Proteinase K

High‑resolution crystal structures of a “half sandwich”‑type Ru(II) coordination compound bound to hen egg‑white lysozyme and proteinase K

Abstract

The high-resolution X-ray crystal structures of the adducts formed between the “half sandwich”-type Ru(II) coordination compound [RuII(1,4,7-trithiacyclononane)(ethane-1,2-diamine)Cl]+ and two proteins, namely hen egg-white lysozyme and proteinase K, are presented. The structures unveil that upon reaction with both enzymes the Ru(II) compound is coordinated by solvent-exposed aspartate residues after releasing the chloride ligand (Asp101 in lysozyme, Asp200 and Asp260 in proteinase K), while retaining the two chelating ligands. The adduct with Asp101 residue at the catalytic cleft of lysozyme is accompanied by residue-specific conformational changes to accommodate the Ru(II) fragment, whereas the complexes bound at the two calcium-binding sites of proteinase K revealed minimal structural perturbation of the enzyme. To the best of our knowledge, proteinase K is used here for the first time as a model system of protein metalation and these are the first X-ray crystal structures of protein adducts of a Ru(II) coordination compound that maintains its coordination sphere almost intact upon binding. Our data demonstrate the role of ligands in stabilizing the protein adducts via hydrophobic/aromatic or hydrogen-bonding interactions, as well as their underlying role in the selection of specific sites on the electrostatic potential surface of the enzymes.

Keywords : Ruthenium anticancer compounds · X-ray crystallography · Lysozyme (HEWL) · Proteinase K · Protein ruthenation

Introduction

Data deposition: Atomic coordinates and structure factors have been deposited in the Protein Data Bank, https://www.rcsb.org/ (PDB ID codes 6TVL and 6TXG). which was later replaced by its more soluble sodium salt [5, 6] (Fig. 1), a plethora of ruthenium compounds have been developed and assessed for their antitumor activity. Among them, two classes of organometallic half-sandwich Ru(II)–arene species have shown promising anticancer activity in vitro and in vivo. The first group is termed RAPTA and extensive structure–activity relationship stud- ies revealed that the lead compound is [RuII(η6-p-cymene) Cl2(1,3,5-triaza-7-phosphadamantane)] (RAPTA-C, Fig. 1) [7]. The second group termed RAED is of the gen- eral formula [RuII(η6-arene)(en)X]+ (en = ethane-1,2-di- amine, X = halide) [8], with representative, lead com- pounds the RAED-C (Fig. 1) and RM175 [9, 10]. Contrary to the established Pt(II)-based drugs that share a com- mon pharmacological behavior, even the most structurally similar ruthenium compounds exhibit distinct anticancer profiles. Specifically, NAMI-A has been shown to inhibit metastasis [11], whereas the structurally similar KP1019 and IT139 display cytotoxicity against platinum-resistant primary tumors [6, 12]. The same applies to the organome- tallic Ru(II)-based compounds, with RAPTA-C exhibiting mainly anti-metastatic activity [13, 14], whereas RAEDs are more cytotoxic and active against primary tumors [15]. The exact mechanism of action and the biological target(s) of ruthenium compounds, even of the most thor- oughly studied like NAMI-A and KP1019, remain largely unknown [16]. However, it is commonly accepted that the majority of the ruthenium complexes, similar to cisplatin and its analogues, are prodrugs that are activated by aquation (functional compounds), i.e., replacement of one or more ligands by water molecule(s), possibly preceded by reduc- tion/oxidation [17]. The active species are thus the aquated metabolites that eventually bind covalently to the biological target(s). Early reports suggested that interactions with DNA were ultimately responsible for the anticancer activity of ruthenium compounds. In recent years, though, several stud- ies evidenced that DNA is not necessarily the primary and/or the only target, but interactions with intra- and extra-cellular proteins play a crucial role on the observed pharmacological profile [18–20]. In this context, the low cytotoxicity and the anti-metastatic activity of NAMI-A have been ascribed to its interaction with proteins, due to its fast ligand-exchange kinetics that restrains internalization by cells [21]. On the other hand, it has been demonstrated that the increased cytotoxicity of KP1019 arises from its slow extra-cellular degradation that allows enhanced cellular uptake and inter- action with DNA [22]. Similarly, RAPTA-C has been shown to target proteins selectively, whereas RAED-C displayed a preference for targeting DNA over proteins [23, 24]. It is, therefore, evident that the peripheral ligands have a consid- erable impact on the chemical properties of the ruthenium compounds, which determine their target preferences and consequent mechanisms of action.

To assess the function of the η6-arene ligand of half-sandwich Ru(II) compounds in the context of their anticancer activity, Alessio and co-workers have developed a series of “half sandwich”-type Ru(II) coordination compounds. In these compounds, the aromatic fragment was replaced by a neutral, six-electron donor, face-capping ligand (such as 1,4,7-trithiacyclononane, [9]aneS3), with the remaining coordination sphere unchanged [25–29]. While the major- ity of these Ru(II) coordination compounds did not reveal any significant anti-proliferative activity, [RuII([9]aneS3) (en)Cl]+ (RuTE-Cl, i.e., the coordination counterpart of RAEDs, Fig. 1) showed moderate in vitro cytotoxicity, indicating that the arene ligand can be effectively replaced by other face-capping ligands without total loss of activity [26–29]. Interestingly, RuTE-Cl displays chemical behavior in aqueous solutions and DNA-binding properties similar to those of the RAED group of organometallic compounds [30]. However, the lower cytotoxic activity of RuTE-Cl in comparison to RAEDs has been attributed to its lower lipophilicity ([9]aneS3 vs arene) and to the faster ligand-ca. tenfold faster hydrolysis rate than RAED-C under the same conditions [30].

The nature of protein–ruthenium adducts at the atomic level has been elucidated by crystallographic studies, which highlight the role and importance of the Ru ligands, and provide clues about a general mode of Ru binding [31–33]. With the aim to obtain structural information on the protein- binding mode of these Ru(II) coordination compounds, and towards a better understanding of their mode of action, we studied the crystal structures of the protein adducts formed upon reaction of RuTE-Cl with hen egg-white lysozyme (HEWL) and proteinase K (PK). HEWL (EC 3.2.1.17) is a 129-amino acid residue enzyme (14.3 kDa) and has been extensively employed as a model protein due to its stabil- ity over a wide range of conditions. More importantly, HEWL has been employed in several structural studies of metal-based complexes, including ruthenium anticancer compounds [32, 34–36], thus allowing for direct compari- sons. PK from the mold Tritirachium album Limber (EC 3.4.21.64) is a commonly used, broad-spectrum protease that belongs to the subtilisin peptidase family S8 [37]. It has two binding sites for Ca2+ that are not directly involved in the catalytic mechanism, but which are necessary for stabil- ity and catalytic efficiency of the enzyme. PK is a suitable structural model for the study of protein metalation, con- sidering its stability under varying conditions and the facile preparation of high-quality diffraction crystals. To the best of our knowledge, the structure of the PK adduct presented herein is the only ruthenium adduct of PK in the literature. Our results provide the evidence of ligand-induced confor- mational changes in the catalytic cleft of HEWL, required to accommodate the {RuII([9]aneS3)(en)}2+ fragment (hereafter RuTE), and illustrate the role of residue-specific hydrophobic/aromatic and hydrogen-bonding interactions. Comparison with other Ru(II)–protein adducts suggests a prominent role of the ligands in site-specific interactions that are driven by electrostatic forces on the surface of the enzymes.

Materials and methods

The complex [RuII([9]aneS3)(en)Cl][CF3SO3] (RuTE-Cl) was synthesized as described in the literature [29]. HEWL (L7651) and PK (P2308) were purchased from Sigma and were used without further purification. Crystals of HEWL and PK were grown by vapor diffusion sitting drop method at room temperature and at 289 K, respectively, in 96-well plates (Molecular Dimensions, UK). Lyophilized powder of HEWL was dissolved in water at a final concentration of 20 mg/mL, and 2 μL of the protein solution was mixed with equal parts of reservoir solution containing 0.8–1.5 M sodium chloride and 0.1 M sodium acetate adjusted to pH = 4.5. Tetragonal crystals appeared within the same day and grew to full size within 3 days. Lyophilized powder of PK was suspended in water at a final concentration of 10 mg/mL, and volumes of 2 μL stock protein solution were mixed with equal parts of reservoir solution containing 0.1 M sodium nitrate and 0.1 M of sodium citrate buffer at pH = 6.5. Tetragonal crystals appeared within 3 days and grew to full size within 2 weeks.

Soaking of HEWL and PK crystals with RuTE-Cl was performed by adding solid aliquots to the crystal drops [38]. A significant coloring of the crystals from colorless to yellow was observed within 10 min after the addition of RuTE-Cl with both enzymes. 1 h after the addition of RuTE- Cl, the crystals were captured in silicon loops (Hampton Research, USA) and cryo-protected by rapid immersion in a solution containing the precipitant and 20% (v/v) 1,2-ethan- ediol. Subsequently, the crystals were flash frozen in liquid nitrogen and shipped inside a Taylor–Wharton CX100 dry shipper to the synchrotron facility. Crystallographic data sets were collected at 100 K and 1.00-Å wavelength at the beam- line X06DA of the Swiss Light Synchrotron (PSI, Villigen, Switzerland) for HEWL–RuTE and at 0.98-Å wavelength at the beamline P13 of PETRA III (EMBL, Hamburg, Ger- many) for PK–RuTE. The reflections were integrated using MOSFLM of CCP4 [39] for HEWL–RuTE and XDS [40] for PK–RuTE, the space group was determined using POINTLESS, and data merging was performed using SCALA of CCP4 [39].

The structures of HEWL–RuTE and PK–RuTE com- plexes were solved by molecular replacement with PHASER [41] using the ligand-free structures of HEWL (PDB ID: 193L) [42] and PK (PDB ID: 2ID8) [43] as search models. Both structures were refined with PHENIX [44] using posi- tional refinement and implementing TLS refinement at the final stages. Manual model building and real-space refine- ment were performed with COOT [45], while the geomet- ric restraints of RuTE-Cl (CCDC Accession Code: 265999) [29] were optimized using eLBOW module of PHENIX. The occupancies of RuTE fragments were freely refined in PHENIX using an unrestraint mode. Specifically, RuTE atoms were assigned with the same residue number to be treated as a single chemical species (molecule A). In the PK–RuTE structure, the alternatively present cluster of cal- cium or sodium ions along with their coordinating water molecules were grouped together (molecule B). The occu- pancy of the two species (molecules A and B) was treated as one refinable parameter. The overall geometry of the refined models was very good, with 100% of the residues lying within the Ramachandran favored or allowed regions, while displaying MolProbity [46] scores of 1.56 and 1.24 for HEWL–RuTE and PK–RuTE, respectively (Table 1). The electrostatic potential of the proteins was calculated using APBS with default parameters [47], and mapped onto the solvent-accessible surface area of the proteins at a range of – 1 (red for negative) to + 1 (blue for positive) using PyMOL.

Fig. 2 a Cartoon representation of the X-ray crystal structure of HEWL (orange) with the bound RuTE fragment shown with a stick and ball representation (magenta Ru, cyan C, blue N, and yellow S). Superimposed in green cartoon is the ultrahigh-resolution X-ray structure of ligand-free HEWL (PDB ID: 2VB1). The side-chain of the coordinating residue Asp101 is shown as sticks (orange C and red O). b Refined electron density of RuTE and of the surrounding HEWL residues, contoured at 1σ. Protein residues and RuTE are colored as in (a). c Comparison of the residues that display signifi- cant conformation changes between the ligand-free (green C) and the RuTE-bound (orange C) HEWL. The values given in Å are the dis- tances between the indicated atoms and the dihedral angle in (°) was calculated between the indole planes of Trp62.

Results

The adduct of RuTE with HEWL

Soaking of tetragonal crystals of HEWL with excess of RuTE-Cl resulted in a color change from colorless to yel- low within 10 min and the adduct formation was completed within 1 h. The final model of the complex was resolved at 1.40 Å with an average B value of 28 Å2 over all protein atoms. All protein atoms were clearly resolved, with the Cα atoms displaying an average root-mean-square devia- tion (RMSD) of 0.48 Å from the ultrahigh-resolution crys- tal structure of ligand-free HEWL (PDB ID: 2VB1) [48]. Backbone and side-chain differences (see below and also Fig. S1) were found mainly at residues 46–47, 70–73, and 99–103 (Fig. 2a and discussion below). The electron density of the RuTE fragment was unambiguously identified (at 90% occupancy and average B value of 31 Å2) bound to the car- boxylate side-chain of Asp101 (Fig. 2a). The complex was refined without restraints and the Ru–O bond was found at a length of 1.84 Å, which is relatively shorter compared to the typical range of Ru–Obond lengths of 2.0–2.6 Å (from RuII coordinated by Asp residues of HEWL, see the Discussion section). Both Ru–N bond lengths were 2.04 Å and the three Ru–S bond lengths were in the range of 2.19–2.24 Å. The geometry around the metal ion is mostly octahedral, with all L–Ru–L angles deviating less than 8° from the orthogo- nal values, except for the equatorial O–Ru–S angle of 163° (Fig. 2b). At this binding site, the RuTE ligands are sand- wiched between Asn103 and Trp62. In particular, one of the N atoms of the en ligand is at 3.55 Å from the side-chain NH2 amide of Asn103 and is hydrogen bonded to a solvent molecule, whereas the other N atom is at 3.43 Å from the backbone C=O of Asp101. On the other side of the complex, the [9]aneS3 ligand is accommodated within an aromatic, hydrophobic pocket/patch comprised of Trp62, Trp63, and Leu75 (Fig. 2b). All methylene groups of [9]aneS3 are engaged in CH2–aromatic interactions with the indole ring of Trp62 (within 3.5–5.0 Å distance). Two of the [9]aneS3 methylene groups also interact with the side-chain of Trp63, one of which displays an additional hydrophobic interaction with the Cδ1 methyl group of Leu75.

Interestingly, comparison of the HEWL–RuTE adduct with the ultrahigh-resolution (0.65 Å) structure of ligand- free HEWL [48] revealed that binding of the complex induces significant conformational changes (Fig. 2c). In particular, the side-chains of Asp101 and Asn103 move outwards to coordinate Ru(II) and accommodate the en ligand, respectively, with a concomitant main-chain shift of residues 99–103 (RMSDCα of 1.4–2.8 Å with respect to the ligand-free HEWL). Moreover, the side-chain of Trp62 is rotated by more than 90° to accommodate the [9]aneS3 ligand of RuTE, as in the case of HEWL with a bound sub- strate model [49]. Trp62, which has been shown to play a key role in substrate binding, usually displays poor electron density in the crystal structures of the native enzyme due to its flexibility. In contrast, RuTE binding and stacking of the [9]aneS3 against Trp62 and Trp63 results in stabilization of the two residues, as evidenced by their strong electron density. It should be noted that in the ultrahigh-resolution HEWL structure, Trp62 was found in a double conforma- tion with an angle of 25° between the two indole planes [48]. The conformational change of Trp62 in our structure is also accompanied by a significant side-chain motion of Arg73 (Fig. 2c).

The adduct of RuTE with PK

The adduct of PK with RuTE was formed within 1 h by soaking crystals of PK in excess of RuTE-Cl. The structure was resolved at 1.37 Å with an average B value of 15 Å2 over all atoms (13 Å2 protein only atoms), which is almost identical to the atomic resolution X-ray structures of ligand- free PK (PDB IDs: 1IC6 [50] and 5KXV [51]), displaying Cα atom RMSDs of less than 0.1 Å (Fig. S2). The extra- electron density found at the two Ca2+-binding sites was assigned to two RuTE fragments bound to Asp200 at site 1 (0.54 occupancy and average B = 21 Å2) and Asp260 at site 2 (0.39 occupancy and average B = 24 Å2), as shown in Fig. 3. The remaining density was assigned to a Ca2+ ion and two water molecules at site 1 (occupancy of 0.46) and a Na+ ion and three water molecules at site 2 (occupancy of 0.61). The residual electron density of site 2, which is known to be weak Ca2+-binding site, was also recognized as a sodium ion on the basis of its coordination geometry [52].

RuTE at site 2 is coordinated to Oδ1 of Asp260 (Ru–O bond unrestrained refined at a length of 2.25 Å), while Oδ2 of Asp260 was found at a distance of 2.60 Å from one NH2 (Fig. 3a, b). The same NH2 is hydrogen-bonded to the backbone carbonyl of Thr16, whereas the other N of en is at a distance of 3.6 Å from the carbonyl of Lys258. The [9]aneS3 ligand of RuTE (site 2) is mainly exposed to the solvent. As a result, while the sulfur atoms of the ligand are well localized in geometrically ideal positions, the bridging ethylene groups are only weakly resolved in the unbiased Fo–Fc difference maps (Fig S3c). RuTE at site 1 is bound to the carboxylate group of Asp200, with a Ru–Obond length of 2.05 Å and an equatorial O–Ru–S angle of 176° (Fig. 3d). One of the NH2 groups of en is stabilized by two hydrogen bonds with the main-chain carbonyl groups of Pro175 and Val177, whereas the other is hydrogen-bonded to a solvent molecule only (Fig. 3a). The [9]aneS3 ligand of RuTE (site 1) exhibits only a few contacts with the side-chains of Pro175 and Val269 (at 4.1–4.6 Å). Still, these contacts stabilize the ligand enough to be sufficiently resolved in the Fo–Fc difference maps (Fig. S3b). The Ru–Obond lengths in the protein adducts were determined without using any restraints and were found to be within the typical range (2.05 Å in RuTE at site 1 and 2.25 Å at site 2).

The two RuTE adducts are bound to PK without causing any significant conformational change to the Ca2+-binding site residues (Fig. S4). At site 1, the Ru(II)-bound side- chain of Asp200 is practically at the same position as in the ligand-free enzyme, where the position of Ca2+ is occu- pied by one of the N atoms of the en ligand. The alternative occupancy (0.46) was attributed to a Ca2+ and two water molecules (Fig. 3e). Similarly, binding of RuTE at site 2 mediates a minor movement of Asp260 and the displace- ment of the four Ca2+-bound water molecules, with the N1 of en close to the position of the calcium cation. However, the alternative occupancy of site 2 (0.61) was assigned to a Na+ ion, which is coordinated to Asp260 and three solvent molecules in a trigonal pyramidal arrangement (Fig. 3c). These observations are in accordance with the previous studies, which showed that one calcium ion binds tightly (pK 7.6 × 10−8 M−1) and the other binds only weakly [53], whereas a sodium ion can occupy a PK calcium-binding site under Ca2+-free conditions [54]. Removal of the cal- cium ions (e.g., using EDTA) results in ~ 80% decrease of PK activity within 3 h, which has been attributed to subtle structural changes of the enzyme and of its hydration prop- erties [53].

Discussion

Recent studies of two Ru(II)-arene compounds have demon- strated that RAPTA-C and RAED-C (Fig. 1) target chroma- tin with different site selectivity [23, 24]. While RAPTA-C is primarily associated with the histone proteins, RAED-C preferentially targets the DNA component of chromatin, albeit sharing some common histone-binding sites. The dif- ferent selectivity of RAPTA-C and RAED-C was mainly attributed to steric constraints of the bulky phosphaadaman- tane ligand of RAPTA-C within DNA, and this has been correlated with their distinct cytotoxicity profiles. While RAED-C exhibits high cytotoxicity and activity against pri- mary tumors that is comparable to cisplatin, RAPTA-C dis- plays low cytotoxicity, but has been shown to have anti-angi- ogenic and anti-metastatic properties [13, 14]. Therefore, it has been suggested that modifications in ligand structure or coordination ability of Ru(II) compounds can modulate their DNA/protein-targeting activity, and thus their cytotoxic potential. With the aim to provide structural evidence on the role of the ligands in protein binding, we employed a Ru(II) coordination compound, RuTE-Cl, in which the arene fragment of the RAEDs is substituted by the face-capping
[9]aneS3 ligand (Fig. 1).

Treatment of HEWL crystals with RuTE-Cl revealed a single adduct at the side-chain of Asp101 (Fig. 2). Upon hydrolysis of the chloride ligand, the positively (+2e) charged RuTE species binds to the carboxylate group of Asp101. Binding of RuTE mediates a conformational
change at the catalytic cleft, similar to that observed in the substrate-receptive conformation of HEWL (Fig. 2c). The en ligand is accommodated by a shift in the side-chain of Asn103, whereas the [9]aneS3 ligand is stacked over the indole ring of Trp62, exhibiting several methylene–aromatic interactions, and is also in contact with Trp63 and Leu75 (Fig. 2b). An examination of the electrostatic potential sur- face of HEWL reveals that although the enzyme comprises an overall positively charged surface, the site of ruthenation by RuTE occurs at the basic site of Asp101, which is part of the catalytic cleft comprising also the active site residues Glu35 and Asp52 (Fig. 4).

For the purpose of comparison with related struc- tures of Ru(II)–HEWL adducts, the first crystal struc- ture of a Ru(II)–arene adduct revealed that the neutral [RuII(η6-p-cymene)Cl2(H2O)] species forms a single adduct that is coordinated by Nε1 of His15, the only histidine res- idue of HEWL [34]. At this site, the neutral Ru(II) fragment is bound at a positively charged patch (Fig. 4), with the arene group displaying contacts with the side-chains of Lys14 and Asp87. It should be noted, however, that ICP- MS analysis has indicated more than one ruthenation sites on HEWL [34]. A more recent crystallographic study of HEWL metalation by dinuclear organometallic complexes of the formula [M(η6-p-cymene)X2]2 [M = Ru(II) or Os(II), X = halide] has revealed a selective interaction of the neu- tral {M(η6-p-cymene)Cl2} fragment with His15, which was independent of the metal center or the halide [32]. Inter- estingly, a second binding site was found, where the posi- tively charged {(η6-p-cymene)Ru(μ-Cl2)Ru}3+ fragment was coordinated by Asp101 that bridged the two Ru centers. Although this binding site was suggested to be weaker, as the complex was refined with 0.5 occupancy, this observa- tion provides evidence that most likely the ligand-induced charge of the Ru fragments can modulate the selectivity of the metalation site. In a later work, the same group presented the first X-ray structure of a Ru–carbene adduct of HEWL using [RuII(η6-p-cymene)(dmb)Cl2] (dmb = 1,3-dimethylb- enzimidazol-2-ylidene), where three distinct binding sites were identified [33]. In the first site, {RuCl2(dmb)(OHx)} was coordinated to both side-chains of Arg14 and His15, whereas at the second site, an adduct of {RuCl2(dmb) (OHx)2} with the side-chain amine of Lys33 was formed. A third site was identified with a {RuCl(OHx)4} fragment displaying weak interactions with the carbonyl oxygen of Ala107 [33]. Although the overall charge of the Ru adducts could not be unambiguously determined (the metal center has been shown to be oxidized upon exchange of the arene ligand, and the nature of the coordinated H2O vs OH− has not been determined), and thus, no correlation between frag- ment’s charge and site selectivity can be extracted; these results clearly show that different fragments target diverse sites.

Similarly, HEWL adducts with carbon monoxide-releas- ing compounds have revealed multiple ruthenation sites, with His15 recognized as the primary site. In particular, soaking of HEWL with fac-[RuII(CO)3Cl(κ2-H2NCH2CO2)] resulted in three adducts of the cis-{RuII(CO)2(H2O)3}2+ fragment bound to His15, Asp18, and Asp52 with occupan- cies of 0.8, 0.5, and 0.4, respectively [55]. In a subsequent work, reaction of HEWL with a series of ruthenium car- bonyl complexes of the general formula fac-[RuII(CO)3Cl2L] (L = neutral monodentate ligand) revealed multiple ruthena- tion sites, still exhibiting His15 as the primary site of inter- action (occupancies of 0.7–1.0), and with Asp18, Asp52, Asp101, and Asp119 identified as secondary sites (occu- pancies of 0.4–0.7) [56]. A more recent crystallographic study of Ru(II) carbonyl complexes of the same general formula, where L = N3-imidazole or N3-methyl-imidazole [57], revealed that upon dissociation of the azole ligand, the two chlorides and one or two CO molecules, the com- plexes bind HEWL at His or Asp residues (His15, Asp18, Asn46 and Asp52, Asp119 and Arg125, or the C-terminus of Leu129) [58]. For all the HEWL adducts studied, His15 was identified as the highest occupancy ruthenation site, followed by Asp18 (a detailed table can be found in the supplementary information of ref. 58). On the other hand, reaction of the mixed-valence diruthenium(II,III) complex [Ru2(μ‐O2CCH3)4Cl] with HEWL revealed two adducts of the {Ru2(μ‐O2CCH3)2}3+ fragment bound to Asp101 and Asp119, with their carboxylate groups bridging the two metal centers [59].

The above observations, taken together with an examina- tion of the ruthenation sites on the electrostatic potential surface of HEWL (Fig. 4), suggest that neutral species, such as the {RuII(η6-p-cymene)Cl2} fragments, are preferentially coordinated by the heterocyclic N atom of His, which is located at a positively charged site. On the other hand, posi- tively charged species, such as RuTE and the diruthenium organometallic compounds, display a preference for the O atom donors of aspartate residues that are located at nega- tively charged sites on the surface of HEWL. In cases where positively charged Ru(II) species were coordinated by His15, such as for the {RuII(CO)x} adducts, alternative ruthenation sites were identified at negatively charged aspartate residues (Fig. 4). Interestingly, from the plethora of potential ruthena- tion sites identified on HEWL, RuTE displays a strong pref- erence for Asp101 under the conditions studied.
These observations are further supported by the second crystal structure of the unrelated enzyme PK (Fig. 3), in which the ruthenation sites coincide with the two calcium- binding sites of the enzyme. Our data show that the two RuTE fragments are coordinated by the carboxylate groups of Asp200 and Asp260, without causing any significant con- formational change of PK. The two RuTE adducts are further stabilized through hydrogen-bonding interactions between the en ligands and main-chain atoms from the surrounding residues (Fig. 3b). Examination of the electrostatic potential surface of PK reveals that the two Ca2+-binding sites are regions of negatively charged electrostatic potential (Fig. 5). In particular, the well-defined site 1 where the highest occu- pancy RuTE adduct was identified, displays also the highest negatively charged potential. In the second, less well-defined Ca2+-binding site, the RuTE adduct was resolved at a lower fraction of the total occupancy. While we cannot make a direct comparison with the other X-ray structures, as this is the first crystal structure of a ruthenated PK adduct, our observations are in accordance with the results obtained from the crystallographic studies of Ru(II)–arene adducts with the nucleosome core particle [23, 24]. Albeit their dif- ferent DNA/protein selectivity that has been attributed to the steric demands of the bulkier ligands, RAPTA-C and RAED-C (Fig. 1) have been shown to share a single common binding site on the histone proteins. At this site (designated as Site 2 in Fig. S5), the positively charged Ru(II)–arene fragments were coordinated by Glu carboxylate groups within a highly acidic cleft formed by the histone H2A–H2B dimer. Even though the coordinating ligands of RAPTA-C and RAED-C are different at the variable sites of interac- tion with chromatin, still, the electrostatic forces appear as the steering force of the positively charged species towards regions of negatively charged potential, as in the case of the PK–RuTE adducts presented here (Fig. 5).

Conclusions

While the previous studies have investigated the protein ruthenation by means of organometallic half-sandwich Ru(II) compounds with arene ligands, the present study concentrates on a coordination analogue to RAED, namely [RuII([9]aneS3)(en)Cl]+ (RuTE-Cl), that bears a face-cap- ping crown thioether ligand (i.e., [9]aneS3) in place of the arene. In this sense, we provide the first structural insights.

Fig. 5 Solvent accessible surface of the HEWL–RuTE crystal structure colored by electrostatic potential. The two RuTE adducts, a triflate counter ion (OTf), two ethylene glycol molecules (EDO), and three nitrate ions are shown as ball- and-sticks and are colored as in Fig. 2. The two Ca2+-binding sites and the active site of the enzyme are indicated for the protein adducts formed by this “half sandwich”-type Ru(II) coordination compound with two model proteins (i.e., HEWL and PK). These X-ray structures are also the first examples, to the best of our knowledge, of adducts between proteins and Ru(II) coordination compounds that maintain their coordination sphere almost intact. Typically, Ru was found as a naked ion when coordination compounds were used (e.g., NAMI-A and KP1019). In the HEWL adduct, RuTE is coordinated by Asp101 at the side of the catalytic cavity, whereas in the PK adduct, RuTE is coordinated at the two calcium-binding sites of the enzyme. In the case of HEWL, conformational rearrangements of the surrounding residues were detected, whereas the two RuTE fragments bind at PK without causing any conformational change. The HEWL–RuTE adduct is stabilized mainly via hydrophobic/ aromatic interactions of the [9]aneS3 ligand, whereas the PK–RuTE adducts display hydrogen-bonding interactions of the en ligand with main-chain residues. Taken together, our analysis suggests that although the hydrophobic forces underpin the dominant interactions between the ligands and protein residues, in accordance with many structural stud- ies of protein–ruthenium adducts, the overall charge of the complex, dictated by its coordination sphere, underlies the steering force towards electrostatically favorable regions on protein surfaces.