Towards better biocatalysts: advances in high throughput screening and data-driven design of immobilized enzymes

14

Introduction

 

Biocatalysis has been for the last couple of decades, a focus area for the development of new synthetic strategies. The excellent selectivity that enzymes exhibit makes them valuable catalysts for the synthesis of key chiral intermediates, complex molecules and in late-stage functionalization (1, 2).

 

Biocatalysis, however, apart from for selected productions, has not experienced the broad applicability that could be expected. The two main reasons behind this are the intrinsic (in)stability and (in)compatibility of enzymes under industrial conditions and the long development times and costs associated with the development of tailored solutions (3). For industrial integration, biocatalysis must necessarily co-exist with organic chemistry in a synthetic route, and the two should be easy to implement into the same multi-step synthesis. However, these two technologies tend to operate in very different conditions – for biocatalysis, typically, the main solvent is water, and the reaction conditions are mild (20-40°C, atmospheric pressure). Organic synthesis, on the other hand, is typically performed in organic solvents, and harsher conditions (temperature, pressure) are the norm. Thus, we must find a way to make the two compatible (4–6).

 

One common way to enhance the productivity and stability of enzymes is enzyme engineering. Novel tools and key advances have been made in the last decades in this field, incorporating the latest developments in data-driven design and AI in the planning of the necessary modifications. As a result, several success stories have been published and applied in the last few years (7–10). Nonetheless, there are certain properties that cannot be engineered.
There are successful examples of engineering higher activity, selectivity or enhanced thermostability, but an easier downstream processing or the capability to reuse the biocatalyst cannot be effectively engineered.

 

Supported enzyme immobilization consists in linking the enzyme to a larger particle, which can be of various different materials, creating a heterogeneous preparation that retains the advantages of enzymatic catalysis but adds new beneficial properties: immobilized enzymes can be easily removed from the reaction mixture by simple filtration, allowing not only easier downstream processing and product purification, but also enabling the reuse of the same biocatalyst in multiple reaction cycles. These two characteristics are key for the efficient integration of biocatalysts in production (11–13).

 

The first description of a protocol for enzyme immobilization dates back to the start of the 20th century (14), and since then the field has advanced tremendously. Novel materials and chemistries to bind the enzymes have been developed, and there has been a keen interest to apply this technology to the broad spectrum of enzyme types. What is surprising though, is that the development of novel immobilized enzymes has changed only slightly – immobilization screening remains mostly tied to a trial-and-error approach, where little prior knowledge is used in the planning.

 

In this opinion article, we aim to review the latest advances in the development of data-driven approaches for enzyme immobilization: novel screening technologies, as well as novel bioinformatic tools that have the potential to improve and speed up the development of better, more compatible biocatalysts for chemical production.

 

Screening Technologies: Integration of High-Throughput Screening (HTS) to Enzyme Immobilization

 

The immobilization of an enzyme is at its core, a combinatorial problem. Each protein exhibits different characteristics – determined by the amino acid composition and it’s inherent physicochemical characteristics– and therefore, it is necessary to find the right combination of material and binding chemistry to ensure the maximum activity and stability. Due to the multiple options available, development times are often long, and tedious experimentation is needed. In other fields that suffer from the same problem, high-throughput screening has indeed vastly relieved these issues. It wasn’t until recently, however, that examples of HTS for immobilization have been described.

 

Schober et al. (15), at Novartis recently exemplified the advances that are possible combining a liquid handling robot with downstream analytical systems. Their workflow allowed for enzyme immobilization screening with multiple resins and enzymes in parallel at μL-scale. Interestingly, the same setup can also be used to optimize the reaction after immobilization selection, which leads to significant cost and time reductions and exemplifies the flexibility of the proposed workflow.

 

Similar setups, taking advantage of microtiter plates (MTP) were published in the same year. Sánchez-Morán et al. (16), described what they called a combinatorial high-throughput enzyme support screening (CHESS). Their technology takes advantage of 384-well plates in which different polymer brushes are synthesized in situ in each of the microwells.
This allows for extensive and fast testing of the efficiency of different binding chemistries, but also characterization of the activity and stability of the resulting immobilized enzyme in the same setup. This approach was used for the immobilization of a lipase from Bacillus subtilis (Lip A), and the results could be translated from the MTP to beads modified with the same polymeric structures with very close agreement. Going further than single enzyme immobilization, López et al (17).
leveraged MTPs to test multiple supports and chemistries to assemble an enzymatic cascade. Cascade immobilization requires even lengthier experiments and testing, to find the shared conditions in which none of the enzymes present are negatively affected by immobilization. In their work, they successfully assembled a 4-enzyme cascade, which showed a 70% higher production than the same system in solution.

 

These examples, illustrate how HTS methods are already transforming the development of immobilized enzymes. Faster and more efficient optimization of immobilization protocols will undoubtedly ease its adoption, and as discussed in some of the previously mentioned works, this also opens the door to the next generation of immobilization protocols – data-driven immobilization design that can combine in silico planning with in vitro validation.

 

Bioinformatic Tools for Enzyme Immobilization

 

Leveraging data and advanced bioinformatic simulations has become the norm for several biotechnological fields. From microbial fermentation to protein design, the advances in various fields are fuelled by the application of machine learning to extract from curated data the key properties and to design and develop better performing processes. But in enzyme immobilization, the progress has not kept up. Only a few examples of bioinformatic analysis of immobilized enzymes are available, and typically, these focus on the analysis of the free enzyme through surface analysis or molecular dynamics simulations to understand it’s properties.

 

One of the first examples, is the one reported by Hudson et al. 20 years ago (18). In their work, they applied a methodical approach were prior in silico analysis of the two tested proteins (Cytochrome C and xylanase) informed their immobilization screening. Similar works have been reported since, in which the amino acid composition of the protein and its surface has been considered in the immobilization planning (19, 20) but the lack of a unified protocol and the gap between bioinformatics and experimental immobilization protocols remain open.

 

To bridge this gap, 4 years ago we started developing bioinformatic tools tailored to the design and planning of immobilization protocols. CapiPy (Computer assisted protein immobilization using Python) is an easy-to-use software, that performs a series of protein-centric analysis. Through surface analysis of the protein, CapiPy allows experienced researchers in the field of immobilization to make an informed decision on the material and chemistries to use (21). Since its publication in 2020, CapiPy has been used in several immobilization development studies, (22–24) and to further simplify its adoption it has since been made publicly available through a website to facilitate the access to academics and industry professionals (25). But CapiPy only performs basic analysis and has no predictive power. Moreover, focusing on the protein only gives information about one of the three components of the immobilization – the enzyme – and it is well known that the material and the binding chemistry greatly affects the immobilization outcome.

 

At inSEIT we have taken this fundamental approach one step further; as well as developing novel tools for in silico immobilization planning, we have now integrated an AI-driven platform for the in silico screening of immobilized enzymes. Informed by data, our bioinformatic tools allow a faster and easier selection of the right combination of material and binding chemistry for any given enzyme, ensuring the immobilization strategy is tailored to the specific enzyme and reaction to ensure the highest activity, stability, and reusability. This data-driven approach also allows a much faster development. In this workflow, we not only study the protein, but to also analyse how the material and the binding chemistry can affect the final immobilized biocatalyst. The combination of these three pillars, allows to not only explain but also make informed decisions in the immobilization design, reducing the time and material use (Figure 2).

 

Starting from the structure of the protein to be immobilized, we perform both simulations in the presence of different materials and reactivity analysis. The data is collected and passed through the second stage – an AI driven platform that pre-screens over 40 different materials and >10 binding chemistries and allows the selection of those with the highest probability to work. The results are expressed in a heatmap as seen in Figure 2, and guided by these results we perform the testing of the selected immobilizations in MTP analogues that allow for a fast, reliable and scalable testing of the selected immobilization conditions.

 

To exemplify this, we recently revisited the immobilization of Bacillus megaterium glucose dehydrogenase (BmGDH). This enzyme can be used in the production of gluconic acid, a key platform chemical from glucose obtained in food waste, but it is also one of the most common enzymes used for the NADPH cofactor recycling in redox reactions.

 

Looking at the previously literature, BmGDH was immobilized in silica and agarose supports, with chemistries targeting both the negatively charged groups (DEAE – diethylaminoethyl) and exposed lysines (aldehyde and glutaraldehyde). Nonetheless, in these conditions the reported expressed activity of the heterogeneous catalyst was less than 3 U/g (26, 27).
Interestingly, when mapping the reactivity upon the surface of the enzyme with the previously used chemistries, the most reactive residues were in the active site of the enzyme. With DEAE binding, the reactive residues are distributed across the whole surface but as stated in the publication, these preparations suffer from low reusability as the binding only relies on ionic bonds. When using aldehyde chemistries for binding (both glutaraldehyde and agarose-aldehyde) either low stability of the enzyme in immobilization conditions or very low recovered activity was observed. Our goal was set to move the reactivity from the active site towards the surface of the enzyme, ensuring the maintenance of the quaternary structure and increasing the activity of the immobilized biocatalyst.

 

Based on the literature data and the initial results of our in silico screening, we selected a combination of aldehyde and amino groups in the surface of the material to bind BmGDH. Upon testing of different materials, we found that methacrylate resins exhibited superior activity, with up to 7 times more activity than in previously reported immobilization strategies. Moreover, after optimization of the immobilization conditions, ensuring not only the tetrameric structure preservation but the immobilization of the active form of the enzyme, we achieved up to 10 times more activity than literature reports (28). This higher activity was not in exchange of its stability, as the novel immobilized BmGDH could be stored for up to 2 weeks at room temperature with minimal loss of activity and exhibited excellent stability upon incubation at 50°C for 30 minutes, compared to the free enzyme (Figure 3).

 

The immobilization of BmGDH is the perfect example of how, in silico planning coupled with in vitro testing of the most promising candidates, allows for the optimization and tailoring of immobilization conditions for each individual enzyme.

 

Conclusions

 

Enzyme immobilization is at a turning point. Traditional methods, hampered by trial-and-error and high costs, will inevitably be substituted with data-driven approaches, powered by innovative use of AI technologies and leveraging the advances of molecular simulations. These innovative technologies will ensure we can predict optimal immobilization conditions, cut development time and enabling tailored biocatalysts for industries from pharmaceuticals to environmental clean-up. Paired with high-throughput screening and the application of robotics, immobilization enzymes have the potential to swiftly make processes faster, cheaper, and greener. In the next decade, enzyme immobilization will become a cornerstone of industrial innovation, driving economic and environmental gains and we at inSEIT believe that, pushing further immobilization will unlock the potential of biocatalysis in the chemical industry.

 

 

Figure 1. Schematic representation of the selected examples of HTS techniques applied to enzyme immobilization.

 

Figure 2. Schematic representation of the bioinformatic workflows available for the selection and development of tailored immobilization protocols.

 

Figure 3. a) Immobilization of BmGDH in different resins and with different chemistries. The figure in the left, shows the predicted reactivity when using the Agarose / Aldehyde combination, while the one on the right indicates the one for HFA403/S / Amino-aldehyde. b) Measured stability of the BmGDH immobilized after incubation at 50°C for 30 mins compared to the free enzyme. c) Storage stability of BmGDH in the immobilized form. Samples of dried resin were stored at the designated temperatures and individual samples of 20 mg were tested at the desired times.

 

References and notes

 

1. Alcántara AR, Domínguez de María P, Littlechild JA, Schürmann M, Sheldon RA, Wohlgemuth R. Biocatalysis as Key to Sustainable Industrial Chemistry. ChemSusChem. 2022;15(9):e202102709.
2. Winkler CK, Schrittwieser JH, Kroutil W. Power of Biocatalysis for Organic Synthesis. ACS Cent Sci. 2021 Jan;7(1):55–71
3. Gallou F, Gröger H, Lipshutz BH. Status check: biocatalysis; its use with and without chemocatalysis. How does the fine chemicals industry view this area? Green Chem. 2023 Aug 14;25(16):6092–107.
4. Chakrabarty S, Romero EO, Pyser JB, Yazarians JA, Narayan ARH. Chemoenzymatic Total Synthesis of Natural Products. Acc Chem Res. 2021;
5. Martinez CA, Hu S, Dumond Y, Tao J, Kelleher P, Tully L. Development of a chemoenzymatic manufacturing process for Pregabalin. Org Process Res Dev. 2008 May;12(3):392–8.
6. Losada-Garcia N, Cabrera Z, Urrutia P, Garcia-Sanz C, Andreu A, Palomo JM. Recent Advances in Enzymatic and Chemoenzymatic Cascade Processes. Catalysts. 2020 Nov;10(11):1258.
7. U. Bozkurt E, C. Ørsted E, C. Volke D, I. Nikel P. Accelerating enzyme discovery and engineering with high-throughput screening. Nat Prod Rep [Internet]. 2025 [cited 2025 May 8]; Available from: https://pubs.rsc.org/en/content/articlelanding/2025/np/d4np00031e
8. Orsi E, Schada von Borzyskowski L, Noack S, Nikel PI, Lindner SN. Automated in vivo enzyme engineering accelerates biocatalyst optimization. Nat Commun. 2024 Apr 24;15(1):3447.
9. Yang YY, Jing SL, Shao JL, Chen JX, Zhang WF, Wan SY, et al. Purification and immobilization of β-glucosidase using surface modified mesoporous silica Santa Barbara Amorphous 15 for eco-friendly preparation of sagittatoside A. Nat Prod Bioprospecting. 2024 Aug 23;14(1):50.
10. Buller R, Lutz S, Kazlauskas RJ, Snajdrova R, Moore JC, Bornscheuer UT. From nature to industry: Harnessing enzymes for biocatalysis. Science. 2023 Nov 24;382(6673):eadh8615.
11. Maghraby YR, El-Shabasy RM, Ibrahim AH, Azzazy HMES. Enzyme Immobilization Technologies and Industrial Applications. ACS Omega. 2023 Feb 14;8(6):5184–96.
12. Papamichael EM, Stergiou PY. Enzyme immobilization strategies and bioprocessing applications. In: Biomass, Biofuels, Biochemicals [Internet]. Elsevier; 2020 [cited 2024 Oct 24]. p. 217–41. Available from: https://linkinghub.elsevier.com/retrieve/pii/B9780128198209000120
13. Robescu MS, Bavaro T. A Comprehensive Guide to Enzyme Immobilization: All You Need to Know. Molecules. 2025 Jan;30(4):939.
14. Nelson JM, Griffin EG. ADSORPTION OF INVERTASE. J Am Chem Soc. 1916 May 1;38(5):1109–15.
15. Schober L, Dreier P, Peschke T. Automated high throughput workflow for rapid implementation of immobilized enzymes in chemical process development. React Chem Eng. 2024;10.1039.D3RE00704A.
16. Sánchez-Morán H, Kaar JL, Schwartz DK. Combinatorial High-Throughput Screening of Complex Polymeric Enzyme Immobilization Supports. J Am Chem Soc. 2024 Apr 3;146(13):9112–23.
17. López IL, Sánchez-Costa M, Orrego AH, Zeballos N, Roura Padrosa D, López-Gallego F. Microtiter Plate Immobilization Screening for Prototyping Heterogeneous Enzyme Cascades. Angew Chem. 2024;136(35):e202407411.
18. Hudson S, Magner E, Cooney J, Hodnett BK. Methodology for the Immobilization of Enzymes onto Mesoporous Materials. J Phys Chem B. 2005 Oct 1;109(41):19496–506.
19. Basso A, Braiuca P, Cantone S, Ebert C, Linda P, Spizzo P, et al. In Silico Analysis of Enzyme Surface and Glycosylation Effect as a Tool for Efficient Covalent Immobilisation of CalB and PGA on Sepabeads®. Adv Synth Catal. 2007;349(6):877–86.
20. Weber E, Sirim D, Schreiber T, Thomas B, Pleiss J, Hunger M, et al. Immobilization of P450 BM-3 monooxygenase on mesoporous molecular sieves with different pore diameters. J Mol Catal B Enzym. 2010 Jun;64(1–2):29–37.
21. Roura Padrosa D, Marchini V, Paradisi F. CapiPy: python-based GUI-application to assist in protein immobilization. Valencia A, editor. Bioinformatics. 2021 Sep 9;37(17):2761–2.
22. Benítez-Mateos AI, Paradisi F. Sustainable Flow-Synthesis of (Bulky) Nucleoside Drugs by a Novel and Highly Stable Nucleoside Phosphorylase Immobilized on Reusable Supports. ChemSusChem. 2022 Jan;15(1).
23. Broumidis E, Paradisi F. Engineering a Dual-Functionalized PolyHIPE Resin for Photobiocatalytic Flow Chemistry. Angew Chem Int Ed. 2024;63(21):e202401912.
24. Roura Padrosa D, Wetzl D, Hildbrand S, Tosatti P, Sedelmeier J, Puentener K, et al. Biocatalyst Immobilization and Process Intensification: Production of (S)-Piperazine-2-carboxylic Acid Dihydrochloride. Org Process Res Dev. 2024 May 17;28(5):1713–24.
25. CapiPy by inSEIT | Protein Surface Analysis [Internet]. [cited 2025 May 15]. Available from: https://capipy.inseit.ch
26. Baron M, Fontana JD, Guimarães MF, Woodward J. Stabilization and reutilization of Bacillus megaterium glucose dehydrogenase by immobilization. Appl Biochem Biotechnol. 1997 Mar 1;63(1):257–68.
27. Dall’Oglio F, Contente ML, Conti P, Molinari F, Monfredi D, Pinto A, et al. Flow-based stereoselective reduction of ketones using an immobilized ketoreductase/glucose dehydrogenase mixed bed system. Catal Commun. 2017 Apr;93:29–32.
28. Benítez-Mateos AI, Lim D, Roura Padrosa D, Marchini V, Wu H, Buono F, et al. Biocatalytic Reduction of Heterocyclic Imines in Continuous Flow with Immobilized Enzymes. ACS Sustain Chem Eng. 2025 Apr 7;13(13):5009–18.