Biotinylation in the hyperthermophile Aquifex aeolicus

Isolation of a cross-linked BPL:BCCP complex

Authors


  • Enzyme: Biotin protein ligase or holocarboxylase synthase (EC 6.3.4.10).

D. Campopiano, School of Chemistry, University of Edinburgh, West Mains Road, Edinburgh, EH9 3JJ, UK. Fax: + 44 131 650 4743, Tel.: + 44 131 650 4712, E-mail: Dominic.Campopiano@ed.ac.uk

Abstract

Biotin protein ligase (BPL) catalyses the biotinylation of the biotin carboxyl carrier protein (BCCP) subunit of acetyl CoA carboxylase and this post-translational modification of a single lysine residue is exceptionally specific. The exact details of the protein–protein interactions involved are unclear as a BPL:BCCP complex has not yet been isolated. Moreover, detailed information is lacking on the composition, biosynthesis and role of fatty acids in hyperthermophilic organisms. We have cloned, overexpressed and purified recombinant BPL and the biotinyl domain of BCCP (BCCPΔ67) from the extreme hyperthermophile Aquifex aeolicus. In vitro assays have demonstrated that BPL catalyses biotinylation of lysine 117 on BCCPΔ67 at temperatures of up to 70 °C. Limited proteolysis of BPL with trypsin and chymotrypsin revealed a single protease-sensitive site located 44 residues from the N-terminus. This site is adjacent to the predicted substrate-binding site and proteolysis of BPL is significantly reduced in the presence of MgATP and biotin. Chemical crosslinking with 1-ethyl-3-(dimethylamino-propyl)-carbodiimide (EDC) allowed the isolation of a BPL:apo-BCCPΔ67 complex. Furthermore, this complex was also formed between BPL and a BCCPΔ67 mutant lacking the lysine residue (BCCPΔ67 K117L) however, complex formation was considerably reduced using holo-BCCPΔ67. These observations provide evidence that addition of the biotin prosthetic group reduces the ability of BCCPΔ67 to heterodimerize with BPL, and emphasizes that a network of interactions between residues on both proteins mediates protein recognition.

Abbreviations
BPL

biotin protein ligase

BCCP

biotin carboxyl carrier protein

IPTG

isopropyl thio-β-d-galactoside

EDC

1-ethyl-3-(dimethylamino-propyl)-carbodiimide

HCS

holocarboxylase synthase

The enzymes of bacterial fatty acid biosynthesis have been suggested as good targets for the development of novel antibacterial agents since several natural product and synthetic inhibitors of this pathway are already known [1]. Moreover, significant differences in fatty acid biosynthesis between bacteria and mammals should allow selective inhibition of the microbial enzymes. The first committed step of bacterial fatty acid biosynthesis is catalysed by a multisubunit acetyl-CoA carboxylase [2]. This biotin-dependent complex is composed of biotin carboxylase, carboxyltransferase and biotin carboxyl carrier protein (BCCP) subunits, the exact composition of which is species-specific. The Escherichia coli acetyl-CoA carboxylase has been intensively studied, because the subunits can be separated or expressed individually in an active form. Biotin is covalently bound to a specific lysine residue in the BCCP subunit [3,4]. Biotinylated enzymes transfer cardon dioxide from bicarbonate to organic acids to form cellular metabolites, using the biotin prosthetic group as a mobile carboxyl carrier [5]. Biotin protein ligase (BPL), also known as holocarboxylase synthase (HCS, EC 6.3.4.10) catalyses this post-translational attachment via a two-step reaction (Scheme 1 [6]).

Genes encoding BPLs have been identified in a number of organisms, but the best-characterized BPL is the 35.3 kDa BirA protein from E. coli[7,8]. BirA is a bifunctional protein that can act as both an enzyme and a DNA-binding protein; it catalyses protein biotinylation when in vivo biotin concentrations are low, but becomes a repressor of the expression of biotin biosynthetic enzymes when biotin concentrations are increased. The crystal structure of the biotin-bound protein, determined at 2.3 Å resolution, shows the enzyme has three domains [9,10]; an N-terminal domain that contains a helix-turn-helix fold for DNA binding; a central catalytic domain, which contains a highly conserved GRGRRG motif shown to be involved in biotin binding [11]; and a small C-terminal domain which has been postulated to mediate dimerization with apo-BCCP [12]. The recent determination of the structure of a BirA dimer in the absence of DNA provides insight into how the N-terminal DNA-binding domain interacts with the 40 bp biotin operator sequence [12]. The structure of the apo- and holo-forms of the biotinylation domain of E. coli BCCP (known as BCCP-87) have been determined by X-ray crystallography and NMR [13–15]. The BCCP domain is a barrel consisting of two antiparallel β-sheets each containing four strands. The N- and C-termini are close together at one end, and the biotinylated lysine is exposed on a tight β-turn at the opposite face of the molecule. Surprisingly, the structures of the apo- and holo- forms are remarkably similar suggesting that biotinylation causes few significant changes in the domain tertiary fold.

To gain further insight into the detailed protein–protein interactions that control biotin transfer we have analysed the reaction between BPL and apo-BCCP from the hyperthermophilic organism Aquifex aeolicus[16]. This bacteria grows optimally at 95 °C on hydrogen, oxygen, carbon dioxide and mineral salts. Enzymes from extremophiles (extremozymes) are offering new opportunities for biocatalysis as a result of their extreme stability [17,18]. Analysis of the A. aeolicus genome identified BirA and BCCP homologues; the predicted BPL is from the group I class (which also includes M. tuberculosis) which lack the N-terminal DNA-binding domain found in E. coli BirA [19]. In E. coli, we have expressed active A. aeolicus BPL, the biotin-binding domain of A. aeolicus BCCP as a His6 N-terminal fusion (BCCPΔ67) as well as an A. aeolicus BCCP mutant lacking the active lysine residue (K117L). Biotinylation of apo-BCCPΔ67 by BPL was most efficient at 70 °C and we have carried out kinetic analyses and proteolysis experiments at this temperature. Furthermore, we describe the isolation of a chemically crosslinked BPL:BCCPΔ67 complex for the first time. This study is the first characterization of post-translational modification complex from a hyperthermophilic organism.

Experimental procedures

Materials

All chemicals used in the preparation of buffers were at least of reagent grade. Nu-PAGE gels were obtained from Invitrogen; restriction endonucleases were purchased from New England Biolabs; [14C]biotin (54 mCi·mmol−1) was from Amersham Biosciences; and 1-ethyl-3-(dimethylamino-propyl)-carbodiimide (EDC) was from Sigma. PCR was performed using Ready To Go PCR™ beads (Amersham Biosciences).

Oligonucleotide primers were purchased from SigmaGenosys. The primer details are as follows (restriction sites are indicated by underlining and mutagenic changes are shown in bold). BPL-for, 5′-TTCTTAACCATGGGCTTCAAAAACCTGAT-CTGG-3′; BPL-rev, 5′-TTAAGGATCCTAAGAACGAGACAGGCTGAACTCTCC-3′; BCCPΔ67, 5′-GTAACCATGGGTGAACAGGAAGAA-3′; BCCP-rev, 5′-GGATCCTTAAACGTTTGTGTC TATAAG-3′; BCCP K117L, 5′-GAAGCTCTACTG GTTATGAAC-3′.

DNA was isolated from agarose using a QIAquick® Gel Extraction Kit, and plasmid DNA was purified using a QIAprep® Spin Miniprep Kit (both Qiagen). A. aeolicus genomic DNA was a kind gift from R. V. Swanson (Diversa, San Diego, USA), R. Huber and K. Stetter (University of Regensburg, Germany). All growth media were prepared following standard procedures [20].

Nucleic acid manipulations

DNA manipulations were performed using standard protocols [20]. Standard conditions were used for restriction endonuclease digestions, agarose gel electrophoresis and DNA ligation reactions, according to the manufacturer's instructions. All nucleic acid constructs were confirmed by commercial DNA sequencing (MWG Biotech).

Cloning of BPL, BCCPΔ67 and BCCPΔ67 K177L from A. aeolicus

The A. aeolicus bpl and bccpΔ67 genes were amplified from A. aeolicus genomic DNA template by polymerase chain reaction using primers BPL-for and BPL-rev; and BCCPΔ67 and BCCP-rev, respectively. The PCR products were cloned into plasmid pCR2.1 (Invitrogen) using standard TOPO cloning procedures, yielding the plasmids pCR2.1/BPL and pCR2.1/BccpΔ67. Positive clones were sequenced to confirm the fidelity of the insert and a restriction digest was performed on the pCR2.1/BPL plasmid using the restriction endonucleases NcoI and BamHI. The isolated 723 bp fragment containing the A. aeolicus BPL gene was cloned in NcoI/BamHI-digested pET28a (Novagen), producing the expression vector pET28a/BPL. An NcoI/BamHI digest was performed on plasmid pCR2.1/BccpΔ67 and the resulting 259 bp fragment containing the truncated Bccp gene was ligated in a NcoI/BamHI-digested pET derivative (Novagen). The resulting expression vector, pET6H/BccpΔ67, produced a His6 fusion at the N-terminus of bccpΔ67.

A bccpΔ67 mutant gene encoding a mutation of the active site lysine to a leucine residue (K117L) was produced by the PCR megaprimer method [21]. The primers used were BCCPΔ67, BCCP-rev and BCCPΔ67 K117L and the plasmid pCR2.1/BccpΔ67 was used as the PCR template. The mutant gene PCR product was cloned into pCR2.1 and the resulting plasmid was named pCR2.1/BccpΔ67 K117L. To express the mutant bccpΔ67 with an N-terminal His6-tag, pET6H/BccpΔ67 K117L, was produced in the same fashion as described for the wild-type protein.

Expression and purification of A. aeolicus BPL

The pET28a/BPL vector was used to transform E. coli BL21(DE3) cells (Novagen). A single colony was used to inoculate 200 mL LB broth supplemented with kanamycin (30 µg·mL−1) and grown overnight at 37 °C and 250 r.p.m. This seed culture was then used to inoculate 4 L of fresh growth medium and grown at 37 °C to D600 = 1.0 before induction with 1.0 mm isopropyl thio-β-d-galactoside (IPTG). After a further 3 h growth the cells were harvested by centrifugation (4000 g for 15 min at 4 °C) and washed with 10 mm Hepes (pH 7.5). The cells were resuspended in 10 mm Hepes buffer (pH 7.5) and disrupted by sonication (15 pulses of 30 s at 30-second intervals) at 4 °C. The cell debris was removed by centrifugation at 27 000 gfor 20 min at 4 °C.

One tablet of Complete™ Proteinase Inhibitor Cocktail (Roche) was added to the cell lysate before it was incubated at 60 °C for 20 min. Precipitated cellular debris was removed by centrifugation at 27 000 g for 20 min at 4 °C. The supernatant was filtered through a 0.45-µm membrane before it was loaded onto a 6-mL Resource-S cation exchange column (Amersham Biosciences) equilibrated with 10 mm Hepes (pH 7.5) at room temperatutre. The BPL protein was eluted with a linear salt gradient (0–1 m NaCl in 10 mm Hepes, pH 7.5) over 20 column volumes (120 mL). Fractions containing BPL (eluting at ≈ 200 mm NaCl) were analysed by SDS/PAGE and those fractions judged to be 95% pure were pooled and stored in 10 mm Hepes (pH 7.5) containing 20% glycerol (v/v) at −20 °C. Protein concentration was determined using the Bio-Rad protein assay kit.

Expression and purification of Apo-BCCPΔ67 and BCCPΔ67 K117L from A. aeolicus

Overexpression of A. aeolicus BCCPΔ67 was achieved by transforming E. coli BL21(DE3) cells with the plasmid pET6H/BccpΔ67. A single colony was used to inoculate 200 mL 2YT supplemented with ampicillin (100 µg·mL−1) and grown overnight at 37 °C and 250 r.p.m. This seed culture was then used to inoculate 4 L of fresh growth medium and grown at 37 °C to D600 = 1.0 before induction with IPTG (1.0 mm final concentration). After a further 3 h the cells were harvested by centrifugation (4000 gfor 15 min at 4 °C) and washed in binding buffer (20 mm Tris/HCl, pH 7.5, 0.5 m NaCl, 5 mm imidazole). The cells were resuspended in binding buffer (5 mL per gram of wet cell paste) and disrupted by sonication (15 pulses of 30 s at 30-second intervals) at 4 °C. The cell debris was removed by centrifugation at 27 000 gfor 20 min at 4 °C, after which the supernatant was filtered through a 0.45-µm membrane prior to chromatography.

The cell lysate was applied to a Hitrap® chelating affinity column (Amersham Biosciences) previously loaded with charge buffer (100 mm NiS04) and equilibrated with binding buffer at room temperature. The column was then washed with 5 column volumes of binding buffer before bound material was eluted using a linear gradient of 0–100% elution buffer (20 mm Tris/HCl, pH 7.5, 0.5 m NaCl, 1 m imidazole). Fractions were analysed by SDS/PAGE and those containing BCCPΔ67 were pooled and dialysed overnight against 4 L of 10 mm Hepes (pH 7.5) at 20 °C.

Apo-BCCPΔ67 and holo-BCCPΔ67 were separated by applying the BCCPΔ67-containing fractions eluted from the nickel column onto a 1-mL Mono-Q column (Amersham Biosciences) pre-equilibrated with 10 mm Hepes (pH 7.5) at room temperature. The column was then washed with 20 column volumes of 10 mm Hepes (pH 7.5), before the protein was eluted with a salt gradient (0–100% 10 mm Hepes, 1 m NaCl, pH 7.5) over 25 column volumes. Fractions containing apo-BCCPΔ67 (confirmed by LC-MS analysis) were pooled and stored in 10 mm Hepes (pH 7.5) containing 20% glycerol (v/v) at −20 °C. Due to the low proportion of aromatic residues in BCCPΔ67, protein concentration was evaluated by measuring the absorbance at 280 nm and using the conversion factor calculated using vector nti5 software.

The expression and purification of the BCCPΔ67 K117L mutant was performed in a similar way to the wild type protein. Elution from the Mono-Q column produced a single, apo-form peak.

Mass spectrometry characterization of proteins

Mass spectrometry was performed on a MicroMass Platform II quadrupole mass spectrometer equipped with an electrospray ion source. The spectrometer cone voltage was ramped from 40 to 70 V and the source temperature set to 140 °C. Protein samples were separated with a Waters HPLC 2690 with a Phenomenex C5 reverse phase column directly connected to the spectrometer. The proteins were eluted from the column with a 5–95% acetonitrile (containing 0.01% trifluoroacetic acid) gradient at a flow rate of 0.4 mL·min−1. The total ion count in the range 500–2000 m/z was scanned at 0.1 s intervals. The scans were accumulated and spectra combined and the molecular mass determined by the maxent and transform algorithms of the mass lynx software (MicroMass).

Assay of A. aeolicus BPL

BPL activity was assayed by measuring the incorporation of [14C]biotin into purified BCCPΔ67, in a similar way to that described previously [22]. Except where stated otherwise, the reaction contained 10 mm Hepes (pH 8.5), 100 µm ATP, 200 µm MgCl2, 10 µm biotin, 1 µm[14C]biotin (specific activity 54 mCi·mmol−1), 0.1 mg·mL−1 bovine serum albumin, and 400 µm apo-BCCPΔ67. The reaction was initiated by the addition of purified BPL to a final concentration of 1 µm, and incubated at 70 °C for 30 min. The reaction was terminated by the addition of ice-cold trichloroacetic acid (final concentration 25% w/v), and incubation on ice for 30 min. The resulting protein precipitate was removed by centrifugation (27 000 g for 10 min). Aliquots of the supernatant were added to 5 mL of scintillation fluid (ICN biomedicals), and radioactivity was measured using a Tri-carb 210 OTR liquid scintillation counter (Packard). The extent of BCCPΔ67 biotinylation was deduced from the decrease in [14C]biotin in the supernatant.

For kinetic analysis each of the substrate concentrations (biotin, ATP, BCCP) was varied accordingly. Values for Km and Vmax were determined by Michaelis–Menten analysis on sigmaplot 2001 software. In some assays, to obtain sufficiently high levels of activity for accurate detection, it was necessary to continue until more than 10% of the limiting substrate had been used. In these instances the data was transformed using the method of Lee and Wilson and plotted as transformed values s′ and v′[23].

To demonstrate the formation of the reaction intermediate, biotinyl-5′-AMP, we employed a streptavidin-binding assay. Briefly, the reaction contained 10 mm Hepes (pH 8.5), 10 µm biotin, 100 µm[8-14C]ATP (specific activity 50–62 mCi·mmol−1), 200 µm MgCl2 and 0.1 mg·mL−1 bovine serum albumin. The reaction was initiated by the addition of purified BPL to a final concentration of 5 µm, and incubated at 70 °C for 30 min. Ice-cold trichloroacetic acid (final concentration 10% w/v) was used to terminate the reaction and the resulting precipitate of BPL was removed by centrifugation. Aliquots of the assay were then spotted onto a single SAM® Biotin Capture Membrane (Promega). Unreacted [α-14C]ATP was removed by washing each membrane four times in 2 m NaCl, four times in 2 m NaCl in 1% H3PO4, and twice in water. Finally the membrane was added to 5 mL of scintillation fluid (ICN biomedicals), and the radioactivity of the retained, bound biotinyl-5′-[α-14C]AMP was measured using a Tri-carb 210 OTR liquid scintillation counter (Packard).

Limited proteolysis of BPL

Proteolysis of apo-BPL and substrate-bound-BPL were investigated using the proteases trypsin (Sigma) and chymotrypsin (Promega). Substrate-bound BPL was prepared by incubating BPL (15 µm) for 20 min at 60 °C with saturating amounts of biotin (40 µm), MgATP (2 mm), or both. The samples were then cooled for 10 min before treatment with protease, with a final protease/substrate ratio of 1 : 20 (w/w), and incubation at 37 °C for 30 min. Digestion was terminated by the addition of SDS sample buffer and boiling for 5 min. The extent of proteolysis was analysed by SDS/PAGE and densitometry analysis of the gel spots was performed using imagemaster total laboratory Software (Amersham Biosciences).

Chemical crosslinking of A. aeolicus BPL and Apo-BCCPΔ67

Purified BPL (15 µm) and either apo-BCCPΔ67, holo-BCCPΔ67 or BCCPΔ67 K117L (45 mm) were covalently cross-linked using 1-ethyl-3-(dimethylamino-propyl)-carbodiimide (EDC, 10 mm) at 60 °C for 60 mins. Aliquots were withdrawn at various time intervals, quenched with ammonium acetate (100 mm), and analysed by SDS/PAGE.

The cross-linked complex was prepared on a larger scale and separated from BPL and BCCPΔ67 by gel filtration. To prepare the complex we incubated 5 mg each of BPL and BCCP, EDC (10 mm) in a final volume of 5 mL 10 mm Hepes (pH 8.5) for 60mins at 60 °C. The mixture was concentrated to 1 mL and then passed through a Superdex 75 column (Amersham Biosciences) equilibrated in 10 mm Hepes (pH 8.5) and 100 mm NaCl. The purified protein was stored at −20 °C.

Results

Analysis of the A. aeolicus genome

The complete genome sequence of A. aeolicus consists of 1512 predicted open reading frames [16]. We performed a blast search on the complete genome and identified two ORFs of 233 aa and 154 aa with high sequence homology to E. coli BirA (20.9% identity, 35.2% similarity) and BCCP (33.8% identity, 46.9% similarity), respectively. The pairwise sequence alignments generated by clustal w[24] are shown in Fig. 1and these enabled us to design PCR primers to clone the A. aeolicus BPL and BCCP genes. We noted from this initial analysis that the A. aeolicus BPL differs from the E. coli BirA in that it lacks an N-terminal DNA-binding domain which places it in the group I class of BPLs along with those from Mycobacterium tuberculosis and Thermotoga maritima[19].

Figure 1.

Sequence alignments of BCCP(A) and BPL(B) from E. coli and A. aeolicus. Pairwise alignment was prepared using clustal w. (A) The start residue of the BCCP-87 domain and the BCCP subtilisin fragment are indicated (↓ and ∇, respectively). The start codon of the BCCPΔ67 domain is shown (↑), and the biotinylated lysine residue is indicated (◆). Secondary structural elements of the BCCP-87 domain are shown and the ‘thumb’ region is indicated (*). (B) Pairs of disordered surface loops which are close in space in the E. coli BirA structure are shown (∼ and +) The trypsin cleavage sites of A. aeolicus BPL are indicated ($) as is the site of subtilisin cleavage of E. coli BirA (*).

Previous studies on full-length E. coli BCCP (156 aa) revealed that the protein forms a tight complex with the biotin carboxylase (BC) subunit in solution, which complicates biochemical studies [25]. In most cases, the biotin carrier domain of biotin-containing enzymes is located at the C-terminal end of the carboxylase, with the biotinyl-lysine about 35 residues from the C-terminus. Structural studies revealed that a 65–70 amino acid fragment of BCCP, previously suggested by deletion mutagenesis, is required to form a minimal structured biotin domain [26]. Various truncated forms of the E. coli BCCP have been used in biochemical and structural studies, containing between 80 and 87 residues from the C-terminus of the protein. Here we expressed A. aeolicus BCCP lacking 67 residues from the N-terminus (BCCPΔ67, Fig. 1) with an N-terminal His6-tag (total length 96 aa). The homology scores between A. aeolicus BCCPΔ67 and E. coli BCCP-87 (a domain containing 87 C-terminal amino acids) are 51.9% identity and 69.6% similarity (Fig. 1).

Cloning, expression and purification of BPL

The A. aeolicus bpl gene was amplified by PCR using A. aeolicus genomic DNA as a template and cloned into plasmid pCR2.1. DNA sequencing confirmed the previously published gene sequence, with the exception of a single base change at position 325 (T→C), which results in the substitution of a cysteine residue with an arginine. Subsequently the bpl gene was cloned into a pET expression vector for expression in various E. coli cells (DE3 lysogens); we found optimum recovery of protein using the BL21(DE3) strain. Cells were grown in shake flasks at 37 °C and expression induced with 1 mm IPTG (see Experimental procedures).

The predicted pI of the A. aeolicus BPL is 9.1 and as the enzyme contains a high proportion of positively charged residues, cation-echange chromatography was used to purify it in a single step (Fig. 2, lanes 2–4). Initially the crude lysate was incubated at 60 °C which resulted in the precipitation of a significant quantity of E. coli proteins. It was then necessary to dialyse the sample overnight (20 °C) against 10 mm Hepes (pH 7.5) as immediate loading of an untreated extract onto a ResourceS column resulted in very poor binding (< 5%). It is unclear why this step was necessary, but after dialysis binding to the cation-exchange column approached 100%. BPL eluted from the column at 200 mm NaCl and we obtained the enzyme with a purity of greater than 95% (as determined by SDS/PAGE). Electrospray mass spectrometry analysis gave the molecular mass of the protein as 26636.8 ± 2.3 Da, consistent with the post-translational removal of the N-terminal methionine residue, and accurate to within experimental error of the predicted value of 26634.6 Da. The final yield of BPL using this method was > 10 mg per litre of cell culture and this protein was used for all subsequent kinetic and cross linking analysis.

Figure 2.

Purification of A. aeolicus BPL, BCCPΔ67 and BCCPΔ67 K117L. Protein purification was analysed by SDS/PAGE under reducing conditions. Lanes 1, 5 and 9, low molecular mass marker; lane 2, BPL cell lysate; lane 3, BPL cell lysate after heat purification; lane 4, BPL after ResourseS purification; lane 6, BCCPΔ67 cell lysate; lane 7, BCCPΔ67 after Ni-affinity purification; lane 8, apo-BCCPΔ67 after Mono-Q purification; lane 10, BCCPΔ67 K117L cell lysate; lane 11, BCCPΔ67 K117L after Ni-affinity purification; lane 12, BCCPΔ67 K117L after Mono-Q purification.

Cloning, expression and purification of BCCPΔ67

We designed primers to clone a truncated domain of the A. aeolicus bccp gene missing the first 201 bp, which encode the N-terminal 67 amino acids of A. aeolicus BCCP (Fig. 1). The truncated gene was amplified from genomic DNA using PCR and cloned into the pCR2.1 vector. DNA sequencing confirmed the expected gene sequence, and the bccpΔ67 gene was subsequently cloned into a pET-derived expression vector with an N-terminal His6-tag. E. coli BL21(DE3) competent cells were used for recombinant expression (described under Experimental procedures) and the BCCPΔ67 cell lysate was first purified by nickel-affinity chromatography (Fig. 2, lanes 6–8). The protein eluted with 200 mm imidazole and, as precipitation had been observed at high concentrations of this eluant, it was immediately diluted 1 : 1 with 10 mm Hepes (pH 7.5) and dialysed against this buffer. SDS/PAGE analysis indicated BCCPΔ67 to be > 90% pure but electrospray mass spectrometry revealed the presence of two distinct species. The first, of molecular mass 10740.1 ± 1.1 Da, corresponded to the predicted mass of apo-BCCPΔ67 (10739.6 Da) while the second corresponding to the holo-form (biotinyated), with a mass increase of 226.1 Da (10965.4 Da; predicted mass 10965.7 Da). This confirmed that the A. aeolicus BCCPΔ67 domain folded correctly, and was recognized and biotinylated by the host E. coli BirA. To separate the apo- and holo-forms of BCCPΔ67 we employed anion exchange chromatography in a similar way to that used for E. coli BCCP-87 [27]. Fractions from the column were analysed by electrospray mass spectrometry and the apo-protein eluted at a slightly lower salt concentration than the holo-form (160–240 mm NaCl vs. 240–320 mm NaCl). Approximately 80% of the apo-BCCPΔ67 was resolved from the holo-form by collecting only the leading fractions of the protein peak. The final yield of apo-BCCPΔ67 was ≈ 5–10 mg per litre of cell culture and ≈ 1 mg per litre of the holo-form.

Cloning, expression and purification of BCCPΔ67 K117L mutant

A mutant of the truncated bccpΔ67 gene, with the active lysine residue (K117) replaced by a leucine residue, was produced using the megaprimer method [21]. The mutation was confirmed by DNA sequencing before the gene was cloned into a pET-derived expression vector with an N-terminal His6-tag and the resulting construct was then transformed into E. coli BL21(DE3) cells for expression (as described in Experimental procedures). The BCCPΔ67 K117L protein was purified using nickel-affinity chromatography and the protein eluted with 200 mm imidazole (Fig. 2, lanes 10–12). Protein-containing fractions were immediately dialysed against 10 mm Hepes (pH 7.5). Further purification on anion-exchange chromatography gave a single species with a mass of 10724.8 ± 1.1 Da, consistent with the predicted mass of apo-BCCPΔ67 K117L of 10724.6 Da. A species was not present at +226 Da, an indication that in vivo biotinylation had not occurred. The yield of the apo-BCCPΔ67 K117L mutant was ≈ 15 mg of protein per litre of cell culture.

Biochemical properties of BPL

Activity assays were performed with BPL by measuring the incorporation of [14C]biotin into the purified apo-BCCPΔ67 biotin-accepting domain [22]. In initial experiments we observed optimal enzyme activity at pH 8.5, and magnesium ions, ATP, biotin and apo-BCCPΔ67 were all required for activity. The activity of the enzyme was also measured at varying temperatures, with optimal activity at 70 °C. Activity was seen to decrease by roughly 50% for every 10 °C drop in temperature, and increasing the temperature above 70 °C resulted in enzyme precipitation, together with a dramatic loss in activity (data not shown). The tolerance of BPL for other nucleotide sources was measured by replacing ATP with UTP, GTP or CTP. No BPL activity was detected for any of these three substrates, suggesting that the enzyme is completely dependent on ATP for its nucleotide supply (data not shown).

In assays performed with BCCPΔ67 K117L as the biotin acceptor no biotinylation was observed, verifying K117 as the active residue and demonstrating the specificity of the BPL catalysed reaction.

Kinetic analysis of BPL

The kinetic constants for d-biotin, MgATP and apo-BCCPΔ67 were determined using steady-state kinetics (Fig. 3). The Km for d-biotin was determined to be 440 ± 70 nm. The Km values for BPLs from other species range from low nanomolar to low micromolar; 67 ± 11 nm (Saccharomyces cerevisiae BPL), 300 nm (E. coli BirA), 130 nm (Arabidopsis thaliana HCS) and 3.3 mm (chicken liver HCS1) [28–31]. The Km for MgATP was 15.1 ± 1.5 µm, which is similar to that determined for the S. cerevisiae BPL (20.9 ± 3 µm) and A. thaliana HCS (4.4 µm). In contrast, the Km for MgATP for E. coli BirA is around 300 µm. It should be noted that the kinetic analyses for each BPL were performed under slightly different reactions conditions, for example an elevated temperature was used in the study presented here. Finally, the Km for apo-BCCPΔ67 was 160 ± 32 µm. A range of biotinylation substrates have been used in assays of BPL activity with cross-species reactivity frequently observed, e.g. S. cerevisiae BPL has a Km of 11.1 ± 1 mm for E. coli BCCP-87. However, we could not test E. coli BCCP-87 as a substrate for BPL because the rate of biotinylation at 37 °C was outside the lower limit of detection in our assay.

Figure 3.

Steady-state kinetic analysis of BPL substrate binding. The activity of A. aeolicus BPL was measured under steady-state conditions at 70 °C. Two substrates were kept at constant saturating levels while the concentration of the third substrate was varied over the ranges shown above in the graphs. From the curves, Km values for biotin (A), MgATP (B) and apo-BCCPΔ67 (C) were determined (see Experimental procedures).

As shown in Scheme 1 the first step in all biotinylation reactions studied thus far involves the synthesis of a biotinyl-5′-AMP intermediate and the release of PPi. This molecule is the substrate for biotin transfer to BCCP and is also the corepressor of E. coli BirA. To prove that A. aeolicus BPL synthesises biotinyl-5′-AMP we incubated BPL with biotin and [14C]MgATP at 70 °C and used streptavidin-coated membranes to capture radioactive biotinyl-5′-[14C]AMP (data not shown). Furthermore, we noted that biotinylation was inhibited by the addition of NaCl in concentrations above 200 mm.

Proteolysis of BPL

We subjected BPL to limited proteolysis in the presence and absence of biotin and MgATP (Fig. 4). Digestion with both trypsin and chymotrypsin resulted in formation of a fragment of ≈ 21 kDa. Chymotrypsin digestion also produced an array of smaller peptide fragments. We found that only 34% of total BPL remained after trypsin cleavage in the absence of substrates. However, preincubation of BPL with saturating amounts of biotin or MgATP separately increased its resistance to digestion (50% and 63% remained, respectively). Moreover, preincubation with both substrates dramatically increased the resistance of BPL to proteolysis with trypsin (98.9% remained). Comparative analysis with chymotrypsin showed that 11% of BPL remained intact after digestion. Preincubation of the enzyme with MgATP afforded little protection (13% of BPL remaining), whereas 34% and 92% BPL remained after preincubation with biotin and biotin and ATP. Taken together these results suggest that the binding of the substrates and/or the formation of the intermediate, biotinyl-5′-AMP, plays a role in protecting BPL from protease cleavage.

Figure 4.

Proteolysis of A. aeolicus BPL.A. aeolicus BPL was treated with trypsin or chymotrypsin either with or without equilibrating the enzyme with 1 mm MgATP and/or 50 µm biotin. Lanes 1–4, Trypsin digest; lane 1 BPL; lane 2, BPL + MgATP; lane 3, BPL + biotin; lane 4, BPL + MgATP and biotin. Lanes 5–8 Chymotrypsin digest; lane 5, BPL; lane 6, BPL + MgATP; lane 7, BPL + biotin; lane 8, BPL + MgATP and biotin.

LC-MS analysis of the peptide fragment produced from BPL after treatment with trypsin revealed the presence of two distinct species of mass 215549.5 ± 2.6 Da and 21678.6 ± 5.9 Da. Primary structure analysis of BPL established these masses corresponded to trypsin cleavage between R44 and K45, and K45 and W46 adjacent to the proposed catalytic centre and biotinyl-5′-AMP binding site.

Chemical crosslinking of BPL and BCCP

Although structures of E. coli BirA and both apo- and holo-BCCP-87 have been determined, our goal was to isolate a BPL:BCCP complex for biochemical and structural studies. Previous work in our laboratory used the chemical crosslinking agent EDC to isolate an E. coli flavodoxin–flavodoxin reductase complex, so we used this reagent to crosslink BPL and various forms of BCCPΔ67 [32]. Initially we incubated BPL and apo-BCCPΔ67 in the presence of excess EDC at room temperature with and without saturating amounts of biotin and MgATP, but we did not observe any crosslinked species of predicted molecular mass ≈ 36 kDa on SDS/PAGE (data not shown). However, a species was observed when the incubation was carried out at elevated temperatures, with 60 °C being the optimum (Fig. 5A). The presence of the substrates had no observable effect on crosslinking. Interestingly, when BPL was incubated with holo-BCCPΔ67 and EDC the amount of crosslinked species generated was significantly reduced compared to the apo form (Fig. 5B). Moreover, the incubation of BPL with the BCCPΔ67 K117L mutant led to the formation of crosslinked complex in comparable amounts to that using apo-BCCPΔ67 (Fig. 5C). Purification of the BPL: BCCPΔ67 complex from unreacted proteins was achieved using size exclusion chromatography, which resolved the mixture into three peaks (Fig. 6). We noted that both BPL, BCCPΔ67 and the complex eluted from the size exclusion column at retention volumes different to that predicted by their molecular masses (45, 35 and 70 kDa, respectively). However, analysis by SDS/PAGE revealed that the BPL:BCCPΔ67 complex eluted from the column first and had a molecular mass of 37 kDa (Fig. 6, inset). Electrospray analysis of the complex gave a molecular mass of 37 200 ± 200 Da which agrees well with the predicted mass of a 1 : 1 heterodimer.

Figure 5.

SDS/PAGE analysis of chemical crosslinking assays. Gel A, crosslinking of BPL and apo-BCCPΔ67. Gel B, crosslinking of BPL and holo-BCCPΔ67. Gel C, crosslinking of BPL and BCCPΔ67 K117L. Lanes 1–5 of each gel, assay after 0, 5, 10, 15 and 30 min respectively. Gel A, lanes 6 and 7, control assays with BCCPΔ67 alone and BPL alone.

Figure 6.

Purification of the chemically crosslinked BPL:apo-BCCPΔ67 complex by size-exclusion chromatography. The chromatogram above was obtained when the cross-linking reaction was applied to a Superdex 75 column. The three peaks correspond to the crosslinked complex (7–8 mL), BPL (10 mL) and BCCPΔ67 (11–12 mL). Insert: SDS/PAGE analysis of the column fractions. Lane 1, cross-linking reaction before purification. Lane 2–11, 1 mL fractions eluting between 6 and 15 mL.

Discussion

The attachment of biotin to the specific lysine residue of the apo- forms of biotin-requiring enzymes is a complex, multistep reaction. The BPL enzyme (also known as holocarboxylase synthetase, HCS) catalysing this process first activates biotin as biotinyl-5′-AMP then transfers the biotin to a specific lysine of the BCCP domain. The BPLs and BCCPs from a diverse range of organisms including E. coli (BirA), yeast, human and plant have been isolated and it has been shown that the BPL from one organism can biotinylate the BCCP domain from another [28]. This suggests some degree of structural homology between these proteins and primary structure analysis reveals there is a high degree of amino acid sequence similarity throughout the catalytic domain of the BPL family and the biotinyl domain of BCCPs [33]. An understanding of the protein–protein interactions that mediate this highly specific reaction requires three dimensional structures of each of the components. The structure of the E. coli BirA monomer in complex with biotinyl-lysine revealed details of the protein–substrate interactions but several loops within the active site were disordered [9]. More recently, the structure of the BirA dimer has provided insights into how the ligase also acts as a transcriptional repressor by binding to the E. coli biotin operon operator [12]. The structures of the apo- and holo- forms of E. coli BCCP-87, determined by X-ray and NMR, are virtually identical and showed that the biotinyl-lysine residue is located at an exposed β-turn, flanked by important, highly conserved methionine residues [13,15]. A more recent NMR study, combined with results from site-directed and random mutagenesis [29,34,35], allowed modelling of the elusive E. coli BPL:BCCP-87 complex and it appears that its formation is dependent on subtle, competing protein–protein interactions [36].

Analysis of the complete genome of the hyperthermophile A. aeolicus revealed the presence of BPL and BCCP homologues (Fig. 1). The A. aeolicus BPL enzyme belongs to the class I group of BPLs since it lacks the DNA-binding domain found in BirA and is the smallest characterized thus far. Eukaryotic BPLs also lack predicted DNA-binding domains but have large N-terminal extensions with unknown functions [33]. The full-length A. aeolicus BCCP has a C-terminus showing high sequence homology to the biotin domains of biotin-carboxylases and contains the eight amino acid ‘thumb’ motif found in E. coli BCCP [33,37,38]. The N-terminus has a large proportion of charged residues, and displays little similarity to any other BCCPs.

Using recombinant proteins isolated from E. coli we have characterized the full-length BPL and BCCP biotinylation domain BCCPΔ67 (with a His6 N-terminal tag) from a hyperthermophile. We have gained insight into this extremely specific post-translational modification reaction at high temperatures and used features of the two A. aeolicus proteins to capture a BPL:BCCP complex. We found A. aeolicus BPL to be monomeric, and thus competing homodimerization interactions found in E. coli BirA are not present. We isolated a mixture of apo- and holo-forms of A. aeolicus BCCPΔ67 and so conclude that it must be a substrate for E. coli BPL in vivo. Biotinylation in hyperthermophiles proceeds via the two-step reaction sequence found in other organisms (Scheme 1). Isolated A. aeolicus BPL could biotinylate apo-BCCPΔ67 at temperatures up to 70 °C albeit at a slow rate. It is interesting to compare the A. aeolicus BPL:BCCPΔ67 biotinylation reaction with that of a mutant E. coli BirA lacking the N-terminal DNA binding domain (BirA65-321) and E. coli BCCP-87. The BirA65-321 mutant could synthesize biotinyl-5′-AMP and transfer biotin to apo-BCCP-87 at the same rate as wild-type BirA. However, the affinity of BirA65-321 mutant for biotin and biotinyl-5′-AMP was decreased 100-fold and 1000-fold, respectively [39]. This suggested that in BirA, the N-terminal domain is somehow involved in tight-binding of the two ligands. In future, it would be interesting to study a BPL:BirA chimera by fusing the DNA-binding domain at the N-terminus of A. aeolicus BPL.

Substrate Km values for BPLs from a number of species have been shown to range from the low nanomolar to low millimolar. In steady-state kinetic assays at 70 °C, the A. aeolicus BPL bound biotin, MgATP and apo-BCCPΔ67 with affinites of 440 nm, 15.1 µm and 160 µm, respectively. The kinetic constant for MgATP suggests that A. aeolicus BPL resembles those from eukaryotic biotin auxotrophs (low micromolar). In contrast, E. coli BirA binds MgATP with a Km in the low millimolar range which reflects its dual function as both repressor of biotin biosynthesis and biotin ligase. It is interesting to note that A. aeolicus contains all the genes required to convert pimelate to biotin (bioW, bioF, bioA, bioD and bioB) suggesting it can synthesize this vitamin but the in vivo concentration within A. aeolicus cells is unknown. The Km for the apo-BCCPΔ67 domain used in this study is high compared to others but this may reflect the fact that the first 67 amino acid residues, which contain a high number of charged residues, could play an important role in tight binding to BPL. Most biochemical studies use these truncated BCCP domains and future work using full length BCCPs should elucidate the role of the N-terminal interaction with BPL. It is also possible that the addition of the His6-tag to the protein has altered its kinetic properties and may contribute to the abnormally high Km for BCCPΔ67. The calculated kcat/Km for biotin of 1.7 ± 0.1 × 104 m−1·s−1 is 300-, 100- and 35-fold smaller than the E. coli BirA, yeast and A. thaliana BPL enzymes, respectively [28,30,40] but reflects the fact that the A. aeolicus BPL kcat is low at 70 °C (cf. A. aeolicus grows optimally at 95 °C).

Limited proteolysis with trypsin produced two fragments of ≈ 20 kDa, differing in length by only one residue (Fig. 4). Mass spectrometry revealed that cleavage had occurred after residues R44 and K45 which, by comparison with E. coli BirA, are predicted to lie near the putative intermediate binding site (Fig. 1). Treatment of BPL with trypsin and chymotrypsin in the presence of biotin or MgATP decreased the susceptibility to cleavage by a small but noticeable amount. However, incubation of the enzyme in the presence of both substrates rendered A. aeolicus BPL protease-resistant. The same region is protease-sensitive in S. cerevisiae BPL and is also protected by incubation with both biotin and MgATP [28]. The E. coli BirA structure contains five surface loops, four of which are in the central domain with loop regions (110–128, 212–233) and (140–146, 193–199) close together in three-dimensional space [6]. The region containing 110–128 in E. coli BirA is highly analogous to residues 32–50 in A. aeolicus BPL whereas the other loop regions have low pairwise sequence homology. A protease-sensitive site has been reported between residues 217 and 218 of BirA. In contrast, A. aeolicus BPL is not cleaved at this site but is cleaved in the adjacent loop region (32–50). This suggests that this highly conserved region forms an exposed loop near the biotinyl-5′-AMP binding site (Fig. 1). These flexible, unstructured regions are also involved in BCCP binding and are believed to become more rigid upon substrate-binding [6,34].

A recent combined mutagenesis/biological selection approach identified two single glutamate residues E119 and E147 of E. coli BCCP-87 that appear to interact with BPL [22]. A BCCP-87 E119K mutant is inactive as a substrate for BirA, whereas the E147K protein could be biotinylated, albeit poorly. It is presumed that these acidic BCCP-87 residues interact with basic BirA counterparts and mutation of BirA residues K277 and R317 were found to effect biotinylation and ATP-binding, respectively. This surprising result suggested that the C-terminal domain of BirA, which had been ascribed no biochemical function, also plays a significant role in apo-BCCP and substrate recognition [29].

It has been shown that ion pair networks are a common feature in heat-resistant proteins and are believed to play important roles in their increased thermal stability [17,41]. As both the A. aeolicus BPL and BCCP contain a large number of charged residues and we observed inhibition of biotinylation at high salt concentrations, we presume that ionic interactions are involved in the formation of the hyperthermophilic BPL:BCCPΔ67 complex. To investigate the formation of the BPL:BCCPΔ67 heterodimer we used the chemical cross-linking agent EDC to capture a BPL:BCCP complex for the first time. The zero-length EDC reagent activates acidic residues on one protein to form an unstable urea derivative [42]. This derivative then reacts with a nucleophile (such as lysine) on another protein to form an amide link between the two proteins. Incubation of BPL and apo-BCCPΔ67 in the presence of EDC led to the time-dependent appearance of a species of ≈ 37 kDa on SDS/PAGE gels (Fig. 5A), which is in agreement with the predicted mass of a 1 : 1 complex of BPL and apo-BCCPΔ67. We noticed that BPL, BCCPΔ67 and the complex eluted earlier than predicted from the size-exclusion column. Future studies will analyse the proteins by equilibrium sedimentation experiments in a similar way to that described for the BCCP-87 and BCCP [25]. Nevertheless, the complex was easily separated from the unreacted proteins using this procedure (Fig. 6) and allowed us to confirm its mass by electrospray mass spectrometry. Interestingly, the complex was not formed between BPL and holo-BCCPΔ67 (Fig. 5B) suggesting that biotinylation had either caused a conformational change in BCCPΔ67 such that it no longer bound to BPL or that the biotin moiety had somehow blocked residues that react with the EDC reagent. Furthermore, a complex was formed between the BCCPΔ67 K117L mutant and BPL both in the absence and presence of saturating amounts of biotin and MgATP (Fig. 5C). This demonstrates that the active lysine residue does not take part in the cross-linking reaction and saturating amounts of both substrates do not inhibit complex formation.

Although the published 3D structures of the apo- and holo- forms of BCCP-87 show no major structural differences, some structural studies (both NMR and X-ray) have concluded that the lack of any major differences between them might not be wholly reflected in their behaviour in solution [15]. NMR titration experiments were carried out with BirA and apo-BCCP-87 and, in light of our data, it would be interesting to repeat this work with BirA and holo-BCCP-87 to determine if any differences arise. Recent elegant studies by Cronan and Solbiati et al. highlight a difference in the stability of apo-BCCP-87 and holo-BCCP-87 to proteolysis and stress the importance of the essential so-called ‘thumb’ domain of BCCP-87 (residues 91–100) which had previously been shown to interact with the ureido ring of the attached biotin moiety [37,43]. Studies using chemically biotinylated BCCP-87 recently confirmed that this increased stability is an inherent property of holo-BCCP-87 and not due to a conformational change imparted by BPL. Furthermore, thumbless holo-BCCP-87 mutants exhibit little increased stability over their apo- counterparts, implying the majority of this increased stability is due to the thumb–biotin interaction. The authors conclude that the more protease sensitive apo- BCCP has a more dynamic form than the holo- protein. The A. aeolicus BCCPΔ67 also contains a well-conserved thumb domain (Fig. 1) and we are currently producing thumbless BCCPΔ67 mutants for analysis by EDC cross-linking with BPL (D. Clarke and D. Campopiano, unpublished results).

A recent study suggested that the C-terminal domain of BirA is essential for the catalytic activity of the enzyme and plays a role in ATP and BCCP binding [29]. Also, a model of the E. coli BirA:holo-BCCP-87 complex has been suggested based upon structural studies, sequence analysis, mutagenesis and limited proteolysis experiments [36]. The model (PDB code 1K67) was built using the coordinates of the BirA dimer in the presence of biotin (PDB code 1HXD) and holo-BCCP-87 (PDB 1BIA). Residues in both E. coli proteins thought to be responsible for BirA:BCCP-87 complex formation are conserved in A. aeolicus BPL and BCCPΔ67 (Fig. 1). A current goal is to identify the charged residues taking part in the EDC-mediated crosslinking reaction and A. aeolicus BPL and BCCPΔ67 mutants are currently being studied using high-temperature in vitro biotinylation and chemical crosslinking assays.

Acknowledgements

We wish to thank Profs. K. Stetter and R. Huber (University of Regensburg) for the gift of A. aeolicus chromosomal DNA. The Nuffield Foundation Bursary Scheme is acknowledged for its support of (J. C.). This work was supported by the Biotechnology and Biological Sciences Research Council, UK, and the University of Edinburgh.

Ancillary