Mutation in the Drosophila melanogaster adenosine receptor gene selectively decreases the mosaic hyperplastic epithelial outgrowth rates in wts or dco heterozygous flies

Abstract

Adenosine (Ado) is a ubiquitous metabolite that plays a prominent role as a paracrine homeostatic signal of metabolic imbalance within tissues. It quickly responds to various stress stimuli by adjusting energy metabolism and influencing cell growth and survival. Ado is also released by dead or dying cells and is present at significant concentrations in solid tumors. Ado signaling is mediated by Ado receptors (AdoR) and proteins modulating its concentration, including nucleoside transporters and Ado deaminases. We examined the impact of genetic manipulations of three Drosophila genes involved in Ado signaling on the incidence of somatic mosaic clones formed by the loss of heterozygosity (LOH) of tumor suppressor and marker genes. We show here that genetic manipulations with the AdoR, equilibrative nucleoside transporter 2 (Ent2), and Ado deaminase growth factor-A (Adgf-A) cause dramatic changes in the frequency of hyperplastic outgrowth clones formed by LOH of the warts (wts) tumor suppressor, while they have almost no effect on control yellow (y) clones. In addition, the effect of AdoR is dose-sensitive and its overexpression leads to the increase in wts hyperplastic epithelial outgrowth rates. Consistently, the frequency of mosaic hyperplastic outgrowth clones generated by the LOH of another tumor suppressor, discs overgrown (dco), belonging to the wts signaling pathway is also dependent on AdoR. Our results provide interesting insight into the maintenance of tissue homeostasis at a cellular level.

Electronic supplementary material

The online version of this article (doi:10.1007/s11302-014-9435-2) contains supplementary material, which is available to authorized users.

Introduction

Adenosine (Ado) is a ubiquitous metabolite that plays a prominent role as a paracrine homeostatic signal of metabolic imbalance within tissues [1]. Ado can be formed extracellularly by the dephosphorylation of adenine nucleotides that are released from stressed or damaged cells [2]. Ado is also formed intracellularly from ATP and transported to the extracellular space via bidirectional nucleoside transporters [2]. The adenosine gradient then acts in neighboring cells, allowing them to deal more effectively with potentially stressful events [1].

While there is a lot of information on the effects of Ado and various metabolically stable adenosine receptor (AdoR) analogs at the organismal and organ levels, our knowledge of adenosine signaling in maintaining tissue homeostasis at a cellular level remains poor. Recent studies have shown that Ado transport, metabolism, and signaling are all parts of the Ado homeostatic signaling, which plays a more important biological role in diverse organisms than previously thought. For example, the local concentration of extracellular Ado was shown to provide a signal for the localization of hematopoietic progenitor cells in the cortical zone of the Drosophila lymph gland [3] or to regulate the regeneration of pancreatic beta cells in zebrafish and mice in vivo [4].

AdoRs, which belong to the G-protein-coupled receptor family [5] mediate most of the physiological effects of extracellular adenosine in vertebrates and invertebrates. Four distinct AdoR isoforms were identified in mammals. Two of them are positively coupled to adenylate cyclase (A2A and A2B) and the other two (A1 and A3) negatively regulate the enzyme. Drosophila has a single AdoR, which is positively coupled to adenylate cyclase [6]. Its sequence shows about 38 % identity with the N-terminal region of 350 amino acids of the A2A human homologue [7]. Drosophila AdoR mutants are viable and show no obvious phenotypic changes [7, 8].

Several other conserved gene products are involved in adenosine signaling and modulate the extracellular concentration of adenosine, including nucleoside transporters and Ado deaminases [9, 10]. Our previous research showed genetic interactions between them including compensatory changes in the expression of the equilibrative nucleoside transporter 2 (Ent2) in the AdoR mutants and adenosine deaminase growth factor A (Adgf-A) in the Ent2 mutants [11]. The Adgf-A mutant is homozygously lethal, causing elevated levels of extracellular Ado [12, 13]. Null mutants in Ent2 are also homozygously lethal, while Ent2 hypomorphs show similar defects in synaptic transmission and associative learning as the AdoR¹ mutant [11].

Since Ado exerts a broad range of cytoprotective, growth-promoting, and immunosuppressive effects and was observed at a high concentration in a number of human tumors, it was suggested to be an important factor regulating tumor growth [14–17]. In this report, we examined the effect of three Drosophila genes involved in Ado signaling on the incidence of somatic mosaic clones, including AdoR, Ent2, and Adgf-A. We show that genetic manipulations with these genes cause dramatic changes in the frequency of hyperplastic outgrowth clones by the loss of heterozygosity (LOH) of the warts (wts) tumor suppressor gene, while they have no or very little effect on control clones. The Drosophila wts clones are formed in the epithelium of imaginal discs, neural epithelium, and histoblast nests [18, 19]. The wts homologue in humans is called LATS1 and its downregulation has been reported in a number of tumors [20]. This report is the first to demonstrate how changes in Ado signaling influence the frequency of the wts mosaic hyperplastic epithelial outgrowth clones in Drosophila.

Materials and methods

Fly stocks

The AdoR¹ mutant is viable and fertile and was generated earlier by homologous recombination [12]. We prepared an AdoR¹ stock isogenized to white¹¹¹⁸ flies by serial backcrosses.

Two additional AdoR alleles were received from public stock collections. The AdoR deletion mutant designated AdoR^KG0396ex was generated by Wu et al. [8]. The mutation-designated AdoR^{Df(3R)Exel6214}, which contains a larger deletion located in 3R (99D5-99E2), encompasses approximately 17 genes including the AdoR locus and was from the Exelixis collection (Harvard University, USA). The Drosophila RNA interference strains carrying upstream activation sequence (UAS -AdoR^RNAi-VDRC, line number 1385 and UAS-Ent2^RNAi-VDRC, line number 7618) hairpins were received from the Vienna Drosophila RNA interference (RNAi) Center (VDRC). The UAS-AdoR^RNAi-TRIP (TRIP, Bloomington stock BL-27536) flies were received from the Transgenic RNAi Project of Harvard Medical School. The wts^x1 mutant was received from Dr. Tian Xu (Yale University). The UAS-AdoR, UAS-Ent2, and UAS-ADGF-A strains were constructed earlier[7, 11]. All other Drosophila strains, including Act-Gal4, Tub-Gal80, en-Gal4, y, discs overgrown (dco³) were obtained from the Bloomington Drosophila stock center.

Drug treatment

For testing cisplatin and 2,6-diaminopurine (DAP), 10 female and 4 male flies were placed into standard rearing vials for mating and egg laying. After 36–40 h, the parents were transferred to fresh medium. The first instar larvae were treated by adding 0.3 ml of the solutions or suspensions of tested substances onto the surface of the food. Cisplatin, cis-diaminedichloroplatinum (Sigma-Aldrich) was used in a 0.2 mg/ml aqueous (0.67 mM) solution unless otherwise specified. The DAP assays (Sigma-Aldrich) were performed using 12–80 mM (toxicity assay) and 40 mM (Somatic Mutation and Recombination test, SMART) concentrations in 5 % dimethyl sulfoxide, (DMSO, Sigma-Aldrich). Because of its low solubility, DAP was added as a suspension to larval food. The toxicity of paraquat (Sigma-Aldrich, 20 mM solution in 5 % DMSO) was tested on adults for 3 days (males, 24 h after eclosion). Control flies were treated with the equivalent concentration of DMSO. Lethality was expressed in the percentage of surviving flies (with respect to the original number of embryos taken to the experiment).

SMART and clonal analysis

SMART, as described by Stern [21], is based on LOH and mitotic crossing-over. We used a variation of SMART based on flies heterozygous for marker genes (y/+ or wts^x1/+), in which the mosaic clones can be detected over the entire body surface of the adults and more than 70 % of the wts hyperplastic outgrowths are found in larval wing imaginal discs or in adult wings, notum, and sternopleura [22]. We examined either spontaneous or mutagen (cisplatin, diaminopurine, paraquat)-induced somatic recombination and compared the frequencies of clones that appeared in the examined genetic backgrounds (AdoR⁺, AdoR¹ or AdoR⁺/AdoR¹). Cisplatin showed a strong statistically significant recombinogenic effect (P < 0.01) at the doses that had low toxicity (Fig. S1A). We therefore chose cisplatin to induce mitotic recombination in most of our experiments.

The wts^x1 clones were classified into four size categories as follows: (1) “Smallest detectable”—The clones of the minimal detectable size (∼20 cells, based on the wing hair counting), (2) “Small”—Readily detectable, 4–5 times bigger than the smallest (∼100 cells), (3) “Average”—3–10 times bigger than the small (up to one third of the total organ size), and (4) “Large”—2–3 times bigger than the average (up to the total size of the organ).

In order to label the clones with green fluorescent protein (GFP) and/or to express various transgenes, we used the mosaic analysis with a repressible cell marker (MARCM) system [23], based on the loss of Gal80 expression in clones and subsequent derepression of Gal4-dependent transcription (Figs. 1b–f and S4). For the MARCM expression in wts^x1 or AdoR¹ clones, we used the Gal80 transgene localized in the cytological position 84B, in the chromosomal arm 3R. In wts⁺Tub-Gal80/wts^x1 + heterozygous flies, the mosaic wts^x1 clones lose the repressor and express UAS constructs driven by the Act-Gal4 driver. The following UAS constructs were used: UAS-GFP, UAS-Adgf-A, UAS-Ent2, UAS-Ent2^RNAi, UAS-AdoR, and UAS-AdoR^RNAi.

Fig. 1.

Examples of somatic mosaic clones in adult flies and larval imaginal discs; a–c adult flies, d–f imaginal discs. The clones are indicated by arrows. a yellow (black arrows point to yellow bristles) and wts ^x1 clones (red arrow) on the thorax of a y/+; wts ^x1 AdoR ¹/+ + imago. b A large wts ^x1 GFP clone (white arrow) in the wing (Act-Gal4 UAS-GFP/+ +; + wts ^x1/Tub-Gal80 +); c non-hyperplastic GFP clone (white arrow) in proboscis labeled using the MARCM system (Act-Gal4 UAS-GFP/+ +; +/Tub-Gal80). d UAS-AdoR ^RNAi-VDRC UAS-GFP clone (white arrow) in the imaginal disc of third instar larva Act-Gal4 UAS-GFP/+; AdoR ^RNAi-VDRC/Tub-Gal80 genotype; e wts ^x1 clone (white arrow) labeled with GFP using the MARCM system in the imaginal disc of an Act-Gal4 UAS-GFP/+ +; + wts ^x1/Tub-Gal80 + third instar larva. f GFP-labeled clone (white arrow) using the MARCM system in the imaginal disc of third instar larva (Act-Gal4 UAS-GFP/+ +; +/Tub-Gal80); d–f Nuclei are marked by DAPI stain (blue)

Adult flies were screened for hyperplastic outgrowths or color markers under a binocular microscope. The GFP clones were detected with a fluorescent microscope in dissected imaginal discs from wandering stage third instar larvae.

Quantitative PCR

Total RNA was extracted from 60 Drosophila adults using the RiboZol RNA extraction reagent (AMRESCO) followed by NucleoSpin RNA II kit (Macherey-Nagel) including an on-column rDNase I digestion step. Total RNA (1000 ng) was reverse transcribed using PrimeScript Reverse Transcriptase (Takara) with oligo(dT)₁₇ primer. Quantitative PCR was performed using the HOT FIREPol EvaGreen qPCR Mix Plus (Solis BioDyne). Primer sequences are listed in Table S1. RT-qPCR was performed in triplicates and all results are presented with means and SEM from four independent biological samples. We used rp49 and Rack1 messenger RNA (mRNA) levels for normalization.

Statistical data evaluation

Clone frequency p (%) was calculated as (number of clones)/(total number of flies) × 100. Clone frequencies were compared using Student’s t test and Fisher’s correction ϕ = 2 × arcsin(√p) for frequency. The t test was calculated for confidence intervals (CI) for P < 0.05 and P < 0.01, if appropriate. These intervals are shown as error bars in the figures unless stated otherwise. Low-sample t-criterion correction was applied to samples of fewer than 500 flies, according to Urbakh [24]. We applied the χ² criterion to assess the lethality of various heteroallelic AdoR mutant combinations.

Results

AdoR mutant flies are homozygous viable

We reported earlier that the homozygous AdoR¹ mutation generated by homologous recombination was viable and able to partially rescue the lethality of adenosine deaminase growth factor-A (Adgf-A) mutants [12]. The mutation AdoR^KG0396ex lacking the entire AdoR coding region was also reported to be homozygous viable [8]. In contrast, the survival of AdoR^RNAi-VDRC flies with long hairpin constructs was dramatically reduced when driven by the Act-Gal4 driver. We conclude that the lethality of the AdoR^RNAi-VDRC construct was presumably caused by off-target effects. The larger deletion AdoR^{Df(3R)Exel6214} was homozygous lethal in our experiments.

To verify that there is no negative effect of AdoR mutations on fly viability, we prepared heteroallelic combinations of various AdoR mutations. As shown in Table S2, the frequencies among the offspring of all examined combinations, including AdoR¹/AdoR¹, AdoR¹/AdoR^KG03964ex, AdoR^KG03964ex/AdoR^{Df(3R)Exel6214}, and AdoR¹/AdoR^{Df(3R)Exel6214}, were in accordance with the expected Mendelian ratios, showing no significant deviation from the expected frequencies. Our results confirmed that the AdoR¹ and AdoR^KG03964ex mutations have no negative effect on fly viability under standard rearing conditions.

AdoR¹ strongly reduces wts^x1 clone frequency in SMART

In order to examine the effect of the loss of AdoR function on the incidence of somatic mosaic clones in flies in vivo, we adapted a robust test, originally designed for assaying the mutagenicity of chemicals, or the effect of genetic background on the frequency of mutations [25, 26]. The mutations are visualized as somatic mosaic clones generated by LOH of the recessive allele of the marker gene. In our experiments, it was advantageous to induce clone formation with mutagens since the spontaneous LOH occurred at a relatively low frequency. In this way, we were able to reduce the number of flies in the experiments and still reach statistical significance.

First, we compared the effect of AdoR¹ and AdoR⁺ backgrounds on the rate of somatic mosaic clones labeled with two different types of markers, wts^x1 and y. The wts^x1 is a mutant allele of the wts tumor suppressor causing epidermal tissue overgrowth that is visible as small hyperplastic outgrowths, whereas y causes yellow (y) cuticles and bristles in the adult (Fig. 1a, b). The mosaic clones were induced with cisplatin as described in “Experimental Procedures”. The results are shown in Fig. 2a. The frequency of wts^x1 clones in AdoR¹ flies was much lower than in AdoR⁺ flies; only 4.5 % compared to 65.9 %, respectively. In contrast, y mosaic clones occurred at a frequency of 11.1 % in AdoR⁺ flies, whereas in AdoR¹ flies (y/+; AdoR¹wts^x1/AdoR¹ +) it reached 18.5 % (similarly for y/+; AdoR¹/AdoR¹ flies). The higher frequency of y mosaic clones might be connected to the involvement of AdoR in the control of DNA repair or detoxification. The dramatically decreased wts^x1 hyperplastic outgrowth clone frequency in AdoR¹ flies was unexpected.

Fig. 2.

The effect of AdoR on the frequency of somatic mosaic clones. a The incidence of wts ^x1 and y somatic mosaic clones in AdoR ⁺ and AdoR ¹ flies. The gray and empty bars represent spontaneous and 0.67 mM cisplatin-induced wts ^x1 and y mosaic clone frequencies, respectively, in homozygous AdoR ⁺ (left side, y/+; wts ^x1/+) and AdoR ¹ (right side, y/+; + AdoR ¹/+ AdoR ¹) flies. Error bars show 95 % confidence interval (CI). The numbers above each bar indicate the number of flies tested. Most wts ^x1 clones appeared in adult wings, notum, and sternopleura. The yellow clones appear in lower frequency since they must cover a bristle precursor cell for being detected. b The wts ^x1 clone frequency induced by three chemicals in AdoR ⁺ and AdoR ¹ flies. The wts ^x1 clones were induced by 0.67 mM cisplatin (CP), 20 mM paraquat (PQ), and 40 mM diaminopurine (DAP); the solvents (marked in brackets) were water or 5 % aqueous DMSO; CTR control. White and gray bars represent clone frequency in +AdoR ⁺/wts ^x1 AdoR ⁺ and wts ^x1 AdoR ¹/+ AdoR ¹ flies, respectively. Error bars show 95 % CI. The numbers above the right of each bar indicate the number of flies tested. c wts ^x1 clone frequencies in cis- and trans-AdoR ¹ wts ^x1 heterozygous flies. The black bars (left) represent clone frequencies observed in cis- and trans- AdoR ¹ wts ^x1 heterozygous flies. For comparison, wts ^x1 clone frequencies observed in homozygous AdoR ⁺ and AdoR ¹ flies are shown on the right (white and gray bars, resp.). The wts ^x1 clones were induced by 0.67 mM cisplatin. Error bars show 95 % CI. The numbers above each bar indicate the number of flies tested

To show that the observed difference in the incidence of mosaic clones carrying wts^x1 and y markers was not dependent on cisplatin induction, we examined the frequency of spontaneous mosaic clones for both markers. The spontaneous level of wts^x1 clones also dropped by sevenfold in AdoR¹ homozygous flies, compared to the AdoR⁺ controls (from 6.7 to 0.9 %, respectively) showing that the reduction of wts^x1 clone frequency was not caused by cisplatin. The spontaneous frequencies of y/y clones were slightly below the test sensitivity in both fly strains (Fig. 2a).

To further prove that the drop of wts^x1 clone frequency in AdoR¹ flies was not caused by cisplatin, we used two other genotoxic compounds for SMART, diaminopurine (DAP), and paraquat. As expected, both DAP and paraquat produced higher clonal rates in both AdoR⁺ and AdoR¹ flies than the spontaneous frequency. Again, the frequencies of wts^x1 mosaic clones induced by DAP and paraquat in the AdoR¹ homozygous flies (1.8 and 3.3 %, respectively) were 7–2.7-fold lower compared to those in the corresponding AdoR⁺ flies (12.6 and 8.8 %, respectively). The results are shown in Fig. 2b. The chemicals were also assayed for toxicity. The results showed that DAP and paraquat have a much higher impact on fly viability than cisplatin (Fig. S1).

These results suggested that the AdoR¹ mutation had different effects on the frequency of hyperplastic outgrowth and non-tumor mosaic clones represented by the wts^x1 and y markers, respectively. The effect was independent of the mutagen used and also occurred spontaneously. We did not observe any differences in wts^x1 clone morphology when comparing AdoR⁺ and AdoR¹ flies. The differences in SMART results can be expected because chemicals differ in their genotoxicity, toxicity, and which pathways they activate in different target tissues. The dramatically decreased wts^x1 hyperplastic outgrowth frequency in AdoR¹ flies therefore called for further investigation.

Different wts^x1 clone frequencies in cis- and trans-AdoR wts heterozygotes suggest the need of AdoR⁺ after clone formation

Since both AdoR and wts are localized on the distal tip of the 3R chromosomal arm, mitotic crossing-over break points most frequently occur between the centromere and both loci. To find out whether the AdoR¹ mutation affects wts^x1 clone formation by influencing processes such as DNA repair or affects the survival of already established clones, we compared the frequencies of wts^x1 hyperplastic outgrowths in cis- and trans-heterozygotes for AdoR¹ and wts^x1, which produce qualitatively different wts^x1 clones. The cells in cis- and trans-heterozygotes have almost the same genetic background and conditions; if the AdoR¹ affects wts^x1 clone formation we should observe similar hyperplastic outgrowth frequencies in cis- and trans-heterozygotes. Alternatively, if the AdoR¹ influences wts^x1 hyperplastic growth or survival, we should observe different wts^x1 clone frequencies because the mitotic recombination produces different clones: in the cis-heterozygous flies for these genes, the wts^x1 clones lack functional AdoR, whereas in the trans-heterozygous flies, they contain homozygous AdoR⁺ (Fig. S2). The comparison of wts^x1 clone frequency in cis- and trans-heterozygotes for wts^x1 and AdoR¹ therefore may provide important information about whether AdoR-wts^x1 interactions affect clonal formation or survival. The resulting data are shown in Fig. 2c. The wts^x1 mosaic clones containing AdoR⁺ in trans-heterozygotes occurred at high frequency (91.9 %), whereas their rate in cis-heterozygotes lacking functional AdoR was very low (4.7 %). Such a difference strongly suggests that functional AdoR is needed for wts^x1 hyperplastic clone survival/growth after clone formation.

The low frequency of wts^x1/wts^x1 clones formed in cis-heterozygotes also means there was no rescue of wts^x1 hyperplastic outgrowth clones by the surrounding AdoR¹/AdoR⁺ cells. Taking this result into account, the effect of AdoR¹ appeared to be cell autonomous. This provided the opportunity for further experiments, which could use the overexpression of the studied genes within somatic clones only.

To better understand the impact of AdoR¹ on the wts^x1 cells, we also assessed clone sizes. As shown in Fig. 3, we observed significantly higher proportion of larger wts^x1 hyperplastic outgrowths, among the AdoR¹wts^x1/++ flies compared to wts^x1/+ flies.

Fig. 3.

The size distribution of wts ^x1 hyperplastic outgrowth size classes in two Drosophila genotypes. We analyzed 148 spontaneous wts ^x1 hyperplastic outgrowths detected among 1574 wts ^x1/+ flies and 35 spontaneous wts ^x1 AdoR ¹ hyperplastic outgrowths found among 2267 wts ^x1 AdoR ¹/++ flies. The differences between the average and large clones in wts ^x1/+ and wts ^x1 AdoR ¹/++ flies are statistically significant

Confirmation of the specificity of AdoR effect on the frequency of wts^x1 hyperplastic outgrowth clones

To confirm that the observed decrease of the wts^x1 hyperplastic outgrowth clone rate was specific for the loss of the function of AdoR, we used AdoR silencing through the expression of UAS-AdoR^RNAi hairpin RNA.

Expression of the UAS-AdoR^RNAi-TRIP hairpin in the flies using Act-Gal4 led to the decrease of AdoR mRNA level to 33 % of its wild-type level, as shown by qPCR (Fig. S3). These flies had a significantly decreased frequency of wts^x1 clones by 40.0 % (95 % confidence interval, CI ±0.4 %) compared to control flies without AdoR silencing. This result supports the hypothesis stipulating the requirement of functional AdoR in the wts^x1 clones for their high frequency. The weaker effect of RNAi-silencing as compared to the AdoR¹ mutation can be explained by the incomplete downregulation of AdoR.

The AdoR knockdown experiments were also performed with flies expressing the UAS-AdoR^RNAi-VDRC construct together with UAS-GFP. Since the global expression of UAS-AdoR^RNAi-VDRC caused lethality, we used the MARCM system (see “Materials and methods” and Fig. S4) and targeted its expression just to wts^x1 clones. The frequency of GFP AdoR^RNAi-VDRCwts^x1 clones in adults was about threefold lower than in the control flies without UAS-AdoR^RNAi-VDRC (Fig. 4a). Consistently, the frequency of wts^x1AdoR^RNAi-VDRC clones assayed in larval imaginal discs also showed similar decrease (but these small samples lack statistical power) (Fig. S5). The rate of control MARCM-labeled GFP clones in imaginal discs of AdoR⁺ larvae with or without UAS-AdoR^RNAi-VDRC was statistically indistinguishable (27.3, 95 % CI ±8.5 %, compared to 20.0, 95 % CI ±9.3 %) (Fig. S5). These results support the notion that the loss of AdoR function decreases the frequency of wts^x1 clones and exerts no or very little influence on the frequency of control GFP clones.

Fig. 4.

The frequencies of mosaic clones in various genetic backgrounds. a The incidence of wts ^x1 GFP clones in adult AdoR ^RNAi and control flies. The UAS transgenes were expressed by MARCM system. The left bar (AdoR+) y w; act-Gal4 UAS-GFP/+ +; wts ^x1/tub-Gal80; the right bar (AdoR ^RNAi) y w; act-Gal4 UAS-GFP/+ +; UAS-AdoR ^RNAi-VDRC wts ^x1/tub-Gal80. Error bars show 95 % CI. The numbers above each bar indicate the number of flies tested. b AdoR and Ent2 overexpression and Ent2 silencing affect the frequency of wts ^x1 clones. The wts ^x1 clone frequencies, in flies with UAS-AdoR ⁺ (black bar, left), UAS-Ent2 ⁺ (left gray bar) and UAS-Ent2 ^RNAi-VDRC (right gray bar) constructs expressed using MARCM system (see the text). For comparison, wts ^x1 clone frequency observed in homozygous AdoR ⁺ flies is shown on the left (white bar). Error bars show 95 % CI. The numbers above each bar indicate the number of flies tested. c The incidence of dco ³ and control mosaic clones in AdoR ⁺ and AdoR ¹ flies. Black bars (left) the frequency of dco ³ clones in heterozygous dco ⁺/dco ³ flies (Act-Gal4 UAS-GFP/+ +; + AdoR ¹ dco ³/Tub-Gal80 + + and Act-Gal4 UAS-GFP/+ +; + dco ³/Tub-Gal80 +); gray bars (right) control GFP clones labeled by MARCM system in AdoR ¹ and AdoR ⁺ flies (Act-Gal4 UAS-GFP/+ +; + AdoR ¹/Tub-Gal80+ and Act-Gal4 UAS-GFP/+ +; +/Tub-Gal80). Error bars show 95 % CI. The numbers above each bar indicate the number of flies tested. d The ectopic expression of adenosine deaminase ADGF-A rescues wts ^x1 AdoR ¹ clones in UAS-ADGF-A/en-Gal4; wts ^x1 AdoR ¹/++ flies. The en-Gal4 was used to drive UAS-ADGF-A ⁺ expression in the posterior wing (gray shading in the schematic wing drawing above the graph indicates the en expression pattern). The hyperplastic outgrowth clone rates in the anterior and posterior compartments of 1518 wings examined were 0.2 ± 0.1 (left empty bar) and 2.7 ± 0.1 (right gray bar), respectively

To further test the specificity of the effect of AdoR level on the rate of wts^x1 clones, we used AdoR overexpression. Because the ubiquitous overexpression of UAS-AdoR in the entire body was lethal [9], we expressed it locally in the wts^x1 mosaic hyperplastic clones using the MARCM system (Fig. S4). We constructed UAS-AdoR/Act-Gal4 UAS-GFP;+ Tub-Gal80/wts^x1+ flies carrying the UAS-AdoR and Act-Gal4 transgenes and compared their wts^x1 clone frequency with that of the control flies lacking UAS-AdoR (Fig. 4b). Such AdoR overexpression increased the wts^x1 hyperplastic outgrowth clone frequency to 75.4 compared to 66.0 % in AdoR⁺ flies without UAS-AdoR (significant at P < 0.05).

Taken together, these data confirms the specific requirement of AdoR for wts^x1 hyperplastic outgrowth occurrence and also revealed a dose effect: AdoR overexpression further increased the frequency of wts^x1 clones while, conversely, incomplete RNAi-silencing of AdoR was not as effective as the null mutation.

AdoR¹ mutation also affects the frequency of dco³ hyperplastic outgrowth clones

To further confirm the effect of AdoR¹ on hyperplastic outgrowth clone frequency we tested a mutation in another Drosophila tumor suppressor gene discs overgrown (dco³) under the same conditions. Since the morphological outgrowths of dco³ clones are more difficult to detect, we also labeled the clones with GFP using the MARCM system (Act-Gal4 UAS-GFP/+ +; + dco³/Tub-Gal80 +).

As shown in Fig. 4c, the frequency of dco³ mosaic hyperplastic outgrowths in heterozygous dco³/+ flies with AdoR⁺ and UAS-GFP was about 12.1 % and this rate dropped by approximately a factor of four to 3.5 % in the AdoR¹dco³/+ + cis-heterozygotes (Act-Gal4 UAS-GFP/+ +; + AdoR¹dco³/Tub-Gal80 + +). This difference was statistically highly significant (P < 0.01). In contrast, in the control experiment, the frequency of GFP clones was 1.72 % in the AdoR¹ mutant flies (Act-Gal4 UAS-GFP/+ +; + AdoR¹ /Tub-Gal80 +), which is comparable to the 1.1 % value observed in flies that were wild type for AdoR (Act-Gal4 UAS-GFP/+ +; +/Tub-Gal80); the difference was statistically insignificant.

Our results showed that the AdoR¹ mutation dramatically decreased the frequency of hyperplastic outgrowth clones induced by LOH of another tumor suppressor gene dco, suggesting that more types of hyperplastic outgrowth clones may require functional AdoR for their growth. The extent of this effect was, however, lower than that observed in the previous experiments with wts^x1 clones, probably reflecting the different growth properties of these hyperplastic outgrowth types. The frequency of the control GFP-labeled non-tumor clones did not decrease in AdoR¹ flies in any of the experiments.

The effect of Ent2 and Adgf-A on wts^x1 clone frequency

To find out whether other members of the Ado signaling pathway influence wts^x1 hyperplastic outgrowth frequency, we overexpressed Ent2 and Adgf-A. In addition, we silenced the transporter expression by using UAS-Ent2^RNAi.

MARCM expression of UAS-Ent2 in the wts^x1 clones increased the induced mosaic hyperplastic outgrowth rate from 65.8 to 91.6 % (P < 0.01) (Fig. 4b). Conversely, the knockdown of Ent2 expression by RNAi in the UAS- Ent2^RNAi-VDRC ++/+ Act-Gal4 UAS-GFP; wts^x1+/+Tub-Gal80 flies decreased the frequency of wts^x1 hyperplastic outgrowths to 0.9 %. These results showed that both ectopic expression and silencing of the Ent2 transporter exerted an even stronger effect on wts^x1 hyperplastic outgrowth frequency than the changes in the level of the Ado receptor.

We also expressed UAS-Adgf-A in AdoR¹wts^x1/+ + mutants using an engrailed-Gal4 driver and took advantage of the distinct ectopic expression of the UAS construct in wts^x1 hyperplastic outgrowths in the anterior and posterior compartments of the wing (Fig. 4d). There was a statistically significant increase (13.5 times) of the wts^x1 hyperplastic outgrowth rate in the posterior compartment compared to the intact anterior one—from 0.2 to 2.7 % (P < 0.01). The ratio is quite comparable to the 14-fold difference observed in the wts^x1 hyperplastic outgrowth rates between flies with AdoR⁺ and AdoR¹ homozygous backgrounds (Fig. 2a) suggesting that the overexpression of Adgf-A completely rescues the effect of the AdoR¹ mutation in wts^x1 clones. Taken together, our results suggest that the wts^x1 hyperplastic outgrowth clone frequency is regulated by adenosine signaling.

Discussion

Fly genetic methods allow the study of interactions between cells in the mosaic clones and their microenvironment. Our results show that AdoR and extracellular adenosine play key roles in the growth control of mosaic hyperplastic outgrowth clones. This conclusion follows from the observations that the frequency of wts^x1 and dco³ hyperplastic outgrowth clones in the AdoR¹ mutant background is dramatically reduced compared to that in the AdoR⁺ flies. In contrast, the rate of non-hyperplastic clones (yellow or GFP) did not show any reduction, and even showed a significant increase for y. To exclude the possible effect of different recombination frequency on different chromosomes we used yellow on chromosome X and Gal 80 on chromosome 3 (segregation of Gal 80 permits the UAS-GFP expression on chromosome 2).

AdoR seems to influence the SMART results by more than one mechanism. The slightly higher proportion of y mosaic clones in AdoR¹ flies compared to that in the AdoR⁺ ones suggests that the loss of functional AdoR may slightly increase the frequency of LOH. It was reported earlier that adenosine receptor agonists reduced the level of mutations in mouse and human cells treated with alkylating agents or H₂O₂ [27]. The elevated level of y clones in AdoR¹ flies might suggest that the loss of such protective AdoR function slightly increases the rate of somatic mutations.

We found that the lack of functional AdoR causes a dramatic reduction of wts^x1 mosaic hyperplastic outgrowth clone frequency. Our data strongly suggest that AdoR is not needed for wts^x1 clone formation, but rather for their growth/survival. The reduction of wts^x1 clonal rate in AdoR¹ or AdoR^RNAi flies was consistent for clone established by spontaneous mosaicism, as well as for all chemical treatments, but its extent was different for different SMART conditions and also depended on the expression level of AdoR.

Moreover, AdoR interacts with dco, another tumor suppressor gene, which is also a member of the wts signaling pathway. Both the wts^x1 and dco³ mosaic clones form fast- growing hyperplastic outgrowths. The dco is Drosophila casein kinase Iε/δ and affects the upstream regulator of wts called fat [28, 29]. Wts is a serine/threonine protein kinase belonging to the WTS/LATS tumor suppressor family, which suppresses growth by antagonizing the function of the transcriptional co-activator protein Yorkie (Yki) [30]. Mutations in wts causing elevated levels of Yki activity can convert wild type (WT) cells into faster growing supercompetitors that are able to eliminate adjacent WT cells [31, 32]. Our results suggest that the wts^x1 and dco³ mosaic hyperplastic outgrowths lose their growth advantage in the absence of functional AdoR.

Interestingly, changes in the expression of Ent2 equilibrative nucleoside transporter in wts^x1 clones have a similar cell autonomous effect as changes in AdoR expression. The overexpression of Ent2 in wts^x1 clones leads to their higher frequency, whereas Ent2 silencing dramatically decreases wts^x1 clonal rate (Fig. 4a). The normal frequency of wts^x1 clones therefore requires WT function of both AdoR and Ent2, suggesting some form of interaction between them. We previously observed that AdoR¹ and Ent2³ mutant flies have similar defects in associative learning and synaptic function [11]. We even observed some compensatory changes including a dramatic increase in Ent2 mRNA in AdoR¹ mutants or a reduction of adenosine deaminase transcript levels in Ent2 mutants [11]. Although we cannot directly measure local extracellular Ado concentrations, we know that Ado is critical for clone survival since the clonal rate can be rescued by overexpressing the extracellular adenosine deaminase Adgf-A in AdoR¹wts^x1/++ mutants (Fig. 4d). The growth restrictions probably appear early in the clonal development. Significant differences in the distribution of wts^x1 hyperplastic outgrowth sizes between wts^x1AdoR¹/++ and wts^x1/+ flies suggest that larger wts^x1AdoR¹ clones have lower probability for being eliminated than the smaller ones. Alternative explanation that the fewer clones in wts^x1AdoR¹/++ flies exist because the clones have fused is not likely since the overall frequency of wts^x1 clones was too low as well as the proportion of flies having more than one wts^x1 hyperplastic outgrowth.

Earlier, microarray analyses of gene expression in tissue culture cells showed that most of the common Drosophila cell lines, including imaginal disc cells Cl.8+, neuroblasts Bg2-c2, or hematopoietic Mbn2 cells seem to have the Hippo/Warts pathway inactivated [33]. Interestingly, the Cl.8+ imaginal disc cell line also has a very low level of functional AdoR [6]. Our previous tissue culture experiments showed that there is an intriguing connection between the AdoR and Ent2 functions with cell survival. The exposure of Cl.8+ cells to extracellular Ado causes a concentration-dependent block of cell division and apoptosis associated with a loss of mitochondrial membrane potential and excessive ATP production [34]. In contrast, other cell types containing functional AdoR, including Bg2-c2 and Mbn2 cells, are able to decrease Ado uptake upon Ado treatment and avoid apoptosis following adenosine stimulation [34, 35]. Cl.8+ cell death could be prevented by blocking adenosine transport or increasing extracellular adenosine deaminase activity but not by blocking AdoR [34]. These results showed that Ado can induce receptor-independent cell death or block of cell growth.

Our earlier experiments showed that global AdoR overexpression in flies causes lethality [9]. Microinjection of Act-Gal4 and UAS-AdoR flies with the AdoR antagonist SCH58261 was able to partially rescue the flies [6]. Similarly, the overexpression of AdoR in Drosophila Cl.8+ and S2 cells in vitro was cytotoxic [6]. These results suggest that the AdoR-mediated signaling can also cause death of Drosophila cells. Strong effects of both the Ado transport and AdoR signaling on cell viability were recently also described in the experiments involving the regeneration of pancreatic β-cells in zebrafish in vivo [4].

The simplest model that would explain our data on the occurrence of wts^x1 clones is shown in Fig. 5. We postulate that the extracellular Ado represents two opposing signals, which need to be balanced in fast-growing cells. Such hypothesis is supported by earlier experiments in mice brain showing that Ado uptake induces ATP synthesis and stimulates catabolic reactions, while the signal mediated by the AdoR signaling leads to the downregulation of a number of transcripts encoding metabolic enzymes [1]. It was also shown earlier that wts⁻/wts⁻ cells behave as supercompetitors and eliminate adjacent wts⁺ cells at clonal boundaries [36–38]. In our experiments, AdoR⁺wts^x1/AdoR⁺wts⁺ flies are characterized by a relatively high frequency of wts^x1 hyperplastic outgrowth clones in which the wts^x1/wts^x1 cells may behave as supercompetitors. The dying wts⁺ cells become rich sources of extracellular Ado (Fig. 5a, black arrows). Ado is taken up by the wts^x1 hyperplastic outgrowth cells and this uptake is compensated by high AdoR stimulation (Fig. 5a). This extracellular Ado effect is indirectly supported by the experiments with the ADGF-A overexpression (Figs. 5d and 4d). In flies carrying AdoR¹ mutation or Ent2^RNAi, and therefore lacking such balance, wts^x1 hyperplastic outgrowths cannot grow in most cases (Fig. 5b, c). However, there are some escaper AdoR¹wts^x1 or Ent2^RNAiwts^x1 clones which probably grow if they reach the critical threshold while providing their own microenvironment. If we overexpress Adgf-A in the posterior wing compartment of AdoR⁺wts⁺/AdoR¹wts^x1 flies using en-Gal4 and UAS-Adgf-A, the wts^x1 hyperplastic outgrowths in this area occur at the high rate suggesting that the depletion of extracellular Ado eliminates the effect of AdoR mutation and rescues the wts^x1 clones (Fig. 5d).

Fig. 5.

Model of adenosine effect on wts ^x1 hyperplastic outgrowth survival. Hyperplastic outgrowth cells are shown in orange color, cells with WT phenotype in light gray, green shapes represent AdoR, and blue two-headed arrows denote Ent2. a In AdoR ⁺ wts ⁺/AdoR ⁺ wts ^x1 flies, the frequency of wts ^x1 mosaic hyperplastic outgrowths is high. The wts ^x1 hyperplastic outgrowth cells may behave as supercompetitors and eliminate adjacent wts ⁺ cells. High level of extracellular Ado released by the outcompeted cells causes elevated AdoR stimulation as well as high Ado uptake in hyperplastic outgrowth cells. b Mutation in AdoR leads to dramatic decrease of wts ^x1 clone frequency, because Ado uptake in those cells is no longer balanced by AdoR signaling. c Silencing of Ent2 in wts ^x1 hyperplastic outgrowth clones using RNAi also leads to dramatic decrease of wts ^x1 clonal rate (unbalanced AdoR signal). d Overexpression of Adgf-A in AdoR ¹ wts ⁺/AdoR ¹ wts ^x1 flies eliminates the effect of AdoR mutation and rescues the wts ^x1 clones (example of this effect in the posterior wing compartment is shown in Fig. 4d)

The wts downregulation was shown to play a key role in the stem cell-mediated regenerative response to tissue damage in the Drosophila intestine [39, 40]. The fast-growing supercompetitive cells with blocked Wts function and activated Yki may cause secondary tissue damage. The observed effects of AdoR and Ent2 on the frequency of cells which lost their wts tumor suppressor might be part of a normal tissue protection mechanism which limits the extent of regenerative growth.

A number of recent studies have examined the involvement of adenosine in tumor-protective mechanisms in mice, namely triggering immunosuppression via the A2A adenosine receptor (A2AR) on the surface of activated immune cells [16, 41]. While this mechanism probably evolved for the protection of normal tissues from collateral damage caused by the overactive immune cells, the cancer tissue would also be protected. The best evidence for this was provided by experiments using WT and A2A receptor knockout (A2AR⁻) mice inoculated with established melanomas. These hyperplastic outgrowths were completely rejected in 60 % of A2AR⁻ hyperplastic outgrowth-bearing mice while there was no rejection in WT mice [42]. The effects of the A2A receptor were T cell autonomous. In contrast, our results in Drosophila showed that the effects of AdoR¹ on wts^x1 and dco³ mosaic hyperplastic outgrowth are cell autonomous. Results in both experimental systems suggest the involvement of Ado in the interactions between hyperplastic outgrowths and their neighboring cells in tissues.

Our report is the first to show the interaction between the adenosine and Hippo/Warts signaling pathways in Drosophila. We believe that further investigation of adenosine signaling and its impact on competitive cell interactions may clarify important aspects regarding the maintenance of tissue homeostasis at a cellular level.