Influence of immunoglobulin isotype on therapeutic antibody function

Monoclonal antibody (mAb) drugs have transformed cancer therapy over the last 3 decades,¹  with a plethora of recent examples where established and previously untreatable tumors have been eradicated.²  Unmodified, “naked” mAbs can be harnessed to deliver therapy through a number of mechanisms including direct targeting of tumor to elicit immune cell-mediated clearance; agonistic receptor engagement to stimulate tumor immunity or effect tumor cell apoptosis; and blocking of receptor:ligand interactions important for tumor survival or suppression of antitumor immunity. Target specificity, imparted by the mAb variable domains, is clearly of paramount importance in each of these scenarios. However, it is also apparent that the mAb constant region plays a crucial role, much of which is mediated through interaction of the mAb Fc with Fcγ receptors (FcγRs). In this review, we describe how mAb isotype, which dictates FcγR binding specificity and other structural characteristics, critically influences mAb activity and discuss how this knowledge can be used to improve therapeutic efficacy.

The first demonstrations of the importance of isotype selection in therapeutic activity was in studies with mAbs that directly engage their tumor cell targets, such as clinical rituximab (anti-CD20) and trastuzumab (anti-HER2). Early findings observed the impact of isotype on mAb therapy where particular mouse and human isotypes were seen to offer protection in xenograft models, and efficacy was dependent on FcγR and effector cells.^3,4  One of the principal killing mechanisms of these agents is recruitment of activatory FcγR-expressing immune effectors that mediate target cell deletion (Figure 1A). In seminal mouse studies in 2000, Clynes et al⁵  demonstrated that rituximab and trastuzumab required functional activatory FcγR expression for therapeutic activity, whereas, in contrast, the presence of the inhibitory FcγRIIB reduced mAb efficacy.⁵  Later, detailed syngeneic studies were carried out where it was observed that mouse immunoglobulin (Ig)G2a MAbs that engage activatory FcγR with relatively high affinity⁶  provided effective therapy, whereas isotypes with lower affinities were much less effective.⁷  Through these studies, the paradigm was established that a preference for activatory vs inhibitory FcγR engagement (high activatory:inhibitory [A:I] FcγR binding ratio) was critical for therapeutic mAb activity.^6,8  Since these initial observations, many studies using a variety of agents including rituximab, trastuzumab, and cetuximab (anti-EGFR), have demonstrated an absolute requirement in vivo for activatory FcγR interactions to facilitate depletion of both normal and malignant target cells.^7,9-12  Similar to mouse IgG2a, the human IgG1 isotype selected for clinical reagents has a high A:I FcγR binding ratio.

In preclinical mouse models, circulating monocytes^7,13,14  and tissue macrophages^{7,9,11,12,15-18}  have been demonstrated to be the primary effector cells involved in mAb-induced cell killing, although debate still exists regarding which has the dominant role, and this may vary dependent on target cell and location. Roles for natural killer (NK) cells¹⁹  and neutrophils^20,21  have been demonstrated in some models; however, they have not generally been found to be important for efficacy. In humans, the effector populations are less clear. In vitro experiments with blood-borne effectors suggest NK cells play a predominant role.²²  However, these assays do not necessarily reflect the situation in tissues, especially as the absence of macrophages in blood is likely to underestimate their role. The association between functionally relevant FcγR polymorphisms and clinical response to therapy underscores the critical role of FcγR in mAb activity in humans and also supports a role for macrophages. Cartron et al²³  first demonstrated that inheritance of an F to V amino acid change at position 158 in FcγRIIIA, which increases affinity for human IgG1, a receptor expressed on macrophages and NK cells, was associated with enhanced responses to rituximab in follicular lymphoma patients. Subsequently, >40 similar investigations with a range of mAbs in a variety of hematologic and solid cancer settings have been reported,^24,25  and although findings are mixed and sometimes conflicting, many do support a role of FcγR in clinical activity. Negative findings in some studies may be explained by small patient numbers, combined treatment with chemotherapy, or the presence of additional mAb mechanisms (eg, direct inhibitory or cytotoxic effects) that confound the results.

The importance of a high A:I FcγR binding ratio has stimulated considerable efforts to optimize FcγR interaction, particularly with FcγRIIIA, through amino acid substitution or glycoengineering of mAb Fc.^16,26-28  The most clinically advanced agent is the glycoengineered anti-CD20 obinutuzumab (GA101),²⁹  which, in combination with chlorambucil, was recently shown to nearly double progression-free survival in chronic lymphocytic leukemia patients compared with rituximab.³⁰  The ability to augment mAb activity by increasing A:I ratio while maintaining FcRn association and biological half-life^31-33  is likely to be limited, however. Moreover, there are potential negative impacts such as enhanced antigen shaving/trogocytosis,³⁴  an FcγR-dependent process whereby mAb/Ag complexes are removed from the target cells by effectors, thus leaving the target cells invisible to effector mechanisms. In addition to these approaches, other strategies may be possible to enhance efficacy such as that recently suggested by Kinder et al,³⁵  where specific manipulation of FcγRI engagement was used to modulate effector cell cytokine production to allow manipulation of the tumor microenvironment without effects on direct cell killing. In addition, other classes of immunoglobulin molecules, such as IgA and IgE, have been shown to mediate therapeutic effects through engagement of Fcα and Fcε receptors on immune effector cells, respectively.^36,37  Future studies will be required to determine the relative efficacy of these agents compared with Fc engineered IgG.

An unexpected role for activatory FcγR was also suggested in recent studies of immunomodulatory mAbs that provide therapy by stimulating antitumor immunity. Immunomodulatory mAbs are designed to either block key inhibitory pathways suppressing effector T cells (checkpoint blockers) or to agonistically engage costimulatory immune receptors (immunostimulatory). Recent preclinical data suggest that checkpoint blocking anti-CTLA4 and anti-PDL1 and immunostimulatory anti-GITR and anti-OX40 mAbs actually deliver much of their therapeutic effect through deletion of intratumoural T regulatory (Treg) cells, thus releasing CD8 T cell-mediated antitumoral immunity.^38-42  In this context, the mAb function as classical deletors, much as the direct targeting mAbs discussed above, requiring activatory FcγR engagement on effector cells in the tumor microenvironment (Figure 1A). In the case of CTLA-4, these were FcγRIV-expressing intratumoral macrophages.⁴¹ 

There is some supporting evidence that similar mechanisms may operate in human patients. Ipilimumab, a human IgG1 anti-CTLA4, was recently shown to mediate FcγR-mediated cytotoxicity of human Tregs ex vivo.⁴³  In addition, in a small clinical study, melanoma patients responding to ipilimumab had significantly higher baseline frequency of nonclassical monocytes and more activated tumor-associated macrophages expressing FcγRIII, which correlated with lower intratumoral Treg numbers after therapy, suggesting Treg deletion may also occur in patients.⁴³  In contrast, other checkpoint blockers, such as nivolumab and pembrolizumab (both anti-PD1), are IgG4 isotypes selected for minimal FcγR interactions. Recent preclinical data,⁴²  as well as the association of their efficacy with PDL-1 expression in at least some studies,⁴⁴  suggest that these mAbs may work as true blockers. It is important to note, however, that even the nonfunctional isotypes IgG2 and IgG4 are able to bind human FcγR as immune complexes,^45,46  and their activity can be influenced by FcγR as recently demonstrated for the IgG4 TGN1412.⁴⁷  Thus, further work is needed to fully elucidate the in vivo mechanisms of action of these mAbs.

A further potential role for activatory FcγR in therapeutic anticancer mAb activity is a described vaccine effect with direct targeting mAbs. It has been appreciated for some time that passively administered neutralizing antiviral mAbs can potentiate viral CD8 T-cell immunity.⁴⁸  Two recent studies have demonstrated a similar vaccine effect in mice challenged with human CD20-expessing tumor cells treated with anti-CD20 mAbs,^49,50  with induced immunity dependent on interaction of the CD20 mAb with activatory FcγR on dendritic cells.⁵⁰  It will be important to confirm this effect in models using endogenously expressed (and thus less immunogenic) target antigens; however, these studies highlight the multiple and sometimes unexpected effects that activatory FcγR interaction can have on therapeutic mAb activity.

Both positive and negative roles for the inhibitory receptor, FcγRIIB, have also been established with therapeutic mAbs. Early studies demonstrated that the presence of FcγRIIB was detrimental to the therapeutic activity of direct targeting agents, such as rituximab.⁵  Subsequent studies showed that the FcγRIIB interaction can inhibit target cell killing through competition for Fc engagement or by promoting antigenic modulation¹⁰  where opsonizing mAb Fc interacts in cis with FcγRIIB on target B cells, leading to enhanced internalization of mAb/Ag complexes. This process compromises mAb efficacy as it both consumes the mAb and downregulates the target so that cells become invisible to effector mechanisms.^51-53 

In contrast, FcγRIIB engagement has been shown to be requisite for the activity of agonistic mAbs. These agents stimulate signaling through their target receptors, typically members of the tumor necrosis factor receptor (TNFR) superfamily. As cancer therapeutics, they are designed to enhance tumor immunity by engaging costimulatory receptors such as CD40, 4-1BB, or OX40, on APC or T-effector cells, or to promote apoptosis by stimulating death receptors (DRs) such as DR4, DR5, or Fas (CD95) on cancer cells.⁵⁴  In contrast to direct targeting agents, the agonistic activity of these mAbs is dependent on their ability to engage inhibitory FcγRIIB,^55-59  and mAbs with high A:I ratios (eg, mouse IgG2a, human IgG1) are largely inactive in preclinical models, whereas those with low A:I ratios (eg, mouse IgG1 and rat IgG2a) are highly agonistic.^55-60  Signaling through FcγRIIB is not required to confer activity; rather, it provides a crosslinking scaffold for the mAbs to facilitate TNFR clustering and activation^57,61-63  (Figure 1B). Activatory FcγR can also mediate crosslinking in vitro⁵⁷  or in vivo when sufficiently expressed at the target location^58,63; thus, the predominant role of FcγRIIB in vivo may, in part, reflect its bioavailability.

The dependence of agonistic mAb on FcγRIIB presents a challenge when developing therapeutics, as human IgGs have low affinity for this receptor.⁴⁵  Fc engineering of human IgG1 to enhance FcγRIIB affinity increases therapeutic activity of CD40 and DR5 mAbs in human FcγRIIB transgenic mice.^64,65  A potential downside to this approach, however, is the resulting dependence on local FcγRIIB expression for activity.

Not all therapeutic effects require Fc receptor interactions. For example, a recent study from Dahan et al⁴²  demonstrated greater efficacy with blocking anti-PD1 mAbs engineered to prevent FcγR engagement (Figure 1C). Moreover, human IgG2 agonistic anti-TNFR mAbs also do not require FcγR interaction for agonistic activity⁶⁰  (Figure 1B). ChiLob 7/4, a CD40 mAb that has recently completed a phase 1 clinical trial,⁶⁶  is agonistic in human CD40 transgenic mice as IgG2 in the absence of FcγR expression and even as a F(ab′)₂ fragment.⁶⁰  FcγR-independent activity has also been demonstrated for another IgG2 CD40 mAb in a clinical trial: CP870-893.⁶⁷ 

The key to IgG2 agonistic activity lies in the unique configuration of its hinge region,⁶⁰  which can adopt alternative conformations through disulphide rearrangement.^68-70  IgG2 is believed to be synthesized with all 4 heavy chain (HC) hinge cysteines involved in parallel inter-HC disulphide bonds, a conformation designated as IgG2(A). Over time, these bonds rearrange, and a portion of the mAb eventually achieves a more compact conformation, IgG2(B), in which both Fab arms are disulphide linked to the hinge (Figure 2). Using single cys-ser substitutions to “lock” IgG2 into IgG2(A)-like or IgG2(B)-like conformations,^69,71  ChiLob 7/4 IgG2B was found to be highly agonistic, whereas IgG2A was nonagonistic or even antagonistic.⁶⁰  The more compact conformation of IgG2B may promote close packing of TNFR molecules, already present as preformed dimers or trimers,^72,73  initiating signaling. Thus, fine differences in structure induced by single amino acid changes can have profound effects on mAb activity. A key aspect of future work will be to determine whether FcγR-dependent vs -independent mechanisms of agonistic activity are associated with different levels of toxicity and therapeutic efficacy.

The precise mechanisms through which a particular mAb provides therapy are not necessarily clear. CD40 mAbs, for example, can promote tumor rejection through immune stimulation; however, they can also act as direct targeting agents to mediate lymphoma cell depletion or apoptosis or even as antagonistic agents to reduce inflammation.⁷⁴  Similarly, rituximab (anti-CD20) and trastuzumab (anti-HER2) may have therapeutic benefit by promoting target cell death or by blocking receptor function, respectively.^75,76  The optimal FcγR interactions and therefore isotype for each of these mechanisms are likely to differ. In some cases, it may be advantageous to combine different functions. A good example is the bi-functional nature of immunomodulatory mAbs, which potentially deliver immune agonism or blockade via antigen engagement while simultaneously interacting with FcγR-expressing cells to promote target cell deletion. The latter may be advantageous (deletion of Tregs) or deleterious (deletion of effectors). Whether these apparently competitive functions can be combined to good effect in the same mAb, eg, IgG1/2 chimeras that contain an IgG2 hinge/CH1 with an FcγR binding IgG1 Fc,⁶⁰  remains to be seen. Combinations of agents can also be beneficial, such as the use of agonistic mAb to potentiate the effects of direct targeting agents, through activation of immune effector cells involved in target cell clearance.¹⁹  How potential competition for FcγR binding may influence the dynamics of these relationships is unclear. However, defining the role of FcγR is crucial to better understand mechanisms of action and to define effective combination strategies.

In summary, appropriate isotype selection is crucial for therapeutic mAb activity not only because of its influence on FcγR interactions but also through subtle effects on mAb structure that, in the case of agonistic mAbs, can have profound effects on function. Intricate knowledge of the mechanisms by which mAbs deliver therapy will lead us toward the goal of designing agents capable of delivering curative responses in the majority of cancer patients.

This work was supported by grants from Cancer Research UK and Bloodwise and an EU Framework HEALTH-2013-INNOVATION grant.

Contribution: S.A.B. and A.L.W. wrote the manuscript; and M.J.G. edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Ann L. White, Antibody and Vaccine Group, Cancer Sciences Unit MP88, General Hospital, Tremona Rd, Southampton SO16 8AA, United Kingdom; e-mail: a.l.white@soton.ac.uk.