Contribution of CD 39 to the immunosuppressive microenvironment of acute myeloid leukaemia at diagnosis

Acute myeloid leukaemia (AML) is a heterogeneous group of malignant haemopathies involving the defective differentiation of myeloid progenitors (Estey & Dohner, 2006). At diagnosis, AML patients show severe defects in controlling tumour growth. This is partly due to the immunosuppressive environment created by leukaemic blasts (Buggins et al, 2001). Several mechanisms have been implicated in this process, including the expansion of CD25FoxP3(Forkhead-box-P3) regulatory T-cells (Tregs) (Shenghui et al, 2011; Szczepanski et al, 2009). Indeed, a recent report suggested that the T-cell mediated immune response against AML could be predicted based on the composition of Treg subsets (Schick et al, 2013). Among the mechanisms used by Tregs to inhibit immune responses against cancers, increasing evidence points to a role for extracellular adenosine produced by the ectonucleotidases CD39 and CD73 (Beavis et al, 2012). This extracellular adenosine can bind to adenosine receptors, including the A2A receptor (A2AAR), which are expressed on various immune cells. Adenosine suppresses T and Natural Killer (NK) lymphocyte anti-tumour functions (Hausler et al, 2011). To assess the role of Treg subpopulations in creating an immunosuppressive environment in AML, we examined the CD25 and FoxP3 regulatory CD4 T-cell populations in the peripheral blood from 46 AML patients at diagnosis and 21 healthy controls (HCs) by flow-cytometry (Fig S1, Tables SI, SII). The percentage of CD4 cells among CD3 T-cells was comparable in patients and HCs (medians: 77 0% vs. 73 3%, respectively; NS). As in previous studies (Schick et al, 2013; Shenghui et al, 2011; Szczepanski et al, 2009), patients had an increased frequency of circulating CD25 cells (medians: 7 7% vs. 4 3%, respectively; P = 2 10 ) and FoxP3 CD4 T-cells (medians: 5 4% vs. 2 9%, P = 1 10 ) (Fig 1A). Compared to HC samples, all AML categories tested, except FAB M5, had a significantly higher proportion of CD25CD4 T-cells (Fig 1A). A comparable trend for FoxP3 expression was also observed. The increase in FoxP3CD4 Tregs prompted further analysis of the na€ıve CD45RAFoxP3 and memory CD45RAFoxP3 Treg subsets (Miyara et al, 2009) in 10 representative patients and 11 HC (Fig S2A and Table SIII). These samples showed no imbalance in percentages between na€ıve and memory Treg subsets, and both were CD25CD127 compared to conventional CD4 T-cells (Fig S2B). However, CD25 expression on FoxP3 Treg subsets was strongly reduced on na€ıve and memory Tregs in patients compared to controls (Fig 1B), although this expression remained higher than that of the conventional CD4 T-cells from the same patient (Fig S2 and Table SIII). The expression of homing receptors specific for skin (CLA and CCR4), intestine (CD49d, CD103), lymph nodes (CCR7) and inflamed tissues (CCR6) was then analysed on Treg subpopulations. Only CCR4, CCR6, CCR7 and CD49d were detectable on Tregs from both patients and HC (Figs 1C, S2C, Table SIII), and expression was significantly decreased on patients’ T-cells (both conventional and Treg populations). This suggests that all T-cell subpopulations in AML patients at diagnosis have an altered capacity to migrate into tissues. Because the phenotype of all T-cell populations was significantly affected in AML patients compared to HC, we investigated whether blast cells were involved in this remodelling. We examined CD39 and CD73 expression on Tregs and the corresponding AML cells, hypothesizing that both cells could generate an adenosine-enriched environment potentially affecting patient immunity. CD39 was detectable in all T-cell subsets in patients and HC, with higher expression levels on Tregs (Table SIII and Fig S2C). However, CD39 expression levels, but not CD39 Treg percentages, were decreased in patients compared to HC (Fig 2A). This suggests that CD39 function in Tregs could be maintained in AML patients. Finally, only a minority of Tregs in patients and HC expressed CD73 at very low levels. CD39 and CD73 expression was also analysed on CD45CD34 Lin tumour cells (Figs 2B and S3). CD39 was frequently detected on AML blasts, with high inter-patient variability. CD39 expression was equivalent on these cells compared to mature CD45CD34 Lin blood cell populations. In contrast, CD73 expression was generally lower on blasts than on mature populations (medians: 1 7% vs. 19 0%, respectively; P = 3 10 ), except in two patients who had a vast majority of CD73 leukaemic cells resulting in about 50% of tumour cells being positive for both ectonucleotidases. Ectonucleotidase function was then examined in AML cells by stimulating carboxyfluorescein succinimidyl ester-labelled peripheral blood lymphocytes from HC with a CD39CD73 HL60 AML cell line. The CD39-specific inhibitor, ARL67156 (Hausler et al, 2011), was added to some cultures. After a 5 day incubation, the proliferation index showed that proliferation of CD8 but not CD4 T-cells was weakly but consistently correspondence