Coumarins-lipophilic cations conjugates: Efficient mitocans targeting carbonic anhydrases

Bioorganic Chemistry 145 (2024) 107168 Available online 6 February 2024 0045-2068/© 2024 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Coumarins-lipophilic cations conjugates: Efficient mitocans targeting carbonic anhydrases Alma Fuentes-Aguilar a , b , Aday Gonz´ alez-Bakker c , Mirna Jovanovi´ c d , Sofija Jovanovi´ c Stojanov d , Adri´ an Puerta c , Adriana Gargano e , Jelena Dini´ c d , Jos´ e L. VegaB´ aez a , Pen´ elope Merino-Montiel a , Sara Montiel-Smith a , Stefano Alcaro e , f , g , Alessio Nocentini h , Milica Peˇ si´ c d , * , Claudiu T. Supuran h , * , Jos´ e M. Padr´ on c , * , Jos´ e G. Fern´ andez-Bola˜ nos b , ´ Oscar L´ opez b , * a Facultad de Ciencias Químicas, Benem´ erita Universidad Aut´ onoma de Puebla, Ciudad Universitaria, 72570 Puebla, PUE, Mexico b Departamento de Química Org´ anica, Facultad de Química, Universidad de Sevilla, Apartado 1203, E-41071 Seville, Spain c BioLab, Instituto Universitario de Bio-Org´ anica “Antonio Gonz´ alez”, Universidad de la Laguna, C/ Astrofísico Francisco S´ anchez 2, 38206 La Laguna, Spain d Institute for Biological Research “Siniˇ sa Stankovi´ c”, National Institute of the Republic of Serbia, University of Belgrade, Despota Stefana 142, 11108 Belgrade, Serbia e Dipartimento di Scienze della Salute, Universit` a “Magna Græcia” di Catanzaro, Campus Universitario “S. Venuta”, Viale Europa, 88100 Catanzaro, Italy f Net4Science Academic Spinoff, Universit` a “Magna Græcia” di Catanzaro, Campus Universitario “S. Venuta”, Viale Europa, 88100 Catanzaro, Italy g Associazione CRISEA – Centro di Ricerca e Servizi Avanzati per l’Innovazione Rurale, Localit` a Condoleo, 88055 Belcastro (CZ), Italy h NEUROFARBA Department, Sezione di Scienze Farmaceutiche e Nutraceutiche, University of Florence, 50019 Florence, Italy ARTICLE INFO Keywords: Coumarin Mitochondriotropic agent Carbonic anhydrase inhibitor Cytostatic agent Apoptosis ABSTRACT Being aware of the need to develop more efficient therapies against cancer, herein we disclose an innovative approach for the design of selective antiproliferative agents. We have accomplished the conjugation of a coumarin fragment with lipophilic cations (triphenylphosphonium salts, guanidinium) for providing mitochondriotropic agents that simultaneously target also carbonic anhydrases IX and XII, involved in the development and progression of cancer. The new compounds prepared herein turned out to be strong inhibitors of carbonic anhydrases IX and XII of human origin (low-to-mid nM range), also endowed with high selectivity, exhibiting negligible activity towards cytosolic CA isoforms. Key interactions with the enzyme were analysed using docking and molecular dynamics simulations. Regarding their in vitro antiproliferative activities, an increase of the tether length connecting both pharmacophores led to a clear improvement in potency, reaching the submicromolar range for the lead compounds, and an outstanding selectivity towards tumour cell lines (S.I. up to >357). Cytotoxic effects were also analysed on MDR cell lines under hypoxic and normoxic conditions. Chemoresistance exhibited by phosphonium salts, and not by guanidines, against MDR cells was based on the fact that the former were found to be substrates of Pglycoprotein (P-gp), the pump responsible for extruding foreign chemicals; this situation was reversed by administrating tariquidar, a third generation P-gp inhibitor. Moreover, phosphonium salts provoked a profound depolarization of mitochondria membranes from tumour cells, thus probably compromising their oxidative metabolism. To gain insight into the mode of action of title compounds, continuous live cell microscopy was employed; interestingly, this technique revealed two different antiproliferative mechanisms for both families of mitocans. Whereas phosphonium salts had a cytostatic effect, blocking cell division, guanidines led to cell death via apoptosis. * Corresponding authors. E-mail addresses: [email protected] (M. Peˇ si´ c), [email protected] (C.T. Supuran), [email protected] (J.M. Padr´ on), [email protected] (´ O. L´ opez). Contents lists available at ScienceDirect Bioorganic Chemistry journal homepage: www.elsevier.com/locate/bioorg https://doi.org/10.1016/j.bioorg.2024.107168 Received 21 November 2023; Received in revised form 22 January 2024; Accepted 30 January 2024 Bioorganic Chemistry 145 (2024) 107168 2 1. Introduction Cancer is currently considered to be one most complex diseases [1], with an etiology not completely elucidated [2]. Undoubtedly, there have been outstanding advances in cancer treatment in the last few decades [3], both for therapeutic and diagnosis purposes. Thus, alone themselves, or combined with classical treatments (chemotherapy, radiotherapy, surgery), nanodelivery carriers (e.g. nanoparticles [4], extracellular vesicles [5]), adjuvants/neoadjuvants [6], targeted [7] and gene [8] therapies, magnetic hyperthermia [9], or immunotherapy [10], including cancer vaccines [11] and monoclonal antibodies [12] have increased life expectancy of patients. Despite that, most of the current treatments are still based on the use of chemotherapeutic agents, which in many occasions are devoid of selectivity, leading to severe side-effects [13], and also frequently undergo chemoresistance [14]. Among the numerous potential targets against cancer, mitochondria have become an attractive one in Medicinal Chemistry. Mitochondria are subcellular organelles that play key biological functions: they do not only produce ATP through the oxidative phosphorylation pathway, what constitutes 95 % of cell energy supply, but they also keep ion homeostasis, mediate in the biosynthesis of macromolecule intermediates, regulate apoptosis, and scavenge Reactive Oxygen Species (ROS) [15]; moreover, there is a close relationship between dysfunctional mitochondria and tumorigenesis, through a complex crosstalk with other organelles [16]. This has triggered the development of drugs targeting mitochondria [17] which have been proved to be useful for diminishing the injuries caused by ischaemia, or for the treatment of inflammation, neurodegenerative and lung diseases, or cancer, among others. Compounds showing high affinity towards mitochondria, because of their particular physicochemical properties, are called mitochondriotropic agents [18]; when such compounds exhibit anticancer properties, they are called mitocans (mitochondria +cancer) [18,19]. Although there are a series of scaffolds acting as selective vehicles to mitochondria, like dequalinium vesicles, or mitochondria-penetrating peptides (e.g. Szeto-Schiller peptides), the most common ones are delocalized lipophilic cations (triphenylphosphonium (TPP + ), guanidinium, rhodamine, heterocyclic aromatic cations) [18,20]; the latter ones, due to the large and negative mitochondrial membrane potential (ΔΨ mit =150–180 mV) compared to plasma (ΔΨ plasma =30–60 mV), largely accumulate within mitochondria matrix (up to 500-fold), according to the Nernst equation [20]. This effect is magnified in cancer cells, due to hyperpolarization of their Fig. 1. Coumarin-lipophilic cations hybrids accessed herein. Scheme 1. Retrosynthetic analysis for accessing targeted coumarin-appended phosphonium salts and guanidines. A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 3 membranes [21]. Another target we have exploited herein are carbonic anhydrases (CAs); they comprise a series of metalloenzymes that usually bear Zn(II) as the prosthetic group. CAs catalyse CO 2 hydration to furnish HCO 3 – + H + [22], a crucial reaction within living beings. Within the phylogenetic tree, only a limited number of Gram-negative bacteria and one Archaeon lack CAs [22]. CAs are distributed along eight gene families [23], α -CA being the only one encoded in mammals along 15 different isoforms (12 of them catalytically active) [24], and take part in a plethora of biological pathways, like ureagenesis, gluconeogenesis, lipogenesis, pathogen virulence, and pH regulation [25]. The three remaining mammal isoforms (hCA VIII, X and XI) have no known biological role, and are known as CA-related proteins [26]. Currently, inhibition of CAs constitute a validated target for treating glaucoma [27], bacterial infections [28], obesity [29], or cancer [30]. Even, an involvement of CAs in Alzheimer’s disease has been postulated [31]. In connection with cancer, SLC-0111, a ureido-containing sulfonamide was found to strongly inhibit CA IX and XII isoforms; recently, it entered clinical trials (phase II) against pancreatic ductal adenocarcinoma with overexpression of CA IX, in a combination therapy [30]. α -CAs IX and XII are up-regulated in hypoxic metastatic tumours, promoting metastasis and tumour growth [30]. Herein, we envisioned the possibility of preparing new chemotherapeutic agents endowed with enhanced selectivity, and therefore, with potential reduced side-effects. For that purpose, an innovative dualtargeting approach has been followed, by combination of a mitochondria-targeted scaffold, which acts as a selective vehicle towards cancer cells mitochondria, and a pharmacophore acting on carbonic anhydrases (CAs). 2. Results and discussion 2.1. Chemistry As aforementioned, in this manuscript we have appended a phosphonium salt or a guanidine scaffold (delocalized lipophilic cations) to a coumarin core for achieving a novel family of tumour-targeting compounds. With this kind of structure, we pursued a dual effect on tumour cells; on the one hand, the coumarin scaffold could selectively inhibit CAs involved in tumour progression, namely CA IX and XII. Although the most typical chemotype used for inhibiting this metalloenzymes are sulfonamides and their isosters, we have selected coumarins instead, due to their frequent high selectivity towards the tumour-associated CAs isoforms [32,33]. The lipophilic cation could act as a vehicle for the selective vectorization of the drug to the tumour cells, and also cause disruption of mitochondrial membranes of malignant cells (Fig. 1). Access to such kind of derivatives involves (Scheme 1) a Pechmann condensation [34] between resorcinol and different β-ketoesters for accessing the coumarin core; modification of C-3 and C-4 substituents might modulate the potency and selectivity in both, CA inhibition and antiproliferative properties. Moreover, it has also been reported that incorporation of alkyl residues on C-3 and C-4 can reduce the hepatotoxicity of this kind of compounds by hindering the P-450-mediated formation of transient coumarin 3,4-epoxides in their metabolic routes [35]. Subsequent O-alkylation with different α , ω -dibromoalkanes can Scheme 2. Preparation of phosphonium salts 5b-g. Scheme 3. Plausible mechanism for the spontaneous degradation of phosphonium derivative 6. A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 4 Scheme 4. Preparation of coumarin-derived guanidines 14a,c-g. Scheme 5. Plausible mechanism for the formation of urea 11 via iodine-mediated desulfurization of thiourea 10c. A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 5 provide a flexible tether that might improve the fitting of the pharmacophore within hCA active site and improve the non-covalent interactions. This intermediate was used for the derivatization into phosphonium salts and guanidines, as depicted in Schemes 2 and 4. Thus, the preparation of phosphonium salts 5b-g is depicted in Scheme 2; reaction of resorcinol with β-ketoesters 2 catalysed by H 2 SO 4 afforded coumarins 3, either mono-substituted on C-4 position (3a,b), or disubstituted on C-3 and C-4 positions (3c,d). Subsequent Williamson etherification reaction under basic conditions (K 2 CO 3 ) involving the free OH on C-7 with an excess (10 equiv.) of a series of linear α , ω -dibromoalkanes (n =2, 3, 5, 12) furnished bromo-derivatives 4a-g. Biological properties might depend on the distance between the coumarin and the phosphonium moieties; moreover, increase of conformational flexibility might also favour the appropriate fitting within the CA active site. Final nucleophilic displacement with PPh 3 on 4a-g (Scheme 2) in acetonitrile using a Fischer-Porter tube (T =120 ◦C) afforded phosphonium salts 5b-g (59 %–quantitative yields). Attempts to prepare the phosphonium salt with a 2-carbon tether (starting from bromo-coumarin 4a) failed; unexpectedly, reaction of PPh 3 with 4a using the reaction conditions depicted in Scheme 2 furnished coumarin 3a, with loss of the bromoalkyl chain. Scheme 3 depicts the tentative mechanism for explaining such observation. Therefore, the anchimeric assistance of the oxygen atom on C-7 was considered, whose intramolecular nucleophilic attack con position C-2 of the ethyl chain would afford a transient charged-epoxide. Opening of such epoxide mediated by the bromide anion would lead to the formation of a vinyl ether that could be hydrolysed in the presence of the in situ generated HBr to give coumarin 3a, with a free OH on C-7 position. The presence of the triphenyl phosphonium moiety in 5b-g was easily evidenced by NMR spectroscopy; thus, besides the incorporation of 15 aromatic protons in 1 H NMR spectra, phosphorous provoked the splitting of vicinal aromatic and aliphatic carbon signals. For the preparation of coumarin-derived guanidines 14a,c–g, we envisioned the use of thioureas 10a,c–g as the key intermediates (Scheme 4). Such derivatives can be easily accessed starting from bromo-coumarins 4a,c–g by first, NaN 3 -promoted nucleophilic substitution, followed by heterogeneous hydrogenolysis and reaction of the transient terminal amine with commercially-available p-tolyl isothiocyanate (Scheme 4). Attempts to carry out the coupling with refluxing EtOH afforded 10c in a 63 % yield with long reaction times (18 h) and partial degradation of the isothiocyanate. Changing the solvent to MeCN and rt provoked extensive degradation of the isothiocyanate and large amounts of remaining 9c were observed. Shift to refluxing THF afforded 10c in an excellent yield (82 %) and short reaction times (3 h); such optimized conditions were extended to the preparation of the remaining thioureas 10. The formation of 10a,c–g was evidenced by resonance at roughly 180 ppm in 13 C NMR, attributed to the thioxo (C – – S) group. Once the thioureas were prepared, several conditions were attempted to optimize their transformation into targeted guanidines 14 upon a desulfurization step. Firstly, thiourea 10a was treated with H 2 O 2 (Scheme 4, conditions a), in the presence of p-toluidine and tetrabutylammonium iodide (TBAI), under similar conditions reported [36] by Yadav and co-workers for the transformation of o-phenolic thioureas into 2-aminobenzoxazoles. Isosteric urea 11 was obtained in an acceptable yield (56 %) instead the expected guanidine. Scheme 5 depicts a plausible for the formation of such compound; reaction of H 2 O 2 and TBAI furnishes hypoiodous acid (HIO) which oxidizes the thioureido moiety to give iodinated derivative 15. We postulate that instead of taking place the nucleophilic attack of p-toluidine, it was the more nucleophilic hydroxide anion from tetrabutylammonium hydroxide the one that reacted with 15 to yield intermediate 16, which in turn after protonation, and loss of sulfur and iodide anion, evolved to urea 11. NMR spectra clearly did not show incorporation of the p-toluidine fragment, and resonance of the quaternary carbon shifted from roughly 180 ppm (thiourea) to 156.7 ppm, typical for an ureido motif [37]. In order to avoid the generation of strongly basic conditions obtained by the above methodology, yellow HgO was used for the desulfurization reaction; in our research group, such reagent has been previously used for the preparation of 2-amino-1,3-oxazolines from carbohydrates [38,39] and steroids [40], and also ureas and guanidines derived from the natural sulfoaminoacid taurine [41]. In all cases, and starting from thioureas, reaction was proposed to proceed through a carbodiimide as the key intermediate. Treatment of thiourea 10c with HgO and pTable 1 Inhibition constants (K i , nM) of compounds 5, 14 against hCAs I, II, IV, VII, IX, XII. Compound hCA I hCA II hCA IV hCA VII hCA IX hCA XII 5b R 1 =CH 3 , R 2 = H n =3 >10000 >10000 794 >10000 82.5 80.6 5c R 1 =CH 3 , R 2 = H n =5 >10000 >10000 nt nt 54.1 9.2 5d R 1 =CH 3 , R 2 = H n =12 >10000 >10000 689 >10000 460 88.3 5e R 1 =Ph, R 2 =H n =5 >10000 >10000 796 >10000 408 89.3 5f R 1 =R 2 =CH 3 n =5 >10000 >10000 nt nt 7.8 8.1 5g R 1 =CH 3 , R 2 = Cl n =5 >10000 >10000 nt nt 7.9 58.2 14a R 1 =CH 3 , R 2 = H n =2 >10000 >10000 891 >10000 143 93.6 14c R 1 =CH 3 , R 2 = H n =5 >10000 >10000 nt nt 8.1 619.1 14d R 1 =CH 3 , R 2 = H n =12 >10000 >10000 852 >10000 415 95.5 14e R 1 =Ph, R 2 =H n =5 >10000 >10000 834 >10000 57.8 319 14f R 1 =R 2 =CH 3 n =5 >10000 >10000 958 >10000 685 54.5 14g R 1 =CH 3 , R 2 = Cl n =5 >10000 >10000 792 >10000 313 74.4 AAZ 250.0 12.0 74.0 2.5 25.0 5.7 nt: Not tested. A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 6 toluidine afforded a more polar compound as evidenced by TLC, which was obtained in low yield and as a non-resolved mixture of compounds (Scheme 4, conditions b). Considering that p-toluidine exhibits a moderate nucleophilicity to react with the intermediates generated upon the desulfurization reaction, it was replaced with benzylamine. To our delight, under these conditions (yellow HgO +BnNH 2 ), thiourea 10c was successfully transformed into guanidine 14c in an 84 % yield (Scheme 4, conditions e) after 66 h. Attempts to reduce the reaction time by the use of more reactive intermediates (S-methyl thiouronium salt [42] 12, conditions c; tosylated thiourea [43] 13, conditions d) failed, as no reaction was observed (Scheme 4). Accordingly, conditions e from Scheme 4 were applied to the complete series of compounds, furnishing the corresponding guanidines 14a,c–g in excellent yields (81 %–quantitative) after chromatographic purification. Resonance at roughly 151 ppm, and disappearance of the signal at 180 ppm (C – – S) confirmed the proposed structures. 2.2. Biological assays 2.2.1. Carbonic anhydrase inhibition The 12 mitochondria-directed derivatives prepared herein (phosphonium salts 5b-g, and guanidines 14a,c–g) were assayed as potential inhibitors of tumour-associated CAs, and their activities were compared with acetazolamide (AAZ, positive control). Cytosolic (I, II and VII), and membrane-bound (IV) enzymes were also tested to analyse the selectivity. Inhibition constants (K i ’s) were obtained (Table 1) from the stopped-flow CO 2 hydration protocol. Data in Table 1 allowed the establishment of relevant structure–activity relationships: Fig. 2. Three-dimensional representation of hCA XII (faded green) and hCA IX (faded yellow) in complex with the model compounds 5c,f,g, 14c. In particular, (A, C) constitute the top views and (B, D), the side views of the ligands within the binding site. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 2 G-scores (kcal/mol) of the best dual hits related to both hCA isoforms for coumarins 5c,f,g, 14c. Compound hCA XII hCA IX 5c −6.658 −6.852 14c −6.732 −5.952 5f −4.410 −7.001 5g −4.807 −6.581 A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 7 Fig. 3. 3D plots of the best poses of 5c (A, E; green sticks), 14c (B, F; cyan sticks), 5f (C, G; magenta sticks), and 5g (D, H; orange sticks) into the binding pockets of hCAs XII and IX. The ligands are shown in sticks, hCA IX and hCA XII are shown as a faded yellow-green and faded green ribbon respectively; the key enzyme residues interacting with title compounds are represented as grey carbon sticks. π -Cation, π – π stacking, halogen bond and hydrogen bonding interactions are indicated with dashed lines (green, cyan, violet, and yellow, respectively). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 8 i. Phosphonium salts 5 and guanidines 14 exhibited negligible activity towards cytosolic hCAs (I, II, VII; K i >10,000 nM for all cases). Submicromolar activities were found against CA IV. ii. Low-to-mid nM activities were found for the lead compounds against tumour-related CA IX and XII. Therefore, remarkable selectivities were found for such cases. iii. Regarding the influence of the tether length (compounds 5bd and 14a,c,d), a five-carbon spacer was found to be optimal for inhibiting CA IX in both series of derivatives, and for inhibiting CA XII in the case of the phosphonium salts. iv. Concerning the influence of C-3 and C-4 substitution pattern, disubstitution on these positions with short alkyl fragments (Me) furnished the lead compound (5f), with K i =7.8 and 8.1 nM against CAs IX and XII, respectively. The use of a Ph residue on C4 as a bulky substituent (5e) was detrimental for the inhibition of CA IX. v. In general, phosphonium bromides 5 were stronger CA inhibitors than guanidines 14. vi. A considerable improvement in selectivity was achieved compared to AAZ, an arylsulfonamide in clinical use. 2.2.2. Docking simulations In order to clarify the trends observed for the inhibitory potency of the designed derivatives against the CAs tested, molecular modelling studies were performed; for this purpose, compounds 5c,f,g, 14c were selected. Coumarins have been reported to act as suicide inhibitors of hCAs; they are hydrolysed within the enzyme cavity [44] to give 2hydroxycinnamic acid derivatives which are the actual derivatives occluding the CA cavity entrance [32]. Initially, docking poses and docking scores were evaluated (Fig. 2, Table 2); all compounds showed binding energy values comparable to the ones obtained for known inhibitors (Table S1). Strong interactions with hCA IX and XII binding pocket residues were found, what was evidenced by the high number of contacts detected using the Maestro interface [45] (Tables S2, S3). In connection with hCA XII, the open form of compound 5c established a H-bonding interaction between the COOH moiety and Thr198, a π - π stacking interaction involving the aryl residue of the triphenylphosphonium group, and two different electrostatic interactions between the oxygen atoms of the COOH group and the metal cation (Fig. 3A). In a very similar manner, regarding hCA IX, the open form of compound 5c formed a H-bonding interaction between COOH and Thr200, and a further H-bond between the phenolic hydroxyl group and His94; additionally, two electrostatic interactions between COOH and the Zn 2+ ion were also found (Fig. 3E). The same interactions were also found for compound 14c in both isoforms, but due to the structural diversity of the two molecules, 14c is further stabilised by several additional interactions, specifically a H-bonding interaction between the phenolic hydroxyl group and Gln89, a H-bond interaction between the ether group oxygen and Asn64, a H-bond between the NH + residue and Ser133, a π -cation interaction involving the methyl group of the tolyl moiety and Lys69 in hCA XII (Fig. 3B). In hCA IX, the additional interactions involve the phenolic hydroxyl group, which establishes a two different H-bond with Thr201 and Pro202; moreover, Pro202 also establishes another H-bond with NH + (Fig. 3F). With regard to the open forms of compounds 5f and 5g, there are substantial differences between the two isoforms; in hCA XII they arrange the triphenylphosphonium cation within the binding site and the coumarin moiety outside the site, whereas in hCA IX they continue to arrange themselves with the same binding mode observed for 5c. Specifically, in hCA XII, 5f forms a π -cation interaction between P + and His91, a π - π stacking interaction involving the coumarin scaffold and His66, a H-bond between the free OH on the coumarin residue and Asn64, a H-bond between the carbonyl group of the carboxylic acid moiety and Lys69 (Fig. 3C). In contrast, in hCA IX, 5f forms three different π - π stacking interactions, including one between the coumarin phenyl residue and His94 and two between the central and one lateral aromatic ring of the triphenylphosphonium salt and Trp9. It also establishes a H-bond between the phenolic hydroxyl group and Gln92 and a H-bond involving the carbonyl group of the carboxylic acid moiety and Thr200; in addition, two electrostatic interactions are also formed between the carboxylic group and the Zn 2+ ion (Fig. 3G). Finally, considering hCA XII, 5g establishes two π -cation interactions, one between P + and His91 and one between the central ring of the triphenylphosphonium salt and Lys69, several H-bonds, specifically, one between the phenolic hydroxyl group and His66 and one between the carbonyl group from the COOH moiety and Trp4, a halogen interaction between the chlorine atom and Ser2, and a salt bridge between the hydroxyl of the carboxylic acid moiety and Lys168 (Fig. 3D). In contrast, in hCA IX, 5g forms 4 different H-bonds, including one between the carbonyl of the COOH moiety and Thr200, one between the ether group oxygen and Gln71, and two different ones between the phenolic hydroxyl group, His68 and His94; again, two electrostatic interactions between the carboxylic acid and the Zn 2+ ion were also found (Fig. 3H). Subsequently, the different binding modes and interactions were further investigated through molecular dynamics studies. As far as 5c is concerned, after an initial settling period, it forms a sufficiently stable complex in both isoforms with fluctuations expressed in terms of RMSD that are maintained in the range of 2 Å (Fig. 4). In hCA XII, 5c is stabilised by interactions formed during the course of the dynamics with residues His93, Glu104, Th198 and that are maintained throughout the simulation time. In hCA IX, the complex is strongly stabilised by a large number of interactions that are maintained almost throughout the course of the dynamics, again establishing more than one contact with the same residue, and in particular with Tyr11, Asn66, His96, Glu106, Thr200, Thr201. A similar situation is evidenced by the interaction of 14c with both isoforms, which in both cases forms two stable complexes, characterised Fig. 4. RMSD (Å) trend obtained on the C α of hCA XII (A) and hCA IX (B), in complex with 5c (green line), 14c (cyan line), 5f (magenta line), 5g (orange line). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 9 by not very pronounced fluctuations; in fact, a conspicuous number of stabilising interactions are formed and maintained throughout the 200 ns simulation, in particular with Tyr6, Gln89, His93, Glu104, Thr198 in hCA XII and with residues His94, His96, Glu106, His119, Thr200 and Thr201 in hCA IX. As aforementioned, important differences in binding mode with the two isoforms can be seen for compound 5f In particular, in hCA XII, through the graphic visualisation of the trajectory, the movement of the ligand can be seen, which rotates until the coumarin portion is disposed in the binding site, leaving the triphenylphosphonium moiety on the outside. Although this does not affect the stability of the protein, which is maintained throughout, it justifies not only the fluctuations, but also the number of interactions with Thr199, one of the key residues in the binding pocket, which intensify after the first 100 ns of the dynamics. In contrast, in hCA IX, the ligand is stable within the binding site due to the interactions that form several contacts with residues His96, Glu106, Thr200, Thr201 and are maintained for more than 98 % of the duration of the entire molecular dynamics. This same situation also occurs for compound 5g; in this case, it can be seen that initially in hCA XII the ligand arranges the triphenylphosphonium group in the binding site and that subsequently there is a reversal of the binding mode which allows the coumarin portion to enter the pocket as demonstrated by visual inspection; after the first 150 ns of the simulation, the contacts with residues Thr199 intensify. As is the case for 5f, also for 5g in hCA IX the complex remains stable thanks to the various contacts established with His96, Glu106, Thr200 maintained for over 97 % of the entire simulation. 2.2.3. Antiproliferative activities Potential antiproliferative activities of phosphonium bromides 5 and Table 3 Antiproliferative activity of compounds 5, 14. GI 50 (µM) LC 50 (µM) Compound A549 HBL-100 HeLa SW1573 T-47D WiDr BJ-hTERT (non-small cell lung) (breast) (cervix) (non-small cell lung) (breast) (colon) (human fibroblasts) Sensitive lines Multidrug resistant lines Non-tumour 5b 5.5±0.9 20±3 11±1 16±1 24±1 25±3 n.t. R 1 =CH 3 , R 2 =H n =3 5c 2.1±0.9 3.9±0.3 3.8±1.1 2.7±0.6 3.6±0.4 3.4±0.4 >100 R 1 =CH 3 , R 2 =H n =5 5d 0.36±0.08 0.44±0.09 0.32±0.10 0.28±0.06 0.29±0.06 0.32±0.05 >100 R 1 =CH 3 , R 2 =H n =12 5e 1.6±0.4 1.1±0.3 0.60±0.14 0.95±0.14 0.72±0.32 0.58±0.19 >100 R 1 =Ph, R 2 =H n =5 5f 2.8±0.8 3.0±1.0 2.1±0.1 2.2±0.7 2.4±0.1 2.6±0.7 >100 R 1 =R 2 =CH 3 n =5 5g 2.5±0.4 1.5±0.4 1.6±0.5 1.5±0.4 2.5±1.1 2.6±0.7 >100 R 1 =CH 3 , R 2 =Cl n =5 14a 13±3 15±1 13±1 8.8±1.6 5.2±0.6 4.5±0.1 n.t. R 1 =CH 3 , R 2 =H n =2 14c 6.4±3.1 9.0±2.3 3.8±1.3 6.1±2.0 3.4±0.3 2.8±0.4 >100 R 1 =CH 3 , R 2 =H n =5 14d 1.2±0.1 1.3±0.2 0.85±0.21 0.81±0.09 0.87±0.22 0.41±0.06 >100 R 1 =CH 3 , R 2 =H n =12 14e 1.8±0.2 1.8±0.1 1.7±0.1 1.4±0.1 1.1±0.1 0.46±0.30 >100 R 1 =Ph, R 2 =H n =5 14f 2.0±0.3 2.0±0.1 2.0±0.3 1.9±0.4 1.9±0.3 1.3±0.2 >100 R 1 =R 2 =CH 3 n =5 14g 1.8±0.1 1.7±0.1 1.7±0.1 1.4±0.1 1.5±0.1 0.90±0.28 >100 R 1 =CH 3 , R 2 =Cl n =5 n.t.: Not tested. A. Fuentes-Aguilar et al. Bioorganic Chemistry 145 (2024) 107168 16 [55] S. Ivanov, S.Y. Liao, A. Ivanova, A. Danilkovitch-Miagkova, N. Tarasova, G. Weirich, M.J. Merrill, M.A. Proescholdt, E.H. Oldfield, J. Lee, J. Zavada, A. Waheed, W. Sly, M.I. Lerman, E.J. Stanbridge, Expression of hypoxia-inducible cell-surface transmembrane carbonic anhydrases in human cancer, Am. J. Pathol. 158 (2001) 905–919. [56] Z. Chen, L. Ai, M.Y. Mboge, C. Tu, R. McKenna, K.D. Brown, C.D. Heldermon, S. C. Frost, Differential expression and function of CAIX and CAXII in breast cancer: A comparison between tumorgraft models and cells, PLoS One 13 (2018) e0199476. [57] O. Sedlakova, E. Svastova, M. Takacova, J. Kopacek, J. Pastorek, S. Pastorekova, Carbonic anhydrase IX, a hypoxia-induced catalytic component of the pH regulating machinery in tumors, Front. Physiol. 4 (2014) 400. [58] T. Wang, D. Eskandari, D. Zou, L. Grote, J. Hedner, Increased carbonic anhydrase activity is associated with sleep apnea severity and related hypoxemia, Sleep 38 (2015) 1067–1073. [59] H. Thomas, H.M. Coley, Overcoming multidrug resistance in cancer: an update on the clinical strategy of inhibiting p-glycoprotein, Cancer Control 10 (2003) 159–165. [60] S. Salucci, M. Battistelli, S. Burattini, F. Sbrana, E. Falcieri, Holotomographic microscopy: a new approach to detect apoptotic cell features, Microsc. Res. Tech. 83 (2020) 1464–1470. [61] F.J. Hicke, A. Puerta, J. Dini´ c, M. Peˇ si´ c, J.M. Padr´ on, ´ O. L´ opez, J.G. Fern´ andezBola˜ nos, Straightforward access to novel mitochondriotropics derived from 2-arylethanol as potent and selective antiproliferative agents, Eur. J. Med. Chem. 228 (2022) 113980. [62] M.C. Padilla-P´ erez, M. Elena, A. S´ anchez-Fernandez, A. Gonzalez-Bakker, J. M. Puerta, F. Padr´ on, A.I. Martín-Loro, J.M.G. Arroba, J.M. García Fern´ andez, C. Ortiz Mellet, Fluoro-labelled sp 2 -iminoglycolipids with immunomodulatory properties, Eur. J. Med. Chem. 255 (2023) 115930. A. Fuentes-Aguilar et al.