Accurate detection of liabilities to de-risk antibody development

How to identify sequence-based liabilities

Category:
Science
Date:
October 14, 2022
Read time:
8
min
An illustration of an IgG antibody with sequence-based liabilities and post-translational modifications shown in the CDRs, C-regions and FRs of the antibody

The accurate annotation, definition and characterization of antibody sequences is a crucial step in antibody drug discovery with direct consequences in the final stages of the development process. The slightest oversight in the annotation of the sequences, detection of liabilities and identification of mutations will undoubtedly lead to a reduction in the efficacy of any promising lead candidates. Thus, efficient and exact annotation tools are needed to ensure the developability[1] of antibodies and decrease the risk of producing non-effective therapeutics.

PipeBio’s annotation pipeline is an extremely powerful and configurable tool to precisely analyze various kinds of antibody sequences and identify liabilities that impact the antibody’s affinity and specificity to its target. The customizable scaffold system allows annotation of traditional and custom sequence regions with a wide variety of germlines, annotation definitions and liabilities to choose from.

Identify CDRs, framework regions and germline genes accurately

PipeBio’s annotation engine is able to efficiently detect and label CDRs and framework regions in antibody sequences from both the heavy and light (kappa and lambda) chains. The variable (V), joining (J), diversity (D) and constant (C) gene segments are also identified thanks to our manually-curated germline databases (Fig. 1). Furthermore, custom databases can be created to identify and annotate additional regions in the sequence, such as linkers, tags, loops or any others. We provide germline databases for human, mouse, alpaca, rat, rabbit and chicken, and can easily generate databases for other species not mentioned.

Sequence viewer showing nucleotide and amino acid sequences of an IgG heavy chain on PipeBio
Figure 1. PipeBio’s sequence viewer displaying the CDRs, framework regions and germline genes identified by the annotation pipeline.

Detect liabilities and flag incorrect sequences

One of the most powerful and advantageous characteristics of our annotation tool is its ability to detect a wide range of liabilities in the sequences that can affect the binding of the antibody with the relevant antigen. These include post-translational modifications[2] (PTMs), quality validations and structural verifications that can be applied to specific regions of the sequence as needed. These validations are specified in the scaffold system with a designated liability score (Fig. 2). Additionally, custom validations and liabilities can be easily added to the whole sequence or to desired regions.

The scaffold and errors flagged by PipeBio's sequence liability detection configurations
Figure 2. a) The validations for CDR-H1 are defined in the scaffold system which will be used by the annotation engine to detect liabilities (errors and warnings) in the sequences analyzed. b) A list of all the errors found in the annotated sequences is displayed in the output document

Here we will list and briefly describe some of the different validations and liabilities that can be incorporated in your analysis using PipeBio to optimize the developability of antibodies.

Sequence quality validations

Somatic hypermutations (or SHMs) increase antibody diversity and are essential for the adaptation of the immune system against new foreign elements. It might become difficult to differentiate between those mutations and sequencing errors caused by low quality base calls of nucleotides. We compute the Phred quality score (Q) system as an indicator of sequence quality (Fig. 3), which is logarithmically related to the probability of a wrong base call during DNA sequencing. For instance, a Q30 is equivalent to the probability of 1 incorrect base call in 1000 (99.9% accuracy), while a Q20 indicates 1 wrong base call in 100 (99% accuracy). To ensure high sequence quality, we provide the following validations:

  • Ambiguous bases: nucleotides different from A, C, G, T and U are flagged.
  • Nucleotide quality: marks individual bases with a Phred score lower than a specified threshold.
  • Average quality region: if a certain % of a region in the sequence has a Phred score lower than the specified threshold, the region will be flagged.
  • Secondary peaks: highlights heterozygote base calls in Sanger reads.
The Phred quality score is calculated for each region of a sequence on PipeBion
Figure 3. The Phred quality score is calculated for each region of the sequence. Typically, sequences with a high amount of detected liabilities (warnings) will have lower Phred scores, as seen in the table

Sequence structure validations

We examine sequence modifications in multiple regions to guarantee the integrity of an antibody (Fig. 4).

  • Missing regions: A sequence will be flagged if any of the CDRs, frameworks or other custom regions defined in the scaffold are missing
  • Truncation: The annotation engine is able to detect truncated framework regions in the beginning and in the end of the sequence. The frameworks can be reconstructed using the germline gene as reference
  • Frameshifts: Detects sequences with regions containing frameshifts due to mutations
  • Region length: Useful when regions are shorter or longer than a specified length
Example sequence on PipeBio that is flagged as INCORRECT due to a TAA stop codon, out of frame and missing regions
Figure 4. Example sequence flagged as INCORRECT due to a TAA stop codon, out of frame and missing regions

Post-translational modifications (PTMs)

PTMs refer to the modifications that affect the amino acid chain after the biosynthesis of an antibody (Fig. 5). This process provides additional sequence variation and alters the antibody binding properties, which might be troublesome when developing antibodies designed against a specific target. PTMs include the addition of functional groups in the amino acid chain, the modification of existing amino acids in the sequence or the cleavage of bonds, among others [3, 4]. PTM liabilities can be either added to specific regions of the antibody cover the whole sequence. PipeBio can detect the following PTMs based on the amino acid sequences:

  • Asparagine deamidation
  • Aspartate isomerization
  • N-linked glycosylation
  • Lysine glycation
  • Methionine oxidation
  • Tryptophan oxidation
  • Aspartic acid – Proline cleavage
  • Hydrolysis
An illustration of an IgG antibody with sequence-based liabilities and post-translational modifications shown in the CDRs, C-regions and FRs of the antibody
Figure 5. Post-translational modifications can occur in both the constant and variable regions of IgGs and can be risk factors that decrease the stability of monoclonal therapeutic antibodies

Additional sequence validations

The scaffold system offers the flexibility to add custom validations and detect PTMs in order to support a wide range of structural needs, additional motifs and quality checks. Other liabilities that we can include during the annotation of sequences are, among many others:

  • Stop codons: Depending on the codon table used, the annotation tool will detect stop codons in the sequences.
  • Cysteine count: Different options are available for flagging sequences depending on their C count. Free sulfhydryls (odd cysteines or unpaired cysteines) may result in lower stability, structural changes and a higher risk of aggregation in monoclonal antibodies. We provide support for both antibody and non-antibody configurations with scaffolds that do not flag sequences with a large number of cysteines.
  • Hydrophobic patches: which may aggregate during production.
  • Amino acid repeats: sequences will be marked whenever an amino acid repeat is present that does not correspond to a specific pattern (e.g., “GSS” in a linker region).

Bring your sequences and PipeBio will do the rest

Whether your analysis includes classical IgG human antibodies or non-antibody scaffolds engineered in silico, the PipeBio annotation engine offers a flexible system for accurately analyzing your sequences to fit any structural requirement. We take antibody developability very seriously, and we like to ensure that the candidates coming out of our bioinformatics pipeline have the best prospects for being successful therapeutically.

References

  1. Raybould, M., & Deane, C. M. (2022). The Therapeutic Antibody Profiler for Computational Developability Assessment. Methods in molecular biology (Clifton, N.J.), 2313, 115–125. https://doi.org/10.1007/978-1-0716-1450-1_5
  2. Ramazi, S. & Zahiri, J. (2021). Post-translational modifications in proteins: resources, tools and prediction methods. Database, Volume 2021, baab012. https://doi.org/10.1093/database/baab012
  3. Lu, X., Nobrega, R. P., Lynaugh, H., Jain, T., Barlow, K., Boland, T., Sivasubramanian, A., Vásquez, M., & Xu, Y. (2019). Deamidation and isomerization liability analysis of 131 clinical-stage antibodies. mAbs, 11(1), 45–57. https://doi.org/10.1080/19420862.2018.1548233
  4. Hattori, T., & Koide, S. (2018). Next-generation antibodies for post-translational modifications. Current opinion in structural biology, 51, 141–148. https://doi.org/10.1016/j.sbi.2018.04.006


Identify liabilities, de-risk antibody development and obtain high-quality therapeutic antibody candidates?

Other recent posts