Recombinant Dimeric IgA Antibodies against the Epidermal Growth Factor Receptor Mediate Effective Tumor Cell Killing

Dimeric IgA Abs contribute significantly to the humoral part of the mucosal immune system. However, their potential as immunotherapeutic agent has hardly been explored. In this article, we describe the production, purification, and functional evaluation of recombinant dimeric IgA against the epidermal growth factor receptor. Human joining chain-containing IgA was produced by nonadherent Chinese hamster ovarian (CHO)-K1 cells under serum-free conditions. Purification by anti-human κ and anti–His-tag affinity, as well as size exclusion chromatography, resulted in a homogenous preparation of highly pure IgA dimers. Functional studies demonstrated dimeric IgA to be at least as effective as monomeric IgA in triggering Ab-dependent cellular cytotoxicity by isolated monocytes or polymorphonuclear cell and in human whole-blood assays. Importantly, dimeric IgA was more effective in F(ab)-mediated killing mechanisms, such as inhibition of ligand binding, receptor downmodulation, and growth inhibition. Furthermore, only dimeric but not monomeric IgA or IgG was directionally transported by the polymeric Ig receptor through an epithelial cell monolayer. Together, these studies demonstrate that recombinant dimeric IgA Abs recruit a distinct repertoire of effector functions compared with monomeric IgA or IgG1 Abs.

A ntibodies constitute an integral part of the adaptive immune system (1), in which IgA Abs as part of the mucosal system protect large serosal surface areas from invasion by pathogens and toxins (2). In humans, two IgA isotypes, IgA1 and IgA2, are distinguished, which differ, for example, in their glycosylation profiles, the length of their hinge regions, F(ab) arm orientation, H and L chain linkage, and functional activity (3,4). In contrast with IgG, IgA Abs bind to distinct cellular receptors (5) and exist as monomeric, dimeric, and secretory isoforms. Although monomeric IgA is predominantly produced by plasma cells in the bone marrow, plasma cells in the lamina propria mainly secrete dimeric IgA. This dimer formation requires the presence of the so-called tailpiece, an 18 aa C-terminal extension of the IgA Fc part (6). Penultimate cysteines of two IgA monomers covalently bind to cysteines in the 15-kDa joining (J) chain to form dimeric IgA. Incorporation of the J chain into dimeric IgA is required for its transport onto mucosal surfaces, because the presence of J chain is essential for IgA binding to the polymeric Ig receptor (pIgR) on the basolateral surface of epithelial cells (7). After endocytotic internalization and transcytosis, pIgR is cleaved at the luminal surface, releasing secretory IgA, which consists of dimeric IgA covalently bound to the extracellular domain of pIgR (8).
Effector functions of IgA Abs include pathogen neutralization, receptor blockade, oxidative burst, phagocytosis, and Ab-dependent cellular cytotoxicity (ADCC). Although many of these functions are similar to those of IgG Abs, significant differences have also been described. Although IgG1 Abs efficiently activate the classical complement pathway, IgA Abs do not bind C1q. Furthermore, IgA Abs bind to a different set of cellular FcRs compared with IgG Abs. Among these heterogeneous IgA-binding molecules (5), the myeloid receptor for IgA (FcaRI, CD89) is probably the best characterized receptor, which is expressed by monocytes/ macrophages and polymorphonuclear cells (PMN). Interestingly, phagocyte activation by IgA is often stronger than by IgG Abs (9), although both receptor systems use the same ITAM-dependent intracellular signaling cascades (10). In addition, dimeric and secretory IgA are tetravalent for their respective Ags, which distinguishes them from bivalent IgG Abs. Increased Ab valency has been demonstrated to improve the efficacy of artificially dimerized or engineered IgG Abs (11,12).
IgA is the second most prevalent Ab isotype in blood and has been estimated to be the most abundantly produced Ig in humans (3). Generation of IgA responses has been recognized as an important aim in vaccine studies (13). However, passive immunotherapy with IgA Abs has hardly been investigated (14), whereas targeted therapy with IgG Abs is a rapidly evolving field (15).
One of the most widely investigated tumor target Ags is the epidermal growth factor receptor (EGFR), which is a signaling membrane tyrosine kinase (16). Today, two IgG Abs against the EGFR are FDA approved (17), and 14 others are tested in clinical trials. With the exception of panitumumab, which is of human IgG2 isotype, all these EGFR Abs are of human IgG1 isotype. Human IgG1 has been the preferred isotype for tumor therapy, because IgG1 effectively recruits NK cells for ADCC and may trigger complement-dependent cytotoxicity (CDC) (18). However, human IgG1 was significantly less effective than monomeric IgA in recruiting human neutrophilic phagocytes for tumor cell killing (19)(20)(21)(22). Phagocytes were demonstrated to constitute an important effector cell population for Ab therapy in syngenic animal models (23), and the contribution of PMN for tumor surveillance is actively investigated (24).
In this article, we describe the production, purification, and biochemical, as well as functional, characterization of recombinant dimeric IgA Abs against EGFR. Thus, J chain-containing dimeric IgA proved more effective in F(ab)-mediated killing mechanisms compared with monomeric Abs. Dimeric and monomeric IgA were similarly effective in triggering ADCC by monocytes and PMN. Interestingly, dimeric but not monomeric IgA or IgG was actively transported through an epithelial monolayer, suggesting therapeutic dimeric IgA may reach serosal surfaces. Thus, dimeric IgA may constitute a promising Ab isotype for cancer immunotherapy, which displays enhanced tumor cell killing by different modes of action and altered pharmacokinetics compared with IgG1 Abs.

Materials and Methods
Experiments reported in this article were approved by the Ethical Committee of Christian-Albrechts-University (Kiel, Germany) in accordance with the Declaration of Helsinki.

Cell lines
Human epidermoid carcinoma cell line A431 (German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany), baby hamster kidney 21 (BHK-21) cells cotransfected with FcaRI (CD89) and FcR g-chain (20), and murine BaF3 cells transfected with human EGFR were all kept in RPMI 1640; Madin-Darby canine kidney (MDCK) cells (European Cell Culture Collection, Salisbury, U.K.) in MEM supplemented with 1% L-glutamine (100 mM; Invitrogen, Carlsbad, CA); and human colon carcinoma cell line DiFi (European Collection of Cell Cultures) and human lung carcinoma cell line Calu3 (American Type Culture Collection, Rockville, MD) in DMEM. All culture media were supplemented with 10% heat-inactivated FCS, 100 U/ml penicillin, and 100 mg/ml streptomycin (all from Invitrogen). Selection pressure for BHK transfectants was maintained by adding 1 mg/ml geneticin (PAA, Pasching, Austria) for FcaRI, and 20 mM methotrexate (Sigma, St. Louis, MO) for FcR g-chain; for human pIgR transfected MDCK cells by adding 1 mg/ml hygromycin B (PAA); and for EGFR transfected BaF3 cells by adding 1 mg/ml geneticin and 5% (v/v) supernatant of WEHI-3B (German Collection of Microorganisms and Cell Cultures) cells to supply murine IL-3.

225-IgA production and purification
225-IgA1 was produced from the variable regions of the 225 (cetuximab) Ab as described previously (25). DNA coding for the human J chain (26) was first subcloned into the pcDNA6His vector (Invitrogen) to add an N-terminal 6 3 polyhistidine region, and further subcloned into the pIR-ESpuro3 vector (Clontech, Mountain View, CA) containing the internal ribosome entry site of the encephalomyocarditis virus to enhance J chain expression under puromycin (Invitrogen) selection pressure. Best producing monomeric 225-IgA clones were cotransfected with this vector for production of J chain-containing 225-IgA dimers. Selected transfectomas were grown in disposable CELLine CL 1000 bioreactors (Sartorius, Goettingen, Germany) under serum-free suspension culture conditions. For determination of specific production rates, 5 3 10 6 cells/ml were seeded in 20 ml per bioreactor and grown for 7 d. A total of 500 ml supernatant was collected after 1 h as initial titer and at day 7 as harvest titer. For control experiments, mouse/human chimeric 225-IgG1 (cetuximab) was bought from Merck (Darmstadt, Germany). 225-IgA was affinity-purified as described previously (25) using Capture Select Fab k chromatography media (Capture Select, Naarden, The Netherlands) and prepacked Superdex 200 26/60 columns (GE Healthcare). To separate IgA associated with the hexa-histidine-tagged J chain from IgA without J chain, we performed Ni 2+ immobilized metal ion affinity chromatography using HisTrap HP columns (GE Healthcare) according to manufacturer's instructions. Subsequently, gel filtration with a Superdex 200 26/600 column was performed to separate dimeric from monomeric and polymeric IgA. For analytical size exclusion chromatography, the Superdex 200 26/600 column was used. All purification steps were run on an Ä KTAprime liquid chromatography system (GE Healthcare). UV absorbance at 280 nm, pH, and conductivity of the effluent stream were continuously recorded and analyzed using Unicorn 4.11 software (GE Healthcare). Determination of Ab concentrations and specific production rates was done as described earlier (25).

Gel electrophoresis
Purified Ab preparations were resolved by SDS-PAGE as described earlier (25). For detection of the hexa-histidine-tagged J chain, monoclonal mouse anti-Penta-His Ab was used according to manufacturer's instructions (Qiagen). Molecular mass and purity of Ab preparations were analyzed by silver staining of proteins separated under denaturing, nonreducing conditions on 3-8% Tris-acetate gels using Roti-Black-P Kit (Roth, Karlsruhe, Germany) or by Coomassie staining of proteins separated under native conditions on 4-16% Bis-Tris gels using Native PAGE Kit (Invitrogen), both according to manufacturer's instructions.

Flow cytometry, inhibition of epidermal growth factor binding, and EGFR downmodulation
To measure EGFR and FcaRI binding, flow cytometric analyses were done as described earlier (25). For ligand-blocking experiments, A431 cells were incubated with epidermal growth factor-Alexa Fluor 488-biotinstreptavidin complex (Invitrogen) and increasing concentrations of 225-IgA or control Abs. Downmodulation of EGFR was analyzed on murine BaF3 cells transfected with human EGFR (25), which were incubated with increasing concentrations of 225-IgA and control IgA for 24 h. Residual surface EGFR was detected with Alexa Fluor 488-labeled 425 Ab, using the DyLight Fluor Ab Labeling Kit (Pierce, Rockford, IL). Results were calculated using relative fluorescence intensities (RFI) as follows: % EGFR downmodulation = 100 2 (RFI m425-FITC/RFI sample) 3 100. All samples were analyzed on a Coulter EPICS XL-MCL flow cytometer (Beckman Coulter, Fullerton, CA), collecting 1 3 10 4 events for each experimental value. Data were analyzed using XL-System II V3.0 software. RFI were calculated as the ratio of mean linear fluorescence intensity of relevant to irrelevant isotype-matched Abs.

Growth inhibition and induction of apoptosis assays
Growth inhibition of DiFi colon carcinoma cells was analyzed using the 3-(3,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium inner salt (MTS) assay. Cells were seeded at 4 3 10 4 cells/ well and treated with serial dilutions of EGFR or control Abs. After 72 h, MTS substrate was added, and absorption at 490 nm was measured after 24 h. All experimental points were set up in triplicates, and experiments were performed at least three times. Viable cell mass in the presence of control Ab served as reference (100% cell growth) to calculate growth inhibition by EGFR Abs according to the formula: absorption (EGFR Ab)/ absorption (control Ab) 3 100. To measure apoptosis induction, we seeded 5 3 10 5 cells in 2 ml culture media in six-well plates. Control Abs were added at final concentrations of 20 mg/ml. After incubation for up to 72 h, total protein preparations of cells were collected and analyzed by bicinchoninic acid assay (Bio-Rad). A total of 15 mg proteins was resolved under denaturing reducing conditions on 4-12% Bis-Tris gels (Invitrogen). After gel electrophoresis, proteins were blotted onto polyvinylidene difluoride membranes (GE Healthcare). Poly(ADP-ribose) polymerase (PARP) cleavage was detected using a monoclonal rabbit Ab specific for full-length and cleaved human PARP, and a peroxidase-labeled polyclonal goat anti-rabbit IgG Ab (both Cell Signaling Technology, Boston, MA).

Transcytosis assay
MDCK cells were transfected with cDNA coding for the human pIgR gene, which was kindly provided by Dr. Charlotte Kaetzel (University of Kentucky, Lexington, KY) (27). Expression of pIgR was confirmed by immunofluorescence using mouse anti-human-pIgR Ab (SPM217; GeneTex, Irvine, CA). For transcytosis assays, cells were trypsinized and seeded onto 0.4-nm porous filters (Costar Corning Transwell-Clear-Filters; Corning Life Sciences, Lowell, MA). After 3 d (7 for Calu3 cells), confluence of monolayers was determined by light microscopy and transepithelial resistance measurement using the Millicell-ERS system (Millipore, Billerica, MA). Monolayers were used for transcytosis experiments when transepithelial resistance was .400 V (1.5 kV for Calu3 cells) (28). Monomeric or dimeric 225-IgA or control IgA (50 mg each) was added into basolateral compartments. 225-IgG1 served as leakage control in all basolateral media. In control experiments, dimeric IgA was added into the apical medium to confirm the direction of transcytosis. Untransfected MDCK cells served as negative control. After 24 h of incubation, supernatants of both apical and basolateral compartments were collected and subjected to immunoblot analyses. Thus, 5 ml supernatants and 200 ng control proteins were loaded onto denaturing 3-8% Tris-Acetate gels (Invitrogen). After gel electrophoresis, proteins were blotted onto polyvinylidene difluoride membranes (GE Healthcare). Membranes were probed with HRP-conjugated monoclonal mouse anti-human IgA, IgG, or secretory component (SC) Abs (all Sigma), respectively.

ADCC assays
ADCC was measured using a 51 Cr release assay as described previously (21). In brief, citrate-anticoagulated blood from healthy volunteers was layered over a discontinuous Percoll (Biochrom, Berlin, Germany) gradient consisting of 70 and 63% Percoll. After centrifugation, mononuclear cells were collected from the plasma/Percoll interface and PMN from the interface between the two Percoll layers. Monocytes were further isolated from mononuclear cells by CD14 + selection using magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany). Whole blood or isolated effector cells and sensitizing Abs at various concentrations were added to round-bottom microtiter plates (Wallac, Turku, Finland). Assays were started by adding effector (E) and target (T) cells at an E:T ratio of 80:1. After incubation at 37˚C (3 h for whole blood and PMN assays, 12 h for monocytes), aliquots of supernatants were transferred into 96-well plates containing a scintillation mixture (OptiPhase Scintillator Supermix; Per-kinElmer). 51 Cr release was measured in cpm using a scintillation counter (MetaBase TriLux; PerkinElmer). Percentage of cellular cytotoxicity was calculated using the formula: % specific lysis = (experimental cpm 2 basal cpm)/(maximal cpm 2 basal cpm) 3 100, with maximal 51 Cr release determined by adding Triton X-100 (1% final concentration) to target cells, and basal release as measured in the absence of sensitizing Abs and effector cells. Ab-independent cytotoxicity (effectors without target Abs) was observed in whole-blood assays, but not with PMN or monocytes.

Data processing and statistical analyses
Data are displayed graphically and analyzed statistically using GraphPad Prism 5.0. Group data are reported as mean 6 SEM. Differences between groups were analyzed by unpaired (or, when appropriate, paired) Student t tests. EC 50 values were calculated from dose-response curves, reported as mean 6 SEM and compared by unpaired Student t test to calculate significant differences between data groups. Significance was accepted when p values were ,0.05.

Results
Production of J chain-containing dimeric 225-IgA Nonadherent Chinese hamster ovarian (CHO)-K1 cells expressing the EGFR-directed 225-IgA Ab under serum-free conditions (25) were cultured with 10% FCS for 24 h before transfection with a vector containing human J chain and the puromycin resistance gene. After transfection, J chain and IgA coexpressing clones were selected by puromycin and L-methionine sulfoximine double selection, and readapted to serum-free conditions. In the following weeks, best producing clones were isolated by limiting dilution cloning and checked for Ab production by an IgA-specific ELISA. Best producing clones were cultured in special cell line production flasks. Using this serum-free suspension culture system, we obtained median Ab concentrations of 200.8 (673.3) and 295.2 (6139.5) mg/ml for monomeric and dimeric 225-IgA, resulting in total Ab yields of 6.2 and 7.9 mg/wk/flask for monomeric and dimeric 225-IgA, respectively (Supplemental Table I). The IgA1 isotype was chosen for this study because disulphide bridges are formed between H and L chains, similar to IgG Abs and in contrast with IgA2m(1), which forms covalently bound homodimers of H or L chains, respectively. These biochemical characteristics allow IgA1 Abs to be correctly analyzed also under denaturing conditions, for example, in SDS-PAGE.

Purification and biochemical characterization of dimeric 225-IgA
For purification of J chain-containing dimeric 225-IgA, two different affinity and one size exclusion chromatography were combined. First, a human k-L chain-specific affinity chromatography was used to isolate human IgA Abs from serum-free cell culture supernatants. Next, immobilized metal ion affinity chromatography separated His-tagged J chain-containing molecules from monomeric IgA or from spontaneous IgA dimers without incorporated J chain. Finally, size exclusion chromatography with a preparative Superdex column served to select dimeric 225-IgA (elution volume, 128-150 ml) from higher polymeric forms (elution volume, 110-128 ml). In Supplemental Fig. 1, eluates of the different purification steps were analyzed by gel filtration and Western blots, using anti-human a-chain Abs for detection. Thus, anti-k chromatography resulted in three fractions (Supplemental Fig. 1A), which represented polymeric, dimeric, and monomeric IgA (fraction 1), monomeric IgA (fraction 2), and a-H and -L chain heterodimers (fraction 3; Supplemental Fig. 1B). After Histag-directed affinity chromatography of fraction 1, two fractions were obtained (Supplemental Fig. 1C), which contained polymeric, dimeric, and monomeric IgA (fraction1), and predominantly dimeric IgA (fraction 2), respectively (Supplemental Fig.  1D). When fraction 2 was reanalyzed by size exclusion chromatography (Supplemental Fig. 1E1), a distinct peak was detected, which was shown to represent predominantly dimeric IgA (Supplemental Fig. 1F, lane 1). Purified monomeric IgA is shown for comparison (Supplemental Fig. 1E2, 1F, lane 2).
Molecular mass and purity of J chain-containing dimeric 225-IgA were analyzed by gel electrophoresis under denaturing and native conditions using silver stain or Coomassie blue for protein detection, respectively (Fig. 1A, 1B). Both methods demonstrated the purified 225-IgA preparation to consist predominantly of a molecule of the expected molecular mass of ∼320 kDa (Fig. 1A,  lane 5, 1B, lane 4). Association with the human J chain was demonstrated by staining for the introduced His-tag. Thus, Western blots and indirect immunofluorescence with EGFR-expressing A431 cells stained positive with His-tag Abs in the presence of dimeric 225-IgA (Fig. 1C, lane 4, 1D, second panel), but not in the presence of control Ab preparations. Western blots with antihuman a-H or k-L chain Abs confirmed the correct assembly of 225-IgA (Fig. 1E, lane 4).
To investigate Fc-mediated effector functions, we analyzed binding of recombinant 225-IgA isoforms to FcaRI/FcRgcotransfected BHK cells by indirect immunofluorescence. Expression of human FcaRI on transfected BHK cells was confirmed by staining with CD89 Abs (data not shown). All three IgA Ab preparations demonstrated concentration-dependent binding to transfected BHK cells, whereas the 225-IgG control Ab did not bind (Fig. 3A). Binding of dimeric 225-IgA was significantly more effective than binding of monomeric 225-IgA (EC 50 : 498.1 6 0.57 versus 1137 6 29.37 nM; p , 0.001; Supplemental Fig. 2C, 2D). In this assay, binding of control IgA (EC 50 : 717.5 6 0.57 nM) was lower than that of dimeric 225-IgA, but higher than that of monomeric 225-IgA, which is probably explained by the high polymer content of control IgA (see Fig. 1A, lane 2). In ADCC assays with isolated monocytes as effector cells, all three EGFR Abs were similarly effective in triggering tumor cell killing (EC 50 : 1.82 6 0.37, 0.79 6 0.26, and 1.96 6 0.14 nM for 225-IgG, dimeric 225-IgA, and monomeric 225-IgA, respectively), whereas a control IgA Ab was not effective (Fig. 3B). However, EC 50 of dimeric 225-IgA was significantly lower than EC 50 of monomeric 225-IgA (p , 0.01). In contrast, both 225-IgA isoforms required lower Ab concentrations than the IgG1 Ab (EC 50 : 5.61 6 1.48, 3.66 6 0.45, and 9.98 6 1.90 nM for monomeric IgA, dimeric IgA, and 225-IgG, respectively) with PMN effector cells (Fig.   3C), with dimeric 225-IgA being significantly more effective than its monomeric counterpart (p , 0.05). Next, ADCC assays with human whole blood as effector source were performed. In these assays, both IgA preparations were more effective then their IgG1 counterpart, with dimeric IgA triggering particularly effective tumor cell lysis (Fig. 3D).

Transcytosis of dimeric 225-IgA
J chain is required for the directed transport of dimeric IgA through an epithelial cell monolayer-a process mediated by pIgR and called transcytosis (7). Polarized MDCK cells were stably transfected with human pIgR, to investigate transcytosis. Expression of transfected pIgR was documented by indirect immunofluorescence (RFI: 40.1 6 1.1). Next, binding of different IgA preparations to pIgR-transfected MDCK cells was investigated (Fig.  4A). As expected, only J chain-containing dimeric 225-IgA demonstrated concentration-dependent binding with an EC 50 of 5.9 mg/ ml, whereas monomeric 225-IgA and a control IgA did not bind to pIgR. A dimeric IgA preparation, which was separated by size exclusion chromatography from a non-J chain-expressing transfectoma, did not bind to pIgR, confirming the requirement of J chain for pIgR binding.
In classical transcytosis assays (Fig. 4B), media from the basolateral and apical compartments were analyzed by Western blot after the indicated time periods for the absence or presence of SC, IgG, or IgA. In Fig. 4B, the left panel represents the staining controls, whereas the right panel was obtained from nontransfected MDCK cells. 225-IgG was added to the basolateral compartment and served as leakage control. After 24 h (t24), dimeric IgA was quantitatively transported from the basolateral to the apical compartment (↑), whereas no transport in the other direction (↓) was observed. Interestingly, transported dimeric 225-IgA demonstrated the expected shift in molecular mass and stained positive for SC. Furthermore, secretory 225-IgA recovered from the apical compartment was capable to mediate growth inhibition of DiFi cells (Fig. 4C). As expected, monomeric IgA was not transcytosed.
In addition, we performed transcytosis assays with human Calu3 cells (Supplemental Fig. 3A), which endogenously express human pIgR and low levels of EGFR (Supplemental Fig. 3B). Also in this model, selective and time-dependent transport of dimeric 225-IgA was observed with a calculated median transportation rate of 0.35 nM/h (Supplemental Fig. 3C).

IgA Abs for immunotherapy
In this article, we describe effector functions of a recombinant J chain-containing dimeric IgA Ab against EGFR and compare its efficacy with monomeric IgA and IgG1 Abs. Dimeric and secretory IgA have been demonstrated to play an important role in the protection of serosal surfaces (2), and generation of an IgAmediated immune response has been recognized as an important aim for many vaccine studies (13). However, passive immunotherapy with IgA Abs has hardly been explored, although Ab therapy with IgG Abs is clinically successful in many different diseases (15). Several potential reasons can be identified to explain the slow progress of clinical development of IgA Abs: production and purification are still considered more difficult for IgA than for Results from three experiments are presented as mean 6 SEM of "% inhibition of EGF binding." C, Ab-induced EGFR downmodulation was investigated using EGFR-transfected BaF3 cells and increasing concentrations of the indicated Abs. After 24 h, residual EGFR was detected using fluorescein-labeled m425. Dimeric 225-IgA effectively mediated receptor downmodulation. Results from three different experiments are presented as mean 6 SEM of "% EGFR downmodulation." D, Growth inhibition of DiFi colon carcinoma cells was analyzed by MTS assay. Results from three experiments are presented as "% growth inhibition." In all four assays, dimeric 225-IgA was significantly more effective than monomeric 225-IgA, whereas a control IgA Ab had no effect. Significant effects compared with control Ab are indicated by asterisks, whereas significant differences in EC 50 between dimeric 225-IgA and monomeric 225-IgA are marked by number signs. E, EGFR Ab-induced PARP cleavage was investigated as marker for apoptosis induction in DiFi cells. The indicated Abs were added for 6, 24, or 72 h. Isolated proteins were loaded on denaturing gels and stained with monoclonal rabbit anti-human PARP, recognizing full-length and cleaved PARP, and HRP-conjugated goat anti-rabbit-IgG-Ab. Both IgA Abs mediated similar induction of apoptosis with increasing incubation periods, whereas unspecific control Abs did not induce PARP cleavage. b-actin served as loading control.
IgG Abs, although the generation of recombinant monomeric and dimeric IgA has been described (26,29,30). In addition, IgA Abs are heavily glycosylated (31), which poses challenges on a reproducible and stable production technology. Because of the lack of FcRn binding, IgA has a shorter serum t 1/2 compared with IgG. Serum t 1/2 was critical for the efficacy of IgG Abs (32) but may be less relevant for IgA Abs, which display altered pharmacokinetics and may reach tumors via alternative routes. Furthermore, mice do not express an FcaRI homolog (5), which makes the establishment of relevant small animal models more complicated. Consequently, no IgA Ab has obtained regulatory approval, adding to the business risks of their development. Recently, we described the generation of recombinant monomeric IgA, using production and purification technologies that are similar to those that are well established for IgG Abs (26). In this article, we extend these studies to the production of recombinant human J chain-containing IgA dimers. As target Ag we selected the EGFR, one of the most extensively studied target Ags in oncology against which an IgG1 and an IgG2 Ab have obtained FDA approval (15,16). However, additional studies are required to determine the characteristics of an optimized IgA Ab for clinical application (e.g., monomeric versus dimeric IgA, selection of IgA1 or one of three IgA2 isoforms, impact of Ab glycosylation and protease resistance).

Role of the Ab valency: F(ab)-and Fc-mediated effects of dimeric IgA
Dimeric IgA consists of two monomeric IgA molecules connected by the J chain, forming a near-planar structured molecule (33). The resulting dimer consists of four F(ab) and two Fc domains, and therefore has four potential Ag binding sites. Thus, dimeric IgA is expected to cross-link tumor target Ags more effectively than monomeric IgA or IgG Abs. Consequently, dimeric IgA was more effective than monomeric IgA in EGFR binding, receptor blockade, receptor downmodulation, and growth inhibition ( Fig.  2A-D), particularly at lower Ab concentrations. At saturating concentrations, both Ab preparations were similarly effective, which is probably explained by steric hindrance leading to bivalent or monovalent binding of dimeric IgA. This later mechanism may also explain similar efficacy of dimeric and monomeric IgA in PARP cleavage (Fig. 2E), because this assay was performed at saturating Ab concentrations. Results from our experiments with dimeric IgA are in agreement with studies in which artificially dimerized IgG Abs or tetravalent human CD22 Abs demonstrated enhanced tumor cell killing compared with wild-type Abs (11,12).
In ADCC assays with isolated monocytes or PMN, both dimeric and monomeric 225-IgAs were similarly effective, indicating that the presence of human J chain does not interfere with IgA binding to FcaRI. These functional studies are supported by structural data demonstrating that J chain does not interfere with the FcaRI binding site (34). Unexpectedly, dimeric IgA appeared to be more effective than monomeric IgA in whole-blood assays. Interestingly, IgA was also reported to mediate inhibitory activity (35), which we did not observe under our assay conditions. Increased valency for FcRs may explain the enhanced ADCC activity of IgA compared with IgG Abs; in contrast with IgG Abs, where one Ab Fc part binds to one FcgR, monomeric IgA interacts with FcaRI in a 1:2 stoichiometry (36). Thus, dimeric IgA could theoretically bind to four FcaRI molecules simultaneously, thereby activating and stimulating effector cells more efficiently than monomeric IgA. This tetravalency could be a possible advantage for dimeric compared with monomeric IgA or IgG Abs.

Contribution of ADCC to the effector mechanisms of therapeutic Abs
Functional studies revealed that monomeric IgA Abs recruit a different spectrum of effector mechanisms compared with IgG Abs (21). In addition to direct effector mechanisms such as blockade of growth factor-induced signaling and induction of growth inhibition or apoptosis (37), ADCC is increasingly recognized as an important mechanism of action for therapeutic cancer Abs (38). For example, tumor-directed Abs lost their therapeutic efficacy in animals with disrupted FcgR signaling (39). Because human IgG1 has been demonstrated to optimally recruit NK cells as effector cells for ADCC (17), this isotype has been chosen for most cancer Abs. However, also, human IgG2 Abs, such as panitumumab, trigger ADCC, although not by NK cells but by myeloid cells such as monocytes and PMN (40). Currently, it is unclear which effector cell type is most relevant for the therapeutic efficacy of Abs. Because most preclinical investigations used human IgG1 Abs, which preferentially interact with NK cellexpressed FcgRIIIa, results from these studies were in favor of NK cells (41). These observations triggered substantial efforts to enhance IgG binding to NK cell-expressed FcgRIIIa (42). However, syngenic animal models also demonstrated a significant contribution of phagocytic myeloid cells (22), including monocytes and tissue macrophages (43), as well as granulocytes (24). Clinical evidence for the potential relevance of myeloid cells is derived from the repeated observation that the functionally relevant 131H/R polymorphism of the myeloid cell-expressed FcgRIIa receptor significantly affects response to therapeutic Abs against different target Ags such as CD20 (44), human EGFR 2 (45), and EGFR (46). As demonstrated in Fig. 3, 225-IgA was as effective in recruiting monocytes and significantly more effective in recruiting PMN than human IgG1, suggesting that IgA Abs may be therapeutic in vivo.

Pharmacokinetics of IgA versus IgG Abs
In addition to preferentially recruiting myeloid instead of NK cells, IgA Abs display different pharmacokinetics compared with IgG Abs. IgG Abs bind to FcRn, which prevents lysosomal degradation, and thereby extends the serum t 1/2 of IgG (47). Monomeric IgA does not bind to FcRn and, therefore, has a reduced serum t 1/2 of approximately 7 d (47). In contrast, dimeric IgA, but not monomeric IgA or IgG, binds to the pIgR, which transports dimeric IgA from the basolateral to the apical compartment of epithelia (8), thereby dimeric IgA is transported into secretions where it exerts important immunological functions (2). Whether epithelial cancers can be effectively attacked from their luminal side requires further investigation, but our results (Fig. 4C) demonstrate that transported dimeric IgA is biologically active.
Monomeric IgA and IgG1 Abs reach their target Ags via the tumor vasculature. Physiologically, dimeric IgA is produced by B cells in the lamina propria, which is in close proximity to the basolateral side of epithelial cells (48). After J chain-mediated binding to pIgR, dimeric IgA is then transported through the epithelial cells to the luminal surface (7). Murine models were used to investigate whether systemically administered dimeric IgA was able to cross the epithelial barrier. For example, i.v. applied or FIGURE 4. Transcytosis of dimeric IgA. A, Binding of monomeric (mIgA) and dimeric 225-IgA (dIgA) to pIgR + MDCK cells was analyzed by indirect immunofluorescence using an anti-human k-L chain-specific FITC-labeled Ab for detection. Only J chain-containing dimeric (225-IgA+J), but not monomeric or dimeric IgA without J chain (225-IgA-J), demonstrated concentration-dependent binding to pIgR-expressing MDCK cells. Results are presented as mean RFI 6 SEM obtained from three independent experiments. B, Transcytosis of mIgA and dIgA through a confluent monolayer of human pIgRtransfected MDCK cells (MDCK-pIgR + ) was analyzed by Western blot experiments using SC, IgG-, or IgA-specific Abs for detection. 225-IgG, added to the basolateral compartment, served as leakage control and did not enter the apical compartment. As demonstrated, mIgA was retained in the basolateral compartment, whereas dIgA was transcytosed from the basolateral to the apical compartment (↑) within 24 h. This transport was unidirectional because dIgA was not transported from the apical to the basolateral compartment (↓). Untransfected MDCK cells did not transport dIgA. After transcytosis, dIgA stained positive for SC and demonstrated a higher molecular mass compared with dIgA, demonstrating in vitro generation of secretory 225-IgA. C, Supernatants from apical wells of transcytosis assays from B were used in growth inhibition assays with DiFi cells. Only dimeric IgA transported by pIgRtransfected MDCK cells (dIgA ↑ pIgR + ) mediated growth inhibition of DiFi cells, whereas apical supernatants from wells with monomeric 225-IgA (mIgA ↑ pIgR + ), dIgA in the absence of pIgR (dIgA1 ↑ pIgR 2 ), or MDCK medium alone (medium control) did not mediate growth inhibition. backpack hybridoma-produced dimeric IgA was detected on mucosal surfaces and neutralized their respective target Ags (49)(50)(51). Whether i.v. applied dimeric IgA would also reach basolateral epithelial sites in humans and then be transported to the mucosal surface is an unresolved question. As demonstrated in Fig. 4 and previously by others (7), transcytosis is an efficient process. Thus, it can be expected that dimeric IgA reaching the basolateral compartment will be quantitatively transported to the luminal side.
In conclusion, monomeric IgA has been demonstrated to effectively recruit phagocytic cells for tumor cell killing (20)(21)(22). In this article, we demonstrate that J chain-containing dimeric IgA is at least as effective in triggering ADCC as monomeric IgA. Importantly, dimeric IgA is more effective than monomeric Abs in direct effector mechanisms, such as EGFR blockade, EGFR downmodulation, and growth inhibition. Furthermore, dimeric IgA is actively transported through an epithelial monolayer. Together, these results demonstrate that recombinant dimeric IgA triggers important biological functions and constitutes an interesting Ab format for tumor therapy.