Temporal remodeling of the cell cycle accompanies differentiation in the Drosophila germline

Abstract

Development of multicellular organisms relies upon the coordinated regulation of cellular differentiation and proliferation. Growing evidence suggests that some molecular regulatory pathways associated with the cell cycle machinery also dictate cell fate; however, it remains largely unclear how the cell cycle is remodeled in concert with cell differentiation. During Drosophila oogenesis, mature oocytes are created through a series of precisely controlled division and differentiation steps, originating from a single tissue-specific stem cell. Further, germline stem cells (GSCs) and their differentiating progeny remain in a predominantly linear arrangement as oogenesis proceeds. The ability to visualize the stepwise events of differentiation within the context of a single tissue make the Drosophila ovary an exceptional model for study of cell cycle remodeling. To describe how the cell cycle is remodeled in germ cells as they differentiate in situ, we used the Drosophila Fluorescence Ubiquitin-based Cell Cycle Indicator (Fly-FUCCI) system, in which degradable versions of GFP::E2f1 and RFP::CycB fluorescently label cells in each phase of the cell cycle. We found that the lengths of the G1, S, and G2 phases of the cell cycle change dramatically over the course of differentiation, and identified the 4/8-cell cyst as a key developmental transition state in which cells prepare for specialized cell cycles. Our data suggest that the transcriptional activator E2f1, which controls the transition from G1 to S phase, is a key regulator of mitotic divisions in the early germline. Our data support the model that E2f1 is necessary for proper GSC proliferation, self-renewal, and daughter cell development. In contrast, while E2f1 degradation by the Cullin 4 (Cul4)-containing ubiquitin E3 ligase (CRL4) is essential for developmental transitions in the early germline, our data do not support a role for E2f1 degradation as a mechanism to limit GSC proliferation or self-renewal. Taken together, these findings provide further insight into the regulation of cell proliferation and the acquisition of differentiated cell fate, with broad implications across developing tissues.

1. Introduction

Multicellular organisms develop via the coordinated regulation of cellular differentiation and proliferation. Molecular control of the cell cycle is therefore intimately associated with developmental differentiation cues (Bunce and Capel, 2016; Ruijtenberg and van den Heuvel, 2016; Soufi and Dalton, 2016). Most eukaryotic cells achieve replication and subsequent orderly distribution of genetic material to new cells via discrete cell cycle phases (G1/S/G2/M). Yet studies of pluripotent and multipotent stem cells, which have the dual responsibility of mitotically dividing while maintaining an undifferentiated identity, have suggested that loss or shortening of G1 enables cells to be refractory to differentiation (Boward et al., 2016; Julian et al., 2016). Further, temporal remodeling of the cell cycle accompanies the stepwise acquisition of cell fate and function in many tissues, including the early Drosophila melanogaster syncytial embryo (Duronio, 2012; Farrell and O’Farrell, 2014) and the C. elegans germline (Hansen and Schedl, 2013; Kimble, 2011). Despite the wealth of knowledge on the molecular control of the cell cycle, and the implications of cell cycle research on human disease and tissue homeostasis, the mechanisms through which the cell cycle is coordinated with the acquisition of cell fate remain unclear (Boward et al., 2016; Julian et al., 2016; Ruijtenberg and van den Heuvel, 2016; Soufi and Dalton, 2016).

Drosophila oogenesis provides an excellent model to study how cell cycle control is regulated in concert with differentiation. The adult Drosophila ovary is composed of 14–16 ovarioles: strings of progressively developing follicles, each containing a single oocyte surrounded by somatic follicle cells (King, 1970; Spradling, 1993). Oogenesis is fueled by the activity of germline stem cells (GSCs), which lie in a structure called the germarium at the anterior tip of each ovariole (Fig. 1A–B) (Spradling, 1993; Xie, 2013). A GSC divides asymmetrically to generate a new GSC and a cystoblast, which remains connected to the GSC until G2 of the next cycle (de Cuevas and Spradling, 1998). Following abscission, the cystoblast undergoes four synchronized rounds of mitotic division with incomplete cytokinesis to form inter-connected 16-cell cysts (Fig. 1C) (de Cuevas et al., 1997; Ong and Tan, 2010; Spradling, 1993). Cyst mitotic divisions are rapid, polarized, and uncoupled from cell growth, such that individual cyst cells (cystocytes) divide in a stereotypical pattern with a progressive reduction in cell size (King, 1970; Lilly et al., 2000; Spradling, 1993). Cystocytes are held together by stable actin-rich ring canals, which function as intercellular bridges between cells (Robinson and Cooley, 1996). Following the fourth mitotic division, all 16 cystocytes enter premeiotic S phase; however, only one cell ultimately differentiates as the oocyte, while the rest differentiate into nurse cells (Fig. 1A) (Spradling, 1993). Nurse cells then transition into an endocycle: a variant cell cycle composed of repeated rounds of synthesis and gap phases (Edgar et al., 2014). At the posterior of the germarium, cysts are surrounded by follicle cells and subsequently pinch away from the germarium to form follicles, which support the continued growth and development of the oocyte (Fig. 1A).

Fig. 1. Drosophila ovarian germline stem cells give rise to oocytes and nurse cells.

(A) Germline stem cells (GSCs; pink) are anchored to a niche (composed of cap cells and terminal filament cells) in a structure called the germarium at the anterior tip of each ovariole. Early germ cells are characterized by the presence of the fusome (red), which extends as germ cells divide. Escort cells (blue) signal to germ cells to promote differentiation. Follicle stem cells (dark green) divide to form follicle cells (fc, green), which surround the 16-cell germline cyst, giving rise to a follicle that leaves the germarium. (B) A representative nos-Gal4 > GFP::αTubulin germarium labeled with anti-GFP (green) and DAPI (blue) demonstrates that nos-Gal4 drives robustly and specifically in the early germline. Dotted lines demarcate GSCs (white) and cystoblasts (yellow); solid lines demarcate differentiating cysts. Scale bar, 5 μm. (C) GSCs divide to form daughter cells (cystoblasts, CB), which divide mitotically four additional times (M1–M4) to form 16-cell germline cysts. (D) Morphogenesis of the fusome is coordinated with the GSC cell cycle. Following asymmetric division of the GSC, the GSC and pre-cystoblast (pCB) remain connected as both enter S phase of the next cell cycle. A small plug of fusome material forms in the ring canal between the two cells, and grows anteriorly to meet the GSC fusome (also called a spectrosome) as S phase proceeds. Abscission occurs in late G2, separating the GSC and cystoblast. (E) The Fly-FUCCI system uses fluorescently tagged fragments of E2f1 and CycB to monitor the progress of cells through the cell cycle. The APC/C targets RFP::CycB for degradation, resulting in cells expressing GFP::E2f1 only (green), whereas CRL4 targets GFP::E2f1, resulting in cells expressing RFP::CycB only (red). Cells with low APC/C and CRL4 activity express both reporters (yellow).

Central to early germ cell proliferation and differentiation is an unusual germ cell-specific organelle called the fusome (Fig. 1A, D). This vesicle-rich organelle is thought to help coordinate spindle orientation and cell cycle synchrony during early germ cell mitotic divisions (de Cuevas et al., 1997; McKearin, 1997). Fusomes contain a variety of proteins, including the adducin-like Hu li tai shao (Hts), that are critical not only for morphogenesis of the fusome, but also for cyst formation and oocyte specification (de Cuevas et al., 1997; Lighthouse et al., 2008). Cystoblasts contain a spherical fusome that grows and branches during cyst formation, extending through the ring canals into each cystocyte (Ong and Tan, 2010). The fusome is also prominent in GSCs (frequently referred to as a spectrosome), where it is invariantly located at the anterior margin of the GSC closest to the adjacent cap cells (Fig. 1A, D) (Lin et al., 1994). Because the morphology of the fusome changes synchronously with the phases of the GSC cell cycle, quantification of fusome morphologies is a valuable method for phase identification in GSCs (Ables and Drummond-Barbosa, 2013; de Cuevas and Spradling, 1998; Hsu et al., 2008; Kao et al., 2015).

While the molecular regulation of proliferation and self-renewal has been extensively characterized in GSCs, much less is known about how the cell cycle is remodeled from the cystoblast to the 16-cell cyst to accommodate differentiation, incomplete cytokinesis, and the initiation of specialized cell cycles (e.g., meiosis or endocycle). We used the Drosophila Fluorescence Ubiquitin-based Cell Cycle Indicator system (Fly-FUCCI) (Zielke et al., 2014) to ask whether and how the cell cycle is remodeled during the course of germline differentiation. Cell cycle progression depends on precisely controlled production and destruction of many proteins, including the Cyclins and the E2f family of transcription factors (Teixeira and Reed, 2013; Thurlings and de Bruin, 2016). For example, E2f1 is up-regulated during G1 to promote the expression of genes required for DNA synthesis and destroyed by the Cullin4 (Cul4)-containing RING E3 ubiquitin ligase (CRL4) to promote exit from S phase (Thurlings and de Bruin, 2016). Similarly, the transition from G2 to M is controlled by the Cyclin B (CycB)/Cdk1 complex, which is negatively regulated by the action of the anaphase-promoting complex/cyclosome (APC/C) (Teixeira and Reed, 2013). The Fly-FUCCI system uses an E2f1 fragment containing the CRL4-targeted “PIP-box” destruction degron fused to Green Fluorescent Protein (GFP) to mark G2, M, and G1 phases, and a fragment of Cyclin B (CycB) containing the APC/C-targeted destruction box (D box) fused to Red Fluorescent Protein (RFP) to mark S, G2, and M phases (Fig. 1E) (Zielke et al., 2014). Since these nonfunctional proteins are targeted for destruction in the same manner as the endogenous proteins, they serve as useful markers of cell cycle phase for cells in situ.

In this study, we demonstrate that the Fly-FUCCI system labels cell cycle phases in most germ cells in situ; however, the system must be used in combination with other known cell cycle markers for most accurate cell cycle assessment. We find that the lengths of the G1, S, and G2/M phases of the cell cycle change dramatically over the course of differentiation. In contrast to GSCs and their immediate daughters, germ cells in the second and third mitotic divisions undergo a progressive increase in the length of M and S phases. By the 8-cell stage, the majority of germ cells express neither Fly-FUCCI reporter, suggesting that the activity of CRL4 and APC/C are highest at this stage. Additionally, we identify E2f1 and Cul4 as key regulators of the cell cycle in the early germline. E2f1 is necessary for proper GSC proliferation and self-renewal, and development of GSC progeny. In contrast, while Cul4 is essential for developmental transitions in the early germline, it is not required for GSC proliferation or self-renewal, suggesting that E2f1 degradation does not limit GSC proliferation or self-renewal. Taken together, these data suggest that the 4/8-cell cyst is a key developmental transition state in which germ cells prepare for specialized cell cycles. We speculate that alterations in the length of gap phases may provide critical windows for germ cell differentiation.

2. Materials and methods

2.1. Drosophila strains and culture

Flies were maintained at 22–25 °C in standard cornmeal/molasses/yeast/agar medium (NutriFly MF; Genesee Scientific) supplemented with yeast. To facilitate identification of GSCs, cystoblasts, and cysts, the protein-trap allele vsg^CA07004 (referred to as vsg-GFP) (Buszczak et al., 2007) was recombined with the Fly-FUCCI transgene UASp-GFP-E2f1_1–230^#26 UASp-mRFP1-CycB_1–266^#81 (Bloomington #55100 and #55101) (Zielke et al., 2014). For cell cycle analyses in the germline, the resulting FUCCI; vsg-GFP was crossed with the germline-specific drivers nosGAL4::VP16 (III) (referred to as nos-Gal4) (Rørth, 1998; Van Doren et al., 1998) or GAL4-nos. NGT (40) (referred to as nos. NGT-Gal4) (Barrett et al., 1997). Driver expression in the early germline was confirmed using UASp-GFPS65C-αTub84B (Grieder et al., 2000). Additional analyses utilized the Fly-FUCCI transgene UASp-GFP-E2f1_1–230^#64 UASp-mRFP1-NLS-CycB_1–266^#5 (Bloomington #55111), in which RFP::CycB is retained in the nucleus (Zielke et al., 2014). To generate E2f1 degradation-resistant GSCs, UASp-GFP-E2f1. PIP-3A^#10 (non-degradable mutant) or UASp-GFP-E2f1. WT^#4 (control) flies (Davidson and Duronio, 2012) were crossed with nos-Gal4. Flies were collected one to two days after eclosion and maintained on either standard medium supplemented with wet yeast paste (“nutrient-rich diet”) or on molasses agar (“nutrient-poor diet”) (Drummond-Barbosa and Spradling, 2001) for four or 14 days prior to ovary dissection.

2.2. Genetic mosaic generation and stem cell analyses

For genetic mosaic analyses using flippase (FLP)/FLP recognition target (FRT), the following alleles containing locus-appropriate FRT chromosomes were obtained: E2f1⁷²⁹ (Ambrus et al., 2007; Duronio et al., 1995), Cul4^11L (Hu et al., 2008), and Cul4^KG02900 (Lee et al., 2010; Lin et al., 2009). Other genetic tools are described in FlyBase (Ashburner and Drysdale, 1994). Genetic mosaics were generated by FLP/FRT-mediated recombination in 2–3-day old females carrying a mutant allele in trans to a wildtype allele (linked to a Ubi-GFP marker) on homologous FRT arms, and a hs-FLP transgene, as described (Laws and Drummond-Barbosa, 2015). Flies were heat shocked at 37 °C two times per day for 2 days (E2f1 mutants) or 3 days (Cul4 mutants), and incubated at 25 °C for 6–8 days with daily transfers to freshly yeasted vials (standard media, initially supplemented with dry yeast, then supplemented with wet yeast paste on the last 2 days prior to dissection). Cul4^11L mosaic germaria were analyzed at 6 days after clone induction due to severe disruption of cyst development. Wild-type alleles were used for generation of control mosaics. GSCs were identified based on the juxtaposition of their fusomes to the junction with adjacent cap cells (de Cuevas and Spradling, 1998). Stem cell loss was measured as the percentage of total mosaic germaria showing evidence of recent stem cell loss; namely, the presence of GFP-negative daughters (cystoblasts/cysts generated from an original GFP-negative stem cell) in the absence of the GFP-negative mother stem cell (Laws and Drummond-Barbosa, 2015). Results were subjected to Chi-Square analysis using Microsoft Excel. Early germline cysts were identified based on fusome morphology (de Cuevas and Spradling, 1998; Spradling, 1993).

2.3. Immunofluorescence and microscopy

Ovaries were prepared for immunofluorescence microscopy as described (Ables and Drummond-Barbosa, 2013). Ovaries were dissected and ovarioles teased apart in Grace’s medium without additives (Lonza) and fixed in 5.3% formaldehyde in Grace’s medium for 13 min at room temperature. They were then washed extensively in phosphate-buffered saline (PBS, pH 7.4; Fisher) with 0.1% Triton X-100, and blocked for three hours in blocking solution [5% bovine serum albumin (Sigma), 5% normal goat serum (MP Biomedicals), and 0.1% Triton X-100 in PBS] at room temperature. To detect cells in S phase, dissected ovaries were incubated for one hour at room temperature in Grace’s media containing 10 μM 5-ethynyl-2′-deoxyuridine (EdU; Life Technologies). Following fixation, ovaries were washed extensively in 0.1% Triton X-100 in PBS, ovarioles were teased apart, and samples were permeabilized in 0.5% Triton X-100 in PBS and blocked as described above. The following primary antibodies were diluted in blocking solution and used overnight at 4 °C: chicken anti-GFP (#13970, Abcam; 1:2000), rabbit anti-dsRed (#632496, Clontech; 1:500), rabbit anti-phosphoHistone H3 (pHH3; #06–570, Millipore; 1:200), and mouse anti-CycB [F2F4, Developmental Studies Hybridoma Bank (DSHB); 1:20]. Primary antibodies mouse anti-Hts (1B1, DSHB; 1:10) and mouse anti-Lamin C (LC28.26, DSHB; 1:100) were incubated over two nights at 4 °C. Following a two hour incubation at room temperature with AlexaFluor 488-, 568- or 633-conjugated goat species-specific secondary antibodies (Life Technologies; 1:200), EdU was detected (if necessary) using AlexaFluor-594 or −647 via Click-It chemistry, following the manufacturer’s recommendations (Life Technologies). Ovaries were stained with 0.5 μg/ml 4′−6-diamidino-2-phenylindole (DAPI; Sigma) in 0.1% Triton X-100 in PBS, and mounted in 90% glycerol mixed with 20% n-propyl gallate (Sigma). Confocal z-stacks (1 μm optical sections) were collected with a Zeiss LSM 700 laser scanning microscope using ZEN Black 2012 software. Images were analyzed using Zeiss ZEN Blue 2012 software, and minimally and equally enhanced via histogram using ZEN and Adobe Photoshop CS6.

2.4. Quantification of Fly-FUCCI fluorescence intensity in GSCs

Fluorescence intensity in confocal sections was measured via ZEN Blue 2012 (Zeiss) by manually demarcating individual GSC nuclei (for GFP::E2f1) or cytoplasms (for RFP::CycB), and measuring intensity mean values (gray value/pixel) at the z-level containing the largest nuclear diameter. Because of slight variations in pixel intensity among stain sets, intensity mean values for each fluorescent protein were calculated for a minimum of 400 individual GSCs from two to five independent experiments. Individual GSCs were designated as “hi” or “lo” expression for each reporter depending on whether intensity mean value for that GSC fell in the upper 75th percentile (“hi”) or lower 25th percentile (“lo”) of all observations. Statistical analysis was performed using one-way ANOVA in Prism (GraphPad).

3. Results

3.1. Fly-FUCCI reporters are dynamically expressed in the Drosophila germline

While cell differentiation has been extensively studied in the early germline, the temporal dynamics of the cell cycle in mitotically dividing Drosophila germ cells have not been characterized. We therefore used the Fly-FUCCI system (Zielke et al., 2014) to visualize cell cycle phases in early germ cells as they divide and differentiate in situ. The Fly-FUCCI system uses two reporters that mark different phases of the cell cycle. An N-terminal fragment of CycB fused to RFP (RFP::CycB) marks cells in S, G2, and M phases, while a N-terminal fragment of E2f1 fused to GFP (GFP::E2f1) marks cells in G2, M, and G1 phases (Fig. 1E) (Zielke et al., 2014). To specifically express Fly-FUCCI transgenes in GSCs and their progeny, we used two well-characterized germline-specific drivers (nos-Gal4 and nos. NGT-Gal4) that are sufficient to promote transgene expression in GSCs, all mitotically dividing cysts, and post-mitotic 16-cell cysts in the posterior of the germarium (Fig. 1B). Notably, the strength of nos-Gal4 activation varies across these developmental stages: levels of a UASp-GFP::α-tubulin reporter under the control of the nos-Gal4 driver were lowest and most variable in GSCs and cystoblasts, while reporter expression was consistently high in 4-, 8-, and 16-cell cysts (Fig. 1B). We co-labeled cells in nos-Gal4 > Fly-FUCCI and nos. NGT-Gal4 > Fly-FUCCI ovaries with antibodies against Hts (fusomes and follicle cell plasma membranes) and Lamin C (cap cell nuclear membranes) to unambiguously identify GSCs and cystocytes at each stage of early development based on their fusome structure and position relative to the somatic cells. We also used antibodies against GFP and RFP to visualize the reporter constructs (GFP::E2f1 and RFP::CycB, respectively) in fixed tissue. GFP::E2f1 was highly expressed in early germ cells (Fig. 2), and frequently co-expressed with both cytoplasmic RFP::CycB (Fig. 2A; Fly-FUCCI line 55101) and a RFP::CycB variant that includes a nuclear localization signal (Fig. 2B; Fly-FUCCI line 55111) (Zielke et al., 2014). Co-localization of the Fly-FUCCI reporters with an antibody against Drosophila CycB demonstrated that the Fly-FUCCI system largely, but not completely, recapitulates endogenous CycB expression in the early germline (Fig. 3). In general, RFP::CycB expression appeared more sensitive to endogenous CycB levels than the anti-CycB antibody, likely due to high background fluorescence in samples immunostained with anti-CycB. Further, RFP::CycB expression co-localized with the fusome in GSCs and cysts (Fig. 3A), mirroring the expression of endogenous CycB (Fig. 3B) (Mathieu et al., 2013). In particular, the Fly-FUCCI reporters yielded GSCs containing GFP expression, RFP expression, or a combination of both (Fig. 2). The majority of GSCs were in G2, while a smaller percentage were in G1 and S (Fig. 2C–D and Table 1), consistent with previous studies (Ables and Drummond-Barbosa, 2013; Hsu et al., 2008). Surprisingly, a small proportion of GSCs did not express either marker. Taken together, these data demonstrate that the Fly-FUCCI reporters are dynamically expressed in early germ cells, suggesting that they may serve as accurate indicators of cell cycle phase of germ cells in situ.

Fig. 2. Fly-FUCCI reporters are dynamically expressed in GSCs and mitotically dividing cysts.

(A-B″) nos-Gal4 > UASp-GFP::E2f1,UASp-mRFP::CycB (A-A″) or nos-Gal4 > UASp-GFP::E2F1,UASp-mRFP::nls-CycB (B-B″) germaria labeled with anti-GFP (green; GFP::E2f1), anti-RFP (red; RFP::CycB), anti-Hts (blue; fusomes and follicle cell membranes), and anti-LamC (blue; nuclear envelope of cap cells). Dotted lines demarcate GSCs (white) and cystoblasts (yellow). Asterisks indicate cells that co-express the Fly-FUCCI reporters. (C-D) Percentage of cells/cysts containing GFP::E2f1 only (green), RFP::CycB only (red), both (yellow), or neither (white) in nos-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB, vsg-GFP (C) or nos.NGT-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB (D) females at 6 days after eclosion. The number of cells/cysts analyzed is shown above bars. Scale bar, 5 μm.

Fig. 3. Germ cells display a decreasing window of CycB expression coincident with exit from the mitotic cell cycle.

(A-A″) nos-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB, vsg-GFP germarium labeled with anti-CycB (green; cytoplasm), anti-Hts (green; fusomes and follicle cell membranes), anti-LamC (green; nuclear envelope of cap cells), and anti-RFP (red; RFP::CycB). Asterisks indicate cells/cysts that co-express endogenous CycB and RFP::CycB. Arrows indicate cells that express RFP::CycB, but low endogenous CycB. (B-Bʹ) vsg-GFP germarium labeled with labeled with anti-CycB (green), anti-Hts (red; fusomes and follicle cell membranes), and anti-LamC (red; nuclear envelope of cap cells). (C) Percentage of cells/cysts containing endogenous CycB (pink) or RFP::CycB (red). (RFP::CycB data reproduced for ease of comparison from Fig. 2C and Table 1.) The number of cells/cysts analyzed is shown above bars. Scale bar, 5 μm.

Table 1.

The Fly-FUCCI system and cell cycle markers reveal altered cell cycle dynamics coincident with differentiation.

	FUCCIa expression combinations				Cell cycle markers
Germarium cell types	% E2f1hi, CycBlo	% E2f1lo, CycBhi	% E2f1hi, CycBhi	% E2f1lo, CycBlo	%pH-H3+	%EdU+
GSC	11.07 (54/488)b	11.89 (58/488)	63.93 (312/488)	13.11 (64/488)	5.24 (12/229)	6.35 (31/488)
Pre-CB&CB	10.04 (23/229)	10.04 (23/229)	73.80 (169/229)	6.11 (14/229)	14.02 (15/107)	9.17 (21/229)
2-cell cyst	9.02 (12/133)	29.32 (39/133)	57.14 (73/133)	4.51 (6/133)	34.55 (19/55)	26.32 (35/133)
4-cell cyst	1.43 (1/70)	55.71 (39/70)	14.29 (10/70)	28.57 (20/70)	60.00 (18/30)	28.57 (20/70)
8-cell cyst	1.36 (2/147)	12.93 (19/147)	6.12 (9/147)	79.60 (117/147)	28.85 (15/52)	42.18 (62/147)
16-cell cyst	25.72 (142/552)	7.78 (43/552)	50.91 (281/552)	15.58 (86/552)	0.00 (0/292)	14.13 (78/552)

^aQuantification of nos-Gal4 > Fly-FUCCI ovaries.

^bThe number of positive GSCs/CBs/cysts over total number analyzed is shown in parenthesis.

3.2. Temporal regulation of the cell cycle is coordinated with germ cell differentiation and mitotic exit

Early germ cells transition through several different cell cycles as they differentiate. For example, GSCs have a rapid cell cycle with an asymmetric mitotic division, while 16-cell cysts exit a canonical cell cycle and enter pre-meiotic S phase and endocycles, coincident with terminal differentiation to oocyte or nurse cell fates. We therefore asked whether we could use the Fly-FUCCI system to visualize changes in cell cycle dynamics as germ cells differentiate. By measuring the proportion of each Fly-FUCCI expression pattern for germ cells at each stage of development (GSCs through 16-cell cyst in Regions 1 and 2a/2b, prior to encapsulation), we directly compared cell cycle phase lengths over the course of differentiation (Fig. 2C–D and Table 1). The cell cycle of cystoblasts closely resembled that of the GSCs: most cells exhibited a G2 Fly-FUCCI expression pattern, and a smaller proportion were in G1 and S (Fig. 2C–D). In contrast, 4-cell and 8-cell cystocytes had a strikingly different pattern (Fig. 2C–D). GFP::E2f1 levels, followed by RFP::CycB levels, rapidly decreased across the second and third mitotic divisions, such that the majority of 8-cell cystocytes expressed neither transgene. Similarly, the number of cysts expressing endogenous CycB progressively decreased from the first to last mitotic division (Fig. 3C). Expression of both GFP::E2f1 and RFP::CycB then resumed in the 16-cell cysts in Region 2a (Fig. 2C–D), presumably reflecting the entrance of these cells into premeiotic S phase. These results indicate that the cystocyte cell cycle is dramatically remodeled concurrent with mitotic exit and the onset of differentiation, particularly during the 4-cell and 8-cell cyst stages.

3.3. Fly-FUCCI patterns suggest that up-regulation of APC/C and CRL4 activity precedes germ cell differentiation and mitotic exit

The Fly-FUCCI system relies on the timing of GFP::E2f1 and RFP::CycB degradation (Zielke and Edgar, 2015). Since the temporal regulation of protein degradation could differ from one cell type to another, we examined whether expression of the Fly-FUCCI transgenes in early germ cells correlated with well-known molecular markers of the cell cycle. Based on the kinetics of GFP::E2f1 and RFP::CycB degradation in Drosophila S2 cells and wing imaginal disc cells, we expected that cells in mitosis should express both fluorescent transgenes (Fig. 1E) (Zielke et al., 2014). We therefore asked whether the Fly-FUCCI transgenes were co-expressed with the mitotic marker, phosphorylated Histone H3 (pHH3), in GSCs and their daughters (Fig. 4A, D). All GSCs, cystoblasts, and 2-cell cystocytes in mitosis also co-expressed both Fly-FUCCI reporters (Fig. 4A, D); these results are consistent with previous studies (Zielke et al., 2014). In contrast, 4-cell cysts in mitosis displayed all four combinations of Fly-FUCCI reporters (Fig. 4D). Moreover, 8-cell cystocytes that expressed pHH3 were consistently negative for both reporters (Fig. 4D). These results demonstrate that the Fly-FUCCI reporters faithfully reflect M phase in early germ cells, but the pattern deviates after the second mitotic division.

Fig. 4. Temporal regulation of the cell cycle is altered coincident with differentiation in the germline.

(A-C″) nos-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB, vsg-GFP germarium labeled with anti-GFP (green; GFP::E2f1, vsg::GFP), anti-RFP (red; RFP::CycB), anti-Hts (blue; fusomes and follicle cell membranes), anti-LamC (blue; nuclear envelope of cap cells) and anti-pHH3 (red; cells in M phase) (A-A″) or EdU (blue; cells in S phase) (B-C″). Dotted lines demarcate GSCs (white) and cytoblasts (yellow). Arrows indicate cells that co-express GFP::E2f1, RFP::CycB, and pHH3 (A) or RFP::CycB and EdU (B); arrowheads indicate cells that express EdU alone (B-C). Scale bar, 5 μm. (D-E) Percentage of pHH3 positive (D) or EdU positive (E) cells/cysts containing GFP::E2f1 only (green), RFP::CycB only (red), both (yellow), or neither (white) in nos-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB, vsg-GFP females at 6 days after eclosion. The number of cells/cysts analyzed is shown above bars.

Since degradation of E2f1 accompanies entry into S phase (Shibutani et al., 2008), we hypothesized that cells in S would predominately express high levels of RFP::CycB, but low levels of GFP::E2f1. We therefore compared Fly-FUCCI expression with the incorporation of 5-ethynyl-2′-deoxyuridine (EdU), a marker for cells in S phase. The total percentage of GSCs that incorporated EdU (Table 1) were consistent with previous reports (Ables and Drummond-Barbosa, 2013; Hsu et al., 2008), suggesting that the presence of the Fly-FUCCI reporters does not impede cell cycle progression. Most EdU-positive GSCs, cystoblasts, and 2-cell cysts expressed RFP::CycB alone, as expected (Zielke et al., 2014) (Fig. 4B, E). About half of the EdU-positive 16-cell cystocytes also expressed RFP::CycB (Fig. 4E), likely due to the entry of these cells into pre-meiotic S (Spradling, 1993). A smaller proportion of S phase GSCs and cystoblasts, however, did not express either reporter (Fig. 4C, E). Furthermore, the percentage of double-negative S phase cystocytes increased in the 8-cell stage, similar to M phase cells (Fig. 4D). The presence of cells with double-negative reporter expression has not been reported in the Fly-FUCCI system (Zielke et al., 2014), but could be indicative of a transitional period in early S where GFP::E2f1 is being degraded and RFP::CycB is low (Fig. 1D). While we cannot rule out the possibility that some GSCs and germ cells lack the Fly-FUCCI reporters due to incomplete expression of the reporter transgenes, these data suggest that the activities of the ubiquitin ligases APC/C and CRL4 (which degrade CycB and E2f1, respectively) are very high at the 4/8-cell cyst stage, coincident with the end of the mitotic divisions.

Importantly, comparison of the proportion of GSCs/cysts in M phase (pHH3-positive) and S phase (EdU-positive), independent of the Fly-FUCCI reporters, revealed a dramatic shift in the timing of mitosis and synthesis phases concurrent with differentiation (Fig. 4D–E and Table 1). In stark contrast to GSCs, in which less than 10% of cells express either pHH3 or EdU, more than 50% of 4-cell cystocytes and 40% of 8-cell cystocytes were observed in active M or S phase, respectively. Taken together, these results suggest that GSCs, cystoblasts, and 2-cell cystocytes contain a short G1, a long G2, and short replication and mitotic phases, likely facilitating the rapid turnover of these early stages. Following the second mitotic division, the cell cycle is dramatically remodeled, such that the lengths of M and S phases increase as germ cells approach the terminal mitotic division. The absence of the Fly-FUCCI reporters at the 4- and 8-cell stages may indicate that germ cells undergo a progressive shortening of the G2 phase after the second mitotic division, effectively stalling the cell cycle at the last mitotic division prior to initiating more specialized cell cycles.

3.4. The Fly-FUCCI system confirms that GSCs have a very short G1

Cell cycle progression has been extensively characterized in GSCs, and can be tracked by examining the morphology of the fusome, which changes synchronously with the cell cycle in GSCs (Fig. 1D) (Ables and Drummond-Barbosa, 2013; de Cuevas and Spradling, 1998; Hsu et al., 2008; Kao et al., 2015; Mathieu et al., 2013). As an independent assessment of the temporal dynamics of GFP::E2f1 and RFP::CycB degradation in early germ cells, we therefore compared the fusome morphologies of individual GSCs to the expression of the Fly-FUCCI reporters (Fig. 5). We incorporated a protein trap transgene (vsg-GFP) that localizes specifically to ring canals (including the transient ring canal that forms between the GSC and the pre-cystoblast; see Fig. 1D) into the Fly-FUCCI reporters, allowing us to unambiguously identify fusome morphology in each GSC. Individual GSCs expressed each Fly-FUCCI reporter at varying intensity, making it difficult to assess whether a reporter in a given cell was truly “off” or “on.” To accurately quantify the proportion of GSCs in each phase of the cell cycle, we therefore measured fluorescence signal intensities for each Fly-FUCCI reporter in more than 450 individual GSCs (Fig. 5A–C). We established whether a given cell displayed low or high expression of the reporter protein: “lo” expressing cells fell in the lower 25th percentile of fluorescence intensity mean value, while “hi” expressing cells fell in the upper 75th percentile. As expected, RFP::CycB levels were lowest in G1/S phase GSCs, and increased as cells progressed from S into G2/M (Fig. 5A). GFP::E2f1 expression was also lowest in G1/S phase GSCs, and increased as cells entered into G2 (Fig. 5B). Intriguingly, while most GSCs with “round” fusomes (previously characterized as an indicator of G2/M) expressed high levels of both reporters, a small population of GSCs with “round” fusomes expressed only the GFP::E2f1 reporter (10%; 27/268 “round” fusome GSCs) (Fig. 5C, E). This suggests that degradation of GFP::E2f1 in GSCs begins slightly before fusome extension at the M/G1 transition. Furthermore, we consistently detected GSCs with low levels of both reporters, regardless of the cell cycle stage (Fig. 5C). The low percentage of GFP::E2f1^hi/RFP::CycB^lo GSCs with a round fusome, and the low levels of GFP::E2f1 in very early S phase GSCs (i.e., those with “plug” and “bar” fusome morphologies), support the assertion that G1 is likely very short in GSCs (Ables and Drummond-Barbosa, 2013; Hsu et al., 2008; Kao et al., 2015).

Fig. 5. Fusome morphogenesis in newly divided GSC/cystoblast pairs lags behind GFP::E2f1 degradation at the M/G1 transition.

(A-B) Average RFP::CycB (A) or GFP::E2f1 (B) fluorescence intensity in GSCs at each stage of the cell cycle (as determined by fusome morphology). Bars represent s.e.m. (C-D) Percentage of GSCs containing GFP::E2f1 only (green), RFP::CycB only (red), both (yellow), or neither (white) in nos-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB, vsg-GFP females at 6 days after eclosion on a nutrient-rich (C) or nutrient-deficient (D) diet. The number of cells analyzed is shown above bars. (E-E″) nos-Gal4 > UASp-GFP::E2f1, UASp-mRFP::CycB, vsg-GFP germarium labeled with anti-GFP (green; GFP::E2f1, vsg::GFP), anti-RFP (red; RFP::CycB), anti-Hts (blue; fusomes and follicle cell membranes), and anti-LamC (blue; nuclear envelope of cap cells). Dotted lines demarcate GSCs. Arrow indicates “round” fusome GSC expressing GFP::E2f1 and RFP::CycB; arrowhead indicates “round” fusome GSC expressing GFP::E2f1 alone. Scale bar, 5 μm.

Previous studies have demonstrated that GSCs are sensitive to maternal diet (Hsu et al., 2008). The observation that some GSCs identified as G2/M phase by fusome morphology may actually be in G1 (GFP::E2f1^hi/RFP::CycB^lo) prompted us to re-examine how the GSC cell cycle changes in response to altered environmental cues. GSCs from females fed a nutrient-poor diet displayed different temporal kinetics of Fly-FUCCI reporter degradation than those from well-fed females (Fig. 5D). We reasoned that if G1 was longer in nutrient-deficient GSCs, the percentage of GFP::E2f1^hi/RFP::CycB^lo GSCs with a “round” fusome would increase. While the number of GFP::E2f1^hi/RFP::CycB^lo GSCs with a “round” fusome was only modestly increased in nutrient-poor conditions (13.6%; 31/228 GSCs), we found no GFP::E2f1^hi/RFP::CycB^lo GSCs with a “plug” fusome, suggesting that CRL4 activity has increased, effectively stalling cells at the G1/S transition. Further, more GSCs in G1/early S co-expressed the Fly-FUCCI reporters, suggesting that the timing of RFP::CycB degradation (and thus, APC/C activity) is altered in GSCs from a nutrient-deficient environment. Consistent with previous studies (Hsu et al., 2008; Kao et al., 2015), these data indicate that GSCs from poorly fed females delay entry into S, spend proportionally more time in early S, and have a longer G2 than GSCs from well-fed females. These results suggest that CRL4 and APC/C play important roles in regulating GSC proliferation rates in response to environmental conditions.

3.5. E2f1 is necessary for GSC proliferation and self-renewal

The intriguing pattern of GFP::E2f1 degradation in GSCs and their progeny suggests that E2f1 and CRL4 have important roles in the control of GSC function. E2f1 is necessary for GSC proliferation and the regulation of nurse cell endocycles (Jin et al., 2008; Royzman et al., 2002). Cyclin E (CycE), a target of E2f1, is also required for GSC proliferation, and is essential for proper GSC self-renewal and cyst development (Ables and Drummond-Barbosa, 2013). Moreover, CycB and components of the APC/C are also necessary for GSC self-renewal and the division of GSCs and their progeny (Chen et al., 2009; Wang and Lin, 2005). In contrast, previous studies have not supported a role for E2f1 or its obligate binding partner (encoded by Dp) in the control of GSC self-renewal or cyst mitotic divisions (Jin et al., 2008; Myster et al., 2000; Royzman et al., 2002). We therefore re-visited the hypothesis that E2f1 is necessary for GSC self-renewal and cyst formation. We used the Flippase (FLP)/FLP Recognition Target (FRT) genetic mosaic technique (Laws and Drummond-Barbosa, 2015) to create E2f1-deficient GSCs and early germ cells. GSCs and their progeny were identified by co-immunofluorescent detection of Hts and Lamin C, and E2f1 mutant (or wildtype, in case of controls) germ cells were recognizable by the absence of a GFP marker (Fig. 6A). We first confirmed that E2f1 was necessary for GSC proliferation by analyzing EdU incorporation in genetic mosaic females carrying E2f1-deficient GSCs (Fig. 6B–C, H). As expected, the percentage of E2f1 mutant GSCs that incorporated EdU (1.9%; 1/51 GSCs) was significantly less than wild-type GSCs in control “mock” mosaics (11.2%; 10/89 GSCs; Fig. 6H), indicating that E2f1 is required for GSC proliferation. Intriguingly, the few E2f1 mutant GSCs that did incorporate EdU failed to progress past early S phase, as evidenced by weak, punctate EdU fluorescence (Fig. 6C). While most E2f1 mutant mosaics contained only GFP-negative cystoblasts or 2-cell cysts (Fig. 6C), we occasionally found E2f1 mutant 8-cell and 16-cell cysts, suggesting that cyst divisions can proceed, albeit infrequently, in the absence of E2f1. We did not, however, find E2f1 mutant cysts with greater than 16 cells, as was observed in Dp mutants (Myster et al., 2000). Additionally, unlike CycE-deficient GSCs (Ables and Drummond-Barbosa, 2013), E2f1 mutant GSCs were of comparable size to adjacent wild-type GSCs, suggesting that normal growth controls are not disrupted.

Fig. 6. E2f1, but not Cul4, is necessary for GSC proliferation and self-renewal.

(A) The FLP/FRT technique was used to generate genetic mosaics. Mitotic recombination is mediated by heat-shock-induced expression of flippase (hsFLP). Homozygous E2f1 or Cul4 mutant (mut) cells are identified by the absence of a GFP marker, which is linked to the wildtype allele. (B-G). Representative control (B, E), E2f1⁷²⁹ mutant (C-D), Cul4^KG02900 mutant (F), or Cul4^11L mutant (G) mosaic germaria labeled with anti-GFP (green; wild-type cells), anti-Hts (red; fusomes and follicle cells), anti-LamC (red; nuclear envelope of cap cells), and EdU (red; cells in S phase). Lines demarcate wild type (dashed) or mutant (solid) GSCs (white) and cystoblasts (yellow). Arrows indicate GFP-negative cysts. Arrowhead indicates E2f1⁷²⁹ mutant GSC that initiates S phase. Scale bar, 5 μm. (H) Percentage of GFP-negative GSCs that are positive for EdU in mosaic germaria at 6 or 8 days after heat shock (AHS). Numbers in the bars represent the number of GFP-negative GSCs analyzed. (I) Percentage of germline-mosaic germaria with a GSC loss event 6 or 8 days after clone induction. Numbers in the bars represent the number of germline-mosaic germaria analyzed. *p < 0.05.

We then tested whether E2f1 is required for GSC self-renewal by analyzing GSC loss in control versus E2f1 mutant mosaic germaria (Fig. 6B–D, I; see Section 2). To measure GSC loss, we calculated the percentage of germaria with a mosaic germline that contained GFP-negative cystoblasts/cysts, but lacked an accompanying GFP-negative GSC (i.e., a GSC loss event). In control “mock” mosaics, where all cells are wild type, few GSC loss events were observed (Fig. 6B, I). In contrast, a small, yet significantly increased percentage of E2f1 mosaic germaria had mutant cystoblasts/cysts in the absence of an E2f1 mutant GSC (Fig. 6D, I). Taken together, our results indicate that E2f1 is required for GSC self-renewal and the proliferation of GSCs and their progeny. We speculate that E2f1-deficient GSCs fail to be maintained as a consequence of their failure to progress through the cell cycle.

3.6. E2f1 degradation by CRL4 is essential for germline cyst divisions, but is unlikely to be a mechanism limiting GSC proliferation or self-renewal

The rapid degradation of GFP::E2f1 immediately following the M/G1 transition in GSCs and at the third mitotic division in cysts suggests that CRL4 activity is a critical determinant of early germ cell development. In somatic cells, such as those in the developing wing disc, E2f1 is required for S phase entry, but must be degraded for replication to proceed properly (Asano et al., 1996; Davidson and Duronio, 2012; Heriche et al., 2003; Reis and Edgar, 2004; Shibutani et al., 2008). Indeed, overexpression of E2f1 or expression of stable versions of degradation-resistant E2f1 result in ectopic S phase initiation and apoptosis (Asano et al., 1996; Davidson and Duronio, 2012; Shibutani et al., 2008). Cul4 encodes the primary scaffolding subunit of the CRL4 ubiquitin ligase complex; cells with insufficient Cul4 protein levels do not properly degrade E2f1 (Lee et al., 2010; Shibutani et al., 2008). We reasoned that if CRL4 activity is necessary for progression through the cell cycle in GSCs, then loss of Cul4 should delay or abrogate GSC proliferation. Using the FLP/FRT system, we created genetic mosaics carrying homozygous mutations in Cul4, and assayed Cul4-deficient GSCs for incorporation of EdU (Fig. 6E–H). Intriguingly, the percentage of GSCs in S phase was not statistically different between wild-type GSCs in control “mock” mosaics and those harboring either a weak hypomorphic Cul4 allele (Cul4^KG02900) or a null Cul4 allele (Cul4^11L) (Fig. 6H). These results indicate that Cul4 is not necessary for GSC proliferation. In stark contrast, cyst divisions were significantly blunted in the absence of Cul4, as evidenced by a dramatic decrease in cyst formation in Cul4^11L mutant germline mosaic germaria (Fig. 6G). Importantly, the average number of 16-cell cysts formed per GSC in Cul4^KG02900 germline mosaic germaria (0.64; Fig. 6F) was also significantly less than the number formed in wild-type “mock” controls (1.41; Fig. 6E; p < 0.0001, Student’s paired T-test), indicating that 8-cell cysts are particularly sensitive to Cul4 function. These results indicate that CRL4 is necessary for cyst divisions, yet dispensable in GSCs for their progression through the cell cycle.

We then tested whether Cul4 is necessary for GSC self-renewal. Cul4^KG02900 hypomorphic mutant germline mosaic germaria displayed GSC loss equivalent with wild-type “mock” control mosaics (Fig. 6E–F, I). GSC loss in Cul4^11L null mutant germline mosaic germaria was also statistically indistinguishable from controls, despite the fact that cyst divisions were clearly impaired (Fig. 6G, I). These data demonstrate that Cul4 is not required for GSC maintenance. We cannot, however, rule out the possibility that another Cullin family member could compensate for loss of Cul4 within the CRL4 ubiquitin ligase complex.

As an additional method to test whether E2f1 degradation is necessary for GSC proliferation or maintenance, we expressed a non-degradable, GFP-tagged E2f1 transgene (GFP::E2f1^PIP−3A) specifically in GSCs and their progeny. GFP::E2f1^PIP−3A harbors a mutation in the E2f1 “PIP-box” destruction degron, rendering it insensitive to CRL4 (Fig. 1E) (Davidson and Duronio, 2012; Shibutani et al., 2008). GSCs harboring wild-type (GFP::E2f1) and non-degradable (GFP::E2f1^PIP−3A) reporters were identified by co-immunofluorescent detection of Hts, Lamin C, and GFP, and ovaries were sampled from young (6 days after eclosion) and mature (16 days after eclosion) female flies. In wild-type GFP::E2f1 germaria, 78.6% (young females) or 55.3% (mature females) of GSCs expressed GFP (Table 2); the difference between groups is likely due to changes in cell cycle regulation that occur with aging (Kao et al., 2015). In contrast, in germaria from either maternal age expressing the degradation-resistant GFP::E2f1^PIP−3A, more than 90% of GSCs expressed GFP, consistent with an inability to be degraded. To test whether preventing E2f1 degradation abrogated S phase entry, we measured EdU incorporation in wild-type and GFP::E2f1^PIP−3A germaria. Surprisingly, there was no statistically significant difference in the percentage of GSCs in S phase between the wild-type GFP::E2f1 and the non-degradable GFP::E2f1^PIP−3A mutant (Fig. 7A–C; Table 2). Furthermore, while wild-type GFP::E2f1 GSCs did not co-express EdU (Fig. 7A), GFP::E2f1^PIP−3A mutant GSCs frequently co-expressed EdU (Fig. 7B), suggesting that degradation of E2f1 is not required for entry into S phase. Moreover, we detected no obvious signs of GSC loss due to premature differentiation or death: the average number of GSCs per germarium in GFP::E2f1^PIP−3A mutants was not statistically different from controls (Fig. 7D; Table 2), and the rate of GSC loss as flies age was comparable to expression of the wild type GFP::E2f1 transgene (Fig. 7D; Table 2). Even when female flies were subjected to suboptimal nutritional conditions, we found no significant differences in GSC proliferation or number in germaria from wildtype GFP::E2f1 and GFP::E2f1^PIP−3A mutants (Table 2). Taken together with the phenotype of Cul4 mutant GSCs, these data suggest that E2f1 degradation is not a major contributing factor to the regulation of GSC proliferation or self-renewal.

Table 2.

Expression of a stable version of E2f1 yields no change in GSC proliferation or maintenance.

Genotype	Maternal diet	Maternal agea	% GFP-positive GSCs (n)b	% EdU-positive GSCs	Average number of GSCs per germarium (n)c
GFP::E2f1	Nutrient-rich	6	78.61 (519)	13.68	2.88 (180)
		16	55.33 (291)	10.65	2.31 (126)
	Nutrient-poor	6	70.61 (592)	7.26	2.90 (204)
		16	46.69 (559)	4.83	2.80 (200)
GFP::E2f1PIP3A	Nutrient-rich	6	93.06 (519)	11.75 (p = 0.20)	2.75 (189) (p = 0.13)
		16	96.63 (326)	10.43 (p = 0.90)	2.20 (148) (p = 0.31)
	Nutrient-poor	6	99.01 (406)	8.13 (p = 0.50)	2.96 (137) (p = 0.55)
		16	95.32 (491)	3.67 (p = 0.23)	2.63 (187) (p = 0.05)

^aMaternal age (in days after eclosion) at which ovaries were dissected. Females were maintained on the indicated diet beginning one to two days after eclosion through the day of dissection.

^bThe total number of GSCs analyzed is shown in parentheses.

^cAverage number of GSCs per germarium ± SEM. The total number of germaria analyzed is shown in parentheses.

Fig. 7. E2f1 degradation by CRL4 appears largely dispensable for GSC maintenance and proliferation.

(A-B) nos-Gal4 > UASp-GFP::E2F1 (A-A′; E2f1.WT) and nos-Gal4 > UASp-GFP-E2f1.PIP-3A (B-B′; E2f1.PIP-3A) germaria labeled with anti-GFP (green; GFP::E2f1), anti-Hts (red; fusomes and follicle cells), anti-LamC (red; nuclear envelope of cap cells), and EdU (red; cells in S phase). Lines demarcate GSCs (white) and cystoblasts (yellow) in wild type (dotted) or mutant (solid) germaria. Arrows indicate cells that co-express GFP and EdU; arrowheads indicate cells that express EdU alone. Scale bar, 5 μm. (C) Percentage of GSCs that are EdU positive in E2f1.WT and E2f1.PIP-3A germaria at 6 or 16 days after eclosion (DAE) on a nutrient-rich diet. (D) Average number of GSCs in E2f1.WT (gray line) or E2f1.PIP-3A (red line) 6 and 16 days after eclosion (DAE) on a nutrient-rich diet. Bars represent s.e.m.

4. Discussion

Although cell cycle regulation of the early germline has been well-studied, little is known about how the temporal dynamics of the cell cycle are remodeled to accommodate germline differentiation. Our study, using the Fly-FUCCI system (Zielke et al., 2014) to investigate the cell cycle dynamics of germ cells within their cellular environment, provides novel observations about the molecular control of the cell cycle in GSCs and their differentiating progeny. Our results suggest that GSCs, cystoblasts, and early cysts (2-cell) contain a very short G1, a long G2, and short replication and mitotic phases, likely facilitating the rapid turnover of these early stages. Following the second mitotic division, the cell cycle is dramatically remodeled, such that 4-cell and 8-cell cysts have longer mitotic and replication phases with relatively short gap phases. The absence of the Fly-FUCCI reporters at the 4- and 8-cell stages likely indicates that activity of the APC/C and CRL4 ubiquitin ligases are very high at the 4/8-cell transition. In support of this observation, we find that cyst development is delayed at the 8-cell stage in the absence of Cul4, which encodes a critical subunit of CRL4. Taken together, our results suggest the model that increased APC/C and CRL4 activity signals the end of the mitotic divisions and/or stalls the cell cycle at the last mitotic division to accommodate the initiation of the specialized cell cycles of the oocyte and nurse cells.

Importantly, the Fly-FUCCI reporter dynamics do not perfectly match the fusome cycle in GSCs. For example, GSCs with a “round” fusome morphology have been previously associated with the G2/M phases of the cell cycle (Ables and Drummond-Barbosa, 2013; de Cuevas and Spradling, 1998; Hsu et al., 2008; Kao et al., 2015). In the Fly-FUCCI system, we expected GSCs with “round” fusomes to express both GFP::E2f1 and RFP::CycB; however, we identified “round” fusome GSCs that express GFP::E2f1 only. We also identified a population of GSCs that do not express either Fly-FUCCI reporter. Two possibilities may explain these results. The first is that gaps between reporter expression or delayed reporter degradation simply reflect technical problems with the system. For example, GSCs with low reporter levels could be due, at least in part, to incomplete expression of nos-Gal4 in GSCs. In addition, our data indicate that fusome morphogenesis lags slightly behind the molecular control of cell cycle progression, resulting in small (but useful) discrepancies between fusome morphology and reporter expression when quantifying the percentage of GSCs in a given cell cycle phase. Future studies using the Fly-FUCCI transgenes in combination with a fusome marker in live imaging could resolve these technical issues.

The second, and more intriguing possibility was that the kinetics of GFP::E2f1 and RFP::CycB degradation (and thus, CRL4 and APC/C activity) in GSCs are different than in other cell types. We therefore considered whether E2f1 degradation could be a critical contributing factor to the regulation of GSC self-renewal and proliferation. We tested this hypothesis by expressing an E2f1 protein that cannot be degraded by the CRL4 ubiquitin ligase in GSCs. Stable versions of E2f1 in wing and eye imaginal discs cause cell death and an inability to proceed into S phase (Davidson and Duronio, 2012). In contrast, we found that while E2f1 is necessary for proliferation and self-renewal of GSCs, its degradation is largely dispensable for GSC function. Although it is possible that another E3 ubiquitin ligase could substitute for CRL4 (Heriche et al., 2003), these results may suggest that alternate mechanisms other than E2f1 degradation predominate in GSCs to promote progression through the cell cycle. Maintained levels of E2f1 in GSCs may also explain why CycE, a transcriptional target of E2f1, is expressed throughout much of the GSC cell cycle, instead of being confined to G1 as in other cell types (Ables and Drummond-Barbosa, 2013; Hsu et al., 2008).

Interestingly, despite the clear positive feedback between E2f1/Dp and CycE in other cell types, we observed quite striking differences between E2f1-, Dp- and CycE-deficient GSCs. Most notably, we did not observe the increased cell size in E2f1-deficient GSCs that characterizes CycE-deficient GSCs (Ables and Drummond-Barbosa, 2013; Wang and Lin, 2005). E2f1 mutant GSCs were also more likely to self-renew than either CycE or Cdk2 mutant GSCs, and E2f1 mutant cystoblasts were able to divide into multicellular cysts. Previous studies demonstrated that Dp, the obligate heterodimeric transcriptional partner of E2f1, is not necessary for cyst production; in fact, a small percentage of Dp mutant cysts are able to undergo a fifth mitotic division, resulting in cysts with 32 cystocytes (Myster et al., 2000). Unlike mammals, the Drosophila genome encodes only two E2f transcription factors. E2f1 functions at a small set of promoters as a transcriptional activator, while E2f2 binds a much broader set of promoters and functions as a repressor; both must dimerize with Dp for proper association with chromatin (Korenjak et al., 2012). Although Dp mutants should theoretically phenocopy E2f1 mutants, studies from somatic cells demonstrated that blocking E2f2 function rescues entry into S phase in E2f1 mutants; thus, Dp mutants represent loss of function of both E2f transcription factors more than either one alone (Frolov et al., 2001, 2005). In the germline, loss of E2f1 may permit ectopic formation of E2f2/Dp complexes, repressing genes that E2f1/Dp would normally activate. Taken together, these data suggest that E2f1/Dp, E2f2/Dp, and CycE/Cdk2 may control different targets in undifferentiated versus differentiated germ cells that enable GSCs to progress through the cell cycle while limiting cyst divisions. Furthermore, we have previously demonstrated that CycE also impacts BMP signaling, blocking cystoblast differentiation (Ables and Drummond-Barbosa, 2013). This suggests that E2f1 and CycE together bias GSCs toward self-renewal, potentially by extending the window of time through which BMP signals can be received. Indeed, mounting evidence suggests that cell cycle dynamics are also key determinants of stem cell fate in mammalian stem cells (Boward et al., 2016; Julian et al., 2016). Further investigation into the targets and kinetics of E2f1 and CycE/Cdk2 control in GSCs are therefore likely to provide important clues regarding the link between cell cycle and differentiation, with applications to a broad variety of stem cells.

Our study supports an emerging molecular model describing how germ cells switch between modes of cell cycle control in a precise temporal manner. In GSCs, E2f1, CycB, and CycE are expressed at high levels through much of the cell cycle, resulting in a very short G1 and a long G2. Accordingly, stabilization of E2f1 or CycB in GSCs has little impact on GSC self-renewal, though both are necessary for production of daughter cells (this study; Chen et al., 2009; Wang and Lin, 2005). In contrast, Cyclin A (CycA), a G2/M cyclin enriched in GSCs with “round” fusomes, must be degraded for proper GSC self-renewal (Ables and Drummond-Barbosa, 2013; Chen et al., 2009; Lilly et al., 2000). GSCs express very low levels of a pro-differentiation factor, encoded by bag of marbles (bam), which is directly repressed by BMP signaling (Chen and McKearin, 2003; Song et al., 2004). Recent studies indicate that Bam, together with an acetyltransferase (encoded by Gcn5) and a deubiquitinase [encoded by ovarian tumor (otu)], promote stabilization of CycA in differentiating cystoblasts (Ji et al., 2017; Liu et al., 2017). Our data demonstrate that levels of the APC/C increase across the second and third mitotic divisions, coincident with differentiation and decreasing levels of Bam (McKearin and Ohlstein, 1995). CycA and CycB, which are both degraded by the APC/C (Teixeira and Reed, 2013), are largely absent from 16-cell cysts (this study; Ohlmeyer and Schupbach, 2003), consistent with high APC/C activity at the 8-cell stage. Further, ectopic expression of CycA, CycB, or CycE forces cysts into a fifth mitotic division (Doronkin et al., 2003; Lilly et al., 2000; Ohlmeyer and Schupbach, 2003). Moreover, CRL4, which targets E2f1 and Dacapo (an inhibitor of CycE), likely reinforces the terminal mitotic division by limiting CycE mRNA production (via E2f1) and CycE/Cdk2 activity (via Dacapo). Indeed, stabilized expression of E2f1 in early germ cells did not result in a fifth mitotic division (Fig. 7), suggesting that degradation of E2f1 alone is not sufficient to limit CycE/Cdk2 activity. Taken together, these data support the model that specific Cyclins are critical for GSC self-renewal and regulation of the spatiotemporal switch from mitotic to meiotic divisions in the Drosophila germline. Future studies will be necessary to test whether different Cyclin/Cdk complexes regulate specific sets of targets to control the switch to meiosis in response to environmental or hormonal cues.