Glycolysis is intrinsically required in the follicle stem cell lineage but not in female germline stem cells or their early progeny in Drosophila melanogaster

Abstract

Multiple tissue stem cells depend on glycolysis or β-oxidation for cell fate decisions. However, how universal these requirements are and how they change as stem cell daughters undergo differentiation remains unclear. The Drosophila ovary is a powerful stem cell model with two distinct stem cell populations: germline stem cells (GSCs), which produce oocytes to perpetuate the species, and follicle stem cells (FSCs), a somatic lineage. Several studies have begun addressing the roles of metabolism within the Drosophila female GSC lineage, but direct systematic analyses of glycolysis and/or mitochondrial fatty acid β-oxidation requirements across these lineages have been lacking. Here, using genetic mosaic analysis with null alleles, we found that genes encoding key regulatory glycolytic enzymes—Phosphofructokinase (Pfk) and Pyruvate kinase (Pyk)—are not cell autonomously required for GSC maintenance, proliferation, or early differentiation through 16-cell germline cyst formation and oocyte specification. Although germline cysts lacking Pfk or Pyk function can develop through early vitellogenesis, they grow slowly and display impaired nurse cell chromatin dispersal. By contrast, FSCs and their early daughters require Pfk (but not Pyk) for normal survival, while later follicle cells need both Pfk and Pyk for survival and only Pfk for proliferation, suggesting that follicle cells predominantly require glycolytic intermediates upstream of Pyk. Surprisingly, mitochondrial β-oxidation was dispensable in both lineages. These findings uncover an unusual metabolic state in GSCs and their early daughters, with marked differences from the neighboring FSC lineage and other somatic stem cells.

ARTICLE SUMMARY

Tissue stem cells often rely on glycolysis or β-oxidation, yet how universal these requirements are or how they change during differentiation remains unclear. Here, we genetically delineate glycolytic and mitochondrial β-oxidation requirements along neighboring Drosophila germline stem cell (GSC) and somatic follicle stem cell (FSC) lineages. Key glycolytic enzymes are dispensable in GSCs and their early differentiating daughters but support later germline growth and are needed for survival and/or proliferation of FSCs and their daughters, while neither lineage requires mitochondrial β-oxidation. These findings reveal a unique metabolic state in GSCs and striking metabolic differences between germline and somatic stem cells.

INTRODUCTION

The roles of local signals and systemic factors in controlling tissue stem cell lineages have been studied for decades (Morrison and Spradling 2008; Nakada et al. 2011; Ables et al. 2012), while the importance of their metabolic regulation has gained attention only more recently (Chandel et al. 2016; Jackson and Finley 2024). Several studies have used genetic approaches to examine metabolic requirements of tissue stem cells in vivo. For example, mitochondrial metabolism of pyruvate, which is produced from glucose oxidation through glycolysis, limits the proliferation of mouse neural stem cells (Petrelli et al. 2023), mouse and Drosophila intestinal stem cells (Schell et al. 2017), and mouse hair follicle stem cells (Flores et al. 2017). In mouse hematopoietic stem cells, activation of mitochondrial pyruvate metabolism stimulates proliferation and disrupts stemness instead (Takubo et al. 2013). Correlative data also suggest that mouse spermatogonial stem cells require glycolysis for self-renewal in vivo (Kanatsu-Shinohara et al. 2016). There is also evidence for key roles of fatty acid β-oxidation in stem cells. Rodent neural stem cells rely on fatty acid oxidation to maintain aerobic respiration and limit proliferation (Stoll et al. 2015; Knobloch et al. 2017)—unlike differentiated neurons, which depend on glucose and lactate metabolism (Wyss et al. 2011; Lundgaard et al. 2015; Li et al. 2023). Transcriptional regulation of genes involved in β-oxidation support self-renewal of mouse intestinal stem cells and hematopoietic stem cells (Ito et al. 2012; Chen et al. 2020). In Drosophila, several studies have suggested roles of fatty acid β-oxidation in male germline stem cell (GSC) maintenance and in intestinal stem cell survival and proliferation (Sênos Demarco et al. 2019; Zhang et al. 2022; Zipper et al. 2022). And in Drosophila larvae, fatty acid β-oxidation is required for differentiation of hematopoietic progenitors in the lymph gland (Tiwari et al. 2020). However, much remains unknown about how universal these stem cell metabolic requirements are and how they change along tissue stem cell lineages as stem cell daughters gradually undergo complex differentiation processes.

The Drosophila ovary is a powerful model for addressing the connection between metabolism and adult stem cell lineages in vivo. Each ovary is subdivided into ~15 ovarioles (Fig. 1A), each containing a germarium (Fig. 1B) followed by progressively more developed egg chambers. Each egg chamber is composed of a germline cyst—one oocyte and 15 supporting nurse cells—surrounded by follicle cells (Spradling 1993). The germarium houses distinct types of tissue stem cells, GSCs and somatic follicle stem cells (FSCs), that together give rise to egg chambers. In region 1 of the germarium, two-to-three GSCs are attached to a niche composed primarily of somatic cap cells. As each GSC divides, it self-renews and generates a cystoblast that begins differentiation. The cystoblast undergoes four rounds of mitoses with incomplete cytokinesis to form a two-, four-, eight-, and, finally, 16-cell cyst, marking the beginning of region 2A. In region 2A, the newly formed 16-cell cyst undergoes pre-meiotic S phase and, by region 2B, the oocyte is in early meiosis and nurse cells begin endoreplicating to support oocyte growth and development. Two FSCs located immediately anterior to the border between regions 2A and 2B generate follicle cells that begin enveloping the lens-shaped germline cysts in region 2B. Region 3 of the germarium represents a stage 1 egg chamber, composed of a round 16-cell cyst fully surrounded by a monolayer of mitotic follicle cells. As egg chambers leave the germarium, they continue to develop through 14 recognizable stages. During the first four stages, nurse cells are polytene with their chromatin organized as five “blobs” per nucleus, corresponding to the five major chromosome arms (Dej and Spradling 1999). Starting in stage 5 and completed by stage 6, nurse cell sister chromatids separate, and the chromatin appears more evenly dispersed. Stage 8 is defined by the onset of vitellogenesis, when the oocyte begins taking up yolk proteins and lipids. During later stages, glycogen accumulates, and oocyte meiotic maturation occurs to form a mature stage 14 oocyte (Wessel and Drummond-Barbosa 2025). Diet and other physiological signals regulate GSC and FSC maintenance and proliferation, and the proliferation, growth, and survival of their differentiating daughters (Drummond-Barbosa 2019; Wessel and Drummond-Barbosa 2025).

Fig. 1.

Mutant alleles used for investigating the metabolic requirements of the germline and follicle stem cell lineages in the Drosophila ovary. (A) Drosophila ovariole showing the germarium and chronologically arranged egg chambers. Each egg chamber consists of a germline cyst—15 nurse cells and one oocyte—surrounded by somatic follicle cells, which also form stalks separating egg chambers. Oocytes begin taking up yolk during stage 8 of oogenesis, marking the onset of vitellogenesis. (B) Germarium showing germline stem cells (GSCs) within a niche made up primarily of cap cells (region 1). Each GSC divides to self-renew and give rise to a cystoblast, which undergoes four rounds of mitoses with incomplete cytokinesis (region 1) to form a 16-cell germline cyst (region 2A). Follicle stem cells (FSCs) give rise to follicle cells that envelop the 16-cell cyst (region 2B) to form a new egg chamber ready to leave the germarium (region 3). (C) Simplified diagram of glycolysis, β-oxidation, and the tricarboxylic acid (TCA) cycle. (D to I) Diagram of Pfk (D), Pyk (E), whd (F), CPT2 (G), Mtpα (H), and Mtpβ (I) genes showing mRNA isoforms and mutant alleles. For mRNA isoforms, grey boxes represent exons, with coding regions in darker shade. Insertional alleles are indicated by red arrowheads. Frameshift mutations caused by small insertions and deletions (indels) are indicated by vertical lines with arrows, with premature stop codons shown as red hexagons. The deletions in CPT2^raeg, Mtpα^KO, and Mtpβ^KO alleles are indicated as black lines with bracketed missing segments. In (D'), part of nucleotide sequences of the wildtype Pfk and newly generated null Pfk¹ alleles shows where the indel occurred in Pfk¹, causing a frameshift and changing the amino acid sequence. Underlined amino acid residues (amino acids 120-123) are directly involved in ATP binding within the Pfk active site, which extends further downstream. For Mtpα^KO (H) and Mtpβ^KO (I), the white gene sequence was used to replace the deleted region (Kishita et al. 2012).

Although several studies have begun addressing the roles of metabolism within the Drosophila female GSC lineage, many questions remain. Mitochondria remodel during early stages of germline differentiation in the germarium to increase electron transport chain activity (Cox and Spradling 2003; Wang et al. 2019; Garcez et al. 2021; Monteiro et al. 2023; Wang et al. 2023), while in the final stages of oogenesis mitochondria remodel again to go into respiratory quiescence (Sieber et al. 2016). A recent study concluded that high levels of glycolysis are required for female GSC self-renewal through clonal analysis of Aldolase and Enolase—showing that mutant GSC clones are lost at higher rates than control clones—and through RNAi of other glycolytic enzymes, which did not include precise counting of the number of GSCs per germarium (Rojas-Ríos et al. 2024). However, Aldolase and Enolase control reversible, non-regulatory glycolysis reactions and both have glycolysis-independent "moonlighting" roles—e.g., regulation of cytoskeletal proteins and gene expression, among others (Didiasova et al. 2019; Bian et al. 2022; Horvat et al. 2024; Yu et al. 2025), bringing into question whether the observed GSC loss was indeed caused by changes in glycolysis. Another germline RNAi-based study suggested that knockdown of Hexokinase A, encoding the enzyme that catalyzes the first step of glycolysis, but not of other regulatory glycolytic enzyme genes, caused reduced egg laying (Carvalho-Santos et al. 2020). It has also been proposed, based on indirect evidence, that β-oxidation might be required in female GSCs (Amartuvshin et al. 2020). However, direct systematic analyses of the intrinsic requirements for glycolysis and/or mitochondrial fatty acid β-oxidation at different stages along the GSC and FSC lineages have not been done.

Here, we uncover very distinct metabolic requirements for the GSC and FSC lineages in the Drosophila ovary using extensive genetic mosaic analyses. We find that genes encoding two key regulatory enzymes in glycolysis—Phosphofructokinase (Pfk) and Pyruvate kinase (Pyk) (Fig. 1C)—are not required in the germline for early processes, from GSC maintenance and proliferation through the formation of 16-cell germline cysts. Strikingly, egg chambers containing Pfk or Pyk mutant germline cysts can still grow and undergo vitellogenesis; however, they grow more slowly and their nurse cells do not undergo normal chromatin dispersal. By contrast, there is a strong requirement for Pfk throughout the FSC lineage for cell survival and, in later follicle cells, for proliferation, while Pyk is only required in later follicle cells for survival. Surprisingly, we found no requirements for mitochondrial β-oxidation genes in either the GSC or FSC lineage, even when glycolysis is disrupted. These findings indicate that the GSC lineage meets a large part of its significant demand for energy and macromolecules through a distinct metabolic strategy that appears to be largely independent of the cell-autonomous oxidation of sugars or lipids. Instead, they suggest that the germline largely relies on alternative sources of energy and/or intensive metabolic cooperation with other tissues for most of its processes.

MATERIALS AND METHODS

Drosophila strains and culture conditions

Drosophila stocks were maintained at 22°C on standard medium composed of 4.64% w/v yellow cornmeal (Quaker, Chicago, IL, USA; #030000570203), 5.8% v/v unsulfured cane molasses (Sweet Harvest Feeds, Cannon Falls, MN, USA; #F682), 1.74% w/v active dry yeast (Red Star, Milwaukee, WI, USA; #117929157002), 0.93% w/v agar (Apex BioResearch Products, El Cajon, CA, USA; #20-273), 1.05% Tegosept (Apex Chemicals, El Cajon, CA, USA; #20-258) and 0.36% propionic acid (Apex Chemicals, El Cajon, CA, USA; #20-271). During experiments, females were kept with males in an incubator (Darwin Chambers Company, St. Louis, MO, USA) at 25°C, 80% humidity, under 12-hour light cycles. Flies were transferred daily to fresh vials with standard medium supplemented with dry yeast or, starting two days prior to dissection, yeast paste. For experiments using 1% sugar, the concentration of molasses (~75% sugar) was adjusted to 1.33% instead. The following alleles have been previously described: Pfk⁰⁶³³⁹ (Tennessen et al. 2011); Pyk³¹ (Heidarian et al. 2024); whd¹ (Strub et al. 2008); Mtpα^KO and Mtpβ^KO (Kishita et al. 2012) (Supplementary Table 1). nos-Cas9, MTD-Gal4, UAS transgenes, FRT and hs-FLP transgenes, PBac{GV-CH321-187O16}VK00031, and other genetic elements are described in FlyBase (Öztürk-çolak et al. 2024) (Supplementary Table 1). FRT recombinant lines were generated by standard crosses and verified by complementation tests and/or PCR.

Generation of null Pfk¹ and CPT2^raeg alleles and Pfk genomic rescue construct

Flies carrying nanos-cas9 and sgRNA-NIG-FLY.Pfk were used as described (Ren et al. 2013) to generate the null Pfk¹ allele, which failed to complement the lethality of Pfk⁰⁶³³⁹ and was further verified by sequencing using primers 5'-CGGCAAGGTCTACTTCATTC-3' and 5'-CGATGGAGCCAACCTACA-3'. To generate the CPT2^raeg deletion allele, we used the pCFD5 vector (a gift from Simon Bullock; Addgene plasmid # 73914) as described (Port and Bullock 2016) to build a construct containing two sgRNAs (5’-GGATAGCCATGTTTCGGACC-3’ and 5’-ATGGGAAGGTGTTCATAGAC-3’) targeting opposite ends of CPT2, which was sent to Rainbow Transgenics for generation of transgenic line P{v⁺=CPT2.dgRNA}attP40. This line was crossed to nanos-Cas9 to generate CPT2^raeg, which was identified using PCR. CPT2^raeg homozygous escapers had the typical "withered" wings phenotypes associated with disrupted fatty acid transport into mitochondria (Strub et al. 2008). For the Pfk genomic rescue construct, BAC clone CH322-13F12, obtained from the BACPAC Resources Center (now operating as BACPAC Genomics, Inc., Redmond, Washington, USA), was confirmed by sequencing and sent as a bacterial stab to Best Gene, Inc. (Chino Hills, CA, USA), for integration into the attP docking site VK00033 on chromosome 3L using ΦC31 integrase, as described (Venken et al. 2009). The Pfk genomic insertion transgene is referred to as PBac{CH322-13F12}VK00033.

Genetic mosaic analyses

Genetic mosaics were generated by FLP-FRT-mediated recombination (Xu and Rubin 1993). One- to two-day-old females carrying a mutant allele in trans to a wildtype allele (linked to either a Ubi-GFP or His-RFP marker) on homologous FRT arms and a hs-FLP transgene were heat-shocked for 1 hour at 37°C twice daily for 3 consecutive days, as described (Laws and Drummond-Barbosa 2015). In control mosaics, wildtype instead of mutant alleles were used. Ovaries were dissected 10 days after the last heat shock to ensure that only stem cell clones were analyzed (Margolis and Spradling 1995; Laws and Drummond-Barbosa 2015). GSCs were identified based on the juxtaposition of their fusomes to adjacent cap cells (de Cuevas and Spradling 1998), while FSCs were identified as the anterior-most, marker-negative follicle cell in the region immediately adjacent to germarium region 2A–2B border (Margolis and Spradling 1995; Fadiga and Nystul 2019). Early germline cysts were staged based on fusome morphology (de Cuevas and Spradling 1998). Egg chambers were staged based on size and morphology (Spradling 1993).

Genetic mosaic phenotypic analysis was performed as previously described (Laws and Drummond-Barbosa 2015). Briefly, to quantify GSC and FSC loss, we analyzed all germaria containing GFP- or RFP-negative stem cell daughters and calculated the percentage of germaria that no longer contained the original stem cell that gave rise to those daughters (i.e., GSC or FSC loss events). To objectively score egg chamber growth delay, we first created standards by recording stages flanking GFP- or RFP-negative egg chambers in FRT control mosaic ovarioles, as described (LaFever and Drummond-Barbosa 2005). We scored experimental egg chambers as delayed if they were followed by an egg chamber older than the range of stages that normally follow an FRT control egg chamber and not delayed if they fell within the standards. For chromatin dispersal defects, GFP- or RFP-negative germline cysts in stage 6 or older egg chambers were analyzed and were scored as "non dispersed" if any of the nurse cells retained the "five-blob" chromatin morphology (Dej and Spradling 1999). For categorial data, including stem cell loss, presence of 16-cell cysts in premeiotic S-phase, growth delay, presence of follicle cell apoptosis, and encapsulation defects, statistical analysis was performed using Fisher's exact test or Chi-square test (Xu et al. 2010); except for data in Fig. 5J, where mixed-effects logistic regression and estimated marginal means (Atwill et al. 1995) were used to account for ovariole-specific random effects and interdependence of outcomes within each ovariole. For numerical data, including quantification of the numbers of cystoblasts, two-, four-, eight- and 16-cell germline cysts, follicle cells, and EdU-positive follicle cells, data were presented as mean±SEM and statistical analysis was performed using unpaired, two-tailed Student’s t-test. (See Supplemental Data File 1.)

Immunostaining, EdU labeling, and microscopy

Ovaries were dissected in Grace's Insect Medium (Gibco, Waltham, MA; #11595-030), teased apart, and fixed for 15 min at room temperature in 5.3% formaldehyde (Pierce, Waltham, MA, USA; #28908) in Grace's. Samples were rinsed and washed three times for 15 min each in PBST [0.1% Triton X-100 in PBS (10 mM NaH₂PO₄/NaHPO₄, 175 mM NaCl, pH 7.4)]. Samples were blocked for 3 h at room temperature or overnight at 4°C in 5% normal goat serum (MP Biochemicals, Irvine, CA, USA; #642921) and 5% bovine serum albumin (Sigma-Aldrich, St. Louis MO, USA; #9048-46-8) in PBST and then incubated at 4°C overnight with the following primary antibodies diluted in blocking solution: 1:20 mouse anti-α-Spectrin [Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA, USA; #3A9]; 1:100 mouse anti-Lamin C (DSHB; LC28.26); 1:1000 chicken anti-GFP (Abcam, Cambridge, UK; #ab13970); 1:100 rabbit anti-cleaved Drosophila Dcp-1 (Asp 215) (Cell Signaling Technology, Danvers, MA, USA; #9578); 1:400 rabbit anti-Histone H2AvD phosphoS137 (Rockland, Limerick, PA, USA; #600-401-914); 1:20 mouse anti-Orb (DSHB; #4H8). Ovaries were then washed in PBST and incubated with 1:400 Alexa Fluor^™ 488-conjugated goat anti-rabbit (ThermoFisher Scientific, Waltham, MA, USA; #A-11034), Alexa Fluor^™ 488-conjugated goat anti-chicken (ThermoFisher Scientific, #A-11039), Alexa Fluor^™ 488-conjugated goat anti-mouse (ThermoFisher Scientific, #A-11001), Alexa Fluor^™ 568-conjugated goat anti-mouse (ThermoFisher Scientific, #A-11004), Alexa Fluor^™ 568-conjugated goat anti-rabbit (ThermoFisher Scientific, #A-11011), Alexa Fluor^™ 633-conjugated goat anti-rabbit (ThermoFisher Scientific, #A-21070), or Alexa Fluor^™ 633-conjugated goat anti-mouse (ThermoFisher Scientific, #A-21050) in blocking solution for 2 h at room temperature. Samples were washed in PBST and mounted in Vectashield with 1.5 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Newark, CA, USA; #H-1200-10). For 5-ethynyl-2'-deoxyuridine (EdU) incorporation assays, ovaries were dissected in Grace's medium and incubated in Grace's with 100 μM EdU (Molecular Probes, Eugene, OR, USA; #C10339) at room temperature for 1 h prior to being teased apart, fixed, and stained as above. EdU was labeled with AlexaFluor-594 via Click-It chemistry according to the manufacturer's instructions prior to secondary antibody incubation. All images were collected using a ZEISS (Oberkochen, Germany) LSM900 with Airyscan 2 confocal laser microscope. Objective Plan-Apochromat 63x/1.40 Oil DIC M27 (ZEISS; #420782-9900-799) was used for images shown in Fig. 2B to E, H, J, and K, Fig. 3A to E, Fig. 4A to E, J and K, Fig. 5D, G, and H, Fig. 6A, and Supplementary Fig. 2A, B, D, and E. Objective Plan-Apochromat 40x/1.3 Oil DIC M27 (ZEISS; #420762-9800-799) was used for images shown in Fig. 2L and M, Fig. 3A to D and F, Fig. 4L, Fig. 5A, B, and E, Fig. 6D, G, I, K, and L, Supplementary Fig. 2C, F, and G, and Supplementary Fig. 3D and E.

Fig. 2.

Glycolysis is not required in germline stem cells or their early differentiating daughters. (A) Diagram of FLP-FRT technique used to generate homozygous mutant stem cells in the context of a heterozygous tissue. A wildtype allele (+) of the gene of interest linked to an ubiquitously expressed GFP or RFP is placed in trans to a mutant (or control wildtype) allele (star). Flippase-mediated recombination between FRT sites (white squares) after DNA replication can generate unequal sister chromatids; if both mutant (or control) alleles segregate into a stem cell, this homozygous mutant (or control) stem cell and its daughter cells can be recognized by the absence of GFP or RFP. (B to E) Maximal projection images of two or three optical slices of control (B and D), Pfk¹ (C), or Pyk³¹ (E) mosaic germaria at 10 days after clone induction. GFP-negative GSCs are outlined. GFP-negative cystoblasts and cysts are indicated by arrows. GFP (green) labels cells carrying a wildtype allele of the relevant gene. Lamin C labels cap cell nuclear envelopes (fuschia). α-Spectrin (fuchsia) labels fusomes. Scale bar: 5 μm. (F) Frequency of germaria with a GSC loss event (i.e., germaria where GFP-negative cystoblasts and/or cysts are present but the mother GFP-negative GSC has been lost) in Pfk¹ and Pyk³¹ mosaic germaria and their respective FRT42D and FRT82B controls 10 days after clone induction. The numbers of germline mosaic germaria analyzed are indicated above bars. No significant differences; Fisher’s exact test. (G) Average numbers of GFP-negative (or GFP-positive) cystoblasts, two-, four-, eight-, and 16-cell cysts normalized per GFP-negative (or GFP-positive) GSC in Pfk¹ and Pyk³¹ mosaic germaria and their respective FRT controls. The numbers of germaria containing at least one GFP-positive GSC and one GFP-negative GSC analyzed are indicated above bars. Data are shown as mean±SEM. *P <0.05; Student’s t-test. (H) Maximal projection image of six optical slices of a Pfk¹ mosaic germarium. A GFP-negative 16-cell cyst undergoing pre-meiotic S phase in region 2A is outlined. GFP (green) labels wildtype cells. Lamin C (blue) labels cap cell nuclear envelopes. α-Spectrin (blue) labels fusomes. EdU (fuchsia) labels cells in S phase. DAPI (white) labels nuclei. Single channels for GFP (H'), Lamin C and α-Spectrin (H"), and EdU (H"') are also shown in grayscale. Scale bar: 5 μm. (I) Frequencies of EdU-positive 16-cell cysts in Pfk¹ or Pfk⁰⁶³³⁹ compared to FRT42D control mosaics. The numbers of region 2A 16-cell cysts analyzed are indicated above bars. No significant differences; Chi-square test. (J) Maximal projection image of five optical slices of a Pfk⁰⁶³³⁹ mosaic germarium. GFP (green) labels wildtype cells. Lamin C (blue) labels cap cell nuclear envelopes. α-Spectrin (blue) labels fusomes. γH2Av (fuchsia) labels DNA breaks. DAPI (white) labels nuclei. Single channels for GFP (J'), Lamin C and α-Spectrin (J"), and γH2Av (J"') are also shown in grayscale. A Pfk⁰⁶³³⁹ homozygous GFP-negative 16-cell cyst labeled with γH2Av is outlined in yellow, while a neighboring wildtype GFP-positive 16-cell cyst also labeled with γH2Av is indicated by an arrowhead. Scale bar: 5 μm. (K) Maximal projection image of three optical slices of a Pfk⁰⁶³³⁹ mosaic stage 2 egg chamber. Arrow indicates the oocyte in the GFP-negative germline cyst. GFP (green) labels wildtype cells. Lamin C (fuchsia) labels stalk cell nuclear envelopes. Orb (fuchsia) is an oocyte-enriched protein. DAPI (white) labels nuclei. Single GFP channel is also shown in grayscale (K'). Scale bar: 5 μm. (L and M) Single confocal slices of FRT42D control (L) and Pfk¹ GFP-negative germline cysts in stage 7 egg chambers showing oocytes with normal karyosomes (arrows). GFP (green) labels wildtype cells. DAPI (white) labels nuclei. Scale bar: 20 μm. (See Supplementary Data File 1 for raw data.)

Fig. 3.

Glycolysis is required in germline cysts for egg chamber growth and nurse cell chromatin dispersal. (A to D) Single tiled optical slices of FRT42D control (A), Pfk¹ (B and C), and Pyk³¹ (D) mosaic ovarioles. Arrows indicate GFP-negative germline cysts. Asterisks indicate delayed egg chambers. Insets show representative nurse cell nuclei. GFP (green) labels cells carrying a wildtype allele of the relevant gene. DAPI (white) labels nuclei. Scale bar: 20 μm. (E) Frequencies of GFP-negative germline cysts showing delayed egg chamber growth in Pfk¹ and Pyk³¹ mosaics compared to controls. The numbers of egg chambers with GFP-negative germline cysts analyzed are indicated above bars. ****P <0.0001; Fisher’s exact test. (F) Single optical slice of a stage 9 egg chamber with a homozygous Pfk¹ germline cyst labeled as in (A to D). Differential interference contrast image of the same egg chamber (F') shows the presence of yolk in the oocyte (arrow). Scale bar: 20 μm. (G) Frequency of FRT control, Pfk¹, or Pyk³¹ mosaic ovarioles with vitellogenic egg chambers containing GFP-negative germline cysts. Note that for Pfk¹ and Pyk³¹, all of these vitellogenic egg chambers were in stage 8 or 9 of oogenesis. The numbers of germline-mosaic ovarioles analyzed are indicated above bars. ***P <0.001, *P<0.05; Chi-square test. (H) Frequencies of stage 6 or later egg chambers with GFP-negative germline cysts showing nurse cells that failed to undergo normal chromatin dispersal. The numbers of stage 6 or later egg chambers with GFP-negative germline cysts analyzed are indicated above bars. ****P <0.0001; Fisher’s exact test. (See Supplementary Data File 1 for raw data.)

Fig. 4.

Pfk but not Pyk is required in FSCs and their early daughters. (A to E) Maximal projections of two or three optical slices of FRT42D control (A), Pfk¹ (B to D), and Pyk³¹ (E) mosaic germaria at 10 days after clone induction. In (C), two Pfk genomic rescue transgenes are also present (Rescued Pfk). GFP-negative FSC clones (containing FSC and/or its follicle cell daughters) are outlined in yellow. GFP (white) labels wildtype cells. Arrowheads indicate position of GFP-negative FSCs. In D, fusomes and cleaved Dcp-1 are labeled (fuchsia). The arrow indicates an apoptotic GFP-negative follicle cell. Scale bar: 5 μm. (F) Frequency of germaria with FSC loss events [i.e., germaria where GFP-negative follicle cells are present but there is no GFP-negative FSC at the 2A/2B border, such as in (B)] in FRT control, Pfk¹, and Pyk³¹ mosaics at 10 days after clone induction. In the middle graph, germaria from Pfk¹ mosaic females carrying two genomic transgenes carrying wildtype copies of Pfk (Rescued Pfk) were compared with germaria from sibling Pfk¹ mosaic females without these rescue constructs. The numbers of follicle cell mosaic germaria analyzed are indicated above bars. *P<0.05; Chi-square test or Fisher’s exact test. (G) Total numbers of GFP-negative follicle cells present in the mosaic germaria of same genotypes as in (F). The number of follicle cell mosaic germaria containing a GFP-negative FSC analyzed are indicated above bars. Data are shown as mean±SEM. ****P<0.0001, **P<0.01; Student's t-test. (H) Percentages of GFP-negative follicle cells that are EdU-positive within the germaria of FRT42D control, Pfk¹, or Pfk⁰⁶³³⁹ germaria at 5 or 10 days after clone induction. The numbers of follicle cell mosaic germaria analyzed are indicated above bars. Data are shown as mean±SEM. No significant differences; Student’s t-test. (I) Frequencies of germaria with at least one GFP-negative cleaved Dcp-1-positive follicle cell in FRT42D control and Pfk¹ mosaics. The numbers of follicle cell mosaic germaria analyzed are indicated above bars. **P<0.01; Fisher’s exact test. (J to L) Maximal projections of one or two optical slices showing abnormally encapsulated Pfk¹ mosaic egg chambers where two (J) or more (K) GFP-positive germline cysts are enveloped at least partially by GFP-negative follicle cells in contrast to normally encapsulated Pyk³¹ egg chambers (L). GFP (green) labels wildtype cells. DAPI (white) labels nuclei. GFP-negative follicle cell clones are outlined in yellow. Scale bar: 20 μm. (M) Quantification of encapsulation defects mosaic egg chambers containing GFP-negative follicle cells in FRT control, Pfk¹, or Pyk³¹ mosaics. The numbers of egg chambers covered by at least 50% GFP-negative follicle cells analyzed are indicated above bars. **P <0.01; Fisher’s exact test. (See Supplementary Data File 1 for raw data.)

Fig. 5.

Pfk and Pyk are cell autonomously required for the survival of later follicle cells but only Pfk is necessary for their normal proliferation. (A and B) Single optical slices of FRT42D control (A) and Pfk¹ (B) mosaic egg chambers containing GFP-negative follicle cell clones (outlined). GFP (white) labels wildtype cells. Scale bar: 20 μm. (C) Quantification of the fraction of GFP-negative follicle cells covering stage 4 to 6 egg chambers in Pfk¹ or Pyk³¹ mosaic females compared to FRT controls. The numbers of follicle cell mosaic egg chambers analyzed are indicated above bars. Data are shown as mean±SEM. ****P <0.0001, *P <0.05; Student’s t-test. (D and E) Single tiled optical slices of Pfk¹ mosaic ovarioles with a GFP-negative follicle cell clones (outlined) containing dying follicle cells (arrows). In (E), arrowheads indicate growth delayed GFP-positive germline cysts that have lost their follicle cell monolayer. GFP (green) labels wildtype cells. Cleaved Dcp-1 (fuchsia) labels apoptotic cells. DAPI (white) labels nuclei. Single channel for DAPI (E') is also shown in grayscale. Scale bars: 20 μm. (F) Fraction of stage 4 to 6 egg chambers with one or more cleaved Dcp-1-positive GFP-negative follicle cells in Pfk¹, Pyk³¹, or control FRT mosaics. The numbers of follicle cell mosaic egg chambers analyzed are indicated above bars. ** P<0.01; Fisher’s exact test. (G and H) Single optical slices of EdU-labeled FRT42D control (G) or Pfk¹ (H) mosaic egg chambers containing GFP-negative follicle cells. Follicle cell clones are outlined. GFP (white) labels cells with a wildtype copy of Pfk. EdU (fuchsia) labels cells in S-phase. Scale bars: 20 μm. (I) Fraction of GFP-positive or GFP-negative follicle cells labeled with Edu in Pfk¹, Pyk³¹, or control FRT mosaic stage 4 to 6 egg chambers. The numbers of follicle cell mosaic egg chambers analyzed are indicated above bars. Data are shown as mean±SEM. *P <0.05; Student’s t-test. (J) Distribution of egg chambers containing GFP-negative follicle cells among 30 FRT42D control and 30 Pfk¹ mosaic ovarioles. *P<0.05, **P<0.01, ***P<0.001; Mixed effect logistic regression and estimated marginal means. (K) Fraction of egg chambers covered by at least 50% GFP-negative follicle cells showing growth delay in Pfk¹, Pyk³¹, or FRT control mosaics. The numbers of egg chambers containing 50% or more GFP-negative follicle cells analyzed are indicated above bars. *P <0.05, ***P <0.001 Fisher’s exact test. (See Supplementary Data File 1 for raw data.)

Fig. 6.

Mitochondrial β-oxidation is not required in the GSC or FSC lineages. (A) Maximal projections of three optical slices of a whd¹ mosaic germarium showing a GFP-negative GSC (outlined) and its daughter cells (arrows). GFP (green) labels wildtype cells. Lamin C labels cap cell nuclear envelopes (blue). α-Spectrin (blue) labels fusomes. Scale bar: 5 μm. (B) Frequency of germaria with a GSC loss event in FRT42D control or whd¹ mosaics (left), CPT2^raeg mosaics (without or with one copy of Pfk⁰⁶³³⁹ in the background) (middle), and Pfk¹ mosaics or Pfk¹ Mtpβ^KO double mosaics (right). The numbers of germline mosaic germaria analyzed are indicated above bars. No significant differences; Fisher’s exact test. (C) Quantification of total number of cystoblasts and cysts per GSC for same genotypes as in (B). The number of germaria containing at least one RFP/GFP-positive GSC and one RFP/GFP-negative GSC analyzed are indicated above bars. Data are shown as mean±SEM. No significant differences; Student’s t-test. (D) Single tiled optical slice of a whd¹ mosaic ovariole with GFP-negative germline cysts (arrows). GFP (green) labels wildtype cells (green). DAPI (white) labels nuclei. Scale bar: 20 μm. (E) Frequencies of GFP-negative germline cysts showing delayed egg chamber growth in FRT42D control and whd¹ mosaic ovarioles. The numbers of egg chambers with GFP-negative germline cysts analyzed are indicated above bars. No significant differences; Fisher’s exact test. (F) Frequencies of FRT42D control and whd¹ mosaic ovarioles with vitellogenic egg chambers containing GFP-negative germline cysts. The numbers of germline-mosaic ovarioles analyzed are indicated above bars. No significant differences; Chi-square test. (G) Single tiled optical slice of a CPT2^raeg mosaic ovariole with one copy of Pfk⁰⁶³³⁹ in the background showing normal RFP-negative germline cysts (arrows). RFP (green) labels wildtype cells. DAPI (white) labels nuclei. Scale bar: 20 μm. (H) Frequency of phenotypically normal CPT2^raeg germline mosaic ovarioles with or without a copy of Pfk⁰⁶³³⁹ in the background. The numbers of germline-mosaic ovarioles analyzed are indicated above bars. No significant differences; Fisher’s exact test. (I) Single optical slice of whd¹ mosaic egg chambers containing GFP-negative follicle cell clones (outlined). GFP (white) labels wildtype cells. Scale bar: 20 μm. (J) Percentage of GFP-negative follicle cells covering stage 4 to 6 egg chambers in whd¹ or Mtpα^KO mosaic females compared to FRT controls. The numbers of follicle cell mosaic egg chambers analyzed are indicated above bars. Data are shown as mean±SEM. *P <0.05; Student’s t-test. (K and L) Single optical slices of CPT2^raeg mosaic egg chambers containing RFP-negative follicle cells (outlined) in females without (K) or with (L) one copy of Pfk⁰⁶³³⁹ in the background. RFP (white) labels wildtype cells. Scale bar: 20 μm. (M) Percentage of RFP-negative follicle cells covering stage 4 to 6 egg chambers in mosaic females of genotypes as in (K and L). The numbers of follicle cell mosaic egg chambers analyzed are indicated above bars. Data are shown as mean±SEM. No significant differences; Student’s t-test. (See Supplementary Data File 1 for raw data.)

Quantitative RT-PCR and genomic PCR analysis

For quantitative RT-PCR analysis, 0- to 2-day-old females were incubated at 29°C with males on standard medium supplemented with dry yeast for 5 days and then yeast paste for 2 days before dissection. Ten pairs of ovaries were dissected and incubated in RNAlater Stabilization Solution (Thermo Fisher Scientific; #AM7021) for 10 min. After RNAlater removal, 250 μl lysis buffer from the RNAqueous-4PCR Total RNA Isolation Kit (Thermo Fisher Scientific; #AM1914) were added, samples were homogenized using a motorized pestle, and RNA extraction proceeded according to the manufacturer’s instructions. Complementary DNA was synthesized from 1 μg total RNA using oligo (dT) primers and SuperScript II Reverse Transcriptase (Thermo Fisher Scientific; #18090010) according to the manufacturer’s instructions. Complementary DNA was diluted 1:500 and amplified in a StepOnePlus Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) through a 40-cycle reaction following PowerUp SYBER Green Master Mix (Applied Biosystems; #A25742) instructions (95°C for 30s, 52°C for 30s and 72°C for 60s) using previously described primers for: Hexokinase A (5′-AGTGTGTACCGCTTCCATCC-3′ and 5′-CAGCATCAGATCGAAGGTGA-3′) (Carvalho-Santos et al. 2020); Pfk (5’-GAACTTCGATCAACGCATCC-3’ and 5’-ATACCCTCCTCCTCGTAGGC-3’) (Carvalho-Santos et al. 2020), and RpL32 (also known as RP-49) (5′-CAGTCGGATCGATATGCTAAGC-3′ and 5′-AATCTCCTTGCGCTTCTTGG-3′) as a normalization control. Three to four technical replicates were included for each genotype. For CPT2, a similar procedure was followed, except that RNA was extracted from five 1-day-old whole flies and cDNA was diluted 1:1000 and amplified through a 40-cycle reaction (95°C for 3s, 56°C for 3s and 72°C for 20s) in a QuantStudio 7 Flex Real-Time PCR system (Applied Biosystems) using primers 5’- AGCAGCAGTAGCAAGAAGTA-3’ and 5’-CCCTGTTACGTTAAAGCCAT-3’. One to three technical replicates were included for each genotype. Standard deviation was calculated for the relative ΔΔCT quantification method (See Supplemental Data File 1).

For genomic DNA PCR, five to 10 flies were homogenized in 100 μl of lysis buffer [0.5% SDS, 100 mM Tris (pH 7.4), 100 mM NaCl and 100 mM EDTA (pH 8.0)]. After addition of 900 μl of lysis buffer, samples were incubated at 65°C for 30 min, 400 μl of homogenate was transferred to a new tube, 800 μl of ice-cold 1.43 M LiCl/4.3 M KAc solution were added, and samples were incubated on ice for 10 min. PCR was performed following Econotaq Plus Green 2x Master Mix (LGC Biosearch Technologies, Petaluma, CA, USA; #30033-1) instructions using the following primers for: CPT2 (5'-ACGGAAACGTAGTCCTTCGG-3' and 5'-CAAATGTCAAGCCAGCAGCAT-3'); Mtpβ (5’-TTGATCCCCAATAAGCCGTCA-3’ and 5’-GCGAGCACAGCTACCAGAATA-3’); and lacZ (for Pfk⁰⁶³³⁹ allele) (5'-CGCTGATTGAAGCAGAAGCC-3' and 5'-TGATGGACCATTTCGGCACA-3').

RESULTS

Validation and generation of genetic tools for the study of glycolysis and mitochondrial β-oxidation

To dissect the metabolic requirements of the GSC and FSC lineages in the Drosophila ovary, we chose to target genes encoding key enzymes in the glycolytic and mitochondrial β-oxidation pathways. For glycolysis, we focused on the regulatory enzymes Phosphofructokinase (Pfk, rate-limiting enzyme) and Pyruvate kinase (Pyk, which catalyzes the last step of glycolysis to produce pyruvate) (Chandel 2021) (Fig. 1C). Unfortunately, there were issues with several available tools (see Material and Methods for details). For example, the recessive lethal Pfk insertional allele, Pfk⁰⁶³³⁹ (Tennessen et al. 2011) (Fig. 1D), had background mutations that caused loss of GSCs and increased death of region 2A cysts (Supplementary Fig. 1A and B), while published RNAi lines against Pfk and the regulatory enzyme gene Hexokinase A did not cause efficient knockdown in the germline (Supplementary Fig. 1C to D'). We therefore used the CRISPR-Cas9 system to target the active site of Pfk using an available single guide RNA line (sgRNA-NIG-FLY.Pfk) and generate the null Pfk¹ allele (Fig. 1D). Pfk¹ is an indel allele that causes a frameshift and premature stop codon, severely disrupting the Pfk active site (Fig. 1D'). Pfk¹ fails to complement the lethality of Pfk⁰⁶³³⁹, and its lethality is rescued by genomic transgenes carrying a wildtype copy of Pfk (PBac{GV-CH321-87O16} or PBac{CH322-13F12}) (Supplementary Table 1). In addition, we obtained the previously published null Pyk³¹ allele (Heidarian et al. 2024) (Fig. 1E) and confirmed that it fails to complement the lethality of another null allele, Pyk²³ (Heidarian et al. 2024).

For mitochondrial β-oxidation, we focused on proteins required for the transport of fatty acids into the mitochondria, carnitine palmitoyltransferase 1 [CPT1, the rate-limiting enzyme encoded by withered (whd)] and CPT2, and the β-oxidation reaction enzymes mitochondrial trifunctional protein α subunit (Mtpα) and Mtpβ (Houten et al. 2016) (Fig. 1C). We validated the previously characterized null whd¹, Mtpα^KO, and Mtpβ^KO alleles (Strub et al. 2008; Kishita et al. 2012) by complementation tests and/or genomic PCR and showed that the CPT2^f02667 insertional allele drastically reduces CPT2 expression (Fig. 1F to I and Supplementary Fig. 1E). We also generated the deletion allele CPT2^raeg (Fig. 1G) by targeting opposite ends of the coding region with two guide RNAs through the CRISPR-Cas9 system.

Glycolysis is not cell autonomously required for GSC maintenance

We first asked if glycolysis is required in GSCs for their survival and/or self-renewal by generating GFP-negative GSCs homozygous for Pfk¹ or Pyk³¹ (or wildtype alleles, as controls) in the context of GFP-positive wildtype tissue using FLP-FRT-based genetic mosaic analysis as previously described (Ables and Drummond-Barbosa 2010; Laws et al. 2015; Laws and Drummond-Barbosa 2016) (Fig. 2A to E). We quantified the frequency of GSC loss events based on the percentage of germline-mosaic germaria that had GFP-negative cystoblasts and/or germline cysts but not the GFP-negative GSC that gave rise to them 10 days after clonal induction (Laws and Drummond-Barbosa 2015). The germaria of Pfk¹ and Pyk³¹ mosaic females had a similar frequency of GSC loss events as their respective controls (Fig. 2F), demonstrating that glycolysis is not cell autonomously required for GSC maintenance.

Glycolysis is not cell autonomously required for proliferation or survival of GSCs, cystoblasts, or early germline cysts

The presence of homozygous mutant Pfk¹ and Pyk³¹ cystoblasts and cysts in the germaria of mosaic females (Fig. 2B to E) indicated that glycolysis-defective GSCs can divide; however, it remained unclear if their division rates were altered. We therefore analyzed mosaic germaria in which both GFP-negative and -positive GSCs were present by counting the numbers of GFP-negative and positive cystoblasts and cysts as a measure of the number of recent mitotic divisions undergone by their respective mother GSCs. In control, Pfk¹, and Pyk³¹ mosaic germaria, there was no significant difference between the numbers of GFP-negative compared to -positive cystoblasts and cysts per GSC, other than a slight increase in Pfk¹ cystoblast numbers (Fig. 2G). We conclude not only that GSCs and their early daughters do not require glycolysis to divide at normal rates and form 16-cell cysts, but also that they survive equally well with or without a functional glycolytic pathway.

Pfk is not cell autonomously required for entry into meiosis or oocyte specification

We also tested if glycolysis is required in the germline for establishment of the oocyte fate following 16-cell cyst formation. We initially analyzed Pfk¹ and Pfk⁰⁶³³⁹ mosaic germaria for the presence of newly formed 16-cell cysts in region 2A undergoing pre-meiotic S phase based on incorporation of the thymidine analog 5-ethynyl 2'-deoxyuridine (EdU). GFP-negative (homozygous for Pfk¹ or Pfk⁰⁶³³⁹) and GFP-positive 16-cell cysts (carrying wildtype Pfk) in region 2A incorporated EdU at similar frequencies, just as was the case in control mosaic germaria (Fig. 2H and I and Supplementary Fig. 2A), indicating that glycolysis is not required for entry into pre-meiotic S phase. Pfk⁰⁶³³⁹ homozygous 16-cell cysts in regions 2A and 2B stained positive for γH2Av, a marker for DNA breaks (which occur during meiosis prophase I), in 22 out of 24 germaria (Fig. 2J and J’’), consistent with what we observed in control mosaic germaria (18 out of 19 germaria). In addition, the oocyte marker Orb could be clearly seen in all Pfk⁰⁶³³⁹ homozygous cysts by region 3 of the germarium (Fig. 2K and K'), similar to their FRT controls (Supplementary Fig. 2B) (number of region 3 GFP-negative germline cysts: 25 for Pfk⁰⁶³³⁹ and 22 for FRT42D control). Accordingly, all GFP-negative homozygous Pfk¹ and Pyk³¹ germline cysts in later egg chambers had an oocyte with a typical nucleus (or karyosome) similar to their respective controls (Fig. 2L and M) (31 to 55 germline mosaic ovarioles analyzed per genotype). Therefore, glycolysis is not required for entry into meiosis or oocyte specification.

Glycolysis is required in the germline for normal rates of egg chamber growth and nurse cell chromatin dispersal

Following egg chamber formation, nurse cells become highly polyploid to support rapid egg chamber growth (Spradling 1993), generating a large demand for energy and biomolecules. We therefore wondered if glycolysis might become required in these later stages of germline differentiation. Egg chambers containing homozygous Pfk¹ or Pyk³¹ germline cysts are still able to grow and develop. However, ~22% of those egg chambers had a significant growth delay compared to control egg chambers (Fig. 3A to E), indicating that glycolysis cell autonomously supports high growth rates of germline cysts in developing egg chambers. Surprisingly, Pfk¹ or Pyk³¹ homozygous germline cysts were able to reach vitellogenic stages (Fig. 3F and F'), albeit less frequently than controls—26% of Pfk¹ and 17% of Pyk³¹ germline-mosaic ovarioles had GFP-negative germline cysts in vitellogenic stages compared to 66% and 37% of their respective controls (Fig. 3G). Moreover, unlike their controls, Pfk¹ and Pyk³¹ vitellogenic germline cysts were never observed past stages 8 or 9 of oogenesis. The relatively lower frequency of vitellogenic egg chambers with Pfk¹ or Pyk³¹ germline cysts could be due to their reduced growth rate, although we cannot rule out egg chamber death.

In addition to growing more slowly, Pfk¹ or Pyk³¹ homozygous germline cysts often failed to undergo normal nurse cell chromatin dispersal during stage 5. Specifically, while nearly all control GFP-negative germline cysts disperse their nurse cell polytene chromatin by stage 6 (Fig. 3A inset and Fig. 3H), 67% of Pfk¹ and 93% of Pyk³¹ homozygous germline cysts retained the “five-blob” polytene nurse cell chromatin morphology (Fig. 3C and D insets and Fig. 3H). These results suggest that both the growth and the development of glycolysis-defective germline cysts are delayed, although a more direct requirement of glycolytic products for chromatin dispersal is also possible.

Pfk but not Pyk is required for FSC maintenance and early follicle cell survival

To investigate the requirement for glycolysis during early stages of the somatic FSC lineage, we generated GFP-negative FSCs homozygous for Pfk¹ or Pyk³¹ using genetic mosaic analysis (Fig. 4A to E). The frequency of FSC loss events was quantified as the percentage of follicle cell mosaic germaria that contained GFP-negative daughter follicle cells but had lost the GFP-negative FSC that gave rise to them 10 days after clonal induction (Laws and Drummond-Barbosa 2015). While only ~14% of control GFP-negative FSCs were absent, 23% of Pfk¹ homozygous FSCs had been lost from mosaic germaria (Fig. 4A, B, and F), and the Pfk¹ FSC loss phenotype could be rescued by the genomic transgenes PBac{GV-CH321-87O16} and PBac{CH322-13F12} (Fig. 4C and F). By contrast, Pyk³¹ homozygous FSCs were lost at a similarly low frequency as GFP-negative FSCs in control mosaics (Fig. 4E and F), suggesting a differential requirement for Pfk and Pyk in FSCs.

We then asked if early FSC daughters require Pfk and/or Pyk for various processes. First, we quantified the total number of GFP-negative follicle cells present in germaria that still had the GFP-negative mother FSC (e.g., Fig. 4A, C, D, and E). Pfk¹ mosaic germaria had one-fourth of the number of GFP-negative follicle cells present in control mosaics, and this phenotype was partially rescued by genomic rescue transgenes PBac{GV-CH321-87O16} and PBac{CH322-13F12} (Fig. 4G). In contrast to the Pfk¹ results, Pyk³¹ homozygous and control GFP-negative follicle cells were present at comparable numbers in mosaic germaria (Fig. 4G). Surprisingly, GFP-negative Pfk¹ and Pfk⁰⁶³³⁹ homozygous follicle cells incorporated EdU at a similar frequency as their counterparts in control mosaics (Fig. 4H), indicating that the reduced number of Pfk¹ mutant follicle cells is not due to proliferation defects. On the other hand, 38% of Pfk¹ mosaic germaria contained at least one GFP-negative follicle cell labeled with cleaved Dcp-1, an apoptosis marker (Song et al. 1997), compared to about 5% of control mosaic germaria (Fig. 4D and I). Interestingly, among the relatively rare examples of egg chambers covered by ~50% or more Pfk¹ homozygous follicle cells, 30% showed multiple germline cysts that had been enveloped together (Fig. 4J, K, and M; compare to control mosaic in Supplementary Fig. 2C). This defect was independent of the germline genotype and was rarely observed in control or Pyk³¹ follicle cell mosaics (Fig. 4L and M and Supplementary Fig. 2D). It is likely that these encapsulation defects are a secondary consequence of the death of Pfk mutant follicle cells, although other roles of Pfk are possible.

Altogether, these results indicate that the early stages of the FSC lineage require Pfk for survival but not proliferation, and they also contrast with our findings that glycolysis is not required in GSCs, cystoblasts, or early germline cysts. The differential requirement for Pfk and Pyk in FSCs and their early differentiating daughters might reflect stronger effects of disrupting Pfk on the pentose phosphate pathway, serine/glycine biosynthesis, or other biosynthetic pathways compared to mutation of the more downstream enzyme encoded by Pyk (Chandel 2021).

Follicle cells in growing egg chambers require Pfk and Pyk for survival but only Pfk for proliferation

We next interrogated the requirement for glycolytic genes in later follicle cells after egg chambers leave the germarium. We first quantified the average proportion of GFP-negative follicle cells present in stage 4–6 egg chambers (Fig. 5A to C). In FRT42D and FRT82B control mosaics, GFP-negative follicle cells covered ~50% of the surface of egg chambers (Fig. 5A and C), consistent with the equal contribution from GFP-negative and GFP-positive FSC daughters. By contrast, in both Pfk¹ and, to a lesser extent, Pyk³¹ mosaics, a smaller fraction of follicle cells was GFP-negative (Fig. 5B and C). The frequency of egg chambers with cleaved Dcp-1-positive GFP-negative follicle cells was significantly higher in Pfk¹ and Pyk³¹ mosaics compared to their respective controls (Fig. 5D to F and Supplementary Fig. 2E)—and, in extreme cases, no follicle cells remained around germline cysts (Fig. 5E; compare to control mosaics in Supplementary Fig. 2C and D)—showing that Pfk and Pyk are important for follicle cell survival.

We also measured Edu incorporation in GFP-negative follicle cells (compared to GFP-positive follicle cells) in control, Pfk¹, and Pyk³¹ mosaics. Interestingly, unlike the case for earlier follicle cells in the germarium, a smaller proportion of Pfk¹ homozygous mutant follicle cells relative to controls was EdU-positive (Fig. 5G to I). However, Pyk³¹ mutant follicle cells incorporated EdU at similar rates as controls (Fig. 5I), potentially explaining the more severe reduction in the numbers of Pfk¹ relative to Pyk³¹ follicle cells (Fig. 5C). In agreement with a Pfk requirement for follicle cell proliferation and survival, the frequency of Pfk¹ follicle cell clones found in stage 7 and later egg chambers was drastically lower than that of FRT42D control clones (Fig. 5J). Finally, we also observed that among those individual egg chambers that contained 50% or more Pfk¹ or Pyk³¹ follicle cells surrounding a wildtype germline cyst, 20% or 13% were growth delayed, respectively (Fig. 5E and K). This observation is consistent with previous studies showing that follicle cells regulate the growth of germline cysts (Maines et al. 2004; LaFever et al. 2010; Laws and Drummond-Barbosa 2016). These results reveal changes in the metabolic requirements of follicle cells between the time they contribute to egg chamber formation in the germarium, when Pfk but not Pyk is needed for their survival, and later, during egg chamber growth, when follicle cells require Pfk for proliferation and both Pfk and Pyk for survival.

Mitochondrial β-oxidation is not required in the GSC or FSC lineages even when glycolysis is reduced

To determine if mitochondrial β-oxidation is required in GSC or FSC lineages, we analyzed whd, CPT2, Mtpα, and Mtpβ mutant mosaic females using similar assays as described above. Surprisingly, mosaic germaria containing homozygous mutant germ cells lacking functional mitochondrial β-oxidation genes did not display a consistent increase in GSC loss (Fig. 6A and B and Supplementary Fig. 3A; compare to Fig. 2B). To address the possibility that GSCs might be switching between glycolysis and mitochondrial β-oxidation depending on which pathway is inactivated, we asked if GSCs lacking mitochondrial β-oxidation might be lost under reduced glycolysis conditions. However, there was no increase in CPT2^f02667 GSC loss in mosaic females on a 1% sugar diet as opposed to the standard 5% sugar diet (Supplementary Fig. 3B); in CPT2^raeg GSCs in mosaic females carrying one copy of Pfk⁰⁶³³⁹ in the background (i.e., ~50% dosage of the rate limiting glycolytic enzyme gene Pfk) (Fig. 6B); or in Pfk¹ Mtpβ double null mutant GSCs, where both glycolysis and mitochondrial β-oxidation are lacking (Fig. 6B). These findings rule out the possibility of a glycolysis-β-oxidation switch. Mitochondrial β-oxidation genes were not required in cystoblasts or early dividing cysts for proliferation or survival (Fig. 6C and Supplementary Fig. 3C), or in later germline cysts for follicle growth or vitellogenesis (Fig. 6D to F and Supplementary Fig. 3D to H), even when glycolysis was reduced (Fig. 6C, G, and H). Similarly, we found no requirements for mitochondrial β-oxidation genes in the FSC lineage (Fig. 6I to M and Supplementary Fig. 3I). Based on these results, we conclude that neither the GSC nor the FSC lineage require mitochondrial β-oxidation during Drosophila oogenesis.

DISCUSSION

Carbohydrates and lipids are widely considered primary sources of cellular energy (Alberts et al. 2002; Liu et al. 2025) and their breakdown through glycolysis and β-oxidation, respectively, is required in many tissue stem cells (Chandel et al. 2016; Jackson and Finley 2024). The Drosophila ovary has been a leading model system for the study of fundamental aspects of stem cell biology owing to its powerful genetics and two populations of well characterized stem cells that support oogenesis (Lin and Spradling 1993; Spradling 1993; Margolis and Spradling 1995). FSCs are more akin to typical tissue stem cells, giving rise to an epithelium of somatic follicle cells, while GSCs give rise to germline cysts that develop in linear order to ultimately generate oocytes—highly specialized cells responsible for the perpetuation of the species (Lin and Spradling 1993; Spradling 1993; Margolis and Spradling 1995). Although previous studies have proposed that glycolysis and mitochondrial β-oxidation have roles in Drosophila female GSCs (Amartuvshin et al. 2020; Rojas-Ríos et al. 2024), direct genetic analyses targeting critical regulatory enzymes along the GSC and FSC lineages have been lacking. Through the power of genetic mosaic analysis using null alleles for genes encoding two key glycolytic enzymes—Pfk, the rate-limiting enzyme, and Pyk, which catalyzes the last, irreversible reaction of glycolysis—we provide strong genetic evidence that GSCs and their early mitotically dividing daughters do not require glycolysis for their normal survival or proliferation. Instead, glycolysis becomes necessary in the germline during later stages for egg chamber growth and normal dispersal of nurse cell chromatin (Fig. 7). In stark contrast, FSCs and their early progeny have a strong requirement for Pfk for their survival, while later, during egg chamber growth, follicle cells require Pfk both for survival and proliferation. Interestingly, Pyk is only required in later follicle cells for survival, suggesting that the additional requirements for Pfk likely reflect roles independent of pyruvate production (Fig. 7). Surprisingly, neither lineage required mitochondrial β-oxidation at any stage, even when glycolysis was impaired, suggesting that they do not switch between these two pathways. These findings provide an important step towards a deep understanding of the metabolic requirements of these stem cell lineages, highlight that tissue stem cells can have quite diverse metabolic strategies even within a common tissue, and open several new questions for future investigation illustrated below.

Fig. 7.

Summary of requirements for glycolytic genes in Drosophila ovarian stem cell lineages. In the germline (blue), glycolysis is not required in GSCs or their early daughters but becomes required in later germline cysts for egg chamber growth and nurse cell chromatin dispersal. In the somatic follicle cell lineage (purple), Pfk but not Pyk is required for FSC maintenance and survival of their early progeny, while in later follicle cells both Pfk and Pyk are required for follicle cell survival and Pfk becomes required for follicle cell proliferation. Loss of Pfk or Pyk either directly or indirectly impair the growth of underlying germline cysts. β-oxidation is not required in either lineage.

GSCs have unusual metabolic requirements

Our findings reveal that female GSCs are quite uncommon in that they do not intrinsically require glycolysis or mitochondrial β-oxidation and have no defects even when both metabolic pathways are simultaneously blocked. These findings are particularly surprising considering that GSCs proliferate often (Margolis and Spradling 1995; Drummond-Barbosa and Spradling 2001; Morris and Spradling 2011) and glucose metabolism is typically important in cells that proliferate rapidly both for ATP production and for production of sugar backbone and reducing equivalents required for nucleotide synthesis (Chandel 2021). Accordingly, most tissue stem cells studied to date require either lactate production or mitochondrial pyruvate metabolism downstream of glycolysis for regulation of their stemness and proliferative ability (Chandel et al. 2016; Jackson and Finley 2024), and high levels of glycolysis are considered critical for the pluripotency of embryonic stem cells and induced pluripotent stem cells (Tsogtbaatar et al. 2020). GSCs use ATP, reducing equivalents, and macromolecules as they proliferate independent of intrinsic glycolysis (or β-oxidation), indicating that they either cooperate metabolically with neighboring somatic cells that make up their niche or more distant tissues, use distinct sources of energy and building blocks, or a combination of these strategies. In an interesting parallel to the lack of glycolysis requirement in Drosophila GSCs—which give rise to oocytes capable of generating a whole new organism after fertilization—pharmacological inhibition of Pfk function during preimplantation mouse embryonic development does not affect the progression of the 1-cell stage to the blastocyst stage (Chi et al. 2020), tempting us to speculate that this unusual metabolic state might possibly have a connection with the ultimate potency of these cells.

Our results disagree with a key conclusion from a previous study, which reported that glycolysis was required for Drosophila female GSC maintenance (under regulation of piRNAs) (Rojas-Ríos et al. 2024). The main basis for this conclusion relied on the analysis of the function of genes encoding enolase and aldolase. Aldolase catalyzes the conversion of fructose-1,6-bisphosphate to glyceraldehyde 3-phosphate and dihydroxyacetone phosphate, while enolase converts 2-phosphoglycerate to phosphoenolpyruvate; both reactions are reversible and non-regulatory. Interestingly, both aldolase and enolase are ubiquitous and highly expressed proteins shown to have important roles outside of glycolysis in other organisms, although this has not been explored in Drosophila. Enolase has glycolysis-independent roles related to microtubule interactions, gene expression, mitochondrial membrane potential, and stress responses, among others (Didiasova et al. 2019; Bian et al. 2022; Horvat et al. 2024), while aldolase has moonlighting functions in modulating the actin cytoskeleton dynamics, translation, cell signaling, and DNA repair (Yu et al. 2025). Our genetic evidence—based on targeting key regulatory enzymes—argues strongly against a role for glycolysis in GSC maintenance or proliferation, thereby indicating that aldolase and enolase affect GSCs independently of their roles in glycolysis. On a related note, a recent study based on genetic mosaic analysis found that GSCs lacking the estrogen-related receptor have reduced levels of Enolase and are lost at higher rates (Zike et al. 2025), suggesting that ERR regulates GSC maintenance at least in part through the glycolysis-independent role of Enolase. Interestingly, an earlier study found that ATP synthase complex components are required for differentiation of early GSC daughters past the four-cell cyst stage independent of their ATP synthesis role (Teixeira et al. 2015), providing another example of non-canonical roles for metabolic enzymes in early female Drosophila germ cells. Future studies should explore the basis for the roles of aldolase and enolase in GSC maintenance and of ATP synthase in early cyst differentiation.

Pyruvate requirement for oocyte growth in Drosophila and mammals

Our findings that glycolysis becomes required in the germline during egg chamber growth align well with previous mRNA in situ and single cell RNA sequencing data showing that the expression levels of glycolytic enzymes increase in more differentiated germ cells compared to GSCs and their early dividing daughter cells (Carvalho-Santos et al. 2020; Slaidina et al. 2020; Martin et al. 2022; Pang et al. 2023). In vitro studies show an increase in pyruvate consumption in growing mouse oocytes (Harris et al. 2009), suggesting that the requirement for glycolysis during oocyte growth might be evolutionarily conserved. One distinction is that, in Drosophila, the nurse cells remain attached to oocytes during the growth phase and the germline cysts intrinsically require Pfk and Pyk (which are critical for production of pyruvate), while in mammals the germline cysts break down long before the follicle growth phase (Spradling 2024), and the oocyte-supporting follicular cells instead produce pyruvate, which is then transferred to the oocyte (Gu et al. 2015). Drosophila follicle cells might transport additional pyruvate into germline cysts to help support their rapid growth. In agreement with that possibility, we found that wildtype germline surrounded by large numbers of Pfk mutant follicle cells had delayed growth, although this phenotype could also reflect more indirect mechanisms.

Transport of metabolites between neighboring cells can occur through metabolite transporters or through gap junctions. For example, monocarboxylate transporters transfer pyruvate and lactate from glia to neurons in Drosophila and mammals (Jha and Morrison 2018; González-Gutiérrez et al. 2020; Liu et al. 2017). In Caenorhabditis elegans, strong genetic evidence demonstrates that malonyl-CoA is transported from somatic gonadal cells to the germline through gap junction to support fatty acid synthesis and embryonic development (Starich et al. 2020), and a recent study proposed that gap junctions transport metabolites between follicle cells and the oocyte in Drosophila (Vachias et al. 2025). Additional experiments targeting metabolite transporters and gap junctions will be needed to determine whether follicle cells transfer pyruvate to the germline.

The requirement for pyruvate production within germline cysts to support egg chamber growth raises additional questions for further investigation, as pyruvate can have multiple fates within a cell. Pyruvate can be reduced to lactate by NADH produced during glycolysis to regenerate NAD⁺ to allow for rapid rounds of glycolysis when glycolytic intermediates are in high demand for the synthesis of biomolecules such as nucleotides, lipids, amino acids, and NADPH (Chandel 2021). Pyruvate can also be transported into the mitochondria and converted to acetyl-CoA, which fuels the tricarboxylic acid (TCA) cycle to generate reducing equivalents NADH and FADH₂ for production of large amounts of ATP through oxidative phosphorylation (Chandel 2021). Alternatively, when TCA cycle intermediates are needed as precursors for the biosynthesis of various biomolecules, pyruvate is converted into oxaloacetate to help replenish these depleted intermediates (Inigo et al. 2021). Understanding the contributions of pyruvate, glycolytic intermediates, and TCA cycle intermediates produced within germ cells to egg chamber growth will require the genetic dissection of the requirements for the relevant enzymes during oogenesis.

Other fuel sources driving germline growth

Our findings indicate that the growing germline meets its high metabolic needs through additional mechanisms beyond β-oxidation (which is not required) and glycolysis (the disruption of which slows but does not arrest the growth of egg chambers, at least through stage 9). One possibility is that pyruvate produced in follicle cells might support germline cyst growth, akin to the cooperation between glia and neurons that occurs in Drosophila and mammals (Backer and Kadow 2022). Indeed, we found that a subset of egg chambers with Pfk or Pyk mutant follicle cells have delayed growth—although the death of these follicle cells confounds the interpretation of these results. It is also conceivable that amino acids might serve as a key energy source fueling the TCA cycle, considering that glutaminolysis drives the production of bone mass and of hair follicles in mice (Yu et al. 2019; Kim et al. 2020), and rapidly growing cancers also rely on glutamine metabolism (Jin et al. 2023). Consistent with this possibility, a recent study in Drosophila showed that mutation of a putative amino acid transporter in follicle cells or disruption of gap junctions reduced the growth of underlying germline cysts (Vachias et al. 2025). It is also possible that other metabolites—such as ketone bodies, which can serve as alternate sources of acetyl-CoA (Li et al. 2025)—might be released from other tissues and transported into germ cells to help support their growth.

Glycolysis and chromatin regulation

Relatively little is known about what controls whether cells become polytene or polyploid following endoreplication, including in the case of Drosophila nurse cells, which undergo a developmental transition from polyteny to chromatin dispersal during stage 5 of oogenesis. Mutations in several genes, including those encoding heterogeneous ribonucleoproteins, splicing factors, a deubiquitinase component, and a mediator of insulin signaling, disrupt nurse cell chromatin dispersal (Klusza and Deng 2010; Wang et al. 2010; Klusza et al. 2013), but the specific mechanisms remain largely unclear. Our finding that blocking glycolysis also disrupts this transition adds a new clue. Various glucose-derived metabolites (e.g., acetyl-CoA, succinyl-CoA, or lactate) can be used for posttranslational modifications of proteins, including histones, while other metabolites (e.g., α-ketoglutarate) serve as co-factors for histone-modifying enzymes (Ryall et al. 2015; Diskin et al. 2021). Histone modifications—and consequent changes in chromatin structure—are associated with developmental transitions during embryogenesis across organisms (Liu and Schneider 2025). Changes in histone methylation and chromatin silencing also occur during Drosophila oogenesis (Deluca et al. 2020), and distinct patterns of histone lactylation were recently described during mouse oocyte maturation and Drosophila germ cell differentiation (Yang et al. 2024; Hayashi et al. 2025). It will be very interesting to investigate whether the mechanism linking glycolysis and the transition of nurse cells from polyteny to dispersed chromatin involves changes in key metabolites and/or specific histone modifications.

Glycolysis and oocyte meiosis

We did not find any obvious indication of early meiotic defects in Drosophila Pfk or Pyk mutant oocytes, although we cannot rule out more subtle defects during their prophase I arrest. Mouse oocyte-specific knockout of Pdha1 (which encodes a subunit of a key enzyme in mitochondrial pyruvate metabolism) leads to severe meiotic maturation defects (Johnson et al. 2007). However, given that none of Pfk or Pyk mutant Drosophila oocytes had progressed past stage 9 of oogenesis in our analyses, we could not determine if glycolysis might be required for oocyte meiotic maturation, which occurs during stage 13 (Von Stetina et al. 2008). Therefore, the role of glycolysis in Drosophila meiosis remains an open question.

Distinct requirements for Pfk and Pyk in the FSC lineage

Pfk and Pyk are key enzymes that catalyze irreversible, regulatory steps of glycolysis. Pfk catalyzes the third and rate-limiting reaction of glycolysis (the conversion of fructose 6-phosphate into fructose 1,6-bisphosphate), while Pyk converts phosphoenolpyruvate into pyruvate, the final product of glycolysis (Chandel 2021). Our findings that Pfk but not Pyk is required for survival and proliferation in early stages of the FSC lineage and for proliferation in later follicle cells outside the germarium indicate that cell-autonomous pyruvate production is not required for those processes. Glycolytic intermediates downstream of Pfk and upstream of Pyk have important anabolic roles. For example, dihydroxyacetone phosphate can be converted to glycerol 3-phosphate, which is important for the biosynthesis of structural phospholipids and triacylglycerols, while 3-phosphoglycerate can be used for production of sphingolipids and amino acids serine, cysteine, and glycine (Kierans and Taylor 2024). Some glycolytic intermediates, including fructose-1,6-bisphosphate, dihydroxyacetone phosphate, and phosphoenolpyruvate have also been proposed to have diverse signaling roles in mammalian cells (Kierans and Taylor 2024). Similarly, it is likely that one or more of these glycolytic metabolites are important for various processes in the FSC lineage, although we cannot rule out potential glycolysis-independent, non-canonical roles of Pfk (Bian et al. 2022). Future studies should address these various possibilities.

Further questions

Beyond the questions raised above—What other fuel sources support early stages of the GSC lineage and contribute to the growth and development of later stages? Why are aldolase and enolase required in GSCs (Rojas-Ríos et al. 2024) while glycolysis is not? What tissues metabolically cooperate with the germline? How is glucose metabolism linked to nurse cell chromatin reorganization? What is the role of metabolism (if any) in meiosis? Why is Pfk but not Pyk required for certain processes in the FSC lineage—many more open questions remain. For example, a recent study reported that RNAi-based knockdown of pentose phosphase pathway genes Hexokinase A, Glucose-6-phosphate dehydrogenase, or Phosphogluconate dehydrogenase causes a reduction in the number of eggs laid (Carvalho-Santos et al. 2020); however, this study did not include an in-depth analysis of oogenesis. Although we found no evidence that mitochondrial β-oxidation is required in either of the Drosophila ovarian stem cell lineages in our experiments, it is possible that more drastic changes in physiology might cause a shift in their metabolic needs. Along similar lines, it would be interesting to investigate if and how physiological signaling pathways known to regulate oogenesis, such as AMPK, TOR, ecdysone, and insulin signaling (Drummond-Barbosa 2019) interact with metabolic pathways in the ovary. In vivo genetic studies should also explore if there are additional examples of tissue stem cells besides Drosophila GSCs that do not require glycolysis or if, perhaps, this might be a peculiarity of female germ cell precursors more generally.

DATA AVAILABILITY

Many Drosophila strains used in this study are publicly available. New strains and plasmids generated during this study are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, tables, and supplemental data file.