblots. Ami-no-terminal sequencing was performed with the Model 492 ProciseProtein Sequencer (Applied Biosystems) using the preprogrammedpulsed-liquid PVDF method.

Affinity Chromatography of GAA

Twenty �g of purified human placental GAA and rhGAA was ana-lyzed by affinity chromatography using a Superdex 75 HR 10/30 column(Amersham Biosciences). GAA is retarded because of its affinity for thedextran backbone of the Superdex matrix. The column was equilibratedwith 20 mM NaHPO4 (pH 6.0) and 200 mM NaCl at a flow rate of 0.5ml/min. Peaks were detected at 215 and 280 nm with a Model 1100HPLC system equipped with a diode array detector (Agilent). Thecolumn was calibrated with Bio-Rad gel filtration standard prior tosample analysis. Fractions were collected at 2-min intervals and ana-lyzed by SDS-PAGE.

Fibroblast Uptake of HP-rhGAA

For each time point, �0.5 million Pompe’s disease fibroblasts(GM00248) in Dulbecco’s modified Eagle’s medium plus 10% fetalbovine serum were incubated with 50 nM highly phosphorylated GAAprepared as described above. At designated time points, the cells wereremoved and washed five times with phosphate-buffered saline andstored at �80 °C. At 24 h, the cells were washed, and fresh mediumthat did not contain GAA was added. After the final time point, all cellpellets were thawed and lysed simultaneously with 0.25% TritonX-100. Cellular debris was pelleted, and Western blot analysis wasperformed on supernatants from each time point with several differ-ent antibodies to GAA. The blot was developed as described under“Western blot analysis.”

RESULTS

Placental GAA Contains Low Molecular Mass PolypeptidesDerived from GAA—GAA was purified from human placenta byconcanavalin A-Sepharose, phenyl-Sepharose, and Superdex200 chromatography. The final step was affinity chromatogra-phy, which takes advantage of the retardation of GAA becauseof its affinity for the dextran backbone of the Superdex matrix(11, 14, 23). Analysis of purified human placental GAA by silverstaining/SDS-PAGE demonstrated two consistent low molecu-lar mass polypeptides with apparent masses of 19 and 10 kDain addition to the expected 76- and 70-kDa bands (data notshown). Amino-terminal sequence analysis of PVDF-trans-ferred protein indicated that the amino termini of the 19- and10-kDa forms are APREPAIHSEGQ and MGQPXXFFPP, re-spectively. A data base search using the BLASTP algorithm(24) indicated that the sequences of the 19- and 10-kDapolypeptides were from human GAA and correspond to aminoacids 792–803 and 122–131, respectively.

Since processed GAA is significantly retarded on Superdex200, it was uncertain if the apparent coelution of the 19- and10-kDa polypeptides with the 76- and 70-kDa proteins indi-cated a physical association. To provide evidence that the 19-and 10-kDa polypeptides were physically associated with the76- and 70-kDa polypeptides, human placental GAA was chro-matographed on Superose 6 after Superdex 200. Unlike Super-dex 200, Superose 6 does not have a dextran backbone and doesnot function as an affinity matrix for GAA. GAA eluted as asymmetric peak with an apparent molecular mass of 75,000 Da(Fig. 1A) as reported previously (14). As shown in Fig. 1B,analysis of the Superose 6 fractions by SDS-PAGE demon-strated that the 19- and 10-kDa polypeptides coeluted with the76- and 70-kDa proteins at an apparent molecular mass of75,000 Da, suggesting that they are physically associated.

Characterization of the rhGAA Precursor and Processing In-termediates—It has been shown previously that recombinantCHO cells can be engineered to secrete the 110-kDa precursorform of human GAA into tissue culture medium (25). In addi-tion to the 110-kDa precursor form, recombinant CHO cellsalso release proteolytically processed forms of GAA into the

medium by an undetermined mechanism. In Fig. 2A, the rh-GAA precursor (lane 2) and processing intermediates (lane 3)and mature GAA from placenta (lane 4) are compared by SDS-PAGE. The intermediate fraction contains bands of �108, 95,and 76 kDa. The availability of these intermediates provided aunique opportunity to study the processing/activation of GAA.The preparations shown in Fig. 2A were analyzed by affinitychromatography on Superdex 200 (Fig. 2B). In each case, vir-tually all the protein was derived from GAA. Inspection of thechromatograms demonstrated that the 110-kDa precursoreluted earlier than placenta-derived 76/70-kDa mature GAAand that the intermediate preparation contained both earlyand late eluting material. SDS-PAGE analysis of the late elut-ing material revealed that the 76- and 70-kDa forms were notseparated by affinity chromatography (data not shown). Thelater elution of the processed forms suggests a higher affinityfor the matrix, which is consistent with previous reports thatthe 76/70-kDa species has a Km for glycogen that is 7–10-foldlower than that of the 110-kDa precursor (3, 17).

Characterization of the rhGAA Precursor and Processing In-termediates by Mass Spectroscopy and Amino-terminal Micro-

FIG. 1. Chromatography of human placental GAA on Superose6 and analysis by SDS-PAGE. GAA purified as described under“Experimental Procedures” was applied to a Superose 6 gel filtrationcolumn. A, each fraction was assayed for protein (A280) and GAA activ-ity. Arrows denote elution volumes of gel filtration standards. B, shownis a silver-stained reducing SDS-polyacrylamide gel (4–20% acrylam-ide) of the load and indicated fractions.

Acid �-Glucosidase Consists of Four Peptides 6783

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

sequencing—To fully characterize GAA processing, we utilizeda combination of MALDI-TOF-MS, amino-terminal microse-quencing, capillary LC/MS, and capillary LC/MS/MS of trypticpeptide maps. This allowed us to determine the amino andcarboxyl termini for each polypeptide and the structure of theN-glycans. The results of these analyses are summarized inTable I.

When the recombinant GAA precursor was examined byMALDI-TOF-MS, a molecular mass of 111,600 Da was deter-mined. Amino-terminal microsequencing identified amino acid57 as the amino terminus. Tryptic digestion followed by re-verse-phase HPLC and peptide identification by mass spectros-copy identified amino acid 952 as the carboxyl terminus. Theglycosylation is heterogeneous and estimated at �2000 Da/glycan. These results for the precursor (mass of 111,600 Da) are

in reasonable agreement with the calculated molecular mass ofthe polypeptide and carbohydrate of 113,372 Da.

When the processing intermediates containing predomi-nantly 95- and 76-kDa species upon reducing SDS-PAGE (Fig.2A, lane 3) were analyzed unreduced by MALDI-TOF-MS, ma-jor species with masses of �103.5 and �83.6 kDa were ob-served. After reduction, masses of �99.8 and �79.7 kDa wereobserved, indicating a consistent mass shift of �3.9 kDa, con-sistent with an interchain disulfide bond.

Analysis of human placental GAA by reducing SDS-PAGE(Fig. 1) identified polypeptides with apparent molecularmasses of �95, 76, 70, 19, and 10 kDa. MALDI-TOF-MS anal-ysis of reduced placental GAA determined that the correspond-ing masses were �99.8, �79.7, 69.3, 19.4, and 10.3 kDa.MALDI-TOF-MS analysis of nonreduced placental GAA re-

FIG. 2. Characterization of pro-cessed GAA forms. The rhGAA precur-sor (P), recombinant GAA intermediates(I), and mature human placental GAA (M)were purified as described under “Exper-imental Procedures.” The precursor andintermediates were obtained from the pu-rification of rhGAA. A, shown is a silver-stained reducing SDS-polyacrylamide gel(4–20%) of the GAA precursor, intermedi-ate, and mature form. B, the GAA precur-sor and processed forms were analyzed bySuperdex 75 HR 10/30 affinity chroma-tography. The rhGAA precursor eluted asa single peak (upper panel). The recombi-nant GAA intermediate from the purifica-tion of rhGAA eluted as two species, withthe first peak eluting at the same time asthe GAA precursor (middle panel). Ma-ture human placental GAA eluted as amuch broader and later eluting peak(lower panel). mAU, milli-absorbanceunits.

Acid �-Glucosidase Consists of Four Peptides6784

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

vealed polypeptides with masses of �103.5, �83.6, 69.3, 19.4,and 14.3 kDa. These data show the �103.5-, �83.6-, and 14.3-kDa polypeptides all shifted by �3.9 kDa after disulfide bondreduction, again suggesting the presence of an interchain di-sulfide bond. This experiment was also repeated by LC/MSusing a QSTAR QQ-TOF mass spectrometer. The �3.9-kDamass shift seen in all reduced samples correlated with a newpeak in the spectra with a mass of 3927 Da.

The GAA sequences found in each of the identified polypep-tides were determined by a combination of microsequencing,tryptic digestion, capillary LC/MS, and LC/MS/MS. The 3.9-kDa polypeptide with a determined mass of 3927 Da corre-spond to amino acids 78–113. The 10.3-kDa polypeptide wasidentified as amino acids 122–199. The 70-kDa polypeptidecorresponds to amino acids 204–782. The 19.4-kDa polypeptidewas identified as amino acids 792–952. These results are sum-marized in Table I.

Modification of N-Glycans during GAA Maturation—TheGAA precursor contains seven consensus sites for N-glycosyla-tion that are all utilized. In recombinant GAA, on average,these sites are occupied by five to six complex-type glycans andone to two high mannose-type glycans. In contrast, GAA ex-pressed in human placenta contains predominantly high man-nose-type glycans (18). In either case, the newly synthesizedglycans each contain �9–11 hexoses/hexosamines and have anaverage molecular mass of �2000 Da.

By mass spectroscopy, the recombinant precursor contains�14,000 Da of carbohydrate, indicating an average of �10monosaccharides/glycan (Table I). The 95-kDa intermediatecontains 7300 Da of carbohydrate at these same seven glyco-sylation sites, indicating that the glycans have been truncatedto an average of approximately five monosaccharides, consist-ent with an average core structure composed of two GlcNAcand three mannose residues, which can be derived from eithercomplex-type or high mannose-type glycans. These results sug-gest that the conversion of the 110-kDa precursor to the 95-kDaintermediate has occurred in a subcellular compartment con-taining a variety of glycosidases capable of cleaving a complex-type glycan. This pattern of finding only highly truncated gly-cans on the processed GAA is repeated for all the componentsof the 76- and 70-kDa complexes (Table I). For the fragmentscontaining one, two, or four glycosylation sites, the compositioncan be directly confirmed to contain two HexNAc and threehexose residues/site, consistent with the structure GlcNAc2-Man3, which may also contain fucose. The complete absence ofproteolytically processed intermediates whose glycans have notbeen truncated strongly suggests that this processing occurs in

a compartment containing both proteases and glycosidases,most likely in the lysosome or late endosome.

Identification of Subunits with Anti-peptide Antibodies—Toindependently evaluate the model for GAA maturation derivedfrom mass spectroscopy, polyclonal, monoclonal, and anti-peptide antibodies were generated against peptide sequencesQQGASRPGPRDAQAHPGR (amino acids 57–74), VPTQCD-VPPNSRFDCAP (amino acids 78–94), and IKDPANRRYEV-PLETPRV (amino acids 183–200) as described under “Experi-mental Procedures.” The resulting affinity-purified antiseraare referred to as anti-GAA-(57–74), anti-GAA-(78–94), andanti-GAA-(183–200), respectively. Additionally, a goat poly-clonal antibody to placental GAA (goat anti-GAA) and a mono-clonal antibody to the 70-kDa polypeptide (GAA1) were gener-ated as described under “Experimental Procedures.”

As shown in Fig. 3, Western blot analysis was performedwith a variety of antibodies on a blot containing the 110-kDaHP-rhGAA precursor (P), an intermediate preparation (I) of95/76-kDa forms from the purification of HP-rhGAA, and the76/70-kDa mature species (M) purified from human placenta.HP-rhGAA is a highly phosphorylated form of rhGAA derivedfrom in vitro phosphorylation as described under “Experimen-tal Procedures.” Monoclonal antibody GAA1 recognized the110-, 95-, 76-, and 70-kDa species (Fig. 3A, lanes 1–3). The blotwas then probed again with goat anti-GAA antibody, whichdemonstrated the presence of the 19-kDa fragment (Fig. 3A,lanes 4–6). The 19.4-kDa fragment exists as a doublet in therecombinant intermediate preparation due to differences inglycosylation (data not shown). The blot was then probed againwith anti-GAA-(183–200) antibody (Fig. 3A, lanes 7–9). Thisrevealed the presence of a 10-kDa polypeptide in the interme-diate and mature GAA samples (Fig. 3A, lane 8 and 9). Theseantibodies were helpful to understand the order of the cleavageevents as shown in Fig. 3B.

Anti-GAA-(57–74) antibody identified the 110-kDa precursorenzyme, which is consistent with the MS and amino-terminalmicrosequence analysis of this preparation, but the antibodydid not recognize the intermediate preparation, containing the95/76-kDa forms, or the mature preparation, containing the76/70-kDa mature forms (Fig. 3B, lanes 1–3). Since amino acids57–74 were not detected in the intermediate preparation, it canbe concluded that the cleavage to release this sequence occursearly in the maturation process. It has not been possible toidentify this small peptide by Western blotting using eithernitrocellulose or PVDF membranes. It is possible that thissmall peptide does not bind tightly to the membrane; however,it is more likely lost during processing since the peptide was

TABLE ISummary of the polypeptides and glycans present for each species of GAA

The first column lists the apparent molecular mass of the major polypeptide that represents the common designation for this species. In thesubsequent columns, each polypeptide that is a component of the GAA complex is listed on a separate line. Theoretical, calculated, and observedmasses are as indicated.

Apparentmolecularmass by

SDS-PAGE(reducing)

Polypeptidespresent(aminoacids)

No. ofglycosylation

sitesObserved carbohydrate

structures by MSMolecularmass of

carbohydrate

Theoreticalmolecularmass of

polypeptide

Calculatedmolecular massof polypeptide

� carbohydrate

Observedmolecular massof polypeptide

� carbohydrateby MS

Calculatedmolecularmass ofcomplex

Observedmolecular massof complex by

MS(nondenaturing/

nonreducing)

Source

kDa Da Da Da Da Da Da

110 57–952 7 Heterogeneous glycoforms �14,000 99,372 �113,372 111,600 �113,372 111,600 Recombinant95 78–113 0 None None 3927 3927 3927 103,727 103,551 Recombinant

122–952 7 Heterogeneous glycoforms �7300 92,459 �99,780 99,80076 78–113 0 None None 3927 3927 3927 103,074 NDa Placenta

122–782 5 Heterogeneous glycoforms �6000 74,277 �80,277 79,700792–952 2 4 HexNAc, 6 Hex, 2 Fuc 2078 17,370 19,448 19,447

70 78–113 0 None None 3927 3927 3927 103,094 ND Placenta122–199 1 2 HexNac, 5 Hex 1216 9177 10,393 10,392204–782 4 8 HexNac, 18 Hex, 1 Fuc 4686 84,639 69,325 69,328792–952 2 4 HexNac, 6 Hex, 2 Fuc 2078 17,370 19,448 19,447

a ND, not determined.

Acid �-Glucosidase Consists of Four Peptides 6785

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

not detected by tryptic digestion/MS/MS.Monoclonal antibody GAA1 (Fig. 3B, lane 4–6) detected the

110-, 95-, 76-, and 70-kDa species, whereas anti-GAA-(183–200) antibody (lane 9) detected only the 110-, 95-, and 76-kDapolypeptides and did not detect the 70-kDa species. A 10.3-kDafragment was detected by anti-GAA-(183–200) antibody that isnot recognized by the monoclonal antibody GAA1. The cleavageevent that converted the 76-kDa enzyme to the 70-kDa formresulted in the formation of a 10-kDa fragment, which wasshown in Fig. 1 to co-chromatograph with GAA activity. Sincethe 10.3-kDa fragment was only faintly observed in the 95/76-kDa intermediate preparation, containing predominantly 95-and 76-kDa forms, we conclude that the formation of the 10.3-kDa polypeptide occurs late in the maturation of GAA.

The presence of the 3.9-kDa polypeptide, which was origi-nally identified by mass spectroscopy, was confirmed by West-

ern blotting in Fig. 4. The blots contain the 110-kDa HP-rhGAAprecursor (P) (lanes 1 and 4), a preparation of 95/76-kDa inter-mediate forms (I) from the purification of rhGAA (lanes 2 and5), and the 76/70-kDa mature species (M) purified from humanplacenta (lanes 3 and 6). The Western blot shown in Fig. 4Awas probed with anti-GAA-(57–74) antibody. The intermediatepreparation from the purification of rhGAA was enriched forthe 95/76-kDa intermediates; however, the preparation con-tained some precursor (Fig. 4A, lane 2), which migrated slowerthan the precursor in lane 1. The difference in mobility is dueto glycosylation differences between the HP-rhGAA and rhGAAprecursors (data not shown).

The blot shown in Fig. 4A was then probed with anti-peptideantibody for the 3.9-kDa polypeptide (amino acids 78–94) asshown in Fig. 4B. The antibody detected the 95- and 76-kDapolypeptides in the absence of �-mercaptoethanol (Fig. 4B,

FIG. 3. Identification of the 19.4- and 10.3-kDa polypeptides byWestern blotting. The rhGAA precursor (P), rhGAA intermediates (I),and mature human placental GAA (M) were purified as described under“Experimental Procedures.” The precursor and intermediates were ob-tained from the purification of HP-rhGAA. All panels are reducingSDS-polyacrylamide gels (4–20% acrylamide) subjected to Westernblotting. A, the GAA precursor, rhGAA intermediates, and mature GAAwere probed with anti-GAA-(57–74) antibody (lanes 1–3), monoclonalantibody GAA1 (lanes 4–6), and anti-GAA-(183–200) antibody (lanes7–9). B, a blot of the GAA precursor, rhGAA intermediates, and matureGAA was probed with monoclonal antibody GAA1 (lanes 1–3). The sameblot was then probed with an affinity-purified goat (gt) polyclonal an-tibody raised against GAA purified from human placenta (lanes 4–6).The blot was then probed with anti-GAA-(183–200) antibody (lanes7–9). All blots were developed as described under “ExperimentalProcedures.”

FIG. 4. Identification of a disulfide-bound 3.9-kDa polypeptide.All panels are Western blots of reducing and nonreducing SDS-poly-acrylamide gels loaded with the same samples as described in thelegend to Fig. 3. A and B are from a 6% acrylamide gel, and C is froma 4–20% acrylamide gel. A, the GAA precursor (P), intermediates (I),and mature GAA (M) were probed with anti-GAA-(57–74) antibody(lanes 1–3) and monoclonal antibody GAA1 (lanes 4–6). B, the blotshown in A was probed with anti-GAA-(78–94) antibody (lanes 1–6). C,the blot from a 4–20% SDS-polyacrylamide gel was probed with anti-GAA-(78–94) antibody. All blots were developed as described under“Experimental Procedures.” �ME, �-mercaptoethanol.

Acid �-Glucosidase Consists of Four Peptides6786

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

lanes 5 and 6). The samples were run on a 4–20% SDS-poly-acrylamide gel as shown in Fig. 4C. We were not able to detectthe 3.9-kDa fragment when run under reducing conditions;however, when the samples were not reduced, an �14-kDapolypeptide was observed (Fig. 4C, lane 6). The �14 kDapolypeptide is composed of the 3.9-kDa polypeptide disulfide-bonded to the previously observed 10-kDa polypeptide.

GAA Exists as a Complex in Other Organisms—To determinewhether cleaved polypeptides are associated with GAA in othertissues, we purified endogenous hamster GAA from CHO cells.To our knowledge, hamster GAA had never before been isolatedfrom a cultured cell line. The purification is described in detailunder “Experimental Procedures.” Silver staining/SDS-PAGEanalysis of hamster GAA revealed a predominant protein withan apparent molecular mass of 70 kDa (Fig. 4). The 70-kDaprotein was isolated by reverse-phase chromatography, and theamino terminus was microsequenced. The amino-terminal se-quence determined (APSQLYSVEFSEEPF) is highly homolo-gous to the amino terminus for the human 70-kDa matureenzyme from amino acids 204 to 218 (APSPLYSVEFSEEPF).

The 110-kDa rhGAA precursor, GAA purified from placenta,and hamster GAA purified from CHO cells were compared bySDS-PAGE (Fig. 5A, lanes 1–3). Unlike placental GAA, only the70-kDa species was present in the highly purified preparationof hamster GAA from CHO cells. Western blot analysis wasperformed on hamster GAA with anti-GAA-(183–200) antibody(Fig. 5B, lanes 1 and 3) and goat anti-GAA antibody (lanes 2and 3). The blots show that the 19- and 10-kDa fragments werepresent in a highly purified preparation of hamster GAA fromCHO cells. The presence of these subunits in highly purifiedGAA preparations from human placenta and CHO cells dem-onstrates that the formation and association of these subunitsare not tissue- or species-specific.

The 19-kDa polypeptide in CHO preparations appears as adoublet. The cDNA for human GAA encodes two glycosylationsites in the 19-kDa fragment. To determine whether carbohy-drate differences are responsible for the observed doublet, hu-man placental and hamster GAA were denatured and digestedwith endoglycosidase H and peptide N-glycosidase F. PeptideN-glycosidase F is an amidase that cleaves between the inner-most GlcNAc and asparagine residues of high mannose-, hy-brid-, and complex-type oligosaccharides from N-linked glyco-proteins (26). Endoglycosidase H is a glycosidase that cleavesthe chitobiose core of high mannose-type and some hybrid-typeoligosaccharides from N-linked glycoproteins (26). Endoglyco-sidase H digestion did not have any affect on the 19.4-kDapolypeptide from human placenta or hamster GAA (Fig. 5C,lanes 1, 2, and 6). This suggests that the carbohydrate is nothigh mannose-type. However, peptide N-glycosidase F diges-tion decreased the apparent size of the 19.4-kDa polypeptidefor both human placental and hamster GAA (Fig. 5C, lanes 3, 4,7, and 8). The 19-kDa doublet in the hamster preparation wasreduced to a single band, which suggests that glycosylationdifferences are responsible for the doublet.

Proteolytic Processing of the 110-kDa Precursor after Uptakein Fibroblasts—To directly demonstrate that the proteolyticprocessing of GAA in the cell proceeds via the pathway de-scribed, we utilized mannose 6-phosphate receptor-mediatedendocytosis of the GAA precursor and a GAA-deficient fibro-blast cell line (GM00248) derived from a Pompe’s disease pa-tient. To increase the amount of GAA internalized, HP-rhGAAwas used. The protein sequence of HP-rhGAA is identical tothat of the rhGAA precursor, but the structure differs in thatthe N-glycans of HP-rhGAA contain 7–11 mannose 6-phos-phate groups/molecule of GAA. This was determined by LC/MS/MS. The evidence for this will be described elsewhere.2 At

FIG. 5. Hamster acid �-glucosidase is a multisubunit complex.rhGAA, human placental GAA, and hamster GAA were purified asdescribed under “Experimental Procedures.” A, silver-stained reducingSDS-polyacrylamide gel (10% acrylamide) of rhGAA (lane 1), humanplacental GAA (lane 2), and hamster GAA (lane 3). B, Western blot of areducing SDS-polyacrylamide gel (4–20% acrylamide) of hamster GAA.Lane 1, blot probed with anti-GAA-(183–200) antibody; lane 2, blot probedwith affinity-purified goat (gt) anti-GAA antibody raised against GAApurified from human placenta; lane 3, blot probed with antibodies used inlanes 1 and 2. C, Western blot of a reducing SDS-polyacrylamide gel(4–20% acrylamide) of human placental GAA undigested (lanes 1 and 3)and digested with endoglycosidase H (Endo H; lane 2) and peptide N-glycosidase F (PNGase, lane 4) and hamster GAA undigested (lanes 5 and7) and digested with endoglycosidase H (lane 6) and peptide N-glycosidaseF (lane 8) with affinity-purified goat anti-GAA antibody. All blots weredeveloped as described under “Experimental Procedures.”

Acid �-Glucosidase Consists of Four Peptides 6787

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

the designated time points, cell lysates were prepared andanalyzed by Western blotting using anti-GAA-(57–74) antibody(Fig. 6A), a monoclonal antibody to the 70-kDa polypeptide(Fig. 6B), and anti-GAA-(183–200) antibody (Fig. 6C). In Fig.6A, only the GAA precursor standard was identified by anti-GAA-(57–74) antibody. Fragments containing this sequencewere not identified in any of the lysate-derived samples. This isconsistent with rapid release of this peptide upon internaliza-tion. In Fig. 6B, the same blot probed with monoclonal antibodyGAA1 shows the conversion of the 110-kDa precursor to thetransient 95-kDa intermediate, followed by slower processingto the 76- and 70-kDa forms. In Fig. 6C, a Western blot of thelysates was probed with anti-GAA-(183–200) antibody, demon-strating that the appearance of the 10.3-kDa fragment corre-lates with the conversion of the 76-kDa form to the 70-kDaform. These results suggest that processing takes place by anordered and sequential series of steps.

DISCUSSION

It is well known that GAA is synthesized as a 110-kDaprecursor of 952 amino acids, yet it exists in lysosomes as majorspecies of 76 and 70 kDa. It is generally agreed that the con-version from the 110-kDa form to the 76- and 70-kDa forms isa result of proteolytic processing at both the amino and car-boxyl termini. However, the structures of the 76- and 70-kDamature forms of the enzyme have not been determined. Fur-thermore, the details and sequence of the processing steps areincomplete. In this study, we have demonstrated that fullyprocessed GAA consists of four associated polypeptides all de-rived from the GAA precursor. The ability to determine thestructure of each fragment using sensitive mass spectroscopic

techniques allows us to propose the model for GAA maturationthat is shown in Fig. 7 and supported by Table I.

The GAA predicted primary translation product of 952 aminoacids containing seven N-glycans is probably present only tran-siently and was not detected. The first proteolytic processingstep is likely cleavage between amino acids 28 and 29 by signalpeptidase; again, the product of this cleavage was not detected.The first precursor identified in this study contains amino acids57–952, apparently resulting from a proteolytic cleavage be-tween amino acids 56 and 57 (Fig. 7 and Table I). A similarprecursor beginning at amino acid 70 has been identified inhuman urine (10). The first intermediate identified, designatedintermediate 1, consists of a 95-kDa polypeptide (amino acids(aa) 122–952) covalently linked via a disulfide bond to a 3.9-kDa polypeptide (aa 78–113). Multiple cleavages are requiredto generate these fragments. The identification of the 3.9-kDadisulfide-linked polypeptide is entirely novel. Sequences 57–78and 113–122 are not found, apparently having been lost duringprocessing. The availability of anti-GAA-(57–74) antibody al-lowed confirmation that this sequence is not present in inter-mediate 1 (Fig. 3B, lane 2).

Interestingly, almost all of the reduction in molecular massgoing from the 110-kDa precursor to intermediate 1 is attrib-uted to the loss of carbohydrate due to extensive glycan trim-ming. We have not observed glycans on intermediate 1 thatwere not trimmed. This suggests that the processing that yieldsintermediate 1 takes place in a late endosome or lysosome,containing multiple glycosidases.

In the next step, the 95-kDa intermediate is proteolyticallycleaved at the carboxyl terminus by an unknown protease,

FIG. 6. Fibroblast uptake and proc-essing of HP-rhGAA. As described under“Experimental Procedures” GAA-deficientPompe’s disease fibroblasts (GM00248) wereincubated with 50 nM HP-rhGAA. At thedesignated time points, the fibroblasts werecollected and frozen at �80 °C. The cell ly-sates were probed with a panel of anti-GAAantibodies. A, Western blot of a reducingSDS-polyacrylamide gel (8% acrylamide).The blot was probed with anti-GAA-(57–74)antibody. B, Western blot of a reducing SDS-polyacrylamide gel (8% acrylamide). Theblot was probed with monoclonal antibodyGAA1. C, Western blot of a reducing SDS-polyacrylamide gel (4–20% acrylamide). Theblot was probed with anti-GAA-(183–200)antibody. P, GAA precursor; M, matureGAA.

Acid �-Glucosidase Consists of Four Peptides6788

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

FIG. 7. Model for the maturation of GAA.

FIG. 8. Multiple sequence alignment of human family 31 glucohydrolases. Shown is an alignment of human GAA, maltose isoamylase(MGAM), sucrose isomaltase (SIM), glucosidase II (G2AN), and neutral �-glucosidase C (GANC). The putative active site (WIDMNE) is highlighted(orange) along with sequences removed during proteolytic processing (green).

Acid �-Glucosidase Consists of Four Peptides 6789

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

which we have designated as protease 3. The cleaved carboxyl-terminal 19.4-kDa fragment (aa 792–952) and the 3.9-kDafragment remain associated with the 76-kDa polypeptide (aa122–781). In the final step, the 76-kDa polypeptide is thenproteolytically cleaved near the amino terminus by an un-known protease, which we have designated as protease 4, togive the final complex consisting of the following polypeptides:70 kDa (aa 204–781), 19.4 kDa (aa 792–952), 10.3 kDa (aa122–200), and 3.9 kDa (aa 78–113). Interestingly, at eachcleavage site, four to nine amino acids are not present in thefinal products (aa 114–121, 200–203, and 783–791); whetherthis is the result of two specific proteolytic steps or a singlecleavage followed by nonspecific aminopeptidase or car-boxypeptidase activity has not been determined.

Interestingly, there was no evidence that the three proteo-lytic processing steps could occur in any order except thatproposed in our model. Had these cleavages occurred in adifferent order, additional intermediates that were not ob-served would have formed. The understanding that GAA mat-uration proceeds through a series of discrete ordered steps mayalso have application as the basis for diagnostic tests for gly-cogen storage disease type II. In particular, the identification offragments with masses of 3,927, 10,392, and 19,447 Da mightbe a sensitive indicator that GAA is present and that matura-tion has proceeded successfully. Additionally, the detection ofsuch polypeptides could be used to monitor enzyme replace-ment therapy for Pompe’s disease.

The association of the proteolytically cleaved fragments ofGAA is not unique for lysosomal enzymes. The lysosomal �-man-nosidase enzyme is synthesized as a single chain precursor thatis processed into three glycopeptides of 70, 42, and 15 kDa. The70-kDa glycopeptide is further partially proteolyzed into threemore peptides that are joined by disulfide bridges (27).

The previously described GAA2 allele (D91N) (23, 28) islocated in the 3.9-kDa fragment (aa 78–113). This substitutionresults in a decreased affinity of the mature enzyme for starch,leading to the suggestion that a fragment containing the sub-stitution might be associated with the mature enzyme. Ourfindings present direct physical evidence for both the existenceof this fragment and its association with mature GAA. Exceptfor frameshifts, no disease-causing mutations have been re-ported in the 10.3-kDa fragment. Several mutations that resultin a severe phenotype have been described in the 19.4-kDafragment (aa 792–952). The in-frame deletion of one amino acid(Lys903) and the common deletion of exon 18 that results in a55-amino acid deletion from amino acids 828 to 882 both resultin the fatal infantile form of the disease (29). Other mutationsin the 19.3-kDa fragment include R854X and a frameshiftmutation occurring at Pro913, both resulting in severe pheno-types. Prior to this study, it was believed the common exon 18deletion contained the carboxyl-terminal proteolytic cleavagesite based on the belief that the carboxyl-terminal cleavageoccurs between amino acids 816 and 881 (3). We have shownthat the cleavage site is actually 36 amino acids upstream fromthe beginning of this common deletion.

The maturation process is not dependent on an active GAAenzyme. A patient with infantile type 2 glycogenosis was found tobe homozygous for a E521K substitution (30). When COS cellswere transfected with constructs containing the E521K muta-tion, the GAA was inactive, and GAA maturation was prevented.However, when Glu521 was changed to Gln, maturation of GAAwas restored, whereas there was still a loss of activity.

Since mature GAA consists of four polypeptides, this mayhave implications in treating glycogen storage disease type II.One current strategy for treating the disease is to use enzymereplacement therapy. Using GAA�/� knockout mice (12, 31),

several groups have attempted enzyme replacement therapywith precursor and mature forms of recombinant GAA. The110-kDa rhGAA precursor purified from the milk of transgenicmice and rabbits demonstrated a therapeutic effect in knockoutmice (13, 17). Analysis of the tissues revealed that the 110-kDaprecursor from rabbit and mouse milk was converted to the76-kDa lysosomal form. Neither enzyme preparation appearedto be converted significantly to the 70-kDa form. Another studyreported using the mature form of GAA (the composition wasnot described) to reverse muscle weakness in a knockout mouse(15). Our improved understanding of the processing pathwaymay help us to develop more effective second generation can-didates for enzyme replacement therapy.

The sequences of five human family 31 glucohydrolases(GAA, maltose isoamylase, sucrose isomaltase, glucosidase II,and neutral �-glucosidase C) (16) have been aligned usingClustalW (Fig. 8) (30). Since the maltase isoamylase and su-crose isomaltase sequences have undergone duplication andcontain two active sites, they have been divided into the amino-and carboxyl-terminal ends. Aligned in this fashion, the activesite (WIDMNE) and many of the cysteines have been con-served. Sequences lost from GAA during processing are indi-cated in green. It is apparent from the alignment that GAA,maltose isoamylase, and sucrose isomaltase are most closelyrelated, whereas glucosidase II contains a number of insertionsand deletions reducing the overall homology. GAA is the onlyfamily 31 enzyme that is found in the lysosome and is known toundergo proteolytic maturation/processing. Interestingly,two of the four sequences removed during processing arefound in the only two insertions in the GAA sequence relativeto the family 31 consensus. It will be interesting in the futureto determine how these sequences alter GAA structure andfunction.

Acknowledgments—We thank Sheri Rayl, Mike Brem, Dennis Burian,and the Genzyme Analytics Group for technical assistance. We thankRobert Mattaliano and Jean Elmendorf for helpful discussions andadvice. We also thank K. Jackson (Molecular Biology Resource Center,Oklahoma Center for Molecular Medicine) for protein sequencing.

REFERENCES

1. Hoefsloot, L., Hoogeveen-Westerveld, M., Kroos, M. A., van Beeumen, J.,Reuser, A. J., and Oostra, B. A. (1988) EMBO J. 7, 1697–1704

2. Hasilik, A., and Neufeld, E. F. (1980) J. Biol. Chem. 255, 4937–49453. Wisselaar, H. A., Kroos, M. A., Hermans, M. M., van Beeumen, J., and Reuser,

A. J. (1993) J. Biol. Chem. 268, 2223–22314. Reuser, A. J., Kroos, M. A., Ponne, N. J., Wolterman, R. A., Loonen, M. C.,

Busch, H. F., Visser, W. J., and Bolhuis, P. A. (1984) Exp. Cell Res. 155,178–189

5. Scheibe, R., Wenzel, K. W., and Hofmann, E. (1985) Biomed. Biochim. Acta 44,1279–1286

6. Tashiro, K., Iwamasa, T., Kato, H., Ogata, S., and Anai, M. (1986) J. Biochem.99, 693–701

7. Murray, A. K., Brown, B. I., and Brown, D. H. (1978) Arch. Biochem. Biophys.185, 511–524

8. Matsui, H., Sasaki, M., Takemasa, E., Kaneta, T., and Chiba, S. (1984) J. Bio-chem. 96, 993–1004

9. Chambers, J. P., and Williams, J. C. (1983) Enzyme 29, 109–11910. Oude Elferink, R. P., Brouwer-Kelder, E. M., Surya, I., Strijland, A., Kroos, M.,

Reuser, A. J., and Tager, J. M. (1984) Eur. J. Biochem. 139, 489–49511. Radin, N. S., Shukla, A., Shukla, G. S., and Sano, A. (1989) Biochem. J. 264,

845–84912. Raben, N., Nagaraju, K., Lee, E., Kessler, P., Byrne, B., Lee, L., LaMarca, M.,

King, C., Ward, J., Sauer, B., and Plotz, P. (1998) J. Biol. Chem. 273,19086–19092

13. Bijvoet, A. G. A., Van Hurtim, H., Kroos, M. A., Van de Kamp, E. H. M.,Schoneveld, O., Visser, P., Brakenhoff, J. P. J., Weggeman, M., van Corven,E. J., Van der Ploeg, A. T., and Reuser, A. J. J. (1999) Hum. Mol. Genet. 8,2145–2153

14. Martiniuk, F., Honig, J., and Hirschorn, R. (1984) Arch. Biochem. Biophys.231, 454–460

15. Martiniuk, F., Chen, A., Donnabella, V., Arvanitopoulos, E., Slonim, A. E.,Raben, N., Plotz, P., and Rom, W. N. (2000) Biochem. Biophys. Res. Com-mun. 276, 917–923

16. Hirschhorn, R., Huie, M. L., and Kasper, J. S. (2002) Proc. Natl. Acad. Sci.U. S. A. 99, 13642–13646

17. Bijvoet, A. G. A., Kroos, M. A., Pieper, F. R., Van der Vliet, M., De Boer, H. A.,Van der Ploeg, A. T., Verbeet, M. P., and Reuser, A. J. J. (1998) Hum. Mol.Genet. 7, 1815–1824

Acid �-Glucosidase Consists of Four Peptides6790

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

18. Mutsaers, J. H. G. M., Halbeek, H. V., Vliegenthart, J. F. G., Tager, J. M.,Reuser, A. J. J., Kroos, M. A., and Galjaard, H. (1987) Biochim. Biophys.Acta 911, 244–251

19. Hirschorn, R., and Reuser, A. J. J. in The Metabolic & Molecular Bases ofInherited Disease (Scriver, C. R., Beaudet, A. L., Sly, W. S., and Valle, D.,eds), Eigth Edition Ed., pp. 3389–3420, III vols., McGraw-Hill, New York

20. Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H.,Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk,D. C. (1985) Anal. Biochem. 150, 76–85

21. Blum, H., Beier, H., and Gross, H. (1987) Electrophoresis 93–9922. Qiu, H., Edmunds, T., Baker-Malcolm, J., Karey, K. P., Estes, S., Schwarz, C.,

Hughes, H., and Van Patten, S. M. (2003) J. Biol. Chem. 278, 32744–3275223. Van Hove, J. L., Yang, H. W., Oliver, L. M., Pennybacker, M. F., and Chen,

Y. T. (1997) Biochem. Mol. Biol. Int. 43, 613–62324. Altschul, S. F., and Gish, W. (1996) Methods Enzymol. 266, 460–48025. Van Hove, J. L. K., Yang, H. W., Wu, J. Y., Brady, R. O., and Chen, Y. T. (1996)

Proc. Natl. Acad. Sci. U. S. A. 93, 65–7026. Maley, F., Trimble, R., Tarentino, A., and Plummer, T. (1989) Anal. Biochem.

180, 195–20427. Nilssen, O., Berg, T., Riise, H. M. F., Ramachandran, U., Evjen, G., Hansen,

G. M., Malm, D., Tranebjaerg, L., and Tollersrud, O. K. (1997) Hum. Mol.Genet. 6, 717–726

28. Swallow, D., Corney, G., Harris, H., and Hirschhorn, R. (1975) Ann. Hum.Genet 38, 391–406

29. Boerkoel, C. F., Exelbert, R., Nicastri, C., Nichols, R. C., Miller, F. W., Plotz,P. H., and Raben, N. (1995) Am. J. Hum. Genet. 56, 887–897

30. Hermans, M. M. P., De Graaff, E., Kroos, M. A., Wisselaar, H. A., Willemsen,R., Oostra, B. A., and Reuser, A. J. J. (1993) Biochem. J. 289, 687–693

31. Bijvoet, A. G. A., Van de Kamp, E. H. M., Kroos, M. A., Ding, J. H., Yang, B. Z.,Visser, P., Bakker, C. E., Verbeet, M. P., Oostra, B. A., Reuser, A. J. J., andVan der Ploeg, A. T. (1998) Hum. Mol. Genet. 7, 53–62

32. Canfield, W. M. (November 4, 2003) U. S. Patent 6,642,038

Acid �-Glucosidase Consists of Four Peptides 6791

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from

CanfieldKaren L. Lee, Robert D. Cauthron, Kevin Brewer, Tim Edmunds and William M.

Rodney J. Moreland, Xiaoying Jin, X. Kate Zhang, Roger W. Decker, Karen L. Albee,Single Chain Precursor

-Glucosidase Consists of Four Different Peptides Processed from aαLysosomal Acid

doi: 10.1074/jbc.M404008200 originally published online November 1, 20042005, 280:6780-6791.J. Biol. Chem. 

  10.1074/jbc.M404008200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/280/8/6780.full.html#ref-list-1

This article cites 29 references, 8 of which can be accessed free at

by guest on Decem

ber 28, 2018http://w

ww

.jbc.org/D

ownloaded from