Following of the chemical diversity of 4-thiazolidinones the in-house library of new heterocycles have been designed and synthesized (more 7000 compounds). Anticancer, antitrypanosomal, antituberculosis and antiviral activity screening led to SAR database formation; lead-compounds identification; design of focused sub-libraries; formation and validation of hypotheses for structure optimization: i) complications of C5 fragment and/or functionalization of N3 position; ii) creation of the hybrid molecules; iii) fixation of 5-ene-4-thiazolidinones in fused heterocycles via annulation (thiopyrano[2,3-d]thiazoles were found as a cyclic isosteric mimetics of 5-ene-4-thiazolidinones); iv) the leukemia panel was detected to be the most sensitive among all cancer cell lines. Following the in silico and pharmacological data for the investigation of molecular mechanism of anticancer effect the argument in favor of the apoptotic related and mild prooxidant actions for active compounds have been found.
The main milestones of the project involved synthetic investigations, biological activity studies, rational approaches (QSAR-analysis, docking, molecular modelling etc.) to “drug-like” molecule design [1–3]. Initially, various activity types were studied (anti-inflammatory [4, 5], antimicrobial , anticonvulsant , choleretic , etc.), then investigations focussed on the study of anticancer, antimycobacterial, antiviral and antitrypanosomal activities. More than 2000 biological assays were conducted, allowing the identification of at least 200 hit compounds with anticancer action, 40 hits with antimycobacterial activity and 30 hits with the inhibitory activity against different virus strains. The early stage results demonstrated that the compounds possessed anti-inflammatory activity, antimicrobial action, antioxidant, choleretic and anticonvulsant activities.
The in-house library of heterocycles has been an object for the study of anticancer activity within the NCI, NIH protocol [9–11]. Among the 1,750 tested compounds, 525 (30.0%) have successfully passed the pre-screening phase (Figure 2). After passing the second testing phase, 14 compounds were submitted for consideration by the NCI Biological Committee, among them, 8 compounds were affirmed for in-depth in vivo preclinical trials as potential anticancer agents.
Figure 3 presents selected hit compounds from different groups that possess high antimitotic effect in vitro at submicromolar concentrations (10-5–10-8 М) and are characterised by low in vivo toxicity. It is important to note that these compounds are representative of 5-ene-4-thiazolidinones (“Biological way”) [12–25] and thiopyrano[2,3-d]thiazoles (“Chemical way”) [36–30].
Interestingly, in the anticancer selectivity rating, the leukaemia cell lines were the most sensitive to 4-thiazolidinones and related heterocyclic systems following the analysis of the in-depth in vitro research results. A series of cell lines, such as leukaemia lines (CCRF-CEM, HL-60(TB), RPMI-8226, SR, K-562, MOLT-4), CNS cancer line (U251), non-small cell lung cancer line (HOP-92), renal cancer cell lines (UO-31, 786-O), colon cancer line (HCT-116) as well as breast cancer line (MDA-MB 231) were found to be the most sensitive to the test compounds. Thus, based on the obtained results, it was hypothesised that the heterocycles containing a “thiazolidinone matrix” have specific anti-leukaemia activity.
In silico approaches for anticancer activity data analysis
The COMPARE analysis [10,11] was performed for the active compounds to investigate the similarity of their cytotoxicity pattern (mean graph fingerprints) with those of known anticancer standard agents, NCI active synthetic compounds and natural products. For some synthesised heterocyclic substances, there was established correlation with the inhibitors of tubulin polymerisation, RNA polymerase, p-glycoprotein or topoisomerase II, inductors of apoptosis and activators of caspases. It is of note that the significant values of the correlation coefficients of thiazolidinone derivatives from different sub-librariesto the S-trityl-L-cysteine, aminoacyl-tRNA synthetases inhibitor with antiproliferative effect against leukaemia.
Following the analysis of anticancer activity profiles of different thiazolidinones using modern computational methods, like principal components analysis, neutral networks and cluster analysis, it was found that two different mechanisms and a "mixed" mechanism were responsible for the anticancer activity [3, 31–33].
In cooperation with the Institute of Cell Biology, NAS of Ukraine (Prof Rostyslav Stoika), the hypothesis regarding theapoptosis-dependent mechanism of antitumor activity was confirmed. Moreover, the proapoptotic activity of the tested compounds was observed. Studies have been conducted regarding the effects of compounds on some cytokines, in particular, initiatory effector groups of caspases and cytokines involved in the development of caspase-independent apoptosis. Different apoptotic-related pathways were detected. Also, it should be noted that there were various implementations of such an effect, involving the “classic” apoptotic pathway (mediated by Bax & Caspase-7), caspase-independent apoptosis(mediated by AIF) and “mixed-type” apoptosis (mediated by AIF, Bax & Caspase-9) (Figure 4) [34–37]. This data correlated well with the results obtained in in silico studies[3, 31–33].
Based on the analysis of the biological and in silico data, we proposed a pharmacophore model for the design of potential anticancer agents [38, 39]. The pharmacophore (Figure 5) consists of two aromatic or π-ring system centres, a hydrophobic group and the two projections of the hydrogen bond donors (electron pair acceptors) (error rate 0.8%, accuracy 87.5% and precision = 99.5%).
A new sub-library was created (690 structures) by varying the substituents in the 2,4 and 5 positions of 4(2)-thiazolidinone. Pharmacophore and Random Forest models predicted 101 and 32 hit compounds from the virtual sub-library, respectively. Analysis of the distances to the model showed that all predicted compounds were active, falling into the applicability domain, while 47 out-of-domain compounds were inactive. The Gaussian processes model predicted the cytotoxic effect on tumour cells for 60 structures. The choice of virtual screening hits was based on the consensus between all predictions, thereby predicting 29 hit compounds using these three screening models, of which, 14 structures were selected for synthesis (Figure 6).
Using the above-mentioned approach, 14 novel derivatives 92–105 were selected and synthesised (Figure 7).
The success of the purposeful strategy was confirmed by biological assays, according to which, the synthesised compounds inhibited cancer cell growth, even at micromolar concentrations. The four most potent synthesised compounds 94, 96, 97 and 103 showed IC50 values between 0.16–10 μM in MTT assays of rat glioma cells C6, Mino cells, Jurkat and L1210 cells.
The study of antimicrobial activity was conducted on the Mycobacterium tuberculosis H37Rv (ATCC 27294) within the Tuberculosis Antimicrobial Acquisition & Coordination Facility (TAACF) Programme at the National Institute of Allergic and Infectious Diseases (NIAID, USA). Among the tested compounds, 40 active substances with an IC90 ≤10 μg/mL were identified, for which the cytotoxicity of mammalian cells (CC50) was determined on VERO cells and a selectivity index (SI = CC50/IC90) was calculated. For the in-depth studies, 7 derivatives (Figure 8) from different sub-libraries (5-ene-4-thiazolidinones and thiopyrano[2,3-d]thiazoles) were selected.
Screening of the antiviral activity of the synthesised compounds was also conducted within the Antimicrobial Acquisition & Coordination Facility (AACF, NIAID, USA) Programme. As a result, 30 compounds were identified with significant antiviral activity and sufficient selectivity indices [26, 40–43]. In addition, a group of 4-thiazolidinone-related compounds (Figure 9) with a strong effect on hepatitis B (HBV) and C (NSV) and Flu A & B viruses were identified, as well as one high-level SARS and Tacaribe strain (biological weapons virus).
Derivatives of rel-(5R,5aR,11bS)-2,6-dioxo-3,5a,6,11b-tetrahydro-2Н,5H-chromeno[4’,3’:4,5]thiopyrano[2,3-d][1,3]thiazole-5-carboxylic acids belong to the most promising group of compounds (Fig. 10). The substituents in the positions 8 and 10 and the ester group in position 5 are desirable for antiviral activity. The increase of the alkyl moiety length leads to an increased effect, thus, compound 106 showed a higher activity against Influenza Virus Type A(H3N2, Perth strain) with an EC50 = 0.6 ÷ 2.5 µg/ml and SI = 40 ÷> 170 than derivative 107 with an EC50 = 0.31 ÷ 0.32 µg/ml and SI => 310 ÷> 320 .
The study of antitrypanosomal activity was conducted at the National Museum of Natural History (Prof Philippe Grellier, France) on Trypanosoma brucei brucei (TBB) and Trypanosoma brucei gambiense (TBG) strains. Among more than 200 tested compounds, a series of hits with pronounced antiparasitic activity in cellulo with IC50 values of 0.03–15.64 μg/ml were identified (Figure 11) [27, 43–48].
Conclusions and further perspectives
- 4-Thiazolidinones possess a variety of biological activities, both in screening campaigns and directed experiments, hence are considered as a tool for the synthesis of related heterocycles, simplified analogues and diversity complex heterocycles within various approaches.
- Among the variety of thiazolidinone subtypes, 5-ene-thiazolidinones are of special interest as hit- and lead-compounds possessing anticancer, antimicrobial, antiviral, and antitrypanosomal activities.
- Assigning 5-ene-thiazolidinones as pan assay interference compounds (PAINS) due to their possible Michael acceptor functionality must be analysed in-depth: experimental confirmation is essential to claim target compounds as PAINS; the positive aspect of such covalent modifiers should not be discarded.
- Annealing of a thiazolidine core into thiopyranothiazole analogues is one of prominent optimisation directions which will allow conservation of the activity pattern of synthetic precursors (5-ene-4-thiazolidinones), decrease the toxicity and avoid the Michael acceptor properties.
- The main directions for 4-thiazolidinones optimisation are: i) complication of the fragment in the C5 position; ii) introduction of the substituents in the N3 position (especially fragments with carboxylic group or its derivatives); iii) annealing in complex heterocyclic systems; iv) combination with other pharmacologically attractive fragments within a hybrid pharmacophore approach.
- 4-Thiazolidinones are useful tools for medicinal chemistry and should not be regarded as useless per se.