Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Acute Leukemias

Generation, selection and preclinical characterization of an Fc-optimized FLT3 antibody for the treatment of myeloid leukemia


The therapeutic efficacy of humanized or chimeric second-generation antitumor antibodies is clearly established, but often limited. In recent years, defined modifications of the glycosylation pattern or the amino-acid sequence of the human immunoglobulin G1 Fc part have resulted in the development of third-generation antibodies with improved capability to recruit Fc receptor-bearing effector cells. The first antibodies of this kind, currently evaluated in early clinical trials, are directed against lymphoma-associated antigens. Fc-engineered antibodies targeting myeloid leukemia are not yet available. We here report on the generation and preclinical characterization of an Fc-optimized antibody directed to the FMS-related tyrosine kinase 3 (FLT3), an antigen expressed on the leukemic blasts of all investigated patients with acute myeloid leukemia (AML). This antibody, termed 4G8SDIEM, mediated markedly enhanced cellular cytotoxicity against FLT3-expressing cell lines as well as blasts of AML patients. FLT3 expression levels on AML cells varied between 300 and 4600 molecules/cell and, in most cases, were substantially higher than those detected on normal hematopoietic precursor cells and dendritic cells (approximately 300 molecules/cell). Antibody-mediated cytotoxicity against these normal cells was not detectable. 4G8SDIEM has been produced in pharmaceutical quality in a university-owned production unit and is currently used for the treatment of leukemia patients.


The chimeric monoclonal antibody Rituximab, directed to the CD20 antigen expressed on normal and malignant cells of the B-lymphocytic lineage, has considerably improved treatment of B-cell-derived lymphomas. In contrast, treatment of non-hematopoietic solid tumors with monoclonal antibodies to different tumor-associated antigens has been less successful. The success of Rituximab has been attributed to several particularities: First, the antibody effectively mediates Fc-dependent effector mechanisms, such as complement- and antibody-dependent cellular cytotoxicity (ADCC). In addition, it induces apoptotic cell death at high concentrations.1 Last, and probably the most important, hematopoietic malignancies, in contrast to solid tumors, are more easily accessible to drugs and immunologic effector cells.

Because of their well-documented -but nevertheless still limited- therapeutic activity, antilymphoma antibodies are currently the first to benefit from newly developed strategies that use antibody engineering technologies to further enhance the therapeutic efficacy of antitumor antibodies.2 Many of these strategies focus on the modulation of the human immunoglobulin (Ig)G1 Fc part that is contained in the majority of antitumor antibodies in clinical use. By defined genetic engineering of the glycosylation pattern or the amino-acid sequence of the Fc part, antibody variants with markedly improved ADCC activity have been generated.3, 4 Different companies use different strategies: Roche (Basel, Switzerland), in cooperation with Glycart (Schlieren, Switzerland), has developed an improved, glyco-engineered CD20 antibody, termed GA101.5 Two other antibodies directed to the lymphoma-associated antigens CD19 (XmAb5574)6, 7 and CD30 (XmAb2513),8 developed by Xencor (Monrovia, CA, USA), carry the amino-acid exchanges S239D and I332E (SDIE modification). All these antibodies were reported to exert markedly enhanced ADCC. At present, these and other1 third-generation antibodies directed to lymphoma-associated antigens are in late preclinical and early clinical developmental stages, respectively.

Leukemias share with lymphomas the advantage of easy accessibility for drugs and immunologic effector cells, and the need for effective new treatment modalities is even more pressing. In addition, the therapeutic activity of antileukemia antibodies is particularly easy to monitor. Nevertheless, efforts to develop antibodies for the treatment of leukemia have been less intense and less successful, compared with those directed to lymphomas: a carefully characterized, humanized antibody directed to the pan-myeloid marker CD33 turned out to be of limited efficacy in clinical trials.9 More recently, humanized antibodies against more specific antigens on leukemic cells, such as the FMS-related tyrosine kinase 3 (FLT3, CD135; clinical trials ID NCT00887926) and the α-chain of the interleukin-3 receptor (CD123),10 have been developed to the stage of early clinical trials. These antibodies, however, have not been optimized by recombinant antibody technology, and the antigens recognized are known to be expressed on dendritic cells (DCs), normal hematopoietic stem cells,11 and —in the case of CD123— cultured mast cells12 and basophils,13, 14 raising concerns about potential toxicity towards these cells.

We have focused our efforts on the SDIE modification mentioned above and substituted this modification by adding a C-terminal tag (M-tag) to facilitate antibody detection in vitro and in vivo. After cloning of the respective variable genes, this modification (SDIEM) was used to generate optimized human IgG1 versions of two FLT3 antibodies developed at our institution15 and of the CD19 antibody 4G7 originally generated at the University of California, Stanford.16 In all these cases, we found a markedly enhanced ADCC activity of the SDIEM-modified antibodies compared with that of the unmodified, chimeric versions. Here we report on the preclinical characterization of the two FLT3 antibodies, BV10 and 4G8. FLT3 is a class III receptor tyrosine kinase, which frequently carries mutations in patients with acute myeloid leukemia (AML) resulting in constitutive activation.17 It has an important role in the homeostasis of DCs, and its ligand, FLT3L, has been used in several studies to activate these cells in vitro and in vivo.18, 19, 20, 21, 22 Thus, during characterization of SDIEM-modified FLT3 antibodies we put particular emphasis on the evaluation of antibody specificity and toxicity towards DCs and hematopoietic stem cells in addition to the exploration of ADCC activity. We believe that the results obtained in this study provide a solid basis for clinical pilot studies.

Materials and methods

Production and purification of recombinant and Fc-optimized antibodies

The messenger RNA of mouse FLT3 antibodies BV10 and 4G8, as well as of the CD19 antibody 4G7 (all Igγ1/κ), was isolated from hybridomas with the RNeasy Kit (Qiagen, Hilden, Germany). Unknown variable regions of heavy (VDJ) and light (VJ) chains were identified by sequencing of inverse PCR amplicons generated as previously described,23 using specific primers for mouse constant genes of light (Ck-for: 5′-IndexTermTGTTCAAGAAGCACACGACTGAGGCACCTCC-3′; Ck-back: 5′-IndexTermACTTCTACCCCAAAGACATCAATGTCAAG-3′) and heavy chains (gamma1-for: 5′-IndexTermCAAGGCTTACAACCACAATCCCTGG-3′; gamma1-back: 5′-IndexTermCATATGTACAGTCCCAGAAGTATCATCTG-3′). The sequences for the variable genes of 4G7 were published (GenBank no. AJ555479 and AJ555622). The cloning of the variable genes from the hybridoma 9.2.27 (GenBank no. AJ459796 and AJ459797), producing an Igγ2a/κ antibody directed to the melanoma-associated chondroitin sulfate proteoglycan, has also been described previously.24 For the generation of chimerized and optimized antibodies, the VJ and VDJ elements were re-amplified and cloned into eukaryotic expression vectors. Besides the amino-acid exchanges at S239D and I332E, the optimized γ1 Fc part contains a C-terminal M-tag, which was derived from the amino-acid sequences 455–466 of the human Igα1 tailpiece and the c-myc epitope EQKLISEEDLLR.25 Parental and recombinant BV10, 4G8 and 4G7 antibodies were purified from culture supernatant of hybridoma cells, and Sp2/0 and CHO-transfectants, respectively, using protein A affinity chromatography (GE Healthcare, Munich, Germany). In the case of 4G8SDIEM, a large batch of the antibody (15 g) was produced in GMP (good manufacturing practice)-compliant clean rooms using disposable technology including a 100-l biowave reactor (Sartorius, Goettingen, Germany) for fermentation and an ÄKTAready system for purification by protein A, ion exchange and hydrophobic interaction chromatography (MabSelect Sure and CaptoAdhere columns, GE Healthcare).

A bispecific Fab2 fragment (bsFab2) with FLT3 × CD3 (4G8 × UCHT1) specificity was produced using a previously described conjugation procedure.26

Cells and media

Peripheral blood mononuclear cells (PBMCs), isolated by density gradient centrifugation using LSM 1077 solution (Lonza, Basel, Switzerland), hybridoma cells and NALM-16 cells (ACC 680, DSMZ, Braunschweig, Germany) were kept in RPMI 1640, mouse Sp2/0-Ag14 cells (ATCC, Manassas, VA, USA) in IMDM medium (Lonza). All media were supplemented with 10% heat-inactivated fetal calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, non-essential amino acids, 2 mM L-glutamine and 57 nM β-mercaptoethanol.

FLT3 transfectants

Full-length complementary DNA of human FLT3 (GenBank no. BC126350) was obtained from ImaGenes (Berlin, Germany). The N-terminally FLAG (DYKDDDDK)-tagged derivatives coding either for the complete FLT3 molecule (aa 1–970) or for variants with truncated ectodomains, Δ1 (aa 162–970), Δ2 (aa 244–970), Δ3 (aa 346–970), Δ4 (aa 435–970), were cloned into the pcDNA3 vector using added BamHI and XbaI restriction sites and transfected into Sp2/0-Ag14 cells by electroporation.

Antibodies and flow cytometry

CD33-PE/Cy5 (clone WM53), CD34-APC (clone 581), CD-45-FITC (clone HI30), CD123-PE/Cy5 (clone 9F5), CD11c-PE (clone B-ly6) and isotype control antibodies were purchased from BD Biosciences (Heidelberg, Germany), the CD303-FITC antibody from Miltenyi Biotec (Bergisch-Gladbach, Germany) and FLAG-FITC (clone M2) from Sigma-Aldrich (Deisenhofen, Germany). All antibodies were incubated with cells for 30 min at 4 °C. For indirect immunofluorescence, PE- or APC-conjugated goat-anti-mouse F(ab)2 fragments and goat-anti-human F(ab)2 fragments, respectively, were used (Jackson ImmunoResearch, West Grove, PA, USA). In several experiments, we combined indirect and direct immunofluorescence for multi-dimensional analysis by adding labeled antibodies in the final step. Cells were analyzed on a FACSCanto II or a FACSCalibur (BD Biosciences).

Beads for the quantitative analysis of indirect immunofluorescence (QIFIKIT) were purchased from Dako (Hamburg, Germany) and used according to the manufacturer's protocol. For quantification of humanized antibodies suitable beads were not available. Thus, a specific fluorescence index was calculated by dividing the mean fluorescence intensity obtained with 4G8SDIEM by that detected with the non-binding, SDIE-modified control antibody 9.2.27. For these experiments, PE-conjugated antibodies generated with the Lynx rapid PE antibody conjugation kit (AbD Serotec, Düsseldorf, Germany) were used. Bone marrow cells were stained using CD34-APC/Cy7, CD38-PE/Cy7 (all from BioLegend, San Diego, CA, USA) and CD45-AmCyan (BD Biosciences) antibodies.

For direct assessment of stem cell toxicity, 5 × 105 bone marrow cells were incubated with 1 μg/ml of 4G8SDIEM, 9.2.27SDIE or FLT × UCHT-1 bsFab2, respectively, in triplicate. After incubation for 24 or 48 h the cells were stained and resuspended in FACS buffer containing 7-AAD (BioLegend) and BD negative control compensation particles (BD Biosciences). Stem cells were gated according to the ISHAGE protocol.27 Absolute numbers of CD34+CD38+ and CD34+CD38− cells were determined by acquiring the same numbers of compensation beads for all samples.

Recombinant FLT3 ligand (rFLT3L) was purchased from Peprotech EC (London, Great Britain, UK). For competition experiments various concentrations of rFLT3L were incubated with NALM-16 cells and BV10SDIEM or 4G8SDIEM (1 μg/ml) for 30 min at 4 °C and analyzed by indirect immunofluorescence and flow cytometry.

3[H]-methyl-thymidine uptake assay

2 × 105 AML blasts were seeded in triplicate in 96-well plates and incubated with various concentrations of optimized antibodies. After 24 h, cells were pulsed for another 20 h with 3[H]-methyl-thymidine (0.5 μCi/well) and harvested on filter mats. Incorporated radioactivity was determined by liquid scintillation counting in a 2450 Microplate counter (Perkin-Elmer, Waltham, MA, USA).

51[Cr]-release assays

NALM-16 cells and primary AML blasts were used as targets. To separate blasts and effector cells from the PBMC preparations of leukemia patients, cells were labeled with CD34 and CD33 microbeads and separated on LD columns following the manufacturer's (Miltenyi Biotec) protocol. The number of contaminating blast cells in the negatively selected effector cell population was determined by FACS analysis and varied between 1 and 10% depending on the initial blast contamination. In some experiments labeled DCs were used as target cells.

Chromium release assays were performed as previously described.28 Briefly, labeled target cells and PBMCs were incubated at 37 °C for 4 or 8 h in 96-well flat-bottom plates in the presence of various concentrations of antibodies at a PBMC : target ratio of 50:1. Percentage of specific 51[Cr] release was calculated according to the formula (c.p.m. (test)–c.p.m. (spontaneous)/(c.p.m. (triton lysis))–c.p.m. (spontaneous)).

Antigen shift

NALM-16 cells or AML blasts were incubated with various concentrations of 4G8SDIEM or BV10SDIEM in RPMI 1640 medium. After 24 or 48 h the cells were washed with ice-cold FACS buffer, incubated with a saturating concentration of 4G8SDIEM (2 μg/ml) for 30 min at 4 °C, stained with PE-conjugated goat-anti-human F(ab)2 fragments and analyzed by FACS. Relative surface expression of FLT3 was calculated, defining the mean fluorescence intensity of cells preincubated without antibody as 100%.

DC isolation and maturation

DCs were isolated from buffy coat preparations of healthy individuals using the blood DC isolation kit II according to the protocol of the manufacturer (Miltenyi Biotec). Myeloid DC and plasmacytoid DC subsets were stained with a mixture of CD11c-PE, CD303-FITC and CD123-PE/Cy5 antibodies. For in vitro generation of myeloid DC, 1 × 108 PBMCs of healthy individuals were seeded in 10 ml X-Vivo15 medium (Gibco, Darmstadt, Germany). After 2 h at 37 °C adherent cells were cultured in RPMI 1640 medium supplemented with 50 ng/ml GM-CSF and interleukin-4 (20 ng/ml) for 5 days. On day 6 lipopolysaccharide (100 ng/ml) was added. Cells were harvested on day 7 and analyzed by flow cytometry.

Colony-forming unit assay

Bone marrow cells were obtained by lavage of the femoral head from patients undergoing hip surgery. The cells were purified by density gradient centrifugation and seeded at 1 × 107/ml in RPMI 1640 medium containing 5 μg/ml of 4G8SDIEM or 9.2.27SDIE. After 24 h of incubation, cells were transferred into antibody-containing (5 μg/ml) methylcellulose medium (10 000 cells/ml, MethoCult H4434 classic, Stemcell Technologies, Grenoble, France). The assay was performed in triplicate. After 12 days, colonies were counted and classified.


FLT3 specificity

The parental mouse antibodies 4G8 and BV10 were originally described and characterized as recognizing the FLT3 protein,15 a class III receptor tyrosine kinase with five extracellular immunoglobulin-like domains.29 Epitope mapping using different deletion variants of the molecule revealed binding of BV10 and 4G8 antibodies to domain 2 and 4, respectively (Figure 1a). This is consistent with previous observations that both antibodies do not cross-block each other in competition experiments (data not shown). Both antibodies saturated FLT3 molecules on NALM-16 cells at concentrations below 1 μg/ml. Binding of the chimerized 4G8 antibody was stronger than that of BV10. This is not due to chimerization or optimization as a similar difference was observed when binding of the mouse parental versions of 4G8 and BV10 was compared. No differences in binding between the chimeric and the SDIEM-modified chimeric versions of the antibodies were detected (Supplementary Figure 1). A SDIE-modified antibody, termed 9.2.27SDIE, directed to the melanoma-associated surface antigen chondroitin sulfate proteoglycan, did not bind to NALM-16 cells and was used as a negative control in this and several subsequent experiments.

Figure 1

Differential binding of 4G8 and BV10 antibodies to the FLT3 molecule, effect of 4G8 on proliferation of leukemic blasts. (a) Binding of 4G8 and BV10 to different FLAG-tagged variants of the FLT3 molecule transfected into Sp2/0 cells was determined by flow cytometry. Δ1–Δ4 variants lack ectodomains 1, 1+2, 1+2+3 and 1+2+3+4, respectively. In the case of Δ4, which is recognized by neither antibody, expression is verified by staining with an antibody directed against the FLAG-tag. (b) NALM-16 cells were incubated with 4G8SDIEM or BV10SDIEM at 1 μg/ml in the presence of the indicated concentrations of the recombinant FLT3 ligand and the amount of bound antibody was determined by indirect immunofluorescence and flow cytometry. (c) AML blasts isolated from the peripheral blood of three different patients by density gradient centrifugation were incubated with the indicated concentrations of 4G8SDIEM for 24 h and proliferation was assessed using a 3[H]-thymidine uptake assay. Bars on the right represent proliferation in the absence of the antibody. SFI, specific fluorescence index.

Competition with binding of the FLT3 ligand

In general, interference with binding of the natural ligand may contribute to the therapeutic activity of an antibody. Figure 1b shows that recombinant FLT3L partly inhibits binding of 4G8SDIEM, but not of BV10SDIEM to NALM-16 cells. Therefore, we evaluated the effect of 4G8SDIEM on the spontaneous proliferation of the leukemic blasts of three different patients in vitro using the non-binding SDIE-modified 9.2.27 antibody as a control. Whereas spontaneous proliferation of the primary AML cells varied substantially, significant effects of the antibodies on cell proliferation were not observed up to a concentration of 10 μg/ml (Figure 1c).

Antibody-dependent cellular cytotoxicity

The hallmark of various SDIE-modified antibodies evaluated previously was their impressively increased activity in ADCC assays. Figure 2a shows that the ADCC activity of PBMCs against NALM-16 cells is markedly enhanced in the presence of the SDIEM-modified antibodies as compared with that of the unmodified chimeric antibody versions. In several experiments, the concentrations required to achieve comparable lysis by unmodified and SDIEM-modified antibodies differed by a factor of at least 100. Killing by the 4G8SDIEM antibody was significantly better than that achieved by BV10SDIEM, in particular at low concentrations. This may be due to the moderately lower binding avidity of BV10 (Supplementary Figure 1) and/or the different FLT3 epitopes recognized (Figure 1a). The superior activity of 4G8SDIEM prompted us to focus on this antibody for further characterization. In Figure 2b the ADCC activity of 4G8SDIEM is compared with that of the optimized CD19-antibody 4G7SDIEM. After 4 h a significant difference was observed in favor of the CD19 antibody corresponding to the fourfold-higher expression level of CD19 on NALM-16 cells (Supplementary Figure 2). After 20 h, however, the activity of both antibodies appeared to be comparable. In Figure 3a the ADCC activity of 4G8SDIEM is depicted using PBMCs of three additional healthy donors (PBMC I–III). In these experiments, the SDIE-modified monoclonal antibody 9.2.27 was used as a negative control. The cytolytic activity in the presence of this reagent did not exceed that of natural killer (NK) cells in the absence of antibodies, which varied between 0 and 20%.

Figure 2

ADCC activity of unmodified and SDIEM-modified versions of the FLT3 antibodies 4G8 and BV10. (a) 51[Cr]-labeled NALM-16 cells were incubated for 4 h with PBMCs of a healthy donor (PBMC IV) in the presence of the indicated concentrations of the unmodified chimeric (χ) or SDIEM-modified versions of 4G8 and BV10 at a PBMC : target cell ratio of 50:1. Killing of the target cells was determined using a standard 51[Cr]-release assay. One representative result of six independent experiments with PBMCs from different healthy donors is depicted. (b) Experimental conditions as in (a). ADCC activity of 4G8SDIEM is compared with that of 4G7SDIEM after 4 and 20 h. The results of one experiment that is representative of three experiments are shown.

Figure 3

ADCC activity of 4G8SDIEM against leukemic cells. Cytolytic activity of the PBMCs of three different healthy donors (PBMC I, II, III) against NALM-16 cells (a) and of the PBMCs of donor II against leukemic blasts of three different patients (AML 1, 2, 11) (b) was determined in a 4- and 8-h 51[Cr]-release assay, respectively. In (c) the cytolytic activity after 8 h against AML blasts 1 and 21 is depicted using autologous PBMCs of the respective patients as effector cells. Filled and open symbols indicate ADCC mediated by 4G8SDIEM and non-binding control antibody 9.2.27SDIE, respectively. Filled bars on the right (NK) indicate NK activity in the absence of antibody.

In Figure 3b the ADCC activity of PBMCs from a healthy donor (PBMC II) against leukemic blasts of three different patients is shown. ADCC activity mediated against these blasts (AML 1, AML 2, AML 11), carrying 3750, 4000 and 2600 FLT3 molecules per cell, respectively, was less pronounced than that against cultured NALM-16 cells. It required 8 rather than 4 h to become clearly detectable. Generally, the ADCC as well as the NK activity against NALM-16 cells and leukemic blasts continued to rise after 8 h (see Figure 2b).

We next wanted to evaluate the ADCC activity of PBMCs isolated from the blood of AML patients against autologous blasts. To this end, we depleted leukemic blasts from PBMC preparations and used the depleted PBMCs as effector cells against the positively selected blasts (see Materials and Methods). Under these conditions, we detected significant lysis in two (AML 1, 21) out of five independent experiments with blasts and autologous PBMCs of the respective patients (Figure 3c). It should be noted that this procedure has possible technical shortcomings: (i) contamination of effector cells with residual blasts may result in the deviation of E:T (effector cell to target cell) ratios in favor of the blasts and (ii) lysis may not be complete after 4 and potentially even 8 h (Figure 2b). To overcome the latter restriction, we tried to establish a long-term cytolytic assay covering 24 rather than 4 or 8 h. These experiments failed because of an exceedingly high spontaneous death rate of AML blasts. The extraordinary fragility of AML blasts under standard culture conditions became apparent already after 8 h of culture, when added allogeneic effector cells in most cases inhibit the spontaneous release of chromium rather than enhancing it by exerting NK-mediated lysis. This stabilizing effect resulted in a formally negative NK effect in 51[Cr]-release assays (Figure 3b).

Binding to normal and leukemic cells, antigen shift

Figures 4a and c show binding of the parental mouse 4G8 antibody and 4G8SDIEM, respectively, to a panel of leukemic cells obtained from patients suffering from the indicated subtypes of AML. Gated CD33+CD45dim or CD34+CD45dim cells were analyzed. FLT3 was detected in all 24 patient samples. The number of molecules per cell determined by indirect immunofluorescence and quantitative flow cytometry varied from 350 to 4700. On NALM-16 cells approximately 6800 molecules/cell (mpc) were detected (Figure 4a). Our panel contained only four samples with confirmed positive internal tandem repeat status (Figure 4b), two of them with an intermediate and two with a low expression of FLT3. Thus, these data are not conclusive regarding a correlation between the expression level and the mutation status of the FLT3 molecule. In Figure 4c, 4G8SDIEM-PE rather than mouse 4G8 was used for staining. In this case, a specific fluorescence index value was calculated to quantify antibody binding. For blasts from 7 of the 24 donors, this index was not determined because of high, unspecific reactivity with the control antibody 9.2.27SDIE-PE. As expected, specific fluorescence index values of the evaluable samples closely matched the numbers of molecules determined by quantitative FACS (Figure 4c).

Figure 4

FLT3 expression and antigen shift on leukemic cells of different origin. (a) AML blasts from 24 patients were incubated with mouse 4G8 (10 μg/ml), washed and analyzed by indirect immune fluorescence and flow cytometry. The amount of bound antibody molecules was determined by comparison with calibrated beads (QIFIKIT). (b) FLT3 mutation status of some of the leukemic samples used (ITD: internal tandem duplication; TKD: tyrosine kinase domain). (c) The AML blasts used in (a) were incubated with PE-conjugated 4G8SDIEM or non-binding PE-conjugated 9.2.27SDIE antibody (10 μg/ml) and analyzed by direct immunofluorescence and flow cytometry. The SFI of four samples was not determined (n.d.) because of high binding of the 9.2.27SDIE control antibody. (d) NALM-16 cells and blasts from two different AML patients were incubated with the indicated concentrations of 4G8SDIEM. After 48 h cells were washed, re-incubated with 2 μg/ml of 4G8SDIEM and analyzed by indirect immunofluorescence and flow cytometry. FLT3 expression detected on cells preincubated without antibodies was defined as 100%. SFI, specific fluorescence index.

Modulation of target antigen expression upon antibody binding is a phenomenon often observed during antibody therapy. In particular, a sustained and complete loss has been reported upon treatment of AML patients with a saturating dose of the CD33 antibody Lintuzumab.9 Figure 4d depicts the antigen shift induced after incubation of NALM-16 cells or primary leukemic blasts of two patients (AML 1 and 2) with various concentrations of 4G8SDIEM for 48 h. On all these cells, a moderate antigen shift was observed, which was already completed after 24 h of incubation (data not shown).

Figures 5a–c show that binding of mouse 4G8 to CD11c-positive myeloid DCs and to CD303-positive plasmacytoid DCs purified from normal PBMCs was marginal. The numbers of FLT3 molecules expressed on these cells were below 300/cell. In addition, we generated DCs from normal PBMCs. Although these cells expressed large amounts of the DC-associated markers CD80, CD86 and CD123, binding of 4G8 antibodies was again barely detectable (data not shown). Next, we evaluated binding of mouse 4G8 to CD34-positive cells in normal bone marrow. Again, binding of the antibody to bone marrow cells of three different donors was marginal, corresponding to less than 600 molecules per cell (Figures 5d and e). In summary, binding of FLT3 antibodies to normal DCs and bone marrow cells was significantly lower than that to most FLT3-expressing leukemic cells examined (see Figures 5f–h). In addition, we did not observe binding of FLT3 antibodies to thrombocytes, erythrocytes and granulocytes (data not shown). This is in accordance with our failure to detect FLT3 on a wide array of normal and malignant cryopreserved tissues by immunohistology using 4G8 as well as BV10 antibodies (data not shown).

Figure 5

Expression of FLT3 on normal DCs and bone marrow cells. (a) DCs isolated from the peripheral blood of healthy donors by magnetic cell separation were incubated with mouse 4G8, washed, stained with a labeled secondary antibody, washed again and incubated with a mixture of differently labeled CD11c and CD303 antibodies. Cells were then analyzed by flow cytometry. Binding of 4G8 to the CD303+ plasmacytoid dendritic cell and the CD11c+ myeloid dendritic cell subpopulation is depicted in (b) and (c), respectively. (d, e) Similar to (ac), normal bone marrow cells isolated by density gradient centrifugation were incubated with mouse 4G8, washed, stained with labeled secondary antibody and a mixture of differently labeled CD34−, CD45− and CD38 antibodies. Binding of 4G8 to the CD34+CD45dimSSClowCD38− and CD38+ subpopulation is depicted in (d) and (e), respectively. (fh) Staining of leukemic blasts of AML samples 2 (high FLT3 expression, 4050 mpc), 16 (intermediate FLT3 expression, 1700 mpc) and 3 (low FLT3 expression, 500 mpc) with 4G8 for comparison. Gray and black histograms represent primary staining with isotype control, and mouse 4G8, respectively. Representative results from one of three experiments with DCs and bone marrow cells from different healthy donors are shown.

Toxicity in vitro

Despite the relatively low levels of 4G8SDIEM binding to normal bone marrow precursor cells and DCs, we attempted to assess the potential toxicity of this antibody towards such cells. To this end, we incubated bone marrow cells with saturating concentrations of 4G8SDIEM and 9.2.27SDIE, and determined the influence of these antibodies on the capacity of the bone marrow cells to give rise to colonies (colony forming units) in semi-solid medium. No significant influence of the antibodies on the capacity to form colony-forming units was detected in two experiments with bone marrow cells from different healthy donors (Figure 6a). To assess toxic effects of 4G8 SDIEM on human hematopoietic precursor cells more directly, we incubated bone marrow cells of healthy donors with various antibodies at 1 μg/ml for 24 and 48 h and determined the percentage of the CD34+CD38− as well as CD34+CD38+ subpopulations by flow cytometry. In these experiments, a chemically hybridized antibody with FLT3 × CD3-specificity served as a positive control. This antibody activates T cells in the presence of FLT3-expressing target cells (data not shown), resulting in a significant reduction of the CD34+CD38− as well as the CD34+CD38+ population. The control antibody 9.2.27SDIE as well as 4G8SDIEM did affect hematopoietic progenitor cells in these experiments neither after 24 (Figure 6b) nor after 48 h (data not shown).

Figure 6

Cytotoxic activity of 4G8SDIEM against normal cells. (a) Human bone marrow cells from two different healthy donors (black and white bars) were incubated with 5 μg/ml of 4G8 SDIEM and colony forming units were determined after 12 days of incubation in semi-solid medium. Numbers of colony-forming units were related to untreated controls. (b) Bone marrow cells of healthy donors were incubated with 1 μg/ml 4G8SDIEM, 9.2.27SDIE (negative control) and a bispecific Fab2 fragment with 4G8 × UCHT1 specificity (positive control). After 24 h survival of CD34+CD38− and CD34+CD38+ cells was determined by flow cytometry. Representative results of one of three experiments with bone marrow cells of different donors are shown. (c) DCs isolated from the PBMCs of healthy donors by magnetic cell separation and NALM-16 cells were used as targets for 4G8SDIEM in a 4-h51[Cr]-release assay (PBMC : target ratio 100:1). One representative experiment of three with DCs and autologous PBMCs from different donors is shown.

Finally, we incubated human DCs with autologous PBMCs as effector cells. Although 4G8SDIEM mediated effective ADCC against NALM-16 cells, when used as positive control, no killing of autologous DCs was observed (Figure 6c).


Despite their undisputed success in lymphoma therapy, up to date, effective monoclonal antibodies for the treatment of myeloid leukemias have not been developed. The FMS-like receptor tyrosine kinase FLT3 (CD135), expressed on the leukemic cells of most AML patients, has been considered a promising target for small drug molecules,30 such as tyrosine kinase inhibitors, as well as for monoclonal antibodies.31 Antibodies have the potential advantage of superior efficiency due to induction of immune effector mechanisms. In addition, their activity does not rely on the FLT3 mutation status.

Here we report on the optimization of two FLT3 antibodies using recombinant antibody technology. Initially, we considered two different strategies for optimization of the Fc part: modifications of the glycosylation pattern and the amino-acid sequence, respectively. We chose the latter strategy because (i) the stable fermentation of producer cells with altered glycosylation patterns is delicate; (ii) we found in previous experiments that the ADCC activity of a SDIEM-modified CD19 antibody was comparable to that of an otherwise identical reagent with optimized, low-fucose glycosylation (data not shown). Thus, for the experiments described here, we focused on the SDIEM modification described above and generated (i) chimeric versions of the parental mouse antibodies carrying constant regions of the human IgG1 subtype and (ii) SDIEM variants of these versions with two additional amino-acid modifications in the CH2 domain of the human Fc part to enhance its interaction with Fc-receptor-expressing effector cells.

As expected, the chimerized and SDIEM-modified antibodies showed no difference in binding avidity; however, their capability to mediate ADCC against FLT3-expressing target cells differed markedly: the SDIEM-modified versions of both antibodies were active in vitro at concentrations as low as 1 ng/ml, whereas the unmodified chimeric versions rapidly lost activity below 10 μg/ml. The modified 4G8 antibody was consistently superior to its BV10 counterpart with respect to target antigen binding and ADCC induction. Whether the close proximity to the membrane of the epitope recognized by the 4G8 antibody is —at least in part—responsible for its superior activity remains at present unclear. In any case, the remarkable ADCC activity of 4G8SDIEM may be properly ranged if (i) its activity against NALM-16 cells is compared with that of a SDIEM-optimized antibody directed to CD19 (4G7-SDIEM), an antigen expressed with a fourfold excess (6000 vs 25 000 molecules per cell) on this particular cell line. It should also be noted in this context that the optimized CD19 antibody is significantly more effective than the well-established CD20 antibody rituximab in mediating ADCC against Raji cells expressing comparable amounts of CD19 and CD20 (data not shown). In general, the expression levels of FLT3 on cultured cells as well as leukemic blasts as reported here (<5000 molecules per cell) are clearly lower than the levels reported for CD19 and CD20 antibodies (10 000–35 000 and 65 000–300 000 molecules per cell, respectively).32

When leukemic blasts isolated from the peripheral blood of patients rather than cultured NALM-16 cells were used as targets for allogeneic and autologous PBMCs, killing was less pronounced. We do not think that this is merely due to the lower expression of FLT3 on leukemic blasts but rather reflects, at least in part, a heightened resistance of these primary cells towards antibody-mediated killing. In addition, detection of ADCC against autologous AML blasts is hampered by an exceedingly high spontaneous death rate of these cells in long-term (>8 h) assays. In any case, therapeutic activity of ADCC-inducing antibodies may be limited to situations where favorable target:effector ratios prevail, such as in minimal residual disease.

One of our antibodies, 4G8, partly blocks binding of the physiological FLT3 ligand. However, it does not affect proliferation of cultured NALM-16 cells or AML blasts at concentrations up to 10 μg/ml. This may be due to incomplete blocking. Alternatively, FLT3 signaling may not be crucial for proliferation of the respective cells in vitro. The latter possibility is supported by a report that also highlights the importance of ADCC for the in vivo activity of FLT3 antibodies. It describes two humanized, but not Fc-optimized FLT3 antibodies originally developed by ImClone. Although both of these antibodies interfere with FLT3 signaling in primary AML blasts, only one of them, termed EB10, mediates ADCC in vitro and prolongs survival of SCID mice injected with leukemic cells.31 It is difficult, however, to compare the ADCC activity of EB10 with that of the antibodies described in our study as, for EB10, cytotoxicity data at concentrations lower than 10 μg/ml were not provided. Despite our impression that therapeutic effects of antibody-mediated FLT3 blockade may be limited, we have chosen 4G8SDIEM, the antibody partly blocking binding of the FLT3 ligand, for further characterization, mainly because of its superior binding and ADCC activity.

Antigen shift, that is, loss of surface expression due to internalization or shedding after antibody binding, is frequently observed upon application of therapeutic antibodies in vitro and in vivo. Internalization of antigen/antibody complexes may be desirable during therapy with immunotoxins. However, for strategies relying on antibody-mediated effector mechanisms such as complement activation or ADCC, antigen loss by whatever mechanism may be detrimental. In fact, a profound and long-lasting antigen shift after therapeutic application of saturating antibody doses may have been a major reason for the limited therapeutic efficiency of the CD33 antibody Lintuzumab in clinical trials.9 Very recently, it was reported that this phenomenon may also limit the therapeutic activity of the widely used CD20 antibody Rituximab.33 Although a ‘shaving’ mechanism on chronic lymphocytic leukemia cells has been postulated,34 CD20 has been described in several reports as relatively resistant towards shifting. However, in most cases, the respective in vitro assays have been terminated after several hours, which is possibly too early to detect delayed shifting.33 The antibody 4G8SDIEM described here clearly modulates the surface expression of FLT3 on NALM-16 cells and on leukemic blasts. However, this modulation, which is completed after 24 h, does not reduce expression levels below 50%. Thus, we do not expect this phenomenon to seriously hamper the therapeutic activity of the antibody, although it should certainly be monitored carefully during clinical application.

With respect to antibody toxicity, the specificity of the target antigen recognized is a central issue. FLT3 has been reported to be expressed on normal hematopoietic precursor cells as well as on DCs. In our experiments, FLT3 expression on DCs was hardly detectable (<300 molecules per cell) and was marginal on CD34+ bone marrow cells (<500 molecules per cell). The low expression of FLT3 on DCs is particularly surprising. A number of articles describe successful activation of these cells by recombinant FLT3 ligand resulting in increased antitumor effects mediated by the activated DCs.18, 22, 35 We are aware that our results are not easily reconciled with these findings and propose the hypothesis that activation of DCs via FLT3 requires only few molecules on the surface of the cells. In any case, given the low number of FLT3 molecules on normal DCs and hematopoietic precursor cells, the lack of cytotoxic activity mediated by 4G8SDIEM against such cells in 51[Cr]-release and colony-forming unit assays is not surprising. With respect to the potential toxicity of the 4G8SDIEM antibody these findings are confirmative. Of note, the expression of the interleukin-3 receptor, an alternative target on leukemic cells, on DCs, was found to be considerably higher than that of FLT3, which implies that the specificity of FLT3 for leukemic cells might be superior to that of CD123. In conclusion, we believe that the results discussed above provide a solid basis for the clinical evaluation of 4G8SDIEM.

Recently, we succeeded in the production of 4G8SDIEM in pharmaceutical quality and quantity in a university-owned, GMP-compliant production unit. At present we treat selected AML patients with this antibody on a compassionate need basis. In the first patient we observed NK cell activation and complete but transient clearance of leukemic blasts from the peripheral blood. A human anti-human response was not detected up to 3 months after completion of the therapy. We envisage to start a phase I study with 4G8SDIEM in the second half of 2012. Thus, it appears possible to develop innovative therapeutic antibodies within university institutions up to the stage of clinical pilot studies. We strongly feel that substantial participation of academic institutions will improve the speed and efficiency of antibody development. In any case, studies performed with 4G8SDIEM and other third-generation antibodies, designed for increased ADCC-activity, might soon enable us to judge more reliably the role of this activity for the therapeutic efficiency of antitumor antibodies in humans.

Accession codes




  1. 1

    Oflazoglu E, Audoly LP . Evolution of anti-CD20 monoclonal antibody therapeutics in oncology. MAbs 2010; 2: 14–19.

    Article  PubMed  PubMed Central  Google Scholar 

  2. 2

    Beck A, Wurch T, Bailly C, Corvaia N . Strategies and challenges for the next generation of therapeutic antibodies. Nat Rev Immunol 2010; 10: 345–352.

    CAS  Article  PubMed  Google Scholar 

  3. 3

    Shinkawa T, Nakamura K, Yamane N, Shoji-Hosaka E, Kanda Y, Sakurada M et al. The absence of fucose but not the presence of galactose or bisecting N-acetylglucosamine of human IgG1 complex-type oligosaccharides shows the critical role of enhancing antibody-dependent cellular cytotoxicity. J Biol Chem 2003; 278: 3466–3473.

    CAS  Article  Google Scholar 

  4. 4

    Lazar GA, Dang W, Karki S, Vafa O, Peng JS, Hyun L et al. Engineered antibody Fc variants with enhanced effector function. Proc Natl Acad Sci USA 2006; 103: 4005–4010.

    CAS  Article  Google Scholar 

  5. 5

    Robak T . GA-101, a third-generation, humanized and glyco-engineered anti-CD20 mAb for the treatment of B-cell lymphoid malignancies. Curr Opin Investig Drugs 2009; 10: 588–596.

    CAS  PubMed  Google Scholar 

  6. 6

    Awan FT, Lapalombella R, Trotta R, Butchar JP, Yu B, Benson Jr DM et al. CD19 targeting of chronic lymphocytic leukemia with a novel Fc-domain-engineered monoclonal antibody. Blood 2010; 115: 1204–1213.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  7. 7

    Horton HM, Bernett MJ, Pong E, Peipp M, Karki S, Chu SY et al. Potent in vitro and in vivo activity of an Fc-engineered anti-CD19 monoclonal antibody against lymphoma and leukemia. Cancer Res 2008; 68: 8049–8057.

    CAS  Article  Google Scholar 

  8. 8

    Foyil KV, Bartlett NL . Anti-CD30 antibodies for Hodgkin lymphoma. Curr Hematol Malig Rep 2010; 5: 140–147.

    Article  PubMed  Google Scholar 

  9. 9

    Feldman EJ, Brandwein J, Stone R, Kalaycio M, Moore J, O’Connor J et al. Phase III randomized multicenter study of a humanized anti-CD33 monoclonal antibody, lintuzumab, in combination with chemotherapy, versus chemotherapy alone in patients with refractory or first-relapsed acute myeloid leukemia. J Clin Oncol 2005; 23: 4110–4116.

    CAS  Article  Google Scholar 

  10. 10

    Roberts AW, He S, Ritchie D, Hertzberg MS, Kerridge I, Durrant ST et al. A phase I study of anti-CD123 monoclonal antibody (mAb) CSL360 targeting leukemia stem cells (LSC) in AML. J Clin Oncol (Meet Abstr) 2010; 28 (15_Suppl): e13012.

    Article  Google Scholar 

  11. 11

    Kikushige Y, Yoshimoto G, Miyamoto T, Iino T, Mori Y, Iwasaki H et al. Human Flt3 is expressed at the hematopoietic stem cell and the granulocyte/macrophage progenitor stages to maintain cell survival. J Immunol 2008; 180: 7358–7367.

    CAS  Article  PubMed  Google Scholar 

  12. 12

    Dahl C, Hoffmann HJ, Saito H, Schiotz PO . Human mast cells express receptors for IL-3, IL-5 and GM-CSF; a partial map of receptors on human mast cells cultured in vitro. Allergy 2004; 59: 1087–1096.

    CAS  Article  PubMed  Google Scholar 

  13. 13

    Toba K, Koike T, Shibata A, Hashimoto S, Takahashi M, Masuko M et al. Novel technique for the direct flow cytofluorometric analysis of human basophils in unseparated blood and bone marrow, and the characterization of phenotype and peroxidase of human basophils. Cytometry 1999; 35: 249–259.

    CAS  Article  PubMed  Google Scholar 

  14. 14

    Lim LH, Burdick MM, Hudson SA, Mustafa FB, Konstantopoulos K, Bochner BS . Stimulation of human endothelium with IL-3 induces selective basophil accumulation in vitro. J Immunol 2006; 176: 5346–5353.

    CAS  Article  PubMed  Google Scholar 

  15. 15

    Rappold I, Ziegler BL, Kohler I, Marchetto S, Rosnet O, Birnbaum D et al. Functional and phenotypic characterization of cord blood and bone marrow subsets expressing FLT3 (CD135) receptor tyrosine kinase. Blood 1997; 90: 111–125.

    CAS  PubMed  Google Scholar 

  16. 16

    Meeker TC, Miller RA, Link MP, Bindl J, Warnke R, Levy R . A unique human B lymphocyte antigen defined by a monoclonal antibody. Hybridoma 1984; 3: 305–320.

    CAS  Article  PubMed  Google Scholar 

  17. 17

    Weisberg E, Sattler M, Ray A, Griffin JD . Drug resistance in mutant FLT3-positive AML. Oncogene 2010; 29: 5120–5134.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  18. 18

    Dong J, McPherson CM, Stambrook PJ . Flt-3 ligand: a potent dendritic cell stimulator and novel antitumor agent. Cancer Biol Ther 2002; 1: 486–489.

    Article  PubMed  Google Scholar 

  19. 19

    McKenna HJ . Role of hematopoietic growth factors/flt3 ligand in expansion and regulation of dendritic cells. Curr Opin Hematol 2001; 8: 149–154.

    CAS  Article  PubMed  Google Scholar 

  20. 20

    Waskow C, Liu K, Darrasse-Jeze G, Guermonprez P, Ginhoux F, Merad M et al. The receptor tyrosine kinase Flt3 is required for dendritic cell development in peripheral lymphoid tissues. Nat Immunol 2008; 9: 676–683.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  21. 21

    Harada S, Kimura T, Fujiki H, Nakagawa H, Ueda Y, Itoh T et al. Flt3 ligand promotes myeloid dendritic cell differentiation of human hematopoietic progenitor cells: possible application for cancer immunotherapy. Int J Oncol 2007; 30: 1461–1468.

    CAS  PubMed  Google Scholar 

  22. 22

    Kingston D, Schmid MA, Onai N, Obata-Onai A, Baumjohann D, Manz MG . The concerted action of GM-CSF and Flt3-ligand on in vivo dendritic cell homeostasis. Blood 2009; 114: 835–843.

    CAS  Article  PubMed  Google Scholar 

  23. 23

    Herrmann T, Grosse-Hovest L, Otz T, Krammer PH, Rammensee HG, Jung G . Construction of optimized bispecific antibodies for selective activation of the death receptor CD95. Cancer Res 2008; 68: 1221–1227.

    CAS  Article  Google Scholar 

  24. 24

    Grosse-Hovest L, Hartlapp I, Marwan W, Brem G, Rammensee HG, Jung G . A recombinant bispecific single-chain antibody induces targeted, supra-agonistic CD28-stimulation and tumor cell killing. Eur J Immunol 2003; 33: 1334–1340.

    CAS  Article  PubMed  Google Scholar 

  25. 25

    Evan GI, Lewis GK, Ramsay G, Bishop JM . Isolation of monoclonal antibodies specific for human c-myc proto-oncogene product. Mol Cell Biol 1985; 5: 3610–3616.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  26. 26

    Jung G, Freimann U, Von MZ, Reisfeld RA, Wilmanns W . Target cell-induced T cell activation with bi- and trispecific antibody fragments. Eur J Immunol 1991; 21: 2431–2435.

    CAS  Article  PubMed  Google Scholar 

  27. 27

    Sutherland DR, Anderson L, Keeney M, Nayar R, Chin-Yee I . The ISHAGE guidelines for CD34+ cell determination by flow cytometry. International Society of Hematotherapy and Graft Engineering. J Hematother 1996; 5: 213–226.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  28. 28

    Otz T, Grosse-Hovest L, Hofmann M, Rammensee HG, Jung G . A bispecific single-chain antibody that mediates target cell-restricted, supra-agonistic CD28 stimulation and killing of lymphoma cells. Leukemia 2009; 23: 71–77.

    CAS  Article  PubMed  Google Scholar 

  29. 29

    Ullrich A, Schlessinger J . Signal transduction by receptors with tyrosine kinase activity. Cell 1990; 61: 203–212.

    CAS  Article  PubMed  Google Scholar 

  30. 30

    Sanz M, Burnett A, Lo-Coco F, Lowenberg B . FLT3 inhibition as a targeted therapy for acute myeloid leukemia. Curr Opin Oncol 2009; 21: 594–600.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  31. 31

    Piloto O, Levis M, Huso D, Li Y, Li H, Wang MN et al. Inhibitory anti-FLT3 antibodies are capable of mediating antibody-dependent cell-mediated cytotoxicity and reducing engraftment of acute myelogenous leukemia blasts in nonobese diabetic/severe combined immunodeficient mice. Cancer Res 2005; 65: 1514–1522.

    CAS  Article  Google Scholar 

  32. 32

    Ginaldi L, De Martinis M, Matutes E, Farahat N, Morilla R, Catovsky D . Levels of expression of CD19 and CD20 in chronic B cell leukaemias. J Clin Pathol 1998; 51: 364–369.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  33. 33

    Beers SA, French RR, Chan HT, Lim SH, Jarrett TC, Vidal RM et al. Antigenic modulation limits the efficacy of anti-CD20 antibodies: implications for antibody selection. Blood 2010; 115: 5191–5201.

    CAS  Article  Google Scholar 

  34. 34

    Beum PV, Kennedy AD, Williams ME, Lindorfer MA, Taylor RP . The shaving reaction: rituximab/CD20 complexes are removed from mantle cell lymphoma and chronic lymphocytic leukemia cells by THP-1 monocytes. J Immunol 2006; 176: 2600–2609.

    CAS  Article  PubMed  Google Scholar 

  35. 35

    Davis ID, Chen Q, Morris L, Quirk J, Stanley M, Tavarnesi ML et al. Blood dendritic cells generated with Flt3 ligand and CD40 ligand prime CD8+ T cells efficiently in cancer patients. J Immunother 2006; 29: 499–511.

    Article  PubMed  Google Scholar 

Download references


We thank Sabrina Grimm and Sandra Drescher for expert technical assistance. This work was supported by the GO-Bio programme of the Federal Ministry for Education and Research (BMBF), Germany, and in part by a grant from the Deutsche Forschungsgemeinschaft (SFB 685, project C10).

Author information



Corresponding author

Correspondence to G Jung.

Ethics declarations

Competing interests

LG-H, SA, H-GR and GJ are co-owners of a recently founded company, Synimmune GmbH. The company will develop antitumor antibodies optimized by gene technology. Its foundation is a prerequisite for further funding by the GO-Bio programme of the Federal Ministry for Education and Research (BMBF), Germany.

Additional information

Supplementary Information accompanies the paper on the Leukemia website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Hofmann, M., Große-Hovest, L., Nübling, T. et al. Generation, selection and preclinical characterization of an Fc-optimized FLT3 antibody for the treatment of myeloid leukemia. Leukemia 26, 1228–1237 (2012).

Download citation


  • FLT3
  • AML
  • optimized antibody
  • tumor immunotherapy
  • ADCC

Further reading


Quick links