CD200 expression suppresses natural killer cell function and directly inhibits patient anti-tumor response in acute myeloid leukemia

Abstract

Upregulation of the immunosuppressive cell surface glycoprotein, CD200, is a common feature of acute myeloid leukemia (AML) and is associated with poor patient outcome. We investigated whether CD200 overexpression on AML cells could specifically compromise patient natural killer (NK) cell anti-tumor responses. We found that CD200^hi patients showed a 50% reduction in the frequency of activated NK cells (CD56^dimCD16⁺) compared with CD200^lo patients. Additionally, NK receptor expression (NKp44 and NKp46) on these cells was also significantly downregulated in CD200^hi patients. To assess whether NK cell activity was directly influenced by CD200 expression, we examined the effect of ectopic expression of CD200. These assays revealed that both NK cell cytolytic activity and interferon-γ response were significantly reduced toward CD200⁺ leukemic targets and that these targets showed increased survival compared with CD200⁻ cells. Similarly, NK cells isolated from AML patients were less functionally active toward CD200^hi autologous blasts from both cytolytic and immunoregulatory perspectives. Finally, blocking CD200 alone was sufficient to recover a significant proportion of NK cell cytolytic activity. Together, these findings provide the first evidence that CD200 has a direct and significant suppressive influence on NK cell activity in AML patients and may contribute to the increased relapse rate in CD200⁺ patients.

Introduction

CD200 is a trans-membrane glycoprotein belonging to the type-1 immunoglobulin superfamily.¹ In adults, CD200 is highly expressed in immune-privileged sites, such as the central nervous system, as well as leukocytes (including dendritic cells and T and B lymphocytes).² In both mice and humans, interaction of CD200 with its receptor, CD200R, which is expressed on immune competent cells² imparts an immunosuppressive signal leading to inhibition of macrophage function,^3,4 induction of regulatory T cells,⁵ switching of cytokine profiles from Th1 to Th2 and inhibition of tumor-specific T-cell immunity.⁶ Consistent with this, CD200-deficient mice are susceptible to tissue-specific autoimmunity.⁴

The overexpression of CD200 has been implicated in the pathogenesis of solid tumors^7,8 and hematological malignancies including acute myeloid leukemia (AML),⁹ lymphoma,¹⁰ chronic lymphocytic leukemia,¹¹ hairy cell leukemia¹² and myeloma.¹³ In addition, we have shown that CD200 upregulation in AML is a poor prognostic indicator in non-core binding factor leukemias.¹⁴ Recently, studies have demonstrated that expression of this protein is a common characteristic of cancer stem cells and is associated with tumor progression.^15,16 Furthermore, CD200 has a central role in immune tolerance by protecting critical tissues and stem cells from immune damage, a characteristic that may be exploited to minimize graft rejection through selection of stem cells that have high CD200 expression.^17,18 Therefore, these data are consistent with a hypothesis in which residual disease evades immune-recognition if CD200 is being expressed and indeed there is evidence that viruses encode CD200-type molecules as an immunoevasion strategy.¹⁹

In AML, there is evidence that a state of immunosuppression exists and that an anti-leukemia response can be effective in the treatment of residual disease.^20-22 Natural killer (NK) cells, are important immune cells that modulate the initial recognition and clearance of virus-infected and malignant cells through the release of cytolytic vesicles.^23-25 NK cells constitute approximately 10% of circulating lymphocytes in health and are identified generally as CD45^{+ +} CD19⁻ CD3⁻ CD56⁺ cells. Their activation and immunosurveillance is tightly regulated through a complex network of cytokines and a large and diverse repertoire of membrane receptors that deliver both inhibitory signals (such as NKG2A/CD94 and KIRs) and stimulatory signals (such as NKG2D and the natural cytotoxicity receptors (NCRs): NKp30, NKp44 and NKp46) (Lakshmikanth et al.²⁶; Hecht et al.²⁷). It is therefore unsurprising that defective NCR expression and NK cell dysfunction has been associated with poor patient outcome in many cancers, including AML.^28,29 Five distinct NK cell sub-populations have been identified based on expression of CD56 and CD16 (reviewed in Poli et al.,³⁰): (1) CD56^brightCD16⁻ (normally~15% of NK cells), (2) CD56^brightCD16⁺ (rare), (3) CD56^dimCD16⁻ (rare), (4) CD56^dimCD16⁺ (~80%) and (5) CD56⁻CD16⁺ (rare). However, the frequency of these populations and their activating receptor repertoire/cytolytic activity remains to be elucidated within AML and the effect of CD200 expression on these parameters is unknown.

Given the existing evidence that NK cell function influences AML blast clearance and long-term survival in AML, we investigated the possibility that CD200 expression may directly suppress anti-tumor immunity in this disease. We show that CD200^hi AML patients have a reduced frequency of CD56^dimCD16⁺ NK cells. Moreover, CD200^hi AML patients display an NK cell phenotype that differs from CD200^lo and are also dysfunctional in terms of activation and effector action. Further, our findings suggest that CD200 expression on leukemic blasts have an influential role in suppressing NK cell cytolytic activity making CD200 a potential therapeutic target for CD200^hi AML.

Materials and methods

Normal and AML patient sample materials

Peripheral blood or bone marrow samples were collected at diagnosis, before drug treatment and following informed consent from AML patients treated in the UK Medical Research Council and National Cancer Research Institute AML 10, 11, 12, 14 and 15 clinical trials in accordance with the 1964 Declaration of Helsinki. Fresh peripheral blood was also collected from healthy individuals with informed consent. AML blasts and peripheral blood mononuclear cells (PBMCs) were separated on a Ficoll–Hypaque density gradient (Sigma, Poole, Dorset, UK).

Cell surface immunophenotyping of AML patient blast and NK cell sub-populations

Multiparameter flow cytometry was performed in order to analyze CD200 protein expression on AML blasts and NK cell receptor repertoire from AML patients. All incubations were carried out for 30 min at 4 °C unless otherwise stated. Cells were washed twice with staining buffer (phosphate-buffered saline, 1 mm EDTA, 2% fetal calf serum (FCS)) before immunostaining with the following anti-human antibodies (purchased from BD Biosciences, Oxford, UK, unless otherwise stated). For CD200 analysis: CD200-phycoerythrin (PE), CD45-fluorescein isothiocyanate (FITC), CD34-PerCP.Cy5.5 (BioLegend, Cambridge, UK); for NCR analysis: CD45-APC.H7, CD19-AmCyan, CD3-Pacific Blue, CD56-PC7 (Beckman Coulter, High Wycombe, UK), CD16-PerCP.Cy5.5, NKG2D-FITC (AbCam, Cambridge, UK), NKp30-APC, NKp44-APC, NKp46-PE and CD200R-PE (BioLegend). Data acquisition and analysis is described below.

Generation of K562 cells expressing CD200

Complementary DNA for CD200 was kindly provided by IMAGE consortium complementary DNA clone ID 5299899 (Lennon et al.³¹) and subsequently sub-cloned into the PINCO retroviral expression vector (gift of Pier Pelicci, European Institute of Oncology, Milan, Italy), which co-expresses green fluorescent protein (GFP) from an internal cytomegalovirus promoter.³² Replication-defective retrovirus was generated by transient transfection of Phoenix packaging cells (gift of Garry Nolan, Stanford University School of Medicine, Stanford, CA, USA). The K562 human immortalized cell line was retrovirally transduced as previously described.⁹ In this way three cultures were generated: K562-mock, K562-CD200⁻ (PINCO-expressing GFP alone) and K562-CD200⁺ (PINCO co-expressing GFP and CD200). Transduced K562 cells were cultured in RPMI-1640 (Sigma) containing 10% (v/v) FCS (Biosera, East Sussex, UK), 2 mm l-glutamine (Sigma) and cell sorted for GFP positivity using a MoFlo high speed cell sorter (Dako, Cambridgeshire, UK). Gene transduction (GFP⁺), CD200 protein expression and cell sorting purity were confirmed by flow cytometry as described below (see Data acquisition and analysis).

Assessment of NK cell activity

In a 96-well flat-bottomed culture plate, 10⁵ PBMC from normal healthy individuals were co-cultured in triplicate with 5 × 10⁴ K562-CD200⁺ target cells (or equivalent CD200⁻ controls) in the presence of mouse anti-human CD107a-FITC (BD Biosciences) or isotype-matched antibody immunoglobulin G (IgG)_1κ-FITC (BD Biosciences) for 6 h at 37 °C and 5% CO₂ with the addition of monensin GolgiStop (BD Biosciences) for the last 5 h. To control for spontaneous degranulation, PBMC were incubated at an equivalent density without K562 target cells. Following incubation, cells were harvested, washed twice with staining buffer and incubated with anti-human CD3-PerCP-Cy5.5, CD16-PE (BD Bioscience) and mouse anti-human CD56-APC (Miltenyi Biotec, Bergisch Gladbach, Germany) for 30 min at 4 °C. Following NK cell surface staining, the cells were washed (as above) and analyzed immediately by flow cytometry (see Data acquisition and analysis).

In order to determine the effect of AML patient blasts on NK cell degranulation, cryopreserved AML blasts (of known CD200 expression status) were thawed, washed in pre-warmed RPMI-1640 containing 10% (v/v) FCS and 2 mm l-glutamine and assayed for CD107a expression as above. To block CD200, 20 μg of unconjugated mouse anti-human CD200 (BD Biosciences) was added to the assay and compared with an IgG1 isotype control. Cell viability and apoptosis were assessed by flow cytometry following incubation with annexin-V and 7AAD (BioLegend). To identify AML blast apoptosis, cells were stained with CD45-APC.H7 as detailed above before annexin-V/7AAD staining was performed.

NK cell interferon (IFN)γ response toward CD200⁺ and CD200⁻ K562 targets was assessed by ELISPOT as previously described with minor adjustments.³³ Briefly, Millipore Multiscreen-HA 0.45 μm ELISPOT plates were incubated with 10 μg/ml anti-human IFNγ capture antibody (Mabtech, Stockholm, Sweden) in sterile phosphate-buffered saline overnight at 4 °C, followed by incubation with RPMI-1640 medium adjusted to contain 2 mm l-glutamine and 10% (v/v) FBS for 1 h at room temperature. In respective wells, 5 × 10³ CD200⁺ or CD200⁻ K562 targets (T) were subsequently incubated at 37 °C and 5% CO₂ with lymphocyte effectors (E) from healthy donors overnight (18–20 h) at the following E:T ratios: 20:1, 10:1 and 5:1. Cells were harvested and sterile dH₂O added to the wells for 10 min before consecutive incubations with 1 μg/ml biotinylated mouse anti-human IFNγ (Mabtech) and 1 μg/ml streptavidin–alkaline phosphatase in phosphate-buffered saline 1% bovine serum albumin at room temperature for 2 h. Wells were developed with alkaline phosphatase substrate (Bio-Rad, Hertfordshire, UK) in accordance with the manufacturer’s instructions, washed and allowed to dry overnight. Spots were counted using an inverted stereoscopic microscope. Positive results were considered >10 spots per well with significantly greater number of spots than the negative control (P<0.05, exceeding (sample–background)±2 s.d. To compare NK cell IFNγ response toward autologous blasts in CD200^hi and CD200^lo AML patients (see Data acquisition and analysis), sufficient PBMC to give 10⁵ lymphocytes were incubated overnight in Millipore Multiscreen-HA 0.45 μm ELISPOT plates, prepared as above. To confirm that the source of IFNγ was NK cell derived, replicate incubations were carried out using NK cell-depleted PBMC.

NK cell depletion from AML patient PBMC

NK cells from AML patient PBMCs or bone marrow mononuclear cells (BMMCs) were depleted by MACS (Miltenyi Biotec) column separation in accordance with the manufacturer’s instructions. Briefly, AML patient PBMC or BMMC from cryopreserved stocks were thawed and rested overnight in RPMI-1640 containing 10% (v/v) FCS and 2 mm l-glutamine at 37 1° and 5% CO₂. Cells were washed (×2) with 5 ml column buffer (phosphate-buffered saline, 0.5% bovine serum albumin and 2 mm EDTA) and incubated with CD56 microbeads (Miltenyi Biotec). Cells were washed with column buffer and subjected to magnetic separation using an LD column with VarioMACS magnetic unit. The effluent that contained the CD56 depleted cells was washed with warm RPMI-1640. All incubations were carried out at 4 °C. Aliquots from effluent and column bound fractions assessed by flow cytometry revealed >90% NK depletion.

Data acquisition and analysis

Data were acquired using an Accuri C6 cytometer (Accuri, St Ives, UK) or FACScanto II cytometer (BD Biosciences). Data analysis was performed using FCS express v3.0 (DeNovo Software, Los Angeles, CA, USA). Background fluorescence was established by isotype-matched controls; autofluorescence of mock-infected cultures defined the threshold for GFP positivity. For analysis of CD200 expression on AML patient blasts, lymphocytes were excluded from the analysis based on low side scatter and high CD45 expression. The level of CD200 expression is broadly expressed on AML blasts (Supplementary Figure S1); to define CD200 expression level, CD200^lo relates to data inclusive of minimum to Q1 (0–0.68) and CD200^hi relates to data range from Q3 to maximum (2.87–16.14). For AML patient NK cell surface receptor expression analysis, cells were selected for analysis based on the following phenotype: low side scatter/CD45^{+ +} (thereby excluding myeloblasts), CD19⁻ CD3⁻ CD56⁺ CD16⁺. The NK cell activation markers (NKG2D, NKp30, NKp44 and NKp46) and CD200R expression was analyzed within NK cell sub-populations defined by CD56 vs CD16 bivariate analysis. Data reported are median fluorescence fold increase relative to an isotype-matched control. Significance of difference was determined using one-way analysis of variance with Turkey’s multiple comparison test, or paired t-tests using Minitab v15 (Minitab, State College, PA, USA).

Results

Expression of CD200 on AML blasts is associated with reduced NK cell frequency

To establish whether CD200 protein expression on leukemic blasts was associated with NK cell activation status in AML, we compared NK frequency in CD200^lo and CD200^hi patients (see Supplementary Figure S1 for cytometric gating strategy and CD200 expression levels). We found a twofold reduction in overall NK frequency in CD200^hi patients (3.7±2.5% vs 6.3±4.7% P<0.05). As the CD56⁺ CD3⁻ CD19⁻ NK population consists of sub-populations with different cytolytic potentials, we further examined the relative frequencies of the five sub-populations of NK cells defined by expression of CD56 and CD16 (Poli et al.³⁰) (Figure 1a). These consist of two populations of cells with limited cytolytic activity (1 and 2) and three cytolytic populations (3, 4 and 5). In healthy individuals, population 4 comprises the bulk of NK cells and this was also the case for CD200^lo AML patients. CD200^hi expression was associated with a significant twofold reduction in the frequency of population 4 (P<0.001; Figure 1b). We observed no significant differences in NK cell frequency of sub-populations 1, 2, 3 and 5 (Figure 1b). These data demonstrate that CD200^hi expression on AML blasts is associated with a significant decrease in the frequency of NK cells found within population 4, which is associated with cytolytic activity in healthy subjects.

Figure 1.

Expression of CD200 on AML blasts relates to a change in the frequency of NK cell subsets. (a) Representative flow cytometric plots of NK cells in CD200^lo and CD200^hi AML patients (CD200 expression level is defined in Data acquisition and analysis). Five NK cell sub-populations are shown based on the bivariate expression of CD56 and CD16. Plots are gated on CD45^{+ +} CD19⁻ CD3⁻ lymphocytes. (b) Summary box and whisker plot summarizing the frequency of the five NK cell sub-populations (as percentage of total lymphocytes) between CD200^lo (filled boxes) and CD200^hi (n=12 per group). ***P<0.001 analyzed by one-way analysis of variance (ANOVA) with Turkey’s multiple comparison test.

NCR repertoire is altered between CD200 ^hi and CD200^lo AML patients

To further define our findings, we compared NCR expression in CD200^lo and CD200^hi AML patients within the more cytolytically active NK cell sub-populations (3–5). This important family of receptors is directly involved in tumor recognition and NK cell triggering. A total of three NCRs (NKp30, NKp44 and NKp46) were evaluated in this study. In addition, the well-established NK cell triggering receptor, NKG2D, was investigated (representative flow cytometric data are shown in Supplementary Figure S2). We found that select NCRs were significantly reduced in certain NK cell sub-populations of CD200^hi patients compared with their CD200^lo counterparts (Figure 2). In NK cell population 4, NKp44 expression level in CD200^hi patients was half that observed in CD200^lo patients (Figure 2a; P<0.05). A similar reduction was also observed for NKp46 (P<0.01) and NKp30, although the reduction in NKp30 did not reach significance (Figures 2b and c). A significant reduction in NKp44 expression was also observed for population 5 (Figure 2a; P<0.05). In contrast, the expression of NKG2D was found to be unaffected between CD200^lo and CD200^hi patients (Figure 2d). Taken together, these data demonstrate that an increase in CD200 expression on AML blasts is associated with reduction in NCR expression, particularly within NK cell population 4.

Figure 2.

Comparisons of NK cell activation marker repertoire between CD200^lo and CD200^hi AML patients. NK cells from CD200^lo and CD200^hi AML patients were examined by eight color flow cytometry (see Materials and methods). Sub-populations 3, 4 and 5 in CD200^lo (filled bars) and CD200^hi (open bars) patients were assessed for expression of (a) NKp44, (b) NKp46, (c) NKp30 and (d) NKG2D. Data represent mean±1 s.d. (n=12, per group). *P<0.05 analyzed by one-way analysis of variance (ANOVA) with Turkey’s multiple comparison test.

Functional assessment of NK cell activity toward CD200⁺ and CD200⁻ K562 targets

We next determined whether CD200 could suppress NK cell activity and target cell recognition in vitro. K562 is an NK-sensitive, CD200-negative leukemic cell line, which we stably transduced with CD200 using a retroviral expression vector (Supplementary Figure S3). We analyzed NK cell activity toward K562-CD200⁺ cells compared with K562-CD200⁻ controls by assessing degranulation of cytotoxic granules through measurement of lysosomal-associated membrane protein-1 (CD107a) expression and cytokine release using IFNγ ELISPOT assays.

NK cell populations 3–5 from healthy subjects degranulated less in response to CD200⁺ K562 targets cells compared with CD200⁻ controls (Figure 3a). The largest decrease in degranulation was observed for NK cell population 4, which showed a significant reduction of 55% (Figure 3b; P<0.05). Furthermore, we observed a significant reduction in the number of NK cells producing IFNγ in response to CD200⁺-expressing cells when compared with CD200⁻ K562 targets at all E:T ratios examined (Figure 3c and Supplementary Figure S4C). These findings demonstrate that CD200 expression on leukemic cells can suppress both NK cell cytolytic function and NK pro-inflammatory cytokine secretion.

Figure 3.

Functional assessment of NK cell activity toward K562-CD200⁻ and K562-CD200⁺ targets. (a) Summarized data illustrating percentage of degranulating NK cells in sub-populations 3, 4 and 5 (defined as in Figure 1a) in response to K562-CD200⁻ or K562-CD200⁺ targets (Supplementary Figure S3). Background CD107a expression (PBMC control cultures with no targets) is also indicated. Flow cytometric gating strategy for this assay is shown in Supplementary Figure S4). (b) Representative flow cytometric plots illustrating expression of CD107a (open histograms) in sub-population 4 NK cells in response to K562-CD200⁻ and K562-CD200⁺ targets, or PBMC control. Filled histograms represent isotype-matched controls. (c) IFNγ release in response to K562-CD200⁻ targets compared with K562-CD200⁺ targets as measured by ELISPOT at different effector to target (E:T) ratios (see Materials and methods; representative data shown in Supplementary Figure S4). Data represent mean±1 s.d. (n=6, per group). *P<0.05 analyzed by one-way analysis of variance (ANOVA) with Turkey’s multiple comparison test.

To confirm that reduced NK cell degranulation and IFNγ production resulted in correspondingly lower levels of target cell killing, the viability of CD200⁻ and CD200⁺ K562 target cells were also determined using 7AAD and annexin-V, following an 18-h incubation with effectors. As expected, we found a significant decrease in the frequency of apoptotic K562-CD200⁺ compared with K562-CD200⁻ cells (Figures 4a and b), which corresponded to a decrease in NK CD107a expression in the same culture (Figure 4c). These data therefore confirmed that decreased NK degranulation corresponds to an increased viability of CD200⁺ target cells.

Figure 4.

Reduced NK cell degranulation corresponds to increased viability of CD200⁺ targets. K562-CD200⁻ or K562-CD200⁺ targets were co-cultured with PBMC from healthy donors, for 18 h at 37 °C and 5% CO₂ at an effector:target ratio of 2:1 (or in the absence of PBMC; K562 controls). Following this, cells were stained with annexin-V/7AAD and analyzed by flow cytometry, excluding lymphocytes, (deselected based on low FSC/side scatter (SSC)). (a) Summarized data (mean+1 s.d.; n=8). (b) Representative annexin-V/7AAD flow cytometric plots from PBMC co-cultured with K562-CD200⁻ and PBMC co-cultured with K562-CD200⁺ targets. (c) Representative flow cytometric plots confirming that NK cells degranulated significantly more toward K562-CD200⁻ targets compared with K562-CD200⁺ targets under the same assay conditions. Figures represent mean percent degranulation for each target type (n=8, P<0.05). *P<0.05 analyzed by one-way analysis of variance (ANOVA) with Turkey’s multiple comparison test.

CD200R expression is expressed on NK cell populations 3–5 in AML patients

The simplest mechanistic explanation for our data would be that CD200 directly engages CD200R on NK cells, resulting in suppression of NK cell function. To investigate this hypothesis, we established whether CD200R was detectable on NK cell populations in AML patients. Representative flow cytometric plots illustrated that CD200R was indeed detectable on NK cells (Figure 5a), although the level of expression varied depending on the sub-population of NK cells studied. Interestingly, the frequency of sub-populations 1 and 2 are rare in AML patients making analysis of CD200R technically impossible.

Figure 5.

CD200R is expressed on AML patient NK cell sub-populations. (a) Representative flow cytometric plots reveal that CD200R expression is expressed on NK cells on sub-populations 3, 4 and 5 in AML patients. (b) Summary data illustrating the CD200R relative expression level within the indicated NK cell sub-populations. Data represent mean+1 s.d. (n=20).

CD200 blocking antibody recovers NK cell activity in CD200^hi patients

The observation of CD200R expression on NK cells led us to investigate whether blocking CD200–CD200R interaction could relieve suppression of NK cell activity. Degranulation of NK cells isolated from either CD200^hi or CD200^lo AML patients was therefore studied in the presence of anti-human CD200 or isotype control. As expected, the frequency of NK cell degranulation was significantly reduced in response to autologous AML blasts in CD200^hi compared with CD200^lo AML patients (Figure 6a); however, NK degranulation could be augmented significantly (twofold) by incubating in the presence of anti-CD200. In contrast, CD200 antibody had no effect on the frequency of degranulation in NK cells from CD200^lo patients (Figure 6a). Similarly, we also found that the NK cell IFNγ response was reduced in CD200^hi AML patients (Figure 6b). Again anti-CD200 augmented NK cell IFNγ production only for CD200^hi AML, although this did not reach significance for this group of patients (Figure 6b). These findings support the assertion that CD200 mediates immunosuppression of NK cell activity in AML, but also demonstrates that this suppression is reversible if interaction with CD200 is blocked.

Figure 6.

CD200 blocking antibody promotes NK cell activity in CD200^hi patients. (a) BMMC from CD200^lo and CD200^hi AML patients were incubated in the presence of CD107a with 5 μg/ml of unconjugated anti-human CD200 or respective IgG isotype-matched control for 6 h at 37 °C and 5% CO₂. The percentage of CD107a-positive NK cells in response to autologous AML blasts was determined by flow cytometric analysis. (b) NK cell IFNγ release from BMMC of AML patients in response to CD200^lo or CD200^hi autologous blasts, as measured by ELSIPOT. Incubations were carried out in the presence of unconjugated anti-human CD200 or respective IgG isotype-matched control for 18–20 h at 37 °C and 5% CO₂. Data represent mean+1 s.d. (n=6). *P<0.05 one-way analysis of variance (ANOVA) with Turkey’s multiple comparison test.

Discussion

Previous studies have documented decreased expression levels of NK cell-triggering receptors (NKG2D and the NCR’s) in patients diagnosed with AML.^29,34 Furthermore, research has also shown that NK cell cytolytic activity is decreased and associated with poor overall survival and reduced AML clearance.²⁸ As CD200 is associated with immunosuppression and its expression on AML blasts is indicative of poor prognosis in AML,¹⁴ we compared NK cell function and phenotype in AML patients that were CD200^lo or CD200^hi.

NK cells are notably heterogeneous and it is documented that expression of CD16 (FcγRIII) in conjunction with CD56 discriminates at least five functionally distinct NK cell sub-populations from PBMCs: (1) CD56^brightCD16⁻ (naïve, immunoregulatory ~15% of all NK cells in health), (2) CD56^brightCD16⁺ (rare, immunoregulatory and cytolytic), (3) CD56^dimCD16⁻ (activated, role not well defined, minority in health), (4) CD56^dimCD16⁺ (activated, cytolytically potent, ~80% of all NK cells in health) and (5) CD56⁻CD16⁺ (activated, rare in health, expanded in immunodeficiency), reviewed by Poli et al.³⁰ Using this strategy, we are the first to show that population 4 is the dominant NK cell sub-population for AML patients, with little or no population 1 cells. We also show that NK cell population 4 is reduced by 50% in CD200^hi compared with CD200^lo AML patients, an important observation given that this NK sub-population is considered to be the dominant cytolytic NK sub-population in health. Thus, although CD16 expression remains largely unchanged, the frequency of CD56^dimCD16⁺ NK cells is less for CD200^hi AML patients. Populations 3 and 5 were detectable at low frequency in all AML patients, however, no significant differences were observed between CD200^lo and CD200^hi AML patients.

NK cell activation is a complex process involving both ‘missing-self’ and ‘altered-self’ mechanisms, whereby the latter delivers activation signals through triggering receptors such as NKG2D and the NCR’s; NKp30, NKp44 and NKp46, which serve to identify viral infection or malignant cell markers induced by ‘stress’ to the cell.³⁵ In the context of this study, the NCRs were the main focus, because this triggering receptor family is directly involved in NK cell/leukemic blast cell recognition and activation.^27,36,37 The analysis of NCR’s within the five NK cell sub-populations is yet to be reported in AML and to this end our data reveal important information regarding NK cell sub-population phenotypic heterogeneity between CD200^lo and CD200^hi AML patients. We show that the overall NCR expression level is lower for CD200^hi AML blasts compared with CD200^lo in these NK cell phenotypes. Furthermore, we show a ~50% decrease in the expression level of NKp44 and NKp46 in population 4 for CD200^hi AML patients. Therefore, NK sub-population 4 (which potentially has the most potent anti-leukemic activity)³⁸ appears central to CD200 mediated NK cell suppression in AML, displaying both a reduced frequency and decreased levels of NCR expression, which jointly could contribute to decreased leukemic cell recognition by NK cells. Interestingly, NKp30, NKp44 and NKp46 have previously been reported to be downregulated on AML patient NK cells as a whole to a similar extent.^29,34 Costello et al., showed that most AML patients had an NCR^lo phenotype and both reported around 50% decrease in NKp46 expression.²⁸ In a study by Fauriat et al.³⁴, NKp46 was downregulated on NK cells in around 75% of the AML patients examined using a cohort of 71 patients (mainly from the FAB M2 and M4 subtypes). Previously, we reported CD200 upregulation in 43% of patients diagnosed with AML, with almost all M2 and M4 FAB types overexpressing this protein.¹⁴ It is therefore possible that Fauriat et al.³⁴ mainly analyzed CD200 overexpressing patients, a cohort which we have previously identified to have a poor prognosis. Furthermore, Fauriat et al.³⁴, also reported that AML patients with an NK cell NKp30^dim and NKp46^dim phenotype were prognostically disadvantaged, with around 30% AML patient survival after 5 years. Although NK activating receptor expression is generally reduced in malignancy (possibly through the chronic exposure to tumor ligands),³⁹ we demonstrate additional NCR reduction in CD200^hi AML, potentially exacerbating NK cell immune dysfunction and providing some possible functional insight into why CD200^hi AML patients have a worse prognosis.

The association of CD200 expression with alterations in NK sub-population frequency together with the expression of CD200 receptor on NK cells suggested that CD200 may directly regulate NK cells. To address this, we asked whether NK cell function was altered toward K562 targets transduced with CD200. Several studies have examined NK cell degranulation toward leukemic targets using CD107a (LAMP-1) expression to monitor NK cell cytolytic activity.^40,41 The latter study reported that NK cell population 3 (CD56^dimCD16⁻) was the major cytolytic population toward K562 targets, with little degranulation in population 4. However, it was subsequently noted that the identified NK cell sub-populations were not pure and possibly contaminated with CD3⁺ or CD19⁺ lymphocytes, including CD8⁺ NKT cells, which are also cytolytically potent toward tumor targets.⁴² In our study (gating NK cells on CD45^{+ +} CD19⁻CD3⁻CD56⁺ lymphocytes), we show that NK cell population 4 is cytolytically active toward K562 target cells. Moreover, this activity is reduced by ~50% toward CD200⁺ K562 targets, a finding confirmed by the analysis of target cell survival. Aside from their cytotoxic function, NK cells have an important immune-regulatory role through the secretion of type-I and type-II cytokines.⁴³ This is particularly apparent during the primary immune response in which NK cells are believed to be the principal source of IFNγ.⁴³ Cytokine secretion is generally associated with the CD56^bright NK cell subset, however, CD56^dim NK cells also have the capacity to produce type-I cytokines in direct correlation to cytolytic activity⁴³ and are capable of producing substantial amounts of IFNγ under similar conditions to that employed in this study.⁴⁴ By means of ELISPOT, we also showed a significant reduction in the number of NK cells producing IFNγ in response to CD200⁺ leukemic targets and primary autologous AML blasts from CD200^hi patients, providing the first evidence that engagement of CD200 with its cognate receptor may be responsible for suppression of NK cell activity from both cytolytic and immunoregulatory perspectives. Furthermore, we found that we could recover a significant proportion of NK cell cytolytic activity toward CD200^hi AML blasts using a CD200 antibody. Similarly, studies exploring the role of CD200 in hematologic malignancies have previously demonstrated a capacity to restore tumor immunity through blocking the CD200–CD200R interaction,^45,46 however, these studies focused on the adaptive immune system and T-cell response. Here, we show for the first time that blocking CD200–CD200R interaction in CD200⁺ AML may have therapeutic value in terms of the innate immune response. In a separate study, we have shown that CD200 expression is also associated with suppression of T-cell-mediated immunity in AML (SC, manuscript in preparation).

Together these data suggest that CD200 blocking antibody treatment may promote both innate and T-cell-mediated anti-tumor responses in AML. The observation that CD200-expressing AML patients are at increased risk of relapse, suggests CD200 therapy may be effective in remission when lymphocyte populations are beginning to repopulate and in particular where CD200⁺ AML stem cells are present.