Expression of a biologically relevant, truncated form of the human ATXN3 gene containing an expansion of the glutamine region (trSCA3 Q78, containing 78 glutamines, hereafter referred to as SCA3polyQ78) in Drosophila eyes resulted in progressive degeneration of photoreceptor cells [21], accompanied by eye depigmentation as well as depigmentation accompanied by black necrotic spots (depigmented and necrotic, Fig. 1a; arrow indicates necrotic spots). No such effects were seen in control eyes or eyes expressing SCA3polyQ27, containing a non-pathogenic length of glutamines (SCA3polyQ27; Fig. 1a) [21]. Quantification of the fraction of the eyes with a normal appearance, displaying depigmentation or necrotic spots is shown in Fig. 1b. Only expression of SCA3polyQ78 resulted in depigmentation or necrotic spots. We quantified the extent of degeneration by determining the fraction of the eyes containing necrotic spots. This quantification provides a quick and reproducible means of analysis. To see whether SCA3polyQ78-induced degeneration was accompanied by decreased solubility of SCA3polyQ78, lysates of fly heads were analyzed. Expression of SCA3polyQ78 but not SCA3polyQ27 resulted in the presence of high molecular weight insoluble proteins (Fig. 1c), which have previously been identified as aggregates using immunofluorescence [22].

Analysis and quantification of the effects induced by expression of SCA3polyQ78. a Representative pictures of a control eye, and eyes expressing SCA3polyQ27 or SCA3polyQ78. Expression of SCA3polyQ78 but not SCA3polyQ27 in Drosophila eyes resulted in loss of pigment (“depigmented”), as well as the presence of necrotic patches (“necrotic”, indicated with arrow). b Quantification of the fraction of the eyes as shown in a that have a normal appearance, display a loss of pigment or contain necrotic patches; the fraction of SCA3polyQ78 eyes containing necrotic patches was used as a quantitative measure for degeneration. Data are representative of at least three independent experiments. SEM = 0.863. c Western blot analysis of fly head lysates expressing eye-specific, HA-tagged SCA3polyQ27, or SCA3polyQ78. A fraction of SCA3polyQ78 but not SCA3polyQ27 was SDS-insoluble. Tubulin was used as a control for equal loading. d Expression of mCD8-GFP in the eye to analyze SCA3polyQ78-induced eye degeneration and analysis of SCA3polyQ78 expression on the localization of astrocytes (expressing RFP). Top panel: GFP fluorescence of a representative control eye or eyes coexpressing SCA3polyQ27 or SCA3polyQ78. Bottom panel: the localization of myr-RFP-expressing (RFP) astrocytes was analyzed in control eyes or eyes expressing SCA3polyQ27, or SCA3polyQ78. e Lysates of control flies or flies expressing mCD8-GFP in the eyes and myr-RFP in astrocytes together in the absence or presence of SCA3polyQ27 or SCA3polyQ78 were analyzed for expression of HA-tagged SCA3, GFP, and RFP. Tubulin was used as an equal loading control. Genotypes in a–c: control, GMR-QF2/+. SCA3polyQ27, GMR-QF2/+; QUAS-SCA3polyQ27/+. SCA3polyQ78, GMR-QF2/+; QUAS-SCA3polyQ78/+. Genotypes in d, top panel: control, GMR-QF2/+; QUAS-mCD8-GFP/+. SCA3polyQ27, GMR-QF2/+; QUAS-SCA3polyQ27/+; QUAS-mCD8-GFP/+. SCA3polyQ78: GMR-QF2/+; QUAS-SCA3polyQ78/+; QUAS-mCD8-GFP/+. Bottom: control, GMR-QF2/+; alrm-Gal4::UAS-myr-RFP/+. SCA3polyQ27, GMR-QF2/+; alrm-Gal4::UAS-myr-RFP/QUAS-SCA3polyQ27. SCA3polyQ78: GMR-QF2/+; alrm-Gal4::UAS-myr-RFP/QUAS-SCA3polyQ78. Genotypes in e: as in d except that, where indicated, both QUAS-mCD8-GFP (eyes) or alrm-Gal4::UAS-myr-RFP (astrocytes) were expressed

We used another robust, sensitive way to quantify degeneration, by coexpressing membrane-targeted mCD8-GFP [23] with SCA3polyQ27 or SCA3polyQ78 in the eyes. Expression of SCA3polyQ78 but not SCA3polyQ27 resulted in a decrease of mCD8-GFP levels, indicative of degeneration (Fig. 1d, top panel). Analysis of mCD8-GFP expression levels in SCA3polyQ78-expressing eyes in lysates of fly heads allows for quantification of degeneration (Fig. 1e). Both the fraction of necrotic eyes and GFP levels were used as readouts to determine the effect of downregulation of specific genes in astrocytes on degeneration.

To analyze a putative cell-non-autonomous role for astrocytes in SCA3polyQ78-induced eye degeneration, we used two binary expression systems that act independently [20], UAS-Gal4 and QUAS-QF2 (Additional file 1a). Indeed, expression of fluorescent proteins in astrocytes (RFP) or the eyes (GFP) shows specific non-overlapping expression in the respective tissues (Additional file 1b). We analyzed whether the location of astrocytes (expressing RFP) was altered upon expression of SCA3polyQ78 in the eyes. We observed the presence of astrocytes in SCA3polyQ78-expressing eyes but not in control eyes or eyes expressing SCA3polyQ27 (Fig. 1d, bottom panel). The response in SCA3polyQ78-expressing flies is quite robust, as they all displayed expression of RFP in the eyes (Additional file 2). The number of astrocytes in all samples in Fig. 1d is comparable, as total RFP levels in fly heads were similar (Fig. 1e). Thus, the presence of astrocytes in SCA3polyQ78-expressing eyes suggests a response in astrocytes, which warrants further investigation into a putative cell-non-autonomous role of astrocytes in degeneration.

Our Drosophila SCA3 model allows specific analysis of cell-non-autonomous contributions of astrocytes to the degenerative SCA3 eye phenotype. We selected genes (many of them evolutionarily conserved; Additional file 3) potentially involved in (1) communication between neurons and astrocytes, such as neuropeptides and their cognate receptors (2) immune signaling pathways, such as nuclear factor kappa B (NF-κB) and (3) potential receptors for cytokines, neuropeptides, aggregates, or damage-associated molecular patterns. Some of the genes that were selected have either been implicated in cell-autonomous activation of astrocytes [6] or are upregulated in the degenerative fly brain [24, 25]. However, the role of these genes in cell-non-autonomous activation of astrocytes in response to proteotoxic stress in neurons and subsequent contribution to neurodegeneration still remained to be determined. In a candidate RNAi screen, RNAi constructs targeting a set of 156 selected genes were expressed exclusively in astrocytes in flies expressing SCA3polyQ78 exclusively in the eyes, and the extent of eye degeneration was analyzed by determining the fraction of necrotic spots. The screen revealed both suppressors (n = 26; RNAi against these genes enhanced degeneration) as well as enhancers (n = 20; RNAi against these genes reduces degeneration) of the SCA3 eye phenotype (Additional file 3). These results underscore the relevance of astrocytes in the progression of SCA3polyQ78-induced eye degeneration.

One of the genes identified in the screen was NF-κB transcription factor Relish (orthologous to mammalian NF-κB1). Downregulation of Relish expression by RNAi, but also downregulation of several other genes in the Relish pathway, suppressed the SCA3 phenotype. While NF-κB has been implicated in activation of astrocytes in neurodegeneration (reviewed in [6, 26]), a cell-non-autonomous role of NF-κB activation in astrocytes through signals from neurons and subsequent effect on neurodegeneration remains to be determined. In Drosophila, there are two independent NF-κB pathways, activating transcription factors Relish or Dif/Dorsal respectively. We examined expression of anti-microbial peptides (AMPs) in our SCA3 model, transcriptional NF-κB targets which help fight infection [27]. In inflammatory responses, these peptides are considered to aid in activating and recruiting immune cells (reviewed in [28]). While both NF-κB pathways are activated upon eye-specific expression of SCA3polyQ78, activation of Relish was stronger: AMPs specific for Relish (CecA and DptA) were upregulated more than Dif/Dorsal-specific AMPs (IM1 and IM2; Fig. 2a and shown in [29]). Activation of Relish in SCA3polyQ78-expressing eyes occurred in astrocytes: GFP under control of the promoter of a Relish-dependent gene was expressed predominantly in the astrocytes that were present in SCA3polyQ78-expressing eyes (Additional file 4a). In control eyes, no induction of GFP was seen.

Effect of NF-κB transcription factor Relish in astrocytes on SCA3polyQ78-induced degeneration. a Quantification of Relish or Dif/Dorsal activation in the head upon eye-specific expression of SCA3polyQ78. Heads of control flies or flies expressing SCA3polyQ78 in the eyes were analyzed for expression of Relish target genes (CecA, DptA) or Dif/Dorsal target genes (IM-1, IM-2). b Effect of modulation of Relish expression in astrocytes on SCA3polyQ78-induced eye degeneration. Representative images of fly eyes expressing SCA3polyQ78 (−) or SCA3polyQ78 together with either astrocyte-specific expression of two independent RNAi constructs targeting Relish (Relish RNAi #1 and Relish RNAi #2), a Relish overexpression construct (Relish overexpression) or flies heterozygous for Relish (Relish −/+). c Quantification of the fraction of degeneration of the eyes shown in b. In these and other eye quantification experiments, at least 80 eyes of females were counted per genotype. Data are representative of at least three independent experiments ± SEM. d The effect of modulation of Relish expression in astrocytes on GFP fluorescence in SCA3polyQ78-expressing eyes together with eye-specific mCD8-GFP and on the localization of astrocytes (expressing myr-RFP). Representative images of control eyes, eyes expressing mCD8-GFP, and astrocyte-specific myr-RFP (RFP) or together with coexpression of SCA3polyQ78 in the absence or presence of Relish RNAi constructs (Relish RNAi #1 and Relish RNAi #2) or Relish overexpression in astrocytes. e Quantification of the effect of Relish signaling in astrocytes on SCA3polyQ78-induced degeneration by using eye-specific expression of mCD8-GFP. Levels of mCD8-GFP were determined in lysates of fly heads of flies expressing SCA3polyQ78 together with mCD8-GFP in the eyes and the effect of astrocyte-specific expression of Relish RNAi constructs (Relish RNAi #1 and Relish RNAi #2) in astrocytes, heterozygosity for Relish (Relish −/+), or overexpression of Relish in astrocytes was analyzed. Tubulin was used as a control for equal loading. Quantification of western blots is of at least three independent experiments. f Analysis of myr-RFP expression in astrocytes in lysates of control flies or flies expressing myr-RFP in astrocytes together with eye-specific SCA3polyQ78 in the presence or absence of astrocyte-targeted Relish RNAi (Relish RNAi #1 or #2) or Relish overexpression. Quantification of western blots in two independent experiments. Genotypes: a control, GMR-QF2/+. SCA3polyQ78, GMR-QF2/+; QUAS- SCA3polyQ78/+. b, c -, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/+. Relish RNAi #1, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/UAS-Relish RNAi #1. Relish RNAi #2, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/UAS-Relish RNAi #2. Relish overexpression, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/UAS-GFP-Relish; Relish−/+, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4; Relish E20/+. d–f As in b, but with additional eye-specific expression of QUAS-CD8-GFP or astrocyte-specific expression of UAS-myr-RFP

We further focused on the cell-non-autonomous role of astrocyte-specific Relish signaling in SCA3. We first confirmed the results of our screen by using two independent Relish RNAi constructs (efficacy of knockdown, see Additional file 4b) that both decreased SCA3polyQ78-induced degeneration (Fig. 2b). Inversely, the opposite effect was seen upon overexpression of Relish. Modulation of Relish expression in astrocytes did not affect morphology of control eyes (Additional file 4c). In flies heterozygous for Relish, degeneration was attenuated, similar to astrocyte-specific Relish RNAi. The effect of Relish in astrocytes on eye degeneration was quantified (Fig. 2c) by determining the fraction of the eyes containing necrotic spots. Importantly, the effects of Relish on degeneration are cell-non-autonomous, mediated via astrocytes: when coexpressing SCA3polyQ78 and RNAi constructs targeting Relish in the eyes, eye degeneration was not attenuated (Additional file 4d).

When membrane-targeted mCD8-GFP was used as a readout for degeneration [23] in eyes expressing SCA3polyQ78, the protective or enhancing effects of the astrocyte-specific down- and upregulation of Relish levels were confirmed (Fig. 2d for GFP images of different genotypes). To see whether modulating of Relish expression in astrocytes affected the localization of astrocytes, we coexpressed myr-RFP in astrocytes. We did not see an effect of modulation of Relish expression on the localization of the RFP signal (Fig. 2d for RFP images of different genotypes, Fig. 2f for western blot analysis of lysates). These data suggest that Relish in astrocytes does not have an effect on the localization of astrocytes in SCA3polyQ78-expressing eyes (additional images are provided in Additional file 4e). We quantified the effect of modulating Relish expression in astrocytes on mCD8-GFP levels in SCA3polyQ78-expressing eyes by western blot (Fig. 2e). GFP levels in flies heterozygous for Relish were similar to those of Relish RNAi in astrocytes, suggesting the importance of astrocyte-specific Relish signaling in SCA3polyQ78-induced degeneration. The effect of Relish on mCD8-GFP levels in the eye is specifically related to SCA3polyQ78-induced degeneration, as in control flies (not expressing SCA3polyQ78) GFP levels were similar, irrespective of Relish expression (Additional file 4f). Together, these experiments demonstrate that in our SCA3 model, Relish is activated (and Dif/Dorsal to a lesser extent) and that Relish signaling in astrocytes enhances SCA3polyQ78-induced degeneration. Thus, eye-specific expression of SCA3polyQ78 promoted Relish activation in astrocytes, and this Relish activation enhanced degeneration in a cell-non-autonomous manner.

In the brain, Relish is activated in glia in aging flies as well as in flies that are mutant for A-T mutated (ATM), a kinase associated with ensuring genomic integrity and neurodegeneration [30, 31]. We analyzed contribution of astrocytes to overall Relish signaling in the fly head induced by eye-specific expression of SCA3polyQ78. For this, we examined levels of Relish-dependent AMPs in fly heads expressing SCA3polyQ78 in the eyes together with astrocytes-specific targeting of Relish RNAi or Relish overexpression constructs, or in SCA3polyQ78-expressing flies heterozygous for Relish. Expression of Relish RNAi constructs in astrocytes in flies expressing eye-specific SCA3polyQ78 resulted in decreased expression of Relish-specific AMPs (DptA and AttC) in the head to levels comparable to flies not expressing SCA3polyQ78 (Fig. 3a), whereas overexpression of Relish enhanced expression. SCA3 flies heterozygous for Relish expressed levels of Relish target genes comparable to those expressing Relish RNAi constructs in the astrocytes. Dif/Dorsal-specific target genes were unaffected by \modulating Relish levels in astrocytes (Additional file 5). These data highlight the importance of Relish signaling in astrocytes regarding Relish-dependent immune signaling in the head.

Relish signaling in astrocytes influences SCA3polyQ78-induced gene expression but not the extent of SCA3 aggregation. a Analysis of the effect of modulating Relish expression in astrocytes or Relish heterozygosity on Relish activity in heads of flies expressing SCA3polyQ78 in the eyes. Expression of Relish target genes (DptA or AttC) was determined in heads of control flies (control) and flies expressing SCA3polyQ78 in the eyes. The effect of Relish RNAi targeted to astrocytes (Relish RNAi #1 and Relish RNAi #2), Relish overexpression or heterozygosity for Relish (Relish −/+) in flies expressing eye-specific SCA3polyQ78 was determined. b Lysates of fly heads described in (a) were analyzed on Western blot to determine levels of soluble and aggregated HA-tagged SCA3polyQ78. c Quantification of the SDS-soluble/SDS-insoluble ratio of SCA3polyQ78 of three independent experiments shown in b. n.s., not significant. Genotypes in a and b: control, GMR-QF/+; for other genotypes, see Fig. 2b

The presence of aggregated or misfolded proteins is toxic to neurons [32]. Possibly, astrocytes could have an effect on levels of protein aggregates. To see whether Relish-specific signaling in astrocytes could have an effect on solubility of SCA3polyQ78 in the eyes, we collected heads of flies expressing eye-specific SCA3polyQ78 2 days after eclosion and analyzed the cell extract. Modulating Relish expression in astrocytes or heterozygosity for Relish did not affect total levels or solubility of SCA3polyQ78 (Fig. 3b, quantification of three independent experiments in Fig. 3c). This suggests that the effect of Relish on SCA3polyQ78-induced degeneration is not the result of either a clearance of SCA3polyQ78 aggregates or of a shift of the balance of soluble-insoluble SCA3polyQ78 towards more soluble SCA3polyQ78.

To see whether Relish can modulate SCA3-induced eye degeneration via Relish-dependent AMPs, we expressed RNAi constructs targeting two Relish-dependent AMPs (AttA and CecA; efficacy of knockdown in Additional file 6a). Astrocyte-specific downregulation of these AMPs attenuated the degenerative SCA3 eye phenotype (Fig. 4a; quantification of degeneration in Fig. 4b), similar to decreasing Relish activity in astrocytes (Fig. 2a, b). No effect on eye morphology was seen when RNAi targeting AMPs was expressed in astrocytes (Additional file 6b).

Relish target genes influence SCA3polyQ78-induced degeneration but not SCA3polyQ78 aggregation or SCA3polyQ78-induced relocation of astrocytes. a Representative images of flies expressing eye-specific SCA3polyQ78 compared to flies coexpressing RNAi constructs in astrocytes targeting Relish target genes AttA or CecA. b Quantification of the fraction of degeneration of the eyes shown in a. Data are representative of at least three independent experiments. c Representative images of the effect of downregulation of Relish target genes AttA or CecA in astrocytes on SCA3-induced degeneration or the relocation of astrocytes upon expression of eye-specific SCA3polyQ78 by using eye-specific expression of mCD8-GFP and astrocyte-targeted expression of myr-RFP (RFP). d Quantification of the effect of Relish target genes in astrocytes on SCA3-induced degeneration by using membrane-targeted mCD8-GFP. Lysates of fly heads expressing mCD8-GFP and SCA3polyQ78 in the eyes were analyzed on western blot and compared to flies coexpressing RNAi constructs targeting AttA or CecA in astrocytes. Quantifications are of at least three independent experiments. e No effect on Relish target genes on levels of astrocytes expressing myr-RFP. Lysates of fly heads of control flies or flies expressing myr-RFP in astrocytes together with SCA3polyQ78 in the eyes in the presence or absence or AMP RNAi constructs (AttA RNAi or CecA RNAi) were analyzed on western blot for expression for myr-RFP and tubulin. Quantifications are of two independent experiments. f Analysis of the effect of Relish target genes in astrocytes on SDS-solubility of SCA3polyQ78 expressed in the eyes. Head lysates of flies expressing HA-tagged SCA3polyQ78 in the eyes were compared to flies coexpressing RNAi constructs targeting AttA or CecA in astrocytes. Quantifications of the SDS-soluble/SDS-insoluble ratio are of at least three independent experiments. Genotypes a -: GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/+. AttA RNAi, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/UAS-AttA RNAi. CecA RNAi, GMR-QF2/+; QUAS-SCA3polyQ78:: alrm-Gal4/+; UAS-CecA RNAi/+. c, d Con: GMR-QF2/+, the rest of the genotypes as in a but with coexpression of QUAS-mCD8-GFP or UAS-myr-RFP. e, f Con: GMR-QF2/+; other samples as in a

We next analyzed mCD8-GFP levels in the eyes as a measure of degeneration in SCA3polyQ78-expressing eyes and examined the effect of astrocyte-specific knockdown of Relish-specific AMPs. Levels of mCD8-GFP were increased upon knockdown of AMPs, demonstrating attenuation of SCA3polyQ78-induced eye degeneration (Fig. 4c; analysis on western blot and quantification of three independent experiments in Fig. 4d). In control eyes, no effects of downregulation of Relish-specific AMPs expression in astrocytes on mCD8-GFP levels in the eyes were found (Additional file 6c). We also examined whether astrocyte-specific knockdown of Relish-specific AMPs could influence the presence of astrocytes in SCA3polyQ78-expressing eyes, but did not see an effect (Fig. 4c), and the numbers of astrocytes were also not affected, as analyzed by levels of myr-RFP in astrocytes (Fig. 4e).

When we examined the effect of astrocyte-specific downregulation of AMPs on SCA3polyQ78 levels or solubility, we did not observe an effect (Fig. 4f; quantification of three independent experiments), similar to modulating Relish levels in astrocytes. These data show that the effect of Relish on degeneration in our SCA3 model is, to a significant extent, mediated via the generation of AMPs, and that the effect of AMPs does not occur via alterations in SCA3polyQ78 solubility.

We next tested whether the observed effects of the Relish pathway in astrocytes on (short term) degeneration also translates into a (long term) survival benefit in flies expressing SCA3polyQ78 in neurons. Astrocyte-specific Relish inhibition may also extend lifespan in neurons expressing SCA3polyQ78, given our observations of Relish in astrocytes on the SCA3 eye phenotype. We expressed SCA3polyQ78 pan-neuronally (not in astrocytes) and examined effects of modulation of Relish expression specifically in astrocytes (specificity of expression shown in Additional file 7a) using the following two independent binary expression systems (Additional file 7b): UAS-Gal4 and QUAS-QF2. To avoid potential detrimental effects of SCA3polyQ78 during development, we expressed SCA3polyQ78 in adult flies only, using the inducible “Q system,” in which the neuronally expressed transcription factor (QF2) that induces SCA3polyQ78 expression is coexpressed with QF-suppressor QS [20] (Additional file 7b). This way, expression of SCA3polyQ78 is suppressed during development (Fig. 5a). Feeding quinic acid suppresses QS, thus alleviating inhibition of QF2 resulting in the neuron-specific expression of SCA3polyQ78, of which the extent of aggregation increased over time (Fig. 5a). The lifespan of flies expressing SCA3polyQ78 in neurons was significantly shortened (Fig. 5b). Quinic acid by itself did not affect lifespan (Additional file 7c), excluding non-specific effects. Relish-dependent gene expression was increased in flies that expressed SCA3polyQ78 in neurons (Fig. 5c), similar to expression of SCA3polyQ78 in the eyes (Fig. 2a).

Analysis of SCA3polyQ78 expression in neurons and the effect of Relish in astrocytes on neuronal expression of SCA3polyQ78 or Abeta42. a Inducible expression of SCA3polyQ78 in neurons can be suppressed during development. Control larvae or larvae expressing inducible SCA3polyQ78 were cultured on food with and without quinic acid. Flies expressing inducible SCA3polyQ78 were cultured on food with quinic acid for the times indicated to induce SCA3polyQ78 expression. Levels and aggregation of HA-tagged SCA3polyQ78 were determined in lysates of larvae or fly heads on western blot. Tubulin was used to verify equal loading. b Effect of inducible SCA3polyQ78 on lifespan. Control flies or 4-day-old flies expressing inducible SCA3polyQ78 were cultured on food containing quinic acid and the lifespan was analyzed. c Expression of Relish target genes AttC, DptA and CecA was analyzed in heads of 15-day-old flies expressing SCA3polyQ78 in neurons. d Analysis of modulation of Relish expression specifically in astrocytes on the lifespan of flies expressing SCA3polyQ78 in neurons. Flies expressing SCA3polyQ78 in neurons induced as in b were compared to flies coexpressing astrocyte-targeted Relish RNAi (Relish RNAi #1 or #2) or overexpressing Relish in astrocytes. To induce changes in Relish expression only in adult flies, crosses were kept at 18 °C. 4-day-old progeny were shifted to 25 °C to allow expression of Relish constructs in astrocytes. e Downregulation of Relish expression in astrocytes extends lifespan in an Alzheimer model. Lifespan of control flies and flies expressing Abeta42 in neurons were compared to flies expressing Abeta42 (Aβ42) in neurons and Relish RNAi constructs in astrocytes. Induction of Relish RNAi was done as in d. Genotypes: a -, tub-QS/+; alrm-Gal4/+; nSyb-QF2::tub-QS/+. SCA3polyQ78, tub-QS/+; alrm-Gal4::QUAS-SCA3polyQ78/+; nSyb-QF2::tub-QS/+. b, c Control, tub-QS/+; alrm-Gal4/+; nSyb-QF2::tub-QS/+. SCA3polyQ78, tub-QS/+; alrm-Gal4::QUAS-SCA3polyQ78/+; nSyb-QF2::tub-QS/+. d -, tub-QS/+; alrm-Gal4/+; nSyb-QF2::tub-QS/+. SCA3polyQ78, tub-QS/+; alrm-Gal4::QUAS-SCA3polyQ78/+; nSyb-QF::tub-QS/+; Relish RNAi #1, Relish RNAi #2, or Relish overexpression, as in SCA3polyQ78, but with expression of UAS-Relish RNAi #1 or #2 or UAS-GFP-Relish in astrocytes. e -, alrm-Gal4/+; nSyb-QF2. Aβ42: alrm-Gal4::QUAS-Abeta42/+; nSyb-QF2/+. Relish RNAi (#1 or #2), alrm-Gal4:: QUAS-Abeta42/UAS-Relish RNAi #1 or #2); nSyb-QF2/+

We next wished to examine whether modulating Relish levels in astrocytes in adult flies could have an effect on lifespan of flies that expressed SCA3polyQ78 pan-neuronally. For this, levels of Relish were modulated in astrocytes of adult flies to exclude effects on development. Hereto, we used the transcription factor Gal4 to drive the expression of the Relish constructs. At a lower temperature (18 °C), activity of Gal4 is low and Relish RNAi constructs do not decrease Relish expression (Additional file 7d). 4 days after eclosion, we induced SCA3polyQ78 expression in neurons by adding quinic acid to the food. Flies were shifted to 25 °C to induce expression of the Relish constructs in astrocytes. Importantly, reducing Relish expression in astrocytes extended the lifespan of SCA3 flies, whereas overexpression shortened the lifespan (Fig. 5d). In control flies, downregulating Relish expression in astrocytes had no effect on lifespan (Additional file 7e), indicating that the effects of Relish RNAi are linked to SCA3polyQ78-mediated degeneration of neurons. However, overexpression of Relish alone in astrocytes significantly shortened lifespan (Additional file 7e), precluding analysis on Relish overexpression on SCA3polyQ78-mediated shortening of lifespan.

In parallel, SCA3-related effects on motor function were analyzed, measured as climbing ability. The SCA3polyQ78-induced decrease in climbing ability was partially alleviated by knockdown of Relish in astrocytes (Additional file 7f). As was observed with modulating Relish expression in astrocytes in SCA3polyQ78-expressing eyes, no effects of Relish in astrocytes on total brain SCA3polyQ78 aggregation load or levels could be detected (Additional file 7g).

The effects of Relish modulation on SCA3-related degeneration were not related to alterations of the levels of SCA3polyQ78 aggregates (Fig. 3b and Additional file 7g). Therefore, we hypothesized that modulation of the Relish pathway in astrocytes could be more generic and relevant to a neurodegenerative disease associated with the presence of extraneuronal aggregates, such as in Alzheimer’s disease. For this, we turned to an established Drosophila model of Alzheimer’s disease [18], in which the Amyloid beta (Abeta42; Aβ42) peptides are fused to a secretion signal peptide (of the necrotic gene), resulting in the secretion of Aβ42 peptides. When expressed in neurons (Additional file 7h), these Aβ42 peptides are secreted and form extracellular aggregates in flies, reminiscent of the plaques in human AD [33]. In this AD fly model, flies displayed progressive neurodegeneration, with mobility and memory deficits, and reduced lifespan [18, 34]. In line with our hypothesis, the lifespan of flies expressing Aβ42 in neurons was shortened, but extended when Relish expression was suppressed in astrocytes, using two independent RNAi lines targeting Relish (Fig. 5e). This, together with the data on SCA3, demonstrates that modulating Relish signaling in astrocytes has beneficial effects in neurodegeneration, irrespective of the aggregation process that initiates it.

Our data show that astrocytes can modulate the rate of neurodegenerative disease progression in a cell-non-autonomous manner. This is the first time to show that neuron-specific expression of a pathological, degeneration-inducing protein results in activation of astrocytes, which can subsequently influence symptoms of degeneration.