Pharmacokinetics of Engineered Antibodies

  In the previous article, we covered the basics of antibody pharmacokinetics. In this second article, we’ll introduce FcRn binding-enhanced engineered antibodies. This article will explore the history of research on half-life extension through FcRn-enhancing mutations, case studies in cynomolgus monkeys and humans, and methods for predicting human antibody pharmacokinetics from animal models.

  Antibodies, being large molecules of approximately 150 kDa, are not easily filtered by the glomeruli. They also exhibit long plasma half-lives due to a recycling mechanism mediated by the neonatal Fc receptor (FcRn). After being taken up by vascular endothelial cells and other tissues, antibodies bind to FcRn within acidic endosomes. This binding prevents their transport to lysosomes and protects them from proteolytic degradation. Subsequently, through exocytosis, they are recycled to the cell surface, where the neutral pH of the bloodstream causes them to dissociate from FcRn and re-enter circulation. This cycle repeats, resulting in antibodies having a longer half-life (2–3 weeks) compared to other proteins. Leveraging this property, technologies have been developed to extend antibody half-life by introducing amino acid mutations that enhance FcRn binding, thereby improving recycling efficiency (Figure 1). Among these, the YTE (M252Y/S254T/T256E) and LS (M428L/N434S) mutations have demonstrated half-life extension effects in preclinical studies using human FcRn transgenic mice and cynomolgus monkeys. These modifications have since been applied to several approved drugs following clinical trials (Table 1).

  The YTE mutation was first reported by Dall’Acqua et al. in 2006. Compared to unmodified antibodies, YTE variants bind approximately 10 times more strongly to human and cynomolgus FcRn at pH 6.0, but show negligible binding at pH 7.4. This means YTE-modified antibodies bind tightly to FcRn under acidic conditions in endosomes and dissociate under neutral conditions in the bloodstream. As a result, their recycling efficiency is enhanced, and their plasma half-life in cynomolgus monkeys is extended by approximately fourfold [6]. In humans, motavizumab-YTE, a YTE-modified version of the anti-RSV antibody motavizumab, also demonstrated a fourfold increase in half-life [7].

  The LS mutation, reported by Jonathan Zalevsky et al. in 2010, enhances FcRn binding at pH 6.0 by approximately 11-fold. This extends plasma half-life in both human FcRn transgenic mice and cynomolgus monkeys, with the latter showing a threefold increase [8]. In humans, LS-modified VRC01, an anti-HIV antibody, exhibited a fourfold improvement in half-life [9].

  While these FcRn-enhancing modifications successfully extend half-life, they may inadvertently increase binding to rheumatoid factor (RF), an autoantibody that targets the Fc region of IgG. RF is commonly detected in patients with rheumatoid arthritis and other autoimmune diseases and is known to form immune complexes that help eliminate pathogens [10][11]. For example, a humanized anti-CD4 IgG1 antibody with the N434H mutation showed significantly increased RF binding compared to its parent antibody. Similar concerns have been raised for YTE and LS modifications [12]. Enhanced RF binding may affect pharmacokinetics and immunogenicity, posing potential risks in autoimmune disease treatments.

  To address these concerns, we developed novel FcRn-enhancing mutations that do not increase RF binding. We identified five variants—ACT1 (N434A/Y436T/Q438R/S440E), ACT2 (N434A/Y436V/Q438R/S440E), ACT3 (M428L/N434A/Y436T/Q438R/S440E), ACT4 (M428L/N434A/Y436V/Q438R/S440E), and ACT5 (M428L/N434A/Q438R/S440E)—which bind 4 to 12 times more strongly to human FcRn at pH 6.0 and extend plasma half-life by 2 to 4 times in cynomolgus monkeys [12]. When ACT3, ACT5, and LS mutations were introduced into antibodies targeting antigen X, the ACT variants demonstrated half-life extension comparable to LS. ACT3- and ACT5-modified antibodies have progressed into clinical trials, where their extended half-life has been confirmed. For instance, NXT007, a bispecific antibody designed to mimic activated factor VIII (FVIII) for hemophilia treatment, incorporates the ACT5 mutation and has demonstrated a plasma half-life of approximately 70 days in clinical trials [13].

  Antibody half-life is influenced by FcRn interactions, which vary across species. Rodents are commonly used in preclinical studies, but human IgG binds more strongly to mouse FcRn than to human FcRn, complicating predictions. In contrast, human IgG binding to cynomolgus FcRn closely resembles that of human FcRn, making cynomolgus monkeys a preferred model. Human pharmacokinetics are often predicted using allometric scaling based on body weight differences. For example:

human CL = monkey CL × (BW human/ BW monkey)^α　・・・Equation 1

  The scaling exponent α is determined by analyzing multiple pharmacokinetic datasets. We analyzed 24 antibodies using a two-compartment model and found that α = 0.8 yielded the most accurate predictions [14]. This method enables the precise extrapolation of human pharmacokinetics from cynomolgus monkey data. However, for FcRn-enhanced antibodies, α = 0.8 was insufficient, and α = 0.55 was found to be optimal [15].

  Due to ethical and cost concerns, the use of cynomolgus monkeys is becoming more limited. Recently, Tg32 mice expressing human FcRn have been adopted as an alternative model. These mice lack the mouse FcRn α-chain and express the human FcRn α-chain, resolving species differences in FcRn binding. Tg32 mice offer high throughput and are widely used in research, with predictive accuracy for human pharmacokinetics comparable to that of cynomolgus monkeys [16].

  However, endogenous mouse IgG binds weakly to human FcRn, resulting in minimal competition during recycling. This reduces the clearance difference between wild-type and FcRn binding-enhanced antibodies, leading to discrepancies with human data. To address this, we co-administered intravenous immunoglobulin (IVIG) to mimic human IgG competition. This improved the accuracy of clearance predictions and revealed that α = 0.73 is optimal for Tg32-based scaling [17].

  With the establishment of predictive models using cynomolgus monkeys and Tg32 mice, it is now possible to accurately forecast human pharmacokinetics from preclinical data. The potential of antibody therapeutics is expected to expand further through the development of additional half-life extension technologies and the refinement of mathematical models for predicting pharmacokinetics. FcRn-enhancing modifications significantly prolong half-life, reduce dosing frequency, and contribute to improved patient quality of life.