Increasing cleavage specificity and activity ofrestriction endonuclease KpnIKommireddy Vasu1 Easa Nagamalleswari1 Mai Zahran2 Petra Imhof3

Shuang-yong Xu4 Zhenyu Zhu4 Siu-Hong Chan4 and Valakunja Nagaraja15

1Department of Microbiology and Cell Biology Indian Institute of Science Bangalore 560 012 India2Department of Chemistry New York University NY USA 3Department of Physics Free University BerlinArnimallee 14 14195 Berlin Germany 4New England Biolabs Inc 240 County Road Ipswich MA 01938 USAand 5Jawaharlal Nehru Centre for Advanced Scientific Research Bangalore 560 064 India

Received May 13 2013 Revised July 11 2013 Accepted July 25 2013

ABSTRACT

Restriction enzyme KpnI is a HNH superfamilyendonuclease requiring divalent metal ions for DNAcleavage but not for binding The active site of KpnIcan accommodate metal ions of different atomicradii for DNA cleavage Although Mg2+ ion higherthan 500kM mediates promiscuous activity Ca2+

suppresses the promiscuity and induces highcleavage fidelity Here we report that a conservativemutation of the metal-coordinating residue D148 toGlu results in the elimination of the Ca2+-mediatedcleavage but imparting high cleavage fidelity withMg2+ High cleavage fidelity of the mutant D148E isachieved through better discrimination of the targetsite at the binding and cleavage steps Biochemicalexperiments and molecular dynamics simulationssuggest that the mutation inhibits Ca2+-mediatedcleavage activity by altering the geometry of theCa2+-bound HNH active site Although the D148Emutant reduces the specific activity of the enzymewe identified a suppressor mutation that increasesthe turnover rate to restore the specific activity of thehigh fidelity mutant to the wild-type level Our resultsshow that active site plasticity in coordinating differ-ent metal ions is related to KpnI promiscuousactivity and tinkering the metal ion coordination isa plausible way to reduce promiscuous activity ofmetalloenzymes

INTRODUCTION

Restriction endonucleases (REases) exhibit high sequencespecificity in substrate binding and use versatile DNA

cleavage mechanisms and thus are excellent modelsystems for understanding DNA recognition and phospho-diester bond hydrolysis Many REases studied so farbelong to the PD-(DE)XK superfamily of nucleases (1)However based on structural biochemical and sequenceanalyses it was shown that REases use not only the PD-(DE)XK fold but also unrelated catalytic domains foundin other classes of nucleases namely phospholipase D(PLD) GIY-YIG half pipe and HNH families (1ndash7) Ofthese five evolutionarily distinct active site motifs theHNH superfamily of REases forms the second largestgroup and is characterized by the presence of the bba-Mefinger fold (27) Nucleases of this superfamily are presentin all three kingdoms of life and exhibit broad specificityranging from sequence- or structure-specific to completelynon-specific to sequence or structure (89)

Studies carried out to date on HNH nucleases imply asingle metal ion-mediated catalysis (910) For hydrolysisa conserved His in the active site acts as a general base toactivate the nucleophilic water molecule which thenattacks the scissile phosphodiester bond The role of thecoordinated metal ion serves two purposesmdashstabilizationof the transition state and coordination of a watermolecule to protonate the leaving group (11) It hasbeen suggested that this single-metal ion mechanismimposes a less stringent requirement on the spatial coord-ination of the metal ion than in the two-metal ion mech-anism used by many PD-(DE)XK REases and thereforeallows HNH nucleases to exhibit a broader metal ion co-factor requirement (81213) In general HNH enzymesexhibit preference for metal ions of a particular type ieeither alkaline earth or transition group (101114ndash18)KpnI the first REase identified as a member of HNHsuperfamily however can use a battery of divalentmetal ion co-factors belonging to both alkaline earthand transition groups for DNA cleavage (19)

To whom correspondence should be addressed Tel +91 80 23600668 Fax +91 80 23602697 Email vrajmcbliiscernetinCorrespondence may also be addressed to Siu-Hong Chan Tel +1 978 380 7267 Fax +1 978 921 1350 Email channebcom

The authors wish to be known that in their opinion the Erst two authors should be regarded as joint First Authors

9812ndash9824 Nucleic Acids Research 2013 Vol 41 No 21 Published online 19 August 2013doi101093nargkt734

The Author(s) 2013 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby30) whichpermits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

Many REases show relatively low degree of promiscu-ous activity under standard reaction conditions in thepresence of Mg2+ but exhibit significant star activityunder altered conditions namely low ionic strengthelevated pH or when Mg2+ is substituted with Mn2+

(20) KpnI however exhibits significant promiscuousactivity in the presence of Mg2+ (21) Interestingly thepromiscuous cleavage is suppressed when the metal ionis replaced by Ca2+ in the presence of Ca2+ (ionicradius=114 A) the enzyme is specific even at high con-centrations of the enzyme or the metal ion whereas in thepresence of Mg2+ (ionic radius=086 A) KpnI exhibitspromiscuous activity even at low enzyme concentrations(21) Promiscuous activity has been proposed to offer anadaptive fitness to the host bacteria (22) and a gateway tothe evolution of new enzyme activities (2324) Thedramatic effect of metal ion cofactor on the KpnIactivity is intriguing and suggests that altering the activesite interactions with the metal ion could affect theenzymersquos specificity In our previous studies we reportedthat the mutation of several residues in KpnI altered itsmetal ion binding properties and enhanced cleavagefidelity (25ndash27) We describe here the generation of aKpnI variant with an extremely low promiscuity byintroducing a conservative mutation of a metal-coordinating residue The mutant also becomes inactivewith Ca2+ Although the high fidelity mutant showed alower specific activity a suppressor mutation thatenhances the specific activity of the enzyme has beenidentified Our results show that altering the metal ion-binding properties of KpnI and possibly other HNHREases can alter its substrate specificity and reinforcethe notion that the bound metal ion is correlated to thepromiscuous activity of the enzyme

MATERIALS AND METHODS

Enzymes chemicals and DNA

KpnI and its mutants were purified from Escherichia coliNEB Express (New England Biolabs) expressing the KpnIor mutant proteins from a plasmid under the control of alac promoter and MKpnI from a compatible plasmidunder the control of a tet promoter (unpublishedresults) About 9 g of wet cell paste were resuspended in100ml of Buffer A [20mM TrisndashHCl (pH 80) 50mMNaCl 1mM ethylenediaminetetraacetic acid (EDTA)10mM phenylmethylsulfonyl fluoride] and sonicatedAfter centrifugation at 15 000 rpm for 30min at 4C in aJA-17 rotor (Beckman Coulter) the supernatant wasloaded onto a Heparin HyperDM column (20ml Pall)Protein was eluted by a linear gradient of 50ndash1000mMNaCl in Buffer A over 20 column volumes at a flow rateof 2mlmin Peak fractions were pooled and diluted 4-foldinto 20mM potassium phosphate buffer (pH 70) 50mMNaCl 1mM EDTA (Buffer B) and then loaded onto aceramic hydroxylapatite column (7ml Bio-Rad) Proteinwas eluted using a linear gradient of 50ndash500mM potas-sium phosphate in Buffer B over 40 column volumes at aflow rate of 2mlmin Peak fractions were pooledconcentrated and buffer-exchanged to a buffer containing

40mM TrisndashHCl (pH 80) 500mM NaCl 1mM EDTAusing Vivaspin 15 (Sartorius Stedim Biotech) Afteraddition of equal volume of 100 glycerol theconcentrated proteins were stored at 20C Protein con-centration was determined by measuring its absorbance at280 nm and using a molar extinction coefficient of45 000M1 cm1 The purity of the proteins was gt95as judged by SDS-PAGE followed by Coomassie BluestainingThe enzymes were diluted in a buffer containing

20mM TrisndashHCl (pH 80) 25mM NaCl and 5mM2-mercaptoethanol for all the experimentsOligonucleotides (Sigma) were gel purified and 50-endlabeled using T4 polynucleotide kinase (New EnglandBiolabs) and [g-32P] ATP (6000Cimmol PerkinElmer)The labeled oligonucleotides were purified using G-25Sephadex spin column chromatography

Mutagenesis and screening for higher cleavage activity

Site-directed mutagenesis was carried out essentially asdescribed using a modified inverse PCR scheme (28) Tofind a suppressor mutation that can reverse the reducedactivity of the high fidelity mutant D148E selectedresidues were mutated to Ala using Vent DNA polymerase(New England Biolabs) Escherichia coli NEB Expresscarrying a plasmid expressing MKpnI was used as com-petent cells for transformation by the inverse PCRproducts Mutants with higher specific activity werescreened for as follows 1ml of overnight culture ofNEB Express carrying the mutants were harvestedsonicated and centrifuged Ten-fold dilutions of the super-natant were assayed using plasmid pXba (a pUC19 deriva-tive carrying a XbaI fragment of adenovirus DNA) assubstrate in NEBuffer 4 (New England Biolabs) at 37Cfor 1 h followed by 08 agarose gel electrophoresisMutation(s) were verified by DNA sequencing of theentire kpnIR gene

Intrinsic fluorescence

Fluorescence emission spectra were recorded using Jobin-Yvon fluorometer FluoroMax 3 (HORIBA Jobin-YvonSpex Division Longjumeau France) thermostated at25C The change in the tryptophan fluorescence wasmeasured at excitation and emission wavelengths of 295and 340 nm respectively with a 5 nm slit width For allthe studies proteins were dialyzed against EDTA toremove the intrinsically bound metal cofactor andEDTA was removed by dialysis against 10mM TrisndashHCl(pH 74) buffer EDTA-treated wild-type (WT) or mutantenzymes (2mM each) were incubated in a buffercontaining 10mM TrisndashHCl (pH 74) and 5mM2-mercaptoethanol with different concentrations(0ndash10mM) of Mg2+ or Ca2+ for 15min at 25C andthen fluorescence emission spectra were recordedControl titrations were recorded in the presence of amonovalent cation Na+ All fluorescence emissionspectra and fluorescence intensities from the titrationswere corrected by subtraction of control spectra andcontrol titrations

Nucleic Acids Research 2013 Vol 41 No 21 9813

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

Electrophoretic mobility shift assay for DNA binding

Different concentrations of KpnI WT and its mutantswere incubated with 50 end-labeled double-stranded oligo-nucleotides (1 nM) containing canonical or non-canonicalsites (20 bp) in binding buffer [10mM TrisndashHCl (pH 74)and 5mM 2-mercaptoethanol] for 15min on ice The freeDNA and the enzyme-bound complexes were resolved on8 non-denaturing polyacrylamide gel in TBE buffer(89mM TrisndashHCl 89mM boric acid and 1mM EDTA)Gels were visualized using a phosphorimager (FujifilmFLA5100)

Surface plasmon resonance spectroscopy

DNA binding kinetics study was carried out by usingsurface plasmon resonance (SPR) A streptavidin matrix-coated sensor chip (BIAcore) was preconditioned withthree consecutive injections of 1M NaCl and furtherequilibrated with binding buffer [10mM TrisndashHCl (pH74) 1mM EDTA 25mM NaCl] at a flow rate of 10 mlmin A 72mer 50 biotinylated oligonucleotide (50-ACTCTA GAG GAT CCC CGG TTG GAC TAA GTCCCC AGG CCC CTT AGC CAC CAC AAC GTGGGT ACC GAG CTC GAA-30) was annealed with itscomplementary strand and was in turn immobilized ontothe preconditioned streptavidin chip loaded into flow cell2 Flow cell 1 contained an unmodified streptavidin chipas a reference surface All binding studies were carried outin binding buffer using the BIAcore program KinjectIncreasing concentrations of the WT KpnI and itsmutants were passed on to the chip at a flow rate of50mlmin and change in Response Units was monitoredThe binding surface was regenerated each time by shortpulses of 5 ml of 005 SDS Data analysis was performedusing BIA evaluation software version 31 (BIAcore AB)using 11 Langmuir interaction with mass transfer modelAll SPR experiments were carried out at 25C

DNA cleavage and kinetic analysis

DNA cleavage reactions were carried out by incubatingWT or mutant KpnI proteins with pUC18 DNA (14 nM)(substrate carrying a single KpnI site) or 50 end labeledoligonucleotide duplexes (10 nM) (SupplementaryTable S1) in a reaction buffer containing 10mM TrisndashHCl (pH 80) and 2mM MgCl2 or CaCl2 at 37C for1 h The reactions were terminated using a stop dye con-taining 5 glycerol 10mM EDTA 0025 bromophenolblue and 0025 xylene cyanol The cleavage products ofplasmid DNA and oligonucleotides were analyzed on 1agarose and 12 urea polyacrylamide gel electrophoresisrespectively Chemical modification reactions were carriedout using diethylpyrocarbonate (DEPC) which modifieshistidine residues to N-carbethoxy histidine The WTenzyme and the mutants were pre-incubated with 5mMof CaCl2 or MgCl2 for 15min at 4C in a 4-(2- hydroxyethyl) piperazine ethane sulfonic acid (HEPES)-containingbuffer before incubated with different concentrations ofDEPC (25ndash100 mM) on ice for 2min and the treatedenzymes were used for DNA cleavage in the presence ofMg2+ The Ca2+chase reactions were carried out at a fixed

concentration of Mg2+ (2mM) and increasing concentra-tions of Ca2+ (0ndash10mM) The cleavage products wereanalyzed on 1 agarose gel electrophoresis FidelityIndex defined as the ratio of the maximum enzymeamount showing no star activity to the minimumamount needed for complete digestion at the canonicalrecognition site was determined as described (20) Forthe metal ion-dependence assays the reactions wereinitiated by adding substrate DNA and transferring thereaction mixture to 37C after incubating the enzymewith different concentrations of Mg2+ (001ndash10mM) onice for 5min The cleavage products of plasmid DNAand oligonucleotides were analyzed on 1 agarose geland 12 urea PAGE respectively

Kinetic analysis was carried out at DNA concentrationsof 75ndash250-fold molar excess over dimeric enzyme(1ndash10 nM) at 37C in an assay buffer containing 10mMTrisndashHCl (pH 74) 25mM NaCl 5mM2-mercaptoethanol and 2mM Mg2+ Aliquots weretaken out at different time points for the determinationof initial velocity The reactions were terminated byadding an equal volume of stop dye containing 95formamide 10mM EDTA 1mM NaOH 0025 xylenecyanol 0025 bromophenol blue The kinetic param-eters were determined by fitting the velocity with the sub-strate concentration using GraphPad Prism version 4 Theturnover number (kcat) was calculated as the ratio of Vmax

to the dimeric enzyme concentration used (29)

Molecular dynamics simulations

Four models of KpnI were created based on a homologymodel of KpnI complexed to its target DNA (4) KpnIWT with Ca2+ ions (WT-Ca2+) WT with Mg2+ ions (WT-Mg2+) and KpnI D148E mutant with Ca2+ ions (D148E-Ca2+) and Mg2+ ions (D148E-Mg2+) respectively Themetal ions have been placed at the same positions asobserved in the crystal structure of I-PpoI complexed toDNA (PDB accession 1A73) and crystal water moleculesobserved in the active site of I-PpoI have been added Themutant was modeled by replacing residue D148 with Gluin silico We performed two control simulations of I-PpoIwith Mg2+ions and Ca2+ions with the same settings as forKpnI

The systems are each solvated in a cubic water box of size(x=y=z=120 A) using the explicit TIP3P water model(30) Forty-two Na+counter ions were added to neutralizethe system and a further excess of Na+ and Cl- ions wasadded so as to obtain a physiological concentration of150mM NaCl The addition of the ions was carried outby random substitution of water oxygen atoms

The simulations were run using the program NAMD(31) with the CHARMM27 force field (32) Simulationswere performed using periodic boundary conditions andlong-range electrostatic interactions were treated using theParticle Mesh Ewald method (33) on a 128 128 128charge grid with a non-bonded cutoff of 12 A Shortrange electrostatics and van der Waals interactions weretruncated at 12 A using a switch function starting at 10 AThe solvated structures were minimized using 5000 stepsof the steepest descent followed by minimization with the

9814 Nucleic Acids Research 2013 Vol 41 No 21

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

conjugate gradient algorithm with solute atoms harmon-ically reconstrained until an energy gradient of 001 kcal(molA) was reached The system was then graduallyheated for 30 ps to 300K with 1K temperature stepswith harmonic restraints on the solute atomsEquilibration was performed in three stages with thenumbers of particles pressure (1 bar) and temperature(300K) kept constant (NpT isothermal-isobaricensemble) during a 75-ps window In the first 25 psvelocities were rescaled every 01 ps and in the second25 ps Langevin dynamics were used to maintainconstant temperature Pressure control was introducedin the third 25 ps and in the production run using theNose-Hoover Langevin piston (34) with a decay periodof 500 fs The harmonic restraints were gradually lifted[to 05 025 and 005 kcal(molA2)] in the three-equilibra-tion stages After equilibration the NPT production runswere performed for 20 ns The integration time step was2 fs and coordinates were saved with a sampling intervalof 2 ps All covalent bond lengths involving hydrogen werefixed using SHAKE algorithm The two outer base pairsof the DNA were harmonically restrained to have thecenter of mass of the atoms forming WatsonndashCrickhydrogen bonds of around 3 A As the starting modelonly represents the inner (DNA binding) part of theprotein restraints were put on the Ca-atoms of residuesA97-G141 D156-D163 and K179-N190 so as to mimic thestabilization by the missing residues All other atoms werefree to move All molecular images were generated withPymol [DeLano Scientific Palo Alto CA]

RESULTS

A conservative mutation of a metal ion-coordinatingresidue increases cleavage fidelity with Mg2+

The HNH active site of KpnI exhibits promiscuouscleavage activity in the presence of Mg2+ or high fidelitycleavage with Ca2+ (21) Sequence alignments structuralmodeling and mutational analysis showed that H149 is thefirst histidine of the HNH motif in the enzyme that mightfunction as a general base to activate a water molecule forthe nucleophilic attack on the scissile phosphodiester bond(48) D148 and Q175 correspond to the metal-bindingresidues that coordinate the active site metal ion whichstabilizes the pentacovalent phospho-anion intermediateand assists in the protonation of the leaving group(Figure 1A) Previous studies with the enzyme revealed acorrelation between ion size and promiscuous activity inthe case of alkaline earth group of metal co-factors (19) Ingeneral it was observed that metal ions of larger ionicradii induce high fidelity DNA cleavage To address therole of metal ion coordination on the promiscuousactivity we mutated residue D148 to Glu such that theside chain is extended by a ndashCH2 group without alteringits negative charge at neutral pH When the DNA cleavageproperty of the mutant D148E was analyzed in thepresence of 2mM Mg2+ the enzyme showed no promis-cuous activity even at high enzyme concentrations(Figure 1B) Fidelity index (FI) is defined as the ratio ofthe maximum amount of enzyme showing no star activity

to the minimum amount of enzyme needed for completedigestion at the canonical recognition site (20) The FI ofthe mutant is 4000 in NEBuffer 4 [20mM Tris acetate10mM magnesium acetate 50mM potassium acetate1mM DTT (pH 79)] compared with 4 for WT KpnIunder the same reaction conditions (data not shown)The mutation decreased the specific activity of the

enzyme on the canonical substrate by 6-fold (Table 1)We carried out steady state kinetic studies using the ca-nonical substrate to investigate the role of the mutation onthe catalysis The DNA cleavage reactions proceeded ac-cording to MichaelisndashMenten kinetics and the kinetic par-ameters were determined as described in lsquoMaterials andMethodsrsquo section The kinetic constants derived for theWT enzyme in the presence of Mg2+ were consistentwith the previously reported values (29) The KM valueof the mutant decreased 2-fold from the WT value of174 to 90 nM whereas its kcat value decreased 10-foldfrom 022 to 002 resulting in a decrease of kcatKM

value by 4-fold (Table 2)

Mutation D148E abolishes cleavage activity with Ca2+

Surprisingly the mutant D148E did not show detectableDNA cleavage activity in the presence of 2mM Ca2+

(Supplementary Figure S1) To determine whether theD148E mutation abrogated Ca2+ binding Ca2+ or Mg2+

were titrated into mutant D148E and intrinsic fluores-cence of the protein was measured Figure 2A showsthat the mutant binds with similar affinity to both Mg2+

and Ca2+ but the affinity is lower compared with the WTsuggesting that the increase in the length of the side chainaffects metal binding to some extent but does not abrogateCa2+ binding We have shown earlier that the change inthe intrinsic fluorescence of the protein is mainly due tothe metal ion binding to the HNH active site (25)Consistent to the comparable binding affinity of Ca2+

Figure 1 Mutant D148E exhibits high fidelity DNA cleavage(A) HNH motif of KpnI The KpnI monomer model is shown as aribbon with helices as spirals and strands as arrows The Mg2+ andZn2+ ions are shown as spheres The side chains of active site residuesD148 H149 and Q175 are shown as sticks and labeled (B) Effect ofD148E mutation of KpnI on DNA cleavage specificity Plasmid DNApUC18 (14 nM) was incubated with various concentrations (25 50 and100 nM) of WT or mutant enzyme for 1 h at 37C in the presence of5mM Mg2+ and analyzed on 1 agarose gel C indicates DNAcleavage reaction in the absence of enzyme Nc L and Sc indicatethe positions of nicked circular linear and supercoiled forms of theplasmid respectively The asterisk denotes the promiscuous DNAcleavage products

Nucleic Acids Research 2013 Vol 41 No 21 9815

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

andMg2+ Ca2+could replace theMg2+bound at the HNHactive site resulting in the loss of activity (Figure 2B)These results suggest that Ca2+ ion binds to the active siteof the D148E mutant as strongly as Mg2+ but forms acleavage-incompetent enzyme-metal ion complexTo investigate whether the Ca2+-bound D148E mutant

enzyme has a different DNA binding property electro-phoretic mobility shift assays (EMSA) were carried outusing 50-end labeled 20 bp oligonucleotide duplexes con-taining a KpnI site Ca2+ neither enhanced nor decreasedthe enzyme affinity for the canonical recognition sequence(Supplementary Figure S2) indicating that the metal ioncould be affecting the next step ie the DNA cleavage stepWe next tested whether the binding of Ca2+ affects the

solvent exposure of H149 (3637) through DEPC modifi-cation of histidine residues In these experiments westarted with a low DEPC concentration to limit the modi-fication to the active site histidine residue When D148EKpnI was pre-incubated with Ca2+ and then treated with50 mM of DEPC the DNA cleavage activity with Mg2+

was inhibited (Figure 2C) Cleavage activity of WT KpnIwas not affected at the same DEPC concentrations sug-gesting that in mutant D148E the general base H149becomes more solvent exposed when Ca2+ is boundHowever in the presence of Mg2+ H149 was protectedfrom DEPC inactivation in both WT and D148E(Figure 2D) The likely increase in solvent accessibilityof H149 in the mutant suggests that the mutationinduces a different Ca2+ coordination geometry not per-missive to the formation of an active phosphodiester bondhydrolysis configuration

Molecular basis for high fidelity of KpnI D148Ewith Mg2+

Next we investigated the molecular basis of the highfidelity achieved by the D148E mutant with Mg2+ Tounderstand the effect of the increase in the side chainlength at the 148 position on the Mg2+-mediatedcleavage characteristics of the enzyme the metal ion

activation profile of the mutant D148E was comparedwith the WT enzyme The proteins were pre-incubatedwith increasing concentrations of Mg2+ and the reactionswere initiated by the addition of the substrate DNA TheWT enzyme exhibited complete cleavage of 14 nM ofpUC18 DNA at Mg2+ concentrations as low as 50 mMHowever mutant D148E required 4-fold higher Mg2+

ion concentration for complete cleavage (Figure 3A)Both the DNA binding and the cleavage steps can

account for the fidelity of the REases Increased discrim-ination of canonical from non-canonical sites and therelative rate of cleavage of the canonical and non-canon-ical sites determine the fidelity of an enzyme We thereforecarried out detailed kinetic studies for the binding andcleavage steps of WT and D148E KpnI EquilibriumDNA-binding experiments using the canonical site in theabsence of metal ions revealed that mutant D148E bindsto the canonical site with an affinity of 2 nM comparedwith very poor binding at a non-canonical site (GaTACC)(Figure 3B) This is in contrast to Kd values obtained forthe WT enzyme which binds to both canonical and non-canonical sites with high affinity of 9 and 13 nM respect-ively (29) The differential binding of mutant D148E to thecanonical and non-canonical sequences indicates a role forthe residue D148 in target sequence discrimination Athigher enzyme concentrations mutant D148E exhibitedweak binding to the non-canonical DNA (Figure 3B)but the complex is resistant to cleavage (Figure 3C) sug-gesting that the mutation also suppresses promiscuousactivity at the catalytic step

Molecular dynamics simulations show changes in activesite geometry

To understand how the mutation D148E alters the activesite geometry molecular dynamics simulations were per-formed for both Ca2+ and Mg2+ ions bound to WT (WT-Ca2+ and WT-Mg2+) and the mutant (D148E-Ca2+ andD148E-Mg2+) respectively The simulations were per-formed based on the theoretical model of dimerizedKpnI HNH active site bound to the target DNA (4)Representative snapshots of the catalytic pocket withinthe active sites of WT-Ca2+ WT-Mg2+ D148E-Ca2+andD148E-Mg2+are shown in Figure 4A and B respectivelyA superimposition of the WT and mutant D148E boundto Ca2+ or Mg2+ is shown in Figure 4C and D respect-ively The distance between the ions and key residuesalong the simulation time is given in SupplementaryFigure S3

Replacing Ca2+ with Mg2+ in the WT leads to smallchanges in the active site geometry and in particular tothe coordination of the metal ion From the WT-Ca2+

simulations the metal ion is found to be coordinated bysix atoms throughout the simulation time [coordinationnumber (CN)=6] one oxygen atom of the phosphategroup of the scissile phosphodiester bond of the DNA(Cyt13) one carboxyl oxygen atom of D148 and threewater molecules throughout the simulation time(Supplementary Figure S3) A sixth coordination site isfilled with Q175 occasionally replaced by a fourth watermolecule Simulations for the WT-Mg2+ complex indicate

Table 2 Kinetic parameters for canonical DNA cleavage

Enzyme KM (nM) kcat (s1) kcatKM (s1M1)

(106)WT 174 022 1264plusmn04D16N 569 121 2126plusmn02D148E 90 002 322plusmn06D16ND148E 669 109 1631plusmn06

Table 1 Specific activity of the WT and mutants

Enzyme Specific activity(unitsmg)

WT 12 107

D148E 18 106

D16N 28 107

D16ND148E 12 107

9816 Nucleic Acids Research 2013 Vol 41 No 21

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

no change in CN but Q175 coordinates the metal ionthroughout the course of the simulation In the controlsimulation of I-PpoI with Ca2+ the correspondingresidue N119 was also observed to transiently leave themetal ion coordination sphere whereas it remained closeto Mg2+ in the entire simulation (Supplementary FigureS4) The distance between the metal ion and H149(measured at the putative proton-acceptor nitrogen-atom) fluctuates 45 A in both WT simulations adistance that allows one water molecule in between Oursimulations showed that mutation D148E introduces asignificant difference of the catalytic pocket between thetwo models with different metal ions and to the respectiveWT conformation (Figure 4C and D) The mutatedresidue Glu148 coordinates the Ca2+ ion with bothcarboxyl oxygen atoms throughout the simulation timein both subunits and coordinates the Mg2+ in onesubunit The extra coordination by the second Gluoxygen atom results in a CN of 7 whereas in the

D148E-Mg2+ simulations a CN of 6 is maintained byhaving Q175 leaving the metal ion In the D148E-Ca2+

simulations the coordination by Q175 is lost andreplaced by a fourth water molecule (SupplementaryFigure S3) All other coordination bonds fluctuate 2 Aindicating strong and consistent interactions between theDNA Glu148 and the water molecules bound to the Ca2+

or Mg2+ ion

A suppressor mutation increases the specific activity ofmutant D148E

Although the D148E mutation resulted in an enzyme withhigher specificity it had lower specific activity and kcatKM

than the WT (Tables 1 and 2) We therefore set out toidentify suppressor mutations that can reverse thereduced specific activity while retaining high fidelity Bytargeted mutagenesis in the D148E background eightmutants were found to have higher specific activity (datanot shown) Surprisingly DNA sequencing revealed that

Figure 2 Ca2+ ion binds to KpnI D148E but does not support DNA cleavage (A) Fluorescence emission spectra of WT and mutant D148E in thepresence of Ca2+ and Mg2+with increasing amounts of CaCl2 (Blue) or MgCl2 (Red) (0ndash10mM) Representative spectra and the calculated Kd valuesare shown (B) Metal ion competition assay Reactions were carried out by assaying Mg2+-mediated DNA cleavage activity in the presence oftitrating Ca2+ ions Reactions contained 25 nM D148E and competition was performed by pre-incubation of the enzyme in a buffer containing10mM TrisndashHCl (pH 74) on ice with the indicated metal ions The reactions were initiated by addition of pUC18 DNA (14 nM) and incubated at37C for 1 h Enhanced solvent accessibility of His149 in the presence of Ca2+ KpnI WT or mutant D148E enzymes were pre-incubated with 5mMof (C) CaCl2 or (D) MgCl2 and treated with increasing concentrations (25ndash100 mM) of DEPC Residual activity was assayed by cleavage of pUC18DNA (14 nM) in the presence of 2mM MgCl2

Nucleic Acids Research 2013 Vol 41 No 21 9817

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

Figure 3 Molecular basis for the high fidelity of variant D148E (A) Mg2+-dependent DNA cleavage of mutant D148E Different concentrations ofmetal ions (0ndash10mM) were incubated with 1 unit of WT or mutant D148E in a buffer containing 10mM TrisndashHCl (pH 74) on ice for 5min DNAcleavage reaction was initiated by the addition of (i) pUC18 DNA (14 nM) or (ii) end labeled canonical oligonucleotide (10 nM) and incubated at37C for 1 h (iii) Graphical representation of the Mg2+ activation profile of WT and D148E variant on the oligonucleotide substrate Mutant D148Erequires a Mg2+ concentration of 200mM for the 90 digestion of the substrate compared with 50 mM required by WT showing that the decrease inbinding affinity toward Mg2+ (Figure 2A) directly affects the cleavage activity of KpnI (B) Canonical versus non-canonical discrimination by mutantD148E EMSA was carried to determine the binding affinity of mutant D148E to end-labeled canonical (GGTACC) and non-canonical (GaTACC)substrates over a range of enzyme concentrations between 0 and 256 nM The intensity of the shifted DNA band is expressed as percentage bound(C) Mutant D148E does not cleave star substrate at binding concentrations End-labeled oligonucleotide duplex containing the 50-GaTACC-30

sequence (10 nM) was incubated with WT (25ndash10 nM) or mutant D148E (100ndash500 nM) at 37C for 30min in the presence of 2mM Mg2+ Thecleavage products were analyzed on 12 urea-PAGE Lane C contains the substrate DNA without the enzyme S and P represent the substrate andproduct respectively

9818 Nucleic Acids Research 2013 Vol 41 No 21

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

all the higher activity mutants had acquired an additionalmutation D16N suggesting that the higher specificactivity was due to the D16N substitution but not thetargeted mutations

Site-directed mutagenesis was carried out to generatemutants D16N and D16ND148E In the case ofD16ND148E double mutant the D16N suppressormutation did not affect the other two properties of theD148E mutant namely the high substrate specificity inMg2+catalyzed reaction and the Ca2+-mediated inhibitionof the cleavage activity (See Figure 1B and SupplementaryFigure S1) The increase in the activity of the enzymecaused by mutation D16N could be the result of compen-sational interactions with the D148E mutation orenhanced activity independent of the D148E mutation ora combination of both To delineate the mechanism theDNA cleavage properties of the suppressor D16N wasevaluated in the context of WT and D148E mutation

Mutation D16N causes faster turnover

To better understand the role of the mutation D16N inimproving specific activity we carried out steady state

kinetic analysis of mutants D16N D148E D16ND148Eand the WT enzyme using the canonical substrate(Table 2) The KM values of the mutants D16N andD16ND148E were 569 and 669 nM respectivelycompared with 174 and 90 nM of the WT and mutantD148E respectively The kcat values of mutants D16Nand D16ND148E were 121 and 109 s1 respectivelycompared with 022 and 002 s1 of the WT and mutantD148E respectively Compared with the WT enzyme theKM value of mutant D16N increased by 3-fold whereasthe kcat value increased by 6-fold (Table 2) Whencompared with mutant D148E the KM value of mutantD16ND148E increased by 7-fold whereas the kcatvalues increased by 37-fold resulting in 5-fold im-provement in kcatKM value When the substrate concen-tration is higher than the KM value as in most commonlyused cleavage conditions the D16ND148E enzymecleaves the canonical site more efficiently than the D148Emutant owing to its higher kcat value Thus introduction ofD16N into the D148E background restored the kcatKM

value in the double mutant to the WT level (Table 2)To analyze the molecular basis for the increased specific

activity the DNA-binding property of the mutant

Figure 4 Molecular dynamics simulations for the interactions of KpnI WT and mutant D148E with Mg2+ and Ca2+ Representative snapshots ofthe catalytic pocket within the active sites of KpnI WT bound to Ca2+ and Mg2+ ion (A) and mutant D148E bound to Ca2+ and Mg2+ (B) Asuperimposition of WT and mutant D148E bound to Ca2+ or Mg2+ is shown in (C) and (D) respectively

Nucleic Acids Research 2013 Vol 41 No 21 9819

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

proteins was assessed by EMSA As shown in Figure 5irrespective of their cleavage efficiency all mutants boundto the canonical sequence with equilibrium bindingaffinity comparable with that of the WT enzyme Tofurther investigate the basis for the higher turnovernumber of the D16N mutation the association and dis-sociation rates of WT mutants D16N D148E and D16ND148E with the canonical recognition sequence weredetermined (Figure 6) At the association step the WTenzyme and the mutant D148E exhibited comparable as-sociation rate (lsquoonrsquo rate) In contrast the D16N andD16ND148E mutants exhibited a 7- and 28-fold fasterlsquoonrsquo rates respectively compared with the WT enzymeindicating that the D16N mutation increases the rate ofassociation with the substrate DNA At the dissociationstep when compared with WT enzyme the D148E andD16ND148E mutants exhibited 32- and 18-fold fasterdissociation from the canonical DNA complex respect-ively In addition to a faster association rate the D16Nmutant exhibited a slower lsquooffrsquo rate from the substrateshowing that the D16N mutation does not only speedup the formation of the KpnI-canonical DNA complexbut also stabilizes the complexLastly limited proteolysis analysis of the WT D16N

D148E and D16ND148E was carried out to examinewhether the enzymes have undergone major structuralchange From Supplementary Figure S5 it is apparentthat the trypsin cleavage patterns of the mutants weresimilar to that of WT indicating that the increase inactivity and sequence specificity is not due to major struc-tural change to the protein

DISCUSSION

In this study we show that a conservative substitution ofAsp to Glu in the metal ion coordinating residue D148 ofthe catalyticHNH motif of KpnI results in changes in twomajor aspects in its DNA cleavage activitymdasha drasticincrease in cleavage fidelity with Mg2+ and eliminationof cleavage activity with Ca2+ We show that the highfidelity cleavage was contributed by the enhanced discrim-ination of non-canonical sequences from the canonical atthe binding step of the enzyme and at the catalytic stepCa2+-mediated cleavage activity of the D148E mutant wasabolished at the catalytic step Molecular dynamics simu-lation data suggest that when Ca2+ is coordinated by Gluinstead of Asp it is more tightly bound (as indicated bythe higher CN) and hence less likely to move toward theO30 atom of the leaving group and facilitate its departureby compensating the negative charge The D148Emutation decreases the turnover rate and kcatKM valueof the enzyme whereas the D16N mutation reverts thekcatKM value to the WT level through a higher turnoverrateGenerally Ca2+ ions do not support the cleavage

activity of PD-(DE)-XK REases However it is welldocumented that members of the HNH superfamily ofnucleases exhibit a wider metal ion preference and manyof them can use Ca2+ ion as a cofactor (111014ndash18)although the molecular basis of Ca2+-mediated DNA

cleavage is not completely understood In the case ofKpnI Ca2+ not only facilitates cleavage but also induceshigh cleavage fidelity (21) The binding of Ca2+ to KpnImutant D148E however results in a catalytically inactivecomplex (Figure 2A and B and Supplementary Figure S1)Our results show that when a Ca2+ ion is bound to theD148E mutant the side chain of H149 becomes moreaccessible to DEPC modification than the Ca2+-boundWT or the Mg2+-bound mutant D148E We interpretthe DEPC inactivation of mutant D148E by Ca2+ asmis-positioning of H149 the general base of the HNHactive site (4) such that it cannot activate a nucleophilicwater molecule for the cleavage reaction to occurAlternatively the ordered water molecule that acts as anucleophile in the active site might be excluded due to theincreased length of the acidic side chain of D148E when aCa2+ ion is bound Both of these mechanisms of inhibitionhave been proposed for other enzymes (35) For examplein the case of HincII and E coli ribonuclease H1 it wasobserved that Ca2+ coordinates the general base requiredfor catalysis (3637) Comparison of the Mg2+versus Ca2+

bound structure of the E coli phosphoenolpyruvatecarboxykinase revealed that the binding of a Ca2+ ionresults in the exclusion of two water molecules at theactive site (38) The increased accessibility of H149 toDEPC modification in mutant D148E favors our hypoth-esis that in D148E KpnI H149 is mis-positioned such thatit cannot act as a general base to activate the nucleophilicwater molecule for cleavage to occur

Based on kinetic structural andor biochemical studiesthree possible mechanisms have been proposed to explainpromiscuous DNA cleavage by REases In the case ofEcoRI it has been shown that an overall increase inbinding affinity of the enzyme to canonical non-canonicalor non-specific DNA sequence coupled with increasedcleavage rate constants enhances the probability ofcleavage at non-canonical sites (39) Studies with BamHIand its isochizomer OkrAI have suggested that a struc-tural element the presence of a flexible C-terminal helix(as seen in BamHI) or its absence (in OkrAI) facilitatesthe formation of a non-canonical cleavage competentcomplex (40) Our previous studies have shown thatbinding of an additional Mg2+ to KpnI induces promiscu-ous activity (26) Mechanistically a higher degree of rec-ognition specificity can be the result of the enzymersquosdiscrimination of non-canonical sequences from the ca-nonical one at the binding cleavage or both steps Fromthe results presented here it is evident that the high fidelityvariant D148E exhibits reduced binding at non-canonicalsites The mutant D148E does bind to the substrateGaTACC at high enzyme concentrations but does notcut (Figure 3B and C) suggesting that the mutation alsosuppresses the catalytic step when a non-canonical sub-strate is bound It is possible that the binding of thesecond Mg2+ ion required for inducing promiscuousactivity (2627) is hampered in the mutant The changein the Mg2+ activation profile and the binding affinitytoward Mg2+ and Ca2+ (Figure 3A) suggests thatmutation D148E alters the metal ion coordination appar-atus of the enzyme Molecular dynamics simulations ofthe distances between the metal ions Ca2+ and Mg2+ and

9820 Nucleic Acids Research 2013 Vol 41 No 21

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

atoms of active site residues (Figure 4 and SupplementaryFigure S3) agree with this proposition In a typical one-metal ion mechanism both the activation of the watermolecule that becomes the nucleophile (by donating aproton to an acceptor residue such as a nearby histidine)and the compensation of the negative charge accumulatingat the leaving group oxygen atom are achieved by thesame metal ion The leaving group can become protonatedby a second metal-activated water molecule or directlycoordinated to the metal ion on dissociation of the P-O30 bond In either case the metal ion must movesomewhat from the attacking site to the departure siterequiring a subtle balance between metal ion mobilityand coordination Our molecular dynamics simulationsshow that the CN of the metal ion increases from 6 to 7when Ca2+ is bound to mutant D148E The higher CNmay prevent ion movement affecting the catalysis Oursimulation also indicated that mutation D148E introducessignificant changes of the active site pocket and the modeof Mg2+ ion coordination with respective to the WT con-formation (Figure 4D and Supplementary Figure S3)How these changes in the D148E-Mg2+ complex arerelated to the reduced specific activity and better discrim-ination against non-canonical substrates More detailedsimulations and biochemical experiments will be neededto investigate this issue

Although D148E eliminates cleavage activity with Ca2+

and promiscuous activity with Mg2+ and retains similaraffinity to both the metal ions we have shown previouslythat mutating D148 to Gly eliminated KpnIrsquos cleavageactivity with Mg2+ by preventing the binding of themetal ion to the active site The latter mutant showedlower cleavage activity with non-canonical sites withMn2+ and lower ds cleavage activity toward canonicalsubstrate leading to the accumulation of nicked inter-mediates (25) These properties of D148 mutants under-score the important role of the residue in the plasticity ofthe metal ion cofactor selection and the enzymersquos promis-cuous activity and suggest a correlation between thebound metal ion cofactor and promiscuous activity inKpnIIt is noteworthy that other residues or motifs outside of

the HNH motif also play a role in the cleavage fidelity andCa2+-mediated activity of KpnI First mutation of residueD163 to Ile resulted in the loss of Ca2+-mediated activityand high cleavage fidelity with Mg2+ (2627) Secondmutation in a separate motif ExDxD (E132 D134 andD136) which is involved in Ca2+ coordination reducedthe Ca2+-mediated cleavage activity of canonical substrateand Mg2+-mediated promiscuous activity (27) Althoughit is known that Ca2+ coordinates with a larger number ofamino acid residues compared with Mg2+due to its larger

Figure 5 Suppressor mutant exhibits comparable DNA-binding affinity EMSA analyses for the DNA-binding affinity of WT KpnI and mutantsDifferent concentrations of WT or KpnI mutants (0ndash256 nM) were incubated with 1 nM of end-labeled canonical (-GGTACC) oligonucleotide inbinding buffer [20mM TrisndashHCl (pH 74) 25mM NaCl and 5mM 2-mercaptoethanol] on ice for 15min before loading onto a 8 polyacrylamidegel The intensity of the shifted DNA band is expressed as percentage bound

Nucleic Acids Research 2013 Vol 41 No 21 9821

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

atomic radius structural information is required for abetter insight into the distinct features of different metalion-mediated activity in KpnIThe higher kcat values of mutants D16N and D16N

D148E (compared with WT and mutant D148E respect-ively) indicate that the D16N mutation increases theturnover of the enzyme Without an atomic structure ora reliable structural model of the N-terminus of KpnI weare not able to suggest a molecular mechanism of how theD16N mutation increases the turnover of the enzyme onits canonical substrate The higher on rate and lower off

rate of the mutant D16N toward the cognate substratesuggests that the D16N mutation alters the DNA-binding interactions Recent studies with acylphos-phatases have shown that removal of a salt-bridge nearthe active site decreases the activation barrier of the cata-lytic reaction (41)

To conclude the present study describes a mutationthat results in altered metal ion coordination and elimin-ation of promiscuous activity for restriction enzyme KpnITogether with our previous studies we propose that theplasticity of the KpnI active site for metal ion cofactor is

Figure 6 Binding kinetics to the canonical site DNA-binding kinetics of WT KpnI and the mutants was assayed by SPR spectroscopy Biotinylatedoligonucleotide harboring the canonical (-GGTACC-) site was immobilized onto a streptavidin chip The proteins were passed over the DNA withrestricted flow rates to measure the association and dissociation rates as described in lsquoMaterials and Methodsrsquo section (A) Sensorgrams of WTD16N D148E and D16ND148E Graphical representation of the association (B) and dissociation (C) rates of the proteins is shown

9822 Nucleic Acids Research 2013 Vol 41 No 21

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

associated with its promiscuous activity Substitution(s) ofresidues that are involved in co-factor coordination in theenzymes could be a general strategy to improve the sub-strate specificity of HNH REases and possibly other HNHnucleases Demonstration that spontaneous mutationsdistal to the active site of the enzyme influence the rateof catalysis opens up further avenues for enzyme engin-eering Approaches discussed here for the generation ofvariant enzymes would thus form the basis for futurestudies The tremendous effect of a single pointmutation on the abolition of non-canonical DNAcleavage but retention of high sequence specificity empha-sizes the role of promiscuous activity in divergence offunction while maintaining native interactions

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online

ACKNOWLEDGEMENTS

The authors thank P Uma Maheswari for technical as-sistance Dr Saravanan and VN laboratory members forcritical reading of the manuscript and William Jack forhelpful comments KV was a recipient of ResearchAssociate Fellowship from the Indian Institute ofScience ENM is a recipient of Senior ResearchFellowship from Council of Scientific and IndustrialResearch Government of India VN is a JC Bosefellow of the Department of Science and TechnologyGovernment of India

FUNDING

Work carried out at the Nagaraja lab in India Institute ofScience was supported by the JC Bose fellowship grantfrom Department of Science and TechnologyGovernment of India Work carried out at New EnglandBiolabs was supported by NEB Funding for open accesscharge New England Biolabs will pay for the publicationcharges

Conflict of interest statement Several of the authors(SHC ZZ and SYX) are employees of NewEngland Biolabs the KpnI enzymes are commercialproducts sold by that company

REFERENCES

1 PingoudA FuxreiterM PingoudV and WendeW (2005) TypeII restriction endonucleases structure and mechanism Cell MolLife Sci 62 685ndash707

2 AravindL MakarovaKS and KooninEV (2000) Survey andsummary holliday junction resolvases and related nucleasesidentification of new families phyletic distribution andevolutionary trajectories Nucleic Acids Res 28 3417ndash3432

3 SapranauskasR SasnauskasG LagunaviciusA VilkaitisGLubysA and SiksnysV (2000) Novel subtype of type IIsrestriction enzymes BfiI endonuclease exhibits similarities to theEDTA-resistant nuclease Nuc of Salmonella typhimurium J BiolChem 275 30878ndash30885

4 SaravananM BujnickiJM CymermanIA RaoDN andNagarajaV (2004) Type II restriction endonuclease RKpnI is a

member of the HNH nuclease superfamily Nucleic Acids Res32 6129ndash6135

5 MiyazonoKI WatanabeM KosinskiJ IshikawaK KamoMSawasakiT NagataK BujnickiJM EndoY TanokuraMet al (2007) Novel protein fold discovered in the PabI family ofrestriction enzymes Nucleic Acids Res 35 1908ndash1918

6 BujnickiJM RadlinskaM and RychlewskiL (2001) Polyphyleticevolution of type II restriction enzymes revisited two independentsources of second-hand folds revealed Trends Biochem Sci 269ndash11

7 OrlowskiJ and BujnickiJM (2008) Structural andevolutionary classification of Type II restriction enzymes based ontheoretical and experimental analyses Nucleic Acids Res 363552ndash3569

8 PommerAJ CalS KeebleAH WalkerD EvansSJKuhlmannUC CooperA ConnollyBA HemmingsAMMooreGR et al (2001) Mechanism and cleavage specificity ofthe H-N-H endonuclease colicin E9 J Mol Biol 314 735ndash749

9 BiertumpfelC YangW and SuckD (2007) Crystal structure ofT4 endonuclease VII resolving a Holliday junction Nature 449616ndash620

10 KeebleAH HemmingsAM JamesR MooreGR andKleanthousC (2002) Multistep binding of transition metals tothe H-N-H endonuclease toxin colicin E9 Biochemistry 4110234ndash10244

11 PommerAJ KuhlmannUC CooperA HemmingsAMMooreGR JamesR and KleanthousC (1999) Homing in onthe role of transition metals in the HNH motif of colicinendonucleases J Biol Chem 274 27153ndash27160

12 YangW (2008) An equivalent metal ion in one- and two-metal-ion catalysis Nat Struct Mol Biol 15 1228ndash1231

13 WittmayerPK and RainesRT (1996) Substrate binding andturnover by the highly specific I- PpoI endonucleaseBiochemistry 35 1076ndash1083

14 KuWY LiuYW HsuYC LiaoC-C LiangP-HYuanHS and ChakK-F (2002) The zinc ion in the HNHmotif of the endonuclease domain of colicin E7 is not requiredfor DNA binding but is essential for DNA hydrolysis NucleicAcids Res 30 1670ndash1678

15 GalburtEA ChevalierB TangW JuricaMS FlickKEMonnatRJ and StoddardBL (1999) A novel endonucleasemechanism directly visualized for I-PpoI Nat Struct Biol 61096ndash1099

16 DrouinM LucasP OtisC LemieuxC and TurmelM (2000)Biochemical characterization of I-CmoeI reveals that this H-N-Hhoming endonuclease shares functional similarities with H-N-Hcolicins Nucleic Acids Res 28 4566ndash4572

17 ChanS-H OpitzL HigginsL OrsquoloaneD and XuS-Y (2010)Cofactor requirement of HpyAV restriction endonuclease PLoSOne 5 e9071

18 KriukieneE (2006) Domain organization and metal ionrequirement of the Type IIS restriction endonuclease MnlI FEBSLett 580 6115ndash6122

19 VasuK SaravananM and NagarajaV (2011) EndonucleaseActive Site Plasticity Allows DNA Cleavage with DiverseAlkaline Earth and Transition Metal Ions ACS Chem Biol 6934ndash942

20 WeiH TherrienC BlanchardA GuanS and ZhuZ (2008)The Fidelity Index provides a systematic quantitation of staractivity of DNA restriction endonucleases Nucleic Acids Res 36e50ndashe50

21 ChandrashekaranS SaravananM RadhaDR and NagarajaV(2004) Ca2+-mediated site-specific DNA cleavage and suppressionof promiscuous activity of KpnI restriction endonuclease J BiolChem 279 49736ndash49740

22 VasuK NagamalleswariE and NagarajaV (2012) Promiscuousrestriction is a cellular defense strategy that confers fitnessadvantage to bacteria Proc Natl Acad Sci USA 109 E1287ndash93

23 KhersonskyO and TawfikDS (2010) Enzyme promiscuity amechanistic and evolutionary perspective Annu Rev Biochem79 471ndash505

24 Rockah-ShmuelL and TawfikDS (2012) Evolutionarytransitions to new DNA methyltransferases through target siteexpansion and shrinkage Nucleic Acids Res 40 11627ndash11637

Nucleic Acids Research 2013 Vol 41 No 21 9823

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from

25 VasuK SaravananM RajendraBVRN and NagarajaV(2010) Generation of a manganese specific restrictionendonuclease with nicking activity Biochemistry 49 8425ndash8433

26 SaravananM VasuK and NagarajaV (2008) Evolution ofsequence specificity in a restriction endonuclease by a pointmutation Proc Natl Acad Sci USA 105 10344ndash10347

27 NagamalleswariE VasuK and NagarajaV (2012) Ca2+binding to the ExDxD motif regulates the DNA cleavagespecificity of a promiscuous endonuclease Biochemistry 518939ndash8949

28 ChiuJ MarchPE LeeR and TillettD (2004) Site-directedLigase-Independent Mutagenesis (SLIM) a single-tubemethodology approaching 100 efficiency in 4 h Nucleic AcidsRes 32 e174ndashe174

29 SaravananM VasuK KanakarajR RaoDN andNagarajaV (2007) RKpnI an HNH superfamily REase exhibitsdifferential discrimination at non-canonical sequences in thepresence of Ca2+ and Mg2 Nucleic Acids Res 35 2777ndash2786

30 JorgensenWL ChandrasekharJ MaduraJD ImpeyRW andKleinML (1983) Comparison of simple potential functions forsimulating liquid water J Chem Phys 79 926ndash935

31 PhillipsJC BraunR WangW GumbartJ TajkhorshidEVillaE ChipotC SkeelRD KaleL and SchultenK (2005)Scalable molecular dynamics with NAMD J Comput Chem 261781ndash1802

32 MacKerellAD BanavaliN and FoloppeN (2001) Developmentand current status of the CHARMM force field for nucleic acidsBiopolymers 56 257ndash265

33 DardenT YorkD and PedersenL (1993) Particle mesh EwaldAn N log(N) method for Ewald sums in large systems J ChemPhys 98 10089ndash10092

34 EvansDJ and HolianBL (1985) The NosendashHoover thermostatJ Chem Phys 83 4069ndash4074

35 YangW (2010) Nucleases diversity of structure function andmechanism Quart Rev Biophys 44 1ndash93

36 EtzkornC and HortonNC (2004) Ca2+ binding in the activesite of HincII implications for the catalytic mechanismBiochemistry 43 13256ndash13270

37 BabuCS DudevT CasarenoR CowanJA and LimC (2003)A combined experimental and theoretical study of divalentmetal ion selectivity and function in proteins application toE coli ribonuclease H1 J Am Chem Soc 125 9318ndash9328

38 TariLW MatteA PugazhenthiU GoldieH andDelbaereLTJ (1996) Snapshot of an enzyme reactionintermediate in the structure of the ATPndashMg2+ndashoxalate ternarycomplex of Escherichia coli PEP carboxykinase Nat Struct Biol3 355ndash363

39 LesserD KURPIEWSKIM and Jen-JacobsonL (1990) Theenergetic basis of specificity in the Eco RI endonucleasemdashDNAinteraction Science 250 776ndash786

40 VanameeES ViadiuH ChanS-H UmmatA HartlineAMXuS-Y and AggarwalAK (2011) Asymmetric DNArecognition by the OkrAI endonuclease an isoschizomer ofBamHI Nucleic Acids Res 39 712ndash719

41 LamSY YeungRCY YuT-H SzeK-H and WongK-B(2011) A rigidifying salt-bridge favors the activity of thermophilicenzyme at high temperatures at the expense of low-temperatureactivity PLoS Biol 9 e1001027

9824 Nucleic Acids Research 2013 Vol 41 No 21

at Indian Institute of Science on Novem

ber 25 2013httpnaroxfordjournalsorg

Dow

nloaded from