SUPPLEMENTARY INFORMATION

Size: px

Start display at page:

Download "SUPPLEMENTARY INFORMATION"

Carmel Cox
5 years ago
Views:

1 DOI: /NCHEM.1290 Metal-directed, chemically tunable assembly of one-, two- and threedimensional crystalline protein arrays ,3 Jeffrey D. Brodin 1, X. I. Ambroggio 2, Chunyan Tang 1, Kristin N. Parent 1, Timothy S. Baker 1,3 and F. Akif Tezcan 1 1 Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, CA Rosetta Design Group LLC, Fairfax, VA Division of Biological Sciences, University of California, San Diego, La Jolla, CA Computational interface redesign. RIDC3 was designed starting from the Zn 4 :MBPC1 4 platform as previously described 1, with the following modifications. During rotamer optimizations, the standard Lennard Jones van der Waals repulsive term of RosettaDesign did not include any non-canonical Dunbrack rotamers; and the final design was selected from redesigns of multiple models generated through symmetric rigid body perturbations of the monomers of the MBPC1 crystal structures in a procedure described in detail below. Building off of previously reported design methodologies 2, 3, we sought to expand the search space for the redesigns and compensate for explicit energy-based modeling of rotational, translational, and backbone degrees of freedom by generating multiple starting models through plausible rigid body rotation trajectories and selecting the redesigns with the most favorable properties. For the rigid body rotations, two monomers of MBPC1 making up the desired C2- dimer geometry (initial dimer geometry was based on the symmetric halves of Zn 4 :MBPC1 4 ) were rotated symmetrically inwards or outwards from each other around the Zn-Zn axis. In the first stage, starting models were generated from rotations of one degree increments from 10 degrees inwards to 10 degrees outwards using PDBCUR 4 and the interface residues of the models were redesigned using RosettaDesign as described above. In the second stage, redesigns were performed for models generated by 0.1 degree increment rotations throughout the rotation NATURE CHEMISTRY 1

2 range (2-4 degrees inwards) which resulted in first stage redesigns with computed energies below zero and SASApack 5 values below 0.8. The effect of each individual amino acid substitution of the redesign, with the lowest SASApack value, on the predicted ΔΔG of binding, total energy, and SASApack score was evaluated, and those substitutions which did not significantly affect the scores were reverted to the starting amino acids. Through this design procedure, 10 surface mutations (K27E/D28K/T31E/R34L/L38A/Q41L/H59R/D66A/V69M/ L76A) to MBPC1 were predicted for the construct we named RIDC3. Docking of the crystallographic model into the reconstructed RIDC3 nanotube volume. Docking of dimers of C2-dimers into the reconstructed density map and production of figures was performed using Chimera 6. The dimer of C2-dimers was chosen as a model for docking studies because it was observed in both Type-1 and Type-2 crystal structures and is formed by pairwise, Zn1-mediated interactions that are firmly anchored on rigid helices and should therefore be less flexible than Zn2 and Zn3-mediated interactions, which involve loops. In addition to the orientation depicted in Figure 5, where there is a two-fold screw axis along the length of the subunit, an arrangement with all tetramers facing the same end of the tube was also tested. In this model, every other tetramer exhibited a relatively poor fit, verifying the alternating arrangement of tetrameric subunits. Supplementary References 1. Salgado, E. N., Ambroggio, X. I., Brodin, J. D., Lewis, R. A., Kuhlman, B. & Tezcan, F. A. Metal-Templated Design of Protein Interfaces. Proc. Natl. Acad. Sci.USA 107, (2010). 2. Kortemme, T., Joachimiak, L. A., Bullock, A. N., Schuler, A. D., Stoddard, B. L. & Baker, D. Computational redesign of protein-protein interaction specificity. Nat. Str. Mol. Biol. 11, (2004). NATURE CHEMISTRY 2

3 3. Joachimiak, L. A., Kortemme, T., Stoddard, B. L. & Baker, D. Computational Design of a New Hydrogen Bond Network and at Least a 300-fold Specificity Switch at a Protein- Protein Interface. J. Mol. Biol. 361, (2006). 4. Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta. Cryst. D. 67, (2011). 5. Leaver-Fay, A., Butterfoss, G. L., Snoeyink, J. & Kuhlman, B. Maintaining solvent accessible surface area under rotamer substitution for protein design. J. Comp. Chem. 28, (2007). 6. Ferrin, T. E., Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M. & Meng, E. C. UCSF chimera - A visualization system for exploratory research and analysis. J. Comput. Chem. 25, (2004). NATURE CHEMISTRY 3

4 Supplementary Tables Supplementary Table S1. Crystallization, X-ray data collection and refinement information and statistics for the PEG-precipitated (Type 1) and Zn-directed (Type 2) RIDC3 crystals. * denotes highest resolution shell. Crystallization Conditions Precipitant Solution Type 1 Crystals (PEG-Precipitated) 20 % PEG 3350, 2.1 mm ZnCl 2, 0.2 M CaCl 2, and 100 mm Bis-Tris (ph 6.5) [RIDC3] (mm) V prot :V precipitant (µl) 2:1 2:1 Data Collection Location SSRL BL 7-1 SSRL BL 9-2 Unit Cell Dimensions (Å) α=β=γ=90 Space Group P C2 Resolution (Å) X-ray wavelength (Å) Number unique reflections Redundancy Completeness (%)* 99.0 (99.8) 94.3 (92.8) <I/σI>* 7.7 (2.2) 4.7 (1.5) R symm (%)* 7.0 (34.9) 10.8 (48.9) R work /R free (%) 19.0/ /28.6 B-factors (Å 2 ) Protein Ligands/ions Water R.m.s. deviations Bond lengths (Å) Bond angles ( ) Ramachandran plot (%) Most favored Allowed Generously allowed Disallowed Type 2 Crystals (Zn-directed) 8 mm ZnCl 2 and 200 mm Bis- Tris (ph 6.0) α=γ=90, β=112.6 NATURE CHEMISTRY 4

5 Supplementary Table S2. Cryo-EM data collection and image reconstruction statistics. Cyclic symmetry (Cn) 9 Pixel size (Å) Objective lens defocus range (µm) Total number of micrographs recorded 139 Total number of boxed helices 229 Number of tubes used in the reconstruction 5 Number of segments 1780 Segment length (pixels) 200 Shift between segments (pixels) 4 Padded segment size (pixels) 350 Range of initial ΔΖ (Å) Refined value of ΔΖ (Å) 25.8 Range of initial Δφ (º) Refined value of Δφ (º) 5.3 Smallest diameter (Å) 520 Largest diameter (Å) 635 NATURE CHEMISTRY 5

6 Supplementary Figures Supplementary Figure S1. The derivation of the C2-dimer geometry from the Zn 4 :MBPC1 4 architecture. MBPC1 is a derivative of cytochrome cb 562, decorated with two metal-chelating motifs (H59/H63 and H73/H77) on its surface. Upon equimolar Zn 2+ addition, MBPC1 assembles into a D 2 symmetric tetramer, Zn 4 :MBPC1 4, which is held together by four equivalent Zn ions coordinated to H73/H77 from one monomer, H63 from a second, and D74 from a fourth. Because of its D 2 symmetry, Zn 4 :MBPC1 4 contains three C 2 symmetric interfaces (i1, i2, i3), of which only i1, i2 are shown above. This means that Zn 4 :MBPC1 4 can be halved in three different orientations to obtains three alternative sets of C 2 symmetric dimers. The particular C2-dimer geometry that constitutes the focus of this study (and is stabilized by surface mutations to produce RIDC3) is obtained by halving Zn 4 :MBPC1 4 along i2, which yields two equivalent, three-coordinate Zn coordination sites formed by H73/H77 from one monomer and H63 from the second. The coordination vectors originating from these coordination sites are depicted as red and blue arrows. NATURE CHEMISTRY 6

7 Supplementary Figure S2. Sedimentation coefficient distributions for RIDC3 as determined by analytical ultracentrifugation. Samples were prepared at 5 µm (light blue) or 600 µm (blue) RIDC3 in the presence of equimolar Zn or 600 µm (red) RIDC3 in the presence of 5 mm EDTA. We attribute the tetrameric species populated at 600 µm RIDC µm Zn to the Zn1-linked dimer of C2-dimers observed in both the PEG-precipitated (Fig.1) and Zn-directed (Fig.4) crystals. NATURE CHEMISTRY 7

8 Supplementary Figure S3. Superposition of Rosetta-predicted (magenta) and crystallographically-determined (grey) Zn2:RIDC32 (C2-dimer) structures. NATURE CHEMISTRY 8

Supplementary Figure S4. Time course of Zn-directed RIDC3 self-assembly monitored by TEM. For sample preparation, 3 µl aliquots of a solution containing 100 µm RIDC3 and 300 µm Zn at ph=5.

9 Supplementary Figure S4. Time course of Zn-directed RIDC3 self-assembly monitored by TEM. For sample preparation, 3 µl aliquots of a solution containing 100 µm RIDC3 and 300 µm Zn at ph=5.5 were pipetted onto carbon-coated Cu grids and stained with uranyl acetate. These images reveal mostly small, disordered aggregates until Day 2, after which large crystals (shown with arrows) begin to appear. NATURE CHEMISTRY 9

10 Supplementary Figure S5. Width distributions of negatively stained RIDC3 nanotubes. a, Solutions containing 450 µm RIDC3 and 4.5 mm Zn were deposited on carbon-coated Cu grids and stained with uranyl acetate. After imaging at a nominal magnification of 25,000x, 125 independent width measurements were taken manually using Image J, binned into 4 nm ranges and the frequency of each range was plotted. b, A solution of 100 µm RIDC 3 and 300 µm Zn was imaged and analyzed identically to (a). NATURE CHEMISTRY 10

Supplementary Figure S6. Zn-induced RIDC3 self-assembly in solution characterized by light microscopy and TEM after negative staining. For all images, [RIDC3]=450

11 Supplementary Figure S6. Zn-induced RIDC3 self-assembly in solution characterized by light microscopy and TEM after negative staining. For all images, [RIDC3]=450 µm and ph=5.5. At a 1:1 Zn:protein ratio (first column), only macroscopic crystals are observed and were characterized by light microscopy with (top) and without (bottom) polarizers. For all other cases, the top and the middle rows show low and the high magnification TEM images, respectively, with the Fourier transforms of the latter given in the bottom row. The overlapping lattice axes in the tubular structures are illustrated as black and red arrows. NATURE CHEMISTRY 11

12 Supplementary Figure S7. Light micrographs of Zn-directed, 3D RIDC3 crystalline arrays used for X-ray diffraction studies. NATURE CHEMISTRY 12

Supplementary Figure S8. Superposition of Zn1-linked dimers of C2-dimers onto a 2D projection map of negatively-stained, monolayered RIDC3 sheets obtained at ph 8.5.

13 Supplementary Figure S8. Superposition of Zn1-linked dimers of C2-dimers onto a 2D projection map of negatively-stained, monolayered RIDC3 sheets obtained at ph 8.5. The opposing rows composed of two-fold symmetrical, bilobed densities can be well modeled with Zn1-linked dimers of c2-dimers (Zn4:RIDC34) present both in Type 1 (PEGprecipitated) and Type 2 (Zn-directed) crystals. NATURE CHEMISTRY 13

14 Supplementary Figure S9. Cross-section normal to the RIDC3 nanotube axis. A planar section through the reconstructed volume of an RIDC3 nanotube was generated using Chimera and shows 36 discrete densities, each consistent with a C2-dimer. Two pairs of C2-dimers compose a single subunit in the helical reconstruction. NATURE CHEMISTRY 14

15 Supplementary Figure S10. Comparison of metal-linked interfaces in nanotubes and 2D sheets from the X-ray crystallographic structure. Comparison of Zn2 (a) and Zn3 (b) interfaces in nanotubes (top, colored in cyan and magenta) with those in 2D sheets (bottom, colored in shades of grey), highlighting the curvature of RIDC3 nanotubes and the resulting changes in intersubunit distances based on D21 residues. NATURE CHEMISTRY 15

Supplementary Figure S11. Comparison of the crystallographically observed Zn2 interface (gray backbone; orange coordination sphere; see Fig.

16 Supplementary Figure S11. Comparison of the crystallographically observed Zn2 interface (gray backbone; orange coordination sphere; see Fig. 3b) with that based on docking into the reconstructed nanotubes (cyan). Two dimers of Zn1-linked c2-dimers from each structural model were aligned using Pymol (RMSD = 1.63 Å over 848 Ca s) and show little change in their Zn2 coordination modes. NATURE CHEMISTRY 16

17 Supplementary Figure S12. Comparison of the crystallographically observed Zn3 interface (gray backbone, red coordination sphere; see Fig. 3b) with that based on docking into the reconstructed nanotubes (cyan and magenta). Two Zn3-linked C2-dimers (cyan or magenta) were docked into the reconstructed density map, yielding a ridge comparable to that seen in the Type 2 crystal structure. Because of the asymmetry introduced by bending the lattice into a tube, the ridges pointing towards the outside (left) and inside (right) of the tube are different, as indicated by a structural alignment using Pymol (RMSD = 4.28 Å and 3.7 Å over 424 C! s for the exterior and interior ridges, respectively). The bending and twisting necessary for the curving of the 2D sheet into a tube result in a displacement of the residues involved in the primary coordination sphere (bottom left) and generation of a new potential metal binding site (bottom right). NATURE CHEMISTRY 17

18 Supplementary Figure S13. Time course for the maturation of rhodamine-c21ridc3 crystals at ph 8.5. a, TEM images of negatively stained samples at the indicated time points show the presence of crystalline arrays at Day 3 (arrows) and an increase in the number and relative proportion of arrays versus aggregate by day 10. b, The emergence of a CD signal from samples containing 50 µm rhodamine- C21 RIDC3 and 500 µm Zn coincides with the presence of crystalline arrays as determined by TEM. c, The UV-visible absorbance spectrum of the same sample shows a blue-shifted band immediately after the addition of Zn and a time-dependent increase in turbidity. NATURE CHEMISTRY 18

19 Supplementary Figure S14. Alternative views of vitrified, unstained RIDC3 nanotubes obtained at ph=5.5. The samples shown are frozen on Quantifoil grids rather than lacey carbon grids (which were used for image reconstruction) for a clearer view of collections of RIDC3 nanotubes. The grid circle shown in (a) is 2 µm in diameter. NATURE CHEMISTRY 19

$Supplementary Figure S15. Diffraction pattern of RIDC3 nanotubes.$

20 Supplementary Figure S15. Diffraction pattern of RIDC3 nanotubes. The incoherently averaged power spectrum from all images used in the reconstruction (left) compared with the Fourier transform of a 2D projection of the final reconstruction (right) shows matching layer line positions (see red circles at 1/48 Å -1 and 1/55 Å -1 ), providing evidence that the reconstruction is correct. The increase in resolution obtained during the reconstruction is apparent from a comparison of the high-resolution diffraction limits (arrows) of 2D projections before and after image reconstruction. NATURE CHEMISTRY 20

Designing Metal-Templated de novo Protein Complexes

Designing Metal-Templated de novo Protein Complexes Xavier (Javi) I. Ambroggio RosettaCon, August 4 th 2010 More Parlor Tricks with Metals Interact dammit. What is metal-templating? 1. Starting material: