Chemistry & Biology
Brief Communication IgGA: A ‘‘Cross-Isotype’’ Engineered Human Fc Antibody Domain that Displays Both IgG-like and IgA-like Effector Functions William Kelton,1 Nishant Mehta,2 Wissam Charab,1 Jiwon Lee,1 Chang-han Lee,1 Takaaki Kojima,5 Tae Hyun Kang,2 and George Georgiou1,2,3,4,* 1Department
of Chemical Engineering of Biomedical Engineering 3Section of Molecular Genetics and Microbiology 4Institute for Cellular and Molecular Biology University of Texas at Austin, Austin, TX 78712, USA 5Laboratory of Molecular Biotechnology, Graduate School of Bioagricultural Sciences, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan *Correspondence: [email protected] http://dx.doi.org/10.1016/j.chembiol.2014.10.017 2Department
SUMMARY
All clinically approved antibodies are of the IgG isotype and mediate the clearance of target cells via binding to Fcg receptors and complement (C1q). Even though IgA can elicit powerful cytotoxic action via FcaRI receptor binding, IgA antibodies have not been amenable to therapeutic development. Here, we report the engineering of a ‘‘cross-isotype’’ antibody, IgGA, which combines the effector functions of both IgG and IgA. IgGA binds to FcaRI with an affinity comparable to that of IgA, and to the activating Fcg receptors, FcgRI and FcgRIIa, with high affinity, and displays increased binding to C1q compared to IgG. Unlike trastuzumab-IgG, trastuzumab-IgGA potently activates both neutrophils and macrophages to kill Her2+ cancer cells. Furthermore, IgGA mediates greater complement-dependent cytotoxicity than IgG1 or IgA antibodies. The multitude of IgGA effector functions could be important for therapeutic purposes and highlights the concept of engineering antibodies that combine effector functions from multiple antibody isotypes.
INTRODUCTION Extensive clinical and laboratory evidence has demonstrated that immune effector cytotoxic mechanisms (antibody-dependent cell cytotoxicity [ADCC], antibody-dependent cell phagocytosis [ADCP], and complement-dependent cytotoxicity [CDC]) play a crucial role in the clearance of cancer cells by therapeutic antibodies (Hogarth and Pietersz, 2012). Immunoglobulin (Ig)G immune complexes bind to the activating Fcg receptors (FcgRs), FcgRI (CD64), FcgRIIa (CD32a), and FcgRIIIa (CD16a) that are expressed at varying levels on leukocyte subsets (Bruhns et al., 2009), triggering phosphorylation of the intracellular immunoreceptor tyrosine-based activating motif (ITAM) and, in turn,
downstream signaling events that ultimately result in the destruction of pathogens by ADCC and ADCP. Most leukocytes also express the inhibitory receptor FcgRIIb that signals through an immunoreceptor tyrosine-based inhibitory motif (ITIM) to induce anti-inflammatory responses. It has been proposed that the ratio of binding affinity to the activating FcgRs relative to FcgRIIb (referred to as the activation to inhibition ratio), dictates whether stimulation with immune complexes will trigger inflammatory or anti-inflammatory responses (Nimmerjahn and Ravetch, 2012). The pharmaceutical industry has invested heavily in the engineering of antibodies that display higher affinity or binding selectivity for FcgR and, thus, confer enhanced killing of target cells by ADCC, ADCP, and/or CDC. At present, there are >15 Fc-engineered antibody therapeutics in clinical trials (phase II or III) and two that have been approved as cancer drugs: obinutuzumab/Gazyva (anti-CD20) and mogamulizumab/Poteligeo (antiCCR4) (Beck and Reichert, 2012; Desjarlais and Lazar, 2011). Improvements to the binding affinity, or selectivity, of antibodies for FcgRs have been accomplished either through protein engineering or through glycoengineering via the modification of the critical glycan appended to residue N297 of the IgG Fc domain (Desjarlais and Lazar, 2011; Nimmerjahn and Ravetch, 2012). Examples of engineered Fc domains displaying two or more orders of magnitude improvement in FcgR binding affinity or selectivity have been reported (Chu et al., 2008; Jung et al., 2013). Currently, all antibody therapeutics are of the IgG isotype (predominantly IgG1), which is the most abundant antibody isotype in human serum. These IgG antibodies do not recognize the FcaRI receptor (CD89) that mediates the effector functions of IgA (both for IgA1 and IgA2), the second most abundant antibody isotype in human circulation and the most prevalent in mucosal secretions (Woof and Kerr, 2006). FcaRI is constitutively expressed on many cells of the myeloid lineage, such as neutrophils, macrophages, Kuppfer cells, and eosinophils (van Egmond et al., 1999). Recent studies have clearly demonstrated that engagement of FcaRI by IgA can elicit effective clearance of tumor cells in vitro (Lohse et al., 2012) and in animal models (Boross et al., 2013). This has been shown to occur via macrophage-mediated ADCP and via ADCC by neutrophils. In
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1603
Chemistry & Biology Brief Communication
particular, the FcaRI-dependent activation of neutrophils by IgA has been reported to be far more effective at tumor cell killing than FcgR-mediated activation by IgG antibodies (Dechant et al., 2007). The ability to capitalize on neutrophils for the elimination of pathogenic cells, especially in cancer therapy, is particularly exciting because neutrophils constitute the most abundant leukocyte subset in blood and can infiltrate many solid tumors (Gregory and Houghton, 2011). Unfortunately, the use of IgA antibodies as therapeutics has been hampered by several limitations. (1) The hinge region of IgA is heavily glycosylated, and producing proteins with multiple complex glycans can be problematic for bioprocessing and quality control during drug development (Woof and Kerr, 2006). (2) IgA does not bind to C1q; thus, it cannot mediate CDC through the classical complement pathway (van Egmond et al., 2001). (3) IgA also does not bind to Fcg receptors and, hence, cannot take advantage of the diverse mechanisms for immune cell activation that are pertinent to IgG1 therapeutics. (4) IgA exhibits a shorter circulation half-life than IgG because of a lack of binding to the FcRn receptor, which is important for intracellular recycling. Here, we used a semirational protein engineering strategy to develop a human ‘‘cross-isotype’’ Fc domain, termed IgGA, that displays many of the immunobiological functions of both IgA and IgG isotype antibodies. Not only do engineered IgGA antibodies bind to human FcaRI with an affinity similar to that of IgA antibodies, they also display binding to human FcgRI, FcgRIIa/b, and C1q with an affinity similar to that of IgG1 antibodies. With human neutrophils as effector cells, trastuzumab-IgGA displays much higher ADCC activity compared to clinical-grade trastuzumab (IgG) and ADCC activity comparable to that of trastuzumab formatted as an IgA. Furthermore, trastuzumab-IgGA was much more potent than its IgG counterpart in eliciting phagocytosis (ADCP) of MDA-MB-453 breast cancer cells, which express moderate levels of Her2. Additionally, unlike IgA, which can activate complement only weakly via the alternative and lectin-mediated pathways (Roos et al., 2001), we show that IgGA is at least as effective as IgG1 in mediating CDC. RESULTS AND DISCUSSION The FcaRI-binding site of the IgA1 Fca domain is composed of residues in the AB helix/loop of the CHa12 domain and multiple regions within CHa13 (crystal structure, Protein Data Bank [PDB] ID: 1OW0) (Herr et al., 2003a). We used Ala scanning mutagenesis to identify the relative importance of FcaRI contact residues within the a1 loop of CHa12 (L257, L258), ‘‘noncontact’’ residues within the crystallographically poorly resolved a2 loop of CHa12 (E313, N316, and H317), and certain contacting residues within 4 A˚ of FcaRI in the CHa13 domain: R382 (C strand), E389 (CC0 loop), M433 (F strand), E437 (FG loop), L441 (FG loop), and F443 (G strand) (Figures S1A and S1B available online). Mutant Fca1 domains were expressed in human embryonic kidney 293 (HEK293) cells, and binding to purified FcaRI-GST, also produced in HEK293 cells, was determined by ELISA (Figure S1C). Consistent with earlier reports (Carayannopoulos et al., 1996), an L-Ala substitution at L258 in the a1 loop of the CHa12 domain was found to abrogate FcaRI binding. Further inspection of the Fca L258A protein by nonreducing SDS-PAGE
revealed that this mutation markedly impacts Fca1 assembly. In contrast, an Ala mutation of the adjacent residue (L257) had essentially no effect on Fca1 assembly or FcaRI binding. Mutation of either E313 or N316 in the a2 loop abolished FcaRI binding, possibly through the disruption of hydrogen bonding with Y86 of FcaRI. Within the CHa13 domain, charged residues in the C beta sheet (R382 and E389) were moderately important for binding, while M433, which lies within a hydrophobic binding pocket, was essential for FcaRI binding. However, other residues within the hydrophobic binding pocket (E437, L441, and F443) were more tolerant of Ala substitution. Fc domains comprising the CHa13 fused to CHg12 do not bind to FcaRI (Chintalacharuvu et al., 2001). We constructed a series of chimeric Fc domains using this CHg12-CHa13 fusion as a template in order to preserve important CC0 loop, F strand, FG loop, and G strand contacts identified in the cocrystal structure and by Ala scanning. Because the CHg13 domain is not directly involved in binding to the FcgRs, we reasoned that exchange of this domain would have a minimal effect on FcgR affinity. Both the a2 loop and parts of the a1 loop of CHg12 were exchanged with the respective sequences from CHa12, since these two loops are also important for binding to the receptor, as shown earlier. One construct, which we termed IgGA (Figures 1A and 1B), had a1 loop residues 245–258 (PKPKDTLMISRTPE) of CHg12 and 340–447 (KGQPR. LSPGK) of CHg13 replaced with CHa12 residues 251–263 (PALEDLLLGSEAN) and CHa13 residues 341–450 (SGNTF. KTIDR). In addition, a Gly residue was inserted following CHa12 residue 257 (to ensure that the length of the grafted loop from CHa12 is consistent with that of CHg12). When the IgGA Fc was expressed in HEK293 cells, it was found to assemble efficiently and to bind to both FcaRI and to certain FcgRs by ELISA (data not shown; see surface plasmon resonance [SPR] analysis discussed later). The Fab domains of trastuzumab were fused to the IgGA Fc (trastuzumab-IgGA) or, alternatively, to Fca1 lacking the IgA tailpiece, a C-terminal sequence required for J-chain-mediated dimerization and transcytosis of secretory IgA across epithelial barriers (Woof and Kerr, 2006) (deletion of residues 455–472, PTHVN. DGTCY, antibody designated trastuzumab-IgA). Trastuzumab-IgGA could be expressed at a high yield and showed >90% correct assembly into the dimeric form, compared to 40% for trastuzumab-IgA (Figure 1C). Both glycan crowding and the deletion of the tailpiece have been previously shown to result in poor assembly of intrachain disulfide bonds in the hinge region of monomeric IgA1 (Atkin et al., 1996; Mattu et al., 1998), accounting for the poor assembly of trastuzumab-IgA that we observed. We also expressed and purified trastuzumab (an IgG1 antibody, designated herein as wild-type [WT] trastuzumab) and trastuzumab containing an N297D substitution. The mutation of N297 in IgG abolishes glycosylation, leading to a loss of binding to Fcg receptors as well as C1q and, thus, displaying background levels of ADCC, ADCP, and CDC. The dissociation constants (KD) of trastuzumab-IgGA, trastuzumab-IgA, and clinical-grade WT trastuzumab with purified human Fc receptors (FcgRI, FcgRIIa, FcgRIIb, and FcaRI) and with C1q were determined by SPR analysis (Table 1; Figures S2A and S2B). Notably, trastuzumab-IgGA bound to FcaRI with an apparent KD of 502 nM, an affinity only slightly lower than the value determined for binding to trastuzumab-IgA
1604 Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved
Chemistry & Biology Brief Communication
Figure 1. Engineering of IgGA (A) Sequence of the Fc domain of IgGA aligned with IgG1. Residues within the IgG1 CHg12 and CHg13 domains were exchanged with the underlined residues in orange from IgA1 to create IgGA. The Gly insertion is shown in black. (B) Model of IgGA structure. IgG1 residues in gray and IgA1 residues in orange (PDB ID: 3DO3 for IgG1 Fc fragment). (C) Expression of trastuzumab-IgGA (T-IgGA), trastuzumab-IgA (T-IgA), WT trastuzumab (T-IgG), and trastuzumab-N297D lacking the Fc glycan in HEK293 cells (T-IgG N297D). Antibodies were expressed in HEK293 cells, purified by Protein L chromatography (T-IgA, T-IgGA) or Protein A chromatography (T-IgG, T-IgG N297D), and separated by nonreducing SDS-PAGE on a 4%–20% gel.
(KD = 233 nM), which is consistent with the literature (Herr et al., 2003b). In contrast to trastuzumab-IgA, which did not bind to any FcgRs or to C1q, trastuzumab-IgGA displayed high-affinity binding to FcgRI, FcgRIIa, and FcgRIIb. Notably, trastuzumabIgGA bound to FcgRI with very high affinity (KD = 0.83 nM), only slightly lower than that of clinical-grade trastuzumab (KD = 0.36 nM). However, we could not detect the binding of trastuzumab-IgGA to FcRn or FcgRIIIa (CD16a) by ELISA, even though, in the case of the latter, all the Fc:FcgRIIIa contact residues observed in the cocrystal structure are preserved (PDB ID: 3SGJ) (Ferrara et al., 2011). Thus, it appears that conformation changes at the CH2/CH3 domain interface, possibly caused by the replacement of CHg13 with CHa13, may be responsible for abolishing binding to FcgRIIIa. The loss of binding to FcgRIIIa will interfere with natural-killer-cell-mediated ADCC, which is strongly implicated in the mechanism of some clinical antibodies. Unexpectedly, trastuzumab-IgGA bound C1q with a higher apparent affinity relative to IgG1 (tras-
tuzumab-IgGA, KD = 28 nM, compared with WT trastuzumab, KD = 84 nM), which, as an IgG1 isotype antibody, recruits C1q very efficiently. IgGA Mediates Potent Cancer Cell Killing by Fca and Fcg-Dependent Mechanisms Neutrophils are potent immune mediators of ADCC function, expressing high levels of FcaRI, FcgRIIa, and FcgRI when activated via inflammatory cytokines (Bakema et al., 2011). Even in their immature state (intermediate CD11b, low CD16 expression), neutrophils can kill tumor cells opsonized by IgA antibodies but not by IgG antibodies (Otten et al., 2005). Therefore, there is considerable interest in harnessing neutrophils for tumor cell elimination (Pascal et al., 2012; Souto et al., 2011; van Egmond and Bakema, 2013). ADCC assays were performed by detecting chromium 51 (51Cr) released from Her2-overexpressing breast cancer cells (SkBr3) following incubation with neutrophils and antibody (Figure S3A). Neutrophils from three healthy donors
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1605
Chemistry & Biology Brief Communication
Table 1. SPR Analysis of Fc Receptor Binding KD (nM) Variant
FcgRI
FcgRIIa
FcgRIIb
FcgRIIIa
FcRn
C1q
FcaRI
IgG
0.36 ± 0.06
1,140 ± 50
2,550 ± 140
ND
ND
84 ± 10
–
IgA
–
–
–
–
–
ND
233 ± 11
IgGA
0.83 ± 0.16
2,020 ± 110
4,330 ± 240
–
–
28 ± 7
502 ± 41
Equilibrium binding analysis was performed for trastuzumab-IgGA, trastuzumab-IgA, and WT trastuzumab (IgG1) with purified monomeric FcgRIIa, FcgRIIb, or FcaRI. Kinetic analysis was used to determine binding affinities for the high-affinity FcgRI and C1q interactions. For all interactions, antibodies were amine coupled to CM5 Biacore chips, and receptors were introduced at concentrations up to 10 mM. ND indicates that the affinity was not determined by Biacore, whereas a dash indicates that no binding was detected by ELISA (data are presented as mean ± SD; n = 3). See also Figure S2 for sensograms.
were first individually activated overnight with IFN-g to upregulate the expression of FcgRI (Figure S3D). 51Cr-labeled SkBr3 tumor cells were mixed with trastuzumab-IgA, trastuzumab-IgG, trastuzumab-IgGA, or trastuzumab-IgG-N297D (as a control for Fc independent activity) at concentrations of 0.2, 2, and 20 mg/ml. Control wells were included, lacking either antibodies (no Ab control), or SkBr3 target cells (spontaneous release control). Wells containing lysis buffer were also included (complete lysis control). Neutrophils were added at an 80:1 effector:tumor cell ratio and incubated for 4 hr. We observed that trastuzumab IgGA induced high levels of neutrophil-mediated ADCC (up to 50%) at antibody concentrations as low as 0.2 mg/ml (Figure 2A). With neutrophils as effector cells, trastuzumab IgGA was far more effective than IgG-formatted trastuzumab (WT trastuzumab). The latter displayed essentially no cytotoxic activity and was indistinguishable from the trastuzumab-IgG-N297D antibody control. CD11b+ macrophages express high levels of FcaRI/FcgRs, are able to infiltrate the tumor microenvironment, and eliminate pathogenic cells primarily via phagocytosis (ADCP) (Simpson et al., 2013). We evaluated the ability of monocyte-derived macrophages—differentiated from monocytes isolated from whole blood by incubation for 7 days with granulocyte macrophage colony-stimulating factor (GM-CSF) cytokine—to ingest Her2expressing cancer cells (Figure S3B). All blood donors were genotyped as homozygous for the R131 FcgRIIa allele, which has lower affinity for IgG1 than the H131 allele and are, therefore, less likely to mount IgG1-mediated ADCP responses as robustly as homozygous H131 donors. In these experiments, we used MDA-MB-453 cells expressing moderate levels of Her2 antigen, as these cells have been reported to be resistant to elimination by ADCP using trastuzumab in vitro (Narayan et al., 2009). MDA-MB-453 cells were fluorescently labeled with PKH67 and opsonized with trastuzumab-IgA, trastuzumab-IgG, or trastuzumab-IgGA, each at concentrations of 0.05, 0.5, and 5 mg/ml; and macrophages were added at a 5:1 effector:tumor cell ratio. After 2 hr of incubation, the extent of phagocytosis was evaluated using flow cytometry by determining the percentage of cells staining double positive for PKH67 and anti-CD11b/CD14. WT trastuzumab was unable to mediate phagocytosis of MDA-MB453 over background. In contrast, IgGA was able to effectively clear up to 50% of the tumor cells targeted (Figure 2B), a degree of phagocytosis comparable to that observed with trastuzumabIgA. Consistent with these observations, a number of studies have established that the activation of macrophages via FcaRI
can strongly elicit ADCP of cancer cells (Keler et al., 2000; van Egmond et al., 1999). Unlike IgA, but similarly to IgG1, IgGA binds C1q with high affinity. Complement activation via the classical antibody-dependent pathway is implicated as a key mechanism in the function of type I anti-CD20 antibodies such as rituximab (Zhou et al., 2008). After initially performing CDC assays with IgGA, IgG, and IgA antibodies in trastuzumab format, in which we saw no release of fluorescent dye from calcein-AM-loaded MDA-MB453 cells, we fused the rituximab Fab to each of these respective Fc domains for expression in HEK293 cells. Calcein-AM-loaded Raji cells that express high levels of CD20 were incubated with rituximab-IgGA, rituximab-IgA, or WT rituximab (IgG1) and fresh human serum for 1 hr, at which point the degree of release of the dye in the supernatant was determined (Figure S3C). Trastuzumab-IgG, was included as a negative control and, in addition, the spontaneous release of the dye was measured. IgGA exhibited highly potent CDC, resulting in tumor cell lysis at higher antibody concentrations (up to 70%) and detectable killing even at a very low concentration (0.625 mg/ml, up to 18% lysis) (Figure 2C). Consistent with the higher affinity of IgGA for C1q in vitro (Table 1), rituximab-IgGA was more efficient than WT rituximab at lower concentrations of antibody, e.g., up to 65% lysis compared to a maximum of 35% with rituximab at 2.5 mg/ml. We also observed a low but repeatable degree of CDC using rituximab-IgA at the highest concentration tested. Despite a lack of C1q binding, it has been reported that IgA can weakly activate complement via the lectin-mediated pathway and also, indirectly, recruit C1q during tumor cell apoptosis (apoptotic events are reported to induce antibody-independent C1q deposition) (Pascal et al., 2012). As expected, heat inactivation of the serum prior to use in these assays abrogated all CDC activity (Figure S3E). SIGNIFICANCE It is well established that engagement of the FcaRI receptor by IgA immune complexes elicits potent ADCC by activated neutrophils and eosinophils, as well as high phagocytic potency by macrophages (Dechant and Valerius, 2001; van Egmond et al., 2001). However, the challenges associated with IgA expression and purification and its inability to capitalize on the cytotoxic mechanisms that depend on FcgR or the activation of the classical complement pathway have limited the use of this antibody isotype as a therapeutic. We have
1606 Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved
Chemistry & Biology Brief Communication
Figure 2. ADCC, ADCP, and CDC Assays (A) ADCC of 51Cr loaded SkBR-3 Her2+ cells sensitized using anti-Her2 (trastuzumab) antibody variants having different Fc domains with IFN-g activated human neutrophils as effector cells. An 80:1 effector (neutrophil):tumor cell ratio was used, with neutrophils first activated by overnight culture with IFN-g (Otten et al., 2005). The dashed line indicates the values for the no-antibody control. (B) ADCP assays with monocyte-derived human macrophages and PKH67-labeled, Her2+ medium-expressing MDA-MB-453 tumor cells (Rhodes et al., 2002) at a 5:1 effector:tumor cell ratio. The dashed line indicates the no-antibody negative control. (C) CDC assays with fresh human serum using rituximab antibody variants to target calcein-loaded Raji CD20+ tumor cells. Trastuzumab-IgG was used as a negative control. All data are presented as the mean ± SD (n = 3).
now demonstrated the engineering of a ‘‘cross-isotype’’ IgGA antibody that displays high affinity to FcaRI and retains binding to the activating Fcg receptors FcgRI and FcgRIIa. Furthermore, we show that IgGA exhibits a 3-fold lower dissociation constant with C1q as compared to IgG1, which translates to improved CDC activity. While IgGA does not show affinity for FcgRIIIa, nor does it bind to FcRn in a pH-dependent manner, ongoing protein engineering efforts in our lab are focused on identifying improved IgGA variants that display the full spectrum of Fc-binding activities desirable for therapeutic applications. Despite this loss of binding, the combination of the robust ADCC activity of IgGA with neutrophils, coupled with the excellent ability of IgGA to elicit ADCP by macrophages and its high CDC activity, are likely to present a great advantage for the clearance of pathogenic cells, especially in can-
cer. In particular, the ability of IgGA to unleash the cytotoxic potential of neutrophils is of considerable interest because neutrophils represent the most abundant leukocyte subpopulation in blood and readily infiltrate solid tumors. Furthermore, neutrophils can be rapidly expanded and activated to express elevated FcgRI via the administration of granulocyte CSF (van Egmond and Bakema, 2013). Activated neutrophils have added benefits for therapy in vivo as they secrete a multitude of cytokines and chemokines to recruit other effector immune cell subsets (Mantovani et al., 2011). Finally, we note that the concept of cross-isotype antibodies demonstrated here with the engineering of IgGA may be extended to the creation of other types of cross-isotype antibodies that combine effector functions normally displayed by different isotypes, for example IgG and IgE, for increased therapeutic potency.
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1607
Chemistry & Biology Brief Communication
EXPERIMENTAL PROCEDURES
Egmond, M. (2011). Targeting FcaRI on polymorphonuclear cells induces tumor cell killing through autophagy. J. Immunol. 187, 726–732.
See Supplemental Experimental Procedures for further details. Studies involving human blood donors were performed under the oversight of the Institutional Review Board of The University of Texas at Austin (IRB# 2012-080031).
Beck, A., and Reichert, J.M. (2012). Marketing approval of mogamulizumab: a triumph for glyco-engineering. MAbs 4, 419–425.
SPR Affinity Measurements IgGA, IgA, or IgG antibodies were immobilized on CM5 amine chips in pH 4.5 or 5.0 sodium acetate buffer. For equilibrium binding KD measurements, monomeric FcgRIIa-R131, FcgRIIb, or FcaRI was passed over the chip surface until equilibrium was reached. A 1:1 Langmuir isotherm model was fitted to the data. For kinetic binding KD measurements, FcgRI or C1q was injected onto the chip at 30 ml/min. A 1:1 Langmuir isotherm model was fitted to FcgRI data, and a two-state reaction model was fit to the C1q data. ADCC Assays Neutrophils were purified from fresh human blood by Histopaque density gradient centrifugation. After red blood cell lysis, the neutrophil fraction was cultured overnight in RPMI medium containing 10% fetal bovine serum (FBS) and 50 ng/ml IFN-g. Activated neutrophils were combined at an 80:1 effector:tumor cell ratio with 51Cr-labeled SkBR-3 cells and each of the trastuzumab variants. After 4 hr at 37 C, the release of 51Cr into the media was determined by scintillation counting. ADCP Assays PBMCs were purified from fresh human blood by Histopaque density gradient centrifugation, and CD14+ monocytes were isolated by magnetic bead separation. Monocytes were differentiated into macrophages by culture for 7 days in RPMI medium containing 15% FBS and 50 ng/ml GM-CSF before combination at a 5:1 effector:tumor cell ratio with PKH67-labeled MDA-MB-453 cells and each of the trastuzumab variants. After 2 hr at 37 C, the cells were labeled with anti-CD11b-APC and anti-CD14-APC. Flow cytometry was used to quantify the number of phagocytosed double-positive events. CDC Assays Fresh human serum was mixed with each of the rituximab variants tested and with Raji tumor cells that were loaded with calcein AM for 1 hr. Supernatants were isolated by centrifugation and the released fluorescence was quantified at 535 nm. SUPPLEMENTAL INFORMATION Supplemental Information including Supplemental Experimental Procedures and three figures and can be found with this article online at http://dx.doi. org/10.1016/j.chembiol.2014.10.017. ACKNOWLEDGMENTS We are grateful to Dr. Oana Lungu for help with Rosetta structural modeling and Chhaya Das for her help with mammalian expression. Financial support was provided by the University of Texas at Austin, and the Clayton Foundation (to G.G.) and by a Top Achiever Doctoral Scholarship from the Tertiary Education Commission of New Zealand (to W.K.). We have filed a provisional patent application that relates to this work. Received: August 30, 2014 Revised: October 15, 2014 Accepted: October 22, 2014 Published: December 11, 2014 REFERENCES Atkin, J.D., Pleass, R.J., Owens, R.J., and Woof, J.M. (1996). Mutagenesis of the human IgA1 heavy chain tailpiece that prevents dimer assembly. J. Immunol. 157, 156–159. Bakema, J.E., Ganzevles, S.H., Fluitsma, D.M., Schilham, M.W., Beelen, R.H.J., Valerius, T., Lohse, S., Glennie, M.J., Medema, J.P., and van
Boross, P., Lohse, S., Nederend, M., Jansen, J.H.M., van Tetering, G., Dechant, M., Peipp, M., Royle, L., Liew, L.P., Boon, L., et al. (2013). IgA EGFR antibodies mediate tumour killing in vivo. EMBO Mol. Med. 5, 1213– 1226. Bruhns, P., Lannascoli, B., England, P., Mancardi, D.A., Fernandez, N., Jorieux, S., and Dae¨ron, M. (2009). Specificity and affinity of human Fcg receptors and their polymorphic variants for human IgG subclasses. Blood 113, 3716–3725. Carayannopoulos, L., Hexham, J.M., and Capra, J.D. (1996). Localization of the binding site for the monocyte immunoglobulin (Ig) A-Fc receptor (CD89) to the domain boundary between Calpha2 and Calpha3 in human IgA1. J. Exp. Med. 183, 1579–1586. Chintalacharuvu, K.R., Vuong, L.U., Loi, L.A., Larrick, J.W., and Morrison, S.L. (2001). Hybrid IgA2/IgG1 antibodies with tailor-made effector functions. Clin. Immunol. 101, 21–31. Chu, S.Y., Vostiar, I., Karki, S., Moore, G.L., Lazar, G.A., Pong, E., Joyce, P.F., Szymkowski, D.E., and Desjarlais, J.R. (2008). Inhibition of B cell receptormediated activation of primary human B cells by coengagement of CD19 and FcgammaRIIb with Fc-engineered antibodies. Mol. Immunol. 45, 3926– 3933. Dechant, M., and Valerius, T. (2001). IgA antibodies for cancer therapy. Crit. Rev. Oncol. Hematol. 39, 69–77. Dechant, M., Beyer, T., Schneider-Merck, T., Weisner, W., Peipp, M., van de Winkel, J.G.J., and Valerius, T. (2007). Effector mechanisms of recombinant IgA antibodies against epidermal growth factor receptor. J. Immunol. 179, 2936–2943. Desjarlais, J.R., and Lazar, G.A. (2011). Modulation of antibody effector function. Exp. Cell Res. 317, 1278–1285. Ferrara, C., Grau, S., Ja¨ger, C., Sondermann, P., Bru¨nker, P., Waldhauer, I., Hennig, M., Ruf, A., Rufer, A.C., Stihle, M., et al. (2011). Unique carbohydrate-carbohydrate interactions are required for high affinity binding between FcgammaRIII and antibodies lacking core fucose. Proc. Natl. Acad. Sci. USA 108, 12669–12674. Gregory, A.D., and Houghton, A.M. (2011). Tumor-associated neutrophils: new targets for cancer therapy. Cancer Res. 71, 2411–2416. Herr, A.B., Ballister, E.R., and Bjorkman, P.J. (2003a). Insights into IgA-mediated immune responses from the crystal structures of human FcalphaRI and its complex with IgA1-Fc. Nature 423, 614–620. Herr, A.B., White, C.L., Milburn, C., Wu, C., and Bjorkman, P.J. (2003b). Bivalent binding of IgA1 to FcalphaRI suggests a mechanism for cytokine activation of IgA phagocytosis. J. Mol. Biol. 327, 645–657. Hogarth, P.M., and Pietersz, G.A. (2012). Fc receptor-targeted therapies for the treatment of inflammation, cancer and beyond. Nat. Rev. Drug Discov. 11, 311–331. Jung, S.T., Kelton, W., Kang, T.H., Ng, D.T.W., Andersen, J.T., Sandlie, I., Sarkar, C.A., and Georgiou, G. (2013). Effective phagocytosis of low Her2 tumor cell lines with engineered, aglycosylated IgG displaying high FcgRIIa affinity and selectivity. ACS Chem. Biol. 8, 368–375. Keler, T., Wallace, P.K., Vitale, L.A., Russoniello, C., Sundarapandiyan, K., Graziano, R.F., and Deo, Y.M. (2000). Differential effect of cytokine treatment on Fc a receptor I- and Fc g receptor I-mediated tumor cytotoxicity by monocyte-derived macrophages. J. Immunol. 164, 5746–5752. Lohse, S., Brunke, C., Derer, S., Peipp, M., Boross, P., Kellner, C., Beyer, T., Dechant, M., van der Winkel, J.G., Leusen, J.H., and Valerius, T. (2012). Characterization of a mutated IgA2 antibody of the m(1) allotype against the epidermal growth factor receptor for the recruitment of monocytes and macrophages. J. Biol. Chem. 287, 25139–25150. Mantovani, A., Cassatella, M.A., Costantini, C., and Jaillon, S. (2011). Neutrophils in the activation and regulation of innate and adaptive immunity. Nat. Rev. Immunol. 11, 519–531.
1608 Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved
Chemistry & Biology Brief Communication
Mattu, T.S., Pleass, R.J., Willis, A.C., Kilian, M., Wormald, M.R., Lellouch, A.C., Rudd, P.M., Woof, J.M., and Dwek, R.A. (1998). The glycosylation and structure of human serum IgA1, Fab, and Fc regions and the role of N-glycosylation on Fca receptor interactions. J. Biol. Chem. 273, 2260–2272.
Roos, A., Bouwman, L.H., van Gijlswijk-Janssen, D.J., Faber-Krol, M.C., Stahl, G.L., and Daha, M.R. (2001). Human IgA activates the complement system via the mannan-binding lectin pathway. J. Immunol. 167, 2861–2868.
Narayan, M., Wilken, J.A., Harris, L.N., Baron, A.T., Kimbler, K.D., and Maihle, N.J. (2009). Trastuzumab-induced HER reprogramming in ‘‘resistant’’ breast carcinoma cells. Cancer Res. 69, 2191–2194.
Simpson, T.R., Li, F., Montalvo-Ortiz, W., Sepulveda, M.A., Bergerhoff, K., Arce, F., Roddie, C., Henry, J.Y., Yagita, H., Wolchok, J.D., et al. (2013). Fcdependent depletion of tumor-infiltrating regulatory T cells co-defines the efficacy of anti-CTLA-4 therapy against melanoma. J. Exp. Med. 210, 1695–1710.
Nimmerjahn, F., and Ravetch, J.V. (2012). Translating basic mechanisms of IgG effector activity into next generation cancer therapies. Cancer Immun. 12, 13.
Souto, J.C., Vila, L., and Bru´, A. (2011). Polymorphonuclear neutrophils and cancer: intense and sustained neutrophilia as a treatment against solid tumors. Med. Res. Rev. 31, 311–363.
Otten, M.A., Rudolph, E., Dechant, M., Tuk, C.W., Reijmers, R.M., Beelen, R.H., van de Winkel, J.G., and van Egmond, M. (2005). Immature neutrophils mediate tumor cell killing via IgA but not IgG Fc receptors. J. Immunol. 174, 5472–5480.
van Egmond, M., and Bakema, J.E. (2013). Neutrophils as effector cells for antibody-based immunotherapy of cancer. Semin. Cancer Biol. 23, 190–199.
Pascal, V., Laffleur, B., Debin, A., Cuvillier, A., van Egmond, M., Drocourt, D., Imbertie, L., Pangault, C., Tarte, K., Tiraby, G., and Cogne´, M. (2012). AntiCD20 IgA can protect mice against lymphoma development: evaluation of the direct impact of IgA and cytotoxic effector recruitment on CD20 target cells. Haematologica 97, 1686–1694. Rhodes, A., Jasani, B., Anderson, E., Dodson, A.R., and Balaton, A.J. (2002). Evaluation of HER-2/neu immunohistochemical assay sensitivity and scoring on formalin-fixed and paraffin-processed cell lines and breast tumors: a comparative study involving results from laboratories in 21 countries. Am. J. Clin. Pathol. 118, 408–417.
van Egmond, M., Hanneke van Vuuren, A.J., and van de Winkel, J.G.J. (1999). The human Fc receptor for IgA (Fc a RI, CD89) on transgenic peritoneal macrophages triggers phagocytosis and tumor cell lysis. Immunol. Lett. 68, 83–87. van Egmond, M., Damen, C.A., van Spriel, A.B., Vidarsson, G., van Garderen, E., and van de Winkel, J.G.J. (2001). IgA and the IgA Fc receptor. Trends Immunol. 22, 205–211. Woof, J.M., and Kerr, M.A. (2006). The function of immunoglobulin A in immunity. J. Pathol. 208, 270–282. Zhou, X., Hu, W., and Qin, X. (2008). The role of complement in the mechanism of action of rituximab for B-cell lymphoma: implications for therapy. Oncologist 13, 954–966.
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1609
Brief Communication IgGA: A ‘‘Cross-Isotype’’ Engineered Human Fc Antibody Domain that Displays Both IgG-like and IgA-like Effector Functions William Kelton,1 Nishant Mehta,2 Wissam Charab,1 Jiwon Lee,1 Chang-han Lee,1 Takaaki Kojima,5 Tae Hyun Kang,2 and George Georgiou1,2,3,4,* 1Department
of Chemical Engineering of Biomedical Engineering 3Section of Molecular Genetics and Microbiology 4Institute for Cellular and Molecular Biology University of Texas at Austin, Austin, TX 78712, USA 5Laboratory of Molecular Biotechnology, Graduate School of Bioagricultural Sciences, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan *Correspondence: [email protected] http://dx.doi.org/10.1016/j.chembiol.2014.10.017 2Department
SUMMARY
All clinically approved antibodies are of the IgG isotype and mediate the clearance of target cells via binding to Fcg receptors and complement (C1q). Even though IgA can elicit powerful cytotoxic action via FcaRI receptor binding, IgA antibodies have not been amenable to therapeutic development. Here, we report the engineering of a ‘‘cross-isotype’’ antibody, IgGA, which combines the effector functions of both IgG and IgA. IgGA binds to FcaRI with an affinity comparable to that of IgA, and to the activating Fcg receptors, FcgRI and FcgRIIa, with high affinity, and displays increased binding to C1q compared to IgG. Unlike trastuzumab-IgG, trastuzumab-IgGA potently activates both neutrophils and macrophages to kill Her2+ cancer cells. Furthermore, IgGA mediates greater complement-dependent cytotoxicity than IgG1 or IgA antibodies. The multitude of IgGA effector functions could be important for therapeutic purposes and highlights the concept of engineering antibodies that combine effector functions from multiple antibody isotypes.
INTRODUCTION Extensive clinical and laboratory evidence has demonstrated that immune effector cytotoxic mechanisms (antibody-dependent cell cytotoxicity [ADCC], antibody-dependent cell phagocytosis [ADCP], and complement-dependent cytotoxicity [CDC]) play a crucial role in the clearance of cancer cells by therapeutic antibodies (Hogarth and Pietersz, 2012). Immunoglobulin (Ig)G immune complexes bind to the activating Fcg receptors (FcgRs), FcgRI (CD64), FcgRIIa (CD32a), and FcgRIIIa (CD16a) that are expressed at varying levels on leukocyte subsets (Bruhns et al., 2009), triggering phosphorylation of the intracellular immunoreceptor tyrosine-based activating motif (ITAM) and, in turn,
downstream signaling events that ultimately result in the destruction of pathogens by ADCC and ADCP. Most leukocytes also express the inhibitory receptor FcgRIIb that signals through an immunoreceptor tyrosine-based inhibitory motif (ITIM) to induce anti-inflammatory responses. It has been proposed that the ratio of binding affinity to the activating FcgRs relative to FcgRIIb (referred to as the activation to inhibition ratio), dictates whether stimulation with immune complexes will trigger inflammatory or anti-inflammatory responses (Nimmerjahn and Ravetch, 2012). The pharmaceutical industry has invested heavily in the engineering of antibodies that display higher affinity or binding selectivity for FcgR and, thus, confer enhanced killing of target cells by ADCC, ADCP, and/or CDC. At present, there are >15 Fc-engineered antibody therapeutics in clinical trials (phase II or III) and two that have been approved as cancer drugs: obinutuzumab/Gazyva (anti-CD20) and mogamulizumab/Poteligeo (antiCCR4) (Beck and Reichert, 2012; Desjarlais and Lazar, 2011). Improvements to the binding affinity, or selectivity, of antibodies for FcgRs have been accomplished either through protein engineering or through glycoengineering via the modification of the critical glycan appended to residue N297 of the IgG Fc domain (Desjarlais and Lazar, 2011; Nimmerjahn and Ravetch, 2012). Examples of engineered Fc domains displaying two or more orders of magnitude improvement in FcgR binding affinity or selectivity have been reported (Chu et al., 2008; Jung et al., 2013). Currently, all antibody therapeutics are of the IgG isotype (predominantly IgG1), which is the most abundant antibody isotype in human serum. These IgG antibodies do not recognize the FcaRI receptor (CD89) that mediates the effector functions of IgA (both for IgA1 and IgA2), the second most abundant antibody isotype in human circulation and the most prevalent in mucosal secretions (Woof and Kerr, 2006). FcaRI is constitutively expressed on many cells of the myeloid lineage, such as neutrophils, macrophages, Kuppfer cells, and eosinophils (van Egmond et al., 1999). Recent studies have clearly demonstrated that engagement of FcaRI by IgA can elicit effective clearance of tumor cells in vitro (Lohse et al., 2012) and in animal models (Boross et al., 2013). This has been shown to occur via macrophage-mediated ADCP and via ADCC by neutrophils. In
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1603
Chemistry & Biology Brief Communication
particular, the FcaRI-dependent activation of neutrophils by IgA has been reported to be far more effective at tumor cell killing than FcgR-mediated activation by IgG antibodies (Dechant et al., 2007). The ability to capitalize on neutrophils for the elimination of pathogenic cells, especially in cancer therapy, is particularly exciting because neutrophils constitute the most abundant leukocyte subset in blood and can infiltrate many solid tumors (Gregory and Houghton, 2011). Unfortunately, the use of IgA antibodies as therapeutics has been hampered by several limitations. (1) The hinge region of IgA is heavily glycosylated, and producing proteins with multiple complex glycans can be problematic for bioprocessing and quality control during drug development (Woof and Kerr, 2006). (2) IgA does not bind to C1q; thus, it cannot mediate CDC through the classical complement pathway (van Egmond et al., 2001). (3) IgA also does not bind to Fcg receptors and, hence, cannot take advantage of the diverse mechanisms for immune cell activation that are pertinent to IgG1 therapeutics. (4) IgA exhibits a shorter circulation half-life than IgG because of a lack of binding to the FcRn receptor, which is important for intracellular recycling. Here, we used a semirational protein engineering strategy to develop a human ‘‘cross-isotype’’ Fc domain, termed IgGA, that displays many of the immunobiological functions of both IgA and IgG isotype antibodies. Not only do engineered IgGA antibodies bind to human FcaRI with an affinity similar to that of IgA antibodies, they also display binding to human FcgRI, FcgRIIa/b, and C1q with an affinity similar to that of IgG1 antibodies. With human neutrophils as effector cells, trastuzumab-IgGA displays much higher ADCC activity compared to clinical-grade trastuzumab (IgG) and ADCC activity comparable to that of trastuzumab formatted as an IgA. Furthermore, trastuzumab-IgGA was much more potent than its IgG counterpart in eliciting phagocytosis (ADCP) of MDA-MB-453 breast cancer cells, which express moderate levels of Her2. Additionally, unlike IgA, which can activate complement only weakly via the alternative and lectin-mediated pathways (Roos et al., 2001), we show that IgGA is at least as effective as IgG1 in mediating CDC. RESULTS AND DISCUSSION The FcaRI-binding site of the IgA1 Fca domain is composed of residues in the AB helix/loop of the CHa12 domain and multiple regions within CHa13 (crystal structure, Protein Data Bank [PDB] ID: 1OW0) (Herr et al., 2003a). We used Ala scanning mutagenesis to identify the relative importance of FcaRI contact residues within the a1 loop of CHa12 (L257, L258), ‘‘noncontact’’ residues within the crystallographically poorly resolved a2 loop of CHa12 (E313, N316, and H317), and certain contacting residues within 4 A˚ of FcaRI in the CHa13 domain: R382 (C strand), E389 (CC0 loop), M433 (F strand), E437 (FG loop), L441 (FG loop), and F443 (G strand) (Figures S1A and S1B available online). Mutant Fca1 domains were expressed in human embryonic kidney 293 (HEK293) cells, and binding to purified FcaRI-GST, also produced in HEK293 cells, was determined by ELISA (Figure S1C). Consistent with earlier reports (Carayannopoulos et al., 1996), an L-Ala substitution at L258 in the a1 loop of the CHa12 domain was found to abrogate FcaRI binding. Further inspection of the Fca L258A protein by nonreducing SDS-PAGE
revealed that this mutation markedly impacts Fca1 assembly. In contrast, an Ala mutation of the adjacent residue (L257) had essentially no effect on Fca1 assembly or FcaRI binding. Mutation of either E313 or N316 in the a2 loop abolished FcaRI binding, possibly through the disruption of hydrogen bonding with Y86 of FcaRI. Within the CHa13 domain, charged residues in the C beta sheet (R382 and E389) were moderately important for binding, while M433, which lies within a hydrophobic binding pocket, was essential for FcaRI binding. However, other residues within the hydrophobic binding pocket (E437, L441, and F443) were more tolerant of Ala substitution. Fc domains comprising the CHa13 fused to CHg12 do not bind to FcaRI (Chintalacharuvu et al., 2001). We constructed a series of chimeric Fc domains using this CHg12-CHa13 fusion as a template in order to preserve important CC0 loop, F strand, FG loop, and G strand contacts identified in the cocrystal structure and by Ala scanning. Because the CHg13 domain is not directly involved in binding to the FcgRs, we reasoned that exchange of this domain would have a minimal effect on FcgR affinity. Both the a2 loop and parts of the a1 loop of CHg12 were exchanged with the respective sequences from CHa12, since these two loops are also important for binding to the receptor, as shown earlier. One construct, which we termed IgGA (Figures 1A and 1B), had a1 loop residues 245–258 (PKPKDTLMISRTPE) of CHg12 and 340–447 (KGQPR. LSPGK) of CHg13 replaced with CHa12 residues 251–263 (PALEDLLLGSEAN) and CHa13 residues 341–450 (SGNTF. KTIDR). In addition, a Gly residue was inserted following CHa12 residue 257 (to ensure that the length of the grafted loop from CHa12 is consistent with that of CHg12). When the IgGA Fc was expressed in HEK293 cells, it was found to assemble efficiently and to bind to both FcaRI and to certain FcgRs by ELISA (data not shown; see surface plasmon resonance [SPR] analysis discussed later). The Fab domains of trastuzumab were fused to the IgGA Fc (trastuzumab-IgGA) or, alternatively, to Fca1 lacking the IgA tailpiece, a C-terminal sequence required for J-chain-mediated dimerization and transcytosis of secretory IgA across epithelial barriers (Woof and Kerr, 2006) (deletion of residues 455–472, PTHVN. DGTCY, antibody designated trastuzumab-IgA). Trastuzumab-IgGA could be expressed at a high yield and showed >90% correct assembly into the dimeric form, compared to 40% for trastuzumab-IgA (Figure 1C). Both glycan crowding and the deletion of the tailpiece have been previously shown to result in poor assembly of intrachain disulfide bonds in the hinge region of monomeric IgA1 (Atkin et al., 1996; Mattu et al., 1998), accounting for the poor assembly of trastuzumab-IgA that we observed. We also expressed and purified trastuzumab (an IgG1 antibody, designated herein as wild-type [WT] trastuzumab) and trastuzumab containing an N297D substitution. The mutation of N297 in IgG abolishes glycosylation, leading to a loss of binding to Fcg receptors as well as C1q and, thus, displaying background levels of ADCC, ADCP, and CDC. The dissociation constants (KD) of trastuzumab-IgGA, trastuzumab-IgA, and clinical-grade WT trastuzumab with purified human Fc receptors (FcgRI, FcgRIIa, FcgRIIb, and FcaRI) and with C1q were determined by SPR analysis (Table 1; Figures S2A and S2B). Notably, trastuzumab-IgGA bound to FcaRI with an apparent KD of 502 nM, an affinity only slightly lower than the value determined for binding to trastuzumab-IgA
1604 Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved
Chemistry & Biology Brief Communication
Figure 1. Engineering of IgGA (A) Sequence of the Fc domain of IgGA aligned with IgG1. Residues within the IgG1 CHg12 and CHg13 domains were exchanged with the underlined residues in orange from IgA1 to create IgGA. The Gly insertion is shown in black. (B) Model of IgGA structure. IgG1 residues in gray and IgA1 residues in orange (PDB ID: 3DO3 for IgG1 Fc fragment). (C) Expression of trastuzumab-IgGA (T-IgGA), trastuzumab-IgA (T-IgA), WT trastuzumab (T-IgG), and trastuzumab-N297D lacking the Fc glycan in HEK293 cells (T-IgG N297D). Antibodies were expressed in HEK293 cells, purified by Protein L chromatography (T-IgA, T-IgGA) or Protein A chromatography (T-IgG, T-IgG N297D), and separated by nonreducing SDS-PAGE on a 4%–20% gel.
(KD = 233 nM), which is consistent with the literature (Herr et al., 2003b). In contrast to trastuzumab-IgA, which did not bind to any FcgRs or to C1q, trastuzumab-IgGA displayed high-affinity binding to FcgRI, FcgRIIa, and FcgRIIb. Notably, trastuzumabIgGA bound to FcgRI with very high affinity (KD = 0.83 nM), only slightly lower than that of clinical-grade trastuzumab (KD = 0.36 nM). However, we could not detect the binding of trastuzumab-IgGA to FcRn or FcgRIIIa (CD16a) by ELISA, even though, in the case of the latter, all the Fc:FcgRIIIa contact residues observed in the cocrystal structure are preserved (PDB ID: 3SGJ) (Ferrara et al., 2011). Thus, it appears that conformation changes at the CH2/CH3 domain interface, possibly caused by the replacement of CHg13 with CHa13, may be responsible for abolishing binding to FcgRIIIa. The loss of binding to FcgRIIIa will interfere with natural-killer-cell-mediated ADCC, which is strongly implicated in the mechanism of some clinical antibodies. Unexpectedly, trastuzumab-IgGA bound C1q with a higher apparent affinity relative to IgG1 (tras-
tuzumab-IgGA, KD = 28 nM, compared with WT trastuzumab, KD = 84 nM), which, as an IgG1 isotype antibody, recruits C1q very efficiently. IgGA Mediates Potent Cancer Cell Killing by Fca and Fcg-Dependent Mechanisms Neutrophils are potent immune mediators of ADCC function, expressing high levels of FcaRI, FcgRIIa, and FcgRI when activated via inflammatory cytokines (Bakema et al., 2011). Even in their immature state (intermediate CD11b, low CD16 expression), neutrophils can kill tumor cells opsonized by IgA antibodies but not by IgG antibodies (Otten et al., 2005). Therefore, there is considerable interest in harnessing neutrophils for tumor cell elimination (Pascal et al., 2012; Souto et al., 2011; van Egmond and Bakema, 2013). ADCC assays were performed by detecting chromium 51 (51Cr) released from Her2-overexpressing breast cancer cells (SkBr3) following incubation with neutrophils and antibody (Figure S3A). Neutrophils from three healthy donors
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1605
Chemistry & Biology Brief Communication
Table 1. SPR Analysis of Fc Receptor Binding KD (nM) Variant
FcgRI
FcgRIIa
FcgRIIb
FcgRIIIa
FcRn
C1q
FcaRI
IgG
0.36 ± 0.06
1,140 ± 50
2,550 ± 140
ND
ND
84 ± 10
–
IgA
–
–
–
–
–
ND
233 ± 11
IgGA
0.83 ± 0.16
2,020 ± 110
4,330 ± 240
–
–
28 ± 7
502 ± 41
Equilibrium binding analysis was performed for trastuzumab-IgGA, trastuzumab-IgA, and WT trastuzumab (IgG1) with purified monomeric FcgRIIa, FcgRIIb, or FcaRI. Kinetic analysis was used to determine binding affinities for the high-affinity FcgRI and C1q interactions. For all interactions, antibodies were amine coupled to CM5 Biacore chips, and receptors were introduced at concentrations up to 10 mM. ND indicates that the affinity was not determined by Biacore, whereas a dash indicates that no binding was detected by ELISA (data are presented as mean ± SD; n = 3). See also Figure S2 for sensograms.
were first individually activated overnight with IFN-g to upregulate the expression of FcgRI (Figure S3D). 51Cr-labeled SkBr3 tumor cells were mixed with trastuzumab-IgA, trastuzumab-IgG, trastuzumab-IgGA, or trastuzumab-IgG-N297D (as a control for Fc independent activity) at concentrations of 0.2, 2, and 20 mg/ml. Control wells were included, lacking either antibodies (no Ab control), or SkBr3 target cells (spontaneous release control). Wells containing lysis buffer were also included (complete lysis control). Neutrophils were added at an 80:1 effector:tumor cell ratio and incubated for 4 hr. We observed that trastuzumab IgGA induced high levels of neutrophil-mediated ADCC (up to 50%) at antibody concentrations as low as 0.2 mg/ml (Figure 2A). With neutrophils as effector cells, trastuzumab IgGA was far more effective than IgG-formatted trastuzumab (WT trastuzumab). The latter displayed essentially no cytotoxic activity and was indistinguishable from the trastuzumab-IgG-N297D antibody control. CD11b+ macrophages express high levels of FcaRI/FcgRs, are able to infiltrate the tumor microenvironment, and eliminate pathogenic cells primarily via phagocytosis (ADCP) (Simpson et al., 2013). We evaluated the ability of monocyte-derived macrophages—differentiated from monocytes isolated from whole blood by incubation for 7 days with granulocyte macrophage colony-stimulating factor (GM-CSF) cytokine—to ingest Her2expressing cancer cells (Figure S3B). All blood donors were genotyped as homozygous for the R131 FcgRIIa allele, which has lower affinity for IgG1 than the H131 allele and are, therefore, less likely to mount IgG1-mediated ADCP responses as robustly as homozygous H131 donors. In these experiments, we used MDA-MB-453 cells expressing moderate levels of Her2 antigen, as these cells have been reported to be resistant to elimination by ADCP using trastuzumab in vitro (Narayan et al., 2009). MDA-MB-453 cells were fluorescently labeled with PKH67 and opsonized with trastuzumab-IgA, trastuzumab-IgG, or trastuzumab-IgGA, each at concentrations of 0.05, 0.5, and 5 mg/ml; and macrophages were added at a 5:1 effector:tumor cell ratio. After 2 hr of incubation, the extent of phagocytosis was evaluated using flow cytometry by determining the percentage of cells staining double positive for PKH67 and anti-CD11b/CD14. WT trastuzumab was unable to mediate phagocytosis of MDA-MB453 over background. In contrast, IgGA was able to effectively clear up to 50% of the tumor cells targeted (Figure 2B), a degree of phagocytosis comparable to that observed with trastuzumabIgA. Consistent with these observations, a number of studies have established that the activation of macrophages via FcaRI
can strongly elicit ADCP of cancer cells (Keler et al., 2000; van Egmond et al., 1999). Unlike IgA, but similarly to IgG1, IgGA binds C1q with high affinity. Complement activation via the classical antibody-dependent pathway is implicated as a key mechanism in the function of type I anti-CD20 antibodies such as rituximab (Zhou et al., 2008). After initially performing CDC assays with IgGA, IgG, and IgA antibodies in trastuzumab format, in which we saw no release of fluorescent dye from calcein-AM-loaded MDA-MB453 cells, we fused the rituximab Fab to each of these respective Fc domains for expression in HEK293 cells. Calcein-AM-loaded Raji cells that express high levels of CD20 were incubated with rituximab-IgGA, rituximab-IgA, or WT rituximab (IgG1) and fresh human serum for 1 hr, at which point the degree of release of the dye in the supernatant was determined (Figure S3C). Trastuzumab-IgG, was included as a negative control and, in addition, the spontaneous release of the dye was measured. IgGA exhibited highly potent CDC, resulting in tumor cell lysis at higher antibody concentrations (up to 70%) and detectable killing even at a very low concentration (0.625 mg/ml, up to 18% lysis) (Figure 2C). Consistent with the higher affinity of IgGA for C1q in vitro (Table 1), rituximab-IgGA was more efficient than WT rituximab at lower concentrations of antibody, e.g., up to 65% lysis compared to a maximum of 35% with rituximab at 2.5 mg/ml. We also observed a low but repeatable degree of CDC using rituximab-IgA at the highest concentration tested. Despite a lack of C1q binding, it has been reported that IgA can weakly activate complement via the lectin-mediated pathway and also, indirectly, recruit C1q during tumor cell apoptosis (apoptotic events are reported to induce antibody-independent C1q deposition) (Pascal et al., 2012). As expected, heat inactivation of the serum prior to use in these assays abrogated all CDC activity (Figure S3E). SIGNIFICANCE It is well established that engagement of the FcaRI receptor by IgA immune complexes elicits potent ADCC by activated neutrophils and eosinophils, as well as high phagocytic potency by macrophages (Dechant and Valerius, 2001; van Egmond et al., 2001). However, the challenges associated with IgA expression and purification and its inability to capitalize on the cytotoxic mechanisms that depend on FcgR or the activation of the classical complement pathway have limited the use of this antibody isotype as a therapeutic. We have
1606 Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved
Chemistry & Biology Brief Communication
Figure 2. ADCC, ADCP, and CDC Assays (A) ADCC of 51Cr loaded SkBR-3 Her2+ cells sensitized using anti-Her2 (trastuzumab) antibody variants having different Fc domains with IFN-g activated human neutrophils as effector cells. An 80:1 effector (neutrophil):tumor cell ratio was used, with neutrophils first activated by overnight culture with IFN-g (Otten et al., 2005). The dashed line indicates the values for the no-antibody control. (B) ADCP assays with monocyte-derived human macrophages and PKH67-labeled, Her2+ medium-expressing MDA-MB-453 tumor cells (Rhodes et al., 2002) at a 5:1 effector:tumor cell ratio. The dashed line indicates the no-antibody negative control. (C) CDC assays with fresh human serum using rituximab antibody variants to target calcein-loaded Raji CD20+ tumor cells. Trastuzumab-IgG was used as a negative control. All data are presented as the mean ± SD (n = 3).
now demonstrated the engineering of a ‘‘cross-isotype’’ IgGA antibody that displays high affinity to FcaRI and retains binding to the activating Fcg receptors FcgRI and FcgRIIa. Furthermore, we show that IgGA exhibits a 3-fold lower dissociation constant with C1q as compared to IgG1, which translates to improved CDC activity. While IgGA does not show affinity for FcgRIIIa, nor does it bind to FcRn in a pH-dependent manner, ongoing protein engineering efforts in our lab are focused on identifying improved IgGA variants that display the full spectrum of Fc-binding activities desirable for therapeutic applications. Despite this loss of binding, the combination of the robust ADCC activity of IgGA with neutrophils, coupled with the excellent ability of IgGA to elicit ADCP by macrophages and its high CDC activity, are likely to present a great advantage for the clearance of pathogenic cells, especially in can-
cer. In particular, the ability of IgGA to unleash the cytotoxic potential of neutrophils is of considerable interest because neutrophils represent the most abundant leukocyte subpopulation in blood and readily infiltrate solid tumors. Furthermore, neutrophils can be rapidly expanded and activated to express elevated FcgRI via the administration of granulocyte CSF (van Egmond and Bakema, 2013). Activated neutrophils have added benefits for therapy in vivo as they secrete a multitude of cytokines and chemokines to recruit other effector immune cell subsets (Mantovani et al., 2011). Finally, we note that the concept of cross-isotype antibodies demonstrated here with the engineering of IgGA may be extended to the creation of other types of cross-isotype antibodies that combine effector functions normally displayed by different isotypes, for example IgG and IgE, for increased therapeutic potency.
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1607
Chemistry & Biology Brief Communication
EXPERIMENTAL PROCEDURES
Egmond, M. (2011). Targeting FcaRI on polymorphonuclear cells induces tumor cell killing through autophagy. J. Immunol. 187, 726–732.
See Supplemental Experimental Procedures for further details. Studies involving human blood donors were performed under the oversight of the Institutional Review Board of The University of Texas at Austin (IRB# 2012-080031).
Beck, A., and Reichert, J.M. (2012). Marketing approval of mogamulizumab: a triumph for glyco-engineering. MAbs 4, 419–425.
SPR Affinity Measurements IgGA, IgA, or IgG antibodies were immobilized on CM5 amine chips in pH 4.5 or 5.0 sodium acetate buffer. For equilibrium binding KD measurements, monomeric FcgRIIa-R131, FcgRIIb, or FcaRI was passed over the chip surface until equilibrium was reached. A 1:1 Langmuir isotherm model was fitted to the data. For kinetic binding KD measurements, FcgRI or C1q was injected onto the chip at 30 ml/min. A 1:1 Langmuir isotherm model was fitted to FcgRI data, and a two-state reaction model was fit to the C1q data. ADCC Assays Neutrophils were purified from fresh human blood by Histopaque density gradient centrifugation. After red blood cell lysis, the neutrophil fraction was cultured overnight in RPMI medium containing 10% fetal bovine serum (FBS) and 50 ng/ml IFN-g. Activated neutrophils were combined at an 80:1 effector:tumor cell ratio with 51Cr-labeled SkBR-3 cells and each of the trastuzumab variants. After 4 hr at 37 C, the release of 51Cr into the media was determined by scintillation counting. ADCP Assays PBMCs were purified from fresh human blood by Histopaque density gradient centrifugation, and CD14+ monocytes were isolated by magnetic bead separation. Monocytes were differentiated into macrophages by culture for 7 days in RPMI medium containing 15% FBS and 50 ng/ml GM-CSF before combination at a 5:1 effector:tumor cell ratio with PKH67-labeled MDA-MB-453 cells and each of the trastuzumab variants. After 2 hr at 37 C, the cells were labeled with anti-CD11b-APC and anti-CD14-APC. Flow cytometry was used to quantify the number of phagocytosed double-positive events. CDC Assays Fresh human serum was mixed with each of the rituximab variants tested and with Raji tumor cells that were loaded with calcein AM for 1 hr. Supernatants were isolated by centrifugation and the released fluorescence was quantified at 535 nm. SUPPLEMENTAL INFORMATION Supplemental Information including Supplemental Experimental Procedures and three figures and can be found with this article online at http://dx.doi. org/10.1016/j.chembiol.2014.10.017. ACKNOWLEDGMENTS We are grateful to Dr. Oana Lungu for help with Rosetta structural modeling and Chhaya Das for her help with mammalian expression. Financial support was provided by the University of Texas at Austin, and the Clayton Foundation (to G.G.) and by a Top Achiever Doctoral Scholarship from the Tertiary Education Commission of New Zealand (to W.K.). We have filed a provisional patent application that relates to this work. Received: August 30, 2014 Revised: October 15, 2014 Accepted: October 22, 2014 Published: December 11, 2014 REFERENCES Atkin, J.D., Pleass, R.J., Owens, R.J., and Woof, J.M. (1996). Mutagenesis of the human IgA1 heavy chain tailpiece that prevents dimer assembly. J. Immunol. 157, 156–159. Bakema, J.E., Ganzevles, S.H., Fluitsma, D.M., Schilham, M.W., Beelen, R.H.J., Valerius, T., Lohse, S., Glennie, M.J., Medema, J.P., and van
Boross, P., Lohse, S., Nederend, M., Jansen, J.H.M., van Tetering, G., Dechant, M., Peipp, M., Royle, L., Liew, L.P., Boon, L., et al. (2013). IgA EGFR antibodies mediate tumour killing in vivo. EMBO Mol. Med. 5, 1213– 1226. Bruhns, P., Lannascoli, B., England, P., Mancardi, D.A., Fernandez, N., Jorieux, S., and Dae¨ron, M. (2009). Specificity and affinity of human Fcg receptors and their polymorphic variants for human IgG subclasses. Blood 113, 3716–3725. Carayannopoulos, L., Hexham, J.M., and Capra, J.D. (1996). Localization of the binding site for the monocyte immunoglobulin (Ig) A-Fc receptor (CD89) to the domain boundary between Calpha2 and Calpha3 in human IgA1. J. Exp. Med. 183, 1579–1586. Chintalacharuvu, K.R., Vuong, L.U., Loi, L.A., Larrick, J.W., and Morrison, S.L. (2001). Hybrid IgA2/IgG1 antibodies with tailor-made effector functions. Clin. Immunol. 101, 21–31. Chu, S.Y., Vostiar, I., Karki, S., Moore, G.L., Lazar, G.A., Pong, E., Joyce, P.F., Szymkowski, D.E., and Desjarlais, J.R. (2008). Inhibition of B cell receptormediated activation of primary human B cells by coengagement of CD19 and FcgammaRIIb with Fc-engineered antibodies. Mol. Immunol. 45, 3926– 3933. Dechant, M., and Valerius, T. (2001). IgA antibodies for cancer therapy. Crit. Rev. Oncol. Hematol. 39, 69–77. Dechant, M., Beyer, T., Schneider-Merck, T., Weisner, W., Peipp, M., van de Winkel, J.G.J., and Valerius, T. (2007). Effector mechanisms of recombinant IgA antibodies against epidermal growth factor receptor. J. Immunol. 179, 2936–2943. Desjarlais, J.R., and Lazar, G.A. (2011). Modulation of antibody effector function. Exp. Cell Res. 317, 1278–1285. Ferrara, C., Grau, S., Ja¨ger, C., Sondermann, P., Bru¨nker, P., Waldhauer, I., Hennig, M., Ruf, A., Rufer, A.C., Stihle, M., et al. (2011). Unique carbohydrate-carbohydrate interactions are required for high affinity binding between FcgammaRIII and antibodies lacking core fucose. Proc. Natl. Acad. Sci. USA 108, 12669–12674. Gregory, A.D., and Houghton, A.M. (2011). Tumor-associated neutrophils: new targets for cancer therapy. Cancer Res. 71, 2411–2416. Herr, A.B., Ballister, E.R., and Bjorkman, P.J. (2003a). Insights into IgA-mediated immune responses from the crystal structures of human FcalphaRI and its complex with IgA1-Fc. Nature 423, 614–620. Herr, A.B., White, C.L., Milburn, C., Wu, C., and Bjorkman, P.J. (2003b). Bivalent binding of IgA1 to FcalphaRI suggests a mechanism for cytokine activation of IgA phagocytosis. J. Mol. Biol. 327, 645–657. Hogarth, P.M., and Pietersz, G.A. (2012). Fc receptor-targeted therapies for the treatment of inflammation, cancer and beyond. Nat. Rev. Drug Discov. 11, 311–331. Jung, S.T., Kelton, W., Kang, T.H., Ng, D.T.W., Andersen, J.T., Sandlie, I., Sarkar, C.A., and Georgiou, G. (2013). Effective phagocytosis of low Her2 tumor cell lines with engineered, aglycosylated IgG displaying high FcgRIIa affinity and selectivity. ACS Chem. Biol. 8, 368–375. Keler, T., Wallace, P.K., Vitale, L.A., Russoniello, C., Sundarapandiyan, K., Graziano, R.F., and Deo, Y.M. (2000). Differential effect of cytokine treatment on Fc a receptor I- and Fc g receptor I-mediated tumor cytotoxicity by monocyte-derived macrophages. J. Immunol. 164, 5746–5752. Lohse, S., Brunke, C., Derer, S., Peipp, M., Boross, P., Kellner, C., Beyer, T., Dechant, M., van der Winkel, J.G., Leusen, J.H., and Valerius, T. (2012). Characterization of a mutated IgA2 antibody of the m(1) allotype against the epidermal growth factor receptor for the recruitment of monocytes and macrophages. J. Biol. Chem. 287, 25139–25150. Mantovani, A., Cassatella, M.A., Costantini, C., and Jaillon, S. (2011). Neutrophils in the activation and regulation of innate and adaptive immunity. Nat. Rev. Immunol. 11, 519–531.
1608 Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved
Chemistry & Biology Brief Communication
Mattu, T.S., Pleass, R.J., Willis, A.C., Kilian, M., Wormald, M.R., Lellouch, A.C., Rudd, P.M., Woof, J.M., and Dwek, R.A. (1998). The glycosylation and structure of human serum IgA1, Fab, and Fc regions and the role of N-glycosylation on Fca receptor interactions. J. Biol. Chem. 273, 2260–2272.
Roos, A., Bouwman, L.H., van Gijlswijk-Janssen, D.J., Faber-Krol, M.C., Stahl, G.L., and Daha, M.R. (2001). Human IgA activates the complement system via the mannan-binding lectin pathway. J. Immunol. 167, 2861–2868.
Narayan, M., Wilken, J.A., Harris, L.N., Baron, A.T., Kimbler, K.D., and Maihle, N.J. (2009). Trastuzumab-induced HER reprogramming in ‘‘resistant’’ breast carcinoma cells. Cancer Res. 69, 2191–2194.
Simpson, T.R., Li, F., Montalvo-Ortiz, W., Sepulveda, M.A., Bergerhoff, K., Arce, F., Roddie, C., Henry, J.Y., Yagita, H., Wolchok, J.D., et al. (2013). Fcdependent depletion of tumor-infiltrating regulatory T cells co-defines the efficacy of anti-CTLA-4 therapy against melanoma. J. Exp. Med. 210, 1695–1710.
Nimmerjahn, F., and Ravetch, J.V. (2012). Translating basic mechanisms of IgG effector activity into next generation cancer therapies. Cancer Immun. 12, 13.
Souto, J.C., Vila, L., and Bru´, A. (2011). Polymorphonuclear neutrophils and cancer: intense and sustained neutrophilia as a treatment against solid tumors. Med. Res. Rev. 31, 311–363.
Otten, M.A., Rudolph, E., Dechant, M., Tuk, C.W., Reijmers, R.M., Beelen, R.H., van de Winkel, J.G., and van Egmond, M. (2005). Immature neutrophils mediate tumor cell killing via IgA but not IgG Fc receptors. J. Immunol. 174, 5472–5480.
van Egmond, M., and Bakema, J.E. (2013). Neutrophils as effector cells for antibody-based immunotherapy of cancer. Semin. Cancer Biol. 23, 190–199.
Pascal, V., Laffleur, B., Debin, A., Cuvillier, A., van Egmond, M., Drocourt, D., Imbertie, L., Pangault, C., Tarte, K., Tiraby, G., and Cogne´, M. (2012). AntiCD20 IgA can protect mice against lymphoma development: evaluation of the direct impact of IgA and cytotoxic effector recruitment on CD20 target cells. Haematologica 97, 1686–1694. Rhodes, A., Jasani, B., Anderson, E., Dodson, A.R., and Balaton, A.J. (2002). Evaluation of HER-2/neu immunohistochemical assay sensitivity and scoring on formalin-fixed and paraffin-processed cell lines and breast tumors: a comparative study involving results from laboratories in 21 countries. Am. J. Clin. Pathol. 118, 408–417.
van Egmond, M., Hanneke van Vuuren, A.J., and van de Winkel, J.G.J. (1999). The human Fc receptor for IgA (Fc a RI, CD89) on transgenic peritoneal macrophages triggers phagocytosis and tumor cell lysis. Immunol. Lett. 68, 83–87. van Egmond, M., Damen, C.A., van Spriel, A.B., Vidarsson, G., van Garderen, E., and van de Winkel, J.G.J. (2001). IgA and the IgA Fc receptor. Trends Immunol. 22, 205–211. Woof, J.M., and Kerr, M.A. (2006). The function of immunoglobulin A in immunity. J. Pathol. 208, 270–282. Zhou, X., Hu, W., and Qin, X. (2008). The role of complement in the mechanism of action of rituximab for B-cell lymphoma: implications for therapy. Oncologist 13, 954–966.
Chemistry & Biology 21, 1603–1609, December 18, 2014 ª2014 Elsevier Ltd All rights reserved 1609
