Purification and Properties of Human a-Galactosidases*

The thermolabile oc-galactosidase (cy-galactosidase A) and thermostable a-galactosidase (cr-galactosidase B) were separated and purified from human placenta. A homogeneous a-galactosidase B preparation was obtained, but the a-galactosidase A preparation contained small amounts of contaminating protein and various other acid hydrolase activities. Each preparation had a molecular weight of approximately as estimated Sephadex filtration. Both


SUMMARY
The thermolabile oc-galactosidase (cy-galactosidase A) and thermostable a-galactosidase (cr-galactosidase B) were separated and purified from human placenta.
A homogeneous a-galactosidase B preparation was obtained, but the a-galactosidase A preparation contained small amounts of contaminating protein and various other acid hydrolase activities.
Each preparation had a molecular weight of approximately 150,000, as estimated by Sephadex filtration.
cY-Galactosidase A had a Km of 3.4 mM for the artil%cial substrate, 4-methylumbelliferyl-cr-n-galactopyranoside, and of 40.6 mu for melibiose. a-Galactosidase B hydrolyzed 4-methylumbelliferyl-cY-n-galactopyranoside with first order kinetics and appeared to have no activity with melibiose. Both enzymes had maximal enzyme activity at pH 4.5, but a-galactosidase A had a broad pa-activity curve, while that of the B enzyme was sharply peaked.
cr-Galactosidase A was inhibited by myoinositol; cr-galactosidase B was not. The isoelectric point of a-galactosidase A was 4.70 =t 0.07; the isoelectric point of oc-galactosidase B was 4.42 f 0.04. Antibodies were produced against both the cY-galactosidase A and a+galactosidase B preparations. No cross reactivity between the two enzyme preparations was found on double immunodiffusion.
Neither antiserum neutralized enzyme activity, but the anti-cr-galactosidase A serum precipitated oc-galactosidase A activity from solution and the anti-crgalactosidase B serum precipitated oc-galactosidase B activity from solution.
Treatment of a-galactosidase A with neuraminidase does not change its immune reactivity or kinetic properties.
These studies lend no support to the concept that cu-galactosidase A is the neuraminyl derivative of galactosidase B or that the two enzymes are closely structurally related.
Fabry's disease is an inborn error of metabolism in which the glycolipid, ceramide trihexoside, accumulates in various tissues (I). Brady et al. (2) showed that normal human small intestinal mucosa had the capacit,y to enzymatically hydrolyze ceramide trihexoside, and that this enzyme activity was lacking in specimens from two patients with this disorder.
Subsequently, a decrease in the capacity of leukocytes and fibroblasts to hydrolyze artificial ar-galactosides was demonstrated in patients with * This work was supported, in part, by Grant AM 14755 from the National Institutes of Health.
Fabry's disease (3). Investigating the biochemical genetics of Fabry's disease, we found that human fibroblasts and human leukocytes contained two a-galactosidases (4,5). The major component, designated cr-galactosidase A, was thermolabile and was absent from fibroblasts and lymphocytes of patient,s with Fabry's disease. The other component, cr-galactosidase B, was somewhat increased in activity in Fabry's disease, was thermostable, and could, in addition, be distinguished from oc-galactosidase A on the basis of its higher K, for the artificial substrate and different electrophoretic mobility (4,5). Similar findings have now also been reported by others (6,7).
The relationship between the two cu-galactosidases, which are found also in other human tissue (8), is of importance in furthering our understanding of the genetic basis of this disease. It has been proposed that ol-galactosidase B is a precursor of cr-galactosidase A (6, 7). We have now undertaken purification of human cY-galactosidase A and a-galactosidase 15 to clarify the relationship between these enzymes and to provide pure enzyme for possible replacement therapy of patients with Fabry's disease. MATERIAL AND METHODS Refrigerated human placentas, no more than 48 hours old, served as starting material.
They were stripped of their outer membranes, cut into 2-to 3-inch pieces, washed with cold 0.9% sodium chloride, blotted on filter paper, and ground in a meat grinder.
Ammonium sulfate fractionation and column chromatography were performed at 4" using standard techniques.
A standard containing 0.1325 pM 4-methylumbelliferone in the glycine buffer was read with each assay. Protein determinations were carried out by the method of Lowry et al. (9) when determining specific activity of enzyme in the crude placental extract and by the ratio of absorbance at 280 nm to that at 260 nm (10) in monitoring column elutions.
Activity was expressed as microunits per mg of protein Cpicomoles of substrate hydrolyzed per min per mg of protein).
Hydrolysis of melibiose was studied in a system containing 0.025 M citrate buffer, pH 4.0; 0.770 albumin; and 5 to 500 mM melibiose.
After incubating for 2 hours at 37", glucose released was measured spectrophotometrically at 340 nm using hexokinase ATP, MgC12, glucose 6-phosphate dehydrogenase, and NADP, measuring the amount of NADP reduced.
Isoelectric focusing was performed by the method of Vesterberg and Svensson (11) with LKB 8100 electrofocusing equipment with a 110.ml column.
The ampholyte concentration was 1% with a pH range from 4 to 6 in a sucrose gradient.
The column jacket was cooled to O-l".
Voltage was started at 500 volts and increased to 700 volts after 4 hours.
The run was completed in 44 to 48 hours and eluted in 1.7-to 2.0.ml aliquots.
Acrylamide electrophoresis was carried out with a Polyanalyst apparatus (Buchler Instruments) employing the method of Simons and Bearn (12) with 7% acrylamide gels modified to give a running pH of 7.2 with 0.01 M potassium phosphate buffer. The electrophoresis was performed at 2 ma per gel for 90 to 100 min. Proteins were stained by placing the gels in 0.5% Amido black in 7% acetic acid for 30 to 60 min followed by electrophoretie destaining.
Enzymatic activity was detected by incubating the gel in 5 mM 4-methylumbelliferyl-c-n-galactopyranoside in 0.5 M sodium citrate buffer, pH 4.0, for 30 min at 37" followed by replacement of the buffered substrate solution with 1 M glycine buffer, pH 10.7 in order to visualize, under long wave ultraviolet light, fluorescent bands representing or-galactosidase activity.
Neuraminidase treatment of enzyme preparations was done by incubating the enzyme in 0.04 M acetate buffer, pH 5.0, with 0.25 mg per ml of Clostridium perfringens neuraminidase (Sigma Chemical Company) for 60 min at 37". The mixture was then placed on ice and an equal volume of 0.2 M potassium phosphate buffer, pH 7.0, was added.

Purijkation Preliminary
Concentration of ol-Galactosidase from Human Placenta-Ground placenta was homogenized at a concentration of approximately 25% (w/v) in 0.15 in potassium chloride using a Sorvall Omni-Mixer at high speed for 5 to 10 min. In order to determine the amount of the two isozymes present in the original starting material, thermal stability assays were done on the crude homogenate.
The amount of stable enzyme, extrapolated to zero time, represented the initial activity of cy-galactosidase B and the difference of this from the total initial activity was then the a-galactoside A activity. The homogenate was centrifuged at 5,000 x g for 30 min. A 25 to 50Y0 ammonium sulfate cut was taken on the supernatant, dissolving the protein cake in 0.01 M potassium phosphate buffer at pH 6.5. The solution was dialyzed for 20 to 24 hours against the same buffer and the precipitate removed by centrifuging at 75,000 x g for 45 min. At this point there is a 6-to 7-fold purification of total cY-galactosidase activity with a 25 to 307, yield from crude, ground placenta.
Separation of a-Galactosidase A and or-Galactosidase B-DEAEcellulose (DE52, Whatman) was equilibrated with 0.01 M potassium phosphate buffer, pH 6.5, in a column 2.5 X 45 cm. The supernatant from the dialyzed enzyme was applied, and the column was washed with approximately 150 ml of buffer. The enzyme was then eluted with a 700-ml sodium chloride gradient, 0 to 0.5 M, in the same buffer. The gradient was prepared using seven chambers of a gradient maker (Buchler Instruments) with 100 ml of the following concentrations of NaCl in 0.01 M potassium phosphate buffer, pH 6.5; 0, 0.04 M, 0.08 M, 0.12 M, 0.16 M, 0.20 M, 0.50 M. As shown in Fig. 1, this procedure resulted in complete separation of two peaks of cr-galactosidase activity.
The first peak of activity eluted at approximately 0.08 M sodium chloride was found to be thermolabile.
This peak represented a-galactosidase A activity. The second peak, eluted at approximately 0.15 M sodium chloride, was thermostable and represented ol-galactosidase B activity.
Further Purification of a-Galactosidase A (Table I)  5 IBM potassium phosphate buffer, pH 6.0. The dialysate was allowed to pass slowly through a bed volume of 100 ml of CMcellulose (ClM52, Whatman) which was previously equilibrated with the pH 6.0 buffer in a 500-ml Buchner funnel.
The CMcellulose was washed with an equal volume of the buffer, and the washing and sample eluate containing the enzyme were combined.
The CM-cellulose-treated fraction was then concentrated by collecting the protein precipitated between 30 and 50% (NH& SOc saturation. Ascending chromatography of the redissolved precipitate was carried out on G-200 Sephadex in a column, 2.5 x 36.5 cm, with a flow adaptor (Pharmacia).
The enzyme was eluted at a position corresponding to a molecular weight of 150,000.
The peak enzyme activity was pooled and applied to a column of ECTEOLA-cellulose1 (Sigma Chemical Company) which had been equilibrated with the pH 7.5 buffer. Chromatography was carried out in a column 1.5 X 30 cm, with enzyme elution with 400 ml of a 0 to 0.2 M sodium chloride gradient: the first chamber of the gradient maker contained 100 ml of buffer followed by 100 ml each of 0.05 M, 0.1 M, and 0.2 ~1 sodium chloride in buffer. The peak of enzyme activity was concentrated by ultrafiltration for use in immunological studies and determining properties of the enzyme.
Further PuriJication of a-Galactosidase B (Table II)-The fractions comprising the second peak of the DEAE-cellulose chromatography were pooled and dialyzed against 5 mM potassium phosphate buffer, pH 6.0. A 35 to 50% saturated ammonium sulfate fraction was subjected to G-200 Sephadex ascending chromatography, as described in the purification of the h enzyme except that the pH 6.0 buffer was used. The enzyme was eluted at a position corresponding to a molecular weight of 150,000. The fractions comprising the peak of enzyme activity were then pooled and subjected to isoelectric focusing.
After dialysis against 0.01 M potassium phosphate buffer, pH 7.5, ECTEOLA-cellulose chromatography was performed as described in the purification of the A enzyme.
The or-galactosidase B was concentrated by ultrafiltration and used for immunological studies and determining properties of the enzyme.

Purity of Enzyme Preparation
Purity of the final enzyme preparations was tested with polyacrylamide disc electrophoresis.
The a-galactosidase B enzyme showed a single band when staining both for enzymatic activity and for protein.
The a-galactosidase A preparation showed one major and two to four minor bands when stained for protein.
The single band of enzyme activit,y corresponded to the major protein band.
Although protein determinations on the crude extract gave essentially identical values whether measured by the 280 : 260 nm method or Lowry's method, protein estimation based on the 280:260 nm method on either of the purified preparations gave values approximately 3-to 3.5.fold higher than when measured by the Lowry technique.
Study of distribution of these acid hydrolases in the fractions of the final (ECTEOLA-cellulose) chromatographic step demonstrated that the position of elution of the hydrolases was not identical with that of ac-galactosidase A, indicating that these represented small amounts of contaminating enzymes rather than activities intrinsic to the a-galactosidase A preparation.
a-Galactosidase B appeared to have first order kinetics with 4-methylumbelliferyl-a-n-galactopyranoside and was unable to hydrolyze melibiose even at concentrations up to 500 mM. The apparent first order relationship between 4-methylumbelliferyl-cY-n-galactopyranoside concentration and the velocity of the cu-galactosidase B reaction could well be more apparent than real. Such apparent first order kinetics are observed if the K, of the enzyme for substrate is high and all levels tested are well under the half-saturating concentration. Assay of the enzyme activity at higher substrate concentrations was 7198 galactopyranoside as substrate to that with ceramide trihexoside as substrate to be approximately 6.4. pH-activity curves (Fig. 3) were determined using 0.02 M citrate buffer at pH 3.0 to 6.5. Both enzymes manifest maximal activity at pH 4.5; however, while cr-galactosidase A shows a broad curve, the B enzyme gives a rather sharp peak of activity. Fig. 4 presents the effect of temperature on ar-galactosidase A and oc-galactosidase B as an Arrhenius plot. It is evident that the characteristics of the two enzymes are quite different.
The energy of activation of ar-galactosidase A was approximately 15,700 cal per mole, while a-galactosidase B produced a very unusual plot manifesting optimal activity at 25". The low activities observed at higher temperatures were obviously not due to instability of the enzyme, since ol-galactosidase B is quite stable even at 50", and were probably due to a conformational change in the enzyme.

7199
The separated enzymes after DEAE-cellulose chromatography untreated enzyme. Neuraminidase-treated cu-galactosidase A were tested for inhibition by myoinositol (13). oc-Galactosidase did not react with anti-B serum in the double immunodiffusion A was found to be 43% inhibited when concentrations up to system, and enzyme activity was not precipitated from solution 750 mM were used. cr-Galactosidase B was not inhibited by by the antiserum. Its reaction wit,h anti-A serum was unthis concentration of myoinositol but showed a slight increase altered. in activity.
The purified preparations of both enzymes were found to require albumin in order to maintain activity on assay. DISCUSSION

Isoelectric Points
Fibroblasts from patients with Fabry's disease were shown to contain residual a-galactosidase activity which was different Isoelectric focusing revealed the isoelectric points to be 4.70 from those of normal cells (4)(5)(6)(7)14). This was originally inf 0.07 (mean & 1 S.E.) for a-galactosidase A and 4.42 f 0.04 for ar-galactosidase B. This was not used as a purification step for the A enzyme because of an excessive loss of enzyme activity in this procedure.

Immunological Studies
The final enzyme preparations were mixed with equal parts of complete Freund's adjuvant to obtain an emulsion and injected into the hind foot pads of rabbits.
Protein injected ranged from 15 to 50 pg. Injections were given approximately every 2 weeks, and the rabbits were bled from an ear vein at least 7 days following the last injection.
The serum was heated at 56" for 30 min and then filtered through 0.2-p Millipore filters. Serum from rabbits injected with the purified ac-galactosidase A preparation was designated as "anti-A serum"; serum from the rabbits injected with purified a-galactosidase B was designated as "anti-B serum"; and serum from normal rabbits or rabbits injected only with Freund's adjuvant was designated as "control serum." When tested with standard double immunodiffusion techniques, anti-A serum reacted with ar-galaetosidase A, but not with cr-galactosidase B preparations.
Conversely, anti-B serum reacted with ar-galactosidase B and not with ar-galactosidase A.
Mixing antiserum with enzyme had no appreciable effect on the activity of either cr-galactosidase A or B, but when the mixture of enzyme and antiserum was diluted 1: 10 in 10 mM phosphate buffer, pH 7, and centrifuged at 40,000 x g for 1 hour, anti-B enzyme removed a-galactosidase B activity and anti-A serum removed a-galactosidase A activity. A 1:4 dilution of anti-B serum removed SOolc of a-galactosidase B activity and no cr-galactosidase A activity; a 1:4 dilution of anti-A serum removed 75*1$ of a-galactosidase A activity and no a-galactosidase B activity from solution.
IJndiluted anti-B serum did appear to remove approximately 10% of A enzyme from solution, but this capacity was lost evcu by 1:4 dilution.
The ol-galactosidase activity of fibroblasts from a patient with Fabry's disease was almost completely removed by treatment with anti-B serum, while only a small proportion of enzyme from normal fibroblasts was removed by this antibody.

Effect of Neuraminidase
Treatment on Properties of cu-Galactosidase A terpreted as indicating that Fabry's disease represented a structural mutation of the cr-galactosidase locus, and that an abnormal enzyme was produced (14). Subsequently, however, we were able to show that the residual enzyme in Fabry's disease fibroblasts seemed to be identical to a minor a-galactosidase component found also in normal cells. We named the normal thermolabile cr-galactosidase which was missing in Fabry's disease cY-galactosidase A and designated the other, thermostable component Ly-galactosidase B (4,5). Wood and Nadler (6) and Ho et al. (7) suggested, however, that ol-galactosidase B was also abnormal and that there might be a close genetic relationship between oc-galactosidase A and B. It was proposed that cr-galactosidase B might be the precursor of a-gdactosidase 8, the latter enzyme representing the neuraminyl derivative of a-galactosidase B.
Separation and purification of these two forms of cr-galactosidase have now made it possible to study separately their properties and to produce an antibody against each of these forms. These investigations showed that the biochemical properties of or-galactosidase A and B correspond fairly closely to the properties deduced from studies on crude extracts of normal fibroblast and fibroblasts from patients with Fabry's disease (5). a-Galactosidase A is, indeed, thermolabile and is found to have a K, of 3.4 mM for 4-methylumbelliferyl-cr-galactopyranoside, a K, of 4.9 mM having been estimated previously (5). cr-Galactosidase B, previously thought to have a K, of about 20 m&f for 4-methylumbelliferyl-oc-galactopyranoside (5), apparently reacts with this substrate with first order kinetics.
Since both enzymes appear to have the same molecular weight on Sephadex filtration, the possibility that one isoenzyme represents a polymer of the other is ruled out. It was previously predicted (5) that Lu-galactosidase B, which is active in Fabry's disease, would have no activity with respect to the glycolipid which accumulates in this disease. Indeed, although or-galactosidase A was found to be active with ceramide trihexoside, no activity was found with a-galactosidase B. It is of interest, in this respect, that cu-galactosidase A was also found to be active against melibiose, the cr-galactoside of glucose, while a-galactosidase B manifested no activity against this compound.
In the case of another glycolipid storage disorder, Tay-Sachs disease, there is reason to believe that the fundamental defect may be failure to transform the B isozyme of hexosaminidase to Since it has been suggested that ar-galactosidase B might be the A isozyme (15,16). Recently, it has been found that antithe aneuraminyl derivative of cr-galactosidase A, ac-galactosidase bodies produced against either isozyme of hexosaminidase react A preparations were treated with C. perjringens neuraminidase, also with the other isozyme (17,18). This was not found to be and the properties of the converted enzyme were studied.
Ear-the case with or-galactosidase: antibody against a-galactosidase lier investigations had shown that such treatment alters the A showed no reaction against cY-galactosidase B and antiserum electrophoretic mobility of cr-galactosidase A (7,8) and alters its to the B enzyme gave only minimal reaction with cY-galactosidase isoelectric point to resemble that of cr-galactosidase B (7). The A. Furthermore, removing neuraminic acid from cY-galactothermal stability of cy-galactosidase A was unaltered by neura-sidase A did not convert it to a form which made it active against minidase treatment.
The K, of the treated enzyme for 4-the anti-cY-galactosidase B antibody. Neither did removal of methylumbelliferyl-c-n-galactopyranoside was 2.3 mM, and that neuraminic acid from a-galactosidase A change its kinetic propfor melibiose was 37.1 InM, in close agreement with that of the erties with respect to 4-methylumbelliferyl-a-n-galacto-7200 pyranoside or melibiose nor alter its sensitivity to heat inactivation. Thus, it is clear that a-galactosidase B is not merely the aneuraminyl derivative of a-galactosidase A. Presumably, this enzyme does have a natural function as an cr-ga.lactosidase, but as yet its natural substrate remains unknown.
We have previously presented indirect evidence that cr-galactosidase B is not merely an associated hydrolytic activity of one of the other acid hydrolases (5), and t'he present studies show that the homogeneous preparation is free of fi-glucosidase, fl-galactosidase, ,&hesosaminidase, acid phosphatase, and cy-mannosidase activities.
Efforts to t.reat Fabry's disease by enzyme replacement have consisted of plasma infusion (19) and of kidney transplantation (20, 21). The cr-gala.ctosidase -1 activity of normal human plasma is approximately 145 microunits per ml. Thus, 2.2 liters of plasma or approximately 8 pints of whole blood would have to be infused to provide the amount of enzyme activity in 1 mg of our final placental cr-galactosidase h preparation.
Since the infusion of purified placental enzyme is a much simpler procedure than kidney transplantation, and normal human placentas are freely available, it is possible that purified a-galactosida.se may prove to be of some value in the treatment of this serious disorder.
It must be emphasized, however, that even infusions of massive amounts of or-galactosidase h may prove to be of 110 benefit to paCents with Fabry's disease. For exogenous enzyme to be of help to these individuals, it would be necessary for the infused enzyme to reach the sites of ceramide trihexoside accumulation and t.o survive degradation for a sufficiently long period of time to remove accumulated glycolipid.
Further studies a.re needed to determine whether such replacement therapy is feasible.