Journal of Structural Biology 188 (2014) 249258Contents lists available at ScienceDirect

Journal of Structural Biology

journal homepage: www.elsevier .com/locate /y jsbiCrystal structure of the VapBC-15 complex from Mycobacteriumtuberculosis reveals a two-metal ion dependent PIN-domain ribonucleaseand a variable mode of toxinantitoxin assemblyqhttp://dx.doi.org/10.1016/j.jsb.2014.10.0021047-8477/ 2014 Elsevier Inc. All rights reserved.

q Database: The crystal structure of VapBC-15 complex has been deposited inwwPDB with accession code 4CHG. Corresponding authors at: School of Life Sciences, Jawaharlal Nehru University,

New Delhi 57, India. Tel.: +91 11 2670 4513 (S. Gourinath). Department ofBiophysics, All India Institute of Medical Sciences, New Delhi 29, India. Tel.: +91 112659 4240 (A. Srinivasan).

E-mail addresses: sgourinath@mail.jnu.ac.in (S. Gourinath), srini@aiims.ac.in(A. Srinivasan).Uddipan Das a, Vivian Pogenberg b, Udaya Kumar Tiruttani Subhramanyam c, Matthias Wilmanns b,Samudrala Gourinath d,, Alagiri Srinivasan a,aDepartment of Biophysics, All India Institute of Medical Sciences, New Delhi, Indiab European Molecular Biology Laboratory, Hamburg, GermanycCentre for Structural Systems Biology (CSSB), DESY Campus, Hamburg, Germany and Forschungszentrum Jlich GmbH, Juelich, Germanyd School of Life Sciences, Jawaharlal Nehru University, New Delhi, Indiaa r t i c l e i n f o

Article history:Received 15 July 2014Received in revised form 29 September2014Accepted 6 October 2014Available online 22 October 2014

Keywords:ToxinantitoxinVapBC-15VapC-15ToxinVapB-15Mycobacterium tuberculosisX-ray crystallographya b s t r a c t

Although PIN (PilT N-terminal)-domain proteins are known to have ribonuclease activity, their specificmechanism of action remains unknown. VapCs form a family of ribonucleases that possess a PIN-domainassembly and are known as toxins. The activities of VapCs are impaired by VapB antitoxins. Here we pres-ent the crystal structure of the VapBC-15 toxinantitoxin complex from Mycobacterium tuberculosisdetermined to 2.1 resolution. The VapB-15 and VapC-15 components assemble into one heterotetramer(VapB2C2) and two heterotrimers (VapBC2) in each asymmetric unit of the crystal. The active site of VapC-15 toxin consists of a cluster of acidic amino acid residues and two divalent metal ions, forming a wellorganised ribonuclease active site. The distribution of the catalytic-site residues of the VapC-15 toxinis similar to that of T4 RNase H and of Methanococcus jannaschii FEN-1, providing strong evidence thatthese three proteins share a similar mechanism of activity. The presence of both VapB2C2 and VapBC2emphasizes the fact that the same antitoxin can bind the toxin in 1:1 and 1:2 ratios. The crystal structuredetermination of the VapBC-15 complex reveals for the first time a PIN-domain ribonuclease protein thatshows two metal ions at the active site and a variable mode of toxinantitoxin assembly. The structurefurther shows that VapB-15 antitoxin binds to the same groove meant for the binding of putative sub-strate (RNA), resulting in the inhibition of VapC-15s toxicity.

2014 Elsevier Inc. All rights reserved.1. Introduction

Toxinantitoxin (TA) loci were first discovered on prokaryoticplasmids as factors that prevent segregational loss in low copynumber plasmids (Gerdes et al., 1986). These loci are ubiquitousin microbial genomes (Pandey and Gerdes, 2005), yet their biolog-ical role remains poorly understood. TA genes are arranged in abicistronic operon (Anantharaman and Aravind, 2003) thatencodes a stable toxin and a labile antitoxin: the toxin targetsdiverse cellular machineries and the antitoxin binds and restrainsthe action of the toxin (Gerdes, 2000). Most of these antitoxins pos-sess the ability to bind to their own promoters, thus regulatingtheir own expression (Gerdes et al., 1986). During unfavourableconditions such as stress and nutrient deprivation, cellular prote-ases degrade the antitoxins and thus allow the free toxins to slowdown the growth of the cell or kill the cell (Hayes, 2003).

The TA systems are classified into type I, II, III and IV accordingto the type of the antitoxin in the TA complex (Blower et al., 2011;Fozo et al., 2010; Leplae et al., 2011; Wang et al., 2012). VapBC is amember of the type II TA system, which also includes MazEF,RelBE, ParDE, HigBA, CcdBA, HipBA and Doc/PhD. The VapC,MazF, RelE, HipA and Doc families are known to inhibit translation(Christensen and Gerdes, 2003; Cline et al., 2012; Gazit and Sauer,1999; Korch and Hill, 2006; Zhang et al., 2003). ParE and CcdB actby inhibiting DNA gyrase (Dao-Thi et al., 2005; Jiang et al., 2002)and HipA is a protein kinase that acts on the elongation factor

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jsb.2014.10.002&domain=pdfhttp://dx.doi.org/10.1016/j.jsb.2014.10.002mailto:sgourinath@mail.jnu.ac.inmailto:srini@aiims.ac.inhttp://dx.doi.org/10.1016/j.jsb.2014.10.002http://www.sciencedirect.com/science/journal/10478477http://www.elsevier.com/locate/yjsbi

250 U. Das et al. / Journal of Structural Biology 188 (2014) 249258EF-Tu (Korch and Hill, 2006). Mycobacterium tuberculosis (Mtb)harbours 88 putative TA loci, which belong to five out of the eightfamilies of type II TA identified so far (Ramage et al., 2009).Mycobacterium smegmatis, a non-pathogenic strain, contains onlyone loci in contrast to Mtbs large and diverse repertoire of VapBCs.This marked difference between organisms raises questions aboutthe role of such a large number of VapBC loci in the pathogenicityof Mtb.

VapC proteins (Virulence associated proteins C) belong to thePIN-domain family of proteins, which are known to exhibit ribonu-clease activity (Arcus et al., 2011). Upon expression, these VapCproteins function as toxins which is deleterious to the cell, butcan be inhibited by VapB antitoxins. PIN-domain proteins, includ-ing VapC, have low sequence homology with one another, yetretain a catalytic core that is functionally similar (Arcus et al.,2005). This catalytic core is orchestrated by a cluster of acidicamino acid residues that take part in the hydrolysis of the phos-phodiester bonds of RNA. The active site is also marked by thepresence of divalent metal ions necessary for the catalysis to occur;the catalysis may depend on either one or two metal ions in theactive site. A bioinformatic analysis predicted that PIN-domainproteins are structurally homologous with various exonucleasessuch as T4 phage RNase H (Mueser et al., 1996), Methanococcusjannaschii (Mja) FEN-1(Hwang et al., 1998) and the exonucleasedomain of Taq DNA polymerase (Eom et al., 1996). Although somestructural and biochemical data are available for the VapC toxins,their precise physiological roles in Mtb need to be explored.

Structural insights are provided by a number of VapBC proteinsfrom a broad range of organisms. The first VapBC structurereported was that of VapBC-5 from Mtb (Miallau et al., 2009). InVapBC-5, the toxin VapC-5 is a PIN-domain protein inhibited bya tightly interacting VapB-5 antitoxin. Although no metal ion wasseen, the putative active site of VapC-5 was proposed to be homol-ogous to the nuclease domain of FEN-1 endonuclease. Subse-quently three more VapBC structures were reported: VapBC-3from Mtb (Min et al., 2012), VapBC from Shigela flexneri (Sfl)(Dienemann et al., 2011) and VapBC2 from Rickettsia felis (Rfe)(Mat et al., 2012). The VapBC-3 from Mtb has a PIN-domainassembly with one Mg2+ ion at the active site of the toxin, confirm-ing the ribonuclease catalytic site. The VapBC2 from Rfe was boundto its promoter, providing strong evidence that complexes of theVapBC family can bind to their own promoters and regulate theirown expression.

Although PIN-domain proteins are known to have ribonucleaseactivity (Arcus et al., 2011), the general mechanism of substraterecognition and catalysis is still unknown. Here we present thecrystal structure of the VapBC-15 complex from M. tuberculosis.Our VapC-15 is the first PIN-domain protein to have been visual-ized with two divalent metal ions at the active site, and mightfunction in a similar way as does T4 RNase H and Mja FEN-1. Theoverall structure shows a variable mode of toxinantitoxin assem-bly and inhibition. From our crystal structure we are able to showfor the first time how the same antitoxin can bind to the toxin in a1:1 and 1:2 ratios and inhibit the active site in different modes.2. Materials and methods

2.1. Cloning and co-expression of vapB-15 and vapC-15 genes

The co-expression of Rv2009, which encodes VapB-15 (anti-toxin) and Rv2010, which encodes VapC-15 (toxin) was carriedout using a modified pET28b vector. vapB-15 and vapC-15 geneswere cloned in the MCS 2 (between NdeI and XhoI) and MCS 1(between NcoI and HindIII) sites respectively of the pETDuet-1 vec-tor. The pETDuet-1 vector was then reamplified and digested withNcoI and XhoI and the entire fragment (vapC-15-T7 promoter-vapB-15) was cloned between the NcoI and XhoI sites of the pET28bvector with vapB-15 gene terminating with a C-terminal (His)6 tag.The pET28b-vapBC-15 construct was confirmed by DNA sequenc-ing and transformed into E. coli BL21(DE3) cells for proteinexpression.

2.2. Protein purification

The transformed E. coli cells were grown in autoinductionmedia (Studier, 2005) supplemented with 100 mg/L kanamycin.The cultures were grown initially at 37 C for 3 h until the OD600reached 0.8, after which the temperature was lowered to 24 Cand the cells were grown for additional 24 h for expression.

All purification steps were carried out at 4 C. The cells wereharvested and resuspended in binding buffer containing 20 mMTrisHCl (pH 7.4), 500 mM NaCl, 10 mM imidazole and 5 mMb-mercaptoethanol to which complete EDTA-free proteaseinhibitor cocktail (Roche) was added. The cell lysis was performedby lysozyme treatment (0.1 mg ml1 for 30 min on ice) followed byultrasonication. The lysate was centrifuged at 30,000 g for 45 minat 4 C. The clarified supernatant was filtered through 0.45 lmfilters and passed through a TALON metal affinity (Clontech) resinpre-equilibrated with binding buffer. The columnwas washed with25 column volumes of washing buffer containing 20 mM TrisHCl(pH 7.4), 500 mM NaCl, 20 mM imidazole and 5 mM b-mercap-toethanol to remove non-specifically bound proteins. The proteincomplex was eluted with the elution buffer containing 20 mMTrisHCl (pH 7.4), 300 mM NaCl, 150 mM imidazole and 5 mMb-mercaptoethanol. Protein purity was checked using 12.5%SDSPAGE and the identity of the protein was confirmed usingMALDI-TOF. The pooled fractions were concentrated to 2 ml usingAmicon Ultra 10-kDa cut-off filters (Millipore) and further purifica-tion was performed by size exclusion chromatography using aHiLoad 16/60 Superdex 75 column (GE Healthcare) pre-equili-brated with buffer containing 20 mM TrisHCl (pH 7.4), 100 mMNaCl and 5 mM b-mercaptoethanol. The peak fractions werecollected and analysed by SDSPAGE and protein concentrationwas determined spectroscopically (OD280).

Protein was derivatized with selenomethionine (SeMet) accord-ing to the protocol of Van Duyne et al. (1993). BL21 (DE3) cells har-bouring pET28b-vapBC-15 were grown overnight at 37 C in 40 mlof M9 minimal media. The following day, 4 L M9 minimal mediawas inoculated with the overnight culture and cells were grownat 37 C for 8 h until the OD600 reached 0.8. At the desired OD,the temperature was lowered to 24 C and 100 mg/L of lysine,phenylalanine, threonine and 50 mg/L of isoleucine, leucine, valineand selenomethionine were added to the culture. After 15 min,protein expression was induced using a final concentration of1 mM isopropyl b-D-1-thiogalactopyranoside (IPTG) for 24 h atthe same temperature. Protein was purified following the sameprotocol as described for the native protein.

2.3. Crystallisation and data collection

The VapBC-15 protein complex was concentrated to 7 mg/mlfor the initial crystallisation trials. Crystals were obtained by thesitting-drop vapour-diffusion method using a Mosquito liquid-dispensing robot (TTP labtech) in 96-well crystallisation plates.The native crystals of VapBC-15 appeared within three weeks at16 C in drops containing 200 nl of protein complex and 200 nl ofa precipitant solution consisting of 100 mM Na-HEPES (pH 7.0),8% PEG 8000 and 10 mM b-mercaptoethanol. SeMet VapBC-15crystals appeared after approximately two months in the sameconditions. Both the native and SeMet crystals were cryoprotectedin a solution containing 25% PEG 400 in addition to the mother

Table 1Data collection and refinement statistics.

Data collection SeMet VapBC-15 Native VapBC-15

Source P13, DESY BM14, ESRFSpace group P212121 P212121

Cell dimensionsa, b, c () 86.2,131.2,149.3 85.6,131.1, 148.9a, b, c () 90,90,90 90, 90, 90Wavelength () 0.9773 0.9788Resolution () 60.092.1 (2.12

2.1)a38.922.56(2.652.56)

Rmerge 0.13 (0.64) 7.5 (35.9)CC(1/2) 0.998 (0.845)I/rI 14.1 (4.5) 34.12 (4.1)Completeness (%) 99.1(99.8) 99.6 (99.7)Redundancy 12.6(13.1) 7.4 (7.2)B-factor from Wilson plot

(2)25.3 48.18

RefinementResolution () 2.1 2.56No. reflections 189,140 52,927Rwork/Rfree 16.7/19.9 (26.0/

31.8)21.9/26.2 (26.3/32.9)

No of Residues 928 928Ions 8 8Water 551 125

B-factors (2)Protein 33.1 53Ions 23.4 47.5Water 38.7 51.1

R.M.S. deviationsBond lengths () 0.007 0.009Bond angles () 1.02 1.15

Ramachandran plot (%)Favoured regions 97.4 97Outliers 0 0

a Values in parenthesis are of highest resolution shell.

U. Das et al. / Journal of Structural Biology 188 (2014) 249258 251liquor and then flash frozen in liquid nitrogen before data collec-tion. The native dataset was collected to 2.56 resolution atBM14 beamline of European Synchrotron Radiation Facility,Grenoble (Das et al., 2013). For the SeMet VapBC-15 proteincrystal, a single-wavelength anomalous dispersion (SAD) datasetwas collected to 2.1 resolution at the Se peak wavelength of0.9733 at the P13 beamline of the European Molecular BiologyLaboratory at DESY, Hamburg.

2.4. Structure determination and refinement

X-ray diffraction data from the SeMet crystals were processedwith iMOSFLM (Battye et al., 2011) and scaled with SCALA(Evans, 2006). The structure was solved using the SAS protocol ofAuto-Rickshaw (Panjikar et al., 2005). The SHELX C/D/E platform(Sheldrick, 2010) was used to locate the heavy atoms, and calculatephases to 2.9 resolution by first determining a substructure ofselenium atoms. The correct hand for the substructure was deter-mined using the programs ABS (Hao, 2004) and SHELX E (Sheldrick,2002)). Density modification, phase extension and NCS-averagingwere carried out using the program DM (Cowtan, 1994). A modelcorresponding to 70% of the protein sequence was built usingthe program ARP/wARP (Perrakis et al., 2001). PHENIX.refine(Afonine et al., 2012) and Coot (Emsley and Cowtan, 2004) wereused for iterative rounds of model refinement and manual exten-sions. The translation/libration/screw (TLS) parameters wereincluded in the final step of refinement and final coordinates werevalidated using MolProbity (Chen et al., 2010). Electrostatic surfacepotentials were calculated by using pdb2pqr server (Dolinsky et al.,2004) using CHARMM forcefield using default parameters (pH 7.0,ionic strength 0.15 M, temperature 310 K). The figures were pre-pared using PyMol (DeLano, 2002).

2.5. Activity assay of VapC-15 toxin

For the activity assay, VapC-15 toxin was separated from theVapBC-15 complex using the denaturation and refolding protocolas described by Winther (Winther and Gerdes, 2011). A smallamount of VapC-15 could be refolded successfully, which was ver-ified by circular dichroism spectroscopy. The refolded VapC-15toxin was finally dialysed against 1 PBS (pH 7.4), 5% glyceroland 2 mM DTT. The ribonuclease activity was tested against thetotal RNA extract of E. coli K12. The reaction buffer contained10 mM HEPES (pH 7.4), 15 mM KCl and 10 mM MgCl2 with 1 lgVapC-15 against 1 lg RNA substrate in a final volume of 10 ll.The control reaction contained everything except VapC-15 protein.A reaction was also set up with the VapBC-15 complex. The reac-tions were incubated at 37 C for 5, 10 and 20 min and was stoppedusing 50 mM EDTA. The samples were analysed on a 0.8% Trisborate-EDTA agarose gel stained with ethidium bromide.

The coordinates and the structure factor for VapBC-15 complexfrom M. tuberculosis are deposited in the wwPDB with accessioncode 4CHG.3. Results

3.1. Overall structure of VapBC-15 complex

The genomic DNA of M. tuberculosis encodes an 80-residueVapB-15 antitoxin (Rv2009) and a 132-residue VapC-15 toxin(Rv2010). The co-expressed proteins formed a hetero-complex.The complex was purified to homogeneity by affinity and size-exclusion chromatography in yields of approximately 5 mg/L ofculture. The SeMet-substituted protein complex gave a very lowyield of 0.4 mg/L of culture. The crystals of both native andSeMet-substituted VapBC-15 belong to the orthorhombic spacegroup P212121. The initial attempt to solve the structure by molec-ular replacement was unsuccessful (Das et al., 2013); hence theSeMet-labelled protein crystals were produced and a SAD datasetwas collected to solve the structure. The final refinement provideda model with Rwork and Rfree of 16.7% and 19.9% respectively. Thefinal structure includes four magnesium (Mg2+) ions, four manga-nese (Mn2+) ions and 551 waters molecules. The structure has agood stereochemistry with 97% of the residues residing in thefavoured region of the Ramachandran plot. The data collectionand the refinement statistics for the SeMet and native VapBC-15crystals are shown in the Table 1 and the structures are comparedby superimposition (Fig. S3).

The asymmetric unit (ASU) of the VapBC-15 complex (Fig. 1 A)consists of 6 toxin chains and 4 antitoxin chains arranged asassemblies of one heterotetramer (VapB2C2) and two heterotrimers(VapBC2). In the VapB2C2 structure (Fig. 1 B) each of the toxinmonomer is spanned by an antitoxin monomer, whereas in thecase of the heterotrimer (VapBC2), only one toxin monomer isbound to an antitoxin monomer (Fig. 1 C). In the crystal, two pairsof metal ion bind VapB2C2, in the channel generated by the toxindimer, whereas only one pair binds VapBC2.

The 14.7 kDa VapC-15 toxin monomer (Fig. 1 D) forms a com-pact globular protein with the a/b/a fold that is archetypical forPIN-domain proteins. All six toxin chains of the ASU are completelytraced into the electron density, including all residues from posi-tions 1132 (an Ala at 0 position is an extra amino acid codedby the vector). The structure consists of 12 secondary structure ele-ments, namely b1 (residues 13), a1 (residues 513), a2 (residues1829), b2 (residues 3336), a3 (residues 3746), a4 (residues 5062),b3 (residues 6466), a5 (residues 7287), a6 (residues 96106),

Fig.1. Structure of VapBC-15 complex from Mycobacterium tuberculosis. (A) The asymmetric unit consists of six VapC-15 toxin molecules arranged as three homodimers,represented by light grey and magenta (ribbon diagrams), and four VapB15 antitoxin molecules represented by blue. The ASU therefore has one heterotetramer (VapB2C2) andtwo heterotrimers (VapBC2). (B) Structure of the heterotetramer (VapB2C2): notice how the toxin dimer is spanned by two an antitoxin monomer. (C) Structure ofheterotrimer (VapBC2), here only one antitoxin monomer spans the toxin dimer. (D) Expanded view of the structure of a monomer of VapC-15 toxin inhibited by a monomerof VapB-15 antitoxin. The Mg2+ and Mn2+ are represented by green and purple spheres respectively.

252 U. Das et al. / Journal of Structural Biology 188 (2014) 249258b4 (residues 110112), a7 (residues 115124) and b5 (residues128130). The central domain is made up of a five-strandedparallel b-sheet, with the strands in the order b3 b2 b1 b4 b5,surrounded by four a-helices (a1a4) on one side and three(a5a7) on the other. The six toxin chains in the ASU are organisedas three dimers, with each dimeric interface consisting primarily ofthe a3, a4 and a5 helices of each monomer. Each dimeric interfaceburies 1294 2 of monomeric surface area and the interactionbetween the two monomers is further stabilized by a total of 16salt bridges and 16 hydrogen bonds.

VapB-15 has 80 residues and a molecular weight of 8.8 kDa (88residues and 9.9 kDa if including C-terminal 6 His tag), but morethan than half of this antitoxin could not be modelled into the elec-tron density. The 40 N-terminal amino acid residues and between5 and 11 of the C-terminal residues could not be traced for anyof the antitoxin chains. A 9.9 kDa peak did not appear in aMALDI-TOF spectrum of the crystal contents (Fig. S1A), suggestingthat the VapB-15 protein had undergone proteolysis during crys-tallisation; this proteolysis may in part explain the inability tosee much of the structure. Only residues 4169 of VapB-15 of theheterotetramer and residues 4176 of VapB-15 of the heterotri-mers could be modelled. This observed region includes a loop con-necting two short a-helices: a1, which spans residues 4854 in allfour VapB-15 chains and a2, which spans residues 6367 ofVapB-15 chain of the heterotetramer and residues 6370 in eachof the VapB-15 chains of the heterotrimers. In the latter case, thea2 helix is followed by a short additional loop.

3.2. Geometry of the active site

PIN-domain proteins are predicted to have ribonuclease activ-ity, and metal ions are essential for the catalytic activity of thesenucleases. We observed four two-metal coordination sites in theasymmetric unit of our crystal: two two-metal sites in the hetero-tetramer and one two-metal site in each of the heterotrimers.(Metal cations are missing from one of the VapC-15 monomers ofeach of the VapBC2 heterotrimers). The metal ions were identifiedbased on previous PIN-domain protein structures (PDB ID: 2FE1and 3H87) and the anomalous difference map (Fig. 2). The activesite of the VapC-15 monomer that contains metals consists ofone Mg2+ site and one Mn2+ site, and the Mg2+ and Mn2+ ions areseparated by a distance of 3.8 . These cations occupy and arepartially neutralised by a negatively charged (acidic) groove ofVapC-15.

A cluster of six acidic residues (5 from VapC-15 and 1 fromVapB-15) surrounds the two divalent cations at the base of thechannel generated by the toxin dimer (Fig. 3A, inset II). Theresidues of VapC-15 involved in this cluster are Asp-4, Glu-42,

Fig.2. Stereo view of the active site of VapBC-15 with superimposed 2F0 Fc electron density map (cyan, contoured at 1.5r) and with anomalous difference Fourier electrondensity (yellow, contoured at 4r) shown for the Mn ion. The grey and blue sticks represent residues from VapC-15 and VapB-15 respectively.

U. Das et al. / Journal of Structural Biology 188 (2014) 249258 253Asp-96, Asp-114 and Asp-116, which were found to be conservedin a series of PIN-domain proteins (Fig 4A). Of these five residues,Asp-4 is located near the N-terminus of helix a1, and Glu-42belongs to helix a3, Asp-96 to a6 and Asp-114 and Asp-116 to helixa7. The residue provided by VapB-15 to the metal coordination siteis Glu-67, which protrudes from helix a2 of VapB-15 and projectstowards the negatively charged groove made by the acidic residuesof VapC-15.

TheMg2+ ion is directly coordinated to carboxylate oxygen atomsof Asp-96 (VapC) and of Glu-67 (of VapB), as well as to four watermolecules, together forming an octahedralmetal-bindinggeometry.Also, the carboxylate groups of Asp-4 and Glu-42 (both from VapC)indirectly coordinate the Mg2+ by forming hydrogen bonds withthree of the Mg2+-coordinating water molecules (Fig. 3A, inset II).The Mn2+ is directly coordinated by carboxylate oxygens of Asp-96, Asp-114, Asp116 (all from VapC) and Glu-67 (VapB).

It would appear that Glu-67 of VapB-15 is particularly impor-tant to the function of the antitoxin: it is one of only two acidic res-idues that directly coordinate both a Mg2+ and a Mn2+

simultaneously, and its location in the metal-containing active siteis very similar to the location of a more C-terminal VapB-15 argi-nine residue in a site that is missing the metal cations (see inset1 of Fig. 3A).

A structural superimposition of the VapBC-15 heterotetramerwith the VapBC-15 heterotrimer results in an rmsd (all residues)of 0.4 . This small rmsd indicates that in the crystal there are neg-ligible differences between the detailed structures of the twoassemblies of the ASU. (There is of course the overall differencein the assemblies involving the 1:1 ratio of VapC-15 toxin toVapB-15 antitoxin in the heterotetramer and the 2:1 ratio in theheterotrimer), When the heterotetramer and heterotrimer areoverlaid, the C-terminal loop following the a2 helix of the antitoxinmonomer in the heterotrimer superimposes well with the a2 helixof the antitoxin monomer of the heterotetramer coming from theopposite direction (Fig. 3 B).

3.3. Structural homologues

A search for structural homologues using the DALI server (Holmand Rosenstrm, 2010) and using VapC-15 as the search modelidentified various hits, most of which belong to PIN-domainproteins. The most significant structural homologues are VapC-3from M. tuberculosis (PDB: 3H87, % identity: 25, rmsd: 2.2,Z = 16.2), VapC-2 from S. flexneri (PDB: 3TND, % identity: 22, rmsd:2.0, Z = 16.6), FitB from Neisseria gonorrhoeae (PDB: 2BSQ, % iden-tity: 21, rmsd: 2.1, Z = 15.5) and VapC-2 from R. felis (PDB: 3ZVK,% identity: 14, rmsd: 2.3, Z = 15.4).

Despite a very low sequence identity with its homologues,VapC-15 still adopts the prototypic PIN-domain fold. Secondarystructure alignment of VapC-15 was carried out using the DALI hitsand using the PROMALS3D server (Pei et al., 2008). The resultshows the conservation of a cluster of known acidic residues con-fined to the specific secondary structures (Fig. 4 A). The conservedresidues in VapC-15 are Asp-4, Glu-42, Asp-96 and Asp-114. Thefifth residue, Asp-116, was found to be conserved in VapC-3 andVapC-5 of Mtb (Miallau et al., 2009; Min et al., 2012). Together,these residues cluster with divalent metal ions to form a putativeactive site, as described above. We also observed a threonine (high-lighted in green) and a serine (highlighted in cyan) residue, directlyfollowing the first aspartate (Asp-4) of VapC-15, to be conserved insome PIN-domain proteins. In our crystal structure we found thisserine (Ser-6) making a hydrogen bond with Asp-4, and the threo-nine (Thr-5) making a hydrogen bond with Glu-42 in the VapC-15toxin. There is also a conservation of an aromatic residue followingAsp-116 (highlighted in cyan). In the case of VapC-15 it is Tyr-117while in the case of T4 RNase H and other PIN-domain proteins it isphenylalanine. According to our observations, these residues (Thr-5, Ser-6) might help in the positioning of the catalytic acidic resi-dues in the active site, and their hydroxyl groups (Thr-5, Ser-6and Tyr-117) might serve as a potential hydrogen donor in thetwo-metal-catalysed ribonuclease reaction.

Secondary structure alignment of VapC-15 with known nuc-leases such as T4 RNase H and Mja FEN-1 also inferred the conser-vation of key residues involved in the metal binding (Fig. 4B).Although the three structures have broad structural differences,their active sites superimposed well (Fig. 4C), suggesting that theyare all nucleases, and probably function in a similar manner.

3.4. Enzyme activity

To verify its possible RNase activity, we purified the VapC-15toxin, separating it from the VapB-15 antitoxin by denaturation.

Fig.3. The active site geometry of VapBC-15 complex. A: In this surface representation of a heterotrimer (VapBC2), the antitoxin VapB15 (blue) spans almost 3/4 of thechannel of the VapC-15 toxin dimer (light grey), inhibiting both active sites marked as inset I and II. The figure in inset I indicate the inhibition of the second active site ofVapC-15 by VapB15s Arg-74 that shows the absence of metal ions. The figure in inset II shows the active site of VapC-15 toxin that has the metal cations. The Mg2+

coordinates directly with Asp-96 (VapC), Glu-67 (VapB) and four water molecules, each with a bond distance of approximately 2.1 , and forms an octahedral geometry. TheMn2+ is coordinated by Asp-96, Asp-114, Asp-116 (all from VapC) and Glu-67 (VapB), each with a bond distance of approximately 2 . The residues of VapB-15 and VapC-15are represented by blue and light grey sticks. The Mg2+ and Mn2+ are represented by green and purple and the water molecules by yellow spheres. The black dotted linesrepresent potential hydrogen bonds. B: Superimposition of VapB2C2 (VapB in magenta and VapC in grey) and VapBC2 (VapB in blue and VapC in green). The magnified image(to the right) shows how, in this superposition, the terminus of the side chain of Arg-74 of the heterotrimers antitoxin and the terminus of the side chain of Glu-67 of theheterotetramers antitoxin occupy nearly the same locations in the respective VapC active sites to which they bind. (For interpretation of the references to colour in this figurelegend, the reader is referred to the web version of this article.)

254 U. Das et al. / Journal of Structural Biology 188 (2014) 249258A circular dichroism spectrum indicates a predominantly alpha-helical fold (Fig. S1B) as observed in the crystal structure, whichindicates that the VapC-15 toxin refolded properly.

Ribonuclease activity assay was therefore carried out in order toassess the catalytic activity of the VapC-15 toxin. According to theagarose gel electrophoresis results (Fig. 5 B), we observed thatVapC-15 toxin degrades E. coli RNA, and that the activity seemsto be time dependent and to accelerate with increase in the timeof reaction. VapBC-15 did show some cleavage activity but to a les-ser extent than did VapC-15 after incubation for the same durationof time. VapC-15 with 50 mM EDTA did not show any activity,which is due to the chelating of metal ions by EDTA, indicative ofthe fact that metal ions are necessary for the activity.4. Discussion

4.1. VapBC-15 structure demonstrates variable modesof toxinantitoxin assembly

The three VapBC-15 complexes of the asymmetric unit areformed by two types of assemblies: one of the complexes is aheterotetramer (VapB2C2) and the other two are heterotrimers(VapBC2). The heterotetrameric assembly is equivalent to thecrystal structures of PAE0151 (Bunker et al., 2008) and VapC-3(Min et al., 2012). The same kind of assembly was also reportedfor VapBC of S. flexneri (Dienemann et al., 2011), where two hetero-tetramers combined to form a heterooctameric assembly. If weconsider one of the heterotrimer of VapBC-15, then this assemblycorresponds to the half of the hexameric form of VapBC2 fromR. felis (Mat et al., 2012). All these structures indicate that theantitoxin can bind with different stoichiometries to the toxin.4.2. Two metal ions are involved for the catalytic activity of VapC-15

The VapC proteins have until now been reported to have eithera single metal ion or no metal ions at the active sites (Fig. S2)(Bunker et al., 2008; Dienemann et al., 2011; Mat et al., 2012;Mattison et al., 2006; Miallau et al., 2009; Min et al., 2012). Ourcrystal structure of VapC-15 is the first PIN-domain toxin withtwo metal ions at the active site. Comprehensive analysis ofnucleases by Yang (Yang et al., 2006) suggests that the presenceof two metal ions at the active sites of nucleases increases the

Fig.4. Structural superimposition and secondary structure alignment of VapC-15 toxin. (A) Secondary structure-based sequence alignments of VapC-15 toxin with other PIN-domain proteins. Secondary structure elements are based on the VapC-15 structure. The columns marked with yellow (D4, E42, D96, D114 and D116 in VapC15) are highlyconserved acidic residues in PIN-domain proteins that take part in metal binding. (B) Sequence alignment of VapC-15 toxin with T4 RNase H and Mja FEN-1. (C) Stereo view ofthe active-site residues of VapC-15 toxin (green) superimposed with those of T4 RNase H (blue) and Mja FEN-1(magenta). The green and purple spheres represent Mg2+ andMn2+ of the VapC-15 toxin. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

U. Das et al. / Journal of Structural Biology 188 (2014) 249258 255substrate specificity and versatility of these enzymes. Secondarystructure alignment of the VapC-15 with various known nucleasesalso showed the conservation of key catalytic residues surroundingthe metal ions, suggesting that it too has ribonuclease activity(Fig. 4B).Assigning roles to the individual metal ions will be completelyappropriate only after a crystal structure of the complex betweenVapC-15 and RNA is available. However, based on the previousstudies on various nucleases and structural alignments, we canspeculate about the mode of substrate binding to the active site.

Fig.5. (A) Electrostatic potential surface of the VapC-15 dimer showing the absence of metal ions. The two catalytic cores at the centre are marked by negative potential (red)and the putative channel is marked by black arrows which is gatekeepered by clusters of basic amino acid residues (K47 and R56) and are denoted by positive (blue) potential.(B) Electrostatic potential surface of the VapC-15 dimer showing the presence of metal ions. The shift from red to blue can be observed due to the presence of positivelycharged metal ion at the catalytic core. The electrostatic potential is coloured with a blue to red gradient from +10kBT/e to10kBT/e. (C) Activity assay of VapC-15 toxin. Lane 1and 2, control reaction at 0 and 30 min time interval; lane 3, VapC-15 and substrate with 50 mM EDTA; lane 4, VapBC-15 and substrate at 30 min time interval; lane 57,VapC-15 and substrate at 5, 10 and 20 min.

256 U. Das et al. / Journal of Structural Biology 188 (2014) 249258Since the general topology of the active site of VapC-15 is similar tothat of T4 RNase H and of Mja FEN-1 (Fig. 4C), we propose thatcatalysis should take place in a similar manner. The two metal ionsin the current structure are separated by a distance of 3.8 andtherefore can be coordinated one side by the non-bridging oxygenof the scissile phosphate backbone of the incoming RNA substrateand the other side by conserved Asp-96 of VapC-15. The watermolecule interacting with Mg2+ might facilitate the nucleophilicattack by generating a hydroxyl ion and the transient pentacova-lent species formed as a reaction intermediate would be furtherstabilised by the presence of Mn2+. Therefore based on the previousVapC structures and VapC-15 of this study, it is possible that theVapC proteins might have different variants, which employ eitherone (Min et al., 2012) or two divalent metal ions for their nucleaseactivity.

4.3. The VapB-15 antitoxin may simultaneously inhibit the two activesites of the VapC-15 toxin dimer

We believe that the functional unit of VapC-15 toxin is ahomodimer. Not only does each assembly in the crystal consist,in part, of a VapC-15 homodimer, but homodimerisation also gen-erates a continuous channel which serves as the antitoxin bindingsite and presumably also the RNA binding site. The VapB-15 anti-toxin occupies this channel generated by the toxin dimer(Fig. 3A). This channel harbours one pair of divalent metal ions ineach VapBC2, and two pairs in VapB2C2 that are separated by a dis-tance of approximately 16 .

The N-terminal 40 residues of VapB-15 were not traceable inany of the three complexes of the crystal structure, and neitherwere portions of the C-terminus, although the length of the C-ter-minal region not observed differed in the heterotetrameric VapB2-C2 and heterotrimeric VapBC2 complexes (see Results). Thesedifferent lengths appear to be related to the different stoichiome-tries of the complexes. The inhibition of the VapC-15 toxin tran-spires in two ways. In the case of the heterotetramer, the C-terminal Glu-67 of each VapB-15 antitoxin projects into theactive-site groove of the VapC-15 toxin and coordinates bothMn2+ and Mg2+ (Fig. 3A, Inset II). In this assembly, only 29 residuesof each of the two antitoxins are observed, and they togetheroccupy the complete channel formed by the VapC-15 dimer, result-ing in VapB2C2. This inhibition is the physical occlusion of theactive site as well as presumably the RNA-binding channel. In con-trast, in the case of each of the heterotrimers, a larger potion, i.e.,36 residues, of VapB-15 is observed, and spans more than half,i.e., about three quarters, of the channel of VapC-15 (Fig. 3A);

U. Das et al. / Journal of Structural Biology 188 (2014) 249258 257hence in this case one antitoxin inhibits both active sites of thetoxin. The inhibition of the first active site of the heterotrimeroccurs in a manner similar to that described above (Fig. 3A insetII). The second active site is partly inhibited by the additional C-ter-minal 7 residues observed for VapB-15 in the heterotrimericassembly. Arg-74 of the antitoxin, observed only in this assembly,fits tightly in the second active-site groove and, being positivelycharged, repels the essential metal cations needed for catalysis(Fig. 3A inset I), hence explaining the absence of the second pairof metal ions in the case of the heterotrimers. A similar kind ofinhibition by the C-terminal arginine of antitoxins was alsoobserved in the case of the VapBC of S. flexneri (Dienemann et al.,2011) and FitAB of N. gonorrhoea (Mattison et al., 2006).4.4. VapC-15 belongs to the PIN-domain superfamily and acts as aribonuclease

Although we see two divalent metal ions in the active site of theVapC-15, we could not establish the sequence-specific cleavagespecificity (if associated) of our VapC-15 toxin. A ribonucleaseactivity assay of VapC-15 on E. coli total RNA resulted in the degra-dation of the total RNA, suggesting that VapC-15 might act as anendo/exo ribonuclease.

Despite low sequence homology, the structures of the activesites of VapC-15, T4 RNase H and Mja FEN-1 are very similar(Fig. 4C) indicating that these proteins share a similar nucleaseactivity. A comparison of the active-site residues of VapC-15 withT4 RNase H and Mja FEN-1 revealed that Asp-4, Asp-96, Asp-114and Asp-116 of VapC-15 correspond to Asp-19, Asp-132, Asp-155and Asp-157 of T4 RNase H and to Asp-27, Glu-154, Asp-173 andAsp-175 of Mja FEN-1. Based on our activity assay and structuralcomparison of VapC-15 with the known nucleases, we suggest thatVapC-15 belongs to the ribonuclease family and needs divalentmetal ions for its activity.4.5. The VapB-15 antitoxin obstructs the substrate binding and activesite of the VapC-15 toxin

Based on the surface electrostatic potentials, we propose thatthe substrate (RNA) and the inhibitor (antitoxin) should bind tothe same channel of the VapC-15 toxin dimer. The entry to thechannel has positive potential (blue in Fig. 5A and B) and is madeup of basic amino acids such as Arg-56 and Lys-47, which couldprovide effective charge complementarity to the negative chargeof RNA. Therefore these amino acids can hold the RNA along thechannel. The two active-site pockets are marked with negativepotential (red in Fig. 5A), and are separated by a distance ofapproximately 16 (Fig. 5A). Since the width of the channel atthe entry site is roughly 8 , we presume that the VapC-15 toxinmay catalyze single-stranded RNA rather than double-strandedRNA, but a co-crystal structure of VapC-15 with an RNA substrateis needed to confirm this hypothesis.

As mentioned above, the antitoxin also binds to the same chan-nel as the does the substrate, therefore during unfavourable condi-tions, the cellular proteases degrade the antitoxin in a regulatorymanner, which clears the channel. In this state the toxin is freeto act as a ribonuclease and carry out its biochemical function. Alsoin the absence of antitoxin, the acidic grooves in the channel areexposed, and can thus bind divalent metal ions.Author contributions

UD planned and performed the experiments. UD, VP, AS and SGanalysed the data. UKTS helped with the experiments. MW, AS andSG provided the facility to perform the experiments. UD wrote thepaper taking inputs from all the authors.

Acknowledgements

We thank Dr. Guillaume Pompidor of EMBL Hamburg for theassistance in data collection at P13 beamline of PETRA III at DESY,Hamburg. UD acknowledges European Molecular Biology Organi-sation for the short term fellowship and European Molecular Biol-ogy Laboratory, Hamburg for all the laboratory assistance. Weacknowledge Dr. Rosemary Wilson of EMBL Hamburg for the care-ful reading, comments and corrections of the manuscript.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jsb.2014.10.002.

References

Afonine, P.V., Grosse-Kunstleve, R.W., Echols, N., Headd, J.J., Moriarty, N.W.,Mustyakimov, M., Terwilliger, T.C., Urzhumtsev, A., Zwart, P.H., Adams, P.D.,2012. Towards automated crystallographic structure refinement withphenix.refine. Acta Crystallogr. D Biol. Crystallogr. 68, 352367. http://dx.doi.org/10.1107/S0907444912001308.

Anantharaman, V., Aravind, L., 2003. New connections in the prokaryotic toxinantitoxin network: relationship with the eukaryotic nonsense-mediated RNAdecay system. Genome Biol. 4, R81.

Arcus, V.L., McKenzie, J.L., Robson, J., Cook, G.M., 2011. The PIN-domainribonucleases and the prokaryotic VapBC toxinantitoxin array. Protein Eng.Des. Select. PEDS 24, 3340. http://dx.doi.org/10.1093/protein/gzq081.

Arcus, V.L., Rainey, P.B., Turner, S.J., 2005. The PIN-domain toxinantitoxin array inmycobacteria. Trends Microbiol. 13, 360365. http://dx.doi.org/10.1016/j.tim.2005.06.008.

Battye, T.G.G., Kontogiannis, L., Johnson, O., Powell, H.R., Leslie, A.G.W., 2011.IMOSFLM: a new graphical interface for diffraction-image processing withMOSFLM. Acta Crystallogr. D Biol. Crystallogr. 67, 271281. http://dx.doi.org/10.1107/S0907444910048675.

Blower, T.R., Pei, X.Y., Short, F.L., Fineran, P.C., Humphreys, D.P., Luisi, B.F., Salmond,G.P.C., 2011. A processed noncoding RNA regulates an altruistic bacterialantiviral system. Nat. Struct. Mol. Biol. 18, 185190. http://dx.doi.org/10.1038/nsmb.1981.

Bunker, R.D., McKenzie, J.L., Baker, E.N., Arcus, V.L., 2008. Crystal structure ofPAE0151 from Pyrobaculum aerophilum, a PIN-domain (VapC) protein from atoxinantitoxin operon. Proteins 72, 510518. http://dx.doi.org/10.1002/prot.22048.

Chen, V.B., Arendall, W.B., Headd, J.J., Keedy, D.A., Immormino, R.M., Kapral, G.J.,Murray, L.W., Richardson, J.S., Richardson, D.C., 2010. MolProbity: all-atomstructure validation for macromolecular crystallography. Acta Crystallogr. DBiol. Crystallogr. 66, 1221. http://dx.doi.org/10.1107/S0907444909042073.

Christensen, S.K., Gerdes, K., 2003. RelE toxins from bacteria and Archaea cleavemRNAs on translating ribosomes, which are rescued by tmRNA. Mol. Microbiol.48, 13891400. http://dx.doi.org/10.1046/j.1365-2958.2003.03512.x.

Cline, S.D., Saleem, S., Daines, D.A., 2012. Regulation of the vapBC-1 toxinantitoxinlocus in nontypeable Haemophilus influenzae. PLoS One 7, e32199. http://dx.doi.org/10.1371/journal.pone.0032199.

Cowtan, K., 1994. Joint CCP4 and ESF-EACBM newsletter on protein crystallography.Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography 31, p 3438.

Dao-Thi, M.-H., Van Melderen, L., De Genst, E., Afif, H., Buts, L., Wyns, L., Loris, R.,2005. Molecular basis of gyrase poisoning by the addiction toxin CcdB. J. Mol.Biol. 348, 10911102. http://dx.doi.org/10.1016/j.jmb.2005.03.049.

Das, U., Kumar, N., Gourinath, S., Srinivasan, A., 2013. Preliminary crystallographicanalysis of recombinant VapBC-15 toxinantitoxin complex fromMycobacterium tuberculosis. Acta Crystallogr. F Struct. Biol. Cryst. Commun.69, 12421245. http://dx.doi.org/10.1107/S1744309113024822.

DeLano, W.L., 2002. The PyMOL Molecular Graphics System. Schrdinger LLCwwwpymolorg Version 1, .

Dienemann, C., Bggild, A., Winther, K.S., Gerdes, K., Brodersen, D.E., 2011. Crystalstructure of the VapBC toxinantitoxin complex from Shigella flexneri reveals ahetero-octameric DNA-binding assembly. J. Mol. Biol. 414, 713722. http://dx.doi.org/10.1016/j.jmb.2011.10.024.

Dolinsky, T.J., Nielsen, J.E., McCammon, J.A., Baker, N.A., 2004. PDB2PQR: anautomated pipeline for the setup of Poisson-Boltzmann electrostaticscalculations. Nucleic Acids Res. 32. http://dx.doi.org/10.1093/nar/gkh381.

Emsley, P., Cowtan, K., 2004. Coot: model-building tools for molecular graphics.Acta Crystallogr. D Biol. Crystallogr. 60, 21262132. http://dx.doi.org/10.1107/S0907444904019158.

Eom, S.H., Wang, J., Steitz, T.A., 1996. Structure of Taq polymerase with DNA at thepolymerase active site. Nature 382, 278281. http://dx.doi.org/10.1038/382278a0.

http://dx.doi.org/10.1016/j.jsb.2014.10.002http://dx.doi.org/10.1107/S0907444912001308http://dx.doi.org/10.1107/S0907444912001308http://refhub.elsevier.com/S1047-8477(14)00211-1/h0010http://refhub.elsevier.com/S1047-8477(14)00211-1/h0010http://refhub.elsevier.com/S1047-8477(14)00211-1/h0010http://dx.doi.org/10.1093/protein/gzq081http://dx.doi.org/10.1016/j.tim.2005.06.008http://dx.doi.org/10.1016/j.tim.2005.06.008http://dx.doi.org/10.1107/S0907444910048675http://dx.doi.org/10.1107/S0907444910048675http://dx.doi.org/10.1038/nsmb.1981http://dx.doi.org/10.1038/nsmb.1981http://dx.doi.org/10.1002/prot.22048http://dx.doi.org/10.1002/prot.22048http://dx.doi.org/10.1107/S0907444909042073http://dx.doi.org/10.1046/j.1365-2958.2003.03512.xhttp://dx.doi.org/10.1371/journal.pone.0032199http://dx.doi.org/10.1371/journal.pone.0032199http://dx.doi.org/10.1016/j.jmb.2005.03.049http://dx.doi.org/10.1107/S1744309113024822http://www.pymol.org.%20doi:citeulike-article-id:240061http://www.pymol.org.%20doi:citeulike-article-id:240061http://dx.doi.org/10.1016/j.jmb.2011.10.024http://dx.doi.org/10.1016/j.jmb.2011.10.024http://dx.doi.org/10.1093/nar/gkh381http://dx.doi.org/10.1107/S0907444904019158http://dx.doi.org/10.1107/S0907444904019158http://dx.doi.org/10.1038/382278a0

258 U. Das et al. / Journal of Structural Biology 188 (2014) 249258Evans, P., 2006. Scaling and assessment of data quality. Acta Crystallogr. D Biol.Crystallogr. 62, 7282. http://dx.doi.org/10.1107/S0907444905036693.

Fozo, E.M., Makarova, K.S., Shabalina, S.A., Yutin, N., Koonin, E.V., Storz, G., 2010.Abundance of type I toxinantitoxin systems in bacteria: searches for newcandidates and discovery of novel families. Nucleic Acids Res. 38, 37433759.http://dx.doi.org/10.1093/nar/gkq054.

Gazit, E., Sauer, R.T., 1999. The Doc toxin and Phd antidote proteins of thebacteriophage P1 plasmid addiction system form a heterotrimeric complex.J. Biol. Chem. 274, 1681316818. http://dx.doi.org/10.1074/jbc.274.24.16813.

Gerdes, K., 2000. toxinantitoxin Modules May Regulate Synthesis ofMacromolecules during Nutritional Stress. J. Bacteriol. 182, 561572. http://dx.doi.org/10.1128/JB.182.3.561-572.2000.Updated.

Gerdes, K., Rasmussen, P.B., Molin, S., 1986. Unique type of plasmid maintenancefunction: postsegregational killing of plasmid-free cells. Proc. Natl. Acad. Sci.U.S.A. 83, 31163120.

Hao, Q., 2004. ABS: A program to determine absolute configuration and evaluateanomalous scatterer substructure. J. Appl. Crystallogr. 37, 498499. http://dx.doi.org/10.1107/S0021889804008696.

Hayes, F., 2003. Toxins-antitoxins: plasmid maintenance, programmed cell death,and cell cycle arrest. Science (New York, NY) 301, 14961499. http://dx.doi.org/10.1126/science.1088157.

Holm, L., Rosenstrm, P., 2010. Dali server: conservation mapping in 3D. NucleicAcids Res. 38, W545W549. http://dx.doi.org/10.1093/nar/gkq366.

Hwang, K.Y., Baek, K., Kim, H.Y., Cho, Y., 1998. The crystal structure of flapendonuclease-1 from Methanococcus jannaschii. Nat. Struct. Biol. 5, 707713.http://dx.doi.org/10.1038/1406.

Jiang, Y., Pogliano, J., Helinski, D.R., Konieczny, I., 2002. ParE toxin encoded by thebroad-host-range plasmid RK2 is an inhibitor of Escherichia coli gyrase. Mol.Microbiol. 44, 971979. doi:2921 (pii).

Korch, S.B., Hill, T.M., 2006. Ectopic overexpression of wild-type and mutant hipAgenes in Escherichia coli: effects on macromolecular synthesis and persisterformation. J. Bacteriol. 188, 38263836.

Leplae, R., Geeraerts, D., Hallez, R., Guglielmini, J., Drze, P., Van Melderen, L., 2011.Diversity of bacterial type II toxinantitoxin systems: a comprehensive searchand functional analysis of novel families. Nucleic Acids Res. 39, 55135525.http://dx.doi.org/10.1093/nar/gkr131.

Mat, M.J., Vincentelli, R., Foos, N., Raoult, D., Cambillau, C., Ortiz-Lombarda, M.,2012. Crystal structure of the DNA-bound VapBC2 antitoxin/toxin pair fromRickettsia felis. Nucleic Acids Res. 40, 32453258. http://dx.doi.org/10.1093/nar/gkr1167.

Mattison, K., Wilbur, J.S., So, M., Brennan, R.G., 2006. Structure of FitAB fromNeisseria gonorrhoeae bound to DNA reveals a tetramer of toxinantitoxinheterodimers containing pin domains and ribbon-helix-helix motifs. J. Biol.Chem. 281, 3794237951. http://dx.doi.org/10.1074/jbc.M605198200.

Miallau, L., Faller, M., Chiang, J., Arbing, M., Guo, F., Cascio, D., Eisenberg, D., 2009.Structure and proposed activity of a member of the VapBC family of toxinantitoxin systems. VapBC-5 from Mycobacterium tuberculosis. J. Biol. Chem. 284,276283. http://dx.doi.org/10.1074/jbc.M805061200.Min, A.B., Miallau, L., Sawaya, M.R., Habel, J., Cascio, D., Eisenberg, D., 2012. Thecrystal structure of the Rv0301-Rv0300 VapBC-3 toxinantitoxin complex fromM. tuberculosis reveals a Mg2+ ion in the active site and a putative RNA-bindingsite. Protein Sci. 21, 175467. doi:10.1002/pro.2161.

Mueser, T.C., Nossal, N.G., Hyde, C.C., 1996. Structure of bacteriophage T4 RNase H, a50 to 30 RNADNA and DNADNA exonuclease with sequence similarity to theRAD2 family of eukaryotic proteins. Cell 85, 11011112.

Pandey, D.P., Gerdes, K., 2005. Toxinantitoxin loci are highly abundant in free-livingbut lost from host-associated prokaryotes. Nucleic Acids Res. 33, 966976.

Panjikar, S., Parthasarathy, V., Lamzin, V.S., Weiss, M.S., Tucker, P.A., 2005. Auto-Rickshaw: an automated crystal structure determination platform as anefficient tool for the validation of an X-ray diffraction experiment. ActaCrystallogr. D Biol. Crystallogr. 61, 449457. http://dx.doi.org/10.1107/S0907444905001307.

Pei, J., Tang, M., Grishin, N.V., 2008. PROMALS3D web server for accurate multipleprotein sequence and structure alignments. Nucleic Acids Res. 36, W30W34.http://dx.doi.org/10.1093/nar/gkn322.

Perrakis, A., Harkiolaki, M., Wilson, K.S., Lamzin, V.S., 2001. ARP / wARP andmolecular replacement. Acta Crystallogr. D Biol. Crystallogr. 57, 14451450.http://dx.doi.org/10.1107/S0907444901014007.

Ramage, H.R., Connolly, L.E., Cox, J.S., 2009. Comprehensive functional analysis ofMycobacterium tuberculosis toxinantitoxin systems: implications forpathogenesis, stress responses, and evolution. PLoS Genet. 5, e1000767.

Sheldrick, G.M., 2002. Macromolecular phasing with SHELXE. Z. Kristallogr.. http://dx.doi.org/10.1524/zkri.217.12.644.20662.

Sheldrick, G.M., 2010. Experimental phasing with SHELXC/D/E: combining chaintracing with density modification. Acta Crystallogr. D Biol. Crystallogr. 66, 479485. http://dx.doi.org/10.1107/S0907444909038360.

Studier, F.W., 2005. Protein production by auto-induction in high-density shakingcultures. Protein Expr. Purif.. http://dx.doi.org/10.1016/j.pep.2005.01.016.

Van Duyne, G.D., Standaert, R.F., Karplus, P.A., Schreiber, S.L., Clardy, J., 1993. Atomicstructures of the human immunophilin FKBP-12 complexes with FK506and rapamycin. J. Mol. Biol. 229, 105124. http://dx.doi.org/10.1006/jmbi.1993.1012.

Wang, X., Lord, D.M., Cheng, H.-Y., Osbourne, D.O., Hong, S.H., Sanchez-Torres, V.,Quiroga, C., Zheng, K., Herrmann, T., Peti, W., Benedik, M.J., Page, R., Wood, T.K.,2012. A new type V toxinantitoxin system where mRNA for toxin GhoT iscleaved by antitoxin GhoS. Nat. Chem. Biol.. http://dx.doi.org/10.1038/nchembio.1062.

Winther, K.S., Gerdes, K., 2011. Enteric virulence associated protein VapC inhibitstranslation by cleavage of initiator tRNA. Proc. Natl. Acad. Sci. U.S.A. 108, 74037407. http://dx.doi.org/10.1073/pnas.1019587108.

Yang, W., Lee, J.Y., Nowotny, M., 2006. Making and breaking nucleic acids: two-Mg2+-ion catalysis and substrate specificity. Mol. Cell 22, 513. http://dx.doi.org/10.1016/j.molcel.2006.03.013.

Zhang, Y., Zhang, J., Hoeflich, K.P., Ikura, M., Qing, G., Inouye, M., 2003. MazF cleavescellular mRNAs specifically at ACA to block protein synthesis in Escherichia coli.Mol. Cell 12, 913923. http://dx.doi.org/10.1016/S1097-2765(03)00402-7.

http://dx.doi.org/10.1107/S0907444905036693http://dx.doi.org/10.1093/nar/gkq054http://dx.doi.org/10.1074/jbc.274.24.16813http://dx.doi.org/10.1128/JB.182.3.561-572.2000.Updatedhttp://dx.doi.org/10.1128/JB.182.3.561-572.2000.Updatedhttp://refhub.elsevier.com/S1047-8477(14)00211-1/h0115http://refhub.elsevier.com/S1047-8477(14)00211-1/h0115http://refhub.elsevier.com/S1047-8477(14)00211-1/h0115http://dx.doi.org/10.1107/S0021889804008696http://dx.doi.org/10.1107/S0021889804008696http://dx.doi.org/10.1126/science.1088157http://dx.doi.org/10.1126/science.1088157http://dx.doi.org/10.1093/nar/gkq366http://dx.doi.org/10.1038/1406http://refhub.elsevier.com/S1047-8477(14)00211-1/h0140http://refhub.elsevier.com/S1047-8477(14)00211-1/h0140http://refhub.elsevier.com/S1047-8477(14)00211-1/h0140http://dx.doi.org/10.1093/nar/gkr131http://dx.doi.org/10.1093/nar/gkr1167http://dx.doi.org/10.1093/nar/gkr1167http://dx.doi.org/10.1074/jbc.M605198200http://dx.doi.org/10.1074/jbc.M805061200http://refhub.elsevier.com/S1047-8477(14)00211-1/h0170http://refhub.elsevier.com/S1047-8477(14)00211-1/h0170http://refhub.elsevier.com/S1047-8477(14)00211-1/h0170http://refhub.elsevier.com/S1047-8477(14)00211-1/h0170http://refhub.elsevier.com/S1047-8477(14)00211-1/h0170http://refhub.elsevier.com/S1047-8477(14)00211-1/h0175http://refhub.elsevier.com/S1047-8477(14)00211-1/h0175http://dx.doi.org/10.1107/S0907444905001307http://dx.doi.org/10.1107/S0907444905001307http://dx.doi.org/10.1093/nar/gkn322http://dx.doi.org/10.1107/S0907444901014007http://refhub.elsevier.com/S1047-8477(14)00211-1/h0195http://refhub.elsevier.com/S1047-8477(14)00211-1/h0195http://refhub.elsevier.com/S1047-8477(14)00211-1/h0195http://dx.doi.org/10.1524/zkri.217.12.644.20662http://dx.doi.org/10.1524/zkri.217.12.644.20662http://dx.doi.org/10.1107/S0907444909038360http://dx.doi.org/10.1016/j.pep.2005.01.016http://dx.doi.org/10.1006/jmbi.1993.1012http://dx.doi.org/10.1006/jmbi.1993.1012http://dx.doi.org/10.1038/nchembio.1062http://dx.doi.org/10.1038/nchembio.1062http://dx.doi.org/10.1073/pnas.1019587108http://dx.doi.org/10.1016/j.molcel.2006.03.013http://dx.doi.org/10.1016/j.molcel.2006.03.013http://dx.doi.org/10.1016/S1097-2765(03)00402-7

Crystal structure of the VapBC-15 complex from Mycobacterium tuberculosis reveals a two-metal ion dependent PIN-domain ribonuclease and a variable mode of toxinantitoxin assembly1 Introduction2 Materials and methods2.1 Cloning and co-expression of vapB-15 and vapC-15 genes2.2 Protein purification2.3 Crystallisation and data collection2.4 Structure determination and refinement2.5 Activity assay of VapC-15 toxin

3 Results3.1 Overall structure of VapBC-15 complex3.2 Geometry of the active site3.3 Structural homologues3.4 Enzyme activity

4 Discussion4.1 VapBC-15 structure demonstrates variable modes of toxinantitoxin assembly4.2 Two metal ions are involved for the catalytic activity of VapC-154.3 The VapB-15 antitoxin may simultaneously inhibit the two active sites of the VapC-15 toxin dimer4.4 VapC-15 belongs to the PIN-domain superfamily and acts as a ribonuclease4.5 The VapB-15 antitoxin obstructs the substrate binding and active site of the VapC-15 toxin

Author contributionsAcknowledgementsAppendix A Supplementary dataReferences