Open Access Article
Maria
Romero-Fernandez
a and
Francesca
Paradisi
*ab
aSchool of Chemistry, University of Nottingham, University Park, NG7 2RD, Nottingham, UK
bDepartment of Chemistry, Biochemistry and Pharmaceutical Sciences, University of Bern, Freiestrasse 3, Bern, Switzerland. E-mail: francesca.paradisi@dcb.unibe.ch
First published on 24th May 2021
As an alternative to classical synthetic approaches for the production of betazole drug, a one-pot biocatalytic system for this pharmaceutical molecule from its alcohol precursor has been developed. An ω-transaminase, an alcohol dehydrogenase and a water-forming NADH oxidase for in situ cofactor recycling have been combined to catalyse this reaction, yielding 75% molar conversion in batch reactions with soluble enzymes. This multienzyme system was then co-immobilised through a newly established protocol for sequential functionalization of a methacrylate-based porous carrier to enable tailored immobilisation chemistries for each enzyme. This pluri-catalytic system has been set up in a continuous flow packed-bed reactor, generating a space–time yield of up to 2.59 g L−1 h−1 with 15 min residence and a constant supply of oxygen for in situ cofactor recycling through a segmented air–liquid flow. The addition of an in-line catch-and-release column afforded >80% product recovery.
Betazole is an established drug in the pharmaceutical industry and a member of the pyrazole derivative family. This pharmaceutical molecule is an analogue of histamine and a H2-receptor agonist with the ability of stimulating the secretion of gastric acid.6,7 Clinical studies confirmed that betazole stimulation allowed an increase in both serum pepsinogen and gastric acid secretion even in patients with chronic Chagas’ disease subjected to surgical treatment, where reduced gastric acid secretion is relevant, since chagasic achalasia surgical treatment alters the esophagogastric junction anatomy.8
The first synthesis of substituted pyrazoles was carried out in 1883 by Knorr.9 For betazole, a two step synthesis was proposed by Jones, R.G.7, which evolved from a longer synthetic route developed in 1949.10 Hydrazine was first reacted with γ-pyrone in an alcohol solution, affording 3-pyrazoleacetaldehyde hydrazone in practically quantitative yields. A Ni-mediated catalytic hydrogenation of the latter compound afforded betazole in 80% yield.7 Nowadays, several relevant synthetic methods are being used to access the substituted pyrazoles.11
As an alternative to classical synthetic approaches, the design of a biocatalytic synthetic route for betazole has the potential to increase the sustainable production of this drug, but to date it has not been explored. In particular, the biocatalytic synthesis of the primary amine of betazole is relevant, since biocatalysis is a proven powerful tool in the synthesis of (enantiopure) amines, which are key intermediates in the fabrication of numerous active pharmaceutical ingredients (APIs).1 Both ω-transaminases (TA)12–17 and amine dehydrogenases (AmDHs)18,19 catalyse the selective production of primary amines from carbonyl compounds.
Furthermore, the synthesis of amines from the corresponding alcohols offers significant advantages since the alcohol is mostly easily accessible.20 In fact, different metal-catalysed methodologies have been proposed to carry out this reaction.21 No enzyme is however known to directly aminate an alcohol, and multienzyme systems have been proposed to catalyse this reaction in a two-step process: alcohol oxidation to form a carbonyl group, followed by reductive amination of the latter to produce a primary amine. Despite the apparent simplicity of the process, the technical challenges involved in establishing a reliable, sustainable and highly yielding system are significant. To date, only very few such multienzyme systems have been reported: galactose oxidase/TA/alanine dehydrogenase (AlaDH);22 alcohol dehydrogenase (ADH)/TA/AlaDH;23,24 ADH/TA/lactate dehydrogenase;24 ADH/AmDH;25 and alcohol oxidase/TA/AlaDH/catalase.26
The ADH-catalysed oxidation of alcohols requires stoichiometric amounts of oxidised nicotinamide adenine dinucleotide cofactors i.e. NAD(P)+, which represent a major cost contribution to the system. Therefore, an in situ regeneration system for the oxidised cofactor is essential for ensuring the cost-efficiency of the process. The integration of water-forming NAD(P)H oxidase (NOX) enzymes would appear to be ideal,27 as NAD(P)H is oxidized at the expense of the most innocuous and cheapest oxidant, O2, and producing H2O as a by-product.27 Several water-forming NOX have been described in the literature.28–30 However, the application of water-forming NOX to in situ cofactor recycling in continuous flow methodologies, which are increasing in popularity in the pharmaceutical industry, is not a straightforward process. O2 transfer from the gas to the aqueous liquid phase faces several well-known restrictions.31 Although efficient O2–liquid transfer can be achieved in conventional small-scale batch reactors by high levels of aeration and agitation,32 providing O2 to enzymatic reactions is especially challenging in standard biocatalytic flow reactors operating only with a liquid phase, where there is little gas (air)–liquid contact.
The aim of this work is to expand the applicability of synthetic biocatalysis by overcoming existing challenges that limit its implementation, adopting as a case study the sustainable production of betazole, a pyrazole derivative of pharmacological interest. For this purpose, biocatalysis and flow chemistry have been integrated to develop a one-pot multienzyme system in a continuous flow packed-bed reactor (PBR). Moreover, a novel protocol of multienzyme co-immobilisation on a polymethacrylate-based porous bead carrier has been developed. This system has been successfully applied to catalyse the direct synthesis of the primary amine of betazole from the corresponding alcohol.
![]() | ||
| Scheme 1 One-pot multienzyme system catalysing the functional group interconversion of the alcohol group of 2-(1H-pyrazol-3-yl)ethanol to the corresponding primary amine to produce a betazole drug. | ||
Initially, the catalytic activity of this multienzyme system on the alcohol group of 2-(1H-pyrazol-3-yl)ethanol was tested in batch reactions at a 10 mM scale with a catalytic amount of NAD+ (0.1 mol equivalents), using isopropylamine (IPA) as the amino-donor (2 mol equivalents) and the required enzymatic cofactors. High molar conversion (m.c.), 75%, to betazole was achieved after 21.5 hours under the specified reaction conditions (Table 1). HLADH and HEWT can therefore catalyse the functional group interconversion of the alcohol of 2-(1H-pyrazol-3-yl)ethanol into the corresponding primary amine to yield betazole in a one-pot system. The presence of LpNOX, despite its known low stability,29 increases the reaction velocity, meaning that in situ NAD+ recycling pushes the reaction equilibrium towards alcohol oxidation (Fig. S1†). This is likely due to the high redox potential of the O2/H2O couple that results in a strong thermodynamic driving force.27
In addition to the high conversions obtained by multistep flow biocatalysis in serial reactors for the synthesis of different molecules,37 one-pot flow biocatalysis based on a multistep enzymatic synthesis in a single reactor offers significant advantages.1,4,20 Moreover, immobilising the different enzymes on the same intrapore bead surface improves further the performance of biocatalytic cascade systems due to facilitated intrapore diffusion of substrates and cofactors between different enzyme active sites.42–44 Although co-immobilisation of multienzyme systems is highly challenging since there is no universal immobilisation chemistry or carrier, selective and optimal immobilisation of the different enzymes can be achieved by functionalization of the carrier surface with different reactive groups.42,44,45 However, no examples have been reported on polymethacrylate-based carriers, which are relevant for continuous flow applications.36,46,47 Therefore, a strategy for the co-immobilisation of this multienzyme system was developed on a porous bead carrier based on a polymethacrylate polymer matrix.
HLADH, HEWT and LpNOX enzymes were co-immobilised using tailored immobilisation chemistries for each of them. This was achieved by a novel protocol of sequential functionalization of a methacrylate carrier with different reactive groups (Scheme 2) to pursue optimal immobilisation for each of the enzymes and facilitate the diffusion of substrates between them. The epoxy groups displayed on the carrier surface were partially hydrolysed to diol groups in a controlled manner, and diol groups were then oxidised to glyoxyl groups. The carrier initial epoxy group density was determined to be 94 μmol gcarrier−1. After partial hydrolysis with H2SO4 and oxidation with NaIO4 treatment, the resulting glyoxyl group density was 78.3 μmol gcarrier−1, of which 14% resulted from the transformation of epoxy into glyoxyl groups, and 86% resulted from the transformation of diol groups initially present in the carrier into glyoxyl groups. The initial loading of epoxy plus diol groups of the carrier was determined to be 161 μmol gcarrier−1 (Table S2†). Since this strategy also makes use of the diol groups already present in the carrier, it offers the advantage of a better usage of the carrier surface and a higher efficiency of the resulting catalyst.
The resulting glyoxyl groups were used to immobilise the first enzyme, HLADH, by the multipoint-covalent attachment methodology previously described (Fig. S2†).40,46 This methodology is developed under alkaline conditions, where amine groups from lysine residues of the enzyme react with glyoxyl groups to form a Schiff's base which yields a secondary amine upon reduction with NaBH4.40 A minor proportion of HLADH was expected to immobilise via epoxy groups (Fig. S4†). Some of the remaining epoxy groups (∼12%) were then modified to amine groups upon reaction with ethylenediamine (EDA). The second enzyme, HEWT, was covalently immobilised by the two-step amino-epoxy strategy previously reported.48,49 This two-step approach is based on the first stabilisation of enzyme molecules by ionic interaction between their anionic residues and amine groups of the carrier, prior to the covalent bond formation between their nucleophilic groups and nearby epoxy groups on the carrier.48 The aim of this strategy is to overcome the low reactivity of epoxy groups found in enzyme immobilisation.48 Upon addition of polyethyleneimine (PEI), at least some of the unreacted epoxy groups covalently reacted with the amine groups of PEI, which was also used as a scaffold for the immobilisation by ionic adsorption43,50,51 of the third enzyme, LpNOX. With this strategy, ionic adsorption of enzyme molecules is established by ionic interactions between amine groups of PEI and aspartic and glutamic acid residues of the enzyme.43
Overall good values of immobilisation yield and recovered activity were obtained for the three enzymes (Table 2). Note that the low reported recovered activity for the HLADH could be possibly attributed to a mass transfer limitation at this high catalyst loading52 (Fig. S3†). High stability of HLADH and HEWT was achieved upon covalent immobilisation, and sufficient stability of LPNOX was reached by ionic adsorption on PEI (Fig. S5†). As a proof of concept, LpNOX could be efficiently desorbed by incubation with 0.2 M NaCl and fresh LpNOX could be immobilised on the same catalyst yielding 35% of the initial specific immobilised activity while preserving HLADH and HEWT immobilised activities (Table S3†) demonstrating the feasibility of the approach.
| Immobilisation yielda (%) | Recovered activitya (%) | Specific immobilised activitya (U gcarrier−1) | |
|---|---|---|---|
| a Immobilisation yield (%), recovered activity (%) and specific immobilised activity (U gcarrier−1) were calculated as described in the Experimental section. | |||
| HLADH | 100 | 9 | 2.3 |
| HEWT | 93 | 34 | 6.1 |
| LpNOX | 85 | 33 | 11.2 |
The catalytic activity of the resulting co-immobilised multienzyme system was evaluated on the alcohol group of 2-(1H-pyrazol-3-yl)ethanol in batch reactions under the same conditions as used with soluble enzymes. High m. c., 70%, to the corresponding primary amine was also achieved after 3 hours of reaction under specified conditions (Table 3). To demonstrate the reusability of this co-immobilised HLADH-HEWT-LpNOX biocatalyst, it was used to catalyse 3 batch reaction cycles. Remarkably, the catalytic efficiency of this multienzyme system was practically unaltered for 2 batch reaction cycles of 3 h (Table 3). After the second reaction cycle, the catalytic efficiency of the co-immobilised multienzyme system decreased considerably. This is likely due to the low stability of LpNOX as the activity of immobilised LpNOX on this multienzyme system dropped to 35% after 2 h of incubation under batch reaction conditions of temperature, pH and agitation, and to 1.6% after 24 h (Fig. S5†).
In systems requiring O2 as a substrate, as is the case for NOX catalysts, standard flow reactors can be highly limiting unless constant supply of O2 to the liquid phase can be guaranteed (e.g. tube-in-tube reactors,53 segmented air–liquid flow reactors32), or a strategy to increase the O2 concentration in the liquid phase (e.g. pressurized flow reactors31) is implemented. A simple solution when a packed-bed reactor is employed can be achieved with a segmented gas–liquid flow strategy composed of substrate solution/air. To date this was only reported with whole cell catalysts.32 Here this approach was trialled with cell-free immobilised catalysts to ensure a constant reservoir of oxygen to supply the liquid phase for the LpNOX-catalysed in situ cofactor regeneration in this PBR (Scheme 3, red).
A segmented 50
:
50 substrate solution/air flow was flowed into the PBR with 15 min residence time. This biphasic flow allowed alcohol amination with 50% m. c. To our knowledge, only one example of enzymatic amination of alcohols in continuous flow has been previously reported.54 However, the present work is the first example in the literature of a multienzyme system applied to amination of alcohols in continuous flow that uses TA, which is mostly relevant given the broad range of applications of this type of enzyme in the production of APIs and their intermediates.17
Recirculation of the reaction mixture through the PBR increased the m. c. to 72% highlighting a possible limitation in the residence time. A recirculation strategy of the reaction mixture (Scheme 3, green) was then further implemented.
A biphasic flow of 50
:
50 substrate solution/air and a residence time of 4 × 15 min pushed the conversion to 84%. Very low increment of space time yield (STY) with further passes was observed after the first pass, where substrate concentration drops to 5 mM. This suggests a km issue and poor enzymatic efficiency below such a threshold (≈5 mM) (Fig. S6†). When the reaction scale was increased to 50 mM, and the same flow reaction conditions were applied, 23% m. c. to betazole was obtained after 1 pass through this PBR and 66% after 4 passes. STY and catalyst productivity were higher at the 50 mM scale, although conversion was lower (Table 4). At the 30 mM scale, a compromise between conversion and space time yield was found. At this substrate concentration, 16% m. c. to betazole was obtained after 1 pass and 69% after 4 passes. After 20 passes, 85% m. c. was achieved.
| Scale (mM) | Residence time (min) | M. c.a (%) | STYb (gbetazole L−1 h−1) | Catalyst productivityb (μmolbetazole h−1 mgenzyme−1) |
|---|---|---|---|---|
| a Determined by HPLC. b STY (gbetazole L−1 h−1) and catalyst productivity (μmolbetazole h−1 mgenzyme−1) calculated as described in the ESI.† | ||||
| 10 | 1 × 15 min | 50 | 1.11 | 0.40 |
| 2 × 15 min | 72 | 1.61 | 0.58 | |
| 4 × 15 min | 84 | 1.86 | 0.67 | |
| 30 | 1 × 15 min | 16 | 1.07 | 0.39 |
| 2 × 15 min | 35 | 2.28 | 0.84 | |
| 4 × 15 min | 69 | 4.53 | 1.67 | |
| 20 × 15 min | 85 | 5.60 | 2.07 | |
| 50 | 1 × 15 min | 23 | 2.59 | 0.96 |
| 2 × 15 min | 44 | 4.91 | 1.81 | |
| 4 × 15 min | 66 | 7.37 | 2.72 | |
Assuming a volumetric composition of oxygen in air of 21%, the input oxygen flow rate was 0.025 mL min−1. At 30 °C and 1 atm, this would be equivalent to 1.012 μmoloxygen min−1 flowing to the PBR in the gas phase. The molar conversion to betazole at the 10 mM scale and 15 min residence time after 1 pass was 50%, which corresponded to 0.60 μmol product per min. At this reaction scale, the input NAD+ concentration was 1 mM; therefore, 0.12 μmolNAD+ min−1 was flowed into the system. Since the stoichiometry of the ADH-catalysed alcohol oxidation and concomitant NAD+ reduction is 1 to 1, 0.48 μmolNAD+ min−1 must be recycled in 1 pass (at least). This would indicate that 0.24 μmoloxygen min−1 were consumed in the enzymatic reaction under these conditions. In the case of the 50 mM scale reactions, the oxygen intake rate by the enzymatic reaction in the first pass was estimated to be 0.39 μmoloxygen min−1. Moreover, a segmented liquid–gas flow was required to ensure a constant reservoir of oxygen to supply the liquid phase and make the reaction work (only 10% m. c. was obtained without air supply in the 10 mM reaction). Therefore, 0.24 μmoloxygen min−1 were then supplied from the gas to the liquid phase in the 10 mM scale reaction, and 0.39 μmoloxygen min−1 in the 50 mM scale reaction, under the applied flow conditions, and consumed by the enzymatic reaction.
At the 50 mM scale without any recirculation, this one-pot multienzymatic system yields 2.59 gbetazole L−1 h−1 in continuous flow, which is mostly relevant since it represents the first example of a biocatalytic synthetic step in the fabrication of betazole. As opposed to classical synthetic routes to this pharmaceutical molecule,7,11 this metal-free biocatalytic approach allows the synthesis of the primary amine of betazole from the corresponding alcohol by means of a highly yielding system using only water as the solvent at 30 °C and atmospheric pressure, and without the formation of unwanted side-products. Moreover, the co-immobilisation strategy of this multienzyme system increased biocatalyst reusability, thus increasing the catalyst productivity in continuous flow. Considering a window of 5 h of operational stability of the co-immobilised multienzyme system under specified flow conditions (Table 4), the accumulated catalyst productivity is 4.8 μmolbetazole mgenzyme−1.
Finally, a step to trap the primary amine-containing molecules (which include also unreacted IPA as well as betazole) was integrated in the system. A tailor-made aldehyde resin was packed in a column and placed downstream of the recirculating flow reaction. This amine catch-and-release step had an efficiency of 93% with respect to the recovered betazole after desorption with 0.2% HCl solution (Fig. S8†). The IPA could be successfully removed by freeze-drying the eluted fractions, as shown by 1H NMR and 13C NMR. The betazole-HCl salt form of the product was isolated with an apparent 94% yield. The difference with the reaction conversion is due to the presence of residual phosphate salts from the buffer as confirmed by 31P NMR.
000, 50% aq. solution, branched) was purchased from Acros Organics. 2-(1H-Pyrazol-3-yl)ethanol and betazole were acquired from Fluorochem.
![[G with combining low line]](https://www.rsc.org/images/entities/char_0047_0332.gif)
![[C with combining low line]](https://www.rsc.org/images/entities/char_0043_0332.gif)
![[T with combining low line]](https://www.rsc.org/images/entities/char_0054_0332.gif)
![[A with combining low line]](https://www.rsc.org/images/entities/char_0041_0332.gif)
![[G with combining low line]](https://www.rsc.org/images/entities/char_0047_0332.gif)
TGGAGCCACCCGCAGTTCGAAAAAGGCGCCATGAGCACCGCAG-3′) and RV-HLADH (5′-TAGTTA![[G with combining low line]](https://www.rsc.org/images/entities/char_0047_0332.gif)
![[A with combining low line]](https://www.rsc.org/images/entities/char_0041_0332.gif)
![[T with combining low line]](https://www.rsc.org/images/entities/char_0054_0332.gif)
![[A with combining low line]](https://www.rsc.org/images/entities/char_0041_0332.gif)
![[T with combining low line]](https://www.rsc.org/images/entities/char_0054_0332.gif)
ATTAAAAGGTCAGAAT-3′). The primers were designed to incorporate NheI and EcoRV restriction sites, respectively (underlined). The manufacturer's protocol for GoTaq® DNA polymerase PCR in 50 μL reaction was used. PCR was carried out with 10 μL 5× GoTaq® buffer; 2.5 μL 10 mM FWD-HLADH primer; 2.5 μL 10 mM RV-HLADH primer; 32.5 μL nuclease-free water; 1 μL 100 ng μL−1 plasmid containing the HLADH gene; and 1 μL GoTaq® DNA polymerase. PCR conditions were 2 min of at 72 °C and 35 cycles of 1 min denaturation at 95 °C, 1 min annealing at 60 °C and 2 min extension 72 °C. The 35 cycles were followed by 7 min of final elongation at 72 °C. The amplified DNA was purified with a QIAquick PCR Purification Kit. The purified PCR product was digested with NheI-HF® and EcoRV-HF® for 2 h at 37 °C. The digested gene was gel-extracted using a QIAquick Gel Extraction Kit and cloned into the pASK-IBA5plus vector harbouring Lactobacillus brevis ADH, kindly donated by Prof. Kroutil, digested with the same restriction enzymes. Ligation was developed with T4 DNA ligase following the manufacturer's guidelines and the resulting product was used to transform E. coli XL10-Gold® Ultracompetent cells. The construct was verified by sequencing and the obtained plasmid was named pASK-IBA5plus-HLADH.
500 rpm for 60 min. The supernatant was filtered (0.45 μm) and loaded into a StrepTrap HP 5 mL column using an ÄKTA™ start FPLC (GE Healthcare). The protein was eluted with 20 mM sodium phosphate buffer, 0.28 M NaCl, 6 mM KCl and 2.5 mM D-desthiobiotin pH 7.4. HLADH was dialysed twice against 50 mM Tris buffer pH 8 at 4 °C. For HEWT, cells were resuspended in 50 mM potassium phosphate buffer, 0.1 M NaCl, 0.1 mM PLP and 30 mM imidazole pH 8. The same sonication and centrifugation conditions as in the case of HADH were used. The supernatant was filtered (0.45 μm) and HEWT was purified using the ÄKTA™ start FPLC as previously described.35 HEWT was dialysed twice against 50 mM potassium phosphate buffer 0.1 mM PLP pH 8 at 4 °C. Cells containing expressed LpNOX enzyme were resuspended in 50 mM potassium phosphate buffer, 0.5 M NaCl, 20 mM imidazole, 10% (v/v) glycerol, and 0.01 mM flavin adenine dinucleotide (FAD) pH 8. After cell disruption and insoluble fraction separation under the same conditions as before, the filtered supernatant was loaded into a HisTrap HP 1 mL column using the ÄKTA™ start FPLC. The protein was eluted with 50 mM potassium phosphate buffer, 0.5 M NaCl, 0.5 M imidazole, 10% (v/v) glycerol, and 0.01 mM FAD pH 8, and dialysed twice against 50 mM Tris buffer 0.01 mM FAD pH 7.5 at 4 °C.
For the immobilised enzyme, the activity was measured using the same conditions as for the free enzyme. For HLADH and HEWT, 20 mg of the immobilised biocatalyst were added to a 10 mL reaction mixture. At regular intervals of time during 10 min, a sample was taken to measure the absorbance at 340 nm or 245 nm, for HLADH or HEWT, respectively. For LpNOX, 20 mg of the immobilised biocatalyst were added to 40 mL of reaction mixture, and the same protocol was followed. The specific activity of HLADH, HEWT and LpNOX immobilised biocatalysts (U gcarrier−1) was defined as: the formation of NADH in μmol per minute and g of biocatalyst (HLADH); the formation of acetophenone in μmol per minute and g of biocatalyst (HEWT); and the depletion of NADH in μmol per minute and g of biocatalyst (LpNOX).
HLADH and HEWT protein concentration were determined by UV absorption at 280 nm using an Epoch Take3 Micro-Volume Plate. The extinction coefficients 22
460 M−1 cm−1 and 62
340 M−1 cm−1, at 280 nm, measured in water, were respectively estimated for HLADH and HEWT using the ExPASy ProtParam tool, accessible from the ExPASy website (http://www.expasy.ch). The LpNOX concentration was determined using a Bradford assay with bovine serum albumin (BSA) as the standard, as previously reported.29
The immobilised activity (U gcarrier−1) was calculated as the difference between initial offered activity per gram of carrier and remaining activity in the liquid phase per gram of carrier at the immobilisation end point. The immobilisation yield (%) was calculated as the ratio between immobilised activity and initial offered activity per gram of carrier. The recovered activity (%) was calculated as the ratio between the specific activity of the immobilised enzyme and the immobilised activity.
:
50 substrate solution/air segmented flow was directed into the PBR. Samples were taken from the exiting flow streams of single-phase and two-phase flow systems and analysed by HPLC, as described in the Analytical method section. For the two-phase flow reactions with recirculation, a 50 mM substrate solution 5 mM NAD+, 100 mM IPA, 0.1 mM PLP, 1 mM FAD in 50 mM potassium phosphate buffer pH 8.0; and a 30 mM substrate solution 3 mM NAD+, 60 mM IPA, 0.1 mM PLP, 1 mM FAD in 50 mM potassium phosphate buffer pH 8.0 were prepared. 9.151 mL of the 10 mM, the 30 mM or the 50 mM substrate solution were flowed at 122 μL min−1 to allow a residence time of 15 min. The exiting flow stream was placed in the same container as the substrate solution. Air was flowed into the PBR at the same flow rate. After one pass of the 9.151 mL substrate solution (after 75 min), a sample was taken and analysed by HPLC. The other samples were taken after 2 and 4 (after 2.5 h and 5 h, respectively).
Footnote |
| † Electronic supplementary information (ESI) available. See DOI: 10.1039/d1gc01095f |
| This journal is © The Royal Society of Chemistry 2021 |