Serum-free generation and quantification of functionally active Leukemia-derived DC is possible from malignant blasts in acute myeloid leukemia and myelodysplastic syndromes

Patients’ characteristics, sample collection and diagnosis

Mononuclear cells (MNC) from heparinized blood (PB) or bone marrow (BM) (PB-MNC, BM-MNC) were isolated from the interphase by density-gradient centrifugation (Ficoll-Hypaque, Biochrom, Germany), washed and suspended in PBS without Ca²⁺ and Mg²⁺ (Biochrom). Diagnosis of AML and MDS cases was based on the French-American-British (FAB) classification [4, 5]. Samples were collected in active stages of the disease from 100 AML patients (83 at first diagnosis, 7 in persisting disease and 10 at relapse), from 55 MDS patients as well as from 38 healthy donors after obtaining informed consent (Table 1). The median age of the AML Patients was 56 years (range: 20–76 years), the female:male ratio was 1:1.8. The median age of the MDS patients was 65 years, the female:male ratio 1:1. Cytogenetic analyses were performed according to the standard protocols and criteria defined by the International System for Human Cytogenetic Nomenclature [41]. Patients were categorized in risk groups as described [26, 27]: ‘Favorable risk’ AML patients had presented with a t(8;21), t(15;17), inv(16), t(16;16); ‘poor risk’ AML patients with −5/5q-, −7/7q-, t(11q23), inv(3), t(3;3), 17p abnormalities or a complex aberrant karyotype (=three abnormalities); ‘intermediate risk’ AML patients had presented with a normal karyotype or with any of the remaining aberrations. ‘Favorable risk’ MDS patients had presented with a normal karyotype, a del(5q) only, a del(20q) only or a – Y only; ‘poor risk’ MDS patients had presented with −7/7q aberrations or a complex aberrant karyotype; ‘intermediate risk’ patients were MDS patients with any of the remaining aberrations.

In addition, DC were generated from four different AML or MDS cell lines: Kazumi and HL60 (AML-M2), Mutz-3 (AML-M4) and Mutz-1 (MDS-RAEB).

DC generation

MNC were incubated in 12-well multiwell tissue culture plates in Xvivo 15 (BioWhittaker, Europe) FCS-free medium at a concentration of 2.5×10⁶ cells/ml for 7 (healthy samples) −14 days (AML/MDS samples), containing 800 U/ml GM-CSF (Sandoz, Germany), 500 U/ml IL-4 (CellConcepts, Germany), 40 ng/ml FL (FLT3-Ligand, PromoCell, Germany) and for the last 2 days 200 U/ml TNFα (CellConcepts). In parallel experiments, DC were generated in ‘MCM-Mimic’ medium containing 800 U/ml GM-CSF, 500 U/ml IL-4, 10 ng/ml IL-1β (CellConcepts), 1,000 U/ml IL-6 (CellConcepts), 200 U/ml TNFα, 1 μg/ml PGE₂ (Pharmacia biotech, Germany) and in addition 40 ng/ml FL. Half medium exchange was performed every 4–7 days with fresh cytokine-supplemented medium. After 7–14 days in culture, DC were harvested for subsequent experiments. If available, autologous CD3⁺ cells were positively selected using anti-CD3 antibodies and immunomagnetic MACS separation according to the manufacturer’s instructions (Miltenyi Biotec, Germany) and used for mixed lymphocyte reactions. The purity of CD3⁺ cells was 98%.

Flow cytometry

Flow cytometric analyses with a panel of mouse monoclonal antibodies (moAbs) directly conjugated with fluorescein isothiocyanate (FITC), phycoerythrin (PE) or with tandem Cy5-PE conjugation (PC5) were performed to evaluate and quantify amounts and phenotypes of leukemic cells, DC, B cells, T cells and NK cells in the PB/BM samples analyzed. Antibodies were purchased from Beckman Coulter^*, Becton Dickinson^**, R&D^***, Serotec^**** and Biozol^***** (CD1a^* ; CD3^* [T3]; CD14^* [LPS-R]; CD15^* [Lewis X]; CD19^* [B4]; CD28^* [CD80/86-ligand]; CD33^* ; CD34^* ; CD40^* ; CD56^* [N-CAM]; CD71^* [Transferrin receptor]; CD80^**** [B7.1]; CD83^* [HB15]; CD86^**** [B7.2]; CD117^* [c-kit]; CD135^* [FLT3]; CD137^** [4-1BB]; CD137L^***** [4-1BBL]; CD152^* [CTLA-4]; CD154^** [CD40L]; CD197^*** [CCR7]; CD209^*** [DC-sign]; HLA-Dr^* [MHC class II]; 7.1^* [NG2-antigen]). In MDS cases, amounts of CD34⁺ cells were regularly determined and analyzed to evaluate the amounts of undifferentiated cells/blasts [12, 10]. The MNC or cultured cells were suspended in PBS with 20% FCS (Biochrome) and incubated with moAbs according to the manufacturer’s instructions. Appropriate isotype controls were used and at least 5,000 events were evaluated on a FACS Calibur Flow Cytometer (Becton Dickinson, Heidelberg, Germany) using the Cell Quest data acquisition and analysis software (Becton Dickinson). For analysis and quantification of leukemic cells and DC, the total MNC/DC fractions were gated. A leukemic BM sample was considered as ‘positive’ for a surface marker if the percentage of positive events in a gate surrounding blasts, lymphocytes and monocytes was more than 20%, as described [12]. For T-cell analyses a ‘lympho gate’ was created, including lymphocytes and activated lymphocytes (Fig. 1c). Proportions of positive events in defined gates compared with the isotype controls were calculated using the CellQuest Software (Becton Dickinson).

Simultaneous fluorescence in situ hybridization (FISH) and immunophenotyping analysis (‘FISH-IPA’)

AML or MDS samples known to exhibit numerical cytogenetic aberrations were selected for FISH-IPA analyses. Aliquots of MNC or DC created were centrifuged onto slides, air-dried and frozen. Surface-antigen staining of aceton fixated slides was performed using FITC or PE labeled (CD1a) DC markers [29]. Detection of chromosome aberrations was performed according to the manufacturer’s instructions after methanol/acetic acid fixation and denaturation of chromosomes, using centromere-specific probes for chromosome 7, 8 (Oncor, Germany) 9 or 21 (Abbot/Vysis). Hybridizations were carried out overnight at 37°C. Cy3-conjugated streptavidin and AMCA-conjugated antibodies were used for the detection of hybridized DNA probes and cells were counterstained with DAPI. Slides were analyzed using a Zeiss microscope and attached CCD camera with an Image analysis software (MetaSystems). CD1a⁺ DC carrying the clonal chromosomal markers were finally differentiated from CD1a⁺ DC without clonal markers and from clonal, cytogenetically aberrant cells not expressing DC antigens (Fig. 1b, right side): proportions of blasts converted to DC_leu as well as proportions of DC of leukemic origin were quantified in cases with more than 20 clonal and/or CD1a⁺ cells.

Viability analysis

To evaluate the amounts of viable cells or DC after the culture procedure, MNC or DC were gated and viable (Annexin⁻ PI⁻) cells quantified [33]. The test was carried out according to the manufacturer’s instructions (BioVision, BioCat): 1–3×10⁵ cells were incubated in 300 μl binding buffer for 5 min with 3 μl Annexin V-FITC and 3 μl PI in the dark. Without washing, the cells were measured in an FACS Calibur flow cytometer (Becton Dickinson), using the Cell Quest data acquisition and analysis software (Becton Dickinson).

Quantification and characterization of DC

The DC were generated as described, harvested, counted and DC quantified by Flow cytometry: cells which displayed the typical scatter of DC and expressed typical DC antigens (e.g. CD86, CD80, CD40, CD1a and CD83) were counted in a gate surrounding the total MNC/DC fraction analyzed. For quantification, those DC markers were selected that were not expressed on naïve blasts. The DC coexpressing CD83 were defined as ‘mature DC’. Other maturation criteria were the loss of CD14 or CD209 positivity. The DC coexpressing, e.g. CD86 or CD137L were defined as DC coexpressing costimulatory antigens. In AML cases with an immunophenotypically detectable blast population in MNC fractions, amounts of blasts being converted to DC coexpressing specific blast antigens (e.g. 7.1, CD56 and CD117) were quantified (Fig. 1b, left side). In cases with no available specific blast marker combination, but less than 5% CD14⁺ cells in the MNC-fraction, CD33 or CD13 antibodies were used to estimate proportions of blast cells converted to DC_leu. In MDS cases with >10% CD34⁺ or CD117⁺ cells, those markers were used to evaluate proportions of cells being converted to DC_leu. Microscopical controls were regularly performed and revealed the DC-typical morphology of large cells with irregular shapes and cytoplasmic projections (Fig. 1a).

Mixed lymphocyte cultures (MLR)

Positively selected CD3⁺ T cells (Milteney Biotech, 1×10⁶ cells/well) from MNC and AML or MDS patients were cocultured with autologous leukemia-derived DC, MNC or without DC or MNC as a control (1×10⁵) in 1 ml RPMI 1640 medium (Biochrom) containing 10% FCS, as described. After 5 days of coculture (DC-induced), T-cell activation with respect to proliferation and expression of DC-contact relevant antigens was measured by Flow cytometry.

T-cell stimulatory, proliferation-inducing function of leukemia-derived DC

The T-cell activating capacity of DC was evaluated by FACS analyses comparing the expression of costimulatory receptors on T cells before and after autologous MNC or DC contact and by evaluating the proliferation activity of T cells. T cells coexpressing, e.g. CD28, CD137 and CD154 were defined as ‘T cells coexpressing costimulatory receptors’. T-cell proliferation was evaluated by the quantification of T cells coexpressing HLA-Dr, CD71 or CD25 before and after DC contact as described [43] (Fig. 1c).

Generation of cytotoxic autologous T cells

Positively selected CD3⁺ T cells (2×10⁶) were cocultured with 5×10⁴ DC or MNC in a 24-well plate in a final volume of 2 ml RPMI+ 10% FCS, supplemented with IL-7 (day 0–21) (10 μg/ml; CellConcepts) and IL-2 (23 IU/ml (day 7–21) Proleukin R5, Chiron). T cells were restimulated with 5×10⁴ DC or MNC on day 7 and on day 14. Half medium exchange was carried out every 3–4 days. After six days the last restimulation, cells were harvested and the cytotoxicity assay was carried out.

Cytotoxicity assay (Fluorolysis)

The lytic activity of effector T cells was measured by a Fluorolysis assay by counting viable target cells, labeled with specific fluorochrome-labeled antibodies, before and after effector cell (E) contact. T cells (E) from AML or MDS patients or healthy donors were stimulated with DC or MNC, harvested and an aliquot cocultured in 1.5-ml Eppendorf tubes with thawed autologous MNC or DC or other target cells (T) at E:T ratios of 20:1, 10:1, 6:1 and 2:1 for 3 h at 37°C and 5% CO₂. Before culture, MNC-target cells were stained for 15 min with two FITC- and/ or PE-conjugated ‘blast’-specific antibodies and cocultured for 3 h with effector cells (T cells or DC as target cells were stained with T-/ DC-specific antibodies). As a control target effector cells were cultured separately and afterwards mingled with T cells. To evaluate the amounts of viable (7AAD⁻) target cells and to quantify the cell loss after the 3-h incubation time, cells were harvested, washed in PBS and resuspended in a FACS flow solution containing 7AAD (Becton Dickinson, Biosciences Pharmingen) and a defined number of Calibrate APC-labeled beads (Becton Dickinson). Viable cells were gated in a SSC/7AAD gate and afterwards viable cells coexpressing specific blast marker (combinations) were quantified taking into account defined counts of APC-labeled calibration beads as described. Cells were analyzed in a FACS Calibur Flow Cytometer using the CELL Quest software (Becton Dickinson). The percentage of lysis was the difference between proportions of viable blasts before and after the effector cell contact [32].

Statistical methods

Mean and standard deviation, median and range and two-tailed t-tests were performed with a personal computer using Excel 97 (Microsoft). Differences were considered as significant, if the P-value was <0.05.

Characterization of mononuclear cell fractions obtained from AML, MDS and healthy probands

Before culture, we analyzed the mononuclear cells from 100 AML cases in active stages of the disease, 55 MDS and 38 healthy samples by Flow cytometry to quantify monocytes, lymphocytes and blasts and to evaluate or to confirm the blast phenotype. On average, AML samples presented with 11% B cells, 9% T cells, 14% monocytes (5% in non-monocytoid differentiated leukemias) and 59% leukemic cells; in MDS samples, 6% B cells, 15% T cells, 9% NK cells, 12% monocytes and 6% CD34⁺ cells could be detected. In MNC samples from healthy probands, on average, 10% B cells, 39% T cells and 28% monocytes were found.

Mature and vital DC can be generated from AML and MDS mononuclear cell fractions

On average, 34% DC from AML samples, 20% from MDS samples and 25% from healthy samples could be generated. Quantification of DC was performed as described (Fig. 2). Between 43 and 58% of these DC coexpressed CD83, a marker for mature DC. Even more of these DC were negative for monocyte-marker CD14 (84% of DC in AML, 77% in MDS and 67% in healthy samples) or were negative for CD209⁻ (76% of DC in AML, 56% in MDS and 50% in healthy samples). Viability of cells analyzed after 7–14 days of culture varied between 38% and 64% with fresh or frozen/thawed samples used as cell sources. The highest DC counts from cell lines were obtained with the Mutz-3 (AML-M4) line. The maturity of these DC was even higher in cell lines (except from Mutz-3) than that in patients’ DC.

We can conclude that mature and vital DC can be generated both from AML and MDS-MNC samples.

DC harvest is highest in monocytoid subtypes, but independent of the cytogenetic risk

Quantification of DC was performed by Flow cytometry as described. The DC harvests were similar in non-monocytoid AML- or MDS-FAB types with, on average, 30% DC in AML or 20% DC in MDS. The Highest DC-counts were obtained from the (myelo)-monocytoid subtypes of AML (on average 40% DC) or CMML (36% DC, Fig. 3a) and also from the AML-M4 line Mutz3 (Fig. 2b). No differences in DC harvests were seen in AML and MDS cases subdivided into cytogenetic risk groups (Fig. 3b).

In 6% of AML cases, 31% of MDS cases and in 8% of healthy donors, less than 10% could be generated. With cases subdivided into different FAB types, we found that in 9% of cases with undifferentiated AML (M0-M2), but in none of the remaining AML-FAB types and in 31% of cases with RAEB or RAEBt, less than 10% DC could be generated (data not shown).

This means that the DC harvests are independent of cytogenetic risk groups and highest in monocytoid FAB types in AML and MDS and cannot be generated from all MNC samples.

Autologous T-cell activation is mediated by expression of costimulatory ligands on DC and costimulatory receptors on T cells

DC obtained from 38 healthy as well as from 100 AML and 55 MDS samples coexpress comparable amounts of costimulatory ligands (e.g. CD137L⁺, CD40⁺, CD86⁺, CD80⁺ and CD1a, Fig. 4a, left side). These costimulatory ligands were also expressed on DC generated from AML or MDS-cell lines (Fig. 4a, right side). Costimulatory receptors (e.g. CD28, CD154 and CD137) on T cells were upregulated after contact with autologous DC in MLR compared to controls (autologous T cells, cultured in parallel for 5 days). Compared to cocultures of autologous T cells with MNC, a coculture with DC increased amounts of T cells coexpressing costimulatory receptors (Fig. 4b).

This means that DC/T-cell contact-mediating antigens are expressed and upregulated on DC and autologous T cells from AML/MDS-MNC in comparable amounts as from healthy PB-MNC, and a conversion of MNC to DC in addition increases the expression of these antigens.

Leukemic origin of DC generated from AML and MDS samples can be proven by FISH-IPA or Flow cytometry

To prove the leukemic/clonal derivation of DC generated from AML or MDS-MNC, ‘FISH-IPA’ analyses were performed in cases with a clonal numerical aberration. The codetection of the clonal marker in CD1a⁺ cells was interpreted as proof for ‘leukemia-derived DC’ (Fig. 5a). We could demonstrate the clonal monosomy 7 in CD1a⁺ DC obtained from a case with AML-M6 and in two cases with sAML and the monosomy 9 in CD1a⁺ cells in a case with AML-M4. The clonal trisomias of chromosomes 8 could be detected in CD1a⁺ cells in two cases with MDS-RAEB and in a case with MDS-RAEBt. In another case of CMML, the monosomy 21 was demonstrated in CD1a⁺ cells. Figure 1b (right side) gives an example for the codetection of the clonal trisomy 8 in CD1a⁺ DC from a patient with MDS-RAEB. Moreover, a coexpression of the 7.1 blast-characterizing antigen, the cell surface protein product of 11q23 aberrations, on CD83⁺ mature DC derived from a patient with an AML-M5 (characterized by an 11q23 aberration at first diagnosis) is given (Fig. 1b, left side). Most of the leukemic cells in AML are CD33⁺, thus characterizing the myeloid blast phenotype. In addition, blasts regularly express ‘stem cell’ antigens like CD117, CD34 or non-myeloid lineage antigens like CD56; in MDS, CD34⁺ or CD117⁺ cell populations can be detected. In cases with a gain of ‘DC markers’ on the blast population after a 10 to 14-day culture-period, resulting in cells coexpressing blast (e.g. 7.1, CD56) with DC markers, proof for the leukemic derivation was given. In cases with a CD33⁺ blast phenotype, only cases with less than 5% CD14⁺ cells in the MNC fraction were included for analyses to exclude normal monocyte-derived DC. In 32 cases with AML and in 12 cases with MDS, a meaningful evaluation was possible and in all/of those cases, a coexpression of blasts and DC markers was seen, proving the leukemic derivation of DC.

FISH-IPA and Flow cytometry allow quantification of blasts converted to DC and of DC of non-leukemic origin

An important question was whether clonal/leukemic cells in AML and MDS were quantitatively converted to leukemia-derived DC (‘DC_leu’). In cases with more than 20 evaluable CD1a⁺ and/ or FISH-evaluable cases, we quantified clonal cells with/without coexpression of CD1a after conversion to DC. We could demonstrate that on average, 53% of clonal cells in AML/MDS (77% of leukemic cells in AML-DC cultures and 37% of clonal cells in MDS-DC cultures (range 18–54)) coexpressed CD1a, giving rise to leukemia-derived DC (Fig. 5b, left side). Quantitative evaluations by Flow cytometry were performed by including 32 AML cases with a CD117⁺, CD56⁺ and CD34⁺ blast phenotype or with a CD33⁺ AML-blast phenotype and less than 5% CD14⁺ cells in the MNC fraction. We could demonstrate that on average, 57% of those blasts coexpressed DC antigens after 10–14 days of culture, thus confirming data described above. In 12 MDS samples with >10% immunocytologically detectable blasts (CD34⁺ or CD117⁺), we found that 64% of these cells coexpressed DC antigens after 10–14 days of culture (Fig. 5b, middle).

Figure 5c gives some examples of Flow cytometry of MNC/DC in cases with AML and MDS: in cases 1–3, we demonstrate that the leukemic cell population in the MNC fraction, positive for CD33 (AML-M0), CD117 (MDS-RAEBt) or CD34 (MDS-RAEB) can be redetected in the DC culture after 10–14 days of culture, with those cells still being positive for these markers, but in addition characterized by an increased sideward scatter (SSC), typical for DC and an expression of DC markers. The coexpression of these ‘leukemic markers’ on DC demonstrate the conversion of these cells to leukemia-derived DC. In cases 4 and 5, we demonstrate that the leukemic population does not quantitatively gain the DC-typical increased sideward scatter or the coexpression of DC antigens on all blasts, but leukemic cell populations are still detectable at the same ‘MNC-Scatter position’.

In cases with clonal aberrations, we could demonstrate that beside clonal DC also CD1a⁺ cells without the clonal aberration of the leukemic cells were generated (Fig. 5b, right side).

We can conclude and prove that we can regularly create leukemia-derived DC in AML/MDS cases with clonal or leukemia-specific markers being copresented with DC markers. Moreover, we can summarize that clonal/leukemic cells as well as DC of non-leukemic origin are detectable in the DC culture: about 60% of clonal/leukemic cells in AML or MDS are convertible to DC and vice versa, with about 50% of DC created being of leukemic/clonal origin.

Conversion of MNC to DC increases autologous T cell activation with respect to proliferation, migration and production of leukemia-cytotoxic T cells

We studied the proliferation characteristics of T cells’ healthy AML and MDS probands without DC contact compared to that of T cells mixed with autologous DC for 5 days in an MLR setting. The results are given in Fig. 6a: compared to non-DC-stimulated T-cell controls, DC-stimulated T cells showed an increased stimulation of proliferation. Compared to cocultures of autologous T cells with MNC, a coculture with DC increased amounts of proliferating T cells (Fig. 6a). Figure 1c shows an example of activated T cells after a 5-day stimulation of T cells with autologous DC in a case of AML-M5: amounts of T cells coexpressing CD71 (signaling proliferation) and CD28 (signaling DC contact) are increased.

CD62L is expressed on CCR7⁺ or CCR7⁻ T cells. Both the expressions of CD62L and CCR7 on T cells mediate their homing to lymph nodes. We could demonstrate that the expression of these migration antigens is upregulated (6–23) on T cells after autologous DC contact in cases with AML as well as in healthy donors compared to T cells after autologous MNC contact (data not shown).

The specific lytic activity of T cells prepared from AML patients after an autologous DC-activation step was studied in a Fluorolysis assay. In three cases with AML, a specific lysis of naïve blasts expressing patient-specific blast antigens was shown, but no lysis of autologous T cells or DC (Fig. 6 b):

• Case 1: MNC from a patient with AML-M1 containing 53% CD33⁺ /CD117⁺ blasts were cocultured with autologous DC-primed T cells in the ratio of 2:1 and 6:1. After only 3 h, a specific lysis of 3%/24% CD33⁺ /CD117⁺ % blasts was seen (Fig. 6b, upper part, left side). These T cells, however, did not lyse autologous DC after a 3-h coculture in a ratio of 2:1 (Fig. 6b, lower part).

• Case 2: MNC from a patient with AML-M0 containing 23% CD34⁺ blasts were cocultured with autologous DC-primed T cells in the ratio of 2:1. After 3 h, a specific lysis of 19% of CD34⁺ blasts was seen (Fig. 6b, upper part).

• Case 3: MNC from a patient with AML-M2 containing 85% CD34⁺ blasts were cocultured with autologous DC-primed T cells in the ratio of 20:1. After 3 h, a specific lysis of 82% of CD34⁺ blasts was seen (Fig. 6b, upper part, right side). These T cells did not lyse autologous DC after a 3-h incubation in the ratio of 20:1(Fig. 6b, lower part).

DC-activated T cells from two healthy donors did not lyse autologous T cells or DC after 3 h, thereby representing another negative control (Fig. 6b, lower part). Using DC-primed or non-DC-primed T cells from three healthy donors, a lysis of 30–90% of allogeneic AML-blasts expressing a patient-specific blast phenotype was seen, thereby representing a positive control (data not shown). In three other cases with AML, no lysis of blasts by autologous or allogeneic DC-activated T cells was seen. In these three cases, less than 5% viable T cells were available. In another AML case, only 4% DC could be generated. Autologous T cells from this patient could not lyse naïve blasts, thereby demonstrating that T cells are necessary to lyse allogeneic or autologous blasts and DC are necessary to activate autologous T cells.

We can conclude that T cells proliferate after autologous contact with DC generated from AML- and MDS-MNC fractions. Moreover, they give rise to cytotoxic subtypes, which are able to specifically lyse (autologous) leukemic cells without lysing autologous (unstimulated) T cells or DC—supposing that enough DC and T cells are available.

Taken together, these data point to the crucial role of stimulated T cells and DC_leu to enable a specific lysis of naïve autologous blasts.

Expression of migration antigens on DC can be improved by the cultivation of DC in PGE₂ containing ‘MCM-Mimic’ medium

Only DC coexpressing migration markers like CCR7 (CD197) are able to migrate to lymph nodes and to present antigens to naïve T cells. Therefore we studied and compared the expression of migration antigens on DC created in GM-CSF, IL-4, TNFα and FL-standard medium (as described) with DC created in ‘MCM-Mimic’ medium containing GM-CSF, IL-4, TNFα, FL, IL-1â, IL-6 and PGE₂. We could show that MCM-Mimic not only increases the percent harvest of mature DC, but also especially proportions of CCR7⁺ DC compared to DC cultured in ‘standard’ medium (Fig. 7). Proportions of DC- coexpressing costimulatory antigens were not different in the groups compared (data not shown). These data prompted us to change DC-standard medium to ‘MCM-Mimic’.