Phosphorylation disrupts long-distance electron transport in cytochrome c

Article https://doi.org/10.1038/s41467-022-34809-1 Phosphorylation disrupts long-distance electron transport in cytochrome c AlexandreM.J.Gomila 1,2,9 , Gonzalo Pérez-Mejías 3,9 ,AlbaNin-Hill 4,9 , Alejandra Guerra-Castellano 3 , Laura Casas-Ferrer 1,8 , Sthefany Ortiz-Tescari 1 , Antonio Díaz-Quintana 3 , Josep Samitier 1,2,5 ,CarmeRovira 4,6 , Miguel A. De la Rosa 3 , Irene Díaz-Moreno 3 ,PauGorostiza 1,2,6 , Marina I. Giannotti 1,2,7 & Anna Lagunas 1,2 It has been recently shown that electron transfer between mitochondrial cytochrome cand the cytochrome c 1 subunit of the cytochrome bc 1 can proceed at long-distance through the aqueous solution. Cytochrome cis thought to adjust its activity by changing the affinity for its partners via Tyr48 phosphorylation, but it is unknown how it impacts the nanoscopic environment, interaction forces, and long-range electron transfer. Here, we constrain the orientation and separation between cytochrome c 1 and cytochrome cor the phosphomimetic Y48pCMF cytochrome c, and deploy an array of singlemolecule, bulk, and computational methods to investigate the molecular mechanism of electron transfer regulation by cytochrome cphosphorylation. We demonstrate that phosphorylation impairs long-range electron transfer, shortens the long-distance charge conduit between the partners, strengthens their interaction, and departs it from equilibrium. These results unveil a nanoscopic view of the interaction between redox protein partners in electron transport chains and its mechanisms of regulation. Many electron transfer (ET) reactions occurring in vivo implicate pairs of proteins in which at least one of them is free to diffuse, like the small electron-carrying proteins plastocyanin, ferredoxin or cytochrome c (Cc)1.Specific recognition and binding between proteins occur widely in biology. To reconcile an overall sustained and efficient ET process demands a high turnover rate. Therefore, the complex formed between redox proteins in the electron transport chain (ETC) must be transient, and balance specificity and binding strength2,3.Therecent direct experimental observation of electrochemically gated, longdistance ET between cytochrome c 1 (Cc 1 , subunit of the cytochrome bc 1 )andCcthrough the aqueous solution suggested that such tradeoff is achieved without direct contact, by molding the ionic distribution and electric field between their redox-active sites4. Thus, these redox protein partners might conciliate high specificity with weak binding, to keep the high turnover rate required by their biological function5. Ccis a mitochondrial hemeprotein that exerts multiple functions, and its location depends on the cell condition6–10. Under homeostasis, Cclocalizes in the intermembrane mitochondrial space, transferring electrons from cytochrome bc 1 complex (Cbc 1 , complex III) to cytochrome coxidase (CcO, complex IV) in the ETC. Two binding sites for Received: 18 July 2022 Accepted: 8 November 2022 Check for updates 1 Institute for Bioengineering of Catalonia (IBEC), The Barcelona Institute for Science and Technology (BIST), Barcelona, Spain. 2 CIBER-BBN, ISCIII, Barcelona, Spain. 3 Institute for Chemical Research–cicCartuja, Universidad de Sevilla, Consejo Superior de Investigaciones Científicas (CSIC), Sevilla, Spain. 4 University of Barcelona, Department of Inorganic and Organic Chemistry, Institute of Theoretical Chemistry (IQTCUB), Barcelona, Spain. 5 Department of Electronics and Biomedical Engineering, University of Barcelona (UB), Faculty of Physics, Barcelona, Spain. 6 Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain. 7 Department of Materials Science and Physical Chemistry, University of Barcelona (UB), Faculty of Chemistry, Barcelona, Spain. 8 Present address: Laboratoire Charles Coulomb (L2C), UMR 5221 CNRS-Université de Montpellier, Montpellier, France. 9 These authors contributed equally: Alexandre M. J. Gomila, Gonzalo Pérez-Mejías, Alba Nin-Hill. e-mail: [email protected];[email protected];[email protected]; [email protected];[email protected] Nature Communications | (2022) 13:7100 1 1234567890():,; 1234567890():,; Ccon Cc 1 and CcO have recently been reported, a functional site (proximal site) enabling the correct orientation of redox centers to optimize the ET route, and a non-productive one (the so-called distal site) which is implicated in the turnover of Ccmolecules11,12.Ccalso takes part in other redox reactions within mitochondria, including reactive oxygen species (ROS) scavenging, redox-coupled protein import via Erv1-Mia40 pathway, and reduction of p66Shc generating ROS8. Upon distinct stimuli, such as oxidative stress or DNA damage, Cccan relocate to several cell organelles, interacting with proteins governing cell life and death13–20. Regarding oxidative stress, around 0.2-2% of the electrons involved in the ETC yield ROS via complex I (mainly), II, and III21. ROS have a dual role in cell signaling showing regulatory behavior at low levels and inducing apoptosis through oxidative stress at high ROS concentration. When mitochondrial production of hydrogen peroxide increases, Ccfunctionsasaproapoptotic protein via its pseudo-peroxidase activity22. Around 15% of the Ccpopulation is anchored to the inner mitochondrial membrane through binding to the lipid cardiolipin23,24. The pseudo-peroxidase function of the cardiolipin-Cccomplex is activated at high levels of hydrogen peroxide and leads to the release of the hemeprotein from the mitochondria into the cytosol where it triggers apoptosis25,26.In mammals, Ccinteracts in the cytosol with the apoptosis protease activating factor-1 protein (Apaf-1), to form the apoptosome and leading to downstream caspase activation and cell death8,27,28. These moonlighting Ccactivities are tightly regulated by posttranslational modifications (PTM)29,30. Notably, phosphorylation occurs in physiological and disease conditions and is highly tissue specific. Up to five tissue-specific phosphorylation sites of Cchave been characterized in mammals, including Thr28, Ser47, Tyr48, Thr58 and Tyr9731. Phosphorylation of Cchas been described to exert a cytoprotective and antiapoptotic role by downregulating ETC flux, thereby preventing mitochondrial membrane potential hyperpolarization, decreasing ROS production, and avoiding caspase activation. Phosphorylation of Cchas also been related to several pathologies, such as ischemia/reperfusion injury, neurological disorders, and cancer progression32. Although phosphorylation mainly occurs on serine and threonine residues with high frequency, phosphorylation of tyrosine residues has a crucial role in the regulation of mechanisms directly relevant to mitochondrial and cancer signaling, such as metabolic homeostasis, differentiation, and proliferation33,34.Among the two Cctyrosine residues that can be phosphorylated in vivo, Tyr48 residue is located at the Ω-loopinthevicinityofthehemecleft, whereas Tyr97 lays far from the heme cleft in the C-terminal αhelix of Cc.Ccphosphorylated on residue Tyr48 was primarily isolated from bovine liver tissue and it has been related to cancer35. Moreover, it has been found that Tyr48 can be mutated by histidine, which causes a mild human disease—thrombocytopenia—that results in a lower level of blood platelets36. Therefore, discerning the molecular mechanisms that regulate the function of phosphorylated Ccat position 48 is of great interest for understanding the role of Ccin pathology. Functional studies revealed that Tyr48-phosphorylated Ccpreserves the heme environment (no changes are observed in the Fe-Met80 absorption band at 695 nm) and that both Tyr48- phosphorylated and nonphosphorylated Ccspecies produced a hyperbolic kinetic response and the maximal turnover number of non-phosphorylated Ccwas two times higher than that of phosphorylated Cc(3.7 s−1and 8.2 s−1, respectively)35.Thetechnicaldifficulties to preserve the in vivo protein phosphorylation after purification, the large amount of phosphoprotein needed for structural and biophysical assays and the lack of information onspecificCc- Tyr48 kinases requires adopting other approaches to study this phosphorylated protein. Over the last years, some authors have developed engineered aminoacyl-tRNA synthetase systems that can incorporate O-phosphotyrosine. However, this approach presents multiple challenges as phosphotyrosine poorly penetrates the cells due to its charge, it is unstable inside the cells and the incorporation efficiency by the aminoacyl-tRNA synthase is relatively low, producing a low yield of labelling (<5%) that is incompatible with single molecules techniques37. A convenient approach is using phosphomimetic mutants for functional studies38.TheCcY48E mutant, in which Glu mimics the negative charge of the phosphate group, showed similar reaction kinetics to the non-phosphorylated Ccbut presented a −45 mV midpoint redox potential shift. This mutation also inhibits the activation of caspase downstream signaling in vitro39.Nevertheless, the Y48E mutation leads to ca. 20% decrease in the volume of residue Tyr48, whereas it is increased upon phosphorylation. Thus, using Y48E to mimic phosphorylated Y48 underrepresents volume-related effects40. Therefore, the incorporation of a noncanonical amino acid that presents volume and charge values closer to those of phosphotyrosine is a better alternative. Among the different noncanonical amino acids that can be used to mimic Tyr-phosphorylation, p-car- boxy-methyl-L-phenylalanine (pCMF) is one of the best substitutes, as it is stable, efficiently incorporated by E. coli cells, and preserves a similar volume than the phosphotyrosine residue41. This mutation does not affect the overall structure and heme iron environment as Tyr48-phosphorylated Cc(Supplementary Fig. 1)42. In terms of functionality, the Y48pCMF Ccshows lower binding affinity for the distal sites of Cc 1 and CcO, altering the diffusion pathway of Ccmolecules through Cbc 1 -CcO, thus modulating the ETC flux42. Finally, the Y48pCMF mutation effectively enhances peroxidase activity and induces an antiapoptotic function of Cc42. Our recent observation of inter-protein long-distance ET between Cc 1 and Ccthrough the aqueous solution4can be explained by the presence of a charge conduit at the active interface between the two proteins (Gouy-Chapman conduit). These results suggestthat ET could occur already when the two proteins are approaching (i.e., in the encounter state) without the need of establishing a well-defined, static protein complex, thereby reconciling high specificity with weak binding, and allowing to keep the high turnover rate required in the ETC5. Within this scenario, the presence of a charge conduit is of paramount importance for a sustained electron transport. The spatial extent of the charge conduit was found to depend on the redox potential of the interacting proteins and on the charge distribution between them. Here we aim to investigate the molecular mechanism of ET regulation by phosphorylation in Cc-Cc 1 . We deploy an array of singlemolecule, bulk, and computational methods to address this question in wildtype, phosphomimetic, and phosphorylated Cc.Ourfindings show that phosphorylation strengthens the interaction with Cc 1 (lower equilibrium dissociation constant—K D —) and disrupts the Gouy- Chapman conduit associated with long-distance charge transport. These results could be of general significance for inter-protein ET, its regulation under diverse conditions in the cell, and the switching between multiple tasks. Results Long-distance ET between Cc-Cc 1 partners is impaired by Cc Tyr48 phosphorylation The effects of Tyr48 replacement by the phosphomimetic pCMF in the inter-protein ET between Ccand the mitochondrial complex III, or Cbc 1 , were studied using the cross complex between human Ccand the soluble domain of plant cytochrome Cc 1 —which contains all residues of the physiological mature (without mitochondrial transit peptide) protein until Glu265, at which the C-terminal hydrophobic helix starts12—by means of electrochemical scanning tunneling microscopy (EC-STM)4.Ccand Cc 1 were immobilized on the probe and the sample electrodes of EC-STM, respectively, with their redox active sites facing the solution, thereby allowing protein interaction. Cc 1 was immobilized through its native N-terminal Cys10 to the Au(111) sample electrode of the EC-STM, and Ccor Y48pCMF Ccwere anchored to the gold Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 2 probe of the EC-STM through the C-terminal mutation E104C, thus facilitating the homogenous orientation of the proteins, while still preserving rotational freedom and good conductivity43. The experiments were conducted as described4in an electrochemical cell under bipotentiostatic control of the probe and sample electrodes versus an Ag/AgCl (SSC) reference electrode, spanning a potential range around the redox potentials of Cc 1 and Cc,andalways keeping Cc 1 reduced (e.g. sample potential (U S )=−0.20 V) and Cc oxidized (e.g. probe potential (U P ) = 0.60 V). These values yield a constant positive bias U bias =U P -U S = 0.80 V by which the electrons are transferred from Cc 1 to Cc,inanalogytophysiologicalET. Current-distance (I-z) electrochemical spectroscopy was conducted between Cc 1 -Ccand Cc 1 -Y48pCMF Ccin 50 mM phosphate buffer pH 6.5. By positioning the EC-STM probe over the sample at a current set point of 0.4 nA, and after a stabilization period, the feedback loop was briefly disconnected, and the probe current recorded while the piezoelectric scanner holding the probe was retracted 15 nm fromthesetpointat12nm·s −1. Then, the feedback was restored, and the probe reengaged, allowing subsequent recordings at different areas of the sample. Faradaic leakage current was maintained below a few pA over the entire recording range through probe insulation44.A nearly exponential current decay was observed in both cases from the initial current set point up to the distance-independent faradaic leakage current. For Cc 1 -Ccthe current decay spans several nanometers as previously reported4, while for Cc 1 -Y48pCMF Ccthe current decays more abruptly, reaching leakage values 3 nm away from the set point (Fig. 1a). Fitting the exponential region of the curves yields the distance decay factor (β) distributions shown in Fig. 1b, which are centered at β=1.3±0.8nm −1for Cc 1 -Cc,andatβ 1 =1.3±0.2nm −1and β 2 =3.1±0.4nm −1for Cc 1 -Y48pCMF Ccshowing a bimodal distribution (Fig. 1c). Although β 1 values indicate that long-distance ET still takes place for a small population of recordings (up to 22%) in Cc 1 -Y48pCMF Cc,β 2 population doubling the βobtained for Ccsuggests that the replacement of Tyr48 by the phosphomimetic pCMF causes a severe alteration of the inter-protein ET. The presence of the phosphomimetic group can affect inter-protein ET by perturbing the two main steps involved in the process. Namely, modifying intra-protein conduction pathways at the protein level or altering the interface between the two proteins and thus affecting the charge conduit. When we compared the βvalues between bare gold and either Ccor Y48pCMF Cc,wefoundnosignificant differences (Supplementary Fig. 2), thus indicating that the Y48pCMF mutation does not modify protein conduction, and that the changes observed in Fig. 1a, b may involve the charge conduit that is established between Cc 1 and Cc. Electrochemical gating experiments were conducted for the Cc 1 - Y48pCMF Ccinteraction. We recorded I-z plots at different probe and Probe current (A) Piezo retraction (nm) Counts β(nm -1) ab Cc/Y48pCMF Cc ∆z Cc1 d c β(nm -1) EC Gate (V) CcY48pCMF Cc β(nm -1) **** UP(V) 76 2 . . . . . 0.5 .04.0 0. 0. . ... . . . . . . 0.3 . Fig. 1 | Current-distance electrochemical tunneling spectroscopy of Cc 1 -Cc. aEnsemble of semi-logarithmic current-distance (I-z) curves during probe retraction, showing a more abrupt current decay for Cc 1 -Y48pCMF Cc. Sample and probe electrodes are represented by a square and a triangle, respectively. Us = −200 mV vs. SSC and at constant bias of 800 mV. bHistograms of distance decay factors (β) quantified from individual curves in a.cplot of the averaged βvalues (mean ± s.d.) β=1.3±0.8nm −1for Cc 1 -Cc(n = 103 I-z curves from 2 independent experiments, red) and β=3.1±0.4nm −1Cc 1 -Y48pCMF Cc(n=64I-z curves from 2 independent experiments, blue), respectively, showing significant differences (two-sample ttest, t Statistic = 19.65, degrees of freedom = 195, **** P<0.0001).dAveraged β values (mean ± s.d. of n= 80) vs. EC gate potential at 200 mV constant bias obtained for Cc 1 -Cc(red) and Cc 1 -Y48pCMF Cc(blue). The midpoint redox potentials for Cc 1 at the sample (0.276 V; green dashed line) and Ccor Y48pCMF Cc(red and blue dashed lines, respectively) at the probe are indicated, showing the βminimum for Cc 1 -Ccis located within the midpoint redox potentials of Cc 1 and Cc. Despite being redox-active at similar potential than Cc,Y48pCMF Ccdid not display a βminimum within the measurement range. All experiments were performed in 50 mM phosphate buffer, pH 6.5. Initial current set point 0.4 nA. Averaged βvalues for Cc 1 -Cc (red) are reproduced from reference 44. Source data are provided as a Source Data file. Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 3 sample potentials keeping the bias constant. The potential of the reference electrode was adjusted to modulate the gate voltage in analogy to the gate electrode in a field effect transistor. As the EC-STM probe is grounded in our STM electronic configuration, the EC gate corresponds to -U S at any given bias45.ForCc 1 -Ccwe have previously observed that βdisplays a marked potential dependence at a moderate bias of 200 mV. Averaged βvalues showed a minimum (β= 0.5 ± 0.3 nm−1)atanECgatepotentialof−0.25 V, which was located between the midpoint redox potentials of Cc 1 (0.28 V vs. SSC) and Cc(0.35 V vs. SSC), respectively (Fig. 1d, red curve). This potential-dependent effect allowed to measure current spanning morethan 10 nm away from the departure setpoint and highlights the contribution of the redox sites in the ET between Cc 1 and Cc4.Theβ values of Cc 1 -Y48pCMF Ccdo not depend on the EC gate within the considered redox potential window (Fig. 1d, blue curve), in striking contrast to Cc 1 -Cc. The midpoint redox potential of immobilized Y48pCMF Cc(0.36 V vs. SSC), is very similar to that of Cc(Supplementary Fig. 2), as previously reported41, and contrary to the Y48E substitution, which shows a ∼80 mV reduction of its midpoint redox potential with respect to Cc39. The redox potential values obtained are within the previously reported values for electron-transfer active Cc directly immobilized on an Au electrode46,47. Therefore, Y48pCMF Cc maintains its redox activity but cannot be engaged in long-distance ET through the aqueous solution. This is likely due to an alteration in the charge conduit between the protein partners rather than coarse protein structural changes. Ccstructure consists of five cooperative folding-unfolding units (foldons)48 and its midpoint redox potential is very sensitive to protein folding and heme group coordination49,50.As Met80 is located in the second most unstable foldon (foldon IV) of Cc, subtle alterations of Ccstructure affect the Met80-Fe coordination bond and, therefore, the redox potential value. Indeed, several authors have demonstrated the important role of Met80-Fe coordination on Ccredox potential, as the loss of this bond shifts the redox potential towards negative values (−200 mV)51. Based on our voltammograms (Supplementary Fig. 2), we can rule out the hypothesis of partial or substantial denaturation of the protein after its immobilization on the Au(111) surface, since no voltammetric signal is observed at negative redox potential values. To complement current-distance spectroscopy measurements, we studied current blinks in the static break-junction approach52,53 to measure the spontaneous interaction between protein partners (Fig. 2). By setting the current set point to 0.30 −0.35 nA, and transiently disconnecting the feedback loop at a constant bias of 0.8 V, we recorded current vs. time (I-t)tracesonCc 1 -Ccand Cc 1 -Y48pCMF Ccinteraction (Fig. 2a, red and blue, respectively). In the I-t curves, sudden jumps appear in the measured current (blinks), which are attributed to the spontaneous contact between the probe and the sample. The collected blink currents (I blink ), after subtracting thebaselinecurrent(I baseline )were set to a common time origin (initial blinking time = 0) and plotted with the blinking lifetime in 2D-histograms (Fig. 2b). These 2D-blinking maps show the conductance dispersion of the partners’interaction. Taking the most probable conductance peak, we found that Cc 1 -Y48pCMF Ccinteraction led to a significantly higher conductance (P < 0.05) (Fig. 2c). Gaussian fitting of main conductance peaks yields G=(4.3·10 −6±9·10 −7)·G 0 for Cc 1 -Ccand G = (7.4·10−6±5·10 −7)·G 0 for Cc 1 - Y48pCMF Cc(with G 0 =2e 2/h). In Fig. 2d, I blink is plotted against its corresponding I baseline showing that most probable Cc 1 -Ccinteraction (G = 4.3·10−6±9·10 −7)·G 0 can occur below the set point current (dashed line), thereby at longer distances than with Y48pCMF Cc,inagreement with I-z measurements. CcTyr48 phosphorylation strengthens the Cc-Cc 1 interaction Since EC-STM results suggest that Ccphosphorylation alters the interface between the two proteins, we further evaluated the interprotein interaction using bulk and single molecule measurements. We assessed the effect of Tyr48 replacement by the phosphomimetic pCMF on the interaction strength between Ccand Cc 1 by analyzing the forced unbinding between human Ccand the soluble domain of plant cytochrome Cc 1 using single molecule force spectroscopy with an atomic force microscope (AFM)54. In addition, binding affinities of the Cc 1 -Cccomplex were calculated from solution-based surface plasmon resonance (SPR) ensemble measurements, and the impact on Cbc 1 complex (complex III) activity was assessed in mammalian mitochondrial extracts. To carry out single-molecule unbinding experiments, Cc 1 was immobilized on Au(111) substrates in a similar way as described for the EC-STM experiments. Either Ccor Y48pCMF Cc,bothwiththe C-terminal mutation E104C, were conjugated to the maleimide group of a PEG 27 heterobifunctional crosslinker previously anchored to amino-functionalized AFM silicon probes. In the force spectroscopy measurements, the tip and substrate are approached and forced into contact up to a maximum force of 200 pN and then retracted away from the surface at constant velocity, in 50 mM phosphate buffer pH 6.5. The force-separation retraction traces at 1 μm·s−1velocity occasionally displayed unbinding events like those in Fig. 3a, which allows quantifying the force (F r )andlength(l r ) (see representative example in Fig. 3a inset) at which the contact is ruptured. The fact that the rupture length is maintained around 10–20 nm (Fig. 3b) indicates that protein unfolding is not significant, considering that the linker length is about 8 nm (monomer length of approx. 0.3 nm)55,56, and the measured distances may be due to deformation and elongation of the proteins prior to dissociation. No correlation is found between F r and l r for neither Cc 1 -Ccnor Cc 1 -Y48pCMF Cc(Fig. 3b). For the pair Cc 1 -Cc,F r is centered at 55 ± 17 pN (fitting a gaussian distribution) with a mean value of 75 pN (Fig. 3b, c). These relatively low rupture forces support the idea of the transient nature of the Cc 1 -Ccinteraction. In contrast, Cc 1 -Y48pCMF Cc display higher unbinding forces (P< 0.05), centered at 77 ± 24 pN (gaussian fit) with a mean value of 91 pN (Fig. 3b, c). We complemented these results with force-mediated dissociation kinetics studies. The lifetime of a bond sustained by weak noncovalent interactions decreases when subjected to force, which is conceptualized as force causing a tilt on the energy landscape, lowering energy barriers57. The unbinding force generally depends on the loading rate, as theory predicts for many biological interactions like receptor-ligands bonds58. In that case, the measured unbinding force corresponds to an individual point in a continuous spectrum of bond strengths. Therefore, we measured the rupture force at varying retraction velocities (from 0.5 to 4 μm·s−1). To avoid transmission of the load through the polymer linker and bring high inaccuracy to the loading rate values, either Ccor Y48pCMF Ccwere immobilized directly onto Au-coated probes through the thiol-Au chemistry. Here, the protein immobilization and orientation are comparable to those of theECSTMexperiments.Thetwoproteinsarethenforcedintocontact and F r and l r are evaluated at different retraction velocities. These measurements only carry direct information of the kinetics of unbinding (not binding). The force of Cc 1 -Ccunbinding is constant for the pulling velocities studied (contour maps F r vs.l r ,Fig.3d). In contrast, Cc 1 - Y48pCMF Ccunbinding becomes force resistant, and the unbinding force scales logarithmically with the pulling velocity and the loading rate (Fig. 3d). Cc 1 -Ccdissociation occurs at low forces independently of the pulling velocity (mean F r vs velocity in log scale, Fig. 3e). This behavior is associated with two bodies being pulled apart during a near-equilibrium mode59–61, in which complexes may have short lifetime and unbinding-rebinding is faster than pulling. This allows quick dissociation of the complex at biologically accessible loading rates and forces, in accordance with the requirement of a rapid turnover. In contrast, the rupture force between Cc 1 -Y48pCMF Ccdepends logarithmically on the loading rate (r) (Fig. 3f), meaning that the measurements are done in a Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 4 kinetic regime (non-equilibrium bond rupture), as predicted by the Bell-Evans model based on a single transition state theory57,59,60,62. The most probable unbinding force plotted against the loading rate on a log scale results in a straight line. Kinetic parameters can be obtained assuming that the energy barrier observed in dynamic force spectroscopy (DFS) remains rate limiting below the loading rate range tested. Following the Bell-Evans model57,59,62, the rate of the molecular complex force-activated unbinding (ko u ) and the distance from the bound state to the barrier that is rate limitng for unbinding (x u ) can be calculated from a linear fittothedataonlyfor the Cc 1 -Y48pCMF Ccpair (ko u =48s; −1x u = 0.31 nm). Overall, the DFS results indicate that the Y48pCMF mutation in Ccstrengthens the interaction with Cc 1 , thereby hindering a fast turnover. Cc 1 -Ccbinding affinities were calculated from solution-based surface plasmon resonance (SPR) ensemble measurements. Protein binding affinities are commonly measured and described by the 910 -6 0 Current (nA) Time (s) Time (s) a b 0.0 8.1 10-6 1.6 10-5 4.3 10-6 Conductance (G0) 0.0 0.1 0.2 0.3 Blinking lifetime (s) 0.0 0.5 1.0 7.410-6 Iblink (nA) 0.0 0.1 0.2 0.3 Blinking lifetime (s) 0.0 8.1 10-6 1.6 10-5 Conductance (G0) 0100 c Conductance (G0) 610 -6 310 -6 0 CcY48pCMF Cc *d Iblink (nA) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Ibaseline (nA) 0 2 4 6 80.3 nA 0.35 nA 0 3.2 10-5 6.5 10-5 9.7 10-5 1.3 10-4 Conductance (G0) . . . . . . . ........ . Fig. 2 | Blinking experiments. a Representative blinking recordings for Cc 1 -Cc (red) and Cc 1 -Y48pCMF Cc(blue). Blinks are shown as transient jumps of the current flowing between the two electrodes at a constant distance (imposed by initial current set point) and bias (800 mV). bTwo-dimensional histograms of blink current (I blink ) and blink lifetime for Cc 1 -Cc(left, n = 100980 conductance values included) and for Cc 1 -Y48pCMF Cc(right, n = 252193 conductance values included). Counts have been normalized to a color scale, with 100 counts representing the maximum, and 0 representing the minimum. The corresponding conductance (G) profiles are shown. Baseline correction was applied in a and b. G = I blink /V bias is used to obtain the conductance values. G 0 =2e 2/h = 77.5 μS, with hthe Planck constant and ethe elementary charge. cPlot of the averaged G values from gaussian fit of the most probable conductance peaks (mean ± s.d.) G = 4.3·10−6±9·10 −7G 0 for Cc 1 -Cc(red) and G = 7.4·10−6±5·10 −7G 0 for Cc 1 -Y48pCMF Cc(blue), respectively, showing significant differences (n = 6 conductance values from 2 independent experiments, two sample t-test, t Statistic = 2.777, degrees of freedom = 10.00, *P = 0.0195. dPlot of I blink against its baseline current (I baseline )for a set of blinks of Cc 1 -Cc(red) and Cc 1 -Y48pCMF Cc(blue) (mean ± s.d.). The dashed vertical lines indicate the corresponding current set points at which feedback was turned off. Sample and probe electrodes are represented by a square and a triangle, respectively. All experiments were performed in 50 mM phosphate buffer, pH 6.5. Source data are provided as a Source Data file. Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 5 equilibrium dissociation constant (K D ),whichevaluatesthestrengthof biomolecular interactions (i.e., the greater the K D value, the weaker the attraction between both protein partners). Fitting of the SPR curves for the interaction between Cc(WT and Y48pCMF) and Cc 1 to a 1:1 binding model (Fig. 3g) allows to obtain the association (k on )anddissociation (k off ) rates for the interaction (Table 1). The calculated association rate k on of Cc 1 -Y48pCMF Cccomplex is nearly 2-fold larger than that of the Cc 1 -Cccomplex, whereas the differences between spontaneous offrate k off values are small (but statistically significant, P < 0.05). Differences found between ko u values determined in single-molecule DFS experiments and spontaneous off-rates k off values determined by SPR have been attributed to the complexity of the energy landscape that may cause high differences between the forced unbinding mechanism and the thermally activated one63,64. The differences in K D (k off/ k on ) values between Ccand Y48pCMF Ccin complex with Cc 1 go along with the single molecule force spectroscopy results, and agree with published NMR and ITC measurements42, demonstrating a stronger interaction force between Cc 1 -Ccwhen the Tyr48 residue is replaced by pCMF. We further investigated if the observed differences in the unbinding forces and K D values could modulate the mitochondrial complex III activity (Supplementary Fig. 4). The biochemical assay to study the complex III activity relies on following the changes in Ccredox state from oxidized to reduced, by recording the absorption spectrum of Ccat 550 nm. Our results with mammalian mitochondrial extracts showed that Y48pCMF Ccdisplays a 1.6- fold reduction (P< 0.001) of the net complex III activity (0.28 nmol·min−1μg−1vs.0.17nmol·min −1·μg−1for WT and Y48pCMF Cc, respectively). Such results support the hypothesis that the observed differences in unbinding forces and K D values could be physiologically relevant in the modulation of the ET behavior of Tyr48-phosphorylated Cc. Fig. 3 | Unbinding between Cc 1 -Ccstudied by AFM-FS single molecule and surface plasmon resonance (SPR) ensemble measurements. a Representative force-separation unbinding events upon tip retraction at 1 μm·s−1for Cc 1 -Cc(red) and Cc 1 -Y48pCMF Cc(blue), when Cc 1 is directly anchored to Au(111) and Ccis anchored to the AFM tip through a PEG 27 linker. The inset shows the parameters F r and l r , that are collected from each curve and displayed in bas F r vs.l r with the corresponding histograms in each axis. F r distributions are fitted to a Gaussian. cPlot of the averaged F r values (mean ± s.d.) for Cc 1 -Cc(n = 134 force-separation curves from 2 independent experiments, red) and Cc 1 -Y48pCMF Cc(n = 88 forceseparation curves from 3 independent experiments, blue), showing significant differences (two sample t-test, t Statistic = 2.34, degrees of freedom = 220, *P = 0.02). dF r vs.l r contour maps for the unbinding at different pulling rates (0.5 − 4.0 μm·s−1) for Cc 1 -Cc(top, n = 212, 185, 55 and 226 for 0.5, 1.0, 2.0 and 4.0 μm·s−1, respectively) and Cc 1 -Y48pCMF Cc(bottom, n = 194, 83, 72 and 97 for 0.5, 1.0, 2.0 and 4.0 μm·s−1,respectively),whenCc 1 is directly anchored to Au(111) and Ccis directly anchored to the Au-coated AFM tip. eMean F r vs. tip velocity (in log scale). fAverage F r (mean ± s.d.) vs. ln (loading rate) for the Cc 1 -Y48pCMF Ccunbinding fitted to a linear function (blue line) (n = 194, 83, 72 and 97, for 28000, 56000, 112000 and 224000 pΝ·s−1, respectively). Forced-activated unbinding parameters are calculated following the Bell-Evans model57,59,62. All AFM experiments were performed in 50 mM PBS, pH 6.5. gBinding kinetics determined from SPR measurements. SPR sensograms obtained with 5 µM (squares) and 2.5 µM (circles) of Cc (red) and Y48pCMF Cc(blue) analyte and immobilized Cc 1 . Lines correspond to a simultaneous, global fit of the two curves to a one-site binin model. Data is representative of a total of 16 spots. The binding assays were performed in 10 mM PBS, pH 6.5. Source data are provided as a Source Data file. Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 6 Ccphosphorylation disrupts the Gouy-Chapman conduit for long-distance ET with Cc 1 Long-distance currents between Ccand Cc 1 can be explained by a low ionic concentration in the aqueous gap between the two protein redox sites, compared to the solution bulk4. Cation depletion near the redoxactive sites impairs charge screening in the solution and allows the electric field to extend several nanometers within the diffuse ionic Gouy-Chapman layer. To account for the changes in βobserved when Tyr48 is replaced by the phosphomimetic pCMF in the EC-STM experiments, we performed molecular dynamics (MD) calculations of the pairs Cc 1 -Cc(WT) and Cc 1 -Y48pCMF Cc(phosphomimetic), as well as Cc 1 -pY48 Cc(phosphorylated Tyr48: O-phospho-L-tyrosine) with charge −1and−2. Note that the latter “true”phosphorylated Ccforms offer valuable insights as they are not yet experimentally accessible. The proteins were separated 3.2 nm with the redox active sites facing each other, in a solvent with explicit water molecules and 50 mM NaCl concentration (Supplementary Fig. 3a–d) Cation depletion near the redox-active sites in the aqueous gap between the proteins is observed for Cc 1 -Cc, and the concentration is nearly zero in the proximity of the redox active sites (Fig. 4a), in accordance with our previous studies4.Incontrast,forCc 1 -Y48pCMF Ccan accumulation of cations is found around the proteins and in the region between them (Fig. 4b, h), likely due to the extra negative charge at position 48 that alters the surface electrostatic potential. This is reflected in the computed electrostatic potential in a 50 mM ionic concentration, which shows equipotential lines that extend between Ccand Cc 1 proteins, forming a conduit (Fig. 4c), but this is reduced between Y48pCMF Ccand Cc 1 (Fig. 4d). For Cc 1 -Ccthe space between both heme groups had the lowest average concentration of sodium ions (0.018 M) in comparison to Cc 1 -Y48pCMF Cc(0.067 M). Thus, MD calculations show that the Y48pCMF mutation of Cccauses the disruption of the Gouy-Chapman conduit that has been associated to long-distance ET through the aqueous solution in agreement with EC-STM results. Interestingly, the phosphorylation of Tyr48 in native Ccmay shift the electrostatic field even more than the Y48pCMF mutation, as evidenced when thecomputedaveragesodiumconcentration (average sodium concentration of 0.044 M for charge −1 and 0.052 M for charge −2, Fig. 4h, Supplementary Fig. 3e, f) and electrostatic potentials are obtained for pY48 Ccwith charge −1and−2 (Figs. 4e, f, h). In both cases, the equipotential lines do not extend between Ccand Cc 1 . We also performed Brownian dynamics (BD) simulations to calculate binding kinetics using the MD structures of Ccand Y48pCMF Cc complexes with Cc 1 . For each computation, two curves are represented (Fig. 4i), corresponding to 4 independent pair-distances criteria for reaction, as reported previously65. We estimated association rate (k on ) values at a distance of 0.6 nm as reaction criterion (vertical line in Fig. 4i). Ccand Y48pCMF Ccspecies show different behaviors across the simulation volume (expressed as distance cut-off dependencies of observables): i) the long-range attraction between Y48pCMF Ccand Cc 1 is larger than between Ccand Cc 1 , and ii) below 12 nm, the estimated apparent k on values for Y48pCMF Ccdecay more steeply than in the case of Cc, probably due to bias of electrostatic steering. This matches the differences in force values found in DFS experiments, indicating that the Y48pCMF mutation strengthens the interaction with Cc 1 , as pointed out before. The ratio between the k on values decreases hyperbolically with the distance, reflecting the impact of changes in the energetic landscape of interaction. The rate constant k trigger was set at the cut-off distance (2.5 nm) at which ET can be triggered in the EC-STM experiments. At the contact level, reckoned k on for Cc 1 -Y48pCMF Ccare ca. 10-fold larger than those estimated for Cc 1 -Cc(Table 2,Fig.4i). At the cut-off distance of 2.5 nm, k trigger values increase 2-3-fold in the trajectories corresponding to Cc 1 -Y48pCMF Cc compared to Cc 1 -Cc. In summary, Cc 1 attracts Y48pCMF Ccstronger than the WT Cc, even at long distances. Discussion Phosphorylation is one of the most prevalent mechanisms of regulation of protein activity. In particular, tyrosine phosphorylation plays a crucial role in cellular processes such as metabolism or cellular growth. Ccand its partner CcO are among the mitochondrial proteins whose function is highly regulated by phosphorylation. Ccphosphorylation has been suggested to perform a crucial role in regulating mitochondrial respiration and cell fate30. Tyr48 phosphorylation has been reported to affect the binding affinity of Cctowards its molecular partners, altering the diffusion pathway of Ccmolecules through Cbc 1 -CcO and affecting the ETC flux by partially inhibiting the mitochondrial respiration42.However,the molecular mechanism through which Ccphosphorylation impacts the ETC is unknown. The multimodal experiments on phosphomimetic Cc and calculations reported here show that upon phosphorylation the long-distance charge conduit established between Cc-Cc 1 is shut off and their interaction is strengthened and departed from binding/ unbinding equilibrium in the measured conditions. Overall, these results are in full agreement with the consensus evidence published on the topic and unveil a nanoscopic view and mechanistic framework that offers deeper insights into the interaction between redox protein partners of the respiratory ETC. Notably, these results indicate a regulatory advantage of using electrochemically gated, long-distance electron transport through the solution between redox partner proteins. Besides reconciling high specificity and electron transport efficacy with weak binding to keep high turnover rates4, the Gouy-Chapman conduit bearing reduced ionic density in the confined volume between the proteins can be externally controlled to reduce charge flow. When phosphorylated at Tyr48 (herein mimicked by replacement with pCMF), Ccsliding from Cbc 1 to CcO is impaired42. The stronger interaction between the partners upon phosphorylation was considered to hinder the Ccturnover (exchange with the partner) as it would lead to a frozen complex as concluded from calorimetry data and NMR spectroscopy results42. Note that our scanning probe microscopies experiments allow probing interactions at the single protein level, and in the case of EC-STM (I-z and blinking studies) it is possible to keep Cc 1 reduced, and Ccoxidized to mimic the physiological situation of ET from Cc 1 to Cc. Independent control of probe and sample electrochemical potential in the single molecule force spectroscopy AFM setup is more challenging66,67 and constitutes one aspect to improve experimentally in future studies. We observe that electrochemically gated long-range ET is impaired in single-molecule experiments with phosphomimetic Y48pCMF Cc, with current decay distances reducing to half (to 3 nm or less, Fig. 1a–c) and abolishing EC gate potential control (Fig. 1d). MD simulations show that the negative charge introduced by the pCMF residue on the Ccsurface prevents cation exclusion between the proteins’redox active sites, thereby enabling charge screening in this confined interface and disrupting the Gouy-Chapman conduit (Fig. 4). The effects are even stronger in simulations with phosphorylated residues, which bear high relevance as they are not yet experimentally accessible. Besides, we observe that the single molecule unbinding of Table 1 | Analysis of surface plasmon resonance (SPR) curves Spot Analyte k on (M−1s−1)k off (s−1)K D (M) Cc 1 Cc6.54·103± 8.8·1013.18·10−3±5·10 −55.27·10−7±1.4·10 −8 Y48pCMF Cc 1.00·104± 1.57·1023.58·10−3±6·10 −53.57·10−7±9·10 −9 Association rate (k on )andspontaneousoff-rate(k off ) constants and dissociation equilibrium constants (K D )werecalculatedfromthefits of SPR curves to 1:1 binding model. The errors reported are the standard errors derived from the fitting of the experimental data to a 1:1 binding model. R2are 0.9975 and 0.9954 for WT and Y48pCMF Ccfitting, respectively. Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 7 Y48pCMF Ccfrom the redox partner Cc 1 occurs at higher forces (Fig. 3). These results are consistent with the higher binding affinities of phosphomimetic Cccompared to the wildtype in bulk experiments (Fig. 3g, Table 1) and BD calculations (Fig. 4i, Table 2). Figure 5recapitulates all these results and provides a mechanistic framework showing how phosphorylation of Ccalters several functions of this multifaceted protein involved in electron transport and cell fate. First, by shutting down its ability to exchange charge with pY48 Cc(q-1) - Cc1pY48 Cc(q-2) - Cc1 Cc - Cc1 Y48pCMF Cc - Cc1 Y48pCMF Cc - Cc1 Cc - Cc1 ab cd ef Cc [ Na+](mM) Cc Y48 pCMF Cc pY48 Cc(q -1 ) pY48 Cc (q -2 ) 0 50 100 ghi [Na + ] (M) U (kT/e) Cc1 CcCc1 0102030 10 6 10 7 10 8 10 9 10 10 kon( M -1 s-1) Cut-off (Å) Fig. 4 | Molecular dynamics (MD) calculations of Cc 1 and Ccdisplay an interprotein Gouy-Chapman conduit that is disrupted by phosphorylation. a,bSide view of the averaged sodium concentration map (M) at the X-axis plane dissecting both Cc 1 -Ccredox active sites (Ccand Y48pCMF Cc, respectively) facing each other at 3.2 nm separation taking the most external hydrogen atoms from the -CH 3 distal pairs from each heme group, that is 4.2 nm taking the Fe-to-Fe distance. MD simulations were performed with explicit water (TIP3P solvent model) and 50 mM ionic concentration (NaCl) over 36 ns. Each contour line displays an increase of 0.15 M. (c–f) Side view of computed electrostatic potential with APBS76 from –1kT/e (red) to 1 kT/e (blue) for Cc 1 -Ccpairs: Cc 1 -Cc(c),Cc 1 -Y48pCMF Ccin silico (d),Cc 1 - pY48 Cc(O-phospho-L-tyrosine) with charge –1(e) and Cc 1 -pY48 Cc(O-phospho-L- tyrosine) with charge –2(f). Ion concentration is 50mM, pH 6.5. Each contour line displays an increase of 0.2 kT/e. gVirtualization of the Cc 1 -Ccsystem in 3D. Representation of the prism (blue) used to calculate the average concentration of sodium ions throughout the simulation. The image shows a single frame of the Cc 1 -Ccsystem depicting protein (cyan, translucent grey surface), heme groups (white with Iron atom in red), sodium ions (green spheres), and omitting water molecules. The closest backbone residues to the generated prism vertices are His18, Pro76, Lys79 and Ile85 of the Y48pCMF Cc, and Leu134, Leu94, Glu101 and Ala160 of Cc 1 .hAverage sodium concentration over 36 ns in a 12.9 nm3volume prism (prism dimensions for x, y, z = 15, 43, 20 units) centered between both heme groups of Ccand Cc 1 for all systems. Volmap Tool Plugin 1.1 from VMD75 was used. iDerivation of binding kinetics from the BD computations of Cc 1 -Cc(red) and Cc 1 - Y48pCMF Cc(blue) complexes. For each computation, two curves are represented, corresponding to 4 independent pair-distances criteria for reaction, as reported previously65. Vertical lines correspond to the distances used as reaction criterion to estimate k on and k trigger values. Source data are provided as a Source Data file. Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 8 Cc 1 (complex III) at long distances through the aqueous solution and in an electrochemically selective manner4. And second, by promoting a stronger contact with Cc 1 that disturbs the reversible binding found in equilibrium conditions (low forces) and likely reduces the turnover of molecules at complex III. In this scenario, even small amounts of phosphorylated Ccwill lead to one of them being stuck at complex III and effectively preventing others from transferring charge. The dissociation constant (K D )islowerfortheCcY48pCMF variant, which must get closer to complex III than the wildtype in order to achieve similar current levels (due to the higher βof the former, see Fig. 1a). In this situation, the interaction forces (unbinding force, Fig. 3;binding affinity, Table 1and Fig. 3g; association rate, Table 2and Fig. 4i) are higher and thereby the overall turnover of Ccmolecules is reduced, lowering the complex III activity (Supplementary Fig. 4). The higher conductance observed in the Y48pCMF variant compared to the WT species (Fig. 2c) might be caused by such closer interaction. In conclusion, we have performed a multimodal study of the interaction between redox partners Ccand Cc 1 that covers single protein conductance and force measurements, bulk binding and biochemical assays, and molecular dynamics calculations. By studying a phosphomimetic mutation, we have focused on the impact of phosphorylation in all the observables and integrated the obtained results within the available biochemical evidence, aiming to provide a nanoscopic view and mechanistic framework that was lacking in this topic. Phosphorylation of Ccimpairs electrochemically gated long-distance electron transport through the solution and disturbs the binding/unbinding equilibrium between the partners, strengthening their interaction. Calculations show that the cation exclusion volume that is established between the partners (Gouy-Chapman conduit) is disrupted by the negative charges introduced near the Ccredox site by phosphorylation and its experimental molecular mimics. These results allow a deeper understanding of the tradeoff between binding, transfer, and turnover that is established between redox protein partners in the respiratory ETC and provide mechanistic insights into the molecular sites that regulate cell signaling circuits. Methods Protein expression and purification Escherichia coli (E. coli) BL21 (DE3) cells were transformed with pBTR1-WT plasmid—that contained the CYCS gene-coding for human Ccand, the CYC3 gene of the yeast Ccheme lyase, required for the maturation of Cc—to recombinantly express Homo sapiens WT Ccfor Surface Plasmon Resonance (SPR) experiments. For the expression of E104C mutant, used for Electrochemical Scanning Tunneling Microscope (EC-STM) and Atomic Force Microscopy-Force Spectroscopy (AFM-FS) measurements, pBTR-1 plasmid was mutated by replacing the GAA triplet corresponding to Glu104 by TGT, which codifies to cysteine residue. The primers for PCR were pBTR1-E104C fw (5- AAGGCGACGAACTGTTGATAAGGTACCA-3) and pBTR1-E104C rv (5- TGGTACCTTATCAACAGTTCGTCGCCTT-3). For this purpose, onestep mutagenic PCR with Accusure™DNA Polymerase (Bioline) was used following the manufacturer’s instructions. The new plasmid, containing the mutated sequence, is designated as pBTR1-E104C. Expression and purification of both Ccspecies was carried out as previously described68.Briefly, transformed bacteria were incubated for 20 h at 150 rpm and 30 °C in 2.5 L standard lysogeny broth (LB) rich medium (10 g·L−1tryptone, 5 g·L−1yeast extract and 10 g·L−1NaCl), supplemented with 100 μg·mL−1ampicillin. Cells were harvested via centrifugation at 5000 x g for 10 min at 4 °C, resuspended in 10 mM tricine buffer pH 8.5 supplemented with cOmplete protease inhibitor cocktail (#11836145001, Roche), ruptured by sonication for 5 min, and then centrifuged at 14,000 x g for 60 min. The supernatant was loaded onto a Nuvia S column (#7324720, Biorad) and eluted with a 25 to 500 mM NaCl gradient in 10 mM tricine buffer pH 8.5. Phosphomimetic Y48pCMF Ccwas expressed using the evolved tRNA synthetase technique, as previously described41. Briefly, E. coli BL21 (DE3) cells were co-transformed with either pBRTR1-Y48AMBER or pBTR1-Y48AMBER/E104C, and pEVOL/pCMF/tRNA plasmids—which contains the unnatural tRNA/aminoacyl-tRNA synthetase pair that recognizes the pCMF—and were cultured for 20 h at 150 rpm and 30 °C in M9 minimal medium supplemented with 100 μg·mL−1ampi- cillin, 20 μg·mL−1chloramphenicol, 16.7 μg·mL−1δ-aminolevulinic acid hydrochloride and 263 μg·mL−1non-canonical amino acid pCMF. Protein expression was induced with 1 mM isopropyl-β-D- thiogalactoside and 0.02% arabinose. Cells were harvested by centrifugation and purified similarly as previously described for Cc41. The soluble domain of Arabidopsis thaliana Cc 1 was expressed and purified as previously described12. Briefly, E. coli BL21 (DE3) cells were cotransformed with pET-Cc 1 and pEC86 plasmids (which constitutively express the eight cytochrome cmaturation genes ccmABCDEFGH)to produce the soluble domain of Cc 1 (residues 64–265, Gene ID 834081). The cells were grown for 24 h at 150 rpm and 30 °C in 2 L of standard LB rich medium, supplemented with 50 μg·mL−1kanamycin, 12 μg·mL−1chloramphenicol and 16.7 μg·mL−1δ-aminolevulinic acid hydrochloride. Harvested cells were resuspended in 20 mM Tris-HCl buffer pH 8.0 supplemented with cOmplete protease inhibitor cocktail (#11836145001, Roche) and physically ruptured by sonication for 5 min. The supernatant was loaded onto a Nuvia Q column (#7324721, Biorad) and eluted with a 25 to 500 mM NaCl gradient in 20 mM Tris-HCl buffer pH 8.0. Fractions containing Cc 1 were further purified by size exclusion with a ENrich SEC 70 column (#7801070, Biorad). All protein samples were dialyzed against 10 mM sodium phosphate buffer at pH 6.5. Electrochemical scanning tunneling microscopy-spectroscopy Sample preparation. Atomically flat Au(111) single-crystal disks (10 mm diameter and 1 mm thickness, MaTecK) were flame annealed and electrochemically polished45. The Au(111) electrodes were incubated with Cc 1 in sodium phosphate buffer 10 mM, pH = 6.5 at a 56 μM concentration. The EC-STM probes (Au wire 0.25 mm in diameter; GoodFellow) were insulated with ApiezonTM45, and were incubated with a12.5μMsolutionofeitherCc(E104C) or Y48pCMF Cc(E104C), in sodium phosphate buffer 10 mM, pH = 6.5. All incubations were conducted overnight and at 4 °C. EC-STM measurements. All experiments were performed with a PicoSPM microscope head and a PicoStat bipotentiostat (Agilent, USA) controlled by Dulcinea electronics (Nanotec Electronica, Spain) using the WSxM 4.13 software. A custom-made electrochemical liquid cell with a standard sample plate was used in fourelectrode configuration with the STM Au(111) sample and probe as working electrodes, a 0.25 mm diameter Pt80/Ir20 wire as counter electrode and a miniaturized ultralow leakage membrane Ag/AgCl (SSC) reference electrode filled with 3 M KCl. The potentials of the Au(111) electrode sample (U S ) and EC-STM probe (U P )were expressed against this reference. The electrochemical cell was cleaned with piranha solution (3:1 v/v solution of H 2 SO 4 and H 2 O 2 ) before each measurement. Caution: piranha solution is a strong oxidizer and a strong acid. It should be handled with extreme care, as it reacts violently with most organic materials. The experiments Table 2 | Analysis of brownian dynamics (BD) simulations for the Cc 1 -Ccinteraction Cc 1 -CcCc 1 -Y48pCMF Cc k on (M−1s−1)(0.6nm) 9.50·106±7.90·10 62.10·108±9.6·10 7 k trigger (M−1s−1) (2.5 nm) 3.20·109±8.6·10 76.80·109± 1.4·108 Estimated association rate at contact level (k on ) and at the cut-off distance of 2.5 nm (k trigger ). The errors reported correspond to the standard deviation. Article https://doi.org/10.1038/s41467-022-34809-1 Nature Communications | (2022) 13:7100 9