Abstract
The TNF family member receptor activator for NF-κB ligand (RANKL) and its receptors RANK and osteoprotegerin are key regulators of bone remodeling but also influence cellular functions of tumor and immune effector cells. In this work, we studied the involvement of RANK–RANKL interaction in NK cell–mediated immunosurveillance of acute myeloid leukemia (AML). Substantial levels of RANKL were found to be expressed on leukemia cells in 53 of 78 (68%) investigated patients. Signaling via RANKL into the leukemia cells stimulated their metabolic activity and induced the release of cytokines involved in AML pathophysiology. In addition, the immunomodulatory factors released by AML cells upon RANKL signaling impaired the anti-leukemia reactivity of NK cells and induced RANK expression, and NK cells of AML patients displayed significantly upregulated RANK expression compared with healthy controls. Treatment of AML cells with the clinically available RANKL Ab Denosumab resulted in enhanced NK cell anti-leukemia reactivity. This was due to both blockade of the release of NK-inhibitory factors by AML cells and prevention of RANK signaling into NK cells. The latter was found to directly impair NK anti-leukemia reactivity with a more pronounced effect on IFN-γ production compared with cytotoxicity. Together, our data unravel a previously unknown function of the RANK–RANKL molecule system in AML pathophysiology as well as NK cell function and suggest that neutralization of RANKL with therapeutic Abs may serve to reinforce NK cell reactivity in leukemia patients.
Introduction
Natural killer cells are cytotoxic lymphocytes that play an important role in anti-tumor immunity (1). Their involvement in immunosurveillance of hematopoietic malignancies and in particular acute myeloid leukemia (AML) is highlighted by studies on haploidentical stem cell transplantation \(SCT), where the recipient’s leukemia cells fail to inhibit donor NK cells via killer Ig-like receptors (KIRs), and KIR disparity is associated with powerful graft versus leukemia effects and better clinical outcome (2–4). The observation that leukemia cells may downregulate HLA class I molecules (5, 6), presumably to escape adaptive immunity, suggests that NK cells are also involved in controlling leukemia in an autologous setting. This is also supported by the observation that NK cell counts and activity are reduced in patients with leukemia and that activity levels of autologous NK cells are associated with survival of leukemia patients (7–9). Because NK cell reactivity is governed by a balance of multiple inhibitory and activating receptors, interaction of NK cells and leukemia cells is dependent on various immunoregulatory molecules far beyond HLA class I–specific inhibitory KIR receptors (10, 11). Among others, many members of the TNF/TNFR family influence NK cell activation, and several TNF/TNFR family members have been found to be expressed on AML cells and influence anti-leukemia reactivity of NK cells that express their respective counterpart (12–14).
The TNFR family member receptor activator for NF-κB (RANK; TNFRSF11A) and its ligand (receptor activator for NF-κB ligand; RANKL) are mainly known for their key role in regulating bone metabolism (15, 16) but were also found to influence the interaction of dendritic cells (DCs) and T cells as well as the pathophysiology of hematopoietic malignancies and metastasis of solid tumors (17–25). Notably, available data indicate that RANK is also expressed on NK cells (26), but to date nothing is known regarding the functional relevance of RANK–RANKL interaction for NK cell reactivity. We report in this study that AML cells express RANKL in a high proportion of cases and that NK cells of AML patients display upregulated expression of its counterpart RANK. To account for the fact that several TNF/TNFR family members may mediate different effects in mice and men (14, 27, 28), we set out to study the role of RANKL in AML pathophysiology by using primary malignant cells of leukemia patients and examined the functional relevance of RANK–RANKL interaction for NK cell immunosurveillance by using PBMCs of healthy donors as effector cells thereby mimicking the situation in patients that undergo allogeneic SCT.
Materials and Methods
Patients
PBMCs of AML patients obtained at the time of diagnosis before therapy and of healthy donors were isolated by density gradient centrifugation after obtaining informed consent in accordance with the Helsinki protocol. In functional analyses, PBMCs of patients with >80% blast count according to differential blood count in blood smears were used without further purification to avoid potential artifacts by Ab-based isolation techniques. The study was performed according to the guidelines of the local ethics committee.
Transfectants and cell lines
The RANKL-transfectants (L-RANKL) as well as parental controls (L cells) were previously described (29). The human AML cell line HL-60 was obtained internally at Eberhard Karls University Tuebingen. Authenticity was determined by validating the respective immunophenotype described by the provider using FACS every 6 mo and specifically prior to use in experiments.
Abs and reagents
The mAbs against RANKL (MIH24 and MIH23) and the RANK–Ig fusion protein were previously described (29, 30
Flow cytometry
FACS was performed using specific mAb, RANK–Ig, and the respective isotype controls at 10 μg/ml followed by specific PE-conjugates (1:100). Analysis was performed using a FC500 (Beckman Coulter, Krefeld, Germany). Where indicated, specific fluorescence indices (SFI) were calculated by dividing median fluorescences obtained with specific mAb by median fluorescences obtained with isotype control. To exclude potential artifacts due to unspecific Ab binding, a threshold for defining surface positivity was set at SFI ≥ 1.5.
PCR analysis
RT-PCR was performed as described previously (31). The following primers were used for nested PCR of RANKL splice variants: membrane-bound RANKL (accession number NM_003701; http://www.ncbi.nlm.nih.gov/nuccore/), 5′-CGTCGCCCTGTTCTTCTATT-3′ and 5′-TATGGGAACCAGATGGGATG-3′ (step 1; 353 bp) and 5′-TCAGAAGATGGCACTCACTG-3′ and 5′-TGAGATGAGCAAAAGGCTGA-3′ (step 2; 268 bp); soluble RANKL (accession number NM_033012; http://www.ncbi.nlm.nih.gov/nuccore/), 5′-CTTAGAAGCCACCAAAGAATTG-3′ and 5′-TATGGGAACCAGATGGGATG-3′ (step 1; 347 bp) and 5′-TCAGAAGATGGCACTCACTG-3′ and 5′-TGAGATGAGCAAAAGGCTGA-3′ (step 2; 268 bp). Primers for 18S rRNA were 5′-CGGCTACCACATCCAAGGAA-3′ and 5′-GCTGGAATTACCGCGGCT-3′ (186 bp).
Determination of soluble RANKL, cytokines, and metabolic activity of AML cells
Levels of soluble RANKL (sRANKL) in sera of AML patients and healthy donors were analyzed using the commercially available ELISA BI-20452 from Biomedica (Vienna, Austria) according to the manufacturer’s instructions. Results shown are means of duplicates.
TNF, IL-6, IL-8, and IL-10 in culture supernatants, sometimes after dilution, were determined by ELISA using OptEIA sets from BD Biosciences (San Diego, CA) or by Matched Pairs for ELISA from ImmunoTools GmbH according to the manufacturer’s instructions. Cytokine concentrations in supernatants are expressed as means of triplicates.
Metabolic activity of AML cells was measured using the Cell Proliferation Reagent WST-1 set from Roche according to the manufacturer’s instructions. Absorbance was measured at 450 nm with 650 nm as reference wavelength, and results are shown as means of triplicate measurements.
NK cell activation, degranulation, cytotoxicity, and cytokine production
Upregulation of CD69 and CD107a as markers for NK activation and degranulation, respectively, was analyzed by FACS. NK cells within PBMCs were selected by staining for CD56+CD3−. Cytotoxicity was analyzed by 2 h BATDA europium release assays as previously described (13). IFN-γ production was analyzed using the ELISA mAb set from Thermo Scientific (Rockford, IL) according to the manufacturer’s instructions. Lysis rates and cytokine concentrations in supernatants are shown as means of triplicate measurements in each experiment.
Results
Expression of RANKL in AML
Recent studies suggested an involvement of RANKL in disease pathophysiology of chronic lymphoid leukemia (CLL) and multiple myeloma as well as metastatic spread of solid tumors (19–25, 32), but nothing was yet known on the role of the RANK–RANKL molecule system in AML. In this study, we analyzed RANKL expression on primary AML cells by flow cytometry and selected malignant cells among PBMCs of leukemia patients by staining for CD33 and CD34. Overall, substantial surface expression (SFI ≥ 1.5) was detected in 53 of 78 (68%) investigated AML patient samples. Individual SFI levels and the clinical characteristics of each patient are given in Table I. In Fig. 1A, RANKL surface expression on AML cells from selected patients is shown. Because RANKL expression patterns differed between individual patients, we studied whether RANKL was associated with certain AML French–American–British classification (FAB) types. The fractions of RANKL-positive (SFI ≥ 1.5) samples among that of the investigated patients with different FAB types were as follows: M0, 2 of 5 (40%); M1, 10 of 11 (91%); M2, 16 of 26 (62%); M3, 2 of 5 (40%); M4, 15 of 21 (71%); and M5, 8 of 10 (80%). Statistical analysis revealed that RANKL expression was not significantly (p = 0.33, one-way ANOVA) associated with any specific FAB type (Fig. 1B). Moreover, no significant association of RANKL expression with cytogenetic risk, a particular cytogenetic abnormality, blast count, or white blood count was observed (data not shown). RANKL expression of leukemia cells was also confirmed at the mRNA level by RT-PCR. Amplicons of membrane-bound RANKL (mRANKL) were detected in all 30 investigated samples. Notably, this also comprised samples without detectable surface expression on leukemic cells, which could be due to contamination with RANKL-expressing healthy cells like B and T cells (19, 29), but may also indicate that RANKL surface expression is regulated posttranscriptionally. The mRNA splice variant specifically coding for sRANKL was detected in 5 (17%) of the samples (Fig. 1C and Table I). Comparative analysis using sera of leukemia patients and healthy donors (where mRNA for sRANKL was never detectable in PBMCs, data not shown) revealed that sRANKL protein levels are not significantly (p = 0.65, Mann–Whitney U test) elevated in AML (Fig. 1D). Moreover, sRANKL was also not detectable in supernatants of patient AML cells upon in vitro culture (data not shown). Notably, these analyses comprised those AML patient samples that displayed amplicons of the alternatively spliced mRNA of sRANKL. This indicates that presence of mRNA for RANKL does not necessarily result in detectable protein expression, which lends further evidence that RANKL expression may substantially be influenced by posttranscriptional and/or posttranslational mechanisms in individual AML patients.
RANKL expression in AML. (A and B) RANKL surface expression on patient AML cells was investigated by FACS using the RANKL mAb MIH24 with mouse IgG2b serving as isotype control. Malignant cells were gated based on CD33 and CD34 staining. (A) Representative results from exemplary AML patients with different FAB types; numbers in histograms represent uniform patient number (UPN) as shown in Table I. Shaded peaks, anti-RANKL; open peaks, isotype control. (B) SFI levels obtained by analysis of 78 AML patients grouped according to their FAB subtypes and medians of results obtained with all patients analyzed (bar). The dotted line represents SFI 1.5 as defined threshold for surface positivity. No significant association with specific FAB types was observed (p = 0.33, one-way ANOVA). (C) PBMCs from AML patients (>80% blast count) were investigated for RANKL mRNA expression by RT-PCR with 18S rRNA serving as control. Exemplary results obtained with RANKL surface-positive and RANKL surface-negative cells of patients with different FAB types (identified by their respective UPN) are shown. mRANKL, splice variant for membrane-bound RANKL; sRANKL, splice variant for soluble RANKL; +, positive control. (D) Levels of sRANKL in sera of healthy donors and AML patients (n = 20 and 44, respectively) were analyzed by ELISA. Results obtained with single patients are depicted; the bar indicates the median of the results. Statistical analysis was performed using the Mann–Whitney U test.
RANKL transduces activating signals into AML cells
Many TNF family members including RANKL are able to mediate bidirectional signals (19, 33, 34). Therefore, we studied whether AML-expressed RANKL was able to transduce reverse signals that influence the cellular activity of the leukemia cells. To this end, primary AML cells of 10 different patients were cultured alone, on isotype control or immobilized RANK–Ig, which enables RANKL multimerization. Subsequent analysis by ELISA revealed that RANKL signaling significantly (all p < 0.05, Mann–Whitney U test) enhanced the release of TNF, IL-6, IL-8, and IL-10, which are associated with AML pathophysiology (35, 36), into the culture supernatants (Fig. 2A). No effects were observed with RANKL-negative AML cells, confirming that the effects of the RANK–Ig were specifically due to RANKL signaling (data not shown). Notably, we observed substantial interindividual differences concerning the cytokine release of AML cells upon RANKL signaling: Only with 8 of 17 investigated samples we observed release of all four cytokines by the AML cells. Release of TNF, IL-6, IL-8, and IL-10 was observed with 16, 14, 15, and 10 of the 17 samples, respectively (Fig. 2B). RANKL may thus (variably) contribute to the cytokine milieu associated with AML. Moreover, RANKL signaling potently and statistically significantly (p < 0.05, Mann–Whitney U test) stimulated cellular activity of AML cells as revealed by WST-1 proliferation assays performed with eight RANKL-positive patient samples, whereas no significant (p = 0.7, Mann–Whitney U test) effects were observed in analyses with leukemia cells of three RANKL-negative patients. Notably, basal proliferation rates observed with samples of different patients varied substantially without a significant association with RANKL positivity (Fig. 2C and data not shown). Together, these data indicate that signaling via RANKL may contribute to AML pathophysiology.
RANKL signaling induces cytokine release and metabolic activity of AML cells. PBMCs of AML patients (all >80% blast count) were cultured alone, on immobilized RANK–Ig or human IgG1 as control. (A and B) The levels of TNF (after 6 h), IL-6, IL-8, and IL-10 (all after 24 h) in culture supernatants were determined by ELISA. (A) Results obtained with RANKL-positive AML cells of 10 different patients and medians of results (bar) for each cytokine are shown. (B) Analysis of results obtained with 17 RANKL-positive patient samples with regard to release of specific cytokine combinations. Positive response (+) was defined as ≥3-fold increase of each individual cytokine upon RANKL signaling. The percentage of samples responding with the indicated cytokine pattern is depicted. (C) Metabolic activity of AML cells was determined by WST proliferation assays after 48 h. Results of one representative experiment each with leukemia cells of RANKL-positive and RANKL-negative patients and combined results obtained in independent experiments with AML cells of different RANKL-positive (n = 8, upper panels) and RANKL-negative (n = 3, lower panels) patients are shown. *p < 0.05 (Mann–Whitney U test).
AML-derived factors released upon RANKL signaling inhibit NK cell reactivity
Next, we studied how the factors released by AML cells upon RANKL signaling influence NK cell anti-tumor immunity. PBMCs of healthy donors were cultured with the RANKL-negative (data not shown) leukemia cell line HL-60 in the presence of supernatants from patient AML cells that had been generated upon RANKL-signaling as described earlier. Target cell lysis was markedly decreased by the factors released upon culture on RANK–Ig compared with experiments with supernatants derived from AML cells cultured alone or on human IgG1 as isotype control. This effect was clearly dependent on RANKL expression by AML cells, as no effects were observed with supernatants from RANKL-negative leukemia cells (Fig. 3A). Reduced anti-leukemia reactivity upon exposure to RANKL-induced factors was also observed when CD107a expression as surrogate marker for granule mobilization on CD56+CD3− cells among treated PBMCs was analyzed, which also confirmed that specifically the reactivity of NK cells was affected (Fig. 3B). Statistical analysis of the effects on cytotoxicity and degranulation in independent experiments with RANKL-positive and RANKL-negative patient cells (cytotoxicity: n = 46 and n = 8, respectively; degranulation: n = 34 and n = 7, respectively) revealed that only supernatants obtained upon reverse signaling into RANKL-positive patient cells caused a statistically significant (both p < 0.05, Mann–Whitney U test) reduction of NK cell reactivity (Fig. 3C). Thus, RANKL enables leukemia cells to release factors that impair NK cell immunity.
Factors released by AML cells upon RANKL signaling impair NK cell reactivity. Primary RANKL-positive and RANKL-negative AML cells of patients with >80% blast count were cultured alone, on immobilized RANK–Ig or human IgG1 as control for 24 h before harvest of culture supernatants (sn). Then, PBMCs of healthy donors as effectors were cultured with HL-60 cells in culture supernatants diluted 1:3 with fresh medium. Exemplary results with leukemia cells of the indicated (UPN) patients in analyses of (A) cytotoxicity by 2 h BATDA europium release assays and (B) CD107a upregulation on NK cells (CD56+CD3−) after 3 h by FACS (percentages of CD107a-positive NK cells are given) are shown. (C) Combined analysis of the effects of supernatants from RANKL-positive and RANKL-negative patient AML cells. Results obtained with untreated PBMCs in each individual data set were set to 1. Left, Cytotoxicity (n = 46 and n = 8, respectively). Right, CD107a expression (n = 34 and n = 7, respectively). *p < 0.05 (Mann–Whitney U test).
Denosumab blocks immunomodulatory effects of RANKL signaling
On the basis of the central role of RANKL in bone metabolism, the mAb Denosumab, capable of blocking RANKL, was developed and recently proved to be effective for treatment of nonmalignant and malignant osteolysis (37, 38). We found that Denosumab specifically bound to patient AML cells, thereby also confirming our results on RANKL expression by primary AML cells (Fig. 4A). We further confirmed that Denosumab is able to prevent binding of RANK to its ligand in our experimental system (Supplemental Fig. 1). Next, we studied whether Denosumab was able to block RANK-induced signaling via RANKL into primary AML cells. Denosumab treatment statistically significantly (all p < 0.05, Mann–Whitney U test) diminished the RANKL signaling–induced release of TNF, IL-6, IL-8, and IL-10 by the AML cells, whereas isotype control had no effect (Fig. 4B). Importantly, also the inhibitory effects of the factor(s) released by AML cells upon RANKL signaling on NK reactivity as determined by analyses of degranulation were significantly (p < 0.05, Mann–Whitney U test) reduced when Denosumab was present (and thus able to block RANK–RANKL interaction) during generation of the culture supernatants that were subsequently used in assays with RANKL-negative HL-60 cells and PBMCs of three independent donors (Fig. 4C). Thus, Denosumab may serve to decrease the release of RANKL-induced factors from AML cells that contribute to disease pathophysiology and immune evasion from NK cell reactivity.
Denosumab blocks RANKL signaling into malignant cells and its NK-inhibitory effects. (A) Binding of Denosumab to the indicated RANKL-positive or RANKL-negative primary AML cells was analyzed by FACS. Shaded peaks, Denosumab; open peaks, isotype control. (B) PBMCs of AML patients (all >80% blast count) were cultured alone, on immobilized RANK–Ig to induce RANKL signaling (black bars) or human IgG1 as control (white bars). Where indicated, AML cells were treated for 1 h with Denosumab or isotype control (10 μg/ml each) followed by washing prior to induction of RANKL signaling. Release of TNF (after 6 h), IL-6, IL-8, and IL-10 (all after 24 h) was determined by ELISA of culture supernatants. Data of one representative experiment each are shown in the upper panels, and lower panels depict results of experiments with leukemic cells of at least four different patients. (C) PBMCs of healthy donors were cultured with HL-60 cells in the presence or absence of culture supernatants (sn) generated as described in (B) diluted 1:3 with fresh medium. CD107a expression levels on NK cells (CD56+CD3−) were analyzed after 3 h by FACS. To account for donor variation and to enable statistical analysis, combined data of three independent experiments with PBMCs of different donors after normalization by setting results obtained with PBMCs cultured in control supernatants to 100% are shown. *p < 0.05 (Mann–Whitney U test).
AML-derived factors induce RANK expression on NK cells
Numerous TNF/TNFR family members are expressed by NK cells and are upregulated in malignant disease, where they influence NK reactivity upon interaction with their target cell–expressed counterparts (13, 14, 39–41). As available data indicated that RANK can be expressed by NK cells (26), we comparatively analyzed RANK expression on NK cells of healthy donors and our AML patients. Only low levels of RANK were detected on NK cells of healthy donors, whereas significantly (p < 0.05, Mann–Whitney U test) increased expression was observed on NK cells of the leukemia patients (Fig. 5A). To unravel the mechanism(s) responsible for the differential RANK expression on NK cells of patients, freshly isolated PBMCs of healthy donors were incubated with supernatants of primary AML cells that had been cultured alone, on immobilized RANK–Ig or isotype control. Afterwards, RANK expression on NK cells was determined by FACS analyses. Notably, NK cells cultured in fresh medium acquired low levels of RANK, likely due to unspecific effects of the isolation procedure and/or in vitro culture. Control supernatants derived from AML cells only slightly modulated RANK expression, whereas that obtained upon RANKL signaling into the leukemia cells significantly (p < 0.05, Mann–Whitney U test) induced RANK upregulation on NK cells (Fig. 5B). Next, we treated PBMCs from healthy donors with the cytokines that we had found to be released upon RANKL signaling by AML cells. Again, only low levels of RANK were detected on resting NK cells and expression increased over time upon in vitro culture in medium alone. Whereas TNF, IL-6, and IL-8 had no effect, RANK expression on NK cells was strongly upregulated in the presence of IL-10. Induction of RANK expression was not associated with NK cell activation, as no substantial upregulation of the activation marker CD69 was observed upon IL-10 treatment (Fig. 5C and data not shown). Notably, RANK was also upregulated by supernatants of AML cells that did not respond with secretion of IL-10 to RANKL signaling. This indicates that IL-10 is sufficient but not necessary to induce RANK, and other yet unidentified factors that are released upon RANKL signaling also modulate NK cell RANK expression (Fig. 5D). Taken together, RANK expression on NK cells can be induced by factors released by AML cells, which turn may facilitate RANK–RANKL interaction that subsequently could directly influence NK cell reactivity via RANK.
Factors released by AML cells upon RANKL signaling enhance RANK expression on NK cells. Expression of RANK on CD56+CD3− NK cells was analyzed by FACS. (A) Comparative analysis of PBMCs from AML patients and healthy donors (n = 60 each). The percentage of RANK-positive NK cells is indicated. Results obtained with single patients are depicted; bar indicates medians of measurements. (B) Results obtained with PBMCs of healthy donors after 72 h of culture in the presence of 1:3 diluted supernatants (sn) generated by incubation of AML cells of different patients alone, on immobilized RANK–Ig or human IgG1 as control (n = 12). Results were normalized in each individual data set by setting expression in PBMCs cultured in medium to 1. (C) Results obtained upon incubation of healthy PBMCs in the presence or absence of recombinant IL-10 (10 ng/ml) for the indicated times. The activation marker CD69 was determined in parallel. The percentages of RANK-positive and CD69-positive NK cells are indicated. (D) Comparative analysis of RANK induction on NK cells within PBMCs as determined in (B) upon exposure to supernatants of AML cells that did (IL-10 +, n = 9) and did not (IL-10 −, n = 3) respond to RANKL signaling with release of IL-10. *p < 0.05 (Mann–Whitney U test).
RANK directly impairs NK reactivity upon interaction with RANKL-expressing target cells
To determine whether and how forward signaling via RANK directly influenced NK cell reactivity, we used murine tumor cells transfected with human RANKL (L-RANKL) or RANKL-negative control cells (L cells) as targets for NK cells within PBMCs that were previously cultured for 72 h to induce RANK expression. Cytotoxicity assays revealed that lysis was profoundly reduced by target cell–expressed RANKL. Similarly, release of IFN-γ, the second major mechanism by which NK cells contribute to anti-tumor immunity, was significantly reduced in the presence of target-expressed RANKL (Fig. 6A). Next, we used primary AML cells and determined whether Denosumab was able to improve the reactivity of RANK-expressing NK cells within allogeneic PBMCs. This was facilitated by the fact that Denosumab does not affect NK cell reactivity via its Fc part allowing for attribution of its effects in our functional analyses specifically to blocking RANK–RANKL interaction (Supplemental Fig. 1). In cytotoxicity assays, we observed increased NK lysis of primary RANKL-expressing AML cells upon blocking RANK–RANKL interaction with Denosumab (Fig. 6B). In addition, IFN-γ production was substantially increased upon blocking RANK–RANKL interaction (Fig. 6C). Statistical analysis of the effects on cytotoxicity and cytokine production in 36 and 13 independent experiments, respectively, revealed that both NK effector functions were significantly (both p < 0.05, Mann–Whitney U test) enhanced by Denosumab (Fig. 6B, 6C). Notably, the stimulatory effect of RANKL blocking (defined as >20% increase) on cytokine production occurred in a statistically significantly (p < 0.05, Mann–Whitney U test) higher number of experiments compared with cytotoxicity [10 of 13 (77%) versus 12 of 36 (33%) independent experiments, respectively] (Fig. 6D). When RANKL-negative AML cells were used as targets, NK cytotoxicity and cytokine production were not affected by the presence of Denosumab, thereby excluding that the above-described effects were due to modulation of effector cells by the Ab (Supplemental Fig. 1). Thus, RANK inhibits both cytotoxicity and cytokine production of NK cells upon interaction with its leukemia-expressed ligand with the effect of RANKL on the two major NK effector functions being more pronounced with regard to cytokine production.
RANK mediates inhibition of NK cells, and disruption of RANK–RANKL interaction by Denosumab reinforces NK cell anti-leukemia reactivity. PBMCs were cultured for 72 h to induce RANK expression on NK cells. Then (A) cocultures with RANKL-transfectants (L-RANKL) or RANKL-negative control cells (L cells) were performed. Cytotoxicity was determined by 2 h BATDA europium release assays (left); IFN-γ levels in supernatants were analyzed by ELISA after 24 h (right). (B–D) Cocultures with RANKL-positive primary AML cells (>80% blast count) in the presence or absence of Denosumab or isotype control (10 μg/ml each): (B) Cytotoxicity was determined by 2 h europium release assays. Results of one exemplary experiment with pronounced effects of RANKL blocking (left) and combined results of 36 independent experiments (right) are shown (E:T ratio 80:1). (C) IFN-γ levels in supernatants as determined by ELISA after 24 h. Left, Results of one representative experiment. Right, Combined results of 13 independent experiments (E:T ratio 6:1). (D) Comparative analysis of the effects of Denosumab on cytotoxicity and cytokine production. Lysis rates and cytokine levels obtained with untreated PBMCs in each individual data set were set to 1 to account for effects due to the allogeneic setting in the different experiments. More than 20% increase by blockade of RANK–RANKL interaction in each individual assay (dotted line) was defined as positive response. *p < 0.05 (Mann–Whitney U test).
Discussion
Activity of both tumor cells and immune effector cells including NK cells is substantially influenced by various members of the TNF/TNFR family (12). RANKL and its receptors RANK and osteoprotegerin play a central role in regulating bone turnover (16), but available data indicate that RANKL also contributes to the pathophysiology of malignant diseases. It has been shown that the RANK–RANKL system substantially influences regulation of cancer cell migration and metastasis of solid tumors (21–25, 32). With regard to hematopoietic malignancies, RANKL was reported to be expressed in membrane-bound and soluble form in multiple myeloma and may contribute to disease pathology, for example, by induction of osteolysis (20, 42, 43). In CLL, RANKL expression has been reported to influence the release of IL-8, which acts as autocrine and paracrine growth and survival factor for the malignant cells (19).
In this study, we report for the first time to our knowledge that RANKL is expressed on primary leukemia cells in a high proportion of AML cases. No significant association with specific FAB types or parameters associated with disease severity was observed. The latter is in line with results regarding the expression of other TNF family members in AML (e.g., Refs. 13, 14). RANKL can also be released as soluble form due to alternative splicing or by shedding from the cell surface due to the activity of metalloproteinases. We detected expression of the mRNA for the alternatively spliced form of RANKL in 17% of the investigated cases. However, neither release of sRANKL protein by AML cells in vitro nor elevated RANKL levels in sera of AML patients were observed in our study. The mRNA for mRANKL was found in all investigated AML samples including cases where surface expression was not detectable. Although the latter may be due to contamination with RANKL-expressing healthy cells, the lack of correlation between mRNA expression and prevalence of the respective soluble and membrane-bound protein points to a potential regulatory or mutational blockade of RANKL expression by posttranscriptional and/or posttranslational mechanisms in the individual patients. This again is in line with findings on the expression and release of other TNF family members in leukemic cells (13, 14, 40, 41).
When we set out to determine whether the leukemia-expressed RANKL was functional, we found that RANKL signaling significantly induced the release of TNF, IL-6, IL-8, and IL-10 as well as metabolic activity of primary AML cells. This is in line with findings that many ligands of the TNF family mediate bidirectional signals (33) and with reports that RANKL alters cytokine production of T cells and CLL cells (19, 34). The cytokines that we found to be induced by RANKL signaling into AML cells are known to act as autocrine/paracrine growth and survival factors in AML and are, at least in part, associated with development and progression of the disease (35, 36). Notably, RANKL signaling did not always induce release of the same cytokines with all RANKL-positive patient samples. Rather, we found distinct patterns of cytokine release upon RANKL signaling, and whereas TNF, IL-6, and IL-8 were released in most of the investigated cases, IL-10 was only released by 59% of the patient samples. Only 47% of the patient AML cells released all four cytokines investigated in our study, but all investigated RANKL-positive AML patient samples responded to RANKL signaling by release of at least one of the cytokines. Together with our finding that RANKL stimulates leukemia cell metabolism, these data strongly point to an involvement of RANKL in AML pathophysiology. The differing cytokine response to RANKL signaling could be due to regulatory or mutational blockades of RANKL or the associated signaling pathways or cytokine production in the leukemia cells that may be associated with development and progression of disease. Thus, RANKL expression and/or signaling may play a particular role in individual patients.
The factors released by AML cells upon RANKL signaling were further found to suppress NK cell reactivity, which points to a functional relevance of RANKL for NK cell immunity in AML. This is of importance as NK cells play an important role in immunosurveillance of AML, both in an autologous setting and after therapeutic intervention, for example, with allogeneic SCT (2–4). Notably, certain TNF family ligands may transduce signals even in the absence of their cognate counterpart, as exemplified by the role of CD137 ligand in promoting cytokine production by macrophages (44). In addition, RANK has been shown to be expressed by various cell types in peripheral blood and bone marrow and may thus readily be available to interact with AML-expressed RANKL (17, 26). By inducing the release of immunomodulatory factors, RANKL may thus substantially affect the reciprocal interaction of leukemia cells with the immune system, and preventing RANKL signaling might serve to influence the clinical course of AML. Our data indicate that this can be achieved with the mAb Denosumab capable of blocking RANKL, which was recently approved for treatment of osteolysis (37, 38). Denosumab reduced the release of immunomodulatory factors by AML cells and this reinforced NK cell anti-leukemia reactivity in our analyses.
Further rationale for neutralization of RANKL as an approach to enhance NK cell reactivity against AML cells is provided by our analyses on the expression and function of RANK on NK cells. RANK was previously shown to be expressed on NK cells, but nothing was yet known regarding its functional role (15, 17, 26). We report in this study that RANK is upregulated on NK cells of AML patients compared with healthy controls. This may, at least in part, be due to factors released from AML cells, as RANK was found to be induced on healthy NK cells by factors released from AML cells upon RANKL signaling. In line, IL-10, a cytokine known for its distinct immunomodulatory/immunosuppressive effects that we found to be released by AML cells upon RANKL signaling, caused upregulation of RANK on NK cells. However, supernatants of AML cells that did not release IL-10 in response to RANKL signaling also caused upregulation, indicating that other factors produced by AML cells upon RANK–RANKL interaction and that yet remain to be elucidated also modulate RANK expression in NK cells. It is tempting to speculate that interaction of NK cell–expressed RANK with its AML-expressed ligand may cause a “vicious RANK–RANKL cycle” of NK immunosubversion, disruption of which by blocking RANKL may restore NK cell reactivity. This is even more as we found that RANKL also directly diminishes the reactivity of NK cells by mediating inhibitory forward signals via RANK. The inhibitory effects of RANK in NK cells were revealed by analysis of cytotoxicity and IFN-γ production in experiments with RANKL-transfectants as targets. Moreover, blocking RANK–RANKL interaction in cultures of allogeneic NK cells and primary RANKL-positive AML cells by Denosumab increased target cell lysis and IFN-γ production. Notably, we confirmed that Denosumab did not induce Fc-mediated effects in NK cells (45), and no modulation of NK reactivity by Denosumab was observed when RANKL-negative cells were used as targets. This clearly attributes the effects of Denosumab in our functional analyses to disruption of RANK–RANKL interaction. Notably, the effects of blocking AML-expressed RANKL by Denosumab on NK function were significantly less pronounced with regard to cytotoxicity compared with cytokine production. NK cell reactivity is governed by a balance of multiple inhibitory and activating receptors that mediate their effects by at least partially different molecular mechanisms (10, 46). Induction of NK cell cytotoxicity and cytokine release are distinct events and selectively controlled (47). In addition, the assays used for analysis of cytotoxicity and cytokine production differed with regard to the time NK cells were cultured with their leukemia targets. The molecular mechanisms by which TNFR family members in general and RANK in particular influence NK reactivity are yet largely undefined, and it is thus possible that RANK activation may in fact mediate more pronounced effects on NK cell cytokine production compared with cytotoxicity, which requires further elucidation. Notably, beyond affecting NK cells, AML-expressed RANKL could also influence other components of immunity, as RANK protein was detected, among others, on the surface of DCs and T cells. With regard to its immune-modifying properties in the interaction of the latter cell types, RANK–RANKL interaction has been shown to enhance survival and the immunostimulatory capacity of DCs and to modulate effector T cell activation. In contrast, the RANK–RANKL system has been shown to control numbers and function of regulatory T cells (reviewed in Ref. 48). Further work is required to elucidate whether and how RANKL expression, besides modulating NK cell anti-leukemia reactivity, directly or via altered DC function, also affects T cell immunity in AML.
Overall, we provide the first evidence, to our knowledge, for the nvolvement of RANKL in AML pathophysiology. Our data indicate that RANKL also subverts NK cell immunosurveillance, which can be prevented by the clinically available RANKL Ab Denosumab, at least in RANKL-positive AML cases, which accounted for ∼two-thirds of the patients in our analyses. Notably, species-specific functional differences, which were observed by us and others with regard to the function of other TNFR family members (13, 14, 27, 28, 49), were excluded in our study by using human NK cells and primary leukemia cells of AML patients. Our data suggest that modulation of RANK–RANKL interaction may serve to prevent NK cell suppression thereby holding promise to optimize the reactivity of autologous or allogeneic NK cells in AML patients.
Disclosures
The authors have no financial conflicts of interest.
Footnotes
This work was supported by grants from Deutsche Forschungsgemeinschaft (SA1360/7-1, SFB685 TP A7), Wilhelm Sander-Stiftung (2007.115.3), and Deutsche Krebshilfe (109620). P.S. is supported by grants from the Swiss National Science Foundation.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- AML
- acute myeloid leukemia
- CLL
- chronic lymphoid leukemia
- DC
- dendritic cell
- FAB
- French–American–British classification
- KIR
- killer Ig-like receptor
- mRANKL
- membrane-bound RANKL
- RANK
- receptor activator for NF-κB
- RANKL
- receptor activator for NF-κB ligand
- SCT
- stem cell transplantation
- SFI
- specific fluorescence index
- sRANKL
- soluble RANKL.
- Received June 28, 2012.
- Accepted November 9, 2012.
- Copyright © 2013 by The American Association of Immunologists, Inc.
References
In this issue
