A Cytotoxic Bis(1,2,3-triazol-5-ylidene)carbazolide Gold(III) Complex Targets DNA by Partial Intercalation

Abstract

The syntheses of bis(triazolium)carbazole precursors and their corresponding coinage metal (Au, Ag) complexes are reported. For alkylated triazolium salts, di- or tetranuclear complexes with bridging ligands were isolated, while the bis(aryl) analogue afforded a bis(carbene) Au^I -CNC pincer complex suitable for oxidation to the redox-stable [Au^III (CNC)Cl]⁺ cation. Although the ligand salt and the [Au^III (CNC)Cl]⁺ complex were both notably cytotoxic toward the breast cancer cell line MDA-MB-231, the Au^III complex was somewhat more selective. Electrophoresis, viscometry, UV-vis, CD and LD spectroscopy suggest the cytotoxic [Au^III (CNC)Cl]⁺ complex behaves as a partial DNA intercalator. In silico screening indicated that the [Au^III (CNC)Cl]⁺ complex can target DNA three-way junctions with good specificity, several other regular B-DNA forms, and Z-DNA. Multiple hydrophobic π-type interactions involving T and A bases appear to be important for B-form DNA binding, while phosphate O⋅⋅⋅Au interactions evidently underpin Z-DNA binding. The CNC ligand effectively stabilizes the Au^III ion, preventing reduction in the presence of glutathione. Both the redox stability and DNA affinity of the hit compound might be key factors underpinning its cytotoxicity in vitro.

Keywords: anticancer; cytotoxicity; gold complexes; mesoionic carbenes; metallodrugs.

Scheme 1

Synthesis of the bis(triazolylidene) complexes 1 b and 1 c via a transmetallation route starting from the triazolium salt L1. The tetranuclear analogues of the monocarbene silver(I) complexes 1 a and 2 were also prepared from their respective triazolium salts, L1 and L2 (see text). The reaction conditions were as follows: a) 4 eq Ag₂O, 10 eq KCl, CH₂Cl₂, rt, 7d. b) 1.5 eq Au(tht)Cl (tht=tetrahydrothiophene), 1 d. c) 7 days in solution.

Scheme 2

Metalation of L3 and L4 to form the Au^I CNC pincer complexes 3 a and 4 a, followed by oxidation to the corresponding Au^III complexes, 3 b and 4 b. The reaction conditions were as follows: a) 5 eq KN(Si(CH₃)₃)₂, 1.5 eq Au(tht)Cl, THF, −60 °C, 3 d. b) 1.1 eq PhICl₂, CH₂Cl₂, rt, 10 min–3 h.

Figure 1

Molecular structures of a) 4 a and b) 4 b showing 50 % probability ellipsoids and partial atom‐numbering schemes. For clarity, the hydrogen atoms and counterions are omitted and the wingtip functionalities are displayed as wireframes. Selected bond lengths (Å) and angles (°) for 4 a: Au1−C1 2.024(3), Au1−C16 2.021(3), Au1−N4 2.445(3), C1−Au1−C16 169.85(15), C1−Au1−N4 86.12(14), C16−Au1−N4 84.65(13). For 4 b: Au1−C1 2.054(3), Au1−C16 2.040(4), Au1−N4 2.010(3), Au1−Cl1 2.266(10), C1−Au1−C16 171.54(14), N4−Au1−Cl1 172.97(9), C1−Au1−Cl1 92.83(10), C16−Au1−Cl1 90.24(10), C1−Au1−N4 89.86(12), C16−Au1−N4 88.01(12), C1−C2−C3−C8 4.6(6), C9−C14−C15−C16 −21.3(6).

Figure 2

a) Experimental and DFT‐calculated (inset) electronic absorption spectra of C ₂‐symmetry 4 b in DMSO. The absorption maxima (λ_max) are indicated for the experimental spectrum at 303, 336, 363 and 398 nm (red line). The absorption envelope for the DFT‐calculated spectrum is plotted with a band width of 2200 cm⁻¹ (full width at half maximum intensity, FWHM). b) Molecular orbitals involved in the three most intense absorption bands in the DFT‐calculated electronic spectrum of 4 b in DMSO. The percentage contribution of the electronic transition to each band is indicated. The transition dipoles (not shown) will be polarized in the plane spanning the carbazole and triazole ring systems for the two lowest‐energy transitions.

Figure 3

Electronic absorption spectra of 15 μM 4 b in a water (10 % V/V) and DMSO mixture containing GSH at a molar ratio of 1 : 7 [4 b]:[GSH]. The inset shows the change in absorbance at 415 nm with time. The data are fitted to a standard double exponential kinetic function, A=B ₁ e ^(−x/t1)+B ₂ e ^(−x/t2)+A _∞, where B ₁ and B ₂ are adjustable pre‐exponential factors, t ₁ and t ₂ are the time constants, and A_∞ the limiting absorbance. The derived parameters are k _n=1/t _n and τ_n=t _n ln(2); R ² for the fit is 0.9998.

Figure 4

a) UV‐vis absorption spectra of 4 b in 1 x PBS (10 % V/V DMSO) before (30.8 μM, blue spectrum) and after sequential additions of ctDNA (final [ctDNA]=87.7 μM, red spectrum). The spectra have been corrected for dilution and only selected intermediate spectra are plotted; the sloping (yet unchanging) background absorption reflects some aggregation of the chromophore in the buffer. Inset: change in the absorbance at 363 nm as a function of [ctDNA] fitted to the Hill binding isotherm. b) EMSA agarose gel for 4 b with pUC57 plasmid DNA (12.5 ng/well, 1x TAE buffer, 5 % V/V DMSO). The DNA forms are supercoiled (SC), nicked‐open circular (NOC), and linear (Lin). Control lanes (1 and 15, C) contain pUC57 DNA only. Lanes with increasing concentrations of ethidium bromide (EB, 2–4), 4 b (5–11), and Hoechst 33258 (HS, 12–14) compare the behaviour of a DNA intercalator, the Au^III complex, and a DNA minor groove binder, respectively. Experiments were done in triplicate and a representative assay is shown.

Figure 5

a) Plot of the change in relative viscosity of solutions of calf thymus DNA (ctDNA) as a function of r, where r is the analyte/DNA mole ratio (based on the DNA base pair concentration). The measurements were conducted at 37 °C in 1x phosphate‐buffered saline (PBS) containing 10 % (V/V) DMSO. Hoechst 33258 (HS) and ethidium bromide (EB) are the groove‐binder and intercalator controls, respectively. The data comprise several independent experiments and were fitted to single‐ and double‐Hill sigmoidal functions. For 4 b, r ₅₀=2.9(1) while for EB, r ₅₀₍₁₎=4.7(1.5) and r ₅₀₍₂₎=1.9(5)×10³, where r ₅₀ is the mole ratio at which 50 % dissociation occurs. b) ctDNA melt curves recorded at equivalent concentrations of added EB and 4 b. The ΔT _m values for EB and the gold complex are +1.8(2) and +2.7(2) °C, respectively, under non‐saturating conditions. The positive ΔT _m values reflect intercalation and stabilization of the DNA duplex; however, neither represent limiting ΔT _m ^max values. The data are fitted to Boltzmann sigmoidal functions.

Figure 6

Circular and linear dichroism spectra delineating the interaction of 4 b with ctDNA. All spectra are smoothed (8‐point fast Fourier transform) and were recorded in a Couette flow cell which orients the DNA colinear with the shear flow axis (i. e., flow direction). a) Difference CD spectra as a function of the mole ratio r of the gold complex to ctDNA base pairs (10 μM throughout) in PBS‐DMSO (10 % V/V) buffer at 37 °C. The difference spectra are corrected for dilution (normalized to an effective r=1) and have the native ctDNA CD spectrum subtracted throughout. The inset shows the best fit of the data at 207 nm to the Hill function (K _A=2.6(3)×10⁵ M⁻¹, Hill coefficient n=1.9(3)). b) Difference LD spectra (37 °C) for 4 b bound to ctDNA. Spectra for Hoechst 33258 (minor groove binder) and ethidium bromide (intercalator) are included. The samples, DNA concentration, and buffer were the same as in Part a).

Figure 7

a) Lowest‐energy pose for 4 b docked within the pocket of the DNA three‐way junction taken from PDB code 3I1D. The carbazole ring system of the metal complex is intercalated at the 5′‐TA‐3′ step within each strand. The shape complementarity between the metal complex (light blue) and the DNA target (purple) is discernible from the two views down the 3‐fold axis. Molecular surfaces were calculated using a 1.4 Å probe radius. b) View of the noncovalent interactions between the metal complex and the DNA bases (A and T) lining the triangular cavity of the three‐way junction (H atoms omitted for clarity). C−H⋅⋅⋅π, π⋅⋅⋅σ, and π⋅⋅⋅π interactions between the carbazole and Dipp groups of the complex and the DNA bases are dominant. A π⋅⋅⋅anion interaction is also present between the carbazole nitrogen and closest stacked thymine residue.