Depletion of miR-143 and miR-145 supports clonal expansion of human HSPC
To examine the role of miR-143 and miR-145 in human cells, we first confirmed that their expression was lower in CD34+ cells from del(5q) MDS patient marrow compared to MDS marrow diploid at chromosome 5q, as previously reported15 (Fig. 1A). Expression of both miRNAs was higher in the CD34+ stem/progenitor cell fraction compared to the more differentiated CD34− cells, suggesting a role for miR-143 and miR-145 in HSPC (Fig. 1B). To model miR-143/miR-145 depletion in human HSPC, we knocked down miRNA expression using lentiviral decoy constructs in CD34+ HSPCs (Supplemental Fig. 1A, B). In functional assays for hematopoietic progenitor activity, the miR-143/miR-145 decoy-transduced cells produced more colonies than vector-transduced cells in both primary and serial replating progenitor assays (Fig. 1C). The effect on colony formation was seen in the more-primitive erythroid (BFU-E) with a trend towards an increase in the primitive mix (CFU-GEMM) colonies (Supplemental Fig. 1C). The decoy-transduced cell cultures had enhanced viability, as estimated by Annexin V and propidium iodide (PI) staining (Fig. 1D). Thus, depletion of miR-143 and miR-145 enhances viability of CD34+ HSPC and promotes clonogenic potential in vitro.
To determine the impact of miR-143 and miR-145 depletion on hematopoiesis in vivo, we injected CD34+ HSPC, two days post-lentiviral transduction with either the decoy or control vectors, into the femurs of sublethally-irradiated NSG mice. Human lymphomyeloid reconstitution in the marrow of recipient mice was monitored by serial marrow aspiration at weeks 3, 8, 20 and 34 post-transplant (Fig. 1E). Between 8 and 34 weeks post-transplant, decoy-transduced cell populations showed increased marrow chimerism compared to vector-transduced cells (Fig. 1F). This growth advantage included both myeloid and lymphoid GFP+ populations (Fig. 1F). In contrast, the proportions of myeloid and lymphoid cells in the non-transduced (GFP−) human cell compartment in the same mice was similar between the two groups of transplants (Fig. 1G), consistent with miRNA depletion providing a cell intrinsic competitive advantage in vivo. These findings support a role for haploinsufficiency of miR-143 and miR-145 in the clonal expansion of malignant cells in del(5q) MDS.
miR-143 and miR-145 target IGF-1R in human hematopoietic progenitor cells
We have previously determined which pathways are targeted by both miRNAs using Ingenuity Pathway Analysis (IPA)33 on the non-redundant set of predicted targets of miR-143 and miR-14516. Here we used gene expression data from a previously-published cohort of MDS patients34, ranked samples on the basis of either the expression of miR-143 or the expression of miR-145, and performed Gene Set Enrichment Analysis (GSEA)28 to determine which BIOCARTA pathways were differentially upregulated with low expression of miR-143 and miR-145 (Supplemental Table 1). The IGF-1 signaling pathway was the only pathway common to both analyses (Fig. 2A). In addition, the IGF-1 receptor (IGF-1R) was predicted to be a target of both miR-143 and miR-145 based on the miRNA-target prediction algorithm, TargetScan (Supplemental Table 2).35
To validate the predicted interaction between miR-143/miR-145 and the 3’-UTR of IGF-1R mRNA, we performed luciferase assays using the wildtype (WT) or mutated IGF-1R 3’UTR containing mutations in the predicted binding sites, cloned downstream of a luciferase reporter. Overexpression of miR-143 or miR-145 markedly reduced luciferase activity in cells with the WT IGF-1R 3’UTR but not the mutated IGF-1R 3’UTR (Fig. 2B). Furthermore, an miRNA pulldown assay demonstrated that biotinylated miR-143 and miR-145 transfected into del(5q) MDS-L cells directly bound to IGF-1R mRNA, but not to insulin receptor (INSR) mRNA, confirming the specificity of the interaction (Supplemental Fig. 2, Fig. 2C). In addition, surface expression of IGF-1R was higher in del(5q) myeloid cell lines compared to myeloid cell lines diploid for chromosome 5q (Fig. 2D). To confirm that the miR-143/145-IGF-1R interaction is functionally relevant in primary hematopoietic cells, we transduced CD34+ HSPC with control or miRNA-decoy vectors. Surface IGF-1R expression was increased in miRNA-decoy cells, particularly in the CD34+ HSPC fraction compared to the differentiated CD34− cells (Fig. 2E). Conversely, overexpression of each of the miRNAs in the MDS-L del(5q) cell line led to decreased protein expression of IGF-1R (Fig. 2F). MDS-L cells overexpressing either miRNA had reduced colony formation in CFC assays, as well as reduced proliferation and a trend towards reduced viability (Supplemental Fig. 3). These data suggest that miR-143 and miR-145 regulate cell progenitor activity and cell proliferation in primitive human hematopoietic cells through the regulation of IGF-1R expression.
IGF-1R inhibition reduces progenitor activity of miR-143/miR-145-haplodeficient del(5q) cells
To determine which effects of miR-143/miR-145 deficiency are mediated by IGF-1R, we first asked whether specific inhibition of IGF-1R signaling in del(5q) MDS-L cells could abrogate cell expansion in vitro. To address this, we treated MDS-L cells with a small molecule IGF-1R inhibitor, BMS-536924, at various concentrations. At all tested doses of the inhibitor, there was a significant increase in cell death by 48 h of treatment (Fig. 3A). Proliferation was also significantly reduced, in a dose- and time-dependent manner (Fig. 3B). Cells treated with inhibitor appeared to mainly target cycling cells as cells were less likely to be in S-phase with more cells in G0/G1 of the cell cycle compared to controls (Supplemental Fig. 4). At the highest dose, cells were virtually eliminated following 3 days of treatment (Supplemental Fig. 5). Consistent with the above, MDS-L cells pre-treated with an intermediate concentration (1 µM) of IGF-1R inhibitor for 48 h had reduced output in CFC assays (Fig. 3C).
To extend this result to primary cells, CD34+ HSPC were transduced with either miRNA-decoy or control vector, treated with BMS-536924 (1 µM) and subjected to CFC assays. As observed previously, miRNA-decoy-transduced cells had increased colony output (Figs. 2E, 3D). However, IGF-1R inhibition reduced colony formation only in decoy-transduced cells, but had no effect on vector-transduced cells (Fig. 3D), potentially implicating a differential effect in del(5q) MDS cells compared to normal cells. Next, we used a complementary genetic approach, by targeting IGF-1R through lentiviral-delivered shRNA to reduce its expression in the del(5q) MDS-L cell line. Knockdown of IGF-1R resulted in reduced IGF-1R protein expression compared to the control shRNA-transduced cells (Fig. 3E). Cells transduced with shIGF-1R had reduced colony formation in clonogenic assays compared to controls (Fig. 3F). Together, these studies suggest that pharmacologic or genetic inhibition of IGF-1R signaling reduces cell expansion observed with loss of miR-143 and miR-145, consistent with the hypothesis that derepression of IGF-1R is functionally relevant in human del(5q) MDS.
To determine whether targeting IGF-1R would prolong survival in mice xenografted with human del(5q) MDS-L cells, we used a xenograft model in which NRG-3GS immunodeficient mice expressing three human growth factors were transplanted with MDS-L cells. To determine the specific effect of IGF-1R depletion in this model, we transplanted MDS-L cells transduced with shIGF-1R or shControl into NRG-3GS mice. Depletion of IGF-1R resulted in significantly prolonged survival of xenografted mice compared to the controls (median survival 57 days vs. 45 days, P = 0.0004) (Fig. 3H). To confirm that pharmacological inhibition of IGF-1R would have the same effect, we xenografted parental MDS-L cells into NRG-3GS mice, and injected engrafted mice intraperitoneally with BMS-536924 (40 mg/kg) 3 times weekly starting at 2 weeks post-transplant. As with IGF-1R knockdown, chemical inhibition of IGF-1R resulted in significant improvement in median survival, compared to controls (52 days vs 44 days, P = 0.0016) (Fig. 3I). Thus, either genetic knockdown or pharmacologic inhibition of IGF-1R inhibits the expansion of del(5q) MDS-L cells in a preclinical model of haploinsufficiency for miR-143 and miR-145.
IGF-1R inhibition bypasses LEN resistance in del(5q) MDS cells
Given that a significant proportion of del(5q) MDS patients do not respond to LEN or become resistant over time, we attempted to determine whether IGF-1R might be a relevant target in LEN-resistant del(5q) cells. To address this, we examined the efficacy of IGF-1R inhibitor, BMS-536924, along with LEN on four del(5q) and three non-del(5q) myeloid cell lines by measuring apoptosis and proliferation. Targeting IGF-1R reduced the viability of all del(5q) cell lines (Fig. 4A), including those known to be resistant to LEN (KG1, HL60).21 In contrast, survival of non-del(5q) cell lines was not affected by the addition of BMS-536924 (Fig. 4B).
We have previously shown that loss of function mutations in TP53 or RUNX1 cause LEN resistance.13 To further assess the efficacy of IGF-1R inhibition in the context of LEN resistance we used CRISPR/Cas9 MDS-L cells targeted for TP53 or RUNX1 that are resistant to LEN13. Despite LEN resistance, BMS-536924 caused significant cell death and reduced the clonogenic output of TP53- or RUNX1-targeted MDS-L cells (Fig. 4C, D). CSNK1A1 haploinsufficiency is required for LEN-induced cell death in del(5q) MDS cells11,36. As miR-143/miR-145 function is likely independent of CSNK1A1, we sought to determine if IGF-1R inhibition is effective via a pathway that is distinct from LEN by using the OCI-AML3 cell line that expresses diploid copies of CSNK1A1, miR-143 and miR-145. The OCI-AML3 cell line was transduced with the miR-143/miR-145 decoy construct or empty vector, or shRNAs against CSNK1A1 or a control shRNA, and colony output of these cells was measured in the presence of either LEN, BMS-536924, or vehicle (Supplemental Fig. 6). miR-143/miR-145 decoy-transduced cells had decreased clonogenic output with BMS-536924 treatment while LEN had no effect (Fig. 4E). In contrast, cells transduced with shCSNK1A1, had reduced colony number with LEN treatment but BMS-536924 had no effect (Fig. 4F). These results verify that IGF-1R inhibition acts via a mechanism distinct from LEN/CK1α to suppress del(5q) MDS cells.
Del(5q) MDS cells targeting TP53 or RUNX1 are sensitive to Abl or MAPK inhibitors
IGF-1R is frequently overexpressed in different cancers, but results of clinical trials with IGF-1 or IGF-1R have not shown convincing efficacy, resulting in the lack of further development of these compounds. 28,29,30. We thus sought to determine whether targeting IGF-1R-related pathways also exposed vulnerabilities in del(5q) MDS cells. Abl is involved in autoregulation of IGF-1R activity40,41. Interestingly, our GSEA analysis showed the BIOCARTA_GLEEVEC geneset, containing genes involved in Abl signaling, to be the top differentially upregulated pathway in MDS patients with low miR-143 or miR-145 (Supplemental Table 2, Fig. 5A). We also examined whether other pathways downstream of IGF-1R were differentially upregulated with miR-143 and miR-145 haploinsufficiency. We performed GSEA analysis on previously published RNA expression data of MDS samples34, and showed the BIOCARTA_MAPK and BIOCARTA_MTOR pathways to both be significantly activated in patients with low expression of either miR-143 or miR-145 (Fig. 5B, C). In contrast, the PI3K/Akt pathway was not activated with low expression of miR-143/miR-145 (Supplemental Fig. 7). These results suggest that the Abl, MAPK and mTOR signaling pathways may provide a potential therapeutic target for LEN-resistant del(5q) MDS.
Given that Abl, MAPK and MTOR signaling pathways were differentially upregulated with low expression of miR-143/miR-145, we sought to determine the dependency of del(5q) cells on these pathways. The Abl inhibitor, imatinib, reduced the viability and colony output of LEN-resistant MDS-L cells (Fig. 5D-F). Additionally, we assessed the effect of FDA approved inhibitors of MAPK and MTOR signaling pathways. Trametinib (MEK1/2 inhibitor), everolimus and temsirolimus (mTOR inhibitors) reduced the viability of del(5q) KG1a cells significantly (Fig. 5G), but only trametinib induced significant cell death in MDS-L cells (Fig. 5H). As trametinib was the only effective inhibitor in both del(5q) cell lines, we evaluated the sensitivity of LEN-resistant MDS-L cells to this inhibitor. Trametinib treatment resulted in reduced viability of LEN-resistant MDS-L cells, similar to wild-type control cells, in cells deficient for either TP53 or RUNX1 (Fig. 5I, J).