COMPUTATIONAL ENGINEERING OF A NANOBODY (sdAb-ECA2-𝛼) SPECIFIC AGAINST THE BIOPHARMACEUTICAL L-ASPARAGINASE

Nanobodies are single-domain antigen-binding domains derived from camelid heavy-chain antibodies. They present a reduced size (12–16 kDa), high stability, solubility, specificity, and low immunogenicity due to their structural similarity to human antibodies. Furthermore, they can be efficiently expressed in bacterial systems, which drastically reduces production costs compared to traditional monoclonal antibodies. The objective of this work is to generate, in silico, a nanobody directed at the enzyme asparaginase, a crucial biopharmaceutical used in the treatment of acute lymphoblastic leukemia. The development of a specific nanobody against this target aims to enable a more sensitive and precise identification of the medication in different biological samples from patients.

The computational design was performed through the RFantibody pipeline executed via Google Colab. This platform integrates three advanced softwares: RFdiffusion, which generates the three-dimensional geometry of the hypervariable regions (CDRs) responsible for binding; ProteinMPNN, which determines the ideal amino acid sequence for the interaction interface; and RoseTTAFold2 (RF2), in charge of performing the initial structural validation and discarding incorrect folds. As structural templates, the folding of a pre-existing nanobody from the group (sdAb-mrh-IgG) and the monomer of Escherichia coli type 2 asparaginase (PDB ID: 3ECA) were used.

Within the workflow, RFdiffusion generated 5 geometric scaffolds, ProteinMPNN proposed 10 sequences for each, and RF2 validated the quality of each complex 6 times. The process was repeated until generating more than a thousand distinct sequences, which were subsequently evaluated in triplicate by AlphaFold3. Models with the highest structural confidence were selected, using ipTM (interface) and pTM (chain) values greater than 0.75 as cutoff criteria. The screened structures underwent affinity analyses in Prodigy and biophysical evaluations in ProtParam and Ramachandran plots.

As a result, 19 promising sequences were obtained that met all validation criteria. The next step will consist of heterologous expression in E. coli BL21(DE3) using the pET28(a)+ vector for the execution of experimental binding tests with L-asparaginase.

We would like to thank CAPES, CNPq, Finep, and FAPERJ for the financial support that made this project possible.