C. elegans ADAMTS ADT-2 regulates body size by modulating TGFβ signaling and cuticle collagen organization

Abstract

Organismal growth and body size are influenced by both genetic and environmental factors. We have utilized the strong molecular genetic techniques available in the nematode C. elegans to identify genetic determinants of body size. In C. elegans, DBL-1, a member of the conserved family of secreted growth factors known as the Transforming Growth Factor β superfamily, is known to play a major role in growth control. The mechanisms by which other determinants of body size function, however, is less well understood. To identify additional genes involved in body size regulation, a genetic screen for small mutants was previously performed. One of the genes identified in that screen was sma-21. We now demonstrate that sma-21 encodes ADT-2, a member of the ADAMTS (a disintegrin and metalloprotease with thrombospondin motifs) family of secreted metalloproteases. ADAMTS proteins are believed to remodel the extracellular matrix and may modulate the activity of extracellular signals. Genetic interactions suggest that ADT-2 acts in parallel with or in multiple size regulatory pathways. We demonstrate that ADT-2 is required for normal levels of expression of a DBL-1-responsive transcriptional reporter. We further demonstrate that adt-2 regulatory sequences drive expression in glial-like and vulval cells, and that ADT-2 activity is required for normal cuticle collagen fibril organization. We therefore propose that ADT-2 regulates body size both by modulating TGFβ signaling activity and by maintaining normal cuticle structure.

Introduction

Body size is a fundamental feature of an organism critical to survival and fitness, yet the mechanisms underlying its regulation remain incompletely understood. Existing studies show that body size is in part genetically controlled (Oldham et al., 2000). For example, in Drosophila, the insulin/insulin-like growth factor 1 (IGF1) signaling pathway regulates body size, so that mutants are smaller than wild-type flies (Bohni et al., 1999; Leevers et al., 1996; Neufeld and Edgar, 1998). Flies mutant for Drosophila S6 kinase (DS6K) and d-Myc also have slow growth, reduced cell size and small body size (Gallant et al., 1996; Johnston et al., 1999; Montagne et al., 1999). The Drosophila homolog of the target of rapamycin (TOR) affects growth by modulating the activity of DS6K. Mutant cells are small in size (Neufeld, 2003; Oldham et al., 2000). In contrast, the Salvador–Warts–Hippo (SWH) pathway represses tissue size. Lack of SWH pathway component activity results in overgrowth of the adult structures (Badouel et al., 2009; Tyler and Baker, 2007; Willecke et al., 2006).

In C. elegans, there are several known small body size mutants. Components of the DBL-1 Transforming Growth Factor β (TGF-β) pathway play a major role in the regulation of body size. Mutations in any component of the pathway, dbl-1 (ligand), sma-6 (type I receptor), daf-4 (type II receptor), sma-2, sma-3, sma-4 (Smad transcription factors) and sma-9 (transcription co-factor) result in smaller than wild-type body size (Estevez et al., 1993; Krishna et al., 1999; Liang et al., 2003; Savage-Dunn, 2005; Savage-Dunn et al., 2003; Savage et al., 1996; Suzuki et al., 1999).

In addition to the DBL-1 pathway, other genes have been identified that play less critical roles in regulation of body size. One such group is genes that are expressed in sensory neurons, including che-2, che-3, egl-4, tax-6 and cnb-1 (Bandyopadhyay et al., 2002; Fujiwara et al., 2002; Kuhara et al., 2002). The che-2 and che-3 mutations cause small body size due to defects in sensory perception. egl-4 (cGMP-Dependent Protein Kinase) acts downstream of che mutants to regulate body size by repressing the DBL-1 pathway (Fujiwara et al., 2002). tax-6 and cnb-1 encode the catalytic and regulatory subunits of calcineurin, respectively. tax-6 interacts with kin-29 (ser/thr kinase) and mef-2 (MADS box transcription factor) to regulate body size (Singaravelu et al., 2007).

Feeding defective mutants also have small body size. These include pha-2 and pha-3 with abnormal pharyngeal anatomy, eat-1, eat-2, eat-3 with reduced pumping rates and eat-10 with inefficient pharyngeal pumping (Morck and Pilon, 2006). Mutations of components of the TORC2 complex result in small body size (Jones et al., 2009; Soukas et al., 2009). Intriguing recent evidence implicates the cell death machinery in the regulation of cell and body size (Chen et al., 2008). In addition, mutations that affect the structure of the cuticle can change the body size of the animal because the cuticle encapsulates the body. Some examples are dpy-2, dpy-7, dpy-10, dpy-13, sqt-1, sqt-3 and lon-3. All of these genes encode cuticular collagens (Johnstone et al., 1992; Kramer, 1994; Kramer and Johnson, 1993; Kramer et al., 1988; Levy et al., 1993; Nystrom et al., 2002; Suzuki et al., 2002; van der Keyl et al., 1994; von Mende et al., 1988). Finally, mutations in a small number of genes which have yet to be assigned to a particular pathway, including sma-1 (βH-spectrin), sma-5 (MAP kinase BMK1/ERK5 homolog) and rnt-1 (RUNX family transcription factor), also cause small body size (Ji et al., 2004; McKeown et al., 1998; Watanabe et al., 2005).

To understand the genetic basis of body size regulation, a forward genetic screen for small body size mutants was carried out (Savage-Dunn et al., 2003). In that screen, alleles of many of these genes were identified. Furthermore, one of the novel small body size mutants isolated was sma-21, which we now show is allelic with ADAMTS family member adt-2.

ADAMTS (a disintegrin-like and metalloprotease with thrombospondin type I motif) are secreted metalloproteases that bind to the extracellular matrix (ECM) (Kuno and Matsushima, 1998; Porter et al., 2005; Tang, 2001). ADAMTS are related to the ADAM (a disintegrin and metalloprotease) subfamily of transmembrane proteins. Both ADAM and ADAMTS proteins are Zn dependent metalloproteases (Jones and Riley, 2005; Kaushal and Shah, 2000; Stocker et al., 1995). In mammals, the ADAMTS proteases are believed to function in ECM assembly (procollagen N-proteinases: ADAMTS-2, -3 and -14) and ECM degradation (aggrecanases: ADAMTS-1, -4, -5, -8, -9 and -15) (Colige et al., 1995; Colige et al., 1997; Collins-Racie et al., 2004; Fernandes et al., 2001; Kuno et al., 2000; Somerville et al., 2003; Tortorella et al., 2000; Tortorella et al., 2005; Wang et al., 2003). In C. elegans, some of these proteases are involved in organogenesis by remodeling the ECM. For example, GON-1 and MIG-17 are involved in distal tip cell migration in gonad development (Blelloch and Kimble, 1999; Ihara and Nishiwaki, 2007). Another protease ADT-1 is involved in ray morphogenesis by rapid remodeling of ECM (Kuno et al., 2002).

In this study we describe the role of C. elegans ADAMTS gene adt-2 in the control of body size. Genetic interactions show that adt-2 is likely involved in multiple pathways that regulate body size. We show that ADT-2 is synthesized in glial cells of sensory neurons and in the vulva, and is required to promote DBL-1 signaling activity and for normal cuticle structure.

Results

Isolation of sma-21 mutants

To identify genes required for body size regulation, a forward genetic screen for small body size mutants was carried out (Savage-Dunn et al., 2003). In that screen, N2 hermaphrodites were mutagenized with ethyl methanesulfonate (EMS) and the F2 progeny worms were screened for small body phenotype. In the screen, sma-20 mutants were identified. In the course of mapping, we found that the sma-20 strain has two mutations, sma-20(wk31) and sma-21(wk156). Furthermore, 23 semi-small segregants were isolated from the double mutant strain and it was found that they were all allelic to sma-21. This finding led us to the conclusion that sma-21 regulates body size, and that sma-20 is an enhancer of the sma-21 phenotype with no apparent phenotype on its own. All of the analyses in this paper were performed with a sma-21(wk156) segregant from which the sma-20 enhancer mutation was outcrossed.

sma-21 encodes an ADAMTS (disintegrin and metalloprotease with thrombospondin repeats) family member ADT-2

We used single nucleotide polymorphism (SNP) mapping (Davis et al., 2005) to locate sma-21 on the X chromosome in the region between 7,439,984 (cosmid C01C10) and 7,982,355 (cosmid F45E1). Then, we employed array comparative genomic hybridization (aCGH) to identify any polymorphisms in this region. An oligonucleotide chip spanning this region was designed. This was hybridized with sma-21 and wild-type genomic DNA at Roche NimbleGen (Maydan et al., 2009). This analysis identified two polymorphisms in the sma-21 mutant: one in K09F5.1 and one in adt-2. Sequence analysis of PCR fragments in sma-21 verified the two SNPs: one at 7590038 bp (G to A) in the adt-2 gene and the other at 7740620 (G to A) in K09F5.1. To determine which of these mutations is responsible for the small body size phenotype in sma-21, we used RNAi and transformation rescue. We inactivated the corresponding genes by feeding N2 (wild type) and rrf-3 (RNAi hypersensitive) worms on adt-2 RNAi and K09F5.1 RNAi plates (Kamath et al., 2001). adt-2 RNAi fed worms are small unlike the K09F5.1 RNAi (Fig. 1). Then, we introduced fosmid clones containing adt-2 wild-type sequences into sma-21 mutants by microinjection (Mello et al., 1991). The fosmid clone WRM0636aH08 which contains the full length adt-2 wild-type gene rescues the body size of sma-21(wk156) (Fig. 1). Finally, we performed a complementation test between sma-21 and a deletion allele of adt-2(tm975) obtained from the CGC. The complementation test shows that these two mutations fail to complement for body size. Putting all the data together, we can conclude that sma-21 encodes adt-2. We will therefore refer to sma-21 as adt-2.

Fig. 1.

sma-21 encodes ADT-2. (A,B,C) RNAi inactivation of adt-2 results in small body size. K09F5.1(RNAi) treated worms are indistinguishable from control worms. rrf-3 RNAi hypersensitive strain is used for body size measurements. (D,E,F) Fosmid WRM0636aH08 containing adt-2 gene rescues the body size of sma-21(wk156). A,B,D and E are adult worms photographed at the same magnification. Body length was measured in adult worms. Each value represents a mean of 30 – 60 worms. Error bars indicate the standard deviation. Scale bars = 0.2mm. **indicates p<0.01

In order to verify the exon-intron structure of adt-2, we sequenced cDNA clones obtained from Dr Yuji Kohara. Sequencing of multiple cDNA clones gave no evidence for alternative splicing. The sequence of EST yk1586e04 contains the full-length transcript revealing that adt-2 consists of 20 exons (exon 1 is non-coding) that encode a protein of 1020 amino acids. ADT-2 belongs to the ADAMTS family of secreted extracellular metalloproteases. These proteins bind to extracellular matrix (ECM) and are believed to be involved in remodeling of the ECM (Porter et al., 2005). As shown in Table 1, the catalytic domain of ADT-2 shows 21% to 41% homology to other ADAMTS family members. These proteins are comprised of several domains. Starting with the N terminus, ADT-2 has a signal peptide (predicted by SignalP 3.0 Server); a catalytic domain with a reprolysin- type zinc-binding motif, HEX1X2HX3X1GX1XHD [where X1 is typically hydrophobic aa, X2 is glycine or a hydrophobic aa and X3 is asparagine (http://www.lerner.ccf.org/bme/apte/adamts/domain_organization.php)]; a cysteine-rich domain; a central TS (thrombospondin type I-like) domain and six C-terminal TS repeats (predicted by SBASE release 14, Sept 2006) (Fig. 2A). The domain organization of mammalian ADAMTS-2 is shown for comparison (Fig. 2A). Table 1 shows the amino acid identity in the zinc binding motif, catalytic domain and in the first thrombospondin type I-like repeat of C. elegans ADT-2 compared to mammalian and C. elegans ADAMTS. The metalloprotease domain of ADT-2 shows the highest similarity to ADAMTS-2, -3, and -14 among mammalian ADAMTS family members (Table 1 and Fig. 2D).

Table 1.

Amino acid identity of C. elegans of ADT-2 to mammalian and C. elegans ADAMTS. Calculated using Clustalw.

	Catalytic domain	First TSP type 1 motif	Zn binding motif
C. elegans GON-1	26	35	33
C. elegans ADT-1	41	37	36
Mouse ADAMTS-1	21	33	36
Human ADAMTS-4	25	33	54
Human ADAMTS-5	22	32	31
Human ADAMTS-2	29	37	36
Human ADAMTS-3	26	35	36
Human ADAMTS-14	25	32	36

Fig. 2.

Structure of the ADT-2 protein. (A) Domain organization of ADT-2 protein. Domain organization of mammalian ADAMTS-2 is shown for comparison (predicted by SBASE and SignalP 3.0 Server). (B) Amino acid sequence of ADT-2. Different domains are highlighted. In red is the signal sequence; in violet, the prodomain; in brown, the catalytic domain. Underlined is the Zn binding motif. In green is the conserved methionine residue downstream of the metal binding motif which forms ‘Met-turn’ (Porter et al., 2005). In yellow is the cysteine rich domain. Thrombospondin type I-like repeats are highlighted in blue. The location of the deletion in tm975 and the missense mutation (Gly³⁶⁴ – Ser) in wk156 are indicated by a shaded box and an asterisk, respectively. (C) Zinc binding motif and Met turn of C. elegans ADT-2 are aligned with other mammalian and C. elegans ADAMTS family members. The conserved zinc binding motif and methionine residue downstream of the metal binding site are shaded. Asterisks indicate conserved amino acids. The glycine in red is the conserved residue mutated in adt-2(wk156). (D) Phylogenetic tree showing relationships between C. elegans and human ADAMTS sequences (constructed using BioEdit sequence Alignment Editor and MEGA 4.0.2). The numbers above the nodes indicate the percent bootstrap values in 500 replicates of the data.

adt-2(wk156) is a missense mutation that changes a conserved glycine residue to a serine (highlighted in red in Fig. 2C) within the metal binding motif. Conservation of this glycine residue suggests that it is functionally important. adt-2(tm975) has a deletion of 475bp (WormBase; Fig. 2B) which deletes 92 amino acids within the catalytic domain including the metal binding motif. adt-2(tm975) is lethal. Homozygous adt-2(tm975) mutants die at the 3-fold stage of embryogenesis or during hatching (Fig. 3B and C).

Fig. 3.

Phenotypes of adt-2 mutants. (A) adt-2(wk156) mutant has small body size. Growth curve of adt-2 showing reduced growth rate compared to wild-type N2 worms. Each time point represents a mean of 25–50 animals. Error bars indicate the standard deviation. Body size of adt-2 mutant at each point except 24 hrs is significantly different from that of wild-type. (B,C) Lethal phenotype of adt-2(tm975) deletion homozygotes. Images show an unhatched embryo (three-fold stage; B) and a partially hatched larva (C). D) Reduced lifespan of adt-2 mutants. A Kaplan-Meier survival curve shows the lifespan of wild-type N2 (blue; median survival 16 days), mutant adt-2(wk156) (brown; median survival 10 days; logrank statistical comparison p<0.0001 relative to N2), adt-2(RNAi) (orange; median survival 8 days; logrank statistical comparison p<0.0001 relative to N2) and adt-2(RNAi) initiated in adulthood (green; median survival 14 days; logrank statistical comparison p<0.0001 relative to N2). For each strain, n=120 worms.

Phenotypic characterization of adt-2

To characterize the body size phenotype of adt-2(wk156), we created growth curves that indicate the body length of wild-type N2 worms and adt-2(wk156) worms at various times after egg collection (Fig. 3A). Newly hatched adt-2 mutant larvae have the same body length as control animals, similar to dbl-1 pathway mutants. adt-2 mutant worms grow more slowly after 24 hours resulting in adult worms with about a 20 percent reduced body length when compared to wild type. Interestingly, adult adt-2 mutants show a significant decrease in body length between 96 and 120 hours, whereas wild-type animals continue to grow larger during adulthood. The reduction in body length in adt-2 mutants suggests that the gene is required for maintenance of body length as well as for increase in body length during growth stages.

We have also examined the body width of adt-2 mutants. Typical small mutants, such as those defective in the DBL-1 pathway, have reduced width as well as reduced length. In contrast, dumpy mutants, such as those defective in some cuticle collagens, have reduced length but not reduced width. adt-2(wk156) mutants show a significantly increased body width, 0.072 mm +/− 0.005 mm vs 0.066 mm +/− 0.005 mm for N2 wild-type controls grown concurrently under the same conditions, p<0.01 (see also Fig. 1E). In contrast, adt-2(RNAi) animals show a significantly reduced body width, 0.072 mm +/− 0.006 mm vs 0.089 mm +/− 0.006 mm for control animals grown concurrently under the same conditions, p<0.01 (see also Fig. 1B). Thus, the partial loss-of-function adt-2(wk156) mutant resembles a dumpy mutant, whereas more severe loss-of-function due to RNAi inactivation results in a more characteristic small phenotype.

We next evaluated the effects of adt-2 on lifespan. Both adt-2(wk156) and adt-2(RNAi) have significantly reduced lifespans compared to wild type, with a median survival of 10 and 8 days from adulthood, respectively, compared with 16 days for wild type (Fig. 3D). Since early lethality could be due to developmental defects associated with adt-2 loss of function, we also measured lifespan of animals grown under normal conditions until adulthood and then transferred to adt-2(RNAi) conditions. These animals also show reduced adult longevity, with a median survival of 14 days and a maximum survival reduced from 21 days to 18 days (Fig. 3D). Overall, the median lifespan of adt-2 mutants and adt-2(RNAi) under conditions tested is reduced by 13% to 50% relative to wild type. Thus, adt-2 activity is required to promote normal adult longevity.

The reduction in body length of adt-2 mutants could be due to decreased cell number, cell size or alterations in cellular or extracellular morphology. In order to determine which cause is responsible in this case, we first examined the number of seam cells, the number of intestinal cells, and the number of hypodermal nuclei in wild-type and adt-2 mutant animals. The number of seam cells was measured using the seam cell marker::GFP reporter, and found to be not significantly different in adt-2(wk156) or adt-2(RNAi) compared to controls (Table 2). DAPI staining was done in order to count intestinal nuclei, which are the same in adt-2(wk156) and adt-2(RNAi) compared with controls (Table 2). We also counted the number of nuclei in the hypodermis of fourth larval stage (L4) worms of adt-2(wk156), adt-2(RNAi), and control animals. Hypodermal nuclei were counted using the dpy-7p::2Xnls::yfp marker, which is expressed in hypodermal cells beginning at the L1 stage (Myers and Greenwald, 2005). The number of hypodermal nuclei is not significantly different in adt-2 mutants compared to control animals (Table 2). For comparison, we determined the number of hypodermal nuclei in dbl-1 mutants, which is also not significantly different from wild type (114.6 +/− 9.4). These results are consistent with previous reports that DBL-1 pathway mutants have normal numbers of hypodermal nuclei (Nagamatsu and Ohshima, 2004).

Table 2.

Cell and nuclei measurements of adt-2 and wild-type worms.

		N2	adt-2(wk156)	RNAi control	adt-2(RNAi)
A	Seam cell number	16 +/− 0	16 +/− 0	16.1 +/− 0.7	16.2 +/− 0.4
	(adult)	(n=30)	(n=30)	(n=10)	(n=20)
	Intestinal nuclei number	31.4 +/− 1.2	31.5 +/− 1.4	32.7 +/− 1.4	33.4 +/− 1.0
	(adult)	(n=31)	(n=38)	(n=10)	(n=20)
	Hypodermal nuclei number	116.3 +/− 11.9	114.1 +/− 10.7	145.4 +/− 6.3	145.9 +/− 4.0
	(L4)	(n=33)	(n=31)	(n=5)	(n=10)
	Seam cell perimeter (mm)	0.073 +/− 0.011	0.067** +/− 0.011
	(L3)	(n=78)	(n =76)
	Pharynx length (mm)	0.109 +/− 0.006	0.108 +/− 0.009
	(L3)	(n=18)	(n=18)
	Body size (mm)	0.651 +/− 0.053	0.575** +/− 0.065
	(L3)	(n=23)	(n=27)

^**p<0.01 compared to control

We also measured the dimensions of two accessible tissues (seam cells and the pharynx) in adt-2 and wild type. We crossed an ajm-1::gfp seam cell marker into adt-2(wk156) or fed the strain adt-2(RNAi) bacteria and observed fluorescence in the L3 larval stage. The ajm-1:: gfp marker localizes at the adherens junctions in the seam cells (Mohler et al., 1998). The pharynx length and body length of the same L3 worms were also measured. The length measurements for adt-2(RNAi) animals are not significantly different from controls at the L3 stage (data not shown). The perimeters of individual seam cells and the overall body length of adt-2(wk156), however, are reduced significantly compared to wild-type animals, while the length of the pharynx is not affected (Table 2). These data are consistent with the hypothesis that adt-2 mutants are smaller at least partly due to decreased cell size or altered cell morphology, rather than decreased cell number. Our previous analysis of DBL-1 pathway mutants similarly revealed size reductions in the seam cells but not in the pharynx (Wang et al., 2002). Thus, the adt-2 size defects are similar to those seen in mutants of the DBL-1 signaling pathway.

Genetic interactions of adt-2 and other body size mutants in C. elegans

The DBL-1 pathway plays a major role in body size regulation in C. elegans. Mutations in any components of the pathway, dbl-1 (ligand), sma-6 (type I receptor), daf-4 (type II receptor), sma-2, sma-3, sma-4 (Smad transcription factors), sma-9 (transcription co-factor) will result in smaller bodies than wild type (Estevez et al., 1993; Krishna et al., 1999; Liang et al., 2003; Savage-Dunn, 2005; Savage-Dunn et al., 2003; Savage et al., 1996; Suzuki et al., 1999). Double mutants were created in order to determine the relationship between adt-2 and dbl-1 pathway components. The dbl-1 pathway mutants we used for this are null or strong alleles. Double mutants of adt-2(wk156) combined with dbl-1(wk70), sma-6(wk7), daf-4(e1364), sma-2(e502), sma-3(wk30), sma-4(e729) or sma-9(wk55) have a smaller body size than the respective single mutants (Fig. 4A), indicating that adt-2 may act in a parallel pathway to regulate body size. To further validate this result, we looked at the dbl-1 overexpression [dbl-1(++)] phenotype in the adt-2 mutant background. dbl-1(++) results in a long (Lon) body phenotype in an otherwise wild-type background, but the small (Sma) phenotypes of sma-2, sma-3, sma-4, sma-6 and daf-4 are epistatic to the Lon phenotype of dbl-1(++). This epistasis places the activity of these sma genes downstream of the DBL-1 ligand activity (Suzuki et al., 1999). We fed dbl-1(++) worms on adt-2 RNAi plates. However, rather than the expected Sma phenotype, if the adt-2 gene activity were downstream, the body length is intermediate (Table 3). This is evidence that adt-2 acts at least partially independently of the DBL-1 pathway.

Fig. 4.

Genetic interactions of adt-2 and other small body size mutants in C. elegans. In each graph first two bars represents N2 and adt-2(wk156). Subsequent bars represent the indicated single mutant and corresponding double mutant with adt-2(wk156) or adt-2(RNAi). Data are shown as a percentage of wild-type body length. Each bar represents a mean of more than 27 adult animals measured at 96 hrs after embryo collection (except daf-4 and daf-4;adt-2 which were grown at 15°C for 144hrs). Error bars indicate the standard deviation. ** indicates p<0.01 for the pairwise comparison between the single mutant and the corresponding double mutant with adt-2.

Table 3.

Epistasis analysis between adt-2 and dbl-1 overexpression.

Genotype	Body length in adult animals	n
N2	1.193 +/− 0.065	33
adt-2	1.004 +/− 0.048	28
ctIs40[pTG96(sur-5::gfp) ; dbl-1(++)]	1.336 +/− 0.069	33
adt-2; ctIs40[pTG96(sur-5::gfp) ;dbl-1(++)] (1, 2)	1.177 +/− 0.07	33

¹Value is not significantly different from N2, p>0.05

²Value is significantly different from both adt-2 and ctIs40, p<0.0001

To determine whether adt-2 interacts with other pathways that regulate body size, we next tested mutants in sensory processing. che-2 and che-3 mutants have impaired sensory cilia and therefore defects in sensory perception. They also have a small body size phenotype. EGL-4 cGMP-dependent protein kinase acts downstream of che genes to regulate body size by repressing the dbl-1 pathway (Fujiwara et al., 2002). The tax-6 and cnb-1 genes encode calcineurin A and B subunits. Both tax-6 and cnb-1 are expressed in sensory neurons. Mutations in these genes cause pleiotropic defects including small body size and defects in sensory neuronal behavior (Bandyopadhyay et al., 2002; Kuhara et al., 2002). cnb-1 also has a transparent appearance because of thinning of the cuticle (Bandyopadhyay et al., 2002). In order to see whether adt-2 functions in the chemosensory pathways, double knockdowns between adt-2 and egl-4, che-2, tax-6, and cnb-1 were made by combining existing chromosomal mutations and RNAi inactivation of adt-2. Alleles chosen were likely molecular nulls or the characterized reference alleles for each gene. The egl-4(n477);adt-2(RNAi) double knockdowns have an intermediate body size compared to egl-4 and adt-2 single mutant animals. The che-2(e1033);adt-2(RNAi), tax-6(ok2065);adt-2(RNAi), and cnb-1(p675);adt-2(RNAi) double knockdowns also show additive effects (Fig. 4B). The feeding defective mutants pha-2 and pha-3 (Morck and Pilon, 2006) were also analyzed, and found to have additive effects on body size with adt-2 (Fig. 4B). A deletion allele of pha-2 is lethal, precluding the possibility of testing for body size phenotypes in the pha-2 homozygous mutant background.

To test whether adt-2 functions in the same pathways as rnt-1 and sma-1, rnt-1(ok351);adt-2(wk156) and sma-1(e30);adt-2(wk156) double mutants were constructed, using a deletion allele of rnt-1 and a nonsense allele of sma-1. The rnt-1 gene is the C. elegans homologue of mammalian RUNX transcription factors (Ji et al., 2004). sma-1 encodes for βH-spectrin (McKeown et al., 1998). Mutations in these genes cause a small body size phenotype. Double mutants of rnt-1(ok351);adt-2(wk156) and sma-1(e30);adt-2(wk156) show additive effects (Fig. 4C). Thus, we can conclude that adt-2 acts independently of all tested pathways, or is involved in multiple pathways that regulate body size.

Since adt-2 may act in multiple pathways to regulate body size, we tested whether it has an effect on a DBL-1 pathway transcriptional reporter. The RAD-SMAD reporter (a generous gift of Jun Liu) consists of tandem Smad GTCT binding sites driving expression of nuclearly localized GFP (Tian et al., 2010). GFP expression levels from this construct are positively regulated by DBL-1 signaling (Tian et al., 2010) (Fig. 5A, C, E). We tested whether RAD-SMAD expression is regulated by adt-2(RNAi). adt-2(RNAi) reduces the expression of RAD-SMAD in an otherwise wild-type background (Fig. 5A and B). LON-2 is a glypican that negatively regulates DBL-1 activity (Gumienny et al., 2007). In the lon-2 mutant background, RAD-SMAD expression is increased (Tian et al., 2010) (Fig. 5C), and this expression is also reduced by adt-2(RNAi) (Fig. 5D). In the dbl-1 mutant background, RAD-SMAD expression is reduced to a few dimly fluorescing nuclei in the posterior (Fig. 5E) and this expression is not diminished further by adt-2(RNAi) (Fig. 5F). Thus, adt-2 is a positive regulator of a DBL-1-responsive reporter. Furthermore, unlike the additive effect on body size regulation, the effects of dbl-1 and adt-2 on RAD-SMAD regulation are not additive, indicating that ADT-2 regulates this reporter via the DBL-1 pathway. We therefore conclude that ADT-2 directly or indirectly regulates DBL-1 signaling activity.

Fig. 5.

Expression of RAD-SMAD, a DBL-1 pathway transcriptional reporter. GFP expression was assessed at the L2 stage in hypodermal nuclei. (A, B) Wild-type strain fed control or adt-2(RNAi) bacteria; (C, D) lon-2 mutant fed control or adt-2(RNAi) bacteria; (E, F) dbl-1 mutant fed control or adt-2(RNAi) bacteria. Body length of the same strains was measured in adulthood, and the mean length shown in mm in each panel. In each case, the adt-2(RNAi) treatment reduced body size compared to the respective control strain significantly (p < 0.01).

adt-2 may regulate body size in part by modification of the external cuticle

Our results thus far indicate that ADT-2 may regulate body size in part by modulation of DBL-1 signaling activity, but also in part independently of the DBL-1 pathway. Since ADAMTSs are known to process ECM proteins, we considered whether adt-2 might regulate body size by targeting the cuticle. The cuticle of C. elegans is a collagen-containing extracellular matrix that surrounds the body and lines the four major openings to the exterior (pharynx, excretory pore, vulva, and anus). The cuticle is known to play a role in determining body size and morphology. In particular, the genes dpy-2, dpy-7, dpy-10, dpy-13, sqt-1, sqt-3 and lon-3 that encode cuticular collagens are required for normal body length (Johnstone et al., 1992; Kramer, 1994; Kramer and Johnson, 1993; Kramer et al., 1988; Levy et al., 1993; Nystrom et al., 2002; Suzuki et al., 2002; van der Keyl et al., 1994; von Mende et al., 1988). In addition, the adt-2(wk156) mutant phenotype is similar to the dumpy phenotypes displayed by some of these mutants. To test the hypothesis that adt-2 functions via the cuticle, we first determined the gene’s expression pattern using a transcriptional fusion of GFP to putative adt-2 promoter sequences. No expression from the transgene is seen in embryonic stages. It is possible that adt-2 is maternally provided for embryonic development, but that adt-2p::gfp expression is silenced in the maternal germline. GFP expression is detected in glial cells associated with amphid, phasmid, labial and posterior deirids (PDE) sensory neurons as early as the L1 stage and continuing throughout larval development (Fig. 6A–C). Expression is also seen in vulval tissue in L4 and adult animals (Fig. 6D). We note that these glial and vulval cells are in close association with the external cuticle, so it is possible that ADT-2 exerts its influences by proteolytic processing of cuticle proteins such as collagen. It is also possible that adt-2 is expressed in additional locations, if this construct does not contain all of the sequences necessary for normal expression of adt-2.

Fig. 6.

Expression pattern of an adt-2 reporter construct adt-2p::gfp. adt-2 is expressed in the glial cells of amphid and labial neurons (A), phasmids (B), and postdeirid (C) and in the vulva (D).

To determine whether ADT-2 is secreted to the cuticle, we examined the localization of an ADT-2::GFP fusion protein. We used fosmid recombineering (Tursun et al., 2009) to insert GFP coding sequences at the C-terminal end of the adt-2 coding sequence. Mixtures of recombineered fosmids were injected into adt-2(wk156) animals, yielding a total of nine independent transgenic lines. Six of nine lines rescue the small body size phenotype of adt-2(wk156), indicating that the GFP fusion is functional. Nonetheless, no consistent green fluorescence above background levels is seen in these lines (data not shown). The lack of GFP could indicate that ADT-2 protein does not accumulate to a high level, or that the C-terminus is proteolytically processed and degraded.

To test whether ADT-2 acts to regulate cuticle collagen organization, we used a COL-19::GFP protein fusion (Thein et al., 2003) to visualize the cuticle collagen fibrils in adt-2 mutant and RNAi animals. This reporter construct is expressed in annuli and lateral alae of the adult cuticle [(Thein et al., 2003); Fig. 7A]. In adt-2(wk156) mutants, which have the least severe body size phenotype, no significant disruption of the COL-19::GFP organization is observed. In these mutants, the annuli are constricted (Table 4), which could be either a primary or secondary defect, but is also seen in some collagen mutants (Thein et al., 2003). As shown in Fig. 7C, animals treated with adt-2(RNAi) show significant disruptions of the lateral cuticle overlying the seam cells. We also detect constrictions of the annuli in adt-2(RNAi) worms (Fig. 7C and Table 4). We analyzed COL-19::GFP localization in adt-2(tm975/wk156) transheterozygous animals, revealing patches of disorganized annuli (Fig. 7E) and structural defects in the alae. The defects are discontinuous alae and alae with two or four ridges as compared to three ridges in the wild-type worms (Fig. 7B, D and F). These disruptions may partially explain the reduced body size in adt-2 mutants and knockdowns.

Fig. 7.

Aberrant COL-19::GFP localization in adt-2 mutants. (A,C,E) COL-19::GFP expression. (B,D,F) Nomarski microscopy image of cuticular alae. In wild-type animals, COL-19::GFP fusion protein is expressed in annuli extending over most of the dorsal and ventral cuticle (A). The cuticle of adult wild-type hermaphrodite shows lateral alae with three characteristic ridges (B). RNAi inactivation of adt-2 causes an expanded region of disruption in the lateral cuticle (double headed arrow; C). Patches of disorganization of the annuli are also evident in the adt-2(tm975/wk156) adult worms (E). Nomarski microscopy images of the cuticle of adt-2(tm975/wk156) hermaphrodites show abnormal numbers of alae (arrowheads; four in D and two in F), as well as discontinuous alae (D; arrows from first asterisk mark region in which third ridge from top is discontinued; second asterisk marks region in which top ridge is discontinued, transitioning from four alae to three alae).

Table 4.

adt-2(wk156) and adt-2(RNAi) worms have constricted annuli.

Genotype	RNAi treatment	Distance between annuli (µm)	Number of annuli	Number of worms
kaIs12[COL-19::GFP]	none	1.809 +/− 0.18	390	15
kaIs12[COL-19::GFP];adt-2(wk156)	none	1.478 +/− 0.14**	688	20
kaIs12[COL-19::GFP]	empty vector control	1.656 +/− 0.20	459	17
kaIs12[COL-19::GFP]	adt-2(RNAi)	1.437 +/− 0.16**	690	21

^**p<0.01

Discussion

sma-21, a gene required for normal body size in C. elegans, encodes ADT-2 ADAMTS secreted metalloprotease

To characterize the genetic regulation of body size, we undertook the molecular cloning of sma-21, a gene required for normal larval and adult body size in the nematode C. elegans. Four independent pieces of evidence demonstrate that sma-21 encodes ADT-2. First, sma-21 mutants contain a missense mutation in a conserved glycine residue in the catalytic domain of ADT-2. Second, inactivation of ADT-2 by RNAi phenocopies the small body size defect of sma-21 mutants. Third, genomic clones containing adt-2 are sufficient to rescue the body size defect of sma-21 mutants. Fourth, a lethal deletion allele of adt-2 fails to complement sma-21(wk156) for the body size defect.

adt-2 mutants and knockdowns display a graded series of defects. The original wk156 mutation likely results in a partial loss of function, since it causes a mild body size phenotype that is enhanced in trans with a deletion allele. The wk156 mutational lesion results in an amino acid substitution for a conserved glycine residue in the zinc binding motif of the catalytic domain. The resulting mutant protein apparently retains partial activity. Knockdown of adt-2 by RNAi causes a more severe body size phenotype than in the wk156 mutant, as well as some lethality. Finally, the deletion allele tm975, which removes a large part of the catalytic domain including the metal binding motif, causes late embryonic lethality.

The molecular identification of SMA-21/ADT-2 demonstrates a role for ADAMTS function in the regulation of body size in C. elegans. This function in body size regulation may be a conserved activity for ADAMTS proteins. Consistent with this model, ADAMTS mutations and variants are associated with normal and pathological variations in human height. For example, ADAMTS2 is associated with connective tissue disorders such as EDS (Ehlers-Danlos syndrome) in humans and with dermatosparaxis in cattle. Both disorders are characterized by fragility of skin and short stature (Colige et al., 1999; Lenaers et al., 1971). A mutation in ADAMTS10 is associated with Weill-Marchesani syndrome (WMS), a disorder that is characterized by the short body size, short fingers and toes, joint stiffness and eye anomalies (Dagoneau et al., 2004). In addition, a mutation in ADAMTSL2 (ADAMTS-like2) leads to geleophysic dysplasia, a condition characterized by short stature and digit abnormalities (Le Goff et al., 2008). Finally, gene variants in human ADAMTS10, ADAMTS17 and ADAMTSL3 (ADAMTS-like3) are associated with variation in human height (Gudbjartsson et al., 2008; Lettre et al., 2008; Weedon et al., 2008).

ADT-2 is required for normal cuticle collagen fibril structure and DBL-1 signaling activity

In mammals, ADAMTS proteases can be divided into two classes: those involved in ECM assembly (procollagen N-proteinases: ADAMTS-2, -3 and -14) and those involved in ECM degradation (aggrecanases: ADAMTS-1, -4, -5, -8, -9 and -15). ADT-2 is one of four members of the ADAMTS family present in C. elegans. Two of these, GON-1 and MIG-17, are more similar to the aggrecanases involved in ECM degradation. These two proteins are involved in cell migrations during organogenesis, and may function by locally degrading the basement membrane ECM. In contrast, ADT-1 and ADT-2 are more similar to the collagenases that are required for ECM assembly. ADT-1 is required for the morphogenesis of male copulatory organs, a process that requires rapid remodeling of the ECM.

In C. elegans one ECM tissue, the cuticle, plays a major role in determining body size due to the fact that it encapsulates C. elegans. adt-2 mutants show a series of defects in the cuticle. We have observed the COL-19::GFP collagen fusion protein marker in adt-2 mutant and RNAi backgrounds, and found that the recruitment of this collagen fusion protein into fibrils in the cuticle is aberrant. We see major disruptions of the COL-19-containing fibrils, particularly in the lateral cuticle underlying the alae. These defects are associated with shorter distances between the rings of annuli indicating excessive longitudinal constriction of the cuticle. Defects in the alae can also be seen by Nomarski DIC microscopy, including the presence of two or four lateral ridges instead of the normal three, and the presence of short discontinuities in the alae ridges. These defects are reminiscent of defects reported in a subset of cuticle collagen mutants such as dpy-5, dpy-13, and bli-1, and in col-19(RNAi) animals. Interestingly, adt-2 was also previously identified in an RNAi screen for genes required for molting of the cuticle (Craig et al., 2007; Frand et al., 2005). Taken together, these results are consistent with a focus of action for ADT-2 in the cuticle. Due to the defects in the cuticle of adt-2 mutants, the cuticle may not be able to maintain its integrity, which leads to the resultant defect in body size. Also, adt-2(wk156) mutants lose length rather than growing during adulthood, which could be caused by a lack of cuticle integrity. These defects in the cuticle can also explain the reduced lifespan of adt-2 mutants as an increased vulnerability to injuries and infection.

To test for an influence of ADT-2 on the DBL-1/BMP signaling pathway, the major size regulating pathway in C. elegans, we employed a DBL-1-responsive transcriptional reporter, RAD-SMAD (Tian et al., 2010). Intriguingly, adt-2 inactivation leads to a reduction in expression of this reporter, indicating that ADT-2 also regulates body size in part by direct or indirect modulation of DBL-1 signaling. Since TGFβ ligands are often secreted in inactive forms that must be activated by proteolysis (Lawrence, 2001), it possible that ADT-2 is directly involved in activation of DBL-1. Based on the other roles of ADT-2 in the cuticle, however, we find it more likely that ADT-2 plays an indirect role. Consistent with this hypothesis, ECM is known to influence the activity of signaling pathways; for example, ECM collagens act as regulators of BMP signaling in Drosophila (Wang et al., 2008). It is therefore possible that ADT-2 modification of the ECM leads to changes in DBL-1 bioavailability.

Materials and Methods

Strains

C. elegans strains were grown at 20°C using standard methods (Brenner, 1974), except for daf-4, which was grown at 15°C to prevent constitutive dauer formation. All of the analyses reported here were performed with a sma-21(wk156) segregant from which the sma-20 enhancer mutation was outcrossed. In addition to strains generated in this study the following strains were used:

• N2 and HA (wild type)

• LG I: rnt-1(ok351)

• LG II: cnb-1(ok276)

• LG III: sma-2(e502), sma-3(wk30), sma-4(e729), daf-4(e1364)

• LG IV: tax-6(ok2065), egl-4(n477), pha-3(ad607)

• LG V: dbl-1(wk70), sma-1(e30), him-5(e1490)

• LG X: adt-2(tm975) obtained from Dr. Shohei Mitani, National Bioresource Project for the nematode, Tokyo Women's Medical University School of Medicine, Japan; sma-9(wk55), che-2(e1033), pha-2(ad472).

• Transgenics: dbl-1(ctIs40) (Suzuki et al., 1999); TP12 [COL-19::GFP] obtained from Dr. Antony Page; wIs51 [SCM::GFP, unc-119(+)]; jcIs1 [ajm-1::gfp] obtained from Dr. Jeff Simske; arIs99 [dpy-7p::2Xnls::yfp] obtained from Dr. Iva Greenwald; jjIs2277[RAD-SMAD + mec-7p::rfp], lon-2(e678);jjIs2277, and sma-6(wk7);jjIs2277 (Tian et al., 2010) obtained from Dr. Jun Liu.

Mapping and cloning of sma-21

We used SNP mapping (Davis et al., 2005) and aCGH (Maydan et al., 2009) as previously described. Briefly, a microarray was designed for a 0.6 MB region delimited by the standard SNP mapping protocol using an application at http://hokkaido.bcgsc.ca/SNPdetection/. Processing of the microarray was done by Roche/NimbleGen. To confirm the polymorphisms identified by aCGH, we generated PCR fragments flanking the mutation sites and directly sequenced them. Primer sequences available on request. The PCR amplification produces 752 bp fragment for the wild-type allele and a 277 bp fragment for the deletion allele. For rescue, 10 ng/µl fosmid DNA was injected into the gonadal syncytia of sma-21(wk156) hermaphrodites with myo-3::mcherry (kindly provided by Dr. Hannes Bülow) as a marker (Mello et al., 1991). Total concentration of the DNA at injection was adjusted to 100 ng/µg using Bluescript SK. yk cDNAs spanning this region were obtained from Dr. Y. Kohara. The yk1586e04 was sequenced and the predicted transcript structure was verified. Primers used for sequencing the yk clone are available upon request.

Measurement of nuclear and cell numbers

We counted the number of nuclei in the hypodermis, seam cells and intestine in wild-type and adt-2 mutant animals. DAPI staining was used to label the intestinal nuclei. N2, adt-2 worms were fixed with acetone, washed with PMB (50mM Na₂HPO₄ pH 7.5, 1mM MgCl₂) and immersed in 500 µl/ml DAPI solution for 30 minutes. Worms were then washed twice with PMB and the intestinal nuclei were counted. The number of seam cell nuclei was counted using the transgene wIs51 [SCM::GFP, unc-119(+)], a seam cell marker in adult animals (Clucas et al., 2002). The number of nuclei in the hypodermis of L4 animals was counted using the transgene arIs99 [dpy-7p::2Xnls::yfp], a hypodermal reporter, which is expressed in hyp7 and other hypodermal cells (Myers and Greenwald, 2005). Individual seam cell perimeters were measured in L3 larvae using the ajm-1::gfp marker (Mohler et al., 1998) which localizes to adherens junctions. The length of the pharynx and the body size of the same L3 worms were also measured. These markers were introduced into adt-2(wk156) by standard genetic crosses. Images were taken using a confocal or epifluorescence microscope.

RNAi feeding

RNAi feeding was performed as described in (Kamath et al., 2001). Six L4 animals were transferred to feeding plates, incubated overnight, transferred to fresh plates and the progeny were scored.

adt-2 expression

We used PCR fusion based approach to create adt-2p::gfp construct (Hobert, 2002). We amplified 4492bp region of genomic DNA upstream of adt-2 to the nearest adjacent gene and fused with a GFP reporter. Primers used for fusion were:

• adt-2A: 5’ GC GGA TCC TAA AAC TAT AGG AAA TTC GGA 3’

• adt-2A*: 5’ GC GGA TCC TTT ATG TAA TAC TAA TAC TGG 3’

• adt-2B: 5’ GAA AAG TTC TTC TCC TTT ACT CAT aat gtt gtt ctg gag ttg gca gaa 3’

• adt-2B*: 5’ GAA AAG TTC TTC TCC TTT AC TCA Ttg aga ata tgc aga ttt cac aac g 3’

20 ng/µl adt-2p::gfp construct was microinjected to N2 hermaphrodites along with myo-3::mcherry as a transformation marker. Total concentration of the DNA at injection was adjusted to 100 ng/µl using Bluescript SK.

Analysis of body size measurements

To characterize the body size phenotype of adt-2, we have created growth curves by measuring the body lengths of wild type N2 worms and adt-2(wk156) at various times after embryo collection. Worms were grown at 20°C and photographed using Axio Vision 3.00 software. The length of each worm was determined by drawing a segmented line along the midline using the same software. To analyze body width, adt-2(wk156) and adt-2(RNAi) adult worms were photographed at 96 hrs after embryo collection and the width measured at a position through the vulva in the center of the body.

For body length measurements of the single and double mutants, 96 hrs old worms (144 hrs for daf-4 and daf-4;adt-2 double mutant grown at 15°C) were photographed using QC Capture 2.73.0 and the length was measured using Image-Pro Express 5.1.0.12 software.

Lifespan assay

Synchronous animals were obtained by bleaching of gravid hermaphrodites. 120 L4 worms from wild type, adt-2(wk156) and adt-2(RNAi) were picked. Adult N2 worms also treated with adt-2 RNAi were tested for their lifespan. From each strain, the L4 worms were divided into 12 plates with 10 worms per plate. Worms were transferred to fresh plates every other day and the number of surviving worms was counted. Worms were marked as dead if they failed to respond to the nose tap with a platinum wire. Kaplan-Meier survival analysis was used to compare the median lifespan of different strains of animals.

Generation of double mutants

Double mutants containing mutations in the sma-6, daf-4, sma-2, sma-3, sma-4, sma-9, sma-1 and rnt-1 genes were created by crossing adt-2(wk156);him-5(e1490) males into the hermaphrodites of each mutant and scoring progeny for the expected phenotypes. The rnt-1(ok351) deletion mutation was confirmed by PCR. All the other mutations were confirmed by complementation tests. che-2(e1033);adt-2(RNAi), egl-4(n477);adt-2(RNAi), tax-6(ok2065);adt-2(RNAi), cnb-1(ok276);adt-2(RNAi), pha-2(ad472);adt-2(RNAi), and pha-3(ad607);adt-2(RNAi) double knockdowns were made by feeding che-2, egl-4, tax-6, cnb-1, pha-2, and pha-3 mutant worms with adt-2 dsRNA expressing bacteria.

COL-19::GFP localization in adt-2(wk156)

In order to see the cuticle of adt-2(wk156), an adult specific marker COL-19::GFP was used (Thein et al., 2003). kaIs[col-19::gfp];adt-2(wk156) was constructed by crossing COL-19::GFP males into adt-2(wk156) hermaphrodites. The small hermaphrodites with COL-19::GFP were picked in the F2 generation. Plates bearing smalls and COL-19::GFP were selected and the cuticle was observed under the confocal microscope. kaIs[col-19::gfp];adt-2(tm975/wk156) was made by crossing kaIs[col-19::gfp];adt-2(wk156) males into adt-2(tm975)/+ hermaphrodites. We also treated COL-19::GFP worms on adt-2(RNAi).

Research Highlights

• ➢Growth and body size are influenced by both genetic and environmental factors.
• ➢We identify adt-2, a gene required for regulation of body size in C. elegans.
• ➢adt-2 encodes a member of the ADAMTS family of secreted metalloproteases.
• ➢ADT-2 controls size by modulating TGFβ signaling and cuticle collagen organization.