A feedforward loop between ACLY and MYC supports T-ALL progression in vivo

Key Points

• •MYC transcription is supported by ACLY-generated acetyl-CoA, and ACLY is a direct target of MYC, establishing a feedforward loop in T-ALL.
• •Secondary Acly loss or ACLY inhibition results in significant therapeutic effects in vivo, unveiling ACLY as a novel target in T-ALL.

Visual Abstract

Abstract

T-cell acute lymphoblastic leukemia (T-ALL) is a hematological malignancy in need of novel therapeutic approaches. Here, we identify ATP-citrate lyase (ACLY) as overexpressed in human T-ALL and as a promising therapeutic target for its treatment. To test the effects of ACLY in leukemia progression, we developed an isogenic model of NOTCH1-induced Acly conditional knockout leukemia. Importantly, we observed intrinsic antileukemic effects upon loss of ACLY, which further synergized with NOTCH1 inhibition in vivo. Metabolomic profiling upon ACLY loss revealed a metabolic crisis with reduced acetyl-coenzyme A (acetyl-CoA) levels and decreased oxygen consumption rate. Gene expression profiling analyses showed that the transcriptional signature of ACLY loss very significantly correlates with the signature of MYC loss in vivo. Mechanistically, the decrease in acetyl-CoA led to reduced H3K27ac levels in Myc, resulting in transcriptional downregulation of Myc and drastically reduced MYC protein levels. Moreover, pharmacological inhibition of ACLY led to reduced MYC levels and antileukemic effects in human T-ALL cell lines and patient-derived xenografts. Interestingly, our analyses also revealed a reciprocal relationship whereby ACLY itself is a direct transcriptional target of MYC, thus establishing a feedforward loop that is important for leukemia progression. Overall, our results identified a relevant ACLY-MYC axis and unveiled ACLY as a novel promising target for T-ALL treatment.

Introduction

T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive hematological disease of immature T-cell progenitor cells. Although most patients nowadays get cured through intensive chemotherapy regimens, 20% to 50% of patients still have relapse and a dismal prognosis.¹ Targeting different metabolic routes in patients has resulted in notable antileukemic effects, as exemplified by the clinical use of methotrexate, 6-mercaptopurine, or L-asparaginase.² Moreover, we recently identified the antileukemic effects of either serine hydroxymethyltransferase inhibition³ or mitochondrial uncoupling,⁴ highlighting the value of further exploring metabolic vulnerabilities in leukemia. We previously demonstrated that the metabolic consequences of NOTCH1 inhibition in T-ALL are critical for its antileukemic effects.⁵ Because NOTCH1-activating mutations occur in >60% of patients,⁶ we hypothesized that thorough analyses of the metabolic enzymes being downregulated on NOTCH1 inhibition might yield novel therapeutic targets for T-ALL treatment and identified the ATP-citrate lyase (ACLY), a critical metabolic and epigenetic regulator,⁷ as significantly downregulated upon NOTCH1 inhibition, suggesting a relevant role for ACLY in NOTCH1-driven leukemia. However, nothing is known on the role of ACLY in T-cell development or T-ALL.

Methods

In vivo models of NOTCH1-driven mouse T-ALLs

Animals were maintained in ventilated caging in specific pathogen-free facilities at New Brunswick Rutgers Biomedical and Health Sciences campus. All animal housing, handling, and procedures involving mice were approved by Rutgers Institutional Animal Care and Use Committee, in accordance with all relevant ethical regulations.

To generate Acly conditional inducible knockout leukemias, we used bone marrow progenitor cells (fresh cells provided by K.E.W.) from Acly^flox/flox mice (JAX #043555) harboring a tamoxifen-inducible Cre recombinase from the ubiquitin C locus (JAX #007001). Then, we performed retroviral transduction of lineage-negative–enriched cells with a retrovirus encoding a ΔE-NOTCH1-GFP oncogenic activated form of NOTCH1 with concomitant expression of GFP, as previously described.⁵ Cells were then transplanted through retro-orbital injection into lethally irradiated (7.5 Gy) recipient mice.

For leukemia-progression studies, already generated ΔE-NOTCH1-GFP–induced Acly^flox/flox-Cre-ERT2/+ leukemias (1 × 10⁶ leukemia cells) were transplanted from primary recipients to sublethally irradiated (4.5 Gy) 6- to 8-week-old secondary recipient C57BL/6 mice (Taconic Farms) by retro-orbital injection. After 48 hours from leukemic cell transplantation, recipient mice were treated with vehicle only (corn oil; Sigma, C8267) or tamoxifen (Sigma, T5648, 3 mg per mouse in corn oil), to induce isogenic loss of Acly in established leukemias. Subsequently, 5 days posttransplantation, mice in each arm were divided randomly into 2 different groups: control groups were subsequently treated with vehicle only (2.3% DMSO in 0.5% methylcellulose, 0.1% Tween 80) or with DBZ (5 mg/kg in vehicle solution; Syncom, 29762) on a 4-day-on and 3-day-off schedule, as previously described.⁵ For BMS-303141 survival experiments using a human T-ALL patient-derived xenograft (PDX), 1 × 10⁶ cells were injected into female 10-week-old NRG mice (Jackson Laboratory, stock #007799). Three days after transplantation, mice were divided randomly into 2 different groups and treated daily through gavage with vehicle (3% DMSO in 0.5% sodium carboxymethyl cellulose) or with BMS-303141 (50 mg/kg in vehicle; Selleckchem, S0277). Investigators were not blinded to group allocation. Animals were monitored for signs of distress or motor function at least twice daily until they were terminally ill, whereupon they were euthanized.

For Acly acute deletion analyses in mouse primary leukemias, we transplanted lymphoblasts from spleens of Acly conditional knockout NOTCH1-induced T-ALL–bearing mice into a secondary cohort of recipient mice, as before. We monitored mice until they presented clear leukemic signs with >60% GFP-positive leukemic cells in the peripheral blood; then, mice were treated with vehicle or tamoxifen, and 48 or 72 hours after treatment, mice were euthanized, and spleen samples were collected for further analyses.

For T-cell development and evaluation of bone marrow subsets, we bred Acly conditional knockout mice with Vav-iCre mice (JAX #008610) to obtain the different genotypes of interest. T-cell development was analyzed in 6- to 8-week-old mice, and bone marrow populations were analyzed in 8- to 10-week-old mice.

Luciferase reporter assays

Reporter assays were performed using a pGL4.10 Vector (Promega, Madison, WI, E6651) luciferase construct alone or coupled with the human ACLY promoter sequence (hg38; chr17: 41918547-41919547; DNA sequence was synthesized by GENEWIZ), cloned in the forward orientation, and following our previously described protocol.⁸ Constructs were transfected into 293T together with a pMSCV-IRES-mCherry control vector or a pMSCV-cMYC-IRES-mCherry, and a plasmid driving the expression of the Renilla luciferase gene (pCMV-Renilla) was used as an internal control. Then, 48 hours after transfection, cells were treated with JQ1 (500 nM; MedChemExpress, HY-13030) for 24 hours, as indicated. We measured luciferase activity 72 hours after electroporation with the Dual-Luciferase Reporter Assay kit (Promega, E1980).

Flow cytometry analysis

To analyze leukemic spleen samples, single-cell suspensions were prepared by disrupting spleens through a 70-μm filter. The red cells were lysed by incubation with ammonium–chloride–potassium (ACK) lysing buffer (155 mM NH₄Cl, 12 mM KHCO₃, and 0.1 mM EDTA) for 5 minutes on ice. Apoptotic cells in leukemic spleens were quantified with PE-Annexin V Apoptosis Detection Kit I (BD Pharmingen, 559763).

To analyze thymic populations, single-cell suspensions of total thymocytes were prepared by disrupting the tissues through a 70-μm filter. The red cells were removed by incubation with the ACK lysing buffer for 5 minutes on ice. Single cells were stained with the following anti-mouse fluorochrome-conjugated antibodies: CD4 PE-eFluor 610 (1:400, RM4-5, eBioscience), CD8a PE (1:200, 53-6.7, BD Pharmingen), CD44 PerCP-Cy5.5 (1:400, IM7, eBioscience), and CD25 Alexa Fluor 488 (1:1000, 7D4, eBioscience). Total thymocytes were visualized in a CD4 vs CD8a plot, and CD4 SPs (CD4⁺CD8a⁻), CD8 SPs (CD4⁻CD8a⁺), CD4/CD8 DPs (CD4⁺CD8a⁺), and CD4/CD8 DNs (CD4⁻CD8a⁻) were gated. To differentiate specific stages of early T-cell development, CD4/CD8 DNs were plotted in a CD44 vs CD25 plot to distinguish DN1 (CD44⁺CD25⁻), DN2 (CD44⁺CD25⁺), DN3 (CD44⁻CD25⁺), and DN4 (CD44⁻CD25⁻) populations.

To analyze bone marrow stem and progenitor populations, single-cell suspensions of total bone marrow cells were prepared by crushing the long bones and passing the cells through a 70-μm filter. Erythrocytes were eliminated using the ACK lysing buffer. Absolute quantification of cell numbers was performed using a Countess II FL instrument. Single cells were stained with the following anti-mouse fluorochrome-conjugated antibodies: lineage cocktail fluorescein isothiocyanate (1:10, 145-2C11; RB6-8C5; RA3-6B2; Ter-119; M,1/70, BioLegend), Ly-6A/E (Sca1) PE (1:400, E13-161.7, BD Pharmingen), CD117 (c-Kit) APC (1:100, 2B8, eBioscience), CD34 eFluor 450 (1:40, RAM34, eBioscience), CD135 (Flt3) PerCP-eFluor710 (1:50, A2F10, eBioscience), and CD16/32 PE-Cy7 (1:100, 93, eBioscience). Flow cytometry analysis of different hematopoietic stem cells (HSCs) and bone marrow progenitors subsets: LT-HSCs (Lin⁻ Sca1⁺ c-Kit⁺ CD34⁻ Flt3⁻), ST-HSC (Lin⁻ Sca1⁺ c-Kit⁺ CD34⁺ Flt3⁻), multipotent progenitor population (MPP) (Lin⁻ Sca1⁺ c-Kit⁺ CD34⁺ Flt3⁺), megakaryocytic-erythroid progenitor (MEP) (Lin⁻ Sca1⁻ c-Kit⁺ CD34⁻ CD16/32⁻), CMP (Lin⁻ Sca1⁻ c-Kit⁺ CD34⁺ CD16/32⁻), and GMP (Lin⁻ Sca1⁻ c-Kit⁺ CD34⁺ CD16/32⁺).

All flow cytometry data were acquired using an Attune NxT flow cytometer (Thermo Fisher Scientific) and analyzed with FlowJo v.10.6.2 software (BD).

RNA sequencing gene expression profiling

Acly conditional knockout ΔE-NOTCH1–induced T-ALL–bearing mice were treated with vehicle only (corn oil; Sigma, C8267) or tamoxifen (Sigma, T5648; 3 mg per mouse in corn oil) to induce isogenic loss of Acly through intraperitoneal injection. After 72 hours, single-cell suspensions of total leukemic splenocytes were prepared by pressing leukemic spleens through a 70-μm filter. We removed the red cells in the spleen samples by incubation with ACK lysing buffer for 5 minutes on ice. RNA was extracted using QIAshredder (QIAGEN, 79656) and RNeasy Mini (QIAGEN, 74106) kits. RNA library preparations and next-generation sequencing were performed using Illumina Next-Seq platform (Illumina). We estimated gene-level raw counts and performed differential expression and/or gene set enrichment analyses (GSEA) as previously described.⁸

ChIP-seq analysis

Analyses of genome-wide H3K27ac, H3K4me3, and H3K9ac marks in leukemic cells isolated 72 hours after vehicle or tamoxifen treatment in leukemic mice harboring Acly conditional knockout leukemias were performed by Active Motif, following well-established protocols and using ChIP-seq–validated antibodies. ChIP-seq reads were trimmed using Trim Galore! (v.0.6.7) with the default parameters. Thereafter, quality control was performed using FastQC (v.0.11.9). Reads were then aligned to the mm39 reference genome using BWA (v.0.7.17) followed by Samtools (v.1.17) to sort the aligned reads. Picard (v.3.0.0) was used to mark the duplicates. The aligned reads were then filtered using BAMtools (v.2.5.2) to remove reads mapped to blacklisted region, reads marked as duplicates, reads not marked as primary alignments, reads unmapped, and reads mapped but with a low mapping quality (multiple hits, secondary alignments, etc). Normalized and Relative Strand Cross-correlation was computed using Phantompeakqualtools (v.1.2.2). BigWig files were then created using deepTools (v.3.5.1). Finally, both narrow peak calling and broad peak calling were performed using magnetic-activated cell sorting 2 (v.2.7.1) with the q-value threshold set to 0.1. Differential peak analysis was performed using edgeR (v.4.20). H3K27ac ChIP-seq from long-term leukemic samples was performed in-house as described previously⁹ (using Thermo Fisher Scientific antibody, MA5-2351).

Statistical analysis

Statistical analyses were performed with Prism 8.0 (GraphPad). Unless otherwise indicated in figure legends, statistical significance between groups was calculated using an unpaired 2-tailed Student t test. Survival in mouse experiments was represented with Kaplan-Meier curves, and significance was estimated with the log-rank test.

Results

ACLY is overexpressed in T-ALL

We previously demonstrated the critical effects of the metabolic consequences of NOTCH1 inhibition in vivo for its antileukemic effects.⁵ Interestingly, we found Acly among the metabolic genes downregulated by NOTCH1 inhibition in T-ALL in vivo (Figure 1A). ACLY has been previously suggested as an important therapeutic target in solid tumors, such as pancreatic cancer,¹⁰ and recent studies have described certain roles for ACLY in the normal physiological function of CD4 and CD8 T cells.^11,12 However, its putative role in the normal development of T cells and other hematological lineages, including in leukemic transformation and progression, remains largely unknown. ACLY controls both metabolic and epigenetic processes, given its critical role as the rate-limiting enzyme in fatty acid synthesis and its generation of acetyl-coenzyme A (acetyl-CoA), which is critical for histone acetylation, as a product of its catalytic reaction.⁷ Because both metabolic and epigenetic mechanisms of resistance to NOTCH1 inhibition in T-ALL have been previously described,^5,8,13 our findings suggested a potential role for ACLY in response to NOTCH1 inhibition and, more broadly, as a relevant player in T-ALL. Next, we analyzed available gene expression data from patients with T-ALL¹⁴ and observed a broad upregulation of ACLY as compared with normal T-cell subsets (Figure 1B). In line with this, ACLY levels were similarly high across all different T-ALL clinical subgroups. However, samples with NOTCH1 or FBXW7 mutations had significantly increased ACLY levels as compared with non-mutated cases¹⁵ (supplemental Figure 1A-B). ACLY overexpression in T-ALL was confirmed at the protein level, with human T-ALL cell lines having higher expression than normal human thymus or peripheral blood cells (Figure 1C). Taken together, these results suggest a previously unknown but important role for ACLY in T-ALL.

Figure 1.

ACLY is a novel therapeutic target in T-ALL. (A) mRNA levels of Acly upon NOTCH1 inhibition with the DBZ gamma-secretase inhibitor in mouse T-ALL in vivo⁵ (∗∗P < .01 calculated with 2-tailed Student t test). (B) Box plot showing ACLY expression among T-ALL samples (n = 57) and physiological thymocyte subsets (n = 21).¹⁰ Quantile normalization was performed across samples. Boxes represent first and third quartiles, and lines represent the median. Whiskers represent the upper and lower limits (P < .001 using Mann-Whitney U test; FDR <.05 using Benjamini-Hochberg correction). (C) Western blot analysis of ACLY and ACTIN expression in human PBMNCs, CD4⁺ T cells, or healthy human thymocytes, as compared with human T-ALL. (D) Schematic illustration of retroviral-transduction protocol for the generation of NOTCH1-induced T-ALLs from inducible Acly conditional knockout mice, followed by transplant into secondary recipients treated with vehicle (Acly+) or tamoxifen (Acly−), with or without DBZ. (E) Kaplan-Meier survival curves of mice harboring Acly-positive and Acly-deleted isogenic leukemias (n = 10 per group). (F) Kaplan-Meier survival curves of mice harboring Acly-positive and Acly-deleted isogenic leukemias treated with 4 cycles of DBZ (5 mg/kg) on a 4-day-on (blue blocks at the bottom) and 3-day-off schedule (n = 10 per group). (G) Western blot analysis of ACLY, ACSS2, and ACTIN expression in leukemic spleens from terminally ill mice from the survival curve in panel E. ∗∗∗P < .005 in panels B-C calculated with log-rank test; ∗P < .05; ∗∗P < .01; ∗∗∗P < .005 in panels D-F calculated with log-rank test calculated with 2-tailed Student t test. d, day; PBMNC, peripheral blood mononuclear cell.

ACLY is critical for T-ALL progression in vivo

Next, we hypothesized that targeting ACLY in T-ALL might confer therapeutic effects. To test this hypothesis, we generated a model of NOTCH1-induced Acly conditional knockout leukemia using a well-established protocol⁵ of retroviral transduction of an oncogenic form of NOTCH1 in bone marrow progenitor cells from Acly conditional knockout mice, followed by transplantation into lethally irradiated recipients (Figure 1D). In time, these mice developed NOTCH1-induced leukemias, which were subsequently transplanted into a secondary cohort of mice and treated with vehicle (control) or tamoxifen, to induce isogenic loss of Acly. Mice were further subdivided into control treated or treated with the gamma-secretase inhibitor DBZ, to test the response to NOTCH1 inhibition in vivo with or without ACLY expression (Figure 1D). Notably, our results demonstrated intrinsic antileukemic effects for ACLY loss (Figure 1E), which further synergized with NOTCH1 inhibition resulting in a 20% cure rate of leukemic mice (Figure 1F). Interestingly, leukemias eventually developing in tamoxifen-treated mice were not genetic escapers and did not have a concomitant feedback upregulation of ACSS2 (Figure 1G) or PDHA1 (supplemental Figure 1C). Upregulation of both ACSS2 and PDHA1 has been previously found to mediate resistance to ACLY loss in certain contexts,^16,17 suggesting that T-ALL cells can still survive ACLY loss via alternative mechanisms.

ACLY is dispensable for normal T-cell development

Many relevant therapeutic targets of T-ALL progression also show an effect in normal T-cell development, and the potential effects of ACLY loss in normal healthy hematopoietic cells would also be important to predict potential toxicities associated to therapies aimed at inhibiting ACLY. To this end, we bred the Acly conditional knockout mice with Vav-iCre–expressing mice,¹⁸ which express Cre in the whole hematopoietic system. Mice with hematopoietic-specific ACLY loss were viable and had no overt phenotype. Detailed immunophenotypic analyses of thymi from these mice revealed no differences in thymus weight or cellularity (Figure 2A-B), and we also failed to detect any significant difference in the numbers or proportions of different thymocyte subsets (Figure 2C-F). We then tried to generate NOTCH1-induced leukemias from these mice with germinal loss of Acly in the hematopoietic system. However, although the total numbers of lineage-negative progenitor cells were unchanged, cells with ACLY loss failed to proliferate and had drastically reduced survival in tissue culture conditions in vitro as compared with wild-type cells, hindering our ability to infect them and to generate leukemias from them (supplemental Figure 2A-B). Next, we performed detailed bone marrow immunophenotypic profiling from these mice. Interestingly, our results revealed reduced numbers of long-term HSCs, short-term HSCs, and MPPs, whereas the numbers of more committed populations, such as MEPs, CMPs, and GMPs, were unchanged (supplemental Figure 2C-D). Overall, our data suggest that ACLY is dispensable for normal T-cell development but plays a more relevant role in early bone marrow progenitor cells and in established T-ALL.

Figure 2.

ACLY loss is dispensable for normal T-cell development in vivo. (A) Thymus weight in Acly^flox/flox-Vav-iCre and Acly^+/+-Vav-iCre mice. (B) Total thymocyte count in Acly^flox/flox-Vav-iCre and Acly^+/+-Vav-iCre mice. (C-D) Representative plots (C) and quantification (D) of CD4⁻CD8⁻ (DN), CD4⁺CD8⁺ (DP), CD4⁺CD8⁻ (CD4), and CD4⁻CD8⁺ (CD8) populations in the thymus from Acly^flox/flox-Vav-iCre and Acly^+/+-Vav-iCre mice. (E-F) Representative plots (E) and quantification (F) of CD44⁺CD25⁻ (DN1), CD44⁺CD25⁺ (DN2), CD44⁻CD25⁺ (DN3), and CD25⁻CD44⁻ (DN4) populations in the thymus from Acly^flox/flox-Vav-iCre and Acly^+/+-Vav-iCre mice. No comparison was significant using 2-tailed Student t test. N.S., not significant.

ACLY loss leads to a metabolic crisis and reduced tumor burden

To explore the mechanistic effects of ACLY loss, we next performed an acute deletion experiment in which mice harboring Acly conditional knockout leukemias were first allowed to become fully leukemic before treating them with vehicle (control) or tamoxifen (to induce ACLY loss), followed by euthanasia 72 hours later (Figure 3A). In this setting, acute ACLY loss translated into reduced Acly levels (Figure 3B) and led to reduced tumor burden (Figure 3C). Mice harboring Acly-deleted leukemias had significantly reduced infiltration across tissues, including the spleen or the bone marrow and, most notably, also had reduced central nervous system infiltration in the meninges (Figure 3D). These results were mainly driven by cytotoxic effects, as revealed by increased annexin V staining upon ACLY loss (Figure 3E).

Figure 3.

Metabolic consequences of ACLY loss in T-ALL. (A) Schematic of acute Acly deletion experiment in leukemic mice in vivo. (B) Quantitative RT-PCR analysis of Acly mRNA expression in tumor cells isolated from Acly conditional knockout leukemia-bearing mice 72 hours after being treated with vehicle only (Acly+) or tamoxifen (Acly−) in vivo. (C-D) Tumor burden in Acly conditional knockout leukemia-bearing mice 72 hours after being treated with vehicle only (Acly+) or tamoxifen (Acly−) in vivo as revealed by total spleen weight (C) or the infiltration of GFP-positive leukemic cells in the spleen, bone marrow, or meninges (D). (E) Representative flow cytometry plots of annexin V (apoptotic cells) and 7-AAD (dead cells) staining (left) and quantification of apoptosis (right) in leukemic spleens from Acly conditional knockout leukemia-bearing mice 48 hours after being treated with vehicle only (Acly+) or tamoxifen (Acly−) in vivo (n = 4-5 per treatment in panels B-E; ∗P < .05 and ∗∗∗P < .005 using 2-tailed Student t test). (F-G) Relative acetyl-CoA (F) and AICAR (G) abundance upon isogenic loss of Acly in leukemic spleens 48 to 72 hours after being treated with vehicle only (Acly+) or tamoxifen (Acly−) in vivo. (H) Relative abundance of the indicated pyrimidine intermediates upon tamoxifen-induced isogenic loss of Acly in leukemic spleens from mice treated as in panel A. (I) Relative abundance of the indicated glycolytic intermediates upon tamoxifen-induced isogenic loss of Acly in leukemic spleens from mice treated as in panel A. (J) Relative abundance of glutathione disulfide upon tamoxifen-induced isogenic loss of Acly in leukemic spleens from mice treated as in panel A. (K-L) Growth curve (K) and western blot analyses of ACLY and ACTIN expression (L) in an Acly conditional knockout T-ALL cell line in vitro upon treatment with ethanol (control) or 4-hydroxytamoxifen (4OHT) to induce ACLY loss. (M) Oxygen consumption rate OCR in Acly conditional knockout T-ALL cells, under basal conditions or after 4OHT-induced loss of ACLY, measured in real-time using a Seahorse XF24 instrument. Data are presented as mean ± standard deviation of n = 4 wells. n = 6 to 8 per treatment in panels F-G and n = 4 to 5 per treatment in panels H-J; ∗P < .05; ∗∗P < .01; ∗∗∗P < .005 using 2-tailed Student t test. AICAR, 5-aminoimidazole-4-carboxamide ribonucleotide; d, day; min, minutes; FCCP, carbonyl cyanide-p-trifluoromethoxyphenylhydrazone; OCR, oxygen consumption rate.

Next, given the well-known metabolic role of ACLY, we performed global metabolomic analyses upon acute ACLY loss in vivo. Interestingly, we observed an expected reduction in acetyl-CoA levels (Figure 3F) and a strong accumulation of 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) (Figure 3G), a purine intermediate and a well-described activator of AMPK.¹⁹ In addition, we detected accumulation of the pyrimidine intermediate N-carbamoyl-L-aspartate with a concomitant global reduction in pyrimidines (Figure 3H), overall suggesting a block in nucleotide biosynthesis. Finally, we also observed an accumulation of glucose and glycolytic intermediates (Figure 3I) and an increase in oxidized glutathione levels (Figure 3J). Consistent with these metabolic effects, loss of ACLY in an Acly conditional knockout T-ALL cell line generated from our primary leukemia led to impaired proliferation (Figure 3K-L), together with a significant reduction in oxygen consumption rate (Figure 3M). Overall, our results suggest that ACLY loss leads to a metabolic crisis in T-ALL cells.

An ACLY-MYC feedforward loop regulates T-ALL progression

Next, gene expression profiling analyses revealed a strong transcriptional signature upon acute ACLY loss (Figure 4A; supplemental Table 1). Pathway analyses using enrichplot revealed strong downregulation of pathways related to oxidative phosphorylation and translation (Figure 4B), consistent with the metabolic changes previously observed. To gain a deeper understanding of the transcription factor/s that may be governing these transcriptional effects, we analyzed our gene expression profile data using Enrichr²⁰ and Dorothea.²¹ Intriguingly, both algorithms identified a downregulation of MYC activity as potentially related to the changes observed (Figure 4C; supplemental Figure 3A). Previous reports suggested a potential regulation of MYC by ACLY in endothelial cells²² and the reverse regulation of ACLY by MYC in prostate cancer.²³ To unravel the directionality of the regulation in our T-ALL setting, we performed GSEA²⁴ against the signature of MYC loss in T-ALL upon deletion of the N-Me enhancer, which controls Myc expression downstream of NOTCH1 in T cells.²⁵ Our results revealed a very strong correlation between both signatures (Figure 4D), and western blot analyses confirmed a drastic downregulation in MYC protein levels (Figure 4E), which may help explain the strong antileukemic effects observed, given the critical role of MYC in T-ALL.²⁶ Because ACLY has a well-described role in epigenetic regulation, and our own results showed reduced acetyl-coA levels upon ACLY loss (Figure 3F), we next performed ChIP-seq epigenetic profiling experiments, which revealed a global reduction in H3K27ac mark with a concomitant milder reduction in H3K4me3 levels (Figure 4F). Interestingly, the effect in histone acetylation might be specific for H3K27ac, as we did not observe major changes in H3K9ac (Figure 4F). Importantly, these epigenetic changes were evident in the Myc promoter, which had markedly reduced levels of H3K27ac and a milder decrease in H3K4me3 (Figure 4G), consistent with the changes in MYC activity previously observed.

Figure 4.

ACLY-MYC axis in T-ALL. (A) Heat map representation of the top differentially expressed genes between control (Acly+) and tamoxifen-treated (Acly−) Acly conditional knockout T-ALL. Cutoffs used: top 40 up and downregulated genes; P adjusted value <.0001; log-fold change >1. Scale bar illustrates color-coded differential expression, with red indicating higher and blue indicating lower levels of expression. (B) Pathway analyses of significantly downregulated or upregulated genes upon ACLY loss using enrichplot (https://bioconductor.org/packages/release/bioc/html/enrichplot.html). (C) Enrichr analyses of transcription factors controlling downregulated pathways upon ACLY loss. (D) GSEA of genes downregulated by N-Me loss in control (vehicle-only treated; ACLY ON) or Acly-deleted leukemias (tamoxifen treated; ACLY OFF) in vivo. (E) Western blot analyses of ACLY, MYC, and ACTIN expression 72 hours after tamoxifen-induced deletion of Acly in vivo. (F) Heat map representation of genome-wide enrichment for H3K27ac (left), H3K4me3 (middle), and H3K9ac (right) marks in control (vehicle treated; ACLY ON) or Acly-deleted leukemias (tamoxifen treated; ACLY OFF) in vivo. The color of the heat map indicates the number of CPM-normalized reads within ±3000 bp of annotated Transcription Start Site (TSS) for each gene. (G) Epigenetic profiling around the Myc promoter revealing ChIP-seq tracks for the indicated histones marks in mouse T-ALL leukemic samples treated as in panel F. (H) GSEA of genes downregulated by ACLY loss in control (vehicle treated; N-Me ON) or N-Me–deleted leukemias (tamoxifen treated; N-Me OFF) in vivo.²⁰ (I) Luciferase reporter activity in 293T cells of a pGL4 promoter empty construct (pGL4-Luc) or a pGL4 promoter plus the human ACLY promoter, in the presence or absence of MYC overexpression or treatment with JQ1. Data from 3 independent transfection replicates are shown ∗∗∗P < .005 using 2-tailed Student t test. (J) Western blot analyses of ACLY, MYC, and ACTIN expression 48 hours after tamoxifen-induced deletion in vitro of N-Me using an N-Me conditional knockout T-ALL cell line.²⁰

Conversely, reverse GSEA analyses also revealed that the signature of MYC loss upon N-Me deletion is very similar to the one obtained upon ACLY loss (Figure 4H), suggesting that MYC might also be upstream of Acly. Indeed, luciferase reporter assays testing this regulation showed that MYC overexpression positively regulates the transcriptional activity of the ACLY promoter, whereas treatment with the BRD4 inhibitor JQ1, which prominently inhibits MYC activity,²⁷ resulted in reduced transactivation of ACLY (Figure 4I). In line with this, epigenetic profiling of the ACLY promoter using publicly available ChIP-seq data in human T-ALL cells uncovered the binding of multiple relevant T-cell–specific transcription factors, including both MYC and NOTCH1 (supplemental Figure 3B). Finally, we took advantage of our mouse N-Me conditional knockout primary T-ALL cell line²⁵ to demonstrate that MYC loss in T-ALL resulted in significantly reduced ACLY protein levels (Figure 4J), further supporting that MYC also regulates Acly expression. This is consistent with the broad ACLY overexpression we previously observed in T-ALL (Figure 1B) and with its downregulation upon NOTCH1 inhibition (Figure 1A), given the tight link between NOTCH1 and MYC in T-ALL.²⁵

Resistance to ACLY loss is associated with restored MYC levels

Even if targeting ACLY has strong intrinsic antileukemic effects in vivo, tamoxifen-treated leukemic mice eventually relapse and succumb to their disease, and these leukemias are not genetic escapers nor have compensatory upregulation of ACSS2 or PDHA1 (Figure 1G; supplemental Figure 1C), supporting alternative mechanisms of resistance. Interestingly, western blot analyses of relapsing leukemias revealed restored expression of MYC protein levels (supplemental Figure 3C). Moreover, acetyl-CoA levels were also restored in these relapsing leukemias (supplemental Figure 3D), leading to a complete rescue of the effects on H3K27ac at the Myc promoter level (supplemental Figure 3E). Similarly, gene expression profiling analyses of relapsing leukemias confirmed the restoration of MYC-driven transcriptional programs and additional pathways that were significantly changing upon acute ACLY loss (supplemental Figure 3F-G).

Pharmacological inhibition of ACLY has antileukemic effects in human T-ALL

Finally, to further assess the clinical relevance of our findings, we tested the effects of the ACLY inhibitor BMS-303141 in human T-ALL cells. Interestingly, treatment with BMS-303141 in vitro led to impaired proliferation and a strong reduction in MYC levels in a broad panel of T-ALL cell lines, regardless of PTEN or NOTCH1 mutational status (Figure 5A-B; supplemental Figure 4A-D). As expected, similar results were observed in our mouse T-ALL cell line (supplemental Figure 4E-F). Next, we decided to test its effects in 2 different PTEN-positive human T-ALL PDXs^3,4 upon short-term culture in vitro, and we observed similar results (Figure 5C-D; supplemental Figure 4G-H). Finally, we tested the potential antileukemic effects of BMS-303141 in vivo in mice harboring one of these PDXs, and, most notably, BMS-303141 treatment led to a mild but significant extension in survival (Figure 5E). Overall, our results demonstrate that ACLY inhibition leads to antileukemic effects associated with decreased levels of MYC in human T-ALL cell lines and PDXs, underscoring the potential clinical relevance of ACLY inhibitors in T-ALL treatment.

Figure 5.

Antileukemic effects of ACLY pharmacological inhibition in human T-ALL. (A) Effects on cell survival in different human T-ALL cell lines upon 3 days of treatment with ethanol (Et-OH; control) or the ACLY inhibitor BMS-303141 in vitro. (B) Western blot analyses of ACLY, MYC, and ACTIN expression 24 hours after treatment with ethanol or BMS-303141 in the same human cell lines. (C) Effects on cell survival in human T-ALL PDX PDTALL 19 upon 3 days of treatment with ethanol or BMS-303141 in vitro. (D) Western blot analyses of ACLY, MYC, and ACTIN expression 24 hours after treatment with ethanol or BMS-303141 in the same PDX. (E) Kaplan-Meier survival curves of immunodeficient mice transplanted with PDTALL 19 and treated with vehicle or BMS-303141 on a 5-day on and 2-day off schedule (n = 7-8 per group). ∗∗∗P < .005 in panels A,C using 2-tailed Student t test; ∗P < .05 in panel E calculated with log-rank test.

Discussion

Collectively, our results identify a therapeutically relevant ACLY-MYC feedforward loop in T-ALL in which ACLY provides acetyl-CoA units needed for the sustained transcriptional regulation of MYC, whereas MYC itself further reinforces this loop by transcriptionally upregulating ACLY. Importantly, although our results identified a critical role for ACLY in T-ALL progression in vivo, ACLY is dispensable for normal T-cell development, and mice lacking ACLY in the heme compartment have no gross abnormalities, suggesting that targeting ACLY might have limited toxicities for normal healthy cells. Still, detailed immunophenotypic analyses revealed a reduction in the numbers of earlier populations of hematopoietic stem cells, and lineage-negative, ACLY-deleted cells failed to proliferate in vitro. However, more committed downstream populations such as MEPs, CMPs, and GMPs were unchanged. These results suggest that, although ACLY loss leads to defective early progenitor cells, these cells can largely compensate for this in their normal physiological context (as opposed to tissue culture conditions).

MYC and AKT are 2 of the most relevant oncogenic factors downstream of NOTCH1 in T-ALL.^5,25 Intriguingly, AKT has also been previously found to directly phosphorylate and positively regulate ACLY,²⁸ and we verified that this happens in the context of both mouse and human T-ALL (supplemental Figure 4A-B). Thus, the transcriptional program controlled by NOTCH1 in T-ALL leads to increased ACLY activity through both MYC and AKT pathways, highlighting the key role of ACLY in leukemogenesis. Although the increased phospho-ACLY levels in PTEN-negative T-ALLs suggest that PTEN-positive and PTEN-negative leukemias might have different dependencies on ACLY, we observed similar antileukemic effects upon ACLY inhibition in a broad panel of human T-ALL cell lines regardless of PTEN mutational status. It is also interesting to note that acute loss of ACLY in T-ALL led to a striking massive reduction in MYC protein levels. Thus, it is tempting to speculate that, beyond the epigenetic downregulation of Myc transcription due to reduced H3K27ac of its promoter, additional mechanisms may also contribute to this phenotype. Related to this, MYC stability is also known to be regulated by direct acetylation of several lysine residues,²⁹ and lower levels of MYC acetylation have been associated with decreased MYC stability.³⁰ Therefore, the reduced availability of acetyl-CoA upon ACLY loss may also contribute to the reduction in MYC levels by directly affecting its stability. Finally, it is also interesting to speculate whether the metabolic effects observed upon ACLY loss are MYC driven or ACLY driven. Although it is difficult to disentangle this effect, given that our results demonstrate that MYC and ACLY regulate each other, we specifically investigated the transcriptional changes related to metabolism upon ACLY loss (in our Acly conditional knockout leukemias) or upon MYC loss (in our N-Me conditional knockout leukemias). Commonly downregulated pathways upon MYC or ACLY loss are mainly related to TCA and oxacid/organic acid metabolism and, to a lesser extent, to nucleotide metabolism, whereas exclusively downregulated genes upon loss of ACLY are more enriched in nucleotide-related pathways and exclusively downregulated genes upon MYC loss are more enriched in oxacid/organic acids metabolism and TCA pathway (supplemental Figure 5). Still, there is overlap between all these metabolic processes. Thus, the metabolic effects observed upon ACLY loss are likely a combination of changes driven by MYC depletion and changes driven by ACLY depletion unrelated to MYC.

Overall, our results demonstrate that targeting ACLY constitutes a novel and attractive therapeutic option for leukemia treatment, as verified using BMS-303141 in human T-ALL cell lines and PDXs. Still, BMS-303141 is only a tool compound.^31,32 Although several other ACLY inhibitors have been described to date, these are either liver-specific inhibitors, such as bempedoic acid,³³ or potent but not cell-permeable inhibitors, such as NDI-091143.^34,35 In this context, our unequivocal genetic and pharmacological results demonstrating a therapeutic role for targeting ACLY in leukemia in vivo will hopefully spur additional efforts to develop clinical-grade ACLY inhibitors that can be used in patients with leukemia in the near future.

Conflict-of-interest disclosure: The authors declare no competing financial interests.