Nanoprobe XRF analysis reveals the intracellular distribution of organometallic rhenium complexes with different directing ligand

The intracellular localization of metal-based anticancer agents can provide important indications for their biological mode of action and intracellular targets. Here, we investigated the uptake, the intracellular distribution and cellular responses to three cytotoxic rhenium 3 complexes with different ligand environments in HeLa cells using cryogenic X-ray fluorescence (XRF) nanoanalysis combined with fluorescence microscopy. Despite their distinct ligand functionalities and expected targeting properties, all three Re complexes showed a similar intracellular distribution. The close agreement between fluorescence microscopy and XRF imaging of Re and Br supports the preservation of the spatial association of the complexes and their ligands following cellular uptake. Quantitative pixel-by-pixel XRF analysis at the single-cell level revealed elevated intracellular potassium levels after treatment, consistent with a global cellular stress response, but no local correlation between Re and K. In contrast, Zn showed a spatial correlation with Re, pointing to Zn-rich cellular structures as potential sites of interaction. The presented work characterizes the uptake of medicinally relevant metal-based drug candidates containing unnatural metals with different ligand spheres and correlates the data with the resulting endogeneous biometal levels and their localization. The observed Re–Zn association provides a basis for investigating potential Zn-associated targets of emerging metallodrugs.

The increasing number of cancer types (RKI, 2026) and the development of resistances against standard chemotherapeutic agents led to a situation that in several cases established platinum-based therapy has only a limited probability to cure the diseases (Kelland, 2007). Thus, new drugs are urgently needed for an improved treatment with higher success rates. Metal complexes in particular offer a largely untapped reservoir as novel lead structures for anti-cancer drugs. The biological activity of metal complexes may be linked to their intrinsic reactivity, either by ligand hydrolysis (Kelland, 2007) or even as catalytic drugs (Gasser et al., 2011; Nussbaumer et al., 2011). It may also be rooted in the photophysical properties of a particular metal-ligand combination e.g., for photodynamic therapy (Gasser et al., 2011). Finally, the three-dimensional arrangement of ligands especially around many transition metal complexes makes them well suited to fill “pockets” in biological targets (Cohen and Lippard, 2001; Meggers, 2007; Morrison et al., 2020) as has been shown impressively for kinase inhibitors (Meggers, 2009; Debreczeni et al., 2006; Dörr and Meggers, 2014; Meggers et al., 2007). In the area of metal-based drugs, binding to DNA and initiation of programmed cell death, or apoptosis, following futile attempts on DNA repair has been established as the major mode of action of the clinically used Pt anticancer drugs (Jamieson and Lippard, 1999; Wang and Lippard, 2005). However, more recently other cellular structures and modes of action have been identified as targets of metal-based drugs (Holtkamp and Hartinger, 2018; Boros et al., 2020) e.g., inhibition of mitochondrial enzymes leading to initiation of the intrinsic pathway to apoptosis (Köster et al., 2012), or even endoplasmatic reticulum (ER) stress, which often also leads to induction of a systemic immune response (ICD–immune-induced cellular death). In many cases a major challenge is to understand the exact mode of action of new metal-based drug candidates. The cellular uptake kinetics, the quantities accumulated in cells, and the intracellular distribution and localization of a given compound will provide a first clue about possible biological interaction partners and molecular targets of this compound (Karges and Metzler-Nolte, 2023).

If the metal compounds are autofluorescent or if they can selectively be stained, the intracellular distribution of the drugs can be analyzed by fluorescence microscopy. For complexes where this method is not applicable, or if a higher degree of quantification is desired, X-ray fluorescence (XRF) nanoprobe imaging is uniquely suited to quantify the uptake and the intracellular localization of metal complexes in cells. While nanoprobe-XRF is a powerful and established tool for the analysis of the elemental composition of materials, its application to imaging biologically active metal complexes inside cells under in situ conditions is still in its infancy. Among the challenges are sample treatment and fixation protocols and high throughput data collection and analysis. Spatially resolved XRF imaging has successfully been applied to study classical platinum-based cytotoxic drugs such as cisplatin (Shi et al., 2024; Bolitho et al., 2021; Hall et al., 2003). The uptake by platinum-resistant cancer cells was analyzed by XRF and reduced platinum levels compared to cells without resistance were found, indicating that the resistance seems to be connected to a reduced drug uptake (Shimura et al., 2005). Bimetallic gadolinium-platinum complexes were examined and their accumulation in the cell nucleus could be proven (Crossley et al., 2010). As another recent example, the distribution of an osmium complex in A2780 cells has been imaged (Sanchez-Cano et al., 2017). This Os complex targets the transcription of the mitochondrial chromosome, together with a fast cellular response involving a mitochondria-linked oxidative stress response pathway (NRF2), and a downregulation of key genes (Hearn et al., 2015). A different mechanism of action than the platinum complexes that attack DNA has been concluded, which offers the possibility to overcome cisplatin resistance. In line with this assumed mode of action, in XRF images of A2780 cells Os appeared to be strongly compartmentalized in small elliptical areas that were identified as mitochondria.

These examples demonstrate the tremendous potential of using XRF to help elucidate the mode of action of non-biometals. Among the metal complexes currently studied as alternatives to cytotoxic Pt-based compounds, rhenium (Re) complexes (Langdon-Jon et al., 2014; Balasingham et al., 2012; Amoroso et al., 2007) have received considerable attention, as they are hypothesized to reveal new modes of action. Several cytotoxic Re carbonyl complexes were already synthesized and described (Leonidova et al., 2014; Horiguchil et al., 1992; Gantsho et al., 2020; Leonidova and Gasser, 2014) and the analysis of the cellular uptake of Re compounds revealed their intracellular localization (Kelland, 2007; Zhang et al., 2012; Clède et al., 2016; Alderden et al., 2007; Hummer et al., 2013; Aitken et al., 2012; Waern et al., 2005; Niehoff et al., 2016; Levina et al., 2024; Bader et al., 2024). When the uptake and stability of two different Re(l) TRIP complexes were studied in HeLa cells, iodine could be used as marker for the ligands and observing a colocalization of Re and I is indivcative of the complexes remaining largely intact after uptake into the cell (Konkankit et al., 2020). This study used a [2 + 1] ligand approach with phenanthroline and isonitrile ligands. A different ligand set using pyridine-triazole, synthesized by Schibli’s “click-to-chelate” approach was used to install triphenylphosphonium groups on Re compounds (Mindt et al., 2006), intended for mitochondrial localization. An XRF microscopy study of these Re carbonyl complexes on A549 cells (PFA-fixed) indeed showed a spotted distribution that was suggested to be mitochondrial but also revealed how simple modifications of the triphenylphosphonium group strongly altered the cellular uptake. In addition to direct imaging and quantification of non-biometal complexes, XRF provides quantitative access to a range of naturally occurring biological metal ions like K +. XRF imaging on freeze-dried 22Rv1 cells investigated the uptake of Re(l) complexes and a disturbed homeostasis of zinc and potassium was confirmed (Wedding et al., 2017).

As described above, individual metal complexes show different intra-cellular localization which can be imaged by XRF-nanoscopy. On the other hand, there is no systematic study that uses the same metal complex core but directs it to different cellular compartments by strategically modifying its ligand set. In the present work we attach three different substituents to the Re(CO) 3 core, which have been reported to induce differential intracellular localization (see Figure 1). Sulfonamide derivatives were chosen which have been associated with cytosolic localization, while pyrene moieties in the ligand shell are expected to promote DNA targeting. In contrast, triphenylphosphonium groups are known to facilitate mitochondrial localization (Zielonka et al., 2017; Ross et al., 2008).

The pyridine-triazole ligand is used in the synthetic approach because of the proven stability of its Re complexes under biological conditions. Moreover, this ligand is readily and directly functionalized with the directing groups in a facile one-step Cu-assisted azide-alkyne cycloaddition reaction (Cu-AAC) (Mindt et al., 2006). Finally, the fact that Re1 and Re2 are fluorescent allows to apply a recently developed correlative imaging workflow using fluorescence microscopy and XRF nanoscopy. This precision-indexed approach allows to first investigate selected cells by fluorescence microscopy on SiN wavers, and then select and zoom in on the exact same cells on the XRF nanoscope under cryogenic conditions after rapid vitrification of the cells. In an effort to capitalize on the advanced cryogenic imaging facilities (REF) at the P06 Beamline (Schroer et al., 2010) at DESY we aimed to correlatively investigate the intracellular distribution of metal complexes with different ligand sphere and compared effects of metal ion distribution on a relatively large number of cells. The combination of a Rococo detection module (Rumancev et al., 2020) with high acceptance angle and an incoming photon flux of 5.7x10 9 Ph/s at 17 keV enables rapid measurement times down to 15 min per vitrified cell enabling the analysis of an ensemble of cells and a detailed statistical pixel-by-pixel analysis of metal content and colocalization. With this correlative approach the intracellular amounts and distribution of Re complexes Re1 – 3 in HeLa cells were analyzed. Re1 and Re2 complexes are fluorescent in the visible wavelength range, allowing a correlative assessment of their distribution in the cells by fluorescence microscopy. This served to verify if the fluorescence microscopy images correspond to the Re distribution in the XRF images. Re3, in turn, is not autofluorescent and XRF is the technique that allows to unravel its intracellular distribution.

Syntheses of Rhenium complexes

Pentacarbonylrhenium(I) bromide (1 eq.) and 2-pyridyl-1,2,3-triazole ligand (1 eq.) were dissolved in ethanol (7 mL) and the reaction mixture was refluxed at 78 °C for 24 h in the absence of light. The suspension was cooled to room temperature and solvent was reduced by half under vacuum. Rhenium complexes were precipitated as yellow solids by the addition of diethyl ether, collected by filtration and dried under vacuum. Further details of ligand and metal complex syntheses as well as characterizing data (including mass spectra, NMR spectra, and elemental analyses) are provided in the SI.

Cell culture and incubation with the Rhenium complexes

HeLa cell were seeded on ibidi 8-well microplates prepared with sterilized silicon nitride membranes in the wells with a concentration of 30,000 cells/250 μL cell culture medium. The cells were incubated for 24 h at 37 °C and 10% CO 2, before the compounds were added to reach a final volume of 350 µL per well (with DMSO concentrations of 1%). A final compound concentration of 10 µM was used for the incubation. Beforehand a range determination study was carried out revealing that at 10 µM compound concentration a suitable amount of living cells was left for imaging and a sufficient uptake for imaging of the compounds inside the cells occurred. After 45 min incubation time in the Re solution, the cells were washed with PBS (Dulbecco’s Phosphate-Buffered Saline) and covered with phenol red free medium for light microscopy imaging. The controls were prepared in the same manner, just without the addition of a metal complex.

Microscopy and cryo-fixation

The cryofixation of the cells after the incubation in the presence of the Rhenium complexes followed previously established protocols (Rumancev et al., 2020). Prior to cryofixation, the HeLa cells on the silicon nitride membranes were imaged using a Nikon Eclipse Ti fluorescence microscope. Immediately after microscopy, the silicon nitride membranes were blotted with cellulose paper and the samples were rapidly immersed in a liquid ethane/propane mixture at liquid nitrogen temperature (Kasas et al., 2003). After cryofixation the samples were stored in liquid nitrogen until the X-ray fluorescence measurement at the P06 beamine (Rumancev et al., 2020; Henke et al., 2005).

X-ray fluorescence measurement and data analysis

Nanoprobe-XRF measurements were carried out at the microprobe-branch of the Hard X-ray Micro/Nanoprobe Beamline P06, DESY (Hamburg) (Schroer et al., 2010). A previously described cryogenic vacuum chamber (Rumancev et al., 2020; Walter et al., 2014) was used for measurements in vacuum at a base pressure between 10 -6 mbar to 10 -8 mbar at a typical sample temperature of 120 K. The synchrotron beam (17 keV) was focused to a size of 440 nm × 500 nm providing a flux of 5.7x10 9 Ph/s. The entrance window into the cryogenic vacuum chamber was made of black Kapton with a thickness of 75 µm (Dupont, Hertfordshire, England). The fluorescence signal was detected by a Rococo 2 detector (PN detectors, München, Germany), which uses four monolithically integrated SDD sensor elements (15 mm 2 each) in an annular cloverleaf shape centered around a hole with a 1.8 mm diameter to allow transmission of the synchrotron beam. The detector was positioned in a backscatter geometry, 4 mm upstream of the sample (detected solid angle ≈0.9 Sr). The 2D area scans were measured in a continuous scan mode in the horizontal direction and after each horizontal line incrementally moved in the vertical direction to form a snake pattern. During measurements, the spectral data were collected within pre-defined distance of 400 nm in the horizontal and vertical direction using an Xspress 3 pulse processor (Quantum Detectors, Oxford). The position of the stages was determined by optical encoders and the intensity of the excitation X-ray beam was monitored using a photodiode and an ionization chamber. The spectra of the four active elements of the Rococo 2 detector were summed up and normalized to the intensity of the excitation beam. Subsequently, spectra were assigned to pixel positions on a Cartesian grid according to the recorded encoder positions. All spectra were analyzed by peak deconvolution and fitting using the fast XRF stacking function of the software PyMCA (European Synchrotron Radiation Facility, ESRF) (Solé et al., 2007). Spatial mass densities were calculated based on spectra obtained from a thin film XRF reference sample “RF17-200-S4218-41” (AXO Dresden, Germany). The thickness of the amorphous ice of the rapid plunge frozen HeLa cells were determined by comparing the intensity of the Si K emission from the silicon nitride substrate at the position of the cells to the Si K signal emitted by an empty silicon nitride membrane using Lambert-Beer’s law as described previously (Rumancev et al., 2020). As attenuation length of the Si - K emission with a photon energy of 1.74 keV in water λ = 11.6 µ m was used and we found that ice thicknesses were between 1 and 8 µm (Henke et al., 2005). For quantification of the XRF signal, the mass fractions were corrected for the emission energy depending attenuation through the ice layer. As self-absorption within the cell contributed, e.g., for K by less than 5%, we refrained from a self-absorption correction (Gräfenstein et al., 2023). For the correlation analysis of the average densities of two different elements in different cells, the average area densities for the two elements under consideration have been analyzed in all investigated cells using manual segmentation and plotted against each other. In addition, the intracellular correlation of two different elements within one single cell was obtained by determining the element density in each recorded pixel within a single cell for two elements and plotting the obtained area concentrations against each other. To determine if observed differences in metal levels were statistically significant, an ANOVA with post-hoc Tukey test was used (Origin Pro, α = 0.05).

Synthesis of the Re complexes and treatment of HeLa cells

For an easy synthetic access of suitably derivatized Re complexes a one-step cycloaddition reaction to assemble a complexing pyridine-pyrazole ligand was employed. Re complexes were prepared by refluxing Re(CO) 5 Br with the respective ligands for 24 h in EtOH. The ligands were prepared from 2-ethynyl-pyridine and the respective azides by Cu-catalyzed azide-alkyne cycloaddition reaction (Cu-AAC). The substituents on the ligands were chosen to facilitate differential intracellular localization, based on literature reports. Sulfonamide derivatives (Re1) have been reported to localize in the cytosol, while the pyrene derivative in Re2 should target DNA, and triphenylphosphonium groups (Re3) are expected to localize in mitochondria (Zielonka et al., 2017; Ross et al., 2008). All complexes were obtained in good yield around 75%, and fully characterized spectroscopically as well as by elemental analysis. NMR data and mass spectrometry confirm the proposed composition of Re1 - Re3 (see Supplementary Material). To analyze the uptake of the Re complexes, HeLa cells attached on silicon nitride membrane windows were incubated for 45 min with the complexes and subsequently quickly cryofixated. Frozen samples were stored in liquid N 2 and transported to DESY for the measurements.

Correlation of fluorescence microscopy and XRF nanoimaging

Vitrified HeLa cells were imaged under cryogenic conditions using XRF nanoprobe analysis to compare the metal distribution after treatment with the different Re complexes Re1 - Re3 (Figure 1). Untreated cells were used as controls. In total 13 replicates of the untreated cells were imaged by XRF, for Re1 29 cells were imaged, 14 cells for Re2, and 20 cells for Re3 (example images for the four cases are provided in the Supplementary Material, Supplementary Figure S16). The analysis comprised 76 cells in total, providing a dataset for comparison of the intracellular elemental distributions. In Figure 2, light microscopy (LM), light fluorescence imaging (LF) and XRF images of HeLa cells treated with three Re complexes Re1 – Re3 are additionally included for comparison. Slight deviations between LM/LF and XRF could be a result of subtle motions during the cryofixation step. In the LF images, complexes Re1 and Re2 can be visualized, and their intracellular localization can be mapped. As Re3 does not show any fluorescence in the visual wavelength range, its distribution can only be assessed by the XRF measurements. Comparing the phase contrast micrographs and the fluorescence microscopy images reveals that Re1 and Re2 accumulate predominantly in the cytosol and the vicinity of cell nuclei. The nucleoli seem to have much lower levels of the fluorescent complexes as compared to their surroundings.

The XRF images show the distribution of rhenium, and additionally the one of the physiologically relevant elements potassium and zinc. The potassium distribution throughout the cell seems to be rather homogeneous and potassium levels decrease towards the cellular periphery, reflecting the decreasing thickness of the cells. For compounds Re1 and Re2, an accumulation of the Re signal at rather distinct spots in the cells can be observed that correlates very well with the fluorescence microscopy images. For Re1 and Re2 complexes, the Re and the Zn distribution appear to be colocalized. As shown in the Supplementary Figure S21, Br was found at similar spots as Re, suggesting that Br is very likely to still remain in the ligand sphere after cellular uptake of the complex.

As pointed out above, in contrast to the Re1 and Re2 complexes, no fluorescence microscopy signal is visible for Re3 as this complex does not show fluorescence in the visible wavelength range. The XRF images show again that Re is localized at specific spots within the cells. In contrast to Re1 and Re2, after Re3 treatment zinc shows a rather homogeneous distribution throughout the cells without any noticeable localization at a specific part of the cell. Potassium is distributed throughout the cells, albeit not as homogeneous as in the case of Re1 and Re2. Interestingly, the overall potassium levels seem to vary between the five cells in the field of view.

From the XRF data, the average metal area concentrations within individual cells have been determined. The cell area was selected based on the XRF signal of requested element and comparison with light microscopy, using ImageJ. Each point in Figures 2b–d corresponds to the average within a single cell and the diagrams allow to compare the amount of Re accumulated from the 10 µM solutions of the three complexes and to compare effects on the intracellular potassium and zinc contents. Different Re levels (Figure 2b) were found for the three different complexes Re1-Re3. Re1 showed a relatively broad distribution of area concentrations between 14 and 106 ng/cm 2, with average contents of 38 ± 26 ng/cm 2, which was well above the one of the controls (0.234 ng/cm 2) (Table 1). For Re2, the metal levels spreaded less (between 25 and 47 ng/cm 2) with a slightly lower average area concentration of 34 ± 7 ng/cm 2. Treatment with Re3 showed again a relatively broad distribution of metal levels (between 0.3 and 26 ng/cm 2) (Table 1) with average values of 7 ± 8 ng/cm 2. The Re contents of all complexes (Re1, Re2, Re3) were found to be statistically significantly different (p < 0.05) relative to untreated cells. Thus, Re1, Re2, and Re3 were taken up to a similar extend, with a relatively broad distribution compared to the untreated cells, despite their chemically different ligand sphere.

The effect of the Re complexes on the potassium and zinc levels is shown in Figures 2c,d. The complexes Re1 and Re2 show increased potassium values of 668 ± 231 and 377 ± 123 ng/cm 2, respectively, compared to the 148 ± 35 ng/cm 2 of the controls (Figure 3b; Table 1). Re3 showed a broad distribution of potassium levels with average potassium area concentrations of 532 ± 620 ng/cm 2 (Table 1). Generally, we noted a higher variation of the potassium content in all three treatments compared to the controls. Potassium contents after treatment with Re1 – Re3 were significantly different compared to the untreated cells (p < 0.05). For zinc (Figure 2c), less differences than for the other two metals were detected. Increased zinc contents of 16 ± 2 ng/cm 2 were seen in Re2 treated cells compared to the 10 ± 3 ng/cm 2 of the untreated ones (Table 1). Re1 showed similar or slightly elevated zinc levels, on average 14 ± 6 ng/cm 2, compared to the untreated ones with a broader distribution than found for Re2. For Re3, Zn showed narrower distribution than Re1 and Re2 with area concentrations of 9 ± 9 ng/cm 2. Area concentrations of zinc after treatment with Re1, Re2, were significantly different compared to the untreated cells (p < 0.05). For Re3 the distribution of concentrations suggests the presence of at least two groups. While this grouping is pronounced for zinc, it is less obvious in the case of potassium. When grouping the cells according to the zinc distribution, the average concentrations could be separately analyzed and for the group with high zinc levels average zinc concentrations of 19 ± 4 ng/cm 2, average potassium concentrations of 1008 ± 667 ng/cm 2, and rhenium concentrations of 149 ± 0.5 ng/cm 2 were found, while for the cells with low zinc levels, zinc concentrations were as low as 1.5 ± 0.5 ng/cm 2, potassium 143 ± 8 ng/cm 2, and rhenium 0.9 ± 0.6 ng/cm 2.

Analysis of the intracellular colocalization of elements

The discussion of the quantitative data shown in Figure 3 and the images of the local element area concentrations within the cells already suggested that the intracellular distribution of some of the elements seemed to be colocalized (see also correlative color-coded representation in the Supplementary Figures S16-S18). Such a colocalization can be considered as an important indicator for the mode of action of the used cytotoxic complexes. For a detailed colocalization analysis, single cells were analyzed, and the intracellular content of different elements were compared at all analyzed intracellular positions. Thus, the analysis answers the question if throughout the analyzed cell certain metal levels are locally correlated. In Figures 3a,b, the average intracellular amounts of potassium and zinc are plotted correlatively against rhenium (K/Re and Zn/Re, data from Table 1) in all analyzed cells. In addition, panels c and d show the corresponding plots for the individual pixels within three single cells treated either with Re1, Re2, or Re3; the cells analyzed are shown Figure 3e (additional correlation plots are shown in S20). In general, a strong positive linear correlation is reflected by Pearson correlation coefficients approaching 1, while a deviation points towards uncorrelated data (Pearson coefficients much smaller than 1). For all complexes, Re1 Re2, and Re3 in Figure 3a, the data appears to be rather uncorrelated, which is also reflected in the low correlation coefficients of 0.08 (Re1), 0.25 (Re2) and 0.348 (Re3) (Supplementary Table S1). For the Zn/Re correlation in Figure 3b, cells with higher average Re area concentrations also tended to show higher average Zn area concentrations. Re1 shows a correlation coefficient of 0.88. Re2 is slightly weaker correlated with a coefficient of 0.52 and Re3 similar to Re2 with a coefficient of 0.48. This analysis shows that cells with a higher rhenium content seem to also show elevated zinc levels, while the local K levels showed only a weak association with Re. As in the analysis of the averaged intracellular area concentrations discussed above (e.g., Table 1), the rhenium treatment increased the intracellular potassium levels for all three compounds compared to the wild type cells, the weak correlation of the intracellular rhenium and potassium levels was unexpected.

While the above analysis answers the question if average intracellular rhenium uptake systematically affected the average intracellular zinc and potassium levels of single cells, the elemental area concentration distributions shown in Figure 2 suggest that in most cases rhenium seems to be very localized at specific positions inside the cells and, in some cases, even correlated with the localization of zinc. To analyze this in greater detail, Figure 3c and d show correlation plots of the metal levels in individual pixels of the elemental maps within single cells to answer the question if positions with high local rhenium content also correlate with elevated potassium or zinc levels. Correlation plots for K/Re (c) and Zn/Re (d) of one single cell that was treated with Re1, one single cell treated with Re2, and one single cell treated with Re3 were analyzed. The K/Re plots show that K levels barely change even at intracellular positions with high Re uptake for Re1 and Re2. In both cases, the correlation coefficients of 0.48 and 0.19, respectively, were very low (Supplementary Table S1). For Re3, the K/Re plots shows different potassium area concentrations at positions with similar rhenium levels, leading also to a low correlation coefficient of 0.40. Zn/Re correlation plots in panel d show a different picture as a higher correlation for Re1 (0.62) and Re2 (0.84) was found, which confirms the visual impression of the elemental distribution maps in Figure 2. The Zn/Re plots for Re3 show a slightly broader distribution with a smaller correlation coefficient of 0.55. Thus, the correlation analysis supports the visual impression that potassium levels seem to be rather homogeneous across the entire cell without any noticeable local colocalization with rhenium. Zinc in turn is stronger colocalized with rhenium and elevated zinc levels occur in positions within the cells where rhenium accumulates (Figure 3d). It is unclear at present whether the higher degree of Re/Zn co-localization is a consequence of rhenium treatment of the cells, or rather a coincidence in the sense that target for the rhenium compounds is located in areas of higher zinc content. It is clear, however, that there is no strong structure-activity relationship of this observation related to the structure of the rhenium complexes, since all three complexes studied herein show a rather similar pattern and correlation coefficients.

Previous work has demonstrated the potential of Re complexes as novel cytotoxic components. The fact that these compounds may overcome resistances towards established platinum-based drugs indicates a different mode of action which is currently remaining elusive. Tracking their intracellular localization may at least provide clues for their intracellular targets. So far correlative fluorescence microscopy and X-ray fluorescence nanoanalysis studies were limited to phenanthroline-derived compounds to prove the intracellular integrity of the Re complexes (Konkankit et al., 2020). Here we extend this work towards Re complexes with pyridine-triazole ligands, which are chemically more versatile and allow for an easy functionalization, e.g., with variation of the directing ligands. The present study uses a one-step click chemistry approach to install three different intracellular directing groups on the Re complexes and uses a statistical multicellular and even pixel-by-pixel correlation analysis to gain a deeper understanding of intracellular effects of the Re complexes.

The distribution of cytotoxic Re complexes in HeLa cells and its influence on the intracellular potassium and zinc levels were quantitatively analyzed using cryogenic X-ray fluorescence nanoanalysis. The cryogenic environment allowed to preserve the cells as close to their natural state as possible and resembles the gold standard for spatially resolved nanoanalytical experiments. Three different complexes were analyzed with different substituents on the Re (pyridine-triazole) core, that should lead to differential localization intracellularly: Sulfonamide derivatives (Re1) have been reported to localize in the cytosol, while the pyrene derivative in Re2 should target DNA, and triphenylphosphonium groups (as in Re3) are expected to localize in mitochondria (Zielonka et al., 2017; Ross et al., 2008).

For cells treated with the compounds Re1 and Re2 which show visible light fluorescence, a very good correlation of the fluorescence microscopy images and the rhenium and bromine distribution according to the XRF images were found. First of all, this excellent correlation is a strong indication of the structural integrity of compounds Re1 and Re2, even after incubation and cell uptake. This result also verifies that the used method and experimental protocol enables a marker-free intracellular localization of the rhenium complexes. An immediate application is the analysis of the intracellular distribution of compounds with triphenylphosphonium groups, a complex that does not show fluorescence in the visible wavelength range.

The rhenium intracellular distribution of the three complexes was in all cases similar. This finding is indeed rather surprising since the design principle with different substituents on the metal complex was chosen to induce a differential distribution intracellularly. One possible explanation is that the overall physicochemical properties of the complexes, including size, lipophilicity, charge, and interactions with cellular membranes and proteins, may dominate over the contribution of the individual targeting groups. Thus, the different ligand functionalities did not result in clearly distinguishable intracellular distributions under the conditions investigated. It thus appears that the presence of the same large, relatively lipophilic metal core in all three used compounds is decisive for the intracellular localization, and the presence of so-called “directing ligands” (such as pyrene or the triphenylphosphonium group) has only a minor influence. Also, it has been shown that large hydrophobic Re(I) complexes, such as those shown in Figure 1, are likely to enter cells by endocytosis and be initially localized in the lysosomes, which is the common cellular protection

n mechanism against exogenous toxins (Zhai and El Hiani, 2020). A perinuclear localization of Re, as shown in Figure 2, is consistent with such lysosomal uptake, which is also in line with the rather short incubation time of 45 min used in our experiments (Levina et al., 2024; Bader et al., 2024). Longer cell treatment times can lead to re-distribution of Re to other organelles, such as mitochondria (Levina et al., 2024).

Besides the intracellular distribution of the elements, changes in the potassium and zinc levels due to the rhenium treatment have been analyzed by nanoprobe X-ray fluorescence analysis. The average potassium content within the cells increased by 450% after the treatment with complex Re1 compared to untreated controls, by 370% for Re2, and with a relatively large variation by 350% for Re3. The level of zinc was also affected, although to a lesser degree, with increases of 40% for Re1, 60% for Re2, and decrease of 10% for Re3 compared to the untreated cells.

Potassium is an important physiological marker and changes in its intracellular concentration indicates malfunction of the potassium homeostasis and/or a stress response of cells (Gräfenstein et al., 2022). Changes in the potassium channel function have in the past been observed in cancer diseases (Liu et al., 2019). It was also reported that increased K + concentrations support the development of resistances in cancer. Also, several neurodegenerative diseases are associated with altered intracellular K + and Cl − concentrations (Ariano et al., 2005; Angulo et al., 2004). K + has also been identified to control the function of enzymes involved in the induction of apoptosis. Physiological K + concentration inhibits apoptosis inducing enzymes like caspase – 3 or endonuclease. Low potassium concertation in turn lead to an activation of caspase-3 or endonuclease which is the first step in apoptosis with an associated reduction in cell volume. We recently found that the overexpression of Huntingtin proteins in HeLa cells lead to increased potassium concetracion (Gräfenstein et al., 2022). Thus, the increased potassium levels are most likely a protective stress response of the cells, which we observed in this work to be a consequence of the presence of the cytotoxic rhenium complexes. The effect of the three Re complexes on the intracellular potassium levels was different. Re1 led to an increase of intracellular potassium content to 450% of untreated cells. Re2 showed an increase of 250% and Re3 an increase of 350%. Thus, the change in potassium level in the cells depended on the ligand sphere of the Re complexes and might suggest a different propensity to activate the protective stress responses of the cells.

The elemental distribution maps also suggested a colocalization of the rhenium with zinc. Potassium, in turn, was homogeneously distributed throughout the cell. While the average potassium and zinc levels were significantly enhanced in the total cell population, the total rhenium level in single cells was not correlated with the potassium level. This leads to the interpretation that a certain rhenium threshold is enough to trigger a stress response, but more rhenium does not additionally enhance this effect. The elemental distribution maps also suggested a colocalization of rhenium with zinc. Potassium, in turn, was homogeneously distributed throughout the cell. While the average potassium and zinc levels were enhanced in the total cell population, the total rhenium level in single cells was not correlated with the potassium level. This indicates that the cellular increase in potassium is not simply proportional to the total intracellular Re content. In contrast, zinc levels were strongly correlated with rhenium levels, suggesting that Re tends to be spatially associated with Zn-rich cellular regions. However, XRF colocalization can only provide evidence, not a final molecular proof for the direct interactions of the Re complexes with Zn-containing proteins.

This observation holds true even if single cells were analyzed pixel-by-pixel. In agreement with the visual impression of the XRF maps, local intracellular potassium contents were not correlated with the ones of rhenium. Enhanced intracellular levels of zinc, in turn, were for all three compounds colocalized with rhenium (Figure 3d). Zn-containing proteins, including zinc-finger proteins, represent one possible class of structures that could contribute to this association; however, their involvement cannot be established from the present XRF data. Zinc finger peptides are mainly involved in DNA transcription signal transduction, DNA repair and cell migration (Kamaliyan and Clarke, 2024; Cassandri et al., 2017). Furthermore, these proteins are involved in cancer development and the formation of metastasis (Bendellaa et al., 2024). The zinc finger family includes both tumor suppressor genes and oncogenes (Cassandri et al., 2017; Jen and Wang, 2016). The observed Re/Zn colocalization may indicate an association of the Re complexes with Zn-rich cellular structures, including Zn-containing proteins such as zinc-finger proteins (Ye et al., 2020; Abbehausen, 2019), suppresses it leading to cell death, or Re complexes are bound to DNA at the same site, leading to disruption of replication. Also, DNA repair mechanisms may be disrupted which can also lead to cell death. The zinc levels seemed to be slightly increased after Re1 and Re2 treatment compared to the one in untreated cells. This may be due to the blocking of zinc finger proteins by Re complexes and the cell trying to compensate for the deficiency by forming new zinc peptides, requiring further uptake of zinc. Our mechanistic suggestions are solely based on the correlation analysis in the XRF data and in future studies it would be interesting to explore further experimental evidence if such interactions occur on the molecular scale.

In this investigation, the distribution of Re complexes with different ligand structure in HeLa cells and its influence on the intracellular potassium and zinc distributions was investigated. The cells were cryofixed, no chemical fixation was used, which allowed to study the element distribution as close as possible to the natural biologically relevant state. Correlations could be made between fluorescence microscopy and XRF images, providing evidence for the structural integrity of the complexes and establishing the basis for intracellular localization analysis. Despite the different ligands, the intracellular rhenium distribution was largely similar for the different complexes. It appears that the presence of the same large, relatively lipophilic metal core in all three compounds dominates their intracellular localization, and the presence of so-called “directing ligands” (such as pyrene or the triphenylphosphonium group) had only a minor influence. Meaningful correlations could be detected between rhenium and the biometals potassium and zinc. Rhenium and potassium, on the other hand, appear not to be correlated. Both, the concentration analysis on the cellular level and the statistical multi-cellular and pixel-by-pixel analyses provided important insight into the distribution of Re based complexes and suggests the future extension of such studies to further biochemical and mechanistic studies of new metallodrug candidates to complement biochemical and cell biology work to understand their mode of action.