English

• Evolutionarily selected protein channels are often characterized by extraordinarily high selectivity in conducting molecular transport across cellular membrane. One prominent example is a family of water channel proteins called aquaporins, transporting water molecules at rate of 10⁹ molecules per second while rejecting both ions and protons [1–3]. Other remarkable examples include potassium channel KcsA [4], exhibiting stunningly high K⁺/Na⁺ selectivity of 10⁴, and proton channel M2 [5], which is > 10⁵-fold more permeable to protons than to other monovalent cations.

  Serving as a rich source for biomimetic chemistry, these natural channel systems further set a high standard in membrane transport performance extremely challenging for manmade channel systems to reach. In fact, despite tremendous developments over the past four decades in artificial channel [6–15], to our best knowledge, there is only one artificial channel system that truly imitates its natural counterparts, i.e., aquaporins, in terms of high water transport selectivity (high rejection of both salts and protons) while surpassing them by 150% in water transport rate [16]. Recently, significant advances also have been made in artificial proton channel [17,18], culminating in high selectivity factors of 167.6, 122.7 and 81.5 over Cl⁻, Na⁺ and K⁺ ions, respectively, and a proton transport rate 1.22 times that of gramicidin A [18]. Other than these rare cases, efforts to replicate high selectivity seen in nature have not been very successful in developing artificial channels toward selective transport of K⁺ [19–25], Na⁺ [26–28], Cl⁻ [29–36], ClO₄⁻ [37] and I⁻ [38–40]. Particularly, while we are not aware of any Na⁺-selective channel that can achieve Na⁺/K⁺ selectivity of > 10 obtained on the basis of single channel current traces, impressive progresses in recent years have led to artificial K⁺-channels that not only transport K⁺ ions at ultrafast rates (Table S1 in Supporting information) but also exhibit K⁺/Na⁺ selectivity factors of 9.8 in 2017 [20], 14.0 [22] and 16.3 [23] in 2020, and 18.2 [24] and up to 18.9 [25] in 2021. Nevertheless, compared to KcsA's extraordinarily high K⁺/Na⁺ selectivity, these very modest numbers of < 20 certainly point to a great room of improvement for artificial channel systems.

  In this article, we show that pairing crown ethers with mutually compatible tert-butyl groups via a social self-sorting process efficiently generates heteromeric channels that line up crown ethers and tert-butyl groups alternatively on the same side, rather than homomeric ones consisting of either crown ethers or tert-butyl groups. This alternative arrangement greatly alleviates the steric repulsions that otherwise exist among the vertically aligned crown ethers or tert-butyl groups in their homomeric channels. Significantly, the resultant heteromeric channels consistently exhibit better ion transport activities and higher K⁺/Na⁺ selectivities than the corresponding homomeric channels that contain only vertically arrayed crown ethers. We further found that a small change in lipophilic aliphatic side chain of the channel molecule drastically boosts its K⁺/Na⁺ selectivity. As such, we demonstrate that high K⁺/Na⁺ selectivity of 16.0 can be attained through side chain tuning, and this high selectivity can be further augmented quite substantially via social self-sorting to 20.1, a value that is the 5F8 (Fig. 1a) was recently demonstrated to function as an excellent K⁺-selective channel, with K⁺/Na⁺ selectivity of 9.8 determined using single channel conduction measurement [20] and an EC₅₀ (K⁺) value of 6.2 µmol/L determined at the liposome level. Presumably, molecules of 5F8 with built-in H-bond functionalities self-assemble via intermolecular H-bonds into one-dimensional structures (5F8)_n, aligning appended 15-crown-5 units into an ordered array to create a transmembrane pathway for transporting K⁺ ions (Fig. 1a). Formation of such H-bonded 1D structures can be supported by the crystal structures of related molecules (e.g., Fmoc-Phe-C4 [41]) and many other structurally similar molecules that are able to form fibers and gel oils with high ability [42–45].

  Figure 1

  

  Figure 1.  (a) Self-assembled homo-oligomeric 1D structures (5F8)_n might undergo a structural transformation to form heteromeric structures (5F8·BF8)_n via a social self-sorting process. (b) Computationally determined molecular heights of tert-butyl group in Boc, benzo-15-crown-5 in C5, and 18-crown-6 in C6. (c) Computationally obtained near-perfect compatibility in geometry between benzo-15-crown-5 in 5F8 and tert-butyl group in BF8 in the hetero-ensemble (5F8·BF8)_n, having perfectly superimposable crown ether units. (d) Molecular design for combinatorial discovery of channel molecules that can pair with BF8 to produce heteromeric ensembles with enhanced ion transport activity and selectivity.

  DownLoad: Full-Size Img PowerPoint

  Consistent with an inter-chain separation distance of 5.0 Å seen in (Fmoc-Phe-C4)_n [41], such distance in (5F8)_n was computationally determined to be 4.9 Å using the COMPASS force field [46]. This value, however, is significantly smaller than the molecular height of 6.0 Å computationally determined at the B3LYP/6–31G* level for 15-crown-5 group from C5 (Fig. 1b), indicating a certain degree of conformational derivation in the crown ethers found in (5F8)_n from that in C5. Computationally, this expected conformational deviation turns out to be quite significant (Fig. S1a vs. Fig. S1b in Supporting information), and every 15-crown-5 unit in (5F8)_n was calculated to destabilize the H-bonded ensemble (5F8)_n by 3.03 kcal/mol with chloroform as the explicit solvent.

  To minimize the extent of structural crowdedness involving crown ethers in homo-ensemble (5F8)_n, one likely effective way is to introduce a steric code [47–49] to promote social self-sorting [50–52] in a way that hetero-ensemble (5F8·BF8)_n (Fig. 1a) highest among all hitherto reported artificial potassium channels and transporters [19–25,53–56] with a larger inter-chain separation distance of 9.8 Å can be formed to measurable or desirably dominant extents with respect to (5F8)_n and (BF8)_n. We envisioned that a steric code comprising a tert-butyl group in Boc and a benzo-15-crown-5 unit in C5 (Fig. 1b) might help achieve desired social self-soring among molecules of Boc-containing BF8 and C5-containing 5F8, leading to preferential formation of hetero-ensemble (5F8·BF8)_n (Figs. 1a and c). This is possible since the tert-butyl group, having a molecular height of 6.2–6.8 Å (Fig. 1b and Fig. S1d in Supporting information), must destabilize homo-ensemble (BF8)_n, and thus should favor an alternative alignment with a benzene group of 3.4 Å in thickness (Figs. 1a-c). The latter arrangement generates hetero-ensemble (5F8·BF8)_n with average molecular heights of 4.8–5.1 Å involving the tert-butyl and benzene groups, which are compatible with the inter-chain distance of 4.9 Å. Further, upon forming (5F8·BF8)_n, the separation distance between the two adjacent crown ethers increases from 4.9 Å to 9.8 Å (Fig. 1c), thereby eliminating the molecular crowdedness among crown ether units and allowing them to adopt the energetically favored conformation like that in C5. Indeed, in sharp contrast to (5F8)_n, the conformation of crown ether units in (5F8·BF8)_n is now found to be near-identical to that in C5 (Fig. S1a vs. Fig. S1c in Supporting information). Similar analysis also holds true for 6F8, containing an 18-crown-6 unit of 7.0 Å in molecular height (Fig. 1b). Consequently, mixing 6F8 with BF8 also should form hetero-ensemble (6F8·BF8)_n to good extents. This concept can be additionally generalized to include non-octyl side chains of different lengths (e.g., (5Fm·BFm)_n or (6Fm·BFm)_n, where m = 10, 12, etc.). With this hypothetic social self-sorting-mediated formation of (5Fm·BFm)_n or (6Fm·BFm)_n (m = 8, 10, 12, etc.) using mutually compatible bulky tert-butyl and benzo-crown ether (e.g., benzo-15-crown-5 or benzo-18-crown-6) groups, we further hypothesized that hetero-ensembles such as (5Fm·BFm)_n with an enlarged separation distance among crown ethers might differ from home-ensembles such as (5Fm)_n in terms of ion transport activity and selectivity.

  In our early report [20], 5F8 and 5F10 were identified among 15 channel molecules to be the most selective and the most active K⁺ transporters, respectively. In our initial screening, we therefore used these two channel molecules to pair with BF8 and BF10 to examine the above hypothesis. MF8 and MF10, both having a methyl rather than a bulky Boc group at R₁ position, are used to compare with BF8 and BF10.

  Following the established protocols (Figs. 2a-c) [20,57], a pH-sensitive HPTS (8-hydroxypyrene-1,3,6-trisulfonic acid) assay was carried out to investigate the potassium transport activities of eight hetero-ensembles ((5Fm·BFm)_n, (6Fm·BFm)_n, (5Fm·MFm)_n and (6Fm·MFm)_n (m = 8 and 10)) and eight homo-ensembles ((5Fm)_n, (6Fm)_n, (BFm)_n and (MFm)_n (m = 8 and 10)). To evaluate their potassium transport activities, large unilamellar vesicles (LUVs), containing HPTS (100 µmol/L) and Na₂SO₄ (100 mmol/L) at pH 7.0, was diluted into a buffer, containing 100 mmol/L K₂SO₄ at pH 8.0, to generate a pH gradient and ion gradients (Na⁺ and K⁺) across LUVs. Upon addition of channel molecules, changes in fluorescence intensity of HPTS were monitored over duration of 5 min after which time triton was added to lysate the liposomes. For sodium transport study, the extravesicular K₂SO₄ was changed to Na₂SO₄.

  Figure 2

  

  Figure 2.  (a) LUV-based HPTS assay for evaluating ion transport properties of ion channels. (b) Typical K⁺ transport curves observed for hetero- and homo-ensembles such as those involving 5F10, BF10 and MF10 at 1.8 µmol/L. (c) Comparative K⁺ transport efficiencies (R_K⁺) between hetero-ensembles ((nFm·BFm)_n and (nFm·MFm)_n) and homo-ensembles ((nFm)_n, (BFm)_n and (MFm)_n (n = 5, 6, m = 8, 10). (d, e) Comparative values of R_K⁺ and EC₅₀ between (nXm·BXm)_n and (nXm)_n where n = 5, 6, m = 8, 10, X = F, L, V, I or A. HPTS = 8-Hydroxypyrene-1,3,6-trisulfonic acid.

  DownLoad: Full-Size Img PowerPoint

  Among four channel molecules (5F8, 5F10, 6F8 and 6F10), 5F10 produces the most active hetero-ensemble (5F10·BF10)_n, with a K⁺ transport efficiency (R_K) reaching 98.5% at a very low channel concentration of 1.8 µmol/L, and the remaining three channels require relatively high concentrations of 5.0, 9.0 and 4.0 µmol/L to generate similarly active hetero-ensembles with BFm (96.5%, 96.4% and 99.0%, respectively, Figs. 2b and c, Fig. S2 in Supporting information). These R_K values are much higher than the sums of R_K values from the two respective homo-ensembles (e.g., 31.7%, 29.5%, 52.0% and 57.1% for (5F8)_n+(BF8)_n, (5F10)_n+(BF10)_n, (6F8)_n+(BF8)_n and (6F10)_n+(BF10)_n, respectively) at the same concentrations. The highest enhancement by 2.3-fold was observed for 5F10 (98.5% for (5F10·BF10)_n vs. 29.5% for (5F10)_n + (BF10)_n). For comparison, for ensembles involving MFm (m = 8 and 10) that contains a methyl group at R₁ position, not only no significant differences were seen between homo- and hetero-ensembles, but also all ensembles exhibit low K⁺ transport activities of 25.8%–58.5% (Fig. 2c and Fig. S2).

  Having verified the above design hypothesis that a pair of mutually compatible bulkier Boc and benzo-crown ether groups indeed leads to substantially enhanced potassium transport activities, we subsequently looked into and compared the EC₅₀ values and K⁺/Na⁺ selectivity in fractional activity (e.g., R_K/R_Na) among the homo- and hetero-ensembles ((nFm)_n vs. (nFm·BFm)_n, n = 5 and 6, m = 8 and 10). Our recent works have demonstrated reliability of using R_K/R_Na to quickly gage the relative ion selectivity across different classes of channel molecules [20,55]. However, caution needs to be taken that R_K/R_Na could only serve as a good approximation of true ion selectivity intrinsic to any channel molecule, and the best method to determine the ion selectivity is via the use of single channel current measurement. From data gathered in Figs. 2d and e, Figs. S3-S8 (Supporting information), 5F10 combines with BF10 to afford the most active and selective hetero-ensemble (5F10·BF10)_n. In other words, compared to an EC₅₀ value of 2.8 µmol/L and a R_K/R_Na value of 9.1 in fractional activity for (5F10)_n, the corresponding values increase by 57% and 56% to 1.2 µmol/L and 14.2, respectively, for (5F10·BF10)_n. In general, hetero-ensembles containing 15-crown-5 units were found to be more selective than 18-crown-6-containing hetero-ensembles.

  An inspection of four more channel molecules derived from four other types of amino acids (5X10, X = A, V, L and I, Figs. 2d and e) establishes (5F10·BF10)_n, having an EC₅₀ value of 1.2 µmol/L and a R_K/R_Na value of 14.2, as the best hetero-ensemble in terms of both activity and selectivity. Based on (5F10·BF10)_n, the R₃ group was further varied to include straight alkyl chains of C₁₂H₂₅, C₁₄H₂₉ and C₁₆H₃₃, and our results show that (5F10·BF10)_n exhibits the highest potassium transport activity of 98.2% at 1.8 µmol/L and that R_K values are 59.5%, 11.3% and 0.8% for hetero-ensembles (5F12·BF12)_n, (5F14·BF14)_n and (5F16·BF16)_n, respectively. Nevertheless, with a slightly larger EC₅₀ value of 1.5 µmol/L (Fig. S9a in Supporting information), (5F12·BF12)_n turns out to possess the highest R_K/R_Na value of 34.7 at 2.6 µmol/L (Figs. 3a). This high selectivity is followed by (5F12)_n, having the second highest R_K/R_Na value of 24.6 and an EC₅₀ value of 2.4 µmol/L (Fig. S9b in Supporting information). In addition, deviations in 5F12:BF12 molar ratio from 1:1 drastically reduce ion selectivities of (5F12)_2n·(BF12)_n and (5F12)_n·(BF12)_2n by 47% and 79% to 18.5% and 7.4% (Fig. 3b and Fig. S10 in Supporting information), respectively. Selective ion transport could cause an asymmetrical accumulation of ions across LUVs, resulting in membrane polarization that can be probed by using voltage-sensitive Safranin O. For this experiment, LUVs, containing 100 mmol/L K₂SO₄, was diluted into a solution, containing 100 mmol/L Na₂SO₄ and 60 nmol/L Safranin O. After system equilibrium for 50 s, (5F12·BF12)_n or Valinomycin (VA, a highly selective K⁺ carrier) at various concentrations were added, and changes in fluorescence intensity were monitored for 300 s with fluorescence intensity at t = 0 s set to be 1 (Fig. 3c). Compared to control experiments where both intra- and extravesicular salts are kept as K₂SO₄ (denoted as "K₂SO₄ out" in Fig. 3c) as well as other channel/carrier concentrations (1.25–10 µmol/L for (5F12·BF12)_n and 0.01–0.08 µmol/L for VA, Fig. S11 in Supporting information), channel (5F12·BF12)_n at 6.5 µmol/L and carrier VA at 0.5 µmmol/L induce the largest changes of 157% and 146% in fluorescence intensity of Safranin O, respectively. These comparative data establish high K⁺/Na⁺ selectivity in K⁺ transport mediated by (5F12·BF12)_n.

  Figure 3

  

  Figure 3.  (a) Ion selectivity determined for (5F12·BF12)_n at 2.6 µmol/L of 5F12. (b) Ion selectivity (R_K⁺/R_Na⁺) determined for (5F12·BF12)_n at total concentrations of 4.9 µmol/L for 2:1 molar ratio, 5.2 µmol/L for 1:1 molar ratio and 5.4 µmol/L for 1:2 molar ratio. (c). Changes in fluorescence intensity of Safranin O (λex = 522 nm, λem = 581 nm; 60 nmol/L) after addition of (5F12·BF12)_n from 6 µmol/L to 7 µmol/L and Valinomycin (VA) from 0.03 µmol/L to 0.05 µmol/L. The 10 mmol/L HEPES buffer at pH 7 inside LUVs contains 100 mmol/L K₂SO₄ while external HEPES buffer at pH 7 contains either 100 mmol/L K₂SO₄ or 100 mmol/L K₂SO₄. The starting background intensity F₀ at t = 0 was set to be 1. (d) Single channel current traces, recorded in asymmetric baths (cis chamber = 0.25 mol/L K₂SO₄; trans chamber = 0.25 mol/L Na₂SO₄), for (5F12)_n and (5F12)_n·(BF12)_m. (e) Determination of K⁺/Na⁺ selectivity for (5F12)_n·(BF12)_m that may exist in various forms at diluted concentrations using linear ohmic current-voltage (I-V) curve.

  DownLoad: Full-Size Img PowerPoint

  Membrane integrity in the absence and presence of channels (5F12)_n or (5F12·BF12)_n was then evaluated using a self-quenching dye made up of a mixture of 5-carboxyfluorescein and 6-carboxyfluorescein (CF). These two CF isomers have dimensions of 9.2 × 11.4 Å and 10 Å × 10 Å (Fig. S12 in Supporting information), respectively. For this assay, CF dyes were trapped inside LUVs at a high concentration of 50 mmol/L such that they exist in mostly non-fluorescent dimer state. If channel molecules form a pore larger than 1 nm or result in membrane lysis, the leaked CF dimer molecules will largely dissociate to form more fluorescent monomers. As shown in Fig. S12, addition of (5F12)_n (EC₅₀ = 2.4 µmol/L) or (5F12·BF12)_n (EC₅₀ = 1.5 µmol/L) at 15 µmol/L into dye-containing LUVs does not produce any noticeable changes in fluorescence intensity relative to the background signals. These results are in sharp contrast to efflux of 59% and 99% dye molecules caused by Melittin at 0.25 and 1 µmol/L, respectively, thereby confirming that the observed changes in fluorescence intensity of HPTS or Safranin O in the presence of channel molecules are a consequence of the selective transport of K⁺ ions, rather than membrane lysis, by channel molecules.

  Single channel conductance measurements were then carried out on (5F12)_n and (5F12·BF12)_n, having the highest selectivities among all channels studied herein. The observation of a series of multiple single channel current traces, which were recorded at different voltages in unsymmetrical baths (cis chamber = 0.2 mol/L K₂SO₄ and trans chamber = 0.2 mol/L Na₂SO₄, Fig. 3d and Figs. S13-S15 in Supporting information), unambiguously confirm single channel mechanism underlying the selective K⁺ transport by both (5F12)_n and (5F12·BF12)_n. Plotting current vs. voltage (I-V curve, Fig. S13) gives rise to high K⁺/Na⁺ selectivity (P_K/P_Na) of 16.0 for (5F12)_n. Interestingly, for (5F12·BF12)_n, four sets of I-V curves (curves a-d, Fig. 3e and Fig. S14) could be identified from which P_K/P_Na values were determined to be 9.9, 20.1, 11.6 and 14.7 respectively. Coupled with the molar ratio-dependent changes in ion selectivity presented in Fig. 3b and high K⁺/Na⁺ selectivity of 16.0 for (5F12)_n, we believe a high P_K/P_Na value of 20.1 should result from (5F12·BF12)_n, and value of 14.7 can be assigned to channels containing more 5F12 than BF12 (e.g., (5F12)_n·(5F12·BF12)_m where n ≥ 1), and both 11.6 and 9.9 likely might correspond to (5F12·BF12)_m·(BF12)_n where n ≥ 1.

  Comparing changes in chemical shift of H-bond-forming protons is a well-established method for studying the association among molecules that contain H-bond donors and acceptors [58–61]. Here, an equal molar mixing of 5F12 and BF12 at 50 mmol/L produces not only well-defined ¹H NMR signals, but also noticeable downfield shifts of 0.11, 0.05, 0.04 and 0.10 ppm for protons H_a, H_b, H_c and H_d of hetero-ensemble (5F12·BF12)_n relative to those found in homo-ensembles (5F12)_n and (BF12)_m (Figs. 4a and b). These signify the formation of (5F12·BF12)_n from molecules 5F12 and BF12. Subsequent concentration-dependent ¹H NMR investigations from 0.78 mmol/L to 50 mmol/L lead to signals that can be fitted using a non-linear regression analysis method, suggesting these four protons be involved in forming defined H-bonds (Fig. 4c). But it needs to be pointed out that fitting using a simple dimerization equation to obtain association constants was not attempted since other oligomeric species may co-exist in equilibrium. Signal fitting is further complicated by the fact that formation of (5F12·BF12)_n requires self-assembled (5F12)_n to break first, a process that requires a very sophisticated treatment, which is beyond the scope of the current work. Interestingly, while non-linear regression analyses of protons H_a and H_b also support the formation of H-bonded homo-ensemble (5F12)_n (Fig. 4d), the same analyses on protons H_c and H_d prove inability of BF12 to form well-defined self-assembled (BF12)_n (Fig. 4e). As such, we believe monomer and other oligomers involving BF12 might co-exist in a poorly defined fashion. This likelihood can be corroborated by its large size of 6.2–6.8 Å (Fig. 1b), which is much larger than the inter-chain separation distance of 4.9 Å (Fig. 1a), and further by the molecular dynamics simulation (Fig. 4h).

  Figure 4

  

  Figure 4.  (a) Structure of hetero-ensembles (5F12·BF12)_n. (b) Comparative partial ¹H NMR spectra of (5F12)_n, (BF12)_n and (5F12·BF12)_n at 50 mmol/L at room temperature. (c-e) Concentration-dependent changes in chemical shift for protons a-d found in (5F12·BF12)_n, (5F12)_n and (BF12)_n from 0.78 mmol/L to 50 mmol/L at room temperature. (f-h) Illustrate molecular dynamics-simulated structures of (5F12·BF12)₃, (5F12)₆ and (BF12)₆ in POPC membrane. Small balls in gray (H-atoms) and red (O-atoms) represent atoms that are within H-bonding distance of < 2.5 Å and side chains R2 and R3 as well as non-amide H-atoms have been removed for clarity of view. POPC = 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine.

  DownLoad: Full-Size Img PowerPoint

  To obtain its possible structures in the membrane, a hybrid K⁺-selective ion channel (5F12·BF12)_n was computationally optimized using the COMPASS force field [46]. The resultant structure shown in Fig. 1c was then placed in a bilayer of 128 POPC molecules (1-palmitoyl-2-oleoyl-sn-glycero-3-phospho choline) solvated on two sides by 2 × 2397 water molecules (Fig. 4a) and subjected to MD simulation. Given the hydrophobic thickness of ~28 Å for POPC bilayer [62] and an inter-chain separation distance of 4.9 Å for (5F12·BF12)_n, n value was set to be 3 for MD simulation. After equilibration steps, the production run was carried out for 20 ns. Similar simulation steps were also performed on homo-ensembles (5F12)₆ and (BF12)₆. Highly populated and representative structures seen in the last 5 ns trajectories for (5F12·BF12)₃, (5F12)₆ and (BF12)₆ were presented in Figs. 4f-h.

  Structural analyses of these MD-simulated structures pleasantly reveal features that validate our above premise in using mutually compatible bulky groups (benzene-modified crown ether and tert-butyl group) to promote the formation of hetero-ensembles (e.g., (5Fm·BFm)_n or (6Fm·BFm)_n, m = 8, 10, 12, etc.) via a social self-sorting process. That is, an alternative arrangement of three molecules of 5F12, containing a bulky benzo-crown ether, with three molecules of BF12, containing a bulky tert-butyl group, indeed leads to an H-bonded hetero-ensemble (5F12·BF12)₃, possessing nine intermolecular H-bonds with a loss of one H-bond (Fig. 4f). In contrast, the bulky crown ether units in (5F12)₆ result in a loss of four H-bonds that are immediately adjacent to crown ethers, but appear to exert little influence on the remaining more distant five H-bonds (Fig. 4g). For (BF12)₆, the six bulky tert-butyl groups break the hexamer into three isolated species, i.e., trimer, dimer and monomer (Fig. 4h), which are consistent with the NMR dilution experiment (Fig. 4e). Although some structural differences for (5F12·BF12)₃ in the gas phase and in the POPC membrane can be noted (Fig. 1c vs. Fig. 4f), the three crown ethers are well separated from each other with ample spaces around them to mediate a faster and more selective transport of K⁺ ions (P_K/P_Na = 20.1, EC₅₀ = 1.5 µmol/L). On the other hand, a very crowded arrangement of six crown ethers in (5F12)₆ must greatly diminish the channel's ion transport activity and selectivity (P_K/P_Na = 16.0, EC₅₀ = 2.4 µmol/L).

  The ability to control self-assembly of two or more components is crucial in creating smart peptide biomaterials. With solid and consistent evidence, we have demonstrated here the steric code-mediated social self-sorting as an effective strategy for enhancing both ion transport activities and selectivities of K⁺-selective ion channels. In the first round of screening that focuses on eight K⁺-selective channels containing C₈H₁₇ or C₁₀H₂₁ side chains, this strategy achieves both substantial enhancements in K⁺ transport activity by 68% to 230% for all channels and significant increases in K⁺/Na⁺ selectivity, particularly for phenylalanine-based channels. From this screening, hetero-ensemble (5F10·BF10)_n emerge as the best channels in terms of both ion transport activity and selectivity. On this basis, the subsequent screening using the same strategy, which is in further combination with the side chain tuning, culminated in the discovery of highly sought-after hetero-oligomeric K⁺ channel (5F12·BF12)_n. While the HPTS-based assay gives rise to an excellent K⁺ transport activity (EC₅₀ = 1.5 µmol/L = 1.8 mol% relative to lipids, or 0.5 µmol/L, 0.62 mol% in terms of single channel concentration) for (5F12·BF12)_n with n likely equal to 3 in the lipid membrane, single channel current measurements yield the highest K⁺/Na⁺ selectivity (P_K/P_Na = 20.1) among all hitherto reported K⁺-selective channels.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  Acknowledgments

  This work was supported by the National Natural Science Foundation of China (No. 22271049), Fuzhou University, Xiamen University and Northwestern Polytechnical University.

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.108355.

  
  
• 1. [1]

     P. Agre, Proc. Am. Thoracic Soc. 3 (2006) 5–13. doi: 10.1513/pats.200510-109JH

  2. [2]

     K. Murata, K. Mitsuoka, T. Hirai, et al., Nature 407 (2000) 599–605. doi: 10.1038/35036519

  3. [3]

     E. Tajkhorshid, P. Nollert, M. Ø. Jensen, et al., Science 296 (2002) 525–530. doi: 10.1126/science.1067778

  4. [4]

     D.A. Doyle, J. Morais Cabral, R.A. Pfuetzner, et al., Science 280 (1998) 69–77. doi: 10.1126/science.280.5360.69

  5. [5]

     J.A. Mould, H.C. Li, C.S. Dudlak, et al., J. Biol. Chem. 275 (2000) 8592–8599. doi: 10.1074/jbc.275.12.8592

  6. [6]

     A. Vargas Jentzsch, A. Hennig, J. Mareda, S. Matile, Acc. Chem. Res. 46 (2013) 2791–2800. doi: 10.1021/ar400014r

  7. [7]

     J. Montenegro, M.R. Ghadiri, J.R. Granja, Acc. Chem. Res. 46 (2013) 2955–2965. doi: 10.1021/ar400061d

  8. [8]

     T.M. Fyles, Acc. Chem. Res. 46 (2013) 2847–2855. doi: 10.1021/ar4000295

  9. [9]

     F. Otis, M. Auger, N. Voyer, Acc. Chem. Res. 46 (2013) 2934–2943. doi: 10.1021/ar400044k

  10. [10]

      G.W. Gokel, S. Negin, Acc. Chem. Res. 46 (2013) 2824–2833. doi: 10.1021/ar400026x

  11. [11]

      Y. Zhao, H. Cho, L. Widanapathirana, S. Zhang, Acc. Chem. Res. 46 (2013) 2763–2772. doi: 10.1021/ar300337f

  12. [12]

      B. Gong, Z. Shao, Acc. Chem. Res. 46 (2013) 2856–2866. doi: 10.1021/ar400030e

  13. [13]

      Y.P. Huo, H.Q. Zeng, Acc. Chem. Res. 49 (2016) 922–930. doi: 10.1021/acs.accounts.6b00051

  14. [14]

      J.Y. Chen, J.L. Hou, Org. Chem. Front. 5 (2018) 1728–1736. doi: 10.1039/C8QO00287H

  15. [15]

      S.P. Zheng, L.B. Huang, Z. Sun, M. Barboiu, Angew. Chem. Int. Ed. 60 (2021) 566–597. doi: 10.1002/anie.201915287

  16. [16]

      A. Roy, J. Shen, H. Joshi, et al., Nat. Nanotech. 16 (2021) 911–917. doi: 10.1038/s41565-021-00915-2

  17. [17]

      T. Yan, S. Liu, J. Xu, H. Sun, S. Yu, J. Liu, Nano Lett. 21 (2021) 10462–10468. doi: 10.1021/acs.nanolett.1c03858

  18. [18]

      J. Shen, R.J. Ye, Z. Liu, H.Q. Zeng, Angew. Chem. Int. Ed. 61 (2022) e202200259. doi: 10.1002/anie.202200259

  19. [19]

      C. Lang, X. Deng, F. Yang, et al., Angew. Chem. Int. Ed. 129 (2017) 12842–12845. doi: 10.1002/ange.201705048

  20. [20]

      C.L. Ren, J. Shen, H.Q. Zeng, J. Am. Chem. Soc. 139 (2017) 12338–12341. doi: 10.1021/jacs.7b04335

  21. [21]

      M. Barboiu, Acc. Chem. Res. 51 (2018) 2711–2718. doi: 10.1021/acs.accounts.8b00311

  22. [22]

      L.Z. Zeng, H. Zhang, T. Wang, T. Li, Chem. Commun. 56 (2020) 1211–1214. doi: 10.1039/C9CC08396K

  23. [23]

      F. Chen, J. Shen, N. Li, et al., Angew. Chem. Int. Ed. 59 (2020) 1440–1444. doi: 10.1002/anie.201906341

  24. [24]

      S. Qi, C. Zhang, H. Yu, et al., J. Am. Chem. Soc. 143 (2021) 3284–3288. doi: 10.1021/jacs.0c12128

  25. [25]

      D. Qiao, H. Joshi, H. Zhu, et al., J. Am. Chem. Soc. 143 (2021) 15975–15983. doi: 10.1021/jacs.1c04910

  26. [26]

      A. Nakano, Q. Xie, J.V. Mallen, L. Echegoyen, G.W. Gokel, J. Am. Chem. Soc. 112 (1990) 1287–1289. doi: 10.1021/ja00159a083

  27. [27]

      G.W. Gokel, A. Mukhopadhyay, Chem. Soc. Rev. 30 (2001) 274–286. doi: 10.1039/b008667n

  28. [28]

      F. Otis, C. Racine-Berthiaume, N. Voyer, J. Am. Chem. Soc. 133 (2011) 6481–6483. doi: 10.1021/ja110336s

  29. [29]

      V. Gorteau, G. Bollot, J. Mareda, A. Perez-Velasco, S. Matile, J. Am. Chem. Soc. 128 (2006) 14788–14789. doi: 10.1021/ja0665747

  30. [30]

      C.R. Yamnitz, S. Negin, I.A. Carasel, R.K. Winter, G.W. Gokel, Chem. Commun. 46 (2010) 2838–2840. doi: 10.1039/b924812a

  31. [31]

      A. Vargas Jentzsch, S. Matile, J. Am. Chem. Soc. 135 (2013) 5302–5303. doi: 10.1021/ja4013276

  32. [32]

      T. Saha, A. Gautam, A. Mukherjee, M. Lahiri, P. Talukdar, J. Am. Chem. Soc. 138 (2016) 16443–16451. doi: 10.1021/jacs.6b10379

  33. [33]

      X. Wei, G. Zhang, Y. Shen, et al., J. Am. Chem. Soc. 138 (2016) 2749–2754. doi: 10.1021/jacs.5b12698

  34. [34]

      C. Ren, X. Ding, A. Roy, et al., Chem. Sci. 9 (2018) 4044–4051. doi: 10.1039/C8SC00602D

  35. [35]

      C.L. Ren, F. Zeng, J. Shen, et al., J. Am. Chem. Soc. 140 (2018) 8817–8826. doi: 10.1021/jacs.8b04657

  36. [36]

      W.L. Huang, X.D. Wang, Y.F. Ao, Q.Q. Wang, D.X. Wang, J. Am. Chem. Soc. 142 (2020) 13273–13277. doi: 10.1021/jacs.0c02881

  37. [37]

      L. Yuan, P. Jiang, J. Hu, et al., Chin. Chem. Lett. 33 (2022) 2026–2030. doi: 10.1016/j.cclet.2021.09.089

  38. [38]

      B.P. Benke, P. Aich, Y. Kim, et al., J. Am. Chem. Soc. 139 (2017) 7432–7435. doi: 10.1021/jacs.7b02708

  39. [39]

      L. Yuan, J. Shen, R.J. Ye, F. Chen, H.Q. Zeng, Chem. Commun. 55 (2019) 4797–4800. doi: 10.1039/C9CC00248K

  40. [40]

      A. Roy, H. Joshi, R.J. Ye, et al., Angew. Chem. Int. Ed. 59 (2020) 4806–4813. doi: 10.1002/anie.201916287

  41. [41]

      C.L. Ren, G.H.B. Ng, H. Wu, et al., Chem. Mater. 28 (2016) 4001–4008. doi: 10.1021/acs.chemmater.6b01367

  42. [42]

      C.L. Ren, J. Shen, F. Chen, H.Q. Zeng, Angew. Chem. Int. Ed. 56 (2017) 3847–3851. doi: 10.1002/anie.201611852

  43. [43]

      C.L. Ren, F. Chen, F. Zhou, Langmuir 32 (2016) 13510–13516. doi: 10.1021/acs.langmuir.6b04027

  44. [44]

      J.T. Li, Y.P. Huo, H.Q. Zeng, J. Mater. Chem. A 6 (2018) 10196–10200. doi: 10.1039/C8TA03415J

  45. [45]

      J.T. Li, Y.P. Huo, H.Q. Zeng, Langmuir 34 (2018) 8058–8064. doi: 10.1021/acs.langmuir.8b01643

  46. [46]

      H. Sun, J. Phys. Chem. B 102 (1998) 7338–7364. doi: 10.1021/jp980939v

  47. [47]

      B.H. Northrop, Y.R. Zheng, K.W. Chi, P.J. Stang, Acc. Chem. Res. 42 (2009) 1554–1563. doi: 10.1021/ar900077c

  48. [48]

      M.M. Safont-Sempere, G. Fernández, F. Würthner, Chem. Rev. 111 (2011) 5784–5814. doi: 10.1021/cr100357h

  49. [49]

      J. Shen, R. Ye, H. Zeng, Angew. Chem. Int. Ed. 60 (2021) 12924–12930. doi: 10.1002/anie.202101489

  50. [50]

      W. Jiang, C.A. Schalley, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 10425–10429. doi: 10.1073/pnas.0809512106

  51. [51]

      J.M. Lehn, Angew. Chem. Int. Ed. 54 (2015) 3276–3289. doi: 10.1002/anie.201409399

  52. [52]

      H. Jędrzejewska, A. Szumna, Chem. Rev. 117 (2017) 4863–4899. doi: 10.1021/acs.chemrev.6b00745

  53. [53]

      Z. Sun, M. Barboiu, Y.M. Legrand, E. Petit, A. Rotaru, Angew. Chem. Int. Ed. 127 (2015) 14681–14685. doi: 10.1002/ange.201506430

  54. [54]

      C. Ren, F. Chen, R.J. Ye, et al., Angew. Chem. Int. Ed. 58 (2019) 8034–8038. doi: 10.1002/anie.201901833

  55. [55]

      R.J. Ye, C.L. Ren, J. Shen, et al., J. Am. Chem. Soc. 141 (2019) 9788–9792. doi: 10.1021/jacs.9b04096

  56. [56]

      N. Li, F. Chen, J. Shen, et al., J. Am. Chem. Soc. 142 (2020) 21082–21090. doi: 10.1021/jacs.0c09655

  57. [57]

      H.Q. Zhao, S. Sheng, Y.H. Hong, H.Q. Zeng, J. Am. Chem. Soc. 136 (2014) 14270–14276. doi: 10.1021/ja5077537

  58. [58]

      H.Q. Zeng, R.S. Miller, I. Robert A. Flowers, B. Gong, J. Am. Chem. Soc. 122 (2000) 2635–2644. doi: 10.1021/ja9942742

  59. [59]

      H.Q. Zeng, H. Ickes, I. Robert A. Flowers, B. Gong, J. Org. Chem. 66 (2001) 3574–3583. doi: 10.1021/jo010250d

  60. [60]

      H.Q. Zeng, X.W. Yang, I. Robert A. Flowers, B. Gong, J. Am. Chem. Soc. 124 (2002) 2903–2910. doi: 10.1021/ja010701b

  61. [61]

      H.Q. Zeng, X. Yang, A.L. Brown, et al., Chem. Commun. 13 (2003) 1556–1557. doi: 10.1039/b301791e

  62. [62]

      Y. Guo, S. Pogodin, V.A. Baulin, J. Chem. Phys. 140 (2014) 174903. doi: 10.1063/1.4873586

• 

  Figure 1  (a) Self-assembled homo-oligomeric 1D structures (5F8)_n might undergo a structural transformation to form heteromeric structures (5F8·BF8)_n via a social self-sorting process. (b) Computationally determined molecular heights of tert-butyl group in Boc, benzo-15-crown-5 in C5, and 18-crown-6 in C6. (c) Computationally obtained near-perfect compatibility in geometry between benzo-15-crown-5 in 5F8 and tert-butyl group in BF8 in the hetero-ensemble (5F8·BF8)_n, having perfectly superimposable crown ether units. (d) Molecular design for combinatorial discovery of channel molecules that can pair with BF8 to produce heteromeric ensembles with enhanced ion transport activity and selectivity.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) LUV-based HPTS assay for evaluating ion transport properties of ion channels. (b) Typical K⁺ transport curves observed for hetero- and homo-ensembles such as those involving 5F10, BF10 and MF10 at 1.8 µmol/L. (c) Comparative K⁺ transport efficiencies (R_K⁺) between hetero-ensembles ((nFm·BFm)_n and (nFm·MFm)_n) and homo-ensembles ((nFm)_n, (BFm)_n and (MFm)_n (n = 5, 6, m = 8, 10). (d, e) Comparative values of R_K⁺ and EC₅₀ between (nXm·BXm)_n and (nXm)_n where n = 5, 6, m = 8, 10, X = F, L, V, I or A. HPTS = 8-Hydroxypyrene-1,3,6-trisulfonic acid.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) Ion selectivity determined for (5F12·BF12)_n at 2.6 µmol/L of 5F12. (b) Ion selectivity (R_K⁺/R_Na⁺) determined for (5F12·BF12)_n at total concentrations of 4.9 µmol/L for 2:1 molar ratio, 5.2 µmol/L for 1:1 molar ratio and 5.4 µmol/L for 1:2 molar ratio. (c). Changes in fluorescence intensity of Safranin O (λex = 522 nm, λem = 581 nm; 60 nmol/L) after addition of (5F12·BF12)_n from 6 µmol/L to 7 µmol/L and Valinomycin (VA) from 0.03 µmol/L to 0.05 µmol/L. The 10 mmol/L HEPES buffer at pH 7 inside LUVs contains 100 mmol/L K₂SO₄ while external HEPES buffer at pH 7 contains either 100 mmol/L K₂SO₄ or 100 mmol/L K₂SO₄. The starting background intensity F₀ at t = 0 was set to be 1. (d) Single channel current traces, recorded in asymmetric baths (cis chamber = 0.25 mol/L K₂SO₄; trans chamber = 0.25 mol/L Na₂SO₄), for (5F12)_n and (5F12)_n·(BF12)_m. (e) Determination of K⁺/Na⁺ selectivity for (5F12)_n·(BF12)_m that may exist in various forms at diluted concentrations using linear ohmic current-voltage (I-V) curve.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) Structure of hetero-ensembles (5F12·BF12)_n. (b) Comparative partial ¹H NMR spectra of (5F12)_n, (BF12)_n and (5F12·BF12)_n at 50 mmol/L at room temperature. (c-e) Concentration-dependent changes in chemical shift for protons a-d found in (5F12·BF12)_n, (5F12)_n and (BF12)_n from 0.78 mmol/L to 50 mmol/L at room temperature. (f-h) Illustrate molecular dynamics-simulated structures of (5F12·BF12)₃, (5F12)₆ and (BF12)₆ in POPC membrane. Small balls in gray (H-atoms) and red (O-atoms) represent atoms that are within H-bonding distance of < 2.5 Å and side chains R2 and R3 as well as non-amide H-atoms have been removed for clarity of view. POPC = 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine.

  下载: 全尺寸图片 幻灯片
•