Download Pharmacokinetics of Antibodies: Enhanced Half-Life through FcRn Interaction and more Slides Engineering in PDF only on Docsity!
Introduction: Antibody Structure and Function
Arvind Rajpal, Pavel Strop, Yik Andy Yeung, Javier Chaparro-Riggers, and Jaume Pons
Introduction to Antibodies
Antibodies, a central part of humoral immunity, have increasingly become a dominant class of biotherapeutics in clinical development and are approved for use in patients. As with any successful endeavor, the history of monoclonal antibody therapeutics benefited from the pioneering work of many, such as Paul Ehrlich who in the late nineteenth century demonstrated that serum components had the ability to protect the host by “passive vaccination” [1], the seminal invention of monoclonal antibody generation using hybridoma technology by Kohler and Milstein [2], and the advent of recombinant technologies that sought to reduce the murine content in therapeutic antibodies [3]. During the process of generation of humoral immunity, the B-cell receptor (BCR) is formed by recombination between variable (V), diversity (D), and joining (J) exons, which define the antigen recognition element. This is combined with an immunoglobulin (Ig) constant domain element (m for IgM, d for IgD, c for IgG (gamma immunoglobulin), a for IgA, and e for IgE) that defines the isotype of the molecule. Sequences for these V, D, J, and constant domain genes for disparate organisms can be found through the International ImMunoGeneTics Information System^1 [4]. The different Ig subtypes are presented at different points during B-cell maturation. For instance, all naïve B cells express IgM and IgD, with IgM being the first secreted molecule. As the B cells mature and undergo class switching, a majority of them secrete either IgG or IgA, which are the most abundant class of Ig in plasma. Characteristics like high neutralizing and recruitment of effector mechanisms, high affinity, and long resident half-life in plasma make the IgG isotype an ideal candidate for generation of therapeutic antibodies. Within the IgG isotype, there are four subtypes (IgG1–IgG4) with differing properties (Table 1.1). Most of the currently marketed IgGs are of the subtype IgG1 (Table 1.2).
1
Therapeutic Fc-Fusion Proteins, First Edition. Edited by Steven M. Chamow, Thomas Ryll, Henry B. Lowman, and Deborah Farson. Ó 2014 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2014 by Wiley-VCH Verlag GmbH & Co. KGaA.
Table 1.1 Subtype properties.
Property IgG1 IgG2 IgG3 IgG Heavy chain constant gene c 1 c 2 c 3 c 4 Approximate molecular weight (kDa) 150 150 170 150 Mean serum level (mg/ml) 9 3 1 0. Half-life in serum (days) 21 21 7 21 ADCC þ þ þ/ CDC þþ þ þþþ Number of disulfides in hinge 2 4 11 2 Number of amino acids in hinge 15 12 62 12 Gm allotypes 4 1 13 Protein A binding þþþ þþþ þ þþþ Protein G binding þþþ þþþ þþþ þþþ Abbreviations: ADCC, antibody-dependent cellular cytotoxicity; CDC, complement-dependent cytotoxicity.
Table 1.2 Marketed antibodies and antibody derivatives by target.
Trade name International non- proprietary name
Target Type Indication
Benlysta^1 Belimumab BLyS Human IgG1l SLE Soliris^1 Eculizumab C5 Humanized IgG2/4 PNH Raptiva^1 Efalizumab CD11a Humanized IgG1k Psoriasis Amevive^1 Alefacept CD2 CD2-binding domain of LFA3---IgG1 Fc fusion
Psoriasis
Rituxan^1 Rituximab CD20 Chimeric IgG1k NHL, CLL, RA, GPA/MPA Zevalin^1 Ibritumomab tiuxetan
CD20 Murine IgG1k---Y90/In conjugate
NHL
Bexxar^1 Tositumomab-I131 CD20 Murine IgG2al---I conjugate
NHL
Arzerra^1 Ofatumumab CD20 Human IgG1k CLL Orthoclone- OKT3^1
Muromonab-CD3 CD3 Murine IgG2a Transplant rejection Adcetris^1 Brentuximab vedotin
CD30 Chimeric IgG1k- conjugated MMAE
Hodgkin’s lymphoma Mylotarg^1 Gemtuzumab ozogamicin
CD33 Humanized IgG4k--- calicheamicin conjugate
Leukemia
Campath- 1H^1
Alemtuzumab CD52 Humanized IgG1k Leukemia
Orencia^1 Abatacept CD80/ CD
CTLA4---IgG1 Fc fusion RA
Nulojix^1 Belatacept CD80/ CD
CTLA4---IgG1 Fc fusion Transplant rejection Yervoy^1 Ipilimumab CTLA4 Human IgG1k Metastatic melanoma
(^2) 1 Introduction: Antibody Structure and Function
The ability of antibodies to recognize their antigens with exquisite specificity and high affinity makes them an attractive class of molecules to bind extracellular targets and generate a desired pharmacological effect. Antibodies also benefit from their ability to harness an active salvage pathway, mediated by the neonatal Fc receptor (FcRn), thereby enhancing their pharmacokinetic (PK) life span and mitigating the need for frequent dosing. The antibodies and antibody derivatives approved in the United States and the European Union (Table 1.2) span a wide range of therapeutic areas, including oncology, autoimmunity, ophthalmology, and transplant rejection. They also harness disparate modes of action like blockade of ligand binding and subsequent signaling, and receptor and signal activation, which target effector functions (antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC)), and delivery of cytotoxic payload. Antibodies are generated by the assembly of two heavy chains and two light chains to produce two antigen-binding sites and a single constant domain region (Figure 1.1, panel a). The constant domain sequence in the heavy chain designates the subtype (Table 1.1). The light chains can belong to two families (l and k), with most of the currently marketed antibodies belonging to the k family. The antigen-binding regions can be derived by proteolytic cleavage of the antibody to generate antigen-binding fragments (Fab) and the constant fragment (Fc, also known as the fragment of crystallization). The Fab comprises the variable regions (variable heavy (VH) [11] and variable light (VL)) and constant regions (CH 1 and Ck/Cl). Within these variable regions reside loops called complementarity determining regions (CDRs) responsible for direct interaction with the antigen (Figure 1.1, panel b). Because of the significant variability in the number of amino acids in these CDRs, there are multiple numbering schemes for the variable domains [12,13] but only one widely used numbering scheme for the constant domain (including portions of the CH 1, hinge, and the Fc) called the EU numbering system [14]. There are two general methods to generate antibodies in the laboratory. The first utilizes the traditional methodology employing immunization followed by recovery of functional clones either by hybridoma technology or, more recently, by recombinant cloning of variable domains from previously isolated B cells displaying and expressing the desired antigen-binding characteristics. There are several variations of these approaches. The first approach includes the immuniza- tion of transgenic animals expressing subsets of the human Ig repertoire (see review by Lonberg [15]) and isolation of rare B-cell clones from humans exposed to specific antigens of interest [16]. The second approach requires selecting from a large in vitro displayed repertoire either amplified from natural sources (i.e., human peripheral blood lymphocytes in Ref. [17]) or designed synthetically to reflect natural and/or desired properties in the binding sites of antibodies [18,19]. This approach requires the use of a genotype–phenotype linkage strategy, such as phage or yeast display, which allows for the recovery of genes for antibodies displaying appropriate binding characteristics for the antigen.
(^4) 1 Introduction: Antibody Structure and Function
Figure 1.1 Structureand featuresofthe IgGand its interactions. (a) The structure of a full-length IgG is shown in ribbon representation with transparent molecular surface. One heavy chain is shown in blue and one light chain in magenta. The other heavy chain and light chain are shown in gray for clarity. In this orientation, two Fab domains sit on top of the Fc domain and are connected in the middle by the hinge region. The Fab domain is composed of the heavy chain V (^) H and C (^) H1 domains and the light chain VL and C (^) L domains---Protein Data Bank (PDB) [5] code 1HZH [6]. (b) Each variable domain contains three variable loops (L1---L3 on light chain and H1---H3 on heavy chain) that make up the antigen-binding site---PDB code 1HZH [6]. (c) The Fc region is composed of the dimer of C (^) H2 and C (^) H3 domains. The C (^) H3 domains form a tight interaction while the C (^) H2 domains interact throughproteinprotein,proteincarbohydrate,
and carbohydratecarbohydrate contacts PDB code 1HZH [6]. (d) The hinge region is composed of a flexible region covalently tied togetherthroughdisulfidebridges.Structuresof the FccRIIIa and FccRIIa bound to the Fc are shown. The structuresreveal that bothreceptors bind to the C (^) H2 domain near the hinge and carbohydrates and upon their binding create an asymmetry such that the second FccR is unable to bind. In this panel, FccRIII is shown in green, and the FccRII is shown in purple---PDB codes 3RY6 [7] and 1T83 [8]. (e) The crystal structure of the complex between the Fc and FcRn reveals that FcRn binds between the C (^) H2 and C (^) H 3 domains in the Fc. FcRn chains are shown in red and orangePDB code 1FRT [9]. (f) Interestingly, the same region also binds to bacterial Protein A commonly used for purificationPDB code 1FC2 [10].
1.1 Introduction to Antibodies 5
Protein G through charged and polar interactions, Proteins A and G bind to a similar site on Fc domain and compete with each other (Figure 1.1, panel f). Interestingly, the binding occurs between the C (^) H2 and C (^) H3 domains of the Fc and largely overlaps with the FcRn binding site. ADCC function is mediated by the interaction of the Fc region with Fcc receptors (FccRs). Biochemical data and structures of Fc in complex with FccRIII and FccRII reveal that the FccRs bind to the combination of the Fc CH2 domain and the lower hinge region (Figure 1.1, panel d) [7,8,25]. Members of the Fcc family have been found to bind to the same region of Fc [20,26,27] and form a 1 : 1 asymmetric complex where one FccR interacts with the dimer of Fc. The binding of one FccRIII to Fc induces asymmetry in the Fc region and prevents a second interaction. While the detailed structural understanding is not available for the Fc–C1q interaction, biochemical data suggest that C1q binds mainly to the CH 2 domain with an overlapping, but nonidentical, binding site of FccRIII [28]. The
IgG1 IgG
IgG
IgG
IgG
IgG
IgG1 Ig 2
IgG
Ig 1
(a) (b)
(c) (d)
Figure 1.2 Interchain disulfide topology in human IgG subclasses. Only H---H hinge and H---L chain disulfides are shown. (a) IgG1, (b), IgG2, (c) IgG3, and (d) IgG4.
1.2 General Domain and Structure of IgG 7
details of the interaction between the Fc and Fcc receptors, as well as the engineering of effector function, are further discussed in Section 1.4.2.1.
1.2.1.2 Fc Glycosylation The Fc region of IgG has a conserved glycosylation site in the C (^) H 2 domain at position N297 (Figure 1.1, panel c). Glycosylation of the C (^) H 2 domain is important in achieving optimal effector function [29] and complement activation; it also contributes to overall IgG stability [30]. Antibodies purified from human serum have been found to contain heterogeneous oligosacchar- ides where each C (^) H 2 domain can contain one of many potential glycans [31]. Therapeutic Fc-containing proteins that are expressed in Chinese hamster ovary (CHO) or human embryo kidney 293 (HEK293) cells typically contain a mixture of glycoforms, with G0F being the most abundant, followed by G1F and G2F [32,33]. The attachment of the glycans at position Asn297 in the C (^) H 2 domain positions the carbohydrates to interact with each other and to form a part of the Fc dimer interface. Because of carbohydrate sequestration into the space between the two C (^) H 2 domains and significant carbohydrate–carbohy- drate and carbohydrate–protein contacts, the carbohydrates in the Fc crystal structures are relatively well ordered. The glycosylation of the Fc has been found to influence biological activity as well as stability of IgGs [34,35]. The removal of the core fucose enhances ADCC activation of FccRIIIa on natural killer (NK) cells but does not change the binding of FccRI or C1q [36]. Increased ADCC has also been observed with the presence of bisecting N-acetylglucosamine in the context of fucosylated IgG, although the effect appears to be smaller than removal of the core fucose [37]. Sialylated IgGs have been suggested to enhance anti- inflammatory properties [38]; however, more work is needed to understand this effect and potential mechanism.
1.2.1.3 Hinge and Interchain Disulfide Bonds The hinge region of human IgGs (IgG1, IgG2, and IgG4) differs between the subtypes both in the hinge length (12–15 residues) and in number of disulfides linking the two heavy chains together (2–4 residues) (Figure 1.2). In addition, the position of the light chain–heavy chain linkage differs among the human IgG subtypes (Figure 1.2). In human IgG1, two disulfides link the heavy chains together while human IgG2 contains four disulfides and a shorter hinge. The presence of an increased number of disulfides as well as a shorter hinge likely decreases the flexibility of hIgG2 Fab regions relative to hIgG1. The hinge can have a profound impact on antibody properties. For example, the sequence in the hinge near the disulfides has been found to be important in the ability of IgG4s to exchange half molecules in vivo and under certain conditions in vitro [39,40]. The absence of one of the proline residues in the hinge of IgG4 coupled with substitution in the CH 3 domain allows IgG4 to form half-antibodies and form bispecific antibodies by exchanging with other IgG4s (Figure 1.2).
(^8) 1 Introduction: Antibody Structure and Function
Table 1.3 (Continued)
Mutation(s) (EU numbering)
IgG isotype
Target antigen
FcRn affinity increase at pH 6. (fold)
Serum half- life (fold of WT)
Clearance (Fold of WT)
Source
M428L/ N434S
IgG1 a-VEGF 11 (human) b)^ 3.2 (cyno) 0.32 (cyno)
[53]
V259I/ V308F
IgG1 a-VEGF 6 (human) b)^ 1.7 (cyno) 0.63 (cyno)
[53]
M252Y/ S254T/ T256E
IgG1 a-VEGF 7 (human) b)^ 2.5 (cyno) 0.42 (cyno)
[53]
V259I/ V308F/ M428L
IgG1 a-VEGF 20 (human) b)^ 2.6 (cyno) 0.39 (cyno)
[53]
M428L/ N434S
IgG1 a-EGFR 11 (human) b)^ 3.1 (cyno) 0.31 (cyno)
[53]
N434H IgG1 a-VEGF 4 (human) b)^ 1.6 (cyno) 0.62 (cyno)
[54]
5 (cyno)b) T307Q/ N434A
IgG1 a-VEGF 18 (human) b)^ 2.2 (cyno) 0.52 (cyno)
[54]
10 (cyno)b) T307Q/ N434S
IgG1 a-VEGF 10 (human) b)^ 2.0 (cyno) 0.49 (cyno)
[54]
12 (cyno)b) T307Q/ E380A/ N434A
IgG1 a-VEGF 13 (human) b)^ 1.9 (cyno) 0.57 (cyno)
[54]
15 (cyno)b) V308P/ N434A
IgG1 a-VEGF 26 (human) b)^ 1.8 (cyno) 0.57 (cyno)
[54]
34 (cyno)b) N434H IgG (N297A)
CD4 3 (human) b)^ N/A 0.50 (baboon)
[55]
3 (baboon) b) V308P IgG4 5 unknown targets
40---390 (cyno)c)^ 2.0---3.3 (cyno)
0.22---0.74 (cyno)
[56]
T250Q/ M428L
IgG4 5 unknown targets
11---110 (cyno)c)^ 0.9---2.6 (cyno)
0.31---0.89 (cyno)
[56]
Abbreviations: EGFR, endothelial cell growth factor receptor; FcRn, neonatal Fc receptor; HBV, hepatitis B virus; N/A: not available; RSV, respiratory syncytial virus; TNF, tumor necrosis factor; VEGF, vascular endothelial cell growth factor. a) IC 50 binding ratio performed on FcRn-transfected cells. b) Monovalent interaction: injecting FcRn over surface-conjugated antibodies. c) Bivalent interaction: injecting antibodies over surface-conjugated FcRn. d) No statistically significant difference.
(^10) 1 Introduction: Antibody Structure and Function
FcRn is also known as Fc receptor-protection (FcRp) or Fc receptor-Brambell (FcRB) [57] after F.W. Rogers Brambell who first described it and its protective function. In the late 1950s, Brambell proposed that a saturable receptor existed for transporting IgG from mothers to infants through the yolk sacs and intes- tines [57,58]. Observing the similarity between passive transmission and catabo- lism of IgG, Brambell later postulated that a similar or identical receptor system was responsible for the protection of IgG from catabolism [59]. It was not until 1989 that FcRn was finally cloned from the epithelial cells of the small intestine of a rat and confirmed to carry out important functions of both transporting IgG across cellular barrier (transcytosis) [58] and rescuing IgG from catalytic degradation (homeostasis) [60,61] (Figure 1.3). Even though transcytosis and homeostasis processes are both mediated by FcRn, the respective processes are regulated differently inside cells [62]. Studies have also shown that FcRn is involved in the internalization [63], presentation, and cross-presentation of immune complex onto the major histocompatibility complex (MHC) [64–66]. FcRn, which is structurally homologous to MHC class I molecules, is a heterodimer consisting of a 50 kDa transmembrane a chain and 12 kDa b2-microglobulin (b2m) (Figure 1.1, panel e), which is required for the FcRn expression. FcRn mostly resides intracellularly, but can be exposed to the extracellular environment through vesicle trafficking [9,67]. Unlike MHC molecules, the counterpart of the MHC peptide-binding groove in FcRn is occluded by its own residues, so FcRn is incapable of binding peptides and hence does not present peptide–MHC complex to T cells [9,67]. FcRn can simultaneously bind both IgG and albumin, but the binding stoichiometries are different, with a 2 : 1 ratio for FcRn– IgG and a 1 : 1 ratio for FcRn–albumin [21,68]. It was previously shown that bivalent binding of an IgG to FcRn enhances the rate of IgG recycling inside a cell [69]. Although a crystal complex structure of human FcRn–Fc is still not available, the major contact residues in the human complex can be deduced from the crystal structure of a rat FcRn–Fc complex [21,70] and site-directed mutagenesis studies [71–74]. At the protein level, FcRn binds to the Fc portion of IgG at the CH 2 – qC (^) H 3 interface, which is distinct from the binding sites of FccR and C1q component of complement (Figure 1.1, panel e) [21,39,67,72]. It is worth noting that Fc alone can mediate binding to FcRn, so in addition to IgG, FcRn can protect Fc-fusion proteins from degradation. Molecularly, major contact residues of the human FcRn are Glu115, Asp130, Trp131, Glu133, and Leu135 on the a chain and Ile1 on the b2m. On the Fc side, residues Ile253, Ser254, His310, His435, and Tyr436 are important for the interaction, as alanine substitution at these positions results in a significant reduction in binding to FcRn [72]. Meanwhile, FcRn has been found to bind albumin around His166, opposite from the Fc binding region [75], which may explain why FcRn can simultaneously bind both IgGs and albumins [71]. FcRn protects IgGs from catabolism through a pH-dependent binding mechan- ism. IgGs bind FcRn with high affinity at acidic pH (pH 6 – 6.5). As the pH is raised to neutral (pH 7.4), the binding affinity drops considerably. The pH-dependent interaction is mainly attributed to the titration of histidine residues
1.3 The Neonatal Fc Receptor 11
(H310 and H433) on human Fc and their subsequent interaction with acidic residues on FcRn [76]. Specifically, pinocytosed IgGs are captured by FcRn in the acidic endosome [77], recycled to the cell surface [78], and then released into the circulation at a physiological pH of 7.4 [79]. IgGs that are not bound by FcRn are transported to the lysosome and degraded [77]. Studies in knockout mice have demonstrated that the serum half-lives of IgGs and albumin in FcRn- or b2m- deficient mice are greatly reduced [45,60]. It has also been observed in familial hypercatabolic hypoproteinemia patients that their low levels of serum IgGs and albumin were caused by the reduction of FcRn expression, resulting from b2m deficiency [61]. Functional FcRn expression has been reported in a variety of tissues and cells such as vascular endothelium [80], hematopoietic cells (monocytes, macrophages, dendritic cells, polymorphonuclear leukocytes, and B cells) [81], intestinal epithelium [82], brain and choroids plexus endothelium [83,84], podocytes [85], placental endothelium [86,87], lung epithelium [88], and vaginal epithelium [64]. FcRn has been shown in studies to either recycle IgG or transport IgG across the cellular barriers. As FcRn is ubiquitously expressed in multiple cell types like endothelial, epithelial, and hematopoietic cells at various body sites, it is of interest to understand which cell types and tissues are responsible for the IgG recycling. First, Akilesh et al. [89] used bone marrow chimeric mice with FcRn-deficient and FcRn-sufficient cells to demonstrate that a significant fraction of IgG protection is mediated by hematopoietic cells. Later, by using a mouse strain in which FcRn is conditionally deleted, Montoyo et al. [90] were able to show that transgenic mice without FcRn expression in endothelial and/or hematopoietic cells did not protect IgGs from degradation, indicating that both hematopoietic cells and endothelial cells are the primary sites for maintaining IgG homeostasis in mice.
1.3. Species Difference in FcRn
FcRns of multiple mammalian species have been cloned [91], and functional expression of FcRn has been reported in mammals like the rat, mouse, rabbit, sheep, bovine, possum, horse, pig, nonhuman primate, and human [57,92–95]. An FcRn orthologous molecule has also been described in chicken [96]. In terms of sequence homology of FcRn in difference species, primate FcRn is the closest to human FcRn. For example, cynomolgus monkey FcRn has a 96% sequence identity (98% similarity) to human FcRn and the two receptors bind human IgG with similar affinity. Meanwhile, rat and mouse FcRns are 91% identical but have only 69% identity (80% similarity) and 70% identity (80% similarity) with human FcRn, respectively. Molecularly, rodent FcRn contains four N-linked glycans while human FcRn has only one N-linked glycan [73]. In rodent FcRn, carbohydrate residues in the glycan at Asn 128 participate in binding mouse Fc [73]. Along with glycan differences, the amino acid differences between rodent and human FcRns give rise to different cross-binding specificity and affinity between rodent and human FcRn–Fc [74]. For example, mouse FcRn, which is considered the most
1.3 The Neonatal Fc Receptor 13
promiscuous, was able to bind IgGs of human, rabbit, bovine, rat, sheep, and guinea pig, whereas human FcRn was unable to bind mouse and rat IgGs, showing restricted binding to only human and rabbit IgGs [97]. In addition, human IgGs have higher affinity toward murine FcRn than murine IgG [98]. Because of the binding affinity and specificity differences, it has been challenging to use rodent models to evaluate PK of human IgG Fc variants that were engineered for human FcRn binding (see more detailed discussion in Section 1.3.3) [48].
1.3. Engineering to Modulate Pharmacokinetics
1.3.3.1 Fc Engineering As FcRn interaction is responsible for the PK of IgG and Fc-fusion therapeutics, engineering the FcRn–Fc interaction is one of the methods for modifying the PK and pharmacodynamics (PD) of an IgG or Fc fusion [22,99]. The one area that is most researched and attractive to the pharmaceutical industry is the attempt to increase the serum half-life of the therapeutic IgG and Fc fusions. Advantages of increased serum half-life include increasing transcytosis to maximize drug delivery at specific tissues, minimizing adverse reactions caused by high doses, decreasing production cost, and reducing frequency of injection, thereby potentially increasing the compliance of patients taking the drugs. A number of studies have demonstrated that balancing the affinity/kinetic improvements of Fc variants at pH 6.0 and pH 7.4 is critical for engineering variants with improved half-life. Increasing affinity at pH 6.0 can increase the capture of IgG variants by FcRn in the endosome, thereby reducing degradation and increasing the recycling chances. However, affinity increases of variants at pH 6.0 and pH 7.4 are often coupled, so a substantial increase in a variant’s FcRn affinity at pH 6.0 will lead to an undesirable increase in affinity at pH 7.4 [50,51]. High levels of binding at pH 7.4 hinder the release of an FcRn-bound IgG variant back into circulation and increase the binding of circulating IgGs to the cell surface-expressing FcRn, effectively accelerating the clearance of IgG and canceling out the benefit of increased affinity at acidic pH. Therefore, FcRn binding affinities/kinetics at both acidic and physiologic pHs are important determinants to balance the PK engineering of an IgG or Fc fusion. The first proof-of-concept study of engineering Fc for improved half-life was performed by Ghetie et al. [100], who observed that an engineered murine Fc variant (T252L/T254S/T254F) with a threefold increase in murine FcRn affinity at pH 6.0 exhibited approximately a 30–60% extension in serum half-life in mice. Subsequent studies have identified and validated various favorable mutations on human IgG1 Fc residues, such as Thr250, Met252, Thr254, Ser256, Thr307, Val308, Glu380, Met428, and Asn434 [49,54,56,72,101]. These mutations can be combined synergistically to give Fc variants of different affinities, yielding half-life improvements of varying extents (Table 1.3). One important aspect of engineering the half-lives of IgGs is the use of suitable preclinical animal models to evaluate the variant’s half-life in vivo and predict the
(^14) 1 Introduction: Antibody Structure and Function
CD40) and abundantly synthesized soluble antigens like PCSK9 and IgE. In addition to studies that have shown that engineering optimized FcRn binding can extend the serum half-life of antibodies [53], other work has demonstrated that the half-life of some antibodies can be extended by engineering their interactions with antigens for increased pH dependency [106–108]. Such pH-sensitive antibodies would bind tightly to the receptor/antigen in the plasma (pH 7.4); once trafficked inside cells, the antibody–antigen complex would dissociate in the acidic endosome and the antibody would recycle back into the circulation by FcRn, effectively avoiding lysosomal degradation and allowing the same antibody to undergo further binding cycles. Such pH-engineered antibodies were shown to have improved half- lives in vivo compared to their parental counterparts, particularly at lower doses [109,110]. Besides Fc–FcRn and antibody–antigen interactions, other biochemical factors affect the PK of an antibody, such as molecular size, molecular charges, stability of the IgG, and glycosylation patterns, along with route of delivery [111]. A recent study suggested that engineering the isoelectric point (pI) of the antibody variable region could offer an alternative way to improve PK [107]. The typical pI of an antibody is 8–9. It has been shown that antibodies with higher pI values tend to have faster clearance in vivo [95]. This reduced PK was hypothesized to be due to the unfavorable electrostatic interaction between the anionic cell membrane and the positive charge of an antibody. Igawa et al. [107] engineered variants with pI values 1 – 2 units lower than the wild type and showed that these pI variants displayed longer half-lives and reduced clearance in mice. This pI-lowering effect can also be achieved using conjugation of charge moieties [112]. Overall, there are multiple methods for engineering the PK of an antibody. However, each antibody and its target antigen have very different biochemical, biophysical, and cellular properties on which antibody PK is highly dependent. Therefore, a one-size-fits-all engineering method to improve the PK of every antibody may not exist. Choosing the appropriate PK engineering method depends on understanding the properties and interactions between an antibody and its antigen, and the desired antibody exposure and distribution levels.
Introduction to FccR- and Complement-Mediated Effector Functions
Antibodies are multifunctional molecules resulting from adaptive immunity. Part of the variable region binds to the target antigen and different parts of the constant region are responsible for diverse effector functions. This enables the antibody to bridge between the target antigen and the body’s immune system. The antibody is able to recruit cellular and noncellular immune responses, which interact with each other in a complex way. In this chapter, the effector functions leading to cell lysis, phagocytosis, immune activation, and T-cell activation properties, which are essential for therapeutic antibodies against tumor cells and for infectious diseases, will be discussed.
(^16) 1 Introduction: Antibody Structure and Function
Cell Lysis and Phagocytosis Mediation
Cell lysis can be mediated in two ways: (1) ADC mediated by FccR-expressing NK cells. Immune complex-mediated activation of the NK cells by FccR results in cytokine release such as interferon-c (IFNc) and cytotoxic granules. These perforin- and granzyme-containing granules enter the target cell and induce cell death through apoptosis. ADCC has also been induced by monocytes and eosinophils [113]. (2) CDC is initiated when C1q, the initiating component of the classical complement pathway, is fixed to the Fc portion of target-bound antibodies. Once C1q binds to the antibody Fc, the complement cascade is activated and leads to a membrane-spanning, multiprotein pore complex. This pore is called the membrane attack complex (MAC) and can lyse the cells (for more detailed information, see Ref. [114]). Phagocytosis can be mediated in two ways: (1) Antibody dependent cellular phagocytosis (ADCP) is mediated by FccR-expressing monocytes/macrophages, neutrophils, and dendritic cells (DCs). After activation of the phagocytes by FccR, the antibody-coated cell is engulfed and degraded once the phagosome fuses to the lysosome [115]. (2) Complement-dependent cellular cytotoxicity (CDCC) can be initiated when opsonic membrane-bound components (C3b, iC3b, C4b) of the complement cascade interact with complement receptors (CR1, CR3, CR4, CRIg) on immune cells. This interaction can result in phagocytosis by monocytes/ macrophages, neutrophils, and DCs, while cell lysis is mediated by NK cells [114]. CDCC only occurs as a response to yeast- and fungi-produced proteins because CR3 requires dual ligation to iC3b and the cell wall b-glucan, which is only found on yeast or fungi [116].
FccR-Mediated Effector Functions
1.4.2.1 FccR Biology Antibodies interact with effector cells via binding of their Fc region to cellular FccR. In humans, six known members in three subgroups of FccR (FccRI, FccRIIabc, FccRIIIab) exist, with differences in expression, signaling, and affinities for the IgG subclasses. Four activating receptors with cytoplasmic immunoreceptor tyrosine-based activation motif (ITAM) exist, which are genetically encoded or associated. The inhibitory FccRIIb contains an immunoreceptor tyrosine-based inhibition motif (ITIM), which is genetically encoded in the same molecule. The glycosyl phosphatidyl inositol (GPI)-anchored FccRIIIb does not signal. ITAM/ITIM signaling is based on receptor clustering induced by binding to immune complexes. No clustering occurs for FccRII/III binding to monomeric IgG because of the low affinity. The high affinity of FccRI can poorly differentiate between monomeric IgG and immune complexes. FccRIIa and FccRIIIa have allelic forms, which can impact their affinity to different IgG subclasses (Table 1.4) and ultimately their biological function. FccRIIa contains a polymorphism at position
1.4 Introduction to FccR- and Complement-Mediated Effector Functions 17
T cells. The most relevant for FccR-induced effector functions of therapeutic antibodies are NK cells, monocytes/macrophages, DCs, and neutrophils. NK cells are unique because in most people they typically only express the activating receptor FccRIIIa while NK cells in some individuals express the activating FccRIIc receptor. The main FccR-induced functions of NK cells are the cytolysis of target cells through lytic granule release (granzyme, perforin), apoptosis via secretion of tumor necrosis factor (TNF) family ligands, and production of cytokines such as IFNc. A series of activating receptors, like NKG2D, and inhibitory receptors of the killer Ig-like receptor (KIR) family regulate the NK cell activity. On normal cells, the killing is suppressed because the KIRs interact with autologous MHC class I molecules. Killing is induced if matching MHC molecules are missing. Antibody-coated target cells can be killed by FccRIIIa engagement because KIR inhibition is partially overridden. Monocytes/macrophages, neutrophils, and DCs (myeloid cell lineage) have overlapping FccR expression profiles and all of them express FccRIIa and FccRIIb. FccRI and FccRIIIa are also expressed depending on their source and activation state by monocytes/macrophages and DCs; for example, after G-CSF activation neutrophils express FccRIIIb rather than FccRI and FccRIIIa. Upon FccR engagement, macrophages and neutrophils can phagocytose opsonized target cells. They lyse target cells by releasing cytolytic granules or inducing apoptosis via release of reactive nitrogen and oxygen intermediates. Besides target-cell destruction, macrophages and DCs are also professional antigen-presenting cells and can present peptides of target cell antigens on MHC class II to CD4þ^ T cells. DCs can additionally present peptides of target antigens on MHC class I and activate cytotoxic T cells (CD8þ) by cross-priming, which can lead to long-lasting adaptive antitumor immunity and long-term remission. This was observed for the anti-CD20 antibody rituximab [119].
1.4.2.3 Therapeutic Relevancy The most compelling data are obtained from associations of clinical outcomes with functionally relevant receptor polymorphisms. FccR polymorphisms have been associated with infectious and autoimmune disease, or with disease sever- ity [120,121]. In humans, the FccRIIa-H131 allotype is known to interact efficiently with complex human IgG2, whereas the FccRIIa-R131 allotype does so only poorly. This polymorphism may therefore have implications for IgG2-mediated phagocy- tosis of encapsulated bacteria and susceptibility to bacterial infections. FccRIIa- R131 is associated with greater susceptibility to infectious diseases [120,122]. Polymorphism association studies have been applied to cancer therapies using monoclonal antibodies [115]. Significant response differences between high-affinity V158 and low-affinity F158 FccRIIIa alleles have been observed with rituximab (anti-CD20) for the treatment of follicular non-Hodgkin’s lymphoma [123], Waldenstr€om’s macroglobulinemia [124], and in two out of three studies in diffuse large B-cell lymphoma [125–127]. No FccR polymorphism was observed for rituximab treatment in chronic lymphocytic leukemia (CLL) [128,129]. Apart from the clinical relevancy of FccRIIa for infectious diseases [120,122], the clinical outcome of murine IgG3 anti-GD2 antibody 3F8 in the solid tumor neuroblastoma
1.4 Introduction to FccR- and Complement-Mediated Effector Functions 19
is correlated with FccRIIa-R131 [130]. Murine IgG3 has preferential binding of human FccRIIa-R131 over H131, which are only expressed on neutrophils, macrophages, and DCs, while NK cells do not express FccRIIa. In a following clinical study, 3F8 was administered in combination with granulocyte macrophage colony-stimulating factor (GM-CSF) and suggested that granulocytes (e.g., neutro- phils) are relevant effector cells [131]. Future studies will perhaps bring a clearer picture of the importance of activating FccR since many ADCC-enhanced therapeutic antibodies are in the clinic [132,133].
Complement
1.4.3.1 C1q Biology The lectin pathway and alternative pathways are generally activated by pathogens and not by cell-bound antibodies and are therefore not further discussed. The classical pathway of the complement system can be activated following binding of monoclonal antibodies to tumor cells. This pathway can be initiated on binding of the C1q component of the C1 complex to the Fc of the antibody on the cell membrane. The initiation of the pathway by the antibody is dependent on the subclass of bound antibody (IgG3 > IgG1 IgG2 > IgG4), the membrane proxi- mity of the antibody epitope, the membrane protein number per cell, and the affinity of the antibody, which leads to simultaneous binding of one C1 complex to at least two Fcs [116,134]. The initiation of the pathway results in the deposition of C3b, which is subsequently converted to iC3b and can lead to the formation of the cytolytic MAC. This complex results in CDC, which is a noncellular activity. The initiation of the pathway by antibodies also leads to the deposition of opsonic proteins (C3b, iC3b, and C4b) on the cell surface, which can lead to two cellular complement activities. First, cell surface-bound opsonins can bind the complement receptors (CR1, CR3, CR4, and CRIg) on phagocytes and NK cells, which can trigger CDCC. This only occurs in cases where the cell wall b-glucan from yeast or fungi is present. Therefore, tumor cells do not trigger CDCC. Second, opsonin–CR interaction and C5a trigger the enhancement of ADCC. C5a function is a chemoattractant for effector cells and lowers the threshold for FccR activation by upregulating the expression of activating FccR and downregulating the inhibitory FccR. To prevent uncontrolled activation and consumption of complement components, complement activation is tightly regulated by complement regulatory proteins (CRPs). CRPs are present as soluble proteins and as membrane-bound complement regulatory proteins on most cell types [114,116].
1.4.3.2 Therapeutic Relevancy The importance of the complement system for antibodies against infectious diseases has been shown in mouse models [135]. The role of the complement system for anticancer antibodies is not well understood and may be even detrimental. Data from mouse models and clinical trials are contradictory [113,136]. The complement activation of anti-CD20 antibodies was found to correlate with
(^20) 1 Introduction: Antibody Structure and Function
