Background
Acute myeloid leukemia (AML) is the most common form of acute leukemia in adults [
1] with only ~ 20% of patients expected to survive past 5 years after diagnosis [
2]. Despite significant advances in the field of AML pathophysiology, only a few novel therapies for AML have moved into the clinic for a subset of AML cases; consequently, AML relapse remains a significant issue that adversely impacts AML patient survival [
3‐
6]. Cytarabine (AraC) and daunorubicin (DNR) are conventional chemotherapy drugs widely used over the past three decades for induction therapy, which aims to eliminate the bulk of AML blasts by targeting rapidly proliferating cancer cells. Many AML patients achieve initial remission and will receive consolidation therapy, such as high-dose AraC in order to target the remaining AML blasts. Unfortunately, these therapeutic regimens are extremely intensive and toxic [
7‐
9], making them unfeasible for debilitated elderly patients. Despite the improved prognostic information obtained from identifying key cytogenetic and molecular abnormalities to help guide treatment selection, progress on new treatments has not advanced as much as our understanding of the factors that drive the disease [
10,
11]. Thus, new therapeutic strategies with lower toxicities are needed to effectively eliminate chemotherapy-resistant AML so as to improve patient survival.
Donor T cells can effectively target AML cells, as evidenced by the strong and curative graft-versus-leukemia effects following allogeneic hematopoietic stem cell transplantation (alloHSCT) or occasionally following donor lymphocyte infusions, which help prevent disease relapse and increase disease-free survival rate [
12‐
14]. T cell-based therapies have therefore been viewed as having potential in curing AML by targeting relapse-initiating AML. However, alloHSCT and donor lymphocyte infusions have a significant drawback in that they can also cause crippling graft versus host disease (GvHD), where the activity of donor cells against host cells is not limited to transformed cells [
13,
15].
Our lab was the first to identify CD4 and CD8 double negative T cells (DNTs) in mice [
16] and demonstrate the anti-leukemic effects of their
ex vivo expanded human counterpart
in vitro and in vivo [
17,
18]. We showed that
ex vivo expanded allogeneic human DNTs can selectively target AML cells, including those obtained from chemotherapy-resistant patients, without causing toxicity towards normal cells and tissues in an in vivo mouse model [
18]. Accordingly, a first-in-human phase I clinical trial using allogeneic DNTs to treat patients with high-risk AML has been initiated (NCT03027102). Although DNTs target a wide range of primary AML samples, blasts from approximately 22% of AML patients are not sensitive to DNT-mediated cytotoxicity
in vitro. Furthermore, administering DNTs as a stand-alone therapy is not curative in patient-derived xenograft models [
18].
Induction chemotherapy is administered to most AML patients with curative intent; there is increasing evidence that the cures are in part due to enhanced anti-tumor immune responses [
19‐
21]. Given this, it is reasonable to explore combining standard of care chemotherapy with immune-mediated killing. To the best of our knowledge, there are no reports of combining conventional chemotherapy with adoptive T cell therapy against AML in a xenograft model. Given that DNTs have the potential to be used as an off-the-shelf adjuvant cellular therapy due to their non-HLA-restricted, non-TCR-dependent mode of action [
18] and ability to broadly target AML cells from some, but not all chemotherapy-resistant patients, it is of interest to know whether conventional chemotherapy would increase the effectiveness of DNTs against chemotherapy-resistant forms of AML. Furthermore, since about 30% of AML patients do not respond to conventional chemotherapy and a significant portion of their AML cells can be targeted by DNTs [
18], it is important to know whether DNT therapy would be complementary to conventional chemotherapy to increase response rate and survival.
Methods
Human samples and cell lines
Human myeloid leukemia cell lines OCI-AML-2, OCI-AML-3, KG1a, and MV4–11 were obtained from ATCC. AML2 and AML3 were cultured in alpha-MEM supplemented with 10% fetal bovine serum (FBS), KG1a was cultured in RPMI-1640 supplemented with 10% FBS and MV4–11 was cultured in IMDM supplemented with 10% FBS. All cell lines were incubated at 37 °C in 5% CO2. Human blood samples were obtained from healthy adult donors and AML patients, respectively, after obtaining written informed consent and were used according to University Health Network (UHN) Research Ethics Board (05–0221-T) and NHLBI approved protocols. Peripheral blood mononuclear cells (PBMCs) from healthy donors (HDs) or AML patients were separated by Ficoll (GE Healthcare) density gradient. AML patient samples were viably frozen in 10% DMSO, 40% fetal calf serum (FCS) and alpha MEM at the Princess Margaret Leukemia Bank and stored in the vapor phase of liquid nitrogen until used.
Chemotherapy drugs and treatment
Chemotherapy drugs AraC and DNR (Sigma-Aldrich) were reconstituted in 0.2 μm filtered water and stored in aliquots at − 20 °C. Chemotherapy was added to target cells for 24 h, then incubated at 37 °C in 5% CO2. The cells were then washed with RPMI-1640 before use in experiments.
Ex vivo expansion of human DNTs
Peripheral blood samples were obtained from healthy donors under a UHN-REB approved protocol (05–0221-T). DNTs were enriched from the whole blood by using CD4 and CD8 RosetteSep depletion kits according to the manufacturer’s instructions (StemCell Technologies). The samples were then layered on Ficoll-Paque (GE Healthcare) and centrifuged at 1200 x g for 20 min. The enriched DNTs were expanded
ex vivo as described previously [
17]. DNTs from d12 to d20 of culture were used in experiments.
Flow cytometry
The following anti-human antibodies for staining of cell surface markers were used: CD3 (HIT3a), CD33 (WM53), CD45 (HI30), CD34 (561), CD112 (TX31), CD155 (SKII.4), MIC-A/B (6D4), Annexin V, and 7AAD, which were all purchased from BioLegend, and ULBP4 (709116) from R & D Systems. Data acquisition was performed using C6 Accuri (BD Biosciences), LSRII (BD Biosciences), or Attune NxT (ThermoFisher) flow cytometers and data were analyzed using FlowJo version 10.
Cytotoxicity assays and blocking experiments
The cytotoxic activity of DNTs was measured by a 2 h or 4 h flow-based killing assay. Target cells were labelled with PKH-26 (Sigma-Aldrich) according to the manufacturer’s instructions, and then co-incubated with DNTs at appropriate effector to target (E:T) ratios in U-bottom 96-well plates (Corning). Dead cells were identified as the PKH
+CD3
−AnnexinV
+ by flow cytometry. Gating strategies for patient leukemic blasts varied according to the phenotype of the AML cells. Percent specific killing was calculated using the formula:
$$ \% Specific\kern0.5em Killing\kern0.5em =\frac{\left(\%{AnnexinV}_{with\kern0.5em DNT}-\%{AnnexinV}_{Without\kern0.5em DNT}\right)}{\left(100\%-\%{AnnexinV}_{with out\kern0.5em DNT}\right)}\times 100\kern0.5em \% $$
Blocking antibodies for NKG2D and DNAM-1 (CD226), or the isotype control (BioLegend) were incubated with DNTs at a final concentration of 10 μg/mL for 30 min and washed before co-incubation with target cells.
Xenograft models
NOD.Cg-
Prkdc
scid
Il2rg
tm1Wjl
/SzJ (NSG) mice (Jackson Laboratories) were maintained at UHN animal facility in accordance with the guidelines of the Animal Care Committee of UHN and the Canadian Council on Animal Care. On day 0, 8- to 12-week-old female NSG mice were irradiated (225 cGy) and then injected with 4 × 10
6 KG1a cells intravenously (i.v.). On day 5, mice were administered a “5 + 3” chemotherapy regimen as described by Wunderlich et al. [
22], but at an adjusted lower dose (8 mg/kg AraC + 0.24 mg/kg DNR). 20 × 10
6 DNTs were then injected i.v. on days 12, 15, and 18. rIL2 (Proleukin, 10
4 IU/mouse) was administered i.v. along with DNT infusions and was also given intraperitoneally on days 21, 24, and 27. Mice were sacrificed 6 weeks after engraftment of KG1a and bone marrows were harvested and processed using standard techniques. Leukemic engraftment was determined by flow cytometry gating on the human CD45
+CD34
+ population.
Statistical analysis
Statistical analyses were performed using GraphPad Prism version 6 (San Diego, CA, USA). Data were expressed as means + standard deviation (SD). Two-tailed unpaired or paired Student t tests, one-way ANOVAs with Newman-Keul multiple comparisons test correction, and repeated measures ANOVAs with Holm-Sidak’s multiple comparisons test correction were performed, where appropriate, to identify significant differences between groups in our experiments.
Discussion
We previously demonstrated the feasibility of expanding therapeutic quality and quantity of DNTs and the capabilities of DNTs against AML, among other forms of leukemia and lymphoma [
18]. Herein we explored the use of DNTs in a combinatorial approach with conventional chemotherapy against chemotherapy-resistant AML. Using KG1a, an AML cell line that is resistant to NK cell lysis and chemotherapy [
24], and CD34
+ primary AML samples, which are resistant to apoptosis [
31], the data presented further supports the effectiveness of DNTs against therapy-resistant cells. Moreover, these results show that prior treatment with chemotherapy such as DNR sensitizes AML cells to DNT killing.
KG1a is resistant
in vitro to apoptosis induced either by chemotherapy (Figs.
1a and
b) or DNT-mediated cytotoxicity (Fig.
1c). We also found that the cell line is resistant to DNT therapy in vivo (Fig.
3b). However, our results show that engraftment of KG1a in the bone marrow can be significantly reduced by chemotherapy in vivo (Fig.
3b). This may be due to administering a 5-day regimen that consists of both chemotherapy drugs, compared to when single chemotherapy drugs were added for 24 h
in vitro. The pharmacokinetics and therapeutic effects of the drugs in vivo over a prolonged period would conceivably be different from a 24 h
in vitro treatment, and the combination of the two drugs may have additive or synergistic effects in targeting KG1a. A previous report also described similar characteristics of this cell line
in vitro [
24]. Importantly, despite the resistance of KG1a to conventional therapies relative to other AML lines, a greater anti-leukemic effect was observed both
in vitro (Fig.
2b) and in vivo (Fig.
3b) when we used a combinatorial approach. Furthermore, we examined, through cytotoxicity assays, the effectiveness of the combination therapy on CD34
+ primary AML samples. Most notably, we observed a significant increase in specific killing of these cells by DNTs after pre-treatment with DNR in approximately half (6/13) of the samples (Fig.
4b). Similar to what we demonstrated with KG1a (Fig.
2b), AraC pre-treatment did not elicit a sensitizing effect comparable to that of DNR (Figs.
4a and
c). These observations are in keeping with reports in the literature that the family of chemotherapy drugs encompassing DNR is known to elicit immunogenic cell death by calreticulin translocation and the release of high-mobility-group box 1 [
36].
AML is known to be an extremely heterogeneous disease; this is reflected in our finding that some of the primary AML samples appeared to be targeted more effectively by DNTs than others after chemotherapy (Figs.
4a and
b). The % specific killing calculation (see Additional file
2 Figure S2) takes into account the spontaneous and chemotherapy-induced cell death to ultimately determine the proportion of cells that are solely targeted by DNTs. Since we detected specific killing of all primary AML samples by DNTs, it is expected that combining DNTs and chemotherapy can target more AML cells than chemotherapy alone. Accordingly, we saw the effect of DNTs in reducing the proportion of viable AML blasts
in vitro after chemotherapy treatment (Figs.
2c and
d). Nevertheless, 1 out of 13 primary AML samples became significantly less sensitive to DNTs after AraC treatment (Fig.
4a). To circumvent the issue of potential antagonism between the two therapies, pre-screening patients after they have undergone chemotherapy to determine the sensitivity of their AML cells to DNTs
in vitro may help to stratify patient selection or regimen.
In our in vivo experiments, mice were administered a “5 + 3” chemotherapy regimen as described by Wunderlich et al. [
22], but at an adjusted lower dose (8 mg/kg AraC + 0.24 mg/kg DNR), which we established through titrating the drugs in vivo (see Additional File
6 Figure S4). Our in vivo studies demonstrated that DNT therapy alone was ineffective at reducing engraftment of KG1a. While low-dose chemotherapy treatment significantly reduced KG1a engraftment in the bone marrow, we observed an even greater reduction with the combination of DNT therapy and chemotherapy (Fig.
3b). In the clinic, almost all AML patients receive chemotherapy, which is effective in reducing the bulk of AML cells. Since our
in vitro and in vivo data indicate that chemotherapy can also prime the remaining AML blasts to be more susceptible to DNT-mediated cytotoxicity, it suggests that DNTs can be used as an adjuvant and administered shortly after chemotherapy in order to take advantage of the sensitizing effects of chemotherapy to eliminate chemotherapy-resistant residual AML cells. Based on our model using a reduced chemotherapy dose, which was 16% of the maximum tolerated dose in NSG mice [
22], perhaps a lower dose can be used in clinic as well when combined with DNT therapy, in the hope of reducing the various side-effects and toxicities of chemotherapy. This would greatly benefit elderly patients, who have much poorer prognosis than the rest of the population and have additional risk factors that prevent them from being eligible for remission style therapy [
37‐
39]. Additionally, there are current efforts by others to optimize conventional chemotherapy drug delivery in AML patients to reduce toxicities [
40], which have led to a phase III clinical trial of CPX-351, using a liposomal formulation of daunorubicin and cytarabine to treat elderly patients with high risk (secondary) AML (NCT01696084). The advent of newer technologies that can more efficiently administer chemotherapeutics to patients while avoiding side-effects can pave the way for more effective combination therapies.
Chemotherapeutic agents are known to influence our immune system in various ways [
19]. Specifically, chemotherapeutics can induce expression of various markers on the surface of cancer cells to facilitate their lysis by cytotoxic immune cells or induce the release of soluble factors that in turn stimulate immune responses [
20,
21]. There is also evidence that anthracyclines, a family of chemotherapy drugs that DNR is part of, have strong, immunogenic effects [
36]. The role of NKG2D and DNAM-1 receptor-ligand interactions in cell-based immunotherapies is well-described [
41,
42]. Likewise, the blocking experiments in this study demonstrated a role of NKG2D and DNAM-1 on DNTs in the targeting of chemotherapy-treated KG1a (Fig.
5c). We also observed the ability of DNR and, to a lesser extent, AraC, to increase the expression of NKG2D and DNAM-1 ligands in KG1a (Fig.
5a). PBMCs from healthy donors, however, did not express nor upregulate the ligands after chemotherapy pre-treatment (Fig.
5b). The blocking assay using anti-NKG2D and anti-DNAM-1 antibodies significantly reduced but did not fully abrogate the targeting of DNR-treated KG1a by DNTs (Fig.
5c), which suggests that other pathways may be involved. As there are many ways that chemotherapy drugs are able to influence the immune system and immune function [
19], future studies are required to explore the full range of their immunogenic effects so as to identify other mechanisms involved in the chemotherapy-induced susceptibility of AML cells to DNTs.