Highly Active Halogen Bonding and Chalcogen Bonding Chloride Transporters with Non‐Protonophoric Activity

Abstract Synthetic anion transporters show much promise as potential anti‐cancer agents and therapeutics for diseases associated with mis‐regulation of protein anion channels. In such applications high activity and anion selectivity are crucial to overcome competing proton or hydroxide transport which dissipates cellular pH gradients. Here, highly active bidentate halogen bonding and chalcogen bonding anion carriers based on electron deficient iodo‐ and telluromethyl−triazole derivatives are reported. Anion transport experiments in lipid bilayer vesicles reveal record nanomolar chloride transport activity for the bidentate halogen bonding anion carrier, and remarkably high chloride over proton/hydroxide selectivity for the chalcogen bonding anionophore. Computational studies provide further insight into the role of sigma‐hole mediated anion recognition and desolvation at the membrane interface. Comparison with hydrogen bonding analogues demonstrates the importance of employing sigma‐hole donor motifs in synthetic anionophores for achieving both high transport activity and selectivity.


Materials and methods
All reagents and solvents were purchased from commercial sources and used without further purification. Lipids were purchased from Avanti polar lipids and used without further purification. Where necessary, solvents were dried by passing through an MBraun MPSP-800 column and degassed with nitrogen. Column chromatography was carried out on Merck® silica gel 60 under a positive pressure of nitrogen. Where mixtures of solvents were used, ratios are reported by volume. NMR spectra were recorded on a Bruker AVIII 400, Bruker AVII 500 (with He cryoprobe) and Bruker AVIIIHD 500 spectrometers. Chemical shifts are reported as δ values in ppm. Mass spectra were carried out on an Agilent 6120 bench-top single quadrupole, a Waters LCT Premier XE benchtop (oa-TOF) and a Thermo Exactive High-Resolution Orbitrap FTMS spectrometer. Fluorescence spectroscopic data were recorded using a Horiba Duetta fluorescence spectrophotometer, equipped with Peltier temperature controller and stirrer. Experiments were conducted at 25°C unless otherwise stated. Vesicles were prepared as described below using Avestin "LiposoFast" extruder apparatus, equipped with polycarbonate membranes with 200 nm pores. GPC purification of vesicles was carried out using GE Healthcare PD-10 desalting columns prepacked with Sephadex G-25 medium.

Synthesis and characterization
Warning! Low molecular weight organic azides used in this study are potentially explosive and should be used on a small scale. Appropriate protective measures should always be taken when handling these compounds.

General Procedure 2
The appropriate alkyne (0.30 mmol) was dissolved in the minimum amount of MeOH (ca. 10 mL per 0.30 mmol of alkyne) to which was added AgNO3 (1 equivalent per alkyne) followed by one drop of concentrated NH4OH (0.1 mL) immediately inducing precipitation of the desired silver acetylide. The reaction mixture was left to stir for 30 minutes at room temperature, after which time the solvent was removed in vacuo. The silver acetylide was subsequently suspended in anhydrous THF (20 mL) under an atmosphere of nitrogen. In a separate flask Me2Te2 (0.50 equivalents per alkyne) was dissolved in THF (5 ml) and cooled to 0°C, to the solution of the ditelluride 1 M of Br2 in CH2Cl2 (0.50 equivalents per alkyne) was added dropwise, after which time the ditelluride solution darkened indicating the formation of MeTeBr which was added immediately to the suspension of the silver acetylide dropwise. The reaction mixture was left to stir for 30 minutes at room temperature in the absence of light, after which time it was filtered through Celite and the reaction mixture was concentrated to dryness in vacuo. The crude telluromethyl alkyne was then used immediately for the following CuAAC reaction. Iododecyne 6. Under N2, decyne (3.623 mmol) and KOH (9.058 mmol) were stirred in MeOH for 10 mins, followed by the addition of I2 (3.986 mmol). This was stirred at rt for 48 hours. The reaction was diluted with water and extracted with EtOAc (20 mL). The organic mixture was washed with saturated Na2S2O3 solution (2 x 10 mL), followed by brine (20 mL) and then dried over MgSO4. The solvent was removed in vacuo to give the pure product (quant.). Characterisation data aligns with the literature. [     Carrier 1•XB. 1,3-diethynyl-5-nitrobenzene (100 mg, 0.584 mmol) was dissolved in anhydrous THF to which was added CuI (11 mg, 0.0583 mmol) and N-iodomorpholine hydroiodide (656 mg, 1.93 mmol) and left to stir for 4 hours in the absence of light. After which time the reaction mixture was diluted with DCM (300 mL) and filtered through a DCM saturated pad of alumina, the filtrate was concentrated in vacuo and adsorbed onto silica. The dry loaded compound was eluted with 10% EtOAC/hexane mixtures (500 mL) and the filtrate concentrated to dryness in vacuo. The iodoalkyne 9 was isolated as a yellow solid in 83% yield and then immediately used in a CuAAC reaction with perfluorbenzene azide according to general procedure 1, to yield 1•XB (428 mg, 0.509 mmol, 87%) as an off white solid. 1 H NMR (500 MHz, CDCl3) δ 9.20 (t, J = 1.6 Hz, 1Hb), 9. Figure S23. 13 C NMR Spectrum of 2•ChB (CDCl3, 298 K, 126 MHz).

Single crystal X-ray diffraction experiments
Crystal Structure Determination Deposition Number(s) 2075766 (for 1•XB) and 2075767 (for 1•ChB) contains the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service www.ccdc.cam.ac.uk/structures." Single-crystal X-ray diffraction intensities for 1•XB and 1•ChB were collected at 150 K on Oxford Diffraction/Agilent SuperNovae diffractometers with Cu-Kα (λ = 1.54184 Å) radiation equipped with a nitrogen gas Oxford Cryosystems Cryostream unit. [6] A suitable crystal was chosen and mounted on a 200 μm MiTeGen loop using perfluoropolyether oil. The CrysAlisPro [7] software was used for data collection and integration. All structures were solved using SuperFlip [8] and refined using full-matrix least-squares refinement within the CRYSTALS [9] suite. All non-hydrogen atoms were refined anisotropically. The hydrogen atoms were positioned at geometrically sensible positions and refined using a riding model.
Transport assays with HPTS [10], [11] In a typical experiment, the LUVs containing HPTS (25 µL, final lipid concentration 31.3 µM) were added to buffer (1950 µL of 100 mM NaCl, 10 mM HEPES, pH 7.0) at 25°C under gentle stirring. A pulse of NaOH (20 µL, 0.5 M) was added at 40 secs to initiate the experiment. At 100 s the test transporter (various concentrations, in 5 µL DMSO) was added, followed by detergent (25 µL of Triton X-100 in 7:1 (v/v) H2O-DMSO) at 300 secs to calibrate the assay. The fluorescence emission was monitored at λem = 510 nm (λex = 405/460 nm). The fractional fluorescence intensity (Irel) was calculated from equation (S1), where Rt is the fluorescence ratio at time t, R0 is the fluorescence ratio at time 0, and Rd is the fluorescence ratio after the addition of detergent.
The fractional fluorescence intensity (Irel) at 288 s just prior to lysis, defined as the fractional activity y, was plotted as a function of the ionophore concentration (x / µM). Hill coefficients (n) and EC50 values were calculated by fitting to the Hill equation (S2): where y0 is the fractional activity in the absence of transporter, ymax is the fractional activity in with excess transporter, x is the transporter concentration in the cuvette. Hill plots were fitted to at-least 8, and up to 13 data points spanning the required concentration range, and each individual concentration was repeated at-least twice and averaged.
Experiments with DPPC lipids were conducted in the same way. For elevated temperature studies, the buffer was equilibrated at 45°C (using the Peltier temperature controller) for 5 minutes prior to initiating the experiment. Cation selectivity experiments were conducted by external ion exchange, by adding the POPC vesicle solution (prepared as above) to buffer (100 mM M + Cl -, 10 mM HEPES, pH 7.0), where M + = Li + , Na + , K + , Rb + .
Experiments with NMDG-Cl (100 mM) in place of NaCl in both the external and internal buffer was also carried out according to the above procedures, except that the POPC lipid concentration was increased to 0.1 mM and NMDG (20 µL of 0.5 M giving a final concentration 5 mM in the cuvette) was used for the base pulse in place of NaOH. This assay was carried out with and without the addition of the proton channel Gramicidin D (0.1 mol%, added in 4 μL DMSO), which when added was done so 30 seconds after the base pulse.

HPTS assay data for all transporters
In

NMR titration experiments
All binding constants were measured by 1 H NMR titrations in a Bruker AVIII 500 spectrometer at 500 MHz and 298 K. The host (anionophores 1 and 2) was dissolved in acetone-d6, or 2.5% D2O-acetone-d6 (v/v) mixtures, at 1 mM concentration and a known volume (0.5 mL) added to the NMR tube. Known volumes of anion guest (added as the TBA salt) in acetone-d6, or 2.5% D2O-acetone-d6 (v/v) mixtures, were added and the spectra were recorded after each addition. The chemical shift perturbations of the host spectra were monitored as a function of guest concentration. In the case of 1•XB, 1•ChB and 1•HB the internal aryl signal a was monitored, in the case of 2•XB, 2•ChB and 2•HB; the methylene signal b, the telluromethyl signal c and the C-H triazole signal d, respectively were monitored ( Figure S49). The data was analysed using a global fit procedure using the Bindfit [14], [15] program, using non-linear least squares analysis to obtain the best fit between observed and calculated chemical shifts for the 1:1 binding stoichiometry. In all experiments the association of guest and host was fast on the NMR timescale and the chloride association constants are summarised in Error! Reference source not found.3. Figure S49. The structures of the bidentate and monodentate carriers with perturbations of the labelled proton signal used for fitting to determine chloride anion association constants.

Dehydration energies of anions
Implicit solvation models are known to underestimate the solvation free energy change associated with the transfer of anions between ideal gas and aqueous phase, primarily due to the lack of short-range solvent-solute interactions (e.g. hydrogen bonds) in continuum models. [30] Therefore, it came as no surprise that using the SMD solvent model the computed desolvation energy (∆Gdesolv) of OHwas underestimated by ~13 kcal mol -1 (91.9 kcal mol -1 versus the experimental value of 105.0 kcal mol -1 ). [31] While we anticipate the error in the dehydration (i.e. H2O to CHCl3) energy considered in this study may be lower than the error in desolvation (i.e. H2O to vacuum), a systematic underestimation of this value is likely to remain and be on the order of 10 kcal mol -1 . However, this error will not change the qualitative trend in ∆Gtotal observed in this study.