Complexities of the interaction of Ni(II), Pd(II) and Pt(II) pyrrole-imine chelates with human serum albumin

Human serum albumin (HSA) is the most abundant blood protein and is responsible for the transport of many exogenous compounds, including clinically deployed and investigational drugs: most are organic. Here, Ni(II), Pd(II) and Pt(II) chelates of a tetradentate bis(pyrrole-imine) ligand, H 2 PrPyrr, were used to delineate how the identity of the d 8 metal ion impacts the compound’s affinity for HSA. Fluorescence quenching data acquired on the native protein and HSA bound to site-specific probes showed the compounds target sites close enough to its single Trp-214 residue (subdomain IIA) to quench the fluorophore. The bimolecular quenching rate constants, (cid:1) (cid:2) , were of the order of 10 1 to 10 4 times higher than the maximum diffusion-controlled collision constant of a biomolecule in water (10 10 M  1 s  1 ), reflecting a static fluorescence quenching mechanism. The Stern-Volmer constants, (cid:3) (cid:4)(cid:5) , spanned the range 10 4 M  1 to 10 6 M  1 at 37  C, while the affinity constants, (cid:3) (cid:6) , ranged from  3  10 3 M  1 to  8  10 7 M –1 at 37  C, and followed the order Pd(PrPyrr) >> Pt(PrPyrr) > Ni(PrPyrr) > H 2 PrPyrr. The thermodynamics reflect enthalpically driven ligand uptake, hinging mainly on London dispersion forces (metal ion dependent), along with general multi-site binding. Notably, two reactive species exist for the Pd(II) system, affording the complex HSA•{Pd a }{Pd b }. Molecular docking simulations (G LIDE XP) support the spectroscopic data, confirming that all ligands can target multiple binding sites in silico —all within  20 Å of Trp-214. Although far-and near-UV CD spectroscopy indicated that the optically inactive ligands negligibly perturb the secondary and tertiary structure of HSA, substantial induced CD (ICD) spectra were recorded for the protein-bound ligands and could be simulated by hybrid QM:MM TD-DFT methods. This study highlights how appropriately chelated square planar d 8 metal ions neither decompose nor demetallate after uptake by HSA, proving that metallodrug transport and delivery by HSA might be more feasible than generally acknowledged.


Introduction
Pyrrole-based compounds have been extensively synthesized and studied due to their unique roles in pharmaceuticals and intrinsic bioactivity. 1,2Pyrrole-imine Schiff bases are popular ligands formed by condensing pyrrole-2-carboxaldehyde (or a pyrrole-2-ketone) with primary alkyl-or arylamines. 3The resulting pyrrole-imine Schiff-bases chelate a variety of metal ions, forming stable metal complexes with metal-pyrrolide and metal-imine bonds after concomitant deprotonation of the pyrrole NH group.5][6] Such ligands are suitable for stabilizing metal ions in both typical and high oxidation states. 7Examples include crystallographically characterized complexes of Co(II/III), [8][9][10][11] Rh(I/III), [12][13][14][15] Ir(III), 16,17 Ni(II), [18][19][20] Pd(II), [20][21][22] Pt(II), [23][24][25] Cu(II), [26][27][28][29] Ag(I), 30 and Au(III). 31yrrole-based compounds are important building blocks for numerous organic pharmaceuticals, 32 as exemplified by the clinically-approved drugs Tolmetin (a non-steroidal antiinflammatory agent), 33 Sunitinib (multi-targeted receptor tyrosine kinase inhibitor for cancer chemotherapy), 34 and Glimepiride (used to manage type 2 diabetes mellitus by stimulating insulin release), 35 to name a few.However, studies over the last two decades suggest that pyrrole-imine metal chelates exhibit significant medicinal potential.The best-known example is the pentadentate macrocycle texaphyrin, 36 which complexes numerous transition and main group metal ions, including lanthanides.Texaphyrin's Lu(III) 37 and Gd(III) 38 complexes are promising (but not FDA-approved) compounds for photodynamic cancer chemotherapy 39 and MRI contrast 40 applications, respectively.Although further behind on the clinical development trajectory, patented tetradentate bis(pyrrolide-imine) chelates of Au(III) are promising investigational compounds for cancer 31 and mycobacterial chemotherapy. 41Despite these advances, limited work has been done to elucidate how these compounds bind to human serum albumin (HSA), which is a key step to understanding the plasma distribution of metal chelate drug candidates. 42SA is the most abundant plasma protein present at concentrations of around 600 µM. 43Two of the main functions of HSA are regulating colloidal osmotic pressure and the transportation of exogenous and endogenous compounds. 44SA is a heart shaped macromolecule that consists of a single polypeptide chain (585 amino acids) with a molar mass of 66.5 kDa. 45As depicted in Fig. 1b, the structure is dominated by αhelices within the three domains, namely I (residues 1-195), II  (196-383), and III (384-585).8][49] Apart from the two main drug binding sites, there are several other binding sites that populate the protein, including 7 fatty acid, 4 thyroxine, and several known metal ion binding sites. 50Delineating the binding of a medicinal compound to HSA is central to understanding its pharmacodynamic and pharmacokinetic data in vivo given the abundance of HSA and its transporter role in blood plasma. 472][53] Generally, protein binding by metal complexes is accompanied by ligand dissociation [54][55][56] while nondissociative binding is less common, [57][58][59][60] often though not exclusively 61 hinging on multidentate ligands that provide kinetic inertness and thermodynamic stability. 62,63n this study, we synthesized N,N'-bis[(1E)-1H-pyrrol-2ylmethylene]propane-1,3-diamine (H2PrPyrr) and its complexes with three Group 10 metal ions, specifically Ni(II), Pd(II) and Pt(II), to generate the corresponding isoelectronic (nd 8 ) squareplanar M(PrPyrr) chelates (Fig. 1a).Spectroscopic methods (fluorescence and CD spectroscopy) were used to probe the binding of H2PrPyrr and the three metal chelates by HSA under physiological conditions to gauge how the metal ion influences uptake of the isoelectronic and isomorphic complexes by HSA and to delineate their preferred binding site(s).Finally, molecular docking simulations were used to corroborate the spectroscopically elucidated binding site locations.

Metal chelate synthesis
Modified literature methods were used to metalate H2PrPyrr with Ni(II) 64 and Pt(II). 65For Pt(PrPyrr), we found that simply stirring the solution of the free ligand with K2[PtCl4] and sodium acetate in a binary mixture of DMSO-acetonitrile (50%, V/V) for 4 days at ambient temperature (as opposed to the 4-h reaction at 80 C in DMF employed for the cyclohexyl-bridged ligand 65 ) afforded the product cleanly without significant formation of metal-containing by-products.The difference in reactivity of H2PrPyrr towards Pt(II) compared with the cyclohexyl-bridged ligands likely reflects the higher flexibility of the propyl-bridged analogue employed here.For Pd(PrPyrr), the protocol depicted in Fig. 1a was followed with the use of Pd(II) acetate affording the product cleanly; the free acetate ions serve as a suitable base for accepting the pyrrole NH protons released during metal ion chelation.We found it unnecessary to first dissolve the free ligand in ethanol, as reported elsewhere for this reaction, obviating the need for a mixed solvent system. 66

Protonation reactions
The tridentate hemiprotonated complex PdCl(HPrPyrr) was unexpectedly obtained from a reaction in which Pd(PrPyrr) had been synthesized as depicted in Fig. 1a.The crystals which grew from a mixture of CH2Cl2 and hexane were heterogeneous comprising Pd(PrPyrr) as the majority species and PdCl(HPrPyrr) as a minor component.The source of chloride ions for the latter reaction product may derive from the commercial solvent 67 or the hydrolysis of CH2Cl2 to form HCl and formaldehyde (Scheme 1), 68 possibly catalysed by residual acetic acid present as a byproduct of the metalation of H2PrPyrr with Pd(O2CCH3)2.
Protonation of coordinated N-donor ligands in Pd(II) pincer complexes is known and can be achieved by simply using HCl•OEt2 to effect the reaction. 69Here, experimental evidence in support of the mechanism depicted in Scheme 1 was acquired by investigating the reactions of Pd(PrPyrr) and H2PrPyrr with HCl using UV-visible and 1 H NMR spectroscopy.As shown in Fig. 2a, progressively increasing the concentration of HCl in an acetonitrile solution containing 20 M Pd(PrPyrr) converts the complex into the protonated adduct PdCl(HPrPyrr).A significant excess of HCl was required to bring about complete reaction.Notably, from the spectra of the free ligand (H2PrPyrr) recorded as a function of [HCl] (Fig. 2b), addition of HCl to Pd(PrPyrr) does not lead to demetallation of the chelate and hence formation of the protonated fee base, either [H3PrPyrr] + or [H4PrPyrr] 2+ .The species assignments above were confirmed by the TD-DFT spectra of Pd(PrPyrr) and PdCl(HPrPyrr) calculated in acetonitrile (Fig. S13), which show remarkably good agreement with the experimental spectra.
Since the structural hallmark of PdCl(HPrPyrr) is its lack of symmetry and a single protonated pyrrole group, we used proton NMR spectroscopy to study the reaction of HCl with Pd(PrPyrr).The 1 H NMR spectrum of Pd(PrPyrr) before and after addition of 2 eq HCl in CD3CN confirmed the appearance of two new, moderately sharp signals at 8.12 and 8.21 ppm, which may be assigned to the pyrrole NH proton and imine CH proton of the chelate closest to the protonated pyrrole ring.Both exhibit exchange broadening.In contrast, the imine CH proton of the coordinated pyrrolide-imine moiety is sharp and resonates downfield (8.03 ppm) from the signal for this proton in the parent complex (7.70 ppm).The spectrum of Pd(PrPyrr) recorded in acidified acetonitrile is thus consistent with the structure of PdCl(HPrPyrr).The remaining signals for the protonated complex were in excellent agreement with those calculated by DFT methods (Fig. S14), confirming the intrinsic asymmetry of the complex and the assignment of its structure.From Fig. 2, an additional minor species of a similar structure to PdCl(HPrPyrr) is present (signals marked with an asterisk), quite possibly the solvent-substituted salt [Pd(HPrPyrr)(NCCH3)]Cl.Since two methylene group resonances occur at 3.70 and 3.81 ppm for this minor species, we can rule out double-protonation of Pd(PrPyrr) as this would lead to either dissociation of the ligand from the Pd(II) ion or formation of a symmetric structure with metal-coordination by the central chelate ring alone (i.e., there would be two dissociated pyrrole rings).Both options would give a single methylene resonance in this region of the spectrum, inconsistent with the experimental data.Collectively, the UV-visible and NMR spectroscopic data strongly support the mechanism proposed in Scheme 1.

X-Ray crystal structures
The ambient temperature X-ray structure of Ni(PrPyrr) has been previously described. 64The novel low temperature X-ray structures of Pd(PrPyrr) and PdCl(HPrPyrr) delineated here are shown in Fig. 3. Pd(PrPyrr) crystallized in the space group P21/c (Table S1, Fig. S15) and is structurally similar to the room temperature X-ray structure of the compound (reported in the non-standard space group setting P21/a, see Fig. S16). 66The asymmetric unit (ASU) for Pd(PrPyrr) comprises four independent molecules (A-D), each having the square planar  ).There are no bands below 250 nm or above 350 nm in the spectrum.(b) 1 H NMR spectra of Pd(PrPyrr) recorded in CD3CN in the absence (blue) and presence (red) of 2 eq of HCl in D2O recorded at 37 C.The DFT-calculated structure of PdCl(HPrPyrr) is shown along with signal assignments in the experimental NMR spectrum that are based on the isotropic shielding tensors calculated for the protons in the complex.DFT-calculated bond distances are in Å.The signal from the central methylene group of the propyl bridge (not shown) is overlapped with the solvent peaks at 1.97 ppm (see Fig. S2).The minor signals marked with an asterisk possibly belong to the solvated complex (Cl  exchange).
Pd(II) ion chelated by the tetradentate bis(pyrrolide-imine) ligand.The Pd-Npyrrole and Pd-Nimine distances average 2.017(6) and 2.010(11) Å, respectively (Table S2).These bond distances are statistically equivalent to the mean Pd-Npyrrole and Pd-Nimine values of 2.02(3) Å and 2.02(4) Å, respectively, calculated from 27 similar structures in the Cambridge Crystallographic Database (CSD). 70The relevant literature data are collated in Table S3 and plotted in Fig. S17.Interestingly, despite the higher temperature data for the literature structure of PdPrPyrr, 66 which has mean Pd-Npyrrole and Pd-Nimine distances of 2.011(4) and 2.006(10) Å, respectively, the coordination group bonds are equivalent (within 1) to those determined here at 173 K.The somewhat nonplanar conformations of the ligands mainly reflect nonbonded crystal packing interactions and the formation of discrete -stacked dimers, as discussed in the ESI (Fig. S18).Notably, DFT simulations (Fig. S19) suggest the observed stacking involves significant London dispersion forces.
The X-ray structure of PdCl(HPrPyrr) is shown in Fig. 3b.The square planar Pd(II) ion is chelated by three of the four nitrogen donors of the monoanionic tetradentate chelate, with the protonated pyrrole ring unbound and oriented away from the chloride ligand occupying the fourth coordination site at the metal ion.The unbound pyrrole N-H group participates in intermolecular N-HCl hydrogen bonding (2.40 ± 0.02 Å), which leads to the formation of a one-dimensional H-bonded chain in the solid state.The polymer chain axis is colinear with the lattice 2-fold screw axis, giving the extended chain 21 symmetry overall (Fig. S20).The Pd-Npyrrole and mean Pd-Nimine distances measure 1.993(1) and 2.015(18) Å, respectively.The individual Pd-Nimine bond distances, measuring 1.997(1) and 2.032(1) Å, vary significantly depending on whether the flanking pyrrole ring is metal-bound or not.The shorter distance (Pd1-N2) occurs when the pyrrole ring is metal-bound, and the imine nitrogen resides within the 5-membered chelate ring incorporating the metal ion.This reflects the binding constraints imposed by the 5-membered chelate ring on N2, which are relaxed for N3 within the larger 6-membered chelate ring (propyl bridge).The Pd-Cl distance measures 2.3204(4) Å and is typical for coordination of Pd(II) by chloride in pyrrole-based pincers. 69Interestingly, there are no X-ray structures for comparison with PdCl(HPrPyrr) in the literature, commensurate with the rather unusual nature of this species.

Fluorescence quenching measurements
The mechanism of fluorescence quenching for a protein can be described as either static, dynamic, or a combination of both.For HSA, the intrinsic fluorescence emission is largely the result of a single tryptophan residue (Trp-214,  ex = 280 nm) located in subdomain IIA, and to a lesser extent the 18 tyrosine residues present in the protein. 71We investigated quenching of the intrinsic emission spectrum of HSA (300-400 nm) induced by varying concentrations of the free ligand (H2PrPyrr) and the corresponding Ni(II), Pd(II), and Pt(II) chelates (Figs. 4 and S21A-S21C).The emission maximum for HSA was observed at 343 nm and the fluorescence intensity decreased monotonically with increasing concentrations of the added ligands.(The term "ligand" hereafter is used in a biomolecular sense, referring to the metalated or metal-free compound that binds to the macromolecule.)The spectroscopic data suggest that all four ligands bind reasonably close to Trp-214 (subdomain IIA), changing the microenvironment around the fluorophore and inducing a marked dose-dependent (i.e., "static" or bindingrelated) quenching of the intrinsic fluorescence of HSA. 72,73or HSA, the wavelength of the fluorescence emission maximum ( ) was dependent on the identity of the bound ligand.At the maximum doses investigated before quenching and band broadening limited fluorescence detection, H2PrPyrr and Pd(PrPyrr) induced blue shifts relative to the native protein of 2.0 and 11.2 nm, respectively (same buffer system, identical temperature).In contrast, the Ni(II) and Pt(II) chelates red-shifted by up to +2.3 nm in the case of the Pt(II) derivative.While shifts of 1 nm in are close to the wavelength resolution of the instrument (i.e., statistically insignificant), the notable blue shift in the emission maximum for Pd(PrPyrr) ostensibly reflects the development of a significantly less polar environment around Trp-214 after ligand uptake. 74,75From the Hill function 76 used to fit the variation in HSA emission wavelength with [Pd(PrPyrr)] (Fig. S22), the wavelength shift is 48% complete when [Pd(PrPyrr)] = 6 M.The full wavelength shift cannot be measured in the present system because the emission signal is strongly quenched when [Pd(PrPyrr)] > 10 M.Consequently, the estimated limiting value for is 317 ± 19 nm when [Pd(PrPyrr)] > 100 M.The full shift in the emission wavelength maximum for HSA bound to Pd(PrPyrr) at saturation is thus around 26 nm.Though large, the magnitude of the blue shift observed here (relative to native HSA) is akin to that found for T4 lysozyme, staphylococcal nuclease, and bacterial chemotaxis protein CheY (relative to for Trp in these proteins in vacuo) 77 caused by changes in (i) the orientational polarization of the Trp fluorophore induced by ordered water molecules within 15-25 Å of the residue and (ii) electronic polarization of the Trp indole ring.For HSA•{ligand} complexes, electronic polarization of Trp-214 induced by the bound ligand and perturbation of the ordered water structure around the chromophore both likely underpin the observed shifts in , particularly considering the electronic structures of the metal chelates.Whether the bound ligand induces a blue-or red shift in would depend on the orientation and centre-to-centre distance of the ligand's dipole moment relative to the transition dipole moment for the lowest energy 1 A1  1 La ground state transition of Trp's indole ring. 74This transition involves transient charge migration from the pyrrole ring to the benzene ring of the chromophore.)In effect, the precise way in which the ligand is bound relative to Trp-214 will govern the sign and magnitude of Δ .The small red shift in the emission maximum of 2.3 nm for the HSA•{Pt(PrPyrr)} complex suggests that the environment around Trp-214 becomes somewhat more hydrophilic (polar) upon ligand uptake. 75As noted above, this shift likely reflects several contributing, intricate factors such as rearrangement of the positions or orientations of H-bonded water molecules within a 15-25 Å radius of Trp-214.(Even though Trp and Tyr are readily excited at 280 nm, most of the intrinsic fluorescence of HSA is due to Trp-214 since it has a higher quantum yield and more efficient resonance energy transfer than Tyr; 71 see Table S6.)The key conclusion garnered from the present wavelengthshift data is that Pd(PrPyrr) probably binds to HSA in a distinctly different manner to the Ni(II) and Pt(II) analogues.

Fluorescence quenching mechanism
Stern-Volmer model.Fluorescence quenching is normally described by the Stern-Volmer (SV) equation (eq.1), 78 1 1 ⁄ (1) where is the fluorescence intensity of HSA in the absence of the added quencher (metal chelate or free base ligand in this work) and is the fluorescence intensity of HSA in the presence of the ligand.is the Stern-Volmer constant (M 1 ), is the molar concentration of the quencher, is the bimolecular quenching rate constant (M 1 s 1 ), and is the average lifetime of HSA fluorescence in the absence of any quencher (5.28  0.03 ns, 79 5.60  0.10 ns, 80 6.72  0.07 ns; 81 mean = 5.87  0.76 ns).
Stern-Volmer plots recorded as a function of temperature and [Pd(PrPyrr)] are presented in Fig. 5.The measurements were conducted in triplicate and the graphs plot the mean fluorescence ratio recorded at each dose of the metal chelate.Although eq. 1 is normally linear for a single quenching mechanism, either dynamic or static, the plots for Pd(PrPyrr) are nonlinear, showing a marked positive deviation from linearity with increasing [ligand].The data were thus analysed using the extended nonlinear SV equation reported by Campbell et al. 82 and, in rearranged format, by Ciotta et al., 83 eq. 2, where and are the experimental concentrations of the fluorophore (HSA) and quencher (ligand), respectively, and incorporates the Stern-Volmer constant, eq.3: 1 In distinct contrast to Pd(PrPyrr), linear SV plots were obtained for H2PrPyrr, Ni(PrPyrr), and Pt(PrPyrr).The variable temperature data were, furthermore, adequately described by eq. 1 (Fig. S23a).From least-squares fits of eqs. 1 or 2 to the dose-dependence of ⁄ , the bimolecular fluorescence quenching rate constant for the HSA•{ligand} complex can be straightforwardly deduced from eq. 4, provided that is known (vide supra). (4) The calculated and values for the interaction of the four ligands with HSA are summarized in Table 1.Typically, linear SV plots are associated with a single class of quenching, either dynamic (diffusion-limited collisional) or static (bindingrelated). 84,85The specific class of quenching can be distinguished by the temperature dependence of .For dynamic quenching, the values increase with increasing temperature mainly due to an increase in the collisional frequency and diffusion rate, while for static quenching, values decrease with increasing temperatures because the equilibrium formation constant for the complex decreases with increasing temperature. 86From Table 1, the values all decrease with increasing temperature, reflecting a static quenching mechanism.Furthermore, the bimolecular quenching rate constants, , all exceed the diffusioncontrolled limit (1 × 10 10 M 1 s 1 ) 87 by 3 to 4 orders of magnitude, consistent with stable complex formation and a static fluorescence quenching mechanism.
For Pd(PrPyrr), the SV plot of Fig. 5a is linear at low ligand concentrations (< 2 µM), but exhibits a marked upward curvature as [Pd(PrPyrr)] increases (also see Fig. S24).The spectroscopic behaviour of this system suggests that two static quenching mechanisms related to two distinct proteinligand binding modes are probably operative for the Pd(II) chelate since it clearly behaves similarly to the isostructural Ni(PrPyrr) derivative only at low concentrations (< 2 µM).
Positive deviations from linearity (upward curvature) in SV plots are not uncommon and have been experimentally and theoretically delineated since the 1950s; 88-90 they often signal the involvement of both static and dynamic quenching mechanisms or a combination of distinct static mechanisms. 84e interpret the nonlinear SV plot for the HSA-Pd(II) system as reflecting two overlapping binding equilibria.The first dominates at low ligand concentration (< 2 µM) and involves uptake of the tetradentate chelate of Pd(II) (represented structurally in Fig. 3a) at a site close enough to Trp-214 (within 20 Å) to commence fluorescence quenching, culminating in the initial linear phase of the SV plot.The behaviour of Pd(PrPyrr) in this binding event is thus akin to the isostructural complex Ni(PrPyrr).The second equilibrium commences either in parallel or at a slightly higher [Pd II ] and involves formation of a different HSA-ligand complex, possibly involving uptake of the metal complex at a second binding site with a different spatial ), bimolecular quenching rate constants ( ), association constants ( ) and stoichiometric coefficients ( ) for the interaction of H2PrPyrr, Ni(PrPyrr), Pd(PrPyrr) and Pt(PrPyrr) with HSA at different temperatures in 50 mM KH2PO4 buffer at pH 7.50.

Compound
Temp. a values (Stern-Volmer constants) were determined from fitting the data to eq. 2 in the case of Pd(PrPyrr); all other values were obtained by fitting linear eq. 1 to the data.A mean excited state lifetime, 0, of 5.87 (76) ns for HSA was used to calculate the bimolecular quenching rate constant, .c The estimated standard uncertainties of the least significant digits are given in parentheses.d Ligand:HSA binding stoichiometry from the fit of the data to eq. 7. Error bars are ESD's based on the average of three independent determinations.The data are described by eq. 7, which affords the affinity constant and stoichiometric coefficient for the reaction (Table 1).
relationship to the fluorophore (Trp-214), or uptake of a different Pd(II) chelate species at an exclusive location.The second binding event could well hinge on initial ligand exchange, which is unique to the more labile Pd(II) ion, as evidenced by the structure of PdCl(HPrPyrr) shown in Fig. 3b.In effect, the metal chelate species bound by HSA in the second equilibrium is likely to be dissimilar to that bound in the first and will therefore quench the Trp-214 fluorescence differently, such that !≠ . The values of and determined here for the interaction of native HSA with H2PrPyrr and its Ni(II) and Pt(II) chelates are from 2 to 10 times larger than, for instance, those reported for HSA interacting with berberine, 91 epinastine hydrochloride, 92 primethamine and trimethoprim, 93 as well as malvidin-3-glucoside. 84 However, the interaction of this Pd(II) system with HSA affords and values that are unique, being about two orders of magnitude larger than is typical for simple quenchers and probably due to dual-site/dual-species binding of the Pd(II) chelate(s) with the protein (vide infra).
Sphere of action model.A convenient interpretation of the nonlinear SV plot for the HSA-Pd(II) system may be made using this model, 87,94 which invokes the presence of a sphere of a finite volume around Trp-214 in which the probability of the quencher (Q) inducing fluorescence quenching is unity. 84The "sphere of action model" is described by eqs. 5 and 6, where // is the apparent static fluorescence quenching constant.By taking the natural logarithm of eq. 5, linear plots with a zero intercept (eq.6) are expected for increasing [Q] in systems that exhibit static quenching.Analysis of the data for the Pd(II) system with eq.6 is particularly effective (Fig. S23b), and confirms fluorescence quenching by a static mechanism.Importantly, we have used eqs.2, 6, and 7 together with sitespecific drug probes to develop a better understanding of the reaction mechanism for the interaction of Pd(PrPyrr) and the related compounds with HSA (vide infra).
From Table 1, the values follow the order Pd(PrPyrr) > Ni(PrPyrr) > H2PrPyrr > Pt(PrPyrr).The values of delineated from the fits of eq. 2 to the data for Pd(PrPyrr) are 5-fold larger than those determined from eq. 6 (Fig. S23b) at each temperature, reflecting some of the approximations inherent in the sphere of action model.Notably, the identity of the d 8 metal ion influences the HSA fluorescence quenching mechanism in the present system, which likely has a significant FRET (Förster resonance energy transfer) 95 component.Pt(PrPyrr) is the least efficient HSA fluorescence quencher (smallest ).Because Pt(II) bis(pyrrolide-imine) chelates, but not the Pd(II) or Ni(II) congeners, are brightly phosphorescent, they have singlet excited states 1 (-*) that relax by internal conversion and intersystem crossing to the lowest-energy emissive triplet excited state, 3 (-*). 65,96If Trp-214 fluorescence decay proceeds via a singlet state pathway involving the excited singlet state(s) of the quencher, then a Pt(II) chelate with a low-energy 3 (-*) excited state will be a less efficient Trp-214 quencher than its non-phosphorescent Ni(II) congener, in accord with the spin selection rule for FRET. 97(FRET requires conservation of angular momentum for the transitions of the donor and acceptor chromophores, i.e., the excited states of both chromophores must have the same spin multiplicity.) The quenching efficiency order here ultimately reflects the smaller spin-orbit coupling constant for Ni(II) vs. Pt(II), which means intersystem crossing to a low-lying excited triplet state is negligible for Ni(PrPyrr).Since H2PrPyrr has no heavy elements to facilitate intersystem crossing, it readily quenches Trp-214's excited singlet state when noncovalently bound to HSA and, for a comparable spectral overlap integral, will be more efficient than Pt(PrPyrr), as seen in Table 1.Finally, Pd(PrPyrr) is expected to be the most efficient native HSA quencher because (a) its low-lying excited states overlapping the emission from Trp-214 are 1 (-*) excited states and (b) two metal chelate species simultaneously bind to the protein, which statistically increases the quenching rate by at least a factor of 2 (two quenchers are better than one, all other aspects being equal).The actual quenching rate increases 10-fold for the Pd(II) system relative to Ni(PrPyrr) because the molecular and electronic structures of the two quencher species are different.

Ligand binding equilibrium constants
Changes in HSA fluorescence intensity as a function of ligand dose can be used to calculate thermodynamic data for the binding equilibrium between the protein and the titrant ligand.The biophysical parameters that describe the ligand's affinity for the protein ( ) and the stoichiometry of the reaction ( ) are delineated from a log plot of the emission data as a function of increasing ligand concentration (eq.7), 98 ,01 -. ,01 ,01 where the intercept and slope of the curve give the affinity constant and stoichiometry, respectively.The data are summarized in Table 1 and have been plotted for Pd(PrPyrr) in Fig. 5b.(Similar plots for the remaining compounds are given in Fig. S25.)For each compound, the affinity constant decreases with increasing temperature, consistent with a static quenching mechanism. 99,100The stoichiometric coefficients from the slopes of the straight line graphs obtained for Pd(PrPyrr) were all > 1 (e.g., 1.63  0.08 at 288 K), strongly suggesting that the reaction stoichiometry tends towards 2:1 (ligand:HSA) as the temperature is lowered (Table 1).The behaviour of Pd(PrPyrr) with HSA sharply contrasts that for H2PrPyrr, Ni(PrPyrr), and Pt(PrPyrr) since these compounds all gave 0.8 < < 1.2.
From Table 1, the log values follow the order Pd(PrPyrr) ≫ Pt(PrPyrr)  Ni(PrPyrr) > H2PrPyrr.The value of ,01 for the Pd(II) chelate ranges from 9.36-6.65 (288-315 K) and is akin to the product equilibrium constant 3 , where ,013 ,01 !,01 , which accounts for why the affinity constant is 2-3 orders of magnitude larger than the values measured for the Pt(II) or Ni(II) congeners.It is noteworthy that ! and for the reaction of HSA with the Pd(II) system are unresolvable (distinct binding steps are not observed in the titration data), consistent with concurrent ligand binding.Since the isostructural neutral metal chelates Ni(PrPyrr) and Pt(PrPyrr) are similar, and neither readily undergo ligand exchange at the metal ion, it is expected that they will bind noncovalently to the same site in HSA with comparable affinity.This is affirmed by their log values which overlap and fall in the range 5.41-4.07(288-315 K).

Reaction mechanism and binding site specificity
Hydrolysis.A feasible and key reaction for Pd(PrPyrr), given its lability and known solution chemistry (Scheme 1, Fig. 2), is hydrolysis of the complex to form an aqua complex, eq.8: 2][103] If this occurs, then two Pd(II) species will be present in solution to react with HSA and, because they are structurally distinct, will likely bind in two independent sites within range of Trp-214 to collectively quench its fluorescence.
Experimental evidence for the spontaneous formation of [Pd(HPrPyrr)(OH2)] + was obtained by simply recording the CD spectrum of Pd(PrPyrr) in an acetonitrile-water mixture as a function of time.The spectra are shown in Fig. S27 along with the DFT-calculated spectrum of this species.Because Pd(PrPyrr) has exact Cs symmetry it has no CD spectrum.The structure of [Pd(HPrPyrr)(OH2)] + , in contrast, has C1 symmetry with enantiomers that differ in energy by 50 kcal mol 1 , thwarting racemization.The chiral hydrolysis product therefore exhibits a distinct CD spectrum that is easily measured upon its formation.
To understand how the species distribution in eq. 8 affects the reaction of the Pd(II) system with HSA, we carried out a competitive blocking assay by fluorescence spectroscopy in which the compounds were titrated into solutions of both native HSA and solutions that were pre-equilibrated with (i) warfarin and (ii) ibuprofen (Fig. 6).As revealed by X-ray crystallography, warfarin binds specifically in Sudlow's site I (subdomain IIA), 104 while ibuprofen primarily binds in Sudlow's site II (subdomain IIIA), with a secondary site in the region between subdomains IIA and IIB (IIA-IIB). 105By selectively and sequentially blocking the two main drug binding sites of HSA, the experiment can potentially delineate a ligand's binding site specificity.One caveat is that HSA has multiple drug binding sites. 105Blocking a primary binding site may simply redirect ligand uptake to a secondary binding site within the protein (or simply thwart incoming ligand binding by allosteric inhibition).However, changes in the affinity constant, stoichiometry, , and substitution of the blocking drug itself might occur, thereby revealing a possible binding site for the ligand.The system is intricate, but we have summarized the experimental strategy used and the results obtained pictorially in Fig. 7 (which also incorporates our in silico binding site affinity data, vide infra).
Binding of Pd II (X) by HSA.When the equilibrium mixture of Pd(PrPyrr) and [Pd(HPrPyrr)(OH2)] + (hereafter Pd eq II ) reacts with native HSA, all possible binding sites are available for ligand uptake.The Trp-214 fluorescence quenching (which is static over the full range), reaction stoichiometry, and log data (Figs.6a and 6b) indicate that a nominally 2:1 metal:protein complex, abbreviated HSA•{Pd II (X)}2, where X represents the ligand combinations in eq. 8, is formed by simultaneous ligand uptake in a single step (logK1 = 9.01  0.25, n = 1.57 0.04, 298 K).Titration of HSA•{warfarin} with Pd eq II while monitoring the fluorescence emission from warfarin ( ex = 320 nm,  em = 382 nm) reveals, firstly, that step-wise binding of the Pd(II) chelate(s) occurs and, secondly, that warfarin is displaced from the protein when [Pd eq II ] > 5 M.The latter event is highlighted by the switch from static quenching of the emission from protein-bound warfarin to dynamic quenching of free warfarin in the solvent when [Pd eq II ] > 25 M (Fig. 6a).The switch in the slope of the double-log plot of Fig. 6b reflects the stepwise uptake of the Pd(II) chelate(s) by HSA•{warfarin}, which is characterized by independent affinity constants for each step (log ! 6= 5.08  0.05, n1 = 0.98  0.01; log 6 = 1.82  0.17, n2 = 0.34  0.04) and an overall binding constant log3 6 = 6.90, which is 2 log units lower than log3 measured for native HSA.The data from these experiments are consistent with the simultaneous uptake of two Pd(II) complexes at distinct sites located near Trp-214 in native HSA.When warfarin is bound to HSA in Sudlow's site I (subdomain IIA), only one Pd(II) complex is initially bound by HSA-presumably at a site somewhat removed from the warfarin-binding pocket, but close enough to quench its fluorescence by a FRET-based mechanism.Since Sudlow's site II in subdomain IIIA is vacant at lower [Pd eq II ], ligand uptake could target this binding site or a region close to it.However, we cannot rule out the binding of a Pd(II) species at a unique location close to the warfarin binding site, e.g., between subdomains IB and IIA (indomethacin secondary binding site), IA-IIA (fatty acid binding site, FA2), or even the secondary ibuprofen binding site (IIA-IIB). 105Importantly, at a sufficiently high concentration of Pd eq II , warfarin is displaced from the protein by the uptake of a second partial equivalent (n2 < The signature switch from static to dynamic quenching of the fluorophore when it is released into the bulk solvent is shown in Fig. 6a.The data confirm that when ibuprofen blocks Sudlow's site II (broadly defined), the first  S9) and Ni(PrPyrr) gives similar results (Figure S28; Tables S8 and S9).Although spectroscopy and simulations are insightful here, X-ray crystallography will ultimately be required to definitively assign the proposed binding sites.
ligand uptake equilibrium involves binding of the Pd(II) chelate(s) most probably within the unoccupied pocket of subdomain IIA (Sudlow's site I), or another site in the protein not occupied by ibuprofen.The binding site location will be within 20 Å of the bound ibuprofen fluorophores to quench their emission by FRET.The second ligand binding equilibrium for the Pd(II) chelate(s) involves competition for one or both of the ibuprofen binding sites, culminating in substitution of the blocking drug with log 7 < log ! 7and concomitant release of ibuprofen into the bulk solvent, as evidenced by the switch to dynamic quenching of the fluorophore.Since log 7 > log 6 , ibuprofen is more readily displaced than warfarin from HSA.One caveat is that metal chelate binding by HSA may induce conformational changes that trigger release of the blocking drug (fluorophore), even if the metal chelate does not target the drug's binding site directly (i.e., probe dissociation induced by allosteric effects).Although there is still some uncertainty regarding the precise binding sites of the Pd(II) chelates, the evidence collectively points to Sudlow's sites I and II as being among the most probable locations for ligand uptake.Such multi-site ligand binding is not uncommon-organic antiinflammatory drugs such as diflunisal are known (X-ray crystallography) to bind at multiple sites in HSA (specifically Sudlow's I and II, and between subdomains IIA and IIB). 105It is worth emphasizing here that ligands with multi-site binding to a promiscuous protein like HSA may fractionally populate two or more preferred binding sites in a concerted uptake equilibrium (single step) to give an apparent reaction stoichiometry of n 1:1 (ligand:protein) with a single log value, potentially complicating interpretation of the biophysical data.
The analysis of the dual ligand binding mechanism for the reaction of native HSA with the Pd(II) chelate above hinges on aquation of the metal ion and the resultant two species in solution (eq.8).A precedent for this type of dual-species uptake with subsequent Trp-214 fluorescence quenching exists for 3hydroxyflavone (3-HF) with HSA. 106Specifically, 3-HF has two tautomers (one anionic) in equilibrium which have similar, resolvable overlapping binding constants with HSA ( ! = 7.2  10 5 M 1 and = 2.5  10 5 M 1 ).Moreover, the two 3-HF tautomers bind in different binding sites (subdomains IIIA and IIA), engendering different Trp-214 quenching rate constants. 106his is similar to what we observe for the two Pd(II) species, Pd(PrPyrr) and [Pd(HPrPyrr)(OH2)] + , with the notable exception that we cannot resolve ! and for the binding of the two structurally distinct metal chelates due to their simultaneous uptake by the protein.Interestingly, their docking scores are virtually the same for their top-scoring and independent binding sites (Table S9), consistent with the experimental data.
Binding of Ni(PrPyrr) by HSA.The analogous titration data for Ni(PrPyrr) (Table S8, Fig. S28) reflect an apparent single binding equilibrium in which uptake of the metal chelate is diminished in the ibuprofen-bound target, HSA•{ibuprofen}n, but seemingly unaffected in the warfarin-bound target relative to native HSA.This suggests Ni(PrPyrr) either uniquely targets Sudlow's site II (binding in or close to the ibuprofen primary binding site in subdomain IIIA), or that the Ni(II) chelate targets both sites with the location of the Ni(II) chelate in the larger warfarin binding pocket (subdomain IIA) being displaced from the position occupied by warfarin itself.Thus, Ni(PrPyrr) and warfarin may co-locate in subdomain IIA of HSA, which would afford a similar affinity constant for uptake of the Ni(II) chelate by the two targets-native HSA and HSA•{warfarin}.Interestingly, and as discussed later, our in silico docking data confirm the feasibility of Ni(PrPyrr) co-binding adjacent to warfarin in subdomain IIA.From the "sphere of action" plot of Fig. S28, Ni(PrPyrr) begins to displaces both ibuprofen and warfarin from HSA at higher doses since these probes exhibit enhanced dynamic quenching by the solvent when [Ni(PrPyrr)] > 30 M.The spectroscopic data collectively suggest that Ni(PrPyrr) exhibits single-step multi-site binding to HSA, simultaneously targeting both subdomain IIA and IIIA, with a cumulative binding stoichiometry of n  1.
Binding of Pt(PrPyrr) by HSA.The picture is slightly more complicated (yet still conceptually accounted for by Fig. 7) for the binding of Pt(PrPyrr) and H2PrPyrr to HSA pre-equilibrated with the fluorescent drug probes (Figs.S29 and S30).These plots show that Pt(PrPyrr) binds with an approximately 1:1 stoichiometry to native HSA.At no point does Pt(PrPyrr) displace warfarin or ibuprofen from HSA•{warfarin} and HSA•{ibuprofen}n; however, the presence of either drug enhances the binding of Pt(PrPyrr) as the log values increase from 5.35  0.20 (ntotal = 1.22  0.05) in the native protein to 6.80  0.16 (ntotal = 1.50  0.04) and 6.76  0.15 (ntotal = 1.41  0.03) in HSA•{warfarin} and HSA•{ibuprofen}n, respectively.One interpretation is that the Pt(II) chelate can bind in both Sudlow's site I and II of the native protein (i.e., half-or fractionally-saturate the two sites so that n1 + n2 = ntotal = 1.22), but that when a drug is already located in one of these two sites, the reciprocal unbound site is fully saturated by the metal chelate.Furthermore, the bound drug (warfarin or ibuprofen) subtly alters the protein conformation to enable the uptake of an additional half equivalent of Pt(PrPyrr) at a third site, as evidenced by ntotal 1.5 for either protein target.This points to a classic structural synergistic or allosteric effect (cooperative binding).It is also possible that Pt(PrPyrr) may bind alongside warfarin in the larger Sudlow's I pocket (subdomain IIA), like Ni(PrPyrr), and that warfarinPt(PrPyrr) London forces of attraction enhance the microscopic binding constant at this site.This notion is well-supported by our in silico docking experiments (vide infra) and could at least partly explain why Pt(PrPyrr) does not induce the dissociation of warfarin at higher doses.Regarding the existence of additional drug binding sites, many are known for HSA 105 and are located in subdomains IIIB (propofol binding site, 98 close to Sudlow's site II in subdomain IIIA) and IB (bilirubin 107 and hemin binding site 46 ).From the steep initially linear and thence upwardly-curved Stern-Volmer response function for Pt(PrPyrr) binding to HSA•{ibuprofen}n, Pt(PrPyrr) is possibly partially taken up in subdomain IIIB, which is adjacent to subdomain IIIA in the protein, thereby facilitating efficient additive quenching of the fluorescence from bound ibuprofen (the first equivalent of the quencher being bound in the adjacent pocket, Sudlow's site I).
Binding of H2PrPyrr by HSA.In the case of H2PrPyrr, similar but somewhat lower affinity binding to that seen for Pt(PrPyrr) with native HSA is observed (log = 4.44  0.05, ntotal = 0.93  0.01).Likewise, H2PrPyrr cannot displace warfarin or ibuprofen (Fig. S30).As before, the presence of either drug enhances the binding of the ligand H2PrPyrr (via a structural synergistic effect) as the log values increase by more than 2.5 log units for the warfarin-loaded target (log = 6.95  0.29, ntotal = 1.50  0.06, HSA•{warfarin}) and ibuprofen-bound target (log = 8.35  0.19, ntotal = 1.80  0.04, HSA•{ibuprofen}n).As with Pt(PrPyrr), half-saturation of Sudlow's site I and II of the native protein can account for ntotal = 1, and when a drug is already located in one of these two sites, the reciprocal site is fully saturated as the ligand dose increases (keeping ntotal = 1).The presence of a prebound drug (warfarin or ibuprofen) clearly facilitates the uptake of an additional near-full equivalent of H2PrPyrr at a third site (ntotal is 2 when HSA•{ibuprofen}n is the target).The third site is presently difficult to pinpoint from the available spectroscopic data; our in silico docking study (Table S9, vide infra) suggests the neutral free ligand targets subdomain IA in the absence of ibuprofen, which is adjacent to the hemin and bilirubin binding site (subdomain IB). 107When ibuprofen is present, however, H2PrPyrr targets subdomain IIA while the protonated species [H3PrPyrr] + and [H4PrPyrr] 2+ are surfacebound in the IB-IIIA cleft (Table S9).Although a pH of 7.5 is maintained experimentally in the titrations, protonation of the Schiff base free ligand is possible upon ligand uptake by HSA, depending on the specific binding site and which polar residues interact with the ligand.Interestingly, [H3PrPyrr] + binds adjacent to warfarin in subdomain IIA of HSA•{warfarin} in silico (Table S9), suggesting that if this cationic species is formed under the present experimental conditions, then dissociation of warfarin from HSA would be thwarted at higher doses.
Mass spectroscopy of HSA complexes.Although details regarding the binding stoichiometry of the present metal chelates and H2PrPyrr are quite clear from the above analysis, direct experimental proof gleaned via another experimental method is desirable.We chose to analyse the complexes of HSA formed with the metal chelates by ESI-MS to prove (i) that uptake of fully intact metal chelates occurs, (ii) multi-site binding is possible (and at least consistent with 1  ntotal  2), and (iii) that the Pd II (X) system behaves uniquely due to dual species uptake by the protein.From Fig. S31 it is clear that the species HSA•{Ni(PrPyrr)}2 was formed and dominant in the sample.In the case of Pt(PrPyrr), an entirely reasonable species distribution was evident: HSA (47%), HSA•{Pt(Pr(Pyrr)} (43%), and HSA•{Pt(Pr(Pyrr)}2 (10%).For the HSA-Pd(II) system, the resolution of the data was not sufficient to distinguish between the exact chemical species bound, i.e., Pd(PrPyrr) and/or [Pd(PrPyrr)(OH2)] + .However, a similar component distribution pattern to that seen for the Pt(II) system was obtained, albeit with a lower fraction of the unbound protein: HSA (26%), HSA•{Pd(X)} (38%), and HSA•{Pd(X)}2 (35%).Notably, the mass spectrum of the Pd(II) system was unique in that satellite peaks appeared on the main bands at low m/z.Although minor in abundance, these peaks are likely due to covalent adducts formed when surface residues such as His 62 and Glu substitute Pd-bound water in [Pd(PrPyrr)(OH2)] + .The kinetic inertness of Pt(PrPyrr) and Ni(PrPyrr) precludes the formation of such peaks.

Thermodynamics of ligand binding
The departure from 1:1 reaction stoichiometry for the native HSA-Pd II (X) system, where X represents the ligand combinations for Pd eq II in eq. 8, was strikingly confirmed by the nonlinear van't Hoff relationship obtained from the temperature dependence of the values (Fig. 8b).The data are satisfactorily described by the second order van't Hoff equation, eq. 9 (where 8, 9, and : are coefficients determined by least squares fitting), 108 which confirms the entropy and enthalpy change for the reaction (∆< and ∆=, respectively) are not constant over the full temperature range.In the present case, ∆< and ∆= change with temperature according to eqs. 9 and 10 (where > is the gas constant) 108 because of the non-singular mode of interaction of the metal chelate species with the native protein.
∆< > @8 : ?A For the HSA-Pd II (X) system, the coefficients of eq. 9 may be used to calculate the change in enthalpy and entropy for the reaction at any temperature (eqs.10 and 11), allowing delineation of the ligand binding thermodynamics.The remaining ligands follow the linear van't Hoff relationship, and plots for triplicate measurements are given in Fig. 8.Under non-standard conditions, the enthalpy change (∆=), entropy change (∆<), and Gibbs free energy change (∆B), may be deduced from the temperature dependence of the affinity constants ( ) using eqs.12 and 13: ∆B ∆= ?∆< Table 2 summarises the thermodynamic data for reaction of the compounds of interest with HSA in phosphate buffer at neutral pH.The reactions are all exergonic, 98 with ∆B ranging from 26.4 kJ mol 1 for H2PrPyrr to 51.4 kJ mol 1 for Pd eq II at 298 K.
As graphically portrayed in Fig. 8, the values of ∆B fall in a narrow range for three of the four compounds (26 to 29 kJ mol 1 at 298 K).The much larger free energy change measured for the formation of HSA•{Pd II (X)}n reflects a more complex interaction with HSA.Although Pd(PrPyrr) is isoelectronic and isomorphous with its Ni(II) and Pt(II) congeners, neither the Ni(II) nor the Pt(II) chelate exhibits close to a 2:1 (ligand:HSA) reaction stoichiometry with the native protein.The unique behaviour of the Pd(II) derivative reflects the uptake of two distinct metal species, as postulated in Fig. 7.Moreover, the dual binding equilibrium between Pd eq II and native HSA involves overlapping equilibria because the plot of eq.7 is linear over the full concentration range (Fig. 6b), pointing to simultaneous uptake of the metal chelate species.
Regarding the ∆= values for the reactions, which are exothermic and range from 45 kJ mol 1 for Ni(PrPyrr) to 104 kJ mol 1 for Pd II (X), Fig. 8a highlights the dominance of the enthalpy term.Since the drug binding sites known as Sudlow's I and II (subdomains IIA and IIIA) in the protein are mostly hydrophobic, 45 the favourable heats of reaction suggest that there is minimal disruption of the protein's ordered water or backbone structure and that London dispersion forces of attraction (and H-bonding for H2PrPyrr) likely underpin binding of the present ligands in these sites.This notion is indeed confirmed by our molecular docking simulations (vide infra).
By using the Gibbs-Helmholtz relationship (eq.13) with the experimental data (Table 2), the role played by the d 8 metal ion in determining the thermodynamics governing ligand uptake by HSA for this group of compounds may be delineated (Fig. 9).Specifically, when ?∆</∆B is plotted against ∆=/∆B, the data fit the linear relationship in the upper quadrant of the graph where ∆= < 0, consistent with enthalpic control of the reaction.In this quadrant, spontaneity is assured as changes in ∆=/∆B are compensated for by commensurate changes in ?∆</∆B.Importantly, ?∆</∆B and ∆=/∆B increase together in the order: Ni(PrPyrr) < Pd II (X) < Pt(PrPyrr), which follows the increase in the principal quantum number for the metal ion (3d < 4d < 5d).The trend indicates that the thermodynamics of metal chelate binding by HSA depend on increasing the electron density present in the complexes, which occurs naturally as the size of the metal ion and its polarizability (i.e., softness) increases. 109,110Since London dispersion forces (LDF) of attraction in biomolecules significantly control the tertiary structure of the macromolecule, 111 and will increase in the order 3d < 4d < 5d 110 for this class of pyrrolide-imine metal chelates, Fig. 9 demonstrates that LDF attraction governs the interaction of the current metal chelates with HSA largely by a The estimated standard uncertainties of the least significant digits are given in parentheses.b X represents the ligand combinations shown in eq. 8.
Fig. 9 Plot of the Gibbs-Helmholtz relationship (eq.13) for the reaction of H2PrPyrr, Ni(PrPyrr), Pd II (X), and Pt(PrPyrr) with HSA at 298 K in 50 mM KH2PO4 buffer at pH 7.50.The straight line fit of the data gives R 2 = 0.9997 with a slope and intercept of 1.For all reactions, ∆B < 0. The plot highlights how the identity of the d 8 metal ion influences the reaction thermodynamics.(X represents the ligand combinations shown in eq.8.) modulation of the ∆= term.The free ligand (H2PrPyrr) is located between Pd II (X) and Pt(PrPyrr) on the Gibbs-Helmholtz function, reflecting the contribution to binding made by both the electron density of the pyrrole-imine groups as well as the pyrrole N-H groups, which are excellent H-bond donors. 112,113ormation of strong N-HX hydrogen bonds between H2PrPyrr (or its conjugate acids) and polar residues within the central region of Sudlow's drug binding site I 114 will clearly favour exothermic ligand binding (∆= C 0), albeit at a loss of entropy.
The ∆< values for the d 8 Group 10 metal chelates listed in Table 2 become increasingly negative with progression down the group (3d, 4d, 5d), ranging from 57 J K 1 mol 1 (Ni 2+ chelate) to 222 J K 1 mol 1 (Pt 2+ chelate).The change in entropy for the binding of the free ligand (H2PrPyrr) is also large and negative (135 J K 1 mol 1 ).The data suggest that there is neither substantial desolvation of the incoming ligand (none of them are particularly hydrophilic) nor significant loss of ordered water from the target binding site which accommodates the ligand, culminating ordering of the system upon ligand uptake.The ligand binding thermodynamics here mirror data reported from microcalorimetry studies on the binding of drugs such as ibuprofen, 115 benzodiazepines, 116 and phenothiazines by HSA, 117 which have moderately negative ∆= (50 to 75 kJ mol 1 ) and ∆< values as well as sizeable !values 10 5 M 1 . 118n the first association equilibrium, these particular drugs all bind in hydrophobic Sudlow's site II (subdomain IIIA) of HSA.Interestingly, a second ligand association occur (dual-site binding, also in subdomain IIIA) with values 10 3 M 1 and similarly negative ∆= and ∆< values. 118Warfarin, which binds in Sudlows's site I (subdomain IIA), 104 has broadly comparable thermodynamics 118 (,01 != 5.15, ∆= = 49 kJ mol 1 , ∆< = 12 J K 1 mol 1 ) to ibuprofen (Sudlow's site II, ,01 != 5.04, ∆= = 67 kJ mol 1 , ∆< = 34 J K 1 mol 1 ) and Ni(PrPyrr) (,01 = 4.92, ∆= = 45 kJ mol 1 , ∆< = 57 J K 1 mol 1 ), suggesting that assigning the specific binding site of the ligand requires additional structural or spectroscopic data.

Far-UV CD spectroscopy of HSA
Since the UV-region CD spectra of HSA (and other proteins) are sensitive to changes in the secondary (and tertiary) structure of the protein caused by ligand binding (i.e., perturbations of helices, -sheets, hairpin turns and unordered coils), 119,120 CD spectroscopy (185-260 nm) was used to delineate how H2PrPyrr and its metal chelates impact the secondary structure of the protein (Fig. 10).The CD spectrum of native HSA exhibits a double minimum (208  1 nm and 222  1 nm), consistent with a dominantly α-helical structure for the protein. 121The electronic transitions in this region are assigned to -* (208 nm minimum), n-* (222 nm minimum), and -* (196 nm maximum) transitions of the amide groups of the peptide backbone. 121,122From Figs. 10a-10d it is clear that HSA binds the present group of compounds not only with a high affinity (Table 1), but also with relatively minor changes in its secondary structure elements since the spectrum of the native protein is similar, though not identical, to the spectra in the presence of saturating ligand doses.Specifically, the wavelengths of the band maxima and minima are largely unshifted, with only slight changes in intensity evident (most noticeably for the 196-nm band).The isodichroic points are generally quite difficult to locate, but those that are discernible appear to be somewhat sensitive to the identity of the ligand bound by HSA.In particular, there is some agreement between the wavelengths of the isodichroic points for HSA•{H2PrPyrr} and HSA•{Pd II (X)} as well as those for HSA•{Ni(PrPyrr)} and HSA•{Pt(PrPyrr)}.
The CD data were analysed further (JASCO Spectra Manager™) to calculate the percentage composition of α-helical coils, β-sheets, turns, and unordered coils present for each HSAligand complex (Table S7).The dominant secondary structure domains are -helices (50%), and unordered coils (31%), while -sheet domains (8-19%) show the largest variation with the identity of the bound ligand.Interestingly, the solution state secondary structure composition for HSA differs from that of native HSA in the solid state 45 (68.5 % -helix, 0% -sheet, 9.6% turns, and 21.9% unordered coils; PDB code 1BM0 analysed with BESTSEL 123 ).Our CD data are, however, consistent with solution state spectral decompositions reported by others, 124- 126 and it is generally accepted that enhanced subdomain mobility and general thermal motion/disorder accounts for the decrease of α-helicity. 120,121n Fig. 10e, we have plotted the change in the secondary structure components of HSA as the percentage difference between the domain content for the native protein and that determined for the ligand-bound complex at saturation (Table S7).For the -helix, turn, and coil regions of the structure, the ligand-induced perturbation falls within the range  1.3%, which is comparable to the X-ray data for HSA bound to warfarin. 104owever, for the -sheet domains, the ligand-induced perturbation is significantly greater with a range of  3.1%.The complexes HSA•{H2PrPyrr} and HSA•{Pd(PrPyrr)} exhibit the largest increase and decrease in -sheet content, respectively.Importantly, the ligand-induced changes in the protein secondary structure correlate with the thermodynamics of ligand binding measured independently by fluorescence spectroscopy.Thus, as shown in Fig. 10f, the change in the helix composition, ∆(-helix), is linearly dependent on the enthalpy and entropy changes for the reaction.Specifically, as the ∆= and ∆< values become more negative and positive, respectively, the -helix content increases.From eq. 13, the relationship involving ∆(-helix) reflects a linear inverse correlation with ∆B for the reaction.Simply put, a more exergonic reaction correlates with an increase in -helicity (and thus intrinsic protein stability 127 ) induced by ligand binding.Because -helices are usually more stable thermodynamically than -sheets, 128,129 the expectation is that an increase in sheet content for HSA in solution induced by ligand binding will correlate with a less negative ∆B for the reaction.Although statistically poorer than the 3D fit for ∆(-helix), ∆(-sheet) does correlate linearly with ∆= and ∆< for the reactions (Fig. S33, R 2 = 0.743), with the expected inverse relationship to that delineated for ∆(-helix).

Longer wavelength CD spectroscopy and in silico data
Near-UV CD spectra provide a valuable fingerprint of protein tertiary structure perturbations, specifically those involving Trp (285-300 nm), Tyr (280 nm), and Phe (250-270 nm).Additionally, induced circular dichroism (ICD) spectra are informative when an optically inactive guest interacts with a protein and gains chirality from its chiral host. 122In this section, we will present the experimental CD spectra for HSA bound to the ligands of interest and include an integrated interpretation of the data based on the results of our GLIDE XP 130,131 (Schrödinger Release 2022-3) molecular docking and TD-DFT simulations (ONIOM 132 CAM-B3LYP 133 /SDD 134 /GD3BJ: 135 UFF 136 ) of the electronic structures of the protein-bound metal chelates.Fig. 11 shows the near-UV and visible CD spectra of HSA recorded in native form and in the presence of identical saturating doses (12 M) of H2PrPyrr, Ni(PrPyrr), Pt(PrPyrr), and Pd II (X), where X represents the ligands of eq. 8. Using difference spectroscopy, the ICD spectra were resolved for all ligands.The intensity of the ICD response followed the order Pd II (X) > Pt(PrPyrr) > Ni(PrPyrr) > H2PrPyrr, which is identical to the order of the affinity constants (log , Table 1) for the compounds.
Evidently, the stronger the HSA-ligand interaction, the greater the induced chirality in the formally achiral ligand.Importantly, the ICD spectra indicate that the complexes neither demetallate nor suffer imine hydrolysis or ligand dissociation when binding to HSA.The evidence is discussed further below and supported by macromolecular simulation data.Since ligand dissociation often accompanies HSA binding by simple metal complexes such trans-[RuCl4(1H-indazole)2]  (KP1019) 54 and cisplatin, 137 the present findings are significant and highlight the potential for fully reversible transport of the present class of compounds by HSA and, more generally, polydentate chelates and macrocyclic complexes of metal ions.
For HSA•{H2PrPyrr} and HSA•{Ni(PrPyrr)}, the weaker ICD spectra of the ligands permit delineation of subtle changes in the fine structure CD bands of the protein host (Fig. S34).Clear perturbation of the signal from Tyr (277 nm) is evident in the difference spectrum for HSA•{H2PrPyrr} (Fig. 11a), consistent with this ligand and/or its protonated forms binding adjacent to one or more tyrosine residues, perhaps aided by H-bonding. 112rom our in silico docking data obtained with different grids (vide infra, Fig. S35, Tables S10 and S11) there are several possible binding sites found within 10-27 Å of Trp-214 that can account for (i) FRET-based quenching of this fluorophore and (ii) location of the ligand within 4-10 Å of at least one Tyr residue.Interestingly, none of the binding sites are directly in the main drug-binding pockets of the protein (i.e., Sudlow's sites I and II).This evidently explains why blocking these sites with warfarin and ibuprofen neither inhibited the uptake of H2PrPyrr (and/or its protonated forms) nor resulted in displacement of the probe drugs at higher ligand doses (Table S8, Fig. S30).
For HSA•{Ni(PrPyrr)}, the near-UV CD signals from Phe (269 nm), Tyr (279 nm), and Trp-214 (290 nm) showed significant perturbations (Fig. 11b), commensurate with binding of the Ni(II) chelate in one or more sites containing Phe and Tyr residues and within range of Trp-214 for FRET-based quenching (1-3 nm in an isolated biomolecule). 138From our GLIDE XP data, Ni(PrPyrr) favours binding in Sudlow's site I (Tables S9, S11) and possibly also Sudlow's site II.Medium-range dipole-dipole interactions between Ni(PrPyrr) and the closest Phe and Tyr residues within the binding pocket of Sudlow's site I (Table S10, Fig. S36) are thus feasible given the spatial proximity of the chromophores (e.g., Tyr-150Ni, 7.0 Å; Phe-149Ni, 8.5 Å), accounting for the experimental near-UV CD data.
Regarding assignment of the near-UV and visible ICD spectrum of HSA•{Ni(PrPyrr)}, peaks with positive (300-340 nm; 375-430 nm) and negative ellipticity (457 and 477 nm) are observed (Fig. 11b).Since unbound Ni(PrPyrr) is achiral (exact Cs symmetry), it has no CD spectrum.The electronic absorption spectrum of the protein-free Ni(II) chelate is characterized by absorption maxima with  > 4  10 3 M 1 cm 1 at 313, 369, 388, 409, and 433 nm in pure acetonitrile (Fig. S37).Considering the TD-DFT data for Ni(PrPyrr) (Table S12), these bands may be assigned to MLCT (metal-to-ligand charge transfer) transitions from filled MOs with mainly metal 3d-orbital character (and some ligand  character) to vacant MOs with mainly ligand * character.(A broad, weaker band with  200 M 1 cm 1 at 523 nm encompassing one or more d-d transitions of the complex is also evident.)The MLCT absorption bands at 369 and 388 nm red-shift by 20-30 nm upon uptake of the Ni(II) chelate by HSA, giving rise to the positive ICD bands at 404 and 421 nm.Similarly, those at 409 and 433 nm equate with the negative ICD bands at 457 and 477 nm (Fig. S37).The fact that signals from Ni(PrPyrr) are discernible in the experimental CD spectrum of the protein complex, HSA•{Ni(PrPyrr)}, strikingly confirms that the intact Ni(II) chelate is noncovalently bound within the protein.Significantly, TD-DFT simulations with the best-scoring pose for Ni(PrPyrr) docked within subdomain IIA of the protein (Table S9, Fig. S38) support this conclusion by reproducing the main features of the experimental ICD spectrum.
The ICD spectrum recorded for HSA•{Pt(PrPyrr)} is akin to that of HSA•{Ni(PrPyrr)} and may be similarly assigned after taking into account the blue shift in the transition energies associated with the heavier metal ion.Indeed, the ICD spectra suggest that Pt(PrPyrr) targets the same binding site(s) as Ni(PrPyrr), consistent with our in silico docking data (Table S10).Pt(PrPyrr).The spectra were recorded at 298 K in 50 mM KH2PO4 buffer at pH 7.50 and smoothed using a relatively fine Lowess function (0.07 span).The graphical insets are plots of the difference spectra generated by subtraction of the spectrum of native HSA from that of the HSA-ligand complex; they represent the induced CD (ICD) spectrum of each achiral protein-bound ligand.(For the Pd(II) complex, X represents the ligand combinations of the equilibrium species depicted in eq.8.) Perturbations in the protein structure around Phe (250-270 nm), Tyr (280 nm), and Trp (285-300 nm) residues may also be resolved in some difference spectra.Key maxima and minima are indicated (in nm).H2PrPyrr has no visible absorption spectrum; undulations > 350 nm are noise.Selected unsmoothed spectra are available in the ESI (Fig. S34).
The ICD spectrum of HSA•{Pd II (X)} is the most intense of the series and shows features reminiscent of the above M(PrPyrr) chelates, specifically the peaks with positive ellipticity at 358 and 373 nm and the peak with negative ellipticity at 414 nm.One clear difference is the stronger set of positive ellipticity bands from 280-305 nm ( 1.7 mdeg).These bands overlap the CD signals expected from perturbed Tyr and Trp residues; in the present system, they also reflect the interaction of HSA with the C1-symmetry aqua complex, [Pd(HPrPyrr)(OH2)] + .We used GLIDE XP to dock both species depicted in eq. 8, i.e., Pd(PrPyrr) and [Pd(HPrPyrr)(OH2)] + , into HSA (PDB code: 1HA2) using a large 40 3 Å 3 grid centred on Trp-214 or the warfarin binding site as well as a similarly-constructed grid for the diazepam adduct of HSA (PDB code: 2BXF).The two best-scoring for the Pd(II) chelates are located at sites in subdomain IIA of HSA (Tables S9-S11, Fig. S39).Specifically, Pd(PrPyrr) favours the larger Sudlow's I site (Gdock = 5.56kcal mol 1 ) in a region directly adjacent to the warfarin binding site (Fig. S40), while [Pd(HPrPyrr)(OH2)] + binds towards the protein surface between two -helical coils belonging to subunits IA and IIA (Gdock = 5.53kcal mol 1 ; Fig. S39).Given the reaction stoichiometry and fluorescence quenching data for the HSA-Pd(II) system (Table 1), any attempt to calculate the ICD spectrum of the system would need to include both Pd(II) chelates.
We used a hybrid QM:MM TD-DFT simulation (ONIOM method 139 ) in which the two Pd(II) chelates were assigned to the quantum mechanics (QM) layer of a model taken directly from the optimized GLIDE XP structure (Fig. S39).Three side-chains belonging to residues closely interacting with Pd(PrPyrr) in Sudlow's site I (Arg-257, His-288, and Tyr-150) were added to the QM layer to calculate the most accurate CD spectrum (Fig. S41).The remining atoms of the protein were simulated using molecular mechanics (UFF 136 ).The results of the simulation are illustrated in Fig. 12.Briefly, the calculated CD spectrum of the system broadly matches the key features of the experimental ICD spectrum.Significantly, the analogous TD-DFT CD spectra for each Pd(II) chelate independently bound to HSA poorly matched the experimental ICD spectrum (Fig. S41), firmly supporting our understanding of the reaction between HSA and the Pd II (X) system (dual chelate binding).One caveat is that we do not know with certainty whether [Pd(PrPyrr)(OH2)] + binds non-covalently (as modelled) or covalently (e.g., via a surface residue such as His, Glu, or Cys substituting Pd-bound water) to HSA.Although including noncovalently bound [Pd(PrPyrr)-(OH2)] + in the TD-DFT simulation was obligatory to match the positive ellipticity bands from 275 to 325 nm in the experimental CD spectrum, only X-ray data can answer this question.
Finally, the CD spectra for all HSA-ligand complexes closely match the shoulder at 262 nm of the native protein, suggesting the 17 disulphide bonds of the protein remain chemically intact upon ligand uptake.This reflects the conformational integrity observed for the protein from its far-UV CD spectra (Fig. 10) and confirms that any release of the d 8 metal ion via chelate dissociation, culminating in side-reactions to form possible metal-thiolate adducts with Cys residues, is negligible.

Macromolecular simulations
Docking studies (GLIDE XP) were used to predict possible HSA binding sites for both the present ligand set and control compounds (warfarin, diazepam, ibuprofen).Since rigid X-ray structures of proteins are used as targets for flexible ligand docking, such methods offer mainly approximate insights on how a group of ligands may bind to the target.Notably, docking scores usually do not correlate quantitatively with experimental thermodynamic data (G values), 140,141 because a docking experiment does not try to emulate a physical reaction between the macromolecule and incoming ligand in solution.Thus, only qualitative parallels are drawn between the thermodynamic data (Tables 1, 2, and S8) and docking scores of Tables S9 and  S11, with a specific focus on in silico data that add insights to the experimental facts (CD spectra, values, and fluorescence quenching data).The key trends are enumerated below.
1.The existence of multiple binding sites within HSA for the current ligand set is predicted from the narrow range of docking scores (Gdock  3 to 7 kcal mol 1 ) and repeatedly favoured binding locations (7 in total including Sudlow's I and II), consistent with X-ray data for diverse drug and ligand types bound by HSA. 1052. The binding site locations were all within 5-20 Å of Trp-214, accounting for facile quenching of its emission by the ligands (Table 1).Furthermore, all were within 3.5-15 Å of tyrosine and/or phenylalanine residues (Table S10), affording observable CD signal perturbations between 250 and 295 nm for at least two HSA-ligand complexes (H2PrPyrr and Ni(PrPyrr), Fig. 11).3. Warfarin bound within HSA (subdomain IIA), generally enhances the docking scores for ligand uptake when forming a multi-ligand complex (Table S9).Thus, Gdock improves from 5.17 to 6.00 kcal mol 1 when Ni(PrPyrr) co-binds with warfarin (Fig. S36) due to enhanced lipophilic van der Waals and electrostatic energy terms.In contrast, the docking score worsens from 5.56 to 4.01 kcal mol 1 for Pd(PrPyrr) when warfarin is co-bound in Sudlow's site I due to van der Waals repulsion between the metal chelate and the probe (Fig. S40).These trends qualitatively mirror the experimental log data for several ligands (Table S8). 4. Ibuprofen bound within Sudlow's site II (subdomain IIIA, as well as at its secondary site, IIA-IIB) generally decreases the docking scores for the ligands with HSA.This trend is mirrored by the reaction thermodynamics (e.g., Ni(PrPyrr) and Pd II (X) binding to HSA, Table S8).Although these ligands do not compete with ibuprofen for the same binding sites (IIIA and IIA-IIB) in the docking experiment (Table S9), ibuprofen seemingly redirects their uptake to less-favourable binding sites relative to those in native HSA.In reality, Pd II (X) and Ni(PrPyrr) easily displace ibuprofen from HSA (Figs. 6 and S28), suggesting that these ligands can bind in subdomain IIIA (Table S11) or that the release of ibuprofen is triggered by allosteric effects when the ligands bind at the alternative locations.5.The grid chosen for the docking simulation affects the docking-score order and in silico site specificity of the ligands for HSA (Tables S9 and S11).This limitation is well known with the use of a rigid receptor grid, as done here, considered best for some metalloproteins. 142For this reason, and because there is no other way to guide the grid choice for novel ligands when X-ray data are absent, we have mainly used the data generated with a rigid receptor grid (Table S9) as opposed to that obtained with a soft receptor and ligand potential (Table S11).The results obtained with the softer potential may, however, be equally valid.

Conclusions
The interaction of H2PrPyrr and its Ni(II), Pd(II), and Pt(II) chelates with human serum albumin (HSA) was investigated by complementary spectroscopic techniques to understand how the identity of the d 8 metal ion affects ligand uptake by the protein.All ligands quenched the intrinsic Trp-214 fluorescence of HSA via a static quenching mechanism and gave values typical of small molecules binding to HSA (10 4 -10 6 M 1 ). 143,144e markedly higher affinity of Pd(PrPyrr) for HSA ( 10 9 M 1 ) coupled with a reaction stoichiometry tending towards 2:1 (ligand:HSA) at 288 K was explained in terms of simultaneous uptake of two Pd(II) species from an equilibrium mixture comprising Pd(PrPyrr) and [Pd(HPrPyrr)(OH2)] + , the latter structure being related to the crystallographically-characterized chlorido complex, PdCl(HPrPyrr).Evidence supporting this hypothesis was obtained directly from the induced CD spectrum of HSA bound simultaneously to the pair of Pd(II) complexes-a binding model well-described by TD-DFT calculations on the macromolecular complex predicted by docking simulations.Moreover, ESI-MS data directly confirmed formation of HSA•{ligand}n species with n = 1 and 2 for the Ni(II), Pd(II), and Pt(II) chelates.The present ligands bind to HSA with negative ∆=, ∆B and ∆< values, reflecting a spontaneous enthalpydriven process governed by van der Waals (London dispersion) forces as well as H-bonding, particularly for the metal-free ligand (H2PrPyrr).Our docking simulations support the notion (gleaned from spectroscopy) that each ligand has more than one binding site in HSA and that ligand uptake minimally perturbs the protein's secondary structure.Finally, the induced CD spectra recorded for the metal chelates bound to HSA strikingly confirm uptake of the intact complexes without imine group hydrolysis or demetallation.

Fig. 1
Fig. 1 (a) Structures of the bis(pyrrole-imine) ligand H2PrPyrr (N,N'-bis[(1E)-1H-pyrrol-2ylmethylene]propane-1,3-diamine) and its neutral metal chelates with square planar d 8 metal ions relevant to this work.The method for metalation of the ligand is illustrated for the synthesis of Pd(PrPyrr); the solvate water molecules enter the bulk solvent.(b) Xray structure of HSA bound to ibuprofen (redrawn from PDB code 2BXG) illustrating the two main small molecule binding sites.Sudlow's site I is larger than Sudlow's site II and compounds that bind in this pocket perturb the fluorescence from Trp-214.The protein secondary structure elements are depicted schematically, coloured by domain, and labelled with Roman numerals and Arabic letters.H atoms (calculated positions) were added to the bound drug and Trp-214 to enhance their visualization.

Fig. 2
Fig.2(a) Protonation of Pd(PrPyrr) and H2PrPyrr studied by UV-visible spectroscopy in acetonitrile solution at 37 C.The spectra were smoothed with an FFT filter algorithm (5point window, OriginPro 2021).The electronic spectrum of the protonated free base ligand, [H4PrPyrr] 2+ , was deconvoluted into three Gaussian bands (fit R 2 = 0.998,  2 = 1.6 × 10 4 ).There are no bands below 250 nm or above 350 nm in the spectrum.(b)1 H NMR spectra of Pd(PrPyrr) recorded in CD3CN in the absence (blue) and presence (red) of 2 eq of HCl in D2O recorded at 37 C.The DFT-calculated structure of PdCl(HPrPyrr) is shown along with signal assignments in the experimental NMR spectrum that are based on the isotropic shielding tensors calculated for the protons in the complex.DFT-calculated bond distances are in Å.The signal from the central methylene group of the propyl bridge (not shown) is overlapped with the solvent peaks at 1.97 ppm (see Fig.S2).The minor signals marked with an asterisk possibly belong to the solvated complex (Cl  exchange).

Fig. 3
Fig. 3 (a) Partly labelled view of the low-temperature X-ray structure of Pd(PrPyrr).Two of the four independent molecules (A and B) in the asymmetric unit are illustrated.(b) View of the low temperature X-ray structure of PdCl(HPrPyrr).Thermal ellipsoids are rendered at the 35% and 50% probability levels for Pd(PrPyrr) and PdCl(HPrPyrr), respectively.Hydrogen atoms are drawn as spheres with an arbitrary radius.

Fig. 4
Fig. 4 Emission spectra of human serum albumin (HSA, 5.0 M) recorded as a function of the concentration of Pd(PrPyrr) at 298 K (pH 7.50).The spectra are fitted by single Gaussian functions to locate the emission maxima.Correlation coefficients, R 2 , ranged from 0.993 to 0.997.The emission intensity was arbitrarily set equal to zero at 300 nm for all spectra.Analogous spectra for the reaction of HSA with H2PrPyrr, Ni(PrPyrr), and Pt(PrPyrr) are given in Figure S21 (ESI).The wavelength shift accompanying ligand uptake (Δ ) is plotted as the inset to the main figure (upper right).

Fig. 5
Fig. 5 (a) Stern-Volmer (SV) fluorescence intensity ratio plots for human serum albumin (HSA, 5.0 M) recorded as a function of the concentration of Pd(PrPyrr) and temperature (50 mM KH2PO4, pH 7.50).Error bars are ESD's based on the average of three independent determinations.For clarity, data recorded at 315 K are not shown.The data are well-fitted by eq. 2 for nonlinear SV emission behaviour with appreciable static quenching.(b) Double logarithm plot of the fractional change in fluorescence intensity for human serum albumin (HSA, 5.0 M) recorded as a function of the concentration of Pd(PrPyrr) and temperature (50 mM KH2PO4, pH 7.50).Error bars are ESD's based on the average of three independent determinations.The data are described by eq. 7, which affords the affinity constant and stoichiometric coefficient for the reaction (Table1).

Fig. 7
Fig.7Proposed mechanism of interaction of a pH 7.5 equilibrium mixture of Pd(PrPyrr), comprising the tetradentate Pd(II) chelate "Pd a " and its tridentate aqua complex "Pd b ", with human serum albumin (HSA).The experimental affinity constants and fluorophore excitation wavelengths are indicated; the diagram summarizes our best interpretation of the experimental data (Tables1 and S8; Figs.5, 6 and S26) and in silico binding site specificity data.In Part A, the reaction of native HSA with the two Pd(II) chelate species is shown.In addition to the main drug binding sites in subdomains IIA and IIIA of HSA (Sudlow's I and II, respectively), the existence of a third binding site unique to the Pd(II) aqua complex is required to account for the spectroscopic and thermodynamic data.This binding site accommodates [Pd(HPrPyrr)(OH2)] + and was found to be at the interfacial region between subunits IA-IIA/IIIA-IIA of the protein (GLIDE XP, TableS9).Pd(PrPyrr) preferentially binds in subdomain IIA, but also has some affinity for subdomain IIIA (indicated by different intensity purple triangles).All bound Pd(II) chelates will contribute to efficient quenching of the emission from Trp-214 (dashed arrows).In Part B, HSA•{warfarin} is the reaction target.The binding pockets have diminished affinity due to allosteric effects (partial inhibition; dotted grey lines) from bound warfarin.This leads to step-wise binding of the Pd(II) chelates with the affinity order [Pd(HPrPyrr)(OH2)] + > Pd(PrPyrr).Because the fluorescence titration monitors the emission from the drug probe directly (warfarin), its displacement and subsequent dynamic quenching by the solvent is readily detected.Pd(PrPyrr) fractionally occupies Sudlow's sites I and II in the product.Part C depicts the analogous reaction when the macromolecular target is HSA•{ibuprofen}n.Notably, displacement of both warfarin and ibuprofen is mediated by Pd(PrPyrr) since [Pd(HPrPyrr)(OH2)] + has a low affinity for these drug binding sites (TableS9) and Ni(PrPyrr) gives similar results (FigureS28; TablesS8 and S9).Although spectroscopy and simulations are insightful here, X-ray crystallography will ultimately be required to definitively assign the proposed binding sites.

Fig. 8
Fig.8(a) Comparison of the thermodynamic parameters governing the reactions of the free base ligand and metal chelates with HSA (T = 298 K).All measurements were done in triplicate.Derived parameters were individually averaged.(b) Linear van't Hoff plots (eq.12) for the reactions of H2PrPyrr, Ni(PrPyrr), and Pt(PrPyrr) with HSA in 50 mM KH2PO4 buffer at pH 7.50.The curve for Pd II (X), where X represents the ligand combinations given in eq. 8, is nonlinear and is well-described by the modified van't Hoff model (eq.9).Error bars are estimated uncertainties of the mean for both graphs.

Fig. 10
Fig. 10 Plots of the far-UV CD spectra of native HSA and the protein incubated with saturating doses of (a) H2PrPyrr, (b) Ni(PrPyrr), (c) Pd II (X), and (d) Pt(PrPyrr) recorded at 298 K in 50 mM KH2PO4 buffer at pH 7.50.(For the Pd(II) complex, X represents the ligand combinations of the equilibrium species depicted in eq.8.) The spectra were smoothed using a Lowess function (0.07 span); unsmoothed spectra at ligand doses ranging from 4 M to 12 M are given in ESI Fig. S32.Band maxima and minima (black numerals) as well as isodichroic points (blue numerals) are indicated (in nm).(e) Graph of the percentage change in the protein secondary structure delineated by fitting the CD spectra in Parts (a)-(d).Data for warfarin were determined from the X-ray structures of the native (PDB code: 1BM0) and ligand-bound (PDB code: 2BXD) proteins using the program BESTSEL.Although the protein is minimally perturbed by ligand uptake, the -sheet content decreases by as much as 3% in the case of the complex formed with the Pd(II) chelate.(f) Three-dimensional graph showing how the change in the -helical coil content of the protein correlates with the experimentally determined values of ∆= and ∆< for the reaction of each compound with HSA (T = 298 K).The relationship, ∆(-helix) = 2.80(25) 0.0766(64)  ∆= + 0.0242(25)  ∆<, reflects the fact that ∆B for the reaction becomes more favourable (negative) as the -helix content of the protein increases, which is dependent on the identity of the bound ligand.

Fig. 11
Fig.11Plots of the near UV-visible CD spectra of native HSA (5 M) and the protein incubated with a saturating dose (12 M) of (a) H2PrPyrr, (b) Ni(PrPyrr), (c) Pd II (X), and (d) Pt(PrPyrr).The spectra were recorded at 298 K in 50 mM KH2PO4 buffer at pH 7.50 and smoothed using a relatively fine Lowess function (0.07 span).The graphical insets are plots of the difference spectra generated by subtraction of the spectrum of native HSA from that of the HSA-ligand complex; they represent the induced CD (ICD) spectrum of each achiral protein-bound ligand.(For the Pd(II) complex, X represents the ligand combinations of the equilibrium species depicted in eq.8.) Perturbations in the protein structure around Phe (250-270 nm), Tyr (280 nm), and Trp (285-300 nm) residues may also be resolved in some difference spectra.Key maxima and minima are indicated (in nm).H2PrPyrr has no visible absorption spectrum; undulations > 350 nm are noise.Selected unsmoothed spectra are available in the ESI (Fig.S34).

Fig. 12 (
Fig. 12 (a) Comparison of the experimental ICD spectrum (  9 mdeg) recorded for HSA•{Pd II (X)} at pH 7.5 (Fig. 11c) and the CD spectrum calculated using hybrid QM:MM TD-DFT simulations (CAM-B3LYP/SDD/GD3BJ:UFF) for the top-scoring docked poses of Pd(PrPyrr), Pd a , and [Pd(HPrPyrr)(OH2)] + , Pd b , co-bound to HSA (PDB code: 1HA2).Both poses were required to get a reasonable match with the experimental data (Fig. S41); a band width of 2500 cm 1 (hwhm) was used for the TD-DFT data.(b) Structural model used for the TD-DFT simulations showing the location of the two Pd(II) chelates (orange C atoms, van der Waals radii) and Trp-214 (purple C atoms) in different regions of subdomain IIA (see also Fig. S39).(c) View of the metal chelate binding sites containing the best poses of the two Pd(II) complexes and their closest residues (1-letter codes).His-247 hydrogen bonds to the coordinated water molecule of [Pd(HPrPyrr)(OH2)] + (O-HN, 2.08 Å, dashed yellow cylinder).Only polar hydrogen atoms are shown.