An Insight into the Anticancer Activities of Ru(II)-Based Metallocompounds Using Docking Methods

Unlike organic molecules, reports on docking of metal complexes are very few; mainly due to the inadequacy of force fields in docking packages to appropriately characterize the metal atoms that consequentially hinder the rational design of metal-based drug complexes. In this study we have made used Molegro and Autodock to predict the anticancer activities of selected Ru(II) complexes against twelve anticancer targets. We observed that introducing the quantum calculated atomic charges of the optimized geometries significantly improved the docking predictions of these anticancer metallocompounds. Despite several limitations in the docking of metal-based complexes, we obtained results that are highly correlated with the available experimental results. Most of our newly proposed metallocompounds are found theoretically to be better anticancer metallocompounds than all the experimentally proposed RAPTA complexes. An interesting features of a strong interactions of new modeled of metallocompounds against the two base edges of DNA strands suggest similar mechanisms of anticancer activities similar to that of cisplatin. There is possibility of covalent bonding between the metal center of the metallocompounds and the residues of the receptors DNA-1, DNA-2, HDAC7, HIS and RNR. However, the general results suggest the possibility of metals positioning the coordinated ligands in the right position for optimal receptor interactions and synergistic effects, rather than forming covalent bonds.


Introduction
There have been several ruthenium-based complexes synthesized as anticancer drugs [1][2][3] to serve as alternative to cisplatin, which is the most widely used and efficient anticancer agent [1,4]. One of the major limitations in the rational design of organometallic anticancer complexes is the lack of proper knowledge about their anticancer targets [5][6][7][8][9]. In our research group we have been making efforts to predict the possible targets of metal-based complexes as anticancer agents using theoretical docking methods. However, unlike their organic counterparts, there is a very serious limitation in making use of docking tools for metal complexes, mainly due to lack of proper force fields to accommodate the metal centre [10]. Even in the quantum computation field, the volume of works on metal-based complexes are very low compared to organic compounds due to the high computational demand and difficulties of finding an appropriate method for their optimization [11].
In our previous work [12,13], we have made used of docking packages like Glide, Gold and Autodock whereby some interesting results were obtained. In this work we have made use of another docking package called Molegro, which recognizes the ruthenium atom as a metal centre better than Glide, Gold and Autodock. Also, we have enhanced its performance including that of Autodock by introducing the atomic charges of each metallocompound obtained from the quantum optimized structures of the compounds. Other new metallocompounds are designed which we have never reported before and the binding of co-crystallized compounds that accompanied each of the receptors is set as references in this computational docking. In addition to the number of receptors considered in our previous works [12,13], we have also included DNA as one of the targets to see the possibility of some of these metallocompounds having favourable interaction properties like that of cisplatin. Other receptors considered in these work besides DNA are thioredoxin reductase (TrxR) [14], histone protein in a nucleosome core particle (HP-NCP) [15], BRAF Kinase [16], recombinant human albumin (rHA) [17,18], thymidylate synthase (TS) [19], ribonucleotide reductase (RNR) [20], histone deacetylase (HDAC7) [20], cathepsin B (CatB) [14], topoisomerase II (TopII) [21,22] and DNA gyrase. All these have been reported to play significant roles in cancer growth and metastasis, except DNA gyrase, which is a notable bacterial enzyme [23,24], considered in this project for the purpose of seeing if some of these metallocompounds can play a dual role.

General Features of the Binding Activities of the Metallocompounds
In this paper the binding affinities and conformations of 30 metallocompounds including their hydrated forms (Figure 1) are reported. The 30 metallocompounds are docked against twelve receptors using the relaxed Molegro, constrained Molegro and Autodock docking approaches. The summary of 72 poses of the best two metallocompounds in their interaction with the receptors are shown in Table 1, indicating the hydrogen bonding (HB) and the possible metal-receptor (MR) interactions with the receptor binding site residues. The results obtained from the docking of metallocompounds to twelve receptors using Autodock, relaxed Molegro without metal-residue covalent constraints and constrained Molegro with covalent constraints are shown in Tables 2 and 3, 4 and 5, and 6 and 7, respectively. The docking results obtained from both relaxed and constrained Molegro docking (Tables 4-7) give a better ranking than the Autodock which cannot not be easily ranked since many of the Autodock values (Tables 2 and 3) are within its standard error margin of ~2.177 kcal/mol [25,26]. Tables 2, 4 and 6 contain  binding affinities of non-hydrated complexes, while Tables 3, 5 and 7 contain the affinities of their respective hydrated complexes. The hydrated complexes are named after their respective non-hydrated type with the suffix "a".  The PP = 4-Phenylpyridine, RA = Ruthenium-(6-arene); etdia = ethylenediamine, phen = phenanthroline; pyr = pyridine; pyraz = pyrazole, pta = 1,3,5-triaza-7-phosphaadamantane; Cp = cyclopentadienyl, Bk = β-diketone, piano = model of piano-stool.
The proposed mechanism of activation of the metallocompounds through hydration [2,18,27,28] is further confirmed through the docking results. In most of the receptor interactions, Autodock indicates an enhanced binding energy with receptors for all the hydrated complexes (Tables 4 and 5). The same  thing is observed for the covalent constrained docking of metallocompounds to DNA-1,DNA-2, CatB,  RNR and TrXR and also for relaxed Molegro docking to DNA-1 and DNA-2 (Tables 2-5).In other receptors, only a few hydrated metallocompounds increase their binding energy as a result of the hydration and this depends on the type of the receptor targets. The binding activities of the hydrated forms of RAPTA complexes were greatly enhanced. Also, the bidentate form of RAPTA complexes are suggested to competitively bind strongly to receptors as the hydrated forms of RAPTA complexes, especially metallocompound 14a, though their activities are still lower than the modeled metallocompounds using the results of the three methods of docking (Tables 2,3; 4,5; 6,7). In many of the receptor interactions, metallocompounds 16 and 16a are predicted to have better binding affinity than 15 and 15a (Tables 4,5) except for receptors RNR which further supports the experimental report of higher anticancer activities of the first pair [27].
In many instances, the binding of the relaxed Molegro docking are stronger than the constrained Molegro docking, also the relaxed Molegro agrees better with the available experimental reports (Tables 8,9) and the experimentally proposed activation of metallocompounds by hydration (Tables  4,5 and 6,7). This is an indication that the metal center will give preference to holding the coordinated ligands in appropriate position for optimal receptor residues interaction and synergistic effect than forming covalent bonding. In the absence of covalent constraints, the features of the affinities of the metallocompounds to receptors depends greatly on the type of metallocompounds from the results of strong binding interactions of 1 and 1a, 8 and 8a, 9 and 9a, 10 and 10a, 11 and 12 to most of the receptors. Even though the ranking of the metallocompounds against the receptors in the covalent constrained docking is similar to that of the unconstrained (Tables 4,5 and 6,7) type, yet the features of the metallocompounds' affinity depend greatly on the type of receptors.
All the docking methods clearly indicate that the best target of the hydrated cisplatin (Cisp-W1 and Cisp-W2) is DNA as they are found having higher binding energy with DNA compare to other receptors. The hydrated cisplatin, especially the doubly hydrated form (Cisp-W2) has the highest DNA binding compared to the other forms of cisplatin which further confirmed the need for hydration before activation [3,28]. However, in protein receptors, there is decrease in binding affinity of hydrated cisplatin compare to its non-hydrated type that further confirms that the protein is not its target. Considering the ranking of both relaxed and constrained Molegro docking, the best targets of the metallocomplexes follow the order HDAC7 > DNA-1 > rHA > CatB > DNA-2 > Gyrase > TrXR > TopII > TS > RNR > HiS > Kinase except in few cases like complex 12 that prefers CatB to rHA. The poorest binding affinity of the complexes is observed in the covalent constrained docking to rHA receptor. This suggest that the interaction of the metallocompounds with rHA as a transport receptor that is responsible for the pharmacokinetic availability of a wide range of drugs, including metallodrugs and consequentially determine their bioavailability and toxicology [29] is better described in terms of other interacting forces rather than covalent interaction of ruthenium metal with its residues.

The Interacting Poses of the Best Two Metallocompounds in the Receptor Binding Sites
To have a better understanding of the binding site orientation poses and interaction of the metallocompounds with the residues of each receptor, we have selected two metallocompounds predicted by the relaxed Molegro, covalent constrained Molegro and Autodock methods to have the best binding to each of the receptors ( Table 1). The binding site interactions are elaborated in terms of the hydrogen bonds (HB) and the metal-residues (MR). The MR considers possible covalent bonding interactions of the metal with a nucleophilic Centre of receptor residue, which was considered during the constrained docking. There is a possibility of metal-residue (MR) interaction of complex 4a with DNA-1 according to Autodock prediction ( Table 1). The relaxed Molegro and Autodock dockings suggest a different binding site from that of constrained docking in the binding of the metallocompound to DNA-1 ( Figure 2a). However, in DNA-2 both relaxed and Autodock dockings suggest similar binding sites to that of covalent constrained docking for many of the metallocompound interactions. Interestingly, the suggested conformation of complex 12 binding to DNA-2 from both the relaxed and covalent constrained dockings is perfectly superimposed (Figure 2b). There is also an interesting feature of possible intercalation of complex 12 with the two based pair nitrogen atoms (N7) of the guanine in DNA-2 by both relaxed and constrained docking as shown in Figure 2b. Ranked among the best binding metallocompounds against the DNA-1 and DNA-2 are complexes 10a and 12 by both the relaxed and covalent constrained docking. Autodock also predicted complex 10a among the best five inhibitors of DNA within the interval of 0.60 Kcal/mol. The features of complex 10a in Autodock docking to CatB shows a very strong H-bond interaction of one of its carboxylic groups with the SG of the CYS 29, which supposed to be, center of metal-residue covalent interaction (Table 1).
When covalent constraints were applied on the metallocompounds' interaction with CatB, the possibility of metallocompound 12 forming HB with the receptor was traded for the formation of covalent bonds (Table 1) which subsequently led to a lower activity (comparing metallocompound 12 in Tables 4,5 and 6,7). The same thing was observed for the interaction of metallocompound 1 with the TrXR receptor. Also, the tridentate ligand part of metallocompound 12 in the relaxed Molegro docking (cyan) and Autodock (magenta) fits properly into the hydrophobic pocket of the CatB as shown in Figure 2c, which was not the case in the constrained docking (green). This further suggest the possible reason why the binding affinities of the relax docking are higher than that of the constrained docking. The three methods of docking predicted 12 as one of the best binding metallocompounds to CatB. Also, the orientation and the interactions suggested Autodock and relaxed Molegro docking are the same with the structure of 12 on the binding site almost superimposed (Figure 2c).
Both relaxed and constrained Molegro docking show a superimposed conformation for complex 10a in which the arene unit is pointing toward the mouth of the inner pocket of the HDAC7 while the Autodock preferentially locates its outer pocket (Figure 2d). Complex 1a is rated to have the best binding affinity to HP-NCP in relaxed Molegro docking but it has no visible H-bond or metal-receptor interaction. The same is observed for Autodock docking of complex 5a to BRAF Kinase and relaxed Molegro docking of complex 10 to TopII. These suggest the significant importance of other interacting forces like Van der Waal and close contact interactions. The ruthenium metal appears to have interaction with the oxygen atom of residue GLU 61G of the HP-NCP from the Autodock prediction and constrained docking. The conformation of complex 1a as suggested by the relaxed Molegro docking fits in totally into one side of the binding site pockets than the conformation suggested by the constrained and Autodock docking to HP-NCP. In Figure 2e, the experimental conformation from the crystal structure of RAPTA complex 14a in the HP-NCP binding site is compared to the conformation obtained from the docking results. This complex was rated best among the RAPTA complexes by Autodock; it is the fourth of the RAPTA complexes from relaxed Molegro while it is ranked low by constrained Molegro. However, the orientation suggested by the covalent constrained docking agrees strongly with one of the experimental orientations as shown in Figure 2e. Also the Autodock locates the same binding sites in agreement with the crystal structure but associated with the reverse orientation of complex 14a while relaxed Molgro located only one of the pockets suggested by the crystal structure. In Autodock docking of complex 10 to rHA we observed possible covalent interactions of ruthenium with the NH 3 unit of the LYS 195A residue characterized by a distance of 1.78 Å ( Table 1). The three methods of docking suggested either the hydrated complex 10a or non-hydrated complex 10 as the best binding metallocompound to rHA. They all predicted complex 1 as the best for RNR. The suggested conformation for the non-hydrated and hydrated complexes 1 and 1a, respectively, by constrained docking in RNR is almost superimposed with a little variation in the orientation of their arene unit. The conformation suggested for the docking of RAPTA complex 3 to RNR is shown in Figure 2f; both Autodock and Molegro suggest a very close position for the ruthenium atom but a different position for its coordinated ligands. In the constrained docking of complexes 9a and 8a to TrXR, there is possibility of the ruthenium atom forming a covalent bond with the bridged surphur (SG) residue CYS 64A or strong close contact van der Waal interactions with the two bridged surphur (SG) atoms of the residues CYS 64A and CYS 59A.

The Binding Affinity of the Metallocompounds to DNA
An insight into the interaction of the metallocompounds with the two types of DNA (DNA-1 and DNA-2) is given for the Autodock (Tables 2,3), relaxed Molegro (Tables 4,5) and constrained Molegro (Tables 6,7) models. There is improved in the binding of many of the complexes to DNA-1 and DNA-2 due to their hydration. Typical examples are complexes 1a, 8a, 9a, 10a and all the hydrated RAPTA complexes. In agreement with the literature, the best target of cisplatin is DNA [3,28] as the binding affinity decreases significantly in protein receptors compared to DNA-1 and DNA-2. The features of cisplatin in all the dockings shows that its activity significantly increases when it is doubly hydrated (Cisp-W2) than when it is singly hydrated (Cisp-W1). The newly proposed model of complexes appears to have strong DNA interactions that suggest DNA as part of their possible targets. Most of the RAPTA complexes will target CatB or HDAC7 preferentially to DNA. The spectrum of the interactions of RAPTA complexes with DNA range from −116.96 to −49.53, with CatB from −124.88 to −60.05 and with HDAC7 from −127.16 to −55.13 considering both the realaxed and constrained docking which further confirm thats RAPTA complexes will target proteins preferably to DNA [14,15].

The Binding Affinities of the Metallocompounds to CatB and TrxR
The results obtained from relaxed Molegro and Autodock docking (Tables 2-5) show that 12, 1 and 1a, 10 and 10a, 9 and 9a and 11 bind strongly to CatB than its co-crystallized compound. Both Molegro and Autodock included the hydrated RAPTA complexes as part of the best binding metallocompounds to CatB while only Molegro ranked bidentate RAPTA complexes 2 and 3 better than the hydrated ones. In ranking, we give preference to the Molegro dockings than the Autodock ones, especially where there is a very close binding affinity prediction from the Autodock docking considering the standard error margin of ~2.177 kcal/mol [25,26] which is an indication that metallocompounds within this range of binding can be reordered. In the binding affinities of the metallocompounds to TrXR, the new models are still rated best followed by the bidentate RAPTA complexes (compounds 3 and 2) and the hydrated RAPTA complexes. The features of the predicted binding affinities of RAPTA complexes give a better insight into their reported experimental activities in CatB and TrXR. The experimental activities of the RAPTA complexes against CatB follow the order 15, 14, 2, while 3 and 13 rarely show any appreciable binding affinity and that of TrXR followed the order 2, 3, 14 and 15 while 13 has the lowest activity (Table 8) [14]. Selecting only the metallocompounds that have experimental values, the predicted binding affinities of the metallocompounds against TrXR using Molegro agree better with the experimental order of their inhibitory activities as metallocompound 13 is rated least in activities toward TrXR followed by metallocompound 15. Contrary to the experimental report, metallocompound 3 was predicted to bind strongly to CatB. However, the ranking of metallocompounds 14a and 15a among those that have the best binding to CatB and metallocompound 13 rated least agrees well with the experiment ( Table 8). The rating of both hydrated and non-hydrated forms of metallocompound 14a among those that bind strongly to CatB, further give insight into their experimental finding as the best anticancer agents among the RAPTA complexes [5]. The three docking methods show that the interaction of the metallocompounds with CatB is stronger than with TrXR, which agrees well with the experimental report [14]. The binding affinities of the hydrated metallocompounds with CatB and TrXR are significantly enhanced (Tables 2-7)

The Binding Affinity of the Metallocompounds with Gyrase, HDAC7, HP-NCP, BRAF Kinase, rHA, RNR, TopII and TS
The features of the binding affinities of the metallocompounds against gyrase from both docking methods show that co-crystallized compound still preferentially bind to gyrase better than any of the metallocompounds, even though the activities of the new modeled metallocompounds are competitively close to that of its co-crystallized compound. The metallocompounds that have the best binding to gyrase using Molegro are 1 and 1a, 9 and 9a, 10 and 10a, 11 and 12 (Tables 4-7). Autodock included the hydrated RAPTA complexes 14a, 13a and 15a immediately after  metallocompounds 12, 9a, 10, 4a, 8a, 1a, 10a and 5a (Tables 2-3). The binding affinities of the metallocompounds to HDAC7 show that the models 1, 12, 11, 10, 9 and their respective hydrolytic forms (where applicable) have better binding than its co-crystallized compound. This is also applicable to the RAPTA complexes. There have been experimental reports on HP-NCP also being a possible target of RAPTA complexes [15], but the new models metallocompounds 1, 9, 10, 11, 12 and their hydrated forms (where applicable) are predicted to bind better than RAPTA complexes by all three methods of docking. The crystal structure of HP-NCP shows RAPTA-C (metallocompound 14a) binding to two different sites on chain G and H of the receptor (Figure 2e). In order to make a parallel comparison of the poses of metallocompound 14a with the available experimental crystal structure, its docking features are shown in Figure 2f. The constrained Molegro and Autodock docking of complex 14a gives preference to the first binding site, which suggests that complex 14a will bind preferentially to the first site than the second. Also, the constrained Molegro docking shows that the PTA unit of this metallocompound locates the same receptor pocket as the crystal structure (Figure 2e). The new models of metallocompounds 1, 9, 10, 11, 12 and 8 (include their hydrated forms) are predicted among those that have the best binding to BRAF kinase (Tables 4,5). BRAF kinase has been proposed as a target of the "piano stool" type of Ru complexes [16]. This "piano stool" which was treated as a co-crystallized compound during docking was predicted in the relaxed Molegro docking (Tables 4,5) to have the affinity closest to the best binding complexes 1a and 12. The best binding metallocompounds to rHA from the three methods of docking are still 10, 12, 1, 3, 11, 8, 9, 3 and 14 (including their hydrated forms where applicable). The rHA is predicted as an average target, suggesting that many of the Ru(II)-based complexes that are considered will on average be kinetically favourable which further support the experimental findings [3,5,27].
In the interaction of the metallocompounds with RNR, the binding site co-crystallized compound is found to bind preferentially to all the metallocompounds. The predicted best binding metallocompounds to RNR are 1, 10, 8, 11, 12 and 9 including the bidentate RAPTA complexes 3, 2 and hydrated 14a. The suggested best binding metallocompounds to TopII after the co-crystallized compound are 1, 10, 9, 12, 11 and the RAPTA complexes 2, 3, 14a and 16a. Commonly predicted best binding metallocompounds to TS by the three methods of docking are 12, 10, 11, 9 and the bidentate RAPTA complexes 2 and 3 while only Molegro rated 1 as part of the best for TS.

Metal-Receptor Residues Covalent Interaction Using Constrained Molegro Docking
Since the targets of many organometallic complexes are not yet known [5][6][7][8][9], there is yet to be clear evidence of their metal centre forming covalent bonding with the residues of the receptors. In docking, the possible covalent interaction of the metal with any of the target residues can best be computed using constraints [14] but the application is usually limited to the systems in which such interactions have been experimentally proven. Generally, it is understood that the metal centre (the acidic centre) in metallocompounds may interact with nucleophilic centres (the basic centres) like the sulphur atom of the thiol in cysteine and the thioether in methionine, the imidazole nitrogen atom in histidine (i.e., for protein receptors) and the N7 atom of the guanine in DNA [11,30]. There have been both reports of inhibitory activities of metal complexes where there is no observed metal-residues covalent interaction even within the 4.00 Å distance from the metal centre [16,31] and in cases in which there is the possibility of a covalent bond between the metal centre of metallocompounds and the receptor residues [15]. There is a clear indication that the metal atoms in many cases prefer to hold the ligands together for synergistic effects [17] and for better interaction with the receptor residues than forming covalent bonds. In order to apply the possibility of covalent bonding of the metal centre with the receptor residues, we apply a distance constraint of 2.00 to 4.60 between the metal and the nucleophilic center of residues that makes up the receptor-binding site. In this study, based on the available information on the binding site residues of each receptor, a nucleophilic atom was selected as centre of the interaction with the metal centre of the metallocompounds. The cysteine sulphur atom (SG) was selected for CatB (Cys A 29) [32], BRAF kinase (CYS A 531) [16], RNR (CYS B 439) [33], TrXR (CYS A 59) [34] and TS (CYS A 198) [35]; a histidine nitrogen atom was used for HDAC7 (HIS A 709) [36], HP-NCP (HIS H 106) [15] and TopII (HIS B 20) [37]; while for gyrase [23] and rHA [31] where none of the HIS, CYS and MET are within the binding site residues, the nucleophilic nitrogen atoms of ASN A 46 and ARG A 144 were used, respectively. For DNA (4DL7) [38], the nitrogen atom (N7) of the guanine (DG T 7) was used.
The general feature is that the binding affinity is lower when the covalent constraint is applied (Tables 6,7) compare to when it is not (Tables 4,5). This therefore shows that the ruthenium metal in many of the metallocompounds will prefer to position the ligands for optimum receptor interactions than to form covalent bonds with the residues of protein receptors. However, the binding affinities of some metallocomponds against targets like DNA-1, DNA-2, HDAC7, HIS and RNR during the covalent constrained docking is comparatively high as that of the relaxed docking, indicating the possibility of the ruthenium atom forming covalent interactions with the receptor residues. Whether the constraint is applied or not, some of our model metallocompounds like 1, 9, 10, 11 and 12 (including their hydrated forms where applicable) are still ranked among those that have the best binding affinities to all the receptors. The binding behaviour of many of the metallocompounds when the covalent constraint is applied or omitted shows that the functions of the metal centre of the metallocompounds are receptor dependent.
In summary, the new models, especially complexes 1, 9, 10 and 12 (and their respective hydrated forms where applicable) are predicted by all the methods of docking between the best two inhibitors of all the receptors. Also, complex 11 is rated among the best three during the relaxed docking for gyrase and during the constrained docking for CatB and HDAC7. Specific to Autodock docking, the hydrated complexes 4a and 5a are suggested among the best inhibitors of DNA-1, DNA-2, HP-NCP, Kinase, TopII, TrXR and TS. Ranked high among the RAPTA complexes in many of the receptor interactions are the bidentate complexes 3 and 4 followed by the hydrated complexes 14a and 16a. This gives further insight into the reported better anticancer activities of complexes 14a [5] and 16a [27]. Considering the ranking of the three methods of docking, the best five metallocompounds predicted as inhibitors of the receptors are mostly within the proposed new models except in TopII, RNR and rHA where some of the RAPTA complexes are ranked among the best five. However, there is none of the RAPTA complexes appear among the best two proposed inhibitors of the receptors considered in this work using the three methods of docking (Table 1) except in rHA and RNR, where the ranked first and the second complexes are the non-hydrated and hydrated form of the same compound (Tables 4,5 and 6,7). This suggests the possibility of the new models acting a better anticancer than the RAPTA complexes.

The Correlation of Factors that Determined the Binding Activities of the Metallocompounds
The docking method of Molegro gives a more comprehensive insight into the factors that determine the binding affinities of the metallocompounds. Their correlation Tables 8 and 9 are constructed using the statistical package called R [39]. The most significant factors that influence the metallocompounds interaction with the DNA and many of the protein receptors are intramolecular van der Waal (E.Intra.vdw.), pose energy, steric and long range electrostatic interaction (ElectroLong in Tables 8,9). In all the protein receptors, the inter protein-ligand and the pose energies significantly determine the total binding energy of the metallocompounds. The presence of higher number of heavy atoms is proposed to favour better binding energy in all the receptor-ligand interaction. The presence of halogens also averagely favours the ligand-receptor interaction. A higher torsional bonds and lower steric hindrance are shown to favour the binding interaction of the metallocompounds. In the unconstrained docking, the short electrostatic interaction (Electro in Table 8) plays significant role in the binding affinity of the metal complexes to the receptors gyrase and RNR and the long range electrostatic interactions (ElectroLong in Table 8) influence the binding to DNA, TrXR and TS. The HB strongly affects the metallocompounds binding to TrXR while Van der Waal forces strongly influence metallocompounds binding to HP-NCP (VdW in Tables 8 and 9). In the constrained system, soft constraints strongly influence the binding affinities of metallocompounds to BRAF kinase and rHA. Also, steric interaction energies (Steric in Tables 8 and 9) between the receptors and the metallocompounds strongly influence the binding affinities in both constrained and unconstrained docking.

Computational Methods
The geometries of the metallocompounds were first optimized in Gaussian 03/09 [40,41] using PBE0 hybrid density functional [42] with two different basis sets of 6-31G* for all atoms order than Ru, Cl and P, which were defined with SBKJC VDZ [43] with effective core potential. The atomic charges were recalculated using b3lyp hybrid functional [44] and all electron minimal basis set 3-21g [45] in order to have a uniform representation of the charges. The atomic charges were incorporated into the docking packages for each of the metallocompounds. The docking activities were done using Autodock 4.2 [46] and Molegro [47] that combines differential evolution with a cavity prediction algorithm. The parameter file of the Autodock was further modified to incorporate ruthenium metal van der Waals and other needed parameters which were obtained from the Autodock website [48] but Molegro has built-in parameters that recognise the ruthenium metal. In Autodock, the number of grid points in x, y, z-axes was set to 60 × 60 × 60 with each point separated by 0.375 Å. Docking calculations were carried out with Lamarkian Genetic Algorithm (LGA).
Step sizes of 2Å for translation and 50° for rotation were chosen, the maximum number of energy evaluations was set to 2, 500,000 and for each of the 20 independent runs, a maximum number of 27,000 GA operations were generated on a single population of 100 individuals. In Molegro, the docking scoring function of MolDock which make use of piece-wise linear potential (PLP) was used. The maximum iteration was set to 2,500 against the default 1,500 and the population number also increased from default values of 50 to 100. Five maximum poses were selected and a more stringent re-scoring applied on them for a better prediction of the binding activities of the metallocompounds. MolDock scoring considers the hydrogen bonding, inter molecular protein-ligand and intra molecular ligand interactions and has been successfully applying for molecular docking [49]. Since the Molgro docking package is designed for the ligand-protein interaction, the simple trick employed to consider the ligand-DNA interaction is to include DNA as a cofactor of the protein. This is achieved by incorporating protein into the DNA pdb file 3LPV (named DNA-1) or using pdb 4DL7 input which contain both DNA and protein (named DNA-2) obtained from protein data base [50]. The view of the docking results and analysis of their surface with graphical representations were done using UCSF Chimera package [51].

Conclusion
In this paper we have presented the binding affinities of new model metallocompounds and some of the experimentally available metallocompounds, especially those of RAPTA complexes using the Molegro and Autodock methods of docking. Making use of the atomic charges of the metallocompounds obtained from the optimized quantum calculation significantly improved the predicted activities and ranking of the metallocompounds, which strongly agrees with the available experimental results. In summary, the new models, especially the complexes 1, 9, 10 and 12 (and their respective hydrated versions where applicable) are predicted by all the methods of docking to be between the best two inhibitors of all the receptors. Ranked high among the RAPTA complexes in many of the receptor interactions are the bidentate complexes 3 and 4 followed by the hydrated complexes 14a and 16a in agreement with reported better anticancer activities of complexes 14a [5] and 16a [27]. Considering the ranking of both relaxed and constrained Molegro docking, the best targets of the metallocompounds are in the order HDAC7 > DNA-1 > rHA > CatB > DNA-2 > Gyrase > TrXR > TopII > TS > RNR > HiS > Kinase except in few cases like complex 12 that prefers CatB to rHA. The general features are that the binding affinity is lower when the covalent constraint is applied (Tables 6,7) compare to when it is not (Tables 4,5). However, the binding affinities of some metallocomponds against targets like DNA-1, DNA-2, HDAC7, HIS and RNR during the covalent constrained docking is competitively as high as that of the relaxed docking, indicating the possibility that the ruthenium atom forms covalent interactions with the receptor residues. Also, only the new model metallocompounds are found to display strong binding affinities against DNA either as cofactor in Molegro or pure DNA as in Autodock docking of the metallocompounds against DNA-1 and DNA-2. The RAPTA complexes are suggested to target proteins like CatB and HDAC7 better than DNA, which further confirms the experimental suggestion that DNA may not be the target of RAPTA complexes [14]. The ranking order of the Molegro agrees better with the available experimental reports on some of the RAPTA complexes than the Autodock one, which is found to be within the error margin of Autodock predictions. The proposed mechanism of activation of metallocompound through hydration [2,28] is supported by the three methods of docking. Many of the hydrated forms of the complexes are found to significantly bind stronger than the non-hydrated complexes in many of the receptors. Also, the binding affinities of both hydrated and non-hydrated metallocompounds 16 and 16a respectively are higher than that of metallocompounds 15 and 15a by the Molegro, which further agrees with the experimental reports [27]. Those predicted among the RAPTA complexes to have the best binding to many of the receptors are bidentate and hydrated RAPTA complexes 2, 3, 14a, 15a and 16a. The lower binding affinities of the metallocompounds in the constrained metal-residues covalent bond type of docking compare to the relaxed Molegro docking and better correlation of the relaxed Molgro to the available experimental reports show that the metal atom in these type of metallocompounds will give preference to positioning the coordinated ligands in the right position for optimal receptor interactions and synergistic effects on each other than forming a covalent bond. Also from the Molegro docking, there is clear indication that the pose energy, inter protein ligand energy, intra molecular van der Waal and molecular weight strongly determine the final binding energy of the metallocompounds to the receptors.