MYC Induces Oncogenic Stress through RNA Decay and Ribonucleotide Catabolism in Breast Cancer

During tumorigenesis, cancer cells acquire genetic and epigenetic alterations that lead to widespread reprogramming of transcriptional and metabolic processes. Such changes fuel uncontrolled proliferation and other protumor hallmarks (1–6). However, large-scale deregulation of these processes may also create new dependencies and vulnerabilities in cancer cells (5, 7). In some cases, these counterbalancing protumor and stress hallmarks result from the same oncogenic alteration. MYC is a salient example of an oncogene that induces stresses across multiple cellular processes, from DNA replication to energy metabolism, that may be exploited for anticancer therapies (8–12).

Amplification and/or hyperactivation of MYC occurs in 20% to 40% of human malignancies and is associated with poor prognoses. As a transcription factor, MYC regulates many of its protumor functions by inducing or repressing gene expression programs (13–19). Oncogenic MYC activation has also been shown to elevate total RNA and protein production in many disease contexts to support the metabolic demands of cell growth and proliferation; the mechanisms for this elevation in macromolecular synthesis are subjects of intensive study (20–30).

While MYC-driven increased production of cellular RNA endows cancer cells with proliferative advantages, such a widespread increase in RNA synthesis may also create a burden on cells to maintain homeostasis of these macromolecules (23, 24, 30). For instance, MYC-driven cancers support the metabolic demands of increased transcription by increasing expression of de novo nucleotide biosynthesis pathways (31–34). In contrast, much less is known about how such elevated RNA synthesis is counterbalanced by RNA decay pathways and how this impacts downstream catabolic pathways that handle RNA breakdown products.

Here, we demonstrate that increased production of RNA catabolites and catabolic byproducts such as reactive oxygen species (ROS) is a critical source of MYC-induced cell death and oncogenic stress. In the setting of MYC hyperactivation, the RNA exosome and its catalytic ribonuclease DIS3L contributes to oncogenic cell death by facilitating increased RNA decay and subsequent accumulation of RNA catabolites. DIS3L mutations observed in human cancers can compromise this function of DIS3L and disable MYC-driven cell death. Furthermore, we find that purine salvage, which counterbalances ribonucleotide catabolism, is a compensatory mechanism on which MYC-driven cancers become dependent to tolerate such catabolic stress. MYC-driven breast cancers are especially sensitive to genetic and pharmacologic inhibition of purine salvage, which may represent a new therapeutic approach.

In many cell lineages and experimental systems, oncogenic activation of MYC has been shown to amplify the synthesis of cellular RNA through direct and indirect mechanisms (25, 26, 35–39). Several important studies have demonstrated that MYC-driven cancers support the metabolic demand of this global increase in RNA synthesis through coordinated transcriptional and translational control of nucleotide synthesis (23, 24, 30). These MYC programs have been the subject of intense therapeutic discovery (23, 40). In contrast, the impact of MYC-amplified RNA synthesis on downstream RNA decay and ribonucleotide catabolism in cancer and how this impacts tumor cell fitness is largely unknown.

To determine if MYC-driven tumors have altered ribonucleotide catabolism, we analyzed 61 primary human breast cancers for which broad metabolomics profiling was previously performed (32). Primary tumors were separated into MYC-high and MYC-low subsets via a MYC transcriptional signature (41) and assessed for levels of nucleic acid metabolites (Fig. 1A; Supplementary Fig. S1A–S1C). Nucleic acid metabolites, including those specific to nucleotide catabolism, such as uracil and β-alanine, were significantly more abundant in MYC-high breast cancers (Fig. 1A; Supplementary Fig. S1A–S1C). Strikingly, MYC-high breast cancers exhibited significantly higher levels of catabolites selective to RNA breakdown (i.e., not DNA breakdown), such as xanthosine, suggesting that RNA degradation is elevated in MYC-driven tumors.

To assess more directly whether MYC hyperactivation alters the accumulation of ribonucleotide catabolites, we performed targeted metabolomics profiling of human mammary epithelial cells (HMEC) expressing the MYC-estrogen receptor fusion protein, in which MYC activity can be rapidly induced with 4-hydroxytamoxifen (4-OHT; Supplementary Fig. S1D). Aberrant activation of MYC increased the abundance of nucleosides and free bases that may result from RNA or DNA degradation (e.g., xanthine, uracil, and β-alanine; Fig. 1B–E). Importantly, MYC also increased ribonucleotide catabolites specific to RNA degradation (e.g., xanthosine) and the end products of nucleotide degradation (e.g., uric acid). Prior reports suggest eukaryotic RNA turnover is at least 10- to 100-fold higher than that of DNA (43–46), leading us to hypothesize that after MYC hyperactivation, the increase in nucleotide catabolites was primarily the consequence of increased RNA degradation (as opposed to DNA degradation). This prediction was also supported by mathematical modeling of purine metabolism (46, 47), where increasing total cellular RNA by 25% led to a robust increase in hypoxanthine (∼4 to 11 fold) and xanthine (∼4 to 6 fold) concentration within 5 to 8 hours but increasing DNA by 25% did not impact hypoxanthine and xanthine concentrations (Supplementary Fig. S1E–S1G). To directly test whether oncogenic MYC affected global RNA decay, we labeled nascent RNA with the uridine analog 5-ethynyl uridine (EU) and quantified RNA synthesis (through EU incorporation) and RNA decay (through pulse-chase experiments; Fig. 1F). In agreement with the reported role of MYC in enhancing cellular RNA synthesis, MYC induction resulted in increased nascent RNA production as measured by fluorescent detection of EU immediately following EU labeling (Fig. 1G). Importantly, pulse-chase quantification of EU-labeled RNA over time revealed an increase in RNA decay rate in the MYC-hyperactivated state (Fig. 1H), suggesting that MYC hyperactivation increases both global RNA synthesis and RNA decay.

Taken together, these results indicate that MYC hyperactivation leads to an increase in RNA decay and a concomitant accumulation of many ribonucleotide catabolic intermediaries and terminal products. Furthermore, ribonucleotide catabolites are elevated in MYC-driven breast cancers, suggesting that this imbalance in RNA decay exists in the context of oncogenic MYC hyperactivation.

MYC-driven oncogenic stress pathways and cell death have been shown to be major barriers to tumorigenesis. Notably, in addition to accumulation of RNA catabolites, MYC induction also led to a significant increase in cell death, including apoptosis (Supplementary Fig. S1H–S1J). As some RNA catabolic products can be cytotoxic (48–50), these findings led us to investigate whether heightened RNA decay and ribonucleotide catabolism contribute to MYC-driven cell death and thus may be a form of oncogenic stress. More specifically, we hypothesized that regulators of RNA degradation may impact cell survival in the setting of MYC hyperactivation.

To investigate whether pathways that control RNA decay can also impact cell survival in the context of oncogenic MYC hyperactivation, we analyzed genome-wide loss-of-function screen data we previously conducted in HMECs with and without MYC hyperactivation (10, 51). In particular, we were interested in genes whose knockdown increased cell viability selectively upon MYC induction. These genes may encode proteins that are critical facilitators of MYC-induced oncogenic stress. Thus, short hairpin RNAs (shRNA) targeting these genes would result in decreased MYC-induced cell death, resulting in greater shRNA abundance specifically in the MYC-induced cell population (Fig. 2A). A total of 223 genes (Supplementary Fig. S1K; Supplementary Table S1) were identified as candidate facilitators of MYC-induced cell death (genes identified had 2+ shRNAs per gene with a log2 fold change >0.5 in MYC-induced state compared to MYC-normal state, and P < 0.1). To systematically identify cellular components that participate in MYC-induced cell death, we performed gene ontology analysis (52). Notably, from these 223 candidates, the top scoring protein complex was the cytoplasmic RNA exosome, a multi-subunit complex that executes exonucleolytic decay of single-strand RNA in the 3′-5′ direction into single nucleotides (Fig. 2B and C; Supplementary Fig. S1E; Supplementary Table S2). This highlighted the hypothesis that the RNA exosome may be a conduit of cellular stress in context of MYC hyperactivation by increasing cellular accumulation of RNA breakdown products. Multiple shRNAs targeting scaffold components (e.g., EXOSC6, EXOSC8) and the catalytic subunit (DIS3L) of the cytoplasmic exosome were enriched in the primary screen (Supplementary Fig. S1K; Supplementary Tables S1 and S2). Given that the original genetic screen was performed with a shallow library (mean of 2.31 shRNAs/gene), we sought to test whether DIS3L and other exosome components were on-target hits using several orthogonal approaches. In individual shRNA validation experiments, multiple shRNAs targeting these subunits suppressed MYC-induced cell death (Fig. 2D; Supplementary Fig. S2A–S2H). Inducible expression of shRNA-resistant DIS3L restored MYC-induced cell death (Fig. 2E; Supplementary Fig. S2D), thereby validating that the phenotype is an RNAi on-target effect.

Given that the exosome scaffold components are shared between the nuclear and cytoplasmic exosome complexes (53–55), we investigated whether the nuclear exosome also mediates MYC-induced cell death. Interestingly, knockdown of the nuclear exosome exonuclease (DIS3) did not impact MYC-induced cell death (Supplementary Fig. S2I and S2J), even though significant RNA turnover is known to occur in the nucleus (56). This suggests that cytoplasmic, rather than nuclear, RNA degradation by the exosome is a critical mediator of MYC-induced cell death. To test whether DIS3L, which is localized exclusively in the cytoplasm, impacts MYC-induced increase in RNA decay, we performed EU pulse-chase experiments with and without knockdown of DIS3L. Depletion of DIS3L suppressed the MYC-induced increase in RNA decay rate (Fig. 2F and G) without affecting MYC-induced increase in RNA synthesis (Supplementary Fig. S2K). This indicates that cytoplasmic RNA is the major source of RNA whose degradation is impacted by MYC hyperactivation.

Collectively, these data suggested that MYC-induced cell death is, in part, dependent on cytoplasmic RNA decay (and potentially downstream ribonucleotide catabolites) that is mediated by the DIS3L ribonuclease. However, because mRNA decay can also be induced by apoptosis (57), we evaluated an alternative hypothesis that MYC-induced RNA decay is a downstream consequence of MYC-induced apoptosis. To test this, we evaluated the timing of both MYC-induced RNA decay and MYC-induced cell death. Notably, we found that while MYC-induced RNA decay and ribonucleotide catabolism occurs within 4 hours of MYC hyperactivation (Figs. 1B-D and 1H), MYC-induced cell death is observed long after this time point (24 hours; and Fig. 2D). To verify this timing, we measured MYC-induced apoptosis by annexin V, caspase 3/7 assay, and time-lapse imaging. We did not find significant induction of apoptosis within 4 hours post-MYC-hyperactivation (Supplementary Fig. S3A–S3D) when RNA decay was observed. These observations indicated that MYC-induced RNA decay and ribonucleotide catabolism precede MYC-induced cell death. Moreover, ectopic expression of BCL2 [which inhibits MYC-driven cell death (58, 59)] did not alter MYC-induced RNA decay (Supplementary Fig. S3E-F). These data are consistent with the hypothesis that MYC-induced RNA decay is not a consequence of cell death but instead may contribute to MYC-induced apoptosis.

Given our observation that cytoplasmic RNA decay may contribute to MYC-induced cell death, we sought to investigate the pools of RNAs that might be dysregulated by MYC hyperactivation specifically when DIS3L is impaired. In particular, we hypothesized that such pools of RNA may contribute to the downstream accumulation of RNA degradation products and ribonucleotide catabolites that are observed when MYC is aberrantly activated (Fig. 1). Given that rRNA turnover rates (half-life ∼2 to 4 days) are significantly less than that of mRNA (average half-life ∼9 hours; refs. 60–63), we sequenced rRNA-depleted RNA. We compared MYC-induced changes in RNA expression levels in cells with and without DIS3L knockdown. Strikingly, when we compared the MYC inducibility of RNAs in each decile of steady-state abundance, the most abundant transcripts in the baseline state were also the most highly induced by MYC in the DIS3L-depleted state but not in the control-shRNA state (Fig. 2H; Supplementary Fig. S3G). In addition to coding mRNAs, these high abundance transcripts included many small RNAs like snoRNAs and microRNAs (Fig. 2I). In contrast, MYC-induced expression changes for lowly expressed transcripts were less impacted by knockdown of DIS3L (Fig. 2H; Supplementary Fig. S3G). This suggests that the ability of MYC to induce RNAs that are already highly abundant is counterbalanced by DIS3L-mediated degradation. It also suggests that this highly abundant pool of RNAs may be a prominent source of ribonucleotide catabolites and oncogenic stress in the context of MYC hyperactivation. Notably, MYC hyperactivation had only modest or no effect (0%–20% change) on expression levels of the RNA exosome itself (including DIS3L; Supplementary Fig. S3H), suggesting that MYC-induced RNA decay likely does not occur through direct upregulation of the RNA exosome.

Taken together, these results suggest that MYC-induced RNA synthesis is followed by widespread cytoplasmic RNA decay mediated by DIS3L that maintains homeostasis of highly expressed RNAs. This also supports the hypothesis that MYC-induced RNA decay produces a collateral effect of increased accumulation of downstream ribonucleotide catabolites, ROS, and cell death (explored in more detail below).

Oncogenic MYC activity leads to apoptotic signaling that serves as a barrier to cancer progression in the absence of cooperating mutations (8, 9). Our discovery that loss of DIS3L function can suppress MYC-induced cell death raised the hypothesis that DIS3L may be a tumor suppressor. Thus, we evaluated whether DIS3L function may be compromised in human cancer. We noted that somatic mutations occur in the DIS3L locus in breast cancer and other malignancies with a pan-cancer frequency of 1.4% in 10,953 patients from cBioPortal (http://www.cbioportal.org; Supplementary Fig S4A; ref. 64). To assess which mutations, if any, are likely to impact DIS3L function, we performed an evolutionary analysis on DIS3L somatic mutations from pan-cancer collections. Evolutionary action (EA) predicts the effect of genotype perturbations on protein fitness and function (65). Evolutionary trace (ET) scores that indicate the functional role of each genotype position were aligned and mapped onto the crystal structure of DIS3L’s structural and functional homolog, Saccharomyces cerevisiae Dis3 (Fig. 3A; ref. 66). EA analysis revealed a strong enrichment of deleterious mutations (EA score > 60) within the RNA-interacting domains including the exonuclease domain (Supplementary Fig. S4B; Supplementary Table S3; ref. 66). Notably, several mutations with high EA scores were recurrent across cancers (Supplementary Fig. S4A and S4B) including mutations within the exonuclease domain, suggesting that the ability of the cytoplasmic exosome to degrade RNA may be compromised in multiple cancer types. To better understand the impact of cancer-derived mutations in DIS3L, we analyzed the residues that are conserved in S. cerevisiae Dis3 and predicted their impact on DIS3L functionality. In general, mutations in DIS3L fall into five categories- (i) protein misfolding/instability (e.g., V347E), (ii) exosome core interaction defect (e.g., R104Q), (iii) direct and indirect RNA-binding defects (Fig. 3B and C; e.g., D486N and G786S), (iv) incomplete DIS3L expression (e.g., R262*), and (v) unexplained defects (e.g., R412C). Overall, these structure-based predictions suggest that DIS3L function may be compromised in DIS3L-mutated human cancer.

To test the functional consequence of these tumor-derived DIS3L mutations, we evaluated three DIS3L mutations with a high EA score (>60), for their impact on MYC-induced cell death. To this end, endogenous DIS3L was knocked down in MYC-ER HMECs with an shRNA targeting the 3′UTR of DIS3L, and DIS3L wild-type or mutant cDNAs were expressed in these cells. Notably, all three mutations (R412C, D486N, and G786S) impaired MYC-induced cell death (Fig. 3D and E). In addition, mutation of one of the DIS3L catalytic residues (D486; refs. 53, 54) abrogated MYC-induced cell death (Fig. 3E), strongly suggesting that RNA degradation by the exosome is required, in part, for MYC-induced cell death.

These data suggest that the ability of DIS3L to degrade RNA and contribute to MYC-induced cell death may be partially compromised in human cancers. However, we noted that the frequency of these somatic mutations were relatively low (1.4% pan cancer). Thus, we evaluated whether the DIS3L locus is compromised by additional mechanisms. Notably, we found that ∼30% of breast cancers harbor shallow genomic deletions in DIS3L that associate with lower DIS3L mRNA levels (Supplementary Fig. S4C–S4E). Importantly, DIS3L deletions also associate with MYC amplification (Supplementary Fig. S4D), consistent with the hypothesis that MYC-amplified tumors may select for reduced DIS3L function. To further explore this possibility, we analyzed DIS3L mRNA and MYC copy number in cohorts of breast tumors. Notably, we found that DIS3L mRNA is significantly downregulated in MYC-amplified tumors compared to MYC-nonamplified tumors (Supplementary Fig. S4F). These data suggest DIS3L downregulation is a common feature of MYC-driven breast tumors. More broadly, these data suggest that DIS3L-mediated RNA decay is required for MYC-induced cell death, at least in some contexts, and that this function of DIS3L may be partially compromised in human cancer. Further studies are needed to determine whether RNA degradation is compromised by other mechanisms (in addition to DIS3L mutation and expression) in MYC-driven malignancies.

The genetic data and mutational analysis presented above indicate that DIS3L mediates an increase in RNA decay and cell death in the context of MYC hyperactivation. Notably, the process of ribonucleotide catabolism produces cytotoxic ROS, several catabolites, and intermediates (e.g., xanthine, uric acid; example diagram in Fig. 1E; refs. 48–50). Given our data that MYC upregulates ribonucleotide catabolites (Fig. 1), we hypothesized that this increase is the consequence of RNA decay mediated by the cytoplasmic exosome (and DIS3L). Indeed, depletion of DIS3L alone significantly suppressed MYC-induced levels of both purine and pyrimidine ribonucleotide catabolites, suggesting that DIS3L-mediated RNA decay is critical for aberrant accumulation of these catabolites (Fig. 4A and B). Notably, knockdown of DIS3L did not significantly lower the effect of MYC-induction on the expression of genes implicated in ribonucleotide catabolism and synthesis or the genes known to play antiapoptotic roles (Supplementary Fig. S4G, S4H, and S5A). This supports the model that DIS3L regulates catabolite levels via widespread degradation of RNA rather than regulation of specific gene products.

MYC hyperactivation leads to increased ROS in many cell and cancer types (68–70). ROS accumulation has been proposed to be a major oncogenic stress in the context of MYC oncogenesis and is a target of current therapeutic efforts against MYC-driven cancers (40). However, the mechanisms underlying MYC-induced ROS generation have remained unclear. Prior observations have suggested that MYC-induced ROS may arise from several mechanisms including induction of p53 target genes, induction of a MYC-target gene ODC, biochemical imbalances in mitochondrial metabolism, overexpression of NADPH oxidases, and increased oxygen consumption (68–70). Because ribonucleotide catabolism produces ROS at several steps, we hypothesized that RNA decay and ribonucleotide catabolism may be a major source of MYC-induced ROS (Fig. 4C). Similar to other cell types, MYC hyperactivation in HMECs led to robust ROS accumulation [Fig. 4D (top) panels, quantification in Fig. 4E]. Depletion of DIS3L completely suppressed MYC-induced ROS but did not impact baseline levels of ROS (Fig. 4D and E), suggesting that RNA decay is required for ROS generation in response to MYC hyperactivation.

To determine if MYC-induced ROS is dependent on downstream ribonucleotide catabolism, we measured ROS and subsequent cell death in the context of depletion of xanthine dehydrogenase (XDH), an enzyme that functions in the last steps of purine catabolism. XDH converts hypoxanthine into xanthine and uric acid and at both catalytic steps reduced XDH reacts with molecular oxygen (O₂) to produce superoxides (O₂⁻; ref. 71). Similar to DIS3L depletion, reduction in XDH levels via shRNA mitigated MYC-induced ROS and cell death (Fig. 4F and G). Similarly, XDH-inhibitor allopurinol suppressed MYC-induced apoptosis (Supplementary Fig. S5B). To directly test if an increase in these RNA catabolites may contribute to cell death, we treated MYC-amplified TNBC cells with exogenous xanthosine, which is converted to cytotoxic catabolites xanthine and uric acid in cells, and also increases the production of toxic superoxides (O₂⁻) in the process. Consistent with the model that ribonucleotide catabolite accumulation gives rise to oncogenic stress and cell death, xanthosine treatment led to growth inhibition and apoptosis in TNBC cells, and increased ROS accumulation and MYC-induced cell death in MYC-ER HMECs (Supplementary Fig. S5C–S5F). Collectively, these data support the hypothesis that aberrant MYC hyperactivation leads to enhanced RNA decay, ribonucleotide catabolism, and ROS generation that contribute to MYC-induced cell death and may serve as an oncogenic stress that limits tumor progression.

The results above indicate that MYC-hyperactive cancers may harbor increased levels of ribonucleotide catabolites and ROS, and thus depend on mechanisms to mitigate accumulation of such potentially toxic catabolites (model in Fig. 4H). During normal homeostasis, the accumulation of ribonucleotide catabolites is primarily counterbalanced by nucleotide salvage pathways, especially via the purine salvage enzyme HPRT1, which converts hypoxanthine and guanine into IMP and GMP, respectively (50). Therefore, we hypothesized that MYC hyperactivation may lead to a heightened dependency on HPRT1 in a manner that depends on elevated RNA decay. To evaluate this hypothesis, we first tested whether MYC induction causes sensitivity to HPRT1 inhibition. In this experimental context, we lowered the MYC-ER induction (using reduced levels of 4-OHT) in order to observe potential MYC-synthetic lethality. Indeed, HPRT1-depletion with multiple shRNAs was synthetic lethal with MYC hyperactivation in HMECs (Fig. 4I; Supplementary Fig. S5G and S5H). HPRT1-depletion also impaired anchorage-independent proliferation of MYC-dependent breast cancer cells in contrast to MYC-independent breast cancer cell lines (Fig. 4J and K). Likewise, doxycycline-inducible shRNAs targeting HPRT1 significantly impaired progression of MYC-hyperactive breast cancer xenografts, as measured by pooled in vivo tumor competition assays (data for three independent shRNAs in Fig. 4L). In contrast, shRNAs targeting DIS3L significantly enhanced progression of MYC-hyperactive breast cancer xenografts (Supplementary Fig. S5I). These data suggest that MYC-hyperactive breast cancer cells harbor an enhanced dependency on HPRT1. We hypothesized that in the context of human breast cancers, MYC-hyperactive tumor cells exhibit increased reliance on HPRT1 to mitigate stress from increased RNA catabolism and thus may select for higher HPRT1 levels during tumorigenesis. Consistent with this hypothesis, MYC-high tumors in both The Cancer Genome Atlas (TCGA) and METABRIC cohorts had a higher expression of HPRT1 (Fig. 4M and N).

The observations above suggest MYC-hyperactive cancers may be sensitive to inhibition of HPRT1. Notably, 6-mercaptopurine (6-MP) is an HPRT1 competitive inhibitor that has been studied and used for decades as an anticancer agent and immunosuppressant (72, 73) in the clinic. However, the core mechanism proposed for the ability of 6-MP to kill cancer cells and suppress the immune system is as a DNA incorporating poison, not as an HPRT1 antagonist. More specifically, studies suggest cancer and immune toxicities are induced by 6-MP because it is converted by HPRT1 into thio-dGTP, which in turn is a substrate for DNA polymerase that causes cytotoxicity once incorporated into DNA (72). However, our observations that genetic perturbation of HPRT1-function alone may selectively kill MYC-driven cancers raised the provocative hypothesis that 6-MP may have anticancer activity through a second unanticipated mechanism: exacerbation of purine catabolism in MYC-hyperactive cells, resulting in increased production of cytotoxic ribonucleotide catabolites and ROS. To evaluate this hypothesis, we first investigated whether oncogenic MYC increases sensitivity to 6-MP, and whether 6-MP enhanced purine ribonucleotide catabolism in MYC-hyperactive breast cancer cells. Notably, in MYC-ER HMECs, MYC-induction significantly increased sensitivity to 6-MP (decrease in GI50 by nine-fold) (Fig. 5A), suggesting that MYC hyperactivation is sufficient to drive sensitivity to 6-MP. Similarly, MYC-dependent cancer cell lines were more sensitive to 6-MP than MYC-independent models (Fig. 5B; Supplementary Fig. S6A–S6D). Finally, single-agent 6-MP impaired progression of MYC-driven breast cancer xenografts (Fig. 5C). These data suggest that MYC hyperactivation is at least one mechanism that drives sensitivity to 6-MP in human cancers.

These observations raise the hypothesis that 6-MP anticancer activity may depend, in part, on enhanced ribonucleotide (purine) catabolism. Notably, 6-MP caused an increase in xanthine (Fig.5D) and uric acid (Supplementary Fig. S6E), consistent with impaired HPRT1-mediated salvage, but did not impact accumulation of uracil, which is a pyrimidine catabolite independent of HPRT1 activity (Fig. 5D). This was accompanied by increased levels of ROS (Fig. 5E), consistent with impaired purine salvage and enhanced purine catabolism. Importantly, shRNA-depletion of DIS3L abrogated 6-MP-induced ROS (Fig. 5E) and conferred resistance to 6-MP (Fig. 5F), indicating that partially decreased RNA decay is sufficient to lower the flux through this catabolic pathway and enhance cell survival in the context of MYC hyperactivation. Further supporting this model, depletion of the enzyme XDH that catalyzes hypoxanthine and xanthine and produces ROS conferred resistance to 6-MP-driven cell death in MYC-hyperactivated conditions (Supplementary Fig. S6F). Overall, these data suggest that RNA decay and ribonucleotide catabolism by XDH are required for 6-MP induced cancer cell death in MYC-hyperactivated settings. As an alternative hypothesis, we tested if inhibiting HPRT1 may also cause cytotoxicity through depletion of the purine nucleotides. Notably, the levels of guanine and GMP were not impacted by 6-MP treatment (Supplementary Fig. S6G), suggesting that inhibition of ribonucleotide salvage enhances terminal purine ribonucleotide catabolism and may primarily cause toxicity by accumulation of ribonucleotide catabolites. Consistent with this hypothesis, supplementation with IMP had a minor effect and was not able to fully rescue cell viability in MYC-dependent LM2 cells. In addition, supplementation with GMP had no effect on 6-MP-induced cytotoxicity (Supplementary Fig. S6H). Likewise, a second MYC-high TNBC cell line (SUM159) did not show significant rescue of 6-MP induced cytotoxicity by IMP or GMP supplementation (Supplementary Fig. S6I). Taken together, these results indicate that 6-MP may impair the survival and progression of MYC-driven cancers through an unanticipated mechanism of action independent of DNA incorporation.

While 6-MP is effective in some cancers (72), its efficacy has been largely hampered by dose-limiting toxicities in normal tissues, and these toxicities have been attributed to the DNA-incorporation properties of 6-MP metabolites [Fig. 5G (left panel)]. Our data suggest that sensitivity to 6-MP in MYC-driven cancers may be dependent on inhibition of purine salvage as opposed to DNA incorporation/DNA damage. Thus, we were interested in a pharmacologic strategy that specifically targets purine salvage without incorporating into DNA, thus avoiding potential patient toxicities associated with this mechanism of action (Fig. 5G, model contrasted in left and right panels). To test this, we evaluated acyclic nucleoside phosphonate (ANP) DA-XV-50, a potent competitive inhibitor of human HPRT1 (i.e., K_i = 10 nmol/L; ref. (74)). For these studies, a prodrug form of DA-XV-50, identified as DA-XV-55 [Fig. 5G (middle) panel], was synthesized to enhance cell permeability. Once inside the cell, the lipophilic attachments are cleaved by endogenous esterases to release DA-XV-50, which then inhibits HPRT1. Because of its stable chemical structure and bulk, DA-XV-50 is incapable of being a substrate for DNA polymerase and therefore cannot incorporate into DNA [Fig. 5G (right) panel]. In MYC-ER HMECs, we tested if DA-XV-55 is MYC-synthetic lethal. Notably, DA-XV-55 treatment resulted in a greater increase in caspase-3 and -7 activity in MYC-hyperactivated cells compared to MYC-normal cells (Fig. 5H). Likewise, DA-XV-55 induced greater cell death in MYC-dependent breast cancer cells (LM2) compared to MYC-independent breast cancer cells (HCC1937; Fig. 5I). To test if DA-XV-55 driven cell death is dependent on DIS3L-mediated RNA decay, we performed a competition assay in MYC-dependent breast cancer cells (SUM159 and LM2) with DIS3L or control shRNA. In line with our hypothesis, depletion of DIS3L conferred dose-dependent resistance to DA-XV-55 in MYC-dependent breast cancer cells (Fig. 5J). Overall, these results suggest that the HPRT1 inhibitor DA-XV-55, which cannot incorporate into DNA, is synthetic lethal with MYC hyperactivation and drives cell death through DIS3L-mediated RNA decay. Because the pharmacokinetic properties of DA-XV-55 prevent in vivo testing, additional medicinal chemistry and preclinical studies will be required to evaluate this hypothesis in the setting of MYC-driven tumor progression. More broadly, these results provoke the hypothesis that pharmacologic strategies to inhibit HPRT1 and exacerbate ribonucleotide catabolism (but preventing DNA incorporation) may be a new approach to ameliorating MYC-driven cancers.

Hyperactivation of MYC is one of the most common events in human cancer and thus there have been major efforts to delineate how this oncogene drives tumorigenesis and to subvert these mechanisms therapeutically. Particular attention has been paid to how MYC hyperactivation imposes a metabolic stress on cancer cells. Prior studies have shown that this stress is, in part, due to the biosynthetic demand of increased production of macromolecules (e.g., nucleic acids and protein) necessary to sustain cell proliferation, survival, and other protumorigenic programs (20–22, 25, 26, 28, 29). In contrast, little attention has been given to the catabolic pathways that counterbalance the oncogenic programs driving widespread macromolecular synthesis (23, 24, 30). Herein, we have discovered that oncogenic activation of MYC induces cell death through a new mechanism: widespread elevation in RNA decay and downstream accumulation of toxic ribonucleotide catabolites and associated ROS. Moreover, we show that this new form of MYC oncogenic stress, that we term RNA catabolic stress (RCS), can be exploited therapeutically by inhibiting HPRT1, which removes the compensatory nucleotide salvage pathway that enables cancer cells to tolerate RCS.

MYC has been known to induce ROS for two decades and oxidative stress has been suggested as a therapeutic vulnerability in MYC-driven cancers, though the mechanism by which MYC induces oxidative stress is unclear (40). Our data indicate that MYC hyperactivation induces oxidative stress through increased RNA decay by the cytoplasmic exosome and downstream elevation in purine ribonucleotide catabolism by XDH. Given this novel understanding of MYC-induced oxidative stress, our results suggest the possibility of exacerbating oxidative stress-induced cell death in MYC-driven cancer by toggling RNA metabolism toward increased RNA degradation. Intriguingly, XDH inhibitor allopurinol is given as a supportive measure during chemotherapy which could inadvertently be helping MYC-driven tumor cells survive RCS. Further studies will be needed to understand whether other oncogenic contexts (e.g., oncogenic transcription factors like the estrogen receptor) produce aberrant ribonucleotide catabolism and oxidative stress through similar mechanisms.

Targeting DNA metabolism has been a mainstay in cancer therapeutics for more than 75 years, where rapidly replicating tumor cells exhibit enhanced sensitivity to chemotherapies that impede DNA synthesis, induce DNA damage, or inhibit DNA repair. Unfortunately, these therapies also impose undesired toxicity in actively dividing nonmalignant cells (like the immune compartment and GI tract), and these dose-limiting toxicities are often a major limitation to providing efficacious cancer treatment. For instance, the HPRT1 antagonist 6-MP is used to treat ALL and CML and is known to cause cytotoxicity through conversion into thio-GTP and activation of mismatch repair upon incorporation into DNA (72, 73). Our data suggest an alternative, proliferation-independent mechanism whereby 6-MP causes cancer cytotoxicity through increased ribonucleotide catabolism in MYC-driven malignancies. While this mechanism requires validation and further study, it invokes the provocative idea that therapies specifically targeting oncogenic stress due to enhanced RNA turnover may provide new entry points for cancer therapy. Development of new modalities that exploit this mechanism may provide new therapeutic approaches that avoid the toxicities associated with DNA poisons and perhaps even kill nondividing cancer cells.

Commercially available pGIPZ shRNAs targeting DIS3L (V2HS_ 99882 and V3LHS_391635), EXOSC8 (V2LHS_ 61553 and V2HS_ 202870), EXOSC2 (V3LHS_368199, V2HS_197390 and V2HS_6007), XDH (V2LHS_92691), and HPRT1 (V2LHS_82409, V3LHS_364983 and V3LHS_364984) were obtained from Open Biosystems. shRNAs targeting the 3′ UTR region of DIS3L were designed using the RNAiCentral (http://katahdin.mssm.edu/siRNA/RNAi.cgi?type=shRNA) algorithm (shRNA sequence 5′- TGCTGTTGACAGTGAGCGAAGAGCGAGACTCCATCTCAAATAGTGAAGCCACAGATGTATTTGAGATGGAGTCTCGCTCTGTGCCTACTGCCTCGGA -3′). For inducible RNAi experiments, shRNAs were cloned into the pINDUCER11, and for inducible cDNA experiments, wild-type and mutant cDNAs were cloned into the pINDUCER22 doxycycline-inducible lentiviral expression system (75). BCL2-pcw107-V5 was a gift from David Sabatini and Kris Wood (Addgene plasmid # 64630; http://n2t.net/addgene:64630; RRID:Addgene_64630). DIS3L mutants were generated by site-directed mutagenesis using QuikChange II (Agilent) according to manufacturer’s protocol. Lentiviruses and retroviruses were produced by transiently transfecting shRNA or cDNA constructs into HEK293T cells using Mirus Bio TransIT (Mirusbio) and viral supernatants were collected 48 hours after transfection.

All cell lines were incubated at 37°C in 5% CO₂. MYC-ER HMECs were cultured in MEGM media (Lonza) and treated with 4 to 60 nM 4-OHT to induce MYC hyperactivation. MDA-MB-231 LM2 and 293T were cultured in DMEM media (Gibco) supplemented with 10% FBS. SUM159 were cultured in F12 media (Thermo Fisher) supplemented with 5% FBS, 10 mmol/L HEPES, 5 µg/mL insulin, and 1 µg/mL hydrocortisone. HCC1937 were cultured in RPMI 1640 media (Gibco) supplemented with 10% FBS. SKBR3 cells were cultured in McCoy’s 5A medium (Corning) supplemented with 10% FBS. MYCN3 were cultured in RPMI 1640 media (Gibco) supplemented with 10% FBS and treated with 1 uµ/mL doxycycline to induce MYCN expression. For P493-6 cell line with pMYC-tet construct, MYC was reduced by treating cells with 0.1 mg/mL tetracycline (Sigma) for 72 hours. MYC was re-induced by washing P493-6 cells with PBC twice, then culturing cells in RPMI-1640 medium with 10% Tet System Approved FBS (Clontech) and 1% GlutaMAX. 293T, SKBR3, and HCC1937 cells were obtained from ATCC #cat nr. CRL-3216, HTB-30, and CRL-2336, respectively); SUM159 (female) cells were obtained from BioVT (SUM-159PT); MDA-MB-231-LM2 were obtained from (76); HMEC (HME1) cells were obtained from Infinity (hTERT-HME1); and MYC-ER HMEC (HME1 MYC-ER) were obtained from (10). Cells were tested for mycoplasma every 6 months. Last testing was done using MycoAlert kit (Lonza) in November 2022 and tested negative. Xanthosine (CDS020790), IMP (57510), and GMP (G8377) were purchased from Sigma (CDS020790).

Stable cell lines (MYC-ER HMECs and breast cancer cell lines MDA-MB-231-LM2 and SUM159) expressing pGIPZ or cDNA constructs were generated by infecting cells at a multiplicity of infection of 0.3 (pGIPZ shRNA) or 0.5 (pINDUCER cDNA) in the presence of 8 µg/mL polybrene. Transduced cells with pGIPZ shRNA plasmid were selected using 1 µg/mL puromycin. Single cell clones were generated for cells transduced with pINDUCER22 plasmid containing cDNA.

For Western blotting, cells were lysed in 1x Laemmli SDS sample buffer, incubated at 95°C for 10 minutes, centrifuged at 13,000 × g, and equal cell numbers were loaded onto criterian TGX stain-free precast gels (Bio-Rad) and separated by electrophoresis. Proteins were transferred on methanol-activated 0.45 µmol/L PVDF membrane (Thermo scientific) using semi-dry transfer by Trans-Blot (Bio-Rad). Following transfer, membrane was blocked in 5% BSA in TBS-T (0.1% Tween-20 in 1 × TBS) for 1 hour and incubated with primary antibody overnight at 4°C. The membrane was then washed three times in 1 × TBS-T and incubated with secondary antibody conjugated with fluorophore for 1 hour at room temperature. Following secondary antibody washing, membrane was imaged in respective fluorescence channel using chemi-doc (Bio-Rad). Antibody for BCL2 was purchased from Cell Signaling (Cat#4223; RRID: AB_1903909) and vinculin from Sigma (Cat#V9131; RRID: AB_477629)

For cell death assays in MYC-ER HMECs in response to inhibition of purine salvage (using shHPRT1, 6-MP, or DA-XV-55), MYC was induced with 4 to 12 nmol/L 4-OHT to allow for observation of a MYC-synthetic lethal effect. For cell death assays in MYC-ER HMECs using all other shRNAs or cDNAs, MYC was induced with 60 nmol/L 4-OHT. For cell death assays using propidium iodide (PI), cells were seeded at a density of 20,000 onto 12-well plates and 3,000 onto 96-well plates (Corning). MYC was induced the following day and 24 hours after MYC hyperactivation, the supernatant from the cells was collected. Cells were then trypsinized, mixed with the previously collected supernatants, and stained with PI (2 µg/µL) for 5 minutes. PI-positive dead cells were analyzed by flow cytometry (BD LSRFortessa) or by imaging using the Celigo Imaging Cell Cytometer (Brooks).

For annexinV assay, cells were trypsinized and stained with annexinV-APC conjugates from Thermo Fisher (Cat#A35110) according to manufacturer’s protocol. Stained cells were analyzed by flow cytometry (BD LSRFortessa).

For caspase-3/7 activity measurement, cells were cultured in white-walled 96-well plate (Corning) in 100 μL cell culture media. The next day, 100 µL Caspase-Glo 3/7 reagent (Promega, G8090) was added to the cell media, mixed briefly, and incubated in the dark for 30 to 60 minutes. Luminescence was measured on a luminometer (Molecular Devices). Luminescence readings were normalized to cell number determined by Hoechst 33342 (Life Technologies) staining of a duplicate 96-well plate (Greiner), followed by nuclei counting using the Celigo Imaging Cell Cytometer (Brooks).

For anchorage-independent growth assays, breast cancer cells were suspended in 0.3% low-melting agarose (Invitrogen) and plated on already solidified 0.5% low-melting agarose in culture medium. Media were changed every 3 days and macroscopic colonies (>125 µmol/L) were quantified after 2 weeks in respective treatment conditions.

For MTT assays, cancer cells MDA-MB-231-LM2 (LM2), SUM159, P493-6, MYCN3, SKBR3, and HCC1937 were seeded at a density of 3,000 onto 96-well plates (Greiner). After 6-MP treatment, cells were fixed and stained with MTT solution, lysed in 150 µL DMSO, and OD was measured at 550 nmol/L.

RNA isolation was performed with the RNeasy Mini kit (Qiagen) and TRIzol method (Invitrogen). Reverse transcription was performed using the High Capacity RNA-to-cDNA Master Mix (Applied Biosystems), and qPCR was performed using SYBR Green MasterMix (Applied Biosystems). qPCR primers were designed from PrimerBank (https://pga.mgh.harvard.edu/primerbank/; ref. 77). The following primers were used: DIS3L forward: 5′-GCCCACGAGCTTTGTGATTCT-3′, reverse: 5′-GCTGGCTCCTTGAAGTCGAA-3′; EXOSC8 forward: 5′-ACCCCTGTGTTCATCGAGATT-3′, reverse: 5′-CATCGTAGTCGAGGCAAATGAG-3′; EXOSC2 forward:

5′-CAATCACTACGGACACAGGATTC-3′, reverse: 5′-TTCGTCCCACTACGATGTCTC-3′; ACTB forward: 5′-TCCCTGGAGAAGAGCTACCAG-3′, reverse: 5′-TCGATGCCACAGGACTCCAT-3′; XDH forward: 59-CCAAATTGCTGCATGAACCAG-3′, reverse: 5′-TGCTTCCGAGGAGTGTCTTTC-3′; HPRT1 forward: 5′- CCTGGCGTCGTGATTAGTGAT-3′, reverse 5′-AGACGTTCAGTCCTGTCCATAA-3′.PVRL4 forward: 5′- GGAGGAGAGTGTAGGGCTGA-3′, reverse 5′- CGTGGTCAGCGTGGAGTAAC-3′.

Metabolites were extracted from the cells and quality controls as following. For cell lines with the indicated genotypes, 1 to 3 × 10⁶ cells were trypsinized, washed with 1x PBS, centrifuged, and snap frozen in liquid nitrogen. For metabolite extraction, frozen cell pellets were thawed at 4°C and were subjected to three freeze–thaw cycles on liquid nitrogen and ice, to rupture the cell membrane. Next, 750 µL ice-cold methanol:water (4:1) containing 20 µL spiked internal standards were added to each cell extract. Next, ice-cold chloroform and water were added in a 3:1 ratio to make a final proportion of 4:3:2 (methanol:chloroform:water). The organic (methanol and chloroform) and aqueous layers were dried and resuspended in 50:50 methanol:water. Next, the extracts were deproteinized using a 3-kDa molecular filter (Amicon Ultracel-3K Membrane; Millipore Corporation) and the filtrate containing the metabolites was dried under vacuum (Genevac EZ-2plus; Gardiner). Prior to mass spectrometry, the dried extracts were resuspended in identical volumes of injection solvent composed of 1:1 water:methanol and subjected to liquid chromatography-mass spectrometry (32, 78).

A total of 10 µL of suspended samples was injected and analyzed using a 6490 triple-quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA) coupled to an HPLC system (Agilent Technologies, Santa Clara, CA) via single reaction monitoring of the selected metabolites. Source parameters were as follows: gas temperature was 250°C; gas flow was 14 L/min; nebulizer was 20psi; sheath gas temperature was 350°C; sheath gas flow was 12 L/min; capillary was 3000 V positive and 3000 V negative; nozzle voltage was 1500 V positive and 1500 V negative. Approximately 8 to 11 data points were acquired per detected metabolite.

ESI positive mode was used in method A. The HPLC column was Waters XBridge Amide 3.5 µm, 4.6 × 100 mm. Mobile phase A and B were 0.1% formic acid in water and acetonitrile respectively. Gradient: 0 minute, 85% B; 3–12 minutes, 85% to 10% B, 12–15 minutes, 10% B, 16 minutes, 85% B, followed by re-equilibration end of the gradient, the 23 minutes to the initial starting condition 85% B. Flow rate: 0.3 mL/minutes.

For reverse phase separation, a Zorbax Eclipse XDB-C18 column (50 × 4.6 mm i.d.; 1.8 µm particle size, Agilent Technologies, Santa Clara, CA) maintained at 37°C, was used. The mobile phase used for the reverse phase separation contained 0.1% formic acid/water (v/v; A) and 0.1% formic acid/acetonitrile (v/v; B). The gradient conditions were 0 to 6 minutes, 2% B; 6.5 minutes, 30% B; 7 minutes, 90% B; 12 minutes, 95% B; followed by re-equilibration to the initial starting condition.

Metabolite peak areas were normalized to the internal standard metabolite peak area and log-transformed (base 2) to obtain the relative metabolite abundance.

MYC-ER HMECs engineered with DIS3L or control shRNA ±60 nmol/L 4-OHT were labeled with 1 mmol/L EU for 1 hour and fixed in 4% paraformaldehyde (PFA). For chase experiment, fresh media without EU but containing 2 µg/mL ActinomycinD was added to the cells. Cells were harvested 2 to 3 hours post chase by fixing in 4% PFA. EU Click-iT staining was performed according to manufacturer’s instructions (Thermo Fisher). EU signal was visualized by fluorescence microscopy and multiple images were collected using Nikon Eclipse Ti-E inverted microscope with 20X air objective and Andor Zyla 4.2 sCMOS camera. Fluorescence intensity for each cell was analyzed using NIS-Element AR 4.30 software. For each image, EU signal intensity was adjusted for background fluorescence by subtracting the mean extracellular pixel value.

HMEC cells engineered with DIS3L, XDH, or control shRNA ±60 nmol/L 4-OHT or ±6-MP were incubated with CellROX Deep Red reagent at a final concentration of 5 µmol/L (life technologies) for 30 minutes at 37°C. Cells were washed three times with PBS and fixed with 4% formaldehyde for 15 minutes. Fixed cells were visualized and imaged using a Nikon Ti-E inverted microscope with 20X air objective and Andor Zyla 4.2 sCMOS camera. For each treatment condition, cells were analyzed for mean DeepRed fluorescence intensity using NIS-Element AR 4.30 software. For each image, CellROX DeepRed signal intensity was adjusted for background fluorescence by subtracting the mean extracellular pixel value.

LM2 and SUM159 cells engineered with pINDUCER11-shDIS3L or control shRNA, which allows doxycycline-inducible shRNA and red fluorescence protein (RFP) expression. Each cell population was mixed with non-RFP LM2 or SUM159 cells. Mixed population was treated with increasing doses of 6-MP or DA-XV-55 and the proportion of shRNA expressing (RFP+) cells was analyzed by flow cytometry (BD LSRFortessa).

LM2 breast cancer cells were individually transduced with shRNAs (targeting RNA metabolism and nucleotide salvage genes) in pINDUCER11 backbone (75) at a multiplicity of infection of 1.2 to 1.5. All cell lines were sequentially pooled in equal ratios. The pool of cells was then subcutaneously transplanted (3 × 10⁶ cells per mice) into athymic nude mice (female mice, 4–6 weeks old, Harlan labs). Mice were randomized onto water with or without doxycycline (±DOX) 3 days post-transplantation. Tumors were measured using calipers and harvested at 1,000 mm³. Genomic DNA from dissected tumors was collected using the QIAamp DNA mini kit (Qiagen). Genomic DNA from tumor samples was isolated and shRNA library was amplified with the following primers: Forward: 5′- TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG TAG TGA AGC CAC AGA TGT A-3′, reverse: 5′- GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGCGCGGAGGCCAGATCTT-3′. Amplified shRNA library obtained from each tumor was indexed using Nextera Index Kit (96 indices) from Illumina and purified using PippinHT. Library was quantified using KAPA Library Quantification Kit and sequenced on Illumina HiSeq 2500 platform (∼10 × 10⁶ reads per tumor with a read coverage of >10,000 reads per shRNA per tumor).

Reads were processed to remove adaptor sequences using cutadapt (79) and then aligned to the reference library using Bowtie2 (80) in end-to-end mode allowing up to a maximum of three mismatches/indels compared to the reference sequence. The raw number of reads mapping to each shRNA in each sample was then extracted from the SAM files and DESeq2-v.1.14.0 (81) was used to determine the normalized abundance of each shRNA in the vehicle and doxycycline-treated tumors.

Investigators responsible for monitoring and measuring the xenografts of individual tumors were not blinded. Simple randomization was used to allocate animals to experimental groups. All animal studies were performed in accordance with institutional and national animal regulations. Animal protocols were approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine. Power analysis was used to determine appropriate sample size to detect significant changes in animal survival, which were based on previous survival analyses in our laboratory. All animals were included in analyses.

Total RNA (1 ug/sample) was used as input for the NEBNext rRNA Depletion Kit (Human/Mouse/Rat; NEB). After rRNA depletion, libraries were made using the NEBNext Ultra Directional RNA Library Prep Kit for Illumina, quantified using KAPA Library Quantification Kit, and sequenced on the NextSeq 500/550. After demultiplexing, reads were processed using SAMtools (v1.4) and aligned using Hisat2 (v2.0.4) with default parameters. The reference genome sequences (hg38/GRCh38) and gene annotation files were obtained from the UCSC Genome Browser. Gene expression FPKM values used for differential gene expression analyses were obtained using the cufflinks suite (v2.2.1). Analysis of gene expression deciles and plots were made using R (v4.0.3). RNA sequencing data for this experiment are deposited to the Gene Expression Omnibus (GEO) repository with the GEO accession nr. GSE223046.

The relative evolutionary importance of DIS3L residues was estimated by the ET method using homologous sequences with MUSCLE (65). The ET method computed importance rankings by systematically correlating residue variations with evolutionary divergences, in effect linking perturbations in sequence with perturbations in function as judged by phylogenetic distances. TCGA database was used to obtain mutation information on DIS3L residues. The impact of amino acid substitutions was estimated using the EA equation (65). ET ranks were aligned and colored onto the crystal structure of the human DIS3L homolog, Saccharomyces cerevisiae Dis3 (PDB ID 4IFD, chain J; ref. 66), using PyETV (PMID: 20929911). The Dis3 G833S mutation (G786S in DIS3L) was modeled using the serine rotamer with the least clashing in Coot (82). Figures were prepared with the PyMol visualization software (The PyMOL Molecular Graphics System, version 2.4.2 Schrödinger).

All experiments were performed on biological replicates (n ≥ 3) unless otherwise specified. Sample sizes for experimental groups and conditions are reported in the appropriate figure legends and methods. For cell culture experiments, sample size was not predetermined, and all samples were included in analyses.

Principal component analysis was performed on the catabolite data from Terunuma and colleagues (32) measuring the levels of 22 catabolites across 61 samples. A linear model was used to assess significance of the association between principal component 1 and MYC status.

For comparison of control and experimental groups, significance was determined using two-tailed unpaired Student t test unless otherwise specified in figure legends. Data were checked for similar variance before statistical analysis.

Predictions of hypothetical purine nucleotide flux (Supplementary Table S4) were developed using the published model (46, 47). The model was implemented as described except for several corrections necessary to match the reported steady state; the corrections, code to match the results in the original study (46), and code to create the figures in this paper, are available upon request. After running to steady state, we modeled addition of 25% of either RNA, over the course of 1 day, while allowing the model to relax back to steady state; the addition was in the form of a beta function to represent additional production due to MYC hyperactivation, with peak addition occurring around 5 to 8 simulated hours after starting the run. Implementation was in R using the routine “ode” from the package “deSolve” (83), with the default options.

Expression, copy number, and mutation data from TCGA were downloaded using TCGAbiolinks (84). RSEM normalized gene expression data from TCGA was obtained from the Broad GDAC Firehose. Copy number was obtained from TCGA as GISTIC data (85). Copy number analysis was done by segregating tumors based on a GISTIC score of >0 (“gain”) or 0 (“normal”) or <0 (“Loss”).

For gene ontology enrichment analysis, MYC-synthetic viable candidates (n = 223 genes) from the genome-wide screen were analyzed using PANTHER (http://pantherdb.org/; ref. 86) for enrichment using the C2 cellular component database. Overrepresentation was determined using a binomial test with FDR correction.

Requests for resources and reagents should be directed to T.F. Westbrook (thomasw@bcm.edu) and J.K. Meena (meena@bcm.edu). RNA sequencing data are deposited to the GEO with the accession number GSE223046. Metabolomics data related to Fig. 1A are published and available to the reader (32). Other data and codes used in this study are available upon request.

Description of specific terms used in this manuscript: Ribonucleotide catabolism, Catalytic breakdown of nucleotides containing ribose sugar; RNA catabolites, nucleotide base containing metabolites resulting from RNA degradation; MYC hyperactivation, increasing the levels of MYC protein in nucleus to supraphysiological levels; Oncogenic MYC, levels/activity of MYC that are associated with human cancer; MYC-dependent cells, cell lines that require MYC for cell growth; MYC-independent cells, cell lines that do not require MYC for cell growth.

J.K. Meena reports grants from Susan G. Komen during the conduct of the study; personal fees from Affirmativ Diagnostics DBA ADx Health outside the submitted work. E.A. Bowling reports grants from NIH/NCI during the conduct of the study. K. Eagle reports personal fees from Flare Therapeutics and personal fees from Dana Farber Cancer Institute outside the submitted work. A.D. Yang reports grants from CPRIT and NIH/NCI during the conduct of the study. S.J. Kurley reports grants from CDMRP and The Helis Foundation during the conduct of the study; personal fees from Optum Genomics (UnitedHealth Group) and personal fees from Castle Biosciences outside the submitted work. C.M. Olson reports personal fees from Marlinspike Therapeutics outside the submitted work. K.L. Meerbrey reports personal fees from Marlinspike Therapeutics outside the submitted work. A. Sreekumar reports grants from Agilent Foundation, non-financial support from Sri Sathya Sai Institute for Higher Learning, India, and personal fees from Karkinos Health Care Pvt. Ltd., India, outside the submitted work. D. Rejman reports grants from The Ministry of Education, Youth and Sports during the conduct of the study. C.Y. Lin reports other support from Kronos Bio outside the submitted work. M.C. Pillon reports grants from Cancer Prevention & Research Institute of Texas, National Institute of Environmental Health Sciences, and National Institute of General Medical Sciences during the conduct of the study. T.F. Westbrook reports grants, personal fees, and other support from Marlinspike Therapeutics outside the submitted work. No disclosures were reported by the other authors.

J.K. Meena: Conceptualization, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing, This author is also a co-corresponding author on this manuscript. J.H. Wang: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. N.J. Neill: Software, formal analysis, investigation, methodology, writing–review and editing. D. Keough: Resources, writing–review and editing. N. Putluri: Formal analysis. P. Katsonis: Software, formal analysis, writing–review and editing. A.M. Koire: Formal analysis. H. Lee: Investigation. E.A. Bowling: Investigation, writing–review and editing. S. Tyagi: Investigation. M. Orellana: Investigation. R. Dominguez-Vidaña: Software, formal analysis. H. Li: Software, formal analysis. K. Eagle: Software, formal analysis. C. Danan: Data curation, software, formal analysis. H.-C. Chung: Investigation, writing–review and editing. A.D. Yang: Investigation. W. Wu: Investigation. S.J. Kurley: Investigation. B.M. Ho: Formal analysis. J.R. Zoeller: Formal analysis, writing–review and editing. C.M. Olson: Investigation. K.L. Meerbrey: Investigation. O. Lichtarge: Formal analysis. A. Sreekumar: Investigation. C.C. Dacso: Resources, supervision, funding acquisition. L.W. Guddat: Resources, writing–review and editing. D. Rejman: Resources. D. Hocková: Resources. Z. Janeba: Resources, writing–review and editing. L.M. Simon: Data curation, software, formal analysis. C.Y. Lin: Data curation, software, formal analysis, writing–review and editing. M.C. Pillon: Software, formal analysis, writing–review and editing. T.F. Westbrook: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, visualization, methodology, writing–original draft, project administration, writing–review and editing.

J.K. Meena is supported by Susan G. Komen PDF Basic/Translational and Clinical Funding Program (PDF17487931). A.D. Yang is supported by NIH/NCI Ruth L. Kirschstein National Research Service Award (NRSA) Individual Fellowship (1F30CA278316-01). J.R. Zoeller is supported in part by Baylor Research Advocates for Student Scientists. D. Rejman is supported by The Project National Institute Virology and Bacteriology (Programme EXCELES, Project No. LX22NPO5103) grants from the European Union-Next Generation EU. J.H. Wang is supported in part by Baylor Research Advocates for Student Scientists. E.A. Bowling is supported in part by Baylor Research Advocates for Student Scientists and the NIH (T32GM120011). M.C. Pillon is supported by the Cancer Prevention & Research Institute of Texas grant (CPRIT, RR200076), the National Institute of Environmental Health Sciences (NIEHS, R00ES030735), and the National Institute of General Medical Sciences (NIGMS, R35GM147123). This project was supported in part by the following Advanced Technology Cores at Baylor College of Medicine: the Cytometry and Cell Sorting Core with funding from CPRIT Core Facility Support Award (CPRIT-RP180672), NCI P30 Cancer Center Support Grant (CA125123), and NIH (RR024574), the Metabolomics Core at Baylor College of Medicine with funding from the CPRIT Core Facility Support Award (RP210227), NIH (P30 CA125123), and Dan L. Duncan Cancer Center, and the Integrated Microscopy Core with funding from NIH (DK56338 and CA125123), CPRIT (RP150578 and RP170719), the Dan L. Duncan Comprehensive Cancer Center, and the John S. Dunn Gulf Coast Consortium for Chemical Genomics. C.Y. Lin is supported by the Cancer Prevention Research Institute of Texas (RR150093), the NIH, and NCI (1R01CA215452-01), and is a Pew-Stewart Scholar for Cancer Research (Alexander and Margaret Stewart Trust). C.C. Dacso is supported by Philip J. Carroll Excellence Fund, Rene and Kay Joyce Family Foundation, Flour Family Foundation. J.H. Wang and T.F. Westbrook are supported in part by The McNair Medical Institute (The Robert and Janice McNair Foundation). T.F. Westbrook is supported by the NIH and NCI (U01CA214125, 1R01CA215226, and 1R01CA215452), DOD (1W81XWH-18-1-0573), The Welch Foundation (Q-0007), Merck KgaA Open Innovation Program, and the CRUK Grand Challenge and the Mark Foundation for Cancer Research (C5470/A27144 to T.F. Westbrook as a member of the SPECIFICANCER Team).