Presented by Jacob Kofoed, Principal Scientist at Novo Nordisk A/S
Transcribed by Tia Byer
Novo Nordisk conduct pioneering scientific breakthroughs to expand access to medicines and ultimately cure serious chronic diseases such as diabetes, obesity, and rare blood disorders. Among its core expertise is protein and peptide technology, including expression, synthesis, engineering, and formulation. Global Research Technologies at Novo Nordisk A/S aims to automate the research and development of these treatment modalities. As part of the Peptide Congress held in April 2021, Novo Nordisk’s Global Research Technologies presented its automation platform for high-throughput protein and peptide purification.
The Platform: AI-Driven Characterization Technology
Novo Nordisk is constructing a 3200 square foot automation laboratory of 8 fully automated cells for recombinant production, chemical modification, as well as sample storage and logistics of peptides and proteins. It also provides functional analysis and biophysical characterization. The setup is integrated with automated peptide synthesis and purification, as well as LC and LC/MS analysis. The platform aims to integrate the set-up with machine learning capabilities and uses the generated data for building Structure Activity-Relationship (SAR) models, which will ultimately aid in the discovery of better medicines.
Opportunities for Drug Discovery in Untapped Chemical Space
Novo Nordisk has found that in using peptide arrays, that one can construct a ligand fingerprint to understand the important interactions between the receptor and its peptide ligand. This is valuable information for generating Structure-Activity Relationship models. The platform also promises the opportunity for therapeutic discovery in the untapped chemical interface space between peptide and proteins in the 60 to 100 amino acid residue range.
To access this untapped chemical space, there are various synthetic opportunities available. One such opportunity is chemical ligation in which thioester condensed with cysteine-containing peptides form a thioester-linked intermediate which rearranges into a native peptide structure. Another technique is to use the synthesis of long peptides, boosting them into the molecular range of 60 to 100 amino acid residues accomplished by using excess reagents, prolonged coupling time, and increased reaction temperature.
The Challenge of Longer Synthetic Peptides
Using automated peptide synthesizers, scientists produce peptides in 96-deep-well plates (96DWPs). However, synthesizing longer peptides is not without challenges. As seen in Figure 1, longer synthetic peptides suffers from decreasing purity with increasing chain length.
Figure 1. The challenge with longer synthetic peptides as evidenced by RP-UPLC analysis (214 nm) of crude synthetic peptides of various lengths showing broadening of peaks due to deletions, isomerization and deprotection biproducts for the longer sequences.
High-Throughput Peptide Purification Process: Hardware Set Up and Plate Flow
Hence, longer synthetic peptides needs to be purified and Novo Nordisk’s novel automation platform uses MS-coupled RP-HPLC purification strategies to purify libraries of synthetic peptides in 96DWP format. Scientific data management software is used to register the structure of each peptide which in turn provides metadata about the molecular weight and spatial position on the plates. Incoming crude peptide plates and outgoing fraction plates then undergo barcode scanning to streamline the plate management process.
Using the chromatography hardware SDK the raw data is then converted into a textual representation, and read into a Python script which uses AI to perform automated selection of the fractions of interest for each peptide. The Python script compiles all the collected data for each incoming 96DWP and its associated outgoing fractionation plates into data package. The data package is then fed into a browser where scientists can visualize and interact with the data. The AI component also allows the creation of pick lists for a liquid handler which further reorganises purified peptides into new barcoded 96DWPs. Scientists can then concentrate these solutions using a vacuum concentrator to produce pure, lyophilized peptide substances ready for further processing.
The Verdict on High-Throughput Peptide Purification: Does it matter?
Peptide purification is important. Upon conducting a purification of hydrophobic 50-mer peptides, Novo Nordisk found that their high-throughput platform was successful in obtaining sufficiently pure peptides. Peptides up with up to 80 residues have so far been produced synthetically and purified using this novel automation platform. It was consistently found that peptides purified via the automated process gave higher potencies as compared to their crude counterparts, for example in reporter gene assays.
Novo Nordisk concluded that automated peptide purification via AI improves data quality. The key to this success was using plate barcodes streamlining the plate management process. Supporting IT infrastructures was also paramount for the structural registration of compounds and associated data. Furthermore, using vendor raw data readers access to fraction results data at the scan level was possible, which allowed for rapid fraction selection using AI, and thus without any human intervention.
Dr. Jacob Kofoed received his PhD in bioorganic chemistry from the University of Berne working in the field of synthetic peptide dendrimers. Kofoed joined Novo Nordisk A/S in 2007 as a research scientist before being appointed head of various Protein and Peptide Chemistry departments. Key research activities include therapeutic peptide half-life extension, multivalent protein-peptide conjugates, orally bioavailable peptides, chemistry automation, and artificial intelligence within drug discovery.