We present two different staining protocols for NKG2D ligand (NKG2DL) detection in human primary acute myeloid leukemia (AML) samples. The first approach is based on a fusion protein, able to recognize all known and potentially yet unknown ligands, while the second protocol relies on the addition of multiple anti-NKG2DL antibodies.
Within the same patient, absence of NKG2D ligands (NKG2DL) surface expression was shown to distinguish leukemic subpopulations with stem cell properties (so called leukemic stem cells, LSCs) from more differentiated counterpart leukemic cells that lack disease initiation potential although they carry similar leukemia specific genetic mutations. NKG2DL are biochemically highly diverse MHC class I-like self-molecules. Healthy cells in homeostatic conditions generally do not express NKG2DL on the cell surface. Instead, expression of these ligands is induced upon exposure to cellular stress (e.g., oncogenic transformation or infectious stimuli) to trigger elimination of damaged cells via lysis through NKG2D-receptor-expressing immune cells such as natural killer (NK) cells. Interestingly, NKG2DL surface expression is selectively suppressed in LSC subpopulations, allowing these cells to evade NKG2D-mediated immune surveillance. Here, we present a side-by-side analysis of two different flow cytometry methods that allow the investigation of NKG2DL surface expression on cancer cells i.e., a method involving pan-ligand recognition and a method involving staining with multiple antibodies against single ligands. These methods can be used to separate viable NKG2DL negative cellular subpopulations with putative cancer stem cell properties from NKG2DL positive non-LSC.
NK cells are important effectors of the innate immune system that can recognize and eliminate malignant cells or stressed healthy cells (e.g., by a viral infection) without prior antigen stimulation1. This process is tightly regulated via a complex repertoire of activating receptors —such as natural cytotoxicity receptors (NCRs), NKG2D and CD16— and inhibitory receptors that are largely represented by killer immunoglobulin-like receptors (KIRs)2. Binding of KIRs to human leukocyte antigen (HLA) class I molecules on somatic cells ensures self-recognition and conveys NK cell tolerance. On the other hand, absence of self-recognition and increased binding of activating receptors to their ligands on the target cells trigger the release of cytotoxic granules leading to NK cell-mediated cytotoxicity1. Finally, NK cells can exert antibody-dependent cellular cytotoxicity (ADCC) by binding of the activating receptor CD16 to targets expressing the Fc portion of Ig (FcR)2. Apart from direct cytotoxicity, NK cells can also trigger cytokine release bridging the innate with the adaptive immune system3.
NKG2D is a major activating receptor expressed on NK, NKT, γδ T, and naïve CD8+ T cells4 that enables such cytotoxic immune cells to recognize and lyse NKG2D ligand (NKG2DL) expressing target cells. Healthy cells commonly do not express NKG2DL. Instead, NKG2DL expression is upregulated on malignant or virus-infected cells to make these amenable to immune clearance5.
The human NKG2DL family comprises eight known molecules among which the two MHC I chain-related molecules A and B (MICA and MICB6) and the cytomegalovirus UL16-binding proteins 1–6 (ULBP1-67). The expression of NKG2DL is regulated on the transcriptional, post-transcriptional as well as the post-translational levels8. As such, while NKG2DL expression is commonly not detectable on the surface of healthy cells, NKG2DL mRNA9 and intracellular protein expression were reported in healthy tissues. The functional relevance of such expression and the mechanisms underlying such discrepant expression patterns remain to be defined10.
The mechanistic regulation of NKG2DL expression in the cancer cells is a fascinating area of investigation. Pathways known to be involved in either cellular stress, e.g., the heat shock stress pathway9, or DNA damage-associated pathways, such as the ataxia telangiectasia mutated (ATM) and Rad3 related (ATR) pathway11, as well as viral or bacterial infections have been directly linked to the induction of NKG2DL expression12. However, even if surface expression of NKG2DL has been effectively induced, this expression can be lost again through proteolytic-mediated shedding, a mechanism associated with immune escape and poor clinical prognosis in some cancers13.
The absence of cell surface NKG2DL may also play important roles in patients with AML. Here, treatment with intensive chemotherapy often induces remission, but relapse often occurs from leukemic stem cells (LSC), which selectively survive chemotherapies and evade immune response. As we recently showed, LSCs, for example, escape NK cell lysis by suppressing NKG2DL surface expression14.
Inversely, the absence of surface NKG2DL expression can be used as a method to identify and viably isolate putative stem-like subpopulations of cells from bulk counterpart leukemic subpopulations. Here, we present two flow cytometric approaches that can be used to detect NKG2DL surface expression and thereby identify NKG2DL negative stem cells in leukemia and perhaps also in other cancers: A method for pan-ligand surface recognition and a method involving staining with single or pooled antibodies recognizing individual known NKG2DL proteins.
Patient samples were collected following approval from the Ethics Review Board of the University Hospitals of Basel and Tuebingen.
1. Biotinylation of the NKG2D fusion protein
NOTE: This step is performed with a biotinylation kit (see Table of Materials) according to the manufacturer’s instruction. This step of the protocol must be performed at least 24 h prior to the staining. The biotinylated NKG2D fusion protein should be stored at -20 °C.
2. Thawing of primary AML cells
NOTE: Approximately 5,000,000 frozen leukemic cells per patient stored in liquid nitrogen were thawed and then used for the assays right away. Leukemic cells were obtained and frozen as previously described15. Briefly, peripheral blood samples collected from patients with AML and high blast cell percentages (>90% of blasts among mononuclear cells) were processed with a density gradient separation to obtain mononuclear cells and subsequently frozen in fetal calf serum (FCS) containing 10% dimethyl sulfoxide solution (DMSO). Cell numbers highly varied between patients in dependence on the leukocyte concentration in the patient (range: 1,000,000 to 30,000,000 leukemic cells per mL blood).
3. Cell counting
4. Staining of primary AML cells using the biotinylated NKG2D fusion protein
5. Staining of single or pooled single anti-NKG2DL antibodies
6. Data acquisition
NOTE: Make sure that weekly quality controls of the flow cytometer are performed to ensure that lasers are functioning properly. For the protocol presented here, a weekly Cytometer and Tracking (CTS) process is performed with relative beads to check laser performances on all channels. After the CTS, 10,000 events are recorded using the 8-peak beads as an internal control. All peaks should be in the same position inside their gates and well separated from each other.
Both the protocols presented here allow the enrichment of AML LSC by flow cytometric analyses using CD34, a known marker of LSC17, in combination with NKG2DL surface expression by either utilizing pan-ligand recognition or staining with pooled antibodies against individual ligands. In Figure 1, we show that the analyzed AML samples are positive for CD34 and NKG2DL, but negative subpopulations also exist, which is shown by the presence of four different populations in total. Our data shows the typical gating strategy for such a staining, starting with the selection of the main population of cells via their FSC and SSC. Doublets and dead cells are excluded in the downstream analysis (Figure 1A). Gates are adjusted using FMO controls to ensure proper identification of positive cells (Figure 1B).
In Figure 1C, we highlight the fluorescence intensities of cells positive for CD34 (APC, Y-axis) vs NKG2DL (PE, X-axis). AML cells that are positive for CD34 show a lower surface expression of NKG2DL (Figure 1C), which is in line with previous findings from our laboratory, showing that NKG2DL expression is associated with lack of stemness14.
Figure 2 indicates the robustness of both the NKG2DL staining methods. In Figure 2A, we investigate three primary AML samples side-by-side showing 19.8, 49.8, and 89.4%, respectively of positive events for the fusion protein staining vs 20.4, 50.4, and 90.6% for the pooled anti-NKG2DL antibody staining (pooled antibodies against MICA, MICB, ULBP1, ULBP2/5/6, and ULBP3), respectively. On the other hand, single ligand (against either MICA, MICB, ULBP1, ULBP2/5/6, or ULBP3) staining shows a range of positive events up to 92% (Figure 2B). This percentage varies depending on the ligand itself and the patient, for example, ULBP3 (Figure 2B). Of note, a single cell can express multiple NKG2DL.
When multiple anti-NKG2DL antibodies are pooled and combined into one single tube, the percentage of positive cells is comparable to the percentage of the fusion protein (Figure 2A). As such, both protocols work well and may provide similar results regarding NKG2DL expression on primary AML cells. In some AML, the fusion protein method may allow higher sensitivity since it can allow the recognition of further NKG2DL proteins.
Figure 1: Typical gating strategy of primary AML samples. (A) Cells are first gated based on their size (X-axis) and complexity (e.g., granularity, Y-axis). Doublets are then excluded by plotting the height or width against the area for forward scatter. Finally, living cells are selected based on their size (X- axis) vs their negativity for 7AAD (PerCP-Cy5.5, Y-axis). (B) FMO controls are used to set up the gates helping to discriminate between positive and negative populations for indicated markers. The gating strategy remains identical as depicted in Figure 1A. (C) Flow cytometry plot showing positive and negative events for CD34 (APC, Y-axis) vs NKG2DL (PE or AlexaFluor488, X-axis). Please click here to view a larger version of this figure.
Figure 2: Staining protocols comparison. (A) Bar graphs showing the percentage of positive events for the fusion protein vs the pooled anti-NKGD2L antibodies (n = 3 AML samples) from gated single living cells. (B) Bar graphs displaying the percentage of positive events for single ligand staining for all known anti-NKG2DL antibodies and pooled single ligands (see Table of Materials for detailed antibody information). Gating was again performed on single living cells. Please click here to view a larger version of this figure.
Tube name | Antigen | Fluorophore | Dilution/concentration |
Unstained cells | – | – | – |
Single stain PE |
NKG2D-Fusionprotein | PE | 1:10 for primary step 1:100 for secondary step |
Single stain 7AAD (PerCP-Cy5.5) |
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 |
Single stain APC |
CD34 | APC | 1:25 |
Full stained cells | CD34 | APC | 1:25 |
NKG2D-Fusionprotein | PE | 1:10 for primary step 1:100 for secondary step | |
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 |
Table 1: Tubes required to stain samples with the NKG2D fusion protein. This table shows the tubes required in order to set up the experiment.
Tube name | Antigen | Fluorophore | Dilution/concentration |
Unstained cells | – | – | – |
Single stain Alexa Fluor 488 |
ULBP-1 | Alexa Fluor 488 | 10 µg/mL for primary step 4 µg/ml for secondary step |
ULBP-2/5/6 | |||
ULPB-3 | |||
MICA | |||
MICB | |||
single stain 7AAD (PerCP-Cy5.5) |
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 |
Single stain APC |
CD34 | APC | 1:25 |
Full stained cells | CD34 | APC | 1:25 |
NKG2DL | Alexa Fluor 488 | 10 µg/mL for primary step 4 µg/ml for secondary step |
|
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 |
Table 2: Tubes required to stain samples with the anti-NKG2DL antibodies. This table shows the tubes required in order to set up the experiment.
Tube name | Antigen | Fluorophore | Dilution/concentration |
Unstained cells | – | – | – |
FMO_APC | ULBP-1 | Alexa Fluor 488 | 10 µg/mL for primary step 4 µg/mL for secondary step |
ULBP-2/5/6 | |||
ULPB-3 | |||
MICA | |||
MICB | |||
NKG2D-Fusionprotein | PE | 1:10 for primary step 1:100 for secondary step |
|
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 | |
FMO_PE | CD34 | APC | 1:25 |
ULBP-1 | Alexa Fluor 488 | 10 µg/mL for primary step 4 µg/mL for secondary step |
|
ULBP-2/5/6 | |||
ULPB-3 | |||
MICA | |||
MICB | |||
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 | |
FMO_Alexa Fluor 488 | CD34 | APC | 1:25 |
NKG2D-Fusionprotein | PE | 1:10 for primary step 1:100 for secondary step |
|
Live/Dead | 7AAD (PerCP-Cy5.5) |
1:1000 |
Table 3: Tubes required in order to set up the essential controls. This table shows the tubes required in order to set up proper FMO, primary, and secondary antibody controls. All the required controls (FMO) must be added by the experimenter to obtain valid results and interpretable data.
Here we present two flow cytometric methods that can detect NKG2DL surface expression on human primary AML cells. We show that both detection methods can be used in conjunction with other antibody stainings (e.g., detecting CD34 expression). Similar stainings may also be performed on other primary cell types and cell lines.
We recently showed that the absence of NKG2DL on the surface of AML patient blasts can enrich LSC14. In AML, NKG2DL negative but not NKG2DL positive leukemic subpopulations possess clonogenic and in vivo leukemia-initiating capacities. Future research will show whether absence of NKG2DL surface expression can also be used as a tool to enrich stem cells in other cancer types.
CD34 has been classically shown to mark LSCs in AML, but AML is a highly heterogeneous disease and not all AML patients show CD34 expression (i.e., CD34 negative AML18). We have previously shown that absence of NKG2DL surface expression provides a novel tool to identify LSC in this subgroup of CD34 negative AML. Here, in contrast, we focus on the subgroup of CD34-positive AMLs and show that different subpopulations can occur. Future research will show whether co-staining for NKG2DL in conjunction with such other markers (e.g., CD34) can more effectively enrich LSC.
In flow cytometry, it is important to include adequate controls such as FMO to help the experimenter to correctly discriminate between positive and negative events. In addition, positive controls for every used antibody are necessary to ensure that the absence of staining is not due to a malfunctioning antibody. The protocol we present here uses only a limited number of surface markers, which can be extended according to the experimenter’s need, for example, by addition of CD33 or potential LSC markers19. Furthermore, the fluorophores used in this protocol can be adapted. Importantly, such changes should be accompanied by an antibody titration to determine the best dilution.
One critical step before the staining is the thawing of primary AML cells. Indeed, in addition to their fragility, such samples are conserved in dimethyl sulfoxide (DMSO), which is per se toxic for the cells. Therefore, the samples should be handled with care and should be rinsed thoroughly to avoid prolonged exposure to DMSO. Although our samples were processed and frozen according to a previously established protocol15, it is important to carefully follow the steps described in the manuscript to ensure a high cell recovery. Working with fresh samples will ensure improved viability and is thus recommended. However, in most cases this is not practicable.
Another factor influencing the described method is sample-specific deviation that can result in different cell scattering profiles observed via flow cytometry. Due to disease heterogeneity, this is unavoidable but should be considered in the analysis. Additional variability can be introduced by freezing/thawing or other mechanical steps in sample processing. In general, rapidity is key to avoid cell death. Yet, dying cells and debris cannot be avoided and need to be taken into account and excluded in the analysis. Gating based on FSC/SSC should be done with thoroughness and a critical eye as it is the first step of sample analysis.
One major advantage of the fusion protein is its potency to detect all known and potentially unknown NKG2DLs, while specific anti-NKG2DL antibodies (e.g., MICA, MICB) introduce a selection bias. Both methods are equally feasible. For the analyzed limited number of samples, they also showed very comparable results. Using the NKG2D fusion protein has some major benefits: Less time-consuming, decreased reagent number, and number of staining samples as well as a lower probability of human mistakes (i.e., pipetting mistakes). Moreover, this method may also be more effective in capturing NKG2DL expression in some samples, since it should also capture expression of unknown proteins with the NKG2DL function, which may be expressed in some AML cases. While providing insight into the surface presence of NKG2DL, these method does not provide information about the intracellular levels of NKG2DL, their trafficking or regulation, and other assays should therefore be performed to gain further knowledge on such questions.
The authors have nothing to disclose.
This study was supported by grants from the Swiss National Science Foundation (179239), the Foundation for Fight Against Cancer (Zuerich), the Wilhelm Sander Foundation to CL (2019.042.1), and the Novartis Foundation for medical-biological research to C.L. Furthermore, this project has received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 765104. We thank the Flow Cytometry Facility in Basel for support.
7-amminoactinomycin D (7-AAD) | Invitrogen | A1310 | Viability dye |
96 well plate U bottom | Sarstedt | 833925500 | 96 Well plate for our Flow cytometer |
APC Mouse Anti-Human cd34 | BD | 555824 | Antibody detecting CD34 RRID: AB_398614 |
Bovine Serum Albumin | PanReac AppliChem | A1391,0050 | Component of the staining buffer |
Ethylenediaminetetraacetic acid | Roth | 8043.1 | Component of the staining buffer |
Fetal Calf Serum (FCS) | BioConcept | 2-01F10-I | Component of the supplemented RPMI medium |
FlowJo 10.2 | BD | / | Software enabling data analysis for flow cytometry experiment |
Goat- anti-Rabbit IgG (H+L) Alexa Fluor 488 | Thermo Scientific | A21222 | Secondary antibody detecting the primary antibodies for MICA and MICB RRID: AB_1037853 |
Human NKG2D Fc Chimera Protein, CF | R&D | 1299-NK-050 | Fusion Protein detecting all NKG2DLs RRID: |
Human ULBP-1 Antibody | R&D | AF1380 | Antibody detecting ULBP1 RRID: AB_354765 |
Human ULBP-2/5/6 Antibody | R&D | AF1298 | Antibody detecting ULBP2/5/6 RRID: AB_354725 |
Human ULBP-3 Antibody | R&D | AF1517 | Antibody detecting ULBP3 RRID: AB_354835 |
MICA Polyclonal Antibody | Thermo Scientific | PA5-35346 | Antibody detecting MICA RRID: AB_2552656 |
MICB Polyclonal Antibody | Thermo Scientific | PA5-66698 | Antibody detecting MICB RRID: AB_2663413 |
One-step Antibody Biotinylation Kit 1 strip, for 8 reactions | Miltenyibiotec | 130-093-385 | Biotinylation kit for the NKG2DL fusion protein |
Phosphate Buffered Saline | Sigma Aldrich | D8537-500ML | Component of the staining buffer |
Rabbit anti-Goat IgG (H+L) Alexa Fluor 488 | Thermo Scientific | A11034 | Secondary antibody detecting the primary antibodies for the ULBPs RRID: AB_2576217 |
Rainbow Calibration Particles (8-peaks) 3.0 um | Spherotech Inc. | RCP-30-20A | Beads used for flow cytometry device maintainance |
RPMI medium | Sigma Aldrich | R8758-500ML | Cell culture medium |
RayBright Universal Compensation Beads | Raybiotech | 137-00013-100 | Beads used to create the compensation matrix |
Streptavidin, R-Phycoerythrin Conjugate (SAPE) – 1 mg/mL | Invitrogen | S866 | Seondary step for the biotinylated NKG2DL fusion protein detection |