PEDS Advance Access published online on June 21, 2007
Protein Engineering Design and Selection, doi:10.1093/protein/gzm018
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Probing the structural plasticity of an archaeal primordial cobaltochelatase CbiXS
Department of Biochemistry, Molecular Biology and Biophysics, University of Minnesota, St Paul, MN 55108, USA
1 To whom correspondence should be addressed. E-mail: schmi232{at}umn.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Insertion of metal ions into tetrapyrrole macrocycles is catalyzed by a diverse group of enzymes called chelatases. Structures are known for several chelatases catalyzing metal insertion into protoporphyrin IX or sirohydrochlorin. Despite a lack of significant amino acid sequence similarity, these ferro- and cobaltochelatases share a high degree of structural similarity. Cobaltochelatase CbiK and ferrochelatase HemH are bilobial enzymes with two
/ß domains, which were suggested to origin from a common ancestral protein via gene duplication. Small, single-domain chelatases (CbiXS) were recently described in archaea and are believed to represent primordial chelatases. Here, we tested the structural plasticity of an archaeal cobaltochelatase CbiXS by rearranging its structure with a novel method producing random in-frame deletions, duplications and insertions. A number of functional chelatase variants with insertion of duplicated sequence stretches, encompassing from one to nine secondary structural elements, were obtained. CbiXS was found to tolerate large sequence rearrangements in four out of the nine loop regions of the protein, indicating a high degree of structural plasticity. The predicted topologies of two variants (M51 and M518) are strikingly similar to CbiK and HemH, suggesting that we recreated duplication events that are believed to have created the bilobial chelatases.
Keywords: CbiK/CbiX/chelatase/directed evolution/HemH
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Insertion of metal ions (e.g. iron, cobalt, nickel and magnesium) into the center of tetrapyrrole macrocycles is catalyzed by a diverse group of enzymes called chelatases (Brindley et al., 2003
). Structures are known for several chelatases catalyzing metal insertion (cobalt or iron) into protoporphyrin IX or sirohydrochlorin. Despite a lack of significant amino acid sequence similarity, these ferro- and cobaltochelatases share a high degree of structural similarity.
The protoporphyrin IX ferrochelatase from B. subtilis (HemH) is probably the best studied chelatase (Hansson, 1994; Lecerof et al., 2000, 2003; Olsson et al., 2002). This enzyme is organized into two domains with a similar Rossman-type /ß-fold that form a deep cleft between them lined with several invariant residues important for catalysis (Kohno et al., 1994; Gora et al., 1996) (Fig. 1). The similarity of the folds and conservation of residues between the two domains suggests that they origin from a common ancestral protein via gene duplication (AlKaradaghi et al., 1997). Co-crystallizations of the enzyme with an inhibitor and metalation studies of this transition-state analog provided details on protoporphyrin IX binding, macrocycle distortion and subsequent metal insertion (Lecerof et al., 2000, 2003; Shipovskov et al., 2005). Two invariant residues Glu264 and His183 are most important for catalysis. His183 has been shown to be the primary metal coordination site. Glu264 has also been shown to participate in metal coordination together with His183, but mutations studies also suggested a role in proton abstraction from the porphyrin macrocycle (Kohno et al., 1994; Gora et al., 1996).
|
The anaerobic cobaltochelatase from Salmonella typhimurium (CbiK) shares a high degree of structural similarity with the ferrochelatase HemH, despite a low sequence identity (Schubert et al., 1999
). Both proteins have the same bilobial topology, although two additional helices in the N-terminal domain and a tilting of this domain make the active site cleft relatively inaccessible in HemH compared with CbiK. Residues His207 and His145 in CbiK, which have been shown to be important for cobalt ion selectivity, occupy positions equivalent to the metal coordinating residues Glu264 and His183 in HemH (Fig. 1).
More recently, two small archaeal cobaltochelatases (CbiXS) from Methanosarcina barkeri and Methanobacter thermoautotrophicum, only about half the size of other known ferro- and cobaltochelatases, have been described (Brindley et al., 2003). These enzymes align well with both domains of another cobaltochelatase from Bacillus megaterium (CbiXL), indicating that CbiXL (and also CbiK and HemH) may have evolved from two CbiXS proteins fused together where one domain of the fused protein maintained the catalytic residues and the other domain evolved other functions (Brindley et al., 2003). When we generated a structural model of the M. barkeri CbiXS enzyme from a recently published structure of a hypothetical protein from Archaeoglobus fulgidus (PDB entry 1TJN) (Yin et al., 2006), the structural similarity of CbiXS to the domains of the larger chelatases became obvious (Fig. 1). Two histidines, His12 and His78, analogous to the cobalt ion coordinating histidine residues in CbiK can be identified in CbiXS. All these data suggest that CbiXS represents a primordial chelatase leading to the emergence of other chelatases by means of gene duplication.
In this study, we aimed to reproduce the same type of events that might have led to the emergence of the bilobial chelatases and evaluate the structural plasticity of the hypothesized primordial chelatase CbiXS from M. barkeri. We developed a method of random in vitro gene sequence rearrangements that can produce random sequence duplications, deletions, insertions and inversions while maintaining the reading frame. Screening of the generated library utilized a functional selection scheme in siroheme deficient E. coli that relied on the ability of CbiXS to insert both cobalt and to a lesser extend iron into its tetrapyrrole substrate sirohydrochlorin (Brindley et al., 2003). A number of functional CbiXS variants were obtained that contained extensive duplications, deletions and insertions of various sequences. Two of the selected clones closely resembled CbiK and HemH.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Chemicals and reagents
Porphobilinogen was from Porphyrin Products Inc.; restriction and modification enzymes were purchased from MBI Fermentas, Vent DNA polymerase was purchased from New England Biolabs, Nickel-NTA resin and gel filtration columns were from Amersham Biosciences, Pfu DNA polymerase from Stratagene, medium components from Fisher Scientific and oligonucleotide primers were from Integrated DNA Technologies.
Bacterial strains, media and growth conditions
E. coli JM109 was used as a host both for cloning and overexpression of proteins for metal affinity chromatography. A cysG deletion strain E. coli JM109 
cysG was used for complementation experiments and library screening. This strain was obtained according to the method described by Datsenko and Wanner (2000). For complementation experiments, E. coli cysG transformants were first grown overnight in LB medium supplemented with 100 µg/l of ampicillin at 37°C. About 0.5 ml of the overnight culture was then centrifuged for 2 min at 6000 rpm in a bench top centrifuge. The pellet was washed twice with 1 ml of M9 medium (NaCl 0.5 g/l, Na2HPO4 6 g/l, KH2PO4 3 g/l, NH4Cl 1 g/l, glucose 2 g/l, MgSO4 2 mM and CaCl2 0.1 mM) to eliminate all traces of cysteine. Washed cells were resuspended in 0.5 ml of M9 medium and 10 µl of the suspension was inoculated in 4 ml of M9 medium supplemented with 100 µg/l of ampicillin. Tubes were shaken at 30°C for 5 days with constant growth monitoring. Positive controls contained 50 mg/l of cysteine.
Cloning and plasmid construction
Siroheme synthase cysG gene was amplified from genomic DNA of E. coli JM109. Truncated cysG variant cysGa (residues 203–457) (Fazzio and Roth, 1996) was generated following the QuickChangeTM protocol (Stratagene) and cloned into pUCmod, a derivative of pUC19 obtained by replacing a DNA stretch encompassing the lac-operator through the lacZ gene with a new multiple cloning site (XbaI, SmaI, EcoRI, NcoI and NotI) (Schmidt-Dannert et al., 2000). This vector facilitates constitutive expression of cloned genes when provided with a Shine–Dalgarno sequence. The resulting plasmid was called pUC-cysGa. The region downstream of the cysGa gene was then replaced with a new multiple cloning site (SmaI, PstI, HindIII, XhoI, EcoRI and NotI). This plasmid (pUC-cysGa-L1, Fig. 2B) was used to express synthetic cbiXS gene encoding the cobalt chelatase from M. barkeri and all its variants.
|
Synthesis of cbiXS gene
The synthetic gene encoding CbiXS from M. barkeri (GeneBank accession #NZ_AAAR01001953) was synthesized by PCR. The amino acid sequence of CbiXS was backtranslated using the E. coli K-12 codon usage (Codon usage database at http://www.kazusa.or.jp/codon/) and the resulting nucleotide sequence divided into 18 overlapping 45 bp-long oligonucleotides. Oligonucleotides were assembled by PCR using Pfu polymerase. PCR conditions were: initial denaturation at 95°C for 2.5 min followed by 90 cycles of denaturation at 94°C for 20 s, annealing at 50°C for 20 s and extension at 72°C for 40 s. The assembled gene was amplified with a forward primer containing a HindIII restriction site and a reverse primer with an XhoI restriction site for cloning into plasmid pUC-cysGa-L1 to give plasmid pUC-cysGa-CbiXs. The nucleotide sequence of the synthetic gene was verified by sequencing.
Two sets of forty 21 bp-long (each covering exactly seven codons and spaced two codons apart) oligonucleotides that evenly cover the CbiXS gene sequence (coding for 130 amino acids) in both forward and reverse direction were used to amplify three different segments of the CbiXS gene (Fig. 3). Two PCR reactions (left and right segments) contained either the 5' forward or 3' reverse gene sequence flanking oligonucleotides and the corresponding oppositely priming 39 oligonucleotides. The third reaction (middle segment) contained all oligonucleotides except for the two flanking oligonucleotides. Internal oligonucleotides contained a KasI restriction site overhang whereas the two flanking oligonucleotides contained a HindIII (5' forward primer) or XhoI restriction site. Each amplification reaction contained 40 pmol of each oligonucleotide and 100 ng of pUC-cysGa-CbiXs as template. Amplifications were carried out with Vent DNA polymerase in the manufacturer-supplied reaction buffer (Promega). The PCR mix (25 µl) was initially heated at 95°C for 2.5 min followed by 30 cycles of denaturation at 94°C for 15 s, annealing at 55°C for 15 s and extension at 72°C for 40 s. Following amplification, fragments from PCR 1, 2 and 3 were gel purified, mixed with each other (
200 ng of each fragment), digested with KasI and ligated. One microliter (100 ng of DNA) of the ligation mixture was then used in another PCR reaction (using the same general amplification conditions as before) to amplify ligation products containing 5' and 3' sequences of wild-type CbiXS with gene sequence flanking oligonucleotides containing HindIII and XhoI sites for cloning of the rearranged amplification products into the expression vector pUC-cysGa-L1. The resulting cbiXS library was then transformed into electrocompetent E. coli JM109 cysG. Transformed cells were allowed to recover in 1 ml of SOC medium at 37°C for 1.5 h before being washed with M9 medium and plated on LB (naïve library) or M9 agar plates containing 100 µg/l ampicillin (selected library). M9 plates contained Noble agar to avoid traces of cysteine.
|
DNA Sequencing
Sequencing was performed at the BioMedical Genomics Center at the University of Minnesota using the ABI PRISM BigDye Terminator Sequencing kit and DNA analyzer ABI3130x1.
Sirohydrochlorin was generated as described in Schubert et al. (2002) by incubating porphobilinogen with porphobilinogen deaminase (HemC), uroporphyrinogen III synthase (HemD), uroporphyrinogen III methyltransferase (YlnD) and precorrin-2 dehydrogenase (YlnF). Cloning of HemC and HemD into the constitutive expression vector pUCmod was described previously (Kwon et al., 2003). YlnD and YlnF were cloned from B. subtilis into pUCmod and identified as homologs of previously reported SirA (uroporphyrinogen III methyltransferase) and SirC (precorrin-2 dehydrogenase) of B. megaterium (Johansson and Hederstedt, 1999). E. coli transformed with hemC, hemD, ylnD or ylnF cloned into pUCmod were grown for 24 h at 30°C. Cells were then pelleted, suspended in 50 mM Tris buffer pH 8, containing 50 mM NaCl and sonicated. Cell lysate was cleared by centrifugation and the resulting crude protein extracts were used for sirohydrochlorin synthesis (Schubert et al., 2002).
CbiXS proteins were purified from bacterial lysates using metal chelate affinity chromatography. After application of the crude cell lysates to the metal chelate resin, the column was washed with 50 mM phosphate buffer, pH 8.0, containing 10 mM imidazole and 0.3 M NaCl. Bound CbiXS was eluted in buffer containing 150 mM imidazole and 0.3 M NaCl. Proteins were desalted by gel filtration on a PD10 column (Amersham) with a 50 mM Tris–HCl buffer, pH 8.0, containing 50 mM NaCl. Chelatase activity was measured by monitoring the disappearance of sirohydrochlorin (
max 376 nm) using an extinction coefficient of 2.4 x 105 M–1 cm–1. CbiXS protein (50 µg) was incubated in 1 ml reaction volume of 50 mM Tris–HCl, pH 8 with sirohydrochlorin (2.5 µM) and CoCl2 (20 µM). Assays were performed in triplicate, and initial rates were recorded with an Ultrospec 3300pro spectrophotometer (Amersham Biosciences) at 37°C.
Protein structure homology modeling
The model structure of CbiXS from M. barkeri (GeneBank accession #NZ_AAAR01001953) (Brindley et al., 2003) was obtained with the recently published X-ray structure of a hypothetical protein from A. fulgidus (PDB entry 1TJN) and using the automated homology modeling server SWISS-MODEL (http://swissmodel.expasy.org//SWISS-MODEL.html).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Selection scheme for the identification of functional CbiXS variants
Successful identification of functional CbiXS variants from sequence libraries requires an efficient screening or selection system. In the case of the cobaltochelatase CbiXS, we can utilize its ability to insert both cobalt and to a lesser extend iron into its tetrapyrrole substrate sirohydrochlorin for the development of a functional selection scheme (Brindley et al., 2003). Insertion of iron into sirohydrochlorin produces siroheme; the prosthetic group of sulfite reductase SiR required in E. coli for cysteine biosynthesis. Strains deficient in siroheme biosynthesis are therefore cysteine auxotrophs.
Siroheme biosynthesis in E. coli is accomplished by a trifunctional siroheme synthetase CysG that converts uroporphyrinogen III (Uro'gen III) via three enzymatic reactions to siroheme (Spencer et al., 1993) (Fig. 2A). Uro'gen III is first methylated and the resulting precorrin-2 is then oxidized to sirohydrochlorin by a NAD+-dependent precorrin-2 dehydrogenase, which is not absolutely necessary in E. coli as precorrin-2 spontaneously oxidizes (Leech et al., 2002; Raux et al., 2003). Finally, the ferrochelatase activity of CysG catalyzes the insertion of iron into precorrin-2 to produce siroheme. The CysG protein can be split into two functional subunits, CysGa (C-terminal residues 203-457) and CysGb (N-terminal residues 1-202) that independently catalyze the methylation and oxidation/metal insertion (Fazzio and Roth, 1996). We constructed a cysG deletion strain and complemented its cysteine auxotrophy on minimal medium with a complementing plasmid expressing CysGa and CbiXS (pUC-cysGa-cbiXS) (Fig. 2B). Insertion of a CbiXS gene library in place of the wild-type CbiXS gene on the complementing plasmid and transformation into E. coli cysG provides a simple screen for functional CbiXS variants.
Generation of in-frame sequence rearrangements
To replicate protein sequence rearrangements, duplications and indels that may have led to the evolution of bilobial chelatases from a primordial, small chelatase like CbiXS, we needed to develop a method that would allow us to introduce at a high frequency random deletions, insertions and duplications into a gene sequence while maintaining its reading frame and without inserting many new mutations. Current methods mainly focus on the recombination of gene sequences and either require a high nucleotide sequence homology [e.g. DNA shuffling (Stemmer, 1994; Zhao and Arnold, 1997)] to generate recombination at relatively low frequencies or randomly recombine sequences from unrelated genes [e.g. ITCHY (Ostermeier et al., 1999) or NRR (Bittker et al., 2002)] resulting in a large fraction of nonsense frame-shift mutations. We designed two sets of 21-bp long (each covering exactly seven codons) oligonucleotides that evenly cover the CbiXS gene sequence (coding for 130 amino acids) in both forward and reverse direction (Fig. 3). No structural considerations were used for primer design. The CbiXS gene was synthesized with a codon composition optimized for expression in E. coli, but eliminating codons that could produce stop-codons during sequence rearrangements.
Segments of the CbiXS gene were then amplified in three PCR reactions; two reactions contained either the 5' forward or 3' reverse gene sequence flanking oligonucleotides and the corresponding oppositely priming 40 oligonucleotides, and the third reaction contained all oligonucleotides except for the two flanking oligonucleotides (Fig. 3). As a result, amino-, carboxyterminal and internal gene fragments are obtained that are spaced two codons apart.
Purified fragments of all three PCR reactions were mixed and ligated to obtain in-frame CbiXS sequences with duplications, deletions and insertions. Because blunt-ended ligation of the PCR products proved inefficient, we added a KasI restriction site to all internal oligonucleotides. This site was chosen because it encodes glycine and alanine, residues without bulky side groups that are easily incorporated into protein folds. Following fragment ligation, a second PCR reaction was performed with CbiXS gene flanking oligonucleotides to amplify fragments containing 5' and 3' sequences of CbiXS. The resulting library of CbiXS sequences was then cloned into the expression vector pUC-cysGa-L1 (Fig. 2B) and transformed into E. coli cysG. About 20 000 clones were obtained on LB agar plates. Fifty clones were randomly picked for sequence analysis and cultivation in minimal medium. One clone (mutant M10, Fig. 4) was found to grow in minimal medium, indicating the expression of a functional chelatase capable of complementing the cysteine auxotrophy of E. coli cysG. This clone (M10, Figs 4 and 5) had an extensive internal duplication. The other sequenced clones had different kinds of deletions and duplications (not shown). Only one clone had a frame-shift mutation, showing that this method is efficiently producing in-frame sequence rearrangements.
|
|
Selection of functional CbiXS variants
The CbiXS gene library was then plated on minimal agar plates to select for clones expressing functional chelatase variants. About 500 E. coli clones grew and were further analyzed by colony PCR to identify CbiXS sequences that were different in size from the wild-type sequence as the result of sequence rearrangements. Fifty clones were identified that contained CbiXS sequence variants that were mostly larger than the wild-type gene sequence. Sequence analysis of these variants showed that many contained N-terminal and C-terminal duplications as shown for three representative variants M51, M136 and M200 in Figs 4 and 5. Other variants, however, had various internal duplications (M10, M54, M502, M508, M518 and M519; Figs 4 and 5). Two variants (M507 and M510; Figs 4 and 5) had a duplication and insertion. The insertion in M507 has the same sequence as the region where it is inserted, which together with a duplication in the same sequence region results in the triplication of a short sequence stretch (Fig. 4). Short deletions and substitution were found in two variants (M76 and M509; Fig. 4).
Topology models of CbiXS variants
In order to map the sequence rearrangements found in the different CbiXS variants on the wild-type CbiXS structure, a structural model of CbiXS from M. barkeri was generated based on the structure of the recently crystallized but yet uncharacterized hypothetical protein af0721 from A. fulgidus (PDB entry 1TJN). This protein shows sequence similarity (50% similarity and 30% identity) to the characterized anaerobic CbiXS cobaltochelatases from M. barkeri and Methanobacter thermoautotrophicum (Brindley et al., 2003). Using the automated homology modeling server SWISS-Model, a structural model of the M. barkeri CbiXS was obtained (Fig. 1). The topological architecture of CbiXS is as expected very similar to that of the individual domains of bilobial chelatases CbiK and HemH (Schubert et al., 1999). However, CbiXS has one additional ß-sheet (ß4) as depicted in the topological diagrams shown in Fig. 5.
Because of the extensive rearrangements in the obtained CbiXS variants, no reasonable structural models could be produced. Instead, we mapped the deletions, duplications and insertions present in the different CbiXS variants on the topological diagram obtained for the wild-type enzyme. The resulting topological cartoons depicting the location and extent of the sequence rearrangements in the variants are shown in Fig. 5. The topological diagrams of two CbiXS variants (M51 and M518) show them to be strikingly similar to the topologies of CbiK or HemH. Both variants contain large duplications located at the C-terminus of M51 and between ß-sheet ß5 and helix 5 in M518. The large size of the duplications comprising most of the CbiXS sequence (1–5 in M51 and 1–ß5 in M518) except for the N-terminal ß-sheet suggests a bilobial topology similar to the one observed in CbiK and HemH (Schubert et al., 1999).
The other variants contained mostly less extensive insertions at different sites of the CbiXS protein sequence. Interestingly, except for the short terminal duplications in M200 and M136, sequence rearrangements occurred in or close to predicted loop regions in the structural model of CbiXS. Three of the variants (M508, M519 and M510) contained insertions in a loop between helices 3 and 4. Other loop regions in CbiXS found to tolerate insertions include those between ß-sheet ß3 and helix 3 (M502), between ß-sheets ß4 and ß5 (M54), ß-sheet ß1 and helix 1 (M507) as well as the large duplication of variant M518 between ß-sheet ß5 and helix 5. Thus, at least four out of nine loop regions present in CbiXS can tolerate insertions and the loop between helices 3 and 4 appears to be particularly tolerant to modifications, as three of the selected variants contained insertions in this region. Only two variants have rearrangements in their secondary structure elements: M76 has half of its ß4 is deleted and M507 has an insert of helix 1 in the middle of the same helix (Figs 4 and 5). In variant M509, the C-terminal helix 5 is replaced by a different sequence (Fig. 4) as the result of a frame-shift mutation. When this helix is deleted in the wild-type enzyme, the resulting truncated variant (WT5) is inactive and unable to complement the cysG deficiency of E. coli cysG (Table I). Thus, the substitution in M509 must have preserved the functional role of helix 5.
|
Characterization of CbiXS variants
Our selection scheme for functional cobaltochelatase CbiXS variants made use of its ability to insert both Co2+ and Fe2+ into sirohydrochlorin; a property which has also been reported for other cobaltochelatases (Raux et al., 1997). Ferrous iron, however, is inserted at only a rate of 10% of that of cobalt and produces siroheme (Brindley et al., 2003). Because Co2+ is the preferred substrate, addition of exogenous cobalt to cysteine auxotrophic E. coli overexpressing cobaltochelatases prevents complementation by presumably producing cobalt-sirohydrochlorin instead of siroheme (Raux et al., 1997). Since the complementation of E. coli cysG may require only low ferrochelatase activity for the synthesis of sufficient amounts of siroheme, chelatase activities of CbiXS variants were compared to the wild-type enzyme. E. coli transformants expressing wild-type and mutant CbiXS were grown in liquid M9 minimal medium at 30°C for 5 days to compare their growth rates that should reflect their chelatase activities. Three groups of mutants can be distinguished (Table I): mutants that grew at a rate comparable to the wild-type (M508, M136 and M200), mutants (almost all other mutants) with a somewhat impaired growth rate and one mutant (M502) with a significantly reduced growth rate. Addition of exogenous cobalt (0.1 mM) to the growth medium inhibited cell growth of all transformants, suggesting that the metal preference of the CbiXS variants was not changed.
To perform in vitro activity assays with purified protein, wild-type CbiXS and variants were histidine-tagged at their C-terminus. Two histidine-tagged variants (M76 and M509) lost their ability to complement the cysG deficiency of E. coli cysG, suggesting complete inactivation or loss of Fe2+ chelating activity (Table I). All CbiXS variants were expressed in E. coli at levels similar to the wild-type enzyme (data not shown) and present in the soluble cell fraction, indicating that they were well folded. The CbiXS proteins were readily purified by metal affinity chromatography and most proteins migrated on a SDS–PAGE gel in accordance with their expected molecular mass (Fig. 6). However, two variants (M507 and M508) showed a reduced mobility. A similar unexpected protein mobility has previously been reported for the Methanobacter thermoautotrophicum CbiXS and attributed to unusual protein properties (Brindley et al., 2003).
|
Chelatase activity of purified CbiXS proteins was measured with sirohydrochlorin and Co2+ as substrates (Table I). The specific activity measured for the recombinant wild-type CbiXS was 119 nmol min–1 mg–1 which agrees well with a previous report of 122 nmol min–1 mg–1 (Brindley et al., 2003
). Except for three variants that either inefficiently complemented the cysG deficiency (M502) or were inactivated (M76 and M509) by the addition of a histidine-tag, all other variants exhibited appreciable levels of cobalt chelating activity ranging from 5% to 90% of the wild-type activity. Not surprisingly, the two variants with short terminal duplications comprised of only one structural element (5 in M136 and ß1 in M200) have specific activities very similar to the wild-type enzyme (Table I). Insertions in loop regions expectedly decreased chelatase activity 5–20-fold (M508, M519, M510 and M507). Variants M51 and M518 with large duplications retained 22% and 6%, respectively, of the specific activity measured for the wild-type enzyme, meaning that at least one if not two functional active sites are present in the enzymes. The Km of the more active variant M51 was measured with sirohydrochlorin (Km = 0.43 ± 0.18 µM) and found to be quite similar to the wild-type protein (Km = 0.62 ± 0.09 µM), suggesting that substrate binding was not altered by the duplication in M51.
The presumed active site histidines in the wild-type enzyme are His12 and His78 which correspond to active site residues His145 and His207 in CbiK (Schubert et al., 1999) and His183 and Glu264 in HemH (Ferreira et al., 2002; Lecerof et al., 2003). In addition to His12 and 78, variants M51 and M518 have a histidine residue (His192 in M51 and His177 in M518) that corresponds to His78 in their duplicated part. First, His12 and His78 in wild-type CbiXS were substituted with alanine to establish their roles in catalysis. Two mutants were created and transformed into E. coli CysG for complementation. Mutant H12A was unable to complement the cysG deficiency of E. coli cysG, whereas slow growth was observed with mutant H78A (Table II). These results suggest that only His12 is absolutely essential for CbiXS activity whereas His78 appears to increase catalytic activity. Next, the role of active site histidines in variants M51 and M518 was investigated by replacing His78 and its counterpart in the duplicated parts of the proteins with alanine (Table II). Similar to the wild-type enzyme, substitution of His78 decreased the complementation efficiency of variants M51 and M518. However, mutation of His192 in M51 and His177 in M518 had no apparent influence on CbiXS activity, indicating that the duplicated domain in both variants is inactive.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Here, we tested the structural plasticity of the archaeal cobaltochelatase CbiXS by introducing extensive in frame sequence rearrangements. Such rearrangements, including domain duplications, deletions and substitutions, are a major driving force of protein evolution and contribute to the observed diversity of protein topologies and functions (Grishin, 2001
; Aravind et al., 2002; Koonin et al., 2002; Bjorklund et al., 2005; Orengo and Thornton, 2005; Vogel et al., 2005). Sequence analysis and structural information of proteins from superfamilies suggest that complex structures evolved from simple ancestral core folds by elaboration of their folds through the acquisition of new sequence elements inserted mostly into loop regions (Grishin, 2001; Aravind et al., 2002, 2006).
The mechanisms of protein evolution are increasingly explored for the investigation of protein functions as well as creation of new protein properties using rational or random (in vitro evolution) design strategies (Hopfner et al., 1998; Schmitz et al., 2001; Guntas and Ostermeier, 2004). In vitro protein evolution is a powerful protein engineering strategy, as it does not rely on structural information. The introduction of amino acid point mutations and homologous sequence crossovers is most widely used to generate protein sequence diversity (Stemmer, 1994; Zhao and Arnold, 1997). More recently, several strategies have been developed that produce sequence rearrangements by random ligation of DNA fragments (Ostermeier et al., 1999; Lutz et al., 2001; Sieber et al., 2001; Bittker et al., 2002, 2004; Kawarasaki et al., 2003). However, random ligation results in up to 66% out of frame sequences in the initial sequence library. In another approach, oligonucleotides encoding each desired crossover are synthesized (O'Maille et al., 2002). This approach is expensive, requires time-consuming primer design and produces fewer crossovers. We developed a method that would produce in frame sequence rearrangements to investigate the structural plasticity of CbiXS and reproduce duplication events that may have led to the emergence of the bilobial chelatases.
Sequencing of randomly picked clones from the unselected CbiXS library showed that this method effectively produced a wide range of in-frame sequence rearrangements. Selection for functional CbiXS variants and screening for clones with sequences larger than the wild-type (suggesting duplication events) identified 10 functional variants with insertions of duplicated sequence stretches that encompassed from one to nine secondary structural elements (Figs 4 and 5). Interestingly, most variants not only preserved appreciable cobalt chelatase activity in vitro compared with the wild-type enzyme, but also the purified proteins were soluble (Fig. 6). This suggests that the duplicated segments and their hydrophobic residues are either integrated into the overall fold or exposed hydrophobic residues are shielded from the solvent by, for example, oligomerization. Oligomerization was proposed as a mechanism of protein stabilization during the evolution of new protein architectures (Riechmann et al., 2005). Wild-type CbiXS was also found to form homodimers or homotrimers (Brindley et al., 2003), suggesting that the variants likewise are prone to oligomerize which needs to be further investigated in future work. The majority of insertions occurred in loop regions of CbiXS (Figs 4 and 5), which agrees with previous observations that loop regions of proteins are more tolerant to sequence rearrangements than secondary structural elements (Freimuth et al., 1990; Sondek and Shortle, 1990, 1992; Ladant et al., 1992; Beernink et al., 2001; Bittker et al., 2004). The relative high number of functional variants obtained that contained extensive insertions in loop regions suggests that CbiXS has a robust core-fold that is not easily perturbed by changes in regions that connect secondary structural elements. Additional random in vitro gene sequence rearrangement experiments with other proteins will be necessary to investigate whether this holds true for other proteins as well.
Although most variants were 10–20-fold less active than the wild-type enzyme, their iron chelating activity was sufficient to complement the cysG deletion in E. coli (Table I). Attachment of a C-terminal histidine-tag led to the inactivation of variants M76 and M509. Variant M76 had a deletion of several amino acids in the fifth ß-sheet and variant M509 contained a substitution of the last helix. These rearrangements located near the C-terminus may cause unfavorable interactions with the added histidine stretch.
The most interesting protein variants obtained in this study were variants M51 and M518 which may assume a topology similar to that of bilobial chelatases like the cobalt chelatase CbiK and ferrochelatase HemH (Schubert et al., 1999) (Fig. 4). None of the two variants are the result of a simple head-to-tail fusion of two cbiXS gene copies. Variant M51 has a C-terminal duplication of a CbiXS sequence that encompasses a sequence from helix 1 to helix 5. Variant M518 on the other hand contains an insertion of a sequence that encompasses helix 1 to helix 5 between ß-sheet ß5 and helix 5. However, neither variant has ß-sheet ß1 duplicated, which is located at the N-terminus of wild-type CbiXS and contains the active site His12. Replacement of His12 in wild-type CbiXS, M518 and in M51 completely abolished activity (Table II), confirming its essential role in catalysis. Substitution of His78 on the other hand only reduced CbiXS activity, which agrees with findings reported for substitution of equivalent residues in yeast ferrochelatase and CbiK (Gora et al., 1996; Schubert et al., 1999). Substitution of His192 in M51 and His177 in M518 as the equivalents to His12 in the duplicated domain expectedly had no influence on the proteins activity. Only the N-terminal domain of M51 and M518 contains therefore catalytic functional groups. These two catalytic residues are likewise only present in one of the two domains of CbiK, HemH and CbiXL, whereas the other domain has evolved different functions and participates, e.g. in porphyrin binding (AlKaradaghi et al., 1997). However, which of the duplicated domains maintains the catalytic residues is different: in CbiK and HemH, it is the C-terminal domain whereas in CbiXL (like in M51 and M518), it is the N-terminal domain (Brindley et al., 2003).
In conclusion, we developed a method of random in vitro in-frame sequence rearrangement to probe the structural plasticity of a hypothesized ancestral archaeal chelatase CbiXS and recreate duplication events that are believed to have created the bilobial chelatases (Brindley et al., 2003). We show that CbiXS tolerates large sequence rearrangements in four out of the nine loop regions of the protein and that random in vitro sequence rearrangement result in two variants with topologies that are strikingly similar to that of bilobial ferrochelatases and cobaltochelatases. These rearrangements, in particular the large loop insertions found in several variants, may now serve as starting points for the evolution of new functions such as, e.g. binding of new substrates. Substrate binding may require a larger active side and/or interaction with additional residues that could be provided by the inserted loop sequences. Additional in vitro evolution studies aimed at creating new chelatase activities in these variants are currently ongoing. The approach presented here may be applied to other proteins as well for structure–function investigation and/or to engineer new protein functions.
|
Footnotes |
|---|
Edited by Evamaria Peterson
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
This research was supported by the National Institute of Health (grant NIH/1R01-GM65471-01) and the David and Lucile Packard Foundation (grant #2001-18996).
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
AlKaradaghi S., Hansson M., Nikonov S., Jonsson B., Hederstedt L. Structure (1997) 5:1501–1510.[Medline]
Aravind L., Mazumder R., Vasudevan S., Koonin E.V. Curr. Opin. Struct. Biol. (2002) 12:392–399.[CrossRef][Web of Science][Medline]
Aravind L., Iyer L.M., Koonin E.V. Curr. Opin. Struct. Biol. (2006) 16:409–419.[CrossRef][Web of Science][Medline]
Beernink P.T., Yang Y.R., Graf R., King D.S., Shah S.S., Schachman H.K. Protein Sci (2001) 10:528–537.[CrossRef][Web of Science][Medline]
Bittker J.A., Le B.V., Liu D.R. Nat. Biotechnol. (2002) 20:1024–1029.[CrossRef][Web of Science][Medline]
Bittker J.A., Le B.V., Liu J.M., Liu D.R. Proc. Natl. Acad. Sci. USA (2004) 101:7011–7016.
Bjorklund A.K., Ekman D., Light S., Frey-Skott J., Elofsson A. J. Mol. Biol. (2005) 353:911–923.[CrossRef][Web of Science][Medline]
Brindley A.A., Raux E., Leech H.K., Schubert H.L., Warren M.J. J. Biol. Chem. (2003) 278:22388–22395.
Datsenko K.A., Wanner B.L. Proc. Natl. Acad. Sci. USA (2000) 97:6640–6645.
Fazzio T.G., Roth J.R. J. Bacteriol. (1996) 178:6952–6959.
Ferreira G.C., Franco R., Mangravita A., George G.N. Biochemistry (2002) 41:4809–4818.[CrossRef][Medline]
Freimuth P.I., Taylor J.W., Kaiser E.T. J. Biol. Chem. (1990) 265:896–901.
Gora M., Grzybowska E., Rytka J., Labbe-Bois R. J. Biol. Chem. (1996) 271:11810–11816.
Grishin N.V. J. Struct. Biol. (2001) 134:167–185.[CrossRef][Web of Science][Medline]
Guntas G., Ostermeier M. J. Mol. Biol. (2004) 336:263–273.[CrossRef][Web of Science][Medline]
Hansson M.H.L. Eur. J. Biochem. (1994) 220:201–208.[Web of Science][Medline]
Hopfner K.P., Kopetzki E., Kresse G.B., Bode W., Huber R., Engh R.A. Proc. Natl Acad. Sci. USA (1998) 95:9813–9818.
Johansson P., Hederstedt L. Microbiology (1999) 145(Pt 3):529–538.[CrossRef][Web of Science][Medline]
Kawarasaki Y., Griswold K.E., Stevenson J.D., Selzer T., Benkovic S.J., Iverson B.L., Georgiou G. Nucleic Acids Res. (2003) 31:e126.
Kohno H., Okuda M., Furukawa T., Tokunaga R., Taketani S. Biochim. Biophys. Acta (1994) 1209:95–100.[CrossRef][Medline]
Koonin E.V., Wolf Y.I., Karev G.P. Nature (2002) 420:218–223.[CrossRef][Medline]
Kwon S.J., de Boer A.L., Petri R., Schmidt-Dannert C. Appl. Environ. Microbiol. (2003) 69:4875–4883.
Ladant D., Glaser P., Ullmann A. J. Biol. Chem. (1992) 267:2244–2250.
Lecerof D., Fodje M., Hansson A., Hansson M., Al-Karadaghi S. J. Mol. Biol. (2000) 297:221–232.[CrossRef][Web of Science][Medline]
Lecerof D., Fodje M.N., Alvarez Leon R., Olsson U., Hansson A., Sigfridsson E., Ryde U., Hansson M., Al-Karadaghi S. J. Biol. Inorg. Chem. (2003) 8:452–458.[Web of Science][Medline]
Leech H.K., Raux-Deery E., Heathcote P., Warren M.J. Biochem. Soc. Trans. (2002) 30:610–613.[CrossRef][Web of Science][Medline]
Lutz S., Ostermeier M., Moore G.L., Maranas C.D., Benkovic S.J. Proc. Natl Acad. Sci. USA (2001) 98:11248–11253.
O'Maille P.E., Bakhtina M., Tsai M.D. J. Mol. Biol. (2002) 321:677–691.[CrossRef][Web of Science][Medline]
Olsson U., Billberg A., Sjovall S., Al-Karadaghi S., Hansson M. J. Bacteriol. (2002) 184:4018–4024.
Orengo C.A., Thornton J.M. Annu. Rev. Biochem. (2005) 74:867–900.[CrossRef][Web of Science][Medline]
Ostermeier M., Shim J.H., Benkovic S.J. Nat. Biotechnol. (1999) 17:1205–1209.[CrossRef][Web of Science][Medline]
Raux E., Thermes C., Heathcote P., Rambach A., Warren M.J. J. Bacteriol. (1997) 179:3202–3212.
Raux E., et al. Biochem. J. (2003) 370:505–516.[CrossRef][Web of Science][Medline]
Riechmann L., Lavenir I., de Bono S., Winter G. J. Mol. Biol. (2005) 348:1261–1272.[CrossRef][Web of Science][Medline]
Schmidt-Dannert C., Umeno D., Arnold F.H. Nat. Biotechnol. (2000) 18:750–753.[CrossRef][Web of Science][Medline]
Schmitz H.P., Jockel J., Block C., Heinisch J.J. J. Mol. Biol. (2001) 311:1–7.[CrossRef][Web of Science][Medline]
Schubert H.L., Raux E., Wilson K.S., Warren M.J. Biochemistry (1999) 38:10660–10669.[CrossRef][Medline]
Schubert H.L., Raux E., Brindley A.A., Leech H.K., Wilson K.S., Hill C.P., Warren M.J. Embo. J. (2002) 21:2068–2075.[CrossRef][Web of Science][Medline]
Shipovskov S., Karlberg T., Fodje M., Hansson M.D., Ferreira G.C., Hansson M., Reimann C.T., Al-Karadaghi S. J. Mol. Biol. (2005) 352:1081–1090.[CrossRef][Web of Science][Medline]
Sieber V., Martinez C.A., Arnold F.H. Nat. Biotechnol. (2001) 19:456–460.[CrossRef][Web of Science][Medline]
Sondek J., Shortle D. Proteins (1990) 7:299–305.[CrossRef][Web of Science][Medline]
Sondek J., Shortle D. Proteins (1992) 13:132–140.[CrossRef][Web of Science][Medline]
Spencer J.B., Stolowich N.J., Roessner C.A., Scott A.I. FEBS Lett. (1993) 335:57–60.[CrossRef][Web of Science][Medline]
Stemmer W.P. Proc. Natl Acad. Sci. USA (1994) 91:10747–10751.
Vogel C., Teichmann S.A., Pereira-Leal J. J. Mol. Biol. (2005) 346:355–365.[CrossRef][Web of Science][Medline]
Yin J., Xu L.X., Cherney M.M., Raux-Deery E., Bindley A.A., Savchenko A., Walker J.R., Cuff M.E., Warren M.J., James M.N. J. Struct. Funct. Genomics. (2006) 7:37–50.[CrossRef][Medline]
Zhao H., Arnold F.H. Nucleic Acids Res. (1997) 25:1307–1308.
Received January 11, 2007; revised March 21, 2007; accepted March 30, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||