Synthesis and Characterization of an Epidermal Growth Factor Receptor‐Selective RuII Polypyridyl–Nanobody Conjugate as a Photosensitizer for Photodynamic Therapy

Abstract There is a current surge of interest in the development of novel photosensitizers (PSs) for photodynamic therapy (PDT), as those currently approved are not completely ideal. Among the tested compounds, we have previously investigated the use of RuII polypyridyl complexes with a [Ru(bipy)2(dppz)]2+ and [Ru(phen)2(dppz)]2+ scaffold (bipy=2,2′‐bipyridine; dppz=dipyrido[3,2‐a:2′,3′‐c]phenazine; phen=1,10‐phenanthroline). These complexes selectively target DNA. However, because DNA is ubiquitous, it would be of great interest to increase the selectivity of our PDT PSs by linking them to a targeting vector in view of targeted PDT. Herein, we present the synthesis, characterization, and in‐depth photophysical evaluation of a nanobody‐containing RuII polypyridyl conjugate selective for the epidermal growth factor receptor (EGFR) in view of targeted PDT. Using ICP‐MS and confocal microscopy, we could demonstrate that our conjugate has high selectivity for the EGFR receptor, which is a crucial oncological target because it is overexpressed and/or deregulated in a variety of solid tumors. However, in contrast to expectations, this conjugate was found to not produce reactive oxygen species (ROS) in cancer cells and is therefore not phototoxic.


Introduction
The use of photodynamic therapy (PDT) has expandedt he possible techniques in medicinet otreat various types of cancer (e.g.,l ung, bladder,e sophageal and brain cancer) as well as bacterial, fungal or viral infections. Its effect is caused by ac ombination of an ideally nontoxic photosensitizer (PS), oxygen and light. Upon light exposure,t he PS is able to produce reactive oxygen species( ROS), such as singlet oxygen ( 1 O 2 )o rother radicals. Due to the high reactivity of the latter, these can cause oxidatives tress and damage in different cellular compartments( e.g.,m embrane, nucleus, endoplasmic reticulum, lysosome, mitochondria), leadingu ltimately to cell death. [1] Next to the already approved PDT PSs, which are based on a tetrapyrrolic scaffold (i.e.,p orphyrins, chlorins, phthalocyanines), the development of Ru II polypyridyl complexes as PDT PSs is receiving more attention due to their ideal photophysical and photochemical properties, which include, among others, high water solubility,h igh chemicals tability and photostability, intense luminescence,l arge Stokes shifts, high 1 O 2 production. [1a-d, 2] These attractive features have allowed one of such complexes, namely TLD-1433, to enter into clinical trial as aP DT PS against bladder cancer. [3] Phase II has been recently started. [2f] In this context, our group was able to demonstrate that Ru II complexes of the type [Ru(bipy) 2 (dppz)] 2 + (bipy = 2,2'-bipyridine, dppz = dipyrido [3,2-a:2',3'-c]phenazine) and [Ru(phen) 2 -(dppz)] 2 + (phen = 1,10-phenanthroline) weree ffective PDT PSs ( Figure 1). [1a, 2c, 4] As ah ighlight,w ec ouldd emonstrate that some of these complexes were nontoxic in the dark and highly toxic upon light irradiation with IC 50 values in the low micromolar range and ap hototoxic index of up to > 150. [2c] Based on the extended planar p-systemo ft he dppz ligand, which is able to intercalate into the base pairs of the DNA, these compounds showed ap referable nuclear localization.U ponl ight exposure, these complexes caused oxidative stress, as wella s There is ac urrent surge of interest in the development of novel photosensitizers (PSs) for photodynamic therapy (PDT), as those currently approved are not completely ideal. Among the tested compounds, we have previously investigated the use of Ru II polypyridyl complexes with a[ Ru(bipy) 2 (dppz)] 2 + and [Ru(phen) 2 (dppz)] 2 + scaffold (bipy = 2,2'-bipyridine;d ppz = dipyrido [3,2-a:2',3'-c]phenazine;p hen = 1,10-phenanthroline). These complexes selectively target DNA. However,b ecause DNA is ubiquitous, it would be of great interestt oi ncreaset he selectivity of our PDT PSs by linking them to at argeting vector in view of targeted PDT.Herein, we present the synthesis, char-acterization, and in-depthp hotophysical evaluation of an anobody-containing Ru II polypyridyl conjugates electivef or the epidermal growth factor receptor (EGFR) in view of targeted PDT. Using ICP-MS and confocal microscopy,w ec ouldd emonstrate that our conjugate has high selectivity for the EGFR receptor, which is ac rucial oncological target because it is overexpressed and/ord eregulated in av ariety of solid tumors.H owever,i nc ontrast to expectations,t his conjugate was found to not produce reactive oxygen species( ROS) in cancerc ells and is therefore not phototoxic.
DNA photocleavage, suggesting that they impaired replication and integrity of the genetic material. [1a, 2c, 4] Highly proliferating cells like cancerc ells are generally preferably targeted by such compounds over healthyc ells, as it is the case for cisplatin. [5] However,o ther frequentlyd ividing cells in the organism (e.g.,h air follicles, gastrointestinal tract, bone marrow)c an be affected, leading to severes ide effects for the patients. [4a, 6] Thus, it is extremely important to increaset he selectivity of PDT PS, for example, with the development of a suitable delivery system. So far,t he examples of Ru II polypyridyl complexes for targeted PDT are scare, if we do not take into account polymer encapsulation/nanoparticle attachment. [4a, 7] The group of Lilge could recently demonstrate that the premixing of TLD-1433 with transferrin was able to increaset he extinction coefficient, prolongst he absorption range, reduced photobleaching, cellular uptake as well as overall toxicity of the compound. [8] Our group previously demonstrated the efficiency of the coupling of am etal-based PDT PS to peptides, which are known to bind specifically to abundant molecular targets on malignant cells. More precisely,i nt hose studies, bombesin, that is known to target the human gastrin-releasing peptider eceptor as well as an uclear localization signal peptidethat facilitates the intracellular transport into the nucleusw ere coupled to Ru-based PDT PSs. We were able to demonstrate an increased uptake of the conjugate in the receptor-expressingc ells in comparison with the free complex. [4a] The groups of Weil and Rau were able to link the peptide hormone somatostatin to aP Sa nd could show a1 00-fold increased efficiencyf or somatostatin receptorexpressing cells relative to the free PS. [7a] Recently,t he same authors described am acromolecular plasma protein serum albumin-PS conjugate with several Ru complexesb ound to the protein surface. Using the protein as an anocarrier,t he PSs were delivered selectively to the mitochondria, where it showeda ni mpressive phototoxicity with IC 50 values in the nanomolar range. [7c] Notably,avariety metal complexes as for example Re I ,P t II ,R u II or Ir III compounds have been successfully coupled to peptides to increasereceptor selectivity. [9] Among the differente stablished classes of delivery systems [10] (e.g.,o il dispersions, encapsulation in polymericp articles/lysosomes, targetingp eptide-PS conjugates, polymer-PS conjugates), the conjugation of PS to monoclonal antibodies (mAb) takes advantage of the excellent target specificity of the latter.H owever,d espite their clinical success, the concept of using mAb-PSc onjugates is afflictedw ith severali mportant drawbacks. These vectorm olecules are known for their high stabilitya nd prolonged serum half-life, slow pharmacokinetics and clearance from the body.T his leads to an increase of the absolute level of the mAb-PS conjugate in the tumor alongside with an increased nonspecific uptake in non-target tissues. [11] Additionally,t he treatment of solidt umors is limited due to penetration problemso ft he large conjugate into the tumor causedb yp oor vascularization, drainage, interstitial pressure and dense stroma. [12] An attractive strategy to circumvent these limitations is the use of smaller oncotropicv ector molecules like antibody fragments or nanobodies (NBs). [13] NBs represent the antigen-binding domain of heavy-chain-only antibodies that occur in species belonging to the familyo fCamilidae. Their smalls ize, stability,s olubility,f ast pharmacokinetics as well as high specificity and affinity for their cognate antigens make them powerful targeting agents for diagnostic imaging and targeted therapy. [14] Notably in this context, Caplacizumab, ab ivalent anti-vonWillebrand factor NB, is currently in phase III clinical trials against acquired thrombotic thrombocytopenic purpura. [15] Ar ecent study has highlighted the high tumor uptake, rapid blood clearance and low liver uptake of a 99m Tc -labeled NB as an imaging probe for epidermal growth factor receptor (EGFR) positive tumors. [16] Thisreceptor,which is involved in many cellular processes such as proliferation, differentiation and cell survival, represents ac rucial target in oncology as it is overexpressed and/or deregulated in av ariety of solid tumors,i ncluding head andn eck, breast, non-small-celll ung and pancreatic cancer.T herefore, EGFR is am ajor target for cancert herapy. [16,17] Notably,t he successful conjugation of the PS IR-Dye700DX-maleimide to nanobodies for hepatocyte growth factor receptor targeted PDT was recently demonstrated. [18] With this in mind, we report herein the design,s ynthesis, characterization and in-depthb iological evaluation of aN Bcontaining Ru II polypyridyl conjugate. The conjugatec onsists of three building blocks:1 )A [Ru(phen) 2 (dppz)] 2 + complex, which is known to have an excellent phototoxicity, [1a, 2c, 4] 2) a 7C12 NB, which is known for specific binding to EGFRe xpressing cells [16,19] and 3) ap eptide chain with ap oly-glycine unit, Figure 1. Structures of [Ru(bipy) 2 (dppz)] 2 + and [Ru(phen) 2 (dppz)] 2 + complexesa sP Ss developed by our group. [1a, 2c, 4] ChemBioChem 2020, 21,531 -542 www.chembiochem.org 2019 The Authors. Publishedb yWiley-VCH Verlag GmbH &Co. KGaA, Weinheim which is necessary for an efficient and site-specific conjugation by as ortase A( SrtA)-mediated trans-peptidationr eactionl eading to an 1:1N B:PS ratio. [20] To the best of our knowledge,w e report herein the first NB-containing Ru II polypyridylc onjugate as aP DT PS for EGFR-targeted PDT.A sc an be seen below, thanks to this design,ahighly selectiveN B-containing [Ru-(phen) 2 (dppz)] 2 + conjugate Ru-NB could be unveiled.

SortaseA-mediated conjugation
Site-specific attachment of the [Ru(phen) 2 (dppz-7-maleimidemethyl-S-Cys-(Ser) 2 -(Gly) 5 -NH 3 )] 3 + complex to the EGFR-specific NB 7C12 by sortase Ar equires protein engineering to endow the desired conjugation site at the C-terminal end of the NB with the unique sortase recognition motif. To this end, the NB was produced with its Cterminus tagged with a( GGGGS) 3 spacer followedb yaStrep-tag, the LPETGG sortase motif, another (GGGGS) 3 spacer and ah exahistidine purification tag (His 6 ). As successful sortase A-mediated conjugation leads to the eliminationo ft he His 6 tag, this design allows the removal of the unreactedN Ba sw ell as of the His 6 -tagged enzymeb y affinity chromatography (Scheme 2).

Photophysical properties
With the conjugate in hand, we performed photophysical measurementst oe valuate its potentiala saPDT agent. At first, the absorptions of [Ru(phen) 2 (dppz-7-maleimidemethyl)](PF 6 ) 2, [Ru(phen) 2 (dppz-7-maleimidemethyl-S-Cys-(Ser) 2 (Gly) 5 -NH 3 )]-(TFA) 3 and Ru-NB werem easured to investigate if the peptide chain or the NB conjugation had an influence on the photo-physicalp roperties of the Ru II polypyridyl complexes. Because the conjugate is insoluble in CH 3 CN, the measurements of Ru-NB were performed in DMSO. The comparison betweent he absorption spectra ( Figure S6) shows small differences that can be explained by solvent effects. As all major bands are still comparable, we assumet hat the conjugation did not change the photophysical properties of the Ru II polypyridyl complex. As as econd experiment,t he luminescence of the conjugate was investigated upon excitation at 450 nm in DMSO. The maximum of the emission of the complex ( Figure S7) was determined to be at 633 nm. Consequently,t here is al arge Stokes shift which results in minimal interference between excitation and luminescence. The luminescence quantum yield (F em )w as measured upon excitation at 450 nm by comparison with the model complex [Ru(bipy) 3 ]Cl 2 in CH 3 CN (F em = 5.9 %). [24] The luminescence quantum yield (F em )o ft he conjugate Ru-NB with av alue of 3.3 %w as found to be in the same range as other complexes of the type [Ru(bipy) 2 (dppz)] 2 + and [Ru(phen) 2 (dppz)] 2 + . [2c, 4] For ad eeper investigationo ft he excited state, the luminescence lifetimes were determined in degassed and air saturated DMSO upon excitation at 450 nm to investigate the influence of the presence of oxygen. As expected, the luminescence lifetimei nadegassed solution was much longer (589 ns, Figure S8) than in an aerated solution( 134 ns, Figure S9). This showst hat oxygen has as ignificant influence on the lifetimeo ft he excited state and indicates that 3 O 2 can interactwith the triplet state of the complex.

Singletoxygen generation
Knowingt hat the triplet excited state of the conjugate is able to interactw ith oxygen, we were interested in determining the singlet oxygen quantum yield F( 1 O 2 )o fR u-NB using two methodsp reviously described by our group, [25] namely: 1) Direct by measurement of the phosphorescenceo f 1 O 2 at 1270 nm. Notably, this methodi sd ependento nt he used setup. With the used equipmenti no ur laboratory,w ec an only detect F( 1 O 2 ) > 0.20;2 )Indirect by measurement of the change in absorbance of ar eporter molecule which is monitored by UV/VIS spectroscopy.B ecause the measurements were performed in DMSO and aqueous solution,o nly rather small values (Table 1) could be measured. This is not surprising and has already been investigated for severalo ther [Ru(bipy) 2 -(dppz)] 2 + and [Ru(phen) 2 (dppz)] 2 + derivatives. [2c, 4a, b] In-depth [a] Averageo ft hree independent measurements, AE 10 %( n.d. = not detectable).

In vitro evaluation of EGFR targeting after conjugation
To investigate the targeting ability of the functionalized NB, uptake in the human epithelial cell line A431 originating from an epidermoid carcinoma of the skin was examined by confocal fluorescencem icroscopy. These squamous carcinoma cells express approximately 2 10 6 EGFR molecules per cell, [27] which represents ah igh expression level. Confocal imaging of A431 cells showed co-localizationo fR u-NB with EGFR ( Figure 3), thus indicating the preserved targeting ability of 7C12 after site-specific modification.N otably,R u-NB showeda predominant membrane staininge ven after 48 ho fi ncubation at 37 8C, and only very little intracellularf luorescence was observed. However,i th as been shown recently that the free amine ruthenium complex is characterized by ap oor cellular uptake even at high micromolar concentrations. [4a] Cellular uptake of the bioconjugates The presence of am etal ubiquitous in ac ellular environment as an essential component of the PS allowsi nvestigating the cellular accumulation of the bioconjugate by inductively coupled plasma-masss pectrometry (ICP-MS). [28] To demonstrate the receptor-specific uptake, EGFR-positive (A431)a nd EGFRnegative (MDA-MB-435S) cells were incubated for different periodso ft ime (4, 24, and 48 h) with variousc oncentrationso f the bioconjugatei nt he dark at 37 8C. The amount of cell-associated ruthenium was determined by ICP-MS and related to the cellular protein content (Figure4). Although ruthenium was detectable in the cell lysate of both cell lines after 24 and 48 h, respectively,t he amount of the metal strongly correlated with the level of EGFRe xpression. There was more of ruthenium in the EGFR-overexpressingcell line than in the EGFR-negative one. This finding confirmed that cell associationw as primarily mediated by the NB andn ot by the PS. An identicalc ell uptakes tudy was performed with the complex [Ru(bipy) 2 (dppz-OMe)](PF 6 ) 2 , [2c] resulting in similar ruthenium levels for the A431 cell line ( Figure S10 and Ta ble 2). The amount of ruthenium detected in MDA-MB-435S cells upon incubation with this non-targeted Ru complex was highera t each time point relative to the EGFR-targeting Ru-NB conjugate. This result is unsurprising, as the latter cells lack these receptorp roteins at their surface.   To confirmthe receptor specificity of the ruthenium accumulation, A431 cells were incubated in the presence or absence of cetuximab in addition to Ru-NB. The epitope for 7C12 partially overlaps the cetuximab epitope on domain III of the EGFR extracellular region and an excess of the mAb can block its interaction with the receptor. [16,29] After 24 and 48 ho fi ncubation with 200 nm of Ru-NB at 37 8C, 0.77 and 2.74 ng ruthenium per mg protein (Table 3), respectively,w ere detected in the cell lysates. Upon co-incubation of EGFR-overexpressing A431 cells with Ru-NB and cetuximab, no cell-associated ruthenium was detectable even after 48 h.
These latter findings corroborate the hypothesis that cellular ruthenium association occurs in ar eceptor-mediated manner. Overall, Ru-NB targetsE GFR specifically.I mportantly,t he free water-soluble PS exhibits only poor cell binding capacity and lacks cell line selectivity,u ntil their conjugation to targeting moieties. These facts together strongly providet he basis for tumor-specific PDT.

Dark cytotoxicity and phototoxicity of Ru-NB
To evaluate the potencyo ft he bioconjugate Ru-NB as aP DT agent, its cytotoxicity in the dark and upon light irradiation was determined. For these experiments, the A431 cell line had to be chosen due to the strong light sensitivity of the MDA-MB-435S( EGFR negative) cell line that precluded it from phototoxicity studies. To avoid light sensitivity in A431 cell line, irradiation at 480 nm was performed in sequences. 6 3.5 min of irradiation with 15 min gap in between (6.741 Jcm À2 )w ere used. Darkt reatment and surprisingly light irradiation of the A431 cells (48 hi ncubation with Ru-NB) at 480 nm did not cause any cytotoxic effect (IC 50 dark > 25 mm,I C 50 light > 25 mm, see FigureS11)f or Ru-NB.W en ote that we could not go for higher concentration due to conjugate precipitation at 50 mm. Adding polyethylene glycols pacers, changing the ionic strength or the pH could possibly affect the conjugate solubility,a nd consequently help solving this problem. Lack of cytotoxicitye ncouraged us to try to enhancet he internalization of the conjugate into the cells. For that purpose, an additional step was used, namely temperature change. [30] Cells treated with Ru-NB were incubated for 1hat 4 8C. As EGFR internalization is an energy-dependent process, incubating cells at 4 8C inhibits the endocytosisp rocesses but not the binding of Ru-NB to the receptor.Atemperature shift to 37 8C( for 1h)a l-lowed then for efficient endocytosis of the receptor with the boundc onjugate. This step enablesahigher accumulation of Ru-NB in the cells. Due to conjugate precipitation,t he highest concentrationt estedw as 35 mm.R u-NB was again found to be nontoxic in the dark (IC 50 > 35 mm). Unfortunately,l ight irradiation at 480 nm (6 3.5 min with 15 min gapb etween irradiations) again did not cause any phototoxic effect (IC 50 > 35 mm, see FigureS12).

Cellular ROS production by Ru-NB
The lack of phototoxicity of Ru-NB led us to investigate whether this conjugate could produce ROS in irradiated cells. For that purpose, we have stainedA 431 cellsw ith the knownR OS probe DCFH-DA (2',7'-dichlorodihydrofluorescein diacetate). Cells were then treated with Ru-NB (35 mm)u sing the receptor internalization protocol, irradiated (480 nm light for 3.5 min; 1,124 Jcm À2 )a nd suspended in PBSb uffer.T he DCFH-DA signal was detected using flow cytometry instrument. As can be seen in Figure S13, there was no ROS production in the A431 cells that were treatedw ith Ru-NB and then irradiated, as distinct from the H 2 O 2 treatedc ontrol. This unexpected result might be caused by the impairmento ft he internalization of Ru-NB into the cells. Another explanation would be that the ROS produced are directly reactingw ith the NB itself. However, this hypothesis is unlikely,a s 1 O 2 was detectedd uring the 1 O 2 production measurements.

Conclusion
In summary,i nt his article, we presentt he synthesis, characterization as well as photophysical and biological evaluation of a novel nanobody containing Ru II polypyridine conjugate. As a benefito ft he linkaget oa7 C12 nanobody,t he conjugates electively accumulated at the epidermal growth factor receptor (EGFR). The investigation of the uptake via ICP-MS indicated that the conjugate has been successfully internalized inside cancerous A431 cells. Photophysical studies in cuvette suggested that the photophysicalp roperties of the conjugate remain unchanged in comparison with the compound alone. However, DCFH-DAs taininge xperiments indicatedt hat no significant ROS was produced inside the cells. Consequently,p hotocytotoxicityi nvestigationsd id not showa ny significant effect. Focus of future work will be the successful developmento fa nanobody-containing Ru II polypyridine conjugate with ROS and photocytotoxicity inside cancerous cells.
Instrumentation and methods: 1 Ha nd 13 CNMR spectra were recorded on aB ruker 400 MHz NMR spectrometer.E SI-MS experiments were carried out using aL TQ-Orbitrap XL from Thermo Sci- Table 3. Amount of cell-associatedr uthenium after incubation of EGFRpositive A431 with 200 nm of Ru-NB for 24 or 48 hat3 78C. [a] ng Ru per mg protein cetuximab -1 mm 24 h0.77 AE 0.10 < LOD 48 h2.74 AE 0.12 < LOD [a] The level of ruthenium in cell lysates of A431 co-incubatedw ith 1 mm of the EGFR-blocking antibody cetuximab were belowt he limit of detection (LOD).
ChemBioChem 2020, 21,531 -542 www.chembiochem.org entific (Thermo Fisher Scientific, Courtaboeuf, France) and operated in positive ionization mode, with as pray voltage at 3.6 kV.N o sheath and auxiliary gas was used. Applied voltages were 40 and 100 Vf or the ion transfer capillary and the tube lens, respectively. The ion transfer capillary was held at 275 8C. Detection was achieved in the Orbitrap with ar esolution set to 100 000 (at m/z 400) and a m/z range between 150-2000 in profile mode. Spectrum was analyzed using the acquisition software XCalibur 2.1 (Thermo Fisher Scientific). The automatic gain control (AGC) allowed accumulation of up to 2 10 5 ions for FTMS scans, Maximum injection time was set to 300 ms and 1 mscan was acquired. 10 mLw as injected using aT hermo Finnigan Surveyor HPLC system (Thermo Fisher Scientific) with ac ontinuous infusion of methanol at 100 mLmin À1 .F or analytic and preparative HPLC the following system has been used:2 Agilent G1361 1260 Prep Pump system with Agilent G7115A 1260 DAD WR Detector equipped with an Agilent Pursuit XRs 5C18 (Analytic:1 00 ,C 18 5 mm2 50 4.6 mm, Preparative:100 ,C 18
[Ru(phen) 2 (dppz-7-aminomethyl)](PF 6 ) 2 (25 mg, 1.0 equiv) and maleic anhydride (46 mg, 20.0 equiv) were suspended in acetic acid (10 mL) under an itrogen atmosphere. The mixture was held at reflux for 10 h. The solution was then cooled, and as aturated aqueous solution of NH 4 PF 6 was added. The crude product, which precipitated as aP F 6 salt, was collected by filtration and washed three times with H 2 Oa nd Et 2 O. The product was purified by column chromatography on silica gel with aC H 3 CN/aq. KNO 3 (0.4 m)s olution (10:1). The fractions containing the product were united and the solvent was removed. The residue was dissolved in CH 3 CN and undissolved KNO 3 was removed by filtration. The solvent was removed and the product was dissolved in H 2 O. Upon addition of NH 4 PF 6 the product precipitated as aP F 6 salt. The solid was obtained by centrifugation and was washed with H 2 Oa nd Et 2 O. Yield:8 6%.E xperimental data fits with the literature. Purity of the sample was assessed by NMR and HPLC analysis. RP-HPLC: t R = 16.2 min.
Cultivation and expression of recombinant proteins: Freshly transformed E. coli SHuffle T7 Express or E. coli BL21(DE3) harboring the plasmids pET-28b:7C12-Strep-Sortag-His 6 or pGBMCS-SortA were inoculated in 10 mL of LB broth containing 50 mgmL À1 of kanamycin or 100 mgmL À1 of ampicillin, respectively,a nd cultivated at 30 8Co vernight in an orbital shaker with 50 mm offset and shaking speed of 200 rpm. After that, 5mLo ft his pre-culture were transferred into 125 mL MagicMedia E. coli Expression Medium (Life Te chnologies) in 1000 mL baffled-bottom glass flasks and grown at 30 8Cf or 24 h. For final harvest, cultures were chilled on ice for 5min and centrifuged for at least 15 min at 6000 g and 4 8C. After removal of the supernatant, cell pellets were either stored at À20 8Co rs ubjected to purification procedure immediately.
Purification of recombinant proteins: Ah igh-capacity Ni-iminodiacetic acid (IDA) resin in combination with an ¾KTAp ure chromatography system (GE Healthcare) was used for purification of hexahistidine tagged proteins by immobilized metal affinity chromatography (IMAC) under native conditions. Efficient cell lysis was achieved by addition of 1mLR IPAc ell lysis buffer (G-Biosciences) supplemented with EDTA-free protease inhibitor cocktail (Roche Diagnostics), 500 mgl ysozyme (Sigma-Aldrich) and 25 Ue ndonuclease (Thermo Scientific Pierce) per 200 mg bacterial cell pellet. Prior to incubation on ice for at least 15 min, the pelleted cells were resuspended completely by vortexing or pipetting up and down until no cell clumps remained. After centrifugation at 10 000 g and 4 8Cf or 20 min to remove cellular debris, the clarified supernatant was loaded using an automated sample pump with a flow rate of 0.5 mL min À1 .I MAC was performed on ap refilled 5-mL His60 Ni Superflow cartridge (Clontech Laboratories) at af low rate of 5mLmin À1 in equilibration buffer (50 mm Tris·HCl, 150 mm NaCl, pH 7.5). Before elution of the hexahistidine-tagged proteins by addition of 8CVe lution buffer (50 mm Tris·HCl, 150 mm NaCl, 500 mm imidazole, pH 7.5), the column was washed with 8CV equilibration buffer and 7CVw ash buffer (50 mm Tris·HCl, 150 mm NaCl, 35 mm imidazole, pH 7.5). Removal of imidazole and buffer exchange after IMAC was achieved by dialysis against sortase buffer (50 mm Tris·HCl, 150 mm NaCl and 10 mm CaCl 2 ,p H7.5) using ac ellulose ester membrane with am olecular weight cut-off of 3.5-5 kDa (Spectrum Laboratories). Gel electrophoresis: Denaturing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out according to as tandard protocol. [33] For each gel, PageRuler Plus Prestained Protein Ladder (Thermo Fisher Scientific) was used as molecular weight ladder standard. After electrophoresis, gels were imaged with aD -DiGit Gel Scanner (LI-COR Biosciences) and subsequently stained with PageBlue protein staining solution (Thermo Fisher Scientific) according to the manufacturer's instructions.
Protein determination: Protein concentration was determined with the DC Protein Assay (Bio-Rad Laboratories) according to the manufacture'sm icroplate assay protocol using bovine serum albumin in sortase buffer (50 mm Tris·HCl, 150 mm NaCl and 10 mm CaCl 2 ,p H7.5) as protein standard.
Purification of conjugation reactions: In the first purification step, all remaining hexahistidine tagged proteins were eliminated from the reaction mixture by IMAC using prepacked His60 Ni Gravity Columns (Clontech Laboratories). After collection of the flowthrough, the gravity-flow column was washed twice with equilibration buffer (50 mm Tris·HCl, 150 mm NaCl, pH 7.5). These wash fractions as well as the flow-through were analyzed for the presence of the Ru-NB conjugate by SDS-PAGE. Remaining unconjugated [Ru(phen) 2 (dppz-7-maleimidemethyl-S-Cys-(Ser) 2 (Gly) 5 -NH 3 )] 3 + was removed in as econd purification step by size-exclusion chromatography using Zeba Spin Desalting Columns (7 KM WCO, Thermo Scientific) with elution in PBS. The purified conjugate was sterile filtered using Whatman Puradisc FP 30 cellulose acetate syringe filter units with ap ore size of 0.2 mm( GE Healthcare Life Sciences) and stored at 4 8C.
Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry of purified sdAb conjugates: 2,5-Dihydroxyactetophenone (2,5-DHAP,B ruker Daltonik) was used as matrix for MALDI-TOFM S. For solubilization of the matrix, 7.6 mg of 2,5-DHAP were dissolved in 375 mLo fa bsolute ethanol. After this, 125 mLo fa n1 8mgmL À1 aqueous solution of diammonium hydrogen citrate (Sigma-Aldrich) were added. Protein samples were desalted using mixed cellulose esters membrane filters with a pore size of 0.025 mma nd ad iameter of 25 mm (MF-Millipore Membrane Filter VSWP,M erck Chemicals). Briefly,t he filter was placed on the water surface of ab eaker filled with distilled water. A2mLa liquot of the protein sample was carefully pipetted on top of the membrane. After incubation at room temperature for at least 10 min, 2 mLo ft he dialyzed sample was mixed with 2 mLo f 2% TFAs olution. After addition of 2 mLo ft he matrix solution, the mixture was pipetted up and down until the crystallization starts and the solution became cloudy.F inally,0 .5 mLo ft he crystal suspension was spotted onto the ground steel target plate and the droplet was air-dried completely at room temperature.
Spectra were acquired with an autoflex II TOF/TOF (Bruker Daltonik) in positive linear mode in combination with the flexControl software (Version 3.3, Bruker Daltonik) and analyzed with the flex-Analysis software (Version 3.3,Bruker Daltonik). Theoretical molecu-lar weights were calculated using the Compute pI/Mw tool on the ExPASy Server. [34] Spectroscopic measurements: The absorption of the samples was measured in ac uvette with aL ambda 800 UV/VIS Spectrometer (PerkinElmer Instruments) or in 96 well plates with aS pectraMax M2 Spectrometer (Molecular Devices). The emission was measured by irradiation of the sample in fluorescence quartz cuvettes (width 1cm) using aN T342B Nd-YAG pumped optical parametric oscillator (Ekspla) at 450 nm. The luminescence was focused and collected at ar ight angle to the excitation pathway and directed to a Princeton Instruments Acton SP-2300i monochromator.A sadetector aXPI-Max 4C CD camera (Princeton Instruments) was used.
Luminescence quantum yield measurements: For the determination of the luminescence quantum yield, the samples were prepared in aC H 3 CN solution with an absorbance of 0.1 at 450 nm. This solution was irradiated in fluorescence quartz cuvettes (width 1cm) using aN T342B OPO pulse laser Nd-YAG pumped optical parametric oscillator (Ekspla) at 450 nm. The emission signal was focused and collected at ar ight angle to the excitation pathway and directed to aPrinceton Instruments Acton SP-2300i monochromator.A sad etector aX PI-Max 4C CD camera (Princeton Instruments) was used. The luminescence quantum yields were determined by comparison with the reference [Ru(bipy) 3 ]Cl 2 in CH 3 CN (F em = 0.059) [24] applying the following formula [Eq. (1)]: F em = luminescence quantum yield, F = fraction of light absorbed, I = integrated emission intensities, n = refractive index, A = absorbance of the sample at irradiation wavelength.
Lifetime measurements: For the determination of the lifetimes, the samples were prepared in an air saturated and in ad egassed CH 3 CN solution with an absorbance of 0.1 at 450 nm. This solution was irradiated in fluorescence quartz cuvettes (width 1cm) using a NT342B Nd-YAG pumped optical parametric oscillator (Ekspla) at 450 nm. The emission signal was focused and collected at ar ight angle to the excitation pathway and directed to aP rinceton Instruments Acton SP-2300i monochromator.A sadetector aR 928 photomultiplier tube (Hamamatsu) was used.

Singlet oxygen measurements
Direct evaluation: The samples were prepared in an air-saturated DMSO or D 2 Os olution with an absorbance of 0.2 at 450 nm. This solution was irradiated in fluorescence quartz cuvettes (width 1cm) using am ounted M450LP1 LED (Thorlabs) whose irradiation, centered at 450 nm, was focused with aspheric condenser lenses. The intensity of the irradiation was varied using aT -Cube LED Driver (Thorlabs) and measured with an optical power and energy meter.T he emission signal was focused and collected at ar ight angle to the excitation pathway and directed to aP rinceton Instruments Acton SP-2300i monochromator.Along-pass glass filter was placed in front of the monochromator entrance slit to cut off light at wavelengths shorter than 850 nm. The slits for detection were fully open. As ad etector an EO-817 LI R-sensitive liquid nitrogen cooled germanium diode detector (North Coast Scientific Corp.) was used. The singlet oxygen luminescence at 1270 nm was measured by recording spectra from 1100 to 1400 nm. For the data analysis, the singlet oxygen luminescence peaks at different irradiation intensities were integrated. The resulting areas were plotted against the percentage of the irradiation intensity and the slope of the linear regression calculated. The absorbance of the sample was corrected with an absorbance correction factor.A sr eference for the measurement in an CH 3 CN solution phenalenone (F phenaleone = 0.95) [35] and for the measurement in aD 2 Os olution [Ru(bipy) 3 ]Cl 2 (F RuðbipyÞ 3 Cl 2 = 0.22) [36] was used and the singlet oxygen quantum yields were calculated using the following formula [Eq. (2)]: F = singlet oxygen quantum yield, S = slope of the linear regression of the plot of the areas of the singlet oxygen luminescence peaks against the irradiation intensity, I = absorbance correction factor, I 0 = light intensity of the irradiation source, A = absorbance of the sample at irradiation wavelength.
Indirect evaluation: For the measurement in DMSO:T he samples were prepared in an air-saturated DMSO solution containing the complex with an absorbance of 0.2 at the irradiation wavelength and 1,3-diphenylisobenzofuran (DPBF,3 0 mm). For the measurement in PBS buffer:T he samples were prepared in an air-saturated PBS solution containing the complex with an absorbance of 0.2 at the irradiation wavelength, N,N-dimethyl-4-nitrosoaniline aniline (RNO, 20 mm)a nd histidine (10 mm). The samples were irradiated on 96 well plates with an Atlas Photonics LUMOS BIO irradiator for different times. The absorbance of the samples was measured during these time intervals with aS pectraMax M2 Microplate Reader (Molecular Devices). The difference in absorbance (A 0 ÀA)a t 415 nm for the DMSO solution and at 440 nm for the PBS solution was measured and plotted against the irradiation times. From the plot the slope of the linear regression was calculated as well as the absorbance correction factor determined. The singlet oxygen quantum yields were calculated using the same formulas as used for the direct evaluation.
Cell culture: Cell culture flasks, dishes and plates (CELLSTARS) were supplied by Greiner Bio-One GmbH. The adherent human tumor cell lines A431 (ATCC number:C RL-1555) and MDA-MB 435S (ATCC number:H TB-129) were maintained as previously reported. [31,37] All cell lines were confirmed to be mycoplasma-negative using the Venor GeM Advance Mycoplasma Detection Kit (Minerva Biolabs) and were tested monthly.
Cell uptake studies: At otal of 300 000 MDA-MB 435S cells and 450 000 A431 cells were seeded in T25 cell culture flasks in 5mL DMEM supplemented with 10 %f etal calf serum (FCS), respectively, and incubated in ah umidified atmosphere of 95 %a ir/5 %C O 2 at 37 8C. After 48 ho fi ncubation, cells were washed twice with warm PBS. The buffer was then replaced by fresh DMEM supplemented with 10 %F CS and different concentrations of the Ru-NB conjugate or [Ru(bipy) 2 (DPPZ-OMe)](PF 6 ) 2 .F ollowing incubation at 37 8Cf or certain time periods, medium was removed and the cells washed three times with warm PBS and trypsinized. After resuspension in warm DMEM with 10 %F CS, the pellets were collected by centrifugation at 200 g for 5min and washed once with warm PBS. The cell pellets were resuspended in 500 mLo fPBS, lysed by ten freezethaw cycles, and sonicated in an ice-cold ultrasonic bath for 20 min (SONOREX SUPER 10P digital, Bandelin). After determination of the protein content, the lysates were lyophilized on an Alpha 2-4L SC plus (CHRIST).

ICP-MS studies:
After digestion of samples in distilled ultrapure 65 %H NO 3 (Roth) and dilution in 1% HNO 3 ,I CP-MS measurements were performed on an iCap RQ ICP-MS spectrometer (Thermo Fisher Scientific) equipped with aS C-2DX autosampler (ESI). Calibration was done with Ru single element standard (Merck 170347). Rh and Sc were used as internal standards. Limit of detection (LOD) was 50 ng L À1 Ru.
Confocal microscopy: At otal of 100 000 A431 cells were seeded in 35 mm imaging dishes (IBIDI) in 2mLD MEM supplemented with 10 %f etal calf serum (FCS), and incubated in ah umidified atmosphere of 95 %a ir/5 %C O 2 at 37 8C. After 24 ho fi ncubation, media was refreshed and cells were incubated with 100 nm of Ru-NB at 37 8Cf or up to 48 h. Afterward, cells were washed thrice with icecold PBS, fixed with 4% paraformaldehyde and 2.5 %s ucrose in PBS, and permeabilized with 0.25 %T ritonX-100 in PBS for 10 min.
To prevent unspecific antibody binding, cells were incubated with 10 %F CS in PBS overnight at 4 8C. Cells were then incubated with rabbit anti-EGFR (D38B1) Alexa Fluor 647 monoclonal antibody (Cell Signaling Technology) and with StrepMAB-Classic Chromeo 488 conjugate (IBA Lifesciences) for 2h at RT in the dark. Cells were again washed three times with PBS, and the nuclei were stained using Hoechst 33258. Fluorescence microscopy was performed with the Fluoview 1000 confocal laser scanning microscope (Olympus) using a60 (NA 1.35) oil objective.
Dark cytotoxicity and phototoxicity: The dark and light cytotoxicity of the Ru II -containing conjugates was assessed by fluorometric cell viability assay using resazurin (ACROS Organics). For dark and light cytotoxicity with the EGFR internalization step, [38] A431 cells were seeded in triplicates in 96-well plates at ad ensity of 4000 cells per well in 100 mL, 24 hp rior to treatment. Cells were then treated with serum free DMEM medium containing 0.3 %o fB SA for 1h at 37 8C. The medium was then replaced with increasing concentrations of Ru-NB, then cells were incubated on ice for 1h. After that time, cells were transferred for 1hat 37 8C. The medium was then replaced by fresh complete medium. For the dark and light cytotoxicity without the EGFR internalization step, A431 cells were seed in triplicates in 96-well plates at ad ensity of 4000 cells per well in 100 mL, 24 hp rior to treatment. The medium was then replaced with increasing concentrations of Ru-NB for 44 h.
Cells used for the light cytotoxicity experiments with Ru-NB were exposed to 480 nm light for 6 3.5 min with 15 min gap in between irradiations or in a9 6-well plate using aL UMOS-BIO photoreactor (Atlas Photonics). Each well was individually illuminated with a5lm LED at constant current (6.741 Jcm À2 ). After 44 hi nt he incubator,t he medium was replaced by fresh complete medium containing resazurin (0.2 mg mL À1 final concentration). After 4hincubation at 37 8C, the fluorescence signal of the resorufin product was read by SpectraMax M5 microplate reader (l ex = 540 nm, l em = 590 nm). IC 50 values were calculated using GraphPad Prism software.
Cellular ROS production: 10 cm cell culture plates were seeded with A431 cell line and allowed to adhere overnight. Next, the cells were incubated with aD CFH-DA solution (100 mm)i nD MEM media for 30 min at 37 8C. Cells were then washed and treated with serum-free DMEM medium containing 0.3 %o fB SA for 1hat 37 8C.
The medium was then replaced in the plates with either Ru-NB dilution, 0.1 mm H 2 O 2 or medium. Cells were then incubated on ice for 1h.A fter that time, the cells were transferred for 1h at 37 8C.
The medium was then replaced by fresh complete medium. The cells used for the light experiments were exposed to 480 nm light for 3.5 min using aL UMOS-BIO photoreactor (Atlas Photonics; 1.124 Jcm À2 ). All cells were then washed, collected and gated