1. Cell Biology
  2. Chromosomes and Gene Expression

PGAM5 promotes lasting FoxO activation after developmental mitochondrial stress and extends lifespan in Drosophila

  1. Martin Borch Jensen
  2. Yanyan Qi
  3. Rebeccah Riley
  4. Liya Rabkina
  5. Heinrich Jasper  Is a corresponding author
  1. Buck Institute for Research on Aging, United States
  2. Genentech, United States
https://doi.org/10.7554/eLife.26952
Share this article
Cite this article
  1. Martin Borch Jensen
  2. Yanyan Qi
  3. Rebeccah Riley
  4. Liya Rabkina
  5. Heinrich Jasper
(2017)
PGAM5 promotes lasting FoxO activation after developmental mitochondrial stress and extends lifespan in Drosophila
eLife 6:e26952.
https://doi.org/10.7554/eLife.26952
  2. Download BibTeX
  3. Download .RIS

Abstract

The mitochondrial unfolded protein response (UPRmt) has been associated with long lifespan across metazoans. In Caenorhabditis elegans, mild developmental mitochondrial stress activates UPRmt reporters and extends lifespan. We show that similar developmental stress is necessary and sufficient to extend Drosophila lifespan, and identify Phosphoglycerate Mutase 5 (PGAM5) as a mediator of this response. Developmental mitochondrial stress leads to activation of FoxO, via Apoptosis Signal-regulating Kinase 1 (ASK1) and Jun-N-terminal Kinase (JNK). This activation persists into adulthood and induces a select set of chaperones, many of which have been implicated in lifespan extension in flies. Persistent FoxO activation can be reversed by a high-protein diet in adulthood, through mTORC1 and GCN-2 activity. Accordingly, the observed lifespan extension is prevented on a high-protein diet and in FoxO-null flies. The diet-sensitivity of this pathway has important implications for interventions that seek to engage the UPRmt to improve metabolic health and longevity.

https://doi.org/10.7554/eLife.26952.001

Introduction

A wide range of studies in genetically accessible model systems have led to the realization that aging is a malleable process, responsive to both genetic and pharmacological interventions. An integrated view of the aging process has emerged from these efforts, spurred by the identification of a select group of biological processes and pathways that drive, influence, and regulate the physical decline characteristic of the aging process (Kennedy et al., 2014; López-Otín et al., 2013). Many of these pathways involve mitochondria: either through their role in metabolism (López-Otín et al., 2016), as a source of reactive oxygen species (Balaban et al., 2005), or as signaling hubs (Chandel, 2015).

In recent years, the mitochondrial unfolded protein response (UPRmt) has emerged as a unifying mechanism for several of these pathways (Jensen and Jasper, 2014). As the name implies, the UPRmt is a conserved cellular mechanism that serves to restore proteostasis in mitochondria. The response was first described in mammalian cells more than a decade ago (Martinus et al., 1996; Zhao et al., 2002) and has since been studied primarily in Caenorhabditis elegans. In worms, strong evidence suggests that the UPRmt is involved in delaying aging and promoting adult lifespan (Baker et al., 2012; Durieux et al., 2011; Houtkooper et al., 2013; Merkwirth et al., 2016; Mouchiroud et al., 2013; Tian et al., 2016; Yang and Hekimi, 2010). This work has revealed the UPRmt to be reminiscent of, but distinct from, the cytoplasmic (heat shock) and endoplasmic reticulum unfolded protein responses (Baker et al., 2012; Haynes et al., 2007; 2010). Through the UPRmt, mitochondrial stress induces a nuclear transcriptional response that promotes the expression of a group of mitochondrial chaperones and proteases (Aldridge et al., 2007; Yoneda et al., 2004). The primary transcription factor in C. elegans is ATFS-1, which is regulated by membrane potential-dependent import into and degradation in mitochondria (Nargund et al., 2012). Mitochondrial stress blocks this import and ATFS-1 instead moves to the nucleus, where it interacts with DVE-1 and UBL-5 to turn on the transcriptional response (Benedetti et al., 2006; Haynes et al., 2007). In addition to proteostatic elements, this response includes a metabolic reconfiguration to increase glycolytic capacity while restoring oxidative phosphorylation (Nargund et al., 2015). Meanwhile, a separate branch of the UPRmt mediates a general downregulation of translation, through GCN-2 and eIF2α (Baker et al., 2012).

Evidence for the evolutionary conservation of the UPRmt has emerged in recent years. In mice, perturbation of mitochondrial translation has been implicated in long lifespan (Houtkooper et al., 2013), and recent findings have revealed the conservation of the ATFS-1 regulated transcriptional response (mediated by ATF5 in mice (Fiorese et al., 2016)). In Drosophila, a UPRmt-like response was first described in a paradigm where a misfolding ornithine transcarbamylase (ΔOTC) was overexpressed, resulting in upregulation of mitochondrial chaperones and induction of mitophagy (Pimenta de Castro et al., 2012). Furthermore, knocking down electron transport chain (ETC) components has been shown to extend lifespan and to induce the UPRmt (Copeland et al., 2009; Owusu-Ansah et al., 2013). However, the signaling pathway mediating UPRmt activation in Drosophila remains to be clarified. ETC knockdown results in induction of the insulin signaling inhibitor ImpL2, and promotes the expression of target genes of the insulin-regulated transcription factor Forkhead Box O (FoxO) (Owusu-Ansah et al., 2013). FoxO activation is a well-established lifespan-extending condition, yet its involvement and specific regulation in this context remain to be established (Kappeler et al., 2008; Kenyon et al., 1993; Kimura et al., 1997; Selman et al., 2008; Tatar et al., 2001).

Activation of the UPRmt has also been implicated in the lifespan-extending effects of various drugs, including resveratrol (Houtkooper et al., 2013), rapamycin (Houtkooper et al., 2013) and NAD precursors (Mouchiroud et al., 2013). Nevertheless, it is not clear which cellular consequences of the UPRmt contribute to organismal health, nor whether resistance to a particular type of stress underlies lifespan extension by the UPRmt. Indeed, the generic view that stimulating this response invariably leads to longer life has been called into question (Bennett et al., 2014). Understanding how conserved elements of UPRmt signaling connect to longevity pathways will be useful in resolving this contention.

An interesting observation from both fly and worm studies is that lifespan extension by the UPRmt is contingent upon its activation during development (Durieux et al., 2011; Owusu-Ansah et al., 2013). This is consistent with earlier observations that lifespan extension by disrupting the ETC in some cases require developmental treatment (Copeland et al., 2009; Dillin et al., 2002; Rea et al., 2007). An important step toward understanding this lasting effect came from work in worms, in which the UPRmt leads to changes in histone methylation and chromatin organization (Merkwirth et al., 2016; Tian et al., 2016). The H3K27 demethylases jmjd-1.2 and −3.1 were found to be required for UPRmt activation and lifespan extension, while the histone methylase met-2 is required for induction of most UPRmt genes. This effect involves global chromatin condensation, while opening up specific sites for occupation by DVE-1 (Tian et al., 2016). The loci thus revealed have yet to be characterized, so the lasting cellular changes and longevity pathways influenced by these epigenetic mechanisms remain unclear.

Here, we have explored the signaling pathway regulating the transcriptional response to mitochondrial proteostatic stress in Drosophila, and have identified a role for persistent FoxO activation in promoting longevity after developmental mitochondrial stress. We find that Phosphoglycerate Mutase 5 (PGAM5) is required to activate a pathway that includes ASK1, JNK, Relish and FoxO and promotes protective gene expression in response to mitochondrial stress. Activation of this pathway during development leads to persistent FoxO activation and increased expression of chaperones in adult flies, and is required for longevity. We further find that lasting FoxO activation is sensitive to dietary conditions, as it can be abolished by elevated protein intake and elevated mTORC1 and GCN-2 activity. Our findings identify a new diet-sensitive pathway of lifespan regulation by mitochondrial stress. Since the identified pathway components are evolutionarily conserved, we anticipate that these results inform our understanding of similar interactions in vertebrate systems, including humans.

Results

Acute mitochondrial stress upregulates proteostatic, immune and stress signaling pathways through PGAM5- and JNK-dependent FoxO activation

To gain an overview of the gene expression changes induced by the UPRmt in Drosophila, we sequenced mRNA extracted from adult female thoracic tissue after 24 hr of mitochondrial stress. We induced mitochondrial stress ubiquitously, using GeneSwitch for temporal control, either by overexpressing a misfolding variant of the human mitochondrial enzyme ornithine transcarbamylase (ΔOTC) (Pimenta de Castro et al., 2012) or by knocking down the ETC complex I component ND75 (Owusu-Ansah et al., 2013). 23% of the ΔOTC-induced genes and 15% of the genes downregulated by ΔOTC expression were co-regulated by both conditions, and we propose that these genes compose a core UPRmt in Drosophila (Figure 1a). The main functional categories enriched among the induced genes were: several stress signaling pathways, chaperones and proteases, and the innate immune system (Figure 1b). Activation of immune responses is broad in this case, encompassing target genes of both the Toll and Imd pathways.

Figure 1 with 3 supplements see all
Download asset Open asset
The Drosophila signaling pathway and transcriptional response to mitochondrial proteostatic stress.

(A) Mitochondrial proteostatic stress was triggered by overexpression of a misfolding mitochondrial protein (ΔOTC) or knockdown of the complex I subunit ND75, as previously reported. The ubiquitous DaGS::GeneSwitch driver was activated for 24 hr at day 7 of adulthood, and female thoracic tissue collected for RNAseq analysis. Venn and bar diagrams show genes upregulated 3-fold/downregulated 2-fold with both treatments relative to controls, and with control FKPM > 10. Full raw data found in Supplementary File 1. (B) Gene Ontology analysis of the shared stress response; upregulated genes include chaperones, stress pathways, and antimicrobial peptides/innate immune genes. Redundant GO terms were trimmed using REVIGO. (C) Mitochondrial proteostatic stress response genes from RNAseq analysis verify by qPCR and are not activated by heat shock or induction of the ER-UPR through RNAi of Hrd1. p-values for ΔOTC ± RU486 are 0.109, 0.130, 0.255 and 0.008. (D) Knocking down the transcription factors FoxO or Relish, or the JNK kinase, during ΔOTC expression blocks induction of response genes. The kinase ASK1 and the mitochondrial membrane protein PGAM5 are also required for the mitochondrial proteostatic stress response. See supplementary for verification of additional UPRmt genes. p-values relative to ΔOTC + RU are (left to right)<0.0001,<0.0001,<0.0001,<0.0001,<0.0001, 0.0185 and 0.9888; 0.002,<0.0001,<0.0001, and <0.0001. (E) Proposed signaling pathway for immune activation by mitochondrial proteostatic stress. All error bars are SEM of 3+ independent experiments. ANOVA with Dunnett’s multiple comparisons test; *p<0.05, **p<0.01, then each *=0.1 x.. 200 µM RU486 was used in all experiments.

https://doi.org/10.7554/eLife.26952.002
Figure 1—source data 1

The Drosophila signaling pathway and transcriptional response to mitochondrial proteostatic stress.

https://doi.org/10.7554/eLife.26952.006

We confirmed the induction of several immune genes by quantitative real-time PCR (qPCR), and further demonstrated that transcriptional induction of these genes was specifically triggered by mitochondrial stress and not by conditions that trigger the cytoplasmic- (heat shock) or endoplasmic reticulum (loss of ER-associated degradation after knockdown of Hrd1, (Bordallo et al., 1998)) unfolded protein responses (Figure 1c). We also used qPCR to confirm that ΔOTC expression leads to induction of mitochondrial chaperones and proteases associated with the UPRmt, under conditions used in earlier studies (Figure 1—figure supplement 1a). As further confirmation of immune activation, we tested the ability to fight off pathogenic bacteria introduced either orally or by abdominal pricking, following 24 hr of ΔOTC expression. Resistance to oral infection was mildly increased, while the speed by which flies succumb to humoral infection was unaffected; we speculate that this may be because the modest induction of immune genes is not sufficient to clear bacteria introduced directly to the hemolymph (Figure 1—figure supplement 2a). Initial experiments confirmed these phenotypes in both male and female flies (data not shown), and we opted to use females for all following experiments.

Activation of a transcriptional program that includes immune response genes by the UPRmt has previously been noted in worms, where both immune and proteostatic gene induction is dependent on the ATFS-1 transcription factor (Nargund et al., 2012; Pellegrino et al., 2014). Since a homologous mediator of the response to mitochondrial stress in flies had not been described, we set out to identify transcription factors that could mediate the UPRmt in this organism. We performed a targeted RNAi screen of transcription factors known to regulate immune and stress genes. Of the two methods initially used to induce the mitochondrial proteostatic stress, we opted to use ΔOTC for further experiments to minimize potential secondary effects of inducing ETC dysfunction. Because of its robust, specific and stable induction, we focused on the expression of the antimicrobial peptide metchnikowin (Mtk) as a reporter of mitochondrial stress. Knocking down the transcription factors foxo and Relish/NF-κB, but not jun or kayak/Fos, strongly suppressed Mtk induction (Figure 1d), consistent with upregulation of several FoxO target genes in our RNAseq data (Figure 1a). Because RNAseq further suggested activation of the c-Jun N-terminal kinase (JNK) pathway, and since JNK has been reported to regulate FoxO (Wang et al., 2005), we asked whether knocking down the Drosophila JNK basket (bsk) would prevent Mtk induction. Indeed, Mtk induction by mitochondrial stress was significantly impaired in bsk loss-of-function conditions (Figure 1d).

We next explored mechanisms by which mitochondrial stress could activate JNK. One of the JNK-activating kinases, Apoptosis Signaling Kinase 1 (ASK1), had been reported to interact with and be activated by the mitochondrial Serine/Threonine phosphatase Phosphoglycerate Mutase 5 (PGAM5) (Takeda et al., 2009). Knocking down ASK1 and PGAM5 blocked induction of Mtk after mitochondrial stress, suggesting that PGAM5, ASK1, and JNK are components of a pathway responsible for transducing mitochondrial stress to downstream transcription factors (Figure 1d). The lack of Mtk induction in PGAM5 homozygous null mutants (PGAM51/1) experiencing mitochondrial stress confirmed this finding (Figure 1d). To determine whether this pathway indeed represented UPRmt signaling, we tested and confirmed that its disruption also blocks induction of classical UPRmt genes such as hsp60 and hsp10 (Figure 1—figure supplement 1b–d).

We then asked whether PGAM5 or ASK1 also play a role in canonical pathways for antimicrobial peptide (AMP) induction, and measured induction of Mtk following exposure to the pathogenic Pseudomonas entomophila. Ubiquitously knocking down either PGAM5 or ASK1 had no effect on this response (Figure 1—figure supplement 2b), suggesting that the pathway mediating AMP induction in response to mitochondrial stress is distinct from the canonical Toll and Imd pathways.

Because PGAM5 has been reported to regulate apoptosis (Ishida et al., 2012; Wang et al., 2012), we further investigated whether ΔOTC-induced stress signaling was the result of increased apoptosis. Although PGAM5 null flies showed higher baseline levels of apoptosis, in line with previous reports (Ishida et al., 2012), ΔOTC expression stress did not induce apoptosis in our experiments (Figure 1—figure supplement 2c).

Since ASK1 can be activated by redox signaling (Saitoh et al., 1998), we also tested whether mitochondrial reactive oxygen species (ROS) production was required for transcriptional activation of AMPs. We blocked mitochondrial ROS production from sites on complexes I and III of the ETC using previously characterized compounds (Brand et al., 2016; Orr et al., 2013) but did not see any effect on Mtk induction after mitochondrial stress (Figure 1—figure supplement 3a). Along these lines, overexpression of the antioxidant enzymes Catalase and jafrac1/Trx-1 during developmental ΔOTC expression does not affect acute or lasting FoxO activation (Figure 1—figure supplement 3b & c). This suggests that direct activation of ASK1 by ROS is not the basis of ΔOTC-mediated AMP induction.

Together, these data indicate that the transcriptional response to mitochondrial unfolded proteins in flies is mediated by a pathway involving PGAM5 and ASK1, as well as JNK, Relish and FoxO (Figure 1e).

Mitochondrial proteostatic stress does not induce longevity via improved adult immune function, or via changes to the microbiome

In both flies and worms, the UPRmt has been reported to extend lifespan when activated during development (Durieux et al., 2011; Owusu-Ansah et al., 2013). To confirm this observation in our system, we used the ubiquitous RU486-inducible daughterless GeneSwitch (DaGS) driver to express ΔOTC throughout larval development only, and measured adult lifespan. As shown in Figure 2a,g—figure supplement 2, this treatment consistently extended maximum and median lifespan. Confirming observations in C. elegans, ΔOTC expression limited to adulthood did not lead to extended lifespan (Figure 2b). To determine whether the pathway we identified in the experiments described above is required for this longevity effect, we assessed lifespan in PGAM5 homozygous null flies (Figure 2c). The absence of PGAM5 prevented lifespan extension after developmental ΔOTC expression, suggesting that the protective effects responsible for lifespan extension are downstream of PGAM5. In line with this, the adult lifespan of FoxO double heterozygous null flies was significantly reduced after developmental ΔOTC expression (Figure 2d). Both PGAM5 and especially FoxO nulls also showed reduced survival through larval development when subjected to mitochondrial proteostatic stress (Figure 2—figure supplement 1). This suggests that FoxO activity plays an integral part in promoting resilience to mitochondrial stress in these conditions.

Figure 2 with 4 supplements see all
Download asset Open asset
Developmental mitochondrial stress extends adult lifespan.

(A) Expressing ΔOTC throughout larval development extends median and maximal adult lifespan by 21/14%, respectively. RU486 at 200 µM does not affect lifespan in genetic control flies. (B) Expressing ΔOTC throughout adulthood does not extend lifespan. (C) Null mutants for the mitochondrial protein PGAM5 do not live longer following developmental ΔOTC expression. (D) Null mutants for FoxO do not live longer following developmental ΔOTC expression. These flies also showed strongly reduced survival through developmental mitochondrial stress (see supplementary). (E) Flies given antibiotics throughout adulthood still show lifespan extension by developmental ΔOTC expression. (F) Flies raised in a sterile environment from egg stage still show lifespan extension by developmental ΔOTC expression. (G) Statistical analysis of lifespan experiments. Each graph contains data pooled from 2 + independent experiments. Individual experiments, mortality graphs and life tables are shown in Supplement 4 and File 2.

https://doi.org/10.7554/eLife.26952.007

In worms, the UPRmt also induces the expression of immune genes (Pellegrino et al., 2014), but it has not been explored whether UPRmt-induced longevity is a result of improved immune function. Since our data showed that the pathway responsible for activating immune-response genes during mitochondrial stress is also required for longevity, we hypothesized that the UPRmt might extend lifespan by improving the ability to fight off infections and/or by altering the microbiome. To test this hypothesis, we repeated our lifespan experiments in two conditions where flies are not exposed to microbes; in one condition, eggs were washed in bleach (HClO) to eliminate microbes, then transferred to a sterile hood and reared on autoclaved food throughout development and adult life. In the second condition, we added a cocktail of antibiotics shown to eliminate all culturable strains of the microbiome (Li et al., 2016) to the food of adult flies. Both conditions prevent pathogen exposure and eliminate the adult microbiome, while the first condition further prevents the formation of a larval microbiome. In both cases, we observed increased median and maximal lifespan after developmental ΔOTC expression, just as in flies reared normally (Figure 2e–f). Larval ΔOTC expression also did not affect pathogen resistance at day 7 of adulthood (Figure 2—figure supplement 3a). We further analyzed the intestinal microbiome load (as assessed by colony-forming units, CFUs) at different points of life after developmental ΔOTC expression, and found no significant difference relative to controls (Figure 2—figure supplement 3b). To test effects on microbiome composition, we performed 16S sequencing in young and old flies after developmental ΔOTC expression, and similarly found no major changes relative to controls (Figure 2—figure supplement 3c). Altogether, these data indicate that activation of the immune system following developmental ΔOTC expression does not contribute to the observed increase in adult longevity.

Developmental mitochondrial proteostatic stress affects the metabolic state of adult flies and leads to persistent FoxO activation in the fat body

To further explore potential physiological mechanisms conferring lasting protective effects of developmental mitochondrial stress, we assessed whether metabolism in adult flies is affected (Figure 3, Figure 3—figure supplement 1). We first tested the ratio of NAD/NADH, as a measure of mitochondrial function, but did not detect changes in adults after developmental ΔOTC expression (Figure 3a). We also did not see changes in MitoTracker intensity in muscle tissue from adults after developmental ΔOTC expression (Figure 3—figure supplement 1a). However, we found that expressing ΔOTC during development leads to reduced triglyceride (TAGs) concentrations (relative to total protein) in adult flies (Figure 3b); this phenotype is consistent with activation of FoxO, which is known to induce lipases and increase fat metabolism (Karpac et al., 2013). PGAM5 null mutants did not show a decrease in TAG levels, supporting the notion that this metabolic shift results from UPRmt signaling.

Figure 3 with 1 supplement see all
Download asset Open asset
Developmental mitochondrial stress changes adult metabolism and leads to lasting activation of FoxO.

(A) Developmental ΔOTC expression does not affect the ratio of NAD/NADH in adult flies. (B) Developmental ΔOTC expression lowers steady-state triglyceride levels in adult flies, consistent with fat body FoxO activation. In PGAM5 null flies this effect is not seen. p-values are 0.011 and 0.567. (C) ΔOTC expression throughout development leads to increased nuclear localization of FoxO in the fat body of third instar larvae. This effect is neutralized in PGAM5 null larvae. p-values are <0.0001 and 0.0001. (D) Adult fat bodies show persistent nuclear FoxO localization after developmental stress. Flies were dissected 7 days after eclosion, with no adult ΔOTC expression. p-values are 0.002 and 0.002. (E) Larval and lasting FoxO translocation by the mitochondrial stress can be blocked by simultaneously administering the JNK inhibitor SP600125 (far left and left). p-values are 0.001 and <0.0001;<0.0001 and 0.0002. Similarly, 24 hr mitochondrial stress in adult flies leads to FoxO translocation, which is blocked by simultaneous expression of dominant negative bsk/JNK (right). p-values are <0.0001 and<0.0001. In contrast to developmental treatment, FoxO activation induced by adult ΔOTC expression does not persist 1 week after treatment is stopped (far right). p-values are <0.0001 and 0.96. All error bars are SEM from two independent experiments. Student’s t-test (B) or ANOVA with dunnett’s multiple comparisons (A, C–E); *p<0.05, **p<0.01, then each *=0.1 x. 200 µM RU486 used in all experiments. .

https://doi.org/10.7554/eLife.26952.012
Figure 3—source data 1

Developmental mitochondrial stress changes adult metabolism and leads to lasting activation of FoxO.

https://doi.org/10.7554/eLife.26952.014

The fat body is a major metabolic organ in Drosophila, and we have previously shown that JNK activation in the larval fat body leads to nuclear translocation and activation of FoxO (Wang et al., 2005). We therefore used immunohistochemistry to explore the dynamics of FoxO localization in the fat body in response to mitochondrial stress. ΔOTC expression during development strongly induced nuclear translocation of FoxO in late L3 larval fat bodies (Figure 3c). Again, this was not evident in PGAM5 null mutant larvae. FoxO is a known regulator of stress responses and organismal lifespan, but in order to directly promote longevity in response to developmental UPRmt, FoxO would have to remain active during adulthood. To test this, we imaged adult fat bodies one week after eclosion and found that flies retained nuclear-localized FoxO after developmental ΔOTC expression (Figure 3d). We next tested whether this lasting effect on FoxO was dependent on JNK activity during development, by adding the JNK inhibitor SP600125 to the larval food. This treatment inhibited both larval and adult translocation of FoxO to the nucleus (Figure 3e, left). To confirm that this was due to specific inhibition of JNK/bsk, we also used genetic tools to block this pathway. Developmental expression of dominant negative bsk (bskDN) was lethal, but expressing bskDN simultaneously with ΔOTC in adult flies limited FoxO nuclear translocation (Figure 3e).

Since developmental activation is crucial for UPRmt-mediated longevity ((Durieux et al., 2011) and Figure 2b), we tested whether the lasting effect on FoxO depends on the timing of ΔOTC expression. We expressed ΔOTC for 24 hr on day 3 of adulthood, which lead to acute but not lasting nuclear translocation of FoxO (Figure 3e, right). The absence of a persistent response to mitochondrial stress after adult treatment adults is consistent with the lack of lifespan extension in these conditions (Figure 2b). Our results suggest that the persistent activation of FoxO in response to the developmental activation of the UPRmt is the mechanism by which developmental but not adult mitochondrial stress can extend lifespan.

Heat-shock proteins are persistently upregulated in adult flies after developmental stress

To test this hypothesis, we then explored the transcriptional consequences of developmentally activated, persistent FoxO activation in adults. We performed RNA sequencing of fat bodies from both larvae acutely expressing ΔOTC, and 7-day-old adults that had experienced larval ΔOTC expression. To distinguish the specific transcriptional response responsible for longevity from more general adaptations to mitochondrial stress, we performed this experiment in both wild type and PGAM5 null mutant larvae and adult flies. We observed a large number of upregulated genes in the acutely stressed larvae (Figure 4a; using a minimum baseline FPKM value of 10 and a cutoff of 3-fold induction relative to controls). These genes were predominantly involved in energy metabolism, and showed considerable overlap between the wild type and PGAM5 null mutant data sets. Only a small number of these (predominantly heat-shock and stress response genes) remained upregulated in adult flies. Notably, none of the genes induced persistently in wild-type flies remained induced in PGAM5 nulls (Figure 4b,c). These genes include twelve chaperones, some of which are FoxO targets (Donovan and Marr, 2016; Wang et al., 2003) and have previously been shown to extend lifespan when upregulated (Liao et al., 2008; Morrow et al., 2004; Tatar et al., 1997; Wang et al., 2003; Zhao et al., 2005). The only non-chaperone-encoding genes in this list (Tsp42E and Obp99) are also regulated by FoxO and are involved in lifespan regulation (Alic et al., 2014; Bai et al., 2013). Additional FoxO targets, such as peroxiredoxin 3 (Chiribau et al., 2008), were also upregulated in adults following developmental ΔOTC expression, but fell below our 3-fold cutoff. Similarly, targets that are negatively regulated by FoxO were found to be downregulated in response to developmental mitochondrial stress (Supplementary File 1).

Figure 4
Download asset Open asset
Lasting transcriptional activation of FoxO targets after developmental stress.

(A) Venn diagram shows genes upregulated ≥3 x in fat bodies from third instar larvae undergoing mitochondrial proteostatic stress relative to untreated controls, where control FKMP ≥10. We separate genes whose upregulation depends on the identified stress signaling pathway (red) from those induced by other means (green), by comparing the transcriptional profile of PGAM5 null mutant larvae. Bar graphs show overrepresented Gene Ontology terms within each group of genes, showing mainly metabolic changes. Redundant GO terms were trimmed using REVIGO. (B) Venn diagram shows genes whose expression levels are persistently upregulated ≥3 x after developmental stress. All these genes were dependent on functional UPRmt signaling (i.e. not induced in PGAM5 nulls). Most of these genes correspond to heat-shock proteins. (C) Expression of the persistently upregulated genes is shown for all conditions, with FoxO target genes highlighted in red. Several of these have been shown to extend lifespan when overexpressed (see text). Full raw data found in Suppl. File 1. 200 µM RU486 was used.

https://doi.org/10.7554/eLife.26952.015

Persistent FoxO activation and lifespan are diet-dependent and regulated by mTORC1 and GCN-2 activity

To identify the biological mechanism promoting persistent FoxO activation, we first analyzed potential post-translational modifications of FoxO. It was previously reported that nicotinamide adenine dinucleotide (NAD+) precursors activate the UPRmt in worms, via the deacetylase sir-2.1 (Mouchiroud et al., 2013). Knocking down Sirt2 (the fly homologue of sir-2.1 and SIRT1) during developmental ΔOTC expression, however, did not affect FoxO nuclear localization, and neither Sir2 overexpression nor supplementation with the NAD precursor nicotinamide riboside were sufficient to induce nuclear translocation of FoxO (Figure 5—figure supplement 1b–d). Drosophila FoxO can also be regulated by HDAC4 (Wang et al., 2011), but knocking down this deacetylase during developmental ΔOTC expressionalso did not affect FoxO localization (Figure 5—figure supplement 1e).

We next asked whether lasting FoxO activation was the result of nutrient response signaling. FoxO activity is regulated by insulin signaling, in flies mediated by the Drosophila Insulin-like Peptide (dilp) family (Kannan and Fridell, 2013). During starvation, reduced availability of dILPs promotes FoxO nuclear translocation in the fat body, by reducing insulin receptor and protein kinase B activities. Upon refeeding, this signaling network is reset, promoting FoxO translocation to the cytoplasm. To test whether reduced insulin signaling activity was carried over from developmental stress into adulthood, we performed starvation and refeeding experiments on these flies. As shown in Figure 5a, FoxO translocates to the nucleus after developmental ΔOTC expression. After 9 hours of starvation (only water was offered to flies), the fat body of both stressed and control flies exhibit nuclear FoxO. 24 hr of refeeding resets this activation in control flies, but FoxO remained nuclear in flies that had experienced developmental stress. We were also unable to detect changes in transcription of dilp2, −3,–6 and −8 during/after ΔOTC expression. Further supporting the idea that the mitochondrial proteostatic stress affects FoxO independently of insulin signaling, switching flies to a diet of pure sucrose (5% in water) for 5 days did not abolish the persistent nuclear localization of FoxO caused by developmental ΔOTC expression (Figure 5b).

Figure 5 with 2 supplements see all
Download asset Open asset
ΔOTC-mediated FoxO activity is regulated by diet, through mTORC1 and GCN-2.

(A) Adult flies show nuclear localization of FoxO after developmental ΔOTC expression. After 9 hr of wet starvation, FoxO is nuclear in treated and untreated flies alike. 24 hr of refeeding returns FoxO to the cytoplasm in untreated flies, but does not reset the effect of developmental stress. p-values are (pre to starved) 0.517 and 0.031, (starved to refed) 0.723 and 0.001. (B) FoxO nuclear localization following developmental stress is reset by 5 days on a high-protein diet, but not a pure sucrose diet. Feeding rapamycin at 200 µM to inhibit mTORC1 with the high-protein diet blocks this effect. p-values are (±RU) 0.007, (+RU vs diets) 0.790, 0.915,<0.0001 and 0.338, (HPD ± rapa) 0.001. (C) Knocking down TSC1 during developmental ΔOTC expression blocks persistent activation of FoxO in adults (left). Conversely, overexpressing the mTORC1 inhibitors TSC1 and two in adult flies is sufficient to change FoxO localization absent mitochondrial stress (right). p-values are <0.0001 and 0.0027. (D) Knocking down GCN-2 during developmental ΔOTC expression is sufficient to block persistent FoxO activation (left). p-values are 0.0005 and 0.0006. Overexpressing the GCN-2 kinase in larvae produces FoxO nuclear localization similar to the effect of mitochondrial stress (right). p-values are (ΔOTC vs UAS-GCN2) 0.703 and (UAS-GCN2 ±RU)<0.0001. (E) Inhibiting mTOR with 200 µM rapamycin is sufficient to induce nuclear FoxO localization. This effect is blocked by RNAi of GCN-2, suggesting a downstream role for this kinase. p-values are 0.032 and 0.602. (F) Developmental ΔOTC expression does not extend lifespan when flies are fed a high-protein diet throughout adulthood, consistent with a requirement for persistent FoxO activity. All error bars are SEM from 2 + independent experiments. ANOVA with Tukey’s (A-B) or dunnett’s (D) post-hoc, or student’s t-test (C and E); *p<0.05, **p<0.01, then each *=0.1 x. 200 µM RU486 used in all experiments.

https://doi.org/10.7554/eLife.26952.016
Figure 5—source data 1

ΔOTC-mediated FoxO activity is regulated by diet, through mTORC1 and GCN-2.

https://doi.org/10.7554/eLife.26952.019

However, flies switched to an alternate diet containing ~5 x higher protein levels than control food showed a nearly complete loss of nuclear FoxO in the fat body after 5 days (Figure 5b). To test whether this high-protein diet affects FoxO localization through the main amino-acid-sensing pathway mTORC1, we repeated this experiment with supplementation of the mTORC1 inhibitor rapamycin at 200 µM. This prevented the loss of nuclear FoxO (Figure 5b), suggesting that mTORC1 activation can override the persistent nuclear translocation of FoxO produced by developmental ΔOTC expression. In support of this role for mTORC1, activating mTORC1 during developmental stress (by RNAi of its inhibitor TSC1) prevents lasting FoxO activation (Figure 5c, left), while reducing developmental survival in the same manner as FoxO deficiency (Figure 5—figure supplement 2). Furthermore, simultaneous over-expression of TSC1 and 2 to inhibit mTORC1 activity in adult flies is sufficient to promote nuclear translocation of FoxO (Figure 5c, right). mTORC1 activity has previously been reported to inhibit the FoxA ortholog Fork Head by preventing its nuclear localization (Bülow et al., 2010). This involves signaling through the protein kinase GCN-2, which has previously been implicated in the UPRmt in C. elegans (Baker et al., 2012). Indeed, overexpression of GCN-2 was sufficient to induce nuclear localization of FoxO, while knocking down GCN-2 during developmental ΔOTC expression abrogates the lasting effect on FoxO (Figure 5d). To test whether GCN-2 acts downstream of mTORC1 in this context, we knocked down GCN-2 in flies treated with rapamycin and observed a reduction in nuclear FoxO (Figure 5e).

Because FoxO activity is required for lifespan extension (Figure 2d) and a high-protein diet erases the persistent FoxO activation seen after developmental ΔOTC expression (Figure 5b), we hypothesized that adult diet would affect the longevity effects of mitochondrial proteostatic stres. To test this hypothesis, we repeated our lifespan experiment with developmental ΔOTC expression but switched flies to the high-protein diet after eclosion. The high-protein diet reduced the lifespan of the control group relative to our standard fly food, consistent with previous literature, but also completely abolished the lifespan extension normally induced by ΔOTC expression (Figure 5f). This supports our proposed pathway for lifespan extension, and suggests diet as an important regulator of beneficial UPRmt effects.

Discussion

Our data identify a signaling pathway that responds to mitochondrial proteostatic stress through the phosphatase PGAM5, leading to activation of JNK and FoxO. This results in the FoxO- and Rel-mediated induction of immune, antioxidant and metabolic gene expression, as well as of genes encoding heat shock proteins. When activated during development, FoxO remains active throughout life, inducing a select group of genes that extend lifespan. Interestingly, this activity is subject to regulation by GCN-2/mTORC1-dependent nutrient sensing, providing a clue for how diet may interfere with lifespan extending stress signaling mechanisms (Figure 6).

Figure 6
Download asset Open asset
Proposed model for UPRmt-mediated longevity.
https://doi.org/10.7554/eLife.26952.020

Our genetic studies suggest a pathway of UPRmt activation from the mitochondrial membrane protein PGAM5, through ASK1 and JNK, to the FoxO transcription factor. FoxO increases the expression of Relish, and thereby induces antimicrobial peptide expression. Persistent activation of FoxO also leads to lasting upregulation of chaperones, which improves proteostasis and extends lifespan. Meanwhile, sensing of amino acids can activate mTORC1 and GCN-2, which negate the persistent activation of FoxO and block lifespan extension.

PGAM5 has not been fully characterized but plays a role in regulating cell death pathways in flies and cultured cells (Ishida et al., 2012; Wang et al., 2012; Zhuang et al., 2013). It has been reported to localize to either the inner (Lo and Hannink, 2008) or outer mitochondrial membrane, where it is cleaved upon loss of membrane potential (Sekine et al., 2012). While a member of the PGAM family based on sequence, it displays phosphatase rather than mutase activity in culture (Takeda et al., 2009). PGAM5 also interacts with the mitophagy factor PINK1, and loss of PGAM5 rescues the muscle degeneration phenotype of dPINK1 mutant flies (Imai et al., 2010). Loss of PGAM5 inhibits mitophagy in vitro, and leads to Parkinson’s-like symptoms in mice (Lu et al., 2014). Regulation of mitophagy by PGAM5 may depend on its interaction with the PARL protease, which is normally responsible for cleaving PINK1 to prevent mitophagy of healthy mitochondria (Sekine et al., 2012). Indeed, induction of the UPRmt in HeLa cells has been reported to trigger PINK1 accumulation, Parkin recruitment and mitophagy (Jin and Youle, 2013). In our experiments, PGAM5 does not seem to regulate longevity directly by stimulating apoptosis, however, suggesting a more complicated biological role for this phosphatase. It is tempting to speculate that it may act as a rheostat to dictate whether mitochondrial stress results in UPRmt activation, mitophagy, or cell death. Further exploration of the intra-mitochondrial signals that activate PGAM5, as well as its immediate downstream partners, should be interesting topics for future studies of mitochondrial stress.

The importance of developmental activation of the UPRmt for lifespan extension was previously observed in studies exploring this mitochondrial stress response and the response to ETC dysfunction in C. elegans and Drosophila (Copeland et al., 2009; Dillin et al., 2002; Durieux et al., 2011; Rea et al., 2007). Our findings confirm this and identify persistent FoxO activation and FoxO-induced gene expression as associated with the lasting benefits of developmental mitochondrial stress in the adult organism. Two recent papers have identified a role for the UPRmt in regulating histone methylation and chromatin remodeling (Merkwirth et al., 2016; Tian et al., 2016). The studies show that demethylase activity during developmental UPRmt is required for lifespan extension, and that atfs-1 acts synergistically with these epigenetic changes. It will be interesting to explore the interplay between elevated FoxO activity and demethylase-regulated chromatin accessibility in future studies.

The hypothesis that a combination of signaling and epigenetic changes are required for UPRmt longevity could help explain recent reports that commonly used UPRmt reporters in C. elegans correlate poorly with longevity (Bennett et al., 2014; Ren et al., 2015): the UPRmt likely involves several pathways, which can also be activated by other stimuli. However, a crucial combination of signaling and epigenetic changes could be induced by specific types of mitochondrial stress. This would prompt caution about using a single gene reporter as a measure of ‘UPRmt activation’ and suggest using multiple assays to determine its role in each biological process. To our knowledge, this is the first report of an organism other than C. elegans showing improved longevity from UPRmt activation that does not involve direct ETC disruption. This lends support to the UPRmt as a conserved longevity mechanism, which overlaps with but is not fully explained by ETC function.

Our work also provides further support for the observation that the UPRmt triggers the innate immune system (Pellegrino et al., 2014) and shows that this feature is conserved outside of C. elegans. As in worms, this activation provides acute resistance to infection in flies, but we find that immune function is not responsible for the effects of the UPRmt on longevity. The activation of antimicrobial gene expression in response to FoxO activation is not surprising, as it has previously been reported that the FoxO transcription factor can induce the expression of antimicrobial peptides independently of classical innate immune pathways in response to starvation (Becker et al., 2010). Moreover, we have shown previously that FoxO activation in the larval Drosophila fat body can induce Relish/NF-κB signaling by stimulating Rel expression (Karpac et al., 2011). Our data indicate that mitochondrial proteostatic stress activates this FoxO/Relish cassette, but that the FoxO-dependent upregulation of chaperones in adult flies is likely responsible for lifespan extension.

Our results are also consistent with reports that the UPRmt can be activated by rapamycin (Houtkooper et al., 2013) and identify high-protein diet as a critical intervention that activates mTORC1 and overrides the positive effects of developmental mitochondrial stress. It was also previously reported that NAD precursors and sirt-2.1 activate the UPRmt and extend lifespan in a daf-16/FOXO-dependent manner (Mouchiroud et al., 2013). Our experiments did not suggest a role for NAD precursors or Sirt2 in regulating the Drosophila UPRmt, but it remains possible that exploring different sirtuins or dietary conditions could reveal such a role. This is especially true in light of the observed effect of diet on UPRmt activity and lifespan extension. The regulation of FoxO by amino acid sensing/mTORC1 rather than insulin signaling is surprising, but is supported by a previous observation that mTORC1 activity regulates longevity in C. elegans through activation of daf-16/FOXO (Robida-Stubbs et al., 2012).

These findings thus reveal a critical vulnerability of developmental UPRmt-mediated physiological changes that promote longevity: they can be erased by a high-protein diet. If the observed mTORC1-induced signaling interactions are conserved in vertebrates, this has important implications for the development of interventions that aim to engage the UPRmt to increase metabolic health and extend health- or lifespan.

Materials and methods

Drosophila stocks

Request a detailed protocol

The following lines were obtained from the Bloomington Drosophila Stock Center: w1118, y1 w1, Da::GS, tub::GS, PGAM5 RNAi (#34744), ASK1 RNAi (#32646), foxo21 (Jünger et al., 2003), foxoΔ94 (Slack et al., 2011), bskDN (#6409).

The following lines were obtained from the Vienna Drosophila RNAi Center: Hrd1 RNAi (#6870), PGAM5 RNAi (#51657), ASK1 RNAi (#110228), TSC1 RNAi (#110811), bsk RNAi (#34138), foxo RNAi (#30556).

The following lines were gifts from other labs: UAS-ΔOTC (Martins lab, Pimenta de Castro et al., 2012), PGAM51 (Ichijo lab, Imai et al., 2010), jun and fos RNAi (Bohmann lab, Hyun et al., 2006)), UAS-GCN-2 and GCN-2 RNAi (Leopold lab, Bjordal et al., 2014), UAS-TSC1 and TSC2 (Tatar lab, Hwangbo et al., 2004), ND75 and Rel RNAi (Perrimon lab, Agaisse et al., 2003; Owusu-Ansah et al., 2013)

Fly husbandry and demographics

Request a detailed protocol

Standard fly food was prepared with the following recipe: 1 l distilled water, 13 g agar, 22 g molasses, 65 g malt extract, 18 g brewer’s yeast, 80 g corn flour, 10 g soy flour, 6.2 ml propionic acid, 2 g methyl-p-benzoate in 7.3 ml of EtOH. High-protein food was prepared with the following recipe: 1 l distilled water, 10 g agar, 80 g brewer’s yeast, 20 g yeast extract, 20 g peptone, 51 g sucrose, 6.2 ml propionic acid, 2 g methyl-p-benzoate in 7.3 ml of EtOH (Musselman et al., 2011). 100% sucrose diet is 50 g/l sucrose in water. For GeneSwitch experiments, 86 mg RU486 was additionally dissolved in the EtOH for 200 µM final concentration. For antibiotic experiments, 50 mg of the following antibiotics were added during food preparation: ampicillin, tetracycline, erythromycin, kanamycin. For axenic experiments, fly food bottles were autoclaved at 121°C for 30 min, and 1 ml EtOH ±5 mg RU486 added after cooling, previously determined to yield a final concentration of 200 µM (Biteau et al., 2010). Eggs laid over a 24 hr period were collected and sterilized for 3 min in 2.7% sodium hypochlorite, then washed twice with sterile, distilled water for 1 min and transferred to sterile bottles in a laminar flow hood. Eclosed flies were kept on sterilized food in the flow hood. Flies were maintained at 25°C and 65% humidity, on a 12 hr light/dark cycle, unless otherwise indicated. For lifespan experiments, up to 30 flies per vial were flipped thrice weekly, with dead flies counted visually.

Immunostaining and microscopy

Request a detailed protocol

Fat bodies were dissected from adult females or from third instar larvae pre-wandering. Tissue was fixed at room temperature for 45 min in 100 mM glutamic acid, 25 mM KCl, 20 mM MgSO4, 4 mM Sodium Phosphate, 1 mM MgCl2, 4% formaldehyde. Washes were done in PBS with 0.5% BSA, 0.1% Triton X-100 at 4°C. Primary incubation was done overnight 4°C using a rabbit N-terminal FoxO antibody (RRID:AB_2569227, gifted from O. Puig) at 1:500. Secondary antibodies were from Jackson Immunoresearch. TUNEL staining was performed using In Situ Cell Death Detection Kit (Roche). DNA was stained using Hoechst, and tissue mounted on slides using Mowiol mounting medium (Sigma). Imaging was done on a Zeiss LSM700 confocal microscope at 40x magnification.

NAD/metabolism assays

Request a detailed protocol

All assays were done on samples of five whole flies/larvae. NAD and NADH levels were quantified from whole flies using the NAD/NADH Quantitation Colorimetric kit from BioVision. Glucose and glycogen were quantified using the Glucose (HK) Assay kit from Sigma. Lipids were quantified with the Triglyceride LiquiColor kit from Stanbio. Protein levels were quantified with the BCA assay.

Infection

Request a detailed protocol

P. entomophila and P. aeruginosa were cultured in LB medium for 16 hr at 37°C and 10 hr at 30°C, respectively. For oral infections, flies were starved for 2 hr before treatment. 6 mL culture was pelleted at 5000 g for 10 min, then resuspended in 500 uL 5% sucrose and added to a vial containing Whatman filter paper. 20 flies per vial were maintained at 29°C, and deaths tracked daily or twice daily. 100 uL 5% sucrose was added to the vials each day. For humeral infections, a tungsten needle was dipped in 30x concentrated culture and flies were pricked in the periphery of the abdomen. They were then transferred to normal food vials. Negative controls were fed/pricked with 5% sucrose alone.

RNA extraction and quantitative RT-PCR analysis

Request a detailed protocol

Muscle tissue was isolated by removal of head, abdomen, wings, legs and intestine from fly thoraces. Fat bodies were manually isolated from fixed abdomens/larvae. 5–8 flies were used per sample, and total RNA was extracted from using Trizol (Life Sciences) according to manufacturer’s protocol. cDNA was synthesized using Superscript III (Life Sciences). Results represent four biological samples, each with triplicate technical repeats. Real-time PCR was performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) on a Bio-Rad Real-Time CFX96 system. Expression levels were calculated as ΔΔCt normalized to RP49.

RNA sequencing

Request a detailed protocol

Muscle or fat body tissue was isolated and RNA extracted as described above, using 15 flies per sample. TruSeq RNA Library Prep Kit (Illumina) was used to prepare libraries, and sequencing was performed on an Illumina MiSeq system. Raw data were analyzed with the Tuxedo suite (RRID:SCR_013194) and reads were mapped to Drosophila genome release 5.2. Expression was recorded as FPKM: fragments per kilo-base per million reads.

Statistics and bioinformatics

Request a detailed protocol

Significance in two-condition experiments was evaluated by student’s t-test. Multiple condition experiments were evaluated by one-way ANOVA, with Dunnett’s post-hoc comparing to induced sample or Tukey’s post-hoc when multiple conditions are compared. Gene Ontology analysis of RNAseq experiments was done at flymine.org, using REVIGO (RRID:SCR_005825) to trim redundant terms (Allowed similarity: Medium).

CAFE assay

Request a detailed protocol

Feeding rates were quantified by the CAFE assay (Ja et al., 2007), measuring consumption of a 5% w/v sucrose solution accessible from a capillary tube. Each vial contained 10 flies per vial, and evaporation was measured with an empty vial. Consumption was measured at 8–12 hr intervals and normalized to uL/fly/hr.

16S sequencing

Request a detailed protocol

Flies were washed in 70% EtOH for 60 s to kill external bacteria, and intestines dissected out in sterile PBS. Ten guts were collected per sample. Commensal genomic DNA was extracted using UltraClean Microbial DNA Isolation Kit (MO BIO). This DNA was used as template for limited cycle PCR with primers targeting V3/V4 regions (Forward 5’-CCTACGGGNGGCWGCAG-3’ and Reverse 5’-GACTACHVGGGTATCTAATCC −3’) to generate a 16S metagenomic sequencing library. The following reaction conditions were used: 94°C for 5 min, followed by 30 cycles of 94°C for 1 min, 48°C for 2 min, and 72°C for 2 min, and a final extension at 72°C for 5 min. Illumina Miseq paired-end (2 × 300 bp) sequencing was performed and Miseq Reporter Software was used for primary analysis and classification based on Greengenes database (RRID:SCR_002830).

Commensal quantification

Request a detailed protocol

Intestines were dissected out as described above, pooled in groups of 10 and crushed with a motorized pestle. These samples were diluted 1000x and plated on non-selective LB Amp plates. Sterile PBS was used as a negative control. Colonies were counted after 60 hr at 30°C.

Primers used

Request a detailed protocol

RP49 F: 5’-TCCTACCAGCTTCAAGATGAC-3’

RP49 R: 5’-CAGGTTGTGCACCAGGAACT-3’

Metchnikowin F: 5’-AATCAATTCCCGCCACCGAG-3’

Metchnikowin R: 5’-GACCCGGTCTTGGTTGGTTA-3’

Drosomycin F: 5’-CGTGAGAACCTTTTCCAATATGAT-3’

Drosomycin R: 5’-TCCCAGGACCACCAGCAT-3’

Turandot A F: 5’-GCACCCAGGAACTACTTGACATCT-3’

Turandot A R: 5’-GACCTCCCTGAATCGGAACTC-3’

Relish F: 5’-ACAGCCCACATTCCCATCAG-3’

Relish R: 5’-GAGCCGCACCTGGTTCAA-3’

Hsp60 F: 5’-GACCAGATCGAGGACACCAC-3’

Hsp60 R: 5’- GCCGAGTTTCTGATCCTCGTTG-3’

Hsc70-5 F: 5’-CTGCGTTACAAGTCCGGTGA −3’

Hsc70-5 R: 5’- GCAGCACATTAAGACCAGCG-3’

Hsp10 F: 5’-CCCGCATCTAGCGAGAATAG −3’

Hsp10 R: 5’-CTCCTTTCGTCTTGGTCAGC −3’

ClpX F: 5’-AAAATGCTCGAAGGCACAGT −3’

ClpX R: 5’-TTGAGACGACGTGCGATAAG −3’

References

Decision letter

  1. Utpal Banerjee
    Reviewing Editor; University of California, Los Angeles, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "PGAM5 promotes lasting FoxO activation after developmental mitochondrial stress and extends lifespan in Drosophila" for consideration by eLife. Your article has been favorably evaluated by Jonathan Cooper (Senior Editor) and three reviewers, one of whom, Utpal Banerjee, is a member of our Board of Reviewing Editors. The following individual involved in review of your submission has agreed to reveal their identity: Edward Owusu-Ansah (Reviewer #2).

The reviewers have discussed the reviews with one another and appreciate the importance of the findings and the overall quality of the data. The summary of these reviews and discussions are summarised below by the Reviewing Editor.

A set of specific issues raised by the reviewers that we believe can be addressed by new experiments and/or rewriting the manuscript are summarized below as "other concerns". However, there is one major issue that needs special attention. During the post-review discussions, the reviewers unanimously agree that the data lack direct evidence for a sustained FOXO activation (which, as the title suggests, is central to the conclusions derived from this work).

Summary:

In this paper Jensen et al. investigate how mitochondrial dysfunction during Drosophila development leads to a stress response that has lasting effects on the flies, most notably on extension of lifespan. The authors characterize the steps that signal mitochondrial dysfunction to modulate nuclear gene expression, using an immune response gene as an output. In adult flies that were subjected to mitochondrial stress during development, a subset of stress response genes are expressed (even though the original stimulus is removed); it is proposed that these genes (all FoxO targets) mediate lifespan extension. The lasting effects of developmental mitochondrial dysfunction can be reversed through exposure to a high protein diet during adulthood.

Essential revisions:

(major concerns):

The key weakness regards the study's (implied) inference on the role of lasting FoxO activation as the longevity mechanism. Epistasis analysis is needed to test this idea.

The important required study is to conduct mtUPR (in any way) in FoxO-null epistasis. Alternatively, the authors could test the effect of co-expression of the OTC construct and RNAi to FoxO, using the gene-switch system. This approach allows the genotype without RU to serve as its own co-isogenic control. One needs to measure the longevity of the OTC expression alone, using geneswitch again, to confirm there is a longevity benefit in that period. And, one should do the same by driving the FoxO RNAi with the same geneswitch (which should do very little to the survival).

Essential revisions:

(other concerns):

1) Are phenotypes due to a UPRmt stress response per se? How much is due to a more general response caused by mitochondrial dysfunction. This is unclear because of the nature of mitochondrial stressors. ND75 knockdown will cause mitochondrial dysfunction (possibly affecting ATP, Ca2+, ROS, NADH, NAD+ etc.). The ΔOTC mutant in flies also severely impacts mitochondrial function (ATP, oxygen consumption etc.). Therefore, it is not clear that a UPRmt or some other type of mitochondrial stress response is being studied. Presumably studies in worms, where chaperone up-regulation has clear linkage to UPRmt provides the inspiration, but this is not yet in flies. For example, ΔOTC causes ETC dysfunction in flies (Pimenta de Castro et al. 2012). If this is not true for this system, then some functional data are required, or the description of the model should be modified.

2) Is the PGAM5/ASK/JNK signaling pathway a bona fide UPRmt pathway (analogous to the ATFS-1)? If a chaperone important in the UPRmt response affecting proteostasis directly was chosen as a marker rather than a gene involved in the immune response, then the relation of this new signaling pathway to UPRmt could be better assessed. Either a more specific UPRmt output marker should be used, or discussion of UPRmt should be reserved for the Discussion.

3) In the same vein, it is unclear whether the canonical markers of the UPRmt, such as Hsp60, Hsp10, mortalin (mt-Hsp70) and the ClpP protease are induced under the conditions used to activate the UPRmt in this study. The authors should comment on this, and provide a possible explanation for this apparent disparity.

4) The exact dosages of RU456 used in the gene-switch experiments should be listed in the figure legends.

5) The link between UPRmt and longevity is not yet settled, and while this work provides important advancements, the authors will do well to tone down the enthusiasm with which they link expression of the OTC construct with UPRmt induction. OTC induces other genes as well (such as antimicrobial peptides). Therefore several bold statements such as the subheading "UPRmt-mediated longevity is not caused by improved adult immune function, or by changes to the microbiome" and "Developmental UPRmt affects the metabolic state of adult flies and leads to persistent FoxO" or (based on point 1 above), "Of the two methods initially used to induce the UPRmt, we opted to use ΔOTC for further experiments to avoid potential secondary effects of inducing ETC dysfunction" should be modified to reflect the fact that the results observed are due to OTC expression and not necessarily all due to the UPRmt.

6) ADaGSxOTC itself yields less than 50% eclosion. Survivors of this cohort may live longer because of frailty selection: weak larvae that would produce shorter-lived adults do not eclose. Some rescue-type data (Figure 2C) argues against this potential confound, but the evidence is thin.

7) UPRmt induced in adults activates FoxO, but such cohorts are not long-lived when RU is constantly applied. That FoxO is only transiently activated when RU is given for just one week is not a satisfactory explanation.

8) The data with HPD is unclear. Presumably, HPD resets both this TF and the lifespan, and this is used to infer causality between the induced longevity and FoxO activation. But in the sole survival experiment to this point (Figure 5F), the shape of the plots is a concern: they are too linear, suggesting that age-independent mortality is the overriding cause of death. This can mask any potential impact on age-dependent mortality; and there are no data to rule this out (or in). Perhaps the HPD is toxic, and all flies die for reasons besides aging. There are no other data to address the relevance of activated FoxO (by any of the interesting, observed mechanisms) as relevant to the larval UPRmt impact on adult longevity.

https://doi.org/10.7554/eLife.26952.024

Author response

Essential revisions:

(major concerns):

The key weakness regards the study's (implied) inference on the role of lasting FoxO activation as the longevity mechanism. Epistasis analysis is needed to test this idea.

The important required study is to conduct mtUPR (in any way) in FoxO-null epistasis. Alternatively, the authors could test the effect of co-expression of the OTC construct and RNAi to FoxO, using the gene-switch system. This approach allows the genotype without RU to serve as its own co-isogenic control. One needs to measure the longevity of the OTC expression alone, using geneswitch again, to confirm there is a longevity benefit in that period. And, one should do the same by driving the FoxO RNAi with the same geneswitch (which should do very little to the survival).

We agree that this is an important experiment. Per the reviewer’s suggestions, we have now carried out duplicate lifespan experiments where DaGS was used to drive ΔOTC expression in a FoxO null background (the foxoΔ94/21 transheterozygotes already described in the manuscript). In parallel, we replicated the DaGS x ΔOTC lifespan experiment already included in the paper. In support of our model, developmental mitochondrial stress leads to a significant shortening of lifespan in FoxO null flies, in sharp contrast to the lifespan extension observed in normal flies. The FoxO null data is presented in the revised Figure 2D, alongside statistics for the positive control in 2G. The graph for the positive control is shown in Figure 2—figure supplement 2. Combined with our observations of drastically reduced survival through developmental mitochondrial stress in FoxO nulls (now in Figure 2—figure supplement 1), this solidifies FoxO activity as a necessary factor for the lifespan extension elicited by developmental mitochondrial stress.

Essential revisions:

(other concerns):

1) Are phenotypes due to a UPRmt stress response per se? How much is due to a more general response caused by mitochondrial dysfunction. This is unclear because of the nature of mitochondrial stressors. ND75 knockdown will cause mitochondrial dysfunction (possibly affecting ATP, Ca2+, ROS, NADH, NAD+ etc.). The ΔOTC mutant in flies also severely impacts mitochondrial function (ATP, oxygen consumption etc.). Therefore, it is not clear that a UPRmt or some other type of mitochondrial stress response is being studied. Presumably studies in worms, where chaperone up-regulation has clear linkage to UPRmt provides the inspiration, but this is not yet in flies. For example, ΔOTC causes ETC dysfunction in flies (Pimenta de Castro et al. 2012). If this is not true for this system, then some functional data are required, or the description of the model should be modified.

We agree that distinguishing general mitochondrial stress from a more specific UPRmt is difficult in any organism. In our original manuscript, we decided to follow the nomenclature of previous studies (Pimenta de Castro et al. 2012, Owusu-Ansah et al. 2013). However, to our knowledge (and as suggested by the reviewers), all methods of inducing the UPRmt in any organism leads to mitochondrial dysfunction of some sort. It does not yet seem clear whether and which such dysfunctions should be classified as causative or part of the UPRmt. For example, some parts of the UPRmt have been characterized as ROS-dependent. We originally chose to define the UPRmtas any response occurring in two distinct types of mitochondrial proteostatic stress, but we agree with the reviewers’ point that the Drosophila UPRmt remains imperfectly described. We have thus rewritten several parts of the manuscript to refer to ‘mitochondrial stress’ rather than UPRmt.

2) Is the PGAM5/ASK/JNK signaling pathway a bona fide UPRmt pathway (analogous to the ATFS-1)? If a chaperone important in the UPRmt response affecting proteostasis directly was chosen as a marker rather than a gene involved in the immune response, then the relation of this new signaling pathway to UPRmt could be better assessed. Either a more specific UPRmt output marker should be used, or discussion of UPRmt should be reserved for the Discussion.

3) In the same vein, it is unclear whether the canonical markers of the UPRmt, such as Hsp60, Hsp10, mortalin (mt-Hsp70) and the ClpP protease are induced under the conditions used to activate the UPRmt in this study. The authors should comment on this, and provide a possible explanation for this apparent disparity.

Although the fluorescent reporters available for hsp-6 and hsp-60 in C. elegans respond strongly to induction of the UPRmt, Drosophila studies (Pimenta de Castro et al. 2012, Owusu-Ansah et al. 2013) have shown only moderate (~1.5-fold) induction of these genes. For this reason, the markers mentioned by reviewers were not included in the set of 3-fold upregulated genes in Figure 1A, and we chose metchnikowin as our marker for subsequent screens due to its higher dynamic range of induction.

To address the pertinent comments by the reviewers, we have added a supplement to Figure 1 (Figure 1—figure supplement 1) showing that expression of ΔOTC using the regime reported in Owusu-Ansah et al. leads to a similar upregulation of hsp60, hsc70-5 and hsp10. We further show that this induction is blocked in the absence of PGAM5, bsk or FoxO.

4) The exact dosages of RU456 used in the gene-switch experiments should be listed in the figure legends.

200 μm RU486 was used throughout, and this has now been added to figure legends.

5) The link between UPRmt and longevity is not yet settled, and while this work provides important advancements, the authors will do well to tone down the enthusiasm with which they link expression of the OTC construct with UPRmt induction. OTC induces other genes as well (such as antimicrobial peptides). Therefore several bold statements such as the subheading "UPRmt-mediated longevity is not caused by improved adult immune function, or by changes to the microbiome" and "Developmental UPRmt affects the metabolic state of adult flies and leads to persistent FoxO" or (based on point 1 above), "Of the two methods initially used to induce the UPRmt, we opted to use ΔOTC for further experiments to avoid potential secondary effects of inducing ETC dysfunction" should be modified to reflect the fact that the results observed are due to OTC expression and not necessarily all due to the UPRmt.

As noted in our response to comment #1, we have rewritten parts of the manuscript to refer to mitochondrial stress rather than UPRmt. In response to the comment that “OTC induces other genes as well (such as antimicrobial peptides).”, we note that Pellegrino et al. 2014 showed that AMPs are induced by ATFS-1 activation even absent mitochondrial stress, and were thus described as part of the core UPRmt in C. elegans.

6) ADaGSxOTC itself yields less than 50% eclosion. Survivors of this cohort may live longer because of frailty selection: weak larvae that would produce shorter-lived adults do not eclose. Some rescue-type data (Figure 2C) argues against this potential confound, but the evidence is thin.

This is an important concern, and indeed activation of the UPRmt by constitutively active ATFS-1 in C. elegans also leads to reduced survival through development (Cole Haynes, personal communication). However, our new FoxO null lifespan data shows that FoxO activity not only correlates with, but is required for lifespan extension. It is pertinent to this specific concern that while FoxO null mutants show strongly reduced survival through developmental ΔOTC expression (Figure 2—figure supplement 1), even the survivors showed reduced adult lifespan (Figure 2D). The same is true for PGAM5 nulls (Figure 2C and Figure 2—figure supplement 1).

7) UPRmt induced in adults activates FoxO, but such cohorts are not long-lived when RU is constantly applied. That FoxO is only transiently activated when RU is given for just one week is not a satisfactory explanation.

We apologize for not making this clear in the original manuscript, as indeed the transience of FoxO activation by mitochondrial stress does not explain why continuous mitochondrial stress in adults does not extend lifespan. Our interpretation is that the negative effects of chronically impairing mitochondrial function through expression of ΔOTC has detrimental effects that overshadow the improved proteostasis conferred by activating the FoxO pathway. This is consistent with observations by Pimenta-Castro et al. 2012, who show impaired mitochondrial function and survival with chronic adult ΔOTC expression. In the case of developmentally expressed ΔOTC, on the other hand, the mitochondrial stress is eliminated in adults while FoxO activation is retained chronically.

8) The data with HPD is unclear. Presumably, HPD resets both this TF and the lifespan, and this is used to infer causality between the induced longevity and FoxO activation. But in the sole survival experiment to this point (Figure 5F), the shape of the plots is a concern: they are too linear, suggesting that age-independent mortality is the overriding cause of death. This can mask any potential impact on age-dependent mortality; and there are no data to rule this out (or in). Perhaps the HPD is toxic, and all flies die for reasons besides aging. There are no other data to address the relevance of activated FoxO (by any of the interesting, observed mechanisms) as relevant to the larval UPRmt impact on adult longevity.

As mentioned above, we have now obtained additional evidence that FoxO activity is required for lifespan extension by developmental ΔOTC expression (new lifespan data for null mutants). We have edited the manuscript to merely note that the lack of extension on HPD is consistent with our demonstration that HPD blocks FoxO activation and with the fact that FoxO is required for lifespan extension by developmental mitochondrial stress.

https://doi.org/10.7554/eLife.26952.025

Article and author information

Author details

  1. Martin Borch Jensen

    Buck Institute for Research on Aging, Novato, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8875-0345

  2. Yanyan Qi

    Buck Institute for Research on Aging, Novato, United States
    Contribution
    Investigation, RNA and 16S sequencing experiments
    Competing interests
    No competing interests declared

  3. Rebeccah Riley

    Buck Institute for Research on Aging, Novato, United States
    Contribution
    Investigation, RNA sequencing experiments
    Competing interests
    No competing interests declared

  4. Liya Rabkina

    Buck Institute for Research on Aging, Novato, United States
    Contribution
    Investigation, Metabolite experiments
    Competing interests
    No competing interests declared

  5. Heinrich Jasper

    1. Buck Institute for Research on Aging, Novato, United States
    2. Immunology Discovery, Genentech, South San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Methodology, Project administration, Writing—review and editing
    For correspondence
    hjasper@buckinstitute.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6014-4343

Funding

National Institute on Aging (R01 AG028127)

  • Yanyan Qi
  • Rebeccah Riley
  • Heinrich Jasper

American Federation for Aging Research (Breakthroughs in Gerontology Award)

  • Heinrich Jasper

Alfred Benzon Foundation (Postdoctoral Fellowship)

  • Martin Borch Jensen

National Institute on Aging (R01 AG050104)

  • Yanyan Qi
  • Rebeccah Riley
  • Heinrich Jasper

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by the National Institute on Aging (NIH R01 AG028127 and R01 AG050104), the American Federation for Aging Research (Breakthroughs in Gerontology award to HJ), and by an Alfred Benzon fellowship to MBJ. We thank Drs. Oscar Puig, Hidenori Ichijo, Dirk Bohmann, Pierre Leopold, Samantha H Loh, Norbert Perrimon, Marc Tatar and the Bloomington and Vienna stock centers for antibodies and flies. We thank Suzy Jackson for assistance with metabolite experiments. We further thank Olivia Murillo, Maurice Brady, Margot Devincenzi and Meredith Nix for fly maintenance.

Reviewing Editor

  1. Utpal Banerjee, University of California, Los Angeles, United States

Version history

  1. Received: March 17, 2017
  2. Accepted: September 8, 2017
  3. Accepted Manuscript published: September 11, 2017 (version 1)
  4. Version of Record published: September 26, 2017 (version 2)
  5. Version of Record updated: April 10, 2018 (version 3)

Copyright

© 2017, Borch Jensen et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,289
    Page views
  • 720
    Downloads
  • 37
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Martin Borch Jensen
  2. Yanyan Qi
  3. Rebeccah Riley
  4. Liya Rabkina
  5. Heinrich Jasper
(2017)
PGAM5 promotes lasting FoxO activation after developmental mitochondrial stress and extends lifespan in Drosophila
eLife 6:e26952.
https://doi.org/10.7554/eLife.26952

Categories and tags

Research organism

Further reading

    1. Cell Biology
    2. Medicine

    Glucagon-like peptide-1 receptor activation stimulates PKA-mediated phosphorylation of Raptor and this contributes to the weight loss effect of liraglutide

    Thao DV Le, Dianxin Liu ... Julio E Ayala
    Research Article Updated

    The canonical target of the glucagon-like peptide-1 receptor (GLP-1R), Protein Kinase A (PKA), has been shown to stimulate mechanistic Target of Rapamycin Complex 1 (mTORC1) by phosphorylating the mTOR-regulating protein Raptor at Ser791 following β-adrenergic stimulation. The objective of these studies is to test whether GLP-1R agonists similarly stimulate mTORC1 via PKA phosphorylation of Raptor at Ser791 and whether this contributes to the weight loss effect of the therapeutic GLP-1R agonist liraglutide. We measured phosphorylation of the mTORC1 signaling target ribosomal protein S6 in Chinese Hamster Ovary cells expressing GLP-1R (CHO-Glp1r) treated with liraglutide in combination with PKA inhibitors. We also assessed liraglutide-mediated phosphorylation of the PKA substrate RRXS*/T* motif in CHO-Glp1r cells expressing Myc-tagged wild-type (WT) Raptor or a PKA-resistant (Ser791Ala) Raptor mutant. Finally, we measured the body weight response to liraglutide in WT mice and mice with a targeted knock-in of PKA-resistant Ser791Ala Raptor. Liraglutide increased phosphorylation of S6 and the PKA motif in WT Raptor in a PKA-dependent manner but failed to stimulate phosphorylation of the PKA motif in Ser791Ala Raptor in CHO-Glp1r cells. Lean Ser791Ala Raptor knock-in mice were resistant to liraglutide-induced weight loss but not setmelanotide-induced (melanocortin-4 receptor-dependent) weight loss. Diet-induced obese Ser791Ala Raptor knock-in mice were not resistant to liraglutide-induced weight loss; however, there was weight-dependent variation such that there was a tendency for obese Ser791Ala Raptor knock-in mice of lower relative body weight to be resistant to liraglutide-induced weight loss compared to weight-matched controls. Together, these findings suggest that PKA-mediated phosphorylation of Raptor at Ser791 contributes to liraglutide-induced weight loss.

    1. Cell Biology
    2. Developmental Biology

    Cylicins are a structural component of the sperm calyx being indispensable for male fertility in mice and human

    Simon Schneider, Andjela Kovacevic ... Hubert Schorle
    Research Article

    Cylicins are testis-specific proteins, which are exclusively expressed during spermiogenesis. In mice and humans, two Cylicins, the gonosomal X-linked Cylicin 1 (Cylc1/CYLC1) and the autosomal Cylicin 2 (Cylc2/CYLC2) genes, have been identified. Cylicins are cytoskeletal proteins with an overall positive charge due to lysine-rich repeats. While Cylicins have been localized in the acrosomal region of round spermatids, they resemble a major component of the calyx within the perinuclear theca at the posterior part of mature sperm nuclei. However, the role of Cylicins during spermiogenesis has not yet been investigated. Here, we applied CRISPR/Cas9-mediated gene editing in zygotes to establish Cylc1- and Cylc2-deficient mouse lines as a model to study the function of these proteins. Cylc1 deficiency resulted in male subfertility, whereas Cylc2-/-, Cylc1-/yCylc2+/-, and Cylc1-/yCylc2-/- males were infertile. Phenotypical characterization revealed that loss of Cylicins prevents proper calyx assembly during spermiogenesis. This results in decreased epididymal sperm counts, impaired shedding of excess cytoplasm, and severe structural malformations, ultimately resulting in impaired sperm motility. Furthermore, exome sequencing identified an infertile man with a hemizygous variant in CYLC1 and a heterozygous variant in CYLC2, displaying morphological abnormalities of the sperm including the absence of the acrosome. Thus, our study highlights the relevance and importance of Cylicins for spermiogenic remodeling and male fertility in human and mouse, and provides the basis for further studies on unraveling the complex molecular interactions between perinuclear theca proteins required during spermiogenesis.