Bi20 (fBTA05), a novel trifunctional bispecific antibody (anti-CD20 × anti-CD3), mediates efficient killing of B-cell lymphoma cells even with very low CD20 expression levels

Abstract

Trifunctional bispecific antibodies can efficiently mediate tumor cell killing by redirecting T cells and immune accessory cells to the tumor cell. Here, we describe the new trifunctional antibody, Bi20 (FBTA05, anti-CD20 × anti-CD3), that connects B cells and T cells via its variable regions and recruits FcγRI⁺ accessory immune cells via its Fc region. Bi20 mediated efficient and specific lysis of B-cell lines and of B cells with low CD20 expression levels that were derived from CLL patients. Remarkably, T-cell activation and tumor cell killing occurred in an entirely autologous setting without additional effector cells in 5 of 8 samples. In comparison, rituximab, a chimeric monoclonal CD20 antibody, demonstrated a significantly lower B-cell eradication rate. Additionally, Bi20, but not rituximab, upregulated the activation markers CD25 and CD69 on both CD4⁺ and CD8⁺ T cells in the presence of accessory immune cells. CD14⁺ accessory cells and the monocyte cell line THP-1 were activated via binding of the Fc region of Bi20, given that T cells were simultaneously engaged by the antibody. Bi20 induced a strong Th1 cytokine pattern characterized by high IFN-γ and very low IL-4 secretion. In conclusion, Bi20 may offer new immunotherapeutic options for the treatment of B-cell lymphomas. © 2008 Wiley-Liss, Inc.

Non-Hodgkin's lymphoma (NHL) and chronic lymphocytic leukemia (CLL) are among the most prevalent malignancies. NHL is the fifth most common malignant disease in the United States with an annual incidence rate of 18–20/100,000 individuals. The incidence of NHL increased annually by 5–10%1 until the mid-1990s but has remained constant since then.2 CLL belongs to the frequent NHLs and furthermore is the most common leukemia in the Western world with an annual incidence of 3–4/100,000 individuals.

Currently, only limited curative therapeutic options exist for NHL and especially for CLL, despite the fact that major improvements in therapy were achieved after the approval of monoclonal antibodies such as rituximab and alemtuzumab.3, 4 Although rituximab monotherapy has not been highly successful, particularly, in CLL, the combination of a monoclonal antibody with a standard chemotherapy regimen significantly improved patients' outcome.5 However, many patients eventually relapse,6 and new therapeutic strategies are desperately sought. Many new drug candidates such as fully humanized CD20 antibodies or antibodies targeted against other antigens (e.g., CD19, CD22, CD23, CD80, or HLA-DR) are currently under intensive investigation.7, 8

In addition to the search for new targets, another approach to improving antibody therapy was the development of bispecific antibodies. Various bispecific formats have been developed and tested, e.g., full-length antibodies produced in quadroma cells, chemically crosslinked F(ab)₂ fragments, single chain antibodies,9 and diabodies.10 Some of these antibodies have been tested for treatment of human lymphomas. However, despite promising in vitro efficacy, these molecules showed limited clinical benefit.11–13

We previously reported the development of a new class of intact bispecific antibodies, trifunctional antibodies (trAbs), that features a special combination of mouse IgG2a and rat IgG2b isotypes, allowing the simultaneous redirection and activation of FcγRI/III⁺ accessory cells and T cells.14, 15 Model trAbs showed efficient tumor cell elimination in different immunocompetent B-cell lymphoma mouse models.16, 17 The potency of this new class of bispecific antibodies could be further demonstrated with the anti-EpCAM trAb catumaxomab and anti-HER-2 trAb ertumaxomab. These trAbs share with Bi20 an identical anti-CD3 binding arm and a homologous Fc part that can effectively kill human carcinoma cells in vitro and in vivo.18–20

Here we describe the production and functional characterization of Bi20, a new trifunctional bispecific antibody directed against human CD3 and human CD20.

CD20 is a 35-kDa nonglycosylated phosphoprotein that has been proposed to act as a calcium channel, playing a role in B-cell differentiation.21 However, its exact function is still unknown as CD20-deficient mice show normal B-cell development and are also phenotypically normal.22 CD20 was chosen as a target for directed immunotherapy for several reasons: (i) it is expressed exclusively on normal and malignant B cells but not on hematological precursor cells or cells in other human tissues,23 (ii) it is expressed on most lymphoma B cells,24 and (iii) it is not shed or secreted upon antibody binding.25, 26 Furthermore, the suitability of CD20 as a target for tumor therapy has been clinically validated by the successful use of rituximab, especially in combination with conventional chemotherapy.

In our present investigations, we analyzed the in vitro properties of Bi20 in detail, demonstrating that this trAb efficiently kills human B-cell tumor cells as well as cells derived from CLL patients that express only low levels of CD20. No preactivation of effector cells was necessary for tumor cell elimination, which was accompanied by a typical Th1 cytokine pattern and strong activation of T cells and monocytes. As an important prerequisite for clinical application, Bi20 can be purified easily and produced in sufficient amounts. Therefore, this trAb might offer new treatment options for thus far incurable NHL and CLL patients.

ADCC, antibody-dependent cell-mediated cytotoxicity; BL, Burkitt's lymphoma; CD, clusters of differentiation; CLL, chronic lymphocytic leukemia; CTL, cytotoxic T lymphocyte; FcR, Fc receptor; IEF, isoelectric focusing; kDa, kilodalton; MFI, mean fluorescence intensity; NHL, non-Hodgkin's lymphoma; PBMC, peripheral blood mononuclear cells; PBL, peripheral blood lymphocytes; trAb, trifunctional bispecific antibody.

Material and methods

Antibodies and reagents

The mouse cell line TPA10, producing a CD20-specific monoclonal IgG2a antibody, was fused with 26II6, a rat cell line secreting a CD3-specific IgG2b antibody, resulting in the quadroma cell line TPBs05 producing Bi20 (FBTA05). Catumaxomab (Removab®) was used as a trifunctional control antibody, consisting of the 26II6-derived CD3-specific arm and an anti-EpCAM specificity. Rituximab (MabThera), a chimeric CD20 antibody, was purchased from Roche (Basel, Switzerland). CD20 antibody and the corresponding isotype were obtained from Beckman Coulter Immunotech (Krefeld, Germany). All other antibodies used for FACS analysis were purchased from BD Biosciences (Heidelberg, Germany). Propidium iodide was obtained from Sigma Chemicals (Deisenhofen, Germany).

Cell lines and PBMC preparation

The CLL cell line Mec1 was kindly provided by Dr. Michael Hallek (GSF, Munich, Germany). The Burkitt's lymphoma (BL) cell line Ramos was obtained from ATCC (USA). The BL cell line Raji, the NHL cell lines DOHH-2 and Granta-519 and the AML cell line THP-1 were purchased from the DSMZ (Braunschweig, Germany). Cells were grown in RPMI-1640 media (PAN-Biotech, Aidenbach, Germany) supplemented with 10% heat-inactivated fetal calf serum (FCS), nonessential amino acids, sodium pyruvate and L-glutamate (PAN-Biotech). Granta-519 cells were grown in DMEM medium instead of RPMI-1640. Human peripheral blood mononuclear cells (PBMCs) were isolated from heparinized blood of healthy donors by density gradient centrifugation using PANCOLL (PAN-Biotech). Peripheral blood samples from CLL patients were obtained after informed consent. The diagnosis of CLL was based on standard clinical and laboratory criteria. After purification, CLL cells were either immediately used or cryo-conserved for later use. CLL cells were cultivated in IMDM (PAN Biotech) supplemented with 10% FCS. For assays in the autologous setting, only freshly prepared CLL cells were used.

Generation and purification of Bi20

Bi20 was produced using the quadroma technology.27 Supernatants of quadroma cells were tested for trAb binding to target cells by FACS analysis. TrAb-producing cells were subcloned several times to increase antibody production stability, and a master cell bank was established. TrAbs were purified from the quadroma cell culture supernatant by protein A affinity and ion exchange chromatography as described28 using an ÄKTA purifier 100 (Amersham Biosciences, Uppsala, Sweden).

Isoelectric focusing

Antibody samples were spotted onto a pH 7–11 IsoGel-Agarose-IEF (Cambrex Bio Science, Rockland, USA) and separated for 75 min at 1,500 V, 50 mA and 25 W. Antibody isoforms were visualized by staining with Coomassie brilliant blue (Sigma-Aldrich, MO), and gels were scanned for analysis.

FACS analysis

Binding of antibodies to target cells, activation of effector cells and cell subtype composition were analyzed by flow cytometry using a FACScalibur (BD Biosciences, Heidelberg, Germany) equipped with a 488-nm argon laser. Data were analyzed using Cellquest Software (BD Biosciences).

To analyze antibody binding, 5 × 10⁵ cells (Ramos, Jurkat or THP-1) were incubated with 190 ng Bi20 for 30 min at 4°C, washed with PBS, supplemented with 2% heat-inactivated FCS and incubated with a FITC-labeled mouse anti-rat IgG detection antibody (Det-AB) using Ramos or a FITC-labeled rat anti-mouse detection antibody in the case of Jurkat and THP-1 cells for 30 min at 4°C. For competition experiments with rituximab, cells were preincubated with 20 μg, 4 μg, 0.8 μg, 170 ng and 5.7 ng (resulting in a 105 down to 0.03-fold excess of rituximab).

To characterize the activation properties of Bi20, 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml human tumor cells were incubated with the indicated concentrations of antibodies for 1, 2 and 3 days at 37°C in RPMI medium. Cells were harvested, washed and incubated for 30 min at 4°C with FITC- or PE-labeled detection antibodies directed against CD25 and CD69.

Cell subtype composition was determined by staining 3 × 10⁶ PBMCs with the following anti-CD antibody mixtures (FITC/PE/APC): 14/19/5, 4/8/5, 4/8/25, 45/3/5 and 16/56/5.

B-cell depletion assays

Bi20-mediated B-cell depletion was determined using a bioassay. PBMCs (1 × 10⁶/ml) obtained from healthy donors were supplemented with 2 × 10⁵/ml of the indicated tumor cells (Raji, Ramos, Mec1, DOHH-2 or Granta-519) or CLL patient cells (effector-to-target ratio 5:1) and incubated in the presence of the different antibodies at the indicated concentrations in 1 ml total volume. At Days 1, 2 and 3, cells were collected and washed, and the percentage of viable B cells (normal and tumor) was analyzed by FACS analysis using a PE-conjugated anti-CD19 detection antibody. The total number of viable cells was determined by trypan blue exclusion counting.

To examine antibody-mediated B-cell depletion in autologous CLL samples, PBMCs isolated from CLL patients were plated at a density of 2 × 10⁶ cells/ml, and antibodies were added at the indicated concentrations on Days 0, 3 and 6. FACS measurement and trypan blue exclusion counting were performed at the indicated time points. Experiments with patient CLL 15 were performed at EUFETS AG under similar conditions.

Measurement of cytokines

Bi20-induced cytokine release was determined under the same conditions used for the B-cell depletion experiments. Tumor B-cell lines and PBMCs from healthy donors were incubated with 50 ng/ml of the indicated antibodies. After 3 days, the supernatants were collected. Cytokines were measured with the human Th1/Th2 cytometric bead array (CBA, BD Biosciences, Heidelberg, Germany), which allows simultaneous analysis of INFγ, TNFα, IL-2, IL-4, IL-6 and IL-10. Data acquisition and analysis were performed according to the manufacturer's protocol.

Results

Generation and purification of Bi20

The trAb Bi20 was produced in quadroma cells and captured by protein A affinity chromatography. Parental rat antibodies do not bind to protein A and thus are absent from the eluate. Then, impurities such as parental mouse antibodies were removed by cationic exchange chromatography (CIEX) (Fig. 1a). Mouse antibodies eluted at 148 mM NaCl (Fig. 1a, Peak 1), whereas trAbs eluted at 320 mM NaCl. Isoelectric focusing of purified trAbs showed a distinct banding pattern with a pI range of ∼8.65–8.15 (Fig. 1b). TrAbs were found only in peak 2 of the CIEX separation. No contaminating parental antibodies were detected.

Figure 1. Purification of Bi20. (a) Cationic exchange chromatography (CIEX). Parental mouse antibodies (Peak 1) were separated from trifunctional antibodies (Peak 2) on a HR10/10 high performance SP-Sepharose column. (b) Isoelectric focusing (IEF) of Bi20 trifunctional antibody purification. Parental rat antibodies (26II6) were removed after protein A affinity chromatography (AC Bi20), and parental mouse antibodies (TPA10) were separated from trifunctional antibodies by CIEX (CIEX P1 Bi20). Purified Bi20 was found in Peak 2 of the cationic exchange chromatography (CIEX P2 Bi20).

Binding specificities of Bi20

To demonstrate the trifunctional nature of Bi20, we examined its binding to specific target cells. We used the CD20⁺ Burkitt's lymphoma cell line Ramos to confirm the functionality of the CD20-binding arm, the CD3⁺ T-cell line Jurkat for the CD3-binding arm and the monocyte cell line THP-1 (CD20⁻, CD3⁻) to verify Fcγ-RI binding (Fig. 2). To confirm binding of Bi20 to CD20, competitive binding assays with rituximab were performed. Bi20 binding on Ramos cells could be completely blocked by preincubation of the cells with a 21-fold excess of rituximab (Fig. 2a). Similarly, binding of Bi20 to THP-1 cells could be competed by rituximab (Fig. 2b). Binding of Bi20–CD3 on Jurkat cells was comparable to Bi20 binding to CD20 on Ramos cells (data not shown). Because the CD3-binding arm of Bi20 is identical to the corresponding arms in the well-characterized trAbs catumaxomab and ertumaxomab, no further analysis of the specificity of the CD3-binding arm was performed. In summary, these data confirm the trivalent binding mode of Bi20 and the specificity of binding.

Figure 2. Bi20 specifically binds to target and effector cells. Bi20 (190 ng) was incubated with 5 × 10⁵ target cells, and binding was measured by FACS analysis. Ramos cells (a) were used as target cells for the CD20 arm, and THP-1 cells (b) were used for the Fc part. For competition assays, cells were preincubated with rituximab (Rit) at the indicated excess of antibody for 30 min. Rituximab is shown as triangle or rectangle, depending on the excess of rituximab applied. Det-Ab: FITC labeled secondary detection antibody.

Bi20 mediates B-cell elimination

To study the effects of Bi20 on B-cell killing, we developed a FACS-based bioassay. Freshly prepared PBMCs without any prestimulation or costimulation were mixed with tumor B-cells at a ratio of 5:1 and 50 ng/ml of the indicated antibodies and incubated for 3 days. Cells were collected, and the composition of the subpopulations was determined by FACS analysis. Trypan blue exclusion counting was performed to obtain absolute cell numbers.

Using the CD20⁺ Burkitt's lymphoma cell line Raji as the target cell, effective B-cell elimination (normal and tumor B cells) was observed with 50 ng/ml Bi20 (Fig. 3a). As controls, we used the trAb catumaxomab (anti-EpCAM × anti-CD3), which does not bind to B cells (data not shown), and the parental antibodies 26II6 (anti-CD3) and TPA10 (anti-CD20). Catumaxomab (Catum) and the combination of the parental antibodies induced some B-cell depletion, probably due to the release of cytokines such as INF-γ and TNF-α (Table I), indicating that the antitumor activity of Bi20 is indeed caused by its trifunctional design. Under these experimental conditions (low antibody concentration and E:T ratio), rituximab mediated only a minor reduction in B-cell numbers. Comparable data were obtained with the CLL cell line Mec1, which has 2.5-times lower CD20 expression than Raji cells (MFI 179 versus 428, data not shown).

Figure 3. Bi20 mediated efficient B-cell depletion in the presence of effector cells. (a) 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml tumor cells (Raji, Mec1) were incubated with 50 ng/ml of Bi20, catumaxomab (Catum), rituximab, or parental antibodies TPA10 and 26II6. B-cell elimination was determined on Day 3 by trypan blue exclusion counting and FACS analysis of monocytes, B cells and T cells. The absolute B-cell number in the assays without antibody (neg) was set as 100%. Mean values and standard deviations of 3 independent experiments with different PBMC donors are shown. Differences between Bi20 and negative control, catumaxomb, rituximab and parental antibodies both with Raji or Mec1 cells are statistically significant as analyzed by Student's t-test (p < 0.05) (b) 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml Raji cells were incubated with increasing amounts of Bi20 (0.5 ng/ml up to 1 μg/ml) and rituximab (0.5 ng/ml up to 50 μg/ml). The absolute B-cell count was determined on Day 3 by trypan blue exclusion and FACS analysis. The absolute B-cell number in the assays without antibody (neg) was set as 100%. Mean values and standard deviations of 3 independent experiments with different PBMC donors are shown. Differences between Bi20 and rituximab are statistically significant as analyzed by Student's t-test (p < 0.05) (c) 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml Raji cells were incubated with increasing amounts (0.5–250 ng/ml) of Bi20. The absolute B-cell count was determined on Days 1, 2 and 3 by trypan blue exclusion and FACS analysis, and the percentage of B-cell depletion was calculated. Mean values and standard deviations of 3 independent experiments with different PBMC donors are shown.

Table I. Bi20 induced cytokine secretion

		INFγ	IL-6	TNFα	IL-10	IL-4	IL-2
1 × 10^6/ml PBMCs and 2 × 10^5/ml tumor cells (Raji, Ramos, Mec1, Granta, DOHH-2) were incubated with 50 ng/ml Bi20, catumaxomab (Catum), rituximab, or parental antibodies TPA10 and 26II6 for 3 days. Supernatants were harvested, and cytokine secretion was measured with a CBA assay and FACS analysis. Cytokine secretion is indicated as pg/ml. Mean values of two independent experiments are shown.
Raji	neg	8947	821	420	81	16	1299
	Bi20	143631	7503	1261	2012	101	11559
	Catum	157624	3505	741	433	16	10
	Rituximab	24148	3702	411	90	13	748
	26II6/TPA10	61910	5050	112	555	10	8
Mec1	neg	13086	1109	486	641	20	1257
	Bi20	49905	5773	1347	3101	70	3457
	Catum	131770	2962	1758	1222	12	10
	Rituximab	32225	3666	205	268	17	700
	26II6/TPA10	43951	2393	428	652	19	8
Ramos	neg	6336	214	264	114	46	112
	Bi20	67806	4524	1033	3298	22	1256
	Catum	97615	2408	762	1156	53	10
	Rituximab	13413	1570	188	148	24	73
DOHH-2	neg	68	117	14	30	5	17
	Bi20	110309	4314	402	1543	22	13
	Catum	116321	2980	168	578	6	20
	Rituximab	242	440	16	24	10	22
Granta	neg	8393	355	391	157	16	228
	Bi20	92807	6968	2479	1652	39	3939
	Catum	184439	3364	1518	444	19	6
	Rituximab	16259	3496	154	105	12	74

For a more detailed comparison of in vitro efficacy, rituximab and Bi20 were used in different concentration ranges: 0.5 ng/ml up to 50 μg/ml for rituximab and 0.5 ng/ml up to 1 μg/ml for Bi20 (Fig. 3b).

The dose response analysis revealed that even at the highest rituximab concentration of 50 μg/ml 35% of B cells were still viable. In contrast, 95% up to 100% of B cells were already killed at a Bi20 concentration of 50 ng/ml, depending on the PBMC donor. These data indicate that the mechanism of Bi20-mediated B cell depletion via T cell and accessory cell recruitement is much more efficient than rituximab-mediated killing.

These results encouraged us to analyze Bi20-mediated B-cell depletion in more detail. Time-kinetic analyses showed depletion of Raji cells after only 24 hr (Fig. 3c), when at least 35% of the B cells were eliminated at antibody concentrations as low as 0.5 ng/ml. At Bi20 concentrations of 250 ng/ml and 50 ng/ml, complete B-cell depletion was observed on day 2 and day 3, respectively. Comparable results were found with Mec1 cells (data not shown).

Furthermore, we investigated whether B-cell lines representing other lymphoma entities could be effectively killed. Efficient B-cell depletion was detected when DOHH-2 (follicular lymphoma), Granta (mantle cell lymphoma) or Ramos (Burkitt's lymphoma) were used as target cells (data not shown).

Bi20 mediates activation of T cells independent of concurrent B-cell binding

In contrast to previously described bispecific anti-lymphoma antibodies that display B-cell cytotoxicity only after pre-stimulation with, e.g., anti-CD28 antibodies,29 no pre-activation of T cells or monocytes was necessary for efficient B-cell lysis induced by Bi20. Therefore, we asked whether effector cells could be activated upon Bi20 binding. PBMCs and Raji cells were incubated with the indicated antibodies, and upregulation of the activation marker CD25 on CD4⁺ and CD8⁺ T cells was measured by flow cytometry (Fig. 4a). Both Bi20 and catumaxomab induced a significant upregulation of CD25 on CD4⁺ and CD8⁺ T cells, indicating that the simultaneous binding of the trAb to CD3 and CD20 is not necessary for T-cell activation. Upregulation of CD25 was also observed with a combination of the parental antibodies 26II6 (anti-CD3) and TPA10 (anti-CD20). As expected, rituximab did not induce any T-cell activation (Fig. 4a).

Figure 4. Bi20 mediated activation of T cells. (a) 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml Raji tumor cells were incubated with 50 ng/ml of Bi20, catumaxomab (Catum), rituximab, or parental antibodies TPA10 and 26II6 for 3 days. MFI of CD25 expression on CD4⁺ and CD8⁺ cells was determined by FACS analysis. (b) PBMC and Raji cells were incubated with increasing amounts (0.5–250 ng/ml) of Bi20. MFI of CD25 expression on CD4⁺ T cells was measured on Days 1, 2 and 3. Mean values and standard deviations of 3 independent experiments with different PBMC donors are shown. Differences between Bi20 and negative control and rituximab in (a) are statistically significant as analyzed by Student's t-test (p < 0.05).

As shown in Figure 4b, T-cell activation depended on incubation time and antibody concentration, with maximal CD25 expression was observed at Day 3 for the highest Bi20 concentration (250 ng/ml). Other activation markers such as CD69 were also upregulated, although to a lower extent and for a shorter time period (data not shown).

Bi20 induces T-cell proliferation

In addition to T-cell activation, we determined the increases in CD4⁺ and CD8⁺ T-cell numbers induced by the different antibodies (Fig. 5). Incubation with Bi20 resulted in the preferential proliferation of T cells of the CD8⁺ subset compared with the CD4⁺ subset (148% versus 91% expansion of T cells), changing the CD4⁺:CD8⁺ ratio from 2.0:1 to 1.5:1. Catumaxomab and the parental antibodies also induced T-cell proliferation but to a lower extent than Bi20. As expected, rituximab did not stimulate T-cell proliferation. Notably, no stimulating agent other than the antibodies was necessary to induce proliferation.

Figure 5. Bi20 induced proliferation of T cells. 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml Raji cells were incubated with Bi20, catumaxomab (Catum), rituximab, or parental antibodies TPA10 and 26II6 at a concentration of 50 ng/ml for 3 days. T-cell number was determined by trypan blue exclusion counting and FACS analysis with anti-CD5/4 and anti-CD5/8 antibodies. Mean values and standard deviations of 3 independent experiments with different PBMC donors are shown. Differences between Bi20 and negative control and rituximab are statistically significant as analyzed by Student's t-test (p < 0.05).

Monocyte activation is dependent on concurrent T-cell binding but independent of simultaneous B-cell binding

Next, we addressed the question of whether Fcγ-RI/III⁺ cells such as monocytes/macrophages could also be activated by Bi20. PBMCs and Raji cells were incubated in the presence of different antibodies, and the activation of CD14⁺ cells was measured. As illustrated in Figure 6a, both trAbs Bi20 and catumaxomab induced upregulation of CD25 on CD14⁺ monocytes/macrophages, indicating that simultaneous B-cell binding is not necessary for monocyte activation. In contrast, rituximab, which binds B cells and monocytes/macrophages but not T cells via its Fc region, did not induce activation of CD14⁺ accessory cells. From these results, we concluded that T-cell engagement is necessary for activation of accessory cells.

Figure 6. Bi20-mediated activation of monocytes. (a) 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml Raji tumor cells were incubated with 50 ng/ml of Bi20, catumaxomab (Catum), or rituximab for 1 day. MFI of CD25 expression by CD14⁺ monocytes was determined by FACS analysis. (b) 1 × 10⁶/ml PBMCs and 1 × 10⁵/ml THP-1 cells were incubated with 50 ng/ml of Bi20, catumaxomab (Catum), rituximab (Rit), TPA10, or 26II6 for 1 day. Activation of CD33⁺ THP-1 cells was determined via CD25 or CD40 expression. Mean values and standard deviations of at least 3 independent experiments are shown. Differences in (a) between Bi20 and negative value and rituximab are statistically significant as analyzed by Student's t-test (p < 0.05). Differences in (b) between Bi20 and negative control, catumaxomab and rituximab are statistically significant as analyzed by Student's t-test (p < 0.05).

We confirmed this result using a slightly different system in the absence of malignant B cells. PBMCs and THP-1 cells were incubated in the presence of different antibodies, and the activation of CD33⁺ THP-1 cells was monitored via CD25 and CD40 expression (Fig. 6b). All antibodies that simultaneously bind THP-1 cells and T cells (Bi20, TPBs01 and 26II6) activated THP-1 cells. In contrast, no activation was observed when antibodies were used that bind THP-1 and B cells but not T cells (TPA10, rituximab). THP-1 cells in the absence of PBMCs were not activated by any of the antibodies (data not shown), indicating that concurrent binding of T cells and monocytes is a prerequisite for monocyte activation in our experimental setting.

Bi20 induces a Th1-like cytokine profile

Since cytokine release is one of the immunological reactions that occurs after the therapeutic use of monoclonal antibodies,30 we studied the cytokine profile induced by Bi20. PBMCs were incubated with different tumor cells (Raji, Ramos, DOHH-2, Granta or Mec 1) and 50 ng/ml of antibodies. After 3 days, supernatants were collected and analyzed for the amounts of IFNγ, TNFα, IL-2, IL-4, Il-6 and IL-10 (Table I).

Bi20 induced high levels of IL-6, which also was detected after incubation with the other antibodies at significantly lower levels. TNF-α was detected after Bi20 or catumaxomab treatment but not after incubation with rituximab or the parental antibodies. A strong increase in IL-2, the most important autocrine growth factor for T cells, could be detected only after incubation with Bi20, confirming earlier investigations with trAbs showing that IL-2 was released only when all 3 binding partners (tumor cells, T cells and accessory cells) are present,15 except for DOHH-2 where no IL-2 increase could be detected. IL-4, a marker of the Th2 lineage, was only marginally secreted, whereas INF-γ, a Th1-specific cytokine, was strongly produced. Interestingly, IL-10, which can act as an inhibitory cytokine, was also released, suggesting an immunological counter-reaction. Similar results were obtained with all cell lines (Raji, Ramos, Mec1, DOHH-2, Granta; Table I).

Bi20 mediates elimination of B cells derived from CLL patients

We next investigated whether Bi20-mediated B-cell depletion could be confirmed by using cells derived from CLL patients. Cells from 9 different donors were incubated with PBMCs from healthy donors and Bi20 (5, 50 or 250 ng/ml) or control antibodies as illustrated in Figure 7. Increasing concentrations of Bi20 led to improved elimination (up to 99%) of CLL B cells. Rituximab also induced significant B-cell depletion, but even at a concentration of 250 ng/ml, nearly half of the B cells were viable. This shows that, at least in our experimental setting, the trAb Bi20 is much more potent than rituximab with respect to the elimination of primary tumor cells from CLL patients. In this case, unspecific B-cell depletion mediated by catumaxomab was substantially decreased compared with that in previous experiments with B-cell lines (Figs. 3 and 7).

Figure 7. Bi20 induced depletion of CLL B cells in the presence of healthy donor effector cells. 1 × 10⁶/ml PBMCs and 2 × 10⁵/ml CLL tumor cells were incubated with 5, 50, or 250 ng/ml Bi20, 250 ng/ml catumaxomab (Catum), or 250 ng/ml rituximab (Rit) for 3days. B-cell elimination was determined on day 3 by trypan blue exclusion counting and FACS analysis of monocytes, B cells and T cells. The absolute B-cell number in the assays without antibody (neg) was set as 100%. MV: mean value of CLL-1 to CLL-9. Differences between Bi20 (50 ng/ml) or Bi20 (250 ng/ml) and rituximab (250 ng/ml) are statistically significant as analyzed by Student's t-test (p < 0.05).

Remarkably, even when CLL B cells expressing very low levels of CD20 were tested (CLL 2, 3 and 4; Fig. 7), nearly all B cells were eliminated by Bi20 (99%, 93% and 98%, respectively). In contrast, rituximab showed less or no efficacy using cells from the same patient group (76%, 79% and 1%, respectively).

Next, we addressed the crucial question of whether Bi20 is able to induce killing of CLL B cells in an autologous setting without the addition of effector cells from healthy donors. Therefore, we incubated PBMCs derived from CLL patients with Bi20 and control antibodies. Because of the unfavorable E:T ratio and the known T-cell anergy in CLL, cells were incubated for at least 6 days (except for patient CLL 10), and antibodies were added every third day. Analyses of B-cell depletion and T-cell activation were performed between Days 6 and 10, except in the case of patient CLL 10 (Day 3).

As summarized in Table II, efficient B-cell depletion (>80%) was detected in 4 of 8 patient samples analyzed. In 1 sample (patient CLL 13), 65% tumor B-cell depletion was observed.

Table II. Bi20 mediated elimination of CLL B cells in vitro in an autologous setting

Patient	B-cells (%)	T-cells (%)	CD20 MFI	E:T	[Ab] ng/ml	Day	B-cell depletion (%)		T cell activation
							Bi20	Rit
Percentages of B and T cells, CD20 expression (MFI) on B cells, and E:T ratio were determined immediately after CLL PBMC preparation and before incubation. CLL cells were incubated with the indicated amounts of Bi20 or rituximab on days 0,3,6 and 9. B-cell elimination was determined as shown between day 3 and 10 depending on the patient sample. B-cell elimination is indicated as % B cells depleted. T-cell activation: CD25 expression on CD4^+ and CD8^+ T cells is at least five-fold (++) up regulated. Effector/target ratio (E:T) was defined as the number of monocytes, NK cells and T cells versus the number of B cells.
CLL 2	89.3	4.6	40	1:16	250	6	18	16	++
CLL 4	90	3.4	15	1:22	500	6	3	7	++
CLL 10	22.2	54.1	120	3.5:1	200	3	88	n. d.	++
CLL 11	62	21.9	13	1:2.1	500	6	89	36	++
CLL 12	56	24.2	22	1:1.7	500	9	100	0	n.d.
CLL 13	86	15.6	31	1:4.3	500	10	65	10	–
CLL 14	76.8	10.9	41	1:5.2	250	6	0	0	++
CLL 15	96	4.5	139	1:14.1	200	6	83	2	++

Interestingly, B-cell depletion was observed even at very low CD20 expression levels (patients CLL 11, 12, 13), confirming the results obtained in the allogeneic setting (Fig. 7). Control experiments with rituximab at the same concentrations showed no or a very low reduction in CLL B cells.

Efficient B-cell depletion seems at least in part to depend on the effector-to-target ratio (E:T). Patient samples with the best responses (CLL 10, 11 and 12) also had the highest percentages of effector cells. In particular, the sample with the most favorable E:T ratio (3.5:1; derived from patient CLL 10) displayed efficient B-cell elimination after only 3 days. In the case of Patient 15 showing an unfavorable E:T ratio also effective B-cell depletion was observed with Bi20. In contrast, rituximab failed to induce B cell killing. To verify if better response rates might be achieved with higher rituximab concentrations, a dose response analysis was performed (Fig. 8). Even at concentrations up to 100 μg/ml rituximab completely failed to show B cell depletion in the autologous setting, thus confirming our data with PBMCs and Raji cells (Fig. 3b).

Figure 8. Antibody mediated elimination of CLL B-cells in vitro in an autologous setting. 4 × 10⁶ CLL cells (CLL Patient 15) were incubated with the indicated amounts of Bi20 or rituximab on Days 0 and 3. The absolute B-cell count was determined on Day 6 by trypan blue exclusion and FACS analysis. The absolute B-cell number in the assays without antibody (neg) was set as 100%. The experiment was performed in duplicates and the average values are shown.

CLL T cells are known to be anergic, notably the CD4 subset.31 Surprisingly, when we analyzed activation of the CD4⁺ and CD8⁺ T-cell subsets by CD25 upregulation at the end of the incubation period, we found that in 6 of 7 patient samples efficient activation of both the CD4⁺ and CD8⁺ subsets had occurred (Table II). No activation was seen when samples were treated with rituximab (data not shown). These results show that, at least in vitro, Bi20 could overcome T-cell anergy in CLL samples, even in the presence of tumor cells.

Discussion

Despite promising results in the treatment of B-cell lymphomas using the anti-CD20 monoclonal antibody rituximab (especially in combination with chemotherapy), patients eventually relapse.6, 32 Therefore, there is a need for further therapeutic options. One improvement might be the use of bispecific antibodies, which redirect effector cells such as T cells to the tumor cells.

Bispecific antibodies directed against B-cell-specific antigens such as CD19 or CD20 have been under intensive investigation for the last 10–15 years, but limited clinical data showing only moderate responses11, 33, 34 are available thus far. Moreover, the complicated and inefficient manufacturing process makes it difficult to produce sufficient amounts of antibody for clinical use. Here, we describe the generation and production of Bi20, a new trifunctional bispecific antibody directed against CD20 and CD3. The use of the isotype combination mouse IgG2a and rat IgG2b enables the high-yield production of correctly paired bispecific antibodies due to species-restricted heavy/light chain pairing.27 These trAbs are characterized by simultaneous binding of tumor cells, T cells and FcγRI/III⁺-accessory cells, thus enhancing tumor cell eradication.14, 15

Bi20 was shown to be very efficient in B-cell elimination in vitro without the need for any additional costimulation. B-cell depletion assays demonstrated that even at an antibody concentration as low as 0.5 ng/ml, nearly 70% of Raji tumor B cells could be eliminated after 24 hr, with complete tumor cell eradication after 3 days. By using B-cell lines representative of other NHL entities (Mec1, Granta or DOHH-2), similar results were obtained.

Furthermore, B cells derived from CLL patients also were efficiently killed (Fig. 7, Table II), despite the increased apoptosis resistance and low CD20-expression levels that are characteristic of CLL B cells.24, 35, 36 Again, no additional preactivation or coactivation of T cells was necessary for efficient CLL B-cell elimination, in contrast to many other bispecific antibodies that have been described. For example, the bispecific mouse antibody Bis20 (mouse IgG1 CD20 × mouse IgG2b CD3) induced nearly 60% B-cell depletion at the same E:T ratio used in our experiments but with a 10-fold higher antibody concentration, preactivated T cells and a shorter incubation time.37 In another study using an anti-CD20 × anti-CD3 diabody, lysis of tumor cells (Raji) could only be achieved when PBLs were prestimulated with IL-2. Even at a diabody concentration of 1,000 ng/ml and an E:T ratio of 20:1, only about 35% of cells were lysed.38 Furthermore, Cochlovius and colleagues showed for an anti-CD3 × anti-CD19 diabody that preactivated PBLs and T-cell costimulation via CD28 are necessary for efficient B-cell lysis at antibody concentrations of 1–2.5 μg/ml.29, 39

Bi20 induced efficient tumor cell killing in 5 of 8 patient samples, even in an autologous setting without addition of healthy donor PBMCs. Remarkably, in contrast to rituximab, Bi20-mediated B-cell depletion is even effective in the presence of very low CD20 expression levels by the CLL tumor cells, which holds true for the autologous as well as the allogeneic setting (Table II and Fig. 7). This finding is in accordance with clinical data of CLL studies in which rituximab shows only limited efficacy as a monotherapy, possibly due to low CD20 expression by the CLL tumor cells.40, 41

In contrast to several other bispecific antibodies,42, 43 Bi20 efficiently activated both CD4⁺ and CD8⁺ T-cell subsets, even in the absence of any costimulatory molecules such as anti-CD28 antibodies or IL-2. Consequently, we detected high levels of INF-γ, which is known to be released by stimulated T cells and which is an indicator of a Th1-type response.44 Full activation of T cells, as seen by the production of IL-2, was only detected with Bi20 and is, therefore, dependent on the presence of all 3 binding partners, reducing the risk of intolerable systemic IL-2 concentrations during in vivo application. T-cell activation was also accompanied by proliferation of both T-cell subsets.

T-cell activation was not only observed using healthy donor T cells but also with T cells from B-CLL patients (Table II). These T cells are often characterized by anergy, probably due to low CD28 and T-cell receptor expression.45, 46 In our experiments, activation of the CD4⁺ and CD8⁺ T-cell subsets was detected even in those samples in which no substantial CLL B-cell depletion was observed. The limited B-cell depletion in some samples (CLL-2, CLL-4 and CLL-14) might be due to the short incubation period that is restricted by the limited viability of B-CLL cells in vitro. A comparable B-cell depletion based on coactivator-independent T-cell activation was recently described with a single chain CD19 × CD3 bispecific antibody, further confirming the potency of the bispecific antibody format.47, 48 Thus, our results indicate that Bi20 might have the potential to overcome CLL T-cell anergy not only in vitro but also in vivo.

In addition to T cells, Bi20 also binds to and activates Fcγ-RI⁺ cells. Activation is dependent on concomitant binding of Bi20 to T cells because binding of tumor target cells and Fcγ-RI⁺ cells is not sufficient for activation, as shown with the parental antibody TPA10 (Figs. 6a and b). The importance of the Fc region of trAbs is supported by our recent observation that in an immunocompetent mouse B-cell lymphoma model, the induction of long-lasting antitumor immunity was dependent on a functional Fc part of the trAb. In these experiments, the use of a F(ab)₂ fragment of a trAb significantly diminished survival of these mice.17 This result is consistent with a recent publication49 showing that a diabody combination of anti-CD19 × anti-FcγRIII and anti-CD19 × anti-CD3 showed higher antitumor activity compared with a single diabody in vitro and in a mouse model, although in both cases additional costimulation of T cells via an anti-CD28 antibody was necessary. These results indicate that simultaneous binding and activation of T cells and FcγR⁺ effector cells as mediated by Bi20 allows more efficient tumor cell killing than binding to tumor cells and T cells alone especially in vivo.

Taken together, our results show that the new trAb Bi20 is a potent inducer of tumor B-cell depletion that is not dependent on any preactivation or coactivation, even when CD20 expression by the tumor cells is very low. The simultaneous activation of T cells and accessory immune cells and their redirection to the tumor cells may offer the possibility to overcome the T-cell anergy described for CLL. Therefore, Bi20 could offer new therapeutic options to treat non-Hodgkin's lymphoma and CLL. First clinical experience in compassionate use trials with Bi20 in combination with allogeneic donor lymphocyte transfusion in 3 CLL and 3 high-grade NHL patients suggested the safe clinical application of Bi20, which is now further tested in a phase I clinical trial.50

Acknowledgements

We would like to thank Dr. Jens Herold and Dr. Jürgen Hess for helpful discussions and Mr. Reinhard Liedtke for statistical analysis. Expert technical assistance from Ms. Cornelia Wagner is gratefully acknowledged.