Attenuating and ablating FcγR‐related functions of IgG
There are circumstances where binding to FcγR is unnecessary or undesirable in the MOA of a therapeutic mAb. Unmodified IgG irrespective of its subclass or intended therapeutic effect has the potential to engage an FcγR which may lead to suboptimal therapeutic performance or to unexpected and catastrophic consequences.57, 59 Clearly reducing or eliminating FcγR interactions, when they are not required for therapeutic effect, may be desirable. Indeed, this had been addressed by the choice of IgG subclass or by modifying the Fc region. Indeed, most efforts in Fc engineering mAbs that have translated to an approved drug have focused on the reduction or elimination of FcγR interactions (Table 4).
One approach to minimize interactions with the activating FcγR has been the use of IgG4 or IgG2 backbones, which show a more restricted specificity for the activating FcγR and consequently have been traditionally, and simplistically, viewed as “inert” IgG subclasses. Notwithstanding the unexpected, and FcγR‐dependent, severe adverse reaction induced by the IgG4 TGN1412 mAb, the IgG4 or IgG2 backbones have been successfully used in many settings. Indeed, checkpoint inhibitors, such as mAbs targeting CTLA‐4 or the PD‐L1/PD‐1 interaction for the suppression of inhibitory signals that contribute to immune tolerance in the tumor microenvironment, are formatted on an IgG4 backbone. Pembrolizumab, nivolumab and cemiplimab are all anti‐PD‐1 antibodies currently used for cancer therapy and have been formatted on an IgG4 backbone95, 96, 97 with a stabilized core hinge (S228P) to prevent half‐IgG4 exchange. Similarly, the checkpoint inhibitor tremelimumab is an anti‐CTLA‐4 antibody formatted on an IgG2 backbone to avoid potential ADCC killing of target cells.98
However, the use of IgG2 and IgG4 as “inert” subclasses is problematic. Both bind to the activating receptors FcγRIIa‐H131 and FcγRI, respectively (Table 2), and initiate effector functions such as neutrophil activation and apoptosis induction.75, 99 Interestingly, in experimental systems, cross‐linking of anti‐PD‐1 IgG4‐based mAb by FcγRI switched its activity from blocking to activatory.10 Moreover, IgG4 binds to FcγRIIb, which may scaffold the therapeutic mAb. Although scaffolding may be beneficial in some contexts, for example, in immune agonism,43 it can be disastrous and unexpected in others as it was for the anti‐CD28 TGN1412 mAb.59 Thus, the IgG2 and IgG4 subclasses are not the optimum choice for “FcR‐inactive” mAbs, and so modifying the Fc is a more direct approach.
The complete removal of the heavy‐chain N‐linked glycan is well known to ablate all FcγR binding by dramatically altering the Fc conformation.36, 67, 101, 102 Atezolizumab, an IgG1 anti‐PD‐L1 checkpoint inhibitor mAb, utilizes this strategy and eliminates FcγR and also complement activation.13
Modification to the Fc amino acid sequence of the FcγR‐contact regions can also be used to reduce FcγR binding. A widely used modification of IgG1 is the substitution of leucine 234 and 235 in the lower hinge sequence (L234 L235 G236 G237) with alanine (L234A L235A). It is often referred to as the “LALA mutation” and effectively eliminates FcγR binding by more than 100 fold104, 105 and is used in teplizumab and spesolimab (Table 4).
A separate strategy has used combinations of amino acid residues from the FcγR‐binding regions of IgG2 and IgG4, which have restricted FcγR specificity, together with other binding‐inactivating mutations. The lower hinge amino acids of the IgG1 mAbs durvalumab (anti‐PD‐L1) and anifrolumab (anti‐interferon‐α receptor; Table 4) were modified to mimic the lower hinge of IgG4 (L234F). They additionally incorporated L235E in the lower hinge and P331S in the F/G loop of the CH2 domain to ablate FcγR binding by disrupting two major FcγR contact sites7 and also coincidently decreasing C1q activation.16
IgG4 mAbs have been similarly engineered to eliminate their interaction with FcγRI and FcγRIIb. The IgG4 anti‐PD‐1 antibody tislelizumab has had its FcγR contact residues in the lower hinge E233, F234, L235 substituted with the equivalent residues of IgG2 P, V, A (E233P, F234V, L235A) as well as the additional D265A mutation which disrupts a major FcγR contact in CH2. It also has substitutions in the core hinge S228P and the CH3 L309V and R409K to stabilize the H‐chain disulfides and CH3 interactions, respectively, thereby preventing half‐Ig exchange characteristic of natural IgG4. Collectively, these modifications create a stable IgG4 with no FcγR binding nor complement activation.17
Thus, Fc engineering is an effective way to remove FcγR effector functions and may be preferable to using unmodified IgG2 or IgG4 backbones that have a more restricted repertoire of FcγR interactions but which are still able to induce certain effector functions.