No keto for AML stem cells!

In this issue of Blood, Yusuf et al report a new metabolic dependency of leukemic stem/progenitor cells (LSPCs) on lipid metabolism-associated detoxifying enzyme aldehyde dehydrogenase 3a2 (ALDH3A2) (see figure), which could be exploited to enhance therapeutic eradication of aberrant stem cells, which are the cause of acute myeloid leukemia (AML).¹ 

The hallmark of AML is abnormal proliferation of nonfunctional, differentiation-impaired myeloid blast cells originating from a rare population of therapy-refractory LSPCs.²  LSPCs develop after the acquisition of multiple genetic and epigenetic alterations in hematopoietic stem cells (HSCs).³  Systematic molecular profiling of AML genomes has uncovered the disease’s complex genomic landscape and its substantial molecular heterogeneity. This profiling has also demonstrated the coexistence of multiple preleukemic and leukemic clones with linear and nonlinear clonal evolution trajectories during disease evolution and progression.^3,4 

Via various molecular pathway alterations, leukemogenic lesions dictate how aberrant cells proliferate, differentiate, and interact with their microenvironment. These aberrations also permit evasion of commonly encountered apoptotic stimuli to allow outgrowth and annihilation of normal hematopoiesis. Upregulation of metabolic pathways fueling anabolic cell growth is a functional prerequisite, and importantly, a converging feature of various oncogenic drivers during all disease stages (reviewed by Rashkovan and Ferrando).⁵  Although a growing body of studies has uncovered unique metabolic dependencies and vulnerabilities of malignant cells across several cancers, including hematologic malignancies, the genetic and clonal heterogeneity of AML has hampered a comprehensive understanding of how metabolic plasticity impacts leukemia initiation and progression, especially at the LSPC level. This gap in our knowledge has largely precluded the successful development of metabolism targeting therapies against AML, with perhaps 1 exception: small-molecule inhibitors of the oncometabolite-generating, mutant isocitrate dehydrogenase enzymes IDH1 and IDH2.⁵ 

Lipid metabolism has emerged as a central molecular pathway necessary for both healthy and leukemic (stem) cell maintenance. It also helps with the adaptation to specialized microenvironments, particularly during antileukemic therapy.⁵  Lipids, essential structural building blocks of membranes, are also the main source for the production of adenosine triphosphate, reduced nicotinamide adenine dinucleotide phosphate (NADPH), and specialized lipid-based signaling molecules. Noncancerous cells can satisfy their lipid requirement through uptake of lipoproteins and free fatty acids (FAs); in contrast, malignant cells additionally require the de novo synthesis of FA, which also generates highly reactive fatty aldehydes, metabolic intermediates that act as signaling molecules and cytotoxic products in cells.⁵ 

Microsomal homodimer ALDH3A2 (or fatty aldehyde dehydrogenase) belongs to the superfamily of detoxifying ALDHs, whose 19 members are redundantly found in all subcellular compartments of cells. ALDH3A2 catalyzes the nicotinamide adenine dinucleotide (NAD)⁺-dependent oxidation of long-chain aliphatic aldehydes produced at cellular membranes as a consequence of lipid metabolism and oxidative stress–associated lipid peroxidation.⁶  Loss of ALDH3A2 function is not compensated by other ALDH enzymes⁷  and known to cause the rare autosomal recessive neurocutaneous Sjögren-Larsson syndrome (MIM 270200). ALDH3A2 function in HSCs or LSPCs had been unexplored until now.

Yusuf and colleagues used a comprehensive RNA interference screen of literature-curated canonical rate-limiting metabolic effectors and metabolic regulators found overexpressed in LSPCs vs normal counterparts to identify metabolic enzymes conferring LSPC-specific dependencies. The authors used a well-established and characterized mosaic MLL fusion oncogene-driven AML mouse model allowing for the isolation of leukemic stem and progenitor cells (leukemia granulocytic-monocytic progenitors) along with their normal counterpart (granulocytic-monocytic progenitor).⁸  This screen identified 6 high-confidence targets, among them Aldh3a2. Validation experiments employed in vivo mouse models, short-term cultures of human AML patient-derived cell lines, and primary cells to assess the consequences of Aldh3a2 ablation in healthy hematopoiesis. These experiments uncovered a specific dependency of LSPCs on ALDH3A2, independent of MLL fusion oncogene presence, while leaving nonleukemic cells and hematopoiesis functionally unaffected. This finding provides a quite remarkable proof of concept for the use of AML mouse model–derived LSPCs for the identification of novel metabolic vulnerabilities relevant across several human AML (stem) cell populations, which has posed a considerable experimental challenge in the past.

Characterizing the mechanics of Aldh3a2 dependency in LSPCs, Yusuf and team found an increased accumulation of fatty alcohols, precursors of reactive fatty aldehydes, in Aldh3a2-deficient LSPCs (vs their Aldh proficient leukemic counterparts). Enhanced reactive Aldh3a2 lipid substrate levels occurred alongside with the accumulation of cellular reactive oxygen species and increased oxidative damage of DNA and proteins, which the authors attributed to a failure to produce sufficient amounts of antioxidant NADPH via shunting glucose through the pentaphosphate pathway (PPP). PPP is known to drive glucose addiction in and survival of AML blast cells.⁵  Albeit the authors did not provide further mechanistic insight into this observation, it is conceivable that Aldh3a2-dependent metabolites play a role in modulating the activity of glucose-6-phosphate dehydrogenase, the PPP rate-liming enzyme. It is also still unclear what constitutes the molecular effects of Aldh3a2 loss in healthy HSCs.

Importantly, ablation of Aldh3a2 in LSPCs reduced 18 carbon FAs (including arachidonic acid precursor linoleic acid) and rendered arachidonic acid undetectable, along with a reduction in their respective lipid products. The authors predicted altered lipid composition to be a result of reduced de novo FA synthesis and compensatory lipid peroxidation in the absence of Aldh3a2, which sensitized LSPCs to nonapoptotic, iron-dependent cell death via ferroptosis. Indeed, ferroptosis inhibition functionally rescued Aldh3A2 loss. Conversely, genetic and pharmacologic inhibition of ferroptosis inhibitory glutathione peroxidase 4 (GPX4) and cytotoxic therapy triggered increased elimination of Aldh3a2-deficient LSPCs and extended AML-free survival in vivo compared with aberrant stem cells expressing the detoxifying enzyme. This finding provides further evidence that AML relies on increased metabolic plasticity, particularly enhanced FA metabolism, for desensitizing leukemic clones to cell death⁹ ; it also underscores that distinct metabolic adaptations are likely at play in LSPCs and AML blast cells as suggested by Jordan and colleagues.¹⁰ 

Future studies are needed to continue identifying LSPC-specific metabolic adaptations and vulnerabilities, particularly in the context of therapeutic interventions for AML. It will also be of critical importance to gain insights into the role of metabolic dependencies in pre-LSPCs, as they are not only the cellular source of primary disease initiation, but they provide a pool of highly transformation-susceptible cells during antileukemic therapy.^3,4