Elemental mapping of half-sandwich azopyridine osmium arene complexes in cancer cells †

Transition metal complexes are often prodrugs which undergo activation by ligand exchange and redox reactions before they interact with target sites. It is therefore important to understand the roles of both the metal and the ligands in their activation, especially in cells. Here we use a combination of synchrotron nanoprobe X-ray ﬂ uorescence (XRF) from Os L 3 M 5 and Br KL 3 emissions and inductively coupled plasma-mass spectrometry (ICP-MS) detection of 189 Os, 79 Br, and 127 I, to investigate the time-dependent accumulation and localization of osmium as well as the monodentate ligand and the chelated phenylazo-pyridine in A2780 human ovarian cancer cells treated with the potent anticancer complexes [Os( η 6 - p - cymene)(4-R 2 -phenyl-azopyridine-5-R 1 )X]PF 6 , with R 2 = NMe 2 or OH, R 1 = H or Br, and X = Cl or I. The data con ﬁ rm that the relatively inert iodido complexes are activated rapidly in cancer cells by release of the iodido ligand, probably initiated by attack by the intracellular tripeptide glutathione ( γ - L -Glu-l-Cys- Gly) on the azo double bond. The bond between osmium and the azopyridine appears to remain stable in cells for ca. 24 h, although some release of the chelated ligand is observed. Interestingly, the complexes seem to be degraded more rapidly in normal human cells, perhaps providing a possible mechanism for selective cytotoxicity towards cancer cells.

Recently, we have reported promising Os(II) arene and Ir(III) cyclopentadienyl half-sandwich anticancer complexes containing azopyridine ligands, which are activated in cancer cells and modulate their metabolism through redox processes.In particular, Os(II) complexes in the family [Os(η 6 -p-cymene) (4-R 2 -phenylazopy-5-R 1 )X]PF 6 (where Azpy = p-( phenylazo)pyridine, R 2 = NMe 2 or OH, R 1 = H or Br, and X = Cl or I; Table 1) can be up to 49 times symbol more potent than cisplatin in a panel of over 800-cancer cell lines, including platinum-resistant cancer cells. 1,2Moreover, they exhibit lower in vivo toxicity against zebrafish embryos, 13 and act through a different mechanism of action compared to cisplatin.4][15][16][17] The combination of time-resolved X-ray absorption spectroscopy (XAS) and DFT theoretical approaches has shown that the activation of this family of drug candidates involves a catalytic ligand-mediated mechanism. 18The azo-bond (NvN) in the chelated azopyridine reacts directly with GSH, promoting the dissociation of the monodentate iodide from osmium.The same reaction can occur when other monodentate ligands such as Cl, H 2 O, OH − or GS − are coordinated to Os instead of iodide.As such, the nature and chemical properties of the ligands coordinated to the Os are important for the rate of activation of the complexes, as they control the stability of the Os-X bond and overall reactivity of the azo-bond towards GSH.Moreover, this also affects their biological properties. 1,2,13,14,19or example, the Os-Cl bonds are more labile than Os-I bonds, and complexes containing chlorido ligands show slower cellular accumulation 19 and are generally significantly less active as anticancer agents than their iodido analogues; 1-PF 6 is ca.12× less potent than 2-PF 6 against A2780 ovarian cancer cells; IC 50 = 1.8 and 0.15 μM, respectively; Tables 1  and 2. 1,2,13,19 Similarly, complexes with phenylazopyridine ligands with R 1 = OEt and R 2 = H undergo hydrolysis of their Os-X bonds (X = Cl or I) in presence of GSH nearly 3× faster than those with R 1 = H and R 2 = NMe 2 (1-PF 6 and 2-PF 6 , Table 1). 15Complexes with both these types of R 1 substituents are activated at a similar rate inside cells, but complexes with R 1 = OEt are at least 6× less active than R 1 = NMe 2 complexes. 15uch differences in activity might be caused by drug deactivation before reaching the intracellular target site, due to an increased lability of the Os-X bond (Os-Cl vs. Os-I) and reactivity of the azopyridine ligand.
X-ray Fluorescence (XRF) has allowed studies with subcellular resolution of the localisation of this type of osmium complex in chemically-fixed A2780 human ovarian cancer cells. 17,20This was achieved by mapping Os L 3 M 5 emissions, and showed apparent mitochondrial localisation of the complexes after 24 h treatment. 17Equally, in-cell XAS studies using nanofocused synchrotron radiation have indicated the possible presence of an Os(II)/Os(III) redox cycle in those areas of A2780 cells with high concentration of the complex. 21Such redox reactions may be involved in the generation of the observed ROS. 21Studies are now needed to investigate the importance of the halide and phenylazopyridine ligands coordinated to the Os in the cellular mechanism of action of these complexes.The lability of the Os-X ligand for example, may affect the processes for cellular localisation of these complexes, and their biological activity.Our previous studies suggest that the azopyridine ligand mediates the intracellular activation of the complexes. 15,18Although the redox properties of the phenylazopyridine could also have a role in the ROS production observed in treated cells, it is not known if this chelated ligand remains bound to Os after activation of the complex inside cells.Hence, probing the intracellular stability of these complexes can contribute to our understanding of their detailed cellular mechanism of action.
Recently we reported the tracking of osmium diamine transfer hydrogenation catalysts in cancer cells by monitoring osmium and a bromine substituent on a chelated ligand using XRF and ICP-MS, which provided insight into dissociation of the chelated ligand inside cells. 22Here we have used a similar combination of ICP-MS, for monitoring 189 Os, 79 Br, 127 I, and XRF detecting Os L 3 M 5 and Br KL 3 emissions at the I14 nanoprobe beamline (Diamond Light Source) to track and map the time-dependent localisation and ligand dissociation reactions of potent anticancer complexes 1-4 (Table 1) in human ovarian cancer cells and normal human cells.We have monitored osmium, the iodide monodentate ligand and a Br substituent on the chelated phenylazopyridine ligand, in a position that has little effect on activity.This has enabled us to gain new insights into the activation of these complexes in cells, the extent to which the chelated phenylazopyridine ligand remain bound to Os, and the difference in behaviour in normal cells compared to cancer cells.
XRF mapping.XRF maps were acquired with incident beam energies of 12 keV (1-PF 6 and 2-PF 6 ) or 15 keV (4-PF 6 ), using a 100 nm step size and 0.1 s exposure, at the I14 beamline (Diamond Light Source, UK).XRF data were collected by a 4-element silicon drift detector (SGX-RaySpec, UK) laid out in backscatter geometry, covering 0.6-0.8sr solid angle and capable of 1.5 Mcps/channel.Data fitting and analysis were performed using PyMCa software developed by the ESRF. 24he fitted data were analysed in ImageJ software to gain information on cell size, morphology, elemental distribution and co-localisation statistics.

General cellular accumulation protocol
10 or 20 mL of a suspension of 4 × 10 6 cells per mL was seeded in 100 or 145 mm cell culture Petri dishes (respectively) for 24 h (310 K, 5% CO 2 ).Prior to cell treatment, stock solutions of osmium compounds were prepared in 5% v/v DMSO and 95% v/v DMEM, and their Os concentrations determined by ICP-OES (see ESI †).Cells were treated under varying conditions (temperature, concentration, time) as specified below.Then, they were washed with PBS and 0.25% trypsin/EDTA (2 mL) was added to each Petri dish (5 min, 310 K, 5% CO 2 ), which was quenched with a known volume of DMEM to form a single cell suspension.Cells were counted in duplicate (2 × 10 µL) and pelleted by centrifugation (1000 rpm, 5 min, 298 K).The supernatant was removed, pellets resuspended in PBS (1 mL) for further centrifugation (1200 rpm, 5 min), and the new pellets were analysed by ICP-MS.
189 Os ICP-MS cellular accumulation studies.Cell pellets were digested in 72% v/v nitric acid (200 µL) for ca. 12 h (353 K) and then diluted to 3.6% v/v nitric acid using a solution containing 10 mM thiourea and 10 mg mL −1 L-ascorbic acid.
(i) Time: A2780 cells were treated with 1× IC 50 of 4-PF 6 (0.42 ± 0.03 µM) for 4, 8 18 or 24 h (without further recovery time), or 24 h followed by a period of 24, 48 or 72 h recovery in drugfree DMEM.Every two Petri dishes were combined, and cells counted from the combined solutions to form larger pellets for reliable bromine detection by ICP-MS.Cell pellets were digested using the alkaline digestion method described above.

Time-dependent localisation of complexes carrying various halogen ligands in cancer cells
We used nano-focused XRF to probe the effect of different halogen ligands on the time-dependent accumulation and localisation of half-sandwich Os(II) azopyridine complexes in cancer cells.A2780 ovarian cancer cells were treated for various times with equipotent concentrations (7× IC 50 ) of complexes with almost identical structures, but containing Cl (1-PF 6 ; 7× IC 50 : 12 µM; treated for 4 or 8 h) or I (2-PF 6 ; 7× IC 50 : 1 µM; treated for 4, 8 or 24 h) as a monodentate coordinated ligand, Table 1.Then, maps based on Os L 3 M 5 emissions were acquired with a 100 × 100 nm 2 resolution from cryo-fixed, dehydrated cells (Fig. 1; ESI, Fig. S1-14 †).This Os distribution was compared with the localisation of native elements such as P or Zn, which were used to define the cellular limits within the maps (e.g.][27] As expected, Os L 3 M 5 emissions were not detected in untreated A2780 cells (ESI, Fig. S1 †).Control samples (Fig. 1a; ESI, Fig. S2-4 †) also exhibited clear cell outlines with rounded shapes (mean roundness factor, RF: 0.77 ± 0.14; where RF = 1 implies perfect circularity), large cell nuclei, and overall area ranged between 215-439 µm 2 (mean area = 303 ± 93 µm 2 ).All of this correlated well with the rounded morphologies and nuclei, [28][29][30] and typical size (<20 µm in diameter; expected area around 314 µm 2 ) 31 reported for this cell line.Moreover, observation of the nuclei of the control cells mapped (shown by the concentrated regions of Zn), 32 suggested that the wide variation between their areas (ESI, Table S1 †) could be explained by their different stages of the cell cycle.This includes cells in the interphase (ESI, Fig. S2 †), but also in the anaphase (ESI, Fig. S3 †) and telophase/early cytokinesis (where the two daughter cells are separated; ESI, Fig. S4 †) of mitosis.
Treatment of A2780 cancer cells with 7× IC 50 1-PF 6 or 2-PF 6 for 4-8 h did not cause significant alterations to the size (mean area = 327 ± 22, 396 ± 55, 403 ± 275 and 403 ± 275 µm 2 ; ESI, Fig. S5-12, Table S1 †) or morphology (mean RF: 0.79 ± 0.04, 0.83 ± 0.04, 0.82 ± 0.05 and 0.84 ± 0.10; ESI, Table S1 †) of A2780 cells when compared with untreated controls (t-tests p > 0.05).However, cells treated with iodido complex 2-PF 6 for 24 h were significantly larger (mean area = 1156 ± 361 μm 2 ), and presented substantial cellular damage (Fig. 1f; ESI, Fig. S13-14 †).Cellular alterations observed included membrane blebbing and the collapse of nuclear integrity, suggesting the initiation of programmed cell death pathways, as previously reported from XRF studies using chemically-fixed A2780 cells. 17Equally, Os maps acquired from cells treated with the complexes confirmed previous ICP-MS studies, and showed that the accumulation of both 1-PF 6 and 2-PF 6 into A2780 cells was time-dependent and slower for the chlorido complex 1-PF 6 . 19As such, 2-PF 6 was always found at higher levels inside cells than its chlorido analogue 1-PF 6 (Fig. 1).Unfortunately, due to synchrotron beamline time-constraints the experimental setup could not be calibrated to an AXO GmbH standard, thus quantities of osmium ( pg mm −2 ) could not be determined.Still, the XRF maps allowed it to be determined that cells treated with 1-PF 6 for 4 h contained negligible quantities of Os (Fig. 1b; ESI, Fig. S5-6 †), and only small amounts of Os from the same complex were found after 8 h treatment (Fig. 1b; ESI, Fig. S7-8 †).A similar trend was observed for 2-PF 6 , for which osmium accumulation was more pronounced after 8 h compared to 4 h (Fig. 1d and e; ESI, Fig. S9-12 †).Yet, there was a decrease in the amounts of intracellular Os after 24 h treatment (Fig. 1f; ESI, Fig. S13-14 †), probably caused by the loss of cellular integrity due to the initiation of cell death processes.
XRF maps also indicated cytosolic localisation for both 1-PF 6 and 2-PF 6 in cells, with no statistical correlation between Os and Zn (as determined by Pearson's R-value, R∼0, and Spearman's Rank Correlation, R s ∼ 0; ESI, Table S2-3 †).The lack of Os reaching the cell nuclei, together with their inability of these complexes to form adducts with nucleobases, 16 suggests that chromosomal DNA is not a primary therapeutic target for this type of anticancer compound.Moreover, 2-PF 6 was observed to accumulate in small elliptical areas ( previously identified as mitochondria) 17 after just 8 h (Fig. 1e; ESI, Fig. S12 †). 17 This finding correlates closely with that observed for glutaraldehyde-fixed A2780 cells after 24 h treatment with iodido complex 2-PF 6 and analysed by XRF, 17 and supports cell fractionation studies using ICP-MS and proteomic analysis for 2-PF 6 . 2,14Still, no particular localisation into cellular compartments or organelles could be detected for chlorido complex 1-PF 6 under the experimental conditions used.Therefore, the presence of a monodenate Cl bound to Os(II) in these half-sandwich azopyridine complexes both decreases the cellular accumulation and slows down the ability of the complex to reach cellular targets, when compared with their iodido analogues.h).Data were fitted in PyMCa, 24 and images generated in ImageJ software. 33The scale bar = 10 µm.It is notable that natural Zn is highly concentrated in the cell nucleus, whereas Os from the anticancer complexes is not.

Cellular stability of osmium azopyridine complexes in cancer cells
We assessed the intracellular stability of these half-sandwich Os(II) complexes by labelling their phenylazopyridine ligands with bromine-tags and determining the relative Os and Br accumulation and localisation in cancer cells using ICP-MS and nanofocused XRF.A new brominated chelating ligand was synthesised by reacting p-benzoquinone and 5-bromo-2-hydrazinopyridine in the presence of perchloric acid, and used to generate [Os(η 6 -p-cymene)(4-Br-phenylazopy-5-OH)I]PF 6 4-PF 6 a labelled analogue of the active half-sandwich complex [Os(η 6p-cymene)( phenylazopy-5-OH)I]PF 6 3-PF 6 (Tables 1 and 2).The Br-tag in the azopyridine ligand of 4-PF 6 is located on the pyridine ring, which provides high resistance towards nucleophilic substitution of the C-Br bond by biological thiols. 34This was done to avoid unwanted release of the Br group under cellular conditions, so most of the bromine detected using ICP-MS or XRF on cells treated with 4-PF 6 is expected to correspond to the intact ligand or complex, or fragmented chelating ligand (Scheme 1).
Bromination of Azpy-OH caused little alteration in the chemical properties of 4-PF6 when compared to non-brominated 3-PF 6 .Both were stable in phosphate buffer in concentrations of NaCl equivalent to those found in cells (∼25 mM), and showed similar acid dissociation constants ( pK a ) for their phenolic OH substituents (6.41 ± 0.02 and 6.78 ± 0.02 for 3-PF 6 and 4-PF 6 , respectively; ESI, Fig. S15-16 †), which exist as zwitterionic species under physiological conditions ( pH = 7.4).
Equally, 3-PF 6 and 4-PF 6 were readily converted to their chlorido and hydroxido analogues in the presence of 10 mol equiv. of GSH (ESI, Fig. S17-18 †).Moreover, they can form thiolato (Os-SG) and sulfenato (Os-SOG) adducts in presence of molar excesses of GSH (as previously observed for 2-PF 6 ; ESI, Fig. S17-18 †), 15 suggesting that they are activated in cells by the same mechanism.3-PF 6 and 4-PF 6 also possess equivalent redox properties linked to the irreversible two-electron reduction of the azo-bond (NvN) in their chelating ligands.The half-wave reduction potentials, determined using an Ag/ Ag + reference electrode in acetonitrile with tetrabutylammonium hexafluorophosphate (0.1 M) as a supporting electrolyte (−0.66 and −0.74 V for 3-PF 6 and 4-PF 6 , respectively; ESI Fig. S19 †), are comparable with those previously reported for analogous osmium arene complexes (including complex 2-PF 6 , −0.64 V), 35,36 and indicate that reduction of the phenylazopyridine might occur under physiological conditions.However, 4-PF 6 was slightly more hydrophobic than 3-PF 6 (HPLC capacity factors K f = 1.51 ± 0.02 and 0.90 ± 0.01, respectively; Fig. 2a), likely owing to the large size, polarisation and increased London dispersion forces of the Br substituent.
The anticancer properties of 3-PF 6 and 4-PF 6 correlated strongly with those from other osmium azopyridine complexes of the same family carrying Os-I bonds (Fig. 2, Table 2; ESI Table S4 †), which showed similar antiproliferative potencies and cellular accumulations, and were largely taken up by active transport mechanisms such as endocytosis. 14,19Yet, 3-PF 6 and 4-PF 6 were less selective towards non-cancer cells than 2-PF 6 .They inhibited the proliferation more (Table 2) and were accumulated in greater quantities by MRC-5 lung fibroblasts (Fig. 2; ESI, Table S4 †).This could be due to differences in the interaction between Azpy-OH complexes 3-PF 6 and 4-PF 6 (zwitterionic at physiological pH) and 2-PF 6 (cationic) with the surfaces of membranes from cancer cells (overall charge) and healthy cells (neutral). 37,38omplex 4-PF 6 is therefore a suitable model to study the incell stability and distribution of this type of half-sandwich Os(II) drugs using ICP-MS ( 189 Os,79 Br, 127 I) and XRF (Os L 3 M 5 , Br KL 3 ).First, the relative cellular accumulation of the different components of the complexes (i.e.Os, chelated azopyridine and halido leaving group) was determined by measuring the quantity of osmium ( 189 Os), bromine ( 79 Br) and iodine ( 127 I) in cancer cells using ICP-MS.A2780 cells were treated with 4-PF 6 (1× IC 50 , 0.42 µM) for 4, 8 and 24 h, or for 24 h followed by 24, 48 or 72 h recovery in complex-free medium (Fig. 3; ESI, Table S5 †).Then, cell pellets were collected and digested in tetramethylammonium hydroxide (TMAH).TMAH was used instead of nitric acid to avoid oxidation of bromide to volatile Br 2 , and to overcome the high ionisation potential of the halogens (Br = 11.8 eV and I = 10.5 eV). 39s predicted based on our previous experiments using iodine ( 131 I) radiolabelling, the intracellular activation and cellular efflux of the iodido ligand occurred very fast. 15Hence it was not possible to detect iodine ( 127 I) from 4-PF 6 inside A2780 cells even at the shortest exposure time of the experiment (4 h, Fig. 3a).On the contrary, both osmium ( 189 Os) and bromine ( 79 Br) from the azopyridine ligand were detected inside cancer cells treated with 4-PF 6 .Accumulation of 4-PF 6 in A2780 cells followed the same pattern as previously observed for other organo-osmium complexes of the same family (i.e.1-PF 6 and 2-PF 6 ), 19,40 but it showed faster initial influx and efflux rates.Maximum amounts of intracellular osmium and bromine were found after just 4 h exposure (∼ 32 ± 1 and 9.2 ± 0.3 ng Os or Br/10 6 cells respectively; Fig. 3a; ESI, Table S5 †), and then decreased at a fast rate until the removal of the drug from the medium.
During the exposure of A2780 ovarian cancer cells to 4-PF 6 (4-24 h, no recovery), the molar ratio between 79 Br/ 189 Os remained relatively constant (ca.0.65), with less bromine than osmium inside cells (Fig. 3, ESI, Table S5 †).This suggests the partial in-cell degradation of the complex.Moreover, the lower amount of intracellular Br might arise from a faster efflux of the brominated azopyridine ligand (or fragments of it) compared to osmium-containing fragments.Yet, efflux of osmium at each timepoint seemed to be always higher than that of bromine (ESI, Table S5 †), ruling out that possibility.Alternatively, higher levels of Os than Br might be achieved by rapid internalisation of both intact complex and cationic Os fragments from the degradation of 4-PF 6 (occurring inside cells followed by excretion, or in the extracellular medium).Nevertheless, the 79 Br/ 189 Os ratio remained constant throughout the incubation with 4-PF 6 , indicating the presence of significant quantities of azopyridine ligand bound to Os, and probably intact complex inside A2780 cells.The intracellular level of Br and Os approached equal quantities (molar 79 Br/ 189 Os became ca. 1) just 24 h after 4-PF 6 was removed from the medium, and remained the same even after 72 h recovery (Fig. 3b).As efflux from A2780 of fragments containing Os and Br occurs at different rates, most of the remaining intracellular osmium should be still coordinated to the azopyridine ligand a long time after it was initially internalised by the cancer cells.ICP-MS experiments also showed that MRC-5 healthy lung fibroblast cells accumulated much more Os than Br (molar ratio  Fig. 4 Synchrotron-XRF elemental maps of cryo-fixed and freeze-dried A2780 (human ovarian) cancer cells grown on silicon nitride membranes treated with 7 × 50 of brominated complex 4-PF 6 (3 μM) for 4, 8 or 24 h (no recovery), showing phosphorus, osmium and bromine as represented using the 16-colour image setting in Image J software. 33Data were obtained using an incident energy of 15 keV: (a) untreated, (b) 4-PF 6 (4 h), (c) 4-PF 6 (4 h).Data were fitted in PyMCa, 24 and images generated in ImageJ software. 33The scale bar = 10 µm.
The stability of the Os-azopyridine bond found in nated complex 4-PF 6 inside A2780 ovarian cancer cells was further investigated using nanofocused XRF by monitoring the relative distribution of Os (L 3 M 5 ) and Br (KL 3 ) emissions in a time-dependent manner.Cells were treated with 7× IC 50 of 4-PF 6 for 4 and 24 h before cryo-fixation and freeze-drying.Acquisition of elemental maps from those cells allowed investigation of whether the complex remains intact once internalised.As expected, XRF emissions from Os and Br were not detected in untreated A2780 cells (Fig. 4a; ESI, Fig. S23-S27 †), but were clearly visible even after just 4 h treatment with 4-PF 6 (Fig. 4b; ESI, Fig. S28-S33).As previously observed for A2780 cells treated with 1-PF 6 and 2-PF 6 , cells treated with 4-PF 6 for 4 h did not show significant alterations in their size and morphology, which were similar to those of the untreated cells (ESI, Table S6 †).However, cells treated with the brominated complex for 24 h were significantly damaged, maintaining the roundness of the cells, but with changes in their total area (as seen for 2-PF 6 , ESI, Table S6 †).Moreover, Os and Br co-localised strongly in cells treated with 4-PF 6 for both 4 and 24 h exposure times (R = 0.33 ± 0.04 and 0.38 ± 0.08, respectively; ESI Table S7 †).This is consistent with the results of ICP-MS studies, reinforcing the suggestion that most of the intracellular osmium is coordinated to the azopyridine ligand.Thus, significant amounts of this type of complex might still remain intact even long periods of time after they have been internalised by cells.

Conclusions
2][43][44][45][46] Improvements in the design of metallodrugs require consideration not only of the role of the metal and its oxidation state, but also that of the ligands, both monodentate and chelating ligands.Investigation of the redox and ligand exchange chemistry of metallodrugs inside cells is a major challenge, but an important step towards optimising structure-activity relationships for both efficacy and unwanted side-effects.
Here we have probed reactions of half-sandwich Os(II) anticancer complexes [Os(η 6 -p-cymene)(R 1 -PhAzPy-R 2 )X]PF 6 in human ovarian cancer cells by monitoring Os, Br and I by ICP-MS and XRF.These measurements have allowed us to investigate the accumulation and localisation of the complexes in cancer and normal cells, as well as the release of the monodentate iodido ligand and chelated bromo-azopyridine.These results confirm that this family of Os(II) azopyridine complexes are activated rapidly in cancer cells via hydrolysis of the Os-I bond, most likely due to attack by intracellular GSH on the azo-bond.We have also shown for the first time that the osmium-azopyridine fragment generated after activation is remarkably stable in cancer cells over 24 h, although it is still unknown if it is part of a larger fragment also containing the bound arene, ({Os( p-cymene)(azopyridine)} 2+ ).Release of the arene might be expected to facilate oxidation of Os(II) to Os(III). 21These complexes also appear to be degraded more rapidly in normal (non-cancerous) cells, perhaps providing a possible mechanism for selective antiproliferative activity towards cancer cells, which may reduce unwanted patient side effects.This work demonstrates how halogen tags can be used to probe the in-cell stability of metal complexes, providing crucial insights into their mechanisms of action necessary for clinical progression.

Fig. 3
Fig. 3 (a) Time-dependent cellular accumulation (nmol/10 6 cells) of 189 Os (orange), Br (blue) and 127 I (green) in A2780 cells treated with 1 × IC 50 of 4-PF 6 (0.42 μM) for 4, 8 and 24 h, and additionally 24 h with 24, 48 and 72 h recovery, as determined by ICP-MS.Data were normalised to the number of cells collected in each pellet.Error bars show the standard deviation from the mean.(b) Time-dependent intracellular bromine-toosmium molar ratios (Br/Os) of A2780 cells treated with 1 × IC 50 of 4-PF 6 (0.42 μM) for for 4, 8 and 24 h, and additionally 24 h with 24, 48 and 72 h recovery, as determined from the ICP-MS data in (a).