MAB-10/NAB acts with LIN-29/EGR to regulate terminal differentiation and the transition from larva to adult in C. elegans

C. elegans heterochronic genes have been characterized based in large part upon their affects in regulating the biology of lateral hypodermal cells known as seam cells. Seam cells undergo a stem-cell-like pattern of asymmetric cell division during each larval stage (Sulston and Horvitz, 1977). At the end of each larval stage, the hypoderm generates a new cuticle and the animal sheds its old cuticle. After the final (L4) larval stage, the hypoderm undergoes a process of terminal differentiation that comprises four events: (1) synthesis of the adult-specific cuticle; (2) exit from the molting cycle; (3) seam cell fusion; and (4) seam cell exit from the cell cycle. Terminal differentiation is initiated via the downregulation of the Hunchback/Ikaros homolog hbl-1 and the TRIM-NHL gene lin-41 by the let-7 family of miRNAs. This downregulation triggers the activity of the Kruppel family zinc-finger protein LIN-29, which promotes all four aspects of terminal differentiation (Fig. 1B) (Reinhart et al., 2000; Abrahante et al., 2003; Lin et al., 2003). Seam cell fusion, exit from the cell cycle, and synthesis of the adult-specific cuticle occur during the L4 stage, but despite the presence of LIN-29 activity, exit from the molting cycle does not occur until after the animal becomes an adult. This observation suggests that exit from the molting cycle is not solely dependent on the presence of LIN-29 activity and that there might be other factors that work with LIN-29 to ensure its proper regulation.

C. elegans was grown as described previously (Brenner, 1974) and maintained at 20°C unless otherwise noted. N2 was the wild-type strain. The mutations used in this study were: LGI, smg-1(e1228) (Hodgkin et al., 1989), lin-28(n719) (Ambros and Horvitz, 1984), lin-41(n2914) (Slack et al., 2000); LGII, dpy-10(e128) (Brenner, 1974), mab-10(e1248) (Hodgkin, 1983), mab-10(n5117), mab-10(n5118), mab-10(tm2497), lin-29(n546) (Ambros and Horvitz, 1984), lin-29(n836) (Papp et al., 1991) and rol-1(e91) (Brenner, 1974); LGIV, him-8(e1489) (Hodgkin et al., 1979); LGV, him-5(e1467) (Hodgkin et al., 1979).

Information about tm2497 (kindly provided by S. Mitani, Tokyo Women's Medical University, Japan) can be found at www.wormbase.org. The following balancer chromosomes were used: hT2 [qIs48] LGI; LGIII, mIn1 [mIs14] LGII and mnC1 LGII (Edgley et al., 2006).

The F2 progeny of mutagenized him-8(e1489) hermaphrodites were screened clonally using a dissecting microscope for adult males with an extra cuticle. We screened about 4500 genomes and isolated two mutants (n5117 and n5118), both of which failed to complement mab-10(e1248).

Cuticles were observed using a Zeiss Axioskop 2 with Nomarski optics. Seam cell fusion was assayed in early L4 and young adult animals using the ajm-1::gfp (wIs78) (Koh and Rothman, 2001) reporter. Adult molting was assayed by picking L4 males to plates and scoring 24 hours later for the presence of an extra cuticle. Synchronized wild-type and mab-10(e1248) males and hermaphrodites were grown until the mab-10 mutants had synthesized an extra cuticle. These animals were then fixed and electron microscopic analysis was performed as described previously. Seam cell divisions were followed using either Nomarski microscopy or fluorescence microscopy to observe scm::gfp (wIs78) (Koh and Rothman, 2001) or col-19::gfp (maIs105), as noted in figure legends. To assay adult seam cells, we picked late L4 animals to fresh plates and scored 24 hours later.

pDH04 contains about 9 kb of the mab-10 genomic locus on LG II from 10,525,019 to 10,534,380 with gfp inserted in place of the stop codon. pDH36 contains about 18 kb of the lin-29 genomic locus on LG II from 11,917,276 to 11,936,077 with mCherry inserted in place of the stop codon. mab-10 cDNA was isolated by RT-PCR from a mixed-stage RNA preparation and cloned into the PCR8GW Gateway entry vector creating pDH08. pDH23 contains the mab-10 cDNA fused downstream of gst and was created by the Gateway LR reaction between pDH08 and pDestGST2TK. Full-length lin-29 cDNA was amplified from the plasmid AAC37255 (Open Biosystems) and cloned into the PCR8GW entry vector, creating pDH06. pDH24 contains the lin-29 cDNA fused downstream of 6XHis tag and was created by the Gateway LR reaction between pDH06 and pDest17. All plasmids containing lin-29 deletion variants were generated by site-directed mutagenesis of pDH24.

The oligos used were: pDH06 (5′-atggatcaaactgttctagattcggc-3′ and 5′-ttaataggaatgatttttcatattat-3′; pDH08 (5′-atgtcatcatcgtcgtcgtcgtcgtta-3′ and 5′-tcaagattccgggagctcacccttcattttttcgattatcgccgcccat-3′; pDH25 (LIN-29 Del 1-144) (5′-aaagcaggctccgaattcgcccttatgcagatgcgggaagcaaaaccttacaagt-3′ and 5′-acttgtaaggttttgcttcccgcatctgcataagggcgaattcggagcctgcttt-3′); pDH26 (LIN-29 Del 147-300) (5′-caaaattgctggccttcgatcgatccgctgcttcccgcatctgcatctgttgctg-3′ and 5′-cagcaacagatgcagatgcgggaagcacaaactgataagccattcaaatgtaac-3′); pDH27 (LIN-29 Del 301-459) (5′-catgacaaagcacgcggatcgatcgtaaaagggcgaattcgacccagctttc-3′ and 5′-gaaagctgggtcgaattcgcccttttacgatcgatccgcgtgctttgtcatg-3′); pDH29 (LIN-29 Del 371-389) (5′-cagtggggttatcatgttgaacccaggaatattaaagagctgactgaacgcact-3′ and 5′-agtgcgttcagtcagctctttaatattcctgggttcaacatgataaccccactg-3′); pDH30 (LIN-29 Del 390-406) (5′-acgagtacgaagaatggagaacgagcttcatcttcctcggccacggctgtcgta-3′ and 5′-tacgacagccgtggccgaggaagatgaagctcgttctccattcttcgtactcgt-3′); pDH31 (LIN-29 DEL 404-440) (5′-ctggagaacatccaacgctacaacgggcagggaggcgtgttcaacccacaatca-3′ and 5′-tgattgtgggttgaacacgcctccctgcccgttgtagcgttggatgttctccag-3′); pDH32 (LIN-29 441-459) (5′-tcctcgtcagcaggttcgtcctcaagttacccagctttcttgtacaaagtggtt-3′ and 5′-aaccactttgtacaagaaagctgggtaacttgaggacgaacctgctgacgagga-3′); pDH37 (LIN-29 Del 147-232) (5′-cagcaacagatgcagatgcgggaagcacaaactgataagccattcaaatgtaac-3′ and 5′-gttacatttgaatggcttatcagtttgtgcttcccgcatctgcatctgttgctg-3′); pDH38 (LIN-29 Del 233-300) (5′-aatctccaatctcactctcgatgccatgcggatcgatcgaaggccagcaat-3′ and 5′-attgctggccttcgatcgatccgcatggcatcgagagtgagattggagatt-3′).

The integrated arrays wIs78 (Koh and Rothman, 2001), which contains scm::gfp and ajm-1::gfp, and maIs105 (Feinbaum and Ambros, 1999), which contains col-19::gfp, were used to assay seam cell fusion and seam cell number. The mab-10::gfp array (nEx1655) was formed by the co-injection of pDH04 at 35 ng/μl and pSN359 [pgp-12::gfp (Zhao et al., 2005)] at 40 ng/μl. The lin-29::mCherry array (nEx1681) was formed by the injection of PCR product (50 ng/μl) containing LG II sequence from 11917298 to 11927996 from template pDH36 with ttx-3::gfp (Hobert et al., 1997) at 40 ng/μl. nEx1681 was integrated into the genome using γ irradiation to form nIs408.

GST pull-downs were performed as previously described (Reddien and Horvitz, 2000) and quantified using a Typhoon phosphorimager (GE Healthcare). In short, ³⁵S-methionine-labeled LIN-29 protein was transcribed and translated in vitro using a TNT Quick Coupled Transcription/Translation Kit (Promega) and bound to MAB-10::GST fusion protein purified from E. coli. Binding efficiency is expressed as the percent of wild-type binding after correcting for binding to GST alone.

Single-molecule FISH (fluorescence in situ hybridization) was performed based on protocols described previously (Raj et al., 2008) and at www.singlemoleculefish.com. Briefly, synchronized populations of N2 animals were grown on NGM agar plates seeded with OP50. Animals were harvested by the addition of 2-3 ml of M9 solution and subsequent collection in 15 ml conical tubes. Animals were washed twice with PBS and transferred to a 1.5 ml microfuge tube. This procedure yielded approximately 35 μl of packed animals per large NGM plate. PBS was removed and replaced with 1 ml of fixation solution (4% formaldehyde in PBS). Animals were fixed for 45 minutes and washed twice in PBS. After the second wash, the PBS was removed and replaced with 70% ethanol. Animals were kept in 70% ethanol for at least 1 night.

Hybridizations were performed essentially as described previously (Raj et al., 2008) and at www.singlemoleculefish.com. For mab-10 and cki-1 probes conjugated to Cy5, a probe dilution of 1/10,000 from the original stock solution was used. For nhr-23 and nhr-25 probes conjugated to Cy5, a probe dilution of 1/5000 from the original stock solution was used. Probes for mab-10, nhr-23, nhr-25 and cki-1 were designed following the guidelines specified at www.singlemoleculefish.com and purchased from Biosearch Technologies (Novato, CA).

Images were filtered using MATLAB to enhance automated transcript identification as previously reported (Raj et al., 2008). To quantify transcript numbers within the hypoderm specifically, we used MATLAB to manually define a region-of-interest around the hypoderm in three dimensions using DAPI staining as a guide. The lateral hypoderm is readily identifiable based on its morphology and position within the animal. Whole-animal hypoderm counts are the summation of five to eight individual partially overlapping stacks. Care was taken when selecting a region-of-interest to minimize double counting of transcripts within image overlaps.

RNAi by feeding was performed as reported previously (Ashrafi et al., 2003). We used RNAi constructs from the ORFeome library for mab-10, lin-29 and cki-1. For RNAi of mab-10, L4 him-8 animals were picked individually to seeded RNAi plates containing 1 mm IPTG. Animals were moved to another RNAi plate 72 hours later. Progeny from both plates were scored for extra molts. For RNAi of lin-29, L4 him-5 and L4 wIs78-containing animals were picked individually to seeded RNAi plates containing either 1 nm or 1 μM IPTG. Animals were transferred to new plates as described above and progeny were scored for extra molts, extra seam cell divisions, adult alae and seam cell fusion. We performed postembryonic RNAi of cki-1 by picking multiple late L1/L2 wIs78-containing animals, which hatched on OP50, onto seeded RNAi plates containing 1 mm IPTG. These animals were scored for extra molts, extra seam cell divisions, adult alae and seam cell fusion.

To identify new genes required for the terminal differentiation of the hypoderm, we performed a screen for mutants that failed to exit the molting cycle and inappropriately initiated an adult molt. Because extra molts are difficult to observe in adult hermaphrodites (as described below), we focused on the identification of mutant males. We identified two mutations, n5117 and n5118, both of which failed to complement the previously isolated mutation mab-10(e1248). mab-10(e1248) was originally identified in a screen for morphologically abnormal males (male abnormal) (Hodgkin, 1983), and mab-10 males were later observed to undergo an extra molt (C. Link, personal communication).

We found that mab-10 mutants synthesized an adult cuticle appropriately at the end of the fourth larval stage but subsequently generated a second adult cuticle. All mab-10 males underwent an extra molt, whereas not all mab-10 hermaphrodites underwent the extra molt (Fig. 1C-F). Adult mab-10 hermaphrodites that had begun to synthesize the second adult cuticle were typically consumed by internally hatched progeny, making the extra molt difficult to observe.

Seam cells fused normally at the larval-to-adult transition, although the fused seam occasionally appeared wider than that of the wild type (Fig. 1G-J), and the seam cell lineage during larval development was normal (data not shown). However, approximately five of the 11 V lineage-derived seam cell nuclei underwent an extra division by 24 hours after the larval-to-adult transition (Fig. 1K,L) (Table 1). The timing of the extra divisions ranged from ∼14 to 20 hours post-adult, and seam cells occasionally divided twice within this interval. This result indicates that mab-10 is required to promote seam cell exit from the cell cycle. Although present within a syncytium, the different seam cell nuclei were not equally likely to undergo an extra division (Fig. 1M).

Although the mab-10 mutant phenotype was originally thought to be male-specific, we have shown that mab-10 males and hermaphrodites undergo extra molts and the seam cells of mutant males and hermaphrodites continue to divide after the larval-to-adult transition. Thus, mab-10, like lin-29, is required to prevent extra molts and extra seam cell divisions and both mab-10 and lin-29 have a retarded heterochronic defect. However, mab-10, unlike lin-29, is not required for adult-specific cuticle synthesis or seam cell fusion.

lin-28 and lin-41 mutants exhibit adult characteristics one or sometimes two stages prematurely. In the case of lin-28, an L2-specific developmental program is skipped, resulting in a relatively complete transition to adult prematurely. Early development of lin-41 animals proceeds normally through the L3 stage but animals then prematurely display adult characteristics, including patches of adult alae and seam cells that have exited the cell cycle.

lin-28; mab-10 animals, like lin-28 animals, skip the L2-specific developmental program and take on adult characteristics prematurely, including the formation of complete adult alae (see Table S1 in the supplementary material); however, the seam cells continue to divide in a mab-10-like fashion (Table 1). This result suggests that mab-10 acts downstream of or parallel to lin-28 to control seam cell differentiation.

Similarly, seam cells of lin-41; mab-10 animals stopped dividing at the end of the L3 stage, but then divided again in a mab-10-like fashion (Table 1). Surprisingly, far fewer lin-41; mab-10 animals displayed precocious patches of adult alae when compared with lin-41 animals alone and the patches of those lin-41; mab-10 animals that had precocious alae were far smaller than those in lin-41 single mutants (see Table S1 in the supplementary material). This suppression demonstrates that mab-10 has the capacity to promote adult alae formation despite not being required for normal adult alae formation. Although mab-10 mutants made normal adult alae, we did observe a modest decrease in the expression of col-19, an adult-specific cuticle collagen (see Fig. S1A-C in the supplementary material).

These data indicate that mab-10 likely functions downstream of or parallel to both lin-28 and lin-41 and that MAB-10 can promote aspects of the larval-to-adult transition beyond exit from the cell cycle and exit from the molting cycle.

Using a combination of SNP mapping, genomic rescue and DNA sequence determination, we determined that mab-10 is the gene R166.1(NM_063867.1) (Fig. 2A; see Fig. S2 in the supplementary material) and encodes a protein similar to NAB transcriptional co-factors (Fig. 2B,C). The NAB-specific NCD1 (Nab conserved domain) domain is thought to be required for physical interactions with the R1 (repressor 1) domains of the mammalian immediate early genes EGR-1, EGR2 and EGR-3 (Russo et al., 1995), and the NCD2 domain has been shown to function both as a transcriptional activator and as a repressor (Swirnoff et al., 1998; Sevetson et al., 2000). The two strongest alleles, mab-10(n5117) and mab-10(n5118), are both early nonsense mutations located within the NCD1 domain and are probably nulls. The weaker allele, mab-10(e1248), causes a serine-to-phenylalanine amino acid substitution within the NCD2 domain (Fig. 2C).

Mammalian NAB proteins act as co-factors for EGR proteins to transcriptionally activate or repress genes required to regulate the terminal differentiation and function of several cell types: keratinocytes, chondrocytes, Schwann cells and macrophages (Swirnoff et al., 1998; Topilko et al., 1998; Sevetson et al., 2000; Le et al., 2005; Laslo et al., 2006). In the gonadotrope lineage of the anterior pituitary gland, NAB proteins act with EGR1, in conjunction with steroidogenic factor 1 (SF1), to promote transcription of the β-subunit of luteinizing hormone and the onset of puberty (Dorn et al., 1999; Wolfe and Call, 1999).

Given the similar phenotypes of mab-10 and lin-29 mutants, we hypothesized that MAB-10 acts with LIN-29 to promote the late-occurring aspects of terminal differentiation during the larval-to-adult transition.

To determine whether MAB-10 and LIN-29 are co-expressed, we generated rescuing MAB-10::GFP and LIN-29::mCHERRY fusion proteins and performed single-molecule FISH of endogenous mab-10 transcripts. Expression of LIN-29::mCHERRY largely resembled the previously published pattern of LIN-29 expression (Bettinger et al., 1996). MAB-10 and LIN-29 were co-expressed in multiple cells, including specific pharyngeal cells throughout development (Fig. 3A-C), vulval precursor cells during the third larval stage (Fig. 3D-F), and the seam cells and hypodermal nuclei derived from the seam cell lineages throughout the L4 stage and adulthood (Fig. 3G-I).

Because LIN-29 is the master regulator of the larval-to-adult transition, we wanted to determine whether mab-10 expression was dependent on LIN-29. We expressed MAB-10::GFP in lin-29(n836) mutants and found that the pattern of MAB-10::GFP expression was similar to the wild type with the notable exception of the seam cells. During the L4 stage, MAB-10::GFP was present within the nuclei of the syncytial hypoderm but was not visible within the seam cells (Fig. 3J-L).

We also expressed LIN-29::mCHERRY in mab-10 mutants and found that removal of mab-10 had no effect on LIN-29::mCHERRY accumulation (Fig. 3M-O). These results demonstrate that mab-10 probably does not promote lin-29 activity by regulating lin-29 expression and suggest that either mab-10 transcription is controlled by LIN-29 specifically in the seam cells or that LIN-29 promotes MAB-10::GFP nuclear accumulation post-transcriptionally.

To determine whether LIN-29 promotes mab-10 transcription in the seam cells, we performed single-molecule FISH (Raj et al., 2008) to detect endogenous mab-10 transcripts in wild-type and lin-29(n836) animals.

We observed during the L1 stage mab-10 transcripts in the anterior and posterior bulbs of the pharynx as well as a small number of transcripts throughout the hypoderm (Fig. 4A). L2-stage animals showed increased expression of mab-10 mRNA in hyp7 and the rectal epithelium in addition to the nerve ring and the ventral nerve cord, where several cells contained one or two transcripts (Fig. 4B). During the early L3 stage, mab-10 was weakly expressed in hyp7 but showed high expression in the vulval and uterine precursor cells (Fig. 4C). mab-10 mRNA was also present within the distal tip cells and a pair of bilaterally symmetric cells that we believe to be the CAN neurons, based on their position and neuronal nuclear morphology. The CAN neurons are associated with the excretory canal and are thought to regulate excretory function (Sulston et al., 1983; Nelson and Riddle, 1984; Forrester et al., 1998).

By the late L3 stage, mab-10 expression was almost absent from hyp7 but was high in the seam cells, distal tip cells and developing vulva (Fig. 4D). mab-10 transcript was also detected in the gonadal sheath cells. During the L4 stage, mab-10 transcripts dramatically increased in abundance throughout the pharynx, hypoderm and somatic gonad, including the distal tip cells (Fig. 4E).

We found no change in the level or pattern of mab-10 transcription in a lin-29(n836) mutant background compared with the wild type (Fig. 4F), suggesting that mab-10 is not a transcriptional target of LIN-29 and that the absence of MAB-10::GFP from the nuclei of L4 seam cells reflects a post-transcriptional regulation of MAB-10 by LIN-29.

Using in vitro pull-down assays, we found that MAB-10::GST interacted with in vitro-translated LIN-29 but did not interact with luciferase (Fig. 5A; see Fig. S3A in the supplementary material). A similar interaction between Drosophila NAB and the two Drosophila LIN-29 homologs RN and SQZ was reported recently (Terriente Felix et al., 2007). Deletion of a 17 amino acid sequence (390-406) (Fig. 5B; see Fig. S3B in the supplementary material) within the LIN-29 C terminus decreased MAB-10 binding by about 80% (Fig. 5C, lane 3). Alignment of LIN-29 with the closest homologs from the distantly related nematodes Brugia malayi and Pristionchus pacificus revealed significant identity within only the zinc-finger domains and the region of LIN-29 important for interaction with MAB-10 (see Fig. S4 in the supplementary material). Strikingly, alignment of this 17 amino acid LIN-29 sequence with the R1 domains of mammalian EGR1, EGR2 and EGR3 revealed a conserved motif (Fig. 5D). Furthermore, and unexpectedly, we identified the same conserved region within the domains of Drosophila RN and SQZ previously shown to bind to Drosophila NAB, but not previously known to be similar to EGR proteins. Mutation of the completely conserved isoleucine (268) to asparagine in this region of EGR2 causes a recessive form of the hypomyelinating neuropathy Charcot-Marie tooth disease in humans (Warner et al., 1998) and eliminates NAB/EGR binding in vitro (Warner et al., 1999). Based on these observations, we conclude that LIN-29, RN, SQZ and the mammalian EGR proteins share a conserved NAB-interacting domain.

To determine whether the LIN-29 C terminus is required in vivo for lin-29 function, we attempted to generate transgenic animals expressing an engineered form of LIN-29 missing the C terminus or specifically missing the putative MAB-10 interaction domain. These attempts were unsuccessful. As an alternative, we used lin-29(n546), an allele that carries a nonsense mutation immediately following the last zinc-finger domain (Fig. 5E). In an otherwise wild-type background, lin-29(n546) animals are completely defective in all four aspects of hypodermal terminal differentiation. However, some lin-29 activity can be restored to lin-29(n546) mutants by disrupting nonsense-mediated mRNA decay using a mutation in a smg (suppressor with morphological effect on genitalia) gene (Hodgkin et al., 1989). The lack of nonsense-mediated decay presumably stabilizes the lin-29(n546) mRNA and allows for the translation of a LIN-29 product missing the C terminus. An antibody raised against the LIN-29 C terminus was unable to detect LIN-29 protein (A. Rougvie, personal communication), suggesting that the restoration of LIN-29 activity was the result of the production of truncated protein and not the result of the production of full-length LIN-29 protein via read-through of the nonsense mutation.

We characterized the extent of terminal differentiation in smg-1(e1228); lin-29(n546) double mutant animals and found them to be normal with respect to seam cell fusion, adult cuticle synthesis and exit from the cell cycle. However, 93% of smg-1(e1228); lin-29(n546) males failed to exit the molting cycle and generated an extra adult cuticle (Fig. 5F,G) (Table 1), indicating that the LIN-29 C terminus is required for these events.

Although the absence of the LIN-29 C terminus causes extra molts, it is not clear whether the extra molts are the result of decreased MAB-10 binding, or of a MAB-10-independent general decrease in lin-29 function. To address this, we partially decreased lin-29 function using RNAi. We found that a partial reduction of lin-29 function could induce an extra molt in males and inappropriate adult seam cell divisions in hermaphrodites, while having little to no effect on adult alae formation (see Fig. S5A-C in the supplementary material). This result demonstrates that a reduction of general lin-29 activity is sufficient to phenocopy mab-10 mutants and that adult alae synthesis, seam cell fusion, seam cell exit from the cell cycle and exit from the molting cycle probably require different levels of LIN-29 activity. These results, combined with our observation that mab-10 promotes precocious adult alae formation in lin-41 mutants without being required for normal adult alae formation, raises the possibility that MAB-10 promotes general LIN-29 activity and does not act specifically on target genes controlling molting cycle exit and seam cell exit from the cell cycle.

Previous reports implicated the nuclear hormone receptors NHR-23 (nuclear hormone receptor) and NHR-25/SF1 as the primary positive regulators of molting (Kostrouchova et al., 1998; Asahina et al., 2000; Gissendanner and Sluder, 2000; Frand et al., 2005; Hayes et al., 2006). Both nhr-23 and nhr-25 are expressed at high levels in the hypoderm, and it has been proposed that LIN-29 represses their expression during adulthood, preventing the execution of further molts (Hayes et al., 2006). We performed genetic mosaic analysis, the result of which suggested that mab-10 functions in the hypoderm to prevent extra molts (see Fig. S6 in the supplementary material). We used single molecule FISH to quantify the endogenous expression of nhr-23 and nhr-25 in the hypoderm of wild-type, lin-29 and mab-10 adults (Raj et al., 2008). As expected, we detected a low number of nhr-23 and nhr-25 transcripts in the hypoderm of adult wild-type hermaphrodites and males (Fig. 6A-D). We found no significant increase of nhr-23 expression relative to the wild type (Fig. 6E,J) in adult mab-10 hermaphrodites. However, in adult mab-10 males, nhr-23 expression was increased 19-fold (Fig. 6F,J). We found a near threefold increase in expression of nhr-25 in mab-10 hermaphrodites (Fig. 6G,L) and an 11-fold increase in mab-10 males (Fig. 6H,L). The difference in nhr-23 and nhr-25 expression between mab-10 hermaphrodites and males correlates with the difference in penetrance of the extra molt defect between the sexes. By comparison, we found that nhr-23 and nhr-25 were upregulated about 12-fold relative to the wild type in adult-aged lin-29 hermaphrodites (Fig. 6I-L). These results are consistent with the hypothesis that MAB-10 acts with LIN-29 to prevent extra molts by repressing the expression of nhr-23 and nhr-25 in the adult hypoderm.

During the L4 stage, LIN-29 upregulates the cell cycle inhibitor cki-1 (cyclin-dependent kinase inhibitor) in seam cells (Hong et al., 1998). cki-1 and its homolog p27 promote terminal differentiation in C. elegans and mammals, respectively (McArthur et al., 2002; Fukuyama et al., 2003; Thomas et al., 2004; Fujita et al., 2007). We hypothesized that the extra seam cell divisions in mab-10 mutants might be the result of reduction of cki-1 expression during the larval-to-adult transition. RNAi of cki-1 by injection can induce extra larval seam cell divisions (Hong et al., 1998). We found that postembryonic RNAi of cki-1 by feeding was sufficient to induce inappropriate adult seam cell divisions (Table 1; see Fig. S6D in the supplementary material) without causing extra larval divisions. We used single-molecule FISH to detect endogenous cki-1 transcripts in wild-type, mab-10 and lin-29 animals.

We observed a pulse of cki-1 expression in all seam cells following normal seam cell divisions in wild-type, mab-10 and lin-29 strains. In wild-type animals and mab-10 mutants, a second pulse of cki-1 expression occurred late during the L4 stage after the seam cells had fused. This L4-specific increase was not observed in lin-29 animals, although there was a pulse of cki-1 expression associated with the extra seam cell divisions. In wild-type and mab-10 animals, L4 expression was highest in regions of the seam containing those seam cell nuclei least likely to undergo extra divisions in mab-10 mutants: those at the anterior and posterior ends of the animal, as well as those directly dorsal to the vulva (Fig. 6M,N). The total number of cki-1 transcripts within the seam of mab-10 animals (556±70) was on average two-thirds of that found in wild-type animals (863±166). In lin-29 animals, the seam cells underwent an extra cell division and displayed the typical pattern of seam cell cki-1 expression observed following a normal cell division. These observations are consistent with the hypothesis that MAB-10 acts with LIN-29 to prevent extra seam cell divisions by promoting the expression of cki-1.

We have identified mab-10 as a new heterochronic gene that is required for specific aspects of the larval-to-adult transition, specifically molting cycle exit and seam cell exit from the cell cycle. We cloned mab-10 and found that it encodes the only C. elegans NAB transcriptional co-factor. NAB proteins are thought to physically interact with Kruppel family EGR transcription factors to regulate their activity.

Previous work demonstrated that MAB-10 (then known only as the C. elegans NAB protein R166.1) could interact with mammalian EGR proteins in a yeast two-hybrid assay; no corresponding C. elegans EGR protein was identified (Svaren et al., 1996). We demonstrate that MAB-10 interacts with the terminal differentiation factor LIN-29 through an evolutionarily conserved NAB binding domain (R1 domain) and that MAB-10 is required for a subset of LIN-29-dependent activities. Our work identifies LIN-29 as a C. elegans EGR-like protein and demonstrates that the C. elegans heterochronic pathway controls the timing of NAB/EGR-mediated differentiation.

Several experiments using mammalian tissue culture suggest that NAB proteins negatively regulate EGR activity by binding EGR proteins at specific target genes and preventing EGR-mediated transcription (Russo et al., 1995; Svaren et al., 1996). However, loss of either EGR2 function or NAB function in mice and humans results in hypomyelination, suggesting that EGR and NAB proteins need not act antagonistically in vivo.

We found that in C. elegans, MAB-10 and LIN-29 both act to promote terminal differentiation and the onset of adulthood. Furthermore, we showed that mab-10 promotes the formation of precocious adult alae in a lin-41 mutant background, suggesting that MAB-10 does not specifically act to control genes required for exit from the molting cycle and seam cell exit from the cell cycle, but more likely acts as a general enhancer of LIN-29 activity.

EGR and NAB proteins have been shown to operate in a negative-feedback loop wherein an EGR protein promotes the expression of its NAB co-factor, which then inhibits EGR activity (Kumbrink et al., 2005). We found that mab-10 transcription does not depend on LIN-29, despite a dramatic increase of mab-10 transcription during the L4 stage. Thus, mab-10 is not a transcriptional target of LIN-29.

Whereas mab-10 is not a transcriptional target of LIN-29, MAB-10::GFP localization to seam cell nuclei during the L4 stage required LIN-29, indicating that LIN-29 might promote MAB-10 seam cell nuclear localization via a post-transcriptional mechanism or via direct physical interaction.

Our work demonstrates that MAB-10 and LIN-29 do not operate in a negative-feedback loop. We propose that other components of the heterochronic pathway directly regulate mab-10 transcription to temporally regulate MAB-10/LIN-29 activity and that LIN-29 or some factor downstream of LIN-29 controls MAB-10/LIN-29 activity by promoting the accumulation of MAB-10 in seam cell nuclei.

By showing that MAB-10 acts with LIN-29 through an evolutionarily conserved EGR R1 domain, we identify LIN-29 and the Drosophila LIN-29 homologs RN and SQZ as EGR-like molecules. We propose that NAB proteins and EGR proteins act together in temporal developmental programs to control terminal differentiation. In Drosophila, the LIN-29 homolog SQZ acts with Drosophila NAB to control neuroblast differentiation (Baumgardt et al., 2009). We showed that in C. elegans, LIN-29 and MAB-10 act together to control the differentiation of a hypodermal stem cell lineage during the transition from larva to adult by regulating the expression of the nuclear hormone receptors nhr-23 and nhr-25 and the cell cycle regulator cki-1 (Fig. 7A). Recently, a study of C. elegans demonstrated that nhr-25 is itself a heterochronic gene and possibly functions with lin-29 to promote aspects of the larval-to-adult transition, including seam cell exit from the cell cycle (Hada et al., 2010). Though the mechanism by which nhr-25 regulates seam cell exit from the cell cycle is not known, we speculate that LIN-29 and NHR-25 might act together to promote cki-1 expression (Fig. 7A).

EGR proteins were originally identified as immediate-early genes and generally have been regarded as differentiation factors. Like mab-10 and lin-29 mutants, Nab and Egr mutant mice are defective in the terminal differentiation of several cell lineages. For example, in Schwann cells, EGR2 promotes the expression of P27, the homolog of C. elegans CKI-1, and acts with NAB proteins to promote terminal differentiation (Fig. 7B) (Parkinson et al., 2004; Baloh et al., 2009). Mammalian homologs of other C. elegans heterochronic genes also control differentiation. Similar to the role of LIN-28 in C. elegans, mammalian LIN28 and LIN28B promote stem cell identity and prevent differentiation by repressing the let-7 microRNA gene (Yu et al., 2007; Heo et al., 2008; Viswanathan et al., 2008). As in C. elegans, increasing levels of let-7 drive differentiation, and the mouse homolog of LIN-41, LIN41, has been shown to be a let-7 target acting in stem cell niches to prevent premature differentiation (Rybak et al., 2009).

Mammalian LIN-28 controls the timing of the onset of puberty in mice and possibly humans (Fig. 7B) (Ong et al., 2009; Zhu et al., 2010). Mice lacking EGR1 function, like lin-29 mutants of C. elegans, fail to undergo puberty (Topilko et al., 1998). EGR1 and NAB proteins act with SF1, the homolog of C. elegans NHR-25, in the gonadotrope lineage of the pituitary gland to regulate the expression of luteinizing hormone and the onset of puberty (Topilko et al., 1998). The molecular mechanism by which mammalian LIN-28 regulates the onset of puberty is not known. Our work raises the possibility that homologs of C. elegans heterochronic genes might act in an evolutionarily conserved pathway that controls the terminal differentiation of cell lineages and the onset of adulthood by regulating the activity of NAB and EGR proteins (Fig. 7B).

This work was supported by NIH training grant T32GM007287 and by the Howard Hughes Medical Institute. H.R.H. is an Investigator of the Howard Hughes Medical Institute. EM imaging was performed by V. Hatch. We thank C. Link for unpublished information regarding mab-10, A. Rougvie for strains and suggestions regarding lin-29, S. Russel and G. Ruvkun for discussions, A. van Oudenaarden for the use of microscopes, C. Engert for help with FISH, and B. Galvin and D. Denning for providing comments concerning the manuscript. Deposited in PMC for release after 6 months.