The crystal structure of the Mycobacterium tuberculosis Rv3019c-Rv3020c ESX complex reveals a domain-swapped heterotetramer

Abstract

Mycobacterium tuberculosis encodes five gene clusters (ESX-1 to ESX-5) for Type VII protein secretion systems that are implicated in mycobacterial pathogenicity. Substrates for the secretion apparatus are encoded within the gene clusters and in additional loci that lack the components of the secretion apparatus. The best characterized substrates are the ESX complexes, 1:1 heterodimers of ESAT-6 and CFP-10, the prototypical member that has been shown to be essential for Mycobacterium tuberculosis pathogenesis. We have determined the structure of EsxRS, a homolog of EsxGH of the ESX-3 gene cluster, at 1.91 Å resolution. The EsxRS structure is composed of two four-helix bundles resulting from the 3D domain swapping of the C-terminal domain of EsxS, the CFP-10 homolog. The four-helix bundles at the extremities of the complex have a similar architecture to the structure of ESAT-6·CFP-10 (EsxAB) of ESX-1, but in EsxRS a hinge loop linking the α-helical domains of EsxS undergoes a loop-to-helix transition that creates the domain swapped EsxRS tetramer. Based on the atomic structure of EsxRS and existing biochemical data on ESX complexes, we propose that higher order ESX oligomers may increase avidity of ESX binding to host receptor molecules or, alternatively, the conformational change that creates the domain swapped structure may be the basis of ESX complex dissociation that would free ESAT-6 to exert a cytotoxic effect.

Figure 1

Purification of dimeric and tetrameric forms of EsxRS. (A) Typical chromatographic profile for preparative SEC of EsxRS. Peak 1 contains an EsxRS complex with a molecular mass consistent with a tetramer while peak 2 contains an EsxRS complex with a molecular mass consistent with a dimer. (B) SDS—PAGE (20% gel) analysis of the two peaks from SEC purification. Lane M, Broad-range molecular weight markers; lane 1, peak 1 from SEC purification; lane 2, peak 2 from SEC purification. EsxRS complexes exist in two distinct oligomeric states, dimer and tetramer, that contain the EsxR and EsxS subunits in equal amounts.

Figure 2

The structure of the domain-swapped EsxRS complex. (A) Ribbon representation of the EsxRS structure. EsxS subunits are colored in red and orange and the EsxR subunits are in purple and teal. (B) Superposition of the two four-helix bundles of the EsxRS complex shows strong structural similarity. The orientation of the four-helix bundle at the top of the panel is in the same orientation as in Panel A. Subunit colors are the same. (C) Electrostatic surface potential of the EsxRS complex calculated with an ionic strength of 100 mM with red and blue representing negative and positive potentials, respectively. Fully saturated colors indicate a potential of magnitude ≥ ±3 kT. The first view in panel C is the same representation as in Panel A. The surface of EsxRS lacks the intense accumulation of charge associated with nucleic acid binding proteins implying that the EsxRS complex binds to specific target molecules as has been proposed for the EsxAB homolog. The structural similarity and symmetrical charge distribution between the four-helix bundles suggests that both extremities of EsxRS are capable of binding to receptor proteins, which has implications for inhibiting host cell processes that involve protein dimerization. An interactive view is available in the electronic version of the article.PRO451 Figure 2

Figure 3

Structural similarity of EsxRS and related ESX molecules. (A) Stereoview of the superposition of EsxRS with M. tuberculosis EsxAB (ESAT-6·CFP-10) in worm representation with EsxRS oriented and colored as in Figure 2. The EsxA (ESAT-6) subunit is colored in green and EsxB (CFP-10) is colored in yellow. (B) Stereoview of the superposition of Staphylococcus aureus EsxA (SaEsxA) on the EsxR subunit of the EsxRS complex. EsxRS is colored as in Panel A and SaEsxA is colored pink. For clarity only the termini of EsxAB and SaEsxA are labeled. Strong structural similarity between the EsxRS and EsxAB complexes suggests that mycobacterial ESX complexes have related functions whereas the divergence in the SaEsxA structure suggests that more distantly related ESX homologs may have evolved new functions. An interactive view is available in the electronic version of the article.PRO451 Figure 3

Figure 4

Sequence alignment of CFP-10 (bottom) and ESAT-6 (bottom) homologs. The alignment was prepared with ClustalW and colored using BOXSHADE. Secondary structure elements are represented by sinusoidal lines representing α-helices with 3₁₀-helices represented by the lighter color. Disordered or missing regions of the structure are represented by black lines. The position of the hinge loop in EsxS (residues 41–46) is indicated as are the salt bridges formed between EsxS domains (red triangles, Arg26-Asp70; black triangles, Glu33-His55) and the intermolecular salt bridge formed between EsxS Lys21 and EsxR Glu35 (blue triangles). Residues involved in the intermolecular interface are designated by black circles. From the structure presented here and from the high conservation in residues 41–46 (this figure) we can speculate that the EsxGH complex will also be capable of domain swapping.