Prolonged quiescence delays somatic stem cell-like divisions in Caenorhabditis elegans and is controlled by insulin signaling

Aging Cell. 2020;19:e13085.   | 1 of 13 https://doi.org/10.1111/acel.13085 wileyonlinelibrary.com/journal/acel Received:28August2019 | Revised:30October2019 | Accepted:13November2019 DOI:10.1111/acel.13085 ORIGINAL ARTICLE Prolonged quiescence delays somatic stem cell-like divisions in Caenorhabditis elegans and is controlled by insulin signaling María Olmedo1 | Alejandro Mata-Cabana1 | María Jesús Rodríguez-Palero2,3 | Sabas García-Sánchez1 | Antonio Fernández-Yañez2,3 | Martha Merrow4 | Marta Artal-Sanz2,3 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2019 The Authors. Aging Cell published by the Anatomical Society and John Wiley & Sons Ltd. 1Departamento de Genética, Facultad de Biología, Universidad de Sevilla, Seville, Spain 2Andalusian Center for Developmental Biology,ConsejoSuperiordeInvestigaciones Científicas/Junta de Andalucía/Universidad PablodeOlavide,Seville,Spain 3Department of Molecular Biology and Biochemical Engineering, Universidad Pablo deOlavide,Seville,Spain 4InstituteofMedicalPsychology,Facultyof Medicine, LMU Munich, Munich, Germany Correspondence MaríaOlmedo,DepartamentodeGenética, Facultad de Biología, Avd. Reina Mercedes s/n, 41012, Sevilla, Spain. Email: mariaolme[email protected] Marta Artal-Sanz, Centro Andaluz de Biología del Desarrollo, Carretera de Utrera Km 1, 41013 Seville, Spain. Email: marts[email protected] Funding information SecretaríadeEstadodeInvestigación, DesarrolloeInnovación,Grant/Award Number:BFU2012-35509,BFU2016- 74949-P,RYC-2010-06167andRYC- 2014-15551;FP7Ideas:EuropeanResearch Council,Grant/AwardNumber:ERC-2011- StG-281691andPEOPLE-2013-IEF-627263 Abstract Cells can enter quiescence in adverse conditions and resume proliferation when the environment becomes favorable. Prolonged quiescence comes with a cost, reducing the subsequent speed and potential to return to proliferation. Here, we show that a similar process happens during Caenorhabditis elegans development, providing an in vivo model to study proliferative capacity after quiescence. Hatching under starvation provokes the arrest of blast cell divisions that normally take place during the first larval stage (L1). We have used a novel method to precisely quantify each stage of postembryonic development to analyze the consequences of prolonged L1 quiescence. We report that prolonged L1 quiescence delays the reactivation of blast cell divisions in C. elegans, leading to a delay in the initiation of postembryonic development.ThetranscriptionfactorDAF-16/FOXOisnecessaryforrapidrecovery after extended arrest, and this effect is independent from its role as a suppressor of cellproliferation.Instead,theactivationofDAF-16bydecreasedinsulinsignalingreduces the rate of L1 aging, increasing proliferative potential. We also show that yolk provisioning affects the proliferative potential after L1 arrest modulating the rate of L1 aging, providing a possible mechanistic link between insulin signaling and the maintenance of proliferative potential. Furthermore, variable yolk provisioning in embryos is one of the sources of interindividual variability in recovery after quiescence ofgeneticallyidenticalanimals.OurresultssupporttherelevanceofL1arrestasan in vivo model to study stem cell-like aging and the mechanisms for maintenance of proliferation potential after quiescence. KEYWORDS arrest, C. elegans, development, insulin signaling, proliferation, quiescence 2 of 13 | OLMEDO Et aL. 1 | INTRODUCTION Innaturalconditions,organismsareoftensubjectedtochangesin nutrientavailabilitythatmodulatesgrowthandproliferation.Inthe absence of growth-sustaining factors, cells can enter quiescence, a reversible,nonproliferativestate.Interestingly,cellularquiescence also occurs in normal development, with many cells spending most of their lifetime in this state. Under certain conditions, quiescent cells re-enter the cell cycle and resume proliferation (McCulloch & Till,2005;Siminovitch,McCulloch,&Till,1963).Despitethelackof proliferation, quiescence is not a homogeneous state. Cells move progressively into a deeper level of quiescence over time (Kwon et al., 2017). As a consequence, the duration of quiescence affects cell viability and proliferation potential. Understanding the molecular processes that regulate the maintenance of proliferation potential and the mechanisms mediating reactivation of proliferation after quiescence will contribute to comprehension of stem cell aging. Furthermore, failure of the programs that negatively regulate cell division upon withdrawal of growth signals would lead to uncontrolled proliferation, a hallmark of cancer cells (reviewed in (Hanahan & Weinberg, 2011)). InthenematodeC. elegans, newly hatched larvae enter a quiescent state when subjected to food deprivation. These L1 larvae (first larval stage) have 558 nuclei, 53 of which are in somatic blast cells that divide further to reach the 959 somatic cells of the adult (Sulston&Horvitz,1977).Inthepresenceoffood,blastcellsdivide over four stages of postembryonic development (L1-L4), and somatic cell divisions end with the transition to adulthood. When embryos hatch in the absence of food, they arrest at the L1 stage. This arrest involves quiescence of the blast cells that would normally divide during L1. The penetrance of the arrest depends on the cyclin-de- pendentkinaseinhibitorCKI-1/CIP/KIP/p27thatholdsthecellcycle atG1(Hong,Roy,&Ambros,1998).TheactivationofCKI-1intheabsenceoffoodisinpartmediatedbyDAF-16/FOXO,atranscription factor activated in conditions of low insulin signaling. A fraction of daf-16 mutant starved L1 larvae partially initiate divisions of postembryonicdevelopment(Baugh&Sternberg,2006),althoughthisonly occurs in the presence of small amounts of ethanol (Fukuyama, Kontani, Katada, & Rougvie, 2015). Arrested L1 larvae can survive several weeks without food and show increased resistance to stress (Baugh, 2013). Following feeding after prolonged L1 starvation, animals take longer to reach adulthood and show increased variability in the time it takes to reach this developmental stage (Jobson et al., 2015; Lee, Hendrix, Kim, Yoshimoto, & You, 2012). During L1 arrest, larvae undergo a process of aging, manifested by the accumulation of protein aggregates, increasedreactiveoxygenspecies(ROS)accumulation,andmitochondrial fragmentation. The addition of food to arrested L1 leads to the reversion of these aging phenotypes and the initiation of larval development. However, extended L1 arrest reduces the potential for recovery (Roux, Langhans, Huynh, & Kenyon, 2016). Sleep during L1 arrest counteracts aging phenotypes and increases survival rates (Wu, Masurat, Preis, & Bringmann, 2018). Thus, the nematode offers an exceptional tractable model system to study the mechanisms that impact cell arrest and proliferation in a multicellular organism. Here, we show that prolonged L1 arrest delays the reactivation of the cell divisions that mark the initiation of the postembryonic developmentalprogram.Oncetheprogramisinitiated,ournoveland highly quantitative assay shows that developmental timing is not affected. This contrasts with previous interpretations concerning the delay in reaching adulthood observed in animals after prolonged starvation. The specific effect of prolonged quiescence in recovery ismaintainedoverthelifespanofarrestedL1.Inourefforttounderstand the signaling pathways controlling quiescence, we investigated the role of insulin signaling in recovering from L1 arrest. Low insulin signaling in the insulin receptor mutant daf-2 leads to faster recovery and reduced aging, suggesting a shallower level of quiescence.ThetranscriptionfactorDAF-16isnecessaryfortheeffectof the daf-2mutationinrecovery.DAF-16functionintheprocessisnot related to a possible role as a suppressor of proliferation but to its role activating stress responses. Finally, we have discovered that maternal provisioning, that is also regulated by insulin signaling, reduces aging and improves proliferation after prolonged arrest. With this new approach to study re-initiation of cell proliferation after arrest, we have added an in vivo model that will contribute to tackle the mechanisms controlling maintenance of proliferation potential after arrest in multicellular organisms. 2 | RESULTS 2.1 | Time to recover from L1 arrest increases with prolonged starvation When L1 larvae are arrested for prolonged periods of time, animals take longer to reach adulthood once they are fed (Jobson et al., 2015; Lee et al., 2012). However, the duration of each of the four larval stages after extended starvation remains unexplored, hindering the analysis of the effects of starvation on the reactivation of the larval developmental program. To investigate in detail the timing of development after starvation, we used a quantitative and novelassaybasedontheuseofaluciferasereporterstrain(Olmedo, Geibel, Artal-Sanz, & Merrow, 2015). The enzyme luciferase, which is constitutively expressed in this strain, catalyzes the oxidation of the substrate luciferin, provided with the food. During the transitions between the larval stages, C. elegans ceases feeding and does not incorporate the substrate, leading to a rapid reduction in the luminescence signal. The profile of light emission can be used to precisely determine the timing of larval stages and molts. We measured the duration of all larval stages after different periods in L1 arrest (Figure 1a,b). As previously observed, the entry into adulthood was increasingly delayed with longer times in arrest (Figure 1c). When we analyzed the duration of each stage of development, we found that, of all larval and molt stages, only L1 is affected by the duration of the arrest (Figure S1). Delays in reaching adulthood resulted from an extended recovery time, defined as the time from the exposure | 3 of 13 OLMEDO Et aL. to food until the entry into the first molt (L1; Figure 1d). Prolonged starvation also resulted in increased variance of recovery time (F test p-value < .0001, when comparing day 2 and day 8) (Figure 1d). Starvation time had no effect on the amount of time between the first and the last molt (M1-M4) (Figure 1e). The differential effect of prolonged starvation on recovery time and development becomes evident when comparing the effect size for both processes (Figure 1f). Furthermore, we extended the time of starvation up to 27 days to analyze whether the effect on recovery and development was the same for any duration of the arrest. Despite a modest increase in the duration of development after very prolonged arrest, the effect is not comparable to the lengthened recovery time (Figure 1g,h). The maximum delay in development is about 30%, while the extensionoftherecoveryperiodreaches250%(Figure1i).Insummary, developmental timing is largely resilient to extended arrest, whereas recovery time increases with the duration of the arrest. 2.2 | Recovery time reflects proliferative potential after quiescence InordertounderstandthenatureoftheextendedL1stage,wefitted the recovery times after different starvation durations using linear regression. The stage L1, by definition, starts with hatching of the embryos and ends with the transition to L2. However, this definition does not consider the time needed to launch the postembryonic developmental program. As a consequence, L1 appears longer than the rest of larval stages. The relative molt/larva duration is FIGURE 1 Prolonged quiescence delays recovery time. (a) Past and current experimental designs to study the consequences of prolonged L1 arrest. (b) Representative plots of the duration of development for animals arrested as L1 for 2 days (left) and 8 days (right). Recovery time (L1) is defined as the time between the addition of food to starved L1 animals and the initiation of the first molt. Developmental timing is defined as the period between the beginning of the first molt and the end of the last molt, or initiation of adulthood (M1-M4). Total time to reach adulthood is the time between the addition of food and the end of the last molt (L1-M4). (c–e) Total time to adulthood (c), recovery time (d), and developmental time (e) for larvae arrested for 2–10 days. Average values per experiment are indicated with a black dot, and valuesfromsingleanimalsareindicatedwithacoloreddot.Weperformedone-wayANOVAontheaveragesof3biologicalreplicates (** p < .01). (f) Effect size of prolonged starvation in recovery and development. The data show the ratio of the average duration relative to day 2, for the three independent experiments shown in 1c-e. (g–h) Recovery (g) and developmental (h) time of L1 arrested up to 27 days. The plots show data from two independent replicates. (i) Effect size of prolonged starvation in recovery and development 246810 0 20 40 60 Days in L1 arrest Recovery time, L1 (hr) 246810 0 20 40 60 Days in L1 arrest Developmental time, M1-M4 (hr) 246810 0 20 40 60 80 100 Days in L1 arrest Time to adulthood, L1-M4 (hr) (a) (c) (d) Recovery time, L1 Developmental time, M1-M4 Time to adulthood, L1-M4 01224364860 0 500 1,000 1,500 Time on food (hr) Luminescence (a.u.) After 8 days of arrest 01224364 86 0 0 500 1,000 1,500 Time on food (hr) Luminescence (a.u.) (e) After 2 days of arrest ** n.s. ** (h)(g) 4681046810 0 1 2 3Recovery Development Days in L1 arrest Effect size (fold change to day 2) (i) (f) Feeding Adulthood ? L1arrest (variable duration) Precise quantification of development L1arrest (variable duration) (b) 0102030 0 20 40 60 Exp. 1 Exp. 2 Time in L1 arrest (days) Recovery time, L1 (hr) 0102030 0 20 40 60 Exp. 1 Exp. 2 Time in L1 arrest (days) Developmental t ime, M1-M4 (hr) 0102 03 0 0 1 2 3 Recovery, L1 Development, M1-M4 Time in L1 arrest (days) Effect size (fold change) 4 of 13 | OLMEDO Et aL. markedly smaller for M1/L1, compared to all the other larval stages (Olmedoetal.,2015).Fromthelinearregression,wecalculatedthe time (y value) needed to reach the first molt when time in arrest (x) equals 0. This provides an estimate of the bonafide duration of L1 for nonstarved larvae of ca. 10 hr (Figure S2 a). With the bonafide L1 duration determined from our data, the M1/L1 ratio reached the samevalueasalltheotherlarvalstages(FigureS2b).Ourhypothesis is that prolonged L1 arrest does not affect the duration of the bonafide L1, but it determines the delay between re-introduction of food and reactivation of postembryonic cell divisions. Alternatively, after extended arrest, reactivation of cell divisions could take place upon food addition without a delay and the complete L1 stage be prolonged (Figure S2 c). To differentiate between these two possibilities, we analyzed the relative timing of specific events during L1, namely, the divisions of seam and M cells. Seam cells undergo a single round of asymmetric self-renewing division at the beginning of each larval stage (L1-L4). The first of these divisions takes place about five hours after hatching. M blast cells present in the L1 larvae give rise to the complete mesodermal cell lineage. The first division of the M cell takes place midway through this larval stage (Sulston & Horvitz, 1977). We used fluorescent reporters to monitor the timingofseamcells(bothVandHlineages)andM-celldivisionafter one and four days of arrest. Specifically, we used Pscm::gfp reporters to follow the division of seam cells and a Phlh-8::gfp reporter to monitor M cells divisions. After a single day of arrest, seam cells of theV1-4andHlineagesstarttodivideafterfiveandeighthours, respectively, following exposure to food (Figure 2a–b). M cells divide after 9 hr (Figure 2c). After four days of arrest, the divisions were delayedonaverage5.75hrinthecaseofseamcellsand6.43hrinthe case of the M cells. These results support the first scenario, in which relative timing of divisions is similar whether animals are arrested only briefly or for extended periods. To compare the time of blast cell divisions within the same animals, we made a double-seam and M-cellreporterandmonitoredV-andM-celldivisionsuponfeeding of larvae arrested for one and four days (Figure 2d). We calculated the time needed for 50% of animals to achieve these cell divisions in the two conditions (Figure 2e). While the time between the addition offoodandVcelldivisionsisalmostdoubledinL1sarrestedforfour days, the time between V- and M-cell divisions remains constant (Figure 2f). We concluded that reactivation of blast cells is delayed after prolonged quiescence in C. elegans. This result supports our hypothesis that the bonafide L1 is constant over time in arrest, as are the rest of larval stages. As a consequence, recovery time can be used to quantify proliferative potential after quiescence. These findings support L1 arrest as an in vivo model to study proliferation reactivation potential after cell quiescence. We asked whether delayed cell division after prolonged arrest resulted from a failure to release the repression exerted by DAF- 16.Thenumberofanimalswithcytoplasmiclocalizationincreases rapidly in response to food independently of the duration of the arrest (Figure 2g), suggesting that exposure to food is not sufficient to resume the larval developmental program. The cyclin kinase inhibitorCKI-1mediatesdevelopmentallytimedcellcyclearrest,aswell as the arrest meditated by starvation, partially under the control of DAF-16(Baugh&Sternberg,2006).Wevisualizedcki-1 expression in the seam cells of arrested L1, which was higher after prolonged arrest (Figure 2h). When we fed L1 larvae after arrest for either one or four days, those subjected to shorter arrest reached basal cki-1 levels after less than 10 hr, while L1 arrested for four days needed more than 20 hr to reach similar levels (Figure 2i). These results imply that therapidcytoplasmicrelocationofDAF-16uponfeedingafterfour days of arrest is not sufficient to produce an immediate reduction of cki-1expression.OtherprocessesmustbehinderingCKI-1reduction in response to food and delaying proliferation after prolonged L1 quiescence. 2.3 | Insulin signaling modulates L1 aging and recovery One possibleexplanation for thedelayedreactivationof celldivisions is that the arrested larvae need to address the damage caused during prolonged arrest. Since quiescent L1 develop age-related phenotypes(Rouxetal.,2016),weinvestigatedtheprocessofL1aging and whether it influences reactivation of the postembryonic development.Insulin/IGF-1-likesignaling(IIS)isaprominentpathwayin the control of aging in the adult (Kenyon, 2010) and a modulator of the survival of arrested L1 animals (Muñoz & Riddle, 2003). Reduced IIS leads to activation of DAF-16/FOXO as it favors its translocationtothenucleus(Lin,Dorman,Rodan,&Kenyon,1997;Oggetal., 1997; Lin et al., 2001; Henderson & Johnson, 2001). Mutations in the sole C. elegans insulin receptor daf-2 increase survival during L1 arrest, while mutations in daf-16reduceit(Baugh&Sternberg,2006; Muñoz & Riddle, 2003). We measured recovery time in daf-16(mu86), daf-2(e1370), and daf-16(mu86);daf-2(e1370) double mutant. After one day of starvation, the recovery time of the daf-16 mutant was similar to that of the wild-type strain. However, after four days of starvation, the daf-16 mutant showed a remarkable delay in recovery. For some animals, recovery took more than 50 hr (Figure 3a, left). Despite the long recovery time, the developmental timing of daf-16 is only mildly delayed compared to wild-type (Figure 3a, right). The differential effect of the duration of quiescence on recovery and development is evident when calculating the fold change of the durationofbothprocesses(FigureS3a).ThisresultsuggeststhatDAF-16 is necessary to maintain the potential to reactivate cell proliferation. WetheninvestigatedwhetherincreasedDAF-16activation,as that found in daf-2(e1370) mutants, leads to enhanced reactivation of proliferation.DuringprolongedL1arrest,DAF-16relocalizestothe cytoplasm(Weinkove,Halstead,Gems,&Divecha,2006),butdaf-2 mutantsshowprolongedactivationofDAF-16(FigureS3b).When we measured recovery after one and four days of arrest, we did not observe a faster recovery of the daf-2 mutant (Figure S3 c). However, when we maintained L1 larvae in starvation for 14 days, the daf-2 animals recovered more rapidly than the wild-type (Figure 3b left and Figure S3 d). This result is remarkable since low insulin signaling is generally related to proliferative defects (Michaelson, Korta, | 5 of 13 OLMEDO Et aL. Capua, & Hubbard, 2010). The developmental time (M1-M4) of daf- 2 mutants is increased compared to wild-type animals (Figure 3b, right),aspreviouslydescribed(Olmedoetal.,2015;Ruaud,Katic,& Bessereau,2011).Importantly,thedurationofarrestdidnothavea strong impact on growth rate, neither in wild-type nor in daf-2 mutants (Figure 3b, right). As expected, daf-16 is epistatic to daf-2 for recovery from L1 arrest (Figure 3c), suggesting that the protective functionoflowinsulinsignalingrequiresDAF-16activity. The delayed recovery of daf-16 mutants seems surprising, in lightoftheroleofDAF-16asarepressorofproliferation(Baugh& Sternberg,2006;Kaplanetal.,2015).Weinvestigatedwhetherthe delayed recovery of daf-16 was caused by a defect to resume proliferation of blast cells by following seam cell division during recovery. After one day of arrest, seam cell divisions occurred at the same rate in the wild-type and the daf-16 mutant. After four days of arrest, the delay in seam cell proliferation was much larger in the daf-16 mutant (Figure 3d). Even after 45 hr of recovery, many daf-16 animals continued arrested, with complete absence of blast cell divisions (Figure S3 e). After seam cell division, the time to M-cell division was unaltered in the daf-16 mutant, even after four days of arrest (Figure S3 f). We then investigated how activation of the transcription factor DAF-16enhancesrecovery.Intheabsenceoffood,DAF-16mediates both quiescence of blast cells through activation of the cyclin-de- pendent kinase inhibitor CKI-1 and the activation of stress resistance and maintenance pathways (Baugh, 2013; Baugh & Sternberg, 2006). Interestingly, the effect of DAF-16 in cell quiescence has been separated from its effect in L1 survival. DBL-1, a TGF-β ligand, is necessary for reactivation of divisions during arrest in the daf-16 mutant. A double-mutant daf-16;dbl-1, which does not undergo cell division during arrest, has the same L1 survival as the daf-16 mutant (Kaplan et al., 2015). This indicates that daf-16 mortality during FIGURE 2 Recovery time reflects reactivation of the developmental program. (a–c) Timing of seam cells (a,b) and M cell (c) division upon addition of food to L1 larvae arrested for one day or for four days. The plots show data from 3–4 biological replicates, and values represent the percentage of animals showing division. Curves represent the fit assuming a cumulative Gaussian distribution. The dashed linesindicatethevalueof50%ofanimalswithdividedcells.(d)TimingofdivisionofVseamcellsandMcellsuponadditionoffoodtoL1 larvae arrested for one day or for four days. We analyzed a double reporter to assay both divisions in the same animals, performing three biological replicates. (e) Representative plot showing the calculated intervals between cell divisions. (f) Prolonged arrest delays the division ofVseamcellsbutnotthetimingbetweenV-andM-celldivisions.(g)PercentageofanimalswithcytoplasmiclocalizationofDAF-16inthe first five hours upon the addition of food after one or four day of L1 arrest. The plot shows the mean (±SD) of four biological replicates. (h) Representative images of cki-1 activation after one and four days of L1 arrest. (i) Activation of cki-1 during recovery after one or four days of arrest. The plot shows the average (±SD) of three independent replicates (a) 04812162024 0 25 50 75 100 day 4 day 1 Time on food (hr) % divided H seam 04812162024 0 25 50 75 100 day 1 day 4 Time on food (hr) % divided M cells (b) (c) 04812 16 20 24 0 25 50 75 100 day 1 day 4 Time on food (hr) % divided V seam (d) (e) 04812162024 0 25 50 75 100 M V Time on food (hr) % divided cells 04812162024 0 25 50 75 100 M V Time on food (hr) % divided cells 048121620 0 25 50 75 100 Start to V V to M Time on food (hr) % divided cells 1414 0 5 10 15 20 Start to V-cell divisions V- to M-cell divisions ** n.s. Days in L1 arrest Time (h) day 1day 4 (f) (g) (h) 010203040 0 25 50 75 100 day 1 day 4 Time on food (hr) induced cki-1 (fraction of animals ) 012345 0 25 50 75 100 day 1 day 4 Time on food (hr) % L1 with cytoplasmic DAF-16 Pcki-1::NLS-GFP day 1day 4 (i) 6 of 13 | OLMEDO Et aL. FIGURE 3 Low insulin signaling during quiescence ameliorates proliferative potential and attenuates L1 aging. (a) Recovery and developmental time for the wild-type strain and the daf-16(mu86) mutant after one and four days of arrest. Average values per experiment are indicated with a black dot, and values from single animals are indicated with a colored dot. We performed t test on the averages of 3 biological replicates (* p < .05). (b) Recovery and developmental timing for the wild-type strain and the daf-2(e1370) mutant after 8 and 14 days of arrest. We performed t test on the averages of 3 biological replicates (* p < .05, *** p < .001). (c) Recovery and developmental timing for the wild-type strain, the daf-16(mu86) mutant and the double-mutant daf-2(e1370);daf-16(mu86) mutant after 1 and 4 days of arrest. (d) Seam cell division in wild-type and daf-16larvaeduringrecoveryafteroneorfourdaysofarrest.Valuesrepresentthepercentage of animals showing division in three biological replicates. (e) Recovery and developmental timing for animals categorized as having low orhighDHEstainingafter8daysofarrest.Weperformedone-wayANOVAontheaveragesof4biologicalreplicates(**p < .01). (f) QuantificationofROSaccumulationinthewild-typestrain,daf-2 (e1370), and daf-16(mu86). Plots show mean (±SD) of 3–4 biological replicates.Fordays1,4,and8,weperformedone-wayANOVAontheaveragesofbiologicalreplicates,followedbyDunnett´sMultiple Comparison test to detect significant differences between the mutants and the wild-type. For day 14, we performed t test (** p < .01). (g) Quantification of amyloids in the wild-type strain, daf-2(e1370), and daf-16(mu86). Plots show mean (±SD) of four biological replicates. Statistics were performed as in (g) (* p < .05 and *** p < .001). (h) Recovery and development for larvae arrested during eight days in the presenceofammoniumchlorideandacridineorange(AO).(i)Recoveryanddevelopmentofwild-typeanddaf-16 mutant L1 arrested during fourdaysinthepresenceof(AO) 0 30 60 90 day 1day 4 Days in L1 arrest Recovery time, L1 (hr) 0 30 60 90 day 1day 4 wt daf-16 Days in L1 arrest Developmental time M1-M4 (hr) 0 30 60 90 day 8 day 14 Days in L1 arrest Recovery time, L1 (hr) 0 30 60 90 wt daf-2 day 8day 14 Days in L1 arrest Developmental time M1-M4 (hr) (a) (b) n.s. *n.s. *n.s. ******* (c) 0 30 60 90 day 1day 4 Days in L1 arrest Recovery time, L1 (hr) n.s. 0 30 60 90 day 1day 4 wt daf-16 daf-2;daf-16 Days in L1 arrest Developmental time, M1-M4 (hr) (d) 0481216202428323640 0 25 50 75 100 wt day 1wt day 4 daf-16 day 1daf-16 day 4 Time on food (hr) % divided seam cells Low High 0 40 80 120 ROS Recovery time, L1 (hr) Low High 0 20 40 60 80 ROS Developmental time, M1-M4 (hr) 0 500 1000 1500 wt daf-2 daf-16 day 1day 4day 8day 14 ** n.s. n.s. n.s. DHE staining (pixel intensity) 0 200 400 600wt daf-2 daf-16 day 1day 4day 8day 14 *** n.s * n.s. n.s. NIAD-4 staining (pixel intensity) ** (f) (g)(e) 0 20 40 60 day 1day 8 Recovery time, L1 (hr) 0 20 40 60 day 1day 8 control Ammonium chloride Acridine orange Developmental time, M1-M4 (hr) * n.s. (h) (i) 0 25 50 75 100 125 –+ –+ –+–+ 1414 wt daf-16 AO Days in L1 arrest Recovery time, L1 (hr) 0 25 50 75 100 125 wt daf-16 14 14 –+ –+ –+ –+ Developmental time, M1-M4 (hr) | 7 of 13 OLMEDO Et aL. arrest is not a consequence of inappropriate divisions. We looked at recovery of the double-mutant daf-16;dbl-1 and observed a delay in recovery greater than that of the single daf-16 mutant (Figure S3 g,h), indicating that dbl-1 is not epistatic to daf-16 for recovery from arrest. This result already suggests that, as it happens with survival, the delayed recovery of daf-16 is not a consequence of inappropriate cell divisions. When we analyzed cki-1 activation in the daf-16 mutant, we could not observe differences with the wild-type strain (FigureS3i).TheseresultspointtoaroleofDAF-16incellularmaintenance during L1 arrest that is independent of its role in the control of cell proliferation. 2.4 | Markers of aging correlate with increased recovery time Inlightofthepreviousresults,weaskedwhetherrecoverytime,asa proxy for reactivation of quiescent cells, correlated with the process of L1 aging. This would support the idea that recovery time reflects the level of activation of stress resistance and maintenance pathways during arrest. We investigated if recovery time was related to theaccumulationofROS,oneofthemarkersofL1aging(Rouxetal., 2016).WestainedL1arrestedlarvaewithDihydroethidium(DHE),a compoundthatfluorescesinthepresenceofROS.Weobservedan increasing variability in DHE signal over time of arrest (Figure S4 a). After 8 days of L1 arrest, we could visually establish two categories of larvae with high and low fluorescence in the head, which yielded significant differences when we quantified the actual fluorescent signal (Figure S4 b). We selected animals based on these two categories and analyzed their recovery, demonstrating that animals with higher DHE signals recovered significantly more slowly than those with lower signals (Figure 3e, left) but did not show developmental delay (Figure 3e, right). This result suggests that the capacity to recover depends on the accumulation of age-related phenotypes (Rouxetal.,2016). We suspected that insulin signaling modulates the rate of accumulation of aging markers during L1 quiescence. The increase in the DHE signal was less pronounced in the daf-2 mutant, showing a significant reduction relative to the wild-type animals after 14 days of arrest (Figure 3f). We monitored DHE signals in the daf-16 mutant up to day 8 of arrest, due to the increased mortality of these animals (Kaplan et al., 2015). At that time, the DHE signal was similar tothatofwild-type(Figure3f),suggestingthateitherROSaccumulation does not account for the delay in reactivation of proliferation ofthismutantorDHEstainingdoesnotreflectallROSpresentin the nematode. We also checked DHE signal at day 11 of starvation and obtained similar results. At day 11, many daf-16 animals were dead, and they show very high DHE staining (Figure S4 c). However, this seemed to be a consequence, rather than the cause of death, since we do not observe a continuous transition in staining from live to dead L1s. The DHE signal in live worms is similar between wildtype and daf-16mutants(FigureS4c).Next,wecheckedwhether the delayed recovery of daf-16 mutants is due to a pronounced accumulation of other markers of aging. We analyzed the formation ofamyloidsusingthedyeNIAD-4(Habchietal.,2016).Inthiscase, the differences in staining in the daf-2 mutant after 14 days of arrest are even more pronounced. The mutant daf-16 showed significant differencesinNIAD-4stainingafter8daysofarrest(Figure3g),suggesting that increased protein aggregation could delay recovery. AnotherpossibleconnectionbetweenDAF-16functionandL1 recoveryislysosomalfunction.DAF-16promoteslysosomeacidification in C. elegans (Baxi, Ghavidel, Waddell, Harkness, & Carvalho, 2017). Furthermore, impaired lysosomal function leads to delayed proliferation in rat embryonic fibroblast (Fujimaki et al., 2019). We tested whether lysosomal function had an effect on recovery time by applying compounds that alkalize lysosomes, namely ammonium chloride (NH4Cl) and acridine orange (AO) (Artal-Sanz, Samara, Syntichaki, & Tavernarakis, 2006). After 8 days of treatment, L1 thatwerearrestedinthepresenceofacridineorange(AO)showed a slower recovery than control animals. The treatment did not affect the subsequent developmental timing (Figure 3h and Figure S4 d). This experiment is of special relevance, as animals are exposed to AOonlyduringL1arrest.Thisresultimpliesthatrecoverytimecan reflect the conditions during L1 arrest, independently of the conditions during the actual recovery period, which in this case are the same.WhenweanalyzedtheeffectofAOinthedaf-16 mutant arrested for four days, we observed an additional delay in recovery (Figure 3i), indicating that lysosomal function has a wider role in recovery from starvation. Altogether, these results suggest that the physiological decline, similar to the aging process, that takes place specifically during L1 arrest leads to slower recovery once animals are fed. 2.5 | Reduced maternal provisioning delays recovery in the progeny daf-2 mutants produce longer, better-provisioned embryos which, after prolonged starvation, produce more progeny than wild-type (Hibshman,Hung,&Baugh,2016).Reducedinsulinsignalinginthis mutant increases maternal yolk provisioning to embryos and mitigates the reproductive abnormalities consequence of L1 starvation (Jordan et al., 2019). We investigated whether modulation of maternal provisioning per seaffectedrecoveryoftheprogeny.Vitellogenin/yolkproteinsuptake by the oocyte occurs through RME-2 mediated endocytosis (Grant & Hirsh, 1999). Although yolk is not necessary for efficient reproduction in C. elegans(VanRompayetal.,2015),efficientyolkproteintransport and storage contribute to L1 survival (Chotard, Skorobogata, Sylvain, Shrivastava, & Rocheleau, 2010). We tested the efficiency of the rme- 2knockdownbymeasuringVIT-2::GFPinembryosofanimalstreated with rme-2RNAi(FigureS5a,b).WeobtainedL1larvaeaftertreatment of the mothers with rme-2RNAiandwithcontrolbacteriaandmaintained them in arrest for 8 days. Larvae from embryos with reduced vitellogenin showed longer recovery times than larvae from mothers treated with control RNAi (Figure 4a, left). Despite the pronounced 8 of 13 | OLMEDO Et aL. delay in recovery, developmental timing was similar between conditions(Figure4a,right).Ourresults,therefore,confirmthatyolkprovisioning is important during L1 starvation and provides a mechanistic link between insulin signaling and recovery from L1 arrest. Interestingly,differentialembryoprovisioningofvitellogeninor yolk proteins, as a consequence of maternal age is one factor introducing phenotypic variability in development after extended arrest. Embryos from older mothers contain more yolk proteins and develop faster (Perez, Francesconi, Hidalgo-Carcedo, & Lehner, 2017). A recurrent observation for both recovery time and aging markers is an increase in the coefficient of variation as the values increase. This means that processes that increase recovery time show an important interindividual variability. The causes of such variability in genetically identical organisms reared in the same environmental conditions are only starting to be elucidated. We investigated whether naturally occurring differences in maternal provisioning could explain the variability we observed in recovery time and aging markers. We obtained embryos from gravid adults in their first, second, and third day of egg laying, maintained them in L1 arrest for 8 days, and measured its recovery time. We found that larvae from older mothers recovered significantly faster than those of younger mothers (Figure 4b). Although larvae from day 3 mothers were on average similar to those from day 2 mothers, this population contained the fastest animals in terms of recovery. When analyzing the 50% fastest animals in each condition, average recovery was progressivelyfaster,withrecoverytimesof22.52,19.65,and17.56hr respectively (Figure S5 c). This effect was more prominent when we analyzed recovery of the very first embryos produced by the mothers (day 0.5, Figure S5 d). Maternal age does not have an effect on developmental timing after extended arrest (Figure 4c and Figure S5 d). We also analyzed whether embryos from older mothers were protected against the accumulation of L1 aging markers. After 8 days of arrest, larvae from day 2–3 mothers showed reduced DHE accumulation compared to larvae from day 1 mothers (Figure 4c). WefoundnodifferencesinNIAD-4staining(FigureS5e).These results support a role for maternal age in preventing L1 aging, leading to maintenance of proliferation potential. To explore a possible link between insulin signaling and yolk provisioning, we analyzed thelocalizationofDAF-16duringL1arrest,inlarvaetreatedwith rme-2RNAi(FigureS5f).Larvaewithreducedyolkprovisioning hadsimilarlevelsofnuclearDAF-16thanthosetreatedwithcontrol RNAi. This result indicates that reduced L1 aging of larvae from older mothers is unlikely to be due to an increased activation ofDAF-16. 3 | DISCUSSION The balance between cell quiescence and proliferation is crucial for maintenance of stem cell pools, which hold the potential to maintaintissuehomeostasisorreplacedeadcellsafterinjury.On the one hand, stimulation of proliferation of quiescent cells provokestheexhaustion ofstem cells (Li & Clevers,2010).On the other, the degenerative processes in stem cells and the systemic cues that regulate their activity have been connected to the agedependent decline in regenerative potential of tissues (Ahmed, FIGURE 4 Maternal provisioning modulates recovery from L1 arrest. (a) Recovery time after 8 days of arrest for L1 larvae from mothers fed either control or rme-2RNAibacteria.Weperformed one-wayANOVAfollowedbyBonferronitestingontheaveragesof 3 biological replicates (* p < .05). (b) Recovery and developmental time after 8 days of arrest of L1 larvae from day 1–3 progeny. We performedone-wayANOVAfollowedbyBonferronitestingonthe averages of 8 biological replicates (** p<.01).(c)ROSaccumulation after 1 or 8 days of arrest for day 1–3 progeny. We performed onewayANOVAfollowedbyBonferronitestingontheaveragesof3–4 biological replicates (* p < .05) control rme-2 0 25 50 75 100 RNAi Developmental time, M1-M4 (hr) 123 0 25 50 75 Maternal age (days) Recovery time, L1 (hr) ** n.s. ** 123 0 25 50 75 Maternal age (days) Developmental time, M1-M4 (hr) n.s n.s. n.s. (c) 123 123 –250 0 250 500 750 1,000 1,250 1,500 Maternal age (days) DHE staining (pixel intensity) * * n.s n.s n.s. n.s day 1day 8 controlrme-2 0 25 50 75 100 RNAi Recovery time, L1 (hr) * (b) (a) n.s. | 9 of 13 OLMEDO Et aL. Sheng, Wasnik, Baylink, & Lau, 2017). Proliferative potential ofstemcellsisreducedovertimeinquiescence.Now,wehave shown that, also in C. elegans, prolonged arrest delays blast cell divisions. This finding parallels that in cultured cells, where delayed reactivation of divisions after prolonged starvation is proposed to be due to quiescence deepening (Kwon et al., 2017). This suggests that blast cells in L1 larvae progress to deeper levels of quiescence throughoutL1arrest.InC. elegans, we have monitored this process over 27 days of arrest, showing that the selective effect of prolonged quiescence on recovery extends over the survival of arrested L1. These results support the relevance of L1 arrest upon starvation as a model to study proliferative potential after quiescence in vivo. Roux et al. proposed that quiescent L1 suffer a physiological declinesimilartotheagingoftheadult(Rouxetal.,2016).Most aging markers accumulated by L1 larvae during arrest, including ROS,werereversedbyfeeding,andthecompetenceoftheanimal to clear those signs of aging determined the capacity to recover from L1 arrest. Here, we show that we can measure the proliferative potential of quiescent L1, not only in terms of recovery per se, but through precise quantitation of recovery time. Through determination of the recovery time of daf-2 mutants, we found that low insulin signaling plays a maintenance role during quiescence, as daf-2 mutants recover faster than the wild-type after prolonged arrest. This finding seems surprising considering the role of this pathway in proliferation. However, we have observed reduced ROS accumulation and protein aggregation in the daf-2 mutant. Agedstemcellsaccumulatereactiveoxygenspecies(ROS),DNA damage, aggregated proteins, and they show mitochondrial misfunction(reviewedin(Oh,Lee,&Wagers,2014)).Interestingly,a recent work has determined that lysosomal gene expression increases as quiescence deepens. Lysosomal function favors a shallowerlevelofquiescencebyreducingROSaccumulation.Ascore for gene expression during the deepening of quiescence parallels cellular senescence and aging (Fujimaki et al., 2019). This recent findingandourownresultssuggestthatROSaccumulationfavors deeper quiescence. We have shown that lysosome malfunction also increases recovery time after prolonged quiescence in C. elegans, which represents an important parallelism to the recent findings in rat embryonic fibroblasts (Fujimaki et al., 2019). Since DAF-16driveslysosomalfunctioninC. elegans (Baxi et al., 2017), this could explain the observation that daf-2(e1370) mutants, whichpresentaconstitutiveDAF-16activation,showlowerlevels ofROSandfasterrecovery.AlthoughFoxOdeficienthematopoieticstemcellspresentincreasedROS(Tothovaetal.,2007),we donotobservedhigherROSaccumulationinadaf-16mutant.It is possible that other factors, as protein aggregation, are limiting survivalofthismutantbeforetheincreasedaccumulationofROS becomesevident.OurresultsindicatethatactivationofDAF-16 confers an advantage for recovery, possibly by reducing the rate of L1 aging. Despite the proliferative role of insulin signaling, its repression could be beneficial for maintenance of cells by conferring a shallower level of quiescence, allowing faster recovery upon feeding.Itisalsoimportanttonoticethatinsulinsignalinghasalready been connected to the progression of L1 aging through the regulation of sleep. However, daf-16 mutation alone does not impair L1 sleep (Wu et al., 2018), and therefore, sleep loss could not explain the defective recovery of this mutant. The delayed L1 recovery after prolonged starvation involves a delay in reactivation of blast cell divisions. This process is, in principle, controlled by insulin signaling, as daf-16 mutants are arrest defective. However, it is important to recall that the defective-arrest phenotype of daf-16 requires the presence of ethanol (Fukuyamaetal.,2015).Inourexperiments,wedonotincludeethanol in the buffer used for the starvation assay, indicating that the phenotypes we observe are not related to reactivation of cell divisions during arrest. Actually, we have found that daf-16 mutants show activation of the cell cycle inhibitor cki-1/CIP/KIP/p27,unlike what is observed when ethanol is present in the starvation media (Baugh & Sternberg, 2006). This means that, as previously suggested by other authors, additional pathways could contribute to the regulation of cki-1/CIP/KIP/p27duringL1arrest(Baugh,2013; Kniazeva, Euler, & Han, 2008). Interestingly, protein aggregates usually sequester proteins involved in cell cycle control (Lutz & Peng, 2018; Zhou et al., 2004). The increased protein aggregation that we observed in daf-16 could lead to a reduction of cell cycle regulators and delay its progression. Onefeatureofdaf-2 animals is that they produced longer, better-provisioned embryos than the wild-type strain. We modulated maternalprovisioningbyperformingRNAiagainstthevitellogenin/ yolk transporter rme-2 and by controlling maternal age of the L1 larvae subjected to arrest. These interventions were sufficient to modulate L1 aging and affect recovery time. This observation provides a possible mechanistic link between reduced insulin signaling and the regulationofL1aging.TheeffectofmaternalageinROSaccumulation and in recovery also explains our observation of increased variability upon prolonged starvation in the experiments where we did not control for maternal age. The heterogeneity in recovery of larvae with different maternal provisioning is reminiscent of the effect of cell growth and divisions history in subsequent quiescence exit in rat embryonic fibroblasts (Wang et al., 2017). Although we have not tested this relationship specifically, it is tempting to speculate that maternal dietary conditions and growth rate will affect recovery of the progeny. 4 | EXPERIMENTAL PROCEDURES 4.1 | Culture conditions and strains We cultured stock animals according to standard methods (Brenner, 1974), maintaining them at 20°C on nematode growth medium (NGM)withalawn ofEscherichia coliOP50-1.Theonlyexception is in Figure 1a–d, where animals were maintained at 18°C previous to the experiment. A detailed description of the strains used in this work is presented in the Supporting information.