Purification and Initial Characterization of Intrinsic Membrane-bound Alkaline Phosphatase from Chicken Epiphyseal Cartilage *

Alkaline phosphatase has been purified from microsomes of chicken epiphyseal cartilage by first selectively extracting certain adventitious proteins with 0.25 M trichloroacetate. The membrane-bound enzyme was then solubilized by 1% cholate in buffered 33% saturated ammonium sulfate and purified by column chromatography on Bio-Gel A-5m, extraction with 1butanol, and ion exchange chromatography on DEAEBio-Gel A. The purified alkaline phosphatase from the cartilage membrane had a subunit molecular weight of 53,000 and a holoenzyme weight of 207,000-220,000, indicating a tetramer. The pH optima forp-nitrophenylphosphate, ATP, and pyrophosphate hydrolysis were 10.3, 9.0, and 8.5, respectively.  Values of Vmax (in micromoles/min/mg) were 220,3.1, and 0.8, respectively. Substrate inhibition was pronounced at values of pH below 8.5. Inhibition of p-nitrophenylphosphate  hydrolysis at pH  10.3 showed that phosphate and arsenate were competitive inhibitors (KI = 1.88 and 0.15 m ~ , respectively) and levamisole was an uncompetitive inhibitor (KI = 0.32 m), while L-phenylalanine and ZnClz were mixed inhibitors (K, = 15.8 and 0.02 m ~ , respectively). Inhibition by preincubation in 1 m~ EDTA was reversible by readdition of 0.25 m~ MgClz and 20 pM ZnClz. The data indicate that this membrane-bound alkaline phosphatase from chicken epiphyseal cartilage is a Zn2+ and possibly Mgz+-containing enzyme. While the subunit molecular weight and kinetic properties of the enzyme are quite typical of vertebrate alkaline phosphatases, the tightness of binding to the membrane lipids, the extreme sensitivity to substrate inhibition, and the tetrameric conformation of the holoenzyme are unusual.

Alkaline phosphatase in matrix vesicles appears to be very tightly associated with the membrane as an intrinsic membrane component. It has been isolated previously from matrix vesicles and chondrocytes that were obtained by crude-collagenase digestion of epiphyseal cartilage (11). The extremely low subunit molecular weight (M, = 18,000) and the instability of the enzyme following treatment with metal ion chelators, however, suggests that the enzyme had been altered by proteases known to contaminate the crude collagenase (12, 13). While enzyme activity is not destroyed by treatment of the tissue with proteases, it is likely that this nevertheless damages the noncatalytic portion of the protein. This may affect characteristics critical to its function in calcification Alkaline phosphatase has been isolated from rabbit fracture-callus cartilage which had not been treated with proteases (14). Two antigenically distinct forms were obtained. However, because of the method employed (ie. extraction of whole tissue), it was impossible to ascertain the subcellular source of the two forms. While one may have been derived from chondrocytes and the other from matrix vesicles, it is equally possible that one form may be a soluble and the other, a membrane-associated form of the enzyme. This was not explored.
Recently, we found that microsomes isolated from chicken epiphyseal cartilage by non-protease-dependent methods not only were able to support rapid mineralization using physiological synthetic cartilage lymph but also were enriched &fold in alkaline phosphatase (15). Mineralization of this fraction was significantly retarded by inhibitors of this enzyme (15,16). We now report the isolation and initial characterization of this alkaline phosphatase, an integral protein of the microsomal membranes of chicken epiphyseal cartilage. This work represents the fwst successful isolation of the membranebound enzyme from mineralizing cartilage, without resort to the use of proteases. The enzyme, unlike that isolated from protease-treated tissue, has a signifcantly higher subunit molecular weight and different holoenzyme structure (17). The catalytic activity (pH optima, kinetic behavior, response to inhibitors, etc.) are, however, similar to that of other bone alkaline phosphatases (14, 18, 19). The enzyme is unusual in that it is a tetramer, and displays physical characteristics typical of an intrinsic membrane protein.

Purification of Alkaline Phosphatase
All steps prior to the chromatography columns were performed at 0-4 "C unless otherwise noted. The ammonium sulfate solution was saturated at room temperature and all per cent saturations are relative to this. Chromatographic columns were operated at room temperature. Protein concentrations were estimated by the procedure of Lowry et al. (20).
Isolation of Microsomes-Slices of epiphyseal cartilage (100-120 g wet weight) from the proximal end of the metatarsus of 8-to 10week-old broiler-strain chickens were obtained by a published procedure (21). The slices were suspended in 175 ml of an isolation medium composed of 10% (w/v) sucrose, 50 mM Tris-HC1 (pH 8.0) and 2.5 mM MgC12, and homogenized for 6 min with a Tekmar Tissuemizer equipped with a type SDT-180EN probe. Differential centrifugation was employed as previously described (13). The pellet from the 85,000 X g centrifugation was resuspended in the isolation medium with the aid of a Potter-Elvehjem homogenizer and adjusted to a total volume of 15 ml.
Solubilization-To the microsomal pellet suspension was added, at room temperature and with vigorous stirring, 0.11 volume of 2.5 M sodium trichloroacetate (pH 8) to give a final concentration of 0.25 2. This was stirred for 30 min and then centrifuged at 85,000 X g for 60 min. The supernatant was discarded and the pellet was resuspended to a total volume of 10 ml with isolation medium.
The phosphatase was extracted by dropwise addition of 0.5 volume of saturated ammonium sulfate followed by 0.08 volume of 20% (w/v) potassium cholate (pH 8). The mixture was stirred for 15 min and then centrifuged at 85,000 X g for 60 min. The pellets were discarded and the supernatants were saved. The enzyme can be frozen overnight at this stage, with no apparent loss of activity.
Gel Filtration-The cholate-soluble material was precipitated by the addition of an equal volume of saturated ammonium sulfate. It was stirred in an ice bath for 15 min and then centrifuged at 120,000 X g for 60 min. The subnatants were removed by puncturing the bottoms of the tubes. The coagulum was dissolved in isolation medium (total volume, 5.5 ml), any insoluble material being removed by centrifugation at 120,000 X g for 30 min. The clear supernatant was applied to a Bio-Gel A-5m column (3.2 X 50 cm) which had been equilibrated with a solution composed of 0.5% (w/v) cholic acid, 50 m~ Tris-HC1 (pH 8.0), and 2.5 mM MgCI,. The column was eluted with the same buffer and the peak activity fractions were pooled.
Ion Exchange Chromatography-The pooled fractions from the Bio-Gel A-5m column were extracted at room temperature with an equal volume of I-butanol by vortexing seven times at 2-min intervals. The aqueous phase was collected and diluted with an equal volume of distilled water. The diluted enzyme was applied to a DEAE-Bio-Gel A column (0.8 X 8 cm) which had been equilibrated with a 50 mM Tris-HC1 (pH 8.0), 2.5 m~ MgCL buffer. The column was washed with the equilibration buffer until the absorbance at 280 nm returned to the base-line, the phosphatase being eluted with a linear gradient composed of the above buffer and 1.0 M NaCl in the same buffer (50 ml each). The active fractions were pooled and concentrated using an Amicon YM-10 membrane.

Kinetic Determinations
Hydrolysis ofp-Nitrophenylphosphate-Assays were performed at 37 "C in a GCA/McPherson model EU-701 spectrophotometer set at 410 nm. The instrument was operated in the single beam mode with the display updated every 5 s. Absorbance readings were entered into a Wang model 2200s microcomputer until 18 points (high activity) or 54 points (low activity) were collected and the slopes obtained by a linear least squares fit. Activities were converted from AA410/s to micromoles/min by the use of extinction coefficients determined by dilution of a stock solution of p-nitrophenol with the appropriate buffer.
Hydrolysis of ATP-This was measured by determining P, released. Tubes containing 1.5 ml of buffer and about 0.5 pg of enzyme protein were preincubated at 37 "C for 10 min. A 5O-pl aliquot of an ATP solution was then added and the incubation continued for 30 rnin. The reaction was terminated by the addition of 50 p1 of 70% perchloric acid. Pi was separated and determined by the isobutanol/ benzene extraction method of Martin and Doty (22). Correction for nonenzymatic hydrolysis was made through the use of simultaneous incubations containing no enzyme.
Hydrolysis of PPi-This was assayed by the method described for ATP substituting [32P]PPi as substrate. The amount of 32Pi produced was determined by liquid scintillation spectrometry of an aliquot of the phosphomolybdic acid-containing organic phase. Plastic tubes were used for all assays to minimize association of PPi and P, with the walls of the tubes.
Estimation of Kinetic Constants-The data from the activity measurements were analyzed by a BASIC language translation of the HYPER fit program developed by Cleland (23). All fits were based on 5-8 substrate concentrations with velocities determined in triplicate at each concentration. insoluble "p-Nitrophenylphosphate hydrolysis at pH 10.3 (see "Materials and Methods").

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis
This was performed in slab gels (130 X 200 X 1.5 mm) essentially as described by OFarrell(24), using gels containing 12.5% acrylamide. Samples from each of the various purification steps were lyophilized and dissolved in 50 pl of sample buffer. The cholate-soluble and Bio-Gel eluate fractions were dialyzed overnight against 100 ml of sample buffer to remove excess salt. Electrophoresis was carried out at 35 mA until the tracking dye (bromphenol blue) was within 5 cm of the bottom. The gel was stained for protein by the method of Blakesley and Boezi (25).

Nondissociative Polyacrylamide Gel Electrophoresis
A discontinuous system, essentially as described by Davis (26) was used. Samples of purified alkaline phosphatase (10 pg of protein) were applied in the buffer with which they were eluted from the DEAE-Bio-Gel A column. Other proteins (hemoglobin, catalase, and aldolase) were dissolved in 50 mM Tris-HC1 (pH 8.0). The running buffer was 0.192 M glycine, 25 mM Tris. Electrophoresis was carried out using the same type of slab gels, tracking dye (bromphenol blue), and protein staining as described above.

Assay of Column Fractions
Column fractions were assayed in 0.7 mMpNPP,' 0.125 M 2-amino-2-methyl-I-propanol buffer (pH 10.3) at 37 "C. Assays were started by addition of enzyme and absorbance at 410 nm was monitored continuously with a GCA/McPherson model EU-701 spectrophotometer.

Sucrose Density Gradient Centrifugation
Purified alkaline phosphatase was layered onto a continuous sucrose density gradient (5-20%, w/v, dissolved in 50 mM Tris-HC1 buffer, pH 8.0, containing 0.1 M NaCl and 2.5 mM MgC12). Centrifugation was carried out for 12 h at 40,000 rpm using a SW 50.1 rotor.
Standards (bovine serum albumin, catalase, and lactic acid dehydrogenase) were run on parallel gradients. Fractions of 0.1 ml were collected and assayed by standard procedures for catalase (27) or lactic acid dehydrogenase (28) activity. The bovine serum albumin peak was detected by protein concentration. Sedimentation data were analyzed by the method of Martin and Ames (29).

Chemicals
Levamisole, L-phenylalanine, p-nitrophenol (25 mM standard solution), cholic acid, acrylamide, N,N'-methylenebisacrylamide, N,N,N',N',-tetramethylethylenediamine, and 2-amino-2-methyl-lpropanol were obtained from Sigma;pNPP was from Technicon; Bio-Gel A-5m and DEAE-Bio-Gel A were products of Bio-Rad; AcA34 was obtained from LKB Produkter; and [3zP]PP, was purchased from New England Nuclear. AU other chemicals used were of reagent grade and were supplied by Fisher. Table I  the microsomal pellet was achieved. Recovery of activity from the microsomal pellet ranged from 15-25%. The fact that only a 30-fold purification was required to achieve homogeneity suggests that a substantial proportion of the protein in the microsomal pellet was alkaline phosphatase. Since the microsomal pellet itself provided a 5-fold enrichment of alkaline phosphatase activity relative to the supernatant from the 600 X g centrifugation ( E ) , the total purification from the tissue was greater than 150-fold.

Purification of Membrane-bound Alkaline Phosphatase-
A typical elution profile from Bio-Gel A-5m is shown in Fig.  1. The hydrolytic activity did not correspond to any discrete peak with absorbance at 280 nm. A large degree of purifkation was obtained at this stage, however. Extraction with 1-butanol could not precede this step as the enzyme was then inactivated. In addition, it was critical that cholate be included in the eluting buffer to prevent irreversible aggregation of the enzyme and concomitant loss of activity.
Ion exchange chromatography on DEAE-Bio-Gel A (Fig. 2) resulted in the removal of the final contaminating proteins. Dilution of the 1-butanol-extracted aqueous phase was necessary in order for the enzyme activity to bind to the column. The inability of the enzyme to bind appeared to be caused by The results of sodium dodecyl sulfate-polyacrylamide gel electrophoresis of aliquots from each of the purification steps are shown in Fig. 3. There was no obvious reduction in the number of protein bands following extraction of microsomes with 0.25 M trichloroacetate; however, incorporation of this step was found to be crucial. Its omission resulted in the presence of several contaminating proteins after the DEAE-Bio-Gel A step and these could not be removed by varying the conditions for elution from the columns or by inclusion of a concanavalin A-Sepharose chromatography step.
Physical Characterization-Calibration of the sodium dodecyl sulfate-polyacrylamide gel electrophoresis yielded a molecular weight (M,) for the subunits of this alkaline phosphatase of 53,000, a value near the range reported for other vertebrate alkaline phosphatases (Mr = 55,000-70,000) (17).
The molecular weight of the holoenzyme was estimated to be 207,000 by sucrose-density gradient ultracentrifugation, and 220,000 by gel filtration on AcA34 (data not shown). These findings suggest that the enzyme exists as a tetramer in its native configuration.
Electrophoresis of the enzyme under nondissociative conditions was also performed using a graded series of acrylamide concentrations. When these results were analyzed through the use of Ferguson plots (30), an apparent molecular weight of approximately 80,000 was obtained (Fig. 4). The low molecular weight observed by this method may be caused by a high charge density for the enzyme and/or the loss of metal ions  normally part of the enzyme structure (17), allowing the subunits to dissociate.
Kinetics-The pH profiles for ATP, PPi and pNPP hydrolyses are shown in Fig. 5. As expected, the pH optimum was dependent on the substrate used. With pyrophosphate the pH optimum was 8.5. At this pH the Vmax was 0.8 pmol of PPi hydrolyzed/min/mg of protein, the K, was 85 p~. The most alkaline pH optimum (10.3) was obtained with pNPP. For this substrate the Vmax was 220 pmol/min/mg of protein, with a K,,, of 0.7 mM. The optimum for ATP was intermediate at pH 9.0, its values for V,,, (3.1 pmol/min/mg of protein) and K,,, (0.23 mM) also being intermediate between those for PP, and pNPP. Substrate inhibition with PPi was quite marked substrate concentrations had to be maintained below 100 p~ to obtain linear reciprocal plots. With ATP, concentrations up to 0.5 mM could be used. When pNPP hydrolysis was measured, substrate inhibition was not observed with 1 mM substrate, except when the pH was below 8.5. At these lower pH values the highest concentration used was 50 p~.
The apparent optimum pH was affected by the concentration of the substrate, as is shown in Fig. 6 for hydrolysis of pNPP. This was due to several effects, primarily the increase in substrate inhibition and the decrease in K,,, at decreasing pH. In Fig. 7 , A and C, the log ( Vmax/Kn) and (-)log K,, respectively, are plotted as functions of pH. The K , forpNPP hydrolysis decreased by almost 4 orders of magnitude between pH 10.5 and 7.5. In the same range Vmax changed by only slightly more than 2 orders of magnitude (Fig. 7B). These effects combine to shift the apparent pH optimum toward neutrality at low substrate concentrations. Plots of log ( VmaX/ K,) uessus pH are generally considered to reflect two features of the interaction between the substrate and the enzyme: 1) the inflection points occur at the apparent pK, of the functional groups involved, and 2) the magnitude of the slope reflects the number of groups involved in substrate binding. The observed plots (Fig. 7A) are consistent with the involvement of two ionizable groups. One is evident in the plots for all three substrates with an apparent pK, between 8 and 8.5.
A second group with a pK, between 9.5 and 10.0 can be seen in the plot for pNPP. Since these two apparent pK, values are near each other, the values should be considered to be upper and lower limits for the lower and upper pK, values, respectively. The effects of various inhibitors of pNPP hydrolysis are summarized in Table 11. Sodium phosphate was a competitive inhibitor (KI = 1.88 mM, Fig. 8) as was sodium arsenate, a   potent inhibitor tested, however, based on plots of l/Vmax versus inhibitor concentration, was ZnCl2 with a Kr of 0.02 mM.
Neither MgC12 nor CaClz was found to have any significant effect on the activity of the purified epiphyseal alkaline phosphatase at concentrations up to 1 mM (data not shown). Preincubation of the enzyme (approximately 25 pg protein/ ml) with 1 mM EDTA at pH 8.0 for 15 min at room temperature resulted in the loss of greater than 95% of the activity. However, approximately 70% of the activity could be restored by the addition of 20 PM ZnClp (an amount equimolar with the EDTA transferred from the preincubation) and 0.25 mM MgC12 to the assay mixture. MgC12 alone was not sufficient to restore activity. This observation is in marked contrast with the studies of Fortuna et al. (31,32) in which their chondrocyte and matrix vesicle alkaline phosphatases were irreversibly inactivated by EDTA.

DISCUSSION
This paper describes the isolation and characterization of a membrane-bound alkaline phosphatase from microsomes prepared from mineralizing epiphyseal cartilage. The enzyme is tightly bound and can only be solubilized by treatment of the membranes with detergents. Since neither treatment with high ionic strength salt solutions nor the chaotrope, trichloroacetate, solubilizes it, this enzyme is properly classed as an intrinsic membrane protein.
The epiphyseal cartilage alkaline phosphatase which we describe here is a tetramer composed of apparently identical subunits with a M, of 53,000. While this subunit molecular weight is similar to that reported for other vertebrate alkaline phosphatases (17), the oligomeric structure is significantly different in that the typical structure is a dimer. A tetrameric alkaline phosphatase from pig kidney with subunits of M, = 39,000 has been described by Wachsmuth and Hiwada (33). However, Ramaswamy and Butterworth later reported this enzyme to be a dimer with a total molecular weight of about 185,000 (34). While it is possible that aggregation of our cartilage enzyme may have occurred, as is known to happen with the placental enzyme (35), this appears unlikely considering the low protein concentrations employed (less than 50 pg/ml). Further, 0.1 M NaCl was included in the buffers during determinations of molecular weights by the hydrodynamic method to minimize the possibility of aggregation. In an experiment where the salt was omitted, alkaline phosphatase activity eluted from a Bio-Gel A-15m column in several distinct peaks corresponding approximately to multiples of the tetrameric enzyme, the largest appearing to be of 16 subunits. In the presence of 0.1 M NaCl the enzyme activity eluted as a single symmetrical peak corresponding to the tetrameric configuration.
As noted in the introduction, we feel that several characteristics of matrix vesicles isolated from crude collagenase-digested epiphyseal cartilage indicate that they have been damaged by this protease treatment. These include the necessity of high Ca X Pi ion products ( 7 ) and/or high pH (7, 8,10) and/or the presence of alkaline phosphatase substrates (7-10) to initiate significant 45Ca uptake by the isolated vesicles. The alkaline phosphatase isolated from such vesicles also revealed evidence of proteolytic damage (11). For example, the active subunits had a very low molecular weight (Mr = 18,000), although the total weight of the active and inactive subunits (63,000, assuming a 1:1 relationship) would be well within the range of values for mammalian alkaline phosphatases. Further, the instability reported for the enzyme when Zn2+ was removed by EDTA treatment (31-33) probably also stems from proteolytic scission of the enzyme. Removal of the metal ion apparently allows the enzyme fragments to irreversibly denature, much as occurs with insulin when its disulfide bonds are cleaved. It should be recalled that the membrane-bound enzyme we report in this paper could be renatured after EDTA treatment by the readdition of Zn2+ and M$'. Our findings thus suggest that the molecular weight of the epiphyseal cartilage alkaline phosphatase is not greatly different from that of enzymes isolated from other vertebrate sources, in agreement with findings of Arsenis et al. (14).
In several respects the phosphatase I from rabbit cartilage reported by Arsenis et al. (14) is similar to the membranebound enzyme which we have isolated from chicken cartilage. Assuming a dimeric structure for their enzyme (no subunit determination was reported), a M , of 63,000 would be obtained for the subunits of their enzyme. This would be in reasonable agreement with the value reported here, the small difference perhaps being a species variation. Substrate specificities of both enzymes are similar in that both are active toward PP,, ATP, and pNPP with similar pH optima. Nevertheless, a significant difference is apparent in that their enzyme required the addition of M$+ to the assay medium, while the alkaline phosphatase we isolated from chicken epiphyseal cartilage was fully active in the absence of exogenous Mg2' ions.
This membrane-bound enzyme showed kinetics quite typical of other alkaline phosphatases. For example, the pH optimum varied with the type of substrate, ranging from pH 10.3 for pNPP to about 8.5 for PPi. Inactivation by preincubation with EDTA and reconstitution by the addition of ZnCl2 suggest that the enzyme is a Zn2+-containing protein. As is generally true for alkaline phosphatases, substrate inhibition was noted with this enzyme. It appears, however, that this alkaline phosphatase is more sensitive to inhibition by PPi than are the other reported cartilage alkaline phosphatases.
We observed obvious inhibition at pH 8.5 when PPi concentrations exceeded 100 PM. By contrast, Fortuna et al. (31,32) saw inhibition only at concentrations above 10 mM, and Arsenis et al. (14) routinely assayed pyrophosphatase activity using 2 mM PPi at pH 8.5. Under these conditions, our enzyme showed barely detectable activity. This difference may be of physiological significance since PPi is a known inhibitor of hydroxyapatite crystal growth (36).
In conclusion, because of the apparent close association between the membrane-bound cartilage alkaline phosphatase and the mineralization induced by matrix vesicles (15), any studies of the physical properties of this enzyme must be done on an enzyme that is fully preserved during the isolation procedure. Unfortunately, in the previously reported matrix vesicle alkaline phosphatases (11, 14, 18), the proteases present in the crude collagenases required for release of the matrix vesicles from which the enzyme was isolated make it almost certain that these had undergone some degree of proteolysis. That this is indeed true is suggested by differences we now report in the molecular weight, stability, and kinetic behavior of this membrane-bound cartilage enzyme.