AP-3 and the V-ATPase Modulate CTP Synthase Assembly Through Spatial Association at the Yeast Vacuole

Abstract

The compartmentalization of metabolic enzymes into membraneless filaments termed cytoophidia represents a conserved regulatory mechanism, exemplified by cytidine triphosphate synthase (CTPS), which assembles into pH-sensitive cytoophidia in the cytosol. In Saccharomyces cerevisiae, nutritional deprivation both triggers CTPS cytoophidia assembly and disassembles the vacuolar H⁺-ATPase (V-ATPase) that acidifies vacuoles (lysosomes), yet whether these processes are functionally linked remains unknown. We demonstrate spatial proximity between the yeast CTPS homologs Ura7/Ura8, the V-ATPase, and the AP-3 adaptor complex that mediates vesicular transport to vacuoles. We show Ura7, the major CTPS isoform in yeast, localizes to vacuoles under both nutrient-rich and starvation conditions. Genetic disruption of AP-3 function altered Ura7 assembly dynamics in starved cells, reducing total structures yet dramatically enhancing Ura7 cytoophidia elongation (~5-fold), suggesting a dual regulatory role for AP-3 that both promotes Ura7 assembly and restrains elongation. Moreover, combining nutritional and pharmacological V-ATPase inhibition triggered massive Ura7 cytoophidia formation. These findings reveal a previously unrecognized spatial coupling between metabolic enzyme compartmentalization, vacuolar trafficking, and the pH regulation machinery, suggesting a new organizational principle whereby CTPS assembly dynamics respond to vacuolar function.

INTRODUCTION

The AP-3 (adaptor protein complex-3) pathway mediates vesicular transport from late Golgi/endosomal compartments to lysosomes and lysosome-related organelles (Cowles et al., 1997; Theos et al., 2005; Robinson, 2015). Unlike AP-1 and AP-2, AP-3 remains on vesicles during transit, releasing only upon arrival at their target organelle where they facilitate tethering via the HOPS complex (Angers and Merz, 2009; Cabrera et al., 2010; Schwartz et al., 2017; Schoppe et al., 2020). This prolonged coat association may enable spatial organization of cytosolic factors to sites of membrane trafficking and fusion. AP-3 proximity-labeling has revealed numerous proteins beyond canonical cargoes that reside in close spatial proximity to this trafficking machinery (Schoppe et al., 2020), raising questions about whether AP-3 coordinates additional cellular processes beyond vesicle transport. Among these proximity-detected proteins are metabolic enzymes, suggesting potential links between vesicular trafficking and metabolic regulation.

Cytidine triphosphate synthase (CTPS) is one of the metabolic enzymes identified in the AP-3 proximity dataset (Schoppe et al., 2020). CTPS exemplifies an emerging regulatory mechanism through self-assembly into membraneless filaments termed cytoophidia (Noree et al., 2010; Liu, 2016). In Saccharomyces cerevisiae, the two CTPS isoforms, Ura7 and Ura8, catalyze the ATP-dependent conversion of UTP to CTP and form cytoophidia in response to stressors including glucose starvation. Filament formation sequesters the enzyme in an inactive state and is promoted by elevated CTP levels, providing a mechanism by which cytoophidia tune enzymatic output to cellular nucleotide status (Barry et al., 2014; Noree et al., 2014). CTPS cytoophidium formation is exquisitely sensitive to intracellular pH, with acidification triggering rapid assembly (Hansen et al., 2021). This pH sensitivity suggests potential coordination with the vacuolar H⁺-ATPase (V-ATPase), which regulates both organellar and cytosolic pH through reversible assembly of cytoplasmic (V1) and membrane (V0) sectors (Kane, 2006; Martinez-Munoz and Kane, 2008; Forgac, 2007). Notably, glucose starvation – the same condition that induces CTPS cytoophidia–triggers rapid V1-V0 disassembly, silencing proton transport and causing cytosolic acidification (Kane, 1995; Parra et al., 2000).

Despite extensive characterization of these individual systems, how AP-3 trafficking, CTPS assembly dynamics, and V-ATPase regulation might be functionally integrated remains unknown. Multiple V-ATPase subunits appear in AP-3 proximity-labeling (Schoppe et al., 2020) and affinity chromatography datasets (Leih et al., 2026), and V-ATPase mutants exhibit AP-3 trafficking defects (Anand et al., 2009), suggesting physical association between the trafficking machinery and pH regulatory apparatus. Whether this proximity extends to metabolic enzymes like CTPS remained unexplored.

By systematically interrogating published AP-3 interactomes in yeast using bimolecular fluorescence complementation (BiFC), we discover spatial associations between the Ura7/Ura8 CTPS isoforms, AP-3, and the V-ATPase. Genetic ablation of AP-3 sensitizes Ura7 to stress-induced filamentation, and V-ATPase inhibition using both nutritional and pharmacological methods triggers massive Ura7 assembly, resulting in extensive cytoophidia formation. These findings reveal previously unrecognized connections between the vacuolar trafficking machinery, pH regulatory apparatus, and metabolic enzyme compartmentalization.

RESULTS AND DISCUSSION

Spatial association between AP-3 and CTPS

To identify candidate AP-3-associated proteins, we cross-referenced AP-3 proximity-labeling (Schoppe et al., 2020) and affinity chromatography datasets (Leih et al., 2026), yielding 98 proteins identified by both methods (Supplemental Table S1). We screened these candidates for AP-3 association by BiFC, resulting in 6 positives (Supplemental Figure S1A–D). Among these, we focused on the Ura7 CTPS, motivated by its strong AP-3 association compared to other candidates (Supplemental Figure S1D) and by its glucose starvation-induced cytoophidium formation (Noree et al., 2010), a condition that also inhibits the V-ATPase (Kane, 1995; Parra and Kane, 1998). To assess whether Ura7 proximity involves the assembled AP-3 complex, we tested BiFC between Ura7 and each of the two AP-3 large subunits (Apl6 and Apl5); both pairings produced significantly more BiFC-positive cells and greater total fluorescence than the negative control (Figure 1A–C). Each AP-3 large subunit also showed BiFC with Ura8 (Figure 1D–F), the minor CTPS isoform that is 78% identical to Ura7 but half as abundant (Ozier-Kalogeropoulos et al., 1994; Nadkarni et al., 1995). The comparable BiFC signals between each CTPS and both AP-3 large subunits confirm that their spatial association involves the assembled AP-3 complex. This dual association mirrors the redundancy in cytoophidium formation, where Ura7 and Ura8 colocalize (Noree et al., 2010). Neither CTPS isoform has a canonical AP-3 sorting signal, suggesting they associate with AP-3 at trafficking sites rather than as transported cargo.

Figure 1. BiFC validates spatial proximity between AP-3 and Ura7.

(A,D) Confocal images of cells expressing the indicated N- and C-terminal fragments of the Venus fluorescent protein (VN and VC, respectively). His2 serves as negative control. Dashed ovals indicate cellular outlines. Scale bar, 5 μm. (B,E) Mean percentage of cells containing ≥1 BiFC punctum. (C,F) Total puncta fluorescence per puncta-positive cell (log-transformed); the horizontal dashed line indicates the median BiFC fluorescence in negative-control cells expressing His2-VC with Ura7-VN (C) or Ura8-VN (F). Graphs show means ± SEM from 3 independent biological replicates (circles). Statistical comparisons were performed on biological replicate means by one-way ANOVA with Tukey’s post-hoc test (B,E) or by unpaired t-test (C,F); p values are shown in the graphs.

BiFC reveals spatial associations linking AP-3, CTPS, and the V-ATPase

Since CTPS cytoophidia and V-ATPase disassembly are both triggered by glucose starvation (Kane, 1995; Parra and Kane, 1998; Noree et al., 2010), we asked if AP-3 is also spatially associated with V-ATPase components. AP-3 proximity labeling identified six V-ATPase subunits (Schoppe et al., 2020), including Vma10, a peripheral stalk subunit (Forgac, 2007). We found that BiFC pairings between Vma10 and each AP-3 large subunit produced fluorescent puncta adjacent to vacuoles, with significantly more BiFC-positive cells and greater total fluorescence than the negative control (Figure 2A–C). Given AP-3 proximity to Ura7/Ura8 and Vma10, we tested if Vma10 and each CTPS isoform are also in close spatial proximity; BiFC confirmed this (Figure 2D–F), consistent with independent proteomic identification of Ura7 in V-ATPase immunoprecipitates (Khan et al., 2021). Notably, Ura7 showed no BiFC significantly above the negative control when paired with endosomal ESCRT components Vps36 (Supplemental Figure S1E,G) or Vps60 (unpublished results), though it showed weak but detectable BiFC with Vps8, a subunit of the CORVET endosomal tethering complex (Supplemental Figure S1F,G). Because CORVET tethers endosomes to the vacuolar membrane (Peplowska et al., 2007), Vps8 is transiently positioned at the endosome-vacuole interface under normal conditions; hence, this low-level BiFC signal might reflect incidental proximity of Vps8 to vacuole-associated Ura7 rather than a specific endosomal interaction.

Figure 2. AP-3 and Ura7 associate with the V-ATPase subunit Vma10.

Confocal images of cells expressing the indicated VN/VC fusions (A,D) Scale bar, 5 μm. (B,E) Mean percentage of cells containing ≥1 BiFC punctum. (C,F) Total puncta fluorescence per puncta-positive cell (log-transformed). Graphs and statistics as in Figure 1B,C.

The results described above suggest AP-3, CTPS, and the V-ATPase function within a spatially organized network at the vacuole. The V-ATPase itself does not traffic to vacuoles via the AP-3 pathway (Piper et al., 1997; Bowers et al., 2005), so its presence in the AP-3 proximity cloud is consistent with AP-3 coat retention until vesicles reach the vacuole (Angers and Merz, 2009; Schoppe et al., 2020). Because the V-ATPase is the primary regulator of cytosolic and organellar pH (Martinez-Munoz and Kane, 2008), and CTPS filament formation is triggered by pH drops (Hansen et al., 2021), spatial proximity between Vma10 and Ura7/Ura8 suggests a mechanism by which CTPS assembly responds to the V-ATPase structural state.

Ura7 localizes to vacuoles under nutrient-rich and starvation conditions

We next examined whether Ura7 spatial association with the V-ATPase reflects functional vacuolar localization. In cells coexpressing Ura7-GFP and the vacuolar marker Vph1-mCherry, Ura7-GFP events (defined as Ura7-GFP observed either as punctate structures or as filamentous cytoophidia) were infrequent under rich growth conditions, as expected (Noree et al., 2010), but more than one-third of these events colocalized with Vph1-mCherry at the vacuolar membrane (Figure 3A,D). Starvation increased Ura7-GFP events more than two-fold (Figure 3B–C), with a similar proportion colocalized with Vph1-mCherry (Figure 3D), though this proportional trend did not reach significance (p=0.084). Notably, Ura7 was independently identified as a vacuolar protein by unbiased mass spectrometry of purified vacuoles (Eising et al., 2022), providing orthogonal biochemical evidence for vacuolar association that does not rely on fluorescence proximity. These observations suggest a substantial fraction of Ura7 structures associates with the vacuolar membrane regardless of metabolic state.

Figure 3. Ura7 cytoophidia localize to the vacuole under both nutrient-rich and starvation conditions.

Confocal images of cells expressing Ura7-GFP and Vph1-mCherry in rich medium (A) or starvation medium (B). Scale bar, 5 μm. CMAC marks the vacuolar lumen in rich medium but distributes cytosolically under starvation. Mean number of Ura7-GFP events (puncta or cytoophidia) per cell (C) and mean percent of total Ura7-GFP events colocalizing with Vph1-mCherry (D) in rich or starvation medium. Object-based colocalization analysis (in D) described in Materials and Methods. Graphs show means ± SEM from 3 independent biological replicates (circles in C; triangles in D). Statistical comparisons were performed on biological replicate means by unpaired t-test; p values are shown in the graphs.

AP-3 loss promotes Ura7 cytoophidia elongation under starvation

Having established that Ura7 localizes to vacuoles, we examined whether AP-3 function affects Ura7 assembly dynamics using Ura7-mCherry, which exhibits lower basal assembly compared to Ura7-GFP, facilitating detection of starvation-induced changes. We examined Ura7-mCherry in wild-type, apl5Δ, and apl6Δ cells cultured at log phase in rich growth medium or starved. Log-phase cells showed diffuse cytosolic/nuclear Ura7-mCherry across all three strains with minimal punctate or filamentous cytoophidia structures (Figure 4A–C). After starvation, wild-type cells showed a mixture of puncta and cytoophidia, while AP-3 mutants dramatically shifted toward elongated cytoophidia, increasing the filament-to-puncta ratio ~5-fold (Figure 4G). Ratios in apl5Δ and apl6Δ were indistinguishable from one another, consistent with complete loss of AP-3 function. Interestingly, AP-3 mutants showed significantly fewer total Ura7-mCherry events per cell compared to wild-type (Figure 4H), indicating that AP-3 function promotes Ura7 assembly. Expression of dominant-negative vps4-E233Q, which disrupts endosomal trafficking (Babst et al., 1997), did not significantly alter the filament:puncta ratio or events per cell relative to wild-type (Supplemental Figure S2A–D), indicating that endosomal dysfunction does not drive these assembly changes. Together, these results are consistent with a dual role for AP-3 in modulating Ura7 assembly dynamics under starvation: AP-3 may promote the formation of Ura7 structures while simultaneously restraining their elongation into extended cytoophidia. The dramatic shift in the filament-to-puncta ratio despite reduced total events suggests that, once nucleated in the absence of AP-3, Ura7 structures more readily form elongated cytoophidia.

Figure 4. Loss of AP-3 function promotes Ura7 hyper-assembly under starvation conditions.

Confocal images of wild-type (A,D), apl6Δ (B,E), and apl5Δ (C,F) cells expressing Ura7-mCherry in rich medium (A-C) or after starvation (D-F). Scale bar, 5 μm. CMAC marks the vacuolar lumen in nutrient-rich conditions but distributes cytosolically under starvation. (G) Mean Ura7-mCherry filament-to-puncta ratios in starved cells (filaments = aspect ratio ≥1.5, puncta = aspect ratio <1.5). (H) Mean number of Ura7-mCherry events per cell after starvation. Graphs show means ± SEM from 3 independent biological replicates (circles). Statistical comparisons were performed on biological replicate means by one-way ANOVA with Tukey’s post-hoc test; p values are shown in the graphs.

The opposing effects of AP-3 loss (decreased Ura7-mCherry events but increased filament-to-puncta ratio) suggest AP-3 function influences multiple aspects of Ura7 regulation. Depletion of Ura7 from purified vacuoles isolated from AP-3 mutants (Eising et al., 2022) is consistent with fewer total Ura7 events in these strains, suggesting a vacuole-associated Ura7 pool that depends on AP-3. Loss of this pool, combined with stress from impaired vacuolar trafficking (Darsow et al., 1998; Sun et al., 2004; LaGrassa and Ungermann, 2005), could lower the threshold for filament assembly. One possibility is that AP-3 vesicles deliver a vacuolar receptor or adaptor to which Ura7 binds, such that AP-3 loss reduces Ura7 docking and lowers the local concentration threshold for cytoophidium nucleation and elongation, a model that future studies can evaluate directly.

To connect Ura7 assembly dynamics to a physiological output, we measured intracellular CTP in wild-type, apl5Δ, and apl6Δ cells under log phase and starvation conditions by LC-MS/MS metabolomics (Supplemental Figure S3). CTP levels rose ~40-fold under starvation across all genotypes, suggesting that the starvation-induced collapse in biosynthetic demand for CTP – in nucleic acid synthesis, phospholipid production, and other anabolic processes (Ostrander et al., 1998; Chang and Carman, 2008; Sun and Liu, 2019) – causes CTP to accumulate even as Ura7 is sequestered in cytoophidia, consistent with filament formation serving as a feedback response to rising CTP rather than a mechanism limiting its availability. CTP levels in AP-3 mutants were not statistically distinguishable from wild-type under either condition – despite the ~5-fold shift toward elongated cytoophidia in these strains (Figure 4G) – indicating that this morphological change does not produce a correspondingly large change in bulk CTP pools, consistent with AP-3 influencing Ura7 spatial organization without being rate-limiting for CTP synthesis. UTP levels were also measured, but these values fell below the detection limit under starvation (unpublished results), precluding CTP/UTP ratio calculations for that condition.

Since autophagy is strongly induced by the same nutrient deprivation that triggers Ura7 cytoophidia formation, we examined whether autophagic flux contributes to Ura7 assembly dynamics. Autophagy mutants atg1Δ and atg8Δ showed an opposite phenotype relative to AP-3 mutants: significantly more Ura7-mCherry events per cell but a markedly lower filament:puncta ratio under starvation (Supplemental Figure S2E–H), suggesting that autophagic flux influences the balance between Ura7 puncta accumulation and filament elongation. Notably, the elevated filament:puncta ratio characteristic of AP-3 single mutants was completely suppressed in atg8Δ apl5Δ and atg8Δ apl6Δ double mutants, which were indistinguishable from atg8Δ alone (Supplemental Figure S2E–H). This epistatic relationship indicates that functional autophagy is required for the AP-3 loss phenotype to manifest, consistent with these pathways influencing Ura7 assembly through distinct but interacting mechanisms, the basis of which remains an important question for future investigation.

V-ATPase inhibition induces Ura7 assembly and cytoophidia formation

We next asked whether V-ATPase inhibition affects Ura7 assembly. Wild-type cells expressing Ura7-mCherry were subjected to glucose withdrawal (V1-V0 disassembly; Kane, 1995), concanamycin A (ConcA; proton translocation block; Dröse and Altendorf, 1997), combined treatment, or mock treatment. Either single treatment induced Ura7 aggregates with some filaments (Figure 5B,C), but combined treatment produced a dramatically enhanced phenotype – marked increases in puncta number and size plus extensive cytoophidia formation (Figure 5D–H) – demonstrating that dual V-ATPase inhibition potently induces Ura7 cytoophidia elongation. This synergistic effect fits a model in which Ura7 assembly responds to vacuolar acidification defects and pH dysregulation. CTPS filamentation is triggered by cytosolic acidification (Hansen et al., 2021), and the V-ATPase is the primary cytosolic pH regulator (Martinez-Munoz and Kane, 2008). Glucose withdrawal disassembles the V-ATPase and acidifies the cytosol (Kane, 1995; Dechant et al., 2010); ConcA pretreatment would block compensatory proton pumping during this transition. However, pH was not directly measured here, and alternative mechanisms such as ATP depletion or metabolic stress signaling could contribute.

Figure 5. Dual inhibition of the V-ATPase potently induces Ura7 assembly and filament formation.

Confocal images of wild-type cells expressing Ura7-mCherry subjected to (A) mock treatment, (B) glucose withdrawal, (C) ConcA treatment, or (D) dual treatment. Scale bar, 5 μm. CMAC marks the vacuolar lumen in nutrient-rich conditions but distributes cytosolically under starvation. (E) Mean fluorescence intensity (integrated density) of Ura7-mCherry structures per cell. (F) Mean Ura7-mCherry events per cell. (G) Mean Ura7-mCherry event size (area, μm²). (H) Mean number of filamentous Ura7-mCherry structures (aspect ratio (AR) ≥1.5) per cell. Graphs show means ± SEM from 3 independent biological replicates (circles). Statistical comparisons were performed on biological replicate means by one-way ANOVA with Tukey’s post-hoc test; p values are shown in the graphs.

Collectively, our data establish spatial associations among the AP-3 vesicle coat complex, the V-ATPase, and CTPS, revealing a previously unrecognized organizational principle that links vesicular trafficking, pH regulation, and metabolic enzyme compartmentalization. Given that cytoophidium formation is conserved across eukaryotes (Liu, 2016; Noree et al., 2019), these connections may reflect broader principles of how membrane trafficking influences metabolic enzyme assembly, a possibility that future studies in other systems can now evaluate.

MATERIALS AND METHODS

Yeast Strains and Culture Conditions

All S. cerevisiae strains used in this study were derived from the SEY6210 background (Robinson et al., 1988) unless otherwise noted and were constructed using standard techniques for yeast growth and genetic manipulation. Yeast strains were generated by one-step PCR-based integration (Longtine et al., 1998). Unless otherwise specified, cells were grown in rich medium (YPD: 1% yeast extract, 2% peptone, 2% glucose) at 26°C with shaking at 200 rpm. Strains used in this study were: GOY559 (MATa APL6-VC::KANMX6; this study); GOY1041 (MATα URA7-VN::HIS3MX6; this study)); MYMY109 (MATaα (GOY559 × GOY1041, where “×” denotes mating of the indicated strains); this study); GOY999 (MATα APL5-VC::KANMX6; this study); VMA10VN (BY4741 background (MATa VMA10-VN::KlURA3; Sung et al., 2013)); GOY1100 (MATaα (GOY999 × VMA10VN; this study)); MYMY110 (MATaα (GOY999 × GOY1041; this study)); GOY664 (MATα HIS2-VC::TRP1; this study); MYMY027 (BY4741 background (MATa URA7-VN::KlURA3; Sung et al., 2013)); GOY1012 (MATaα (GOY664 × MYMY027; this study)); GOY1009 (MATα APL6-VC::KANMX6 APL5-mCherry2Bx6::0; this study); GOY1109 (MATaα (GOY1009 × VMA10VN; this study)); GOY1124 (MATaα (VMA10VN × GOY664; this study)); GOY1019 (MATα VMA10-VC::KANMX6; this study); MYMY151 (MATaα (MYMY027 × GOY1019; this study)); MYMY106 (MATa URA8-VN::HIS3MX6; this study); MRBY033 (MATaα (MYMY106 × GOY1019; this study)); GOY661 (MATa HIS2-VN::TRP1; this study); MRBY035 (MATaα (GOY1019 × GOY661; this study)); GOY558 (MATα APL6-VC::KANMX6; this study); MRBY008 (MATaα (MYMY106 × GOY558; this study)); MYMY172 (MATaα (MYMY106 × GOY999; this study)); MRBY024 (MATaα (MYMY106 × GOY664; this study)); GOY1374 (MATa VPH1-mCherry::HPHMX6; this study)); MYMY001 (MATα URA7-GFP::KANMX6; this study)); GOY1394 (MATaα (GOY1374 × MYMY001; this study)); MYMY10 (MATα URA7-mCherry::HPHMX6; this study)); MYMY175 (MATα URA7-mCherry::HPHMX6 apl5Δ::HIS3MX6; this study); MYMY174 (MATα URA7-mCherry::HPHMX6 apl6Δ::HIS3MX6; this study); DCBY24 (MATα VPS8-VC::TRP1; this study); MYMY190 (MATaα (MYMY027 × DCBY24; this study)); MYMY202 (MATα vac8Δ::KANMX6 URA7-mCherry::HPHMX6; this study); MYMY203 (MATα atg18Δ::LEU2 URA7-mCherry::HPHMX6; this study); MYMY204 (MATα atg8Δ::HIS3MX6 URA7-mCherry::HPHMX6; this study); MYMY205 (MATα URA7-mCherry::HPHMX6 atg1Δ::HIS3MX6; this study); MYMY206 (MATα atg8Δ::HIS3MX6 URA7-mCherry::HPHMX6 apl6Δ::HIS3MX6; this study); MYMY207 (MATα atg8Δ::HIS3MX6 URA7-mCherry::HPHMX6 apl5Δ::HIS3MX6; this study). GOY1254 (MATα URA7-VC::KANMX6; this study); RTY078 (MATa HIS2-VN::HIS3MX6; this study); MYMY187 (MAT aα (GOY1254 x RTY078; this study)); GOY1078 (MATa URA7-VC::KANMX6; this study); RTY086 (MATα VPS36-VN:HIS3MX6; this study); MYMY195 (MATaα (GOY1078 x RTY086; this study)); MYMY210 (MATα URA7-mCherry::HPHMX6; pRS416; this study); MYMY211 (MATα URA7-mCherry::HPHMX6; pMB103(vps4[E233Q]); this study).

BiFC Analysis

BiFC reconstitutes a fluorescent Venus protein when complementary N-terminal (VN) and C-terminal (VC) fragments are brought into proximity by interacting or spatially adjacent proteins (Kerppola, 2008). To generate strains for BiFC analysis, we integrated genes encoding Venus C-terminal (VC) or Venus N-terminal (VN) fragments at the chromosomal loci of genes of interest. For BiFC experiments, MATα haploid strains expressing Apl6-VC or Vma10-VC were mated with MATa haploid strains expressing VN fusions to candidate proteins from the Bioneer VN collection (Sung et al., 2013) or custom-constructed VN fusions. Resulting diploids were sporulated, and haploid progeny expressing both VC and VN fusion proteins were selected and confirmed by genetic markers. For experiments with heterozygous diploids (Figures 1 and 2), MATα and MATa haploid strains carrying individual VC or VN fusions were mated, and the resulting diploid cells (expressing both tagged and untagged versions of each protein) were used directly for imaging. Prior to imaging, cells were grown to log phase (OD₆₀₀=0.4–0.6), stained with CMAC-blue vacuolar marker (7-amino-4-chloromethylcoumarin; 50 μM final concentration) for 20 minutes at 26°C, washed, and immediately imaged. Note that CMAC-blue accumulates in the vacuolar lumen in nutrient-replete cells but is excluded from vacuoles and stains the cytosol in starved cells, as vacuolar acidification is reduced under nutrient deprivation.

Nutrient Starvation Treatments

For starvation buffer experiments (Figures 3 and 4), cells were grown to log phase in YPD, collected by centrifugation at 3,000 × g for 3 minutes, washed once with sterile water, and resuspended at the original cell density in citrate-phosphate buffer (0.15 M, pH 6.0). Cells were then incubated with shaking at 26°C for 16–20 hours before imaging. For acute glucose withdrawal experiments (Figure 5), cells were grown to log phase in YPD, collected by centrifugation at 3,000 × g for 3 minutes, washed once with sterile water, and resuspended in YP medium (lacking glucose) at the original cell density. Cells were incubated with shaking at room temperature for 20 minutes, stained with CMAC-blue, and imaged.

Concanamycin A (ConcA) Treatment

For ConcA treatment alone (Figure 5), cells were grown to log phase in YPD and concanamycin A (LC Laboratories) was added directly to the culture at a final concentration of 0.5 μM. Cells were incubated with shaking at room temperature for 90 minutes, stained with CMAC-blue, and imaged. For combined ConcA and glucose withdrawal experiments (Figure 5), cells were first treated with 0.5 μM ConcA for 90 minutes at room temperature as described above. After 90 minutes, cells were collected by centrifugation, washed once with water, and resuspended in YP medium (lacking glucose) supplemented with 0.5 μM ConcA. Cells were incubated for an additional 20 minutes at room temperature, stained with CMAC-blue, and imaged.

Fluorescence Microscopy

All live-cell imaging was performed on a Nikon Ti2E inverted microscope equipped with a CSU-X1 Yokogawa spinning disk confocal scanner unit, OBIS 405, 488, 561, and 640 nm lasers, and an ANDOR iXon Ultra 512 × 512 EMCCD camera. Images were acquired using a Nikon Plan Apo λ 100× oil immersion objective (1.45 N.A.) and controlled with Micro-Manager 2.0 software. For z-stack acquisitions, three optical sections separated by 0.5 μm were collected and collapsed into 2D representations using maximum intensity projections in Fiji (ImageJ, NIH). Images were processed and assembled into figures using Fiji and Adobe Illustrator. All data are presented as mean ± SEM for biological replicate means. Statistical significance was assessed at α = 0.05 using GraphPad Prism software.

Image Quantification and Statistical Analysis

All image quantification was performed using Fiji/ImageJ. For each experimental condition, ≥100 cells were analyzed per biological replicate, and all experiments were performed in biological triplicate. All datasets were first assessed for variance heterogeneity across conditions. Brown-Forsythe tests and F tests confirmed homogeneity of variance across conditions. Statistical comparisons were performed using one-way ANOVA with Tukey’s post-hoc test for experiments involving three or more groups, and unpaired t-tests for two-group comparisons. All analyses were performed on biological replicate means unless otherwise noted. Exact statistical tests and post hoc comparisons used for each dataset are specified in the figure legends and stated below.

BiFC puncta quantification (Figures 1B, 1E, 2B, 2E) and BiFC fluorescence quantification (Figures 1C, 1F, 2C, 2F):

Individual cells were manually segmented using region-of-interest (ROI) outlines, excluding cells with abnormally high diffuse autofluorescence indicative of compromised membrane integrity. Using a custom macro, cells were scored for the presence of ≥1 BiFC puncta per cell, and total fluorescence intensity values of puncta in each cell were recorded. Total puncta fluorescence per cell was divided by the median total puncta fluorescence measured in the His2 negative control from the same imaging session. Statistical comparisons of the average intensities from each replicate were performed using one-way ANOVA with Tukey’s post-hoc test.

Ura7 cytoophidia-vacuole colocalization (Figure 3D):

Multi-channel z-stacks were collapsed by maximum intensity projection and channels were separated for independent processing. Binary masks were generated for each channel using automated thresholding. To identify overlapping structures, the Image Calculator AND operation was applied to generate an intersection mask representing colocalized pixels. Objects in each mask were quantified using the Analyze Particles tool with size exclusion filters (0.05 μm² to infinity) and circularity constraints to exclude noise. The percentage of Ura7-GFP cytoophidia colocalizing with Vph1-mCherry was calculated as: (number of objects in intersection mask / total number of Ura7-GFP objects) × 100. Statistical analysis was performed using unpaired t-test.

Ura7 filament-to-puncta ratio (Figure 4G):

Multi-channel images were separated and the Ura7-mCherry channel was processed using Otsu thresholding. Automated particle analysis was performed on binary masks with a size filter of 0.05 μm² minimum to eliminate background noise. Ura7-mCherry structures were classified by aspect ratio (AR): AR ≥1.5 was categorized as filaments, AR <1.5 as puncta. Automated counts were manually verified against raw data. Statistical comparisons were performed using one-way ANOVA with Tukey’s post-hoc test.

Ura7 events per cell quantification (Figure 4H):

Total Ura7-mCherry events per cell were quantified in starved cells using the same thresholding and particle analysis approach described above for filament-to-puncta analysis. Events were defined as discrete structures with area ≥0.05 μm² detected above threshold. The total number of events per cell was counted for each strain. Statistical comparisons were performed using one-way ANOVA with Tukey’s post-hoc test.

Ura7 assembly quantification following V-ATPase inhibition (Figures 5E–H):

For quantification of Ura7 structures (puncta and filaments), the Ura7-mCherry channel was thresholded using the Otsu method. The threshold value determined for wild-type mock-treated cells was applied uniformly to all experimental conditions to ensure consistent detection. Particle analysis was performed to quantify the number, area, integrated density, and aspect ratio of Ura7 structures. Filamentous structures were defined as objects with AR ≥1.5 and area ≥0.5 μm². Statistical analysis was performed using one-way ANOVA with Tukey’s post-hoc test.

Classification of Ura7 structures:

Following established criteria (Noree et al., 2010; Barry et al., 2014), Ura7 structures were classified based on morphology. Structures <0.75 μm in length are defined as puncta (foci), representing nucleation sites or smaller assemblies, while structures ≥0.75 μm are defined as filaments (cytoophidia). For automated quantification of filament-to-puncta ratios, we used aspect ratio (AR) as a proxy for elongation, with AR ≥1.5 corresponding to filamentous structures and AR <1.5 corresponding to punctate structures. This approach was validated by manual inspection confirming that structures with AR ≥1.5 consistently represented elongated filaments rather than circular puncta.

CTP Measurements

Measurements of CTP and UTP from yeast cells were performed by the University of Colorado Anschutz School of Medicine Metabolomics Core. Each cell pellet was lysed with cold 50:30:20 methanol:acetonitrile:water to a final concentration of 40 million cells/mL. Additionally, 1 mm glass beads were added, then the samples were vortexed for 20 seconds at 4 °C and immediately placed on dry ice for 30 seconds. This was repeated a total of 3 times, followed by centrifugation for 5 min at 11,000 rpm at 4 °C. Supernatant was placed in autosampler vials for analysis by UHPLC (Vanquish, ThermoFisher) coupled to a mass spectrometer (Orbitrap Exploris 120, ThermoFisher). Separation was achieved using a Waters Acquity UHPLC BEH C18 column (1.7 μm, 2.1 × 100 mm) with 2 μL injection. A 5-minute negative polarity method was run at 45°C with mobile phases (A) 10 mM ammonium acetate in water and (B) 10 mM ammonium acetate in 50:50 acetonitrile:methanol and the following gradient: 0.45 mL/min, 0% B at 0 min, 0% B at 0.5 min, 100% B at 1.1 min, 100% B at 2.75 min, 0% B at 3 min, 0% B at 5 min. Samples were analyzed in randomized order with a technical mixture injected interspersed throughout the run to qualify instrument performance. For data analysis, acquired data was converted from raw to mzXML file format using RawConverter, then uploaded into El Maven, an open-source software. Peak areas for UTP and CTP were annotated and integrated alongside an in-house compound library.

Abbreviations:

AP-3

adaptor protein complex 3

BiFC

bimolecular fluorescence complementation

CTPS

cytidine triphosphate synthase

V-ATPase

vacuolar H⁺-ATPase