Xudong
Ren
and
Anthony P.
Davis
*
University of Bristol, School of Chemistry, Cantock's Close, Bristol, BS8 1TS, UK. E-mail: anthony.davis@bristol.ac.uk
First published on 5th July 2025
The selective recognition of carboxylates in water, the biological solvent, could have various applications in biology and medicine. Of particular interest is the design of antibiotics which mimic the glycopeptides such as vancomycin through binding C-terminal peptide units involved in bacterial cell wall synthesis. Here we report a general approach to carboxylate receptors with structures capable of encapsulating and interacting with all parts of their substrates. The synthesis involves elaboration of a diamino bridge unit into a bicyclic system incorporating a tetralactam anion binding site. Water-solubility can be achieved in a final step which introduces two dendrimeric nonacarboxylate units via Cu(I)-catalysed azide–alkyne cycloaddition. Three examples have been prepared and found to bind simple carboxylates and polar inorganic anions with Ka up to ∼400 M−1 in water at near-neutral pH, despite the presence of polycarboxyl solubilising groups. Selectivities are modest, probably because of the flexible bridge units employed, but the versatile synthesis should allow access to a wide range of variants including some with potential for medical applications.
A selective receptor for carboxylate RCO2− will presumably need a binding site for the anionic group with extensions that can interact with group R, increasing affinity and conferring selectivity (see Fig. 1b). As indicated, cyclisation to link the extensions is likely to reduce flexibility and help preorganisation. We considered that core binding sites with well-defined 3D architectures would be useful to preorganise the extensions for binding to R, and the disubstituted tetralactam unit 418 (Fig. 1c) emerged as a favoured candidate. As described previously,19 versions of 4 with R′′ = aromatic groups did not perform well as receptors for simple carboxylates in water. We therefore decided to link the two R′′ groups to generate bicyclic structures, beginning with a bis(phenoxy)phenyl bridge derived from diamine 5 (Fig. 1d). Unexpectedly, the synthesis led to the tricyclic, dimeric structure 6 with two equivalent binding sites. Divalent receptor 6 was found to perform well as a receptor for carboxylates in water – the first example of a synthetic charge-neutral binding site with this capability.19 Notably, this was achieved despite the presence of multiple solubilising carboxylate groups.
Although 6 was remarkably effective, the dimeric structure is too large and complex to serve as a starting point for receptor development. Moreover, calculations suggested that the structure would adopt a folded conformation in water, lacking an open hydrophobic cavity. We therefore wanted to show we could access structures which align with our original concept, with just one tetralactam binding site and cavities capable of enclosing carboxylate substrates. Here we describe the synthesis of a series of bicyclic receptors 7–13 (Fig. 2), which conform to Fig. 1b. We show that they are effective in both organic media and water, and provide evidence for the expected binding mode. We also show that the design can be adapted to provide fluorescence sensing of anions in water. The work serves as an important step towards selective carboxylate recognition under biological conditions, including mimicry of the glycopeptide antibiotics.
![]() | ||
Scheme 1 General synthetic scheme for bicyclic carboxylate receptors based on tetralactam core 4. P1 = Fmoc for 7–11 (remove with NaOH/MeOH) and Teoc for 12–13 (remove with TBAF). |
In principle the scheme is highly versatile. The only limitation on the blue diamine component is that the two amino nitrogens must be close enough together for the final cyclisation, although it is also desirable that the amide groups in target bicycle 18 should all be available for H-bond donation to a carboxylate. In particular, the diamine need not be symmetrical; the C2 symmetry of the core tetralactam unit ensures that only one product is possible. In the present work, we chose diamino bridges that might assist binding and selectivity through hydrophobic and/or steric interactions with the carboxylate R-group, as well as generating a fluorescence response for use in sensing.
For initial trials, we chose the phenyloxy-p-xylyl-oxyphenyl bridge in 7–9, derived from diamine 19. Monte Carlo molecular mechanics searches suggested that, unlike 6, these structures should possess persistent amphiphilic cavities in water, defined by the tetralactam ring and the bridge aromatic surfaces (Fig. S104†). The apolar cleft formed by the bridge could accommodate straight-chain carboxylates up to butanoate (see Fig. S105–S108†). Diamine 19 was prepared and converted into intermediate 21via reaction with commercially available 20, as shown in Scheme 2a. Intermediate 21 was then carried forward to bicycles 7 and 8, as indicated in Scheme 1. For studies in water, the dendrimeric solubilising group X1 (Fig. 2) was added to 8via Cu-catalysed azide alkyne cycloaddition, as described previously,19 to give 9.
To generate a second example, the benzene unit in the bridge was mutated into a naphthalene, as in 10 and 11. This enlarged the cavity slightly, adding apolar surface, and introduced fluorescence in the region 300–480 nm. The intermediate 23 required to prepare 10 and 11 was synthesised via diamine 22, as shown in Scheme 2b. Intermediate 23 was converted into organic-soluble receptor 10 as indicated in Scheme 1, and thence into water-soluble 11.
Finally, apolar surface and fluorescence emission were further increased by introducing an anthracene unit, as in 12 and 13. The synthesis of 12/13 was complicated by the sensitivity of the anthracenyl CH2–O unit to the acid conditions needed to remove N-Boc groups, as discussed below, as well as solubility issues. Intermediate 25 was therefore prepared via a route which did not involve strong acid, and employed the Teoc-protected starting material 24 to increase solubility (Scheme 2c).
The acid-sensitivity of the anthryl CH2–O bond also raised issues later in the sequence, and these were resolved by an unusual stratagem. The source of the problem is the nature of the carbocation derived from CH2–O cleavage. As shown in Scheme 3a, acid-catalysed loss of ArOH from an intermediate such as 25 generates a cation stabilised by extended conjugation. Anthracenes are known to undergo Diels–Alder reactions across positions 9 and 10, and these can often be reversed by heating.20,21 The Diels–Alder adduct should not be acid-sensitive, and this could provide a straightforward way to protect the system from damage.
The scheme proved quite easy to implement. Tetracyanoethene (TCNE, 27)22 and close relatives23,24 were known to react efficiently and reversibly with 9,10-disubstituted anthracenes. Trial experiments on model compound 26 showed that the Diels–Alder, acid-catalysed Boc deprotection and reverse Diels–Alder all worked well (Scheme 3b). Accordingly, intermediate 25 was converted to bis-PFP ester 28 and carried through to diprotonated diamine 29 as shown in Scheme 4. 29 was then cyclised to 12 through addition to a tertiary amine, following the pattern in Scheme 1. For studies in water, the dendrimeric solubilising group X1 (Fig. 2) was installed via Cu-catalysed azide alkyne cycloaddition to give 13.
![]() | ||
Scheme 4 Conversion of intermediate 25 into receptors 12 and 13, employing TCNE 27 for anthracene protection. |
Schemes showing each sequence in full are available in the ESI.†
![]() | ||
Fig. 3 (a) 1H NMR spectra (500 MHz) from a titration of receptor 7 with TBA acetate in DMSO-d6. (b) Chemical shift measurements from the spectra in (a) fitted to a 1![]() ![]() |
The movements of protons j and k seemed to suggest a direct interaction between the host bridge and the guest, as expected if the acetate is positioned in the cavity. The separation of protons l suggests a loss of conformational freedom in the bridge, also consistent with intracavity binding. To provide further evidence of this binding geometry, a NOESY spectrum was acquired for a 1:
3 mixture of 7 and TBA acetate. As shown in Fig. 3c, a strong cross-peak was recorded between acetate CH3 and receptor proton j. The complex was modelled using molecular mechanics, confirming that these protons should be in close proximity (Fig. 3d and see also S106†). The results thus suggest that the substrate does indeed enter the cavity where, in principle, it can be affected by the choice of bridge.
The binding studies with TBA propionate and butyrate as substrates produced fairly similar results, the main difference being that the bridge NHa moved very little. The affinities are shown in Table 1, and were very similar to that for acetate. Again, NOESY spectra provided evidence that the anions entered the cavity of 7. For both anions, connections were revealed between α-CH2 and receptor protons j, and between the terminal CH3 and receptor protons m (Fig. S101 and S102†). For benzoate, protons b–d moved downfield, and k upfield, as for the other anions. However, in this case NHa moved upfield by ∼0.3 ppm, and bridge CHj also moved upfield significantly. The measured affinity was somewhat lower, at 3400 M−1. These differences are again consistent with the proposed mode of binding. Modelling suggests that benzoate is too large to enter the cavity but will clash with the bridge (Fig. S109†). The structure of the complex should therefore differ somewhat from those of the others. The lower binding constant could also be due to these steric interactions.
Substratea | Binding constants Kab (M−1) | |||
---|---|---|---|---|
7 | 8 | 10 | 12 | |
a As tetrabutylammonium salts. b Analysed using Bindfit.25 | ||||
Acetate | 10![]() |
11![]() |
10![]() |
10![]() |
Propionate | 10![]() |
10![]() |
||
n-Butyrate | 10![]() |
9100 (±1.9%) | ||
Benzoate | 3200 (±0.9%) | 3400 (±1.3%) | 3000 (±2.9%) | 2400 (±2.6%) |
Receptor 8 was also studied with all four anions, giving very similar results to 7. Receptors 10 and 12 were tested with just acetate and benzoate, showing similar affinities to 7 and 8. The full set of data is listed in Table 1.
1H NMR binding studies were performed in H2O/D2O 9:
1 at near neutral pH (7.30–7.65), with a range of anionic substrates (see Fig. 4a and Table 2). The guest solutions were carefully adjusted to match the pH of the host. The host concentration was always below the threshold for aggregation, as determined by the dilution studies (see above). Signal movements were observed, consistent with 1
:
1 binding which is fast on the 1H NMR chemical shift timescale. Fig. 4a shows an example involving receptor 11 titrated against L-lactate. Protons b, c and d, expected to interact directly with the carboxylate, were observed to move downfield for all receptor–substrate combinations, although the movements for 13 were smaller than for the other two receptors. Proton a generally moved slightly upfield, as did protons j and, to a lesser extent, k. Protons l again changed from a singlet to an AB quartet suggesting a loss of conformational freedom in the complex. For each titration, several signals could be followed and were analysed using Bindfit25 based on the 1
:
1 binding model (see for example Fig. 4b).
Substrate | 9 | 11 | 13 | ||||
---|---|---|---|---|---|---|---|
1H NMRa | ITCb | 1H NMRa | Fluorescencea | ITCb | 1H NMRa | Fluorescencea | |
a Analysed using Bindfit.25 b Analysed using an Excel spreadsheet developed in-house, see ESI. | |||||||
Formate | 70 (±4.5%) | 93 (±2.9%) | |||||
Acetate | 79 (±2.5%) | 33 | 123 (±3.6%) | 102 (±3.0%) | 54 | 154 (±4.5%) | |
Propionate | 100 (±1.7%) | 154 (±6.6%) | 144 (±4.6%) | 182 (±4.1%) | |||
n-Butyrate | 126 (±3.0%) | 93 | 154 (±5.3%) | 69 | 148 (±4.6%) | ||
Iso-butyrate | 134 (±3.6%) | 198 (±5.4%) | 182 (±2.7%) | ||||
Pivalate | 146 (±3.1%) | 174 (±6.4%) | |||||
Benzoate | 135 (±3.5%) | 110 (±8.0%) | 176 (±3.3%) | ||||
L-Lactate | 135 (±4.6%) | 185 (±5.4%) | 170 (±7.8%) | 161 (±4.7%) | 143 (±5.6%) | ||
D-Lactate | 140 (±3.9%) | 191 (±8.0%) | 181 (±3.3%) | ||||
Chloride | 96 (±1.7%) | 54 | 192 (±4.8%) | 207 (±10.3%) | 77 | 239 (±7.8%) | |
Bromide | 137 (±3.0%) | 203 (±5.4%) | |||||
Iodide | 112 (±3.1%) | 215 (±4.6%) | 205 (±4.9%) | ||||
Sulphate | 232 (±2.5%) | 406 (±7.0%) | |||||
Nitrate | 100 (±1.6%) |
The resulting Ka values are listed in Table 2. The affinities are generally consistent with those measured for dimeric receptor 6, considering that the latter benefits from a statistical contribution due to the two identical binding sites.27 As in the case of 6, selectivities between different carboxylates are modest. This suggests that binding is dominated by polar interactions, while hydrophobic and steric interactions with the bridge play a relatively minor role. Inorganic anions are also bound, especially the well-hydrated sulphate, again suggesting that polar interactions predominate. Differences between the receptors are also quite small, but seem to be significant in some cases. In particular, 13 is generally slightly stronger than 9, both for carboxylates and inorganic anions.
As discussed earlier, the introduction of fluorescence signalling was a key objective of this work. The naphthalene unit in 11 generated useful levels of emission between 300 and 500 nm if excited between 250 and 300 nm. Excitation at 250 nm was preferred for binding studies as this minimised the Raman emission of the solvent. Addition of anions caused a roughly two-fold increase in emission intensity, as exemplified for 11 + L-lactate in Fig. 4c. The changes were again consistent with 1:
1 binding, and analysis with Bindfit gave closely similar affinities to those measured by NMR (e.g.Fig. 4d). The anthracene unit in 13 generated peak emission at longer wavelengths than 11. However addition of L-lactate induced just a ∼1.3-fold increase in fluorescence. This could be analysed to obtain an affinity, but the low sensitivity discouraged further investigations. As further confirmation of binding, we also performed ITC measurements on 9 and 11 with acetate, isobutyrate and chloride. The results are listed in Table 2. Sensitivity limitations necessitated host concentrations above the aggregation threshold, so it is unsurprising that the values are slightly lower than those obtained by the other methods.
Finally, we sought direct evidence that the substrates enter the cavities of these water-soluble macrocycles, as already shown for 7 + acetate, propionate and butyrate in DMSO. NOESY spectroscopy was complicated by the low binding constants and the intense signals from the solubilising groups, especially as these tended to overlap with substrate protons of interest. However a NOESY spectrum of 9 + propionate showed cross-peaks which, despite overlaps, could be assigned with reasonable confidence to connections between propionate α-CH2 and receptor protons d and j, as expected for enclosed substrate. Further details are given in the ESI (Fig. S103).†
A second conclusion is that the tetralactam binding site 4, when constrained by a polycyclic architecture, is confirmed to be effective at binding carboxylates and other hydrophilic anions in water even in the presence of 18 solubilising carboxyl groups. The fluorescence results for 11 and 13 are especially helpful in this respect, providing complementary data that was not available for 6.
Thirdly, NOESY data support the proposed binding mode for carboxylates, in which the substrate enters the cavity as indicated in Fig. 1b. The shifts in bridge proton signals during 1H NMR titrations and the emission changes during fluorescence titrations are also supportive of intracavity binding. It might seem surprising that the bridge NHa NMR signals do not move consistently downfield on binding, but both DMSO and water are good H-bond acceptors which must be displaced by the carboxylates. Depending on the strength of the H-bonds formed on binding, substantial downfield movements are not necessarily expected. Similar behaviour was observed for the corresponding protons in 6.19
Fourthly, simple linearly connected bridges as in 7–13 are relatively ineffective in controlling selectivity, even in water where hydrophobic interactions can contribute to binding. These tris-aromatic straps are quite flexible and it seems that they cannot provide effective, well-defined hydrophobic pockets. However, given the versatility of the synthesis it should be possible to generate analogues with more structured cavities, capable of both polar and hydrophobic interactions. Improved discrimination should thus be possible, perhaps leading to selectivity for C-terminal peptides and ultimately to mimics of the glycopeptide antibiotics.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5sc04104j |
This journal is © The Royal Society of Chemistry 2025 |