Transcriptional profiling of polyQ Atrophins. (A) Illustration of the crossing and ageing scheme used to obtain total RNA extracts from fly heads for the transcriptional profiling and all successive qPCR assays. Expression of different Atrophin forms with the GMR driver was induced, owing to a temperature-sensitive mutant Gal80 repressor. F1 flies were allowed to develop at 18°C; at this temperature the Gal80 repressor keeps transgenes silent. Newly eclosed flies (0–48 h) were collected and killed immediately (0d) or aged for 2 or 14 days at 29°C. This inactivates Gal80 and transgenes are switched on by GMR-Gal4. This protocol allows comparing siblings that differ exclusively in their age and transgenes expression. Control flies crossed to GMR-Gal4; UbiGal80^ts in all experiments are from the w¹¹¹⁸ stock in which all UAS transgenes have been generated. (B) Tangential eye sections of flies representative of all the different populations used in the microarray analysis at all different time points. Weak degeneration is only visible after 14 days with polyQ Atro; in particular with Atro75QN there is an initial loss of photoreceptors (PR, arrow), 30.7% of the ommatidia has lost at least 1 PR, that is only 5.1% of all neuronal PR have been lost at this stage (N=333).

Global analysis of transcript profiling. (A) Heat maps for three subsets of the pattern clustering analysis generated using the Euclidean Distance metric. Upregulated genes are in red and downregulated genes are in green. The downregulating transcriptional activity increases with time. The main branch in clustering of the different Atro mutants is along time points. See also . (B) Venn diagrams for the subset of genes affected at the two time points for the different Atro versions with respect to control flies. The overlap between all Atro forms and the two polyQ versions increases with time. The genes in common for all Atro versions or between the polyQ Atro mutants at both times are listed.

Regulation of ft/Hpo levels is functionally relevant to neurodegeneration by Atrophins. Statistical significance is marked as follows: ^***P<0.001, ^**P<0.01; ^*P<0.05 in two-tailed t-test. All qPCR results here and in all figures are the average of three reactions of independent biological replicas. (A) qPCR analysis of the fold changes±s.d. of ft transcription. Downregulation starts at 2 days and progresses at 14 days. (B) qPCR analysis of the enrichment±s.d. of different regions of the ft regulatory elements in ChIP for Atro from BG3 neuronal cell extracts. Enrichment is calculated as the percentage of DNA immunoprecipitated from extracts of cells in which Atro expression has been induced, with respect to the amount immunoprecipitated from uninduced cell extracts according to the (Atro–No Ab)_induced/(Atro−No Ab)_uninduced formula. The DNA region from IV chromosome has been previously shown not to be immunoprecipitated in ChIP for Atro () and has been used as a negative control. (C) Histograms showing the number of ommatidia with a full complement (7) of PR in flies expressing Atro75QN with the Rhodopsin1 driver in either a control (w¹¹¹⁸) or different mutant backgrounds and aged at 29°C for 28 days. Heterozygosis for two independent ft alleles and a sav allele significantly enhances the loss of PR. Mild overexpression of Wts via a GMR–wts transgene, which does not display any strong phenotype per se, significantly suppresses the loss of PR. No interaction was detected in this assay with wts^x1 and yki^B5 alleles in heterozygosity. N=430–963 from at least four eyes. (D) Tangential eye sections and histograms showing the degeneration of GMR>Atro75QN, Ubi–Gal80^ts flies in combination with different UAS transgenes and aged 14 days as in . UAS–EGFP is used as a negative control. Either Ft or Wts overexpression strongly suppresses loss of PR caused by Atro75QN. Lower panels are provided as examples of quantification. A more modest but still significant rescue is observed by overexpressing a Yki RNAi contruct or Hpo; however, in this case it is to be considered that the overexpression of Hpo per se brings about the loss of at least one PR in ∼5% of ommatidia (data not shown). No effect is detected with overexpression of Sav. Two-tailed t-test: ^*P<0.05; ^**P<0.01. The wts^EPG4808 is a previously uncharacterised EP element insertion at the 5′ of the wts gene. Rare homozygous escapers display larger eyes, whereas in combination with GMR–Gal4 this line gives rise to smaller rough eyes, and, finally, qPCR analysis of GMR>wts^EPG4808 indicates a 10-fold increase in the head content of wts mRNA (data not shown).

Neurodegeneration by mutations in ft. (A) Tangential eye sections through ft^fd clones aged 1, 7 and 14 days, of ft^fd,d^DGC13 double mutant clones and of ft^fd clones in a yki^B5 heterozygous background aged 14 days. Clones are marked by the absence of yellow pigment. Arrows point at degenerating photoreceptors, arrowheads point at intact wt photoreceptors in mosaic ommatidia. The circled ommatidium is genetically wt but non-autonomously flipped in its polarity and has not degenerated. Lower panels are masks that show clonal borders and quantification of PR number per ommatidia. After 14 days, almost all ft^fd cells have degenerated, whereas almost all ft^fd,d^DGC13 and some ft^fd,yki^B5/+ neuronal photoreceptors have not. (B) Histograms showing the quantification of degeneration in ft^fd, ft^fdd^DGC13 and ft^fdyki^B5/+ clones. Loss of ft leads to statistically significant loss of neurons and this is suppressed at 14 days by yki^B5and much more dramatically by d^DGC13. ^***P<0.001, ^**P<0.01; ^*P<0.05 in two-tailed t-test. (C) Degeneration in mosaic ommatidia in ft^fd clones. At this stage of degeneration approximately half of the non-pigmented (genotypically mutant) PR cells display a degenerative phenotype, whereas virtually all pigmented (genotypically wt) PR cells are normal. χ²-test: 82.38; P<0.001. Total PR N=310 (pigmented) and 327 (not pigmented). (D) Tangential eye sections through MARCM clone mutants for ft^fd that express either no transgene or an RNAi construct against yki (UAS–yki^IR) with Tub–Gal4 and aged 14 days at 29°C. Because the transgene UAS–yki^IR is on chromosomal arm 2 L, the same as ft, clones generated with this system carry two copies of UAS–yki^IR and therefore downregulate yki very effectively. No other cell outside the clones expresses the construct and is therefore wt for yki. Mask panel is presented on the right of each section. A quantification of photoreceptor numbers (far right) displays a very significant increase in the number of wt ommatidia (N=210 and 219 from four eyes). ^***P<0.001 in two-tailed t-test. If all classes of ommatidia are considered, χ²-test=232.30; P<0.001 for 3 degrees of freedom.

Neurodegeneration by mutations in Hippo pathway genes. All Hippo pathway mutant clones display a remarkable increase in interommatidial space as a consequence of failure in lattice programmed cell death and these lattice cells also degenerate with age (see also . Arrows point at degenerating neuronal photoreceptor cells. (A) Tangential eye sections through wts^X1 mutant fly eyes. In clones for the null wts^X1 allele, degeneration inside the clones is at severe stages at 14 days. (B) Tangential eye sections through sav³ clones. In sav³ after enclosure many photoreceptors are intact but after 14 days at 29°C most sav mutant photoreceptors have degenerated. (C) Histograms showing the quantification of ommatidia with the full complement of photoreceptors in hpo, wts and sav mutant clones. The progression of loss of photoreceptors is evident and statistically significant in all cases. Mutants for hpo^BF33 do not survive for 14 days at 29^oC and corresponding sections are shown in . N=195–268. ^***P<0.001, ^**P<0.01; ^*P<0.05 in two-tailed t-test.

Uncoupling of neurodegeneration and overgrowth. (A) Regression analysis of the overgrowth versus the neurodegeneration observed in loss of function mutants (left) and overexpression of transgenes (right). Overgrowth quantification is shown in . For loss of function mutant, the neurodegeneration was quantified at 14 days in clones (as shown in and ). For UAS-overexpression transgenes neurodegeneration was quantified at 14 days when expressed with GMRGal4; UbiGal80^ts. For CycD+Cdk4, these specific data are missing and have not been plotted; however, shows that these mutants do not cause neurodegeneration in a different setup in which they are expressed during both development and adult life. In both cases the regression coefficient r² is extremely low (0.21 and 0.03), indicating an absence of correlation between the two processes. (B) Tangential eye sections through the eyes of flies expressing with GMR–Gal4, Ubi–Gal80^ts either UAS–EGFP or UAS–Yki or UAS–Yki^{S111A,S168A,S250A} and aged 1 or 28 days. Arrows point at missing or degenerated photoreceptors. The UAS–Yki^{S111A,S168A,S250A} is so effective that even the very low expression leaking out with this system is enough to affect development and generate mild eye roughness and overproliferation of lattice cells. Histograms showing the quantification of cell loss in eyes shown in A and also eyes from flies aged at the intermediate 14-day stage. Mild but significant degeneration is observed for both Yki forms, with respect to the negative control; however, no statistical difference is observed between the two Yki proteins that have dramatically different effects on overgrowth. ^***P<0.001, ^**P<0.01; ^*P<0.05 in two-tailed t-test.

Neurodegeneration by polyQ Atro partially requires dachs. Tangential eye sections through d^GC13 clones (marked by the absence of yellow pigment) either in control flies or those expressing Atro75QN with the Rhodopsin1 driver and aged 28 days. Mask panels below each section exemplify clonal boundaries and PR no. for the ommatidia. Early signs of degeneration indicated by reduced complement of photoreceptors are detected specifically in the pigmented area (arrowhead). At this stage, inside d mutant clones all ommatidia are unaffected and display a wt arrangement (arrows). (B) Histograms showing the quantification of cell loss in eyes shown in (A). A statistically significant rescue is obtained both for the number of wt ommatidia (^*P<0.05 in two-tailed t-test) and for all other categories of ommatidia (χ²=44.76; P<0.001 for 2 degrees of freedom).

Autophagic modifications in ft mutants. (A) EM scan of a wt (left) and ft^fd mutant ommatidium (right) of a ft^fd clone after 14 days. Scale bar: 1 μm. High-magnification panels (far right, top to bottom) display autophagosomes with undigested debris, damaged mitochondria and forming phagophores (arrowheads), found in ft^fd mutant cells. Scale bar: 0.2 μm for zoom-in panels. (B) Graphs of the quantification of autophagic vesicles (AV) per photoreceptor found in EM sections of 7-day (left)- and 14-day (right)-old ft^fd clones. Significant accumulation of AV is found in mutant cells (not pigmented) with respect to genotypically wt (pigmented) cells. ^**P<0.01 and ^***P<0.001 in one-tailed t-test. N=12 pigmented cells versus 20 non-pigmented cells for the 7-day graph and 13 pigmented cells versus 21 non-pigmented cells for the 14-day graph. (C) Confocal pictures of whole-mount retinae of a ft^fd clone aged 7 days and expressing GFP∷Atg8a ubiquitously with Tub–Gal4. Red is phalloidin marking rhabdomeres, green is GFP and the clone is marked by the absence of β-gal staining (blue). Small GFP∷Atg8a dots accumulate specifically inside ft mutant cells (arrow). (D) Confocal pictures of whole-mount retinae of a ft^fd clone aged 7 days (left) and 14 days (right). Red is phalloidin marking rhabdomeres, blue is p62 and the clone is marked by the absence of GFP staining (green). p62 starts to gather in small dots specifically inside ft mutant cells (arrow) and then accumulates massively (arrow) in the ft mutant clones as many cells degenerate.

Autophagy in sav mutants. (A) EM scan of a 7-day-old sav³ mutant ommatidium and zoom-in on autophagosomes containing undigested debris (arrowhead). Scale bar: 1 μm for panel on the left and 0.2 μm for zoom-in panel on the right. (B) Graph of the quantification of autophagic vesicles (AV) per photoreceptor found in EM sections of 1-week-old sav³ clones. Significant accumulation of AV is found in mutant cells. ^***P<0.001 in one-tailed t-test. N=45 pigmented cells versus 42 non-pigmented cells. (C) Model for the relationship between Atro, Ft/Hippo and neuronal homeostasis. Atro wt overexpression, and more effectively polyQ Atro overexpression, causes a downregulation of ft transcription and of other genes important for photoreceptor neuron function and homeostasis. This results in downregulation of Ft/Hippo signalling, which affects neurodegeneration. Healthy Ft/Hippo signalling contributes to neuronal homeostasis through effective autophagy. Atro may also exert Ft-independent effects through other factors (like Tetraspanins) and polyQ Atro also affects neurons through polyQ-specific factors (like misfolding and aggregation), which put additional pressure on the autophagic system. Other polyQ proteins may not feed into the Ft/Hippo pathway, as in the case of Htt-ex1-93Q in flies.