Case study
For a technical application, an enzyme was required that could functionalise a substrate and operate at elevated process temperature. Based on an initial enzyme that was already able to catalyse the relevant reaction but lacked sufficient process stability, the MetXtra platform was used to identify new enzymes that exhibited improved performance at higher temperatures.
Discovery began with the exploration of the MetXtra database and design. Initially, a sequence space of more than 120,000 enzyme sequences was considered.
The discovery workflow combined several filters, including an AI tool for thermostability prediction and molecular dynamics simulations to validate application suitability, leading to 36 prioritised enzyme candidates for experimental evaluation. The genes for these were synthesised and expressed in a microbial production strain.
Detailed biochemical characterisation of the enzymes showed that 78 % of all candidates exhibited the desired activity, the best candidate demonstrated an improvement in performance of over 300 % at elevated temperatures under process-relevant conditions. The high hit rate regarding activity and the discovery of significantly improved performance in only 36 tested enzymes demonstrates the strength of this approach.
To ensure the success of discovery projects, it is important to know the requirements of the application in detail. Accordingly, the bioinformatics and AI tools used are prioritised and combined flexibly.
When discovering enzymes for applications such as pharmaceuticals, for example, we prioritise regio- and enantioselectivity to ensure high product purity. In these cases, highthroughput structure predictions via complementary AI and modelling approaches including substrate dockings are prioritised in MetXtra discover to identify the most promising enzymes.
Fine-tuning via enzyme engineering
If the identified enzymes require further refinement to suit a particular application, enzyme engineering offers a wide range of possibilities. We use advanced bioinformatics and AI for enzyme engineering through rational design. These in silico predictions help us to identify the positions(‘ hot spots’) in the enzyme and the appropriate amino acid substitutions that lead to enzyme improvements.
At the primary sequence level, analyses such as amino acid conservation and correlated mutation studies are performed. Structural elucidation includes investigating the active site, potential substrate access tunnels, the enzyme surface, and intra- and intermolecular contact regions. Simulations ranging from docking studies to molecular dynamics, along with AI-based predictions, provide additional insights that guide the engineering strategy.
Since both discovery and engineering rely on similar tools, we can use these technologies sequentially or simultaneously( Figure 2). This gives us tremendous flexibility and the ability to obtain the perfect enzyme for an application using both strategies.
Microbial enzyme production strains
Discovering and engineering enzymes is only the first step. For the application of enzymes to be feasible, they must be produced in sufficient quantity and quality. We focus on microbial production strains allowing high cell densities and high space-time yields.
We draw mostly on the bacteria E. Coli and Bacillus, the yeast Pichia( Komagataella), and the fungus Aspergillus to cover a broad diversity. Strains that can secrete enzymes into the fermentation medium are particularly popular, as this enables simple and cost-effective DSP.
Once new enzymes have been selected, the question of which production strain to choose for achieving high enzyme titres always arises. Our strategy is to evaluate
Figure 2- Complementary enzyme discovery & engineering approach
© BRAIN Biotech AG
40 SPECIALITY CHEMICALS MAGAZINE ESTABLISHED 1981