Expression and Purification of Proteins

NATURE m/s 2006-01-00642A

SUPPLEMENTARY INFORMATION

METHODS

Expression and Purification of Proteins

A full-length construct of yeast Hsp90 harbouring the mutation A107N and with residues 221-255 in the charged linker region deleted and replaced by LQHMASVD; an N-terminal and middle-segment (M-C) construct of yeast Hsp90 (residues 273-709); and full-length yeast p23 (Sba1) were inserted with addition of N-terminal His6-tag and PreScission protease cleavage site into pRSETA, and expressed in E.coli BL21(DE3) pLysS. For Hsp90 constructs, expressed protein was purified by metal affinity chromatography on Talon resin, ion-exchange on Mono-Q, and size exclusion chromatography on a Superdex S200 HR column equilibrated in 20mM Tris (pH8.0), 10% glycerol, 200mM NaCl, 1mM EDTA. The His6-tag was cleaved with PreScission protease (Stratagene), which was subsequently removed by passing protein mixture through a GSTrap, HisTrap and Superdex S200 HR columns. The resulting highly purified and homogenous protein was concentrated using 10K Vivaspin concentrators. Sba1 was purified by metal affinity on talon resin, ion-exchange using Source-Q, and size exclusion chromatography on a Superdex S75HR column.

Crystallisation, Data Collection and Structure Determination

Crystals of the Hsp90-p23/Sba1 complex were grown at 16ºC in hanging-drop vapour diffusion experiments from a mixture of Hsp90, p23/Sba1 and Mg2+-AMP-PNP at a final concentration of 0.2 : 0.2 :1 mM respectively, with a precipitant solution containing 18-20% polyethylene glycol 4000, 10% glycerol, 10% iso-propanol and 100 mM Hepes (pH 7.5). Initial crystals only diffracted to 8Å but after extensive optimization of every step, diffraction was improved to beyond 3.5Å. Crystals are thin rods, of typical dimensions being 40 x 40 x 300 µm. Crystals have space group P41212 with unit cell dimensions a =126.9Å, c = 279.9Å. Analysis of the Matthews coefficient suggests a solvent content 56%, with one dimer of Hsp90 and two Sba1 molecules present within the asymmetric unit with a combined weight of ~220 000 daltons. Crystals of the M-C construct were grown at 14ºC in hanging-drop vapour diffusion experiments with protein at 7.5 – 15 mg/ml, with a precipitant solution containing 1.28-1.45 M tri-sodium citrate, 100 mM Hepes (pH 8.0), and transferred to 1.7M tri-sodium citrate. M-C crystals have space group P43212 with unit cell dimensions a = 105.6Å, c = 289.3Å, and three molecules in the asymmetric unit.

Crystals of the M-C construct were flash-cooled in liquid nitrogen and data collected on station ID29 at ESRF Grenoble, to 3.0Å. All data were processed and reduced in MOSFLM1and SCALA2. The structure was determined by molecular replacement using Phaser2, with the structure of yeast Hsp90 middle segment (PDB code 1HK7) as the search model. The three non-crystallographic symmetry related middle segments were initially refined as rigid bodies, and then subjected to simulated annealingusing CNS3. The structure of the C-terminal domain was then built manually in O4using clear difference Fourier maps and refined using Refmac2with tight three-fold non-crystallographic symmetry, and with several cycles of manual intervention. TLS refinement of protein domains was introduced in the latter stages.

Crystals of the Hsp90-p23/Sba1 complex were cryo-protected with the addition of 30% glycerol and data collected on station ID29 at ESRF Grenoble. Diffraction from these crystals is very anisotropic, with reflections at 3.0Å along one axis, but only to 3.5Å in other directions. Data was processed as above and the structure solved by molecular replacement using the N-terminal domain and middle segments of yeast Hsp90 (PDB codes 1AMW, 1HK7), the C-terminal domain derived from the M-C crystals, and a homology model of Sba1 based on the crystal structure of human p23. Individual domains were initially refined as rigid bodies, and then subjected to simulated annealingusing CNS3. Difference Fourier maps unambiguously identified changes relative to the isolated domains, and the structure was rebuilt manually using O4 and COOT5. The rebuilt structure was then refined with tight two-fold non-crystallographic symmetry restraints relating equivalent domains using Refmac2, with cycles of manual intervention. TLS refinement of protein domains was introduced in the latter stages. More than 98% of the structure’s 1464 residues have backbone conformations in the favoured and allowed regions in the Ramachandran plot - remaining residues with disfavoured conformations are in poorly ordered segments of the structure : loop-tips of Sba1 not involved in interaction with Hsp90, the visible segment of the charged linker connecting the N-domain and middle segment of Hsp90, the linker between the Hsp90 C-domain and middle segment that undergoes an order-disorder transition in going from the unconstrained (M-C) to ATP-bound conformations, and the projecting amphipathic loop that is disordered in one Hsp90 protomer, or have backbone conformations determined by interaction with the nucleotide, e.g. Phe120. Statistics for data collection and refinement are given in Table 1. Coordinates and structure factors have been deposited in the PDB with accession codes 2CGE and 2CG9.

‘Omit’ maps (FIGURE 3) for the catalytic loop and bound AMP-PNP were generated from weighted Fo-Fc difference Fourier maps following maximum likelihood refinement of the structure, with residues 370-390 or the ATP-analogue omitted from refinement and structure factor, and contoured at 1.75. The difference density clearly shows the ordered conformation of the catalytic loop and position of the side-chain of Arg380, which is able to make a polar interaction with the -phosphate of the bound ATP analogue. The omit density for the ATP-analogue clearly shows the position of the -phosphate and is consistent with the presence of a hydrated Mg2+ bridging the  and  phosphates, but this cannot be reliably modelled at this resolution.

1.Leslie, A. G. W. MOSFLM Users Guide (MRC Laboratory of Molecular Biology, Cambridge, U.K., 1995).

2.CCP4. Programs for protein crystallography. Acta Crystallographica D50, 760-763 (1994).

3.Brunger, A. T. et al. Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallographica D54, 905-921 (1998).

4.Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallographica A47, 110-119 (1991).

5.Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr 60, 2126-32 (2004).