Forty Years after the Discovery of Its Nucleolytic Activity: [Cu(phen)2]2+ Shows Unattended DNA Cleavage Activity upon Fluorination

Abstract [Cu(phen)2]2+ (phen=1,10‐phenanthroline) is the first and still one of the most efficient artificial nucleases. In general, when the phen ligand is modified, the nucleolytic activity of its CuII complex is significantly reduced. This is most likely due to higher steric bulk of such ligands and thus lower affinity to DNA. CuII complexes with phen ligands having fluorinated substituents (F, CF3, SF5, SCF3) surprisingly showed excellent DNA cleavage activity—in contrast to the unsubstituted [Cu(phen)2]2+—in the absence of the otherwise required classical, bioabundant external reducing agents like thiols or ascorbate. This nucleolytic activity correlates well with the half‐wave potentials E 1/2 of the complexes. Cancer cell studies show cytotoxic effects of all complexes with fluorinated ligands in the low μm range, whereas they were less toxic towards healthy cells (fibroblasts).

acid as ar educing agent. Thea ctive species wass hown to be the corresponding Cu I complex. Thus, reducing agents are mandatory for initiating the generationofr eactive oxygen species (ROS), which are required for exertingo xidatived amage to DNA. [1] Different groups have tried to enhancea nd manipulate this nucleolytic activity,f or example, via substitution of hydrogen atoms on the ligand scaffold. DNA cleavage can even be inhibited as by 2,9-dimethylphenanthroline (neocuproine), which scavenges Cu I andb locks the redox cycle involving the production of ROS. All [CuL 2 ] 2 + species( L = phen derivatives like dipyrido[3,2-d:2',3'-f]quinoxaline (dpq) and dipyrido[3,2a:2',3'-c]phenazine (dppz))i nvestigated until today requirede xternalr educing agents foroxidative DNA cleavage like thiols or ascorbate. [2][3][4][5] The only reportede xamples where no addition of external reductants was necessary were obtained by linkage of two [Cu(phen) 2 ] 2 + moieties by aliphatic and aromatic dicarboxylates.A"self-activation" mechanismf or ROS formation based on al igand radical species was proposed. [6] Our group has focusedo nt he manipulationo ft he redox properties of [Cu(phen) 2 ] 2 + complexes via substitution of the ligand scaffold with fluorine or fluorine-containing groups.I n this report, we are presenting the synthesis of new homoleptic Cu II complexes with phen ligands havingf luorine or fluorinated substituents at positions 5a nd/or 6( Figure 1). The interaction of these complexes with DNA, their nucleolytic activity as well as their cytotoxic properties are described herein. Fluorine-containing phen derivatives were obtained via Skraup synthesis as described previously,s tarting from prefunctionalized nitroaniline and nitroquinoline precursors. [7] Such phen complexesa re barely known to the literature even with other metals.T ot he best of our knowledge, the only examples up to date represent ah omolepticC u I complex with two to four CF 3   groups in the phen moiety and Pd II complexes [PdCl 2 (L)] and [PdCl 2 (PPh 3 )(L)] with L = 4,7-dichloro-5-fluoro-2,9-dimethyl-1,10phenanthroline. [8,9] Besides that, theoretical investigations have been carried out for Ru complexes with 5,6-difluoro-1,10-phenanthroline (F 2 phen)a nd 5-trifluoromethyl-1,10-phenanthroline (CF 3 phen). [10,11] The corresponding Cu II complexes were obtained by allowing copper(II) nitrate to react with the fluorinated phen derivatives (cf. S-1). Depending on the crystallization conditions, two differents olvates of the complex [Cu(F 2 phen) 2 (NO 3 )]NO 3 were obtained, the solid-state molecular structures of whichw ere determined by X-ray diffraction (see Experimental Section). Crystallization from water/ methanol/diethyl ether affordeda dihydrate (Cu(F 2 phen) 2 (a)), in which the Cu center is hexa-coordinated by two bidentate F 2 phen ligands, one monodentate nitrate ligand,a nd one aqua ligand in at ypical square-bipyramidal fashion, with the axial positions being defined by an N atom and the nitrate Oa tom (Figure 2a). This arrangementi s most likelys upported by an intramolecular OÀH···O bond between the nitrate and aqua ligands. Ad ifferent picturewas observed when crystallization was conducted in the absence of water,a ffordingam ono-MeOH solvate (Cu(F 2 phen) 2 (b)). In this case, the Cu II ion displays as omewhat square-pyramidal coordination by the two F 2 phen ligandsa nd one nitratol igand, with an Na tom defining the apical position ( Figure 2b). However,t he structure of the MeOH solvate can be interpreted as an intermediate case between penta-and hexa-coordination, as the sixth coordination site is occupied by aw eak, additional Cu···O contact to the nitrato ligand. The complex [Cu(Fphen) 2 (ONO 2 )]NO 3 ,e ven thoughc rystallizing from wet methanol with one equivalent of crystal water,a lso displays only a penta-coordination (cf. Figure S-3.3).H owever,i nt his case the arrangement of the ligandsi sb etter described as trigonal-bipyramidal, with two Na toms defining the axial positions. This coordination has been observed mostf requently in previously reported [Cu(phen) 2 ] 2 + derivatives. [12][13][14][15] The Cu II complexes with fluorinated phen ligandss howed remarkable changes of half-wave potentials E 1/2 depending on the substituents of their ligands (Table 1, Figure S-4). The Cu II complex with two fluorine atoms at the phen ligand was the easiest to reduce. As ar esult, substitution with as trongly electron-withdrawing group like Fo rC F 3 (and even more with two of them)s tabilizes the corresponding Cu I species. Similarly, substitution of the phen scaffold with CF 3 groups in positions 2a nd 9l ed to the highest potential for Cu II /Cu I ever measured in ar eversible redox process for am ononuclear copperc omplex (+ 1.1 Vv s. FcH/FcH + ;F cH = ferrocene). [16] The electronwithdrawing capability of the substituents (H < F < CF 3 % SCF 3 < SF 5 ) [17,18] correlates with the increase in half-wavep otential of the corresponding copperc omplexes.As imilar correlation of the electron-withdrawing properties of substituents and the reduction potential has been observedb efore for example,for metalloporphyrins. [19] With these trends in mind, the nucleolytic activity of the complexes in the presence and absence of reducing agentst owards supercoiled plasmid DNA was investigated.I nt he presence of ascorbic acid (    redox activity,b ut also affinity to DNA has to be considered. DNA affinity was determined with DNA melting experiments (cf. section S-6) and ethidiumb romide (EB) displacement assays with CT-DNA (cf. section S-7, Ta ble 2).
The meltingt emperature T m of CT-DNA was increased by 13 8Cb yt he most efficient DNA cleaving agent [Cu(phen) 2 ] 2 + , whereas for the fluorinated species DT m of only 1 8Ct o8 8C was observed (cf. S6). Likewise, EB displacement showedh ighest affinity,t hat is, highest K SV and K app constants ( Table 2) for the complexes of highest nucleolytic activity ([Cu(phen) 2 ] 2 + > [Cu(Fphen) 2 ] 2 + ). However, the affinity of the least active sulfurcontaining species ([Cu(SCF 3 phen) 2 ] 2 + > [Cu(SF 5 phen) 2 ] 2 + )w as comparable to the one of the more active difluorinatedand trifluoromethylated species( [Cu(CF 3 phen) 2 ] 2 + > [Cu(F 2 phen) 2 ] 2 + ). Small differences in the outcome of the DNA melting and EB displacements tudies probably result from the fact that different mechanismsu nderlie the two methods. Nevertheless,b oth experiments indicated ad ependenceo ft he strength of DNA bindingo nt he steric bulk of the complexes. The larger the substituent (e.g.,C F 3 vs. F) and the higher the grade of substitution (2 Fv s. F) on the phen scaffold, the lower was the affinity of the resulting complexes to DNA. Accordingly,a lso a lower nucleolytic activity of the complexes wasobserved.
Interestingly,i nt he absence of any added externalr educing agents unexpectedly high nucleolytic activity in case of the fluorine-containing complexes was observed-in contrast to the parentc ompound ( [ Cu(F 2 phen)] 2 + was 2.5 times more active than [Cu(phen) 2 ] 2 + ,w hich had an activity barely above the background (DNA reference). When Tris-HClw as used insteado f MOPS as ab uffer ( Figure S-5.3),t he trends were the same as in MOPS, however,p lasmidD NA was even cleaved into its linear form by [Cu(F 2 phen)] 2 + .I tshould be mentioned thought hat Tris (tris(hydroxymethyl)-aminomethane), which is commonly used in such assays,isap otential competitive ligand for Cu II . [20] According to the literature-reported association constantsl ogK (Table S-5) also ternary complexes with Tris and phen as ligands are conceivable to play ar ole in the DNA cleavage reaction. (In case of the cleavage reaction in the presence of ascorbate, the bufferchange did not reveal any differences.) The nucleolytic activity in the absence of added external reducing agents correlates linearly with the half-wave potentials E 1/2 of the complexes (Table 1, cf. FigureS-5.5 correlation coefficient r = 0.86). With the reduction potential becomingl ess neg-ative the Cu I species is expected to be stabilized by the respective electron-withdrawing substituents at the phen moiety. Such as tabilization apparently leads to higher DNA cleavage rates. As ac onsequence,ahigherd egree of fluorination caused ah ighera ctivity of the complexes( cf. [Cu(F 2 phen) 2 ] 2 + vs. [Cu(Fphen) 2 ] 2 + )i nD NA cleavage.
Cleavage withoutt he participation of external reducing agents like ascorbate might indicate ah ydrolytic cleavage mechanism, however,a lso photocleavage or oxidative mechanisms initiated by reducing species in the incubation solution are conceivable. Hydrolytic cleavagew as excluded due to demonstrated O 2 -a nd light-dependence of the cleavage reaction:U nder O 2 -depleted conditions DNA cleavagea ctivity of [Cu(F 2 phen) 2 ] 2 + decreasedi ndicating an oxidative mechanism. When the reaction was carried out in the dark, the activity decreasede ven more (additive effects of light and O 2 )p ointing to ap hotocleavage process (Figure S-5.4).W hereas the results were not conclusive for the parentc ompound-probably due to low cleavage activity-a processs imilar to the one of type I photosensitizers couldb eproposed for [Cu(F 2 phen) 2 ] 2 + . [21] Also, there is indication from the literature that photoreduction of Cu II to Cu I in phen complexes is possible. In line with what we have observed, the parent compound is photochemically rather inert, but highera ctivity is expected for species with lessn egative redox potentials, that is, higher tendency for forming Cu I . [22] Such Cu I speciesa re pronet oR OS generation. [23] Alternatively, DNA itself could serve as ar educing agent. DNA bases like guanine exhibit ar elatively low oxidation potential, [24] especially when considering Watson-Crick G·C base pairing. [25,26] Ap athway involving Cu III species like in the case of Cu II hydroxysalenec omplexes [27] is improbable due to the electron-withdrawing property of the fluorine-containing functional groups. [28] To test for ROS involved in the cleavage reactions cavenging experiments were carriedo ut. The quenching of cleavage ac-  showedt he participation of hydrogen peroxide and superoxide ( Figure 5). No quenching was observed in case of DMSO and NaN 3 whichmakes it unlikely that freely diffusible hydroxyl radicals or singlet oxygen are involved as ROS. It seems that the formation of ROS proceeds in as imilarm anner like the formation of ROS in the presence of external reducing agents like ascorbate including the stepwise one-electron reduction of oxygen to superoxide andf urthermore to ap eroxo species. [29] In order to evaluate the influence of fluorine-containing substituents on cytotoxicity,t he MTT assay was carried out with MCF-7 breast cancerc ells. Figure 6s hows cell viability in the presenceo f1 0mm Cu II complexes with fluorinated ligands in comparison to [Cu(phen) 2 ] 2 + with unsubstituted phen.
There is no correlation between the cytotoxicity and nucleolytic activity and DNA affinity of the complexes so that also other mechanisms have to be considered in inducing cell death. Indeed, lipophilicity seems to be decisive here to explain the cytotoxicity trends for the fluorinated compounds: CF 3 ,S CF 3 and SF 5 show higher lipophilicity incrementsi nc omparisont oF , [31] which was confirmed for the complexes through the water/n-octanol partition experiment [32,33] (log P for complexes with phen substituents H < F < 2 F < CF 3 < SCF 3 < SF 5 ,T able S-9). The compounds with the latter substituents (CF 3 ,S CF 3 ,S F 5 )s howedh igher (less negative) log P values (higherl ipophilicity) than the ones with the first ones (F,2 F) which correlates wellw itht he observed cytotoxic behavior. The fact that the parentc ompound representing the least lipophilic complex, [Cu(phen) 2 ] 2 + ,i ss imilarly cytotoxic as the most lipophilic ones, confirms that lipophilicity is only one factor determining cytotoxicity.
It has to be mentioned that recent findings indicatet hat Cu II phen complexes,a lthoughs table at pH 7.4 and 37 8Ci nw ater (as demonstrated in Figure S-10 by UV/VIS spectroscopy) most probablyd ecompose in cell culture media. Interaction with cell components and cell death may be due to Cu ions and phen acting separately. [34] Also, thiol-rich molecules such as metallothioneins and glutathione, abundant in the cytosola nd nucleus, might reduce Cu II and scavenge reduced Cu I even before the complex reaches itsp otentialt arget, the DNA. [35] Such behavior is even more probablef or longer incubation times (> 24 h). However, in these reported studies the amount of Cu accumulated in cell compartments wasd ifferent for different phen derivatives, indicating that efficiency of uptake, and eventually also cell death, indeed depends on the ligand moiety. [34] In conclusion, we have synthesized Cu II complexes of the type [Cu(phen) 2 ] 2 + with fluorinated phen ligands, which act as chemicaln ucleases without the otherwise required classical, bioabundant external reducing agents. This modification showsahigh value for the construction of such nucleases via simpled erivatization of ligands, changing the electronic properties of the [Cu(phen) 2 ] 2 + complex. Substitution of the ligands with fluorine-containing groups is responsible for al ess negative reduction potentialo ft he complexes thus enabling an activationo ft he nucleasew ithoute xternal reducing agent. All complexes showedh igh cytotoxicity in two different breast cancer cell lines (IC 50 < 10 mm), and the more lipophilic the ligand,t he higher the cytotoxic effect was. Among the fluorinated complexes, despite of being the least cytotoxic compound, [Cu(F 2 phen) 2 ] 2 + stood out by showingt he most distinctive differentiation between cancera nd healthyc ells. This complex was also the most efficient DNA cleaver outperforming the parent compound [Cu(phen) 2 ] 2 + ,w hich does need an

Experimental Section
For the synthesis of complexes, methods and molecular and cell biological studies see the Supporting Information. Human dermal fibroblasts isolated from neonate foreskin biopsies after ethical approval (EA1/081/13, Ethics Committee from the CharitØ Campus Mitte, Berlin) and with informed parental consent, were provided by the Institute of Pharmacy (Freie Universität Berlin).
Deposition numbers 2032216 (Cu(F 2 phen) 2 (a))a nd 2032217 (Cu(F 2 phen) 2 (b))c ontain the supplementary crystallographic data for this paper.T hese data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.