Dan E.
Wise
a,
Aimee J.
Gamble
b,
Sham W.
Arkawazi‡
b,
Paul H.
Walton
b,
M. Carmen
Galan
a,
Michael P.
O'Hagan
a,
Karen G.
Hogg
c,
Joanne L.
Marrison
c,
Peter J.
O'Toole
c,
Hazel A.
Sparkes
a,
Jason M.
Lynam
*b and
Paul G.
Pringle
*a
aSchool of Chemistry, University of Bristol, Cantock's Close, Bristol, BS8 1TS, UK. E-mail: paul.pringle@bristol.ac.uk
bDepartment of Chemistry, University of York, Heslington, York, YO10 5DD, UK. E-mail: jason.lynam@york.ac.uk
cImaging and Cytometry Laboratory, Bioscience Technology Facility, Department of Biology, University of York, UK
First published on 30th September 2020
We report cytotoxic ruthenium(II) complexes of the general formula [RuCl(cis-tach)(diphosphine)]+ (cis-tach = cis–cis-1,3,5-triaminocyclohexane) that have been characterised by 1H, 13C and 31P{1H} NMR spectroscopy, mass spectrometry, X-ray crystallography and elemental analysis. The kinetics of aquation and stability of the active species have been studied, showing that the chlorido ligand is substituted by water at 298 K with first order rate constants of 10−2–10−3 s−1, ideal for potential clinical use as anti-tumour agents. Strong interactions with biologically relevant duplex and quadruplex DNA models correlate with the activity observed with A549, A2780 and 293T cell lines, and the degree of activity was found to be sensitive to the chelating diphosphine ligand. A label-free ptychographic cell imaging technique recorded cell death processes over 4 days. The Ru(II) cis-tach diphosphine complexes exhibit anti-proliferative effects, in some cases outperforming cisplatin and other cytotoxic ruthenium complexes.
The seminal work on the anti-cancer properties of (η6-arene)Ru complexes has spurred the investigation of many coordination complex analogues of organometallic piano-stool complexes. For example, Alessio et al. replaced the arene with [9]aneS3 to give complex F with minimal loss of biological activity compared to its organometallic analogues.24 Furthermore, Ru-diphosphine complexes such as [(κ3-tpm)RuCl(diphos)]PF6 (G) showed activity in vitro.25 However, despite the variety of facially capping ligands available which could be used to modulate activity, this aspect of the complexes has received far less attention than modification of the other ancillary ligands on the metal.26–28 From a biological perspective, the narrow range of face-capping ligands that have been used limits the rate and extent of the substitution of the halido ligands by water. This rate is known to correlate with in vitro activity,8,29,30 and expanding its range is thus a critical factor in maximising the clinical potential of ruthenium complexes in the treatment of cancer.
The ligand cis-tach (cis–cis-1,3,5-triaminocyclohexane) forms face-capping complexes with many transition metals including ruthenium(II).31–35 The labile complex [RuCl(dmso-S)2(cis-tach)]Cl (1) is the precursor to N,N-chelate complexes 2a–c and P,P-chelate complexes 3a–f (Scheme 1) that have been previously reported.35 The influence of the cis-tach ligand was evaluated by comparison of the structural data with those for (η6-arene)Ru complexes. It has previously been demonstrated35,36 that cis-tach is a strong σ-donor, as would be expected due to the three nitrogen atoms coordinated to the metal. For instance, reaction of the DMSO complex 1 with diamines yields dicationic complexes in which one coordinated DMSO is retained, whereas 1 with diphosphines gave complexes with one chlorido ligand retained (Scheme 1). This difference in behaviour was rationalised on the basis of the different bonding characteristics of diamines and diphosphines. The electron-richness of the Ru created by the cis-tach, is augmented by the diamine σ-donors which leads to strong π-back-donation to the DMSO ligand, strengthening the Ru–S interaction. By contrast, diphosphines, which are better π-acceptors than DMSO, favour the coordination of chloride, which is presumed to be a π-donor. Importantly, for the use of these complexes as anti-cancer agents, it was reasoned that the Ru(cis-tach)(diphos) moiety may promote the rapid aquation of the Ru–Cl bond which might then result in enhanced in vitro activity.31,37
![]() | ||
Scheme 1 Synthesis of complexes 2a–c and 3a–f.35 |
In addition to their potential as effective anti-cancer agents, Ru complexes of cis-tach have other features that make them attractive for medicinal chemistry. For instance, the cationic Ru cis-tach complexes are readily prepared as chloride salts, obviating the use of the toxic PF6− anion in potential pharmaceuticals.38 The NH2 groups of the cis-tach ligand enhance the water solubility of the complexes and, moreover, may strengthen any binding to DNA through hydrogen-bonding interactions, in a similar manner to the DNA binding with [RuCl(η6-biphenyl)(en)]PF6 (C).8,9 Finally, the cyclohexane ring provides a hydrophobic face to the complex, giving steric protection to the hydrophilic metal centre.
It is in this context that we now report a detailed investigation of the in vitro activity of ruthenium cis-tach complexes. It is shown that a range of diphosphine derivatives exhibit activity against three tumour cell lines, in some cases with potency exceeding that of cisplatin or established anti-cancer ruthenium complexes. The extended aromatic backbones of the new diphosphines L1–L3 (Fig. 2) are shown to allow detailed insight into the nature of the biological interactions with their Ru-complexes via a range of physical inorganic and biological measurements including UV/visible, fluorescence and NMR spectroscopy as well as label-free cellular imaging techniques.
The X-ray crystal structure of [3h]PF6 (Fig. 3) demonstrates that the addition of the larger, planar aromatic quinoxaline diphosphine ligand L2 does not significantly alter the geometry of the (cis-tach)Ru complex, as shown by the overlap with the phenylene diphosphine analogue (3f) illustrated in Fig. 3. The cis-tach ligand adopts the expected κ3-coordination mode and there are intramolecular interactions detected between the N(4)H2 and the centroids of the phenyl rings of the PPh2 groups. In addition to the lipophilic cyclohexane ring, the PPh2 groups provide further hydrophobicity to the complex and have the potential to interact with biomolecules (see below).
Ruthenium mono-phosphine complexes containing an η6-arene, have previously been shown11,15,39–42 to be active against cancer cells. We now report clear antiproliferative activity with the P,P-chelates 3a–f as shown in Fig. 4 and some trends can be discerned from the data. (1) The activity generally increases with increasing chelate ring size: 3a < 3b < 3c ≃ 3d; complexes 3c and 3d are over twice as active as cisplatin against the A549 cell line and are equipotent to cisplatin against the A2780 cell line. (2) Complex 3b is over twice as active as complex 3e against both cell lines; although both 3b and 3e are 5-membered chelates, the more active chelate 3b has a less rigid backbone. (3) Complex 3f is significantly more active (by factors of ca. 10 and 7 against the two cell lines) than the ostensibly similar complex 3e. Although both 3e and 3f are rigid, 5-membered chelates, the phenylene backbone in 3f will make the complex more lipophilic. It was also speculated that the intercalating potential of the planar aromatic backbone present in 3f may also be a contributing factor in its higher activity than 3e.
![]() | ||
Fig. 4 Cell viability data in A549 and A2780 cells treated with cisplatin and 3a–f to show influence of ligand backbone on cytotoxicity. Antiproliferative activities were determined by MTT assay and dose response curves are given in Fig. S5 and 6.† The IC50 calculated is the concentration of drug required for 50% growth inhibition over a 72-hour period. The error bars represent one standard deviation from three independent experiments. |
To explore this hypothesis further, the (cis-tach)Ru complexes 3g–i containing diphosphines with extended aromatic surfaces: terthiophenyl diphosphine (L1), quinoxaline diphosphine (L2) and dibenzo[f,h]quinoxaline diphosphine (L3) (see Fig. 2 and Scheme 2) were tested. It was postulated that these novel complexes might exhibit dual-function cytotoxicity by covalently binding to biomolecules and by intercalation with DNA in a similar way to the functioning of the cytotoxic Pt complex phenanthriplatin.43,44 If this were the case, it was reasoned that 3g–i would be expected to be more active than the first-generation Ru cis-tach complexes 3a–f.
We assessed the antiproliferative activity of 3g–i against A549 cells by a 72-hour MTT assay and found that the complexes were comparable in activity to the most active tested P,P-chelate complexes 3c–d with IC50 values of 1.83 ± 0.66 μM for 3g, 11.81 ± 1.23 μM for 3h, and 5.06 ± 1.01 μM for 3i. Taking into account the experimental errors inherent in the MTT assays, the difference between the activity of 3g and 3c–d is not statistically significant. Encouraged by the activities of these more lipophilic derivatives, we sought to understand their interactions with a variety of biomolecules and study their activity in vitro. The MTT data for complexes 3g–i (Table S2†) demonstrated that their activity was comparable to the most active (cis-tach)Ru complexes. However, detailed time course studies using in situ cellular imaging revealed further details of the behaviour of the complexes and indicated that 3g–i are significantly more active against A549 and 293T cells when compared to cisplatin and 3f (see below).
Two potential modes of action of the Ru complexes 3a–i were investigated. Firstly, aquation of the Ru–Cl bond to the labile Ru–OH2 complex followed by covalent interactions with nucleosides; and secondly, the ability of the complexes to interact with DNA by non-covalent interactions.
![]() | ||
Fig. 5 Stacked 31P{1H} NMR spectra of a 500 μM solution (pH 7.4) of 3c with various chloride concentrations to assign the resonances corresponding to the Ru chlorido (3c) and Ru aqua (3c′) complexes. |
Complex | T/K | k/10−3 s−1 | t ½/s |
---|---|---|---|
a Measurements for the aquation of 3b, 3c, 3f and 3h (300 μM) in aqueous solution buffered at pH 7.4 (10 mM sodium phosphate). | |||
3b | 288 | 2.09 ± 0.02 | 331 ± 3 |
3b | 293 | 3.60 ± 0.08 | 192 ± 5 |
3b | 298 | 6.55 ± 0.06 | 106 ± 1 |
3b | 303 | 10.7 ± 0.20 | 65 ± 1 |
3b | 310 | 21.0 ± 0.70 | 33 ± 1 |
3c | 298 | 63.9 ± 6.0 | 10 ± 2 |
3f | 298 | 2.23 ± 0.12 | 311 ± 17 |
3h | 298 | 1.02 ± 0.01 | 679 ± 3 |
The rate constants for the aquation of 3b (6.55 ± 0.06 × 10−3 s−1) and 3c (63.9 ± 6.0 × 10−3 s−1) at 298 K are approximately 5 and 15 times faster than the η6-biphenyl complex [RuCl(η6-bip)(en)]+ (C) (1.28 × 10−3 s−1) respectively. This difference could be attributed to a weakening of the Ru–Cl bond, as shown by its lengthening in 3b (2.4431(14) Å) and 3c (2.4404(4) Å) when compared to RAen complex C (2.405(6) Å), due to the large trans-effect of the nitrogen donors in the cis-tach ligand.
The reaction rates for aquation and anation at different temperatures allowed for the determination of the Arrhenius activation energy (Ea), activation enthalpy (ΔH‡) and activation entropy (ΔS‡). The Ea and ΔH‡ for the aquation reaction are comparable to those reported for complex C (Table 2), consistent with the fundamental mechanistic steps in the aquation process being the same. Both 3b and 3c were found to be stable for the duration of a typical 72 h MTT assay experiment. Furthermore, over a two-week period at 37 °C, the 1H NMR spectrum of 3b in 10% D2O/90% H2O did not change. Since the rate of aquation of the Ru–Cl is rapid, the biological activity is likely more dependent on the binding to biomolecules once this aquation step has taken place.
Complex | E a/kJ mol−1 | ΔH‡/kJ mol−1 | ΔS‡/J K−1 mol−1 |
---|---|---|---|
a Arrhenius activation energy (Ea), activation enthalpy (ΔH‡) and activation entropy (ΔS‡) for the aquation and anation of 3b and 3b’ at pH 7.4. b Complex C is shown in Fig. 1; values taken from ref. 29. | |||
3b | 79.8 ± 0.7 | 77.3 ± 0.7 | −27.6 ± 4.8 |
3b′ | 84.9 ± 1.0 | 82.4 ± 1.0 | 24.4 ± 3.4 |
C![]() |
75.6 ± 0.6 | 73.1 ± 0.6 | −55.7 ± 2.0 |
C′![]() |
76.7 ± 1.3 | 74.1 ± 1.3 | −13.6 ± 4.5 |
The acid-dissociation constants for these complexes are higher than those commonly obtained for (η6-arene)ruthenium(II) complexes, including [Ru(OH2)(η6-biphenyl)(en)]2+ (7.71 ± 0.01), [Ru(OH2)(η6-tha)(en)]2+ (8.01 ± 0.03, tha = tetrahydroanthracene), and [RuCl(OH2)(η6-C6H6)(PTA)]2+ (9.2 ± 0.03). The highest pKa values reported are obtained for O,O-chelates, such as [Ru(OH2)(η6-p-cymene)(malonate)]+ (9.23 ± 0.02) and [Ru(OH2)(η6-p-cymene)(acac)]+ (9.41 ± 0.01).8,10,48,49 The conclusion drawn from the pKa measurements of complexes 3b and 3c is that the deprotonated forms are physiologically inaccessible by more than 3 pH units and aquation therefore affords exclusively the aqua complex.
One likely step in the mechanism of the antiproliferative effect of the Ru complexes 3a–i is through coordination of a biomolecule to the site on Ru initially occupied by the chlorido ligand. The Ru complexes share a structural feature with cisplatin – a nitrogen donor trans to a chlorido ligand and have favourable aquation kinetics. Additionally, the amine groups of the cis-tach ligand are located cis to the chlorido ligand and may have a role in strengthening interactions with a bound molecule through hydrogen bonding. As a result of this reasoning, the series of DNA binding experiments described below were carried out.
The equilibrium constant, K, for the aquation of 3b (K = 30.6 ± 1.7 × 10−3 M) is significantly greater than for 3c (K = 5.9 ± 0.1 × 10−3 M) and results in a considerable difference in speciation (Fig. S3 and Table S1†). The two equilibrium constants span those of the RAen complexes. The equilibrium constant for the aquation of 3b may result in the formation of the aqua species in the blood, cytoplasm, and cell nucleus, leading to a greater possibility for deactivation reactions to occur. For example, cisplatin is deactivated by glutathione binding prior to reaching the nucleus51 and a similar process could account for the 10-fold reduction in activity of 3b compared to 3c.
The combination of a smaller equilibrium constant for aquation and a lower pKa of the aqua species results in a favourable proportion of 3c′ formed under different physiological environments. It is predicted for 3c that a lower proportion of aqua species would be present in the blood, but a high proportion in the cell nucleus compared to RAen complexes. Therefore, the proportion of aqua species for 3c is greater than for many other Ru complexes that have been evaluated for anti-cancer activity (Table S1†). This produces a good balance between protection of the complex outside the cell and a higher degree of activation inside the cell. As demonstrated by the kinetic study, the rate at which the complex is aquated once inside the cell is very rapid and is not a factor in the intracellular speciation.
Substitution of the chorido ligand with a guanine derivative was demonstrated by the reaction of 3b with guanosine (Guo) after incubation at 37 °C for 24 h in water. The solution was diluted to 0.1 mM with 50% methanol in water and the ESI mass spectrum recorded. An ion with mass and isotope pattern corresponding to [M–Cl + Guo–H]+ (30%) and [M−Cl + Guo]2+ (30%) was observed at m/z 911.1 and 456.2 respectively along with chlorido (100%) and hydroxy (30%) species. In contrast to 3b, the reaction of 3c with EtG did not produce an observable adduct in the 1H NMR spectrum after 24 h at 37 °C. Therefore, it is plausible that coordination to 3c of the N7 of guanine may not be involved in the mechanism by which this complex inhibits proliferation.
![]() | ||
Fig. 6 Chemical structures of the DNA intercalating dye ethidium bromide and the known intercalating complex [Ru(bpy)2(DPPZ)]2+. |
For titrations with complexes 3g, 3h, 3i, a gradual decrease in emission was observed (Fig. 7 and Fig. S7†) implying that these complexes outcompete the EB to interact with DNA. The same titration was performed with 3b and 3f but no appreciable decrease in emission was observed for the dppe complex indicating little intercalative interaction is present (Fig. S7†).
![]() | ||
Fig. 7 (A) Emission spectra of CT-DNA (50 μM) and EB (5 μM) competition assay with 3h (0–112.5 μM). (B) Stern–Volmer plots EB-CT-DNA vs. the concentration of 3g, 3h and 3i. |
For 3f, the decrease was so small that a binding constant could not be calculated and therefore intercalation is not considered a viable mode of action for this complex. The binding constants (Kapp) for 3g, 3h and 3i are 3.07 ± 0.07 × 105 M−1, 1.16 ± 0.04 × 105 M−1 and 5.11 ± 0.25 × 105 M−1 respectively (Table 3), indicating that increasing the aromatic surface of the ligand backbone (3h to 3i) gave a five-fold increase in binding affinity.
These binding constants are comparable, and in some cases superior to, rigid dinuclear (η6-arene)Ru complexes previously reported.59 The apparent binding constant of the known DNA intercalator [Ru(bpy)2(DPPZ)]2+ was calculated as 1.75 ± 0.02 × 106 M−1, only one order of magnitude higher than the (cis-tach)Ru complexes.
This therapeutic hypothesis has led to many groups designing G-quadruplex binders as potential anticancer agents67–70 and many metal complexes are known to bind G-quadruplexes effectively through covalent and non-covalent interactions.71 For example, Liu et al. found that ruthenium polypyridyl complexes containing 4idip (4-indoleimidazo[4,5-f][1,10]phenanthroline) ligands were able to selectively stabilise the human telomeric G-quadruplex structure.68,69
As a result of the intercalating ability of 3g–i indicated by the EB assays, we investigated the ability of the complexes to stabilise G-quadruplex DNA and duplex DNA structures. The extent of stabilisation was quantified by performing a fluorescence resonance energy transfer (FRET) assay initially reported by De Cian et al.72 (see ESI†). The change in DNA melting temperature (ΔT1/2) induced by a Ru complex compared to that of the oligonucleotide in the absence of complex provides an indication of the capacity of the complex to stabilise the G-quadruplex structure. We chose to investigate three models of G-quadruplex DNA and one of duplex DNA (see Fig. 9). The human telomeric sequence (F21T) was studied in potassium- and sodium-containing buffer owing to the known influence of the metal ion on the polymorphism of this sequence.73,74 The G-quadruplex sequence found in the c-myc oncogene promoter (FmycT) was selected as a model of this anticancer target.
The results of the FRET experiments are shown in Fig. 9 and representative raw data in the ESI (Fig. S8†). Complexes 3g and 3h did not induce any appreciable stabilisation of quadruplex DNA (ΔT1/2 < 3 °C at 1 μM complex) but 3i did stabilise F21T (ΔT1/2 = + 7.5 ± 2.3 °C) in sodium-containing buffer. Additionally, 3i was selective for quadruplex DNA structures, stabilising the quadruplex sequence FmycT (ΔT1/2 = + 6.2 ± 1.5 °C) whilst stabilisation of the duplex sequence F10T (Fig. 9) was negligible. Meanwhile, the same complex did not significantly stabilize F21T in K+-rich buffer (ΔT1/2 < 3 °C), suggesting that as well as G4/duplex selectivity, the complex can also discriminate between different G-quadruplex topologies to some extent. As a control, the well-known DNA intercalator complex [Ru(bpy)2(DPPZ)]2+ did not significantly stabilise quadruplex DNA (ΔT1/2 = + 1.6 ± 0.4 °C for F21T) or duplex DNA (ΔT1/2 = + 1.2 ± 0.3 °C for F10T) as previously reported.67
The dibenzo[f,h]quinoxaline moiety in 3i provides a large aromatic surface that may selectively stabilise G-quadruplex DNA through preferential association with the large G-tetrads over intercalation with Watson–Crick base pairs, akin to the 4idip examples previously reported. Complex 3i showed stronger binding to quadruplex DNA than 3g and 3h in experiments with quadruplex DNA as well as CT-DNA as previously shown. This interesting discovery that Ru cis-tach complexes interact with quadruplex DNA warrants further study.
This technique enables quantification of the total cellular dry mass as an indicator of cell death and growth. For A549 and 293T cells, the effect of treatment with cisplatin and 3i is shown at 36 and 72-hour time-points (Fig. 10 and 11). For 0-hour time points and compounds 3f–h see Fig. S11–13.† The MTT assay is an endpoint colourmetric readout of cell viability and does not provide any visual characteristics of the cells state. In this preliminary study, integrated image analysis software (Livecyte Cell Analysis Toolbox) was used to extract real-time changes in morphology and dry mass of each cell over time. The summed mass of the cellular components excluding water (known as dry mass, see Fig. S9 and 10†) was calculated and for each treated population of cells used as a measure of the combined growth and proliferation. For A549 cells (Fig. 10D), the reduction in dry mass is greatest for 3g and 3i followed by the slightly less active derivative 3h (which shows comparable results to cisplatin), and 3f gave the smallest decrease in dry mass of those tested. This difference in dry mass reduction between cisplatin and 3f is notable as the MTT assay derived IC50 values are within error of each other (IC50 = 2.7 μM). For 293T cells (Fig. 11D), this effect is more pronounced with 3f causing significantly less reduction in dry mass than cisplatin.
The other (cis-tach)Ru complexes 3g–i caused significant cell death at a concentration of 6.25 μM as is evident from the cell images and dry mass curves, and this greater activity is consistent with 3g–i exhibiting cytotoxicity by multiple mechanisms.
New analogues with planar aromatic backbones have been shown to intercalate strongly with CT-DNA models, and, in the case of 3i also selectively stabilise G-quadruplex DNA over duplex DNA. The anti-proliferative effect has been monitored by LiveCyte, label-free, time-lapse imaging and stark differences are observed between the phenylene derivative 3f and the extended aromatic derivatives (3g and 3i). Overall, these preliminary biological studies suggest that (cis-tach)Ru diphosphine complexes exhibit a dual-action cytotoxic effect, targeting cellular DNA by intercalation, as well as by modes of action involving covalent binding with DNA. This robust, water-soluble molecular architecture could be further developed to produce next generation ruthenium chemotherapeutic agents. Further cell studies exploiting the tunability of phosphine ligands that result in targeted metallodrugs are ongoing.
Footnotes |
† Electronic supplementary information (ESI) available: Experimental, characterisation, X-ray crystallography and detailed assay procedures. Table S3, CCDC 1959465. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0dt02612c |
‡ Current address, Department of Chemistry, University of Garmian, Kurdistan Region, Iraq. |
This journal is © The Royal Society of Chemistry 2020 |