The Rab GTPase activating protein TBC-2 regulates endosomal localization of DAF-16 FOXO and lifespan.

FOXO transcription factors have been shown to regulate longevity in model organisms and are associated with longevity in humans. To gain insight into how FOXO functions to increase lifespan, we examined the subcellular localization of DAF-16 in C. elegans. We show that DAF-16 is localized to endosomes and that this endosomal localization is increased by the insulin-IGF signaling (IIS) pathway. Endosomal localization of DAF-16 is modulated by endosomal trafficking proteins. Disruption of the Rab GTPase activating protein TBC-2 increases endosomal localization of DAF-16, while inhibition of TBC-2 targets, RAB-5 or RAB-7 GTPases, decreases endosomal localization of DAF-16. Importantly, the amount of DAF-16 that is localized to endosomes has functional consequences as increasing endosomal localization through mutations in tbc-2 reduced the lifespan of long-lived daf-2 IGFR mutants, depleted their fat stores, and DAF-16 target gene expression. Overall, this work identifies endosomal localization as a mechanism regulating DAF-16 FOXO, which is important for its functions in metabolism and aging.

Introduction

Insulin/insulin-like growth factor signaling (IIS) is an evolutionarily conserved pathway that plays an important role in lifespan, development, metabolism, immunity and stress responses from Caenorhabditis elegans to humans [1-4]. IIS-mediated regulation of C. elegans FOXO, DAF-16, was first identified via genetic characterization of mutants affecting dauer (an alternative stress resistant larval stage) and life span [5-8]. For example, animals that have a mutation in their insulin-like growth factor receptor (IGFR), DAF-2, live almost two times longer than wild-type animals in a DAF-16-dependent manner [6]. Under favorable conditions, the C. elegans DAF-2 IGFR signals through a conserved PI3K/Akt pathway to phosphorylate and inhibit nuclear accumulation of DAF-16 FOXO [1]. In short, DAF-2 IGFR activation leads to the activation of AGE-1 PI3K which phosphorylates PI(4,5)P2 to generate PI(3,4,5)P3, which can bind and recruit PDK-1 and AKT-1/2 kinases. DAF-18 PTEN, a lipid and protein phosphatase, countacts AGE-1 PI3K and thus negatively regulates signaling. PDK-1 activates AKT-1/2 which in turn phosphorylates DAF-16 creating binding sites for PAR-5 and FTT-2 14-3-3 scaffold proteins that sequester DAF-16 FOXO in the cytoplasm. During adverse conditions such as starvation, IIS is suppressed and DAF-16 FOXO enters the nucleus to induce the expression of stress response genes. As such, DAF-16 FOXO mediates most daf-2 IGFR mutant phenotypes.

Upon activation, the Insulin/IGF receptor is internalized into endosomes, where it can disassociate from its ligand and recycle back to the plasma membrane, or it can be targeted for lysosomal degradation [9]. The identification of activated IGFR on endosomes suggested that endosomes can serve as a platform for signaling [10]. Subsequently, several components of the IIS pathway have been shown to localize on endosomes. PTEN localizes on PI(3)P positive endosomes through its C2 domain and has been demonstrated to regulate endosome trafficking via dephosphorylation of Rab7 [11,12]. Akt2 localizes to Appl-1 and WDFY-2 positive endosomes to get fully activated and regulate Akt2 specific downstream substrates [13,14]. 14-3-3 proteins can interact with several Akt phosphorylation targets to regulate their subcellular localization and have been found on endosomes [15,16]. However, a role for endosome trafficking in regulation of FOXO transcription factors has not been demonstrated to the best of our knowledge.

Rab5 and Rab7 GTPases localize to early and late endosomes respectively and are critical regulators of trafficking to the lysosome, an organelle important for cargo degradation and metabolic signaling [17,18]. Like many small GTPases, Rabs cycle between a GTP-bound active state and GDP-bound inactive state. This cycling requires guanine nucleotide exchange factors for activation and GTPase Activating Proteins (GAPs) to catalyze GTP hydrolysis and hence Rab inactivation. We previously characterized C. elegans TBC-2 as having in vitro GAP activity towards RAB-5, and some activity towards RAB-7 [19]. Mutations in tbc-2 result in enlarged late endosomes in several tissues including the intestine, an important site of IIS and metabolic regulation [20]. In addition to early to late endosome maturation, TBC-2 regulates phagosome maturation [21], dense core vesicle maturation [22] endosome recycling as an effector of RAB-10 and CED-10/Rac [23,24] and yolk protein trafficking in oocytes and embryos [25]. Yolk protein is prematurely degraded in tbc-2 mutants. As such, tbc-2 mutant larvae hatched in the absence of food had reduced survival during L1 (first larval stage) diapause [25].

Here we explore the subcellular localization of DAF-16 and show that DAF-16 localizes on early and late endosomes in C. elegans intestinal cells. We found that endosome localization of DAF-16 is regulated by nutrient availability and IIS. We show that endosome localization of DAF-16 is increased through mutations in tbc-2, at the expense of nuclear localization. The increased endosomal localization of DAF-16 in tbc-2 mutants decreases lifespan, fat storage and DAF-16 target gene expression in daf-2 IGFR mutant animals. These results demonstrate a role of endosomal localization in the regulation and function of DAF-16 FOXO.

Results

TBC-2 and the RAB-5 and RAB-7 GTPases regulate DAF-16 localization to endosomes in the intestine

We previously reported that tbc-2 is required for survival during L1 diapause [25]. Since daf-16 is also required for survival during L1 diapause [26,27] we sought to test whether TBC-2 might regulate the nuclear versus cytoplasmic localization of DAF-16. Although we determined that loss of tbc-2 did not result in precocious development during L1 diapause as seen daf-16 mutants [25,27], we found that TBC-2 does in fact regulate DAF-16 localization. Unexpectedly, we found that DAF-16a::GFP (zIs356) localized to numerous amorphous vesicles in the intestinal cells of tbc-2(tm2241) deletion mutant animals (Fig 1B). Furthermore, we found that DAF-16a::GFP localized to cytoplasmic vesicles in the intestine of wild-type animals (Fig 1A). The percentage of hermaphrodites with DAF-16 vesicles increased during larval development, peaking at L4 and young adults (Figs 1I and S1A). tbc-2(tm2241) animals were about twice as likely to have DAF-16 vesicles than wild type (Figs 1J and S1A). The number of DAF-16 positive vesicles in wild type can range from zero to hundreds (Fig 1I). DAF-16 positive vesicles can be distributed throughout all 20 intestinal cells or be present in high numbers in just a few cells. DAF-16a::GFP localization is not sex specific as we found similar numbers of DAF-16 positive vesicles in males as in hermaphrodites (S1B Fig). Although DAF-16a::GFP nuclear localization is low in normal growth conditions, we find that tbc-2(tm2241) intestinal nuclei appear to have less nuclear DAF-16a::GFP than wild type (Fig 1A and 1B). We quantified the fluorescence intensity of DAF-16a::GFP in nuclei of tbc-2(tm2241) intestinal cells comparing nuclei of cells with vesicles versus nuclei of cells without vesicles (Fig 1K). We found that nuclei of cells with DAF-16 positive vesicles have significantly less nuclear DAF-16a::GFP than cells without vesicles. Thus, DAF-16 localizes to vesicles in wild-type and tbc-2(tm2241) animals, and tbc-2(tm2241) animals have less nuclear DAF-16a::GFP, likely due to sequestration to cytoplasmic vesicles.

Fig 1

DAF-16 FOXO localizes to vesicles in intestinal cells.

(A-H) Representative confocal and differential interference contrast (DIC) images of an intestinal cell of wild-type (A, C, E, G) and tbc-2(tm2241) (B, D, F, H) animals expressing DAF-16a::GFP (zIs356). DAF-16a::GFP (green) is present on vesicles in both wild-type and tbc-2(tm2241) intestinal cells (A and B) that are positive for autofluorescence (blue) present in the the endolysosomal system (E and F). Corresponding DIC (C and D) and merged (G and H) images are shown. A representative vesicle is shown (arrow head) and the two nuclei of the binucleate cell are marked (n). Note that DAF-16a::GFP is excluded from the large nucleoli (A). (I-J) Grouped bar graphs quantifying the percentage of wild-type (I) and tbc-2(tm2241) (J) animals with 0, 1–10, 11–50 or >50 DAF-16a::GFP (zIs356) positive vesicles at the different larval stages (L1-4) or 1 and 3 day old adults (A1 and A3). Raw data is available in S1 Data. (K) A Tukey boxplot of the nuclear intensity in artificial units (a.u.) of DAF-16a::GFP in tbc-2(tm2241) intestinal cells with and without DAF-16a::GFP positive vesicles. Raw data is available in S1 Data. n, total number of animals *** P<0.001 in an unpaired t test (Prism 8). Scale bar (A), 10μm.

To determine if the localization of the GFP tag or the splice variant or expression levels affected DAF-16 vesicular localization we analyzed three other transgenic strains with lower expression levels; GFP::DAF-16a (muIs71), DAF-16a::RFP (lpIs12) and DAF-16f::GFP (lpIs14, with an alternative N-terminus)[28-30]. Using the three fluorescent reporters, we detected DAF-16 vesicles in the tbc-2(tm2241) mutant, albeit not in the wild-type background (S2G–S2X Fig). Overexpression of GFP from vha-6 intestinal specific promoter, vhEx1[Pvha-6::GFP], did not show significant vesicular localization in wild-type animals (S3A Fig). In tbc-2(tm2241) animals GFP showed some vesicular localization and potential aggregates, but much less than seen with DAF-16a::GFP, indicating that the vesicular localization is due to DAF-16 and not the GFP tag. Of note, we found that GFP expression in the tbc-2 background was visibly stronger than in wild type (S3B–S3E Fig), consistent with vha-6 ranking amongst the top downregulated DAF-16-responsive genes [31] and consistent with TBC-2 facilitating DAF-16 nuclear localization.

To determine if endogenous DAF-16 localizes to vesicles we analyzed daf-16(hq23), a DAF-16::GFP line generated by CRISPR/Cas9 genome editing [32], and found that endogeneously tagged DAF-16 localized to vesicles in both wild type and tbc-2(tm2241) mutants (S2A–S2F Fig). To determine if the GFP/RFP tag is driving vesicular localization of DAF-16 we analyzed two daf-16 alleles endogenously tagged with evolutionarily distant mNeongreen and mKate2 fluorescent proteins [33,34]. We found that DAF-16::mNG and DAF-16::mK2 both localized to intestinal vesicles in wild-type animals (Fig 2). DAF-16 positive vesicles are marked with arrowheads and are distinct from intestinal autofluorescence in the other fluorescence channels. Furthermore, the percent wild-type and tbc-2(tm2241) animals with endogenous DAF-16::mNG vesicles were comparable to that of the overexpressed DAF-16a::GFP (Fig 2I). Therefore, DAF-16 localization to vesicles is not due to overexpression and unlikely to be an artifact of the fluorescent tag.

Fig 2

Endogenously tagged DAF-16::mNeongreen and DAF-16::mKate2 localize to vesicles.

Representative confocal and differential interference contrast (DIC) images of intestinal cells of daf-16(ot853[daf-16::linker::mNeongreen::3xFlag::AID]) (A,C,E,G) and daf-16(ot821[daf-16::mKate2::3xFlag]) (B,D,F,H). Arrows mark DAF-16::mNeongreen positive vesicles in the green channel (A) that are distinct from autofluorescence in the red (C) and blue channels shown as a merge (E). Arrows mark DAF-16::mKate2 positive vesicles in the red channel (B) that are distinct from autofluorescence in the green (D) and blue channels shown as a merge (F). For additional context the fluorescent channels were merged with their corresponding DIC image (G and H). Bar graphs displaying the percent wild-type and tbc-2(tm2241) animals with DAF-16::mNG positive vesicles (I). Raw data is available in S1 Data. Fisher’s exact test (graphpad.com) was used to determine the statistical difference between conditions. n, total number of animals. **** P<0.0001. Scale bar (A), 5μm.

The finding that DAF-16 localizes to enlarged vesicles in tbc-2 mutants and that DAF-16 vesicles contain autofluorescent material (Figs 1E–1H, 2E and 2F), a hallmark of intestinal endolysosomes [Mech Ageing Dev. 1986 ">35-37], suggested that DAF-16 vesicles are endosomal. To determine if the DAF-16 vesicles are endosomes, we co-expressed DAF-16a::GFP with RFP::RAB-5, an early endosomal marker, or with mCherry::RAB-7, a late endosomal marker [38-40]. We found that in wild-type animals, DAF-16a::GFP localizes to a subset of RAB-5 and RAB-7-positive endosomes (Fig 3A–3F). Thirty percent of DAF-16a::GFP vesicles were RAB-5 positive (n = 959) while 5% were RAB-7 positive (n = 365). Furthermore, we tested the genetic requirements for rab-5 and rab-7 for DAF-16 localization by RNAi. We found that rab-5(RNAi) and rab-7(RNAi) knockdown significantly decreased the number of wild-type and tbc-2(tm2241) animals with DAF-16a::GFP vesicles (Fig 3G and 3H). Together, these data are consistent with DAF-16 localizing to a sub-population of endosomes or endosome-like vesicles in the intestinal cells. It further implicates endosomal regulators TBC-2, RAB-5 and RAB-7 as regulators of DAF-16.

Fig 3

RAB-5 and RAB-7 GTPases promote DAF-16 FOXO localization to endosomes.

(A-F) Representative confocal images of the intestinal cells of animals expressing DAF-16a::GFP (zIs356) (green) and either RFP::RAB-5 (pwIs480) (A-C) or mCherry::RAB-7 (pwIs429) (magenta) (D- F). Arrowheads mark examples of vesicles positive for both DAF-16a::GFP and either RFP::RAB-5 or mCherry::RAB-7. Arrows mark examples of DAF-16a::GFP vesicles that are not positive for either RFP::RAB-5 or mCherry::RAB-7. (G and H) Bar graphs displaying the mean (and SEM) of the percent wild-type and tbc-2(tm2241) animals with DAF-16a::GFP (zIs356) vesicles fed bacteria expressing control empty vector (ev) RNAi (black bars) compared to animals fed rab-5(RNAi) or rab-7(RNAi) (white bars) from three independent experiments. Raw data is available in S1 Data. Fisher’s exact test (graphpad.com) was used to determine the statistical difference between conditions. n, total number of animals. * P<0.05, *** P<0.001, **** P<0.0001.

Acute starvation suppresses DAF-16 localization to endosomes

Nuclear localization of DAF-16 is modulated by nutrient availability. As such, starvation promotes DAF-16 cytoplasmic-to-nuclear shuttling [28]. To determine if endosomal DAF-16 can translocate to the nucleus, we tested the effect of acute starvation on DAF-16a::GFP localization to endosomes. We found that starvation strongly suppressed the localization of DAF-16 to endosomes in both wild type and tbc-2(tm2241) mutants (Figs 4A and S4). Upon re-feeding, DAF-16 relocalized to endosomes after 1–2 hours, in both wild type and tbc-2(tm2241) mutants. Thus, DAF-16 localization on endosomal membranes is regulated by nutrient availability.

Fig 4

DAF-16 FOXO endosome localization is regulated by IIS and nutrient availability.

(A) Bar graphs of the percent wild-type (black) and tbc-2(tm2241) (red) animals with DAF-16a::GFP (zIs356) vesicles in fed animals, animals starved between 4 and 5 hours and starved animals that have been refed for 1 to 2 hours. (B) Diagram of IIS-mediate regulation of DAF-16/FOXO. (C) Grouped bar graphs quantifying the percentage of wild-type and daf-2(e1370) L4 larvae (cumulative data from 3 independent strains) with 0, 1–10, 11–50 or >50 DAF-16a::GFP (zIs356) positive vesicles at 15°C or shifted overnight to 25°C. (D) Bar graphs of percent wild-type and daf-18(e1375) (cumulative data from 3 independent strains) and daf-18(ok480) (cumulative data from 5 independent strains) with DAF-16a::GFP (zIs356) vesicles. (E) Bar graphs of percent wild-type and akt-2(ok393) (cumulative data from 2 independent strains) with DAF-16a::GFP (zIs356) vesicles. (F) Bar graphs of percent wild-type animals treated with control empty vector RNAi and akt-1(RNAi) with DAF-16a::GFP (zIs356) vesicles. (G) Bar graphs of percent wild-type, sgk-1(ok538) and sgk-1(ft15) with DAF-16a::GFP (zIs356) vesicles. (H) Bar graphs of the percent wild-type (black) and tbc-2(tm2241) (red) animals fed control empty vector RNAi, par-5(RNAi) and ftt-2(RNAi) with DAF-16a::GFP (zIs356) vesicles. (I) Bar graph of the precent tbc-2(tm2241) animals with GFP::DAF-16a (muIs71) or DAF-16aAM::GFP (muIs113)-positive vesicles. Raw data is available in S1 Data. Representative images of GFP::DAF-16a (muIs71) (top) and DAF-16aAM::GFP (muIs113) (bottom) are shown. GFP::DAF-16a vesicles (arrows) and the nuclei (n). Fisher’s exact test (graphpad.com) was used to determine the statistical difference between conditions. n, total number of animals. ns, not significant, * P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001.

Insulin/IGF signaling regulates DAF-16 localization to endosomes

To determine if IIS regulates DAF-16 localization to endosomes we analyzed the effect of disrupting IIS on DAF-16a::GFP localization (see Fig 4B for reference). Firstly, we analyzed a hypomorphic mutant of the insulin/IGF receptor, daf-2(e1370), in which IIS is reduced, particularly at higher temperatures [41]. We compared DAF-16a::GFP localization in three independent daf-2(e1370) strains at 15°C and shifted overnight to 25°C to enhance disruption of DAF-2. All three showed a significant reduction in the number of DAF-16 positive vesicles at 25°C, while the wild-type strain did not (Fig 4C), indicating that DAF-2 promotes DAF-16 localization to endosomes.

DAF-18, the homolog of human tumor suppressor protein PTEN, acts as a negative regulator of the IIS pathway counteracting AGE-1 PI3K signaling by dephosphorylating the 3’phosphate on the PI(3,4,5)P 3 converting it to PI(4,5)P2 [42]. We analyzed DAF-16a::GFP localization in the daf-18 reference allele, e1375. We generated three independent daf-18(e1375); zIs356 strains, but only two strains had increased endosome localization of DAF-16. However, combined data from the three strains had a statistically significant increase in the number of animals with DAF-16 endosomes (Fig 4D). Since daf-18(e1375) is possibly a non-null allele, we analyzed a deletion allele, daf-18(ok480). We found that five independent daf-18(ok480); zIs356 strains had more DAF-16 endosomes than wild type that when combined was statistically significant (Fig 4D). Since DAF-16 endosome localization is mildly increased in the background of two distinct daf-18 alleles, it is consistent with increased IIS causing an increase in DAF-16 localization to endosomes.

C. elegans AKT-1 and AKT-2 function upstream of, and can phosphorylate, DAF-16 [43,44]. The SGK-1 Serum Glucorticoid Kinase homolog interacts with AKT kinases and has been shown to regulate DAF-16 nuclear localization [44]. We tested which of these kinases might regulate DAF-16 localization to endosomes. We found that animals with an akt-2(ok393) deletion mutation, there was an insignificant decrease in DAF-16 endosomal localization in two independent strains (Fig 4E). However, in akt-1(RNAi) animals, we observed a significant decrease in DAF-16 endosomal localization where DAF-16 localized mainly to the nucleus, compared to the control animals in three independent experiments (Fig 4F). Neither an sgk-1(ok538) deletion nor an sgk-1(ft15) gain-of-function allele affected DAF-16 localization to endosomes [44,45](Fig 4G). Therefore, AKT-1 is the main kinase regulating DAF-16 localization to endosomes.

Phosphorylation of FOXO proteins by Akt creates binding sites for 14-3-3 proteins, which promote cytoplasmic retention of FOXO [46]. C. elegans has two genes that encode for 14-3-3 proteins that interact with DAF-16, par-5 and ftt-2 [47-49]. We found that ftt-2(RNAi), but not par-5(RNAi), suppressed the localization of DAF-16 to endosomes (Fig 4H), suggesting that FTT-2 regulates DAF-16 endosomal localization.

To determine if AKT phosphorylation regulates DAF-16 localization to endosomes we tested if DAF-16aAM::GFP muIs113, in which the four consensus AKT phosphorylation sites (T54A, S240A, T242 and S314A) are mutated to alanine [29], could still localize to vesicles in tbc-2(tm2241) mutants. Since the expression levels are lower than DAF-16a::GFP zIs356, we used GFP::DAF-16a muIs71 with comparable expression levels as control (S5 Fig). We found that the number of tbc-2(tm2241) animals with DAF-16aAM::GFP positive vesicles was significantly less than the GFP::DAF-16a control (Fig 4I). Together, these data demonstrate that IIS regulates endosomal DAF-16 through AKT-1 specific phosphorylation and FTT-2 14-3-3.

TBC-2 is required for lifespan extension and increased fat storage of daf-2(e1370) mutants

To determine if the increased endosomal localization of DAF-16 seen in tbc-2 mutants affects DAF-16 activity, we tested if TBC-2 regulates adult lifespan. We found that the tbc-2(sv41) and tbc-2(tm2241) deletion alleles had similar lifespans as compared to wild type (Fig 5A and 5B). As expected, daf-2(e1370) animals lived significantly longer than wild type. We found that both tbc-2 alleles significantly shortened the lifespan of daf-2(e1370) animals (Fig 5A and 5B). Thus, TBC-2 is not required for normal lifespan, but is partly required for the extended lifespan of daf-2(e1370) mutants.

Fig 5

TBC-2 is required for full lifespan extension and increased fat storage resulting from decreased insulin/IGF signaling.

(A,B) Percent survival curve of adult animals of wild-type and daf-2(e1370) with either tbc-2(tm2241) and tbc-2(tm2241); daf-2(e1370) (A) or tbc-2(sv41) and tbc-2(sv41); daf-2(e1370) (B) genotypes. The survival of all strains are statistically different from daf-2(e1370) animals as determined by a Mantel-Cox log-rank test: tbc-2(tm2241); daf-2(e1370) (P<0.001) and tbc-2(sv41); daf-2(e1370) (P = 0.0314). Cumulative data from three independent replicates of 50 young adults for a total of 150. (C,E) Representative images of wild-type, daf-2(e1370), tbc-2(tm2241) and tbc-2(tm2241); daf-2(e1370) L4 hermaphrodites fixed and stained with Nile Red (C) or Oil Red O (E). (D,F) Log2 fold change in Nile Red (NR) fluorescence intensity (D) or Oil Red O staining intensity (F) comparing wild-type, daf-2(e1370), tbc-2(tm2241) and tbc-2(tm2241); daf-2(e1370) fixed L4 hermaphrodites. Raw data is available in S1 Data. Statistical analysis was done using a student’s t-test with a one way analysis of variance (ANOVA). ns, not significant. ** P <0.01, **** P <0.0001.

To assess TBC-2’s contribution to other daf-2 mutant phenotypes, we tested whether TBC-2 is required for the increased fat storage of daf-2(e1370) mutants. Consistent with previous findings we found that daf-2(e1370) had significantly more fat than wild-type animals as determined by Nile Red and Oil Red O staining of fixed L4 larvae [50,51] (Fig 5C–5F). We found that tbc-2(tm2241) larvae had less fat than wild type with Nile Red staining, but not with Oil Red O staining (Fig 5D and 5F). This difference might reflect the lower sensitivity of Oil Red O for quantifying lipid abundance as compared to Nile Red [52]. Interestingly, we found that tbc-2(tm2241) significantly suppressed the daf-2(e1370) increased lipid staining by both Nile Red and Oil Red O (Fig 5C–5F). Thus, TBC-2 is required for the increased fat storage of daf-2(e1370) mutants.

The fact that TBC-2 is required for the increased longevity and increased fat storage of daf-2(e1370) mutants suggests that TBC-2 could be a negative regulator of IIS. Therefore, we tested if tbc-2 mutants regulate DAF-16 target gene expression in daf-2(e1370) animals. We used qRT-PCR to measure the expression of DAF-16 target genes in wild type, daf-2(e1370), tbc-2 mutants and tbc-2; daf-2 double mutants. Consistent with previous reports, the expression of six of the DAF-16 target genes were upregulated in daf-2(e1370) animals (Fig 6) [31,53-55]. While tbc-2(tm2241) and tbc-2(sv41) did not appreciably alter DAF-16 target gene expression relative to wild type, both reduced the expression of sod-3, dod-3, gpd-2 and icl-1 in daf-2(e1370) mutants, while mtl-1 and ftn-1 were not significantly decreased (Fig 6). The fact that TBC-2 was required for the increased expression of four of six DAF-16 target genes with elevated expression in daf-2(e1370) mutants suggests a more specific role for TBC-2 in regulating DAF-2 to DAF-16 signaling.

Fig 6

TBC-2 is required for the increased DAF-16 target gene expression resulting from decreased insulin/IGF signaling.

Quantitative RT-PCR analysis of DAF-16 target gene expression in wild type, daf-2(e1370), tbc-2(tm2241), tbc-2(tm2241); daf-2(e1370), tbc-2(sv41), and tbc-2(sv41); daf-2(e1370). Statistical analysis was performed using an Ordinary one-way ANOVA–Tukey’s multiple comparisons test. Shown are comparisons of tbc-2(tm2241); daf-2(e1370) and tbc-2(sv41); daf-2(e1370) versus daf-2(e1370) single mutants. Raw data is available in S1 Data. ns, not significant. * P <0.05, ** P <0.01, *** P <0.001.

Discussion

Endosome trafficking and signal transduction are intimately linked processes regulating signal propagation, specificity and attenuation [9,56]. However, there remains a large knowledge gap in the spatial regulation of cell signaling and where downstream transcription factors are regulated. We identified a previously unknown localization for the DAF-16 FOXO transcription factor on endosomes in C. elegans. Endosome localization is limited by the TBC-2 Rab GAP. Loss of C. elegans TBC-2 results in increased endosomal localization of DAF-16 at the expense of nuclear localization. As such, C. elegans TBC-2 is partly required for several daf-2 IGFR mutant phenotypes including lifespan extension, increased fat storage, and increased DAF-16 target gene expression that result from DAF-16 nuclear translocation. DAF-16 endosome localization is largely dependent on IIS consistent with this being a phosphorylated, inactive pool of DAF-16. Together our data show a role for the TBC-2 Rab GAP in regulating the balance of nuclear versus endosomal localization of the DAF-16 transcription factor.

DAF-16 localizes to endosomes

We were surprised to find that DAF-16 FOXO localizes to a subset of RAB-5 and RAB-7 endosomes in wild-type animals. Many studies have used DAF-16 cytoplasmic versus nuclear localization to assess IIS activity under various conditions. We assume that DAF-16 positive endosomes were not discovered earlier as nuclear translocation can be assessed at low magnification where DAF-16 positive endosomes are not apparent, and DAF-16 positive endosomes are not present in every cell or every animal. Furthermore, we first took notice of DAF-16 positive endosomes in the tbc-2 mutant background where they are more prominent. These are likely not an artefactual consequence of overexpression, as we see these vesicles in the endogenously tagged daf-16::GFP and we do not see similar vesicles in a GFP overexpression strain. DAF-16::mNG and DAF-16::mK2 also localize to vesicles indicating that membrane localization is unlikely to be an artifact of the GFP tag. Additionally, DAF-16-vesicles are regulated by IIS.

Many components of the IIS pathway localize to endosomes in mammalian cells including active insulin receptor and downstream signaling components such as PI3K, Akt, PTEN and 14-3-3 proteins [9,11,15,57-61]. In the case of Akt and PTEN, both have demonstrated roles in regulation of endosome trafficking independent of IIS [12,62-64]. On the other hand, the PI3Ks are Rab5 effectors and Rab5 has been shown to promote Akt activity on endosomes [60,61,65-67]. While endosomal localization of FOXO proteins has not been reported to the best of our knowledge, knockdown of Rab5 in mouse liver results in a strong increase in phosphorylated FOXO1 [68]. This is contrary to the finding that Rab5 promotes Akt phosphorylation [61,66], which could be a consequence of indirect regulation or suggest tissue-specific regulation. The fact that TBC-2 is a RAB-5 GAP is consistent with increased RAB-5 activity promoting DAF-16 localization on endosomes. This is further supported by the fact that rab-5 and rab-7 RNAi knockdown reduces the number of animals with DAF-16 vesicles in both wild type and tbc-2 mutants. Given the importance of RAB-5 for endosome trafficking, it is difficult to parse whether RAB-5 is promoting a platform for DAF-16 localization or if it also has a role in IIS.

We demonstrated that DAF-16 localized to a subset of RAB-5 and RAB-7 positive endosomes. Since RAB-5 and RAB-7 promote trafficking to the lysosome and promote receptor tyrosine kinase degradation, it is possible that DAF-16-positive endosomes are signaling endosomes, in which case we would expect other upstream signaling components might be present. Consistent with that hypothesis, knockdown of IIS reduces the number of animals with DAF-16 vesicles. However, when analyzing DAF-16 localization in daf-2(e1370) mutants at 15°C vs. 25°C, we find that while there is a reduction in the number of DAF-16 endosomes, these endosomes are noticeably fainter at 25°C. This suggests that there are not necessarily less endosomes being generated, but rather less DAF-16 on the vesicles which may be inconsistent with these being signaling endosomes derived from DAF-2 internalization at the plasma membrane. On the other hand, the fact that Akt and 14-3-3 can localize to endosomes in mammalian cells [15,59,61] and that AKT-1 and FTT-2 promote DAF-16 localization to endosomes, suggests that these proteins might recruit DAF-16 onto endosomes rather than DAF-16 interacting directly with membranes. Future studies should test whether DAF-2 IGFR and downstream IIS components actively recruit DAF-16 to endosomes or whether IIS has a passive role. IIS inhibition of nuclear DAF-16 could result in increased DAF-16 in the cytoplasm where it can bind endosomes.

If these are not signaling endosomes, then what are the DAF-16 endosomes? Since RAB-5 and RAB-7 are also regulators of autophagy, so it is possible that these endosomes contribute to degradation of inactive excess DAF-16. There is precedent for selective autophagy in the degradation of the GATA4 transcription factor [69]. Alternatively, these endosomes could serve as a reservoir of inactive DAF-16 that can be quickly mobilized if environmental stress is encountered. For example, we found that acute starvation is a potent regulator of DAF-16 endosome localization, even in the tbc-2 mutant background.

We find it interesting that there is such variability within the population, and amongst the intestinal cells in a given animal, as to whether there will be DAF-16 positive endosomes or not. It suggests that each intestinal cell autonomously senses changes in IIS, or possibly other nutrient and stress sensing pathways, to regulate DAF-16 localization. Then, why is endosomal DAF-16 more prominent in tbc-2 mutants? One explaination would be that the expansion of endosomal membranes in a tbc-2 mutant create more storage space for inactive DAF-16. Another would be that IIS or other pathways are more active in tbc-2 mutants, or some combination of the two. The fact that loss of IIS does not eliminate DAF-16 localization from endosomes suggests that additional signaling pathways could regulate DAF-16 endosome localization. GLP-1/Notch signaling in the germline regulates longevity in a DAF-16-dependent manner as well as DAF-16 nuclear translocation [70], and IIS post-translationally regulates GLP-1 signaling [71], thus it would be interesting to determine if DAF-16 localization to endosomes are regulated by GLP-1/Notch signaling and if TBC-2 regulates GLP-1 to DAF-16 target gene expression [72]. Additionally, AMPK, JNK and LET-363/mTor signaling regulate DAF-16 and could regulate DAF-16 localization to endosomes or be subject to regulation by TBC-2, particularly mTor which localizes to lysosomes [70,73-77].

A direct role for TBC-2 in regulating daf-2(e1370) IGFR mutant phenotypes

Our finding that tbc-2 was required for the extended lifespan and increased fat storage of daf-2(e1370) mutants suggests that TBC-2 might have a more specific role related to IIS. However, TBC-2 is not the first endosomal regulator required for lifespan extension of daf-2(e1370) mutants. C. elegans BEC-1, a homolog of human Beclin1, is a regulator of autophagy and endosome trafficking, and bec-1 mutants accumulate large late endosomes in the intestinal cells [78-80]. Mutations in bec-1 suppress the increased lifespan of daf-2(e1370), and were reported to be required for the increased fat storage [78]. Similarly, other autophagy regulators, atg-7 and atg-12, have been shown to be required for daf-2 longevity [81]. An RNAi screen identified regulators of endosome to lysosome trafficking, including RAB-7, and components of the ESCRT and HOPS complexes as being required for the lifespan extension phenotypes of daf-2 IGFR mutants [82]. However, the mechanisms by which they regulate lifespan are not known. In daf-2(e1370) mutants, there is an increase in autophagy and lysosome function, both of which are required for the extended lifespan [78,83-85], consistent with increased stress resistance contributing to increased longevity. Since the human homologs of TBC-2 are implicated in autophagy [86-89], and tbc-2 mutants accumulate autophagy protein LGG-1 (LC3/Atg8) in enlarged endosomes [19], it is possible that TBC-2 also regulates autophagy and thus daf-2(-) longevity. However, our findings that TBC-2 regulates DAF-16 nuclear vs. endosome localization and that TBC-2 is required for the DAF-16 target gene expression in daf-2 mutants demonstrate that TBC-2 has a more direct role in IIS, as opposed to being required for downstream cellular responses. It will be interesting to test if TBC-2 regulates DAF-16-independent mechanisms of longevity as well as to determine if other endosome and autophagy-regulating genes can regulate DAF-16 localization to endosomes.

In conclusion, we demonstrate that the DAF-16 FOXO transcription factor localizes to endosomes. This endosomal localization of DAF-16 FOXO is regulated by both IIS and the TBC-2 Rab GAP. TBC-2 promotes the nuclear localization of DAF-16 FOXO, and this localization has functional effects on both longevity and metabolism through modulation of DAF-16 target gene expression. Our data suggest that endosomes serve as an important location for DAF-16 FOXO transcription factor regulation, and suggest that endomembranes may function as a site of transcription factor regulation.

Methods

C. elegans genetics and strain construction

C. elegans strains were cultured as described in Wormbook (www.wormbook.org). The C. elegans N2 Bristol strain was the wild-type parent strain and HB101 E. coli strain was used as a food source. Both were obtained from the Caenorhabditis Genetic Center (CGC) as were many of the strains used in this study (S1 Table). New strains were constructed using standard methods and the presence of mutations were confirmed by PCR and DNA sequencing.

RNAi experiments

C. elegans RNAi feeding experiments were conducted essentially as described in [90]. RNAi feeding clones were obtained from the Ahringer RNAi library and confirmed by sequencing (S1 Table)[91,92]. The L4440 empty RNAi feeding vector transformed into HT115(DE3) was used as a negative control [93].

Microscopy

DAF-16 vesicular localization was analyzed in the intestinal cells of hermaphrodites. Hermaphrodite worms at the L4 stage were imaged alive at room temperature unless it is stated otherwise. Animals were picked and mounted onto 4% agarose pads, animals were anesthetized with levamisole.

Differential interference contrast (DIC) and fluorescent imaging were performed with an Axio Imager A1 compound microscope with a 100×1.3 NA Plan-Neofluar oil-immersion objective lens (Zeiss) and images were captured by using an Axio Cam MRm camera and AxioVision software (Zeiss). Confocal microscopy was performed on an Axio Observer Z1 LSM780 laser scanning confocal microscope with a 63×1.4 NA Plan-Apochromat oil-immersion objective lens (Zeiss) in a multi-track mode using an argon multiline laser (405 nm excitation for autofluoresence, 488 nm excitation for GFP and a 561/ 594 nm excitation for mCherry/RFP). Images were captured with a 32 channel GaAsP detector and ZEN2010 image software. Raw data was analyzed using Fiji (ImageJ) or Zen 2010 Lit programs, and images were modified by using Fiji (ImageJ).

To compare DAF-16 (zIs356) nuclear intensity in cells with or without DAF-16 vesicles, animals at L4 stage were imaged using an LSM780 scanning laser microscope. To ensure consistency, only the anterior most intestine cells were imaged. First, the nucleus of intestinal cells were focused under bright field and without changing the position, GFP, autofluoresence and bright field signals were imaged. Each animal was imaged using the same confocal settings. After data collection, each nucleus was categorized as a nucleus with adjacent DAF-16-positive vesicles or a nucleus without any DAF-16 positive vesicles. Total GFP intensity inside the nucleus was measured using Fiji (Image J) software. The nucleus is circled using DIC/Bright Field and autofluorescence channels as reference. Since intestinal cells have two nuclei per cells, if there are two nuclei within the focus, their GFP intensity is averaged for statistical analysis. Prism 8 (GraphPad) were used to graph the data and determine statistical analysis using an unpaired t-test.

Starvation-refeeding and temperature shift experiments

For starvation and refeeding experiments animals were synchronized at the L1 stage and grown on NGM plates with HB101 E. coli till the L4 stage. Then animals were collected washed 3 times for 5 mins with M9 buffer to remove bacteria in their gut. After the third wash animals were plated to regular NGM plates with or without HB101 E. coli, and incubated for 4–5 hours before scoring. Animals are scored for the presence of DAF-16::GFP-positive vesicles the intestine using an A1 Zeiss microscope. After 4h-5h of starvation, animals were harvested with M9 buffer from the starved plates and washed once with M9 buffer and plated to NGM plates with HB101 E. coli and incubated for 1–2 hours at 20°C before scoring.

Life span analysis

Replicate strains were maintained for several generations prior to beginning the lifespan assays which were conducted at 20°C. For each strain 25 young adult hermaphrodites were picked to two NGM plates without FUDR seeded with HB101 E. coli for three independent replicates totaling 150 animals. Strains were coded and scored blindly to reduce bias. Animals were transferred to fresh NGM plates to avoid contamination and getting crowded out by their progeny. Animals were scored every 2–3 days and were considered dead when they stop exhibiting spontaneous movement and fail to move in response to 1) a gentle touch of the tail, 2) a gentle touch of the head, and 3) gently lifting the head. Animals that die of unnatural causes (internal hatching of embryos, bursting, or crawling off the plate) are omitted. Graphs and statistics were done using Graphpad Prism. None of the strains used in the lifespan assays carry the fln-2(ot611) mutation found in a N2 male stock strain and found to extend median lifespan [94,95].

Fat staining

L4 animals were fixed for staining with Nile Red (Invitrogen) or Oil Red O (Sigma-Aldrich) as previously described [52]. Imaging and analysis was done as previously described [96]. Graphs and statistics were done using Graphpad Prism.

Quantitative real-time RT-PCR

C. elegans RNA was isolated from young adults maintained at 15°C using TRIZOL reagent (Invitrogen). 1 ug of RNA was reverse transcribed into cDNA using the High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems). Quantitative real-time PCR was performed using 1 μl of the cDNA preparation with SYBR-Green Reagents and a Vii7 qPCR analyzer (Applied Biosystems). Each DAF-16 target gene was amplified using PCR primers as described in [97] and compared to act-3 (S1 Table).

Statistical analysis

Statistical analysis was carried out using GraphPad Prism software. For the analysis of two groups, a student t test was performed using two-tailed distribution for analysis involving two groups of samples. Fishers’ exact test was used for comparing groups of four. For each analysis, P<0.05 was considered as significant.

DAF-16::GFP localizes to vesicles in wild-type and tbc-2 animals at all post-embryonic developmental stages and in wild-type males.

(A) Bar graph of the percent animals with DAF-16a::GFP (zIs356) positive vesicles in the intestinal cells at larval stages L1-L4 and young adults of wild-type and tbc-2(tm2241) animals at 20°C. Fisher’s exact test (graphpad.com) was used to determine that there is a significant increase the number of tbc-2(tm2241) animals with DAF-16a::GFP as compared to wild type at each developmental stage (L1: P<0.05, L2-adult: P<0.0001, n = 23 to 47 animals). (B) Bar graph of the percent zIs356/+ L4/young adult hermaphrodites and males with DAF-16a::GFP positive vesicles. Raw data is available in S1 Data. ns, not significant, n = 47–50 animals.

(TIF)

Click here for additional data file.

Representative confocal and differential interference contrast (DIC) images of intestinal cells of wild-type (A-C, G-I, M-O and S-U) and tbc-2(tm2241) (D-F, J-L, P-R and V-X) animals expressing DAF-16::GFP daf-16(hq23) (A-F), muIs71 GFP::DAF-16a (G-L), lpIs12 DAF-16a::RFP (M-R) and lpIs14 DAF-16f::GFP (S-X). Endogenously tagged DAF-16::GFP daf-16(hq23) is present on vesicles in both wild-type and tbc-2(tm2241) intestinal cells (A and D arrowheads). Arrows mark bright autofluorescent lysosome-related organelles that bleed through the GFP channel in these lower expressing strains. Vesicular localization of GFP::DAF-16a (G and J), DAF-16a::RFP (M and P) and DAF-16f::GFP (S and V) was only seen in tbc-2(tm2241) animals and not visible in wild-type backgrounds. Scale bars (A, G, M, S), 10μm.

(TIF)

Click here for additional data file.

GFP localization and expression in wild-type and tbc-2 mutants.

(A) Grouped bar graph quantifying the number of wild-type and tbc-2(tm2241) L4 larvae with 0, 1–10, 11–50 or >50 GFP (vhEx1[Pvha-6::GFP]) positive vesicles or aggregates (aggregates were included in the analyses which are not often seen with DAF-16a::GFP). Raw data is available in S1 Data. (B-E) DIC and epifluorescence images of wild-type and tbc-2(tm2241) (B,C) as well as wild-type and tbc-2(sv41) (D,E) animals expressing GFP under an intestine specific promoter, vhEx1 [Pvha-6::GFP]. White and yellow arrowheads mark the anterior of the intestine of wild-type and tbc-2 mutant animals, respectively. Both tbc-2 mutants have increased GFP expression as compared to wild-type animals. tbc-2 mutants were distinguished from wild-type by the presence of enlarged vesicles using the 100X objective (not shown).

(TIF)

Click here for additional data file.

Endosomal DAF-16 is suppressed by acute starvation.

Epifluorescence (A,C,E,G,I,K) and DIC (B,D,F,H,J,L) images of wild-type (A-F) and tbc-2(tm2241) (G-L) animals under fed (A,B,G,H), 4–5 hours of starvation (C,D,I,J) and after 1–2 hours of refeeding (E,F,K,L). Scale bar (A), 10μm.

(TIF)

Click here for additional data file.

GFP::DAF-16 and DAF-16AM::GFP have similar expression levels.

Bar graph depicting the mean pixel intensities of GFP fluorescence in the intestine of QR508 tbc-2(tm2241); muIs71[GFP::DAF-16] and QR697 tbc-2(tm2241); muIs113[DAF-16AM::GFP] animals. Raw data is available in S1 Data. The difference was determined to be not significant (ns) in an unpaired t test.

(TIF)

Click here for additional data file.

List of key resources used in this study including bacterial strains, C. elegans strains, RNAi-feeding clones, and oligonucleotides.

(DOCX)

Click here for additional data file.

Raw numbers and statistical analyses summarized in the bar graphs.

(XLSX)

Click here for additional data file.