Ensemble structure description of Lys63-linked diubiquitin☆

The data described herein are related to the article entitled “Lys63-linked ubiquitin chain adopts multiple conformational states for specific target recognition” [1], and to the coordinates for the ensemble structure of Lys63-linked diubiquitin (PDB code 2N2K). A Lys63-linked diubiquitin exists in three conformational states with different orientations for the two subunits, each responsible for binding to a target protein and encoding a specific cell signal. An atomic entry in the ensemble structure file consists multiple lines, representing alternative locations of the atom and recapitulating the dynamics of the protein. Experimental details about obtaining strictly intramolecular paramagnetic restraints and determining the relative occupancies of the conformational states are presented. The experimental design and procedures in this Data article can be useful for characterizing the structure and dynamics of other multi-domain proteins.

3. Data, experimental design, materials and methods

Experiment design
Paramagnetic NMR, in particular paramagnetic relaxation enhancement (PRE), can provide long-range distance information and allows the depiction of protein conformational fluctuations [2]. Paramagnetic NMR is mostly performed with the conjugation of an extrinsic paramagnetic probe. As the linker between protein backbone and the paramagnetic center involves multiple rotatable bonds, the exact position of the paramagnetic center is not fixed with respect to the protein. The PRE value is related to the inverse sixth power of the distance between the paramagnetic center and a protein nucleus, and therefore a small fluctuation in the position for the paramagnetic center can cause a large variation in the PRE value. The observed PRE value for a protein nucleus in a certain conformational state is a product of the PRE values and the relative populations [3]. Thus it is difficult to deconvolute the occupancy of each protein conformational state from the back-calculated PRE.
Lys63-linked diubiquitin(referred to as Ub 2 ) comprises two ubiquitins covalently linked via an isopeptide bond between Lys63 in one subunit and Gly76 in the other subunit, and is involved in many nondegradative signaling cascades in cell [4]. It has been found that Ub 2 can exist in both open and closed conformations in solution [1,5,6]. Using the PRE, we characterized the structures of the two alternative closed states of Ub 2 . To assess the relative occupancies for each conformational state, we integrated the data from small angle X-ray scattering (SAXS) and isothermal calorimetry (ITC). SAXS data define the overall silhouette of the protein system, but lack structural details. ITC measures the binding affinities between Ub 2 with its protein ligands. In a conformational selection mechanism, a point mutation (E64R in one subunit of Ub 2 ) that perturbs the relative populations of the preexisting conformational states would affect the binding affinities towards the respective ligands. As such, the changes in binding affinities allow the extrapolation of the relative populations of the preexisting Ub 2 conformational states.

Sample preparation
UbiquitinTAB2 and Ub 2 proteins were prepared according to the established protocol [7]. Specific ligands for Ub 2 , including tandem ubiquitin-interacting motif (tUIM), the fourth zinc-finger domain of A20 (A20 ZnF 4 ), and the NZF domain of TAB2 protein (TAB2 NZF), were purified with GST affinity column followed by the removal of the GST tag with TEV protease. Single-point mutations were introduced using the QuikChange method. To specifically introduce the paramagnetic probe at a desired site, a Ub 2 cysteine mutant (note that the mutations do not affect the enzyme-catalyzed formation of di-ubiquitin) was reacted with 5-times excess of S-(1-oxylÀ 2,2,5,5-tetramethyl-2,5dihydro-1H-pyrrol-3-yl methyl methanesulfonothioate (MTSL, from Toronto Research Chemicals, Canada) for two hours at room temperature. Excess probe was removed through desalting, and the completeness of the conjugation was confirmed by mass spectrometry. The sites for paramagnetic conjugation (N25C and K48C both in the distal subunit of Ub 2 , one at a time) and single-point mutation (E64R at the proximal subunit of Ub 2 ) are illustrated in Fig. 1.

Intramolecular PRE data
We used a two time-point NMR pulse scheme to measure the transverse relaxation rates for backbone amide protons [8], with a 4-ms delay between the two time points. In an Ub 2 protein sample, one subunit is unlabeled and paramagnetically tagged, while the other subunit is [U-15 N]labeled for PRE measurement. The observed PRE rate for a particular nucleus can arise both intramolecularly (between two subunits) or inter-molecularly (between two separate Ub 2 molecules). Hence we prepared a second sample comprising an equimolar mixture of Ub 2 proteins, with paramagnetic tagging and isotope labeling on different subunits of different Ub 2 molecules (Fig. 2). The PRE values measured for the [U-15 N]-labeled subunit in this mixed sample should only arise from inter-molecular, noncovalent interactions between different Ub 2 proteins [6], which was subtracted from the PRE measured for first sample for strictly intramolecular inter-subunit PRE (Table 1).

SAXS measurement
Small-angle X-ray scattering data were collected on a SAXSess mc 2 instrument (Anton Paar, Graz, Austria) equipped with a sealed-tube source and a CMOS diode array detector. The proteins were extensively dialyzed. To remove any high molecular-weight aggregates, Ub 2 was centrifuged at N25C K48C Proximal Distal E64 Fig. 1. Illustration of the mutations introduced to Lys63-liked diubiquitin. The proximal subunit of Ub 2 is shown as blue cartoon with transparent surface, and the distal subunit is shown as red cartoon. The MTSL paramagnetic probe conjugated at the engineered cysteine residue (one at a time) is shown as sticks, and the oxygen atoms carrying the unpaired electron are shown as red spheres. E64 in the proximal subunit where a charge reversal mutation (E64R) is introduced is also indicated. 15,000 rpm for 30 min prior, and the upper portion of the supernatant was carefully collected for SAXS measurement. To ensure the stability of the protein sample, SAXS data were collected every 30 min; the frames were compared, and were combined if there are no differences among them. As Ub 2 adopts multiple conformations in solution, the SAXS data could not be fitted to any single known structure. The SAXS is rather insensitive to the transient noncovalent interactions between Ub 2 proteins, and the data are the direct averaging (as opposed to or -6 4 averaging for PRE) of the scattering profiles for the conformational states in solution.

Binding affinities between Ub 2 and its ligands
The binding affinity between wild type Ub 2 (or the E64R mutant) and its cognate ligands (tUIM or TAB2 NZF) were measured on a VP-ITC instrument (GE Healthcare, Piscataway, NJ). The ligand protein (300 μM) was placed into a syringe and titrated into the reservoir containing wildtype or mutant Ub 2 proteins (20 mM). The titration for each pair of interactions was repeated four times. The binding affinity between Ub 2 and A20 ZnF 4 was too weak to be assessed using ITC. Therefore   A charge reversal mutation away from the binding interface to its partner protein was introduced at the interface between two ubiquitin subunits (Fig. 1). The E64R mutation destabilizes the preexisting closed-state conformations of Ub 2 and promotes the open-state conformations. As a result, the binding affinity towards the open-state ligand (tUIM) increases, at the cost of decreasing binding affinities towards the closed-state ligands A20 ZnF 4 and TAB2 NZF ( Table 2; the binding affinities between Lys63-linked diubiquitin and A20 ZnF4 were obtained by fitting NMR chemical shift perturbations). Importantly, the changes in binding affinities are entropic, meaning that the mutation leaves the interaction between Ub 2 and its ligands unperturbed and likely modulates the relative populations of the preexisting conformational states.

Ensemble refinement of Ub 2 structure
We refined the Ub 2 ensemble structure against intramolecular inter-subunit PRE restraints using Xplor-NIH [9]. The paramagnetic probe was represented as three-conformer ensemble and was The residue is broadened out beyond detection in the paramagnetic spectrum and gives a very large PRE value, with the lower limit at 120 s À 1 .
optimized. We fixed one subunit of Ub 2 (the distal unit with paramagnetic probes conjugated), and allowed the other subunit (the proximal unit) to move as a rigid body, with torsional freedom given to the covalent linker between two subunits. The intramolecular PRE could not be fully accounted for with a single conformation for the Ub 2 protein. Therefore we invoked ensemble representation, and we systematically incremented the number of conformers representing the closed state. The intersubunit PREs could only be accounted for with a four-conformer representation for the closed state. However, the exact population of closed state was less certain, as the agreement between the observed and calculated PREs are all reasonably good at closed-state population 4 30%. Owing to the steep distance dependence of PRE, a small movement in the relative position of the two Ub units can compensate for the offset in population. At population o30%, however, the van der Waals repulsive term builds up, which would prevent further compensation and result in an disagreement between observed and back-calculated PREs. The determination of the exact population for the closed state was based on several sources of evidences. (1) It has been shown previously that ubiquitin monomer noncovalently interacts with each other at an apparent K D value of $ 5 mM. When covalently linked with an isopeptide linkage, the effective concentration would drive $70-80% of Ub 2 to the closed state. (2) The SAXS profile is a direct average of the scattering profiles for the constituting conformers. In this case, it appears closer to the profile of the close state than to the profile of the open state. However, due to limited resolution, SAXS cannot tell whether multiple closed states exist or how the two Ub units interact with each other in the closed states. (3) A minimum of four-conformer ensemble is required to account for the intramolecular PREs. In the ensemble structure, three conformers are similar to each other, and are different from the other conformer. The four conformers could be clustered to two states (C1 and C2), and the number of conformers in each cluster represents the relative population and also serves as the scaling factor for the PRE. (4) The PRE profile for the E64R mutant is similar to that of the wildtype Ub 2 protein, yet the magnitude is halved. This means the mutation decreased the closedstate population without perturbing the closed-state conformations. (5) The mutant proteins has decreased binding affinities towards the respective ligands for C1 and C2 closed states, and increased binding affinity towards the ligand for the open state. As the changes in binding affinities and binding free energies arise entropically (Table 2), the differences in conformational free energies, i.e. the changes in the relative populations of C1, C2 and open states, should be solely responsible for the changes in binding free energies. Using all the information above, we could determine the closedstate population at $ 70%, and the open-state population at $ 30%.

Description of the Ub 2 ensemble structure
A concatenated file containing 70 refined structures was deposited at the PDB. The 70 structures in the PDB file are calculated with various starting coordinates, and are slightly different owing to structural convergence problem. With the distal Ub unit fixed, each structure comprises four conformers, representing alternative locations of the proximal Ub unit. As the Ub 2 cannot be in all four locations at the same time, the atomic entries were given occupancy of 0.25 (the 9th column in the PDB file).