Recognition of chronic myelogenous leukaemia cells by autologous T lymphocytes primed in vitro against the patient's dendritic cells

Authors

  • Ludmila Müller,

    1. Tübingen Ageing and Tumour Immunology Group (TATI), Section for Transplantation Immunology,
      University of Tübingen, and
    2. Second Department of Internal Medicine, University of Tübingen Medical School, Germany
    Search for more papers by this author
  • Carmelinda Provenzani,

    1. Tübingen Ageing and Tumour Immunology Group (TATI), Section for Transplantation Immunology,
      University of Tübingen, and
    2. Second Department of Internal Medicine, University of Tübingen Medical School, Germany
    Search for more papers by this author
  • Christoph Faul,

    1. Second Department of Internal Medicine, University of Tübingen Medical School, Germany
    Search for more papers by this author
  • Graham Pawelec

    1. Tübingen Ageing and Tumour Immunology Group (TATI), Section for Transplantation Immunology,
      University of Tübingen, and
    2. Second Department of Internal Medicine, University of Tübingen Medical School, Germany
    Search for more papers by this author

Ludmila Müller, Ph.D., Section for Transplantation Immunology, Second Department of Internal Medicine, University of Tübingen Medical School, Otfried-Müller Str.10, D-72076 Tübingen, Germany. E-mail: ludmila.mueller@med.uni-tuebingen.de

Abstract

Defects in immune responses are common in patients with chronic myelogenous leukaemia (CML). However, using dendritic cells (DCs) to promote T-cell immunity in vitro may nonetheless elicit potent specific anti-tumour responses for use in immunotherapy. Here, we show that DCs generated from CML patients had a typical dendritic phenotype and were able to stimulate autologous T cells. Three primed T-cell lines were studied in more detail in one patient. They were stimulated by autologous CML cells, but not by normal non-leukaemic cells from the patient's HLA-identical sibling. This was blocked by HLA-DR-specific, but not HLA-DQ- or HLA-DP-specific antibodies. CML-stimulated cytokine secretion, including interferon-γ and granulocyte macrophage-colony stimulating factor, suggested a Th1-type phenotype for these sensitized anti-leukaemic T cells. This study therefore shows that cells with a functional dendritic phenotype can be generated from the blood of CML patients and are potent inducers of T-cell responses to tumour cells. This approach allows sensitization of patients' T cells by their own particular tumour without the need to identify the exact leukaemia antigens involved, and may find application in immunotherapy of CML.

Chronic myelogenous leukaemia (CML) is a proliferative disorder of the haematopoietic system. It is characterized by clonal expansion of a primitive pluripotent stem cell that has the capacity to differentiate into the myeloid, monocyte, megakaryocyte and erythrocyte lineages (Pasternak et al, 1998; Faderl et al, 1999; Gordon et al, 1999). In CML, normal haematopoiesis is increasingly replaced by the production of leukaemic cells. During disease progression, multiple genetic changes occur, potentially generating novel antigens that are only present in leukaemic cells. Theoretically, these novel proteins and the junction-spanning sequences of oncogene fusion proteins, such as the characteristic 9:22 translocation of the CML Philadelphia chromosome, might make ideal targets for therapeutic immune responses because they are not present in normal cells.

There is increasing evidence from studies in vitro that specific T-cell immune responses can be generated against CML antigens (Ten Bosch et al, 1995; Boccia et al, 1996; Pawelec et al, 1996; Mannering et al, 1997; Yotnda et al, 1998; Osman et al, 1999). The lack of a T-cell response sufficient to protect perfectly against CML in vivo may be related to many factors, including inadequate antigen presentation by antigen presenting cells (APCs), failure of the tumour to express MHC antigen, or the abnormal expression of adhesion or accessory molecules leading to T-cell non-responsiveness (anergy). Some of these difficulties may be overcome by providing an optimal activation environment, using, for example, dendritic cells (DCs), which are considered to be the most effective professional APCs. DCs are specialized to internalize, process and present antigen, and they express high levels of MHC class I and II molecules, as well as accessory and adhesion molecules. Thus, appropriately pulsed DCs may be useful for inducing specific immune responses. Mannering et al (1997) utilized peptides from the bcr/abl fusion region (b3–a2) presented by DCs to raise specific CD4+ T cells from the peripheral blood mononuclear cells (PBMCs) of normal donors. CD4+ T cells that were restricted by MHC class II were obtained; they proliferated specifically either to DCs pulsed with peptides or to APCs exposed to lysates of leukaemia cells harbouring the specific translocation (Mannering et al, 1997). However, these investigators did not report the recognition of native CML cells by these sensitized T cells.

Clinical immunotherapy of cancer to date has largely been restricted to patients carrying particular HLA class I alleles and has relied on application of peptides representing known antigens, thereby excluding many patients who may have benefited from this approach. Advances in our understanding of anti-tumour immunity and the genetic alterations that accumulate in the progression of malignancy have recently provided unforeseen opportunities for the development of more selective and safer therapeutic approaches. One such strategy involves the use of dendritic cell-based vaccines. DCs could be pulsed in vitro with characterized tumour-associated antigens or with uncharacterized antigens extracted from individual tumours, such as cell lysates (Mannering et al, 1997; Nestle et al, 1998; Mackensen et al, 2000). An analogous approach, using electrofusion techniques, relies on the generation of hybrids between autologous tumour cells and allogeneic DCs. These hybrids may present antigens expressed by the tumour in concert with the co-stimulating capabilities of DCs (Kugler et al, 2000). Such an approach is an example of individualized immune therapy, which has been applied with some success in renal carcinoma.

In haematological malignancies such as CML, both malignant cells and DCs originate from the same haematopoietic precursor, they could have the same genetic defects and, therefore, may also express tumour-specific antigens (Heinzinger et al, 1999; Dietz et al, 2000). It has been recently reported that the percentage of bcr–abl+ dendritic cells derived from PBMCs of CML patients was more then 98% (Dietz et al, 2000). It might therefore be possible not only to prime T cells with DCs pulsed with CML antigen, but also to stimulate T cells directly with DCs that possessing the malignant phenotype themselves. The latter approach has resulted in the generation of T cells cytotoxic to autologous CML cells (Choudhury et al, 1997). However, the use of this approach to generate MHC class II-restricted CD4+ helper T-cell responses has not yet been reported.

We report here the derivation of T cells reactive to autologous tumour cells, using unpulsed patient DCs for priming, as well as DCs pulsed with whole tumour cell lysate. One advantage of this approach is that both endogenous and exogenous leukaemic antigens might be presented to T cells and responses could be achieved in an autologous system without a requirement for prior knowledge of the leukaemic antigens involved.

Patients and methods

Isolation of peripheral blood mononuclear cells After informed consent, peripheral blood from patients with CML in the chronic phase of disease and from their HLA-identical siblings was obtained. PBMCs from CML patients attending the Department of Internal Medicine II, University Hospital, Tübingen as outpatients or their normal healthy sibling donors were separated from heparinized peripheral blood on Ficoll discontinuous density gradients (FicoLite H; LINARIS, Bettingen am Main, Germany). Interphase cells were harvested, washed three times and cultured, or stored in liquid nitrogen until use.

HLA typing PBMCs from CML patients and their normal healthy sibling donors were typed for human leucocyte antigen (HLA) class I and II alleles in the context of the bone marrow transplantation programme. Standard molecular methods were used to establish the HLA-A and B types, as well as HLA-DR and DQ.

Generation of dendritic cells DCs were derived from PBMCs isolated from CML patients (Fig 1). Cells were counted, assayed for viability using trypan blue exclusion and monocytes were separated using plastic adherence in serum free X-Vivo 15 medium (BioWhittaker, Walkersville, MD, USA) for 2 h at 37°C in a 5% CO2 atmosphere. Non-adherent cells were washed off by gentle rinsing with Roswell Park Memorial Institute (RPMI) medium. Adherent cells were cultured in X-Vivo 15 with 75 ng/ml of granulocyte macrophage-colony stimulating factor (GM-CSF; Leucomax 300, Novartis, Nürnberg, Germany) and 500 U/ml interleukin 4 (IL-4; Novartis, Basel, Switzerland) for 6–8 d, followed by maturation with 10 ng/ml tumour necrosis factor-α (TNF-α; PharMingen, Hamburg, Germany). Cultures were maintained at 37°C in 5% CO2 by replacing culture medium and cytokines every 2 d. The development of DCs was monitored by the detection of large, stellate cells with typical motile veils in phase-contrast microscopy. Before adding TNF-α, one aliquot of the immature DCs was pulsed with tumour cell lysate from the same patient (Fig 1). For this purpose, the PBMCs of patient CML-3, containing 51·2% CD 33+ cells were freeze/thawed three times at 1 × 106 cells/well using liquid nitrogen and a 37°C water bath. No further treatment of the lysates was undertaken; thus, they contained all the components of necrotic tumour cells. These lysates were added for 6 h to the 2 × 105 dendritic cells. The second aliquot of DCs was left untreated.

Figure 1.

Schematic chart showing the generation of dendritic cells from CML-PBMCs, pulsing with CML cell lysate and sensitization of autologous T cells: After preparation of PBMCs using density gradient centrifugation, the adherent cell fraction of PBMCs was cultured in X-Vivo 15 with GM-CSF and IL-4 for 6–8 d. Before adding TNF-α, some of the immature DCs were pulsed with freeze-thawed tumour cell lysate from the same patient; the remainder was not treated with CML cell lysate at all. The maturation of DCs occurred with addition of TNF-α. In the next step, autologous PBMCs, depleted of tumour cells by pretreatment with ara-C in vitro, were stimulated with these mature DCs. The T-cell lines obtained in this way were cloned using limiting dilution. The effector functions of the T-cell lines and T-cell clones were screened for proliferation and cytokine production.

Generation of T-cell lines Autologous PBMCs at 1 × 106 cells/ml, depleted of tumour cells by pretreatment with 100 μg/ml cytosine arabinoside (ara-C) for 1 h in vitro (Pawelec et al, 1989), were stimulated with DCs at 2 × 105 cells/ml. Two approaches were used to sensitize patients' autologous T cells: (i) against leukaemic DCs at 1 × 106 cells/well alone; (ii) against leukaemic DCs pulsed with tumour cell lysate (Fig 1). Culture medium was X-Vivo 15, in 2 ml cluster plates (Greiner, Nürtingen, Germany). After 10 d, stimulated T cells were cloned using limiting dilution.

T-cell cloning T cells were plated at 0·45 per well in 1 mm diameter microplate wells (Terasaki plates, Greiner, Frickenhausen, Germany) containing 1 × 104 30 Gy-irradiated pooled PBMCs from 21 random normal donors as feeder cells in X-Vivo 15 medium. Contents of positive wells were transferred to 96-well 7 mm diameter flat-bottomed microtitre plates containing fresh X-Vivo 15 medium with IL-2 (20 U/ml) and 1 × 105 pooled PBMC feeder cells. After 10–12 d in culture, the contents of wells that were observed to be at least one third full of growing cells were transferred to 16 mm diameter 24-well plates with 5 × 105 feeder cells per well in X-Vivo 15 with IL-2. Cultures were given fresh medium and feeder cells every 7–10 d. The fraction of transferred clones achieving a population size of greater than 1 × 106 cells was studied further.

Phenotypic analysis of DCs Purity of the DC fraction was confirmed by assessment of morphology under phase-contrast microscopy and by immunophenotyping. Surface marker analysis of in vitro cultured cells was performed using a FACScan (Becton Dickinson). The monoclonal antibodies (mAbs) used were specific for HLA-DR (L243) obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA); HLA-DP (B7/21), a kind gift of Dr I. Trowbridge, San Diego, USA; HLA-DQ (Tü22), a local mAb; CD3 (OKT3), CD8 (OKT8), CD1a (OKT6) and MHC I (W6/32) were also obtained from the ATCC; CD4 (TT1) was kindly provided by Prof. G. A. Müller (University of Göttingen, Germany); CD80 (B7-24) was from Immunogenetics, Ghent, Belgium; CD83 was from Immunotech, Hamburg; and CD86 (IT2·2) was from PharMingen, Hamburg. The class I mAb W6/32 was used as a positive control (W6/32.HL) and its non-binding variant (W6/32.HK) as a negative control. Labelled cells were developed with DTAF [fluoroscein isothiocyanate (FITC]-conjugated F(ab)2 rabbit anti-mouse IgG and analysed on a FACS.

T-cell proliferation assays Proliferation was assessed using a standard [3H]-thymidine incorporation assay. Sensitized T lymphocytes (3 × 104 cells per well) or cloned T cells (2 × 104 cells/well) were cultured with 5 × 104 cells/well irradiated autologous PBMCs from CML patients or an HLA-matched healthy sibling in 96-well U-bottomed plates (Greiner, Nürtingen, Germany). For blocking experiments, anti-HLA-DR, -DQ and -DP antibodies were added at ≈ 2·5 μg/ml as hybridoma supernatants. After 48 h, 37 kBq/well [3H]-thymidine (Amersham Buchler, Braunschweig, Germany) was added for a further 18 h before harvesting onto glass fibre filter mats and counting in a β-scintillation counter (LKB-Wallac, Turku, Finland). Results are expressed as means of triplicate wells.

ELISA measurements of cytokines Enzyme-linked immunosorbent assays (ELISAs) were used for assessing concentrations of cytokines released into culture media. Cells were cultured at 6 × 105/ml in X-Vivo 15. Cell-free supernatants of these T cells, CML-PBMCs or T cells restimulated with autologous PBMCs were assayed. Antibody pairs and cytokine controls were purchased from PharMingen (Hamburg, Germany). ELISAs were performed according to the manufacturer's instructions using 1 μg/ml capture antibody and 0·5 μg/ml biotinylated detection antibody.

Results

In vitro generation of DCs from the adherent fraction of PBMCs from CML patients

It was established that DCs could be reliably generated from the adherent cell fraction of peripheral blood from 17 different CML patients (data not shown). Here, we describe in detail the generation of DCs and their use in T-cell sensitizations in one of these patients. Monocytes were differentiated using GM-CSF and IL-4, followed by maturation with TNF-α. One aliquot of immature DCs was pulsed with freeze/thawed autologous tumour cell lysate, whereas a second was left untreated (Fig 1). In the early stages of culture, we observed that the cells were dispersed and spindle-shaped, but gradually over time they seemed to form two groups of cells. The first group formed cell clusters of non-adherent dendritic cells. A second type formed an adherent cell population of flat cells spreading on the plastic substratum. The non-adherent fraction of cells generated in these cultures displayed typical DC morphology, exhibiting long, thin cytoplasmic projections of lamellopodia and increased size. Using flow cytometry, we demonstrated that these DCs, generated from PBMCs of CML patients, had a typical dendritic surface phenotype (as shown for patient CML 3 in Fig 2). They were highly positive for HLA-DR, CD86 and CD80, positive for CD1a and CD83, and negative for the monocytic marker CD14.

Figure 2.

Surface phenotype of dendritic cells derived from adherent cell fraction of patient CML-3. Flow cytometric analysis was performed after 8 d in culture with IL-4 and GM-CSF, followed by 2 d incubation with TNF-α. The class I mAb W6/32 was used as a positive control (HL) and its non-binding variant W6/32.HK (HK) as a negative control. Labelled cells were developed with DTAF (FITC)-conjugated F(ab)2 rabbit anti-mouse IgG and analysed on a FACS.

T-cell stimulation by autologous DCs

Autologous peripheral mononuclear cells, depleted of tumour cells by pretreatment with ara-C in vitro, were added to the DCs obtained as described above. After 7 d of co-cultivation, we observed proliferating cells in the vicinity of clustered DCs. These were propagated until d 10, when some were used for limiting dilution cloning and the remainder were tested for their responses to autologous CML cells.

Three cell lines, all of which showed some evidence for recognition of native autologous CML cells (see below), were phenotyped for T-cell markers. As expected, they all consisted of 100% T cells (CD3+). Two of them were essentially pure CD4+ (TL-DC3-0 and TL-DC3-2), whereas the third (TL-DC3-1) clearly consisted of a mixture of CD4- and CD8-single positive T cells (Fig 3).

Figure 3.

Phenotypic profile of T-cell lines obtained by stimulation with unpulsed DCs (TL-DC3-0) and with DCs pretreated with tumour cell lysate (TL-DC3-1 and TL-DC3-2).

Autocrine proliferative responses of DC-primed T cells

These three T-cell lines obtained by stimulation with CML-DCs manifested different patterns of autocrine proliferative activity and cytokine secretion after restimulation with autologous CML cells. While the line TL-DC3-1, obtained by stimulation with lysate-pulsed DCs, clearly proliferated more strongly when challenged with autologous CML cells than in the absence of stimulation (stimulation index 4·8), two other lines, designated TL-DC3-0 and TL-DC3-2, did not show appreciable autocrine proliferative activity when stimulated thus (Fig 4). One of these was derived from sensitization against untreated DCs and the other from lysate-pulsed DCs (see Fig 1).

Figure 4.

T-cell lines obtained by stimulation with CML-DCs show different patterns of autocrine proliferation after restimulation with autologous CML cells (CML-3). The responders were T-cell lines TL-CML3-0, TL-CML3-1 and TL-CML3-2. The stimulators were autologous CML cells. The column labelled CML-3 represents thymidine incorporation by irradiated CML cells alone.

Because the capacity for clonal expansion (autocrine proliferation) is essential for successful immune responses, further studies first focused on the proliferative line TL-DC3-1. Cells from this line were rechallenged with autologous CML cells (CML3) and with cells from the patient's HLA identical sibling (D). As shown in Fig 5, the latter failed to stimulate a proliferative response, showing lack of autoreactivity against, for example, MHC class II molecules by TL-DC3-1. Restimulation with allogeneic CML revealed positive but lower responses against these cells (Fig 5). This allogeneic CML patient (CML2) shared an HLA-DRB1 allele and both HLA-DQB1 alleles, as well as HLA-A2, with the TL-DC3-1 donor (Table I). Further investigations on the HLA restriction of these responses were therefore necessary and mAb inhibition experiments were performed using autologous CML cells as stimulators. Clearly, only HLA-DR (L243)-specific, but not DQ (TÜ22)- or DP (B7/21)-specific mAbs blocked stimulation of TL-DC3-1 cells (Fig 6).

Figure 5.

Proliferative activity of TL-DC3-1 T cell line (R) after restimulation with autologous CML cells (R+CML3), with HLA-identical bone marrow donor (R+D), with allogeneic partly MHC-matched CML cells (R + CML2), and with addition of phytohaemaglutinin (PHA) as a control, yielding maximum responses. The responder was T-cell line TL-CML3-1.

Table I.  HLA types of CML patients.
 HLA-AHLA-BHLA-CHLA-DRB1HLA-DRB3/4HLA-DRB5HLA-DQB1
CML3A2B7Cw-DRB1*1501 DRB5*0101DQB1*0603
A32B61Cw2DRB1*1302DRB3*0202 DQB1*0602
CML2A*01xB*39xCw*03xDRB1*1301DRB3*0101 DQB1*0602
A*02xB*40xCw*0702DRB1*1501 DRB5*0101DQB1*0603
Figure 6.

Inhibition of restimulation by autologous PBMCs with HLA-specific antibodies: Resp.-T cells from TL-DC3-1 alone; CML3, stimulator cells alone; R+CML3, autocrine proliferation of T cells after restimulation with autologous tumour cells; R+CML3 + B7/21, the same with addition of anti-DP-Ab; R+CML3+Tü22, the same with addition of anti-DQ-Ab; R+CML3+l243, the same with an addition of anti-DR-Ab

Cytokine secretion by DC-primed T cells

Antigen-specific IFN-γ secretion by the three T-cell lines was tested. Patterns of response different from autocrine proliferation emerged when IFN-γ was assayed. Now, the TL-DC3-0 cells, derived from sensitization against untreated CML-DCs, which had not been able to proliferate, showed evidence of recognition of autologous CML cells. Thus, in the absence of stimulation they secreted only 20 pg/ml of IFN-γ, but when rechallenged with autologous CML cells this increased to 200 pg/ml (Fig 7). The TL-DC3-2 T-cell line showed an even higher level of CML-stimulated IFN-γ secretion, reaching 440 pg/ml under the same conditions, whereas the T-cell line TL-DC3-1 was not able to secrete this cytokine at all.

Figure 7.

IFN-γ secretion of T-cell lines after restimulation with autologous CML-3 cells. The responders were T-cell lines TL-CML3-0, TL-CML3-1 and TL-CML3-2. The stimulators were autologous CML-3 cells. The column labelled CML-3 depicts spontaneous GM-CSF secretion by the CML cells alone.

Cloning the DC-primed T-cell lines

We attempted to clone cells from the three lines using limiting dilution. After transfer to microtitre plates, many clones were short-lived and could not be propagated extensively. We were not able to obtain clones from either the TL-DC3-0 or TL-DC3-1 T-cell lines, but 78 clones were derived from the TL-DC3-2 line (all tested clones were CD4-single positive). Six of these showed an increased secretion of the general activation marker cytokine GM-CSF in response to restimulation with autologous tumour cells as opposed to non-stimulated cells (Fig 8A). Only four of these six clones (Fig 8B) also showed high IFN-γ release after restimulation with autologous CML cells. However, those that did secrete cytokine secreted very large amounts. Indeed, in some cases, cytokine production of specifically stimulated cells was higher than cytokine production with the addition of phytohaemagglutinin (PHA) (maximal control stimulus).

Figure 8.

GM-CSF (A) and IFN-γ (B) release of T-cell clones generated using limiting dilution from T-cell line TL-DC3-2 after restimulation with autologous CML-PBMCs

Discussion

We show here that functional DCs can be generated from adherent PBMCs of CML patients in the chronic phase by culturing them with GM-CSF and IL-4, followed by TNF-α. Such differentiated CML-DCs are capable of sensitizing T cells to recognize autologous tumour cells and to respond by secreting IFN-γ by themselves. However, the T cells sensitized in this way did not undergo clonal expansion on rechallenge with autologous CML cells. To obtain autocrine proliferative cell lines, it was necessary to pulse the CML-DCs with tumour cell lysate. This resulted in a line, TL-DC3-1, which proliferated strongly to autologous tumour cells, but more weakly to HLA partly matched allogeneic CML cells and not at all to HLA-identical sibling PBMCs. Using antibody blocking, the response to autologous CML cells was shown to be HLA-DR-restricted, but the nature of the antigenic epitopes recognized is thus far unknown. Various approaches may be taken to identify the antigens recognized, including eluting and sequencing peptides from the tumour cells, an approach currently being exploited in our laboratory (Halder et al, 2000). However, by using tumour lysates and potent APCs, tumour-specific T cells with potential use in immunotherapy can be generated from each individual patient without knowledge of the exact nature of the antigens recognized. This may offer a significant advantage over immunotherapeutic approaches using defined synthetic peptides.

Evidence that human T lymphocytes can discriminate between leukaemia cells and normal cells continues to accumulate slowly from our and other studies. Several recent investigations have involved the use of DCs as APCs in an attempt to stimulate both primary and secondary immune responses to poorly immunogenic tumours (Choudhury et al, 1997; Nieda et al, 1998; Osman et al, 1999; Dietz et al, 2000). Some progress has been made using DCs to initiate anti-tumour activity in mouse models (Paglia et al, 1996; Young & Inaba, 1996; Fields et al, 1998) and in humans (Choudhury et al, 1997; Eibl et al, 1997; Mannering et al, 1997; Nieda et al, 1998; Osman et al, 1999). In CML it was shown that CD8+ cytotoxic effector cells specific for bcr–abl fusion peptides can be generated from normal peripheral blood after stimulation with autologous DCs pulsed with 16 mer b3a2 fusion peptide (Nieda et al, 1977). However, particularly for HLA class I-restricted responses, synthetic peptides of a given amino acid sequence can be loaded only onto DCs expressing a particular MHC haplotype. In such strategies, DCs could be pulsed with peptides specifically designed to fit the HLA-binding motifs of individual patients. Alternatively, if DCs could be generated from PBMCs of CML patients and presented endogenous antigens, immunotherapy would not be limited by the requirement to match peptide with patients' HLA types (Heinzinger et al, 1999). Dietz et al (2000) compared DC preparations obtained from patients suffering from CML with DCs prepared from normal CD14+ mononuclear cells. They studied normal DC and bcr–abl+ leukaemic DC yields, expression of membrane molecules, differentiation status and ability to stimulate T cells. CML-DCs and normal DCs were indistinguishable in expression of CD83 and in their yields, and were equally effective in stimulating proliferation of allogeneic and autologous T cells when stimulated by DCs pulsed with keyhole limpet haemocyanin. Thus, functional DCs could be obtained from monocytes of CML patients under conditions established for the generation of normal DCs, but this study did not show the recognition of autologous leukaemic cells. The percentage of bcr–abl+ cells in PBMCs varied among patients between 65% and 97%, and the final CML-DC preparations were > 98% bcr–abl+, the highest purity of bcr–abl+ cells to date (Dietz et al, 2000). However, some dendritic cell cultures stimulated strong allogeneic responses, but not peptide- and tumour cell-specific responses (Heinzinger et al, 1999; Wang et al, 1999; Dietz et al, 2000). In contrast, we have found that sensitization with patient DCs loaded with tumour cell lysate can elicit T cells that recognize autologous tumour cells. The DC phenotype of generated DCs was defined as upregulation of the co-stimulatory molecules CD80 and CD86, and the expression of CD1a and CD83, as with healthy controls. We found that DCs generated from a CML patient primed autologous T cells effectively so that these T cells proliferated or produced IFN-γ and GM-CSF upon co-culture with their own leukaemia cells.

We used the patient's peripheral blood adherent cells as the source of DCs in our study. According to Osman et al (1999), DCs generated from peripheral blood adherent cells had an equivalent or superior activity in priming T lymphocytes, compared with DCs isolated by immunomagnetic separation. In addition, the adherent cell-derived DCs are easier to obtain and clinically more applicable. We did not enrich mature DCs further after their generation from adherent cells in PBMCs (thus, they may be present at different maturation stages). After culture with GM-CSF and IL-4, followed by maturation with TNF-α, the resulting cells consist of weakly or non-adherent DCs and an adherent cell population of flat cells. We did not separate the former for use as APCs, but used the entire mixture in situ, because we did not want to destroy a potentially beneficial microenvironment created by these cells in which in vitro immunity could be developed. Recently, Fujii et al (1999) noted, in their report on the induction of cytotoxic T lymphocytes (CTLs) against acute myeloid leukaemia (AML) cells using autologous DCs generated from CD34+ haematopoietic progenitor cells, that DC clusters seemed to be essential for developing T-cell responses against target cells. They found that stromal cell-like cells forming the foundation of the clusters may efficiently support development of DC for antigen uptake and presentation (Fujii et al, 1999).

In our experiments, sensitization against leukaemic DCs alone gave rise to specific cytokine-secreting cells (TL-DC3-0), but these cells were unable to undergo autocrine proliferation. This is a very unusual phenotype for CD4+ T cells, probably reflecting the lack of ability to secrete autocrine growth factors, the most common of which is IL-2. It is presently unclear why these cells do not proliferate, but lack of IL-2 production is a hallmark of anergic cells or tolerized cells. It is conceivable that sensitization of T cells by CML-DCs alone may represent a tolerogenic signal; alternatively, the presence of a small fraction of CD8+ cells in this line may result in suppressive interactions. This needs to be investigated by depleting the CD8+ cells or by cloning experiments. However, separation of CD4+ cells by cloning resulted in loss of the line and no clones were obtained. An interaction between the different cell populations in this line may therefore be necessary to maintain growth.

In contrast to TL-DC3-0, sensitization against leukaemic DCs pulsed with tumour cell lysate gave rise to two types of T cell lines: those unable to proliferate but which did secrete cytokine (TL-DC3-2), and, most importantly, those which could proliferate in an autocrine fashion when challenged by autologous CML cells (TL-DC3-1), as above. Interestingly, like TL-DC3-0, this line also consisted of a mixture of CD4+ and CD8+ cells, but with far more of the latter. The role of distinct lymphocyte populations in the control of CML remains poorly defined; however, the nature of anti-CML immunity probably varies with the stage of disease, as well as from one patient to another. Although it is well established that CD8+ CTLs play an important role in the suppression of cancer cell growth, recent evidence indicates that CD4+ T cells are an equally critical component of the anti-tumour immune response (Pardoll & Topalian, 1998). Giralt et al (1995) had demonstrated that infusing donor lymphocytes depleted of CD8+ T cells could induce remissions with a low rate of graft-vs.-host disease (GvHD) in patients with CML who had relapsed after bone marrow transplantation (BMT), suggesting that CD4+, but not CD8+ T lymphocytes, are essential for a graft-vs.-leukaemia (GvL) effect in patients with CML. Successful immunity to cancer will therefore require activation of tumour-specific CD4+ T cells. MHC class I-restricted cytotoxic lymphocytes can recognize and destroy tumour cells in vitro and in vivo, but usually require CD4+ MHC class II-restricted help for optimal responses. Because CD4+ T cells, particularly those of the Th1 phenotype, are commonly cytotoxic, they may represent the cells of choice for adoptive immunotherapy of class II-bearing CML (Pawelec et al, 2000). The line TL-DC3-1 may therefore represent an optimal reagent for immunotherapy.

We found that in vitro stimulation of tumour cell-depleted PBMCs from CML patient 3 by tumour cell lysate-pulsed DCs also resulted in the generation of specific GM-CSF-producing T-cell lines. This finding is of particular interest because previous studies in mice and humans suggested a correlation between tumour-specific GM-CSF secretion by TIL or tumour-draining lymph node cells and their anti-tumour therapeutic efficacy in vivo (Aruga et al, 1997). These latter investigators also showed that mixing of the two purified T-cell populations enhanced the anti-tumour activity of the CD8+ T cells in vivo. Attempts to clone the different T-cell lines obtained in the present study were successful only with TL-DC3-2, the line secreting the greatest amounts of GM-CSF. This cytokine can lead to the recruitment and activation of other lymphocytes, monocytes, neutrophils and DCs, provide critical signalling to other effector cells and help in the generation of optimal CD8+ T-cell cytotoxic responses. A majority of the GM-CSF-secreting clones also produced large amounts of the Th1 cytokine IFN-γ. According to Reuben et al (2000), the synthesis of Th1 cytokines in CML patients during remission was accompanied by normalization in the percentage and number of T cells and restoration of their functions. This type of line may therefore also be useful for immunotherapy.

In vitro studies have demonstrated that synthetic bcr–abl chimaeric peptides can bind to certain HLA class I and II alleles on APCa, and T cells can be generated that recognize targets preincubated with these peptides. However, recognition of native tumour cells in the absence of exogenous peptide has only been demonstrated in a minority of cases and reports are conflicting on whether CTLs can lyse leukaemic cells displaying endogenous bcr–abl peptides on their MHC molecules, and whether Th cell recognition occurs. We report here the isolation of T cells that proliferated or produced IFN-γ and GM-CSF upon co-culture with autologous leukaemia cells, but not HLA-identical sibling cells, suggesting that in vitro primed T cells recognized the same epitopes on the fresh leukaemia cells that had been presented by immunogenic dendritic cells. This approach includes a broad range of polyclonal T cells reactive with autologous leukaemic cells without the need to identify the exact leukaemia antigen involved in the T-cell response.

Acknowledgments

This work was supported by the Mildred Scheel Foundation (Deutsche Krebshilfe) grant number 10–1173-Pa3 and the Dieter Schlag Foundation. We thank patients who provided blood samples.

Ancillary