Open Access Article
B. Ivonne.
Vergara-Arenas
,
J. Antonio.
Morales-Serna
and
Andrew. J.
Surman
*
Department of Chemistry, King's College London, 7 Trinity Street, London, SE1 1DB, UK. E-mail: andrew.surman@kcl.ac.uk
First published on 18th February 2026
Use of homogeneous catalysis – typically based on scarce precious metals – remains a dominant approach to afford good yields of enantiopure compounds. Combining typical strengths of heterogenous catalysts (low cost, sustainable, recyclable) with those of precious metal-mediated homogenous catalysis (amenability to design for selectivity) is desirable: several approaches have been demonstrated (chiral material surfaces, modification of surfaces with chiral auxiliaries, immobilisation of chiral catalysts), but it remains a challenge. Here we present a systems catalysis approach, with a heterogeneous material providing catalytic activity, and a separate host species controlling access to the catalyst to impart ‘prosthetic’ chiral selectivity. Since this non-covalent analogue to conventional covalent protecting group strategies is modular, the same substrate/host combination may be applied to a range of catalytic surfaces. The potential of this approach to achieve effective kinetic resolution is demonstrated in stereoselective synthesis of the drug (R)-cinacalcet.
By contrast, heterogeneous (insoluble/solid) catalysts are widely used in industrial chemistry, and typically are relatively cheap, sustainable, stable, recyclable, and easy to produce/obtain.34 However, heterogenous catalysts typically lack precisely-defined sterically-controlled active sites, akin to those which mediate control in homogenous catalysts, making design for (stereo)selectivity challenging.29,35–41
Attempts to achieve “the best of both worlds” – the selectivity of homogenous precious metal complexes, with the low cost and recyclability of heterogenous catalysts – have been numerous, but their success and adoption limited. Approaches to impart selectivity to catalytic surfaces by modifying them have included the tethering otherwise-soluble metal complexes to surfaces,42 or depositing chiral compounds on their surface,29,35–41 have met with some success, but are not widely reported in routine use. In all these approaches, the catalyst – or catalytic surface – is engineered for selectivity: a costly and laborious process, even when successful.
In this work, we explore an alternative: a modular systems approach, where instead of engineering the catalytic species for selectivity, we separate the roles of catalysis and selection (Fig. 1). A non-selective catalyst may be used, and “prosthetic” selectivity (selectivity not mediated by the catalyst) is imparted by molecular recognition of one enantiomer by stoichiometric amounts of a Supramolecular Protecting Group (SPG), preventing its reaction in a manner analogous to a covalently-bound protecting group. Using SPGs to control reactivity, where molecular recognition prevents access of reagents or catalysts to a substrate – or a region of a substrate – has been reported for some time,43 including for regioselectivity43–50 and kinetic resolution reactions,51,52 particularly following a landmark publication by Gibb et al. applying this principal to control a simple, otherwise-unselective, ester hydrolysis reactions.51 Indeed, the approach might be more well-recognised but for the diverse range of terms used to describe it (‘Supramolecular Inhibitors’, ‘Noncovalent Auxiliaries’, ‘Shadow Mask’, among others).53 However we are not aware of examples incorporating this into a catalytic system employing an unselective solid catalyst, which may be recycled.
Specifically, here we demonstrate how a Supramolecular Protecting Group may be applied to impart stereoselectivity to an otherwise-achiral catalytic system, a reductive amination: a common reaction in pharmaceutical synthesis.54 We employ an unmodified, commercially-available, solid acid catalyst, which is active, but not stereoselective, and an established host molecule (a modified β-cyclodextrin) which provides stereoselectivity by selectively recognising one enantiomer of the starting material selectively, preventing its reaction. We demonstrate this system in kinetic resolution in a model reductive amination reaction, and in the stereoselective synthesis of the drug (R)-cinacalcet, sold under the names Sensipar®, Mimpara®, or Regpara® in enantiopure form,55,56 from a racemic amine.
![]() | ||
| Scheme 1 Model reductive amination reaction, which may be carried out over a range of temperatures (see SI, Section 2). | ||
The amine, 1, was chosen, since a stereoselective host/SPG for this amine (6-O-triisopropylsilylated β-cyclodextrin, “TIPS-β-CD”, see Fig. 3) is already established by Kida et al.,52 achieving 2
:
1 stereoselective recognition of the (S)-1 and kinetic resolution in more simple model reactions at low temperatures (−20 °C). Amberlyst 15, a polymeric resin with strongly acidic sulfonic acid groups, with was chosen as an available, affordable, achiral, and reusable solid acid catalyst.57,58
Exploring reaction conditions in the absence of TIPS-β-CD, in toluene (see SI, Table S1; protic solvents absent) we found that the imine (3a) formation proceeds at reduced temperatures only in the presence of the Amberlyst 15 catalyst (Brønsted acid) (Table S1, Entries 4 & 5). Furthermore, the subsequent reduction of the imine (without isolation) to yield the amine (4a) product requires the catalyst, even at room temperature (Table S1, Entries 7 & 8): likely as the catalyst acts as a source of H+ (4.7 mmol H+ per g)59 for initiation of the reduction by NaBH4 in an aprotic solvent.60 Taken together, in the low-temperature conditions required for selective recognition of (S)-1 by TIPS-β-CD, the reaction only proceeds when catalysed by Amberlyst 15.
An important advantage of solid catalysts is recyclability. To screen the recyclability of the Amberlyst 15, two reactivation processes were applied to the material after catalysing reaction between 1 and 2a at in our reaction conditions (see SI, Fig. S1). While reuse of the catalyst after simple washing with an organic solvent (CH2Cl2) led to decreased activity in subsequent reactions, an additional wash with dilute aqueous hydrochloric acid maintained the efficiency of the catalyst for at least three subsequent reactions.
In our initial experiments (all 1 h imine formation; 1 h NaBH4 reduction), the reaction was conducted with 2 equivalents of (R/S)-1 at room temperature (relative to aldehyde 2a), and we obtained 90% conversion observed with 4% enantiomeric excess (ee). This is consistent with observations of mild enantioselectivity in 1/TIPS-β-CD interactions at room temperature.52 Progressively lowering temperature (all reactions: 1 h imine formation, 1 h reduction) led to progressively increasing chiral selectivity in product 4a, up to 90% ee, with only a minimal drop in conversion (to ca. 80%). Decreasing the amount of limiting reagent (aldehyde) to 0.2 eq. at the lowest temperature led to negligible change in selectivity (ee = 91%), and the use of a different aldehyde (2b, 4-nitrobenzaldehyde) provided similarly high yield and selectivity profiles to 2a (see SI, Section 7).
While fresh samples of TIPS-β-CD were used here, the host (used in stoichiometric amounts) can be readily recovered and reused after each reaction. To remove the amine 1, an extraction can be performed using ethyl acetate and an acidic aqueous solution, with the amine retained in the aqueous layer. After two extraction cycles, the TIPS-β-CD was not distinguishable from fresh host (unused as SPG) by 1H NMR (see SI, Section 9).
At room temperature, progressive changes in the chemical shift of the well-resolved host proton resonances (H1, H3, H6(a), H6(b), see Fig. 3 and SI Section 6; others are not well-resolved) on guest addition reveals binding with a fast exchange regime, in contrast to reports which observed slow exchange for 1/TIPS-β-CD binding in cyclohexane/benzene.52 This difference may be attributed to inclusion of bulkier solvent molecules (toluene) in the host cavity, analogous to inclusion of benzene in reported XRD-derived structures (no crystallisation was observed in our system, despite many attempts).52
Comparing chemical shift changes at multiple host protons (Fig. 3c and d) reveals the nature of the binding of (S)-1 and (R)-1 is qualitatively distinct. While binding of (S)-1 leads to marked changes in all four protons (suggesting binding deep in the cyclodextrin cavity to perturb H6), binding of (R)-1 affects H6(a) and H6(b) notably less, suggesting a distinct binding conformation for (R)-1 (likely sandwiched more ‘flatly’ between the two hosts) resulting from a poorer ‘fit’ in the host. This is also reflected in the broadening of some (S)-1 aromatic protons on binding TIPS-β-CD, not observed for the corresponding resonances from (R)-1 (see SI, Fig. S3). Rotating Frame Overhauser Effect Spectroscopy (ROESY) NMR spectra of the (S)-1 and (R)-1 recognition by TIPS-β-CD in the same conditions (see SI, Section 6) show interactions between the aromatic protons of 1 and cavity protons (H-3 and H-5) of TIPS-β-CD, with notably more intense cross-peaks with (S)-1 reflecting closer interaction.
Consistent with all other reports of guest binding by TIPS-β-CD binding in other solvents being 2
:
1 “sandwich-like”, a Job's plot suggests 2
:
1 host
:
guest binding (see SI Section 6).52,61–63 We applied a standard stepwise 2
:
1 binding model to fit binding (a simultaneous fit of the chemical shifts of H1, H3, H6(a), and H6(b) (see Fig. 3b, and SI, Section 6; we note that, since chemical shift changes are not exclusively monotonic responses, the unusual one-step 2
:
1 binding model applied elsewhere,52 or a 1
:
1 model, are not appropriate here). The binding constants estimated show modestly increased binding constants for (S)-1 (estimating K11 as 0.22 M−1 and K12 as 2000 M−1), relative to (R)-1 (estimating K11 as 0.08 M−1 and K12 as 1220 M−1), and in both cases these are consistent with 2
:
1 binding predominating in reaction conditions (i.e. in the absence of large excess of TIPS-β-CD). The small differences between these values reflect the scant selectivity observed in SPG-mediated reactions at room temperature.
Since greater selectivity in our SPG-mediated reactions is observed at −20 °C (−253 K), a series of 1H spectra were acquired at this temperature (see SI, Fig. S8; all resonances broadened at low temperature). While all room-temperature studies manifest fast guest exchange regimes, these spectra suggest slow guest exchange for the second (K12) binding of (S)-1, but fast exchange for the first binding (K11); contrastingly, both binding steps of (R)-1 appear in fast exchange. Furthermore, the disappearance of the resonance corresponding to unbound host on the addition of 1 equivalent of (S)-1 suggest complete guest binding in these conditions. Together these observations explain increased (S)-1 protection at low temperatures, however the mixed fast/slow regime observed for (S)-1 binding prevents fitting under these conditions using established models.
Overall these results support the reaction selectivity we observe being derived from selective binding of (S)-1 by TIPS-β-CD.
To establish this, we performed a series of reactions employing a range of solid acid catalysts (Montmorillonite, Graphene Oxide, MCM-41, SiO2), without further optimisation of conditions. We observe that in all cases where measurable yield is observed, selectivity remains almost constant, at around 90% ee. Very similar results were observed on varying the aldehyde to 2b (see SI, Table S3). This demonstrates the modularity of the system. While selectivity is unaffected, yield varies to a great extent, reflecting the optimisation of conditions/catalyst loading for Amberlyst 15 (Table 1).
| Entry | Aldehyde | Catalyst | Conv.b (%) | % eeb |
|---|---|---|---|---|
| a Reagents: (a) amine 1 (0.1 mmol), aldehyde 2a (0.05 mmol), TIPS-β-CD (0.4 mmol) and toluene (2 mL), 15 min; (ii) solid acid (50% mmol), 1 h; (iii) NaBH4 (0.20 mmol), 1 h at −20 °C. b The conversion of 2a to 4, and % ee of product 4a was determined by HPLC (see SI, Section 8). | ||||
| 1 | 2a | Amberlyst 15 | 78 | 90 |
| 2 | 2a | Graphene oxide | 28 | 87 |
| 3 | 2a | Montmorillonite | 20 | 87 |
| 4 | 2a | MCM-41 | — | — |
| 5 | 2a | SiO2 | — | — |
Following the approach developed in our model reactions, we performed implemented the reaction as shown in Scheme 2. Varying conditions (see SI, Table S4), we find similar responses to temperature: at room temperature no enantioselectivity is observed in the system. Lowering temperature to favour TIPS-β-CD acting as an SPG for (S)-1 (preventing access to catalyst and reagents), we are able to produce (R)-cinacalcet (4c) in good yield and high enantiopurity (up to 94% ee with 80% aldehyde conversion) from the racemic amine.
While we are not aware of previous applications to systems employing solid catalysts, we have noted that the use of Supramolecular Protecting Groups to control reactivity is not a new phenomenon (though nomenclature may vary). Since early reports,43 however, concepts and implementation of catalytic systems for synthesis have developed enormously. Where refining/engineering catalysts (the selectivity and activity of singles species) was long the overarching focus, increasingly catalytic systems are becoming accessible. Whether they employ multiple enzymes in a ‘cascade’,31 chemo–enzymatic combinations,65 or the use of multiple chemical catalysts,66 all these approaches share the division of roles into modules, comprising a (catalytic) system.67 We suggest that the value of SPGs is not as stand-alone alternatives to ‘traditional’ selective catalysts, but promising modules for use in developing more advanced catalytic systems. We hope this demonstration of their use in a catalytic system incorporating a solid catalyst, and multiple steps, will lead to the exploration of more complex catalytic systems.
| This journal is © The Royal Society of Chemistry 2026 |