Structure of the Mycobacterium Tuberculosis Ompatb Protein: a Model of an Oligomeric Channel in the Mycobacterial Cell Wall

proteins STRUCTURE O FUNCTION O BIOINFORMATICS

Structure of the Mycobacterium tuberculosis OmpATb protein: A model of an oligomeric channel in the mycobacterial cell wall Yinshan Yang,1,2 Daniel Auguin,1,2,3 Ste´phane Delbecq,4 Emilie Dumas,1,2 Ge´rard Molle,1,2 Virginie Molle,5 Christian Roumestand,1,2* and Nathalie Saint1,2

1 Centre de Biochimie Structurale, CNRS UMR 5048, Universite´ Montpellier 1 et 2, F34090 Montpellier, France 2 INSERM U554 F34090 Montpellier, France 3 INRA, USC2030 ‘Arbres et Re´ponses aux Contraintes Hydrique et Environnementales’ (ARCHE), F-45067 Orle´ans Cedex 02, France 4 Dynamique des Interactions Membranaires Normales et Pathologiques, CNRS UMR 5235, Universite´ Montpellier 2, F34095 Montpellier Cedex 5, France 5 Institut de Biologie et de Chimie des Prote´ines, Universite´ de Lyon, CNRS UMR 5086, F69367 Lyon, France

ABSTRACT INTRODUCTION

The pore-forming outer membrane protein The etiological agent of tuberculosis (TB), Mycobacterium OmpATb from Mycobacterium tuberculosis is a viru- tuberculosis, causing nearly 2 million deaths per year, is presently lence factor required for acid resistance in host one of the most important infectious agents implicated in mortal- phagosomes. In this study, we determined the 3D ity worldwide. TB has emerged as a major public health threat structure of OmpATb by NMR in solution. We because of a significant increase in multiple-drug-resistant TB and found that OmpATb is composed of two independ- synergism between human immunodeficiency virus and M. tuber- ent domains separated by a proline-rich hinge culosis infection.1,2 One of the principle problems in TB therapy region. As expected, the high-resolution structure of is the slow uptake of drugs across the thick mycobacterial cell the C-terminal domain (OmpATb198–326) revealed a wall made of unique lipid and glycolipid moities.3,4 This cell wall module structurally related to other OmpA-like pro- is extremely hydrophobic and forms an exceptionally strong per- teins from Gram-negative bacteria. The N-terminal meability barrier, 100-fold less permeable than the outer mem- domain of OmpATb (73–204), which is sufficient to brane of Escherichia coli.5,6 Water-filled protein channels, called form channels in planar lipid bilayers, exhibits a porins, are considered to represent the main pathway for entry of fold, which belongs to the a1b sandwich class fold. hydrophilic drugs through the outer membrane of Gram-negative Its peculiarity is to be composed of two overlapping 7 subdomains linked via a BON (Bacterial OsmY and bacteria. Porin-like proteins have also been identified in the cell wall of several mycobacterial species such as M. chelonae,8 M. Nodulation) domain initially identified in bacterial 9 10 5,11 12 proteins predicted to interact with phospholipids. phlei, M. smegmatis, M. bovis BCG, and M. tuberculosis Although OmpATb73–204 is highly water soluble, by assaying organic solvent- or detergent-solubilized cell wall frac- current–voltage measurements demonstrate that it tions for channel-forming activity in artificial lipid bilayers. In is able to form conducting pores in model mem- most cases, the reported studies were in agreement concerning branes. A HADDOCK modeling of the NMR data the paucity of porins in mycobacterial membranes, which could gathered on the major monomeric form and on the contribute to the low permeability observed in vivo. The low level minor oligomeric populations of OmpATb73–204 sug- of mycobacterial porins in solubilized cell wall fractions has also gest that OmpATb73–204 can form oligomeric rings impeded their biochemical and structural characterization. How- able to insert into phospholipid membrane, similar ever, the most advanced characterization of a mycobacterial porin to related proteins from the Type III secretion sys- tems, which form multisubunits membrane-associated rings at the basal body of the secretion ma- Additional Supporting Information may be found in the online version of this article. Yinshan Yang and Daniel Auguin contributed equally to this work. chinery. Daniel Auguin’s current address is Universite´ d’Orle´ans, UFR-Faculte´ des Sciences, Laboratoire de Biologie des Ligneux et des Grandes Cultures, UPRES EA 1207, rue de Chartres, BP 6759, 45067 Proteins 2011; 79:645–661. Orle´ans Cedex 02, France. VC 2010 Wiley-Liss, Inc. *Correspondence to: Christian Roumestand, Centre de Biochimmie Structurale, 29 rue de Navacelles, 34090 Montpellier Cedex 5, France. E-mail: [email protected]. Key words: membrane protein; porin; NMR struc- Received 27 July 2010; Revised 23 September 2010; Accepted 24 September 2010 Published online 12 October 2010 in Wiley Online Library (wileyonlinelibrary.com). ture; oligomeric assembly; HADDOCK calculation. DOI: 10.1002/prot.22912

VC 2010 WILEY-LISS, INC. PROTEINS 645 Y. Yang et al. was achieved on the MspA protein representing the channels with similar pore properties to those of entire major general diffusion pathway for hydrophilic com- OmpATb were observed, suggesting that the 72 N-termi- pounds in the fast growing Mycobacterium smegmatis.13 nal proximal residues are not necessary for the pore- In 2004, Faller et al.14 succeeded in resolving the 3D forming activity. Similarly, it was shown that the C-ter- structure of this mycobacterial porin by using a recombi- minal part of OmpATb was not essential for the channel nant MspA protein expressed in E. coli cells. The MspA activity because an OmpATb73–220 construct was able to crystal structure revealed a homooctameric goblet-like form pores in lipid bilayers.22 In a very recent report, architecture with a single central channel of 10 nm in the pore-forming activity of OmpATb was questioned length. This structure is completely different from that of because the protein is apparently not folded in a beta- the known trimeric porins of Gram-negative bacteria, barrel.25 which have one channel per monomer and are 4nm In this study, with the goal of characterizing the archi- long.15–17 Homologs of MspA have been identified in tecture of OmpATb responsible for its channel activity, other rapid-growing mycobacterial species such as M. we carried out NMR studies on the soluble form of the phlei but not in the slow-growing mycobacteria M. tuber- protein lacking the first 72 N-terminal residues, 18 culosis and M. bovis BCG. OmpATb73–326. Based on the complete NMR assignment In the slow-growing pathogen M. tuberculosis, a pro- of this 254-residue protein, extensive analysis of NOESY tein with significant sequence homology to the major spectrum revealed that OmpATb is composed of two in- outer membrane protein OmpA of E. coli was identified dependent domains separated by a proline-rich hinge about 10 years ago.19 This protein, OmpATb, has been region. The C-terminal module is structurally related to shown to form ion channels in planar lipid bilayers and other OmpA-like proteins, whereas the structure of the was considered as a porin-like protein from M. tuberculo- N-terminal domain displays an original fold unexpected sis.20 Furthermore, it was proposed that OmpATb corre- for a porin-like proteins family member. These results sponds to the major functioning porin at low pH are in agreement with the very recent NMR study of because deletion of the ompATb gene caused a decrease OmpATb proposed by Teriete et al.,25 which appeared in of the permeability to several small water-soluble sub- the literature during the drafting process of this article. stances under reduced pH conditions.21 Recent electro- The authors report the characterization of the secondary physiological studies with recombinant OmpATb pro- structure and the dynamics of the two domains duced in E. coli demonstrated that the channel activity of (OmpATb73–220 and OmpATb196–326) as well as the OmpATb is modulated by pH.22 It was also observed detailed 3D structure of the N-terminal domain that OmpATb channels exhibit more frequent and more (OmpATb73–220). In addition to this previously reported prolonged closure events at acidic pH. This particular structure of the N-terminal domain, we report here the behavior of OmpATb at low pH conditions was proposed 3D structure of the C-terminal domain, apparently to be beneficial to M. tuberculosis survival in the mildly related to the crystal structure of the OmpA-like domain acidic environment encountered in the phagocytotic of the RmpM protein from N. meningitidis.26 The struc- vacuole of host macrophages. Moreover, it was shown ture described in these two different studies cannot that OmpATb appears to be expressed only in pathogenic explain the porin activity of OmpATb: this highly soluble species (i.e., the members of the M. tuberculosis complex) protein is unlikely to insert into phospholipidic mem- underscoring its role in the virulence of these mycobacte- branes as is. From careful inspection of the gel chroma- rial strains.22 tography elution profile of the recombinant protein, we Since its identification in the genome of M. tuberculo- found that oligomeric minor species of OmpATb N-ter- sis, the structural characterization of OmpATb has been a minal domain coexist in solution with the monomeric crucial objective. Initial circular dichroism experiments major species. Based on NMR data combined with HAD- revealed that OmpATb presents a significant a-helical DOCK (High Ambiguity Driven DOCKing) bases model- and b-sheet content, in agreement with that was ing calculations, we propose here a multimeric model of observed for OmpA of E. coli.20,23 More recent studies OmpATb that can form a pore inside the cell wall of M. revealed that the N-terminal proximal residues of tuberculosis. OmpATb are essential to target the protein to the membrane. Indeed, the production of a recombinant MATERIALS AND METHODS OmpATb73–326 protein in mycobacterial strains such as M. smegmatis and M. bovis yielded to the expression of Bacterial strains the protein essentially in the cytosol of the mycobacte- 24 ria. Small amounts of OmpATb73–326 could be Strains used for cloning and expression of recombinant expressed in the membrane when the protein construct proteins were E. coli DH5a (Clontech laboratories, Palo was fused to the E. coli OmpA signal peptide.22 Addi- Alto, CA) and E. coli BL21(DE3)omp8, a universal tionally, when this truncated protein was purified from expression host lacking the major E. coli outer membrane the membrane in the presence of detergent, functional proteins LamB, OmpA, OmpC, and OmpF.27

646 PROTEINS Structure of OmpATb from M. tuberculosis

Amplification and cloning of OmpATb73–326, and glucose as nitrogen and carbon source, respectively, OmpATb73–204, OmpATb1–204, and as well as ampicillin and kanamycin were inoculated with OmpATb 198–326 the recombinant strains. The fermentation was conducted 8 The 765-bp ompATb73–326 gene fragment, encoding a at 37 C in fed-batch mode with the pH maintained at truncated OmpATb protein lacking the first 72 N-termi- pH 7.0 and at 30% of dissolved oxygen. According to the 15 13 nal residues, with appropriate sites at both ends, was syn- label needed, NH4Cl or [ C]glucose has been used in thesized by PCR amplification using M. tuberculosis place of nonlabeled compound. After induction with H37Rv genomic DNA as a template, with the following IPTG (final concentration 0.5 mM) during 4 h, cells were primers: #241, 50-TATGGATCCGGCGCTTCT GCGTTGT harvested by centrifugation, disrupted in lysis buffer [50 CCTTG-30 (containing a BamHI site in bold); #242, 50- mM Tris-HCl, pH 8.0, 150 mM NaCl, 10% glycerol, 1 TATAAGCTTTTAGTTGACCACGATCTCGACGCGAC-30 mM mercaptoethanol, and antiproteases cocktail (containing a HindIIIsiteinboldandastopcodon (Roche)] using lysozyme and sonication, and the clarified underlined). The 396-bp ompATb73–204 gene fragment, supernatant was incubated with Ni-NTA agarose suspen- encoding a truncated OmpATb protein lacking the first sion (Qiagen). The protein–resin complex was packed 72 N-terminal residues and the last 121 C-terminal resi- into a column and washed extensively with 20 mM Tris- dues, with appropriate sites at both ends, was synthesized HCl, pH 8.0, 150 mM NaCl, and 10 mM imidazole. The by PCR amplification using M. tuberculosis H37Rv protein–resin complex in the same buffer is then treated genomic DNA as a template, with the following primers: with thrombin (Amersham) overnight. The proteins were #241 and #553, 50-TATAAGCTTTTAGGCCGGGGGTCCT eluted with 20 mM Tris-HCl, pH 8.0, 150 mM NaCl and GGCG GTGCCTG-30 (containing a HindIII site in bold concentrated using Centriprep with a 10-kDa cutoff and a stop codon underlined). The 612-bp ompATb1–204 (Millipore, France) before being loaded onto a gel filtra- gene fragment, encoding a truncated OmpATb protein tion column (Superdex 75). The permeation gel buffer lacking the last 121 C-terminal residues, with appropriate was 20 mM NaPi, pH 7.5, 50 mM NaCl. The eluted sites at both ends, was synthesized by PCR amplification fractions were analyzed by SDS-PAGE, and purified pro- using M. tuberculosis H37Rv genomic DNA as a template, tein was concentrated before being processed for NMR with the following primers: #243, 50-TATGGATCC analysis. GTGGCTTCTAAGGCGGGTTTG-30 (containing a BamHI site in bold and an initiation codon underlined) and #553. Dynamics light scattering experiments The 390-bp ompATb198–326 gene fragment, encoding a truncated OmpATb protein lacking the first 197 N-terminal Dynamics light scattering (DLS) experiments were residues, with appropriate sites at both ends, was synthe- recorded on a MALVERN Instrument Zetasizer Nano-S sized by PCR amplification using M. tuberculosis H37Rv spectrophotometer and processed with the Dispersion genomic DNA as a template, with the following primers: Technology software 4.20. #565, 50-TATGGATCCGCACCGCCAGGACCC CCGGCC- 0 3 (containing a BamHI site in bold) and #242. The PCR NMR spectroscopy products were digested with BamHI and HindIII and ligated into the pETSIG vector previously digested with the All NMR experiments were carried out on Bruker 500 same enzymes, thus yielding, respectively, pETSIG- and 600 MHz spectrometers equipped with 5-mm z- 1 13 15 ompATb73–326,pETSIG-ompATb73–204,pETSIG-ompATb1–204, shielded H, C, N triple-resonance cryogenic probes and pETSIG-ompATb198–326. The pETSIG vector is a at several temperatures ranging from 293 to 308 K. Sam- pET28a (Novagen) derivative that allows the His-tagged ples were generally 0.5–1.0 mM in concentration, dis- protein of interest to be expressed as an N-terminal fusion solved in 20 mM phosphate buffer, 50 mM NaCl, at pH 1 product with the signal peptide of E. coli OmpA porin.28 ranking from 3.5 to 7.2. H chemical shifts were directly A thrombin cleavage site is added between the His-tag and referenced to the resonance of 2,2-dimethyl-2-silapen- 13 15 the sequence of the protein to get rid of the tag after the tane-5-sulfonate sodium salt, whereas C and N chem- purification. ical shifts were indirectly referenced with the absolute fre- quency ratios X(13C/1H) 5 0.251449530 and X(15N/1H) 5 0.101329118. NMR spectra were processed with Gifa Expression and purification of recombinant 29 OmpATb proteins (version 4.4) software utility. Sequential backbone resonance assignment of OmpATb E. coli BL21(DE3)omp8 were transformed with was obtained by combined uses of 15N-edited three- pETSIG-ompATb73–326, pETSIG-ompATb73–204, pETSIG- dimensional NOESY/TOCSY and heteronuclear correla- 30 ompATb1–204, and pETSIG-ompATb198–326. The recombi- tion experiments HNCA, CBCA(CO)NH, and HNCO. nant strains were selected on LB supplemented with 100 The side-chain assignments were achieved by 1H two- 2 2 lgmL 1 ampicillin and 50 lgmL 1 kanamycin. For the dimensional NOESY and 13C-edited three-dimensional 1 15 fermentation, 2 L of minimal medium containing NH4Cl NOESY/TOCSY using protein samples in D2O. H, N

PROTEINS 647 Y. Yang et al.

Table I Statistics for the Calculations of the Solution Structures of the N- and C-Terminal Domains of OmpATb

NMR restraints OmpATb73–204 OmpATb198–326 Number of restraints per residue 12.16 11.09 Distance restraints 1605 1431 Intraresidue 357 296 Short range 547 445 Medium range* 284 402 Long range 417 288 (H-bonds) (30) (45) Torsion angle restraints F 96 90 C 00 v1 017 Structure statistics (10 structures) Maximum NOE upper violation 0.28 0.21 Average deviation from distance restraints () 0.081 0.057 Maximum dihedral angle violation 34.66 32.5 Average deviation from dihedral restraints (8) 12.13 9.39 Average Amber energy (kcal mol21) 23797.09 24349.75 Ramachandran plot Residues in most favored regions 84.3% 86.9% Residues in additional allowed regions 14.6% 12.6% Residues in generously allowed regions 0% 0.2% Residues in disallowed regions 1.1% 0.3% CING ROG analysis (all residues) (132) (129) Red [number of residues (proportion)] 31 (23%) 24 (19%) Orange [number of residues (proportion)] 30 (23%) 26 (20%) Green [number of residues (proportion)] 71 (54%) 79 (61%) Overall NOE completeness (Wattos analysis) 40.6% 36.6% What IF analysis Structure z-scores, positive is better than average First-generation packing quality 21.096 0.065 21.232 0.134 Second-generation packing quality 21.521 0.169 0.162 0.215 Ramachandran plot appearance 22.339 0.321 21.241 0.238 chi-1/chi-2 rotamer normality 24.108 0.514 24.080 0.270 Backbone conformation 20.980 0.256 20.052 0.120 RMS z-scores, should be close to 1.0 Bond lengths 1.206 0.003 1.196 0.001 Bond angles 0.762 0.007 0.784 0.019 Omega angle restraints 1.049 0.019 1.070 0.036 Side chain planarity 0.805 0.109 0.921 0.053 Improper dihedral distribution 0.986 0.024 1.046 0.030 Inside/Outside distribution 0.977 0.012 1.000 0.010 Root-mean-square deviation For residues 78–195 For residues 208–26 Mean global backbone 0.46 0.06 0.71 0.1 Mean global heavy 1.08 0.08 1.18 0.12 Mean backbone for secondary structures residues 0.45 0.06 0.55 0.10

Root-mean-square deviations (RMSD) were calculated over mean coordinates by MOLMOL.36

HSQC experiments were used to observe protein signal bone dihedral angle restraints (F and C) were derived changes in each circumstance. from chemical shift analysis using program PREDI- 31 TOR. Side-chain dihedral angle restraints (v1) were based on NOE patterns and used for residues like Val, Structure calculations Tyr, and Leu (v1 52608 or 1808). The 1H–1H distance restraints were derived from Two hundred structures were calculated using the 32 aforementioned NOESY spectra, all with 100-ms mixing program CYANA-2.1 for OmpATb73–204 and time. NOE intensities were converted into interproton OmpATb198–326. Thirty structures with the lowest target distances: 2.4, 2.8, 3.6, 4.4, and 5.0 A˚ , corresponding to function values were further energy minimized with Sander very strong, strong, medium, weak, and very weak inten- (AMBER8)33 using the force-field ff99 and Generalized sities, respectively. Based on H/D exchange experiments, Born solvation model GBOBC-II.34 The set of distance and hydrogen-bond restraints were used for the residues in angle restraints exploited here was the same than the one the regular secondary structures (rNHO 5 1.8–2.2 A˚ used before. The 10 final structures were analyzed with in a-helix and rNHO 5 1.8–2.4 A˚ in b-sheet). Back- INSIGHT and checked up with AQUA suite35 as

648 PROTEINS Structure of OmpATb from M. tuberculosis with more recent CING (http://nmr.cmbi.ru.nl/cing/Credits. have regular surfaces at both sides, and (3) the conduct- html) and PSVS37 online suites. Complete CING statistics ance value measured for the pore inserted in planar lipid are available at http://nmr.cmbi.ru.nl/NRG-CING/data/kg/ bilayers (see Results section). In this third phase, starting 2kgs/2kgs.cing/2kgs/HTML/index.html and http://nmr.cmbi. from a monomer from the last and best run of HAD- ru.nl/NRG-CING/data/kg/2kgw/2kgw.cing/2kgw/HTML/ DOCK used to build the pentameric water-soluble com- index.html for the N- and C-terminal domains of plex, we ran SymmDock40 to modeled a complex with OmpATb, respectively. A compilation of the results given seventh-order symmetry (not yet implemented in the by the different check procedures (CING ROG, WHAT online HADDOCK version), keeping the orientation of IF, and Wattos) is presented in Table I. Research of N- the side chains at the interface as it was on the HAD- terminal domain fold analogs within the structural data- DOCK model. This allowed for less disfavorable energetic base was done with DALI38; the figures were generated terms for the heptameric form. As a final stage, the with PyMOL (Warren L. DeLano, ‘‘The PyMOL Molecu- model was minimized with gromacs to remove possible lar Graphics System,’’ DeLano Scientific LLC, San Carlos, van der Waals clashes coming from the rigid body dock- CA; http://www. pymol.org). ing described above. The interfaces were checked to be still consistent with the NMR-derived data and displayed eight potential hydrogen bonds and 10 potential salt Modeling of the water-soluble pentamer bridges across each interface according to PDBePISA At this stage, D@ppm observed when comparing the (http://pdbe.org/pisa/), as expected for electrostatic 1H,15N-HSQC spectra obtained from the species corre- attenuation. sponding to the major peak and one of the minor peaks of the gel filtration step [Fig. 5(A,B)] were interpreted as possible contact consequences between two protomers Planar lipid bilayer recordings and hence finally used as ambiguous restraints into the HADDOCK webserver.39 The modeling task was driven Virtually solvent-free planar lipid bilayers were formed as a two-phase process (preceding a third phase yielding over a 125–200 lm hole in a polytetrafluoroethylene film the heptameric complex). Consistent with gel permeation (10-lm thick) pretreated with a mixture of 1:40 (v/v) chromatography, DLS measurements, and NMR data that hexadecane/hexane and sandwiched between two half suggest the existence of pentameric OmpATb symmetrical glass cells. Phosphatidylcholine from soy beans (azolectin species (see Results section) in solution, a model for an from Sigma type IV S), dissolved in hexane (0.5%), was intermediary water-soluble state—probably preceding a spread on the top of the electrolyte solution (1M KCl/10 functional lipid-soluble state—was built using rigid pre- mM HEPES, pH 6.0) in both compartments of the meas- diction and seventh-order symmetry with SymmDock uring cell. Bilayer formation was achieved by lowering (fully random search phase or Phase 1). This first model and raising the level up in one or both compartments allows us to sort out ‘‘active’’ residues on each monomer and monitoring capacity responses. Voltage was applied interfaces, lowering the ambiguity of the restraints fur- through Ag/AgCl electrodes in the two compartments of ther used in HADDOCK computations (see Supporting the measuring cell. The applied voltage was the potential Information Table IS). These ambiguous restraints were of the cis-side compartment and the trans-side was then used in a flexible HADDOCK modeling, through grounded. The purified recombinant OmpATb proteins the online version kindly updated on demand to allow were added to the cis side (5–100 ng/mL). Single-channel computation with fifth-order symmetry (semirandom currents were recorded with a BLM 120 amplifier (Bio- search or Phase 2). We used the C5 symmetry mode Logic, France) and stored via a DRA 200 interface (Bio- offered by the software in the so-called ‘‘guru mode.’’ Logic, Claix, France) on a CD recorder for off-line analy- Otherwise, the other parameters were left to their default sis. CD data were then analyzed by the WinEDR and values. The first result has been chosen for its best score Biotools softwares (Bio-Logic, France). In macroscopic (Haddock score 2183.7 9.2) and its quite low RMSD conductance measurements, the doped membranes were from the overall lowest energy structure (1.7 1.3 A˚ ). subjected to slow ramps of potential (10 mV/s), and transmembrane currents were fed into an amplifier (BBA-01, Eastern Scientific, Rockville, MD). Current– Building a putative membrane-soluble heptamer with Symmdock voltage curves were stored on a computer and analyzed with Scope software (Bio-Logic, France). All measure- A priori, the criteria to select a ‘‘plausible’’ (i.e., mem- ments were performed at room temperature. brane-soluble) heptamer were mostly: (1) the quenching The coordinates have been deposited in the PDB: of the charges (already observed-although estimated 2KGS for the OmpATb73–204 and 2KGW for the insufficient for membrane immersion with the HAD- OmpATb198–326. The coordinates for the pentameric and DOCK pentamer) incompatible with an apolar medium heptameric complexes are available upon request to the insertion, (2) the alignment on a plan of the modules to authors.

PROTEINS 649 Y. Yang et al.

RESULTS AND DISCUSSION that OmpATb73–326 is a two-domain protein with a central proline-rich hinge. OmpATb73–326 is a two-domain protein The full-length OmpATb protein of M. tuberculosis was The C-terminal domain of OmpATb produced as a recombinant protein in outer membranes 73–326 22 shares a 3D fold with proteins of the of E. coli as done previously. However, after unsuccess- OmpA-like family ful attempts to purify sufficient amounts of the entire OmpATb to carry out structural studies, we focused our Initially, OmpATb was identified in the genome of M. tuberculosis based on its significant sequence homology analysis on the truncated OmpATb73–326, which is produced in large quantities as a soluble protein in the bac- with the C-terminal region of the OmpA protein family 19 terial cytosol. Previous studies showed that OmpA members. Therefore, it is tempting to propose that OmpATb and OmpA-like proteins share a common Tb73–326 was able to form ion channels in lipid bilayers similarly to the full-length OmpATb,22 confirming that structural fold in their C-terminal elements. 1 15 this truncated form corresponds to a functional protein. In the H, N-HSQC spectra (Fig. 1), the segment I223- The NMR backbone resonance sequential attribution L240 located in the C-terminal domain of OmpATb73–326 8 was carried out on 15N- and 15N,13C-uniformly labeled exhibited incomplete resonances even at 25 C and pH 6.0, precluding the calculation of a high-resolution structure. OmpATb73–326 samples, using standard double- or triple- resonance experiments, respectively. At pH 7.2 and 358C, Suspecting rapid amide proton exchange with water, we 210 of 254 residues were assigned, the misassigned regions carried out additional NMR experiments on a sample of 15 8 corresponded mainly to the five peptide segments G88- OmpATb198–326 N-labeled C-terminal domain at 25 C 1 15 N89, S168-E170, G196-D210, I223-L240, and S266-G268, and pH 3.5. Comparing the H, N-HSQC spectra of which showed weak signals or no signal at all, probably OmpATb198–326 at pH 6.0 [Supporting Information Fig. because of fast amide proton exchange with water or con- 1S(B)] and at pH 3.5 (Supporting Information Fig. 2S), formational changes on the NMR time scale. To assign no major changes were observed, excepting the segment these remaining residues, a new series of NMR measure- I223-L240, which was stabilized at acidic pH as revealed by ments was carried at pH 6.0 and at 258C. A 1H,15N HSQC the appearance of additional NMR resonances correspond- spectrum with numbered residues is depicted in Figure 1, ing to several backbone and side-chain atoms. Combining revealing that most 1H/15N cross peaks were recovered information obtained at these two pH, a virtually complete under these conditions, excepting the segment I223-L240. resonance assignment has been obtained for the C-termi- 13C-edited 3D-TOCSY were used to identify side-chain nal domain of OmpATb73–326. protons for each residue, whereas 15N- and 13C-edited The three-dimensional structure of OmpATb198–326 was 3D-NOESY allowed us to establish proton–proton dis- then determined using 1538 experimental restraints tance constraints. The chemical shift table was deposited derived from NMR spectroscopy (Table I). An ensemble of in the bmrb databank (accession number 16237). the 10 lowest energy NMR structures [Fig. 2(A,B)] reveals b a Our detailed NOE analysis confirmed that OmpA that OmpATb198–326 belongs to the / structural class. b Tb is composed of two domains (G73-T195 and This domain is composed of a mixed -sheet formed from 73–326 b b b C208-N326 fragments) presenting both b-strands and a- parallel and antiparallel -strands ( 1to 4), which are helices. The G73-T195 and C208-N326 protein fragments flanked by four helices (a1toa4) [Fig. 3(A)]. The four b- did not show any close contact, suggesting that these seg- strands combined with helices a2 and a3 constitute a ments are both mobile and independent. Moreover, the bababb motif, which was described previously in the proline-rich central linker (G196-P207) connecting the crystal form of the OmpA-like domain of the RmpM pro- two domains seems to be relatively flexible as reflected by tein from N. meningitidis,26 confirming for the first time the narrow and strong peaks of its residues G196, Q197, that the C-terminal region of OmpATb has the same fold A198, G201, and A204 in the 1H,15N-HSQC spectrum than other OmpA-like domains. Both structural modules (Fig. 1). To demonstrate that the two domains of can be superimposed satisfactorily [Fig. 3(B)]. OmpATb73–326 are independent, both fragments Nevertheless, comparison of the folds of the OmpATb (OmpATb73–204 and OmpATb198–326) were produced sep- C-terminal domain and RmpM revealed slight structural arately as recombinant proteins in E. coli and purified differences. The two proteins differ regarding the N-ter- a under the same conditions as OmpATb73–326. The minal proximal residues, which are organized as an -he- 1 15 H, N-HSQC spectra obtained from the gel filtration lix (a1) in OmpATb198–326 and a b-strand (b1) in chromatography major peak of OmpATb73–204 and RmpM. Moreover, OmpATb198–326 did not exhibit an 1 15 OmpATb198–326 were nearly identical to the H, N- elongated molecular shape as does RmpM, because of a b b HSQC of OmpATb73–326—in spectral superimpositions— shorter loop connecting 3 and 4 strands. According to confirming the lack of interaction between the two a previous sequence alignment performed with several domains (Supporting Information Figs. 1S and 2S). Alto- proteins containing a C-terminal OmpA-like domain, gether these data confirm the results by Teriete et al.25 this extension loop appears to vary in length.26

650 PROTEINS Structure of OmpATb from M. tuberculosis

Figure 1 1 15 15 OmpATb73–326 NMR fingerprint. [ H, N] HSQC spectrum of OmpATb73–326 recorded at 600 MHz, 258C, and pH 6.0 on a 1.0 mM N-uniformly labeled sample. Cross-peak assignments are indicated using the one-letter amino acid code and number. The side-chain amine groups are differentiated by the mention NH2 following their position number. The green squares are meant to highlight the narrow and strong resonances observed for residues of proline-rich central linker (G196-P207). Nine amide resonances (F225, G229, L232, I233, A235, D236, Y237, E238, and L240) are absent because of conformational exchanges. These missing cross peaks were recovered at pH 3.5 as shown in Supporting Information Figure 2S.

The N-terminal domain of OmpATb73–326 OmpATb protein, indicating that alone, the N-terminal reveals an unusual folding among the module of this protein can permit ion diffusion. In this porin-like proteins family study, before investigating the structure of the It has been shown previously that the N-terminal frag- OmpATb73–204, we verified that this shorter segment ment OmpATb73–220 could form channels in planar lipid could also form channels in lipid bilayers. When water- bilayers similar to the channels observed for the entire soluble OmpATb73–204 was reconstituted into planar lipid

PROTEINS 651 Y. Yang et al.

Figure 2

Solution structures of the N- and C-terminal domains of OmpATb. Upper panel: OmpATb198–326, lower panel: OmpATb73–204.(A) and (C) overlay of 10 NMR deposited structures with lowest energy from AMBER41 calculation, (B) and (D) a cartoon representation of the energy-minimized structure.

652 PROTEINS Structure of OmpATb from M. tuberculosis

Figure 3

Comparison of the OmpATb198–326 with RmpM of Nesseiria meningitidis.(A) Topology of OmpATb198–326,(B) superposition with RmpM: ˚ OmpATb198–326 is colored in green, RmpM in pink. The effective superimposition with a value of 1.514 A over 92 aligned residues has been accomplished with the ‘‘YASARA view’’42 version of Mustang.43 With this method, these two domains are evaluated to share a 31.52% sequence identity. bilayers no ion channel activity was observed. However, to FATCAT.44 A superimposition of the two NMR struc- addition of detergent, such as octyl-polyoxyethylene ture is given as Supporting Information (Supporting In- (final concentration 0.5%), to the protein sample resulted formation Fig. 3S). The OmpATb73–204 fold belongs to in the formation of ion channels (Fig. 4). As expected, the a1b sandwich structural class. It is composed of a the electrical properties of this construct were similar to mixed b-sheet formed from parallel and antiparallel b- those observed previously for the fragment 73–220,22 strands (b1tob6), with three helices (a1toa3) packed with a single-channel conductance value around 1000 pS on the same side of the b-sheet [Fig. 5(A,B)]. The topol- and a preferred orientation of the OmpATb73–204 mole- ogy presents a twofold symmetry (1808 rotation) and can cules in the lipid bilayer as documented by the asymmet- be divided in two b-sheet subdomains B1 (b1tob3) ric current–voltage curves we obtained. However, the and B2 (b4tob6) that superimpose satisfyingly [Fig. absolute orientation of OmpATb73–204 could not be 5(C)]. B1 and B2 are both associated to one a-helix, a1 determined from these current–voltage recordings. and a3, respectively, which links the second and third In our NMR experiments, unlike the C-terminal do- strands of each subdomain. The subdomain B2 is related main, which as discussed above required to work at low to the BON (Bacterial OsmY and Nodulation) domain pH to stabilize a well-defined conformation, the N-termi- family [Fig. 5(B)], previously described as a putative nal domain OmpATb73–204 yielded well-resolved and membrane-binding domain found in a family of osmotic complete NMR spectra at pH 6.0 [Fig. 1 and Supporting shock protection proteins and a family of hemolysins. It Information Fig. 1S(A)]. The three-dimensional structure was also reported to be associated to membrane pore- of OmpATb73–204 was determined using 1701 experimen- forming domains of secretins and mechanosensitive 45 tal restraints: it is similar to that published by Teriete channels. In the OmpATb73–204 structure, the BON do- et al.25 during the processing of this manuscript, with P main is clearly associated with what we previously desig- 2 value of 3.89 3 10 12 and an RMSD of 3.07 A˚ measured nated as subdomain B1, which shares nearly the same between the two structures and without twists according topology, minus the missing first N-terminal helix of the

PROTEINS 653 Y. Yang et al.

Figure 4

Ion channel activity of OmpATb73–204. (A) A recording of current–voltage (I/V) curve obtained after addition of OmpATb73–204 at a final concentration of around 100 ng/mL to the cis-side of azolectin bilayers submitted to slow voltage ramps (10 mV/s). The arrows indicate the direction of the applied voltage ramp. The electrolyte was 1M KCl, 10 mM HEPES, pH 6.0. (B) Current fluctuations of a single channel of OmpATb73–204 at an applied voltage of 2160 mV giving rise to conductance values of about 1000 pS. The electrolyte is the same than in (A). BL represents the baseline. All the measurements were done at room temperature.

canonical sequence. Consequently, OmpATb73–204 can be MspA water-soluble monomer (184 amino acid) from M. described as a two BON domain–containing module and smegmatis, which assemble in an octameric complex in not as a classical porin whose architecture corresponds to the mycobacterial membrane.14,48 Recently, another oli- a hollow b-barrel.47 gomeric channel structure has been proposed for the These data, as well as the results recently obtained by 10.6-kDa soluble form of Corynebacterium glutamicum Teriete et al.,25 raised the question of how OmpA PorB protein belonging, like M. tuberculosis, to the group Tb73–204 could form a functional channel in lipid bilayers. of actinomycetes. In this case, the pore would be formed 49 Obviously, monomeric OmpATb73–204, even with a com- by a pentameric a-helical arrangement. When search- plete modification of its 3D structure due to the hydro- ing for structural homologs in the PDB, DALI returned phobic environment (detergent and/or lipid bilayers), more than 500 solutions presenting both a low z-score cannot form large ion channels such as those observed in lying between 4.1 and 2.0 (in the DALI evaluation, a z- our planar lipid bilayers. In a previous study, Teriete score lower than 2 is considered insignificant) and a low et al. suggested that the physiological function of sequence identity lying under 22%. A rapid inspection OmpATb is not related to a porin activity and proposed revealed that most of them displayed a partial and mod- alternative models for its organization in the bacterial en- est structural similarity, but were referring repeated times velope. On the basis of the evidences presented below, we to identical molecules due to the redundancy in the postulate that OmpATb73–204 must oligomerize to form a PDB. Nevertheless, one candidate, EscJ, proposed by functional channel-forming structure as observed for the DALI’s ranking retained our attention, not based on the

654 PROTEINS Structure of OmpATb from M. tuberculosis

statistics, but rather for its role in the enteropathogenic E. coli bacteria cell membrane. A few years ago, the crystal structure together with concomitant biochemistry approaches of this protein revealed that EscJ was capable of associating into large assemblies of 24 subunits forming a 170-A˚ diameter ring displayed on the periplasmic face of the inner membrane and anchored by an N-terminal lipidation and a C-terminal transmembrane helix.50 Additional structures of homologs have appeared recently confirming the importance of this type of architecture in the formation of such rings.41,51 Therefore, it is tempting to propose, based on structural similarities with EscJ and on the apparent necessity of the OmpATb73–204 two BON domain–containing modules to oligomerize to form a channel, that OmpATb73–204 could assemble into a macromolecular complex of closed rings inside the thick mycobacterial cell wall.

Evidences for oligomerization of

OmpATb73–204

To investigate OmpATb73–204 oligomers, we turned to size exclusion chromatography. Indeed, contrary to other protein constructs in this study, the OmpATb73–204 chromatogram profile presented, in addition to the major monomer peak, some minor peaks corresponding to elution of oligomers [Fig. 6(B)]. The 1H,15N-HSQC spectrum of the smallest oligomer peak (corresponding to a 2.3 mer according to the calibration) gave rise to similar chemical shifts as those observed for the monomeric form, indicating that the global folding remained unchanged. Straightforward signal assignments of the oligomer were achieved via comparison with the known assignments of the monomer. However, the resonances we measured were broadened, confirming that oligomerization of OmpATb73–204 molecules occurred in this fraction [Fig. 6(A)]. A more detailed analysis of the 1H,15N-HSQC spectra revealed that 20% of the residues presented significant amide proton chemical shift differences (>0.05 ppm for 1Hor15N). For all these residues, Figure 5 beside the major broadened peaks corresponding to the The N-terminal domain is an a1b sandwich with a mainly antiparallel oligomeric form of OmpATb , minor peaks corre- b-sheet. (A) OmpATb73–204 reveals two subdomains one being a 73–204 topological image of the other. A twofold symmetry (1808 rotation) is sponding to the monomeric form are detectable on the observed and is virtually splitting this module into two subdomains HSQC spectrum, suggesting that the oligomeric and the denoted by a two color code (green for the first subdomain and blue monomeric forms are in slow equilibrium. Most of the for the second). The axis of rotation is passing perpendicularly through the center of the b-sheet. The second subdomain has been identified as concerned residues are located at the N-terminal pole of 45 a member of the BON domain fold class. (B) The topological OmpATb73–204, indicating that this portion of the mole- diagram shows that the two subdomains are joined together by strand 1 cule represents the main interface involved in the oligo- and strand 4 in an antiparallel fashion and results in a six-strand sheet merization process [Fig. 6(C,D)]. Indeed, the extreme N- formation (same color code). The BON domain topology is delineated by a black dotted square. (C) In this stereoview, the two subdomains terminal residues A74, S75, and A76 did not show any were satisfyingly superimposed with an 1.8-A˚ RMSD on C-alpha atoms resonance in the monomeric form at pH 7.2 because of 46 using Dalilite (a z-score of 3.9 has been returned: 45 residues were fast solvent exchange, whereas they exhibited intense effectively aligned with an apparent identity of 22%). The segment that appears twice is colored in light blue with respect to the first peaks in the oligomeric form. This result suggests that subdomain. Those data are in favor of a possible duplication to these three residues are no longer exposed to solvent, but originate the whole domain explaining the twofold symmetry found. rather participate in oligomer assembly, further sup-

PROTEINS 655 Y. Yang et al.

Figure 6 1 15 Evidences of oligomerization of OmpATb73–204.(A) Superposition plot of the H- N HSQC spectra of the monomeric form (magenta) and oligomeric form (black) of OmpATb73–204. Cross-peak assignments are indicated using the one-letter amino acid code and number. The side-chain amine groups are differentiated by the mention NH2 following their position number. The star designates the NH of the indole group of tryptophan 183 and the NeH of the guanidine group of arginines. The juxtaposition of two residue labels, one of them in parenthesis indicates that both resonances are affected by the oligomerization process. The green squares highlight the shifted—or recovered—resonances as a consequence of oligomerization. (B) Profile of the steric exclusion column chromatogram carried out on a pure OmpATb73–204 fragment sample. Based on a calibration, it results that OmpATb73–204 can oligomerize and exist not only as a monomer (major peak highlighted in magenta) but also as a dimer, pentamer, or heptamer (minors peaks highlighted in black). (C) Plot of the amide chemical shift differences between the monomeric and oligomeric forms of OmpATb73–204. In the insert, all residues involved in the oligomerization interface are shown in green. (D) Electrostatic surface 52 53 potential of OmpATb73–204 generated using PDB2PQR and APBS with parameters at ‘‘standard’’ (ionic strength 155 mM, pH 7). Electrostatic potential is represented as a tricolor gradient going from blue (11.8 kT) over white (neutral) to red (21.8 kT). Interestingly, the two involved surfaces appear complementary in surface and charge.

ported by relatively large chemical shift changes observed monomers. It should be noted that the NeH of the gua- for their neighboring residues (segment L77-S83). nidine group of arginine is usually not detectable at pH Further analysis of the NMR spectrum of the oligo- 7.2, except if it is involved in a salt bridge or buried meric form also revealed that several other residues dis- inside the protein structure. Of the four arginines R86, tributed on either side of the a1b sandwich monomer R131, R157, and R177, only R86 presented a side-chain presented a shift or recovered their amide proton signal (Fig. 6) in the monomeric form, indicative of a resonances. The residues involved are D96, Q123, H125, buried position between an a-helix and a b-strand. Inter- D127, and R157 on the b-sheet side of the monomer estingly, R157 and R131 showed not only backbone am- and D99, R131, D185, and K187 on the a-helical side. ide chemical shift changes but also exhibited side-chain These are mainly charged residues that could plausibly be resonances in the oligomeric form. Hence, these residues involved in salt bridges or hydrogen bonds between two must be involved in electrostatic interactions with nega-

656 PROTEINS Structure of OmpATb from M. tuberculosis

Figure 7

Pentameric architecture of OmpATb73–204. The docking resulted from an HADDOCK approach using amide chemical shift differences between the monomeric and oligomeric forms of OmpATb73–204 as ambiguous constraints and a symmetry of order 5. (A) Cartoon representations of the pentameric assemblage. (B) Close-up on the residues bridging two monomers indicating that most of them are polar or charged. tively charged residues or be buried during molecular as- umn. However, the residues distribution, concerned by a sembly process. Analysis of the electrostatic surface of the shift or the apparition of their amide proton resonances, monomer using APBS53 revealed a large acidic region on on the two opposite sides of the a1b sandwich sug- the a-helical side of the monomer and a basic patch on gested to us that each monomer must have two neigh- the b-sheet side [Fig. 6(D)]. This distribution of charged bors. Moreover, with only one set of resonances observed residues would favor interactions between monomers, the eventuality of an asymmetrical object must be with the a-helical side of one monomer binding to the rejected; indeed, the oligomer must be formed in a sym- b-sheet of the second via salt bridges or hydrogen bonds metrical fashion. The broadened NMR lines observed in as mentioned above. the oligomer spectra more likely reflect population of at The oligomer peak analyzed by NMR was predicted to least as a trimer of OmpATb73–204 (396 residues) but be a dimer from the calibration of the size exclusion col- could arise from a tetramer (528 residues) or even a pen-

PROTEINS 657 Y. Yang et al. tamer (about 600 structured residues). These obser- vations forced us to consider that the dimeric OmpA Tb73–204 we collected from the chromatography step might have assembled into larger oligomers during the concentration procedure before NMR measurements. This was further supported by DLS experiments (Sup- porting Information Fig. 4S) recorded on the NMR sample, yielding a gyration radius of 32 A˚ , a value which is not compatible with a dimeric assembly of the molecule. Preliminary modeling on a trimer or a tetramer was carried out with Symmdock40 using the 32 marked residues noted in the Figure 6(A) as input. At this stage, analysis of the obtained docking structures revealed that OmpATb73–204 might pile up with a ring-like organization similarly to EscJ protein family with all the N-terminal poles in proximity in the middle of the macromolecular complex and with the a-helical side of one monomer interacting with the b-sheet side of the neighboring molecule. A close inspection of the docking results suggested that the trimer or tetramer was not compact enough to be compatible with our NMR data. The extensive areas delineated by the NMR footprint [Fig. 6(A,C)] are consistent with multiple contacts between the subunits, as expected for a pentameric form with at least half of the b-sheet surface involved in the interface. Taking into account the extent of the NMR cross peak broaden- ing and the molecular size of the oligomeric state that must remain detectable, we focused on a one step higher order oligomerization docking calculation, that is, a pentameric form, using the new version of HADDOCK web- server54 with the ambiguous constraints derived from the D@ppm (Supporting Information Table IS) and imposing a C5 symmetry. As displayed in Figure 7, the resulting pentamer exhibited a more compact complex, consistent with the NMR spectral changes. In addition, the gyration radius calculated on this pentameric model with YASARA (27.9 A˚ ) is in good agreement with the one estimated from DLS experiments. In this model, residues of the a helical side of one monomer can easily interact with part- ners from the b-sheet side of neighboring molecule via a hydrogen bond network. The pairs of residues, which could be involved in these interactions, are D99/H125, R131/D96, D134/K154, and D185/R157 (in agreement with the chemical shift changes). It is noteworthy that residues D96, D99, and D127 showed unusual chemical shift perturbations (>0.20 ppm for 1Hor15N). For clarity, the resulting H-bond network combination out of the Haddock modeling is illustrated in Figure 7. Figure 8 Although the pentameric OmpATb73–204 displayed a central aperture that could act as a diffusion pore, this Electrostatic surface potential of a heptameric form of OmpATb73–204. The electrostatic surface charge of the heptamer (result from Simmdock complex only represents a water-soluble oligomer corre- calculation; cf. ‘‘Modeling section’’ in ‘‘Materials and Methods’’) was sponding to the main population present in the NMR calculated using PDB2PQR52 and APBS,55 with parameters in standard sample but does not appear compatible with insertion conditions (ionic strength 155 mM, pH 7). Electrostatic potential is 1 into a lipid bilayer to form a membrane channel. Indeed, represented as a tricolor gradient going from blue ( 8 kT) over white (neutral) to red (28 kT). a more pronounced neutralization of the surface charges between adjacent monomers on the external side of the

658 PROTEINS Structure of OmpATb from M. tuberculosis ring would be necessary to create a hydrophobic complex on the high sequence identity of its carboxy-terminal capable of interacting with the membrane. As we part with the carboxy-terminal domain of OmpA from observed oligomeric forms of OmpATb73–204 up to hep- E. coli. Similarly to its E. coli homolog, this membrane tamers (see chromatography step), we modeled a more protein was expected to be located, at least for half of it, compact heptameric complex, with a larger protein–pro- inside the outer membrane of virulent strains. We report tein interface (Fig. 8). In such an assembly, many of the here the complete structural study of OmpATb73–326. charges are neutralized, resulting in an extended, mostly The N-terminal domain OmpATb73–204, which defi- nonpolar perimeter surface all around the ring. This type nitely does not fold as a b-barrel in accordance with a 25 of assembly should allow OmpATb73–204 to insert into a recent study, is capable to interact intimately with the biological membrane of 30 A˚ thickness, extracting its membrane and form channels. It could insert in the perimeter from the solvent and leaving the two charged membrane as a heptameric ring formed by a close inter- surfaces left exposed on both sides of the bilayer. The action of the a1b structured monomer, allowing ion width of this heptameric complex is almost two times diffusion as observed in planar lipid bilayers. OmpATb larger than dimensions observed for classical porins from has been reported to be associated with acid resistance Gram-negative bacteria, but is in the same range as the and virulence of TB. Although the assumption that the size of the octameric MspA from M. smegmatis.14 A 10- OmpATb channel activity involved in this acid response A˚ pore diameter can be measured on this model, consist- cannot be fully confirmed from our model, the pH mod- ent with conductance values previously measured: an av- ulation of the OmpATb pore-forming properties erage conductance of 1600 pS has been measured for observed previously22 is explained by partial or full neu- OmpATb extracted from mycobacterial cell wall,24 tralization (depending on how low the pH gets in situ) roughly corresponding to a 7.6-A˚ diameter in the case of of negatively charged residues sitting at entrances and/or an ideal cylinder, and Senaratne et al.20 published a pore inside our modeled channel inducing electrostatic poten- diameter of 14–18 A˚ . The channel lumen is mostly tial changes and modification of the gating properties. hydrophilic like known porin-like proteins and is slightly The role of the C-terminal domain OmpATb198–326 acidic (in particular because of D127), consistent with remains unclear. Previously, the C-terminal OmpA 22 the cation selectivity measured for the native OmpATb. Tb212–326 was shown not to be necessary for the pore Other acidic residues are located at the external surface function, suggesting another role for this domain.22 of the pore (D99, E100, and D122; Supporting Informa- Recent molecular dynamic simulations support the inter- tion Fig. 5S) and could also contribute to increase the action of a C-terminal domain of an OmpA-like protein, concentration of cations at the channel mouth. Moreover, the PmOmpA protein of Pasteurella multocida, with a polar residues as S75 and S78, located in the lumen near lipid bilayer via contacts between charged/polar residues the channel entrance, might also favor fluxes of positively and lipid head groups.58 Therefore, we hypothesize that charged ions. In the MspA porin, which is also cation the presence of charged residues in the C-terminal mod- selective, the eyelet of the channel is constituted by two ule of OmpATb could allow such contacts with lipids rings of aspartate residues and a nonpolar ring consisting present in the mycobacterial cell wall. Alternatively, the of leucines and isoleucines located on the external side of OmpATb C-terminal domain could interact with the ara- the eyelet.14 In our modeled pore, a hydrophobic bottle- binogalactan-peptidoglycan layer of the mycobacterial en- neck (V129) is also located above the D127 ring near the velope. Indeed, it has been proposed for the OmpA-like outside entry. The neighboring charged residues engaged domain of RmpM that 10 highly conserved residues in salt bridges in this area are similar to the one present within the family, located in a hydrophilic groove, might in the E. coli OmpA channel for which the central E52- be involved in a putative peptidoglycan-binding site.26 R138 salt bridge was suggested to form alternate ion The presence of these conserved residues in the sequence pairs with K82 and E128 to modulate structural transi- of OmpATb198–326 suggests that, as for the OmpA-like tion between opened and closed states.56,57 For this rea- domain of RmpM, the OmpATb protein could interact, son, even if the measured diameter at the apex of the via its C-terminal domain, with the pseudo-peptidogly- two methyl group of V129 is to be around 10 A˚ ,itis can layer of the mycobacterial envelope. conceivable to suppose some movement as part of the gating mechanism, which could both enlarge or reduce ACKNOWLEDGMENTS the aperture size of OmpATb. The authors thank Sylvie Campagna and Laurent Kremer for technical assistance in preparation of mycobacterial cell wall. CONCLUDING REMARKS REFERENCES OmpATb was earlier characterized as a porin family member based on its putative role in facilitating several 1. Wells CD, Cegielski JP, Nelson LJ, Laserson KF, Holtz TH, Finlay A, water-soluble substances across the mycolic acid wall and Castro KG, Weyer K. HIV infection and multidrug-resistant tuber-

PROTEINS 659 Y. Yang et al.

culosis: the perfect storm. J Infect Dis 2007;196 (Suppl 1):S86– 23. Sugawara E, Steiert M, Rouhani S, Nikaido H. Secondary structure S107. of the outer membrane proteins OmpA of Escherichia coli and 2. Frieden TR, Sterling TR, Munsiff SS, Watt CJ, Dye C. Tuberculosis. OprF of Pseudomonas aeruginosa. J Bacteriol 1996;178:6067–6069. Lancet 2003;362:887–899. 24. Alahari A, Saint N, Campagna S, Molle V, Molle G, Kremer L. The 3. Jarlier V, Nikaido H. Mycobacterial cell wall: structure and role in N-terminal domain of OmpATb is required for membrane translo- natural resistance to antibiotics. FEMS Microbiol Lett 1994;123:11– cation and pore-forming activity in mycobacteria. J Bacteriol 18. 2007;189:6351–6358. 4. Brennan PJ, Nikaido H. The envelope of mycobacteria. Annu Rev 25. Teriete P, Yao Y, Kolodzik A, Yu J, Song H, Niederweis M, Marassi Biochem 1995;64:29–63. FM. Mycobacterium tuberculosis Rv0899 adopts a mixed alpha/beta- 5. Chambers HF, Moreau D, Yajko D, Miick C, Wagner C, Hackbarth structure and does not form a transmembrane beta-barrel. Bio- C, Kocagoz S, Rosenberg E, Hadley WK, Nikaido H. Can penicillins chemistry 2010;49:2768–2777. and other beta-lactam antibiotics be used to treat tuberculosis? 26. Grizot S, Buchanan SK. Structure of the OmpA-like domain of Antimicrob Agents Chemother 1995;39:2620–2624. RmpM from Neisseria meningitidis. Mol Microbiol 2004;51:1027– 6. Daffe M, Draper P. The envelope layers of mycobacteria with refer- 1037. ence to their pathogenicity. Adv Microb Physiol 1998;39:131–203. 27. Prilipov A, Phale PS, Van Gelder P, Rosenbusch JP, Koebnik R. 7. Nikaido H. Molecular basis of bacterial outer membrane permeabil- Coupling site-directed mutagenesis with high-level expression: large ity revisited. Microbiol Mol Biol Rev 2003;67:593–656. scale production of mutant porins from E. coli. FEMS Microbiol 8. Trias J, Jarlier V, Benz R. Porins in the cell wall of mycobacteria. Lett 1998;163:65–72. Science 1992;258:1479–1481. 28. Siroy A, Molle V, Lemaitre-Guillier C, Vallenet D, Pestel-Caron M, 9. Riess FG, Dorner U, Schiffler B, Benz R. Study of the properties of Cozzone AJ, Jouenne T, De E. Channel formation by CarO, the a channel-forming protein of the cell wall of the gram-positive bac- carbapenem resistance-associated outer membrane protein of terium Mycobacterium phlei. J Membr Biol 2001;182:147–157. Acinetobacter baumannii. Antimicrob Agents Chemother 2005;49: 10. Trias J, Benz R. Permeability of the cell wall of Mycobacterium smeg- 4876–4883. matis. Mol Microbiol 1994;14:283–290. 29. Pons JL, Malliavin TE, Delsuc MA. Gifa V.4: a complete package 11. Lichtinger T, Heym B, Maier E, Eichner H, Cole ST, Benz R. Evi- for NMR data set processing. J Biomol NMR 1996;8:445–452. dence for a small anion-selective channel in the cell wall of Myco- 30. Grzesiek S, Bax A. Improved 3D triple-resonance NMR techniques bacterium bovis BCG besides a wide cation-selective pore. FEBS Lett applied to a 31 kDa protein. J Magn Reson 1992;96:432–440. 1999;454:349–355. 31. Berjanskii MV, Neal S, Wishart DS. PREDITOR: a web server for 12. Kartmann B, Stenger S, Niederweis M, Stengler S. Porins in the cell predicting protein torsion angle restraints. Nucleic Acids Res wall of Mycobacterium tuberculosis. J Bacteriol 1999;181:6543–6546. 2006;34(Web Server issue):W63–W69. 13. Stahl C, Kubetzko S, Kaps I, Seeber S, Engelhardt H, Niederweis M. 32. Guntert P. Automated NMR structure calculation with CYANA. MspA provides the main hydrophilic pathway through the cell wall Methods Mol Biol 2004;278:353–378. of Mycobacterium smegmatis. Mol Microbiol 2001;40:451–464. 33. Case DA, Cheatham TE, III, Darden T, Gohlke H, Luo R, Merz 14. Faller M, Niederweis M, Schulz GE. The structure of a mycobacte- KM, Jr, Onufriev A, Simmerling C, Wang B, Woods RJ. The Amber rial outer-membrane channel. Science 2004;303:1189–1192. biomolecular simulation programs. J Comput Chem 2005;26:1668– 15. Cowan SW, Schirmer T, Rummel G, Steiert M, Ghosh R, Pauptit 1688. RA, Jansonius JN, Rosenbusch JP. Crystal structures explain func- 34. Onufriev A, Bashford D, Case DA. Exploring protein native states tional properties of two E. coli porins. Nature 1992;358:727–733. and large-scale conformational changes with a modified generalized 16. Schirmer T, Keller TA, Wang YF, Rosenbusch JP. Structural basis for born model. Proteins 2004;55:383–394. sugar translocation through maltoporin channels at 3.1 A resolu- 35. Laskowski RA, Rullmannn JA, MacArthur MW, Kaptein R, Thorn- tion. Science 1995;267:512–514. ton JM. AQUA and PROCHECK-NMR: programs for checking the 17. Basle A, Rummel G, Storici P, Rosenbusch JP, Schirmer T. Crystal quality of protein structures solved by NMR. J Biomol NMR structure of osmoporin OmpC from E. coli at 2.0 A. J Mol Biol 1996;8:477–486. 2006;362:933–942. 36. Koradi R, Billeter M, Wuthrich K. MOLMOL: a program for dis- 18. Niederweis M. Mycobacterial porins—new channel proteins in play and analysis of macromolecular structures. J Mol Graph 1996; unique outer membranes. Mol Microbiol 2003;49:1167–1177. 14:29–32,51–55. 19. Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, Harris D, 37. Bhattacharya A, Tejero R, Montelione GT. Evaluating protein struc- Gordon SV, Eiglmeier K, Gas S, Barry CE, III, Tekaia F, Badcock K, tures determined by structural genomics consortia. Proteins 2007; Basham D, Brown D, Chillingworth T, Connor R, Davies R, Devlin 66:778–795. K, Feltwell T, Gentles S, Hamlin N, Holroyd S, Hornsby T, Jagels K, 38. Holm L, Sander C. Dali: a network tool for protein structure com- Krogh A, McLean J, Moule S, Murphy L, Oliver K, Osborne J, parison. Trends Biochem Sci 1995;20:478–480. Quail MA, Rajandream MA, Rogers J, Rutter S, Seeger K, Skelton J, 39. de Vries SJ, van Dijk M, Bonvin AM. The HADDOCK web server Squares R, Squares S, Sulston JE, Taylor K, Whitehead S, Barrell for data-driven biomolecular docking. Nat Protoc 2010;5:883–897. BG. Deciphering the biology of Mycobacterium tuberculosis from the 40. Schneidman-Duhovny D, Inbar Y, Nussinov R, Wolfson HJ. Patch- complete genome sequence. Nature 1998;393:537–544. Dock and SymmDock: servers for rigid and symmetric docking. 20. Senaratne RH, Mobasheri H, Papavinasasundaram KG, Jenner P, Nucleic Acids Res 2005;33(Web Server issue):W363–W367. Lea EJ, Draper P. Expression of a gene for a porin-like protein of 41. Sani M, Allaoui A, Fusetti F, Oostergetel GT, Keegstra W, Boekema the OmpA family from Mycobacterium tuberculosis H37Rv. J Bacter- EJ. Structural organization of the needle complex of the type III iol 1998;180:3541–3547. secretion apparatus of Shigella flexneri. Micron 2007;38:291–301. 21. Raynaud C, Papavinasasundaram KG, Speight RA, Springer B, 42. Krieger E, Koraimann G, Vriend G. Increasing the precision of Sander P, Bottger EC, Colston MJ, Draper P. The functions of comparative models with YASARA NOVA—a self-parameterizing OmpATb, a pore-forming protein of Mycobacterium tuberculosis. force field. Proteins 2002;47:393–402. Mol Microbiol 2002;46:191–201. 43. Konagurthu AS, Whisstock JC, Stuckey PJ, Lesk AM. MUSTANG: a 22. Molle V, Saint N, Campagna S, Kremer L, Lea E, Draper P, Molle multiple structural alignment algorithm. Proteins 2006;64:559–574. G. pH-dependent pore-forming activity of OmpATb from Mycobac- 44. Ye Y, Godzik A. Flexible structure alignment by chaining aligned terium tuberculosis and characterization of the channel by peptidic fragment pairs allowing twists. Bioinformatics 2003;19 (Suppl 2): dissection. Mol Microbiol 2006;61:826–837. ii246–ii255.

660 PROTEINS Structure of OmpATb from M. tuberculosis

45. Yeats C, Bateman A. The BON domain: a putative membrane-bind- 52. Dolinsky TJ, Czodrowski P, Li H, Nielsen JE, Jensen JH, Klebe G, ing domain. Trends Biochem Sci 2003;28:352–355. Baker NA. PDB2PQR: expanding and upgrading automated prepa- 46. Holm L, Park J. DaliLite workbench for protein structure compari- ration of biomolecular structures for molecular simulations. Nucleic son. Bioinformatics 2000;16:566–567. Acids Res 2007;35(Web Server issue):W522–W525. 47. Schirmer T. General and specific porins from bacterial outer mem- 53. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA. Electro- branes. J Struct Biol 1998;121:101–109. statics of nanosystems: application to microtubules and the ribo- 48. Niederweis M, Ehrt S, Heinz C, Klocker U, Karosi S, Swiderek KM, some. Proc Natl Acad Sci USA 2001;98:10037–10041. Riley LW, Benz R. Cloning of the mspA gene encoding a porin 54. de Vries SJ, van Dijk AD, Bonvin AM. The HADDOCK web server from Mycobacterium smegmatis. Mol Microbiol 1999;33:933–945. for data-driven biomolecular docking. Nat Protoc 2010;5:883–897. 49. Ziegler K, Benz R, Schulz GE. A putative alpha-helical porin from 55. Baker NA. Poisson-Boltzmann methods for biomolecular electro- Corynebacterium glutamicum. J Mol Biol 2008;379:482–491. statics. Methods Enzymol 2004;383:94–118. 50. Yip CK, Kimbrough TG, Felise HB, Vuckovic M, Thomas NA, 56. Bond PJ, Faraldo-Gomez JD, Sansom MS. OmpA: a pore or not a Pfuetzner RA, Frey EA, Finlay BB, Miller SI, Strynadka NC. Struc- pore? Simulation and modeling studies. Biophys J 2002;83:763– tural characterization of the molecular platform for type III secre- 775. tion system assembly. Nature 2005;435:702–707. 57. Hong H, Szabo G, Tamm LK. Electrostatic couplings in OmpA ion- 51. Spreter T, Yip CK, Sanowar S, Andre I, Kimbrough TG, Vuckovic channel gating suggest a mechanism for pore opening. Nat Chem M, Pfuetzner RA, Deng W, Yu AC, Finlay BB, Baker D, Miller SI, Biol 2006;2:627–635. Strynadka NC. A conserved structural motif mediates formation of 58. Khalid S, Bond PJ, Carpenter T, Sansom MS. OmpA: gating and the periplasmic rings in the type III secretion system. Nat Struct dynamics via molecular dynamics simulations. Biochim Biophys Mol Biol 2009;16:468–476. Acta 2008;1778:1871–1880.

PROTEINS 661