An in vivo model is used to define Fc motifs engaged by mAbs to deplete target cells. Human IgG1 and human IgG4 were very potent, and mutations within a motif critical for FcγR binding (glutamate 233 to proline, leucine/phenylalanine 234 to valine, and leucine 235 to alanine) completely prevented depletion. Mouse IgG2b was also potent, and mutations to prevent complement activation did not impair depletion with this isotype, as previously shown for human IgG1. In contrast, a mutation that impaired binding to mouse FcγRII (glutamate 318 to alanine) eliminated effector function of mouse IgG2b and also reduced the potency of human IgG4. To reveal potential contributions of domains other than CH2, domain switch mutants were created between human IgG1 and rat IgG2a. Two hybrid mAbs were generated with potencies exceeding anything previously seen in this model. While their mechanism of depletion was not defined, their activity appeared dependent upon interdomain interactions in the Fc region.

Monoclonal Abs offer potential for the therapy of malignant and autoimmune diseases. In malignant disease, their use has largely focused on killing of target tumour cells (1); in autoimmunity, there are opportunities to exploit either depleting or nondepleting mAbs (2). Protein engineering offers the possibility of fashioning mAbs to given therapeutic situations (3); however, to fully exploit such technology, a thorough understanding of the structural basis of mAb effector function is required. Historically, selection of the constant region (C region) of therapeutic mAbs has often been based upon in vitro comparisons of matched sets of chimeric mAbs for effector functions such as complement-mediated lysis (CML)5 or Ab-dependent cell-mediated cytotoxicity (ADCC) (4, 5, 6). There has been an increasing awareness, however, that what is measured in vitro may not completely reflect what is relevant in vivo (7). Furthermore, rules generated for one Ag may not translate to another (8), nor to the same Ag at a different density (9), nor allow for polymorphisms within a population (10).

In view of these limitations, it is important to be able to evaluate mAb function in vivo. This is not easily achieved in the clinical setting; what is needed is a preclinical model that might allow us to predict the biologic effects of mAb administration. One could then correlate in vivo and in vitro data so as to devise predictive in vitro tests of in vivo function. A sensitive system of this kind would also enable a fuller description of the structural basis of in vivo activity.

In a previous paper, we described a mouse model designed for this purpose (11). We compared a number of chimeric CD8 mAbs for their ability to deplete CD8+ PBL in CBA/Ca mice. We observed all human IgGs (hIgG) as well as rat IgG1 (rIgG1) and rIgG2b to be potent depleting agents, while rIgG2a was intermediate in potency, and hIgA2 and hIgE were inactive. With mutants of hIgG1, we were able to document the importance of the N-linked carbohydrate at residue N297 for in vivo effector function and also to show that complement was not necessary for in vivo killing. Thus, a mutant of hIgG1 that lacked the C1q binding motif (E318 K320 K322) (12) remained highly lytic both in complement-replete and in complement-depleted animals.

In the current paper, we build on these experiments using mutants of hIgG1 and hIgG4 to demonstrate the importance of FcγR interactions for in vivo killing. We also investigate an entirely homologous system using mouse IgG2b (mIgG2b) and mutants thereof. Finally, using domain switch mutants between hIgG1 and rIgG2a, we generate a chimeric protein of superior potency. These latter data suggest that interdomain interactions may be an important factor in determining mAb effector function.

The methods employed were broadly as previously published (11).

Chimeric mAbs were created by transfection of the plasmid pSV-V169 encoding chimeric CD8 heavy chains into the heavy chain loss variant YTS 169.4L (11). Table I lists mutants of hIgG1, IgG4, and mIgG2b that were generated and includes details of mutated amino acid residues. A mutated residue is denoted AnB, where n signifies the position of the mutated residue, A is the wild-type (WT), and B the mutant amino acid. The origin of WT hIgG1 and hIgG4 C regions and the derivation of mutants by site-directed mutagenesis was described previously (11). The oligonucleotide used to mutate the C1q binding site of hIgG1 and hIgG4 was 5′-TTT GTT GGA GAC CGC GCA CGC GTA CGC CTT GCC ATT CA-3′. The lower hinge FcγR binding site was mutated using the oligonucleotide 5′-GAC TGA CGG TCC CCC CGC GAC TGG AGG TGC TGA GGA-3′. The mIgG2b WT and mutant constructs were described previously (12, 13).

Table I.

Details of mutant mAbs used in these investigations

mAbMutations
hIgG1FcγR E233P L234V L235A    
hIgG1C1qFcγR E233P L234V L235A E318A K320A K322A 
hIgG4C1q    E318A K320A K322A 
hIgG4FcγR E233P F234V L235A    
hIgG4C1qFcγR E233P F234V L235A E318A K320A K322A 
mIgG2bC1q      K322A 
mIgG2bFcγRII    E318A   
mIgG2bFcγRI+   E235L    
mAbMutations
hIgG1FcγR E233P L234V L235A    
hIgG1C1qFcγR E233P L234V L235A E318A K320A K322A 
hIgG4C1q    E318A K320A K322A 
hIgG4FcγR E233P F234V L235A    
hIgG4C1qFcγR E233P F234V L235A E318A K320A K322A 
mIgG2bC1q      K322A 
mIgG2bFcγRII    E318A   
mIgG2bFcγRI+   E235L    

Genes for domain switch mAbs were constructed using a hIgG1 genomic clone mutated to incorporate restriction sites between the exons encoding CH1 and hinge (XbaI), hinge and CH2 (XhoI), and CH2 and CH3 (KpnI) (10). A rIgG2a genomic clone was similarly mutated and cloned into pUC18 with a SalI site upstream and an EcoRI site downstream. Exchange of exons between hIgG1 and rIgG2a and subsequent cloning into pSV-V169 provided the mAbs illustrated in the left panel of Figure 5.

FIGURE 5.

Depletion by domain switch hybrid mAbs. Groups of four mice were administered the mAbs shown. The figure shows depletion as measured on day 21. Results are expressed as mean ± SD and are pooled from three experiments. The left panel is a diagrammatic representation of the hybrid mAbs administered: white domains are human, black are rat. 0.5 μg panel: Group A vs B, p = 0.0007; A vs C, p = 0.0019. 5.0 μg panel: Group L vs G, p = 0.0040; L vs H, p = 0.0107; L vs J, p = 0.0279. (Mann-Whitney U test).

FIGURE 5.

Depletion by domain switch hybrid mAbs. Groups of four mice were administered the mAbs shown. The figure shows depletion as measured on day 21. Results are expressed as mean ± SD and are pooled from three experiments. The left panel is a diagrammatic representation of the hybrid mAbs administered: white domains are human, black are rat. 0.5 μg panel: Group A vs B, p = 0.0007; A vs C, p = 0.0019. 5.0 μg panel: Group L vs G, p = 0.0040; L vs H, p = 0.0107; L vs J, p = 0.0279. (Mann-Whitney U test).

Close modal

Transfectants were cloned once in soft agar, and high secreting clones (detected by isotype-specific ELISA) were grown to stationary phase in roller cultures. Supernatants were concentrated by ammonium sulfate precipitation and dialysed against PBS. hIgG1 and hIgG4 mAbs and mutants thereof were further purified on protein A, but other mAbs were used as crude precipitates.

mAb concentrations in all preparations were calculated by competitive binding assay using the WT rIgG2b mAb YTS 169.4 as previously described (11). SDS-PAGE analysis confirmed all mAbs to be H2L2 dimers (not shown), and therefore binding equivalents translated directly into μg/ml as used throughout this paper (11).

The preparation of PBL from peripheral blood and dual-color immunofluorescence has been described previously (14). mAbs used in the current study were biotin-coupled MTF 171 (aglycosyl hIgG1 anti-mouse CD8) (11) with either YTS 177.9 (rIgG2a anti-mouse CD4) (15) or KT3-1.1 (rIgG2a anti-mouse CD3) (16). Detection reagents were MARG2A-FITC (anti-rat IgG2a, Serotec MCA 278F, Oxford, U.K.) and streptavidin-phycoerythrin. Cells were fixed and stored after staining for subsequent analysis using a FACScan incorporating a FACSmate robotic sampler (Becton Dickinson, Mountain View, CA). The percentage of CD8+ PBL was calculated as the mean of CD4CD8+ and CD3+CD8+ values.

Single-color immunofluorescence (see Table III) employed YTS 177.9, KT3-1.1, and YTS 105.18 (rIgG2a anti-mouse CD8) using MARG2A-FITC for detection as previously described (11). Circulating lymphocytes coated with mAb were detected using MAR 18.5 (anti-rat κ light chain).

Table III.

Kinetics of CD8+ PBL clearance and coatinga

P/BD1D4D7D14
Group A. hIgG1, 5.0 μg      
CD8 15 2.5 
CD4 69 69 81 75 71 
CD3 85 70 83 82 78 
MAR 18.5 ND 1.5 
Group B. hIgG1FcγR, 5.0 μg      
CD8 19 18 15 15 
CD4 67 63 64 59 63 
CD3 84 83 84 80 84 
MAR 18.5 ND 20 18 1.5 
P/BD1D4D7D14
Group A. hIgG1, 5.0 μg      
CD8 15 2.5 
CD4 69 69 81 75 71 
CD3 85 70 83 82 78 
MAR 18.5 ND 1.5 
Group B. hIgG1FcγR, 5.0 μg      
CD8 19 18 15 15 
CD4 67 63 64 59 63 
CD3 84 83 84 80 84 
MAR 18.5 ND 20 18 1.5 
a

Mice were bled at intervals after mAb administration and stained (single color) for CD4, CD8, or CD3, or with an antiglobulin (MAR 18.5), which stained cells coated with “therapeutic” mAb (11 ). CD8+ PBL disappeared within 24 h following administration of depleting mAbs, with minimal evidence of coating (group A). Nondepleting mAbs also resulted in an early loss of CD8+ staining but with a compensatory increase in MAR 18.5 staining, indicating that circulating cells were “coated.” CD8+ staining reappeared and MAR 18.5 staining disappeared as therapeutic mAb was lost from the cell surface, with dual staining on day 4 (D4) indicating partial coating at that time (group B). Numbers denote the percentage of PBL staining at each time point. ND, not determined. P/B, pre-bleed.

CBA/Ca mice were maintained at the animal facility of the Department of Pathology, University of Cambridge. They were thymectomized at 4 to 5 wk of age (17) and used for experiments in age- and sex-matched groups. When required, depletion of complement C3 component was achieved by administration of 75 μg cobra venom factor in four divided doses 48 to 24 h before mAb administration. This dose is at least six times larger than required to deplete mice of C3 for 96 h after the fourth dose of cobra venom factor (18).

Test mAb was administered via a tail vein at time zero. mAb preparations were centrifuged before injection (13,000 rpm for 20 min in a bench top microfuge), and only the top one-third of MAb was used, to avoid administering large aggregates. Mice were bled at intervals following mAb administration for staining of PBL. Each experimental group comprised four CBA/Ca mice, and data were pooled from more than one experiment for analysis. Intergroup statistical comparisons utilized the Mann-Whitney U test.

In previous work, we demonstrated that a hIgG1 mAb was very efficient at killing mouse PBL, but an aglycosylated variant lost this capacity. Additionally, we showed that complement activation was not necessary for depletion with this mAb (11). A critical motif for the interaction of Igs with the high affinity human FcγR (hFcγRI, CD64) has been identified (13, 19), and the same motif has been shown to be important for binding mFcγRI (20). The motif (E233 L234 L235 G236 G237 P238) is shared by hIgG1 and hIgG3 (Table II). hIgG4 differs by one amino acid (F234) and has a 10-fold reduced affinity for hFcγRI, while hIgG2 (P233 V234 A235) does not bind (19). We created mutants of hIgG1 and hIgG4 with the hIgG2 sequence at residues 233 to 235 and assessed them in vivo. This manipulation completely abolished their depleting capacity (Fig. 1), but targeted PBL were coated with CD8 mAb for 4 to 7 days following 5 μg of mAb (Table III) and up to 14 days after 25 μg (not shown). Residues 233 to 235 thus play a critical role in the effector function of these isotypes. A motif implicated in Ig binding to mFcγRII and probably to mFcγRIII has been localized to this same region (21), and while our data strongly implicate FcγR-mediated clearance mechanisms, they do not denote a specific receptor subtype.

Table II.

Ig CH2 domain sequences for hIgG1-4, mIgG2b, and the mAb mutants used in these studiesa

240250260270280290300310320330340
 ****       ** ** 
hIgG1 APELLGGPSV FLFPPKPKDT LMISRTPEVT CVVVDVSHED PEVKFNWYVD GVEVHNAKTK PREEQYNSTY RVVSVLTVLH QDWLNGKEYK CKVSNKALPA PIEKTISKAK 
C1q ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- -------A-A -A-------- ---------- 
FcγR --PVA----- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- 
C1qFcγR --PVA----- ---------- ---------- ---------- ---------- ---------- ---------- ---------- -------A-A -A-------- ---------- 
hIgG2 --PVA_---- ---------- ---------- ---------- ---Q------ ---------- -----F---F --------V- ---------- ------G--- --------T- 
hIgG3 ---------- ---------- ---------- ---------- ---Q-K---- ---------- ---------F ---------- ---------- ---------- --------T- 
hIgG4 ---F------ ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- ---------- ------G--S S--------- 
C1q ---F------ ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- -------A-A -A----G--S S--------- 
FcγR --PVA----- ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- ---------- ------G--S S--------- 
C1qFcγR --PVA----- ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- -------A-A -A----G--S S--------- 
mIgG2b --N-E----- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS---F- ---N--D--S ---R----I- 
FcγRI+ --N------- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS---F- ---N--D--S ---R----I- 
C1q --N-E----- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS--AF- ---N--D--S ---R----I- 
FcγRII --N-E----- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS---F- -A-N--D--S ---R----I- 
240250260270280290300310320330340
 ****       ** ** 
hIgG1 APELLGGPSV FLFPPKPKDT LMISRTPEVT CVVVDVSHED PEVKFNWYVD GVEVHNAKTK PREEQYNSTY RVVSVLTVLH QDWLNGKEYK CKVSNKALPA PIEKTISKAK 
C1q ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- -------A-A -A-------- ---------- 
FcγR --PVA----- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- 
C1qFcγR --PVA----- ---------- ---------- ---------- ---------- ---------- ---------- ---------- -------A-A -A-------- ---------- 
hIgG2 --PVA_---- ---------- ---------- ---------- ---Q------ ---------- -----F---F --------V- ---------- ------G--- --------T- 
hIgG3 ---------- ---------- ---------- ---------- ---Q-K---- ---------- ---------F ---------- ---------- ---------- --------T- 
hIgG4 ---F------ ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- ---------- ------G--S S--------- 
C1q ---F------ ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- -------A-A -A----G--S S--------- 
FcγR --PVA----- ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- ---------- ------G--S S--------- 
C1qFcγR --PVA----- ---------- ---------- -------Q-- ---Q------ ---------- -----F---- ---------- -------A-A -A----G--S S--------- 
mIgG2b --N-E----- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS---F- ---N--D--S ---R----I- 
FcγRI+ --N------- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS---F- ---N--D--S ---R----I- 
C1q --N-E----- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS--AF- ---N--D--S ---R----I- 
FcγRII --N-E----- -I----I--V ----L--K-- -------ED- -D-QIS-F-N N----T-Q-Q THR-D----I ----T-PIQ- ---MS---F- -A-N--D--S ---R----I- 
a

Residues previously shown to affect either complement activation or FcγR binding are marked by an asterisk (see text). Identity to hIgG1 is denoted by −. The _ corresponds to an “absent” residue in the shortened hIgG2 lower hinge.

FIGURE 1.

hIgG1 and hIgG4 mutated in an FcγR binding motif are impotent in vivo. Groups of four mice were administered the mAbs shown. Depletion of CD8+ PBL was measured on day 14. Results are expressed as the mean ± SD and are pooled from three experiments.

FIGURE 1.

hIgG1 and hIgG4 mutated in an FcγR binding motif are impotent in vivo. Groups of four mice were administered the mAbs shown. Depletion of CD8+ PBL was measured on day 14. Results are expressed as the mean ± SD and are pooled from three experiments.

Close modal

hIgG4 does not bind C1q (22); mutations in the C1q binding motif (E318A K320A K322A) (12) were therefore not predicted to influence depleting potency. Surprisingly, depletion was reduced by ∼50% compared with WT but was not further compromised by pretreatment of mice with cobra venom factor (Fig. 2). A possible explanation is the involvement of residue 318 in interactions with mFcγRII. Thus, mIgG2b with the mutation E318A no longer bound mFcγRII (21), and the same may hold for hIgG4. The equivalent mutation did not influence depletion by hIgG1 (11), suggesting either lack of involvement of FcγRII or redundancy of clearance mechanisms with that isotype.

FIGURE 2.

hIgG4 mutated in the C1q binding motif has impaired depleting capacity. Groups of four mice were administered the mAbs shown. Depletion of CD8+ PBL was measured on day 14. Results are expressed as the mean ± SD and are pooled from three experiments.

FIGURE 2.

hIgG4 mutated in the C1q binding motif has impaired depleting capacity. Groups of four mice were administered the mAbs shown. Depletion of CD8+ PBL was measured on day 14. Results are expressed as the mean ± SD and are pooled from three experiments.

Close modal

The data accumulated thus far suggest that in this model depletion requires interaction with Fcγ receptors, and circumstantial evidence implicates both FcγRI and FcγRII. To clarify these findings we investigated the depleting capacity of mIgG2b and mutants thereof. WT mIgG2b binds mFcγRII but not mFcγRI (20), and it activates complement (12). It depleted CD8+ PBL with a potency equivalent to hIgG1 and hIgG4 (Fig. 3): 70 to 80% depletion was achieved with 5 μg mAb per mouse and ∼50% less when 0.5 μg was administered. As with hIgG1, inactivating the C1q binding site (K322A) did not impair this activity in either complement-sufficient or -depleted mice. A mutation that enabled binding to mFcγRI (E235L) (20) did not enhance potency, but inactivation of the FcγRII binding motif (E318A) (21) significantly reduced depletion. These data clearly demonstrate that depletion can occur via interaction with mFcγRII.

FIGURE 3.

mIgG2b depletes via an FcγRII-dependent mechanism. Groups of four mice were administered the mAbs shown. Depletion of CD8+ PBL was measured on day 14. Results are expressed as mean ± SD and are pooled from two experiments.

FIGURE 3.

mIgG2b depletes via an FcγRII-dependent mechanism. Groups of four mice were administered the mAbs shown. Depletion of CD8+ PBL was measured on day 14. Results are expressed as mean ± SD and are pooled from two experiments.

Close modal

The data presented above suggest that hIgG4 may interact with both mFcγRI and mFcγRII. Thus, the E233P L234V L235A mutant failed to deplete, and the E318A K320A K322A mutant was less potent than WT. Surprisingly, a double mutant mAb encompassing both sets of changes retained weak activity (Fig. 4). This result suggested that the two sets of mutations balanced each other in some way, although an equivalent effect was not seen with hIgG1, a double mutant of which failed to deplete (Fig. 4).

FIGURE 4.

Mutation within the C1q binding motif restores cytotoxicity to hIgG4 mutated in an FcγR binding motif. Groups of four mice were administered the mAbs shown. The figure shows depletion as measured on day 14. Results are experienced as mean ± SD and are pooled from three experiments.

FIGURE 4.

Mutation within the C1q binding motif restores cytotoxicity to hIgG4 mutated in an FcγR binding motif. Groups of four mice were administered the mAbs shown. The figure shows depletion as measured on day 14. Results are experienced as mean ± SD and are pooled from three experiments.

Close modal

The data in the previous section imply that discrete loci within an Ig molecule can interact and contribute to effector function. Domain switch mAbs, in which an entire Fc domain is substituted by its equivalent from an alternative isotype, can provide similar information. In previous work, we showed the rIgG2a isotype to possess unusual depleting properties. Not only did it have an intermediate potency compared with other isotypes, but depletion occurred slowly over 10 days (11). To further dissect these findings, we produced rIgG2a/hIgG1 domain switch mAbs. Figure 5 illustrates these hybrids and the results of their in vivo administration to mice. The 5.0-μg data broadly confirm the in vitro work of other authors and demonstrate that the CH2 domain is the dominant determinant of effector function (10, 23, 24): Any hybrid possessing a hIgG1 CH2 domain achieved 80 to 90% depletion, although there was more variability among hybrids with a rIgG2a CH2 domain. Thus, adding a hIgG1 hinge to rat WT (group G) substantially improved potency, as did the combination of hIgG1 CH1 and CH3 domains (groups H and J). The hinge has been implicated previously in mAb effector function (25), but a contribution from CH1 or CH3 domains was not predicted. Equivalent results were also seen with the administration of 0.5 μg of hybrid mAb per mouse. Most mAbs were unable to deplete CD8+ PBL by >20 to 30% at this dose, but two were consistently more potent, exceeding by two- to threefold the depleting capacity of hIgG1 WT. These mAbs shared rat CH1 and CH3 domains (groups B and C), and regardless of mechanism, clearly demonstrated an influence of domains outside the hinge and CH2 on effector function. These are the most potent mAbs thus far tested in this model, and to our knowledge, this is the first demonstration of bivalent mAbs with effector potencies in vivo superior to the “best” WT isotypes.

These experiments reinforce and extend our previous observations concerning potential mechanisms by which mAbs kill lymphocytes in the mouse. Our model comprises the administration of recombinant CD8 mAbs to thymectomized mice, and subsequently monitoring the fate of CD8+ PBL. We previously showed that hIgG1 had potent depleting properties that did not require complement activation, but that an aglycosylated version of that isotype was unable to clear cells. The current work provides similar information for mIgG2b and unambiguously reveals the importance of FcγR binding for all isotypes tested. Depleting isotypes cleared cells from the circulation with rapid kinetics (Table III), presumably via interactions with reticuloendothelial cells in the liver and spleen. In the current work, PBL populations were monitored for 14 to 21 days and in our previous work for several months. Depleted cells did not reappear during these time periods, strongly suggesting that they had been killed and not merely sequestered.

Care is necessary in interpreting the role of each FcγR class, but minimal conclusions are as follows. 1) mFcγRII (or mFcγRIII; see below) mediates clearance with mIgG2b. This potent isotype does not bind mFcγRI (20), and preventing complement activation (K322A) had no effect on its activity. However, blocking binding to FcγRII (E318A) reduced the level of depletion (Fig. 3). 2) Mutating residues 233 to 235 in both hIgG1 and hIgG4 abolished depletion by these isotypes (Fig. 1), and these residues are critical for mFcγRI (20) and mFcγRII (21) binding. The mutation E318A also impaired the activity of hIgG4 (Fig. 2) but not hIgG1 (11). Thus, it seems likely that both hIgG1 and hIgG4 use mFcγRI to deplete, although hIgG4 may additionally interact with mFcγRII.

Unexpectedly, mutations toward the carboxyl terminus of IgG4 CH2 (residues 318, 320, and 322) partially restored the activity of the impotent lower hinge (233–235) mutant (Fig. 4). Loci that are distant in primary sequence have previously been shown to interact in both complement activation (10, 23) and FcγR binding (24), purportedly by becoming close neighbors in Ig quaternary structure. In the current example, both regions have been implicated in FcγR interactions, and it seems likely that we fortuitously restored a FcγR binding capability, although direct binding studies would be required to confirm this. The spatial proximity of these two sets of residues are illustrated in Figure 6, a raster space-filling model of a hIgG1 mAb (26). There have been no detailed studies of binding motifs on Igs for mFcγRIII, but mFcγRII and III share >95% homology in their extracellular domains and an identical hierarchy for binding to WT mouse and human Igs (27). Therefore, mutations that affect binding to mFcγRII may also affect binding to mFcγRIII. It should also be noted that in a single recent report, the lower hinge was implicated in complement activation as well as FcγR binding (28). Overall, however, our data are not consistent with a complement-mediated mechanism for mAb cytotoxicity in this model.

FIGURE 6.

A raster space-filling model of a human IgG1 Ab, highlighting residues discussed in the text. See Reference 26 for its derivation. The basic model is in the public domain and available at the following website: http://www.path.cam.ac.uk/∼mrc7/functions/ig1func.html.

FIGURE 6.

A raster space-filling model of a human IgG1 Ab, highlighting residues discussed in the text. See Reference 26 for its derivation. The basic model is in the public domain and available at the following website: http://www.path.cam.ac.uk/∼mrc7/functions/ig1func.html.

Close modal

By comparing structure and function, data such as ours can be used to implicate novel residues in Ig effector function. In our previous work, however, all four human IgG subclasses efficiently depleted mouse PBL (11), and our current data appear surprising in that context: the mutations that abolished depletion by hIgG1 and hIgG4 mimic the equivalent residues of hIgG2, another depleting subclass (Table II). Table IV and Figure 6 highlight nonconserved residues within the CH2 domains of hIgG2, hIgG4, hIgG1FcγR and hIgG4FcγR, which might account for these paradoxical data. Although several are located close to previously defined effector motifs, no residues, either singly or in combination, adequately resolve the disparate properties of these four mAbs. Interpretation is slightly complicated by a single amino acid truncation of the hIgG2 hinge, but an alternative explanation is that mAb effector function is modulated by Ig regions outside the CH2 domain, as suggested by our final set of experiments in which we created domain switch mutants between hIgG1 and rIgG2a. These confirmed CH2 as a critical determinant of effector function, but also exposed influences from other domains. For example, substitution of a human hinge on rIgG2a improved the latter’s potency, but so too did exchange of CH1 and CH3 domains for their human counterparts (but not either alone; Figure 5, right panel, hybrids G to L). Similarly, replacement of human CH1 and CH3 with the equivalent rat domains improved potency of hIgG1 (Fig. 5, middle panel, hybrids A–F).

Table IV.

Nonconserved CH2 domain residues between hIgG1 and IgG4 FcγR mutants and WT hIgG2 and IgG4a

233234235268274296300309327330331339
hIgG1FcγR (nondepleting) 
hIgG4FcγR (nondepleting) 
hIgG2 (depleting) 
hIgG4 (depleting) 
233234235268274296300309327330331339
hIgG1FcγR (nondepleting) 
hIgG4FcγR (nondepleting) 
hIgG2 (depleting) 
hIgG4 (depleting) 
a

No residue or combination of residues satisfactorily explains the differing properties of these four mAbs (see text).

Multidomain influences on effector function are not without precedent. Thus, a study of hIgG1/IgG4 domain switch hybrids of CD52 mAbs concluded that CH1 and CH2 influence CML potency (10), and there is also evidence that hinge, CH1, and CH2 domains influence hFcγRI binding (24). In contrast, both CH2 and CH3 domains of hIgG1 were needed to fully restore hFcγRI and hFcγRII binding activity to a hapten-binding mIgE (29). Such effects presumably arise from improved binding to complement components or FcγR, either by provision of a more stable binding motif or perhaps through stabilization of a favorable quaternary structure. Alternatively, synergy between domains may result from overlapping but distinct functions. For example, ADCC via hFcγRIII requires an IgG CH3 domain to bind hFcγRIII and a CH2 motif to trigger cytotoxicity (30). Similarly, our in vivo readout may reflect more than one effector function. For example, a C3b bridge between effector cell and target could facilitate binding to Fcγ receptors and, since CH1/CH2 interactions have been invoked in CML (10) and CH2/CH3 in ADCC (30), this could also explain a multidomain influence on effector function. Finally, multimeric mAbs display improved effector functions via clustering of Ig Fc regions at the target cell surface. This is achieved naturally in IgM and IgA but can also be created artificially. Thus, dimerizing mAb by means of a C-terminal disulfide bridge improved effector function substantially (31, 32), and addition of a μ-tailpiece to IgG4 even permitted CML with this isotype (33). Although we quantitated our mAb preparations by competitive binding to CD8, demonstrated that they were H2L2 tetramers by PAGE, and centrifuged them before injection, we cannot exclude the possibility that interdomain interactions resulted in multimerization of our domain switch chimeras, leading to increased potency. In future work, different mAb preparations should be analyzed by size exclusion to appraise a potential role for multimers in dictating effector function.

Our data provide important, generalizable messages for mAb engineers. Nondepleting mAbs are very powerful tools for modulating immune responses (2, 34) and are fashionable agents for manipulation of human autoimmunity. The best route to creating such mAbs has not been identified, however. It had been assumed that “impotent” isotypes such as hIgG4 could be used, but we previously argued that in vitro activity may not predict in vivo function (11), and furthermore, population FcγR polymorphisms may lead to interindividual variations in biologic activity (10). In contrast, aglycosylated forms of mAb are generally devoid of effector function in vivo (11) and should not provoke cytokine release reactions (35). Furthermore, fears concerning their immunogenicity appear unfounded (36). Similarly, our current work suggests that if a mAb depletes via FcγRs, the mutation of one or two amino acids may completely abolish effector function. The immunogenicity of such agents is likely to be similar to the humanized mAbs from which they are derived (37, 38). At the other extreme, the more cytolytic a mAb, the more effective it should be at targeting and killing tumor cells (1). The domain switch mAbs described above have highlighted interactions between the Ig constant domains and include hybrids of exceptional potency, which should guide additional experiments in this area.

These data again emphasize the importance of an in vivo model for the investigation of mAb effector function. Although largely derived from a heterologous system, they have clearly demonstrated the in vivo effects of specific mutations and revealed complexities of interaction that would be extremely difficult to reproduce in vitro. We would argue that such a model provides a critical bridge between in vitro and clinical studies, although at our present state of knowledge, no model can currently substitute for small in vivo pilot trials (39).

We thank Dr. Greg Winter for supplying plasmid DNA encoding mIgG2b and mutants thereof and Dr. Mike Clark for providing the raster space-filling model of hIgG1 (Fig. 6).

1

This work was supported by a U.K. Medical Research Council (MRC) Programme Grant. J.D.I. was a U.K. MRC Clinician Scientist throughout the course of this work and a Research Fellow of Downing College, Cambridge, U.K.

5

Abbreviations used in this paper: CML, complement-mediated lysis; ADCC, Ab-dependent cell-mediated cytotoxicity; m, mouse (e.g., mIgG); h, human; r, rat; FcγR, Fc receptor for IgG; WT, wild type.

1
Dyer, M. J. S., G. Hale, F. G. J. Hayhoe, H. Waldmann.
1989
. Effects of CAMPATH-1 antibodies in vivo in patients with lymphoid malignancies: influence of antibody isotype.
Blood
73
:
1431
2
Cobbold, S. P., S. Qin, L. Y. W. Leong, G. Martin, H. Waldmann H.
1992
. Reprogramming the immune system for peripheral tolerance with CD4 and CD8 monoclonal antibodies.
Immunol. Rev.
129
:
165
3
Neuberger, M. S., G. T. Williams, R. O. Fox.
1984
. Recombinant antibodies possessing novel effector functions.
Nature
312
:
604
4
Bruggemann, M., G. T. Williams, C. I. Bindon, M. R. Clark, M. R. Walker, R. Jefferis, H. Waldmann, M. S. Neuberger.
1987
. Comparison of the effector functions of human immunoglobulins using a matched set of chimeric antibodies.
J. Exp. Med.
166
:
1351
5
Bruggemann, M. C., C. Teale, M. R. Clark, C. Bindon, H. Waldmann.
1989
. A matched set of rat/mouse chimeric antibodies: identification and biological properties of rat H chain constant regions μ, γ1, γ2a, γ2b, γ2c, ε, and α.
J. Immunol.
142
:
3145
6
Reichmann, L., M. Clark, H. Waldmann, G. Winter.
1988
. Reshaping human antibodies for therapy.
Nature
332
:
323
7
Alters, S. E., K. Sakai, L. Steinman, V. Oi.
1990
. Mechanisms of anti-CD4-mediated depletion and immunotherapy: a study using a set of chimeric anti-CD4 antibodies.
J. Immunol.
144
:
4587
8
Bindon, C. I., G. Hale, H. Waldmann.
1988
. Importance of antigen specificity for complement-mediated lysis by monoclonal antibodies.
Eur. J. Immunol.
18
:
1507
9
Michaelsen, T. E., P. Garred, A. Aase.
1991
. Human IgG subclass pattern of inducing complement-mediated cytolysis depends on antigen concentration and to a lesser extent on epitope patchiness, antibody affinity and complement concentration.
Eur. J. Immunol.
21
:
11
10
Greenwood, J., M. Clark, H. Waldmann.
1993
. Structural motifs involved in human IgG antibody effector functions.
Eur. J. Immunol.
23
:
1098
11
Isaacs, J. D., M. R. Clark, J. Greenwood, H. Waldmann.
1992
. Therapy with monoclonal antibodies. An in vivo model for the assessment of therapeutic potential.
J. Immunol.
148
:
3062
12
Duncan, A. R., G. Winter.
1988
. The binding site for C1q on IgG.
Nature
332
:
738
13
Duncan, A. R., J. M. Woof, L. J. Partridge, D. R. Burton, G. Winter.
1988
. Localization of the binding site for the human high-affinity Fc receptor on IgG.
Nature
332
:
563
14
Cobbold, S. P., G. Martin, H. Waldmann.
1990
. The induction of skin graft tolerance in major histocompatibility complex-mismatched or primed recipients: primed T cells can be tolerized in the periphery with anti-CD4 and anti-CD8 antibodies.
Eur. J. Immunol.
20
:
2747
15
Qin, S., M. Wise, S. P. Cobbold, L. Leong, Y.-C. Kong, J. Parnes, H. Waldmann.
1990
. Induction of tolerance in peripheral T cells with monoclonal antibodies.
Eur. J. Immunol.
20
:
2737
16
Tomonari, K..
1988
. Human monoclonal antibodies.
Immunol. Today
9
:
113
17
Monaco, A., M. Wood, P. S. Russel.
1966
. Studies on heterologous anti-lymphocyte serum in mice. II. Effect on the immune response.
J. Immunol.
96
:
229
18
Pepys, M..
1972
. Role of complement in induction of the allergic response.
Nat. New Biol.
237
:
157
19
Chappel, M. S., D. E. Isenman, M. Everett, Y. Xu, K. J. Dorrington, M. H. Klein.
1991
. Identification of the Fcγ receptor class I binding site in human IgG through the use of recombinant IgG1/IgG2 hybrid and point-mutated antibodies.
Proc. Natl. Acad. Sci. USA
88
:
9036
20
Wawrzynczak, E. J., S. Denham, G. D. Parnell, A. J. Cumber, P. T. Jones, G. Winter.
1992
. Recombinant mouse monoclonal antibodies with single amino acid substitutions affecting C1q and high affinity Fc receptor binding have identical serum half-lives in the BALB/c mouse.
Mol. Immunol.
29
:
221
21
Lund, J., J. D. Pound, P. T. Jones, A. R. Duncan, T. Bentley, M. Goodall, B. A. Levine, R. Jefferis, G. Winter.
1992
. Multiple binding sites on the CH2 domain of IgG for mouse FcγRII.
Mol. Immunol.
29
:
53
22
Bindon, C. I., G. Hale, M. Bruggemann, H. Waldmann.
1988
. Human monoclonal IgG isotypes differ in complement activating function at the level of C4 as well as C1q.
J. Exp. Med.
168
:
127
23
Tao, M.-H., S. M. Canfield, S. L. Morrison.
1991
. The differential ability of human IgG1 and IgG4 to activate complement is determined by the COOH-terminal sequence of the CH2 domain.
J. Exp. Med.
173
:
1025
24
Canfield, S. M., S. L. Morrison.
1991
. The binding affinity of human IgG for its high affinity Fc receptor is determined by multiple amino acids in the CH2 domain and is modulated by the hinge region.
J. Exp. Med.
173
:
1483
25
Oi, V. T., T. M. Vuong, R. Hardy, J. Reidler, J. Dangl, L. A. Herzenberg, L. Stryer.
1984
. Correlation between segmental flexibility and effector function of antibodies.
Nature
307
:
136
26
Clark, M. R..
1997
. IgG effector mechanisms.
Chem. Immunol.
65
:
88
27
Ravetch, J. V., J.-P. Kinet.
1991
. Fc receptors.
Annu. Rev. Immunol.
9
:
457
28
Morgan, A., N. D. Jones, A. M. Nesbitt, L. Chaplin, M. W. Bodmer, J. S. Emtage.
1995
. The N-terminal end of the CH2 domain of chimeric human IgG1 anti-HLA-DR is necessary for C1q, FcγRI and FcγRIII binding.
Immunlogy
86
:
319
29
Shopes, B., M. Weetal, D. Holowaka, B. Baird.
1990
. Recombinant human IgG1-murine IgE chimeric immunoglobulin: construction, expression, and binding to human Fcγ receptors.
J. Immunol.
145
:
3842
30
Sarmay, G., R. Jefferis, E. Klein, M. Benzcur, J. Gergely.
1985
. Mapping the functional topography of Fc with monoclonal antibodies: localisation of epitopes interacting with binding sites of Fc receptor on human K cells.
Eur. J. Immunol.
15
:
1037
31
Shopes, B..
1992
. A genetically engineered human IgG mutant with enhanced cytolytic activity.
J. Immunol.
148
:
2918
32
Greenwood, J., S. D. Gorman, E. G. Routledge, I. S. Lloyd, H. Waldmann.
1994
. Engineering multiple-domain forms of the therapeutic antibody CAMPATH-1H: effects on complement lysis.
Ther. Immunol.
1
:
247
33
Smith, R. I. F., M. J. Coloma, S. L. Morrison.
1995
. Addition of a μ-tailpiece to IgG results in polymeric antibodies with enhanced effector functions including complement-mediated lysis by IgG4.
J. Immunol.
154
:
2226
34
Qin, S., S. P. Cobbold, H. Pope, J. Elliott, D. Kioussis, J. Davies, H. Waldmann.
1993
. “Infectious” transplantation tolerance.
Science
259
:
974
35
Bolt, S., E. Routledge, I. Lloyd, L. Chatenoud, H. Pope, S. D. Gorman, M. Clark, H. Waldmann.
1993
. The generation of a humanized, non-mitogenic CD3 monoclonal antibody which retains in vitro immunosuppressive properties.
Eur. J. Immunol.
23
:
403
36
Routledge, E. G., M. E. Falconer, H. Pope, I. S. Lloyd, H. Waldmann.
1995
. The effect of aglycosylation on the immunogenicity of a humanized therapeutic CD3 monoclonal antibody.
Transplantation
60
:
847
37
Isaacs, J. D., H. Waldmann.
1994
. Helplessness as a strategy for avoiding antiglobulin responses to therapeutic antibodies.
Ther. Immunol.
1
:
303
38
Isaacs, J. D., R. A. Watts, B. L. Hazleman, G. Hale, M. T. Keogan, S. P. Cobbold, H. Waldmann.
1992
. Humanised monoclonal antibody for rheumatoid arthritis.
Lancet
340
:
748
39
Isaacs, J. D., M. Wing, J. D. Greenwood, B. L. Hazleman, G. Hale, H. Waldmann.
1996
. A therapeutic human IgG4 monoclonal antibody that depletes target cells in humans.
Clin. Exp. Immunol.
106
:
427