A secreted microRNA disrupts autophagy in distinct tissues of Caenorhabditis elegans upon ageing

Macroautophagy, a key player in protein quality control, is proposed to be systematically impaired in distinct tissues and causes coordinated disruption of protein homeostasis and ageing throughout the body. Although tissue-specific changes in autophagy and ageing have been extensively explored, the mechanism underlying the inter-tissue regulation of autophagy with ageing is poorly understood. Here, we show that a secreted microRNA, mir-83/miR-29, controls the age-related decrease in macroautophagy across tissues in Caenorhabditis elegans. Upregulated in the intestine by hsf-1/HSF1 with age, mir-83 is transported across tissues potentially via extracellular vesicles and disrupts macroautophagy by suppressing CUP-5/MCOLN, a vital autophagy regulator, autonomously in the intestine as well as non-autonomously in body wall muscle. Mutating mir-83 thereby enhances macroautophagy in different tissues, promoting protein homeostasis and longevity. These findings thus identify a microRNA-based mechanism to coordinate the decreasing macroautophagy in various tissues with age.

A geing results from the disruption of homeostasis, such as protein homeostasis (proteostasis). Proteostasis is the functional balance of the proteome of a living organism. Its disruption is closely associated with age and leads to various ageing phenotypes and ageing-related disorders 1 . One of the critical factors in maintaining proteostasis is macroautophagy (hereafter referred to as autophagy) 1 . During autophagy, cytosolic cargo (e.g., protein aggregates) is engulfed into autophagosomes and degraded in autolysosomes after autophagosomes fuse with lysosomes 2 . Autophagy is enhanced in various longevity models, whereas multiple lines of evidence suggest a gradual dysregulation of autophagy with ageing, impairing proteostasis in aged animals 3,4 . However, the mechanism underlying the decrease in autophagy has scarcely been studied.
An intriguing feature of the decreasing proteostasis with ageing is that it is systematically coordinated in tissues with distinct proteomes and undergoing disparate stresses. Consistently, two crucial prsocesses for proteostasis (i.e., the unfolded protein response and the chaperone signaling response) function in a tissue non-autonomous manner when challenged with tissue-specific perturbations, such as polyglutamine in neurons and misfolded myosin heavy chain in muscle [5][6][7][8][9] . The ageing-induced impairment of autophagy is also systematically controlled among tissues 10 , implying that tissues engage in crosstalk with each other to coordinate their autophagic activities. Previous studies on autophagy in tissues have focused on tissue-specific changes in autophagy and their functions in ageing 3 . However, little is known about the intertissue regulation of autophagy.
MicroRNAs are important regulators in various biological processes, including ageing and autophagy 11,12 . They are not only present in cells but have also been found to be secreted into different body fluids. Secreted microRNAs are widely proposed to be biomarkers of various patho-physiological conditions, including ageing 13,14 . Herein, we report a secreted microRNA, mir-83/miR-29, as a cross-tissue regulator of the decrease in autophagy with age. Upregulation of mir-83 in aged intestine by hsf-1/HSF1 impairs intestinal autophagy through inhibition of the mir-83 target CUP-5/MCOLN. Moreover, this autophagy regulation mechanism in the intestine also controls autophagy in body wall muscle (BWM). mir-83 is transported from the intestine into BWM potentially via extracellular vesicles, suppressing autophagy there by targeting BWM cup-5 per se. Our results thus reveal a secreted microRNA-based mechanism dysregulating autophagy in distinct tissues with ageing.
Results mir-83 is induced by hsf-1 during ageing in the intestine. To explore the mechanism impairing autophagy among tissues with ageing, we examined transcriptome changes in Caenorhabditis elegans at different ages. Wild-type (WT) worms at four stages of adulthood were analyzed: day 1 (D1, the start of adulthood), day 7 (D7, around the end of the reproductive period), day 14 (D14), and day 21 (D21, around the mean lifespan of WT worms). From microRNA-Seq analysis and qRT-PCR validation, mir-83, homologous to mammalian miR-29, was found to increase with ageing 15 (Fig. 1a and Supplementary Fig. 1a-b). A GFP reporter driven by the native promoter of mir-83 confirmed the upregulation of this microRNA with ageing (Fig. 1b). Based on previous reports and our observations, mir-83 is mostly expressed in the intestine and neurons ( Fig. 1c and Supplementary Fig. 1c-d) 16,17 . Intriguingly, we observed significant induction of mir-83 only in the intestine during ageing but not in neurons (Fig. 1c, d and Supplementary Fig. 1c-d), indicating that mir-83 is upregulated specifically in the intestine by ageing.
We next questioned which transcription factor controls mir-83 expression during ageing. mir-83 has been reported to be upregulated by heat shock 18 and harbors heat shock elements in its promoter ( Supplementary Fig. 1e) 19 . hsf-1/HSF1 is the master transcription factor of the heat shock response 20 . Upon using genome editing to tag endogenous hsf-1 with GFP::3xFLAG, we observed increases in HSF-1 protein levels in worms at D4 and D10 ( Supplementary Fig. 1f). Consistent with the specific upregulation of mir-83 in the aged intestine, further analysis of the fluorescent signal from HSF-1:: GFP::3xFLAG in live worms indicated that HSF-1 was upregulated in the intestines but not in the neurons of aged animals ( Supplementary Fig. 1g). We then examined whether hsf-1 controls mir-83 during ageing. Suppressing hsf-1 by RNAi completely blocked the upregulation of mir-83 expression in worms at D4 and D7 but had little effect on mir-83 levels at D1 ( Fig. 1e and Supplementary Fig. 1h). Accordingly, hsf-1 RNAi suppressed the GFP reporter of mir-83 transcription in the intestine at D7 but not at D1 (Fig. 1f). Taken together, these data indicate that hsf-1 promotes the expression of mir-83 in the aged intestine.
cup-5/MCOLN is a target of mir-83. MicroRNAs function through binding to the 3′-UTRs of target genes mRNA and inhibiting their translation. By TargetScan 21 , two autophagyrelated genes, cup-5/MCOLN and atg-4.2/ATG4D 22,23 , were predicted as potential targets of mir-83 ( Supplementary Fig. 2a). Transfection of mir-83 mimic, but not a control oligo, into HEK293T cells inhibited the expression of the luciferase reporter with the 3′-UTR of either cup-5 or atg-4.2 ( Fig. 2a and Supplementary Fig. 2b). When the mir-83 target site in the cup-5 or atg-4.2 3′-UTR was mutated, the mir-83 mimic no longer inhibited the expression of the luciferase reporter of the 3′-UTR ( Fig. 2a and Supplementary Fig. 2b). These results therefore demonstrate the interactions of mir-83 with these two genes. To confirm their interactions in vivo, we followed our previous reports and prepared transgenic animals with dual fluorescence, expressing mCherry with the native promoter and the 3′-UTR of cup-5 or atg-4.2, and a ubiquitously expressed GFP as an internal control 24,25 . If mir-83 interacts with the 3′-UTR of interest in vivo, the ratio of mCherry versus GFP (mCherry/GFP) would be expected to increase in mir-83(-) mutants (with deletion of the mir-83 gene) (Fig. 2b). As expected, the dual fluorescent reporter of cup-5 showed an elevated mCherry/GFP ratio upon mir-83 deletion (Fig. 2c). However, the mCherry/GFP ratio of the atg-4.2 3′-UTR reporter was unchanged in mir-83(-) mutants (Supplementary Fig. 2c-d), indicating that mir-83 does not interact with the 3′-UTR of atg-4.2 in vivo. We next assessed the protein level of CUP-5. Upon using an extrachromosomal transgene of mCherry::cup-5 controlled by its native promoter and 3′-UTR, we observed an elevated level of mCherry::CUP-5 in the intestines of mir-83(-) mutants at different ages (Fig. 2d). We further knocked in a 3xFLAG:: WrmScarlet tag at the 5′ of the cup-5 gene with CRISPR/Cas9 technology for more precise quantification of endogenous CUP-5 levels. Consistent with the upregulation of mCherry:: CUP-5 in mir-83(-) mutants ( Fig. 2d and Supplementary  Fig. 2f), the endogenously labeled CUP-5 was also increased in mir-83(-) mutants at different ages (Fig. 2e). Therefore, mir-83 inhibits CUP-5. The mRNA level of cup-5 was unchanged with age, with a mild but non-significant increase upon mir-83 mutation ( Supplementary Fig. 2e), implying that mir-83 inhibits cup-5 mainly by blocking its translation but not by degrading its mRNA. Collectively, the results show that cup-5 is a target of mir-83.
mir-83 impairs intestinal autophagy through cup-5. The target of mir-83, cup-5/MCOLN, encodes a calcium channel localized to the lysosomal membrane and controls multiple steps of autophagy 22,26,27 . Therefore, mir-83 could control autophagy by inhibiting CUP-5. As mir-83 is upregulated in the aged intestine ( Fig. 1c-d and Supplementary Fig. S1c-d), we first analyzed autophagy in this tissue. GFP::LGG-1 (GFP-LC3)-labeled autophagosomes (APs) are well-recognized autophagy markers 28 . Indeed, more GFP::LGG-1 puncta (APs) were observed in the intestines of mir-83(-) mutants than in the intestines of WT worms (Fig. 3a). Increases in APs could be due to either induction or inhibition of autophagy 28 . We then treated worms with chloroquine to distinguish between these two possibilities. When autophagy is active, chloroquine treatment increases AP numbers because it inhibits the transition of APs to autolysosomes (ALs) and the degradation of protein in ALs associated with active autophagy flux 29 . When autophagy is already blocked, chloroquine treatment can no longer change the number of APs. At 1 h post treatment with chloroquine, the numbers of GFP::LGG-1labeled APs were increased in the intestines of both the WT worms and the mir-83(-) mutants (Fig. 3b), indicating that autophagy flux was not blocked upon mutation of mir-83. Thus, mutation of mir-83 upregulates AP numbers to promote autophagy.
We next used mCherry-GFP-tagged LGG-1 to further examine the effect of mir-83 on autophagy. GFP, but not mCherry, in this reporter is quenched in ALs, making APs and ALs respectively visible as yellow and red puncta 10 . In the intestines of mir-83(-) mutants, the numbers of both APs and ALs were increased ( Fig. 3c), in accordance with our previous observation (Fig. 3a, b). We also validated these results with intestine-specific dual GFP (dFP)-tagged LGG-1, which releases monomeric GFP (mFP) in ALs. The mFP/dFP ratio correlates with the activity of autophagy 30 . Consistent with our observations with other autophagy metrics, increased mFP/dFP ratios were detected in mir-83(-) mutants ( Supplementary Fig. 3a).
To determine whether mir-83 controls autophagy through cup-5, cup-5-related autophagy phenotypes in WT worms and mir-83 (-) mutants were first examined. If mir-83 dysregulates autophagy by inhibiting CUP-5, RNAi against cup-5 should suppress the increase in CUP-5 and abolish the enhancement of cup-5-related autophagy phenotypes in mir-83(-) mutants. The numbers of lysosomes were first determined, since cup-5(-) mutants exhibit were normalized against the one at day 1. n = 7, 4, 3, 3, and 3 independent experiments for samples of day 1, 4, 7, 14, and 21, respectively. b Immunoblots of mir-83p::GFP at indicated ages. Blots against α-tubulin and day 1 sample serve as controls for loading and normalization respectively. n = 3 independent experiments. c Representative images of mir-83p::gfp transgenic animals at indicated ages. n: neuron, i: intestine. d GFP intensity of mir-83p::gfp in the intestine and head neurons at day 1, day 4, and day 7 of adulthood. Samples of day 1 serve as controls for normalization. n = 3 independent experiments containing at least 41 worms. e qRT-PCR of miR-83 in WT worms treated with luc2 or hsf-1 RNAi. RNAi treatment was performed from hatching and worms were examined at indicated ages. Samples of day 1 serve as controls for normalization. n = 3 independent experiments. f Expression of mir-83p::gfp under luc2 or hsf-1 RNAi. Worms were treated with RNAi from hatching and examined for GFP intensity at indicated ages. day 1 worms subjected to luc2 RNAi serve as control for normalization. n = 3 independent experiments containing at least 39 worms. Scale bars: 100 μm. Statistical significance was calculated by One-way ANOVA in (a and b), or Two-way ANOVA in d-f. ns: non-significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data are provided as a Source Data file NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12821-2 ARTICLE NATURE COMMUNICATIONS | (2019) 10:4827 | https://doi.org/10.1038/s41467-019-12821-2 | www.nature.com/naturecommunications decreased numbers of lysosomes due to impaired budding and maturation of lysosomes 22,31 . Consistently, we observed increases in the numbers of lysosomes in the intestines of mir-83(-) mutants, which was suppressed upon cup-5 RNAi (Supplementary Fig. 3c-d). Therefore, mir-83 controls lysosome biogenesis by targeting cup-5. We next examined the protein level of LGG-1/ LC3, since the mammalian orthologue of cup-5, MCOLN, activates the transcription factor EB (TFEB) to promote the expression of LC3 and other key autophagy genes 26,27 . GFP:: LGG-1 was indeed increased in mir-83(-) mutants (Supplementary Fig. 3e). RNAi against cup-5 in mir-83(-) mutants restored the expression of GFP::LGG-1 to WT levels ( Supplementary  Fig. 3f), indicating that mir-83 controls LGG-1 expression via cup-5 and implying that the MCOLN-TFEB axis may be conserved in C. elegans. Because LGG-1 and lysosomes are key to the formation of APs and ALs, we finally examined the numbers of APs and ALs in the intestines of WT worms and mir-83(-) mutants upon cup-5 RNAi. Using the mCherry-GFP::LGG-1 reporter, we found that RNAi against cup-5 completely inhibited the increases in ALs in the intestines of mir-83(-) mutants ( Supplementary Fig. 3g), indicating that cup-5 is required for the enhanced autophagy flux in mir-83(-) mutants. Intestine-specific RNAi treatment against cup-5 abolished the elevated autophagy flux in the intestines of mir-83(-) mutants as well (Fig. 3d), indicating cell-autonomous regulation of cup-5-mediated autophagy in the intestine by mir-83. Taken together, mir-83 regulates intestinal autophagy with ageing by downregulating cup-5 cellautonomously. c Fluorescent signals and immunoblots of the dual fluorescence reporter of cup-5 3′-UTR in WT worms and mir-83(-) mutants at day 1 of adulthood. Quantification is from the western blots. GFP blots and WT worms serve as controls for loading and normalization respectively. Scale bar: 100 μm. n = 3 independent experiments. d Fluorescent signals of the cup-5 translational reporter (mCherry::CUP-5) in the posterior intestine (dashed lines) and nearby BWM (yellow lines) of WT worms and mir-83(-) mutants at indicated ages. WT signals at day 1 serve as controls for normalization. Scale bar: 10 μm. n = 3 independent experiments. e The protein level of the endogenously tagged 3xFLAG::WrmScarlet::CUP-5 is increased upon mutating mir-83 at indicated ages. α-tubulin and WT blots at day 1 serve as controls for loading and normalization respectively. n = 5 independent experiments. Statistical significance was calculated by unpaired t-test in a, c, e, or two-way ANOVA in d. ns: non-significant, *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file Intestinal mir-83 disrupts autophagy in body wall muscle. Although mir-83 is not expressed in body wall muscle (BWM) ( Fig. 1c and Supplementary Fig. 1c-d) 16,17 , we unexpectedly found that autophagy in BWM is also under the control of mir-83. Upon examining the signals from GFP::LGG-1 and mCherry-GFP:: LGG-1, we observed increases in both APs and ALs in the BWM cells of mir-83(-) mutants (Fig. 4a, b). Chloroquine treatment increased the numbers of APs in BWM of WT worms and mir-83(-) mutants at D1 (Fig. 4c), indicating that autophagy is active rather than blocked in BWM upon mir-83 mutation. We LGG-1 reporter (mFP/dFP, mFP vs dFP::LGG-1) in neurons responds to the treatment of 5 mM CQ but not mir-83 mutation at day 1 of adulthood. WT samples with control treatment serve as controls for normalization. n = 3 independent experiments. g-h APs (yellow arrows) and ALs (white arrow heads) in the head neurons surrounding the posterior pharynx bulb (dashed lines). Worms of indicated genotypes were examined at day 1 and day 5 of adulthood. A CQ treatment (5 mM) was performed as indicated. n = 3 independent experiments containing at least 15 worms. Scale bars: 10 μm. Statistical significance was calculated by Poisson regression in a-d and g-h, unpaired t-test in e, or one-way ANOVA in f. ns: non-significant, *p < 0.05, ***p < 0.001. Source data are provided as a Source Data file NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12821-2 ARTICLE NATURE COMMUNICATIONS | (2019) 10:4827 | https://doi.org/10.1038/s41467-019-12821-2 | www.nature.com/naturecommunications further used a BWM-specific SQST-1::GFP transgene to confirm the elevated autophagy flux in mir-83(-) BWM 28 . As expected, SQST-1::GFP puncta in BWM were decreased in mir-83(-) mutants compared with WT worms (Fig. 4d).
Since mir-83 is induced in the intestine with ageing and is not expressed in BWM ( Fig. 1c and Supplementary Fig. 1c-d) 16,17 , we hypothesized that intestinal mir-83 may regulate autophagy in BWM. An intestine-specific mir-83 transgene indeed restored the increased ALs to WT levels in the BWM of mir-83(-) mutants while further increasing the APs in the WT worms to the levels in mir-83(-) mutants (Fig. 4e, f), indicating that intestinal mir-83 suppresses the AP-AL transition in BWM. Therefore, the induced mir-83 in the intestine is responsible for the disrupted autophagy flux in BWM during ageing.
mir-83 regulates autophagy in the intestine via cup-5 (Fig. 3), and cup-5 is also expressed in BWM (Fig. 2d) 32 . We therefore speculated that the enhanced BWM autophagy in mir-83(-) mutants could occur through cup-5 in BWM. Indeed, RNAi f The intestine-specific expression of mir-83 abolishes the increased ALs (white arrow heads) in BWM cells (dashed lines) of mir-83(-) mutants. Note that APs (yellow arrows) were increased in WT worms due to a block of the autophagy flux. n = 4 independent experiments containing at least 19 worms. Scale bars: 10 μm. Statistical significance was calculated by Poisson regression. ns: non-significant, *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file against cup-5 in various tissues or specifically in BWM restored the numbers of SQST-1::GFP puncta in the BWM of mir-83(-) mutants to WT levels ( Fig. 5a and Supplementary Fig. 4a), indicating that mir-83 regulates BWM autophagy in a cup-5dependent manner.
mir-83 is transported across tissues. Since mir-83 directly interacts with the cup-5 3′-UTR in BWM but is not expressed in this tissue (Figs. 1c and 5 and Supplementary Figs. 1c-d and 4) 16,17 , we speculated that mir-83 could be transported into BWM from other tissues, such as the intestine. To verify this hypothesis, we isolated BWM cells and performed single-cell qPCR (Fig. 6a). rgef-1, vha-6, and myo-3 (three genes specifically expressed in neurons, intestine and BWM, respectively) were also examined in isolated BWM cells ( Supplementary Fig. 5a-d). rgef-1 and vha-6 were not detected, therefore excluding contamination from the two mir-83-enriched tissues (i.e., the neurons and intestine). Indeed, mir-83 was detected in BWM cells from WT animals but not in those from mir-83(-) mutants ( Fig. 6b and Supplementary  Fig. 5e-f). In the isolated BWM cells, gfp expressed from a BWMspecific transgene was detected, but TagRFP-T expressed by the mir-83 promoter was not ( Supplementary Fig. 5c-d), confirming that mir-83 was not transcribed in BWM. In addition, BWM cells from mir-83(-) mutants carrying a transgene specifically overexpressing mir-83 in the intestine also had elevated mir-83 levels (Fig. 6b), whereas TagRFP-T driven by the same intestine-specific promoter was not detected in the isolated BWM cells (Supplementary Fig. 5c-d). Collectively, these data indicate that mir-83 is transported from the intestine into BWM.
How is mir-83 transported from the intestine into BWM? The pseudocoelom in C. elegans could mediate this transportation because it provides a medium for communication across tissues similar to the circulatory system in higher organisms and is located between the intestine and BWM. Coelomocytes in the pseudocoelom continually take in pseudocoelom fluids and are likely to be enriched with secreted mir-83 33 Fig. 1c-d) 16,17 , it was detected in isolated coelomocytes by singlecell qPCR (Fig. 6c and Supplementary Fig. 5g-h). We further examined LMP-1-labeled lysosomes in coelomocytes and found that they turned into large vacuoles at day 5 of adulthood, reminiscent of the phenotypes observed upon inhibition of cup-5 22 (Fig. 6d). As in the intestine, mutation of mir-83 suppressed the enlargement of LMP-1-positive vacuoles in the aged coelomocytes, and cup-5 RNAi restored the sizes of LMP-1positive vacuoles in the coelomocytes of mir-83(-) mutants to WT levels (Fig. 6d, e). Hence, mir-83 is also present in coelomocytes and could control lysosomes by inhibiting cup-5.
Previous studies have suggested that extracellular vesicles (EVs) transfer microRNAs among cells 34 . Worm EVs, mostly exosomes given their size, were isolated to determine whether mir-83 is also transferred in EVs (Supplementary Fig. 6a-c). mir-56 and mir-82, two microRNAs proposed to be enriched in EVs 35 , were abundant in our isolated EVs (Fig. 6f), further validating the EV preparation. As expected, mir-83 was detected in the isolated EVs from WT worms but not in those from mir-83 (-) mutants ( Fig. 6f and Supplementary Fig. 6d). These data therefore imply that mir-83 could be shuttled into BWM by EVs.

mir-83 regulates ageing by inhibiting autophagy and cup-5.
Autophagy is crucial for protein homeostasis and longevity 36 . Consistent with the effects of mir-83 on autophagy, mir-83 deletion reduced the numbers of polyQ aggregates in the intestines and BWM of worms at days 5 and 3 of adulthood, respectively (Fig. 7a, b), indicating the importance of mir-83 in protein homeostasis. Consistent with the role of mir-83 in protein homeostasis, mutation of mir-83 extended the lifespan of WT animals ( Fig. 7c; Supplementary Data 1), as recently reported 37 . mir-83(-) mutants also exhibited a mild egg-laying defect (Supplementary Fig. 7), suggesting that this microRNA functions in reproduction. Inhibiting cup-5 by RNAi abolished the decrease in polyQ aggregates and the extension of lifespan in mir-83(-) mutants ( Fig. 7d-f; Supplementary Data 1), further indicating that mir-83 promotes ageing by inhibiting cup-5 and dysregulating autophagy. Accordingly, disruption of autophagy with chloroquine reduced the lifespan of mir-83(-) mutants to that of WT worms ( Fig. 7g; Supplementary Data 1). As ageing upregulated mir-83 in the intestine (Fig. 1c, d and Supplementary  Fig. 1c-d), intestinal overexpression of mir-83 blocked the longevity of mir-83(-) mutants (Fig. 7h). Taken together, these data indicate that the ageing-related upregulation of mir-83 in the intestine impairs protein homeostasis and longevity by disrupting autophagy.

Discussion
Autophagy is gradually impaired with ageing, resulting in the disruption of proteostasis 3 . As ageing progresses, autophagy systematically declines in distinct tissues in C. elegans 10 . However, the machinery co-ordinately decreasing autophagy across tissues with age has been poorly studied. In this study, we identified a secreted microRNA, mir-83/miR-29, as an age-controlled messenger that dysregulates autophagy among tissues by directly inhibiting cup-5/MCOLN (Fig. 7i). Autophagy is generally a three-step process that involves sequestration of cytoplasmic components, degradation, and reuse of the degraded products 2 . A target of mir-83, cup-5, is a key regulator in these steps. cup-5 was first recognized for its essential role in the degradation step of autophagy 22 , and mir-83 indeed suppresses the degradation of autophagy substrates through cup-5. The mammalian homolog of cup-5, MCOLN, was also found to be an important regulator of TFEB, a master transcription factor controlling multiple steps of autophagy 26,27 . A recent report suggested hlh-30/TFEB is under the regulation of mir-83 37 . Our results showed that mir-83 inhibited GFP::LGG-1 expression through cup-5, implying that the interaction between cup-5/ MCOLN and hlh-30/TFEB could be well conserved in evolution and indicating that mir-83 also disrupts autophagy by affecting its initiation. Consistently, mir-83 reduced the numbers of APs and ALs through cup-5. Therefore, mir-83 impairs multiple steps of autophagy by inhibiting CUP-5.
With ageing, autophagy is systematically dysregulated in distinct tissues in C. elegans 10 , implying the existence of inter-tissue crosstalk regulating autophagy. Secreted microRNAs have been proposed to be potential messengers in inter-tissue signaling 38 . After using two independent transcriptional reporters, we and two other labs report that mir-83 is not transcribed in BWM 16,17 . qPCR of isolated BWM cells confirmed that the mir-83 promoter is inactive in BWM cells. However, mir-83 is detectable in isolated BWM cells, indicating that it is a secreted microRNA. We cannot exclude the possibility that a trace amount of mir-83 below the detection limit may still be transcribed in BWM, as suggested by a recent report 37 ; however, mir-83 expressed specifically in the intestine, but not gfp driven by the same intestine-specific promoter, was also detected in BWM. Together with the detection of mir-83 in EVs, this finding indicates that mir-83 is a secreted microRNA. Its translocation from the intestine into BWM impairs autophagy in the receiving tissue by suppressing CUP-5. Our results therefore reveal a mechanism of inter-tissue regulation of autophagy and identify mir-83, a secreted microRNA, as its modulator. In worms, several neurotransmitters and a Wnt ligand have been identified as signals that are secreted from the nervous system to control the unfolded protein response, another essential element of proteostasis, in other tissues 7,8 . Our results not only show that a secreted microRNA is an inter-tissue messenger controlling autophagy for proteostasis but also indicate that tissues other than nervous system (e.g., the intestine) broadcast signals for proteostasis throughout the body. Similarly, transcellular chaperone signaling from muscle to intestine and neurons is important in the response against proteotoxic stress 5,6 .
mir-83 was present in purified EVs and coelomocytes. Given this finding and the changes in lysosome morphology in coelomocytes, which were due to the interaction of mir-83 and cup-5, we speculate that the transportation of mir-83 from the intestine to BWM potentially occurs through EVs in pseudocoelom fluids, the body fluids of worms. Coelomocytes, which are active in endocytosis, may scavenge mir-83 in pseudocoelom fluids and thereby control its transportation across tissues. In addition, other microRNAs were also detected in the coelomocytes and EVs secreted by worms in our study, suggesting that inter-tissue transportation is not specific for mir-83. In mammals, secreted microRNAs are also abundant in body fluids and are linked to ageing 13 . Our results suggest that molecules in this class are not only ageing biomarkers but could also function as key messengers in regulating ageing progression (e.g., the disruption of autophagy) in distinct tissues. It will be intriguing to study in the future how microRNAs are transported across tissues.
The levels of pri-miR-83 are much lower than those of mature miR-83 in worms, potentially leading to the transport of many more mature miR-83 molecules than pri-miR-83 molecules. This could be why we did not detect pri-miR-83 in isolated BWM cells, coelomocytes, or EVs. Nevertheless, the precursors of mir-83 (priand pre-miR-83) could still be transported across tissues in addition to the mature form. Future studies on the detailed species of mir-83 undergoing inter-cellular transportation could help elucidate the mechanism of microRNA transport across tissues.
Evolution is likely to enrich alleles crucial to youth fitness while neglecting the detrimental effects of these alleles after reproduction due to declining natural selection pressure with ageing 39 . Consistent with previous RNA-Seq studies 40 , our study presented multiple lines of evidence that mir-83 is upregulated in aged worms without drug treatment, whereas Dzakah et al. observed decreases in mir-83 in FUdR-treated worms 37 . In untreated worms, the upregulation of the hsf-1-mir-83 axis with ageing could be a good example of the antagonistic pleiotropy hypothesis. mir-83 contributes to the development of the germline in larvae and to egg-laying in reproductive adults 17 , but its upregulation by HSF-1 in aged worms impairs autophagy in multiple tissues. hsf-1/HSF1 is a multi-faceted transcription factor for cell survival and growth. It is a master regulator that promotes autophagy and protein homeostasis, especially under stresses 41 , whereas its inactivation causes small cell/body sizes and infertility 42,43 . In the first 10 days of adulthood, C. elegans grow by 70% in length solely by increasing cell size 44 . The increase in HSF-1 in aged worms might be a response to this change. However, this increase induces detrimental pleiotropic side effects i mir-83 is upregulated in the aged intestine by increased HSF-1 and transported across tissues to disrupt autophagy through inhibiting CUP-5. See Discussion for details. Scale bar: 100 μm. Statistical significance was calculated by unpaired t-test in a, b or Two-way ANOVA in d, e. ns: non-significant, *p < 0.05, **p < 0.01, ***p < 0.001. See Supplementary Data 1 for lifespan statistics (c and f-h). Source data are provided as a Source Data file of mir-83 upregulation and subsequent impairment of autophagy and proteostasis. In young worms, hsf-1 upregulates autophagy following hormetic heat stress 41 . Our results therefore suggest that hsf-1 plays a dual role in autophagy, promoting autophagy in young worms upon heat stress and disrupting it in aged animals. Consistent with this speculation, heat shock upregulates multiple autophagy genes in young worms but not in aged animals 41 . Similar to our findings, inactivation of pha-4/FOXA, a key transcription factor in dietary restriction-induced longevity, in worms post reproduction actually extends lifespan 45 .
In addition to the increase in HSF-1 in aged worms, epigenetic factors may also contribute to the late activation of the hsf-1-mir-83 axis in aged worms. Trimethylation of Lys 27 on H3 (H3K27me3) has been reported to interfere with HSF-1 binding to genomic DNA and to suppress the transcription of HSF-1 targets 46 . The transcription of mir-83 could be under similar control, since ChIP-Seq results indicate strong binding of H3K27me3 and another histone modification anti-correlated with gene expression (H3K9me3) at the promoter of mir-83 in young adults and larvae 47 . Because histone modifications undergo dramatic changes with ageing and have remarkable variation across tissues 48,49 , the specific spatial and temporal pattern of histone modification at the mir-83 promoter may thus suppress the response of mir-83 to hsf-1 in young animals but facilitate the upregulation of mir-83 by hsf-1 in intestines of aged animals.
In summary, we have identified mir-83/miR-29 as an ageinginduced microRNA messenger from the intestine that systematically disrupts autophagy in multiple tissues by inhibiting cup-5/ MCOLN. mir-83 is well conserved in evolution, and its mammalian orthologue, miR-29, is upregulated in the brains and the serum of aged patients with hypertrophic cardiomyopathy 50,51 . In addition, although the 3′-UTR of MCOLN does not harbor miR-29 binding sites, miR-29 is predicted to target a few other autophagy genes, including ATG4D, the mammal orthologue of atg-4.2 52 . Therefore, it will be interesting to explore the function of miR-29 as an inter-tissue messenger in impaired autophagy in aged individuals. In addition, as miR-29 is reported to interfere with biological processes other than autophagy, it will also be interesting to explore the other functions of secreted mir-83 in worms.

Methods
C. elegans strains and culture. C. elegans strains used in this study are listed in Supplementary Data 3. C. elegans were grown with standard techniques at 20°C, unless otherwise noted 53 . Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). The transgenic lines of HZ589 and bpIs193 were kindly provided by Hong Zhang lab 54 , and the transgenic lines expressing dFP::lgg-1 (DLM3 and DLM11) were gifts from Miller lab 30 . To collect worms at different ages, worms were synchronized by collecting eggs and transferred to 60 mm plates (20 worms per plate). Worms were transferred to fresh plates every other day until day 10 to avoid the contamination from progenies, and subsequently monitored every other day for dead worms. Worms undergoing internal hatching, bursting vulva, crawling off the plates, or contamination as well as dead worms were removed from the plates.
Lifespan assays. Adult lifespan analysis was performed as reported 24 . For lifespan assays under chloroquine treatment, worms were incubated on NG plates containing 5 mM chloroquine diphosphate salt (Sigma-Aldrich) from eggs. During reproductive period, worms were handpicked onto fresh plates every other day to avoid progeny contamination. Worms not responding to prodding were scored as dead. Worms undergoing internal hatching, bursting vulva, crawling off the plates, or contamination were censored. Statistical analysis was performed with the Mantel-Cox Log Rank method.
To generate mir-83p::gfp::mir-83, 3110 bp of mir-83 promoter and 500 bp of mir-83 stem loop with flanking sequence were amplified from N2 genomic DNA and cloned into the 5′ and 3′ region of gfp in L3781, respectively. The Fire Lab C. elegans Vector Kit was a gift from Andrew Fire.
To generate ges-1p::mir-83, L3781 was first linearized by PCR to remove the gfp within. 2.2 kb of ges-1 promoter, 500 bp of mir-83 stem loop with flanking sequence were amplified from N2 genomic DNA and cloned into the linearized L3781.
To generate mir-83p::TagRFP-T and ges-1p::TagRFP-T, L3781 was first linearized by PCR to remove the gfp within. TagRFP-T was amplified from pDD284 (a gift from Bob Goldstein, Addgene # 66825) and cloned into the linearized L3781. 2.2 kb of ges-1 promoter and 3,110 bp of mir-83 promoter were amplified from N2 genomic DNA and cloned into the 5′ region of TagRFP-T respectively.
The cup-5(syb1028)III allele was generated by SunyBiotech using CRISPR/Cas9 technology. 3xFLAG and wrmScarlet (789 bp) were insert into the N terminal of the endogenous cup-5 gene. The strain was verified by DNA sequencing.

RNA interference.
RNAi experiments were based on a reported protocol 56 . In brief, worms were grown on HT115 expressing dsRNA against indicated genes from egg until corresponding time. HT115 expression dsRNA against luc2 (a real gene but irrelevant to worm biology) served as a control. The strains of HT115 Cell transfection and luciferase assay. HEK293T cells were seeded into a 24-well plate and transfected using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific) following the manufacturer's instruction one day post seeding. When testing the interaction between cel-mir-83-3p and 3′-UTRs of interest in HEK293T cells, 14 pmol of microRNA mimic, 50 ng of the 3′-UTR reporter plasmid, and 5 ng of Renilla luciferase plasmid were added into each well. MicroRNA mimics were ordered from GenePharma. Sequences of these mimics are listed in Supplementary Data 4.
Cells were collected and examined for luciferase activity 48 h post transfection by Dual-Luciferase® Reporter Assay System (Promega) and Synergy TM NEO (Bio Tek) as the manufacturer instructed. At least three independent experiments were performed.
Isolation of worm cells. Worms were lysed following a modified protocol from Murphy lab 57 . In brief, about 2000 synchronized worms of day 1 adulthood were washed by M9 to remove excess bacteria. Pelleted worms were incubated in 1 ml lysis buffer (200 mM DTT, 0.25% SDS, 20 mM HEPES, 3% sucrose) with rocking at 700 rpm for 7 min at 20°C. The worms were washed by 1 ml PBS for 6 times to remove the lysis buffer. The worm pellet was incubated in 400 μl Pronase (20 mg/ ml) with rocking at 1200 rpm for 10 min at 20°C. The proteinase was removed by washing worms with 1 ml PBS for 6 times.
To isolate coelomocytes,~100 μl digested worms were placed on a glass slide. On an Olympus IX73 microscope, coelomocytes, identified by unc-122p::DsRed, were picked using a broken microinjection needle (Harvard Apparatus) and transferred into 200 μl PBS with 0.1% BSA. Cells were further transferred with the needle into~4 μl PBS with 0.1% BSA. About 20 coelomocytes were collected for each sample.
To isolate BWM cells, worms were further digested in 400 μl 0.2% collagenase IV with rocking at 1200 rpm for about 40 min at 20°C. When most worm bodies were dissociated and lots of single BWM cells were identified by GFP::MYO-3, the collagenase was removed by centrifugation. Pelleted cells and worm debris were resuspended in 1 ml PBS with 0.1% BSA. BWM cells were picked as described above. About 30 BWM cells were collected for each sample.
Quantitative RT-PCR. qRT-PCR was performed following a previous report 24 . In brief, for mRNA, total RNA was prepared by RNeasy Mini kit (QIAGEN) from 100 worms. cDNA was subsequently generated by iScript TM Reverse Transcription Supermix for RT-qPCR (Bio-Rad). For microRNA, miRNeasy Mini kit (QIAGEN) and TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific) were used for total RNA and cDNA preparation, respectively. qRT-PCR was performed with Bestar® Sybr Green qPCR Master Mix (DBI Bioscience) or 2xNovoStart®SYBR qPCR SuperMix Plus (Novoprotein) on a QuantStudioTM 6 Flex Real-time PCR System (Applied Biosystems) or a CFX384 Touch TM Real-Time PCR Detection System (Bio-Rad). Four technical replicates were performed in each reaction. A combination of ama-1, act-1 and cdc-42 was used as reference for mRNA quantification, whereas Sno-RNA U18 as a reference for microRNA qRT-PCR. The results were from at least three biological replicates. qRT-PCR primers for mir-83 was designed as reported 58,59 . The sequences of all reported and self-designed primers are listed in Supplementary Data 4 24,60,61 .
For qRT-PCR of singled cells, isolated cells were first collected in~4 μl PBS with 0.1% BSA and lysed at room temperature for 5 min using 6 μl of lysis buffer (miScript Single Cell qPCR Kit, QIAGEN). Cell lysates were first incubated at 75°C for 5 min to inactivate DNase and then divided into two sets of samples respectively for mRNA and microRNA analysis.

RNA-Seq.
In each one of the three biological replicates,~300 N2 worms at indicated ages were handpicked into TRIzol (Invitrogen) for each time point. Total RNA isolation was performed using QIAGEN miRNeasy Mini (QIAGEN) following the manufacturer's instruction. Small RNA library preparation and sequencing were performed with Illumina sequencing technology by BGI, as reported 63 . The reads were first aligned to the C. elegans reference genome (WBcel215.67) using Tophat 64 . Aligned reads were then mapped against known microRNA sequences by miRDeep 65 . Differentially expressed microRNAs were figured out using the R package DESeq 66 .
Microscopy. For microscopy experiments, worms were mounted on 5% agar pads and anesthetized using 5 mM levamisole for imageing LGG-1 puncta, or 0.1% sodium azide for the rest assays. The images were captured using a Leica TCS SP8 or an Olympus BX53 microscope. Fluorescence intensity were measured by Adobe Photoshop. Background signal was subtracted as reported 24 .
Autophagy analysis. GFP:: LGG-1 puncta of posterior intestine cells were counted from about 5 slices with 1 μm step size, the Z-position was selected where intestinal nucleus could be seen clearly. GFP:: LGG-1 puncta of BWM cells were counted from 1 slice where the striation could be clearly seen. The puncta were counted using ComDet v.0.3.7 in ImageJ. For each genotype, at least 22 animals at corresponding stages from three independent experiments were scored.
Images of mCherry::GFP:: LGG-1 puncta were scored using the same method as examining GFP:: LGG-1 puncta. The mCherry::GFP::LGG-1 puncta of head neurons were counted from 1 slice where most neurons surrounding the posterior pharynx bulb are clearly seen. The number of APs was calculated as the GFPpositive puncta, and the number of ALs was calculated as the difference of the mCherry-positive puncta and the mCherry-GFP-positive puncta. For each genotype, at least 16 animals at corresponding stages from at least three independent experiments were scored.
To capture SQST-1::GFP images, head neurons and BWM cells were identified by morphology using differential interference contrast microscopy. The head neurons between two bulbs of pharynx and the BWM cells near the tail of worm were examined. For each genotype, at least 45 animals at corresponding stages from at least three independent experiments were scored.
Worms were transferred onto NGM plates containing 5 mM chloroquine and incubated for 4 h or 1 h to block the autophagy flux in the neurons or in other tissues.
Analysis of lysosome number and morphology. Lysosome numbers of anterior intestine cells were counted from about 5 slices with 1 μm step size, the Z-position was selected where intestinal nucleus could be clearly seen. For each genotype, at least 20 animals at corresponding stages from at least three independent experiments were scored.
The images of lysosome morphology in coelomocytes were captured from the pair near worm tail. The diameters of large LMP-1::GFP positive vacuole of WT worms and mir-83(-) mutants were quantified by LAS X. The diameters of large LMP-1::GFP positive vacuole of worms under RNAi treatment were quantified by ImageJ. For each genotype, at least 45 animals at corresponding stages from at three independent experiments were scored.
Analysis of polyQ strains. The numbers of intestinal and muscular polyQ aggregates in WT worms and mir-83(-) mutants were counted in individual worms at indicated days of adulthood. For each genotype, at least 45 animals at corresponding stages from at three independent experiments were scored.
Western blotting. Synchronized worms at corresponding stages of adulthood were harvested in M9. After three rounds of freezing and thaw, worms were lysed by adding 4x loading buffer. Proteins were separated by reducing SDS-PAGE and transferred to PVDF membranes. Membranes were then blotted with primary antibodies against GFP (1:200 dilution, Cat# sc-9996, Santa Cruz), FLAG (1:2000 dilution, Cat# F7425, Sigma-Aldrich), α-tubulin (1:2000 dilution, Cat# T5168, Sigma-Aldrich), or mCherry (1:2,000 dilution, Cat# NBP1-96752, Novus Biologicals), and secondary antibodies against Rabbit IgG (1:5000 dilution, Cat# G-21234, Thermo Fisher Scientific) or Mouse IgG (1:5000 dilution, Cat# G-21040, Thermo Fisher Scientific). To harvest worms at day 5 and day 7 of adulthood, each sample was prepared from about 120 synchronized L4 worms on six 60 mm plates. The worms were transferred to fresh plates every two days until reaching desired ages. Worms undergoing internal hatching and vulva bursting were discarded. Signals of western blotting were measured using Adobe Photoshop. Background signals were subtracted as reported 24 . All the uncropped and unprocessed images were shown in the Supplementary Fig. 8.
Isolation of extracellular vesicles. Synchronized worms were grown in liquid culture with the density of 3 worms per microliter and fed with 25 mg OP50 per milliliter until day 1 of adulthood 67 . Because a proportion of extracellular vesicles are secreted out of the worms [68][69][70] , the synchronized worms were soaked in M9 to collect EVs. In brief, for each sample, 300,000 worms were collected and washed by M9 to remove excess bacteria. Worms were subsequently cultured in 25 ml M9 at 20°C for 12 h with shaking at 180 rpm. Worms were sedimented by gravity, whereas the supernatant was collected and filtered by a 0.22 μm Millex-GP Syringe Filter Unit (Millipore). Extracellular vesicles (EVs) were isolated from the filtered medium using Total Exosome Isolation Reagent (Thermo Fisher Scientific) following the manufacturer's instruction. For gene expression, the isolated EVs were suspended in 700 μl QIAzol Lysis Reagent (QIAGEN). EVs were suspended in 600 μl PBS for other experiments.
Transmission electron microscopy of isolated extracelluar vesicles. In total 10-20 μl EVs suspended in PBS were deposited onto a carbon coated electron microscopy grid, stained with 2% uranyl acetate, and examined using a Tecnai G2 Spirit microscope (Thermo Fisher Scientific).
Nanoparticle tracking of isolated extracelluar vesicles. EVs in PBS were diluted 100-fold in PBS and subjected to nanoparticle tracking using a NanoSight NS300 (Malvern Instruments).
Statistical analysis. Results are presented as Mean ± SD unless otherwise noted. Statistical tests were performed as indicated using GraphPad Prism (GraphPad software). The statistics of Poisson regression was performed with R Core Team as reported 10 . Detailed statistics for lifespan assays and other experiments were respectively listed in Supplementary Data 1 and 2.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The source data underlying all figures are provided as a Source Data file. The associated data are also available at figshare