Abstract
Many tissue-specific stem cells maintain the ability to produce multiple cell types during long periods of non-division, or quiescence. FOXO transcription factors promote quiescence and stem cell maintenance, but the mechanisms by which FOXO proteins promote multipotency during quiescence are still emerging. The single FOXO ortholog in C. elegans, daf-16, promotes entry into a quiescent and stress-resistant larval stage called dauer in response to adverse environmental cues. During dauer, stem and progenitor cells maintain or re-establish multipotency to allow normal development to resume after dauer. We find that during dauer, daf-16/FOXO prevents epidermal stem cells (seam cells) from prematurely adopting differentiated, adult characteristics. In particular, dauer larvae that lack daf-16 misexpress collagens that are normally adult-enriched. Using col-19p::gfp as an adult cell fate marker, we find that all major daf-16 isoforms contribute to opposing col-19p::gfp expression during dauer. By contrast, daf-16(0) larvae that undergo non-dauer development do not misexpress col-19p::gfp. Adult cell fate and the timing of col-19p::gfp expression are regulated by the heterochronic gene network, including lin-41 and lin-29. lin-41 encodes an RNA-binding protein orthologous to LIN41/TRIM71 in mammals, and lin-29 encodes a conserved zinc finger transcription factor. In non-dauer development lin-41 opposes adult cell fate by inhibiting the translation of lin-29, which directly activates col-19 transcription and promotes adult cell fate. We find that during dauer, lin-41 blocks col-19p::gfp expression, but surprisingly, lin-29 is not required in this context. Additionally, daf-16 promotes the expression of lin-41 in dauer larvae. The col-19p::gfp misexpression phenotype observed in dauer larvae with reduced daf-16 requires the downregulation of lin-41, but does not require lin-29. Taken together, this work demonstrates a novel role for daf-16/FOXO as a heterochronic gene that promotes expression of lin-41/TRIM71 to contribute to multipotent cell fate in a quiescent stem cell model.
Introduction
Tissue-specific stem cells divide as needed to replenish cells lost due to injury or normal wear and tear. Many stem cell types retain the capacity to produce multiple cell types during lengthy periods of quiescence, or non-division, and increased cell division can lead to compromised multipotency and stem cell maintenance (Orford and Scadden 2008). However, the connections between quiescence and multipotency are incompletely understood. Caenorhabditis elegans (C. elegans) development can be interrupted by a quiescent and stress-resistant stage called dauer in response to adverse environmental conditions (Cassada and Russell 1975). Stem cell-like progenitor cells maintain or re-establish multipotency during dauer, a situation analogous to mammalian stem cells (Liu and Ambros 1991; Euling and Ambros 1996; Karp and Greenwald 2013).
Dauer formation is regulated by three major signaling pathways (Fielenbach and Antebi 2008; Baugh and Hu 2020). One of these pathways is insulin/IGF signaling, where favorable environmental cues lead to the production of multiple insulin-like peptides in sensory neurons. (Pierce et al. 2001; Cornils et al. 2011; Murphy and Hu 2013; Ritter et al. 2013; Fernandes-de-Abreu et al. 2014; Zheng et al. 2018). These signals are released and then received in multiple tissues where insulin signaling blocks the activity of the downstream DAF-16/FOXO transcription factor (Kimura et al. 1997; Ogg et al. 1997; Lin et al. 1997; Pierce et al. 2001). In adverse environments, DAF-16 is active and regulates the expression of genes that promote dauer formation (Lin et al. 2001; Lee et al. 2001).
The role of the DAF-16 transcription factor in promoting dauer formation is analogous to the role of the FOXO proteins in promoting quiescence in mammalian stem cells (Tothova and Gilliland 2007). In addition, DAF-16/FOXO is required for stem cell maintenance and the ability of stem/progenitor cells to differentiate into the correct cell types in both systems (Narbonne and Roy 2006; Baugh and Sternberg 2006; Tothova et al. 2007; Zhang et al. 2011; Murphy and Hu 2013; Karp and Greenwald 2013; Liang and Ghaffari 2018). For example, in C. elegans dauer larvae, daf-16 is required in a set of multipotent vulval precursor cells to block EGFR/Ras and LIN-12/Notch signaling from prematurely specifying vulval cell fates (Karp and Greenwald 2013).
Another group of progenitor cells that must remain quiescent and multipotent during dauer is the lateral hypodermal seam cells that make up part of the worm skin. Seam cells undergo self-renewing divisions at each larval stage and then terminally differentiate at adulthood (Sulston and Horvitz 1977). Larval vs. adult seam cell fate is regulated by a network of heterochronic genes (Ambros and Horvitz 1984). In general, heterochronic transcription factors and RNA-binding proteins that specify early cell fates are expressed early in development. These early cell fate-promoting factors are then downregulated by microRNAs in order to allow progression to later cell fates (Rougvie and Moss 2013). This pathway culminates with the expression of the LIN-29 transcription factor. LIN-29 is the most downstream regulator of adult cell fate and directly activates the expression of adult-specific collagens such as col-19 (Ambros 1989; Liu et al. 1995; Rougvie and Ambros 1995; Abete-Luzi et al. 2020). LIN-29 protein is not expressed in the hypodermis until late in larval development due to the combined action of the early-promoting heterochronic genes lin-41 and hbl-1 (Rougvie and Ambros 1995; Slack et al. 2000; Azzi et al. 2020). lin-41 encodes an RNA-binding protein that binds to the lin-29 mRNA and blocks translation (Slack et al. 2000; Aeschimann et al. 2017). lin-29 is also required for adult cell fate in post-dauer animals. However, many genes that act earlier in the heterochronic pathway are dispensable for post-dauer development, suggesting that the regulation of adult cell fate differs in continuous and dauer life histories (Liu and Ambros 1991).
Here, we examine the role of daf-16 in maintaining multipotent seam cell fate during dauer. We find that daf-16 is required to block the expression of multiple adult-enriched collagens during dauer, including col-19, defining a novel role for daf-16 as a dauer-specific heterochronic gene. Using the adult cell fate marker col-19p::gfp as a readout, we find that daf-16 acts via lin-41 to regulate adult cell fate. Surprisingly, lin-29 plays at most a minor role in the regulation of col-19p::gfp during dauer. mRNA-seq experiments identified 3603 genes that are regulated by daf-16 during dauer, including 112 transcription factors that may mediate repression of adult-enriched collagens during dauer downstream of daf-16.
Results
daf-16 blocks adult-specific collagen expression in dauer larvae
Progenitor cells remain multipotent throughout dauer diapause, and daf-16 is required to maintain or re-establish multipotency in the VPCs of dauer larvae (Karp and Greenwald 2013). We asked whether daf-16 might also promote multipotency in lateral hypodermal seam cells, another multipotent progenitor cell type. The well-characterized adult cell-fate marker col-19p::gfp, where GFP expression is driven by the adult-specific col-19 promoter, is expressed in differentiated seam cells and the hyp7 syncytium in adult worms but is not expressed in multipotent seam cells or hyp7 in larvae. The stage-specific expression of col-19p::gfp is controlled by heterochronic genes (Liu et al. 1995; Abrahante et al. 1998; Feinbaum and Ambros 1999). To determine if daf-16 promotes multipotent, larval seam cell fate during dauer, we assayed larvae with reduced or absent daf-16, termed “daf-16(-)”, for precocious expression of col-19p::gfp. Specifically, we tested two daf-16 null alleles, daf-16(mu86) and daf-16(mgDf50), as well as daf-16(RNAi) dauer larvae. As daf-16(-) larvae are dauer-defective, we used the dauer-constitutive allele daf-7(e1372) to promote dauer formation in daf-16(-) and control larvae (Vowels and Thomas 1992; Larsen et al. 1995). For simplicity, daf-7(e1372) is not always mentioned, but this allele was present in all daf-16(-) and control dauer larvae (see Table S1 for a complete list of strains used in this study). We found that daf-16(-) dauer larvae displayed penetrant col-19p::gfp expression in seam cells and hyp7 (Fig 1A). col-19p::gfp was expressed throughout the hypodermis, but was consistently brighter in the posterior third of the worm, gradually dimming toward the anterior. These findings indicate that daf-16 is necessary to block col-19p::gfp expression during dauer. For the remainder of the paper, “daf-16(0)” is used to indicate the daf-16(mgDf50) null allele, a large deletion that removes most of the daf-16 coding sequence, including the DNA-binding domain (Ogg et al. 1997).
Wild type daf-16 promotes dauer formation largely in parallel to the DAF-3/SMAD-DAF-5/Ski complex (Fielenbach and Antebi 2008; Baugh and Hu 2020). To test whether this complex is also required to block col-19p::gfp expression during dauer, we examined daf-5(0) dauer larvae. Since daf-5(0) larvae are dauer-defective, we used the daf-2(e1370) allele to drive dauer formation (Vowels and Thomas 1992; Larsen et al. 1995). We found that daf-5(0); daf-2(e1370) dauer larvae never expressed col-19p::gfp (Fig S1). Therefore, while daf-16 and daf-5 act in parallel to promote dauer formation, the regulation of adult cell fate is specific to daf-16.
We next asked whether daf-16 regulates col-19p::gfp expression during stages other than dauer. First, we examined the timing of col-19p::gfp expression in larvae before, during, and after their entry into dauer. In wild-type animals, col-19p::gfp expression begins during the L4 molt, peaks soon after molting to adulthood, and then declines over time. Similarly, in daf-16(0) mutants, col-19p::gfp expression begins during the L2d-to-dauer molt, peaks soon after larvae enter dauer, and then declines over time in dauer (Fig 1B). Next, we asked whether larvae that develop continuously express precocious col-19p::gfp. For this experiment, we used larvae that were wild-type for daf-7. col-19p::gfp expression was never observed in daf-16(0) larvae during continuous development, and col-19p::gfp was expressed normally in adults (Fig 1C). Finally, we asked whether daf-16(0) larvae in L1 arrest express col-19p::gfp. Dauer shares some similarities with L1 arrest, which is a developmentally arrested and quiescent stage occurring when L1 larvae hatch in the absence of food (Johnson et al. 1984). daf-16 is important for L1 arrest; arrested daf-16(0) mutant larvae precociously initiate post-embryonic development including V lineage seam cell divisions (Baugh and Sternberg 2006). However, we saw no precocious col-19p::gfp expression in these larvae (Fig 1C).
We next wondered which daf-16 isoforms block col-19p::gfp expression during dauer. There are three major isoforms of daf-16: a, b, and f (Ogg et al. 1997; Lin et al. 1997; Kwon et al. 2010). The a and f isoforms together are the major players with respect to the role of daf-16 in longevity, dauer formation, and stress resistance (Chen et al. 2015). Isoform a plays a larger role than f, but the f isoform still contributes because mutants lacking isoforms a and f display a stronger phenotype than mutants lacking either single isoform. By contrast, the b isoform is dispensable because animals lacking a and f isoforms were indistinguishable from null mutants for the phenotypes tested (Chen et al. 2015). In contrast, the b isoform regulates neuronal morphology and function (Christensen et al. 2011; Sun et al. 2019). To determine which daf-16 isoforms contribute to maintaining seam cell multipotency during dauer, we used the same isoform-specific alleles used by Chen et al. (2015) to assess col-19p::gfp expression during dauer. Loss of only daf-16a produced a stronger phenotype than loss of only daf-16f, and loss of both daf-16a and f produced an even stronger phenotype than loss of either individual isoform examined, indicating that both isoforms contribute (Figs 1D, S2). Loss of a and f together caused expression of col-19p::gfp at high penetrance (Fig 1D), however this expression was noticeably dimmer and in fewer cells than the expression observed with in the null background (Fig S2). Therefore, since loss of daf-16a/f did not produce a phenotype as strong as that of the complete null, the b isoform must also play a role. Taken together, we conclude that all three isoforms of daf-16 contribute to blocking col-19p::gfp expression during dauer.
To extend these findings beyond col-19p::gfp reporter expression, we performed mRNA-seq and differential gene expression analysis on daf-16(0) vs. control dauer larvae. Consistent with the data from col-19p::gfp experiments, we found that endogenous col-19 was highly and significantly upregulated in daf-16 mutants. Furthermore, 19/22 other adult-enriched collagens were also upregulated (Fig 1E, Table S2). Consistent with the upregulation of adult-enriched collagens, 23/24 dauer-enriched collagens were downregulated in daf-16(0) dauer larvae (Fig 1E), indicating an overall shift to genes required for adult cuticle formation in daf-16(0) dauer larvae.
In wild-type larvae, the onset of col-19p::gfp expression coincides with several other characteristics of adult seam cell fate, including the production of adult alae, cell-cycle exit, and seam-cell fusion (Sulston and Horvitz 1977; Ambros and Horvitz 1984; Podbilewicz and White 1994). Although daf-16(0) dauer larvae express adult collagens, they do not display these other adult characteristics. Consistent with previous reports, we observed that daf-16(0) dauer larvae display dauer alae, which are distinct from adult alae (Vowels and Thomas 1992; Ogg et al. 1997). Using the wIs78 transgenic strain that expresses GFP in both seam cell nuclei (scm::gfp) and apical junctions (ajm-1::gfp) (Abrahante et al. 2003), we found that seam cells in daf-16(0) dauer larvae remained unfused. Additionally, seam cells in daf-16(0) dauer larvae failed to maintain quiescence and underwent cell divisions. These cell divisions were similar to the asymmetric divisions that normally occur during the L1, L3, and L4 stages (Fig S3). Seam cell division does not occur during either adult or dauer stages in wild-type animals, however daf-16 is required for cell-cycle arrest in other contexts (Narbonne and Roy 2006; Baugh and Sternberg 2006). Therefore, during dauer daf-16 both promotes seam cell quiescence and blocks adult cell fate, where its role in regulating seam cell fate is predominantly to control stage-specific collagen expression.
daf-16 promotes expression of lin-41 during dauer
During continuous development, the timing of col-19p::gfp expression is regulated by the heterochronic gene network (Ambros and Horvitz 1984; Liu et al. 1995; Feinbaum and Ambros 1999). To determine how daf-16 interacts with heterochronic genes, we began by asking whether heterochronic genes that oppose adult cell fate during continuous development might also be required to block col-19p::gfp expression during dauer. We focused on the three most downstream of these heterochronic genes, lin-14, hbl-1, and lin-41 (Fig 2A). We found that knockdown of lin-14 and hbl-1 resulted in little to no col-19p::gfp expression in hypodermal cells whereas lin-41 RNAi resulted in penetrant col-19p::gfp expression in lateral hypodermal cells during dauer (Fig 2B). Similar to daf-16(RNAi), col-19p::gfp expression in lin-41(RNAi) dauer larvae was brightest in the posterior of the worm, and dim or absent in the anterior. However, overall col-19p::gfp expression was dimmer in lin-41(RNAi) dauer larvae compared to daf-16(RNAi). In addition, some col-19p::gfp expression was observed in non-hypodermal cell types in hbl-1(RNAi) and lin-41(RNAi) dauer larvae (Fig S4).
Since both daf-16 and lin-41 block col-19p::gfp expression in dauer larvae, we asked whether daf-16 regulates lin-41 expression. We examined our mRNA-seq data and also performed qPCR on daf-16(0) and control dauer larvae. In these experiments, lin-41 was downregulated approximately 2-fold in daf-16(0) dauers compared to control dauer larvae (Fig 3A). In contrast, there was no significant change in mRNA levels of the other core heterochronic genes (Fig S5). Together, these data suggest that during dauer, daf-16 specifically promotes lin-41 expression to block col-19p::gfp and adult cell fate, and that in daf-16(-) dauers, lower levels of lin-41 lead to precocious col-19p::gfp expression.
To test whether reduced lin-41 levels are required for precocious col-19p::gfp expression in daf-16(0) dauers, we performed daf-16 RNAi on a strain in which lin-41 is misexpressed. To do this, we took advantage of the lin-41(xe8[Δ3’UTR]) allele which prevents the normal downregulation of lin-41 in late larval stages and causes a strong gain-of-function phenotype (Ecsedi et al. 2015; Aeschimann et al. 2019). We found the col-19p::gfp phenotype produced by daf-16(RNAi) was markedly suppressed in lin-41(xe8) dauer larvae, consistent with the hypothesis that daf-16 works through lin-41 to block col-19p::gfp expression during dauer (Fig 3B). To ensure that the observed suppression was not due to the lin-41(xe8) strain being less sensitive to RNAi, we tested the response of lin-41(xe8) and control strains to unc-22 RNAi and found a similar response in both strains (Fig S7).
lin-29 is not required for col-19p::gfp expression in lin-41(-) dauer larvae
During continuous development, lin-41 regulates col-19p::gfp and adult cell fate by directly regulating the translation of the LIN-29 transcription factor (Slack et al. 2000; Aeschimann et al. 2017). LIN-29 is in turn a direct activator of col-19 transcription and is thought to be the most downstream regulator of other aspects of adult cell fate (Ambros 1989; Rougvie and Ambros 1995; Azzi et al. 2020). Consistent with the regulation of lin-29 by lin-41 during continuous development, loss of lin-29 completely suppresses lin-41(-) precocious phenotypes (Slack et al. 2000). If the col-19p::gfp expression observed in lin-41(RNAi) dauer larvae is due to misexpression of LIN-29, then loss of lin-29 should prevent col-19p::gfp expression. To ask this question, we used the lin-29(xe37) deletion allele that removes all but 27 amino acids of LIN-29 (Aeschimann et al. 2019). Surprisingly, we saw no effect of the loss of lin-29 on the expression of col-19p::gfp in lin-41(RNAi) dauer larvae (Fig 4A). As a control, we established that our strain produced the expected lin-29(0) phenotypes in adults: a drastic reduction of col-19p::gfp and a complete lack of adult alae formation (Fig S8).
We next confirmed the lin-41 RNAi results using the null allele lin-41(n2914). Since lin-41(0) hermaphrodites are sterile (Slack et al. 2000), we used an available transgene that is integrated close to the lin-41 locus as a balancer: nIs408[lin-29::mCherry, ttx-3p::gfp] (Harris and Horvitz 2011). This transgene had the added advantage of rescuing some of the lin-29(0) defects, making the strain easier to maintain (see Methods). When we induced dauer formation in the progeny of lin-41(0)/nIs408 and lin-41(0)/nIs408; lin-29(0) mothers, we found that lin-41(0); lin-29(+) homozygous larvae, recognized by the lack of ttx-3p::gfp expression, were largely dauer defective (see (Cale and Karp 2020)). Of the two lin-41(0) dauer larvae we recovered, one expressed col-19p::gfp (Fig 4B). Interestingly, 62% of larvae expressing ttx-3p::gfp displayed col-19p::gfp expression. This percentage is close to the 2/3 of transgene-containing larvae segregating from the heterozygous parent that would be predicted to be lin-41(0)/nIs408 heterozygotes, suggesting that lin-41 is haploinsufficient in its role in blocking col-19p::gfp during dauer. To bolster this supposition, we confirmed that dauer larvae homozygous for nIs408 do not express col-19p::gfp (Fig 4B).
Although lin-41(0) homozygous larvae display a dauer-defective phenotype, lin-41(0); lin-29(0) larvae do not (Cale and Karp 2020). We were therefore able to test lin-41(0); lin-29(0) dauer larvae for col-19p::gfp expression. Approximately half of these larvae expressed col-19p::gfp (Fig 4B). Although the dauer-defective phenotype of lin-41(0) single mutants prevented us from determining the extent to which loss of lin-29 suppresses the col-19p::gfp phenotype in lin-41(0) homozygous mutant dauer larvae, these results do confirm that col-19p::gfp can be misexpressed in lin-41(0) dauer larvae even in the presence of a lin-29 null allele.
To better understand the relationship between lin-41 and lin-29 during dauer, we asked whether lin-41 is required to block LIN-29 expression in dauer larvae. During continuous development, the lin-29a isoform is translationally repressed by LIN-41, and therefore lin-41 mutants display precocious hypodermal expression of LIN-29 (Slack et al. 2000; Aeschimann et al. 2017). We used lin-29(xe61[lin-29::gfp]), which tags both isoforms of lin-29 (Aeschimann et al. 2017) to assess the effect of lin-41 RNAi during dauer. Using settings that allow unambiguous visualization of hypodermal lin-29::gfp during L4 and adult stages, we saw only extremely dim lin-29::gfp expression in lin-41(RNAi) dauer larvae, just at the edge of detection (Fig 4C). We used confocal microscopy as a more sensitive assay to confirm that lin-29::gfp is expressed, albeit at very low levels in lin-41(RNAi) but not control dauer larvae (Fig 4D). In control RNAi experiments run in parallel, lin-41(RNAi) was able to cause strong col-19p::gfp expression (24/27 larvae examined), demonstrating that there were no technical problems in the RNAi experiment that would lead to such low expression. Finally, we confirmed that lin-41(RNAi) produces the expected precocious lin-29::gfp expression in L3 staged larvae that developed continuously (Fig 4C). Taken together, contrary to the role of lin-29 during continuous development, our data indicate that lin-41 regulates col-19p::gfp expression in dauer larvae and that lin-29 does not play a significant role in this regulation.
daf-16 acts at least partially independently of lin-29 to regulate col-19p::gfp during dauer
As described above, daf-16 promotes lin-41 expression during dauer, and misexpression of lin-41 suppresses the precocious col-19p::gfp expression observed in daf-16(0) dauer larvae. Since lin-29 was not required for misexpression of col-19p::gfp in lin-41(-) dauer larvae, we hypothesized that lin-29 would also be dispensable for col-19p::gfp expression in daf-16(-) dauer larvae. Neither of two lin-29 null alleles affected the penetrance of col-19p::gfp expression in daf-16(0) mutant dauers, demonstrating that the misexpression of col-19p::gfp in daf-16(0) dauer larvae does not depend on lin-29 (Fig 5A). However, when we compared levels of expression between the strains, we found a small but statistically significant decrease in expression in dauer larvae that lack lin-29, indicating that the presence of lin-29 bolsters col-19p::gfp expression slightly (Fig 5B). We next asked whether loss of daf-16 affects expression of lin-29. Examining our mRNA-seq data, no significant difference in lin-29 mRNA levels were observed in daf-16 vs. control dauer larvae (Fig S5). However, during continuous development, regulation of lin-29a by lin-41 occurs translationally and may not be evident from mRNA levels (Bettinger et al. 1996; Aeschimann et al. 2017). We next examined lin-29::gfp expression in daf-16(0) dauer larvae. Unlike the dim lin-29::gfp expression we observed in lin-41(RNAi) dauer larvae, lin-29::gfp expression in the hypodermis was completely undetectable in daf-16(0) dauer larvae. Using confocal microscopy, 0/24 daf-16(0) dauer larvae displayed detectable lin-29::gfp. Taken together, these experiments demonstrate that daf-16 regulates col-19p::gfp expression largely independently of lin-29.
daf-16 regulates the expression of many genes during dauer
Since the key downstream regulator of col-19 expression is not required for the misexpression of col-19p::gfp in daf-16(0) dauer larvae, one or more different regulators must activate or derepress col-19p::gfp expression in this context. One possibility is that DAF-16 itself is a direct activator. Two pieces of indirect evidence initially argued against this possibility. First, the promoter sequence used in the col-19p::gfp transgene does not contain a canonical DAF-16 binding element (TGTTTAC) (Furuyama et al. 2000). Furthermore, ChIP-seq experiments performed as part of modENCODE and displayed on WormBase did not show binding of DAF-16 to the col-19 promoter (Fig S9) (Gerstein et al. 2010). However, this evidence is not conclusive. The consensus sequence for DAF-16 binding elements is very broad, leaving open the possibility that DAF-16 could recognize a non-canonical site (Tepper et al. 2013). In addition, the ChIP-seq experiments were carried out during the L4/young adult stage, whereas the only time that we see aberrant col-19p::gfp expression in daf-16(0) mutant larvae is during dauer. To more definitively test DAF-16 binding at the col-19 promoter during the dauer stage, we used an endogenously tagged daf-16::zf1::wrmScarlet::3xFLAG strain to perform ChIP-qPCR during dauer. As expected, the promoter for a confirmed DAF-16 target, mtl-1 (Barsyte et al. 2001; Li et al. 2008) showed 18-30 fold enrichment in the daf-16::3xflag ChIP sample compared to wild type. By contrast, the promoter for col-19 showed only 2-4-fold enrichment in the daf-16::3xflag ChIP sample compared to wild type (Figs 6A, S10). This lower level of DAF-16 enrichment at the col-19 promoter was comparable to the DAF-16 enrichment at the coding region of the heterochromatinized gene bath-45, suggesting that the DAF-16 ChIP-qPCR experiments were inherently noisy. These data indicate that daf-16 blocks col-19p::gfp expression indirectly.
To identify candidate regulators of col-19p::gfp and endogenous adult collagen expression during dauer, we looked more broadly at the changes in gene expression that occur downstream of daf-16 using our mRNA-seq data comparing daf-16(0) vs. control dauer larvae. We found that there are many genes whose expression levels are regulated by daf-16 during dauer. Specifically, we found that 2027 genes were downregulated ≥2-fold in daf-16(0) dauers, and 1576 genes were upregulated ≥2-fold in daf-16(0) dauers (FDR ≤ 0.05) (Fig 6B). Interestingly, when we performed functional annotation clustering on these differentially expressed genes using the Database for Annotation, Visualization, and Integrated Discovery (http://david.abcc.ncifcrf.gov/), terms related to collagens and cuticle structure were highly enriched among both downregulated and upregulated genes (Fig S11). Signaling-related terms were also highly enriched, including terms related to ion transport associated with the downregulated genes and terms related to protein kinases and phosphatases associated with the upregulated genes (Fig S11). Of the genes whose expression was affected, 112 encode transcription factors (91 downregulated and 21 upregulated) and are candidate genes to directly regulate col-19p::gfp expression and adult cell fate during dauer (Fig S12).
Discussion
Heterochronic genes specify stage-specific seam cell fate at each larval stage and at adulthood (Ambros and Horvitz 1984). During continuous development, heterochronic genes function as a cascade where successive microRNAs act as molecular switches to downregulate early cell fate-promoting transcription factors and RNA-binding proteins, allowing progression to the next cell fate (Rougvie 2001; Rougvie and Moss 2013). The decision to enter dauer drastically alters the timing of developmental progression (Cassada and Russell 1975). Perhaps for this reason, extensive modulation of the heterochronic pathway has been identified in pre- and post-dauer stages. (Liu and Ambros 1991; Hammell et al. 2009; Karp and Ambros 2012; Ilbay and Ambros 2019). For example, many heterochronic genes that are essential for stage-specific cell fate specification during continuous development are dispensable after dauer (Liu and Ambros 1991; Abrahante et al. 1998, 2003; Karp and Ambros 2012). However, the mechanisms that act to prevent precocious specification within the dauer stage itself have not been addressed. Here, we find a novel role for daf-16 as a dauer-specific heterochronic gene. daf-16 opposes adult cell fate during dauer via a modified heterochronic pathway involving lin-41 but not lin-29.
The heterochronic gene lin-41 encodes an RNA-binding protein that promotes larval cell fate and opposes adult cell fate during continuous development. lin-41 is expressed early in larval development and downregulated by the let-7 microRNA during the L4 stage to allow progression to adult cell fate (Reinhart et al. 2000; Slack et al. 2000). We found that lin-41 acts downstream of daf-16 to block adult cell fate during dauer. Specifically, both daf-16 and lin-41 are required to prevent precocious expression of the col-19p::gfp adult cell fate marker during dauer, and lin-41 expression is downregulated in daf-16(0) dauer larvae (Figs 1A, 2B, 3A). In contrast, the other early-promoting heterochronic genes tested, lin-14 and hbl-1, do not appear to play a role in regulating col-19p::gfp hypodermal expression during dauer, and their expression was not significantly affected in daf-16(0) dauer larvae (Figs 2B, S5). Finally, a lin-41 gain-of-function allele partially blocked the misexpression of col-19p::gfp caused by daf-16 RNAi. All together, these data are consistent with a linear pathway whereby daf-16 positively regulates lin-41 during dauer to oppose col-19p::gfp and adult cell fate (Fig 6D). These findings do not exclude the possibility that daf-16 also regulates col-19p::gfp and adult cell fate in parallel to lin-41.
The mechanism by which daf-16 promotes lin-41 expression is currently unknown. One possibility is that daf-16 regulates expression of lin-41 via the let-7 microRNA. let-7 is the major regulator of hypodermal lin-41 expression during continuous development, where let-7 opposes lin-41 expression by binding to the lin-41 3’UTR and mediating silencing (Reinhart et al. 2000; Slack et al. 2000). The ability of the lin-41(xe8[Δ3’UTR]) allele to interfere with the daf-16(-) phenotype suggests the possibility that let-7 is involved in this regulatory pathway.
During continuous development lin-41 blocks adult cell fate by directly repressing the translation of the most downstream component in the heterochronic pathway, the LIN-29 transcription factor that promotes all aspects of adult cell fate (Ambros and Horvitz 1984; Ambros 1989; Rougvie and Ambros 1995; Aeschimann et al. 2017; Azzi et al. 2020). LIN-29 expression remains low in the lateral hypodermis during early larval development due to the combined action of lin-41 and hbl-1 which regulate distinct lin-29 isoforms (Slack et al. 2000; Aeschimann et al. 2017; Azzi et al. 2020). Surprisingly, we found that daf-16 and lin-41 regulate adult cell fate mostly independently of lin-29 (Fig 6D). During continuous development, the precocious phenotypes observed in lin-41(-) larvae are completely suppressed by compromising lin-29 activity (Slack et al. 2000). Indeed, the phenotypes of every precocious heterochronic mutant tested during continuous development are suppressed by lin-29(-) (Ambros 1989; Abrahante et al. 1998, 2003; Slack et al. 2000; Lin et al. 2003). In contrast, loss of lin-29 had little to no effect on the precocious col-19p::gfp phenotype observed in daf-16(-) or lin-41(-) dauer larvae (Figs 4A, 4B, 5A, 5B). Therefore, whereas lin-29 is a direct transcriptional activator of col-19 and other adult-enriched collagens in the context of adults (Liu et al. 1995; Rougvie and Ambros 1995; Abete-Luzi et al. 2020), our work demonstrates that during dauer, daf-16 blocks adult collagen expression through factor(s) other than lin-29. We found over 3600 genes whose expression changed in daf-16(0) dauer larvae, including 112 transcription factors. One or more of these factors may control the expression of the col-19p::gfp transcriptional reporter.
In addition to regulating col-19p::gfp, we found that daf-16 is required during dauer to block expression of nearly all adult-enriched collagens. The expression of adult-enriched collagens in daf-16(0) dauer larvae is accompanied by a concomitant decrease in the expression of dauer-enriched collagens (Fig 1E). The effect of daf-16 mutations on the dauer cuticle may explain some of the previously observed defects in daf-16(0) dauer larvae. Two defining features of dauer larvae are the presence of dauer alae on the cuticle and resistance to treatment with detergents such as SDS. SDS-resistance depends in part on the specialized dauer cuticle (Cassada and Russell 1975). daf-16(0) dauer larvae possess dauer alae that are slightly less defined than wild-type, and daf-16(0) dauers are only partially SDS-resistant (Vowels and Thomas 1992; Ogg et al. 1997; Nika et al. 2016). It is possible that the shift from dauer-enriched collagens to adult-enriched collagens in daf-16(0) dauer larvae is responsible for these phenotypes. Notably, our functional annotation clustering analysis found that collagen-related terms were highly enriched among both upregulated and downregulated genes. This finding suggests that regulation of stage-specific collagen expression is a key role of daf-16 during dauer.
Dauer interrupts development midway through the larval stages (Cassada and Russell 1975). During dauer, progenitor cells that have not yet completed development must maintain or re-establish multipotent fate, neither differentiating prematurely nor losing their previously acquired tissue identity. The role we identified for daf-16 in blocking adult cell fate in lateral hypodermal cells during dauer appears to contribute to the maintenance of multipotency in lateral hypodermal seam cells. We have previously described an analogous role for daf-16 in promoting multipotent VPC fate (Karp and Greenwald 2013). In both contexts, daf-16 activity is important to prevent precocious adoption of cell fates that normally occur later in development, after recovery from dauer. However, the mechanisms by which these cell fates are adopted differ in each cell type. Adult seam cell fate is regulated by the heterochronic genes, whereas VPC fate specification is regulated primarily by EGFR/Ras and LIN-12/Notch signaling (Ambros and Horvitz 1984; Sternberg 2005; Rougvie and Moss 2013). The ability of daf-16 to influence these distinct developmental pathways suggests that daf-16 has a broad role in promoting multipotent cell fate during dauer. As one of the major regulators of dauer formation, daf-16 is well-positioned to coordinate the decision to enter dauer with the necessary alterations to developmental pathways that are paused during dauer.
daf-16 is the sole C. elegans ortholog of the genes encoding the FOXO proteins (Ogg et al. 1997; Lin et al. 1997). FOXO transcription factors regulate both stem cell quiescence and stem cell plasticity across species, from Hydra to mammals (Boehm et al. 2012; Liang and Ghaffari 2018). However, the mechanisms by which FOXO promotes multipotency in stem cells are still emerging. The protein encoded by the lin-41 ortholog, LIN41/TRIM71 also promotes cell fate plasticity in mammalian stem cells (Worringer et al. 2013). Our work provides the first connection between daf-16/FOXO and lin-41/TRIM71 and may be relevant to mammalian stem cells.
Materials and Methods
Strains and maintenance
A full list of strains and their genotypes used in this study is located in Table S1. Balanced strains are described in more detail here. Homozygous lin-41(0) hermaphrodites are sterile, therefore for experiments involving the null allele lin-41(n2914), lin-41 was balanced with transgene nIs408[ttx-3p::gfp, lin-29::mCherry] (Harris and Horvitz 2011), which we found to be closely linked to lin-41. At each generation, larvae that expressed ttx-3p::gfp were singled out and their progeny were monitored for segregation of lin-41(0) homozygotes that were Dpy, Ste, and lacked ttx-3p::gfp. The rescuing lin-29::mCherry enabled more robust growth in the lin-41(0)/nIs408; lin-29(0) strain than typical strains homozygous for lin-29(0). Although lin-29(0) mutants are homozygous viable, they are Egl, Pvl, and have reduced brood size (Ambros and Horvitz 1984). For experiments, lin-41(n2914) homozygous larvae were identified based on the lack of ttx-3p::gfp expression.
Animals homozygous for the lin-41(xe8[Δ3’UTR]) allele burst at young adulthood (Ecsedi et al. 2015). This gain-of-function allele was balanced over the lin-41(bch28 xe70) allele, where bch28 is a complex insertion of eft-3p::gfp that disrupts the lin-41 locus, and xe70 is a deletion of the lin-41 3’UTR (Katic et al. 2015; Aeschimann et al. 2019). For experiments, lin-41(xe8) larvae were identified based on the lack of eft-3p::gfp expression.
All strains were grown according to standard procedures on Nematode Growth Medium (NGM) plates seeded with the E. coli strain OP50 (Brenner 1974). Strains were maintained at 15°C or 20°C.
Dauer induction
All strains used for experiments involving dauer larvae contained daf-7(e1372), which is a temperature-sensitive, hypomorphic allele that induces dauer entry at 24°C or 25°C (Vowels and Thomas 1992; Karp 2018). Unless otherwise specified, dauer larvae were obtained by allowing 10-20 gravid adult hermaphrodites to lay embryos for 2-8 hours at 24°C, then removing the parents and allowing the embryos to incubate at 24°C. For all experiments except those in figure 1B, dauer larvae were scored approximately 48-52 hours after egg-laying, a time soon after dauer formation (“early dauer” or 0-day dauer larvae). In figure 1B when different stages were examined, L2d larvae were scored at 24 hours after egg-laying; L2d-dauer molt larvae were scored at 39 hours after egg-laying; 1-day dauer larvae were scored at 72 hours after egg-laying. Dauer formation was verified by looking for crisp, defined dauer alae and radial constriction (Karp 2018).
For experiments involving homozygous egg-laying-defective strains, including those with lin-29(0) or lin-29(xe61), embryos were obtained by sodium hypochlorite treatment consisting of two 2-minute incubations in 1M NaOH, 10% Clorox bleach. For experiments with XV254 daf-16(mgDf50); lin-29(xe37); daf-7(e1372); maIs105[col-19p::gfp] and controls, embryos were obtained by dissecting gravid adults.
Nondauer stages
Continuous development (Fig 1C)
Strains XV33 maIs105[col-19p::gfp] and VT1750 daf-16(mgDf50); maIs105 were synchronized by allowing gravid adult hermaphrodites to lay embryos at 24°C or 25°C and then incubating the progeny until the desired stage was reached. Developmental stage was evaluated by the extent of gonad and/or vulval development and scored for col-19p::gfp or lin-29::gfp expression.
L1 arrest (Fig 1C)
To induce L1 arrest, embryos from strains XV33 and VT1750 were obtained by sodium hypochlorite treatment and then incubated at 20°C in M9 in a shaking incubator. L1 larvae were isolated each day for 7 days and scored for col-19p::gfp expression.
Adults (Fig S8)
Strains VT1777 daf-7(e1372); maIs105 and XV253 lin-29(xe37); daf-7(e1372); maIs105 were synchronized by sodium hypochlorite treatment. Embryos were placed on seeded NGM plates and grown at 20°C for 72-75 hours. Adults were scored for presence of stage-specific alae and col-19p::gfp expression.
RNAi
The RNAi bacteria were from the Ahringer library (Source Bioscience), except the lacZ RNAi clone was pXK10 (Karp and Greenwald 2003). RNAi plates were prepared by adding 300µL of an overnight culture of RNAi bacteria to 60mm NGM plates containing 50µg/mL carbenicillin and 200µg/mL IPTG. Embryos obtained from sodium hypochlorite treatment were plated onto seeded RNAi plates and incubated at 24°C to allow daf-7(e1372) larvae to enter dauer or daf-7(+) larvae to develop continuously.
Compound Microscopy
Animals were picked onto slides made with 2% agarose pads and paralyzed with 0.1 M levamisole. A Zeiss AxioImager D2 compound microscope with HPC 200 C fluorescent optics was used to image worms. DIC and fluorescence images were obtained using a AxioCam MRm Rev 3 camera and ZEN 3.2 software. GFP was visualized with a high efficiency GFP shift free filter.
Confocal Microscopy
Animals were picked onto slides made with 2% agarose pads and paralyzed with 0.1 M levamisole. A Nikon A1R scanning laser confocal light microscope was used to image worms using an excitation laser set to at wavelength of 488 nm. To visualize lin-29::gfp, the laser power was set to 70%.
Phenotypic Data Collection and Analysis
All phenotypic data presented are from at least two independent experiments, typically performed by independent researchers. Any experiments that involved subjective decisions were blinded by having a lab member not involved in the experiment code the strains and/or images before scoring. Statistical analyses were performed on Graphpad Prism (version 9.1) and specific tests are described in individual figure legends. P-values < 0.05 were deemed statistically significant.
Sample Collection for qPCR and RNA sequencing
Synchronized populations of VT2317 daf-16(mgDf50); daf-7(e1372) and control CB1372 daf-7(e1372) dauer larvae were obtained by incubating embryos isolated by sodium-hypochlorite treatment at 24°C for 52 hours. Because some daf-16; daf-7 animals grown under these conditions fail to enter dauer (Nika et al. 2016), dauer larvae from both strains were handpicked into M9 solution, washed twice, and then pelleted to a volume of 100µl packed worms. TRIzol reagent (Invitrogen) was added to the samples at a 10:1 ratio of TRIzol to worm pellet. The samples were then frozen in dry-ice/ethanol. Two biological replicates were obtained for each strain for mRNA-seq. For qPCR, we obtained two biological replicates of the wild-type sample and three biological replicates of the daf-16(0) mutant sample.
RNA Isolation
Total RNA isolation from dauer sample was conducted using TriReagent (Ambion) protocol with the following modifications: pelleted C. elegans in TRI-Reagent were subjected to three freeze/thaw/vortex cycles prior to BCP addition to improve extraction efficiency, isopropanol precipitation was conducted in the presence of glycogen for 1hr at −80°C, RNA was pelleted by centrifugation at 4°C for 30 minutes at 20,000 x g; the pellet was washed three times in 70% ethanol and resuspended in water. BioAnalyzer assay (Agilent Technologies) was used for quality control of the RNA sequencing samples prior to library creation, with a minimum RIN of 8.5. NanoDrop2000 (Thermo Scientific) was used to quantify and assess the RNA quality of samples analyzed by qPCR.
RT-qPCR
cDNA was synthesized from 250ng total RNA using SuperScript III Reverse Transcriptase (Invitrogen) and analyzed with a CFX96 Real-Time System (BioRad) using Absolute Blue SYBR Green PCR MasterMix (Life Technologies). Relative lin-41 mRNA levels were calculated based on the ΔΔ2Ct method (Nolan et al. 2006) using eft-2 for normalization. Results presented are the average values of independent calculations from biological replicates.
RT-qPCR Primers
eft-2 F ACGCTCGTGATGAGTTCAAG
eft-2 R ATTTGGTCCAGTTCCGTCTG
lin-41 F GGTTCCAAATGCCACAAGAG
lin-41 R AGGTCCAACTGCCAAATCAG
mRNA-seq Library Preparation and Sequencing
Libraries were constructed using the TruSeq RNA Library Prep Kit v2 (Illumina). The DNA concentration and fragment size of sequencing libraries were analyzed using the BioAnalyzer assay (Agilent Technologies). High-throughput sequencing was performed on the Illumina HiSeq 2500 platform to generate paired-end reads of 100 bp.
mRNA-seq Analysis
Basecalling and base call quality were performed using Illumina’s Real-Time Analysis (RTA) software, and CIDRSeqSuite 7.1.0 was used to convert compressed bcl files into compressed fastq files. Using Trimmomatic v. 0.39, Illumina adapters were clipped from raw mRNA-seq reads, followed by quality trimming (LEADING: 5; TRAILING: 5; SLIDINGWINDOW:4:15) (Bolger et al. 2014). Reads with a minimum length of 36 bases were retained (MINLEN: 36). Processed reads were aligned to C. elegans reference genome WBcel235 using STAR v. 2.4.2a with default parameters and -- twopassMode Basic (Dobin et al. 2013). RSEM v. 2.1 was used to quantify gene abundance based on mapped reads (Li and Dewey 2011). Outlier samples were identified by assessing the quality of raw and processed reads with FastQC v. 0.11.7 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and by Principal Component Analysis (PCA) on log2-transformed normalized gene expression counts generated using DESeq2 v. 1.30.0 (Love et al. 2014). Differential gene expression analysis was performed using DESeq2 with a significance threshold of FDR ≤ 0.05 and absolute value of log2-transformed fold changes of at least 2.
DAVID analysis
Differentially upregulated and downregulated genes with at least a log2 transformed fold change of 2 and FDR ≤ 0.05 were individually analyzed using DAVID Functional Annotation Clustering (https://david.ncifcrf.gov/summary.jsp. Accessed 20210507). Terms from InterPro and Gene Ontology sub-ontologies cellular component (CC), biological process (BP), and molecular function (MF) were plotted by −log10 (FDR) and ranked by fold enrichment..
ChIP-qPCR
GS8924 (daf-16(ar620[daf-16::zf1-wrmScarlet-3xFLAG)), a gift from Katherine Luo and Iva Greenwald (Columbia University) and control N2 embryos obtained after sodium-hypochlorite treatment were hatched overnight in M9 buffer. The L1 arrested worms were grown on NGM plates seeded with HB101 at 25°C. After 5 days, the starved worm population was collected in M9 buffer and nutated in 1% SDS solution for 30 minutes to isolate dauers. The worm pellet was washed 4x with autoclaved DI water and the worms were plated on an NGM plate with no food. The live dauers crawled out and were collected as a ∼500μL packed pellet in M9 buffer.
The animals were nutated for 30min at room temperature in 12mL of 2.6% formaldehyde in autoclaved DI water for live crosslinking. The crosslinking was quenched with 600μL of 2.5M glycine for 5min at room temperature. The worms were washed 3x in water before flash freezing and storing at −80°C. Frozen pellets were ground twice for 1min in the Retsch MM400 cryomill at a frequency of 30/s. The worm powder was lysed in 2mL of 1xRIPA (1x PBS, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS) with 1xHALT Protease and phosphatase inhibitor (Thermo Scientific 78443) for 10min at 4°C. The crosslinked chromatin was sonicated in the Diagenode BioruptorPico for 3 3min cycles, 30s ON/OFF at 4°C. The chromatin was diluted to a concentration of 20-30ng/uL. 10% of the volume was processed as the input sample. 10ug of chromatin was incubated with 2ug of Monoclonal Anti-Flag M2 antibody (Sigma-Aldrich F3165) overnight at 4°C and then for 2 hours with Dynabeads M-280 Sheep Anti-Mouse IgG (Invitrogen 11201D). After 3 800uL washes in LiCl buffer (100mM Tris HCl pH 7.5, 500mM LiCl, 1% NP40, 1% sodium deoxycholate), the samples were de-crosslinked by incubating with 80ug of Proteinase K (Thermo Scientific 25530015) in Worm Lysis buffer (100mM Tris HCl pH 7.5, 100mM NaCl, 50mM EDTA, 1% SDS) at 65°C for 4 hours. The samples were subjected to phenol-chloroform extraction and the DNA pellet was resupended in TE buffer. RNase A (Thermo Scientific 12091021) treatment was performed for 1 hour at 37°C.
Quantitative PCR for promoter regions of interest was performed with Absolute Blue SYBR Green (Thermo Scientific AB4166B) on the CFX96 Real Time System Thermocyclers (Biorad) using custom primers. The cycle numbers in the ChIP samples were normalized to respective input values. The log 2 transformed fold change values in the daf-16::3xFLAG::wrmScarlet samples were normalized to the respective N2 samples. 2 biological replicates with 2 technical replicates each were done for the promoter regions of mtl-1 and col-19. Each biological replicate was analyzed separately.
ChIP-qPCR primers
mtl-1p F TGAGCACTCTAATCCTTTGCAC
mtl-1p R ACGTGAATGTTGCAAACACCT
col-19p F TCCATCTCTCTTGGAAACACAT
col-19p R ACACCTTCAAACCTAACCAGTGT
mtl-1 F ATTTAATTTGTTTTCAGAAGGCAGC
mtl-1 R GTGCTTTCTGCATCAGAGCC
Data Availability
mRNA-seq data have been deposited in NCBI under GEO accession number GSE179166. Processed data and scripts used for analysis are available at https://github.com/starostikm/DAF-16.
Acknowledgements
We are grateful to Brooklynne Watkins for advice about confocal microscopy and to other members of the Schisa lab (Central Michigan University) for helpful discussions. mRNA-seq library preparation, sequencing, and data processing into fastq format were conducted at the Genetic Resources Core Facility, Johns Hopkins Institute of Genetic Medicine, Baltimore, MD. Computational resources were provided by the Maryland Advanced Research Computing Center (MARCC). Many thanks to Katherine Leisan Luo in the Greenwald lab at Columbia University for sharing daf-16(ar620) prior to publication. We thank WormBase. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). This work was supported by NIH R01GM118875 and R01GM129301 (to JKK), NIH R15GM117568 (to XK), NSF CAREER 1652283 (to XK).