Artificial enzymes in place of protecting groups
Can artificial enzymes modify sensitive molecules selectively enough to make protecting groups unnecessary?
Summary
Why it is open
Protecting groups are not a minor convenience. Michael Schelhaas and Herbert Waldmann wrote in 1996 that the choice of protecting groups "is one of the decisive factors in the successful realization of a complex, demanding synthetic project", and that the protecting groups used "influence the length and efficiency of the synthesis and are often responsible for its success or failure" [1]. The cost is in steps: Reinhard Hoffmann stated in 2006 that the introduction and removal of protecting groups "are steps that unnecessarily increase the length of multistep syntheses" [3], and Lisa Kennedy and colleagues noted in 2024 that removal "often requires excess reagents, extensive time and can have low % yield" [14].
The question has no single bar, and three things are being asked at once. First, what counts as an artificial enzyme: an engineered natural enzyme [7], a protein carrying a synthetic metal cofactor [8], or a small-molecule catalyst with a recognition domain that mimics an enzyme [11]. Second, how selective is selective enough, since a replacement must distinguish one site among several of the same chemical kind in a real substrate [2, 10]. Third, whether the catalyst is available when needed: Alison Narayan and colleagues wrote in 2015 that the application of such biocatalytic tools in organic synthesis "remains under-utilized due to limitations in substrate scope and scalability" [7].
The problem can also be reduced without any new catalyst. Protecting-group-free routes exist for complex targets [5], so each demonstration removes protecting groups from one synthesis rather than from synthesis in general [3, 5]. On 16 October 2026 the named source still listed the question as open.
What is known
Protecting-group-free synthesis (2006 to 2009). Hoffmann argued that "efficient syntheses are those that avoid the use of protecting groups" and reviewed strategies for minimising them [3]. In 2007 Phil Baran, Thomas Maimone and Jeremy Richter reported preparative-scale, enantioselective total syntheses of members of the hapalindole, fischerindole, welwitindolinone and ambiguine families, "each constructed without the need for protecting groups", writing that "the use of such groups adds considerably to the cost and complexity of syntheses"; molecules that had previously taken twenty or more steps to make racemically in milligram amounts were obtained as single enantiomers in ten steps or fewer [5]. A complementary route is to oxidise late rather than protect early. Mark Chen and Christina White reported an iron catalyst using hydrogen peroxide whose "predictable selectivity is achieved solely on the basis of the electronic and steric properties of the C-H bonds, without the need for directing groups", and showed oxidation of complex natural products at specific carbon-hydrogen bonds [4]. Ke Chen and Baran then synthesised five oxidised eudesmane terpenes by carbocycle formation followed by site-selective carbon-hydrogen oxidations, modelled on the cyclase and oxidase phases of terpene biosynthesis, and wrote that "much remains to be learned if the oxidase phase is to be mimicked in the laboratory" [6].
Site-selective catalysts that mimic enzymes (2006 to 2018). In 2006 Chad Lewis and Scott Miller showed that simple peptide-based nucleophilic catalysts "can perturb the inherent reactivity hierarchy of the polyol natural product erythromycin A", that is, they derivatised a hydroxyl group that reagents without a catalyst do not favour [2]. Christopher Shugrue and Miller reviewed this field in 2017 as late-stage functionalisation of natural product scaffolds by non-enzymatic catalysts, and concluded that "challenges for the design of very broad classes of catalysts, with even broader selectivity demands (e.g., stereoselectivity, functional group selectivity, and site-selectivity) persist" [10]. Moira Flanagan, Dmitry Gorin and colleagues tethered an imidazole catalyst to a DNA aptamer and increased the rate of hydrolysis of a target ester "by >100-fold compared with equimolar untethered imidazole", with other esters unaffected and enzyme-like kinetics [11].
Engineered and artificial metalloenzymes (2015 to 2016). Narayan, Gonzalo Jiménez-Osés, David Sherman, Kendall Houk and colleagues modified the monooxygenase PikC by computationally guided protein and substrate engineering and applied it to the oxidation of unactivated methylene carbon-hydrogen bonds, with molecular dynamics and quantum calculations used to predict substrate scope, site selectivity and stereoselectivity [7]. Hanna Key, Paweł Dydio, Douglas Clark and John Hartwig replaced the iron of haem proteins with noble metals, and an iridium-methyl myoglobin catalysed carbon-hydrogen functionalisation by carbene insertion; directed evolution produced mutants giving either enantiomer [8]. Dydio and colleagues then reconstituted the P450 enzyme CYP119 with an iridium porphyrin and reported up to 98% enantiomeric excess, 35,000 turnovers and a turnover frequency of 2,550 per hour, noting that the artificial metalloenzymes developed until then had activities "much lower than those of natural enzymes" [9].
Enzymes already replacing protection in practice (2018 to 2026). In carbohydrate chemistry, where protecting-group patterns dominate chemical routes, Liuqing Wen and colleagues stated in 2018 that "enzymatic glycosylation can proceed stereo- and regiospecifically without protecting group manipulations", while adding that automated enzymatic synthesis "is not as well developed" as the chemical kind [12]. Mark Huffman and colleagues built a three-step, nine-enzyme in vitro cascade for the HIV drug candidate islatravir, with five enzymes engineered by directed evolution to act on non-natural substrates, requiring "fewer than half the number of steps of the previously reported routes" [13]. Kennedy and colleagues took the opposite approach and used enzymes to remove protecting groups, combining a Bacillus esterase and a Sphingomonas amidohydrolase in a one-pot, two-step deprotection, and observed that reported deprotectase biocatalysts "are not commonly used in mainstream synthetic routes" [14]. In March 2026 Shixuan Cao, Yuxuan Ye and colleagues engineered ene-reductases for the direct desaturation of cyclic ketones and reported late-stage dehydrogenation of terpenoids "with complementary site selectivity to existing methods" [15].
Main approaches
Redesign the route instead. Invent reactions and sequences in which no group needs masking [3]. Strongest evidence: the protecting-group-free syntheses of 2007 and the step counts they report [5]. Main difficulty: each case needs its own invention, and the oxidation stage of such routes is the part still least understood [6].
Site-selective small-molecule catalysts and enzyme mimics. Add a recognition element to a simple catalyst so that it picks one site [2, 11]. Strongest evidence: the catalyst-controlled derivatisation of erythromycin A [2] and the DNA-tethered imidazole that accelerates one ester and leaves others unaffected [11]. Main difficulty: no broad classes of such catalysts have been designed, and the combined selectivity demands remain open [10].
Engineered natural enzymes. Start from an enzyme and engineer it, with computation or directed evolution, for the substrate and the site wanted [7, 15]. Strongest evidence: the predictive model for site and stereoselectivity in PikC oxidations [7], the site-divergent ene-reductase desaturations of 2026 [15] and the islatravir cascade in manufacture [13]. Main difficulty: substrate scope and scalability limit the use of these tools [7], and five of the nine enzymes in the islatravir cascade had to be engineered by directed evolution before they would act on the non-natural substrates [13].
Artificial metalloenzymes. Put a synthetic metal cofactor into a protein scaffold, then evolve the scaffold [8]. Strongest evidence: iridium in place of iron gave carbene insertions into unactivated carbon-hydrogen bonds with native-like kinetic parameters [9], and either enantiomer from evolved myoglobin mutants [8]. Main difficulty: artificial metalloenzymes were generally much less active than natural enzymes before that result [9], and in scaffolds where the metal complex occupies the substrate binding site the substrate interactions differ from those of the native protein [8].
Enzymatic deprotection. Keep the protecting groups but remove them selectively with enzymes under mild conditions [14]. Strongest evidence: the one-pot, two-step deprotection of a doubly protected amino acid [14]. Main difficulty: this shortens the deprotection step rather than removing protection, and such biocatalysts are not in mainstream use [14].
Open sub-questions
- One site among equals. For a substrate with several hydroxyl groups of similar reactivity, which recognition element added to a simple catalyst predicts which site is derivatised, as the peptide catalysts did for erythromycin A [2, 10]?
- From mimic to useful scale. Does a catalyst with a nucleic acid or peptide recognition domain keep its site selectivity at preparative concentrations, where the >100-fold rate difference was measured on a small scale [11]?
- Activity of artificial metalloenzymes. Can an artificial metalloenzyme reach the turnover frequency reported for iridium CYP119 on a polyfunctional substrate that would otherwise need protection [9]?
- Steps actually saved. For a published route that protects and deprotects a group, how many steps fall away when one enzymatic step replaces the masked one, measured as in the islatravir comparison [13]?
- Orthogonal deprotectases. Which pairs of protecting groups can be removed in one pot by two enzymes without either touching the other's substrate [14]?
- Where the catalysts come from. Can a catalyst for a selective step on a given substrate be designed rather than found, which is the open question on the Papex page De novo enzyme design [7, 8]?
- Automated enzymatic glycosylation. What prevents the enzymatic glycosylation that needs no protecting groups from being automated as chemical glycosylation has been [12]?
Work on this in Papex
If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.
Key references
- Schelhaas, M. & Waldmann, H. (1996). Protecting group strategies in organic synthesis. Angewandte Chemie International Edition in English 35(18), 2056–2083. 10.1002/anie.199620561 · Checked 16 Oct 2026
- Lewis, C. A. & Miller, S. J. (2006). Site-selective derivatization and remodeling of erythromycin A by using simple peptide-based chiral catalysts. Angewandte Chemie International Edition 45(34), 5616–5619. 10.1002/anie.200601490 · Checked 16 Oct 2026
- Hoffmann, R. (2006). Protecting-group-free synthesis. Synthesis 2006(21), 3531–3541. 10.1055/s-2006-950311 · Checked 16 Oct 2026
- Chen, M. S. & White, M. C. (2007). A predictably selective aliphatic C–H oxidation reaction for complex molecule synthesis. Science 318(5851), 783–787. 10.1126/science.1148597 · Checked 16 Oct 2026
- Baran, P. S., Maimone, T. J. & Richter, J. M. (2007). Total synthesis of marine natural products without using protecting groups. Nature 446(7134), 404–408. 10.1038/nature05569 · Checked 16 Oct 2026
- Chen, K. & Baran, P. S. (2009). Total synthesis of eudesmane terpenes by site-selective C–H oxidations. Nature 459(7248), 824–828. 10.1038/nature08043 · Checked 16 Oct 2026
- Narayan, A. R. H., Jiménez-Osés, G., Liu, P., Negretti, S., Zhao, W., Gilbert, M. M. et al. (2015). Enzymatic hydroxylation of an unactivated methylene C–H bond guided by molecular dynamics simulations. Nature Chemistry 7(8), 653–660. 10.1038/nchem.2285 · Checked 16 Oct 2026
- Key, H. M., Dydio, P., Clark, D. S. & Hartwig, J. F. (2016). Abiological catalysis by artificial haem proteins containing noble metals in place of iron. Nature 534(7608), 534–537. 10.1038/nature17968 · Checked 16 Oct 2026
- Dydio, P., Key, H. M., Nazarenko, A., Rha, J. Y.-E., Seyedkazemi, V., Clark, D. S. & Hartwig, J. F. (2016). An artificial metalloenzyme with the kinetics of native enzymes. Science 354(6308), 102–106. 10.1126/science.aah4427 · Checked 16 Oct 2026
- Shugrue, C. R. & Miller, S. J. (2017). Applications of nonenzymatic catalysts to the alteration of natural products. Chemical Reviews 117(18), 11894–11951. 10.1021/acs.chemrev.7b00022 · Checked 16 Oct 2026
- Flanagan, M. L., Arguello, A. E., Colman, D. E., Kim, J., Krejci, J. N., Liu, S., Yao, Y., Zhang, Y. & Gorin, D. J. (2018). A DNA-conjugated small molecule catalyst enzyme mimic for site-selective ester hydrolysis. Chemical Science 9(8), 2105–2112. 10.1039/c7sc04554a · Checked 16 Oct 2026
- Wen, L., Edmunds, G., Gibbons, C., Zhang, J., Gadi, M. R., Zhu, H. et al. (2018). Toward automated enzymatic synthesis of oligosaccharides. Chemical Reviews 118(17), 8151–8187. 10.1021/acs.chemrev.8b00066 · Checked 16 Oct 2026
- Huffman, M. A., Fryszkowska, A., Alvizo, O., Borra-Garske, M., Campos, K. R., Canada, K. A. et al. (2019). Design of an in vitro biocatalytic cascade for the manufacture of islatravir. Science 366(6470), 1255–1259. 10.1126/science.aay8484 · Checked 16 Oct 2026
- Kennedy, L., Sajjad, M., Herrera, M. A., Szieber, P., Rybacka, N., Zhao, Y. et al. (2024). Developing deprotectase biocatalysts for synthesis. Faraday Discussions 252, 174–187. 10.1039/d4fd00016a · Checked 16 Oct 2026
- Cao, S., Zhu, Y., Lei, J., Dai, R., Zhu, T. & Ye, Y. (2026). Biocatalytic site- and stereoselective carbonyl desaturation for late-stage functionalization of cyclic ketones. Nature Chemistry 18(4), 731–740 (published online 13 March 2026). 10.1038/s41557-026-02086-w · Checked 16 Oct 2026
Source
Open, as listed by Wikipedia · Unsolved problems in chemistry. Checked 08 Oct 2026.
The status is the source’s. Papex does not decide whether a problem is solved.
Records on this problem
No public record addresses this problem yet.