Denatured State Conformational Biases in Three-Helix Bundles Containing Divergent Sequences Localize near Turns and Helix Capping Residues

Abstract

Rhodopseudomonas palustris cytochrome c', a four-helix bundle, and the second ubiquitin-associated domain, UBA(2), a three-helix bundle from the human homologue of yeast Rad23, HHR23A, deviate from random coil behavior under denaturing conditions in a fold-specific manner. The random coil deviations in each of these folds occur near interhelical turns and loops in their tertiary structures. Here, we examine an additional three-helix bundle with an identical fold to UBA(2), but a highly divergent sequence, the first ubiquitin-associated domain, UBA(1), of HHR23A. We use histidine-heme loop formation methods, employing eight single histidine variants, to probe for denatured state conformational bias of a UBA(1) domain fused to the N-terminus of iso-1-cytochrome c (iso-1-Cytc). Guanidine hydrochloride (GuHCl) denaturation shows that the iso-1-Cytc domain unfolds first, followed by the UBA(1) domain. Denatured state (4 and 6 M GuHCl) histidine-heme loop formation studies show that as the size of the histidine-heme loop increases, loop stability decreases, as expected for the Jacobson-Stockmayer relationship. However, loops formed with His35, His31, and His15, of UBA(1), are 0.6-1.1 kcal/mol more stable than expected from the Jacobson-Stockmayer relationship, confirming the importance of deviations of the denatured state from random coil behavior near interhelical turns of helical domains for facilitating folding to the correct topology. For UBA(1) and UBA(2), hydrophobic clusters on either side of the turns partially explain deviations from random coil behavior; however, helix capping also appears to be important.

Figure 1.

(A) Yeast iso-1 cytochrome c (red, PDB ID: 2YCC) UBA(1) (blue, PDB: 1IFY) fusion protein. UBA(1) is inserted between Phe(-3) and Lys(-2) (white spheres) of iso-1-Cytc. Single histidine substitution sites in UBA(1) are shown with orange spheres. Stick models are used to represent the heme and its points of attachment to iso-1-Cytc at Cys14, Cys17, His18 and Met80. His31 was mutated to Asn for the other single His variants. A Y27Q variant was prepared in the presence of His31. (B) Schematic representation showing how His-heme loop size varies in the denatured state for single histidine variants at positions 7, 11, 15, 24, 27, 31 or 35 in the sequence of UBA(1) when the histidine displaces the water in the 6^th coordination site of the heme of the fusion protein.

Figure 2.

GuHCl denaturation curves at pH and 25 °C for variants of UBA(1) – iso-1-Cytc, pWT (red circles), T7H (blue triangles), R24H (pink inverted triangles), and His31 (cyan squares). Plots show the dependence of corrected ellipticity at 222 nm, θ_222corr on the concentration of GuHCl. The curves (black) show the fits to eq 2 (Experimental Procedures) except for the His31 (WT) protein, which is fit to a two-state transition. Parameters from the fits can be found in Table 1.

Figure 3.

Loop formation for UBA(1) – iso-1-Cytc shown schematically. At low pH (left) the His-heme loop is broken. At high pH (right), the His-heme loop is formed.

Figure 4.

Denatured state (6 M GuHCl) pH titration data for selected UBA(1) – iso-1-Cytc variants. The dependence of A_398corr on pH is shown for the pWT (red circles), T7H (blue triangles), R24H (pink inverted triangles), and His31 (cyan squares) variants. The curves (black) show the fits to eq 3 (Experimental Procedures).

Figure 5.

Values of k_b at 25 °C and 6 M GuHCl plotted against loop size, N, for UBA(1) – iso-1-Cytc (blue squares). Data at 6 M GuHCl for UBA(2) – iso-1-Cytc (red triangles), and alanine inserts (KAAAAA)_n (n = 1 – 5, black circles) between Phe(-3) and Lys(-2) of iso-1-Cytc are included for comparison.

Figure 6.

Loop stability, pK_loop(His), as a function of loop size (logarithmic scale) under denaturing conditions (4 M Gu HCl) for the three-helix bundles UBA(1) (blue squares) and UBA(2) (red triangles) inserted at the N-terminus of iso-1-Cytc. pK_loop(His) data at 4 M (open circles) and 6 M (black circles) GuHCl for homopolymeric segments of alanine, (KAAAAA)_n, inserted into iso-1-Cytc between Phe(-3) and Lys(-2) are displayed for comparison. The UBA(1) and UBA(2) variants with slow loop breakage rates (See Figure 5) are labeled. Fits to eq 7 are shown for the UBA(2) data (red line, R = 0.912), the poly(Ala) data at 4 M (dotted black line) and 6 M (black line) GuHCl. Scaling exponents ν₃ obtained from the fits to eq 7 are 2.4 ± 0.5, 2.3 ± 0.1 and 1.97 ± 0.05, respectively. The dotted red line shows the effect of fitting the UBA(2) data without the E27H data point (ν₃ = 2.5 ± 0.4, R = 0.959). The blue dotted line is a fit to the combined UBA(1) and UBA(2) data with neither the E15H, His31, His31/Y27Q and E35H data points nor the E27H UBA(2) data point included (ν₃ = 2.2 ± 0.3, R = 0.938).

Figure 7.

Plot of hydrophobicity versus sequence position for WT (His31) UBA(1) – iso-1-Cytc (black line) versus pWT (C26A) UBA(2) – iso-1-Cytc for the (A) normalized Kyte-Doolittle and (B) hybrid Miyazawa-Jernigan/Hopp-Woods (MJHW) scales. Hydrophobicity was calculated using a nine-residue window. Reported average values of Kyte-Doolittle hydrophobicity, <KD>, and MJHW hydrophobicity, <MJHW>, and for HpC are for the regions between the N-terminus and Cys14 of the fusion proteins (i.e., the portion of the protein involve in loop formation). The sequence numbering corresponds to that used for the UBA(1) domain in Figure 1. Hydrophobic clusters used to calculate HpC correspond to regions of the MJHW plots with values above zero (dashed line). The secondary structure of the UBA domains in the fusion proteins is shown above the plots. Blue rectangles correspond to helices and blue lines correspond to the turns. The portion of the sequence derived from Cytc is labeled and denoted with a red line.

Figure 8.

Portions of the ILV cluster of UBA(1) that pack around each turn of UBA(1). (A) Turn 1 of UBA(1). The hydrogen bonds from the carbonyl of the N-cap residue Glu15 to amide NH atoms of Val19 and Arg18 are shown as yellow dashed lines. (B) Turn 2 of UBA(1). The hydrogen bonds from the carbonyl of the N-cap residue Pro30 to the amide NH of Val34 and from His31 to the amide NH of Glu35 are shown as yellow dashed lines. The positions of the histidine variants are shown with orange spheres.

Figure 9.

Plot of k_b for His-heme loop breakage versus the degree of hydrophobic clustering, HpC, of the portion of the UBA(1) – iso-1-Cytc N-terminal to the point of heme attachment. The solid red line shows the correlation of k_b with HpC for loop breakage in 6 M GuHCl (R = 0.96). The solid blue line is the same correlation at 4 M GuHCl (R = 0.89). The correlations do not include the His variants adjacent to turn 1 (E15H) and turn 2 (His31 and His31/Y27Q).