1 Introduction

It is more than 30 years [1] that Bcl-2 (B-cell lymphoma 2) has been studied in search of anti-cancer drugs. The reason is linked to its role in apoptosis (programmed cell death) [2]. Among many aspects which have been studied [3, 4], finding small molecule inhibitors of Bcl-2 proteins has drawn special attention for discovering anti-cancer drugs [5, 6]. Such compounds are potential candidates for various cancer treatments [5, 7, 8]. More recently, interest has been shown to discover Bcl-2 inhibitors as anti-oral cancer drugs [9,10,11,12]. Various computational methods have been used in this purpose, although structure-based design (SBD) methods have been used widely perhaps due to the ability of such methods to study drug-receptor interactions helping understand drug activity more clearly. It is noteworthy that one of the most useful Bcl-2 protein inhibitors, an ABT series of drugs, ABT-263 (Navitoclax), has been discovered using SBD methods [5].

As of now, researchers have worked on discovering new series of compounds as Bcl-2 inhibitors [5,6,7,8,9,10,11,12]. However, we aim to evaluate known anti-neoplastic agents as Bcl-2 protein inhibitors in search of potential anti-oral cancer drugs. This approach is believed to help identify potential compounds differently from a carefully curated important set of known anti-neoplastic agents. Therefore, we have compiled a set of 276 such compounds taken from PubChem [13] and the compounds have been docked on a human Bcl-2 protein (PDB id: 6QGH), taken from Protein Data Bank (PDB) [14] using LibDock small molecule docking program available in Discovery Studio (DS) (ver. 4.1) [15]. The protein is complexed with a ligand ABT-263. Therefore, the ligand has been detached first and then the ligand-free protein and 276 compounds, collected from PubChem [13], have been prepared for docking using corresponding tools available in DS (ver. 4.1) [15].

For carrying out the docking studies, we have considered Methotrexate, a well-known anti-cancer drug, as a reference compound having a LibDock score of 114.76. Subsequently, 25 compounds that have returned scores higher than 100 (> 100) have been identified as the high-scoring compounds, and 114 compounds that produced LibDock scores between 80 and 100 have been identified as the next level of high-scoring compounds. The main reason for identifying the next level of high-scoring compounds is the apoptosis-related property that several compounds in this set possess and therefore can be of interest for the present purpose. Moreover, their scores are not too low either (compared to the high-scoring compounds) unlike those obtained for many other low-scoring compounds considered for the present docking studies. The results of these two sets of high-scoring docked compounds are given in two tables (Tables 1 and 2). The drug-receptor interactions have also been carried out for selected high-scoring compounds as well as some low-scoring compounds and that has been illustrated with docked poses along with the interacting bonds for the compound giving the highest LibDock score as well as those for a low-scoring compound for having a comparative picture. Finally, a set of 10 most probable potential anti-oral cancer drug candidates have been identified based on Lipinski’s Rule-of-Five [16] filtering and a few other considerations [13] from the compounds given in Tables 1 and 2.

2 Methods

We have carried out the docking studies using LibDock, a small molecule docking program, available in DS (ver. 4.1) software [15]. The LibDock docking program is based on a binding site comprising of lists of polar and apolar hot spots [15]. To carry out the docking studies, we have considered a human Bcl-2 protein obtained from Protein Data Bank (PDB id: 6QGH) [14] complexed with a ligand ABT-263. The ligand ABT-263 has been detached to get the ligand-free protein for docking studies. A set of 276 known anti-neoplastic compounds, carefully curated from PubChem [13], have been considered for docking. In this docking study, the receptor is fixed and the ligands/small molecules are flexible which allows the generation of several conformations/poses. The docking studies have been carried out using the default parameters, e.g., Number of Hotspots: 100; Docking Preference: High Quality, onto the first site in the protein molecule identified by the site finding tool available in DS (ver. 4.1) [15]. Subsequently, some selected compounds have been filtered through Lipinski’s “Rule-of-5” (Molecular Weight (MWT) not greater than 500; Calculated Partition Co-efficient, Clog P, not greater than 5.0; Hydrogen Bond Donor, NHs and OHs, not more than 5 and Hydrogen Bond Acceptor, Ns and Os, not more than 10) [16] by considering relevant PubChem [13] information to identify most probable and potential anti-oral cancer drug molecules. It may be noted that Lipinski et al. [16] have pointed out that the compound classes that are substrates of biological transporters are exceptions to the rule. It may also be noted that although Methotrexate violates the H-bond acceptor number to some extent (12 H-bond acceptors) [13] and may also have some notable side effects, it has been considered as a reference compound since it is a well-known anti-cancer drug. Lipinski et al. [16] have shown that some existing drugs, e.g., Erythromycin (MWT = 733.95, H-bond Acceptor = 14), do not obey all four points of the Rule-of-Five.

3 Results and discussion

In this section, we have first furnished and discussed docking results obtained by docking 276 known anti-neoplastic compounds, obtained from PubChem [13], on the ligand-detached Human Bcl-2 protein (PDB id: 6QGH) [14] using the methods available in DS (ver. 4.1) [15] as described earlier. This is followed by the views (screenshots) of the docked poses and the tables on non-bonded interactions of two selected compounds, the compound that have returned the highest LibDock score and one of the low-scoring compounds, for comparative analyses. Finally, several compounds which have been screened based on their LibDock scores and apoptosis-related properties [13] have been filtered through Lipinski’s Rule-of-Five [16] and some other considerations [13] to identify 10 most probable potential anti-oral cancer drug candidates based on the above-mentioned criteria adopted for the present study.

3.1 Docking and LibDock score

The 276 anti-neoplastic compounds, collected from PubChem [13] have been docked onto the first binding site, identified by the site-searching algorithm available in DS (ver. 4.1) [15], of the ligand-free Human Bcl-2 protein (PDB id: 6QGH) obtained from Protein Data Bank (PDB) [14]. To identify potential anti-oral cancer drugs, we have considered the well-known anti-cancer drug Methotrexate as a reference compound having a LibDock score of 114.76. Hence, we have first identified 25 compounds (including Methotrexate), considering them as highly potential, which have scored greater than 100 (> 100) as given in Table 1.

Table 1 The 25 anti-neoplastic agents as potential Bcl-2 inhibitors with LibDock scores greater than 100
Full size table

We have also identified a second set of 114 compounds which have returned scores between 80 and 100 (Table 2). The main reason for considering these compounds is based on the finding that several compounds of this set are apoptosis inducers and this property is believed to be an important factor for the present purpose. It has also been found that the number of apoptosis-inducing compounds having LibDock scores higher than 90 (> 90) is more in number than those having LibDock scores higher than 80 (> 80). 60 compounds have returned 90 + scores and those scores are much higher than the scores returned by several other docked compounds such as the two low-scoring compounds shown at the bottom of Table 2 (Nos. 115 and 116). Some other low-scoring compounds may possess apoptosis-inducing properties. However, they have not been considered here as potential compounds due to their low scores.

Table 2  A list of 114 Bcl-2 inhibitors with LibDock scores between 80 and 100 and 2 inhibitors with scores less than 80 (the last two compounds underlined)
Full size table

It may also be noted that there are compounds that have not docked at all and therefore they are beyond the scope of further discussion in this paper.

3.2 Views and information for selected docked compounds

In addition to getting LibDock scores of the docked compounds, we have also taken screenshots of two docked compounds (binding pocket shown in hydrophobicity scale) along with the details of their non-bonded interactions. Accordingly, we have furnished here the above-mentioned information for the highest scoring (LibDock score) compound Olaparib (Figs. 1 and 2; Table 3) which has been taken from Table 1 (No. 2). We have also considered a low-scoring compound Tucatinib from Table 2 (No. 116) for a comparative view and analyses (Figs. 3 and 4; Tables 4, 5).

Fig. 1
figure 1

Non-bonded interactions of the highest scoring pose of Olaparib (LibDock score: 124.36) of Bcl-2 target protein (PDB id: 6QGH)

Full size image
Table 3 Details of the non-bonded interactions of the highest-scoring pose of Olaparib
Full size table
Fig. 2
figure 2

Side view of docked Olaparib

Full size image
Fig. 3
figure 3

Non-bonded interactions of the highest scoring pose of Tucatinib (LibDock score: 54.80—low score) of Bcl-2 target protein (PDB id: 6QGH)

Full size image
Table 4 Details of the non-bonded interactions of the highest-scoring pose of Tucatinib
Full size table
Table 5 Details of the unfavorable non-bonded interactions of the highest scoring pose of Tucatinib
Full size table
Fig. 4
figure 4

Side view of docked Tucatinib

Full size image

It is clear from Fig. 1 that the highest scoring (124.36) compound, Olaparib, has made several favorable non-bonded interactions (Fig. 1; Table 3) with the receptor protein and is found to be well placed in the binding groove (Fig. 2). On the other hand, one of the low-scoring (54.80) compounds, Tucatinib, along with forming favorable non-bonded interactions (Fig. 3; Table 4), has formed an unfavorable bond as well (Fig. 3; Table 5). Moreover, a large portion of the compound is found to be outside the binding pocket (Fig. 4). These comparative analyses seem to go along with the finding that Olaparib has got high (highest among the compounds considered for the present study) LibDock score while Tucatinib has failed to achieve that.

3.3 Lipinski’s rule-of-five filtering results

In this section, we have reported the findings obtained from Lipinski’s Rule-of-Five [16] filtering and some other considerations [13] for the compounds screened from the docking studies to identify the most probable potential anti-oral cancer drug molecules. To obtain these potential compounds from Tables 1 and 2 (docking results), we have investigated those compounds that possess apoptosis-related properties, as described in PubChem [13], taking into consideration the role Bcl-2 plays in apoptosis [2] which seems to be relevant in the present context. In the process, we have found nine such compounds from those shown in Table 1. Out of those nine compounds, we have found seven compounds (Compounds 1–7 in Table 6) to be most suitable in that we have considered those compounds which are already approved by FDA or other such agencies and/or do not have any major side effects such as liver toxicity/injury [13]. Subsequently, we have identified three such compounds from Table 2 as well which have returned quite high LibDock scores; i.e., their LibDock scores are much closer to 100 compared to many other compounds docked. Moreover, these compounds are devoid of the side effects mentioned above. All the compounds shown in Table 6 have been found to satisfy Lipinski’s Rule-of-Five [16]. It may be noted that in PubChem [13], the partition coefficient (logP) has been given as computed XlogP3-AA or XlogP3 values. Now, that the compounds shown in Table 6 have got high LibDock scores [15] and satisfy Lipinski’s Rule-of-Five [16], these compounds may be regarded as the most probable potential anti-oral cancer drug molecules obtained from the present study.

Table 6 Identified most probable 10 potential anti-oral cancer drugs
Full size table

4 Conclusion

The purpose of the present study is to identify potential anti-oral cancer drug molecules from known anti-neoplastic agents, a particularly important class of compounds, through small molecule docking studies and Ripinski’s Rule-of-Five filtering. This is believed to help identify potential anti-oral cancer drugs as these compounds already show anti-neoplastic properties. For example, Olaparib, the most probable potential anti-oral cancer drug, identified from the present study, works by taking advantage of a defect in Deoxyribonucleic Acid (DNA) repair in cancer cells with BRCA (breast cancer gene) mutation and inducing cell death [13]. Moreover, Olaparib is a drug in use for the treatment of various cancers like ovarian cancer, breast cancer, pancreatic cancer, and prostate cancer [13]. All these seem to indicate that the identification of Olaparib as a potential anti-oral cancer drug is reasonable. Another important role that a Bcl-2 inhibitor can play is to control the over-expression of Bcl-2 which may confer resistance to chemotherapeutic drug treatment e.g., resistance of oral tongue squamous cell carcinoma (OTSCC) cells to cisplatin [9]. Therefore, the identification of effective Bcl-2 inhibitors is important, and the present approach is believed to find useful applications in this regard. Furthermore, it would be interesting to see by carrying out experimental work e.g., through studies on a cell line (in vitro) of interest, whether the identified compounds are able to exhibit anti oral-cancer activity, alone or in combination with other anti-oral cancer drugs.

Finally, the results obtained from these studies seem to be encouraging in identifying the most probable potential anti-oral cancer drugs as apparent from the docked poses shown in Figs. 1, 2, 3 and 4 for comparative analyses. While we have identified the 10 most probable potential anti-oral cancer drug molecules, one can always consider other compounds of one’s interest from those given in Tables 1 and 2. It appears that the findings of the present study and the approach followed in this purpose may significantly help discover anti-oral cancer drugs as Bcl-2 inhibitors for the protein target considered here and possibly for other relevant receptor targets as well by screening known anti-neoplastic agents.