Oncogenic KrasG12D causes myeloproliferation via NLRP3 inflammasome activation

Oncogenic Ras mutations occur in various leukemias. It was unclear if, besides the direct transforming effect via constant RAS/MEK/ERK signaling, an inflammation-related effect of KRAS contributes to the disease. Here, we identify a functional link between oncogenic KrasG12D and NLRP3 inflammasome activation in murine and human cells. Mice expressing active KrasG12D in the hematopoietic system developed myeloproliferation and cytopenia, which is reversed in KrasG12D mice lacking NLRP3 in the hematopoietic system. Therapeutic IL-1-receptor blockade or NLRP3-inhibition reduces myeloproliferation and improves hematopoiesis. Mechanistically, KrasG12D-RAC1 activation induces reactive oxygen species (ROS) production causing NLRP3 inflammasome-activation. In agreement with our observations in mice, patient-derived myeloid leukemia cells exhibit KRAS/RAC1/ROS/NLRP3/IL-1β axis activity. Our findings indicate that oncogenic KRAS not only act via its canonical oncogenic driver function, but also enhances the activation of the pro-inflammatory RAC1/ROS/NLRP3/IL-1β axis. This paves the way for a therapeutic approach based on immune modulation via NLRP3 blockade in KRAS-mutant myeloid malignancies.

A ctivating mutations of NRAS and KRAS genes were reported to occur in 18-32% of acute myeloid leukemia (AML) 1,2 , in 11-38% of chronic myelomonocytic leukemia (CMML) 3,4 and in 25-35% of juvenile myelomonocytic leukemia (JMML) patients 5,6 . JMML is an aggressive myeloproliferative disease (MPD) of early childhood characterized clinically by the overproduction of myelomonocytic cells 7 . Other mutations found in this disease include mutations in the tumor suppressor gene NF1 8 , CBL and mutations in PTPN11 9 , which also activate the RAS/extracellular-signal-regulated kinase (ERK) signaling pathway 10 , reviewed in ref. 11 . CMML is a clonal stem cell disorder associated with peripheral blood monocytosis, myeloid cell proliferation with myeloid cell dysplasia and ineffective hematopoiesis, and can transform to AML 12 .
Since the RAS/RAF/MEK/ERK cascade couples signals from cell surface receptors to transcription factors in non-malignant immune cells 13 , we hypothesized that in addition to the oncogenic signal, inflammatory mechanisms may drive the proliferation of myeloid cells and cause fever, wasting syndrome and failure to thrive, which is frequently observed in JMML, CMML and more rarely in AML patients. This hypothesis was also based on our previous work showing that signaling caused by FLT3-ITD and JAK2-V617F oncoproteins can affect the expression of the inflammation-related molecules IL-15 14 and PD-L1 15 , respectively. In support of the concept that RAS/ERK signaling plays a role in inflammation, recent studies had shown that ERK1/2 phosphorylation is critical for allogeneic immune activation during graft-versus-host disease (GVHD) 16 and MEK inhibition blocked these pro-inflammatory events 17 . A major inflammatory mediator in GVHD is the NLRP3 inflammasome 18 and caspase-1, the effector protease of the NLRP3 inflammasome, which is essential for pro-IL-1β maturation 19 .
We found in microarray-based studies that NLRP3 expression was upregulated in murine hematopoietic bone marrow (BM) cells harboring active inducible Kras G12D allele. In agreement with a functional role of NLRP3 in the myeloid compartment, Kras G12D BM-derived dendritic cells (BMDCs) showed increased IL-1β production and caspase-1 activation compared to wildtype (WT) cells. While mice expressing active Kras G12D selectively in the hematopoietic system developed cytopenia and myeloproliferation, these disease features were abrogated in Kras G12D mice lacking NLRP3 in the hematopoietic system. The findings in the mouse models could be recapitulated in patient samples of JMML, CMML, and AML patients carrying activating KRAS mutations. This study shows that oncogenic Kras G12D leads to activation of the RAC1/ROS/NLRP3/IL-1β axis, which could be the basis for therapeutic approaches.
NLRP3 deficiency reverses Kras G12D -induced cytopenia. To investigate a functional role of the NLRP3 inflammasome for the effects induced by oncogenic Kras G12D , we next generated mice with Kras G12D activation and NLRP3 deficiency by crossing Rosa26Cre-ER T2 ; LSL-Kras G12D mice onto a NLRP3-deficient background (Nlrp3 −/− ). The strain was termed Rosa26Cre-ER T2 ; LSL-Kras G12D ; Nlrp3 −/− mice (Kras G12D ; Nlrp3 −/− ). To avoid a possible indirect effect of active Kras in non-hematopoietic cells, we generated BM chimera that had either WT or Kras G12D or Kras G12D and Nlrp3 −/− in the hematopoietic compartment (Fig. 2a). Mice with Kras G12D expression in hematopoietic system were termed Kras G12D BM mice and mice with Kras G12D and Nlrp3 −/− in the hematopoietic system were termed Kras G12D ; Nlrp3 −/− BM mice. We found that the frequency of CD45 positive cells was equal in all three groups and we could show that the myeloablative conditioning followed by the BM transfer led to full engraftment ( Supplementary Fig. S1E, F). We then administered tamoxifen to recipient mice to induce oncogenic Kras G12D and later analyzed the phenotype in the peripheral blood (PB), BM and spleen. We observed an increase in lactate dehydrogenase (LDH) release in KrasG12D BM compared to WT BM, while there was no difference in LDH release in KrasG12D BM compared to KrasG12D; Nlrp3 −/− BM ( Supplementary Fig.  S1G). Kras G12D BM mice developed anemia (decreased hemoglobin concentration and hematocrit) and an increase of reticulocytes (immature red blood cells) that were identified based on their higher size compared to mature erythrocytes and the scattered reticulum network in the cytoplasm which is visible as a blue granular precipitate 21 (Fig. 2b-e). This phenotype was not seen in Kras G12D ; Nlrp3 −/− BM mice (Fig. 2b-e). In addition, Kras G12D BM mice developed low platelet counts and giant platelets were found in the peripheral blood and were not seen in Kras G12D ; Nlrp3 −/− BM mice (Fig. 2f-h).
NLRP3 absence counteracts Kras G12D induced myeloproliferation. We then analyzed the leukocytes in more detail and found increased frequencies of myeloid CD11b + cells in the PB of Kras G12D BM mice which were not seen in Kras G12D ; Nlrp3 −/− BM mice (Fig. 3a, b). Neutrophil granulocytes with a dysplastic phenotype and blasts were more frequent in Kras G12D BM mice compared to WT or Kras G12D ; Nlrp3 −/− BM mice (Fig. 3c-f). Dysplasia was identified via the presence or absence of hyperlobation of the nucleus of neutrophils. Next, we asked whether NLRP3 deficiency would also influence other Kras G12D mediated effects. We found increased amounts of blasts and promonocytes   1 Oncogenic Kras G12D leads to NLRP3 inflammasome activation in murine BM cells. a The heatmap represents the expression of inflammationrelated genes in bone marrow-derived dendritic cells (BMDCs) isolated from either WT (n = 4) or Kras G12D (n = 4) mice following treatment by tamoxifen.
Color code represents the Z-score log2 intensity. b The heatmap represents the expression of inflammasomes upregulated genes from the Reactome gene set "inflammasome" in BMDCs isolated from either WT (n = 4) or Kras G12D (n = 4) mice following treatment by tamoxifen. Color code represents the Z-score log2 intensity. c Western blot shows the amount of caspase-1 (p20 subunit) in WT or Kras G12D BMDCs. The blot is representative for three independent experiments. d The ratio of caspase-1 (p20 subunit)/β-actin in WT (n = 3) or Kras G12D (n = 3) BMDCs normalized to the WT. e Western blot shows the amount of cleaved IL-1β (p17) in WT or Kras G12D BMDCs. The blot is representative for three independent experiments. f The ratio of cleaved IL-1β (p17)/ β-actin in WT (n = 3) or Kras G12D (n = 3) BMDCs normalized to the WT. g The histogram shows mean fluorescence intensity (MFI) for IL-1β in WT or Kras G12D BMDCs. One representative experiment from four experiments with a comparable pattern is shown. h The graph displays the fold change of IL-1β expression as measured by flow cytometry in WT (n = 6) or Kras G12D (n = 6) BMDCs, normalized to the WT in both stimulation conditions. Pooled data of six biological replicates per group is shown. All data are shown as mean ± SEM.  in the BM of Kras G12D BM mice compared to WT or Kras G12D ; Nlrp3 −/− BM mice (Fig. 3g). The number of blasts in the PB was increased in Kras G12D ; Nlrp3 −/− BM mice compared to WT mice (Fig. 3g). When analyzing the BM morphology, Kras G12D BM mice exhibited hypercellularity with reduced lipid-rich adipose cells and clusters of immature granulocytic cells (Fig. 3h, i).
Moreover, Kras G12D BM mice developed splenomegaly, which was not seen in Kras G12D ; Nlrp3 −/− BM mice (Fig. 4a, b). In agreement with this finding, splenomegaly is a disease feature of JMML driven by KRAS activation 22 . In contrast to the spleen, organs that do not belong to the hematopoietic/lymphatic system (e.g. kidney, heart) were not enlarged indicating that the effect of   KRAS activation was restricted to the hematopoietic/lymphatic system. Also, the splenic architecture comprising white pulp lymphoid follicles with germinal centers and marginal zone areas in WT mice was disrupted in Kras G12D BM mice but not in Kras G12D ; Nlrp3 −/− BM mice (Fig. 4c, d). We found abnormal infiltrates of myeloid progenitor cells in the spleens of Kras G12D BM mice but not in Kras G12D ; Nlrp3 −/− BM mice (Fig. 4e). Moreover, myeloid CD11b + cells were increased in the spleens of Kras G12D BM mice but not in Kras G12D ; Nlrp3 −/− BM mice (Fig. 4f). Consistent with these findings in the spleen and BM, we observed higher numbers of granulocytes in the PB of Kras G12D BM mice but not in Kras G12D ; Nlrp3 −/− BM mice (Fig. 4g).
Conversely, the percentages of T cells of all CD45 + cells were decreased in Kras G12D BM mice (Fig. 4h, i).
To evaluate whether the connection between oncogenic Kras G12D and NLRP3 is cell-autonomous, we analyzed if the myeloproliferation caused by the KRAS/NLRP3 axis in vivo would also be active in vitro in the absence of a microenvironment. We observed that proliferation of myeloid cells in vitro was higher when the cells were derived from Kras G12D mice compared to WT mice ( Fig. 5g-i). We also found that proliferation in BMDCs generated from Kras G12D was higher compared to BMDCs generated from Kras G12D ; Nlrp3 −/− mice indicating that NLRP3 deficiency reverses the Kras-induced phenotype (Fig. 5i). However, proliferation in BMDCs generated from Kras G12D ; Nlrp3 −/− mice was higher compared to BMDCs generated from WT mice (Fig. 5i). These findings support the concept that the connection between oncogenic Kras G12D and NLRP3 is partly cell-autonomous and partly via impacting the microenvironment. An immunosuppressive, pro-tumorigenic role of IL-1β in the tumor microenvironment has been reported before 23 .
NLRP3 inhibition reverses myeloproliferation by Kras G12D . To study the functional role of the NLRP3/IL-1β axis, we treated Kras G12D BM mice with the IL-1 receptor type 1 antagonist Anakinra (recombinant IL-1RA) or the NLRP3 inhibitor MCC950 for 4 weeks, starting 3 weeks post-tamoxifen when the disease was established until the day before analysis. We observed increased hematocrit and hemoglobin concentrations in Kras G12D BM mice treated with Anakinra or MCC950 (Fig. 6a, b), compared to Kras G12D BM mice treated with vehicle. In addition, there was a decrease in the percentage of immature red blood cells in PB of Anakinra-treated Kras G12D BM mice, compared to vehicle-treated Kras G12D BM mice (Fig. 6c). Platelet counts increased in Kras G12D BM mice treated with Anakinra or MCC950 (Fig. 6d). Moreover, Anakinra or MCC950 treatment could reverse the spleen and BM phenotype in Kras G12D recipient mice (Fig. 6e, f). Consistent with a reduced KRAS/NLRP3/IL-1β axis, we observed decreased amounts of IL-1β in BMDCs (enrichment purity: Supplementary Fig. S2D) generated from Anakinra-treated or MCC950-treated Kras G12D BM mice compared to vehicle-treated Kras G12D BM mice (Fig. 6g, h). Importantly the inhibitor treatment did not lead to a depletion of a certain cell type (Fig. 6i, Supplementary Fig. S3A, B) but only to a mild reduction of mature neutrophils that did not reach significance (Fig. 6j). These findings indicate that NLRP3 and IL-1R are therapeutic targets to interfere with KRAS-driven myeloproliferation.
Myeloproliferation is mediated via the RAC1/ROS/NLRP3 axis. To understand the mechanism by which oncogenic KRAS leads to NLRP3 activation, we performed an unbiased gene expression analysis of WT or Kras G12D BMDCs. We observed increased expression of genes for Gene Ontology (GO) terms related to NADPH oxidase, an enzyme required for the production of extracellular/endosomal ROS (Fig. 7a). In agreement, we found increased cellular ROS production in Kras G12D BMDCs compared to WT (Fig. 7b, c). Blocking ROS production by PDTC or Ebselen (enrichment purity and viability: Supplementary  Fig. S4A, B) abrogated the increased caspase-1 cleavage seen in Kras G12D BMDCs (Fig. 7d, e).
NLRP3 inflammasome activation in human KRAS-mutant (Kras mut ) leukemia cells. To determine whether the observations made in mouse cells were also seen in human cells, we isolated PBMCs of AML, JMML or CMML patients with KRAS mutations (Kras mut ) and gated on CD11b + cells (Supplementary Fig. S5A). In contrast to AML patients without KRAS mutations (Non- Kras mut ), we observed increased cleaved caspase-1 in Kras mut PBMCs (Fig. 8a, b), indicating increased NLRP3 inflammasome activity in human Kras mut cells compared to non-mutant cells. In agreement with inflammasome activation, we also observed increased IL-1β production in human Kras mut cells compared to non-mutant cells in JMML, AML, and CMML ( Fig. 8c-g). The cell number and phenotype was not affected by the stimulation with LPS ( Supplementary Fig. S5B).
To determine if the KRAS mutations were found in CD11b positive cells, we sorted for these cells, and analyzed the resulting DNA using targeted Next-Generation Sequencing (NGS) and digital droplet PCR (ddPCR) for the respective KRAS mutations. We could show that the KRAS mutations were found in CD11b + sorted cells at a high variant allele frequency (VAF; 45%, 49% in representative samples, Supplementary Fig. S5C-E). In addition, we found increased cleaved caspase-1 in human Kras mut CMML cells compared to non-Kras mut cells (Fig. 8h, i). CMML samples with NRAS mutations (Nras mut ) displayed only a small increase of IL-1β ( Supplementary Fig. S5F).
Increased ROS production in human Kras mut myeloid leukemia cells. Comparable to the observations made in mice, we found that ROS production was increased in human Kras mut cells compared to non-Kras mut cells derived from AML patients in the cytoplasm (Fig. 9a, b) and in the mitochondria (Fig. 9c, d). Blocking ROS production by PDTC or Ebselen reduced IL-1β production (Fig. 9e), and also abrogated the increased caspase-1 cleavage seen in Kras mut JMML PBMCs (Fig. 9f, g). Increased ROS production was also seen in Kras mut cells, but not Nras mut cells, from CMML patients compared to non-Kras mut CMML cells (Fig. 9h, i, Supplementary Fig. S5G). Blocking ROS production by PDTC or Ebselen reduced IL-1β production in CMML patient Kras mut PBMCs (Fig. 9j). In addition, and similar to our findings in murine models, blocking RAC1 in Kras mut cells, but not Nras mut cells, from CMML patients caused a reduction of IL-1β expression and ROS production (Fig. 9k, l, Supplementary  Fig. S5H, I).
These observations in patient samples show that the KRAS/ ROS/NLRP3/IL-1β axis was also active in human AML, CMML and JMML cells, which is in agreement with our observations in mouse models.
Our findings indicate that oncogenic KRAS leads to RAC1 activation, causing NADPH activation, resulting in ROS production, which in turn activates the NLRP3 inflammasome causing cleavage of caspase-1 and IL-1β production ( Fig. 10; proposed model).

Discussion
Improving the treatment of JMML, CMML, and KRAS-mutant AML is still a major unmet clinical need. Conventional chemotherapy in JMML has little effect 24 and the probability of event-free survival at 5 years is only 50% after allogeneic hematopoietic cell transplantation (allo-HCT) 25 . Similarly, allo-HCT remains the only curative option for patients with CMML, with an overall survival at 3 years of 32% 12 . A major clinical feature of patients with JMML are inflammatory symptoms including hepatosplenomegaly in greater than 90%, lymphadenopathy in 76%, pallor in 64%, fever in 54%, skin rash in 36% 22,26 . Also in CMML hepatomegaly, splenomegaly and features of myeloproliferation, such as fatigue, symptoms from organomegaly, night sweats, weight loss and cachexia have been reported 12 . The pathomechanism causing these symptoms in KRAS-mutant myeloid neoplasms was not fully elucidated. Here we report that multiple inflammation-related genes are upregulated upon oncogenic KRAS activation in myeloid cells. A major inflammatory pathway we identified was the NLRP3/ caspase-1/IL-1β axis. This is consistent with previous reports showing that IL-1β was increased in JMML 27,28 . These reports, however, did not clarify the mechanism underlying the increased IL-1β production, because the role of the inflammasome for IL-1β activity was unknown at this time. Our work identifies a critical role of the NLRP3 inflammasome in Kras G12D induced inflammation, as the hematological phenotype caused by oncogenic Kras G12D expression is reversed by NLRP3 deficiency. Our studies delineate that oncogenic KRAS activates RAC1 which in turn leads to the production of ROS. The impaired hematopoiesis and splenomegaly that we observed when the NLRP3/IL-1β axis was active upon KRAS activation is very similar to the hepatosplenomegaly and pancytopenia reported in patients with activating mutations of NLRP3 that suffer from Cryopyrin-associated periodic syndrome (CAPS) 29 . In agreement with our findings, CAPS patients experience an improvement of their symptoms upon IL-1β blockade 29 . The regulation of cytokine production via oncogene/growth factor receptor activation has been previously reported for FLT3-ITD signaling blocking IL-15 production 14 , promoting lytic cell death and inflammasome activation 30 , cMYC induced IL-10 31 and TNF-α 32 production and EGFR induced IL-6 production 33 . The production of these cytokines may shape the tumor microenvironment in a favorable fashion for tumor cells. It is possible that besides IL-1β, also IL-18 plays a role because attenuating IL-1β with Anakinra produces a less robust protective effect compared to blocking NLRP3.  6 IL-1R-or NLRP3-inhibition in Kras G12D mice leads to reduced myeloproliferation. a The percentage of hematocrit in PB of vehicle-treated Kras G12D (n = 15), Anakinra-treated Kras G12D (n = 10) or MCC950-treated Kras G12D (n = 15) BM mice. Both Anakinra and NLRP3 inhibitor (MCC950) treatment started 3 weeks post-tamoxifen and ended a day before sacrificing the mice (treatment was in total for 4 weeks). b The concentrations of hemoglobin in PB of vehicle-treated Kras G12D (n = 15), Anakinra-treated Kras G12D (n = 10) or MCC950-treated Kras G12D (n = 15) BM mice. c The percentage of immature red blood cells observed in PB of vehicle-treated Kras G12D (n = 15), Anakinra-treated Kras G12D (n = 10) or MCC950-treated Kras G12D (n = 15) BM mice. d Shown are the number of platelets in PB of vehicle-treated Kras G12D (n = 15), Anakinra-treated Kras G12D (n = 10) or MCC950-treated Kras G12D (n = 15) BM mice. e The number of blast cells counted in BM isolated from vehicle-treated Kras G12D (n = 5), Anakinra-treated Kras G12D (n = 6) or MCC950-treated Kras G12D (n = 15) BM mice. The counts represent percentage of 100 analyzed cells. f The number of spleen follicles counted in spleens isolated from vehicle-treated Kras G12D (n = 5), Anakinra-treated Kras G12D (n = 6) or MCC950-treated Kras G12D (n = 15) BM mice, per 1 high power field (HPF). g The histogram shows mean fluorescence intensity (MFI) for IL-1β in BMDCs of vehicle-treated Kras G12D , Anakinra-treated Kras G12D or MCC950-treated Kras G12D BM mice. One representative experiment from three experiments is shown. h The graph displays the fold change of IL-1β expression as measured by flow cytometry in BMDCs of vehicle-treated Kras G12D (n = 15), Anakinra-treated Kras G12D (n = 10) or MCC950-treated Kras G12D (n = 15) BM mice, normalized to vehicletreated Kras G12D in both stimulation conditions. i FlowSOM-guided metaclustering of CD45 + BM cells (gated on live/single cells/CD45 + ) per condition: vehicle-treated Kras G12D (n = 4), Anakinra-treated Kras G12D (n = 5) or MCC950-treated Kras G12D (n = 5) BM mice. j Frequency of FlowSOM-guided mature neutrophils cluster among CD45 + BM cells in vehicle-treated Kras G12D (n = 4), Anakinra-treated Kras G12D (n = 5) or MCC950-treated Kras G12D (n = 5) BM mice. All data are shown as mean ± SEM.
Regarding the relevant cell type, the scenario that activation of NLRP3 in non-leukemic cells contributes as a driver of the disease is conceivable.
We also clarify the mechanism by showing that KRAS-induced ROS production leads to NLRP3 activation. Previous work had shown that ROS can activate NLRP3 34 . Functional inhibition of NADPH oxidase-derived ROS prevented ATP-induced caspase-1 activation and IL-1β production in alveolar macrophages 35 . Consistently we had observed NADPH gene expression being increased upon KRAS activation and that blocking ROS production reduced caspase-1 activation and IL-1β cleavage in KRAS-mutant mouse and human cells which has important clinical implications. Our findings support the concept that oncogenic KRAS does not only act via its oncogenic driver function but also enhances activation of the NLRP3/IL-1β axis, which explains the clinical features of JMML and CMML patients suffering from inflammation-related symptoms. Even when NLRP3 inhibition or IL-1R blockade could not eliminate the malignant clone, this therapy could reduce the symptoms of patients such as massive splenomegaly, unexplained fever and failure to thrive, which would be a major benefit for the patients and bring them into a condition that is more stable before undergoing the potentially curative allo-HCT.
In summary, the data suggest NLRP3 and IL-1R inhibition as a strategy for the treatment of KRAS-driven hematological malignancies.

Methods
Human subjects. Human sample collection and analysis were approved by the Institutional Ethics Review Board of the Medical center, University of Freiburg, Germany (protocol numbers 10024/13, 26/11, 509/16). Written informed consent was obtained from each patient. All analysis of human data was carried out in compliance with relevant ethical regulations. The characteristics of patients are listed in Supplementary Tables S1-S3.

Isolation of patient-derived peripheral blood mononuclear cells (PBMCs).
PBMCs were isolated from patients' blood by Ficoll gradient centrifugation, according to the manufacturer's protocol (Sigma-Aldrich), and frozen in Fetal Calf Serum (FCS) with 10% Dimethyl sulfoxide (DMSO). Once thawed, cells were plated in 6-well plates at 0.5 × 10 6 cells per well, and left for 24 h to recover. If stated, PBMCs were stimulated with 200 ng/ml of LPS followed by 5 mM of ATP, with a minimum of 4 h in between. Cells were gated on CD11b + cells.
Bone marrow transplantation (BMT) and generation of BM chimera. Recipient mice were lethally irradiated with split doses of 5 Gy and 4.5 Gy (137 Cs source), with a minimum period of 4 h in between, followed by retro-orbital venous plexus injection of 5 × 10 6 donor BM cells. Peripheral blood (PB) samples were obtained at different time points, and PB counts were performed using the scil Vet abc analyzer (Henry Schein). For BMT studies, the experiment was performed 3 times and pooled data of the indicated number of biological replicates per group is shown.
Treatments. Tamoxifen free base (Sigma T5648) was prepared as 20 mg/ml in corn oil and administrated to mice at a single dose of 1 mg per mouse, via oral gavage. The administration of tamoxifen induces nuclear translocation of Cre-ER T2 , causing the deletion of the LSL elements in Cre-ER T2 , Kras G12D mice 37 . A proof of successful recombination was determined using the polymerase chain reaction (PCR) protocol established by the Jacks laboratory (https://jacks-lab.mit. edu/protocols/genotyping/kras_cond) and Western blotting ( Supplementary  Fig. S1A, B). To assess the role of NLRP3 and IL-1 (IL-1 α and ß) in the disease phenotype, mice were treated with an IL-1 type 1 receptor antagonist or a NLRP3 inflammasome inhibitor. Anakinra (Kineret®; Swedish Orphan Biovitrum -sobi) was administered once daily via subcutaneous injection at a dose of 10 mg/kg. MCC950 38 was administered once daily via intraperitoneal injection at a dose of 30 mg/kg. Mice from the vehicle group were treated with phosphate-buffered saline (PBS). Both Anakinra and NLRP3 inhibitor (MCC950) treatment started 3 weeks post-tamoxifen and ended one day before sacrificing the mice (treatment was in total for 4 weeks).

Generation of bone marrow-derived dendritic cells (BMDCs) and stimulation.
Bone marrow cells were cultured at 6 × 10 6 cells in the presence of 40 ng/ml GM-CSF in 10 ml of RPMI 1640 medium (Gibco) in 10 cm dishes. On day 3, 10 ml of fresh medium containing 40 ng/ml GM-CSF was added. On day 7, 10 ml medium was replaced with fresh medium containing 40 ng/ml GM-CSF. Cells were used on days 7 or 8 with a CD11c + purity of >80%, as determined by flow cytometry. If stated, BMDCs were stimulated with 200 ng/ml of Lipopolysaccharide (LPS, Sigma-Aldrich) for 4 h, followed by 5 mM of Adenosine 5′-triphosphate disodium salt (ATP; Sigma A7699-1G) for 30-60 min. For ELISA and colorimetric assay, BMDCs were primed with 200 ng/ml E.coli K12 ultra-pure LPS (InvivoGen) for 3-9 h and subsequently stimulated with either 5 µM of Nigericin (Sigma-Aldrich) for 90 min, 5 mM of ATP for 90 min or left without secondary stimulus. All stimulations were performed at least in biological and technical triplicates.
Immunohistochemical and histological analysis. Tissue sections of tibia, femur, and spleen were collected at the terminal point of the experiment, fixed in 4% paraformaldehyde for 12 to 16 h at 4°C and washed in 70% ethanol. Tissues were embedded in paraffin and cut into 5-μm sections, followed by the standard protocol of Hematoxylin/Eosin (H/E) staining. Blood smears were stained with Pappenheim stain. The results were assessed, graded and quantified by an investigator blinded to the experimental groups.
Flow cytometry. The fluorochrome-conjugated antibodies mAbs (Clone) used for flow cytometric analysis are shown in Supplementary Table S4. To examine cell viability and exclude dead cells, the LIVE/DEAD Fixable Dead Cell Stain kit (Molecular Probes) or the Zombie NIR Fixable Viability kit (Biolegend) were used. For intracellular staining, the BD Cytofix/Cytoperm kit (BD Biosciences) was used following the manufacturer's protocol. Data were acquired on a BD LSR Fortessa flow cytometer (BD Bioscience) and analyzed using the FlowJo (Flowjo 10.4, LLC) software. Multiparameter high-dimensional data was acquired on a FACSSymphony (BD Bioscience) and compensated in FlowJo (V10.6), then live, single CD45 + cells were exported and analyzed using the R environment 39 . Data were processed for FlowSOM clustering 40 . For dimensionality reduction the UMAP package was used 41 . Fig. 7 Kras G12D causes inflammasome activation RAC1 and reactive oxygen species (ROS) production. a The heatmap shows genes for upregulated GO terms for NADPH in BMDCs generated from either WT (n = 4) or Kras G12D (n = 4) mice following treatment by tamoxifen. Color code represents the Zscore log2 intensity. b The histogram shows mean fluorescence intensity (MFI) for H2DCF-DA in WT or Kras G12D BMDCs. c The graph displays the fold change of H2DCF-DA expression as measured by flow cytometry in WT (n = 9) and Kras G12D (n = 9) BMDCs, normalized to WT in both stimulation conditions. d Western blot shows the amount of caspase-1 (p20 subunit) in BMDCs isolated from Kras G12D mice treated with DMSO, Ebselen or PDTC, after stimulation with 200 ng/ml LPS and 5 mM ATP. The blot is representative for two independent experiments. e The ratio of caspase-1 (p20 subunit)/ β-actin in Kras G12D BMDCs treated with DMSO (n = 3), Ebselen (n = 3) or PDTC (n = 3), normalized to DMSO-treated Kras G12D BMDCs. f Western blot shows the amount of caspase-1 (p20 subunit) in BMDCs generated from Kras G12D mice treated with DMSO, 50 µM or 100 µM of RAC1 inhibitor NSC 23766, after stimulation with 200 ng/ml LPS and 5 mM ATP, normalized to DMSO-treated Kras G12D BMDCs. The blot is representative for two independent experiments. g The ratio of caspase-1 (p20 subunit)/Vinculin in Kras G12D BMDCs treated with DMSO (n = 3), 50 µM (n = 3) or 100 µM (n = 3) of RAC1 inhibitor NSC 23766, normalized to DMSO-treated Kras G12D BMDCs. h The graph displays the fold change of IL-1β expression as measured by flow cytometry in Kras G12D BMDCs treated with DMSO (n = 6), 50 µM (n = 6) or 100 µM (n = 6) of RAC1 inhibitor NSC 23766, after stimulation with 200 ng/ml LPS and 5 mM ATP, normalized to DMSO-treated Kras G12D BMDCs. i Western blot shows the amount of active RAC1-GTP in BMDCs generated from WT or Kras G12D mice. j The ratio of active RAC1-GTP/Vinculin in BMDCs generated from WT (n = 3) or Kras G12D (n = 3) BM, normalized to WT. k The graph displays the fold change of H2DCF-DA expression in BMDCs generated from Kras G12D mice, treated with DMSO (n = 6), 50 µM (n = 6) or 100 µM (n = 6), normalized to DMSO-treated Kras G12D BMDCs. All data are shown as mean ± SEM. Reactive oxygen species (ROS) analysis and inhibition. 2′, 7′-dichlorofluorescein diacetate (H2DCF-DA; Sigma D6883) was used to detect cellular ROS. One vial was dissolved in dimethyl sulfoxide (DMSO) to prepare 20 mg/ml stock solution. Briefly, cells were pelleted and resuspended in 1.16 µg/ml of H2DCF-DA and incubated at 37°C in the dark for 30 min. The cells were pelleted and resuspended in fresh medium along with stimulus and incubated at 37°C in the dark for 60 min. After washing, cells were stained to their corresponding surface antigen at 4°C in the dark for 20 min and washed before acquisition on flow cytometer. MitoSOX TM red reagent (Thermofisher M36008) was used as an indicator of mitochondrial ROS (mtROS) in live cells. One vial was dissolved in DMSO to make a 5 mM MitoSOX TM stock solution. Cells were incubated in 5 µM MitoSOX TM in HBSS/Ca 2+ /Mg 2+ solution (Gibco 14025) at 37°C for 30 min. ROS signal was measured with a BD LSR Fortessa flow cytometer and analyzed using FlowJo software. For inhibition of ROS, 30 μM of Ebselen (Enzo Life Sciences), 50 μM ammonium pyrrolidinedithiocarbamate (PDTC; Enzo Life Sciences) or DMSO as a control, were added 2.5-3 h after priming with LPS and 20-30 min before stimulation with ATP.
Microarray analysis. RNA was isolated from BMDCs using the RNeasy Mini Kit (Qiagen). RNA quality was assessed by Agilent 2100 Bioanalyzer (Agilent 141 Technologies). RNA samples with an RNA integrity number (RIN) greater than 8 were further processed. Affymetrix Clariom S Mouse arrays were normalized using Single-Channel Array Normalization and exon expression summarized to the gene level using the R/Bioconductor package pd.clariom.s.mouse. Differential gene expression analysis between Kras G12D BM versus WT was calculated using the R/ Bioconductor limma package with p-values corrected for multiple testing using Benjamini & Hochberg. Gene set enrichment analysis (GSEA) was performed using the R/Bioconductor package GAGE which tests for significant differential regulation of a gene set instead of individual genes. The complete gene expression data are available in the GEO repository under the access ID GSE131885 (https://www. ncbi.nlm.nih.gov/geo/query/acc.cgi?&acc=GSE131885).
Proliferation analysis. On day 7, the number of BMDCs generated from WT or Kras G12D mice was counted per 10 cm dish. Biological replicates each mouse was considered, and the average was taken. For the proliferation analysis, the Click-iT™ EdU Pacific Blue™ Flow Cytometry Assay kit (Invitrogen™; C10418) was used, following the manufacturer's protocol. Briefly, 10 µM of EdU was added to the culture medium and incubated for 1-2 h. After cells were harvested and washed with 1% BSA in PBS, extracellular staining proceeded. Following, cells were fixed using the Click-iT™ fixative for 15 min at room temperature, and subsequent washes were performed using the 1X Click-iT™ saponin-based permeabilization and wash reagent. Lastly, cells were incubated for 30 min at room temperature protected from light with the Click-iT™ reaction cocktail composed of CuSO 4 , Pacific Blue dye, reaction buffer additive and PBS, following the manufacturer's instructions. Cells were later analyzed by flow cytometry. Fig. 9 The KRAS/RAC1/ROS/NLRP3 axis is active in human leukemia cells. a The histogram shows mean fluorescence intensity (MFI) for H2DCF-DA in CD11b + cells of AML patients with a KRAS mutation (Kras mut ) or without a KRAS mutation (non-Kras mut ). b The graph displays MFI fold change of H2DCF-DA expression in Kras mut (n = 11) and non-Kras mut (n = 10) AML patient cells, normalized to non-Kras mut in both stimulation conditions. c The histogram shows MFI for MitoSOX in Kras mut or non-Kras mut AML patients cells. d The graph displays MFI fold change of MitoSOX expression in Kras mut (n = 16) and non-Kras mut (n = 11) patient cells, normalized to non-Kras mut in both stimulation conditions. e The graph displays MFI fold change of IL-1β expression in Kras mut JMML patients cells, treated with DMSO (n = 5), Ebselen (n = 5) or PDTC (n = 5), normalized to DMSO-treated Kras mut . f Western blot shows the amount of caspase-1 (p20 subunit) in PBMCs isolated from Kras mut JMML patients treated with DMSO, Ebselen or PDTC after stimulation with 200 ng/ml LPS and 5 mM ATP. g The ratio of caspase-1 (p20 subunit)/β-actin in Kras mut PBMCs derived from JMML patients, treated with DMSO (n = 3), Ebselen (n = 3) or PDTC (n = 3), normalized to DMSO-treated Kras mut cells. h The histogram shows MFI for H2DCF-DA in Kras mut or non-Kras mut CMML patients cells. i The graph displays MFI fold change of H2DCF-DA expression in Kras mut (n = 5) and non-Kras mut (n = 13) CMML patients cells, normalized to non-Kras mut in both stimulation conditions. j The graph displays MFI fold change of IL-1β expression in Kras mut CMML patients cells, treated with DMSO (n = 5), Ebselen (n = 5) or PDTC (n = 5), normalized to DMSO-treated Kras mut cells. k The graph displays MFI fold change of IL-1β expression in Kras mut CMML patients cells, treated with DMSO (n = 5) or 100 µM of RAC1 inhibitor NSC 23766 (n = 5), after stimulation with 200 ng/ml LPS and 5 mM ATP, normalized to DMSO-treated Kras mut . l The graph displays MFI fold change of H2DCF-DA expression in Kras mut CMML patients cells, treated with DMSO (n = 5) or 100 µM of RAC1 inhibitor NSC 23766 (n = 5), after stimulation with 200 ng/ml LPS and 5 mM ATP, normalized to DMSO-treated Kras mut . All data are shown as mean ± SEM.  Fig. 10 Proposed model. The scheme shows the proposed mechanism. Oncogenic KRAS leads to NLRP3/ASC transcription and RAC1-mediated production of reactive oxygen species (ROS). This licenses the NLRP3 inflammasome and activates the inflammasome by ROS leading to caspase-1 activation, ASC recruitment and consecutively to the release of bioactive IL-1β. The KRAS/NLRP3/IL-1β axis can be inhibited by blocking IL-1R or by inhibiting the NLRP3 inflammasome.
Targeted Next Generation Sequencing (NGS). The Illumina TruSight Tumor 15 and TruSight Myeloid Panels were used for targeted resequencing and processed as described by the manufacturer (Illumina Inc.). Sequencing libraries were sequenced paired-end on an Illumina NextSeq 550 with 2 × 150 base reads and FASTQ-files were further analyzed with the SeqNext software (JSI Medical Systems). We used a significance threshold of 3% for the detection of missense mutations, with a minimum coverage of 500 reads and 50 reads per variant.
Digital droplet polymerase chain reaction (ddPCR). ddPCR was used to validate selected mutations in KRAS (G12V, G12A). The ddPCR reaction volume was 20 µL and composed of 10 µL of dPCR supermix for probes (Bio-Rad Laboratories GmbH), 900 nmol/L of each primer and 250 nmol/L probes. Each reaction mixture was partitioned into approximately 20,000 droplets using a QX200 automated droplet generator (Bio-Rad Laboratories GmbH) and then cycled under set conditions. Cycled droplets were read in the QX 200 droplet-reader and the analysis of the data was performed using the QuantaSoft analysis software (Version 1.0, Bio-Rad Laboratories GmbH). The threshold between the positive and negative droplet clusters was manually set for both fluorochrome channels for each sample. Data analysis was performed in those samples were 10,000 or more accepted droplets were obtained.
Lactate dehydrogenase (LDH) release assay and enzyme-linked immunosorbent assay (ELISA). Lytic cell death was quantified by determining LDH levels in the cell-free supernatant using the CytoTox 96 Non-Radioactive Cytotoxicity kit (Promega G1780), according to the manufacturer's instructions. Absorbance of medium was measured to serve as blank value and was subtracted from the sample values. Results are shown as percentage of 100 % dead cells after lysis with 0.8% Triton X-100 for 45 min. IL-1β quantification of cell-free supernatants was performed using ELISA, following the manufacturer's instructions (Invitrogen 88701377). Measurements were performed at least in technical triplicates and absorbance was detected using a microplate reader.
Reticulocyte counting. Reticulocytes were distinguished from red blood cells by 2 criteria: 1. Size: reticulocytes are larger than erythrocytes: 10-15 mm vs. 6-8 mm 2. Morphology: Reticulocytes have a scattered reticulum network in the cytoplasm which is visible as a blue granular precipitate. All experiments were reproducible and were performed at least 3 times.
Statistical analysis. GraphPad Prism v7.01 was used for statistical analysis. A sample size of at least n = 5 per group was determined by 80% power to reach a statistical significance of 0.05 to detect an effect size of at least 1.06. For statistical analysis an unpaired t-test (two-sided) was applied. All data were tested for normality applying the Kolmogorov-Smirnov test. If the data did not meet the criteria of normality, the Mann-Whitney U test was applied. Data are presented as mean and s.e.m. (error bars). Differences were considered significant when the p-value was <0.05.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.