Protein disorder: conformational distribution of the flexible linker in a chimeric double cellulase

Abstract

The structural properties of the linker peptide connecting the cellulose-binding module to the catalytic module in bimodular cellulases have been investigated by small-angle x-ray scattering. Since the linker and the cellulose-binding module are relatively small and cannot be readily detected separately, the conformation of the linker was studied by means of an artificial fusion protein, Cel6BA, in which an 88-residue linker connects the large catalytic modules of the cellulases Cel6A and Cel6B from Humicola insolens. Our data showed that Cel6BA is very elongated with a maximum dimension of 178 A, but could not be described by a single conformation. Modeling of a series of Cel6BA conformers with interdomain separations ranging between 10 A and 130 A showed that good Guinier and P(r) profile fits were obtained by a weighted average of the scattering curves of all the models where the linker follows a nonrandom distribution, with a preference for the more compact conformers. These structural properties are likely to be essential for the function of the linker as a molecular spring between the two functional modules. Small-angle x-ray scattering therefore provides a unique tool to quantitatively analyze the conformational disorder typical of proteins described as natively unfolded.

FIGURE 1

(a) Schematic cartoon (not to scale) of the modular organization of the Cel6B and Cel6A cellulases and of the chimeric variant Cel6BA from H. insolens. (b) Prediction of a long-disorder region (thick black line) in Cel6BA by PONDR, and sequence of the 88-residue linker of Cel6BA predicted as disordered.

FIGURE 2

Fit of the experimental scattering curves with the crystal structures (red line) for the isolated catalytic modules in (a) Cel6A and (b) Cel6B.

FIGURE 3

Guinier plot of the scattered intensity of Cel6BA in 50 mM sodium phosphate, pH 8.5 buffer. The radius of gyration R_g is inferred from the slope of the straight line fitting the data in the q-range qR_g ≤ 1.0. Protein concentration (from top to bottom): 13 mg/ml, 10.1 mg/ml, 6.9 mg/ml, 4.9 mg/ml, and 2.1 mg/ml.

FIGURE 4

Experimental distance distribution profile P(r) of the double cellulase Cel6BA (solid line), superimposed with the distance distribution profiles of the catalytic modules Cel6A (dotted line) and Cel6B (dashed line).

FIGURE 5

Calculated scattering curves for two-module Cel6A and Cel6B structures with and without the 88-residue intermodule linker. (a) The 12 scattering curves correspond to every 10th model generated in which the C-terminal α-carbon atom of the Cel6A module was separated by 10 Å to 120 Å in 10 Å steps from the N-terminal α-carbon atom of the Cel6B module. No linker is present. The minimum at 0.17 Å⁻¹ is indicated by an arrow (see text). The experimental curve is indicated by the dotted line. (b) The 12 scattering curves correspond to the 12 models of a, but now have linkers attached between the Cel6A and Cel6B modules. Although the same general features are observed as in a, the minimum at q of 0.17 Å⁻¹ (arrow) now varies from curve to curve. (c) The effect of simulations with glycosylated linker structures is shown. (d) The goodness-of-fit R-factors calculated from the comparison of the modeled and experimental scattering curves are shown as functions of the separation of the Cel6A and Cel6B modules with and without the linker peptide.

FIGURE 6

The refinement of a compact linker structure between the Cel6A and Cel6B modules. A total of 1000 linker conformations were generated in the course of an energy minimization by a molecular dynamics procedure. The α-carbon molecular views of the first and last models are shown as insets, together with a 20 Å scale bar.

FIGURE 7

Analysis of P(r) profiles for models of the Cel6A and Cel6B module structure. (a) The P(r) profiles of three models with intermodule separations of 10 Å, 60 Å, and 120 Å were represented by histograms of the distribution of their inter-Cα–Cα distances (red, green, and blue). These are compared with the experimental P(r) profile calculated using GNOM with an assumed maximum length of 180 Å (black dotted line). (b) The modeled P(r) profile (red) was calculated from a weighting scheme based on 124 models with intermodule separations between 6 Å and 129 Å in 1 Å steps. Each model possessed a linker that was energy-minimized (see Fig. 6) in order that each linker adopted a stereochemically reasonable conformation. These models have R-factors as shown in Fig. 6. The weighted summation of the 124 models generated the P(r) profile with a single peak maximum as shown in red. The experimental P(r) curve is shown as the dotted line. The relative weights for the 124 models are shown as an inset, with that at 15 Å set as 1.

FIGURE 8

Comparison between the experimental (dotted line) and calculated (solid line) scattering curves I(q). The calculated I(q) curve corresponds to the weighted summation of the 124 models used to generate the P(r) profile of Fig. 4. The R-factor is 7.4%. The α-carbon views of four typical molecular structures that were used for the weighted summation are shown below the curve fit.