Clonal analysis in Drosophila has led to the discovery of many genes regulating major cellular processes such as growth, differentiation, and cell competition. Studying genetic mosaics has not only been crucial for understanding developmental processes but has also provided useful insights into pathological conditions such as cancer, which is inherently a clonal disease. Over the past decades, Drosophila has proven to be an excellent tool for cancer research, with the identification of tumor suppressor genes, development of genetic models of tumorigenesis or metastasis, and establishment of personalized avatars of human cancers for drug screening. One example of a model of early tumorigenesis in Drosophila is the clonal loss of tumor suppressors phosphatase and tensin homolog (Pten) or tuberous sclerosis complex (TSC) subunit 1 (Tsc1) in the eye imaginal discs of larvae raised under nutrient restriction (NR). Pten and Tsc1 mutant cells outgrow the surrounding wild-type cells by different growth mechanisms: Pten mutant cells shift from a hypertrophic overgrowth to a combined hypertrophic and hyperplastic overgrowth under NR, whereas Tsc1 mutant cells overgrow specifically due to an increase in cell size. Differential activation of the transcription factor forkhead box O (FoxO) upon loss of Tsc1 restricts the proliferation of these cells, and the presence of Tsc1-foxo mutant cells results in massively overgrown tissues containing minimal portions of wild-type cells.

In this study, the capacity of Tsc1 or Tsc1-foxo knockdown cells to overgrow when surrounded by other overgrowing cell populations, such as Pten knockdown cells, was explored. A PtenRi construct under LexO control was developed and used in combination with the coinFLP-LexGAD/Gal4 system to generate different knockdown cell populations within the same tissue. Detailed analysis of clonal populations in eye discs demonstrates that the loss of FoxO enhances the proliferation of Tsc1 knockdown cells, thereby increasing their potential to outcompete Pten knockdown cells.

This study aimed to examine the growth of cells with loss of Tsc1 function in the presence of cells with increased growth advantage, such as Pten knockdown cells. The function of FoxO in mediating this growth regulation was also investigated.

To compare the growth properties of Pten knockdown cells with Tsc1-foxo knockdown cells within the same tissue, the coinFLP-LexGAD/Gal4 system was used to generate tissues with clones expressing either Gal4 or LexGAD. The Pten inverted repeat (IR) from UAS-PtenRi (101475 Vienna Drosophila Resource Center (VDRC)) was cloned downstream of LexO to establish LexO-PtenRi that brings the Pten knockdown under LexGAD control. Combined with the Gal4 control of UAS-Tsc1Ri and UAS-foxoRi, this system produces Pten and Tsc1-foxo knockdown clones side-by-side in the same tissue. Since the Pten knockdown should lead to increased cell size and number, the area between veins L3 and L4 was measured upon knockdown of Pten in the Dpp domain using dpp>LHG (Fig. 1A, A’). The area of the anterior crossvein was also measured as it lies in the region of Dpp expression (Fig. 1A’’). In both cases, the area was increased upon knockdown of Pten as compared to the control.

At the molecular level, loss of Pten function results in hyperactivation of Akt due to its enhanced phosphorylation. Phospho-Akt staining of developing eye discs showed higher signal in the Pten knockdown clones generated using the coinFLP-LexGAD/Gal4 system as compared to the control (Fig. S1A), confirming an efficient loss of Pten activity. Finally, to test whether the growth phenotypes of the LexO-PtenRi allele are comparable to the UAS-PtenRi allele, the clonal growth was observed under NR. The Pten knockdown clones grew much bigger as compared to the control clones under NR (Fig. S1B), indicating that the growth phenotypes of LexO-PtenRi and UAS-PtenRi are similar.

The above-mentioned LexO-PtenRi was recombined with LexO-mCherry-CAAX and UAS-CD8-GFP to label LexGAD and Gal4-driven clones, respectively. Using a combination of eyFLP and coinFLP-LexGAD/Gal4 systems, ctrl, foxo, Tsc1 or Tsc1-foxo knockdown clones were generated in eye discs under Gal4 control, and ctrl or Pten knockdown clones were generated under LexGAD control (Fig. 1B). Since the LexGAD-expressing clones occupy 80% of the disc area, the manipulations using LexGAD are referred to as background of the disc. Whereas ctrl and foxo knockdown clones were present in the ctrl background, they were almost absent in the Pten background. This suggests a compromised growth of ctrl or foxo knockdown cells in the presence of Pten knockdown cells, as has been described earlier. Also consistent with earlier studies, Tsc1 and Tsc1-foxo knockdown produced massive discs composed primarily of knockdown clones in the ctrl background. Interestingly, despite the generation of Pten knockdown clones in 80% of disc area under LexGAD control in the Pten knockdown background, Tsc1 knockdown clones occupied more than 50% of the total disc area in 34.5% of discs analyzed. This percentage was further increased to 50% in Tsc1-foxo knockdown discs (Fig. 1B’).

Furthermore, patches of Gal4-driven GFP-positive cells could be observed in the ventral nerve cord (VNC) in the presence of Tsc1 or Tsc1-foxo knockdown clones in the ctrl or Pten knockdown background (Fig. 1B’’). To check if these patches were due to an invasion of epithelial cells from eye disc to the VNC or developed from brain-specific lineage, the brains were stained with Repo, a marker for glial cells. The positive staining of cells in the VNC with Repo argues against the invasion of knockdown cells from the eye discs, and points towards leakiness of the Gal4/UAS system.

The large disc area occupied by Tsc1 or Tsc1-foxo knockdown clones in Pten knockdown background could be because of larger cell size or higher proliferation of Tsc1 or Tsc1-foxo knockdown cells, or reduced viability of Pten knockdown cells in the presence of Tsc1 or Tsc1-foxo knockdown cells. To explore cell viability, eye discs with Tsc1 or Tsc1-foxo knockdown clones in Pten knockdown background were stained with cleaved Dcp-1 (Fig. 1C). Both Tsc1 and Tsc1-foxo knockdown clones showed a high number of cleaved Dcp-1-positive cells. Taken together with results in figure 1B’, this implies greatly increased proliferation of Tsc1 and Tsc1-foxo knockdown cells to account for elevated apoptosis within the clones. It was also noted that Pten knockdown background cells showed reduced levels of apoptosis in the presence of Tsc1-foxo knockdown cells, suggesting a further enhanced proliferation rate of Tsc1-foxo knockdown cells. These results add to the important role of FoxO in restricting the proliferation of Tsc1 knockdown cells when surrounded by wild-type cells as well as by highly proliferative Pten knockdown cells.

The results of this study have been interpreted using a system that generates mosaic clones in an uneven proportion. The observed apoptosis in Tsc1 and Tsc1-foxo knockdown clones could also be due to the toxicity of Gal4 expression. Although the outcompeting of Pten knockdown clones that occupy 80% of the tissue by Gal4-driven clones is evidence of enhanced growth of Tsc1 and Tsc1-foxo knockdown clones, the analogous experiment with LexO-driven Tsc1 and Tsc1-foxo knockdowns and Gal4-driven Pten knockdown will provide additional insights into the regulation of growth by these different tumor suppressors.

Cloning of LexO-PtenRi

The sequence targeted by LexO-PtenRi was PCR-amplified from y w genomic DNA using primer set PCR1, as used for the UAS-PtenRi construct by VDRC. Restriction digestion sites for EcoRI and XbaI were introduced by using primer set PCR2, and IR was generated by digesting with EcoRI and self-ligation. The IR construct was subcloned into pLOTattB by using the XbaI site to give pLOTattBLexO-PtenRi.

Fly husbandry

All crosses were maintained at 25°C on normal fly food unless otherwise stated. Normal fly food is composed of 100 g fresh yeast, 55 g cornmeal, 10 g wheat flour, 75 g sugar, 8 g bacto agar, and 1.5% antimicrobial agents (33 g/L nipagin and 66 g/L nipasol in ethanol) in 1 L water. NR food was prepared by reducing the amount of yeast to 10% with 1.5% nipagin (100 g/L in ethanol).

Fly lines used: LexO-PtenRi line was generated in this study by injecting the pLOTattBLexO-PtenRi plasmid into line ΦX-22A; dpp>LHG, Act>CD2>LHV2 and LexO-mCherry-CAAX UAS-CD8-GFP tub-Gal80ts; coinFLP-LexGAD/Gal4 (59270 Bloomington Drosophila Stock Center (BDSC)); foxoRi (107786 VDRC); Tsc1Ri (31039 BDSC).

Immunofluorescence and microscopy

Adult wings of females were detached from thorax using forceps 4 days after eclosion and mounted in Euparal (Roth #7356.1). They were imaged on KEYENCE VHX1000 digital microscope. Imaginal discs were dissected 108 h AEL from normal food and 156 h AEL from NR unless otherwise stated.

Imaginal discs and brains were fixed in 4% paraformaldehyde (PFA, 30 min, room temperature (RT)), washed thrice in 0.3% Triton-X in PBS (PBT, 15 min, RT), blocked in 2% Normal Donkey Serum in 0.3% PBT (2 h, 4°C), incubated with primary antibodies (overnight, 4°C), washed thrice in 0.3% PBT (15 min each, RT), incubated with secondary antibodies (2 h, RT), washed thrice in 0.3% PBT (15 min each, RT), stained with DAPI in 0.3% PBT (1:2000, 10 min, RT), and washed once with PBS (10 min, RT). The samples were mounted in VECTASHIELD (Vector Laboratories H-1000). Confocal images were obtained on Leica SPE TCS confocal laser-scanning microscope.

Antibodies used: rabbit anti-phospho-Drosophila Akt Ser505 (Cell Signaling Technology #4054, 1:300), rabbit anti-Repo (gift from Angela Giangrande, 1:500), rabbit anti-cleaved Drosophila Dcp-1 Asp216 (Cell Signaling Technology #9578, 1:200), goat anti-rabbit Alexa Fluor 488 (Thermo Fisher Scientific #A11034, 1:500) and goat anti-rabbit Alexa Fluor 633 (Thermo Fisher Scientific #A21070, 1:500). All antibody dilutions were made in 2% NDS in 0.3% PBT.

Quantification and statistical analysis

The area between veins L3-L4 and area of crossveins from adult wings of females were measured using ImageJ software. The area of Gal4 clones was calculated by dividing the GFP-positive area by DAPI area for individual discs using ImageJ software. Error bars in bar plots represent mean ± standard deviation. The student’s t-test (two-tailed) was used to test for significance; *** indicates a p-value <0.001. n for experiments is indicated in corresponding figure legends.

This work was supported by grants from the Swiss National Science Foundation (SNF 31003A_166680) and the Swiss Cancer League (KLS-3407-02-2014) to HS.

We are grateful to the VDRC and BDSC for flies; Ryohei Yagi for advice regarding the cloning of LexO-PtenRi and sharing useful plasmids and fly lines; Angela Giangrande for sharing the anti-Repo antibody, and Igor Vuillez for technical assistance.

Not applicable.