of April 11, 2018.This information is current as

Pathway Amplification LoopTumor Cells through the AlternativeDecay-Accelerating Factor Expressing Complement-Mediated Lysis ofTargeting IgG3 Triggers Epidermal Growth Factor Receptor

Valerius and Stefanie DererThies Rösner, Stefan Lohse, Matthias Peipp, Thomas

http://www.jimmunol.org/content/193/3/1485doi: 10.4049/jimmunol.1400329June 2014;

2014; 193:1485-1495; Prepublished online 27J Immunol 

Referenceshttp://www.jimmunol.org/content/193/3/1485.full#ref-list-1

, 12 of which you can access for free at: cites 41 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2014 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The Journal of Immunology

Epidermal Growth Factor Receptor Targeting IgG3 TriggersComplement-Mediated Lysis of Decay-Accelerating FactorExpressing Tumor Cells through the Alternative PathwayAmplification Loop

Thies Rosner, Stefan Lohse, Matthias Peipp, Thomas Valerius, and Stefanie Derer

Binding of C1q to target-bound IgG initiates complement-mediated lysis (CML) of pathogens, as well as of malignant or apoptotic

cells, and thus constitutes an integral part of the innate immune system. Despite its prominent molecular flexibility and higher C1q

binding affinity compared with human IgG1, IgG3 does not consistently promote superior CML. Hence the aim of this study was to

investigate underlying molecular mechanisms of IgG1- and IgG3-driven complement activation using isotype variants of the ther-

apeutic epidermal growth factor receptor (EGFR) Ab cetuximab. Both IgG1 and IgG3 Abs demonstrated similar EGFR binding

and similar efficiency in Fab-mediated effector mechanisms. Whereas anti–EGFR-IgG1 did not promote CML of investigated

target cells, anti–EGFR-IgG3 triggered significant CML of some, but not all tested cell lines. CML triggered by anti–EGFR-IgG3

negatively correlated with expression levels of the membrane-bound complement regulatory proteins CD55 and CD59, but not

CD46. Notably, anti–EGFR-IgG3 promoted strong C1q and C3b, but relatively low C4b and C5b-9 deposition on analyzed cell

lines. Furthermore, anti–EGFR-IgG3 triggered C4a release on all cells but failed to induce C3a and C5a release on CD55/CD59

highly expressing cells. RNA interference-induced knockdown or overexpression of membrane-bound complement regulatory

proteins revealed CD55 expression to be a pivotal determinant of anti–EGFR-IgG3–triggered CML and to force a switch from

classical complement pathway activation to C1q-dependent alternative pathway amplification. Together, these data suggest human

anti–EGFR-IgG3, although highly reactive with C1q, to weakly promote assembly of the classical C3 convertase that is further

suppressed in the presence of CD55, forcing human IgG3 to act mainly through the alternative pathway. The Journal of

Immunology, 2014, 193: 1485–1495.

Activation of the human complement system by cell-surface ligated Igs leads to opsonization and destruc-tion of pathogens or malignant cells and constitutes an

integral part of the innate immune system (1). Thus, complement-dependent cytotoxicity (CDC) has been supposed to constitutea pivotal mechanism of action of different therapeutic Abs in tu-mor therapy (2). However, membrane-bound complement regu-latory proteins (mCRPs), such as CD46, CD55, and CD59, areubiquitously expressed in human cells to tightly regulate localcomplement activation and protect against uncontrolled complement-mediated cell destruction. CD46 (membrane cofactor protein) actsas a cofactor of the serine protease “factor I” to degrade C3b orC4b. CD55 (decay-accelerating factor) prevents the formation of

new and accelerates the decay of preformed C3 and C5 con-vertases by associating with C4b or C3b to force dissociation ofC2b (generally designated as C2a) from classical convertases,as well as of factor Bb from alternative convertases. CD59 blocksthe assembly of the membrane attack complex by displacementof C9 (3). Hence mCRPs are often found to be highly upregulatedon tumor cells and, therefore, confer resistance against CDC. In-terestingly, small interfering RNA (siRNA)-induced knockdownof CD55 in lymphoma cells has been demonstrated to overcomeresistance against rituximab-triggered CDC (4). Similar resultshave been recently reported for the therapeutic Her2/neu-directedAbs trastuzumab and pertuzumab (5).CDC against target cells is evoked by the efficient fixation of the

initial complement component C1q on at least two Ig-Fc portions inthe vicinity to the cell membrane. However, based on the findingthat complement activation by Ig does not solely depend on C1qfixation (6), it might be hypothesized that other essential molec-ular features are required besides strong C1q binding for efficientinduction of CDC, such as Ag cell-surface expression levels, thebinding epitope of the Ab, Ags’ mobility within the cell mem-brane, as well as Abs’ isotype (7–9). This suggestion has beenstrengthened by results received from functional characterizationof the therapeutic epidermal growth factor receptor (EGFR)–tar-geting Ab cetuximab, a mouse-human chimeric IgG1 molecule.Unexpectedly, cetuximab does not trigger CDC against solid tu-mor cells as a single agent while being effective in combinationwith a second noncompeting EGFR Ab (10, 11). Besides humanIgG1, human IgG3 has been demonstrated to be highly reactivewith complement (12, 13). Although both human IgG1 and IgG3efficiently initiate CDC against target cells expressing high Ag

Division of Stem Cell Transplantation and Immunotherapy, 2nd Department of Med-icine, Christian-Albrechts-University and University Hospital Schleswig-Holstein,24105 Kiel, Germany

Received for publication February 4, 2014. Accepted for publication May 29, 2014.

This work was supported by the German Research Foundation (Grant DE 1874/1-1)and intramural funding from Christian-Albrechts-University Kiel.

Address correspondence and reprint requests to Dr. Stefanie Derer, Division of StemCell Transplantation and Immunotherapy, 2nd Department of Medicine, Christian-Albrechts-University, Schittenhelmstraße 12, 24105 Kiel, Germany. E-mail address:s.derer@med2.uni-kiel.de

Abbreviations used in this article: AP, alternative pathway; BHK-21, baby hamsterkidney 21; CDC, complement-dependent cytotoxicity; CHO-K1, Chinese hamsterovarian K1; CML, complement-mediated lysis; CRIg, complement receptor of theIg-superfamily; EGFR, epidermal growth factor receptor; mCRP, membrane-boundcomplement regulatory protein; MFI, mean fluorescence intensity; RFI, relative fluo-rescence intensity; SABC, specific Ag-binding sites per cell; w/o, without.

Copyright� 2014 by The American Association of Immunologists, Inc. 0022-1767/14/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1400329

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

levels, human IgG3 gains advantage over human IgG1 againsttarget cells expressing low Ag levels (14, 15). This difference inCDC activity has been ascribed to the 4-fold longer hinge region,as well as to the higher flexibility of human IgG3 compared withhuman IgG1, enabling IgG3 to better span widely spaced Agmolecules (13, 16, 17). Referring to these structural features, anIgG3 isotype switch variant of the therapeutic IgG1 Ab rituximabhas been demonstrated to possess superior complement-activatingcapacity against CD20 low-expressing cells such as chroniclymphocytic leukemia cells (18). However, contrasting results alsohave been reported regarding complement activation by humanIgG1 and IgG3, demonstrating lower complement-activatingcapacities for IgG3 than for IgG1 (12). Referring to these stud-ies, it might be hypothesized that structural differences betweenhuman IgG1 and IgG3 may result in distinct complement-activating mechanisms. However, in-depth analyses elucidatingthese modes of action have not been reported.Hence the aim of this study was to gain novel insights into the

mode of action and to decipher main determinants of human IgG3-driven complement activation. For this purpose, we systematicallyinvestigated underlying molecular mechanisms of IgG3-triggeredcomplement activation in the background of the therapeuticEGFR Ab cetuximab. Thus, we found anti–EGFR-IgG3 to mediatepotent C1q fixation and to promote the classical pathway exclu-sively on CD552 target cells. When target cells express CD55, anti–EGFR-IgG3 mainly triggers CDC through the alternative pathway(AP) amplification route after initial classical pathway activation.

Materials and MethodsStudy population and consent

Experiments reported in this article were approved by the Ethics Committeeof the Christian-Albrechts-University, Kiel, Germany, in accordance withthe Declaration of Helsinki. Blood donors were randomly selected fromhealthy volunteers, who gave written informed consent before analyses.

Cell lines

Human epidermoid carcinoma cell line A431 (German Collection ofMicroorganisms and Cell Cultures, Braunschweig, Germany) and babyhamster kidney 21 (BHK-21) cells transfected with the human EGFR (9)were kept in RPMI 1640. The human colon carcinoma cell line DiFi andglioblastoma cell line A1207 (European Collection of Cell Cultures, Sal-isbury, U.K.) were kept in DMEM, and Chinese hamster ovarian K1(CHO-K1) cells were kept in DMEM select according to manufacturer’sinstructions. Culture media were supplemented with 10% heat-inactivatedFCS, except DMEM select, which was supplemented with dialyzed heat-inactivated FCS. In addition, all culture media were supplemented with100 U/ml penicillin and 100 mg/ml streptomycin (all from Invitrogen,Carlsbad, CA). Selection pressure for BHK-EGFR+ transfectants 1–5 wasmaintained by adding 1 mg/ml geneticin (PAA, Pasching, Austria).

Generation and production of mAbs

Anti–EGFR-IgG1, as well as anti–EGFR-IgG3, was generated based onthe variable regions of the 225 (cetuximab) Ab, whereas control mAbswere derived from the variable regions of C2B8 (rituximab) Ab. In brief,225-k-L chain genes and 225-variable heavy genes were each clonedinto pEE14.4 GS-vector (Lonza Biologics, Slough, U.K.), resulting inpEE14.4-225-VL-k-vector and pEE14.4-225-VH-vector (19). The cDNAsequence encoding the variable k-L chain and variable H chain region ofC2B8 (patent no. US006399061B1) was de novo synthesized (Entelechon,Regensburg, Germany) and cloned in the pEE14.4 GS-vector, resulting inpEE14.4-C2B8-VL-k-vector and pEE14.4-C2B8-VH-vector. Next, theg1- or g3-genes, synthesized de novo, were subcloned into either pEE14.4-225-VH-vector or pEE14.4-C2B8-VH-vector, resulting in pEE14.4-225-VH-IgG1, pEE14.4-225-VH-IgG3, pEE14.4-C2B8-IgG1, or pEE14.4-C2B8-IgG3. The g3-gene was a Gm3 allotype of the Ig G3 family, inwhich a point mutation at position 435 was inserted, resulting in an aminoacid exchange from arginine to histidine. This mutation has been shown toconfer binding to FcRn and, therefore, to increase the in vivo half-life of anIgG3 molecule (20). CHO-K1 cells were stably cotransfected with equal

amounts of L and H chain vectors using Lipofectamine 2000 (Invitrogen)according to the manufacturer’s instructions for CHO cells. For selection,25 mMmethionine sulfoximine (Sigma-Aldrich, St. Louis, MO) was addedfor inhibition of the endogenous glutamine synthetase expression of CHO-K1 cells. The IgG1 and IgG3 mAbs were affinity purified using CaptureSelect Fab anti-human k-L chain chromatography matrices (Capture Se-lect, Naarden, the Netherlands). For analytical size exclusion chromatog-raphy, a Superdex 200 26/600 column (GE Healthcare) was used. Allpurification steps were run on an AKTAprime liquid chromatographysystem (GE Healthcare). UV absorbance at 280 nm, pH, and conductivityof the effluent stream were continuously recorded and analyzed usingUnicorn 4.11 software (GE Healthcare). Ab concentrations were deter-mined by capillary electrophoresis using the Experion system (Bio-RadLaboratories GmbH, Munich, Germany).

Growth inhibition assay

Growth inhibition of DiFi cells was investigated using the MTS assayaccording to the manufacturer’s instructions (Promega, Madison, WI).Cells were seeded in a 96-well microtiter plate at a density of 5 3 103

cells/well and treated with Abs at indicated concentrations. After 72 h ofincubation at 37˚C, MTS was added as a substrate, and absorption at 490nm was measured. Percentage of growth inhibition was calculated usingfollowing formula: % growth inhibition = absorption (EGFR Ab)/absorption(without [w/o] Ab) 3 100.

SDS-PAGE and immunoblotting

Purified Abs were separated by denaturing NaDodSO4 PAGE (SDS-PAGE)on gradient Bis-Tris gels (4–12%; Invitrogen) under reducing or nonre-ducing conditions and either stained with Coomassie brilliant blue simpleSafe stain (Invitrogen) or transferred onto polyvinylidene fluoride mem-branes (Millipore, Billerica, MA) for immunoblotting experiments.Immunoblots were performed using HRP-conjugated goat anti-humank-L chain–specific or HRP-conjugated goat anti-human IgG–specific Ab.Proteins were visualized by chemiluminescence (ECL; Thermo FisherScientific, IL).

Flow cytometric analyses

EGFR binding experiments were carried out by indirect immunofluores-cence as described previously (21). Relative fluorescence intensity (RFI)was determined by the following formula: mean fluorescence intensity(MFI) EGFR Ab/MFI control Ab. EGFR, CD46, CD55, and CD59 ex-pression were quantified using the following murine Abs: EGFR Ab m425,CD46 Ab 0846 (Immunotech Laboratories, Monrovia, CA), CD55 AbS031 (BD, Franklin Lakes, NJ) or the CD59 Ab mem43, and the QIFIKIT(DAKO, Glostrup, Denmark), according to the manufacturer’s instructions.A murine CD20 Ab served as a control. Immunofluorescence was analyzedon a flow cytometer (Epics Profile; Beckman Coulter, Fullerton, CA).

Complement deposition

Deposition of complement components on the cell surface of target cellswas determined by incubation of 1.5 3 105 cells with EGFR-directed orcontrol Abs (all at 66.67 nM), as well as with a combination of two non–cross-blocking EGFR Abs (anti–EGFR-IgG1 and 003, each at 33.34nM), serving as a positive control (10), for 15 min at 4˚C. Subsequently,25% v/v human serum was added, followed by incubation for 10 min at37˚C. After washing, samples were either stained with polyclonal FITC-conjugated C1q or C4(b, c) Abs (both from DAKO) or with mouse anti-human (i)C3b mAb (Thermo Fisher Scientific), as well as with mouseanti-human Factor Bb or C5b-9 mAb (both from Quidel, San Diego,CA), followed by staining with FITC-conjugated goat anti-mouse Fcgfragment-specific F(ab)2 (Jackson Immunoresearch Laboratories, WestGrove, PA). Samples were analyzed on a flow cytometer (Epics Profile;Beckman Coulter).

Anaphylatoxin release

Fresh human plasma (25% v/v), supplemented with 25 mg/ml lepirudin(Refludan; Pharmion, Hamburg, Germany), sensitizing Abs (66.67 nM) orAb combination (33.34 nM each mAb), and RPMI 1640 (10% FCS) wereadded to a 96-round-well plate. As a control, target cells were incubatedonly with RPMI 1640 (10% FCS). The assay was started by adding 5 3104 target cells in 50 ml culture medium, therefore resulting in a finalvolume of 200 ml/well. After 3 h at 37˚C, anaphylatoxin release from thesupernatants was determined by calibrated flow cytometry using the Hu-man Anaphylatoxin Kit (BD Biosciences, San Diego, CA) according to themanufacturer’s instructions.

1486 CD55 CONTROLS IgG3-DRIVEN COMPLEMENT ACTIVATION

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

CDC assays

CDC assays were performed by preincubating target cells with 200mCi [51Cr]for 2 h. Afterward, cells were resuspended in PBS (1X) supplementedwith 0.11 mM Mg2+ (PBS (1X) supplemented with 0.5 mM Mg2+ and0.15 mM Ca2+ in the case of experiments with factor B–depleted serum)or cell culture medium in the case of BHK-EGFR transfectants. To blockcomplement regulatory activity of CD55, we used mouse anti-humanCD55 (66.67 nM, BRIC216, mouse IgG1; Bio-Rad) blocking mAb inCDC experiments at saturating concentrations. To inhibit the AP ofcomplement activation, we used a fusion protein of the human AP inhibitorcomplement receptor of the Ig-superfamily (CRIg) and human IgG1-Fc(Genentech, San Diego, CA) in CDC experiments at 10 mg/ml. As thesource of complement, 25% v/v freshly drawn human serum, 25% v/vC1q-depleted serum, or 12.5% v/v factor B–depleted serum (Com-plement Technology) was used. C1q-depleted serum was reconstitutedwith 100 mg/ml C1q (Complement Technology), whereas factor B–depletedserum was reconstituted with 200 mg/ml factor B (Complement Tech-nology). Percentage of [51Cr] release was calculated using followingformula: % lysis = (experimental cpm 2 basal cpm)/(maximal cpm 2basal cpm) 3 100.

siRNA-induced knockdown experiments

A431 cells were seeded at a density of 1 3 106/well in 10-cm plates.Twenty-four hours later, transfection of siRNA was performed using Lip-ofectamine 2000 (Invitrogen). For all siRNA transfection experiments,standard protocols were used according to the manufacturer’s instructions.

Synthetic siRNAs targeting CD46, CD55, or CD59 were purchased fromInvitrogen. Target sequences were as follows: CD46 siRNA #1 (IDHSS142895) sense 59-CAUGUCCAUAUAUACGGGAUCCUUU-39 andantisense 59-AAAGGAUCCCGUAUAUAUGGACAUG-39; CD46 siRNA#2 (ID HSS181049) sense 59-GGUGAACGAGUAGAUUAUAAGUGUA-39 and antisense 59-UACACUUAUAAUCUACUCGUUCACC-39; CD46siRNA #3 (ID HSS181050) sense 59-CAAAUGGGACUUAGGAGUUU-GGUUA-39 and antisense 59-UAACCAAACUCGUAAGUCCCAUUUG-39; CD55 siRNA #1 (ID HSS102621) sense 59-ACAGUCUGUAACGU-AUGCAUGUAAU-39 and antisense 59AUUACAUGCAUACGUUACA-GACUGU-39; CD55 siRNA #2 (ID HSS175910) sense 59-CACAGUAAAU-GUUCCAACUACAGAA-39 and antisense 59UUCUGUAGUUGGAAC-AUUUACUGUG-39; CD55 siRNA #3 (ID HSS175911) sense 59-GGGUA-CAAAUUAUUUGGCUCGACUU-39 and antisense 59AAGUCGAGCCAA-

AUAAUUUGAACCA-39; CD59 siRNA #1 (ID HSS101611) sense 59-CCAAAGCUGGGUAUCAAGUGUAUAA-39 and antisense 59UUAUA-CACUUGUAACCCAGCUUUGG-39; CD59 siRNA #2 (ID HSS10613)sense 59-AGUUCUUCUGCUGGUGACUCCAUUU-39 and antisense59AAAUGGAGUCACCAGCAGAAGAACU-39. As control, siRNA theLow GC Duplex #1 (Invitrogen) was used. Efficacy of siRNA-inducedknockdown of CD46, CD55, or CD59 was determined by direct immuno-fluorescence using Pacific blue–conjugated mouse anti-human CD46 mAb(Exbio, Vestec, Czech Republic), PE-conjugated mouse anti-human CD55mAb (Beckman Coulter), or FITC-conjugated mouse anti-human CD59mAb (Exbio). Irrelevant Pacific blue/PE/FITC-conjugated Abs served ascontrols.

Overexpression of CD55

BHK-EGFR+ #5 cells were seeded at a density of 1 3 106/well in a 10-cmplate. Next day, cells were transfected with pCMV6-Entry vector (catalogno. PS100001; OriGene Technologies, Rockville, MD) or pCMV6-Entry-CD55 vector (catalog no. RC201615; OriGene Technologies) by lip-ofection using Lipofectamine 2000 (Invitrogen) according to the manu-facturer’s instructions. Expression of CD55 was analyzed by flowcytometry using PE-conjugated mouse anti-human CD55 mAb (BeckmanCoulter).

Data processing and statistical analyses

Graphical and statistical data analyses were carried out using GraphPadPrism 5.0. Curves were adjusted by using a nonlinear regression model witha sigmoidal dose response. Statistical significance was determined bythe two-way ANOVA repeated-measures test with Bonferroni’s posttest.Results are presented as mean 6 SEM of at least three independentexperiments. Correlations between CD46, CD55, or CD59 molecules/celland CDC activity were determined by the Pearson correlation test. The pvalues were calculated and null hypotheses were rejected when p # 0.05.

ResultsAnti–EGFR-IgG1 and anti–EGFR-IgG3 do not differ inFab-mediated effector mechanisms

Human IgG3 in comparison with human IgG1 displays a longerhinge region, 62 versus 14 aa, respectively, and therefore a higher

FIGURE 1. Anti–EGFR-IgG1 and anti–EGFR-IgG3 display similar Ag binding and Fab-mediated effector mechanisms. (A) Structural model of IgG1

and IgG3 molecules. Arrows highlight differences between the hinge regions of IgG1 and IgG3. (B) Monomeric anti–EGFR-IgG1 and anti–EGFR-IgG3

were separated under reducing or nonreducing conditions by SDS-PAGE and were either stained using Coomassie blue (left panel) or immunoblotted (right

panel) against k L chain or human IgG. (C) Size-exclusion chromatography of anti–EGFR-IgG1 and anti–EGFR-IgG3 were performed to separate ag-

gregated from monomeric molecules. (D) Binding of anti–EGFR-IgG1 and anti–EGFR-IgG3 to EGFR expressing DiFi cells were analyzed by indirect

immunofluorescence. Means 6 SEM of three independent experiments are presented. (E) Inhibition of DiFi cell growth was analyzed by MTS assay.

Means 6 SEM of three independent experiments are presented. *p # 0.05, specific mAb versus respective control Ab.

The Journal of Immunology 1487

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

molecular mass, 170 versus 150 kDa, respectively, as demonstratedby biochemical analyses (Fig. 1A, 1B). Purity of monomeric anti–EGFR-IgG1 and anti–EGFR-IgG3, as well as of control-IgG1 andcontrol-IgG3 molecules, was investigated by size exclusion chro-matography. No contamination with aggregated immune complexes,known to encompass stronger complement activation capacities,was observed (Fig. 1C).Monomeric anti–EGFR-IgG1 and anti–EGFR-IgG3 were com-

pared regarding their ability to bind to EGFR on DiFi colorectalcarcinoma cells. No differences in EGFR binding were detectedbetween anti–EGFR-IgG1 (EC50: 1.73 nM) and anti–EGFR-IgG3 (EC50: 1.26 nM; Fig 1D). Similar results were obtainedfrom experiments using A431 cells as a second EGFR+ tumorcell line (data not shown). Furthermore, anti–EGFR-IgG1 (EC50:0.41 6 0.18 nM) and anti–EGFR-IgG3 (EC50: 0.58 6 0.15 nM)similarly induced growth inhibition of DiFi cells (anti–EGFR-IgG1:66.9 6 6.3%; anti–EGFR-IgG3: 68.7 6 6.4%) in a concentration-

dependent manner when compared with the respective control Abs(Fig. 1E).

CDC is induced by anti–EGFR-IgG3, but not byanti–EGFR-IgG1

To analyze the effect of different EGFR expression levels on anti–EGFR-IgG1– or anti–EGFR-IgG3–mediated CDC, we used BHK-EGFR+ #1–#5 cells, displaying varying expression levels of hu-man EGFR (#1 , #2 , #3 , #4 , #5), in CDC assays (9). Toexclude different EGFR binding capacities of anti–EGFR-IgG1and anti–EGFR-IgG3 on these transfectants, we generated dose–response curves for both Abs, as well as for the respective controlAbs by flow cytometry analyses (Fig. 2A). No significant differ-ences were observed between anti–EGFR-IgG1 and anti–EGFR-IgG3, except for significantly different binding of anti–EGFR-IgG1 (MFI = 13.4 6 0.2) and anti–EGFR-IgG3 (MFI = 7.55 62.75) on BHK-EGFR+ #3 cells at 0.53 nM (Fig. 2A).

FIGURE 2. Anti–EGFR-IgG3 in contrast with anti–EGFR-IgG1 triggers CDC. (A) Binding of anti–EGFR-IgG1 or anti–EGFR-IgG3, as well as of

respective control Abs, to BHK cells transfected to express increasing levels of EGFR (BHK-EGFR+ cell lines #1–#5) was analyzed by indirect flow

cytometry at indicated Ab concentrations. (B) In [51Cr] release assays, BHK-EGFR+ cell lines #1–#5 were incubated for 3 h in the presence of 25% v/v

normal human serum and anti–EGFR-IgG1, anti–EGFR-IgG3, or control Abs at increasing concentrations. Means 6 SEM of three independent experi-

ments with different blood donors are presented. (C) CDC against three different tumor cell lines was analyzed by 3-h [51Cr] release assay in the presence of

25% v/v human serum and anti–EGFR-IgG1, anti–EGFR-IgG3, or the respective control Abs at increasing concentrations. Means 6 SEM of at least three

independent experiments are presented. *p # 0.05 anti–EGFR-IgG1/anti–EGFR-IgG3 versus respective control Abs, #p # 0.05 anti–EGFR-IgG3 versus

anti–EGFR-IgG1. (D) Surface expression levels of EGFR were quantified by calibrated flow cytometry. Means 6 SEM of at least three independent

experiments are presented. (E) Correlation between EGFR expression levels and anti–EGFR-IgG3–mediated CDC was calculated for the four cell lines.

1488 CD55 CONTROLS IgG3-DRIVEN COMPLEMENT ACTIVATION

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

FIGURE 3. Anti–EGFR-IgG3 fails to initiate the formation of C5b-9 on the cell surface of A431 cells. (A) Schematic models of the classical pathway

and the AP of complement activation triggered by IgG molecules. (B–F) Deposition of complement components was analyzed on cell lines that were either

resistant (A431) or susceptible (BHK-EGFR+ #5 and DiFi) to anti–EGFR-IgG3–triggered CML. Cells were incubated in the (Figure legend continues)

The Journal of Immunology 1489

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Notably, when compared with the respective control Abs, anti–EGFR-IgG3, but not anti–EGFR-IgG1, triggered significant CDCin a concentration-dependent manner (0.11–66.67 nM) againstBHK-EGFR+ cell lines #3, #4, and #5, whereas no CDC wasdetected against BHK-EGFR+ cell lines #1 and #2 (Fig. 2B).To investigate whether results received from BHK-EGFR+ #1–

#5 cell lines could be transferred to human tumor cell lines, weanalyzed CDC activity evoked by anti–EGFR-IgG1 or anti–EGFR-IgG3 against different tumor cell lines by [51Cr] release assays.Anti–EGFR-IgG3 triggered strongest CDC activity against DiFicells (35.2 6 3.1% maximum lysis), followed by A1207 cells(11.8 6 1.8% maximum lysis). However, no complement-mediatedlysis (CML) of A431 cells was induced by anti–EGFR-IgG3. Inaddition, anti–EGFR-IgG1, as well as the respective control Abs,did not mediate CDC activity against any of the analyzed tumorcell lines (Fig. 2C). CDC triggered by IgG1 Abs directed againstEGFR has been recently demonstrated to depend on cell-surfaceexpression levels of the Ag (9). To analyze whether the extent ofCDC activity induced by anti–EGFR-IgG3 is controlled by EGFRexpression levels, we quantified EGFR cell-surface expressionlevels by calibrated flow cytometry and correlated them to theextent of anti–EGFR-IgG3–triggered CML of the respective cellline. However, no correlation was observed because of similarEGFR-Ab binding sites per cell between all four analyzed celllines (A431 = 1.2 3 106 6 0.1 3 106; BHK-EGFR+ #5 = 1.1 3106 6 0.1 3 106; DiFi = 1.1 3 106 6 0.1 3 106; A1207 = 1.0 3106 6 0.2 3 105; Fig. 2D, 2E).

Anti–EGFR-IgG3 fails to efficiently initiate generation offunctional C3 convertases on A431 cells

Because anti–EGFR-IgG3 triggered CDC against DiFi and BHK-EGFR+ #5 cells, but not against A431 cells, the ability of anti–EGFR-IgG3 to deposit complement components on A431, DiFi,or BHK-EGFR+ #5 cells was analyzed by flow cytometry. In theseexperiments, anti–EGFR-IgG3 was compared with anti–EGFR-IgG1, as well as with a combination of two noncompetingEGFR-directed IgG1 Abs, serving as a positive control (Fig. 3).Interestingly, despite stronger C1q binding by anti–EGFR-IgG3compared with the positive control on A431, DiFi, and BHK-EGFR+ #5 cells (Fig. 3B), lower amounts of C4(b/c) deposition onA431, DiFi, or BHK-EGFR+ #5 cells were observed for anti–EGFR-IgG3 in comparison with the positive control (Fig. 3C).However, in contrast with C4(b/c) deposition, anti–EGFR-IgG3displayed similar C3b deposition as detected for the positivecontrol on A431, DiFi, and BHK-EGFR+ #5 cells (Fig. 3D).Notably, although significantly higher factor Bb deposition wasobtained for anti–EGFR-IgG3 compared with the positive controlon DiFi and A431 cells, significantly lower factor Bb depositionwas found on BHK-EGFR+ #5 cells, pointing to stronger activa-tion of the AP of complement on DiFi and A431 cells than onBHK-EGFR+ #5 cells (Fig. 3E). Furthermore, anti–EGFR-IgG3induced significant C5b-9 levels on DiFi and BHK-EGFR+ #5cells, but not on A431 cells, compared with the respective controlIgG3 Ab (Fig. 3F). In contrast with anti–EGFR-IgG3, anti–EGFR-IgG1 or used control Abs did not trigger significant deposition ofC1q, C4(b/c), C3b, and C5b-9 on all three analyzed cell lines(Fig. 3B–F).

To investigate functional consequences of complement depo-sition on analyzed cell lines, we analyzed the release of C4a, C3a,and C5a anaphylatoxins by calibrated flow cytometry (Table I).Significant release of C4a, C3a, and C5a was triggered by anti–EGFR-IgG3 on DiFi and BHK-EGFR+ #5 cells. Notably, on A431cells, the positive control induced significant C4a, C3a, and C5arelease, whereas anti–EGFR-IgG3 evoked only significant C4a,but not C3a, as well as C5a release. In addition, significant C3arelease was measured in the case of anti–EGFR-IgG1 on A431and DiFi, but not on BHK-EGFR+ #5 cells. However, no signifi-cant augmentation of C4a and C5a levels was detected on all threecell lines by anti–EGFR-IgG1.

The extent of CML evoked by anti–EGFR-IgG3 negativelycorrelates with CD55 and CD59 expression levels

Besides the impact of Ag expression levels on the extent of CDC,cell-surface expression levels of mCRPs such as CD46, CD55, orCD59 have been demonstrated to control CDC activity triggered byHer2/neu-directed Abs (5). To investigate whether the extent ofCDC activity induced by anti–EGFR-IgG3 is controlled by mCRPexpression levels, we performed calibrated flow cytometry toquantify CD46, CD55, or CD59 cell-surface expression levels onA431, DiFi, A1207, and BHK-EGFR+ #5 cells (Fig. 4A). As ex-pected, no expression of human CD46, CD55, or CD59 wasdetected on BHK-EGFR+ #5 cells, whereas significant differenceswere observed for A431, A1207, and DiFi cells. Highest CD46expression levels were identified on DiFi cells (0.4 3 106 6 0.02 3106 specific Ag-binding sites per cell [SABC]), followed by A431and A1207 cells. However, highest expression levels of humanCD55 (0.9 3 105 6 0.1 3 105 SABC) and human CD59 (0.5 3106 6 0.1 3 106 SABC) were measured on A431 cells, followedby A1207 and DiFi cells. Furthermore, correlation coefficients (R)between mCRPs (CD46, CD55, and CD59) cell-surface expressionlevels and relative anti–EGFR-IgG3–induced cytotoxic activitywere calculated for the analyzed cell lines (Fig. 4B). Correlationwas calculated at saturating anti–EGFR-IgG3 Ab concentration(2 mg/ml) by equating cytotoxic activity against BHK-EGFR+

#5 cells with 100% (maximum lysis), which was related to thecytotoxic activity measured for all other cell lines. A negativecorrelation between CD55 and/or CD59 cell-surface expressionlevels and cytotoxic activity triggered by anti–EGFR-IgG3 wasobserved (R = 20.9, p 5 0.06 for CD55, R = 20.9, p 5 0.06 forCD59), whereas no correlation was found for CD46 cell-surfaceexpression levels.Because of results received from correlation analyses, the

impact of CD46, CD55, or CD59 on CML of A431 cells wasanalyzed in more detail by siRNA-induced knockdown experi-ments. Knockdown of CD46, CD55, or CD59 was tested using threedistinct siRNAs specific either for CD46, CD55, or CD59, whereasa nonspecific siRNA served as a control siRNA (data not shown).Based on these data, siRNAs displaying strongest knockdowneffects were chosen for further experiments. Significant siRNA-induced knockdown was achieved for CD46, CD55, or CD59:CD46-specific siRNA (∼73% knockdown), CD55-specific siRNA(∼67% knockdown), or CD59-specific siRNA (∼67% knockdown).A combination of CD46-, CD55-, and CD59-specific siRNAs alsoinduced a significant knockdown of CD46, CD55, and CD59,

presence of 25% v/v human serum and indicated EGFR-Abs or the respective control Abs (66.67 nM for individual mAbs and 33.34 nM each for the positive

control). Deposition of (B) C1q, (C) C4(b/c), (D) (i)C3(b), (E) factor Bb, and (F) C5b-9 were analyzed by direct or indirect immunofluorescence. Relative

deposition levels were calculated by equating RFI measured in the presence of the positive control with 100% for each cell line. Means 6 SEM of at least

three independent experiments are presented. *p # 0.05 anti–EGFR-IgG1/anti–EGFR-IgG3/positive control versus respective control Abs; #p # 0.05

indicated mAb versus anti–EGFR-IgG3.

1490 CD55 CONTROLS IgG3-DRIVEN COMPLEMENT ACTIVATION

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

which was comparable with the extent of knockdown inducedby single siRNAs (CD46: ∼77% knockdown; CD55: ∼55%knockdown; CD59: ∼58% knockdown; Fig. 4C).Subsequently, [51Cr] release assays with increasing concen-

trations of anti–EGFR-IgG1, anti–EGFR-IgG3, or the respectivecontrol Abs in the presence of human serum were performed withcontrol siRNA or with mCRP-specific siRNA-transfected A431cells (Fig. 4D–G). In the case of anti–EGFR-IgG3, siRNA-induced knockdown of CD46 (Fig. 4D, lower panel) did not af-fect CDC activity, whereas siRNA-induced knockdown of CD55(10.4 6 4.4% lysis; Fig. 4E, lower panel) or CD59 (14.0 6 4.3%lysis; Fig. 4F, lower panel) significantly improved CML of A431cells. This improvement was further enhanced by a combinedsiRNA-induced knockdown of CD46, CD55, and CD59 (36.5 62.1% lysis; Fig. 4G, lower panel). Anti–EGFR-IgG1 and respec-tive control Abs did not trigger CDC neither against controlsiRNA-transfected nor against mCRP-specific siRNA-transfectedA431 cells. Results received from RNA interference-inducedknockdown experiments with CD55-specific siRNA were addi-tionally verified by CDC experiments against A431 cells in thepresence of a CD55 blocking Ab at a saturating concentration of66.67 nM. In these blocking experiments, anti–EGFR-IgG3(9.6 6 2.6% lysis), but not anti–EGFR-IgG1, triggered CDCagainst A431 in the presence of the CD55 Ab (Fig. 4H, 4I). Basedon its earlier inhibitory function in the complement cascade incomparison with CD59, these data point to a crucial role of CD55in anti–EGFR-IgG3–evoked complement-dependent cytotoxicactivity.

CD55 constitutes the main regulator ofanti–EGFR-IgG3–driven CML

To further analyze the influence of CD55 on anti–EGFR-IgG3–dependent CML, we transiently transfected human CD552 BHK-EGFR+ #5 cells with a plasmid encoding human CD55, leading tostrong overexpression of CD55 (control vector RFI = 1.2 6 0.2versus CD55-vector RFI = 380.8 6 39.7) on the cell surface(Fig. 5A). Accordingly, dose-dependent CDC activity againstcontrol vector or CD55 vector–transfected BHK-EGFR+ #5 cellstriggered by anti–EGFR-IgG3 was analyzed by [51Cr] releaseassays in the presence of human serum (Fig. 5B). Compared withcontrol vector–transfected BHK-EGFR+ #5 cells, overexpressionof CD55 led to a significant decrease (∼60%) in anti–EGFR-IgG3–induced complement-dependent cytotoxic activity. No CMLof analyzed transfectants was observed in the presence of acontrol IgG3 Ab (Fig. 5B).Activation of the classical pathway and the AP of complement

can be distinguished by analyzing the kinetics of CML, with the APbeing slower than the classical pathway (22, 23). Hence CDCassays using control vector or CD55 vector–transfected BHK-EGFR+ #5 cells were performed at different time points (0–240min) at saturating Ab concentrations (anti–EGFR-IgG3 andcontrol-IgG3, both at 13.3 nM). Notably, significantly slower anti–EGFR-IgG3–promoted CDC kinetics were observed on CD55transfectants (EC50 = 85.00 6 1.95 min) compared with controlvector transfectants (EC50 = 32.41 6 2.71 min; Fig. 5C), pointingto the predominant activation of the alternative complementpathway on CD55 transfectants. The contribution of the alterna-tive complement pathway in anti–EGFR-IgG3–induced CDCagainst CD55- or control vector–transfected BHK-EGFR+ #5 cellswas additionally examined by CDC assays using different serumconcentrations (0–25% v/v). As presented in Fig. 5D, comparedwith control vector–transfected cells, overexpression of CD55 onBHK-EGFR+ #5 cells was accompanied by a statistically signifi-cant decline of anti–EGFR-IgG3–driven CML efficiency at lowerT

able

I.EGFRAb–triggered

anaphylatoxin

release

w/o

EGFR-IgG1

EGFR-IgG3

Positive

Control

Anaphylatoxin

Release

(ng/m

l)C4a

C3a

C5a

C4a

C3a

C5a

C4a

C3a

C5a

C4a

C3a

C5a

DiFi

2556

61.4

5196

17.2

18.6

63.7

2916

43.2

5846

21.6a

256

3.4

4316

51.2a

6026

17.2a

40.2

69.4a

4476

23a

5926

4.9a

466

5.8a

BHK-EGFR+#5

3226

55

4746

49.8

8.1

61.6

3236

50.8

4716

50

8.4

61.7

4246

50a

5556

51.2a

15.1

63.2a

3586

57a

5096

44

14.1

62.1a

A431

3266

55.8

3846

53.6

9.7

62.1

3796

45.4

4526

67a

116

2.2

3936

39.8a

4106

43.6

8.8

61.9

4466

49.4a

4736

43.2a

15.8

63.1a

ap#

0.05anti–EGFR-IgG1/anti–EGFR-IgG3/positive

controlversusw/o

Ab.

The Journal of Immunology 1491

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

FIGURE 4. CML triggered by anti–EGFR-IgG3 negatively correlates with CD55 and CD59 expression levels. (A) Surface expression levels of CD46,

CD55, and CD59 on analyzed cell lines were quantified by calibrated flow cytometry. Means 6 SEM of at least three independent experiments are

presented. (B) Correlations between CD46, CD55, or CD59 and anti–EGFR-IgG3–mediated CDC were calculated for all four cell lines. CDC results at 2

mg/ml Ab concentration were taken from experiments presented in Fig. 2. (C) A431 cells were seeded into 10-cm plates and grown overnight. On the

following day, cells were transfected with 50 nM control siRNA or with single siRNAs specific for CD46, CD55, CD59, or with a combination of all three

mCRP-specific siRNAs for 72 h. Efficiency of siRNA-induced knockdown was analyzed by direct flow cytometry using fluorochrome-labeled, mCRP-

specific Abs (CD46-Pacific blue, CD55-PE, CD59-FITC), or respective control Abs. (D–G) CDC against control siRNA or mCRP-specific, siRNA-

transfected A431 cells was analyzed by 3-h [51Cr] release assays in the presence of 25% v/v human serum and anti–EGFR-IgG1 (upper panels), anti–

EGFR-IgG3 (lower panels), as well as the respective control Abs at increasing concentrations. (H) Concentration-dependent binding of CD55-Ab

(BRIC216, mouse IgG1) to A431 cells was analyzed by indirect immunofluorescence. Results from one representative experiment are presented. (I) CDC

triggered by anti–EGFR-IgG1, anti–EGFR-IgG3, or respective control Abs (all at 66.67 nM) against A431 cells in the presence of saturating concentrations

of CD55-Ab or a control Ab (both at 66.67 nM) was analyzed by 3-h [51Cr] release assays in the presence of 25% v/v human (Figure legend continues)

1492 CD55 CONTROLS IgG3-DRIVEN COMPLEMENT ACTIVATION

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

serum concentrations. To more directly analyze activation of theAP by anti–EGFR-IgG3 on CD55 overexpressing BHK-EGFR+ #5cells, we performed CDC assays in the presence of the alternativecomplement pathway inhibitor CRIg-Fc (Fig. 5E) (24). In linewith CDC experiments presented in Fig. 5C and 5D, no inhibitionof anti–EGFR-IgG3–induced CML of control vector–transfectedBHK-EGFR+ #5 cells was observed in the presence of CRIg-Fc.However, CRIg-Fc strongly inhibited (∼50% inhibition) anti–EGFR-IgG3–triggered CML of CD55-transfected BHK-EGFR+ #5cells. For both transfectants, control vector– or CD55-transfectedBHK-EGFR+ #5 cells, C1q-dependent initiation of CDC wasobserved using C1q-depleted serum in cytotoxic assays (Fig. 5F).

In addition, CDC experiments using DiFi cells that endogenouslyexpress CD55 were performed in the absence and presence ofCRIg-Fc (Fig. 5G). As presented in Fig. 5G, CRIg-Fc also sig-nificantly inhibited anti–EGFR-IgG3–induced CML of DiFi cells(∼80% inhibition) compared with untreated control experiments.Similar results were received from CDC experiments using factorB–depleted serum (Fig. 5H). Although no CDC was inducedagainst DiFi cells by anti–EGFR-IgG3 in the absence of factor B,significant cytotoxicity was triggered when factor B–depletedserum was reconstituted with factor B. Results received from CDCexperiments were further supported by flow-cytometry analysesdemonstrating anti–EGFR-IgG3 to promote significantly more

serum. Results are presented as mean6 SEM of at least three independent experiments with different blood donors. (D–I) *p# 0.05 anti–EGFR-IgG1/anti–

EGFR-IgG3 versus respective control Abs; (D–G) #p # 0.05 specific siRNA versus control siRNA.

FIGURE 5. CD55 dampens anti–EGFR-IgG3–triggered CML and promotes C1q-dependent induction of AP amplification. BHK-EGFR+ #5 cells were

transiently transfected with a control vector or a CD55 vector for 48 h. (A) Cell-surface expression of CD55 was analyzed by direct flow cytometry using

PE-conjugated CD55-specific or control Abs. (B–D) The influence of CD55 overexpression on anti–EGFR-IgG3–mediated CDC was investigated by [51Cr]

release assays either (B) in an Ab concentration–response curve, (C) in a time-dependent manner, or (D) in serum titration experiments. (E–H) The influence

of the alternative complement pathway inhibitor CRIg (E and G), the presence of C1q in serum (F; at 66.67 nM Ab concentration; mean 6 SEM of

triplicates), as well as of factor B (H; 13.33 nM Ab concentration, 12.5% v/v factor B–depleted serum, 200 mg/ml factor B), on anti–EGFR-IgG3–mediated

CDC was analyzed using either (E and F) control vector–transfected or CD55 vector–transfected BHK-EGFR+ #5 cells or (G and H) DiFi cells (66.67 nM

Ab concentration). (I) Deposition of factor Bb on control vector– or CD55 vector–transfected BHK-EGFR+ #5 cells was analyzed by flow cytometry.

Relative deposition levels were calculated by equating RFI measured in the absence of Ab with 100%. Results are presented as mean 6 SEM of at least

three independent experiments with different blood donors. *p # 0.05 anti–EGFR-IgG3 versus respective control Ab; (B–D, I) #p # 0.05 control vector

versus CD55 vector; (E and G) #p # 0.05 without CRIg-Fc versus CRIg-Fc; (H) #p # 0.05 w/o factor B versus with factor B.

The Journal of Immunology 1493

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

factor Bb deposition on CD55-transfected than on control vector–transfected BHK-EGFR+ #5 cells (Fig. 5I, Fig. 6).

DiscussionCDC constitutes a powerful effector mechanism of the immunesystem and is suggested to contribute to the clinical efficacy ofsome tumor-directed therapeutic Abs (3). Unexpectedly, humanIgG1 Abs targeting EGFR lack the capacity to activate the com-plement cascade on solid tumor cells as single agents, whereasbeing potent in activating CDC as noncompeting Ab combinations(10, 25). Because efficient C1q fixation to the CH2 domain of Igmolecules depends on the close proximity of at least two Ig-Fcparts (26, 27), it can be assumed that a combination of non-competing EGFR Abs, in contrast with single EGFR Abs, morepotently enables clustering of closely spaced Ig-Fc parts.To enhance the CDC capacity of single EGFR-targeting Abs, we

generated an IgG3 isotype variant of cetuximab to take advantageof human IgG3-related structural features. Notably, despite strongC1q binding capacity, single EGFR-specific IgG3 Abs triggeredsignificant CDC against CD552 or low-expressing target cells, butnot against CD55 high-expressing cells. This discrepancy betweenstrong human IgG3 triggered C1q deposition on target cells, butrelatively low lytic activity has been also previously reported (12)and further related to inefficient deposition of C4b and C3b ontarget cells (28). Furthermore, lower C4b deposition, although notas much as demonstrated by Bindon et al. (28), but stronger CDCactivity by IgG3 compared with IgG1 has been reported in a recentstudy analyzing CD20-directed Abs on lymphoma cells (29).Notably, in two additional studies, using enzyme immunoassays,either similar (30) or even stronger (31) C4 binding was observedfor IgG3 compared with IgG1. Because C4b has been reported tocovalently bind to Ag-bound Abs in the Fd region, comprising theVH and the CH1 domains (32, 33), it can be hypothesized that

human IgG1 and IgG3 molecules significantly differ in C4b bind-ing. In this study, human IgG3 against EGFR was also found todisplay low C4b in relation to high C1q deposition levels and,therefore, to mainly recruit the AP after initial C1q-dependentcomplement activation on CD55 high-expressing cells (Fig. 6).Hence although human IgG3 has been described to be an activatorof the classical pathway of complement activation (34), recent dataalso demonstrated that human IgG3 directed against factor Hbinding protein expressed by meningococcal strains mainly acti-vates the AP of complement at low Ag expression levels (35).Furthermore, this study shows that CD55 controls anti–EGFR-IgG3–driven complement activation on human cells by interferingwith the assembly of classical C3 convertase and forces anti–EGFR-IgG3 to more efficiently activate the AP. In line with these findings,a previous study reported superior human IgG3-induced CML oftumor cells expressing high CD55 and CD59 but low CD46 levelsby neutralization of CD46, CD55, and CD59 compared with soleneutralization of CD59 on these cells (36). Moreover, recent datarevealed human IgG3 specific for CD20 to more potently triggerCML of CD55-expressing CLL cells than the human IgG1 coun-terpart, potentially pointing to different mechanisms of action ofhuman IgG3 Abs displaying distinct target Ag specificities (18).CD55, the decay-accelerating factor of C3 (C4b2b/C4b2a or

C3bBb) and C5 (C4b2b3b/C4b2a3b or C3bBb3b) convertases, isa GPI-anchored protein that is widely expressed in human cells.However, dysregulated CD55 expression was identified by manystudies in distinct tumor types, with some revealing an upregula-tion and some a downregulation (37). The importance of CD55 intumorigenesis has been underlined by studies revealing high CD55expression levels as a poor prognosis marker in carcinomas of thebreast, the gallbladder, or the colon (38–40). Furthermore, pan-creatic and breast carcinomas were identified to display strongerCD55 cell-surface expression levels compared with oral squamous

FIGURE 6. Overview of complement activation by human anti–EGFR-IgG3 in the context of CD55 expression. On CD55-deficient target cells (left

panel), anti–EGFR-IgG3 mediates strong C3b but low C4b deposition and induces assembly of classical and alternative C3 convertases, predominantly

resulting in the induction of fast and efficient CDC via the classical pathway of complement activation. In contrast, on CD55-expressing target cells (right

panel), CD55 mainly accelerates the decay of low amounts of classical C3 convertases, leading to amplification of the AP and finally to slow and inefficient

CDC induction.

1494 CD55 CONTROLS IgG3-DRIVEN COMPLEMENT ACTIVATION

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

cell or colorectal carcinomas, as well as melanomas (41). In ac-cordance with these results, lowest CD55 expression and, there-fore, strong CDC activity by anti–EGFR-IgG3 was detected on thecolorectal carcinoma cell line DiFi in this study.In conclusion, an isotype switch from human IgG1 to human IgG3

is accompanied by improvement of complement-activating capaci-ties of EGFR targeting Abs. However, despite its strong C1q fixingcapacity, human anti–EGFR-IgG3 lacks the ability to efficientlydeposit C4b on targets’ cell surfaces, and thus to initiate the com-plement cascade on CD55 high-expressing tumor cells. These novelfindings revealed human IgG3 directed against EGFR to encompassimproved cytotoxic potential, and hence to display a promisingstrategy in Ab-based therapy of CD55 low-expressing tumors.

AcknowledgmentsWe thank ChristynWildgrube, Yasmin Brodtmann, and Kathinka T€uxen for

excellent technical assistance. CRIg-Fc was kindly provided by Dr. M. van

Lookeren Campagne (Genentech, South San Francisco, CA). We thank

Dr. F. Beurskens (Genmab, Utrecht, the Netherlands) for critically reading

the manuscript.

DisclosuresThe authors have no financial conflicts of interest.

References1. Dunkelberger, J. R., and W. C. Song. 2010. Complement and its role in innate

and adaptive immune responses. Cell Res. 20: 34–50.2. Adams, G. P., and L. M. Weiner. 2005. Monoclonal antibody therapy of cancer.

Nat. Biotechnol. 23: 1147–1157.3. Gelderman, K. A., S. Tomlinson, G. D. Ross, and A. Gorter. 2004. Complement

function in mAb-mediated cancer immunotherapy. Trends Immunol. 25: 158–164.4. Terui, Y., T. Sakurai, Y. Mishima, Y. Mishima, N. Sugimura, C. Sasaoka, K. Kojima,

M. Yokoyama, N. Mizunuma, S. Takahashi, et al. 2006. Blockade of bulkylymphoma-associated CD55 expression by RNA interference overcomesresistance to complement-dependent cytotoxicity with rituximab. CancerSci. 97: 72–79.

5. Mamidi, S., M. Cinci, M. Hasmann, V. Fehring, and M. Kirschfink. 2013. Lip-oplex mediated silencing of membrane regulators (CD46, CD55 and CD59)enhances complement-dependent anti-tumor activity of trastuzumab and pertu-zumab. Mol. Oncol. 7: 580–594.

6. Bindon, C. I., G. Hale, and H. Waldmann. 1990. Complement activation by im-munoglobulin does not depend solely on C1q binding. Eur. J. Immunol. 20: 277–281.

7. Cragg, M. S., S. M. Morgan, H. T. Chan, B. P. Morgan, A. V. Filatov, P. W. Johnson,R. R. French, and M. J. Glennie. 2003. Complement-mediated lysis by anti-CD20mAb correlates with segregation into lipid rafts. Blood 101: 1045–1052.

8. Teeling, J. L., R. R. French, M. S. Cragg, J. van den Brakel, M. Pluyter,H. Huang, C. Chan, P. W. Parren, C. E. Hack, M. Dechant, et al. 2004. Char-acterization of new human CD20 monoclonal antibodies with potent cytolyticactivity against non-Hodgkin lymphomas. Blood 104: 1793–1800.

9. Derer, S., P. Bauer, S. Lohse, A. H. Scheel, S. Berger, C. Kellner, M. Peipp, andT. Valerius. 2012. Impact of epidermal growth factor receptor (EGFR) cellsurface expression levels on effector mechanisms of EGFR antibodies. J.Immunol. 189: 5230–5239.

10. Dechant, M., W. Weisner, S. Berger, M. Peipp, T. Beyer, T. Schneider-Merck,J. J. Lammerts van Bueren, W. K. Bleeker, P. W. Parren, J. G. van de Winkel, andT. Valerius. 2008. Complement-dependent tumor cell lysis triggered by combina-tions of epidermal growth factor receptor antibodies. Cancer Res. 68: 4998–5003.

11. Klausz, K., S. Berger, J. J. Lammerts van Bueren, S. Derer, S. Lohse,M. Dechant, J. G. van de Winkel, M. Peipp, P. W. Parren, and T. Valerius. 2011.Complement-mediated tumor-specific cell lysis by antibody combinations tar-geting epidermal growth factor receptor (EGFR) and its variant III (EGFRvIII).Cancer Sci. 102: 1761–1768.

12. Br€uggemann, M., G. T. Williams, C. I. Bindon, M. R. Clark, M. R. Walker,R. Jefferis, H. Waldmann, and M. S. Neuberger. 1987. Comparison of the ef-fector functions of human immunoglobulins using a matched set of chimericantibodies. J. Exp. Med. 166: 1351–1361.

13. Dangl, J. L., T. G. Wensel, S. L. Morrison, L. Stryer, L. A. Herzenberg, andV. T. Oi. 1988. Segmental flexibility and complement fixation of geneticallyengineered chimeric human, rabbit and mouse antibodies. EMBO J. 7: 1989–1994.

14. Michaelsen, T. E., P. Garred, and A. Aase. 1991. Human IgG subclass pattern ofinducing complement-mediated cytolysis depends on antigen concentration andto a lesser extent on epitope patchiness, antibody affinity and complementconcentration. Eur. J. Immunol. 21: 11–16.

15. Michaelsen, T. E., I. Sandlie, D. B. Bratlie, R. H. Sandin, and O. Ihle. 2009.Structural difference in the complement activation site of human IgG1 and IgG3.Scand. J. Immunol. 70: 553–564.

16. Brekke, O. H., T. E. Michaelsen, and I. Sandlie. 1995. The structural require-ments for complement activation by IgG: does it hinge on the hinge? Immunol.Today 16: 85–90.

17. Roux, K. H., L. Strelets, and T. E. Michaelsen. 1997. Flexibility of human IgGsubclasses. J. Immunol. 159: 3372–3382.

18. Rosner, T., S. Derer, C. Kellner, M. Dechant, S. Lohse, G. Vidarsson, M. Peipp,and T. Valerius. 2013. An IgG3 switch variant of rituximab mediates enhancedcomplement-dependent cytotoxicity against tumour cells with low CD20 ex-pression levels. Br. J. Haematol. 161: 282–286.

19. Beyer, T., S. Lohse, S. Berger, M. Peipp, T. Valerius, and M. Dechant. 2009.Serum-free production and purification of chimeric IgA antibodies. J. Immunol.Methods 346: 26–37.

20. Stapleton, N. M., J. T. Andersen, A. M. Stemerding, S. P. Bjarnarson,R. C. Verheul, J. Gerritsen, Y. Zhao, M. Kleijer, I. Sandlie, M. de Haas, et al.2011. Competition for FcRn-mediated transport gives rise to short half-life ofhuman IgG3 and offers therapeutic potential. Nat. Commun. 2: 599.

21. Schlaeth, M., S. Berger, S. Derer, K. Klausz, S. Lohse, M. Dechant, G. A. Lazar,T. Schneider-Merck, M. Peipp, and T. Valerius. 2010. Fc-engineered EGF-Rantibodies mediate improved antibody-dependent cellular cytotoxicity (ADCC)against KRAS-mutated tumor cells. Cancer Sci. 101: 1080–1088.

22. Oganesyan, L. P., G. M. Mkrtchyan, S. H. Sukiasyan, and A. S. Boyajyan. 2009.Classic and alternative complement cascades in post-traumatic stress disorder.Bull. Exp. Biol. Med. 148: 859–861.

23. Harboe, M., and T. E. Mollnes. 2008. The alternative complement pathwayrevisited. J. Cell. Mol. Med. 12: 1074–1084.

24. Wiesmann, C., K. J. Katschke, J. Yin, K. Y. Helmy, M. Steffek, W. J. Fairbrother,S. A. McCallum, L. Embuscado, L. DeForge, P. E. Hass, and M. van LookerenCampagne. 2006. Structure of C3b in complex with CRIg gives insights intoregulation of complement activation. Nature 444: 217–220.

25. Koefoed, K., L. Steinaa, J. N. Søderberg, I. Kjær, H. J. Jacobsen, P. J. Meijer,J. S. Haurum, A. Jensen, M. Kragh, P. S. Andersen, and M. W. Pedersen. 2011.Rational identification of an optimal antibody mixture for targeting the epider-mal growth factor receptor. MAbs 3: 584–595.

26. Duncan, A. R., and G. Winter. 1988. The binding site for C1q on IgG. Nature332: 738–740.

27. Painter, R. H. 1984. The C1q receptor site on human immunoglobulin G. Can.J. Biochem. Cell Biol. 62: 418–425.

28. Bindon, C. I., G. Hale, M. Br€uggemann, and H. Waldmann. 1988. Humanmonoclonal IgG isotypes differ in complement activating function at the level ofC4 as well as C1q. J. Exp. Med. 168: 127–142.

29. Natsume, A., M. In, H. Takamura, T. Nakagawa, Y. Shimizu, K. Kitajima,M. Wakitani, S. Ohta, M. Satoh, K. Shitara, and R. Niwa. 2008. Engineeredantibodies of IgG1/IgG3 mixed isotype with enhanced cytotoxic activities.Cancer Res. 68: 3863–3872.

30. Michaelsen, T. E., A. Aase, C. Westby, and I. Sandlie. 1990. Enhancement ofcomplement activation and cytolysis of human IgG3 by deletion of hinge exons.Scand. J. Immunol. 32: 517–528.

31. Mollnes, T. E., K. Høgasen, B. F. Hoaas, T. E. Michaelsen, P. Garred, andM. Harboe. 1995. Inhibition of complement-mediated red cell lysis by immu-noglobulins is dependent on the IG isotype and its C1 binding properties. Scand.J. Immunol. 41: 449–456.

32. Goers, J. W., and R. R. Porter. 1978. The assembly of early components ofcomplement on antibody-antigen aggregates and on antibody-coated erythro-cytes. Biochem. J. 175: 675–684.

33. Campbell, R. D., A. W. Dodds, and R. R. Porter. 1980. The binding of humancomplement component C4 to antibody-antigen aggregates. Biochem. J. 189:67–80.

34. Lucisano Valim, Y. M., and P. J. Lachmann. 1991. The effect of antibody isotypeand antigenic epitope density on the complement-fixing activity of immunecomplexes: a systematic study using chimaeric anti-NIP antibodies with humanFc regions. Clin. Exp. Immunol. 84: 1–8.

35. Giuntini, S., D. C. Reason, and D. M. Granoff. 2012. Combined roles of humanIgG subclass, alternative complement pathway activation, and epitope density inthe bactericidal activity of antibodies to meningococcal factor h binding protein.Infect. Immun. 80: 187–194.

36. Olafsen, T., C. K. Munthe Lund, O. S. Bruland, I. Sandlie, and T. E. Michaelsen.1999. Complement-mediated lysis of cultured osteosarcoma cell lines usingchimeric mouse/human TP-1 IgG1 and IgG3 antibodies. Cancer Immunol.Immunother. 48: 411–418.

37. Spendlove, I., J. M. Ramage, R. Bradley, C. Harris, and L. G. Durrant. 2006.Complement decay accelerating factor (DAF)/CD55 in cancer. Cancer Immunol.Immunother. 55: 987–995.

38. Ikeda, J., E. Morii, Y. Liu, Y. Qiu, N. Nakamichi, R. Jokoji, Y. Miyoshi,S. Noguchi, and K. Aozasa. 2008. Prognostic significance of CD55 expression inbreast cancer. Clin. Cancer Res. 14: 4780–4786.

39. Wu, J., L. Lei, S. Wang, D. Gu, and J. Zhang. 2012. Immunohistochemicalexpression and prognostic value of CD97 and its ligand CD55 in primary gall-bladder carcinoma. J. Biomed. Biotechnol. 2012: 587672.

40. Durrant, L. G., M. A. Chapman, D. J. Buckley, I. Spendlove, R. A. Robins, andN. C. Armitage. 2003. Enhanced expression of the complement regulatoryprotein CD55 predicts a poor prognosis in colorectal cancer patients. CancerImmunol. Immunother. 52: 638–642.

41. Ravindranath, N. M., and C. Shuler. 2007. Cell-surface density of complementrestriction factors (CD46, CD55, and CD59): oral squamous cell carcinomaversus other solid tumors. Oral Surg. Oral Med. Oral Pathol. Oral Radiol.Endod. 103: 231–239.

The Journal of Immunology 1495

by guest on April 11, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from