Selective inhibition of carbonic anhydrase IX and XII by coumarin and psoralen derivatives

Abstract A small library of coumarin and their psoralen analogues EMAC10157a-b-d-g and EMAC10160a-b-d-g has been designed and synthesised to investigate the effect of structural modifications on their inhibition ability and selectivity profile towards carbonic anhydrase isoforms I, II, IX, and XII. None of the new compounds exhibited activity towards hCA I and II isozymes. Conversely, both coumarin and psoralen derivatives were active against tumour associated isoforms IX and XII in the low micromolar or nanomolar range of concentration. These data further corroborate our previous findings on analogous derivatives, confirming that both coumarins and psoralens are interesting scaffolds for the design of isozyme selective hCA inhibitors.


Introduction
The development of cancer is a complex multifactorial process, involving many cellular adaptations and signal transduction pathways [1][2][3][4] . In solid tumours, cancer cells must survive in a low oxygen concentration environment, due to the rapid cellular proliferation and to the impossibility to promptly supply an adequate vascularisation 5,6 . Indeed, many pathways are involved in the hypoxia survival mechanism [7][8][9][10][11] and they all concur in helping cancer cells to escape from apoptosis. These pathways have been investigated in depth and might be inhibited by relatively new classes of anticancer drugs, to contrast the angiogenesis process, such as VEGFR (sunitinib, sorafenib), VEGF directed monoclonal antibodies (bevacizumab), and mTOR (everolimus, temsirolimus) inhibitors [12][13][14][15][16][17] . In this contest, the key role of human carbonic anhydrases, a class of metalloproteins that catalyse the reversible conversion of carbon dioxide to bicarbonate and protons [18][19][20] , has been outlined. In particular, two membrane isoforms, namely hCA IX and XII, are mainly involved in cancer proliferation and invasion [21][22][23][24][25][26][27][28][29][30][31] . Not surprisingly, several inhibitors of membrane bounded hCA isozymes, with diverse structures and mechanisms of action, have been designed and investigated so far. In this respect, both synthetic and natural coumarin derivatives have already demonstrated to possess high selectivity and activity towards specific hCA isozymes [32][33][34][35][36] . Moreover, their interaction and binding mode on this class of metalloenzymes have been investigated in depth 37,38 . Besides, it should be considered that coumarins have been reported to interact with several cancer druggable targets. In particular, coumarin derivatives have shown a variety of biological activities such as CK2 inhibitors 39 , EGFR 40 , PI3K-AKT-mTOR signalling inhibitors [41][42][43] . Furthermore, their anticancer potential, tumour targets, diverse mechanisms of action as well as their advantages and disadvantages have been recently reviewed 44 . In continuation with our previous work and prompted by these considerations, to further explore the influences of structural modifications on the coumarin and psoralene core on the activity and selectivity towards membrane-bound hCA isozyme, we have designed and synthesised a small library of methyl-2-[4methyl-2-oxo-7-(2-oxo-2-arylethoxy)-8-propylchromen-3-yl]acetate and 2-(5-methyl-7-oxo-3-aryl-9-propyl-7H-furo[3,2-g]chromen-6yl)acetic acid derivatives were a propyl group in the position 8 or 9 has been introduced compared to the previously synthesised derivatives.

Materials and apparatus
Starting materials and reagents were obtained from commercial suppliers and were used without purification. All melting points were determined on a Stuart SMP11 melting points apparatus and are uncorrected. Melting points, the yield of reactions, and analytical data of derivatives EMAC10157a-b-d-g and EMAC10160a-bd-g are reported in Table 1. 1 H-NMR and 13 C-NMR spectra ( Table 2) were registered on a Bruker AMX 400 MHz (chemical shifts in d values) operating at 400 MHz and 100 MHz, respectively. All samples were measured in DMSO. Chemical shifts are reported referenced to the solvent in which they were measured. Coupling constants J are expressed in hertz (Hz). Elemental analyses were obtained on a Perkin-Elmer 240 B microanalyser. Analytical data of the synthesised compounds are in agreement within ±0.4% of the theoretical values. TLC chromatography was performed using silica gel plates (Merck F 254), spots were visualised by UV light.
General procedure for the synthesis of compound EMAC10157 a-b-d-g and EMAC10160 a-b-d-g Synthesis of methyl 2-(7-hydroxy-4-methyl-2-oxo-8-propyl-2Hchromen-3-yl)acetate A mixture of propylresorcinol (1 eq.), dimethylacetylsuccinate (1 eq.) and sulphuric acid 98% (2.8 eq.) was vigorously stirred at room temperature. The progression of the reaction was monitored by TLC, using ethyl acetate/n-hexane 2:1. After 30 min a homogeneous sticky solid was obtained which was dissolved in methanol and poured into ice water. The mixture was stirred until ice melting and then filtered off to obtain a light yellow solid. The crude product was washed with ethyl ether giving a white powder that was crystallised from methanol.
The solution was cooled to room temperature and poured into ice water. Concentrated HCl was added to the solution, obtaining a suspension that was filtered and crystallised. Analytical and spectral data of compounds EMAC10157 a-b-dg and EMAC10160 a-b-d-g are reported in Tables 1 and 2. 1 H-NMR and 13 C-NMR spectra are reported in Supplementary material ( Figures S1-S16).

Biological activity
Carbonic anhydrase inhibition assay The CA catalysed CO 2 hydration/inhibition was measured by using a stopped-flow instrument (Applied Photophysics, Oxford, U.K.) as the method described earlier 45 . Initial rates of the CA-catalysed CO 2 hydration reaction were followed for 10-100 s. The CO 2 concentrations ranged from 1.7 to 17 mM for the determination of the inhibition constants. For each inhibitor, at least six traces of the initial 5-10% of the reaction were used for assessing the initial velocity. The uncatalyzed rates were subtracted from the total observed rates. Stock solutions of inhibitors (10 mM) and dilutions up to 0.01 nM were prepared in distilled-deionized water. Inhibitor and enzyme solutions were preincubated together for 15 min at room temperature prior to assay, in order to allow for the formation of the E-I complex. The inhibition constants were obtained by non-linear least-squares methods using PRISM 3 as reported earlier, and represent the mean from at least three different determinations [46][47][48] . hCA I, hCA II, hCA IX (catalytic domain), and hCA XII (catalytic domain) were recombinant proteins produced inhouse using our standardised protocol and their concentration in the assay system was in the range of 3-10 nM (and even lower for highly effective, sub-nanomolar inhibitors) [46][47][48] .

Ligand preparation
The ligands were built using the Maestro GUI 49 . The most stable conformation has been determined by molecular mechanics conformational analysis performed by Macromodel software version 9.2 50 . using the Merck Molecular Force Fields (MMFFs) 51 and GB/ SA water implicit solvation model 52 , Polak-Ribier Conjugate Gradient (PRCG) method, 5000 iterations, and a convergence criterion of 0.05 kcal/(mol Å). All the other parameters were left as default.

Protein preparation
The coordinates for hCA isoforms enzymes were taken from the RCSB Protein Data Bank 53 (PDB codes 3k34 54 , for isoform II; 5fl4 55 , for isoform IX and 5msa, for isoform XII). The proteins were prepared by using the Maestro Protein Preparation Wizard 49 . Original water molecules were removed.

Docking protocol
Molecular docking studies were performed using QMPL workflow protocol 56 . Grids were defined around the refined structure by centring on crystallised ligands. The other settings were left as default.

Post docking protocol
To better consider, the induced fit phenomena, the most energy favoured generated complexes were fully optimised with the OPLS2005 force field in GB/SA implicit water. The optimisation process was performed setting 10,000 steps interactions up to the derivative convergence criterion equal to 0.05 kJ/(molÅ). The resulting complexes were considered for the binding modes graphical analysis with Pymol 57 and Maestro.
All the synthesised compounds show a propyl substituent in the position 8 or 9 of the coumarin and psoralen nucleus, respectively. As in previously reported derivatives, a methylene carboxylic group was placed in position 3 of the chromene which may lead to the formation of a bidentate chelator of the Zn 2þ ion in the catalytic pocket, due to the hCA esterase activity 32,34,63 on the dihydropyranone ring, as illustrated in Figure 2.
The synthesis of the new derivatives is reported in Scheme 1. It consists of the H 2 SO 4 mediated Pechman condensation of dimethylacetylsuccinate and 2-propylresorcinol at room temperature in solvent-free conditions, followed by a Williamson reaction of the phenolic group in the position 7 of the chromene ring with the appropriate a-haloketone in dry acetone and K 2 CO 3 , to generate the asymmetric ether which will lead to the furocoumarin formation by intramolecular electrophilic substitution, mediated by refluxing in NaOH 1 N water solution.
EMAC10157 a-b-d-g-and EMAC10160 a-b-d-g were characterised employing analytical and spectroscopic methods and the results are summarised in Tables 1 and 2. Compounds were then submitted to enzymatic evaluation towards hCA I, II, IX, and XII. The results are reported in Table 3. Interestingly, none of the new EMAC derivatives exhibited any inhibition activity towards hCA I and II isozymes. On the contrary, all of them are submicromolar inhibitors of the hCA membrane isoforms IX and XII. This behaviour is in agreement with our previous findings and with the generally observed selectivity profile of coumarin derivatives 32,33,35,37,38,63 .
Within the tested compounds, EMAC10157 series was the most active towards the hCA IX and XII isoforms. Therefore, its mechanism of action was investigated in more detail by docking experiments followed by energy minimisation of the obtained complexes.
In particular, considering that an interesting hCA esterase mediated inhibition mechanism was recently reported for coumarin derivatives 32,33,63 , we firstly investigated this aspect. Hence, the most promising compound, EMAC10157g, was submitted to docking experiments to evaluate if the dihydropyranone ring of the chromene moiety could reach the bottom of the catalytic cavity of hCA II, IX and XII and, therefore, be hydrolysed by the Zn 2þ activated water molecule, which acts as a very potent nucleophile (Figure 2(a)).
Confirming the selectivity profile already observed in previously investigated analogues docking experiment in hCA II enzyme showed that the EMAC10157g tail is too bulky to access hCA II  cavity, while, in isoforms IX and XII, the dihydropyranone portion of EMAC10157g was able to bind close enough to the Zn 2þ ion (Figures 2(a) and 3(a)) cofactor to undergo the esterase reaction.
Hence, the two open cinnamic acid derivative (Figure 2(a)) configurations (E/Z) were subjected to docking experiments to predict the putative binding mode of the hydrolysed forms in hCA IX and hCA XII (Figures 3(b,c) and 4(b,c).
The predicted affinity of the open compounds was estimated to be better than that of the closed ones. What seems important is the Zn 2þ chelation and the interactions between the newly Table 3. Inhibition data towards hCA I, II, IX, and XII of compounds EMAC10157a-b-d-g and EMAC10160 a-b-d-g.
Compound R  hCA-IX -EMAC10157g-openZ and the relative 2D representation of the complexes stabilising interactions with the binding site residues represented with different colour depending on their chemical-physical properties: green, hydrophobic; cyan, polar; violet, positive; red, negative charged residues; grey, metal atoms. Instead, magenta arrows indicate the formation of a hydrogen bond between protein and ligand, while grey lines indicate the interaction with the complexed ion.
formed carboxylate moiety and the methyl ester with the surrounding residues in the catalytic site.
The Y shape of the compound, with propyl and -biphenyl-2oxoethoxy moiety being the arms, helps to maintain the selectivity but, ultimately, decreases the activity, compared to previously investigated psoralens 32 .
Biphenyl moiety accommodation is well tolerated due to the wide cavity exposed to the solvent, as shown by other hCA inhibitors with bulky tails. In fact, the interactions between this group with adjacent hydrophobic residues stabilise the complexes.
In summary, the computational methods helped to rationalise the good activity of the investigated compound towards the hCA IX and hCA XII isoform and to suggest a reasonable mechanism of action that could be further investigated to be confirmed. If verified, this compound can be considered as a new prodrug candidate with acceptable oral bioavailability properties (Figure 2(b)), and with good drug-like properties. Nevertheless, considering the encouraging predicted ADME properties 64 and the activity data, we are further optimising this scaffold.

Conclusions
We have designed and synthesised a series of methyl-2-[4-methyl-2-oxo-7-(2-oxo-2-arylethoxyi)-8-propylchromen-3-yl]acetate and 2-(5methyl-7-oxo-3-aryl-9-propyl-7H-furo[3,2-g]chromen-6-yl)acetic acid, and evaluate their activity on hCA I, II, IX, and XII isozymes. As a confirmation of the literature reported chromene derivatives selectivity profile, none of the investigated compounds was able to inhibit the off-target I and II isoforms of hCA. On the contrary, they all inhibit the membrane isozymes hCA IX and XII and further corroborate the reported data on chromene derivatives. Considering the acceptable ADME prediction profile and the high potential of coumarin derivatives, not only as hCA selective inhibitors but also as potential multitarget anticancer agents, these data prompted us to further investigate this scaffold to optimise both the activity and the isozyme selectivity.