PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Nat Chem Biol. Author manuscript; available in PMC Mar 18, 2014.
Published in final edited form as:
PMCID: PMC3957331
NIHMSID: NIHMS561456
Redesign of a mononuclear zinc metalloenzyme for organophosphate hydrolysis
Sagar D. Khare, Yakov Kipnis, Per Greisen, Jr., Ryo Takeuchi, Yacov Ashani, Moshe Goldsmith, Yifan Song, Jasmine L. Gallaher, Israel Silman, Haim Leader, Joel L. Sussman, Barry L. Stoddard, Dan S. Tawfik, and David Baker
Sagar D. Khare, Department of Biochemistry, University of Washington, Seattle, USA;
Contact Information: David Baker; dabaker/at/u.washington.edu
These authors contributed equally to this work: Sagar D Khare, Yakov Kipnis, Per Jr. Greisen
The ability to redesign enzymes to catalyze non-cognate chemical transformations would have wide-ranging applications. We developed a computational method for repurposing the reactivity of active site functional groups of metalloenzymes to catalyze new reactions. Using this method, we engineered a zinc-containing murine adenosine deaminase to catalyze the hydrolysis of a model organophosphate with a catalytic efficiency kcat/Km ~104 M−1s−1 after directed evolution. In the high-resolution crystal structure of the enzyme, all but one of the designed residues adopt the designed conformation. The designed enzyme efficiently catalyzed the hydrolysis of the RP-isomer of a coumarinyl analog of the nerve agent cyclosarin, and showed striking substrate selectivity for coumarinyl leaving-groups. Computational redesign of native enzyme active sites complements directed evolution methods and offers a general approach for exploring their untapped catalytic potential for new reactivities.
The redeployment of catalytic machinery in naturally occurring enzyme active sites for non-cognate reactions holds considerable promise for obtaining novel biocatalysts1,2. Alterations of substrate- and stereo-specificity, and the enhancement of pre-existing promiscuous activities of enzymes have been accomplished by library-based directed evolution approaches in several instances3. However, introducing completely new catalytic activities by exploiting native functional groups remains a challenge with these approaches because a starting point with some activity is typically required for directed evolution. Mechanistically related activities have been introduced into enzymes but these approaches require homologs with the desired activity to borrow sequence or structural features from412.
Organophosphate (OP) pesticides and nerve agents, introduced in the environment in the last century, are highly toxic threat agents and there is considerable interest in treating and mitigating their effects. Enzymes that rapidly hydrolyze these compounds with high proficiency when reacting with non-chiral OP pesticides, and with strong stereo-preference when reacting with the more toxic isomers of chiral nerve agents would be valuable as prophylactics and post-exposure treatments13. These enzymes may also serve as useful reagents for the bioremediation of contaminated sites. Naturally occurring enzymes with cognate or promiscuous OP hydrolysis activity have been evolved to target the toxic isomers of some OP compounds14,15.
Computational enzyme design methods have been used to alter enzyme substrate specificity16 and for designing catalysts for reactions that are not known to be catalyzed by natural enzymes1719. In de novo computational enzyme design, constellations of backbones that can support idealized active sites for the target reaction are identified in a set of protein scaffolds, and the sequence of the rest of the binding pocket is optimized for transition state affinity20. Because these efforts have been aimed at building active sites de novo, the reactivity of wild type functional groups already present in the scaffold proteins is not utilized in design.
Metal ions in enzyme active sites are highly reactive, and metal catalysis plays a key role in both enzymatic and abiological reactions. In particular, zinc ions serve as powerful catalysts in several hydrolase enzymes by (a) providing an activated water molecule for nucleophilic attack, (b) polarizing the scissile bond, (c) stabilizing the developing negative charge on the transition state, or a combination of the above21.
Here, we extend our de novo enzyme design methodology to exploit the reactivity of metal ions in existing enzyme active sites, and use this strategy to design an OP hydrolase starting from a functionally diverse set of mononuclear zinc-containing metalloenzyme scaffolds. Starting from a native adenosine deaminase with catalytic efficiency kcat/Km < 10−3 M−1s−1 for the hydrolysis of an OP substrate, we computationally engineered an OP hydrolase with catalytic efficiency kcat/Km ~104 M−1s−1 after activity maturation using directed evolution, a net >107-fold increase in activity.
Computational design of OP hydrolysis activity
We extracted a set of mononuclear zinc enzyme scaffolds, with diverse wild type functions and at least one open coordination site on the zinc atom from the Protein Databank (PDB). The requirement of at least one open zinc coordination site ensured that structural zinc sites were excluded from the scaffold set. The zinc coordination geometry in the scaffold set was either tetrahedral or trigonal bipyramidal.
The OP hydrolysis reaction proceeds by a SN2 mechanism such that the TS geometry is trigonal bipyramidal around the phosphorous center22. We obtained the bond length and bond angles around the phosphorous reaction center in the TS from previously reported gas-phase quantum mechanical calculations23 of the hydrolysis of paraoxon with the modification that the lengths of the bonds being formed and broken around the phosphorous center were set to 1.6Å and 2.0Å, respectively. We created a TS ensemble by sampling the torsional degrees of freedom in the R1, R2 and R3 substituents (Fig. 1) of the TS model while avoiding internal steric clashes. We superimposed models of the TS corresponding to two substrates, methyl paraoxon and DECP respectively, into the active site of each member of the scaffold set. In hydrolytic metallo-enzymes, metal ions act as activating agents for a hydroxyl ion nucleophile, or as Lewis acids for stabilizing the charge on the transition state, or both. Therefore, we generated three corresponding alignments (Fig 1a–c) of the TS model with the wild type metal center. In each case, we used RosettaMatch20 to search for additional hydrogen bonding interactions to the phosphoryl oxygen, the nucleophilic hydroxyl moiety, and the leaving group oxygen. In cases where at least two of these three hydrogen bonding interactions were found, we introduced shape complementary interactions to the TS using RosettaDesign24 (Fig. 1d). We maintained the coordination geometry around the metal ion during the design procedure by enforcing pseudo-covalent restraints between the metal ion and the protein sidechains coordinating the metal. We selected a set of twelve designed proteins (Table I) for experimental characterization based on the number of hydrogen bond interactions to the TS (more than 2), TS shape complementarity25 (Sc > 0.6), and the presence of a docking funnel26 ensuring that alternative orientations of the TS were higher in energy compared to the designed pose. We found that one of these twelve proteins, a redesigned adenosine deaminase (scaffold PDB accession code 1A4L), which we called PT3, hydrolyzed the substrate diethyl 7-hydroxycoumarinyl phosphate (DECP). While OP hydrolysis and adenosine deamination are both hydrolytic reactions (Fig 1e), the transition state geometry, leaving group character and the inherent reactivity of the substrate electrophilic center are quite distinct.
Figure 1
Figure 1
Computational active site redesign
Table I
Table I
List of designed proteins
Characterization and directed evolution of PT3
PT3 has 8 mutations compared to the parent adenosine deaminase – the substitutions F65W, A183I, F61T, D19S, L62I, I299V and D296A provide shape complementarity, and L58Q introduces a hydrogen bond to the nucleophile in the computational model (Fig. 1d). The zinc coordinating residues H15, H17, H214 and D295, and the catalytic residues E217 and H238 were retained from the wild type adenosine deaminase. Kinetic measurements showed that the activity of PT3 was ~7-fold higher than the buffer background with protein and substrate concentrations of 2.5 and 50 μM, respectively. The products of the reaction were identical to those obtained with the wild type bacterial phosphotriesterase enzyme (Fig. S1, Supplementary Information), but the catalytic efficiency of PT3, measured by Michaelis-Menten analysis, was modest, kcat/Km = 4 M−1s−1 (Fig. 2a). Wild type adenosine deaminase exhibited no acceleration of OP hydrolysis at <20 μM enzyme concentration, suggesting a kcat/Km below 10−3 M−1 s−1 (Fig. S2, Supplementary Information). The E217Q variant of an activity-matured variant of PT3 (PT3.1, see below) showed an OP hydrolysis activity with DECP identical to the wild type deaminase (Fig. 2b), indicating that E217 – a residue that is involved in proton shuttling in the deamination reaction27 – is also crucial for catalysis of the newly introduced OP hydrolysis reaction. The wild type murine adenosine deaminase enzyme belongs to the amidohydrolase superfamily28, and while some member enzymes of this superfamily catalyze OP hydrolysis, they use dinuclear rather than mononuclear metal sites29, and do not contain the substitutions introduced to obtain PT3.
Figure 2
Figure 2
Kinetic characterization of PT3
To gain insight into the features missing from the computational design protocol, we performed directed evolution experiments aimed at enhancing OP hydrolysis activity. We chose twelve positions – including five computationally designed positions, but excluding the putative catalytic residues – surrounding the active site for saturation mutagenesis, and screened variants at each position separately (see Methods). We found activity-enhancing mutations at three of the twelve positions. By recombining these substitutions, we obtained the variant PT3.1 (PT3 I62L, V218F, V299E) with a ~40-fold higher catalytic efficiency (Table II). The I62L substitution reverted a computationally designed residue to its wild type identity, V299E changed the identity of a computationally designed residue, and the largest single contributor to the enhancement, the V218F substitution (~20-fold increase), was at a position that was not considered during the computational design process.
Table II
Table II
Kinetic parameters for the hydrolysis of DECP for the wild type deaminase (1A4L), designed variant PT3, and variants identified by activity maturation
We subsequently introduced random mutations in PT3.1 using error-prone PCR, and found increased activity upon substitutions at three positions (S57, P59 and E186). Saturation mutagenesis at these positions led to the identification of the variant PT3.2 (PT3.1 S57D, P59K, E186D) with a ~10-fold higher catalytic efficiency compared to PT3.1 (Table II). Finally, modeling the interaction of the TS with the crystal structure of apo-PT3.1 (see below) suggested that Q58 was suboptimal for activity. We performed saturation mutagenesis at this position and identified the variant PT3.3 (PT3.2 Q58V) with a ~10-fold higher catalytic efficiency compared to PT3.2. The variant PT3.3 had a catalytic efficiency (kcat/Km =9750 M−1s−1) approximately 2500-fold higher than the initial PT3 variant (Table II). Product inhibition leading to low total turnover numbers can be an issue for metallohydrolases, but PT3 proceeded through multiple turnovers – at an enzyme concentration of 350 nM, PT3.3 completely hydrolyzed 50 μM DECP indicating >140 catalytic turnovers per enzyme molecule (Fig. 2c).
As the designed OP hydrolysis activity increased during directed evolution, there was a concomitant decrease in the wild type deamination activity. The variant PT3 had an adenosine deamination activity that was ~50,000-fold less than the wild type deaminase whereas in the evolved variant PT3.1 deamination activity was undetectable (Fig. S3, Supplementary information). To uncover the minimal set of mutations required to endow the deaminase with the designed OP hydrolysis activity, we reverted each position in PT3.3 to its wild type identity, one at a time (PT3toWT set). In parallel, starting from the wild type adenosine deaminase we generated a library of variants (WTtoPT3 set) in which the substitutions corresponding to PT3 were present in random combinations. We screened these libraries as cell lysates for OP hydrolysis activity with DECP, and sequenced select variants. In both cases, we found that four substitutions in the wild type deaminase (D19S, F61T, A183I, and D296A) were required to be simultaneously present for the emergence of OP hydrolysis activity. All variants with detectable DECP hydrolysis activity from the WTtoPT3 set (Table S2, Supplementary Information) had these four substitutions simultaneously.
There was striking spatial clustering (Fig. 3) of the mutations that enhanced PT3 activity – primarily by increasing kcat (Table II) – suggesting that these residues fine-tuned the alignment of the substrate in the active site and/or increased the pKa of E217 by increasing hydrophobicity (e.g. substitutions V218F and Q58V) around this putative catalytic base (Fig. 1d). This residue has an elevated pKa in the wild type deaminase27, and the calculated pKa of E217 in the crystal structure of the evolved variant PT3.1 (see below) was higher by 1.6 ΔpH units than that in the design model (Table S3, Supplementary Information). These results suggest that the increase in activity observed during the course of directed evolution was related to an increase in the basicity of E217.
Figure 3
Figure 3
Spatial clustering of wild type and activity-enhancing residues
Crystal structure of apo-PT3.1
We determined the crystal structure of the variant PT3.1 (PT3 I62L, V218F, V299E) to 2.35 Å resolution. In the crystal structure, the TIM-barrel fold of the deaminase was maintained, and the overall backbone conformation was quite similar (backbone RMSD 0.65 Å) to the PT3 design model (Fig. 4a). The zinc-binding site and other elements of the wild type catalytic machinery retained in the designed protein (e.g. residues H238, E217, D295) maintained their wild type conformations. We observed clear density for a zinc-bound water molecule and found that the trigonal bipyramidal zinc coordination geometry from the wild type was maintained in PT3.1. All computationally designed residues of PT3.1 except Q58 adopted rotamers predicted by the computational model (Figs. 4b and 4c).
Figure 4
Figure 4
Design model and crystal structure of apo-PT3.1
Two loops that border the active site and that connect sequential β–α elements within the core TIM barrel fold (β4–α4 and β5–α5, respectively) each displayed backbone conformational differences compared to the wild-type deaminase structure and to their predicted conformation in the original PT3 computational design model (Fig. 4a). Loop I, composed of residues 183–194, was displaced by approximately 2 Å away from the mouth of the β-barrel. This conformational difference involved only small alterations of backbone dihedral angles, such that the loop was displaced with little overall change in its peptide conformation, and likely resulted from electrostatic repulsion between the D185 and Q58 sidechains.
Loop II, composed of residues 217 to 221, contains the V218F mutation that provided a significant contribution (20-fold) to rate enhancement for the PT3.1 construct. The side chain for F218 was well ordered and was in close proximity to side chains of E217 and Q58. In contrast, the surrounding residues of loop II appeared to exhibit dynamic flexibility resulting in density consistent with two distinct conformations (one of which resembled the structure of the original wild-type deaminase). The side chain of Q58 was found in an alternate rotameric conformation relative to the original computational design, suggesting that the addition of V218F during directed evolution led to the inversion of the Q58 rotamer (Fig. 4d).
Substrate- and stereo-selectivity of PT3
Both the computational design and directed evolution of PT3 were performed with the non-chiral substrate DECP. To investigate the substrate specificity of the designed OP hydrolysis activity, we used several substrates, each with different substituents on the phosphate moiety (see Methods). Surprisingly, we found that PT3 efficiently hydrolyzed substrates with a coumarinyl leaving-group (DECP, DEPCyc and CMP-coumarin, Fig. S4 and Fig. 5) but did not hydrolyze substrates that have substituted phenyl-ring leaving-groups (e.g. paraoxon, 4-nitrophenyl diethyl phosphate), despite their similar intrinsic reactivity. We found that both paraoxon and 4-nitrophenol inhibited the reaction of PT3 with DECP (Ki~250μM and ~150μM, respectively; Fig. S5, Supplementary information), but there was no evidence of product inhibition with DECP (Fig. 2c). Docking calculations indicated that the smaller size of the nitrophenyl group allowed a greater variety of alternative binding modes for paraoxon in the active site pocket, greatly reducing the fraction of bound configurations productively aligned with the catalytic machinery (Fig. S6, Supplementary information). These alternative binding modes, presumably driven by the binding of the phenyl ring, also sterically precluded the simultaneous binding of DECP and may thereby inhibit the reaction of PT3 with DECP.
Figure 5
Figure 5
PT3 variants stereo-selectively catalyze the hydrolysis of a cyclosarin analog
The variants PT3.1-PT3.3 efficiently hydrolyzed the coumarinyl analog of the racemic nerve agent surrogate of cyclosarin, CMP-coumarin. PT3.3 had a marked stereo-preference for the RP isomer of CMP-coumarin, whereas the wild type deaminase showed no detectable activity with the either isomer of CMP-coumarin at the same concentration (Fig. 5a). Despite the pronounced stereo-preference of PT3.3 for the RP isomer, we also observed slow hydrolysis of the SP isomer (a RP/SP kinetic ratio >150). These results are consistent with the modeled alignment of CMP-coumarin in the PT3.3 active site (Fig. 5b). For the actual nerve agent cyclosarin, the more toxic SP-isomer was not hydrolyzed by any PT3 variant (data not shown). However, since the design model is consistent with the observed stereo-preference, and in view of the slow but detectable hydrolysis of the SP-isomer of the cyclosarin surrogate, it should be possible to use the same approach to explicitly design catalysts to hydrolyze the toxic SP-isomer of nerve agents.
The high reactivity and large diversity of reactions carried out by metalloenzymes makes them attractive starting points for the introduction of new activities. We chose zinc-containing enzymes as templates for computational design of OP hydrolysis activity because zinc performs diverse mechanistic roles in enzymes, is redox-stable and is used as a Lewis acid-type catalyst in many natural hydrolases21 including the dinuclear bacterial OP hydrolases. Computational design of mononuclear zinc-containing active sites successfully identified a set of mutations in an adenosine deaminase that endowed it with the target OP hydrolysis activity. We considered only the geometric compatibility of the TS with the active site structure during design, and did not use a priori knowledge about the wild type activity in scaffold identification. Because we found that four simultaneous mutations were required for the emergence of OP hydrolysis activity in the deaminase (Fig. 3), identification of the computationally designed variant by library screening would likely require generating and screening a library of an exceedingly large size. Having obtained the initial low-activity lead from computational design, activity levels were enhanced by approximately 2500-fold using directed evolution, demonstrating the power of computational design followed by directed evolution for enzyme redesign.
In contrast with the dinuclear zinc-containing bacterial phosphotriesterases, PT3 is a mononuclear zinc enzyme. In the dinuclear phosphotriesterases, one zinc ion polarizes the scissile P-O bond and the μ-hydroxyl moiety bridging the zinc ions acts as the nucleophile22. Because we observed only one free zinc coordination site in the crystal structure of apo-PT3, the zinc ion could act either to polarize the P-O bond (Fig. 1b) or to activate the nucleophile (Fig. 1a). In the wild type adenosine deaminase, the zinc ion activates the hydroxyl nucleophile27 whereas in the design model it polarizes the P-O bond along with the sidechain of H238, and E217 activates the nucleophile by functioning as a general base. The increased hydrophobicity around the E217 sidechain during directed evolution, the stereo-preference for the RP-isomer of CMP-coumarin (Fig. 5) and the lack of activity of the E217Q variant (Fig. 2b) of PT3 are consistent with the placement of the TS in the design model (Fig. 1b) and the role of E217 as a general base. The residue E217 shuttles protons in the wild type deaminase27 and metal ion polarization of the P-O bond has been proposed for the OP hydrolase diisopropyl fluorophosphatase which has a single calcium ion in the active site30. However, in the absence of a substrate- or TS-analogue bound structure of PT3, the possibility that the zinc ion and the residues H238 and D295 activate the nucleophile while the protonated form of E217 helps polarize the P-O bond (Fig. 1a) cannot be definitively ruled out.
Shortcomings of the computational enzyme design protocol31, and avenues for further improvement were highlighted by the sequence changes accrued during directed evolution and by comparing the crystal structure of the evolved variant PT3.1 to the design model. Most improvements in catalytic efficiency during directed evolution were increases in kcat, and the largest single increase arose from the substitution V218F (variant PT3.1). The sidechain at position 218 is not expected to directly contact the TS, but V218F increases the hydrophobic bulk around the catalytically critical E217 sidechain, which likely modulates its pKa and reactivity. Furthermore, the crystal structure of PT3.1 shows that the backbone structure of the loop composed of residues 217–221 needs to change conformation to accommodate the bulkier phenylalanine sidechain introduced by the V218F mutation. Since the design procedure is carried out on a fixed (wild type) backbone, the V218F substitution would cause steric clashes in the design model. Allowing backbone flexibility in the design procedure, and incorporating pKa effects using a more accurate electrostatic interaction model are avenues for improving computational enzyme design.
The computational active site redesign method described here is general and can be readily extended to obtaining biocatalytic leads for other reactions of interest. The method complements current library-based directed evolution approaches for enzyme engineering. Optimization of the computationally generated leads with directed evolution approaches has the potential to both significantly enhance the desired activities, and provide valuable feedback for incorporating key missing elements in the computational design methodology. The vast diversity of highly reactive catalytic microenvironments in structurally well-characterized natural enzymes represents a rich resource of reactivity for obtaining novel biocatalysts using computational enzyme redesign.
Protein expression, purification and kinetic measurements
Designed proteins were expressed in Escherichia coli BL21 (DE3) using the pET29b plasmid (Novagene), and purified over a Ni-NTA affinity column. Protein purity and integrity was verified by SDS-PAGE. For PT3 variants, protein concentration was estimated spectrophotometrically using a calculated molar extinction coefficient of 50,000 M−1cm−1. Reaction progress was monitored by following increase in fluorescence of 7-hydroxycoumarin product Excitation (365nm)/Emission(445nm) in 10 mM HEPES, 150 mM NaCl buffer (pH 7.5) at a protein concentration of 2.5 μM. Initial rates of DECP decomposition were determined at 10–500 μM of substrate and fit to the Michaelis-Menten equation equation M1 to obtain kcat and Km. Results of at least three independent measurements were used to calculate reported kinetic parameters and estimate errors.
Library construction and screening
Desired positions in the gene were changed or diversified using either the Kunkel method32 or QuickChange mutagenesis (Stratagene), using mutagenic oligonucleotides (IDT). Error prone PCR was performed using GeneMorphII kit (Stratagene) according to the manufacturer’s instructions. Genes were cloned into the pET29b expression vector, propagated, and constructed libraries were transformed into the E. coli BL21 (DE3) strain. Depending on the estimated complexity of the library 100–800 variants were assayed in screening experiments. Single colonies were grown in 96-deep-well plates. After cell lysis, activity of the protein variants was assayed in clarified lysate with 50 μM DECP. Improved variants were sequenced and mutations responsible for improvements were recombined before proceeding to the next round.
Crystal structure determination
After PT3.1 expression, purification and 6xHis tag cleavage (see Supplementary Information), the purified protein was concentrated to 0.2 mM in 20 mM HEPES-NaOH, 100 mM sodium glutamate (pH 7.5), 0.1 mM zinc acetate, and mixed with a 2.5-fold excess of sodium orthovanadate and 7-hydroxycoumarin. The crystal was obtained in 0.1 M Tris-HCl (pH 7.0), 0.2 M calcium acetate, and 20 % PEG 3000, and frozen by looping and submersion into liquid nitrogen. The crystal diffracted up to approximately 2.35 Å resolution at the ALS beam line 5.0.1. The data set was processed using HKL 2000 package 33, and a murine adenosine deaminase (D295E; PDB accession code1FKW) was used as a search model for molecular replacement. A single copy of the search model was found by PHASER34, and refined using REFMAC535. The final model was deposited in RCSB Protein Data Bank with ID code 3T1G. Statistics for the crystallographic data are shown in Table S4 (Supplementary Information). The anomalous difference Fourier map was used to conclude that PT3.1 contained a heavy metal atom in its active site (Fig. S7, Supplementary Information).
Determination of substrate-specificity
Substrate specificity of the PT3 variants was tested using 2,4-dinitrophenyl diethyl phosphate, 4-diethyl phosphate benzaldehyde, 4-nitrophenyl diethyl phosphate (paraoxon), 4-nitrophenyl dimethyl phosphate (methylparaoxon), and 7-O-diethylphosphoryl-3-cyano-7-hydroxycoumarin (DEPCyC). 0.5–1 mM of the OP compound were incubated with 5–10 μM of protein for 30 min. Change in absorbance or fluorescence at appropriate wavelength was recorded. Inhibitory effect of 2,4-dinitrophenyl diethyl phosphate, paraoxon or 4-nitrophenol on decomposition of DECP was studied by recording rate of hydrolysis of 50 μM DECP in the presence of 0.025–1 mM inhibitor. IC50 was calculated by fitting into equation equation M2.
Determination of stereo-preference with CMP-coumarin
The stereo-preference of the PT3 variants when reacting with chiral OPs was determined with racemic CMP-coumarin as substrate using the previously reported stereo-preference of two engineered mammalian serum paraoxonase variants, rePON1s, 3B3 and 3D814,36. The exact concentration of the CMP-coumarin stock solution was chemically determined by use of NaF as described earlier36. Approximately 10 μM of racemic CMP-coumarin were incubated with 2- to 100 nM PT3 variant in 50 mM Tris-100 mM NaCl, pH 8.0 at 25°C and the release of the leaving group 3-cyano-7-hydroxy-4-methylcoumarin was monitored by measuring absorbance at 400 nm. To identify the configuration of the non-hydrolyzed isomer in solution after the reaction approached an end-point, the mixture was spiked first with 0.1 μM 3B3 and 1 mM CaCl2 that induced the exclusive hydrolysis of residual RP isomer. Additional spiking with 0.1 μM 3D8 revealed the presence and concentration of the intact SP isomer. Detoxification of cyclosarin was carried out by monitoring the loss of anti-acetylcholine esterase potency as described previously14.
Supplementary Material
Supplementary Information
Acknowledgments
We thank Lucas Nivon for assistance with liquid chromatography, and Jiri Damborsky for comments on the manuscript. This work was supported by the Defense Advanced Research Projects Agency, Defense Threat Reduction Agency, and the Howard Hughes Medical Institute. PJG was supported by Novo Nordisk Danmark-Amerika Fondet and Oticon Fonden.
Footnotes
Competing financial interests
The authors declare no competing financial interests.
Author Contributions
SDK developed the computational method for active site redesign, performed computational design and kinetic characterization of PT1-PT12, analyzed the data and wrote the paper. YK designed and performed the directed evolution and library screening, wild type activity measurement, substrate selectivity and inhibition experiments, analyzed the data and wrote the paper. PJG implemented the computational redesign method, performed the computational design of PT1-PT12 and analyzed the data. RT determined the crystal structure of PT3.1. JLG expressed and purified the designed proteins PT1-PT12. YS performed pKa calculations. YA, MG, IS, HL and JLS synthesized nerve agents and nerve agent analogues, screened these with PT3, and determined its stereoselectivity. BLS performed structural analysis and wrote the paper. DST designed the experiments, and analyzed the data. DB designed the computational method and the experiments, analyzed the data, and wrote the paper.
Contributor Information
Sagar D. Khare, Department of Biochemistry, University of Washington, Seattle, USA.
Yakov Kipnis, Department of Biochemistry, University of Washington, Seattle, USA.
Per Greisen, Jr., Department of Physics, Technical University of Denmark, Lyngby, Denmark.
Ryo Takeuchi, Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, USA.
Yacov Ashani, Department of Structural Biology, Weizmann Institute of Science, Rehovot, Israel. Department of Neurobiology, Weizmann Institute of Science, Rehovot, Israel.
Moshe Goldsmith, Department of Biological Chemistry, Weizmann Institute of Science, Rehovot, Israel.
Yifan Song, Department of Biochemistry, University of Washington, Seattle, USA.
Jasmine L. Gallaher, Department of Biochemistry, University of Washington, Seattle, USA.
Israel Silman, Department of Neurobiology, Weizmann Institute of Science, Rehovot, Israel.
Haim Leader, Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot, Israel.
Joel L. Sussman, Department of Structural Biology, Weizmann Institute of Science, Rehovot, Israel.
Barry L. Stoddard, Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, USA.
Dan S. Tawfik, Department of Biological Chemistry, Weizmann Institute of Science, Rehovot, Israel.
David Baker, Department of Biochemistry, University of Washington, Seattle, USA. Howard Hughes Medical Institute, University of Washington, Seattle, USA. Biomolecular Structure and Design Program, University of Washington, Seattle.
1. Toscano MD, Woycechowsky KJ, Hilvert D. Minimalist active-site redesign: Teaching old enzymes new tricks. Angewandte Chemie-International Edition. 2007;46:3212–3236. [PubMed]
2. Gerlt JA, Babbitt PC. Enzyme (re)design: lessons from natural evolution and computation. Current Opinion in Chemical Biology. 2009;13:10–18. [PMC free article] [PubMed]
3. Khersonsky O, Tawfik DS. Enzyme Promiscuity: A Mechanistic and Evolutionary Perspective. Annual Review of Biochemistry. 2010;79:471–505. [PubMed]
4. Yin DL, et al. Switching Catalysis from Hydrolysis to Perhydrolysis in Pseudomonas fluorescens Esterase. Biochemistry. 2010;49:1931–1942. [PMC free article] [PubMed]
5. Terao Y, Miyamoto K, Ohta H. Introduction of single mutation changes arylmalonate decarboxylase to racemase. Chemical Communications. 2006:3600–3602. [PubMed]
6. Schmidt DMZ, et al. Evolutionary potential of (beta/alpha)(8)-barrels: Functional promiscuity produced by single substitutions in the enolase superfamily. Biochemistry. 2003;42:8387–8393. [PubMed]
7. Leopoldseder S, Claren J, Jurgens C, Sterner R. Interconverting the catalytic activities of (beta alpha)(8)-barrel enzymes from different metabolic pathways: Sequence requirements and molecular analysis. Journal of Molecular Biology. 2004;337:871–879. [PubMed]
8. Williams SD, David SS. A single engineered point mutation in the adenine glycosylase MutY confers bifunctional glycosylase/AP lyase activity. Biochemistry. 2000;39:10098–10109. [PubMed]
9. Xiang H, Luo LS, Taylor KL, Dunaway-Mariano D. Interchange of catalytic activity within the 2-enoyl-coenzyme a hydratase isomerase superfamily based on a common active site template. Biochemistry. 1999;38:7638–7652. [PubMed]
10. Park HS, et al. Design and evolution of new catalytic activity with an existing protein scaffold. Science. 2006;311:535–538. [PubMed]
11. Jochens H, et al. Converting an esterase into an epoxide hydrolase. Angew Chem Int Ed Engl. 2009;48:3532–5. [PubMed]
12. Chen B, et al. Morphing activity between structurally similar enzymes: from heme-free bromoperoxidase to lipase. Biochemistry. 2009;48:11496–504. [PubMed]
13. Raushel FM. Chemical biology: Catalytic detoxification. Nature. 2011;469:310–1. [PubMed]
14. Gupta RD, et al. Directed evolution of hydrolases for prevention of G-type nerve agent intoxication. Nat Chem Biol. 2011;7:120–5. [PubMed]
15. Tsai PC, et al. Stereoselective hydrolysis of organophosphate nerve agents by the bacterial phosphotriesterase. Biochemistry. 49:7978–87. [PMC free article] [PubMed]
16. Murphy PM, Bolduc JM, Gallaher JL, Stoddard BL, Baker D. Alteration of enzyme specificity by computational loop remodeling and design. Proc Natl Acad Sci U S A. 2009;106:9215–20. [PubMed]
17. Rothlisberger D, et al. Kemp elimination catalysts by computational enzyme design. Nature. 2008;453:190–U4. [PubMed]
18. Jiang L, et al. De novo computational design of retro-aldol enzymes. Science. 2008;319:1387–1391. [PMC free article] [PubMed]
19. Siegel JB, et al. Computational design of an enzyme catalyst for a stereoselective bimolecular Diels-Alder reaction. Science. 2010;329:309–13. [PMC free article] [PubMed]
20. Zanghellini A, et al. New algorithms and an in silico benchmark for computational enzyme design. Protein Science. 2006;15:2785–2794. [PubMed]
21. McCall KA, Huang C, Fierke CA. Function and mechanism of zinc metalloenzymes. J Nutr. 2000;130:1437S–46S. [PubMed]
22. Aubert SD, Li Y, Raushel FM. Mechanism for the hydrolysis of organophosphates by the bacterial phosphotriesterase. Biochemistry. 2004;43:5707–15. [PubMed]
23. Dyguda-Kazimierowicz E, Sokalski WA, Leszczynski J. Gas-phase mechanisms of degradation of hazardous organophosphorus compounds: do they follow a common pattern of alkaline hydrolysis reaction as in phosphotriesterase? J Phys Chem B. 2008;112:9982–91. [PubMed]
24. Kuhlman B, Baker D. Native protein sequences are close to optimal for their structures. Proc Natl Acad Sci U S A. 2000;97:10383–8. [PubMed]
25. Lawrence MC, Colman PM. Shape complementarity at protein/protein interfaces. J Mol Biol. 1993;234:946–50. [PubMed]
26. Davis IW, Baker D. RosettaLigand docking with full ligand and receptor flexibility. J Mol Biol. 2009;385:381–92. [PubMed]
27. Wang Z, Quiocho FA. Complexes of adenosine deaminase with two potent inhibitors: X-ray structures in four independent molecules at pH of maximum activity. Biochemistry. 1998;37:8314–24. [PubMed]
28. Pegg SC, et al. Leveraging enzyme structure-function relationships for functional inference and experimental design: the structure-function linkage database. Biochemistry. 2006;45:2545–55. [PubMed]
29. Afriat L, Roodveldt C, Manco G, Tawfik DS. The latent promiscuity of newly identified microbial lactonases is linked to a recently diverged phosphotriesterase. Biochemistry. 2006;45:13677–86. [PubMed]
30. Blum MM, Lohr F, Richardt A, Ruterjans H, Chen JC. Binding of a designed substrate analogue to diisopropyl fluorophosphatase: implications for the phosphotriesterase mechanism. J Am Chem Soc. 2006;128:12750–7. [PubMed]
31. Baker D. An exciting but challenging road ahead for computational enzyme design. Protein Sci. 2010;19:1817–9. [PubMed]
32. Kunkel TA. Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc Natl Acad Sci U S A. 1985;82:488–92. [PubMed]
33. Otwinowski Z, Minor W. Processing of X-ray diffraction data collected in oscillation mode. Macromolecular Crystallography, Pt A. 1997;276:307–326.
34. Mccoy AJ, et al. Phaser crystallographic software. Journal of Applied Crystallography. 2007;40:658–674. [PubMed]
35. Winn MD, Murshudov GN, Papiz MZ. Macromolecular TLS refinement in REFMAC at moderate resolutions. Methods Enzymol. 2003;374:300–21. [PubMed]
36. Ashani Y, et al. Stereo-specific synthesis of analogs of nerve agents and their utilization for selection and characterization of paraoxonase (PON1) catalytic scavengers. Chem Biol Interact. 2010;187:362–9. [PMC free article] [PubMed]