Article Cite This: Acc. Chem. Res. 2017, 50, 2870-2878
pubs.acs.org/accounts
Anion Recognition in Aqueous Media by Cyclopeptides and Other Synthetic Receptors Stefan Kubik* Fachbereich Chemie − Organische Chemie, Technische Universität Kaiserslautern, Erwin-Schrödinger-Straße, 67663 Kaiserslautern, Germany CONSPECTUS: Anion receptors often rely on coordinative or multiple ionic interactions to be active in water. In the absence of such strong interactions, anion binding in water can also be efficient, however, as demonstrated by a number of anion receptors developed in recent years. The cyclopeptide-derived receptors comprising an alternating sequence of Lproline and 6-aminopicolinic acid subunits are an example. These cyclopeptides are neutral and, at first sight, can only engage in hydrogen-bond formation with an anionic substrate. Nevertheless, they even interact with strongly solvated sulfate anions in water. The intrinsic anion affinity of these cyclopeptides can be related to structural aspects of their highly preorganized concave binding site, which comprises a wall of hydrophobic proline units arranged around the peptide NH groups at the cavity base. When anions are incorporated into this cavity they can engage in hydrogen-bonding interactions to the NH groups, and complex formation also benefits from cavity dehydration. Formation of 1:1 complexes, in which an anion binds to a single cyclopeptide ring, is associated with only small stability constants, however, whereas significantly more stable complexes are formed if the anion is buried between two cyclopeptide molecules. A major contribution to the formation of these sandwich complexes derives from the addition of the second ring to the initially formed 1:1 cyclopeptide−anion complex. This step brings the apolar proline residues of both cyclopeptides in close proximity, which causes the resulting structure to be stabilized to a large extent by hydrophobic effects. Solvent dependent binding studies provided an estimate to which degree these solvent effects contribute to the overall complex stability. In these studies, bis(cyclopeptides) were used, featuring two cyclopeptide rings covalently connected via linkers that enable both rings to simultaneously interact with the anion. Bis(cyclopeptides) with additional solubilizing groups allowed binding studies in a wide range of solvents, including in water. The systematic analysis of the solvent dependence of anion affinity yielded a quantitative correlation between complex stability and parameters relating to the solvation of the anions and solvent properties, confirming that solvent effects contribute to anion binding. Interestingly, the thermodynamic signature of complex formation in water mirrors that of sulfate binding to a protein complex but is opposite to that of other recently described anion receptors, which also do not engage in ionic or coordinative interactions with the substrate. These receptors not only differ in terms of the thermodynamics of binding from the cyclopeptides but also possess a characteristically different anion selectivity in that they prefer to bind weakly coordinating anions but fail to bind sulfate. Solvent effects likely control the anion binding of both receptors types but their impact on complex formation and anion selectivity seems to be profoundly different. Future work in the area of anion coordination chemistry will benefit from the deeper understanding of these effects and how they can be controlled.
■
INTRODUCTION The efficacy of a molecular recognition event in solution depends on a number of factors, among which the desolvation of the binding partners in regions that approach each other in the complex and the direct attractive interactions thus made possible are the most important ones. The respective binding strength can be understood and even quantitatively estimated by correlating the hydrogen bond donor and acceptor properties of the functional groups that engage in complex formation with the corresponding parameters of the solvent.1 If solute−solute interactions are stronger than solute−solvent interactions, complex formation is an exergonic process. If, however, one of the binding partners prefers to interact with the solvent, solute−solute interactions become inefficient. An interesting situation arises if the solvent−solvent interactions are stronger than solvent−solute interactions. In this case, © 2017 American Chemical Society
solvent molecules prefer to interact with themselves, forcing dissolved solutes to associate. Molecular recognition in water can therefore either be achieved by interactions that overcome the hydration of the binding partners or by making use of solvophobic effects. If the substrate is particularly strongly hydrated, which is often the case with inorganic anions, the first strategy may be the more promising one, explaining why many anion receptors rely on the strongest types of interactions such as multiple ionic or coordinative interactions to overcome solvent competition.2,3 There are, however, receptors that achieve high anion affinity in water without engaging in such interactions. The bambus[6]urils developed in the Sindelar group4,5 and Pittelkow’s Received: September 18, 2017 Published: November 10, 2017 2870
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research structurally related biotin[6]urils6,7 are examples. The relatively pronounced selectivities of these receptors for weakly coordinating anions suggest that their anion affinity could be mainly due to solvophobic effects rather than direct receptor− substrate interactions. Other examples are the anion-binding cyclopeptides developed in the group of the author, which use a combination of hydrogen-bonding and hydrophobic interactions for anion recognition. The fact that these cyclopeptides even bind the efficiently hydrated sulfate anion in water suggests differences in the binding modes of the cyclopeptide and glycoluril-derived systems. In this Account, the evolution of the cyclopeptide-based anion receptors will be presented and their properties related to those of other known anion binders active in water. Insights gained in this context could be of relevance for future research in the field of anion coordination chemistry.8,9
■
Figure 1. Crystal structure of the trihydrate of 1 (a).10 Protons except the ones on amide and proline Cα groups and solvent molecules are omitted for clarity. Figure 1b shows the electrostatic potential surface of the C3 symmetric conformation of 1, ranging from −120 to +120 kJ mol−1, with red and blue signifying values greater or equal to the absolute maximum in negative and positive potential, respectively. Figure 1c shows two arrangements of the amide group at the 6aminopicolinic acid residue with the repulsive interactions in one conformation denoted with a star.
ANION-BINDING CYCLOPEPTIDES The story of the anion-binding cyclopeptides began with the cyclic hexapeptide 1, comprising an alternating sequence of Lproline and 6-aminopicolinic acid subunits (Chart 1).10,11 Chart 1. Structures of Cyclopeptides 1 and 2
aqueous solvents, binding studies concentrated on anions derived from strong acids to avoid the use of buffers whose components potentially compete with the actual substrates for the binding site. Complex formation involves formation of 2:1 complexes in which the anion is sandwiched between two cyclopeptide rings as the crystal structure of the iodide complex shown in Figure 3a illustrates.10 No conformational reorganization of the cyclopeptide rings is necessary for the binding to occur. The iodide anion seems to have the ideal size to fill the available space between the two rings, allowing it to simultaneously engage in interactions with all six NH groups. The cyclopeptide rings are almost perfectly shape-complementary, allowing them to interdigitate, with the proline units approaching each other within van-der-Waals contact. Initial attempts to determine by NMR titrations the two microscopic equilibrium constants describing the formation of the 1:1 complex (K11) and its further conversion into the 2:1 complex (K21) failed because the regression analyses of the obtained binding isotherms proved to be very sensitive to the starting values used. A solution for this problem presented itself when studying the binding properties of the hydroxyprolinecontaining cyclopeptide 2 (Chart 1).12 Peptide 2 was synthesized in an attempt to increase the water solubility of 1, thus allowing binding studies in water. These studies showed that although the preferred conformations of 1 and 2 are overall similar, 2 only forms 1:1 complexes with the investigated anions. The stability of the sulfate complex amounts to 52 M−1 in water, for example, indicating that the bowl-shaped and relatively solvent exposed cavity of 2 provides a suitable environment for anion recognition even in the absence of strong interactions. Moreover, the underlying binding mode allows overcoming the Gibbs free energy required to at least partially dehydrate the sulfate anion, but the driving force that causes the 1:1 complexes of 1 to recruit a second cyclopeptide ring (causing complete sulfate desolvation) seems to be absent in the case of 2. The reasons are presumably steric factors (the hydroxy groups on the proline units prevent the proper mutual arrangement of two cyclo-
This peptide adopts averaged C3 symmetric conformations in polar solvents, characterized by an arrangement of the three aromatic subunits almost parallel to the C3 axis of the macrocycle and cis-conformations at the tertiary amide groups (Figure 1a). The proline rings surround a concave cavity into which the NH and CαH protons converge. To illustrate the steric and electronic environment inside the cavity, the electrostatic potential surface of 1 is depicted in Figure 1b. The positive potential inside this cavity and the conformational bias of the cyclopeptide, which is caused by electronic effects of the ring nitrogen atoms that control the arrangement of the neighboring amide groups (Figure 1c), render 1 well preorganized for anion binding. The interactions between 1 and anions in polar protic media such as 80 vol % D2O/methanol-d4 cause changes in the 1H NMR spectrum of the cyclopeptide. The most characteristic one is the deshielding of the proline Cα protons, which is caused by the spatial proximity of the corresponding protons to the anion in the complex.10 As an example, the effects of iodide on the 1H NMR spectrum of 1 in 80 vol % D2O/methanol-d4 are shown in Figure 2. NMR spectroscopy provided evidence that a variety of anions bind to 1 under these conditions, including chloride, bromide, sulfate, nitrate, acetate, phosphate, and carbonate. Since the latter three anions engage in protonation equilibria in 2871
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research
Figure 2. 1H NMR spectra of 1 in 80 vol % D2O/methanol-d4 (1 mM) in the absence (a) and the presence (b) of 5 equiv of sodium iodide. The signals of the CαH and aromatic C3H, C4H, and C5H signals are marked.
Figure 3. Crystal structure of the iodide complex of 1 (a)10 and of the sulfate complex of 4a (b).16 Protons except the ones on NH and OH groups, solvent molecules, and counterions are omitted for clarity.
increasing from chloride to iodide in the series of the halides and from the singly charged iodide anion to the doubly charged but similarly large sulfate anion. All binding constants K21 associated with the formation of the final complexes are significantly larger than the corresponding K11, indicating that complex formation is highly cooperative, that is, the main contribution to complex stability comes from the second step. The K21 values for the halide complexes again exhibit a small increase from the smallest to the largest halide. The smallest K21 is observed for sulfate binding, thus rendering the iodide complex of 1 the most stable one among the anions studied.
peptide rings) combined with solvent effects (desolvation of the hydroxyproline units renders sandwich complex formation enthalpically difficult). The inability of 2 to form higher complexes ultimately turned out to be advantageous as it allowed estimation of the intrinsic anion affinity of a single cyclopeptide ring in 80 vol % D2O/ methanol-d4. With this information in hand, the NMR titrations performed with 1 could be analyzed and the individual stability constants determined. The results thus obtained are summarized in Table 1.12 Table 1 shows that the stabilities of the 1:1 complexes of 1 and 2 correlate with the size and charge of the anions,
■
ANION-BINDING BIS(CYCLOPEPTIDES) The next step in the evolution of the cyclopeptide-based anion receptors involved covalently connecting two cyclopeptide rings to convert the 2:1 complexes of 1 into easier to characterize 1:1 complexes. The initial investigations focused on derivatives of 1 containing an amino group in one proline unit to enable reaction with dicarboxylates. A number of such bis(cyclopeptides) were studied with linkers of different length and flexibility.13,14 All of these bis(cyclopeptides) turned out to be less soluble than 1 so that the aqueous solvent mixtures used for the binding studies had to contain a higher content of the organic component. At the concentration used for these investigations, only the formation of the expected 1:1 complexes was observed in which an anion is sandwiched between two cyclopeptide rings in a similar
Table 1. Microscopic Stability Constants K11 and K21 and Overall Stability Constants log Ka of Anion Complexes of 1 and Stability Constants Ka of the Respective Complexes of 2 in 80 vol % D2O/Methanol-d4 (T = 298 K)a 1
2
anion
K11
K21
log Ka
Ka
chloride bromide iodide sulfate
5 16 22 96
6770 6820 7380 1270
4.53 5.04 5.21 5.09
8 13 19 95
K11, K21, and Ka in M−1; all anions were used as sodium salts; errors typically ±10%.
a
2872
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research Chart 2. Structures of Bis(cyclopeptides) 3 and 4a,b
(ΔH < 0) as well as entropically (TΔS > 0) favored in 50 vol % water/methanol. Substantial overall affinities thus result, with the dependence of halide affinity on the size of the anion mirroring that of 1. The most stable complex is formed in this case with sulfate but the dependence of the anion affinity of 1 on solvent composition indicates that iodide over sulfate selectivity is rather due to solvent effects than receptor structure. This assumption derives from the observation that the increase of the methanol content of the solvent causes a pronounced improvement of the sulfate affinity of 1 whereas the iodide affinity remains almost unaffected with respect to that in 80 vol % D2O/methanol-d4 (Table 1). As a consequence, sulfate becomes the better substrate in the mixture containing more methanol. Further investigations indeed demonstrated that the cause of the dependence of anion selectivity on solvent composition lies in anion solvation also in the case of the bis(cyclopeptides) (vide inf ra). Besides molecular modeling, also dynamic combinatorial chemistry was used to optimize the linker structure in these bis(cyclopeptides).15 To this end, a dynamic library was generated by treating a thiol derivative of 1 with various dithiols. A solvent mixture comprising 33 vol % water/ acetonitrile was used in these experiments in which all library components remained soluble during the equilibration. The addition of potassium iodide or sulfate to these libraries caused the amplification of bis(cyclopeptides) 4a and 4b (Chart 2), of which 4a afforded a crystal structure that confirmed the expected binding of the anion, in this case sulfate, in the cavity between the two cyclopeptide rings (Figure 3b).16 Calorimetric binding studies showed that 4a and 4b bind the anionic templates used for their selection up to 1 order of
fashion as in the complexes of 1. The formation of higher complexes in which anions bridge different receptor molecules seems to be entropically too unfavorable under these conditions. The most efficient dicarboxylate-linked bis(cyclopeptide), namely, 3 (Chart 2), was identified by using computational de novo structure-based design methods.14 Table 2 summarizes the calorimetrically determined affinities of 3 for a series of anions in 50 vol % water/methanol. For Table 2. Stability Constants log Ka of Anion Complexes of 3 in 50 vol % Water/Methanol Determined by ITC and Microscopic Stability Constants K11 and K21, and Overall Stability Constants log Ka of Selected Complexes of 1 Determined by NMR Titrations in 50 vol % D2O/Methanold4 (T = 298 K)a 3
1
anion
log Ka
K11
K21
log Ka
chloride bromide iodide sulfate
3.39 4.01 4.43 5.97
30 360
7670 8760
5.36 6.50
K11 and K21 in M−1; all anions were used as sodium salts; errors typically ±10%.
a
comparison, the stability constants associated with the formation of the sulfate and iodide complexes of 1 in the same mixture are also specified. Isothermal titration calorimetry (ITC) demonstrated that formation of the investigated complexes of 3 is enthalpically
Figure 4. Dependence of the thermodynamic parameters associated with the formation of the sulfate (a) and iodide (b) complexes of 4b on the mole fraction of water in water/acetonitrile (T = 298 K). The anions were used as potassium salts. 2873
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research Chart 3. Structures of Bis(cyclopeptides) 5a and 5b
■
WATER-SOLUBLE BIS(CYCLOPEPTIDES) Although an impressively high anion affinity log Ka of up to ca. 9 in 33 vol % water/acetonitrile could be achieved by covalently connecting two cyclopeptide rings via two linkers,19 the crucial question remained whether anion affinity is retained in water. To obtain information in this respect, water-soluble derivatives of 3 were synthesized by introducing tri(ethylene glycol) groups into positions where they were not expected to influence anion binding. A comparison of the thermodynamics of iodide and sulfate binding of the respective substituted receptor 5a (Chart 3) and the parent compound 3 in 50 vol % water/methanol indeed confirmed that the effects of the substituents on anion binding are negligible.20 Both new bis(cyclopeptides) 5a and 5b turned out to be water-soluble, with the more easily accessible 5a just sufficiently soluble for ITC titrations in water (0.25 mM). Concentrations up to 10 mM could be achieved with 5b, which rendered binding studies straightforward.17,20 The affinity of 5b for the halides and sulfate in water is summarized in Table 3.17 No binding of nitrate and perchlorate
magnitude more efficiently in 33 vol % water/acetonitrile than 3.14,15 Interestingly, sulfate complexation in the water/ acetonitrile mixture turned out to be endothermic and associated with a large favorable entropy term, whereas iodide complexation exhibits the same thermodynamic signature (ΔH < 0, TΔS > 0) as in water/methanol. The solvent therefore has profound effects on the thermodynamics of binding, an aspect that was studied in later work in more detail.17 Further insight into the factors responsible for the anion affinity of these bis(cyclopeptides) was gained by characterizing the solvent dependence of the iodide and sulfate affinity of 4b.16 The obtained Gibbs free energies, enthalpies, and entropies of binding in different water/acetonitrile mixtures are depicted graphically in Figure 4. Figure 4 shows that increasing the water content in the mixture causes an almost linear reduction of the Gibbs free energies of the binding of both anions in the solvent range studied. The reason is entropic since binding enthalpy benefits from the increase of the water content, going from endothermic to exothermic in the case of sulfate binding. The stronger dependence of sulfate complex stability on solvent composition with respect to iodide complexation thus parallels results obtained for 1 in water/methanol. Assuming that the reduction of the overall binding strength is partially caused by the extent to which anion desolvation becomes more difficult as the water content rises, the observed solvent dependence was correlated with the Gibbs free energies of transferring iodide from water to water/acetonitrile mixtures (respective energies for sulfate in this solvent system are not available).18 Endergonic transfer energies indicate, for example, that anion solvation is thermodynamically less favorable in the solvent mixture than in water, rendering desolvation easier. The results demonstrated that iodide affinity of 4b decreases over the whole solvent range much less strongly than expected on the basis of the respective transfer energies, suggesting that a compensating factor exists, which progressively stabilizes the complexes with increasing water content of the solvent. This stabilization was ascribed to the hydrophobic interactions between the apolar surfaces of the bis(cyclopeptide) proline units approaching each other in the complex that should benefit from the presence of more water in the mixture.
Table 3. Stability Constants log Ka, Binding Enthalpy ΔH, and Entropy TΔS Associated with the Formation of Different Anion Complexes of 5b in Water Determined by ITC (T = 298 K)a anion
log Ka
ΔH
TΔS
chloride bromide iodide sulfate sulfateb
2.15 3.23 3.62 3.31 3.11
10.1 3.6 −3.2 5.9 6.8
22.4 22.0 17.4 24.8 24.6
ΔH and TΔS in kJ mol−1; all anions were used as sodium salts; errors typically ±10%. bMeasured in phosphate buffer (35 mM, pH 7.4). a
anions was observed under the chosen conditions. Phosphate also does not bind, likely because repulsive interactions between the NH groups inside the cavity of 5b and the protons on the H2PO4− or HPO42− anions destabilize the respective complexes. Indeed, sulfate affinity of 5b is practically independent of whether the measurement is performed in water or 35 mM phosphate buffer (pH 7.4), indicating that 2874
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research
stable complex is formed in DMSO, however, which was unexpected. Moreover, iodide affinity changes in the region between water and a pure organic solvent in a nonlinear fashion as illustrated in Figure 5a. Plateau regions exist, in which iodide affinity is only weakly affected by solvent composition like in the case of 4b in water/acetonitrile (Figure 4). In contrast, sulfate affinity increases rapidly in all three solvent mixtures (Figure 5b) as the water content decreases until insufficient solubility of the sulfate salt prevented measurements in solvents with a high fraction of the organic component. The higher affinity of 5a for the weakly over the strongly hydrated anion in water suggests that anion solvation could be one parameter controlling complex stability. The low stability of the iodide complex in DMSO and the solvent dependence of the iodide affinity on solvent composition indicate, however, that anion solvation alone cannot be responsible for the observed effects. An attempt was therefore made to establish a linear free energy relationship in the form ΔG = gΔGtr + aA + ... that correlates the experimental Gibbs free energies of anion complexation ΔG with a combination of parameters describing anion solvation and solvent properties. As a measure for anion solvation, the Gibbs free energies of transferring iodide or sulfate anions into an organic solvent or an aqueous solvent mixture, ΔGtr, were again used.18 Among the parameters tested to describe the solvent properties, the only one that provided a satisfactory correlation turned out to be Kamlet−Taft’s α value, describing the hydrogen bond donor strength of the solvent.21 The multiparametric analysis afforded an equation that reasonably well describes the anion affinity of 5a in the different solvents and solvent mixtures, namely, ΔG = −(1.26 ± 0.44)ΔGtr − (19.94 ± 2.45)α.17 According to this equation, large positive ΔGtr values, which signify more facile anion desolvation with respect to water, are beneficial for complex stability, which agrees with the observations. The correlation between ΔG and α was attributed to the extent to which solvent reorganization at the binding site contributes to anion affinity, and α thus provides a quantitative estimate for the hydrophobic effects that mediate complex formation of this receptor. Insight into receptor hydration prior to complex formation was derived from the crystal structure of the trihydrate of 1.10 The three water molecules in this structure are arranged in a cyclic fashion and interact with one another by three intermolecular O−H···O hydrogen bonds (Figure 6). Their arrangement thus resembles the structure of a trimeric water
phosphate and sulfate anions do not compete for the binding site. The results in Table 3 demonstrate that 5b interacts with anions even in water, although less strongly than 3 in 50 vol % water/methanol. Binding of all anions except the weakly hydrated iodide anion is endothermic and therefore only favored by the large entropic term. Importantly, even sulfate can be bound under these conditions, albeit not as efficiently as iodide. As in the case of 1 and 3, iodide over sulfate selectivity thus seems to reverse when increasing the water content of the solvent, which was subsequently confirmed experimentally by following anion binding in a range of solvent mixtures. Halide affinity shows the typical trend, increasing with increasing size of the anion. To learn more about the effects of the solvent on anion affinity, binding of 5a to iodide was investigated in several organic solvents, namely, in methanol, acetonitrile, and DMSO.17 In addition, iodide and sulfate affinity of 5a was studied in aqueous mixtures of these solvents. The results are summarized in Table 4 and depicted graphically in Figure 5. Table 4. Stability Constants log Ka, Binding Enthalpy ΔH, and Entropy TΔS Associated with the Formation of the Iodide Complex of 5a in Different Solvents Determined by ITC (T = 298 K)a solvent
log Ka
ΔH
TΔS
water methanol acetonitrile DMSO
3.80 6.38 6.24 3.21
−4.8 −34.2 −48.0 7.1
16.9 2.2 −12.4 25.4
ΔH and TΔS in kJ mol−1; sodium iodide was used in water and tetramethylammonium iodide in the organic solvents; errors typically ±10%.
a
For reasons of solubility, the counterion had to be varied in these measurements but reference experiments did not reveal notable effects of the cations on anion binding. Ion-pairing effects therefore do not seem to influence anion affinity of 5a to a large extent, not even in the organic solvents. This result is consistent with other studies in which the counterion effect on anion binding of the cyclopeptides and bis(cyclopeptides) was investigated.12,13 Table 4 shows that, not surprisingly, iodide affinity of 5a is higher in methanol and acetonitrile than in water. The least
Figure 5. Dependence of the Gibbs free energy of formation of the iodide (a) and sulfate (b) complexes of 5a on solvent composition (water/ methanol, black; water/acetonitrile, red; water/DMSO, blue; iodide binding could not be studied reliably in water/DMSO because it is almost athermic). 2875
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research
Figure 6. Arrangement of the water molecules in the crystal structure of the trihydrate of 1 shown in a schematic fashion (a) and as a space-filling model (b).
cluster22 but involves further hydrogen bonds to the cyclopeptide NH groups. In addition, the protons not involved in hydrogen bonds to other water molecules interact in the crystal with carbonyl groups of neighboring cyclopeptides, indicating that hydration of 1 in solution could extend to the apolar regions of the proline rings. Thus, 1 likely features a cluster of water molecules in its cavity in the absence of guests whose release during complex formation should be thermodynamically beneficial. Cavity dehydration may thus be the reason why even monotopic cyclopeptides such as 2 possess a small anion affinity in water. Since the formation of sandwich complexes by 1 and the bis(cyclopeptides) requires desolvation of larger receptor areas, solvent effects can be expected to be more pronounced. The respective thermodynamic contributions to complex formation should be particularly large in water with its α value of 1.17. In solvents or solvent mixtures with smaller α values such as methanol or acetonitrile, solvation effects should be smaller until they completely disappear in DMSO whose α value is zero.
anion receptors feature multiple negative charges to mediate water solubility like the “octa-acid” deep-cavity cavitand developed by Gibb (6),25 Sindelar’s water-soluble bambus[6]uril (7),4,5 and Pittelkow’s biotin[6]uril (8) (Chart 4).6,7 In the Chart 4. Structures of Gibb’s Deep Cavitand 6, Sindelar’s Water-Soluble Bambus[6]uril 7, and Pittelkow’s Biotin[6]uril 8
■
COMPARISON WITH OTHER RECEPTORS This analysis thus indicated that anion affinity of 5a is mainly controlled by two parameters: anion solvation and reorganization of solvent molecules at the binding site during complex formation. Parallel trends observed in all binding studies performed so far such as the weak counterion effect on anion binding and the dependence of iodide over sulfate selectivity on solvent composition indicate that this correlation likely extends to the whole family of these cyclopeptide-derived receptors. It is interesting that the direct anion−receptor interactions seem to be relatively independent of the environment. The fact that the bis(cyclopeptides) bind the strongly hydrated sulfate anion in water but fail to interact with the weakly coordinating perchlorate anion indicates, however, that hydrogen-bonding between the peptide NH groups and the anion in a cavity well shielded from the solvent mediates selectivity to some extent. At this stage, it is interesting to compare the properties of the cyclopeptide-derived receptors to those of other receptors that bind anions in water without having to engage in ionic or coordinative interactions.23 Examples of neutral receptors are cyclodextrins, which have been shown to bind iodide, perchlorate, or anionic dodecaborate clusters.24 Other such
case of the anion-binding rotaxane recently described by Beer, cyclodextrin residues mediate water solubility.26 This receptor, whose iodide affinity in water is comparable to that of 5b, demonstrates that halogen-bonding represents a promising means to achieve anion recognition in water.27 Advantages of this interaction mode are the high directionality and the lower solvent dependence with respect to hydrogen-bonding. Indeed, halogen-bonding receptors often feature a higher anion affinity than the corresponding hydrogen-bonding counterparts.28 Since, however, Beer’s rotaxane also contains two positive charges in the vicinity of the binding site, the differentiation of the extent to which ionic interactions and halogen-bonding 2876
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research
A global understanding of the effects that contribute to the affinity and selectivity of such receptors in water is still missing. The binding studies performed with the cyclopeptide-derived receptors provided deep insight into relevant aspects. It should now be worthwhile to extend these investigations to the other receptor classes to gain more general information about the binding modes that allow anion recognition in water.
contribute to anion recognition is somewhat difficult, and the following discussion will therefore concentrate on receptors 6− 8. Complex formation of receptors 6−8 in water is strongly exothermic with an adverse entropic term, which is in striking contrast to the thermodynamic signatures associated with anion-binding of 5a (Table 4) but consistent with binding mechanisms involving the dehydration of large convex hydrophobic surfaces or the release of high-energywater.23,29,30 In the case of the γ-cyclodextrin/dodecaborate cluster complexes, the driving force of complex formation was indeed attributed to the enthalpically favorable but entropically opposed desolvation of the anions.24 Anion binding of 6 was explained by the balance between the solvation of the receptor cavity and the anion and the fact that anion binding does not require complete desolvation of either binding partner.25 The situation for receptors 7 and 8 is somewhat less clear because the binding studies were performed in phosphate buffer so that effects of the buffer anions on complex formation cannot be excluded. Recent theoretical studies confirm, however, that solvation effects contribute to the binding of these receptors as well,31 with potential additional contributions coming from receptor rigidification and effects of the carboxylate groups in the periphery.5 Solvation effects therefore seem to mediate anion binding of all of the presented receptors, but their impacts on the overall thermodynamics seem to differ. The contributions of anion desolvation on complex formation can be assumed to be similar in all cases if the anion is fully desolvated in the complex, and indeed, halide binding to receptors 7, 8, and 5a exhibits similar trends, becoming enthalpically more favorable with a concomitant adverse effect of entropy from chloride over bromide to iodide. This leaves the effects of receptor solvation and the direct receptor-anion interactions as the main causes for the different thermodynamic signatures of binding. Considering the different structures of the receptors, their hydration in water should indeed profoundly differ. Molecular dynamics simulations have, for example, shown that 6 contains four to five water molecules in its hydrophobic cavity,25 whereas a water dimer was found in the cavity of 8 by crystallography.7 Such linearly arranged water molecules are intrinsically less stable than the cyclic water cluster found in the trihydrate of 1,22 which is furthermore stabilized by direct interactions with the peptide NH groups. It is therefore possible that dehydration of 8 is associated with a favorable enthalpic contribution (nonclassical hydrophobic effect), while it could be more strongly controlled by entropy in the case of the cyclopeptides (classical hydrophobic effect).30 Dehydration of 6 should, however, also be entropically favorable,25 and in the absence of more detailed information about the hydration of all of these receptors in solution it is too early to estimate exactly how anion binding is controlled by solvation effects. It should nevertheless be noted that the thermodynamic signature of sulfate binding to a protein−protein complex resembles that of 5a,32 possibly indicating that this bis(cyclopeptide) imitates the binding mode of natural anion binders. With respect to the direct receptor−anion interactions, anion complexation of the cyclopeptides involves N−H···anion interactions, whereas the receptors in Chart 4 rely on dispersive or C−H···anion interactions.33 The softer nature of the latter types of interactions could be the reason why receptors 6−8 prefer to bind weakly coordinating, singly charged anions such as perchlorate but fail to bind the harder doubly charged sulfate.
■
CONCLUSIONS Anion receptors active in water do not necessarily have to use strong types of interactions to overcome competing effects of the solvent but can also benefit from solvation effects and still perform efficiently. Currently known receptors that presumably do so can be divided into two classes: those whose binding is exothermic but entropically unfavorable and those with the opposite thermodynamic signature of binding. More work needs to be done to understand these differences, how they relate to solvation, and whether they potentially also control anion selectivity to a certain extent. In the case of the cyclopeptides, anion affinity in water likely benefits from the converging arrangement of NH groups inside a concave cavity lined by hydrophobic subunits. These NH groups allow hydrogen-bonding interactions with the guest while the steric and electronic nature of the cavity side walls mediate the solvent effects. The resulting binding mode allows the complexation of even the strongly hydrated sulfate anion in water. The discovery of these cyclopeptides was serendipitous but a detailed understanding of the structural features responsible for their anion affinity could guide the design of new receptors with similar properties as the anion-binding pseudopeptide described some years ago demonstrates.34
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Stefan Kubik: 0000-0003-0526-7014 Notes
The author declares no competing financial interest. Biography Stefan Kubik is full professor of Organic Chemistry at the University of Kaiserslautern. He obtained his Diploma and Ph.D. degrees from Heinrich-Heine-Universität in Düsseldorf. After a postdoctoral stay at M.I.T., he began independent work in Düsseldorf that led to his Habilitation. He had the position of a substitute professor at Bergische Universität Wuppertal before joining the Chemistry Department in Kaiserslautern in 2004. His research interests encompass the design of cyclo(pseudo)peptide and gold nanoparticle-based receptors, and the development of scavengers for neurotoxic organophosphates.
■
ACKNOWLEDGMENTS The author thanks all co-workers and collaborating partners for their invaluable contributions. Funding from the Deutsche Forschungsgemeinschaft is also gratefully acknowledged.
■
REFERENCES
(1) Hunter, C. A. Quantifying intermolecular interactions: guidelines for the molecular recognition toolbox. Angew. Chem., Int. Ed. 2004, 43, 5310−5324.
2877
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
Article
Accounts of Chemical Research (2) Kubik, S. Anion recognition in water. Chem. Soc. Rev. 2010, 39, 3648−3663. (3) Langton, M. J.; Serpell, C. J.; Beer, P. D. Anion recognition in water: recent advances from a supramolecular and macromolecular perspective. Angew. Chem., Int. Ed. 2016, 55, 1974−1987. (4) Yawer, M. A.; Havel, V.; Sindelar, V. A bambusuril macrocycle that binds anions in water with high affinity and selectivity. Angew. Chem., Int. Ed. 2015, 54, 276−279. (5) Havel, V.; Babiak, M.; Sindelar, V. Modulation of bambusuril anion affinity in water. Chem. - Eur. J. 2017, 23, 8963−8968. (6) Lisbjerg, M.; Jessen, B. M.; Rasmussen, B.; Nielsen, B. E.; Madsen, A. Ø.; Pittelkow, M. Discovery of a cyclic 6 + 6 hexamer of Dbiotin and formaldehyde. Chem. Sci. 2014, 5, 2647−2650. (7) Lisbjerg, M.; Nielsen, B. E.; Milhøj, B. O.; Sauer, S. P. A.; Pittelkow, M. Anion binding by biotin[6]uril in water. Org. Biomol. Chem. 2015, 13, 369−373. (8) Sessler, J. L.; Gale, P. A.; Cho, W.-S. Anion Receptor Chemistry; RSC: Cambridge, 2006. (9) Bowman-James, K.; Bianchi, A.; García-Españ a, E. Anion Coordination Chemistry; Wiley-VCH: Weinheim, 2011. (10) Kubik, S.; Goddard, R.; Kirchner, R.; Nolting, D.; Seidel, J. A cyclic hexapeptide containing L-proline and 6-aminopicolinic acid subunits binds anions in water. Angew. Chem., Int. Ed. 2001, 40, 2648− 2651. (11) For an excellent review about cyclopeptide-based anion receptors, see Elmes, R. B. P.; Jolliffe, K. A. Anion recognition by cyclic peptides. Chem. Commun. 2015, 51, 4951−4968. (12) Kubik, S.; Goddard, R. Conformation and anion binding properties of cyclic hexapeptides containing L-4-hydroxyproline and 6aminopicolinic acid subunits. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 5127−5132. (13) Kubik, S.; Kirchner, R.; Nolting, D.; Seidel, J. A molecular oyster: A neutral anion receptor containing two cyclopeptide subunits with a remarkable sulfate affinity in aqueous solution. J. Am. Chem. Soc. 2002, 124, 12752−12760. (14) Reyheller, C.; Hay, B. P.; Kubik, S. Influence of linker structure on the anion binding affinity of biscyclopeptides. New J. Chem. 2007, 31, 2095−2102. (15) Otto, S.; Kubik, S. Dynamic combinatorial optimization of a neutral receptor that binds inorganic anions in aqueous solution. J. Am. Chem. Soc. 2003, 125, 7804−7805. (16) Rodriguez-Docampo, Z.; Pascu, S. I.; Kubik, S.; Otto, S. Noncovalent interactions within a synthetic receptor can reinforce guest binding. J. Am. Chem. Soc. 2006, 128, 11206−11210. (17) Sommer, F.; Marcus, Y.; Kubik, S. Effects of solvent properties on the anion-binding of neutral water-soluble bis(cyclopeptides) in water and aqueous solvent mixtures. ACS Omega 2017, 2, 3669−3680. (18) Marcus, Y. Gibbs energies of transfer of anions from water to mixed aqueous organic solvents. Chem. Rev. 2007, 107, 3880−3897. (19) Rodriguez-Docampo, Z.; Eugenieva-Ilieva, E.; Reyheller, C.; Belenguer, A. M.; Kubik, S.; Otto, S. Dynamic combinatorial development of a neutral synthetic receptor that binds sulfate with nanomolar affinity in aqueous solution. Chem. Commun. 2011, 47, 9798−9800. (20) Sommer, F.; Kubik, S. Anion binding of a neutral bis(cyclopeptide) in water/methanol mixtures containing up to 95% of water. Org. Biomol. Chem. 2014, 12, 8851−8860. (21) Marcus, Y. The use of chemical probes for the characterization of solvent mixtures. Part 2. Aqueous mixtures. J. Chem. Soc., Perkin Trans. 2 1994, 1751−1758. (22) Vaitheeswaran, S.; Yin, H.; Rasaiah, J. C.; Hummer, G. Water clusters in nonpolar cavities. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 17002−17005. (23) Biedermann, F.; Schneider, H.-J. Experimental binding energies in supramolecular complexes. Chem. Rev. 2016, 116, 5216−5300. (24) Assaf, K. I.; Ural, M. S.; Pan, F.; Georgiev, T.; Simova, S.; Rissanen, K.; Gabel, D.; Nau, W. M. Water structure recovery in chaotropic anion recognition: high-affinity binding of dodecaborate clusters to γ-cyclodextrin. Angew. Chem., Int. Ed. 2015, 54, 6852−6856.
(25) Sokkalingam, P.; Shraberg, J.; Rick, S. W.; Gibb, B. C. Binding hydrated anions with hydrophobic pockets. J. Am. Chem. Soc. 2016, 138, 48−51. (26) Langton, M. J.; Robinson, S. W.; Marques, I.; Félix, V.; Beer, P. D. Halogen bonding in water results in enhanced anion recognition in acyclic and rotaxane hosts. Nat. Chem. 2014, 6, 1039−1043. (27) Gilday, L. C.; Robinson, S. W.; Barendt, T. A.; Langton, M. J.; Mullaney, B. R.; Beer, P. D. Halogen bonding in supramolecular chemistry. Chem. Rev. 2015, 115, 7118−7195. (28) Kilah, N. L.; Wise, M. D.; Serpell, C. J.; Thompson, A. L.; White, N. G.; Christensen, K. E.; Beer, P. D. Enhancement of anion recognition exhibited by a halogen-bonding rotaxane host system. J. Am. Chem. Soc. 2010, 132, 11893−11895. (29) Biedermann, F.; Nau, W. M.; Schneider, H.-J. The hydrophobic effect revisited - studies with supramolecular complexes imply highenergy water as a noncovalent driving force. Angew. Chem., Int. Ed. 2014, 53, 11158−11171. (30) Hillyer, M. B.; Gibb, B. C. Molecular shape and the hydrophobic effect. Annu. Rev. Phys. Chem. 2016, 67, 307−329. (31) Cova, T. F. G. G.; Nunes, S. C. C.; Pinho e Melo, T. M. V. D.; Pais, A. A. C. C. Bambusurils as effective ion caging agents: Does desolvation guide conformation? Chem. Phys. Lett. 2017, 672, 89−96. (32) Waldron, T. T.; Modestou, M. A.; Murphy, K. P. Anion binding to a protein-protein complex lacks dependence on net charge. Protein Sci. 2003, 12, 871−874. (33) In the case of the cyclopeptides, contributions from C−H··· anion interactions to binding involving the proline Cα protons cannot be excluded. (34) Krause, M. R.; Goddard, R.; Kubik, S. Anion-binding properties of a cyclic pseudohexapeptide containing 1,5-disubstituted 1,2,3triazole subunits. J. Org. Chem. 2011, 76, 7084−7095.
2878
DOI: 10.1021/acs.accounts.7b00458 Acc. Chem. Res. 2017, 50, 2870−2878
