Destabilizing an interacting motif strengthens the association of a designed ankyrin repeat protein with tubulin

Abstract

Affinity maturation by random mutagenesis and selection is an established technique to make binding molecules more suitable for applications in biomedical research, diagnostics and therapy. Here we identified an unexpected novel mechanism of affinity increase upon in vitro evolution of a tubulin-specific designed ankyrin repeat protein (DARPin). Structural analysis indicated that in the progenitor DARPin the C-terminal capping repeat (C-cap) undergoes a 25° rotation to avoid a clash with tubulin upon binding. Additionally, the C-cap appears to be involved in electrostatic repulsion with tubulin. Biochemical and structural characterizations demonstrated that the evolved mutants achieved a gain in affinity through destabilization of the C-cap, which relieves the need of a DARPin conformational change upon tubulin binding and removes unfavorable interactions in the complex. Therefore, this specific case of an order-to-disorder transition led to a 100-fold tighter complex with a subnanomolar equilibrium dissociation constant, remarkably associated with a 30% decrease of the binding surface.

Figure 1. ELISA analysis of D1 variants after two rounds of error-prone PCR followed by selection using ribosome display.

DARPins at a 30 nM concentration were incubated in 10 nM biotinylated peptide-coupled tubulin-coated wells. For off-rate estimation, after DARPin binding and the washing steps, the wells were incubated either with buffer or with a 100 nM tubulin solution, to prevent the rebinding of detached DARPins to immobilized tubulin. The DARPins were ranked as low-, medium- or high-affinity binders. D1 is the parent DARPin whereas off7 is an unrelated one, not binding to tubulin and used as a negative control. Error bars are standard deviations from duplicate experiments. AU, absorbance unit.

Figure 2. The tubulin-DARPin interaction monitored by fluorescence spectroscopy and surface plasmon resonance (SPR).

(a) Fluorescence variation of 100 nM (left) or 15 nM (right) acrylodan-labeled DARPins as a function of tubulin concentration. The curve is the fit of the experimental points with Equation 1, from which the K_D is extracted. Error bars correspond to standard deviation from duplicate experiments. a.u., arbitrary units. (b) Dissociation of acrylodan-labeled DARPin from tubulin. In the case of D1 (left), 5.5 μM unlabeled D1 was added to a 100 nM labeled D1 and 0.5 μM tubulin mixture. The decrease in fluorescence signal was monitored in a stopped-flow apparatus (30% of the data points are shown). In the case of A-C2 (right), 2 μM unlabeled A-C2 was added to a 20 nM labeled A-C2 and 40 nM tubulin mixture and the fluorescence signal was monitored in a spectrofluorometer. The curve is the fit of the experimental points with a mono-exponential decay equation (Equation 2). (c) Determination of the association rate constant by fluorescence. Tubulin at the indicated concentrations was added to a fixed concentration (50 nM) of labeled D1 (left) or A-C2 (right) in a stopped-flow apparatus (20% of the experimental points are displayed). The data were fitted according to Equation 3. The variation of k_obs as a function of tubulin concentration is shown in inset, from which the k_on (slope of the curve) is derived. (d) The tubulin-DARPin interaction monitored by SPR. D1 (left) and TM-3 (right) were immobilized through their His-tag on the sensor chip. Tubulin at the indicated concentration was applied at time zero for 60 s (D1) or 180 s (TM-3), followed by a washing buffer flow. The black curves are the fit of the experimental data using the Langmuir analysis, from which the k_on and k_off are extracted. In the case of D1, plotting the values at the plateau as a function of the tubulin concentration provided an estimate of the K_D (Supplementary Fig. 5) that is very similar to the k_off/k_on ratio.

Figure 3. Assessing the contribution of the D1 residues mutated in A-C2.

(a) Sequence alignment of D1, A-C2 and the optimized TM-3 DARPin (color code: N-cap, green; 1^st internal repeat, bright blue, 2^nd internal repeat, dark blue; 3^rd internal repeat, cyan; C-cap, yellow; mutated residues relative to D1, red). (b) Structure of tubulin−D1 with the Cα position of the residues mutated in A-C2 highlighted as red spheres. D1 is colored according to panel A. (c) ELISA analysis (binding and off-rate estimation) of the 11 D1 single mutants, 2 double mutants (DM-1: H118R, I122V; DM-2: E127K, V131A) and of 2 triple mutants (TM-1: N74S, I76S, I78L; TM-2: F150Y, I152T, N158S). The experimental conditions were as in Fig. 1. (d) ELISA analysis of the DM-3 double mutant (I152T, N158S) and of the TM-3 triple mutant (with the additional H118R substitution) together with the corresponding single mutants. The experimental conditions were as in panel C, except that the concentration of biotinylated peptide-coupled tubulin used for coating was decreased 5-fold (down to 2 nM).

Figure 4. Conformational changes of D1 upon tubulin binding.

(a) Overview. Uncomplexed D1 (in orange, C-cap in yellow) has been superimposed on D1 in tubulin−D1 (D1 in green with the C-cap in brighter color, β tubulin in beige), taking the N-cap and the internal repeats as a reference. (b) Close-up (only β tubulin and uncomplexed D1 are shown). Without a C-cap rotation, its residues would clash with tubulin. Five distances shorter than 2 Å are highlighted as black solid lines.

Figure 5. The C-cap motif of the high affinity DARPins is mobile in the crystal structure.

(a) The H118R mutation destabilizes the C-cap. The TM-3 structure (blue) was superimposed to uncomplexed D1, colored as in Fig. 4a. The Arg118 side chain adopts two alternate conformations in TM-3 (Supplementary Fig. 6). Both would clash with the C-cap if it were folded as in D1 (3 distances shorter than 1.5 Å are highlighted as black solid lines), as would the many other Arg118 conformations we modeled. (b) Comparison of the complexes of tubulin (beige) with D1 (green) and with A-C2 (blue). The β tubulin subunits have been superimposed. Some side chains are drawn to help visualize the shift between the DARPins. (c) View of the tubulin−D1 interface colored by electrostatic potential. (Left) Electrostatic potential surface of β tubulin (red, negative; blue, positive) with bound D1 (green). The side chains of two acidic residues of the C-cap that are close to an acidic part of the surface of tubulin are shown. The Cαs of residues 148 and 149 are also highlighted as magenta spheres. Subtilisin cleaves the affinity-improved DARPins preferentially after these two residues (see text and Supplementary Fig. 2). (Right) Electrostatic potential surface of D1 centered on the tubulin-interacting surface. Note that the two panels are not to scale.

Figure 6. The C-cap motif of the high-affinity DARPins is disordered.

(a) Gel filtration analysis of 100 μM D1, A-C2 and TM-3. (b) SDS-PAGE analysis of the limited proteolysis of D1, TM-3 and A-C2 by subtilisin. DARPins at 40 μM concentration were incubated at 25 °C with subtilisin at a 1:2000 protease:DARPin molar ratio at the indicated times. (c) Far-UV circular dichroism spectra of 20 μM D1, A-C2 and TM-3 and of 5 μM TM-3 1-149. The spectra were recorded using a 1 mm path length cuvette. They were normalized and are depicted as molar ellipticity. (d) ELISA analysis (tubulin binding and off-rate estimation) of D1, A-C2 and TM-3 along with their C-cap-truncated variants (FL, full length; 1-149, DARPin terminating after residue 149). The experimental conditions were as in Fig. 3d. Here, as in Figs 1 and 3c,d, soluble tubulin at a 100 nM concentration was added after the washing steps to prevent the rebinding of dissociated DARPins.

Figure 7. The C-cap stability of D1 mutants conditions their affinity for tubulin.

SDS-PAGE analysis of limited proteolysis of DARPins in the conditions described in Fig. 6b, before (−) and after (+) incubation with subtilisin for 30 minutes. DARPins are ranked as low- (a) medium- (b) or high-affinity (c) tubulin binders according to Fig. 1.