Kwaku
Twum
a,
Seyed
Iraj Sadraei
b,
Jordan
Feder
a,
S. Maryamdokht
Taimoory
*bc,
Kari
Rissanen
*d,
John F.
Trant
*b and
Ngong Kodiah
Beyeh
*a
aDepartment of Chemistry, Oakland University, 146 Library Drive, Rochester, MI 48309-4479, USA. E-mail: beyeh@oakland.edu
bDepartment of Chemistry and Biochemistry, University of Windsor, 401 Sunset Avenue, Windsor, ON N9B 3P4, Canada. E-mail: Nazanin.Taimoory@uwindsor.ca; j.trant@uwindsor.ca
cUniversity of Michigan, Department of Chemistry, Ann Arbor, MI, USA
dUniversity of Jyvaskyla, Department of Chemistry, P. O. Box 35, 40014 Jyvaskyla, Finland. E-mail: kari.t.rissanen@jyu.fi
First published on 14th January 2022
N-Alkyl ammonium resorcinarene chloride receptors, NARX4, have been shown to act as high-sensitivity detectors of pyrophosphate (PPi), a biomarker of disease, in aqueous media through the chloride-to-PPi exchange [NAR(Cl)4 to NARPPi]. The nature of the anion of the macrocyclic NARX4 (X = Cl−, Br−, triflate OTf−) receptor greatly influences the PPi-affinity in aqueous media. The binding affinity for [NAR (Cl)4] is 3.61 × 105 M−1, while the NAR (Br)4 and NAR (OTf)4 show stronger binding of 5.30 × 105 M−1, and 6.10 × 105 M−1, respectively. The effects of upper rim ammonium cation, –N+H2R substituents (R = 3-hydroxypropyl, cyclohexyl, benzyl, or napththalen-1-ylmethyl), of the macrocyclic resorcinarene hosts have also been evaluated. The highest affinity was obtained using 3-hydroxypropyl groups due to the additional hydrogen bonds and the naphthyl upper-rim group that provides a larger hydrophobic surface area and favorable stacking interaction (i.e., π–π and CH–π). We note that two PPi molecules can bind to the more selective receptors through an additional interaction with the lower rim hydroxyls, making the resorcinarene a divalent binder. Comparing PPi with other phosphate anions (PO43−, AMP, ADP, and ATP) shows that the receptors are more selective for PPi due to the size and charge complementarity. Experimental (1H, 31P NMR, and isothermal titration calorimetry), and computational analyses support the reported trends for PPi selectivity even in highly competing aqueous media.
Our preliminary contributions reported on N-alkyl ammonium resorcinarene salts (NARX4), consisting of a macrocyclic resorcinarene tetra-ammonium cation and four anions. Chloride, viz. NAR (Cl)4, is preferred for resorcinarene conformational stability in organic media over the bromide, nitrate, triflate, or picrate.32–34 This binding preference is due to the chloride's suitable size, the snug fit between two adjacent ammonium moieties, and the strong hydrogen bond circular seam [NH2+⋯–Cl−⋯NH2+⋯–Cl−⋯]2 along the upper rim of the macrocycle that holds it in place. The tetra-cationic NARX4s are, on even a cursory inspection, excellent potential hosts for tetra-anionic pyrophosphate due to size-charge complementarity and chelate cooperativity. In 2018, we reported that an N-alkyl ammonium resorcinarene chloride receptor is capable of selective high-affinity (107 M−1) binding of PPi in pure water.35 Binding is driven by the entropically favorable displacement of the four chloride counter ions upon complexation of PPi. Based on the mechanism, we recognized the potential to improve the receptor's affinity by employing more weakly coordinating counter ions. The upper rim of the ammonium salt receptor is available for structural modification. In our previous study, a seemingly minor change, incorporating an additional methylene group converting the chain from ethanol to propanol, greatly improved PPi affinity.35 In these cooperative systems, even minor differences can drastically affect the thermodynamics of complexation.
We now report the application of less strongly coordinating counter anions, intending to significantly increase NARX4's affinity towards PPi. We synthesized ten new R-NARX4's (R = upper rim substituent, X = counter anion) receptors with three different counter anions: chloride, bromide, and triflate (OTf); and four different upper rim substituents: one with a flexible terminal hydroxyl propyl group (C3OHNARX4, X = Cl, Br, and OTf), the second with a rigid cyclohexyl group at the upper rim (CyNARX4, X = Cl and Br), the third with a flexible benzyl group (BnNARX, X = Cl and Br), and the fourth with a flexible and fluorescent napththalen-1-ylmethyl group (NpNARX4, X = Cl, Br and OTf, Fig. 1). In addition to PPi, the binding properties of the receptors towards a tribasic monophosphate (K3PO4), and a dibasic monophosphate (AMP), diphosphate (ADP), and triphosphate (ATP) are also investigated (Fig. 1). Binding was confirmed through 1H and 31P NMR experiments. Quantifying the thermodynamics of binding was accomplished using a series of isothermal titration calorimetry (ITC) and computational studies. The binding modes were justified using density functional theory (DFT) at the (wB96XD/6-311-G(d,p)) level of theory which was supported by molecular dynamic (MD) simulations.
Fig. 1 The resorcinarene salt receptors C3OHNARX4, CyNARX4, BnNARX4, NpNARX4 (X = Cl−, Br− and CF3SO3−), and the phosphate guests K3PO4, Na4PPi, Na2AMP, Na2ADP, and Na2ATP. |
Probing different functionalization of the upper rim alkyl (R) substituent was necessary to underscore its importance in defining phosphate anion affinity (Table 1). The switch to a combination of solvent systems is a likely explanation for lower Ka values than we earlier reported.35 The NARX4 with naphthalene groups show higher binding when compared to benzyl or cyclohexyl analogs, potentially due to the presence of a series of stacking interactions (Table 1). However, our previous C3OH chains provided the best affinity constant due to the extra hydrogen bond potential of the hydroxyl groups. On all scaffolds, counter anion identity showed the same trend; the K1 value for the complexation of PPi to Np-NARX4 increased by 52.5% and 68.2% (Table 1) upon switching the chloride counterion to bromide and triflate, respectively. The same trend remained when the water was replaced with pH 7.4 Tris buffer, confirming that the observed isotherms do not arise from (de)protonation but host–guest binding processes. Negative ΔG values in all cases confirm that association is spontaneous at 298 K. The ΔH and TΔS results also indicate the first binding event is both enthalpically and entropically favorable. This is slightly unexpected; entropically-driven complexation is the norm as solvent and counterions are liberated upon binding, but the energy of dehydrating pyrophosphate was expected to be substantial. The favorable ΔH term hints at the significant strength of the salt bridges formed in the binding pocket.37,38 Affinity is maintained in these new receptors: the K1 values for PPi are generally more than for AMP, ADP, and ATP (Fig. S23–S33 and Tables S1–S11†).
Complex | K 1 (×105) M−1 | ΔH1 kcal mol−1 | TΔS1 kcal mol−1 | ΔG1 kcal mol−1 | Complex | K 1 (×105) M−1 | ΔH1 kcal mol−1 | TΔS1 kcal mol−1 | ΔG1 kcal mol−1 |
---|---|---|---|---|---|---|---|---|---|
PPi@C3OHNAR(Br)4 | 5.3 ± 1.1 | −6.7 ± 2.2 | 1.09 | −7.81 | PPi@NpNAR(Cl)4 | 2.0 ± 0.5 | −3.3 ± 2.0 | 3.97 | −7.23 |
PPi@CyNAR(Br)4 | 2.6 ± 0.5 | −1.1 ± 0.5 | 6.29 | −7.38 | PPi@NpNAR(Br)4 | 3.0 ± 0.2 | −4.9 ± 0.6 | 2.55 | −7.47 |
PPi@BnNAR(Br)4 | 3.1 ± 0.2 | −0.4 ± 0.1 | 7.06 | −7.48 | PPi@NpNAR(OTf)4 | 3.3 ± 0.9 | 7.0 ± 0.7 | 14.5 | −7.51 |
PPi@NpNAR(Br)4 | 3.0 ± 0.2 | −4.9 ± 0.6 | 2.55 | −7.47 |
Complex | K 2 (×105) M−1 | ΔH2 kcal mol−1 | TΔS2 kcal mol−1 | ΔG2 kcal mol−1 | Complex | K 2 (×105) M−1 | ΔH2 kcal mol−1 | TΔS2 kcal mol−1 | ΔG2 kcal mol−1 |
---|---|---|---|---|---|---|---|---|---|
a ITC was done in H2O (90%) /DMSO (10%) at 298 K. K1 and K2 represent the first and second binding constants. | |||||||||
PPi@C3OHNAR(Br)4 | 1.40 ± 0.03 | 8.4 ± 1.7 | 15.3 | −6.94 | PPi@NpNAR(Cl)4 | 0.59 ± 0.08 | 11.1 ± 1.8 | 17.6 | −6.50 |
PPi@CyNAR(Br)4 | 0.18 ± 0.01 | 26.7 ± 0.8 | 32.2 | −5.60 | PPi@NpNAR(Br)4 | 1.13 ± 0.01 | 32.6 ± 2.1 | 39.6 | −7.00 |
PPi@BnNAR(Br)4 | 0.29 ± 0.03 | 16.0 ± 1.5 | 22.1 | −6.10 | PPi@NpNAR(OTf)4 | 4.6 ± 2.1 | −1.4 ± 0.2 | 7.72 | −9.09 |
PPi@NpNAR(Br)4 | 1.13 ± 0.01 | 32.6 ± 2.1 | 39.6 | −7.00 |
All the data, with few exceptions, were fitted to a two-set-of-sites binding model. As the cavity is committed to binding the first PPi molecule, the second interaction in C3OH-NARX may be allosteric exo binding with the hydroxyls of the top rim, as we previously speculated.35 However, as the other hydrophobic R-NARXs, lacking these hydroxyl hydrogen bond participants, present a similar two binding site event in this solvent mixture, this seems unlikely. There is, however, another allosteric binding site formed from the four dangling hydroxyl chains of the lower rim. To investigate this possibility, we turned to NMR spectroscopy.
Taking the binding of PPi by C3OHNARX4s as an example, up to 0.09 ppm upfield shifts are realized by the methylene protons, and up to 0.22 ppm upfield shifts are observed for the aromatic protons of the resorcinarene core. The changes in the host's signals support the host re-organizing the cavity during the binding process. No significant differences in the 1H NMR of the receptor in PPi@R-NAR(X)4 complex was observed between the chloride, bromide, or triflate counter ions once the counter anions are displaced to accommodate pyrophosphate. The final assembly is expected to be the same PPi@C3OHNAR(X)4, where X represents the initial counterion that has now been fully displaced (Fig. 2 and Fig. S36–S43†). An even better way to access the binding process is to monitor the 31P NMR signals of the phosphate guests upon complexation with the NARX4 receptors. A series of 31P NMR experiments of the phosphate anions and equimolar concentrations of pure phosphates and receptors were used to qualitatively probe the binding processes (Fig. 3 and Fig. S44–S56†). The magnitude of the upfield shift of the phosphorus isotopic signals upon binding the R-NARX4 receptors qualitatively infers how deep the guests’ sit in the host's binding pocket. Of the different R-NARBr4 receptors, PPi experienced the most significant upfield movement with R = naphthalene (Fig. 3). This was followed closely by the benzyl (Bn)–NARBr4 receptor and 1-propanol (C3OH)–NARBr4. Qualitatively, the absence of an aromatic environment or a hydrogen bond potential of hydroxyl groups in the cyclohexyl upper rim modification may explain why it has the smallest effect on the PPi resonance (Fig. 3, Table 1 and Fig. S44–S56†).
Additionally, it is evident how the different counter ions contribute to the shielding of the pyrophosphate anions. The PPi signal was shielded by 1.14 ppm with C3OHNAR(Cl)4, 1.76 ppm with C3OHNAR(Br)4, and 1.85 ppm with C3OHNAR(OTf)4. Larger upfield shifts of 2.42, 2.50, and 2.72 ppm were observed upon switching from chloride to bromide or triflate with the NpNARX4 receptors. The magnitude of these shifts is a qualitative representation of the different affinities. As the product is identical in each case, the changes are better interpreted as a measurement of the population ratios between bound (upfield shifted) and free (unchanged) PPi. This is further complicated by the second binding interaction of the guest (NaPPi) with the lower rim. Computational results suggest that PPi phosphorous atoms in the cavity have a higher point-charge electron density (Natural charge = 2.53e) than a PPi bound to the lower rim (Natural charge = 2.48e, Fig. S57†). The weaker the coordinating anion, the greater the binding affinity for the upper rim. Thus, one would expect the instantaneous relative population of PPi in the upper vs. lower binding sites to grow as the anion coordinating ability drops. The increased shielding observed for phosphorus in the triflate pro-receptor compared to the chloride or bromide strongly supports this inference. The most significant shifts are observed for largely dissociative triflate as this counterion is easiest to displace. This same differential effect is not immediately apparent in the 1H NMR spectra, where we note that the absolute shifts are similar. But they start from different starting points: the chemical shift of the indicated resonances in the parent resorcinarenes are not identical (although the peak shapes are), but upon binding, we do see differences in peak broadness, suggesting that although the PPi-bound complexes are expected to be all identical, we might be observing indicators of the dynamism of the binding process.
Similar 31P experiments were used to probe the binding of the receptors towards other phosphates: PO43−, AMP, ADP, and ATP. PO43− showed similar but much weaker upfield movement than PPi, while AMP showed greater field changes than PPi. Moderate but measurable deshielding was observed for ADP and ATP. The interaction of ADP and ATP with the host might be through an exo-binding mode, explaining the weaker effects; their larger size could be the main factor in this behavior. Table 1 summarizes the 31P signal changes upon complexation with the different R–NARX4s receptors.
Fig. 4 The optimized geometries and the calculated binding energies of three representative complexes of the chloride and triflate salts. NpNARTf4 could not be converged. |
This arrangement is maintained throughout the full 20 ns of our MD sampling. This bulky and more weakly coordinating counterion, by readily dissociating, should greatly increase the binding affinity of incoming PPi for the host. With this initial view of host-counteranions, we next investigated the complex formation with PPi.
The relative value of the predicted solvated binding energy of complex formation for PPi@C3OHNAR was found to be 1.8 kcal mol−1 (ΔΔG) more favorable than for the PPi@NpNAR complex. The very strong hydrogen bonds in the former are met by the formation of new intra-host upper rim π–π interactions between the naphthyl groups. It can be seen from the calculated host–guest and intra-host bond lengths (Fig. 7) that in PPi@NpNAR, four of PPi's oxygen atoms sit deep in the NpNAR cavity, forming strong interactions with the trimethyl ammonium and phenols, leaving only two PPi oxygen atoms to face the solvent. In PPi@C3OHNAR, the oxygen atoms of PPi are also immersed in the C3OHNAR host cavity, but less deeply and potentially more easily interrupted by the solvent. The deeper positioning of the guest in the PPi@NpNAR complex facilitates the intra-host interactions between upper rim OH⋯OH and the π–π interactions between the Np groups, which are obviously not available to the PPi@C3OHNAR complex. The interactions block solvent from almost half of the circumference of the complex, protecting the host–guest interactions from interference.
However, all these calculations simply look at the 1:1 system. With it becoming increasingly clear that a 2:1 PPi:host stoichiometry exists, we reoptimized the trimeric complexes (Fig. 8). As we had considered, the optimal structure does have the second PPi unit, in the form of Na4PPi, localized to the bottom rim of the cavitand. It does not adopt the same conformation for both systems; with C3OHNAR, it sits perpendicular to the axis of the cavitand, roughly parallel with the PPi in the upper cavity, while for NpNAR it sits aligned with the cavitand's axis. This difference is likely because these are both low-lying interactions, and in neither case can all four rim hydroxyls engage with the PPi. However, it would be sensible that the degree of entropic contribution (provided by PPi and cavitand desolvation) would drive this forward. The overall ΔG of binding to form this ternary complex, −60.6 kcal mol−1 for C3OHNAR and −61.6 kcal mol−1 for NpNAR, is roughly double that of the dimer above; binding to the lower rim is less favorable than the initial interaction with the cavity but is still highly exogenic. The reaction actually becomes endothermic on the ΔH term (sum of the binding energy and the zero-point correction energy), with the dimer (Fig. 6) being more favorable than the trimer; trimer formation is driven entirely by entropy. The entropy terms are overestimated in these calculations. This binding could likely be further improved through cooperative anion–cation interactions.
Fig. 6 The optimized geometry of PPi@NpNAR, and PPi@C3OHNAR complexes. BE is binding energy, Ezpe is the zero-point energy; these terms together equate to ΔH. |
In all, the computational data parallels the relative results of the experimental ITC data: binding is favorable for a two-site model, and the NpNAR forms the stronger interactions with PPi.
A detailed series of molecular dynamic simulations (MD) and the density functional theory (DFT) study highlighted the significance of the upper rim substituents and the counter anion as crucial factors for establishing the high binding affinity. These results show that modifying the upper rim and counter anions of the R-NAR(X)4 enhances their affinity and sensor ability towards PPi. Similarly, they support the contention that the second binding interaction with PPi is competitive and occurs with a specific conformation with the lower rim hydroxyls rather than as an exo-interaction with the cavity, as we had previously speculated. These results pave the way to using a supramolecular approach using cavity containing resorcinarene salts receptors as qualitative sensors for PPi in biological media.
Footnote |
† Electronic supplementary information (ESI) available: Synthesis, analytical methods, experimental details, NMR, ITC and UV-Vis/fluorescence data and detailed DFT. See DOI: 10.1039/d1qo01877a |
This journal is © the Partner Organisations 2022 |