Crystal structure of the N-terminal domain of EccA₁ ATPase from the ESX-1 secretion system of Mycobacterium tuberculosis

Abstract

EccA₁ is an important component of the type VII secretion system (T7SS) that is responsible for transport of virulence factors in pathogenic mycobacteria. EccA₁ has an N-terminal domain of unknown function and a C-terminal AAA+ (ATPases associated with various cellular activities) domain. Here we report the crystal structure of the N-terminal domain of EccA₁ from Mycobacterium tuberculosis, which shows an arrangement of six tetratricopeptide repeats that may mediate interactions of EccA₁ with secreted substrates. Furthermore, the size and shape of the N-terminal domain suggest its orientation in the context of a hexamer model of full-length EccA₁.

INTRODUCTION

Mycobacterium tuberculosis employs an army of secreted proteins to subvert the immune system during infection. In order to transport these virulence factors across an unusual two-membrane cell envelope mycobacteria use the specialized type VII secretion system (T7SS).¹ The T7SS machinery is an ~1500 kDa complex composed of a membrane pore and associated proteins including membrane-associated and cytosolic ATPases that are thought to provide the energy for transporting protein cargoes across the membrane. In the genome of M. tuberculosis there are five homologous T7SS clusters named ESX-1 to ESX-5. The most intensely studied region, ESX-1, includes the Rv3868 gene encoding the AAA+ ATPase EccA₁. Like other AAA+ ATPases EccA₁ forms oligomers, possibly hexamers, and hydrolyzes ATP in vitro.² Furthermore, its ATPase activity promotes virulence in vivo through increased mycolic acid synthesis.³ In addition EccA₁ has been shown to be essential in vivo for specific targeting and secretion of EspC and other co-secreted virulence factors such as ESAT-6/CFP-10 through the ESX-1 system.⁴

Sequence analysis of EccA₁ indicates that it contains a C-terminal ATPase domain and a tetratricopeptide repeat (TPR) containing N-terminal domain of unknown function [Fig. 1(A)]. However, there is not yet any structure of an AAA+ ATPase containing a TPR domain, nor is there any reported structure with significant primary sequence identity with the EccA₁ N-terminal domain. To gain insight into the function of EccA₁ and related T7SS ATPases, we solved the structure of the N-terminal TPR domain of EccA₁ from M. tuberculosis.

Figure 1.

EccA₁ N-terminal domain structure. (A) A schematic diagram of EccA₁’s TPR repeats, β finger insert, and C-terminal AAA+ ATPase domain. TPR motifs 1–6 are colored in rainbow colors from red to blue, whereas residues 237-273 are colored purple to highlight a capping C-terminal α helix or a possible TPR motif that was truncated in this construct. (B) Cartoon representation of EccA₁ N-terminal structure. TPR repeat motifs are colored as in (A). The β finger insert is highlighted in gray. (C) Structural conservation of the TPR superhelix motif between EccA₁ (alternating red / orange) and PilF (green). The β finger insert is highlighted in gray.

METHODS

Expression, purification and crystallization

The gene fragment corresponding to the N-terminal domain of EccA₁, residues 1–280, was PCR amplified from genomic DNA of M. tuberculosis H37Rv and cloned into a modified pET-28b vector (EMD Millipore) to encode an N-terminal His₆-tag followed by a tobacco etch virus (TEV) protease cleavage site. The resulting vector was transformed into Rosetta (DE3) cells (EMD Millipore) for expression. The cells were grown in Luria broth at 37°C to OD₆₀₀ 0.6 and induced for 4 h with 0.5 mM isopropyl-1-thio-β-D-galactopyranoside at 25 °C. Cells were harvested by centrifugation and resuspended in buffer containing 20 mM Tris-HCl pH 8.4, 300 mM NaCl, and 20 mM imidazole. The resuspended cells were lysed using an EmulsiFlex-C5 microfluidizer (Avestin) and EccA₁ was purified from the soluble fraction of the lysed cells using Ni-nitrilotriacetic acid agarose (Qiagen) column followed by His₆-tag cleavage with TEV protease. EccA₁ was further purified by size-exclusion using a Superdex200 column (GE Healthcare) in buffer containing 10 mM Tris-HCl pH 8.4, 200 mM NaCl. Crystallization screens were performed using the vapor diffusion method in a hanging drop 96-well plate format with JCSG Core Suites I–IV (Qiagen). The optimized EccA₁ crystals were grown in sitting drop format using 0.1 M sodium citrate pH 5.6, 1.0 M lithium sulfate, 0.5 M ammonium sulfate.

Data collection and structure determination

Crystals were soaked in crystallization solution supplemented with 20% glycerol and flash cooled in liquid nitrogen. To obtain heavy atom derivatives, the crystals were soaked for brief time periods, 30 s to 2 min, in cryo-protectant solution supplemented with compounds from Heavy Atom Screens (Hampton Research).⁵ Data were collected at Southeast Regional Collaborative Access Team (SER-CAT) 22-ID beamline at the Advanced Photon Source, Argonne National Laboratory. Data were processed and scaled using XDS and XSCALE.⁶

The structure of EccA₁ was determined by single wavelength anomalous diffraction (SAD) method using data from a single crystal soaked in the presence of KAu(CN)₂. Initial Au sites were found using SHELXD⁷ at resolution 3.3 Å. Additional Au sites were found and refined using SAD protocol in PHASER.⁸ Following density modification in Parrot,⁹ an initial model was built using Buccaneer.¹⁰ After manual rebuilding in Coot,¹¹ the improved model was refined using REFMAC5¹² against a ‘native’ dataset at resolution 2.0 Å. The ‘native’ dataset was collected from a crystal soaked in the presence of samarium acetate. A preliminary data analysis showed a lack of strong anomalous signal in this dataset, however, three partially occupied samarium ions were identified by peaks in anomalous difference maps and were included in the final model. The final rounds of refinement were performed applying eight translation, libration and screw-rotation displacement (TLS) groups determined by the TLSMD server.¹³ The final model contains two EccA₁ monomers (residues 1-273) in the asymmetric unit, three samarium ions, six sulfate ions and 453 water molecules. The structure was validated using Coot and the Molprobity server (http://molprobity.biochem.duke.edu). Coordinates and structure factors have been deposited at the Protein Data Bank with accession code 4F3V.

Sequence alignments were done using ClustalW2 (http://www.clustal.org) and rendered using the ESPript server (http://espript.ibcp.fr). Structural illustrations were prepared using PyMol (http://www.pymol.org).

RESULTS AND DISCUSSION

The three dimensional structure of the N-terminal domain (NTD) of M. tuberculosis EccA₁, residues 1-280, was determined to 2.0 Å resolution. The protein crystallized in space group P2₁2₁2₁ with two molecules in the asymmetric unit (Table I). Crystallographic phases were experimentally determined by single wavelength anomalous diffraction method using apotassium dicyanoaurate derivative. The final model includes residues 1-273 of EccA₁ in both molecules. The two monomers adopt highly similar structures with an r.m.s.d. of 0.52 Å between the two chains. In the asymmetric unit the two molecules of EccA₁ dimerize via a non-crystallographic two-fold axis through surfaces on the edge of TPR motifs 2 and 3. This dimerization interface probably represents a non-physiological interaction. Analysis using the PISA server¹⁴ shows that the interface between the two chains buries 2200 Ų of surface area. However, a calculated PISA score of 0.110 for the interface indicates that it probably does not represent a physiological interaction. Furthermore, Arg62 and Arg63 from each chain form bridging interactions mediated by a sulfate ion from crystallization solution. Because the two chains are nearly identical and probably do not dimerize in the context of full-length EccA₁ we limit our discussion to a monomer of EccA₁.

Table I.

Data collection and refinement statistics.

	Native (PDB 4F3V)	KAu(CN)2 derivative
Data collection
Wavelength (Å)	1.0000	1.0000
Space group	P2l2121	P2l2121
Cell dimensions
a, b, c (Å)	73.23, 92.51, 105.71	74.32, 92.58, 105.76
α, β, γ (°)	90, 90, 90	90, 90, 90
Resolution (Å)	29.6–2.00 (2.11–2.00)a	29.6–2.49 (2.63–2.49)
R sym	0.101 (0.782)	0.122 (0.508)
I / σI	14.2 (2.5)	13.9 (4.4)
Completeness (%)	99.8 (99.3)	99.4 (96.8)
Multiplicity	5.7 (5.7)	7.3 (7.2)
Anomalous completeness (%)		99.3 (96.1)
Anomalous multiplicity		3.9 (3.8)
Refinement
Resolution (Å)	29.6–2.00
No. reflections (total / free)	49226 / 2525
Rwork / Rfree	0.174 / 0.211
No. atoms
Protein	4135
Ligand/ion	33
Water	453
B-factors
Protein	26.2
Ligand/ion	56.4
Water	36.3
Wilson B	30.8
R.m.s. deviations
Bond lengths (Å)	0.011
Bond angles (°)	1.286
Ramachandran distribution (%)b
Favored	98.5
Outliers	0.0

^aValues in parentheses are for the highest-resolution shell.

^bCalculated using the MolProbity server (http://molprobity.biochem.duke.edu).

The 12 anti-parallel α helices of EccA₁ arrange into 6 tandem TPR motifs of approximately 34 residues each [Fig. 1]. The TPRs associate through hydrophobic interactions between consecutive helices and together form a right-handed superhelix with a pitch of ~60 Å, and a width of ~40 Å that is characteristic of TPR domains¹⁵. A DALI search ranked the Pseudomonas aeruginosa type IV pili system protein PilF as the closest characterized bacterial homolog of EccA₁ with an r.m.s.d of 3.5 Å and 9% sequence identity over 173 aligned residues (PDB 2HO1).^16,17 Despite low homology between two proteins, the TPR helices of EccA₁ superpose with PilF remarkably well, however, unlike PilF, EccA₁ has a β finger insertion after TPR2 that occupies the concave groove of the TPR superhelix [Fig. 1]. This is the same groove commonly used for protein-protein interactions by the TPR motif proteins.¹⁸ PilF contains a conserved Asn ladder within its binding groove that is thought to promote peptide binding through formation of bidentate hydrogen bonds to substrate backbone.^16,19 However, the Asn ladder is missing in EccA₁, and in its place are hydrophobic residues that form a hydrophobic core with β finger insert residues. This hydrophobic core suggests that the insert is a permanent feature within the putative binding groove. It may be that the insert participates in protein-protein interactions by forming β strand complementation interactions with the substrates in an extended conformation within the concave groove. Alternatively, proteins may interact with the side of the TPR bundle instead of the canonical binding groove. This latter binding mechanism by a TPR protein is illustrated by the p67^phox - Rac complex structure (PDB 1E96), which contains a similar β finger motif in its concave groove.²⁰

EccA₁ belongs to the CbxX/CfxQ family of ATPases, however, the homology is limited to the AAA+ domain and the N-terminal TRP domain is unique for EccA family. The nearest homolog of EccA₁ with a known structure is Rubisco activase (PDB 3SYL and 3ZUH),²¹ which is 46% identical to the C-terminal ATPase domain of EccA₁. Similar to Rubisco activase, EccA₁ may adopt a hexameric ring stabilized by contacts between the C-terminal domains [Fig. 2(B)].² Modeling the architecture of full-length EccA₁ using the structure of Rubisco activase (PDB 3ZUH) to align the EccA₁ monomers into a hexamer yielded insight into the possible conformations of the N- and C- terminal domains relative to each other. There is no obvious region of the C-terminal ATPase domain that might fit into the large groove of the N-terminal domain structure. This groove probably remains open and available for interactions with other proteins and/or substrates of the T7SS. Also, the length of the N-terminal TPR domain (~70 Å) is slightly shorter than the radius of the Rubisco activase hexamer. Thus, it is possible that the N-terminal domain lies across the top of the ATPase domain without occluding the central pore. Importantly, this hexameric conformation could be modeled without occluding any of the ATP binding sites indicating that such a configuration is compatible with ATPase activity [Fig. 2(A)].

Figure 2.

Model of EccA₁ hexamer. (A) View looking down the central pore of full length EccA₁ modelled using the hexameric model of Rubisco activase (3ZUH). C-terminal AAA+ ATPase domains are colored blue. N-terminal domain TPR motifs are colored alternating red / orange. Bound ADP is shown as spheres. (B) Same as (A) after rotating the view 90°.

The sequence alignment of EccA₁ homologs from the ESX-1 clusters of mycobacteria shows that EccA₁ proteins are conserved with at least 74% pairwise sequence identity between family members [Fig. S1]. In contrast, the sequence identity ranges between 29–39% for EccA proteins from different ESX clusters: ESX-1, ESX-2, ESX-3 and ESX-5. This may indicate that EccA ATPases interact with diverse components of secretion system or substrates. However, the overall protein architecture with an N-terminal TPR domain and a C-terminal AAA+ domain is clearly present in all EccA proteins. The key elements of the AAA+ proteins including Walker A motif, Walker B motif, pore loop are conserved in EccA₁ family members [Fig. S1]. In addition, Tyr439 (M. tuberculosis EccA₁ numbering) may serve as sensor 1, whereas Arg519 and Arg522 may serve as sensor 2. Arg429 has been suggested as an Arg finger residue and Arg429Ala substitution affected ATP hydrolysis by the C-terminal domain of EccA₁.² Our hexamer model identified Arg456 as another candidate Arg finger residue. Indeed, Arg456 is located closer than Arg429 to the ATP-binding site of adjacent subunit.

In summary, the EccA₁ N-terminal domain adopts a TPR fold indicating it may mediate protein-protein interactions between the C-terminal ATPase domain and substrate proteins. Like other TPR motif proteins, EccA₁ may use its central concave groove to interact with protein cargos, but the presence of a β finger fixed within the groove raises questions about the exact nature of the interaction. Oligomerization interactions were not found in the structure indicating that it is the C-terminal domain that primarily mediates EccA₁ hexamerization. Finally, the dimensions of the N-terminal TPR domain suggest that along with the C-terminal domain it form a compact hexamer while maintaining a central pore that can open and close during ATP hydrolysis. What the mechanism may be for specific protein recognition and energy transfer to substrates passing through the T7SS remains to be investigated.