Identification of interleukin-1 by functional screening as a key mediator of cellular expansion and disease progression in acute myeloid leukemia

Summary

Secreted proteins in the bone marrow microenvironment play critical roles in acute myeloid leukemia (AML). Through an ex vivo functional screen of 94 cytokines, we identified that the pro-inflammatory cytokine interleukin-1 (IL-1) elicited profound expansion of myeloid progenitors in ~67% of AML patients while suppressing the growth of normal progenitors. Levels of IL-1β and IL-1 receptors were increased in AML patients, and silencing of the IL-1 receptor led to significant suppression of clonogenicity and in vivo disease progression. IL-1 promoted AML cell growth by enhancing p38MAPK phosphorylation and promoting secretion of various other growth factors and inflammatory cytokines. Treatment with p38MAPK inhibitors reversed these effects and recovered normal CD34⁺ cells from IL-1-mediated growth suppression. These results highlight the importance of ex vivo functional screening to identify common and actionable extrinsic pathways in genetically heterogeneous malignancies and provide impetus for clinical development of IL-1/IL1R1/p38MAPK pathway-targeted therapies in AML.

Introduction

Acute myeloid leukemia (AML) is a genetically heterogeneous disease characterized by clonal expansion of myeloid progenitors that accumulate in the bone marrow and peripheral blood. Standard-of-care cytotoxic chemotherapy has remained the mainstay of AML treatment for decades, with clinical response to therapy influenced by the cytogenetic and molecular phenotype of the disease (Estey and Dohner, 2006). While ‘favorable risk’ patients exhibit high response rates to chemotherapy, more than one-third of diagnosed patients ultimately succumb to treatment-resistant disease (Ishikawa et al., 2007).

Recent whole-genome sequencing have revealed significant heterogeneity in the molecular abnormalities driving AML (Cancer Genome Atlas Research, 2013). Selective targeted inhibitors have been developed for many of the pathways influenced by these genetic alterations, but successful translation of these agents into clinical practice has been impeded due to disease heterogeneity. Therefore, in order to develop targeted therapies with clinical efficacy, it will be critical to identify unifying mechanisms required for AML progression, irrespective of the diverse genetic and molecular abnormalities. In vitro functional analyses have aided these drug discovery efforts (Tibes et al., 2012), but are commonly conducted in the absence of microenvironmental factors such as secreted cytokines and growth factors known to influence cell survival and response to therapy (Bernasconi et al., 2016).

Cytokines and growth factors secreted in the bone marrow microenvironment play important roles in modulating cell survival, proliferation, differentiation, and the immune response (Welner et al., 2015; Zhang et al., 2012). For example, several pro-inflammatory cytokines such as tumor necrosis factor-alpha (TNF-α) inhibit the growth of normal hematopoietic progenitor cells, yet paradoxically enhance the proliferation of neoplastic cells in patients with myeloproliferative neoplasms (Fleischman et al., 2011), myelodysplastic syndrome (Verma et al., 2002), and Fanconi anemia (Anur et al., 2012; Garbati et al., 2016; Li et al., 2007). Perhaps not surprisingly, there is also an association between chronic inflammation and tumor development or progression (Coussens and Werb, 2002), in which prolonged inflammatory cytokine exposure has the potential to promote tumor cell survival and growth. We therefore hypothesized that inflammatory cytokines play critical but not yet fully defined roles in the leukemic cell expansion and progression of AML. The goal of this study was to identify secreted cytokines and growth factors, and mechanisms by which such factor(s) facilitate the survival of AML cells through a common signaling pathway independent of mutation status, and which could be targeted in order to develop a therapeutic strategy against this disease. In our study, we demonstrate that the pro-inflammatory cytokine IL-1 has a profound autocrine and paracrine growth-promoting effect on progenitor cells from AML patients, contributes to an inflammatory microenvironment, promotes AML disease progression in a mouse model, and can be targeted using small-molecule inhibitors of the p38 mitogen-activated protein kinase (MAPK) pathway.

Results

Ex vivo cell viability screen identifies functionally significant cytokine- and growth factor-driven signaling pathways in AML

To identify cytokine and growth factor signaling pathways critical for AML cell survival, we systematically quantified the growth of 60 primary AML patient samples in the presence of graded concentrations of 94 cytokines using an ex vivo cell viability screen (Figures 1A, S1A-B, and Table S1). Wells with HS-5 conditioned media and high and low concentrations of cytokine cocktail were used as positive controls; no-cytokine wells were used as negative controls. While a number of cytokines, including granulocyte macrophage colony-stimulating factor (GM-CSF), IL-3, M-CSF (macrophage colony-stimulating factor), granulocyte colony-stimulating factor (G-CSF), and TNF-α, stimulated AML cell growth (Figure S2A), IL-1α and IL-1β had the most profound effect, resulting in dose-dependent increase in cell growth up to 15-fold in ~67% (40/60) of the primary AML samples (designated as “IL-1-sensitive patients”) (Figure 1A–C). Notably, IL-1α and IL-1β responses clustered with those of GM-CSF and IL-3, two other growth-associated cytokines, suggesting a role for IL-1 as a key mediator of growth in AML (Figure 1B). In contrast, several cytokines, including transforming growth factor beta (TGF-β)-1/2/3, chemokine (C-C motif) ligand-26 (CCL-26), sCD27L, CCL-24, CCL-11 and Activin-A, suppressed AML cell growth in 20–40% patient samples (Figure S2B).

Figure 1. IL-1 promotes the growth of a subset of primary AML patient samples.

(A) Primary mononuclear cells derived from bone marrow and peripheral blood of 60 AML patient samples were cultured for 3 days in RPMI-1640 supplemented with 5% FBS. The effect of individual cytokines on cell growth was measured by MTS assay (screen layout in Figure S1A). Absorbance values were normalized against the growth of the cells without cytokines. Cytokines causing increased growth equal to or more than the growth response to HS-5 conditioned media were designated as growth-promoting cytokines; those causing decreased growth compared to the no-cytokine control were designated as growth-suppressive cytokines. The percentage of primary samples responding to individual cytokines is shown.

(B) Heatmap representation of cytokine response clustering with IL-1 (full heatmap in Figure S1B). Red indicates increased growth; blue indicates decreased growth. The growth response for selective cytokines is show in Figure S2 and correlation of IL-1-mediated growth response with clinical, demographic, and genetic features of AML patients is shown in Figure S3 and Table S2.

(C) Impact of graded concentrations of IL-1α and IL-1β on the growth of primary AML mononuclear cells (MNCs) from each patient sample screened in Figure 1A by colorimetric MTS assay. In this assay, the absorbance value for untreated cells were considered as maximum viability and fold change over the untreated controls were calculated.

(D) Purified CD34⁺ cells derived from AML patient samples bone marrow were cultured in RPMI-1640 supplemented with 5% FBS with graded concentrations of IL-1α and IL-1β and the effect on growth was measured by MTS and colony formation assays in cytokine free MethoCult supplemented with 10 ng/ml IL-3 and 50 ng/ml SCF.

(E) Purified normal CD34⁺ cells derived from healthy donors (NL) bone marrow were cultured with graded concentrations of IL-1β and effect on growth was measured by colony formation assay in cytokine-free MethoCult supplemented with 10 ng/ml IL-3 and 50 ng/ml SCF. Levels of statistical significance: * p≤0.05, ** p≤0.01, *** p≤0.001.

IL-1 is a master regulator of inflammation, innate immunity, and hematopoiesis (O’Neill, 1995). The IL-1α and IL-1β exhibit similar biological activities, although they differ in the manner in which they are processed and secreted. IL-1α is localized in the cytosol or cell membrane. In contrast, IL-1β is secreted extracellularly. Given the patients with high IL-1β expression have increased risk of metastasis and poor prognosis in solid tumor diseases (Voronov et al., 2003), we focused on understanding the mechanism, functional significance, and therapeutic potential underlying the high frequency of IL-1β sensitivity that we observed in our screening assay. A comparison of clinical, demographic, and prognostic characteristics between IL-1β-sensitive and -nonsensitive AML patients included in the screening assay revealed no significant association between fold-change in IL-1β-mediated cell growth in the viability screen and patient age, sex, ethnicity, white blood cell count, blast percentage, karyotype, or frequently occurring gene mutations (Figures S3 and Tables S2–3). Similarly, none of the canonical IL-1 signaling mediator genes have been reported to be significantly mutated in AML (TCGA analysis) (Cancer Genome Atlas Research, 2013). These results indicate that a large percentage of AML patient samples show cellular expansion in the presence of IL-1 irrespective of mutation status and distinct clinical features, suggesting that targeting this unifying mechanism of IL-1-mediated leukemic cell expansion in AML patients may be applicable across heterogeneous AML subtypes. Our results also highlight the relevance of functional screening to identify extrinsic factors-driven pathways that support the expansion and survival of AML cells.

IL-1 promotes the expansion of AML progenitor cells while suppressing the growth of healthy normal progenitors

Previous studies have indicated a role for IL-1 in the expansion of AML (Bruserud, 1996; Turzanski et al., 2004) and normal progenitors (Fibbe and Willemze, 1991; Haylock et al., 1992; Kovacs et al., 1995; Pietras et al., 2016; Yonemura et al., 1996). However, the functional significance and therapeutic implications of this pathway in AML and normal progenitors has not been fully explored. To further define the role of IL-1 signaling in AML, we first validated the effects of IL-1α and IL-1β on an additional independent set of 26 primary AML samples. Similar to the results of our ex vivo functional screen on 60 primary samples, 61% (16/26) samples showed increased growth on exposure to IL-1α and IL-1β in a concentration-dependent manner (Figure S4A). Similarly, treating CD34⁺ AML cells with IL-1 led to ~6-fold increase in cell growth and a ~2-fold increase in clonogenic potential (Figure 1D). In contrast, IL-1 treatment led to ~5-fold suppression of colony formation of CD34⁺ cells from normal bone marrow donors (Figure 1E). The contrasting growth effect of IL-1 on healthy and AML progenitors might be due to the modulation of cell proliferation, survival, and differentiation (evaluated in Figure 5).

Figure 5. IL-1-promotes the proliferation, survival, and colony formation ability of AML cells which are inhibited by a p38MAPK inhibitor.

(A) Purified CD34⁺ progenitors from the bone marrow of healthy donors and AML patients were cultured in 10 ng/mL IL-1β with and without 500 nM doramapimod, for 6 days in a culture media containing IMDM+10%FBS with 20 ng/ml IL-6, 10 ng/ml IL-11, 50 ng/ml FLT-3L, and 100 ng/ml SCF. Fresh doses of cytokine and/or inhibitor were added to the culture every 3 days. Cells for counted alternate day and total cell number is shown over time.

(B) CD34⁺ cells were labeled with a Violet cell tracer dye and dilution of dye was measured by FACS after 3 days of culturing in above indicated media. % proliferation is shown from a representative sample from three independent experiments from the bone marrow of healthy donors and IL-1 sensitive AML patients. (See also Figure S7H for IL-1 nonsensitive AML sample)

(C) CD34⁺ AML and healthy bone marrow cells were cultured in 10 ng/mL IL-1β with and without 500 nM doramapimod and the effect on apoptosis was determined by measuring annexin V positivity after 2 days of treatment.

(D) The effect of IL-1 and doramapimod on the clonogenic growth of CD34⁺ cells from primary AML patients and healthy bone marrow was measured by colony formation assay. Cytokine-free MethoCult was supplemented with 10 ng/ml IL-3 and 50 ng/ml SCF, and colonies were scored on day 14.

(E) AML and healthy cells were cultured as described in Figure 5A and the effect of treatments were measured on differentiation using indicated surface markers. Histograms are showing mean flouroscence intensity (MFI) for a representative sample and bar graphs are showing mean+SEM for MFI and % of total cells from three samples. * p≤0.05, ** p≤0.01, *** p≤0.001.

AML patients have significantly increased blood and bone marrow IL-1β levels and IL-1β-secreting monocytes promote the survival of AML cells

Consistent with the dichotomous role of IL-1 on the growth of AML and normal progenitors, IL-1β plasma levels were significantly increased in AML peripheral blood and bone marrow samples compared to healthy controls as measured by ELISA (Figure 2A). Further, the secretion and expression of IL-1β was significantly increased in the bone marrow of IL-1β-sensitive AML samples compared to IL-1β-nonsensitive samples (Figure 2B), and there was a trend toward increased expression in the peripheral blood of IL-1β-sensitive AML samples.

Figure 2. IL-1-sensitive AML patient samples have increased IL-1β and IL-1 receptor expression in the bone marrow, and IL-1-secreting monocytes regulate the growth of AML cells in paracrine manner.

(A) IL-1β levels were measured by ELISA in plasma from peripheral blood (PB) from healthy individuals, and from AML peripheral blood and bone marrow samples (BM).

(B) IL-1β level by ELISA and expression by quantitative PCR were measured from IL-1β-sensitive (S) and nonsensitive (NS) AML peripheral blood and bone marrow samples.

(C) IL1R1 expression was measured by quantitative PCR from IL-1β-sensitive (S) and nonsensitive (NS) AML peripheral blood and bone marrow samples.

(D,E) IL-1 receptor (IL1R1, IL1RAcP) expression and intracellular IL-1β levels were measured by flow cytometry in various subpopulations of bone marrow cells from healthy individuals and IL-1β-sensitive and nonsensitive AML patients. Percentage of positive cells for each subpopulation is shown as mean+SEM from three independent bone marrow samples. The representative gating strategy and FMO controls are shown in Figure S4D.

(F) Representative plots showing percentage of IL-1β-positive cells in various subpopulations in bone marrow of healthy individuals and IL-1β-sensitive and nonsensitive AML patients. See also Figure S4 for the percentage of CD14+, CD34+ and CD33+ cells.

(G) CD14⁺ cells were depleted from IL-1β-sensitive primary AML MNCs bone marrow, and the effect of depleting CD14⁺ cells and adding them back to MNCs was measured by MTS assay. * p≤0.05, ** p≤0.01, *** p≤0.001.

Both IL-1α and IL-1β agonists bind to the IL-1 receptor type I (IL1R1), leading to heterodimerization with IL-1R accessory protein (IL1RAcP) (O’Neill, 1995). We found that the expression of IL1R1 and IL1RAcP was significantly increased in IL-1β-sensitive AML bone marrow samples compared to IL-1β-nonsensitive samples (Figures 2C–D, S4D, and S5). These results suggest that IL-1 signaling is aberrantly activated in IL-1 sensitive AML samples due to increased expression of IL-1 and its receptors.

To identify the source of IL-1 secretion, we performed immunophenotyping of primary AML samples using predefined cell surface markers and revealed that IL-1β is primarily secreted by monocytes (CD14⁺), myeloid (CD33⁺), and stem (CD34⁺) cells, but not by T (CD3⁺), B (CD19⁺), or to a lesser extent by stromal (CD45⁻CD34⁻CD90⁺) cells (Figures 2E–F and S5). Furthermore, IL-1 sensitive AML patients had significantly increased numbers of IL-1-secreting monocytes, as well as a trend towards increasing likelihood of FAB M4/M5 myelomonocytic/monocytic disease classification (Figures S4B–C). In contrast, most of the IL-1-nonsensitive AML samples were of M1 disease classification (Table S2).

Given that IL-1 is secreted by monocytes, we next investigated the possibility of a paracrine role for IL-1 in the expansion of AML progenitor cells. We observed that depletion of CD14⁺ monocytes from AML mononuclear cells (MNCs) resulted in 5-fold reduced growth of AML progenitor cells, inhibition which was reversed by adding exogenous IL-1β or by adding back CD14⁺ cells (Figure 2G). Our results suggest that although monocytes represented ~20% of the AML subpopulation, and that upon the depletion of monocytes, AML progenitors should remain IL-1 dependent in a cell-autonomous manner, the depletion of monocytes does have a significant impact on AML growth. This might occur either because exogenous IL-1 secreted by monocytes may promote the secretion of additional growth-stimulating cytokines from progenitors (evaluated in Figure 6), or because monocytes serve as a reservoir for multiple other cytokines (Panoskaltsis et al., 2003). Together, these results support a paracrine role of IL-1 in AML, in which increased production of IL-1 by monocytes promotes the survival of leukemia cells through paracrine signaling, underlying the basis of IL-1 hypersensitivity in AML.

Figure 6. IL-1β promotes the secretion of multiple growth-promoting inflammatory mediators from AML progenitors and their secretion is blocked by p38 kinase inhibition.

Comprehensive evaluation of cytokines produced by IL-1-sensitive and non-sensitive AML CD34⁺ bone marrow progenitors using the 30-plex human cytokines Luminex platform. Purified CD34⁺ progenitors from IL-1-sensitive and non-sensitive AML samples (N=3) were treated with 10ng/ml IL-1β and 500nM doramapimod and cultured in 10% FBS+ RPMI media for 48 hrs and conditioned media were evaluated for cytokine production. Standard curves for each cytokine evaluated met quality control assessment.

(A) Heat map showing the relative cytokine production for the 30 cytokines compared to untreated controls.

(B) The mean absolute values of cytokines showing the profound response to IL-1 stimulation in AML progenitors are shown on a log10 scale as the mean of three independent samples. (See also figure S7I showing the effect of blocking individual cytokine on cell viability)

Absence of IL-1 receptor attenuates the expansion of AML cells in vitro and disease progression in vivo

To further evaluate the functional relevance of IL-1 signaling in AML, IL-1-sensitive AML patient samples were subjected to individual siRNA knockdown by a panel targeting 188 cytokine and growth-factor receptors (Agarwal et al., 2014b), including the IL-1 receptors IL1R1 and IL1RAcP. We found that IL-1-sensitive AML samples were dependent on both IL-1 receptors for survival (Figure S6A). Notably, while the importance of IL1RAcP in AML has been previously reported (Askmyr et al., 2013; Barreyro et al., 2012), its role as a co-receptor is not confined to IL-1 signaling, as it serves a similar role for IL-33 and IL-36. Therefore, to address the specific role of IL-1 signaling in AML, we focused our efforts on establishing the functional significance of the IL1R1. We identified that lentivirus-mediated stable knockdown of IL1R1 in primary AML cells and cell lines reduced cell growth ~2-fold and clonogenicity 3.5-fold (Figures 3A and S6B–C), but not in IL-1-nonsensitive AML samples (Figure S6D). The importance of IL1R1 in AML cell clonogenicity was further demonstrated using murine IL1R1-null and wild-type bone marrow transduced with oncogenes involved in AML pathogenesis (AML1-ETO9a coupled with NRAS^G12D or MLL-ENL) (Zuber et al., 2009). IL1R1-null oncogene-transduced cells demonstrated significantly reduced myeloid colony formation compared to wild-type cells, using either cytokine-supplemented media containing IL-1β (2.5-fold reduction) or IL-1β alone (4.0-fold reduction) (Figures 3B and S6E). By contrast, no significant effect on colony growth was observed in empty vector controls (Figure 3B). Additionally, immunophenotyping of bone marrow from wild-type and IL1R1-null mice revealed no significant differences in stem and progenitor cell populations (Figure S6F). Taken together, these results suggest that IL1R1 is dispensable for normal hematopoiesis, but can modulate AML cell growth.

Figure 3. The absence of IL1R1 attenuates the growth of primary AML cells as well as colony-forming ability and disease progression in a murine AML bone marrow transduction/transplantation model.

(A) Primary AML CD34⁺ cells purified from bone marrow were infected with a combination of two inducible IL1R1 shRNA hairpins (see also Figure S6). After 48 h of infection, cells were sorted for GFP positivity, and those cells stably expressing IL1R1 shRNA were treated with doxycycline. The effect of knockdown was determined on IL1R1 expression by qPCR after 48 h, cell growth by MTS assay after 4 days after culturing cells in IMDM supplemented with10%FBS and 10⁻⁴ M β-ME, and colony formation assay in cytokine-free MethoCult supplemented with 10 ng/ml IL-3 and 50 ng/ml SCF. The experiment is representative of 3 independent experiments.

(B) Myeloid colony formation assay using mouse bone marrow cells from wild-type and IL1R1-null mice transduced with AML1-ETO9a/NRAS^G12D or empty vector and plated with IL-1α and IL-1β together (2 ng/ml, each) with either cytokine-free or cytokine-supplemented MethoCult. Colonies were scored on day 8 (see also Figure S7).

(C) Lethally irradiated wild-type mice were injected retro-orbitally with equal numbers of AML1-ETO9a/NRAS^G12D-transduced GFP+ bone marrow from wild-type and IL1R1-null mice and evaluated for survival using Kaplan-Meier statistics.

(D) Hematoxylin and eosin (H&E) histopathological analysis of mouse tissues from IL1R1-null and wild-type leukemic mice. Insets represent megakaryocytic and myeloid infiltrates. * p≤0.05, ** p≤0.01, *** p≤0.001.

To evaluate the contribution of IL1R1 to disease progression in vivo, we employed a murine AML bone marrow transplantation model in which we transplanted AML1-ETO9a/NRAS^G12D-transduced bone marrow cells from wild-type and IL1R1-null mice into wild-type recipients. Both wild-type and IL1R1-null-transduced marrow recipients developed AML-like disease with leukocytosis and hepatosplenomegaly, but recipients of IL1R1-null marrow survived longer (median: 36.5 days; range: 30–120) than recipients of wild-type marrow (median: 31.5 days; range: 27–61; p=0.030; Figure 3C). Although WBC count, platelet count, and spleen size were comparable between mice transplanted with IL1R1-null and wild-type marrow (data not shown), histopathological analysis revealed a significant reduction of myeloid infiltrates in the liver and lungs in IL1R1-null leukemic mice (Figure 3D). Overall, these results demonstrate a critical role for IL-1 signaling in the growth of AML cells and clearly define the in vivo role for IL1R1 in disease progression.

IL-1β promotes the growth and survival of AML cells by increasing p38MAPK phosphorylation, and p38MAPK inhibitors reverse this effect

We evaluated the influence of IL-1β on downstream signaling in IL-1-sensitive and -nonsensitive AML patients compared to normal CD34⁺ cells. We observed that IL-1-sensitive primary AML MNCs and CD34⁺ cells, along with AML cell lines, showed increased phosphorylation of p38MAPK at the basal level and in the presence of exogenous IL-1β. However, IL-1β had slight or no consistent effects on the expression of other IL-1 signaling mediators such as IRAK1, MYD88, IKKβ, and NF-κB, which were expressed at high basal level, suggesting IL-1 signaling remains intact in AML (Beverly and Starczynowski, 2014) (Figures 4A–F, S7A and S7E). IL-1β stimulation had no significant effect on p38MAPK phosphorylation in normal CD34⁺ cells and IL-1-nonsensitive AML cells (Figures 4C, 4E, and S7B). Interestingly, IL-1-sensitive AML cells treated with a non-IL-1-inclusive cocktail of cytokines (IL-6, IL-11, FLT3L, and SCF) did not show increased phosphorylation of p38MAPK (Figure 4B), suggesting IL-1β is uniquely responsible for increased p38MAPK activation in these cells.

Figure 4. IL-1β promotes the growth of AML cells by increasing p38MAPK phosphorylation and IL-1-dependent growth of AML cells is inhibited by a p38MAPK inhibitor.

(A) Basal p38MAPK phosphorylation of CD34⁺ cells purified from the bone marrow of IL-1-sensitive and -nonsensitive primary AML samples, as well as healthy controls, as measured by phosphoflow analysis. The bar graph shows relative mean fluorescent intensity (MFI) of p38MAPK phosphorylation for 3 IL-1-sensitive and 3 IL-1-nonsensitive AML samples and 2 healthy controls.

(B,C) CD34⁺ cells purified from primary AML samples and normal CD34⁺ cells from a healthy individual were stimulated with 10 ng/ml IL-1β or a non-IL-1β inclusive cocktail of various cytokines (20 ng/ml IL-6, 10 ng/ml IL-11, 50 ng/ml FLT-3L, and 100 ng/ml SCF). The effect of 500 nM of the p38MAPK inhibitor doramapimod (Dora) was determined by measuring p38MAPK phosphorylation by phosphoflow analysis. The results from IL-1-sensitive and -nonsensitive primary mononuclear cells (MNCs) and AML cell lines are shown in Figure S7A,B and S7E. (D) CD34⁺ cells purified from primary AML samples and normal CD34⁺ cells from a healthy individual were stimulated with 10 ng/ml IL-1β. The effect of 500 nM of the p38MAPK inhibitor ralimetinib (Ralim) was determined by measuring p38MAPK phosphorylation by phosphoflow analysis.

(E) The average mean fluorescent intensity (MFI) of p38MAPK phosphorylation, measured by phosphoflow analysis, for 4 IL-1-sensitive AML samples and 2 healthy controls.

(F) Primary CD34+ cells derived from 3 IL-1-sensitive AML bone marrow samples were stimulated with 10 ng/ml IL-1β with and without 500 nM doramapimod for the indicated times. The effect of IL-1 stimulation on downstream signaling by measured by immunoblotting. Relative expression of phospho-p38MAPK and phospho-IRAK-1 normalized to GAPDH controls is shown as the mean+SEM from 3 samples.

(G) CD34⁺ AML cells were cultured in the indicated concentration of IL-1α or IL-1β (ng/mL) with and without 500 nM doramapimod or ralimetinib, and cell viability was measured by MTS assay. Results from IL-1-sensitive and -nonsensitive primary MNCs are shown in Figure S7C-D.

(H) Primary AML MNC, purified CD34⁺, and FACS sorted CD34⁺ IL1R1⁺ and CD34⁺IL1R1⁻ cells were cultured in 10 ng/mL IL-1β with and without 500 nM doramapimod, for 3 days and cell viability was measured by MTS assay (see also Figure S7G). * p≤0.05, ** p≤0.01, *** p≤0.001.

To understand the mechanistic relevance of p38MAPK activation in IL-1-sensitive AML, we treated primary AML cells and cell lines with the small-molecule p38MAPK inhibitor doramapimod (BIRB-796) (Anur et al., 2012) and observed inhibition of p38MAPK phosphorylation in IL-1-sensitive AML samples and cell lines (Figures 4B–F, S7A, and S7E). Similar results were obtained with the newly developed p38 kinase inhibitor ralimetinib, currently in Phase II clinical trials for ovarian cancer (NCT01663857) (Patnaik et al., 2016). Ralimetinib inhibited p38MAPK and the activity of its downstream substrate MK2 (Figure 4D). We observed only marginal inhibition of p38MAPK in normal cells with p38 inhibitor (Figure 4E). One explanation for this observation is that normal CD34⁺ cells show low phosphorylation of p38MAPK at the basal level (Figure 4A). Accordingly, IL-1-induced growth of primary AML MNCs and CD34⁺ cells was blocked upon treatment with doramapimod and ralimetinib (Figures 4G and S7C), but IL-1-nonsensitive samples were unresponsive to p38 inhibition (Figures S7B and S7D). In long-term culture assays, AML cells cultured over 12 days demonstrated sustained cell growth in the presence of IL-1β, an effect that was suppressed by p38MAPK inhibition (Figure S7F). Our data showed that IL-1 promoted the growth of purified CD34⁺IL1R1⁺ progenitors, but not CD34⁺IL1R1⁻ progenitors. Further, IL1R1⁺ progenitors were more sensitive to p38 inhibition compared to IL1R1⁻ progenitors (0.6 versus 0.3 fold inhibition, Figure 4H). However, in bulk MNC culture, p38 inhibition eliminated both IL1R1⁺ and IL1R1⁻ cells, and the effect of the p38 inhibitor was more pronounced in the presence of IL-1 (Figures 4H and S7G). These results indicate that in bulk culture, the engagement of the IL-1/p38AMPK pathway in IL1R1⁺ cells might activate p38MAPK in IL1R1⁻ cells possibly via secretion of additional cytokines from IL1R1⁺ cells in p38MAPK dependent manner (Zarubin and Han, 2005). Next, we evaluated the influence of the IL-1/p38MAPK pathway on cell proliferation, survival, clonogenic potential, and differentiation. IL-1 promoted sustained expansion of IL-1 sensitive AML progenitors over 6 days due to increased proliferation and reduced apoptosis in cultured AML cells, as well as increased their colony formation ability. These effects were reversed after p38MAPK inhibitor treatment (Figure 5A–D). IL-1 had no effect on the proliferation of IL-1 nonsensitive AML cells (Figure S7H). In contrast, while IL-1 promoted a slight short-term expansion of normal progenitors with increased proliferation within 2–4 days (Figure 5A–B), IL-1 also mediated increased apoptosis and suppression of clonogenic potential in normal CD34⁺ cells. Importantly IL-1-mediated suppression of normal CD34⁺ cells was reversed with doramapimod treatment (Figure 5C–D). Our data suggest that this paradoxical effect of IL-1 on normal and AML progenitors might be because IL-1 had differential effect on the maturation of these cells (Figure 5E). IL-1 promoted reduction in the expression of myeloid monocytic marker (CD14⁺) with increased number of CD34⁺ AML progenitors after 6 days. However, normal progenitors gained a significant increase in the expression of myeloid and monocytic maturation markers (CD13⁺, CD33⁺, CD11b⁺, CD14⁺) with a reduction in the number of healthy CD34⁺ progenitors. Treatment with p38MAPK inhibitor suppresses the expression of these myeloid markers in healthy cells. These results demonstrate that IL-1 promoted the growth of AML progenitors by increasing cell proliferation and survival with only slight effects on differentiation. However, IL-1 suppressed the clonogenic potential of normal progenitors possibly by increasing their maturation, leading to their reduced survival in a p38MAPK-dependent manner. Together, these results suggest that p38 kinase inhibitors both selectively suppress AML cell growth and support the expansion of normal hematopoiesis.

IL-1 promotes the secretion of multiple growth-promoting and inflammatory mediators from AML progenitors, and p38MAPK inhibition blocks this secretion

Both IL-1 and p38MAPK are known to influence the expression of inflammatory cytokines (Zarubin and Han, 2005). Therefore, to identify the influence of IL-1 and p38MAPK activity on downstream inflammatory milieu in AML, we screened levels of secreted cytokines in culture media from CD34⁺ progenitors from IL-1-sensitive and non-sensitive AML patients following treatment with IL-1 and the p38 kinase inhibitor doramapimod alone and in combination. We identified that IL-1 stimulation of IL-1-sensitive AML progenitors enhanced the secretion of various cytokines, growth factors and chemokines, but not in IL-1-non-sensitive AML progenitors (Figure 6A). The most pronounced effects observed were in increased levels of IL-6, IL-8, IL-12, GM-CSF, MCP1, MIP1α, MIP1β, and VEGF (Figure 6B). These effects were completely abrogated upon treatment with doramapimod, suggesting IL-1-mediated stimulation of inflammatory signals is p38MAPK dependent in IL-1-sensitive AML samples. Further, blocking individual cytokines, including IL-6, IL-8, GM-CSF, MCP1, with neutralization antibodies caused a slight reduction in IL-1β-mediated growth of AML cells. However, the blocking of IL-1 itself using anti-IL1R1 or anti-IL-1β antibodies reduced the growth of AML progenitors significantly. This is possibly because IL-1 promotes the expansion of AML cells by secreting multiple cytokines, and after blocking individual cytokines compensatory pathways involving other cytokines would be activated (Figure S7I). Overall, these results indicate that increased expression of IL-1 and its receptors in a subset of AML patients promotes p38MAPK phosphorylation, leading to an increase in the secretion of pro-inflammatory cytokines and growth factors that promote the survival of AML progenitors.

Based on our findings, we propose a model in which aberrant IL-1 signaling promote the expansion of leukemic stem and progenitor cells in a large subset of AML patients, while simultaneously inhibiting hematopoiesis of normal stem and progenitor cells. These differential effects of IL-1 on AML and normal cells can be selectively blocked by p38MAPK inhibition (Figure 7). Although, the differential effect of IL-1 on leukemia cells and healthy progenitors needs to be validated in competitive repopulation models, our results provide strong evidence supporting therapeutic targeting of the IL-1/IL1R1/p38MAPK pathway in diverse subtypes of AML.

Figure 7. Summary model of IL-1-mediated promotion of AML cell growth and suppression of normal hematopoiesis.

(A) In AML, increased IL-1β is secreted in the bone marrow by monocytes and leukemic cells. Increased IL-1β promotes the expansion of AML myeloid progenitors while suppressing the growth of normal myeloid progenitors due to the aberrant activation of the IL-1/p38MAPK pathway. This leads to disease progression in AML.

(B) Knocking down IL1R1 and blocking IL-1 signaling with a p38MAPK inhibitors suppresses the growth of AML cells and rescues normal hematopoiesis.

Discussion

In the treatment of AML, standard-of-care cytotoxic chemotherapy can have an initial effect on disease burden for many patients. However, surviving leukemic clones can repopulate the disease (Ishikawa et al., 2007). Part of this treatment challenge stems from the nature of AML as a heterogeneous disease with diverse cytogenetic and molecular abnormalities. Recent whole-genome sequencing studies have increasingly characterized these heterogeneities, which may be useful for designing personalized therapeutics. However, challenges remain for expanding these findings to develop treatment strategies that are broadly applicable to AML patients while still curbing some of the toxicities associated with standard chemotherapy. For example, even FLT3 and NPM1, two of the more frequently mutated genes in patients with AML, occur in only ~25% of patients (Cancer Genome Atlas Research, 2013), warranting critical efforts to identify more ubiquitous mechanisms that are responsible for AML progression.

We hypothesized that a systematic analysis of factors within the leukemia microenvironment might provide an opportunity to find more broadly effective therapeutic targets in AML. By using a unique ex vivo screen, we evaluated cytokine-driven pathways that are critical in promoting or suppressing AML cell growth. Interestingly, among 94 cytokines tested, only a few—including IL-1, GM-CSF, IL-3, M-CSF, G-CSF, and TNF-α—profoundly stimulated growth; members of the TGF-β family suppressed the growth of AML cells. These results suggest that select secreted proteins in the microenvironment influence the expansion of leukemic progenitors. Additionally, this study offers a platform to systematically evaluate a large number of extrinsic factors to identify pathways critical for disease survival using primary AML patient specimens.

Importantly, we identified a large percentage (~67%) of AML patients show expansion of leukemia progenitors in the presence of exogenous IL-1, independent of their mutation and cytogenetic status. Further, our results demonstrated that IL-1-sensitive AML patients clustered with other known myeloid growth-promoting cytokines such as GM-CSF and IL-3. While previous studies describe an effect of IL-1 predominantly on the production of GM-CSF and G-CSF by AML blasts (Bradbury et al., 1990; Cozzolino et al., 1990; Delwel et al., 1989), we show that IL-1 enhances the production of numerous other pro-inflammatory cytokines (e.g. IL-6, IL-8, MCP-1, MIP-1α, MIP-1β) from AML progenitors and blocking IL-1 signaling may mitigate these effects. These findings implicate a role for IL-1 as a key mediator of myeloid growth in AML.

IL-1 is a pluripotent cytokine responsible for various physiological roles including inflammation, immunity, and hematopoiesis, and plays a central role in various auto-inflammatory diseases (Voronov et al., 2003). Both isoforms of IL-1, IL-1α and IL-1β, bind the IL1R1 receptor, which heterodimerizes with IL1RAcP to initiate downstream signaling events (O’Neill, 1995). Previously, the expression of IL1RAcP, a receptor that binds IL-1 along with IL-33 and IL-36, was shown to be a functionally relevant and prognostic marker in CML and AML (Askmyr et al., 2013). However, these studies have not established which of the IL-1 family of cytokines is critical in AML specifically.

We identified a pivotal role of IL-1 for in vitro cell expansion in a large subset of AML patients with diverse genetic subtypes. AML cells exhibit increased secretion of IL-1 and expression of IL-1 receptors (IL1R1 and IL1RAcP) compared to normal marrow and IL-1-nonsensitive AML samples. We demonstrate that genetic knockdown of IL1R1, an IL-1-specific receptor, suppresses the growth of primary AML cells and reduces disease progression in a murine model of AML. A recent study has demonstrated that IL-1 represses the growth of leukemic cells by inducing differentiation in murine models driven by single oncogenes such as FLT3-ITD and AML1-ETO (Hockendorf et al., 2016). Although this study highlights an important context-dependent role of IL-1, possibly at a pre-leukemic state, AML is driven by complex combination of mutations. Therefore, in our study, we evaluated the effect of IL-1 signaling in primary AML samples harboring multiple mutations and in a murine AML model with dual oncogenes AML1-ETO/NRAS^G12D which provide strong evidence for the functional significance of IL-1 and its receptor IL1R1 in AML

IL-1 is known as an autocrine growth factor for AML cells (Cozzolino et al., 1989; Sakai et al., 1987). We identified a role for IL-1 as an AML cell survival factor mediated via paracrine signaling, in addition to its growth-stimulating autocrine effect in AML. We observed that IL-1 is primarily secreted by monocytes and myeloid blasts and showed a role for IL-1-producing monocytes in supporting the survival of leukemic cells. We believe these findings have important implications in treatment strategies, identifying a potential rationale for targeting not only leukemic progenitor cells, but also more differentiated accessory cells involved in disease pathogenesis and progression.

IL-1 driven inflammation is known to contribute to tumor development and invasiveness (Dinarello et al., 2012; Lee et al., 2015). We showed IL-1 promotes the growth of AML progenitors, while suppresses the clonogenic potential of healthy progenitors. Previously, it has been shown that IL-1, in combination of various cytokines, promotes the expansion of healthy progenitors (Fibbe and Willemze, 1991; Haylock et al., 1992; Kovacs et al., 1995). These differential outcomes might be attributed to variation in culture conditions and cell sources. Our findings are consistent with several studies that have shown that the presence of IL-1 restricts the self-renewal potential of murine healthy progenitors due to increased myeloid differentiation, and suppresses colony-forming and in vivo reconstitution abilities of healthy progenitors in a murine model (Pietras et al., 2016; Yonemura et al., 1996).

Importantly, our findings demonstrate a unique, paradoxical, context-dependent role of an IL-1 rich pro-inflammatory environment that may enhance the growth of IL-1-sensitive AML cells and disrupt normal hematopoiesis by a p38MAPK-dependent mechanism (Figure 7). The concept of inflammatory cytokines influencing clonal selection and leukemogenesis has been established previously (Bagby and Fleischman, 2011). A similar paradigm may apply to AML, in which an IL-1 rich environment augments the selective pressure that preferentially stimulates growth of neoplastic clones compared to normal hematopoietic cells. The combination of genetic mutations and cytokine hypersensitivity may create such an environment, one that supports the selection of leukemic over non-leukemic stem cells (Konopleva and Jordan, 2011). However, this proposed model is based on ex vivo experiments using primary AML samples, and murine in vivo competitive repopulation studies are required to further validate the model. In addition, IL-1 stimulation of AML cells promotes the secretion of multiple growth-promoting and inflammatory cytokines in IL-1 sensitive AML progenitors in p38MAPK-dependent manner. While the role of p38MAPK in inflammatory disease is well characterized, recent studies have established its role in leukemia progression (Bachegowda et al., 2016). We demonstrated that IL-1 promotes expansion and survival of AML progenitors with only slight effects on differentiation in a p38MAPK-dependent manner. However, in healthy progenitors, IL-1 promotes short-term proliferation and myeloid differentiation, leading to cell death that is reversed by p38MAPK inhibition, suggesting dependence on this pathway. These results suggest that IL-1 may activate differential transcription programs in AML than in healthy progenitors downstream of p38MAPK. For example, a recent study showed that IL-1 induces the myeloid differentiation of healthy progenitors at the expense of stem progenitors in a PU.1-dependent manner (Pietras et al., 2016). Further, PU.1 activation might be regulated by p38MAPK (Wang et al., 2003). Future studies defining mechanistic differences underlying normal and AML progenitor responses to IL-1 may offer additional insight into the origins of cellular expansion in AML.

Targeting IL-1 signaling has been an attractive therapeutic approach in variety of inflammatory and autoimmune diseases (Dinarello et al., 2012). Blocking IL-1 signaling using an IL-1 receptor antagonist (Estrov et al., 1992), or p38MAPK (Anur et al., 2012), has been implicated as a promising cancer therapeutic including in leukemia. Further, inhibition of IL1RAcP has shown efficacy in suppressing the growth of CML and AML cells (Agerstam et al., 2016; Askmyr et al., 2013). We have identified IL1R1 as an additional specific target in AML, targeting of which ablates in vitro expansion and in vivo disease progression. Furthermore, blocking p38MAPK kinase using a small-molecule inhibitor concomitantly suppresses the growth of AML cells and rescues the expansion of normal progenitor cells. Recent studies show that IL1R1 expression is high in primitive leukemic progenitors in CML and targeting IL-1 signaling eliminates these early progenitors(Agerstam et al., 2016; Zhang et al., 2016). Our data suggests that purified IL1R1 expressing CD34⁺ progenitors are more sensitive to p38MAPK inhibition than IL1R1 non-expressing progenitors. However, in bulk culture both IL1R1 expressing and non-expressing progenitors show sensitivity to p38MAPK inhibition suggesting p38MAPK might be activated by other signaling molecules secreted via IL1R1⁺ cells and CD14⁺ monocytes (Zarubin and Han, 2005). While we cannot rule out that the small-molecule inhibitors used in our studies may block p38MAPK activity through secondary molecules other than IL-1, our data strongly suggest that IL-1 is a principal contributor to the activation of p38MAPK in AML. Additionally, we demonstrate that p38MAPK inhibition suppressed the IL-1-mediated secretion of additional pro-inflammatory cytokines (Figure 6). Given the numerous p38MAPK inhibitors already in clinical development (Bachegowda et al., 2016; Patnaik et al., 2016), our findings provide strong pre-clinical data to extend these efforts to AML. Furthermore, from a clinical application standpoint, our data together suggest that in vitro IL-1 or p38MAPK inhibitor sensitivity, increased expression of IL-1 and its receptors, and/or increased p38MAPK phosphorylation in AML cells may serve as biomarkers for dependence on IL-1 signaling, warranting their potential translation to routinely available clinical screening assays in order to aid stratification of patients for such therapy.

In summary, we demonstrate the functional relevance of aberrant IL-1 signaling in the pathobiology of AML and establish a role of IL1R1 in AML progression. In a malignancy as genetically heterogeneous as AML, our use of in vitro functional screening to identify a unifying pattern of cytokine and growth factor-driven pathway activation underscores the value of this type of assay in interrogating similar dilemmas in other genetically and molecularly complex malignancies. Furthermore, targeting aberrantly activated pathways in the tumor extrinsic environment may offer strategies to circumvent or mitigate the disease. Although additional evidence is required, our findings support a model wherein excessive inflammatory cytokines such as IL-1 may facilitate the selection of neoplastic clones by simultaneously enhancing their growth and exhausting non-neoplastic clones (Figure 7), which is consistent with the suppression of normal stem cell growth observed clinically in AML (Miraki-Moud et al., 2013). Moreover, the potential to selectively reverse both of these effects by p38MAPK inhibition represents a significant advantage over traditional chemotherapy, which is generally toxic to both normal hematopoietic and leukemia cells. Our findings provide a rationale for therapeutic targeting of the IL-1/IL1R1/p38MAPK pathway in AML, with the goal to improve treatment outcomes for a broad range of AML patients. These results may also inform similar strategies in other malignancies.

EXPERIMENTAL PROCEDURES

Detailed methods are in the Supplemental Information.

Primary samples and cell lines

AML samples were obtained from patients evaluated at Oregon Health & Science University, University of Utah, University of Texas Southwestern Medical Center, University of Colorado, and Stanford University. All samples represented unique patients with AML and were categorized according to ex vivo response to IL-I treatment. The clinical, genetic, and demographic information of primary AML samples along with the response to IL-1 is summarized in Table S3. Studies using human cells were approved by the institutional IRBs.

Cytokine library, viability, proliferation, and differentiation assays

Ninety-four recombinant cytokines and growth factors were plated at graded concentrations in 384-well plates (Table S1). Cells were cultured for 72h and subjected to a colorimetric cell viability assay (MTS assay, Promega, Madison). For the MTS assay, the absorbance values of treated cells were compared to untreated control cells, and fold change in viability over untreated cells were evaluated. Conditioned media collected from the HS-5 stromal cells were used as a positive control. To increase stringency, we coded the data based on reproducible response across wells. If an individual sample showed increased growth based on the threshold for sensitivity in both the high and medium concentrations of a given cytokine, it was coded as Category 2 (sensitive samples). If a cytokine showed increased growth in only one well (high or medium concentration) based on the threshold for sensitivity, it was coded as Category 1 (indeterminate sensitivity). Cytokines that were below the threshold for sensitivity in both wells were coded as Category 0 (nonsensitive samples). Cytokines that resulted in decreased growth response compared to no-cytokine wells (based on third quartile of the negative controls) were considered growth suppressive (Categories -1 or -2). The proliferation and differentiation assays are described in the Supplemental Methods.

siRNA, shRNA, and lentivirus

siRNA knockdown was performed as described previously(Agarwal et al., 2014b) and supplemental methods. An inducible shRNA library targeting IL1R1 was obtained from Cellecta, Inc (Mountain View).

Cytokine measurement, quantitative RT-PCR, phosphoflow, intracellular FACS, and immunoblotting

Plasma cytokine levels were measured using IL-1β ELISA kit (R&D Systems, Minneapolis). Conditioned media was evaluated for secreted cytokines by utilizing 30-plex multiplex magnetic bead-based Luminex assay (ThermoFisher Scientific). Phosphoflow, intracellular FACS, immunoblotting, and qPCR analyses were performed as in (Agarwal et al., 2014a) and supplemental methods.

Immunophenotyping of murine bone marrow, colony formation and transplantation assays

All animal procedures were conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee. Bone marrow cells were harvested from 4- to 6-week-old C57/BL6 wild-type and IL1R1-null male mice (Jackson Laboratory, Bar Harbor, ME). Immunophenotyping of murine cells, myeloid colony formation, and bone marrow transplantation were performed as in (Agarwal et al., 2008) and supplemental methods.

Statistical analysis

For siRNA knockdown experiments, a two-tailed Student’s t-test was carried out for each siRNA target in comparison to nonspecific siRNA control. Cell growth, apoptosis, and gene expression were also compared by a two-tailed Student’s t-test for 2 independent treatments or different sample groups. For analysis of demographic, clinical and genetic factors, Fisher’s exact test was utilized for categorical variables (gender, surface markers, complex karyotype, and mutations) and Mann-Whitney test was utilized for continuous variables (age, % blast in bone marrow, % blast in peripheral blood, % monocytes in peripheral blood, and WBC). Level of significance was set at p ≤ 0.05.