Development of a Novel L-Asparaginase with Enhanced Properties Using Ancestral Sequence Reconstruction

The overall survival of children with acute lymphoblastic leukemia (ALL) now exceeds 90% due to the precise disease risk stratification and intensification of multi-agent chemotherapy regimens. L-asparaginase (L-ASNase) is an indispensable biotherapeutic enzyme used in treatment of pediatric ALL. Discontinuation of L-ASNase treatment due to immunological or adverse reactions leads to significantly inferior disease-free survival (73% ±7% vs 92% ± 2%; P <0.01, Gupta et al. JCO 2020). Clinical L-ASNases are bacterial in origin, derived from either Escherichia coli or Erwinia chrysanthemi and are therefore highly immunogenic. Additionally, emerging data suggests that the concurrent glutaminase activity seen in bacterial asparaginases exacerbates its liver and pancreatic toxicity, especially in adults thereby limiting its extensive use beyond children. Therefore, finding alternative sources with preferable biochemical features remains paramount.

The inferior catalytic properties of human L-ASNase restricts its functionality in ALL treatment. However, characterization has shown that guinea pig (GP) L-ASNase demonstrates significant anti-tumor properties and improved enzyme kinetics despite a 69.8% amino acid sequence identity with human L-ASNase, as compared to 30% for bacterial L-ASNases. Additionally, guinea pig L-ASNase has no glutaminase activity. Thus, to overcome the major deficiencies of bacterial L-ASNase and develop a more humanized form of L-ASNase we performed ancestral sequence reconstruction (ASR). ASR is a protein drug discovery / optimization platform that predicts ancient DNA and protein sequences from extant sequences. This approach permits higher-resolution mapping potentially highlighting functional resides. The objective therefore is to exploit functional diversity refined by natural selection using ASR to identify therapeutic variants with superior properties. Recent work on guinea pig (GP) L-ASNase (Schalk et al. J Biol Chem 2014) identified a low micromolar (μM) Michaelis constant (Km) at 57.7 ± 6.4 μM essential for hydrolyzing the extra-cellular pool of asparagine ~50 μM in human blood; thus, equipping us with the ideal mammalian L-ASNase ortholog to serve as a template to resurrect ancient ASNase sequences.

Ancestral L-ASNase sequences were generated utilizing 54 extant L- asparaginase sequences, aligned using MUSCLE and an evolutionary tree inferred using MrBayes. A total of 53 ancestral nodes were identified, and the ten ancestral L-ASNase variants spanning the ancient primate and guinea pig (GP) lineage were resurrected. The sequence identity of these ancestral L-ASNase variants ranged from 81 to 98% when compared to human L-ASNase. E. coli codon optimized complementary DNA (cDNA) sequences were synthesized and subcloned intoan expression vector. Correctly cloned plasmids were transformed into E. coli BL21 (DE3) cells for protein expression. An-ASNase candidates were isolated by Ni ²⁺ affinity chromatography and then further purified by size exclusion chromatography. Asparaginase activity of the An-ASNase candidates were determined via a modified Nessler's reagent assay by using a continuous spectroscopic enzyme-coupled assay. Three of the ancestral L-ASNase candidates, An-88, An-104 and An-107 demonstrated excellent asparaginase activity when tested at an enzyme concentration of 0.1 mg/mL and an excess substrate concentration, with activity within a comparable range to the clinically relevant E. coli and Erwinia asparaginases. An-88 had 81% similarity whereas both An-104 and An-107 ASNases had 88% identity to human L-ASNase, which is in stark contrast to the only 30% similarity seen with bacterial L-ASNases. Detailed enzymatic characterization of these three lead drug candidates is currently being performed. Preliminary cytotoxicity data testing the ancestral L-ASNase candidates against a T-cell ALL cell line CCRF-CEM showed an expected dose response, with An-107 showing the highest cytotoxicity.

Our findings demonstrate that ASR is an efficient protein drug design/optimization platform that we have now extended beyond our previous work with recombinant coagulation factors (Zakas et al. Nat Biotech 2017, Knight et al. Blood Adv 2021) to therapeutic enzymes; thereby, enabling the creation of novel recombinant L-ASNases with translational implications due to improved enzymatic specificity and reduced toxicity.