Isolation and Characterization of Propionyl-CoA Carboxylase from Normal Human Liver EVIDENCE FOR A PROTOMERIC TETRAMER OF NONIDENTICAL SUBUNITS*

We have purified propionyl-CoA carboxylase from normal, postmortem human liver to homogeneity. The isolation procedure, which provided an approximately 3000-fold purification and an overall yield of 268, em-ployed initial centrifugation of a cetyltrimethylammo- nium bromide-treated homogenate, followed by se- quential chromatographic separations using DEAE-cel-lulose, Blue Sepharose, and Bio-Gel A-1.5m. The native enzyme has a molecular weight of -540,000 and is composed of nonidentical subunits (a and p) of M, = 72,000 and 56,000, respectively. When studied with an- alytical isoelectrofocusing techniques, it focuses as a single peak at pH 5.5. Each mole of native enzyme contains 4 mol of bound biotin, virtually all of which is found with the larger (a) subunit. The apparent K , values for ATP, propionyl-CoA, and bicarbonate are 0.08 m ~ , 0.29 m ~ , and 3.0 111~, respectively. The enzyme also catalyzes the carboxylation of acetyl-coA and bu-tyryl-CoA to a limited degree, but not that of crotonyl- CoA. Propionyl-CoA carboxylase is quite stable over a temperature range from -50-37°C and over a pH range from 6.2 to 8.4. It has a broad pH optimum from pH 7.2 to 8.8. Limited proteolysis with trypsin results in slow, time-dependent deactivation of the enzyme

Propionyl-CoA carboxylase (EC 6.4.1.3), a biotin-dependent, mitochondrial enzyme, catalyzes the carboxylation of propionyl-CoA to D-methyhalonyl-CoA, a key step in the catabolic pathway for odd chain fatty acids and for isoleucine, threonine, methionine, and valine (1, 2). The enzyme was initially studied in crude extracts of animal tissues; it was later purified to homogeneity from pig heart mitochondrial fractions (3,4) and, more recently, from bovine kidney mitochondria (5). The native enzyme from pig heart has a molecular weight of about 700,000 and contains 4 mol of biotin/mol of enzyme (4). Propionyl-CoA carboxylase from bovine kidney mitochondria is composed of two nonidentical subunits of molecular weights 74,000 and 56,000; biotin was found bound to the larger subunit ( 5 ) .
The important role of propionyl-CoA carboxylase in human intermediary metabolism is evidence from the often lifethreatening recessively inherited propionic acidemias caused * This work was supported by Grant AM 09527 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. by deficiency of this enzyme (6). A molecular understanding of such propionyl-CoA carboxylase deficiency depends on the analysis of the normal and mutant enzymes. Although propionyl-CoA carboxylase activity has been measured in human skin fibroblasts (6), leukocytes (7), liver (8-lo), and amniotic fluid cells (ll), very little is known about the structure of the human enzyme. Until now the enzyme has been only partially purified from human liver mitochondria (9) and from cultured fibroblasts of both controls and several propionyl-CoA carboxylase-deficient patients (12). Mutant enzymes had pH optima, ionic requirements, and substrate affinities similar to those of the normal enzyme, but were more labile to both cold and heat (12). Further studies of biochemical parameters of mutant enzymes from two major genetic complementation groups (pcc A and pcc C ) were consistent with the notion that propionyl-CoA carboxylase consists of two nonidentical subunits ( 13).
In this communication we report the purification to homogeneity of propionyl-CoA carboxylase from normal human liver and describe some characteristics of the pure enzyme.
Assay of Enzyme Activity-Propionyl-CoA carboxylase was assayed by a modification of a procedure described previously (7). The enzyme was diluted in 10 mM phosphate buffer (pH 7.0) containing 1 mM 2-mercaptoethanol and 0.1 mg/ml of bovine serum albumin. The glutathione, 2 m~ ATP, 100 mM KCl, 10 m~ MgC12, 10 mM [''CI-standard reaction mixture containing 50 mM Tris-HC1 pH 8.0, 5 mM bicarbonate (specific activity 12.4 mCi/mmol), 3 m~ propionyl-CoA, and enzyme was incubated at 37°C for 15 min. The reaction was stopped by the addition of 10% trichloroacetic acid. Following centrifugation at 200 X g, an aliquot of the supernatant was dried slowly under a heat lamp, dissolved in water, and counted in Aquasol. One unit of enzyme activity is defined as that amount of enzyme catalyzing the furation of 1 pmol of bicarbonate/min at 37OC.
Protein Determination-Protein was measured by the method of Lowry et al. (16) or by a fluorometric method (17), using crystdine bovine serum albumin as a standard. Protein concentration in the

Human Hepatic
PropionyL-CoA CarboxyLase 61 fractions after column chromatography was determined by using A2m: Azm absorbance ratios as described by Warburg and Christian (18). Preparation of Antibody-White New Zealand rabbits were kept on a standard diet. One volume of complete Freund adjuvant obtained from Difco was added to 1 volume of protein diluted in 0.9% NaCl to a final concentration of 0.3 mg/ml and emulsified by pumping the mixture repeatedly through an 18-gauge needle using a I-ml syringe. Initially, 1.5 ml of the emulsion was injected into the hind foot pads and abdominal dermis. After 14 days, a booster of 150 pg of antigen in complete Freund's adjuvant was injected into the hind thigh and back. Intracardiac bleeding was performed 14 days after the second immunization. Sera were prepared and stored at -15'C. Purity of the antibody was determined by both Ouchterlony double diffusion and immunoelectrophoretic analysis in agar gels.
Electrofocusing-A sucrose-stabilized linear pH gradient (pH 4 to 6 ) of 1.2% ampholytes was prepared in an LKB (model 8101) isoelectrofocusing column. Either crude extract or enzyme in different stages of purification was added to the denser solution and distributed throughout the gradient. The isoelectrofocusing was run for 60 h at 4OC to a final voltage of 800 V. After completion of the run, 1.5-ml fractions were collected and assayed for activity.
Preparation of Subunits-Separation of the subunits was accomplished by sodium dodecyl sulfate gel electrophoresis on 3-mm-thick vertical slabs (17 X 17 cm), using the buffer system described by Fairbanks et al. (21). Usually, 500 pg of pure enzyme was loaded on one gel. After the run, 4-mm strips were cut from the slab and stained with 0.1% Coomassie blue, 25% 2-propanol, and 10% acetic acid. After destaining, these reference strips were used to localize the position of the subunits in the nonstained remainder of the gel. The subunits were subsequently eluted from the latter with an Isco Sample Concentrator, model 1750, and concentrated by lyophilization.
Biotin Assay-Qualitative estimation of biotin content was made using the method of Swack et al. (22). The enzyme was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, followed by incubation with fluorescent avidin. Quantitative spectrophotometric determination of biotin was performed as described by Green (23) using digestion with pronase at room temperature for 16 h (22).

Purification of Enzyme
Human liver, obtained a t autopsy 5 to 8 h after death, was frozen and stored at -70°C. All operations were carried out at 0-5°C unless otherwise stated. All buffers contained 1 mM 2mercaptoethanol.
Step 1-Three hundred grams of frozen human liver were usually used for the purification of the enzyme. The liver was homogenized in 1500 ml of 10 mM potassium phosphate (pH 7.0) and M NaEDTA (Fraction 1, Table I).
Step 2-Ten per cent (w/v) cetyltrimethylammonium bromide was added to Fraction I to a final concentration of 0.05%.
The suspension was stirred for 10 min, then centrifuged for 15 min a t 15,000 X g. The supernatant was collected and filtered through glass wool (Fraction 11).
potassium phosphate (pH 7.0) and applied to a column (5 X 30 cm) of DEAE-cellulose which had previously been equilibrated with the same buffer at a flow rate of 250 ml/h. After adsorption, the column was washed with 3000 ml each of 10 m~ and 30 mM potassium phosphate (pH 6.5), followed by approximately 12 liters of 50 mM potassium phosphate (pH 6.5) until the absorbance at 280 nm was below 0.035 A units. The enzyme was then eluted at the same flow rate with a linear gradient made up of 4 liters each of 50 mM and 120 mM potassium phosphate (pH 6.5). The propionyl-CoA carboxylase activity was found in a relatively broad peak with the highest activity appearing at approximately 70 mM. The active fractions were then pooled (about 3000 m l ) and concentrated to 400 ml in an Amicon ultrafiltration cell using an XM-100 membrane. The concentrate was dialyzed overnight against 5 mM potassium phosphate buffer, pH 7.0 (Fraction 111).
Step 4-The dialyzed sample was adsorbed onto a Blue Sepharose column (2.5 X 8 cm) equilibrated with the same buffer as used for dialysis at a flow rate of 50 ml/h. The column was then washed with 300 ml of 15 mM potassium phosphate buffer, pH 7.0. The ionic strength was reduced by washing with 100 ml of 5 mM potassium phosphate buffer, pH 7.0. The enzyme was eluted with 150 ml of 5 m~ potassium phosphate buffer containing 3 m~ ATP. The eluate was concentrated in an Amicon ultrafiltration cell and reconcentrated in 10 m~ potassium phosphate (pH 7.0) containing 100 mM NaCl and 10% glycerol to a final volume of 4 ml (Fraction IV). At this stage the enzyme is about 30% pure, containing only three other components as determined by sodium dode-cy1 sulfate-polyacrylamide gel electrophoresis.
Step 5-Fraction IV was applied to a column (2 X 120 cm) of Bio-Gel A-1.5m which had been equilibrated with 10 mM potassium phosphate buffer (pH 7.0), 100 m~ NaC1. Fractions of 3.5 ml each were collected a t a flow rate of 25 ml/h. The enzymatic activity was eluted in a sharp peak which cochromatographed with the first protein peak. Active fractions of the same specific activity were pooled, concentrated, checked for purity, and kept in 50% glycerol at -20°C.

Summary of Purification Procedure
A typical purification experiment starting with 300 g of liver is shown in Table I. Step 3 resulted in about a 40-fold purification; Step 4 resulted in another 20-fold gain. Although the final specific activity of the homogeneous enzyme (2 X 10' units/mg) was constant in liver obtained from several subjects, the net purification factor ranged from 2200 to 4000 (depending largely on the amount of blood sequestered in the liver). The degree of purity of propionyl-CoA carboxylase preparations was assessed by analytical polyacrylamide gel electrophoresis, using either 7.5% or 4% acrylamide. In both cases the protein migrated as a single band even when more than 20 pg was applied (Fig. 1, A and B ) .

Properties of the Purified Enzyme
Molecular Weight and Subunit Structure-The molecular weight of pure, native propionyl-CoA carboxylase was estimated on a Bio-Gel A-1.5m column to be -540,000, with a range of 500,000 to 575,000 from four separate determinations (Fig. 2). When the enzyme was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing conditions, two nonidentical subunits were noted with mobilities corresponding to molecular weights of 72,000 and 56,000 ( Figs. 1 and 3).
Biotin Determination-The biotin content of pure enzyme was determined spectrophotometrically on three different enzyme preparations using a method based on stoichiometric titration of avidin with biotin. About 4 mol of biotin were bound to 1 mol of native enzyme (Table 11). To identify the biotin-carrying subunit, the enzyme was run in 7.5% sodium dodecyl sulfate-polyacrylamide gel under reducing conditions. The gel was then exposed to fluorescent avidin which will bind to any biotin in the gel. The fluorescence generally  appeared only in the position of the larger subunit. In some preparations, however, slight fluorescence appeared in the region of the smaller subunit as well (data not shown). In control experiments, no fluorescence was observed in the bands of bovine serum albumin, ovalbumin, chymotrypsinogen, and lysozyme. T o estimate biotin content of each subunit, the subunits were isolated from a sodium dodecyl sulfatepolyacrylamide gel and assayed for biotin by the same method used for the native enzyme. The result presented in Table I1 shows that the large subunit contains biotin in a 1:l biotin/ protein molar ratio whereas the smaller subunit contains less than 0.1 mol of biotin/mol of subunit.
Limited Proteolysis-To examine the possibility that the smaller subunit might be a degradation product of the larger one, the effect of trypsin on pure propionyl-CoA carboxylase was studied. Both enzyme activity and gel electrophoresis were followed as functions of trypsin concentration and of time. At a trypsin/enzyme ratio of 1:500, no decrease in enzymatic activity was observed. Even at a ratio 1:100, about 84% of the original activity was still present after 3 h of incubation (Fig. 4). Only at a trypsin/enzyme ratio of 1:20 did the activity rapidly decrease with a half-life of about 60 min (Fig. 4). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing conditions did not suggest any conversion of a to /? subunits (Fig. 5). Moreover, the /? subunit was more sensitive to trypsin; only a slight amount was left after 3 h of incubation with trypsin, whereas much of the a subunit was still present (Fig. 5). The amino acid composition of both subunits also precludes the possibility of conversion of a to /? forms (data not shown). and 10 m~ bicarbonate, the apparent K,,, for ATP was found to be 0.08 mM (Fig. 6A). Similarly, at saturating concentrations of the other two substrates (with 2 m~ ATP), the K,,, values for propionyl-CoA and sodium bicarbonate were calculated to be 0.29 m~ (Fig. 6B) and 3.0 m~ (Fig. 6 0 , respectively. These kinetic parameters are similar to those described for pure pig heart propionyl-CoA carboxylase (4) and for crude enzyme from human fibroblasts (12) except that, in the latter, the K , for ATP was an order of magnitude higher. The enzyme had a K,,, for butyryl-CoA of 1.2 m~. Moreover, it carboxylated acetyl-coA, but only at a rate -1.5% that for propionyl-CoA. It did not carboxylate crotonyl-CoA. As noted previously for the pure enzyme from pig heart (24) and bovine liver (25). activity of the human enzyme was stimulated 6-to 7-fold by potassium ions. The estimated K,,, for this K ' stimulation was 9.4 mM, a value nearly 7-fold higher than that reported for the enzyme from pig heart (24).
pH Optimum and pH Stability-The pH optimum for activity of the pure enzyme was very broad, between pH 7.2 and 8.8. Even at a pH as high as 9.6, the enzyme had about half of the maximal activity. No substantial differences in activity were found in two different buffers, potassium phosphate or Tris-HC1, over a pH range 7.4 to 8.2. However, when the enzyme was incubated without substrate at 37OC for 20 min with buffers of different pH values, and then assayed at pH 8.0, it retained more than 60% of activity at pH 6.4, but only 10% of activity at pH 9.2 and 3% of activity at pH 9.6. Isoelectrofocusing-Pure propionyl-CoA carboxylase was electrofocused as described under "Experimental Procedures.'' The major peak of activity was found at pH 5.5 (Fig.  7). However, in all preparations of pure enzyme, there was an additional peak consisting of about 5 to 10% of the total enzyme activity found, at a PI around 5.0. Thinking that these two peaks represented different isoenzymes, we focused crude extracts of liver homogenate, all of the enzyme activity focused at pH 5.0 (Fig. 7). The Same PI value was also found for the fraction focused after DEAE-cellulose chromatography (Fig.  7). Only in focusing fractions after Blue Sepharose chroma- tography did we observe the shift of the major portion of activity to a PI value of 5.5, with a small amount of activity left at PI 5.0 (Fig. 7). It was not possible to focus the sample after treatment with cetyltrimethylammonium bromide because the enzyme became inactive under isoelectrofocusing conditions. In two out of six experiments, crude homogenates were heterogeneous with three main peaks at isoelectric points 4.6, 5.0, and 5.5; although the distribution of activity was uneven, at least 60% was always present at pH 5.0.
Antibody Preparation and Zmmunotitration-The purity of antibody raised against pure human propionyl-CoA carboxylase was tested both by Ouchterlony double diffusion and by immunoelectrophoresis. In both cases only one precipitin line Antiserum ( @ I was observed even when crude homogenate was used as a source of antigen (Fig. 8). Immune titration curves, which were similar for both crude homogenate and pure enzyme, were linear over a wide range of enzyme activity (Fig. 9). Rabbit antisera against pure human propionyl-CoA carboxylase showed a low degree of species specificity, reacting readily with propionyl-CoA carboxylase in crude extracts prepared from monkey, bovine, rat, and mouse liver.

DISCUSSION
We have purified human hepatic propionyl-CoA carboxylase about 3000-fold from crude homogenates and have obtained a homogenous enzyme. Since the human livers available to us were obtained 5 to 10 h postmortem and had, therefore, undergone significant postmorten autolysis, we were forced to develop methods for purification of propionyl-CoA carboxylase from the crude homogenate rather than intact mitochondria. As with the purification of ornithine transcarbamylase from human liver (26), we observed that the addition of cetyltrimethylammonium bromide to the crude extract greatly increased the total recoverable enzyme activity. This step precipitated considerable amounts of protein from the homogenate which could then be removed by low speed centrifugation. The subsequent relatively simple purification scheme depended on affinity chromatography with Blue Sepharose. It was essential during this procedure to keep the ATP concentration at 3 mM. Lower concentrations caused the enzyme to be eluted very slowly and not quantitatively, and higher concentrations caused the release of a t least five additional proteins, as shown by polyacrylamide gel electrophoresis. Furthermore, the descending portion of the propionyl-CoA carboxylase peak eluted from DEAE-cellulose contained another protein which was also bound on Blue Sepharose and was eluted with ATP. This contaminant had a native molecular weight of 470,000 and a subunit structure close to that of the / 3 subunit of propionyl-CoA carboxylase. Removal of this impurity required either rechromatography on Bio-Gel A-1.5m or electrofocusing. T o our knowledge, human propionyl-CoA carboxylase is the f i t biotin enzyme purified in this manner. It is likely that Blue Sepharose can be used for all biotin enzymes which require ATP for their activity.
Analysis of pure human propionyl-CoA carboxylase shows that the native enzyme has a molecular weight of 540,000 and is composed of two nonidentical subunits of molecular weight 72,000 and 56,000. Similar molecular weights have been reported for native propionyl-CoA carboxylase isolated from pig heart (4) and bovine kidney (5) mitochondria. In the latter, the subunit structure and molecular weights of the subunits were also studied and found to be comparable to those of human liver. This subunit composition is not restricted to eukaryotes. Propionyl-CoA carboxylase from Mycobacterium smegmatis also consists of two subunits of molecular weight 64,000 and 53.000 (27). Since nonidentical subunits have now been found for propionyl-CoA carboxylases from organisms as diverse as bacteria and man, it seems likely that such a structure is common to all prokaryotes and eukaryotes. It should be mentioned, however, that such a structure has not been reported for all biotin-dependent carboxylases, some of which have been shown to be composed of identical subunits (28,29), others of nonidentical ones (30).
It could be argued that the smaller (B) subunit found in our studies is produced by in vitro proteolysis of the larger (a) one, and, therefore, that our demonstration of nonidentical subunits is artifactual. This seems most unlikely on several grounds. Our experiments with in vitro treatment of pure carboxylase with trypsin failed to demonstrate any conversion Human Hepatic Propionyl-CoA Carboxylase of the larger subunit to the smaller one, showing rather that the human enzyme is quite resistant to trypsin. Since trypsin is by no means the only possible protease, this finding is inconclusive. More convincing are the following: first, the markedly different amino acid compositions of the two putative subunits found in our studies and to be reported subsequently; and second, the demonstration by Lau et al. ( 5 ) that the molecular weight of the pure bovine kidney enzyme was the same when purified in the presence of protease inhibitors as in their absence.
All available information concerning molecular weight and subunit structure suggests the existence of subunit stoichiometry in both human and bovine propionyl-CoA carboxylases. This thesis is strongly supported by our nearly quantitative elution of the subunits in a molar ratio from sodium dodecyl sulfate-polyacrylamide gels. Furthermore, based on amino acid composition of subunits (data not presented here), which were eluted quantitatively from the gels with formic acid followed by hydrolysis, an approximate molar ratio was observed.
Our analysis shows that 1 molecule of biotin is bound/ molecule of the larger (a) subunit. A small amount of biotin was also found occasionally with the smaller subunit, as shown both by the fluorescent avidin technique and by quantitative assay. Perhaps, as has been proposed for P-methylcrotonyl-CoA carboxylase (31), the smaller subunit may contain a second binding site for free biotin. Alternatively, the minimal binding to the smaller subunit may be artifactual.
Pure propionyl-CoA carboxylas\? is remarkably sensitive to pH. Another biotin enzyme, 3-metl'ylcrotonyl-CoA carboxylase from Achromobacter, has been shown to dissociate into two nonidentical subunits above pH 9.0, thus losing enzymatic activity (32). The dissociation is reversible; reassociation can be stimulated by the substrate, 3-methylcrotonyl-CoA (32). It can be anticipated that our enzyme undergoes similar dissociation. We have already shown that, in the presence of substrate, enzymatic activity is preserved well beyond pH 9.0.
We found a significant difference in isoelectric points between the pure enzyme and the crude enzyme fractions. This change in PI occurred during chromatography on Blue Sepharose. We suggest three possible explanations for this finding: first, that during that step we remove a nucleotide or nucleic acids loosely bound to the enzyme molecule; second, that we break a complex between the enzyme and some acidic protein; and third, that this step selectively changes the conformation of the molecule in such a way as to expose more positive charges. We consider the first two possibilities more likely since, in some crude extracts, the enzyme focused in more than one peak, all between pH 4.6 and pH 5.5. Such variation in extracts could be due to various degrees of formation of some protein complex, or to various amounts of some complexing factor present. These situations in turn could reflect postmortem or freezing effects in the livers.
Recently McKeon et al. reported that the isoelectric point of propionyl-CoA carboxylase measured in crude extracts of normal liver differed from that in liver of a patient with propionyl-CoA carboxylase deficiency (thepcc C complementation group) (33). The authors concluded that these differences in the isoelectric point reflected a structural alteration in the mutant enzyme. In view of our observation, it is unlikely that measurement of isoelectric points in crude extracts is a valid criterion for the comparison of normal and mutant enzyme structure.
At least four distinct genetic complementation groups of human propionyl-CoA carboxylase mutants have recently been reported (34, 35). Biochemical studies in our laboratory of two major complementation groups, pcc C and pcc A, suggested different structural gene mutations (13). Experiments with antisera prepared against pure human propionyl-CoA carboxylase are now in progress to characterize these mutants further.