In vitro antitumor activity of water-soluble copper(I) complexes with diimine and monodentate phosphine ligands

2.1 Synthesis and characterization of copper(I) complexes 1–9

Copper complexes were synthesized via ligand-exchange reactions starting from the labile Cu(I) precursor [Cu(CH₃CN)₄][BF₄]. All of the complexes are deeply colored solid (from yellow to purple), stable in air and in solutions; they are soluble in water, methanol and acetonitrile exhibiting a solvatochromic behavior. A slightly modified synthesis of complex (1) has recently been reported in a nitrogen atmosphere, and the application of (1) in dye sensitized solar cells (DSSC) was described (Wills et al., 2013) The X-ray molecular structure of (1)-like evidenced that the two di-imines were virtually orthogonal adopting a pseudo tetrahedral geometry in the [Cu(BCA)(H₂BCA)]⁻ anion, that was counterbalanced by a [HEt₃N]⁺ cation, both features supporting the occurrence of a Cu(I) complex. Although there was no evidence of the [PF₆]⁻ anion in the unit cell, characterization in solution by multinuclear ³¹P and ¹⁹F NMR spectroscopy and the detection of the peak at m/z 145 in the ESI(−) spectrum of the complex indicated the additional presence of one equivalent of co-crystallized [HEt₃N][PF₆] (Wills et al., 2013).

During our syntheses we utilized more physiologically friendly Na⁺ cations coming from di-imine ligands and [BF₄]⁻ anions coming from the labile Cu(I) precursor. Also in our case, elemental analysis evidenced the presence of co-crystallized NaBF₄ in most compounds (complexes 1, 4–9), in addition to the sodium atoms necessary to the overall charge balance of the complex. The presence of BF₄ was authenticated by detection of a broad band at ca. 1030 cm⁻¹ in the pertinent IR spectra.

According to the characterization presented in the above reference study (Wills et al., 2013), our compounds were characterized by means of multinuclear ¹H, ¹³C and ³¹P NMR spectroscopy and electrospray ionization (ESI) mass spectrometry both in the positive and negative ion modalities. The easy collection of NMR spectra, typical of diamagnetic Cu(I) compounds, rules out metal oxidation phenomena occurring in solution in accordance with the presence of bulky di-imine ligands in homoleptic complexes 1–3, and with the further presence of phosphines in mixed-ligand complexes 4–9.

¹H and ¹³C NMR spectra of bis-substituted di-imine complexes 1–3 were consistent with the proposed formulation. ³¹P NMR spectra of mixed-compounds 4–9 showed each a distinctive singlet signal, downfield shifted and broadened, both features typical of copper coordination, compared to the signal of uncoordinated DAPTA or PTA phosphines (see Fig. 2).

Close

Fig. 2

Comparison of the ³¹P NMR signals of uncoordinated PTA and DAPTA ligands with those arising from mixed-ligand complexes [Cu(BCSNa)(PTA)₂]·NaBF₄ (8) and [Cu(i-BCSNa)(DAPTA)₂]·NaBF₄ (6).

Export to PPT

For illustrative purposes, here follows the discussion of the NMR spectra of the mixed complex (4). Its ¹H spectrum shows a set of coupled aromatic protons in the 8.51–7.79 ppm range arising from the coordinated BCA framework (Fig. S1). Among these signals, the unique downfield shifted singlet corresponds to H_8,8′ (see Fig. 1 for proton caption). A further set of aliphatic signals in the 5.43–3.45 ppm range arises from the diastereotopic methylene protons of coordinated DAPTA. The coupling of diastereotopic protons of each methylene group is distinguishable in the ¹H-¹H COSY map of Fig. S2. Two further singlet signals at 1.91 and 1.77 ppm are due to magnetically unequivalent acetyl protons of DAPTA. Proton integrations indicate that coordination of the different ligands to the Cu(I) ion occurs in the 1:2 BCA:DAPTA ratio. The additional ¹H-¹³C HMQC map allows the unequivocal assignment of ¹³C signals from the relative ¹H-¹³C coupling in accordance to the proton attribution (Fig. S3).

ESI-MS, in both positive and negative ion modalities, were utilized for the characterization of our Cu(I) complexes. ESI(−)MS was employed for homoleptic, bis-substituted compounds 1–3, showing the molecular ion species at m/z 749 for [Cu(HBCA)₂]⁻ and m/z 1101 for [Cu(HBCS)₂]⁻. Bis-substituted species are detected together with mono-substituted adducts at m/z 405 ([Cu(BCA)]⁻) and at m/z 581 ([Cu(BCS)]⁻. Cleavages of CO₂ from the coordinated BCA and of SO₃ from coordinated BCS ligands were also detected.

For illustrative purposes, the ESI(−)MS behavior of complex (2) is below detailed. Full ESI(−) spectrum is outlined in Fig. 3, and the collisionally induced decomposition (CID) pathway is presented in Fig. 4 together with a pictorial sketch of fragmentation. The base peak in the full spectrum corresponds to the doubly charged [(BCS)]²⁻ species at m/z 259, indicating that the complex is labile under ESI conditions. Significant peaks including copper are detected at m/z 1165 (10%), m/z 1101 (30%), m/z 581 (78%) and m/z 367 (30%). They correspond, respectively, to the ions [Cu(HBCS)(CuBCS)]⁻, [Cu(HBCS)₂]⁻ (molecular ion species), [Cu(BCS)]⁻ and triply charged [Cu(HBCS)₂]³⁻. CID of the molecular ion at m/z 1101 displays loss of a neutral di-imine with formation of the fragment ion at m/z 581([Cu(BCS)]⁻), that, in turn, shows subsequent losses of SO₃ (−80 Da) and SO₂ (−64 Da) with formation of the fragment ion at m/z 501 and 437, respectively. As shown in Fig. 5, the attribution of the molecular ion species at m/z 1101 has been authenticated by comparison of the experimental and calculated cluster ion traces.

Close

Fig. 3

Full ESI(−) spectrum of homoleptic, bis-diimine complex (2).

Export to PPT

Close

Fig. 4

Collisionally induced decomposition of the ion at m/z 1101 corresponding to the anion of complex (2).

Export to PPT

Close

Fig. 5

Comparison of the experimental and calculated clusters of the ion at m/z 1101 corresponding to the anion of complex (2).

Export to PPT

Considering mixed-ligand complexes 4–9, ESI(+) spectra gave more informative results, whereas ESI(−) spectra did not show any ions including phosphines. Full ESI(+) spectra, while not showing the molecular ions [Cu(diimine)(phosphine)₂]⁺, exhibited fragment ions of general formula [Cu(diimine)(phosphine)]⁺ generated from the loss of a phosphine group. This behavior resembles that already observed for a family of similar [Cu(N,N-{CH₂(pz)₂}(phosphine)₂]⁺ compounds, in which the molecular ion peak was never observed, but the adduct [Cu(N,N-{CH₂(pz)₂}(phosphine)]⁺ corresponding to the loss of one phosphine ligand were always detected (Porchia et al., 2013).

As an example of DAPTA derivatives, complex (4) displayed abundant [Cu(H₂BCA)(DAPTA)]⁺ at m/z 636, [Cu(HBCANa)(DAPTA)]⁺ at m/z 658 and [Cu(BCANa₂)(DAPTA)]⁺ at m/z 680. These mixed species were flanked by low abundant rearranged ions [Cu(DAPTA)₂]⁺ at m/z 521 and [Cu(DAPTA)₃]⁺ at m/z 751, and by the metal free base peak [(DAPTA)₂ + Na]⁺ at m/z 481 (Fig. S4). CID of the [Cu(BCANa₂)(DAPTA)]⁺ ion at m/z 680 ion showed loss of DAPTA (−229 Da) followed by two subsequent losses of CO₂ with formation of the fragment ions at m/z 451, 407 and 363, respectively (Fig. S5). The latter fragmentation processes were accompanied by addition of water with formation of adducts at m/z 425 and 381, respectively.

Regarding PTA-containing mixed compounds 7–9, the predominant peak in the ESI(+) full spectrum of complex 7 was due to the rearranged ion [Cu(PTA)₂]⁺ at m/z 377, flanked by low abundant [Cu(PTA)₃]⁺ at m/z 533 (25%) and [Cu(BCANa₂)(PTA)]⁺ at m/z 608 (10%). As previously mentioned, ESI(−) spectrum did not show any peak containing PTA, but formation of clusters including BCA and their fragments due to loss of CO₂. Protonated, sodiated and cuprated [Cu(BCA)]₂²⁻ dimer gave abundant ions at m/z 810, 833 and 874, respectively. The ESI(+) spectrum of complex (7) and the MSⁿ fragmentation of [Cu(H₂BCA)(PTA)]⁺ at m/z 564 (Fig. S6) displayed similar profiles to those described for complex 4.

Compounds 8 and 9 showed almost identical ESI(+) spectra giving the base peak at m/z 785 corresponding to [Cu(BCSNa₂)(PTA)]⁺, with low abundant [Cu(PTA)₂]⁺ at m/z 377 (10% for 8 and 35% for 9). As observed in the spectra of complexes 2 and 3, the ESI(−) traces of complexes 8 and 9 showed abundant peaks corresponding to [Cu(BCS)]⁻ and [Cu(i-BCS)]⁻ at m/z 581 due to loss of PTA ligands, and different sodiated and cuprated adducts at m/z 1165, 1187 and 1227, corresponding to [Cu(HBCS)(BCSCu)]⁻, [Cu(BCSNa)(BCSCu)]⁻ and [Cu(BCSCu)(BCSCu)]⁻, respectively (see Fig. S7 for complex 8).

2.2 Cytotoxicity

The in vitro antitumor activity of the newly synthesized copper(I) complexes (1–9) and of the corresponding uncoordinated ligands (PTA, DAPTA, BCANa₂, BCSNa₂ and i-BCSNa₂) were evaluated against a panel of human tumor cell lines derived from solid tumors. The cytotoxicity parameters, in terms of IC₅₀ (the median growth inhibitory concentration calculated from dose–survival curves) obtained after 72 h exposure, are listed in Table 1. Cell lines representative of lung (A549), colon (HCT-15), breast (MCF-7) and pancreatic (BxPC3) cancers, along with melanoma (A375), have been included. For comparison purposes, cytotoxicity of cisplatin, the most widely used metal-based anticancer drug, was assessed under the same experimental conditions.

Cancer cell lines included in the screen were endowed with different degrees of sensitivity to cisplatin and showed a consistently good response to treatment with all tested compounds (concentration- and time-dependent antiproliferative profiles, data not shown), except the uncoordinated PTA ligand, which proved to be ineffective. Also the other uncoordinated di-imine and phosphine ligands, BCANa₂, BCSNa₂ and i-BCSNa₂ and DAPTA proved to be scarcely effective in decreasing cancer cell viability (even if detectable, average IC₅₀ values were always over 50 μM). It’s worth noting that uncoordinated lipophilic diimines, such as bipy and phen, exhibit cytotoxic activities in some cases comparable with that shown by cisplatin, due to their interaction with DNA by aromatic π stacking (Gandin et al., 2013).

A different behavior can be evidenced between the homoleptic complexes 1–3 and the heteroleptic ones 4–9, as the former displayed a lower in vitro antiproliferative activity. Among homoleptic di-imine compounds (1−3), complex 1, bearing the BCA ligand, was the most interesting derivative, yielding average IC₅₀ values very similar to those obtained with cisplatin (7.6 and 8.2 μM, respectively). Complexes 2 and 3 were about 2- and 5-fold less effective than 1, respectively.

Instead, mixed-ligand compounds (4−9) were quite active and in particular complexes bearing BCS and i-BCS ligands showed a marked cytotoxic activity, with IC₅₀ values in the low micromolar range, and slightly lower than those recorded with cisplatin.

Complex 6, the mixed copper(I) complex bearing i-BCS and DAPTA ligands, was the most promising derivative (mean IC₅₀ values being 5.2 and 8.2 μM for 6 and cisplatin, respectively). Against human BxPC3 pancreatic and A549 lung cancer cells, complex 6 was roughly 4- and 5-fold more effective than the reference drug. Similarly, complexes 8 and 9 showed average IC₅₀ values toward the in-house cell panel of 5.8 and 5.9 μM, respectively. Noteworthy, BxPC3 pancreatic cancer cells were the most sensitive cell line to copper(I) complexes. In detail, complex 8 bearing BCS and PTA ligands was roughly 5 times more effective than cisplatin against human pancreatic BxPC3 cell, notoriously poor sensitive to cisplatin.

All copper derivatives have been additionally tested for their in vitro antitumor activity on two human cell line pairs which have been selected for sensitivity/resistance to cisplatin (ovarian cancer cells 2008/C13^∗). Cytotoxicity on sensitive and resistant cells was assessed after 72 h of drug treatment by an MTT test. Table 2 shows the cytotoxicity parameters, in terms of IC₅₀ and the resistance factor (RF), the latter defined as the ratio between IC₅₀ values calculated for resistant cells and those obtained with sensitive ones. All copper complexes exhibited a different cross-resistance profile than cisplatin. Actually, the RF values were from 6- to 33-fold lower than that of cisplatin, indicating the absence of cross-resistance phenomena. These results are consistent with the hypothesis that copper species possess mechanism(s) of action different from that shown by platinum drugs.

The cytotoxicity of tested complexes was also evaluated against non-tumor cells in rapid proliferation, the human embryonic kidney HEK293 cells (Table 3). All copper(I) complexes elicited a cytotoxic activity similar or even lower than that recorded with cisplatin. Interestingly, the selectivity index values (SI = quotient of the average IC₅₀ toward normal cells divided by the average IC₅₀ for the malignant cells) calculated for almost all complexes (excluding derivative 3) were higher than that calculated with cisplatin, thus attesting a preferential cytotoxicity of copper(I) complexes versus neoplastic cells. Among all in the series, the mixed copper di-imine/phosphine compounds [Cu(BCS)(PTA)₂]Na₂BF₄ (8) and [Cu(i-BCS)(DAPTA)₂]Na₂BF₄ (6) distinguished as those possessing the highest SI values.

In addition, 6 and 8 showed SI_BxPC3 values up to 6 times higher than that calcolated with cisplatin, being the SI_BxPC3 the quotient of the IC₅₀ toward normal cells divided by the IC₅₀ for the pancreatic BxPC3 cancer cells.

2.3 Cellular accumulation and partition coefficient

With the aim of identifying a possible correlation between cytotoxic activity and cellular accumulation, the cellular copper content was measured in BxPC3 cells treated for 24 or 36 h with equimolar concentrations (5 µM) of tested compounds. The cellular copper levels were quantified by means of GF-AAS analysis, and the results, expressed as ng Cu per 10⁶ cells, are shown in Fig. 6.

Close

Fig. 6

Cellular uptake. Intracellular accumulation of copper complexes 1–9. BxPC3 cells were incubated with 5 µM of copper complexes for 24 and 36 h, and cellular copper content was detected by GF-AAS analysis. Error bars indicate the standard deviation. *p < 0.05; ^**p < 0.01 compared with the control.

Export to PPT

Although to a different extent, homoleptic and heteroleptic BCS and BCA derivatives (1, 2, 4, 5, 7 and 8) accumulated in a time-dependent manner, whereas cells treated with i-BCS complexes (3, 6 and 9) did not show a time-dependent increase in cellular copper content.

To appreciate whether a correlation between lipophilicity and cellular uptake exists, the octanol/water partition coefficients (log P values outlined in Fig. 7) were measured using the shake-flask method. Among investigated compounds, i-BCS-containing derivatives 3, 6 and 9 showed the highest hydrophilic character (logP = −0.21, −0.34 and −0.19, respectively). Considering the effect of phosphine on the hydrophilic balance of the compounds, PTA-containing complexes 7–9 showed all an elevated hydrophilic character. A linear correlation between lipophilicity and cellular association has been evidenced in the case of mixed-ligand complexes 4–9 (Fig. S8).

Close

Fig. 7

Log P values (octanol/water) of complexes 1–9 obtained by shake flask method. Results are expressed as the mean ± SD of three independent experiments.

Export to PPT

Matching the cytotoxic activity with the cellular uptake, a correlation was evidenced in the case of homoleptic copper(I) complexes (1−3). In fact the cellular copper content of BxPC3 cells treated for 36 h with the most cytotoxic derivative 1 was significantly higher than that recorded with compounds 2 and 3. On the contrary, in the case of mixed complexes (4−9) no direct correlation could be derived.

2.4 Effect of copper(I) complexes on ultrastructure morphology

It has been previously demonstrated that phosphino copper(I) complexes induce endoplasmic reticulum (ER) stress and trigger a nonapoptotic mechanism of programmed cell death (PCD), termed paraptosis (Gandin et al., 2012). On the contrary, mixed phosphino/di-imine derivatives induce apoptosis in different types of cancer cells (Gandin et al., 2013). In order to characterize the cellular morphological changes induced by the present copper(I) compounds, human pancreatic BxPC3 cancer cells were treated with complex 9 as representative of mixed-ligand compounds and analyzed by transmission electron microscopy (TEM) (Fig. 8). 9-Treated BxPC3 cells showed the nucleus with a clearly defined ultrastructure, and regularly distributed chromatin. In addition, mitochondria appeared conserved in shape and internal structures, and no swelling features (increase in size and decrease in turbidity) could be detected. On the other hand, striking morphological changes, such as ER swelling and intracytoplasmic vacuoles, previously characterized as hallmarks of paraptotic cell death were distinctly noticed. Furthermore, 9-treated cells stained with Hoechst 33258 did not show nuclear DNA condensation or apoptotic body formation, classical hallmarks of apoptosis (Fig. 8, panel B).

Close

Fig. 8

Morphological changes. (A) Transmission electron micrographs of BxPC3 cells after 48 h of treatment with IC₅₀ concentrations of compound 9. (a) control cells; (b),(c), and (d) 9. (B) Hoechst staining. BxPC3 cells were incubated for 48 h with IC₅₀ of 9 (b) and cisplatin (c). Panel a represents the control untreated BxPC3 cells.

Export to PPT

Concerning the molecular basis underlying the mechanism of action of antiproliferative copper(I) complexes, some investigations recognized DNA (Gandin et al., 2013) or the proteasome multiprotein complex as the main molecular targets (Gandin et al., 2014; Gandin et al., 2012). On this basis, we evaluated the ability of complex 9 to interact with DNA in vitro by means of UV absorption titration and gel electrophoresis experiments. In addition, we assessed the ability of complex 9 to hamper proteasome proteolytic activities by testing chymotrypsin-like (CTL), trypsin-like (T-L) and caspase-like (C-L) activity of 26S proteasome isolated from rabbit. Neither DNA interaction nor proteasome inhibition were detected (data not shown), thus suggesting that its antiproliferative efficacy does not rely on DNA or proteasome targeting.

4.1 Materials and general methods

Commercially available substances were of reagent grade and used without further purification. The reagents 4,4′-dicarboxy-2,2′-biquinoline disodium salt (BCANa₂), disodium 2,9-dimethyl-4,7-diphenyl-1,10-phenantrolinedisulfonate (BCSNa₂), bathocuproinedisulfonic acid disodium salt (i-BCSNa₂) and 5,6-diphenyl-3-(2-pyridyl)-1,2,4-triazine-4,4″-disulfonic acid sodium salt (ferrozine sodium salt, HFzNa) were obtained from commercial sources (Fluka, Sigma-Aldrich). The Cu(I) precursor [Cu(CH₃CN)₄][BF₄] was prepared by reaction of [Cu(H₂O)₆][BF₄]₂ with metallic copper in acetonitrile. 1,3,5-triaza-7-phosphaadamantane (PTA), 3,7-diacetyl-1,3,7-triazaphosphabicyclo[3.3.1]nonane (DAPTA) and the homoleptic copper derivatives [Cu(PTA)₄]BF₄ (Porchia et al., 2009) and [Cu(DAPTA)₄]BF₄ (Gandin et al., 2015) were synthesized according to published methods. Elemental analyses were performed on a Carlo Erba 1106 Elemental Analyzer. ¹H, ³¹P and ¹³C NMR spectra were recorded on a Bruker AMX-300 instrument (300.13 MHz for ¹H, 75.47 MHz for ¹³C and 121.41 MHz for ³¹P), using SiMe₄ as internal reference (¹H and ¹³C) and 85% aqueous H₃PO₄ as external reference (³¹P). FT IR spectra were recorded on a Mattson 3030 Fourier transform spectrometer in the range 4000–400 cm⁻¹ in KBr pellets. Mass spectra have been recorded by an electrospray LCQ ThermoFinnigan mass spectrometer. The purities of compounds used for biological activity determination, were checked by elemental analysis and were found to be ≥95%.

4.2 Synthesis and characterization of copper(I) complexes

4.3 Experiments with cultured human cells

Copper complexes and the corresponding uncoordinated ligands were dissolved in deionized water. Cisplatin was dissolved in 0.9% sodium chloride solution. MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide), fluorogenic peptide proteasome substrates (N-Suc-Leu-Leu-Val-Tyr-7-amido-4-methylcoumarin (AMC), Boc-Gln-Ala-Arg-AMC, and Z-Leu-Leu-Glu-AMC), and cisplatin were obtained from Sigma Chemical Co, St. Louis, USA.