PEDS Advance Access originally published online on February 3, 2006
Protein Engineering Design and Selection 2006 19(4):163-167; doi:10.1093/protein/gzj015
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Engineering of Escherichia coli L-serine O-acetyltransferase on the basis of crystal structure: desensitization to feedback inhibition by L-cysteine
1Institute of Life Sciences, Ajinomoto Co., Inc., 1-1 Suzuki-cho, Kawasaki-ku, Kawasaki 210-8681, Japan, 2Ajinomoto-Genetika Research Institute, Moscow 117545, Russia and 3Department of Bioscience, Fukui Prefectural University, Fukui 910-1195, Japan
4 To whom correspondence should be addressed. E-mail: eiichiro_suzuki{at}ajinomoto.com
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
L-Serine O-acetyltransferase (SAT) from Escherichia coli catalyzes the first step of L-cysteine synthesis in E.coli and is strictly inhibited by the second step product, L-cysteine. To establish a fermentation process to produce L-cysteine, we embarked on a mutational study of E.coli SAT to desensitize the feedback inhibition by L-cysteine. The crystal structure and the reaction mechanism of SAT from E.coli have shown that the substrate L-serine and the inhibitor L-cysteine bind to the identical region in the SAT protein. To decrease the affinity for only L-cysteine, we first built the structure model of L-serine-binding SAT on the basis of the crystal structure with bound L-cysteine and compared these two structures. The comparison showed that the C
of Asp92 underwent a substantial positional change upon the replacement of L-cysteine by L-serine. We then introduced various amino acid substitutions at positions 89–96 around Asp92 by randomized, fragment-directed mutagenesis to change the position of the Asp92. As a result, we successfully obtained mutant SATs which have both extreme insensitivity to an inhibition by L-cysteine (the concentration that inhibits 50% activity; IC50 = 1100 µmol/l, the inhibition constant; Ki = 950.0 µmol/l) and extremely high emzymatic activities.
Keywords: desensitization/Escherichia coli/feedback inhibition/L-cysteine/L-serine O-acetyltransferase
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
L-Cysteine plays crucial roles in the structure, stability and catalytic function of many proteins, and is also an important amino acid used in the pharmaceutical, food and cosmetics industries. In enteric bacteria such as Escherichia coli and Salmonella typhimurium (Kredich and Tomkins, 1966
; Kredich et al., 1969; Soda, 1987) and higher plants (Smith and Thompson, 1969; Nakamura et al., 1987), L-cysteine is synthesized via O-acetyl-L-serine through the pathway constituted with L-serine O-acetyltransferase (SAT) (EC 2.3.1.30
[EC]
) and O-acetyl-L-serine sulfhydrylase (EC 4.2.99.8
[EC]
). The SAT enzyme catalyses the formation of O-acetylation from acetyl-CoA and L-serine, which is the first reaction in the two-step process of sulfur assimilation.
 | (1) |
 | (2) |
; Kredich et al., 1969). Takagi et al. (1999a,b) have recently reported that L-cysteine is overproduced to some extent in the L-cysteine non-utilizing E.coli strains that express the feedback inhibition-insensitive SATs (Nakamori et al., 1998). Site-directed and PCR random mutagenesis in the cysE gene encoding E.coli SAT was employed to construct the mutant enzymes that cause L-cysteine overproduction as a result of desensitization to feedback inhibition (Nakamori et al., 1998; Takagi et al., 1999b). Expression of two complementary DNAs encoding feedback inhibition-insensitive SAT isoforms of Arabidopsis thaliana (Noji et al., 1998) also resulted in L-cysteine overproduction (Takagi et al., 1999a). However, these SATs showed significant decreases in enzymatic activity relative to the E.coli wild-type enzyme (feedback inhibition-sensitive). Further improvements in L-cysteine production are therefore expected to use an engineered SAT, which shows a higher level of feedback desensitization and a higher catalytic activity.
Recently, the three-dimensional (3D) structure of SAT from E.coli with its inhibitor L-cysteine was determined by X-ray crystallography (Pye et al., 2004). This crystal structure was solved by single-wavelength anomalous dispersion phasing from seleno-L-methionine substituted SAT at 2.2 Å resolution. SAT was shown to be a trimeric structure with 3-fold symmetry, independent of the crystallographic 3-fold symmetry, forming a three-sided pyramid shape. This trimer is likely to interact with another SAT trimer at N-terminal ends. An L-cysteine molecule was observed to bind at the serine substrate site but not at the acetyl-CoA site between monomers in the trimeric interaction.
In contrast, Hindson and Shaw (2003) have performed steady-state kinetic and dead-end inhibition studies for alternative substrates, which showed that the reaction mechanism of SAT is a random-order ternary complex mechanism. Hindson also showed that L-cysteine would interact productively at the serine-binding site and suggested that binding of L-cysteine to SAT may give rise to a reduction in affinity for acetyl-CoA (Hindson, 2003). The reaction mechanism also was proposed to be similar to that described for chloramphenicol acetyltransferase (Moody and Wilkinson, 1990).
In this study, we first observed the crystal structure of SAT from E.coli with its inhibitor L-cysteine. Pye et al. (2004) built a model of the serine-binding SAT on the basis of the crystal structure, and then compared these two structures. We found that the C of Asp92 is the most flexible atom among these structures. In this case, because the substrate L-serine and the inhibitor L-cysteine have a very similar chemical structure, we hypothesized that SAT's interaction site needed a very subtle structural change to adapt its affinity for L-serine rather than L-cysteine. As we assumed that such a subtle structural change cannot be achieved with a simple procedure, amino acid's substitutions were then carried out at eight residues around Asp92 rather than just Asp92. We report here the isolation and characteristics of the mutant SATs that have high activities and are extremely desensitized to L-cysteine.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
Docking study
L-Serine-binding SAT was constructed by the Affinity program within the docking module of InsightII (Molecular simulations Inc., San Diego, CA, USA). Affinity is a suite of programs for automatically docking a ligand (guest) to a receptor (host); the process is an energy-driven method according to the classification of Kuntz et al. (1994). The CVFF force field was used for docking and scoring. The 3D structure of SAT was constructed by removing an L-cysteine molecule and changing Se atoms to S atoms from the crystal structure of SAT (PDB code 1T3D
[PDB]
). The 3D structure of L-serine was extracted from a 3D fragment of Builder/InsightII (Molecular simulations Inc.). Hydrogen atoms were added to these 3D structures by using the same program.
Because Affinity allows predefined atoms of the ‘ligand’ (L-serine) and the ‘binding site’ (L-cysteine-binding site) to relax during docking, we initially ‘pre-docked’ L-serine into the L-cysteine-binding site of SAT-like L-cysteine by manual manipulation. That is, the carbonyl oxygens were at distances equivalent to those of hydrogen bonds with Arg192, the amide nitrogen was at distances equivalent to those of hydrogen bonds with Asp92 and Asp157, and the acyl oxygen was at distances equivalent to those of hydrogen bonds with His158 and His 193.
The binding subset was defined to be the residues present at <6 Å from the ‘pre-docked’ L-serine. The bulk of the SAT, defined as atoms not in the binding subset, was held rigidly during the docking process. Using Monte Carlo minimization, we generated the 10 starting structures. The parameters of minimization were 100 minimization steps performed by the Quartic_vdw_no_Coul method as a non-bond summation procedure, with the convergence of 1 x 10–6 kcal/mol used as the energy test and 1 Å used as the RMSD tolerance threshold. Simulated annealing was then applied for each respective selected complex with 50 stages of 100 fs, the initial and final temperatures being 500 and 300 K, respectively. After we obtained the structure of L-serine-binding SAT, we compared it with L-cysteine-binding SAT using the Superimpose module of InsightII.
Construction of the mutant cysE genes by randomized, fragment-directed mutagenesis
First, we prepared the DNA fragment with the promoter region of the ompC gene or the nlpD gene by PCR using the genomic DNA from E.coli MG1655 as a template, oligonucleotide primers P4 and P6 (Figure 1) for the ompC gene, and primers P5 and P7 (Figure 1) for the nlpD gene. The expected bands (0.3 kb) of the PCR products were digested with SphI and SalI, and ligated to the SphI–SalI sites of plasmid pMW118 (Nippon Gene, Tokyo, Japan), to construct plasmids pMW-PompC and pMW-PnlpD, respectively. The wild-type cysE gene was also obtained by PCR with primers P1 and P2 (Figure 1) by using chromosomal DNA from E.coli MG1655. The unique amplified band of 0.83 kb was digested with SalI and XbaI, and ligated to the large fragment of plasmids pMW-PompC and pMW-PnlpD digested with SalI and XbaI to construct plasmids pMW-PompC-cysE and pMW-PnlpD-cysE, respectively. All oligonucleotide primers were designed on the basis of the available nucleotide sequences. The enzymes used for DNA manipulations and PyrobestTM DNA polymerase used for PCR amplification were obtained from Takara Shuzo (Kyoto, Japan) and used under conditions recommended by the supplier. The nucleotide sequences of the cysE gene were confirmed by DNA sequencing with a model 310 DNA sequencer (PE Biosystems, Foster City, CA, USA) by using the dideoxy chain termination method.
|
To construct the pool of the mutant cysE genes with the randomized nucleotides at positions 285–291, we first amplified by PCR the fragment of cysE gene coding the sequence from 1st to 102nd amino acid residues of SAT using plasmid pMW-PompC-cysE as a template, primer P3 containing six randomized nucleotides depicted by the letter ‘n’ (n indicates a mixture of G, A, T and C; Figure 1), and primer P4. The fixed, 19-nucleotides sequence at the 3' end of primer P3 is homologous to the amino acid sequence upstream of Val95. Twenty nanograms of plasmid pMW-PompC-cysE were added as a template to the reaction mixture (50 µl) containing each of the two primers (10 pmol). Twenty-five PCR cycles (96°C for 0.4 min, 60°C for 0.4 min, 72°C for 1 min) were carried out to obtain the 0.3 kb fragment. At the second step, the unique amplified band of 0.3 kb was purified by agarose gel electrophoresis and used as a ‘primer’ for the primer extension procedure to obtain the full sequence of cysE gene by 10 PCR cycles (96°C for 1 min, 40°C for 1 min, 72°C for 0.5 min). Finally, an aliquot (10 µl) of the PCR product was transferred to a fresh reaction mixture (40 µl) containing the plasmid pMW-PnlpD-cysE used as a template and the primers P4 and P2 at a concentration of 50 pmol each; five additional PCR cycles (94°C for 0.5 min, 57°C for 0.5 min, 72°C for 2 min) were performed. The 0.83 kb fragment containing the mutant cysE genes was purified by agarose gel electrophoresis, digested with SalI and XbaI, and then ligated to plasmid pMW-PompC digested with SalI and XbaI. The resultant plasmid, which had the various mutant cysE genes, was designated pMW-PompC-cysE(random).
Other mutant cysE genes, cysE1, cysE142 and cysE256, having an amino acid substitution [Arg89Pro, Arg89Ser/Thr90Leu and Met256Ile (Denk and Böck, 1987), respectively] in the SAT enzyme, were also constructed by the method of conventional site-directed mutagenesis using PCR.
Isolation of the mutant cysE genes
Escherichia coli strain LE392 cysE::KmR as a host cell was obtained from strain LE392 (Sambrook and Russell, 2001) by disruption of the cysE gene for further experiments. The chromosomal cysE gene was disrupted by using E.coli strain JC7623 to introduce the kanamycin resistance gene (Kushner et al., 1972). The plasmid pMW-PompC-cysE(random) was used to transform E.coli strain LE392 cysE::KmR. The transformants, which express the active mutant SAT with various amino acid substitutions at positions 95 and 96, were selected by complementing L-cysteine auxotrophy of E.coli strain LE392 cysE::KmR on M9 agar plates supplemented with 0.5% glucose and 50 mg/l methionine. The plasmid DNA from each transformant was purified, and the cysE gene in the plasmid was sequenced. The strain LE392 cysE::KmR was retransformed with these plasmids to determine the SAT activity.
Assay of SAT activity
Catalytic properties of mutant SATs were analyzed by a method described previously with some modifications (Kredich and Tomkins, 1966). Acetyl-CoA and other chemicals were purchased from Sigma Chemical Co. (St Louis, MO, USA). To determine mutant SAT activities, E.coli LE392 cysE::KmR cells carrying recombinant plasmids were grown in 40 ml of M9 minimal medium with appropriate supplements up to late exponential phase and harvested by centrifugation. The cells were washed only once with phosphate saline, resuspended in minimal volume of 50 mM potassium phosphate (pH 7.1), 2 mM DTT, 1 mM EDTA and disrupted by sonication on ice. Soluble protein extracts were obtained by centrifugation at 10 000 g and then were subjected to fractional precipitation with ammonium sulfate (20–75% saturation). The precipitates redissolved in minimal volume of buffer A, containing 50 mM Tris–HCl (pH 7.5), 2 mM DTT, 0.5 mM EDTA were dialyzed against the same buffer and load into Q-Sepharose column (2 ml). The absorbed proteins were eluted with linear gradient of KCl (0–0.5 M) in buffer A (20 ml). The combined active fractions were concentrated by precipitation of ammonium sulfate (80% saturation). Resulting pellets redissolved in minimal volume of buffer A were dialyzed against the same buffer using mini dialyzing devices (Pierce Co.). The desalted preparations were used as partially purified SAT for enzymatic activity assay. The enzyme activity of SAT was assayed at 25°C by monitoring the cleavage of the thioester bond of acetyl-CoA as described previously with some modifications (Kredich and Tomkins, 1966). We measured the initial rate of decrease in absorbance at 232 nm of the reaction mixture (final volume, 1 ml) containing 50 mmol/l Tris–HCl buffer (pH 7.5), 5 mmol/l L-serine, 0.1 mmol/l acetyl-CoA and enzyme solution; then, we subtracted the initial rate of decrease obtained for a solution containing the same materials except L-serine (blank). The reaction rate was calculated by using the differential extinction coefficient between acetyl-CoA and CoA of 3.2 mmol/L/cm. One unit of enzyme activity was defined as the amount of enzyme that catalyzed the formation of 1 µmol CoA/min. Protein concentrations were determined with the use of a Bio-Rad protein assay kit (Hercules, CA, USA), with bovine serum albumin as the standard protein.
|
Results and discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
Serine-binding SAT
Observing the determined SAT structure (1T3D), the SAT monomer is composed of two domains: residues 1–140 form eight -helices, and residues 141–262 form a left-handed ß-helical domain. The inhibitor cysteine is located between SAT monomers. The cysteine-binding pocket and key residues are shown in Figure 2A. In the cysteine-binding pocket, the sulfur atom, the carboxyl oxygen atoms and the amino nitrogen atom of the cysteine ligand form hydrogen bonds with His158 and His193, with Arg192, and with Asp92 and Asp157, respectively. Among these residues, Asp92, Asp157 and His158 belong to one monomer, while Arg192 and His193 belong to another monomer.
|
After simulated annealing, we obtained 10 structures of serine-binding SAT that have non-bond energies from –50 to –25 kcal/mol. The lowest energy structure is shown in Figure 2B. In this model, the carbonyl oxygens of the substrate serine form hydrogen bonds with Arg192, that the amide nitrogen hydrogen bonds to Asp92, and that His158 can be acyl transferred similar to the procedure described for chloramphenicol acetyltransferase (Hindson, 2003
). This structure can be considered a serine-binding SAT structure.
Next, we superimposed this serine-binding structure on a cysteine-binding structure and compared the two active site structures. Then, we measured the distances of Cs' positions. The distance of Asp157 Cs is 0.58 Å, the distance of His 158 Cs is 0.62 Å, the distance of His 193 Cs is 0.87 Å, the distance of Arg 192 Cs is 0.70 Å and the distance of Asp92 Cs is 1.25 Å. We thought the Cs are the basic skeleton of the protein structure, that can make changes of the affinity to a ligand. And we selected the most moving one, Asp92 as the primary target for mutation. Because Asp92 is in the middle of a short loop that has about eight residues, to make slightly moving of Asp92, we decided to substitute around this loop: that is, the Arg89 residue to the Asp96 residue (the black residues in Figure 3).
|
The feedback repression mutants, their activity, IC50 and Ki
As a result of the randomized, fragment-directed mutagenesis of the residues from Arg89 to Asp96, we succeeded in creating some serine acetyltransferase mutants which have both high-enzymatic activity as the wild-type and extreme insensitivity to inhibiton by L-cysteine. We show the mutants, their activities, their L-cysteine IC50 and their L-cysteine Ki in Table I; Table II shows the mutants' sequences. This mutagenesis shows that the high-desensitivity mutants are only the mutants of the residues at least three residues removed from Asp92. We considered that the distance of three residues is just as well of making slightly moving of Asp92 that is best suited to the slight difference of the structures of cysteine and serine. Table III shows the comparison with the catalytic properties of the previous studies.
|
|
|
We have studied X-ray structures of a novel acid phosphatase from Escherichia blattae and its complex with the transition-state analog molybdate (Ishikawa et al., 2000
). In that study, significant conformational change upon molybdate binding was observed around the His150 side chain, and consequently, the side chain of Leu70 etc. move remarkably toward the protein. Enhancement of nucleoside phosphorylation activity in this acid phosphatase was successful by mutation to these flexible residues (Ishikawa et al., 2002). Our experience in this study somewhat looks similar to that of this transphosphorylase case as mutation around flexible region is promising, although, in general, the region not well determined is not always flexible.
In summary, we succeeded in creating some serine acetyltransferase mutants which have both high enzymatic activity as the wild-type and extreme insensitivity to inhibiton by L-cysteine from the structure. We compared the calculated structure of L-serine-binding SAT with the crystal structure of L-cysteine-binding SAT. Because the inhibitor L-cysteine and the substarate L-serine have a very similar structure, we designed amino acid substitutions at positions 89–96 around Asp92, the C of which underwent a substantial positional change upon the replacement of L-cysteine by L-serine. We finally obtained the mutant SATs that have both high enzymatic activity as the wild-type and extreme insensitivity to inhibiton by L-cysteine.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
We thank H.Ito (Ajinomoto Co.) and N.Tsujimoto (Ajinomoto-Genetika Research Institute) for this collaboration between Ajinomoto Co. and Ajinomoto-Genetika Research Institute, and U.Tagami (Ajinomoto Co.) for the support of docking study.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
Denk,D. and Böck,A. (1987) J. Gen. Microbiol., 133, 515–525.
Hindson,V.J. (2003) Biochem. J., 375, 745–752.[CrossRef][Web of Science][Medline]
Hindson,V.J. and Shaw,W.V. (2003) Biochemistry, 42, 3113–3110.[CrossRef][Medline]
Ishikawa,K., Mihara,Y., Gondoh,K., Suzuki,E. and Asano,Y. (2000) EMBO J., 19, 2412–2423.[CrossRef][Web of Science][Medline]
Ishikawa,K., Mihara,Y., Shimba,N., Ohtsu,N., Kawasaki,H., Suzuki,E. and Asano,Y. (2002) Protein Eng., 15, 539–543.
Kredich,N.M. and Tomkins,G.M. (1966) J. Biol. Chem., 241, 4955–4965.
Kredich,N.M., Becker,M.A. and Tomkins,G.M. (1969) J. Biol. Chem., 244, 2428–2433.
Kuntz,I.D., Meng,E.C. and Shoichet,B.K.(1994) Acc. Chem. Res., 27, 117–123.
Kushner,S.R., Nagaishi,H. and Clark,A.J. (1972) Proc. Natl Acad. Sci. USA, 69, 1366–1370.
Moody,P.C.E. and Wilkinson,J.A. (1990) In Rickwood,E. (ed.), Protein Engineering in Focus. IRL Press, Oxford, UK.
Nakamori,S., Kobayashi,S., Kobayashi,C. and Takagi,H. (1998) Appl. Environ. Microbiol., 64, 1607–1611.
Nakamura,K., Hayama,A., Masada,M., Fukushima,K. and Tamura,G. (1987) Plant Cell Physiol., 28, 885–891.
Noji,M., Inoue,K., Kimura,N., Gouda,A. and Saito,K. (1998) J. Biol. Chem., 273, 32739–32745.
Pye,V.E., Tingey,A.P., Robson,R.L. and Moody,P.C.E. (2004) J. Biol. Chem., 279, 40729–40736.
Sambrook,J. and Russell,D.W. (2001) Molecular Cloning: A Laboratory Manual, 3rd ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York.
Smith,I.K. and Thompson,J.F. (1969). Biochem. Biophys. Res. Commun., 35, 939–945.[CrossRef][Web of Science][Medline]
Soda,K. (1987) An overview in Methods in Enzymology. Academic Press, New York.
Takagi,H., Awano,N., Kobayashi,S., Noji,M., Saito,K. and Nakamori,S. (1999a) FEMS Microbiol. Lett., 179, 453–459.[Web of Science][Medline]
Takagi,H., Kobayashi,C., Kobayashi,S. and Nakamori,S. (1999b) FEBS Lett., 452, 323–327.[CrossRef][Web of Science][Medline]
Received August 18, 2005; revised December 4, 2005; accepted January 3, 2006.
Edited by Haruki Nakamura
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||