SEM-2/SoxC regulates multiple aspects of C. elegans postembryonic mesoderm development

Abstract

Development of multicellular organisms requires well-orchestrated interplay between cell-intrinsic transcription factors and cell-cell signaling. One set of highly conserved transcription factors that plays diverse roles in development is the SoxC group. C. elegans contains a sole SoxC protein, SEM-2. SEM-2 is essential for embryonic development, and for specifying the sex myoblast (SM) fate in the postembryonic mesoderm, the M lineage. We have identified a novel partial loss-of-function sem-2 allele that has a proline to serine change in the C-terminal tail of the highly conserved DNA-binding domain. Detailed analyses of mutant animals harboring this point mutation uncovered new functions of SEM-2 in the M lineage. First, SEM-2 functions antagonistically with LET-381, the sole C. elegans FoxF/C forkhead transcription factor, to regulate dorsoventral patterning of the M lineage. Second, in addition to specifying the SM fate, SEM-2 is essential for the proliferation and diversification of the SM lineage. Finally, SEM-2 appears to directly regulate the expression of hlh-8, which encodes a basic helix-loop-helix Twist transcription factor and plays critical roles in proper patterning of the M lineage. Our data, along with previous studies, suggest an evolutionarily conserved relationship between SoxC and Twist proteins. Furthermore, our work identified new interactions in the gene regulatory network (GRN) underlying C. elegans postembryonic development and adds to the general understanding of the structure-function relationship of SoxC proteins.

AUTHOR SUMMARY

SoxC transcription factors play important roles in metazoan development. Abnormal expression or function of SoxC factors has been linked to a variety of developmental disorders and cancers. It is therefore critical to understand the functions of SoxC proteins in vivo. C. elegans has a single SoxC transcription factor, SEM-2, that is known to regulate a fate decision between a proliferative progenitor cell vs. a terminally differentiated cell during postembryonic mesoderm development. In this study, we report new functions of SEM-2 in postembryonic mesoderm development via our studies of a partial loss-of-function allele of sem-2. Our work uncovers new regulatory relationships between SEM-2/SoxC and the FoxF/C transcription factor LET-381, and between SEM-2/SoxC and the C. elegans Twist ortholog HLH-8. Our findings suggest that the SoxC-Twist axis, including the downstream targets of Twist, represents an evolutionarily conserved regulatory cassette important in metazoan development.

INTRODUCTION

Metazoan development is characterized by the specification and diversification of multipotent cells, as well as the proper organization of their differentiated descendants. These processes require well-orchestrated interplay between cell-intrinsic transcription factors and cell-cell signaling. The C. elegans postembryonic mesoderm, the M lineage, offers a unique model system to dissect the underlying regulatory logic of cell fate specification and diversification. The M lineage is derived from a single pluripotent precursor cell, the M mesoblast [1]. During hermaphrodite postembryonic development, the M cell undergoes stereotypical divisions to produce fourteen striated body wall muscles (BWMs), two non-muscle coelomocytes (CCs), and two multipotent sex myoblasts (SMs) that subsequently proliferate to produce sixteen sex muscles—four type I and four type II vulval muscles (vm1s and vm2s), as well as four type I and four type II uterine muscles (um1s and um2s)—that are required for egg laying (Fig 1A–B).

Fig 1: The sem-2 P158S mutants exhibit multiple defects during postembryonic development.

A) Diagram of the C. elegans hermaphrodite postembryonic mesoderm, M lineage, showing all differentiated cell types that arise from the M mesoblast cell. a, anterior; p, posterior; d, dorsal; v, ventral; l, left; r, right. B) Schematic of developing C. elegans hermaphrodites showing the locations of M lineage cells. C) An L4 sma-9(0) animal that has four embryonic coelomocytes (arrowheads) labelled by CC::gfp. D–E) Two L4 sem-2(jj152[P158S]); sma-9(0) animals that have four embryonic CCs (arrowheads) and one (D) or two (E) M-CCs (arrows). F) A wild-type gravid adult hermaphrodite. G) A sem-2(jj152[P158S]) gravid adult hermaphrodite, showing the Egl phenotype. H) The vulva (asterisk) region of a wild-type gravid adult hermaphrodite. I) A sem-2(jj321[P158S]) gravid adult hermaphrodite with two vulvae (asterisks). J) Orientation of all animals shown in this and subsequent figures.

Previous studies have identified multiple transcription factors and signaling components essential for the proper development of the M lineage [2]. In particular, LIN-12/Notch signaling is known to function upstream of the single C. elegans SoxC protein, SEM-2, to specify the SM fate in the ventral M lineage, while the zinc finger transcription factor SMA-9 antagonizes BMP signaling to specify the M lineage-derived CC (M-CC for short) fate in the dorsal side [3–6]. SMA-9 functions by activating the expression of the sole FoxF/C transcription factor in C. elegans, LET-381, in the M-CC mother cells. LET-381 then directly activates the expression of the Six homeodomain transcription factor, CEH-34, where LET-381 and CEH-34 function in a feedforward manner to directly activate the expression of genes required for CC differentiation and function [7, 8]. At the same time, SMA-9 and LET-381 are each known to repress the expression of sem-2 in the dorsal side of the M lineage to prevent it from specifying the SM fate [6].

In addition to the factors described above that are important for proper fate specification in the M lineage, the sole C. elegans Twist ortholog, HLH-8, is known to function in proper patterning of the M lineage. HLH-8 is a basic helix-loop-helix (bHLH) transcription factor that is expressed in the undifferentiated cells of the M lineage through regulatory elements located in the promoter, and it is expressed in the vulval muscles by autoregulation through E boxes located in its first intron [9, 10]. HLH-8 is known to have multiple functions during M lineage development: proper cleavage orientation of M lineage cells, proper proliferation of the SMs, and proper differentiation and function of the vulval muscles [11, 12].

In this study, we provide new insight into the relationships between various factors important in M lineage development and the sole SoxC transcription factor SEM-2. SoxC proteins are Sry-related HMG box (Sox)-containing transcription factors that are known to play critical roles in multiple developmental processes [13]. There are three highly conserved SoxC proteins in vertebrates, Sox4, Sox11, and Sox12. Abnormal expression or function of SoxC factors has been linked to a variety of developmental disorders and cancers [13–18]. In particular, mutations in Sox4 and Sox11, most of them being point mutations in the HMG box, are associated with a developmental disorder called Coffin-siris syndrome (CSS) [19–21]. However, the underlying molecular mechanisms are not completely understood.

We identified a new allele of sem-2, jj152, which is a point mutation resulting in a single amino acid change in a highly conserved residue at the end of the DNA-binding domain of SEM-2. We present genetic evidence showing that this single amino acid change results in a partial loss of SEM-2 function. By analyzing the M lineage phenotypes of jj152 mutants, we uncovered an unexpected role of SEM-2 in the dorsal M lineage, where SEM-2 functions antagonistically at the level of expression and function with LET-381 in CC specification. We also found that SEM-2 regulates the expression of hlh-8, possibly directly, in the SMs, and that SEM-2 is essential for the proliferation and the diversification of the SMs. Our work uncovered new interactions in the gene regulatory network underlying C. elegans postembryonic development, some of which are likely conserved in other species, and we add to the general understanding of the functions and structure-function relationship of SoxC proteins.

MATERIALS AND METHODS

C. elegans strains and transgenic lines

C. elegans strains used in this study were maintained at 20°C under normal culture conditions [22]. Analyses of hlh-8 reporters in sem-2(jj152), sem-2(jj321), sem-2(jj382 jj417) and sem-2(jj476) were performed at 25°C unless specifically noted. All strains are listed in Supplementary Table 1.

Microscopy

Epifluorescence and differential interference contrast (DIC) microscopy were conducted on a Leica DMRA2 compound microscope equipped with a Hamamatsu Orca-ER camera using the iVision software (Biovision Technology). Subsequent image analysis was performed using Fiji. For comparison of fluorescence intensities in different genetic backgrounds, images were collected at the same magnification and exposure. Statistical significance was calculated by performing unpaired two-tailed Student’s t-tests.

Isolation and mapping of sem-2(jj152)

sem-2(jj152) was isolated in a large-scale sma-9(cc604) Susm screen (suppression of the sma-9(cc604) M lineage defect) for the restoration of M-derived CCs in sma-9(cc604) mutants [5]. The sma-9(cc604) allele is a nonsense mutation that is considered a genetic null [23] and referred to as sma-9(0). jj152 was mapped to chromosome I using the whole genome sequencing approach described in Liu et al. [5]. Because jj152 animals are egg-laying defective (Egl), we performed complementation tests between jj152 and each of the two previously studied sem-2 alleles, n1343 and ok2422 [6]. All jj152/n1343 trans-heterozygotes are Egl, but they did not exhibit the Susm phenotype. To perform a complementation test between jj152 and ok2422, hT2[qIs48]/jj152; jjIs3900/+; sma-9(0) males were mated with hT2[qIs48]/sem-2(ok2422); sma-9(0) hermaphrodites. Red, non-green jjIs3900/+; jj152/sem-2(ok2422); sma-9(0) cross progeny were scored for CC number. Most jjIs3900/+; jj152/sem-2(ok2422); sma-9(0) cross progeny were dead as embryos. Among those that survived, all (8/8) were Egl and exhibited the Susm phenotype.

Suppression of the sma-9(0) M lineage defect (Susm) assay

For the Susm assay, animals containing a CC::gfp marker (see Supplementary Table 1) were grown at 20°C to the young adult stage. The number of animals with four, five, or six CCs were tallied for each genotype. Statistical significance was calculated by performing unpaired two-tailed Student’s t-tests.

Assays for brood size and embryonic lethality

Brood size and embryonic lethality were scored under the Nikon SMZ1500 fluorescence stereo microscope. Single L4 animals were placed onto NGM plates and allowed to give progeny at 20°C. To score the brood size of wildtype N2 animals, the parent was transferred to a new plate every 24 hours for a total of 3 days, and the eggs and newly hatched L1 larvae on each plate were counted. To score the brood size of the Egl strains, the total number of eggs present in the bagged parent and newly hatched larvae on the plate were counted 24–48 hours post-plating. To assess embryonic lethality, the progeny of all strains were allowed to grow to adulthood, and the total number of adults was noted and compared with the corresponding brood size of each parent. The average brood size and embryonic lethality were calculated for each strain. Statistical significance between the mutant strains and wildtype N2 worms was calculated by performing unpaired two-tailed Student’s t-tests.

Plasmids

All plasmids used in this study are listed in Supplementary Table 2. pMDB28 was generated by performing Gibson Assembly with sem-2 genomic sequences amplified from N2 worms, the sequence of gfp amplified from pDD282 [24], a gBlock containing 2xflag purchased from IDT, and a pBSII/SK+ vector digested with BamHI and HindIII. pMDB37 was generated by performing Gibson Assembly with hlh-8 genomic sequences amplified from N2 worms, a gBlock with the gpd-2 intergenic region and nuclear localization sequence from CEOPX036 [25], the sequence of gfp amplified from pDD282 [24], another nuclear localization signal introduced by primer MDB-87, and a pBSII/SK+ vector digested with BamHI and HindIII.

pAYL11, pAYL21–pAYL25, and pAYL31–pAYL35 were generated using pJKL502 in the Fire Lab Vector Kit as the original template. pJKL502 contains a 1.3kbp fragment (−1.3kbp to −1bp) of the hlh-8 promoter driving gfp expression. As a starting point, deletions were made at either the 5′ or 3′ end of a 517bp fragment (−517bp to −1bp) of the hlh-8 promoter, which was previously shown to be sufficient to drive M lineage expression [9]. These fragments contained HindIII and XbaI restriction sites at their ends and were used to replace the promoter region in pJKL502. Next, the 50bp internal deletion constructs, pAYL21–pAYL25, were generated by ligating the promoter elements from a 5′ deletion construct and a 3′ deletion construct, thus resulting in a set of constructs each missing a contiguous 50 bp sequence of the hlh-8 517bp promoter. Constructs that contain 20bp deletions in the −300bp to −200bp region, pAYL31–pAYL35, were generated by a two-step bridging/fusion PCR scheme where PCR fragments containing an overlapping 20bp region were used as templates for a second round of PCR to bridge the two fragments together. The resulting fragments (flanked by HindIII and XbaI sites) were subcloned to replace the promoter region in pJKL502 in the same way as described for the 5′ and 3′ deletion constructs. All plasmids were confirmed by Sanger sequencing.

CRISPR

CRISPR experiments were conducted either by using plasmids expressing regular Cas9 (pDD162) [24] or the VQR variant of Cas9 (pRB1080) [26] and plasmids expressing sgRNAs in the pRB1017 backbone [27], or by injecting ribonuclear RNP complexes with Cas9 protein, tracer RNA (from IDT), and sgRNA as described in Beacham et al. [28] (sequences listed in Supplementary Table 3). For large insertions, plasmid repair templates (Supplementary Table 2) were used, while ssDNA oligos (Supplementary Table 3) were used as repair templates to introduce point mutations. For injections, PRF4(rol-6(d)) [29] was used as a co-injection marker. Injected animals were singled onto NGM plates seeded with OP50 bacteria. Plates that gave the most roller progeny were selected for screening by PCR. Final CRISPR edits were confirmed by Sanger sequencing.

RNAi

The plasmids for let-381(RNAi) and ceh-34(RNAi) were obtained from the Ahringer RNAi library [30] and confirmed by sequencing. RNAi was conducted by following the protocol of Amin et al. [7]. Synchronized L1 animals of various genotypes were plated on HT115(DE3) bacteria expressing dsRNA against the gene of interest, allowed to grow at 25°C, and scored for M lineage phenotypes 12–48 hours after plating. Bacteria carrying the L4440 empty vector was used as a negative control.

Data availability

Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.

RESULTS

A sem-2 allele, jj152, suppresses the sma-9(0) M lineage phenotype in coelomocyte specification

In a sma-9(0) suppressor screen to identify new factors involved in M lineage development [5], we identified a complementation group on chromosome I that includes a single allele, jj152. jj152 animals showed partial suppression of the sma-9(0) M lineage (Susm) phenotype: 41.2% of jj152; sma-9(0) animals (N=767) have 1–2 M-CCs, instead of zero M-CCs in sma-9(0) single mutant (Fig 1C–E, Table 1A). jj152 animals are also 100% egg-laying defective (Egl), have a smaller brood size than wildtype (WT), display ~25% embryonic lethality and exhibit a 6.9% bivulva (Biv) phenotype (N=360) (Fig 1F–I, Table 1B).

Table 1.

Summary of the phenotypes caused by the sem-2 P158S mutation.

A. The sem-2 P158S mutation suppresses the sma-9(cc604) missing M-derived coelomocyte (CC) phenotype.

Genotype	% 1–2 M-CCsa	% Egl	No. of animals
sem-2(jj152)	100	100	845
sma-9(cc604)	1.4	-	1359b
sem-2(n1343); sma-9(cc604)	0.59%	100	1672
sem-2(jj152); sma-9(cc604)	41.2***	100	767
sem-2(jj152); sma-9(cc604); jjIs1647[sem-2(+)]	0 ND	0	329
sem-2(ok2422)/+; sma-9(cc604)	2.4ND	0	205
sem-2(ok2422)/sem-2(jj152); sma-9(cc604)	100***	100	8c

^aCCs were scored using the CC::gfp described in Supplementary Table 1.

^bData from Liu et al. (2022).

^cMost sem-2(ok2422)/sem-2(jj152); sma-9(cc604) animals are embryonic lethal, with only few animals able to survive through larval development to become young adults, which is when we score the M-CC phenotype. Genotyping four of the eight animals confirmed that they are jj152/ok2422. Details of the complementation tests are described in Materials and Methods.

ND: no difference;

^***:P<0.001, when compared to sma-9(cc604) animals based on unpaired two-tailed Student’s t-test.

B. The sem-2 P158S mutation results in reduced brood size, increased embryonic lethality, and an egg-laying defective phenotype.

Genotype	% Egl (n)	Brood size (n)	% Emb (n)
WT	0 (>100)	303 ± 26.7 (9)	0 (2582)
sem-2(jj152)	100 (>100)	32.4 ± 8.3 (11)***	17.2 ± 9.7 (314)***
sem-2(jj320)	100 (>100)	41.9 ± 5.5 (10)***	20.97 ± 8.4 (419)***
sem-2(jj321)	100 (>100)	37.3 ± 5.2 (10)***	30.31 ± 9.2 (373)***
sem-2(ok2422)	N/A	N/A	100***a

Egl: egg-laying defective; Emb: embryonic lethal.

The average brood size and average percentage of Egl or Emb ± standard deviation are presented in this table. Details on scoring the brood size and embryonic lethality are described in Materials and Methods.

n, number of worms scored. For brood size, n reflects the number of parents singled and their brood sizes scored.

^***P<0.001 when compared to WT based on unpaired two-tailed Student’s t-test.

N/A: not applicable.

^adata from Tian et al. 2011.

Several lines of evidence suggest that jj152 is an allele of sem-2: 1) whole genome sequencing revealed that jj152 maps to chromosome I and contains a cytosine (C) to thymine (T) nucleotide change, which results in a proline (P) to serine (S) residue change in amino acid 158 (P158S) of SEM-2 (Fig 2A–B); 2) jj152 failed to complement the sem-2(ok2422) null allele in both the Egl and the Susm phenotypes (Table 1A); 3) a transgene carrying a fosmid containing the wild-type sem-2 genomic sequences (Tian et al. 2011) rescued both the Susm and the Egl phenotypes of jj152 mutants (Table 1A); and 4) CRISPR-engineered sem-2 P158S mutations in the wild-type background, sem-2(jj320) and sem-2(jj321), recapitulated the mutant phenotypes exhibited by jj152 animals (Table 1B). Given that jj152, jj320, and jj321 mutants exhibit a weaker phenotype than sem-2(ok2422) null or trans-heterozygous mutants (Table 1A–B), or animals that have undergone postembryonic sem-2(RNAi) [6], we concluded that the SEM-2(P158S) mutant protein exhibits a partial loss of SEM-2 function.

Fig 2: The SEM-2 P158S mutation does not affect the expression, localization, or level of the SEM-2 protein.

A) Schematic of the SEM-2 protein with the DNA-binding domain in yellow, the serine-rich region in red, and the transactivation domain (TAD) in purple. The P158S mutation is indicated by an asterisk. Sequence alignment of part of the DNA-binding domain of SEM-2 with SoxC proteins in different vertebrate species. Residue P158 is conserved in all SoxC proteins shown and is highlighted by a green box. B) Structural model of the SEM-2 DNA-binding domain (yellow) with DNA (tan), based on the structure of the Mouse Sox4 DNA-binding domain-DNA complex (PDB code 3U2B [32]. Residue P158 is highlighted in green. C–D) Fluorescence images showing the expression patterns of the endogenously tagged GFP::2xFLAG::SEM-2 (C) and GFP::2xFLAG::SEM-2(P158S) (D) in hermaphrodites at the L3 stage. Arrows point to the migrating SMs in the focal plane.

The single amino acid change in sem-2(jj152), P158S, does not affect the localization or level of SEM-2 protein

We were intrigued by the Susm phenotype of sem-2(jj152[P158S]) because sem-2(n1343), which has a Tc1 transposon insertion that disrupts sem-2 expression in the SM lineage and exhibits a SM to BWM fate transformation phenotype, does not show any Susm phenotype (Table 1, [6]). We therefore decided to further explore the impact of the sem-2(jj152[P158S]) point mutation on SEM-2 function, and the relationship between sem-2 and sma-9 in M-CC fate specification.

sem-2 encodes the sole SoxC protein in C. elegans, whereas vertebrates have three SoxC proteins, Sox4, Sox11, and Sox12 [31]. SoxC proteins have a conserved DNA-binding domain, a serine-rich region, and a transactivation domain (Fig 2A). Amino acid P158 is located near the end of SEM-2′s DNA-binding domain and is conserved in SoxC homologs from vertebrates (Fig 2A). Structural modeling based on the co-crystal structure of mouse Sox4 and its cognate DNA showed that P158 is located in a flexible linker region of the domain but does not directly touch DNA (Fig 2B, [32]). It is likely that the P158S mutation affects the DNA-binding affinity of SEM-2.

To begin to assess the functional consequences of the P158S mutation on SEM-2 protein, we tagged SEM-2 at the endogenous locus with GFP::2xFLAG using CRISPR. This endogenously tagged SEM-2 is functional as GFP::2xFLAG::SEM-2 animals are viable, fertile, and non-Egl. As shown in Fig 2C, GFP::2xFLAG::SEM-2 is nuclear localized in many different cell types, such as cells of the hypodermis, intestine, vulva and pharynx, similar to what we have previously reported using transgenic animals expressing a GFP-tagged SEM-2 in a fosmid backbone [6]. To determine if the P158S mutation affects SEM-2 expression, localization, or stability, we used CRISPR and introduced the P158S mutation into the endogenously tagged GFP::2xFLAG::SEM-2, and generated sem-2(jj382 jj417[GFP::2xFLAG::SEM-2(P158S)]. As shown in Fig 2C–D, SEM-2(P158S) exhibits a similar pattern and level of expression and localization to the wild-type SEM-2 protein. Considering the structural information shown in Fig 2B, we concluded that the P158S mutation likely affects the activity of the SEM-2 protein. This is consistent with our genetic evidence suggesting the SEM-2(P158S) is partially functional.

sem-2 exhibits dynamic expression patterns in the postembryonic M lineage

Having an endogenously tagged GFP::2xFLAG::SEM-2 allowed us to more accurately assess the expression pattern of sem-2 in the M lineage. Consistent with our previous report based on transgenic animals expressing GFP::SEM-2 in a fosmid backbone [6], endogenous GFP::2xFLAG::SEM-2 expression is detectable in the SM mothers, the SMs, and throughout the SM lineage in all SM descendants before they terminally differentiate (Fig 3E–K″). Surprisingly, strong nuclear GFP::2xFLAG::SEM-2 signal is detectable in the M mesoblast, and it remains detectable, but becomes progressively fainter, in all M lineage cells up to the 16-M stage (Fig 3A–E″). At the 16-M stage, GFP::2xFLAG::SEM-2 signal is significantly brighter in the SM mother cells (M.vlpa and M.vrpa) (Fig 3A–E″). At the 18-M stage when the BWMs and CCs are born and become terminally differentiated, GFP::2xFLAG::SEM-2 is only detectable in the two SMs (M.vlpaa and M.vrpaa) and transiently in the SM sister cells (M.vlpap and M.vrpap) before they terminally differentiate into BWMs (Fig 3F–F″). The expression pattern of GFP::2xFLAG::SEM-2 in the M lineage is summarized in Fig 3L.

Fig 3: Endogenously tagged GFP::2xFLAG::SEM-2 exhibits dynamic expression throughout the M lineage.

A–K″) Florescence images showing the expression of GFP::2xFLAG::SEM-2 (A′–K′) in M lineage cells of hermaphrodites labelled by the hlh-8p::nls::mCherry reporter (A–K) at different stages of M lineage development, with the corresponding merged images shown in A″–K″. The SM sisters (M.vlpap and M.vrpap) migrate posteriorly after they are born and become BWMs [1]. Only the left side of an animal is shown in A–K″. The other side is out of the focal plane. YA, young adult. L) Summary of the GFP::2xFLAG::SEM-2 expression pattern in the M lineage. Green lines indicate cells that express GFP::2xFLAG::SEM-2. Cells represented by the light green color have lower levels of GFP::2xFLAG::SEM-2 expression than cells represented by the dark green color. Black lines indicate cells that do not express GFP::2xFLAG::SEM-2.

SMA-9 specifically represses the dorsal M lineage expression of GFP::2xFLAG::SEM-2 at the stage of CC and SM fate specification

We have previously shown that SMA-9 represses the expression of SEM-2 in the dorsal M lineage at the 18-M stage when the SMs are born [6]. Given the unexpected Susm phenotype of sem-2(jj152[P158S]) and the unexpected expression of endogenous GFP::2xFLAG::SEM-2 in the early M lineage prior to the birth of SMs, we sought to determine if the expression level or pattern of GFP::2xFLAG::SEM-2 changes in the M lineage of sma-9(0) mutants. Consistent with our previous report, we observed expression of endogenous GFP::2xFLAG::SEM-2 in the SM-like cells in the dorsal M lineage of sma-9(0) animals (Fig 4B–B‴). However, we did not detect any difference in the expression level or pattern of GFP::2xFLAG::SEM-2 between WT and sma-9(0) animals from the 1-M to the 8-M stage (S1 Fig). These results suggest that SMA-9 functions to repress the dorsal M lineage expression of sem-2 only at the stage of CC and SM fate specification.

Fig 4: SMA-9 and LET-381 repress sem-2 expression in the dorsal M lineage, while ectopically expressed SEM-2 represses let-381 expression in the dorsal M lineage.

A–D″) Florescence images showing GFP::2xFLAG::SEM-2 (A′–D′) in M lineage cells labelled by the hlh-8p::nls::mCherry reporter (A–D) at different stages of M lineage development in WT (A–A″), sma-9(0) (B–B″), let-381(RNAi) (C–C″), or ceh-34(RNAi) (D–D″) hermaphrodites. A″–D″ are corresponding merged images. A‴–D‴) Schematic representation of the fluorescence images. Green circles represent cells expressing GFP::2xFLAG::SEM-2. Green circles with thick, black outlines are SMs. Green circles with thin, black outlines represent SM sisters fated to become BWMs. The cell with thick, burnt orange outline is an SM mother (C‴). Asterisks label the SM cells (M.vlpaa); white, unshaded arrowheads label the SM sister cells (M.vlpap), which are BWMs; arrows label the SM-like cells born in the dorsal side of the M lineage in sma-9(0) (B′) or let-381(RNAi) (C′) hermaphrodites. E–H″) Florescence images showing mNG::LET-381 (E′–H′) in M lineage cells labelled by the hlh-8p::nls::mCherry reporter (E-H) at different stages of M lineage development in WT (E–E″), sem-2(jj476[P158S]) (F–F″), sma-9(0) (G–G″), or sem-2(jj476[P158S]); sma-9(0) (H–H″) hermaphrodites. (E″–H″) are corresponding merged images. E‴–H‴) Schematic representation of the fluorescence images. Blue circles represent CC cells expressing mNG::LET-381. Circles with thick, black outlines are SMs that have migrated to the future vulva region. Yellow asterisks label the SM cells (M.vlpaa); yellow arrow labels the SM-like cells born in the dorsal side of the M lineage in sma-9(0) (G) hermaphrodites; yellow, shaded arrowheads label the CC cells (M.dlpa) in WT (E–E′), sem-2(jj476[P158S]) (F–F′), and sem-2(jj476[P158S]); sma-9(0) (H–H′) hermaphrodites. The fluorescence intensity of the hlh-8p::nls::mCherry reporter is significantly reduced in animals carrying the sem-2(jj476[SEM-2(P158S)] mutation, and was adjusted individually for panels F and H for cell identification. Only the left side of an animal is shown in this figure, while the other side is out of the focal plane.

SEM-2 antagonizes the function of LET-381 in specifying M-derived CCs by repressing let-381 expression

As described above, when the fully functional SEM-2 is ectopically expressed in the dorsal M lineage of sma-9(0) single mutants, no M-CCs are produced. However, around 40% of sem-2(jj152[P158S]); sma-9(0) double mutants have 1–2 M-CCs. The forkhead transcription factor LET-381/FoxF/C is known to function downstream of SMA-9 and upstream of the Six homeodomain transcription factor CEH-34, where LET-381 and CEH-34 function in a feedforward manner to specify the M-CC fate [7, 8]. We therefore hypothesized that the Susm phenotype of sem-2(jj152[P158S]) might be because the partially functional SEM-2(P158S) protein, when ectopically expressed in the dorsal side of the M lineage at the 16-M stage, cannot fully inhibit either the expression or the function of let-381.

To test the above hypothesis, we examined the expression pattern of an endogenously tagged mNG::LET-381 [33] in WT, sma-9(0), and sem-2(jj152[P158S]); sma-9(0) animals. Since sem-2 and let-381 are located close to each other on Chromosome I (sem-2 at −0.27 while let-381 at +1.02), we used CRISPR to introduce the P158S mutation into the let-381(dev205[mNG::LET-381]) strain and generated the sem-2(jj476[P158S]) let-381(dev205[mNG::LET-381]) strain. We then conducted genetic crosses and generated the sem-2(jj476[P158S]) let-381(dev205[mNG::LET-381]); sma-9(0) strain. Similar to the previously reported pattern of expression for the LET-381::GFP transgene (Amin et al., 2010), endogenous mNG::LET-381 is expressed in the dorsal M-CC mothers (M.dlp and M.drp) and the CCs (M.dlpa and M.drpa) in WT animals (100%, N=40) (Fig 4E–E‴) and in sem-2(jj476[P158S]) single mutants (98%, N=50, Fig 4F–F‴), while this M lineage expression disappears in sma-9(0) mutants (98.3%, N=59) (Fig 4G–G‴). Instead, 65.9% of sem-2(jj476[P158S]) let-381([mNG::LET-381]); sma-9(0) mutants examined (N=41) showed expression of mNG::LET-381 in M.dlpa and/or M.drpa (Fig 4H–H‴), cells that are normally fated to become CCs. Consistent with this finding, the M-CCs formed in sem-2(jj152[P158S]); sma-9(0) double mutants require LET-381. As shown in Table 2, while 24.6% of sem-2(jj152[P158S]); sma-9(0) animals on control RNAi with the empty vector L4440 (N=240) had 1–2 M-CCs, only 3.1% of sem-2(jj152[P158S]) let-381(RNAi); sma-9(0) animals (N=291) had M-CCs. Taken together, the above findings demonstrate that SEM-2 antagonizes the function of LET-381 in specifying M-derived CCs by repressing let-381 expression.

Table 2.

LET-381 is required for the formation of M-derived CCs in sem-2(jj152); sma-9(cc604) double mutants.

Genotype	RNAia	% with no M-CCsb	% with 1–2 M-CCs	No. of animals
WT	L4440	7.6	92.2	437c
WT	let-381	93.4	6.6	412
sem-2(jj152)	L4440	0.7	99.3	431
sem-2(jj152)	let-381	92.1	7.9	350
sem-2(jj152); sma-9(cc604)	L4440	75.4	24.6	240
sem-2(jj152); sma-9(cc604)	let-381	96.9	3.1	291

^aPostembryonic RNAi as described in Materials and Methods.

^bCCs were scored using the CC::gfp described in Supplementary Table 1.

^cOne of the 437 animals scored had 3 M-CCs.

LET-381, but not CEH-34, represses the expression of SEM-2 in dorsal M lineage cells fated to become CCs

We have previously shown that in let-381(RNAi) animals, CCs are transformed to SM-like cells in the dorsal side of the M lineage [8], and that LET-381 represses the expression of the GFP::SEM-2 transgene in the dorsal M lineage cells fated to become M-CCs [6]. We found that this result holds true for the endogenously tagged GFP::2xFLAG::SEM-2, as let-381(RNAi) animals had GFP::2xFLAG::SEM-2 ectopically expressed in the dorsal M lineage cells that are normally fated to become M-CCs (Fig 4C–C‴).

Because LET-381 and CEH-34 function in a feedforward manner to specify M-CCs, we also examined the expression pattern of GFP::2xFLAG::SEM-2 in ceh-34(RNAi) animals. As shown in Fig 4D–D‴, ceh-34(RNAi) animals did not exhibit any ectopic expression of GFP::2xFLAG::SEM-2 in the dorsal M lineage. This is consistent with CEH-34 functioning downstream of LET-381 for specifying M-CCs [7, 8]. Thus, LET-381, but not CEH-34, functions to repress sem-2 expression in cells fated to become M-CCs.

SEM-2 regulates the expression of hlh-8 in the M lineage

During our analysis of the mechanistic basis of the Susm phenotype of sem-2(P158S) mutants, we noticed that sem-2(P158S) mutants exhibit significantly reduced levels of expression of the transgenic hlh-8 transcriptional reporter jjIs3900[hlh-8p::nls::mCherry] (Fig 5, S2 Fig). In wild-type animals, hlh-8p::nls::mCherry is expressed in the M mesoblast and all undifferentiated cells in the M lineage from the 1-M to 16-SM stage. The sem-2(jj321[P158S]) mutants display reduced expression of this reporter at all stages of M lineage development (100%, N=234) (Fig 5A–F, S2A–B″ Fig). Similar reduction of hlh-8p::nls::mCherry expression was observed in the M lineage of sem-2(jj152[P158S]) mutants (100%, N=643). Moreover, the expression of another hlh-8 transcriptional reporter ayIs6[hlh-8p::gfp] [9] is also reduced in sem-2(jj152[P158S]) mutants (100%, N=599) (S2C–D″ Fig). Thus, the expression of two independent, transgenic, hlh-8 transcriptional reporters is significantly reduced in sem-2(jj152[P158S]) mutants.

Fig 5: SEM-2 regulates hlh-8 expression in the M lineage.

A–C) Fluorescence images (A–B) of wild-type (A–A″) and sem-2(jj321[P158S]) (B–B″) L1 animals showing expression of hlh-8p::nls::mCherry in the M mesoblast cell. A′–B′ and A″-B″ are corresponding DIC and merged images, respectively. C) Quantification of hlh-8p::nls::mCherry expression level in the M mesoblast of wild-type and sem-2(jj321[P158S]) L1 animals. D–F) Fluorescence images (D–E) of wild-type (D–D″) and sem-2(jj321[P158S]) (E–E″) L3 animals showing expression of hlh-8p::nls::mCherry in SMs. D′–E′ and D″–E″ are corresponding DIC and merged images, respectively. F) Quantification of hlh-8p::nls::mCherry expression in the SMs of wild-type and sem-2(jj321[P158S]) L3 animals. G–I) Fluorescence images (G′–H′) of wild-type (G–G″) and sem-2(jj321[P158S]) (H–H″) L1 animals showing expression of mNG::MLS-2 in the M mesoblast labelled by the hlh-8p::nls::mCherry reporter (G–H). I) Quantification of mNG::MLS-2 in the M mesoblast cell of wild-type and sem-2(jj321[P158S]) animals. J–M″) Fluorescence images of a wild-type animal (J-K″) and a sem-2(P158S) mutant animal, showing GFP::2xFLAG::SEM-2 (J, K, L, M), hlh-8p::nls::mCherry (J′, K′, L′, M′) and the corresponding merged images (J″, K″, L″, M″) in their two SMs. N) Quantification of GFP::2xFLAG::SEM-2 in the SMs of wild-type and sem-2(jj417[P158S]) animals. Arrows point to the M cell, while arrowheads point to the SM cell. For panels C, F, I and N, each dot represents an animal scored. Statistical significance was calculated by performing unpaired two-tailed Student’s t-tests. *** P<0.001, **** P<0.0001, ns, not significant.

To determine if the reduced expression of hlh-8 in SEM-2 P158S mutants is specific to the hlh-8 gene in the M lineage, we examined the expression of mls-2 in sem-2(jj321[P158S]) mutants using an endogenously tagged mNG::MLS-2. MLS-2 is a NK homeodomain protein that regulates the patterning, cell fate specification, and proliferation in the early M lineage [34]. mls-2 is expressed starting at the 1-M stage in the M lineage [34]. We found no change in expression level or pattern of mNG::MLS-2 in sem-2(jj321[P158S]) mutants compared to WT (Fig 5G–I). These results suggest that SEM-2 specifically regulates the expression of hlh-8 in the M lineage.

To determine if a change in hlh-8 expression was specific to the SEM-2 P158S allele, we introduced jjIs3900[hlh-8p::nls::mCherry] into sem-2(ok2422) null mutants. The majority of sem-2(ok2422) null mutants produced by heterozygous sem-2(ok2422)/hT2[qIs48] mothers die as threefold embryos, while a few embryos can hatch but die as L1 larvae [6]. We found that 83.3% sem-2(ok2422) embryos (N=12) express hlh-8p::nls::mCherry in the M cell at similar levels as sem-2(ok2422)/hT2[qIs48] control embryos (92.9%, N=14) (S2E–G Fig). However, none of the sem-2(ok2422) L1 animals (100%, N=14) display any detectable hlh-8p::nls::mCherry expression, unlike the sem-2(ok2422)/hT2[qIs48] control L1s where most of them show bright hlh-8p::nls::mCherry expression in the M cell (98.1%, N=52) (S2H–I‴ Fig). We reasoned that the more robust expression of sem-2 in sem-2(ok2422) null embryos born from sem-2(ok2422)/hT2[qIs48] parents is likely due to the maternal contribution of SEM-2 by sem-2(ok2422)/hT2[qIs48] parents. These results further suggest that SEM-2 is required for maintaining stable hlh-8 expression in the M lineage.

SEM-2 is required for endogenous hlh-8 expression in the M lineage

To determine if SEM-2 is required for endogenous hlh-8 expression in the M lineage, we generated an endogenous hlh-8 transcriptional reporter using the strategy described in Luo et al. [35]. Using CRISPR, we inserted the intergenic trans-splicing acceptor region from CEOPX036 followed by nls::gfp::nls at the end of the hlh-8 coding region. The resulting hlh-8(jj422[hlh-8::sl2::nls::gfp::nls]) (denoted jj422[hlh-8p::gfp]) is a bicistronic allele where hlh-8 and gfp are co-transcribed under the hlh-8 promoter (Fig 6A), and expression of nuclear localized GFP is indicative of endogenous hlh-8 expression. Animals expressing this endogenous hlh-8 transcriptional reporter do not exhibit any overt phenotypes exhibited by hlh-8(0) mutants [12], suggesting that hlh-8 function is not compromised in jj422[hlh-8p::gfp] animals. The GFP signal of this endogenous hlh-8 transcriptional reporter is rather faint, yet it has the same expression pattern as the transgenic jjIs3900[hlh-8p::nls::mCherry] and ayIs6[hlh-8p::gfp] reporters in the M lineage (Fig 6B–D″).

Fig 6: SEM-2 is required for endogenous hlh-8 expression in the M lineage.

A) Schematic representation of an endogenous hlh-8 transcriptional reporter generated by introducing a sl2 spliced leader sequence from CEOPX036 and the sequence of nls::gfp::nls at the end of the HLH-8 coding region. This reporter hlh-8(jj422[hlh-8::sl2::nls::gfp::nls]) is denoted jj422[hlh-8p::gfp]. Gray boxes represent hlh-8 exons with introns separating them. The orange color indicates the sl2 sequence, the blue color indicates the nuclear localization signal (nls), and the green color indicates gfp. B–D″) Fluorescence images showing the expression of jj422[hlh-8p::gfp] (B′-D′), the transgenic jjIs3900[hlh-8p::nls::mCherry] (B-D), and corresponding merged imageds (B″-D″) in the M lineage at the 1-M (B–B″), 2-SM (C–C″), and 8-SM (D–D″) stages. jj422[hlh-8p::gfp] has the same expression pattern as the transgenic jjIs3900[hlh-8p::nls::mCherry], but the jj422[hlh-8p::gfp] signal in the M cell is faint. Arrow in B′ points to the M cell. E–H) Fluorescence images (E′–G′) of wild-type (E–E″) and sem-2(jj321[P158S]) (F–G″) L3 animals showing expression of jj422[hlh-8p::gfp] in SMs labelled by the transgenic hlh-8p::nls::mCherry reporter (E–G). Merged images are shown in E″–G″. H) Quantification of jj422[hlh-8p::gfp] expression in the number of SMs in wild-type and sem-2(jj321[P158S]) animals.

We then examined the expression of the endogenous hlh-8 transcriptional reporter in sem-2(jj321[P158S]) mutants. The expression of jj422[hlh-8p::gfp] in the early M lineage is too faint to make accurate comparisons between WT and sem-2(jj321[P158S]). At the SM stage, the expression level of jj422[hlh-8p::gfp] was significantly reduced in at least one of the two SMs in sem-2(jj321[P158S]) mutants (81.0%, N=42) compared with wild-type animals that express sem-2 in both SMs (94.3%, N=35) (Fig 6E–H). Similarly, In 50% of sem-2(jj321[P158S]) animals, jjIs3900[hlh-8p::nls::mCherry] is undetectable in one of two SMs (N=26), whereas wild-type animals always have detectable expression in the SMs (100%, N=24) (Fig 5J–M″). Thus, based on data using both the transgenic and endogenous hlh-8 transcriptional reporters, SEM-2 is required to regulate hlh-8 expression in the M lineage.

sem-2 P158S mutants exhibit abnormal expression of several HLH-8 direct target genes and have defects in SM proliferation and egg-laying muscle differentiation

HLH-8 is known to directly, but differentially, regulate the expression of several reporter genes in the M lineage, including egl-15p::gfp (expressed in vm1s), arg-1p::gfp (expressed in all vms), and NdEbox::gfp (expressed in all vms and ums) [9–11, 36, 37]. Because hlh-8 expression is reduced in sem-2 mutants carrying the P158S mutation, we asked whether these sem-2 mutants exhibited altered expression of HLH-8 target genes. As shown in Fig 7A–M, while 100% WT animals had arg-1p::gfp (N=30) expression in vms or egl-15p::gfp in vm1s (N=100), only 1.6% of sem-2(jj321[P158S]) mutants examined (N=63) expressed arg-1p::gfp in the vms. While 95% sem-2(jj321[P158S]) mutants expressed the vm1 marker egl-15p::rfp (N=138) (Fig 7A–B′), the levels of egl-15p::gfp expression in sem-2(jj321[P158S]) mutants were variable (Fig 7G). In addition, fewer cells in sem-2(jj321[P158S]) mutants expressed NdEbox::gfp, although the expression level of Ndebox::gfp appeared unchanged in the cells that expressed it (100%, N=25) (Fig 7D–D′ and 7H). Thus, sem-2 P158S mutants exhibit abnormal expression of target genes directly regulated by HLH-8.

Fig 7: SEM-2 P158S mutants have reduced numbers and deformed egg-laying muscles.

A–E′) Images of wild-type (A-E) and sem-2(jj321[P158S]) mutant (A′-E′) gravid adult animals (A-C″) or L4 animals (D-E′) showing the expression of egl-15p::rfp and hlh-29p::gfp (A′-A″), arg-1p::gfp (B-B′), rgs-2p::gfp (C-C″), NdEbox::gfp (D-D′) and hlh-8p::nls::mCherry (E-E′). sem-2(jj321[P158S]) mutants either have no expression (A′) or reduced number of cells (A″) expressing of hlh-29p::gfp, without affecting hlh-29p::gfp expression in the spermatheca (SP). arg-1p::gfp expression is completely lost in the vms, but remains in the head mesodermal cell (HMC) and the enteric muscles (EMs), in sem-2(jj321[P158S]) mutants (compare B and B′) but do not express arg-1p::gfp in the vms, sem-2(jj321[P158S]) mutants also have either have no expression (C″) or reduced number of cells (C′) expressing of the um marker rgs-2p::gfp, without affecting its expression in the nerve cord. At the L4 larval stage, sem-2(jj321[P158S]) mutants have reduced number of cells expressing NdEbox::gfp and hlh-8p::nls::mCherry (D′-E′) compared to WT (D-E). F) Graphs showing the number of vm1s and vm2s as indicated by the expression of egl-15p::rfp (vm1s) and hlh-29p::gfp (vm2s) (F), the expression level of egl-15p::rfp (G), and the number of Ndebox::gfp-expressing cells (H) in wild-type and sem-2(jj321[P158S]) animals. Each dot represents an animal scored. Statistical significance was calculated by performing unpaired two-tailed Student’s t-tests. * P<0.05, **** P<0.0001.

HLH-8 is known to regulate the proper proliferation of the SMs, and the proper differentiation and function of the vulval muscles [11, 12]. Consistent with a role of SEM-2 in regulating the expression of hlh-8, sem-2(P158S) mutants also exhibited defects in SM proliferation and vulval and uterine muscle differentiation. While sem-2(jj321[P158S]) mutants have two SMs, as indicated by the expression of the endogenously tagged GFP::2xFLAG::SEM-2 in sem-2(jj382 jj417[GFP::2xFLAG::SEM-2(P158S)] animals (92%, N=25) (Fig 5J–N), the SMs do not divide at the wild-type rate (Fig 8A–F‴). At late L4 larval stage when wild-type animals have 16 SM descendants, sem-2(jj417[P158S]) mutants have fewer than 16 cells resembling SMs in the developing vulva region (Fig 7E–E′ and Fig 8A–F‴). Consistent with an SM proliferation defect, only 28% of sem-2(jj321[P158S]) mutants (N=139) expressed the vm2 marker hlh-29p::gfp [38], and 50% of sem-2(jj321[P158S]) mutants (N=20) expressed the um marker rgs-2p::gfp [39], while 95% of sem-2(jj321[P158S]) mutants (N=138) expressed the vm1 marker egl-15p::rfp (Fig 7A–A″ and 7C–C″ and 7F). Notably, the vulval muscles in sem-2(jj321[P158S]) appeared deformed (Fig 7A–A″). Similarly, while all wild-type animals have 16 cells expressing NdEbox::gfp, sem-2(jj321[P158S]) mutants have a range of 0–12 NdEbox::gfp-positive cells (Fig 7D–D′ and 7H). Thus, SEM-2 is required for the proper proliferation of the SMs and differentiation of the various egg-laying muscles.

Fig 8: SEM-2 P158S mutants have SM proliferation defects.

Fluorescence and DIC images showing SMs and SM descendants labelled by hlh-8p::nls::mCherry (A–F) and GFP::2xFLAG::SEM-2 (A′–F′) in wild-type (sem-2(jj382)) (A–A‴, C–C‴, E–E‴) and sem-2(jj382 jj417[GFP::2xFLAG::SEM-2(P158S)]) mutant (B–B‴, D–D‴, F–F‴) animals at the L3 (A–B‴), young L4 (C–D‴), and mid L4 (E–F‴) stages. Corresponding merged and DIC images are shown in A″–F″ and A‴–F‴, respectively.

A putative SoxC-binding site is critical for hlh-8 promoter activity in the M lineage

We then sought to determine if SEM-2 might directly regulate the expression of hlh-8 in the M lineage. We identified several regions in the hlh-8 promoter that are highly conserved among multiple nematode species (S3A Fig). One of these conserved regions contains a putative Sox transcription factor binding site, which we named Site1, based on the transcription factor binding site identifier PROMO (https://alggen.lsi.upc.edu). The consensus sequence for SoxC binding is CA/TTTGTT [13, 40, 41]. Site1 contains the exact sequence for SoxC binding AACAAAGaagaag and is located at −272bp to −259bp upstream of the hlh-8 start codon (S3 Fig). Another region that we named Site2, has a sequence that matches the consensus sequence for SoxC binding by 6 out of 7 nucleotides (CTTTCTTttc) and is located at −221bp to −211bp upstream of the hlh-8 start codon (S3 Fig). We then performed transgenic reporter assays by generating a series of deletion constructs in the hlh-8 promoter and testing the ability of each deleted hlh-8 promoter to drive GFP expression in the M lineage. As shown in Fig 9A–H″, we uncovered two 20bp regions required for hlh-8 promoter activity in the M lineage: one between −280bp and −260bp, which we named E1, and another between −220bp and −200bp, which we named E2 (Fig 9A–H″ and S3A Fig). These two regions respectively correspond to the Site1 and Site2 elements described above (S3A Fig).

Fig 9: Two 20bp regions in the hlh-8 promoter are necessary for hlh-8 promoter activity in the M lineage.

A) Schematics of the wild-type construct and 50bp deletions in a hlh-8 transgenic transcriptional reporter that has 517bp of the hlh-8 promoter located immediately upstream of its start codon driving the expression of gfp. B) Schematics of hlh-8 transcriptional reporter constructs with 20bp deletions in a 100bp region located at −300bp to −200bp upstream of the start codon of hlh-8. E1 and E2 contain putative SEM-2/SoxC-binding sites and are indicated by the color magenta. C–E″) Representative fluorescence images of reporter expression (C′-H′) in the M mesoblast cell of L1 animals (C-E″) and in SMs of L3 animals (F-H″) labelled by the hlh-8p::nls::mCherry reporter (C–H). F″–H″ are merged images.

We then tested the importance of the putative SoxC-binding sites in Site1 and Site2 in vivo, by mutating them in the endogenous hlh-8 transcriptional reporter background, jj422[hlh-8p::gfp]. We generated three alleles: jj483 jj422 [hlh-8p(Site1m+Site2m)::gfp], which has a 13bp mutation with the putative SoxC-binding site in Site1 mutated and a 10bp mutation with the putative Sox-Cbinding site in Site2 mutated, as well as jj445 jj422 [hlh-8p(Site1m)::gfp] and jj446 jj442 [hlh-8p(Site1m)::gfp], which each have a 13bp mutation with the putative SoxC-binding site in Site1 mutated (Fig 10A, Supplementary tables 1 and 2). All three alleles resulted in reduced or undetectable levels of GFP in the SMs: jj483 jj422 [hlh-8p(Site1m+Site2m)::gfp] have undetectable levels of GFP in 97.4% of SMs scored (N=39), jj445 jj422 [hlh-8p(Site1m)::gfp] and jj446 jj442 [hlh-8p(Site1m)::gfp] have undetectable levels of GFP in 96.1% of SMs scored (N=52). In contrast, wild-type animals have detectable levels of GFP in 99% of all the SMs scored (N=96). The few animals carrying Site1 and Site2 mutations (jj483) or only Site 1 mutations (jj445/jj446) that expressed jj422[hlh-8p::gfp] in the SMs had significantly reduced expression compared to wild-type animals (Figure 10). These results indicate that the putative SoxC-binding site in Site1 is important for hlh-8 expression in the M lineage. Intriguingly, jj422[hlh-8p::gfp] expression at the 16SM stage in the three mutants appeared comparable to that in wild-type animals (Fig 10E–G″), suggesting that mutating the putative SoxC-binding sites significantly reduces, but does not completely abolish, hlh-8 expression in the M lineage. Consistent to this notion, none of these three mutants exhibited an Egl phenotype, a hlh-8 null-like phenotype. We reasoned that there might be additional SoxC-binding site(s) in the endogenous hlh-8 genomic region that contributed to the activation of hlh-8 expression in the absence of the two putative SoxC-binding sites in Site1 and Site2.

Fig 10: A putative SEM-2/SoxC-binding site is important for robust endogenous hlh-8 expression in the M lineage.

A) A schematic showing mutations made in two putative SEM-2/SoxC-binding sites in the hlh-8 promoter in worms expressing an endogenous hlh-8 transcriptional reporter, jj422[hlh-8p::gfp]. The sequences of E1 (20bp region identified in Fig 9) and E2 (20bp region identified in Fig 9) are listed. The putative SEM-2/SoxC-binding sites are highlighted in gray. Site1 (13bp) and Site2 (10bp) are indicated by uppercase letters. The red nucleotides shown are not included in E1 or E2 but are a part of Site1 or Site2. B–D″) Fluorescence images (B′–D′) of a wild-type (hlh-8(jj422)) (B–B″), a Site1 and Site2 mutant (hlh-8(jj483 jj422)) (C–C″), and a Site1 mutant (hlh-8(jj445 jj422)) (D–D″) L3 animal showing expression of jj422[hlh-8p::gfp] in SMs labelled by the hlh-8p::nls::mCherry reporter (B–D). Merged images are shown in B″–D″. jj483 jj422 [hlh-8p(Site1m+Site2m)::gfp] have undetectable levels of GFP in 97.4% (N=39) of SMs scored, jj445 jj422 [hlh-8p(Site1m)::gfp] and jj446 jj442 [hlh-8p(Site1m)::gfp] have undetectable levels of GFP in 96.1% of SMs scored (N=52), while wild-type animals have detectable levels of GFP in 99% (N=96) of SMs scored. E–L″) Fluorescence images (E′–G′) of wild-type (hlh-8(jj422)) (E–E″), Site1 mutant (hlh-8(jj445 jj422)) (F-F″), and Site1 and Site2 mutant (hlh-8(jj483 jj422)) (G–G″) animals showing expression of jj422[hlh-8p::gfp] in SM descendants labelled by the hlh-8p::nls::mCherry reporter (E–G). Merged images are shown in E″–G″.

DISCUSSION

By taking advantage of a partial loss-of-function allele of sem-2, we have identified additional functions of the single C. elegans SoxC protein, SEM-2, and uncovered previously unappreciated regulatory relationships between SEM-2 and LET-381/FoxF/C, as well as between SEM-2 and HLH-8/Twist. Our work added new subcircuits to the gene regulatory network underlying C. elegans postembryonic development [2].

SEM-2/SoxC functions antagonistically with LET-381/FoxF/C in CC fate specification

The zinc finger transcription factor SMA-9/Schnurri is known to regulate the expression of the forkhead transcription factor LET-381/FoxF/C, and both proteins are required for specifying the M-CC fate in the dorsal M lineage. Loss of either transcription factor results in a fate transformation of M-CCs to SMs due to derepression of sem-2 expression [8, 23]. In this study, we have found that the sem-2[P158S] mutation, which renders the SEM-2 protein partially functional, can suppress the loss of M-CC phenotype of sma-9(0) mutants (Susm), and that this Susm phenotype is dependent on the presence of LET-381 (Tables 1 and 2, Fig 4). These findings support a role of SEM-2 in the dorsal side of the M lineage and a mutually repressive relationship between SEM-2 and LET-381 (Fig 11A). We therefore propose a model shown in Fig 11A. Based on this model, in wild-type animals, SMA-9 functions, either acting through or independently of LET-381, to repress sem-2 expression in the presumptive CC mothers (M.dlp and M.drp) and CCs (M.dlpa and M.drpa) in the dorsal M lineage (Fig 11A). The expression of let-381 in M.dlpa and M.drpa then leads to the activation of ceh-34 expression, where LET-381 and CEH-34 function in a feedforward manner to regulate M-CC specification and function [8]. In sma-9(0) single mutants, ectopic expression of a fully functional sem-2 in the dorsal M lineage leads to the repression of let-381 expression in the presumptive CCs, causing the transformation of these cells to SMs (Fig 11B). sem-2([P158S]) single mutants look like wild-type animals regarding CC specification, because of the actions of SMA-9 and LET-381 in preventing sem-2 expression in the dorsal M lineage (Fig 11C). In sem-2([P158S]); sma-9(0) double mutants, the partially functional SEM-2(P158S) protein being expressed in the dorsal side of the M lineage is not sufficient to fully repress let-381 expression, leading to the expression of let-381 and the formation of M-CCs in a fraction of these double mutant animals (Fig 11D). This model is consistent with our previous findings that forced expression of sem-2 throughout the M lineage leads to the conversion of M derived-CCs and BWMS to SMs [6]. Moreover, it adds the possibility that SMA-9 activates let-381 expression by way of repressing sem-2 expression, forming a double negative gate. Future work will aim to determine whether SEM-2 and LET-381 directly regulate each other’s expression in the M lineage.

Fig 11: A model on the regulatory network involved in CCs specification in the dorsal M lineage.

In WT animals (A), LET-381 is expressed in the dorsal M lineage cells fated to become CCs due to the presence of SMA-9 in these cells, either by repressing the expression of sem-2, thus preventing SEM-2 from repressing the expression of LET-381, or by activating let-381 expression independently of SEM-2. Once expressed, LET-381 further represses the expression of sem-2, while at the same time, directly activates the expression of CEH-34 and functions in a feedforward manner with CEH-34 to activate genes required for the differentiation and function of CCs. In sma-9(0) mutants (B), the de-repression of SEM-2 expression leads to the conversion of CCs to SMs as the fully functional SEM-2 represses let-381 expression. The partially functional SEM-2 protein expressed in sem-2(P158S) mutants (SEM-2*) on its own (C) is not expected to affect the CC fate. However, in sma-9(0) mutants, the de-repressed expression of a partially functional SEM-2(P158S) (SEM-2*) protein is not sufficient to completely repress LET-381 expression, thus leading to a fraction of sem-2(P158S); sma-9(0) double mutants producing CCs in the dorsal M lineage (D). Dotted line in panel D indicates that SEM-2 P158S (SEM-2*) is not able to fully repress let-381 expression.

The repression of sem-2 expression in the M lineage by SMA-9 appears to be stage-specific, as sem-2 expression in the early M lineage (1-M stage to 8-M stage) does not change in sma-9(0) mutants (S1 Fig). SMA-9 does not appear to be the only factor repressing sem-2 expression in the non-SM cells at the 16- to 18-M stage. We have previously shown that HLH-1 and FOZI-1 repress sem-2 expression in the M-derived BWMs, and that this genetic interaction is reciprocal, as SEM-2 is known to repress the expression of hlh-1 and fozi-1 in the SM mother cells and the SMs [6]. Similarly, expression of sem-2 in the SM mother cells and the SMs requires LIN-12/Notch signaling and the zinc finger transcription factor SEM-4 [6, 42], implicating additional levels of complexity in the regulatory network underlying proper fate specification in the M lineage. It is clear that the specification of M-CCs and SMs involves intricate gene regulatory networks that include both positive and negative feedback, feedforward, and mutually antagonistic regulatory subcircuits. As previously suggested [43], such regulatory logic ensures temporal and spatial specificity of gene expression and robust cell fate specification.

SEM-2/SoxC is necessary for the specification, proliferation, and differentiation of the SMs and SM descendants

Previous studies have shown that SEM-2 specifies the multipotent and proliferative SM fate [6] (Tian et al. 2011). Since the SMs are not made in sem-2(n1343) mutants that were used in the previous study, we could not determine if SEM-2 plays a role beyond SM specification in the SM lineage. In most sem-2 P158S mutants, the level of functional SEM-2 in the M lineage is sufficient to specify the SM fate (Fig 5J–M″). However, the SMs in sem-2 P158S mutants exhibit reduced proliferation. These results provide direct evidence supporting a role of SEM-2 in regulating cell proliferation. It is possible that SEM-2 functions upstream of, and/or works cooperatively with, certain cell cycle regulators, including those described in Kipreos and van den Heuvel [44]. This role of SEM-2 in regulating cell proliferation appears to be evolutionarily conserved. Sox4, one of the SoxC proteins in humans, is often amplified and overexpressed in multiple cancers, and Sox4 is known to play crucial roles in cancer development and progression, and has been classified as a “cancer signature” gene [16, 18].

In addition to regulating SM specification and proliferation, SEM-2 is essential for the proper differentiation of multiple non-striated muscle cells derived from the SM lineage. We have shown that while the vm1-specific reporter (egl-15p::rfp) is expressed in over 90% of SEM-2 P158S mutants, a significantly smaller fraction of SEM-2 P158S mutants animals express the vm2-specific reporter (hlh-29p::gfp, 28%), the um-specific marker (rgs-2p::gfp, 50%), or another vm marker (arg-1p::gfp, 1.6%) (Fig 7). These findings suggest that SEM-2 is important for the proper differentiation of the various non-striated muscles derived from the SM lineage. The more prevalent expression of the vm1-specific reporter (egl-15p::rfp) in SEM-2 P158S mutants is similar to our previous report showing that in lin-39(0) mab-5(0) mutants where the M lineage exhibits reduced cell proliferation, the few M lineage cells precociously differentiate to express the vm1 marker egl-15p::gfp, but not other vm2s or ums markers [45]. This shared, reduced proliferation phenotype by lin-39(0) mab-5(0) mutants and SEM-2 P158S mutants is consistent with MAB-5 and LIN-39 directly activating the expression of SEM-2 in the M lineage [6].

The SoxC-Twist axis as a conserved regulatory cassette in metazoan development

Multiple lines of evidence support the role of SEM-2 in regulating the expression of hlh-8/Twist in the M lineage, possibly directly. First, both transgenic hlh-8 transcriptional reporters (hlh-8p::gfp and hlh-8p::nls::mCherry) and an endogenous hlh-8 transcriptional reporter all exhibited significantly reduced expression in sem-2 P158S mutants (Fig 5 and Fig 6). Moreover, the sem-2 P158S phenotype is similar to the hlh-8 mutant phenotype. Both sem-2 P158S mutants and hlh-8 null, nr2061, mutants are Egl due to missing vulval muscles (this study, [12]). Additionally, animals with a semidominant allele of hlh-8, n2170 (E29K), have SMs that often fail to divide [11], similar to the SM proliferation defect in the sem-2 P158S mutants. Finally, sem-2 P158S mutants exhibit reduced, yet differential expression of several HLH-8 target genes, such as egl-15p::gfp and arg-1p::gfp, a phenotype that has been previously observed in various hlh-8 mutants. For example, several hlh-8 point mutants (R103M, R103A, L95F and F99L) express egl-15p::gfp, but do not express arg-1p::gfp, in the vulval muscles [37]. Similarly, worms containing hlh-8(tm726), a 646-nucleotide deletion at the 3′ end of intron one, express egl-15p::gfp (15% of animals) but do not express arg-1p::gfp [10], whereas animals that are heterozygous for hlh-8(n2170) express arg-1p::gfp but do not express egl-15p::gfp [11]. These results are consistent with altered or reduced endogenous hlh-8 expression in sem-2 P158S mutants.

It is likely that SEM-2 directly regulates the expression of hlh-8 in the M lineage, through at least one of the putative SoxC-binding sites (Site1) in the hlh-8 promoter (Fig 10A, S3 Fig). We have shown that Site1 is essential for hlh-8 promoter activity in transgenic reporter assays and is important for hlh-8 expression in the endogenous genomic environment (Fig 9 and Fig 10). Mutating the putative SoxC-binding sites in Site1 and Site2 significantly reduced, but did not completely abolish, hlh-8 expression in the endogenous locus, possibly due to the presence of other putative SoxC-binding sites in the hlh-8 genomic region. Nevertheless, our results collectively are consistent with SEM-2 playing an important, likely a direct, role in regulating the expression of hlh-8 in the M lineage.

In humans Twist1, Twist2, FGFRs and JAG-1/Notch2 are known to play important roles in craniofacial development [46, 47]. The C. elegans homolog of Twist1/2 is HLH-8, the homolog of the FGFRs is EGL-15, and the homolog of JAG-1 is ARG-1 [11, 37, 48]. egl-15 and arg-1 are direct targets of HLH-8, and they have each been shown to be expressed in and/or to work in patterning a subset of mesodermal tissues: the egg-laying muscles and the enteric muscles [11, 12]. Mutations in hlh-8 and egl-15 lead to Egl and/or Con phenotypes, which have been labelled as phenologs of craniofacial defects in humans [37]. The SEM-2 P158S mutants are 100% Egl (Fig 1F–G, Table 1B). Intriguingly, mutations in SoxC proteins, Sox4 and Sox11, are associated with a developmental disorder called Coffin-siris syndrome (CSS), and one key characteristic of CSS patients is craniofacial defects [19–21]. Similarly, SoxC proteins are known to function upstream of Twist1, in some cases directly, in disease initiation and progression in mammals, particularly, in the regulation of epithelial-mesenchymal transition (EMT) [16, 49–51]. Thus, the SoxC-Twist axis, including the downstream targets of Twist, such as FGFRs and JAG-1, represents an evolutionarily conserved regulatory cassette important in metazoan development.