Proc. Natl. Acad. Sci. USA Vol. 93, pp. 65-70, January 1996 Genetics

High-level production of recombinant human lysosomal acid a-glucosidase in Chinese hamster ovary cells which targets to heart muscle and corrects glycogen accumulation in fibroblasts from patients with Pompe disease JOHAN L. K. VAN HOVE*t, HELEN W. YANG*, JER-YUARN Wu*, RoscOE 0. BRADYt, AND YUAN-TSONG CHEN* *Department of Pediatrics, Duke University Medical Center, Durham, NC 27710; and *National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892 Contributed by Roscoe 0. Brady, September 11, 1995

ABSTRACT Infantile Pompe disease is a fatal genetic transfected COS cells is taken up efficiently by cultured patient muscle disorder caused by a deficiency of acid v-glucosidase, cells through the mannose-6-phosphate receptor (3, 10-12). a glycogen-degrading lysosomal . We constructed a Because of the high abundance of these receptors in heart and plasmid containing a 5'-shortened human acid c-glucosidase muscle, bovine testis enzyme injected intravenously into mice is cDNA driven by the cytomegalovirus promoter, as well as the targeted to these tissues (13). Interspecies antigenicity requires aminoglycoside phosphotransferase and dihydrofolate reduc- the use of the human enzyme (7). The low abundance in human tase genes. Following transfection in dihydrofolate reductase- urine makes this source impractical (10). The isolation of a deficient Chinese hamster ovary cells, selection with Geneti- full-length acid a-glucosidase cDNA (1) makes production of cin, and amplification with methotrexate, a cell line producing recombinant human enzyme possible. high levels of the c-glucosidase was established. In 48 hr, the Recombinant human acid a-glucosidase made in bacteria cells cultured in Iscove's medium with 5 mM butyrate secreted was not catalytically active (14), and that made in baculovirus- 11O-kDa precursor enzyme that accumulated to 91 ,ug ml-t in infected insect cells was active but was not taken up efficiently the medium (activity, >22.6 ,umol hr-l ml-"). This enzyme by a Pompe patient's fibroblasts (15). Enzyme secreted by has a pH optimum similar to that of the mature form, but a transiently transfected COS cells was active and taken up lower Vmax and Km for 4-methylumbelliferyl-ca-D-glucoside. It efficiently by a Pompe patient's fibroblasts (3). Thus, mam- is efficiently taken up by fibroblasts from Pompe patients, malian cells are required for correct posttranslational modi- restoring normal levels of acid c-glucosidase and glycogen. fications for use in enzyme replacement. Chinese hamster The uptake is blocked by mannose 6-phosphate. Following ovary (CHO) cells are most frequently used for recombinant intravenous injection, high enzyme levels are seen in heart protein production in a mammalian system. Here we describe and liver. An efficient production system now exists for a system for high-level production of recombinant human acid recombinant human acid ai-glucosidase targeted to heart and a-glucosidase in CHO cells. capable of correcting fibroblasts from patients with Pompe disease. MATERIALS AND METHODS Lysosomal acid a-glucosidase (1,4-a-D-glucan glucohydrolase; Materials. Glycogen (rabbit liver type III), Aspergillus niger EC 3.2.1.20) is an exo-1,4-a-D-glucosidase that hydrolyzes both amyloglucosidase, 4-methylumbelliferone and its derivatives, a-1,4 and a-1,6 linkages of oligosaccharides, liberating glu- the lactate diagnostic kit, and dialyzed fetal bovine serum cose. It catalyzes the complete degradation of glycogen, with (FBS) (Mr cutoff, 20,000) were from Sigma. Con A-Sepharose slowing at branching points. The 28-kb acid a-glucosidase gene 4B and Sephadex G-200 were from Pharmacia. FBS and on human encodes a 3.6-kb mRNA which Geneticin were from Life Technologies (Gaithersburg, MD). produces a 951-aa polypeptide (1, 2). The enzyme receives Tissue culture media were from BioWhittaker. Restriction cotranslational N-linked glycosylation in the endoplasmic re- were from New England Biolabs. T4 DNA was ticulum. It is synthesized as a 110-kDa precursor form, which from Life Technologies, and Epicurian coli TOPP competent matures by extensive modification of its glycosylation, phos- bacteria, from Stratagene. phorylation, and proteolytic processing through a 90-kDa Plasmid Constructs. The full-length acid a-glucosidase endosomal intermediate into the final lysosomal 76- and cDNA (1) was excised with EcoRI from the pSHAG2 vector 67-kDa forms (1, 3-5). (3), a gift from A. Reuser (Erasmus University, Rotterdam), In Pompe disease, a deficiency of acid a-glucosidase causes and ligated into the multicloning site of the mammalian expres- massive accumulation of glycogen in , disrupting sion vector pcDNA3 (Invitrogen) to give pcDNA3-GAA. The cellular function (6). In the most common infantile form, cDNA was also cloned as a polycistronic construct in the EcoRI patients exhibit progressive muscle degeneration and cardio- restriction site of the pMT2 vector (16), a gift from R. J. Kaufman myopathy and die before 2 years of age. Intravenous injection (Genetics Institute, Cambridge, MA), and called pMT2-GAA. of enzyme obtained from human placenta or Aspergillus niger The dihydrofolate reductase (DHFR) gene and its adenovirus corrected enzyme and glycogen levels in liver but not in muscle major late promoter were excised from the pMT2 vector with or heart in patients with Pompe disease (7-9). Acid at-gluco- BamHI and ligated in the Bgl II site of pcDNA3-GAA. The sidase is targeted to lysosomes via the mannose-6-phosphate resultant plasmid, pJW-GAA, had the DHFR gene in the same receptor as well as by sequences associated with delayed cleavage orientation as the acid a-glucosidase gene. A 547-bp fragment, of the signal peptide (6). Mannose 6-phosphate-containing en- starting 18 bp 5' of the ATG start codon and ending 529 bp into zyme from bovine testes, human urine, or medium of transiently the coding sequence of acid a-glucosidase, was prepared by PCR

The publication costs of this article were defrayed in part by page charge Abbreviations: CMV, cytomegalovirus; DHFR, dihydrofolate reduc- payment. This article must therefore be hereby marked "advertisement" in tase; FBS, fetal bovine serum; MTX, methotrexate. accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 65 Downloaded by guest on September 29, 2021 66 Genetics: Van Hove et al. Proc. Natl. Acad. Sci. USA 93 (1996)

amplification using the primers GAA224(+) (5'-TCC-AGG- the medium to pH 6.6) in 25-cm2 confluent flasks with 0.2 ml CCA-TCT-CCA-ACC-AT-3') and GAA751(-) (5'-TCT-CAG- of medium per cm2. TCT-CCA-TCA-TCA-TCA-CG-3') and cloned into pCRII (In- Enzyme Purification. Acid a-glucosidase was purified from vitrogen). The resultant plasmid, pCRII-1, contained the Hindlll human placenta (18) and human urine (10). Antiserum to the site 5' relative to the ATG start codon and a Sac II site 320 bp placental enzyme was raised in rabbits. For purification of into the coding sequence of acid a-glucosidase. pJW-5' sGAA, recombinant acid a-glucosidase, the method used for urinary containing the 5'-shortened version of the acid a-glucosidase enzyme was employed but with a 2.6 cm x 100 cm Sephadex cDNA, was then constructed by replacing the HindIII-Sac II G-200 column. After elution, enzyme-containing fractions fragment of pJW-GAA with the HindIII-Sac II fragment of were combined into four fractions. Following the addition of pCRII-1. 0.1 volume of 0.65 M NaCl/0.5 M citrate, pH 6.50, they were Cell Culture, Transfection, and Selection. Plasmids were concentrated in 10-kDa Centriprep cartridges (Amicon). To screened by transient transfection of COS-7 cells with Lipo- examine purity, 10 ,ug of total protein was analyzed by SDS/ fectin (Life Technologies) and assay of enzymatic activity in PAGE and stained with Coumassie blue R. Denatured frac- cells and medium. CHO cells (CHO-Kl) (ATCC CCL 61) and tions (2 gl of enzyme boiled with 1% SDS) were deglycosylated DHFR- CHO cells (ATCC CRL 9096) were maintained in overnight at 37°C with N-glycosidase F (0.5 unit; Boehringer Mannheim). The molecular weight of the enzyme in the CHO minimal essential medium a (MEM-a), supplemented with ribonucleotides and deoxyribonucleotides for the DHFR- cells and culture medium under various culture conditions was and detection with cells. pJW-5'sGAA was transfected into regular CHO-Kl cells. analyzed by SDS/PAGE, Western blotting, a-glucosidase antiserum and chemolumines- pMT2-GAA, pcDNA3-GAA, and pJW-5'sGAA were trans- rabbit anti-acid fected into DHFR- CHO cells. Transfections used 50 ,ug of cence (ECL; Amersham). Enzyme Assays. Acid and neutral a-glucosidase were as- linearized plasmid DNA and 200 ,ug of sheared salmon sperm DNA in 800 ,ul of Hepes-buffered saline, with electroporation sayed using 4-methylumbelliferyl a-D-glucoside in a 96-well plate (19). Acid fB-galactosidase, ,3-glucuronidase, and total at 320 V and 960 ,uF in a Bio-Rad Gene Pulser (17). After 2 hexosaminidase were similarly assayed with the respective days, the cells were subcultured and selection was initiated 4-methylumbelliferyl derivatives (20). a-Glucosidase activity with Geneticin at 700 for plasmids pcDNA3-GAA and jig/ml was also measured by following the release of glucose from pJW-5'sGAA. For pMT2-GAA and pJW-5'sGAA, selection maltose or glycogen (21). The glycogen content of fibroblast in nucleotide-free medium, with 10% dialyzed FBS, was lysates was assayed by the release of glucose after digestion of started after 4 days. For some clones 5 or 25 nM methotrexate 25 ,ll of boiled supernatant with 0.5 unit of Aspergillus niger (MTX) was added. After colonies appeared, single-cell cul- amyloglucosidase at 37°C for 90 min. Total protein was tures were established by use of cloning rings or limiting measured by the Bradford method (22). dilution. Enzyme activity was determined in confluent cultures Uptake of Acid a-Glucosidase by Fibroblasts, Liver, Mus- in 25-cm2 flasks after incubation for 2 days in medium with cle, and Heart. Fibroblasts from two patients with infantile serum in which the acid a-glucosidase enzyme had been Pompe disease, without measurable residual enzyme activity inactivated (1 hr at 56°C and pH 10). Cells were washed twice (<1 nmol/hr per mg of protein), were grown to confluency in with phosphate-buffered saline, harvested by scraping, pel- Eagle's MEM containing 10% FBS in which acid a-glucosidase leted, and suspended in 130 ,ul of water. Cell lysates were was inactivated. The uptake in fibroblasts was examined after prepared by brief sonication; the debris was pelleted by incubation for 24 hr in medium with recombinant enzyme- centrifugation, and glucosidase activity and total protein con- activity at 1000 nmol hr- 'ml-'. Mannose 6-phosphate (5 mM) tent were determined in the supernatant. Both cells and was added to the medium to examine the effect of blockade of medium were analyzed, and the 10 highest-producing clones of the mannose-6-phosphate receptor. To examine the effect on pJW-5'sGAA were grown in 200 nM MTX. Resistant clones intracellular glycogen stores, the cells were incubated in Ea- were further amplified with methotrexate. Clonal homogeneity gle's MEM without glucose or with glucose at 1 mg-ml-1, with was assured by repeated single-cell selection, using cloning 10% dialyzed FBS in which the acid a-glucosidase had been rings initially and later followed by limiting dilution with inactivated, and with or without purified enzyme activity at selection for the highest producers. Stability of cell clones was 1500 nmol-hr-1ml-1. To examine tissue localization of the examined after 3 weeks in culture without the selection agents. enzyme in vivo, the remainder of fraction 11 (460 ,ll) was To optimize production, enzyme activity, glucose, and lactate injected intravenously into an 8-day-old Hartley guinea pig, in the medium were followed over 48 hr under various culture with two littermates as control. After 24 hr the animals were conditions (Iscove's medium, 5 mM butyrate, acidification of sacrificed and acid a-glucosidase activity was determined in Table 1. 1 Recombinant acid a-glucosidase production MEM-a Iscove's plus butyrate CHO cells Vector MTX, nM Cell Medium Cell Medium DHFR+ None 25 <2 DHFR- pMT2 427 4.1 DHFR- pcDNA3 2,800 69 3,000 432 DHFR+ pJW 3,000 277 3,470 1,720 DHFR- OM22 pJW 9,910 836 6,560 2,070 DHFR- 5M25 pJW 5 1,630 320 200 5,050 977 17,300 2,960 800 28,000 2270 35,200 3,080 5,000 33,500 3070 73,500 22,600 20,000 12,400 3560 40,200 11,300 Regular (DHFR+) or DHFR- CHO cells were transfected with pMT2 or pcDNA3, containing a full-length acid a-glucosidase gene, or with pJW, containing a 5'-shortened acid a-glucosidase gene, and amplified with MTX as indicated. Enzyme activity levels in cells (nmolhr-lmg-1) and medium (nmol hr- 'ml-1) were assayed after 48 hr in MEM-a or in Iscove's medium with 5 mM butyrate. Downloaded by guest on September 29, 2021 Genetics: Van Hove et al. Proc. Natl. Acad. Sci. USA 93 (1996) 67

supernatant solutions of muscle, heart, and liver after homog- 25 - i30 enization, sonication, and centrifugation. A 25 2 20 E RESULTS 20 @ Production of Recombinant Enzyme. Expression of the .C 15' 0 0 full-length acid a-glucosidase cDNA in the pcDNA3 vector E1 15- under control of the cytomegalovirus (CMV) promoter and -i 10' 0)0 selected with Geneticin was more efficient than that with the 10 0(o -> polycistronic construct of the pMT2 vector under the adeno- C-, 5 - virus major late promoter and selection based only on the A 5 (5 DHFR gene (Table 1). Incorporation of the DHFR gene into pcDNA3-GAA, giving the pJW vector series, allowed for 0 10 20 30 40 50 0 10 20 30 40 50 MTX-induced amplification. The initial plasmid contained the Time, hr full-length 5' and 3' ends of the acid a-glucosidase cDNA. Transient expression in COS cells of the 5'-shortened cDNA FIG. 1. Influence of culture conditions on acid a-glucosidase incorporated in pJW-5'sGAA showed improved expression production. Time course of acid a-glucosidase activity (circles), glu- (intracellular, 133.4 nmolPhr-' mg-1; medium, 59.1 nmol- cose (diamonds), and lactate (squares) in MEM-a (A) and Iscove's hr-'1ml-1) compared with pcDNA3-GAA (77.1 nmolhr-l1 medium with 5 mM butyrate (B) (mean values of at least four mg- I and 27.9 nmolhr- 'ml- '). When this shortened cDNA in experiments). the pJW vector was used for stable transfection of regular which time a maximum of lactate was achieved. The acid CHO cells with similar single selection with Geneticin, the 24 yield was also improved, particularly in the medium (Table 1). a-glucosidase level increased nearly linearly for hr, then After transfection into DHFR-deficient CHO cells and selec- plateaued until 42-48 hr, after which the acid a-glucosidase tion with Geneticin in ribonucleotide-deficient medium, 10 level dropped and cell death occurred (Fig. 1A). The cells in clones were selected for optimal production. Small amounts of Iscove's medium metabolized glucose throughout the 48 hr MTX in the initial selection (0, 5, or 25 nM) did not increase period, albeit at a lesser rate after 24 hr, when lactate peaked the yield of enzyme. A single clone (OM22) produced the and cells started metabolizing more lactate. This much higher highest amounts of recombinant enzyme and, after renewed lactate production was associated with a greater drop in pH single-cell selection, was used in the initial characterization of (after 48 hr, MEM-a was pH 7.5, and Iscove's medium was pH the recombinant enzyme. This cell line ultimately proved 6.8). The cells in Iscove's medium with butyrate used glucose unstable and activity was lost to the level of the other clones continuously and produced lactate throughout the 48-hr pe- (similar to 5M25; Table 1). When the 10 clones were grown in riod, albeit at a lower rate than in Iscove's medium alone (Fig. 200 nM MTX, recombinant enzyme production was increased 1B). The drop in pH was also less marked (pH 7.1). The acid in only 1 clone (5M25). This cell line was further amplified by a-glucosidase enzyme level increased throughout the 48-hr stepwise increases in MTX, while clonal homogeneity was period but was more variable in Iscove's medium without assured by repeated single-cell cloning. Maximal expression butyrate. There were 60% more cells when the cells were was attained at 5 ,uM MTX, with decreased expression ob- incubated in Iscove's medium without butyrate (11 x 106 cells served in cells resistant to 20 p.M MTX. MTX-induced am- vs. 7 x 106 cells), but with butyrate the cells produced more plification resulted in a 20-fold increase in intracellular enzyme enzyme per cell (40,200 nmol hr- per mg of protein and 4.59 and a 10-fold increase in recombinant enzyme in the medium. pmolbhr-1 per cell vs. 12,400 nmolbhr-1 per mg of protein and Overall, the genetic manipulation of the CHO cells resulted in 3.89 pmol-hr- ' per cell). a 1340-fold increase of acid a-glucosidase intracellularly and a The overproducing cells contained predominantly the 76- 3000-fold increase in the medium compared with the original kDa form of the enzyme intracellularly with significant untransfected CHO cells. At this level, 30% of all enzyme in amounts of the 90-kDa precursor form (Fig. 2). In contrast, the the culture dish was intracellular, the rest being in the medium. enzyme in the medium was almost completely the 110-kDa The relative amount secreted increased with increasing levels of form. Culture conditions did not affect the molecular mass of production. There was no increase in acid ,3-galactosidase, /3-gluc- the enzyme; only in MEM-a with 5 mM butyrate was a small uronidase, or hexosaminidase in the medium of the transformed amount of the 76-kDa form most cells. Clonally homogeneous cells resistant to 800 nM MTX present, likely representing showed stable expression in cells and medium after 3 weeks cell lysis. Under optimized conditions, Iscove's medium with 5 culture without Geneticin. But when cultured for 3 weeks without mM butyrate for 48 hr, the cells produced the enzyme at 22,600 Geneticin and MTX, activity was reduced to the level before nmol-hr-I per ml of the medium (91 p.g/ml), with 73,500 MTX amplification (intracellular, 17,583-nmol-hr-lmg-1; me- dium, 424 nmol hr- 'ml- 1). Medium Cells Optimization of Acid a-Glucosidase Production. Reduction (x (x+But Isc Isc+But ox ot+But Ise Isc+But Urine of the pH of the medium to 6.6 stabilized the enzyme and kDa improved yield in MEM-a from 1135 to 4725 nmolbhr- 'ml-'. This effect was not seen with Iscove's medium. Butyrate ge I 110 increased acid a-glucosidase production both in cells and in as* medium when either MEM-a (intracellular, from 22,900 to 32,900 nmol-hr-lmg-1; medium, from 1140 to 3650 nmol ..... 76 hr-l ml-1) or Iscove's medium (intracellular, from 16,700 to 67 25,700 nmol-hr- 'mg-1; medium, from 4200 to 7260 nmol- hr- was used. The results of the from MEM-a 'ml-') change FIG. 2. Acid a-glucosidase isoforms in transfected CHO cells and to Iscove's medium were variable and dependent on cell in the medium. Western blot analysis of recombinant enzyme in CHO confluency. To analyze this effect in more detail, glucose, cells (10 ,ug of cellular protein) and medium (40 ,ul) under various lactate, and the pH of the medium were analyzed in highly culture conditions: MEM-a (a), MEM-a with 5 mM butyrate confluent cultures. In MEM-a, the glucose level dropped to (a+But), Iscove's medium (Isc), and Iscove's medium with 5 mM below the detection limit (0.8 mM) within 12 hr of culture, at butyrate (Isc+But). Downloaded by guest on September 29, 2021 68 Genetics: Van Hove et al. Proc. Natl. Acad. Sci. USA 93 (1996)

Table 2. Purification of acid a-glucosidase from tissue culture medium Activity, Volume, Total activity, Specific activity, Relative Yield, Km, Vmax, ,umol hr-l ml-1 ml ,umol hr- ,umol-hr-l.mg-1 purity % mM j,mol hr-'mg- Medium 0.93 6000 5608 0.33 1 Con A 5.55 750 4160 9.2 28 74 Con A conc. 281 16 4630 15.0 46 82 Sephadex G-200 48 87 4180 367 1122 75 Fraction I 505 0.70 350 240 732 6 0.62 248 Fraction II 3385 0.71 2400 388 1184 43 0.91 495 Fraction III 1758 0.80 1410 461 1406 25 1.08 548 Fraction IV 218 0.75 160 309 944 2.9 Seph. conc. 1460 2.96 4330 384 1174 77 Tissue culture medium to which an equal volume of 0.5 M NaCl/0.05 M citrate, pH 6.5, was added was purified on a concanavalin A-Sepharose 4B column (Con A). The sample was concentrated and diafiltered with 25 mM NaCl/1 mM EDTA/20 mM acetate buffer, pH 4.5 (Con A conc.), before purification on a Sephadex G-200 column. The final concentrated product (Seph. conc.) is the sum of the four concentrated fractions (I-IV). Kinetic parameters were determined with 4-methylumbelliferyl a-D-glucoside as substrate.

nmol-hr-1 per mg of protein in the cells (13% of cellular with purified recombinant enzyme at 1500 nmol hr-l ml-1, the protein). intracellular glycogen dropped to a level similar to that in normal Purification of Recombinant Enzyme. Acid a-glucosidase control fibroblasts (Fig. 4). This decrease was prevented by was retarded on the Sephadex G-200 column beyond the addition of mannose 6-phosphate to the medium. When glucose- elution of bovine serum albumin, resulting in >200-fold pu- free medium was used for 24 hr to reduce the contribution of rification with a recovery of >70% (Table 2), with no addi- cytoplasmic glycogen, a similar drop in intracellular glycogen was tional bands visible by Coomassie staining (Fig. 3). The first seen with recombinant enzyme. Twenty-four hours after intra- fraction contained almost uniquely the precursor 110-kDa venous injection of 4 mg of purified enzyme (fraction II) into a form, while the later fractions contained increasing amounts of 124-g guinea pig, the acid a-glucosidase activity was strikingly the 76-kDa form (Fig. 3). The specific activity increased with higher in the liver and in the heart than in liver and heart of the increased amount of the 76-kDa form (Table 2). The control animals (Table 4). diffuse nature of the bands on SDS/polyacrylamide gels suggested variable glycosylation. After deglycosylation with DISCUSSION N-glycosidase F, the first fraction showed two bands at 100 and 95 kDa. The more mature 95- and 76-kDa forms in fractions Large amounts of human acid a-glucosidase are required for 2 and 3 similarly showed a second band, at 90 and 68 kDa, therapeutic trials in patients with Pompe disease. Shortening respectively. Analysis of substrate kinetics with 4-methylum- of the 5' end of the acid a-glucosidase cDNA increased the belliferyl a-D-glucoside gave a linear Lineweaver-Burk plot efficiency of expression of this enzyme. Use of a CMV and showed a lower Vma,, and lower Km for fraction 1 than for promoter-driven expression vector with Geneticin selection fractions II and III (Table 2). The pH optimum of the enzyme and DHFR amplification allowed the development of a high- for 4-methylumbelliferyl a-D-glucoside was 4.5, similar to producing cell clone with only three rounds of amplification. enzyme purified from human urine and placenta. The pH This technique considerably shortens the time required with optimum of the recombinant acid a-glucosidase was 4.5 for MTX alone. Increased production of recombinant lysosomal maltose and 4.25 for glycogen. enzymes has been reported with the use of NH4Cl to interfere Correction of Patient's Fibroblasts and in Vivo Animal with lysosomal routing (23), with acidification of the medium Experiment. Addition of tissue culture medium containing to stabilize the lysosomal enzymes (24), and with butyrate, recombinant acid a-glucosidase at 1000 nmol-hr-lml-1 to which increases transcription of stably transfected genes (25, fibroblasts from two patients showed uptake of the enzyme 26). In our work, treatment with NH4Cl did not improve the that reached the intracellular levels seen in normal fibroblasts. production (data not shown). Butyrate increased the produc- The uptake was >99% inhibited by 5 mM mannose 6-phos- tion per cell, both in the cells and in the medium, but toxic phate (Table 3). After a 24-hr incubation of patient's fibroblasts effects limited its use to 48 hr. The low glucose content in MEM-a limited the production of recombinant enzyme to 24 M I II III IV S hr. In high-glucose (4.5 mgml-') Iscove's medium, the re- I Im l A B A B A B A sponse depended on the confluency. If sufficient cells were kDa present at the initiation of the experiment, high production _- 200 resulted from a combination of increased cell numbers (even Table 3. Enzyme uptake in fibroblasts of patients with infantile Pompe disease ,o- 116 l Acid a-glucosidase _- 97 ...... activity, nmol-hr- lmg- 1 0 0 ,, _- 66 Treatment Patient 1 Patient 2 Sham <1.0 <1.0 Sham + M6P 2.2 <1.0 _ -45 CHO Enzyme added 385.0 429.0 CHO Enzyme + M6P 3.6 2.6 FIG. 3. SDS/PAGE analysis of purified acid a-glucosidase. Sam- Fibroblasts were incubated for 24 hr with acid a-glucosidase (ac- ples (20 ,ug of protein) were analyzed by SDS/7.5% PAGE. Purified tivity, 1000 nmol hr-l ml-1) with or without 5 mM mannose 6-phos- fractions I-IV containing acid a-glucosidase are shown before de- phate (M6P), and acid a-glucosidase specific activity was measured. glycosylation (lanes A) and after deglycosylation (lanes B). Lane M, Normal control fibroblasts (n = 22) contained 255 nmol hr-l mg-1 medium before purification; lane S, molecular size standards. (range 66-498). Downloaded by guest on September 29, 2021 Genetics: Van Hove et al. Proc. Natl. Acad. Sci. USA 93 (1996) 69

1.2 enzyme, prepared from medium from which suspended cells 3 had not been removed, contained all forms of the enzyme. - Because of the larger, 110-kDa size of the precursor enzyme, 11 °~~~~~~~~ 1 Sephadex G-200 had to be used instead of the usual Sephadex G-100 for purification. The large precursor was eluted earlier 0. (fraction I), allowing its biochemical characterization. We found t- 0- 2 4 it to have a lower Vmax and Km than mixtures containing the more a) mature forms, but a similar pH optimum with 4-methylumbel- 0.60. a) liferyl a-D-glucoside as substrate. These differences could be due E 0.2 to a larger contribution of catalytically inactive leader peptide- 0- 3 containing enzyme in the first fraction (14) or to an increase in a)0 1 0 specific enzyme activity with maturation (4). 2 For effective therapy, the enzyme has to be taken up by cells, localized to the lysosomes, and there decrease the accumula- tion of glycogen. Our studies show normalization of the acid a-glucosidase level, which is prevented by mannose 6-phos- With glucose Without glucose phate, indicating the participation of the mannose-6-phosphate receptor. Recombinant acid a-glucosidase joins a growing list of FIG. 4. Intracellular glycogen levels (measured as glucose release, recombinant lysosomal enzymes produced in CHO cells that gmolPhr- 1, upon digestion with Aspergillus niger amyloglucosidase) in exhibit high uptake mediated by the mannose-6-phosphate re- patient's fibroblasts after 24-hr treatment with recombinant enzyme (1500 nmol-hr-l'ml-1) in MEM with glucose or without glucose. ceptor (23, 24, 27). The decrease of both lysosomal and cytoplas- Patient's cells treated with enzyme (bars 2) or with enzyme plus 5 mM mic glycogen with enzyme treatment and the strikingly increased mannose 6-phosphate (bars 3) are compared with sham-treated pa- enzyme levels in the heart following intravenous injection illus- tient's cells (bars 1) and control fibroblasts (bars 4). trate the potential of this recombinant enzyme for in vivo targeting and correction of affected tissues in Pompe disease. suspended in culture) and stabilization of the enzyme at the Human enzyme of the appropriate mannose-6-phosphate recep- low pH by production of a large amount of lactic acid. Finally, tor-targeted form is now available to investigate its potential for combining Iscove's medium with cells at high confluency and enzyme replacement therapy in Pompe disease. butyrate for 48 hr proved optimal for enzyme production, with enzyme up to 91.4 ,tg/ml in the medium and constituting up We thank E. W. Suh for technical assistance and the Duke Cancer to 13% of the cellular protein. This level supersedes previously Center Cell Culture Facility for assistance in cell culture. This study published levels for production of lysosomal enzymes in mam- was supported by a generous gift from Sympac Chemicals Ltd. and malian cells (23-25, 27, 28). One possible explanation for the from the Carmen Ross Memorial Fund. J.-Y.W. is a recipient of a fellowship award from the Muscular Dystrophy Association. high production of enzyme was the extensive selection for subclones that not only accumulated the a-glucosidase intra- 1. Hoefsloot, L. H., Hoogeveen-Westerveld, M., Kroos, M. A., van cellularly, but also secreted it efficiently into the medium. Cell Beeumen, J., Reuser, A. J. J. & Oostra, B. A. (1988) EMBO J. 7, clones with similar intracellular levels of enzyme activity had 1697-1704. >10-fold differences in activity in the medium. Yet, as noted 2. Martiniuk, F., Mehler, M., Tzall, S., Meredith, G. & Hirschhorn, with recombinant production of other lysosomal enzymes (24, R. (1990) DNA Cell Biol. 9, 85-94. 25), this increased secretory characteristic did not involve 3. Hoefsloot, L. H., Willemsen, R., Kroos, M. A., Hoogeveen- other, endogenous lysosomal enzymes. Westerveld, M., Hermans, M. M. P., Van der Ploeg, A. T., Oostra, B. A. & Reuser, A. J. J. (1990) Biochem. J. 272, 485-492. Similar to what has been observed with transiently trans- 4. Wisselaar, H. A., Kroos, M. A., Hermans, M. M. P., van Beeu- fected COS cells (4), the medium contained only the 110-kDa men, J. & Reuser, A. J. J. (1993) J. Biol. Chem. 268, 2223-223 1. precursor enzyme even under optimized culture conditions. 5. Hermans, M. M. P., Wisselaar, H. A., Kroos, M. A., Oostra, The broad bands on SDS/polyacrylamide gels suggest a mi- B. A. & Reuser, A. J. J. (1993) Biochem. J. 289, 681-686. croheterogeneous . After deglycosylation, two 6. Hirschhorn, R. (1995) in The Metabolic and Molecular Basis of sharp bands in the region of 95 kDa were seen, confirming not Inherited Disease, eds. Scriver, C. R., Beaudet, A. L., Sly, W. S. & only the extensive contribution of glycosylation but also the Valle, D. (McGraw-Hill, New York), 7th Ed., Vol. 2, pp. 2443- underlying heterogeneity in protein size. The secreted form in 2464. 7. Hug, G., Schubert, W. K. & Chuck, G. (1968) Clin. Res. 16, 345 COS cells starts at His-29, corresponding to cleavage of the (abstr.). leader peptide (4). We have not ascertained whether the 8. de Barsy, T. & Van Hoof, F. (1974) in Enzyme Replacement difference in these two bands reflects a difference in start site Therapy in Lysosomal Storage Diseases, eds. Tager, J. M., Hoogh- (e.g., with or without loss of the leader peptide) or in the winkel, G. J. M. & Daems, W. T. (North-Holland, Amsterdam), carboxyl-terminal position. Intracellularly, the CHO cells con- pp. 277-279. tained typical intermediates of 90 and 76 kDa, but not the 9. Hug, G. & Schubert, W. K. (1967) J. Cell Biol. 35, C1-C6. 67-kDa form seen in fibroblasts, and little of the 1 10-kDa form 10. Oude Elferink, R. P. J., Brouwer-Kelder, E. M., Surya, I., Strij- land, A., Kroos, M., Reuser, A. J. J. & Tager, J. M. (1984) Eur. seen in transiently transfected COS cells (4). The purified J. Biochem. 139, 489-495. 11. Reuser, A. J. J., Kroos, M. A., Ponne, N. J., Wolterman, R. A., Table 4. Enzyme delivery to guinea pig tissues Loonen, M. C. B., Busch, H. F. M., Visser, W. J. & Bolhuis, P. A. Activity, nmolPhr- lmg- 1 (1984) Exp. Cell. Res. 155, 178-189. 12. Van der Ploeg, A. T., Bolhuis, P. A., Wolterman, R. A., Visser, Control I Control 2 Infused J. W., Loonen, M. C. B., Busch, H. F. M. & Reuser, A. J. J. Muscle 35 27 40 (1988) J. Neurol. 235, 392-396. Heart 49 19 176 13. Van der Ploeg, A. T., Kroos, M. A., Willemsen, R., Brons, N. H. C. & Reuser, A. J. J. (1991) J. Clin. Invest. 87, 513-518. Liver 25 62 2116 14. Martiniuk, F., Tzall, S. & Chen, A. (1992) DNA Cell Biol. 11, An 8-day-old guinea pig received purified recombinant enzyme 701-706. intravenously (activity, 12.6 ,umolPhr-1 per gram of animal weight). 15. Wu, J.-Y., Van Hove, J. L. K., Huang, Y. S., Zhang, W. & Chen, The tissues were examined 24 hr later, and acid c-glucosidase activity Y.-T. (1993) Am. J. Hum. Genet. 53, (Suppl.), 963 (abstr.). was measured. Two littermates served as controls. 16. Kaufman, R. J. (1990) Methods Enzymol. 185, 537-566. Downloaded by guest on September 29, 2021 70 Genetics: Van Hove et al. Proc. Natl. Acad. Sci. USA 93 (1996)

17. Barsoum, J. (1990) DNA Cell Biol. 9, 293-300. 23. Anson, D. S., Taylor, J. A., Bielicki, J., Harper, G. S., Peters, C., 18. Reuser, A. J. J., Kroos, M., Oude Elferink, R. P. J. & Tager, J. M. Gibson, G. J. & Hopwood, J. J. (1992) Biochem. J. 284, 789-794. (1985) J. Biol. Chem. 260, 8336-8341. 24. Kakkis, E. D., Matynia, A., Jonas, A. J. & Neufeld, E. F. (1994) 19. Reuser, A. J. J., Koster, J. F., Hoogeveen, A. & Galjaard, H. Protein Expression Purif 5, 225-232. (1978) Am. J. Hum. Genet. 30, 132-143. 25. Ioannou, Y. A., Bishop, D. F. & Desnick, R. J. (1992) J. Cell Biol. 119, 1137-1150. 20. Wenger, D. A. & Williams, C. (1991) in Techniques in Diagnostic 26. Dorner, A. J., Wasley, L. C. & Kaufman, R. J. (1989) J. Biol. Human Biochemical Genetics: A Laboratory Manual, ed. Hommes, Chem. 264, 20602-20607. F. A. (Wiley-Liss, New York), pp. 587-617. 27. Bielicki, J., Hopwood, J. J., Wilson, P. J. & Anson, D. S. (1993) 21. Park, H. K., Kay, H. H., McConkie-Rosell, A., Lanman, J. & Biochem. J. 289, 241-246. Chen, Y.-T. (1992) Prenatal Diagn. 12, 169-173. 28. Grubb, J. H., Kyle, J. W., Cody, L. B. & Sly, W. S. (1993) FASEB 22. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254. J. 7, 1255a (abstr.). Downloaded by guest on September 29, 2021