Expanding on the plecstatin anticancer agent class: exchange of the chlorido ligand for N-heterocyclic ligands

Introduction

Piano-stool metal complexes with the general formula [M(π-bound ligand)(X)(Y)(Z)]⁺ have been investigated as anticancer agents due to the scaffold being highly modifiable to enhance and tailor the biological and chemical properties.^1–5 This is exemplified by the different biological activity of the Ru complexes [Ru(arene)(en)Cl]⁺ (RAED; where en is 1,2-ethylenediamine) and [Ru(arene)(PTA)Cl₂] (RAPTA; where PTA is 1,3,5-triaza-7-phophaadamantane).^1,5–7 The RAED complexes exhibit antiproliferative activity and were found to bind to DNA,^1,6,8 whereas the RAPTA compounds are inactive in in vitro anticancer activity assays but show antimetastatic properties and bind preferentially to proteins over DNA.^5,7,9–14 Consequently, a large number of compounds based on these scaffolds have been reported, often with bioactive ligands coordinated to the metal centres.^15–19

Pyridinecarbothioamides (PCAs) are a class of bioactive molecules,^20,21 and PCA-based organometallics of the general formula [M(L)(PCA)X]ⁿ⁺ (where M is Ru, Os, Rh, Ir; L is cym [η⁶-p-cymene], biphenyl, Cp* [Cp* is η⁵-pentamethylcyclopentadienyl]; X is Cl, Br, I, PTA, PPh₃) exhibited potent antiproliferative activity, showed activity in in vivo assays and displayed unique chemical reactivity.^18,21–28 PCA complexes demonstrated low reactivity to amino acids such as L-cysteine and L-histidine and high stability in acidic solutions. In particular, the high stability under acidic conditions may make them suitable as orally administered anticancer agents.²² We investigated a wide range of related derivatives and functionalised the phenyl group of PCAs, exchanged the chlorido ligand with bromido and iodido ligands, substituted cym for other arenes and replaced the Ru(cym) moiety with Rh(Cp*) and Ir(Cp*) groups.¹⁹ The Ru complexes were highly potent in various cancer cell lines with the p-fluoro-substituted derivative (Scheme 1) being the most active.²⁹ The numerous binding interactions that metal complexes can have in biological systems^30,31 makes it challenging to determine their key biological target.³² Through pull-down studies, the biological target of [Ru(cym)(p-F-PCA)Cl]Cl (Scheme 1) was identified as plectin, a key protein involved in the cytoskeletal framework of cells, and the compound was termed plecstatin-1.^29,32 In an attempt to improve the delivery of such compounds, we also investigated plecstatin-1 as a payload in peptide–drug conjugates through arene functionalisation,³³ while substitution of the chlorido ligand could provide another means of derivatisation in the future.^34,35

Scheme 1.

Synthetic routes A and B to coordinate N-heterocycles to plecstatin-1 to give complexes 4‒6. Counterion X is OTf⁻ or PF₆⁻ depending on if pathway A or B was used respectively.

Monodentately coordinated labile ligands in metal complexes can be replaced with aqua ligands in biological environments, which facilitates covalent interactions with their targets. Although in some cases, altering this ligand can reduce the activity of compounds,^2,36–38 in other cases the substitution for more lipophilic ligands improves the biological activity.³⁷ In the case of the PCA complexes, the chlorido ligand of the p-Cl-PCA analogue of plecstatin-1 was exchanged with the phosphines PTA and triphenylphosphine to examine the influence of hydrophilic and lipophilic ligands on the biological performance of PCA complexes. Interestingly, the triphenylphosphine derivatives were more active than the PTA derivatives and plecstatin-1 in the cancer cell lines tested.²⁸ Consequently, we explore here the impact of substituting the monodentate chlorido ligand of plecstatin-1 with N-methylimidazole, N-methylbenzimidazole and pyridine on its chemical and biological properties.

Results and discussion

In order to substitute the chlorido ligand of plecstatin-1 [Ru(cym)(p-F-PCA)Cl]Cl with N-heterocycles, two pathways were explored (Scheme 1). In route A, we aimed to exploit the strong interaction between the bidentate PCA ligand with metal centres to introduce p-F-PCA by substituting the two chlorido ligands in [Ru(cym)(N-heterocycle)Cl₂] precursors 1–3. For this purpose, the Ru precursor complexes 1–3 were treated with AgOTf in acetonitrile to replace the chlorido ligands with solvent molecules. It was anticipated that this would allow facile coordination of p-F-PCA to the Ru centre to give the target complexes 4–6. Conversely, route B was based on the chlorido ligand of plecstatin-1 being replaced in a methanol and water solution with the N-heterocycle, followed by the addition of NH₄PF₆ to induce precipitation of 4–6 from the solution. Unlike in the former pathway, approach B avoids the use of Ag salts such as Ag₂O or AgX (X is PF₆⁻, OTf⁻) that may compete with the Ru centres in the precursors for ligands, further complicating the reaction and requiring additional purification steps.

Both synthetic methods gave the products in similar yields of 60–80%, but the synthesis by pathway B proved indeed to be more facile. The approach requires a single metal complex as a precursor rather than the preparation of 1–3. Furthermore, the products precipitated from the aqueous solution, simplifying purification to just filtration and washing with diethyl ether.

Inspection of the ¹H NMR spectra of complexes 4–6 in polar solvents such as d₆-acetone or d₄-methanol (for the NMR spectra see Supplementary Fig. S1–S9) revealed two sets of signals that suggested the formation of two species. However, only one species was observed when the ¹H NMR spectrum was collected in CDCl₃. Metal piano-stool complexes bearing the PCA ligand were reported with the thioamide group in protonated and deprotonated forms with the ligand being either charge neutral to anionic respectively.^{22,24,25,27,28} In the case of complexes 4–6 without an anionic co-ligand, this results in the overall charge of the complex being either +1 or +2. The ¹H NMR spectra for complexes 4–6 in CDCl₃ (Fig. 1) showed downfield shifts of protons adjacent to the coordinating nitrogen atom, indicating their coordination to the Ru centre due to π-back donation effects. In addition, the ¹H–¹H nuclear Overhauser effect spectroscopy (NOESY) spectra showed nuclear Overhauser effect (NOE) correlations between the cym and PCA ligand protons and the N-heterocycle, providing further evidence of their coordination (Supplementary Fig. S3, S6 and S9). The aromatic cym protons of 5 and 6 became more chemically distinct as the size of the N-heterocyclic ligand increased, as evident by the larger separation of the four doublets between 5.5 and 6.2 ppm (Fig. 1). This contrasts with 4 for which two of the aromatic proton doublets overlapped to form a multiplet. The pseudo-tetrahedral geometry of the complex results in the protons becoming increasingly more chemically distinct with larger N-heterocyclic ligands. This also extended to the methyl groups on the isopropyl substituent of cym, with the spectra of 5 and 6 displaying them as two doublets whereas in that of 4 they were observed as a triplet due to the overlap of two doublets.

Fig. 1.

¹H NMR spectra of 4‒6 in CDCl₃ showing the downfield shifts of signals assigned to the coordinated N-heterocycle relative to spectra collected for the heterocyclic compounds. The arrows indicate the extent of the downfield shift of the highlighted protons relative to the spectra of free N-heterocycles.

The complexes showed all the expected signals in the ¹³C{¹H} NMR spectra, with the PCA and p-cymene carbon atoms detected with similar chemical shifts. The carbon atoms adjacent to the coordinating N-heterocycle shifted downfield relative to the uncoordinated compound. Exchanging the chlorido ligand with the larger N-heterocyclic ligand, the chemical environment of the carbon atoms on the N-fluorophenyl ring became inequivalent, resulting in C-8 and C-9 being observed in the ¹³C{¹H} NMR spectrum as two signals that were 0.2 ppm apart.

To confirm the coordination of the N-heterocyclic ligands to plecstatin-1, complexes 4–6 were analysed by mass spectrometry. In all cases, the electrospray ionisation (ESI)-mass spectra feature the [M – N-heterocycle]⁺ ion as the N-heterocycle ligands were cleaved during the ionisation process (Supplementary Fig. S10–S12). Analysis of the complexes by elemental analysis provided further evidence of the coordination of the N-heterocyclic ligands to plecstatin-1 with the PCA ligand being deprotonated. The elemental analysis data match compositions with complex cations that are singly positively charged as a result of N-heterocyclic ligand coordination, and the overall charge is balanced by a PF₆⁻ counterion.

A single crystal of 4 was grown from an NMR sample in d₆-acetone, and it crystallised as a pair of enantiomers together with disordered acetone. X-Ray diffraction analysis confirmed the expected piano-stool configuration with the PCA ligand coordinating bidentately through its pyridinyl-N and S atoms (Fig. 2). The thioamide moiety was found to be deprotonated, as evident by shorter C6–N2 and longer S1–C6 bonds compared to plecstatin-1 (Table 1),²² and a single PF₆⁻ counterion being present to balance the charge of the singly charged complex cation. In addition to the cym ligand, the coordination sphere was completed by methylimidazole bonded to the Ru centre by the imine-N atom. The length of the Ru–N_imidazole bond (Ru–N3) at 2.1071(16) Å) was in the similar range as other Ru–N_imidazole bonds^39,40 but slightly longer than the Ru–N_pyridine bond in 4 (Ru–N1; 2.094(16) Å) and much shorter than the Ru–Cl distance found in plecstatin-1 (2.392(9) Å).²² The bond angles around the coordination sphere of the Ru centre increased to accommodate the larger N-methylimidazole instead of the chlorido ligand in plecstatin-1. The torsion angle N1–C5–C6–S1 was 10.5(2)° for 4, suggesting that the pyridine ring and thioamide form a conjugated system and to a greater extent than plecstatin-1 (15.9(4)°), in which however the PCA ligand was found protonated.²²

Fig. 2.

Molecular structure of one of the enantiomers of 4 drawn at 50% probability level. Co-crystallised solvent molecules and counterions were omitted for clarity.

Plecstatin-1 exhibits unique behaviour in aqueous solution. It rapidly forms dimers with PCA-S atoms bridging Ru centres (Fig. 3).²² Reversed phase–high-performance liquid chromatography (RP-HPLC) was used to investigate the behaviour of 4 and 6 in comparison to plecstatin-1 in aqueous solution (H₂O:acetonitrile = 1:1). The retention times (R_t) of 4 and 6 were slightly different from that found for plecstatin-1 due to the N-heterocycle coordinated to the metal centre (Fig. 3, Supplementary Fig. S13–S17). However, both 4 and 6 underwent rapid dimerisation in aqueous conditions as evidenced by the formation of peaks with larger retention times. Co-injection of plecstatin-1, 4 and 6 revealed that they were forming the same dimeric complex, with the μ-S bridging atom inducing the dissociation of the N-heterocyclic or chlorido ligands from the Ru centres (Fig. 3). The differences in peak ratios for the two compounds may be explained by the ligand donor strengths for 1-methylimidazole as compared to pyridine.⁴¹ The former binds more strongly to metal centres than pyridine and accordingly the conversion of the complex into the dimer occurs to a lesser extent.

Fig. 3.

(a) Dimerisation of plecstatin in aqueous solution and conversion of 4 and 6 into the same dimer. (b) From top to bottom: chromatograms of solutions of plecstatin-1, 4, 6 and a mixture of plecstatin-1, 4 and 6 in MeCN/H₂O (1:1).

In another experiment complex 6 was incubated in water as well as in an aqueous solution acidified with HCl (60 mM) to give a pH value of 1. Aliquots taken after 0, 4 and 24 h were analysed by RP-HPLC (Supplementary Fig. S16 and S17). The complex showed rapid formation of a dimeric species in both media with a prominent peak with an intensity similar to 6 in the chromatogram at t = 0 h. In H₂O, it appears as if an equilibrium is present with both complex 6 and the formed dimeric structure at a ratio of ~1:1. In the acidified solution, the initial dimerisation also occurred and after 4 h both complexes were detected at a similar ratio in the chromatogram. However, after 24 h the equilibrium shifted almost completely back to a monomeric species. Owing to the similarity in R_t it is difficult to say whether 6 is formed or the chlorido complex, i.e. plecstatin-1, due to the high chloride concentration. However, this transformation has relevance in a biological setting given the presence of relatively high chloride concentrations in blood serum and therewith affects the mode of action of the compound class.

The antiproliferative activity of 4–6 was determined against a small panel of human cancer cell lines and half-maximal inhibitory concentration (IC₅₀) values in the low micromolar range were observed (Table 2). Complexes 4 and 6 gave IC₅₀ values similar to plecstatin-1 in the cancer cell lines,²¹ whereas 5 was the slightly less active. That all three complexes still displayed good anticancer activity exemplifies the low effect the exchange of the leaving chlorido ligand with an N-heterocycle had on these ruthenium piano-stool complexes. Considering the studies in aqueous solution, the active pharmacophores for the compounds may well be the same and this explains IC₅₀ values in a similar range. Alternatively, it is possible that the mode of action is based on non-covalent interactions between the compounds and their biomolecular target.^19,22

Conclusions

The plecstatin class of metal complexes has shown promise as chemotherapeutic agents with their potent antiproliferative activity, specific interaction with their cellular target plectin and overall unique biological and physicochemical properties. Here, we expanded the diversity of plecstatin complexes by exchanging the chlorido ligand with the N-heterocycles 1-methylimidazole, 1-methylbenzimidazole and pyridine. Two avenues were explored to synthesise complexes 4–6, one which involved the preparation of the [Ru(cym)Cl₂] complexes of the aforementioned N-heterocycles followed by reaction with p-F-PCA and the other in which plecstatin-1 was used as the precursor and the chlorido ligand was substituted for the N-heterocycles. Both approaches proved successful with the latter route deemed more efficient due to the fewer synthetic steps and ease of purification by precipitation. The coordination of the ligand to the metal Ru centres was confirmed through ¹H–¹H NOESY correlations in the 2-D NMR spectrum, single crystal X-ray diffraction and elemental analysis.

Replacing the chlorido ligand in plecstatin-1 for a more strongly coordinated N-heterocycle could have detrimental effects on the biological properties of such complexes as the chlorido–aqua ligand exchange is a key step for biological activity and considered an activation step. However, the synthesised derivatives still exhibited potent antiproliferative activity in the small panel of cancer cell lines, with IC₅₀ values similar to or only slightly higher than those of plecstatin-1. This may be explained by the fact that the complexes with N-heterocycle co-ligands displayed the same behaviour in aqueous and acidic solutions as observed for their chlorido precursor. It appears as if the complexes undergo dimerisation in aqueous solution, whereas acidification with HCl seems to push the equilibrium towards a monomeric derivative. Therefore, exchanging the chlorido ligand with other ligands can provide opportunities for further functionalisation in the future, such as conjugation to delivery vectors with a functionalised N-heterocyclic co-ligand as a conjugation handle that could be cleaved under physiologically relevant conditions.

Supplementary material

Supplementary material is available online.