Biological Characterization of a Stable Effector Functionless (SEFL) Monoclonal Antibody Scaffold in Vitro*

The stable effector functionLess (SEFL) antibody was designed as an IgG1 antibody with a constant region that lacks the ability to interact with Fcγ receptors. The engineering and stability and pharmacokinetic assessments of the SEFL scaffold is described in the accompanying article (Jacobsen, F. W., Stevenson, R., Li, C., Salimi-Moosavi, H., Liu, L., Wen, J., Luo, Q., Daris, K., Buck, L., Miller, S., Ho, S-Y., Wang, W., Chen, Q., Walker, K., Wypych, J., Narhi, L., and Gunasekaran, K. (2017) J. Biol. Chem. 292). The biological properties of these SEFL antibodies were assessed in a variety of human and cynomolgus monkey in vitro assays. Binding of parent molecules and their SEFL variants to human and cynomolgus monkey FcγRs were evaluated using flow cytometry-based binding assays. The SEFL variants tested showed decreased binding affinity to human and cynomolgus FcγRs compared with the wild-type IgG1 antibody. In addition, SEFL variants demonstrated no antibody-dependent cell-mediated cytotoxicity in vitro against Daudi cells with cynomolgus monkey peripheral blood mononuclear cells, and had minimal complement-dependent cytotoxicity activity similar to that of the negative control IgG2 in a CD20+ human Raji lymphoma cell line. SEFL mutations eliminated off-target antibody-dependent monocyte phagocytosis of cynomolgus monkey platelets, and cynomolgus platelet activation in vitro. These experiments demonstrate that the SEFL modifications successfully eliminated Fc-associated effector binding and functions.

Monoclonal antibodies (mAbs) are the largest class of biopharmaceuticals and have diverse clinical applications (1). The choice of therapeutic mAb isotype to develop (IgG1, IgG2, or IgG4) is dependent on the target (cell surface versus soluble), desired biology, safety (risk of immunogenicity and undesired immunological effects), and manufacturability (expression, formulation, and stability). More than 80% of approved therapeutic mAbs are IgG1 isotypes that target cell surface receptors and are effective for oncology indications (2). For these therapeutic approaches, mAb isotypes that can induce cell killing such as complement-dependent cytotoxicity (CDC) 2 and antibody-dependent cell-mediated cytotoxicity (ADCC) are often desirable (3,4). However, for non-oncologic indications, therapeutic mAbs without cytotoxic effector function may be more appropriate because cell killing may not be a goal of therapy.
The Fc portion of IgG has interaction sites for the effector ligands, including Fc␥ receptors (Fc␥RI, Fc␥RII, and Fc␥RIII), C1q complement, and the neonatal Fc receptor (FcRn). IgG isotypes differentially engage Fc␥ receptors and C1q binding to recruit immune effector functions and initiate cytotoxic effector functions, (either ADCC or CDC (5)). Historically, IgG2 or IgG4 isotypes were thought to have minimal cytotoxic effector function, and have been selected for applications where cytotoxic effector function is not required or desirable (5). However, recent evidence suggests that IgG2 is not completely devoid of cytotoxic effector function under certain culture conditions (6,7). In addition, IgG2 can also bind to Fc␥ receptors to induce non-cytotoxic effector functions. For example, IgG2 isotypes can mediate phagocytosis via monocytes/macrophages and neutrophils through interaction with the Fc␥RIIa on platelets (8), and binding to the Fc␥RIIa along with a cell surface antigen has been implicated in off-target effects of mAbs on platelets (9,10).
IgG mAb isotypes differ in stability as well as effector functions. IgG2s have decreased conformational stability relative to IgG1s, which can result in a higher propensity to aggregate compared with IgG1 in the native state (11,12). The three different disulfide bond arrangements possible in the hinge region of IgG2s enable disulfide shuffling, which can result in a more heterogeneous product (13). mAbs based on the IgG4 scaffold can disassociate and pair with endogenous IgG4 antibodies in vivo, leading to bispecific antibodies and variable efficacy of the therapeutic (14).
Improved mAb constructs would combine the best properties of the IgG1 and IgG2 molecules: the homogeneity and stability of an IgG1, a single disulfide bond, and virtually no effector functions. To eliminate effector functions, Fc modifications have been introduced that significantly reduce binding to Fc␥R (15,16). Mutation of the glycosylation site (Asn 297 ) in the CH2 domain of IgG1 generates aglycosylated mAbs, which have very low binding to Fc␥Rs (11,17). Mutations of Asn 297 are advantageous in that the introduction of only a single mutation in the constant domain is likely to minimize the risk of immunogenicity (18). Several aglycosylated IgG1 therapeutic mAbs are currently in clinical trials (i.e. Millennium ACT2 mAb is an IgG1 with N297A mutations to remove ADCC (17)). However, the N297A mutation is less thermally stable compared with natural, unmutated IgG1s (19,20). The Asn 297 mutated to Gly results in an Fc constant region that lacks the ability to interact with the Fc␥Rs (21) and shows improved thermal stability compared with other aglycosylated mAbs with good manufacturing and pharmacokinetic properties (22). These mAbs, termed stable effector functionless (SEFL) were tested for Fc functionality in a variety of biologic systems.
In this article, we describe the characterization of the biology of diverse therapeutic mAbs (16 variants of 5 mAbs) (22) to determine whether the SEFL modifications eliminated Fc␥R binding and Fc effector functions (Table 1). SEFL mAbs have been shown to retain FcRn binding and binding to the target antigen (22). Isotype control mAbs, mAbs that had previously been shown to have effector functions and SEFL variants of these mAbs were evaluated in a variety of assays, including binding strength to human and cynomolgus Fc␥Rs in cell-based flow cytometry assays, binding to cynomolgus peripheral blood granulocytes, cynomolgus PBMC ADCC assay, and a human CDC assay. In addition to the assessment of target-mediated Fc effector function, off-target-mediated Fc effector function of SEFL mAb was assessed in platelet phagocytosis and platelet activation assays. These experiments demonstrated that the SEFL modification of mAbs successfully eliminated both ontarget and off-target Fc-associated effector functions.

Results
Binding of parent mAb and their SEFL variants to human and cynomolgus monkey Fc␥Rs were evaluated using flow-based binding assays.
Cynomolgus Monkey Granulocyte Binding Assay-We have previously observed that some human IgG2 mAb clones will bind nonspecifically to cynomolgus granulocytes despite lack of target expression (for example, see Ref. 7). Therefore we compared the ability of the parental and SEFL forms of such mAbs to bind to cynomolgus peripheral blood granulocytes using flow cytometric analyses. SEFL variants of mAbW (Table 1) and an IgG2 control mAbV.IgG2 (Table 1) did not directly bind to cynomolgus monkey granulocytes (Fig. 3). In contrast, mAbW.IgG1 bound to cynomolgus monkey granulocytes demonstrating that off-target binding to cynomolgus monkey cells occurring with an IgG1 mAb can also be eliminated with the SEFL construct.
ADCC Assay-mAbW.IgG1 causes ADCC in vitro against Daudi cells in an assay using cynomolgus monkey PBMC effector cells (Fig. 4). To confirm that the ADCC activity of the SEFL mAbs was eliminated the mAbW antibody variants (mAbW.SEFL1.1 and mAbW.SEFL2.0; see Table 1) and control (mAbW.IgG2) were assessed in this assay. mAbW.IgG2, mAbW.SEFL1.1, and mAbW.SEFL2.0 did not mediate ADCC in vitro against Daudi cells with cynomolgus monkey PBMC. The slightly negative cytotoxicity seen for mAbW.SEFL1.1 is within the experimental variation due to the use of PBMC from cynomolgus monkey.
a mAbX is referred to as AMG X in Ref. 9. b SEFL mAb are described in Jacobsen et al. (22). c mAbY.IgG2 is referred to as mAbY.1 in Ref. 10. FEBRUARY 3, 2017 • VOLUME 292 • NUMBER 5

JOURNAL OF BIOLOGICAL CHEMISTRY 1877
Rituximab-mediated CDC Assays-Rituximab causes CDC in a CD20 ϩ human Raji lymphoma cell line (23). To further examine the effector functions of the SEFL mAb, the cytotoxic activities of two rituximab SEFL variants (RTX.SEFL1.1 and RTX.SEFL2.0; Table 1) were assessed for CDC activity in this assay (Fig. 5). Rituximab (RTX.IgG1) showed CDC activity of 27% cell death at a mAb concentration of 100 g/ml, whereas both SEFL variants of rituximab (RTX.SEFL1.1 and RTX.SEFL2.0) at a concentration of 100 g/ml were similar to that of the negative control RTX.IgG2.

JOURNAL OF BIOLOGICAL CHEMISTRY 1879
Platelet Activation Assays-mAbX, is an IgG2 mAb that has previously been shown to cause cynomolgus, but not human, platelet activation in vitro through an Fc-dependent process (9). As shown in Fig. 7, parental mAbX.IgG2 (positive control) caused cynomolgus platelet activation at concentrations as low as 1 mg/ml. In contrast, a SEFL variant of mAbX (mAbX.SEFL2.0) caused no activation of cynomolgus platelets at concentrations up to 7.8 mg/ml.

Discussion
Altering the Fc portion of a mAb has been conducted previously to either increase or decrease cytotoxic effector functions (ADCC and CDC) (5, 15, 24 -30). Because IgGs bind similarly to both human and cynomolgus monkey Fc␥R (31), and cynomolgus monkeys are often used in the safety assessment of biotherapeutics, it is helpful to demonstrate that the modifications have similar functional consequences in humans and cynomolgus monkeys. The SEFL mAb variants tested showed decreased binding affinity to human and cynomolgus monkey Fc␥Rs compared with the corresponding wild-type IgG1. The binding affinity of the SEFL IgG1 was decreased 40 -50% for cynomolgus monkey Fc␥RI and decreased by ϳ70% for human Fc␥RI in comparison to wild-type IgG1. In addition, cynomolgus monkey PBMC ADCC activity was eliminated using the SEFL mAb variants suggesting that reduced binding was sufficient to block this cytotoxic effector function. The SEFL modification of a mAb eliminated CDC activity in a human cell system, also indicating that the decreased Fc␥R binding correlated with a loss of cytotoxic effector function. However, under the constraints of the availability of cells for assays, ADCC and CDC activity was not tested in both cynomolgus monkey and human cells.
Several mAbs have now shown off-target effects (8), although the mechanisms involved are not well understood. Our results further support that some of these off-target effects require Fc-Fc␥R binding (10), and that the elimination of the effector function of a mAb may eliminate off-target reactions as well (as shown for mAbX and mAbY (9, 10)). The generation of a SEFL mAb construct offers a path forward for development of IgG1based mAbs when cytotoxicity is not desired, by eliminating the Fc-mediated events. The use of SEFL constructs eliminated mAb binding to human and cynomolgus Fc␥RIIa (including high binding isoform His 131 ) and Fc␥RIIb, whereas the nonmutated IgG1 and IgG2 show significant binding. Because Fc␥RII has been found on platelets and likely contributes to the non-target, non-cytotoxic effector functions seen with two IgG2 mAbs, mAbY and mAbX (9, 10), SEFL mAb constructs should eliminate these effects. Indeed, SEFL variants of mAbY and mAbX could not induce either platelet phagocytosis or platelet activation in contrast with their parental mAb constructs, and SEFL variants of mAbW did not bind to cynomolgus neutrophils. Thus, the change in Fc␥R binding also correlated with decreased off target effector function. Importantly, the SEFL mAb constructs do not impact FcRn binding such that the normal half-life of mAbs was maintained (22).
The SEFL mAb construct used in these studies also has superior manufacturability features compared with IgG2 mAbs, which are described in the accompanying article (22). This combination of stability, homogeneity (through elimination of disulfide variants and Asn 297 glycosylation), and lack of effector functions make this a promising scaffolding on which to build protein therapeutics for diseases where effector functions are not desired, such as autoimmune diseases.

Conclusion
In these experiments, we have shown that novel SEFL mAb constructs can significantly decrease binding to human and cynomolgus monkey Fc␥R, and eliminate cytotoxic and noncytotoxic effector functions in human and cynomolgus monkey in vitro systems. SEFL mutations show similar Fc␥R binding in humans and cynomolgus monkeys. The SEFL mAb variants attenuate ADCC and CDC activity in cynomolgus and human in vitro systems, respectively. Non-cytotoxic effector functions are also eliminated or reduced in a cynomolgus in vitro system (platelet activation, platelet phagocytosis, and nonspecific binding to cynomolgus neutrophils).
This article demonstrates that a mutation that eliminates effector function by decreasing binding to Fc␥R, acts similarly in humans and cynomolgus monkey immune cells in vitro, and is a promising scaffolding on which to build protein therapeutics for diseases where effector functions are not desired and where non-cytotoxic, non-target related effects on platelets may be an issue.
Fc␥ Cell-based Binding Assay-The cell-based IgG1/Fc␥R flow cytometry assay measures the relative binding of IgG1 to 293T cells or AM-1D cells (Amgen Inc.) stably transfected to express a single type of Fc␥R. 293T or AM-1D-Fc␥R cells were incubated with the IgG1 samples at increasing concentrations in 200 l of binding buffer containing 2% FBS in DPBS ϪCa 2ϩ , ϪMg 2ϩ (Gibco). Incubation of the cells and antibody was for 1 h at 4°C. The cells were washed twice with binding buffer and then the 2nd antibody anti-LC -FITC (Sigma, SAB4700605) was added to the cells. The final concentration of anti-LC -FITC for human Fc␥Rs binding was 200 g/ml; for cyno Fc␥Rs binding was 40 g/ml. The bound IgG1 was detected by labeling with an anti-LC -FITC. The complex was then fixed using a diluted formalin solution 10% neutral buffer (Sigma, HT5011) before detection using a flow cytometer (BD Bioscience FACSCalibur TM ). Ten thousand events were captured and the mean fluorescence intensity (MFI) was plotted versus the antibody concentration.
Binding of SEFLs to Cynomolgus Peripheral Granulocytes-SEFL antibodies were assessed for surface binding to cynomolgus leukocytes as follows: heparinized cynomolgus blood was incubated with a dose titration of parental or SEFL antibodies or anti-streptavidin (negative control antibody) at final concentrations of 200 to 0.03 g/ml for 30 min at room temperature. One microliter (1 l) of anti-human IgG-FITC was added to each well (final concentration of 10 g/ml), and incubated an additional 30 min at room temperature. Two ml of 1ϫ ammonium chloride lysing solution (10ϫ solution prepared as 1.5 M ammonium chloride, 100 mM sodium bicarbonate, 10 mM EDTA, pH 7.4) was added to each well to lyse red blood cells (RBC), and the mixture was incubated at room temperature in the dark for 15 min. Leukocytes were centrifuged at 300 ϫ g for 5 min, and washed twice in FACS buffer (2% fetal bovine serum (FBS), 0.1% sodium azide, PBS, pH 7.4) and suspended in 200 l of FACS buffer prior to acquisition on a FACSCantoII TM flow cytometer (BD Biosciences). The granulocyte population was gated using SSC versus forward scatter, and MFI was determined.
Antibody-dependent Cell-mediated Cytotoxicity Assay (ADCC) Assay-Cryopreserved cynomolgus monkey PBMCs were obtained from SNBL, USA, Ltd., and thawed just before testing in the ADCC assay according to the vendor's protocol. Daudi cells (1 ϫ 10 4 cells/well) were seeded in a 96-well plate with or without serially diluted mAbW.IgG1, mAbW.IgG2, mAbW. SEFL1.1, and mAbW.SEFL2.0 in a total volume of 50 l/well. After incubation at room temperature for 30 min, 25 l of the cynomolgus PBMC (2 ϫ 10 5 cells/well) were added to each well of the plate. All cells and antibody molecules were suspended in RPMI1640 medium containing 4% Ultra-low IgG FBS (Gibco). After incubation at 37°C in 5% CO 2 for 22 h, lactate dehydrogenase activities of cell culture supernatant were measured by using the Cytotoxicity Detection Kit from Roche Applied Science. The percentage of cytotoxicity was calculated as described in the manufacturer's protocol.
CDC Assay-Raji lymphoma cells were cultured in medium consisting of RPMI (Life Technologies), 10% FBS (Life Technologies), 100 units/ml of penicillin, 100 g/ml of streptomycin (Life Technologies), and 2 mM L-glutamine (Life Technologies). The Raji cells were plated in RPMI medium at a density of 2.5 ϫ 10 5 cells/well into a 96-well plate. Diluted antibody was added to final concentrations of 1, 10, and 100 g/ml in a final volume of 125 l. The suspension was kept at room temperature for 10 min before addition of the complement serum to the cells. 25 l of human complement serum (QUIDEL, San Diego, CA) was added to the cell suspension in each well and the suspension was incubated overnight at 37°C/5% CO 2 . The next day, propidium iodide (PI) was added to final concentration 5 g/ml, 15 min before flow cytometry analysis. The dead cells that are PI positive were analyzed by FACSCalibur TM (BD Bioscience). Percentage of PI positive cells was measured against total gated cells.
Antibody Reagents for in Vitro Assays-The following detection antibodies used for flow cytometric analyses were purchased from BD Biosciences: CD14-PECy7 (number 557742), CD41a-APC (number 559777; mouse anti-human GPIIb), and CD62P-PE (number 550561; mouse anti-human P-selectin). Anti-human IgG-FITC (number A80-319F) was purchased from Bethyl Laboratories (Montgomery, TX). Antibody mAbV is an in-house generated human IgG2 that was used as an isotype control for SEFL antibodies. The mAbZ.IgG2 used as a positive control for surface binding studies on leukocytes is a proprietary human IgG2 antibody. Test articles (parental and modified antibodies) were supplied in vehicle solution at concentrations of 70 mg/ml (mAbX.IgG2, mAbY.IgG2, and mAbW.IgG1) or 30 -40 mg/ml.
Sample Preparation-Blood from cynomolgus macaques and human donors was collected into acid citrate dextrose for platelet phagocytosis and activation, and into sodium heparin for flow cytometric analyses of leukocyte binding. Blood samples were used within 2 to 4 h of collection for most studies except when indicated otherwise. For studies involving platelets, blood was collected from subjects not exposed to aspirin, ibuprofen, or other anti-inflammatory analgesics for the previous 7 days. Whole blood samples from human donors or cynomolgus macaques were centrifuged at 170 ϫ g for 15 min without braking. Platelet-rich plasma was collected from the top layer and used for platelet activation and phagocytosis experiments.
Phagocytosis Assay-The in vitro cynomolgus monocyteplatelet phagocytosis assay was performed as described elsewhere (10). In brief, peripheral blood leukocytes (PBL) and platelets were isolated from blood samples of cynomolgus macaques and the buffy coat layer was enriched for PBL. Platelets in platelet-rich plasma were labeled with Cell Tracker TM Green CMFDA (5-chloromethylfluorescein diacetate; Invitrogen, Eugene, OR) at a final concentration of 20 M for 30 min at room temperature. After washing with PBS, labeled platelets were resuspended in RPMI1640, 10% FBS with a final concentration of 1 ϫ 10 8 cells/ml. 100 l of CMFDA-labeled platelets were then incubated with 100 l of PBL at 5 ϫ 10 6 cells/ml from an autologous cynomolgus donor with 20 l of various concentrations of test articles in the dark at 37°C. After 6 h, the reaction was stopped by transferring the plates to ice. Extracellular fluorescence was quenched by incubating 2 min with 100 l of 0.1% trypan blue (Life Technologies, number 15250). After washing, cells were labeled with anti-CD14-PE-Cy7 and propidium iodide (Life Technologies, number P3566) for 30 min at 4°C. CMFDA fluorescence in CD14-expressing monocytes was analyzed by a BD LSRII TM flow cytometer (BD Biosciences) and the data analyzed using FlowJo software (Treestar). The increase of CMFDA fluorescence intensity was used to estimate effects of various reagents on the phagocytic capacity of monocytes for platelets.
Platelet Activation Assay-Five l of test articles mAbX.5 (0.2-7.6 mg/ml) and controls (5 mg/ml) mAbV.IgG2 (human IgG2 isotype control), 5 M 4-phorbol 12-myristate 13-acetate (positive control), 5 mg/ml of mAbX.IgG2 (positive control for cynomolgus monkey), and 5 l of vehicle control (A52SuTacetate sucrose Tween buffer at pH 5.2) were incubated with 20 l of acid citrate/dextrose anticoagulated whole blood at the above concentrations for 20 min at room temperature. After incubation, anti-CD41a-APC (for identification of platelets) and anti-CD62P-PE (for assessment of activation) were added and incubated for another 20 min at room temperature. Lyse/ fixative solution was added to each sample as the last step and they were analyzed on a FACSCalibur TM (BD Biosciences) and data were analyzed using FlowJo software (Treestar). Activated platelets were identified by an increase in MFI of CD62P compared with resting platelets as previously described (10).
Generation of SEFL mAbs-The SEFL mAbs were generated at Amgen from unamplified Chinese hamster ovary (CHO) cell pools. The SEFL mAbs were recovered from clarified CHO cell condition media using a three-step process. First, the mAbs were affinity-captured using MabSelect SuRe (32) (GE Healthcare Life Sciences). This was then followed by cation exchange chromatography. Finally, the mAbs were dialyzed into sodium acetate buffer for long term stability. The engineering and characterization of these mAbs are presented in the accompanying article (22).