The expression of heteroligomeric protein complexes for structural studies often requires a special coexpression strategy. The reason is that the solubility and proper folding of each subunit of the complex requires physical association with other subunits of the complex. The genomes of pathogenic mycobacteria encode many small protein complexes, implicated in bacterial fitness and pathogenicity, whose characterization may be further complicated by insolubility upon expression in Escherichia coli, the most common heterologous protein expression host. As protein fusions have been shown to dramatically affect the solubility of the proteins to which they are fused, we evaluated the ability of maltose binding protein fusions to produce mycobacterial Esx protein complexes. A single plasmid expression strategy using an N-terminal maltose binding protein fusion to the CFP-10 homolog proved effective in producing soluble Esx protein complexes, as determined by a small-scale expression and affinity purification screen, and coupled with intracellular proteolytic cleavage of the maltose binding protein moiety produced protein complexes of sufficient purity for structural studies. In comparison, the expression of complexes with hexahistidine affinity tags alone on the CFP-10 subunits failed to express in amounts sufficient for biochemical characterization. Using this strategy, six mycobacterial Esx complexes were expressed, purified to homogeneity, and subjected to crystallization screening and the crystal structures of the Mycobacterium abscessus EsxEF, M. smegmatis EsxGH, and M. tuberculosis EsxOP complexes were determined. Maltose binding protein fusions are thus an effective method for production of Esx complexes and this strategy may be applicable for production of other protein complexes.
Citation: Arbing MA, Chan S, Harris L, Kuo E, Zhou TT, Ahn CJ, et al. (2013) Heterologous Expression of Mycobacterial Esx Complexes in Escherichia coli for Structural Studies Is Facilitated by the Use of Maltose Binding Protein Fusions. PLoS ONE 8(11): e81753. https://doi.org/10.1371/journal.pone.0081753
Editor: Mark J. van Raaij, Centro Nacional de Biotecnologia – CSIC, Spain
Received: March 26, 2013; Accepted: October 15, 2013; Published: November 29, 2013
Copyright: © 2013 Arbing et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the Howard Hughes Medical Institute (www.hhmi.org) and National Institutes of Health (www.nih.gov) grants 23616-002-06 F3:02, TBSGC R01 (A1068135), and TBSGC P01 (AI095208). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Tuberculosis (TB) is the leading cause of death from a single infectious organism worldwide and Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis, is estimated to have infected one-third of the world's population . A high number of small protein complexes are encoded within the Mtb genome and are believed to play critical roles in bacterial virulence and pathogenesis –. Of particular interest are the Esx complexes which are dimers of two subunits (ESAT-6, early secreted antigen of 6 kDa; and CFP-10, culture filtrate protein of 10 kDa) and which are secreted across the cytoplasmic membrane by Type VII secretion (T7S) systems . Mycobacterial ESX secretion systems have been shown to be involved in a variety of physiological processes including conjugation in a non-pathogenic mycobacterium , , iron and/or zinc acquisition by both pathogenic and non-pathogenic mycobacterial species –, and virulence of pathogenic mycobacteria , . Consequently high resolution structural information is necessary to guide biochemical and biophysical studies of these protein complexes.
Structural studies of proteins, in general, are hampered by bottlenecks that exist at the key stages of production of soluble protein and protein crystallization ,  and the study of protein complexes is further complicated by the requirement to produce two, or more, protein subunits in soluble form. Additional complications in the study of Esx protein complexes is that the majority of M. tuberculosis proteins expressed in E. coli are insoluble  and that expression of Mtb protein complexes may require a coexpression strategy to obtain folded, soluble protein complexes . Strategies for coexpression of Mtb protein complexes in soluble form using E. coli expression systems are thus valuable in avoiding the use of other less well-characterized, and potentially cumbersome, bacterial expression systems and to avoid the need to explore laborious refolding and protein complex reconstitution schemes.
Certain highly soluble proteins have, when fused to target proteins, been demonstrated to promote the solubility, and in some instances, the folding of the proteins to which they are fused into biologically active form , . Maltose binding protein (MBP) fusions have been demonstrated to be exceptional in this regard , , thus we evaluated the ability of MBP fusions to promote the solubility and folding of Esx complexes. Our results demonstrate that MBP fusions are an efficient approach for the production of Esx complexes and, in combination with intracellular proteolytic cleavage of the MBP fusion partner, allow production of Esx complexes in amounts and purity suitable for structural studies. Using this expression approach we expressed and purified six Esx complexes and determined the crystal structures of M. abscessus EsxEF (EsxEFma), encoded by the MAB_3112 and MAB_3113 genes, M. smegmatis EsxGH (EsxGHms), encoded by the MSMEG_0620 and MSMEG_0621 genes, and M. tuberculosis EsxOP (EsxOPmt), encoded by the Rv2346c and Rv2347c genes, at resolutions of 1.96 Å, 2.70 Å, and 2.55 Å, respectively.
Materials and Methods
Vector construction and Esx complex cloning
The pET28b (EMD Millipore, Billerica, MA) vector was used as the basis for the construction of expression vectors: pMA507, which contains an N-terminal hexahistidine tag (His6) followed by a tobacco etch virus (TEV) protease cleavage site; pMA510, which has an N-terminal MBP fusion followed by a His6 tag and TEV protease site; pMAPLe3, which allows intracellular processing of an N-terminal MBP fusion by TEV protease to yield a target protein with a C-terminal His6 tag; and pMAPLe4, which allows intracellular processing of an N-terminal MBP fusion by tobacco vein mottling virus (TVMV) protease to yield a target protein with a TEV protease cleavable N-terminal His6 tag. All enzymes were obtained from New England Biolabs (Ipswich, MA).
pMA507 and pMA510 vector construction.
Briefly, pMA507 was constructed by inserting an oligonucleotide linker encoding the sequence MGSDKIGSHHHHHHENLYFQG between the NcoI and XhoI sites of pET28b. pMA510 was constructed by PCR-amplifying the MBP sequence from pMAL-C2 and inserting the PCR product between the NcoI restriction enzyme site and a BamHI restriction enzyme site encoded in the pMA507 linker such that amino acids 1–366 of mature processed E. coli MBP are inserted upstream of the His6 tag and TEV protease site. Both pMA507 and pMA510 contain the ccdB toxin gene downstream of the nucleotide sequence encoding the TEV protease site for use as a negative selection element in downstream cloning.
pMAPLe3 vector construction.
For in vivo processing of MBP fusions the DNA sequence encoding amino acids 1–366 of mature processed E. coli MBP was PCR amplified using primers that incorporated a TEV protease site and His6 tag after the MBP sequence and the PCR product was inserted into the NcoI and XhoI sites of pET28b. The resulting MBP protein has the following sequence at its C-terminus: NSSSENLYFQSTHHHHHH where the underlined amino acids (ST) incorporate a ScaI restriction enzyme site. Two stop codons following the DNA sequence that encodes the His6 tag were encoded in the reverse primer. The gene encoding TEV protease and the promoter that drives its expression was PCR-amplified from pRK603  and digested with the restriction enzymes PshAI and Tth111I whose recognition sequences were encoded in the primer extensions. This PCR product was digested with PshAI and Tth111I, treated with Klenow fragment to create blunt ends, phosphorylated with polynucleotide kinase, and ligated into the MBP containing vector which had been digested with PshAI to create pMAPLe2. To facilitate cloning into pMAPLe2, the ccdB toxin gene was PCR-amplified from pKM596  using primers that incorporated ScaI restriction enzyme sequences in the primer extensions. The PCR product was digested with ScaI and inserted into the ScaI site of pMAPLe2 to make pMAPLe3. Protein complexes cloned into pMAPLe3 have seven amino acids (THHHHHH) appended to the C-terminus of the 3′ gene product and a single serine added to the N-terminus of the of the 5′ gene product after TEV cleavage of the MBP moiety.
pMAPLe4 vector construction.
The TEV protease gene cassette in pMAPLe3 was replaced with the TVMV protease gene cassette, including the promoter that drives its expression, by the SLIC cloning technique using the plasmid pRK1037  as the source of TVMV protease. The nucleotide sequence between the MBP gene and ccdB negative selection element was then modified by PCR-amplification using primers that introduced a TVMV protease sequence to generate pMAPLe4. The resulting MBP protein has the following sequence at its C-terminus: GSETVRFQSHHHHHHSSSENLYFQS. Fusion proteins expressed from this vector are cleaved intracellularly by the TVMV protease expressed from the same vector resulting in the cleaved partner proteins having the following sequence at their N-terminus: SHHHHHHSSSENLYFQS. TEV protease treatment of the partner protein following affinity purification results in the partner protein having a single serine residue present at its N-terminus.
Cloning of Esx complexes.
Primer extensions contain 15 bp of sequence homologous to the flanking regions of the vector so that the PCR products could be inserted into the expression vectors using the SLIC cloning technique . Primer sequences for Esx complexes cloned in this study are listed in Table S1. All PCR reactions were performed using Phusion DNA polymerase and PCR products were purified from agarose gels using the QIAquick gel extraction kit (Qiagen, Germantown, MD) and treated with T4 DNA polymerase following the SLIC protocol. For cloning into pMA507 and pMA510 the vectors were PCR-amplified with primers (PIPE.Vec.For. and PIPE.Vec.Rev.) homologous to either side of the site into which the Esx PCR products were to be inserted and the vector was purified and treated with T4 DNA polymerase in the same manner as the Esx PCR products. For pMAPLe3 and pMAPLe4, the vector was digested with ScaI restriction enzyme and the linearized vector treated with T4 DNA polymerase. PCR products of the Esx gene pairs (90 ng) was added to 45 ng of vector and the DNA mixture was transformed into E. coli DH5α (Invitrogen, Carlsbad, CA). The bicistronic operons encoding the Esx complex gene pairs [MSMEG_0620-MSMEG_0621 (EsxGHms), rv2346c-rv2347c (EsxOPmt), rv3444c-rv3445c (EsxTUmt), and rv3904c-rv3905c (EsxEFmt)] were PCR-amplified from Mycobacterium smegmatis MC2155 and M. tuberculosis H37Rv genomic DNA and cloned into the pMA507, pMA510, and pMAPLe3 expression vectors. The bicistronic operons encoding the genes for three Esx complexes from Mycobacterium abscessus [EsxGHma (MAB_0665-MAB_0666), EsxEFma (MAB_3113-MAB_3112), and EsxTUma (MAB_3754c-MAB_3753c)] were PCR-amplified from M. abscessus genomic DNA and cloned into pMAPLe4. Positive transformants were identified by colony PCR using T7 and T7 terminator primers and the correct sequence of putative positive clones was verified by DNA sequencing (Genewiz, South Plainfield, NJ); the MAB_3112 and MAB_3113 genes were found to have eight mutations (MAB_3112: S23N, R37K, V75A, G93D; MAB_3113: P2A, T51A, E60Q, N77S) versus the protein sequences from the M. abscessus ATCC 19977 type strain.
Small-scale expression and purification of Esx complexes
Expression and solubility of affinity tagged Esx complexes and their ability to bind Ni-NTA agarose was evaluated in 96 well format using a variation of the protocol of Klock et al. . Briefly, E. coli BL21 (DE3) and/or E. coli Rosetta (DE3) harboring the expression plasmids were grown overnight in a 96 well block (Thomson Instrument Company, Oceanside, CA) at 37°C in 1 ml of LB media supplemented with 30 µg/ml kanamycin, and 34 µg/ml chloramphenicol for E. coli Rosetta (DE3), using a Shel Lab SI6R-HS shaking incubator with shaking at 650 rpm. The following day a 96 well block with 1 ml of fresh media supplemented with the appropriate antibiotics was inoculated with 50 µl of the overnight culture. The cultures were grown to an OD600 of 0.5 and protein expression induced by the addition of IPTG to a final concentration of 0.5 mM. The cultures were grown for an additional 4 hours at 37°C or, alternatively, were grown overnight at 18°C. Cultures were harvested by centrifugation and cell pellets were frozen and stored at −20°C pending lysis. The cell pellets were thawed and were resuspended in 500 µl lysis buffer A (20 mM HEPES, pH 7.5, 50 mM sucrose, 1 mM EDTA, 10 mM β-mercaptoethanol) supplemented with protease inhibitor cocktail (Sigma, St. Louis, MO), 1 mM PMSF, and ∼40 U/mL Ready-Lyse lysozyme (Epicentre, Madison, WI). The cell suspensions were shaken at 800 rpm for 20 minutes at 25°C and then MgCl2 was added to 5 mM and DNase I to 20 µg/ml. At this point 500 µl of lysis buffers B (50 mM HEPES, pH 7.8, 0.3 M NaCl), C (50 mM HEPES, pH 7.5, 1 M NaCl), or D (50 mM HEPES, pH 7.5, 0.3 M NaCl, 0.2% LDAO) were added to the cell lysate and the suspensions shaken for an additional 20 minutes. The lysates were clarified by centrifugation and the supernatants were transferred to a fresh block containing 50 µl of Ni-NTA agarose beads (Qiagen, Valencia, CA) and were incubated with shaking for one to two hours at 4°C. After incubation the suspensions were transferred to a 96-well filter plate and the beads were washed three times with 1 ml of wash buffer B (50 mM HEPES, pH 7.8, 0.3 M NaCl, 10 mM imidazole), C (50 mM pH 7.5, 1 M NaCl, 10 mM imidazole), or D (50 mM HEPES, pH 7.5, 0.3 M NaCl, 0.2% LDAO, 10 mM imidazole) prior to elution of the bound complexes by the addition of 100 µl of elution buffers B (50 mM HEPES, pH 7.8, 0.3 M NaCl, 0.3 M imidazole), C (50 mM pH 7.5, 1 M NaCl, 0.3 M imidazole), or D (50 mM HEPES, pH 7.5, 0.3 M NaCl, 0.2% LDAO, 0.3 M imidazole). Eluates were analyzed by SDS-PAGE using Criterion gels (Bio-Rad Laboratories, Hercules, CA).
Large-scale expression and purification of Esx complexes
Selenomethionine(SeMet)-labeled EsxEFma, EsxGHms, and EsxOPmt complexes were expressed as previously described  in E. coli BL21 (DE3) with the following modifications: the EsxEFma and EsxGHms cultures were grown for 18 hours at 18°C after induction of protein expression with 0.5 mM IPTG while the EsxOPmt culture was grown under the same conditions but protein expression was induced with 1.0 mM IPTG. An unlabeled preparation of the EsxEFma complex was prepared using the same expression and purification conditions as for the SeMet-labeled EsxEFma complex with minor modifications as noted. The cells were harvested by centrifugation and the cell pellets were stored at −20°C pending lysis. Cell pellets were resuspended in lysis buffer (EsxEFma: 20 mM Tris, pH 8.0, 300 mM NaCl, 10% glycerol; EsxGHms and EsxOPmt: 50 mM HEPES, pH 7.8, 150 mM NaCl) containing protease inhibitor cocktail (Sigma), 2 mM β-mercaptoethanol, 1 mM PMSF, DNase I (0.5 µg/ml), and lysozyme. The cells were lysed by sonication, and the lysates clarified by centrifugation (30,000 x g for 30 minutes at 4°C). The supernatants were incubated with Ni-NTA agarose beads (Qiagen, Valencia, CA) for one to two hours at 4°C and the suspension was then poured into a gravity column. The beads were washed twice with wash buffer (lysis buffer with 10 mM imidazole and for EsxGHms 10% glycerol was included), once with high salt buffer (same composition as wash buffer but with 1 M NaCl), again with wash buffer containing 50 mM imidazole, and the complexes then eluted with elution buffer (same composition as wash buffer but with 0.3 M imidazole). The complexes were further purified by size exclusion chromatography using a HiPrep 16/60 Sephacryl S-100 column (GE Healthcare, Piscataway, NJ) equilibrated in: 20 mM Tris, pH 8.0, 0.3 M NaCl, 10% glycerol (SeMet-labeled EsxEFma), 20 mM Tris, pH 8.0, 150 mM NaCl, 10% glycerol, 2 mM β-mercaptoethanol (native EsxEFma), 20 mM HEPES, pH 7.8, 150 mM NaCl, 10% glycerol, 2 mM β-mercaptoethanol (EsxGHms), or 50 mM HEPES pH 7.8, 150 mM NaCl (EsxOPmt). The N-terminal His6 tag was cleaved from the native and SeMet-labeled EsxEFma complexes by TEV protease and the cleaved affinity tag and His6-tagged TEV protease separated from the cleaved target protein by passing the sample over Ni-NTA agarose beads prior to the size exclusion chromatography step. The SeMet-labeled EsxEFma complex was dialyzed into buffer IEX-A (20 mM Tris, pH 8.0, 150 mM NaCl, 10% glycerol) and further purified by ion exchange chromatography using a HiTrap Q HP column (GE healthcare) using a linear gradient from 0–100% buffer IEX-B (20 mM Tris, pH 8.0, 500 mM NaCl, 10% glycerol). The pure SeMet-labeled EsxEFma complex was subsequently dialyzed against storage buffer (20 mM Tris, pH 8.0, 0.3 M NaCl, 10% glycerol). The purified complexes were then concentrated for crystallization screening [EsxEFma (native): 11 mg/ml; EsxEFma (SeMet-labeled): 4.5 mg/ml; EsxGHms: 27 mg/ml; EsxOPmt: 16 mg/ml].
Crystallization of Esx complexes
All crystallization reactions were performed using the hanging drop vapor diffusion method at 18°C. Crystals of the EsxEFma native complex were grown by mixing protein stock solution 1∶2 with reservoir solution (0.1 M tri-sodium citrate, pH 5.6, 0.2 M potassium sodium tartrate, 2 M ammonium sulfate) while the SeMet-labeled EsxEFma crystals were grown by mixing the protein solution 1∶1 with reservoir solution (1.8 M tri-ammonium citrate, pH 7.0). Irregular “coffee bean” shaped crystals of the native EsxEFma complex grew in 1 week while “spearhead” shaped crystals of the SeMet-labeled EsxEFma complex grew in 4–6 weeks. Native EsxEFma crystals were cryo-protected in reservoir solution containing 25% glycerol and SeMet-labeled crystals were flash frozen without any additional manipulation. Hexagonal rod crystals of the EsxGHms complex were grown in 2 months by mixing the protein stock solution 1∶1 with reservoir solution (460 mM potassium sodium tartrate, 35% glycerol, 95 mM HEPES, pH 7.5) and were mounted directly from the crystallization drop. Crystals of EsxOPmt in crystal form I were grown by mixing protein stock solution 2∶1 with reservoir solution (9% isopropanol, 90 mM sodium acetate trihydrate pH 4.6, 200 mM CaCl2). Hexagonal rod shaped crystals of this complex grew in 2–3 weeks and were cryoprotected with paraffin oil. Trapezoidal prism crystals of EsxOPmt (form II) were grown by mixing complex at 8 mg/ml in storage buffer (20 mM HEPES, pH 7.8, 150 mM NaCl, 2 mM ZnSO4, 38 mM β-mercaptoethanol) 1∶1 with reservoir solution (11% PEG 3350, 40 mM citric acid pH 3.5). Crystals of EsxOPmt (form II) grew to full size in 2–3 weeks and were cryoprotected using paraffin oil.
Data collection and structure determination
Diffraction data were collected on beamlines 24-ID-C and 24-ID-E at the Advanced Photon Source of Argonne National Lab. Diffraction data for EsxEFma and EsxOPmt (form II) were processed with XDS  while data for EsxGHms and EsxOPmt (form I) were processed and scaled with DENZO and Scalepack . The position of the selenium sites for the EsxEFma and EsxGHms substructures were found with HKL2MAP  while the selenium sites in the EsxOPmt (form I) substructure were determined using Phenix . Initial models were built with Phenix AutoBuild  and subsequent model building was performed manually using Coot . The unlabeled structures of EsxEFma and EsxOPmt (form II) were solved by molecular replacement using the program phenix.phaser  using the models obtained from the respective SeMet-labeled crystals. Refinement of EsxEFma and EsxOPmt (forms I and II) structures was performed with Phenix.refine  and the EsxGHms structure was refined with Refmac . Figures were prepared with PyMOL (), structural homology searches were performed with DALI , electrostatic surface potentials were calculated with APBS , solvent-accessible surface area was calculated with PISA , shape complementarity was calculated with SC , and mapping of sequence identity onto molecular surfaces was performed with the ConSurf server  using sequence alignments generated with Clustal Omega . Unless otherwise mentioned the following structures and chains were used for the calculations described above: EsxEFma, PDBid 4IOX, chains A and B; EsxGHms, PDBid 3Q4H, chains A and B; EsxOPmt, PDBid 4GZR, chains C and D. Protein identifiers for protein sequences used to generate sequence alignments and ConSurf calculations are listed in Table S2. The coordinates and molecular structure factors for the Esx complex structures determined in this study have been deposited in the Protein Data Bank (http://www.rcsb.org) under the accession codes: 4IOX (EsxEFma), 3Q4H (EsxGHms), 3OGI (EsxOPmt, form I), and 4GZR (EsxOPmt, form II).
Expression of mycobacterial Esx complexes
Four mycobacterial Esx complexes were cloned using the native bicistronic operon structure which has the CFP-10 homolog subunit separated from the downstream ESAT-6 homolog by an intergenic region of variable length and sequence (Table S3). The Esx operons were cloned into E. coli expression vectors (Figure 1) that allowed expression of the complexes with either: 1, an N-terminal His6 tag and TEV protease cleavage site on the CFP-10 homolog (pMA507); 2, an MBP fusion protein, His6 tag and TEV protease cleavage site on the N-terminus of the CFP-10 homolog (pMA510); or 3, with a C-terminal His6 tag on the ESAT-6 homolog and an N-terminal MBP fusion and TEV protease site on the CFP-10 homolog that is cleaved intracellularly by concurrent expression of TEV protease during recombinant protein expression (pMAPLe3). To simplify the cloning process we chose to express the Esx complexes from a single bicistronic message as our previous experience with Mtb protein complexes (our unpublished results) and that of others  has shown that E. coli recognizes the M. tuberculosis operon structure and can produce two proteins from a single bicistronic transcript.
Bicistronic operons encoding the Esx complex genes, separated by a naturally occurring intergenic region of variable length, were cloned into three different expression vectors. (A) The two subunits are coexpressed from a single bicistronic transcript with an N-terminal His6 tag and TEV protease site on the CFP-10 homolog. (B) The two subunits are coexpressed from a single bicistronic transcipt with an N-terminal MBP fusion with His6 tag and TEV protease site on the CFP-10 homolog. (C) The two subunits are coexpressed from a single bicistronic transcript with an N-terminal MBP fusion with TEV protease site on the CFP-10 homolog and a C-terminal His6 tag on the ESAT-6 homolog. Concurrent expression of TEV protease (TEVp) cleaves the MBP moiety from the CFP-10 homolog intracellularly at the TEV protease site positioned between the MBP C-terminus and CFP-10 N-terminus.
Maltose binding protein is a potent solubilizing fusion partner and is capable of solubilizing misfolded proteins  thus an assay is required to ascertain whether the protein of interest exists in a folded or unfolded state. Esx complexes lack a defined assayable enzymatic function so small-scale affinity purification was used to assess whether the expressed subunits of the Esx complexes were soluble and properly folded. We reasoned that the ability of the untagged subunit to copurify with the His6-tagged subunit indicates that both complex subunits are properly folded and that the proper dimeric complex has been formed. Small-scale affinity purification screening is particularly well suited for structural genomics projects in that the conditions under which a complex is soluble and can be purified are rapidly determined. However, it does not determine the relative expression and solubility levels of the individual components of the complex. The small-scale affinity purification results are summarized in Table S4. Large-scale protein purification that includes a size exclusion chromatography step was subsequently used to validate the small-scale affinity purification results.
When the Esx complexes were expressed with a His6 tag on the CFP-10 subunit alone none of the four tested Esx complexes were purified by small-scale affinity purification (Figure 2, lanes 1, 4, 7, and 10). With extensive expression optimization we were able to obtain expression of the complexes in soluble form but in trivial amounts insufficient for biochemical characterization (data not shown). This result is expected as the majority of M. tuberculosis proteins expressed in E. coli are insoluble . Moreover, our experience with M. tuberculosis PE-PPE complexes  has shown that even with a coexpression strategy Mtb protein complexes are often insoluble when expressed in E. coli.
(A) SDS-PAGE analysis (Any-kD TGX gel, Bio-Rad) of small-scale expression and affinity purification of His6–tagged proteins: lanes 1–3, EsxGHms complex; lanes 4–6, EsxOPmt complex; lanes 7–9, EsxTUmt complex; and lanes 10–12, EsxEFmt complex. The first lane for each complex is expression of the complex from the pMA507 vector with the His6 tag alone on the N-terminus of the CFP-10 homolog, the second lane is the complex expressed from pMA510 which expresses the CFP-10 homolog with an N-terminal MBP-His6 fusion (indicated by an asterisk), and the third lane is the complex expressed from the pMAPLe3 vector which allows proteolytic cleavage of the MBP moiety in vivo and purification of the complex via a C-terminal His6 tag on the ESAT-6 homolog. Arrows indicate the presence of Esx complex subunits. (B) SDS-PAGE analysis (Any-kD TGX gel, Bio-Rad) of concentrated Esx complex samples from large-scale purification (∼3 µg per lane): 1, EsxOPmt; 2, EsxEFmt; 3, EsxGHms; 4, EsxGHma; 5, EsxEFma; and 6, EsxTUma.
The ability of MBP fusions to promote folding of fused passenger proteins was apparent when we expressed the CFP-10 subunits with an N-terminal MBP fusion. When the CFP-10 homologs were expressed as MBP fusions using the pMA510 vector we found that in all four test cases the CFP-10 homologs were successfully expressed as soluble MBP fusions (Figure 2, lanes 2, 5, 8, and 11). However, in only two instances were the cognate ESAT-6 homolog protein partners copurified using this strategy (Figure 2, lanes 2 and 5). The second MBP fusion protein strategy employed intracellular processing of the MBP fusion proteins to evaluate whether the passenger proteins would be soluble after cleavage of the MBP molecule , . The expression vector used in this approach (pMAPLe3) was engineered to allow simultaneous coexpression of TEV protease to cleave passenger proteins from the MBP moiety in vivo via a TEV protease site positioned between MBP and the CFP-10 passenger protein. This approach allowed three of four Esx complexes to be successfully expressed in soluble form and affinity purified via the C-terminal His6 tag on the ESAT-6 subunit (Figure 2, lanes 3, 6, and 12).
Crystallization and structure determination of mycobacterial Esx complexes
The Esx complexes found to be expressed and soluble in the small-scale screen using the intracellular processing method were grown in large-scale and purified to homogeneity by affinity and size exclusion chromatography (Figure 2B). The relative amounts of purified protein obtained from large-scale purification (EsxGHms, ∼8 mg/L; EsxOPmt, ∼12 mg/L; and EsxEFmt, ∼2.5 mg/L) were consistent with the yields from the small-scale screen. Three additional Esx complexes (EsxGHma, EsxEFma, and EsxTUma) from M. abscessus, a fast-growing pathogen, were cloned into pMAPLe4, an expression vector similar to pMAPLe3 with the exception that the MBP moiety is cleaved from the CFP-10 passenger by TVMV protease leaving a TEV protease cleavable His6 tag on the N-terminus of the CFP-10 subunit. The substitution of an N-terminal cleavable tag on the CFP-10 subunit in lieu of the C-terminal His6 tag on the ESAT-6 subunit was deemed to be more amenable to generating a well ordered crystal lattice as the number of potentially disordered amino acid residues is reduced. The three M. abscessus Esx complexes were found to be soluble using our small-scale expression and purification screening method and were subsequently purified in large-scale in quantities suitable for crystallization screening (Figure 2B).
The six Esx complexes purified in large-scale were subjected to extensive crystallization screening and strongly diffracting crystals were obtained for EsxEFma, EsxGHms, and EsxOPmt. The structures of these complexes were determined by single wavelength anomalous diffraction (EsxEFma and EsxGHms) and multiple wavelength anomalous diffraction (EsxOPmt) at resolutions of 3.0 Å, 2.70 Å, and 2.55 Å, respectively. A crude low resolution model of the SeMet-labeled EsxEFma complex was used to obtain a molecular replacement solution for a high resolution (1.96 Å) EsxEFma native dataset and a second crystal form of EsxOPmt, solved by molecular replacement using the EsxOPmt model from form I, at 2.55 Å was also obtained. Diffraction data, refinement statistics, and model contents are summarized in Tables 1 and 2.
Structures of mycobacterial EsxEFma, EsxGHms, and EsxOPmt complexes
The asymmetric units for the EsxEFma crystal forms have a single complex in the asymmetric unit for the SeMet-labeled protein and six complexes in the asymmetric unit of the native protein crystal form while the EsxGHms and EsxOPmt (both crystal forms) contain two heterodimeric complexes. The overall architecture of the complexes is similar to known Esx structures , , ,  in that each heterodimer forms a four helix bundle with the CFP-10 and ESAT-6 subunit homologs each contributing an α-helical hairpin to the complex (Figure 3). The α-helices of each subunit are connected by a flexible loop containing the canonical ESAT-6/CFP-10 signature motif (Trp-Xaa-Gly; WXG) although the EsxGms CFP-10 homolog contains a modified motif (His-Xaa-Gly). The WXG-containing loops are well ordered in the EsxGHms complexes and in most of the EsxEFma complexes but the majority of the connecting loops are disordered in the EsxOPmt subunit structures (Table 2). Those connecting loops which were found to be ordered have, as in existing Esx structures, the tryptophan sidechain of the WXG motif buried in the subunit interface. Additional regions of disorder are found at the N- and C-termini of the complexes with the EsxFma and EsxOmt subunits having a substantial number of disordered residues at their C-termini.
The CFP-10 homologs are colored red and the ESAT-6 homologs are colored blue. (A) the EsxEFma complex; (B), the EsxGHms complex; and (C), the EsxOPmt complex. The N- and C-termini of individual chains are labeled and the disordered loop region of EsxOmt that connects its two α-helices is indicated by a dashed line. The tyrosine and acidic residues of the secretion signals of the EsxGHms and EsxOPmt complexes are shown in stick representation.
The subunit interfaces in the four-helix bundles of the Esx complexes are similar to those of the previously determined Esx complex structures in that the intermolecular interface between the CFP-10 and ESAT-6 subunits is largely hydrophobic. The average amount of buried surface area in the subunit interfaces varies considerably [EsxEFma, ∼1605 Å2; EsxGHms, ∼1555 Å2; EsxOPmt (form I), ∼1025 Å2; and EsxOPmt (form II), ∼1175 Å2] and the percentage of the surface of each subunit buried in the complex interface is also highly variable, from 13.5–33%, and is related to the degree to which the termini of the subunits are ordered and to the conformations that they adopt. Shape complementarity (Sc) statistics of 0.77 for EsxEFma, 0.74 for EsxGHms, and 0.69 and 0.71 for EsxOPmt forms I and II, respectively, indicate that the interfaces are highly complementary, with biologically significant interfaces such as antibody-antigen interactions generally having values in the range of 0.65–0.85. The complexes are further stabilized by hydrogen bonds and salt bridges.
The heterologous expression of proteins in soluble form is a major bottleneck in structural studies and is particularly problematic for expression of Mtb proteins. A recent analysis of target status data for mycobacterial protein targets (http://www.webtb.org/Targets/), collected by the Tuberculosis Structural Genomics Consortium (TBSGC), found only 48% of Mtb proteins expressed in E. coli are soluble. In comparison, two studies of homologous overexpression of His6-tagged E. coli proteins in E. coli expression hosts found that 60% and 92.5% of the targets were soluble when expressed in small-scale format , . However, in the latter study only 73% of the targets were found to be soluble using large-scale expression and purification methods. Data from the Northeast Structural Genomics Consortium (http://www.spine.nesg.org) found 60% of 697 E. coli targets were expressed in soluble form although this number does not provide statistics on construct engineering approaches targeting individual domains or truncations of constructs to improve solubility. The TBSGC data for mycobacterial species (for which a significant number of target statuses are reported) other than Mtb finds that heterologous expression of target proteins from M. abscessus, M. avium, M. marinum, and M. avium subsp. paratuberculosis produces soluble protein at a similar rate (56–61%) as for homologous overexpression of E. coli proteins. The remaining mycobacterial species are at opposite end of the spectrum in terms of heterologous production of soluble proteins with only 40% of M. leprae proteins expressed in soluble form while over 72% of heterologously expressed M. smegmatis MC2 155 proteins were found to be soluble.
The failure of a protein to express in soluble form in a heterologous expression host is unlikely to be attributable to a single cause as multiple factors at many levels influence protein solubility. High GC-content and secondary structure in the mRNA 5′ region has been shown to have a significant deleterious effect on protein expression levels  suggesting that the 66% GC-content of the Mtb genome, relative to 51% for E. coli, could be partially responsible for the insolubility of heterologously expressed Mtb proteins. However, the GC-content of other mycobacterial genomes is similar to Mtb and the TBSGC data (www.webtb.org/Targets) shows that proteins from these species, with the exception of M. leprae, are expressed in soluble form almost as often as homologously expressed E. coli proteins. Thus GC-content alone is not the sole factor responsible for the poor solubility of heterologously expressed Mtb and M. leprae proteins.
Protein solubility is also affected by differences in organism-specific codon usage (codon usage bias) between the heterologous expression host and the source of the gene to be expressed . Codon usage has a substantive effect on co-translational protein folding by altering the rate at which the nascent polypeptide is synthesized . Rare, or slow, codons are present in specific regions of mRNA messages and function to alter translation rates, and thus protein folding, in mRNA segments that correspond to protein domain boundaries ,  or in the transition from unstructured to structured regions of the nascent polypeptide chain , allowing structural elements additional time to fold. As a result of differences in codon usage bias the relative position of rare codons within a target gene in regard to structural elements is likely to be altered versus the expression host. As a result protein translation rates are likely to be significantly altered with a potential negative effect on protein solubility. Attempts to manipulate the expression of heterologous proteins by mitigating the effects of rare codons, through the use of synthetic genes or accessory plasmids encoding tRNA molecules for rare codons, have had mixed results. The increased translation rates associated with these strategies appear to uncouple the co-translational process of chain elongation and protein folding leading to high amounts of misfolded, insoluble protein , .
The physicochemical properties of the proteins themselves are also an important determinant of protein solubility. An analysis of data from large-scale structural genomics centers found the proteins most likely to be successfully expressed were of shorter length, had lower hydrophobicity, and were moderately acidic . The results from large-scale structural genomic centers may be skewed as the methodology employed by the centers may tend to favor production of proteins with these particular characteristics. Additional complicating factors in heterologous protein expression are the potential for organism-specific chaperones and post-translational protein modifications. Goldstone et al. found that five of eight Mtb proteins that were insoluble when expressed in E. coli were soluble when expressed in M. smegmatis, a close relative of M. tuberculosis . The authors speculate that mycobacterial chaperones may be responsible for the increase in target protein solubility, but there are many other factors, including the effects of codon usage bias, which may be responsible for the dramatic increase in protein solubility.
The difference in success rate between different mycobacterium for the heterologous production of target proteins in soluble form is unusual. One would expect that the relatedness of mycobacterial species and similar GC-content would result in a similar success rate for the heterologous production of target proteins in soluble form. A possible reason for the discrepancy is that target selection for strains other than Mtb and M. leprae may be biased towards targets that have a greater chance of being expressed in soluble form. Structural genomics efforts investigating mycobacterial protein structures have concentrated on Mtb with over 2200 unique proteins having been targeted (http://www.webtb.org/Targets/) while the other mycobacterium represented in the TBSGC data have only 170–370 targets per species. As more data is accumulated the question of the solubility of heterologously expressed mycobacterial proteins may be better answered.
MBP fusions promote Esx complex expression
We used a small-scale affinity purification screen to evaluate the ability of maltose binding protein fusions to produce soluble, folded Esx complexes. Coexpression of the ESAT-6 and CFP-10 subunits homologs without an N-terminal MBP fusion on the CFP-10 subunit failed to produce any of the four test complexes in yields sufficient for structural studies. This result is consistent with our past experiences demonstrating that coexpressing the two subunits of a complex is not necessarily sufficient for the production of Mtb protein complexes in soluble form in E. coli . In contrast, the use of N-terminal MBP fusions to the CFP-10 homologs allowed three of the four tested Esx complexes to be expressed and purified in quantities suitable for structural studies.
The ability of MBP to facilitate production of recalcitrant mycobacterial proteins was evident in our first MBP fusion strategy in which all four CFP-10 homologs were successfully expressed and purified as soluble N-terminal MBP fusions. However, in only two of four instances were both subunits of the heteroligomeric Esx complex copurified using this strategy (Figure 2). Multiple explanations are possible for the failure of the ESAT-6 homologs to copurify with their cognate CFP-10 partners: 1, the ESAT-6 homolog may not be expressed; 2, the ESAT-6 subunit may be insoluble; 3, the CFP-10 subunit may be unable to fold into a conformation allowing productive complex formation; and 4, the transient physical interactions between MBP and its fused passenger protein, believed to prevent aggregation of the passenger protein , may prevent interaction of the CFP-10 molecule and its natural ESAT-6 protein partner. A limitation of screening for soluble complexes is that it does not distinguish amongst these possibilities and specifically whether the untagged subunit has been expressed in either soluble or insoluble form.
In our second MBP fusion strategy we employed intracellular processing of the MBP fusion proteins to evaluate whether the passenger proteins would be soluble after cleavage of the MBP molecule , . As many proteins are insoluble after cleavage of the MBP moiety ,  this strategy determines whether MBP fusions are a viable experimental strategy for producing the passenger protein(s) in soluble form. Interestingly, while the EsxEFmt complex was expressed and purified using this strategy (Figure 2, lane 12) the dimeric complex was not purified using the unprocessed MBP fusion vector (Figure 2, lane 11). This suggests that the failure to copurify the ESAT-6 homologs for the EsxTUmt and EsxEFmt complexes in the absence of intracellular MBP cleavage may be attributable to two distinct pathways. In both instances MBP acts in its “holdase” capacity  maintaining its passenger, the CFP-10 homolog, in a soluble state. In the first instance, of EsxTmt, the passenger protein is likely not properly folded. While in the second instance, of EsxEmt, the passenger appears to fold productively but interactions with the ESAT-6 homolog (EsxFmt) may be physically prohibited by either MBP-EsxEmt interactions or by MBP impeding the EsxEmt-EsxFmt interactions required for complex formation. However, an alternative explanation is that the presence of the C-terminal His6 tag on the EsxFmt subunit, derived from the pMAPLe3 vector, increases the solubility of this subunit thus allowing the complex to form.
While our MBP fusion protein strategy for production of protein complexes is straightforward and time efficient, we did fail to purify the EsxTUmt complex. The screen for soluble protein complexes is particularly well-suited for structural genomics approaches targeting the most tractable targets. However, a major limitation of the screen is that it does not fully address the expression and solubility of the individual subunits and fails to identify conditions under which a component of the complex is expressed but is insoluble. As a result it is difficult to draw general conclusions about the expression and/or solubility of individual components of the complex under different experimental conditions. High value targets should thus be subjected to alternate salvage pathways to determine conditions that produce both complex subunits in soluble form. For mycobacterial protein complexes, in particular, use of the M. smegmatis expression system is a likely next step as the EsxTUmt and the EsxEFmt complexes were both successfully produced using this system .
Esx signal sequences adopt multiple conformations
The extent of disorder present at the N- and C-termini of both subunits of our mycobacterial Esx complex structures is not unprecedented as the structures of the EsxABmt , , EsxGHmt , and EsxRSmt  complexes have considerable amounts of disordered residues (between 4–20 amino acids) at the N- and C-termini of both complex subunits. Disordered subunit termini may mediate protein-protein interactions between Esx complexes and other proteins and there is evidence demonstrating the interaction of the ESAT-6 (EsxAmt) C-terminus with surface receptors of host immune system cells . Additional evidence implicates the C-terminus of CFP-10 homologs in protein-protein interactions as the CFP-10 C-terminus contains a signal sequence directing secretion of the complex across the cytoplasmic membrane  and mutation of either of two conserved amino acids in an amino acid motif (Tyr-Xaa-Xaa-Xaa-Glu/Asp) located within this region abolishes secretion .
The C-termini of the CFP-10 homolog subunits of our EsxGHms and EsxOPmt (forms I and II) structures, which contain the secretion motif, are ordered and adopt helical structures; in the EsxOPmt structure the secretion motif containing helix is an extension of the C-terminal helix of the α-helical hairpin while in EsxGHms the secretion motif is primarily contained in a short, 13 amino acid, helix linked to the α-helical hairpin by a short turn. The sidechains of the conserved amino acids in the secretion motif are oriented outward and away from the bundle in the EsxGHms structure while in the EsxOPmt structure the sidechains are oriented towards the center of the four helix bundle (Figure 3). The crystal structure of M. tuberculosis CFP-10 (EsxBmt; PDBid 3FAV-chain C) also has a partially ordered C-terminus and in this instance the side chains of the conserved amino acids of the secretion motif are in a helical region with the sidechains pointing to the central axis of the helical bundle. While crystal packing is most likely responsible for stabilizing these regions in different helical conformations, it is likely that the helical conformation is biologically relevant. Indeed, the N-terminal Type I signal sequences of proteins exported by the general secretory (Sec) system have been shown to adopt helical conformations in the presence of lipids or upon association with components of the Sec machinery , . The helical structure of the C-termini of multiple CFP-10 homologs and the i+4 spacing of the T7S secretion motif that positions the sidechains of critical amino acid residues on the same face of a helix both suggest that this conformation facilitates a specific protein-protein interaction, likely with another component of the T7S system secretion machinery.
Features of the Esx complex surfaces
The contribution of individual Esx complexes to mycobacterial fitness has yet to be fully elucidated, although the EsxAB and EsxGH complexes have been implicated in virulence ,  and metal acquisition –, respectively. To elicit clues to the function of the Esx complexes whose structures were determined in this study, we mapped electrostatic charge, hydrophobicity, and sequence conservation on the molecular surfaces of the complexes (Figures 4 and 5). Mapping of charge and hydrophobicity dose not reveal any obvious clues to the function of our Esx complexes as all three complexes show an even distribution of both properties on their surfaces. In contrast, our previous structure of the PE25-PPE41 complex, encoded by rv2341c-rv2340c, also an ESX secretion substrate, revealed a hydrophobic stripe on the surface of one face of the complex, suggesting a region involved in protein-protein interactions . The mapping of sequence conservation onto the surfaces of the Esx complexes was more informative. The sequence conservation on the surface of the EsxEFma complex is particularly interesting as the CFP-10 homolog side of the complex shows a high degree of surface variability, while the surface of the ESAT-6 homolog subunit exhibits a diagonal stripe of conserved residues, suggesting a region that may be important for function (Figure 4A). The EsxGHms complex has a region of significant sequence conservation at the end of the complex that includes the hairpin turn of EsxGms and the N- and C-termini of EsxHms (Figure 5B), which correlates with data implicating amino acid residues in this region of the complex in metal binding . At the other end of the EsxGHms complex there is significant sequence conservation in the N- and C-termini of EsxGms which both protrude above the core helical bundle structure (Figure 4B). The C-terminus of EsxGms contains the secretion motif required for complex secretion. Likewise, the C-terminus of EsxPmt, which protrudes above the EsxOPmt helical bundle, and which also contains the secretion motif, shows a high degree of sequence conservation (Figure 4C). There are other isolated patches of conserved amino acids on the surfaces of the structures whose functional significance cannot be interpreted in the absence of mechanistic studies exploring the roles of individual amino acid residues in the currently unknown functions of these complexes.
The complexes are shown in the same orientation as in Figure 3 with the ESAT-6 homolog subunit facing the viewer. (A) the EsxEFma with the surface colored by electrostatic potential (first column), hydrophobicity (second column), and sequence identity (third column). (B) the EsxGHms complex colored as in (A). (C) the EsxOPmt complex colored as in (A). Colored bars under each column indicate: column 1, electrostatic surface potentials of +/− 5 kT calculated at an ionic strength of 150 mM; column 2, hydrophobicity with a gradient of red (most hydrophobic) to blue (least hydrophobic); and column 3, the degree of sequence conservation with variable regions in teal, highly conserved regions in burgundy, and the regions where the degree of conservation could not be assigned with confidence in yellow. Sequence conservation was calculated using alignments of 15 (EsxEFma), 21 (EsxGHms), and 27 (EsxOPmt) pairs of homologous sequences (listed in Table S2).
View of the Esx complex surfaces with the complexes rotated 180° versus the orientations in Figure 3 to show the CFP-10 homolog side of the complex. With the exception of the rotation the parameters are the same as in Figure 4 with EsxEFma in (A), EsxGHms complex in (B), and the EsxOPmt complex in (C).
Comparison with structural homologs
A DALI search of structures deposited in the Protein Data bank was used to find clues to Esx function. However, the lack of unique structural features associated with the helical hairpin of a single CFP-10 or ESAT-6 subunit, or the four-helix bundle of the heterodimeric complex, results in a high degree of structural similarity with many proteins that are unlikely to share any evolutionary or functional overlap with Esx complexes. Nonetheless, the DALI search confirmed that the folds of the EsxEFma, EsxGHms,and EsxOPmt complexes are structurally similar to previously determined Esx complex structures including the homodimeric Esx complexes prevalent in non-mycobacteria (Table 3).
A DALI search, performed using the four helix bundle of the complex as a single chain search model, reveals that the Esx complexes have a high degree of structural similarity with the four helix bundle proteins of the ferritin superfamily (Figure 6 and Table 4). The data implicating Esx complexes in metal acquisition, specifically EsxGH, and the structural similarity between the Esx bundles and those of ferritin suggest a possible evolutionary relationship. The head-to-tail arrangement of the CFP-10 and ESAT-6 homologs of Esx complexes results in the helices of the Esx complexes having the same topology as those of the ferritin-like proteins. Helices 2 and 3 of ferritin-like protein bundles are connected by a long flexible loop. A similar loop joining the C-terminus of the CFP-10 homolog with the N-terminus of the ESAT-6 homolog would effectively generate a four-helix Esx complex bundle with an architecture similar to the ferritin-like helical bundles. However, a sequence and structural comparison of EsxGHms with various ferritins (data not shown) failed to identify the canonical ferridoxidase iron binding center and potential metal binding sites on the EsxGHms surface. Moreover, most ferritins assemble into spherical oligomeric structures of 12 or 24 subunits or head-to-tail dimers  and to date the only oligomeric state of an Esx complex greater than a heterodimer has been a domain swapped tetramer , which does not resemble ferritin-like dimer molecules. Nonetheless, the overall structural similarity and possibility of a gene fusion or fission event leading to the creation of one protein family from the other is intriguing.
The Esx complexes are oriented and colored as in Figure 3. For clarity only the N-termini of the CFP-10 homologs and C-termini of the ESAT-6 homologs are labeled. Ferritin-like proteins are colored orange and N- and C-termini are labeled. (A) Stereo view of the superposition of EsxEFma with a single subunit of the Sulfolobus solfataricus DPS-like dodecamer assembly (PDBid 2CLB, chain A; Z-score of 10.8 with an RMSD of 2.7 Å for the superposition of 126 amino acids with a sequence identity of 5%). (B) Stereo view of the superposition of EsxGHms complex with a single subunit of the E. coli YciE ferritin-like dimer (PDBid 3OGH, chain A; Z-score of 10.5 with an RMSD of 3.0 Å for the superposition of 134 amino acids with a sequence identity of 6%). (C) Stereo view of the superposition of EsxOPmt complex with a single subunit of the Bacillus anthracis BA_0993 hypothetical ferritin-like protein dodecamer (PDBid 2QQY, chain A; Z-score of 9.6 with an RMSD of 3.0 Å for the superposition of 114 amino acids with a sequence identity of 4%).
Our results demonstrate that MBP fusions are efficient means for the production of mycobacterial Esx protein complexes in E. coli. The approach is efficient in that expression of the two subunits from a single bicistronic transcript facilitates cloning, and small-scale affinity purification of complexes, coupled with intracellular TEV processing of the fusion protein, provides a rapid assessment of whether the complex is soluble and reconstituted in the absence of MBP. Because obtaining soluble protein is a major bottleneck in structural studies, this approach is useful as either a first approach for expression of protein complexes or, alternatively, as a salvage pathway when complexes fail to express using other methods.
Primers used for cloning of Esx complexes.
Length and sequence of intergenic region between CFP-10 and ESAT-6 homolog pairs examined in this study.
Data collection at Northeastern Collaborative Access Team beamline ID-24 at the Advanced Photon Source of Argonne National Laboratory was greatly facilitated by M. Capel, K. Rajashankar, N. Sukumar, J. Schuermann and I. Kourinov. We thank the UCLA-DOE Protein Expression Technology Center, UCLA-DOE X-ray Crystallography Core Facility, and the UCLA Crystallization Core Facility for assistance with protein purification and crystallization.
Conceived and designed the experiments: MA SC DE. Performed the experiments: MA SC LH EK TTZ CJA LN QH JL PTM AS MS DC. Analyzed the data: MA SC TH MS DC. Wrote the paper: MA DC DE.
- 1. World Health Organization (2013) Global Tuberculosis Report. 1–289.
- 2. Simeone R, Bottai D, Brosch R (2009) ESX/type VII secretion systems and their role in host-pathogen interaction. Curr Opin Microbiol 12: 4–10
- 3. Ramage HR, Connolly LE, Cox JS (2009) Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress responses, and evolution. PLoS Genet 5: e1000767
- 4. Mukhopadhyay S, Balaji KN (2011) The PE and PPE proteins of Mycobacterium tuberculosis. Tuberculosis (Edinb) 91: 441–447
- 5. Stanley SA, Raghavan S, Hwang WW, Cox JS (2003) Acute infection and macrophage subversion by Mycobacterium tuberculosis require a specialized secretion system. Proc Natl Acad Sci USA 100: 13001–13006
- 6. Coros A, Callahan B, Battaglioli E, Derbyshire KM (2008) The specialized secretory apparatus ESX-1 is essential for DNA transfer in Mycobacterium smegmatis. Mol Microbiol 69: 794–808
- 7. Flint JL, Kowalski JC, Karnati PK, Derbyshire KM (2004) The RD1 virulence locus of Mycobacterium tuberculosis regulates DNA transfer in Mycobacterium smegmatis. Proc Natl Acad Sci USA 101: 12598–12603
- 8. Siegrist MS, Unnikrishnan M, McConnell MJ, Borowsky M, Cheng T-Y, et al. (2009) Mycobacterial Esx-3 is required for mycobactin-mediated iron acquisition. Proc Natl Acad Sci USA 106: 18792–18797
- 9. Ilghari D, Lightbody KL, Veverka V, Waters LC, Muskett FW, et al. (2011) Solution structure of the Mycobacterium tuberculosis EsxG·EsxH complex: functional implications and comparisons with other M. tuberculosis Esx family complexes. J Biol Chem 286: 29993–30002
- 10. Serafini A, Boldrin F, Palù G, Manganelli R (2009) Characterization of a Mycobacterium tuberculosis ESX-3 conditional mutant: essentiality and rescue by iron and zinc. J Bacteriol 191: 6340–6344
- 11. Lewis KN, Liao R, Guinn KM, Hickey MJ, Smith S, et al. (2003) Deletion of RD1 from Mycobacterium tuberculosis mimics bacille Calmette-Guérin attenuation. J Infect Dis 187: 117–123
- 12. Pym AS, Brodin P, Brosch R, Huerre M, Cole ST (2002) Loss of RD1 contributed to the attenuation of the live tuberculosis vaccines Mycobacterium bovis BCG and Mycobacterium microti. Mol Microbiol 46: 709–717.
- 13. Terwilliger TC, Stuart D, Yokoyama S (2009) Lessons from structural genomics. Annu Rev Biophys 38: 371–383
- 14. Christendat D, Yee A, Dharamsi A, Kluger Y, Gerstein M, et al. (2000) Structural proteomics: prospects for high throughput sample preparation. Prog Biophys Mol Biol 73: 339–345.
- 15. Chim N, Habel JE, Johnston JM, Krieger I, Miallau L, et al. (2011) The TB Structural Genomics Consortium: A decade of progress. Tuberculosis 91: 155–172
- 16. Strong M, Sawaya MR, Wang S, Phillips M, Cascio D, et al. (2006) Toward the structural genomics of complexes: crystal structure of a PE/PPE protein complex from Mycobacterium tuberculosis. Proc Natl Acad Sci USA 103: 8060–8065
- 17. Esposito D, Chatterjee DK (2006) Enhancement of soluble protein expression through the use of fusion tags. Curr Opin Biotechnol 17: 353–358
- 18. Waugh DS (2005) Making the most of affinity tags. Trends Biotechnol 23: 316–320
- 19. Kapust RB, Waugh DS (1999) Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused. Protein Sci 8: 1668–1674
- 20. Kataeva I, Chang J, Xu H, Luan C-H, Zhou J, et al. (2005) Improving solubility of Shewanella oneidensis MR-1 and Clostridium thermocellum JW-20 proteins expressed into Esherichia coli. J Proteome Res 4: 1942–1951
- 21. Kapust RB, Waugh DS (2000) Controlled intracellular processing of fusion proteins by TEV protease. Protein Expr Purif 19: 312–318
- 22. Fox JD, Waugh DS (2003) Maltose-binding protein as a solubility enhancer. Methods Mol Biol 205: 99–117.
- 23. Nallamsetty S, Kapust RB, Tözsér J, Cherry S, Tropea JE, et al. (2004) Efficient site-specific processing of fusion proteins by tobacco vein mottling virus protease in vivo and in vitro. Protein Expression and Purification 38: 108–115
- 24. Li MZ, Elledge SJ (2012) SLIC: a method for sequence- and ligation-independent cloning. Methods Mol Biol 852: 51–59
- 25. Klock HE, Koesema EJ, Knuth MW, Lesley SA (2008) Combining the polymerase incomplete primer extension method for cloning and mutagenesis with microscreening to accelerate structural genomics efforts. Proteins 71: 982–994
- 26. Arbing MA, Kaufmann M, Phan T, Chan S, Cascio D, et al. (2010) The crystal structure of the Mycobacterium tuberculosis Rv3019c-Rv3020c ESX complex reveals a domain-swapped heterotetramer. Protein Sci 19: 1692–1703
- 27. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66: 125–132
- 28. Otwinowski Z, Minor W (1997) Processing of X-ray Diffraction Data Collected in Oscillation Mode. Methods in Enzymology 276: 307–326.
- 29. Pape T, Schneider TR (2004) HKL2MAP: a graphical user interface for macromolecular phasing with SHELX programs. Journal of Applied Crystallography 37: 843–844.
- 30. Adams PD, Afonine PV, Bunkóczi G, Chen VB, Davis IW, et al. (2010) PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66: 213–221
- 31. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66: 486–501
- 32. Murshudov GN, Vagin AA, Dodson EJ (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53: 240–255
- 33. Holm L, Kääriäinen S, Rosenström P, Schenkel A (2008) Searching protein structure databases with DaliLite v.3. Bioinformatics 24: 2780–2781
- 34. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA (2001) Electrostatics of nanosystems: Application to microtubules and the ribosome. PNAS 98: 10037–10041
- 35. Krissinel E, Henrick K (2007) Inference of macromolecular assemblies from crystalline state. J Mol Biol 372: 774–797
- 36. Lawrence MC, Colman PM (1993) Shape complementarity at protein/protein interfaces. J Mol Biol 234: 946–950
- 37. Ashkenazy H, Erez E, Martz E, Pupko T, Ben-Tal N (2010) ConSurf 2010: calculating evolutionary conservation in sequence and structure of proteins and nucleic acids. Nucleic Acids Res 38: W529–533
- 38. Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, et al. (2011) Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol 7: 539
- 39. Poulsen C, Holton S, Geerlof A, Wilmanns M, Song Y-H (2010) Stoichiometric protein complex formation and over-expression using the prokaryotic native operon structure. FEBS Lett 584: 669–674
- 40. Sun P, Tropea JE, Waugh DS (2011) Enhancing the solubility of recombinant proteins in Escherichia coli by using hexahistidine-tagged maltose-binding protein as a fusion partner. Methods Mol Biol 705: 259–274
- 41. Nallamsetty S, Waugh DS (2006) Solubility-enhancing proteins MBP and NusA play a passive role in the folding of their fusion partners. Protein Expr Purif 45: 175–182
- 42. Renshaw PS, Lightbody KL, Veverka V, Muskett FW, Kelly G, et al. (2005) Structure and function of the complex formed by the tuberculosis virulence factors CFP-10 and ESAT-6. EMBO J 24: 2491–2498
- 43. Vincentelli R, Bignon C, Gruez A, Canaan S, Sulzenbacher G, et al. (2003) Medium-Scale Structural Genomics: Strategies for Protein Expression and Crystallization. Acc Chem Res 36: 165–172
- 44. Ergin A, Büssow K, Sieper J, Thiel A, Duchmann R, et al. (2007) Homologous high-throughput expression and purification of highly conserved E coli proteins. Microbial Cell Factories 6: 18
- 45. Allert M, Cox JC, Hellinga HW (2010) Multifactorial determinants of protein expression in prokaryotic open reading frames. J Mol Biol 402: 905–918
- 46. Angov E (2011) Codon usage: Nature's roadmap to expression and folding of proteins. Biotechnol J 6: 650–659
- 47. Fedyunin I, Lehnhardt L, Böhmer N, Kaufmann P, Zhang G, et al. (2012) tRNA concentration fine tunes protein solubility. FEBS Lett 586: 3336–3340
- 48. Angov E, Hillier CJ, Kincaid RL, Lyon JA (2008) Heterologous protein expression is enhanced by harmonizing the codon usage frequencies of the target gene with those of the expression host. PLoS ONE 3: e2189
- 49. Zhang G, Hubalewska M, Ignatova Z (2009) Transient ribosomal attenuation coordinates protein synthesis and co-translational folding. Nat Struct Mol Biol 16: 274–280
- 50. Saunders R, Deane CM (2010) Synonymous codon usage influences the local protein structure observed. Nucleic Acids Res 38: 6719–6728
- 51. Rosano GL, Ceccarelli EA (2009) Rare codon content affects the solubility of recombinant proteins in a codon bias-adjusted Escherichia coli strain. Microb Cell Fact 8: 41
- 52. Slabinski L, Jaroszewski L, Rodrigues APC, Rychlewski L, Wilson IA, et al. (2007) The challenge of protein structure determination--lessons from structural genomics. Protein Sci 16: 2472–2482
- 53. Goldstone RM, Moreland NJ, Bashiri G, Baker EN, Shaun Lott J (2008) A new Gateway vector and expression protocol for fast and efficient recombinant protein expression in Mycobacterium smegmatis. Protein Expr Purif 57: 81–87
- 54. Nallamsetty S, Waugh DS (2007) Mutations that alter the equilibrium between open and closed conformations of Escherichia coli maltose-binding protein impede its ability to enhance the solubility of passenger proteins. Biochem Biophys Res Commun 364: 639–644
- 55. Austin BP, Nallamsetty S, Waugh DS (2009) Hexahistidine-tagged maltose-binding protein as a fusion partner for the production of soluble recombinant proteins in Escherichia coli. Methods Mol Biol 498: 157–172
- 56. Jeon WB, Aceti DJ, Bingman CA, Vojtik FC, Olson AC, et al. (2005) High-throughput purification and quality assurance of Arabidopsis thaliana proteins for eukaryotic structural genomics. J Struct Funct Genomics 6: 143–147
- 57. Raran-Kurussi S, Waugh DS (2012) The Ability to Enhance the Solubility of Its Fusion Partners Is an Intrinsic Property of Maltose-Binding Protein but Their Folding Is Either Spontaneous or Chaperone-Mediated. PLoS ONE 7: e49589
- 58. Pathak SK, Basu S, Basu KK, Banerjee A, Pathak S, et al. (2007) Direct extracellular interaction between the early secreted antigen ESAT-6 of Mycobacterium tuberculosis and TLR2 inhibits TLR signaling in macrophages. Nat Immunol 8: 610–618
- 59. Champion PAD, Stanley SA, Champion MM, Brown EJ, Cox JS (2006) C-terminal signal sequence promotes virulence factor secretion in Mycobacterium tuberculosis. Science 313: 1632–1636
- 60. Daleke MH, Ummels R, Bawono P, Heringa J, Vandenbroucke-Grauls CMJE, et al. (2012) General secretion signal for the mycobacterial type VII secretion pathway. PNAS 109: 11342–11347
- 61. Briggs MS, Cornell DG, Dluhy RA, Gierasch LM (1986) Conformations of signal peptides induced by lipids suggest initial steps in protein export. Science 233: 206–208.
- 62. Chou Y-T, Gierasch LM (2005) The conformation of a signal peptide bound by Escherichia coli preprotein translocase SecA. J Biol Chem 280: 32753–32760
- 63. De Leon J, Jiang G, Ma Y, Rubin E, Fortune S, et al. (2012) Mycobacterium tuberculosis ESAT-6 Exhibits a Unique Membrane-interacting Activity That Is Not Found in Its Ortholog from Non-pathogenic Mycobacterium smegmatis. J Biol Chem 287: 44184–44191
- 64. Andrews SC (2010) The Ferritin-like superfamily: Evolution of the biological iron storeman from a rubrerythrin-like ancestor. Biochim Biophys Acta 1800: 691–705