Acute Leukemias

Costimulation improves the killing capability of T cells redirected to tumor cells expressing low levels of CD33: description of a novel modular targeting system


Owing to their clinical success, there is growing interest in novel bispecific antibodies (bsAbs) for retargeting of T cells to tumor cells including for the treatment of acute myeloid leukemia (AML). One potential target for retargeting of T cells to AML blasts is the surface molecule CD33. Here we describe a novel modular targeting platform that consists of a universal effector module (EM) and individual target modules (TMs). Both modules can form an immune complex via a peptide epitope. The resulting targeting complex can functionally replace a conventional bsAb. By fusion of a costimulatory domain (for example, the extracellular CD137 ligand domain) to the TM, the targeting complex can even provide a costimulatory signal to the redirected T cells at their side of interaction with the tumor cell. Furthermore, we observed that an efficient killing of tumor cells expressing low levels of the tumor target CD33 becomes critical at low effector-to-target cell ratios but can be improved by costimulation via CD137 using our novel targeting system.


Over the past two decades, an antigen-specific retargeting of T cells with bispecific antibodies (bsAbs) has emerged as a promising strategy in the fight against tumors.1, 2 Usually, T-cell-engaging bsAbs consist of the variable light (VL) and heavy chains (VH) of two monoclonal antibodies (mAbs) recognizing the CD3 complex on T cells and a tumor-associated surface antigen (TAA). According to the current opinion, the bsAb-mediated cross-link of T cells and tumor cells leads to a major histocompatibility complex (MHC), T cell receptor (TCR) and costimulation-independent activation of polyclonal T lymphocytes, which finally results in an efficient tumor cell lysis.3, 4, 5 The front runner in this field is the bsAb blinatumomab with a dual specificity for CD3 and CD19, which has been successfully tested in pilot clinical trials for the treatment of B-cell lymphomas.6, 7, 8, 9

Prerequisite for an efficient bsAb-mediated serial tumor cell killing with lowest risk of side effects are optimally balanced affinities of the bsAb towards CD3 on T cells and the TAA on the tumor cell. As modifications in one Ab domain can also alter the binding and functional properties of the second, unmodified Ab domain, each novel bsAb has to be optimized for the on/off rate of T-cell/target cell interaction.2, 10 In order to shorten the costly and protracted optimization process during bsAb development, we established an advanced T-cell retargeting complex that splits the dual functionality of conventional bsAbs to two modules: (i) an effector module (EM) with dual specificities for CD3 and a peptide epitope, and (ii) a target module (TM) that represents an epitope fusion protein with affinity to the target cell. Both components can form a complex being able to cross-link T cells and target cells comparable to a conventional bsAb. Using this novel modular system T cells can be easily redirected against any TAA simply by replacing the TM. As the EM can be maintained, there is no need for a repeated optimization of the critical anti-CD3 domain to achieve an optimal efficacy with lowest risk of potential side effects for each novel T-cell retargeting complex. Furthermore, additional functional domains can be fused to components of the modular system, for example, ligands for a costimulatory receptor of the effector cell.

For proof of concept, we here describe for the first time such T-cell-engaging modular systems including a fully humanized (hu) modular system for redirection of T cells to acute myeloid leukemia (AML) cells. AML is the most commonly diagnosed acute leukemia in adults. The high relapse and very low 3-year overall survival rate of AML patients after initial complete remission gained by standard induction chemotherapy11 explain the need of new adjuvant therapeutic strategies in addition to conventional treatment schemes. Examples for immunosurveillance of AML and susceptibility of leukemic cells to T-cell attack have been already described.12, 13, 14, 15, 16, 17, 18 As target antigen for redirection of T cells to AML cells, we selected CD33, which is a 67 kDa transmembrane cell-surface glycoprotein specific for the myeloid lineage and present on more than 80% of myeloid leukemias.19, 20 We compared the functionality and efficiency of the novel T-cell retargeting complex with a recently described fully hu conventional bsAb CD33-CD318 in vitro using CD33+ AML cell lines and patient-derived AML blasts, as well as in vivo using immunodeficient non-obese diabetic (NOD)/severe combined immunodeficiency (SCID) interleukin (IL)2Rγ−/− mice. Although both targeting strategies comparably well eradicated AML cells expressing high levels of CD33 even at low picomolar concentrations, the elimination of CD33low AML cells becomes critical at low effector-to-target cell (e:t) ratios, but can significantly be improved when using a modified modular system that was able to simultaneously provide activation signals via CD3 and costimulatory signals to redirected T cells.

Materials and methods

Isolation of human PBMCs and T-cell subpopulations

The isolation and cultivation of human PBMCs, T-cell subpopulations and the generation of pre-activated PBMCs were performed as described previously.21 Patient-derived AML blasts and autologous T cells were obtained from leukapheresis products from patients with hyperleukocytosis with informed consent and approval by the local institutional review board. After isolation of MNCs by gradient centrifugation over Biocoll (Biochrom, Berlin, Germany), cells were stained with anti-CD3/APC-eFluor780 (eBioscience, Inc., San Diego, CA USA), anti-CD19/APC (BD Biosciences, Heidelberg, Germany) and anti-CD45/AlexaFluor700 (BioLegend, Uithoorn, The Netherlands) mAbs. Living cells were distinguished from dying cells and cellular debris by size exclusion and being 4,6-diamidino-2-phenylindole negative (Sigma-Aldrich, Steinheim, Germany). CD45+CD3+ T cells and CD45+CD3CD19 (defined as AML blasts) cells were sorted by using a FACS-ARIA II cell sorter (BD Biosciences).

Cell lines

MOLM-13, MV4-11 and CHO cells were cultured in complete RPMI 1640 medium.10 OCI-AML3, HEK293T-CD33 and 3T3 cells were kept in complete DMEM medium.10

Construction of recombinant Abs

The generation and humanization of murine anti-CD3(MT-301), anti-CD33(DRB2) or anti-La(5B9) single-chain variable fragments (scFvs) were performed as described previously.10, 21, 22, 23 The structures of recombinant Abs are shown in Supplementary Figure 1a and Supplementary Figure 6. For permanent production, all recombinant Ab constructs were cloned into the lentiviral vector p6NST50 used for transduction of 3T3 or CHO cells as described previously.21

Expression, purification and binding analysis of recombinant Abs

6xhistidine (his)-tagged recombinant Abs were expressed by 3T3 or CHO cells and purified from cell culture supernatants using Ni-NTA affinity chromatography as described previously.10 Protein and binding analysis of the purified Abs was performed using an anti-myc/FITC mAb (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) and a FACSCalibur flow cytometer (BD Biosciences).10, 21 As control mAbs, anti-CD3/PE (Miltenyi Biotec GmbH), anti-CD33/PE (BD Biosciences) and anti-La(5B9)23 were used. For detection of unlabeled murine anti-La(5B9), the PE-conjugated Goat F(ab́)2 anti-Mouse IgG (Fcgamma) Ab (Immunotech, Marseille, France) was used.

Stability of recombinant Abs

In order to test Ab stability, recombinant proteins were incubated at 37 °C for 24 h in complete RPMI 1640 medium supplemented with 10% human serum (HS) (RPMIHS). Subsequent T-cell activation assays were performed with either pre-incubated Abs or Abs stored at 4 °C in PBS. Complete RPMI 1640 or RPMIHS medium served as coculture medium.

Cytotoxicity assay

Cytotoxicity was examined by standard chromium release assay or flow cytometry-based assays using the MACSQuant Analyzer (Miltenyi Biotec GmbH) as recently established.10, 22 To distinguish effector from target cells in flow cytometry-based assays, CD33+ tumor cells were labeled with either 1 μM 5(6)-carboxyfluorescein diacetate N-succinimidyl ester (CSFE) (Sigma-Aldrich) or 5 μM proliferation dye eFluor670 (eBioscience). Statistical analysis was performed by one-way analysis of variance (ANOVA) with the Bonferroni Multiple Comparison test using GraphPad Prism Software (La Jolla, CA, USA) (***P<0.001, **P<0.01, *P<0.05).

T-cell activation, degranulation, proliferation assays and cytokine enzyme-linked immunosorbent assay

To determine the activation state of T cells, the expression of the activation markers CD69 and CD25, the release of tumor necrosis factor (TNF) and interferon (IFN)-γ as well as the upregulation of CD107a as a typical degranulation marker were analyzed. Untouched pan T cells, CD4+ or CD8+ T cells were cocultured with fluorescently labeled MOLM-13 cells in the presence or absence of 5 pmol/ml recombinant Ab at 37 °C. After indicated time points, supernatant of each sample was collected and the cells of one triplet were pooled, washed and stained with a mixture of anti-CD25/PE (BD Biosciences), anti-CD69/FITC, anti-CD3/VioBlue and anti-CD4/PerCP (Miltenyi Biotec GmbH). The concentration of TNF and IFN-γ in the cell culture supernatants was determined using enzyme-linked immunosorbent assay (ELISA).21 T-cell degranulation and proliferation assays were performed as described previously.21, 23, 24

PBMC cocultivation assay

3 × 105 PBMCs were incubated with or without 30 pmol/ml of each recombinant Ab component. After 24 h PBMCs were stained with anti-CD3/PECy7, anti-CD25/PE and anti-CD69/FITC mAbs (Miltenyi Biotech GmbH) to analyze T-cell activation. Concentration of TNF and IFN-γ in cell culture supernatants was analyzed using ELISA.21 In order to investigate killing of myeloid cells, PBMCs of one triplet were pooled and stained with anti-CD3/PECy7, anti-CD13/FITC, anti-CD33/PE and anti-CD45/VioBlue mAbs (Miltenyi Biotech GmbH). Living cells were identified by being propidium iodide negative. CD13+CD33+ or CD3+ cells within the CD45+ subpopulation were considered to be myeloid or T cells, respectively.

Transplantation of human AML blasts into NOD/SCID IL2Rγ−/− mice

Eight- to ten-week-old NOD/SCID IL2Rγ−/− mice were intravenously injected with 1 × 105 MOLM-13 cells. Beforehand, the AML cells were cultured alone or in the presence of 5 × 105 T cells for 6 h at 37 °C. In addition, effector or control Ab constructs were added at a concentration of 0.5 pmol/ml of each construct. Mice were killed when visible tumors developed at injection site and single-cell suspensions from bone marrow obtained from femur and tibia of the left hind leg were prepared. Erythrocytes were removed by lysis and nucleated cells were stained with anti-mouse CD45.1/PECy7 (eBioscience), anti-human CD3/APC-eFluor780 (eBioscience), CD19/APC (BD Bioscience), CD33/PE (eBioscience) and CD45/AlexaFluor700 (BioLegend) mAbs. Doublet discrimination was routinely carried out and dead cells were excluded by 4,6-diamidino-2-phenylindole staining (Sigma-Aldrich). All measurements were performed on a BD LSRII (BD Biosciences). Data were analyzed using FlowJo software (Tree Star, Ashland, OR, USA). The animals were kept under standardized environmental conditions. All experiments were performed according to the German animal protection law with permission from the responsible local authorities (Sächsische Landesdirektion). To determine statistical differences in tumor engraftment, data were analyzed by the Kruskal–Wallis test and post hoc Dunn’s Multiple Comparison test using GraphPad Prism software.


Development and characterization of a novel modular cell targeting system

Conventional bsAbs directly cross-link effector and target cells by binding to CD3 on T cells and TAA on cancer cells (Figure 1a, left panel). The key idea leading to the flexibility of the novel targeting system is the replacement of a bsAb by two Ab components: (i) a universal EM and (ii) exchangeable TMs (Figure 1a, right panel). The EM is a functionally optimized bsAb that binds to CD3 and a 10-amino-acid-long peptide epitope (E5B9).18, 21, 23, 25 As TM, we first exemplarily established a CD33-directed scFv-E5B9 fusion protein (hu scFv CD33-E5B9) for targeting AML blasts. The design of the fully hu EM and TM is schematically compared with the recently described conventional anti-CD3-anti-CD33 bsAb18 in Supplementary Figure 1a. Recombinant Ab constructs were purified as full-length proteins with the expected size of 60 or 40 kDa (Supplementary Figure 1b). Their monomeric character was confirmed using size-exclusion gel filtration chromatography (data not shown). Capability of antigen binding was characterized by flow cytometry analysis as shown in Supplementary Figure 1c. As important prerequisite for functionality of the modular system, we could demonstrate that the epitope E5B9 of the TM is accessible after its binding to CD33+ cells (Supplementary Figure 1ciii, right panel).

Figure 1

Efficient lysis of CD33+ tumor cells by redirection of T cells via the novel modular retargeting system. (a, left panel) A conventional single-chain bsAb is able to cross-link T cells to CD33+ AML blasts. (a, right panel) Alternatively, T cells can be redirected to tumor cells by the novel modular system that is based on the complex of two recombinant Ab derivatives. The first component is an exchangeable target module (TM) composed of an anti-CD33 scFv tagged with a short peptide epitope (E5B9). The second component (universal effector module (EM)) is a bsAb with one binding arm for CD3 on T cells and another binding arm recognizing the peptide epitope E5B9 of the scFv-based TM. (b, c) In standard chromium release assays, freshly isolated T cells were incubated with 51Cr-labeled CD33+ MOLM-13 cells (b), MV4-11 (c) or OCI-AML3 (c) at effector-to-target (e:t) cell ratios of 1:1, 5:1 (b) and 10:1 (b, c) for 48 h either in the absence or in the presence of 30 pmol/ml of indicated Ab components. Average and s.d. of three or four independent donors are shown, respectively. Statistical analysis was performed by one-way ANOVA with the Bonferroni Multiple Comparison test (***P<0.001 and **P<0.01 with respect to controls with no Ab, TM or EM). (d) In a flow cytometry-based cytotoxicity assay, fluorescently labeled MOLM-13 cells and freshly isolated T cells were incubated at an e:t ratio of 5:1 for 20 h. Ab modules were either applied in equimolar concentrations to the coculture (checked bars) or the EM (open bars) or the TM (dashed bars) was used at an Ab-concentration of 5 pmol/ml, whereas the respective counterpart module was applied at increasing concentrations (as indicated on the x axis). The respective number of surviving cells of a control assay performed in the absence of T cells and Ab (not included in the figure) was set to 100%, and the corresponding relative amounts of surviving cells was calculated for the other samples. Summarized data of three individual donors are shown.

Killing efficacy of the novel modular system in vitro

In a first step, we wanted to learn whether the CD33-directed modular system can functionally replace a bsAb. To compare the efficacy of the novel T-cell retargeting strategy with the conventional bsAb CD33-CD3 in vitro, different naturally CD33+ AML cell lines were incubated with T cells at varying e:t ratios. The modular system mediates a significant tumor cell lysis at e:t ratios between 1:1 and 10:1 (Figure 1b). As shown in Figure 1c, the novel CD33-directed T-cell retargeting complex also triggered efficient killing of other CD33+ AML cell lines (OCI-AML3 and MV4-11), although CD33 is less densely expressed on these cells.18 Incubation of T cells and tumor cells only with one module of the modular system does not result in tumor cell eradication. Comparative analysis indicates that tumor cell elimination by the modular system is as efficient as in the presence of the bsAb CD33-CD3. Moreover, we could demonstrate that the modular system efficiently redirects T cells to lyse CD33+ MOLM-13 cells even at picomolar Ab concentrations (Figure 1d).

For the in vitro assays described so far, equimolar concentrations of the TM and EM were utilized. To investigate whether functionality of the modular system is affected by alterations of the EM:TM ratio, T cells were redirected to MOLM-13 cells with constant concentrations (5 pmol/ml) of either the EM or the TM, whereas the concentration of the respective counterpart module was subsequently decreased to 0.025 pmol/ml (Figure 1d). Comparative analysis using equimolar Ab concentrations revealed that maintaining the TM at a constant high concentration leads to an improvement of functionality up to fivefold. According to these data, the killing efficacy of the novel retargeting complex is within the range of previously published conventional bsAbs CD33-CD3.16, 17, 18

Time kinetics of T-cell activation and cytotoxicity

To further characterize tumor cell killing mediated by the modular system in vitro, we analyzed T-cell activation and tumor lysis over a time period of 3–60 h in comparison to the conventional bsAb CD33-CD3. As revealed by CD69 and CD25 expression profiles, the novel CD33-specific retargeting complex mediated a rapid and efficient activation of engaged CD4+ and CD8+ T cells (Figure 2). Neither the TM nor the EM alone was able to mediate T-cell activation. Moreover, the observed activation profile is comparable to that of the bsAb CD33-CD3. In parallel, the secretion of pro-inflammatory cytokines as a characteristic feature of activated effector T cells was investigated. As shown in Figure 3a, upon Ab-mediated cross-linkage T cells secrete significant amounts of TNF and IFN-γ. No or only marginal cytokine amounts were detectable when T cells and tumor cells were cultured alone or with only one component of the modular system. As analyzed exemplarily after 20 h, both CD4+ and CD8+ T cells secreted TNF and IFN-γ after Ab-mediated redirection to CD33+ tumor cells (Figure 3b). Although total amounts of measured cytokines and respective contributions of T-cell subpopulations varied depending on the analyzed donor, the conventional bsAb CD33-CD3 always triggered a higher cytokine release compared with the modular system. As CD4+ T lymphocytes are not typical killer cells but were activated upon the Ab-mediated cross-linkage, we wanted to analyze whether this T-cell subpopulation is able to exhibit redirected tumor lysis. As shown by initial experiments with pan T cells, both retargeting systems mediate an efficient tumor cell killing with similar kinetics (Figure 4a). A decrease of CD33+ MOLM-13 cells was first detectable after 6 h of cocultivation. Within 40 h, nearly all target cells were eradicated. Additional cytotoxicity assays with isolated subpopulations of either CD8+ or CD4+ T cells revealed that after 6 h, mainly CD8+ T cells trigger a significant tumor cell lysis in the presence of cross-linking recombinant Abs (Figure 4b). However, after 20 h, both T-cell subpopulations exhibited similar killing capacities. These results are in accordance with the upregulation of the degranulation marker CD107a on Ab-redirected T cells. After 6 h, especially CD8+ T lymphocytes upregulate CD107a. Within 20 h, the degranulation marker was detectable on CD4+ and CD8+ T lymphocytes upon Ab-mediated cross-linkage with CD33+ tumor cells.

Figure 2

Specific activation of T cells upon cross-linkage with CD33+ MOLM-13 cells by the modular system. Freshly isolated T cells were incubated with eFluor670 proliferation dye-labeled CD33+ MOLM-13 cells at an e:t ratio of 5:1 in the absence or presence of 5 pmol/ml of one or both components of the modular system, as well as in the presence of reference bsAb CD33-CD3. At indicated time points, T cells were stained with fluorescently labeled anti-CD3, anti-CD4, anti-CD25 and anti-CD69 mAbs. The average percentage of CD25+ (lower panel) or CD69+ (upper panel) T cells (left), CD4+ T cells (middle) or CD8+ T cells (right) of three individual donors are summarized in depicted diagrams (TM, target module; EM, effector module).

Figure 3

Cytokine production profile of Ab-redirected T cells. (a) Non-activated T cells were incubated with eFluor670 proliferation dye-labeled CD33+ MOLM-13 cells at an e:t ratio of 5:1 in the absence or presence of 5 pmol/ml of one or both components of the modular system, as well as in the presence of bsAb CD33-CD3. At different time points, secreted amounts of TNF and IFN-γ were assessed by ELISA. (b) In addition, freshly isolated CD4+ or CD8+ T cells were used as effector cells in a 20-h cocultivation. (x, not detectable) (a, b) Average cytokine concentrations and s.d. of triplets for three individual donors are shown (TM, target module; EM, effector module).

Figure 4

Efficient lysis of CD33+ tumor cells by redirected CD4+ and CD8+ T cells. (a) Cocultivation of eFluor670 proliferation dye-labeled CD33+ MOLM-13 cells with freshly isolated T cells (e:t ratio of 5:1) occurred either in the absence or in the presence of 5 pmol/ml of indicated Ab components. The percentage of surviving tumor cells was analyzed by flow cytometry at various time points. (b) In addition, fluorescently labeled MOLM-13 cells were cultured with CD4+ or CD8+ T cells for 6 (b, left panel) and 20 h (b, right panel). The respective number of surviving cells of a control assay performed in the absence of T cells and Ab (not included in the figure) was set to 100% and the corresponding relative amounts of surviving cells was calculated for the other samples. (c) Degranulation of CD4+ or CD8+ T cells was analyzed in the presence of anti-CD107a/PE mAb. After cocultivation of 6 or 20 h, T cells were counterstained with FITC-, VioBlue-, APC- or PECy5-conjugated anti-CD3, anti-CD4 and anti-CD8 mAbs. Percentages of CD4+CD107a+ or CD8+CD107a+ T cells are depicted in diagrams. (TM, target module; EM, effector module). (ac) Data are represented as mean±s.d. of three, four or five individual donors, respectively (***P<0.001, with respect to control with no Ab; one-way ANOVA with the Bonferroni Multiple Comparison test).

Bearing in mind that the modular system results in a lower cytokine release, the killing by redirected T cells might be mainly mediated via the granzyme/perforin pathway, which is in a good agreement with recently published data obtained for a conventional anti-CD3-anti-PSCA bsAb.21

Redirected lysis of AML blasts by autologous T cells

To investigate the capability of the modular system to eradicate patient-derived AML blasts by redirection of autologous T cells, a flow cytometry-based cytotoxicity assay was performed. As shown in Figure 5a, in the presence of the modular system, T cells were specifically activated, whereas no upregulation of CD25 in the presence of the EM or in the absence of any cross-linking Ab was observed after 48 h. Twenty-four hours after cross-linkage of T cells and AML blasts by the modular system, 70% of tumor cells were already eliminated irrespective of the chosen e:t ratio (Figure 5a, lower panel). Within 48 h, the number of surviving cells further decreased to 20%. Thus, the modular system is able to efficiently activate patient-derived T cells for lysis of autologous AML blasts.

Figure 5

Redirection of T cells by the modular system leads to an efficient antitumor effect against patient-derived AML blasts in vitro and CD33+ MOLM-13 in vivo. (a) CFSE-labeled, patient-derived CD33+ AML blasts were incubated with autologous T cells (e:t ratio of 10:1) in the absence or presence of 5 pmol/ml of each recombinant Ab component. (a, upper panel) After 48 h of incubation, the activation status of redirected T cells was examined by cell-surface staining of T lymphocytes with anti-CD3/VioBlue and anti-CD25/PE mAbs. Denoted values represent percentage of CD3+CD25+ T cells. (a, lower panel) After 24 and 48 h, the percentage of surviving CFSE+ AML blasts was determined by flow cytometry analysis. The number of CFSE+ tumor cells in the sample with T lymphocytes alone was equalized to 100% and the relative percentage of surviving cells was calculated. The average of surviving cells and the s.d. of triplets are shown for one representative donor out of three. (b) In vivo antitumor effect of the modular system was analyzed in NOD/SCID IL2Rγ−/− mice. Before retrobulbar injections, freshly isolated T cells and CD33+ MOLM-13 AML blasts were incubated at a ratio of 5:1 for 6 h either in the presence or in the absence of 0.5 pmol/ml Ab component. Both macroscopically visible eye tumors (b, left panel) and percentage of huCD45+CD33+ cells in murine bone marrow (b, right panel) was monitored after 2.5 weeks (*P<0.05, **P<0.01; the Kruskal–Wallis test and post hoc Dunn’s Multiple Comparison test). Experiment was performed in parallel with previously presented work on the conventional bsAb CD33-CD318 (TM, target module; EM, effector module).

Functionality of both retargeting strategies in vivo

In vivo functionality of the modular system was investigated in immunodeficient NOD/SCID IL2Rγ−/− mice. After pre-incubation, mixtures of MOLM-13 cells, T cells and recombinant Abs were administered intravenously according to respective control or treatment groups. Regular monitoring of all animals revealed that after 2.5 weeks, a large proportion of non-treated control mice developed tumors at injection site, whereas such tumors were not visible in mice treated with the modular system or the bsAb CD33-CD3 (Figure 5b, left panel). Thus, the experiment was terminated due to ethical reasons. Final analysis of bone marrow chimerism revealed that numbers of huCD45+CD33+ cells in bone marrow of treated mice were significantly lower in comparison with control animals receiving tumor and T cells alone (Figure 5b, right panel). These results suggest that the novel modular retargeting complex can also mediate a significant antitumor effect in vivo. The efficacy is in the same range as for the recently described fully hu bsAb CD33-CD3.18

Enhancement of T-cell-mediated anti-leukemia response by a costimulatory modular system

The idea to split a conventional bsAb in an EM and a TM also allowed us to experimentally test whether an additional costimulus would affect the tumor killing abilities of T cells redirected to AML cells expressing different levels of CD33.18 Therefore, a costimulatory TM that was C-terminally equipped with the extracellular domain of 4-1BBL (CD137L) was generated (Figure 6a). In combination with the common EM, it facilitates the formation of a T-cell retargeting complex that not only provides a T-cell activation signal by CD3 cross-linkage but also a costimulatory signal via 4-1BB/4-1BBL interaction (Figure 6b). To investigate functionality and potential enhanced antitumor effects of the costimulatory modular system, T-cell activation and proliferation assays were performed with immunoligand-negative HEK293T-CD33 cells. As shown by the upregulation of CD69 and CD25, both CD4+ and CD8+ T cells can be efficiently activated after cross-linkage with CD33+ tumor cells by the costimulatory modular system (Figure 7a). Although T-cell activation was comparable to the conventional CD3-enganging modular system, remarkable differences concerning proliferation and cytokine secretion of redirected T cells were observed (Figures 7b and c). Addition of the CD137 costimulatory signal enhances proliferation of both T-cell subpopulations (Figure 7bi) and causes on average the doubling of total T-cell numbers (Figure 7bii). In addition, it enhances the release of IFN-γ and TNF by a factor of 2 or 4, respectively (Figure 7c). But most importantly, the costimulatory modular system significantly ameliorates the killing of CD33low AML cells especially at low e:t ratios (Figure 8).

Figure 6

Development of a costimulatory modular system. (a) The ECD of 4-1BBL (CD137L) (aa 41 to 253) was amplified from human PBMCs and fused in frame at the C terminus of the CD33-specific TM in between the E5B9 (E) and myc/his-Tag to generate the costimulatory TM hu single-chain variable fragment immunoligand tandem (scFvILTa) construct CD33-E5B9-4-1BBL. (b) The costimulatory modular system redirects T cells to CD33+ tumor cells by the complex of the common EM and a costimulatory CD33-specific TM, providing a costimulatory signal.

Figure 7

Providing the costimulatory signal 4-1BBL via the modular system ameliorates T-cell proliferation, expansion and cytokine secretion. Freshly isolated T cells and HEK293T-CD33 cells were incubated at an e:t ratio of 5:1 with or without 15 pmol/ml recombinant Ab. (a) After 48 h, CD69 (ai) and CD25 (aii) expression levels of CD4+- or CD8+-gated T cells are determined. Data are presented as mean±s.d. of five independent donors (***P<0.001, **P<0.01; one-way ANOVA with the Bonferroni Multiple Comparison test). (b) T lymphocytes were stained with eFluor670 proliferation dye. After 6 days of cocultivation with tumor cells, T cell counts and dilution of proliferation dye was investigated. (bi) Percentages of eFluor670dim CD4+ or CD8+ T cells are depicted in histograms. Data of one representative donor are shown. (bii) Expansion of T cells was determined using total cell numbers. Expansion rates were normalized to the modular system (EM+TM). Data are represented as mean±s.d. of five individual donors. (c) Secretion of pro-inflammatory cytokines by Ab-redirected T cells was determined in coculture supernatants after 48 h. Cytokine concentrations measured in the presence of the modular system providing the costimulatory signal 4-1BBL (EM+TM-4-1BBL) were set to 100%. Normalized IFN-γ and TNF cytokine levels of five individual donors are shown (TM, target module; EM, effector module).

Figure 8

Providing the costimulatory signal 4-1BBL by the modular system improves Ab-mediated killing of CD33low AML cells at low e:t ratios. In a flow cytometry-based cytotoxicity assay, freshly isolated T cells and eFluor670 proliferation dye-labeled CD33+ tumor cells were incubated at an e:t ratio of 1:1 in the absence or presence of 15 pmol/ml recombinant Ab for 48 or 96 h. Average numbers of surviving cells of four independent donors are shown (***P<0.001, **P<0.01 and *P<0.05 with respect to control with EM alone; one-way ANOVA, the Bonferroni Multiple Comparison test) (TM, target module; EM, effector module).

Summing up, analyses of T-cell activation and T-cell-mediated tumor cell killing demonstrate that the modular system works as well as the conventional bsAb CD33-CD3, whereas the risk of cytokine storms may be less for the modular system. Moreover, costimulation improves the killing of CD33low cells by redirected T cells.


T-cell-engaging bsAbs are emerging as the next generation of selective targeted therapeutics. Their efficiency has been shown by several in vitro, in vivo and pilot clinical studies, demonstrating that both solid tumors and hematological malignancies can be efficiently treated with bsAb-redirected T cells.3, 4, 5, 6, 7, 8, 9, 10, 16, 17, 18, 21, 26, 27, 28, 29, 30 Unfortunately, the development of novel bsAbs is cost- and time-intensive as individual optimization steps are required to obtain molecules with optimal reactivity and lowest risk of side effects. In order to accelerate the bsAb optimization process, we established a novel, more flexible targeting system. The modular system splits conventional direct cross-linking bsAbs in two components, a universal EM and a TAA-specific TM, that form a protein complex via a peptide epitope. As a consequence, the interaction of the peptide epitope in the TM with the anti-peptide binding domain of the EM predominantly determines the on/off rate of the T-cell/target cell interaction and, thereby, the serial killing capability of the T-cell retargeting complex. Therefore, the optimized EM described here can be combined with any future TM, and there is no need to repeat the optimization of its anti-CD3 domain for either efficiency or lowest risk of side effects.

For proof of concept, we therefore established such a modular system and compared its properties with a recently described conventional hu bsAb anti-CD33-anti-CD3.18 In general, the novel modular system is able to redirect T cells comparably well as the bsAb CD33-CD3. As seen for conventional bsAbs,21 both CD8+ and CD4+ T cells efficiently eliminated target cells upon cross-linkage by the modular system, also the cytotoxic reaction of CD4+ T cells was delayed in time. Most importantly, only the complex of both components, but neither the TM nor the EM alone induced T-cell activation and redirected tumor cell lysis. Thus, the risk of an unexpected off-target activation and potentially fatal systemic cytokine release induced by the EM alone should be negligible for any future alternative TM/EM complex. Interestingly, activation of T cells mediated by the modular targeting system resulted in a lower cytokine release than conventional bsAb-mediated cross-linkage without loss of killing efficacy. This is in line with previous observations that killing by redirected T cells is mediated via the granzyme/perforin pathway.21 For functionality, a pre-incubation of the two modules to form a functional targeting complex is not required. Moreover, recombinant Abs are stable in HS at 37 °C, indicating that neither Ab stability nor functionality should be negatively influenced within the scope of clinical applications (Supplementary Figure 2). Interestingly, the modular system was further able to activate T cells from patients against autologous AML blasts. This finding is of special interest, as AML blasts are known to be immunosuppressive and can induce apoptosis in T cells via the PD-1/B7-H1 pathway.31 Furthermore, redirected T cells were able to induce lysis of autologous AML blasts. Based on this observation, one can speculate that the administration of CD33-specific bsAb-based systems in AML patients can override immunosuppressive mechanism mediated by AML blasts against T cells. The locally released amount of cytokines are therefore sufficient to turn the immunosuppressive into an inflammatory environment through the redirected T cells, including the reduced levels induced by the modular system, and could perhaps even help to reactivate AML-specific T-cell clones present in the peripheral blood of AML patients.14, 32 In the long run, this could lead to a control or even clearance of residual leukemic blasts. The functionality of the modular system was finally confirmed in a MRD mouse model in which a significant antitumor effect was observed. Relating to our in vivo data, efficiency of the modular system was again comparable to the conventional bsAb CD33-CD3.18 To this end, these findings as well as published data demonstrate that Ab-mediated cross-linkage of redirected T cells via the CD3 complex is sufficient to induce profound tumor lysis both in vitro and in vivo.3, 4, 5, 6, 7, 8, 9 Otherwise, clinical data from immunotherapeutical trials suggest that additional costimulatory signals are required to induce a long-lasting, durable antitumor response in patients. In particular, members of the tumor necrosis receptor family such as CD27, CD134 (Ox40) and CD137 (4-1BB) have an important role in shaping T-cell immune response. The modular organization of our T-cell-engaging system allows the generation of a target-specific immunoligand, which simultaneously triggers CD3 and CD137. As it is well known that AML blasts show varying CD33 expression levels and killing efficiency of bsAb-redirected T cells depends on the TAA expression density, we asked the question whether additional T-cell costimulation can positively affect tumor cell elimination. Using the model cell lines MOLM-13, MV4-11 and OCI-AML3 that express high, medium and low levels of CD33 similar to AML blasts, we were able to show that an additional costimulatory signal can indeed improve the killing of CD33low AML cells at low e:t ratios. Although part of the T cells, notably of the CD8+ population, start to proliferate in response to single CD3 triggering by the modular system, profound proliferation of both CD4+ and CD8+ T cells is only induced in response to combined stimulation of CD3 and CD137 by the immunoligand. It has been described that activation of the CD137 signaling pathway leads to increased expression of anti-apoptotic factors such as Bcl-XL. Prevention of activation-induced cell death by the CD137 pathway-dependent factors such as Bcl-XL could explain, why an increase in cell numbers over time is only observed in presence of the CD33-specific immunoligand. Our findings are in line with data recently published by Horing et al.,33 who showed an enhanced antitumor response by providing a CD137 signal in trans using a second recombinant Ab directed against a non-tumor-specific cellular target. In contrast to this approach, the modular system has the unique advantage to simultaneously and locally activate T cells by CD3, and trigger the costimulatory CD137 molecule upon cross-linkage of the T cell to its target cell by application of one single TM.

According to recent data published by Aigner et al.,17 T cells redirected to the surface molecule CD33 can also eliminate healthy CD33+ myeloid cells, for example, CD14+ monocytes. In line with these studies, we also observed lysis of healthy CD33+ cells by redirected T cells (Supplementary Figure 3). Nonetheless, we recently showed that bsAb-redirected T cells do not affect the hematopoietic potential of human hematopoietic stem cells.18 Therefore, we expect that the myeloid compartment will be fully restored after stop of the treatment.

Overall, the conceptual split of bsAbs in two separate molecules increases flexibility of the Ab-based T-cell retargeting strategy. As one unique feature, bifunctional TMs providing costimulatory signals such as CD137 can be applied within the modular system to locally enhance antitumor activities, prolong immune responses and improve killing properties. Moreover, in context of the novel modular system, a combined application of several TMs with specificities for different TAAs on the same tumor might be conceivable. In order to support the postulated flexibility of the modular system aside of the anti-CD33 TM, we have already developed additional functional TMs, for example, directed to the prostate stem cell antigen, a promising immunotarget for the treatment of prostate cancer,34 and used in combination with the EM for redirecting of T cells to prostate cancer cells (Supplementary Figure 4). Moreover, TMs can be constructed, for example, as bsAbs directed either to the same or different antigens. The resulting multispecific targeting complexes may be useful to reduce the risk of escape variants or to improve the target specificity as recently shown for combined usage of chimeric antigen receptors.35 First, preliminary data indeed support this concept (Bachmann, unpublished data).

Taken together, our in vitro and in vivo data underline the high potential of the novel modular targeting system presented here as powerful tool for redirecting of T cells to tumor cells in general and particularly for an antigen-specific immunotherapy of AML patients. It helps to overcome limitations of bsAbs but maintains at the same time their outstanding advantages:3, 4, 5, 6, 7, 8, 9, 10, 16, 17, 18, 21, 24, 25, 26, 27, 28, 29, 30, 36, 37, 38, 39 (1) high selectivity for tumor and ignorance of healthy TAA-negative tissues; (2) polyclonal stimulation of CD4+ and CD8+ T cells; (3) circumvention of classical tumor evasion mechanisms due to TCR- and MHC-independent mode of action, and flexible targeting of various TAAs; (4) minimal risk for unspecific T-cell activation because of monovalent CD3 binding; (5) easy production with eukaryotic expression system; (6) usage of small Ab fragments reduces unspecific cytotoxic side effects, facilitates tumor penetration and decreases their serum half-life; and (7) providing costimulatory signals, if requested. In particular, the novel modular system is characterized by its high efficacy, flexibility, specificity and potential wide range of application.


  1. 1

    Müller D, Kontermann RE . Bispecific antibodies for cancer immunotherapy: current perspectives. BioDrugs 2010; 24: 89–98.

  2. 2

    Stamova S, Koristka S, Keil J, Arndt C, Feldmann A, Michalk I et al. Cancer immunotherapy by retargeting of immune effector cells via recombinant bispecific antibody constructs. Antibodies 2012; 1: 172–198.

  3. 3

    Hoffmann P, Hofmeister R, Brischwein K, Brandl C, Crommer S, Bargou R et al. Serial killing of tumor cells by cytotoxic T cells redirected with a CD19-/CD3-bispecific single-chain antibody construct. Int J Cancer 2005; 115: 98–104.

  4. 4

    Offner S, Hofmeister R, Romaniuk A, Kufer P, Baeuerle PA . Induction of regular cytolytic T cell synapses by bispecific single-chain antibody constructs on MHC class I-negative tumor cells. Mol Immunol 2006; 43: 763–771.

  5. 5

    Stamova S, Feldmann A, Cartellieri M, Arndt C, Koristka S, Apel F et al. Generation of single-chain bispecific green fluorescent protein fusion antibodies for imaging of antibody-induced T cell synapses. Anal Biochem 2012; 423: 261–268.

  6. 6

    Bargou R, Leo E, Zugmaier G, Klinger M, Goebeler M, Knop S et al. Tumor regression in cancer patients by very low doses of a T cell-engaging antibody. Science 2008; 321: 974–977.

  7. 7

    Nagorsen D, Baeuerle PA . Immunomodulatory therapy of cancer with T cell-engaging BiTE antibody blinatumomab. Exp Cell Res 2011; 317: 1255–1260.

  8. 8

    Topp MS, Kufer P, Gökbuget N, Goebeler M, Klinger M, Neumann S et al. Targeted therapy with the T-cell-engaging antibody blinatumomab of chemotherapy-refractory minimal residual disease in B-lineage acute lymphoblastic leukemia patients results in high response rate and prolonged leukemia-free survival. J Clin Oncol 2011; 29: 2493–2498.

  9. 9

    Klinger M, Brandl C, Zugmaier G, Hijazi Y, Bargou RC, Topp MS et al. Immunopharmacologic response of patients with B-lineage acute lymphoblastic leukemia to continuous infusion of T cell-engaging CD19/CD3-bispecific BiTE antibody blinatumomab. Blood 2012; 119: 6226–6233.

  10. 10

    Feldmann A, Stamova S, Bippes CC, Bartsch H, Wehner R, Schmitz M et al. Retargeting of T cells to prostate stem cell antigen expressing tumor cells: comparison of different antibody formats. Prostate 2011; 71: 998–1011.

  11. 11

    Robak T, Wierzbowska A . Current and emerging therapies for acute myeloid leukemia. Clin Ther 2009; 31: 2349–2370.

  12. 12

    Zhong RK, Lane TA, Ball ED . Generation of T-cell lines to autologous acute myeloid leukemia cells by competitive limiting dilution culture of acute myeloid leukemia mononuclear cells. Exp Hematol 2008; 36: 486–494.

  13. 13

    Draube A, Beyer M, Wolf J . Activation of autologous leukemia-specific T cells in acute myeloid leukemia: monocyte-derived dendritic cells cocultured with leukemic blasts compared with leukemia-derived dendritic cells. Eur J Haematol 2008; 81: 281–288.

  14. 14

    Rezvani K, Yong AS, Tawab A, Jafarpour B, Eniafe R, Mielke S et al. Ex vivo characterization of polyclonal memory CD8+ T-cell responses to PRAME-specific peptides in patients with acute lymphoblastic leukemia and acute and chronic myeloid leukemia. Blood 2009; 113: 2245–2255.

  15. 15

    Bornhäuser M, Thiede C, Platzbecker U, Kiani A, Oelschlaegel U, Babatz J et al. Prophylactic transfer of BCR-ABL-, PR1-, and WT1-reactive donor T cells after T cell-depleted allogeneic hematopoietic cell transplantation in patients with chronic myeloid leukemia. Blood 2011; 117: 7174–7184.

  16. 16

    Stamova S, Cartellieri M, Feldmann A, Bippes CC, Bartsch H, Wehner R et al. Simultaneous engagement of the activatory receptors NKG2D and CD3 for retargeting of effector cells to CD33-positive malignant cells. Leukemia 2011; 25: 1053–1056.

  17. 17

    Aigner M, Feulner J, Schaffer S, Kischel R, Kufer P, Schneider K et al. T lymphocytes can be effectively recruited for ex vivo and in vivo lysis of AML blasts by a novel CD33/CD3-bispecific BiTE antibody construct. Leukemia 2013; 27: 1107–1115.

  18. 18

    Arndt C, von Bonin M, Cartellieri M, Feldmann A, Koristka S, Michalk I et al. Redirection of T cells with a first fully humanized bispecific CD33-CD3 antibody efficiently eliminates AML blasts without harming hematopoietic stem cells. Leukemia 2013; 27: 964–967.

  19. 19

    Dinndorf PA, Andrews RG, Benjamin D, Ridgway D, Wolff L, Bernstein ID . Expression of normal myeloid-associated antigens by acute leukemia cells. Blood 1986; 67: 1048–1053.

  20. 20

    Legrand O, Perrot JY, Baudard M, Cordier A, Lautier R, Simonin G et al. The immunophenotype of 177 adults with acute myeloid leukemia: proposal of a prognostic score. Blood 2000; 96: 870–877.

  21. 21

    Feldmann A, Arndt C, Töpfer K, Stamova S, Krone F, Cartellieri M et al. Novel humanized and highly efficient bispecific antibodies mediate killing of prostate stem cell antigen-expressing tumor cells by CD8+ and CD4+ T cells. J Immunol 2012; 189: 3249–3259.

  22. 22

    Stamova S, Cartellieri M, Feldmann A, Arndt C, Koristka S, Bartsch H et al. Unexpected recombinations in single chain bispecific anti-CD3-anti-CD33 antibodies can be avoided by a novel linker module. Mol Immunol 2011; 49: 474–482.

  23. 23

    Koristka S, Cartellieri M, Arndt C, Bippes CC, Feldmann A, Michalk I et al. Retargeting of regulatory T cells to surface-inducible autoantigen La/SS-B. J Autoimmun 2013; 42: 105–116.

  24. 24

    Koristka S, Cartellieri M, Theil A, Feldmann A, Arndt C, Stamova S et al. Retargeting of human regulatory T cells by single-chain bispecific antibodies. J Immunol 2012; 188: 1551–1558.

  25. 25

    Carmo-Fonseca M, Pfeifer K, Schröder HC, Vaz MF, Fonseca JE, Müller WE et al. Identification of La ribonucleoproteins as a component of interchromatin granules. Exp Cell Res 1989; 185: 73–85.

  26. 26

    Dreier T, Baeuerle PA, Fichtner I, Grün M, Schlereth B, Lorenczewski G et al. T cell costimulus-independent and very efficacious inhibition of tumor growth in mice bearing subcutaneous or leukemic human B cell lymphoma xenografts by a CD19-/CD3- bispecific single-chain antibody construct. J Immunol 2003; 170: 4397–4402.

  27. 27

    Schlereth B, Quadt C, Dreier T, Kufer P, Lorenczewski G, Prang N et al. T-cell activation and B-cell depletion in chimpanzees treated with a bispecific anti-CD19/anti-CD3 single-chain antibody construct. Cancer Immunol Immunother 2006; 55: 503–514.

  28. 28

    Lutterbuese R, Raum T, Kischel R, Lutterbuese P, Schlereth B, Schaller E et al. Potent control of tumor growth by CEA/CD3-bispecific single-chain antibody constructs that are not competitively inhibited by soluble CEA. J Immunother 2009; 32: 341–352.

  29. 29

    Amann M, D'Argouges S, Lorenczewski G, Brischwein K, Kischel R, Lutterbuese R et al. Antitumor activity of an EpCAM/CD3-bispecific BiTE antibody during long-term treatment of mice in the absence of T-cell anergy and sustained cytokine release. J Immunother 2009; 32: 452–464.

  30. 30

    Fortmüller K, Alt K, Gierschner D, Wolf P, Baum V, Freudenberg N et al. Effective targeting of prostate cancer by lymphocytes redirected by a PSMA × CD3 bispecific single-chain diabody. Prostate 2011; 71: 588–596.

  31. 31

    Berthon C, Driss V, Liu J, Kuranda K, Leleu X, Jouy N et al. In acute myeloid leukemia, B7-H1 (PD-L1) protection of blasts from cytotoxic T cells is induced by TLR ligands and interferon-gamma and can be reversed using MEK inhibitors. Cancer Immunol Immunother 2010; 59: 1839–1849.

  32. 32

    Ma Q, Wang C, Jones D, Quintanilla KE, Li D, Wang Y et al. Adoptive transfer of PR1 cytotoxic T lymphocytes associated with reduced leukemia burden in a mouse acute myeloid leukemia xenograft model. Cytotherapy 2010; 12: 1056–1062.

  33. 33

    Hornig N, Kermer V, Frey K, Diebolder P, Kontermann RE, Müller D . Combination of a bispecific antibody and costimulatory antibody-ligand fusion proteins for targeted cancer immunotherapy. J Immunother 2012; 35: 418–429.

  34. 34

    Cunha AC, Weigle B, Kiessling A, Bachmann M, Rieber EP . Tissue-specificity of prostate specific antigens: comparative analysis of transcript levels in prostate and non-prostatic tissues. Cancer Lett 2006; 236: 229–238.

  35. 35

    Kloss CC, Condomines M, Cartellieri M, Bachmann M, Sadelain M . Combinatorial antigen recognition with balanced signaling promotes selective tumor eradication by engineered T cells. Nat Biotechnol 2013; 31: 71–75.

  36. 36

    Dreier T, Lorenczewski G, Brandl C, Hoffmann P, Syring U, Hanakam F et al. Extremely potent, rapid and costimulation-independent cytotoxic T-cell response against lymphoma cells catalyzed by a single-chain bispecific antibody. Int J Cancer 2002; 100: 690–697.

  37. 37

    Haas C, Krinner E, Brischwein K, Hoffmann P, Lutterbüse R, Schlereth B et al. Mode of cytotoxic action of T cell-engaging BiTE antibody MT110. Immunobiology 2009; 214: 441–453.

  38. 38

    Holliger P, Hudson PJ . Engineered antibody fragments and the rise of single domains. Nat Biotechnol 2005; 23: 1126–1136.

  39. 39

    Yokota T, Milenic DE, Whitlow M, Schlom J . Rapid tumor penetration of a single-chain Fv and comparison with other immunoglobulin forms. Cancer Res 1992; 52: 3402–3408.

Download references


We thank Livia Schulze, Kristin Heidel, Barbara Uteß and Christine Gräfe for their excellent technical assistance; Professor Dr Dirk Lindemann for providing us with the lentiviral vector system; and Professor Dr Christian Thiede for providing the CD33+ cell lines MOLM-13, MV4-11 and OCI-AML3. This study was supported by a grant of the Medical faculty of the Technical University Dresden to Marc Cartellieri, a seed grant by the Center for Regenerative Therapies Dresden (CRTD), Technical University Dresden, and the José Carreras Stiftung to Michael Bachmann.

Author information



Corresponding author

Correspondence to M Bachmann.

Ethics declarations

Competing interests

MB, SS and GE have filed provisional patent application related to the antibodies directed to CD33 and La.

Additional information

Supplementary Information accompanies this paper on the Leukemia website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Arndt, C., Feldmann, A., von Bonin, M. et al. Costimulation improves the killing capability of T cells redirected to tumor cells expressing low levels of CD33: description of a novel modular targeting system. Leukemia 28, 59–69 (2014).

Download citation


  • single-chain bispecific antibodies
  • T-cell retargeting
  • acute myeloid leukemia
  • CD33
  • immunotherapy
  • costimulatory immunoligands

Further reading