Impaired proteasomal degradation enhances autophagy via hypoxia signaling in Drosophila

Abstract

Background: Two pathways are responsible for the majority of regulated protein catabolism in eukaryotic cells: the ubiquitin-proteasome system (UPS) and lysosomal self-degradation through autophagy. Both processes are necessary for cellular homeostasis by ensuring continuous turnover and quality control of most intracellular proteins. Recent studies established that both UPS and autophagy are capable of selectively eliminating ubiquitinated proteins and that autophagy may partially compensate for the lack of proteasomal degradation, but the molecular links between these pathways are poorly characterized.

Results: Here we show that autophagy is enhanced by the silencing of genes encoding various proteasome subunits (α, β or regulatory) in larval fat body cells. Proteasome inactivation induces canonical autophagy, as it depends on core autophagy genes Atg1, Vps34, Atg9, Atg4 and Atg12. Large-scale accumulation of aggregates containing p62 and ubiquitinated proteins is observed in proteasome RNAi cells. Importantly, overexpressed Atg8a reporters are captured into the cytoplasmic aggregates, but these do not represent autophagosomes. Loss of p62 does not block autophagy upregulation upon proteasome impairment, suggesting that compensatory autophagy is not simply due to the buildup of excess cargo. One of the best characterized substrates of UPS is the α subunit of hypoxia-inducible transcription factor 1 (HIF-1α), which is continuously degraded by the proteasome during normoxic conditions. Hypoxia is a known trigger of autophagy in mammalian cells, and we show that genetic activation of hypoxia signaling also induces autophagy in Drosophila. Moreover, we find that proteasome inactivation-induced autophagy requires sima, the Drosophila ortholog of HIF-1α.

Conclusions: We have characterized proteasome inactivation- and hypoxia signaling-induced autophagy in the commonly used larval Drosophila fat body model. Activation of both autophagy and hypoxia signaling was implicated in various cancers, and mutations affecting genes encoding UPS enzymes have recently been suggested to cause renal cancer. Our studies identify a novel genetic link that may play an important role in that context, as HIF-1α/sima may contribute to upregulation of autophagy by impaired proteasomal activity.

Figure 1

Depletion of genes encoding various proteasome subunits decreases proteasomal activity in vivo. A) Control cells expressing the proteasome degradation reporter GFP-CL1 are similarly sized as neighboring non-GFP control fat body cells and show diffuse fluorescence. Cell nuclei are labeled with DAPI (blue). B)Rpn2 RNAi cells are smaller than control cells and accumulate large aggregates of GFP-CL1. C) Proteasome RNAi leads to accumulation of GFP-CL1 aggregates relative to control cells shown in panel A. Statistically significant differences are marked by asterisks (Kruskal-Wallis test, n = 5-6 per genotype, ** P < 0.01, * P < 0.05), and error bars denote standard error. The different types of subunits are indicated as 20S core subunits (α and β), and 19S regulatory particle subunits (ATPase and non-ATPase). Pomp is required for assembly of the 20S core. D) Schematic of the clonal expression system. GFP-positive cells of interest are surrounded by wild-type cells in the same tissue of mosaic animals, serving as an internal control in various staining experiments. E) Depletion of Rpn2 leads to accumulation of ubiquitinated protein aggregates in Lamp1-GFP marked fat body cell clones. Note that the numerous small Lamp1-GFP dots representing lysosomes are not ubiquitin-positive. F) Similarly, silencing of Rpn2 leads to cell-autonomous accumulation of p62 aggregates in Lamp1-GFP marked fat body cell clones. Note that Lamp1-GFP dots are not p62-positive either. G) Proteasome RNAi leads to accumulation of endogenous p62 aggregates relative to control cells. n = 5-6 per genotype. Statistically significant differences are marked by asterisks (Kruskal-Wallis test, n = 5-6 per genotype, ** P < 0.01, * P < 0.05), and error bars denote standard error. Boxed areas in E and F are shown enlarged. Scale bar in A equals 20 μm for A, B, E, F.

Figure 2

Proteasome impairment enhances autophagy in Drosophila larvae. A) Depletion of Rpn2 increases punctate Lysotracker Red (LTR) staining in Lamp1-GFP marked larval fat body cell clones of starved animals relative to surrounding control cells. B)Rpn2 knockdown induces autophagy in fat body cell clones of well fed larvae, as LTR dots are only observed in GFP-positive RNAi cells but not in control cells. C) Silencing of genes encoding different proteasome subunits results in a statistically significant induction of punctate LTR in well fed conditions (u test, n = 5-7 per genotype, ** P < 0.01), and error bars denote standard error. D) Immunostaining reveals cell-autonomous activation of punctate endogenous Atg8a labeling in Rpn2 RNAi cells, representing autophagosomes. E) Statistical evaluation of the effect of proteasome subunit RNAi on punctate endogenous Atg8a. Statistically significant differences are marked (u or t test, n = 5 per genotype, ** P < 0.01), and error bars denote standard error. (F) Western blots show that RNAi knockdown of genes encoding different proteasome subunits greatly increases the levels of both the specific autophagy cargo p62 and autophagosome-associated lipidated Atg8a-II. Numbers refer to relative expression levels compared to Tubulin, as determined by densitometric evaluation. Boxed areas in A, B and D are shown enlarged. Scale bar in A equals 20 μm for A, B, D.

Figure 3

Proteasome impairment leads to accumulation of cytoplasmic aggregates and enhances autophagic flux. A) Overexpressed GFP-Atg8a reporter is incorporated into the large protein aggregates containing p62 in Rpn2 RNAi cells. B) The tandemly tagged mCherry-GFP-Atg8a reporter forms numerous small mCherry-labeled autolysosomes, in addition to large aggregates positive for both mCherry and GFP in Rpn2 RNAi cells. C) The lysosome inhibitor chloroquine blocks autophagy-dependent quenching of GFP, as now most puncta are positive for both mCherry and GFP. D) Quantification of data from panels B and C (u or t test, n = 5-8 per genotype, ** P < 0.01, * P < 0.05), and error bars denote standard error. E) Transmission electron microscopy reveals the presence of autolysosomes (arrow) and large protein aggregates in fat body cells with impaired proteasome function in well-fed larvae. The boxed area is shown enlarged in panel E’. F) Quantification of ultrastructural data. Protein aggregates occupy 7% of the total cytoplasm in Rpn2 depleted fat body cells of well-fed larvae, and double-membrane autophagosomes (AP) and degrading autolysosomes (AL) take up 0.06% and 0.24% of the total cytoplasm, respectively. No such structures are recognized in fat body cells of well-fed control larvae (u test, n = 3 per genotype, ** P < 0.01), and error bars denote standard error. Boxed areas in A-C are shown enlarged. Scale bar in A equals 20 μm for A-C. Scale bars equal 1 μm in E, E’.

Figure 4

Activation of hypoxia signaling induces autophagy in Drosophila. A-C) Overexpression of the Drosophila HIF-1α ortholog sima (A) or depletion of Vhl(B) induces the formation of autolysosomes positive for both Lamp1-GFP and LTR in fat body cell clones in well-fed larvae. Quantification of data is shown in panel C. Statistically significant differences are marked (u test, n = 6 per genotype, ** P < 0.01), and error bars denote standard error. D-F) Overexpression of sima (D) or knockdown of Vhl(E) promotes the formation of punctate mCherry-Atg8a in fat body cell clones marked by GFP-nls (nuclear localization sequence) in well-fed larvae. Quantification of data is shown in panel F. Statistically significant differences are marked (u test, n = 6 per genotype, ** P < 0.01), and error bars denote standard error. G-J) Overexpression of sima (G) or depletion of Vhl(H) leads to formation of dots that are mostly positive for mCherry with the mCherry-GFP-Atg8a reporter. Chloroquine treatment blocks the autolysosomal quenching of GFP, as puncta are now positive for both mCherry and GFP in cells overexpressing sima (I) or Vhl RNAi (J). K) Quantification of data from panels G-J. Statistically significant differences are marked (u or t test, n = 5-8 per genotype, ** P < 0.01, * P < 0.05), and error bars denote standard error. L) RT-PCR analyses show that the transcription of BNIP3 is upregulated in both Vhl and Rpn2 RNAi samples relative to controls. Boxed areas in A, B, D, E, G-J are shown enlarged. Scale bar in A equals 20 μm for A, B, D, E, G-J.

Figure 5

Proteasome inactivation-induced autophagy requires Atg genes and sima/HIF-1α. A) Depletion of Rpn2 increases punctate LTR staining in larval fat body cell clones in fed animals, marked by membrane-associated mCD8-GFP. B) Simultaneous silencing of Rpn2 and sima results in a block of punctate LTR staining. C) Overexpression of dominant-negative (DN) Atg4 inhibits LTR dot formation in Rpn2 RNAi cells. D, E) Simultaneous silencing of Rpn2 and p62 using two independent RNAi lines combined (D) or a third independent single p62 RNAi line (E) does not inhibit punctate LTR staining in larval fat body cell clones in fed animals. F) Depletion of p62 blocks aggregate formation in Rpn2 RNAi cells. G) Statistical evaluation of punctate staining in Rpn2, Rpt1 and Prosβ2 knockdown cells. Depletion of sima using 3 independent RNAi transgenes, silencing of BNIP3, Atg1, Atg9, Atg12, or overexpression of dominant-negative Atg1, Vps34, Atg4 inhibits proteasome inhibition-induced autophagy. Depletion of p62 or Atg18b (which is dispensable for starvation-induced autophagy and used as a negative control here) does not block the autophagy-inducing effect of proteasome inactivation. Statistically significant differences are marked (Kruskal-Wallis test or ANOVA, n = 5-8 per genotype, * P < 0.05, ** P < 0.01), and error bars denote standard error. H) Similarly, silencing of sima but not p62 attenuates LTR puncta formation in Prosα1 and Prosα5 RNAi cells. Statistically significant differences are marked (Kruskal-Wallis test or ANOVA, n = 5-8 per genotype, * P < 0.05, ** P < 0.01), and error bars denote standard error. Boxed areas in A-F are shown enlarged. Scale bar in A equals 20 μm for A-F.