Tracking Reactions of Asymmetric Organo‐Osmium Transfer Hydrogenation Catalysts in Cancer Cells

Abstract Most metallodrugs are prodrugs that can undergo ligand exchange and redox reactions in biological media. Here we have investigated the cellular stability of the anticancer complex [OsII[(η6‐p‐cymene)(RR/SS‐MePh‐DPEN)] [1] (MePh‐DPEN=tosyl‐diphenylethylenediamine) which catalyses the enantioselective reduction of pyruvate to lactate in cells. The introduction of a bromide tag at an unreactive site on a phenyl substituent of Ph‐DPEN allowed us to probe the fate of this ligand and Os in human cancer cells by a combination of X‐ray fluorescence (XRF) elemental mapping and inductively coupled plasma‐mass spectrometry (ICP‐MS). The BrPh‐DPEN ligand is readily displaced by reaction with endogenous thiols and translocated to the nucleus, whereas the Os fragment is exported from the cells. These data explain why the efficiency of catalysis is low, and suggests that it could be optimised by developing thiol resistant analogues. Moreover, this work also provides a new way for the delivery of ligands which are inactive when administered on their own.


Introduction
Tr ansition metal catalysts have potential as therapeutic agents to treat cancer and other diseases. [1] Such catalysts might transform multiple substrate molecules in situ, including exogenous prodrugs,d iagnostic agents,a nd endogenous metabolites, [1,2] whilst requiring only low concentrations to achieve the desired activity.T herapeutic strategies based on metal catalysts might help to overcome resistance to chemotherapy and reduce unwanted side effects, [1a,b] both of which are of current clinical concern.
[Os II [(h 6 -p-cymene)(RR/SS-MePh-DPEN)] [1]( MePh-DPEN = tosyl-diphenylethylene-diamine) is ac hiral 16-electron organo-osmium(II) half-sandwich complex structurally derived from the well-established Noyori Ru II catalysts (Figure 1a), [3] which shows high enantioselectivity and conversion rates.For example,reduction of acetophenone is ca. 3fold more efficient (in turnover frequency,T OF) and more stable (over one month under normal atmospheric conditions) than its industrially-used Ru II analogue. [3] Furthermore, once inside cells,a nd in presence of the non-toxic hydride donor formate,t his complex catalyses the enantioselective reduction of pyruvate,a ne ssential precursor in cell metabolism, to natural L-lactate or unnatural D-lactate,depending on the chirality of the catalyst. [1f] It can be assumed that the ability of such catalysts to cause metabolic perturbations in cells requires the presence of intact catalyst, which contributes to the antiproliferative activity and selectivity of 1 towards avariety of human cancer cell lines. [2e,4] Yet, its intracellular catalytic activity is most likely marked by low turnover numbers (TON;p reviously estimated to be % 13), which might suggest some degradation of the complex inside cells. [1f] This might be expected to be ac ommon problem for synthetic metal catalysts,s uch as organometallic complexes designed to work under welldefined chemical conditions,i ncluding inert atmospheres and in organic solvents. [1g,i,2c, 5] In order to optimise the design of synthetic intracellular catalysts,and increase their in cellulo catalytic and biological efficiency,i ti si mportant to investigate their fate in cells.
Previous work using ICP-MS experiments on fractionated cancer cells treated with 1,s howed ac a. 47 %c ytosolic accumulation of the Os, [1f] suggesting that catalysis may take place in the cytosol. Additionally,ca. 48 %ofintracellular Os was present in the membrane/ particulate fraction (which contains organelles and membrane proteins), which may also implicate organelles (i.e.m itochondria or lysosomes) as cellular targets. [1f] However, such studies did not provide information on the intracellular stability of the complex.
To probe this,w eh ave incorporated ah alide substituent on as ulfonylphenyl substituent in the chelated Ph-DPEN ligand, so generating [Os II (h 6 -p-cym)(R 1/2 Ph-DPEN)],c omplexes with R 1 (para) = Br (complexes R,R-and S,S-2), I (R,R-and S,S-3), F( R,R-4), or R 2 (meta) = Br (S,S-5). A combination of nanofocused synchrotron X-ray fluorescence (XRF) and ICP-MS allows not only osmium but also the chelated DPEN ligand to be tracked in cells using the halide label. These experiments shed new light on the chemistry of these organometallic catalysts in cells,and will aid the design of next-generation catalytic drugs.
Halogenation of the phenyl group did not alter significantly the structure of the complexes,asseen by comparison of the X-ray crystal structures of complexes S,S-1, S,S-2 and S,S-3 (Supporting Information, Table S1, Figures S7 and S8), Abstract: Most metallodrugs are prodrugs that can undergo ligand exchange and redox reactions in biological media. Here we have investigated the cellular stability of the anticancer complex [Os II [(h 6 -p-cymene)(RR/SS-MePh-DPEN)] [1] (MePh-DPEN = tosyl-diphenylethylenediamine) whichc atalyses the enantioselective reduction of pyruvate to lactate in cells.The introduction of abromide tag at an unreactive site on aphenyl substituent of Ph-DPEN allowed us to probe the fate of this ligand and Os in human cancer cells by acombination of X-ray fluorescence (XRF) elemental mapping and inductively coupled plasma-mass spectrometry (ICP-MS). The BrPh-DPEN ligand is readily displaced by reaction with endogenous thiols and translocated to the nucleus,whereas the Os fragment is exported from the cells.These data explain why the efficiency of catalysis is low, and suggests that it could be optimised by developing thiol resistant analogues.M oreover, this work also provides an ew way for the delivery of ligands which are inactive when administered on their own. and led only to am oderate decrease in their hydrophobicity (LogP 1.45 AE 0.02 for 1,ca. 1for 2, 3 and 5,and 0.30 AE 0.03 for 4;S upporting Information, Table S2). Density Functional Theory (DFT) calculations related such changes in hydrophobicity to differences in the effective charge distribution at the C À Xb onds due to the electronegativity of the halides (F > Cl > Br > I, Supporting Information, Table S3 Tables S3-6). Equally, 2-5 maintained their ability to act as catalysts for asymmetric transfer hydrogenation (ATH) reactions.T hey catalysed the reduction of acetophenone to (S)-or (R)-1phenylethanol in presence of formic acid, achieving high enantiomeric excesses and conversions (> 93 %), with the Br and Ia nalogues (2, 3)s howing slightly higher catalytic efficiency when compared with 1 (TOF = 81 AE 3, 87 AE 5a nd 63.6 AE 0.6 h À1 , [3] respectively;T able 1). This confirmed that labelling Ph-DPEN ligands with halogen tags did not affect significantly the chemical or catalytic properties of the new complexes.F urthermore, 1 HNMR and UV/Vis studies showed that the halogenated analogues of 1 were stable over 24 hinD MSO or PBS.
Enantiomers of 2 and 3 showed the same antiproliferative activity in vitro against ovarian A2780, lung A549, breast MCF7 and prostate PC3 cancer cells,w hile R,R-4 was more potent against A2780 cells.Still, all of them were slightly less potent than 1. (Figure 1d;Supporting Information, Table S7). Similarly,they were also less toxic to zebrafish embryos,a nd showed slightly improved biocompatibility (Supporting Information, Table S20). [4] ICP-MS showed similar amounts of Os in cells treated with equipotent concentrations of 1-3 after 24 h( dose of 1 IC 50 ,F igure 1d;0 .25-1.5 IC 50 ,S upporting Information, Tables S10 and S11). Moreover,i dentical timedependent influx and efflux of osmium was observed in cells treated with 1 IC 50 of 1 and 2 ( Figure 1b); reaching maximum accumulation after ca. 6-8 hf ollowed by ap eriod of concentration-independent efflux. Both complexes were also taken up to some extent by cells using energy-dependent mechanisms (i.e.e ndocytic pathways leading to lysosomal deposition), as accumulation of intracellular osmium decreased at 277 K( Supporting Information, Table S12). Reduction in accumulation was significantly lower for cells treated with 2.
Thea nticancer activity of R,R-and S,S-2 was improved (cell survival reduced by 74 %) by co-treatment with nontoxic concentrations of hydride donor formate ( Figure 1c; Supporting Information, Figure S10), but not by acetate, which cannot act as ahydride donor (Supporting Information, Figure S11). Thee ffect induced by formate was specific for Table 1: Catalytic conversion (%), enantiomeric excess (%) and turnover frequency (TOF, h À1 )f or the reduction of acetophenone to (S)-or (R)-1phenylethanol for complexes 1-5 in the presence of 5:2f ormic acid/ TEA azeotrope for 24 hat310 Kasdetermined by 1 HNMR and chiral GC. Halfmaximal inhibitory concentrations (IC 50 /mM) of 1-5 towards A2780 (human ovarian) and A549 (human lung) cancer cells upon 24 hcomplex exposure followed by a7 2hrecovery period in fresh media. Osmium cellular accumulation in A2780 cells treated with 1 IC 50 of 1-4 for 24 h(no recovery).
cancer cells compared to normal healthy cells (Supporting Information, Figure S12), and was not caused by an increase in the intracellular levels of Os (p > 0.5;Supporting Information, Table S13), or the disruption of membrane integrity (Supporting Information, Table S8, Figure S13). However,as previously observed for 1, [1f] the presence of 2 induced G 1arrest (Supporting Information, Table S9). These results confirm that halogen substitution does not affect significantly the chemical, structural and catalytic properties of 2-5 compared to 1.T he tagged complexes also show similar biological properties to the parent complex. Complexes 2-5 appear to have the same mechanism of action than 1,b eing capable of performing in-cell transfer hydrogenation. The small decrease in their anticancer potencym ay be be caused by reduced cellular accumulation, possibly related to lower hydrophobicity. Thehalogen tags used (i.e.F ,Brand I) are present in most biological organisms. [7] Bromine is reported to have an important role in connective tissue formation, [8] but is found in lower quantities than the other halogens in the body. [7a] Bromine is also present only at low concentrations in formulations for cell culture media and serum albumin. [7a, 9] Thus,low backgrounds were found when its accumulation and distribution were determined in cellular samples that were cultured in vitro (4.6 ng Br/ 10 6 cells). Furthermore,due to the relatively high stability towards nucleophilic substitution of the CÀBr bond in the BrPh-DPEN ligand in 2, [10] release of free bromide is unlikely to occur in the presence of biological thiols in cellular conditions.T herefore,m ost of the detected bromine should correspond to either complex 2 itself, displaced or fragmented BrPh-DPEN ligand. Hence,weused brominated complex 2 to probe the stability of this catalyst in cells by determining the relative accumulation and localisation of Os and the Br tag, by detection of 189 Os and 79 Br by ICP-MS,and by elemental mapping of Os and Br by detecting nanofocused synchrotron X-ray fluorescence (XRF) emissions of Os (L 3 -M 5 = 8.9 keV) and Br (K-L 3 = 11.9 keV).
Quantification of Br by ICP-MS was challenging due to the high energy of its first ionisation potential (11.8 eV), [11] and the presence of common polyatomic interferences with m/z = 79 or 81 (e.g. 40 Ar 39 K + or 31 P 16 O 3 + for m/z = 79; 32 S 16 O 3 1 H + or 40 Ar 40 Ar 1 H + for m/z = 81). [12] Furthermore, nitric acid could not be used as the cell digestion matrix, as it induces oxidation of bromine to molecular Br 2 .I nstead, pellets of A549 lung cancer cells treated with S,S-2 under different conditions were digested at 353 Kusing an alkaline solution of 25 % w/v tetramethylammonium hydroxide (TMAH). This reduced the loss of volatile analytes. [13] As expected, intracellular levels of Br were much lower for untreated cells than in cells exposed to S,S-2 (4.6 vs.185 ng/10 6 cells after exposure to 30 mMo fS,S-2 for 0o r3 ,h , respectively). Thec ells also accumulated Br and Os in ad ifferent time-dependent manner (Figure 2a). Theh ighest levels of intracellular Os were reached after ca. 4-8 h, and the levels of both Os and Br decreased upon recovery of cells in drug-free media after 24 he xposure (Figure 2a;S upporting Information, Table S14). However, there was aclear influx of Br into cells for the whole duration of the exposure to the complex (24 h). TheB rt ag in the complex was accumulated > 9 more (on amolar basis) than Os after just 3hexposure to S,S-2 (Figure 2a;S upporting Information, Table S14).
ICP-MS measurements using isolated subcellular fractions obtained from cells treated with S,S-2 also showed clear differences in the cellular distribution of Br and Os (Figure 2b;S upporting Information, Tables S15-19). Osmium was mostly (> 65 %) in the fraction containing the membranes and other organelles (i.e.mitochondria or lysosomes) at all of the time points analysed (3-96 h). Lesser amounts of Os were found in the cytoskeletal fraction, perhaps indicating interactions between cationic osmium species and negativelycharged microtubules,w hile very little was present in the cytosol and nuclei of the treated cells (< 6%). In contrast, while most Br was always present in the membrane and organelle fraction (> 58 %), much higher quantities of Br compared to Os were found at all times in the cytosol and nuclei of the treated cells (10-20 %).
ICP-MS experiments showed that not only the levels of Os,but also Br decreased when cells were exposed to S,S-2 at lower temperatures (277 Kv s. 310 K, by ca. 48 %a nd 75 %, respectively;Supporting Information, Figure S15). Lower Br- to-Os ratios found in cells upon inhibition of active transport also suggested the presence of higher quantities of intact complex (Supporting Information, Figure S16). Besides,e xcretion of Os and BrPh-DPEN seemed to occur through different mechanisms.I nhibition of caveolae endocytic pathways using methyl-b-cyclodextrin reduced cellular uptake of the complex (Supporting Information, Figure S17 and S18). Yet, it did not alter the intracellular levels of Br, possibly by inhibiting the cellular efflux of the free ligand. Moreover,Os (but not Br) efflux was reduced when cells recovered in 20 mM of verapamil after exposure to S,S-2 (Supporting Information, Tables S18,19, Figure S19). Verapamil inhibits the ATPdependent efflux membrane pump Pgp (Permeability Glycoprotein-1;awell-known pump involved in detoxification and drug resistance). [14] Thus,u ptake experiments suggested the presence of intracellular degradation of the complex, followed by differential cellular trafficking and efflux for Os and Br-carrying fragments.
It is difficult to differentiate between membrane-bound and internalized elements using 2D mapping techniques.Still, concentration of large quantities of such elements in specific organelles and cellular areas (i.e.n uclei, lysosomes,m itochondria, ER or other cytosolic organelles) can be easily detected, providing vital information on their cellular distribution. As such, the distribution of Os and Br in cancer cells was further studied at sub-cellular spatial resolution (100 100 nm 2 )b ya cquiring XRF elemental maps using nanofocused synchrotron radiation at I14 (Diamond Light Source). A549 lung cancer cells grown on silicon nitride membranes were treated with various concentrations of S,S-2 (1-5 IC 50 concentration) for 24 h, before being cryo-fixed and freeze-dried for subsequent analysis under ambient conditions (Figure 3). Natural intracellular levels of Br were below the detection limit ( Supporting Information, Figure S20). XRF emissions from Br or Os were not detected in untreated cells (Supporting Information, Figure S21-24). These control cells also maintained anormal "stretched out" morphology (as seen from S, Pand Kmaps) typical of this cell line, [17] and had clearly defined nuclei, areas with high accumulation of Zn (Supporting Information, Figures S21-24). On the contrary,X RF maps acquired from cells treated with S,S-2 showed the presence of drug-induced morphological changes,which were concentration-dependent( Figure 3; Supporting Information, Figures S25-S38). Forexample,cells became smaller in size and more rounded in shape when treated with 1-3 IC 50 of S,S-2.I nstead, the use of higher concentrations of the drug (5 IC 50 )l ed to cell swelling ( Figure 3; Supporting Information, Table S21, Figures S39-42) and nuclei with increased size and poorly-defined perimeters due to as parse intracellular distribution of Zn, suggesting rupture of the nuclear membrane.T his indicated significant concentration-dependent cell damage caused by 2.
XRF elemental maps obtained from cells treated with S,S-2 also showed that Os was located mostly in cytosolic regions (from 75-85 %for 1-3 IC 50 treatment, to ca. 65 %at5 IC 50 ; Figure 3, Supporting Information, Table S22, Figures S25-38). Equally,c ells treated with the complex accumulated more Br than Os (5, 7a nd 2 more at 1, 3a nd 5 IC 50 , respectively;S upporting Information, Table S23). Most of that intracellular Br was found in regions of the cytosol (71 AE 9%,6 0AE 7% and 52 AE 7% at 1, 3a nd 5 IC 50, respectively), but as ignificant amount of Br also reached the nuclei of treated cells (29 AE 9%,4 0AE 7% and 48 AE 7% at 1, 3a nd 5 IC 50 ,r espectively;F igure 3, Supporting Information, Table S23). Remarkably,t he percentage of Br in the nuclei calculated from XRF maps (29 AE 9%)was slightly higher than that found when nuclear fractions were isolated from cells treated under the same conditions (1 IC 50 )a nd analyzed using ICP-MS (12 AE 1%). XRF data are based on averaged elemental determinations from between 3-7 individual cryofixed, dried cells,i nc ontrast to ICP-MS,p erformed on digested cell fractions from al arge population of cells.A lso, fractionation kits can introduce elemental leeching and crosscontamination between fractions.M oreover,t he data obtained from XRF maps confirmed the trends found on our previous ICP-MS experiments,a nd implied that intracellular degradation of the complex was occurring. Nevertheless,t he Os and Br from the catalyst were found to co-localise moderately in the cytoplasm (PearsonsC oefficient R = 0.24 AE 0.11, 0.39 AE 0.05 and 0.17 AE 0.01 at 1, 3a nd 5 IC 50 , respectively;S upporting Information, Table S24). Thus,s uggesting that some of the complex should remain intact in those areas after 24 ht reatment, and could support transfer hydrogenation catalysis in the cytosol of cancer cells.F urthermore,b oth Os and Br appeared to accumulate in small (0.65 AE 0.21 mm 2 ,0.6 AE 0.1 mm 2 and 0.78 AE 0.25 mm 2 at 1, 3and 5 IC 50, respectively), cytoplasmic compartments of cells when they were treated with S,S-2 (Figure 4a-c;S upporting Information, Table S25, Figures S43-53). It is likely that these are lysosomes or endosomes,s ince they are known to be similar in size,a nd temperature-dependent accumulation experiments had shown (at least) partial uptake of the Os catalysts through endocytosis( or other active transport mechanisms). Remarkably,B r/Os ratios were lower in those areas than in the rest of the cell (4 AE 2, 4 AE 1and 1.48 AE 0.17 at 1, 3a nd 5 IC 50 ,r espectively,S upporting Information, Tables S23 and S25), suggesting the presence of higher concentrations of intact complex.
Since the catalysts were at least partially taken up by energy-dependent mechanisms and probably reached the lysosomes,w ei nvestigated whether they are stable in acidic, cysteine-rich environments typical of lysosomes (i.e.c ysteine proteases). [18,19] We incubated complexes 1 and 2 for 24 ha t pH 5.5 or 7, with 1or10mML-cysteine at 310 K. MS analysis showed that the complexes alone remained intact at pH 5.5, but in the presence of L-Cys they released their chelated MePh-DPEN or BrPh-DPEN ligands (Supporting Information, Table S26, Figures S54-S57). Thefragment or fragments carrying the Os center remain unidentified, as they could not be isolated. It seems unlikely that dissociation of the Brlabelled chelated ligand occurs in the culture medium used to treat A549 cells since although we have shown that such ad issociation can be induced by thiols,t he level of thiols in the medium is very low,w ith cysteine being present only as oxidised cystine (0.2 mM) which does not react ( Figure S58). Cys34 in the foetal albumin present (ca. 30 mM), which is in ac left, would be expected to be inaccessible to such ab ulky organometallic complex, as found previously for [(biphenyl)Ru(en)Cl] + . [20] Thus,degradation of the complexes by thiols such as GSH seems likely to occur in lysosomes after endocytosis.W et ested this hypothesis by reducing cellular levels of glutathione with low doses of L-buthionine sulfoximine (L-BSO), or inhibiting the activity of lysosomes in A549 cells with chloroquine diphosphate (which reduces activity of lysosomal proteases and prevents endosome maturation). [21] Remarkably,co-administration of L-BSO (5 mM; Figure S59) or pre-incubation with chloroquine (150 mM, 2h;F igure 5a) led to asignificant increase in the anticancer potencyof1 (by 36 %or49%,respectively) and 2 (by 25 %, for chloroquine). This may be due to increased accumulation of catalyst ( Figure 5b), but was not caused by ad ecrease in membrane integrity in cells exposed to both chloroquine and the catalysts  [15] and images generated in ImageJ. [16]  The calibration bar is in pg mm À2 .Data were processedu sing the PyMCatoolkit developed by the ESRF, [15] and images generated in ImageJ. [16] (Supporting Information, Table S27). It appeared that the complexes were degraded less by chloroquine-treated cells,as they accumulated higher quantities of Os and had lower Br/ Os ratios (Figure 5b,c, Supporting Information, Figure S60, Table S28). Interestingly,c hloroquine enhanced the antiproliferative response of 1 to formate,w hich supported the presence of increased concentrations of intact catalyst inside cells (Figure 5d,Supporting Information, Table S28).
Overall, these results provide new insights into the behaviour of this family of asymmetric transfer hydrogenation catalysts in cancer cells ( Figure 6). Thec omplexes are internalised by cells via ac ombination of both passive and active transport, reaching lysosomes,t he cytosol and some other organelles.O nce inside,i nt he presence of ah ydride donor, intact catalysts facilitate the catalytic reduction of pyruvate to lactate in the cytosol, altering the metabolism of cancer cells and inhibiting their proliferation. Thec atalysts also interact with intracellular thiols (e.g.cysteine-containing peptides such as GSH and proteins). This leads to the release of Os-containing fragments and intact (or fragmented) DPEN chelating ligands,w hich are likely to be charged at physiological pH. [22] Thef ragments generated during this degradation exhibit different cellular behaviour,w hich explained differences observed between the accumulation of Os and Br. Osmium-containing fragments are rapidly excreted from cells via Pgp membrane pumps,while chelated ligands show much longer in-cell lifetimes.A ss uch, RPh-DPEN ligands are highly accumulated by cells,r eaching even the cell nuclei before they are excreted by vesicle-related exocytosis.T his raises questions about the role of the free ligands in the activity of the complexes.H owever,d egradation of 1 and its analogues causes the loss of catalytic and biological activity of the complexes,r educing their overall anticancer potential. Furthermore,a lthough RPh-DPEN ligands are readily internalised by cells (Supporting Information, Table S15), they do not possess antiproliferative properties (IC 50 > 150 mM, Supporting Information, Table S7). Still, the interaction observed between 1 and cellular thiols could provide ab asis for developing analogous therapeutic complexes designed for the simultaneous delivery of active ligands and reactive fragments of metal complexes inside cells.
isation of abrominated transfer hydrogenation catalyst inside cells with ICP-MS and nanoscale synchrotron XRF mapping, combined with cellular uptake and mechanistic studies.These experiments showed that the catalyst was degraded in cancer cells,p robably through transport into acidic lysosomes following by reaction with cellular thiols.T he chelated Brligand (or aBr-fragment), but not Os,istranslocated into the nucleus.S uch reactions help to explain the low intracellular TONe stimated for these catalysts.T his work demonstrates the utility of halogen tags as probes for MS and X-ray based techniques which elucidate reactions of organometallic anticancer catalysts in cells.T he next step for improving the efficiency of the catalysts might be to tether the chelated ligand to the arene ring, as trategy which has been used effectively in chemical systems. [26] On the other hand, such ligand release reactions might be used to deliver drugs which would otherwise be poorly taken up into cells and be inactive if administered separately.
Data Availability:The data that support the findings in this study are availablei nt he Warwick Research Archive Portal (WRAP) repository, http://wrap.warwick.ac.uk/147754/.