Escherichia coli JM109 was used as a general cloning strain (Promega Madison, WI, USA). The expression strain E. coli BL21(DE3) was purchased from Invitrogen (Groningen, The Netherlands). The E. coli strains used for assessing restriction activity in vivo comprise NM1261 (r⁻m⁻), NM1049 (restriction system Type IA), NK354 (Type IB), NK402 (Type IC), NM1009 (Type ID) and were a kind gift of Professor Noreen E. Murray (School of Biology, University of Edinburgh, UK). NM1041 (hsd⁺ clpX⁻) was again obtained from Noreen Murray (30). The NK and NM series were converted to DE3 lysogens as described previously (8).

Four ArdB genes were identified in the public sequence database held by the National Center for Biotechnology Information (Bethesda, MD, USA), Table 1. The gene sequences were then synthesised by GeneArt (Regensburg, Germany). The codon usage pattern was optimized in order to maximize expression yield in E. coli. To facilitate cloning, an NdeI restriction site was engineered to overlap the ATG start codon of each synthetic gene. In addition, a unique HindIII site was also included just downstream of the stop codon. The synthesised genes were then ligated into pET24a (kanamycin resistant) at the NdeI and HindIII sites of the expression vector and the anticipated sequence was verified by DNA sequencing on both strands. Two further ardB genes from E. coli DH5α (yjfX and yafX), Table 1, were PCR-amplified from genomic DNA and ligated into pET20b (Novagen, Madison, WI, USA). The primers used to amplify yfjX were: forward primer, 5′-ACGGTGTCATATGACAACACAGACACAGT-3′; reverse primer, 5′-GCTAGAATTCTTAATCAATAATGCGCAT-3′. The primers used to amplify yafX were: forward primer 5′-ACGGTGTCATATGACAACACAGACGCAGC-3′; reverse primer, 5′-GCTAGAATTCTCAGTCGATAATACGCAG-3′. The NdeI and EcoRI recognition sites used in the subsequent ligation are underlined. In both cases an amplified product of the expected size was obtained. The PCR products were then ligated into the NdeI and EcoRI sites of pET20b (ampicillin resistant). The sequence of the insert was verified to ensure no mistakes had been introduced during amplification. For assessing the effect of expressing klcA₁₃₆ in NM1041(DE3) (hsd⁺ clpX⁻) on growth in the presence and absence of 2-AP, klcA₁₃₆ was ligated into an expression vector with a carbenicillin selection marker (pET20b). This was achieved by digesting the KlcA₁₃₆ pET24a expression construct with NdeI and HindIII and ligating the insert into the corresponding sites of pET20b to produce pET20b-klcA₁₃₆. Table 1.

The source of the ardB and klcA genes investigated

The extinction coefficients of the proteins were calculated using the method of Gill and von Hippel (31), Table 1. All concentrations quoted for the proteins are based on the concentrations of monomers. The monomer molecular weights of the proteins are shown in Table 1 and include the N-terminal methionine.

Escherichia coli BL21(DE3) cells were transformed with the recombinant plasmid and transformants selected using LB agar supplemented with 25 μg/ml kanamycin or 100 µg/ml carbenicillin. A single colony was picked and grown in LB medium plus 25 μg/ml kanamycin or 100 µg/ml carbenicillin at 37°C to an optical density at 600 nm (OD₆₀₀) of 0.5. Heterologous gene expression was induced by the addition of IPTG to 1 mM and the growth continued for a further 3 h at the temperature given in Table 1 before harvesting. For four of the chosen genes, a prominent protein band corresponding to the predicted protein mass was visible on SDS–polyacrylamide gels of induced cells of BL21(DE3) containing the recombinant plasmid that was not visible in the negative control of the vector alone in BL21(DE3). The A. denitrificans klcA and L. pneumophila ardB could not be expressed at levels sufficient for purification. For protein production, a single colony of was used to inoculate 100 ml of LB medium supplemented with 25 μg/ml kanamycin or 100 µg/ml carbenicillin. The starter culture was grown to an OD₆₀₀ of 0.5–0.8 at 37°C and used to inoculate 10 L of LB containing 25 μg/ml kanamycin or 100 µg/ml carbenicillin and the cells grown at 37°C to an OD₆₀₀ of 0.5. Heterologous gene expression was induced as described above. Typically 50 g of cells were obtained from 10 l of culture. Cells were harvested by centrifugation (6000g, 4°C, 15 min) and the cell pellet was stored at −20°C until required.

Approximately 20 g of cell pellet was defrosted on ice for 30 min and resuspended in 200 ml buffer A (20 mM Tris–HCl pH 8.0, 7 mM 2-mercaptoethanol). A cocktail of protease inhibitors (Complete Protease Inhibitor Cocktail Tablet; Roche) was added to the cell suspension (1 tablet per 50 ml solution). The cells were disrupted by sonication on ice using a Soniprep 150 sonicator (Sanyo, Tokyo, Japan) fitted with a 9 mm diameter probe (5 × 1 min bursts with 15 s interval between each burst). The cell lysate was centrifuged at 20 000g for 1 h at 4°C and the supernatant was filtered through a 0.45 μm filter unit (Sartorius AG, Goettingen, Germany). The clarified extract was then loaded onto a 30 cm × 1 cm diameter DEAE anion exchange column equilibrated with buffer A, at a flow rate of 60 ml/h. The column was washed with ∼3 column volumes of buffer A to remove unbound material. Bound proteins were then eluted using a 500 ml gradient of 0–0.7 M NaCl in buffer A at a flow rate of 20 ml/h. Fractions that gave a UV absorbance were subsequently analysed by SDS–PAGE and those containing ArdB were pooled and concentrated using a spin concentrator with a 5 kDa MWCO (VivaScience AG, Hannover, Germany) to a final volume of ∼5 ml. The sample was then loaded on a Superdex-200 column (GE Healthcare, Uppsala, Sweden), equilibrated with buffer A + 0.2 M NaCl, and the protein was eluted at a flow rate of 10 ml/h. The fractions were subsequently analysed by SDS–PAGE and those containing ArdB were pooled and concentrated. Purified ArdB/KlcA was stored at −20°C in buffer A supplemented with glycerol to 50% (v/v).

The methods employed for assaying ardB activity in vivo used the efficiency of plating (eop) of phage lambda (virulent), λv, on the various strains of E. coli as described previously in detail (8,13). The NM and NK E. coli strains were transformed with the derivatives of pET24a or pET20b described above but protein overexpression was not induced. Instead, we relied upon the small amount of leaky expression to produce the anti-restriction proteins in sufficient quantity to demonstrate anti-restriction in vivo. All assays were performed in triplicate and at least 50 phage plaques per plate were counted. The standard deviation on the anti-restriction and anti-modification values is ± 25%.

To assess the possible role of the ClpXP in the anti-restriction behaviour of ArdB and KlcA, the ability of E. coli NM1041 (hsd⁺ clpX⁻) transformed with plasmids expressing ardB or klcA to survive on an agar plate containing 2-aminopurine (2-AP) was assessed (32). The strain will not grow in the presence of 2-AP unless the EcoKI RM system is inactivated by the production of active ArdB or KlcA. A single colony from an agar/carbenicillin plate of E. coli NM1041 transformed with either pET20b or pET20b-klcA₁₃₆ was streaked on an agar plate with or without 2-AP and incubated overnight at 37°C. This was repeated in triplicate with different colonies.

Fourier transform ion cyclotron mass spectrometry (33), isothermal titration calorimetry (10,13,33), size exclusion chromatography (34) and assays of EcoKI endonuclease activity (8,13,34) were performed as described previously. To investigate whether ArdB or KlcA interfered with the assembly of the EcoKI nuclease, 42 nM EcoKI Mtase and ∼100 nM HsdR proteins (34) were incubated separately with a 20-fold excess of ArdB or KlcA in the nuclease reaction buffer for 2 min at 37°C. Equal volumes of these two samples were then mixed and pBRsk1 DNA added to 3 nM. After incubation of this reaction mix for 8 min at 37°C, the reaction was stopped as above.

The KlcA protein from B. pertussis was uniformly labelled with ¹⁵N and ¹³C by growing the recombinant strain in M9-minimal medium containing ¹⁵NH₄SO₄ and ¹³C-d-glucose as the sole nitrogen and carbon sources, respectively. The labelled protein was purified as described above. Samples for NMR measurements contained 0.7 mM KlcA, 90% H₂O/10% D₂O and 0.05% NaN₃ in 20 mM deuterated sodium acetate buffer (pH 6.0).

All NMR experiments were carried out at 298 K unless stated otherwise on a Bruker AVANCE 800 MHz with a 5-mm triple resonance cryoprobe. NMR experiments for ¹H, ¹⁵N and ¹³C backbone and side-chain resonance assignments included: two-dimensional (2D) ¹H-¹H TOCSY, ¹H-¹H NOESY, ¹H-¹⁵N HSQC, ¹H-¹³C HSQC, (HB)CB(CGCD)HD/(HB)CB(CGCDCE)HE, three-dimensional (3D) HNCO, HN(CA)CO, CBCANH, CBCA(CO)NH, HBHA(CO)NH, HBHANH, (H)CCH-TOCSY, H(CCO)NH, C(CO)NH (35). 3D ¹⁵N- and ¹³C-edited NOESY-HSQC spectra were acquired to obtain distance constraints (36,37). Aromatic ¹H^δ and ¹H^ε frequencies were obtained from (HB)CB(CGCD)HD/ (HB)CB(CGCDCE)HE spectra (38) where the aromatic ¹H frequencies are correlated with the assigned ¹³C^β frequencies of the aromatic residue.

NMR data were processed using the Azara suite of programs (Wayne Boucher, Department of Biochemistry, University of Cambridge, UK, http://www.ccpn.ac.uk/azara/) and analysed using Ccpnmr Analysis (39). NOE assignments were generated during the initial rounds of structure analysis using CANDID (40) in CYANA 2.1 (41). Dihedral constraints were generated using TALOS (42). The NOE assignments obtained by CANDID were checked manually and from the 100 structures calculated, 20 structures with lowest target function were selected for further analysis. The structures were further refined with the incorporation of residual dipolar couplings (RDCs) using CNS (43). PROCHECK-NMR (44) and WHATIF (45) were used to analyse the quality of the structures. MOLMOL (46) and PYMOL (http://www.pymol.org) were used for visualisation.

A 200 µM ¹⁵N-labelled KlcA sample in 20 mM deuterated sodium acetate, 0.05% NaN_3, pH 6.0 in 90% H₂O/10% D₂O aligned with 6.25 mg/ml of Pf1 filamentous phage (Pf1 protease and RNase free, Profos) was used to measure RDCs. Magnetic alignment was monitored by 1D ²H NMR spectra as a function of phage concentration. A 7 Hz residual quadrapolar splitting of the deuterium signal was achieved with 6.25 mg/ml phage. ¹H-¹⁵N splittings were measured under isotropic and partially aligned conditions using 2D IPAP ¹H-¹⁵N HSQC experiments (47). Analysis and back calculation of the RDCs were performed with REDCAT (48).

Relaxation experiments were performed on a 600 MHz Bruker Avance spectrometer at 25°C with 200 µM ¹⁵N labelled KlcA. For the longitudinal (T₁) and transverse (T₂) relaxation data for the backbone ¹⁵N nuclei of KlcA, water was kept along the z-axis during the entire pulse sequence (49) and extra delays were used in the T₁ experiment to eliminate differential heating arising from the variations of the relaxation delay (50). Eight interleaved spectra were recorded for each relaxation delay T₁ (51 twice, 401, 601, 801, 901, 1001 and 1201 ms) and T₂ (16 twice, 48, 64, 80, 96, 112 and 144 ms). T₁ and T₂ cross-peak intensities were measured from peak heights and fit to an exponential decay function I(t) = I(0)e^−Rt where I(0) and I(t) are the cross-peak intensity at time 0 and time t, where time is the relaxation delay and R (1/T) is the relaxation rate (39). For the 2D ¹H-¹⁵N heteronuclear NOE experiment (49) a 3 s pre-saturation period and a 2 s relaxation delay was used; the control experiment had an equivalent 5 s delay. Spectra from the heteronuclear NOE experiment were extended by linear prediction and zero filling prior to apodisation with a 5% shifted mixed Gaussian/exponential function. NOE peak heights were taken from the Gaussian fits of the frequency-domain data. The NOE was calculated as NOE = I_on/I_off where I_on and I_off are the peak heights with and without ¹H pre-saturation, respectively. T₁, T₂ and NOE data were fitted using ‘Quadric_diffusion’, (AG Palmer, Columbia University, NY, USA, http://biochemistry.hs.columbia.edu/labs/palmer/software/quadric.html) to obtain the rotational diffusion tensor. Model-free analysis of the relaxation data was performed using Tensor2 (51). For this analysis a subset of relaxation data was selected using criteria defined by Tjandra et al. (52). The relative moments of inertia were calculated with ‘pdbinertia’ (http://biochemistry.hs.columbia.edu/labs/palmer/software/pdbinertia.html) and the average rotational correlation time (τ_c) calculated with r2r1_tm (http://biochemistry.hs.columbia.edu/labs/palmer/software/r2r1_tm.html).

The complete set of coordinates for the structure has been deposited at the Protein Data Bank (PDB) (53) with accession code 2 kmg and in the Biological magnetic resonance bank with code 16428.

Secondary structure was identified using STRIDE (54). Fold topology analysis was conducted using HERA (55) under PROMOTIF v2.0 (56) and schematically redrawn using TopDraw (57). The B. pertussis KlcA sequence was subjected to a BLAST search (58) against the PDB and to fold recognition server PHYRE (59) to check if there was any clear or remote homology to an existing structure or fold at the sequence level. Additionally, the 3D-coordinates of the KlcA structure were compared against 3D-fold and structure libraries using DALI (60) and SSM (61).

Sequences related to B. pertussis KlcA (UniProt Accession no: Q08L07) were obtained using a BLASTP search (58) against the UniProt sequence database (62), via the NPS@ server (63). A total of 182 sequences with an E-value of less than 10⁻⁶ were obtained and this list was made non-redundant initially by removing 62 identical sequences. The remaining 120 KlcA-like sequences were retrieved and a multiple sequence alignment was generated using the program PROMALS3D (64). This set of sequences was rendered further non-redundant by removing all 90% pairwise identical sequences in the alignment using the Expasy server (http://www.expasy.ch/tools/redundancy/). A total of 44 non-redundant divergent sequences were thus obtained. Upon analysis, one sequence that lacked the C-terminal helix 5 was discarded and three other sequences, which possessed N-terminal extensions, were trimmed. The remaining 43-aligned sequences were used as input for Consurf (65) (http://consurf.tau.ac.il/) to map sequence conservation on the surface of the protein. The electrostatic surface potential representation was calculated using GRASP (66). The lipophilic surface rendition was generated using MOLCAD (67) under SYBYL (Tripos Associates, St Louis, MO, USA). Solvent-accessibility calculations were performed using GETAREA (68).

It has been shown that unregulated expression of the kil genes leads to killing of the host cells. The cells become elongated with distorted outer membranes and macromolecular synthesis ceases (69). Previous studies of the kilC operon in plasmid RK2 showed that klcA was responsible for the lethality of E. coli host cells in the absence of korA and korC (20). However, we were able to overexpress the klcA₁₃₆ gene in a pET vector in the absence of kor determinants with no apparent harmful effects. Furthermore, in vivo anti-restriction experiments performed on the klcA from RK2 failed to identify any signs of anti-restriction function (20), whereas the klcA₁₃₆ and klcA_ADE showed efficient anti-restriction activity against all four families of Type I restriction systems as described below.

The activity of anti-restriction genes can be tested by determining how many phages successfully infect a bacterial lawn in the presence or absence of host RM systems when the bacterium also contains a plasmid encoding the anti-restriction protein. All of the ardB and klcA genes except ardB_LPP showed significant levels of anti-restriction (>10-fold enhancement in the number of phage infections) despite their low levels of expression in the absence of induction, Table 2. The poor activity of the ardB_LPP and klcA_ADE genes were most likely due to poor expression as neither could be observed on SDS–PAGE. The effectiveness of the other proteins in inhibiting restriction matched levels observed for the ocr (9,10,15,70) and ArdA proteins (8,12,13) in vivo under similar conditions. Table 2.

Effect of expression of six different ardB and klcA on the RM of bacteriophage λv by four different Type I RM systems, each representing one of the four families (IA–ID) of Type I RM systems

However, in contrast to ocr and ArdA, anti-modification by ArdB was generally very weak (<3-fold and negligible given the limitations of the assay), Table 2. In other words, the phages that escaped restriction were easily modified by the methylation activity of the RM system and ArdB had little if any inhibitory effect on methylation. In three cases, some inhibition of methylation was observed; ArdB_CFT and ArdB_YFJ slightly inhibited methylation by the Type ID RM system and ArdB_YAF slightly inhibited methylation by the Type IC RM system. These results show a clear difference between the activities of ArdB and KlcA compared to the DNA mimics, ocr and ArdA, suggesting that they use different mechanisms for interfering with the Type I RM systems.

It has been shown that ClpXP-dependent regulation of the EcoKI endonuclease activity enables bacteria that acquire unmodified chromosomal target sequences to survive. When cells are under stress after prolonged exposure to treatments that damage DNA, such as exposure to 2-AP, the ClpXP-dependent pathway operates to prevent EcoKI from cleaving the chromosome (71). The restriction subunit of the nuclease (HsdR) is degraded in the presence of ClpXP (71). As described later, we did not observe any direct interaction of KlcA₁₃₆ with the EcoKI in vitro; therefore we tested the hypothesis that the anti-restriction activity of KlcA₁₃₆ was dependent on the ClpXP pathway. A clpX⁻ mutant, NM1041(DE3), was plated in the presence and absence of 80 µg/ml 2-AP, Figure 1. The clpX⁻ mutant was also transformed with the plasmid, pET20b-klcA₁₃₆, expressing klcA₁₃₆. As expected, the clpX⁻ strain with no anti-restriction determinant present, grew well in the absence of 2-AP, but no growth was observed in the presence of 2-AP. In contrast, the clpX⁻ strain carrying klcA₁₃₆ grew equally well in both the presence and absence of 2-AP. This finding indicates that KlcA₁₃₆ is able to block EcoKI restriction without involving the ClpXP protease. We assume that this result holds for ArdB and KlcA in general. Figure 1.

The effect of expressing KlcA₁₃₆ in a 2-AP sensitive strain of E. coli. A single colony of the ClpX⁻ strain NM1041(DE3) harbouring either the KlcA expression construct (pET20b-klcA₁₃₆) or the corresponding vector alone (pET20b) was picked with a loop and spread onto an LB agar plate supplemented with 100 μg/ml carbenicillin, 25 μg/ml kanamycin in the absence (A) or presence (B) of 2-AP (80 μg/ml). The plates were then incubated at 37°C for 14 h.

During purification it was noted that ArdB_YAF and ArdB_CFT behaved anomalously on a large volume (60 cm × 1.6 cm) Superdex 200 size-exclusion chromatography column despite showing only a single homogeneous species of the correct molecular weight on SDS–PAGE (data not shown) and mass spectrometry, Supplementary Table S1. Analysis with a small volume analytical column (25 cm × 0.46 cm) revealed two peaks corresponding to a dimer and a monomer for ArdB_YAF while ArdB_CFT showed two peaks on the large volume column and a broad peak with shoulders on the small volume column, Figure 2. ArdB_YFJ and KlcA₁₃₆ showed only a single monomer peak. These results indicate that there was genuine heterogeneity in the quaternary structure for two of the four proteins and the dependence on column volume for the ArdB_CFT protein further suggested that the apparent switching between the monomer and dimer forms was occurring on the minute timescale of the chromatography process. Figure 2.

An overlay of elution profiles for different proteins from a small volume (25 cm × 0.46 cm diameter) size exclusion chromatography column showing the variable quaternary structure of the four different proteins in 20 mM Tris, 20 mM MES, 0.2 M NaCl, 0.1 mM EDTA, 7 mM β-mercaptoethanol, pH 6.5: ArdB_CFT (black, 4 µM)), ArdB_YAF (blue, 3 µM), ArdB_YFJ (red, 5 µM) and KlcA₁₃₆ (green, 4 µM). The proteins were detected using tryptophan fluorescence emission and are arbitrarily scaled vertically.

A circular plasmid containing a single copy of the target sequence for the EcoKI type I RM enzyme is efficiently cleaved by stoichiometric amounts of EcoKI in the presence of the essential cofactors SAM, ATP and Mg²⁺ resulting in a linearised plasmid (34). Stoichiometric amounts of the ocr and ArdA anti-restriction proteins inhibit this activity by binding strongly to and blocking the DNA binding site on EcoKI (9,12,13). Figure 3 shows that addition of a 20-fold excess of ocr or ArdA, expected to ensure complete inhibition, did indeed prevent any nuclease activity of the EcoKI restriction enzyme. Addition of the same concentration and degree of excess of any of the examined ArdB or KlcA₁₃₆ proteins to the endonuclease reaction showed no observable inhibition of the nuclease activity of EcoKI. Thus the anti-restriction activity clearly observed in vivo was not due to an anti-nuclease activity in contrast to what one might expect from the results of the anti-restriction assay. Hence, none of the ArdB or KlcA₁₃₆ tested appeared able to act as an inhibitor of the Type IA RM enzyme EcoKI in DNA cleavage. The order of addition of the reactants can be important for analysing Type I endonuclease activity. For instance ocr is unable to inhibit EcoKI if it is added after the DNA and ATP (72) so for this reason we added the DNA last to the mixture containing EcoKI, cofactors and anti-restriction protein to avoid this problem. Even with this precaution, ArdB and KlcA₁₃₆ were unable to influence the endonuclease activity of EcoKI. Figure 3.

Nuclease assay to test for inhibition of EcoKI by ArdB proteins. Lane 1, undigested DNA; lane 2, EcoKI digested DNA; lane 3, plus ocr; lane 4, plus Orf18 ArdA; lane 5, plus ArdB_YAF; lane 6, plus ArdB_YFJ; lane 7, plus KlcA₁₃₆; lane 8, plus ArdB_CFT; lane M, DNA ladder with sizes in kb. Plasmid DNA used was non-methylated pBRsk1. Ratio of the anti-restriction proteins (monomers) to EcoKI was 20:1. In each case the reaction mixture minus DNA was made up and incubated for 2 min at room temperature. The reaction was then initiated by adding DNA. Digestion was carried out for 8 min at 37°C before quenching at 68°C for 10 min. CC, closed circular plasmid; L, linear plasmid; OC, open circular plasmid.

Biophysical characterization using ITC and fluorescence spectroscopy performed on two other anti-restriction proteins, ocr and ArdA, have shown a strong interaction of the Mtase part of EcoKI with these proteins (13,16). However, ITC experiments performed at 10 and 25°C showed no apparent binding of KlcA₁₃₆ to Mtase, Figure 4. This result agrees with the in vivo modification assays where inhibition of phage DNA modification was not observed. Detecting the binding to the EcoKI nuclease was not feasible using ITC due to limited amounts of the nuclease. However, given the result of the nuclease inhibition assay, no interaction is expected with the surface of the nuclease. Figure 4.

Raw ITC data showing heat changes for the interaction of KlcA₁₃₆ with the EcoKI Mtase compared with data for the interaction of ArdA with the Mtase as previously published (13). The traces show from top to bottom: Mtase being titrated with ArdA at 25°C, buffer being titrated with Mtase at 25°C, Mtase being titrated with KlcA₁₃₆ at 25°C and Mtase being titrated with KlcA₁₃₆ at 10°C. It is clear that there is no interaction between the Mtase and KlcA₁₃₆.

Since the KlcA and ArdB proteins appear unable to interact neither with the complete EcoKI RM enzyme nor with the core Mtase in vitro, we also examined whether they could interact with the surface of the HsdR restriction subunit of EcoKI that becomes buried on binding to the Mtase. This would interfere with the assembly of the complete RM enzyme as originally proposed by Belogurov and Delver (26). Addition of ArdB or KlcA prior to assembling the Mtase and HsdR into an intact EcoKI endonuclease failed to inhibit the endonuclease activity when DNA was subsequently added to the mixture, Figure 5. This strongly suggests that ArdB and KlcA do not interact with HsdR to interfere with assembly of EcoKI. Figure 5.

The effect of adding ArdB or KlcA during the assembly of the EcoKI RM enzyme. Where relevant, Mtase and the R subunit were preincubated at 25°C in the presence of a × 20 excess of ArdB or KlcA protein for ∼5 min prior to mixing and addition of DNA. Reactions were performed at 37°C for 8 min. Lane 1, uncut pBRsk1 (3 nM); lane 2, pBRsk1 (3 nM) digested with EcoKI (30 nM); lane 3, pBRsk1 digested with reconstituted nuclease (equivalent to 30 nM EcoKI); lane 4, nuclease reconstituted in the presence of ArdB_YAF (20-fold excess over nuclease); lane 5, 20-fold excess ArdB_YFJ; lane 6, 20-fold excess KlcA₁₃₆; lane 7, pBRsk1 in the presence of R subunit only; lane 8, pBRsk1 in the presence of Mtase only. M, 1 kbp size marker.

The recorded ¹H-¹⁵N HSQC spectrum, Figure 6, was consistent with that of a folded protein exhibiting well dispersed signals and very little spectral overlap. Backbone and side-chain resonances of KlcA₁₃₆ were assigned based on 2D and 3D triple-resonance experiments. Peaks in the 3D ¹³C and ¹⁵N edited NOESY spectra were picked and the NOE cross-peak assignments were obtained from the automated iterative assignment program CANDID that works in conjunction with 3D structure calculations in the program CYANA 2.1. Using NOE restraints, TALOS dihedral restraints and ¹H-¹⁵N RDCs (Supplementary Figure S1), the lowest energy ensemble of 20 structures of KlcA₁₃₆ were calculated (Supplementary Figure S2). The calculations converged well, yielding root mean square deviations of 0.63 and 0.65 for backbone and all heavy atoms in structured regions (residues 7–142). Only a subset of RDCs, as specified in the ‘Methods’ section for the calculation of the rotational diffusion tensor, was used in the structure calculations. The structures were validated using PROCHECK and WHATIF. In the Ramachandran plot, 98.5% of residues appeared in the most favourable and additionally allowed regions. WHATIF packing analysis yielded an average quality control z-score of −1.432 suggesting a good quality NMR structure. Figure 6.

2D ¹H-¹⁵N HSQC spectrum of KlcA₁₃₆ at 800 MHz.

The KlcA₁₃₆ protein structure is made up of a single domain (73) containing five α-helices (labelled 1–5), a single 3₁₀-helix and five β-strands (labelled A–E) and belongs to the mixed α/β class of proteins, Figure 7. The five β-strands form two separate beta-sheets, the first composed of anti-parallel β-strands A, B and C and the other made up of anti-parallel β-strands D and E. β-Strand A usually occurs as an isolated β-bridge from the ensemble consensus. The two β-sheets closely flank the C-terminal three α-helices (3–5) on either side. These three helices appear to run in an almost anti-parallel fashion to their helical axes in a classical knobs-in-holes arrangement of a three-helical bundle. Helix 3 is the longest of the helices located at the centre of the fold and contains several hydrophobic amino acids at its N-terminus that are deeply buried. A stereo view of the ensemble of the 20 lowest energy structures of KlcA136 is shown in Supplementary Figure S2. Figure 7.

KlcA₁₃₆ secondary and tertiary structure. Colour code: alpha-helices, red; beta-strands, yellow and 3₁₀ helix, purple. (A) Secondary structure elements identified by STRIDE for the ensemble consensus shown as a topology fold schematic (55–57). (B) A cartoon representation of the closest-to-mean 3D structure in the ensemble. Note: loops are rendered smooth for clarity.

An electrostatic surface representation generated by GRASP shows that the surface of KlcA₁₃₆ is predominantly negatively charged, consistent with its theoretical pI (4.76), Figure 8. The negatively charged side-chain residues on the KlcA₁₃₆ surface reveals two distinct clusters (i) an ‘Aspartate cluster’ made up of five linearly arranged residues i.e. D61 (in the loop between β-strands B and C), D110, D112, D115, D119 (in the loop between α-helix 3 and 4 and on α-helix 4) and (ii) a ‘Glutamate cluster’ that contains amino acid residues E47 (in the loop between α-helix 2 and β-strand B), E75 and E77 (on β-strand D), E86 (on β-strand E) and E133 (on α-helix 5). In addition to the five residues present in the ‘Glutamate cluster’ are residues E15, D69 and D90, which are also located nearby in 3D space. A striking linear arrangement of positive charge involving arginine side-chains is also evident, predominantly on the opposite side of the KlcA₁₃₆ protein surface with respect to its negatively charged clusters. This ‘Arginine cluster’ contains seven residues: R29, R33, R43, R44 on α-helix 2, R48 in the loop between α-helix 2 and β-strand B, R84 on β-strand E and R120 on α-helix 4. Even though three distinct regions of charge are apparent from the electrostatic surface representations, only a handful of the charged residues are conserved in the 43 non-redundant sequences identified (Supplementary Figure S5). These include R18, E35, D61, D90, R125, E133 and D142. Figure 8.

Electrostatic surface representation of the KlcA₁₃₆ protein. Two views rotated by 180° about the y-axis of a GRASP electrostatic surface representation of the KlcA₁₃₆ protein. The molecule appears to expose many charged residues, labelled. Negative charge is coloured red and positive charge coloured blue, ranging from −10 k_BT to +10 k_BT (k_B: Boltzmann constant; T: temperature in Kelvin).