Supporting InformationGhasriani et al. 10.1073/pnas.1007141107SI Materials and MethodsIn Vitro MinD ATPase Assay. Reactions contained 2.7 μM Ng-MinDin 65 mM Tris HCl pH 8.2, 50 mM KCl, 80 mMNaCl, 1 mMATP,5 mM MgCl2 with 500 μg∕mL 1,2-dioleoyl-sn-glycero-3-[phos-phor-rac-(1-glycerol)] phospholipids (Avanti Polar Lipids). Con-centrations of purified Ng-MinE and Ng-MinD were determinedusing a bicinchoninic acid assay (Pierce Biotech). Concentrationsof WTand mutant Ng-MinE peptide stock solutions (GenescriptUSA Inc.) were determined by amino acid analysis (AdvancedProtein Technology Centre, Hospital for Sick Children). Forthe full-length Ng-MinE samples, average ATP hydrolysis ratesstimulated by 0.06 μM of the WTor indicated mutant were mea-sured using at least two different protein preparations.

The specific Ng-MinD activity plotted as a function of increas-ing concentration for full-length, or N-terminal fragments of,Ng-MinE generally showed a sigmoidal appearance. Assumingthat the rate of Ng-MinD-stimulated ATP hydrolysis is directlyproportional to the amount of MinE binding, then the ratio ofthe measured rate (vo) versus the maximal rate (Vmax) of MinD-stimulated ATP hydrolysis can be considered to approximate thefraction of available MinE-binding sites that are occupied (θ).Changes in the fraction of occupied sites for cooperative bindingsystems can generally be described by the Hill equation:

θ ¼ ½L�hoKh

0.5 þ ½L�ho;

where K0.5 is the concentration of ligand (L) required to reachhalf-maximal saturation and h is an interaction coefficient thatcan reflect the level of MinE-binding cooperativity. In our systemθ ¼ vo

Vmaxand ½L�o is the concentration of Ng-MinE used. Non-

linear least squares fitting of MinE-stimulated MinD ATP hydro-lysis rates allowed determination of Vmax, K0.5, and h.

Paramagnetic Relaxation Enhancement Experiments. Ng-MinE hasno native cysteine residues, allowing introduction of single-cysteine mutations for site-specific derivatization by sulfhydril-reactive paramagnetic probes. Ng-MinE double mutants E46A/A31C and E46A/E83C were generated for this purpose, becauseboth Ala31 and Glu83 are solvent accessible, proximal to regionsof structural interest, and do not participate in structurally impor-tant interactions. 15N-labeled samples were prepared as done forother Ng-MinE proteins except that 1 mM β-mercaptoethanolwas included in the nickel affinity purification buffers, and1 mM DTT in the size exclusion chromatography purificationbuffers. 1H-15N heteronuclear single quantum coherence(HSQC) spectra were acquired to confirm that only residuesin close proximity to the mutated site showed chemical shiftperturbations, and therefore that no structural change was causedby these mutations. Chemical shift assignments were obtainedusing previously assigned spectra for E46A (1).

After removal of DTT from each sample by buffer exchangein an ultracentrifugation unit, a 10% (wt∕vol) solution of thespin-label (1-oxyl-2,2,5,5-tetramethyl-3-pyrroline-3-methyl)methanethiosulfonate (MTSL, Toronto Research Chemicals) inacetonitrile was added to a 10∶1 spin label∶Ng-MinE molar ratio.The final concentration of acetonitrile in the sample was typicallyless than 1% (vol∕vol). After overnight incubation at roomtemperature, excess spin label was removed by buffer exchangeon a NAP-10 gel filtration column (GE Healthcare Life Sciences)equilibrated in 25 mM Tris pH 7.2, 50 mM NaCl, 200 μM EDTA,200 μM benzamidine, and 0.02% NaN3 in 10% D2O.

1H-15NHSQC spectra of 350-, 500-, and 700-μMparamagneticspin-labeled mutant samples were acquired. The spin label wassubsequently reduced with a 2- to 4-fold molar excess of ascorbicacid and a spectrum of the reduced sample was recorded. Peakfitting was carried out with the nlinLS module in NMRPipe (2)using a Gaussian model, and the ratio of peak intensities in para-magnetic and diamagnetic spectra were determined. Measuredintensity ratios were normalized using the three highest intensityratios as has been done previously (3), to correct for decreasedpeak intensities in reduced spectra due to a small amount ofprecipitation that sometimes occurred during acquisition. Inten-sity ratios for the three samples were averaged and classified asshowing strong (intensity ratio <0.2), intermediate (0.2–0.8), orweak (>0.8) broadening, and mapped onto the structures shownin Fig. S2.

Amide Exchange Experiments. Slowly exchanging amides weremonitored in 2H∕1H exchange experiments that were initiatedby buffer exchange of a 0.7 mM Ng-MinE E46A sample into25 mM Tris pH 7.2, 50 mM NaCl, 0.1 mM EDTA, 99.9%D2O using a NAP-10 column. The sample temperature was heldat 303 K while 14 1H-15NHSQC spectra were recorded over threedays. The decay of integrated peak volumes was fit using non-linear least-squares fitting to extract proton exchange constants.Rapidly exchanging amides were monitored in the CLEANEX-PM (phase-modulated CLEAN chemical EXchange) 1H∕1Hexchange experiments in the same buffer with 10% D2O at303 K. A series of 10 spectra were collected with CLEANEXmix-ing times ranging from 5 to 90 ms and normalized peak volumeswere used to extract proton exchange constants. Amide protonprotection factors were calculated by dividing intrinsic exchangerate constants estimated by the program SPHERE (http://www.fccc.edu/research/labs/roder/sphere/) from experimentally deter-mined proton exchange constants.

Fast Motion Dynamics and Model-Free Analysis. Standard pulsesequences (4) were used for measurements of heteronuclearsteady-state f1Hg15NNOE, longitudinal (15N R1), and transverse(15N R2) relaxation rates for 15N-labeled Ng-MinE (E46A) at298 K, and a magnetic field corresponding to 500 MHz 1HLarmor frequency. The relaxation delays used in the T1 experi-ments were 10, 70, 140, 250, 360, 530, 750 and 1150 ms, and theones in the T2 experiments were 10, 30, 50, 70, 90, 110, 130, and150 ms. Two sets of experiments, with interscan delays of 1 and3 s were recorded. Relaxation rates were determined by nonlinearleast-squares fits of data assuming a monoexponential decay andthe errors in the rates were estimated from the recorded dupli-cates. The f1Hg15N NOE was measured from pairs of spectrawith and without 1H saturation during a 3-s recycle delay. Relaxa-tion data were fitted to the extended Lipari–Szabo spectral den-sity function (5–7) employing TENSOR2 (8) using an axiallysymmetric anisotropic model with axial and equatorial rotationaldiffusion tensor components of Djj ∼ ð1.7� 0.1Þ × 107 s−1 andD⊥ ∼ ð1.2� 0.1Þ × 107 s−1, respectively. The relative dimensionsof this tensor agree exactly with structure-based predictions usingHYDRONMR with Djj∕D⊥ ¼ 1.4 (9).

Analytical Ultracentrifugation. Sedimentation velocity experimentswere performed at the Protein Function Discovery Facility atQueen’s University on a Beckman Optima XL-I instrumentequipped with an An-Ti 60 rotor spun at 50,000 rpm, 293 K.Samples containing 110 μM WT Ng-MinE in 50 mM Tris pH

Ghasriani et al. www.pnas.org/cgi/doi/10.1073/pnas.1007141107 1 of 6

8.0, 30 mM NaCl, or 3 μM Ng-MinE in 25 mM Tris pH 8.5,100 mM NaCl, 0.1 M EDTA were monitored with 150 scansby absorbance at 280 nm, or 450 scans by interference optics,

respectively. Sedimentation profiles were generated by fittingthe data to a continuous distribution cðSÞ Lamm equation modelusing the program SEDFIT (10).

1. Ducat T, Goto NK (2010) 1H, 13C, 15N chemical shift assignments for the Neisseriagonorrhoeae MinE regulator of cell division septum placement. Biomol NMR Assign,10.1007/s12104-010-9247-4.

2. Delaglio F, et al. (1995) NMRPipe: A multidimensional spectral processing system basedon UNIX pipes. J Biomol NMR 6:277–293.

3. Liang B, Bushweller JH, Tamm LK (2006) Site-directed parallel spin-labeling andparamagnetic relaxation enhancement in structure determination of membraneproteins by solution NMR spectroscopy. J Am Chem Soc 128:4389–4397.

4. Farrow NA, et al. (1994) Backbone dynamics of a free and phosphopeptide-complexedSrc homology 2 domain studied by 15N NMR relaxation. Biochemistry 33:5984–6003.

5. Clore MG, et al. (1990) Deviations from the simple two-parameter model-freeapproach to the interpretation of nitrogen-15 nuclear magnetic relaxation of pro-teins. J Am Chem Soc 112:4989–4991.

6. Lipari G, Szabo A (1982) Model-free approach to the interpretation of nuclearmagnetic resonance relaxation in macromolecules. 1. Theory and range of validity.J Am Chem Solc 104:4546–4559.

7. Lipari G, Szabo A (1982) Model-free approach to the interpretation of nuclearmagnetic resonance relaxation in macromolecules. 2. Analysis of experimental results.J Am Chem Soc 104:4559–4570.

8. Dosset P, Hus JC, Blackledge M, Marion D (2000) Efficient analysis of macromolecularrotational diffusion from heteronuclear relaxation data. J Biomol NMR 16:23–28.

9. de la Torre JC, Huertas ML, Carrasco B (2000) HYDRONMR: Prediction of NMR relaxa-tion of globular proteins from atomic-level structures and hydrodynamic calculations.J Magn Reson 147:138–146.

10. Schuck, P (2000) Size-distribution analysis of macromolecules by sedimentationvelocity ultracentrifugation and Lamm equation modeling. Biophys J 78:1606–1619.

Fig. S1. Sample strips from the 13C-filtered/edited NOESY (1) recorded on a 12C∕13C-labeled Ng-MinE heterodimer showing intermolecular NOEs between theN-terminal αA helix and the β-sheet. Negative intensities arise from peaks that are folded in the carbon dimension. The corresponding region of the structure ishighlighted on the right. Note that the Ala27 side chain is directly under helix αA in this view, on the same strand and directly above Ile25.

1. Zwahlen C, et al. (1997) Methods for measurement of intermolecular NOEs by multinuclear NMR spectroscopy: Application to a bacteriophage λ N-peptide/boxB RNA complex. J AmChem Soc 119:6711–6721.

Ghasriani et al. www.pnas.org/cgi/doi/10.1073/pnas.1007141107 2 of 6

Fig. S2. Paramagnetic relaxation enhancement confirms the Ng-MinE fold contains β1 in the dimeric interface and an N-terminal α-helix interacting with theβ-sheet. Mutations to cysteine weremade either to Ala31 (Top) or Glu83 (Bottom), and samples were derivatizedwith the sulfhydril-reactiveMTSL spin label. Theratio of peak intensities in 1H-15N HSQC spectra recorded before and after reduction of the spin label provide a measure of the distance between each amideproton and the spin label (1). Residues with amide peak intensity ratios <0.2 that would be within 15 Å of the nitroxide spin label are colored in red, 0.2–0.8(∼15–20 Å) in yellow, and >0.8 (>20 Å) in green for the Ng-MinE structure (two views shown on the left), or for homologous residues in the Escherichia coli-topological specificity domain (Ec-TSD) structure (two views shown on the right). Residues for which broadening could not be accurately measured due to peakoverlap or missing intensities are shown in gray. Secondary structure elements homologous to those determined in the Ng-MinE structure are labeled in theEc-TSD structure. Regions of the Ec-TSD structure that are not consistent with the spin-label induced broadening are highlighted in blue. (For clarity, only one ofthe two symmetric sites in the dimer is highlighted.) Specifically, for A31C, strong broadening was observed for a cluster of residues that would be at least∼20 Å away from the spin label in the TSD structure, but only 10 Å away in the Ng-MinE structure. For spin-labeled E83C, residues 32–35 are ∼20 Å away fromthe side chain at position 83 in the Ec-TSD structure, but were strongly broadened. In the Ng-MinE structure these residues are 10 Å or closer to the Glu83 sidechain. Also, residues 68–71 showed weak to moderate broadening, but are within ∼10 Å of the side chain atoms of residue 83 in the Ec-TSD structure. In theNg-MinE structure these residues are within 15 to 20 Å of this side chain, which again is consistent with the broadening observed. Overall, these results provideindependent confirmation that the topological specificity domain dimer interface involving β3 is not formed in the structure of the full-length protein. Allstructure figures were prepared with MOLMOL (2).

1. Battiste JL, Wagner G (2000) Utilization of site-directed spin labeling and high-resolution heteronuclear nuclear magnetic resonance for global fold determination of large proteinswith limited nuclear overhauser effect data. Biochemistry 39:5355–5365.

2. Koradi R, Billeter M, Wüthrich K (1996) MOLMOL: A program for display and analysis of macromolecular structures. J Mol Graph 14:51–55.

Fig. S3. Three views of an overlay of backbone traces for the Hp-MinE structure in red (PDB ID code 3KU7) and the lowest energy member of the Ng-MinEensemble (PDB ID code 2KXO) in blue. The Cα rmsd between our NMR ensemble and the 2.8-Å resolution Hp-MinE crystal structure was 1.73� 0.05 Å (calcu-lated using regular secondary structure elements Arg21-Arg30, Leu40-Tyr54, Asn61-Gln66, and Asp73-Leu81 in Ng-MinE; and Arg19-Arg28, Met35-Tyr49, Asp54-Thr59,and Glu67-Leu75 in Hp-MinE ). The first 18 and last 7 residues of the Ng-MinE structure were removed to facilitate visualization. Electron density that is missingfor Hp-MinE residues 68–70 are indicated with dashed lines (Middle). The view on left has C termini in front, whereas the view on the right has the N termini infront. (Middle) The β2 strand from Hp-MinE diverges significantly from that of the Ng-MinE structure, possibly due to crystal packing contacts involving β3, orthe absence of the N-terminal α-helix that was found to interact with this part of the sheet in the Ng-MinE structure.

Ghasriani et al. www.pnas.org/cgi/doi/10.1073/pnas.1007141107 3 of 6

Fig. S4. Sequence alignment from Clustal W (1) for MinE proteins from a range of pathogens, with residues showing 100% identity highlighted in orange.Dark yellow highlights the most commonly occurring residue at each position, and light yellow highlights are homologous residues at these positions.Corresponding secondary structure elements found in the Ng-MinE structure are shown on the top, and for Hp-MinE on the bottom. Residues we identifiedas important for stimulation of MinD-catalyzed ATP hydrolysis are indicated with asterisks. Species abbreviations are Ng, Neisseria gonorrhoeae; Ec Escherichiacoli; Se, Salmonella enterica; Kp, Klebsiella pneumoniae; Yp, Yersinia pestis; Pm, Proteus mirabilis; Vc, Vibrio cholerae; Pa, Pseudomonas aeruginosa; Lp,Legionella pneumophila; Bp, Bordetella pertussis; Ft, Francisella tularensis; Ab, Acinetobacter baumannii; Hp, Helicobacter pylori.

1. Larkin MA, et al. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23:2947–2948.

Fig. S5. Log amide proton protection factors (PF), mapped onto the ribbon structure of Ng-MinE E46A that were determined by 2H∕1H exchange experiments(PF > 6.5, purple, PF > 5.5 dark blue, PF > 4.5 light blue) or 1H∕1H experiments in H2O (PF < 2 dark orange, PF < 1 light orange). Residues that undergoexchange at rates that were too rapid to detect by 2H∕1H exchange and too slow to measure in 1H∕1H exchange experiments are in gray, providing indirectevidence for protection factors in the range of 2–4.5 for these regions of the structure. Proline residues 37, 41, and 82 are also in gray. Protection factors are inexcellent agreement with the 1H-15N HSQC spectrum recorded at pH 9.5 (1), because residues having PF < 2 were also not detected in the high pH spectrum.

1. Ramos D, et al. (2006) Conformation of the cell division regulator MinE: Evidence for interactions between the topological specificity and anti-MinCD domains. Biochemistry45:4593–4601.

Fig. S6. Malachite green assay results for ATP hydrolysis stimulated by 3.0 μMWTor 20 μMmutant Ng-MinE1-22, or 20 μMNg-MinE1–17 peptide. Absorbance at620 nm was measured for aliquots that were removed at four different time points, centrifuged to remove the lipid vesicles, mixed with the malachite greenreagent, and allowed to develop for 15 min prior to taking the absorbance reading. Only theWT peptide showed an increased rate of ATP hydrolysis over basallevels.

Ghasriani et al. www.pnas.org/cgi/doi/10.1073/pnas.1007141107 4 of 6

Fig. S7. Backbone amide 15N relaxation values measured for Ng-MinE (E46A) at 298 K. 15N R1 and R2 relaxation rates (A and B), 1H-15N heteronuclear NOEs (C),and generalized order parameters, S2 (D), are shown as a function of residue number. Residues at the C terminus and in the loop between αA and β1 showedthe presence of significant dynamics on the picosecond to nanosecond time scale. Order parameter and heteronuclear NOE values for residues in the N-term-inal αA helix are similar to those obtained for the main body of the structure, confirming the stable association of αA with the β-sheet.

Fig. S8. Sedimentation velocity experiment showing a predominantly dimeric state for WT Ng-MinE at 110 μM (red), and 3 μM (blue). The sedimentationcoefficient for the main species is 1.91–1.92 S, corresponding to a molecular weight of 22.0–22.2 kDa. The expected molecular mass for the His-tagged MinEdimer is 22.2 kDa.

Ghasriani et al. www.pnas.org/cgi/doi/10.1073/pnas.1007141107 5 of 6

Table S1. Statistics for Ng-MinE structure

Nonredundant distance restraintsLong range* ði − jÞ ≥ 5 726Medium range, 1 < ði − jÞ ≤ 4 520Sequential, ði − jÞ ¼ 1 744Intraresidue, i ¼ j 620Total 2610Hydrogen bond restraints 144Dihedral angle constraints 216Backbone NH RDCs 86Average atomic RMSD (Å)†

Pairwise Versus meanBackbone (N, Cα, C′) 0.72 ± 0.21 0.51 ± 0.14Heavy atoms 1.28 ± 0.24 0.91 ± 0.14Deviations from idealized covalent geometry‡

Bond length rmsd (Å) 0.004 ± 0.000Torsion angle rmsd (°) 0.462 ± 0.031Impropers (°) 0.401 ± 0.051WHAT IF quality indicators (9)rms Z-score bond length 0.487rmsd bond length (Å) 0.011rms Z-score bond angles 0.793rmsd Bond angles (°) 1.542Constraint violationsNOE violations >0.5 Å 0NOE rmsd (Å) 0.022 ± 0.002Torsion angle violations >5° 0Torsion angle rmsd (°) 0.255 ± 0.085NH RDC violations >2 Hz 0NH RDC rmsd (Hz) 0.291 ± 0.038Ramachandran statistics (%)§

Most favored 89.8Additionally allowed 10.2Generously allowed 0Disallowed 0

*Two hundred four restraints of the long-range distance restraints are intermolecular.†The 10 conformers in the ensemble were aligned with respect to regular secondary structures elements,residues 3–8, 21–31, 40–54, 59–69, and 72–81.

‡Final force constants in Xplor-NIH were 50 kcalmol−1 Å−2 for NOE and hydrogen bond square-wellpotentials and 5 kCalmol−1 Hz−2 for RDCs. A noncrystallographic symmetry potential with final forceconst of 10 kcalmol−1 Å−2 was employed. Default values from the standard simulated annealing andrefinement protocols were used for all other constraints.

§Ramachandran statistics (10) is for the regular secondary structure elements.

Ghasriani et al. www.pnas.org/cgi/doi/10.1073/pnas.1007141107 6 of 6