Electrostatic Redesign of the [Myoglobin, Cytochrome b₅] Interface to Create a Well-Defined Docked Complex with Rapid Inter-Protein Electron Transfer

Abstract

Cyt b₅ is the electron-carrier ‘repair’ protein that reduces met-Mb and met-Hb to their O₂-carrying ferroheme forms. Studies of electron transfer (ET) between Mb and cyt b₅ revealed that they react on a “Dynamic Docking” (DD) energy landscape on which binding and reactivity are uncoupled: binding is weak and involves an ensemble of nearly isoenergetic configurations, only a few of which are reactive; those few contribute negligibly to binding. We set the task of redesigning the surface of Mb so that its reaction with cyt b₅ instead would occur on a conventional ‘simple docking’ (SD) energy landscape, on which a complex exhibits a well-defined (set of) reactive binding configuration(s), with binding and reactivity thus no longer being decoupled. We prepared a myoglobin (Mb) triple mutant (D44K/D60K/E85K; Mb(+6)) substituted with Zn-deuteroporphyrin, and monitored cytochrome b₅ (cyt b₅) binding and electron transfer (ET) quenching of the ³ZnMb(+6) triplet state. In contrast, to Mb(WT), the three charge-reversals around the ‘front-face’ heme edge of Mb(+6) have directed cyt b₅ to a surface area of Mb adjacent to its heme, created a well-defined, most-stable structure that supports good ET pathways, and apparently coupled binding and ET: both K_a and k_et are increased by the same factor of ~ 2×10², creating a complex that exhibits a large ET rate constant, k_et = 10^{6 1}s⁻¹, and is in slow exchange (k_off ≪ k_et). In short, these mutations indeed appear to have created the sought-for conversion from DD to simple docking (SD) energy landscapes.

Interactions between macromolecules are fundamental to most of biology. The goal of understanding and controlling protein-protein interactions has inspired numerous efforts to modulate the binding within high-affinity complexes of known structure,^1–6 with most alterations serving to weaken binding. To begin such efforts at the other extreme, with a pair of proteins that bind weakly, and by altering them to create a well-defined, highly functional complex, offers a different level of challenge. We here report dramatic progress in creating such a complex between the weakly interacting physiological electron-transfer (ET) partners, myoglobin (Mb) and cytochrome b₅ (cyt b₅).

Cyt b₅ is the electron-carrier ‘repair’ protein that reduces met-Mb⁷ and met-Hb^8,9 to their O₂-carrying ferroheme forms. Studies of ET between Mb and cyt b₅ revealed that they react on a “Dynamic Docking” (DD) energy landscape (Fig 1, upper),^10–12 on which binding and reactivity are uncoupled: binding is weak and involves an ensemble of nearly isoenergetic configurations, only a few of which are reactive; those few contribute negligibly to binding. The sharp contrast of such behavior with that for the conventional ‘simple docking’ (SD) energy landscape (Fig 1, lower), on which a complex exhibits a well-defined (set of) reactive binding configuration(s), led us to ask whether it was possible to redesign the surface of Mb so that its reaction with cyt b₅ instead would occur on a SD landscape, with binding and reactivity thus no longer being decoupled.

Fig 1.

Conversion of ‘Dynamic Docking’ (DD) to ‘Simple Docking’ (SD) Energy landscapes; rainbows represent ET-active configurations.

The low affinity of these partners is the consequence of poor shape complementarity, weak overall electrostatic attractions between the highly charged cyt b₅ and the nearly neutral Mb (q_b5 = − 5.72 C; q_Mb ~ − 0.29 C), and the absence of optimally placed complementary charge pairs on the surfaces of the partners. Our initial attempt to enhance their affinity electrostatically¹³ by neutralizing the Mb heme propionates through esterification increased the ET rate by ~100-fold but surprisingly, did not significantly alter the affinity; this observation in fact led us to formulate the DD paradigm.¹⁰

To identify surface residues of Mb where mutation would best enhance reactive binding, we performed Brownian Dynamics (BD) simulations¹⁴ of the docking between cyt b₅ and all Mb surface charge reversal mutants (D/E → K). To probe net binding we used a center-of-mass (COM) criterion to define a BD trajectory ‘hit’; to probe reactive binding we defined a hit to have a short distance between one O of a cyt b₅ heme carboxylate and one O of a Mb heme carboxylate (O-O criterion).^15,16 The BD profile for Mb(WT) (Fig 2, top left) shows that cyt b₅ ‘binding hits’ (COM criterion) are distributed over the entire Mb surface, with some clustering in patches away from the heme edge, and few ‘reactive hits’ (OO criterion) near the Mb(WT) heme edge (Fig 2, top, right). The twenty most stable binding hits (COM criterion) fall at a Mb surface patch ‘above and to the left’ of the heme edge. Among these, there is no correlation of the cyt b₅ orientations (backbone RMSD, 11.2 Å) as illustrated by their differing cyt b₅ heme orientations (Fig 2, top center).

Fig 2.

BD simulations for docking Mb(WT) (upper) and Mb(+6) (lower) to cyt b₅ (10⁴ trajectories; pH 7; : = 18mM); mutation sites, magenta. ‘Hit’ profiles using COM (Left) and O-O (Right) reaction criteria; blue dots, cyt b₅ COM at a hit (see refs ¹²,¹⁴). Center: Mb heme (black) plus superpositions of the cyt b₅ hemes for the five most-stable configurations from the COM simulations with Mb(WT) (upper), Mb(+6) (lower); hemes properly oriented relative to the Mb heme. Arrows on left indicate corresponding locations of the cyt b₅ COM relative to Mb face.

Not surprisingly, computations with the mutant Mbs showed that binding of the negatively charged cyt b₅ to Mb is opposed by the three acidic residues surrounding the exposed heme on the Mb front face (D44, D60, and E85). We then tested whether the cumulative effects of reversing the charge at these three sites would stabilize the complex in a reactive configuration with the hemes proximate to one another, thereby reshaping the energy landscape to more closely resemble one of simple docking, Fig 1.

Simulations for the triple mutant (Fig 2, lower) indeed show a high density of hits near the Mb heme edge with both criteria. Moreover, the twenty most-stable cyt b₅ configurations now correspond to a single well-defined structure (RMSD, 0.6 Å),¹⁷ as illustrated by the close overlap of their hemes, (Fig 2, center, lower) with their heme edges facing that of Mb. Thus, the simulations suggest that cyt b₅ might bind strongly to the Mb(+6) triple mutant (D44K/D60K/E85K) as a reactive ‘simple-docking’ complex in which binding and reactivity are coupled.

To test these predictions, we prepared ZnMb(+6), the (D44K/D60K/E85K) triple mutant substituted with Zn deuteroporphyrin, and monitored binding and ET quenching of the ³ZnMb triplet state formed by rapid intersystem crossing from the laser-flash generated singlet state of ZnMb.^18,19,20

Throughout a titration of Mb(WT) with Fe³⁺b₅ at pH 6 (µ = 5.3 mM), the ZnMb triplet decay traces remain exponential, with a decay constant (k_obs) that increases linearly with [Fe³⁺b₅] from the intrinsic decay constant, k_D, without evidence of saturation: k_obs = k_D + k₂ [Fe³⁺b₅]. The slope of this line is the bimolecular quenching rate constant, k₂ = 4.7×10⁶ M⁻¹s⁻¹, similar to the value reported previously.^10,12 The measurements indicate this complex is in fast exchange, k_off ≫ k_et, where k_et represents the apparent intracomplex ET rate constant, and as a result, k₂(WT) = k_et(WT)K_a(WT), where K_a is the overall binding constant. Combination of the measured k₂(WT) and reported value, K_a(WT) ≈ 10³ M⁻¹,^21,22 gives, k_et(WT) ≈ 4–5 ×10³ s⁻¹.

In contrast, the triplet decay traces for ZnMb(+6) are biexponential throughout a titration with Fe³⁺b₅, and the larger rate constant is invariant with [Fe³⁺b₅], k_f = 10⁶ s⁻¹ (Fig 3). These observations indicate that the complex is in slow exchange with its unbound partners (k_off ≪ k_et), and that this phase represents rapid ET quenching within the bound [ZnMb, Fe³⁺b₅] complex: k_f. = k_et(+6) = 10⁶ s⁻¹.²³ The rate constant for the slower process increases linearly with addition of Fe³⁺b₅, k_s = k₂ [Fe³⁺b₅] + k_D, and thus is associated with the bimolecular reaction between free ZnMb(+6) and Fe³⁺b₅; the second-order rate constant is, k₂ = 3.1×10⁹ M⁻¹s⁻¹, corresponding to diffusion-limited ET.

Fig 3.

Titration of ZnMb(+6) with Fe³⁺ cyt b₅. Left. Triplet decay traces as a function of [Fe³⁺ cyt b₅] (:M). Conditions: 5 µM ZnMb; 5 mM KPi, pH 6; 20°C; 475 nm. Right. Zero-time absorbance decrease for cyt b₅ binding to Mb(+6) and fit to eq 1: K_a = 2.2 × 10⁵ M⁻¹; *)A₀^max = 0.042. Inset: Rate constants from fitting triplet decays to bi-exponential.

Typically, one would fit the cyt b₅-dependent fraction of the intra-complex phase, f(Fe³⁺b₅), with a standard binding isotherm to obtain the binding constant, K_a. However, as seen in Fig 3, the total amplitude of the triplet signal, measured as the extrapolated zero-time amplitude, )A₀(Fe³⁺b₅), decreases during a titration, an indication that the bound complex not only exhibits rapid ET from the ZnMb(+6) triplet state to Fe³⁺b₅, but also extremely rapid intra-complex quenching of the singlet state. We instead determined K_a by obtaining f(Fe³⁺b₅) from )A₀(Fe³⁺b₅) during a titration by use of the relationship,

$$\begin{array}{c}){\mathrm{A}}_0(0)-){\mathrm{A}}_0({\text{Fe}}^{3+}{b}_{\mathit{5}})=[){\mathrm{A}}_0(0)-){\mathrm{A}}_0(\infty )]f({\text{Fe}}^{3+}{b}_{\mathit{5}})\\ *){\mathrm{A}}_0=*){{\mathrm{A}}_0}^{\text{max}}f({\text{Fe}}^{3+}{b}_{\mathit{5}})\end{array}$$ (1)

and fitting *)A₀ to a binding isotherm. The resulting binding constant, K_a(+6) = 2.2×10⁵ M⁻¹, is more than two orders of magnitude greater than that for ZnMb(WT).^21,22

Thus, while cyt b₅ reacts with Mb(WT) on a DD energy landscape, with weak binding in multiple configurations - most of which are non-reactive, a low second-order ET rate constant, and rapid exchange (k_off ≫ k_f) between the complex and its unbound components, the three charge-reversals around the ‘front-face’ heme edge of Mb(+6) have directed cyt b₅ to a surface area of Mb adjacent to its heme, and created a well-defined, most-stable structure (Fig 2) that supports good ET pathways. The mutations indeed appear to have coupled binding and ET, increasing both K_a and k_et by a factor of ~ 2×10², and creating a complex with a large ET rate constant, k_et(+6) = 10⁶ s⁻¹, that is in slow exchange (k_off ≪ k_et). In short, it appears that these mutations have created the sought-for conversion from DD to SD energy landscapes.