Acute myeloid leukemia (AML) is an aggressive hematopoietic neoplasm with five-year overall survival rates <30% on standard of care therapy. Relapse is common and thought to originate from AML stem cells that survive in the protective bone marrow (BM) microenvironment. Next generation sequencing (NGS) has revealed the somatic mutation spectrum of AML and the approval of midostaurin for FLT3 ITD- and enasidenib for IDH2 -mutated AML is evidence that the molecular knowledge is being translated clinically.

To identify survival-critical AML genes irrespective of mutational status, we performed an shRNA screen on patient samples (N=12) using a barcoded lentiviral shRNA library. Transduced cells were cultured for 9 days on HS-5 stromal cells (to mimic the BM microenvironment) and subjected to NGS for barcode quantification. Sirtuin 5 (SIRT5), the only known enzyme with lysine desuccinylase, demalonylase, or deglutarylase activity, was amongst the top candidates. SIRT5 is implicated in regulating metabolic pathways, including energy metabolism.

We first confirmed growth inhibition upon SIRT5 knockdown (KD) in cryopreserved AML cells from the original screen. Next, we transduced a panel of 25 AML cell lines with doxycycline (dox) - inducible shSIRT5 (dox-shSIRT5). SIRT5 KD strongly inhibited growth, reduced colony formation and increased apoptosis in 18/25 lines (SIRT5-dependent cell lines). Seven cell lines showed <20% colony reduction despite >80% reduction of SIRT5 protein and were considered SIRT5-independent. SIRT5 dependence was neither correlated with specific mutations nor basal SIRT5 expression. To control for off-target effects we tested three unique shSIRT5s targeting different sequences in three SIRT5-dependent and three SIRT5-independent cell lines, and confirmed the differential sensitivity for all lines and all constructs. We next transduced CD34+ cells from AML patients (n=15) or cord blood (CB, n=5) with dox-shSIRT5 and plated cells in colony assays. SIRT5 KD (~40% in AML and CB) reduced AML colonies by ~50%, with no effect on CB. Colony formation by SIRT5 null mouse bone marrow infected with FLT3-ITD or MLL-AF9 retrovirus was reduced by 50-60% compared to wild type controls.

Metabolic profiling showed that SIRT5 KD induced a profound reduction of oxidative phosphorylation and glycolysis in SIRT5-dependent, but not SIRT5-independent cell lines. Mitochondrial reactive oxygen species (ROS) in SIRT5-dependent cell lines were strongly increased upon SIRT5 KD, and this increase preceded apoptosis. Ectopic expression of superoxide dismutase 2 (SOD2, mitochondrial) abrogated the increase in ROS and rescued cells from apoptosis. Metabolomics and RNAseq suggest that SIRT5-dependent cells exhibit profound metabolic disruption, including reduced TCA cycle activity, and recurrent transcriptional changes, while SIRT5 KD in SIRT5-independent cells is inconsequential.

We next injected mice with CMK-1 cells (SIRT5-dependent) expressing dox-shSIRT5 and luciferase. Mice were randomized to dox-supplemented (dox-water) or control water 18 hours post injection, and monitored by luminescence imaging. Control mice showed abundant luminescence at week 3, and died before week 5, while mice receiving dox-water survived throughout the 13-week experiment without evidence of leukemia. When mice with established leukemia were switched to dox-water following week 3, luminescence rapidly decreased and the mice survived without evidence of leukemia until termination of the experiment (week 13). Dox-water had no effect on mice engrafted with OCI-AML3 cells (SIRT5-independent), all of whom died with extensive leukemic involvement despite downregulation of SIRT5 in leukemia cells. We are currently testing the requirement of SIRT5 in MLL-AF9-mediated AML using a retroviral BM transplant mouse model. Preliminary data suggest that absence of SIRT5 may prolong survival.

Our data suggest that SIRT5 KD preferentially targets AML cells over normal cells. As SIRT5 null mice are viable with only minor metabolic abnormalities at steady state, these data implicate SIRT5 as a potential therapy target in AML and support the development of clinical SIRT5 inhibitors.


Deininger:Pfizer: Consultancy, Membership on an entity's Board of Directors or advisory committees; Blueprint: Consultancy.

Author notes


Asterisk with author names denotes non-ASH members.