The two isomers of a cyclometallated palladium sensitizer show different photodynamic properties in cancer cells

Abstract

This report demonstrates that changing the position of the carbon–metal bond in a polypyridyl cyclopalladated complex, i.e. going from PdL¹ (N^N^C^N) to PdL² (N^N^N^C), dramatically influences the photodynamic properties of the complex in cancer cells. This effect is primarily attributed to the significantly difference in absorbance and singlet oxygen quantum yields between the two isomers.

---

The success of cisplatin, a milestone drug in the treatment of cancers, stimulated the generation of many platinum-based anticancer drugs,^1–3 three of which (carboplatin, oxaliplatin and nedaplatin) are approved worldwide. However, the unselective covalent binding of cisplatin with DNA in cancer cells and healthy cells results in serious side effects and drug resistances, which has encouraged the development of anticancer drugs based on alternative metals.^4–9 In this regard, palladium(II) complexes have been proposed as potential metal-based anticancer drugs for their similar d⁸ coordination sphere and square-planar structure, compared to platinum(II) complexes.^10,11 One of them, called padeliporfin or WST11, was recently approved for photodynamic therapy (PDT) of prostate cancer.¹² PDT is a form of light-activated anticancer treatment. It emerges as a more patient-friendly approach due to the controlled toxicity effect and low invasiveness of light irradiation.^13–17 In PDT, a photosensitizing agent (PS) is irradiated by visible light at the tumor site, where it generates cytotoxic reactive oxygen species (ROS), which induces cancer cell death.¹⁸ Polypyridyl metal complexes typically form excellent PDT sensitizers, provided they strongly absorb visible light.^19,20 The light absorption properties of such complexes can be tuned by changing the metal or the ligands. Critically, good photosensitizers should be photostable, which can be achieved using multidentate ligands. The single coordination bonds in multidentate complexes are no stronger than those of monodentate ligands, they are simply less likely to all decoordinate at once.

Recently, bioactive pincer palladium complexes with tridentate N-heterocyclic carbene ligands have been shown to possess stable metal–carbon bonds and tuneable physicochemical properties.^10,21–24 However, intracellular substitution of the remaining monodentate ligands makes their speciation in biological media and mode-of-action complicated to understand. In addition, due to the smaller ionic radius of Pd²⁺ ions, Pd–ligand bonds are longer and more labile than their Pt–ligand analogues,²⁵ so that anticancer drugs based on palladium(II) are still comparatively rare.⁶ To overcome these drawbacks, we investigated the design and properties of palladium(II) PDT sensitizers built from single tetradentate cyclometallating ligands, which are expected to be more stable in biological media compared with the tridentate N-heterocyclic carbene ligands. Cyclometallation was considered for different reasons. Firstly, the strong Pd–C bond can stabilize these compounds in biological media. Secondly, the lower charge introduced by the cyclometallated ligand can improve the lipophilicity and cellular uptake of the metal complexes.^7,26 Finally, the presence of a Pd–C bond should in principle lead to a bathochromic shift of the visible absorption bands of the metal complex, which is key for PDT applications.²⁷

In polypyridyl metal complexes, introducing a metal–carbon bond usually generates a series of isomers that might have different properties. Herein we investigated two novel cyclopalladated isomers PdL¹ (H₂L¹ = N-(3-(pyridin-2-yl)phenyl)-[2,2′-bipyridin]-6-amine) and PdL² (H₂L² = N-(6-phenylpyridin-2-yl)-[2,2′-bipyridin]-6-amine) (Scheme 1a). In PdL¹, the Pd–C bond was introduced in a pyridyl group that is adjacent to the non-bonded nitrogen bridge of the ligand, while in PdL² it is introduced in one of the terminal aromatic rings. The ligands H₂L¹ and H₂L² were synthesized by Buchwald–Hartwig coupling reactions (Scheme S1 and Fig. S1–S4, ESI†).^28–30 Palladation was achieved in more than 90% yield by reacting the corresponding ligand with palladium(II) acetate in acetic acid (Scheme S1 and Fig. S5–S8, ESI†). Neither ¹H NMR (Fig. S5 and S7, ESI†) nor infrared spectroscopy (Fig. S9, ESI†) showed any sign of a protonated secondary amine bridge, which altogether suggested that these complexes were much more acidic than expected. According to ¹³C-APT NMR (Fig. S2, S4, S6 and S8, ESI†), the ligands H₂L¹ and H₂L² have six quaternary carbon peaks, while their palladium complexes have seven, thus demonstrating that cyclometallation did occur. Altogether PdL¹ and PdL² appear to be neutral complexes; their identical HRMS data also demonstrated they are coordination isomers.

Scheme 1 (a) Synthesis of PdL¹ and PdL²; (b) displacement ellipsoid plot (50% probability level) of PdL² at 110(2) K (bond distance: Pd–N1 2.060(3) Å, Pd–N2 2.028(4) Å, Pd–N4 1.988(3) Å, Pd–C21 2.017(4) Å); angle: N4–Pd1–C21 81.99(15)°, N4–Pd1–N2 92.66(16)°, C21–Pd1–N2 174.65(18)°, N4–Pd1–N1 172.2(2)°, C21–Pd1–N1 105.02(17)°, N2–Pd1–N1 80.33(13)°.

Vapor diffusion of diethyl ether into a methanol solution of PdL² yielded red rectangular crystals suitable for X-ray structure determination (Table S1, ESI,† and Scheme 1b). PdL² crystallized in the centrosymmetric P2₁/n monoclinic space group. Three nitrogen and one carbon atom were coordinated to the palladium(II) cation, with bond lengths in the range 1.988(3)–2.028(4) Å for the three Pd–N bonds, and a Pd–C bond distance of 2.017(4) Å. The coordination sphere was slightly distorted, with a dihedral angle N1–N2–N4–C21 of 2.33°. τ₄, a structural parameter calculated by (360° − (α + β))/(141°), where α and β are the two greatest valence angles of the coordination sphere,³¹ was 0.093 in the structure of PdL², which is typical of an essentially square planar complex. Deprotonation of the nitrogen bridge was evident from the shorter distance between the amine nitrogen atoms and the adjacent pyridine carbon atoms (C10–N3 = 1.353(4) Å and C11–N3 = 1.349(4) Å), compared to that found in metal complexes with protonated nitrogen bridges (N–C distances in the range 1.36 Å to 1.39 Å).^29,30 Also, unlike for [Fe(Hbbpya)(NCS)₂],³² no residual electron density was found near the bridging N atom in the structure of PdL². Finally, the asymmetric unit contained no counter-ions. In summary, single crystal X-ray crystallography is consistent with NMR and IR data, showing that PdL¹ and PdL² are neutral species because deprotonation of the nitrogen bridge becomes very easy upon coordination.

The absorption spectrum of both complexes in PBS : DMSO (1 : 1) solution at 310 K (Fig. S10, ESI†) presented no significant changes over 24 hours, suggesting that the complexes were thermally stable under such conditions. Similar results were obtained in cell-growing medium (Fig. S11, ESI†), demonstrating good stability under such conditions. The partition coefficients (log P_ow) of the palladium complexes were determined by the shake-flask method (Table S2, ESI†). log P_ow was lower for PdL¹ (−0.64) than for PdL² (+0.046), confirming the higher solubility in water of the former, compared to the latter. Their cytotoxicity was tested in lung (A549) and skin (A431) cancer cell lines, both in the dark and upon blue light activation. Low doses of blue light were chosen (455 nm, 5 min, 10.5 mW cm⁻², 3.2 J cm⁻²) which have by themselves no effect on cell growth.³³ The cell growth inhibition effective concentrations (EC₅₀) of PdL¹ and PdL² are reported in Table 1, and the dose–response curves are shown in Fig. 1 and Fig. S12 (ESI†). In the dark both compounds showed significant anticancer activity, with an EC₅₀ around 10 μM for PdL¹ and PdL² in A549 cells, respectively. After blue light activation, PdL¹ showed a notable 13- or 4.0-fold increase in cytotoxicity in A549 and A431, respectively, while PdL² showed a negligible photoindex of 1.3 or 1.4, respectively. The difference in photocytotoxicity between the two coordination isomers was quite intriguing. To investigate the reason for such a difference, we first measured the singlet oxygen (¹O₂) generation quantum yield (φ_Δ) of these two isomers in CD₃OD spectroscopically. φ_Δ was more than twice higher for PdL¹ (0.89) than for PdL² (0.38, Fig. 2b and Table S3, ESI†), and higher than the reference [Ru(bpy)₃]Cl₂ (0.73).³⁴ However, PdL² was still a decent ¹O₂ generator. In methanol, the absorbance spectra of both complexes (Fig. 2) were similar in the 270–300 nm region; however, PdL¹ had a much higher absorption in the blue region with λ^abs_max = 422 nm, compared to PdL² that absorbed in the near-UV region (λ^abs_max = 347 nm). In this solvent the molar absorptivity at 455 nm for PdL¹ and PdL² was 2004 M⁻¹ cm⁻¹ and 133 M⁻¹ cm⁻¹, respectively, indicating a 15-fold enhanced absorption of PdL¹ in the blue region, compared with PdL². Considering their similar lifetime (0.271 vs. 0.333 ns for the main component of their biexponential decay, Table S3 and Fig. S13, ESI†), the difference in ¹O₂ generation efficiency is probably a consequence of the higher phosphorescence quantum yield for PdL¹ (0.0017) vs.PdL² (0.00084, Table S3, ESI†), which points to the slower non-radiative decay pathways for the former, compared to the latter. Altogether, the dramatically higher phototoxicity of PdL¹, compared to PdL², seems to result from the much better absorption of blue light of PdL¹, coupled to its higher phosphorescence quantum yield, which leads to higher ¹O₂ generation efficiency. Although different log P_ow values may lead to different cell uptake and sub-cellular localization for both isomers as well, the better photobiological properties of PdL¹ depend, at least in part, on the much better photodynamic properties of PdL¹ (ε₄₅₅ and φ_Δ), compared to its isomer PdL².

Table 1 The cell growing inhibition effective concentrations (EC₅₀ in μM) of PdL¹ and PdL² towards A549 and A431 human cancer cell lines. 95% confidence interval (CI in μM) and photoindex (PI = EC₅₀, dark/EC₅₀, light) are also indicated

Complexes		EC50 (μM)
		A549	±CI	A431	±CI
Irradiation condition: 455 nm blue light, 5 min, 10.5 mW cm^−2, 3.2 J cm^−2. Data is the mean over three independent experiments.
PdL^1	Dark	12	+3.0	20	+4.0
			−3.0		−3.0
	Light	0.9	+0.8	5.0	+2.0
			−0.5		−1.0
PI		13		4.0
PdL^2	Dark	8.0	+2.0	14	+2.0
			−1.0		−1.0
	Light	6.0	+0.8	10	+1.0
			−0.7		−1.0
PI		1.3		1.4

---

Fig. 1 Dose–response curves for A549 cells incubated with palladium complexes and irradiated 5 min with blue light (blue data points), or in the dark (black data points).

Fig. 2 (a) The molar absorption coefficient (solid line) and emission spectra (dashed line) of PdL¹ (black), PdL² (red) in CH₃OH. (b) Singlet oxygen emission spectra of [Ru(bpy)₃]Cl₂ (blue), PdL¹ (black), PdL² (red) in CD₃OD irradiated with blue light (λ_ex = 450 nm, 50 mW, 0.4 W cm⁻²).

Density functional theory (DFT) calculations were performed to understand why PdL¹ exhibited higher absorption in the blue domain than its isomer PdL². The nature of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbitals (LUMO) is highly relevant to predict the photophysical properties of metal complexes.^34–36 As showed in Fig. 3, the HOMO and LUMO orbitals of both isomers PdL¹ and PdL² had π symmetry and were centered on the ligand, with a negligible contribution of the palladium(II) center. The bridged secondary amine is the major contributor to the HOMO of both palladium complexes, attributing for 20.4 (PdL¹) and 21.8% (PdL²) of the electron density. The rest of the HOMO orbital density was located in the aromatic rings directly connected to the nitrogen bridge. By contrast, the LUMO orbitals for both complexes were centered on the bipyridyl fragment. This suggested that the lowest energy absorption band of both palladium complexes should be of ligand-to-ligand charge transfer character, from the amine to the bipyridyl.

Fig. 3 DFT calculation of HOMOs (bottom) and LUMOs (top) orbitals of PdL¹ and PdL²; occupied orbitals (HOMO) have red and blue lobes, and unoccupied orbitals (LUMO) brown and cyan lobes. Element colour code: grey = C; orange = Pd; blue = N; white = H.

The calculated energies of HOMOs, LUMOs and energy gaps are listed in Table S4 (ESI†). The HOMO of PdL¹ was significantly higher in energy than that of PdL², indicating the higher electron-donating effect of the negatively charged carbon atom of PdL¹, compared with that of PdL² which is further away from the nitrogen bridge. By contrast, the LUMO energy levels of both palladium complexes were similar, because LUMO orbitals are located on the almost equivalent bipyridyl fragments. Such lower energy gap of PdL¹ suggested better absorption of low-energy light, which explains the observed differences in the UV-vis spectra of the two isomers. These results were confirmed by time-dependent density functional theory calculations (TDDFT) for both complexes in methanol, using COSMO to simulate solvent effects (Fig. S14, ESI,† left). The calculated spectrum of PdL¹ (Fig. S14, ESI,† left) showed a lower energy (515 nm) for the HOMO–LUMO transition, compared to PdL² (449 nm). These transition energies were increased (404 and 367 nm, respectively) by protonation of the nitrogen bridge (Fig. S14, ESI,† right), which may happen in the slightly acidic environment of cancer cells; however, the trend between [Pd(HL¹)]⁺ and [Pd(HL²)]⁺ was identical to that seen for PdL¹ and PdL². Overall, calculations clearly demonstrated that a change of the position of the carbon–metal bond had a strong influence on the HOMO–LUMO energy gaps of these cyclometallated palladium complexes.

In summary, the new cyclopalladated complex PdL¹ showed good absorbance in the blue region of the spectrum, low phosphorescence, and excellent singlet oxygen quantum yield (0.89), which altogether translated into high photoindex in human cancer cells. By contrast, its isomer PdL² had low absorption and low singlet oxygen quantum yield (0.38), resulting in negligible activation by blue light in vitro. DFT calculation showed that the higher absorption in the blue region of PdL¹, and thus its lower HOMO–LOMO energy gap, was due to the closer proximity between the electron-rich cyclometallated aromatic cycle and the nitrogen bridge of the ligand, while in PdL² both aromatic rings adjacent to the N bridge are electron-poor pyridine rings, which lowers the HOMO energy level. To the best of our knowledge, this study is the first report that two isomers of organometallic prodrugs have different photobiological properties. These results demonstrate that changing the position of the carbon–metal bond in the coordination sphere of photoactive organometallic prodrugs can be used to tune the energy gap between their frontier orbitals, and hence their absorption in the visible region of the spectrum.

X. Zhou gratefully acknowledges the China Scholarship Council (CSC) for a personal grant (No. 201606200045). This work is supported by an ERC Starting Grant to S. Bonnet.

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. T. C. Johnstone, K. Suntharalingam and S. J. Lippard, Chem. Rev., 2016, 116, 3436–3486.
2. L. Kelland, Nat. Rev. Cancer, 2007, 7, 573–584.
3. D. Wang and S. J. Lippard, Nat. Rev. Drug Discovery, 2005, 4, 307–320.
4. L. Ma, N. Wang, R. Ma, C. Li, Z. Xu, M. K. Tse and G. Zhu, Angew. Chem., Int. Ed., 2018, 57, 1–6.
5. S. Medici, M. Peana, V. M. Nurchi, J. I. Lachowicz, G. Crisponi and M. A. Zoroddu, Coord. Chem. Rev., 2015, 284, 329–350.
6. N. Cutillas, G. S. Yellol, C. de Haro, C. Vicente, V. Rodriguez and J. Ruiz, Coord. Chem. Rev., 2013, 257, 2784–2797.
7. L. Zeng, P. Gupta, Y. Chen, E. Wang, L. Ji, H. Chao and Z. S. Chen, Chem. Soc. Rev., 2017, 46, 5771–5804.
8. F. E. Poynton, S. A. Bright, S. Blasco, D. C. Williams, J. M. Kelly and T. Gunnlaugsson, Chem. Soc. Rev., 2017, 46, 7706–7756.
9. H. Huang, P. Zhang, H. Chen, L. Ji and H. Chao, Chem. – Eur. J., 2015, 21, 715–725.
10. T. T. Fong, C. N. Lok, C. Y. Chung, Y. M. Fung, P. K. Chow, P. K. Wan and C. M. Che, Angew. Chem., Int. Ed., 2016, 55, 11935–11939.
11. M. Fanelli, M. Formica, V. Fusi, L. Giorgi, M. Micheloni and P. Paoli, Coord. Chem. Rev., 2016, 310, 41–79.
12. A.-R. Azzouzi, S. Vincendeau, E. Barret, A. Cicco, F. Kleinclauss, H. G. van der Poel, C. G. Stief, J. Rassweiler, G. Salomon, E. Solsona, A. Alcaraz, T. T. Tammela, D. J. Rosario, F. Gomez-Veiga, G. Ahlgren, F. Benzaghou, B. Gaillac, B. Amzal, F. M. J. Debruyne, G. Fromont, C. Gratzke and M. Emberton, Lancet Oncol., 2017, 18, 181–191.
13. H. Cao, L. Wang, Y. Yang, J. Li, Y. Qi, Y. Li, Y. Li, H. Wang and J. Li, Angew. Chem., Int. Ed., 2018, 57, 7759–7763.
14. Y. Ma, X. Li, A. Li, P. Yang, C. Zhang and B. Tang, Angew. Chem., Int. Ed., 2017, 56, 13752–13756.
15. S. H. Askes, A. Bahreman and S. Bonnet, Angew. Chem., Int. Ed., 2014, 53, 1029–1033.
16. S. L. Higgins and K. J. Brewer, Angew. Chem., Int. Ed., 2012, 51, 11420–11422.
17. H. Bi, Y. Dai, P. Yang, J. Xu, D. Yang, S. Gai, F. He, B. Liu, C. Zhong, G. An and J. Lin, Small, 2018, 14, e1703809.
18. H. Bi, Y. Dai, P. Yang, J. Xu, D. Yang, S. Gai, F. He, G. An, C. Zhong and J. Lin, Chem. Eng. J., 2019, 356, 543–553.
19. J. D. Knoll and C. Turro, Coord. Chem. Rev., 2015, 282-283, 110–126.
20. F. Heinemann, J. Karges and G. Gasser, Acc. Chem. Res., 2017, 50, 2727–2736.
21. J.-Y. Lee, J.-Y. Lee, Y.-Y. Chang, C.-H. Hu, N. M. Wang and H. M. Lee, Organometallics, 2015, 34, 4359–4368.
22. S. Ray, R. Mohan, J. K. Singh, M. K. Samantaray, M. M. Shaikh, D. Panda and P. Ghosh, J. Am. Chem. Soc., 2007, 15042–15053.
23. S. G. Churusova, D. V. Aleksanyan, E. Y. Rybalkina, O. Y. Susova, V. V. Brunova, R. R. Aysin, Y. V. Nelyubina, A. S. Peregudov, E. I. Gutsul, Z. S. Klemenkova and V. A. Kozlov, Inorg. Chem., 2017, 56, 9834–9850.
24. W. Liu and R. Gust, Chem. Soc. Rev., 2013, 42, 755–773.
25. J. Ruiz, V. Rodriguez, C. de Haro, A. Espinosa, J. Perez and C. Janiak, Dalton Trans., 2010, 39, 3290–3301.
26. G. Gasser, I. Ott and N. Metzler-Nolte, J. Med. Chem., 2011, 54, 3–25.
27. S. Bonnet, Comments Inorg. Chem., 2014, 35, 179–213.
28. Z. Fan, J. Ni and A. Zhang, J. Am. Chem. Soc., 2016, 138, 8470–8475.
29. V. H. S. van Rixel, B. Siewert, S. L. Hopkins, S. H. C. Askes, A. Busemann, M. A. Siegler and S. Bonnet, Chem. Sci., 2016, 7, 4922–4929.
30. E. M. Hernández, S. Zheng, H. J. Shepherd, D. S. Yufit, K. Ridier, S. Bedoui, W. Nicolazzi, V. Velázquez, S. Bonnet, G. Molnár and A. Bousseksou, J. Phys. Chem. C, 2016, 120, 27608–27617.
31. L. Yang, D. R. Powell and R. P. Houser, Dalton Trans., 2007, 955–964.
32. S. Zheng, N. R. Reintjens, M. A. Siegler, O. Roubeau, E. Bouwman, A. Rudavskyi, R. W. Havenith and S. Bonnet, Chem. – Eur. J., 2016, 22, 331–339.
33. S. L. Hopkins, B. Siewert, S. H. Askes, P. Veldhuizen, R. Zwier, M. Heger and S. Bonnet, Photochem. Photobiol. Sci., 2016, 15, 644–653.
34. X. Li, J. Zhang, Z. Zhao, L. Wang, H. Yang, Q. Chang, N. Jiang, Z. Liu, Z. Bian, W. Liu, Z. Lu and C. Huang, Adv. Mater., 2018, 30, e1705005.
35. F. F. Hung, S. X. Wu, W. P. To, W. L. Kwong, X. Guan, W. Lu, K. H. Low and C. M. Che, Chem. – Asian J., 2017, 12, 145–158.
36. J. Fernandez-Cestau, B. t. Bertrand, A. Pintus and M. Bochmann, Organometallics, 2017, 36, 3304–3312.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details, graphical results, and videos. CCDC 1534260. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c8cc10134e

---