Biotin Protein Ligase from Saccharomyces cerevisiae

Catalytically active biotin protein ligase fromSaccharomyces cerevisiae (EC 6.3.4.15) was overexpressed inEscherichia coli and purified to near homogeneity in three steps. Kinetic analysis demonstrated that the substrates ATP, biotin, and the biotin-accepting protein bind in an ordered manner in the reaction mechanism. Treatment with any of three proteases of differing specificity in vitro revealed that the sequence between residues 240 and 260 was extremely sensitive to proteolysis, suggesting that it forms an exposed linker between an N-terminal 27-kDa domain and the C-terminal 50-kDa domain containing the active site. The protease susceptibility of this linker region was considerably reduced in the presence of ATP and biotin. A second protease-sensitive sequence, located in the presumptive catalytic site, was protected against digestion by the substrates. Expression of N-terminally truncated variants of the yeast enzyme failed to complement E. colistrains defective in biotin protein ligase activity. In vitro assays performed with purified N-terminally truncated enzyme revealed that removal of the N-terminal domain reduced BPL activity by greater than 3500-fold. Our data indicate that both the N-terminal domain and the C-terminal domain containing the active site are necessary for complete catalytic function.

Biotin (vitamin H) is a protein-bound cofactor required for the synthesis of functional biotin carboxylases and decarboxylases. These enzymes catalyze essential metabolic processes in both prokaryotes and eukaryotes using the biotin prosthetic group as a mobile carboxyl carrier (1,2). The post-translational attachment of biotin to these enzymes via the ⑀-amino group of a specific lysine residue is catalyzed by biotin protein ligase (BPL 1 ; EC 6.3.4.15) in a two-step reaction.

REACTIONS 1 and 2
The best characterized BPL is the 35.3-kDa BirA protein from Escherichia coli, which is a bifunctional protein that also acts as the repressor of the biotin biosynthetic operon (3,4). The crystal structure of the protein, determined at 2.3-Å resolution, shows the domains identified from genetic and biochemical studies (4,5). The N-terminal domain contains a helix-turnhelix fold for DNA binding. The central domain, which contacts biotin and contains the nucleotide triphosphate binding motif GRGRRG, is the catalytic domain. No function has been assigned to the small C-terminal domain. BPL protein (6 -8) and the bpl gene (9, 10) have been isolated from several other prokaryotes, and an increasing number of birA homologues have been identified from genome sequencing projects in organisms as diverse as Methanococcus jannaschii (11) and Archaeoglobus fulgidus (12). Similarity with E. coli BirA at the amino acid level suggests that all of these bacterial proteins are also bifunctional molecules with essentially the same domain structure.
Biotin-accepting enzymes can be recognized and biotinylated by BPL derived from widely divergent sources (13)(14)(15)(16), indicating that the determinants of a functional protein/protein interaction have been highly conserved throughout evolution. However, it is apparent that eukaryotic BPLs, while catalyzing the same essential biotinylation reaction, have different properties from the prokaryotic members of the family. Most of the eukaryotic BPL proteins that have been purified are around twice the size of the bacterial enzyme (reviewed in Ref. 17). The gene for BPL from Saccharomyces cerevisiae encodes a protein of 690 amino acids (predicted mass of 76.4 kDa (18)), and the human enzyme is similar, containing 726 residues (15,19). Plant BPLs are intermediate in size between the bacterial and other eukaryotic enzymes, with a molecular mass of 37-41 kDa (16,20,21). The C-terminal region of the eukaryotic BPL sequences shows homology with BirA and the biotin-binding protein, avidin, suggesting that this contains the catalytic site. Residues in E. coli BirA comprising the ATP-binding motif or residues that, if mutated, cause an increased K m for biotin are all highly conserved between species. Additionally, those residues shown to contact biotin in the BirA crystal structure are invariant (15,16,18). Consistent with biotin metabolism in the different organisms, none of the eukaryotic proteins contains sequences that suggest any DNA binding activity.
While the large additional N-terminal domain present in both the human and yeast BPL proteins shows some sequence similarity between the two species (18), little is known of the function of this region of the eukaryotic enzyme. Evidence for the functional importance of this domain comes from the identification of mutations in human BPL responsible for inherited metabolic disease, multiple carboxylase deficiency. Of the known lesions giving rise to this neonatal onset disorder, several point mutations alter residues in the vicinity of the presumptive biotin binding site and give rise to a decreased affinity for biotin (22,23). However, other mutations, also responsive to biotin in vivo, occur in the N-terminal domain (residues 215-240), some distance from the catalytic site (19,22,23). The loss of BPL activity is apparently not due to either decreased affinity for biotin or biotin-mediated stability (23), and it is not clear why the defect responds to biotin therapy. Currently, no comprehensive structural or functional data are available for the N-terminal region of eukaryotic BPL.
Here we report the recombinant expression in E. coli of active yeast BPL (yBPL) with a C-terminal hexahistidine tag. Limited proteolysis of the purified enzyme has confirmed predictions of the domain structure based on sequence homology. Expression of N-terminally truncated proteins indicated that the presence of both domains was necessary for catalytic function.

EXPERIMENTAL PROCEDURES
Materials-Polyacrylamide was obtained from Bio-Rad; nitrocellulose was from Schleicher & Schuell; restriction endonucleases were purchased from New England Biolabs; Pwo DNA polymerase was from Roche Molecular Biochemicals; Ni-NTA alkaline phosphatase conjugate and Ni-NTA-agarose were from Qiagen; and [ 3 H]biotin (35 Ci/mmol) was from Amersham Pharmacia Biotech. All reagents were of analytical grade or higher.
Nucleic Acid Manipulations-Transformation, isolation of plasmid DNA, restriction enzyme digestion, and agarose gel electrophoresis were carried out using standard methods (24). All nucleic acid constructs were confirmed by DNA sequencing using ABI Prism Dye Terminator sequencing (Perkin-Elmer).
The plasmid pKS(bpl) was produced by ligating a 2.6-kilobase pair BstBI/SnaBI DNA fragment, containing the Bpl coding region, from pCY248 (18) into AccI/SmaI-digested pBluescript KS Ϫ II. A 2.1-kilobase pair NcoI/BamHI fragment, excised from pKS(bpl) and containing the Bpl coding region from base pair ϩ131 to the end of the gene, was cloned into NcoI/BglII-digested pET16b (Novagen), yielding the plasmid pN/ B-Bpl. Subsequent cloning steps introduced the 5Ј coding region and codons for six histidine residues at the 3Ј-end of the gene. A BspHI site was introduced around the initiation codon by polymerase chain reaction mutagenesis using primers Bpl5Ј and Bpl3Ј with pKS(bpl) as template. This product was ligated into pGEM-T (Promega), and the 308-base pair fragment liberated by digestion of the resultant plasmid with BspHI and BstEII was cloned into NcoI/BstEII-treated pN/B-Bpl. This yielded the expression vector pET(Bpl). Six histidine codons were introduced at the 3Ј-end of the gene by polymerase chain reaction mutagenesis using primers BplPst and BplBam with pKS(bpl) as template. The expression vector pET(Bpl-His) was constructed by a threefragment ligation containing the purified polymerase chain reaction product from this reaction that had been treated with PstI/BamHI, the 1.1-kilobase pair PstI/BamHI fragment from pET(Bpl) and PstI-digested pET16b. The expression vectors were transformed into the E. coli strain BL21(DE3)pLysS (Novagen), and transformants were grown on Luria broth (24) supplemented with 100 g/ml ampicillin and 30 g/ml chloramphenicol.
A vector was constructed to allow arabinose-inducible expression of full-length yBPL with the C-terminal hexahistidine (His 6 ) tag. The 2.1-kilobase pair NcoI/BamHI fragment from pET(Bpl-His) was cloned into NcoI/BglII-treated pAra13 (25), yielding pN/B-Bpl-His. This construct was digested with NcoI and BstEII and ligated to the 308-base pair BspHI/BstEII fragment described above to yield p[Met 1 ]Bpl 1-690-His 6 . Vectors for expression of N-terminally truncated forms of yBPL were constructed by introducing translation initiation codons, which also contained an NcoI cloning site, preceding residues 233, 369, and 409 of the full-length protein. Oligonucleotides Bpl233, Bpl369, or Bpl409 were included in separate polymerase chain reactions with oligonucleotide BplPstBack and pKS(bpl) as the template. The purified polymerase chain reaction products were digested with NcoI and PstI and cloned into similarly treated pN/B-Bpl-His. This gave constructs pAra[Met 1 ]Bpl 233-690-His 6 , pAra[Met 1 ]Bpl 369 -690-His 6 , and pAra[Met 1 ]Bpl 409 -690-His 6 , respectively. All constructs were confirmed by DNA sequencing using oligonucleotides AraFor and AraFor2.
Bacterial Strains and Growth Media-For large scale protein production, the E. coli strain used was BL21(DE3), also containing the plasmid pLysS (Novagen). The E. coli strains used for complementation assays were the birA1 bioC strain, CY918 (26) and the birA85 bioC strain BM4062 (27). Complementation was assayed on selective medium containing (per liter) 10 g of Bacto-tryptone, 0.5 g of yeast extract, and 5 g of NaCl (18), with growth scored after 16 h of incubation. Nonselective medium for complementation was Luria broth (24).
Expression and Purification of Apo-yPC-104 -The construction of a vector for expression of the C-terminal 104 residues of yeast pyruvate carboxylase-1 (yPC-104) has been previously described (28). Crude cell lysates containing both the apo and holo forms of yPC-104 were prepared using the method of Chapman-Smith et al. (26), except cells were lysed in buffer A (50 mM Tris-HCl, pH 8.0, 0.1 mM EDTA). The lysate from 1 liter of culture was filtered and passed through a Q-Sepharose column (Amersham Pharmacia Biotech, 12 ϫ 2.6 cm) equilibrated in buffer A, and the unbound material containing apo-yPC-104 was collected. This flow-through fraction was reapplied to the column and fractionated with a 250-ml gradient from 0 to 200 mM NaCl in buffer A, run at 5 ml/min. Fractions containing apo-yPC-104 were pooled, concentrated, and run on a Superdex-75 HR 10/30 (Amersham Pharmacia Biotech) gel filtration column in 2 mM ammonium acetate, pH 7.5. Those fractions containing the purified apobiotin domain were lyophilized and stored at Ϫ20°C.
Expression of Yeast BPL and Preparation of Cell Lysates-Bacterial cultures of E. coli BL21(DE3)pLys harboring pET(Bpl-His 6 ) were grown in shake flasks in 2YT supplemented with 100 g/ml ampicillin and 30 g/ml chloramphenicol. Overnight cultures were diluted 1:100 into 2 liters of fresh medium and grown at 30°C to A 600 0.6 -0.8 before the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.1 mM. After 3 h, the cells were harvested by centrifugation and washed in binding buffer (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 5 mM imidazole), and the pellet was stored at Ϫ80°C overnight. The cell pellet was thawed on ice in 60 ml of binding buffer, with the addition of 1 mM phenylmethylsulfonyl fluoride, resulting in cell lysis by the action of T4 lysozyme expressed from the co-resident pLysS plasmid. The disrupted cell suspension was sonicated and centrifuged at 10,000 ϫ g for 10 min. After a second centrifugation, the supernatant was filtered through a 0.45-m Minisart filter (Sartorius) prior to chromatography.
The expression and purification of the yBPL truncations were performed essentially as above except that the truncations were expressed in E. coli DH5␣ and induced by growth in media supplemented in 0.2% arabinose for 6 h at 30°C. Before chromatography, cells were disrupted by two passes through a French press (82,800 -103,500 kilopascals).
Purification of Yeast BPL-His 6 -tagged material was purified on a 2.5-ml Ni-NTA-agarose (Qiagen), gravity-fed column. After loading the cell lysates onto the charged column equilibrated in binding buffer, the column was washed with 12 column volumes of binding buffer and 12 volumes of wash buffer (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 10 mM imidazole), and bound material was eluted with 6 volumes of elution buffer (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 0.5 M imidazole). Fractions containing yBPL were pooled and dialyzed overnight against 4 liters of storage buffer (50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM dithiothreitol, 5% (v/v) glycerol). The BPL-containing fractions eluted from the nickel column were further fractionated on a Q-Sepharose column (Amersham Pharmacia Biotech; 12 ϫ 2.6 cm) with a 450-ml gradient from 0 to 250 mM NaCl in storage buffer, run at 5 ml/min. Fractions containing yBPL, detected by SDS-PAGE and Ni-NTA Western blot, were pooled and stored at Ϫ80°C.
Assay of Yeast BPL-BPL activity was assayed by measuring the incorporation of [ 3 H]biotin into either apoBCCP-87 or apo-yPC-104 as described by Chapman-Smith et al. (29). Briefly, the reactions contained 50 mM Tris-HCl, pH 8.0, 3 mM ATP, 5.5 mM MgCl 2 , 5 M biotin, 5 pmol of [ 3 H]biotin (specific activity 35-44 Ci/mmol), 0.1 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, and either 20 M apoB-CCP-87 or 5 M apo-yPC-104. The reaction was initiated by the addition of purified yBPL to a final concentration of 13 nM, except where stated otherwise, and incubated at 37°C for up to 30 min, when aliquots of the reaction were spotted onto biotin-and trichloroacetic acid-treated filters. After air drying, the filters were washed twice in 10% ice-cold trichloroacetic acid and once in ethanol and dried, and the acid-insoluble radioactivity was measured. A unit of BPL activity was defined as the amount of enzyme required to incorporate 1 nmol of biotin per min. Values for K m and V max were determined by fitting a plot of substrate concentration against rate to the Michaelis-Menten equation using GraphPad Prism for MacIntosh (GraphPad Software Inc., San Diego, CA). In some experiments, to obtain sufficiently high levels of radioactivity for accurate detection, it was necessary to continue until greater than 10% of the limiting substrate had been utilized. In this case, the data were transformed for altering substrate concentration by the method of Lee and Wilson (30) and plotted as transformed values sЈ and vЈ.
Protein Techniques-PAGE was carried out on 12% polyacrylamide SDS gels with the Tris/glycine system (31), using Mark 12 wide-range protein standards (Novex). His 6 -tagged protein was detected after Western transfer using Ni-NTA alkaline phosphatase conjugate (Qiagen) following the manufacturers' protocol. Quantitation of protein after PAGE was carried out by laser densitometry, using a Molecular Dynamics model 300A densitometer with ImageQuant software (Sunnyvale, CA), with adjustment for the effect of molecular weight on Coomassie Blue staining. Protein concentration was determined using the Bio-Rad protein assay kit. Electroblotting of peptides onto polyvinylidene difluoride was performed as described by Matsudaira (32). N-terminal sequencing of proteins by automated Edman degradation was performed using a Perkin-Elmer Procise 492 protein sequencer.

RESULTS
Protein Expression and Purification-The yeast bpl gene was cloned into the pET expression vector for recombinant expression in the E. coli strain BL21(DE3)pLysS, with the hexahistidine tag on the C terminus to facilitate downstream purification of the enzyme (described under "Experimental Procedures"). Assays for yBPL activity, carried out on crude lysates from cells expressing the protein with and without the His 6 tag, revealed that the presence of the His 6 tag did not affect the activity of the enzyme (data not shown). For optimal recovery of active enzyme from the protein expression system, cells were grown at 30°C in shake flasks, and expression of yBPL was induced with low concentrations of isopropyl-1-thio-␤-D-galactopyranoside (0.1 mM). Initial experiments showed that growth at 37°C and higher isopropyl-1-thio-␤-D-galactopyranoside concentrations led to inclusion body formation, in both shake flasks and a 1-liter fermenter.
As an initial purification step, crude lysate was passed over a Ni-NTA column, which gave a 30 -40-fold enrichment (Table  I). Since His 6 -yBPL eluted from the resin with only 20 mM imidazole, column washings were performed under low stringency conditions (10 mM imidazole). BPL activity was detected in the unbound material, but since no His 6 -tagged protein was detected in this fraction by Western blot analysis, the activity apparently represented endogenous bacterial biotin ligase. The recombinant yBPL obtained by Ni-NTA chromatography was immediately dialyzed overnight against storage buffer, since storage in imidazole resulted in irreversible inactivation of the enzyme. Additional purification using anion exchange chromatography was required following Ni-NTA chromatography, since the low stringency washing conditions failed to remove a number of contaminating proteins (Fig. 1, lane 2). After two chromatography steps, half of the initial enzyme activity had been recovered, and the protein was at least 80% pure. This represented a purification of around 88-fold (Table I). However, in addition to the 77-kDa protein corresponding to full-length His 6 -yBPL, a band of 50 kDa was observed on SDS-PAGE of the pooled fractions from the Q-Sepharose column (Fig. 1, lane  3). This protein was also detected on a Ni-NTA blot, indicating the presence of the C-terminal His 6 tag. N-terminal sequencing showed the 50-kDa protein to be a proteolysis product of the intact yBPL, with cleavage occurring between Lys-248 and Phe-249. The cleaved species and the intact protein also coeluted during hydrophobic interaction chromatography on phenyl-Sepharose 6 resin and were only partially resolved by analytical gel filtration on either Superdex 75 or 200 columns (Amersham Pharmacia Biotech). Analysis by SDS-PAGE of this material after Superdex 75 fractionation demonstrated that the cleavage products did not remain associated after three steps during the purification, since the 50-and 27-kDa fragments could be resolved by gel filtration (data not shown).
In order to obtain a sample of intact yBPL that did not contain the cleaved form, protein collected after anion exchange chromatography was concentrated by ultrafiltration prior to fractionation on a preparative Superdex 75 gel filtration column (26 ϫ 600 mm; Amersham Pharmacia Biotech) run in storage buffer. As the quantity of cleaved yBPL constituted less than 20% of the sample (as determined by laser densitometry of bands detected on SDS-PAGE), approximately 50% of the intact enzyme in the sample was resolved away from the cleaved product by collecting only the leading fractions of the protein peak. These fractions contained full-length yBPL with a purity of greater than 95%, determined by SDS-PAGE (Fig. 1,  lane 4) and N-terminal sequencing. The intact yBPL purified in this way was used for all subsequent kinetic and proteolysis analysis.
Biological Properties of yBPL-The activity of yBPL was determined by measuring the incorporation of [ 3 H]biotin into one of two biotin-accepting domains, either the C-terminal 87 amino acid residues of the E. coli biotin carboxyl carrier protein (BCCP-87) (29) or the C-terminal 104 amino acid residues of S. cerevisiae pyruvate carboxylase 1 (yPC-104) (28). Optimal enzyme activity was observed at pH 8.0 -8.5. The presence of magnesium ions, ATP, biotin, and the apo form of a biotin domain were necessary substrates for activity, and the addition of EDTA inhibited the reaction. Other divalent metal ions could be used by yBPL, magnesium (100%), calcium (109%), nickel (74%), and manganese (53%) ions supporting activity to varying extents. However, cobalt (14%), zinc (1.3%), and copper (0.5%) ions were a poor substitute for magnesium. Inhibition of activity was observed in the presence of the monovalent metal ions sodium and potassium at concentrations of 200 mM (data not shown). The ability of the biotin analogues biocytin, diaminobiotin, desthiobiotin, and iminobiotin to inhibit the enzyme was investigated. At a concentration of 5 M, none of these  1. Purification of recombinant  Enzyme activity was also completely dependent on ATP as the nucleotide source, since substitution with UTP, GTP, and CTP gave 5.6, 0.6, and 0.4%, respectively, of the activity observed with ATP.
Kinetic Analysis of yBPL-The kinetic constants for D-biotin, MgATP, and two different biotin domain substrates, apoB-CCP-87 and apo-yPC-104, were determined using steady state kinetics (Fig. 2). The K m for MgATP was determined to be 20.9 Ϯ 3 M ( Fig. 2A). As has been observed with BPLs from a wide variety of species, the yeast enzyme also had a low K m for biotin; 67 Ϯ 11 nM (Fig. 2B). The yeast substrate, apo-yPC-104, displayed a K m of 1.0 Ϯ 0.2 M (Fig. 2C), whereas the K m for the bacterial biotin domain was more than 10-fold higher (11.1 Ϯ 1 M; Fig. 2D).
In order to determine the kinetically preferred order of addition of substrates, we assayed the activity of yBPL under steady state conditions when the concentrations of two substrates were varied while maintaining the third at saturating levels. The double-reciprocal plots from the velocity measurements are shown in Fig. 3. When biotin and MgATP were the varied substrates, patterns of intersecting lines were obtained (Fig. 3, A and B), indicating a reversible connection between these two substrates in the reaction pathway (33). When the concentration of acceptor protein was varied together with either biotin or MgATP, patterns of parallel lines were observed on the double-reciprocal plots (Fig. 3, C and D). These results implied that both biotin and MgATP combine with yBPL before the acceptor protein (33). Biotinyl-5Ј-AMP is known to react readily with hydroxylamine to form biotinylhydroxamate (34,35). When yBPL was incubated with [ 3 H]biotin, MgATP, and hydroxylamine, [ 3 H]biotinyl-hydroxamate was formed (data not shown), confirming that the reaction pathway proceeds, as shown in Reaction 1, through the formation of the biotinyl-5Ј-AMP intermediate.
Since the formation of biotinyl-5Ј-AMP is accompanied by the release of pyrophosphate, product inhibition studies with pyrophosphate were performed. Pyrophosphate behaved as a competitive inhibitor relative to MgATP and as a noncompetitive inhibitor with respect to biotin (Fig. 4), implying that MgATP binding precedes biotin binding (33). Our data here are in agreement with a double displacement kinetic mechanism for yBPL, where biotinylation proceeds through two partial was treated with trypsin, chymotrypsin, and endoproteinase Glu-C for 2 h, and the products were analyzed by SDS-PAGE (Fig. 5A). All three proteases generated a fragment of about 50 kDa, which contained the C terminus, identified using a Ni-NTA blot to probe for the His 6 tag. N-terminal sequencing of these products revealed that cleavage occurred between Lys-256 and Thr-257 for trypsin, between Leu-254 and Thr-255 for chymotrypsin, and between Glu-243 and Ile-244 for endoproteinase Glu-C (Fig. 5B). Digestion with both trypsin and endoproteinase Glu-C released a second fragment containing the His 6 tag. These products, of approximately 27 kDa, were the result of cleavage between Pro-408 and Glu-409 with endoproteinase Glu-C and between Arg-425 and Gly-426 with trypsin. Since these cleavage points are located around the predicted catalytic site, yBPL was subjected to digestion after equilibration with saturating concentrations of MgATP and biotin (Fig.  5A). Under these conditions, the 27-kDa products were not detected, indicating that cleavage did not occur at Glu-409 or Arg-425. In addition, cleavage at Lys-256, Leu-254, and Glu-243 was considerably slower in the presence of the substrates, since the release of the 50-kDa fragment was markedly reduced in all cases. Together, these data imply that the yBPL molecule contains two protease-sensitive sites, one within an interdomain linker that connects a 27-kDa N-terminal domain with the remaining 50 kDa of the protein and a second region within the catalytic site. The enzyme-biotinyl-5Ј-AMP complex was more resistant to proteolysis at both sites than yBPL alone, which suggested that structural differences exist between the two enzyme forms, with the enzyme complex having a more compact conformation.
Tryptic digestion of either yBPL alone (apoenzyme) or yBPLbiotinyl-AMP complex (holoenzyme) was performed, measuring both the loss of the 77-kDa intact protein and enzyme activity (Fig. 6A). As expected, the holoenzyme was more resistant to proteolysis, and more enzyme activity was retained, compared with apoenzyme (Fig. 6A). For both enzyme forms, digestion of the enzyme and loss of activity occurred at slightly different rates over the first 2 h, whereas with longer reaction times the loss of activity corresponded to loss of intact protein. This most probably reflects cleavage occurring at the two protease-sensitive regions of the molecule determined above. Initially, cleavage occurring in the interdomain linker was such that the digested enzyme retained some activity, whereas proteolysis within the catalytic domain inactivated the enzyme. Analysis of the products by SDS-PAGE revealed that the loss of enzyme activity was indeed coincident with the production of a 30-kDa fragment containing the His tag, the product of tryptic digestion in the active site (Fig. 6B).
In Vivo Characterization of Truncated yBPL-A series of vectors, for expression of N-terminally deleted forms of yBPL containing a C-terminal His 6 tag in E. coli, were constructed as described under "Experimental Procedures." These vectors, derived from pAra13 (25), expressed yBPL, yBPL(⌬1-233), yBPL(⌬1-369), and yBPL(⌬1-409) under the control of an arabinose-inducible promoter. All vectors were transformed into the birA, biotin auxotroph strains CY918 (26) and BM4062 (27). These strains have a high requirement for biotin, and the birA85 mutation in BM4062 also confers a temperature-sensitive phenotype. Under selective conditions only those strains expressing functional exogenous BPL survive, since the essential E. coli enzyme acetyl-CoA carboxylase can be biotinylated and is therefore active.
The CY918 strains bearing the expression vectors were grown on permissive and selective media to test for complementation of the defective bacterial birA1 (see "Experimental Procedures"). Strains CY918 and CY918 harboring the parent expression vector alone, pAra13, did not grow on the selective media supplemented with 0.2% arabinose, but growth was observed on nonselective media. The addition of as little as 3 nM biotin to the selective media restored growth (data not shown). As expected, the strains expressing either full-length yBPL or E. coli BirA, from plasmid pCY216 (26) permitted growth on both media. However, expression of the truncated forms of yBPL failed to complement the mutant strain at either 30 or 37°C on selective media (Fig. 7). For all truncations except yBPL(⌬1-409), an induced His 6 -tagged protein of the expected molecular mass was detected in crude cell lysates using a Ni-NTA blot, indicating that the proteins were being expressed (Fig. 8). Expression of yBPL-409 was not detected, possibly because this truncation, which removes part of the predicted catalytic region, is rapidly degraded.
The observed failure of the yBPL truncations to complement the birA1 mutation at low concentrations of biotin may have been due to the truncated yeast enzymes themselves having a higher biotin requirement, as has been reported for N-terminally truncated BirA (36). Therefore, the complementation assay was carried out in a second bacterial strain, BM4062 (27), where the endogenous bacterial BPL could be heat-inactivated. At 42°C, where the BirA85 protein was nonfunctional (27), only full-length yBPL was able to sustain growth. The assay was performed on media supplemented with increasing biotin concentrations up to 1 mM. However, the additional biotin did not permit the growth of strains expressing truncated yBPL.
In Vitro Characterization of Truncated yBPL-The yBPL truncations were expressed in E. coli DH5␣ cells and partially purified by nickel chelating chromatography. The Ni-NTA-purified material recovered from cells harboring pAra13 displayed no BPL activity, showing that endogenous bacterial BPL had been removed. Full-length yBPL was found to have the highest specific activity (25 nmol/min/mg; Fig. 9). Removal of the N-terminal domain in construct yBPL(⌬1-233) was found to reduce the activity of the enzyme by greater than 3500-fold (7 pmol/min/mg; Fig. 9). This truncation was further purified by anion exchange chromatography and tested for activity in the presence of increasing concentrations of biotin. The addition of up to 500 M biotin in the reaction did not improve the enzyme's activity (data not shown), suggesting the observed decrease in activity for this truncation was not the result of an increased K m for biotin, consistent with the in vivo complementation assays. The truncation yBPL(⌬1-369) had very low specific activity (1.7 pmol/min/mg). As expected, yBPL(⌬1-409), which has part of the proposed ATP binding motif deleted, showed no activity. It is evident that the 3500fold reduction in activity seen upon the deletion of the Nterminal domain reduced the activity of yBPL to a level that is inadequate for viability. DISCUSSION Limited proteolysis of purified yBPL, undertaken to determine the domain structure of the protein, indicated that the protein contained two protease-sensitive regions, the more Nterminal of which lies between residues 240 and 260. The observation that all three proteases tested, as well as an endogenous E. coli protease, cut the protein within this sequence suggests that this region most probably forms an exposed linker sequence between a 27-kDa N-terminal domain and the remaining 50-kDa portion of the protein. The second proteasesensitive region we observed in yBPL occurred within the 50-kDa C-terminal region, which, based on sequence homologies, contains the catalytic center (18). In fact, both the trypsin and endoproteinase Glu-C-sensitive residues mapped in the putative ATP and biotin binding region of the catalytic domain, with the trypsin cleavage site within the proposed ATP binding motif, GRGRGG (18). These cleavage sites were protected from proteolysis and subsequent loss of yBPL activity in the presence of ATP and biotin (Figs. 5 and 6). Correspondingly, the ligand-bound form of yBPL was more resistant to the loss of enzyme activity that accompanied proteolysis than apoenzyme. These observations strongly support the identification of the catalytic site made from sequence homologies. In the crystal structure of the BirA, the region containing the ATP binding motif is one of several poorly defined, solvent-exposed loops found close to the catalytic center (5). A subtilisin-sensitive site lies within an unstructured loop adjacent to the ATP binding loop, and cleavage is inhibited when either biotin or biotinyl-5Ј-AMP is bound to the enzyme (37). We conclude that the results of limited proteolysis presented here indicate that yBPL forms two domains and that the cleavage observed around the ATP binding motif suggests the presence of an exposed loop structure within the 50-kDa C-terminal domain, as is seen in BirA. In addition, and somewhat surprisingly, the presence of ATP and biotin also reduced protease susceptibility within the proposed linker sequence between the two domains. This suggests that substrate binding caused a global conformational change, affecting sequences at some distance from the binding site in the primary structure. Conformational changes in BirA associated with substrate binding have also been demonstrated (4). Interestingly, the N-terminal domain of BirA that is involved in DNA binding appears to affect the affinity of the catalytic domain for both biotin and biotinyl-5Ј-AMP. A truncation mutant in which the N-terminal domain was absent was still able to catalyze biotin transfer but displayed a 100-fold decrease in the affinity for biotin and a 1000-fold decrease in the affinity for biotinyl-5Ј-AMP (36). Biotin binding to this truncation caused no quenching of intrinsic protein fluorescence, as opposed to the 15% quenching observed with the intact enzyme (36). These data suggest that quenching of fluorescence may be the result of the conformational changes that are induced by biotin binding and that the truncated enzyme is compromised in its ability to go through these changes. The interaction between the DNA binding domain and the catalytic domain is accompanied by conformational changes and is thought to relate to repressor function of BirA (5). Our data indicate that although the Nterminal domain in yBPL has no equivalent DNA binding function, there is a functional interaction with the catalytic domain.
Furthermore, expression of N-terminally truncated variants of yBPL in two E. coli strains carrying birA mutations showed that the presence of both domains was necessary to produce a functional enzyme. Our analysis of the activity of the yBPL truncations in vitro is consistent with the results of the complementation assays, and it is evident that the 3500-fold reduction in activity seen upon the deletion of the N-terminal domain reduced the activity of yBPL to a level that is inadequate for viability. It is likely that in the absence of the N-terminal domain conformational changes associated with substrate binding, necessary for enzymatic activity, may occur at a slower rate than in the presence of the domain and therefore affect the overall activity of the protein. This agrees with the observation that tryptic cleavage of yBPL in the linker region produced a form of yBPL that retained some catalytic activity in in vitro assays (Fig. 6). While the precise role of the Nterminal domain is unclear, the results presented here are consistent with the studies of known defects in human BPL. Several point mutations in the N-terminal domain of human BPL result in a defective enzyme (19,22,23), indicating that the integrity of this region of the protein is important for function. Sequence homology between yeast and human BPL in the N-terminal domain is low and allows different alignments, making it difficult to precisely identify analogous residues and therefore to produce point mutations in yBPL that mimic those isolated in the defective human enzyme. However, these mutations in human BPL are found upstream of the proteasesensitive linker region in yBPL identified here (17), consistent with the inability of the N-terminal truncation yBPL(⌬1-233) to complement the birA defects. Structural characterization of the N-terminal domain and the identification of interactions with both the catalytic site and other molecules will aid in determining the role of this domain in enzyme function.
While BPL has been purified from a variety of sources, the low abundance of the enzyme has made purification of the endogenous enzyme a difficult task. Since the availability of recombinant DNA technology has permitted high level production of proteins in suitable hosts with improved yields, protein overexpression has facilitated isolation of BirA (38,39) and Arabidopsis thaliana BPL from E. coli. Here we report recom-binant expression in bacteria of a eukaryotic member of this enzyme family. Cloning a hexahistidine tag onto the C terminus of yBPL permitted the isolation of active enzyme in a rapid two-step purification, with yields comparable with those reported for recombinant production of the bacterial protein (38,39). Whereas partial purification of BPL from S. cerevisiae has been reported previously (40), our system allowed purification to apparent homogeneity (Fig. 1).
The results of our steady-state kinetic analysis of yBPL indicate that biotinylation occurs through a two-step Bi Uni Uni Bi ping-pong mechanism. In the first partial reaction, the enzyme complexes with ATP and biotin and catalyzes the synthesis of biotinyl-5Ј-AMP with subsequent release of pyrophosphate. The addition of the apo acceptor protein then follows in the second partial reaction with the release of the biotinylated protein and AMP. The kinetics of the formation of the adenylated intermediate has been quantitatively analyzed using BirA (41). The enzyme-biotinyl-5Ј-AMP complex is quite stable (41) and is proposed to be the most abundant enzyme form in the cell. Here we demonstrate that the precursors of the intermediate, ATP and biotin, bind to yBPL in an ordered manner. As has been observed with plant BPL (20), ATP binds to the yeast enzyme before biotin. This is in contrast to the reaction pathway catalyzed by bacterial BirA, where biotin is the first ligand to bind the enzyme (41). The order of substrate binding in E. coli is believed to allow more responsive regulation of biotin biosynthesis. When the cellular demand for biotin is low, the BirA-biotinyl-5Ј-AMP complex occupies the bio operator sequence and represses transcription of the biotin biosynthetic operon. As apoBCCP levels increase, biotin is transferred from the enzyme-bound adenylate to the protein-bound form, with concomitant derepression of the biotin operon. Kinetic analysis of the interaction of BirA with biotin and ATP indicates that formation of the repressor complex is highly sensitive to biotin, and the K m for ATP is in the low millimolar range (4,29,41). In contrast, BPLs from biotin auxotrophic species generally bind ATP at lower concentrations, with K m values in the range of 0. 38 -200 M (42-45). The value of 21 M reported here for S. cerevisiae is well below the intracellular concentration of ATP (46) and is consistent with the absence of repressor function in the eukaryotic BPLs.
The formation of an activated enzyme-bound biotinyl intermediate in the first partial reaction of BPL requires a nucleotide triphosphate and a divalent metal ion (34). BPLs from different organisms differ in their specificity for both the NTP source and the divalent metal. For example, magnesium is the preferred metal ion for pea BPL (21), whereas zinc and manganese ions can readily substitute for magnesium ions for the P. shermanii enzyme (6,7). For the yeast enzyme, calcium or magnesium ions were the preferred divalent metals, with reasonable levels of activity also seen in the presence of nickel or manganese, while cobalt, zinc, and copper ions failed to support significant activity. Yeast BPL had an absolute requirement for ATP, since only minimal enzyme activity was detected when ATP was replaced by other nucleotide triphosphates. This is similar to BPLs from P. shermanii, B. stearothermophilus, pig, and rabbit, which are also specific for ATP (7,8,34,42). However, UTP can replace ATP for the enzyme from chicken liver (43,47), whereas CTP is the preferred nucleotide for both the pea (21) and bovine liver (44) biotin ligases.
Kinetic studies on BPLs have reported a wide range of K m values for biotin, ranging from 4.7 nM for the rabbit liver enzyme (42) to 3.3 M in chicken liver (43). Here we observed a low K m for biotin (67 nM) for yeast BPL. Bakers' yeast is auxotrophic for biotin, which is actively transported into the cell via the recently cloned H ϩ -biotin symporter, VHT1 (48). This membrane-bound transport protein displays maximal activity when cells are grown in media containing an extremely low concentration of biotin, 0.8 nM and is inhibited by greater than 8 nM biotin (48). Rogers and Lichstein (49) demonstrated that the cellular concentration of free biotin in yeast can be increased against a concentration gradient. Under conditions that give maximal growth, the free biotin pool can reach a concentration of 70 M (49), which would be saturating for yBPL. The mechanism of transport inhibition is not understood, and the possibility that yBPL interacts with the receptor to regulate biotin metabolism in yeast is both speculative and interesting.
An upper limit for the bimolecular rate constant for the formation of the enzyme-biotin complex can be determined using the values obtained in the kinetic analysis. The calculated k cat /K m value for biotin of 6.0 Ϯ 0.08 ϫ 10 6 M Ϫ1 s Ϫ1 is several orders of magnitude smaller than that predicted for a diffusion-controlled process. This value is 3-fold smaller than that of BirA (50) but 10-fold larger than the value determined for A. thaliana BPL (20), suggesting subtle structural differences in the active sites of these BPLs. yBPL was found to be highly specific for binding biotin, since several closely related biotin analogues and lipoic acid all failed to inhibit the incorporation of biotin even when present at a 100-fold excess over [ 3 H]biotin. This substrate specificity is a common feature of BPLs from a variety of sources (6,21,44). Thus, the decreased activity of biotin-dependent enzymes in rat liver seen after the administration of lipoic acid (51) seems unlikely to be due to a direct effect of lipoic acid on BPL in vivo.
We have used two isolated domains as the biotin acceptor domain in our analysis of yBPL, yPC-104, and BCCP-87. This latter peptide has been shown to be as effective a substrate for biotinylation by BirA as intact BCCP (52), with a K m of 4 M in our assay system (29). In the present study, a similarly low K m value was determined when a yeast biotin domain was assayed with yBPL (1 M). However, we observed a greater than 10-fold higher K m (11 M) when BCCP-87 was the substrate for yBPL. There is evidence of cross-species reactivity in biotinylation reactions (13)(14)(15), but kinetic analysis of the interactions has not been previously performed. It seems likely that the differences in K m for the two acceptor proteins for yBPL reflects subtle changes in substrate recognition or efficiency of biotin transfer between the two proteins in the assay system.