Identification and Validation of Novel Hematopoietic Stem Cell Regulatory microRNAs

Within this study we have developed a novel approach to identify hematopoietic regulatory genes by re-analyzing genome-wide viral integration site (IS) data from a clinical gene therapy trial of 10 Wiskott-Aldrich syndrome patients. We observed a statistically enriched contribution of individual cell clones marked by IS to benign human long-term hematopoiesis within a dataset comprising 130,699 unique IS in the proximity of 12,437 genes. Furthermore, when we assessed the genomic distribution of the IS, we found that clusters of IS (CIS) are significantly over-represented at certain coding and non-coding genetic loci, indicating a viral marking of actively transcribed genes or a proliferation advantage of specific clones due to viral induced deregulation of adjacent genes. Notably, 42 % of the 50 most enriched regions encode for human HSC regulators such as MECOM, LMO2 and HMGA2. Based on a stepwise and stringent selection process evaluating IS number, location and CIS size, we chose the top 12 ranked miRNA genes for further analysis. Of those, three miRNAs (miR-132, miR-142, miR-146a) have described roles in hematopoiesis, strongly validating our selection strategy.

In order to functionally investigate the influence of selected candidate genes on hematopoiesis we developed a lentiviral (LV) based pooled overexpression (OE) approach using uniquely barcoded (BC) vectors encoding for the top 8 candidate miRNA HSC regulators in lin- sca1+ ckit+ (LSK) cells. The LV based library consists of two vectors per gene including corresponding GFP control vectors, and harbor a unique BC cassette serving as a molecular tag to allow its unbiased amplification and high-throughput sequencing of up to 96 indexed samples simultaneously.

To assess the influence of the selected miRNA candidates on hematopoietic differentiation in vitro, we performed serial CFU assays using BC library transduced LSK-SLAM cells as a starting population. Low infection rates were used to ensure one vector copy per cell. We detected a 1.58±0.03 [SEM] fold increase (p=0.003, n=5) in myeloid colony forming potential of miRNA library OE cells compared to GFP control cells without a shift in colony phenotypes. After three rounds of serial replating, colonies expressing 2/8 candidates were significantly overrepresented (1.6±0.23 to 2.0±0.20 fold; p=0.049 and p=0.005; n =5) while BC vectors encoding for two other candidate miRNAs were significantly reduced (0.4±0.11 to 0.5±0.14 fold, p=0.006 and p=0.027) compared to GFP control (fc=1.0±0.12).

To address the influence of candidate miRNA OE on multi-lineage engraftment and self-renewal in vivo, we serially transplanted miRNA or GFP control library transduced LSK cells in a cohort of 18 recipient mice. We revealed efficient candidate BC representation in all mice with a median read count of 7.1E4 (min=2451, max=8.4E5) 20 weeks post-transplant. In line with our results in vitro, we observed a 1.23±0.23 (p=0.044) fold enrichment of the same two miRNA candidates and a depletion of the two other miRNA genes (0.63±0.09 to 0.68±0.13fold; p=0.002 and p=0.033) in PB samples already eight weeks after transplantation. Interestingly, BC sequencing of lineage sorted fractions suggests that one of the overrepresented candidates has a strong influence on T-cell differentiation (fc=1.5±0.2, p=0.029, 20 weeks post TP) whereas the other candidate impacts on myelopoiesis (fc=1.6±0.35, p=0.114).

In summary, our systematic stepwise approach combining the comprehensive analysis of the complete integration site repertoire in a clinical gene therapy study with subsequent functional validation using a pooled lentiviral overexpression approach is a useful strategy to identify and characterize novel HSC regulators.