Protein Surface Mimetics: Understanding How Ruthenium Tris(Bipyridines) Interact with Proteins

Abstract Protein surface mimetics achieve high‐affinity binding by exploiting a scaffold to project binding groups over a large area of solvent‐exposed protein surface to make multiple cooperative noncovalent interactions. Such recognition is a prerequisite for competitive/orthosteric inhibition of protein–protein interactions (PPIs). This paper describes biophysical and structural studies on ruthenium(II) tris(bipyridine) surface mimetics that recognize cytochrome (cyt) c and inhibit the cyt c/cyt c peroxidase (CCP) PPI. Binding is electrostatically driven, with enhanced affinity achieved through enthalpic contributions thought to arise from the ability of the surface mimetics to make a greater number of noncovalent interactions than CCP with surface‐exposed basic residues on cyt c. High‐field natural abundance 1H,15N HSQC NMR experiments are consistent with surface mimetics binding to cyt c in similar manner to CCP. This provides a framework for understanding recognition of proteins by supramolecular receptors and informing the design of ligands superior to the protein partners upon which they are inspired.

ing or any structurali nformationa re lacking;t his is characteristic of all but af ew studies on protein surface recognition by using classic supramolecular scaffolds. [18,45,46] Inhibited ascorbate reduction of cyt c [36,37] is consistentw ith binding to the cyt cp eroxidase (CCP) binding site:t hat is, the haem-exposed edge of cyt c,w here there is ah ydrophobic patch surrounded by ar ing of basic amino acid residues. [47] Here we show that highly functionalized Ru II (bpy) 3 complexes inhibitt he cyt c/CCP interaction and do so through electrostatically and entropically driven binding of cyt c in am anner that replicates the binding of cyt c by CCP.H igher-affinity Ru II (bpy) 3 complexes achieve additional potency through enthalpic effects. Finally,b yu sing high-field NMR we demonstrate that recognition occurs at the haem-exposed edge and hence that PPI inhibition is orthosteric. Collectively,t his provides am ore rational framework for the design of supramolecular receptors for cyt c and for protein surfaces more widely.

Results and Discussion
Synthesis Ru II (bpy) 3 synthesis proceeded by the route shown in Scheme1,w ith use of a tert-butyl ester or methyl ester protectingg roup strategy for complex 1 or 2,r espectively.I nt his genericr oute, the ligand is first assembled by amide bond formation, via aw ater-sensitivea cidc hloride, with subsequent complexation using Wilkinson'sr eagent. [48] The protected complex formed can be purified by conventionals ilica flash columnc hromatography.S ubsequent deprotection with trifluoroacetic acid( TFA) or lithium hydroxide affords complexes 1 and 2,r espectively.D eprotectiono ft he larger complex 2 requires mild conditions and careful reaction monitoring due to the lability of the anilide bond under both basic and acidic conditions.
Complex 2i nhibits the cyt c/CCP PPI Given that the affinity of complex 2 for cyt c that we previously reported [36] is greatert han that of CCP for cyt c [49] we anticipated that 2 would be ap otent inhibitor of the cyt c/CCP interaction. Al uminescence quenching assay was implemented ( Figure 2): the luminescence emission from Zn-protoporphyrinsubstituted CCP [50] is first quenched upon interaction with cyt c and then recovered upon displacementw ith the ruthenium complex. Signalo verlap with the Ru II (bpy) 3 luminescence (l max % 625 nm) complicates interpretation;h owever,s imultaneous  [34] and C) the interaction faces of cyt c (left) and CCP (right), showing ar ing (red circle) of basicamino acid residues( blue) on cyt c and acomplementary patch (blue circle) with acidic amino acid residues (red) on CCP.
loss of MLCT luminescence relative to the complex in the absence of cyt c is observed. An ative agarose gel indicated successful PPI inhibition (see Figure S1 in the SupportingI nformation).

Bindingise ntropically favourable and electrostatic in nature
The bindinga ffinities of complexes 1 and 2 towards cyt c were measured by means of al uminescence quenching assay, [36] in which the luminescence of the ruthenium complexes is quenched on bindingt oc yt c through photoinduced electron transfer to its haem group. Previously,c uvette-based fluorescence wasu sed for binding studies; [36,37] however,o ptimization of the assay on a3 84-wellp late wasr equired for higherthroughput screening of the binding under different conditions. Addition of ab locking agent-bovines erum albumin (BSA)-was found to be required to allow for agreement between the two methods. The addition of BSA accompanied a concurrent decrease in binding affinity (from K d (10.5 AE 0.4) nm to (42.9 AE 3.1) nm for complex 2,F igure S1). Determination of the K d at different temperatures and subsequent van't Hoff analyses ( Figure 3A)p rovided thermodynamicp arameters (Table 1) for binding [Eq. (1)],w ith the assumption that DH and DS are temperature independent These data show that, for complex 1,b inding to cyt c is primarily driven by entropic contributions with as mall favourable enthalpic contribution, whereas for complex 2 it is both entropicallya nd enthalpically driven.I nc omparison, the cyt c/ CCP interaction is entropically driven, and enthalpically is mildly unfavourable. [49] Thus, complex 1,w ith fewerc arboxylate moieties, more closely matches the thermodynamic profile of CCP in binding to cyt c.Aplausible hypothesis for the enhanced binding of complex 2 to cyt c is that the additional carboxylic acids form increased numberso fs alt bridges with the basic amino acids on the cyt c surface.
To aid further understanding of the electrostatic contribution to binding, affinitiesw ere determined at different ionic strengths( I). Cyt c binding by both complexes 1 and 2 is highly dependentu pon ionic strength (Table 2), with binding affinity decreasing with increasing ionic strength, suggesting that electrostaticsd ominate binding. The K d values could be fitted to the Debye-Hückel relationship [Eq. (2), Figure 3B], in this case with use of aG üntelberg approximation [Eq. (3)], which is valid up to I = 100 mm.
m % 00 p I=ð1 þ p IÞð 3Þ   Table 1. Thermodynamic parametersd erived from the van't Hoff analysis for the binding of complexes 1 and 2 to cyt c (errors derived from triplicate experiments), togetherw ith literaturev aluesf or the cyt c/CCP interaction under similar conditions. [49] Complex From this relationship the parameters K 0 d and Z 1 Z 2 can be established ( Table 3), providing an estimate of the affinity at I = 0 and the product of the interacting positive and negative charges. The data wereconsistent with the Güntelberg approximation for both complexes( Figure 3B), and the calculated values of K 0 d show high-affinityb inding for complex 2 and weaker binding for complex 1 at zeroi onic strength.T he product, Z 1 Z 2 ,p rovides an indication of the chargesi nvolved in the interaction, with complex 2 having alarger value than complex 1 andC CP.F rom these data, the chargeo nt he complex interacting with cyt c can be estimated.Arudimentary interpretation of this date is made possible by assuming that cyt c has the same charge in all cases (calculated to be % 6a tp H7.5); [51] the charges on complexes 1, 2 and CCP can thus be calculated as 4.3, 5.9 and 4.8, respectively.
Complex 1 and CCP have relativelys imilarc harges, and this suggestst hey make similar electrostatic interactions with cyt c. Complex 2 has al arger charge,i ndicating increased electrostatic interactions with cyt c.T his is consistent with the van't Hoff analyses.A ccounting for the crudenesso ft he Debye-Hückel approximation,i nw hich small ( % 3 ), evenlyd ispersed chargesa re assumed( even when using the Güntelberge xtension), the data indicatet hat perhaps not all carboxylate moieties are deprotonated under the assay conditions (i.e.,p H7.5) and/or that al imited number of carboxylate moieties are neededf or productive protein surfacer ecognition (even fewer than the number identified in the "deletion" study by the Ohkanda group using heteroleptic complexes). [41] Differences in affinity between cyt c and complex 2 were also studied in different buffers (Table 4). Variation in affinity might discriminate between different contributions to binding because negatively chargeda nions must be displaced from cyt c and positivelyc harged cations from complex 2.I np otassium and sodium phosphate no differencei na ffinity between complex 2 andc yt c is observed, thus indicating that interactions of the cationic buffer components with complex 2 are not significant. For binding of cyt c to complex 2 in phosphate or sulfonica cid buffers (MOPS and HEPES), similar affinities are also observed. This suggests that the nature of the anion and, more importantly,t he hydrophobicity of the buffer are not significant in mediating molecular recognition, and reinforce the conclusions gleaned from Debye-Hückel analysis that the interaction is dominated by electrostatic contributions. For the Tris buffers (Tris and Bis-Tris propane (btp)) as mall decrease in binding affinity is observed. Although ad ifferenceinbehaviour due to the chloride counter anion cannotb ee xcluded,t his mightb ed ue to the ability of btp and Tris to participate in differenti nteractions with both cyt c and complex 2;i na ddition to the ammonium function, the hydroxy groups on the buffer mightm akec helating hydrogen bondsw ith charged residues on either.
Cyt c is as table protein thatd oes not unfold over aw ide range of pH values;h owever,i ts ionization state is affected by pH, [53] so the pH of the solution was expected to affect recognition of cyt c by Ru II (bpy) 3 complexes. To investigate the binding affinity between complex 2 and cyt c over ab road pH regime, btp was used, because it allows for ap Hr ange of 6.5-9.5. The affinity follows an inverted bell-shaped profile (Figure 4A), whichm aps reasonably wello nto the ionization state of cyt c (Figure4A, inset). [53] The affinity between pH 7.0-8.5 is relativelyc onstantw ith decreased binding observed at pH 6.5 and pH 9.0. Residues that become protonated/deprotonated in this pH regime are His33 and Lys79, respectively. [53] Lys79 (green) is at the haem-exposede dge ( Figure 4B)w here binding of complex 2 is thought to occur,w hereas His33 (pink) is on the distalface of cyt c.
An umber of reasons for ad ecrease in binding affinity at this pH are possible:1 )Complex 2 might bind to ad ifferento r multiple sites on cyt c,2 )binding of complex 2 might cause subtle conformational changes that transmitt ot he distal face of cyt c,a ffecting the pK a of His33, 3) protonation of His33 might cause subtle conformational changes that affect binding interactions on the haem-exposed edge, or 4) the protonation state of complex 2 might be changed at pH 6.5. More careful analysiso ft he pH-K d /ionization state profiles reveals ad iscrepancy.T he ionization state of cyt c drops at pH 8.0 rather than pH 8.5, at which the binding diminishes, thus suggesting that binding of complex 2 might mask Lys79 and increasei ts pK a . In contrast, there is no difference in the profiles for K d and ionization state of cyt c in the lower pH range, thus suggesting that the pK a of His33 is not affected by binding and that loss of affinity more likely originates from ac hange in ionization state on complex 2. High-field NMR reveals that the complexes1and 2bind to the CCP bindings ite on cyt c Although the pH data provide somec rude structural information on the cyt c binding site of Ru II (bpy) 3 complexes, more detailed residue-specific, atomic-level data were sought. To identify the binding site of complex 1 and 2 on cyt c,asensitivityenhanced natural abundance 1 H, 15 NHSQCs pectrum of cyt c in the presence and in the absence of complex 1 was recorded, with a9 50 MHz NMRs pectrometer.S odium ascorbate (2 mm) was added to theb uffer,t or educe the iron in cyt c from paramagnetic Fe III to diamagnetic Fe II ,t hus minimizing its influence on the spectrum (i.e.,p aramagnetic line broadening). The binding of the complexes to cyt c forr educed versus oxidized cyt c is similar (forc omplex 2, K d = (92.4 AE 5.5) and (49.6 AE 13.3) nm,respectively,i n5m m phosphate, 2mm sodium ascorbate, 0.2 mg mL À1 BSA). The assignmento ft he 1 H, 15 NHSQC spectrum of horse heart cyt c has previously been achieved. [55] After addition of complex 1,t he NMR data showt hat several crosspeaks have disappeared, whereas others displayc hemical shift changes ranging from 0.015-0.05 ppm indicating the presence of proteinligand interactions ( Figure 5A and B). When these chemical shift changes are mapped onto the structure of cyt c from the cyt c·CCP crystal structure, [54] the data indicate that binding occurs predominantly to one side of the haem group, with the opposite face havingv ery few amino acids with sizeables hifts in their HSQC peaks ( Figure 5C). Theb inding site is in al ocation similar to that of carboxylic-acid-functionalized porphyr-ins. [18] In comparison with the cyt c/CCP interaction ( Figure 5D), it can be seen that the amino acids for which cross-peaks have shifteda re in and around the PPI interface, thus indicating that complex 1 is an effective mimic of CCP,b inding at the same face and capable of acting as an orthosterici nhibitor of the interaction.
Attempts to acquire data in the presence of complex 2 were difficult, due to the high-affinityb inding and the relatively high concentrationsr equired for natural-abundance NMR. At 1:1r atios of cyt c and complex 2,d ata could not be obtained, due to the formation of oligomers and ac oncomitant loss of NMR signal intensity,c aused by significant line broadening.W e found this unsurprising given the potential for aggregation at higher concentrations and the observation of additional binding modes in NMR studies with porphyrins. [18] Furthere vidence of an additional/alternative binding mode fort he larger complex 2 is given by the observation of as econd binding event for complex 2 with yeast cyt c,b ut as ingleb inding event for complex 1 ( Figure S2). Even at ac omplex 2/cyt c ratio of 1:2 multiple signals disappeared, so detailed information as to the binding site could not be gleaned;h owever,o ft he signals present,c hemical shift changes wered etected for regions of the protein backbonel ocated on the same binding face as for complex 1,and on the haem-exposed edge.

Conclusion
We have performed ad etailed study on the cyt c binding properties of two synthetic Ru II (bpy) 3 complexes 1 and 2.T he ruthenium complexes are potent ligandsf or selective protein surfacer ecognition of cyt c and capable of inhibiting the cyt c/ CCP PPI. Binding is shown to be entropically favourable and driven by complementary electrostatic interactions between the basic protein and acidic Ru II (bpy) 3 complexes. This profile is consistentw ith accurate mimicry of the cyt c-binding properties of CCP.H igher-affinity recognitiono ft he protein target can be achieved throught he addition of furthera cidic motifs on the Ru II (bpy) 3 complexes, allowing additional enthalpically favourable electrostatic interactions to occur.F inally,N MR experiments have established that the Ru II (bpy) 3 complexes 1 and 2 bind to the solvent-exposed cyt c surface, thus further underscoring the ability of the complexes to act as mimics of CCP and confirming an orthosteric mode of PPI inhibition. These studies highlight the value of detailed analyses of protein-surface recognition by supramolecular hosts in terms of rationalizing structure-function relationshipsa nd informing subsequentd esigns. Moreover, the conclusions of this study point to af uture need for syntheses/assembly of asymmetrically functionalized Ru II (bpy) 3 complexes to maximize productive protein-ligand contacts and selectivity of protein surfacer ecognition. This and the application of our approach to therapeutically attractive protein targetsw ill form the basis of future studies by our group.  [53] B) Cyt c structure (PDB ID:1 U75) [54] with residues that become protonated at pH 6.5 (His33:p urple) and 9.0 (Lys79:g reen).

Experimental Section
Synthesis:S ynthesis was adapted from the literature. [36] Ar epresentative synthesis of complex 1 is shown below.T he synthesis of complex 2 is described in the Supporting Information.
To exchange the haem for Zn porphyrin, the haem was removed by using the acid butanone method [2] with minor modifications. Haem CCP solution ( % 1mm)i np otassium phosphate buffer (100 mm)w as diluted with four volumes of ice-cold water.T he CCP solution was adjusted to 100 mm fluoride by addition of KF solution (1 m), breaking the haem-protein linkage and turning the solution green. The haem was removed by lowering the pH of the solution to pH 3.2-3.3 by dropwise addition of ice-cold HCl (0.1 m), with gentle stirring. The haem was then extracted by addition of an equal volume of ice-cold butanone, shaking for 30 sa nd centrifugation for 1min at 1000 g.T he brown layer was siphoned away, and the extraction was repeated until the aqueous layer became colourless. The resulting apoCCP solution was diluted with ah alf volume of cold water and dialysed against two or three changes of NaHCO 3 solution (10 mm). It was then dialysed against water,w ith the outer solution being changed every 2h until there was no more discernible butanone ( % 24 h), followed by dialysis into Tris·HCl (pH 7, 10 mm). A4:1 excess of porphyrin was dissolved in KOH (100 mm,2 00-500 mL) and diluted five to ten times with water.T he porphyrin solution was added to the protein solution, and the protein solution was titrated to pH 7.8 with KOH (100 mm). In the dark, the alkaline porphyrin solution was added dropwise with gentle stirring to apoCCP until an approximately twofold excess of porphyrin was present. The solution was allowed to stand at near pH 8f or 20-30 min and then brought to pH 6.5-7.0 by addition of monobasic potassium phosphate (1 m). The protein was exchange into potassium phosphate (pH 6.5, 25 mm)a nd concentrated by ultracentrifugation to 0.5-1.0 mm CPP.T he protein was loaded onto as mall column of DEAE Sepharose CL-6B, preequilibrated with potassium phosphate (pH 6.5, 25 mm). The column was rinsed with around half av olume of loading buffer, and the metalloporphyrin CCP was eluted with potassium phosphate (pH 6.5, 0.6 m).
Luminescence quenching assays:A ll stocks for luminescence intensity assays were made up in phosphate buffer (pH 7.5, 5mm). Ruthenium complex stocks were made up to 2mm.H orse heart and yeast cyt c was obtained from Sigma-Aldrich and used without further purification. Ac ytochrome c stock was made up to % 1mm,a nd the concentration was accurately determined by using the molar extinction coefficients at 550 nm of 2.95 10 4 mol À1 dm 3 cm À1 for horse heart cyt c [56] and 2.11 10 4 mol À1 dm 3 cm À1 for yeast cyt c [56] after reduction by addition of one microspatula of sodium dithionite. Assays with oxidized cyt c in ascorbate-containing buffer used cyt c oxidized with K 3 Fe(CN) 6 followed by dialysis into buffer (pH 7.5, sodium phosphate (5 mm), sodium ascorbate (2 mm)) to remove the excess K 3 (CN) 6 .T he concentration of oxidized cyt c was determined by using the molar extinction coefficient at 410 nm of 1.061 10 5 mol À1 dm 3 cm À1 . [57] All buffers used were at 5mm concentration, pH 7.5, BSA (0.2 mg mL À1 )unless otherwise stated. taken in a4mL quartz cuvette with excitation at 467 nm, and emission was measured over the 575-675 nm range, with 10 nm slit widths on both excitation and emission. Peak maxima were recorded over the entire cyt c concentration gradient.
Plate reader luminescence quenching assays were performed by using aP erkinElmer EnVision 2103 MultiLabel plate reader,w ith excitation at 467 nm, and emission at 630 nm fixed wavelength. A 2/3 dilution regime in a3 84-well plate (Optiplate) was used (total well volume 50 mL), with each result measured in triplicate. The K d range possible for this assay is % 5nm-% 100 mm.
In all assays the ruthenium complex concentration was kept constant, with the concentration of cyt c being varied through the assay,a sd escribed below.R esults obtained were fitted, by use of Origin9, to a1:1 binding isotherm [Eq. (4)]: Where I = change in relative luminescence intensity (I/I 0 ), m = maximum value of I, a = concentration of complex, K = dissociation constant, and b = concentration of protein added.
Protein NMR:S ensitivity-enhanced 1 H, 15 NHSQC NMR correlation spectra of ligand-bound and unbound forms of cyt c,p urchased from Sigma-Aldrich, were carried out at natural abundance with a 950 MHz Bruker Ascend Aeon spectrometer operating at ap roton ( 1 H) resonance frequency of 950.13 MHz and equipped with a Bruker TCI triple-resonance cryoprobe. NMR acquisitions were carried out in buffer (pH 7.25, sodium phosphate (5 mm), sodium ascorbate (2 mm)). For cyt c alone, spectra were taken at 2mm protein concentration. With complex 1,c yt c (1 mm)a nd complex 1 (0.5 mm)w ere used, to at otal volume of 600 mL. Spectra were analysed with the aid of the CcpNmr Analysis software package, and the chemical shift perturbations were calculated as the square roots of the sums of the isotope-weighted shift differences squared [Eq. (5)] where Dd is the overall change in chemical shift, Dd N is the change in the nitrogen dimension, and Dd H is the change in the proton dimension. The change in the proton dimension is scaled by the ratio of the gyromagnetic ratios of 15 N( g N )a nd 1 H( g H )t o account for the larger chemical shift range of nitrogen.