Coordination of cell organelles to promote metabolon formation

Significance

The spatial and temporal relationships between sequential metabolic enzymes in a pathway have revealed how metabolon formation contributes to metabolic control. This study uses fluorescence microscopy to better understand the factors that drive the overall stability of these membrane-less assemblies. Here, we reveal a critical dependency between the morphology of the endoplasmic reticulum (ER) and the localization of the purinosome, a metabolon composed of de novo purine biosynthetic enzymes. Defining the impact that organelle membranes have on regulating these dynamic assemblies provides key insights into how compartmentalized metabolic processes coordinate their activities.

Abstract

The spatial coordination between cellular organelles and metabolic enzyme assemblies represents a fundamental mechanism for maintaining metabolic efficiency under stress. While previous work has shown that membrane-bound organelles regulate metabolic activities and that membrane-less condensates conduct metabolic reactions, the coordination between these two organizations remains unaddressed. By using a combination of proximity labeling, superresolution fluorescence microscopy, and metabolite analyses using isotopic tracing, we investigated the relationships between these metabolic hotspots. Here, we show that nutrient deficiency elongates mitochondria and transforms the ER from a tubular to sheet-like morphology, coinciding with increased mitochondrial respiration and inosine 5′-monophosphate levels. These structural changes promote the colocalization of purinosomes with these organelles, enhancing metabolic channeling. Disruption of ER sheet formation via MTM1 knockout destabilizes purinosomes, impairs substrate channeling, and reduces intracellular purine nucleotide pools without altering enzyme expression. Our findings reveal that organelle morphology and interorganelle contacts dynamically regulate the assembly and function of metabolic condensates, providing a structural basis for coordinated metabolic control in response to nutrient availability.

Cellular organelles are dynamic entities that respond to changes in growth conditions. Recent studies have implied that metabolic processes influence the morphology of mitochondria and the endoplasmic reticulum (ER) (1). Mitochondria respond to changes in energy demands and oxygen availability by altering their activities. These changes in activity have been associated with mitochondrial fusion and dissolution to adapt organelle architecture to fulfill cellular energy requirements (2, 3). The ER serves as a central platform for metabolic regulation that tightly binds nutrient sensing to energy homeostasis. Changes in nutrient availability transform the structure of the ER from tube to sheet-like structures (4, 5). The ER also coordinates with mitochondria via membrane contact sites (6, 7) to regulate ATP production, lipid exchange, and the redox state to adapt energy output to cellular nutrient fluctuations (8, 9). With the growing recognition that organelle morphologies regulate metabolic processes, we asked whether other metabolic processes might also depend on this remodeling.

De novo purine biosynthesis (DNPB) is a tightly regulated process leveraged by cells to meet elevated purine demands, particularly under nutrient stress when salvage processes alone cannot supply sufficient nucleotides (10). In humans, DNPB consists of six enzymes that catalyze the reactions converting phosphoribosyl pyrophosphate into inosine 5′-monophosphate (IMP) (11). Of key importance are the trifunctional enzyme GART (EC 6.3.4.13/2.1.2.2/6.3.3.1), phosphoribosyl formylglycinamidine synthase (PFAS, EC 6.3.5.3), and adenylosuccinate lyase (ADSL, EC 4.3.2.2), who compartmentalize with the other pathway enzymes to form a dynamic metabolon called the purinosome (Fig. 1A) (12, 13). Live-cell imaging has defined purinosomes as phase-separated liquid-like condensates (14, 15) that facilitate clustered substrate channeling (16) and enhance pathway throughput (17).

Fig. 1.

Purinosomes spatially organize with mitochondria and the endoplasmic reticulum (ER) upon nutrient depletion. (A) Schematic of the purinosome and proposed localization near mitochondria and the ER for this study. Created in BioRender. Staub, J. (2026) https://BioRender.com/11ty1zl. (B) Oxygen consumption rates (OCR) were measured using sequential additions of chemical inhibitors (1 μM each of oligomycin, FCCP, rotenone, antimycin A) to characterize the cellular bioenergetic state, including basal respiration, maximum respiration, and ATP production from oxygen. Real-time extracellular flux profiles (Seahorse assay) showing changes in OCR from HeLa cells cultured under normal growth conditions (black open circles) and nutrient-depleted conditions (24 h) with (dark gray open squares). Data represent the average ± SD from N = 4 biological replicates of each condition. (C) Quantification of OCR parameters from the same experiment. Data represent the average ± SD from N = 4 biological replicates of each condition. A one-way ANOVA test was used, and P-values are shown above those comparisons demonstrating statistical significance. (D) HeLa cells expressing mScarlet I-SEC61B (gray) were immunostained for TOMM20 (green) and ADSL (magenta) to show the spatial organization of the ER, mitochondria, and ribosomes, respectively. Nutrient-depletion induced purinosome formation (ADSL clusters) where it localizes with the ER near (ROI 1) or away (ROI 2) from mitochondria. [Scale bar, 10 μm (1 μm for ROIs).] (E) ADSL association with mitochondrial transporter SLC25A38 increases upon nutrient-depletion. Percent of positive cells are determined by having greater than six individual localizations. Data show the average ± SD of three biological replicates of at least 100 cells per replicate. (F) The degree of colocalization between ADSL and mitochondria (TOMM20) upon purine-depletion. Colocalization was determined to be having the shortest distance between an individual purinosome (diameter ≥ 400 nm) and mitochondria ≤ 100 nm. Data represent the average ± SD across N = 4 biological replicates with at least 50 purinosomes analyzed per replicate. (G) The percent of cells with greater than six localizations between ADSL and SEC61B. Data show the average ± SD of three biological replicates of at least 100 cells per replicate. (H) The colocalization of ADSL with the ER (mScarlet I-SEC61B) in N = 690 normal or nutrient-depleted growth conditions as determined by Pearson’s Correlation. A two-tailed Welch’s determined statistical significance t test was performed.

An initial survey of the spatial and temporal features of purinosomes has guided insights into its regulation and further supported its biological function. Purinosomes are stabilized through their associations with microtubules and mitochondria (18, 19). This unique positioning is modulated by changes in mTOR signaling (20) and has been shown to be in close proximity to discrete clusters of solute carrier transporters embedded in the mitochondrial inner membrane (21). Combined, these findings propose a mechanism where mitochondria-derived substrate generation is coupled with their direct use in purine biosynthesis.

In this study, we seek out structural elements that assist in stabilizing the purinosome condensate near mitochondria. By leveraging fluorescence microscopy and metabolic profiling assays, we show that changes in mitochondria and ER morphologies are driven by cellular conditions that activate DNPB. Mitochondria-localized purinosomes are encapsulated in sheet-like structures of the ER, and upon knockout of MTM1, a lipid phosphatase that disfavors the ER transformation, the purinosome becomes destabilized. Further, we demonstrate that this membrane-induced destabilization results in a reduction of channeled IMP biosynthesis favoring the IMP generated by salvage synthesis for adenosine and guanosine 5′-monophosphate (AMP and GMP, respectively) production. Together, these findings expand our understanding of purinosome and DNPB regulation by revealing that nearby organelle membrane dynamics play a critical role in its stabilization and role in enhancing the efficiency of purine biosynthesis.

Results

Activation of DNPB Shifts Organelle Morphology.

Our initial investigations sought to determine whether activating DNPB alters the morphology of mitochondria and the ER, which harbor independent metabolic processes while also associating to regulate cellular homeostasis. Nutrient depletion, a mechanism to activate DNPB (17, 22, 23), induced changes in the morphology of both organelles as revealed by DeepSIM, a form of lattice structured illumination microscopy (SI Appendix, Figs. S1 A and B and S2A). On a per cell basis, the average mitochondrion, immunolabeled with TOMM20-AlexaFluor647, was elongated by 21% after 24 h of nutrient depletion (SI Appendix, Fig. S1C). This elongation coincided with enhanced branching as denoted by a 23% increase in total branch length per mitochondrion without a substantial change in the average number of branches per mitochondrion (SI Appendix, Fig. S1 D and E). The ER, labeled with a silicon-rhodamine Halo tag ligand targeting SEC61B, was also transformed after nutrient depletion, which adopted a more sheet-like structure as reflected in a 1.4-fold change in ER sheet ratio (SI Appendix, Fig. S2B).

Due to the high ATP and 10-formyl-tetrahydrofolate cofactor demands of DNPB, we sought out whether changes in the mitochondrial morphologies correlated with increases in mitochondrial activities, notably ATP production and one-carbon metabolism. Gross mitochondrial (basal) respiration upon nutrient depletion increased 2.0 ± 0.5-fold as well as ATP production from oxygen 1.9 ± 0.4-fold (Fig. 1 B and C). Additionally, a 1.5 ± 0.2-fold increase in the NADP+/NADPH ratio was observed, implying one-carbon (folate) metabolism might be further altered in nutrient-depleted HeLa cells (SI Appendix, Fig. S4). Together, membrane-induced increases in these mitochondria localized processes can provide the necessary substrates to fuel DNPB to meet the purine demands of the cell.

Purinosomes Associate With Mitochondria and Sheet-Like ER Structures.

Activation of DNPB can be microscopically visualized through the condensation of pathway enzymes into purinosomes (13, 24). The spatial arrangement of purinosomes, as denoted by discrete adenylosuccinate lyase (ADSL) structures, with subcellular organelles mitochondria (TOMM20) and the ER (SEC61B) was evaluated using three color Airyscan imaging (Fig. 1D). These studies revealed that in addition to purinosome formation and their association with mitochondria, purinosomes exhibited a tendency to colocalize with the ER when HeLa cells experienced nutrient deficient growth conditions. Pairwise analyses using proximity ligation assays indicated 78 ± 8% of nutrient-deficient cells showed a positive association (≥6 localizations) between purinosomes (ADSL) and mitochondria (mitochondrial glycine transporter protein SLC25A38) compared to 30 ± 6% of cells cultured under normal, nutrient-rich conditions (Fig. 1E). Similar results were determined with other mitochondrial transporter proteins such as the aspartate/glutamate transporter SLC25A13 (65 ± 11% for nutrient-deficient cells versus 9 ± 10% control cells) and the ATP/ADP transporter SLC25A5 (73 ± 14% versus 24 ± 13%) (SI Appendix, Fig. S5 A and B), consistent with a prior report (21). Using two-color DeepSIM imaging of purinosomes (ADSL) with mitochondria (TOMM20), nutrient-deficient HeLa cells showed an 80 ± 10% colocalization and was statistically different when compared to a randomization control (24 ± 5%) (Fig. 1F), a result consistent with previous reports of PFAS-mitochondria colocalization in HeLa cells (20).

To investigate purinosome-ER associations, time-lapse imaging of purinosome formation was monitored over 24 h. Following 8 h of nutrient depletion, cells showed a maximal 1.5-fold increase in the appearance of ER sheet-like structures, while the formation of ADSL-PFAS coclusters was not detected until 24 h (SI Appendix, Fig. S6). At this time point, proximity ligation assays were conducted on cells probed for ADSL-SEC61B associations and showed 96 ± 7% of nutrient-depleted cells having a positive association, which decreased to 37 ± 12% in control cells grown under nutrient-rich conditions (Fig. 1G). A Pearson’s correlation coefficient was computed and showed positive colocalization between immunostained ADSL and transiently expressed mScarlet-SEC61B (Fig. 1H). Although the number of cells showing a positive association between SLC25A13 and SEC61B increased from 6 ± 3% to 19 ± 5% upon nutrient-depletion, the occurrence remained limited (SI Appendix, Fig. S5C). These combined data suggest that the purinosome might serve as a mediator between these two organelles as the resulting interface between mitochondria and sheet-like ER structures under nutrient depletion appears to form confined pockets that could restrict diffusion of mitochondria-derived substrates and facilitate their direct transfer to purinosomes (Fig. 1D, white arrow).

Loss of MTM1 Expression Destabilizes the Purinosome Condensate.

To test whether the sheet-like ER structure is required for purinosome stabilization, we investigated its colocalization in a CRISPR/Cas9-generated MTM1 knockout HeLa cell line (4). MTM1 is a lipid phosphatase that contributes to maintaining the ER morphology under nutrient-depleted conditions. In wild-type HeLa cells, nutrient deficiencies promote an ER membrane transition from a tube to sheet-like structure. Loss of MTM1 expression prevents this structural transition and has been implicated in the reduction of mitochondrial processes such as the β-oxidation of fatty acids and energy metabolism (4).

We first asked whether the loss of MTM1 resulted in a decrease in DNPB enzyme associations with the ER. For these experiments, all cells evaluated were nutrient depleted for 24 h to allow for direct comparison between DNPB and purinosome formation with ER structure. Proximity ligation assays showed ADSL remained associated with the ER (SEC61B) regardless of MTM1 expression (purine-depleted wild-type and MTM1^−/− HeLa cells showed 96 ± 7% and 77 ± 12% PLA-positive HeLa cells, respectively) (Fig. 2A). Despite this, the MTM1^−/− cell line showed a decrease in the association between SLC25A38 and ADSL (77 ± 8% for wild-type versus 31 ± 8% for the knockout cell line) (Fig. 2B) as well as another purinosome enzyme GART (88 ± 8% versus 3 ±5%) (Fig. 2C). Combined with the observation that there were not gross mitochondrial morphological changes under these nutrient-limiting growth conditions (SI Appendix, Fig. S1 C–E), we conclude that changes in ER morphology might play a critical role in stabilizing DNPB enzyme associations with mitochondria.

Fig. 2.

Loss of MTM1 does not permit purinosome formation. Proximity ligation assays showing the number of cells showing at least six localizations between ADSL and (A) the ER (SEC61B) and (B) the mitochondrial glycine transporter SLC25A38. (C) Similar results are shown for the GART-SLC25A38 associations. (D) The number of cells with at least 6 localizations between purinosome enzymes ADSL and PFAS indicating a loss of purinosome formation in MTM1^−/− HeLa cells. (A–D) Data show the average ± SD of three biological replicates of at least 100 cells per replicate. Statistical significance between experimental conditions determined by a two-tailed Welch’s t test with calculated P-values displayed. (E) Nutrient-depleted HeLa cells (Top) with and (Bottom) without MTM1 expression immunostained with three representative DNPB enzymes, ADSL (magenta), PFAS (cyan), and GART (yellow). ROI for a representative arrangement of purinosome enzymes is shown in a white box. [Scale bar, 10 μm (ROI: 1 μm).]

We then asked whether the loss of MTM1 decreased purinosome formation. By investigating the colocalization of ADSL with PFAS, 14 ± 6% of MTM1^−/− HeLa cells showed a positive association, a substantial reduction from wild-type cells (70 ± 6%) (Fig. 2D) without a change in the expression of pathway enzymes (SI Appendix, Fig. S7). To corroborate these findings, we performed three color Airyscan imaging of purinosome enzymes ADSL, PFAS, and GART to show the lack of coclustering in the absence of MTM1 (Fig. 2E). Additional pairwise associations between the purinosome, mitochondria, and ER enzymes have been evaluated and are presented in SI Appendix, Fig. S5D to provide a more comprehensive view of how changes in the ER morphology can destabilize the purinosome.

Destabilization of the Purinosome by Impairing ER Sheet Morphology Alters Purine Nucleotide Levels.

Last, we sought to see whether the loss of the purinosome in MTM1^−/− cells resulted in a change in intracellular purine nucleotide levels and metabolic flux through the de novo purine biosynthetic pathway. As anticipated by a lack of quantifiable change in mitochondrial morphology between wild-type and MTM1^−/− nutrient deficient HeLa cells, no observed change in mitochondria respiration or ATP production was detected (SI Appendix, Fig. S3). Prior work reports that the loss of MTM1 impairs mitochondria metabolism, but that study relied on more stringent nutrient depleted growth conditions, which might account for the discrepancy (4).

To monitor total purine monophosphate levels and metabolic flux, stable isotope incorporation experiments using [¹³C₃,¹⁵N]-serine were performed. In mitochondria, serine and tetrahydrofolate (THF) are converted into glycine and 5,10-methylene THF by serine hydroxymethyltransferase (SHMT2). Both products of the reaction become isotopically labeled and integrated into IMP at three steps: step 2 for [¹³C₂,¹⁵N]-glycine incorporation into 5-PRA to form GAR, and steps 3 and 9 for [¹³C]-labeled one carbon units into GAR and AICAR to form FGAR and FAICAR, respectively (SI Appendix, Fig. S8A). Pathway activating conditions driven by nutrient-depletion showed a 2.0 ± 0.2-fold increase in the total amount of IMP with no significant change to the levels of AMP and GMP (Fig. 3A). The loss of MTM1 in nutrient-deficient HeLa cells lowered IMP levels by 3.2 ± 0.6-fold as well as AMP and GMP levels by 3.4 ± 0.7-fold and 2.8 ± 0.6-fold, respectively. The ratio of labeled to unlabeled IMP also shifted with a smaller fraction of unlabeled IMP remaining in the knockout cells (Fig. 3B and SI Appendix, Fig. S8B). This decrease in IMP showed an initial preference in utilizing unlabeled IMP from an existing pool that was generated by salvage synthesis prior to depletion (25). Although mitochondrial respiration and ATP levels were unaffected, disruptions in ER-mitochondria architecture might still influence additional metabolic processes that indirectly support DNPB. In particular, we noted a 1.5 ± 0.2-fold decrease in the NADP+/NADPH ratio upon MTM1^−/−, which might suggest altered redox and one-carbon metabolism (SI Appendix, Fig. S3). Combined with data generated by proximity ligation assays, the ER morphology-driven destabilization of the purinosome and its association with mitochondria caused cells to preferentially use the existing pool of IMP rather than newly labeled IMP (Fig. 3C) owing to a disruption in substrate channeling.

Fig. 3.

Disruption of sheet-like ER morphology results in decreased substrate channeling and IMP production. (A) The total ion current for each purine monophosphate (IMP, AMP, and GMP) relative to the total ion current in the analyzed sample under various experimental conditions. Data represent the average ± SD of four biological replicates of at least 10⁶ cells. A one-way ANOVA test was used to determine statistical significance. (B) By monitoring the [¹³C₃,¹⁵N]-serine incorporation into IMP, the fraction of labeled IMP was determined by the sum of M + n isotopologues (where n = 1 through 5 after natural abundance subtraction) was compared to unlabeled (existing) isotopologue (M) to indicate a preference for using the existing pool of IMP to maintain AMP and GMP intracellular pools. Data represent the average ± SD of four biological replicates of at least 10⁶ cells. (C) Schematic showing our current model for how nutrient-deficient HeLa cells promote purinosome formation through a sheet-like ER structure stabilization to increase efficiency of purine production and how disruption of the purinosome through altering the morphology of the ER can impact substrate channeling.

Discussion

While considerable progress has been made in characterizing purinosome dynamics and function, several fundamental questions have remained. This study aimed to address one of those questions – What factors stabilize purinosomes near mitochondria? We show that on average nutrient-depleted HeLa cells display more elongated mitochondria, which is thought to be more bioenergetically efficient through the remodeling of the inner mitochondrial membrane (26, 27). This remodeling has been shown to reorganize the mitochondrial transporter proteins and hypothesized to facilitate the direct transfer of substrates to the purinosome (21). Building on this, we found that purinosome localization near mitochondria under nutrient-depleted growth conditions occurs only after the induction of sheet-like ER structures. This result suggests that the change in membrane morphology might drive a microenvironment that promotes condensate formation.

Without a membrane, there has been a longstanding inquiry into how the purinosome can sequester pathway substrates and intermediates to facilitate clustered substrate channeling (16). Here, we show that while the condensate itself is not encapsulated by lipids, its overall stability might rely on the lipid arrangement and structure of cellular organelles. Disrupting the sheet-like ER morphology destabilized the complete purinosome resulting in a reduction in overall cellular purine monophosphate levels and substrate channeling efficiency. These findings suggest that morphological remodeling of the ER and mitochondria facilitate both the stabilization and spatial organization of the purinosome.

Although mitochondria and the ER have independent metabolic activities, their enhanced coordination suggests that this association is central to metabolic homeostasis. From this study, the loss of MTM1 alters many metabolic processes, including one-carbon metabolism and DNPB. Under our experimental conditions, we observe limited associations between our mitochondria (SLC25A13) and the ER (SEC61B) markers. Intriguingly, we find that ADSL is associated with both. Does this suggest that ADSL might be mediating the association between these organelles that this might point to an unrecognized regulatory feature of DNPB? Further, how might the activities of these organelles impact these contacts and direct purine production? Ongoing studies are directed toward understanding these connections to better understand not only purinosome regulation but also the spatial interdependencies between metabolic processes.

Materials and Methods

Details on the materials and methods used in this study are provided in SI Appendix, covering immunofluorescence, image analyses, activity assays, stable isotope incorporation assays, and immunoblotting. All methods detail data and statistical analyses.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Plasmids used in this study are available through Addgene. Datasets are available through Iowa Research Online (https://doi.org/10.25820/data.008179) (28). All other study data are included in the article and/or SI Appendix.