**1. Introduction**

The use of self-assembled molecular containers such as coordination cages as catalysts for reactions that occur in the central cavity has provided some remarkable examples of synthetic hosts providing enzyme-like levels of the rate acceleration of reactions. The range of reactions that has been shown to be catalysed is now extensive [1–10].

Many examples of cage-based catalysis rely on the steric properties of the cavity to provide the catalytic e ffect. Thus, the early examples of the acceleration of Diels–Alder reactions occurred on the basis that co-location of the two components in the same cavity provided a high local concentration of the two reacting partners [11–14]. Unimolecular pericyclic reactions can be accelerated because the folding of the gues<sup>t</sup> allows it to bind in the cage cavity, resulting in a conformation that is close to the transition state [15–19]. Catalytic e ffects based on the electronic properties of the cage have also emerged, with photoinduced electron transfer between components of the cage walls and a bound guest, triggering useful reactions [20–23]; and an improved artificial 'Diels–Alderase' has been demonstrated, based on the electronic activation of the dienophile component by hydrogen-bonding interactions between the cage and guest, showing substantial rate enhancements without the need for the diene to be co-located in the cavity [24]. Possibilities for cage-based catalysis have been extended by the encapsulation of small-molecule catalysts, from mononuclear organometallic species to polyoxometallates, inside cage cavities [25–27]: in these cases, the cage itself is not the catalyst, but it modifies the behaviour of the bound catalyst that operates inside a constricted environment quite di fferent from that in the bulk solution.

*Chemistry* **2020**, *2*

We have recently demonstrated examples of catalysis using an octanuclear, approximately cubic, M8L12 coordination cage host denoted **H** or **Hw**, depending on external substituents (Figure 1) [10,28–30], which in water binds a wide range of hydrophobic guests in the central cavity, driven principally by the hydrophobic effect [31–33]. We note that a diverse range of octanuclear cages with the capacity to bind guests in the cavity is known [34–37]. The basis of gues<sup>t</sup> binding in our hosts **H**/**Hw** in solution is well understood to the extent that we have developed a reliable predictive model for quantifying the gues<sup>t</sup> binding free energies [38,39]. In addition to binding guests in the central cavity, the high positive charge [+16, arising from eight Co(II) ions] results in the accumulation of anions around the cage surface, resulting in a high local concentration of anions surrounding the guest, which is the basis for the catalysis [29,30]. We also showed that the binding of anions to the cage surface depends on how readily the anion can be desolvated, with chloride ions displacing hydroxide, and in turn, phenolate anions displacing chloride ions [30], allowing the nature of the anionic reaction partner surrounding a gues<sup>t</sup> to be controlled. Thus, the cage offers the possibility to co-locate (i) a substrate that binds via the hydrophobic effect, with (ii) a high concentration of anions that accumulate around it via ion-pairing, two orthogonal interactions that, in combination, could promote a wide range of catalysed reactions between organic substrates and anions in water. We note that this accumulation of counter-ions around charged cages that can participate in catalytic reactions has also been exploited by Raymond et al. in the opposite sense: they used highly anionic cages to stabilise protonated forms of cavity-bound substrates, even at high pH values [18,40–42].

**Figure 1.** The octanuclear [Co8L12](BF4)16 cages used in this work (**H**, R = H, [28]; **Hw**, R = CH2OH, [31]). (**a**) A sketch showing the approximate arrangemen<sup>t</sup> of metal ions and the structural formula of the bridging ligands, which span every edge of the cubic array of Co(II) ions; (**b**) view of the cationic cage cavity with each ligand coloured separately (from [28]).

We report here that our cage system can catalyse an aldol reaction: the conversion of indane-1,3-dione to bindone (Scheme 1) [43–46]. This was discovered by accident when we were evaluating the binding constants of a range of possible guests using spectroscopic titrations in solution; the addition of indane-1,3-dione (abbreviated hereafter as **ID**) to an aqueous solution of **Hw** resulted in the gradual appearance of a purple colour, which interfered with the titration experiment, but signalled the formation of the condensation product bindone. This did not occur in the absence of the cage under the same conditions. The facile aldol condensation of **ID** to give not just bindone, but also higher oligomers by multiple aldol-type reactions, has been known for over a century [45,46] and was recently re-studied in detail [43,44]. As **ID** has a p*K*a of close to 7, the reaction can occur under very mild conditions and can even be catalysed by the surface of laboratory glassware, meaning that spectroscopic studies need to be prepared and performed in either plastic or quartz vessels.

**Scheme 1.** Aldol condensation of indane-1,3-dione (**ID**) to bindone.

### **2. Materials and Methods**

The Co(II)-based cage **H w** was prepared as its fluoroborate salt through a previously-published method [28]. Indane-1,3-dione was purchased from Sigma-Aldrich; it reacts slowly with atmospheric moisture, so was dried under high vacuum, and stored in a desiccator. The instrumentation used for routine spectroscopic measurements was as follows: 1H-NMR spectroscopy, a Bruker Avance 300 MHz instrument; UV/Vis absorption spectra, an IMPLEN NanoPhotometer C40 cuvette reader, or BMG CLARIOstar plate-reader. Solution pH measurements were performed with a Hamilton Spintrode pH combination electrode calibrated with standards at pH 4.01 and 7.00.

Measurement of the binding constant of **ID** in **H w** was performed as follows. A series of 13 NMR tubes was prepared containing 0.6 mL of a D2O solution containing 0.2 mM **H w** at pD = 3.8, with the **ID** concentration varying from 0 to 1 mM (i.e., 0 to 5 equivalents) across the series. Spectra were recorded at 298 K and signals where free and bound **H w** could be seen separately were deconvoluted and integrated to allow the calculation of *K* (see main text).

Catalysis experiments were performed at 298 K using aqueous solutions containing **H w** (0.09 mM) and **ID** (0.9 mM) at pH 3.4 in a 1 cm path-length quartz cuvette. Progress was monitored on a UV/Vis spectrophotometer by growth in the absorbance of the product bindone at 550 nm (see main text).

The X-ray crystallographic data for the **H w**/**ID** complex were collected in Experiment Hutch 1 of beamline I-19 at the UK Diamond Light Source synchrotron facility [47]. Full details of the instrumentation, methods used for data collection, and for the solution and refinement of the structure, are as recently published [48]. Crystallographic, data collection, and refinement parameters are collected in Table 1. CCDC deposition number: 1979819.


**Table 1.** Summary of crystallographic, data collection, and refinement parameters for **H**•**(ID)3**
