Journal Pre-proof Multigram scale synthesis of polycyclic lactones and evaluation of antitumor and other biological properties

. An efficient four-step synthesis of tetracyclic lactones from 1,4-benzodioxine-2-carboxylic acid was developed. Ellipticine derivatives exhibit antitumor activity however only a few derivatives without carbazole subunit have been studied to date. Herein, several tetracyclic lactones were synthesized and biologically evaluated. Several compounds ( 2a , 3a , 4a and 5a ) were found to be inhibitors of the Kras-Wnt pathway. The lactone 2a also exerted a potent inhibition of Tau protein translation and was shown to have capacity for CYP1A1-bioactivation. The results obtained are further evidence of the therapeutic potential of tetracyclic lactones related to ellipticine. Molecular modeling studies showed that compound 2a is inserted between helix a 3 and a 4 of the KRas protein making interactions with the hydrophobic residues Phe90, Glu91, Ile9364, Hie94, Leu133 and Tyr137 and a hydrogen bond with residue Arg97. 19 17 (C), 143.7 (C), 197.5 (C). Anal. Calcd. for C 21 H 22 O 3 S: C 71.16%; H 6.26%. Found: C 71.33%; H 6.44%.


Introduction
Ellipticine (5,11-dimethyl-6H-pyrido[4,3-b]carbazole, 1), is a planar tetracyclic alkaloid that presents an interesting cytotoxic activity attributable to multiple mechanisms of action. However, the therapeutic applications of ellipticine remain limited because of the large number of side effects [1]. Interest in the ellipticine pharmacophore has attracted the attention of many research groups because of their multiple biological activities [2] and especially for their capacity to inhibit cancer cell growth [3].
Previously, we have reported on ellipticine derivatives in which the indole nucleus has been replaced by 1,4-benzodioxine as interesting bioactive compounds, but their mechanism of action was not confirmed [4]. In this study, we report the preparation of ellipticine derivatives containing the 1,4-benzodioxine nucleus and a lactone replacing the pyridine system. These structural changes and the presence of different substituents make these compounds interesting and promising antitumor candidates.

KRas Inhibition.
KRas mutations are involved in the development of many human tumors.
Progression to colorectal cancers (CRC) requires a second event such as an activating KRas mutation, which is triggered by undefined interactions between the Wnt signalling pathways and the KRas protein [7]. The KRas synthetic lethal phenotypic assay measures survival of CRC cells carrying mutations that activate both Wnt and KRas signaling relative to those with other driver mutations [7]. The measurement of ATP is an accepted method to estimate the number of viable cells. Human CRC cell lines with KRas mutations (DLD-1, HCT116, and SW480) were used to test the effect of the tetracyclic lactones on cell proliferation in vitro [8].
The results showed that compounds 2a, 3a, 4a and 5a exerted dose-dependent inhibition of CRC cells while compounds 2b, 3b, 4b, 5b, 6b, and 7b generally were less potent active (see Table 1). Interestingly, a limited inhibitory effect was observed at the lowest dose of 2a in HCT116 cells, but it increased substantially in glycogen synthase kinase-3 GSK3-β inhibitor HCT pretreated cells. This result indicates an effect on the Wnt pathway at low dose (0.2 µM) but not at the higher doses investigated (2 and 20 µM).
The data also reveals that, in general, the compounds presenting the carbonyl group of lactone on the same side as the long alkyl chain (series a) are more potent compounds than when it is on the side of the methyl group (series b) (see position of R', Scheme 1). The length of alkyl chain (4, 5 or 6 carbons (2a, 3a and 4a)) modifies cytotoxic activity very little.
Likewise, the introduction of halogen on the alkyl chain (5a) produces only slight modifications in the activity, while the change from lactone to thiolactone leads to a pronounced decrease in potency (6a and 6b). Finally, the introduction of a methoxyl group as a substituent of the first aromatic ring of the tetracyclic lactone (at the C-7 position) also produces a marked loss in cytotoxicity (7a and 7b) attributable to that the substituent difficult the binding to the active site of the target.

Angiogenesis inhibition.
Angiogenesis and vasculogenesis facilitates blood vessel sprouting and tube formation, processes which can be modified using vascular disrupting agents (VDAs) that cause rapid collapse and shut down of established cancer blood vessels leading to regional tumor ischemia and necrosis [9]. Often inhibitors of angiogenesis bind the extracellular domain of vascular endothelial growth factor (VEGF) receptor-2 [10] or the specialized ligands VEGF-A, VEGF-C and VEGF-D [11].
Monoclonal antibodies targeting endothelial factors have been studied and approved by the FDA for the treatment of several tumors [12]; however, it should be noted that small molecules with the same therapeutic use are still not medically used.
The results showed that compounds 2a, 3a, 4a and 5a (a series) possessed a high antiangiogenic activity on tube area but not on nuclear count, demonstrating selectivity of these compounds, while the compounds in the b series (2b, 3b, 4b, 5b, 6b, and 7b) were inactive in both assays (Table 2). In addition, side alkyl chains with more than 5 C lead to inactive compounds. Compounds (2a, 3a, 4a and 5a could be of interest for the reduction of angiogenesis in endothelial tube formation without causing antiproliferative activity. As observed in the KRas inhibition assay, compounds possessing an alkyl substituent on the same side as the carbonyl group of the lactone were also the most potent analogues as antiangiogenic inhibitors.  (Table 3), while compounds 6a and 7a showed no activity at the concentrations tested. All b isomers were also inactive.
It is worthy of note that compounds 7a and 7b containing a methoxy group on the aromatic ring are surprisingly inactive in several cytotoxic tests while the methoxylated derivatives of ellipticine showed high cytotoxicity [13].

Wnt pathway activation.
Osteoporosis is a silent disease that weakens the bones over time and increases the risks of fractures. The mechanism of bone loss is not yet well understood, but the disorder arises from an imbalance in the formation of new healthy bones and the decrease and resorption of bone tissue. The imbalance is partly triggered by a decrease of bone protein matrix and mineral content [14a].
The aim of a Wnt pathway activator is to discover compounds to increase bone development and especially slow down osteoporosis. In this regard, a phenotypic bone formation assay is useful to evaluate compounds for their ability to differentiate murine C2C12 cells, which possess multi-lineage potential and an osteoblast-like phenotype through β-catenin-dependent stimulation of alkaline phosphatase activity [14b]. Compounds of therapeutic interest increase osteoblast formation in rodent and human cellular assays through a non-glycogen synthase kinase (GSK) mechanism. Tetracyclic lactones 2a, 3a, 4a and 5a displayed interesting activity by activation of the Wnt pathway through stimulation of alkaline phosphatase activity but not β-catenin (Table 4). The lactones of the series a with side chains greater than 5 carbon atoms were practically inactive. The compounds of the b series did not show activity either. 2.2.5. Other biological properties [15]. The tau protein is related to neuroprotection and Alzheimer's disease. This protein acts in vivo to induce tubulin assembly and stabilize microtubules, and the activity may be necessary but not sufficient for neuronal morphogenesis. Lactone 2a so do 3a and 4a showed important inhibition of Tau protein translation at 40 µM concentration. In addition, 2a presents inhibition of other haplotypes such as AngptI8 (inhibit lipoprotein lipase and decrease plasma triglyceride) APOC3 (decrease of triglycerides), Nav1.7 (antinociceptive activity related to sodium ion channel) [15] and PCSK9 (cholesterol-lowering drugs [16]) at 40 µM (Table 5).

Activity under hypoxic conditions and 3D cell cultures
Solid tumors are often characterized by areas with inadequate blood supply, lower oxygen tension and nutrient supply. Tumor cells distant from the existing blood vascular capillaries (> 100 µm) often become hypoxic and resistant to both radio and chemotherapy. It is also known that hypoxia contributes to the progression of a more aggressive phenotype via activation of several genes including induction of hypoxia-inducible factor 1-alpha (HIF1-α) and subsequently VEGF that contributes to angiogenesis [17][18][19][20].
As compounds 2a and 4a were consistently shown to produce a biological response in all assays investigated and possessed antiangiogenic properties (Table 2)

CYP1A1 bioactivation
The natural product ellipticine (1) exerts its antiproliferative activity in part via CYP1A1 bioactivation leading to a metabolite capable of covalently binding DNA. CYP1A1 has been shown to be differentially overexpressed in CRC while its expression in normal surrounding epithelia is relatively low [21]. Given scaffold similarity of the lactones presented in this study with ellipticine, interrogation of their potential for CYP1A1 bioactivation were assessed in the CHO/CHO1A1 isogenic cell line pair.
Analogue 2a and the thiofuranone analogue 4a were chosen as a suitable pair for this investigation.
Lactone 2a was found to be 3.7-fold more antiproliferative in CHO1A1 cells compared with parental CHO cells while the thiofuranone 4a had much lower potentiation effect in the presence of CYP1A1 (Table 7). Ellipticine was shown to be approximately 15-fold more potent in CHO1A1 cells, which is in accordance with previous studies that have confirmed CYP1A1 as a bioactivating enzyme responsible for generating a metabolite capable of damaging DNA via adduct formation [22].

Molecular Modeling Studies
Compound 2a (L2a) was selected as a representative member of the diverse benzodioxine analogues of the most active series a described above for the determination of its most probable binding site to KRas.
To achieve this objective, we thoroughly explored the conformational space of the KRas   Table S1 and Figures S10, S15 and S17). The third one is composed by those complexes that present good convergence only for the last 5 ns, but with |∆G b | smaller than 25 kcal/mol (Case [3] in Table S1 and Figures S2, S5, S12 and S13). Finally, the last group corresponds to those complexes that present good convergence during more than 5 ns, but with |∆G b | smaller than 25 kcal/mol (Case [4] in Table S1 and Figures S1, S3, S4, S6, S11 and S14). All complexes included in these four groups (Colored red in Table S1) were discarded from future analysis and only the remaining eight complexes which showed better convergence and |∆G b | values greater than 25 kcal/mol were retained (Yellow, brown and green colors in Table S1).
In a second step, the length of the MD of the eight best complexes was extended to 100 ns to permit a better-induced fit of the protein. In this step, only one structure was rejected because the binding energy for the last 5 ns was worse than -25 kcal/mol (Brown color in Table S1 and Figure S18). Thus, to distinguish between the remaining seven poses the length of the MD was extended to 300 ns. Now, poses 1 and 3 of pocket 1 and poses 1 and 3 of pocket 2 were discarded because of their no-binding or low binding energy (Yellow color in Table S1 and Figures S19 to S22). Finally, for the three remaining poses (Green color in Table S1) the length of the MD was extended to 700 ns. Pose 4 of pocket 1 was clearly stable during the last 300 ns; however, it was discarded because of its worse binding energy as compared to those of the remaining two poses ( Figure S23). The two remaining poses present very similar (both PB and GB) final binding energies (Green color in Table S1 and Figures   Finally, we introduced a methoxy group as a substituent of the first aromatic ring of the tetracyclic lactone (at C-7 position, compound 7a) to compare its stability with the original 2a ligand. As expected, both the MMPBSA and MMGBSA approaches suggest that the substituted compound has a smaller binding energy than the original one although the difference is small; 0.8 kcal/mol and 0.2 kcal/mol for the MMGBSA and MMPBSA methods, respectively. The convergence of ∆G b is shown in figure S28 (supporting information). The final structures of both compounds are displayed in figure S29 (supporting information), in which we can see that the position of the 7a modified compound is displaced from the original starting point. which is known to be expressed in a number of cancer types including colorectal carcinoma (CRC) [25].
As such, lactone 2a represents a good starting point for further medicinal chemistry aimed at generating novel agents to probe the colorectal carcinoma microenvironment.

Melting points (mp) were determined on a Gallenkamp-apparatus (MFB-595010M) with interior
thermometer using open capillary tubes and are reported after correction. IR spectra were obtained using reagents were of high quality or were purified before use. All organic solvents were of analytical grade or were purified by standard procedures before use.

11-Methyl-7-methoxy-4-pentyl-1-oxo-3(H)-isobenzofuro[5,6-b][1,4]benzodioxine (7b).
After 48 h, the cells were counted in triplicate using a Coultronics Coulter Counter and the results were expressed as the compound concentration, which inhibited cell growth by 50% as compared to the control (IC 50 ). The IC 50 values were calculated from regression lines obtained from the probit of the percent cell growth inhibition plotted as a function of the logarithm of the dose. [26]. have the same molecular dynamics length for both compounds, the molecular dynamics of the L2a system was also extended to 900 ns.

Molecular dynamics (MD) simulations
All the KRas-L2a complexes were prepared using the same protocol. The LEaP module of Ambertools v.16 [35] was used to create a cubic box of TIP3P [36] waters with a minimum distance between any atom of the system and the edge of the box of 15 Å and removing water molecules closer than 2.2 Å to any of the complex atoms. Next, the system was neutralized with counterions following a grid-shaped procedure for mapping electrostatic potential surface. Finally, the ParmEd program was used to produce hydrogen mass repartitioning to allow an integration time of 4 fs [37]. The prepared structures were heated to 300 K during 200 ps at a constant rate of 30 K/20 ps with harmonic restrains of 1 kcal mol -1 Å -2 in the protein main atoms. Next, 400 ps at constant pressure were performed to increase the system density using harmonic restrains of 1 kcal mol -1 Å -2 in the protein main atoms. Finally, different production molecular dynamic simulations were done under the canonical ensemble using a Langevin [38] thermostat with a collision frequency of 3 ps -1 for temperature control. Long-range electrostatic energy was computed using the Particle Mesh Ewald summation method [39] with a cutoff of 9 Å for non-bonded interactions. SHAKE algorithm [40] was used to constrain the bonds involving the hydrogen atoms.

MMPB/GBSA Calculations
The binding free energy, ∆G bind , for all the studied protein-ligand systems was evaluated using the MMGB/PBSA (molecular mechanics generalized Born/ Poisson Boltzmann surface area) methods as implemented in the MMPBSA.py module of Ambertools v.16 [35]. In this approach, the total binding free energy is calculated as ∆G binding = ∆H 0 gas -T∆S 0 + ∆G solv , where ∆H 0 gas , the gas phase interaction energy, is calculated as the sum of the internal energy (∆H 0 int ) and two non-bonded terms corresponding to the van der Waals (∆H 0 vdW ) and electrostatic (∆H 0 elec ) molecular mechanics energies: ∆H 0 gas = ∆H 0 int + ∆H 0 vdW + ∆H 0 elec . The solvation free energy (∆G solv ) is obtained by summing the polar (∆G polar ) and nonpolar (∆G nonpolar ) terms: ∆G solv = ∆G polar + ∆G nonpolar . For the MMGBSA method the ∆G polar is calculated using the Onufriev-Bashford-Case (OBC) generalized Born [41] (igb=2) whereas the nonpolar contribution (∆G nonpolar ) is calculated from the solvent accessible surface area (SASA) according to the equation: ∆G nonpolar = γ SASA + β where the values for γ and β were set to 0.0072 kcal mol -1 Å -2 and 0 kcal mol -1 [42]. On the other hand, for the MMPBSA method the ∆G polar is calculated using the Poisson Boltzmann approach while the nonpolar contribution (∆G nonpolar ) is obtained using γ = 0.00542 kcal mol -1 Å -2 and β = 0.92 kcal mol -1 [43]. In both methods, values for interior and exterior dielectric constants were set to 1 and 80, respectively. In order to assess the convergence and stability of the ∆G bind values along the time, MMPB/GBSA computations were performed for the complete MD simulations using 100 structures each nanosecond, extracted each 10 ps, using the cpptraj program as implemented in Ambertools v.16 [35]. The contribution of each KRas residue to the total binding free energy was obtained using the MMGBSA decomposition method [44].