Characterization of Recombinant α-Galactosidase for Use in Seroconversion from Blood Group B to O of Human Erythrocytes

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS

Vol. 327, No. 2, March 15, pp. 324–329, 1996Article No. 0129

Characterization of Recombinant a-Galactosidase for Usein Seroconversion from Blood Group B to Oof Human Erythrocytes1

Alex Zhu, Lin Leng, Catherine Monahan, Zhanfan Zhang, Rosa Hurst,Leslie Lenny, and Jack Goldstein2

Lindsley F. Kimball Research Institute, The New York Blood Center, New York, New York 10021

Received October 10, 1995, and in revised form December 15, 1995

Key Words: a-galactosidase; coffee beans; Pichia pas-toris; recombinant protein, characterization; sero-a-Galactosidase (a-GAL) purified from green coffeeconversion.bean cleaves the terminal galactose residues from

the surface of group B erythrocytes, thereby con-verting these cells serologically to group O cells.Such enzymatically converted red cells have beentransfused into group A and O recipients as part of It is possible to enzymatically convert group B redthe first phase of FDA-approved clinical trials. Re- cells to group O red cells in vitro (B r O) because thecently we expressed the recombinant a-GAL (ra- group B antigen differs structurally from the group OGAL) in large quantities in a methylotrophic yeast

antigen only by the addition of one terminal a-linkedstrain Pichia pastoris and purified the protein to ap-galactose residue (1, 2). a-Galactosidase (a-GAL)3 iso-parent homogeneity by chromatography on a macrolated from green coffee beans has been demonstratedprep S50 column. Purified ra-GAL, migrating as ato be highly active in removing the a-linked galactosesingle band of 41 kDa on a SDS–PAGE, appears to beresidues from the group B red cell surface (3). Afteridentical to its native counterpart in specific activitytreatment with a-GAL, the red cells become serologi-(32 U/mg) and kinetic parameters (Km Å 0.363 mM andcally group O and maintain full erythrocyte integrityVmax Å 46.9 U/mg). Both enzymes demonstrate theand viability in vitro. Such enzymatically converted redsame pH profile in the pH range from 2 to 9, with ancells have been successfully transfused to both groupoptimal pH at 6.4 when tested with the substrate p-A and O recipients as part of FDA-approved clinicalnitrophenol-a-D-galactopyranoside. Furthermore, as

with its native counterpart, ra-GAL specifically trials (4). Some of the recipients have also been chal-cleaves a-linked terminal galactose residues from lenged with the converted cells several months aftergroup B red cells without affecting other major anti- their first transfusion. In all cases, the converted cellsgens on the red cell surface. In addition, we devel- survived normally in vivo and antibodies were not de-oped a method for using RT-PCR to detect possible tectable toward either the converted cells or toward theDNA contamination in the purified protein prepara- enzyme used in the conversion.tion, which is one of the concerns for in vivo studies. Considering the amount of enzyme needed for theThus, with a simple procedure for over-expression B r O blood type conversion (45,000 U a-GAL per unitand purification of ra-GAL from P. pastoris cul- of blood), preparation of purified a-GAL from greenture, one can readily obtain the enzyme needed for coffee beans is both labor intensive and time consuminglarge-scale sero-conversion of red cells. q 1996 Academic (5). We have succeeded in isolating a full-length cDNAPress, Inc. clone coding for coffee bean a-GAL (6). Availability of

the cDNA made it possible to express the protein in a1 Supported in part by Office of Naval Research Grant N00014-

91-J-1180 with funds provided by the Naval Medical Research and3 Abbreviations used: a-GAL, a-galactosidase; ra-GAL, recombi-Development Command to the New York Blood Center.

2 To whom correspondence and reprint requests should be ad- nant a-GAL; na-GAL, native enzyme a-GAL; SDS–PAGE, sodiumdodecyl sulfate–polyacrylamide gel electrophoresis.dressed. Fax: (212) 879-0243.

324 0003-9861/96 $18.00Copyright q 1996 by Academic Press, Inc.

All rights of reproduction in any form reserved.

/ 6b14$$9345 02-10-96 02:17:24 arcas AP: Archives

325CHARACTERIZATION OF RECOMBINANT COFFEE BEAN a-GALACTOSIDASE

pH 7.3, or phosphate-citrate-sodium chloride buffer, pH 5.5. The en-cell system that is designed for high-level expressionzymes at these two different pHs were then incubated at either 267Cof heterogeneous proteins. We expressed the protein inor 377C for up to 120 h. During the incubation, aliquots were taken

Escherichia coli, Saccharomyces cerevisiae, COS cells, at different time points for measuring activity as described underand SF9 insect cells; however, the results were unsatis- Enzyme Assay.factory for a variety of reasons (see Section 1 under Glycosylation analysis. Protein samples (5 mg) were subjected toResults and Discussion). A satisfactory-level of biologi- 12.5% SDS–PAGE and then transferred to PVDF membrane (Bio-

Rad) by electro blotting (Trans-blot SD semidry transfer cells fromcally active recombinant a-GAL (ra-GAL) was pro-Bio-Rad). The blots were subjected to glycosylation analysis usingduced in Pichia pastoris, a methylotrophic yeast strain,the GlycoTrack kit (Oxford GlycoSystems) with transferrin servingand ra-GAL was purified from the P. pastoris culture as positive control.

supernatant by a simple chromatography procedureProtein concentration. Protein concentration was determined by(7). In order to use ra-GAL in B r O blood type conver- BCA protein assay (Pierce Chemicals, Rockford, IL) using bovine

sion studies, it is necessary to characterize the purified serum albumin as a protein standard.enzyme and compare its properties with those of native DNA contamination. DNA contamination of the purified ra-enzyme (na-GAL). GAL preparations was measured by PCR using oligonucleotide

primers specific for the 5* region of the a-GAL gene. An aliquot ofpurified ra-GAL (1.4 U/ml) was added to a PCR mixture and ampli-MATERIALS AND METHODSfied according to the PCR profile described previously (6). In order

Preparation of ra-GAL produced in Pichia pastoris. Our proce- to measure sensitivity of the method, genomic DNA isolated fromdures for high level expression and purification of ra-GAL have been paF-BZ-22, in a 10-fold serial dilution from 100 ng to 1.0 fg, wasreported (7). Briefly, the cDNA encoding a-GAL was subcloned into used as the template for PCR and run simultaneously with thethe vector pPIC9 (Invitrogen Corp., San Diego, CA) for the extracellu- protein sample. The PCR products were visualized by electropho-lar expression of the enzyme in P. pastoris, a methylotrophic yeast resis on a 1% agarose gel.strain. An over-expressing transformant (methanol utilizing positive,

Treatment of red blood cells. Group B red cells were treated withmut/) containing 11 copies of the a-GAL gene (paF-BZ-22) was se-ra-GAL under conditions identical to those used for na-GAL, as pre-lected based on a function assay for enzyme activity. In order toviously described (8). Red cells were incubated with ra-GAL (200 U/prepare the recombinant enzyme in large quantities, a high cell den-ml of cells) in isotonic phosphate-citrate-sodium chloride buffer, pHsity fermentation was performed in a 14-liter fermentor with a final5.5, for 2.25 h at 267C. The enzyme was then removed by washingworking volume of 10 liters (New Brunswick MF-114). By usingwith phosphate-sodium chloride buffer, pH 7.3.methanol induction, biologically active ra-GAL was secreted into the

culture medium and reached approximately 400 mg/liter of the me-dium after 4 days induction. After removing the cells by centrifuga-

RESULTS AND DISCUSSIONtion, the culture supernatant was concentrated and loaded onto aMacro Prep S-50 column (Bio-Rad Laboratories). The recombinant

1. Preparation of ra-GAL Produced in P. pastorisenzyme was bound to the column at lower salt concentration andeluted with a NaCl gradient ranging from 50 to 300 mM. The purified

In order to produce ra-GAL for use in blood-type con-enzyme was stored at 0207C until use.version, we isolated a full-length cDNA clone from cof-Reverse transcription-PCR (RT-PCR). The paF-BZ-22 culture

was induced with methanol for the expression of ra-GAL. The cells fee beans (6). The availability of the cDNA coding forwere then collected and used for total RNA isolation with the Micro- coffee bean a-GAL made it possible to produce the pro-Scale Total RNA Separator kit (Clontech). cDNA was then generated tein in large quantities in an expression system de-using the 1st-strand cDNA Synthesis kit (Clontech) with either ran-

signed for high-level expression of heterologous pro-dom hexamer or oligo (dT)18 primers. Multiple PCR reactions wereset up with primers covering the entire a-GAL gene according to the teins. The a-GAL protein was initially expressed inPCR conditions described previously (6). The amplified PCR products E. coli; however, it formed an inclusion body with nowere sequenced directly with the Applied Biosystems (ABI) 373A enzymatic activity (unpublished data). We then ex-DNA sequencer.

pressed biologically active enzyme in Spodoptera frugi-Enzyme assay. The a-GAL activity was measured at 267C in 0.1

perda (Sf9) insect cells with a baculovirus system. Us-M phosphate citrate buffer, pH 6.5, containing 1.25 mM of p-nitrophe-ing this system and site-directed mutagenesis we werenol a-D-galactopyranoside (PNP-a-gal), synthetic substrate for a-

GAL. After the reaction was terminated by adding 0.2 M sodium able to identify tyrosine at position 108 as an essentialborate buffer, pH 9.8, the amount of the liberated p-nitrophenol was residue for a-GAL activity (9). Because of low level ex-determined by measuring the absorbency at 405 nm. One unit (u) of pression of the enzyme in Sf9 cells, as well as in COSthe enzyme activity is defined as the amount of a-GAL that hydro-

cells and S. cerevisiae, these host cell lines were notlyzes 1 mM of the substrate per minute under the assay conditions.useful for large-scale production of ra-GAL. Recently,Kinetic analysis. The initial hydrolysis rates of the substrate forwe have succeeded in producing high-levels of ra-GALraGAL at various concentrations (0.4–1.0 mM) were measured under

the same conditions as described above. Under such conditions, lin- in P. pastoris and purifying the enzyme by a simpleear rates were observed and the data were analyzed by using the chromatography procedure (7). Approximately 400 mgcomputer program ‘‘ENZPACK’’ (Bio-Soft, Cambridge, UK). Km and (12,000 u) of the enzyme was obtained from 1 liter ofVmax values of the enzyme were then obtained with the Lineweaver–

high cell-density culture with a final yield of at leastBurk double reciprocal plot method.80%, as compared with an initial recovery of 4000 unitsThermostability. Purified native and recombinant a-GALs were

diluted to 18 unit/ml with 50 mM phosphate-sodium chloride buffer, of enzyme from 1 kg of coffee beans and only a 50%


326 ZHU ET AL.

TABLE I gene were PCR-amplified and the products were se-quenced. The sequencing data indicated that ra-GALComparison of Native and Recombinant a-Galactosidasesexpressed in P. pastoris has a sequence identical tothe published sequence for coffee bean a-GAL (6). Asna-GAL ra-GALpreviously reported, N-terminal sequencing of the puri-

Initial recovery 4000 U/kg 12,000 U/liter fied ra-GAL indicated that approximately half of theFinal yield 50% 80% protein molecules have the N-terminus identical to theMolecular weight 41 kDa 41 kDa

native enzyme isolated from green coffee beans;Specific activity 32 U/mg 32 U/mgwhereas, the other half of the protein molecules containOptimal pH 6.4 6.4

Km 0.377 mM 0.363 mM an extra phenylalanine residue at the N-terminus (7).Vmax 44.8 U/mg 46.9 U/mg This is generated by heterogeneous cleavages in P. pas-

toris host cells of the secretion sequence (in the vectorNote. Comparison of native and recombinant a-GALs. MolecularpPIC 9) from the mature a-GAL molecule.weight was estimated according to the protein migration on a SDS–

Recombinant and native a-GALs appear to be identi-PAGE. Determination of specific activity is described under Materi-als and Methods. The maximal enzyme activity toward PNP-a-gal cal in molecular mass, specific activity, optimal pH,was observed at pH 6.4 (see Fig. 1). Apparent Michaelis constant and kinetic parameters (Table I). Furthermore the Km(Km) and maximal velocity (Vmax) were generated by the computer and Vmax values for na-GAL, 0.377 and 44.8 (rÅ 0.998),program ENZPACK based on the method of Lineweaver–Burk and

respectively, and for ra-GAL, 0.363 and 46.9 (r Åexpressed in mM and U/mg, respectively. Initial recovery indicatesthe amount of na-GAL from 1 kg of coffee beans and ra-GAL from 0.997), respectively, are considered within the range of1 liter of P. pastoris culture supernatant. experimental error. The optimal pH for the activity of

ra-GAL was measured by testing its activity towardPNP-a-GAL at different pH values ranging from 3 to9 (Fig. 2). Maximal ra-GAL activity was displayed atfinal yield (Table I). Thus because of its high level ex-pH 6.4 dropping sharply at pHs higher than 7.0 andpression and simple scale-up procedure, P. pastoris wasa second peak was observed at pH 4.5. Thus ra-GALchosen for production of coffee bean a-GAL needed fordemonstrated pH-dependent activity identical to a-use in the study of B r O blood type conversion.GAL over the whole pH range tested. The dual peaksEluted with a gradient from the S-50 column, ra-exhibited in the pH profile of coffee bean a-GAL areGAL demonstrated a single band of 41 kDa which israrely found among a-GALs from various sources withindistinguishable from the native enzyme as deter-the exception of a-GAL from vicia faba seeds (11). a-mined by SDS–PAGE (Fig. 1). Furthermore, the re-GALs isolated from mung beans (12), pinto beans (13),combinant protein cross-reacted with an polyclonal an-watermelon (14), and human (5) demonstrated a singletibody raised against purified native enzyme on a West-peak between pH 4.6 and 6.0. The catalytic mechanismern blot and its purity was further confirmed byof a-GAL and critical residues involved in the activechromatofocusing, reverse phase HPLC, and masssite are under study although a general acid-base catal-spectrometry (data not shown). Both native and recom-ysis has been proposed for exo-glycosidases (see reviewbinant proteins appeared to be unglycosylated as indi-

cated by the glycosylation analysis (Oxford GlycoSys-tems) (data not shown), although one potential N-glyco-sylation site was identified based on cDNA coding forcoffee bean a-GAL.

2. Physico-chemical Properties of ra-GAL

The structure and enzymatic properties of the puri-fied ra-GAL were compared with the native enzyme.By homologous recombination during P. pastoris trans-formation, the DNA sequence encoding a-GAL was in-tegrated into the host genome, generating yeast trans-formants with variable numbers of copies of the gene(10). In order to rule out the possibility that any muta- FIG. 1. SDS–PAGE of purified a-GAL. ra-GAL-containing super-

natant was obtained by removing P. pastoris cells from the culture.tions had occurred during the plasmid subcloning andAfter concentration, the supernatant (lane 1) was loaded onto a cat-the integration of the a-GAL gene in the P. pastorision exchange column. After washing off the unbound proteins, ra-genome, the total RNA from induced P. pastoris cells GAL was eluted with a NaCl gradient ranging from 50 to 300 mM.

was isolated for RT-PCR. Three DNA fragments, de- 8 mg of a-GAL purified from coffee beans was used as a control inlane 3. Lane 4 is the size marker.signed to cover the entire coding region of the a-GAL



to the stability at pH 7.3, as shown in Fig. 3B. Therecombinant and native a-GALs were apparently in-distinguishable after incubating at pH 5.5 for a pro-longed period at 377C. Over 95% activity remainedafter 50 h incubation and only 15% activity was lostafter 120 h incubation at pH 5.5. Furthermore, at pH5.5 and 267C, which are conditions used for B r Oseroconversion, both na-GAL and ra-GAL are stable

FIG. 2. pH profile of native and recombinant a-GALs. Their activi-ties toward PNP-a-gal were measured at different pHs ranging from3 to 9 according to the procedure described under Material and Meth-ods.

16). Site-directed mutagenesis studies of a-GAL shouldprovide insight on possible amino acid residues respon-sible for the enzymatic catalysis and for the unusualpH dependence observed with a-GAL from coffee beansand vicia faba seeds.

The thermostability of ra-GAL was compared to thatof na-GAL at two different pHs, 5.5 and 7.3. The reasonfor choosing these particular pHs is that physiologicalpH of blood is 7.3 and the enzyme treatment of redcells is undertaken at pH 5.5. As shown in Fig. 3A, atpH 7.3 both ra-GAL and na-GAL were stable at 267C,while the activity decreased gradually at 377C. Underboth conditions, it appeared that ra-GAL was slightlyless stable than the native enzyme. In order to ruleout the possibility that ra-GAL preparation may becontaminated with proteases which are active at pH7.3, we carried out the same thermostability assay inthe presence of mixed protease inhibitors (the ProteaseInhibitor Set from Boehringer Mannheim Biochemica).The result indicated that the enzyme activity droppedat the same rate as the control (without protease inhibi-tors), suggesting that no protease activity was detect-able under the experimental conditions. On the otherhand, as described above, although the primary se-quence of ra-GAL was verified by RT-PCR approxi-mately half of the protein molecules have an extra resi-due, phenylalanine, at the N-terminus while the re-maining half are identical to its native counterpart.Whether such a subtle structural difference has anyimpact on the protein stability has yet to be deter-

FIG. 3. Thermostability of native and recombinant a-GALs. (A)mined. All assays were carried out at pH 7.3 and two different incubationSince we intend to use ra-GAL to treat red cells at temperatures, 267C and 377C. (B) All samples were incubated at 377C

and two different pHs, 5.5 and 7.3.pH 5.5, the enzyme stability at pH 5.5 was compared


328 ZHU ET AL.

for months (data not shown), suggesting that ra- cells by coffee bean a-GAL was observed when the pHwas less than 6.0 (2). While ra-GAL exhibited highGAL, although a mixture with two different N-ter-

mini, can be used in seroconversion of red cells with- activity at lower pH, cells were less stable and lysisbegan to occur. The seroconversion by coffee bean a-out further purification.GAL is carried out at pH 5.6, reflecting a compromisebetween red cell viability and optimal a-GAL activity3. Seroconversion of Red Blood Cells by ra-GAL (3). This variation of the pH with different substrateshas been observed in some other glycosidases, such asThe high-level production of the ra-GAL in P. pas-porcine pancreatic a-amylase (18).toris followed by a simple purification procedure has

A small aliquot of red cells (Ç1 ml) was treated withallowed us to prepare a large quantity of purified ra-ra-GAL under conditions described under MaterialsGAL for an in vivo study. The final enzyme preparationand Methods. Removal of group B antigens, the termi-met all Food and Drug Administration requirementsnal a-linked galactose residue, from the red cell surfaceregarding purity and sterility. One of the concerns re-was observed by using anti-B sera in a hemagglutina-garding the clinical use of recombinant proteins is thetion assay as described by Lenny et al. (8). In order topossibility of DNA contamination (19). In order to de-demonstrate the specificity of ra-GAL toward group Btect possible DNA contamination in the final enzymeantigen on the red cell surface, the treated red cellspreparation, it would be ideal to measure the total P.were subjected to extensive antigen profiling, includingpastoris genome DNA by a hybridization assay; how-tests for A, B, Rh, MNS, P1, Lewis, Kell, Lutheran,ever, it is technically difficult to carry out such an ex-Duffy, and Kidd. As expected, B antigens were unde-periment at desired sensitivity. Since there are eleventectable on the ra-GAL-treated cells. All other antigenscopies of a-GAL gene incorporated in the genome of thepresent on the red cells remained unchanged with theyeast paF-BZ-22 (7), this gene can be used as a markerexception of P1, which is a minor antigen also havingfor DNA contamination in preparations of protein de-an a-linked galactose residue at the nonreducing endrived from the supernatant of the yeast cultures. Inof its carbohydrate chain. Furthermore, no neoantigenorder to determine the sensitivity of the PCR profilewas generated after enzyme treatment. These resultsin detecting any DNA contamination in the purifiedare in agreement with the data derived from red cellsprotein preparation, genomic DNA isolated from P.treated with na-GAL. As part of our clinical trials, 1pastoris cells was used to quantitate the lower limit ofU of group B red blood cells (Ç200 ml of packed cells)the amount of DNA needed to generate a distinct PCRwas treated with ra-GAL for seroconversion under theproduct on a ethidium bromide-stained agarose gel.same conditions as used for the native enzyme. TheUnder the assay conditions, the intensity of the PCR-treated red cells, which became serologically group O,amplified band decreased in a parallel fashion withwere successfully transfused to a healthy group O re-reducing amounts of template DNA applied in PCR,cipient (20).and no band was visible using 1.0 fg DNA. Bovine se-

In summary, our efficient expression and purificationrum albumin, at the same concentration as the a-GALprocedures enable us to readily prepare enzymaticallysample, was added to the template DNA in order toactive ra-GAL in large quantities. It is also likely thateliminate the possible adverse effect of proteins on thethe expression level of the enzyme in P. pastoris cansensitivity of PCR. The a-GAL sample (1.4 U) exhibitedbe increased significantly in a large-scale productionno band following PCR, suggesting that 1.0 fg DNA orsetting. Thus, ra-GAL will be used routinely in placeless is present in 1.4 U of the enzyme which can beof its native counterpart, in the seroconversion fromextrapolated to less than 32 pg DNA per 45,000 U ofgroup B to group O erythrocytes.the enzyme preparation. The acceptable limit of con-

taminant DNA in 1 U of converted blood is less than100 pg according to FDA requirements. Under stan- REFERENCESdardized conditions, upon treatment of 1 U of converted

1. Clausen, H., and Hakomori, S-i. (1989) Vox Sang. 56, 1–20.red blood cells with 45,000 U of a-GAL, the enzyme was2. Harpaz, N., Flowers, H. M., and Sharon, N. (1975) Arch. Bio-efficiently removed (Ç106-fold reduction) by multiple-

chem. Biophys. 170, 676–683.step washings of red cells (4). Therefore the trace3. Goldstein, J., Siviglia, G., Hurst, R., and Lenny, L. (1982) Scienceamount of DNA in the enzyme preparation is likely to 215, 168–170.

be removed as well during the wash procedure.4. Lenny, L. L., Hurst, R., Goldstein, J., and Galbraith, R. A. (1994)

Although ra-GAL demonstrates maximal activity to- Transfusion 34, 209–214.ward the substrate PNP-a-Gal at pH 6.4, the optimal 5. Haibach, F., Hata, J., Mitra, M., Dhar, M., Harmata, M., Sun,pH drops to between 3.6 and 4 if the substrate is melibi- P., and Smith, D. (1991) Biochem. Biophys. Res. Commun. 181,

1564–1571.ose, raffinose, or stachyose (17). The efficient removalof the terminal a-galactose residues from group B red 6. Zhu, A., and Goldstein, J. (1994) Gene 140, 227–231.



7. Zhu, A., Monahan, C. Zhang, Z., Hurst, R., Leng, L., and 14. Itoh, T., Uda, Y., and Nakagawa, H. (1986) J. Biochem. 99, 243–250.Goldstein, J. (1995) Arch. Biochem. Biophys. 324, 65–70.

15. Dean, K. J., and Sweeley, C. C. (1979) J. Biolog. Chem. 254,8. Lenny, L. L., Hurst, R., Goldstein, J., Benjamin, L. J., and Jones,9994–10000.R. L. (1991) Blood 77, 1383–1388.

16. Withers, S. G., and Aebersold, R. (1995) Protein Sci. 4, 361–372.9. Zhu, A., Wang, Z-K., and Goldstein, J. (1995) Biochem. Biophys.17. Courtois, J. E., and Petek, F. (1966) Methods Enzymol. 3, 565–Acta 1247, 260–264.

571.10. Clare, J. J., Rayment, F. B., Ballantine, S. P., Sreekrishna, K.,18. Ishikawa, K., Matsui, I., and Honda, K. (1990) Biochemistry 29,and Romanos, M. A. (1991) Bio/Technology 9, 455–460.

7119–7123.11. Dey, P. M., and Pridham, J. B. (1969) Biochem. J. 113, 49–55. 19. Goldman, M., Geier, M., Zehnder, D., Wang, A., and Henriksson,12. Dey, P. M. (1984) Eur. J. Biochem. 140, 385–390. T. (1991) Clin. Chem. 37, 189–193.

20. Lenny, L. L., Hurst, R., Zhu, A., Goldstein, J., and Galbraith,13. Dhar, M., Mitra, M., Hata, J., Butnariu, O., and Smith, D. (1994)Biochem. Mol. Biol. Int. 34, 1055–1062. R. A. (1995) Transfusion 35, 899–902.


Documents

Characterization of Recombinant α-Galactosidase for Use in Seroconversion from Blood Group B to O of Human Erythrocytes