Pnictogen-Bonding Catalysis and Transport Combined: Polyether Transporters Made In Situ

Abstract

The combination of catalysis and transport across lipid bilayer membranes promises directional access to a solvent-free and structured nanospace that could accelerate, modulate, and, at best, enable new chemical reactions. To elaborate on these expectations, anion transport and catalysis with pnictogen and tetrel bonds are combined with polyether cascade cyclizations into bioinspired cation transporters. Characterized separately, synergistic anion and cation transporters of very high activity are identified. Combined for catalysis in membranes, cascade cyclizations are found to occur with a formal rate enhancement beyond one million compared to bulk solution and product formation is detected in situ as an increase in transport activity. With this operational system in place, intriguing perspectives open up to exploit all aspects of this unique nanospace for important chemical transformations.

Reactions that occur in lipid bilayer membranes continue to fascinate us, from early oscillating oxidations,¹ self-replicating vesicles,² templated polymerizations,³⁻⁵ photoredox processes,⁶⁻⁹ and catalytic pores¹⁰ to more recent dynamic covalent chemistry,^11,12 molecular rockets,¹³ multistep cascades,^14,15 and signal transduction systems.¹⁶⁻¹⁸ The direct combination of reactions with transmembrane transport has received less attention. The most developed systems operate with existing biological protein architectures,¹⁹ thus avoiding synthetic efforts and direct participation of the membrane in the process. However, the coupling of catalysis and transport appears promising for several reasons in addition to the obvious unique detectability. Most importantly, lipid bilayer membranes offer a solvent-free environment that should strengthen noncovalent interactions. This removal of the often undesired contributions of solvents to catalysis is reminiscent of transformations in the gas phase or in silico. The volume provided for these solvent-free reactions is small and structured: A confined nanospace that appears ideal to maximize effective concentrations and control selectivity. Moreover, the combination of catalysis with transport adds directionality along a polarity gradient, promising remote control over diffusion²⁰⁻²³ along changing environments to maximize transition-state stabilization, minimize product inhibition, vary selectivity, and so on. Compartmentalization further invites multistep processes^6−10,14,15 and various forms of detection.^10,19 Taken together, these intriguing characteristics promise access to rate enhancements, new selectivity, and, at best, new reactivity. To thus explore the possible combination of catalysis and ion transport within lipid bilayer membranes, we decided to couple two topics of current concern, that is, transport and catalysis with chalcogen, pnictogen, and tetrel bonds and the biomimetic epoxide-opening cascade cyclizations into polyether cation transporters (graphical abstract).

Anion transport²⁴⁻³⁷ with pnictogen bonds³⁸⁻⁴⁴ has been realized recently³¹⁻³³ as logical continuation of the earlier studies with chalcogen^34,35 and halogen^36,37 bonds. These unorthodox interactions all originate from σ holes and, related to σ* antibonding orbitals, extend linearly from all covalent bonds made by the element (Figure 1a, see refs (44) and (38) for electrostatic potential surfaces).⁴⁵⁻⁵⁶ Their strength increases with withdrawing substituents and polarizability of the element, that increases top–down and right–left in the periodic table.³¹

Figure 1.

(a) Structure and anion transport activity of catalyst candidates 1–12, with EC₅₀ values in absolute concentrations and mol% relative to EYPC (HPTS assay, EYPC LUVs in buffer containing NaCl). (b) CX complex 13 and F–Sb distances in the X-ray structure of 9. (c) HX prodigiosin complex. Blue ellipses indicate σ holes.

Fluorophenyl derivatives 1–12 were considered to act as both anion transporters and catalysts. They were prepared following reported procedures (Scheme S1).^31,44 Ion transport was measured in EYPC LUVs with the classical HPTS assay.^57,58 In this assay, EYPC LUVs (egg yolk phosphatidylcholine large unilamellar vesicles) are prepared with entrapped HPTS, a pH-sensitive fluorophore. Then the transporter is added, followed by a base pulse (or vice versa). The dissipation of the resulting pH gradient is then followed by ratiometric changes in HPTS fluorescence, reporting on OH^–/anion and H⁺/cation antiport and on OH^–/M⁺ and H⁺/X⁻ symport. At the end of the experiment, an excess of channel forming peptide gramicidin D is added to determine maximal fluorescence intensity for calibration. Results are summarized in dose–response curves (DRCs) and reported as EC₅₀, the effective transporter concentration needed to reach 50% activity.

According to the HPTS assay, the most active anion transporters were σ-hole donors 1–4 equipped with 3,4,5-trifluorophenyl (FP₃₄₅) substituents (Figure 1a). Anion transport with tetrel bonds as in stannane 1 is novel in this series, and activities were with EC₅₀ = 2.3 ± 0.4 nM, very high (Figure S6). Tris(3,4,5-trifluorophenyl)stibine (Sb(FP₃₄₅)₃) 2 was similarly active, while bismuthane 3 was less powerful because, presumably, of the more dominant metallic character. The still high activity of tellane 4 followed polarizability trends. An only micromolar EC₅₀ of germane 5 was consistent with the previously described⁴⁴ inaccessibility of the σ holes on the too small element.

The FP_2–6 series 6–9³¹ was overall less active compared to 1–5 because intramolecular pnictogen bonding of the ortho fluorines weakens anion binding (Figure 1a,b).⁴⁴ The ortho hydrogens in FP₃₄₅1–5, in contrast, are repelled by the σ holes and can further assist anion binding in the catalyst–anion (CX) complex 13. While FP_2–69 and 10 were unstable, FP₂₄₆ analogues 11 and 12 confirmed that ortho fluorines indeed weaken anion transport significantly. FP₂₄₆11 and 12 and all other compounds were stable in buffer (Figures S2–S5). This included the best pnictogen-bonding catalyst 2 and marked a clear contrast to ligand-exchanging, water-incompatible general Lewis acid catalysts.

To facilitate comparison, some EC₅₀’s are also reported in mol% lipid. Today’s record among small molecule transporters is arguably EC₅₀ = 6.1 × 10^–5 mol% for the tridentate HX binding natural product prodigiosin 14, closely followed by a multidentate oligo-urea macrocycle.³⁰ A 1000 times weaker EC₅₀ = 0.01 mol% has been recently reported for bidentate halogen-bonding transporters as current best in the context of σ-hole transporters.³⁵ The EC₅₀ = 1.8 × 10^–3 and 2.1 × 10^–3 mol% of the monodentate and neutral 1 and 2, only about 30 times weaker than cation 14, thus confirmed the power of anion transport with the hydrophobic, strong and directional tetrel and pnictogen bonds.

Polyether cascade cyclizations were selected as reactions of choice for this study. Polyethers are among the most popular cation transporters.⁵⁹⁻⁶⁵ Examples reach from tetrahydrofuran (THF) oligomers such as the natural product monensin A 15 and early artificial ion channels⁵⁹ to most popular crown ethers⁶⁰⁻⁶⁴ and acyclic motifs (Figure 2).^64,65 In nature, polyether transporters are synthesized by epoxide-opening cascade cyclizations.⁶⁶⁻⁷⁴ These charismatic processes have been studied extensively,⁶⁶⁻⁷¹ recently also with anion-π catalysis.⁷²⁻⁷⁴ To further expand the integration of unorthodox interactions, pnictogen-bonding catalysis has been introduced as noncovalent counterpart of Lewis acid catalysis.⁴⁴ Polyether cascade cyclizations have served well to demonstrate that the two are not the same, just like noncovalent hydrogen bonding and covalent Brønsted acid catalysis have their distinct advantages.⁴⁴

Figure 2.

Structure and cation transport activity of substrate and product candidates 16–19 with EC₅₀ values (EYPC LUVs, NaCl), compared to monensin A 15 and the presence of FCCP (in parentheses).

Oligo-epoxides 16 and 17 cyclize selectively into 18 and 19, respectively (Figure 2).⁷² The four compounds and their analogues were synthesized following reported procedures (Figure S14).⁷² The transport activity of THF oligomers increased with length as expected, from less active monomers and dimers (EC₅₀ = 1.1 ± 0.4 mM) up to tetramer 19 (Figure 2). Compared to 19, monensin A 15 was more active without but less active with the proton carrier FCCP, indicating that 19 is a better sodium transporter and a weaker proton transporter.^57,58 A previously unexplored family, the oligo-epoxides 16 and 17 were identified as weak ion transporters. Multiply methylated analogues were more active (EC₅₀ = 8.9 ± 0.7 μM for tetramers, Figure S19) but less attractive for catalysis studies because of violations of the Baldwin rules⁴⁴ and relatively high background reaction in water.

Because stannane 1 is a weaker catalyst,⁴⁴ anion transporter 2 and cation transporters 17 and 19 were selected to combine transport and catalysis (Figure 3). However, the outstanding transport activity of 2 prevented access to the high concentrations of substrate and catalyst needed to observe sufficiently fast initial rates for direct detection (Figure 3b). The replacement of the chloride anions in the buffer by sulfates, which are more difficult to dehydrate and transport, solved this problem: The EC₅₀ of 2 increased almost 1000 times to EC₅₀ = 14 ± 3 μM (Figures 3c and 1).

Figure 3.

(a) Coupled transport and catalysis in EYPC LUVs with M⁺-transporting substrate 17 and product 19, X^–-transporting catalyst 2, and notional structures of CS 22 and CP 23. (b–j) Selected data for transport in the HPTS assay in LUVs, 1 mM HPTS, 10 mM HEPES, pH 7.0, 25 °C; (b–e) 125 μM, (f–j) 500 μM EYPC; (b) 100 mM NaCl, (c–j) 67 mM Na₂SO₄. (b) DRCs for 17 (squares), 2 (crosses), and 19 (circles) in buffer with NaCl. (c) DRCs for 2 (crosses) and 19 (circles) in buffer with Na₂SO₄. (d) Change in emission I of HPTS (λ_em = 510 nm; λ_ex1 = 404 nm, λ_ex2 = 454 nm) during the addition of 2 (20 μM, green), 19 (20 μM, red), and both (ruby, t = 0 min), NaOH (5 mM, t = 1 min), and gramicidin D (10 nM, t = 5.8 min) to EYPC LUVs. (e,f) DRCs for 17 (cyan) and 19 (ruby) with (e) 20 μM 2, 125 μM EYPC and (f) 100 μM 2, 500 μM EYPC. (g) Change in I upon addition of 2 (100 μM, green), together with 17 (100 μM, cyan) and 19 (100 μM, ruby, t = 0 min), NaOH (t = 0.5 min), and gramicidin D (t = 2 min; 500 μM EYPC, 67 mM Na₂SO₄). (h) Change in I during the addition of NaOH (t = 0.5 min) and gramicidin D (t = 2 min) measured t = 0 (cyan), 15, 30, 45, and 60 min (ruby) after the addition of 17 (100 μM) and 2 (100 μM). (i) Y as a function of the reaction time t for 2 (100 μM, green), 17 (100 μM, cyan; activities were recorded with 100 μM 2, added right before the measurements), and both together (ruby, mean ± standard deviation of three experiments). (j) Initial velocity v (best fit ± error) of changes in Y as a function of the concentration of 17 at constant 2 (100 μM), measured as in (g)–(i) for varied substrate concentrations.

Original HPTS kinetics under these conditions for 2 at the most sensitive 50% activity (green), 19 at minimal detectable activity (red), and both transporting together (ruby) revealed that their activities were overadditive (Figure 3d). Similar overadditivity was found for 2 and 17 (Figure S23). Cooperativity rather than anticooperativity demonstrated that the transporter–ion complexes 20, 21, and 13 are more stable than catalyst–substrate (CS) complex 22 and CP 23 (Figure 3a). Such less stable CS and CP are beneficial to avoid anticatalysis and product inhibition.

In the presence of 20 μM 2, the differences in transport activity between 17 and 19 were well preserved (Figure 3e, Hill coefficients n > 1 especially for 19 suggested that PM 21 might be a 2:1 complex, Tables S3 and S4). To stabilize the system and access higher concentrations for faster conversion, the lipid concentration was increased. With four times more vesicles and 100 μM 2, the transport activities of 17 (EC₅₀ = 200 ± 75 μM) and 19 (EC₅₀ = 48 ± 7 μM) decreased correspondingly, while the range for in situ detection of their conversion increased from 5–30 μM to 40–>200 μM substrate (Figure 3e,f, red areas).

Under these optimized conditions, 2 was nearly inactive (Figure 3g). The difference between 17 and 19 in the presence of 2 remained significant, promising detectability of the cyclization (Figure 3g, red area). The substrate, catalyst, and vesicles were thus mixed together under these conditions and stirred for given times before transport activity measurements. Activities increased with reaction time (Figure 3h,i). With 100 μM substrate and catalyst at 20 °C, full conversion was reached in 1 h (Figures 3g,h, ruby). Controls with either catalyst or substrate alone did not generate similar activity (Figure 3i). Repetition at different substrate and constant catalyst concentrations gave the expected changes in initial velocity (Figure 3j).

Cyclization of 2.0 M monoepoxide with 100 mol% stibine 2 in CD₂Cl₂ takes 48 h at 40 °C.⁴⁴ When reacted at 1.0 M concentrations for 4 days, tetra-epoxide 17 was fully converted into 19 (Figure S25). Comparison of 1.0 M substrate converted in 96 h at 40 °C with 100 μM substrate converted in 1 h at 20 °C, also considering the van’t Hoff equation, implied a rate enhancement beyond 6 orders of magnitude. Based on the lipid concentrations used, the high local concentration in membranes can only account for about 10³-fold increases in rate. The found rate enhancement was thus significant, possibly reflecting a shift from more stepwise to more concerted cascade cyclizations.

In summary, we introduce an operational system that combines catalysis and transport in lipid bilayer membranes, with synergistic anion and cation transporters realizing pnictogen-bonding catalysis under conditions that do not work in bulk solution. With the methods in place, these results provide a solid basis to exploit the many aspects of lipid bilayer membranes as a unique, directionally accessible, solvent-free, and structured nanospace for translocation-coupled molecular transformation.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.1c00345.

• Detailed experimental procedures, materials and methods, compound synthesis and characterization, original fluorescence kinetics traces, dose–response curves, Hill analyses, data summarizing tables (PDF)

Author Present Address

^‡ Chengdu University of Technology, Chengdu, People’s Republic of China

Author Contributions

^† H.V.H. and A.G. contributed equally.

The authors declare no competing financial interest.