Over the last decades, many efforts have been devoted towards the development of novel therapeutic approaches to treat cancer due to the significant number of cases and deaths worldwide. From all types of cancer, breast cancer ranks as the most frequently diagnosed type of cancer among women (2.25 million new cases) and as the first cause of cancer-related death in women (0.68 million) in 2020 [1]. Currently, the most effective options to treat breast cancer consist in a combination of surgery with radiation and medication (chemo, hormonal and/or targeted therapies). In the field of chemotherapy, the discovery of the antitumoral properties of cisplatin in the late 1960s was a landmark [2]. So far, cisplatin and two other platinum(II) drugs (carboplatin and oxaliplatin) are the only metal-based drugs approved worldwide to treat cancer [3]. However, despite their success in the clinic, platinum-based drugs often experience some limitations such severe side effects, limited spectra of action and, importantly, inherent or acquired resistance [4]. In order to tackle these drawbacks and explore the scope of action of other metal-based drugs, many researchers have pursuit for newer therapies, more effective and selective, to improve patient survival rates and quality of life of humans, especially in platinum-resistance tumors. In this setting, ruthenium-based compounds emerged as leading players showing considerable potential to be used as alternative metallodrugs for cancer treatment [[5], [6], [7]].

We have been involved in the development of new ruthenium compounds with the general formula [Ru(η5-Cp)(PPh3)(bipy)]+ (Cp = cyclopentadienyl; bipy = 2,2′-bipyiridine derivatives) as anticancer agents [8,9]. Up to now, the results have been very promising in terms of cytotoxic activity against several human cancer cells lines with different degrees of aggressiveness and resistance. Yet, none of these compounds has been evaluated regarding their plasmatic and enzymatic stability. Besides, a comprehensive knowledge of the metabolic fate of drug candidates is essential to mitigate potential structural liabilities and guide structural optimization [10], thereby avoiding late-stage drug attrition due to toxicity problems [11,12]. Therefore, for the present work we selected two lead compounds, namely LCR134 and PMC79 (Fig. 1). LCR134 emerged as a potential agent against breast cancers (tested in MDA-MB-231 and MCF-7 cells). Yet, the most surprising feature about this compound regards its ability to act as P-gp inhibitor (both in vitro and in vivo in the zebrafish model) [13,14]. PMC79 (the parental compound of LCR134) also showed a good cytotoxicity against both breast cancer cells and proteomic studies revealed that its targets are the proteins involved in the cellular cytoskeleton dynamics [15]. Nonetheless, the most striking characteristic of this compound concerns its selectivity against colorectal cancers with specific mutations [16]. In addition, we also considered for the present study a family of compounds with substituents at the η5-cyclopentadienyl ligand (Fig. 1) [17,18]. Some of these compounds have shown unprecedented collateral sensitivity for cisplatin resistant lung cancer, i.e., the compounds are only cytotoxic in the resistant cells (vs. the sensitive cells) by impairing the correct functioning of ABC pumps [18]. We would like to understand if these compounds are active in breast cancer cell lines and select two lead structures from this sub-family to be also subjected to stability studies.

Studying the antitumor potential of prospective anticancer metallodrugs, evaluated by the half maximal inhibitory concentrations (IC50), is crucial for understanding their pharmacological and biological characteristics. For this purpose, a colorimetric MTT assay, based on metabolic active cells, was performed to assess the cytotoxic effects of the organometallic complexes under study against two genetically different breast carcinoma cell lines: MDA-MB-231 and MCF-7, at 48 h treatment (Table 1; for IC50 curves cf. Supplementary Material Fig. S1). MCF-7 is an estrogen receptor positive (ER+), poorly aggressive and non-invasive cell line with low metastatic potential [19]. These cells are commonly used as a model for hormone-dependent breast cancers that are responsive to endocrine therapy. On the contrary, MDA-MB-231 is a highly aggressive, invasive and poorly differentiated cell line [20]. This cell line is used as a model of Triple Negative Breast Cancer (TNBC), as they lack estrogen receptor, progesterone receptor and human epidermal growth factor receptor 2 (HER2).

In general, all complexes showed cytotoxic activity against both cancer cell lines tested. Moreover, all were found to be more cytotoxic against the MCF-7 cells, which may be related to the different genetic background of these cells. The cytotoxicity of the ligands was also assessed in MDA-MB-231 cells. The IC50 values found were much higher than their corresponding complexes, which indicate they are not active except for Me2bipy that shows moderate activity, leading to the most cytotoxic compounds (RT150 and RT151).

Complexes RT111, RT118, RT150 and RT151 presented higher cytotoxic activity against both cell lines. This finding reveals that the functionalization of the cyclopentadienyl ligand (η5-CpR; R = CH2OH or CHO), in the presence of the same bipyridine, does not significantly change the compound cytotoxicity. LCR134 and PMC79 also showed good cytotoxic activity against both breast cancer cells, as expected from previous studies [13,15,22], but LCR134 was ∼2-fold less cytotoxic than its parental compound PMC79. Complexes RT130, RT131 and RT152 showed only a moderate cytotoxicity against the breast cancer cells, suggesting that the presence of the CH2OH and CH2Biotin groups at the bipyridine ligand hampers the compound cytotoxic activity, whereas the presence of a H or a methyl substituent in the bipyridine co-ligand is highly beneficial against both cell lines. In particular, RT150 and RT151 exhibited the lowest IC50 values among these compounds, especially in the MDA-MB-231 cells, suggesting that the presence of the methyl substituent in the bipyridine ligand is the most advantageous configuration.

To evaluate complexes ̓ selectivity for cancer cells, normal fibroblasts (V79) were also included in this study. The IC50 values obtained at 48 h treatment (Table 1) indicate that all complexes could be considered prospective anticancer drug to be further explored since in all cases a favorable selectivity index >3 was found.

The lipophilicity of all compounds was determined by measuring the extent to which they partitioned between n-octanol and water by the shake-flask method. All complexes displayed a preferential partition in n-octanol over water, as shown by their positive logPo/w values (Table 2). Complexes RT150 and RT151 (bearing the Me2bipy) showed the highest partition in n-octanol, followed by LCR134 and pmc79 (bearing biotinylated- and hydroxymethylated-bipyridine, respectively). On the other hand, RT130 and RT131 are the most hydrophilic compounds of the series and appeared to be significantly more hydrophilic than its relative PMC79, which is probably related to the presence of formyl and hydroxymethyl groups at the cyclopentadienyl moiety. Overall, the most hydrophobic compounds displayed the best cytotoxic activity in breast cancer cells as already observed for other ruthenium-based compounds [23].

One of the major problems regarding the lack of success of some chemotherapeutic treatments relies on the ability of cells to maintain their malignant potential even after several cycles of chemotherapy, leading to cancer recurrence. The colony formation assay determines cells' ability to survive a short exposure to a cytotoxic compound, and to proliferate indefinitely after that agent is removed, thereby retaining its reproductive ability to form a colony [24]. This assay allows an in vitro simulation of what happens during cycles of chemotherapy, and thus is used to evaluate potential chemotherapeutic agents.

To evaluate the effect of ruthenium complexes on cellular clonogenic ability (number of colonies), we treated the MDA-MB-231 breast cancer cells with all the ruthenium-based compounds. This cell line was selected because it is a good model to study compounds with antimetastatic properties. In particular, a recent paper reported that RAPTA-T ([Ru(η6-toluene)Cl2(PTA)]) (PTA = 1,3,5-triaza-7-phosphaadamantane) has selective antimetastatic properties on invasive cell lines, and thus they are more evident on MDA-MB-231 breast cancer cells than on non-invasive MCF7 breast cancer cells [25]. MDA-MB-231 cells were exposed to ¼ IC50 and IC50 concentrations of these compounds for 48 h, after which the cell culture media was removed, and cells were kept in culture for 6–7 days. Our results (Fig. 2) show that the clonogenic ability of all compounds tested is proportional to their concentration. Therefore, when present at the IC50 concentration, all compounds were able to significantly inhibit cells from forming colonies at a higher extent. Remarkably, all compounds, with the exception of RT131 and LCR134, completely inhibited colony formation at the IC50 concentration. Additionally, compound RT111 was also able to reduce cell clonogenic ability when exposed to ¼ of its IC50 concentration.

In order to select the most promising compounds among those with a functionalized cyclopentadienyl ligand to be further studied, both the cytotoxicity activity and clonogenic ability were considered. In that frame, compounds RT150 and RT151 revealed the best results in terms of both IC50 values as well as by completely inhibiting the formation of colonies at the IC50 concentration. These results follow the same trend previously observed for non-small cell lung [18] and colorectal [17] cancer cells, revealing that these compounds are highly cytotoxic against a diverse range of cell types.

The cellular distribution of the selected complexes, measured as the ruthenium accumulation by inductively coupled plasma mass spectrometry (ICP-MS) in the different cellular compartments (cytoskeleton, cytosol, membrane and nucleus) was studied after the incubation of MDA-MB-231 and MCF7 cells to complexes RT150, RT151 and PMC79 for 48 h at a concentration equivalent to their IC50. Results are shown in Fig. 3.

As can be observed from our results, all compounds are retained mainly at the membrane of both cancer cell lines (>86% for MDA-MB-231 and > 79% for MCF7). These findings agree with previous results obtained for the intracellular distribution of LCR134 in both breast cancer cell lines, as well as compound PMC79 in MCF7 cells after incubation for 24 h [26,27]. Regarding ruthenium accumulation on other cellular fractions besides the membrane, results showed slightly differences depending on the cell line used: the nucleus (4%–9%) was the second cellular compartment with higher accumulation of all compounds in MDA-MB-231 cells, while for MCF7 cells it was the cytosol (4%–10%).

The cell death mechanism was assessed through the Annexin V/Propidium Iodide (AV/PI) cytometry-based assay. This experiment is commonly used to distinguish between cell death by apoptosis and necrosis, through differences found in the integrity and permeability of the plasma membrane [28]. Viable cells are negative for both probes (AV-/PI-), early apoptotic cells are positive for annexin V only (AV+/PI-), necrotic cells are positive for propidium iodide only (AV-/PI+) and late apoptotic cells are positive for both probes (AV+/PI+). To determine the preferred cell death mechanism, MDA-MB-231 cells were incubated with compounds RT150, RT151, PMC79 and LCR134 at IC50 concentrations for 48 h, followed by staining with AV/PI. The results obtained (Table 3) showed that all tested compounds induce cell death by apoptosis instead of necrosis. In particular, the majority of the cells died by early apoptosis.

We conducted a comprehensive investigation on the metabolic stability and metabolic profile of RT150, RT151, PMC79, and LCR134 using human plasma and human liver microsome (HLM) incubations, by liquid chromatography coupled with tandem high resolution mass spectrometry (LC-ESI(+)-HRMS/MS). Additional incubations were performed in alamethicin-induced HLM [29], in the presence of the Phase II co-factor UDPGA, and in S9 liver fractions, to identify Phase II glucuronides and to explore the potential involvement of cytosolic metabolizing enzymes in the metabolic degradation of selected compounds, respectively.

PMC79 and RT151 exhibited a remarkable stability against plasmatic and liver microsomal enzymes (cf. Supplementary Material Fig. S3). However, in sharp contrast, LCR134 exhibited half-lives of 15 and 12 min in plasma and HLM incubations, respectively (cf. Supplementary Material Fig. S4), demonstrating to be a high clearance compound (t1/2 < 20 min) [30]. This fast depletion was accompanied by the concomitant increased formation of PMC79 (Fig. 4), plausibly as a product of esterase/carboxylase/hydrolase [26,31] activity, with loss of the LCR134 biotin-containing moiety [14]. Despite previous studies conducted in zebrafish models suggested that this moiety was not cleaved, our results clearly show that to preserve its structural integrity until reaching the biological target, careful consideration of adequate drug delivery strategies is necessary.

Whereas PMC79 showed minimal metabolic degradation, one Phase I metabolite, PMC79 M1, resulting from hydroxylation of the bipyridine moiety (Fig. 5) was identified in HLM incubations. Additionally, PMC79 M2, stemming from the glucuronic acid conjugation of the parent drug, constituted the Phase II metabolite identified in alamethicin-induced HLM incubations. Similar depletion and metabolite profiles were obtained in human liver S9 fraction incubations, thereby suggesting that PMC79 is not prone to degradation by cytosolic metabolic enzymes.

More puzzling, however, was the depletion observed for RT150 during plasmatic stability studies, as no identifiable degradation metabolites were found (cf. Supplementary Material Fig. S5). Likewise, and although two Phase I metabolites were identified in HLM incubations, namely RT150 M1 (major), resulting from pyridine moiety hydroxylation, and the carboxylic acid RT150 M2 (minor) stemming from the Cp substituent oxidation (Fig. 6 A and B), their relative intensities over the incubation time were remarkably low compared to the depletion percentage of RT150 in the incubations. In either instance the percentage of RT150 remaining in incubation, following 180 min, was of approximately 60% (cf. Supplementary Material Fig. S5). These intriguing findings suggest that alternative mechanisms might underlie the parent drug's depletion from the incubations. Given the presence of an aldehyde moiety in the RT150 structure, we formulated a hypothesis that covalent binding could occur with the lysine residues of proteins, upon Schiff Base formation (Fig. 7). Thus, this mechanism could provide a plausible explanation for the observed depletion [32]. To investigate this hypothesis, we conducted incubation experiments with lysine and RT150, both in the presence and absence of cyanoborohydride as reducing agent.

As anticipated, RT150-Lys Schiff Base and RT150-Lys Amine were identified, by LC-ESI(+)-HRMS/MS, in the reactions performed in the absence and under reductive conditions, respectively (cf. Supplementary material Fig. S6). Coherently, RT150-Lys Amine was also identified following pronase E digestion (to amino acids) of Human Serum Albumin (HSA), which was preincubated with RT150 in the presence of cyanoborohydride. These results substantiate our initial hypothesis that RT150 depletion from plasmatic/HLM incubations is due to the covalent binding to proteins. While the covalent modification of plasma proteins has anticipated implications for RT150 bioavailability, its adduction to liver proteins might be a toxicologically relevant event. In fact, the covalent modification of proteins by aldehydes [[33], [34], [35], [36]], negatively impacts proteins function and is associated with the inception of immune-mediated drug-adverse reactions [37]. However, the covalent modification of proteins can also constitute a pharmacologically relevant event for its target activity [38]. Therefore, RT150 toxicity assays should be prioritized to assess risk/benefit relationships before further improvements of this drug candidate and the role of covalent binding to the anticancer activity should be assessed.

Even though minimal degradation was observed for RT151 in HLM incubations (Fig. 6 C), two metabolic pathways were identified: 1) oxidation of the Cp hydroxymethyl substituent to aldehyde, thereby yielding RT150, which is subsequently hydroxylated to its major metabolite RT150 M1; and 2) hydroxylation of bipyridine moiety to RT151 M1, which is subsequently oxidized to RT151 M2, presumably involving RT150 M1 intermediate formation. While no Phase II metabolite was identified for RT150, two RT151-derived glucuronide metabolites were identified: 1) RT151 M3, stemming from the glucuronic conjugation of the parent drug; and 2) RT151 M3, formed upon glucuronidation of its Phase I metabolite RT151 M1. These results suggest that RT151 is prone to glucuronidation [39], and although to a minimal extent, it undergoes metabolic conversion into RT150. Due to their structural features RT150 and RT151 might also be targets for the cytosolic enzymes, aldehyde dehydrogenases, and alcohol dehydrogenases [40]. However, similar depletion profiles were obtained, in S9 liver fraction and HLM incubations.