Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Biology (Cell Biology, Microbiology, Molecular Biology)

Major Professor

Alan List, M.D.

Committee Member

Sheng Wei, M.D.

Committee Member

PK Epling-Burnette, PharmD., Ph.D.

Committee Member

Javier Pinilla-Ibarz, M.D., Ph.D.


caspase-1, NADPH oxidase, pyroptosis, S100A9, β-catenin


Myelodysplastic syndromes (MDS) are genetically diverse hematopoietic stem cell malignancies that share a common phenotype of cytological dysplasia, ineffective hematopoiesis and aberrant myeloid lineage maturation. Apoptotic cell death potentiated by inflammatory cytokines has been considered a fundamental feature of MDS for over two decades. However, this non-inflammatory form of cell death cannot account for the inflammatory nature of these disorders. We report that a hallmark of lower-risk (LR) MDS is activation of the NLRP3 inflammasome, which drives clonal expansion and pyroptosis, a caspase-1-dependent programmed cell death induced by danger-associated molecular pattern (DAMP) signals. Independent of genotype, MDS hematopoietic stem and progenitor cells (HSPC) overexpress pyroptosis-related transcripts, inflammasome proteins and manifest activated NLRP3 inflammasome complexes that direct caspase-1 activation, IL-1β and IL-18 maturation and pyroptotic cell death. Using the S100A9 transgenic (S100A9Tg) mouse model that phenocopies human MDS, we demonstrated that forced expression of S100A9 was sufficient to drive pyroptosis in vivo, implicating pyroptosis as the principal mechanism of HSPC cell death in S100A9Tg mice. The lytic cell death releases intraceullar contents that include alarmins and catalytically active ASC specks, which can propagate bystander inflammation. Notably, MDS mesenchymal stromal cells (MSC) and stromal-derived linages were found to predominantly undergo pyroptosis, with marked activation of caspase-1 and NLRP3 inflammasome complexes. These findings may account for the clusters of both HSPC and stromal cell death previously described in the bone marrows of patients with MDS.

Mechanistically, pyroptosis is triggered by the alarmin S100A9 that is found in excess in MDS HSPC and bone marrow (BM) plasma. Further, both somatic gene mutations and S100A9-induced signaling activate NADPH oxidase (NOX), generating reactive oxygen species (ROS) that initiate cation influx, cell swelling and β-catenin activation. Accordingly, ROS and active β-catenin were significantly increased in MDS BM mononuclear cells (BM-MNC) and S100A9Tg mice compared to normal controls, as well as in human cell lines harboring gene mutations and in murine models of gene mutation knock-in or gene loss. ROS and β-catenin nuclear translocation were significantly reduced by NLRP3 or NOX inhibition, indicating that S100A9 and somatic gene mutations prime cells to undergo NOX1/4-dependent NLRP3 inflammasome assembly, pyroptosis and β-catenin activation. Together, these data explain the concurrent proliferation and inflammatory cell death characteristic of LR-MDS.

Given that loss of a gene-rich area in del(5q) disease results in derepression of innate immune signaling, we hypothesized that this genetic deficit would trigger assembly of the NLRP3 inflammasome complex, akin to the pathobiological mechanism characteristic of non-del(5q) MDS. To this end, we utilized two distinct murine models of del(5q) disease, namely in the context of Rps14 haploinsufficiency and concurrent loss of mDia1 and microRNA (miR)-146a. In both models, pyroptosis was not evident in the HSPC compartment; however, early erythroid progenitors displayed high fractions of pyroptotic cells. This was associated with significant increases in caspase-1 and NLRP3 inflammasome activation, ROS and nuclear localization of β-catenin, which was extinguished by inflammasome or NOX complex inhibition. These data suggest that early activation of the inflammasome drives cell death and prevents terminal maturation of erythroid precursors, accounting for the progressive anemia characteristic of del(5q) disease, whereby hematopoietic defects are primarily restricted to the erythroid compartment. Importantly, these data implicate a similar pathobiological mechanism in del(5q) MDS as is observed in non-del(5q) patients.

The identification of the NLRP3 inflammasome as a pathobiological driver of the LR non-del(5q) and del(5q) MDS phenotype allows for novel therapeutic agent development. Notably, knockdown of NLRP3 or caspase-1, neutralization of S100A9, and pharmacologic inhibition of NLRP3 or NOX suppresses pyroptosis, ROS generation and nuclear β-catenin in MDS, and are sufficient to restore effective hematopoiesis. In del(5q) murine models, inhibition of the NLRP3 inflammasome significantly improved erythroid colony forming capacity by a mechanism distinct from that of lenalidomide, highlighting the translational potential for targeting this innate immune complex in this subset of MDS. Thus, alarmins and founder gene mutations in MDS license a common redox-sensitive inflammasome circuit, which suggests new avenues for therapeutic intervention.

Furthermore, aggregated clusters of the NLRP3 adaptor protein ASC [apoptosis-associated speck-like protein containing a caspase activation and recruitment domain (CARD)] are referred to as ASC specks. During pyroptosis, ASC specks are released from dying cells and function as DAMP signals that propagate inflammation. In this way, specks are a surrogate marker for NLRP3 inflammasome activation and pyroptotic cell death. Given that pyroptosis is the predominant mechanism of cell death in MDS and ASC specks are readily quantified by flow cytometry, we hypothesized that BM or peripheral blood (PB) plasma-derived ASC specks may be a biologically rational biomarker for the diagnosis of MDS.

The percentage of ASC specks were significantly increased in MDS BM plasma compared to normal, healthy donors, which was validated by confocal microscopy. PB plasma-derived ASC specks were significantly greater in LR- versus HR-MDS, consistent with the greater extent of cell death and myeloid-derived suppressor cell (MDSC) expansion in LR disease. As hyperglycemia induces NLRP3 inflammasome activation, plasma glucose levels were measured to adjust for this confounding variable. Subsequently, the percentage of glucose-adjusted, PB plasma-derived ASC specks was measured in a panel of specimens of varied hematologic malignancies. The corrected percentage of ASC specks was significantly increased in MDS compared to normal donors and to each other malignancy investigated, including other myeloid and lymphoid leukemias, myeloproliferative neoplasms and overlap syndromes, like chronic myelomonocytic leukemia (CMML). These data indicate that the glucose-adjusted ASC speck percentage is MDS-specific and may be a valuable diagnostic biomarker. At a cutoff of 0.039, the biomarker minimizes misclassification error and achieves 95% sensitivity and 82% specificity in classifying MDS from normal donors, other hematologic malignancies and T2D. Lastly, the biomarker declined with treatment response to lenalidomide in LR-MDS patients, but not to erythropoietin stimulating agent (ESA) or hypomethylating agent (HMA) therapy. As such, the percentage of ASC specks represents the first biologically rational, diagnostic biomarker for MDS that can be implemented with current diagnostic practices to reduce diagnostic error.

Available for download on Wednesday, April 04, 2018