1 Glutamine antagonism suppresses tumor growth in adrenocortical carcinoma

2 through inhibition of de novo nucleotide biosynthesis

Vasileios Chortis1,2,3,4*#, Kleiton Silva Borges1,2*, Cong-Hui Yao5, Claudio Ribeiro1,2, Luis Fernando Nagano6, Mesut Berber1,2, Alessandro Prete3,4,7, Lukáš Najdekr8, Michail E.

3 4 5 6 Klontzas9, Andris Jankevics8, Pedro Vendramini1,2, Jean Lucas Kremer1,2, Liam Kelley5,

7 Sathuwarman Raveenthiraraj10, Stylianos Tsagarakis11, Magdalena Macech12, Ivana D. Pupovac13, Thomas G Papathomas3,14, Betul Haykir1,2, Catherine Winder8, Marcus Quinkler15, M. Conall Dennedy16, Grethe Å. Ueland17, Felix Beuschlein 18,19,20, Antoine

8 9 10 11 12 13 Breault1,2,25#

Tabarin21, Martin Fassnacht22, Angela E. Taylor3,4, Darko Kastelan13, Urszula

Ambroziak12, Dimitra A. Vassiliadi11, Katja Kiseljak-Vassiliades23, Irina Bancos24, Diana L. Carlone1,2,25, Warwick B. Dunn8,26, Wiebke Arlt27,28, Marcia C. Haigis5, and David T.

14 15 16 17 18 of Medicine and Health, University of Birmingham, Birmingham, UK.

1Division of Endocrinology, Boston Children’s Hospital, Boston, MA, USA.

2Department of Pediatrics, Harvard Medical School, Boston, MA, USA.

3Department of Metabolism and Systems Science, School of Medical Sciences, College

19 4Department of Endocrinology, Queen Elizabeth Hospital, University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK.

20 21 5Department of Cell Biology, Blavatnik Institute, Harvard Medical School, Boston, MA,

22 USA.

23 6Department of Genetics, Ribeirão Preto Medical School, University of São Paulo,

24 Ribeirão Preto, São Paulo, Brazil.

7National Institute for Health and Care Research (NIHR) Birmingham Biomedical

26 Research Centre, Birmingham, UK.

27 8Phenome Centre Birmingham, School of Biosciences, University of Birmingham,

28 Birmingham, UK

29 9Artificial Intelligence and Translational Imaging (ATI) Lab, Department of Radiology,

30 School of Medicine, University of Crete, Heraklion, Greece.

10Division of Newborn Medicine, Boston Children’s Hospital, Boston, MA.

31 32 33 Athens, Greece.

11Department of Endocrinology, Diabetes and Metabolism, Evangelismos Hospital,

12Department of Internal Medicine and Endocrinology, Medical University of Warsaw, Warsaw, Poland.

13Department of Endocrinology, University Hospital Centre Zagreb, Zagreb, Croatia.

14Department of Clinical Pathology, Vestre Viken HF, Norway.

34 35 36 37 38 15Endocrinology in Charlottenburg, Berlin, Germany. 39 16Department of Medicine, Royal College of Surgeons in Ireland, University of Medicine and Health Sciences, Dublin, Republic of Ireland.

40 41 17Department of Medicine, section of Endocrinology, Haukeland University Hospital,

42 Bergen, Norway.

43 18Klinik für Endokrinologie, Diabetologie und Klinische Ernährung, Universitäts-Spital

44 Zürich (USZ) und Universität Zürich (UZH), Zurich, Switzerland.

45 19The LOOP Zurich - Medical Research Center, Zurich, Switzerland

46 20Medizinische Klinik und Poliklinik IV, Ludwig-Maximilians-Universität München, Munich,

21Service d’Endocrinologie, Centre Hospitalier Universitaire, Hopital du Haut Leveque,

47 Germany. 48 49 Pessac, France. 50 51 Hospital, University of Würzburg, Würzburg, Germany.

22Department of Internal Medicine I, Division of Endocrinology and Diabetes, University

23Rocky Mountain Regional Veterans Affairs Medical Center, Aurora, CO 80045, USA.

24Division of Endocrinology, Metabolism, Diabetes and Nutrition, Department of Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA.

25Harvard Stem Cell Institute, Cambridge, MA, USA.

26Centre for Metabolomics Research, Biochemistry, Cell and Systems Biology, Institute of Systems, Molecular, and Integrative Biology, University of Liverpool, Liverpool, L69 7ZB, UK.

59 27Medical Research Council Laboratory of Medical Sciences, London, UK

28Institute of Clinical Sciences, Imperial College London, London, UK.

*These authors contributed equally to this work

Co-corresponding authors

Correspondence: v.chortis@bham.ac.uk (V.C.), david.breault@childrens.harvard.edu

(D.T.B.)

52 53 54 55 56 57 58

60 61 62 63 64 65 66 67 68

Abstract

69 70 71 72 73

74 75 76 77

Dysregulation of cellular metabolism is a hallmark of cancer, which remains poorly understood in adrenocortical carcinoma (ACC). Here, we dissected ACC metabolism by integrating transcriptional profiling from human and mouse ACC, targeted tissue metabolomics from a mouse ACC model, and untargeted serum metabolomics from a large patient cohort, providing cross-species validation of metabolic rewiring in ACC. This study revealed global metabolic dysregulation, involving glutamine-dependent pathways such as non-essential amino-acid and hexosamine biosynthesis, nucleotide metabolism, and glutathione biosynthesis, suggesting glutamine catabolismis a critical metabolic vulnerability in ACC. Treatment with glutamine antagonists 6-Diazo-5-Oxo-L-Norleucine (DON) and JHU-083 elicited robust anti-tumor responses. Mechanistic studies revealed DON’s anti-tumor effect was primarily driven by selective inhibition of glutamine-fueled de novo nucleotide biosynthesis. Additionally, DON led to DNA damage, which yielded potent synergism with inhibition of the DNA damage response pathway. Collectively, this work highlights glutamine metabolism as a central metabolic dependency and therapeutic

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95

target in ACC.

Word Count: 151

Abbreviations

ACC	Adrenocortical Carcinoma
Gln	Glutamine
GOF	gain-of-function
LOF	loss-of-function
ßCat	ß-catenin
DON	6-diazo-5-oxo-L-norleucine

Highlights

. Mouse and human ACC share conserved transcriptional-metabolic programs, revealing Gln metabolism as a central, targetable vulnerability.

· Targeted tissue metabolomic analysis in a mouse model of ACC validates dysregulation in Gln-dependent metabolic pathways.

· Targeting of GIn metabolism with JHU-083 (6-diazo-5-oxo-L-norleucine (DON) pro- drug) achieves marked inhibition of tumor growth in vivo.

· High expression of GIn-metabolizing genes mediating de novo nucleotide biosynthesis is associated with poor prognosis in ACC.

. DON drives nucleotide depletion and DNA damage, leading to potent synergy with inhibition of the DNA damage response.

· Untargeted serum metabolomic analysis in a large cohort of patients with adrenal tumors demonstrates dysregulation of Gln and nucleotide metabolism in ACC.

96 97 98 99 100 101 102 103 104 105

106 107 108 109 110

Introduction

Adrenocortical Carcinoma (ACC) is a rare and aggressive malignancy, originating from the adrenal cortex. Most patients present with, or develop, metastatic disease; for such patients, there are no effective medical treatments and overall survival remains poor with a median survival of less than 15 months with cytotoxic chemotherapy1,2. Despite considerable progress in understanding the molecular biology of ACC3,4, no effective 117 targeted treatments have yet emerged. Therefore, there is a critical unmet clinical need to develop effective medical treatments for patients with inoperable tumors.

Over the last decade, seminal studies have characterized the molecular landscape of ACC, identifying molecular clusters associated with distinct clinical outcomes3-5. This work demonstrated that constitutive activation of the Wnt/B-catenin signaling pathway and cell cycle dysregulation due to TP53/RB1 loss are the two most common driver events in ACC, and are most prevalent in patients with poor prognosis. Notably, concurrent alterations in both pathways are observed in up to 30% of ACC cases6. Building on these findings, we recently developed a genetically engineered mouse model of ACC (BPCre), which combines Wnt/B-catenin activation with p53 loss7. This model faithfully recapitulates major features of advanced ACC, including metastasis and active steroidogenesis as well as epigenetic dysregulation7-9, which accounts for the majority of patients with metastatic disease.

Dysregulation of cellular metabolism is recognized as a hallmark of cancer10,11, which supports cancer cells’ specific metabolic needs, including the rapid generation of biomass and energy, and a robust antioxidant defence system. Despite ACC’s unique metabolic trait as an actively steroidogenic tumor, there is limited understanding of how

111 112 113 114 115 116

118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133

34 ACC tumors rewire their metabolic network and what diagnostic or t therapeutic opportunities this aspect of ACC biology may present.

Here, we present a comprehensive analysis of metabolic dysregulation in ACC by integrating transcriptional profiling from human and mouse ACC, targeted tissue metabolomics from a mouse model of ACC, and untargeted serum metabolomics from a large cohort of patients with adrenal tumors, achieving cross-species validation of metabolic rewiring in ACC. These findings reveal dysregulation in pathways supporting non-essential amino-acid and nucleotide biosynthesis, hexosamine biosynthesis, and antioxidant production, all of which are heavily dependent on glutamine (Gln) as a key metabolic substrate. Given Gln’s central role in fueling anabolic and redox-supporting processes, we used the Gln antagonist 6-diazo-5-oxo-L-norleucine (DON), and its recently developed tumor-selective pro-drug JHU-08312-14, to target this putative metabolic dependency in models of ACC. We demonstrate that Gln antagonism is highly effective against ACC using in vitro and in vivo pre-clinical models. Furthermore, we show that DON’s anti-tumor activity against ACC is specifically contingent on disruption of Gln’s role in de novo nucleotide biosynthesis, a process that is up-regulated in the most aggressive human ACCs. Moreover, we discovered that DON-driven nucleotide depletion leads to DNA damage, resulting in potent synergism with inhibition of the DNA damage response pathway. Consistent with these findings, untargeted serum metabolomics from patients with adrenal tumors identified key changes in nucleotide and amino-acid homeostasis in ACC. These pre-clinical findings reveal Gln antagonism as a promising new therapeutic target for this aggressive cancer.

135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155

156

Results

Cross-species analysis reveals conserved transcriptional-metabolic programs in ACC

To identify actionable metabolic vulnerabilities in ACC, we sought to identify common up-regulated metabolic programs utilized by mouse and human ACC. To achieve this, we transcriptionally profiled our recently developed BPCre genetically engineered mouse model of ACC7 to identify dysregulated metabolic pathways and compared these with pathways also dysregulated in human ACC. First, we performed bulk RNA-sequencing (RNA-seq) on BPCre tumors (n = 7 mice, 5 female and 2 male mice, aged 9-12 months) and control mouse adrenals without mutations (n = 6 mice, all female, aged 5-11 months). Then, we analyzed bulk RNA-seq data from human ACC tumors using the TCGA-ACC database (n = 79)3 and normal human adrenals from the GTEx database (n = 295)15 (Fig. 1a). We focused on the expression of 2,754 defined metabolic genes encoding metabolic enzymes and transporters, using an unbiased, standardized transcriptomic approach to identify uniquely dysregulated tumor pathways 16- 18. Comparison of mouse and human ACC tumors revealed a 68% overlap in up-regulated metabolic pathways and a 75% overlap in down-regulated metabolic pathways (Fig. 1b, Suppl. Fig. 1a, Suppl. Tables 1-2), underscoring the strong conservation of transcriptional-metabolic programs between species in ACC. Strikingly, all 15 of the most significantly up-regulated pathways in mouse tumors were also up-regulated in human tumors (Fig. 1c, Suppl. Table 1), revealing a conserved metabolic signature.

Dysregulated purine and pyrimidine metabolism was a prominent feature in this analysis, with up-regulation of several genes involved in de novo biosynthesis (a Gln-

157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178

dependent multi-step process synthesising nucleotides from basic precursors) and salvage biosynthesis (mediating nucleotide recycling) (Suppl. Fig. 1b-c). Interestingly, amino sugar and nucleotide sugar metabolism (Suppl. Fig. 2a), N-glycan biosynthesis, and glycosaminoglycan biosynthesis were all among the most up-regulated pathways, consistent with an increased requirement for activated sugar donors to support protein glycosylation and glycosaminoglycan biosynthesis. Taken together, these prominent changes indicate an increased demand for Gln, an essential nitrogen donor to both de novo nucleotide biosynthesis and amino sugar (hexosamine) biosynthesis (Fig. 1d). Additional indications of an increased demand for Gln included i) an up-regulation of transamidases channeling Gln-derived nitrogen to aspartate and asparagine biosynthesis (GOT2, ASNS) in both species (Suppl. Table 1-2) and ii) extensive dysregulation of glutathione metabolism, a major Gln-dependent antioxidant system (Suppl. Fig. 2b). One-carbon metabolism, encompassing the folate cycle and serine/glycine metabolism, was also prominently up-regulated, further supporting nucleotide synthesis and redox homeostasis. Overall, this cross-species analysis revealed extensive rewiring of Gln and one-carbon metabolism to support nucleotide and protein biosynthesis, redox balance, and protein glycosylation.

Gln-dependent metabolic pathways are dysregulated in mouse ACC

To quantify tumor-intrinsic metabolic alterations and to validate the functional consequences of the transcriptional alterations observed by RNA-seq, we performed targeted tissue metabolomics in BPCre tumors from both sexes compared with age- matched control adrenals (lacking mutations), as well as adrenals from BPCre mice prior

179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201

202 203 204 205 206 207 208 209 210 to tumor formation (BPCre-Pre) (Fig. 2a)7. Targeted polar metabolomic analysis by liquid chromatography-mass spectrometry (LC-MS), using a library that provides a representative cross-section of core carbon and nitrogen-utilising metabolic pathways, confirmed extensive metabolic rewiring involving key programmes highlighted by the transcriptomic analysis (Fig. 2b). We observed a robust increase in the Gln-dependent amino-acids glutamate, aspartate, proline and ornithine, as well as the major antioxidant glutathione and its analogue ophthalmic acid, in ACC tumors (Fig. 2c, Suppl. Table 3). Elevated hexosamine biosynthesis intermediates, including uridine diphosphate N- acetylglucosamine (UDP-GlcNAc) and uridine diphosphate N-acetylgalactosamine (UDP- 211 GalNAc), were consistent with enhanced Gln-fueled hexosamine biosynthesis to support 212 213 214 215 216 217 218 219 220 221 222 223 224 protein glycosylation, in keeping with the transcriptomic analysis. We also identified prominent changes in nucleotide metabolism, particularly related to alterations in the tissue-level abundance of purine salvage/degradation intermediates. Interestingly, there was a depletion in purine metabolites xanthine, xanthosine, uric acid (potentially reflecting increased recycling to meet biosynthetic demands), but an increase in pyrimidine metabolites thymidine, cytidine, and 3-ureidopropionic acid (suggestive of increased turnover/overflow catabolismespite transcriptional up-regulation of several enzymes mediating de novo purine/pyrimidine biosynthesis, steady-state tissue metabolomics did not show changes in de novo intermediates, with the notable exception of the strong depletion of xanthosine monophosphate (XMP), an obligate substrate of the Gln- dependent enzyme GMP synthetase (GMPS). This pattern could be consistent with increased metabolic flux coupled with tight channelling to down-stream processes (DNA/RNA synthesis, hexosamine biosynthesis). In addition to these changes, there was

a robust increase in the Gln-dependent co-factors nicotinamide adenine dinucleotide (NAD) and reduced NAD (NADH). Collectively, these findings highlight Gln metabolism as a putative metabolic dependency in ACC. Interestingly, several of these metabolites exhibited progressive changes across the tumorigenic trajectory from normal adrenals to BPCre adrenals pre-malignant transformation to established BPCre ACC tumors, suggestive of involvement in early ACC tumorigenesis (Fig. 2d, Suppl. Fig. 3a).

As additional controls, we compared the metabolomes of adrenals from male and female mice, from young and older mice, and from single-mutant mice. No significant sex- specific differences were observed in tumor metabolomes (Suppl. Fig. 3b), although several nucleotide metabolites differed between normal male and female adrenal glands 235 (Suppl. Fig. 3c). Additionally, there were no significant differences between control adrenals from younger mice (3-5 months) and older mice (7-10 months) (Suppl. Fig. 3d), indicating that metabolic differences between BPCre-Pre adrenals and ACC tumors or older control adrenals are not a function of increasing age. Analysis of adrenals from single-mutant (BCre or PCre) mice revealed only modest changes compared to the complex metabolic rewiring seen in BPCre adrenals prior to tumor formation and older mice with BPCre tumors (Supplementary Fig. 4a-c).

ACC is highly sensitive to disruption of Gln metabolism in vitro and in vivo

To assess whether Gln metabolism is a druggable metabolic dependency in ACC, we employed the Gln antagonist 6-diazo-5-oxo-L-norleucine (DON)12-14,19 . To test its efficacy, we treated a panel of ACC cell lines with DON, including a new murine ACC cell line (BCH-ACC3) derived from a BPCre tumor and three human ACC cell lines (NCI-

225 226 227 228 229 230 231 232 233 234

236 237 238 239 240 241 242 243 244 245 246 247

H295R, CU-ACC1, CU-ACC2)20 harboring common ACC driver alterations (B-Catenin gain-of-function and/or p53 loss-of-function) (Suppl. Fig. 5a). All ACC cell lines displayed marked sensitivity to DON, with IC50 values ranging from 0.03 to 4.07 uM (Fig. 3a). Notably, low micromolar concentrations (2.5-5 uM) were sufficient to fully impede colony formation in clonogenic assays across all cell lines (Fig. 3b).

To test the effectiveness of Gln antagonism in vivo, we established a syngeneic subcutaneous (SQ) implant model using BCH-ACC3 cells and immunocompetent male C57BL/6J mice. BCH-ACC3 cells were injected SQ into the flanks of C57BL/6J mice, where they rapidly formed tumors (Fig. 3c, Suppl. Fig. 5b-c). To assess the impact of Gln antagonism in this model, mice harboring established SQ tumors were treated with JHU-083, a DON pro-drug that is selectively activated within the tumor microenvironment12-14. Treatment with JHU-083 was well tolerated (Suppl. Fig. 5d) and led to >80% inhibition of tumor growth after nine days (Fig. 3c), demonstrating potent anti-tumor efficacy, while also markedly prolonging survival (Fig. 3d). To assess the contribution of immune-mediated mechanisms on tumor growth (as seen in other cancer models13,14), we treated immunodeficient NSG mice harboring SQ BCH-ACC3 tumors with JHU-083, which confirmed the strong anti-tumor effect of Gln antagonism, albeit with somewhat lower efficacy than in immunocompetent mice (Fig. 3e-f). These results are consistent with JHU-083 having an effect on ACC tumor growth that is in large part immune cell-independent. Importantly, JHU-083 also significantly inhibited tumor growth in a xenograft model of human ACC (using NCI-H295R implants in NSG mice) (Fig. 3g), underscoring its preclinical efficacy.

248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270

Inhibition of GIn-dependent nucleotide biosynthesis drives DON’s anti-tumor effect in ACC

Broad-spectrum Gln antagonism by DON can target multiple Gln-dependent metabolic pathways essential for tumor cell proliferation and survival, including glutaminolysis, non-essential amino-acid biosynthesis, hexosamine biosynthesis, glutathione biosynthesis, and de novo purine and pyrimidine biosynthesis12,21,22. To identify the metabolic pathway(s) responsible for DON’s cytotoxicity in ACC, we performed rescue experiments with cultured ACC cells to functionally bypass specific Gln- dependent processes. Supplementation of ACC cell lines with a cell-permeable nucleoside (Nuc) cocktail was able to fully rescue cell viability in all four ACC cell lines (Fig. 4a, Suppl. Fig. 6a). In contrast, supplementation with (1) non-essential amino-acids (NEAA), (2) N-acetylglucosamine (GlcNAc) to support protein glycosylation pathways, (3) a-ketoglutarate (aKG) to restore TCA cycle function, or (4) antioxidants such as glutathione (GSH) or vitamin E (Trolox), failed to rescue the effect of DON on cell viability (Fig. 4a-b). These findings identify the blockade of GIn-fueled de novo nucleotide biosynthesis as the principal mechanism underpinning DON’s effect on ACC cell viability.

To assess the metabolic impact of Gln antagonism on ACC in vivo, we treated mice harboring SQ BCH-ACC3 tumors with JHU-083 and applied LC-MS-based targeted metabolomics 48 hours after treatment (Fig. 4c). This analysis revealed a marked depletion of purine metabolites (including guanine, GDP, dGMP, AMP, and hypoxanthine) and, to a lesser extent, pyrimidine intermediates such as cytidine and thymidine monophosphate (TMP) (Fig. 4d-g, Suppl. Fig. 6b). Consistent with on-target inhibition of Gln-dependent purine biosynthesis, JHU-083 treatment led to a marked accumulation

271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293

of xanthosine monophosphate (XMP), a substrate of the Gln-utilizing enzyme GMPS in the purine biosynthetic pathway. JHU-083 treatment also led to decreased glutathione levels, consistent with partial inhibition of Gln-dependent antioxidant synthesis, and a modest trend towards decreased hexosamine biosynthesis. However glutamate, a- ketoglutarate, and TCA cycle intermediates were not depleted by JHU-083, suggesting that mitochondrial metabolism is not broadly suppressed by JHU-083 (Suppl. Fig. 6b). Collectively, these findings validate de novo purine biosynthesis as the dominant Gln- dependent pathway disrupted by JHU-083 in vivo and confirm this metabolic axis as a critical vulnerability in ACC.

High expression of Gln-metabolising enzymes in de novo nucleotide biosynthesis

correlates with poor prognosis

Building on the findings that ACC is highly sensitive to Gln antagonism, we examined the gene expression of all known Gln-utilizing enzymes in human ACC using the TCGA dataset. Remarkably, high expression of four out of five key Gln-dependent enzymes directly involved in de novo purine and pyrimidine biosynthesis-PFAS, GMPS, CAD, and CTPS1-were significantly associated with worse survival (Fig. 4g-h). Interestingly, among 32 other cancer types in TCGA, only five (mesothelioma, low-grade glioma, hepatocellular carcinoma, sarcoma, and cutaneous melanoma) showed a comparable correlation between high expression of two or more Gln-metabolizing enzymes involved in nucleotide biosynthesis and poor prognosis (Suppl. Fig. 7a), highlighting the distinctive dependence of ACC on this metabolic axis, particularly in aggressive tumors with poor prognosis. A significant association with overall survival was

294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316

also observed for ASNS, encoding asparagine synthetase, and a trend was noted for NADSYN1 (coenzyme synthesis), but no differences were observed for enzymes involved in alternative Gln-related pathways such as GLS and GLS2 (glutaminolysis) or GFPT1 (hexosamine synthesis) (Suppl. Fig. 7b).

Inhibition of the DNA damage response synergizes with Gln antagonism to enhance anti-tumor effects in ACC

Nucleotide depletion is a well-established driver of replication stress, leading to replication fork stalling, accumulation of DNA damage, and ultimately cell death23. Given DON disrupts de novo nucleotide biosynthesis, we hypothesized that its anti-tumor activity in ACC might involve induction of replication stress. To test this, we assessed YH2AX levels (a marker of DNA double-strand breaks and replication-associated DNA damage24) in SQ BCH-ACC3 tumors treated with JHU-083 or vehicle, which revealed a striking increase in DNA double-strand breaks compared to controls (Fig. 5a). Consistent with this, we confirmed that DNA damage was also induced by JHU-083 in human NCI- H295R xenografts implanted in NSG mice (Fig. 5b).

Cells counteract the effect of replication stress through activation of the DNA Damage Response (DDR), including the ATR and ATM checkpoint kinase pathways, which promote cell survival by halting cell cycle progression, up-regulating nucleotide synthesis, and activating DNA repair programs24,25. Based on the observation that treatment with the DON pro-drug JHU-083 leads to DNA damage in ACC, we hypothesized that inhibiting ATR signaling could block this protective response and drive synthetic lethality in Gln-deprived cancer cells. To test this, we treated ACC cells with the

317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339

combination of DON and the ATR inhibitor elimusertib, which led to a dramatic increase in apoptotic cell death in both murine (BCH-ACC3) and human (NCI-H295R) ACC cell lines compared to either agent alone (Fig. 5c-d). We further evaluated the effects of combination treatment on cell viability across a matrix of dose combinations in all four ACC cell lines. Remarkably, we observed robust synergy in all p53-deficient cell lines (BCH-ACC3, NCI-H295R, CU-ACC2), and in CU-ACC1 (p53 wild-type) at higher elimusertib concentrations (Fig. 5e). Together, these findings demonstrate that DON- induced replication stress renders ACC cells highly dependent on ATR-mediated DDR for survival. Pharmacological co-inhibition of Gln metabolism and ATR signaling produces synergistic anti-tumor effects, particularly in p53-deficient ACC cells, and highlights a potentially promising combination treatment approach.

Untargeted serum metabolomics reveals dysregulation of nucleotide and amino- acid metabolism in patients with ACC

To validate the findings showing dysregulation of Gln-dependent metabolic pathways in ACC, we performed untargeted serum metabolomics on pre-operative samples from 54 patients with ACC and 291 age- and body mass index-matched patients with benign adrenocortical adenomas (ACA), prospectively collected across 10 centers in Europe and the U.S. (Fig. 6a-b). Differential abundance analysis using Benjamini- Hochberg correction revealed 437 metabolites that differed significantly between ACC and ACA (FDR < 0.05), of which 144 were elevated and 293 were reduced in the serum of patients with ACC (Fig. 6c, Suppl. Table 4). Polar metabolite pathway analysis revealed eight significantly dysregulated pathways (FDR<0.05), including pyrimidine

340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362

metabolism and several amino-acid metabolic pathways directly linked to Gln utilisation (‘Arginine and Proline Metabolism’, ‘Arginine Biosynthesis’, ‘Alanine, Aspartate and Glutamate metabolism’) (Fig. 6d and Suppl. Table 5). Importantly, there was a significant depletion of circulating Gln levels, suggestive of increased uptake of Gln by ACC tumors (Fig. 6e). Focusing on nucleotide metabolites, we found an increase in several purine and pyrimidine intermediates in the salvage/degradation pathways, including hypoxanthine, xanthosine, xanthine (Fig. 6f), cytidine, uridine, and deoxyuridine (Fig. 6g), reflecting enhanced nucleotide turnover consistent with the up-regulated nucleotide biosynthesis programs identified in tumor transcriptional profiles. As expected, there was a marked increase in several steroid conjugates (Suppl. Fig. 8 and Suppl. Table 4), in keeping with previous studies using targeted steroid profiling in patients with adrenal tumors26-28. Together, these serum metabolomic data complement the tumor-specific analyses and reinforce the presence of key changes in nucleotide and amino-acid metabolism.

Discussion

By integrating transcriptional profiling, targeted tissue metabolomics in genetically engineered mouse models, and untargeted serum metabolomics from a large patient cohort, our study provides cross-species validation of metabolic rewiring in ACC. With this approach, we identified Gln metabolism as a central, actionable vulnerability, based on transcriptional up-regulation of common pathways in mouse and human ACC, validation of changes in key metabolites and metabolic pathways in mouse ACC tissues, demonstration of ACC tumor sensitivity to Gln inhibition in vitro and in vivo, and validation

363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385

of key changes in circulating metabolites from patients with ACC. Additionally, we showed that addition of ATR inhibition to Gln antagonism results in potent synergy in multiple ACC cell lines in vitro, introducing a new treatment paradigm.

Consistent with this work, a previous pan-cancer analysis of the TCGA database used a gene signature consisting of Gln transporters and glutaminase isozymes and showed an association with poor prognosis that was stronger for ACC than for nearly any other cancer type29. We selected DON to target Gln metabolism due to its effects on multiple Gln-metabolising pathways (rather than focused targeting of glutaminolysis), and additionally taking into account the highly promising new therapeutic opportunities provided by the recent discovery of DON pro-drugs such as JHU-083. While clinical development of DON was previously hindered by the significant systemic toxicity, these new pro-drugs, which are preferentially activated in the tumor microenvironment through moiety cleavage, provide a new avenue to clinical translation for this metabolic treatment12,19. Indeed, a drug from this new class (DRP-104) is in Phase I-II trials for KEAP1-mutant lung cancer (NCT04471415). Moving forward, it will be critical to identify cancer types that are most likely to respond to Gln antagonism30,31 , as well as identify rational combination treatment approaches.

These data reveal that ACC cells are highly sensitive to Gln antagonism, in vitro, compared with other cancer cell lines13,14,22, and in vivo, using both WT and NSG SQ implant models with murine and human cell lines. DON theoretically has the capacity to inhibit all Gln-metabolizing enzymes, including glutaminase as well as transamidases that participate in amino-acid, nucleotide, co-factor and hexosamine biosynthesis12. Individual enzyme inhibition kinetics are incompletely described and the relative importance of

386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408

different Gln-utilizing pathways in the anti-tumor effects of DON is not well understood. The in vitro rescue experiments suggest that, in ACC, DON primarily affects cell proliferation and viability through inhibition of de novo purine and pyrimidine biosynthesis. Consistent with this, depletion of purine metabolites, in particular, was a prominent change in the metabolome of ACC tumors treated with JHU-083, with a lesser effect on pyrimidine metabolites. This asymmetric nucleotide response is suggestive of preferential functional inhibition of purine de novo glutamine-amidotransferases (PPAT, PFAS, GMPS) and/or with greater salvage buffering on the pyrimidine arm. These findings are consistent with recent studies assessing DON as a treatment in KEAP1-mutant lung cancer31 and prostate cancer32, which also showed that cancer cell-intrinsic sensitivity to Gln antagonism was contingent on inhibition of de novo nucleotide biosynthesis. Conversely, nucleoside substitution had no ability to rescue the effect of DON on pancreatic cancer cell viability, while substitution with the amino-acid asparagine did22. Previous work in a colon cancer cell line using shRNA-mediated knockdown of individual Gln-metabolising enzymes suggested that targeting of three Gln-metabolising enzymes involved in pyrimidine biosynthesis (Cad and Ctps1) and hexosamine biosynthesis (Gfpt1) had the strongest effect on cell viability in comparison to other Gln-metabolising enzymes13. Therefore, an important message of this study is that, although DON may be effective across a number of cancer types, the mechanisms underpinning cancer cell- intrinsic sensitivity to the drug are cancer type-dependent.

Metabolic dysregulation in the nucleotide salvage/degradation pathways was a prominent finding in both the tissue-level analysis in BPCre ACC tumors and in the serum- level metabolomic analysis in the cohort of patients with ACC. Pyrimidine

409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431

salvage/degradation metabolites were increased at both a tissue level (BPCre tumors) and at a serum level in patients with ACC. Conversely, purine metabolites were relatively depleted in BPCre tumors but increased in the serum of patients. This discrepancy may reflect accelerated release of these metabolites into the bloodstream from viable or necrotic ACC cells, or contributions from other tissues. Alternative, it is also possible that there may be differences in purine degradation metabolite handling between mouse and human ACC. The observed dysregulation of nucleotide salvage/degradation pathways may plausibly be related to the exceeding sensitivity of ACC to pharmacological inhibition of the alternative nucleotide biosynthetic pathway (de novo biosynthesis), which requires nitrogen contribution by Gln. Dedicated future studies employing metabolic tracing (e.g. by 13C and 15N-labeled Gln) will be required to probe the dynamics of nucleotide salvage/degradation in ACC.

A key translational question is how to build rational drug combinations that will enhance the efficacy of Gln antagonism. In this work, we demonstrate that Gln antagonism by JHU-083 induces DNA damage in ACC tumors in vivo. Interestingly, the same effect was very recently shown in an independent study using the DON pro-drug DRP-104 in prostate cancer33. Building on this finding, we hypothesized that concurrent inhibition of the DNA damage response through the ATR pathway inhibitor elimusertib could yield synergy with DON. Indeed, we confirmed that this new combination can elicit synergistic cytotoxicity across all p53-deficient ACC cell lines tested and, at higher doses, in a p53 WT line. Given that p53 deficiency sensitizes cells to replication stress34, this likely underlies the enhanced response observed in p53-mutant cell lines. These findings introduce a new treatment paradigm that may also be of value in other p53-deficient

432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454

cancers, or cancers in which the anti-tumor effect of DON is primarily driven by nucleotide depletion, such as prostate cancer or KEAP1-mutant lung cancer31-33.

The clinical study reported here is the largest metabolomic study to date in patients with adrenal tumors and provides a unique metabolomic dataset that will be valuable for future studies exploring new targeted therapeutic approaches or biomarker development. It is important to note that the control group for this clinical study was patients with benign adrenocortical adenomas rather than healthy adrenals, in order to allow exploration of diagnostic biomarkers in future work to distinguish benign from malignant adrenal tumors. A limitation of serum metabolomic analyses is that observed differences may, at least in part, reflect changes in systemic metabolism rather than tumor metabolism (due to a lack of specificity). Additionally, mass-to-charge ratios prior to fragmentation (MS1) were used for putative metabolite identification in most cases, rendering distinction between isomeric compounds difficult, with all putative metabolites being reported. Finally, while the two patient cohorts were well-matched for age and BMI, the cohort of patients with ACAs had a stronger female preponderance; however, no statistically significant differences were observed between male and female patients with ACC.

A previous study in human ACC using spatial metabolomics by MALDI mass spectrometry imaging revealed that high intra-tumoral metabolic heterogeneity is associated with poor prognosis35. We were not able to assess intra-tumoral heterogeneity in this work. Another limitation is that we did not use metabolic tracing to directly quantify changes in the intracellular metabolic fate of Gln in primary ACC tumors, or in SQ tumors treated with JHU-083. These in vivo targeted metabolomic data, however, clearly demonstrate that depletion of the pool of purine metabolites is the prominent metabolic

455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477

change in JHU-083-treated tumors in vivo. Interestingly, there was no apparent depletion of alpha-ketoglutarate or TCA cycle metabolites. This is in keeping with recent metabolic tracing studies in xenograft models of renal cell carcinoma, which showed unimpeded [U- 13C] Gln labeling of TCA cycle metabolites 36.

In summary, this study provides a systematic exploration of metabolic rewiring in ACC, leading to the discovery that this rare cancer is sensitive to targeting of Gln’s contribution to de novo nucleotide biosynthesis through treatment with Gln antagonists. With the recent development of DON pro-drugs with a broad therapeutic window, these robust pre-clinical findings are of high translational potential for this aggressive and treatment-resistant cancer. We also show that DNA damage induced by Gln antagonism is an important mechanism, serving as the basis for potent synergistic anti-tumor effects through combination with ATR pathway inhibition.

478 479 480 481 482 483 484 485 486 487 488 489 490

491 Methods

Genomic Methods

To characterize the molecular similarities between BPCre tumors and human ACC, we performed bulk RNA-seq on BPCre tumors (n = 7, 5 from female and 2 from male mice, aged 9-12 months) and control mouse adrenals without mutations (n = 6, 5 from female ASCre/+ mice and 1 from a C57BL/6J female mouse, aged 5-11 months). We also analyzed publicly available bulk RNA-seq data from human ACC tumors (TCGA-ACC, n=79)3 and normal human adrenals (GTEx, n=295)15. The GTEx and TCGA datasets were normalized and batch-corrected using the Variance Stabilizing Transformation (vst) function in the DESeq2 package37. Differential expression analysis was then performed with DESeq2 as well, restricted to a curated list of 2,754 metabolic genes and their mouse orthologs16. Genes with a Benjamini-Hochberg false discovery rate (FDR) < 0.05 were considered significantly differentially expressed.

For pathway interpretation, lists of significant genes were tested for KEGG over- representation using Enrichr38 (human and mouse separately), and cross-species conservation was assessed at the pathway level. For prespecified pathways (Purine metabolism, Pyrimidine metabolism, Glutathione metabolism, Amino sugar and nucleotide sugar metabolism), we generated heatmaps using variance-stabilized expression values of the significant genes in each pathway. A curated set of genes involved in de novo and salvage nucleotide biosynthesis or nucleotide metabolism was retrieved from the Reactome database via the Molecular Signatures Database (MSigDB, v2025.1), with cross-referencing to the KEGG database. All analyses and plots were produced in R, utilizing the DESeq2, ComplexHeatmap39, and ggplot2 packages40.

492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513

Mice

All animal procedures were approved by the Institutional Animal Care and Use Committees of Boston Children’s Hospital. Mouse strains used in this work were: C57BL/6J (Jackson Laboratory, #000664), ASCre/+ (Cyp11b2tm1.1(cre)Brlt41), Ctnnb1fl(ex3) (Ctnnb1tm1Mmt42), Trp53flox/flox43, and NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice (Jackson laboratory, strain #005557). Cross-breeding of transgenic mice led to the generation of the following adrenal-specific transgenic mice: (1) PCre (ASCre/+: Trp53flox/flox), P53 loss-of-function (LOF) model; (2) BCre (ASCre/+ :: Ctnnbflox(ex3)/+), ß-catenin gain-of-function (GOF) model and (3) BPCre (ASCre/+:Trp53flox/flox :: Ctnnbflox(ex3)/+), p53-LOF/ßcat-GOF bigenic model. Mouse sex is indicated in each figure legend or Methods. Littermate animals were studied whenever possible.

Drugs

6-diazo-5-oxo-L-norleucine (DON) was purchased from Sigma (D2141). JHU-083 (HY-122218) and elimusertib (HY-101566) were purchased from Medchem Express.

Cell Culture

The human ACC cell line NCI-H295R was purchased from the ATCC. Human ACC cell lines CU-ACC1 and CU-ACC2 were provided by Dr. Kisseljak-Vassiliades (University of Colorado)21. The murine cell line BCH-ACC3 was derived from a 538 mg primary ACC tumor with documented lung metastases in a male BPCre9 transgenic mouse. Before using BCH-ACC3 cells in tumor formation experiments, they were first assessed for GFP (as the BPCre mouse model includes the ROSAmT/mG reporter allele) and Cre expression (both known to be immunogenic44,45) and found that both genes had been lost during

514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536

serial SQ passaging in C57BL/6J mice during cell line derivation. NCI-H295R and BCH- ACC3 cells were cultured in Advanced DMEM/F12 (Thermo Fisher, 12634010) supplemented with 2% Nu serum (Corning, 355100), 1% ITS (Corning, 354350), 1x Glutamax (Thermo Fisher, 35050061) and 1% Penicillin-Streptomycin (ThermoFisher, 15140122). CU-ACC1 and CU-ACC2 cells were cultured in 3:1 v/v mix of F12 Nutrient Mixture (Ham) (Thermo Fisher, 11765-054) and DMEM-high glucose, pyruvate (Thermo Fisher, 11995-065), 10% fetal bovine serum (Thermo Fisher, A5256701), 0.4 ug/mL hydrocortisone (Sigma, H-0888), 5 µg/mL insulin (Sigma, 16634), 8.4 ng/ml cholera toxin (Sigma, C9903), 10 ng/ml epidermal growth factor (Invitrogen, PHG0311), 24 µg/mL adenine (Sigma, A2786), and 1% Penicillin-Streptomycin (ThermoFisher, 15140122), as previously described20. Cells were tested routinely for Mycoplasma infection by PCR.

In vitro assessment of cell viability and apoptosis

To assess the in vitro efficacy of DON against murine and human ACC cell lines, cells were plated in 96-well plates (Falcon) at different seeding densities depending on the growth kinetics of each cell line: 3,500 cells per well (BCH-ACC3), 6,000 cells per well (CU-ACC1 and CU-ACC2) and 8,000 cells/well (NCI-H295R). The next day, cell culture medium was removed and replaced by 200 ul treatment medium (DON, elimusertib, or combination) at varying concentrations as specified in individual figure legends. Treatment was applied for a total of 72 hours (BCH-ACC3) or 144 hours (NCI-H295R, CU-ACC1 and CU-ACC2) to allow two doubling times. For rescue experiments, cells were co-treated with individual metabolites as specified in figure legends, including 1x non- essential amino-acid cocktail (Thermo Fisher, 11140050), 1x nucleoside cocktail Embryomax (Millipore, ES-008-D), dimethyl-2-oxoglutarate 5.5 mM (Sigma, 349631),

537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559

GlcNAc 10 uM (Sigma, A8625), 2 mM reduced glutathione (Sigma, G1404), or Trolox 10 uM (Abcam, ab120747). At the end of the treatment course, relative viable cell numbers were measured using the CyQuant Direct Cell Proliferation Assay (Thermo Fisher, C35011) as per the manufacturer’s instructions. Baseline cancer cell line doubling times (in regular growth medium) were calculated by culturing each cell line for 72 hours and using the formula Doubling Time (hours) = (72 * ln(2)) / ln(final cell number / initial cell number). Synergy or antagonism was assessed using the SiCoDEA software and applying the Chou-Talalay Combination Index46,47. Cellular apoptosis was assessed using the Caspase-Glo 3/7 Assay kit (Promega, G8090), a luminescence-based assay measuring Caspase 3 and 7 activity in cell lysates following the manufacturer’s instructions, as described48. To account for differences in cell numbers present at the time of apoptosis measurements, caspase assay results were normalized to relative viable cell numbers, as assessed by the CyQuant Direct Cell Proliferation Assay run in wells treated in parallel.

In vitro clonogenic assays

For 2D clonogenic assays, cells were plated onto 6-well plates at a seeding density of 1,000 cells per well (BCH-ACC3, NCI-H295R, CU-ACC2) and 5,000 cells per well (CU- ACC1, as seeding 1,000 cells per well did not lead to colony development in this latter cell line). The next day, DON was added directly to wells at the desired concentrations, as specified in figure legends (medium was not changed to avoid washing out cells). Medium containing the drug was replenished every 3-4 days. The treatment course was ended once confluent colony formation was observed in vehicle-treated wells or at pre- specified endpoints (whichever was earlier). Pre-specified endpoints were 10 days for

560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582

BCH-ACC3 cells and 28 days for NCI-H295R, CU-ACC1 and CU-ACC2. At the end of the course, plates were washed with Phosphate-Buffered Saline (PBS) and fixed in 4% paraformaldehyde in PBS for 30 min. After washing with PBS again, wells were stained with crystal violet (0.1% crystal violet in 5% ethanol in water) for 30 min, then washed with water and left to dry inverted overnight. The next day, cells were imaged at 2x magnification using an Invitrogen EVOS FL 2 Auto microscope. Colonies comprising at least 50 cells were counted in Fiji using the ‘Analyse Particles’ tool.

JHU-083 treatment in vivo

For experiments involving BCH-ACC3 cells, 3x106 cells in 100 ul PBS/Matrigel (Corning) (1:1) were injected subcutaneously in the right flank of 2-month-old male C57BL/6J or NSG mice (Jackson). Tumors were measured by electronic caliper on day 6 and volume was calculated using the formula volume=length x (width)2/2. Mice were randomized to treatment groups and treatment was commenced on the same day. JHU- 083 was diluted in 100% ethanol and stored at -20℃. Working solution of JHU-083 in PBS with 2.5% ethanol was prepared every day and administered by oral gavage at a dose of 1 mg/kg/day (days 6-10) and 0.33 mg/kg thereafter, as previously described13. Tumor growth and mouse weight was monitored by electronic caliper every 72 hours. For tumor analysis by histology, mice were sacrificed and tumors were collected on day 15. JHU-083 or vehicle was administered 60-90 min before tumor collection.

For experiments involving NCI-H295R cells, 10x106 cells in 200 ul PBS/Matrigel (1:1) were injected subcutaneously in the right flank of 2-month-old female NSG mice (Jackson). Tumors were measured by electronic caliper on day 6 and volume was calculated using the formula volume=length x (width)2/2. Mice were randomized to

583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605

treatment groups and treatment was commenced on the same day. JHU-083 was diluted in 100% ethanol and stored at -20℃. Working solution of JHU-083 in PBS with 2.5% ethanol was prepared every day and administered by oral gavage at a dose of 1 mg/kg/day 6 days per week (days 6-29). Treatment dose was reduced to 0.5 mg/kg/day in animals with weight loss>7.5% from baseline and suspended in animals with weight loss >12.5% from baseline until weight improved (no mice lost >20% of their body weight or reached endpoints for euthanasia under this protocol). Tumor growth and mouse weight was monitored by electronic caliper every 72 hours. For both implant models, mice were sacrificed when tumor volume exceeded 2,000 mm3, tumors became ulcerated or necrotic, or when mice lost >20% weight from baseline.

Targeted tissue metabolomic analysis

For metabolomic analysis comparing primary ACC tumors in BPCre mice with adrenals from other transgenic mice and from wild-type mice, mice were euthanized and tumors were immediately dissected and a snap-frozen on dry ice. Because of the small weight of mouse adrenals, two pooled adrenals were used per sample, either deriving from the same mouse or from two separate mice of the same sex, age and genotype. For ACC tumors up to 10 mg of tissue was used per sample. For metabolomic analysis comparing SQ BCH-ACC3 implants treated with JHU-083 or vehicle, mice were euthanized after 48 h of treatment and tumors were immediately dissected and snap- frozen on dry ice. Up to 10 mg of tissue was used per sample. Samples were extracted with methanol: water: chloroform (1:1:1). The water-methanol phase containing polar metabolites was dried down and reconstituted with water/acetonitrile (1:1) with the volume normalized to tissue wet weight. The polar extract was separated on an iHILIC column (5

606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628

um, 150 mm x 2.1 mm I.D., Nest Group)49. The column was coupled to an Agilent 6546 LC/Q-TOF with an ESI source operated in negative and/or positive mode. The identity of each metabolite was confirmed by matching retention time and MS/MS fragmentation data to standard compounds and/or a database. Metabolite quantification was performed using MassHunter Profinder 10.0 software (Agilent). Data were analysed using Metaboanalyst 6.0. Features with >40% missing data for all groups were removed, and other missing variables were replaced by zero. Peak intensities were normalized to internal standards (D8-phenylalanine for polar metabolites) and to total ion count, log- transformed, and mean-centered before statistical analysis.

Paraffin Section Immunofluorescence

After fixation in 4% paraformaldehyde, harvested ACC implants were embedded in paraffin blocks. Paraffin sections were cut at 5 um thickness. Sections were rinsed in xylene, an ethanol gradient and then phosphate-buffered saline (PBS). Antigen retrieval was performed in 10 mM Sodium Citrate pH 6.0. Sections were blocked in 5% Normal Goat Serum (Sigma, G9023) in PBS for 1 hour at room temperature. Primary Phospho- Histone H2A.X (Ser139) Antibody (Cell Signaling, #2577) was diluted 1:200 in 5% normal goat serum in PBS and incubated on sections at 4℃ overnight. Slides were washed three times for 5 min in 0.1% Tween-20 in PBS. Secondary antibody (Alexa Fluor 594- conjugated goat anti-rabbit IgG antibody) was diluted 1:200 in PBS and incubated on sections at room temperature for 1-2 hours. For nuclear staining, DAPI (4’,6-diamidino-2- phenylindole) was added to the secondary antibody mixture at a final concentration of 1:1000. After three 5-min washes with 0.1% Tween-20 in PBS, slides were mounted with ProLong Gold Antifade Mountant (Thermo Fisher, P36930). Images were acquired using

629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651

a Nikon upright Eclipse 90i microscope and adjusted for brightness and contrast in ImageJ.

TCGA transcriptome analysis

GEPIA (online analysis tool) was used to access and process publicly available transcriptomic data from ACC and 32 other human cancers, collected in the context of The Cancer Genome Atlas (TCGA)50. Survival comparisons were performed by logrank (Mantel-cox) tests.

Clinical study design

All patients were prospectively recruited through the cross-sectional Evaluation of Urine Steroid Metabolomics in the Differential Diagnosis of Adrenocortical Tumors (EURINE-ACT) study, whose primary aim was to validate urine steroid profiling as a new diagnostic test for ACC26. The EURINE-ACT study was registered and approved by the European Network for the Study of Adrenal Tumors (ENSAT), in addition to local institute review board approvals. All patient recruitment centers had ethical approval for pseudonymized phenotype recording in the online ENSAT database and all participants of the EURINE-ACT study provided written informed consent prior to study participation.

Inclusion criteria included adult participants (age ≥18 years) with a newly identified adrenal mass of more than 1 cm diameter. Exclusion criteria were biochemical evidence of pheochromocytoma, pregnancy, lactation, and current or recent (<6 months) intake of drugs known to alter steroid synthesis or metabolism. Patients with an adrenal mass discovered during imaging for cancer staging or monitoring were also not eligible. Additional details on patient recruitment for EURINE-ACT are available here26. Centers that contributed serum samples for this study came from the following countries: Croatia

652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674

(Zagreb), France (Bordeaux), Greece (Athens), Germany (Berlin, Munich, Wurzburg, Dresden), Ireland (Galway), Norway (Bergen), Poland (Warsaw), United Kingdom (Birmingham), and the United States (Mayo Clinic). 54 patients with ACC were able to provide serum samples for metabolomic analysis. As a control group, we selected 291 patients with benign adrenocortical adenomas (ACAs) from the same study (EURINE- ACT), who were age- and BMI-matched for the cohort of ACCs. This included patients with non-functioning adrenocortical adenomas, as defined by clinical biochemistry, and patients with glucocorticoid excess, as defined by clinical biochemistry (1 mg dexamethasone suppression test showing cortisol suppression to <1.8 mg/dl).

Untargeted serum metabolome analysis

For each serum sample, stored at -80℃, 50uL aliquots were used for untargeted metabolome analysis. Two assays were applied to increase the coverage of metabolic features detected; polar (water-soluble) metabolites were analysed by hydrophilic interaction chromatography (HILIC) UHPLC-MS, and non-polar (lipid) metabolites by a C18 reversed-phase lipidomics UHPLC-MS. All samples were analyzed separately in positive and negative ion modes. Raw data files were deconvoluted using the XCMS software. Scatterplot smoothing signal correction was used for signal correction. The mean and relative standard deviation of metabolic feature concentrations were calculated against pooled quality control samples. Following data normalization by probabilistic quotient normalisation, metabolic features were annotated using the in-house BEAMS software and the mass spectral library mzCloud. All non-lipid metabolites were grouped into classes based on KEGG metabolic pathway involvement by applying pathway enrichment analysis in MetaboAnalyst. For lipid-based results, lipids were classified

675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697

according to LipidMaps ontologies manually. Data were visualised using R v4.2.2 (r- project.org) as implemented in R studio v. 2024.04.0 for MacOS. A more detailed description of the method can be found in the Supplemental Material.

Quantification and Statistical Analysis

All statistical analyses were performed and graphs prepared using Prism 10 (GraphPad), Metaboanalyst version 6.051 or R v4.2.2 (r-project.org) as implemented in R 04 studio v. 2024.04.0 for MacOS (for metabolomic data). The statistical test used to assess significance for each experiment is indicated in figure legends. To correct for multiple comparisons in the untargeted metabolomic analysis of human serum samples and targeted metabolomic analysis of murine tissue samples, FDR-corrected p values using the Benjamini-Hochberg method were used. To detect metabolites that showed step-wise progression from control adrenals to BPCre-Pre adrenals to ACC tumors, we used Pearson’s correlation coefficients through the Pattern Finder function on Metaboanalyst. For survival analyses, significance and hazards ratio were assessed by Log-rank (Mantel- Cox) tests. In bar graphs, data are presented as Mean + Standard Error of the Mean (SEM), unless otherwise stated. p<0.05 was considered statistically significant, unless otherwise stated. Data collection and analysis were not performed blind to the experimental conditions. Biorender was used for preparation of schematics.

Sex as a biological variable

RNA-seq analysis and targeted transcriptomics on BPCre tumors was performed on mice from both sexes. The murine ACC cell line was derived from a tumor that developed in a male mouse. Human cell lines were derived from female patients. Male C57BL/6J and NSG mice were used as hosts in the BCH-ACC3 implant model. Female

698 699 700 701 702 703

705 706 707 708 709 710

711 712 713 714 715 716 717 718 719 720

bioRxiv preprint doi: https://doi.org/10.1101/2025.09.28.674326; this version posted September 30, 2025. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

NSG mice were used as hosts for the NCI-H295R xenograft model. Both male and female 722 723 724

721

patients were included in the clinical study. The ACC cohort had a female preponderance (65%), in line with disease demographics. We were not able to fully match the ACA cohort for sex (91% female preponderance).

Acknowledgement

We thank members of the Breault and Haigis Labs and Dr. Esin Isik for helpful discussions. This work was supported by R01 DK123694 to D.T.B., a Michail Papamichail, MD, PhD, Postdoctoral Fellowship at Harvard Medical School to V.C., an Academy of Medical Sciences U.K. Clinical Lecturer Starter Grant SGL020\1018 to V.C., the Wellcome Trust (Investigator Award 209492/Z/17/Z to W.A. and Clinical Research Training Fellowship WT101671 to V.C.), the European Commission under the 7th Framework Program (FP7/2007-2013, grant agreement 259735, ENSAT-CANCER, to W.A., M.F., and F.B.), the UK Medical Research Council (MRC) through the construction of the Phenome Center Birmingham (MR/M009157/1) and core programme grant MC_UP_1605/15 to W.A., and a São Paulo Research Foundation (FAPESP) scholarship 2024/14165-6 to J.L.K. Biorender was used to generate schematics for this work

(https://BioRender.com/y9blqb5,	https://BioRender.com/w6r6cqp,
https://biorender.com/dlOpOrg,	https://biorender.com/k7qp0fh,
https://biorender.com/t50p462 ).

Author Contributions

V.C., K.S.B., W.A., W.B.D., M.C.H. and D.T.B. conceptualized the study and developed the methodology; V.C., K.S.B., J.Y., C.R., M.B., P.V., J.L.K., L.N., A.J. and C.W. performed the experiments; V.C., K.S.B., L.F.N., M.E.K., L.K., S.R. and T.G.P. performed formal data analysis; V.C. and M.E.K. visualised data for figures; V.C., A.P., M.P., M.M., I.D.P., D.D., M.Q., M.C.D., G.A.U., F.B., A.T., M.F., D.K., U.A., D.A.V., K.K.V. and I.B. provided resources (serum samples, cell lines); V.C., A.P., A.E.T. and I.B. provided

725 726 727 728 729 730 731 732

733 734 735 736 737 738 739 740 741 742 743 744 745 746 747

project administration for the clinical study; all authors interpreted the data; V.C., K.S.B. and D.T.B. wrote the original paper with editorial inputs from all authors; W.B.D., W.A., M.C.H. and D.T.B. provided the supervision; W.A., M.C.H. and D.T.B. acquired the funds.

Competing Interest Statement

The authors declare no competing interests.

748 749 750 751 752 753 754 755 756

757

References

1. Fassnacht M, Terzolo M, Allolio B, et al. Combination chemotherapy in advanced adrenocortical carcinoma. N Engl J Med 2012;366(23):2189-97. DOI: 10.1056/NEJMoa1200966.

2. Fassnacht M, Dekkers OM, Else T, et al. European Society of Endocrinology Clinical Practice Guidelines on the management of adrenocortical carcinoma in adults, in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol 2018;179(4):G1-G46. DOI: 10.1530/EJE-18-0608.

3. Zheng S, Cherniack AD, Dewal N, et al. Comprehensive Pan-Genomic Characterization of Adrenocortical Carcinoma. Cancer Cell 2016;29(5):723-736. DOI: 10.1016/j.ccell.2016.04.002.

4. Assie G, Letouze E, Fassnacht M, et al. Integrated genomic characterization of adrenocortical carcinoma. Nat Genet 2014;46(6):607-12. DOI: 10.1038/ng.2953.

5. Mohan DR, Lerario AM, Finco I, Hammer GD. New strategies for applying targeted therapies to adrenocortical carcinoma. Curr Opin Endocr Metab Res 2019;8:72-79. DOI: 10.1016/j.coemr.2019.07.006.

6. Pozdeyev N, Fishbein L, Gay LM, et al. Targeted genomic analysis of 364 adrenocortical carcinomas. Endocr Relat Cancer 2021;28(10):671-681. DOI: 10.1530/ERC-21-0040.

7. Borges KS, Pignatti E, Leng S, et al. Wnt/beta-catenin activation cooperates with loss of p53 to cause adrenocortical carcinoma in mice. Oncogene 2020;39(30):5282- 5291. DOI: 10.1038/s41388-020-1358-5.

8. Mariniello K, Pittaway JFH, Altieri B, et al. Dlk1 is a novel adrenocortical stem/progenitor cell marker that predicts malignancy in adrenocortical carcinoma. Cancer Commun (Lond) 2025;45(6):663-668. DOI: 10.1002/cac2.70012.

9. Mohan DR, Borges KS, Finco I, et al. beta-Catenin-Driven Differentiation Is a Tissue-Specific Epigenetic Vulnerability in Adrenal Cancer. Cancer Res 2023;83(13):2123-2141. DOI: 10.1158/0008-5472.CAN-22-2712.

10. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144(5):646-74. DOI: 10.1016/j.cell.2011.02.013.

11. Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov 2022;12(1):31-46. DOI: 10.1158/2159-8290.CD-21-1059.

12. Lemberg KM, Vornov JJ, Rais R, Slusher BS. We’re Not “DON” Yet: Optimal Dosing and Prodrug Delivery of 6-Diazo-5-oxo-L-norleucine. Mol Cancer Ther 2018;17(9):1824-1832. DOI: 10.1158/1535-7163.MCT-17-1148.

758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791

13. Leone RD, Zhao L, Englert JM, et al. Glutamine blockade induces divergent metabolic programs to overcome tumor immune evasion. Science 2019;366(6468):1013-1021. DOI: 10.1126/science.aav2588.

14. Oh MH, Sun IH, Zhao L, et al. Targeting glutamine metabolism enhances tumor- specific immunity by modulating suppressive myeloid cells. J Clin Invest 2020;130(7):3865-3884. DOI: 10.1172/JCI131859.

15. Consortium GT. The Genotype-Tissue Expression (GTEx) project. Nat Genet 2013;45(6):580-5. DOI: 10.1038/ng.2653.

16. Possemato R, Marks KM, Shaul YD, et al. Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 2011;476(7360):346-50. DOI: 10.1038/nature10350.

17. Panwalkar P, Tamrazi B, Dang D, et al. Targeting integrated epigenetic and metabolic pathways in lethal childhood PFA ependymomas. Sci Transl Med 2021;13(614):eabc0497. DOI: 10.1126/scitranslmed.abc0497.

18. Dang D, Deogharkar A, McKolay J, et al. Isocitrate dehydrogenase 1 primes group-3 medulloblastomas for cuproptosis. Cancer Cell 2025;43(6):1159-1174 e8. DOI: 10.1016/j.ccell.2025.04.013.

19. DeBerardinis RJ. Tumor Microenvironment, Metabolism, and Immunotherapy. N Engl J Med 2020;382(9):869-871. DOI: 10.1056/NEJMcibr1914890.

20. Kiseljak-Vassiliades K, Zhang Y, Bagby SM, et al. Development of new preclinical models to advance adrenocortical carcinoma research. Endocr Relat Cancer 2018;25(4):437-451. DOI: 10.1530/ERC-17-0447.

21. Encarnacion-Rosado J, Sohn ASW, Biancur DE, et al. Targeting pancreatic cancer metabolic dependencies through glutamine antagonism. Nat Cancer 2024;5(1):85-99. DOI: 10.1038/s43018-023-00647-3.

22. Recouvreux MV, Grenier SF, Zhang Y, et al. Glutamine mimicry suppresses tumor progression through asparagine metabolism in pancreatic ductal adenocarcinoma. Nat Cancer 2024;5(1):100-113. DOI: 10.1038/s43018-023-00649-1.

23. Mullen NJ, Singh PK. Nucleotide metabolism: a pan-cancer metabolic dependency. Nat Rev Cancer 2023;23(5):275-294. DOI: 10.1038/s41568-023-00557-7.

24. Saxena S, Zou L. Hallmarks of DNA replication stress. Mol Cell 2022;82(12):2298-2314. DOI: 10.1016/j.molcel.2022.05.004.

25. Qiu Z, Oleinick NL, Zhang J. ATR/CHK1 inhibitors and cancer therapy. Radiother Oncol 2018;126(3):450-464. DOI: 10.1016/j.radonc.2017.09.043.

792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825

26. Bancos I, Taylor AE, Chortis V, et al. Urine steroid metabolomics for the differential diagnosis of adrenal incidentalomas in the EURINE-ACT study: a prospective test validation study. Lancet Diabetes Endocrinol 2020;8(9):773-781. DOI: 10.1016/S2213-8587(20)30218-7.

27. Arlt W, Biehl M, Taylor AE, et al. Urine steroid metabolomics as a biomarker tool for detecting malignancy in adrenal tumors. J Clin Endocrinol Metab 2011;96(12):3775- 84. DOI: 10.1210/jc.2011-1565.

28. Yu K, Athimulam S, Saini J, et al. Serum Steroid Profiling in the Diagnosis of Adrenocortical Carcinoma: A Prospective Cohort Study. J Clin Endocrinol Metab 2025;110(4):1177-1186. DOI: 10.1210/clinem/dgae604.

29. Zou J, Du K, Li S, et al. Glutamine Metabolism Regulators Associated with Cancer Development and the Tumor Microenvironment: A Pan-Cancer Multi-Omics Analysis. Genes (Basel) 2021;12(9). DOI: 10.3390/genes12091305.

30. Yokoyama Y, Estok TM, Wild R. Sirpiglenastat (DRP-104) Induces Antitumor Efficacy through Direct, Broad Antagonism of Glutamine Metabolism and Stimulation of the Innate and Adaptive Immune Systems. Mol Cancer Ther 2022;21(10):1561-1572. DOI: 10.1158/1535-7163.MCT-22-0282.

31. Pillai R, LeBoeuf SE, Hao Y, et al. Glutamine antagonist DRP-104 suppresses tumor growth and enhances response to checkpoint blockade in KEAP1 mutant lung cancer. Sci Adv 2024;10(13):eadm9859. DOI: 10.1126/sciadv.adm9859.

32. Moon D, Hauck JS, Jiang X, et al. Targeting glutamine dependence with DRP- 104 inhibits proliferation and tumor growth of castration-resistant prostate cancer. Prostate 2024;84(4):349-357. DOI: 10.1002/pros.24654.

33. Yu J, Jin C, Su C, et al. Resilience and vulnerabilities of tumor cells under purine shortage stress. Clin Cancer Res 2025. DOI: 10.1158/1078-0432.CCR-25-1667.

34. Dembitz V, Tomic B, Kodvanj I, Simon JA, Bedalov A, Visnjic D. The ribonucleoside AICAr induces differentiation of myeloid leukemia by activating the ATR/Chk1 via pyrimidine depletion. J Biol Chem 2019;294(42):15257-15270. DOI: 10.1074/jbc.RA119.009396.

35. Wang Q, Sun N, Meixner R, et al. Metabolic heterogeneity in adrenocortical carcinoma impacts patient outcomes. JCI Insight 2023;8(16). DOI: 10.1172/jci.insight.167007.

851 852 853 854 855 856 857 858 859 860

36. Kaushik AK, Tarangelo A, Boroughs LK, et al. In vivo characterization of glutamine metabolism identifies therapeutic targets in clear cell renal cell carcinoma. Sci Adv 2022;8(50):eabp8293. DOI: 10.1126/sciadv.abp8293.

826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850

861 862 863

37. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15(12):550. DOI: 10.1186/s13059-014-0550-8.

38. Kuleshov MV, Jones MR, Rouillard AD, et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res 2016;44(W1):W90-7. DOI: 10.1093/nar/gkw377.

864 865 866

39. Gu Z, Eils R, Schlesner M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics 2016;32(18):2847-9. DOI: 10.1093/bioinformatics/btw313.

40. Wickham H. ggplot2 - Elegant Graphics for Data Analysis. Second ed. New York: Springer-Verlag, 2016.

41. Freedman BD, Kempna PB, Carlone DL, et al. Adrenocortical zonation results from lineage conversion of differentiated zona glomerulosa cells. Dev Cell 2013;26(6):666-673. DOI: 10.1016/j.devcel.2013.07.016.

42. Harada N, Tamai Y, Ishikawa T, et al. Intestinal polyposis in mice with a dominant stable mutation of the beta-catenin gene. EMBO J 1999;18(21):5931-42. DOI: 10.1093/emboj/18.21.5931.

43. Marino S, Vooijs M, van Der Gulden H, Jonkers J, Berns A. Induction of medulloblastomas in p53-null mutant mice by somatic inactivation of Rb in the external granular layer cells of the cerebellum. Genes Dev 2000;14(8):994-1004. (https://www.ncbi.nlm.nih.gov/pubmed/10783170).

44. Schultheiss C, Binder M. Overcoming unintended immunogenicity in immunocompetent mouse models of metastasis: the case of GFP. Signal Transduct Target Ther 2022;7(1):68. DOI: 10.1038/s41392-022-00929-9.

45. Dubrot J, Lane-Reticker SK, Kessler EA, et al. In vivo screens using a selective CRISPR antigen removal lentiviral vector system reveal immune dependencies in renal cell carcinoma. Immunity 2021;54(3):571-585 e6. DOI: 10.1016/j.immuni.2021.01.001.

46. Spinozzi G, Tini V, Ferrari A, et al. SiCoDEA: A Simple, Fast and Complete App for Analyzing the Effect of Individual Drugs and Their Combinations. Biomolecules 2022;12(7). DOI: 10.3390/biom12070904.

47. Chou TC. Drug combination studies and their synergy quantification using the Chou-Talalay method. Cancer Res 2010;70(2):440-6. DOI: 10.1158/0008-5472.CAN- 09-1947.

48. Chortis V, Taylor AE, Doig CL, et al. Nicotinamide Nucleotide Transhydrogenase as a Novel Treatment Target in Adrenocortical Carcinoma. Endocrinology 2018; 159(8):2836-2849. DOI: 10.1210/en.2018-00014.

867 868 869 870 871 872 873 874 875 876 877 878 879 880 881

882 883 884 885 886 887 888 889 890 891 892 893 894 895 896

49. Yao CH, Park JS, Kurmi K, et al. Uncoupled glycerol-3-phosphate shuttle in kidney cancer reveals that cytosolic GPD is essential to support lipid synthesis. Mol Cell 2023;83(8):1340-1349 e7. DOI: 10.1016/j.molcel.2023.03.023.

50. Tang Z, Li C, Kang B, Gao G, Li C, Zhang Z. GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses. Nucleic Acids Res 2017;45(W1):W98-W102. DOI: 10.1093/nar/gkx247.

51. . Pang Z, Lu Y, Zhou G, et al. MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Res 2024;52(W1):W398-W406. DOI: 10.1093/nar/gkae253.

897 898 899 900 901 902 903 904 905 906 907

908 Fig. 1

healthy adrenals

ACC

Up-regulated pathways

RNA-seq

BPCre ACC (mouse)

Down-regulated pathways BPCre ACC (mouse)

Human ACC

n=6

n=7

110

n=295 (GTEx)

n=79 (TCGA)

Glutamine

-log10 p value

Lipids + acetyl-CoA Nucleotides

SLC1A5

Glutamine

Nucleotides

GLS/GLS2

Hexosamines

0-2

Asp

Citrate

Glutamate

Glutathione

2-5

5-10

Malate 1 NADPH

GLUD

ATS

OAA

TCA cycle

NADH NH,

Amino Acids

>10

a-KG + a-KG

Mitochondrion

Cytosol

C		BPCre	Human
C		ACC	ACC
Pyrimidine metabolism
	Drug metabolism
Purine metabolism
Glycosaminoglycan biosynthesis
Inositol phosphate metabolism
Phosphatidylinositol signaling system
Cysteine and methionine metabolism
Glycerophospholipid metabolism
	Glycine, serine and threonine metabolism
N-Glycan biosynthesis
Glutathione metabolism
One carbon pool by folate
Fructose and mannose metabolism
Amino sugar and nucleotide sugar metabolism
	Porphyrin and chlorophyll metabolism

Fig. 1: Cross-species transcriptomic analysis of metabolic reprogramming in ACC. a Schematic overview of cross-species transcriptomic analysis of metabolic pathways in ACC. b Venn diagrams showing overlap between significantly (FDR<0.05) up-regulated and down-regulated metabolic pathways in mouse (BPCre) and human ACC tumors. Numbers refer to dysregulated pathways. c Top 15 most significantly up-regulated pathways in mouse ACC; all pathways are also significantly up-regulated in human ACC (TCGA/GTEx samples). d Schematic overview of GIn metabolism. a-KG: a-ketoglutarate, Asp: aspartate, ATs: aminotransferases, NADH: reduced nicotinamide adenine dinucleotide, OAA: oxaloacetate.

909 910 911 912 913 914 915 916 917 918 919

920 Fig. 2

Normal adrenals

Pre-malignant transformation

ACC

Control

BPCre-Pre

BPCre-ACC

BPCre-ACC vs Control

Decreased in ACC

Increased in ACC

Creatine Phosphate

BPCre-ACC

☐

Males

Females

Glutathione

XMP

PC 2 (14.7%)

-log10(p value)

Ophthalmic acid NADH

Ornithine

Control

S-adenosyl-L-homocysteine

L-Aspartate

3-Ureidopropionic acid

Cytidine

BPCre-Pre

Xanthine Xanthosine

Creatine

CAMP

UDP-Glucose

Lysine L-Cysteine

D-Glucose

dTTP

UDP-GIGNAC

UDP-GaINAc

NAD

Galactose 1-P Glucose 1-P

L-Proline

Uric acid

-Glutamate

CDP-Ethanolamine

-5

Thymidineo

-10

-5

PC 1 (28.3%)

-4

-2

log2(FC)

Metabolites correlating with progression Control→ BPCre-Pre →BPCre-ACC

Negative correlation

Positive correlation

Glutathione

CDP-choline

O-Phosphorylethanolamine

Ophthalmic acid

Cytidine

L-Aspartate

NADH

GABA

Ornithine

Cyclic AMP

2-Aminoisobutyric acid

UDP-N-acetylglucosamine

L-Cysteine

UDP-GaINAc

Lysine

D-Fructose 1,6-bisphosphate

D-Galacturonate

S-Adenosyl-L-homocysteine

Glycerol

XMP

Creatine

Galactarate

Citramalate

Saccharate

Creatine Phosphate

-1.0

-0.5

0.0

0.5

1.0

Pearson’s Coefficient

921

Fig. 2: Targeted tissue metabolomics in BPCre tumors. a Schematic of malignant transformation in BPCre mice. b Principal Component Analysis (PCA) comparing the polar metabolome of control adrenals from C57BL/6J mice (7-11 months, n=10), adrenals from young BPCre mice pre-tumor formation (BPCre-Pre, 4 months, n=4), and ACC tumors from older BPCre mice (BPCre-ACC, 8-13 months, n=8). Circles represent female mice; squares represent male mice. For control and BPCre-Pre adrenals, each sample represents tissue from two adrenals taken from either the same mouse or from two mice of the same genotype and sex. c Polar metabolite changes comparing BPCre-ACC vs control adrenal tissue selected by Volcano plot with fold change threshold (x) 1.2 and t- ☒ tests threshold (y) 0.05 (FDR-corrected p values). Dashed lines indicate statistical thresholds. d Top 20 metabolites showing progressive changes from control adrenals → BPCre-Pre adrenals → BPCre-ACC.

922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938

939 940 941 942 943 944 945 946 947

Fig. 3

NCI-H295R

BCH-ACC3

1.2-

1.2

NCI-H295R

Cell viability 72h normalised to Vehicle

1.0-

Cell viability 144h

normalised to Vehicle

1.0

600-

60-

0.8-

0.8

Colonies per well

0.6

400-

0.4

IC50=0.03 µM

0.4

IC50=1.42 µM

200-

20-

0.2-

0.2

₣

0.001

0.01

0.1

DON (UM)

0.1

DON (UM)

100

DON (UM) 0

0.05

0.2

0.5

DON (UM) 0

2.5

CU-ACC1

CU-ACC2

1.2-

CU-ACC1

1.2

CU-ACC2

Cell viability 144h normalised to Vehicle

1.0

Cell viability 144h normalised to Vehicle

1.0

250

0.8

Colonies per well

200

Colonies per well

0.6

150

0.4

IC50=3.57 µM

0.4

IC50=4.07 μΜ

100

10.

0.2

0.01

0.1

100

0.01

0.1

DON (UM)

100

2.5

DON (UM)

DON (UM) 0

2.5

BCH-ACC3 implants (C57BL/6 mice)

3000

100

Tumor volume (mm3)

Vehicle

2000

JHU-083

— JHU-083

Survival (%)

BCH-ACC3 cell injection

Treatment start

JHU-083

Treatment end

1000

Day 0

Day 6

Day 15

Days post implant Treatment end

Days post implant

Treatment start

BCH-ACC3 implants (NSG mice)

Tumor volume with JHU-083 NSG vs C57BL/6

Tumor volume (mm3)

3000

Vehicle

0.6

JHU-083

Tumor volume normalised to Vehicle

2000

0.4-

1000

0.2

…

0.0

NSG

C57BL/6

Days post impant

NCI-H295R implants (NSG mice)

500

Vehicle

Tumor volume (mm3)

400

Vehicle

300

JHU-083

NCI-H295R cell injection

Treatment start

JHU-083

Treatment end

200

100

Day 0

Day 6

Day 29

Days post implant

Fig. 3: Gln antagonism is highly effective against ACC in vitro and in vivo. a Dose- response curves for four ACC cell lines treated with DON in vitro. Data are displayed as mean + S.E.M of at least three independent experiments (pooled data) with 5 replicates per dose per experiment. In each experiment, cell viability was normalized to the viability of vehicle-treated cells. b Clonogenic assays in four ACC cell lines treated with DON in vitro. N=3 replicates per dose; * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. One-way ANOVA with post-hoc Tukey’s tests. Data representative of three independent experiments. c Schematic and tumor growth curve from subcutaneous (SQ) BCH-ACC3 implants in male C57BL/6J mice. Mice were treated with JHU-083 1 mg/kg (or vehicle) from day 6-10 post-implantation and 0.33 mg/kg from days 11-15. Tumor inhibition is shown until first sacrifice (day 15). N=9,13; **** p<0.0001 (t-test at the end of the treatment course). Data are representative of three independent experiments. d Survival curves post SQ implantation of BCH-ACC3 cells in male C57BL/6J mice. JHU-083 was administered at 1 mg/kg from day 6-10 post-implantation and 0.33 mg/kg from days 11- 26. Results were analysed by log-rank (Mantel-Cox) test. Data are representative of two independent experiments. e Tumor growth curve from NCI-H295R cell implants in male immunodeficient NSG mice. 1 mg/kg (or vehicle) from day 6-10 post-implantation and 0.33 mg/kg from days 11-15. N=7,8; ** p<0.01 (t-test at the end of the treatment course). f Tumor volume on day 15 post BCH-ACC3 implant in male C57BL/6J and NSG mice treated in parallel. Values have been normalized to vehicle-treated mice of the same strain. N=8-13; *** p<0.001 (t-test). Data are representative of three independent experiments. g Schematic and tumor growth curve from NCI-H295R cell implants in female immunodeficient NSG mice. JHU-083 was administered six days per week for three weeks (days 6-29 post implant). N=13-14 per group. ** p<0.01 (t-test at the end of the treatment course). Data are representative of two independent experiments.

950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976

977

Fig. 4

BCH-ACC3

NCI-H295R

BCH-ACC3

NCI-H295R

Cell viability 72h normalised to Vehicle

1.5

Cell viability 144h normalised to Vehicle

Cell viability 72h normalised to Vehicle

1.5

Cell viability 144h normalised to Vehicle

1.5

1.0

1.0-

0.5

0.5-

0.0

DON

Nuc +

NEAA +

GlcNAc +

a-KG +

DON

Nuc +

NEAA +

GIcNAc +

a -KG +

DON

GSH +

Trolox +

DON

GSH +

Trolox +

C57BL/6 mice SQ BCH-ACC3 implants

JHU-083 or Vehicle

Decreased with JHU-083

Increased with JHU-083

Tissue metabolomics (LC-MS)

48 h

N-acetylserine

Vehicle

Galactarate

N-formylglycine

TMP

D-glucosamine

Saccharate

-log10(p value)

Guanine®

GDP

PC 2 19.9%

N-Forminino- Cystathionine

-glutamate

N,N-Dimethylarginine Hypoxanthine

Oxalate

dGMP

AMP

Hexitol

D-Glycerate

-2

Xanthine

JHU-083

UDP-D-glucuronate Inosine

-Glutathione L-Proline

D-Malate

Creatine P

PEP

-Aspartate

Urocanate

XMP

tidine

alpha-KG

Raffinose

-4

-6

-15

-10

-5

PC 1 47.3%

-2

-1

log2(FC)

TMP

N-acetylserine Cystathionine Hypoxanthine

log2(FC)

Purine Biosynthesis PRPP

Guanine

V,N-Dimethylarginine -glucosamine

class

JHU-083 Veh

Pyrimidine Biosynthesis

-Glutathione

PPAT

Gin

GDP

Glu

N-Forminino-L-glutamate L-Proline

PRA

Asp+HCO3+ATP

Gin

AMP

CAD

dGMP

De novo biosynthesis

FGAR

PFAS

Gin

D-Malate Phosphoenol Pyruvate

De novo biosynthesis

Glu Dihydroorotate

-1

FGAM

Glu

XMP

Orotate

L-Aspartate

-2

Glu

Gin

Oxalate

N-Formylglycine

GDP

GMP

XMP

IMP

AMP

ADP

Salvage/ Degradation

dTMP ++ UMP +

CTP

D-Glycerate

Salvage/ Degradation

GMPS

CTPS

Jrocanate

guanosine

xanthosine

inosine

adenosine

thymidine

uridine

cytidine

Saccharate Galactarate

3-ureido-

Creatine Phosphate

guanine

xanthine hypoxanthine adenine

uracil

propionate

Hexitol

De novo purine biosynthesis

De novo pyrimidine biosynthesis

PFAS Overall Survival

GMPS Overall Survival

PPAT Overall Survival

CAD Overall Survival

CTPS1 Overall Survival

1.0

0.8

0.6

Low

Survival

0.4

High

0.2

High

0.0

Logrank p=0.046

Logrank p=0.00036

Logrank p=0.11

Logrank p=0.0017

Logrank p=0.024

100

150

100

150

100

Months

150

100

Months

150

100

Months

150

978

Fig. 4: Gln-fueled de novo nucleotide biosynthesis as a metabolic vulnerability in ACC. a BCH-ACC3 and NCI-H295R cell viability normalized to Vehicle (DON-) after treatment with DON (DON +), alone or supplemented with the indicated metabolites. DON was administered at 50 nM for 72h for BCH-ACC3 and 5 UM for 144h for NCI-H295R cells. Nuc: Embryomax nucleoside cocktail 1x, NEAA: non-essential amino-acid cocktail (0.1 mM), NAG: n-acetyl-glucosamine (10 µM), a-KG: dimethyl-2-oxo-glutarate, cell- permeable form of a-ketoglutarate, 5.5 mM. Data representative of three independent experiments (5-8 technical replicates per experiment). **** p<0.0001, one-way ANOVA followed by Tukey’s test for multiple comparisons between DON monotherapy and each other group. b BCH-ACC3 (left) or NCI-H295R (right) viability normalized to Vehicle (DON-) after treatment with DON (DON +), alone or with co-treatment with antioxidants (reduced glutathione (GSH) 2 mM or Trolox 10 uM). DON was administered at 50 nM for 72h for BCH-ACC3 and 5 UM for 144h for NCI-H295R. Data representative of three independent experiments (4-8 technical replicates per experiment). **** p<0.0001, one- way ANOVA followed by Tukey’s test for multiple comparisons between DON monotherapy and each other group. c Schematic of JHU-083 treatment of BCH-ACC3 implants for metabolomic analysis. d Principal Component Analysis (PCA) comparing the polar metabolome of BCH-ACC3 implants in C57BL/6 mice after 48h treatment with JHU- 083 (1 mg/kg once daily, n=5) or vehicle (n=4). e Polar metabolite changes comparing JHU-083 vs vehicle-treated tumors by Volcano plot with fold change threshold (x) >1.2 and t-tests threshold (y) p<0.05. Dashed lines indicate statistical thresholds. Purine metabolite names are highlighted in red font; pyrimidine metabolite names are highlighted in blue font. f Heat-map of the top 25 differential polar metabolites in JHU-083 tumors vs vehicle. Purine metabolite names are highlighted in red font. Pyrimidine metabolite names are highlighted in blue font. g Schematic summarising the two pathways contributing to purine (left) and pyrimidine (right) nucleotide biosynthesis: de novo biosynthesis (Gln- dependent, red arrows) and salvage/degradation pathways (blue arrows). The five enzymes that use Gln as a substrate are indicated in circles. h Survival curves comparing overall survival in patients with high (top 50%) versus low (bottom 50%) levels of expression of indicated Gln-metabolizing enzymes, based on The Cancer Genome Atlas (TCGA). Significance and hazards ratio were assessed by Log-rank (Mantel-Cox) tests.

979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009

1010 Fig. 5

BCH-ACC3 implants

DAPI

YH2AX

DAPI YH2AX

BCH-ACC3

Vehicle

JHU-083

Caspase 3/7 activity normalised to vehicle

DON

Eli

NCI-H295R implants

DAPI

YH2AX

DAPI YH2AX

NCI-H295R

Vehicle

Caspase 3/7 activity normalised to vehicle

JHU-083

DON

Eli

p53 LOF

p53 WT

BCH-ACC3

NCI-H295R

CU-ACC2

CU-ACC1 Synergy Antagonism

Synergy

Antagonism

Synergy

Antagonism

Synergy Antagonism

-1.5

-1 0

-3

-1

Log CI

-0.33

-0.46

-1.29

-0.75

400

0.04

-0.11

-0.32

-0.27

-0.58

-1.46

-3.15

-3.02

100

0.13

-0.02

-1.02

-0.26

Elimusertib (nM)

-1.06

-1.10

-0.88

-1.13

200

-0.12

-0.60

-0.65

-0.31

-0.63

-1.46

-2.58

-2.38

0.26

-0.30

-0.15

-0.98

-0.97

-1.30

-1.56

100

-0.66

-1.11

-1.73

-1.89

12.5

-0.51

-0.78

-2.69

-2.25

0.39

0.45

0.24

-0.21

12.5

-1.59

-1.47

-1.62

-1.74

-0.98

-1.40

-1.65

-1.99

6.25

-1.30

-1.40

-2.25

-1.65

12.5

0.08

0.38

-0.48

6.25

0.013

0.025

0.05

0.1

1.25

2.5

1.25

2.5

1.25

2.5

DON (UM)

Fig. 5: ATR inhibition exhibits synergistic cytotoxicity with Gln antagonism. a Representative phosphorylated histone H2A.X (y-H2AX) staining (grey) of SQ BCH- ACC3 implants in C57BL/6J mice treated with JHU-083 or vehicle for nine days (days 6- 15 post-implant). Blue, DAPI. Scale bars: 50 uM. b Representative phosphorylated histone H2A.X (y-H2AX) staining (grey) of SQ NCI-H295R implants in NSG mice treated with JHU-083 or vehicle (days 6-29 post-implant). Blue, DAPI. Scale bars: 50 uM. c,d Assessment of DON and Elimusertib-induced apoptosis (luminescence-based quantification of activated caspase 3/7) in BCH-ACC3 (c) and NCI-H295R (d) cells, treated for 48 hours (BCH-ACC3) and 96 hours (NCI-H295R). Data are representative of three independent experiments (3-4 technical replicates per experiment). For both cell lines, results were analysed by two-way ANOVA, showing very high statistical significance (p<0.0001) for the effect of DON and Elimusertib on apoptosis, as well as for their interaction (consistent with synergy). Post-hoc pairwise comparisons (Tukey’s) are shown on the graphs. * p<0.05, **** p<0.0001. e Synergistic effects of DON and Elimusertib on ACC cell viability. ACC cell lines were treated with varying doses of DON and Elimusertib for 72 h (BCH-ACC3) or 144 h (NCI-H295R, CU-ACC1 and CU-ACC2), followed by assessment of cell viability. BCH-ACC3, NCI-H295R and CU-ACC2: data are representative of three independent experiments (4-8 technical replicates per experiment). CU-ACC1: Data represent combination (median viabilities) of three independent experiments. Synergy or antagonism was assessed using the SiCoDEA software and applying the Chou-Talalay method Combination Index (CI). Blue color indicates synergy; red color indicates antagonism. Numbers represent log (Combination Index).

1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034

1035 Fig. 6

Serum samples ACC (n=54) ACA (n=291)

UHPLC-MS

Untargeted metabolomics

100

Retention time

	ACC (N=54)	ACA (N=291)
Gender Number (% Total) Male Female	19 (35.2%) 35 (64.8%)	26 (8.9%) 265 (91.1%)
Age (years) Median	54.5 43.5-61.8	60
IQR	54.5 43.5-61.8	49.5-67.5
BMI (kg/m2)
Median	28.0	29
IQR	27.5-32.4	25.4-33.3
Clinical biochemistry Number (% Total)
No hormone excess	11 (20.4%)	104 (35.7%)
Isolated glucocorticoid excess	7 (13.0%)	187 (64.3%)
Isolated androgen/estrogen excess	5 (9.3%)	0 (0%)
Isolated mineralocorticoid excess	0 (0%)	0 (0%)
Mixed hormone excess	28 (51.9%)	0 (0%)
Data not available	3 (5.6%)	0 (0%)
Glucocorticoid (GC) status Number (% Total)
No GC excess	17 (31.5%)	104 (35.7%)
Mild autonomous GC secretion	14 (25.9%)	140 (48.1%)
Cushing's syndrome	20 (37.0%)	47 (16.2%)
Unknown	3 (5.6%)	0 (0%)
ENSAT Stage N (% Total) I-III (localized disease)	42 (77.8%)
ENSAT Stage N (% Total) I-III (localized disease)	42 (77.8%)
IV (metastatic disease)	12 (22.2%)

ACC vs ACA

Decreased in ACC

Increased in ACC

-log10(p-value)

-1

log2(FC)

Pathway Analysis

Pathway	Hits/Total	FDR
Arginine and proline metabolism	11/36	0.0001
Valine, leucine and isoleucine biosynthesis	5/8	0.0009
Phenylalanine metabolism	4/8	0.0097
Arginine biosynthesis	5/14	0.0097
Phenylalanine, tyrosine and tryptophan biosynthesis	3/4	0.0097
Pyrimidine metabolism	8/39	0.0116
Glycine, serine and threonine metabolism	7/33	0.0173
Alanine, aspartate and glutamate metabolism	6/28	0.0282

Purine salvage/degradation metabolites

Glutamine (M147T525)

Hypoxanthine (M273T147)

Xanthosine (M323T367)

Xanthine (M151T233)

normalized Peak Area

12.5

10.0

5.0

7.5

2.5

5.0

0.0

2.5

0.0

ACA

ACC

ACA

ACC

ACA

ACC

ACA

ACC

Pyrimidine salvage/degradation metabolites

Cytidine (M242T451)

Uridine (M243T499)

Deoxyuridine (M246T370)

normalized Peak Area

ACA

ACC

ACA

ACC

ACA

ACC

Fig. 6: Untargeted global serum metabolomics in patients with ACC vs benign adrenocortical adenomas (ACA). a Schematic overview of clinical study. UHPLC-MS: Ultra-high performance liquid chromatography-mass spectrometry. b Clinical characteristics of study participants. c Volcano plot with significantly altered metabolites in the serum of patients with ACC, n=54 versus ACA, n=291 (red dots: increased in ACC, blue dots: decreased in ACC). d Dysregulated pathways (FDR<0.05) derived from analysis of polar metabolites. e-g Box and whiskers plots comparing relative serum abundance of Gln (e) and selected purine (f) and pyrimidine (g) metabolites from the salvage/degradation pathways in patients with ACC vs. ACA. The box edges represent the 1st and 3rd quartile and the upper and lower whiskers represent Q3 + 1.5x Interquartile range (IQR) and Q1 - 1.5x IQR, respectively. The median is depicted as the horizontal line within the box. * p<0.05, ** p<0.01, *** p<0.001 (t-test). For all metabolites, alternative putative annotations can be found in Suppl. Table 4 using peak identifiers indicated on the graph titles.

1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052

Supplemental Methods

Untargeted serum metabolome analysis

The samples were aliquoted on the day of collection and stored locally at -80℃. The samples were transported on dry ice to the University of Birmingham, United Kingdom, and stored at -80℃ prior to untargeted serum metabolome profiling.

Sample Extraction

Serum samples: Samples were randomised using the RAND() function in Microsoft Excel to determine the sample preparation order. This order was assessed to ensure that all the metadata (age, sex, body mass index, and disease classification) was randomised and there were no patterns/correlations with the sample preparation order. Serum samples were thawed on ice on the day of preparation. For the HILIC assay, plasma (50 uL) and 1/1 acetonitrile/methanol (v/v) (150 uL) were mixed by vortex (120 s) and centrifuged (17,000 g, 20 min, 4℃). For lipid assays, plasma (50 uL) and isopropanol (150 uL) were mixed by vortex (120 s) and centrifuged (17,000 g, 20 min, 4℃).

Pooled QC samples: 120 uL serum aliquots from all biological samples were transferred to a single tube and vortex mixed. This single tube sample was called the pooled quality control (QC) sample. Multiple 50 uL aliquots were prepared as described for serum samples.

Process blank samples: Blank extraction solutions were prepared by performing the extraction protocol as described above for serum samples with no serum present.

1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075

Ultra High Performance Liquid Chromatography-Mass Spectrometry: All of the three types of samples (biological, QC and blank) were analyzed by applying two UHPLC-MS methods in positive and negative ion modes separately using a Dionex UltiMate 3000 UHPLC system coupled with a heated electrospray Q Exactive Focus mass spectrometer (Thermo Fisher Scientific). Analysis order for the samples was randomized, and all of the samples were analyzed in two analytical batches. Ten QC samples were injected at the start of each assay/ion mode to condition the analytical system and were then injected after every 6th biological sample, with two QC samples being analyzed at the end of each batch. The blank sample was analyzed as the 5th and as the final injection of each batch.

HILIC assays: These assays used an Accucore 150-Amide-HILIC column (100 x 2.1mm, 2.6 um, ThermoFisher Scientific). For positive ion analysis, mobile phase A was 10 mM ammonium formate dissolved in acetonitrile/water/formic acid (95:4.9:0.1 (v/v)) and mobile phase B was 10 mM ammonium formate dissolved in acetonitrile/water/formic acid (50/49.9/0.1 (v/v)). For negative ion analysis, mobile phase A was 10 mM ammonium acetate dissolved in acetonitrile/water/acetic acid (95:4.9:0.1 (v/v)) and mobile phase B was 10 mM ammonium acetate dissolved in acetonitrile/water/acetic acid (50/49.9/0.1 (v/v)). The gradient elution applied for positive and negative ion mode was t=0.0, 1% B; t=1.0, 1% B; t=3.0, 15% B; t=6.0, 50% B; t=9.0, 95% B; t=10.0, 95% B; t=10.5, 1% B; t=14.0, 1% B. All changes were linear (curve = 5) and the flow rate was 0.50 mL.min-1. Column temperature was 35℃ and injection volume was 2 uL. Data were acquired in positive and negative ionization modes separately (70 - 1050 m/z) with a mass resolution 70,000 (FWHM, m/z 200). Ion source parameters applied were Sheath gas = 55 arbitrary

1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098

units, Aux gas = 35 arbitrary units, Sweep gas = 4 arbitrary units, Spray Voltage = 3.2kV (positive ion) and 2.7kV (negative ion), Capillary temp. = 380°℃, Aux gas heater temp. = 440°C. Thermo Exactive Tune (2.8 SP1, build 2806) software controlled the instruments and data acquisition. All data were collected as MS1 data in the Profile mode setting with the exception of five QC sample injections (injections 6-10) where MS/MS data were collected in the “Discovery mode” setting over different precursor m/z ranges (70-210 m/z; 200-310 m/z; 300-410 m/z; 400-510 m/z; 500-1050 m/z) using stepped normalized collision energies (positive ion mode: 20, 40, 100%; negative ion mode: 40, 60, 130%). All samples were maintained at a temperature of 4℃ in the autosampler. Instrument maintenance (source and column cleaning) was performed between the end of batch 1 and start of batch 2.

Lipid assays: These assays used a reversed-phase Hypersil GOLD C18 column (100 x 2.1 mm, 1.9um; Thermo Fisher Scientific). Mobile phase A was 10 mM ammonium formate dissolved in acetonitrile/water/formic acid (60:39.9:0.1 (v/v)) and mobile phase B was 10 mM ammonium formate dissolved in isopropanol/acetonitrile/water/formic acid (85.5/9.5/4.9/0.1 (v/v)). The gradient elution applied was t=0.0, 20% B; t=0.5, 20% B, t=8.5, 100% B; t=9.5, 100% B; t=11.5, 20% B; t=14.0, 20% B. All changes were linear (curve = 5) and the flow rate was 0.40 mL/min. Column temperature was 55 ℃ and injection volume was 2 uL. Data were acquired in positive and negative ionization mode separately (150 - 2000 m/z) with a mass resolution 70,000 (FWHM, m/z 200). lon source parameters applied were Sheath gas = 48 arbitrary units, Aux gas = 15 arbitrary units, Sweep gas = 0 arbitrary units, Spray Voltage = 3.2kV (positive ion) / 2.7kV (negative ion),

1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121

Capillary temp. = 380℃, Aux gas heater temp. = 450℃. Thermo Exactive Tune (2.8 SP1, build 2806) software controlled the instruments and data acquisition. All data were acquired in Profile mode. All data were collected as MS1 data in profile mode with the exception of five QC sample injections (injections 6-10) where MS/MS data were collected in Discovery mode over different precursor m/z ranges (150-510 m/z; 500-710 m/z; 700-860 m/z; 850-1010 m/z; 1000-2000 m/z) using stepped normalized collision energies (positive ion mode: 20, 40, negative ion mode: 40, 60, 130%). All samples were maintained at a temperature of 4℃ in the autosampler. Instrument maintenance (source and column cleaning) was performed between the end of batch 1 and start of batch 2.

Raw data processing: Vendor format raw data files (.RAW) were converted to the mzML file format using the open-source ProteoWizard software1. Deconvolution was performed by the R package XCMS (version 3.6.1 running in the Galaxy environment)2. The R package IPO3 was used to optimise and obtain XCMS peak picking parameters for each of the different assays:

(a) HILIC negative ion mode assay: min peak width = 6.23; max peak width = 40; ppm = 9.98; mzdiff = - 0.02025; bw = 0.25; mzwid = 0.001; minfrac = 0.5; snr_thr = 10.

(b) HILIC positive ion mode assay: min peak width = 5.70; max peak width = 40; ppm = 7.0; mzdiff = 0.00021; bw = 0.25; mzwid = 0.0059; minfrac = 0.5; snr_thr = 10.

(c) Lipids negative and positive ion mode assays: min peak width = 5.44; max peak width = 40; ppm = 6.5; mzdiff = - 0.0065; bw = 0.25; mzwid = 0.04265; minfrac = 0.5; snr_thr = 10.

1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143

A data matrix of metabolite features (i.e., m/z-retention time pairs) versus samples was constructed for each of four assay types (HILIC-negative, HILIC-positive, Lipids-negative, Lipids-positive).

Data quality assessment and filtering: Each of the four data matrices constructed by raw data processing was filtered based on the data collected for QC and blank samples4. Metabolite features were retained in the data matrix if they were: present in >90% of all the QC samples, had a peak area relative standard deviation (RSD) <30% across all the QC samples (QC11 onwards) and had an extract blank/mean QC area ratio of <5%.

Metabolite annotation: Metabolite annotation was performed applying the Python package BEAMSpy https://github.com/computational-metabolomics/beamspy (RT diff = 2 s, Pearson correlation >0.50; m/z values of all experimentally observed peaks were searched against LIPIDMAPS5 and HMDB6, all matches < 5ppm mass error tolerance were reported). To generate more robust compound annotations using MS/MS data, QC sample UHPLC-MS/MS data were matched to MS/MS databases using the LipidSearch software (lipid assay metabolite annotation; version 4.2.18, Thermo Fisher Scientific) and were searched against the entire in-silico HCD MS/MS database with 5ppm mass error. HILIC MS/MS data were searched in mzCloud using Compound Discoverer (v3.1, Thermo Fisher Scientific) using a >70% match as the criteria for a successful annotation. RT data were matched to an in-house RT library constructed with authentic chemical standards, matches within + 5 seconds were applied as the criteria for a successful match. LipidSearch and mzCloud annotations were aligned to the XCMS outputs using the R

1164 1165 1166

1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163

programming language (https://www.R-project.org), applying a 5ppm mass error for the MS1 data and a +/- 5 second retention time tolerance window. All annotations described are reported to either level 2 or 3 as defined by the Metabolomics Standards Initiative7.

Pathway enrichment analysis: Pathway enrichment analysis was performed in the software MetaboAnalyst8 for HILIC based results. Pathway enrichment analysis applied pathway analysis, scatter plot for visualization method, hypergeometric test for enrichment method, relative-betweenness centrality for topology analysis and Homo sapiens (KEGG) as the pathway library (accessed on 13/05/2024). Only pathways with a p-value (after FDR correction) < 0.05 and with a minimum of five metabolites in the pathway listed as biologically important were reported. For lipids based results, lipids were classified according to LipidMaps ontologies5 manually.

REFERENCES

1. Chambers MC, Maclean B, Burke R, et al. A cross-platform toolkit for mass spectrometry and proteomics. Nat Biotechnol 2012;30(10):918-20. DOI: 10.1038/nbt.2377.

2. Smith CA, Want EJ, O’Maille G, Abagyan R, Siuzdak G. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal Chem 2006;78(3):779-87. DOI: 10.1021/ac051437y.

3. Libiseller G, Dvorzak M, Kleb U, et al. IPO: a tool for automated optimization of XCMS parameters. BMC Bioinformatics 2015;16:118. DOI: 10.1186/s12859-015-0562-8.

4. Lloyd GR, Jankevics A, Weber RJM. struct: an R/Bioconductor-based framework for standardized metabolomics data analysis and beyond. Bioinformatics 2021;36(22- 23):5551-5552. DOI: 10.1093/bioinformatics/btaa1031.

5. Conroy MJ, Andrews RM, Andrews S, et al. LIPID MAPS: update to databases and tools for the lipidomics community. Nucleic Acids Res 2024;52(D1):D1677-D1682. DOI: 10.1093/nar/gkad896.

1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194

6. Wishart DS, Guo A, Oler E, et al. HMDB 5.0: the Human Metabolome Database for 2022. Nucleic Acids Res 2022;50(D1):D622-D631. DOI: 10.1093/nar/gkab1062.

7. Sumner LW, Amberg A, Barrett D, et al. Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics 2007;3(3):211-221. DOI: 10.1007/s11306-007-0082-2.

8. Pang Z, Lu Y, Zhou G, et al. MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Res 2024;52(W1):W398-W406. DOI: 10.1093/nar/gkae253.

1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205

1206 Suppl. Fig. 1

Up-regulated genes

BPCre ACC (mouse)

Human ACC

280

283

853

Purine metabolism (genes involved in de novo biosynthesis, salvage or catabolism

Down-regulated genes

BPCre ACC (mouse)

Human ACC

307

233

778

Pyrimidine metabolism (genes involved in de novo biosynthesis, salvage or catabolism

Human adrenals

Human ACC

ATIC

Mouse adrenals

BPCre-ACC

IMPDH

PAICS

PPAT

Adsl

APRT

Atic

DCK

Gart

DGUOK

Gmps

GMPR

Impdh1

HPRT1

ITPA

Regulation

Pfas

NT5C

Ada

NT5C2

DOWN

Aprt

NUDT1

Dck

NUDT15

Dguok

NUDT18

NUDT5

Itpa

NUDT9

Nt5c2

XDH

-1

Nt5e

NT5M

ADSL

-2

Nudt16

GMPS

Nudt5

PFAS

Nt5m

AMPD1

Nt5c1a

AMPD3

GDA

NT5C1A

NT5C1B

NUDT16

Mouse adrenals

BPCre-ACC

Dhodh

Dut

Umps

Dck

Tk1

Tk2

Uck1

Uck2

Uckl1

Upp2

Dpys

Nt5e

Nt5m

Agxt2

Dpyd

Nt5c1a

Human adrenals

Human ACC

DUT

UMPS

CTPS2

CDA

DCK

TK1

Regulation

TK2

UCK1

DOWN

UPP2

NT5C

NT5M

CAD

-1

TYMP

-2

UCK2

UCKL1

AGXT2

DPYD

DPYS

NT5C1A

UPB1

1207

1208

Suppl. Fig. 1 (Related to Fig. 1): Cross-species comparison of transcriptional metabolic dysregulation in ACC (purine and pyrimidine metabolism). a Venn diagrams showing overlap in significantly up-regulated and down-regulated metabolic genes between human and mouse (BPCre) ACC tumors, as assessed by RNA-seq. Mouse tissue: 6 control adrenals (no mutations) and 7 BPCre ACC tumors. Human tissue: 79 ACC tumors (TCGA) and 295 healthy adrenals (GTEx). b-c Heat-maps illustrating significant expression changes in genes involved in de novo and salvage purine (b) and pyrimidine (c) biosynthesis and catabolismas assessed by RNA-seq. Genes involved in de novo nucleotide biosynthesis are highlighted in bold purple font. Genes involved in salvage biosynthesis are highlighted in bold green font.

1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219

1220

Suppl. Fig. 2 Amino-sugar and nucleotide sugar metabolism

Mouse adrenals

BPCre-ACC

Gfpt1

Galk1

Ugdh

Gnpda1

Gmppa

Pmm2

Uxs1

Npl

Gpi1

Uap111

Amdhd2

Cyb5r1

Uap1

Human adrenals

Human ACC

GFPT1

PMM2

PMM1

RENBP

GCK

GALK1

NAGK

GNPDA1

GNPDA2

GALE

GNE

GPI

HEXA

Regulation

CMAS

UXS1

HEXB

DOWN

HK1

HK2

NANP

NANS

UGP2

UAP1L1

GNPNAT1

UGDH

CYB5RL

-1

GALT

-2

MPI

NPL

UAP1 HK3

CYB5R4

GMDS

GMPPA

GMPPB

FPGT

PGM3

PGM2

CYB5R1

PGM1

Mouse adrenals

BPCre-ACC

Glutathione metabolism

Gstm5

Gstm4

Gstm3

Gstm2

Gstm1

Gpx7

Gss

Ggt5

Rrm2

Rrm1

Idh2

Gclm

Gstp1

Gstt2

Gstt1

Gpx3

Ggt6

Gstk1

Ggt7

Mgst2

Mgst1

Human adrenals

Human ACC

GSTM4

G6PD

GPX4

GPX7

GSS

GPX8

RRM2B

GSTA3

GSTA2

GSTA1

ODC1

GGT6

RRM2

GSTK1

GSTO2

RRM1

GSTO1

IDH2

PGD

SMS

MGST2

MGST3

GSTT2

GSTM5

GPX3

GPX2

NAT8B

GSR

GSTA5

GSTA4

GGT5

IDH1

GGT7

NAT8

GCLM

GSTP1

MGST1

GGT1

Regulation

DOWN

-1

1221

1222

1223

Suppl. Fig. 2 (Related to Fig. 1): Cross-species comparison of transcriptional metabolic dysregulation in ACC (amino sugar and nucleotide sugar metabolism, glutathione metabolism). Heat-maps illustrating significant expression changes in genes involved in amino sugar and nucleotide sugar metabolism (a) and glutathione metabolism (b) in mouse and human ACC, as assessed by RNA-seq. Mouse tissue: 6 control adrenals (no mutations) and 7 BPCre ACC tumors. Human tissue: 79 ACC tumors (TCGA) and 295 healthy adrenals (GTEx).

1224 1225 1226 1227 1228 1229 1230 1231

1232 Suppl. Fig. 3

Decreased in BPCre-Pre

Increased in BPCre-Pre

D-Glucosamine

BPCre-Pre

Uracil

Glutathione

L-carnitine

UDP-Glucose

L-alanine Sarcosine

Phosphorylcholine

CAMP

BPCre-Pre VS Control

N-acetylglutamate

PC 2 (13.9%)

XMP

-log10(p value)

Creatine Phosphate Saccharate Galactarate Creatine

1.5

-2

Control

-4

0.5

-6

-10

-5

PC 1 (38.5%)

-3

-2

-1

log2(FC)

Decreased in Females

Increased in Females

Males Females

BPre-ACC Females VS Males

PC 2 (22%)

-log10(p value)

1.5

-5

0.5

-10

-5

PC 1 (30.4%)

-2

-1

log2(FC)

Decreased in Females

Increased in Females

Guanine 3PG

cis-4-Hydroxy-D-proline

Uric acid

2PG

Hypoxanthine Uridine

Guanidinoacetate

Females

DHAP

Control Females VS Males

PC 2 (15.9%)

Phosphoenol Pyruvate

-log 10(p value)

Inosine

Adenosine 5’-diphosphoribose

Males

NAD Carbamoyl-L-aspartate D-Glycerate L-Dihydroorotate Orotic acid

-2

-4

-6

-4

-2

PC 1 (40%)

-2

log2(FC) 0

Decreased in old Controls

Increased in old Controls

Old Controls Young Controls

Old Controls VS Young Controls

-log 10(p value)

1.5

PC 2 (15.2%)

-2

-4

0.5

-6

-5

-2

-1

PC 1 (30.5%)

log2(FC)

1233 1234

Suppl. Fig. 3 (related to Fig. 2): Tissue metabolome analysis from Control and BPCre mice, including comparisons by sex and age. a-d PCA and volcano plots highlighting changes in the polar tissue metabolome of selected groups. For mice without tumors (a, c and d), each sample represents tissue from two adrenals taken from either the same mouse or from two mice of the same genotype. For the volcano plots, metabolites with fold change threshold (x) 1.2 and t-test threshold (y) 0.05 (FDR-corrected ☒ p values) are annotated (colored red for increase and purple for decrease). Dashed lines indicate statistical thresholds. a Macroscopically normal adrenals from BPCre mice prior to tumor formation (BPCre-Pre) (n=4 samples from male and female mice, 4 months) vs. control adrenals from C57BL/6J mice (n=6 males and n=4 females, 7-11 months). Samples also shown in Fig. 2b. b ACC tumors from female and male BPCre mice (n=4 tumors from each sex, 8-13 months) c Control adrenals from female and male C57BL/6J mice (n=4 female and n=6 male samples, 7-11 months). d Adrenals from control younger (n=5 samples, age 3-5 months) and older (n=6 samples, age 6-10 months) male C57BL/6J mice.

1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250

1251 Suppl. Fig. 4

Cyp11b2-Cre

ß-Catenin GoF (BCre)

P53 LoF (Pcre)

Bigenic (BPCre)

Hyperplasia

Normal

ACC

Decreased in BCre

Increased in BCre

HMG

Control

BCre

BCre VS Control

-log10(p value)

1.5

PC 2 (18.9%)

dGMP UMP AMP ADPR

SAH Phosphorylcholine

NANA oo Citrulline

N-alpha-acetyllysine

· L-Threonine

·L-Proline

-2

0.5

-4

-6

-4

-2

PC 1 (25.5%)

-1

log2(FC)

Decreased in PCre

Increased in PCre

FAD

Control

PCre VS Control

PCre

PC 2 (15.7%)

-log10(p value)

-2

0.5

-4

-6

-4

-2

PC 1 (24.6%)

-1

-0.5

0.5

log2(FC)

1252 1253

Suppl. Fig. 4 (related to Fig. 2): Metabolic dysregulation in mice bearing @-Catenin gain-of-function (GOF) mutations or P53 loss-of-function (LOF) mutations. a Schematic of phenotypes in Cyp11b2 (AS)-Cre mouse models. b,c: PCA and volcano plots highlighting changes in the polar metabolome of BCre mice (bearing ß-GOF mutations) and PCre mice (bearing p53 LOF mutations), as compared to AS-Cre mice (no mutations). Each sample represents tissue from two adrenals taken from either the same mouse or from two mice of the same genotype. For the volcano plots, metabolites with fold change threshold (x) 1.2 and t-test threshold (y) 0.1 (FDR-corrected p values) ☒ ☒

are annotated (colored red for increase and purple for decrease). Dashed lines indicate statistical thresholds. b BCre mouse adrenals (n=6 samples from male mice, 7-11 months) vs. control adrenals from C57BL/6J mice (n=6 samples from male mice, 7-10 months). c Adrenals from PCre mice (n=6 samples from male mice, 7-12 months) vs. control adrenals from C57BL/6J mice (n=6 samples from male mice, 7-10 months).

1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267

1268

Suppl. Fig. 5

Cell line	Species	Mutations	Doubling time (h)
BCH-ACC3	Mouse	Ctnnb1 GOF; p53 LOF	19.5±1.1
NCI-H295R	Human	CTNNB1 GOF; P53 LOF	61.0±11.0
CU-ACC1	Human	CTNNB1 GOF	76.6+13.7
CU-ACC2	Human	P53 LOF; MSH2 LOF	52.3+14.9

BCH- Ctrl ACC3

Subcutaneous BCH-ACC3 tumor (H&E stain)

C57BL/6 mice

Cre

Treatment start

100

☒

% baseline weight

☒

Vehicle

JHU-083

☒

GFP

ßCatA Ex3

Days post-implant

1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281

Suppl. Fig. 5 (related to Fig. 3). ACC cell line description and subcutaneous implant model in C57BL/6 mice. a Table summarising ACC cell line characteristics. Doubling time (mean, S.E.M.) for each cell line was calculated from 4-10 independent cell proliferation experiments, using the formula Doubling Time (hours) = (72 * ln(2)) / ln(final cell number / initial cell number). b PCR analysis showing amplification of AS-Cre and ROSAmT/mG alleles in control samples, and loss of ROSAmT/mG and Cre alleles along with deletion of ß-catenin exon 3 in BCH-ACC3 cells. c Hematoxylin-Eosin staining from BCH- ACC3 subcutaneous implants. d Body weight (as % of weight at the beginning of treatment) in C57BL/6 mice treated with JHU-083 or vehicle. JHU-083 was administered at 1 mg/kg from day 6-10 post-implantation and 0.33 mg/kg from days 11-14. N=11,7; *p<0.05 (t-test).

1282 Suppl. Fig. 6

CU-ACC1

1.5-

Cell viability 144h normalised to Vehicle

1.0

0.5

0.0

DON

Nuc +

CU-ACC2

1.5-

Cell viability 144h normalised to Vehicle

1.0

0.5.

0.0

DON

Nuc

Glutaminolysis

Glutamine

Relative metabolite

abundance

Vehicle

JHU-083

Glutamate

2.0

1.5

1.0-

0.5-

0.0

Vehicle

JHU-083

a-ketoglutarate

2.0

1.5-

1.0-

0.5

0.0

Vehicle

JHU-083

D-Glucosamine

1.5

1.0

0.5

0.0

Vehicle

JHU-083

UDP-GIcNAc

2.0-

1.5-

1.0-

0.5

0.0

Vehicle

JHU-083

UDP-GaINAc

2.0

1.5

1.0

0.5

0.0

Vehicle

JHU-083

Glutathione

1.5

Relative metabolite

abundance

1.0

0.5

0.0

Vehicle

JHU-083

XMP

2.0

1.5

1.0

0.5

0.0

Vehicle

JHU-083

GDP

AMP

1.5

1.0-

1.0

0.5

0.0

Vehicle

JHU-083

Vehicle

JHU-083

ADP

2.0

1.5-

1.0-

0.5

0.0

Vehicle

JHU-083

Pyrimidine biosynthesis

N-Carbamoyl-aspartate

2.0

Relative metabolite abundance

1.5

1.0

0.5

0.0

Vehicle

JHU-083

Orotate

1.5

1.0

0.5

0.0

Vehicle

JHU-083

UMP

1.5

1.0-

0.5-

0.0

Vehicle

JHU-083

Cytidine

1.5

1.0

0.5

0.0

Vehicle

JHU-083

TMP

1.5

1.0-

0.5

0.0

Vehicle

JHU-083

Purine biosynthesis

GMP

Vehicle

JHU-083

Hexosamine biosynthesis

Suppl. Fig. 6 (related to Fig. 4): Glutamine antagonism drives nucleotide depletion in ACC. a Viable CU-ACC1 and CU-ACC2 cell number normalized to vehicle (DON-) after treatment with DON (DON +), alone or supplemented with the indicated metabolites. DON was administered at 5 uM for 144h. Nuc: Embryomax nucleoside cocktail 1x. Data representative of three independent experiments (5-8 technical replicates per experiment). **** p<0.0001, one-way ANOVA followed by Tukey’s test for multiple comparisons between DON monotherapy and each other group. b Relative abundance of selected metabolites in BCH-ACC3 SQ mouse implants after 48h treatment with JHU- 083 or vehicle. N=5,4; ns: not significant, *p<0.05, ** p<0.01, *** p<0.001, t-test.

1284 1285 1286 1287 1288 1289 1290 1291 1292 1293

1294

Purine biosynthesis

	PFAS			GMPS			PPAT
	Logrank p	HR		Logrank p	HR		Logrank p	HR
ACC	0.046	2.2	ACC	<0.001	4.3	KICH	0.012	9
LIHC	0.038	1.4	KIRP	0.010	2.3	LGG	<0.001	2
MESO	0.006	2	LGG	0.002	1.8	LIHC	0.008	1.6
SARC	0.023	1.6	LIHC	0.004	1.7	MESO	0.007	2
SKCM	0.007	1.4	LUAD	0.004	1.6	SARC	<0.001	2.1
UVM	0.068	2.3	MESO	0.006	2	THCA	0.002	7.3
			PAAD	<0.001	2.1

PCPG

0.016

5.8 ℮08

Pyrimidine biosynthesis

CAD

	Logrank p	HR
ACC	0.002	3.7
BLCA	0.043	1.4
KIRC	<0.001	1.7
KIRP	0.016	2.2
LAML	0.046	1.8
LGG	0.043	1.4
LIHC	<0.001	1.9
MESO	0.003	2.1

CTPS1

	Logrank p	HR
ACC	0.0024	2.5
LGG	0.00065	1.9
MESO	0.0038	2
SKCM	0.00021	1.7
UVM	0.01	3.4

ASNS Overall Survival

NADSYN1 Overall Survival

GLS Overall Survival

GLS2 Overall Survival

GFPT1 Overall Survival

0.0 0.2 0.4 0.6 0.8 1.0

.2 0.4 0.6 0.8 1

1.0

Percent survival

Low

Percent survival

Low

Percent survival

0.8

High

Percent survival

0.8

0.6

Low

Percent survival

0.6

Low

High

0.4

Low

0.4

High

0.4

High

0.2

Logrank p=0.02

Logrank p=0.06

0.2

0.0

Logrank p=0.60

0.0

Logrank p=0.52

0.0

Logrank p=0.17

100

150

100

Months

150

Months

100

Months

150

100

150

100

150

Months

1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306

Suppl. Fig. 7 (related to Fig. 4): Associations between Gln-metabolising enzyme expression and survival (TCGA database). a Cancer types with significant (logrank p<0.05) association between high enzyme gene expression levels and poor prognosis, based on The Cancer Genome Atlas (TCGA). HR: Hazard ratios. BLCA: Bladder urothelial carcinoma, KICH: kidney chromophobe cell carcinoma, KIRC: kidney renal clear cell carcinoma, KIRP: kidney renal papillary cell carcinoma, LAML: acute myeloid leukemia, LGG: low-grade glioma, LIHC: hepatocellular carcinoma, LUAD: lung adenocarcinoma, MESO: mesothelioma, PAAD: pancreatic adenocarcinoma, PCPG: pheochromocytoma and paraganglioma, SARC: sarcoma, SKCM: skin cutaneous melanoma, THCA: thyroid carcinoma, UVM: uveal melanoma. b Survival curves comparing overall survival in ACC patients with high (top 50%) versus low (bottom 50%)

gene expression levels of the indicated Gln-metabolizing enzymes, based on The Cancer

Genome Atlas (TCGA). Significance and hazards ratio were assessed by Log-rank (Mantel-Cox) tests.

1307 1308 1309 1310

Suppl. Fig. 8 (related to Fig. 6): Top 10 most up-regulated metabolites in the serum of patients with ACC. Box and whiskers plots comparing relative serum abundance of the 10 most increased metabolites in patients with ACC vs ACA controls (n=54, 291). The box edges represent the 1st and 3rd quartile and the upper and lower whiskers represent Q3 + 1.5x Interquartile range (IQR) and Q1 - 1.5x IQR, respectively. The median is depicted as the horizontal line within the box. *** p<0.001, **** p<0.0001 (t-test).

Androst-4-ene 3,17diol disulfate (M449T72)

Octanoylglucuronide (M379T84)

Androst-2-ene-17-one (M273T303)

3-hydroxyandrostane-17-one (M484T303)

normalized Peak Area

10.0

7.5

5.0

2.5

0.0

ACA

ACC

ACA

ACC

ACA

ACC

ACA

ACC

Pregnanetriol-3-O-b-D-glucuronide (M512T352)

Androstan-3,17diol disulfate (M451T76)

Pregn-5-ene-3,16,20-triol, 3-glucoside (M495T323)

11-hydroxyandrosterone- 3-glucuronide (M481T337)

normalized Peak Area

ACA

ACC

ACA

ACC

ACA

ACC

ACA

ACC

3-hydroxyandrostane-17-one 3-D-glucoronide (M525T352)

ST 27:2;04;/ ST 28:1;04 (M431T91)

normalized Peak Area

ACA

ACC

ACA

ACC

1312 1313 1314 1315 1316 1317 1318 1319

Supplemental Tables

Suppl. Table 1. Significantly up-regulated pathways (Benjamini-Hochberg-corrected FDR<0.05) in murine (BPCre) ACC (A) and human ACC (B). Significantly up-regulated genes are provided for each pathway.

Suppl. Table 2. Significantly down-regulated pathways (Benjamini-Hochberg-corrected FDR<0.05) in murine (BPCre) ACC (A) and human ACC (B). Significantly down-regulated genes are provided for each pathway.

Suppl. Table 3. Significantly dysregulated (Benjamini-Hochberg-corrected FDR<0.05 and >1.2 fold-change) polar metabolites in BPCre-ACC tissue, as compared to Control (normal) adrenals.

Suppl. Table 4: Significantly dysregulated (Benjamini-Hochberg-corrected FDR<0.05) serum metabolic features in patients with ACC, as compared to ACA. CE: cholesterol ester; Cer: Ceramide; CL: cardiolipin; DG: diacylglyceride; FA: fatty acid; HexCer: hexose-ceramide; LPA: lysophosphatidic acid; LPC: lysoglycerophosphocholine; LPE: lysoglycerophosphoethanolamine; LPG: lysoglycerophosphoglycerol; LPI:

lysoglycerophosphoinositol; PC:glycerophosphocholine; glycerophosphoglycerol; PI: glycerophosphoinositol;

LPS: PE:glycerophosphoethanolamine;

lysoglycerophosphoserine;

PG:

glycerophosphoserine; SHexCer: Sulfoglycosphingolipids; SM: sphingomyelin; ST: sterol lipid; TG: triacylglyceride

Suppl. Table 5. Significantly dysregulated (FDR<0.05) metabolic pathways based on analysis of significantly dysregulated polar serum metabolites in patients with ACC vs ACA (with reference to the KEGG database). Significant metabolites in each pathway are highlighted in red font.

1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347

Quartz 4

Explorer

41256360

Abstract

Highlights

Introduction

Results

Cross-species analysis reveals conserved transcriptional-metabolic programs in ACC

Gln-dependent metabolic pathways are dysregulated in mouse ACC

ACC is highly sensitive to disruption of Gln metabolism in vitro and in vivo

Inhibition of GIn-dependent nucleotide biosynthesis drives DON’s anti-tumor effect in ACC

High expression of Gln-metabolising enzymes in de novo nucleotide biosynthesis

correlates with poor prognosis

Inhibition of the DNA damage response synergizes with Gln antagonism to enhance anti-tumor effects in ACC

Untargeted serum metabolomics reveals dysregulation of nucleotide and amino- acid metabolism in patients with ACC

Discussion

Genomic Methods

Mice

Drugs

Cell Culture

In vitro assessment of cell viability and apoptosis

In vitro clonogenic assays

JHU-083 treatment in vivo

Targeted tissue metabolomic analysis

Paraffin Section Immunofluorescence

TCGA transcriptome analysis

Clinical study design

Untargeted serum metabolome analysis

Quantification and Statistical Analysis

Sex as a biological variable

Acknowledgement

Author Contributions

Competing Interest Statement

References

Supplemental Methods

Untargeted serum metabolome analysis

Sample Extraction

REFERENCES

1206 Suppl. Fig. 1

1220

1282 Suppl. Fig. 6

Supplemental Tables

Graph View

Table of Contents

Backlinks