Salt effects on the rates of a thiol cyclisation reaction within a yocto-litre inner-space

ABSTRACT The appreciation of the central role of Coulombic interactions in enzyme catalysis has led to the development of many ‘spin-off’ strategies for controlling chemical reactions. In particular, supramolecular chemistry has become increasingly proficient in using encapsulation/compartmentalisation to control both stoichiometric and catalytic reactions within the inner-spaces of hosts. This noted, there are still many open questions around the design of electrostatic potential fields within such hosts, and how exogenous factors can be used to fine-tune these properties. Here, we report on the cyclisation of 12-bromododecane-1-thiol 2 inside supramolecular capsule 1 2 to give thiacyclotridecane 3, and how the rate of this reaction changes as a function of exogenous salts. We find that this cyclisation is slowed in the presence of exogenous anions, with attenuation being highly dependent on both their nature and concentration. Thus, this work demonstrates how anions at the more-weakly solvated end of the Hofmeister series can associate with the outer walls of the capsule and so attenuate cyclisation. This suggests new ways in which reactions in inner-spaces can be indirectly modulated by exogeneous chemical entities. GRAPHICAL ABSTRACT

Parallel to all of these efforts, supramolecular chemistry has become increasingly proficient in using encapsulation/compartmentalisation to control both stoichiometric and catalytic reactions within the innerspaces of hosts .These advances noted, there are still many open questions regarding how molecular design 'translates' to the precise sculpturing of EPFs, how hosts therefore affect reactions within their innerspaces, and how exogenous factors can be used to finetune such properties [23].
Our work in this area focuses on the assembly of water-soluble deep-cavity cavitands [56][57][58] into yoctolitre supramolecular containers.Driven by the hydrophobic effect, containers such as 1 2 (Figure 1) can promote the cyclisation of both α,ω-thio-alkane halides (X (CH 2 ) n SH) [58] and α,ω-amino-alkane halides (X (CH 2 ) 14 NH 2 ) [59] to form macrocyclic thioethers or amines.With respect to the former, the acidity of the thiol guest -itself primarily dependent on the guest motif and the EPF engendered by the solubilising groups of the container -is key to the observed cyclisation rates.For example, the guest motif can change the acidity of the thiol by up to six pK a units [60, ], [61]and so lead to the spontaneous cyclisation of fourteen-or sixteen-carbon precursors simply by mixing with host 1 [58].However, for the corresponding negatively charged host capsule [55,56] reaction is at least four orders of magnitude slower because its EPF destabilises species along the thiolate-based mechanism [58].
To extend our understanding of these yocto-litre reaction flasks, we report here on the cyclisation of 12bromododecane-1-thiol 2 inside supramolecular capsule 1 2 to give thiacyclotridecane 3 (Figure 1), and how the rate of this reaction changes as a function of the nature of exogenous salts.We envisioned that certain anions would have a significant affinity for the outside of the capsule 1 2 , particularly for the crown of four trimethylammonium groups formed by the pendent groups of 1.As a result of anion binding, the EPF of the capsule would be attenuated, and hence the rate of formation of 3 reduced.We find that this general concept holds, with rates being slowed more in cases where exogenous anions have weaker enthalpies of hydration and therefore can bind more strongly to the outer walls of the container.

Host and guest synthesis, and complex formation
Positand 1 was synthesised as previously described [12].Guest 2 was synthesised from the corresponding diol by bromination (NBS/PPh 3 ), mono-thioacetate formation (KSAc), and acid hydrolysis (HBr in MeOH) [58].Guest 2 was selected so as to be readily fully deprotonated, yet cyclise slowly enough for the reaction to be amenable to [1]H NMR analysis.The slow rate of cyclisation (relative to longer homologues) arises because twelve C-atoms in the mainchain (14 non-hydrogen atoms in total) is close to the ideal length for a guest for it to adopt a linear/ extended motif [62][63][64] in which the terminal thiol and bromide groups are anchored in opposing poles of the capsule [58].In contrast, longer guests undergo motif compaction and begin to sample J-motifs with a turn on the main chain located in one pole; a motif that reduces the end-to-end separation and promotes cyclisation [62][63][64].
The requisite host-guest complex was formed by stirring excess guest 2 with a 1 mM solution of host 1, followed by filtering the solution through a 0.1 Durapore hydrophilic polyvinylidene fluoride (PVDF) filter membrane to remove excess guest [1].H NMR analysis confirmed that the guest possessed the expected linear/extended motif, with bound guest signals evident between 0.9 and −0.7 ppm [58].

Effects of exogenous anions
To examine the effect of ion binding on guest cyclisation we selected a range of monovalent anions covering the Hofmeister series; from small and charge-dense, to much larger and charge-diffuse.Specifically, we examined the effects of the sodium salts of F -, Cl -, Br -, NO 3 -, I -, triflate (TfO -or CF 3 SO 3 -), dichloroacetate (DCA -), SCN -, and ClO 4 -.We have previously observed that some of the anions -particularly the larger, charge-diffuse examples -can bind to the non-polar pocket of 1, a fact that could potentially affect the guest affinity or its binding motif by respectively competitive binding or co-encapsulation [64].To ensure that this was not the case, i.e. confirm that added salts were not directly affecting the bound guest, we first examined the 1 2 2 complex in the presence of 10 mM of the aforementioned salts and in the absence of base (Figures S3 and S4).These experiments confirmed that the guest binding region of the 1 H NMR spectrum of the complex did not change at 10 mM salt, confirming that there was no observable effect on affinity or motif.Thus, guest 2 bound sufficiently more strongly to the inner-space of the capsule than any of the anions that no guest displacement occurs, and that despite there being sufficient space in the 1 2 2 complex for co-encapsulation of charge-diffuse anions, this was not observed.
Building on this, we examined the rates of cyclisation of 2 within the capsule 1 2 as a function of the different salts.To trigger cyclisation, 10 mM NaOH was added to the solution of the host-guest complex.This concentration of base ensured a pH value of ~ 13 and, based on our earlier studies of the pK a of bound thiols [60], an estimated > 99% deprotonation of the guest (pK a estimated to be ~ 10).Interestingly, upon deprotonation there was no evidence of a motif change of the guest, suggesting that large, polarisable thiolates are readily accommodated within the capsule.The mechanism of deprotonation is unclear.The fact that the pK a of bound thiols is highly alkyl-chain dependent confirms deprotonation occurs inside the capsule [60].Studies of cyclisation rates as a function of equivalents of base suggest that hydroxide associates with the outside of the capsule [59], and relatedly, it is known that bromide (but not chloride) has a measurable affinity for the empty pocket of host 1. Hence in the presence of excess hydroxide it is not unreasonable for the base to enter the pocket and deprotonate the guest [64].As we discuss further below, entry of hydroxide does not need to involve complete disassembly of the capsule, however beyond all of these points the precise mechanism remains unclear.
Figures S5-S14 show stacks of the NMR spectra revealing reaction progress in the presence of each salt.Because the guest bound region of the 1 H NMR was relatively complex when both starting material and product were present, reaction progress was monitored using the H d proton signal of the host (Figure 1).This signal is very sensitive to slight changes in hemispherical separation due to changes in the shape and size of the guest and its complementarity with the inner-space of the capsule.As a result, the H b signal is an excellent reporter as the nature of the guest changes.Monitoring H d demonstrated that all reactions were completed within 75 minutes (Figures S15-S24).
We used a first-order kinetic model to fit the data and obtain apparent rate constants for the cyclisation of 2 (Table 1).For discussion here, Table 1 also shows prior data for the apparent rate constants for the cyclisation of the amine Br(CH 2 ) 14 NH 2 [59], as well as anion affinity data for the crown of four trimethylammonium groups formed by the pendent groups of 2 [64].The addition of salts slowed all thiol cyclisations, with first-order rate constants ranging from 35.8-83.7 10 −5 s −1 , compared to 143.7 10 −5 s −1 for the 'salt free' rate.In general, the cyclisation rate of 2 was approximately an order of magnitude faster than the cyclisation of the slightly longer amine Br(CH 2 ) 14 NH 2 , indicating the greater sampling of a J-motif by the slightly longer amine guest did not compensate for the weaker nucleophilicity of the amine group versus the thiolate of 2.
The presence of sodium iodide was noted to uniquely accelerate the cyclisation of Br(CH 2 ) 14 NH 2 (Table 1), a fact attributed to capsule breathing, i.e. its partial opening to allow the entry and egress of small chemical entities such as O 2 and I -without complete disassembly and liberation of the primary guest [53,65].Thus breathing was assumed to allow the entry of weakly-hydrated iodide, halogen exchange of the bound guest (R -Br → R -I), and hence reaction acceleration.Interestingly, this rate acceleration was not observed in the cyclisation of 2. There are two potential reasons as to why this is so.First, iodide ingress may be unfavourable.One piece of pertinent data here is that the I − ingress rate constant for the complex of (singlet-excited-state) pyrene inside a hexadeca-anionic capsule is ~ 6.0 × 10 [5] M −1 s −1 [66].Anion ingress into the positive capsule 1 2 is likely to be much faster, but the fact that the guest is itself charged (whereas the amine is neutral) may completely switch off iodide ingress.Second, halogen exchange may be rate limiting.Thus, halogen exchange with triphase catalysts tend to be relatively slow processes, and so may not be able to compete with fast thiolate attack [67].Unfortunately, our available data does not allow us to identify which of these two factors is rate determining overall.
The general slowing of reactions within 1 2 in the presence of anions can be attributed to anion complexation on the outside of the capsule, a reduction in the net EPF of the container, and hence reduced stabilisation of the internalised thiolate intermediate and the negatively charged transition state.We find that the most strongly hydrated fluoride has the weakest affect on cyclisation, whereas weakly solvated triflate and perchlorate reduce the reaction rate constant the most (Table 1).This point is illustrated by the distinct correlation (R 2 = 0.71) between the (available) [68] enthalpy of hydration values of the anions and the observed cyclisation rates (Figure S25).In other words, the weaker the anion solvation the greater its ability to associate with solutes, and hence the greater attenuation of the reaction within the inner-space of 1 2 .Are specific anion binding sites responsible for this correlation?Binding to the nonpolar cavity is switched off by the bound organic guest (vide supra), so the major route by which anions can affect the ESP is binding to either of the two 'poles' of the capsule, i.e. the two crowns of trimethylammonium groups [64]; as the data in Table 1 shows (K crown a ), even chloride has a marked affinity for these sites, whilst more charge diffuse anions have significant binding constants.Unfortunately, whilst a plot of cyclisation rate versus K crown a shows some correlation (Figure S26), it is obvious that attenuation is dependent on other modes of binding.On this topic, one distinct possibility is that in the dimer capsule there may be relatively strong binding sites for anions at the equatorial region between pairs of (inter-hemispherical) trimethylammonium groups.Such ion bridging would reduce the inter-hemispherical Coulombic repulsion between cavitand subunits.Additionally, it is also possible for some of the more charge-diffuse anions to bind to the under-side of the rim of the host (the upper-most, third row or aromatic rings as depicted in Figure 1).Such sites may arguably be classified as 'non-specific', but regardless of their classification, as neither of these binding possibilities can be readily quantified, their impact of cyclisation rates cannot be assessed.
We considered other potential correlations between the changes in cyclisation rate and various anion properties, but found none stronger than that between reaction rate and the enthalpy of hydration of the exogenous anions.Two worthy of note -for very different reasonsare the free energy of hydration and the Setschenow constant of the anions.In the case of the free energy of hydration, the correlation (data not shown) with the cyclisation rates was similar to that observed with the enthalpy of hydration.In the case of the Setschenow constant -a solubility-based measure of the salting-in/ salting-out properties of ions -surprisingly we observed no correlation with the rate of cyclisation.

Conclusions
Our results described here demonstrate that cyclisation reactions within the confines of host 1 2 can be modulated by exogenous anions added to bulk solution.We find no direct interaction between the cyclising guest and the added salt.Instead, reaction rates are modulated b Each experiment was carried out in duplicate, with the average of the calculated conversion percentage at each time-step used to give the corresponding rate constant.Rate constant errors are estimated to be less that 10%.c Data taken from reference [59].d Data taken from reference [64].
indirectly by anion binding to the capsule and modulation of its EPF.Given the complex nature of the capsule it is hard to accurately quantify anion affinity to any potential site other than the crown of trimethylammonium groups, and as a result correlations between affinity constants and rate constants are not strong.However, we find that in general terms the weaker the anion hydration the stronger its affinity for the capsule and the more a reaction rate is attenuated.Thus, for capsule 1 2 , maximal rate acceleration of thiol 2 cyclisation is attained using fluoride as the counter-anion in the absence of any added salt.Though not shown here, the laws of thermodynamics also dictate that dilution of the complex (to minimise ion-pairing) would also maximise the reaction rate.In contrast, higher concentration conditions [69] in the presence of charge-diffuse counter-ions or added salt maximise attenuation of the EPF of the capsule and slow the reaction to the greatest extent.We continue to investigate the reactivity of internalised guests within these types of capsules and will report on these findings in due course.