IGF2BP1 expression is specific to ETV6-RUNX1 positive patient samples
We had earlier reported the overexpression of IGF2BP1 by RT-qPCR in the bone marrow of patients belonging to the ETV6-RUNX1 subtype in a cohort of Indian patients (n=114) [20]. Additional samples were added to the same cohort (Total n=167). IGF2BP1 expression was found to be significantly higher (>1000 fold) in ETV6-RUNX1 subtype, (median value = 2.34) compared to ETV6-RUNX1 negative samples (median value = 0.0022) (p <0.0001) (Fig. 1A).
The presence of the ETV6-RUNX1 translocation also correlated with lower White Blood Corpuscle (WBC) counts (Fig. 1B). For analyzing prednisolone response, the samples were stratified into ETV6-RUNX1 positive and negative groups. The negative group included patients from the other translocations group (BCR-ABL1, E2A-PBX1 and MLL) and the No Known Sentinel Translocation (NKST) group. We observed a higher proportion of Prednisolone Poor Responders (PPRs) among the ETV6-RUNX1 positive patients (38% vs 26%) which was statistically non-significant (Fig. 1C).
The presence of this translocation correlated with a better overall survival as well as lower percentage of deaths (p=0.034, NKST vs E6R1+ and p<0.0001, E6R1- vs E6R1+) (median survival 39 months for NKST and 21 months for E6R1- patients and >60 months for E6R1+ patients) (Fig. 1D) which corroborates with existing literature [21]. Despite showing a better event free survival (p=0.017, NKST vs E6R1+ and p<0.0001, E6R1- vs E6R1+) (median survival was 56 months for E6R1+ patients compared to 33 months for NKST and 19 months for E6R1- patients), around 23% of the ETV6-RUNX1 positive patients relapsed (Fig. 1E).
IGF2BP1 loss of function inhibits B-ALL cell proliferation and reverses prednisolone resistance
IGF2BP1 expression was observed to be significantly higher in Reh (ETV6-RUNX1 translocated) cell line when compared to other leukemic cell lines (Fig. 2A). To examine the effects of the loss of IGF2BP1, knockout was performed in Reh-Cas9 cell line using 3 different sgRNAs (sg1-3) which were designed and cloned as described previously [18] (Fig. 2B). Knockout was confirmed by Western Blotting with sg1 and sg2 causing complete knockout (Fig. 2C). IGF2BP1 knockout led to a significant decrease in cell proliferation as seen by the MTS assay.
IGF2BP1 knockout reversed prednisolone resistance and cells showed sensitivity to prednisolone with IC50 ~1 μM for both the guides used (Fig. 2D-E). Cells with the non-targeting gRNA (NT) were viable even at a 10 μM concentration of prednisolone, confirming resistance [22].
Integrated transcriptome and immunoprecipitation sequencing reveals overlap between IGF2BP1 and ETV6-RUNX1 regulated oncogenic pathways
To begin to unravel the mechanistic basis of our observations, the gene expression profile of IGF2BP1 KO Reh cells was analysed by RNA-Seq. 88 transcripts showed upregulation and 270 transcripts showed downregulation (at least 2-fold change, p<0.05) after IGF2BP1 knockout. Gene Set Enrichment Analysis (GSEA) [23] revealed a significant negative enrichment of the TNFα signalling via NFκB, hypoxia and cholesterol metabolism pathways (Fig. 2E) in KO cells as compared to non-targeting controls. A significant positive enrichment was found for pathways regulating cell cycle and proliferation including G2M checkpoint and the MYC targets (Fig. 2F).
To gain insights into the mRNA interactome of IGF2BP1, we performed RNA Immunoprecipitation followed by high throughput sequencing (RIP-Seq) (Fig. 3A). Application of SETEN [24], a GSEA-based tool for RNA-binding proteins, revealed enrichment of putative IGF2BP1 targets in numerous pro-oncogenic signaling pathways. The ENRICHR tool [25] classified IGF2BP1 target transcripts to be involved in B-cell and myeloid malignancies using the GWAS catalogue database. GO biological processes and molecular functions revealed enrichment for RNA processing, translation and RNA degradation pathways (Fig. 3B, Supplementary Fig.S1A-D).
An integrated analysis of the IGF2BP1-KO RNA-Seq and IGF2BP1 RIP-Seq in Reh cells was performed. Initially, we did not observe a significant enrichment between IGF2BP1 putative targets (RIP-Seq log2 fold>0, Fig. 3C) and the direction of regulation after IGF2BP1 deletion. However, integrating both datasets using GSEA (Fig. 3D) showed that the genes most downregulated after IGF2BP1 deletion were preferentially classified as IGF2BP1 putative targets (highest RIP-Seq enrichment, familywise error-rate p-value<0.003, Fig. 3D). This suggests that IGF2BP1 promotes RNA stability in Reh cells. Interestingly, genes up-regulated after IGF2BP1 KO were preferentially not enriched in the RIP data (familywise error-rate p-value<0.001, Fig. 3D) suggesting that these genes are indirectly regulated by IGF2BP1 (Fig.3C-D).
Combined pathway analysis of IGF2BP1 KO/RIP gene sets revealed that some of the significantly dysregulated pathways after the knockout also had many genes enriched in the RIP-Seq suggesting direct binding and regulation (Fig.3E, Supplementary Data). Examples include the TNFα/NFκB signaling and Hypoxia pathways. Oxidative phosphorylation was upregulated after knockout while cholesterol metabolism was downregulated. However, both were not RIP targets implying an indirect or downstream effect of IGF2BP1 knockout on those pathways. A deeper analysis of genes within these pathways showed that some of the genes in the metabolism of RNA, oxidative phosphorylation, MTORC1 and MAPK pathways showed enrichment in the RIP dataset (Fig.3F).
TNFα induced NFκB and PI3K-Akt signalling pathways are activated in ETV6-RUNX1 positive B-ALL by IGF2BP1
We then aimed to map IGF2BP1-induced gene expression changes to expression signatures of different patient subtypes. We applied GSEA to compare our IGF2BP1-KO data to differential gene expression between ETV6-RUNX1, E2A-PBX1 and MLL positive patients (from GSE65647, n=44) [16]. We also analyzed a dataset which compared ETV6-RUNX1 positive patients’ gene expression (n=4) with normal CD19 positive B-cells (n=2) [26]. There was a significant overlap between genes overexpressed and pathways enriched in ETV6-RUNX1 positive tumors and genes/pathways downregulated after deletion of IGF2BP1 (Fig.3G, Supplementary Fig.S2-S4).
We used BTYNB, a functional inhibitor of IGF2BP1 [27] to validate its role in regulating the non-canonical TNFα/NFκB pathway. Treatment of Reh with BTYNB led to a dose dependent decrease in cell proliferation. BTYNB treatment led to a decrease in both IGF2BP1 and ETV6-RUNX1 fusion transcript expression by qPCR. BTYNB had no effect on the RL and Jurkat cell lines which express minimal levels of IGF2BP1(Supplementary Fig. S5A-D).
Reh cells were transduced with NFκB-Luc-dTomato plasmid [28] which consists of the NFκB consensus sequence upstream of the Luciferase reporter gene. These cells showed an increase in luminescence after treatment with TNFα. Pre-treatment of these cells with BTYNB led to a loss of this luminescence induction in a dose dependent manner (Fig.4A-B).
We further validated genes from the TNFα/NFκB and PI3K-Akt pathways in our patient samples along with MACS enriched CD19+ B-cells as controls (Fig. 4C-D). Many of these genes including IL6ST, MDM2, CDK6 and NGFR showed significant upregulation in the ETV6-RUNX1 positive patients.
ETV6-RUNX1 fusion transcript is stabilized by IGF2BP1
ETV6-RUNX1 fusion transcript has been shown to affect the functions of wild type ETV6 as well as RUNX1 in a dominant negative fashion [29]. We sub cloned the ETV6-RUNX1 transcript from a pcDNA3.1-ETV6-RUNX1 plasmid (a kind gift from Dr Anthony Ford, ICR, London) into a bicistronic, retroviral, murine overexpression vector (Supplementary Fig. S6A). ETV6-RUNX1 virus particles were produced as described previously [19] and used to overexpress the same in a murine pre-B-ALL cell line, 7OZ/3 in three serial dilutions. There was a dose dependent increase in Igf2bp1 levels (Fig. 5B).
We performed ETV6-RUNX1 KO using guides targeting the ETV6-RUNX1 junction or the 5’ ETV6 region as reported previously (Supplementary Fig. S6B) [30] in Reh cell line. We observed a decrease in cell proliferation and sensitivity to prednisolone similar to the phenotype seen with IGF2BP1 KO (IC50=1.6/2/3.5 μM for sg1/2/3 respectively). This suggested the involvement of both the proteins in conferring a glucocorticoid-resistant phenotype to Reh cells. (Fig. 5B-C).
IGF2BP1-RIP followed by RT-PCR of immunoprecipitated RNA showed significant enrichment of the ETV6-RUNX1 junction, 5’-UTR of ETV6 and ACTB (positive control) in the RIP vs Input samples (Fig. 5D). In order to study the effect of IGF2BP1 on ETV6-RUNX1 stability, we cloned the junction region downstream of firefly luciferase (Supplementary Fig. S6C). Cotransfection of this vector with IGF2BP1 resulted in significantly increased luciferase activity (Fig. 5E). Complete knockout of IGF2BP1 (sg1 and sg2) significantly decreased ETV6-RUNX1 transcript levels in Reh cells (Fig. 5F). This data demonstrated the regulation of IGF2BP1 by ETV6-RUNX1 and its stabilization by the former.
IGF2BP1 and ETV6-RUNX1 synergize to cause clonal progenitor expansion in the murine bone marrow
To directly assess the synergism between ETV6-RUNX1 and IGF2BP1, we undertook an in-vivo experiment to examine the effects of enforced expression of both transgenes. Murine bone marrow transplant experiments were performed using retroviruses, synthesized and used as previously described [19] (SupplementaryFig. S7A). We cloned the human coding sequence of IGF2BP1 into MIG (MSCV-IRES-GFP), a murine stem cell virus–based (MSCV) retroviral vector and used it along with the ETV6-RUNX1 cloned in MICH (MSCV-IRES-mCherry). We confirmed the functionality of the vectors in expressing IGF2BP1 and ETV6-RUNX1 as well as the fluorescent markers (Supplementary Fig. S7B–D and data not shown).
All groups had similar levels of engraftment as seen from the CD45.1 positivity (data not shown). Clonal expansion was observed in the peripheral bleeds of mice expressing both ETV6-RUNX1 and IGF2BP1 (combination group) from weeks 4-16 with a constant increase. This was measured by the ratio of GFP+ mCherry+ cells to the total transfected cells (double positive ratio) (Supplementary Fig. S8A-B).
Complete blood counts (CBC) at week 16 showed a decrease in mature B-cell counts after ETV6-RUNX1 overexpression and in the combination. There was an increase in total WBC counts after IGF2BP1 overexpression. In the combination, the number of mature red blood cells, platelets and neutrophils were significantly lower with an increase in immature reticulocyte counts (SupplementaryFig. S9A-H).
To further characterize these hematopoietic changes, mice were euthanized after 16 weeks and hematopoietic organs collected for analysis. The bone marrow counts were significantly higher in the combination group with significant clonal expansion. Analysis of the progenitors revealed significant increase in the Lin- population, hematopoietic stem cells (HSCs), Lymphomyeloid Primed Progenitors (LMPP) and common lymphoid progenitors (CLPs) (percentages and absolute counts) in the combination group. The ETV6-RUNX1 group also showed a small but significant increase in progenitor output (Fig. 6A-6G). Hardy fraction analysis of the bone marrow in the combination showed an accumulation of the pre-B cell population implying a B-cell developmental block. (Supplementary Fig.S10).
Previous studies have identified the Lin- c-Kit+ or the LSK (Lin- c-Kit+Sca1+) populations as leukemia initiating cells in different mouse models [31]. The bone marrow of the combination group showed a significant increase in both populations (Fig. 7A) with strong Ki67 positivity of the Lin- progenitor population implying an increased proliferation rate (Fig. 7B). The myeloid progenitors (common myeloid progenitors (CMP), granulocyte-monocyte progenitor (GMP) and megakaryocyte-erythroid progenitor (MEP)) were also significantly increased in the combination (Supplementary Fig.S11).
The progenitor populations from the bone marrow were analyzed for fluorescent marker expression (double positive (DP): mCherry+ GFP+ and double negative (DN): mCherry-GFP-). In the combination group, there was an increase in the Lin- population in both the DN and DP fractions implying both a cell extrinsic and intrinsic mechanism for progenitor expansion. Interestingly, the HSCs and LMPPs were only increased in the DP fraction indicating a purely cell intrinsic mechanism (Supplementary Fig.S12).
Histopathological examination of the bone marrow of mice expressing both ETV6-RUNX1 and IGF2BP1 showed marked hypercellularity along with a loss of normal architecture characterized by compressed vascular spaces, reduced fat globules and megakaryocytes (Fig. 7C). Spleens were significantly enlarged with clonal expansion (Supplementary Fig.S13A-C). The spleen was populated by immature cells including HSCs implying extramedullary hematopoiesis (Supplementary Fig.S13D). Histopathological analysis revealed loss of architecture with red pulp expansion and smaller germinal centers in the combination group (Supplementary Fig.S13E).
RT-qPCR from the bone marrow of the mice showed a significant increase in endogenous Igf2bp1 levels after ETV6-RUNX1 overexpression, validating our in-vitro finding. The levels of ETV6-RUNX1 and IGF2BP1 were highest in the combination group (Fig. 7D).