THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2002 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 277, No. 25, Issue of June 21, pp. 22520 –22527, 2002 Printed in U.S.A. Interchangeable Enzyme Modules FUNCTIONAL REPLACEMENT OF THE ESSENTIAL LINKER OF THE BIOTINYLATED SUBUNIT OF ACETYLCoA CARBOXYLASE WITH A LINKER FROM THE LIPOYLATED SUBUNIT OF PYRUVATE DEHYDROGENASE* Received for publication, February 6, 2002, and in revised form, March 29, 2002 Published, JBC Papers in Press, April 15, 2002, DOI 10.1074/jbc.M201249200 John E. Cronan, Jr.‡ From the Departments of Microbiology and Biochemistry, University of Illinois, Urbana, Illinois 61801 * This work was supported in part by National Institutes of Health Grant AI15650. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ To whom correspondence may be addressed: Dept. of Microbiology, University of Illinois, B103 Chemical and Life Sciences Laboratory, 601 S. Goodwin Ave., Urbana, IL 61801. Tel.: 217-333-7919; Fax: 217-2446697; E-mail: j-cronan@life.uiuc.edu. Biotin and lipoic acid are vitamins that play essential roles in central metabolism. The biological activities of both coenzymes are dependent upon their covalent attachment to their cognate proteins (1). In both cases, the site of coenzyme attachment is the ⑀-amino group of a lysine residue centrally located within conserved sequences of ⬃70 amino acid residues that fold to form discrete protein domains. The three-dimensional structures of both biotinoyl (2– 6) and lipoyl domains (7–9) from several biological sources have been determined, and the structures are largely superimposable. Indeed, Reche and Perham (10) have succeeded in altering the Escherichia coli acetyl-CoA carboxylase (ACC)1 biotinoyl domain such that the normally absolutely specific lipoate protein ligase will attach lipoic acid to various mutant biotinoyl domains. Therefore, it seems clear that two enzyme families that catalyze very different reactions use a common structural domain to carry the essential coenzyme. Biotinylated enzymes catalyze carboxylation and decarboxylation reactions and play essential roles in fatty acid synthesis and amino acid degradation, whereas lipoylated enzymes catalyze acyl transfer and single carbon transfer reactions and are required for function of the citric acid and glycine cleavage cycles (1). Both the biotinoyl and lipoyl domains are thought to act as swinging arms that convey covalently bound intermediates between different active sites of a multienzyme complex (1). Swinging arm mobility has two different aspects. Mobility on a small scale is imparted by attachment of the carboxyl of the coenzyme to the ⑀-amino group of a lysine residue located at the tip of a protruding ␤-turn. As first proposed (see Ref. 11) this arrangement gives the “business ends” of the coenzymes a significant reach. Mobility on a much larger scale is imparted by the proline plus alanine-rich sequences adjacent to the lipoyl domains that act as flexible linkers (Fig. 1) (1). In lipoyl enzymes having only a single domain, the domain forms the amino terminus of the protein, and the linker connects the domain to the catalytic domain of the protein. In proteins having multiple lipoyl domains, such as E. coli pyruvate dehydrogenase (PDH), which has three lipoyl domains, the domains are arranged in tandem at the amino terminus with linkers separating the domains from one another and from the catalytic domain (Fig. 1B). Biotinylated proteins are the mirror image of the single lipoyl domain proteins; the modified domain is located at the carboxyl terminus, and a proline plus alanine-rich sequence is found upstream of the domain (Fig. 1A). Acetyl-CoA carboxylase (ACC) catalyzes the first step in fatty acid synthesis, the synthesis of malonyl-CoA from acetylCoA (12) (Fig. 2). ACC is a biotin-dependent enzyme, and like 1 The abbreviations used are: ACC, acetyl-CoA carboxylase; BCCP, biotin carboxyl carrier protein; IPTG, isopropyl-␤-D-thiogalactopyranoside; X-gal, 5-bromo-4-chloro-3-indoyl-␤-D-galactoside; PDH, pyruvate dehydrogenase. 22520 This paper is available on line at http://www.jbc.org Downloaded from http://www.jbc.org/ by guest on April 24, 2020 Biotin carboxyl carrier protein (BCCP) is the small biotinylated subunit of Escherichia coli acetyl-CoA carboxylase, the enzyme that catalyzes the first committed step of fatty acid synthesis. E. coli BCCP is a member of a large family of protein domains modified by covalent attachment of biotin. In most biotinylated proteins, the biotin moiety is attached to a lysine residue located about 35 residues from the carboxyl terminus of the protein, which lies in the center of a strongly conserved sequence that forms a tightly folded anti-parallel ␤-barrel structure. Located upstream of the conserved biotinoyl domain sequence are proline/alanine-rich sequences of varying lengths, which have been proposed to act as flexible linkers. In E. coli BCCP, this putative linker extends for about 42 residues with over half of the residues being proline or alanine. I report that deletion of the 30 linker residues located adjacent to the biotinoyl domain resulted in a BCCP species that was defective in function in vivo, although it was efficiently biotinylated. Expression of this BCCP species failed to restore normal growth and fatty acid synthesis to a temperature-sensitive E. coli strain that lacks BCCP when grown at nonpermissive temperatures. In contrast, replacement of the deleted BCCP linker with a linker derived from E. coli pyruvate dehydrogenase gave a chimeric BCCP species that had normal in vivo function. Expression of BCCPs having deletions of various segments of the linker region of the chimeric protein showed that some deletions of up to 24 residues had significant or full biological activity, whereas others had very weak or no activity. The inactive deletion proteins all lacked an APAAAAA sequence located adjacent to the tightly folded biotinyl domain, whereas deletions that removed only upstream linker sequences remained active. Deletions within the linker of the wild type BCCP protein also showed that the residues adjacent to the tightly folded domain play an essential role in protein function, although in this case some proteins with deletions within this region retained activity. Retention of activity was due to fusion of the domain to upstream sequences. These data provide new evidence for the functional and structural similarities of biotinylated and lipoylated proteins and strongly support a common evolutionary origin of these enzyme subunits. 22521 E. coli BCCP FIG. 1. Diagrams of the E. coli BCCP and PDH E2 proteins and their linker sequences. A, BCCP with the domain shown as an oval, the linker as the wavy line, and the biotin as ␸. A putative interaction domain at the N terminus is depicted by a rectangle. B, the PDH E2 subunit using the same depictions except that coenzyme is depicted by ␥. C, the sequences of the PDH E2 linkers; D, the residue substitutions made in replacing the BCCP linker with that from PDH-1. are essentially the same as those of synthetic linker peptides (17, 23). In this paper, I report that a large deletion of the putative linker region of the E. coli BCCP subunit results in a protein of extremely compromised ACC activity in vivo. However, biological activity was restored upon insertion of a sequence derived from the first linker of the E. coli PDH E2 subunit, indicating that the lipoyl and biotinoyl linker regions have overlapping functions. Deletion and insertion analyses of this chimeric protein and of the native BCCP showed that only a restricted portion of the linker was required for biological activity and that many sequences fail to provide linker function. EXPERIMENTAL PROCEDURES Construction of an accB Gene Encoding a PDH Linker—The synthetic 133-bp linker was constructed in three stages due to its length. First, a cassette encoding the N-terminal end of the linker was produced by annealing the complementary oligonucleotides 1 and 2 (Table I) and ligating this to pMTL22 (26) digested with the HindIII and NgoMIV (the cassette was designed such that complementary ends would result upon annealing). Likewise, a cassette encoding the C-terminal end of the linker was produced by annealing oligonucleotides 3 and 4 and was ligated to pK19 (27) digested with NgoMIV plus SmaI. Ligation to the SmaI end resulted in construction of a BsiWI site. The plasmids encoding the N-terminal and C-terminal halves of the linkers were then digested with NgoMIV and HindIII or BsiWI, respectively; the linkerencoding fragments were purified, mixed in equimolar ratios, and then ligated together with the accB plasmid pCY326 (28) digested with HindIII and BsiWI in a three partner ligation. The resulting plasmids were sequenced and found to contain a 179-bp vector-derived sequence inserted between the two linker-encoding fragments. This vector fragment insert (apparently due to a mutation at the NgoMIV end of the C-terminal cassette) was removed by digesting the plasmid with NgoMIV and BglII and inserting a cassette of annealed oligonucleotides 5 and 6 to give the desired linker sequence. Two other plasmids were also derived by cassette mutagenesis. A cassette encoding the sequence PAAAA was obtained by annealing oligonucleotides 7 and 8 and inserting this cassette into BglI-digested pCY326 (the cassette was designed with complementary ends and included a new BstZI site and retained the BglI site at one end of the inserted sequence). Finally, a cassette designed to delete the EAPAAA sequence adjacent to the biotinoyl domain of the wild type gene was obtained by annealing oligonucleotides 9 and 10 and ligating this product to pCY326 digested with NcoI and BsiWI. This cassette introduced a BglII site and inactivated the BsiWI site. A series of deletions within the region encoding the chimeric linker were made by restriction digestion followed by blunt end ligation done at low DNA concentrations to favor recircularization. In most cases, one or both of the ends resulting from digestion were either filled in or resected by treatment with either phage T4 DNA polymerase plus the Downloaded from http://www.jbc.org/ by guest on April 24, 2020 all biotin enzymes, the cofactor must be covalently attached to ACC for enzyme activity. In E. coli and many other bacteria, the functional enzyme consists of two copies of each of four different subunits (13–15). The biotin cofactor is attached to the biotin carboxyl carrier protein (BCCP) subunit, the sole biotinylated protein of this organism (Fig. 2). In the first partial reaction, the heterocyclic ring of the attached biotin moiety is carboxylated by the biotin carboxylase subunit. The carboxyl group is then transferred from biotin to acetyl-CoA to produce malonyl-CoA by carboxyltransferase, a complex of two proteins called ␣ and ␤. Similar ACCs (referred to as heteromeric ACCs) are found in the chloroplasts of many plants (16). The structure of the biotinoyl domains of the E. coli acetyl-CoA subunit, BCCP (also called AccB) (2, 4 – 6) and a related protein, the Propionibacterium shermanii 1.3 S transcarboxylase subunit (3), have been determined. In both cases, the structures of the protein segments located upstream of the biotinoyl domain could not be determined due to their high degree of mobility. No structures could be detected in these protein segments by either x-ray diffraction or multidimensional heteronuclear NMR spectroscopy. Given the similarities between the biotinoyl and lipoyl domains, it seemed plausible that the putative biotinoyl domain linker regions might play roles similar to those of the lipoyl domain linkers. However, an argument against this notion is that the sequences of biotinoyl and lipoyl linkers cannot be aligned for more than a few residues (Fig. 1). Moreover, in the linkers of some biotinoyl proteins, other small residues (generally serine or glycine) have replaced alanine, and the linkers of lipoyl proteins are richer in charged residues than are the biotinoyl linkers. On the other hand, given the conserved domain structures, it seemed unlikely that the conserved proline/ alanine-rich nature of the neighboring amino acid sequences could be accidental. The E. coli PDH linkers (Fig. 2A) have been extensively studied, and the available data indicate that the combination of proline and alanine results in an extended, but flexible, structure (1, 17–25). In the E. coli PDH complex, the linker regions are known to be highly mobile and exposed to solvent. The mobility of these linker regions was first detected by NMR studies of the intact complex (20, 21). Despite the very high molecular weight of the E. coli PDH complex (5–10 ⫻ 106), a set of sharp resonances are observed that disappear with loss of the lipoyl domains (from proteolytic cleavages within the linkers). Moreover, the spectra observed in the intact complex FIG. 2. The reaction mechanism of E. coli acetyl-CoA carboxylase. BCCP is encoded by the accB gene, whereas biotin carboxylase is encoded by the accC gene (15). The two subunits involved in carboxyltransferase activity are encoded by the accA and accD genes (14). The covalently bound biotin of BCCP carries the carboxylate moiety. 22522 E. coli BCCP TABLE I Oligonucleotides used in this work Oligonucleotide Sequence 1 2 3 4 5 6 7 8 9 10 11 12 5⬘-AGCTTACGACGGCGCAGCAGACGCTGCACCTGCGCAGGCAGAAGAGAAGAAAGAAGCAGCG 5⬘-CCGGCGCTGCTTCTTTCTTCTCTTCTGCCTGCGCAGGTGCAGCGTCTGCTGCGCCGTCGTA 5⬘-CCGGCTGCAGCACCAGCGGCTGCGGCGGCGGAGATCTCCGGTCACATCGTAC 5⬘-GTACGATGTGACCGGAGATCTCCGCCGCCGCAGCCGCTGGTGCTGCAGCCGG 5⬘-CCGGCTGCAGCACCAGCGGCTGCGGCGGCGGA 5⬘-GATCTCCGCCGCCGCAGCCGCTGGTGCTGCAG 5⬘-CGGCCGCTGCAGCTC 5⬘-CTGCAGCGGCCGGAG 5⬘-GAGATCTCCGGTCACATTGTAC 5⬘-AATGTGACCGGAGATCTCCATG 5⬘-CATGGAAGCGCCGGCAGCAGCGGAAATCAGTGGTCACATCGTAG 5⬘-GCGCCTACGATGTGACCACTGATTTCCGCTGCTGCCGGCGCTTC plasmid was digested with NcoI (which cuts within accB) and KasI (which cuts within the 1.3 S coding sequence), and a cassette made by annealing the complementary oligonucleotides 11 and 12 was then inserted by ligation. Screening of Deleted Plasmids—The products of most of the deletion/ insertion ligations were transformed into strain CY1478, which carries a chromosomal bioBFC-lacZ fusion and overproduces lacI repressor from an F⬘ episome. Strain CY1478 was constructed by conjugational mating of strain BM2661 (31) with NovaBlue (Novagen) with selection for recombinants resistant to streptomycin and tetracycline to introduce the lacIQ episome. The transformants were tested on RB medium containing 80 nM biotin and X-gal (40 ␮g/ml), ampicillin, and tetracycline. Colonies were patched onto two plates of this medium, one supplemented with glucose (0.4% final concentration) and the other lacking glucose. The plates were then incubated at 37 °C overnight. Colonies that were very dark blue on the plate lacking glucose but pale blue on the glucose-containing plates were in frame deletions, whereas colonies that were white on both plates were out of frame fusions (the rationale is explained under “Results”). The lacIQ episome was introduced to decrease expression from the vector lac promoter, which is responsible for BCCP expression in pCY326 and its derivatives. Glucose addition was used to further decrease BCCP expression by decreasing cAMP-dependent expression from the lac promoter. Low level BCCP expression was desired, because high level expression of this protein is toxic to E. coli and also to allow colony scoring by use of X-gal. All constructs were confirmed by DNA sequencing done by the Keck Genomics Center of the University of Illinois. Treatment with T4 DNA polymerase (New England Biolabs) was done in the restriction enzyme digestion buffer supplemented with the four deoxynucleotide triphosphates each at 0.25 mM. One unit of polymerase was added, and the reaction was incubated at 16 °C for 15–30 min followed by phenol treatment to inactivate the polymerase. Treatment with mung bean nuclease (New England Biolabs) was done in the restriction enzyme digestion buffer supplemented with 1 mM ZnSO4. Ten units of nuclease were added, followed by incubation at 30 °C for 1 h and then phenol extraction to inactive the nuclease. Protein Expression—Each of the altered accB genes was cloned together with the downstream kanamycin resistance gene into the regulated expression vector pCY465 (28) as previously described. Other Methods—Measurement of protein biotinylation by labeling growing cultures with [8,9-3H]biotin, gel electrophoresis, fluorography, and bacterial media were as described previously. A sample of BCCP-87 mixed with the full-length BCCP was made by IPTG induction of strain TM126, a wild type strain carrying pTM52 (32) in [3H]biotin-containing medium. Plasmid pTM52 was constructed by insertion of the BCCP-87 encoding NcoI-HindIII segment of pLS4 (15) into expression vector pKK233-2 (33) cut with the same enzymes. RESULTS Essential Nature of the BCCP Linker—The determined structures of the BCCP biotin domain begin at residues Gly77 and Ile78, although these residues and the next few residues (up to Ile82) are considerably more mobile than the residues of the tightly folded domain (2, 4 – 6). In preliminary work, the accB gene was modified to introduce a BglII site that overlapped codons 77 and 78. The DNA segment lying between this new site and the naturally occurring HindIII site was then Downloaded from http://www.jbc.org/ by guest on April 24, 2020 four nucleoside triphosphates or by treatment with mung bean nuclease in order to give the proper reading frame. The 3⬘ to 5⬘ exonuclease activity of the polymerase was used to remove 3⬘-extensions, whereas 5⬘-extensions were removed by mung bean nuclease. For brevity, the abbreviations T4 and MB following the restriction enzyme used to generate the extension will be used to signify these manipulations (the conditions used are given below). The constructions were as follows: ⌬1, HindIII(T4) and PstI(T4); ⌬2, FspI and NaeI; ⌬4, FspI and BglII(T4); ⌬5, BstAPI(T4); ⌬6, HindIII(T4) and NgoMIV(T4); ⌬8, BsgI(T4) and PstI(T4); ⌬9, NgoMIV(MB) and BglII(MB); ⌬10, NgoMIV(MB) and PstI(T4); ⌬11, HindIII(MB) and FspI; ⌬12, HindIII(MB) and NaeI; ⌬13, HindIII(T4) and BglII(T4). The ⌬15 plasmid resulted from the same manipulations as ⌬10 but suffered a 27-bp deletion presumably via illicit recombination in vivo. A series of manipulations of the wild type BCCP linker region were also made. The construct encoding the ⌬14 BCCP resulted from digestion of pCY326 with NcoI followed by MB treatment. The MB was inactivated, and the DNA was digested with HindIII, treated with T4, and ligated. The construct encoding the ⌬17 BCCP resulted from annealing the complementary oligonucleotides 9 and 10 and ligating the cassette to pCY326 digested with NcoI plus BsiWI (the annealed cassette was designed such that complementary ends would result). The ⌬18 and ⌬19 BCCP constructs were made by insertion of antibiotic cassettes into the introduced BglII site of the ⌬17 construct. The BamHI tetracycline resistance-encoding fragment of p34s-Tet (29) was ligated to the BglII-cut ⌬17 plasmid followed by selection for resistance to both tetracycline and kanamycin. The tetracycline resistance determinant was then eliminated by SstI digestion and ligation. The construction of ⌬19 followed a similar route except that the SstI tetracycline resistance fragment p34s-Tet was flanked by two copies of the BglII-SstI fragment of the multiple cloning site of pMTL25 (26). The tetracycline determinant was then eliminated by MluI digestion followed by ligation to give the construct encoding ⌬19. Construction of the genes encoding the ⌬20, ⌬21, ⌬22, and ⌬23 BCCPs used a 12-bp cassette made by annealing the complementary oligonucleotides 7 and 8 followed by treatment with MB and phage T4 polynucleotide kinase plus ATP. The blunt-ended cassette was then ligated to pCY326 digested with NcoI plus BsiWI and treated with T4. DNA sequencing showed three different products: insertion of one copy of the cassette in the two possible orientations (⌬22 and ⌬22) and two copies inserted in a head to tail configuration (⌬21). The gene encoding ⌬21 was then cut with PstI and religated to give ⌬23. The genes encoding ⌬24 and ⌬27 were derived from that encoding ⌬23 by insertion of the gentamycin resistance encoding the PstI fragment of p34s-Gm (29) into the PstI site of the ⌬23 construct. The resistance gene was then removed by digestion with XbaI digestion to give ⌬25 or SstI digestion to give ⌬27. The gene encoding ⌬26 resulted from insertion of the same oligonucleotide cassette as ⌬17 except that the sticky ends were removed by T4 treatment, and the resulting blunt end cassette was ligated to pCY326 digested with NcoI plus BsiWI and then treated with T4. The resulting recombinant plasmids were then screened for products with the cassette inserted in the orientation opposite that of ⌬17. The gene encoding the I28 insertion was constructed by annealing the complementary oligonucleotides 7 and 8 and ligating the resulting cassette to pCY326 digested with BglI. The gene encoding the BCCP-P. shermanii 1.3 S chimeric protein was constructed from pCY325 (28) and p1.3t (30) by ligating the SalI-BglII fragment of p1.3t into pMTL21 digested with SalI plus BamHI. The resulting plasmid was digested with EcoRI and BamHI and ligated to the EcoRI-BglII accB fragment of pCY325. This E. coli BCCP FIG. 3. Growth of strain CY1336 carrying plasmids encoding wild type and mutant BCCPs. Early stationary phase cultures (grown at 30 °C in the absence of IPTG except as noted) of strain CY1336 (the temperature-sensitive G133S BCCP mutant) carrying plasmids that encoded various mutant BCCPs were serially diluted in 10-fold steps, and 0.01-ml samples were spotted onto plates of RB medium containing 75 ␮M IPTG plus the antibiotics necessary for plasmid maintenance, and the plates were incubated at 42 °C for 48 h (A) or 24 h (B). The leftmost spot was the undiluted culture, the next spot to the right was a 10-fold dilution, and successive 10-fold dilutions proceeded to the right. The greater growth of ⌬13 in A is due to pregrowth at 30 °C in the presence of 75 ␮M IPTG. The DASMEP protein is defective in biotinylation (10) and thus fails to complement the accB temperature-sensitive mutant strain (28). dehydrogenase (Fig. 1). This linker was chosen over the other two E. coli PDH E2p linkers, since its length was exactly that of the proline/alanine-rich region of BCCP, and thus no artificial truncation or extension was required to fit this linker into BCCP. The restriction sites were included to simplify manipulations of the amino acid sequence. If insertion of the new linker allowed biological function of the chimeric BCCP, then construction of deletions that resulted in loss of function could define the important parts of the linker. If, on the other hand, the chimeric BCCP was nonfunctional, substitutions with segments of the BCCP sequence could be made in attempts to restore function. The striking result was that the chimeric BCCP was fully functional at both 37 and 42 °C in vivo, indicating that the proline/alanine-rich regions of BCCP and PDH linker 2 functioned in a similar manner. Given this result and the prior studies demonstrating mobility and flexibility of the PDH linkers both in situ and as isolated peptides, it can be confidently predicted that upon structural analysis, the proline/alaninerich region of BCCP shall be found to be a mobile linker. Essential Segments of the Chimeric BCCP Linker Region—A series of deletions were made within and upstream of the linker region of the chimeric protein by use of the introduced restriction sites. These deletions were all made upstream of Gly80, the first highly structured residue of the conserved domain (2, 4 – 6), since prior work had shown that deletion of residues at the amino or carboxyl ends of biotinoyl domain sequences results in the loss of biotinylation due to loss of domain structure (37, 38). Although BCCP residues 77–79 have defined secondary structures, these residues were considered candidates for deletion or substitution, since the residues are extended from (and do not interact with) the main body of the domain protein and have been considered as part of the linker region (4). These residues are highly mobile in the NMR analyses and have high crystallographic B values and thus seemed reasonable candidates for deletion or substitution. Moreover, these residues are not conserved in the BCCPs of Pseudomonas aeruginosa (39) and Bacillus subtilis (40), both of which have been shown to Downloaded from http://www.jbc.org/ by guest on April 24, 2020 deleted by restriction digestion followed by treatment with phage T4 DNA polymerase plus the four deoxynucleotide triphosphates and phage T4 DNA ligase. These manipulations removed most of the putative linker region. It should be noted that this construct would not be expected to produce an active BCCP species, because the translation frame would be out-offrame relative to the accB coding sequence. However, such in vitro filling and ligation reactions have a significant error rate such that rare in-frame constructs were expected (34). In-frame constructs were detected by use of the biotin operon regulatory system (35–37). Overproduction of a biotin acceptor protein results in increased expression of the biotin biosynthetic operon due to consumption of biotinyl-AMP by protein biotinylation (37). This increase in transcription is readily detected by use of a host strain (strain CY1478) carrying a promoter fusion construct in which a biotin operon promoter drives expression of E. coli ␤-galactosidase (31). The increased biotin operon expression is detected by increased cleavage of a chromogenic ␤-galactosidase substrate. In the present case, correctly processed constructs would be out-of-frame, and the lack of a biotin domain expression would result in white colonies on medium containing X-gal, whereas the rare in-frame constructs that accepted biotin (and thereby result in increased biotin operon expression) would give blue colonies. This screen was used in the construction of the deletions reported below, and all blue colonies were shown by DNA sequencing to have in-frame sequences (a few pale blue colonies were found to encode out-of-frame constructs). In the case of the HindIIIBglII deletion, DNA sequencing showed an in-frame construct (called ⌬13) that removed 30 residues of the putative BCCP linker and resulted from incomplete filling of the BglII site. The function of the ⌬13 construct and the other constructs in this paper were tested in vivo as described previously (28). In brief, the altered accB genes were inserted into an expression plasmid that expressed the altered BCCP at a level comparable with that from the wild type chromosomal accB gene. The expression constructs where then transformed into E. coli accB mutant strain, CY1336, which encodes a temperature-sensitive BCCP (G133S BCCP) that is rapidly degraded upon shift to 37 °C (or higher temperatures). In the absence of expression of a functional plasmid-encoded BCCP gene, strain CY1336 fails to grow, whereas expression of a functional protein permits growth (hence giving genetic complementation of the host accB mutation). At 37 °C, growth was found to require about 8% of the normal level of BCCP (28). Strain CY1336 expressing the wild type gene grows well at 37 °C and appreciably more slowly at 42 °C. When the ⌬13 BCCP was expressed in strain CY1336, the strain grew very poorly at 37 °C and failed to grow at 42 °C (Fig. 3, A and B). As expected from prior work (28), the barely detectable growth of the strain observed at 37 °C required induction of gene expression with IPTG. Therefore, deletion of the BCCP linker residues resulted in a protein that was virtually without activity in vivo. Functional Replacement of the BCCP Linker with Linker 1 of the E. coli PDH E2p Subunit—Two approaches were considered to dissect the role of the linker region in BCCP function. The first approach was a conventional deletion analysis, but this was inconvenient due to a scarcity of restriction sites in the sequence encoding the linker. A second approach was to substitute a known linker sequence for that of BCCP and thereby obtain a well studied linker. This second approach was taken and the coding sequence of the chosen linker was redesigned to include useful restriction sites. In this construct, 28 residues of the putative linker of BCCP were replaced with the sequence that links the outermost two lipoyl domains of E. coli pyruvate 22523 22524 E. coli BCCP fully replace the function of E. coli BCCP in vivo (the latter protein also has a one-residue deletion within this segment). Finally, none of the BCCP residues upstream of residue 81 show any structure-based alignment with the other biotinyl domain of known structure, the P. shermanii 1.3 S trancarboxylase subunit (28), indicating that these residues do not play a role in biotin domain structure. Upon construction and testing in vivo function of a number of constructs (Figs. 3 and 4 and Table II), BCCP function was found to tolerate rather large deletions, whereas some small deletions gave nonfunctional or poorly functional BCCPs. For example, expression of the ⌬1 and ⌬6 BCCPs, containing deletions of 23 and 19 residues, respectively, resulted in normal growth of CY1336 at 37 and 42 °C, whereas ⌬15, a smaller deletion of only nine residues, completely failed to support growth. The deletions (⌬4, ⌬13, and ⌬15) that were the most defective in supporting growth of CY1336 lacked the sequence of four consecutive alanine residues located just upstream of the biotinoyl domain. This region seemed very sensitive to sequence alterations, since deletions that fused the biotinoyl domain to an AAP sequence (⌬15), an AAPA sequence (⌬4), or a single alanine (⌬13) were all inactive. In contrast, deletions of upstream proline/alanine sequences as in the ⌬1, ⌬2, ⌬5, ⌬6, ⌬11, and ⌬12 BCCPs had little or no effect on growth. The only exception to this picture was ⌬8, a deletion of 29 residues. Although in ⌬8 the sequences just upstream of the biotinoyl domain were left intact, expression of the mutant BCCP resulted in only slow growth of CY1336 at 37 °C. However, interpretation of this result is not straightforward, since expression of the ⌬8 BCCP also allowed growth at 42 °C (all other constructs that grew poorly at 37 °C failed to grow at the higher temperature). Note that an observed lack of activity in the biological test system could be due to degradation of the altered proteins such as endoproteolytic cleavage of the altered linker sequences. To test this possibility, each of the constructs was expressed in strain CY1336 at a nonpermissive temperature Downloaded from http://www.jbc.org/ by guest on April 24, 2020 FIG. 4. Analyses of protein biotinylation in strains expressing wild type and mutant BCCPs. All analyses were of cultures grown at 30 °C in the presence of IPTG and then shifted to 38 °C for 70 min. The panels shown are SDS-polyacrylamide gel analyses of [3H]biotin-labeled proteins. Cells from 0.2 ml of [3H]biotin-labeled cultures were boiled in SDS buffer and applied to SDS 16% polyacrylamide gels as described previously (28). Fluorograms of the gels are shown. The samples loaded are given above each lane (see Table II for sequences). A sample of [3H]biotin-labeled BCCP-87 (43) made by induction of strain TM126 (see “Experimental Procedures”) was included as a marker for possible degradation products. The gel was submitted to a prolonged exposure in order to show the traces of chromosomally encoded G133S BCCP that provide an internal mobility standard. (The lines in the gel are cracks resulting from the drying process.) (so the chromosomally encoded G133S protein would be degraded) and labeled with [3H]biotin, and the radioactive proteins were analyzed by SDS-gel electrophoresis. Expression of each of the deleted BCCPs at 38 °C in the presence of [3H]biotin gave a radioactive band of the expected electrophoretic mobility. Fig. 4 shows some of these data. Two of the constructs (⌬1 and ⌬2, data not shown) showed minor amounts of smaller biotinylated proteins in addition to the full-length proteins consistent with some endoproteolytic processing. However, the strains expressing these proteins grew normally, and thus the loss of protein due to processing was of no consequence. Analysis of the Native BCCP Linker Region—Deletion and insertion studies were also done on the native BCCP linker. First, based on the results obtained with the chimeric protein, ⌬14, a deletion similar to ⌬13, but retaining the proline/alanine-rich sequence adjacent to the biotinoyl domain, was constructed. Expression of the ⌬14 protein gave only modestly reduced growth, and thus restoration of only 7 of the 30 residues deleted in ⌬13 resulted in return of most of the biological activity. Of these seven residues, it seemed likely (given the above results) that the APAAA sequence would play an important role in this restoration of biological activity. Therefore, this sequence was deleted from the native linker to give the ⌬17 protein with the anticipation that this protein should show decreased biological activity. However, expression of the ⌬17 protein gave essentially wild type growth (Fig. 3). Hence, the presence or absence of APAAA sequences adjacent to the biotinoyl domain seemed to give apparently conflicting results in ⌬17 BCCP versus ⌬13 BCCP. This apparent conflict could be explained if in ⌬17 BCCP upstream proline/alanine-rich sequences could functionally substitute for the deleted residues, whereas in ⌬13 no sequences were available to act as surrogates of the deleted residues. To test whether upstream residues were responsible for function of the ⌬17 BCCP, two arbitrary sequences were inserted adjacent to the biotinoyl domain of the ⌬17 protein to act as possible spacers or insulators between the domain and the upstream proline/alanine-rich sequences. The sequences used were those encoded by symmetrical multiple cloning sites from two different cloning vectors, and these were inserted immediately upstream of the biotinoyl domain of ⌬17 to give the ⌬18 and ⌬19 BCCPs. Although expression of the ⌬18 protein allowed growth at 37 °C, growth at 42 °C was greatly decreased. Upon expression of the ⌬19 protein, growth at 37 °C was almost normal, and growth also proceeded at 42 °C, albeit at a rate about 50 – 60% of that seen upon expression of the wild type protein. Therefore, the sequence IPGYRARYPG of ⌬18 acts as an effective barrier between the biotin domain and the upstream sequences. Note that the proteins with these altered linkers were well expressed and stable in vivo (Fig. 4). Other BCCP Constructs—A number of other alterations of the wild type BCCP sequence were made (data not shown). Several encoded proteins were unstable or poorly biotinylated. However, four constructs are worthy of note. Substitution of the EAPAAAEISGHI sequence with NVTGDL (⌬26), LQRPRPLQ (⌬21), LQVDSRVDLQ (⌬25), or NVTGDPRVPSSVPGDL (⌬20) gave proteins that were stable and well biotinylated. However, upon expression, none of the three proteins supported growth of strain CY1336. Therefore, residues 77– 82 of BCCP can be substituted with very different residues without major effects on the structure of the biotinoyl domain as assayed by the sensitive assays of biotinylation and stability to proteolysis in vivo (36), and thus the lack of biological activity must be attributed to the alterations of the linker. Finally, a more ambitious chimeric protein was made in which the BCCP biotinoyl domain was replaced with the only other biotinoyl 22525 E. coli BCCP TABLE II Growth resulting from expression of deleted and inserted BCCP species Deleted residues are denoted by the empty spaces and insertions by lines (/\). The sequences above the line are the BCCP/PDH chimeric linker (designated as PDH) and its deletion variants. The sequences below the line are the BCCP linker and its deletion variants. The first and last BCCP residues shown are Pro37 and Val83, respectively. Growth is denoted by colony size at the temperature given on medium containing 75 ␮M IPTG (28), with pp signifying pinpoint colonies visible to the eye, (pp) signifying pinpoint colonies visible only with a magnifying lens, and O signifying no detectable growth. Before testing the cultures were grown at 30 °C in the absence of IPTG. Downloaded from http://www.jbc.org/ by guest on April 24, 2020 domain of known structure, that of P. shermanii transcarboxylase (3). The fusion junction between the two proteins was a structurally conserved glycine residue such that the last residue of the BCCP linker, Gly80, was replaced with Gly48 of the transcarboxylase 1.3 S subunit. Upstream of this glycine residue were the N-terminal 79 residues of BCCP, and downstream were the 75 C-terminal residues of the 1.3 S subunit biotinoyl domain. Expression of this chimeric protein failed to allow any detectable growth of mutant strain CY1336 at any nonpermissive temperature, although the chimeric protein was stable and efficiently biotinylated. Therefore, despite the conserved do- main structure, the biotinoyl domain of BCCP could not be functionally replaced with a foreign biotinoyl domain. DISCUSSION The proline/alanine-rich region of BCCP is essential for function of the protein as shown by the deletion analyses reported. The finding that a linker region from the E2 subunit of the PDH complex functionally substitutes for the natural region indicates that these protein segments play similar roles in the two enzyme complexes, although the cognate enzymes have markedly differing structures and catalytic activities. Since the 22526 E. coli BCCP Study of the native BCCP linker gave a more complex picture in that deletion of the proline/alanine sequence adjacent to the domain had no effect on protein function. This lack of effect appears due to functional replacement of these residues with upstream proline/alanine sequences because introduction of arbitrary sequences adjacent to the domain blocked BCCP function. Therefore, the PDH linker seems to contain sequences that restrict the use of upstream residues as surrogates of those adjacent to the domain, whereas the native BCCP linker lacks this property. Since mobile linker and loop regions cannot be determined by the currently available techniques, the structures of such protein segments is a major unsolved problem of structural biology. In hopes of gaining structural insight, each of the linker sequences in this paper was submitted to the PSIPRED version 2.2 protein structure prediction program (available on the World Wide Web at bioinf. cs.ucl.ac.uk/psipred/). This program is considered to be the current state of the art for structure prediction from primary sequence (about 80% accurate). (Note that PSIPRED prediction of the BCCP biotinyl domain showed a highly accurate correspondence with the known domain structure, thus engendering confidence in the program.) None of the linker residue alterations reported in this paper changed the predicted pattern of secondary structure elements in wild type BCCP. The only changes seen were in the lengths of the linker regions (predicted as coil by the program). Thus, there are currently no data predicting which sequence or sequences can account for the differing “insulating” properties of the PDH and BCCP linkers. It would have been advantageous to supplement the genetic complementation data presented in this paper with assays of ACC activity. However, the ACC of E. coli readily dissociates upon cell lysis, and hence the overall ACC activity cannot be measured in crude cell extracts (12, 13, 42). REFERENCES 1. Perham, R. N. (2000) Annu. Rev. Biochem. 69, 961–1004 2. Athappilly, F. K., and Hendrickson, W. A. (1995) Structure 3, 1407–1419 3. Reddy, D. V., Shenoy, B. C., Carey, P. R., and Sonnichsen, F. D. (2000) Biochemistry 39, 2509 –2516 4. Roberts, E. L., Shu, N., Howard, M. J., Broadhurst, R. W., Chapman-Smith, A., Wallace, J. C., Morris, T., Cronan, J. E., Jr., and Perham, R. N. (1999) Biochemistry 38, 5045–5053 5. Yao, X., Wei, D., Soden, C., Jr., Summers, M. F., and Beckett, D. (1997) Biochemistry 36, 15089 –15100 6. Yao, X., Soden, C., Jr., Summers, M. F., and Beckett, D. (1999) Protein Sci. 8, 307–317 7. Dardel, F., Davis, A. L., Laue, E. D., and Perham, R. N. (1993) J. Mol. Biol. 229, 1037–1048 8. Green, J. D., Laue, E. D., Perham, R. N., Ali, S. T., and Guest, J. R. (1995) J. Mol. Biol. 248, 328 –343 9. Ricaud, P. M., Howard, M. J., Roberts, E. L., Broadhurst, R. W., and Perham, R. N. (1996) J. Mol. Biol. 264, 179 –190 10. Reche, P., and Perham, R. N. (1999) EMBO J. 18, 2673–2682 11. Reed, L. J. (1998) Protein Sci. 7, 220 –224 12. Alberts, A. W., and Vagelos, P. R. (1972) in The Enzymes (Boyer, P. D., ed) Vol. 6, 3rd Ed., pp. 37– 82, Academic Press, Inc., New York 13. Guchhait, R. B., Polakis, S. E., Dimroth, P., Stoll, E., Moss, J., and Lane, M. D. (1974) J. Biol. Chem. 249, 6633– 6645 14. Li, S. J., and Cronan, J. E., Jr. (1992) J. Biol. Chem. 267, 16841–16847 15. Li, S. J., and Cronan, J. E., Jr. (1992) J. Biol. Chem. 267, 855– 863 16. Ohlrogge, J., Pollard, M., Bao, X., Focke, M., Girke, T., Ruuska, S., Mekhedov, S., and Benning, C. (2000) Biochem. Soc. Trans. 28, 567–573 17. Green, J. D., Perham, R. N., Ullrich, S. J., and Appella, E. (1992) J. Biol. Chem. 267, 23484 –23488 18. Miles, J. S., Guest, J. R., Radford, S. E., and Perham, R. N. (1987) Biochim. Biophys. Acta 913, 117–121 19. Miles, J. S., Guest, J. R., Radford, S. E., and Perham, R. N. (1988) J. Mol. Biol. 202, 97–106 20. Perham, R. N., Duckworth, H. W., and Roberts, G. C. (1981) Nature 292, 474 – 477 21. Radford, S. E., Laue, E. D., Perham, R. N., Miles, J. S., and Guest, J. R. (1987) Biochem. J. 247, 641– 649 22. Radford, S. E., Perham, R. N., Ullrich, S. J., and Appella, E. (1989) FEBS Lett. 250, 336 –340 23. Radford, S. E., Laue, E. D., Perham, R. N., Martin, S. R., and Appella, E. (1989) J. Biol. Chem. 264, 767–775 24. Texter, F. L., Radford, S. E., Laue, E. D., Perham, R. N., Miles, J. S., and Downloaded from http://www.jbc.org/ by guest on April 24, 2020 structures of the biotinoyl and lipoyl domains can be largely superimposed, it seems that not only the domain structures, but also the structure of the linkers that connect these domains to their cognate proteins, have been conserved. Although the PDH linker functionally replaced that of BCCP, only borderline amino acid sequence conservation is seen between the BCCP and PDH linkers (Fig. 1). The lack of sequence conservation is not surprising, since little strict sequence conservation is found among the three linkers that connect the three E. coli lipoyl domains to the remainder of the E2 subunit (Fig. 1). Since there is also only marginal sequence conservation between the structurally homologous biotinoyl and lipoyl domains (41), it clearly is the structures rather than the sequences of these protein domains that have been conserved. BCCP function is not dependent on the length of the linker region, since proteins having a deletion of 23 residues (⌬1) and/or an insertion of five residues (I28) were fully functional. Although a defined overall length was not important, deletions of some linker segments resulted in acute losses of function. Indeed, deletion within the BCCP linker had much larger effects on in vivo function than were observed in similar deletion analyses of PDH complexes containing an E2p subunit having a single lipoyl domain (18, 19). In the PDH case, 31 of the 32 linker residues had to be deleted before growth of an E. coli strain requiring function of the engineered E2p subunit was strongly affected. Smaller deletions had only modest effects on growth. The effects on cell growth were reflected in vitro assays of the overall PDH complex enzyme activity, where a maximal effect of 3-fold was seen (19). The most striking effect in these studies was obtained upon substitution of an arbitrarily designed highly charged linker for the natural sequence. This substitution abolished growth of the E. coli strain and resulted in a PDH complex of greatly reduced activity (19). More conservative substitutions such as linkers composed of virtually all alanine residues or all proline residues resulted in highly active PDH complexes that supported full cell growth, although both of these synthetic linkers seemed less flexible than the native linkers (25). Given the extremely permissive nature of permitted substitutions in the PDH E2 linker regions as well as the weak dependence on the presence of a linker, there seems little doubt that the BCCP linker region could functionally replace one or all of the PDH E2 linkers. However, this possibility has not yet been tested. In contrast to results reported for the PDH complex, when placed in the context of BCCP, small deletions of the PDH linker could result in complete loss of protein function in vivo. The deletion analyses indicate that the sequence of four consecutive alanine residues adjacent to the biotinoyl domain plays the most important role of the linker in BCCP function. Since biotinylation of the domain by E. coli biotin protein ligase is not impaired by deletion of these residues and the altered proteins are stable in vivo, the defects must lie in the function of BCCP in the ACC reaction. A 26-residue synthetic peptide having the sequence of the PDH linker that was introduced into BCCP has been studied by physical techniques (23). Although the peptide has the circular dichroism and proton NMR spectra of a random coil, it cannot be viewed as a “wet noodle,” since ⬎95% of the Ala-Pro sequences are in the all-trans configuration, whereas the value expected for a random coil is lower by 10 –15%. This skew toward the all-trans configuration suggests that the linker has a degree of order (23). From the known properties of proline peptides and alanine peptides, it is thought that the linker has a flexible and extended structure that allows mobility without collapsing upon itself. This picture also pertains to BCCP, since the PDH linker is a fully functional replacement for that of BCCP. E. coli BCCP Guest, J. R. (1988) Biochemistry 27, 289 –296 25. Turner, S. L., Russell, G. C., Williamson, M. P., and Guest, J. R. (1993) Protein Eng. 6, 101–108 26. Chambers, S. P., Prior, S. E., Barstow, D. A., and Minton, N. P. (1988) Gene (Amst.) 68, 139 –149 27. Pridmore, R. D. (1987) Gene (Amst.) 56, 309 –312 28. Cronan, J. E., Jr. (2001) J. Biol. Chem. 276, 37355–37364 29. Dennis, J. J., and Zylstra, G. J. (1998) Appl. Environ. Microbiol. 64, 2710 –2715 30. Murtif, V. L., and Samols, D. (1987) J. Biol. Chem. 262, 11813–11816 31. Barker, D. F., and Campbell, A. M. (1980) J. Bacteriol. 143, 789 – 800 32. Morris, T. W. (1994) Lipoate-Protein and Biotin-Protein Ligases of Escherichia Coli, Ph.D. thesis, University of Illinois, Urbana, IL 33. Amann, E., and Brosius, J. (1985) Gene (Amst.) 40, 183–190 34. Hasan, N., Kur, J., and Szybalski, W. (1989) Gene (Amst.) 82, 305–311 22527 35. Chapman-Smith, A., and Cronan, J. E., Jr. (1999) Trends Biochem. Sci 24, 359 –363 36. Chapman-Smith, A., Morris, T. W., Wallace, J. C., and Cronan, J. E., Jr. (1999) J. Biol. Chem. 274, 1449 –1457 37. Cronan, J. E., Jr. (1988) J. Biol. Chem. 263, 10332–10336 38. Cronan, J. E., Jr., and Reed, K. E. (2000) Methods Enzymol. 326, 440 – 458 39. Best, E. A., and Knauf, V. C. (1993) J. Bacteriol. 175, 6881– 6889 40. Marini, P., Li, S. J., Gardiol, D., Cronan, J. E., Jr., and de Mendoza, D. (1995) J. Bacteriol. 177, 7003–7006 41. Brocklehurst, S. M., and Perham, R. N. (1993) Protein Sci. 2, 626 – 639 42. Davis, M. S., J. Solbiati, and J. E. Cronan, Jr. (2000) J. Biol. Chem. 275, 28593–28598 43. Chapman-Smith, A., Turner, D. L., Cronan, J. E., Jr., Morris, T. W., and Wallace, J. C. (1994) Biochem. J. 302, 881– 887 Downloaded from http://www.jbc.org/ by guest on April 24, 2020 Interchangeable Enzyme Modules: FUNCTIONAL REPLACEMENT OF THE ESSENTIAL LINKER OF THE BIOTINYLATED SUBUNIT OF ACETYL-CoA CARBOXYLASE WITH A LINKER FROM THE LIPOYLATED SUBUNIT OF PYRUVATE DEHYDROGENASE John E. Cronan, Jr. J. Biol. Chem. 2002, 277:22520-22527. doi: 10.1074/jbc.M201249200 originally published online April 15, 2002 Access the most updated version of this article at doi: 10.1074/jbc.M201249200 Alerts: • When this article is cited • When a correction for this article is posted This article cites 42 references, 17 of which can be accessed free at http://www.jbc.org/content/277/25/22520.full.html#ref-list-1 Downloaded from http://www.jbc.org/ by guest on April 24, 2020 Click here to choose from all of JBC's e-mail alerts