Introduction

Multicellular organisms encounter stress conditions throughout their developmental and adult stages. Damaged cells by stresses and abnormal cell division need to be removed to ensure proper development and homeostasis. Apoptosis, programmed cell death, is an evolutionally conserved process for the removal of unwanted or damaged cells, which involves condensations of chromatins, shrinkage of cells, and activation of caspases that degrade cellular components1,2,3. Apoptosis accompanies compensatory proliferation to restore damaged tissues to the right size and shape4,5, for which intercellular communication between dying and surviving cells is essential6,7,8. The wing imaginal disc of Drosophila melanogaster is an excellent system to study apoptosis and compensatory proliferation, because wing discs exhibit little apoptosis during normal larval development but exhibit extensive apoptosis under stress conditions9,10,11. Wing discs with up to 40–60% of extensive cell death can yield normal adult wings, indicating that regeneration process in wing discs is efficient and robust4,7,9,11,12,13.

Diap1 is one of the most important proteins for cell survival under stress conditions. Diap1 is an E3 ubiquitin ligase that blocks cell death by tagging the caspases with ubiquitin for proteasome-mediated degradation14,15. Under severe stress conditions, the activity and the amount of Diap1 protein is decreased by the binding of pro-apoptotic proteins such as Head involution defective (Hid), Reaper (Rpr) and Grim16,17,18,19,20. Especially, binding of Hid stimulates autoubiquitination of Diap1 that results in degradation of Diap114,20,21. Among these pro-apoptotic genes, rpr is expressed in a pattern most similar to that of dying cells16, and irradiation can activate transcription of rpr in dying cells through p53 binding to an enhancer of the rpr gene22,23. Heterozygous Diap1 flies are more sensitive to damages than wild-type flies, demonstrating that the amount of Diap1 correlates with the extent of cell survival, and the cells enter the apoptotic process when the level of Diap1 falls below the critical point because of pro-apoptotic proteins14,20,24. Signaling pathways such as JAK-STAT and Hippo pathways are involved in controlling the transcriptional rate of Diap125,26,27.

We recently reported that a Drosophila ADAMTS Sona is important for fly development and promotes Wg signaling28. Sona is processed to an active form in both intracellular and extracellular regions, and promotes Wg secretion. In general, ADAMTSs are secreted proteases that function in extracellular matrix (ECM). Six fly ADAMTSs are involved in various processes such as cell migration, organogenesis and cell signaling29,30,31. Similarly, nineteen mammalian ADAMTSs serve diverse roles32. Some are involved in processing ECM proteins, and malfunction of these ADAMTSs causes connective tissue disorder, arthritis, and arthrosclerosis. Other ADAMTSs regulate cell proliferation and cell survival, and their malfunction causes tumor development and metastasis. Despite involvement of ADAMTSs in diverse cellular functions, the underlying mechanisms of these ADAMTSs are still largely unknown.

We report here that sona is required for cell survival. sona is expressed in a patchy pattern in the wing disc, and irradiation coordinately changed transcription of both sona and Diap1 with negative correlation. Cells expressing either sona or Diap1 at a high level did not exhibit cell death, indicating these two types of cells are resistant to cell death. Consistent with their response to irradiation, sona exhibited a positive genetic relationship with Diap1 but negative genetic relationship with rpr. Furthermore, Sona upregulated the level of Cyc D and increased cell proliferation in a cell non-autonomous manner. We propose that the intercellular signaling between the two types of surviving cells, one expressing sona and the other expressing Diap1, plays an important role for cell survival and compensatory proliferation.

Results

Loss of sona results in cell death

We previously reported that expression of sona RNAi driven by various Gal4 lines results in lethality and malformed appendages28. sona-RNAi-1 and -2 lines were generated by using two different regions of the sona cDNA, and these RNAi lines driven by various Gal4 lines exhibit same phenotypes but with varied strengths28. For instance, engrailed (en)-Gal4 > sona RNAi-13–23 (en > sona RNAi-13–23) wings were smaller in the posterior region (Supplementary Fig. S1a,b). The average distance between L3 and L4 veins was only about 70% of the control (n = 10), and anterior cross-vein was absent in 40% of patched (ptc) > sona RNAi-111–4 wings cultured at 18 °C (n = 23) (Fig. 1a–c). Hair density in the L3-L4 region, however, was unchanged (Fig. 1a,b). Thus, the loss of sona caused reduction in cell number but not cell size.

Figure 1
figure 1

Loss of sona causes apoptosis. Genotypes of wing discs and the visualized proteins are indicated at the upper and lower right of confocal images in all figures, respectively. (a–c) control ptc > GFP (a) and ptc > sona RNAi-111–4 (b) wings cultured at 18 °C. Arrows in (a,b) indicate presence and absence of anterior cross-veins, respectively. The regions marked with the black boxes in (a,b) are magnified in a’ and b’. (c) The distance between L3 and L4 veins in a’ and b’ were measured and graphed. Sample numbers are indicated at the top of bars. (d,e) Dorsal cells with CC3 and nuclei are marked with arrows in e’ and e”. (f–h) CC3 signals and pyknotic nuclei at the basal region are marked with arrows. Scale bars: (d,e) 60 μm; (f–h) 40 μm.

We then examined whether cell death is responsible for the reduced cell number in sona RNAi-expressing wings. Indeed, sona RNAi expressed by apterous (ap)-, ptc-, and en-Gal4s increased cell death detected by an antibody generated against the cleaved form of human Caspase 3 (CC3) that indicates fly Dronc activity (Fig. 1e,g; Supplementary Fig. S1c)33,34,35. The affected dorsal domain in ap > sona RNAi-111–4 discs exhibited a high level of CC3, and highly condensed nuclei were present in the basal region (Fig. 1d,e; Supplementary Fig. S1d,e). Highly condensed nuclei were also present in the anterior-posterior boundary of ptc > sona RNAi-111–4 discs (Fig. 1f,g). When p35 that inhibits the activity of effector caspases36 was coexpressed with sona RNAi, the level of CC3 was reduced and the morphology of nuclei became normal although the nuclei still remained at the basal region (Fig. 1h). Therefore, p35 largely rescued the loss of sona phenotypes.

Different kinds of sona clones are generated

To further prove that Sona functions in cell survival, we examined whether sona47 wing discs show any sign of cell death. Because Sona is involved in Wg signaling, we wanted to check whether the role of sona in cell survival is specific to Wg-producing cells or general to cells in the pouch. To this end, we recombined sona47 allele with the wg-LacZ marker wgen-1137, and found that the wing pouch region of homozygous sona47 discs had less number of nuclei compared to that of heterozygous sona47/+ discs at the apical region (compare Fig. 2a,c’). At the basal region, condensed nuclei were detected in the sona47 discs but not in sona47/+ discs (compare Fig. 2b’,d’). Furthermore, Wg-LacZ expression was absent in the DV midline at the apical region and barely visible at the basal region of sona47 discs (compare Fig. 2a” and c”, arrows in b” and d”). These results showed that sona is important for the survival of both wing pouch cells and wg-lacZ cells.

Figure 2
figure 2

Mosaic pattern of sona expression may be responsible for the three types of sona clones. Confocal planes are indicated at the lower left. (a–d) The effects of sona47 mutation on the survival of Wg-lacZ cells and the number of nuclei. Wing discs of heterozygous (a,b) and homozygous sona47 allele (c,d) are used. The loss of wg-lacZ cells is marked by arrows at the apical region in a” compared to c” and at the basal region in d” compared to b”. Large, medium-sized, and small sona13 and sona47 clones are marked with black arrows, blue arrows, and blue arrowheads, respectively. The regions marked with white boxes in e–g are magnified in e’–g’. (h) Expression pattern of GFP in sona > GFP wing discs. The region in the white box (h) is magnified in h”. Scale bars: (ad,g,h) 40 μm; (e,f) 60 μm; (e’) 23.7 μm; (f’) 14.5 μm; (g’) 21 μm; (h”) 7.2 μm.

We then examined sona13 and sona47 clones marked by the absence of GFP in discs. Interestingly, the size and cell density of sona clones were widely varied (Fig. 2e–g). Based on the difference in clone size, we arbitrarily assigned the clones to three groups. The first group represented single-cell clones with no nuclei, which may absolutely require Sona (blue arrowheads in Fig. 2e–g). The second group of clones was larger than the first group of clones and their nuclei were sparser than their heterozygous counterparts (blue arrows in Fig. 2f,g). Thus, these cells may require Sona at a lesser extent than the cells in the first group of clones. The third group of clones was large in size and the cell density was same as heterozygous counterparts (black arrows in Fig. 2e–e”). Thus, the cells in these clones did not require Sona.

These different sona clones reminded us of the patchy pattern of GFP driven by the sona-Gal4 line, P{GawB}CG9850. The intensity of GFP in the sona > GFP disc varied widely in a highly magnified image (Fig. 2h), and has been shown to be similar to the in situ hybridization pattern of sona transcripts28. To further prove this point, we co-stained sona > GFP discs for GFP and Sona protein with anti-GFP and Sona-Pro antibody28. We found that the level of GFP correlated with that of Sona protein in sona > GFP wing discs, indicating that GFP driven by sona-Gal4 driver reflects both sona transcript and Sona protein (Supplementary Fig. S2). Therefore, the three different kinds of sona clones may have been generated by the differences in the level of sona expression.

We expected to see the sona clones in the entire pouch region because sona was expressed in the entire wing pouch region. Indeed, the large sona13 clones were generated randomly without any preferred location (Supplementary Fig. S3). Overall shapes of large sona clones were similar to those wedge-typed control clones but sona clones appeared to be somewhat rounder than the wild-type clones. To further identify differences between wild-type clones and sona clones, we next generated sona RNAi clones (Supplementary Fig. S4). We collected embryos from hsFlp > Actin > CD2 > Gal4 UAS-GFP and hsFlp > Actin > CD2 > Gal4 UAS-GFP UAS-sona RNAi-111–4 flies for one day at room temperature, and heat-shocked them for one hour at 37 °C at five different time points (Supplementary Fig. S4a). Examination of GFP-positive wild-type clones and sona RNAi clones revealed that average area of RNAi clones are considerably less than that of wild-type clones (Supplementary Fig. S4b–j). When the clones were generated at the 3rd instar stage that was 29 hours before dissection, wild-type clones were evenly distributed in the wing pouch but sona RNAi clones were absent in some particular regions (Supplementary Fig. S4k–m). These data are consistent with our finding that cells in the pouch region differentially require Sona.

The extent of cell survival is proportional to the level of sona expression in irradiated conditions

To understand the effect of sona level on cell survival, we examined which cells are more prone to cell death under γ-ray irradiation. Based on our clonal analyses, we expected that the cells expressing sona at the high level should be more resistant to cell death. To test this idea, sona > GFP larvae were irradiated with γ-ray at 1,500 or 4,500 rad, and the pattern of GFP and CC3 in their wing discs were examined (Fig. 3a–d). Indeed, CC3 signals were mostly absent in the cells that express a high level of sona (Fig. 3d). Cell death by irradiation was severer in sona > sona RNAi-111–4; GFP discs than control sona > GFP discs, demonstrating that the reduction in sona expression enhances cell death (compare Fig. 3b,e to f). Thus, the transcriptional level of sona seemed to positively correlate with the extent of cell survival under the irradiation condition.

Figure 3
figure 3

Sona-expressing cells are resistant to cell death upon irradiation. Irradiation intensity is indicated at the lower left. (a–d) Control sona > GFP wing disc (a), sona > GFP wing discs from the larvae irradiated at 1,500 rad (b) and at 4,500 rad (c) followed by culture for 24 hrs before dissection. The white box in c is magnified in d. The yellow arrows indicate the cells that have a high level of CC3 signals and a low level of sona. The arrowheads mark the cells with no sona but with CC3. The white arrows mark the cells with a high level of sona without CC3. (e,f) Control (e) and irradiated (f) sona > sona RNAi-111–4 wing disc from the irradiated larvae that were cultured for 24 hours before dissection. Arrows and arrowheads mark the regions with GFP and CC3 signals, respectively. (g,h) The chart displays relationship between sona (GFP) and CC3 levels as described in Supplementary Fig. 2a (g). A 2D histogram showing the inverse correlation between sona (LacZ) and CC3 values, and the three arbitrary groups of cells are indicated by colored boxes (h). Representatives of these three groups are marked in g. Scale bars: (a) 60 μm; (b,c,e,f) 40 μm; (d) 13.2 μm.

To quantify the relationship between the levels of sona and cell death, lines were randomly drawn and the intensities of GFP and CC3 on those lines were measured in irradiated sona > GFP discs (Supplementary Fig. S5a). Consistent with the immunocytochemical data in Fig. 3c, raw data showed that the level of CC3 was low when the level of GFP was high, and vice versa in irradiated discs (16 measurements in 10 discs) (Fig. 3g). To statistically analyze these data, we obtained the Mander’s coefficient using measured intensities from multiple wing discs. The Mander’s coefficient is a widely used statistical method that quantifies the degree of co-localization between two fluorescent channels by the combined measurement of correlation and co-occurrence (values ranging from 0 to 1; 0 for complete mutual exclusion, 0.5 for neutral relation, and 1 for complete co-occurrence)38. The average Mander’s coefficient was 0.13 for Sona+ when CC3 is present and 0.15 for CC3+ when Sona is present (Supplementary Fig. S5b,c), which indicate their strong negative correlation. The analysis with the irradiated en > GFP was carried out as a control to validate the use of Mander’s coefficient for correlation analysis (Supplementary Fig. S6a,b). In sum, Mander’s coefficient values confirmed that Sona and CC3 levels show an inverse correlation after irradiation.

Intensity values of GFP and CC3 in sona > GFP and en > GFP discs were used to generate 2D intensity histograms in order to obtain quantitative relationship. Intensity values of different sample sources generated similar 2D histograms, and representatives are shown in Fig. 3h and Supplementary Fig. S6c. We also created an artificial case that represents perfect positive correlation between the two values and generated 2D histogram for comparison (Supplementary Fig. S6d,e). Unlike these two controls, the 2D histogram of the sona > GFP discs showed inverse correlation between GFP and CC3 as expected from the data of Mander’s coefficient (Fig. 3h). We arbitrarily categorized the values in the 2D histogram into three groups as follows. Cells with the low levels of both Sona and CC3 were assigned to group 1 (yellow box), cells with the low level of Sona and the high level of CC3 were assigned to the group 2 (red box), and cells with the higher level of Sona and the low level of CC3 were assigned to the group 3 (green box). Cells in the groups 2 and 3 were readily identified in the wing discs and were consistent with our finding that the level of sona determines cell survival (Fig. 3d). On the contrary, the 2D histogram clearly identified the group 1 cells that represented the majority of cells in the wing disc, but were undetectable in the stained discs. These cells had no sona or a very low level of sona expression but were resistant to cell death. We reasoned that the group 1 cells may express another protein that inhibits cell death, and we chose Diap1 as a candidate because Diap1 is widely expressed in wing discs and essential for cell survival in a cell autonomous manner19.

Expression patterns of sona, Diap1, and rpr show negative correlation after irradiation

To find out the relationship between sona and Diap1, we examined the expression patterns of Diap1 and LacZ in sona > lacZ discs (Fig. 4a). LacZ and Diap1 exhibited complicated mosaic patterns with weak negative correlation (arrowheads in Fig. 4a’,a”). When the sona > lacZ larvae were irradiated at 4,500 rad, LacZ pattern was changed to cluster-like and Diap1 was more evenly expressed in the wing pouch (Fig. 4a,b). Raw data and 2D intensity histograms of sona > GFP control and irradiated discs revealed that the level of Diap1 was low in the LacZ+ clusters (Fig. 4e–h; Supplementary Fig. S7). This indicated that negative correlation between sona and Diap1 became more strengthened by irradiation. LacZ pattern in sona > GFP; Diap1-LacZ discs was similar to the pattern of Diap1 protein in sona > lacZ discs, demonstrating that the negative correlation between sona and Diap1 is generated at the transcriptional level (Fig. 4c,d). Thus, the transcription of both sona and Diap1 genes are coordinately changed by irradiation with negative correlation.

Figure 4
figure 4

Negative correlation of sona and diap1 patterns is accentuated by irradiation. (a,b) Patterns of Diap1 and LacZ in control (a) and sona > lacZ wing discs irradiated at 4,500 rad (b). Arrowheads mark the region with negative correlation, a high level of Diap1 and a low level of LacZ (a’,a”). (c,d) Patterns of Diap1-lacZ and GFP in control (c c’) and irradiated sona > GFP; Diap1-lacZ wing discs at 4,500 rad (d,d’). (e,f) The charts display the expression level between sona (LacZ) and Diap1 in control (e) and irradiated sona > lacZ wing discs at 4,500 rad (f). (g,h) The 2D histograms show correlation between sona (LacZ) and Diap1 values in the discs before (g) and after irradiation (h). (i,j) Basal plane of the irradiated sona > lacZ disc. The white box in i is magnified in j. Arrow mark CC3 signals in the cell with no Diap1, and arrowheads mark the cell with sona but no CC3 signals. (k–m) The pattern and level of rpr-LacZ and sona (GFP) in control (k) and in irradiated (l) in rpr-lacZ, sona > GFP discs. The region marked with a white box in l is magnified in m. Cells with no GFP, but with a high level of LacZ are marked with yellow arrows, and cells with a high level of GFP, but with no LacZ are marked with white arrows. Scale bars: (ad,i) 60 μm; (j) 15.4 μm; (k,l) 40 μm; (m) 5.7 μm.

Since both Sona and Diap1 are involved in cell survival and their expressions are negatively correlated, we expected that the cells with neither Sona nor Diap1 become apoptotic upon irradiation. Indeed, cells with the high level of CC3 had low levels of sona and Diap1 expression, which may belong to the group 2 cells that had CC3 signals and a low level of sona (Fig. 4i,j; Fig. 3h). Taken together, expression of either sona or Diap1 at a high level is sufficient for cell survival against irradiation.

If sona is expressed at a high level in surviving cells, sona and rpr should also be expressed with the negative correlation because rpr is expressed in dying cells. Consistent with previous reports22, irradiation changed Rpr-LacZ pattern and increased the level of LacZ in rpr-lacZ/sona-Gal4; UAS-GFP/+ wing discs (Fig. 4k,l). Furthermore, rpr was expressed at a high level in the cells with no sona expression (yellow arrows in Fig. 4m,m’). Conversely, cells with a high level of sona expressed no rpr (white arrows in Fig. 4m,m’). In summary, rpr is upregulated in the cells with no sona expression in irradiation conditions.

Sona exhibits genetic relationships with Diap1 and rpr

Our data so far have shown that sona and Diap1 are expressed in different cells but function together for cell survival. To examine their genetic relationship in cell survival, we compared the wing phenotypes of sona RNAi, Diap1 or both, driven by ptc-Gal4. While all ptc > Diap1 wings were normal, 83% of ptc > sona RNAi-111–2 (n = 85) were abnormal with no anterior cross-vein, partially lost L3 vein, and narrowed intervein region between L3 and L4 veins (Fig. 5a–c). When both Diap1 and sona RNAi-111–2 were co-expressed, however, only 10% (n = 97) were abnormal (Fig. 5d). Thus, the wing phenotype of ptc > sona RNAi-111–2 flies was considerably rescued by co-expression of Diap1. In addition, while only 20% (n = 85) of ptc > sona RNAi-111–2 flies reached adult stage, 33% (n = 97) of ptc > sona RNAi-111–2; Diap1 flies could reach adult stage (Fig. 5g). Thus, Diap1 partially suppressed the lethality caused by the knockdown of sona. This suggests that sona and Diap1 exhibit a positive genetic relationship.

Figure 5
figure 5

sona exhibits genetic relationship with diap1 and rpr. Genotypes are indicated at the lower right. Arrowheads and arrows indicate the anterior crossvein and the L3 vein, respectively. (a–f) Adult wing phenotype by single or co-expression of sona RNAi, rpr RNAi, and Diap1 by ptc-Gal4 driver cultured at 25 °C. ptc > sona RNAi-111–2; Diap1 wings (d) have anterior cross-vein (arrowhead) and normal L3 vein (arrow). ptc > sona RNAi-111–4 flies show pupal lethality (h), while ptc > rpr RNAi; sona RNAi-111–4 flies survive to adulthood, and still show reduction in distance between L3 and L4 veins (f). (g,h) Bar graphs show the effects of diap1 and rpr RNAi expression on sona RNAi phenotype.

We next tested whether sona has any genetic relationship with rpr. ptc > rpr RNAi flies had normal wing phenotype with no lethality, and ptc > sona RNAi-111–4 flies cultured at 25 °C exhibited 100% of pupal lethality (Fig. 5e,h). When both rpr RNAi and sona RNAi-111–4 were coexpressed, up to 14% of ptc > rpr RNAi; sona RNAi-111–4 flies (n = 63) reached adult stage with somewhat defective wings (Fig. 5f,h). This indicated that cell death by loss of sona partially depends on the function of rpr. Taken together, sona exhibited the positive genetic relationship with Diap1 but the negative genetic relationship with rpr, which is consistent with our finding that sona promotes cell survival.

sona + clusters are formed by increased sona transcription but not by cell division

We have shown that irradiation changes the pattern of sona-Gal4 expression (Figs 3c, 4b). The most noticeable change in irradiated sona > GFP discs was the appearance of sona+ clusters. sona+ clusters were generated by 4,500 rad but rarely by 1,500 rad (Fig. 3a’–c’). As a result, mostly two major types of cells remained in the wing pouch after irradiation at 4,500 rad: one with the high level of sona and the other with no or very low sona expression (Fig. 3c; Fig. 4b,d,i,l). en > GFP or tubulin > GFP larvae irradiated at 4,500 rad did not show any change in the pattern of GFP in the wing discs, so the GFP-positive clusters in sona > GFP discs were unique to sona-expressing cells (Supplementary Fig. S6a, S8).

Based on the result that sona+ clusters are formed by irradiation, these sona+ clusters may be generated by cell division or induction of sona transcription. To distinguish these possibilities, sona > lacZ larvae were irradiated at 4,500 rad, and then cultured in Bromodeoxyuridine (BrdU)-containing food for 24 hours until dissection (Fig. 6a). In control discs, BrdU was incorporated in the cells of the wing pouch region regardless of sona expression (Fig. 6b,d). In contrast, most sona-expressing cells did not incorporate BrdU but the neighboring cells did in irradiated discs (Fig. 6c,e). The two 2D intensity histograms also confirmed that sona-expressing cells have less BrdU incorporation (Fig. 6f). The histogram also revealed that a weak inverse correlation exists between sona expression and BrdU incorporation even in control discs (Fig. 6f). Therefore, sona-expressing cells divide less frequently than sona cells even in the unirradiated condition. Consistent with the BrdU data, mitotic cells marked with anti-phospho-histone H3 (PH3) antibody were mostly absent in the sona-expressing cells in both control and irradiated wing discs of sona > GFP larvae (Fig. 6g,h). We concluded that sona+ clusters are formed by increased sona transcription but not by cell division.

Figure 6
figure 6

Cells expressing a high level of Sona do not divide upon irradiation. (a) The experimental scheme to examine the effect of irradiation on sona expression and DNA synthesis. (b–e) Wing discs from the control (b) and irradiated (c) sona > lacZ larvae fed with BrdU for 24 hours before dissection. The boxed regions in (b,c) are magnified in (d,e), respectively. The cells expressing a high level of sona are outlined to show the lack of overlapping between LacZ and BrdU signals in e’. An arrow marks a representative sona-expressing cell lacking BrdU incorporation while white and yellow arrowheads mark incorporation of BrdU in the cells with no or a low level of sona expression, respectively. (f) 2D histograms showing the inverse relationship between the levels of BrdU and sona in control and irradiated sona > lacZ discs above. (g,h) Phospho-histone H3 patterns of control (g) and irradiated (h) sona > GFP discs. Arrows mark the cells with PH3 but no GFP. Scale bars: (b,c,g,h) 40 μm; (d,e) 12.6 μm.

To further understand how these sona+ clusters are formed, we generated 3D and combined images from multiple confocal images of unirradiated and irradiated sona > GFP discs (Supplementary Fig. S9). Unlike the single confocal images and the combined images (Supplementary Fig. S9a–a”,b–b”), the 3D images revealed that the sona+ clusters were actually two dimensional confocal images from a continuous cell group that express sona at a high level (arrows in Supplementary Fig. S9c’). An image tilted from the 3D image showed that these continuous cell groups may be clonal (Supplementary Fig. S9d). When the combined images of sona expression were compared between irradiated and unirradiated discs, it appeared that irradiation induced either increase or decrease in the expression level of sona, leading to the greater difference between sona+ and sona cells (Supplementary Fig. S9a”,b”).

Sona promotes development and survival under the irradiation condition

Generation of sona+ cell groups suggested that increase in sona transcription may be required for cell survival in the irradiation condition. To test this, we dissected sona > GFP wing discs at different time points after irradiation. We found that the level of GFP was same in both unirradiated and irradiated discs initially but was about doubly increased by 24 hours after irradiation (n > 30, Supplementary Fig. S10). Furthermore, the level of Sona protein was increased where GFP level is high in irradiated discs (Supplementary Fig. S11). These data raised an interesting possibility that increase in sona transcription is an important required step for survival against irradiation.

To test whether the increase in sona level can promote survival, we examined the effect of overexpressed Sona on irradiated flies. We have previously reported that overexpression of Sona by most Gal4 drivers causes organismal death28. To overcome this Sona-induced lethality, we cultured sona-Gal4/+; Gal80ts/+ control flies and sona-Gal4/+; Gal80ts UAS-sona/+ experimental flies at 18 °C for 6 days until the mid to late third instar stage, and then cultured them at 29 °C for six hours in order to transiently induce Sona expression. Flies were then irradiated with 3,000 rad or 4,500 rad, and cultured at 18 °C until adult stage. Control groups that were treated the same way without irradiation developed at the same pace regardless of Sona, and developed faster than irradiated flies. Interestingly, sona-Gal4/+; Gal80ts UAS-sona/+ flies developed faster than the irradiated control flies (Supplementary Fig. S12). Furthermore, they produced more adults (20.5%, 15 adults out of 74 pupae) than control sona-Gal4/+; Gal80ts/+ flies (13.6%, 8 adults from 59 pupae). Although, Sona was overexpressed only for 6 hours prior to irradiation, both survival and development of flies were noticeably promoted.

sona + cells induce the expression of Cyc D in neighboring cells

As the neighboring cells of sona+ clusters incorporated BrdU, and transient expression of Sona increased survival rate, we hypothesized that the sona+ cells send signals to neighboring cells for cell proliferation and enhance survival. To test whether Sona is able to promote cell proliferation and increases tissue size, we overexpressed Sona using ptc-Gal4 driver at 18 °C. Most ptc > sona flies were lethal, but wings of rare survivors (~70%, n = 16) exhibited enlarged ptc region (Fig. 7a–e). Some wings had enlarged bubble in the ptc region, which supports that Sona is able to increase tissue size. The distances between the L3 and L4 veins of these wings except the one with bubbles are plotted in Fig. 7f.

Figure 7
figure 7

Sona overexpression induces tissue growth and the expression of Cyc D in neighboring cells. (a) Control CS adult wing. (b,c) The overgrowth phenotypes of two ptc > sona adult wings. The phenotype of c seems more severe than b. The boxed regions in (a,b) are magnified in (d,e), respectively. The lines with double arrowheads mark the region between L3 and L4 veins. (f) The quantitative analysis on the distances between L3-L4 regions in CS and ptc > sona wings. (g–i) Control ci > Gal80ts (g) and ci > sona, GFP, Gal80ts (h) larvae were cultured at 30 °C for 10 hours to induce transient overexpression of Sona and then their wing discs were stained for Cyc D. The boxed region in (h) is magnified in (i). Similar results in another disc are shown in Supplementary Fig. S13b. (j,k) en > sona-HA, GFP, Gal80ts larvae were also used to induce Cyc D expression by Sona-HA. Larvae were cultured at 30 °C for 24 hours and stained for HA and Cyc D. Arrows mark the sona cells expressing Cyc D. Arrowheads mark the cells with upregulated Cyc D in the cells that are in the close vicinity of sona+ cells. Scale bars: (g,h) 40 μm; (i,k) 17 μm; (j) 50 μm.

We then examined how Sona is able to increase tissue size. Since prolonged overexpression of sona causes lethality28, we used Gal80ts39 to transiently induce sona in the anterior region by cubitus interruptus (ci)-Gal4 driver. UAS-GFP; UAS-sona, Gal80ts flies were crossed with ci-Gal4 flies and cultured at 18 °C, and the progeny was shifted to 29 °C for 6, 10, and 14 hours during the third instar stage to transiently induce Sona. Because Sona can increase tissue size, we examined whether the level of any Cyclins is increased in the wing discs. Six hour culture at 29 °C did not induce GFP at all, and fourteen hour culture induced signs of tissue damage, so we examined the wing discs after 10 hour induction (data not shown). Among the three larval progeny, ci > Gal80ts larvae and ci > GFP; Gal80ts larvae did not show any change (Fig. 7g; Supplementary Fig. 13a) but ci > GFP; sona, Gal80ts larvae had upregulated Cyc D in wing discs (Fig. 7h; Supplementary Fig. 13b). Cyc D was expressed in the cells that are in the close vicinity of but not in sona-expressing cells in most cases (Fig. 7i). Similar experiments were carried out in the posterior region with en-Gal4 driver, in which GFP was detected after 24 hours of transient expression (Fig. 7j,k). We used UAS-sona-HA instead of UAS-sona for transient expression in order to check the expression of ectopic Sona with anti-HA antibody (Fig. 7k’). Similar to the experiment done in the anterior region with ci-Gal4, Cyc D was upregulated in the cells that show neither GFP nor HA in the posterior region with en-Gal4 (Fig. 7k). Cyc D is a key Cyclin that responds to mitogens before the initiation of DNA synthesis, and its expression is regulated by Wnts in mammalian system40,41,42,43. These data strongly suggest that Sona-producing cells send out signals to the neighboring cells for induction of Cyc D. Whether the signal is Sona itself, Wg, or unknown molecules needs further analysis.

Discussion

We have shown in this report that Sona functions in two different ways. First, Sona is required cell autonomously for cell survival, and the level of sona transcription correlates with the degree of cell survival under the irradiated condition. We found that cells expressing a high level of either sona or Diap1 are resistant to cell death, so there are at least two different types of surviving cells in wing discs. Second, Sona induces cell proliferation cell non-autonomously. Transient expression of sona upregulated Cyc D in neighboring cells but not in sona-expressing cells, suggesting that radiation-resistant sona-expressing cells secrete signaling molecule(s) that upregulates Cyc D in neighboring cells. In conjunction with genetic interactions between sona, Diap1 and rpr, cells expressing sona may dynamically interplay with other cells that express Diap1 and rpr for damage control.

Sona is expressed in the patchy pattern, and sona clones with different sizes were generated in the wing pouch region (Fig. 2). We speculate that the large sona clones are generated where sona is not expressed. Similar result was reported in the loss-of-function clone analysis of the vestigial (vg) gene that is essential for survival and proliferation of wing disc cells43. The authors found that vg RNAi clones are generated where Vg is not expressed or expressed at a low level in wing discs44. That is, the loss-of-function vg clones are rarely formed where Vg is required for cell survival and proliferation. Indeed, Vg drives cell cycle progression by inducing dE2F1 gene whose product is essential for G1 to S transition in cell cycle45. Formation of clones with different sizes may be a general phenomenon for the genes that are expressed unevenly and are required for cell proliferation or cell survival. Varied levels of sona and vg expression in different parts of wing discs may be required to create regional differences in growth rate in order to form wings with proper size and shape.

What are the distinct functions of sona-expressing cells? In normal conditions, sona is expressed in a widely varing level  in a patchy pattern, and acts as a signal to promote cell proliferation for wing development. In irradiated conditions, the transcriptional level of sona dramatically increased among sona-expressing cells that may be required to cope against extensive cell death. This irradiation generated two major cell groups in the wing pouch. The first group of cells are determined to grow and proliferate or, alternatively, to die upon a high level of irradiation. The second group of cells are resistant to irradiation and send out signals to regulate the first group of cells. The first group of cells should express proteins for cell growth, cell proliferation, cell survival and cell death including Diap1and Rpr. The second group of cells should have capacity to sense the changes in environment and send out signals to regulate the first group of cells. In this study, we identified Sona as one of the proteins expressed in the second group of cells in a patchy pattern. Interestingly, a phosphorylated form of ribosomal protein S6 (pS6) that should be regulated in the first group of cells is also present in a patchy pattern46,47. We speculate that pS6 and sona may express in the first and the second groups of cells, respectively.

We showed that Sona induces Cyc D in nearby cells, which is consistent with the finding that overexpressed Sona increases tissue size (Fig. 7). Cyc D is known as a target protein of Wnt signaling in mammals40,41,42,43, and acts in a G1 phase of cell cycle that responds to mitogens48. In flies, overexpression of Cyc D/Cdk4 accelerates cell division in proliferating wing disc cells while increasing the cell size in post-mitotic cells by endoreplication49. Because Cyc D is required for cell division of surviving cells, it should be upregulated in Diap1-expressing cells. Indeed, Diap1 is induced by Yorki and STAT in Hippo pathway and JAK-STAT pathway, respectively, and these two pathways are essential for cell proliferation in a cell autonomous manner25,26. An important question is the identity of the direct signal(s) that upregulate Cyc D. Secreted Sona may act as a direct signal or, as a metalloprotease, may modulate the activity of other signaling molecules such as Wg. It has been reported that irradiation activates a damage-activated enhancer BRV118 of the wg gene50,51. Therefore, irradiation changes the transcription pattern of both sona and wg, and may coordinate the functions of Sona and Wg for induction of Cyc D, although Sona may activate yet another signaling molecules.

An interesting new mechanism to repair tissue damage has been recently discovered, in which cells of the presumptive hinge region in the wing disc migrate into the pouch region after irradiation52. These hinge cells also migrate into the pouch region when the pouch cells are killed by hid expression53. These hinge cells are resistant to cell death due to the activation of Wg signaling and STAT signaling that suppress the transcription of rpr52. Resistance to irradiation and lack of rpr expression are also characteristics of sona-expressing cells, suggesting that the sona-expressing cells in the wing pouch may be functionally related to the radiation-resistant hinge cells. Because intense irradiation physically induces DNA breakage even in these radiation-resistant cells, the mechanism by which these sona-expressing cells cope against DNA damage is an important question to be explored.

Several mammalian metalloproteases in ECM are also shown to prevent apoptosis. Overexpression of MMP-15 (matrix metalloprotease-15) prevents apoptosis of Hela and human adenocarcinoma54. Furthermore, ADAM-12 increases apoptosis of stromal cell but decreases that of tumor cells55. In ADAMTS family, ADAMTS20 is shown to be required for cell survival. ADAMTS20 is mutated in belted (bt) mice that show cell death of melanoblast56. Sona is also involved in cell survival by increasing resistance against irradiation and by promoting cell proliferation of neighboring cells. Identification of Sona substrate and its function will greatly help understand the role of Sona in cell survival, which is currently ongoing in our laboratory.

Materials and Methods

Drosophila strains

All UAS-sona RNAi lines and sona mutants generated in our lab and characterization of sona-Gal4 line (P{GawB}CG9850) are described28. P{lacW}Diap1j5C8/TM3 was used as Diap1-lacZ line57,58. UAS-Diap1, UAS-lacZ, rpr-lacZ, wg-lacZ, UAS-GFP, UAS-p35, hsFLP; FRT42D ubiGFP, FRT42D ubi-GFP, ptc-Gal4, en-Gal4, tubulin-Gal4 and apterous-Gal4 were obtained from Bloomington stock center. UAS-rpr RNAi was obtained from VDRC (101234). ci-Gal4 is described in59. Fly cultures were carried out at 25 °C unless otherwise indicated.

Adult wing mounting

Wing of 2–7 days old adult flies were dissected and mounted in Gary’s Magic Mountant (Mixture of Canada Balsam and methyl salicylate, 4:1).

Generation of mitotic clones and RNAi clones

Mitotic clones were generated by FLP-mediated recombination60. sona13 FRT42D/CyO or sona47 FRT42D/CyO flies were crossed to hsFLP; FRT42D ubi-GFP lines28. Clones were induced by heat shock at 37 °C for 1 hour during first or second instar larval stages, and late 3rd instar larvae were dissected. For the generation of sona RNAi clones, hsFLP > Actin > CD2 > Gal4 flies61 were crossed with UAS-GFP; UAS-sona RNAi-111–4 line and their progeny were heat-shocked for one hour at 37 °C to induce the clones, and dissected when they were late 3rd instar larvae. For control clones, UAS-GFP line was used.

Immunocytochemistry

Wing discs from wandering third-instar larvae were dissected and processed as described62. DAPI staining solution (Boehringer Mannheim) was used in the first washing step after secondary antibody staining. Discs were mounted in Vectashield (Vector Laboratories). The images were acquired using Zeiss LSM laser scanning confocal microscope and presented using Adobe Photoshop.

Primary antibodies were used in the following dilutions: sheep anti-GFP (Abd serotec), 1:100; chicken anti-β-gal (ab 9361), 1:100; rabbit anti-cleaved Caspase-3 (Cell Signaling Technology), 1:250; rabbit anti-cleaved Dcp-1 (Cell Signaling), 1:100; rabbit anti-Sona-Pro28, 1:500; mouse anti-Diap1 (a gift from Dr. Soon Ji Yoo21,) 1:100; mouse anti-Cyc D (Santa Cruz), 1:200; rabbit anti-Dlg63, 1:500; rabbit anti-PH3 (Millipore), 1:200; rat anti-HA (Roche), 1:150. Secondary antibodies were from Jackson Immunoresearches.

Apoptosis induction by irradiation

Flies were grown at 25 °C until their progeny reached second to third instar larvae, and were treated with the indicated doses of γ–radiation (1,500 or 4,500 rad). They were allowed to grow for 24 hours after irradiation and dissected at the wandering 3rd instar larvae.

BrdU feeding

The protocol has been modified based on several reports64,65,66. Embryos of sona > lacZ genotype were cultured at 25 °C until they reached to the 1st or 2nd instar larvae. Right after irradiation at 4,500 rad, 0.5 mg/ml BrdU were poured evenly on top of the food containing irradiated larvae. After 24 hours of BrdU feeding, 3rd instar larvae were dissected in Schneider’s insect culture medium (M3 medium). For control experiments, unirradiated 1st and 2nd instar larvae were fed in BrdU containing food for 24 hours. The discs were fixed for 20 min in PBTx (phosphate buffered saline, 0.3% Triton X-100) containing 5% paraformaldehyde.

Image acquisitions and analysis

To quantify the degree of colocalization, the confocal images from two channels were compared using coloc2 analysis package in Fiji software67. To minimize the background signal, the threshold values were automatically set according to the Cotes method. For testing Cotes’ significance, the point spread function (PSF) was set to 50 with the number of iteration of 10. Mander’s Colocalization Coefficient (MCC) was presented within results. The representative 2D histograms were also generated using Fiji software.