March 1, 2015 (Vol. 35, No. 5)

Early Removal of T-Cell Epitopes Reduces Risk of Adverse Immune Responses in Patients

The development of therapeutic proteins such as humanized antibodies and fully human antibodies has in many indications led to remarkable clinical benefit. In spite of this, immunogenicity in patients still remains a significant drawback. The antidrug-antibodies (ADA) produced against therapeutic proteins, such as infliximab, adalimumab, alemtuzumab, interferon-α, and factor VIII, can be neutralizing, so limiting their efficacy, or can lead to their rapid elimination from the body, so reducing their half-life.

In rare instances, generation of an ADA response can result in life-threatening autoimmune-like toxicity by ADAs cross-reacting with endogenous protein, as was observed in patients treated with epoetin alfa (EPO) and thrombopoietin.

As the industry proceeds to develop more exotic therapeutic proteins, such as immunotoxins, bispecific antibodies, and novel scaffolds, the concerns about immunogenicity remain and steps to reduce the risk of immunogenicity are warranted. Ideally the risk should be mitigated before the product enters clinical development to avoid causing potential harm to patients and to minimize the amount of expenditure on a drug that fails due to immunogenicity.

Early Assessment

An outline of the process leading to the generating of high affinity, isotype switched (e.g., IgG) ADAs is shown in Figure 1. This process involves four key stages: (A) uptake of the protein by antigen-presenting cells (APCs), which undergo maturation after stimulation through pattern recognition receptors (PRR); (B) processing of the protein within APCs and presentation of peptides in the context of major histocompatibility complex II (MHC class II); (C) activation of CD4+ T cells against linear peptides (T-cell epitopes) bound to MHC class II; and (D) help provided by CD4+ T cells to B cells enabling subsequent production of high affinity, isotype switched antibodies.

The help provided by T cells is a requirement for somatic hypermutation and isotype switching in B cells producing ADAs and the T-cell epitopes present in the sequences of proteins is a key driver of immunogenicity.  A number of experimental approaches have been developed for predicting and identifying T-cell epitopes including in silico, in vitro, and in vivo methods.

In silico tools that identify linear peptides within the protein sequence that bind to MHC class II are widely available (e.g., RankPep, Propred, NetMHC, and iTope™). However, in silico methods do not take into account other stages in the generation of T-cell epitopes (uptake, antigen processing, and T-cell activation) and therefore inherently over-predict the presence of T-cell epitopes. 

One recent improvement in the in silico approach has been to utilize sequence information from databases containing data on T-cell epitopes identified using in vitro studies (such as the TCED™ from Antitope). Such databases enable the interrogation of positive hits from MHC class II binding analysis against sequence data from previously identified T-cell epitopes thus improving the accuracy of in silico tools. 

In vitro T-cell assays using ex vivo human peripheral blood mononuclear cells (PBMCs) can provide a more accurate assessment of the presence (in the case of EpiScreen™ time course T-cell assay), and number (in the case of EpiScreen T-cell epitope mapping) of T-cell epitopes. These cell preparations containing APCs and T cells are stimulated with whole proteins or peptides derived from the therapeutic protein, and the capacity of the sample to activate T cells is measured via a variety of activation markers. Different in vitro T-cell assay formats enable testing of proteins that directly modulate T-cell activation (e.g. EpiScreen DC:T cell assay) as well as investigating the effects of product formulations on immunogenicity.

Importantly, data from the EpiScreen in vitro T-cell assays have been shown to demonstrate a good correlation with clinical immunogenicity (Figure 2). In vitro T-cell assays that map the precise location of T-cell epitopes can also enable deimmunization strategies to be formulated to produce protein sequences in which T-cell epitopes have been removed.

Figure 1. Summary of the adaptive humoral immune response for T-dependent antigens.

Composite Human Antibodies

Strategies to mitigate the risk of immunogenicity may require designing protein sequences that are devoid of T-cell epitopes.  Historically, such antibody engineering strategies have focused on an unachievable objective of designing variable domains that are as human-like as possible in order to effectively “tolerize” the protein for the human immune system. Given the uniqueness and specific requirement for nonhuman germline residues in the complementarity determining regions (CDRs) and, frequently, in the framework regions, of therapeutic antibodies, it will be impossible to completely avoid T-cell epitopes and thus the potential for immunogenicity. 

This issue can however be avoided using Composite Human Antibody™ technology in which variable domains are designed from unrelated human antibody sequence segments that are screened using in silico tools (iTope and TCED) to avoid the incorporation of segments containing T-cell epitopes. 

Lead Composite Human Antibodies are generated and tested for functionality as well as immunogenicity using EpiScreen  T-cell assays to confirm the avoidance of T-cell epitopes. Six antibodies created using Composite Human Antibody technology are currently in clinical development for multiple indications.

A similar approach has been applied to nonantibody-based therapeutic proteins (Composite Proteins™). This technology uses a similar strategy to identify key residues within a protein sequence that may constrain downstream protein modifications. The protein sequence is then mapped for T-cell epitopes (using the EpiScreen T-cell epitope mapping assay) to identify T-cell epitopes for subsequent removal.

The identified T-cell epitopes are ranked in order of frequency of responses in the assay, and magnitude, to drive a deimmunization strategy that focuses on removing the most problematic T-cell epitopes first while ensuring structure and function of the starting protein are maintained. Composite Proteins™ technology has been applied to a number of nonantibody proteins including toxins for use in therapeutic products.

Figure 2. Mean frequency of anti-drug antibodies observed clinically (source PubMed) correlates with observed immunogenicity (% donor response) in ex vivo EpiScreen T-cell assays.


Reducing the risk of an immunogenic response to therapeutic proteins is best undertaken prior to clinical development to avoid causing harm to patients through poor efficacy or toxicity and to reduce expenditure on a product that then fails due to immunogenicity. By identifying the T-cell epitopes in therapeutic antibodies and other proteins at the early preclinical stage of development an assessment of their immunogenic potential can be made. Use of assay methods that correlate with clinical immunogenicity will help benchmark the risk and allow for selection of the best lead candidate to take into development. 

T-cell mapping allows for a new version of the therapeutic protein to be designed by avoiding T-cell epitopes. A future generation of therapeutic antibodies and other proteins with reduced potential for immunogenicity is a realistic goal using the approaches described.

Ji-won Choi, Ph.D. ([email protected]), is director of scientific affairs and technology evaluation at Abzena. The author is grateful to the EpiScreen and the antibody and protein engineering teams at Antitope, an Abzena subsidiary, for their assistance. The author also extends his thanks to Matthew Baker, Ph.D., CSO of Abzena, for his help in preparing the tutorial.

Previous articleNeuronal Cloaking Device Hides Synaptic Response
Next articleMusic in Your DNA and a New Species of Human?