A Genome Wide shRNA Screen In Primary Human AML Cells Identifies ROCK1 As a Novel Therapeutic Target

In spite of recent advances, the prognosis especially of elderly AML patients remains unsatisfactory with survival rates of less than 10 % at 10 years. Genome-wide RNA-interference screens systematically interrogating the specific vulnerabilities of leukemic cells could be a valuable tool to identify novel therapeutic targets in this patient population. So far, such screens have only been done in immortalized cell lines and / or at sub-genome scale, which limits their transferability to individual patients.

Therefore, we set out to establish an unbiased genome-wide pooled shRNA screen in primary human AML cells to prove the feasibility and test the possible clinical implications of such an approach.

Lentiviral transduction of primary leukemic blasts from a 67-year old patient with AML FAB M1 with a pooled shRNA library (Mission TRC shRNA library SP1, Sigma) according to a specifically optimized protocol resulted in a transduction rate of 25 %, thus rendering multiple integrants unlikely. An aliquot of the cells was separated for DNA extraction directly after removal of viral supernatant (day 0) and after 9 days of suspension culture (day 9). ShRNA barcodes integrated into the genome of the host cells were read out using PCR-coupled next-generation sequencing (HiSeq 2000, Illumina). Of 7709 shRNA contained in the library, 6626 were recovered with at least 10 reads in the day 0 sample. After 9 days of culture, 25 shRNA targeting a total of 12 genes were identified as potentially lethal to the patient's AML-cells (Table 1). All of these shRNA were subjected to single-shRNA transduction experiments using leukemic cells from the same donor. In fact, 18 of 25 shRNA were validated with respect to viability. Knockdown specificity was documented for all validated shRNA by qPCR. For further analyses we focused only on those 7 genes in which more than 50% of the shRNA identified in the pooled screen could be validated (Table 1). These genes were assessed for druggability using publicly available databases. For exploration of the potential therapeutic implications of our screen we chose ROCK1 as a potential target, because Fasudil, a specific ROCK1 inhibitor, has already been licensed for the treatment of pulmonary hypertension in humans.

Knockdown of ROCK1 in primary leukemic blasts led to rapid cell-cycle arrest and cell-death. Treatment with Fasudil proved to be equally effective in killing leukemic cells. Compared to primary leukemic cells from the original as well as from other AML patients, Fasudil seemed to be less toxic to hematopoietic cells derived from healthy volunteer donors. RNA-sequencing revealed that in comparison to the healthy controls none of the studied AML patients demonstrated a significant overexpression of ROCK1. Moreover there was no indication for a functional ROCK1 mutation in the analyzed AML samples. Feeder based long-term culture initiating cell (LTC-IC) assays further suggested that Fasudil had a significant negative effect on the self-renewal capacity of primary human leukemic stem/progenitor cells (Figure 1). Studies in xenograft-models to assess the stem cell toxicity of ROCK1 inhibition in more detail are currently ongoing.

Figure.

View largeDownload slide

Figure.

View largeDownload slide

Close modal

Taken together our results show that pooled shRNA screens in primary patient-derived leukemic cells are feasible and able to pinpoint novel therapeutic targets, which might be missed in mutation- or overexpression-based approaches. Further optimization of transduction and screening protocols might enable such screens to assist physicians in the selection of optimal therapeutic strategies especially in poor risk AML.