Ligands of the prototypical activating NK receptor NKG2D render cancer cells susceptible to NK cell-mediated cytolysis if expressed at sufficiently high levels. However, malignant cells employ mechanisms to evade NKG2D-mediated immunosurveillance, such as NKG2D ligand (NKG2DL) shedding resulting in reduced surface expression levels. In addition, systemic downregulation of NKG2D on NK cells of cancer patients has been observed in many studies and was attributed to soluble NKG2DL (sNKG2DL), although there also are conflicting data. Likewise, relevant expression of NKG2DL in leukemia has been reported by some, but not all studies. Hence, we comprehensively studied expression, release, and function of the NKG2D ligands MHC class I chain-related molecules A and B and UL16-binding proteins 1–3 in 205 leukemia patients. Leukemia cells of most patients (75%) expressed at least one NKG2DL at the surface, and all investigated patient sera contained elevated sNKG2DL levels. Besides correlating NKG2DL levels with clinical data and outcome, we demonstrate that sNKG2DL in patient sera reduce NKG2D expression on NK cells, resulting in impaired antileukemia reactivity, which also critically depends on number and levels of surface-expressed NKG2DL. Together, we provide comprehensive data on the relevance of NKG2D/NKG2DL expression, release, and function for NK reactivity in leukemia, which exemplifies the mechanisms underlying NKG2D-mediated tumor immunosurveillance and escape.
Natural killer cells are cytotoxic lymphocytes that play an important role in antitumor immunity (1). Reports on reduced NK cell counts and activity as well as their prognostic relevance in leukemia patients have provided evidence for the involvement of NK cells in leukemia immunosurveillance (2–4). Further evidence for the ability of NK cells to combat leukemia in humans has been provided by clinical data from allogenic stem cell transplantation (5, 6). NK cell activation is guided by the principles of “missing self” and “induced self”, implying that NK cells kill target cells with low/absent expression of MHC class I (missing self) and/or stress-induced expression of ligands for activating NK receptors (induced self) (7–9). Meanwhile, missing-self recognition and the function of inhibitory NK cell receptors are well defined. The mechanisms governing NK activation are less well characterized, and, despite substantial progress over the recent years, some ligands that engage activating NK cell receptors are still unknown (10). A notable exception and, with regard to the induced-self recognition mode, paradigmatic example is the activating NKG2D receptor, which apart from NK cells is also expressed on CD8+ cytotoxic T cells, γδ T cells, and, under certain conditions, CD4+ T cells in humans (11, 12). NKG2D interacts with cell stress-induced, MHC class I-related molecules, which, in humans, comprise two members of the MHC class I-related chain (MIC) family (MICA, MICB) and six members of the UL16-binding protein (ULBP) family of proteins (ULBP1–4, RAET1G, RAET1L) (13–16).
Depending on surface expression levels, NKG2D ligands (NKG2DL) potently stimulate antitumor responses of cytotoxic lymphocytes (17, 18). However, tumor cells employ various mechanisms to evade NKG2D-mediated immune surveillance: we found that tumor cells downregulate MICA surface expression by proteolytic shedding, resulting in release of soluble MICA (sMICA) and, therefore, are the likely source of elevated sMICA levels in the sera of cancer patients (19). Groh et al. (20) showed that sMICA levels in sera of patients with MICA-expressing tumors correlate with NKG2D downregulation on CD8+ T cells, and these patient sera decreased NKG2D expression. In the following years, reduced expression of NKG2D on T and NK cells of cancer patients and release of MICA as well as other NKG2DL have been described by multiple studies. However, conflicting results have been reported whether in fact soluble NKG2DL (sNKG2DL) or other serum factors such as TGF-β cause diminished NKG2D expression in cancer patients (reviewed in Refs. 21 and 15). In addition, whereas the relevance of the NKG2D/NKG2DL system in NK cell immunosurveillance of epithelial tumors is well established, its role in leukemia remains controversial. To address these questions, we in this study collected primary samples from >200 patients to comprehensively analyze NKG2DL expression and release as well as NKG2D expression and modulation and their role for NK cell reactivity in leukemia.
Materials and Methods
Blood samples from adult leukemia patients were obtained at time of diagnosis prior to therapy or in complete remission (CR), as indicated. All patients gave their written informed consent in accordance with the Helsinki protocol, and the study was performed according to the guidelines of the local Ethics Committee. Diagnosis was confirmed by study of bone marrow specimens and flow cytometric immunophenotyping. Supplemental Table I summarizes the patients’ characteristics.
The mAb AMO1 (anti-MICA), BMO1 (anti-MICB), BAMO1 (anti-MICA/B), AUMO3 (anti-ULBP1), BUMO1 (anti-ULBP2), CUMO3 (anti-ULBP3), and the respective F(ab′)2 fragments, as well as the blocking NKG2D mAb E4 and 6H7, were previously described (22). The anti-mouse PE conjugate was from Jackson ImmunoResearch Laboratories (West Grove, PA). Anti-CD3 PeCy5, anti-CD3 FITC, anti-CD5 PeCy5, anti-CD19 FITC anti-CD33 PeCy5, anti-CD34 PeCy5, anti-CD56 FITC, and anti-CD56 PeCy7 conjugates were from BD Biosciences. Goat anti-mouse IgG1-HRP and IgG2a-HRP were from Southern Biotechnology Associates. BATDA and Europium solution were from PerkinElmer (Waltham, MA).
Generation of NKG2D fusion protein
An expression vector containing the Fc part of the human IgG1 molecule (P217-K447) lacking the CH1 domain and containing a C220S substitution to avoid intermolecular disulfide bonds was employed to add the extracellular domain of NKG2D (F78-V216) at the C terminus. To this end, a 2-aa linker (GS, encoded by tccgga, BspEI restriction site) was inserted at the end of the CH3 domain. To obtain NKG2D-Ig fusion proteins with abrogated affinity to FcγRIIIa (NKG2D-Fc-KO), the human Fc part was modified by amino acid substitutions E233P/L234V/L235A/ΔG236/A327G/A330S, as previously described (23). The construct was transfected into SP2/0-Ag14 cells by electroporation. Secreted fusion proteins were purified from culture supernatants (sn) by protein A affinity chromatography, and purity was determined by size exclusion chromatography.
Transfectants and generation of sNKG2DL-containing supernatants
C1R cells transfected with MICA*01 or ULBP1, respectively, as well as mock-transfected C1R cells were described previously (19, 24). For the production of sNKG2DL-containing and control sn, C1R transfectants were grown in IMDM medium without any additives for 72 h; then supernatants were collected and concentrated; sNKG2DL levels were determined by ELISA prior to addition in functional assays.
Preparation of NK cells
Polyclonal NK cells (pNK) were generated from nonplastic-adherent PBMC, as previously described (25). Functional experiments were performed when purity of pNK was >90%.
Cells were stained with the respective specific Ab or isotype control (10 μg/ml), followed by goat anti-mouse PE conjugate as secondary reagent (1:100) or with direct fluorescence conjugates. NK and leukemic cells were selected as follows: NK cells, CD56+CD3−CD33−; acute myeloid leukemia (AML), CD33/34+CD56−; B cell acute lymphoid leukemia (ALL), CD19+; T cell ALL, CD7; chronic myeloid leukemia (CML), CD34+; B cell chronic lymphoid leukemia (CLL), CD5+CD19+. Analysis was performed using a FC500 flow cytometer (Beckman Coulter, Krefeld, Germany). Specific fluorescence indices (SFI) were calculated by dividing median fluorescences obtained with the respective specific mAb by median fluorescences obtained with isotype control.
Cytotoxicity of NK cells against leukemia cells from patients was analyzed by 2-h BATDA Europium assays, as previously described (26). Percentage of lysis was calculated as follows: 100 × [(experimental release − spontaneous release)/(maximum release − spontaneous release)].
Determination of IFN-γ and sNKG2DL levels
IFN-γ levels were analyzed by ELISA using OptEIA sets from BD Biosciences, according to manufacturer’s instructions. Levels of sMICA, sMICB, and soluble ULBP2 (sULBP2) in sn and sera were determined by sandwich ELISA, as previously described (19, 24, 27
Where indicated, results were compared using Fisher’s exact test and Wilcoxon rank-sum test. All p values were two-sided, and p < 0.05 was considered statistically significant.
NKG2DL surface expression by leukemia cells
As available results regarding NKG2DL expression on leukemia cells, obtained in smaller patient cohorts, are partially conflicting (24, 28–37), we comprehensively analyzed the expression of the NKG2DL MICA, MICB, ULBP1, ULBP2, and ULBP3, for which highly specific Abs are available, as well as HLA class I expression on primary human AML, ALL, CML, and CLL cells from a total of 205 patients. Expression of individual NKG2DL on malignant cells of all patients is depicted in Fig. 1A, and, together with the clinical characteristics of each patient, summarized in Supplemental Table I. Exemplary results obtained by FACS analysis of patient cells from each leukemia entity are shown in Supplemental Fig. 1. Results from FACS analysis were mirrored by analyses of NKG2DL mRNA levels by quantitative RT-PCR, further confirming that NKG2DL are expressed by primary leukemia cells (Supplemental Fig. 2).
Overall, substantial surface expression of at least one NKG2DL ligand (SFI ≥ 1.5) was detected in 153 of the 205 (75%) investigated patient cases. Leukemic cells of AML (n = 104), ALL (n = 30), CML (n = 11), and CLL (n = 60) patients were positive for at least one NKG2DL in 70, 67, 82, and 85% of the investigated cases, respectively (Table I). Thus, NKG2DL expression on leukemic cells was detected significantly more often in chronic as compared with acute leukemias (Fig. 1B). Within all entities, in most cases only one NKG2DL was expressed, whereas several or even all analyzed NKG2DL were only detected in a small proportion of patients (Fig. 1C). In light of clinical data that indicate that ALL cells are less susceptible to NK reactivity as compared with AML upon haploidentical stem cell transplantation (32), it is noteworthy that, compared with AML cells, which expressed four or five different NKG2DL in 15% of the cases, leukemic cells of ALL patients less frequently expressed several and never more than three NKG2DL. Within the chronic leukemias, expression of multiple NKG2DL was detected in a substantial proportion of CLL cases, whereas CML cells never expressed more than two different NKG2DL (Fig. 1C).
Upon combined analysis of all four entities, MICA was most frequently detected, whereas ULBP2 was the least detectable NKG2DL. Within the subset of acute leukemias (AML and ALL), MICA was detected most frequently, whereas ULBP3 was observed most commonly in CML and CLL. Notably, relevant expression of ULBP3 with negativity for other NKG2DL was observed remarkably often in CML, in which we did not detect ULBP1 or ULBP2 (Table I). In AML, neither positivity nor levels (SFI) of the expressed NKG2DL correlated with clinical parameters like WBC count, blast count, cytogenetic risk, achievement of CR, or survival. Moreover, in CLL, no significant association of NKG2DL expression with disease stage according to Binet classification or white blood/lymphocyte count was observed (data not shown). Due to the relatively lower number of cases, no correlative studies were performed in ALL and CML.
Modulation of NK reactivity by leukemia-expressed NKG2DL
Next, we analyzed the influence of NKG2DL expression on NK antileukemia reactivity. These analyses were performed with primary AML, ALL, and CLL cells as targets, whereas CML cells were not employed for technical reasons. Cytotoxicity assays with allogenic pNK revealed that disruption of NKG2D–NKG2DL interaction by blocking the NKG2D receptor significantly reduced leukemia cell lysis (Fig. 2A). The relevance of NKG2D blocking for NK cell reactivity was also confirmed with K562 target cells (Supplemental Fig. 3), which express NKG2DL under the control of BCR-ABL (35, 36). To elucidate the involvement of single NKG2DL in cases in which more than one ligand was expressed, we employed specific blocking F(ab′)2 fragments. Blockade of one among several ligands reduced NK lysis, and additive effects were observed upon blocking several NKG2DL (Fig. 2B). As NKG2D, besides stimulating cytotoxicity, also promotes cytokine production by NK cells (38), we cultured primary leukemia cells and NK cells in the presence or absence of blocking NKG2D Ab or NKG2DL F(ab′)2 and studied NK cell IFN-γ production. In line with the results of the cytotoxicity assays, these experiments demonstrated the specific contribution of individual NKG2DL on leukemia cells to IFN-γ production by NK cells (Fig. 3). Experiments employing blocking Ab against NKp46, which has previously been shown to be involved in NK cell lysis of AML cells (32), confirmed the validity of our experimental setting (data not shown).
Prevalence of sNKG2DL in sera of the leukemia patients
As release of NKG2DL in soluble form is considered as mechanism by which malignant cells evade NK immunosurveillance, we next studied sNKG2DL levels in sera of 183 leukemia patients obtained at time of diagnosis (AML, n = 104; ALL, n = 22; CML, n = 11; CLL, n = 46). The levels of all five NKG2DL in patient serum were significantly higher as compared with healthy controls (n = 30), except for sULBP1 in AML, sULBP2 in CML, and sULBP3 in CLL (Fig. 4A). Notably, the sera of all patients in our cohort contained elevated levels of at least one sNKG2DL, and, in >90% of the cases, substantial levels of two or more sNKG2DL were detectable (Fig. 4B). No correlation of the levels of sNKG2DL in a given patient’s serum with expression of the respective NKG2DL on the surface of leukemic cells was observed (data not shown). Overall and also in each single leukemia entity, sMICA was detected most frequently, and sMICA levels were significantly higher in acute than in chronic leukemias. sULBP3 was detected least often upon combined analysis of all sera and within the lymphoid leukemia entities, whereas sULBP1 and sULBP2 were less frequent in sera of patients with myeloid leukemias. No further obvious differences with regard to prevalence and frequency of the different sNKG2DL among the four leukemia entities were observed (Table I). Correlative analysis with clinical parameters (again limited to AML and CLL due to the numbers of available samples) revealed that sera of AML patients that achieved a CR upon induction chemotherapy contained significantly (p < 0.05) lower levels of sULBP1 than therapy-refractory patients (Supplemental Table I and data not shown). The levels of sMICB and sULBP2 detectable at time of diagnosis of AML were significantly lower in patients that lived for >1 y after CR as compared with patients that had not survived (Fig. 4C). However, no significant results were obtained with regard to survival for 2 or more years. Moreover, no correlation of positivity for or levels of sNKG2DL with other clinical parameters like WBC count, blast/leukemic cell percentage, French–American–British classification, or cytogenetic risk was observed in AML (data not shown). In CLL, sULBP2 levels (but not levels of other NKG2DL) as well as the proportion of sULBP2-positive cases were significantly (p < 0.05) higher in patients with advanced disease (Binet stage B or C) as compared with Binet stage A (Fig. 4D).
NKG2D expression on NK cells of leukemia patients and modulation by patient serum
Diminished NKG2D expression on cytotoxic lymphocytes of cancer patients has been reported by several studies, whereas other investigators did not observe significant differences to healthy controls (15, 21). We analyzed NKG2D expression on the NK cells of AML (n = 30), ALL (n = 12), and CLL (n = 30) patients enrolled in our study and found that, compared with healthy donors (n = 20), NKG2D levels were significantly decreased in the leukemia patients (Fig. 5A). No correlation of NKG2DL expression on leukemic cells or sNKG2DL concentrations in patient sera and NKG2D expression on NK cells of leukemia patients was observed (data not shown). This could be due to the fact that only five of the eight presently known NKG2DL were analyzed in this study. Notably, NKG2D expression recovered upon remission of disease as revealed by our analyses studying NKG2D levels on NK cells from leukemia patients at time of active disease and in CR. For reasons of availability of sufficient sample numbers, these analyses were limited to AML (Fig. 5B). To determine whether factors contained in patient serum influence NKG2D expression, we exposed PBMC from healthy donors to sera of patients with AML, ALL, or CLL, or healthy volunteers. Significant NKG2D downregulation with maximal effects occurring after 12–16 h of incubation was caused by patient sera, but not by the sera of healthy controls, which indicates that soluble factors derived from the malignant cells modulate NKG2D expression in leukemia (Fig. 5C). Similarly, exposure to patient sera also resulted in reduced NKG2D expression on pNK from healthy donors, which allowed for their use in subsequent functional analyses (Fig. 5D).
Generation of NKG2D-Fc fusion proteins with abrogated FcγRIIIa binding
To unravel the potential involvement of sNKG2DL in NKG2D downregulation by patient serum, we generated NKG2D-Fc fusion proteins that, in contrast to NKG2DL-specific Abs, can neutralize all different sNKG2DL. To exclude that binding of NKG2D-Fc to FcγRIIIa affected NK cell reactivity or NKG2D expression, we introduced amino acid modifications that abrogate the affinity of human IgG1 to its receptor, as described in Materials and Methods (23). A schematic illustration of our construct (NKG2D-Fc-KO) is depicted in Fig. 6A. Binding analyses confirmed that NKG2D-Fc-KO, in contrast to a control construct containing the wild-type Fc part (NKG2D-Fc-WT), did not bind to NK cells, whereas no difference in NKG2DL binding of these two constructs was observed in analyses with C1R-ULBP1 transfectants (and also other NKG2DL transfectants; data not shown) and C1R-mock cells (Fig. 6B). Moreover, functional analyses revealed that NKG2D-Fc-KO, in contrast to NKG2D-Fc-WT, did not induce NK cell activation, as revealed by analysis of the activation marker CD69 (Fig. 6C) or IFN-γ production (Fig. 6D).
Contribution of sNKG2DL to NKG2D downregulation in leukemia
To determine how sNKG2DL contribute to NKG2D downregulation, NK cells were incubated in the presence or absence of sULBP1 derived from C1R-ULBP1 transfectants. sULBP1 clearly downregulated NKG2D, with maximal effects occurring after 12–16 h, whereas control supernatant had no relevant effect, and addition of NKG2D-Fc-KO largely prevented sULBP1-mediated NKG2D downregulation. Similar effects were observed with PBMC and pNK (that were subsequently used in functional experiments) of healthy donors, as well as PBMC of AML patients in CR after chemotherapeutic treatment, and also using supernatants of C1R cells transfected with other NKG2DL. Notably, substantial variation with regard to NKG2D downregulation and the effect of neutralizing sNKG2DL was observed, which may be due to technical issues, but could also have been influenced by donor variability (Fig. 7A, data not shown). These results clearly demonstrated that sNKG2DL can downregulate NKG2D expression and confirmed the neutralizing capacity of our fusion protein. Next, we cultured PBMC of healthy donors in sNKG2DL-containing sera of leukemia patients and healthy controls in the presence or absence of NKG2D-Fc-KO or isotype control. Sera of healthy donors in some cases caused a slight reduction of NKG2D expression that was not affected by the presence of NKG2D-Fc-KO. In contrast, pronounced NKG2D downregulation was caused by patient serum, and this was largely prevented by our fusion proteins (Fig. 7B). Moreover, when we exposed PBMC of leukemia patients obtained at time of CR to autologous serum either obtained at time of active disease or in CR, the first caused significant NKG2D downregulation, whereas the latter had no relevant effect. NKG2D-Fc largely prevented NKG2D downregulation induced by the active disease serum, although no effect was observed upon addition to CR serum. Again these analyses were restricted to patients with AML for reasons of availability of suitable samples (Fig. 7C). Together, these data demonstrate that leukemia-derived sNKG2DL can contribute to diminished NKG2D expression on NK cells of leukemia patients with active disease.
NKG2D downregulation impairs NK cell immunosurveillance in leukemia
Finally, we aimed to confirm that sNKG2DL, via NKG2D downregulation, can cause an impairment of NK cell reactivity in leukemia. To this end, we incubated pNK and PBMC of healthy donors in the presence or absence of sULBP1 with or without NKG2D-Fc, as described above. After modulating NKG2D expression in this manner, the pNK and PBMC were employed in cytotoxicity assays with C1R-mock and C1R-MICA transfectants serving as targets expressing NKG2DL at low and high levels, respectively. The critical dependence of NK reactivity on NKG2D/NKG2DL interaction in this setting was revealed by the significantly higher lysis of C1R-MICA as compared with C1R-mock by untreated NK cells. Antecedent exposure to control supernatant of C1R-mock cells did not alter C1R-MICA lysis. Pretreatment with supernatant-derived sULBP1 at a final concentration of 1 ng/ml, which is representative of sNKG2DL content in patient sera and causes a substantial NKG2D downregulation (Fig. 7A), resulted in significantly reduced (and in some cases with PBMC as effector cells nearly abrogated) cytotoxicity. Neutralization of sULBP1 by our NKG2D-Fc-KO nearly completely restored NK reactivity (Fig. 8A). Similar results were observed with regard to IFN-γ production. The functional relevance of NKG2D downregulation by sNKG2DL was also revealed by analyses with primary leukemia cells from AML patients. Both cytotoxicity and IFN-γ production of pNK in response to NKG2DL-positive patient AML cells were substantially reduced after exposure to sNKG2DL for 16 h, and this could be prevented by NKG2D-Fc-KO. In contrast, NKG2D downregulation by exposure to sULBP1 did not alter NK reactivity when NKG2DL-negative primary AML cells were used as targets (Fig. 8B). Finally, we employed sNKG2DL-containing AML patient sera and our NKG2D-Fc-KO to demonstrate the functional relevance of NKG2D downregulation in analyses with our NKG2DL-high and -low C1R transfectants. Similar to the analyses with sULBP1, NKG2D downregulation by patient serum substantially impaired NK cytotoxicity and cytokine production in response to NKG2DL-expressing targets, and this could be prevented to a large extent by sNKG2DL neutralization (Fig. 8C). Together, these results confirm both the relevance of NKG2DL expression on leukemia cells for NK cell immunosurveillance and the involvement of sNKG2DL in NKG2D downregulation, which results in impaired NK reactivity against leukemia targets.
NKG2D is a conserved homodimeric, type II transmembrane C-type lectin-like receptor that potently stimulates cytotoxicity and production of immunostimulatory cytokines like IFN-γ by NK cells in mice and humans (39). The stress-induced NKG2DL are expressed by various types of cancers, but largely absent from healthy cells, which implied that NKG2D plays an important role in immunosurveillance and made NKG2D one of the most intensively studied immunoreceptors of the past decade (15, 16). Recent studies with NKG2D-deficient mice provided evidence that NKG2D in fact is involved in tumor immunosurveillance (40). However, there is also clear evidence that tumor cells employ efficient strategies to evade NKG2D-mediated NK antitumor reactivity: tumor cells can shed NKG2DL from the cell surface, thereby reducing expression levels and thus the amount of stimulatory signals that determine whether NK responses are initiated or not. In addition, tumor-derived sNKG2DL in cancer patient sera may cause systemic NKG2D downregulation, thereby further impairing NK immunosurveillance. However, at least partially discrepant results have been reported whether in fact sNKG2DL or other tumor-derived factors in patient serum are responsible for NKG2D downregulation on cancer patient NK cells. Moreover, in contrast to epithelial tumors, in which the relevance of NKG2D and its ligands for immunosurveillance is well accepted, at least partially conflicting data on the expression and function of this surveillance system in leukemia were reported. In this study, we performed a comprehensive analysis of NKG2DL surface expression and prevalence of sNKG2DL in >200 patients with AML, ALL, CLL, and CML. We circumstantiate the capacity of leukemia-expressed NKG2DL to stimulate NK cell responses and demonstrate that diminished NKG2D expression, which we observed on NK cells of leukemia patients, can in fact be caused by sNKG2DL and results in impaired antileukemia reactivity.
After our initial report that beyond epithelial tumors also leukemia cells can express NKG2DL (24), expression of one or more NKG2DL on leukemic cells of patients with AML (28–30), ALL (31, 32), CLL (29, 33), and CML (34–36) was also reported in other studies. However, two studies reported only low or even no relevant levels of NKG2DL on AML cells in cohorts of 24 and 30 patients (32, 37). We in this study detected substantial expression of at least one NKG2DL in 75% of the 205 investigated patients (AML, n = 104; ALL, n = 30; CML, n = 11; CLL, n = 60), with the percentage of positive cases being lowest in ALL and highest in CLL. Expression patterns varied substantially among the patients, which may explain seemingly contradictory results of studies that determined expression of only one or a limited number of NKG2DL in smaller cohorts. Also, in partial contrast to other studies that reported a correlation of NKG2DL expression, for example, with later stages of differentiation in AML (28, 37), our statistical analyses revealed no association of NKG2DL expression with certain French–American–British classification types or clinical parameters, such as achievement of CR or survival. These discrepancies may be due to technical issues like sensitivity/specificity of reagents used for NKG2DL analysis, but could also be attributable to differences among the study populations. It is also noteworthy that detection of NKG2DL mRNA did not always translate into detectable NKG2DL surface expression in our study. Although the involved mechanisms certainly comprise NKG2DL shedding, several studies demonstrated that expression of NKG2DL on leukemic cells is upregulated in vitro and in vivo by therapeutic compounds like trichostatin A (31), all-trans-retinoic acid (29), valproic acid/sodium valproate (28, 29), or hydroxyurea (41). These findings, together with data that MIC molecule expression is regulated by microRNA (42), implicate that cell-intrinsic mechanisms may also largely contribute to posttranscriptional regulation of NKG2DL expression.
Notably, cytotoxicity and IFN-γ production of allogenic NK cells in response to primary leukemia cells were significantly impaired upon disruption of NKG2D–NKG2DL interaction. When leukemia cells expressed more than one NKG2DL, additive effects were observed upon blocking several ligands, which is in line with data demonstrating that NK reactivity critically depends on NKG2DL expression levels (17, 18). Our results are in agreement with findings of several other investigators who reported a functional relevance of NKG2DL expression for NK antileukemia reactivity (28, 29, 31, 34, 35, 37). However, other investigators did not observe relevant NKG2DL expression and, in line with this finding, reported that NK-mediated lysis of AML cells occurred independently of NKG2D in their experimental setting (32). Technical conditions or a selection bias of the study population may be responsible for this discrepancy.
Next, we set out to determine whether sNKG2DL impair NK immunosurveillance of hematopoietic malignancies. After the initial studies on sMICA in sera of cancer patients, elevated levels of sNKG2DL have been reported by multiple studies in patients with various solid tumors (reviewed in Refs. 21 and 15), but notably also in patients with hematopoietic malignancies (24, 27–29, 33–35). In our large patient population, significantly elevated levels of all five investigated NKG2DL were detectable in all leukemia entities except sULBP2 in CML, sULBP3 in CLL, and sULBP1 in AML. The last is in line with the results of Diermayr et al. (28), who also observed elevated levels of sMICA, but not of sULBP1 in AML. In our study, sNKG2DL were detectable in sera of patients without expression of the respective NKG2DL on the leukemic cells. NKG2DL can be released in soluble form by various mechanisms, including shedding from the cell surface by proteases, which may serve to explain this finding and indicates that leukemia cells may evade NKG2D-mediated immunosurveillance by reduction of cell surface NKG2DL density (19, 27, 43–46). Notably, all analyzed leukemia patient sera were positive for at least one of the five analyzed sNKG2DL, with sMICA being most frequently detectable and at significantly higher levels in acute as compared with chronic leukemias. Correlative analyses in AML, in which the highest number of patients was available, revealed that low levels of sMICB and sULBP2 correlated with survival for longer than 1 y after CR. Notably, NKG2DL vary with regard to regulation of their expression, surface attachment, affinity to NKG2D, etc. (16). Such still incompletely understood and possibly differing functional properties may provide an explanation as to why the various NKG2DL differ with regard to correlation with clinical parameters. However, despite the statistical significance, such considerations and the lacking association with survival at later time points in our cohort raise some doubt as to whether determination of sNKG2DL levels may in fact be clinically useful for prognostic purposes, at least in AML. Another study recently reported on significantly elevated levels of sMICA/B and sULBP2, but not sULBP1 and sULBP3 in CLL, with sULBP2 being further significantly increased in advanced stage patients and being significantly associated with disease progression (33). In partial contrast, we found that beyond sMICA, sMICB, and ULBP2, also sULBP1 levels were significantly elevated in CLL patient sera. Levels of sULBP2, but not levels of other sNKG2DL, were significantly higher in patients with advanced disease. Discrepancies could again be due to technical differences or the differing study cohorts. Overall, we believe that further validation in substantially larger cohorts is needed before the clinical value of sNKG2DL determination can be judged ultimately.
A central aspect of our study was to clearly elucidate the involvement of sNKG2DL in NKG2D downregulation and immune escape. Several previous studies reported reduced NKG2D expression on cytotoxic lymphocytes of patients with malignancies, but others did not observe differences between healthy donors and cancer patients (reviewed in Refs. 21 and 15). Moreover, several investigators attributed reduced surface NKG2D to downmodulation by membrane-bound NKG2DL or other soluble factors than sNKG2DL contained in patient serum such as TGF-β or l-kynurenine (47, 48). With regard to hematopoietic malignancies, three studies reported on significantly fewer NKG2D+ NK cells or reduced NKG2D levels in patients with CML and AML as compared with healthy donors, whereas no difference was observed in two other reports (30, 33–35, 49). Our analyses revealed a clear reduction of NKG2D expression levels on NK cells of leukemia patients in all investigated leukemia entities, and further evidence that leukemic disease causes altered NKG2D expression derived from our finding that reduced NKG2D levels in AML patients recovered in CR. To determine whether in fact sNKG2DL cause NKG2D downregulation on patient NK cells and whether this affects NK immunosurveillance, we genetically engineered NKG2D-Fc fusion proteins that are capable of neutralizing all different NKG2DL without affecting NK cells. By using this NKG2D-Fc-KO in analyses with sNKG2DL-containing supernatants and leukemia patient sera, we provide clear results on the following issues: 1) Not only surface NKG2DL, but also sNKG2DL are capable of downregulating NKG2D expression on NK cells. 2) sNKG2DL can largely contribute to diminished NKG2D expression on NK cells of leukemia patients with active disease. 3) NKG2D downregulation clearly impairs reactivity of NK cells against NKG2DL-expressing patient leukemia cells. Because the latter was functionally relevant in our allogenic setting, it seems likely that it may be of even greater relevance in an autologous situation in which no relevant MHC class I/killer cell Ig-like receptor mismatch aids NK activation. 4) NKG2D expression is restored in remission of disease, after which NKG2DL expression on leukemic cells may be beneficial to maintain NK immunosurveillance and to enable elimination of residual disease.
Beyond increasing our understanding of the mechanisms that influence NKG2D-mediated NK immunosurveillance in general, our data indicate that due to the relatively widespread expression of NKG2DL in leukemia and their potency to induce NK reactivity, strategies to reinforce NK cell reactivity in malignant hematopoietic diseases should rather focus on preventing NKG2D downregulation or restoring its expression than increasing NKG2DL expression. Such approaches may serve to better implement NK cells in the treatment of leukemia and to ultimately improve the prognosis of patients with hematopoietic malignancies.
The authors have no financial conflicts of interest.
We thank Aline Naumann for biometric-methodic consulting.
This work was supported by grants from Deutsche Forschungsgemeinschaft (SFB-685 TP A7), Wilhelm Sander-Stiftung (2007.115.2), and Deutsche Krebshilfe (109620).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- acute lymphocytic leukemia
- acute myeloid leukemia
- chronic lymphocytic leukemia
- chronic myeloid leukemia
- complete remission
- MHC class I chain-related molecule
- NKG2D ligand
- polyclonal NK cell
- specific fluorescence index
- soluble MIC
- culture supernatant
- soluble NKG2DL
- soluble UL16-binding protein
- UL16-binding protein.
- Received March 13, 2012.
- Accepted June 1, 2012.
- Copyright © 2012 by The American Association of Immunologists, Inc.