Fusion crystallization reveals the behavior of both the 1TEL crystallization chaperone and the TNK1 UBA domain

Abstract

Human thirty-eight-negative kinase-1 (TNK1) is implicated in cancer progression. The TNK1 ubiquitin-associated (UBA) domain binds polyubiquitin and plays a regulatory role in TNK1 activity and stability. No experimentally determined molecular structure of this unusual UBA domain is available. We fused the UBA domain to the 1TEL variant of the translocation ETS leukemia protein sterile alpha motif (TELSAM) crystallization chaperone and obtained crystals diffracting as far as 1.53 Å. GG and GSGG linkers allowed the UBA to reproducibly find a productive binding mode against its host 1TEL polymer and crystallize at protein concentrations as low as 0.2 mg/mL. Our studies support a mechanism of 1TEL fusion crystallization and show that 1TEL fusion crystals require fewer crystal contacts than traditional protein crystals. Modeling and experimental validation suggest the UBA domain may be selective for both the length and linkages of polyubiquitin chains.

Keywords: 1TEL; AlphaFold2-Multimer; ETV6; SAM domain; TELSAM; TNK1; UBA; X-Ray Crystallography; crystallization chaperone; polyubiquitin binding.

Figure 1:. Design, crystallization, and structure of 1TEL-GG-UBA.

A. Design model of 1TEL-GG-UBA. B. Representative crystals of 1TEL-GG-UBA at 15mg/mL. Scale bar is 100 μm. C. 1TEL-GG-UBA molecular replacement solution showing the backbones of the 1TEL domain (magenta), the UBA central α-helix (cyan), the eventual position of the UBA main chain (orange ribbon), and electron density corresponding to the initial molecular replacement solution (slate mesh, contoured to 3σ). D. Superposition of the homology model (yellow) onto the UBA domain (cyan) from the 1TEL-GG-UBA crystal structure. E. View of the 1TEL-GG-UBA crystal lattice parallel to the helical axis of the 1TEL polymers, colored as in A. F. View of two 1TEL-GG-UBA polymers perpendicular to their helical axes, in space filling representation and omitting the UBA domains for clarity. G. Superposition of the 1TEL-GG-UBA crystal structure (magenta and cyan) onto the design model (white), aligned through only the 1TEL domains. H-I. Two views, 180° apart, of the binding interface between the UBA domain (cyan) and the 1TEL polymer (magenta). Hydrogen bonds are shown as black lines while atoms making van der Waals contacts are shown as spheres. J. Superposition of the 1TEL domains of representative 1TEL wwPDB structures (slate, PDB IDs 8FT8, 8FT6) onto structures of 1TEL-GG-UBA (magenta and cyan). The two 1TEL domains from 1TEL-GG-UBA are shown (magenta), while only a single UBA domain is shown (cyan). Selected amino acid side chains are shown as sticks.

Figure 2:. 1TEL fusion allows crystallization at low protein concentrations.

A. Representative crystals of 1TEL-GG-UBA at 2 mg/mL. Scale bar is 100 μm. B. Representative crystals of 1TEL-GG-UBA at 0.5 mg/mL. Scale bar is 100 μm. C. pH ranges at which crystals appeared for various input protein concentrations. D. Heat map of the number of protein crystals observed per well as a function of Mg-formate concentration. The number of crystals observed at different pH values have been averaged. E. As in D. but with only the averages of three concentrations ranges shown for clarity. F. Model of a magnesium ion interacting with the E112 at the hydrophobic inter-subunit interface of the 1TEL polymer. Two neighboring 1TEL monomers (cyan and slate) are shown as they would appear in a polymer. The V112E substitution is highlighted in pink while the putative magnesium ion is represented by a green sphere. G. Heat map of the maximum crystal size (μm, longest axis of the largest crystal) observed per well as a function of Mg-formate concentration. The maximum is the largest crystal observed at that Mg-formate concentration across all pH values tested. H. As in G. but with only the averages of the ranges of shown for clarity. I. Representative crystals of 1TEL-GSGG-UBA. Scale bar is 100 μm.

Figure 3:. 1TEL-UBA structures have nearly identical UBA binding modes against the 1TEL polymers.

A. 15 mg/mL (slate), 2 mg/mL (cyan), and 0.5 mg/mL (olive green) 1TEL-GG-UBA structures superimposed onto the 1TEL-GSGG-UBA structure (salmon), via the 1TEL domain (magenta). B. As in A. but zoomed in on the UBA:1TEL interface. C. Schematic of crystal contacts between a UBA domain (cyan), six 1TEL subunits (yellow, slate, and magenta) and four other UBA domains (wheat, orange, and green). Domains contacting the given UBA domain are shown as cylinders and colored, while other 1TEL subunits are shown as a gray ribbon. D. Crystal packing differences between the four 1TEL-UBA structures, focusing on F123. The structures are colored thus: 15 mg/mL UBA (slate), 15 mg/mL 1TEL (dark purple), 2 mg/mL UBA (cyan), 2 mg/mL 1TEL (blue-green), 0.5 mg/mL UBA (olive green), 0.5 mg/mL 1TEL (brown), GSGG UBA (salmon), and GSGG 1TEL (firebrick). All structures have been aligned through the UBA domain that hosts the F123 shown. This same UBA domain appears in the lower portion of panels D, E, and F (SM = symmetry mate). E. As in D. but focusing on W134. F. As in D. but focusing on the C-terminus of the UBA. See also Figure S1.

Figure 4:. UBA-alone crystal structure and B-factor comparison.

A. Representative crystals of the UBA domain crystallized on its own. The scale bar is 100 μm. B. Asymmetric unit of the UBA-alone crystal structure, with two UBA chains. Amino acid sidechains are shown as sticks. C. Superposition of the two chains from the UBA-alone structure (white) with the UBA domains from the four 1TEL–UBA fusion structures (15 mg/mL–slate, 2 mg/mL–cyan, and 0.5 mg/mL–olive green) and the AlphaFold2 prediction–yellow. D. B-factors of a single chain within the UBA-alone crystal lattice, range 12.7–79.7 Ų. E. B-factors of a single chain within the 2 mg/mL 1TEL-GG-UBA crystal lattice range 27.1–86.5 Ų. F. Three 1TEL-GG-UBA polymers from the 1TEL-flex-UBA 2mg/mL crystal lattice, viewed along their helical axes and colored according to the refined B-factors. The color scale is the same in panels D-F.

Figure 5:. Aligned SEC runs of 1TEL-GG-UBA and ubiquitin.

A. mono-ubiquitin, B. M1-di-ubiquitin, C. M1-tri-ubiquitin and D. M1-tetra-ubiquitin with 1TEL-GG-UBA-alone (green), ubiquitin-alone (red) and combined (blue).

Figure 6:. Predicted binding modes and corresponding experimental data.

A. Western blot of TNK1 UBA mutants incubated with or without K48- or K63-tetra-ubiquitin. B. Predicted ubiquitin binding sites supported by in vitro pull-down data. C-D. Schematic of potential UBA:tetra-ubiquitin complexes. E. Western blot of TNK1 UBA incubated with K63-linked-di-, tri-, or tetra-ubiquitin. F. Gel-quantification ratios between the GST-UBA and bound K63-ubiquitin. G. As in E, but with mono- or M1-linked-di-, tri-, or tetra-ubiquitin. H. As in in F, but with mono- or M1-linked-ubiquitin. See also Figures S2, S3, and S4.