Synthesis, molecular modelling and anticancer evaluation of new pyrrolo[1,2-b]pyridazine and pyrrolo[2,1-a]phthalazine derivatives

Abstract Two new series of heterocyclic derivatives with potential anticancer activity, in which a pyrrolo[1,2-b]pyridazine or a pyrrolo[2,1-a]phthalazine moiety was introduced in place of the 3′-hydroxy-4′-methoxyphenyl ring of phenstatin have been synthesised and their structure-activity relationship (SAR) was studied. Fourteen of the new compounds were evaluated for their in vitro cytotoxic activity by National Cancer Institute (NCI) against 60 human tumour cell lines panel. The best five compounds in terms of in vitro growth inhibition were screened in the second stage five dose-response studies, three of them showing a very good antiproliferative activity with GI50<100 nM on several cell lines including colon, ovarian, renal, prostate, brain and breast cancer, melanoma and leukemia. Docking experiments on the biologically active compounds showed a good compatibility with the colchicine binding site of tubulin.


Introduction
Considerable efforts have been focussed in the past decades, on the design and development of new antiproliferative drugs with improved efficiency, limited toxicity, cost-effectiveness, which are synchronously less prone to develop multidrug resistance [1][2][3] . Among the variety of targets used in this huge anticancer fight, tubulin targeting appears to be a key focus in cancer treatment, the research in this field remaining very active in past years [4][5][6][7] . After the success of Colchicine 8 , combretastatin A-4 9 , vincristine or vinblastine 10 as anticancer drugs acting by inhibiting tubulin polymerisation, research efforts focused on developing new colchicine binding site inhibitors with improved pharmacological profiles 4,11 .
One of the simplest known structures synthesised and tested as an anticancer agent in the past years is phenstatin 12,13 which stand as one of the most potent tubulin polymerisation inhibitors by binding to the colchicine site of the tubulin and thus, interfering with the equilibrium dynamics associated with the cell division 14,15 . Because of its biological properties and structural simplicity, phenstatin continues to be a lead compound for rational design in anticancer therapy, the recent literature being plentiful of such phenstatin analogues [16][17][18] .
Pyrrolo-fused derivatives comprise a class of biologically active heterocyclic compounds which can serve as promising scaffolds for the development of anticancer, antimicrobial, antiviral, antimalarial, antitubercular, anti-inflammatory, and enzyme inhibiting drugs 19 . Among the fused pyrrolo-heterocyclic compounds, pyrrolo [1,2-b]pyridazines and its condensed pyrrolo[2,1-a]phthalazine system are compounds well known for their strong luminescence 20,21 and photochromic properties 22 , and at the same time are promising in the field of drug design 23,24 , some derivatives being reported to have antimicrobial 25,26 , antifungal 25 or anticancer effects 27,28 , or to act as acyl CoA:diacylglycerol acyltransferase (DGAT1) inhibitors 24 , JAK inhibitors 29 , HER-2 tyrosine kinase inhibitors 30 , IRAK4 inhibitors 31 , or MEK inhibitors 32 .
The replacement of one of the substituted phenyl ring of phenstatin with pyrrolo-fused heterocycles has been a major focus in rational drug design in the recent years, as there are several reported biological active phenstatin analogues containing an indole ring 5,19 , an indolizine ring 33 , or a pyrrolo [2,3-d]pyrimidine ring 34 . However, to our knowledge, there are no reported analogues of phenstatin with pyrrolo [1,2-b]pyridazine or pyrrolo [2,1-a]phthalazine scaffolds, respectively.

Chemistry
All commercially available reagents and solvents employed were used without further purification. Melting points were recorded on an A. Kr€ uss Optronic Melting Point Meter KSPI and are uncorrected. Proton and carbon nuclear magnetic resonance (d H , d C ) spectra were recorded on a DRX-500 Bruker or a Bruker Avance 400 DRX spectrometers. The following abbreviations were used to designate chemical shift multiplicities: s: singlet, d: doublet, t: triplet, q: quartet, m: multiplet, bs: broad singlet, as: apparent singlet. All chemical shifts are quoted on the d-scale in ppm. Coupling constants are given in Hz. IR spectra were recorded on a FTIR Shimadzu or Jasco 660 plus FTIR spectrophotometer. Analyses indicated by the symbols of the elements or functions were within ±0.4% of the theoretical values. Thin layer chromatography (TLC) was carried out on Merck silica gel 60F 254 plates. Visualisation of the plates was achieved using a UV lamp (k max ¼ 254 or 365 nm).
General procedure for preparation of compounds 8a-t and 11a-d The cycloimmonium salt (7a-d or 10a-d) (1 mmol) and ethyl propiolate (1.1 mmol) were added to 10 ml of anhydrous acetone and the obtained suspension was stirred at room temperature. Then, a solution of triethylamine (TEA) (3 mmol, 3 equiv.) in anhydrous acetone (3 ml) was added drop-wise over 1 h (magnetic stirring) and the resulting mixture was then stirred overnight at room temperature. Water (10 ml) was added and the formed solid was collected by filtration to give a powder which was washed with 5 ml methanol. The product was crystallised from dichloromethane/methanol (1:1, v/v).    Flexible docking experiments were carried out in Autodock Vina 40 , using a 18x22x22 Å3 grid box centered on the colchicine binding site of the a,b-tubulin heterodimer crystal structure (PDB: 1SA0) 41 . The 3 D structures of the compounds were constructed in Avogadro v1.2.0 42 and were subjected to 10,000 steepest descent steps of energy minimisation in the MMFF94 force field. One hundred poses were generated for each ligand, and the best-ranked models were chosen for further visual inspection in order to assess the consistency of the generated docking solutions relative to the docking poses of known inhibitor colchicine. Molecular graphics and visual analyses were performed in The PyMOL Molecular Graphics System, Version 1.8.2. (Schr€ odinger, LLC). Logp values were calculated using the ChemAxon/Chemicalize server (www.chemicalize.com).

Cell proliferation assay
The compounds were tested against a panel of 60 human cancer cell lines at the National Cancer Institute, Rockville, MD. The cytotoxicity experiments were realised using a 48 h exposure protocol using sulphorhodamine B assay [43][44][45] .
First, pyridazines 1-5 (Scheme 1) were used for the synthesis of their monoquaternary salts with 2-bromoacetophenones 6a-d. While compound 6d is commercially available, compounds 6a-c were synthesised using reported procedures 46 . The quaternisation reactions were carried out at room temperature (r.t.) in a minimal amount of acetone, leading to the formation of salts 7a-t (Scheme 1).
As shown in Scheme 2, ethyl propiolate was reacted with the corresponding pyridazinium ylides 7 0 a-t (in situ generated in basic medium from salts 7a-t) to give the intermediate dihydropyrrolo [1,2-b]pyridazines 8 0 a-t, which in turn underwent oxidative dehydrogenation under atmospheric conditions, yielding the final compounds 8a-t in moderate yields (40-52%) (Scheme 2).

Biological activity
Fourteen of the synthesised compounds (8a, b, d, e, f, h, i, j, k, n, q, and 11a-c) were selected by the National Cancer Institute (NCI) for screening against a panel of 60 human tumour cell lines at a single dose of 10 μM 43 , the representative results for the active compounds being summarised in Table 1.

Molecular modelling
Because both computational and biological models of 3,4,5trimethoxyphenyl-containing phenstatin analogues supported the hypothesis that the antiproliferative effects of these compounds are induced by inhibiting tubulin polymerisation 5,7,33,47,48 , docking experiments were performed on the colchicine binding site of the a,b-tubulin heterodimer (PDB:1SA0), in order to evaluate the shape and electrostatic complementarity between ligands and the a,b-tubulin heterodimer interface, which could account for the observed antiproliferative effects. Compounds 8a and 8b displayed similar docking conformations grouped into two distinct clusters, both having the trimethoxyphenyl subunit overlapping with the one in the co-crystallised DAMA-colchicine ligand (Figure 2(a)), and interacting with the protein through hydrogen bonding with bCys241. The ligands are further stabilised in the binding pocket through hydrophobic interactions with bLeu242, bLeu248, bAla250, bLeu252, bLeu255, and bVal238. The diazine moiety either extended on top of the binding pocket, with the ester functional group orienting towards the dimer interface (Figure 2(b,c)), or was flipped at about 180 , to have the ester group roughly overlapping with the third colchicine ring in the crystal structure. Interestingly, the 4-bromo-substituted compound 8d, which displayed a less pronounced biological activity than 8a and 8b, adopted a conformation in which the p-bromo substituted phenyl was accommodated more deeply in the colchicine binding pocket, resulting in a shift in the position of the central heterocyclic moiety towards the center of the colchicine binding site, which led to the disruption of the hydrogen bond with bCys241 (Figure 2(d)).
Overall, the docking experiments suggest that the removal of the 4-methoxy group does not influence the accommodation of the ligand in the binding pocket, in agreement with the biological data, while the introduction of a bromine atom as substituent can induce a different binding conformation which leads to the disruption of the hydrogen bond between the ligand and bCys241, which could account for the reduced antiproliferative activity of compound 8d.
The 2-methyl-substituted analogues of 8a and 8b (8e and 8f) were accommodated in a similar fashion to that of parent compounds (Figure 3(e,f)), suggesting that the introduction of a methyl substituent does not influence the binding preferences of the compounds in the colchicine binding site, in agreement with the biological data in terms of antiproliferative activity. Compound Table 2. Results of the 5-dose in vitro human cancer cell growth inhibition a for compounds 8a-b, e-f and 11b and compared with standard drug Doxorubicin.

Cell type
Compound ! 8a 8b 8e 8f 11b Doxorubicin c 8h, which displayed a marked reduction in biological activity when compared to parent compound 8d, did not form the two expected well-defined clusters of conformations, but rather had a broad range of unrelated docking poses, the most energetically favourable being similar to the second cluster of compound 8d.
Interestingly, 2-(p-halogeno-phenyl)-substituted compounds 8i, 8j, 8k, and 8n, which showed a marked decrease in growth inhibition activity when compared to unsubstituted analogues, were compatible with the colchicine binding site, and were accommodated in a similar fashion to their unsubstituted or 2-methyl-substituted  analogues with biological activity to the a,b-tubulin heterodimer (Figures 2 and 3). A closer inspection of the basic physicochemical properties of these four compounds reveals, however, a violation of Lipinski's rule of five in terms of logp values 49 (Table 3), which could account for the loss in antiproliferative efficacy in spite of apparent activity at the colchicine binding site 50 .
Docking of pyrrolo[2,1-a]phthalazines 11a-c revealed a single cluster of conformations for each compound, similar to the second cluster obtained for pyrrolo [1,2-b]pyridazines 8a, 8b, 8e, and 8f, in which the heterocyclic subunit is oriented as to have the ester group roughly overlapping with the third colchicine ring in the crystal structure ( Figure 4). The methoxyphenyl subunit is stabilised by a hydrogen bond interaction with bCys241, similar to the case of pyrrolo [1,2-b]pyridazine analogues. Notably, compound 11b adopts a conformation slightly deeper in the hydrophobic pocket, which induces a rotation of the heterocyclic core and facilitates a hydrophobic interaction with bLeu248, which is unique among the three docked pyrrolo[2,1-a]phthalazines. A tighter hydrophobic interaction between 11b and the protein could account for the pronounced antiproliferative activity exerted by 11b among the three tested pyrrolo[2,1-a]phthalazines.
However, for all compounds, complementary tubulin polymerisation assays are needed in order to confirm the proposed molecular mechanism.

Conclusion
In summary, five of the newly synthesised pyrrolo [1,2-b]pyridazine and pyrrolo[2,1-a]phthalazine phenstatin analogues showed in vitro antiproliferative activity, the most potent being compounds 8f with GI 50 values <100 nM on thirteen cell lines including colon, ovarian, renal, prostate, brain and breast cancer, melanoma and leukemia. Notably, compound 8a showed a very good antiproliferative effect on melanoma MDA-MB-435 cell, renal cancer A498, and leukemia SR cell. The substitution of position 2 of pyrrolo [1,2b]pyridazine with a methyl group generally appears to increase the antiproliferative potency of the compounds, while the introduction of a more bulkier substituent is completely detrimental for the growth inhibitory properties, despite the fact that docking studies showed a good compatibility with the colchicine binding site of tubulin. The lack of proliferative activity in the case of the bulkier 2-(4-X-phenyl)-pyrrolo [1,2-b]pyridazines could be explained by the suboptimal lipophilicity and solubility of these compounds. However, further assaying in terms of tubulin polymerisation is needed in order to confirm the proposed antiproliferative mechanism of action of the newly synthetised compounds. Compound 8f could serve as a useful lead compound for further structural optimisation in the development of new anticancer agents.  Values were calculated using the ChemAxon/Chemicalize server (www.chemicalize.com).