Sheldon
Sookai
a and
Orde Q.
Munro
*ab
aMolecular Sciences Institute, School of Chemistry, University of the Witwatersrand, PO WITS 2050, Johannesburg, South Africa. E-mail: S.Sookai92@gmail.com
bSchool of Chemistry University of Leeds, Woodhouse Lane, LS2 9JT, UK. E-mail: O.Munro@leeds.ac.uk
First published on 4th September 2023
Three bis(pyrrolide-imine) Pt(II) chelates were synthesised and characterized with different bridging alkyl groups, specifically 2-hydroxypropyl (1), 2,2-dimethylpropyl (2), and 1,2-(S,S)-(+)-cyclohexyl (3). Novel compounds 1 and 2 were analysed by single-crystal X-ray diffraction (space group P). The asymmetric unit of 1 comprises three independent molecules linked by hydrogen bonds involving the OH groups, forming a trimeric supramolecular structure. The Pt(II) chelates were reacted with human serum albumin (HSA) to investigate how the ligand bound to the Pt(II) ion influences the compound's affinity for HSA. Fluorescence quenching data obtained for native HSA and HSA bound to site-specific probes (warfarin, subdomain IIA; ibuprofen, subdomain IIIA) indicated that the three Pt(II) chelates bind close enough (within ∼30 Å) to Trp-214 to quench its intrinsic fluorescence. The bimolecular quenching constant (kq) was 103–104 -fold higher than the maximum diffusion-controlled collision constant in water (1010 M s−1) at 310 K, while the affinity constants, Ka, ranged from ∼5 × 103 to ∼5 × 105 at 310 K, and followed the order 1 > 3 > 2. The reactions of 1 and 3 with HSA were enthalpically driven, while that for 2 was entropically driven. Macromolecular docking simulations (Glide XP) and binding site specificity assays employing site-specific probes and UV-vis CD spectroscopy indicated that 1 and 2 target Sudlow's site II in subdomain IIIA, minimally perturbing the tertiary structure of the protein. Well-resolved induced CD signals from 1 and 2 bound to HSA in subdomain IIIA were adequately simulated by hybrid QM:MM TD-DFT methods. We conclude that the structure of the bis(pyrrolide-imine) Pt(II) chelate measurably affects its uptake by HSA without detectable decomposition or demetallation. Such compounds could thus serve as metallodrug candidates capable of utilising an HSA-mediated cellular uptake pathway.
Pt(II) complexes are among the most widely studied metal complexes in medicine. Since the FDA approval of cisplatin in 1978 and its significant clinical success, work on a plethora of Pt(II) metallodrug candidates has flourished18 mainly to discover derivatives with reduced toxicity alongside responding to the emergence of cisplatin resistant tumours.19 It is postulated that one of the main reasons for the poor therapeutic index and toxic side effects of Pt(II) compounds is their interaction with serum proteins, in particular human serum albumin (HSA).20 For reference, after intravenous administration, between 68–98% of cisplatin is bound by serum proteins, mostly HSA, and up to five cisplatin molecules bind to a single HSA molecule.21 To maximize the activity and improve the safety of Pt(II) metallodrugs more broadly, it is essential to understand the binding mechanism of Pt(II) complexes to HSA (and other macromolecules) and to develop novel candidates with high biological stability and good solubility.
HSA is synthesised in the liver and is the most abundant serum protein with a concentration of 35–50 g L−1. The protein has three main functions in vivo: (i) aiding to maintain physiological pH, (ii) maintaining osmotic pressure, and (iii) the transport of endogenous and exogenous compounds.22 HSA comprises a single polypeptide chain with a heart-shaped tertiary structure consisting of 585 amino acids, culminating in a molar mass of 66.5 kDa (Fig. 1).23 The protein has three domains (I, II, and III), each made up of 10 α-helices and two subdomains (A and B).23,24 The two main drug binding sites are located in subdomains IIA and IIIA and are commonly known as Sudlow's site I and Sudlow's site II.25,26 Apart from Sudlow's sites I and II, HSA has several other binding sites including 4 thyroxine, 7 fatty acid, and several known metal ion binding sites.27 Delineating the mechanistic details of metal complex uptake by HSA is crucial to understanding in vivo pharmacokinetic and pharmacodynamic data of such systems, given the protein's role as the quintessential drug transporter.28
![]() | ||
Fig. 1 (a) Schematic synthesis procedure for the metalation of the three bis(pyrrole-imine) ligands with K2[PtCl4] relevant to this work. (b) X-ray structure of HSA bound to diflunisal (redrawn from PDB code 2BXE) illustrating the protein's two main drug binding sites, which are commonly called Sudlow's sites I and II. Site I (subdomain IIA) is larger than Site II (Subdomain IIIA). Diflunisal itself has three binding sites and the primary (1°) and tertiary (3°) locations are within Sudlow's sites II and I, respectively. The protein secondary structure elements are depicted schematically, coloured by domain, and labelled with Roman numerals and Arabic letters. |
By employing isoelectronic Ni(II), Pd(II), and Pt(II) (nd8) chelates of the propyl-bridged structural congener of 1 (R = CH2CH2CH2; H2PrPyrr), we recently showed that the thermodynamics governing complex uptake by HSA depended on the identity of the metal ion.29 Specifically, TΔS/ΔG and ΔH/ΔG together became increasingly positive (enthalpically driven) in the order Pt(II) < Pd(II) < Ni(II) < H2PrPyrr, thereby inversely tracking the metal ionic radii and polarizability order (5d > 4d > 3d).29 A key question emerging from this study was whether varying the ligand chelated to a given metal ion (e.g., Pt2+) affects its uptake by HSA and, if so, how? To answer this question, we synthesized three structurally-related, yet discrete, bis(pyrrole-imine) Schiff base ligands, namely 1,3-bis{[(1E)-1H-pyrrol-2-ylmethylene]amino}propan-2-ol, 2,2-dimethyl-N,N′-bis[(1E)-1H-pyrrol-2-ylmethylene]propane-1,3-diamine and (1S,2S)-N,N′-bis[(1E)-1H-pyrrol-2-ylmethylene]cyclohexane-1,2-diamine. The ligands were then metalated with Pt(II) to obtain the desired series of square planar Schiff base chelates 1–3 (Fig. 1a). Steady-state fluorescence and CD spectroscopy were used to investigate the binding of the complexes to HSA under physiological conditions, which revealed how the pyrrole-imine Schiff base ligand architecture influences the affinity and preferred binding site(s) targeted by the Pt(II) chelates. Finally, molecular docking (GLIDE XP) and TD-DFT simulations were employed to confirm the spectroscopically identified binding sites.
![]() | ||
Fig. 2 Spectroscopic characterization and structural assignments for 1. (a) Proton NMR spectrum (298 K, 400 MHz, DMSO-d6). The insets show the imine CH signals for 2 and 3. (b) Carbon NMR spectrum (298 K, 100 MHz) with the labelled and assigned DFT-calculated structure of 1. The inset highlights the 13C–195Pt spin–spin coupling for pyrrole carbon 1. (c) Electronic spectrum of 1 recorded in KH2PO4 buffer (50 mM, pH 7.5, 1% DMSO (v/v)). The TD-DFT calculated spectrum scaled (ε and λ) to best match the experimental spectrum is shown with selected transition assignments (bandwidth = 2500 cm−1, fwhm). The four key frontier MOs accounting for the bands at λ > 325 nm are shown. A full list of transition assignments is given in Table S9.† |
The 1H NMR spectrum for 1 exhibits two additional noteworthy features: (i) a sharp doublet for the hydroxyl proton resonance and (ii) resolvable splitting patterns for the methylene protons of the ligand's alkyl bridge. The former suggests that intermolecular H-bonding limits solvent-mediated exchange broadening of the hydroxyl proton resonance, allowing delineation of the J-coupling to methine proton g (3Jh–g = 4.5 Hz, Fig. 2). Regarding the methylene groups of the ligand's alkyl bridge, the axial (e) and equatorial (f) geminal protons resonate at 3.82 and 3.67 ppm, respectively. The resolved J-couplings for these signals indicate limited dynamic motion for the chelate ring (specifically its half-chair inversion). The experimental coupling constants 2Jf–e (14 Hz, geminal) and 3Jf–g (7.6 Hz) were confirmed by a static DFT calculation of the spin–spin coupling constants with the geometry-optimized structure (Fig. 2b, CAM-B3LYP36/def2-QZVP37/GD3BJ38 level of theory) in a DMSO solvent continuum (−16.1 Hz and 5.1 Hz, respectively). Formation of a supramolecular complex in DMSO (an oligomer), as seen in the asymmetric unit of the X-ray structure of 1 (vide infra), is both feasible and consistent with both hallmark features of the 1H NMR spectrum of 1. Interestingly, the methylene protons of 2 resonate as a broad singlet (Fig. S6†) wherein no geminal or 4J spin–spin coupling is evident, suggesting that dynamic/conformational exchange for the six-membered chelate ring is operative in this system.
The more intense band at 382 nm is similar with 77% 1[Pt(5dxz),π → Pt(6pz),π*] character and a prominent vibronic shoulder (364 nm) overlapping a weaker transition of mainly ligand π → π* character (HOMO−2 → LUMO, Table S9†). Finally, the far-UV bands are the most intense with the 313 nm band in the experimental spectrum (ε = 1.53 × 04 M−1 cm−1) correlating with the 327 nm band of the calculated spectrum. This interesting transition is almost pure MMLCT (metal-to-metal–ligand charge transfer) in character, i.e., 97% 1[Pt(5dz2) → Pt(6pz),π*] with a sharp vibronic satellite at 302 nm. The remaining band at 277 nm may be assigned from the DFT-calculated counterpart at 296 nm, which is based on two closely-spaced transitions: 57% HOMO−2 → LUMO+1 and 53% HOMO−3 → LUMO. Significantly, the latter transition, like the HOMO → LUMO transition, has part MLMLCT character (metal–ligand-to-metal–ligand charge transfer) as it originates from an MO which is a substantial admixture of the Pt 5dyz orbital with a bis(pyrrolide-imine) π MO, i.e., 1[Pt(5dyz),π → Pt(6pz),π*].
Complexes 1 and 2 exhibit Pt–Npyrrole distances of 2.011 (3) and 2.013 (3) Å, respectively, while the Pt–Nimine distances were marginally shorter (∼0.4%), measuring 2.005 (3) and 2.002 (3) Å for complexes 1 and 2, respectively. Shorter Pt–Nimine bond lengths are typical for bis(pyrrolide-imine) chelates.10 The Pt–N bond distances of 1 and 2 are in agreement with the enantiomers of 3 reported by Shan et al.,10 with coordination group distances ranging from 1.936–2.026 Å.
Regarding the coordination group bond angles for 1 and 2, the mean Npyrrole–Pt–Npyrrole bond angle (104.04 ± 0.14°) is ∼8% narrower, and the mean Nimine–Pt–Nimine bond angle (95.70 ± 0.14°) ∼13.9% wider compared with those reported by Shan et al.10 for the structurally-related enantiomers of 3. This reflects the 3-carbon chain linking the imine groups of 1 and 2, which is more flexible than the sterically strained 2-carbon chain of 3, and the resulting 6-membered chelate ring which substantially narrows and widens the Npyrrole–Pt–Npyrrole and Nimine–Pt–Nimine bond angles, respectively. Furthermore, the pyrrole groups are anionic σ-donors with more acute C–Npyrrole–C bond angles averaging 106.98 (3)° compared with the CNimine–C bonds which were more obtuse, averaging 120.87 (4)°. The C
Nimine bonds measured 1.299 (5) Å, reflecting the sp2 hybridized imine carbon atom. Selected bond lengths and bond angles are summarized in Table 1.
1 | 2 | |
---|---|---|
Bond distances (Å) | ||
Pt–Npyrrole | 2.011 (3) | 2.013 (3) |
Pt–Nimine | 2.005 (3) | 2.002 (3) |
C![]() |
1.300 (5) | 1.298 (5) |
Bond angles (°) | ||
Npyrrole–Pt–Npyrrole | 103.82 (14) | 104.25 (14) |
Nimine–Pt–Nimine | 95.65 (14) | 95.75 (14) |
cis-Npyrrole–Pt–Nimine | 80.27 (14) | 79.96 (13) |
trans-Npyrrole–Pt–Nimine | 175.50 (13) | 175.27 (14) |
C–Npyrrole–C | 107.05 (3) | 106.91 (3) |
C–Nimine–C | 120.78 (3) | 120.96 (4) |
The asymmetric unit (ASU) of complex 1 consists of three independent molecules that are linked to each other by hydrogen bonds (Fig. 3b), forming a trimeric supramolecular structure. A similar trimer was observed previously for the isoelectronic and isostructural Ni(II) derivative.30 The OH group acts as both a hydrogen bond acceptor and donor, which results in the formation of a six-membered hydrogen-bonded ring (Fig. 3b). Typically, short bond lengths indicate a stronger bond. However, in the case of hydrogen bonding the bond length does not always correlate with the bond strength.39 The trimeric supramolecular structure of 1 did exhibit significantly shorter (20%) hydrogen bond lengths, averaging 2.18 ± 0.01 Å, compared with the Ni(II) analogue (2.749 ± 0.002 Å). The ASU of 2 consisted of 4 independent molecules A–D (Fig. S13†). Further inspection of the crystal packing of both 1 and 2 indicated molecule A of one ASU forms an inverted face-to-face stacked interaction with molecule B′ from a second ASU (Fig. S12 and S13†). The symmetry-unique intermolecular Pt⋯Pt distances measure 3.723 and 3.771 Å and are greater than 3.5 Å, which is double the van der Waals radius of Pt,39 and thus fall outside the range for proper dz2(Pt)–dz2(Pt) orbital interactions.
![]() | ||
Fig. 4 Emission spectra of human serum albumin (HSA, 5.0 μM) recorded as a function of [1] at 298 K in KH2PO4 (50 mM, pH 7.50). Solid lines are least squares fits of the data to the sum of two overlapping Gaussian bands. Analogous spectra for the reaction of HSA with 2 and 3 are given in Fig. S14 (ESI†). The wavelength shift accompanying ligand uptake (Δλemmax) is plotted as the inset to the main figure (upper right). |
Since protein fluorescence quenching is dependent on the overlap between the donor fluorophore (Trp-214) and acceptor's absorbance as well as the spatial proximity of the donor and quencher, we can deduce that the current Pt(II) chelates bind to HSA within range of Trp-214 to disrupt its microenvironment and quench its fluorescence.43,44 The mechanisms that may cause fluorescence quenching include ground state complex formation, collisional quenching, energy transfer, or molecular rearrangements.45,46 Binding of 2 and 3 to HSA induced band broadening and red shifts in λemmax of +3 and +4 nm, respectively, while 1 did not induce a spectral shift, as illustrated in Fig. 4.
The λemmax shift data suggest that the microenvironment surrounding Trp-214 in HSA has become more hydrophilic.47 The small red shifts in the emission maximum for 2 and 3 are comparable to that of Pt(PrPyrr).29 Two potential explanations accounting for the λemmax red shifts are: (i) the Pt(II) complexes bind to HSA sufficiently close to Trp-214 to induce direct electronic polarization of the indole ring in the residue, and/or (ii) the disruption of the ordered water molecules located within 15–25 Å of Trp-214 can lead to orientation-dependent polarization of the fluorophore. It is possible that both mechanisms impact the lowest energy 1A1 → 1La ground state transition of Trp's indole ring,43 thereby causing the λemmax shift. In the case of 1, no λemmax shift was observed, suggesting that 1 did not induce sufficient direct electronic polarization of Trp-214's indole ring to cause a wavelength shift in the protein's emission spectrum.
I0/I = 1 + KSV[Q] = 1 + kqτ0[Q] | (1) |
Stern–Volmer (SV) plots for the fluorescence emission quenching of HSA as a function of temperature (298 K) and [Pt(II) chelate] are shown in Fig. 5a. From the slope of the least squares fit of eqn (1) to the data, the bimolecular fluorescence quenching rate constant (kq) for the HSA-Pt(II) chelate interaction can be calculated (eqn (2)):
kq = KSV/τ0 | (2) |
![]() | ||
Fig. 5 (a) Stern–Volmer plot for HSA (5 μM) in KH2PO4 buffer (50 mM, pH 7.5) recorded as the fluorescence intensity ratio (I0/I) vs. [1–3] at 298 K. Each plot represents the average of three independent experiments and the error bars represent ESD's. The data are well fitted to linear eqn (1), suggesting a single quenching mechanism is dominant. (b) Double logarithm plot of the fractional change in fluorescence intensity for human serum albumin (HSA, 5.0 μM) recorded as a function of the concentration of 1–3 (50 mM KH2PO4, pH 7.50, 298 K). Error bars are ESD's based on the average of three independent determinations. The data are described by eqn (3), which affords the affinity constant and stoichiometric coefficient for the reaction (Table 2). |
The KSV and kq values describing the interaction of the Pt(II) bis(pyrrolide-imine) chelates and HSA are summarised in Table 2. Linear Stern–Volmer plots typically indicate a single dominant quenching mechanism either static (binding-related) or dynamic (diffusion-limited collisional).40,41 The quenching mechanism may be differentiated by the KSV dependence on temperature. If KSV values decrease with increasing temperature, the dominant quenching mechanism is static because there is a decrease in the HSA·{ligand} complexes’ formation constant (the term “ligand” is used in a biochemical context; metal complexes 1–3 are the ligands or quenchers herein). However, if the reverse is observed the quenching mechanism is dynamic, which is largely due to an increase in the diffusion rate and collisional frequency.51
Compound | Temp. T (K) | 10–5KSV![]() |
10–13kq![]() |
log(Ka/M−1)c | n |
---|---|---|---|---|---|
a K SV values were determined by fitting the data to linear eqn (1) to the data. b A mean excited state lifetime, τ0, of 5.87(76) ns for HSA was used to calculate the bimolecular quenching rate constant, kq. c The estimated standard deviations of the least significant digits are given in parentheses. d Ligand:HSA binding stoichiometry from the fit of the data to eqn (3). | |||||
1 | 288 | 1.89 (0.1) | 3.22 | 5.84 (0.04) | 1.13 (0.02) |
298 | 1.59 (0.2) | 2.46 | 5.67 (0.03) | 1.10 (0.01) | |
310 | 1.16 (0.1) | 2.27 | 5.50 (0.01) | 1.06 (0.02) | |
2 | 288 | 0.68 (0.04) | 1.16 | 3.25 (0.004) | 0.68 (0.01) |
298 | 0.71 (0.04) | 1.22 | 3.45 (0.05) | 0.71 (0.01) | |
310 | 0.95 (0.04) | 1.61 | 3.64 (0.04) | 0.73 (0.004) | |
3 | 288 | 0.54 (0.02) | 0.93 | 5.17 (0.04) | 1.10 (0.01) |
298 | 0.56 (0.01) | 0.95 | 4.81 (0.02) | 1.02 (0.01) | |
310 | 0.65 (0.1) | 1.11 | 4.67 (0.02) | 0.98 (0.01) |
The Stern Volmer plots for 1–3 are linear, indicating that a single quenching mechanism is dominant—either static (binding-related) or dynamic (diffusion-limited collisional).40,41 Notably, the KSV values of 1 decreased with increasing temperature, consistent with a static quenching mechanism. In contrast, the KSV values of 2 and 3 increased with increasing temperature; therefore, a dynamic quenching mechanism is operative.40,41 The kq values for 1–3 all exceed the diffusion-controlled limit (1 × 1010 M−1 s−1)52 by 2 to 3 orders of magnitude, which is consistent with significant HSA⋯ligand binding interactions rather than non-specific collisional interactions.45 The KSV values followed the order 1 > 2 > 3 and ranged between 105–104 M−1, indicating a strong quenching phenomenon. The KSV values are consistent with those of other Pt(II) complexes interacting with HSA.29,53,54 A summary of KSV and kq values obtained for the interaction of the three Pt(II) complexes with HSA is reported in Table 2. The Stern–Volmer plots for 1, 2 and 3 over the full range of temperatures are provided in the ESI (Fig. S15†).
![]() | (3) |
The affinity constant (Ka) for 1 and 3 decreases with increasing temperature, while Ka for 2 increases with increasing temperature. The reaction stoichiometry from the slope of eqn (3) (Fig. 5b and S16†) tended to deviate from unity (0.68 < n < 1.13). From Table 2, the Ka values follow the order 1 > 3 > 2. The Ka for the reaction of Pt(PrPyrr) with HSA was previously reported,29 covering a similar range to that found for 1 here. Of the three Pt(II) bis(pyrrolide-imine) chelates, 2 bound endothermically to HSA, while the binding of 1 and 3 to HSA was exothermic (see Table 3). All three Pt(II) bis(pyrrolide-imine) chelates had moderate binding affinities for HSA, suggesting non-covalent binding. Weaker ligand binding interactions are ideal since one potential application is to use HSA as a carrier protein for metal chelates. Moderate binding affinity could signify faster diffusion rates for a drug/metallodrug within the circulatory system.46
Compound | T (K) | ΔGa (kJ mol−1) | ΔHa (kJ mol−1) | TΔSa (kJ mol−1) |
---|---|---|---|---|
a The estimated standard uncertainties of the least significant digits are given in parentheses. | ||||
1 | 288 | −32.1 (0.2) | ||
298 | −32.4 (0.2) | −26.2 (2.7) | 6.2 (2.6) | |
310 | −32.5 (0.1) | |||
2 | 288 | −19.3 (2.0) | ||
298 | −29.6 (2.0) | 30.3 (4.9) | 47.9 (4.9) | |
310 | −19.6 (1.6) | |||
3 | 288 | −27.8 (0.3) | ||
298 | −25.5 (0.4) | −37.4 (2.4) | –10.2 (1.5) | |
310 | −28.13 (0.4) |
![]() | (4) |
ΔG = ΔH − TΔS | (5) |
![]() | ||
Fig. 6 (a) Linear van't Hoff plots for the reactions of 1, 2, and 3 with HSA in 50 mM KH2PO4 buffer at pH 7.50. (b) Comparison of the thermodynamic parameters (298 K) governing the reactions of 1–3 with HSA. All measurements were done in triplicate, derived parameters were individually averaged, and error bars are standard deviations. (c) Plot of the Gibbs–Helmholtz relationship (eqn (5)) for the reaction of the three Pt(II) bis(pyrrolide-imine) chelates with HSA at 298 K in 50 mM KH2PO4 buffer at pH 7.50. The straight line fit of the data gives R2 = 0.9997 with a slope and intercept of −1. For all reactions, ΔG < 0. |
Table 3 summarises the thermodynamic parameters for the reaction of the current Pt(II) chelates with HSA in KH2PO4 buffer (50 mM, pH 7.5). All three reactions are exergonic,55 with ΔG values of −32.3 (±0.1), −19.5 (±0.08) and −27.2 (±1.4) kJ mol−1 for 1, 2 and 3, respectively at 298 K.
The enthalpy values (ΔH) for reactions of 1 and 3 with HSA are exothermic, measuring −26.2 kJ mol−1 and −37.4 kJ mol−1, respectively. Complex 2, in contrast, has an endothermic reaction with HSA (ΔH = +30.3 kJ mol−1). Fig. 6b highlights the dominance of the enthalpy term for the reactions of 1 and 3 with HSA, while the TΔS term is dominant for 2. It is well established that HSA has two main small molecule binding sites that are mainly hydrophobic,23 namely Sudlow's site I (subdomain IIA)55 and Sudlow's site II (subdomain IIIA).55 The favourable heats of reaction for 1 and 3 suggest that they cause minimal disruption of ordered water molecules within HSA. London dispersion forces (LDF) are likely the dominant binding forces for 1 and 3. In the case of complex 2, a positive enthalpy (ΔH > 0) is suggestive of hydrophobic interactions.55,56
In Fig. 6c we used the Gibbs–Helmholtz relationship (eqn (5)) and the experimental thermodynamic data (Table 3) to determine the influence that the ligand chelated to Pt(II) has on the thermodynamic parameters for complex formation with HSA. From Fig. 6c, when TΔS/ΔG is plotted against ΔH/ΔG, the resulting fit is linear with x- and y-intercepts of exactly 1. Reactions for 1 and 3 with HSA are enthalpically driven (ΔH < 0, upper and lower right quadrants) with the reaction for the hydroxy derivative 1 also being entropically favoured (ΔS > 0). Complex 2, in contrast, exhibits a purely entropically driven reaction with HSA (ΔH > 0, ΔS > 0; lower left quadrant). In all three quadrants, spontaneity is assured (ΔG < 0) since changes in ΔH/ΔG are compensated for by changes in TΔS/ΔG. Exact TΔS values for the three Pt(II) chelates are listed in Table 3. Both 1 and 2 have a positive entropy change, while 3 has a negative entropy change, with TΔS ranging from −10.2 kJ mol−1 to +49 kJ mol−1. Because the reaction for 2 is entropically driven, ordered water molecules are probably displaced from the ligand binding site into the buffer, suggesting hydrophobic and electrostatic interactions55,56 could play a key role in the binding of 2 by HSA.
The HSA binding thermodynamics of 1–3 reported here are similar to data for other Pt(II) chelates,53 which have moderately positive ΔH and ΔS values with both compounds binding specifically to Sudlow's site I. For further comparison, binding of the organic drugs ibuprofen57 and several benzodiazepines58 are characterized by moderately negative ΔH (−50 to −75 kJ mol−1) and ΔS values with Ka1 values ∼105 M−1.59 The latter drugs bind to Sudlow's site II initially, followed by secondary site occupation at higher doses (Ka2 values ∼103 M−1). The ΔH and ΔS values were, however, similar for both drug-binding steps.
One caveat concerning the assay is that HSA has several ligand-binding sites,61 which could make interpreting the data difficult. The idea here is that the displacement of either warfarin or ibuprofen by the incoming Pt(II) chelate would signal binding to either Sudlow's site I or II. However, if the presence of either warfarin or ibuprofen redirects the Pt(II) chelate to an alternate binding site within HSA (or prevents Pt(II) chelate uptake due to allosteric inhibition) the exact binding site may be difficult to elucidate. Although there may be certain limitations, examining changes in the logKa, n, and KSV values, as well as movements of the fluoroprobe, could reveal plausible binding locations for these metal complexes.
For 1–3, the assay for their uptake by HSA in the presence of key site-specific probes has been summarised in Fig. 7 (and Fig. S17 and S18†). Initially, when 1–3 react with native HSA, all ligand binding sites are available for complexation. The Stern–Volmer plots (Fig. 5 and Table 2) reveal 1 and 3 are dynamic quenchers, while 2 is a static quencher of HSA intrinsic fluorescence, while the double log plot (Fig. 5b) indicates a nominally 1:
1 Pt(II) complex
:
HSA reaction stoichiometry.
![]() | ||
Fig. 7 Binding site determination for the reaction of 2 with HSA (5.0 μM protein, 50 mM KH2PO4, pH 7.50, 298 K). (a) Emission spectra of HSA·{warfarin} (5.0 μM warfarin, λex = 320 nm) recorded as a function of the concentration of 2 at 298 K. The spectra are fitted by single Gaussian functions to locate the emission maxima. Correlation coefficients, R2, ranged from 0.990 to 0.999. (b) Emission spectra of HSA·{ibuprofen} (5.0 μM ibuprofen, λex = 228 nm) recorded as a function of the concentration of 2 at 298 K. (c) Stern–Volmer (SV) plot for native HSA, HSA·{warfarin}, and HSA·{ibuprofen} as a function of the concentration of 2 at 298 K. The plots are linear with an intercept of 1.0 when static quenching is operative (i.e., the quencher binds to the target protein and co-bound fluorophore(s)). The excitation and emission wavelengths for the fluorophore probes were: (i) Trp-214, λex = 295 nm, λem = 340 nm (native HSA); (ii) warfarin, λex = 320 nm, λem = 382 nm (HSA·{warfarin}), and (iii) ibuprofen, λex = 228 nm, λem = 332 nm (HSA·{ibuprofen}). (d) Double-log plot (eqn (3)) of the fluorescence quenching data, ![]() ![]() |
Titration of 1 into HSA·{warfarin} (λex = 320 nm, λem = 382 nm) and HSA·{ibuprofen}n (λex = 228 nm, λem = 332 nm) resulted in a decrease in the measured logKa values from 5.67 ± 0.03 (ntotal = 1.10 ± 0.01) in the native protein to 5.28 ± 0.05 (ntotal = 1.03 ± 0.03) and 5.38 ± 0.15 (ntotal = 1.06 ± 0.03) for the probe-bound targets HSA·{warfarin} and HSA·{ibuprofen}n, respectively. Similarly, for 3 the log
Ka values decreased from 4.85 ± 0.20 (ntotal = 1.02 ± 0.05) in the native protein to 3.88 ± 0.16 (ntotal = 0.85 ± 0.04) and 4.06 ± 0.15 (ntotal = 0.87 ± 0.03) for HSA·{warfarin} and HSA·{ibuprofen}n, respectively.
In contrast, the logKa values for the binding of 2 increased somewhat from 3.33 ± 0.20 (ntotal = 1.02 ± 0.05) in the native protein to 3.88 ± 0.16 (ntotal = 0.85 ± 0.04) and 4.06 ± 0.15 (ntotal = 0.87 ± 0.03) for the probe-bound targets HSA·{warfarin} and HSA·{ibuprofen}n, respectively. The Stern–Volmer plots for the binding of 1–3 by the probe-laden targets were linear throughout, consistent with a single mode of fluorescence quenching (Fig. 7d, S17 and S18†).
The foregoing results suggest that over the concentration range examined, neither warfarin nor ibuprofen are displaced to a substantial extent from HSA by the incoming Pt(II) chelate. However, the fluorescence from both site-specific probes is clearly quenched by 1–3, no doubt via a long-range FRET-based mechanism. Our interpretation is that all three Pt(II) complexes bind to HSA in the presence of warfarin or ibuprofen, without fully displacing either probe under the conditions employed for these experiments. Nevertheless, the probes can influence the binding affinity of 1–3. The observation that the probes are not fully displaced is not unexpected, considering that the logKa values for the binding of both warfarin (log
Ka ∼5.5)60 and ibuprofen (log
Ka ∼6.3)61 to native HSA are higher than those determined here for 2 and 3, while only the log
Ka value for ibuprofen uptake surpasses that of complex 1.
One interpretation of the data is that 1–3 can target both Sudlow's sites I and II. However, if either of the site-specific markers is already present in one of these sites, the unbound site is vacant and becomes fully saturated by the metal chelate. Alternatively, the presence of warfarin and ibuprofen may cause subtle changes in the tertiary structure of HSA, resulting in 1–3 favouring alternative binding sites in the protein. Such allosteric redirection of ligand uptake could explain why the logKa values for 1 and 3 decrease, while log
Ka for 2 increases for the probe-bound targets. While the current spectroscopic data cannot pinpoint the preferred binding site with absolute certainty, our in silico docking and CD studies (vide infra) suggest that while both sites are accessible, the ibuprofen binding site (Sudlow's site II) is preferred (Fig. 10 and Table S7;†vide infra). Because of the ambiguity of the fluorescence quenching assays in this case, we recommend that an ICD assay be conducted as a definitive cross-check for the binding site(s) favoured by a metal chelate (see Fig. 10).
The far-UV CD data were further analysed using JASCO Spectra Manager™ to calculate the percentage composition of α-helices, turns, and other secondary structure elements present for each HSA·{PtII(L)} adduct (Table 4). The dominant secondary structure domains are α-helices (∼56%), turns (∼11%), and unordered coils (∼32%). In solution, the secondary structure composition of HSA differs from that of native HSA in the solid state23 (68.5% α-helix, 0% β-sheet, 9.6% turns, and 21.9% unordered coils; PDB code 1BM0 analysed with BeStSel64). However, our CD data are consistent with solution state spectral decompositions reported previously by others.65–67 It is accepted that enhanced subdomain mobility and general thermal motion/disorder account for the decrease of α helicity.50,63
Compound | Conc. (μM) | Helix (%) | Turn (%) | Unordered (%) |
---|---|---|---|---|
1 | 0 | 56.39 | 11.03 | 32.58 |
5 | 55.15 | 10.43 | 32.42 | |
10 | 54.31 | 11.10 | 34.59 | |
20 | 54.04 | 10.78 | 35.18 | |
40 | 53.94 | 11.37 | 34.69 | |
2 | 0 | 57.21 | 10.88 | 31.91 |
5 | 55.72 | 10.67 | 33.61 | |
10 | 55.93 | 11.08 | 32.99 | |
20 | 55.19 | 11.05 | 33.76 | |
40 | 55.14 | 10.85 | 34.01 |
Fig. 8 shows the spectra of HSA in its native form and in the presence of identical saturating doses (40 μM) of 1 and 2. Through difference spectroscopy, we were able to resolve the ICD spectra for both Pt(II) chelates. Notably, the intensity of the ICD response for 1 was greater than that of 2, in accord with the affinity constants (logKa, Table 2) for uptake of the two Pt(II) chelates by HSA. A stronger ligand–HSA interaction results in a stronger ICD signal from the optically inactive ligand. This phenomenon was previously observed for related metal chelates by Sookai and Munro.29 The presence of the ICD band in the UV-visible region suggests that the Pt(II) chelates do not undergo decomposition, demetallation, imine hydrolysis, or ligand dissociation upon binding to HSA in solution. Unlike simple metal complexes such as cisplatin73 and NAMI-A (trans-[RuCl4(1H-indazole)2]−),74 whose ligands can dissociate from the metal, the present class of bis(pyrrolide-imine) Pt(II) chelates exhibit the potential for reversible transport in vivo via an HSA-mediated cellular uptake pathway.75 This finding highlights the possibility that other polydentate chelates and macrocyclic metal complexes could be similarly transported.
The near-UV CD spectrum of native HSA typically shows two minima at 262 and 280 nm and a maximum at 290 nm due to disulphide bonds and aromatic amino acids.76 Perturbations in this region provide insights into the conformational changes occurring in HSA. Phenylalanine and tyrosine residues have low quantum yields and extinction coefficients but high symmetry, making them less sensitive to changes in the residue's microenvironment. Tryptophan residues, in contrast, have high quantum yields and extinction coefficients but possess low symmetry. Therefore, they are most sensitive to changes in their immediate microenvironment.77 All three amino acid residues have π → π* transitions (1La and 1Lb) and have the potential to be directly involved in π bonding. Importantly, the π → π* transitions have charge transfer character and thus respond to dielectric changes within the chromophore microenvironment such that solvent changes may affect the transition.78 Thus, changes in the near-UV CD spectrum can be correlated to tertiary structure changes in the protein due to its interaction with other molecules or due to solvent exposure perturbing the aromatic residue's microenvironment.79
Near-UV CD spectra were recorded from 250–310 nm for HSA·{1} and HSA·{2} (Fig. 8c and d). The 291 nm Trp 1Lb (0,0) band and Tyr and Trp overlap bands at 282 and ∼276 nm (1La; solvated Trp),80,81 respectively, are all present. When 1 and 2 bind to HSA, these marker bands exhibit slight positive changes in ellipticity with a negligible wavelength shift. The change in ellipticity suggests that both Tyr and Trp are more exposed to the solvent and/or affected by the electrostatic field of the bound metal chelates. The former would occur for minor alterations in the secondary structure of HSA (Fig. 8a and b), i.e., the decrease in α-helicity and partial unfolding upon ligand uptake noted earlier. Partial unfolding of HSA is expected to alter both the number and arrangement of ordered water molecules around Trp and Tyr, culminating in the observed, subtle perturbations for these chromophores.
![]() | ||
Fig. 9 (a) Comparison of the experimental ICD spectrum (Δθ) recorded for HSA·{1} at pH 7.5 in 50 mM (KH2PO4) and the CD spectrum calculated using hybrid QM:MM TD-DFT simulations (CAM-B3LYP/SDD/GD3BJ:UFF) for the top-scoring docked pose of 1 bound in subdomain IIIA (PDB code: 2BXF). The DFT-calculated spectrum of the best pose matches the experimental spectrum reasonably well; a bandwidth of 1580 cm−1 (hwhm) was used for the TD-DFT data. (b) Structural model used for the TD-DFT simulations showing the location of 1 within subdomain IIIA. (c) View of the metal chelate binding site containing the best pose of 1 and the closest amino acid residues. |
Regarding the assignment of the near-UV CD and visible ICD spectra of HSA·{1} and HSA·{2}, the fine structure perturbations for the TD-DFT ICD spectra matched those observed in the experimental spectra with an amplitude of 1 > 2, which was mainly due to the larger Ka of HSA·{2} compared to HSA·{1}. The TD-DFT data for unbound 1 (Fig. 2c, Tables S5 and S9†) suggest that the UV-vis absorption bands at 350–410 nm involve 1[π → Pt(6pz),π*] and 1[Pt(5dπ),π → Pt(6pz),π*] transitions of the bis(pyrrolide-imine) chelate, while the visible region absorption bands from 410–500 nm have somewhat more metal-rich charge transfer character (MLMLCT, 1[Pt(5dπ),π → Pt(6pz),π*]). The absorption bands peaking at ∼380 nm exhibit a slight blue shift of ∼3 nm upon uptake of the Pt(II) chelates by HSA, giving rise to the positive ICD bands at ∼378 nm (Fig. 9a).
As shown in Fig. 10a, for HSA·{2} the ICD spectrum in the near-UV region was only weakly perturbed after adding a low dose of warfarin (4.2 μM); subsequent higher doses of the probe failed to displace 2 from the protein. Importantly, the logKa value of 2 for HSA (3.45) is significantly lower than that of warfarin for the protein (5.5),60 which would allow complete substitution of 2 by warfarin if the Pt(II) chelate occupied Sudlow's site I. The results clearly demonstrate that Sudlow's site I is not the main binding site of 2 within HSA. For HSA·{1} (Fig. S19†), there was a similarly negligible change in the ICD spectrum upon titration of warfarin into the solution of the protein complex, suggesting that warfarin did not displace 1. This could be due to the log
Ka of warfarin (log
Ka = 5.560) being slightly lower than that of 1 (log
Ka = 5.67) or indicate that 1 (like 2) does not bind within Sudlow's site I. At the highest concentration of warfarin (54 μM), the emergence of the warfarin-induced CD peak at 306 nm was observed.
When ibuprofen was titrated into HSA·{2}, there was a monotonic decrease in the ICD signal from 2 within the protein (Fig. 10b), consistent with displacement of 2 by ibuprofen into the bulk solution and confirming that 2 bound in either one or both ibuprofen binding sites (subunits IIIA or between IIA and IIIA).27 The evidence collectively demonstrates that both Pt(II) chelates preferentially bind in Sudlow's site II (subdomain IIIA). Notably, in silico data (TD-DFT simulations) endorsed the binding of 2 within Sudlow's site II of HSA by matching the key features of the experimental CD spectrum (Fig. 9a). Despite compelling spectroscopic evidence, however, only X-ray data can delineate the precise binding site(s) of 1 and 2 with absolute certainty.
From Table S7,† HSA has several binding sites for 1 and 2, with docking scores (ΔGdock) ranging from −1.4 to − 5.2 kcal mol−1. Both Pt(II) chelates repeatedly favoured similar binding locations (∼3 in total including Sudlow's I and II) and a third site between subdomain IA–IIA, consistent with X-ray data for diverse drug and ligand types bound by HSA61 and recent data reported for d8 metal ions chelated by a related bis(pyrrolide-imine) ligand system.29
The binding sites targeted by the Pt(II) chelates were all located between 5–30 Å of Trp-214. Distances longer than 20 Å are less likely to cause quenching of Trp-214 fluorescence. In such instances, these may be secondary, less-favourable Pt(II) chelate binding sites. Regarding Tyr and Phe (Table S8†), both amino acids were within 6–19 Å from the docked Pt(II) chelate which afforded observable perturbations in the near-UV CD spectrum between 250 and 295 nm (Fig. 8c and d).
The presence of warfarin and diazepam bound within HSA subdomains IIA and IIIA, respectively, minimally influenced the docking scores for the uptake of 2, while there was a decrease in the docking score for 1 (Table S7†). Thus, ΔGdock for 1 is reduced from −4.62 to −1.39 kcal mol−1 when warfarin is present (Table S7†) and from −5.15 to −2.02 kcal mol−1 when diazepam is present (Table S7†). In both instances, 1 is redirected to less favourable binding sites. Ibuprofen bound within Sudlow's site II (subdomain IIIA, as well as at its secondary site, IIA-IIB) did not affect the docking scores of either 1 or 2. This in silico result accords with the site displacement data for HSA·{ibuprofen} (Fig. 7), since neither 1 nor 2 have sufficient affinity (high enough logKa values) to displace ibuprofen. Consequently, it seems likely that the Pt(II) chelates were redirected to alternate binding site(s) in the fluorescence quenching experiments depicted in Fig. 7.
Pt(II) chelates 1 and 3 bind to HSA with negative ΔH, ΔG, and positive ΔS values, reflecting a spontaneous enthalpy-driven process governed by van der Waals (London dispersion) forces, while 2 had a positive ΔH, negative ΔG and a strongly positive ΔS, consistent with an entropically driven reaction with HSA. Glide XP docking simulations in conjunction with fluorescence and induced CD site specificity assays employing probe ligands showed that 1 and 2 bind preferentially in Sudlow's site II, with possible additional binding sites being accessible. Far- and near-UV CD spectroscopy confirmed that the binding of 1 and 2 minimally perturbs the protein's secondary structure. Finally, the induced CD spectra recorded for HSA·{1} and HSA·{2} distinctly confirm uptake of the intact complexes without imine group hydrolysis or demetallation.
Footnote |
† Electronic supplementary information (ESI) available: Complete experimental details and ESI tables and figures in PDF format, and X-ray crystal structures. CCDC 2271558 and 2271559. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3dt02039h |
This journal is © The Royal Society of Chemistry 2023 |