Light‐Activated Carbon Monoxide Prodrugs Based on Bipyridyl Dicarbonyl Ruthenium(II) Complexes

Abstract Two photoactivatable dicarbonyl ruthenium(II) complexes based on an amide‐functionalised bipyridine scaffold (4‐position) equipped with an alkyne functionality or a green‐fluorescent BODIPY (boron‐dipyrromethene) dye have been prepared and used to investigate their light‐induced decarbonylation. UV/Vis, FTIR and 13C NMR spectroscopies as well as gas chromatography and multivariate curve resolution alternating least‐squares analysis (MCR‐ALS) were used to elucidate the mechanism of the decarbonylation process. Release of the first CO molecule occurs very quickly, while release of the second CO molecule proceeds more slowly. In vitro studies using two cell lines A431 (human squamous carcinoma) and HEK293 (human embryonic kidney cells) have been carried out in order to characterise the anti‐proliferative and anti‐apoptotic activities. The BODIPY‐labelled compound allows for monitoring the cellular uptake, showing fast internalisation kinetics and accumulation at the endoplasmic reticulum and mitochondria.


Introduction
Carbon-monoxide-releasing molecules (CORMs) represent a promisingp rodrug approach since it is known that CO plays a beneficial role in mammals, showing anti-bacterial, anti-apoptotic, anti-proliferative and anti-inflammatory effects. [1] To control the administration of the highly toxic gas in as afe manner, certain carbonyl compounds have been designed in which CO is either covalently bound in organic molecules [2] or coordinated to metal centres [1b, f, 3] in inorganic complexes. Transitionmetal carbonylc omplexes are particularly suitable because they feature ar ich chemistry and the MÀCO bond strength (p acceptori nteraction) can be manipulated, for example, through oxidation or electronic excitation by meansofe xternal light irradiation, being triggered by enzymes,l igand exchange reactions, thermalo rb yp Hc hanges. [1b, f, 3b, c, 4] Several CORMs, especially the widely investigated and prominente xample of transition-metal complexes [Ru(CO) 3 Cl(glycinate)]( CORM-3) and derivatives of boranocarbonate Na 2 [H 3 BCO 2 ]( CO-A1) [5] gained considerable attention due to their interesting properties in vitro and in vivo. [1b, c, 6] However,t hese classes of CORMs tend to release the CO molecules in ah ardly controllable way.
For that reason,p hotoCORMs have been developed inter alia to only liberate the CO upon exposure to light of certain wavelengths, allowing for dosage control in affected tissues/organs as well as liberation at desired locations. [3d, 4, 7] To be useful for therapeutic applications,p hotoCORMs must fulfil various requirements, that is,s tability anda ppropriate solubility under physiological conditions, sufficient high decarbonylation rate constants, non-toxicity of the "non-activated" photoCORM and utilisation of biocompatible wavelengths (preferably near-infrared light for which living tissues are permeable). [8] To test and proof the therapeutic potentialo fm etal-based photoCORMs, many anti-proliferative studies were conducted with rhenium(I)-and manganese(I)-based photoCORMs using UV or bluelight. Exemplary studies revealed an unspecific toxicityi n human cancerous cell lines such as colon adenocarcinoma (HT29), cervical (HeLa), ovarian (A2780), cisplatin-resistant ovarian (A2780CP70) and breast cancer( MDA-MB-231), assuming that the toxicityw as induced by the released CO. [7h, 9] However, the exact mechanism of pharmacological activity is often unclear and needs further investigation.
Although Ru II dicarbonyl complexes based on 2,2'-bipyridine derivatives substituted at 4,4'-1 and 5,5'-positions 2 (Scheme1) are knowna se ffective CO releasers, [10] surprisingly,l ittle if any biological studies on this class of Ru-CORMs are reported so far.T hese systems are characterised to undergo as tepwise decarbonylation upon UV irradiation with relatively high rate constants. [10b, c] Previously,w er eported about the influence of the electron-withdrawing carboxyla nd electron-donating methyl groups attached to the bipyridyl ligand of Ru II dicarbonyl complexes 1 on the rate of photolysis in organic and aqueous solvent systems. Electronic properties of the ancillary ligands enable ad ecrease in metal d-electron density and, thuss upports the release of CO. The study clearly revealed as tepwise decarbonylationp rocess with rate constants showing ad ifference of one order of magnitude between each other and asolvent dependency. [10b] The second rate constant depends in particularo nt he electronic structure of the substituent, that is, with coligands bearing electron-withdrawing-groups, such as carboxylicg roups, the complexes tend to releaset he CO slowert han complexes in which ligands bear electron-donating groups,s uch as methyl groups. This is also true when the carboxylicg roup is replaced by the amide function. Importantly,t he CO release is faster when ac arboxamide group is connectedv ia the amide nitrogen atom to the bipyridyl skeleton (2b)c ompared to C-connection ( 2a). [10d] Ru II dicarbonyl complexes based on 2-(2'-pyridyl)pyrimidine (cppH) ligands 3 are less efficient CO releasers than bipyridine-based complexes 1 and 2.T here are no significant differences in CO release properties when the carboxylic (3a)i sr eplaced by an amide group (3b). [10a] Neither conjugation of biomoleculeso rf luorescent dyes nor detailedi nvestigations of the influence of amide functionalization on the CO releasep roperties of bipyridyl dicarbonyl Ru II complexes 1 have been reported so far and biological properties are scarce.
Herein, we report on the preparation and characterisation of two amide-functionalised (4-position) ruthenium(II) dicarbonyl bipyridyl complexes bearing an alkyne group (complex 5)a nd af luorescence tag (complex 10), consisting of aB ODIPY (boron-dipyrromethene) dye, respectively (Scheme 2). BODIPY dyes are ac lass of organic fluorophores, whiche xhibit excellent photochemical stability and spectroscopic properties useful for biological systems, [11] for example, to study the cellular uptake and localisation of the BODIPY-labelled compounds. As reported recently,c onjugateso fm etal complexesw ith BODIPY derivatives offer easy access to new theranostics. [12] We examined the photolysis of these UV-light-sensitive Ru II complexes and compare their antiproliferativea ctivities against two humanc ell lines, epidermoid carcinoma (A431)a nd embryonic kidney cells (HEK293) with reported structures (Scheme 1). Cellularu ptake behaviour and cellular localisation of the fluorescent Ru-CORM-BODIPY conjugate 10 using laser scanningc onfocal microscopy are reported in detail.

Results and Discussion
Synthesis of alkyne-and BODIPY-functionalised Ru II -CORM complexes Startingf rom 4'-methyl-2,2'-bipyridine-4-carboxylic acid (1b), an amide group was introduced to the bipyridine skeleton in different ways. It is essential to prepare the respective ligand prior to the complexationw ith the ruthenium precursor.
The 1 HNMR spectra of 4 and 5 in CD 3 CN as well as 1 H/ 13 CNMR spectra in [D 6 ]DMSO and FTIR data of complex 5 are displayed in the Supporting Information (Figures S1-S3).A sa result, ad istinctive up-field shift is observed for the proton signals of ruthenium(II) complex 5 compared to ligand 4 in the 1 HNMR spectrum ( Figure S1). Complex 5 shows nine signals. Six resonances are assigned to the aromatic protons of the bi-pyridine skeleton and three signals to the methyl (d = 2.62 ppm), methylene (d = 4.19 ppm) and alkyne (d = 3.27 ppm) protons. The two characteristic carbonyl ligands are observedi nt he 13 CNMR spectrumw ith chemical shifts of 196.3 and1 96.1 ppm in [D 6 ]DMSO ( Figure S2). Additionally,t he vibrations in the IR spectrum at 2073 and 2016 cm À1 correspond to the CO ligands ( Figure S3). The CH-and amid-alkyne vibrations are observed at around 3300 cm À1 .T he infrared vibrationo ft he CC triple bond is weak and is overlaid with one of the strongC Ovibrations.
An elegant synthetic pathway for the introduction of an amide group via biorthogonal click reactioni st he Staudinger ligation, in which an azide reacts with af unctionalised triarylphosphine. [13] Ac orresponding synthesisa pproach to reach the green fluorescent BODIPY-CORM 10 derivative is shown in Scheme 4. At first, the BODIPY moiety was prepared by as tandard procedure starting from 2,4-dimethylpyrrole and 4-(chloromethyl)benzoyl chloride. [14] Next, the chloride function at the benzylp ositiono f6 was substituted by the azide group to yield 7.A fterwards, the azide-containing BODIPY derivative 8 was treated with triphenylphosphine to form ar esonance-stabilised iminophosphorane intermediate 8 (not isolated)u nder elimination of N 2 .S ubsequent reaction of 7 with compound 1b,w hich was convertedb eforehand into the benzotriazolyl ester (not isolated), led to compound 9 in as tepwise Staudinger ligation approach. [15] The last step involved the complexation of 9 with the Ru-precursor [RuCl 2 (CO) 2 ] n under argon and exclusion of light to give complex 10 exclusively as trans-(Cl) isomer in 68 %y ield after crystallisation. The forma-tion was confirmed by 1 HNMR spectroscopy,s howing just six expected signals corresponding to bipyridinep rotons (d = 9.76-7.72 ppm) rather than 12 for the cis-(Cl) isomer. [10d] The characteristic signals of the BODIPY-core werec onfirmed using 1 H, 11 B, and 19 FNMR spectroscopy.T he methyl and heteroaromatic protonsa ppear at d = 1.36, 2.44, and6 .17 ppm. The 11 B and 19 FNMR signals are found in the expected region at d = 0.6 ppm and d = À143.7 ppm, respectively.T he characteristic resonances fort he two CO ligands of 10 appear at d = 196.3 and 196.1 ppm in the 13 CNMR spectrum and were evidenced by the symmetric and anti-symmetric vibrations at 2079 cm À1 and 2008 cm À1 in the IR spectrum.
Light-induced decarbonylation monitored by UV/Vis, FTIR and 13 CN MR spectroscopy UV/Vis absorbance spectra of 5 and 10 in water containing 0.8 %( v/v) DMSO are showni nF igure 1. It must be noted that the complexes are soluble in the respective solventu pt oc oncentrationso f1 00 mm for 5 and 50 mm for 10.E xtinction coefficients at all absorption maxima as wellasat350 nm were summarised in Ta ble 1. Similar absorption maxima have been reported for Ru II dicarbonyl complexes bearingb ipyridyl, pyrimidyl andt erpyridyl ligands. [10a-c] Twoa bsorbance bands at 313 and3 24 nm and as houlder at around3 70 nm (Table 1a nd Figure1,l eft, black line) are observedf or the alkyne-functionalised complex 5.T he transitions above 350 nm are assigned to metal-to-ligand and al igand-to-ligand charge-transfer (MLCT/ LLCT) bands.The other two electronic transitions are attributed to p bpy-to-p*b py LLCT bands. [10c] As found for 5,t he Ru II -CORM-BODIPY complex 10 exhibited the same absorption maxima in the UV range at 313, 324 nm and am ore significant shoulder at approximately 370 nm. A new band in the visible region at 509 nm appeared in the absorbance spectrum (Figure 1, left, red line). Thet ransition band at 509 nm is characteristic for the BODIPYc ore and it is due to p BODIPY-p*B ODIPY transitions. [16] An intensee mission band was observeda t5 39 nm (l ex = 509 nm), which corresponds to ar elativelyl arge stokes shift (Figure 1, right). Typically,n arrow stokes shifts and intensive fluorescencea re characteristic features of BODIPYd erivatives. [11a] To evaluate the photolabilities of both compounds 5 and 10, photolysis experiments were carriedo ut upon light exposure at 350 nm. In general,t he photo-induced decarbonylationo f the complexes [RuCl 2 (L)(CO) 2 ]( L = bpy and its derivatives) causes the cleavage of the CO ligandsf rom the coordination sphere when externall ight of high energy (UV range) is applied. The vacant site is replaced by solventm olecules. The mechanism has been previously elucidated by our group using multivariate curve resolution alternating least-squares (MCR-ALS) as well as by other groups with time-resolvedI Rs pectroscopy and DFT calculations. [10b, c, 17] Usually,t he release of the first CO occurs very rapidly (pico-second range), [10c] while the second CO proceeds substantially slower.I nf act, we have reported that decarbonylation is solvent-dependent andt he rate constantsi na queous systems were an order of magnitude slower than in organic solvents. However,t he first rate constant is still high, accompanying with av ery quick releaseo f the first CO ligand. Interestingly,t he rate constants in aqueous systemsw ere less affected by electronic influences of ancillary ligandsthan in acetonitrile. [10b] Changes in the absorbance of 5 and 10 in water containing 0.8 %( v/v)D MSO upon irradiation at 350 nm werem onitored over time by UV/Vis spectroscopy ( Figure 2). Within the first seconds of exposure to 350 nm (E v % 6mWcm À2 ), the spectra changed substantially for 5.T he bands at 313 and 324 nm diminished andan ew band at 304 nm growsi n. The shoulder at around3 70 nm shifted bathochromically and new broad bands appear between 400 and 550 nm. All these bands reachedamaximum extinction within 30 min, but decreased after 1h of irradiation. As imilar,t ime-dependent spectral behaviour is apparent for ruthenium(II) dicarbonyl compounds bearing similarbipyridyl ligands. [10b] Complex 10 behaves comparably since it exhibits similar electronic excited states in the UV range. The former bands at 313 and3 24 nm becamew eaker immediately and an ew band at 303 nm appeared.T he shoulder at about 370 nm is slightly shifted to around 360 nm and increased over time. The p BODIPY-p*B ODIPY transition at 509 nm underwent a1 0nm hypsochromic shift and increased in intensity.
To solve the mechanism and kinetics of the photo-decarbonylationp rocess, the UV/Vis spectra were analysedb yM CR-ALS, as described previously. [10b] Ak inetic model with three consecutive steps and two individualc ompounds (A, P1 and P2, Scheme 5) was calculated using the dataa t3 50 nm forc omplex 5 in water containing 0.8 %( v/v)D MSO whereas for complex 10 two kinetic modelsw ere fitted;m odel 1t hree species with two rate constants and model 2) two species with one rate constant.
The fitted spectra as well as the concentration profiles of 5 and 10 andt heir photoproducts are shown in FiguresS16-S17. For complex 5,o nly two spectra were available for the fit, because the first absorbance spectrumd rastically changes upon Table 1. Absorption maximaand molar extinction coefficients of complex 5 and 10 in water containing0 . exposure to 350 nm. The rate constant k 1 is considered as a lower limit with k 1 > 4min À1 and k 2 = 0.176 AE 0.05 min À1 .A s mentioned above, we determined two kineticm odels for complex 10.M odel 1d etermined three consecutive steps with two individual components (k 1 = 1.913 AE 0.067 min À1 and k 2 = 0.063 AE 0.001 min À1 ). Changesw erel ess pronouncedf or complex 10 than for 5,a nd we obtained a k 1 % 2min À1 ,asimilar order of magnitudea sf or 5.I ti st hus evident for 5 and 10 that the first rate constants k 1 are around one to two orderso f magnitude larger than k 2 confirming af ast first step, which is in agreement with previously published rate constants. [10b] Since rate constants of 5 are in the same order as for 10,t he kinetics of CO release is little influenced by replacing the carboxylic acid group with ac arboxamide group. It should be noted that the concept of kinetic order forp hotochemical reactions may be misleading. The rate constantsd iscussed here are considered conditional rate constants for the observed exponential decay.However,for comparable measurement conditions and similar extinction coefficients, the determined values of k n are ag ood representation of the quantum yield. [7c, 18] Moreover,t he quantumy ield can be directly derived when the photon flux is known (see Experimental Section). The photo-induced loss of the CO ligand from the coordination sphere was furtheri nvestigated by IR and 13 CNMR spectroscopy. Am onodecarbonylation was confirmed by IR and 13 CNMR spectroscopy for 5 and 10 after 4hof irradiation( Figures S14 and S15). The two characteristicc arbonyl resonances of 5 and 10 of around 196 ppm disappear after irradiation, but several 13 CO signals between d = 200-205 ppm appeared, indicating the formationo fs everal ruthenium(II) monocarbonyl isomerso ft he general formula [Ru II (bpy)Cl 2 (CO)(solvent)].I n addition, the IR spectra confirmed the appearance of the monocarbonyl species after exposure to UV light, showing a single carbonyl vibration band at 1982 cm À1 for both Ru II complexes.
The emission spectrao f9 and 10 before and after irradiation for 120 min to 350 nm light are shown in Figure 3. It is worth noting that for compound 10 an increaseo ff luorescenceo ver time was observed when exposed to UV light,b efore subsequently the fluorescence intensity startedt od ecrease, indicating the decomposition. However,t he control experiment with compound 9 showed completep hotobleaching after 60 min of irradiation at 350 nm, suggesting that the metal centrei ss lowing down the photochemical decompositionprocess. An UV-induced fluorescencee nhancement has been reported in the presence of ap hotosensitizer and oxygen molecules due to the production of reactive oxygen species(ROS). [19] In order to corroborate decarbonylation as the main CO release process, time-dependent gasc hromatography with a thermalc onductivity detector (GC-TCD) was performed. When complex 5 wasd issolved in DMSO/water( 3:1, v/v)a nd was irradiated at 390 nm (0.35 AE 0.02 mEs À1 )u nder N 2 ,t he amount of released CO increased significantly to about 1equiv.a fter 4h ( Figure S18). After 20 h, the amount of CO reached almost 1.5 equiv.p er Ru, corresponding to 75 %o ft he theoretical amount of 2equiv.c arbon dioxide was also detected, but the amount of CO 2 was just slightly increasing within time and retained almost constant at the end. In total, am inor CO 2 amount of 0.15 equiv.w as detected. The oxidation of CO to CO 2 in the presence of ac atalysti na queous systems, known as "water gas shift reaction", can occur. [20] This reactionu sually needs high temperature even in the presence of ac atalyst.

Proliferation studies
The effect of CO is highly cell-specific and is based on mechanisms that are not yetf ully clear.A ccording to the current knowledge,COexerts adecisive influence on mitochondrial ac- Scheme5.Serial mechanisms (model 1a nd 2) for the photoreaction of complex 5 and 10. A = startingc omplex; P1, P2 = photoproducts. Chem. Eur.J.2020, 26,10992 -11006 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH tivity by inhibiting the functiono fc ytochrome-c-oxidase (COX, complex IV), an important enzymei nt he respiratory chain of the cell. Thed isruption of electron transport and the associated oxidative phosphorylationl eads to an overproduction of reactive oxygen species. ROS causess ingle or double strand breaks in the DNA and is ultimately the reason for apoptosis. [6a, 21] The pharmacological potentialo ft he synthesised photo-CORMs 5 and 10 were investigated in two different human cell lines:e pidermoid carcinomac ell line (A431) and ah uman an embryonick idney cell line (HEK293) transformed by an adenovirus type 5( AD5) DNA without( here referred as dark) and with exposure to 350 nm light (E v % 6mWcm À2 )f or 10 min at 37 8C. The compounds wered issolved in water/DMSO 0.8 %( v/ v)a tc oncentrationsf rom 3t o1 00 mm.T he cells were treated with the respective compounds2 4h before exposure to UV light. After 24 ho ft reatment, the cells were irradiatedf or 10 min at 350 nm and stored in the incubator for2 4a nd 48 h prior to cell viabilityd etermination by using MTS assay.U n-treated cells (withouta ddition of compounds 5 and 10)w ith and without DMSO in water as well as kept in the dark and irradiated for 10 min at 350 nm were used as controls.T he results are shown in Figures 4and 5.
The induction of at oxic effect of CO was investigated in both cancerous (A431) and non-cancerous (HEK293) cells. No differences in cell viability was observed when A431 cells were treated with 5,w hile exposed to UV-A light for 10 min at 37 8C. All treated cells remain viable over the full concentration range.
In contrast, the HEK293v iability decreaseds ignificantly after 48 ho fe xposure to 350 nm light for 10 min while the non-irradiated cells remained viable. Concentrations above 60 mm reduce the cell viabilityo fi rradiated cells by more than 50 %. This observation points out that the non-cancerous cell line is more sensitive to CO, which induces at oxic effect not observed in the cancerous A431 cell. The results for the fluorescent complex 10 are comparable (incubated for 24 hi nt he dark in respective cell line, then irradiated at 350 nm for  10 min at 37 8Cw ith subsequently cell viability determination after 24 and 48 he xposure to UV light) in the respective cell lines ( Figure 5). Here too, ad ecrease of HEK293 cell viability is evident already at 12.5 mm after 48 h. At this time, just 68 %o f viable cells were obtained, which showed indeed as ignificant toxicityc ompared to the controls kept in the dark. In contrast to HEK293c ells, the cancerous A431 cell line seems to be unaffected by the treatment of photoCORM 10.N os ignificant decrease in cell viability was observed in the concentration range studied neither at 24 nor at 48 ha fter exposure to UV light.
The resultso btainedw ith Ru II compounds 5 and 10 seem contradictory to published proliferation/cytotoxicitys tudies where anti-proliferative and apoptotic activitiesh ave been described for Mn I -orR e I -based photoCORMs in various cancerous cell lines. [7h, 9a, c, j] It is worth nothingt hat CO applied as ag as, can act paradoxically as inhibitor or mediator for tumour cell growth. [22] Ru II -based photoCORMs may undergo ad ifferent mechanism of action upon CO release than Mn I -a nd Re I -based photoCORMs.
Overall, the resultsa re not easy to interpret, because on the one hand cell lines investigated have differentm etabolism and on the other the compounds contain severale lements, which can have ad ifferent influence on proliferation activity.C oncerning the latter,t here are an umber of biological studies with BODIPY derivatives of metal complexes,w hichi ndicate that the metal itself and the corresponding complexing agent are primarily responsible for their biological activity. [12d] For example,t he anti-proliferative activity of platinum complexes is reducedw hen BODIPY is incorporated. [23] Cellular uptake behaviour BODIPY fluorophores have been used for staining cell compartments foralong time. In this context, fluorescent ceramide derivatives accumulate at theG olgi apparatus and endoplasmic reticulum in particular. [24] Metal complexes with appended BODIPY moieties can also overcome the outer cell membrane and accumulate in the cytosol/cytoplasm. [12a] Platinum and or-ganometallic iridium complexes with BODIPY tags are characterised by mitochondrial targeting. [23,25] Cellular uptake is fast and can be accelerated by the presence of BODIPY. [12a] Ac yclopalladated BODIPYprobe (COP-1)w as used for monitoring the CO releasei nl iving cells (HEK293T). [26] In this way, it could be shown that [Ru(CO) 3 Cl(glycinate)] (CORM-3) was efficiently taken up by the cells andC Ow as released. Very recently it has been shown that ar elativelys imple modification of the BODIPY framework leads to organelle-specific targeting. [27] However, the processes that determine the targeting are not well understood.
Prior to the irradiatione xperiments, we monitored the timedependentu ptake behaviour in A431 and HEK293 cells with the green fluorescent compound 10.W ithout signal quantification, it is nicely visible how the penetration of 10 into the cells (Figure 6) within the first few minutes is observed. Theu ptake is slightly faster in HEK293 cells mostl ikely caused by ad ifferent metabolisms.The intracellular localization is similar in mitochondriaa nd endoplasmic reticulum (ER) after 30 min and 4h as was indicated by Pearsonc orrelation coefficient (PCC), [28] which is used to describe the colocalisation of fluorescent signals in biological microscopy.

Colocalisation
Fluorescence imaging of A431 and HEK293 cells visualises the intracellular uptake of the green fluorescent BODIPY-CORM compound 10 after 30 min (10 mm) ( Figure 7) and 4h (50 mm) (Figures S19, S20) of treatment by laser scanning confocal microscopy (LSCM). To analyse the distribution and the localisation in specific cell compartments, the cells were labelled with the counter stains MitoTracker form itochondria, ER-Tracker Blue-White DPX for endoplasmic reticulum, Hoechst33342 for cell nucleusa nd CellmaskDeep Red for plasma membrane between 15 and 30 min at 37 8C. As already discussed, compound 10 is taken up rapidly by both cell lines and is spread through the cytoplasm. As expected, an internalisation by the living cells with an unspecific accumulationi nm itochondria and endoplasmic reticulum (ER) (Table2)was detected.The mitochondria are an important target for CORMs. The close proximity to the respiratory chain should enhance the therapeutic outcome, once CO is released. The PCCs are close to one for both investigated organelles (Table 2), which corresponds to a high level of colocalisation. To summarise, no differences in localisation between the different cell lines nor ad ependency on incubation time wered etermined. Moreover,t he nucleim embrane are invaginated and the circular shape of nuclei are changed (marked with arrows). However,c ompound 10 did not penetrate the cell nuclei. As ar esult, no uptake in the cell nucleusw as detected and the shapeo fd eformation is not accompanied with increased toxicity. Similar resultsw ere reported elsewhere. [29] The colocalisation studies revealed similar Pearson correlation coefficients (PCC) quantified in mitochondria and endoplasmic reticulum (ER) after 30 min and after 4h ( Table 2). The Pearson correlation coefficient quantifies the degree of colocalisation in the region of interest. So, for example the PCC correlates the intensities gained from red pixels (mitochondria stained by MitoTracker staining) and green pixels (compound 10). Due to the similar PCCs, obtained at different time points, irradiation with UV light can be done already after 30 min of treatment with complex 10.
Apoptosis and necrosis assay A431 cells were incubated with compound 5 and 10 for 0.5 h and then irradiated for 10 min with UV light (UV hand lamp, 6W). As controls, we used A431 cells incubated with compound 5 and 10 for 0.5 h, but without irradiation as well as untreated cellsw hichw erek ept in the dark andi rradiated. The percentage of apoptotic and dead cells were evaluated using flow cytometry (Figure 8a nd Figure S21 A). Without irradiation, we did not find significant differences between the control (untreated cells) and cells treatedw ith both compounds 5 and 10.W edetectedaslightly higherp ercentage of dead cells (Q1) in the control, which can be caused by the preparation of the sample. However,v ery few cells were detected, which were in an early or late stage of apoptosis. In comparison, there is a visible shift of the cell population towards the apoptosis sector ( Figure S21, compartmentQ 2l ate apoptosis and Q3 early apoptosis) of cells treated with compound 5 and 10.H owever, these changes were not significant.  After UV irradiation, we determined an increase of dead and apoptotic cells (Q1+ +Q2+ +Q3) for the control (6 %k ept in the dark vs. 13 %i rradiated, Figure 8), showingt he cytotoxic effect of UV radiation. In comparison to the cells treated withb oth photoCORMs, the percentageo fd ead and apoptotic cells rose to about2 0% (22.1 %f or compound 5;2 0.1 %f or compound 10). These resultsa re significant (P > 0.1 controlv s. 5; P > 0.5 control vs. 10)a nd thus we can assume that the enhanced cytotoxicity is not only ac onsequence of the UV radiation but is also due to the CO release. After irradiation of compound 5 and 10 the cytotoxicity slightly increased to 9% and 7%,r espectively.T hese results reveal the potentialo ft hese compounds to act as photoCORMs. However,f urtheri nvestigations are necessary to clarify the mechanism. As expected, the effect on cell death wass imilar for compound 5 and 10 exhibiting different functionalities. This confirms that the amidefunctionalised moiety (alkyne or BODIPY) has no influence on the cytotoxicity and thus marks the crucial importance of the Ru II dicarbonyl centre for the observed CO release in complex biological conditions.

Conclusions
The application of prodrugs that generate toxic species (e.g., CO, NO, 1 O 2 )u pon exposure to electromagnetic radiation provides an excellent strategy to treat cancerd iseases in ac ontrolled way.D icarbonyl ruthenium(II)c omplexesb ased on bipyridine ligands are suitables torage molecules for carbon monoxide to releaset he gas in as afe and controllable way. Twoa mide-functionalised (4-position) ruthenium(II) dicarbonyl bipyridyl complexes bearing an alkyne group 5 or af luorescence tag 10,c onsistingo faBODIPY (boron-dipyrromethene) dye, wereprepared for comparativestudies on CO release, biological activity andc ellular uptake. Concerningt he latter, BODIPY-based dyes feature as eries of sophisticated properties for in vitro applications,i nter alia, to investigate and to influence the uptake of labelled compounds to specific cell organelles.
The elucidation of the mechanism of CO releasefrom photo-CORMsi sf undamentalt oe xtract necessary information about structurala spects influencing photoreactivity or -toxicity.T he exposure to UV light induced aq uick monodecarbonylation for both complexes, which was confirmed by 13 CNMR and IR spectroscopy.As tepwise mechanism with two individual rate constants was determined by MCR-ALS. Thef irst rate constants were high and due to experimental setupt hey were considered lower limits only.I nc omparison, the second rate constants were determined at least one order of magnitude smaller.
Cell proliferation studies of both complexes were carried out using human embryonal kidney HEK293a nd cancerous A431 cell lines. It was shown that reduction of HEK293c ell viability after exposure to UV light is associatedw ith the bipyridyl dicarbonyl ruthenium complex as compound 5 and 10 exhibited similar results. In contrast, the cancerous cell line showed no significant reduction of viable cells. This result differs from the data described for photoCORMs based on Mn I and Re I .Sev-eral factors must be taken into account to interpret the activity of different compounds in different cell lines. Cell lines are very complex and do not necessarily behavei nt he same way.S o, the non-cancerous HEK293c ell line is very sensitivet oe nvironmentalc onditions, whereas the cancerous cell line A431 is very robust and better able to adapt to changes.
Laser scanning confocal microscopy studies revealed that the fluorescent complex 10 was quicklyt aken up by A431 and HEK293c ells with an unspecific accumulationi nm itochondria and the endoplasmic reticulum. Flow cytometry analysis indicates that after treatment with the Ru II dicarbonyl complexes 5 and 10 after 10 min of UV irradiation, the cancerous A431 cells are already damaged after 4h.T he proportiono fd ead and apoptotic cells is comparable for both complexes,s uggesting that the functionalisation of the bipyridine scaffold is of little if any importance. Further investigationsa re necessary to determine the mechanism of cell damage of the investigated photo-CORMs. Future work in this direction will focus on the investigation of furtherc ell lines and in particularo ni dentification of cell organelles responsible for apoptosis.

Materials and Methods
All syntheses were carried out using standard laboratory glassware. Prior to any light-sensitive reactions, an adequate protection against light was ensured by wrapping the glassware in aluminium foil. All chemicals and solvents were purchased from the following suppliers and used without prior purification:A BCR, Across, Alfa Aesar,A pplichem, Sigma-Aldrich, TCI Europe and Thermo Fisher. Reaction progress was monitored by TLC using Merck silica gel 60 F 254 plates and Macherey-Nagel Polygram SIL G/ UV254 plates with fluorescent indicator.P urification was performed by column chromatography (silica gel, particle size 0.035-0.07 mm, Acros Organics) or automated flash chromatography using aB iotage Isolera Four with ab uilt-in UV/Vis detector.I Rs pectra were recorded using aF isher Scientific Nicolet iS5 FTIR spectrometer featuring a built-in ATRm odule. The measurement was carried out in ar ange from 4000 to 400 cm À1 with ar esolution of 0.482 cm À1 .T he background measurement was carried out on an empty ATRc rystal. NMR spectra were recorded on two Agilent Te chnologies devices À400/54 Premium Shielded (400 MHz for 1 HNMR, 101 MHz for 13 CNMR, 128 MHz for 11 BNMR, 376 MHz for 19 FNMR) and 600/54 Premium Compact (600 MHz for 1 HNMR, 151 MHz for 13 CNMR,192 MHz for 11 BNMR, 564 MHz for 19 FNMR). The samples were dissolved in ad euterated solvent purchased from Deutero. All measurement were carried out at 25 8Ca nd the recorded NMR spectra were calibrated to the residual solvent signal [30] ( 1 Ha nd 13 C) or internal standards ( 19 F: CFCl 3 , 11 B: BF 3 .OEt 2 in CDCl 3 ). The chemical shift d in parts per million (ppm) relative to tetramethylsilane, the signal multiplicity,t he coupling constant J in Hertz, the number of cores and the assignment are given. The following abbreviations were used to describe the signal multiplicity:b r= broad, d = doublet, dd = double doublet, m = multiplet, q = quartet, s = singlet, t = triplet, td = triplet of doublets. Mass spectra with electrospray ionisation (ESI) mode were carried out on aQ uadroLC from Micromass. UV/Vis spectra were recorded using SPECORD 50 (Analytik Jena). The samples were transferred into Hellma quartz cuvettes of 1cmo ptical path length and 3mLv olume. The measurements were carried out in as pectral range from 270 to 700 nm, with a Chem. Eur.J.2020, 26,10992 -11006 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH spectral resolution of 1nm. To determine the extinction coefficients, a1 2.5 mm stock solution of the respective compound in DMSO was first prepared. For measurements in pure DMSO, 24 mL of the stock solution was mixed with 2976 mLD MSO. The resulting 100 mm solution was further diluted with pure DMSO as required. For measurements in 0.8 %( v/v)D MSO in water,2 4 mLo ft he 12.5 mm stock solution were dissolved in 2976 mLwater.The resulting 100 mm solution was diluted with a0 .8 %( v/v) DMSO/water mixture as required. Fluorescence spectra were recorded using LS 55 from PerkinElmer.T he samples were transferred into Hellma quartz cuvettes of 1cmo ptical path length and 3mLv olume. The measurements were carried out in as pectral range from 500 to 700 nm, with an excitation wavelength of 509 nm and spectral resolution of 0.5 nm. Sample preparation was carried out in analogy to UV/Vis spectroscopy.
Photolysis experiments were conducted using at emperature-controlled CCP-ICH2 photoreactor from Luzchem fitted with 16 deuterium lamps with emission wavelengths centred at 350 nm (full width at half maximum, FWHM AE 25 nm; E v % 6mWcm À2 ). Aq uartz fluorescence cuvette from Hellma (V = 3mL; d = 1cm) was used as reaction vessel. The extinction coefficients with 3157 and 11,100 m À1 ·cm À1 at 350 nm are well-suited to induce an efficient photo-decarbonylation reaction.
Ferrioxalate actinometry:F errioxalate actinometry was used to determine how much light is absorbed through ap hotoreaction. [31] Experiments were performed as described in the literature. [10a] A photon flow (I abs )o fa bout 3.8 10 À8 Einstein s À1 was determined. The reaction rate of ap hotoreaction is best described by Equation (1). Conditional rate constants can be derived were (pseudo)first order kinetic is observed. [18a, b] À dc dt where c represents concentration, t time, I abs the absorbed light intensity (photon flux), and F the quantum yield.
CO/CO 2 quantification:1 0mLo f0 .314 mm solution of complex 5 in DMSO/water (3:1, v/v) was added to a6 0mLr eaction vessel together with 200 mLo fp ure CH 4 as internal reference. The solution was irradiated using aL ED lamp (390 nm, photon flux of 0.35 AE 0.02 mEs À1 )f or 20 hu nder N 2 atmosphere. Gas chromatograms were recorded using aV arian CP-3800 gas chromatograph with helium as the carrier gas and a3m 2mmc arboxen-1000, 60/80 column. The gas flow was set to 20 mL min À1 .T he oven was operated isothermally at 100 8C. The 100 mLg as samples of the headspace were injected using aH amilton (1825 RN) gastight microliter syringe. The gases were detected using at hermal conductivity detector (Varian) operated at 150 8C. Calibrations were performed by the injection of known quantities of pure gases (CO or CO 2 )diluted in ar eaction vessel. 200 mLo fp ure CH 4 was used as internal reference to calculate the dilution factor.

Cell experiments
Human A431 (skin epidermoid carcinoma) and HEK293 (human embryonic kidney) cell lines were purchased from CLS (Cell Lines Service GmbH, Eppelheim, Germany) and DSMZ Leibniz Institute (DSMZ-German Collection of Microorganisms and Cell Cultures). The cells were cultivated in Dulbecco modified Eagle medium (DMEM), with or without the phenol red indicator and supplemented with heat inactivated 10 %f etal bovine serum (FBS), penicillin and streptomycin. All ingredients of the cell medium were purchased from BIOCHROM or Thermo Fisher Scientific. Cells were cultivated at 37 8Ci n5%C O 2 .
The MTS assay cell experiments were performed in Cellstar transparent flat bottom 48-well plates. The cells were seeded with a density of 10 000 cells per well. Each well contained 200 mLo f media. Cells were incubated for 24 ha t3 7 8Ci nahumidified atmosphere enriched with 5% CO 2 .
The photoCORMs 5 and 10 were added after 24 h. Af reshly prepared 12.5 mm stock solution of 5 and 10 in DMSO was diluted with cell medium depending on the final concentration. The final concentrations of 5 and 10 were adjusted to 25/50/60/75/100 mm and 3.12/6.25/12.5/25/50 mm by adding 2.4 mLD MSO solution of 5 and 10 and 97.6 mLc ell medium. The final concentration of DMSO in each well was then 0.8 %( v/v). Twon egative controls were used;e ither DMSO-containing cell medium (0.8 %( v/v)) or just cell medium with final volumes of 300 mLp er well. The outer wells were filled with phosphate buffer (PBS) to prevent the samples from evaporating.
24 ha fter the treatment of the cells with the respective complex, the cells were irradiated at 350 nm for 10 min using the Luzchem photoreactor CCP-ICH2. The cell medium was changed immediately after the irradiation-the old medium was removed and replaced by af resh, DMSO-and photoCORM-free one. The well plates were then stored in the incubator.C ells viability was measured using an MTS assay 24 or 48 hafter the irradiation.
Cytotoxicity assay:C ell viability was determined using the commercially available Promega CellTiter 96 aqueous one solution cell proliferation assay MTS kit. 24 and 48 ha fter the irradiation, 50 and 60 mLo ft he MTS reagent were added to each well and left in the incubator for further 1-2 h. The formed formazan was detected at 492 nm using the TECAN microplate reader Sunrise.
Intracellular uptake and colocalisation studies:C ellular uptake and organelle colocalisation of compound 10 was evaluated by laser scanning confocal microscopy (LSCM). Compound 10 (10 or 50 mm)w as incubated for 0.5 ha nd 4h at 378Ci n5%C O 2 atmosphere. Compound 10 was detected using the green colour channel (l ex = 485 nm, filter:4 90-590 nm). Cells were visualised using an Olympus FV10-ASV confocal laser scanning microscope (Olympus Czech group Ltd.,P rague, Czech Republic) equipped with a6 0x oil objective. Image analysis and level of the colocalization were determined using ImageJ software. Pearson correlation coefficients (PCC) were calculated using the same software.
Mitochondria, endoplasmic reticulum and colocalization:T he cells were seeded with ad ensity of 3 10 5 cells per well on 4 chamber glass bottom dishes (35 mm x 20 mm bottom well, 0.13-0.16 mm, Bio-Port Europe s.r.o.,P rague, Czech Republic). To the respective cells were added 10 mm or 50 mm (in 0.1 %o r0 .5 %D MSO) of compound 10 and incubated for 0.5 ho r4ha t3 78C. Then, the cells were washed with Dulbecco's modified PBS (DPBS) and labelled with am itochondria selective probe, MitoTracker (molecular probes, Thermo Fisher Scientific, Czech Republic) using 500 nm and for the endoplasmic reticulum ER-Tracker Blue-White DPX dyes for live-cell imaging (Molecular Probes, Thermo Fisher Scientific, Czech Republic) in af inal concentration of 1 mm.T he cell nuclei were labelled using Hoechst 33342 dye and cell membrane using Cell Mask Deep Red marker (Thermo Fischer Scientific, Czech Republic). The cells were labelled for 15 min at 37 8Ci n5% CO 2 ,t hen the cells were washed twice with DPBS and visualized.
Flow cytometry:F or flow cytometry the cells were seeded in 1mL of media into 24-well plates (TPP,C zech Republic) at density 1 10 5 cells per well. The cells were incubated with the compounds for 0.5 ha nd then irradiated for 10 min with UV light. The final concentration of compound 5 and 10 were 10 mm in 0.1 %D MSO. Medium was replaced by fresh one after the irradiation and the Chem. Eur.J.2020, 26,10992 -11006 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH cells were incubated another 4h before analysis. Then, the cells were detached from the cultivation plate by using 0.05 %t rypsin (Thermo Fisher Scientific, Czech Republic) and washed with PBS, and resuspended in 200 mL1 xb inding buffer (Bb) (10x 0.1 m Hepes/NaOH, 1.4 m NaCl, 25 mm CaCl 2 pH 7.4) and incubated for 10 min at room temperature with Annexin V-APC (ebioscience, Europe) diluted in Bb (5 mLo fA nnexin V-APC in 195 mLB b). Subsequently,t he cells were washed and resuspended in Bb with lethal marker 7-aminoactinomycin (7AAD) (1 mgmL À1 )( Invitrogen, USA) and incubated for 15 min at room temperature. Then, the fluorescent signals were evaluated using BD FACSverse flow cytometer (BD Bioscience, San Jose, USA) and the obtained data were analysed using flowJo software (Ashland, Oregon-based FlowJo LLC, USA) as medians of fluorescence intensity.T he experiments were prepared in four replicates. The statistical analysis of data were performed using Graph Pad Prism software 5.1 (Dunnett post test as part of one-way ANOVAanalysis).
Time dependent cellular uptake:T he cells were seeded with a density of 3 10 5 cells per well on 4-chamber glass bottom dishes (35 mm 20 mm bottom well, 0.13-0.16 mm, Bio-Port Europe s.r.o., Prague, Czech Republic). Compound 10 was added at final concentration of 10 mm containing 0.1 %DMSO and the uptake was visualised by LSCM after 4,8,12,16,20 and 24 min.