A model of simple linear control in the heterochronic pathway predicts that lin-29(0) mutations cause the same phenotypes as upstream mutations that impair the activation of lin-29, such as let-7(0). Indeed, the phenotypes of the two mutant strains overlap extensively (Ambros and Horvitz, 1984; Bettinger et al., 1997; Reinhart et al., 2000). However, they are not identical: Whereas lin-29(0) mutant animals fail to execute seam cell fusion (Bettinger et al., 1997), this process was reported to occur normally in let-7(mn112) mutant animals (Hunter et al., 2013). To validate this unexpected observation, we examined a newly created let-7 null mutant strain, let-7(xe150), which lacks let-7 expression due to deletion of the promoter (a gift from J. Kracmarova). Additionally, we used the temperature-sensitive let-7(n2853) strain (Reinhart et al., 2000) at the restrictive temperature, 25°C. Employing an ajm-1::mCherry marker to visualize the seam cells boundaries, we observed that seam cells fused normally in 100% of the let-7(xe150) and let-7(n2853) mutant animals, confirming that let-7 is indeed dispensable for seam cell fusion (Figure 1B). By contrast, complete loss of lin-29 in lin-29(xe37) null mutant animals (henceforth lin-29ab(∆); Aeschimann et al., 2019) caused failure of seam cell fusion in all animals (Figure 1B).

Contrasting with the normal seam cell fusion in let-7 mutant animals, overexpression of lin-41 had previously been reported to prevent seam cell fusion (Slack et al., 2000). However, when we examined lin-41(xe8[∆LCS]) mutant animals (Figure 1B), which lack the let-7 complementary sites (LCSs) in the lin-41 3’UTR and thus exhibit sustained high LIN-41 levels during the L4 stage (Aeschimann et al., 2017; Ecsedi et al., 2015), seam cell fusion occurred normally. Hence, the activity of the let-7–LIN-41 module cannot account for all the J/A transition events regulated by lin-29.

Recently, we showed that let-7–LIN-41 regulate lin-29a (and mab-10), but presumably not lin-29b (Aeschimann et al., 2017; Aeschimann et al., 2019). Accordingly, in let-7(0) and lin-41(∆LCS) animals, LIN-29a and MAB-10 levels are expected to remain low in the L4 stage while LIN-29b can accumulate. Hence, we wondered whether LIN-29b would suffice for seam cell fusion and thus explain the phenotypic discrepancy between let-7(0) and lin-29(0) mutant animals. To test this possibility, we generated two lin-29b isoform-specific mutations (Materials and methods). One, where we deleted the putative promoter of lin-29b, achieved extensive, but not complete depletion of LIN-29b, while leaving LIN-29a levels unaltered (Figure 2A,B). We will refer to it as lin-29b(lf). The other, lin-29b(Δ), where we altered lin-29b translation initiation to translate it out of frame, caused a complete loss of LIN-29b, but we cannot exclude a modest depletion of LIN-29a (Figure 2A,B). Thus, failure to see a phenotype of interest in the lin-29b(∆) mutant excludes an essential function of the b isoform. Conversely, observation of a phenotype in the lin-29b(lf) mutant reveals an essential contribution of the b isoform to this phenotype, although we might under-estimate the extent of this contribution. We engineered each mutation into two different backgrounds, wild-type animals for functional studies, and animals containing a GFP::3xFLAG-tag at the shared C-terminus of the LIN-29 isoforms for expression analysis by Western blotting and imaging (Key resources table).

We used these, and the previously generated lin-29(xe40) (henceforth lin-29a(∆)) and mab-10(xe44) (henceforth mab-10(0)) mutant strains (Aeschimann et al., 2019) to study the individual contributions of LIN-29a, LIN-29b and their co-factor MAB-10 to the different J/A transition events. MAB-10 itself is thought not to directly bind to DNA, but to bind to, and modulate the activity of, both LIN-29 isoforms (Harris and Horvitz, 2011). Consequently, in the lin-29a- or lin-29b-specific single mutants, either of the two LIN-29 isoforms left can still act together with its co-factor MAB-10, while double mutant animals lacking both MAB-10 and one LIN-29 isoform are left with only the other LIN-29 isoform, acting without the co-factor MAB-10.

To determine the individual roles of the LIN-29 isoforms in seam cell fusion, we used the ajm-1::mCherry marker to count the number of unfused junctions in the late L4 stage, after the last seam cell division and before the last molt (Figure 3A,C). lin-29a(∆) and mab-10(0) single mutant animals, as well as mab-10(0) lin-29a(∆) double mutant animals showed unperturbed seam cell fusion. This is consistent with the functional fusion seen in let-7(0) and lin-41(∆LCS) animals (Figure 1B).

In striking contrast to these findings, we observed penetrant seam cell fusion defects in both lin-29b(lf) and lin-29b(∆) single mutant strains, and this phenotype was not enhanced by concomitant loss of mab-10 expression, that is in mab-10 lin-29b double mutant animals (Figure 3A,C). These findings support the notion that the let-7–LIN-41–LIN-29a/MAB-10 module is dispensable for seam cell fusion. They also reveal an unanticipated specialization of LIN-29 isoforms, with LIN-29b being both necessary and sufficient for wild-type seam cell fusion. We conclude that the two LIN-29 isoforms fulfill distinct and non-redundant functions.

Prompted by the discovery of a non-redundant function of LIN-29 isoforms in seam cell fusion, we surveyed their individual contributions to other J/A transition events. First, we examined exit of seam cells from the cell cycle. As shown previously (Aeschimann et al., 2019; Ambros and Horvitz, 1984), lin-29ab(∆) (lin-29(xe37)) animals exhibit a fully penetrant phenotype: seam cells do not exit the cell cycle but instead continue to divide so that all animals show at least 25 seam cells instead of the canonical 16 at the young adult stage (Figure 3D). By contrast, seam cells exit the cell cycle normally in lin-29a(∆) and mab-10(0) animals. A partially penetrant defect occurs in mab-10(0) lin-29a(∆) double mutant animals (Figure 3D), recapitulating the phenotype of let-7(n2853) and lin-41(∆LCS) animals (Aeschimann et al., 2019). Unlike loss of LIN-29a alone, depletion of LIN-29b alone suffices to permit unscheduled seam cell divisions in young adult animals (Figure 3D). The phenotype is more penetrant in lin-29b(∆) than in lin-29b(lf) animals. Moreover, loss of mab-10 enhances the lin-29b(lf) but not the lin-29b(∆) mutant phenotype.

Collectively, these data thus confirm an involvement of all three factors, LIN-29a, LIN-29b and MAB-10 in seam cell cell cycle exit. The data also suggest a more prominent role for LIN-29b versus LIN-29a, and a function for MAB-10 in promoting LIN-29 activity in this process.

Cell division and terminal differentiation are normally mutually exclusive, tightly coupled events. In fact, exit from the cell division cycle is frequently considered a central aspect of terminal cell differentiation (Myster and Duronio, 2000). Hence, we were surprised to find that nuclear divisions continued to occur after seam cells had fused at the L4 stage in mab-10(0) lin-29a(∆) mutant animals (Figure 3A–D). We investigated this apparent discrepancy further and found that although in the L4 stage, all seam cells fused in all animals, cell junctions were apparent again in the young adult stage. Specifically, following the unscheduled divisions of seam cell nuclei in these syncytia, pairs of seam cell nuclei were separated from other pairs by cell junctions (Figure 3B). Hence, although we have not performed lineaging experiments, we consider it reasonable to assume that these represent newly formed cell boundaries that surround pairs of ‘cousins’ rather than individual sister cell nuclei. We conclude that exit from the cell cycle is not elicited by cell fusion, and vice versa, that cell fusion does not require permanent cell cycle exit.

To understand LIN-29 isoform function in late J/A transition events, we examined the cuticles of young adults of the different genetic mutant backgrounds by DIC microscopy. An adult cuticle is characterized by alae (Figure 4A (I)), ridges along the lateral sides of the whole worm that are secreted by seam cells (Singh and Sulston, 1978). Consistent with previous results (Ambros and Horvitz, 1984), we observed a complete lack of alae in lin-29ab(Δ) young adults (Figure 4B).

lin-29a(∆) mutant animals were partially defective in alae formation, exhibiting weak alae structures (Figure 4A (II)) that either covered the whole worm (‘complete’) or at least 50% of the body length (‘partial’) (Figure 4B). The ‘partial alae’ phenotype appears to reflect a delay in alae synthesis because 4 hr later, all lin-29a(∆) animals had complete, yet still weak, alae (n = 20). Animals depleted of LIN-29b either exhibited only alae patches (Figure 4A (III, IV)) or lacked alae entirely (Figure 4B), and unlike in the lin-29a(∆) animals, complete alae were not observed even 4 hr later (n = 20). lin-29b(Δ) animals had a more penetrant ‘no alae’ phenotype than lin-29b(lf) mutant animals, with almost half of the former animals lacking any detectable alae structure (Figure 4B). However, neither lin-29b mutation fully recapitulated the lin-29ab(∆) phenotype, and the alae defects were qualitatively distinct in the lin-29a(∆) and the lin-29b(∆) mutant animals. Finally, mab-10 mutant animals displayed normal, wild-type alae formation (Harris and Horvitz, 2011), and absence of MAB-10 did not enhance the phenotypes of any of the lin-29a or lin-29b mutations (Figure 4B).

We conclude that both LIN-29a and LIN-29b, but not MAB-10, are required for wild-type alae formation, and that their functions are partially distinct.

The last event of the J/A transition that we examined was molting. Using a high-throughput assay (Meeuse et al., 2020; Olmedo et al., 2015), we counted molts in >20 animals for each genotype (Figure 5A,B). We confirmed a previous report (Ambros and Horvitz, 1984) that lin-29 was required for the exit from the molting cycle by observing that all lin-29ab(∆) animals exhibited at least one extra molt, with ~50% exhibiting two extra molts in the time-frame of the experiment. By contrast, animals with almost all other single or double mutant combinations, that is mab-10(0), lin-29a(Δ), lin-29b(lf), lin-29b(Δ) single mutations and mab-10(0) lin-29b(lf) and mab-10(0) lin-29b(Δ) double mutations, did not display defects in exiting the molting cycle. The only exception were mab-10(0) lin-29a(Δ) animals, of which a small percentage executed one or two extra molts.

We conclude that LIN-29a and LIN-29b have a redundant function in promoting the exit from the molting cycle. Their co-factor MAB-10 has a minor contribution that becomes detectable in a sensitized background.

Since our experiments show that LIN-29a and LIN-29b have partially distinct functions (Figure 6—figure supplement 1), we asked whether their molecular differences, that is the unique N-terminal extension of 142 amino acids of LIN-29a (Figure 2A), are responsible for functional specialization. To address this issue, we genetically engineered the lin-29 locus such that this N-terminal extension was removed by deleting the lin-29a-specific coding exons (exons 2–4) (Figure 6—figure supplement 2A). This mutation created a lin-29a transcript identical to that of lin-29b except for harboring the lin-29a 5’UTR and encoding 23 additional N-terminal amino acids. We confirmed by western blotting that this truncated LIN-29a(∆N) protein had a similar size to LIN-29b and accumulated at roughly wild-type LIN-29a levels (Figure 6A). Moreover, the removal of the N-terminus did not affect the spatiotemporal expression pattern of lin-29a, and its expression remained under control of LIN-41 (Figure 6—figure supplement 2C,D).

We characterized the phenotypes of lin-29a(∆N), alone or in combination with a mab-10(0) allele, for all four J/A transition events. In all assays, deletion of the LIN-29a-specific N-terminus did not compromise LIN-29a function, resulting in a wild-type phenotype for all four J/A transition events (Figure 6B–D, Figure 6—figure supplement 2B). This data suggests that the distinct functions of LIN-29a and LIN-29b are not mediated by sequence differences, although we cannot formally exclude a contribution of the remaining unique 23 amino acids.

Given the apparent absence of differences in molecular function between the two LIN-29 isoforms, we revisited their developmental expression patterns. lin-29a and lin-29b transcripts were previously reported to exhibit largely similar temporal patterns of accumulation, being both detectable from L1 stage on, with increasing levels during development, peaking at the L4 stage, and decreasing in adulthood (Rougvie and Ambros, 1995). Furthermore, a similar spatiotemporal expression pattern was deduced from promoter activity reporter experiments (Bettinger et al., 1996). However, transcript quantification and promoter activity measurements do not account for additional layers of post-transcriptional regulation, such as LIN-41-mediated translational repression of lin-29a (Aeschimann et al., 2017). Previous analysis of LIN-29 protein accumulation by immunofluorescence could not distinguish between the two protein isoforms (Bettinger et al., 1996).

To elucidate the temporal expression pattern of lin-29 isoforms on the protein level, we examined animals carrying a C-terminal GFP::3xFLAG tag that marks both isoforms. Using Western blotting, we could distinguish the two isoforms by their distinct sizes. The levels of the LIN-29b protein agreed with its previously reported pattern of mRNA accumulation, with detectable accumulation throughout larval stages and an increase in levels in the L3 and L4 stages (Figure 7A). By contrast, although the lin-29a transcript is detectable in L1 and abundant in L2 (Rougvie and Ambros, 1995), LIN-29a protein was undetectable in L1 or L2 stage worms, accumulated weakly in L3 stage, and peaked in L4 stage worms. This is consistent with its post-transcriptional regulation by LIN-41 (Aeschimann et al., 2017). We note that in late larval stages and adults, we also detect a third band migrating in between the LIN-29a and LIN-29b bands (Figure 7A, asterisk). We did not find evidence for a transcript encoding this intermediate-size protein in available paired-end gene expression data (F. Gypas, personal communication, September 2019) and the molecular nature of this band remains unclear. For the purpose of this study, we considered only the two previously described isoforms (Rougvie and Ambros, 1995).

To examine the spatiotemporal expression patterns of the two lin-29 isoforms in vivo, we utilized strains with GFP::3xFLAG tags on the individual, endogenous LIN-29 isoforms. For lin-29a, we used a published strain carrying a GFP::3xFLAG tag at the unique N-terminus of LIN-29a (Aeschimann et al., 2019). These animals did not display any overt phenotypes. Since we could not create a LIN-29b-specific tag due to lack of isoform-specific sequence, we used a C-terminal fusion that tags both isoforms and mutated the lin-29a(∆) isoform in this background (Figure 2A). Hence, we studied lin-29b expression in a lin-29a(∆) mutant background. The resulting strain exhibited an increased occurrence of protruding vulva (Pvl) and vulval bursting phenotypes relative to lin-29a(∆) animals without the tag (data not shown) (Aeschimann et al., 2019), indicating that the C-terminal tag may partially impair LIN-29 protein function.

Using live imaging of the GFP-tagged proteins, we observed LIN-29 isoform accumulation. In agreement with immunofluorescence-based analysis (Bettinger et al., 1996), we detected signal in the epidermis (Figure 7B–D) and in other regions such as the head, the tail and the vulva (Figure 7—figure supplement 1). In most sites, patterns of the two LIN-29 isoforms appeared to differ. Considering our interest in the role of LIN-29 in the epidermal J/A transition, we focused on the detailed characterization of LIN-29a and LIN-29b accumulation in the skin. LIN-29b was first detectable in lateral seam cell nuclei of late L3 worms (Figure 7B). Subsequently, in mid-L4 stage animals, it also accumulated, weakly, in the major hypodermal syncytium hyp7 (Figure 7C). By contrast, LIN-29a was not detected in either seam cells or the major hypodermal syncytium hyp7 during the L3 stage (Figure 7B). Instead, LIN-29a accumulation began in both cell types in mid-L4 stage worms (Figure 7C). Finally, protein levels of both LIN-29a and LIN-29b peaked in lateral seam cells and hyp7 in late L4 stage animals and persisted into adulthood (Figure 7D, data not shown).

Taken together, our detailed analysis shows that lin-29a and lin-29b differ in their spatial as well as temporal expression patterns in the epidermis, likely explaining their distinct function in the four J/A transition events.

To formally test if the functional difference between the two lin-29 isoforms stems from their distinct expression patterns rather than molecular differences, we sought to alter the patterns experimentally. To this end, we expressed either lin-29a or lin-29b from a single-copy transgene under the control of the epidermal col-10 promoter, which is active from the first larval stage onwards. To avoid potential toxic effects of temporal misexpression of lin-29 in early larval stages, we added an auxin-inducible degron (AID) tag (Zhang et al., 2015) to the transgenic LIN-29 isoforms and prevented LIN-29 protein accumulation by addition of auxin to maintain the transgenic lines (Materials and methods). We then let animals hatch in the absence of auxin and plated starved, synchronized L1 stage larvae on food to permit LIN-29 protein accumulation, before we examined seam cell fusion, the J/A transition event that relies exclusively on LIN-29b. Strikingly, col-10-driven expression of either isoform induced precocious seam cell fusion. Specifically, we observed individual animals with fused seam cells already at 10 hr after plating at 25°C, that is during the L1 stage. At 15 hr after plating, in early L2, all animals had at least partially fused seam cells, with >90% of animals exhibiting complete fusion (Figure 8). By contrast, expression of mab-10 from the col-10 promoter did not cause any precocious seam cell fusion at this time (Figure 8). We conclude that LIN-29 upregulation alone is sufficient to trigger fusion of seam cells, and that upregulation of either LIN-29 isoform suffices. This data further validates that the LIN-29b-specific function observed in seam cell fusion under physiological conditions (Figure 3A,C) is not due to sequence differences between the two isoforms, but due to differences in their expression.

We wondered how the distinct expression patterns of lin-29a and lin-29b are generated. We have previously shown that lin-29a is translationally regulated by LIN-41 (Aeschimann et al., 2017). However, the regulation of lin-29b expression is poorly understood. We noticed that the alae phenotype seen in lin-29b mutant adults, small patches of well-formed alae, corresponds to that seen one stage earlier in precocious lin-41(0) mutant animals (Abrahante et al., 2003). By contrast, the weaker but more continuous alae seen in lin-29a(∆) mutant adults are reminiscent of the appearance of precociously formed alae in HBL-1-depleted L4 stage larvae (Abrahante et al., 2003). Hence, we wondered whether HBL-1, a transcription factor of the Hunchback/Ikaros family (Abrahante et al., 2003; Lin et al., 2003), and LIN-41 might act on distinct lin-29 isoforms.

To test this, we examined expression of lin-29 using the isoform-specific GFP-tags on LIN-29 at the early L3 stage, when neither LIN-29 isoform is expressed in the epidermis under physiological conditions. Consistent with earlier findings (Aeschimann et al., 2017), we found that LIN-41 depletion caused precocious accumulation of LIN-29a but not LIN-29b (Figure 9A,B). This upregulation was marked in hyp7 by 20 hr after plating but, with a delay, also occurred in seam cells (Figure 9—figure supplement 1). By contrast, early L3 stage animals exposed to hbl-1 RNAi displayed precocious accumulation of LIN-29b but not LIN-29a (Figure 9A,C), and this accumulation was more pronounced in the lateral seam cells than in hyp7 (Figure 9A). Moreover, it was detectable at the level of the lin-29b transcript (Figure 9—figure supplement 2). Finally, consistent with earlier evidence (Aeschimann et al., 2017), we observed precocious accumulation of a partially functional MAB-10::mCherry protein (Pereira et al., 2019) in L3 stage animals depleted of LIN-41 in both seam cells and hyp7, whereas depletion of HBL-1 had no effect (Figure 9D).

We conclude that HBL-1 but not LIN-41 regulates LIN-29b accumulation. Conversely, MAB-10 and LIN-29a accumulation are regulated by LIN-41 but not by HBL-1. Whether the regulation of LIN-29b by HBL-1 is direct remains to be determined.

The distinct function and regulation of lin-29 isoforms that we observed reveals that the heterochronic pathway is not linear but branches into two parallel arms, LIN-41–LIN-29a/MAB-10 and HBL-1–LIN-29b. We wondered where in the pathway this bifurcation occurs. We have previously shown that lin-41 is the only relevant target of the miRNA let-7 (Ecsedi et al., 2015; Aeschimann et al., 2017; Aeschimann et al., 2019), placing let-7 into only one arm of the pathway. The heterochronic gene immediately upstream of let-7 is lin-28, which directly inhibits let-7 biogenesis (Lehrbach et al., 2009; Van Wynsberghe et al., 2011). Interestingly, lin-28 was also reported to positively regulate hbl-1 (Vadla et al., 2012; Ilbay and Ambros, 2019), making it a good candidate for being immediately upstream of the pathway’s branching point. Hence, we analyzed precocious expression of the LIN-29 protein isoforms in a putative null mutant background of lin-28, lin-28(n719), at the early L3 stage. We observed that both isoforms accumulate precociously in lin-28 mutant animals (Figure 10, Figure 10—figure supplement 1). Accumulation is initially particularly strong in the seam but over time, and in particular for LIN-29a, also increases in hyp7. Consistent with acting mainly through HBL-1 and LIN-41, respectively, lin-29b but not lin-29a mRNA levels are increased in lin-28(n719) mutant relative to wild-type animals (Figure 10B,C). After 22 hr, when LIN-28 is also repressed in wild-type animals, comparable lin-29b mRNA levels result. Hence, these data support our hypothesis that heterochronic pathway bifurcation occurs downstream of lin-28.

Worm strains used in this study are listed in Key resources table. Bristol N2 was used as the wild-type strain. Animals were synchronized as described (Aeschimann et al., 2017) and, unless specified otherwise, grown on 2% NGM agar plates with Escherichia coli OP50 bacteria (Stiernagle, 2006). For microscopy, animals were mounted on a 2% (w/v) agarose pad and immobilized in 10 mM levamisol (Fluca Analytical, 31742).

Genome editing was performed as described (Katic et al., 2015) to obtain the following null mutant alleles using the sgRNAs listed in Supplementary file 1B. lin-29(xe116) is a 6724 bp deletion deletion of the putative lin-29b promoter region spanning the intron between exon 4 and exon 5 to disrupt lin-29b transcription. The deletion has the following flanking sequences: 5’ atacggtttcagaattaaggagaca – xe116 deletion – cctagatcaattgagctctaaagat 3’.

lin-29(xe120) is an insertion of an upstream ATG start codon, resulting in out-of-frame translation on the lin-29b mRNA. We introduced three point mutations in exon five to generate the following sequence: 5’ ttcgaacaaaagcc(g→a)ga(c→t)gt(g→c) 3’. This introduces an AUG start codon upstream of the normal lin-29b AUG in an improved Kozak context to increase the efficiency of aberrant translation initiation. All three mutations are silent with respect to the lin-29a ORF.

lin-29(xe200) is a 1187 nt deletion spanning exons 2–4 that results in translation of a LIN-29a protein lacking the first 119 amino acids. 23 N-terminal amino acids remain upstream of the LIN-29b N-terminus. The deletion has the following flanking sequences: 5’ ccaacttcttcaacgcaatg – 1187 nt deleted – ttttcacaatttggaagata 3’.

Injection of the same respective sets of sgRNAs into lin-29(xe61[lin-29::gfp::3xflag]) generated lin-29(xe61 xe114), lin-29(xe61 xe121), lin-29(xe61 xe133), which contain the respective mutations in the context of a GFP-3xFLAG-tagged LIN-29.

All final and intermediate plasmids as well as the primers used in these clonings are listed in Supplementary file 1A. The genomic regions containing coding exons (from the ATG start codon until the end of the 3’UTR) for mab-10, lin-29a and lin-29b were amplified by PCR using Phusion High-Fidelity DNA Polymerase (NEB, M0530S) from purified genomic wild-type C. elegans DNA. The lin-29a region was amplified in two fragments, the first spanning exons 2–4 (from ATG in exon 2), the second spanning exons 5–11 (including the 3’UTR). The large intron 4 was not amplified and exons 4 and 5 were fused into one exon when combining the two fragments during Gibson assembly. lin-29a 5’UTR was excluded as it is non-coding and confers LIN-41-mediated regulation (Aeschimann et al., 2017).

Using Gibson assembly, the PCR products were cloned into the BamHI site of pFA198. Plasmid pFA198 is a pENTR L1-L2 backbone plasmid that was constructed using Gateway cloning (BP reaction) to insert a 3xFLAG tag followed by a BamHI restriction site for subsequent Gibson assembly. Similarly, a FLAG-HA-degron tag followed by a BamHI restriction site was inserted into a pENTR L1-L2 backbone to obtain pFA218, a template for Gibson assembly reactions (Aeschimann et al., 2019). Plasmids with mab-10, lin-29a and lin-29b genomic regions cloned into pFA198 (i.e. pFA199, pFA206, pFA207) were only used as cloning intermediates in this study. From those plasmids, the mab-10, lin-29a and lin-29b genomic regions were re-amplified by PCR using Phusion High-Fidelity DNA Polymerase and inserted into BamHI-digested pFA218 using Gibson assembly. The resulting plasmids pFA219, pFA220 and pFA221 were then used for further Gateway cloning (LR reactions). In those LR reactions, transgenes were combined with the col-10 promoter upstream (Supplementary file 1A) and an operon linker with gfp-h2b::tbb-2_3'UTR (Merritt et al., 2008) downstream and cloned into the destination vector pCFJ150 (Frøkjær-Jensen et al., 2008). The resulting plasmids pFA238, pFA239 and pFA240 were used for injections.

After constructing the plasmid for single copy integration, worm lines with integrated transgenes were obtained by single-copy integration into chromosome II (ttTi5605 locus) or chromosome V (oxTi365), using a protocol for injection with low DNA concentration (Frøkjær-Jensen et al., 2012). In order to perform injections while inducing auxin-mediated degradation of proteins expressed from the introduced transgenes, the Mos1 insertion strains EG6699 and EG8082 were crossed to worms expressing TIR1 from the eft-3 promoter (IFM155) to generate strains HW2349 and HW2350, respectively. Following injections, these were grown on NGM plates supplemented with 1 mM auxin (Sigma-Aldrich, I2886). Injection mixes contained pCFJ601 (Peft-3::transposase) at 10 ng/ul, pGH8 (Prab-3::mCherry) at 10 ng/ul, pCFJ90 (Pmyo-2::mCherry) at 2.5 ng/ul, pCFJ104 (Pmyo-3::mCherry) at 5 ng/ul and the respective targeting vector at 50 ng/µl in water. Integrated transgenic lines were further crossed with worms expressing the ajm-1::mCherry marker.

Arrested L1 larvae were grown at 25°C for 36–38 hr (late L4 stage) or 40–42 hr (young adult stage), with the developmental time assessed by staging of individual worms according to gonad length and vulva morphology. Fluorescent and Differential Interference Contrast (DIC) images were acquired with a Zeiss Axio Observer Z1 microscope using the AxioVision SE64 software and Zen 2 (blue edition). Selections of regions and processing of images was performed with Fiji (Schindelin et al., 2012).

Seam cell quantification was performed by counting all clearly visible fluorescent cells expressing an scm::gfp transgene (Koh and Rothman, 2001) of the upper lateral side in mounted worms.

Seam cell fusion quantification was performed by counting seam cell junctions visible through expression of an ajm-1::mCherry transgene (Lehrbach et al., 2009).

Alae quantification was performed by observing alae structures by DIC with a 100x objective. For worm lines throwing mnC1-balanced progeny, Pmyo-2::GFP-positive (i.e. balancer carrying) animals were excluded from imaging and quantifications. To score let-7(xe150) homozygous animals within a population of balanced animals, all Pmyo-2::mCherry expressing (i.e. balancer carrying) animals were excluded from the analysis. To score lin-41(xe8) homozygous animals within a population of balanced lin-41(xe8)/lin-41(bch28 xe70) animals, all Peft-3::gfp::h2b expressing (i.e. balancer carrying) animals were excluded from the analysis.

Luciferase assays were performed as described (Meeuse et al., 2020). Briefly, embryos were extracted from gravid adults using a bleaching solution. Single embryos were transferred into a well of a white, flat-bottom, 384-well plate (Berthold Technologies, 32505) by pipetting. Animals were left to develop in 90 µL S-Basal medium containing E. coli OP50 (OD₆₀₀ = 0.9) and 100 μM Firefly D-Luciferin (p.j.k., 102111). Plates were sealed with Breathe Easier sealing membrane (Diversified Biotech, BERM-2000). Luminescence was measured using a luminometer (Berthold Technologies, Centro XS3 LB 960) every 10 min for 0.5 s for 90 or 100 hr in a temperature controlled incubator set to 20°C. Luminescence data was analyzed using an automated algorithm to detect the hatch and the molts (Meeuse et al., 2020).

Before acquiring images of representative worms, the GFP signals for at least 10 worms were observed to verify that they were comparable among different worms in each worm line and for each condition. For detection of endogenously tagged LIN-29 (HW1826, HW1882, HW1835) by confocal microscopy, animals were grown at 25°C. For confocal imaging of endogenously tagged LIN-29 (HW1826, HW1882, HW1835) and MAB-10 (HW2047) on RNAi conditions, synchronized arrested L1 stage larvae were grown for 20 or 22 hr at 25°C on RNAi-inducing plates with HT115 bacteria (Ahringer, 2006; Timmons et al., 2001) containing the L4440 RNAi vector either without insert (‘mock RNAi’) or with an insert targeting lin-41 (Fraser et al., 2000) or hbl-1 (Kamath et al., 2003). Worms were imaged on a Zeiss LSM 700 confocal microscope driven by Zen 2012 Software. Differential Interference Contrast (DIC) and fluorescent images were acquired with a 40x/1.3 oil immersion objective (1024 × 1024 pixels, pixel size 156 nm).

For confocal imaging of endogenously tagged LIN-29 in the wild-type and lin-28(n719) mutant background (HW1826, HW1882, HW1835, HW1924, HW1925, HW1926), worms were grown at 25°C for 20, 22, 24 and 22, 24, 26 hr, respectively (to accommodate a modest delay in development of lin-28(n719) mutants). For confocal imaging of endogenously tagged LIN-29a in the wild-type and lin-29a(∆N) background (HW1826, HW2408), worms were grown at 25°C. In both cases, worms were imaged on an Axio Imager M2 (upright microscope) + Yokogawa CSU W1 Dual camera T2 spinning disk confocal scanning unit driven by Visiview 3.1.0.3. Differential Interference Contrast (DIC) and fluorescent images were acquired with a 40x/1.3 oil immersion objective (2048 × 2048 pixels, 16-bits). Using the Fiji software (Schindelin et al., 2012), images were processed after selecting representative regions. Worms of the same worm line were imaged and processed with identical settings.

All strains were grown on OP50-seeded NGM plates supplemented with 1 mM auxin. Gravid animals were bleached and eggs were incubated at room temperature in M9 for hatching overnight. To allow accumulation of LIN-29a, LIN-29b or MAB-10, the animals were not exposed to auxin anymore from this point onward. Synchronized L1 animals were plated and grown at 25°C. Seam cell fusion was scored by observation of the ajm-1::mCherry marker at 15 hr after plating synchronized L1 animals. Fluorescent and Differential Interference Contrast (DIC) images were acquired with a 40x/2.5 oil immersion objective with a Zeiss Axio Observer Z1 microscope using the AxioVision SE64 software and Zen 2 (blue edition). Selections of regions and processing of images was performed with Fiji (Schindelin et al., 2012).

Reverse transcription was performed with the ImpromII Reverse Transcription System (Promega; A3800), according to the manufacturer’s protocol, with 100 ng RNA (for lin-41 and hbl-1 samples) or 80 ng RNA (for lin-28(n719) samples) and random primers (Promega; C1181). Using Power Up SYBR Green Master Mix (Thermo Fisher Scientific;A25742), qPCR was performed on a Light Cycler 480II 384 (Roche) with the primers listed in the Supplementary file 1B. For comparing mRNA levels of lin-29b of worms grown on mock, lin-41 or hbl-1 RNAi bacteria, -ΔΔCt were calculated using act-1 as an internal control mRNA and the mock RNAi condition (first time point) as calibrator. For comparing mRNA levels of lin-29a and lin-29b, of wild-type or lin-28(n719) worms, -ΔΔCt were calculated using act-1 as an internal control mRNA and the wild-type condition (first time point) as calibrator.

Animals were grown for 20 hr at 25°C on RNAi-inducing plates, as described above for confocal imaging, or on NA22 plates (Evans, 2006). Lysates were made by boiling (5 min, 95°C) and sonication in SDS lysis buffer (63 mM Tris-HCl (pH 6.8), 5 mM DTT, 2% SDS, 5% sucrose) and cleared by centrifugation, before separating proteins by SDS-PAGE (loading: 50 µg protein extract per well) and transferring them to PVDF membranes by semi-dry blotting. The following antibodies were used: Monoclonal mouse anti-FLAG M2-Peroxidase (HRP) (Sigma-Aldrich; A8592, dilution: 1:1000). Monoclonal mouse anti-Actin clone C4 (Millipore; MAB1501, dilution 1:10000). Detection was performed with a horseradish peroxidase-conjugated secondary anti-mouse antibody (NXA931), ECL Western Blotting Detection Reagents and an ImageQuant LAS 4000 chemiluminescence imager (all from GE Healthcare).

All values of n indicate numbers of animals. No statistical tests were performed.