Publisher: CELL; Journal: MOLEC:Molecular Cell; Copyright:

Volume: 9; Issue: 5; Manuscript: ft; DOI: ; PII:

TOC Head: ; Section Head: ; Article Type: SHORT ARTICLE

Supplemental Results/Discussion and Experimental Procedures

Relaxation Experiments Reveal Increased Mobility of Wing Residues

The NMR ensemble of gpNu1 DBD displays structural disorder for the wing residues and the D helix (Figure 2A, main text). To determine whether this structural disorder reflects conformational mobility, a set of T1, T2, and 1H-15N NOE experiments were recorded (Figure S1). Even a qualitative inspection of Figure S1 reveals mobility for both the wing residues and the C-terminal region of the DBD. Wing residues (31-39) display decreased 1H-15N NOEs and increased T2 values relative to the corresponding parameters of amino acids in well-defined parts of the protein (helices A, B, and C, for example). Within the wing, the NOE decreases monotonously from residue 31, reaching its minimum value at Lys35, the residue at the tip of the wing. Beyond Lys35, the NOEs gradually return to values found for secondary structure elements. The reduction in 1H-15N NOEs and increase in T2 values compared to those of amino acids in ordered elements suggest significant internal motions on a nanosecond to microsecond time scale for residues in the wing (Tjandra et al., 1995).

Residues within the D helix display a similar pattern, reflecting motions on the NMR timescale. A steep decrease in 1H-15N NOEs beyond residue 53 is paralleled by steadily increasing T2 values toward the C terminus. This supports our conclusion that the DNA binding domain consists of the 55 N-terminal amino acids of gpNu1 and that the C-terminal 13 residues of the DBD are part of an interdomain helical linker (Bain et al., 2001).

Determination of the Diffusion Tensor of gpNu1 DBD

Prior to analysis of the 15N relaxation parameters with Modelfree, the rotational diffusion characteristics of gpNu1 DBD were examined. Diffusion tensors were calculated from the 15N relaxation data as described below in the Experimental Procedures. Table S1 shows that neither the symmetric diffusion model nor the anisotropic diffusion model provided significant statistical improvement over a simple isotropic model. This indicates that the gpNu1 DBD molecule tumbles isotropically in solution. At first glance this is surprising given the almost axial symmetry of the gpNu1 DBD structures. Furthermore, the ratio of the principal components of the inertia tensors as calculated from the NMR ensemble is 1.00:0.87 ± 0.04:0.57 ± 0.07, also arguing against an isotropic diffusion model. The discrepancy between the estimated isotropic diffusion tensors and the asymmetry of the inertia tensors is likely due to the artifactual enrichment of structures with -helical character in the C terminus in the NMR ensemble. This remains an inherent drawback of the structural calculation approach, which attempts to fit all distance restraints to a single structure. While a series of (i, i + 3) NOEs clearly attest to the helical character of D, a set of intense sequential -N NOEs appear at odds with a well-ordered  helix. In addition, relaxation parameters point toward increased mobility of this region (Figure S1 and below), while TALOS (Cornilescu et al., 1999) analysis of backbone chemical shifts fails to identify helical character for these residues. The conformation of the residues 53-68 of gpNu1 is thus partially overdetermined, and gpNu1 DBD is more globular than it appears in the NMR ensemble. Since this segment is merely a vestigial, and incomplete, part of the linker to the second domain, this bears little relevance from a structural perspective of course; however, its presence has dramatic effects on the rotational diffusion dynamics of gpNu1 DBD. In addition, the gpNu1 DBD dimer also behaves larger than expected for a 17 kDa globular protein. The C-terminal segment is likely also responsible for the increased overall rotational correlation time (Table S1), which can be readily appreciated by inspection of Figure S2. The protein is significantly larger due to the presence of the disordered C terminus. Nevertheless, an initial analysis of 15N relaxation parameters already provides insight in the dynamics of gpNu1 DBD.

Modelfree Analysis of Relaxation Parameters of gpNu1 DBD

To determine the order of the timescale of wing residue motions, the relaxation parameters R1, R2, and 1H-15N NOEs were fit using the Modelfree package (Palmer et al., 2001). Relaxation rates could be obtained for 63 of the 68 residues. Thirty-nine of these residues were appropriately described by a model containing a single S2 parameter (Figure S3, top panel). These residues are all located in the secondary structure elements of the DNA binding domain of gpNu1 or in short loops in between them. The average S2 for these residues is 0.92 ± 0.05. The wing residues were best fit with a model that contained e in addition to S2 (Figure S3, center panel). Six residues in the C terminus also required e in addition to S2, while the remaining eight residues required S2f, e, and S2 parameters.

Wing residues 31 and 39 display S2 values of 0.85 and 0.84, respectively, which is close to the average value of 0.92 ± 0.05 for ordered residues. In contrast, S2 values of the other wing residues 32 to 38 lie between 0.42 and 0.64, reflecting significant disorder for these residues. The corresponding e values vary between 1.1 and 1.45 nanoseconds with an average of 1.25 ± 0.14 for residues 31-38. Since the motion of these residues occurs on approximately the same timescale, it seems that the entire wing undergoes a concerted motion on a low nanosecond timescale.

Supplemental Experimental Procedures

Inertia tensors of gpNu1 DBD were determined with the program pdbinertia, which calculates the principle moments of inertia from a pdb file. Diffusion tensors were calculated from 15N relaxation data using the programs r2r1_diffusion and quadric_diffusion (all programs were obtained from These programs are based on approaches to determine the diffusion tensors for spherical (Tjandra et al., 1995), axially symmetric (Bruschweiler et al., 1995), and fully anisotropic (Lee et al., 1997) motional models from nitrogen-15 relaxation data. Only residues listed in Table S1 were used for these analyses, because amino acids in the wing and C terminus display significantly lower 1H-15N NOEs and higher T2 values than average values (Figure S1). T1, T2, and 1H-15N experiments were recorded at 600 MHz as described (Farrow et al., 1994).

T1 experiments were recorded with T1 delays of 11.1, 66.6, 144, 244, 366, 532, 765, 1110, and 1500 milliseconds. T2 experiments were recorded with 16.6, 33.3, 49.9, 66.6, 83.2, 99.8, 116.5, 133.1, 149.8, 166.4, and 183.0 milliseconds. Data were processed using NMRPipe (Delaglio et al., 1995) and analyzed using NMRview software (Johnson and Blevins, 1994). Relaxation parameters R1, R2, and 1H-15N NOEs were fitted using the Modelfree software package as described by Palmer et al. (2001) ( The statistical approach as outlined by Mandel et al. (1995) was used to select appropriate models for each amino acid. An initial estimate of tc = 12.7 nanoseconds was used for testing the following five models: (1) S2 alone; (2) S2 and e; (3) S2 and Rex; (4) S2, e, and Rex; and (5) S2, e, and S2f using an isotropic diffusion model. The final fit in which each amino acid was described with its appropriate model returned an overall correlation time of 12.2 nanoseconds. Results for this run are used for Figure S3. Analysis with an axially symmetric rotational diffusion model yielded essentially the same results (not shown).

Page 1 of 4