Click-tambjamines as efficient and tunable bioactive anion transporters†

A novel class of transmembrane anion carriers, the click-tambjamines, display remarkable anionophoric activities in model liposomes and living cells. The versatility of this building block for the generation of molecular diversity offers promise to develop future drugs based on this design.

Top: from left to right, solid state X-ray structures of compounds 1, 2 and 5 as their hydrochloric salts. Solvent molecules have been omitted for the sake of simplicity. In the case of 5, only one of the two molecules the asymmetric unit consists of is shown. Bottom: structures of the studied compounds.
flat, as the maximum mean deviation from planarity of the plane defined by the atoms of the triazole and pyrrole rings and the imine moiety is 0.034 Å.
The interaction of the protonated receptors with chloride, as well as with nitrate and bicarbonate, was also studied in solution by 1 H NMR spectroscopy. Titration experiments of the hydroperchloric salts of compounds 1-9 with the corresponding tetrabutylammonium salts in DMSO-d6 were carried out. As an example, Fig. 2 displays the 1 H NMR stack plot obtained in the titration of 2·HClO4 with TBACl. Upon addition of chloride, the signals corresponding to the protons of the imine and pyrrole N-H fragments as well as the triazole C-H group undergo a progressive deshielding, in agreement with the interaction found in the solid state structures of 1·HCl and 5·HCl. In order to quantify the binding, the association constants Ka were determined using the Bindfit software, 11 by fitting the titration profiles to a 1:1 (LH:A) model, L being the receptor and A the anion (Table 1). The calculated values indicate that the adducts formed with chloride are more stable than those formed with nitrate, in line with the lower coordinating character of the latter. Bicarbonate induces deprotonation of the compounds and no Ka value could be ascertained in these conditions (see ESI).
Transmembrane anion transport was explored by means of two different techniques: potentiometry (ion-selective electrode, ISE) and emission spectroscopy. In all cases 1palmitoyl-2-oleoyl-sn-glycero-3-phosphocoline (POPC) liposomes were employed. For the experiments involving the use of the chloride-selective electrode liposomes are filled with a sodium chloride buffered aqueous solution and they are suspended in an isotonic, chloride-free solution. The chloride efflux promoted by the studied compound is monitored over time and the resulting concentrations normalised by referring them to the total concentration of chloride in the liposomes, which is obtained after lysing them with a surfactant. The chloride efflux at 300 seconds for each concentration of compound employed is plotted against such concentrations and these data fitted with Hill equation. This fitting provides the EC50 parameter, representing the concentration of compound needed to induce a 50% release of chloride. 12 This parameter is useful for comparing the relative potency of the anionophores; the lower the value of the EC50, the higher the potency of the transporter ( Table 1). The transport activity was found to be dependent of the substitution of the imine. The cyclooctylcontaining compounds 3, 6 and 9 are the most active ones, closely followed by the cyclohexyl-derivatives 2, 5 and 8. The cyclopropyl-derivatives 1, 4 and 7 were found significantly less active (about one order of magnitude). The substituent of the aryl group was found to have little influence in the transport activity. Calculated EC50 values of the most active derivatives were found in the low nanomolar range, highlighting the extraordinary anionophoric activity of the most active click- Table 1 Association constants Ka (M -1 ) for compounds 1-9 in their protonated forms with chloride and nitrate (added as tetrabutylammonium salts), determined from 1 H NMR titration experiments in DMSO-d6 at 293 K, and transport activities expressed as EC50 (nM).

Comp.
Ka tambjamines. The observed trend in the transport activity result could be explained by the relative lipophilicity of the compounds, since 1, 4 and 7 are the less lipophilic derivatives of the nine studied (see calculated logP values, Table S2). 13 The variation of the EC50 parameter when studying the chloride efflux in the presence of external nitrate, bicarbonate or sulfate is consistent with anion exchange as the main mechanism accounting for the transmembrane transport activity elicited by these compounds. Bicarbonate is less lipophilic than nitrate and therefore it is more difficult to extract into the lipid membrane; consequently, the EC50 values were found to be higher in the assays involving the chloride/bicarbonate exchange.
Emission spectroscopy experiments performed with carboxyfluorescein-loaded vesicles (see ESI for details) confirm that the anionophores do not form large non-selective pores in the lipid membrane, i.e., they do not behave as detergents. 14 Additional assays with the pH-sensitive fluorescent dye HPTS (see ESI for details) prove that these compounds efficiently dissipate pH gradients (Fig. 3), at concentrations as low as 0.0001% mol carrier to lipid concentration (see ESI). Compounds 1, 4 and 7 are the less active of the series, whereas the cyclohexyl-and cyclooctyl-derivatives are the ones that provoke a greater alteration of the pH inside the vesicles. This trend is similar to that found for the ISE assays.
We next decided to explore the ability of these compounds to facilitate anion transport in living cells using Fisher Rat Thyroid (FRT) cells expressing a variant of the iodide-sensitive Yellow Fluorescent Protein (YFP). 15 Thus, FRT cells transfected with this YFP are incubated with the compounds (or DMSO, as control) at 37 °C for 30 minutes, and the emission baseline recorded for 2 seconds. Then, a sodium iodide aqueous solution (at pH 6.6 or 7.3, depending on the experiment; see ESI) is injected and the emission measured for 20 seconds. The cellular internalisation of the iodide is signalled by the decay of the YFP's emission. Non-active anion transporters and control experiments with DMSO resulted in no internalisation of iodide and no changes in the observed fluorescence. The quenching rate (QR) at different concentrations of the compounds for both pH values was represented against such concentrations and the resulting curves were fitted to a first-order binding model. 16 Fig. 3 Variation of pH upon addition of the studied compounds to 7:3 POPC:cholesterol vesicles (0.5 mM POPC). Vesicles (loaded with 126.2 mM NaNO3 buffered at pH 7.2 with 10 mM phosphate, and containing 10 µM HPTS; I.S. 150 mM) were suspended in a NaNO3 aqueous solution (126.2 mM NaNO3 buffered at pH 7.2 with 10 mM phosphate; I.S. 150 mM). At t = 30 s an aliquot of a NaOH solution (11 µL, 0.5 M) was added, and at t = 60 s the anion carrier was added (0.0005% mol carrier to lipid concentration). The blank is DMSO (10 µL). Each trace represents the average of at least three trials, performed with three different batches of vesicles.
This fitting provides the maximum quenching rate (mQR), which is a direct indication of the activity of the compound as anion transporter; the higher its value, the more active the carrier is (see ESI for a detailed procedure and the plots). Calculated mQR values are displayed in Table 2. The activity trend is in agreement with the previous results observed in liposomes. Thus, very limited activity was observed for the cyclopropylderivatives 1, 4 and 7, whereas cyclohexyl-and cyclooctylcompounds 2, 3, 5, 6, 8 and 9 display significant activity. The observed transport activity increases when the assays were performed at a slightly acidic external pH. Both observations are evident in Fig. 4, where the QR values for compounds 4, 5 and 6 are plotted against different concentrations of the compounds. This result is also in agreement with the demonstrated ability of these compounds to discharge pH gradients in liposomes and bodes well for potential applications in the case of cystic fibrosis, where an acidic airway surface liquid is a hallmark of the pathophysiology of this condition at the pulmonary level. For comparison purposes, it should be noted that the measured mQR of the activated wild type CFTR protein under these conditions is lower than 0.1 s -1 . 17 Table 2 Transport activities of compounds 1-9 in FRT cells at pH 6.6 and 7.3 (outer medium), expressed as mQR (s -1 ), and cell viability IC50 (µM) of such compounds in human lung (A549) and breast (MCF7) adenocarcinoma cell lines and in human mammary epithelial (MCF10A) cell lines.  Finally, the cytotoxicity of these compounds was analysed. The concentration that provokes 50% growth inhibition (IC50 values) was determined on human lung (A549) and breast (MCF7) adenocarcinoma cell lines, as well as human mammary epithelial (MCF10A) cell lines (Table 2). Compounds 2, 3, 5, 6, 8 and 9 display moderate cytotoxicities, with IC50 values ranging from 3.5 to 10.7 µM, whereas compounds 1, 4 and 7 are considerably less cytotoxic (from 23.5 to 51.4 µM). Little discrimination between normal and cancerous cell lines was observed. These results strongly suggest that the observed toxicity is related to the transmembrane transport ability of these compounds. On the other hand, very significant transport activity was observed in FRT cells for the active compounds at concentrations well below their IC50 values.
In summary, nine click-tambjamine compounds have been synthesised and characterised and their transmembrane anion transport activities in both model liposomes (POPC) and living cells investigated. Based on this design we have identified cyclohexyl-and cyclooctyl-derivatives which are remarkably active anion carriers in both vesicles and living cells. The suitability of the molecular design to produce and explore numerous compounds in an easy manner might constitute a good starting point for the design of a drug oriented to the treatment of anion-transport related diseases, such as cystic fibrosis.
Financial support from the European Union's Horizon 2020 research and innovation programme (TAT-CF project, grant agreement 667079), Instituto de Salud Carlos III (Grant PI18/00441) (co-funded by the European Regional Development Fund (ERDF), a way to build Europe) and "La Caixa" Foundation and Caja Burgos Foundation (CAIXA-UBU004) is gratefully acknowledged. V. S.-C. and R. P.-T. also thank CERCA Programme/Generalitat de Catalunya for institutional support. The authors gratefully acknowledge Andrea Sancho-Medina and Víctor Arnáiz-Lozano for their contributions to transmembrane anion transport experiments and Paula Nadeu for her technical assistance.

Conflicts of interest
There are no conflicts to declare.