Understanding Early Stage Myelodysplastic Syndrome Pathobiology

Understanding myelodysplastic-syndrome (MDS) pathobiology is critical to derive and evaluate new therapeutic interventions. Transforming growth factor-beta (TGFβ) signaling is abnormally active in early-stage MDS patient hematopoietic stem and progenitors, and the level of TGFβ-signal-induced transcriptional changes may have prognostic value. However, the mechanism of TGFβ signaling resulting in ineffective hematopoiesis in MDS is not known. TGFβ signaling is tightly controlled by antagonists (e.g. SMAD7) which block promiscuous activity. Upon TGFβ receptor signaling, these antagonists are transiently eliminated to amplify the signal; however, they are re-induced by downstream signaling and subsequently terminate the TGFβ signal (a negative feedback loop). The Verma lab demonstrated that human MDS marrow cells have elevated levels of microRNA-21 (miR-21) that target SMAD7. Their work identifies miR-21 as one factor in MDS that interferes with the TGFβ negative feedback loop to generate a chronic TGFβ signal.

We used transient transcription assays to reveal that miR-21 targets the SKI corepressor, a TGFβ-signaling antagonist that binds to R-SMADS to block downstream SMAD-target gene activation. Immunofluorescence staining of primary bone-marrow samples from early-stage MDS patients (with elevated miR-21 and chronic TGFβ) shows significant reduction of not only SMAD7 but also SKI protein. TGFβ signaling induces ubiquitin mediated degradation of SKI, so low SKI expression in MDS is the sum of miR-21 and chronic TGFβ effects. To determine the functional impact of low SKI we examined Ski^-/- mice. We show that SKI loss of function is itself sufficient to impair hematopoietic stem cell (HSC) fitness. While flow cytometric evaluation of Ski^-/- hematopoiesis indicates little phenotypic change in HSC fractions, and Ski^-/- HSC engraft lethally irradiated recipients, competitive HSC transplantation reveals profound Ski^-/- HSC fitness defects. Ski-like (Sno/Skil) encodes a SKI-related compressor that is not a miR-21 target. Ski^-/- defects are not phenocopied in Skil^-/- HSC; underscoring the specificity of the HSC defects to the hypothesized miR-21-SKI axis. In agreement with this, retroviral or lentiviral vector forced expression of miR-21 in HSC is counter-selected within 60 days after transplantation. The Growth factor independent-1 (GFI1) transcriptional repressor is required for HSC fitness and is an independent prognostic indicator in AML/MDS. We previously demonstrated GFI1 transcriptional repression of miR21 . Subsequently, SKI protein levels are significantly decreased in Gfi1^-/- bone marrow cells (which express elevated miR-21). SKI expression is rescued in Gfi1^-/-miR-21^-/- knockout mice but Gfi1^-/- HSC fitness is not restored. Thus, it is unlikely that loss of SKI downstream of Gfi1 loss of function explains Gfi1^-/- HSC fitness defects. Instead, we show that the SKI corepressor binds the GFI1 transcriptional repressor. Native GFI1 and SKI proteins are in the same complex by immunoprecipitation, and synthetic-protein pull-down assays validated direct interaction and mapped the interacting domains. ChIP Seq reveals that Ski is coincident with Gfi1 on a subset of common peaks. We conclude that: 1) in normal cells GFI1 repression of miR-21 protects expression of SKI, an HSC-relevant GFI1 corepressor, and 2) chronic TGFβ signaling in early stage MDS directly affects HSC fitness by targeting SKI.