Metabolic profiling reveals channeled de novo pyrimidine and purine biosynthesis fueled by mitochondrially generated aspartic acid in cancer cells

Pareek, Vidhi; Benkovic, Stephen

doi:10.1038/s41467-025-64013-w

Download PDF

Article
Open access
Published: 08 October 2025

Metabolic profiling reveals channeled de novo pyrimidine and purine biosynthesis fueled by mitochondrially generated aspartic acid in cancer cells

Nature Communications volume 16, Article number: 8952 (2025) Cite this article

1410 Accesses
9 Altmetric
Metrics details

Subjects

Abstract

Cancer cells have the unique capability to upregulate the de novo nucleotide biosynthesis supporting cell survival under nucleotide deprivation. We probe the role of metabolic channeling and membrane-less metabolic compartmentalization by mitochondria-proximal dynamic de novo pyrimidine and purine biosynthesis metabolons, the pyrimidinosome and the purinosome, respectively. We designed in-cell stable isotope label incorporation assays (¹³C6 glucose, ¹⁵N2 glutamine) for detection of metabolic channeling, revealing the function and enzymatic composition of these complexes. Moreover, we discovered that the mitochondrially compartmentalized GOT2 dependent generation of aspartic acid feeds the channeled nucleotide synthesis instead of the bulk cytosolic pool or the GOT1 activity. While a low flux diffusive pathway generates the pathway intermediates in an accumulative process, it’s the channeled pathway that successfully generates the end product nucleotides. Our results demonstrate how metabolic channeling and efficient de novo nucleotide biosynthesis is fueled by coordination of mitochondrially compartmentalized metabolic events with cytosolic metabolons in cancer cells.

Mitochondrial metabolism and cancer therapeutic innovation

Article Open access 04 August 2025

Nucleotide metabolism: a pan-cancer metabolic dependency

Article 27 March 2023

Computational modeling of cancer cell metabolism along the catabolic-anabolic axes

Article Open access 10 May 2025

Introduction

A hallmark of cancer cells is their capacity to rewire metabolism and maintain anabolic activities even under nutritional stress^1,2. Vital to their unrestrained growth is the capacity of cancer cells to maintain their cellular nucleotide pools by increasing the nucleotide de novo biosynthesis under the elevated metabolic demands, varying microenvironments, and limited access to serum derived pyrimidines and purines^3,4,5. Enhanced de novo nucleotide synthesis leads to poor cancer prognosis as well as limited response to purine antagonists. How do cancer cells achieve rapid upregulation of de novo nucleotide biosynthesis? This question becomes especially pertinent given that the pyrimidine and purine de novo biosynthesis are energy intensive pathways that involve unstable intermediates, and substrates (PRPP, glutamine, glycine, aspartic acid, formyl tetrahydrofolate) that are utilized by multiple enzymes, posing a challenge to the pathway flux^6,7,8. The recent discovery of the pyrimidinosomes and purinosomes, the respective pyrimidine and purine de novo biosynthesis enzyme complex, or metabolons, places them at the center of this mystery^6,9,10,11.

Metabolons are the dynamic complex of metabolic enzymes that lead to membrane-less compartmentalization of metabolic pathways^12,13. Metabolon formation leads to metabolic channeling characterized by restricted diffusion and increased local concentration of the intermediates generated within the metabolon, aiding in preferential utilization of the metabolon-bound intermediates by the successive enzymes of a pathway. Metabolons are also thought to sequester unstable intermediates, limit utilization of intermediates by alternative pathways, and increase the desired pathway flux, a feature especially significant for the nucleotide de novo biosynthesis. Therefore, active metabolons can significantly modulate and rewire metabolism, create routes for effective utilization of substrate pools compartmentalized within different organelles, and avoid thermodynamic and kinetic challenges to a pathway^12,14. There’s significant interest in understanding the molecular and metabolic basis of cancer cells’ remarkable adaptations and uncovering the role of dynamic metabolons in this process to uncover new anti-cancer therapeutic strategies targeting metabolon formation and function.

Based on fluorescence and electron microscopic imaging-based studies that revealed heterogenous cytosolic distribution of pyrimidine biosynthesis enzymes and their co-elution with the mitochondrial fraction, a mitochondria-proximal pyrimidine biosynthesis metabolon was envisioned^11,15. More recently, combining co-immunoprecipitation using transiently over-expressed tagged pyrimidine biosynthesis enzymes and fluorescence imaging, a pyrimidinosome metabolon was proposed⁹. However, in the absence of clear demonstration of metabolic channeling, distinction between the metabolic role of diffusive enzyme pool, partial enzyme complex, and a functional metabolon remains unclear.

The de novo pyrimidine biosynthesis pathway uses glutamine (Gln), aspartic acid (Asp), bicarbonate, and phosphoribosyl pyrophosphate (PRPP) to generate uridine monophosphate (UMP) in six consecutive steps (Figs. 1A and S1A). Yang et al. suggest that the pyrimidinosome is constituted by the (1) trifunctional enzyme carbamoyl-phosphate synthetase, aspartate transcarboxylase, and dihydroorotase (CAD), (2) mitochondrial membrane localized dihydroorotate dehydrogenase (DHODH), (3) cytosolic UMP synthetase (UMPS), and (4) the cytosolic glutamate-oxaloacetate transaminase or aspartate aminotransferase (GOT1), which together localize near the mitochondrial small molecule transporter VDAC3 to carry out efficient UMP synthesis⁹. The study proposes a role for GOT1 as the direct source of Asp for the pyrimidinosome complex. While GOT1 can carry out reversible conversion between oxaloacetic acid (OAA) and Asp, extensive work on cancer cell lines under normoxic conditions has established that GOT1 activity favors conversion of Asp to OAA, which maintains the malate-aspartate shuttle and the cytosolic and mitochondrial redox balance^16,17. In the context of a pyrimidine biosynthetic metabolon, the GOT1 activity may also appear to compete with CAD, which utilizes Asp. Therefore, direct participation of GOT1 in a pyrimidinosome metabolon needs further examination. Additionally, the steady state metabolomics examination, as reported in the above study, doesn’t allow unequivocal distinction between metabolically channeled versus the diffusive pathway flux. Therefore, we probed the presence of metabolic channeling in the pyrimidine biosynthesis pathway, enabling functional characterization of the pyrimidine biosynthesis metabolon, and its enzymatic constitution? We also explore the source of Asp for the pyrimidinosomes and the metabolic link between the mitochondria and the pyrimidinosome function?

**Fig. 1: Tracing the ¹⁵N labeled glutamine and aspartic acid incorporation in de novo synthesized pyrimidines.**

Similarly, a metabolon for the de novo purine biosynthesis (DNPB), the purinosome, formed by liquid-liquid phase separation, has been discovered in cancer cells⁶. Various fluorescence imaging, co-immunoprecipitation, and ensemble as well as spatial metabolomics by high resolution mass spectrometry imaging studies have revealed that the mitochondria-proximal purinosomes are constituted by at least nine purine synthesis enzymes—PPAT, GART, PFAS, PAICS, ADSL, ATIC, ADSS, IMPDH, and GMPS^6,12,18. The DNPB pathway starts with PRPP and utilizes Gln, glycine (Gly), Asp, CO₂, and formyl tetrahydrofolate (formyl-THF), and purines nucleotides, AMP and GMP, are generated in fourteen steps. Purinosomes carry out the pathway in a highly channeled manner, increasing the DNPB flux (~ten folds increase) compared to a low flux diffusive mechanism¹⁹. We have previously demonstrated that purinosomes preferentially utilize the mitochondrially generated Gly and formate, avoiding the impact of competition by other enzymes that utilize the bulk cytosolic pools of these substrates. Here we probe the source of Asp utilized in the channeled pathway and the metabolic impact of altered mitochondrial metabolism on purine synthesis.

Despite the implied role of metabolons in central metabolic pathways, because of the dynamic nature of the protein-protein interactions involved in the metabolon formation, it has been a challenge to reconstitute and functionally characterize them in vitro. This has also led to a heavy reliance on in situ fluorescence imaging-based assays or co-immunoprecipitation assays, often including tagging and overexpression of the involved enzymes, which fail to discriminate between the active complex versus inactive, partial, or alternative complexes made by a set of enzymes. To overcome these limitations, we designed in-cell isotope incorporation assays that helped us to (1) quantitatively assess the output of the active pyrimidinosome and purinosome metabolon, (2) infer the composition of the pyrimidinosome metabolon, and (3) uncover the coordinated metabolic exchange between mitochondria, a membranous organelle, and the nucleotide biosynthesis metabolons, purinosomes and pyrimidinosomes. Our results demonstrate the effectiveness of in-cell isotope labeling assays to detect and characterize alternative metabolic models of diffusive and channeled metabolism without the need of genetic manipulation or protein overexpression.

Results

In cell ¹⁵N2 glutamine incorporation assay reveals deviation from diffusive pyrimidine synthesis

Cancer cells import Gln from serum and rely on coordination of mitochondrial and cytosolic metabolism to generate aspartic acid to support nucleotide metabolism^20,21. Here, we addressed two aspects of pyrimidine de novo biosynthesis: (1) what is the mechanism of UTP synthesis, diffusive or a metabolon catalyzed channeled mechanism; (2) what is the source of Asp utilized for de novo pyrimidine synthesis-mitochondrial or cytosolic Asp? We leveraged the ¹⁵N2 Gln tracer approach to differentiate two competing models of pathway organization: diffusive versus pyrimidinosome metabolon catalyzed (Fig. 1A–C). A completely diffusive pathway implies the pool of all enzymes, substrates, and intermediates are present as a homogenous solution in the bulk cytosol with each reaction step occurring independently. Therefore, the model predicts uniform isotope incorporation patterns across substrates, intermediates, and products of the pathway (Fig. 1B). On the other hand, a mitochondria-associated pyrimidinosome metabolon would lead to deviations from such a diffusion model. Specifically, since mitochondria are the site of Asp generation from Gln in cancer cells^9,22,23,24, we hypothesized that a mitochondria-proximal metabolon would show enhanced isotope enrichment compared to a purely diffusive mechanism (Fig. 1C). Note that this metabolon relies on GOT2 as its source of Asp and the first pyrimidine de novo biosynthesis pathway enzyme CAD utilizes both Gln and Asp such that the resulting intermediates and end product nucleotides contain two nitrogen atoms (N₁ and N₃ of UMP)—one derived from Gln sidechain amide and the other from Asp backbone amine (Figs. 1A and S1). Based on the de novo pyrimidine biosynthesis pathway’s unique nitrogen incorporation pattern, we used ¹⁵N₂ Gln tracer that reports on both glutamine and the glutamine-derived aspartate-mediated nitrogen transfer events to probe metabolic channeling in the pyrimidine biosynthesis pathway.

We cultured HeLa cells in pyrimidine-depleted media for ~20 h before introducing 240 µM ¹⁵N₂ Gln in MEM-α media and performing metabolomics assessments (Figs. 1D, E and S2). Since Gln was limiting in our assays, we used early purine synthesis intermediate (FGAR, which incorporates only one nitrogen from Gln Fig. S2I) to infer the percentage label incorporation with respect to the Gln amide. FGAR showed 100% Gln derived ¹⁵N. However, both Glu and Asp showed only partial ¹⁵N labeling (~30%) (Figs. 1D and S2A). Further analysis revealed an unexpected labeling pattern, both pyrimidine biosynthesis intermediates and uridine nucleotide showed higher ¹⁵N Asp mediated enrichment compared to the bulk cytosolic Asp (Figs. 1D, E and S2B–E). These results contradict predictions of a purely diffusive model, which would generate homogeneous isotope incorporation across all pathway intermediates and end products (Fig. 1B). Instead, our data suggests that pyrimidine nucleotides arise from an Asp pool with higher label enrichment compared to bulk cytosolic Asp. Given that the labeled cytosolic Asp pool can either be generated by mitochondrial GOT2 or by cytosolic GOT1, this pattern suggests either CAD preferentially accesses mitochondrially-generated Asp (Figs. 2C and S2F), or a GOT1-CAD complex generates labeled Asp proximal to CAD, which is then utilized for pyrimidine synthesis (Fig. S2G). ¹⁵N2 Gln labeling is insufficient to resolve the two scenarios.

**Fig. 2: Detection of channeling in the pyrimidine synthesis pathway.**

Additionally, significant accumulation of the pyrimidine biosynthesis intermediates—carbamoyl aspartate, dihydroorotate and orotate—observed in our assays, also suggests diffusive synthesis, accumulating the intermediates and allowing their equilibration of with the bulk cytosol (Fig. S2H). The ¹⁵N₂ Gln incorporation assay clearly demonstrates that bulk cytosolic Asp may not serve as the direct substrate for pyrimidine biosynthesis and rules out a completely diffusive model (Fig. 1B). However, it cannot definitively distinguish between the partially diffusive (Fig. S2F, G) and channeled pathway mechanisms (Fig. 1C) or clearly differentiate between GOT2 versus GOT1 being the Asp sources. To address this, we developed a complementary ¹³C6 glucose tracing assay.

Metabolic channeling mechanism generates uridine nucleotide

We designed a ¹³C6 Glc labeling experiment that relies on the mitochondrially localized pyruvate and Gln metabolism enzymes leading to spatial compartmentalization and temporal separation of the Asp isotopologues produced within mitochondria from that generated in the cytosol. HeLa cells were seeded in purine depleted media (with 10% dialyzed FBS) and incubated for 20 h before changing to the media without glucose (with 10% dialyzed FBS) spiked with ¹³C6 Glc (final concentration 5.4 mM). ¹³C6 Glc incorporation was performed for 2, 4, 6, 8, 14 h. Executing this labeling at various lengths of time allowed the assessment that incomplete diffusion of metabolites into cells is not biasing our findings. ¹³C6 Glc label incorporation performed for 6 h allowed ¹³C incorporation into various tricarboxylic acid (TCA) cycle metabolites, amino acids, pyrimidine de novo biosynthesis intermediates and end products as shown in (Figs. 2 and S4–S6). By tracing ¹³C label incorporation, we aimed to differentiate the products of the enzymatic pool that utilizes mitochondrially generated Asp from one that utilizes the cytosolic Asp pools. By examining the isotope labeling patterns of TCA cycle metabolites after 6 h label incorporation, we inferred the origins of various isotopologues of Asp (Figs. 2A, B, S5, and S6A, B). We hypothesized that if pyrimidine biosynthesis follows a completely diffusive mechanism utilizing the bulk cytosolic Asp, the isotopologue pattern in various intermediates and uridine nucleotide may be predicted from the pattern observed in Asp (Fig. 2B, C). Figure 2C describes the mathematical model that allowed us to compute the various isotopologue fractional abundance calculations for the pathway intermediates and UTP. The model was applied to each individual experiment and the predictions were compared with the observed values in the same experiment.

Since all carbons from Asp are utilized in the synthesis of pyrimidine biosynthesis intermediates, following diffusive mechanism, our model predicts the same fractional abundance of various isotopologues in the pathway intermediates as that observed in the bulk Asp. Our analysis revealed that indeed the isotopologue fractions in bulk Asp and the various pyrimidine biosynthesis intermediates (carbamoyl aspartate, dihydroorotate, and orotate) are similar, supporting a diffusive and equilibrative mechanism of generation of these bulk intermediate pools (Figs. 2D, E and S6D, E). However, the predicted isotopologue pattern in UTP was significantly different from that observed in Asp and the intermediates (Figs. 2F, G and S6F). This indicates that the bulk Asp and the intermediate pools are not the primary source of the newly generated observed UTP and instead suggest a metabolic channeling mechanism for UTP synthesis (Fig. 2H). Note that a higher abundance of the (M + 8) UTP isotopologue was observed compared to that predicted based on the bulk cytosolic Asp, indicating preferential utilization of the mitochondrially generated Asp. Since Asp (M + 4) isotopologue generates the UTP (M + 8) isotopologue, we inferred that the Asp pool utilized for channeled UTP synthesis must arise from a pool of Asp with a higher percentage of the (M + 4) isotopologue compared to that observed in the bulk cytosolic Asp. Under our experimental conditions, Asp (M + 4) primarily originates from the action of mitochondrial PC, therefore, a higher UTP (M + 8) isotopologue enrichment is indicative of the proximity of the enzyme pool responsible for channelled UTP synthesis to the mitochondria (Figs. 2H and S5). From the steps that showed metabolic channeling, we infer that such a pyrimidinosome complex must consist of the enzymes CAD, DHODH, and UMPS coordinating pyrimidine synthesis with mitochondrial metabolism.

Preferential utilization of mitochondrially generated aspartic acid for channeled pyrimidine synthesis

We carried out additional experiments to test if the pyrimidinosome mediated metabolic channeling indeed depends on the mitochondrial pool of Asp. One of the assumptions of the channeled synthesis model was the compartmentalized generation of Asp (M + 4) isotopologue in mitochondria owing to the mitochondrial PC activity and TCA cycle mediated CO₂ generation working in parallel (Figs. 3A and S5). We tested this assumption by pre-treating cells with antimycin A (6 µM), an electron transport chain poison that increases FADH₂/FAD ratio, for 3 h. Upon this treatment, we observed the accumulation of succinate, and inhibition of TCA cycle progression (Fig. 3B). As expected, despite the initiation of TCA cycle, as indicated by citrate isotopologues (Fig. S6G), upon antimycin A treatment followed by 8 h ¹³C6 Glc label incorporation, we observed no (M + 2) isotopologues in succinate, malate, OAA, and Asp, indicating TCA cycle inhibition (Figs. 3C–E and S6H). Further supporting our assumption, we also observed a high abundance of the (M + 3) isotopologue in OAA, malate, and Asp and absence of the (M + 4) isotopologue in all of these metabolites. Similarly, all the pyrimidine synthesis intermediates also showed high (M + 3) isotopologue and no (M + 2) and (M + 4) isotopologues (Fig. S6I).

**Fig. 3: Channeled de novo UMP synthesis preferentially utilizes mitochondrially generated aspartic acid.**

We further hypothesized that if PC activity is responsible for the generation of the OAA and Asp (M + 4) isotopologues, reduction in the PC activity would lead to reduction in the UTP (M + 8) isotopologue (Fig. 3F). To test this, we carried out siRNA mediated PC knockdown, followed by 6 h ¹³C6 Glc label incorporation. Upholding our hypothesis, we observed a drop in the fractional abundance of the Asp (M + 4) and UTP (M + 8) isotopologue upon PC knockdown compared to that in the control siRNA treated sample (Figs. 3G and S7A–C).

To further interrogate our interpretation that the pyrimidinosome mediated uridine nucleotide generation relies on the mitochondrially generated Asp, which in turn arises from the PC and TCA activities, we hypothesized a temporal dependence of the accumulation of the (M + 4) isotopologues of Asp and the pyrimidine biosynthesis intermediates as well as the (M + 8) isotopologue of UTP. Supporting our hypothesis, we observed that the abundance of the (M + 4) isotopologues was negligible in Asp and orotate at a shorter ¹³C6 Glc labeling times (2 and 3 h), where limited mitochondrial ¹³C CO₂ is expected (Fig. S8D, E). Concurrently, the fractional abundance of (M + 8) UTP isotopologue after 2 h labeling was also significantly lower than that in the 6 h labeling experiment (Fig. 3H). Moreover, we discovered a clear temporal dependence of the (M + 8) UTP isotopologue generation upon 2–8 h of ¹³C6 Glc incorporation (Fig. 3I). This further substantiated a channeling mechanism that preferentially utilizes the mitochondrially generated Asp before it equilibrates with the bulk cytosolic pools.

Next, we reasoned that prolonged ¹³C6 Glc labeling (14 h) in HeLa cells would lead to accumulation and equilibration of the mitochondrial Asp with the cytosolic pool, thus accumulating a significantly greater (M + 4) Asp as well as the pyrimidine biosynthesis intermediates. Unlike labeling patterns observed for 2–8 h, we observed a significant accumulation of (M + 4) isotopologues in Asp as well as all the TCA cycle intermediates at 14 h of labeling (Fig. S8A–D). In line with our prediction, the pyrimidine biosynthesis intermediate orotate also showed a significantly greater fraction of the (M + 4) isotopologue (Fig. S8E). Unlike the early time points till 8 h, the observed UTP (M + 8) isotopologue abundance was less than that predicted by our model (Figs. 3J, 2G, and S8H–J). Our results confirm that from 2 to 8 h, the (M + 3) and (M + 4) Asp isotopologues arise primarily from PC activity, while at 14 h, TCA cycle also contributes to these isotopologues (Fig. S8M, N).

Our model for channeled pyrimidine synthesis (Fig. 2H) implies that the product UMP, but not the diffusive synthesis of the intermediates would be sensitive to the mitochondrial Asp production (Fig. 3K). To test this, we carried out siRNA mediated knockdown of GOT2 in HeLa cells. After 48 h of siRNA treatment, metabolite estimation was performed (Figs. S7D–E and 3L, M). To uncover the effect of GOT2 knockdown, we plotted the substrate/product (Asp/UMP, Fig. 3L), and intermediate/product ratios (dihydroorotate/UMP and orotate/UMP, Fig. 3M) for the non-target control siRNA and GOT2 siRNA treated cells, respectively. Supporting our model where the mitochondria-proximal pyrimidinosomes utilize the mitochondrially generated Asp, we observed an increase in both the substrate/product as well as the intermediate/product ratios. These findings are in line with previous studies showing that cancer cells with low or no GOT2 activity rewire to either increased dependence on GOT1 activity or acquire Asp from the media, thus maintaining their cytosolic pools^25,26. As indicated by the metabolite pool comparisons, since the bulk cytosolic Asp is inadequate at supporting channeled nucleotide synthesis, GOT2 knockdown suppresses UMP generation, while the production of various intermediates by the diffusive mechanism goes unperturbed (Fig. S7E).

As our ¹⁵N2 Gln labeling experiment had indicated plausible involvement of GOT1 in the pathway, we sought to clarify the role of the cytosolic GOT1 as an alternative source of Asp for the channeled or the diffusive mechanisms. We explored the metabolic profile of HeLa cells upon GOT1 knockdown achieved by siRNA treatment (Figs. 3N, Q–S, and S7F–J). Previous studies have revealed a dual role of the reversible cytosolic enzyme GOT1 in cancer cells^16,27,28. On the one hand, GOT1 generates OAA from Asp, which is then converted to malate by the cytosolic malate dehydrogenase (MDH1), playing an important role in regulating cytosolic redox balance and the mitochondria-cytosolic proton gradient necessary for electron transport chain (ETC). The cytosolic conversion of Asp to malate is coupled with the mitochondrial conversion of malate to Asp and a directional flow of metabolites across the mitochondrial membrane through the malate aspartate shuttle (MAS). On the other hand, acting in the reverse direction, GOT1 can also convert malate to Asp, especially in cells lacking GOT2 activity or upon electron transport chain (ETC) inhibition^27,29,30. We hypothesized that if GOT1 primarily acts as the Asp supplier for the channeled nucleotide biosynthesis pathways, GOT1 knockdown may lead to the downregulation of the pyrimidine biosynthesis from top down, since the first pyrimidine biosynthesis enzyme requires Asp (Fig. S1A). Moreover, GOT1 knockdown may also diminish the channeled pathway flux. Conversely, if GOT1 promotes pyrimidine biosynthesis via its role as a redox balancing enzyme (Fig. 3N), GOT1 knockdown may act similar to ETC inhibitors like antimycin A, which inhibit DHODH activity due to its dependence on the availability of mitochondrial oxidized ubiquinone (Fig. 3O, P). Similar to the effect of antimycin A treatment, GOT1 knockdown also led to accumulation of the pyrimidine biosynthesis intermediates carbamoyl aspartate and dihydroorotate upstream of the DHODH activity (Figs. 3Q, R and S7G–I). This is indicated by an increase in the CA to orotate and DHO to orotate ratios, respectively, upon GOT1 knockdown. Additionally, GOT1 knockdown neither significantly alters the Asp to UMP or the orotate to UMP ratios, nor the isotopologue distribution in UTP (Figs. 3S and S7G–I). These results support the interpretation that GOT1 does not act as the primary source of Asp for the channeled nucleotide synthesis but rather acts to maintain the cytosolic redox balance and the diffusive generation of the intermediates.

Overall, our data shows that the pyrimidinosome metabolon-bound intermediates are preferentially processed to generate uridine nucleotides, while in parallel, the diffusive pathway only accumulates the pyrimidine synthesis intermediates. Pyrimidinosome formation disallows intermixing of the metabolon-bound and the bulk cytosolic pools of various intermediates. The two pathways also use two different sources of Asp. The channeled pathway preferentially incorporates the mitochondrially generated Asp, while the diffusive pathway uses the bulk cytosolic Asp. While under our experimental conditions, uridine nucleotides are primarily generated by the channeled pathway, further studies are required to find the contribution of the diffusive pathway to the total UMP flux at steady state condition. Alternatively, the pyrimidine biosynthesis intermediates may be excreted out from the cell, or orotate may plausibly diffuse into the nucleus to support the activity of UMPS localized within the nucleus³¹. Our data suggests that the pyrimidinosome does not utilize the cytosolic GOT1 activity as the primary source of Asp. First, GOT1 mediated conversion of Asp to OAA would compete with the aspartate transcarbamylase activity of CAD, thus hindering the conversion of carbamoyl phosphate (a thermodynamically unstable intermediate) to carbamoyl aspartate, potentially making the pathway counterproductive. Second, as the mitochondrial Asp produced by GOT2 is acquired for the pyrimidinosome function via the mitochondrial aspartate glutamate antiporter (AGC), it may be supported by the exchange of Glu, produced during the first step catalyzed by CAD⁹. The importance of the diffusive pathway, supported by GOT1 activity, may be to allow continued DHODH activity supporting mitochondrial respiration and ferroptosis regulation instead of efficient UMP production^{9,32,33,34,35}, especially under hypoxia or when the electron transport chain is disrupted^9,36.

Preferential utilization of mitochondrially generated aspartic acid for channeled purine synthesis

Next, we probed if the purinosomes utilize bulk cytosolic or the mitochondrially generated Asp, one of the nitrogen donors for the purine ring, for the channeled purine synthesis (Fig. S1B, C). For this, we leveraged ¹⁵N2 Gln labeling as described in the “Methods” section. In this experiment, ¹⁵N2 Gln acts as the nitrogen donor in the first step of de novo purine biosynthesis (DNPB), where the enzyme PPAT converts PRPP to 5-phosphoribosyl amine (5-PRA), and in the third step of DNPB, the enzyme PFAS converts formylglycinamide ribonucleotide (FGAR) to formylglycinamidine ribonucleotide (FGAM) (Fig. 4A). Simultaneously, the ¹⁵N labeled Asp generated during the course of the experiments is also utilized in the reactions catalyzed by the enzymes PAICS and ADSS producing the intermediates phosphoribosyl aminoimidazole succinocarboxamide (SAICAR) and succinyl adenosine monophosphate (SAMP), respectively (Figs. 4A and S1B, C). These two intermediates in the pathway are processed to phosphoribosyl aminoimidazole carboxamide (AICAR) and AMP, respectively, by the enzyme ADSL in non-sequential steps, releasing fumarase in each step, while retaining the nitrogen to build the purine base ring. We modeled the DNPB assuming a diffusive mechanism that allows homogenous ¹⁵N isotope incorporation with respect to the bulk ¹⁵N2 Gln as well as ¹⁵N Asp (Fig. 4A). We assessed the model by examining the observed isotopologue distribution patterns in the DNPB pathway intermediates and the end product nucleotide, AMP.

**Fig. 4: Preferential utilization of mitochondrially generated ¹⁵N labeled aspartic acid for channeled de novo purine synthesis.**

Note, as previously described, our extraction method primarily solubilizes bulk metabolites, leaving the metabolon-bound intermediates trapped within¹⁹. Therefore, the observed DNPB intermediates report on the diffusive synthesis in the cytosol, while AMP and GMP represent the channeled DNPB products. 5-PRA, the first intermediate with Gln mediated ¹⁵N incorporation is not observed due to its instability³⁷. Therefore, based on the downstream stable intermediate FGAR, we ascertained that the ¹⁵N incorporation fraction in the steps that utilize ¹⁵N2 Gln was close to 100% (Fig. 4B). Next, the examination of SAICAR revealed ~23% isotope incorporation fraction for the ¹⁵N Asp derived label (%¹⁵N Asp_SAICAR (x) = 100*peak area [M + 3]/total peak area [M + 2] + [M + 3]) (Fig. 4C). Under these conditions, the cytosolic bulk Asp shows 28–30% ¹⁵N incorporation (Fig. 1D) and the slight label dilution in the successive purine synthesis steps supports diffusive SAICAR synthesis (Fig. 4C). We hypothesized that if the channeled and the diffusive pathway utilize the same source of ¹⁵N Asp, our model that uses the isotopologue fractional abundance in SAICAR (Fig. 4A, C), would allow accurate prediction of the isotopologue patterns in ATP (%¹⁵N Asp_ATP). Therefore, using the %¹⁵N Asp incorporation observed in SAICAR, we calculated the isotopologue distribution expected for ATP (Eq. 5 Supplementary Information). Notably, we observed that thus predicted isotopologue abundance and the observed abundance patterns for ATP were significantly different (Fig. 4D). Moreover, using our model, we also back calculated the %¹⁵N Asp enrichment that must have been utilized to generate the observed ATP isotopologue distribution (Fig. 4E). This comparison revealed that the de novo synthesized AMP had ~20% higher label incorporation with respect to ¹⁵N Asp, compared to that utilized for the diffusively generated SAICAR production. Our results support the purinosome model where it preferentially utilizes the mitochondrially generated Asp, and the intermixing of the purinosome-bound intermediates with the bulk cytosolic pools of the intermediates is restricted (Fig. 4F). Our findings highlight the functional dependence of purinosomes on mitochondrial metabolism to acquire building blocks (Gly, Asp, and formate) for purine synthesis. Based on the metabolically channeled steps, we inferred the composition of the active purinosome to include the DNPB enzymes: PPAT, GART, PFAS, PAICS, ADSL, ATIC, ADSS, IMPDH, GMPS, confirming our previously proposed model for the mitochondria-proximal active purinosomes^6,19.

To confirm the contribution of the mitochondrial GOT2 activity to the channeled purine synthesis, we carried out siRNA mediated knockdown of GOT2 and GOT1 in HeLa cells (Figs. S7F and S9B–E). After 48 h of siRNA treatment, metabolite estimation was performed. We predicted that the channeled nucleotide synthesis flux, but not the diffusive synthesis of the intermediates would depend on the mitochondrial Asp production (Fig. 4G). The ratios of substrate/product (Asp/AMP and Asp/GMP Fig. 4H), and intermediate/product (IMP to AMP/GMP ratios, Fig. 4I) total abundances for the non-target control siRNA and GOT2 siRNA treated cells, respectively, were plotted. Validating our prediction, we observed an increase in both the substrate/product as well as the intermediate/product ratios. Our findings reveal that the bulk cytosolic Asp is inadequate at supporting channeled nucleotide synthesis, and therefore GOT2 knockdown suppresses AMP as well as GMP generation, while the production of various intermediates by the diffusive mechanism goes unperturbed (Figs. 4H, I, and S9B). We also probed a plausible role of the cytosolic GOT1 as the source of Asp in channeled or diffusive purine synthesis. Contrary to the effect of GOT2 knockdown, GOT1 knockdown had no effect on AMP/GMP generation (Fig. S9C–E), substantiating our conclusion that purinosome function relies on GOT2 but not GOT1 activity.

Discussion

There are two prerequisites to advantageously deploying metabolic channeling to achieve dynamic metabolic rewiring and efficient nucleotide synthesis: (1) formation of specific enzyme complexes to ensure effective handover of the intermediates between the successive enzymes; and (2) aligning metabolic pipelines to feed all the required substrates and coenzymes/co-factors for the participating enzymes to allow unhindered pathway operation. The latter can become a metabolic bottleneck, especially in cases where the substrates are generated in a different membraned compartments, like mitochondria, chloroplasts, or lysosomes than its location of consumption. Recent advances in biochemical and cell biological approaches have enabled the discovery of several cytosolic multi-enzyme complexes. Still, understanding the functional status of various active multi-enzyme complexes, probing their role in cellular metabolic networks, and distinguishing them from inactive aggregates or partial complexes remains challenging. Our findings demonstrate metabolic channeling through spatially organized mitochondria-proximal pyrimidinosomes and purinosomes (Figs. 2H and 4F), revealing their composition and highlighting coordinated metabolism across cellular compartments. We relied on the strength of in-cell stable isotope labeling assays for functional characterization of metabolons and probing their interdependence on mitochondria. We hypothesized that Asp generated in mitochondria and that in the cytosol may be differentiated in a limited duration labeling experiment, disallowing complete equilibration of the two pools. Under these conditions, examination of the isotopologue patterns in various metabolites reveals: (1) the effect of label dilution as mitochondrially generated metabolites are exported to cytosol (Malate, Asp), or the unlabeled cytosolic metabolites are imported into the mitochondria (Gln); and (2) the temporal dependence of metabolic events. This allowed us to uncover the molecular origins of the labeled carbon atoms in pyrimidine and purines intermediates and end products. Based on this, we developed simple mathematical models that used the isotopologue distribution in the pyrimidine and purine biosynthesis intermediates to predict that expected in the diffusively generated end products, respectively (Figs. 2C and 4A).

These assays facilitated the examination of the origins of different labeled atoms by considering the known subcellular localization of the correlated enzymes and the observed isotopologue patterns of various substrates, intermediates, and the end products involved in the nucleotide de novo biosynthesis. We infer that while the diffusive synthesis in both the pathways does generate various intermediates in an equilibrative manner in the bulk cytosol, intermixing of the intermediates with the channeled intermediates’ pools is restricted, indicating sequestration of the channeled intermediates by the respective metabolons. Notably, by combining extensive ¹⁵N2 Gln and ¹³C6 Glc label tracing (Figs. 1, 2, and 4A–F) with pharmocological or siRNA mediated purturbation of mitochondrial metabolism (Fig. 3A–G), we show that the mitochondrially generated Asp is preferentially utilized for channeled nucleotide de novo biosynthesis and is dependent on the mitochondrial GOT2 activity rather than the GOT1 activity (Figs. 3K–S and 4H, I). These findings provide insights into the functional importance of the subcellular localization of the nucleotide synthesis metabolons. As shown previously, the cytosolic GOT1 activity may indirectly regulate the pathway via its role in maintenance of the mitochondrial malate-aspartate shuttle, cytosolic redox balance, mitochondrial membrane potential, hence also serine biosynthesis and one carbon metabolism^16,17,27 (Fig. 3N–R). Expanding on the recent work by Yang et al.⁹, our data suggests the composition of the mitochondria-associated functional pyrimidine biosynthetic metabolon, the pyrimidinosome, to be CAD and UMPS along with the mitochondrial enzyme DHODH (Fig. 2H), and confirms our previously proposed model with purinosomes constituted by the following purine biosynthesis enzymes: PPAT, GART, PFAS, PAICS, ADSL, ATIC, ADSS, IMPDH, GMPS (Fig. 4F)^6,19.

Our data doesn’t rule out the possibility of partial complexes or some leakiness in the two metabolons contributing to the generation and accumulation of intermediates in the bulk cytosol. It also remains to be investigated why the channeled pathway successfully generates the end product, while the respective diffusive mechanism primarily accumulates the intermediates. Both in vitro and in vivo studies have uncovered varying tissue specific reliance on the salvage versus the de novo nucleotide biosynthesis and differences between cancerous and non-cancerous cells^38,39,40. However, contributions of the channeled mechanism in vivo de novo nucleotide synthesis remains to be probed. Metabolomics examination at high concentration of labeled substrates (eg. 10 mM Asp, ~1000 folds higher than the plasma concentration, 1–2 mM Gln)⁹ and long label incorporation time (8 h or longer, Figs. 3J and S8K–N)¹⁹ can both lead to substantial label scrambling and potential rewiring of cancer metabolism. This can make distinction between the channeled and the diffusive pathway flux difficult and functional characterization of a metabolon challenging. We envision the use of the non-equilibrium isotope labeling workflow developed here to aid metabolic feature-based detection and characterization of nucleotide biosynthesis metabolons in vivo.

Our results also uncover the coordination of metabolic events compartmentalized within a membranous organelle mitochondrion and nucleotide synthesis metabolons, the cytosolic dynamic membrane-less organelles. Direct uptake and preferential utilization of mitochondrially generated substrates by the cytosolic pyrimidinosome and purinosome metabolons imply the proximity of these metabolons to the mitochondrial transporters, plausibly creating a conduit through the mitochondrial membrane (Figs. 1C, 2H, and 4F). In line with our interpretation, proximity of the pyrimidine biosynthetic enzymes to the mitochondrial VDAC3⁹, and purine biosynthesis enzymes to the mitochondrial membrane has been demonstrated previously⁴¹. Based on our prediction we have recently also confirmed the proximity of purine de novo biosynthesis enzymes to the mitochondrial transporters⁴². The reliance of cancer cells on mitochondrial Asp for nucleotide synthesis highlights the central role played by mitochondria both as the energy powerhouse as well as the source of substrates for anabolic activities in cancer cells^43,44. Through mitochondrial glutaminolysis, followed by Glu processing to generate α-KG and Asp, Gln both provides carbon as well as nitrogen for nucleotide synthesis. These results help to explain the heightened reliance of cancer cells on serum Gln and the correlation of mitochondrial GLS, PC, and GOT2 expression with aggressiveness of certain cancers, their immune evasion, and resistance to chemo and radiotherapies^5,45,46,47. Our results provide impetus for targeting mitochondrial metabolism for regulating channeled de novo nucleotide synthesis in cancer cells.

Since both pyrimidine and purine biosynthesis generate Glu upon utilization of Gln, it’s tempting to postulate that the proximity of the two metabolons to the mitochondrial aspartate/glutamate antiporters may support direct exchange of Glu produced within pyrimidinosome and purinosomes with mitochondria, thereby supporting the acquisition of mitochondrial Asp by the two cytosolic metabolons. In the context of a recent study that revealed different mitochondrial subpopulations with distinct compartmentalized metabolic pathways⁴⁸, it will be interesting to probe if the nucleotides released locally in the vicinity of mitochondria supplement the mitochondrial ATP synthase activity or are imported into the mitochondrial matrix, supporting the mitochondrial metabolism and DNA replication. Further studies will be required to fully uncover the role of metabolic compartmentalization by pyrimidinosomes and purinosomes and the subcellular localization of these metabolons in cancer cells and investigate the differences with the non-tumorigenic cells.

Methods

Cell culture

HeLa cells were maintained and propagated in purine and pyrimidine rich media (P+, DMEM, with glutamine + 10% FBS) without any antibiotics. Typically, HeLa cells passaged upto generation 16 were used for various experiments. For inducing purine and pyrimidine depletion, cells were seeded in purine and pyrimidine depleted (P-, DMEM or RPMI + 10% dialyzed FBS) media for 20 h. For preparing dialyzed FBS (dFBS), FBS was extensively dialyzed against 0.15 M NaCl solution using 1 kDa cut-off membrane and filtered before use.

¹⁵N2 glutamine (¹⁵N2 Gln) label incorporation and tracing

Cells were maintained as described above and before labeling, ~3–3.5 × 10⁶ seeded in a 10 cm dish with P- RPMI media for 20 h. Then, cells were washed gently with 1x PBS and the media was changed to MEM-α (medium without ribosides, deoxyribosides, and glutamine; #15-012CV, Corning Inc.) with 10% dFBS. Media was supplemented with ¹⁵N2 glutamine (¹⁵N2 Gln, NLM-1328-MPT-PK, Cambridge Isotope Laboratory, Inc., USA; final concentration 240 µM) and cells were incubated for another 2 h at 37 °C to allow isotope label uptake and incorporation. After this, metabolite extraction was performed using cold 80% methanol (MeOH).

Dually labeled ¹⁵N2 Gln is utilized in the cytosolic pyrimidine and purine de novo synthesis (Figs. 1, 4, S1A, and S2). The pyrimidine biosynthesis enzyme CAD and the purine biosynthesis enzymes PPAT and PFAS use ¹⁵N2 Gln and generate ¹⁵N glutamic acid (Glu), labeled at the backbone amine. Alternatively, mitochondria metabolize the majority of ¹⁵N2 Gln, acting as the loci for transferring the ¹⁵N isotope label from Gln to Glu and Asp. Once ¹⁵N2 Gln is imported into the mitochondria, through glutaminolysis by the mitochondrial glutaminase (GLS), ¹⁵N Glu is produced. Then, ¹⁵N Glu and OAA are utilized by GOT2 within mitochondria, producing α-KG, transferring the ¹⁵N label to Asp (Figs. 1, 4, S1A, and S3). Some ¹⁵N Glu may be exported to the cytosol to be utilized by the cytosolic GOT1. The redox balance of cytosolic versus mitochondrial metabolism in cancer cells supports GOT2 forward reaction generating Asp within mitochondria, thus creating high ¹⁵N Asp local concentration in mitochondria.

On the other hand, cytosolic Asp is regenerated either from efflux of Asp from mitochondria or lysosomal protein degradation, or the activity of cytosolic GOT1 (Fig. 1B). Cytosolic GOT1 is a reversible enzyme, known to reversibly convert Asp and oxaloacetic acid (OAA), consecutively also interconverting Glu and α-KG. So, when the mitochondrial malate is transported to the cytosol, it is utilized by the cytosolic malate dehydrogenase (MDH1), producing cytosolic OAA (Figs. 1B, 2E and S5). Cytosolic OAA may then be converted to Asp by the cytosolic enzyme GOT1, providing an alternative source of Asp for various processes. Since mitochondrial Asp is utilized for nucleotide synthesis, there’s only limited accumulation and export. Therefore, in short duration experiments, the bulk cytosolic pool of Asp doesn’t fully equilibrate with the mitochondrial pool. The time and concentration of ¹⁵N2 Gln was standardized to achieve ~100% isotope incorporation with respect to glutamine label in the various pyrimidine and purine intermediates (Figs. S2B–E and I, 4B, C), but ~30% isotope label in bulk Glu (Fig. S2A) and Asp (Fig. 1D). In these experiments, glutamine was limiting and hence not always observable in our extracts. Therefore, we relied on the examination of nucleotide biosynthesis intermediates (FGAR and carbamoyl aspartate) to determine the appropriate ¹⁵N Gln concentration that is enough to achieve 100% ¹⁵N incorporation derived from ¹⁵N Gln (Figs. 4B and S2B, I).

Due to a high rate of interconversion of nucleotide-adenosine, guanosine, and uridine-mono, di, and tri phosphates, under our experimental conditions, we observed a rapid equilibration between the three and hence a similar label incorporation level in the three nucleotides within any experiment. Wherever necessary, we preferred the diphosphates or triphosphates for calculating the ¹⁵N isotope label fractions for the following two reasons: (1) The di and triphosphates are more abundant compared to their respective monophosphates allowing robust quantitative measurement of the different isotopologues generated during the course of the experiment; and (2) the LC and mass resolution of the orbitrap allowed clear distinction between the ¹⁵N and ¹³C labeled UDP species (Figs. 1D and S2E), while the mass resolution is somewhat compromised at higher molecular weight range obscuring this distinction in UTP. Since we had limited isotope incorporation in nucleotides, these limitations were considered when different ¹⁵N enriched isotopologue estimations were performed (Figs. 1 and S2), however, these limitations didn’t apply to the estimation of total pools of monophosphates and triphosphates or ¹³C labeling experiments.

In-cell labeling with ¹³C6 glucose (¹³C6 Glc) and extraction of polar metabolites

For the in-cell stable isotope label incorporation assay, ~3–3.5 × 10⁶ HeLa cells were seeded in 100 mm culture dishes in the P- media, and after 20 h (80–90% confluency), media was switched to DMEM-without glucose and pyruvate, supplemented with 10% dFBS and ¹³C6 glucose (¹³C6 Glc, CLM-1396, Cambridge Isotope Laboratory, Inc. USA; final concentration 5.4 mM) (Figs. 2, 3, S4–S8) and labeled tracer uptake was allowed for 2, 3, 6, 8, or 14 h, respectively. Metabolite extraction was performed as described below. To ensure rapid uptake and equilibration, labeled glucose concentration was kept equivalent to that expected in the human serum⁴⁹. ¹³C6 Glc labeling allowed us to trace the isotope incorporation both in purines as well as pyrimidines.

Under the above experimental conditions, the incorporation of ¹³C derived from glucose into various metabolites was traced as described (Figs. 2A, B, S5–S6). ¹³C6 Glc derived label is incoporated into pyrimidine synthesis substrates: ¹³C5 PRPP generated by the pentose phosphate pathway (Figs. S6C and S8L) and ¹³C labeled Asp (Figs. S4I and S8D). The carbon as well as the nitrogen atoms in Asp are retained in the pyrimidine nucleotides, hence both the base as well as the ribose ring are labeled (Figs. 2E, G, S1, S5, and S6A and F).

Description of the TCA cycle metabolite labeling using ¹³C6 glucose

To uncover if the mitochondrial OAA was the source of Asp in the pyrimidine and purine de novo biosynthesis, we carefully probed the TCA cycle metabolites that are generated within mitochondria before some of the metabolites diffuse out into the cytosol (Figs. 2, S1, S5). We interpreted the isotope incorporation data at 6 h in the light of the structures of the analytes of interest and made appropriate positional assignments for the labeled atoms. ¹³C3 pyruvate (Fig. S4A) is acted upon by pyruvate dehydrogenase and ¹³C CO₂ and ¹³C2 acetyl CoA (Ac CoA) which enters first round of TCA cycle. When unlabeled OAA combines with ¹³C2 Ac CoA, the enzyme citrate synthase generates ¹³C2 isotopologues of citrate (Fig. S4B). Citrate is successively acted upon by aconitase and isocitrate dehydrogenase to produce α-KG, releasing an unlabeled or ¹³C labeled CO₂ (Fig. S4C, D). Unlabeled Glu is generated from unlabeled Gln imported into the mitochondria by the action of GLS. Labeled OAA and unlabeled Glu are acted upon by the reversible enzyme GOT2, simultaneously converting OAA to Asp and Glu to α-KG. Therefore, the action of GOT2 significantly dilutes the 13C labeled fraction in α-KG as compared to citrate and isocitrate. α-KG is further converted to succinyl CoA by α-KG dehydrogenase, releasing the third unlabeled or ¹³C labeled CO₂. Now, the action of succinyl CoA synthetase and succinate dehydrogenase generates ¹³C₂ succinate and ¹³C₂ fumarate, respectively (Fig. S4E, F). Since succinate is a symmetrical molecule, two alternate ¹³C2 isotopomers of succinate and all the downstream metabolites are possible. Next, fumarase converts fumarate into malate, and malate dehydrogenase (MDH1) converts it to ¹³C2 OAA (Fig. S4G, H). Additionally, ¹³C3 OAA is directly generated from the action of pyruvate carboxylase (PC) that combines unlabeled CO₂ with ¹³C3 pyruvate. Unlike succinate dehydrogenase, which is a unidirectional enzyme, the next two enzymes of TCA cycle, fumarase and malate dehydrogenase, carry out reversible reactions. Therefore, some back conversion of ¹³C3 OAA to ¹³C3 malate and ¹³C3 fumarate is observed (Fig. S4F, G). ¹³C2 and ¹³C3 OAA enter the second round of TCA cycle, generating ¹³C3, ¹³C4, and ¹³C5 citrate, which is processed by the various enzymes described above. Finally, as more ¹³C labeled CO₂ is accumulated within mitochondria, it generates ¹³C4 OAA, which will enter the TCA cycle generating ¹³C4 and ¹³C6 citrate. As the TCA cycle proceeds, first ¹³C2, then ¹³C3, and finally ¹³C4 OAA is produced, which is consecutively converted to ¹³C2, ¹³C3, and ¹³C4 Asp within mitochondria by the action of GOT2 (Fig. S4H, I). Since the mitochondrial Asp is utilized for nucleotide synthesis, there’s only limited accumulation. Therefore, in our extracts, the bulk cytosolic pool of Asp has a dominating contribution. For the experiments reported here, the ¹³C6 Glc concentration and labeling time were standardized to achieve a non-equilibrium condition where sufficient +2, +3, and +4 isotopologues of OAA are generated, but the bulk cytosolic Asp and malate primarily show +2 and +3 isotopologues (Fig. S4G, I). Based on the observed isotopologues, the positions of ¹³C in various TCA cycle intermediates and the pyrimidine biosynthesis metabolites are inferred (Figs. 2A–C, S5 and S6A, B).

To confirm our interpretation, we also carried out shorter (2 and 3 h) as well as longer (8 and 14 h) label incorporations, respectively (Figs. 3H–J and S8). Examination of succinate versus malate and fumarate isotopologue distribution establishes that at early points between 2 and 8 h, the TCA cycle only contributes to the (M + 2) isotopologue of OAA, while PC activity is responsible for the (M + 3) and (M + 4) isotopologues (Figs. 2A, S4E–H and S8A–C, M).

Modeling the ¹³C Asp mediated incorporation in pyrimidine de novo biosynthesis intermediates and UMP

Per a diffusive mechanism, a homogenous isotope incorporation pattern is expected across the pyrimidine biosynthesis substrate, intermediates, and products (Fig. 2A–C). Therefore, using the observed isotopologue fractional abundance of the isotopologues observed at the 6 h time point in Asp and various pyrimidine biosynthesis intermediates (Fig. 2D, F), the fractional abundance of the isotopologues observed in UMP were predicted (Fig. 2G). Considering the positions of ¹³C atoms in various Asp (and the pathway intermediates) isotopologues (Figs. 2B and S6A, B), we could mathematically treat the abundances of (M + 0) = v; (M + 1) = w, (M + 2) = x, (M + 3) = y, and (M + 4) = z (Fig. 2C, F). There’s 100% conversion of the (M + 0):v and (M + 1):w isotopologues to the (M + 5) and (M + 6) isotopologues of UMP, respectively. Due to the scrambling of isotope labeled positions in succinate and fumarate (Fig. S5), two plausible positions for ¹³C carbons may be expected for the Asp (M + 2) isotopologue (Figs. 2A, B and S6A). Therefore, the (M + 2) isotopologue of Asp:x, would be equally divided into the (M + 6) and (M + 7) isotopologues of UMP. Also, based on the position of ¹³C carbons originating as a result of PC activity, 100% of the (M + 3):y and (M + 4):z isotopologues of the intermediates would generate (M + 7) and (M + 8) isotopologue of UMP, respectively (Figs. 2C and S6A, B). Following the argument, the fractional abundances of the various UMP isotopologues may be predicted using the following equations.

$$\left({\rm{M}}+5\right)=v$$

(1)

$$\left({\rm{M}}+6\right)=w+(x/2)$$

(2)

$$\left({\rm{M}}+7\right)=\left(x/2\right)+y$$

(3)

$$\left({\rm{M}}+8\right)=z$$

(4)

For simplicity, using the isotopologue abundance obtained using Eqs. 1–4, the relative abundance of (M + 6):(M + 7):(M + 8), normalized by the combined abundance of (M + 6) + (M + 7) + (M + 8) (Fig. 2G) was plotted and the predicted distribution was compared to the observed relative distribution of the same isotopologues in uridine nucleotides.

Antimycin A treatment

To achieve ETC inhibition, a complexIII inhibitor, antimycin A treatment was used (Figs. 3A–E, O, P, and S6G–L). HeLa cells were maintained in purine rich (P+) media and for antimycin A treatment, ~3 M cells were seeded in a 10 cm dish in purine depleted (P-) media. After 20 h incubation at 37 °C, cells were treated with 12 µM Antimycin A and incubated for 3 h. Following this, cells were washed with 1x PBS, and the media was replaced by fresh DMEM (with 10% dialyzed FBS, without glucose and pyruvate), and 5.4 mM ¹³C6 glucose and fresh antimycin A were added to the media. Cells were incubated for 8 h to allow glucose uptake and isotope incorporation. Finally, metabolite extraction and LC/MS analysis was performed as described below.

siRNA treatment for transient GOT2/PC/GOT1 knockdown

Approximately 2 million HeLa cells were seeded in 100 mm culture dishes and allowed to grow overnight. At around 60–70% confluency, cells were washed with 1X PBS once and 10 mL fresh RPMI (10% dFBS) was added. siRNA treatment was performed as per manufacturer’s recommendations. We first tested 10 nM to 50 nM siRNA final concentrations for treatment between 24 and 96 h. The appropriate time and concentration of siRNA treatment for achieving robust GOT2/GOT1 and PC knockdown was determined to be 48 h with 25 nM, and 72 h with 50 nM siRNA, respectively (Figs. 3F, G, K–M, Q–S, S7, 4H, I and S9B–E). In brief, either ON-TARGET plus GOT2/PC/GOT1 siRNA smart pool (L-011674-01-005, #L-008950-00-0005, #L-011673-01-0005, Dharmacon, USA) or ON_TARGET plus, non-Target siRNA control pool (#D-001810-10, Dharmacon, USA) were mixed with Xfect RNA transfection reagent (#631450, Takara Bio), respectively. After mixing well, the master mix was allowed to stand at room temperature for 20 min to allow the formation of RNA-lipid complexes and the mix was added to the culture dishes to achieve final siRNA concentration of 25/50 nM. To ensure that cell survival wasn’t drastically affected due to reduced availability or Asp, cells were grown in RPMI media. 48/72 h post treatment, media was removed, cells were given two gentle washes with 5 mL 1xPBS, followed by addition of 1.5 mL cold 80% MeOH directly to the culture dish. Cells were harvested by scraping. Metabolite extraction and LC/MS analysis was performed as described in the previous section and the cell pellet was used for Western blotting. Since Western blot analysis was performed using the cell pellet obtained from methanol precipitation, small amount of aggregates led to aberrations in SDS-PAGE migration of the protein, leading to imperfect bands during Western blot analysis.

Western blot analysis

After siRNA treatment, knockdown of the enzyme PC/GOT2/GOT1 was confirmed by Western blotting (Fig. S7B, D, F). The cell pellets obtained after metabolite extraction using 80% MeOH were reconstituted in 100 μL buffer solution (6 M urea, 2 M thiourea in 4% CHAPS buffer) and incubated at room temperature (RT) for 1 h, with vortexing every 15 min. Lysate was centrifuged at 3000 × g for 20 min and ~30 μL of the soluble supernatant was drawn for sunning SDS-PAGE. For protein transfer to the PVDF membrane (#IPVH00010, Fisher Scientific), the semi-dry protocol was used as per the manufacturer’s recommendations. Membrane was blocked using 5% w/v milk powder in Tris buffered saline with Tween 20 (TBST with 0.1% Tween 20) for 2 h at RT. Before adding the antibodies, the blot was cut in half to stain with the target and a control antibody. Anti-GOT2 primary antibody (rabbit polyclonal anti-GOT2 antibody, #NBP1-81565, Novus Biologicals, Inc.), anti-pyruvate carboxylase (rabbit anti-PC, #NBP1-49536SS, Novus Biologicals, Inc.) anti-GOT1 (rabbit polyclonal anti-GOT1, #PA5-80670 Invitrogen) anti-PFAS primary antibody (rabbit anti-PFAS #A304-218A, Bethyl Labs), anti-α-tubulin (#DM1A, Cell Signaling), or anti-β-actin antibody (mouse anti-β-actin #10R-2927, Fitzgerald) were reconstituted in the blocking buffer at 1:2000 dilution and overnight staining was allowed after which the membrane was washed thrice with TBST buffer thrice at RT. Anti-rabbit or anti-mouse HRP conjugated secondary antibodies (#7074S or #7076S, Cell Signaling) were diluted in the blocking buffer at 1:3000 dilution and the blot was stained for 3 h at RT. Following this, the membrane was washed three times with TBST and developed using the SuperSignal West Pico Plus chemiluminescent substrate (#34580, ThermoFisher Scientific).

Modeling ¹⁵N Asp mediated incorporation in purine de novo biosynthesis

¹⁵N2 mediated label incorporation assay was optimized to achieve ~100% ¹⁵N label incorporation in the steps that utilize the sidechain amide of ¹⁵N2 Gln (Fig. S1B, C). These are the de novo biosynthesis steps catalyzed by the enzymes PPAT, and PFAS (Fig. 4). Under our experimental conditions, both Glu and Asp were only partially labeled with ¹⁵N (Figs. S2A and 1D). Compartmentalization of GLS and GOT2 make mitochondria the site of ¹⁵N Glu and ¹⁵N Asp production and upon being exported into the cytosol, significant dilution of the two labeled analytes may be expected. We hypothesized that if both the diffusive and the purinosome mediated channeled purine biosynthesis utilize the bulk cytosolic ¹⁵N Asp, the two steps that incorporate (SAICAR and SAMP production) ¹⁵N Asp must show same isotope incorporation and may be treated as mutually exclusive and independent events (Fig. 4A). Since there’s 100% enrichment with respect to ¹⁵N Gln incorporation, the ¹⁵N fraction arising from ¹⁵N Asp mediated labeling in SAICAR was calculated as (x) = peak area for (M + 3)/total peak area of (M + 2) + (M + 3). Following the argument, the fractional abundance of (M + 2) SAICAR = (1−x), the isotopologue arising from utilization of unlabeled Asp. Using the value of “x”, we predicted the relative fractional abundance of the (M + 2), (M + 3), and (M + 4) isotopologues in ATP, assuming homogenous label incorporation in the two steps catalyzed by the enzymes phosphoribosyl aminoimidazole succinocarboxamide synthase (PAICS) and adenylosuccinate synthase (ADSS):

$$\left({\rm{M}}+2\right):\left({\rm{M}}+3\right):\left({\rm{M}}+4\right) \, {in} \, {ATP}={(1-x)}^{2}:\left[2 * \left(1-x\right) * x\right]:{(x)}^{2}$$

(5)

(M + 2), (M + 3), and (M + 4) ADP or ATP isotopologues had no overlap with the residual salvage of hypoxanthine. Therefore, for the above computation, the combined peak areas of the (M + 2), (M + 3), and (M + 4) of ADP or ATP was equated to 100% and the fractional abundances for the various isotopologues were plotted (Fig. 4D).

Polar metabolite extraction

Polar metabolites were extracted as previously described¹⁹. After isotope incorporation, cells were washed gently thrice with 1X PBS on the culture dish. After adding 1.5 mL cold 80% methanol covering the culture dish, cells were harvested by scraping. To ensure complete rupture of cells, three cycles of freeze thaw using liq N₂ were carried out. The extract mix was centrifuged for 30 min at 4 °C, at 5700 × g in a hanging bucket rotor and the supernatant was collected and dried under N₂ gas flow. The dried extracts were reconstituted in 30 µL 3% MeOH with 1 µM chlorpropamide (internal standard) and 8–10 µL of the extract was used for each LC/MS run. The abundance of different isotopologues for the various analytes of interest was determined by quantitative LC/MS analysis. The corresponding peak areas were obtained using the Thermo Xcalibur software and the peak areas were used to determine the isotope enrichment for each analyte of interest. Data were plotted using GraphPad Prism 10.

We have shown that the above method of metabolite extraction with cold 80% MeOH only reports on the soluble bulk cytosolic pools of various analytes¹⁹, while the purinosome or pyrimidinosome metabolon-bound intermediates remain largely inaccessible due to rapid precipitation of the metabolons. Therefore, assuming a completely diffusive model of metabolism, a homogenous isotope label incorporation level in the intermediates and the end product nucleotides may be expected. On the other hand, any heterogeneity in the isotope label incorporation patterns would indicate pyrimidinosome or purinosome mediated metabolic channeling, generating the end product nucleotides UMP or AMP and GMP, respectively. Therefore, this experiment allows simultaneous detection of the diffusive and the channeled pathways that run in parallel where the diffusive pathway leads to accumulation of the intermediates of the pathway, while the channeled pathway restricts free diffusion of the intermediates and generates the end product which is then released into the bulk cytosol.

HPLC/ESI MS analysis

Cell extracts were analyzed by LC/MS using a modified version of an ion pairing reversed phase negative ion electrospray ionization method. metabolites were separated on a Phenomenex (Torrance, CA, USA) Hydro-RP C18 column (100 × 2.1 mm, 3 μm particle size) using a water/MeOH gradient with tributylamine and acetic acid added to the aqueous mobile phase and analyzed by Exactive plus orbitrap mass spectrometer controlled by Xcalibur 2.2 software (all from Thermo Fisher Scientific, Waltham, MA). The HPLC column was maintained at 30 °C, and at flow rate of 200 μl/ min. Solvent A is 3% aqueous MeOH with 10 mM tributylamine and 15 mM acetic acid; solvent B is MeOH. The gradient used is 0 min, 0% B; 5 min, 20% B; 7.5 min, 20% B; 13 min, 55% B; 15.5 min, 95% B; 18.5 min, 95% B; 19 min, 0% B; and 25 min, 0% B. The Exactive Plus was operated in negative ion mode at maximum resolution (140,000) and scanned from m/z 72 to m/z 1000 for the first 90 s and then from m/z 85 to m/z 1000 for the remainder of the chromatographic run. The retention times, molecular weights, and molecular formula of various analytes included in this study are provided (Table S1). The expected difference in atomic mass (Δamu) of the metabolites upon incorporation of different stable isotope labeled substrates was calculated using Δamu ¹³C: 1.0003 Da and Δamu ¹⁵N: 0.997 Da.

Data processing and statistical analysis

For each experiment, the data from three or four independent biological replicates were collated to determine the mean and standard deviation. LC/MS results are visualized and processed using the Xcalibur Software, Thermo Scientific. The data is plotted, and statistical analysis performed using GraphPad Prism Version 8.2.1. All schematics were generated using BioRender licensed version.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All the data supporting the findings of this research are available within the article and its supplementary information. Data supporting the findings of this manuscript are also available from the corresponding author upon request. Source data are provided with this paper.

References

DeBerardinis, R. J. & Chandel, N. S. Fundamentals of cancer metabolism. Sci. Adv. 2, e1600200 (2016).
Article ADS PubMed PubMed Central Google Scholar
Pavlova, N. N., Zhu, J. & Thompson, C. B. The hallmarks of cancer metabolism: still emerging. Cell Metab. 34, 355–377 (2022).
Article CAS PubMed PubMed Central Google Scholar
Villa, E., Ali, E. S., Sahu, U. & Ben-Sahra, I. Cancer cells tune the signaling pathways to empower de novo synthesis of nucleotides. Cancers https://doi.org/10.3390/cancers11050688 (2019).
Pareek, V., Pedley, A. M. & Benkovic, S. J. Human de novo purine biosynthesis. Crit. Rev. Biochem Mol. Biol. 56, 1–16 (2021).
Article CAS PubMed Google Scholar
Mullen, N. J. & Singh, P. K. Nucleotide metabolism: a pan-cancer metabolic dependency. Nat. Rev. Cancer 23, 275–294 (2023).
Article CAS PubMed PubMed Central Google Scholar
Pedley, A. M., Pareek, V. & Benkovic, S. J. The purinosome: a case study for a mammalian metabolon. Annu Rev. Biochem. 91, 89–106 (2022).
Article CAS PubMed PubMed Central Google Scholar
Wang, Q., Xia, J., Guallar, V., Krilov, G. & Kantrowitz, E. R. Mechanism of thermal decomposition of carbamoyl phosphate and its stabilization by aspartate and ornithine transcarbamoylases. Proc. Natl Acad. Sci. USA 105, 16918–16923 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Hove-Jensen, B. et al. Phosphoribosyl Diphosphate (PRPP): Biosynthesis, Enzymology, Utilization, and Metabolic Significance. Microbiol. Mol. Biol. Rev. https://doi.org/10.1128/MMBR.00040-16 (2017).
Yang, C. et al. De novo pyrimidine biosynthetic complexes support cancer cell proliferation and ferroptosis defence. Nat. Cell Biol. 25, 836–847 (2023).
Article CAS PubMed Google Scholar
An, S., Kumar, R., Sheets, E. D. & Benkovic, S. J. Reversible compartmentalization of de novo purine biosynthetic complexes in living cells. Science 320, 103–106 (2008).
Article ADS CAS PubMed Google Scholar
Evans, D. R. & Guy, H. I. Mammalian pyrimidine biosynthesis: fresh insights into an ancient pathway. J. Biol. Chem. 279, 33035–33038 (2004).
Article CAS PubMed Google Scholar
Pareek, V., Sha, Z., He, J., Wingreen, N. S. & Benkovic, S. J. Metabolic channeling: predictions, deductions, and evidence. Mol. Cell 81, 3775–3785 (2021).
Article CAS PubMed PubMed Central Google Scholar
Sweetlove, L. J. & Fernie, A. R. The role of dynamic enzyme assemblies and substrate channelling in metabolic regulation. Nat. Commun. 9, 2136 (2018).
Article ADS PubMed PubMed Central Google Scholar
Castellana, M. et al. Enzyme clustering accelerates processing of intermediates through metabolic channeling. Nat. Biotechnol. 32, 1011–1018 (2014).
Article CAS PubMed PubMed Central Google Scholar
Carrey, E. A. et al. Detection and location of the enzymes of de novo pyrimidine biosynthesis in mammalian spermatozoa. Reproduction 123, 757–768 (2002).
Kremer, D. M. et al. GOT1 inhibition promotes pancreatic cancer cell death by ferroptosis. Nat. Commun. 12, 4860 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Broeks, M. H. et al. The malate-aspartate shuttle is important for de novo serine biosynthesis. Cell Rep. 42, 113043 (2023).
Article CAS PubMed Google Scholar
He, J., Zou, L. N., Pareek, V. & Benkovic, S. J. Multienzyme interactions of the de novo purine biosynthetic protein PAICS facilitate purinosome formation and metabolic channeling. J. Biol. Chem. 298, 101853 (2022).
Article CAS PubMed PubMed Central Google Scholar
Pareek, V., Tian, H., Winograd, N. & Benkovic, S. J. Metabolomics and mass spectrometry imaging reveal channeled de novo purine synthesis in cells. Science 368, 283–290 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Vasan, K., Werner, M. & Chandel, N. S. Mitochondrial metabolism as a target for cancer therapy. Cell Metab. 32, 341–352 (2020).
Article CAS PubMed PubMed Central Google Scholar
Sullivan, L. B. et al. Aspartate is an endogenous metabolic limitation for tumour growth. Nat. Cell Biol. 20, 782–788 (2018).
Article CAS PubMed PubMed Central Google Scholar
Mazat, J. P. & Ransac, S. The fate of glutamine in human metabolism. The interplay with glucose in proliferating cells. Metabolites https://doi.org/10.3390/metabo9050081 (2019).
Altman, B. J., Stine, Z. E. & Dang, C. V. From Krebs to clinic: glutamine metabolism to cancer therapy. Nat. Rev. Cancer 16, 619–634 (2016).
Article CAS PubMed PubMed Central Google Scholar
Wise, D. R. & Thompson, C. B. Glutamine addiction: a new therapeutic target in cancer. Trends Biochem Sci. 35, 427–433 (2010).
Article CAS PubMed PubMed Central Google Scholar
Kerk, S. A. et al. Metabolic requirement for GOT2 in pancreatic cancer depends on environmental context. Elife https://doi.org/10.7554/eLife.73245 (2022).
Garcia-Bermudez, J. et al. Adaptive stimulation of macropinocytosis overcomes aspartate limitation in cancer cells under hypoxia. Nat. Metab. 4, 724–738 (2022).
Article CAS PubMed PubMed Central Google Scholar
Birsoy, K. et al. An essential role of the mitochondrial electron transport chain in cell proliferation is to enable aspartate synthesis. Cell 162, 540–551 (2015).
Article CAS PubMed PubMed Central Google Scholar
Son, J. et al. Glutamine supports pancreatic cancer growth through a KRAS-regulated metabolic pathway. Nature 496, 101–105 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Altea-Manzano, P. et al. Reversal of mitochondrial malate dehydrogenase 2 enables anaplerosis via redox rescue in respiration-deficient cells. Mol. Cell 82, 4537–4547.e4537 (2022).
Article CAS PubMed Google Scholar
Do, B. T. & Vander Heiden, M. G. GOT2 consider the tumor microenvironment. Trends Cancer 8, 884–886 (2022).
Article CAS PubMed PubMed Central Google Scholar
Sigoillot, F. D. et al. Nuclear localization and mitogen-activated protein kinase phosphorylation of the multifunctional protein CAD. J. Biol. Chem. 280, 25611–25620 (2005).
Article CAS PubMed Google Scholar
Mao, C. et al. DHODH-mediated ferroptosis defence is a targetable vulnerability in cancer. Nature 593, 586–590 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Mishima, E. et al. DHODH inhibitors sensitize to ferroptosis by FSP1 inhibition. Nature 619, E9–E18 (2023).
Article CAS PubMed Google Scholar
Zhou, Y. et al. DHODH and cancer: promising prospects to be explored. Cancer Metab. 9, 22 (2021).
Article PubMed PubMed Central Google Scholar
Wang, W., Cui, J., Ma, H., Lu, W. & Huang, J. Targeting pyrimidine metabolism in the era of precision cancer medicine. Front Oncol. 11, 684961 (2021).
Article CAS PubMed PubMed Central Google Scholar
Spinelli, J. B. et al. Fumarate is a terminal electron acceptor in the mammalian electron transport chain. Science 374, 1227–1237 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Rudolph, J. & Stubbe, J. Investigation of the mechanism of phosphoribosylamine transfer from glutamine phosphoribosylpyrophosphate amidotransferase to glycinamide ribonucleotide synthetase. Biochemistry 34, 2241–2250 (1995).
Article CAS PubMed Google Scholar
Tran, D. H. et al. De novo and salvage purine synthesis pathways across tissues and tumors. Cell https://doi.org/10.1016/j.cell.2024.05.011 (2024).
Fustin, J. M. et al. Rhythmic nucleotide synthesis in the liver: temporal segregation of metabolites. Cell Rep. 1, 341–349 (2012).
Article CAS PubMed Google Scholar
Fairbanks, L. D., Bofill, M., Ruckemann, K. & Simmonds, H. A. Importance of ribonucleotide availability to proliferating T-lymphocytes from healthy humans. Disproportionate expansion of pyrimidine pools and contrasting effects of de novo synthesis inhibitors. J. Biol. Chem. 270, 29682–29689 (1995).
Article CAS PubMed Google Scholar
French, J. B. et al. Spatial colocalization and functional link of purinosomes with mitochondria. Science 351, 733–737 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Sha, Z. & Benkovic, S. J. Purinosomes spatially co-localize with mitochondrial transporters. J. Biol. Chem. 300, 107620 (2024).
Article CAS PubMed PubMed Central Google Scholar
Wallace, D. C. Mitochondria and cancer. Nat. Rev. Cancer 12, 685–698 (2012).
Article CAS PubMed PubMed Central Google Scholar
DeBerardinis, R. J. & Chandel, N. S. We need to talk about the Warburg effect. Nat. Metab. 2, 127–129 (2020).
Article PubMed Google Scholar
Yang, R. et al. GLS and GOT2 as prognostic biomarkers associated with dendritic cell and immunotherapy response in breast cancer. Heliyon 10, e24163 (2024).
Article CAS PubMed PubMed Central Google Scholar
Kiesel, V. A. et al. Pyruvate carboxylase and cancer progression. Cancer Metab. 9, 20 (2021).
Article PubMed PubMed Central Google Scholar
Yang, S. et al. Mitochondrial glutamine metabolism via GOT2 supports pancreatic cancer growth through senescence inhibition. Cell Death Dis. 9, 55 (2018).
Article PubMed PubMed Central Google Scholar
Ryu, K. W. et al. Cellular ATP demand creates metabolically distinct subpopulations of mitochondria. Nature 635, 746–754 (2024).
Article ADS CAS PubMed PubMed Central Google Scholar
Ackermann, T. & Tardito, S. Cell culture medium formulation and its implications in cancer metabolism. Trends Cancer 5, 329–332 (2019).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

S.J.B. and V.P. acknowledge NIH grant GM024129 (S.J.B.), V.P. acknowledges financial support from the Huck Institutes of the Life Sciences, the Huck Innovative and transformational Seed funds, and the CURE Grant from the Pennsylvania Department of Health (the Department specifically disclaims responsibility for any analyses, interpretations, or conclusions). Authors thank the Penn State Metabolomics core facility and Drs. Sergei Koshkin, and Ashley Shay for help with LC/MS. V.P. thanks Dr Dipankar Bhowmik for standardization of Western blot analysis.

Author information

Authors and Affiliations

Department of Chemistry, The Pennsylvania State University, University Park, PA, USA
Vidhi Pareek & Stephen Benkovic
The Huck Institutes of the Life Sciences, The Pennsylvania State University, University Park, PA, USA
Vidhi Pareek

Authors

Vidhi Pareek
View author publications
Search author on:PubMed Google Scholar
Stephen Benkovic
View author publications
Search author on:PubMed Google Scholar

Contributions

V.P. and S.J.B. conceived the research; all experiments were performed and analyzed by V.P. and results interpreted by V.P. and S.J.B. Manuscript written by V.P. and proofread by S.J.B.

Corresponding authors

Correspondence to Vidhi Pareek or Stephen Benkovic.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Binghui Li, Naama Kanarek and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Transparent Peer Review file

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Pareek, V., Benkovic, S. Metabolic profiling reveals channeled de novo pyrimidine and purine biosynthesis fueled by mitochondrially generated aspartic acid in cancer cells. Nat Commun 16, 8952 (2025). https://doi.org/10.1038/s41467-025-64013-w

Download citation

Received: 24 June 2024
Accepted: 02 September 2025
Published: 08 October 2025
DOI: https://doi.org/10.1038/s41467-025-64013-w