I.W.A. and S.K. contributed equally to this work
T cells raised against allogeneic HLA-A2/CD20 kill primary follicular lymphoma and acute lymphoblastic leukemia cells†
Article first published online: 12 AUG 2011
Copyright © 2011 UICC
International Journal of Cancer
Volume 130, Issue 8, pages 1821–1832, 15 April 2012
How to Cite
Abrahamsen, I. W., Kjellevoll, S., Greve-Isdahl, M., Mensali, N., Wälchli, S., Kumari, S., Loland, B. F., Egeland, T., Kolstad, A. and Olweus, J. (2012), T cells raised against allogeneic HLA-A2/CD20 kill primary follicular lymphoma and acute lymphoblastic leukemia cells. Int. J. Cancer, 130: 1821–1832. doi: 10.1002/ijc.26209
Conflict of interest: None
- Issue published online: 22 FEB 2012
- Article first published online: 12 AUG 2011
- Accepted manuscript online: 31 MAY 2011 08:21AM EST
- Manuscript Accepted: 10 MAY 2011
- Manuscript Received: 22 OCT 2010
- University of Oslo
- The Norwegian Research Council
- Oslo University Hospital (OUH)
- Department of Pediatrics
- Health Region South-East
- ØA Stiftelsen
- Grøseth and Moen Stjern families
- follicular lymphoma;
- acute lymphoblastic leukemia;
T cells mediating a graft-versus-leukemia/lymphoma effects without causing graft-versus-host disease would greatly improve the safety and applicability of hematopoietic stem cell transplantation. We recently demonstrated that highly peptide- and HLA-specific T cells can readily be generated against allogeneic HLA-A*02:01 in complex with a peptide from the B cell-restricted protein CD20. Here, we show that such CD20-specific T cells can easily be induced from naïve precursors in cord blood, demonstrating that they do not represent cross-reactive memory cells. The cells displayed high avidity and mediated potent cytotoxic effects on cells from patients with the CD20pos B cell malignancies follicular lymphoma (FL) and acute lymphoblastic leukemia (ALL). However, the cytotoxicity was consistently lower for cells from two of the ALL patients. The ALL cells that were less efficiently killed did not display lower surface expression of CD20 or HLA-A*02:01, or mutations in the CD20 sequence. Peptide pulsing fully restored the levels of cytotoxicity, indicating that they are indeed susceptible to T cell-mediated killing. Adoptive transfer of CD20-specific T cells to an HLA-A*02:01pos patient requires an HLA-A*02:01neg, but otherwise HLA identical, donor. A search clarified that donors meeting these criteria can be readily identified even for patients with rare haplotypes. The results bear further promise for the clinical utility of CD20-specific T cells in B cell malignancies.
In HLA-matched allogeneic hematopoietic stem cell transplantation (AHSCT), donor T cells may recognize self-HLA presenting so-called minor histocompatibility antigens, representing peptides from polymorphic proteins for which donor and host are disparate.1 This is known as the graft-versus-leukemia/lymphoma (GVL) effect.2 Several polymorphisms have successfully been identified in proteins with cell type-restricted expression that bear promise as therapeutic targets.3, 4 This knowledge may be exploited to target T cell reactivity to leukemia. Identification of polymorphic and immunogenic peptides present in large fractions of the population that are also restricted by frequent HLA molecules is, however, likely to be a tremendous task.
An attractive alternative to achieve specific targeting of leukemia cells is to exploit the immunogenicity of foreign HLA molecules.5 To this end, we have developed a novel protocol to induce allo-reactive T cells specific for peptides derived from cell type–specific self-proteins.6 Using dendritic cells engineered to express a foreign human leukocyte antigen molecule (HLA-A*02:01) in complex with a self-peptide, peptide-specific T cells are generated from HLA-A*02:01neg donors.6, 7 Because of negative selection in the thymus, the T cell repertoire is normally depleted of T cells reactive with self-peptides. In contrast, any peptide is a potential immunogen in the context of a foreign HLA molecule.8 As most tumor antigens are overexpressed self-proteins, this approach increases the number of potential therapeutic targets and eliminates the need to identify tumor-specific antigens. The possibility that cell type–specific proteins can be targeted provides the potential to achieve specific killing of hematopoietic cells in the absence of graft-versus-host disease (GVHD), which is the main cause of transplant-related mortality and morbidity.9, 10 This approach could provide a safer and more effective alternative to a standard AHSCT.
The B cell-restricted protein CD20 represents a well-documented example of a self-antigen that can be successfully targeted to kill malignant cells. In FL, treatment with anti-CD20 antibodies (rituximab) has drastically improved the therapeutic outcome. In combination with chemotherapy, the estimated benefit in terms of risk reduction (hazard ratio) is 0.62 for disease control and 0.65 for mortality.11 Rituximab has a low toxicity and is well tolerated, but the treatment is not curative. In ALL, the effect of rituximab treatment is still unclear, and it is not part of the standard therapeutic regime.12, 13 In both diseases, there is an indisputable need for improved therapeutic options.
In a previous report,7 we demonstrated that allo-reactive T cells reactive to HLA-A*02:01 in complex with a peptide derived from CD20 (hereafter referred to as A2/CD20-specific T cells) could be readily generated from all donors tested. The allo-reactive T cells were furthermore highly peptide and HLA specific, and potently killed primary leukemia cells from patients with chronic lymphocytic leukemia (CLL). In our current study, we show that allo-reactive A2/CD20-specific T cells efficiently kill tumor cells from adult patients with FL and pediatric patients with ALL, and that the T cells are of high avidity and can be generated from naïve T cells. The results show promise for the therapeutic use of CD20-targeted T cells in these B cell malignancies.
Material and Methods
Patients, healthy donors and cell lines
Our study was approved by the Regional Ethics Committee and performed in accordance with the Declaration of Helsinki. After informed, written consent, patients with ALL donated bone marrow, lymphoma patients donated an affected lymph node and the CLL patients and healthy donors donated blood. Human umbilical cord blood was obtained from the maternity ward at Oslo University Hospital Rikshospitalet, Oslo, Norway, after informed, written consent. Bone marrow and blood mononuclear cells and lymphoma cells in single-cell suspension were HLA typed using an HLA-A*02:01 antibody (AbD Serotec, Düsseldorf, Germany) and cryopreserved. The patient characteristics are given in Table 1.
The acute myeloid leukemia cell line THP-1 (HLA-A*02:01pos, CD20neg) was obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen. Epstein-Barr virus-lymphoblastoid cell lines (EBV-LCL) were generated by standard procedures. All cell lines were maintained in RPMI 1640 (Gibco, Invitrogen, Paisley, UK) containing penicillin/streptomycin (50 U/ml/50 μg/ml, Sigma-Aldrich, St. Louis, MO) and 10% fetal calf serum (Gibco).
Generation of monocyte-derived dendritic cells
Monocyte-derived dendritic cells (moDCs) from HLA-A*02:01neg individuals were generated as previously described.6 Briefly, plastic adherence of peripheral blood mononuclear cells (PBMCs) and 6 days culture in CellGro DC medium (CellGenix, Freiburg, Germany) supplemented with IL-4 (500 U/ml, PeproTech EC, London, UK), and GM-CSF (800 U/ml, Berlex Laboratories, Richmond, CA) was followed by 24-hr maturation in TNF-α (2 ng/ml, PeproTech) and lipopolysaccharide (10 ng/ml, Sigma-Aldrich). In one experiment, CD25high cells were depleted before adherence using CD25 Dynabeads (Dynal, Invitrogen).
mRNA transfection with HLA-A*02:01
HLA-A*02:01 mRNA was generated and transfected into moDCs by electroporation (1,250 V/cm for 3 msec), using a BTX ECM 830 square wave electroporator (BTX, Harvard Apparatus, Holliston, MA), as previously described.6 EBV-LCLs were electroporated at 1,250 V/cm for 1 msec.
Peptides and HLA multimers
The peptide SLFLGILSV (CD20188–196), first described in Ref.14, and the corresponding HLA-A*02:01 multimer and the NLVPMVATV/HLA-A*02:01 multimer (human cytomegalovirus pp65495–504) (referred to as A2/CD20 multimer and control multimer, respectively) were synthesized by Proimmune (Oxford, UK) or by the authors, as previously described.7 Multimers were labeled with fluorotags conjugated to APC or PE, as recommended by the manufacturer.
Induction, sorting and expansion of allo-reactive T cell lines
HLA-A*02:01-transfected moDCs were pulsed with 50 μg/ml peptide for 4 hr, irradiated (25 Gy), washed and cocultured with autologous, nonadherent cells at a ratio of 1:10 in CellGro DC medium containing IL-7 (10 ng/ml, PeproTech), IL-12 (50 pg/ml, PeproTech) and 1 μg/ml purified protein derivative-1 (a kind gift from Prof. L. Sollid's laboratory, Oslo University Hospital Rikshospitalet). Cells were restimulated on day 12 with HLA-A*02:01-transfected, peptide-pulsed autologous EBV-LCL in X-Vivo 20 (Lonza, Verviers, Belgium) with 5% pooled human serum and IL-2 (10 U/ml on day 14, R&D systems, Minneapolis, MN). On day 19, cells were stained with A2/CD20 multimer, fluorotag and anti-CD8, and A2/CD20 multimer reactive cells were sorted by flow cytometry (FACS Aria Cell sorter, BD Biosciences, San Jose, CA) or by magnetic beads to a purity of >98%. Multimer reactive cells labeled with fluorotag conjugated to PE were isolated using MACS anti-PE MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) followed by separation over two subsequent MACS MS Columns (Miltenyi Biotec).
Sorted, multimer reactive cells were expanded in the presence of irradiated, allogeneic PBMCs in X-Vivo 20 with 5% pooled human serum, 50 U/ml IL-2, 2 ng/ml IL-15 (PeproTech) and 1 μg/ml phytohemagglutinin (Remel, Lenexa, KS). The cells were restimulated weekly and supplemented with fresh medium containing IL-2, IL-15 and phytohemagglutinin when needed.
Priming of naïve allo-reactive T cells specific for CD20 and HLA-A*02:01
HLA-A*02:01pos, peptide-pulsed moDCs generated from human buffy coats (obtained from the blood bank, Oslo University Hospital Ullevål, Oslo, Norway) were cocultured with nonadherent PBMCs from cord blood of three HLA-A*02:01neg donors, either unsorted or sorted as naïve CD8posCCR7posCD45ROneg CD62Lpos T cells, at a ratio of 2:1, and induction of A2/CD20 multimer reactive cells was performed as described above. When sorted CD8pos naïve T cells were used, equal numbers of CD4pos T cells were sorted from the same donor and added to the coculture to provide help.
Antibodies and flow cytometry
The following antibodies were used: CD3 FITC, CD3PerCP-Cy5.5, CCR7 PE, CD8 PerCP-Cy5.5, CD45RO FITC, CD62L APC, CD19 APC, CD20 PE and interferon-γ (IFN-γ) PE (all from BD Biosciences). The antibodies to HLA-A*02:01 (Serotec), CD107a, CD107b (BD Biosciences) and CD8 (Diatec Monoclonals, Oslo, Norway) were conjugated to fluorescein and/or Alexa 647 and/or Pacific Blue (Molecular Probes, Eugene, OR) by the authors. Cells were analyzed on a FACS Canto II cytometer (BD Biosciences), and data analysis was performed using FACS DiVa (BD Biosciences) and FlowJo (Tree Star, Ashland, OR) softwares.
Functional T cell responses (production of IFN-γ and degranulation as determined by staining with anti-CD107a/CD107b) were measured as previously described6 at effector:target (E/T) ratios of 2:1 or 1:2. For measurements of lysis of primary leukemia/lymphoma cells, the cells were incubated for 2–3 hr in X-Vivo 15 (Lonza) at E/T ratios of 5:1 or 10:1 in 96-well plates, harvested and stained for surface markers, Annexin V FITC (BD Biosciences) and Propidium iodide (Sigma-Aldrich), according to the manufacturer's protocols. CD4pos T cells were isolated from PBMC using a CD4-positive isolation kit (Dynal). Cells within the live scatter gate, and staining negatively for Annexin V and Propidium iodide, were considered live and were quantitated by adding a fixed number of fluorescent beads (BD Biosciences). Each assay was performed in triplicates. The following formula was used to calculate lysis: % specific lysis = 1 − (live target cells in wells with responders/live targets in control wells).
Total RNA was prepared from 106 cells using Absolutely RNA Miniprep kit (Stratagene, La Jolla, CA) following instructions from the manufacturer. cDNA preparation was performed using 1 μg of RNA. SuperScript™ III Reverse Transcriptase and oligodT primers (Invitrogen, Carlsbad, CA) were used to perform reverse transcription. The reaction was run at 50°C for 1 hr. Full-length CD20 cDNA was amplified with specific primers: 5′-CAC CAT GAC AAC ACC CAG AAA TTC AG-3′ and 5′-ATA CTC GAG TCA TTA AGG AGA GCT GTC ATT TTC-3′. Pfx polymerase was used for 25 cycles (1′ 94°C, 1′ 50°C and 1′15″ 68°C). The amplicon was directionally cloned into pENTR-TOPO vector (Invitrogen). Three different kanamycin-resistant bacteria colonies were prepared and sent for bidirectional sequencing (Eurofins MWG Operon, Ebersberg, Germany).
A2/CD20-reactive T cells derive from naïve T cells
We previously demonstrated that A2/CD20-specific T cells are rapidly induced from HLA-A*02:01neg donors.7 In five of six donors, multimer reactive T cells were detected on day 12 of coculture with dendritic cells.7 The rapid generation could indicate that the T cells were expanded in vivo against a pathogen presented on an HLA molecule shared by the donors, and that the reactivity with A2/CD20 represented coincident cross reactivity. However, as the donors did not share any single HLA allele, this explanation was unlikely. Nevertheless, we next wanted to specifically investigate if the A2/CD20-specific T cells could be generated from naïve T cells. If so, their therapeutic potential would also increase.15 To this end, we used human umbilical cord blood cells, which contain high frequencies of naïve, unprimed T cells with a typically naïve phenotype (CCR7+, CD45RA+ and CD62L+).16 The results shown in Figure 1a demonstrate that A2/CD20 multimer reactive T cells were readily generated from two different HLA-A*02:01neg cord blood donors after 19 days of coculture with moDCs and EBV-LCLs expressing A2/CD20. The frequencies of multimer reactive cells were similar to those achieved previously by culture of PBMC T cells from adult healthy donors.7 Highly similar results were furthermore obtained when these experiments were repeated starting with sorted CD8posCCR7posCD45ROneg CD62Lpos T cells from cord blood, excluding the presence of memory T cells (Figs. 1b and 1c).
A2/CD20-specific T cell lines show high avidity for their cognate HLA/peptide complex
The avidity of CTLs used for adoptive cellular therapy is essential to achieve clinical responses.17, 18 Thus, we next investigated the sensitivity of the A2/CD20-specific T cells to peptide stimulation. T cell lines reactive with A2/CD20 were generated from two HLA-A*02:01neg donors according to our previously described approach.7 The peptide specificity of the cell lines was confirmed by coculture with two types of HLA-A*02:01pos CD20neg target cells: allogeneic CD4+ T cells from healthy donors and the acute myelomonocytic leukemia cell line THP-1. CTL responses to these target cells in the absence or presence of pulsing with the CD20-peptide (CD20p) were determined (Fig. 2). Although the CTLs showed negligible responses to unpulsed CD4+ T cells, a strong response was induced after peptide pulsing. The CTL response to peptide-pulsed THP-1 cells was even higher; however, a low background response to unpulsed THP-1 cells was also seen. When analyzing the side scatter high THP-1 cells only, we found that these cells stained positively for antibodies to CD107a/b and IFN-γ. This could be explained by the fact that THP-1 cells are monocytic of origin and express Fc receptors capable of capturing the fluorescent antibodies, and in addition hold large quantities of endogenous CD107/a/b.19 When analyzing for CTL responses, we gated on CD8-positive cells. However, consistent with data showing that monocytes may express CD8α,20 THP-1 cells also expressed quite high levels of CD8. Thus, even if using strict gates on scatter and CD8, we cannot exclude that a low fraction of the THP-1 cells remain amongst the analyzed cells, which could explain the background staining (data not shown).
Next, the avidity of the two A2/CD20-specific T cell lines was measured as production of IFN-γ and degranulation (CD107a/b) in response to coculture with HLA-A*02:01pos CD4+ T cells that were loaded with decreasing concentrations of CD20p and subsequently washed. Figure 2b shows that the peptide was recognized at concentrations as low as 100 pM, indicating high avidity.
Primary follicular lymphoma cells are efficiently killed by A2/CD20-specific T cells
The potential of the A2/CD20-specific T cells to kill tumor cells from patients with FL was investigated. Two patients (FL 1 and FL 2) were HLA-A*02:01pos, whereas the third (FL 3) was HLA-A*02:01neg. Cells from all three patients expressed CD20 (Fig. 3a). T cells specific for A2/CD20 were cocultured with FL cells for 3 hr at low E/T ratios of 5:1 or 10:1. Both T cell lines efficiently killed the HLA-A*02:01pos lymphoma cells (mean 59%, range 37–82%, n = 2), but spared those being HLA-A*02:01neg. Furthermore, CD4+ T cells were unaffected, unless pulsed with the CD20p (Fig. 3b).
A2/CD20-specific T cells kill primary acute lymphoblastic leukemia cells
Next, the ability of the A2/CD20-specific T cells to lyse cells from patients with ALL was tested. Patient tumor samples with detectable surface expression of CD20 were selected, but the levels were generally low and heterogeneous (Fig. 4a). Four patient tumor samples were excluded, as less than 4% of the cells expressed CD20.
ALL cells are known to express low levels of costimulatory molecules, and in some studies agents such as CD40L and IFN-γ are added to increase the expression of costimulatory molecules.21, 22 This was avoided here, to mimic a situation of adoptive T cell therapy as closely as possible. Still, even at low E/T ratios of 5:1 or 10:1, leukemia cells from patients ALL 3 and ALL 14 were effectively killed by both T cell lines. For ALL 3, the average specific lysis was 34% (range 15–53%, n = 4), and for ALL 14 35% (range 29–75%, n = 4) (Figs. 4b and 4c). At the E/T ratios tested, leukemia cells from patients ALL 20 and ALL 21 were killed by the T cell line from donor 1 only (for ALL 20 the average was 26%, range 17–35%, n = 2; for ALL 21 the average was 24%, range 16–32%, n = 2, Fig. 4b). The tumor cells from these four patients all expressed HLA-A*02:01, albeit patients 20 and 21 at somewhat lower levels than ALL 3 and 14 (Fig. 4a). However, neither the difference in HLA-A*02:01 expression nor a variable susceptibility to perforin/granzyme B-mediated killing was likely to explain the differential cytotoxicity, as pulsing of ALL 20, 21 and 14 with high concentrations of CD20p (10 μM) increased killing to an average of 91% with both T cell lines (range 82–97%, n = 1 for each T cell line, n = 2 total, Fig. 4d).
A potential alternative explanation could be differential expression of CD20. However, among the HLA-A*02:01pos donors, the highest expression of CD20 was seen on ALL 20 (Fig. 4a). By sequence analysis, we excluded that this particular patient had a mutation in the CD20 sequence. It is thus possible that differences in antigen processing between the patient cancer cells lead to variations in the amounts and types of peptides presented on the surface. The cells from the HLA-A*02:01neg patients ALL 5 and ALL 6 were not killed. Finally, there were no cytotoxic effects on unpulsed CD4+ T cells, whereas pulsing with 10 μM peptide leads to uniformly high levels of killing (Figs. 4b and 4c).
The observed negative lysis (Figs. 3 and 4) of target cells lacking expression of either HLA-A*02:01 or CD20 is likely a result of a generally improved target cell survival during coculture with T cells, as lysis was determined by the formula 1 − (live target cells in wells with T cells/live targets in control wells). Such an improved survival could be caused by membrane-bound or secreted T cell factors. This could suggest that the levels of killing indeed were even higher than reported. We have previously demonstrated efficient killing of CLL cells by these CTLs,7 and cells from patient CLL 112 in the previous study were included as a positive control in the present assays.
HLA-A*02:01neg but otherwise HLA-matched donors can be identified even for patients with rare haplotypes
A potential use of allo-reactive A2/CD20-specific T cells in therapy of B cell malignancies could be adoptive transfer to HLA-A*02:01pos patients, either following a lymphodepleting conditioning regimen or in the setting of AHSCT combined with a T cell-depleted hematopoietic stem cell graft from the same HLA-A*02:01neg donor. Except for the mismatch on HLA-A*02:01, the patient and donor should be HLA identical, implying that the donor must be homozygous for the second HLA-A allele of the patient. A single HLA mismatch is permitted in the standard transplant protocols in use today.23 However, with such requirements the availability of suitable hematopoietic stem cell donors might represent a limitation. We thus explored the feasibility of identifying donors that are HLA-A*02:01neg but otherwise HLA identical with the patient on the HLA loci A, B, C, DRB1 and DQB1. The results in Table 2 show that such unidirectionally mismatched donors could be identified even for patients with rare haplotypes.
We have recently shown that T cells restricted by allogeneic HLA-A*02:01 and specific for a peptide derived from CD20, exert potent cytotoxic effects on CLL tumor cells.7 In our current study, the aim was to further investigate the applicability of these A2/CD20-specific CTLs in the clinic. To this end, we demonstrate that the A2/CD20-specific T cells (i) can be generated from naïve T cells in cord blood, (ii) show high avidity, (iii) have potent cytotoxic effects on patient-derived FL and ALL tumor cells, and (iv) that donors with a single, unidirectional mismatch on HLA-A*02:01 can be readily identified for HLA-A*02:01-positive patients.
Cellular immunotherapy is dependent on in vivo persistence of the transferred cells. Several studies have shown the close kinetics between detection of the transferred cells and antitumor effects.24–26 We have previously shown that our cells have an early effector memory phenotype even after many weeks of in vitro expansion.7 Here, we show that the A2/CD20-specific T cells can be generated from a repertoire of naïve T lymphocytes. This is beneficial in the setting of adoptive cellular therapy, as the antitumor effects of effector memory cells generated from naïve cells are superior to those derived from central memory cells.15 The results also confirm that the CTLs do not represent cross-reactive memory cells generated against a pathogen.
To be efficacious in cancer immunotherapy, T cells must bind their cognate antigen with high affinity.17, 18 Most tumor antigens represent self-antigens, and a common explanation for the low efficiencies of cancer vaccines is that the responding T cells display low avidity because of thymic deletion of high-avidity self-reactive T cells. For this reason, A2/CD20-specific T cells cannot be easily generated from HLA-A*02:01pos donors.7 Thus, it was not possible to compare the avidity of CD20-specific T cells generated against allogeneic and autologous HLA-A*02:01. However, the group of Schendel recently demonstrated that allo-reactive T cells show 100- to 1,000-fold higher avidities compared to autologous T cells recognizing the same HLA/peptide complex.27 By measuring IFN-γ release after stimulation with peptide-pulsed T2 cells, they showed that allo-reactive T cells responded to peptide concentrations as low as 10−10 M. This is comparable to the high average avidities demonstrated for the A2/CD20-specific T cell lines generated here and further indicates their usefulness in immunotherapy.
The A2/CD20-specific T cells showed efficient cytotoxicity of FL cells. Although a limited number of patient tumor cells were studied, our results support the emerging clinical evidence that FL cells are susceptible to T cell-mediated killing, as seen in AHSCT.28, 29 However, the use of AHSCT in FL is still considered experimental. Although reduced-intensity conditioning is used, treatment-related mortality is 15–20% and the incidence of chronic GVHD 20–60%. To be eligible for this treatment, patients must therefore have an excellent performance status and minimal comorbidity.30 In contrast, the number of individuals eligible for a targeted CTL-based therapy could be greatly increased in this often heavily pretreated and elderly, patient group. Although, the prognosis for patients with FL has been greatly improved by the use of therapeutic anti-CD20 antibodies,31 50% of the patients are resistant to treatment and the large majority ultimately relapse.32 The resistance is usually not related to low expression of CD20 on the tumor cells, but rather to limited susceptibility for antibody-dependent cytotoxicity.32 As T cells mediate killing mainly through the perforin/granzyme B pathway and Fas/FasL interactions,33 T cell-based therapies could provide a promising alternative or supplement for these patients.
The A2/CD20-specific T cell line from donor 1 lysed HLA-A*02:01pos CD20pos samples from all four ALL patients, whereas the T cell line from donor 2 effectively lysed cells from donors 3 and 14 only. In concert with this, CTLs from donor 1 lysed cells from patients 20 and 21 with far lower efficiencies than those from patients 3 and 14. Clinical data from AHSCT of ALL show that although a GVL effect can be demonstrated, it is not as strong as in CLL or FL, possibly suggesting a lower susceptibility to T cell-mediated killing.34 The ALL cells that were poorly lysed (patients 20 and 21) both expressed HLA-A*02:01 and CD20, and peptide loading resulted in very efficient killing. Mutations in the CD20 sequence were not found. Thus, one possible mechanism is that the CD20 peptide is processed and presented at lower levels in these tumor cells. Another possibility relates to the threshold of effector cell activation. Studies suggest that in the presence of sufficient costimulatory molecules, a few copies of the cognate peptide will activate effector T cells.35 Cardoso et al.36 have shown that ALL cells generally express low levels of costimulatory molecules, and we hypothesize that the addition of high concentrations of peptide in our experiments reduced the need for costimulatory signals and led to tumor cell killing. Further studies would be needed to clarify this.
As there is still a mortality of 10–20% in childhood ALL, improved treatment options are warranted. However, adult ALL patients have a significantly worse prognosis, with a mortality of 50–70%. The major reasons for the relatively more favorable prognosis in children are better tolerance for high-intensity chemotherapy and a lower prevalence of high-risk disease.37 The ALL tumor cells used in our experiments were from pediatric patients, and efficient cytotoxic effects were demonstrated on cells from ALL patient 3, a high-risk disease patient, NOPHO intensive treatment group. We therefore speculate that the effect on adult ALL tumor cells might be similar, and that interindividual differences, such as the expression of costimulatory molecules or the ability to process and present antigens, are more important than age-group effects. In fact, recognizing the similarities of pediatric and adult ALL, many Nordic institutions have started to treat selected adult patients with the current pediatric protocol NOPHO 2008. Regardless of the potential utility of the A2/CD20-specific CTLs in adult ALL, only ALL patients with tumor cells that express CD20 and are susceptible to killing by the A2/CD20-specific CTLs in vitro should be selected for a CTL-based therapy.
A potential clinical setting for the therapeutic use of the A2/CD20-specific CTLs in HLA-A*02:01pos lymphoma/leukemia patients could be to infuse isolated A2/CD20 multimer reactive CTLs. The cells could be isolated by reversible multimers38 and then directly infused, in a similar way as previously demonstrated for treatment of patients with cytomegalovirus (CMV)-specific T cells.39 Successful clinical use of this technology to treat CMV reactivation has so far been shown in seven patients, in which the majority has shown clearance of virus and none has developed side effects/GVHD (personal communication, L Germeroth IBA GMbH, Göttingen, Germany and as described by Schmitt et al.).40 Alternatively, the isolated CTLs could be expanded before infusion.
Fifty percent of the Caucasian population is HLA-A*02:01pos.41 Here, we demonstrate the feasibility of identifying donors that are HLA-A*02:01neg, but otherwise HLA identical with the patient, being homozygous for the second HLA-A allele of the recipient. In fact, donors could be identified even for patients with rare haplotypes. As the single HLA antigen mismatch would be in the desired donor-to-recipient direction only, the risk of rejection is significantly reduced and confined to mHags. This risk is probably small, as adoptively transferred allogeneic EBV-specific T cells that are mismatched on multiple HLA antigens are not rejected, and efficiently cure EBV-positive post-transplantation lymphoproliferative disease without GVHD.42, 43 Likewise, infusion of virus-specific CTLs from third-party donor mismatched at up to six antigens did not cause GVHD in 73 patients that were off immunosuppressive treatment at the time of T cell infusion.44 Even so, the patient could receive a standard lymphodepleting regimen, such as fludarabine/cyclophosphamide, to reduce tumor cell load and prevent rejection of the infused A2/CD20-specific allogeneic T cells. Lymphodepletion has been shown to greatly potentiate the efficacy of adoptive T cell therapy by creating space for the infused T cells and breaking tolerance.45 Additionally, pretreatment with rituximab, followed by a washout period, would probably be advantageous to prevent exhaustion of the infused T cells due to massive exposure to the CD20 antigen on normal B cells.46 In addition to eradicating malignant B cells, infused A2/CD20-specific CTLs might deplete the patient of normal B cells. However, experience with rituximab treatment has demonstrated that B cell depletion over prolonged periods can be tolerated without major safety hazards.47
Alternatively, the specific CTLs could be administered in conjunction with a T cell-depleted stem cell graft from the same (HLA-A*02:01neg) donor, giving rise to normal B lymphopoiesis. As these B cells would be HLA-A*02:01neg, the A2/CD20-specific CTLs would leave them unaffected. Recent studies show that the immunodeficiency frequently associated with T cell-depleted grafts can be improved by infusion of donor CD4+ T cells post-transplant.48
In conclusion, the results presented here demonstrate the potential use of allogeneic A2/CD20-specific T cells in the clinic. Follicular lymphoma cells and B cell acute lymphoblastic leukemia cells both express CD20, but represent quite different diseases. Follicular lymphoma patients with stage III/IV disease are considered incurable by standard therapies, and the disease often shows an indolent behavior over years. Hence, this subtype of non-Hodgkin's lymphoma would be suitable for clinical studies involving adoptive CTL treatment. In the case of more aggressive B cell malignancies like ALL, selecting patients with minimal residual disease might be more important. HLA-A2/CD20-specific CTLs could potentially form the basis for a new treatment option for both patient groups and possibly also for other B cell malignancies.
The authors thank R. Fet, G. Flatberg and K. Marshal for exemplary technical assistance. This study was financed by the University of Oslo (B.F.L., I.W.A., N.M.), The Norwegian Research Council (J.O. and S.K.), Medinnova (J.O.), Oslo University Hospital (OUH) (A.K., J.O., M.G.I. and T.E.), Department of Pediatrics, OUH (M.G.I.), Health Region South-East (S.W.) and ØA Stiftelsen (J.O.). The authors are grateful to the Grøseth and Moen Stjern families for inspiration. The authors also thank the National Marrow Donor Program Bioinformatics and funding Bone Marrow Donors Worldwide for the use of their databases.
- 40Adoptive transfer and selective reconstitution of streptamer-selected cytomegalovirus-specific CD8+ T cells leads to virus clearance in patients after allogeneic peripheral blood stem cell transplantation. Transfusion 2011; 51: 591–9., , , , , , , , , , , , et al.
- 48Haploidentical stem cell transplantation after a reduced-intensity conditioning regimen for the treatment of advanced hematologic malignancies: posttransplantation CD8-depleted donor lymphocyte infusions contribute to improve T-cell recovery. Blood 2009; 113: 4771–9., , , , , , , , , , , , et al.