Amino Acid Metabolic Vulnerabilities in Acute and Chronic Myeloid Leukemias

Amino acid (AA) metabolism plays an important role in many cellular processes including energy production, immune function, and purine and pyrimidine synthesis. Cancer cells therefore require increased AA uptake and undergo metabolic reprogramming to satisfy the energy demand associated with their rapid proliferation. Like many other cancers, myeloid leukemias are vulnerable to specific therapeutic strategies targeting metabolic dependencies. Herein, our review provides a comprehensive overview and TCGA data analysis of biosynthetic enzymes required for non-essential AA synthesis and their dysregulation in myeloid leukemias. Furthermore, we discuss the role of the general control nonderepressible 2 (GCN2) and-mammalian target of rapamycin (mTOR) pathways of AA sensing on metabolic vulnerability and drug resistance.


INTRODUCTION
Myeloid leukemias are a group of disorders characterized by the presence of increased numbers of immature myeloid cells in the marrow and peripheral blood. These diseases are most common in adults with survival rates and first-line therapy options varying by age. In this review, we focus on the differences of non-essential amino acid (NEAA) metabolism in myeloid leukemias with an emphasis on the enzymes required for amino acid biosynthesis. While normal cells can synthesize NEAAs, cancers, in order to maintain their rapid growth and proliferation, can alter the expression of genes involved in amino acid biosynthesis and therefore lose the ability to synthesize specific NEAAs. This auxotrophy, or the inability of cancer cells to synthesize certain NEAA amino acids required for growth, leads to a dependency on exogenous sources of NEAAs, which can be pharmacologically targeted.
To comprehensively examine potential metabolic targets in myeloid leukemias within this review, we first introduce the disease and currently available pharmacotherapies to demonstrate the need for novel agents and to identify patients most likely to benefit from pharmacological agents targeting NEAA dependencies. We thereafter discuss the biosynthesis of NEAA, explore gene expression data analysis of potentially dysregulated pathways, and consider the supporting evidence for novel vulnerabilities in myeloid leukemias. Lastly, we detail mechanisms of cancer cell adaptation to nutrient stress and considerations for the optimal design of strategies targeting NEAA metabolic vulnerabilities.

MYELOID LEUKEMIAS AND NEED FOR NOVEL TREATMENT OPTIONS
Novel Drug Therapies Are Needed to Improve Outcomes in Myeloid Leukemia Patients Myeloid leukemias are classified as either chronic or acute based on their proliferation rate. Acute and chronic myeloid leukemias make up 14 and 33% of all estimated new 2020 leukemia cases in the United States, respectively (1).
Acute myeloid leukemia (AML) is the most common acute leukemia in adults. As of 2016, the 5-year overall survival (OS) of AML is approximately 24% (2), but the highest rate of AML deaths in the United States is among older patients (~90% for ages ≥ 65) 1 , underscoring the need for better therapy options. Due to the lower performance status of this demographic and inability to tolerate aggressive chemotherapy, these older patients have more limited curative pharmacotherapy options available.
Younger AML patients with good performance status can receive traditional intensive remission-induction chemotherapy with cytarabine, an anthracycline, and a FLT3 inhibitor depending on FLT3 mutation status. Alternatively, pharmacotherapy options for patients not eligible for intensive induction therapy include targeted approaches such as the hypomethylation agents decitabine or azacitidine and the BCL2 inhibitor venetoclax (3). Current approaches for post-induction therapy for low-risk patients generally include additional therapy using similar induction agents (such as high-dose cytarabine), and higher risk patients proceed to allogenic hematopoietic cell transplantation (HCT).
The main characteristic of chronic myeloid leukemia (CML) is the translocation between chromosomes 22 and 9 (i.e. Philadelphia chromosome) that creates a BCR-ABL fusion oncogene with constitutively activated tyrosine kinase activity (4,5). The cornerstone of CML treatment is the BCR-ABL tyrosine kinase inhibitor (6). However, there remains a need to develop novel therapies for CML patients with unfavorable prognosis, including those that are negative for the Philadelphia chromosome (7) and the 20-30% of CML patients who fail to achieve treatment milestones or develop TKI resistance. Altogether, current responses to standard AML and CML treatment regimens indicate that while many patients attain responses, there remains a need to develop therapies for those patients with poor prognosis.

Targeting AAs Dependencies in Myeloid Leukemias
Novel agents for the treatment of myeloid leukemias with unfavorable prognosis will most likely rely on certain cancer cell vulnerabilities to specifically inhibit their growth and limit treatment toxicity. Differences between the metabolic status of cancerous and normal cells are one of the primary hallmarks of cancer (8), and changes in AA metabolism, glucose utilization, lipid consumption, and ATP generation are adaptations cancer cells can acquire for meeting their increased energy demands (9)(10)(11)(12)(13). These metabolic changes can lead to an "addiction" of particular fuel sources and identification of these vulnerabilities has led to the development of several agents directed toward specific molecular targets (14).
AAs are the building blocks of protein and are necessary for the sustenance of both normal and leukemic cells. There are 20 AAs classified as essential, non-essential, and conditionally essential (15). Essential AAs must be obtained exogenously from diet, whereas the body possesses the metabolic pathways to catalyze the synthesis of NEAAs. Conditionally essential AAs, like arginine and glutamine, are those that can become essential during certain physiological conditions that limit their synthesis (16). AAs can serve as precursors to many biological compounds, including those involved nucleotide synthesis (17), redox balance (18), lipogenesis (19,20), and molecules that can fuel the TCA cycle (21). Furthermore, AAs can modulate the tumor microenvironment (22), global chromatin structures (23,24), and epigenetic factors (25)(26)(27)(28).
Herein, our discussion concerning myeloid leukemias and amino acid metabolism will focus on identifying potential vulnerabilities in AA biosynthesis that could be exploited for therapeutic purposes. Below, we provide a thorough discussion on NEAA biosynthesis for subsequently identifying dysregulations in potential targets that can lead to a dependency on extracellular NEAA sources for survival. Upon identification of an AA dependency, there are several targets within AA metabolic pathways that can be selectively targeted to exploit the vulnerability, including AA transporters, synthases, and transaminases that regulate biosynthesis (13,(29)(30)(31). In addition, enzymatic AA depletion therapy is an effective strategy for preferentially targeting cancer cells dependent on a specific AA for maintaining cell proliferation demands (13). metabolized by glutamate dehydrogenase (GLUD1/2) back to aketoglutarate (33).
Alanine, Aspartate, Glutamine, Proline, and Serine Can Be Synthesized From Glutamate As previously mentioned, glutamate serves as a substrate for several enzymes involved in the synthesis of several NEAAs. For alanine, the enzyme alanine aminotransferase (ALT/GPT) catalyzes the transamination of pyruvate, with glutamate as the amino donor, to synthesize alanine (34). Similar to ALT, aspartate aminotransferase (AST/GOT) transfers an amine group from glutamate to oxaloacetate of the TCA cycle producing aspartate and a-ketoglutarate (35).

Aspartate Is a Precursor for the Biosynthesis of Asparagine and Arginine
Similar to glutamate, other NEAAs, such as aspartate, serve as precursors for the biosynthesis of other AAs. Asparagine is synthesized from aspartate and glutamine via asparagine synthetase (ASNS, Figure 1), which transfers an amine group of glutamine to aspartate (39). Similarly, aspartate can be used for the synthesis of arginine through a multi-enzyme pathway in which the enzyme argininosuccinate synthase 1 (ASS1) initiates the process by catalyzing the formation of argininosuccinate from citrulline, which is an intermediate of urea synthesis, and aspartate ( Figure 1). Subsequent cleavage of argininosuccinate by argininosuccinate lyase (ASL) produces arginine and fumarate (40).

Serine Is a Precursor of Both Cysteine and Glycine
Serine can serve as a precursor for several AAs, including cysteine and glycine ( Figure 1). While cancers can dysregulate cysteine biosynthesis (41-43), typically cysteine is derived from extracellular sources (44)(45)(46). The metabolic pathway that leads to the generation of several sulfur metabolites, including cysteine, is called the transsulfuration pathway, where mammalian cells can only obtain cysteine through the reverse transsulfuration pathway (47)(48)(49). This pathway for generating cysteine in mammals involves the condensation of homocysteine (derived from methionine) with serine to generate cystathionine through the enzyme cystathionine b-synthase (CBS). Cystathionine glyase (CTH) then converts cystathionine to cysteine (50).
Serine can also be converted to glycine by serine hydroxymethyltransferase (SHMT), which transfers the bcarbon of serine to tetrahydrofolate to produce glycine and 5,10-methylene tetrahydrofolate (51). In humans, there are two SHMT genes: cytoplasmic SHMT1 and mitochondrial SHMT2 (52).

Tyrosine Is Synthesized by Phenylalanine Hydroxylases (PAH) From Phenylalanine
While the biosynthesis of other NEAA is linked to each other, mammals synthesize tyrosine from the essential amino acid phenylalanine. The enzyme phenylalanine hydroxylases (PAH) hydroxylates the aromatic side-chain of phenylalanine to generate tyrosine (53).

DYSREGULATION OF ENZYMES INVOLVED IN NEAA BIOSYNTHESIS IN MYELOID MALIGNANCIES Gene Expression Analysis of TCGA AML Samples
The Expression of Enzymes Involved in the Biosynthesis of Glutamate Is Dysregulated in AML A gene expression analysis of enzymes involved in NEAA biosynthesis was performed using the Gene Expression Profiling Interactive Analysis (GEPIA) web server (54,55). GEPIA uses RNA-Seq datasets based on the UCSC Xena project 2 that has a standard processing pipeline allowing users to compare gene and transcript expression from The Cancer Genome Atlas (TCGA) tumor samples to corresponding Genotype-Tissue Expression (GTEx) normal samples. GEPIA includes over 8,000 normal samples from TCGA and the GTEx projects, albeit from unrelated donors (56,57). For AML, GEPIA was used to screen for dysregulated enzymes by comparing transcript expression between AML (n = 173) and corresponding normal GTEx samples (n = 70).
Using GEPIA and TCGA AML samples, a gene expression analysis of enzymes involved in the synthesis of glutamate was performed. Consistent with the central role glutamate plays in the biosynthesis of several NEAA, we found that the expression of the glutaminase isoforms GLS and GLS2 in AML is upregulated relative to normal controls (P <0.01, Figures 2A, B). Similar to GLS, PRODH was significantly upregulated in AML samples (P <0.01, Figure 2C), whereas expression levels of P5CDH (i.e., ALDH4A1, data not shown) showed a similar but nonstatistically significant trend. Interestingly, the two isoforms of BCAT (cytosolic BCAT1 and mitochondrial BCAT2) were downregulated in AML TCGA samples (P <0.01, Figures 2D, E), indicating a potential strong dependence on glutamine and proline for sources of glutamate. For glutamate dehydrogenase, which is involved in glutamate metabolism, no association was identified for GLUD1, whereas the GLUD2 genes was significantly downregulated (P <0.01, Figure 2F), consistent with a glutamate dependency in AML.

The Expression of Enzymes Involved in the Biosynthesis of NEAA Requiring Glutamate as a Precursor Is Decreased in AML
Given that several enzymes of glutamate biosynthesis were dysregulated in AML, it was possible that the corresponding effect on glutamate levels in AML had an influence on other enzymes involved in the biosynthesis of NEAAs requiring glutamate as a precursor. An analysis of the 10 enzymes involved in the biosynthesis of alanine, aspartate, glutamine, proline, and serine identified significant downregulations in AST (GOT1 and GOT2), GLUL, PYCR1, PSAT1, and PHGDH (P <0.01, Figures 3A-F), but no statistical dysregulations in ALT (GPT and GPT2), P5CS (ALDH18A1), or PSPH (data not shown). Overall, there was a downregulation in 6 of 10 enzymes involved in the biosynthesis of these NEAAs, with alanine being the only NEAA with no association. Therefore, while there seems to be a dependence on glutamine and proline for sources of glutamate, many of the pathways involved in the biosynthesis of NEAA requiring glutamate as precursor were repressed.

The Expression of Enzymes Involved in the Biosynthesis of NEAA Requiring Aspartate or Serine as Precursors Is Decreased in AML
Aspartate is a precursor for the biosynthesis of asparagine and arginine, whereas serine is a precursor of cysteine and glycine. Of the seven collective enzymes involved in the biosynthesis of asparagine, arginine, cysteine, and glycine, we found that ASNS, ASS1, ASL, CTH, and SHMT2 were all statistically downregulated in AML samples ( Figures 4A-E), whereas CBS a n d S H M T 1 s h o w e d n o n -s t a t i s t i c a l l y s i g n i fi c a n t downregulations relative to controls (data not shown). Altogether, our analysis using the AML samples available through TCGA shows that there is a consistent decrease in the expression of enzymes involved in the biosynthesis of NEAAs that are interconnected through glutamate. In contrast, for tyrosine, whose biosynthesis is not linked to glutamate, there was no statistical difference in the expression of PAH between AML and control samples.
In summary, using GEPIA and TCGA AML samples we identified statistically significant dysregulations in enzymes involved in the biosynthesis of glutamate, aspartate, glutamine, proline, serine, asparagine, arginine, cysteine, and glycine. Therefore, future efforts should focus on validating these associations across various AML subtypes to best identify specific patient groups with potential NEAA metabolism vulnerabilities, which was not possible using GEPIA.

Gene Expression Analysis of NEAA Biosynthesis in CML
The Expression of Enzymes Involved in Regulating Glutamate, Aspartate, Alanine, Asparagine, and Cysteine Levels Is Upregulated in CML CML samples were not available through the TCGA database; rather, we used the largest publicly available data set disaggregated by CML phase with gene expression for all 26 genes involved in the biosynthesis of NEAA (Gene Expression Omnibus, GEO accession no. GSE47927) 3 . Unlike AML, we found few statistical associations for genes encoding enzymes involved in NEAA biosynthesis ( Figure 5). Of the 26 genes encoding enzymes involved in the biosynthesis of NEAAs used for our AML analysis, only five statistically significant dysregulated genes were identified in the combined CML cohort. We identified a significant downregulation in PRODH expression in combined and chronic phase CML samples CML samples ( Figure 5B, P Blast = 4.8 × 10 −2 ). While no consistent similarities between AML and CML were identified, dysregulation of genes involved in sustaining glutamate (i.e., PRODH), aspartate (i.e., AST1), alanine (i.e., GPT), asparagine (i.e., ASNS), and cysteine (i.e., CTH) levels suggest potential vulnerabilities involving these amino acids.

EVIDENCE SUPPORTING NEAA VULNERABILITIES IN AML AND CML
Various factors can lead to genetic alterations in cancers that result in the downregulation of an enzyme involved in the biosynthesis of a NEAA, thus rendering the cancer cells dependent on extracellular sources of the NEAA for proliferation and growth. Based on the expression changes observed in AML and CML samples described in the previous section, significant NEAA biosynthesis pathways identified will be reviewed to identify high-priority NEAA vulnerabilities in AML and CML.

Restriction of Glutamate or Glutamine Metabolism Via Inhibitors of Glutaminase Are Effective Against Myelodysplastic Syndrome
As previously mentioned, glutaminase (GLS) controls the formation of glutamate ( Figure 1) and is used for various biosynthetic purposes by cells, in addition to NEAA biosynthesis, including to generate TCA cycle intermediates, glutathione (GSH), NADPH, nucleotides, and fatty acids (131). Therefore, successful strategies targeting the role of glutamate in cancers have focused on restricting glutamine metabolism via glutaminase inhibition (132). Studies investigating the noncompetitive allosteric GLS1 inhibitor CB-839 have demonstrated that blocking glutamine metabolism has anticancer activity against triple-negative breast cancer, lung adenocarcinoma, chondrosarcoma, lymphomas, esophageal squamous cell carcinoma, and hepatocellular carcinoma (132)(133)(134)(135)(136)(137)(138). Currently, there are 12 ongoing clinical trials evaluating the CB-839 in patients with colorectal cancer (NCT02861300 and NCT03263429), myelodysplastic syndrome (NCT03047993), advanced stage non-small cell lung cancer (NCT04250545 and NCT03831932), diffuse astrocytoma (NCT03528642), ovarian cancer (NCT03944902), refractory multiple myeloma (NCT03798678), advanced or metastatic solid tumors ( N C T 0 3 9 6 5 8 4 5 ) , m e t a s t a t i c r e n a l c e l l c a r c i n o m a (NCT03428217), malignancies with NF1, KEAP1/NRF2, or STK11/LKB1 mutations (NCT03872427), and non-squamous non-small-cell lung cancer (NCT04265534). Additionally, recent interim results from a phase Ib clinical study of CB-839 in combination with azacitidine in patients with advanced myelodysplastic syndrome demonstrate the possible potential of targeting glutamate sources in leukemias, where the regimen was safely tolerated and 70% of MDS patients achieved a complete response to therapy (139). In addition to CB-839, compound 968 (C968) is another small molecule inhibitor of GLS (140) that has been demonstrated to have anticancer activity in ovarian, brain, pancreatic, and breast cancer (140)(141)(142).

Restriction of Other NEAAs Derived From Glutamate Decrease Myeloid Leukemia Proliferation
The biosynthesis of serine and its derivatives were also identified as possible targets in AML and CML ( Figures 3E, F and Figure 5E), and consistent with the gene expression analysis, substantial evidence supports that serine, cysteine, and glycine play important roles in cancers, including myeloid malignancies. One-carbon metabolism is required for the synthesis of proteins, lipids, and nucleic acids, with serine being the main source of one-carbon units for methylation reactions that occur through the generation of S-adenosylmethionine (SAM) (143). Glycine is also a major source of methyl groups for one-carbon pools and is required for the biosynthesis of GSH and purines (51), whereas cysteine contributes to redox control and ATP production as a carbon source for biomass and energy production (144). Furthermore, restriction of serine (145)(146)(147)(148), glycine (145,148,149), or cysteine (150) can decrease cancer cell proliferation. Strategies suppressing PHGDH, SHMT, or cysteine levels via the enzyme cyst(e)inase have demonstrated encouraging results. For AML, there is evidence that serine restriction or PHGDH inhibition can attenuate cell proliferation and that PHGDH expression is prognostic of AML overall survival (151). In addition, cysteine has been demonstrated to be critical for the survival of leukemic stem cells (LSCs) in AML patients, whereas glycine deprivation has been demonstrated to suppress AML cell proliferation (152). For CML, similar to our analysis, several studies have shown that levels of glutamate, serine, which is a precursor of cysteine, and alanine are increased by CML (153)(154)(155). Consistent with these NEAAs playing an important role in CML, restriction of serine (156,157) and its derivatives glycine (156), or cysteine (158,159) decreases CML proliferation, supporting that targeting these NEAA pathways may be a potential target for CML.
In addition to serine, glycine, and cysteine, several studies support that inhibiting proline biosynthesis in melanoma and breast cancers can impede cell growth (160)(161)(162). Furthermore, PYCR1 inhibitors with anticancer activity have been identified to explore the role of proline biosynthesis in AML (163,164). Nevertheless, limited information is available regarding the effect of inhibiting either enzymes involved in proline biosynthesis or directly restricting proline in AML. Rather AML studies available have indicated that proline uptake is elevated in LSCs isolated from de novo AML patients (165), and that the proline metabolism pathway is significantly impacted by differences in the oncogenic receptor tyrosine kinase FLT3 status of pediatric AML samples (166) or in AML cells overexpressing the proto-oncogene EVI1 (167).
Alanine is secreted by skeletal muscles and provides the carbon source for hepatic gluconeogenesis (168,169), yet limited information is available about the role of alanine metabolism in myeloid leukemias. Nevertheless, evidence supports that alanine provides a source of a-ketoglutarate for pancreatic and breast cancers that can be used to fuel the TCA cycle (170,171) or remodel the extracellular matrix for metastasis (172). Additionally, KRAS mutations have been demonstrated to induce the expression of ALT2/GPT2 and drive a-ketoglutarate production and cell growth (173). Few studies have investigated pharmacological strategies for targeting alanine metabolism, yet l-cycloserine has been identified as an inhibitor of ALT/GPT that can attenuate the in vitro and in vivo growth of LLC1 Lewis lung carcinoma cells (173). Taken together, there is substantial evidence that the NEAA biosynthesis pathways dysregulated in AML and CML are promising targets for myeloid leukemias.

GCN2 AND MTOR PATHWAYS CAN MODULATE RESISTANCE TO STRATEGIES TARGETING NEAA VULNERABILITIES
Strategies targeting NEAA biosynthesis vulnerabilities can be a promising approach for myeloid malignancies. However, cancer cell adaptation to nutrient stress can lead to drug resistance or treatment failure (174). This adaptation can be due to upregulation of amino acid transporters and/or cell changes leading to increased availability of NEAA pools (8,174,175). Therefore, understanding the mechanism by which myeloid leukemias adapt and sense NEAA restriction is important for optimizing therapeutic approaches targeting NEAA metabolic vulnerabilities. Two essential kinases that are involved in the adaptation to nutrient stress are mammalian target of rapamycin (mTOR) and general control nonderepressible-2 kinase (GCN2). Collectively, these pathways can decrease the cellular demand for amino acids while concurrently increasing their synthesis to overcome the nutrient stress and lead to resistance ( Figure 6).

Activation of the Amino Acid Response Pathway (AAR) Can Lead to Nutrient Stress Resistance
The GCN2 kinase activates the amino acid response (AAR) pathway by sensing amino acid insufficiency via direct binding to uncharged tRNAs that accumulate during nutrient stress (176). Upon activation via autophosphorylation, GCN2 phosphorylates its only known substrate, the a subunit of the translation eukaryotic initiation factor (eIF2) (177). Phosphorylation of eIF2a has two essential consequences that can affect cancer cell responses to amino acid deprivation ( Figure 6). First, it attenuates the general protein translation of most mRNAs by blocking the activity of the eIF2B guanylate exchange factor, thereby limiting the availability of ternary complex required for initiating translation (178)(179)(180). Second, it activates the AAR pathway by concomitantly increasing the translation of a subset of mRNAs with upstream ORFs that regulate translation of the downstream main ORF (179,(181)(182)(183). The AAR pathway leads to the upregulation of amino acid transporters, aminoacyl-tRNA synthetases, enzymes involved in the biosynthesis of amino acids, and proteins involved in autophagy (29,(184)(185)(186).
In contrast to GCN2, which directly senses amino acid depletion, mTORC1 indirectly senses amino acid sufficiency through mechanisms that were recently elucidated (187)(188)(189)(190). mTORC1 is a serine/threonine protein kinase that acts as an essential regulator of cell growth and metabolism in response to nutrient changes. SLC38A9, Sestrin1/2, and CASTOR1 have been identified as amino acid sensors upstream of mTORC1 that sense the availability of arginine and leucine (187)(188)(189)(190)(191). SLC38A9 stimulates mTORC1 activity through the regulation of Rag proteins, whereas Sestrin1/2 and CASTOR1 dissociate from GATOR2, which is a positive regulator of mTORC1, in the presence of leucine or arginine. Activated mTORC1 phosphorylates 4EBP1 and leads to the dissociation of 4EBP1 from EIF4E, enabling the formation of the translation initiation complex. In contrast, amino acid deficiency inactivates mTORC1 and suppresses 4EBP1 phosphorylation and protein synthesis (192)(193)(194)(195). In addition to translation, mTORC1 is a negative regulator of autophagy, which is an intracellular process that allows orderly degradation and recycling of cellular components, including amino acids. Autophagy is regulated by mTORC1 through its phosphorylation of UNC-51-like kinase 1 (ULK) (196)(197)(198)(199)(200). Therefore, mTORC1 signaling plays a role in AA deprivation adaptation by limiting protein translation and increasing amino acid pools through autophagy ( Figure 6).
Because mTORC1 and the GCN2/eIF2a pathway provide mechanisms of resistance to amino acid deprivation strategies, it is conceivable that blocking these adaptation pathways would further sensitize cancer cells to the nutrient stress. Consistent with that hypothesis, the development of a GCN2 inhibitor has been shown to enhance the antileukemic efficacy of Lasparaginase (208). Other studies have demonstrated similar adaptations and GCN2-dependent sensitizations (174,(209)(210)(211)(212)(213)(214)(215). While constitutive activation of mTORC1 can lead to resistance against targeted therapies (216), in the context of amino acid vulnerabilities, blocking the generation of amino acids from autophagic protein degradation can sensitize cancer cells to amino acid deprivation approaches (217). Consistent with this concept, blocking autophagy during asparagine (93,(218)(219)(220)(221)(222), arginine (106,108,(223)(224)(225)(226)(227)(228)(229), and glutamine (230) restriction leads to enhanced efficacy. FIGURE 6 | GCN2 and mTORC1 contribute to amino acid deprivation resistance. Upon amino acid deprivation, the accumulation of uncharged tRNA leads to the activation of the GCN2 kinase, which phosphorylates eIF2a. GCN2 activation can lead to a resistance to the amino acid restriction by inhibiting protein synthesis and by upregulating the translation of amino acid transporters, enzymes involved in amino acid biosynthesis, and proteins involved in autophagy. In contrast, amino acid suppression leads to inactivation of mTORC1, which induces autophagy and leads to a decrease in protein synthesis. Collectively, these mechanisms provide amino acid sources to maintain myeloid leukemia cell proliferation upon NEAA deprivation.

CONCLUSIONS AND FUTURE PERSPECTIVES
Cancers including myeloid leukemias use various energy sources to maintain their abnormal proliferation rate and therefore can become auxotrophic for particular NEAAs. Identifying these vulnerabilities in myeloid leukemias can therefore lead to new treatment options and improve survival rates. To date, Lasparaginase is the most successful clinical example of exploiting cancer nutrient dependencies. While outside of the scope of this review, there are several fuel sources aside from NEAA not considered here that contribute to metabolic vulnerabilities, including glucose, essential amino acids, fatty acids, lactate, and acetate (8)(9)(10)(11). Furthermore, additional factors, including the expression of amino acid transporters (231)(232)(233)(234)(235)(236) and the effect of stromal cells on the microenvironment, can contribute to cancer metabolism and proliferation (8). It is also clear that optimal treatment strategies will consider mechanisms of cell adaptation to nutrient stress. Our AML analysis identified that the biosynthesis of glutamate, aspartate, glutamine, proline, serine, asparagine, arginine, cysteine, and glycine are altered in TCGA samples, indicating that targeting multiple NEAAs may have a greater effect than a single agent approach. It is feasible that a combination approach to NEAA metabolic vulnerabilities may require lower drug exposures versus a single agent approach, and therefore can lead to better antileukemic efficacy while decreasing the risk of toxicities. Successful novel regimens with reduced risks of toxicities can lead to the development of new first-line therapy options for myeloid leukemia patients with limited treatment options.