Abstract
Purpose: The purpose of this study was to prepare chimeric antihuman CD22 tetravalent monoclonal antibodies (MAbs) with high functional affinity, long persistence in the circulation, increased antitumor activity, and conserved effector function in vitro.
Experimental Design: We investigated the association/dissociation rates of these tetravalent antibodies using CD22+ Daudi lymphoma cells. We then tested their ability to interact with Fc receptors on a human cell line (U937), to mediate antibody-dependent cellular cytotoxicity with human natural killer cells, to bind human C1q, to inhibit the in vitro growth of CD22 Daudi cells, and to persist in the circulation.
Results: The rate of dissociation of the tetravalent MAbs versus the divalent antibody was considerably slower. These tetravalent MAbs inhibited the in vitro proliferation of CD22 Daudi cells at a concentration that was at least 100-fold lower than that of the divalent murine antibody. The tetravalent MAbs containing both the CH2 and CH3 domains and a chimeric recombinant divalent antibody bound similarly to Fc receptor, C1q, and mediate antibody-dependent cellular cytotoxicity equally well with human natural killer cells. The persistence in the circulation of chimeric tetravalent MAbs was considerably longer than that of chemical homodimers.
Conclusions: The tetravalent anti-CD22 MAbs with intact Fc regions should make effective therapeutic agents for B-cell tumors.
INTRODUCTION
Monoclonal antibodies (MAbs) can elicit antitumor responses by modulating cellular activity and/or by recruiting effector cells or molecules or by delivering cytotoxic agents to cells (1). Several lymphoma-reactive MAbs have entered routine clinical practice to treat B lymphomas in humans (e.g., Rituxan and Zevalin). In previous studies, we have demonstrated that some MAbs (anti-CD19, -CD20, -CD21, and -CD22) that have little or no inherent anti-growth activity on lymphoma cell lines can be converted into potent antitumor agents by using them as tetravalent homodimers (2, 3). These chemically generated homodimers exert at least part of their anti-growth activity by signaling cell arrest and/or apoptosis of the targeted tumor cell line. These activities might be enhanced in vivo by the recruitment of effector cells and/or complement (4, 5), although this has not yet been tested using our homodimers. Despite their remarkable antitumor activity both in vitro and in vivo (2, 3), the chemical homodimers have disadvantages. These include a high molecular mass (300 kDa), heterogeneity (they also contain trimers and tetramers), a short half-life in vivo, and immunogenicity because of their murine origin. To address these drawbacks, we have generated recombinant, chimeric, tetravalent antihuman CD22 MAbs. These MAbs have a molecular mass closer to that of divalent MAbs (200 kDa). They are homogeneous, have a longer half-life in vivo, and should be less immunogenic because most of the molecule is of human origin.
In this study, four recombinant tetravalent MAbs were generated using a procedure recommended for the preparation of bispecific antibodies (6). These chimeric tetravalent molecules were prepared from the Fv portions of the murine antihuman CD22 MAb, RFB4 (7), spliced to either the CH3 domain (cRFB4CH3) or both the CH2 and CH3 domains (cRFB4CH2-CH3) of human IgG1. Two of these “heavy chains” associate spontaneously to yield the tetravalent constructs. As a reference MAb, a divalent recombinant chimeric anti-CD22 MAb (cRFB4) was also generated.
We evaluated the association/dissociation and cytotoxicity of these recombinant constructs using CD22+ target cells (Daudi cells). Their ability to interact with Fc receptors (FcRs) was determined by their binding to U937 cells, and their ability to mediate antibody-dependent cellular cytotoxicity (ADCC) was evaluated using human natural killer (NK) cells. Finally, binding to C1q and persistence in circulation was determined.
Our results demonstrate that both the tetravalent constructs containing CH2-CH3 and CH3 had slower dissociation rates from CD22+ cells than the divalent RFB4 MAb and are cytotoxic to Daudi cells; the divalent RFB4 IgG is not. The constructs containing both CH2 and CH3 domains, but not those lacking CH2, bind to FcRs, mediate ADCC, bind to C1q, and have a longer half-life than chemical homodimers.
MATERIALS AND METHODS
Construction and Expression of Recombinant Chimeric MAbs.
The heavy and light chains of chimeric RFB4 were generated using vectors containing the human IgG1 and κ constant domains, pAH4604 and pAG4622, a generous gift from Dr. Sherie Morrison (7). The tetrameric chimeric CH2-CH3 and CH3 constructs of RFB4 were generated using expression vectors kindly provided by Dr. Roland Kontermann (6). The cloning strategy and configurations of the expressed products are outlined in Figs. 1 and 2, respectively.
The heavy and light variable domain sequences of RFB4 have been reported previously (8). These domains were separately PCR-amplified from cDNA preparations from the hybridoma using primers that anneal in their respective inferred leader peptides and the constant domains (to confirm the reported sequence contained no PCR artifacts) using primers (which include additional flanking restriction sites; the sequences complementary to the cDNA are underlined): heavy leader, 5′-GGGTCTAGATATCCACCATGGACTTCGGGCTCAGCTTGG-3′; heavy constant, 5′-AGGGAATTCA(C/T)CTCCACACACAGG(A/G)(A/G)CCAGTGGATAGAC-3′; light leader, 5′-GGGTCTAGATATCCACCATGATGTCCTCTGCTCAGTTCCTTGGTC-3′; and light constant, 5′-GCGGAATTCGCTCACTGGATGGTGGGAAGATGGA-3′.
These PCR products were separately cloned into pUC18 using XbaI and EcoRI restriction sites, and the variable domain sequences were verified. These cloned fragments were used as the starting point in the generation of the cRFB4, cRFB4CH2-CH3, and cRFB4CH3.
The heavy and light leader sequence primers listed above were also designed to include restriction sites for insertion directly into the Morrison expression vectors (7) to generate chimeric RFB4 (cRFB4; Fig. 2 C). These primers were used in conjunction with the following COOH-terminal primers that include sites for in-frame insertion into these vectors as well as flanking sites for insertion into pUC18 (SphI and MluI, respectively) for sequence verification after PCR: heavy 3′-end, 5′-ACGCATGCTAGCGGCTGCAGAGACAGTGACCAGAG-3′; and light 3′-end, 5′-CGACGCGTCGACTTACGTTTGATTTCCAGCTTGGTGCCTCC-3′.
The heavy and light chains were inserted into pAH4604 and pAG4622, respectively, using EcoRV/NheI and EcoRV/SalI sites. These constructs were used to cotransfect SP2/0 murine nonsecreting myeloma cells using Lipofectamine (Invitrogen, Carlsbad, CA). Stable transfectants were positively selected using 5 mm l-histidinol (Sigma, St. Louis, MO). Secreting clones were selected by ELISA screening for secreted human immunoglobulin.
PCR primers were designed to assemble, through a series of overlap extensions (9) of the heavy and light variable domains, a single-chain diabody (HL) consisting of VH-linker (5-mer)-Vl-linker (15-mer)-Vh-linker (5-mer)-Vl or into tandem repeats of a single-chain variable (LH) domain, Vl-linker (15-mer)-Vh-linker (5-mer)-Vl-linker (15-mer)-Vh.
For the HL configuration (Figs. 1,A and 2 A), the following primers, designated A through F, were used: primer A, 5′-GGGCGGCCCAGCCGGCCAGGCGCGAAGTGCAGCTG GTGGAG-3′; primer B, 5′-TGAACCGCCTCCACCTGCAGAGACAGTGACCAG-3′; primer C, 5′-GGTGGAGGCGGTTCAGACATCCAGATGACACAGAC-3′; primer D, 5′-CC GGATCCGCCTGAACCGCCTCCACCTTTGATTTCCAGCT TGGTGCC-3′; primer E, 5′-GCGGATCCGGCTCTGGCGGTGGCGGATCGGAAGTGCAGCTGGTGGAG-3′; and primer F, 5′-ATAGTTTAGCGGCCGCTTTGATTTCCAGCTTGGTGCC-3′.
For the LH configuration (Figs. 1,B and 2 B), the following primers were used: primer G, 5′-GGGCGGCCCAGCCGGCCAGGCGCGACATCCAGATGACACAGAC-3′; primer H, 5′- CGATCCGCCACCGCCAGAGCCACCTCCGCCTGAACCGCCTCCACCTTTGATTTCCAGCTTGGTGCC-3′; primer I, 5′-GGTGGAGGCGGTTCAGGCGGAGGTGGCTCTGGCGG-TGGCGGATCGGAAGTGCAGCTGGTGGAG -3′; primer J, 5′-GCGGATCCGCCTGAACCTGCAGAGACAGTGACCAG-3′; primer K, 5′-GCGGATCCGACATCCAGATGACACAGAC-3′; and primer L, 5′-ATAGTTTAGCGGCCGCTGCAGAGACAGTGACCAG-3′.
Using the engineered SfiI and NotI restriction sites, these sequences were inserted into the CH3 and CH2-CH3 expression vectors for in-frame expression of the chimeric tetrameric antibodies. These plasmids were separately transfected into HEK293 human embryonic kidney cells using Lipofectamine (Invitrogen). Stable transfectants were selected using 200 μg/ml zeocin (Invitrogen). After subcloning, positive lines were chosen by screening culture supernatants using a human immunoglobulin-specific ELISA.
Purification of Recombinant Chimeric MAbs.
Proteins present in the cell culture supernatant (containing tetravalent antibodies) were precipitated at 4°C with ammonium sulfate at 50% saturation. The precipitate was collected by centrifugation, dissolved in a small volume of distilled water, and dialyzed overnight against 50 mm phosphate buffer with 300 mm NaCl at pH 8. Imidazole (Sigma) and Tween 20 (Sigma) were added to the dialysate to final concentrations of 25 mm and 0.05%, respectively. The mixture was loaded into a column of Ni-NTA-agarose (Qiagen, Valencia, CA) equilibrated with 50 mm phosphate buffer containing 300 mm NaCl, 25 mm imidazole and 0.05% Tween 20 (pH 8). The bound proteins were eluted with the same buffer containing an increased concentration of imidazole (250 mm) and dialyzed overnight against 50 mm phosphate buffer containing 150 mm NaCl (pH 7.4).
To further purify the cRFB4CH2-CH3 constructs, the dialysate was affinity-purified on protein G-Sepharose (Amersham Biosciences, Piscataway, NJ), and the bound proteins were eluted with 3.5 m MgCl2. For the cRFB4CH3 constructs, a column containing a polyclonal rabbit IgG antibody against the RFB4 idiotype (10) coupled to Sepharose 4B (Amersham Biosciences) was used. The bound protein was eluted with 3.5 m MgCl2. In all cases, the eluted proteins were dialyzed against 150 mm NaCl solution and then PBS, concentrated to 1.0 mg/ml, filter-sterilized, and stored at 4°C for a maximum of 1 month.
The cRFB4 MAb was precipitated with ammonium sulfate at 50% saturation and purified by affinity chromatography on protein G-Sepharose as described above. The chemical dimer was obtained by cross-linking RFB4 MAb as described previously (2).
Radioiodination.
MAbs and goat antihuman C1q (Sigma) were radiolabeled with Na125I (Amersham Biosciences) using Iodogen reagents as described previously (11). The free 125I was removed by centrifugation on MicroSpin G-25 columns (Amersham Biosciences). The specific radioactivity of the labeled proteins was in the range of 107cpm/μg with <10% free 125I.
SDS-PAGE.
Proteins were separated by 4–15% SDS-PAGE (Amersham Biosciences) under nonreducing condition using a PhastSystem (Amersham Biosciences). The gel was stained with PhastGel Blue R (Amersham Biosciences).
Pharmacokinetic Analysis.
Female 6–8-week-old Swiss mice (Charles Rivers Laboratories, Wilmington, MA) were used. Lugol solution was added to their drinking water to a concentration of 0.05% from 1 day before injection throughout the entire period of the experiment (168 h). Radiolabeled proteins were injected into the tail veins of five mice in a volume not larger than 150 μl, and whole body radioactivity was measured daily in an AtomLab 100 dose calibrator (Atomic Product Corp., New York, NY). The pharmacokinetic parameters were determined using a noncompartmental model with the PKCALC program (12), using data collected between 24 and 168 h.
Cells.
The CD22+ cell lines Daudi and Namalwa and the CD22− FcR+ U937 cells were maintained in culture by serial passages in RPMI 1640 (ICN Biomedicals, Aurora, OH) containing 25 mm HEPES, 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 μg/ml streptomycin, and 2 mm glutamate (complete medium). The cells were grown in a humidified atmosphere of 5% CO2 and air. Cell viability was determined by trypan blue exclusion.
[3H]Thymidine Assay.
The cells in complete medium (5 × 104 cells/100 μl) were plated in 96-well plates and incubated for 48 h with 100 μl of different concentrations of the MAb constructs (ranging from 10−6 to 10−9 m). After incubating the cells at 37°C, the plates were pulsed for 4 h with 1 μCi of [3H]thymidine (Amersham Biosciences), harvested, and counted in a liquid scintillation spectrometer. The percentage reduction in [3H]thymidine incorporation versus the concentration of MAbs was used to quantitate the cytotoxic effect (expressed in IC50).
Binding of MAbs to Daudi and U937 Cells.
The cells were suspended in complete medium at 5 × 106 cells/ml and treated with different concentrations of radiolabeled MAbs (0.1–5 μg/107 cells/ml). After incubation on ice for 90–120 min, the cells were centrifuged and resuspended in 0.5 ml of complete medium. Three aliquots of 100 μl each were applied to 400 μl of a mixture of phthalic acid/phthalate oil (Sigma) in microcentrifuge tubes of 0.4 ml (VWR Scientific, Suwanee, GA) and centrifuged at 1800 × g for 20 min. The tubes were frozen at −80°C, and the tips of the test tubes were cut off. The radioactivity in both the tip (cell-bound radioactivity) and the remainder of the tube (free radioactivity) was measured. The binding curves represent the amount of protein added (in μg/107cells/ml) versus the amount bound (ng/107 cells). For RFB4 and cRFB4(LH)4CH2-CH3 MAbs, Scatchard plots were generated, and the binding constants (Ka) were calculated as recommended previously (13).
Dissociation of MAbs from Daudi Cells.
Daudi cells were suspended in complete medium at 5 × 106 cells/ml and incubated with 1–5 μg/ml of the radiolabeled MAbs on ice for 1 h. After incubation, the cells were washed with cold complete medium, centrifuged, and resuspended at 5 × 106 cells/ml. Unlabeled (“cold”) RFB4 in complete medium was added to a final concentration of 2 mg/ml, and the cells were incubated in 5% CO2 at 37°C for various intervals of time. The dissociation of the radiolabeled MAbs was also measured in the absence of cold RFB4. At each interval of time, triplicate aliquots of 100 μl were removed and centrifuged, and the radioactivity of the pellet and supernatant was measured. The results were expressed as the percentage reduction of the bound radioactivity versus time.
ADCC with Human NK Cells.
NK cells were obtained from human blood using the RosetteSep NK cell enrichment mixture (StemCell Technologies, Vancouver, Canada) as recommended by the manufacturer. The purity of the NK population was 70–85% as determined by staining the cells with FITC antihuman CD56 (Sigma). The cytotoxicity assay was performed as recommended previously (14). Briefly, human Namalwa cells were coated with cRFB4 MAbs at 50 μg/106 cells in 100 μl at 4°C for 30 min and, after washing, mixed with human NK cells at an E:T ratio of 70:1 in 400 μl of complete RPMI 1640. Cells were incubated for 4 h at 37°C in a 5% CO2 incubator. After incubation, the cells were washed and incubated with human IgG (1 μg/100 μl) to block FcRs. FITC-labeled mouse antihuman CD19 (HD37) MAb was added at 4°C for 20 min. After washing, propidium iodide (10 μl/100 μl; Biosource International, Camarillo, CA) were added, and the samples were analyzed on Becton Dickinson FACS (San Jose, CA). The percentage of dead Namalwa cells was determined.
Binding of Human C1q by Daudi Cells Coated with MAbs.
Daudi cells at 5 × 106 cells/ml in complete medium containing 0.01% sodium azide were treated with 5 μg/ml MAbs and incubated for 1 h at 4°C. After washing the cells with ice-cold barbital-buffered saline (Sigma), the cells were incubated with human C1q (1–2 μg/ml; Sigma) at 4°C for 30 min, followed by centrifugation. The cells were resuspended in the same buffer at the initial concentration, and radiolabeled goat antihuman C1q (Sigma) was added. The mixture was incubated for 30 min at 4°C. After washing out the unbound ligand, the radioactivity in the pellet was measured in a gamma counter, and the values (after subtracting the binding to the negative control) were expressed as a percentage relative to the binding of cRFB4, which was taken as 100%.
RESULTS
Purification of the Recombinant Chimeric MAbs.
The recombinant constructs [cRFB4(LH)4CH2-CH3, cRFB4(HL)4CH2-CH3, cRFB4(LH)4CH3, and cRFB4(HL)4CH3] were purified by affinity chromatography on Ni-NTA-agarose and either protein G-Sepharose (for CH2-CH3-containing MAbs) or a rabbit anti-RFB4 idiotypic antibody conjugated to Sepharose (for CH3-containing MAbs). The cRFB4 was purified directly from the dissolved ammonium sulfate precipitate on protein G-Sepharose. Purity was evaluated by SDS-PAGE; there were no differences in the purity or the molecular masses of the CH2-CH3 and CH3 constructs irrespective of their HL or LH configuration. SDS-PAGE of both CH2-CH3 variants migrated as a major band of 200 kDa (Fig. 3). The CH3-containing constructs migrated with an apparent molecular mass of 70 kDa (Fig. 3). Because these CH3 constructs are devoid of a hinge region disulfide bond, they dissociated in SDS, and, hence, their true molecular mass is 140 kDa. By SDS-PAGE, the molecular mass of murine RFB4 and the cRFB4 were indistinguishable (Fig. 3).
Association and Dissociation of MAbs from Daudi Cells.
The binding of radiolabeled MAbs to Daudi cells showed the differences depicted in Fig. 4. The amount of each construct bound at the highest amount added (5 μg/107 cells/ml) ranged, in decreasing order, from cRFB4(LH)4CH2-CH3 > cRFB4(HL)4CH3 > murine RFB4 > cRFB4(HL)4CH2-CH3 > cRFB4 > RFB4 chemical dimer > cRFB4(LH)4CH3. However, when the functional affinities were measured by Scatchard analysis, cRFB4(LH)4CH2-CH3 (highest binding capacity) and murine RFB4 (reference MAb) were similar (2.5 × 109 m−1 versus 1.5 × 109 m−1; Fig. 4). These results suggest that there are no major differences in the avidity of these MAbs for their CD22 target antigen. In contrast, the dissociation of these radiolabeled MAbs from the surface of Daudi cells in the presence of the same cold competitor (RFB4) showed distinct differences (Fig. 5). From the dissociation curves, the persistence of the MAbs on the cell surface could be calculated and expressed as a half-life of dissociation (t1/2). The t1/2 values of the various MAbs decreased in the following order: cRFB4(LH)4CH2-CH3 (t1/2 = 143 min) > cRFB4(HL)4CH2-CH3 (t1/2 = 105 min) > cRFB4(LH)4CH3 (t1/2 = 91 min) > RFB4 (t1/2 = 32 min). In the absence of competitor, the dissociation of all four MAbs was negligible (<10%).
The results demonstrate that the strength of binding is dependent on the valency of the constructs. All of the tetravalent constructs had a significantly longer t1/2 of dissociation than the divalent constructs, suggesting that their increased persistence was a consequence of their multivalency.
Binding of MAbs to U937 Cells.
The binding of MAbs to human U937 cells (FcR I/III+) indicated that binding is dependent on the presence of an intact Fc region of human origin (Fig. 6). Therefore, only cRFB4, cRFB4(LH)4CH2-CH3, and cRFB4(HL)4CH2-CH3 bound to U937 cells, whereas cRFB4(HL)4CH3, which lacks the FcRI/III binding region, and RFB4, which has a murine Fcγ region, showed very low binding to the human U937 cells. Therefore, the chimeric recombinant cRFB4 constructs with an intact human Fc region should mediate ADCC with human NK cells.
ADCC Mediated by Human NK Cells.
The cytotoxicity of the cRFB4 MAbs is shown in Fig. 7. Both cRBF4 and cRFB4(LH)4CH2-CH3 mediated the killing of Namalwa cells by FcR III+ human NK cells. The MAbs cRFB4(LH)4CH3 and murine RFB4 were devoid of any activity. These results suggest that tetravalent cRFB4 with an intact Fc region should mediate ADCC in vivo by NK cells.
Binding of C1q to MAb-Coated Daudi Cells.
The binding of human C1q to Daudi cells previously incubated with the MAbs was measured by treating the MAb-coated cells with C1q and detecting the bound C1q with a radiolabeled anti-C1q antibody. We used this procedure because the direct binding of radiolabeled C1q was sometimes associated with a high background. As shown in Table 1, only constructs of human origin with intact Fc regions bound human C1q. Thus, cRFB4-coated Daudi cells bound an amount of C1q that was similar to that of both cRFB4(LH)4CH2-CH3 and cRFB4(HL)4CH2-CH3, whereas the CH3-containing cRFB4s lacking CH2 did not bind C1q.
Inhibition of Cell Growth.
The inhibition of the growth of Daudi cells by the various MAb constructs was determined by the inhibition of [3H]thymidine incorporation. The inhibitory activities of cRFB4(LH)4CH2-CH3 and cRFB4(HL)4CH3 were comparable with those of the chemically cross-linked RFB4 homodimer. All three had IC50 values in the range of 10−8 m, whereas the RFB4 monomer was not inhibitory even at 10−6 m (Fig. 8).
Pharmacokinetics.
The clearance curves of the radiolabeled MAbs in mice are shown in Fig. 9. From these curves the pharmacokinetic parameters were calculated, and values are summarized in Table 2. The persistence in the circulation (t1/2 and area under the curve), fractional catabolic rate, and mean residence time of the MAbs demonstrate that none had the pharmacokinetic characteristics of the murine and chimeric RFB4. Thus, the t1/2 values of the cRFB4(HL)4CH2-CH3 and cRFB4(LH)4CH2-CH3 were about 175 h versus 235.7 h for murine RFB4 and 211.9 h for cRFB4. However, they were still longer than that of the chemical homodimer of RFB4 (102.8 h; P < 0.01). As expected, cRFB4(LH)4CH3 has a very short half-life (26.0–33.9 h) due to the absence of the CH2-CH3 domain interface that controls the catabolism of IgG MAbs.
DISCUSSION
The objective of this study was to generate several different recombinant, chimeric, tetravalent, anti-CD22 antibodies using the V-regions from murine RFB4 and the C-regions from human IgGl. This was accomplished by assembling the gene fragments encoding the heavy and light domains by overlap PCR and ligation and inserting them into expression vectors containing either the CH3 domain or the entire Fc fragment. These plasmids were used to transfect HEK293 cells. A chimeric divalent IgG1 κ version of RFB4 was also constructed and used as a control. The secreted antibodies were purified from the culture supernatants by affinity chromatography and tested for their ability to bind to CD22+ cells, dissociate from these cells, bind to FcRs, mediate ADCC, and bind C1q in vitro. Their pharmacokinetics in mice were determined and compared with those of the divalent parent murine antibody. Finally, their cytotoxic activity on CD22+ tumor cells was determined and compared with chemical homodimers and divalent (IgG) antibody. The major findings to emerge from this study are as follows: (a) the binding affinities of tetravalent MAbs to Daudi cells were higher than those of the divalent MAbs (RFB4 and cRFB4), but the differences were relatively small. Thus, the binding affinity of cRFB4(LH)4CH2-CH3 versus RFB4 as determined by Scatchard analysis was not significant. The fact that large differences were not observed is due primarily to the irreversible nature of the binding of these MAbs to the cell surface. This was clearly indicated by the virtual lack of dissociation of these ligands from the surface of Daudi cells at 37°C in the absence of the competitor. Thus, the equilibrium Ka (functional affinity) cannot be accurately determined for irreversible reactions (15, 16) such as this. (b) The tetravalent constructs persist for longer times on the cell surface. Dissociation, as compared with association, seems to provide a more reliable estimate of the strength of binding of ligands (15, 17). However, we were unable to determine whether the tetravalent antibodies engaged all four binding sites. However, due to their slower dissociation rate, they clearly use more than two of their binding sites. A similarly constructed tetravalent MAb with different specificity was reported to involve at least three of the four valencies in binding its corresponding antigen (18). (c) The tetravalent MAbs and the chemical homodimer had comparable inhibitory effects on the growth of Daudi cells (IC50 of 10−8 m). In contrast, the divalent (IgG) RFB4 was devoid of cytotoxicity, even at a concentration that was 100-fold higher. This cytotoxic effect is probably due to the ability of the tetravalent MAbs to hypercrosslink the targeted CD22 antigen (3). (d) Only the tetravalent MAbs containing both the CH2 and CH3 domains bound to FcR+ (I/III+) U937 cells, mediated ADCC with FcR III+ human NK cells, and bound human C1q. The binding to both FcR and C1q requires the presence of an intact hinge region and a CH2 domain (19). The hinge region of both LH and HL cRFB4CH2-CH3 but not cRFB4 is slightly different from that of the human IgG1. Thus, the human IgG1 hinge sequence is EPKSCDKTHTCPPCP, whereas the CH2-CH3 construct hinge sequence is AAADKTHTCPPS. However, this peculiar structure of the hinge of the CH2-CH3-containing MAbs did not influence the binding to C1q or FcR (I/III+). Thus, the binding of both tetravalent CH2-CH3-containing MAbs to C1q or U937 cells was comparable with that of cRFB4, which has a hinge region identical to that of human IgG1. The ability of cRFB4(LH)4CH2-CH3 to mediate ADCC by human NK cells was comparable with that of cRFB4. These results suggest that the tetravalent MAbs with CH2-CH3 domains should mediate NK-dependent ADCC and activate complement in vivo. (e) The half-life of the CH2-CH3-containing MAbs in mice was significantly longer than that of the chemical homodimer but shorter than that of either murine or chimeric RFB4 IgG. The reason for the shorter half-lives of the CH2-CH3-containing MAbs is not clear because they both should have CH2-CH3 interfaces that are identical to that of human IgG1, which has a half-life in mice similar to that of mouse IgG1 (20). As expected, the tetravalent MAbs that were devoid of CH2 domains had very short half-lives, comparable with that of the Fab fragments of IgG (21). Because the persistence of MAbs in the circulation is an important factor for therapeutic activity, the CH2-CH3-containing MAbs should be more effective in vivo.
In summary, the tetravalent recombinant MAbs constructed with the mouse Fv region of the antihuman CD22 MAb, RFB4, and the Fc region of human IgG1 are homogenous proteins with a molecular mass comparable with that of divalent MAbs but with higher functional affinities. Due to their increased valencies, these tetravalent MAbs constructs are 100-fold more cytotoxic to the CD22+ tumor cells than their divalent counterparts and also have conserved effector function (they bind to FcRs/C1q) and persist in the circulation because of their human Fc moiety. It remains to be determined whether these tetravalent MAbs will have improved antitumor activity in xenografted severe combined immunodeficient mice or humans with CD22+ tumors.
Grant support: NIH Grant CA64679-07.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Present addresses are, as follows: Ruiqi Meng, Section of Molecular Bacteriology, Veterinary Institute, University of Agriculture and Animal Sciences, 175 Xi’an Road, Changchun, Jilin Province, People’s Republic of China; Michael Yen, Department of Oncology, Stanford University School of Medicine, Stanford, California 94305.
Requests for reprints: Ellen S. Vitetta, Cancer Immunobiology Center, University of Texas Southwestern Medical Center at Dallas, 6000 Harry Hines Boulevard, NB9.210, Dallas, TX 75390-8576. Phone: (214) 648-1200; Fax: (214) 648-1204; E-mail: [email protected]
RFB4 ligand on Daudi cells . | C1q binding (%)a . |
---|---|
cRFB4 | 100b |
cRFB4(LH)4CH2-CH3 | 93 ± 15 |
cRFB4(HL)4CH2-CH3 | 81 ± 18 |
cRFB4(LH)4CH3 | No binding |
cRFB4(HL)4CH3 | No binding |
RFB4 | No binding |
RFB4 ligand on Daudi cells . | C1q binding (%)a . |
---|---|
cRFB4 | 100b |
cRFB4(LH)4CH2-CH3 | 93 ± 15 |
cRFB4(HL)4CH2-CH3 | 81 ± 18 |
cRFB4(LH)4CH3 | No binding |
cRFB4(HL)4CH3 | No binding |
RFB4 | No binding |
Average of two separate experiments carried out in triplicate.
The binding to cRFB4 was considered to be 100%.
MAbb . | T1/2 (h) . | AUC . | FCR (day−1) . | MRT (h) . |
---|---|---|---|---|
Murine RFB4 | 235.7 ± 28.0 | 36,677 ± 3,655 | 0.071 ± 0.008 | 339.8 ± 40.2 |
cRFB4 | 211.9 ± 11.7 | 22,430 ± 725 | 0.078 ± 0.005 | 301.0 ± 16.4 |
Murine RFB4 dimer | 102.8 ± 7.4 | 11,222 ± 693 | 0.162 ± 0.012 | 147.5 ± 11.0 |
cRFB4(HL)4CH2-CH3 | 175.6 ± 7.6 | 19,435 ± 712 | 0.094 ± 0.004 | 249.4 ± 10.9 |
cRFB4(LH)4CH2-CH3 | 174.6 ± 15.0 | 20,148 ± 1,837 | 0.096 ± 0.008 | 249.0 ± 22.0 |
cRFB4(LH)4CH3 | 33.9 ± 0.7 | 2,727 ± 117 | 0.497 ± 0.011 | 33.7 ± 1.4 |
cRFB4(HL)4CH3 | 26.0 ± 0.2 | 2,229 ± 132 | 0.615 ± 0.005 | 24.7 ± 2.4 |
MAbb . | T1/2 (h) . | AUC . | FCR (day−1) . | MRT (h) . |
---|---|---|---|---|
Murine RFB4 | 235.7 ± 28.0 | 36,677 ± 3,655 | 0.071 ± 0.008 | 339.8 ± 40.2 |
cRFB4 | 211.9 ± 11.7 | 22,430 ± 725 | 0.078 ± 0.005 | 301.0 ± 16.4 |
Murine RFB4 dimer | 102.8 ± 7.4 | 11,222 ± 693 | 0.162 ± 0.012 | 147.5 ± 11.0 |
cRFB4(HL)4CH2-CH3 | 175.6 ± 7.6 | 19,435 ± 712 | 0.094 ± 0.004 | 249.4 ± 10.9 |
cRFB4(LH)4CH2-CH3 | 174.6 ± 15.0 | 20,148 ± 1,837 | 0.096 ± 0.008 | 249.0 ± 22.0 |
cRFB4(LH)4CH3 | 33.9 ± 0.7 | 2,727 ± 117 | 0.497 ± 0.011 | 33.7 ± 1.4 |
cRFB4(HL)4CH3 | 26.0 ± 0.2 | 2,229 ± 132 | 0.615 ± 0.005 | 24.7 ± 2.4 |
Values from 5 mice were averaged.
MAb, monoclonal antibody; t1/2, half life (β); AUC, area under the curve; FCR, fractional catabolic rate; MRT, mean residence time.
Acknowledgments
We thank Tia Bowdler for assistance in preparing the manuscript. We thank Ming-Mei Liu, Ana Firan, Stephanie Kufert, and Linda Trahan for technical assistance. We are indebted to Dr. S. Morrison (UCLA) for providing us with the human IgG1 and κ plasmids and Dr. R. Kontermann (University of Marburg) for the human CH3 and CH2-CH3 expression vectors.